Hospital Medicine

In this issue of the Cleveland Clinic Journal of Medicine,¹ Dr. Steven L. Cohn provides a succinct review of the recently published guidelines by the American College of Cardiology and American Heart Association (ACC/AHA) on perioperative cardiovascular evaluation and management of patients undergoing noncardiac surgery.² Although no drastic changes have been made in these guidelines, several significant modifications have been implemented and are highlighted in his review.

See related article

A BREACH OF SCIENTIFIC INTEGRITY

First, I am pleased Dr. Cohn described how the writing committee of the new guidelines handled the well-publicized breaches of scientific integrity by Dr. Don Poldermans, a prolific perioperative-medicine researcher at Erasmus University in the Netherlands who has contributed an abundance of literature that influenced clinical practice. Although some of his key publications were excluded by the ACC/AHA committee in its overall analysis, it remains unclear to me if simply ignoring some of his work is truly possible. For better or for worse, his publications have significantly shaped clinical practice in addition to guiding subsequent research in this field.

ASSESSING RISK

Along with continuing to endorse the Revised Cardiac Risk Index (RCRI),³ the guidelines now include another option for objective preoperative cardiovascular risk assessment. Dr. Cohn nicely outlines the pros and cons of the surgical risk calculator (often referred to as the “Gupta calculator”) derived from the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) database.⁴

Although the RCRI is not perfect, I agree with Dr. Cohn that the ACS NSQIP tool has limitations, including a cumbersome calculation (requiring a smartphone application or online calculator), lack of external validation, and use of the American Society of Anesthesiologists Physical Status Classification System, which has been notoriously confusing for generalists and has demonstrated poor inter-rater reliability among anesthesiologists.^5,6

A patient may have very different risk-prediction scores depending on which tool is used

Of note, a patient may have very different risk-prediction scores depending on which tool is used. For example, a 66-year-old man with a history of ischemic heart disease, diabetes on insulin therapy, hypertension, and chronic kidney disease with a serum creatinine level greater than 2.0 mg/dL who is scheduled to undergo total hip arthroplasty would have a risk of a perioperative cardiovascular event of about 10% according to the RCRI, but only 1.1% according to the ACS NSQIP calculator. How widely this newer risk-stratification tool will be adopted in clinical practice will be interesting to observe.

In what appears to be an effort to simplify the guidelines, the ACC/AHA now recommends combining the patient’s clinical and surgical risks into estimating an overall perioperative risk for developing major adverse cardiac events. This estimate is now whittled down to only two categories: “low risk” and “elevated risk.” I am concerned that the new guidelines may have become too streamlined and lack the direction to assist providers in making important clinical decisions. Most notably, and as Dr. Cohn appropriately suggests, many patients will be in a gray zone with respect to whether cardiac stress testing should be obtained before surgery.

STRESS TESTING

Significant background knowledge is required to answer the important question in the ACC/AHA algorithm, ie, whether further testing will have an impact on decision-making or perioperative care.² Dr. Cohn provides some of this information by noting the abysmal positive predictive value of preoperative noninvasive cardiac testing (with studies ranging from 0% to 37%) and by correctly stating that no benefit has been observed with preoperative cardiac revascularization.

If this is not widely known, I share Dr. Cohn’s fear that the new guidelines may stimulate increased ordering of preoperative stress tests. We observed this trend with the highly scripted 2002 ACC/AHA perioperative guidelines⁷ and subsequently learned that stress testing before surgery very seldom changes patient management.

A preoperative stress test should be reserved for patients with symptoms suggestive of ischemic heart disease. As a diagnostic study, the value of stress testing is excellent. This is not true when it is used as a screening test for asymptomatic patients, where its ability to predict perioperative cardiovascular events is extremely poor. The only other indication for preoperative stress testing is the rare occasion when further risk stratification is desired for exceptionally high-risk patients. In this scenario, test results may influence the decision to proceed with surgery vs seeking nonoperative approaches or palliative care.

MANAGING MEDICATIONS

Dr. Cohn discusses pertinent issues in the perioperative management of patients’ medications, an important component of the preoperative evaluation.

Despite the inconsistent clinical trial results on perioperative beta-blockers, his assessment of their risks and benefits is clinically accurate and practical. Furthermore, I fully agree with Dr. Cohn’s thoughtful approach regarding perioperative statins, despite the limited data available from randomized controlled trials.

With respect to perioperative aspirin use, I have concerns with Dr. Cohn’s statement that it may be reasonable to continue aspirin perioperatively if the risk of potential cardiac events outweighs the risk of bleeding. Given the result of the recently published second Perioperative Ischemic Evaluation (POISE-2) trial⁸ that showed a significantly higher risk of major perioperative bleeding in patients randomized to low-dose aspirin, it is difficult to advocate continuing aspirin when no cardiovascular protection was found in this very large trial. I agree with Dr. Cohn that this applies only to patients with no history of coronary artery stent placement, as patients with a stent should remain on low-dose aspirin throughout the entire perioperative period.

Although we may desire ‘cookbook’ guidelines, we need to practice the art of medicine

Controversy also surrounds angiotensin-converting enzyme inhibitors and angiotensin receptor blockers. Dr. Cohn agrees with the ACC/AHA guidelines to continue these agents before surgery; however, I favor holding them on the day of surgery. Although the risk of hypotension-induced cardiac events has not been clearly demonstrated, a recent retrospective study involving more than 1,100 patients showed significantly more acute kidney injury (even after adjusting for hypotension) as well as an increased length of hospital stay in the patients exposed to these agents before surgery.⁹ Given these findings, in addition to the postinduction hypotension (which can be profound) commonly observed by our anesthesiology colleagues, I recommend holding angiotensin-converting enzyme inhibitors and angiotensin receptor blockers on the day of surgery, with very few exceptions.

THE SCIENCE AND ART OF MEDICINE

Dr. Cohn acknowledges that we lack scientific data to answer many questions that arise when caring for the perioperative patient and thus we rely on the ACC/AHA guidelines to provide a framework. These scientific knowledge gaps emphasize the importance of the art of medicine in the perioperative arena. Although we may desire “cookbook” guidelines, the significant gaps in the perioperative medicine evidence base reinforce the necessity to provide individual patient-level care in a multidisciplinary environment with our surgery and anesthesiology colleagues. Without the proper balance of science and art in perioperative medicine, we sacrifice our ability to deliver optimal care for this high-risk patient population.

In this issue of the Cleveland Clinic Journal of Medicine,¹ Dr. Steven L. Cohn provides a succinct review of the recently published guidelines by the American College of Cardiology and American Heart Association (ACC/AHA) on perioperative cardiovascular evaluation and management of patients undergoing noncardiac surgery.² Although no drastic changes have been made in these guidelines, several significant modifications have been implemented and are highlighted in his review.

See related article

A BREACH OF SCIENTIFIC INTEGRITY

First, I am pleased Dr. Cohn described how the writing committee of the new guidelines handled the well-publicized breaches of scientific integrity by Dr. Don Poldermans, a prolific perioperative-medicine researcher at Erasmus University in the Netherlands who has contributed an abundance of literature that influenced clinical practice. Although some of his key publications were excluded by the ACC/AHA committee in its overall analysis, it remains unclear to me if simply ignoring some of his work is truly possible. For better or for worse, his publications have significantly shaped clinical practice in addition to guiding subsequent research in this field.

ASSESSING RISK

Along with continuing to endorse the Revised Cardiac Risk Index (RCRI),³ the guidelines now include another option for objective preoperative cardiovascular risk assessment. Dr. Cohn nicely outlines the pros and cons of the surgical risk calculator (often referred to as the “Gupta calculator”) derived from the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) database.⁴

Although the RCRI is not perfect, I agree with Dr. Cohn that the ACS NSQIP tool has limitations, including a cumbersome calculation (requiring a smartphone application or online calculator), lack of external validation, and use of the American Society of Anesthesiologists Physical Status Classification System, which has been notoriously confusing for generalists and has demonstrated poor inter-rater reliability among anesthesiologists.^5,6

A patient may have very different risk-prediction scores depending on which tool is used

Of note, a patient may have very different risk-prediction scores depending on which tool is used. For example, a 66-year-old man with a history of ischemic heart disease, diabetes on insulin therapy, hypertension, and chronic kidney disease with a serum creatinine level greater than 2.0 mg/dL who is scheduled to undergo total hip arthroplasty would have a risk of a perioperative cardiovascular event of about 10% according to the RCRI, but only 1.1% according to the ACS NSQIP calculator. How widely this newer risk-stratification tool will be adopted in clinical practice will be interesting to observe.

In what appears to be an effort to simplify the guidelines, the ACC/AHA now recommends combining the patient’s clinical and surgical risks into estimating an overall perioperative risk for developing major adverse cardiac events. This estimate is now whittled down to only two categories: “low risk” and “elevated risk.” I am concerned that the new guidelines may have become too streamlined and lack the direction to assist providers in making important clinical decisions. Most notably, and as Dr. Cohn appropriately suggests, many patients will be in a gray zone with respect to whether cardiac stress testing should be obtained before surgery.

STRESS TESTING

Significant background knowledge is required to answer the important question in the ACC/AHA algorithm, ie, whether further testing will have an impact on decision-making or perioperative care.² Dr. Cohn provides some of this information by noting the abysmal positive predictive value of preoperative noninvasive cardiac testing (with studies ranging from 0% to 37%) and by correctly stating that no benefit has been observed with preoperative cardiac revascularization.

If this is not widely known, I share Dr. Cohn’s fear that the new guidelines may stimulate increased ordering of preoperative stress tests. We observed this trend with the highly scripted 2002 ACC/AHA perioperative guidelines⁷ and subsequently learned that stress testing before surgery very seldom changes patient management.

A preoperative stress test should be reserved for patients with symptoms suggestive of ischemic heart disease. As a diagnostic study, the value of stress testing is excellent. This is not true when it is used as a screening test for asymptomatic patients, where its ability to predict perioperative cardiovascular events is extremely poor. The only other indication for preoperative stress testing is the rare occasion when further risk stratification is desired for exceptionally high-risk patients. In this scenario, test results may influence the decision to proceed with surgery vs seeking nonoperative approaches or palliative care.

MANAGING MEDICATIONS

Dr. Cohn discusses pertinent issues in the perioperative management of patients’ medications, an important component of the preoperative evaluation.

Despite the inconsistent clinical trial results on perioperative beta-blockers, his assessment of their risks and benefits is clinically accurate and practical. Furthermore, I fully agree with Dr. Cohn’s thoughtful approach regarding perioperative statins, despite the limited data available from randomized controlled trials.

With respect to perioperative aspirin use, I have concerns with Dr. Cohn’s statement that it may be reasonable to continue aspirin perioperatively if the risk of potential cardiac events outweighs the risk of bleeding. Given the result of the recently published second Perioperative Ischemic Evaluation (POISE-2) trial⁸ that showed a significantly higher risk of major perioperative bleeding in patients randomized to low-dose aspirin, it is difficult to advocate continuing aspirin when no cardiovascular protection was found in this very large trial. I agree with Dr. Cohn that this applies only to patients with no history of coronary artery stent placement, as patients with a stent should remain on low-dose aspirin throughout the entire perioperative period.

Although we may desire ‘cookbook’ guidelines, we need to practice the art of medicine

Controversy also surrounds angiotensin-converting enzyme inhibitors and angiotensin receptor blockers. Dr. Cohn agrees with the ACC/AHA guidelines to continue these agents before surgery; however, I favor holding them on the day of surgery. Although the risk of hypotension-induced cardiac events has not been clearly demonstrated, a recent retrospective study involving more than 1,100 patients showed significantly more acute kidney injury (even after adjusting for hypotension) as well as an increased length of hospital stay in the patients exposed to these agents before surgery.⁹ Given these findings, in addition to the postinduction hypotension (which can be profound) commonly observed by our anesthesiology colleagues, I recommend holding angiotensin-converting enzyme inhibitors and angiotensin receptor blockers on the day of surgery, with very few exceptions.

THE SCIENCE AND ART OF MEDICINE

Dr. Cohn acknowledges that we lack scientific data to answer many questions that arise when caring for the perioperative patient and thus we rely on the ACC/AHA guidelines to provide a framework. These scientific knowledge gaps emphasize the importance of the art of medicine in the perioperative arena. Although we may desire “cookbook” guidelines, the significant gaps in the perioperative medicine evidence base reinforce the necessity to provide individual patient-level care in a multidisciplinary environment with our surgery and anesthesiology colleagues. Without the proper balance of science and art in perioperative medicine, we sacrifice our ability to deliver optimal care for this high-risk patient population.

In this issue of the Cleveland Clinic Journal of Medicine,¹ Dr. Steven L. Cohn provides a succinct review of the recently published guidelines by the American College of Cardiology and American Heart Association (ACC/AHA) on perioperative cardiovascular evaluation and management of patients undergoing noncardiac surgery.² Although no drastic changes have been made in these guidelines, several significant modifications have been implemented and are highlighted in his review.

See related article

A BREACH OF SCIENTIFIC INTEGRITY

First, I am pleased Dr. Cohn described how the writing committee of the new guidelines handled the well-publicized breaches of scientific integrity by Dr. Don Poldermans, a prolific perioperative-medicine researcher at Erasmus University in the Netherlands who has contributed an abundance of literature that influenced clinical practice. Although some of his key publications were excluded by the ACC/AHA committee in its overall analysis, it remains unclear to me if simply ignoring some of his work is truly possible. For better or for worse, his publications have significantly shaped clinical practice in addition to guiding subsequent research in this field.

ASSESSING RISK

Along with continuing to endorse the Revised Cardiac Risk Index (RCRI),³ the guidelines now include another option for objective preoperative cardiovascular risk assessment. Dr. Cohn nicely outlines the pros and cons of the surgical risk calculator (often referred to as the “Gupta calculator”) derived from the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) database.⁴

Although the RCRI is not perfect, I agree with Dr. Cohn that the ACS NSQIP tool has limitations, including a cumbersome calculation (requiring a smartphone application or online calculator), lack of external validation, and use of the American Society of Anesthesiologists Physical Status Classification System, which has been notoriously confusing for generalists and has demonstrated poor inter-rater reliability among anesthesiologists.^5,6

A patient may have very different risk-prediction scores depending on which tool is used

Of note, a patient may have very different risk-prediction scores depending on which tool is used. For example, a 66-year-old man with a history of ischemic heart disease, diabetes on insulin therapy, hypertension, and chronic kidney disease with a serum creatinine level greater than 2.0 mg/dL who is scheduled to undergo total hip arthroplasty would have a risk of a perioperative cardiovascular event of about 10% according to the RCRI, but only 1.1% according to the ACS NSQIP calculator. How widely this newer risk-stratification tool will be adopted in clinical practice will be interesting to observe.

In what appears to be an effort to simplify the guidelines, the ACC/AHA now recommends combining the patient’s clinical and surgical risks into estimating an overall perioperative risk for developing major adverse cardiac events. This estimate is now whittled down to only two categories: “low risk” and “elevated risk.” I am concerned that the new guidelines may have become too streamlined and lack the direction to assist providers in making important clinical decisions. Most notably, and as Dr. Cohn appropriately suggests, many patients will be in a gray zone with respect to whether cardiac stress testing should be obtained before surgery.

STRESS TESTING

Significant background knowledge is required to answer the important question in the ACC/AHA algorithm, ie, whether further testing will have an impact on decision-making or perioperative care.² Dr. Cohn provides some of this information by noting the abysmal positive predictive value of preoperative noninvasive cardiac testing (with studies ranging from 0% to 37%) and by correctly stating that no benefit has been observed with preoperative cardiac revascularization.

If this is not widely known, I share Dr. Cohn’s fear that the new guidelines may stimulate increased ordering of preoperative stress tests. We observed this trend with the highly scripted 2002 ACC/AHA perioperative guidelines⁷ and subsequently learned that stress testing before surgery very seldom changes patient management.

A preoperative stress test should be reserved for patients with symptoms suggestive of ischemic heart disease. As a diagnostic study, the value of stress testing is excellent. This is not true when it is used as a screening test for asymptomatic patients, where its ability to predict perioperative cardiovascular events is extremely poor. The only other indication for preoperative stress testing is the rare occasion when further risk stratification is desired for exceptionally high-risk patients. In this scenario, test results may influence the decision to proceed with surgery vs seeking nonoperative approaches or palliative care.

MANAGING MEDICATIONS

Dr. Cohn discusses pertinent issues in the perioperative management of patients’ medications, an important component of the preoperative evaluation.

Despite the inconsistent clinical trial results on perioperative beta-blockers, his assessment of their risks and benefits is clinically accurate and practical. Furthermore, I fully agree with Dr. Cohn’s thoughtful approach regarding perioperative statins, despite the limited data available from randomized controlled trials.

With respect to perioperative aspirin use, I have concerns with Dr. Cohn’s statement that it may be reasonable to continue aspirin perioperatively if the risk of potential cardiac events outweighs the risk of bleeding. Given the result of the recently published second Perioperative Ischemic Evaluation (POISE-2) trial⁸ that showed a significantly higher risk of major perioperative bleeding in patients randomized to low-dose aspirin, it is difficult to advocate continuing aspirin when no cardiovascular protection was found in this very large trial. I agree with Dr. Cohn that this applies only to patients with no history of coronary artery stent placement, as patients with a stent should remain on low-dose aspirin throughout the entire perioperative period.

Although we may desire ‘cookbook’ guidelines, we need to practice the art of medicine

Controversy also surrounds angiotensin-converting enzyme inhibitors and angiotensin receptor blockers. Dr. Cohn agrees with the ACC/AHA guidelines to continue these agents before surgery; however, I favor holding them on the day of surgery. Although the risk of hypotension-induced cardiac events has not been clearly demonstrated, a recent retrospective study involving more than 1,100 patients showed significantly more acute kidney injury (even after adjusting for hypotension) as well as an increased length of hospital stay in the patients exposed to these agents before surgery.⁹ Given these findings, in addition to the postinduction hypotension (which can be profound) commonly observed by our anesthesiology colleagues, I recommend holding angiotensin-converting enzyme inhibitors and angiotensin receptor blockers on the day of surgery, with very few exceptions.

THE SCIENCE AND ART OF MEDICINE

Dr. Cohn acknowledges that we lack scientific data to answer many questions that arise when caring for the perioperative patient and thus we rely on the ACC/AHA guidelines to provide a framework. These scientific knowledge gaps emphasize the importance of the art of medicine in the perioperative arena. Although we may desire “cookbook” guidelines, the significant gaps in the perioperative medicine evidence base reinforce the necessity to provide individual patient-level care in a multidisciplinary environment with our surgery and anesthesiology colleagues. Without the proper balance of science and art in perioperative medicine, we sacrifice our ability to deliver optimal care for this high-risk patient population.

In this issue of the Journal, Dr. Kyle Brizendine reviews the basics of the Ebola virus and its natural history, diagnosis, and management.

Like many of you, I have followed the Ebola story with disquietude. So far, the disease has barely touched our country, with fewer than 10 confirmed cases on US soil, but it has had a big impact on our health care system and our national psyche.

The creation of specialized containment and management units may deplete some hospitals and their communities of intensive care beds. Specially trained caregivers will need to be diverted to staff these units, and the public’s fear may dissuade patients from undergoing elective procedures at hospitals caring for patients with Ebola. All of these pose a financial challenge to the hospitals most capable of dealing with these patients.

We have yet to hear about management guidelines dealing with renal replacement therapy and ventilator support, which may extend life but also pose extra risks to caregivers. Do we understand the disease well enough to know when advanced supportive therapies might be futile? Many lessons were learned from the Liberian patient who died of Ebola in Dallas, but many more clinical questions remain. I had hoped that in our sophisticated ICUs patients treated relatively early with aggressive supportive care would likely survive. We do not yet know if that is true. One death does not make it false, but it does give one pause.

About a half dozen other Ebola patients have survived with treatment here, but they were not African. Does genetic background play a role in disease severity and survival? Were the survivors treated sooner or differently in ways that matter? How much of the end-organ damage from the virus is from direct organ infection that cannot be reversed or prevented by even the best supportive treatment? Does the ability of the virus to suppress the immune system doom patients to opportunistic infections during prolonged supportive therapy? Is the viral-associated immunosuppression enough to prevent some patients from mounting an effective innate (interferon-based) or acquired (viral-specific T-cell or humoral) antiviral response? And is transfusing blood from survivors, presumably conferring passive immunity, actually efficacious?

I was relieved there were no new Ebola cases among the staff caring for Mr. Duncan at his second emergency room visit in Dallas, since at that time he was clearly quite ill, viremic, and contagious. Universal safety precautions must have helped. But how did the other nurses become infected, even though they presumably wore better protection? Hopefully, we will gain further understanding of transmissibility and resistance. We need this knowledge to inform safe and manageable protocols of care, particularly if successful vaccine development is delayed.

In this issue of the Journal, Dr. Kyle Brizendine reviews the basics of the Ebola virus and its natural history, diagnosis, and management.

Like many of you, I have followed the Ebola story with disquietude. So far, the disease has barely touched our country, with fewer than 10 confirmed cases on US soil, but it has had a big impact on our health care system and our national psyche.

The creation of specialized containment and management units may deplete some hospitals and their communities of intensive care beds. Specially trained caregivers will need to be diverted to staff these units, and the public’s fear may dissuade patients from undergoing elective procedures at hospitals caring for patients with Ebola. All of these pose a financial challenge to the hospitals most capable of dealing with these patients.

We have yet to hear about management guidelines dealing with renal replacement therapy and ventilator support, which may extend life but also pose extra risks to caregivers. Do we understand the disease well enough to know when advanced supportive therapies might be futile? Many lessons were learned from the Liberian patient who died of Ebola in Dallas, but many more clinical questions remain. I had hoped that in our sophisticated ICUs patients treated relatively early with aggressive supportive care would likely survive. We do not yet know if that is true. One death does not make it false, but it does give one pause.

About a half dozen other Ebola patients have survived with treatment here, but they were not African. Does genetic background play a role in disease severity and survival? Were the survivors treated sooner or differently in ways that matter? How much of the end-organ damage from the virus is from direct organ infection that cannot be reversed or prevented by even the best supportive treatment? Does the ability of the virus to suppress the immune system doom patients to opportunistic infections during prolonged supportive therapy? Is the viral-associated immunosuppression enough to prevent some patients from mounting an effective innate (interferon-based) or acquired (viral-specific T-cell or humoral) antiviral response? And is transfusing blood from survivors, presumably conferring passive immunity, actually efficacious?

I was relieved there were no new Ebola cases among the staff caring for Mr. Duncan at his second emergency room visit in Dallas, since at that time he was clearly quite ill, viremic, and contagious. Universal safety precautions must have helped. But how did the other nurses become infected, even though they presumably wore better protection? Hopefully, we will gain further understanding of transmissibility and resistance. We need this knowledge to inform safe and manageable protocols of care, particularly if successful vaccine development is delayed.

In this issue of the Journal, Dr. Kyle Brizendine reviews the basics of the Ebola virus and its natural history, diagnosis, and management.

Like many of you, I have followed the Ebola story with disquietude. So far, the disease has barely touched our country, with fewer than 10 confirmed cases on US soil, but it has had a big impact on our health care system and our national psyche.

The creation of specialized containment and management units may deplete some hospitals and their communities of intensive care beds. Specially trained caregivers will need to be diverted to staff these units, and the public’s fear may dissuade patients from undergoing elective procedures at hospitals caring for patients with Ebola. All of these pose a financial challenge to the hospitals most capable of dealing with these patients.

We have yet to hear about management guidelines dealing with renal replacement therapy and ventilator support, which may extend life but also pose extra risks to caregivers. Do we understand the disease well enough to know when advanced supportive therapies might be futile? Many lessons were learned from the Liberian patient who died of Ebola in Dallas, but many more clinical questions remain. I had hoped that in our sophisticated ICUs patients treated relatively early with aggressive supportive care would likely survive. We do not yet know if that is true. One death does not make it false, but it does give one pause.

About a half dozen other Ebola patients have survived with treatment here, but they were not African. Does genetic background play a role in disease severity and survival? Were the survivors treated sooner or differently in ways that matter? How much of the end-organ damage from the virus is from direct organ infection that cannot be reversed or prevented by even the best supportive treatment? Does the ability of the virus to suppress the immune system doom patients to opportunistic infections during prolonged supportive therapy? Is the viral-associated immunosuppression enough to prevent some patients from mounting an effective innate (interferon-based) or acquired (viral-specific T-cell or humoral) antiviral response? And is transfusing blood from survivors, presumably conferring passive immunity, actually efficacious?

I was relieved there were no new Ebola cases among the staff caring for Mr. Duncan at his second emergency room visit in Dallas, since at that time he was clearly quite ill, viremic, and contagious. Universal safety precautions must have helped. But how did the other nurses become infected, even though they presumably wore better protection? Hopefully, we will gain further understanding of transmissibility and resistance. We need this knowledge to inform safe and manageable protocols of care, particularly if successful vaccine development is delayed.

A 50-year-old man who returned from a business trip to Nigeria 24 days ago presents with complaints of the sudden onset of fever, diarrhea, myalgia, and headache. He reports 10 bowel movements per day and has seen bloody stools.

During his trip he flew in to Murtala Muhammed International Airport in Lagos, ate meals only in his hotel, and attended meetings in Lagos central business district. He had no exposure to animals, mosquitoes, ticks, or sick people, and no sexual activity. After returning home, he felt well for the first 3 weeks.

The patient has a history of hypertension. He does not smoke, drink alcohol, or use injection drugs. He is married, works with commercial banks and financial institutions, and lives in Cleveland, OH.

On physical examination his temperature is 100.0˚F (37.8˚C), pulse 98, respirations 15, blood pressure 105/70 mm Hg, and weight 78 kg (172 lb). He appears comfortable but is a little diaphoretic. His abdomen is tender to palpation in the epigastrium and slightly to the right; he has no signs of peritonitis. His skin is without rash, bleeding, or bruising. The remainder of the examination is normal.

His white blood cell count is 17 × 10⁹/L, hemoglobin 15 g/dL, hematocrit 41%, and platelet count 172 × 10⁹/L. His sodium level is 126 mmol/L, potassium 3.8 mmol/L, chloride 95 mmol/L, carbon dioxide 20 mmol/L, blood urea nitrogen 11 mg/dL, creatinine 0.7 mg/dL, and glucose 130 mg/dL. His aminotransferase and alkaline phosphatase levels are normal.

Could this patient have Ebola virus disease?

With Ebola virus disease on the rise in West Africa, physicians who encounter patients like this one need to include it in the differential diagnosis. Because the disease is new, many questions are raised for which we as yet have no answers. Here, I will review what we know and do not know in an effort to remove some of the fear and uncertainty.

A NEW DISEASE

Ebola virus disease is a severe hemorrhagic fever caused by negative-sense single-stranded RNA viruses classified by the International Committee on Taxonomy of Viruses as belonging to the genus Ebolavirus in the family Filoviridae. Filoviruses get their name from the Latin filum, or thread-like structure.

The family Filoviridae was discovered in 1967 after inadvertent importation of infected monkeys from Uganda into Yugoslavia and Marburg, Germany. Outbreaks of severe illness occurred in workers at a vaccine plant who came into direct contact with the animals by killing them, removing their kidneys, or preparing primary cell cultures for polio vaccine production.

Ebola virus was discovered in 1976 by Peter Piot, who was working at the Institute of Tropical Medicine in Antwerp, Belgium. The blood of a Belgian woman who had been working in what is now the Democratic Republic of the Congo (formerly Zaire) had been sent to the institute; she and Mabalo Lokela, a school headmaster and the first recorded victim of Ebola virus, had been working near Yambuku, about 96 km from the Ebola River.

Before the 2014 outbreak, all known outbreaks had caused fewer than 2,400 cases across a dozen African countries over 3 decades.

Five species of Ebola virus

The genus Ebolavirus contains five species, each associated with a consistent case-fatality rate and a more or less well-identified endemic area.¹

Zaire ebolavirus was recognized in 1976; it has caused multiple outbreaks, with high case-fatality rates.

Sudan ebolavirus was seen first in the 1970s; it has a 50% case-fatality rate.

Tai Forest ebolavirus has been found in only one person, an ethologist working with deceased chimpanzees.

Bundibugyo ebolavirus emerged in 2007 and has a 30% case-fatality rate.

Reston ebolavirus is maintained in an animal reservoir in the Philippines and is not found in Africa. It caused an outbreak of lethal infection in macaques imported into the United States in 1989. There is evidence that Reston ebolavirus can cause asymptomatic infection in humans. None of the caretakers of the macaques became ill, nor did farmers working with infected pigs, although both groups seroconverted.

A reservoir in bats?

A reservoir in nonhuman primates was initially suspected. However, studies subsequently showed that monkeys are susceptible to rapidly lethal filoviral disease, precluding any role as a host for persistent viral infection. It is likely that Ebola virus is maintained in small animals that serve as a source of infection for both humans and wild primates. A prominent suspect is fruit bats, which are consumed in soup in West Africa.

Transmission is person-to-person or nosocomial

Ebola virus is transmitted by direct contact with body fluids such as blood, urine, sweat, vomitus, semen, and breast milk. Filoviruses can initiate infection via ingestion, inhalation (although probably not Ebola), or passage through breaks in the skin. Droplet inoculation into the mouth or eyes has been shown to result from inadvertent transfer of virus from contaminated hands. Patients transmit the virus while febrile and through later stages of disease, as well as postmortem through contact with the body during funeral preparations. The virus has been isolated in semen for as many as 61 days after illness onset.

One primary human case generates only one to three secondary cases on average, but the case-fatality rate is high

Ebola virus can also be spread nosocomially. In 1976, a 44-year-old teacher sought care for fever at the Yambuku Mission Hospital. He was given parenteral chloroquine as empiric treatment for presumed malaria, which was routine for all febrile patients. However, he had unrecognized Ebola virus infection. Moreover, syringes were rinsed in the same pan of water and reused, which spread the infection to nearly 100 people, all of whom developed fulminant Ebola virus disease and died. Infection then spread to family caregivers, the hospital staff, and those who prepared the bodies for burial.

Nosocomial transmission was also responsible for an outbreak of Lake Victoria Marburg virus in Uige Province in northern Angola in 2005, with 374 putative cases and 329 deaths. When teams from Médecins Sans Frontières started setting up the Marburg ward, there were five patients with hemorrhagic fever in a makeshift isolation room in the hospital, together with corpses that the hospital staff had been too afraid to remove. Healers found in many rural African communities were administering injections in homes or in makeshift clinics with reused needles or syringes.²

There is no evidence that filoviruses are carried by mosquitoes or other biting arthropods. Also, the risk of transmission via fomites appears to be low when currently recommended infection-control guidelines for the viral hemorrhagic fevers are followed.³ One primary human case generates only one to three secondary cases on average.

EBOLA IS AN IMMUNODEFICIENCY VIRUS

The main targets of infection are endothelial cells, mononuclear phagocytes, and hepatocytes. Ebola virus replicates at an unusually high rate. Macrophages infected with Zaire ebolavirus produce tumor necrosis factor alpha, interleukin (IL) 1 beta, IL-6, macrophage chemotactic protein 1, and nitric oxide. Virus-infected macrophages synthesize cell-surface tissue factor, triggering the extrinsic coagulation pathway.

Ebola is an immunodeficiency virus. Dendritic cells, which initiate adaptive immune responses, are a major site of filoviral replication. Infected cells cannot present antigens to naïve lymphocytes. Patients who die of Ebola virus disease do not develop antibodies to the virus. Lymphocytes remain uninfected, but undergo “bystander” apoptosis induced by inflammatory mediators.

CLINICAL MANIFESTATIONS

The incubation period is generally 5 to 7 days (range 2 to 28 days), during which the patient is not infectious. Symptoms begin abruptly, with fever, chills, general malaise, weakness, severe headache, and myalgia. By the time of case detection in West Africa, most patients also had nausea, vomiting, diarrhea, and abdominal pain. Once symptoms arise, patients have high levels of the virus in their blood and fluids and are infectious. Hemorrhagic symptoms have apparently been uncommon in West Africa, occurring in 1.0% to 5.7%, but “unexplained bleeding” has been documented in 18% of cases.⁴ Among those in whom the disease enters its hemorrhagic terminal phase, there is characteristic internal and subcutaneous bleeding, vomiting of blood, and subconjunctival hemorrhage.⁴

Laboratory findings include lymphocytopenia (often with counts as low as 1.0 × 10⁹/L), thrombocytopenia (with counts in the range of 50 to 100 × 10⁹/L), elevated aminotransferase levels (including aspartate aminotransferase levels 7 to 12 times higher than alanine aminotransferase in fatal cases), low total protein (due to capillary leak), and disseminated intravascular coagulation. Those who survive begin to improve in the second week, during which viremia resolves in association with the appearance of virus-specific antibodies.⁴

DIAGNOSIS

In symptomatic patients, Ebola virus infection is diagnosed by detection in blood or body fluids of viral antigens by enzyme-linked immunosorbent assay, or RNA sequences by reverse transcriptase polymerase chain reaction. The diagnosis is confirmed with cell culture (in a BSL-4 containment laboratory) showing characteristic viral particles by electron microscopy.

CARING FOR PATIENTS

The most detailed descriptions of the care of patients with Ebola virus disease have come from Dr. Bruce Ribner, of Emory University Hospital, in an October 2014 report of his experience caring for Ebola-infected patients at Emory University Hospital in Atlanta, GA.⁵ He described fluid losses of 5 to 10 L/day, profound hyponatremia, hypokalemia, and hypocalcemia, which were associated with cardiac arrhythmias and the need for intravenous and oral electrolyte repletion and hemodialysis. Intensive one-to-one nursing was critical, as was the coordination of many medical subspecialties. The Emory team arranged point-of-care testing near the unit and generally kept laboratory testing to a minimum. The team was surprised to learn that commercial carriers refused to transport specimens even when they were licensed for category A agents. Difficulties with the local water authority and waste disposal contractor required the hospital to dedicate an autoclave to process all materials used in clinical care.

TREATMENT: SUPPORTIVE AND EXPERIMENTAL

Treatment is supportive to maintain circulatory function and blood pressure and to correct coagulopathy. However, a variety of vaccines, antibodies, small-molecule agents, and antiviral agents are undergoing testing, mostly in animals at this point.

Vaccines. A therapeutic vaccine that worked only slightly was a live-attenuated recombinant vesicular stomatitis virus expressing Ebola virus transmembrane glycoproteins, which was tested in mice, guinea pigs, and rhesus macaques who had been exposed to Ebola virus.⁶

Incubation is about 5–7 days, during which the patient is not infectious

A preventive vaccine worked better. Stanley et al⁷ evaluated a replication-defective chimpanzee adenovirus 3-vectored vaccine that also contained Ebola virus glycoprotein. They gave macaques a single injection of this vaccine, and then 5 weeks later gave them a lethal dose of Ebola virus. All the vaccinated animals survived the infection, and half (2 of 4) survived when challenged 10 months later. With a prime-boost strategy (modified vaccinia virus Ankara, a poxvirus), all survived when challenged 10 months later.

KZ52, a neutralizing antibody, did not work. Oswald et al⁸ gave a human IgG monoclonal antibody against Zaire Ebola virus, designated KZ52, to four rhesus macaques, challenged them with the virus 24 hours later, and administered a second shot of KZ52 on day 4. All of them died.

ZMAb is a combination of three murine monoclonal antibodies, designated 1H3, 2G4, and 4G7. Ad-IFN is a human adenovirus, serotype 5, that expresses human interferon alpha. Qui et al⁹ gave ZMAb and Ad-IFN to macaques in several experiments. In experiment 1, eight macaques were infected and then were given ZMAb and Ad-IFN 3 days later, and ZMAb again on days 6 and 9. Seven of the eight survived. In a second experiment, Ad-IFN was given first, when the viral load was still less than the limit of detection of known assays, and then ZMAb was given upon detection of viremia and fever. Two of four macaques survived. Control animals had undetectable levels of IgG, whereas Ebola virus GP–specific IgG levels were detected in all survivors. IFN-gamma ELISpots showed high EBOV-GP–specific T-cell response in all survivors.

ZMapp is another cocktail of monoclonal antibodies, containing two from ZMab (2G4 and 4G7), plus a third, c13C6. In experiments in rhesus macaques, three groups of six animals each received three doses of ZMapp at varying times after being infected with Ebola virus: at 3, 6, and 9 days; at 4, 7, and 10 days, and at 5, 8, and 11 days. All 18 macaques treated with ZMapp survived. Thus, Zmapp extended the treatment window to 5 days postexposure.¹⁰ One of the American health care workers who contracted Ebola virus in Liberia received this medication.

HSPA5-PMO. Endoplasmic reticulum chaperone heat shock 70 kDa protein 5 (HSPA5) is instrumental in the maturation of envelope proteins in hepatitis C and influenza A virus. It plays a role in viral entry for coxsackievirus A9 and dengue virus serotype 2, and it may be involved in Ebola viral budding. Phosphorodiamidate morpholino oligomers (PMOs) are a class of antisense DNA nucleotide analogs.

Reid et al¹¹ reported that mice treated with HSPA5–PMO were completely protected from lethal Ebola challenge. Therefore, HSPA5 appears to be a promising target for the development of antifilovirus countermeasures.

Favipiravir, an antiviral agent also known as T-705, is a pyrazinecarboxamide derivative. Invented in 2002 by Toyama Chemicals as an inhibitor of influenza virus replication, it acts as a nucleotide analog, selectively inhibiting the viral RNA-dependent RNA polymerase, or causes lethal mutagenesis upon incorporation into the virus RNA. Favipiravir suppresses Ebola virus replication by 4 log₁₀ units in cell culture.¹²

Mice were challenged with intranasal inoculation of 1,000 focus-forming units of Ebola virus diluted in phosphate-buffered saline. Until the first day of treatment (postinfection day 6), all mice in the T-705 group lost weight similarly to control mice, developed viremia, and showed elevated serum levels of aspartate aminotransferase and alanine aminotransferase. Within 4 days of T-705 treatment (post-infection day 10), the animals had cleared the virus from blood. Surviving mice developed Ebola virus-specific antibodies and CD8+ T cells specific for the viral nucleoprotein.¹²

The authors hypothesized that suppression of virus replication by T-705 allowed the host to mount a virus-specific adaptive immune response, and concluded that T-705 was 100% effective in the treatment of Zaire Ebola virus infection up to postinfection day 6 but was hardly beneficial at the terminal stage of disease.¹² Of note, favipiravir is undergoing phase 2 and phase 3 trials as an anti-influenza agent in Japan.

THE CURRENT OUTBREAK

The current outbreak is with Zaire ebolavirus. It seems to have started in a 2-year-old child who died in Meliandou in Guéckédou Prefecture, Guinea, on December 6, 2013. On March 21, 2014, the Guinea Ministry of Health reported the outbreak of an illness characterized by fever, severe diarrhea, vomiting, and a high case-fatality rate (59%) in 49 persons. On May 25, 2014, Kenema Government Hospital confirmed the first case of Ebola virus disease in Sierra Leone, probably brought there by a traditional healer who had treated Ebola patients from Guinea. Tracing led to 13 additional cases—all women who attended the burial.¹³

The Center for Systems Biology at Harvard University and the Broad Institute of Massachusetts Institute of Technology generated 99 Ebola virus genome sequences from 78 patients with confirmed disease, representing more than 70% of the patients diagnosed with the disease in Sierra Leone from May to mid-June 2014. They found genetic similarity across the sequenced 2014 samples, suggesting a single transmission from the natural reservoir, followed by human-to-human transmission during the outbreak. Continued human-reservoir exposure is unlikely to have contributed to the growth of this epidemic.¹⁴

As of October 14, 2014, there were 8,914 suspected and confirmed cases of Ebola virus infection, and 4,477 deaths.¹⁵

Favipiravir is undergoing phase 2 and phase 3 trials as an anti-influenza agent in Japan

But how did Zaire Ebola virus make the 2,000-mile trek from Central Africa to Guinea in West Africa? There are two possibilities: it has always been present in the region but we just never noticed, or it was recently introduced. Bayesian phylogenetic analyses and sequence divergence studies suggest the virus has been present in bat populations in Guinea without previously infecting humans.

Why Guinea and why Guéckédou? Guinea is one of the poorest countries in the world, ranking 178th of 187 countries on the Human Development Index of the United Nations Development Programme, just behind Liberia (174th) and Sierra Leone (177th). In Guinea, the life expectancy is 56 years and the gross national income per capita is $440. The region has been systematically plundered and the forest decimated by clear-cut logging, leaving the Guinea Forest Region largely deforested, resulting in increased contact between humans and the small animals that serve as the source of infection.¹

LIMITED CAPACITY, EVEN IN THE UNITED STATES

A few hospitals in the United States have dedicated units to handle serious infectious diseases such as Ebola: Emory University Hospital; Nebraska Medicine in Omaha; Providence St. Patrick Hospital in Missoula, MT; and the National Institutes of Health in Bethesda, MD. However, in total they have only 19 beds.

QUESTIONS, ANSWERS—AND MORE QUESTIONS

(The following is from a question-and-answer discussion that followed Dr. Brizendine’s Grand Rounds presentation.)

Q: Are there any differences between survivors and those who die of the disease? A: We do not know. Patient survival depends on early recognition and supportive care. There are disparities in the care of patients. Schieffelin et al¹⁶ analyzed the characteristics of patients who died or who survived in Sierra Leone and found that the mortality rate was higher in older patients and those with a higher viral load on presentation.

Q: Does the virus block production or release of interferon early in infection? A: Yes, it has been shown¹⁷ that Ebola virus protein VP24 inhibits signaling downstream of both interferon alpha/beta and interferon gamma by indirectly impairing the transport of a transcription factor termed STAT1. VP24 is also able to bind STAT1 directly. The resulting suppression of host interferon very early on in the incubation phase is key to the virulence of the virus.

Mutations are occurring but are not changing the characteristics of the Ebola virus

Q: Does infection with one of the viral species confer immunity from other species? A: No, there is no cross-immunity.

Q: How soon do patients test positive? A: About 5 days after exposure, when they develop a fever. At this time patients are highly viremic, which PCR can detect.

Q: Before the virus is detectable in the blood, where is it? A: The liver, endothelial cells, antigen-presenting cells, and adrenal glands.

Q: Do we really need to quarantine ill patients and health care workers returning from Africa, per CDC recommendations? A: We don’t know everything, and some people do make bad decisions, such as traveling while symptomatic. I support a period of observation, although confinement is not reasonable, as it may pose a disincentive to cooperation.

Q: What is the role of giving plasma from survivors? A: Dr. Kent Brantly (see American citizens infected with Ebola) received the blood of a 14-year-old who survived. We don’t know. It is not proved. It did not result in improvement in animal models.

Q: Is the bleeding caused by a mechanism similar to that in enterohemorrhagic Escherichia coli infection? A: No. That is a bacterial toxin, whereas this is more like disseminated intravascular coagulation, with an intrinsic pathway anticoagulation cascade.

Q: How long does the virus remain viable outside the body? A: In one study,¹⁸ Ebola virus could not be recovered from experimentally contaminated surfaces (plastic, metal or glass) at room temperature. In another in which it was dried onto a surface,¹⁹ Ebola virus survived in the dark for several hours between 20 and 25°C. When dried in tissue culture media onto glass and stored at 4°C, it has survived for over 50 days.

Q: How long does the virus remain in breast milk? A: We know it has been detected 15 days after disease onset and think possibly as late as 28 days from symptom onset.³

Q: How are people actually infected? A: I believe people get the virus on their hands and then touch their face, eyes, or mouth. If you are wearing personal protective equipment, it must occur while doffing the equipment.

Q: Could we increase the sensitivity of the test so that we could detect the virus before the onset of symptoms? A: In theory it may be possible. The virus is somewhere in the body during the incubation period. Perhaps we could sample the right compartment in an enriched mononuclear cell line.

Q: When can patients who recover resume their normal activities? A: After their viral load returns to 0, I would still advise abstaining from unprotected sex and from breastfeeding for a few months. but as for other activities, no special precautions are needed.

Q: Does the virus appear to be mutating at a high rate? A: Looking back to 2004, mutations are occurring, but there is no sign that any of these mutations has contributed to the size of the outbreak by changing the characteristics of the Ebola virus. Can it become aerosolized? It has been suggested that the virus that caused the outbreak separated from those that caused past Ebola outbreaks but does not seem to be affecting the spread or efficacy of experimental drugs and vaccines. So, even though it is an RNA virus and mutations are occurring, no serious changes have emerged.¹⁴

BACK TO OUR PATIENT

The differential diagnosis for the patient described at the beginning of this paper includes travelers’ diarrhea, malaria, typhoid fever, yellow fever, meningococcal disease … and Ebola virus disease, although this is much less likely in view of the epidemiology and incubation period of this disease. When his stool was tested by enzyme immunoassay and culture, it was found to be positive for Campylobacter. He recovered with oral rehydration.

A 50-year-old man who returned from a business trip to Nigeria 24 days ago presents with complaints of the sudden onset of fever, diarrhea, myalgia, and headache. He reports 10 bowel movements per day and has seen bloody stools.

During his trip he flew in to Murtala Muhammed International Airport in Lagos, ate meals only in his hotel, and attended meetings in Lagos central business district. He had no exposure to animals, mosquitoes, ticks, or sick people, and no sexual activity. After returning home, he felt well for the first 3 weeks.

The patient has a history of hypertension. He does not smoke, drink alcohol, or use injection drugs. He is married, works with commercial banks and financial institutions, and lives in Cleveland, OH.

On physical examination his temperature is 100.0˚F (37.8˚C), pulse 98, respirations 15, blood pressure 105/70 mm Hg, and weight 78 kg (172 lb). He appears comfortable but is a little diaphoretic. His abdomen is tender to palpation in the epigastrium and slightly to the right; he has no signs of peritonitis. His skin is without rash, bleeding, or bruising. The remainder of the examination is normal.

His white blood cell count is 17 × 10⁹/L, hemoglobin 15 g/dL, hematocrit 41%, and platelet count 172 × 10⁹/L. His sodium level is 126 mmol/L, potassium 3.8 mmol/L, chloride 95 mmol/L, carbon dioxide 20 mmol/L, blood urea nitrogen 11 mg/dL, creatinine 0.7 mg/dL, and glucose 130 mg/dL. His aminotransferase and alkaline phosphatase levels are normal.

Could this patient have Ebola virus disease?

With Ebola virus disease on the rise in West Africa, physicians who encounter patients like this one need to include it in the differential diagnosis. Because the disease is new, many questions are raised for which we as yet have no answers. Here, I will review what we know and do not know in an effort to remove some of the fear and uncertainty.

A NEW DISEASE

Ebola virus disease is a severe hemorrhagic fever caused by negative-sense single-stranded RNA viruses classified by the International Committee on Taxonomy of Viruses as belonging to the genus Ebolavirus in the family Filoviridae. Filoviruses get their name from the Latin filum, or thread-like structure.

The family Filoviridae was discovered in 1967 after inadvertent importation of infected monkeys from Uganda into Yugoslavia and Marburg, Germany. Outbreaks of severe illness occurred in workers at a vaccine plant who came into direct contact with the animals by killing them, removing their kidneys, or preparing primary cell cultures for polio vaccine production.

Ebola virus was discovered in 1976 by Peter Piot, who was working at the Institute of Tropical Medicine in Antwerp, Belgium. The blood of a Belgian woman who had been working in what is now the Democratic Republic of the Congo (formerly Zaire) had been sent to the institute; she and Mabalo Lokela, a school headmaster and the first recorded victim of Ebola virus, had been working near Yambuku, about 96 km from the Ebola River.

Before the 2014 outbreak, all known outbreaks had caused fewer than 2,400 cases across a dozen African countries over 3 decades.

Five species of Ebola virus

The genus Ebolavirus contains five species, each associated with a consistent case-fatality rate and a more or less well-identified endemic area.¹

Zaire ebolavirus was recognized in 1976; it has caused multiple outbreaks, with high case-fatality rates.

Sudan ebolavirus was seen first in the 1970s; it has a 50% case-fatality rate.

Tai Forest ebolavirus has been found in only one person, an ethologist working with deceased chimpanzees.

Bundibugyo ebolavirus emerged in 2007 and has a 30% case-fatality rate.

Reston ebolavirus is maintained in an animal reservoir in the Philippines and is not found in Africa. It caused an outbreak of lethal infection in macaques imported into the United States in 1989. There is evidence that Reston ebolavirus can cause asymptomatic infection in humans. None of the caretakers of the macaques became ill, nor did farmers working with infected pigs, although both groups seroconverted.

A reservoir in bats?

A reservoir in nonhuman primates was initially suspected. However, studies subsequently showed that monkeys are susceptible to rapidly lethal filoviral disease, precluding any role as a host for persistent viral infection. It is likely that Ebola virus is maintained in small animals that serve as a source of infection for both humans and wild primates. A prominent suspect is fruit bats, which are consumed in soup in West Africa.

Transmission is person-to-person or nosocomial

Ebola virus is transmitted by direct contact with body fluids such as blood, urine, sweat, vomitus, semen, and breast milk. Filoviruses can initiate infection via ingestion, inhalation (although probably not Ebola), or passage through breaks in the skin. Droplet inoculation into the mouth or eyes has been shown to result from inadvertent transfer of virus from contaminated hands. Patients transmit the virus while febrile and through later stages of disease, as well as postmortem through contact with the body during funeral preparations. The virus has been isolated in semen for as many as 61 days after illness onset.

One primary human case generates only one to three secondary cases on average, but the case-fatality rate is high

Ebola virus can also be spread nosocomially. In 1976, a 44-year-old teacher sought care for fever at the Yambuku Mission Hospital. He was given parenteral chloroquine as empiric treatment for presumed malaria, which was routine for all febrile patients. However, he had unrecognized Ebola virus infection. Moreover, syringes were rinsed in the same pan of water and reused, which spread the infection to nearly 100 people, all of whom developed fulminant Ebola virus disease and died. Infection then spread to family caregivers, the hospital staff, and those who prepared the bodies for burial.

Nosocomial transmission was also responsible for an outbreak of Lake Victoria Marburg virus in Uige Province in northern Angola in 2005, with 374 putative cases and 329 deaths. When teams from Médecins Sans Frontières started setting up the Marburg ward, there were five patients with hemorrhagic fever in a makeshift isolation room in the hospital, together with corpses that the hospital staff had been too afraid to remove. Healers found in many rural African communities were administering injections in homes or in makeshift clinics with reused needles or syringes.²

There is no evidence that filoviruses are carried by mosquitoes or other biting arthropods. Also, the risk of transmission via fomites appears to be low when currently recommended infection-control guidelines for the viral hemorrhagic fevers are followed.³ One primary human case generates only one to three secondary cases on average.

EBOLA IS AN IMMUNODEFICIENCY VIRUS

The main targets of infection are endothelial cells, mononuclear phagocytes, and hepatocytes. Ebola virus replicates at an unusually high rate. Macrophages infected with Zaire ebolavirus produce tumor necrosis factor alpha, interleukin (IL) 1 beta, IL-6, macrophage chemotactic protein 1, and nitric oxide. Virus-infected macrophages synthesize cell-surface tissue factor, triggering the extrinsic coagulation pathway.

Ebola is an immunodeficiency virus. Dendritic cells, which initiate adaptive immune responses, are a major site of filoviral replication. Infected cells cannot present antigens to naïve lymphocytes. Patients who die of Ebola virus disease do not develop antibodies to the virus. Lymphocytes remain uninfected, but undergo “bystander” apoptosis induced by inflammatory mediators.

CLINICAL MANIFESTATIONS

The incubation period is generally 5 to 7 days (range 2 to 28 days), during which the patient is not infectious. Symptoms begin abruptly, with fever, chills, general malaise, weakness, severe headache, and myalgia. By the time of case detection in West Africa, most patients also had nausea, vomiting, diarrhea, and abdominal pain. Once symptoms arise, patients have high levels of the virus in their blood and fluids and are infectious. Hemorrhagic symptoms have apparently been uncommon in West Africa, occurring in 1.0% to 5.7%, but “unexplained bleeding” has been documented in 18% of cases.⁴ Among those in whom the disease enters its hemorrhagic terminal phase, there is characteristic internal and subcutaneous bleeding, vomiting of blood, and subconjunctival hemorrhage.⁴

Laboratory findings include lymphocytopenia (often with counts as low as 1.0 × 10⁹/L), thrombocytopenia (with counts in the range of 50 to 100 × 10⁹/L), elevated aminotransferase levels (including aspartate aminotransferase levels 7 to 12 times higher than alanine aminotransferase in fatal cases), low total protein (due to capillary leak), and disseminated intravascular coagulation. Those who survive begin to improve in the second week, during which viremia resolves in association with the appearance of virus-specific antibodies.⁴

DIAGNOSIS

In symptomatic patients, Ebola virus infection is diagnosed by detection in blood or body fluids of viral antigens by enzyme-linked immunosorbent assay, or RNA sequences by reverse transcriptase polymerase chain reaction. The diagnosis is confirmed with cell culture (in a BSL-4 containment laboratory) showing characteristic viral particles by electron microscopy.

CARING FOR PATIENTS

The most detailed descriptions of the care of patients with Ebola virus disease have come from Dr. Bruce Ribner, of Emory University Hospital, in an October 2014 report of his experience caring for Ebola-infected patients at Emory University Hospital in Atlanta, GA.⁵ He described fluid losses of 5 to 10 L/day, profound hyponatremia, hypokalemia, and hypocalcemia, which were associated with cardiac arrhythmias and the need for intravenous and oral electrolyte repletion and hemodialysis. Intensive one-to-one nursing was critical, as was the coordination of many medical subspecialties. The Emory team arranged point-of-care testing near the unit and generally kept laboratory testing to a minimum. The team was surprised to learn that commercial carriers refused to transport specimens even when they were licensed for category A agents. Difficulties with the local water authority and waste disposal contractor required the hospital to dedicate an autoclave to process all materials used in clinical care.

TREATMENT: SUPPORTIVE AND EXPERIMENTAL

Treatment is supportive to maintain circulatory function and blood pressure and to correct coagulopathy. However, a variety of vaccines, antibodies, small-molecule agents, and antiviral agents are undergoing testing, mostly in animals at this point.

Vaccines. A therapeutic vaccine that worked only slightly was a live-attenuated recombinant vesicular stomatitis virus expressing Ebola virus transmembrane glycoproteins, which was tested in mice, guinea pigs, and rhesus macaques who had been exposed to Ebola virus.⁶

Incubation is about 5–7 days, during which the patient is not infectious

A preventive vaccine worked better. Stanley et al⁷ evaluated a replication-defective chimpanzee adenovirus 3-vectored vaccine that also contained Ebola virus glycoprotein. They gave macaques a single injection of this vaccine, and then 5 weeks later gave them a lethal dose of Ebola virus. All the vaccinated animals survived the infection, and half (2 of 4) survived when challenged 10 months later. With a prime-boost strategy (modified vaccinia virus Ankara, a poxvirus), all survived when challenged 10 months later.

KZ52, a neutralizing antibody, did not work. Oswald et al⁸ gave a human IgG monoclonal antibody against Zaire Ebola virus, designated KZ52, to four rhesus macaques, challenged them with the virus 24 hours later, and administered a second shot of KZ52 on day 4. All of them died.

ZMAb is a combination of three murine monoclonal antibodies, designated 1H3, 2G4, and 4G7. Ad-IFN is a human adenovirus, serotype 5, that expresses human interferon alpha. Qui et al⁹ gave ZMAb and Ad-IFN to macaques in several experiments. In experiment 1, eight macaques were infected and then were given ZMAb and Ad-IFN 3 days later, and ZMAb again on days 6 and 9. Seven of the eight survived. In a second experiment, Ad-IFN was given first, when the viral load was still less than the limit of detection of known assays, and then ZMAb was given upon detection of viremia and fever. Two of four macaques survived. Control animals had undetectable levels of IgG, whereas Ebola virus GP–specific IgG levels were detected in all survivors. IFN-gamma ELISpots showed high EBOV-GP–specific T-cell response in all survivors.

ZMapp is another cocktail of monoclonal antibodies, containing two from ZMab (2G4 and 4G7), plus a third, c13C6. In experiments in rhesus macaques, three groups of six animals each received three doses of ZMapp at varying times after being infected with Ebola virus: at 3, 6, and 9 days; at 4, 7, and 10 days, and at 5, 8, and 11 days. All 18 macaques treated with ZMapp survived. Thus, Zmapp extended the treatment window to 5 days postexposure.¹⁰ One of the American health care workers who contracted Ebola virus in Liberia received this medication.

HSPA5-PMO. Endoplasmic reticulum chaperone heat shock 70 kDa protein 5 (HSPA5) is instrumental in the maturation of envelope proteins in hepatitis C and influenza A virus. It plays a role in viral entry for coxsackievirus A9 and dengue virus serotype 2, and it may be involved in Ebola viral budding. Phosphorodiamidate morpholino oligomers (PMOs) are a class of antisense DNA nucleotide analogs.

Reid et al¹¹ reported that mice treated with HSPA5–PMO were completely protected from lethal Ebola challenge. Therefore, HSPA5 appears to be a promising target for the development of antifilovirus countermeasures.

Favipiravir, an antiviral agent also known as T-705, is a pyrazinecarboxamide derivative. Invented in 2002 by Toyama Chemicals as an inhibitor of influenza virus replication, it acts as a nucleotide analog, selectively inhibiting the viral RNA-dependent RNA polymerase, or causes lethal mutagenesis upon incorporation into the virus RNA. Favipiravir suppresses Ebola virus replication by 4 log₁₀ units in cell culture.¹²

Mice were challenged with intranasal inoculation of 1,000 focus-forming units of Ebola virus diluted in phosphate-buffered saline. Until the first day of treatment (postinfection day 6), all mice in the T-705 group lost weight similarly to control mice, developed viremia, and showed elevated serum levels of aspartate aminotransferase and alanine aminotransferase. Within 4 days of T-705 treatment (post-infection day 10), the animals had cleared the virus from blood. Surviving mice developed Ebola virus-specific antibodies and CD8+ T cells specific for the viral nucleoprotein.¹²

The authors hypothesized that suppression of virus replication by T-705 allowed the host to mount a virus-specific adaptive immune response, and concluded that T-705 was 100% effective in the treatment of Zaire Ebola virus infection up to postinfection day 6 but was hardly beneficial at the terminal stage of disease.¹² Of note, favipiravir is undergoing phase 2 and phase 3 trials as an anti-influenza agent in Japan.

THE CURRENT OUTBREAK

The current outbreak is with Zaire ebolavirus. It seems to have started in a 2-year-old child who died in Meliandou in Guéckédou Prefecture, Guinea, on December 6, 2013. On March 21, 2014, the Guinea Ministry of Health reported the outbreak of an illness characterized by fever, severe diarrhea, vomiting, and a high case-fatality rate (59%) in 49 persons. On May 25, 2014, Kenema Government Hospital confirmed the first case of Ebola virus disease in Sierra Leone, probably brought there by a traditional healer who had treated Ebola patients from Guinea. Tracing led to 13 additional cases—all women who attended the burial.¹³

The Center for Systems Biology at Harvard University and the Broad Institute of Massachusetts Institute of Technology generated 99 Ebola virus genome sequences from 78 patients with confirmed disease, representing more than 70% of the patients diagnosed with the disease in Sierra Leone from May to mid-June 2014. They found genetic similarity across the sequenced 2014 samples, suggesting a single transmission from the natural reservoir, followed by human-to-human transmission during the outbreak. Continued human-reservoir exposure is unlikely to have contributed to the growth of this epidemic.¹⁴

As of October 14, 2014, there were 8,914 suspected and confirmed cases of Ebola virus infection, and 4,477 deaths.¹⁵

Favipiravir is undergoing phase 2 and phase 3 trials as an anti-influenza agent in Japan

But how did Zaire Ebola virus make the 2,000-mile trek from Central Africa to Guinea in West Africa? There are two possibilities: it has always been present in the region but we just never noticed, or it was recently introduced. Bayesian phylogenetic analyses and sequence divergence studies suggest the virus has been present in bat populations in Guinea without previously infecting humans.

Why Guinea and why Guéckédou? Guinea is one of the poorest countries in the world, ranking 178th of 187 countries on the Human Development Index of the United Nations Development Programme, just behind Liberia (174th) and Sierra Leone (177th). In Guinea, the life expectancy is 56 years and the gross national income per capita is $440. The region has been systematically plundered and the forest decimated by clear-cut logging, leaving the Guinea Forest Region largely deforested, resulting in increased contact between humans and the small animals that serve as the source of infection.¹

LIMITED CAPACITY, EVEN IN THE UNITED STATES

A few hospitals in the United States have dedicated units to handle serious infectious diseases such as Ebola: Emory University Hospital; Nebraska Medicine in Omaha; Providence St. Patrick Hospital in Missoula, MT; and the National Institutes of Health in Bethesda, MD. However, in total they have only 19 beds.

QUESTIONS, ANSWERS—AND MORE QUESTIONS

(The following is from a question-and-answer discussion that followed Dr. Brizendine’s Grand Rounds presentation.)

Q: Are there any differences between survivors and those who die of the disease? A: We do not know. Patient survival depends on early recognition and supportive care. There are disparities in the care of patients. Schieffelin et al¹⁶ analyzed the characteristics of patients who died or who survived in Sierra Leone and found that the mortality rate was higher in older patients and those with a higher viral load on presentation.

Q: Does the virus block production or release of interferon early in infection? A: Yes, it has been shown¹⁷ that Ebola virus protein VP24 inhibits signaling downstream of both interferon alpha/beta and interferon gamma by indirectly impairing the transport of a transcription factor termed STAT1. VP24 is also able to bind STAT1 directly. The resulting suppression of host interferon very early on in the incubation phase is key to the virulence of the virus.

Mutations are occurring but are not changing the characteristics of the Ebola virus

Q: Does infection with one of the viral species confer immunity from other species? A: No, there is no cross-immunity.

Q: How soon do patients test positive? A: About 5 days after exposure, when they develop a fever. At this time patients are highly viremic, which PCR can detect.

Q: Before the virus is detectable in the blood, where is it? A: The liver, endothelial cells, antigen-presenting cells, and adrenal glands.

Q: Do we really need to quarantine ill patients and health care workers returning from Africa, per CDC recommendations? A: We don’t know everything, and some people do make bad decisions, such as traveling while symptomatic. I support a period of observation, although confinement is not reasonable, as it may pose a disincentive to cooperation.

Q: What is the role of giving plasma from survivors? A: Dr. Kent Brantly (see American citizens infected with Ebola) received the blood of a 14-year-old who survived. We don’t know. It is not proved. It did not result in improvement in animal models.

Q: Is the bleeding caused by a mechanism similar to that in enterohemorrhagic Escherichia coli infection? A: No. That is a bacterial toxin, whereas this is more like disseminated intravascular coagulation, with an intrinsic pathway anticoagulation cascade.

Q: How long does the virus remain viable outside the body? A: In one study,¹⁸ Ebola virus could not be recovered from experimentally contaminated surfaces (plastic, metal or glass) at room temperature. In another in which it was dried onto a surface,¹⁹ Ebola virus survived in the dark for several hours between 20 and 25°C. When dried in tissue culture media onto glass and stored at 4°C, it has survived for over 50 days.

Q: How long does the virus remain in breast milk? A: We know it has been detected 15 days after disease onset and think possibly as late as 28 days from symptom onset.³

Q: How are people actually infected? A: I believe people get the virus on their hands and then touch their face, eyes, or mouth. If you are wearing personal protective equipment, it must occur while doffing the equipment.

Q: Could we increase the sensitivity of the test so that we could detect the virus before the onset of symptoms? A: In theory it may be possible. The virus is somewhere in the body during the incubation period. Perhaps we could sample the right compartment in an enriched mononuclear cell line.

Q: When can patients who recover resume their normal activities? A: After their viral load returns to 0, I would still advise abstaining from unprotected sex and from breastfeeding for a few months. but as for other activities, no special precautions are needed.

Q: Does the virus appear to be mutating at a high rate? A: Looking back to 2004, mutations are occurring, but there is no sign that any of these mutations has contributed to the size of the outbreak by changing the characteristics of the Ebola virus. Can it become aerosolized? It has been suggested that the virus that caused the outbreak separated from those that caused past Ebola outbreaks but does not seem to be affecting the spread or efficacy of experimental drugs and vaccines. So, even though it is an RNA virus and mutations are occurring, no serious changes have emerged.¹⁴

BACK TO OUR PATIENT

The differential diagnosis for the patient described at the beginning of this paper includes travelers’ diarrhea, malaria, typhoid fever, yellow fever, meningococcal disease … and Ebola virus disease, although this is much less likely in view of the epidemiology and incubation period of this disease. When his stool was tested by enzyme immunoassay and culture, it was found to be positive for Campylobacter. He recovered with oral rehydration.

A 50-year-old man who returned from a business trip to Nigeria 24 days ago presents with complaints of the sudden onset of fever, diarrhea, myalgia, and headache. He reports 10 bowel movements per day and has seen bloody stools.

During his trip he flew in to Murtala Muhammed International Airport in Lagos, ate meals only in his hotel, and attended meetings in Lagos central business district. He had no exposure to animals, mosquitoes, ticks, or sick people, and no sexual activity. After returning home, he felt well for the first 3 weeks.

The patient has a history of hypertension. He does not smoke, drink alcohol, or use injection drugs. He is married, works with commercial banks and financial institutions, and lives in Cleveland, OH.

On physical examination his temperature is 100.0˚F (37.8˚C), pulse 98, respirations 15, blood pressure 105/70 mm Hg, and weight 78 kg (172 lb). He appears comfortable but is a little diaphoretic. His abdomen is tender to palpation in the epigastrium and slightly to the right; he has no signs of peritonitis. His skin is without rash, bleeding, or bruising. The remainder of the examination is normal.

His white blood cell count is 17 × 10⁹/L, hemoglobin 15 g/dL, hematocrit 41%, and platelet count 172 × 10⁹/L. His sodium level is 126 mmol/L, potassium 3.8 mmol/L, chloride 95 mmol/L, carbon dioxide 20 mmol/L, blood urea nitrogen 11 mg/dL, creatinine 0.7 mg/dL, and glucose 130 mg/dL. His aminotransferase and alkaline phosphatase levels are normal.

Could this patient have Ebola virus disease?

With Ebola virus disease on the rise in West Africa, physicians who encounter patients like this one need to include it in the differential diagnosis. Because the disease is new, many questions are raised for which we as yet have no answers. Here, I will review what we know and do not know in an effort to remove some of the fear and uncertainty.

A NEW DISEASE

Ebola virus disease is a severe hemorrhagic fever caused by negative-sense single-stranded RNA viruses classified by the International Committee on Taxonomy of Viruses as belonging to the genus Ebolavirus in the family Filoviridae. Filoviruses get their name from the Latin filum, or thread-like structure.

The family Filoviridae was discovered in 1967 after inadvertent importation of infected monkeys from Uganda into Yugoslavia and Marburg, Germany. Outbreaks of severe illness occurred in workers at a vaccine plant who came into direct contact with the animals by killing them, removing their kidneys, or preparing primary cell cultures for polio vaccine production.

Ebola virus was discovered in 1976 by Peter Piot, who was working at the Institute of Tropical Medicine in Antwerp, Belgium. The blood of a Belgian woman who had been working in what is now the Democratic Republic of the Congo (formerly Zaire) had been sent to the institute; she and Mabalo Lokela, a school headmaster and the first recorded victim of Ebola virus, had been working near Yambuku, about 96 km from the Ebola River.

Before the 2014 outbreak, all known outbreaks had caused fewer than 2,400 cases across a dozen African countries over 3 decades.

Five species of Ebola virus

The genus Ebolavirus contains five species, each associated with a consistent case-fatality rate and a more or less well-identified endemic area.¹

Zaire ebolavirus was recognized in 1976; it has caused multiple outbreaks, with high case-fatality rates.

Sudan ebolavirus was seen first in the 1970s; it has a 50% case-fatality rate.

Tai Forest ebolavirus has been found in only one person, an ethologist working with deceased chimpanzees.

Bundibugyo ebolavirus emerged in 2007 and has a 30% case-fatality rate.

Reston ebolavirus is maintained in an animal reservoir in the Philippines and is not found in Africa. It caused an outbreak of lethal infection in macaques imported into the United States in 1989. There is evidence that Reston ebolavirus can cause asymptomatic infection in humans. None of the caretakers of the macaques became ill, nor did farmers working with infected pigs, although both groups seroconverted.

A reservoir in bats?

A reservoir in nonhuman primates was initially suspected. However, studies subsequently showed that monkeys are susceptible to rapidly lethal filoviral disease, precluding any role as a host for persistent viral infection. It is likely that Ebola virus is maintained in small animals that serve as a source of infection for both humans and wild primates. A prominent suspect is fruit bats, which are consumed in soup in West Africa.

Transmission is person-to-person or nosocomial

Ebola virus is transmitted by direct contact with body fluids such as blood, urine, sweat, vomitus, semen, and breast milk. Filoviruses can initiate infection via ingestion, inhalation (although probably not Ebola), or passage through breaks in the skin. Droplet inoculation into the mouth or eyes has been shown to result from inadvertent transfer of virus from contaminated hands. Patients transmit the virus while febrile and through later stages of disease, as well as postmortem through contact with the body during funeral preparations. The virus has been isolated in semen for as many as 61 days after illness onset.

One primary human case generates only one to three secondary cases on average, but the case-fatality rate is high

Ebola virus can also be spread nosocomially. In 1976, a 44-year-old teacher sought care for fever at the Yambuku Mission Hospital. He was given parenteral chloroquine as empiric treatment for presumed malaria, which was routine for all febrile patients. However, he had unrecognized Ebola virus infection. Moreover, syringes were rinsed in the same pan of water and reused, which spread the infection to nearly 100 people, all of whom developed fulminant Ebola virus disease and died. Infection then spread to family caregivers, the hospital staff, and those who prepared the bodies for burial.

Nosocomial transmission was also responsible for an outbreak of Lake Victoria Marburg virus in Uige Province in northern Angola in 2005, with 374 putative cases and 329 deaths. When teams from Médecins Sans Frontières started setting up the Marburg ward, there were five patients with hemorrhagic fever in a makeshift isolation room in the hospital, together with corpses that the hospital staff had been too afraid to remove. Healers found in many rural African communities were administering injections in homes or in makeshift clinics with reused needles or syringes.²

There is no evidence that filoviruses are carried by mosquitoes or other biting arthropods. Also, the risk of transmission via fomites appears to be low when currently recommended infection-control guidelines for the viral hemorrhagic fevers are followed.³ One primary human case generates only one to three secondary cases on average.

EBOLA IS AN IMMUNODEFICIENCY VIRUS

The main targets of infection are endothelial cells, mononuclear phagocytes, and hepatocytes. Ebola virus replicates at an unusually high rate. Macrophages infected with Zaire ebolavirus produce tumor necrosis factor alpha, interleukin (IL) 1 beta, IL-6, macrophage chemotactic protein 1, and nitric oxide. Virus-infected macrophages synthesize cell-surface tissue factor, triggering the extrinsic coagulation pathway.

Ebola is an immunodeficiency virus. Dendritic cells, which initiate adaptive immune responses, are a major site of filoviral replication. Infected cells cannot present antigens to naïve lymphocytes. Patients who die of Ebola virus disease do not develop antibodies to the virus. Lymphocytes remain uninfected, but undergo “bystander” apoptosis induced by inflammatory mediators.

CLINICAL MANIFESTATIONS

The incubation period is generally 5 to 7 days (range 2 to 28 days), during which the patient is not infectious. Symptoms begin abruptly, with fever, chills, general malaise, weakness, severe headache, and myalgia. By the time of case detection in West Africa, most patients also had nausea, vomiting, diarrhea, and abdominal pain. Once symptoms arise, patients have high levels of the virus in their blood and fluids and are infectious. Hemorrhagic symptoms have apparently been uncommon in West Africa, occurring in 1.0% to 5.7%, but “unexplained bleeding” has been documented in 18% of cases.⁴ Among those in whom the disease enters its hemorrhagic terminal phase, there is characteristic internal and subcutaneous bleeding, vomiting of blood, and subconjunctival hemorrhage.⁴

Laboratory findings include lymphocytopenia (often with counts as low as 1.0 × 10⁹/L), thrombocytopenia (with counts in the range of 50 to 100 × 10⁹/L), elevated aminotransferase levels (including aspartate aminotransferase levels 7 to 12 times higher than alanine aminotransferase in fatal cases), low total protein (due to capillary leak), and disseminated intravascular coagulation. Those who survive begin to improve in the second week, during which viremia resolves in association with the appearance of virus-specific antibodies.⁴

DIAGNOSIS

In symptomatic patients, Ebola virus infection is diagnosed by detection in blood or body fluids of viral antigens by enzyme-linked immunosorbent assay, or RNA sequences by reverse transcriptase polymerase chain reaction. The diagnosis is confirmed with cell culture (in a BSL-4 containment laboratory) showing characteristic viral particles by electron microscopy.

CARING FOR PATIENTS

The most detailed descriptions of the care of patients with Ebola virus disease have come from Dr. Bruce Ribner, of Emory University Hospital, in an October 2014 report of his experience caring for Ebola-infected patients at Emory University Hospital in Atlanta, GA.⁵ He described fluid losses of 5 to 10 L/day, profound hyponatremia, hypokalemia, and hypocalcemia, which were associated with cardiac arrhythmias and the need for intravenous and oral electrolyte repletion and hemodialysis. Intensive one-to-one nursing was critical, as was the coordination of many medical subspecialties. The Emory team arranged point-of-care testing near the unit and generally kept laboratory testing to a minimum. The team was surprised to learn that commercial carriers refused to transport specimens even when they were licensed for category A agents. Difficulties with the local water authority and waste disposal contractor required the hospital to dedicate an autoclave to process all materials used in clinical care.

TREATMENT: SUPPORTIVE AND EXPERIMENTAL

Treatment is supportive to maintain circulatory function and blood pressure and to correct coagulopathy. However, a variety of vaccines, antibodies, small-molecule agents, and antiviral agents are undergoing testing, mostly in animals at this point.

Vaccines. A therapeutic vaccine that worked only slightly was a live-attenuated recombinant vesicular stomatitis virus expressing Ebola virus transmembrane glycoproteins, which was tested in mice, guinea pigs, and rhesus macaques who had been exposed to Ebola virus.⁶

Incubation is about 5–7 days, during which the patient is not infectious

A preventive vaccine worked better. Stanley et al⁷ evaluated a replication-defective chimpanzee adenovirus 3-vectored vaccine that also contained Ebola virus glycoprotein. They gave macaques a single injection of this vaccine, and then 5 weeks later gave them a lethal dose of Ebola virus. All the vaccinated animals survived the infection, and half (2 of 4) survived when challenged 10 months later. With a prime-boost strategy (modified vaccinia virus Ankara, a poxvirus), all survived when challenged 10 months later.

KZ52, a neutralizing antibody, did not work. Oswald et al⁸ gave a human IgG monoclonal antibody against Zaire Ebola virus, designated KZ52, to four rhesus macaques, challenged them with the virus 24 hours later, and administered a second shot of KZ52 on day 4. All of them died.

ZMAb is a combination of three murine monoclonal antibodies, designated 1H3, 2G4, and 4G7. Ad-IFN is a human adenovirus, serotype 5, that expresses human interferon alpha. Qui et al⁹ gave ZMAb and Ad-IFN to macaques in several experiments. In experiment 1, eight macaques were infected and then were given ZMAb and Ad-IFN 3 days later, and ZMAb again on days 6 and 9. Seven of the eight survived. In a second experiment, Ad-IFN was given first, when the viral load was still less than the limit of detection of known assays, and then ZMAb was given upon detection of viremia and fever. Two of four macaques survived. Control animals had undetectable levels of IgG, whereas Ebola virus GP–specific IgG levels were detected in all survivors. IFN-gamma ELISpots showed high EBOV-GP–specific T-cell response in all survivors.

ZMapp is another cocktail of monoclonal antibodies, containing two from ZMab (2G4 and 4G7), plus a third, c13C6. In experiments in rhesus macaques, three groups of six animals each received three doses of ZMapp at varying times after being infected with Ebola virus: at 3, 6, and 9 days; at 4, 7, and 10 days, and at 5, 8, and 11 days. All 18 macaques treated with ZMapp survived. Thus, Zmapp extended the treatment window to 5 days postexposure.¹⁰ One of the American health care workers who contracted Ebola virus in Liberia received this medication.

HSPA5-PMO. Endoplasmic reticulum chaperone heat shock 70 kDa protein 5 (HSPA5) is instrumental in the maturation of envelope proteins in hepatitis C and influenza A virus. It plays a role in viral entry for coxsackievirus A9 and dengue virus serotype 2, and it may be involved in Ebola viral budding. Phosphorodiamidate morpholino oligomers (PMOs) are a class of antisense DNA nucleotide analogs.

Reid et al¹¹ reported that mice treated with HSPA5–PMO were completely protected from lethal Ebola challenge. Therefore, HSPA5 appears to be a promising target for the development of antifilovirus countermeasures.

Favipiravir, an antiviral agent also known as T-705, is a pyrazinecarboxamide derivative. Invented in 2002 by Toyama Chemicals as an inhibitor of influenza virus replication, it acts as a nucleotide analog, selectively inhibiting the viral RNA-dependent RNA polymerase, or causes lethal mutagenesis upon incorporation into the virus RNA. Favipiravir suppresses Ebola virus replication by 4 log₁₀ units in cell culture.¹²

Mice were challenged with intranasal inoculation of 1,000 focus-forming units of Ebola virus diluted in phosphate-buffered saline. Until the first day of treatment (postinfection day 6), all mice in the T-705 group lost weight similarly to control mice, developed viremia, and showed elevated serum levels of aspartate aminotransferase and alanine aminotransferase. Within 4 days of T-705 treatment (post-infection day 10), the animals had cleared the virus from blood. Surviving mice developed Ebola virus-specific antibodies and CD8+ T cells specific for the viral nucleoprotein.¹²

The authors hypothesized that suppression of virus replication by T-705 allowed the host to mount a virus-specific adaptive immune response, and concluded that T-705 was 100% effective in the treatment of Zaire Ebola virus infection up to postinfection day 6 but was hardly beneficial at the terminal stage of disease.¹² Of note, favipiravir is undergoing phase 2 and phase 3 trials as an anti-influenza agent in Japan.

THE CURRENT OUTBREAK

The current outbreak is with Zaire ebolavirus. It seems to have started in a 2-year-old child who died in Meliandou in Guéckédou Prefecture, Guinea, on December 6, 2013. On March 21, 2014, the Guinea Ministry of Health reported the outbreak of an illness characterized by fever, severe diarrhea, vomiting, and a high case-fatality rate (59%) in 49 persons. On May 25, 2014, Kenema Government Hospital confirmed the first case of Ebola virus disease in Sierra Leone, probably brought there by a traditional healer who had treated Ebola patients from Guinea. Tracing led to 13 additional cases—all women who attended the burial.¹³

The Center for Systems Biology at Harvard University and the Broad Institute of Massachusetts Institute of Technology generated 99 Ebola virus genome sequences from 78 patients with confirmed disease, representing more than 70% of the patients diagnosed with the disease in Sierra Leone from May to mid-June 2014. They found genetic similarity across the sequenced 2014 samples, suggesting a single transmission from the natural reservoir, followed by human-to-human transmission during the outbreak. Continued human-reservoir exposure is unlikely to have contributed to the growth of this epidemic.¹⁴

As of October 14, 2014, there were 8,914 suspected and confirmed cases of Ebola virus infection, and 4,477 deaths.¹⁵

Favipiravir is undergoing phase 2 and phase 3 trials as an anti-influenza agent in Japan

But how did Zaire Ebola virus make the 2,000-mile trek from Central Africa to Guinea in West Africa? There are two possibilities: it has always been present in the region but we just never noticed, or it was recently introduced. Bayesian phylogenetic analyses and sequence divergence studies suggest the virus has been present in bat populations in Guinea without previously infecting humans.

Why Guinea and why Guéckédou? Guinea is one of the poorest countries in the world, ranking 178th of 187 countries on the Human Development Index of the United Nations Development Programme, just behind Liberia (174th) and Sierra Leone (177th). In Guinea, the life expectancy is 56 years and the gross national income per capita is $440. The region has been systematically plundered and the forest decimated by clear-cut logging, leaving the Guinea Forest Region largely deforested, resulting in increased contact between humans and the small animals that serve as the source of infection.¹

LIMITED CAPACITY, EVEN IN THE UNITED STATES

A few hospitals in the United States have dedicated units to handle serious infectious diseases such as Ebola: Emory University Hospital; Nebraska Medicine in Omaha; Providence St. Patrick Hospital in Missoula, MT; and the National Institutes of Health in Bethesda, MD. However, in total they have only 19 beds.

QUESTIONS, ANSWERS—AND MORE QUESTIONS

(The following is from a question-and-answer discussion that followed Dr. Brizendine’s Grand Rounds presentation.)

Q: Are there any differences between survivors and those who die of the disease? A: We do not know. Patient survival depends on early recognition and supportive care. There are disparities in the care of patients. Schieffelin et al¹⁶ analyzed the characteristics of patients who died or who survived in Sierra Leone and found that the mortality rate was higher in older patients and those with a higher viral load on presentation.

Q: Does the virus block production or release of interferon early in infection? A: Yes, it has been shown¹⁷ that Ebola virus protein VP24 inhibits signaling downstream of both interferon alpha/beta and interferon gamma by indirectly impairing the transport of a transcription factor termed STAT1. VP24 is also able to bind STAT1 directly. The resulting suppression of host interferon very early on in the incubation phase is key to the virulence of the virus.

Mutations are occurring but are not changing the characteristics of the Ebola virus

Q: Does infection with one of the viral species confer immunity from other species? A: No, there is no cross-immunity.

Q: How soon do patients test positive? A: About 5 days after exposure, when they develop a fever. At this time patients are highly viremic, which PCR can detect.

Q: Before the virus is detectable in the blood, where is it? A: The liver, endothelial cells, antigen-presenting cells, and adrenal glands.

Q: Do we really need to quarantine ill patients and health care workers returning from Africa, per CDC recommendations? A: We don’t know everything, and some people do make bad decisions, such as traveling while symptomatic. I support a period of observation, although confinement is not reasonable, as it may pose a disincentive to cooperation.

Q: What is the role of giving plasma from survivors? A: Dr. Kent Brantly (see American citizens infected with Ebola) received the blood of a 14-year-old who survived. We don’t know. It is not proved. It did not result in improvement in animal models.

Q: Is the bleeding caused by a mechanism similar to that in enterohemorrhagic Escherichia coli infection? A: No. That is a bacterial toxin, whereas this is more like disseminated intravascular coagulation, with an intrinsic pathway anticoagulation cascade.

Q: How long does the virus remain viable outside the body? A: In one study,¹⁸ Ebola virus could not be recovered from experimentally contaminated surfaces (plastic, metal or glass) at room temperature. In another in which it was dried onto a surface,¹⁹ Ebola virus survived in the dark for several hours between 20 and 25°C. When dried in tissue culture media onto glass and stored at 4°C, it has survived for over 50 days.

Q: How long does the virus remain in breast milk? A: We know it has been detected 15 days after disease onset and think possibly as late as 28 days from symptom onset.³

Q: How are people actually infected? A: I believe people get the virus on their hands and then touch their face, eyes, or mouth. If you are wearing personal protective equipment, it must occur while doffing the equipment.

Q: Could we increase the sensitivity of the test so that we could detect the virus before the onset of symptoms? A: In theory it may be possible. The virus is somewhere in the body during the incubation period. Perhaps we could sample the right compartment in an enriched mononuclear cell line.

Q: When can patients who recover resume their normal activities? A: After their viral load returns to 0, I would still advise abstaining from unprotected sex and from breastfeeding for a few months. but as for other activities, no special precautions are needed.

Q: Does the virus appear to be mutating at a high rate? A: Looking back to 2004, mutations are occurring, but there is no sign that any of these mutations has contributed to the size of the outbreak by changing the characteristics of the Ebola virus. Can it become aerosolized? It has been suggested that the virus that caused the outbreak separated from those that caused past Ebola outbreaks but does not seem to be affecting the spread or efficacy of experimental drugs and vaccines. So, even though it is an RNA virus and mutations are occurring, no serious changes have emerged.¹⁴

BACK TO OUR PATIENT

The differential diagnosis for the patient described at the beginning of this paper includes travelers’ diarrhea, malaria, typhoid fever, yellow fever, meningococcal disease … and Ebola virus disease, although this is much less likely in view of the epidemiology and incubation period of this disease. When his stool was tested by enzyme immunoassay and culture, it was found to be positive for Campylobacter. He recovered with oral rehydration.

Continued progress in understanding the pathophysiology of acute respiratory distress syndrome (ARDS) is translating into changes in the way we diagnose and manage it. Over the past 20 years, low tidal volume,¹ positive end-expiratory pressure (PEEP),² and fluid restriction³ have become the standard of care. A multidisciplinary approach, including targeted use of sedatives, early mobilization, and protocols for weaning from the ventilator, has also brought about substantial changes in ARDS management and its outcomes.^4–6

In this article, we review the most relevant articles about ARDS in the last 5 years. We include the new definition of ARDS and studies of ventilatory and nonventilatory therapies that have implications in managing patients with ARDS.

A STANDARDIZED APPROACH

ARDS is characterized by damage to the alveolar architecture, severe hypoxemia, and bilateral parenchymal opacities.

The working definition of ARDS developed in 1994 by the American-European Consensus Conference (AECC) was the basis for enrollment in most of the landmark trials and observational studies over the past 20 years.^7,8 However, it was limited in its reliability and validity.

An updated definition

In 2011, the ARDS Definition Task Force, using a novel consensus process, updated the ARDS definition,⁹ focusing on its feasibility, reliability, and validity in predicting response to therapies and outcomes in ARDS. This new “Berlin” definition is not substantially different from the old, but defines the criteria more specifically:

• Bilateral opacities, unexplained by nodules, atelectasis, or effusion, on chest radiography or computed tomography
• New or worsening respiratory symptoms, or a clinical insult associated with ARDS within 7 days of diagnosis
• Objective assessment of cardiac function (eg, with echocardiography) to exclude cardiogenic pulmonary edema
• Hypoxemia, with a partial pressure of arterial oxygen divided by the percentage of inspired oxygen (PaO₂/FiO₂ ratio) of 300 mm Hg or less despite noninvasive or invasive mechanical ventilation with PEEP or continuous positive airway pressure (CPAP) of at least 5 cm H₂O.

In addition, the new definition classifies the severity of disease on the basis of the degree of hypoxemia, ie, the PaO₂/FiO₂ ratio:

• Mild: PaO₂/FiO₂ ratio > 200 and ≤ 300 mm Hg
• Moderate: PaO₂/FiO₂ ratio > 100 and ≤ 200 mm Hg
• Severe: PaO₂/FiO₂ ratio ≤ 100 mm Hg.

The term “acute lung injury” has been eliminated, as has the previous criterion of a pulmonary artery wedge pressure of 18 mm Hg or less.

The panel also evaluated four ancillary variables for predicting outcomes in severe ARDS:

• Compliance of the respiratory system less than or equal to 40 mL/cm H₂O
• Radiographic severity (involvement of three or four quadrants on chest radiography)
• PEEP of 10 cm H₂O or greater
• Corrected expired volume 10 L/min or greater.

The task force evaluated the reliability and validity of this definition in a meta-analysis of 4,400 patients previously enrolled in randomized controlled trials or observational studies.

Findings. The Berlin definition predicted the risk of death better than the AECC definition. The mortality rate increased with the severity of ARDS, from 27% with mild disease to 32% with moderate disease to 45% with severe disease. The four ancillary variables did not contribute to the predictive validity of severe ARDS for mortality and were removed from the definition.

Thille et al¹⁰ retrospectively reviewed autopsy findings from 712 patients and found that the new definition identified a homogeneous group who had severe ARDS.¹⁰

Conclusions. The new definition may overcome some of the limitations of the old one, but it needs to be validated in clinical practice, especially its ability to predict death.

VENTILATORY SUPPORT

Prompt recognition, lung-protective ventilation, and a conservative fluid strategy remain the cornerstones of ARDS management. However, other strategies are being tested.

Prone-position ventilation in severe ARDS: The right therapy in a specific population

Prone-position ventilation was first described almost 30 years ago, but it has been used inconsistently in clinical practice.

Physiologic and observational studies indicated that prone positioning might improve survival in patients with ARDS, but several randomized trials failed to demonstrate any positive effect on outcomes.^11,12 Some trials also reported a higher rate of complications with this intervention.¹³ However, meta-analyses suggested that prone-position ventilation might have a beneficial effect in patients with severe ARDS (defined as a PaO₂/FiO₂ ratio ≤ 100 mm Hg).¹⁴

In view of these findings, investigators conducted a trial of prone-position ventilation exclusively in patients with severe ARDS.

The PROSEVA study

The Proning Severe ARDS Patients (PROSEVA) study was a randomized controlled trial designed to determine whether prone-position ventilation, applied early, would improve outcomes in patients with severe ARDS.¹⁵

In PROSEVA, 466 patients with severe ARDS (defined as a PaO₂/FiO₂ ratio < 150 mm Hg, FiO₂ ≥ 60%, and PEEP ≥ 5 cm H₂O) underwent either at least 16 hours of prone positioning or were left in the supine position after 12 to 24 hours of initial conventional mechanical ventilation. The patients were recruited from centers in France and Spain where prone-position ventilation had been used in daily practice for more than 5 years.

The primary outcome studied was the rate of death at 28 days. The secondary end points were the death rate at day 90, rates of successful extubation, the length of stay in the intensive care unit, and complications.

Findings. At study entry, the patients in the supine group were sicker, more of them required a vasopressor, and fewer of them were receiving neuromuscular blocking agents than those in the prone group. These baseline differences may have influenced the outcomes; the unadjusted 28-day mortality rate was 16.0% in the prone group compared with 32.8% in the supine group (P < .001). However, the hazard ratio for death with prone positioning was 0.39 (95% confidence interval [CI] 0.25–0.63) even after adjusting for severity and the use of vasopressors and neuromuscular blocking agents. Prone-position ventilation was not associated with a higher incidence of complications, and the rate of successful extubation was higher.

Conclusions. In patients with severe ARDS, early use of prolonged prone positioning significantly decreased the 28-day and 90-day mortality rates. This trial has made prone positioning one of the strategies in managing patients with early severe ARDS. To minimize complications such as pressure ulcers and line or tube dislodgement, personnel caring for these patients must follow a protocol and undergo specific training.

These results were corroborated by a meta-analysis by Beitler et al¹⁶ that found a significant decrease in mortality rate with prone-position ventilation even in older studies when lung-protective ventilation strategies were separated from high-tidal-volume ventilation.

High-frequency oscillatory ventilation: No benefit in two trials

Observational data and experimental studies suggested that high-frequency oscillatory ventilation (HFOV) is superior to conventional mechanical ventilation in ARDS patients.^17,18 However, outdated and cumbersome equipment, lack of protocols, and a lack of high-quality evidence led to limited and inconsistent use of HFOV, mainly as a rescue therapy in ARDS.¹⁹

Over the last few years, HFOV has been gaining acceptance, especially earlier in the course of ARDS.²⁰ After preliminary clinical trials reported promising results, two trials conducted in Canada and the United Kingdom compared HFOV vs conventional mechanical ventilation in patients with ARDS.

The OSCAR study

The Oscillation in ARDS (OSCAR) study²¹ was a “pragmatic” trial²² (ie, it had minimal exclusion criteria) of the safety and effectiveness of HFOV as a primary ventilatory strategy for ARDS. It included 795 patients randomized to receive conventional ventilation (n = 397) or HFOV (n = 398). Research centers followed detailed algorithms for HFOV management and adopted their usual practice for conventional ventilation. Medical care was given according to the clinician’s judgment.

The primary outcome studied was survival at 30 days. The secondary outcomes were all-cause mortality in the intensive care unit and the hospital, duration of mechanical ventilation, and use of antimicrobial, sedative, vasoactive, and neuromuscular-blocking drugs.

Findings. The patient baseline characteristics were similar in both groups.

There was no significant difference in intensive care unit mortality rates, hospital mortality rates, or mortality rates at 30 days (41.7% in the HFOV group vs 41.1% in the conventional ventilation group; P = .85, 95% CI 6.1–7.5) even after adjustments for center or severity of illness.

The duration of mechanical ventilation was similar in both groups (14.9 ± 13.3 days in the HFOV group vs 14.1 ± 13.4 days in the conventional ventilation group, P = .41). However, sedatives and neuromuscular-blocking drugs were used more often and longer in the HFOV group than in the conventional ventilation group. There was no difference in the use of vasoactive or antimicrobial medications.

Conclusions. This multicenter randomized control trial did not demonstrate any benefit from using HFOV for routine management of ARDS. Its pragmatic design made it less likely to reach a firm conclusion,²² but it at least made a case against routinely using HFOV in patients with ARDS.

The OSCILLATE study

The Oscillation for Acute Respiratory Distress Syndrome Treated Early (OSCILLATE) study²³ assessed the safety and efficacy of HFOV as a treatment for early-onset moderate-to-severe ARDS.

The inclusion criteria were similar to those in the OSCAR trial except that pulmonary symptoms had to be present less than 2 weeks and ARDS assessment was done under standard ventilator settings. As this was an efficacy trial, it had more exclusion criteria than the OSCAR trial. A total of 548 patients were randomized to receive conventional ventilation (n = 273) or HFOV (n = 275). The baseline characteristics were similar between groups.

Conventional ventilation was given according to a protocol used in an earlier trial² and included recruitment maneuvers. HFOV was given in centers that had experience in this treatment, and there were protocols for ventilation management, hemodynamic optimization, and weaning. All other care was left to the clinician’s choice.

The primary outcome studied was in-hospital mortality. The investigators also evaluated whether there were interactions between the treatment and baseline severity of lung injury and center experience with HFOV.

Findings. The trial was stopped after an interim analysis found that HFOV might be harmful, although the statistical threshold for stopping was not reached. The in-hospital mortality rate was 47% in the HFOV group and 35% in the control group (relative risk of death with HFOV 1.33, 95% CI 1.09–1.64, P = .005). HFOV was worse than conventional ventilation regardless of the severity of disease or center experience. The HFOV group had higher mean airway pressures but similar FiO₂ compared with the conventional ventilation group.

The HFOV group received significantly more vasopressors, sedatives, and neuromuscular blockers. This group’s fluid balance was higher as well, but not significantly so. Refractory hypoxemia (defined as PaO₂ < 60 mm Hg for 1 hour with an FiO₂ of 1.0 and neuromuscular blockade) was more frequent in the conventional ventilation group, but the number of deaths in the subgroup with refractory hypoxemia was similar with either treatment.

Conclusions. This multicenter randomized controlled trial demonstrated that HFOV was harmful when used routinely to manage ARDS. The trial’s protocol was based on the results of a pilot study carried out by the same investigators, which provided the best evidence available regarding the safety of HFOV at that time.

The results of the OSCAR and OSCILLATE trials have quelled enthusiasm for early, routine use of HFOV in ARDS. Although there are concerns that the protocol (ie, the way HFOV was implemented) rather than HFOV itself may have led to worse outcomes, there is no signal to support its routine use. We need further studies to define if it remains a viable rescue therapy.

Extracorporeal membrane oxygenation: Is it a viable option in severe ARDS?

Extracorporeal membrane oxygenation (ECMO) uses cardiopulmonary bypass technology to provide gas exchange. In patients with severe hypoxemia, ECMO can ensure adequate oxygenation and ventilation while ensuring the optimization of lung-protective ventilation. But ECMO was never as successful in adults with ARDS as it was in children and neonates.²⁴

The first two trials of ECMO in ARDS^24,25 reported equal or worse survival rates compared with conventional ventilation, and the overall mortality rate in these studies was staggeringly high. However, these studies were carried out before the era of lung-protective ventilation and at a time when ECMO technology was relatively primitive.

With new technology such as venovenous circuits and smaller cannulas, ECMO has gained more acceptance. It was used in patients with severe or refractory hypoxemia associated with ARDS during the H1N1 pandemic.^26,27

The CESAR trial

The Conventional Ventilatory Support Versus Extracorporeal Membrane Oxygenation for Severe Adult Respiratory Failure (CESAR) trial²⁸ assessed the safety, clinical efficacy, and cost-effectiveness of ECMO in managing severe ARDS. It compared best standard practice vs a protocol that included ECMO. The trial was conducted from 2001 to 2006.

Patients with severe ARDS, as defined by a Murray score²⁹ greater than 3 or uncompensated hypercapnea, were prospectively randomized and recruited from an ECMO center and 148 tertiary intensive care units and referral hospitals in England. This was a pragmatic trial, with minimal exclusion criteria (essentially, mechanical ventilation with high pressures and high FiO₂ for more than 7 days, intracranial bleeding, or contraindication to heparinization).

A total of 180 patients were randomized in a one-to-one ratio to receive ECMO or conventional management. The ventilator management in the conventional treatment group was not done according to a protocol but in general was low-volume and low-pressure. All patients randomized to ECMO were transferred to the ECMO center and treated according to a standardized ventilation protocol. After 12 hours, if predefined goals were not reached, venovenous ECMO was started. Patients assigned to conventional management could not cross over to ECMO.

The primary outcomes were death or severe disability at 6 months after randomization, and cost-effectiveness. The secondary outcomes were hospital resource use (eg, rescue techniques, length of stay, duration of ECMO) and health status after 6 months.

Findings. The groups were similar at baseline. Sixty-eight (75%) of the 90 patients randomized to receive ECMO actually received it. Of the 22 patients who did not receive ECMO, 16 (18% of the 90) improved on conventional therapy, 5 (6%) died during or before transfer, and 1 could not receive heparin.

Two patients had severe complications in the ECMO group: one had an arterial puncture, and one had an oxygen delivery failure during transport. In each case, these events contributed to the death of the patient.

More patients in the ECMO group received lung-protective ventilation, 84 (93%) vs 63 (70%).

The primary outcome, ie, death or severe disability at 6 months, occurred in 33 (37%) of the 90 patients in the ECMO group and in 46 (53%) of the patients in the conventional management group (relative risk 0.69, 95% CI 0.05–0.97, P = .03). More patients in the ECMO group survived, but the difference was not statistically significant (relative risk of death 0.73, 95% CI 0.52–1.03, P = .07). The most common cause of death in the ECMO group was multiorgan failure (42%), whereas in the conventional management group, the most common cause of death was respiratory failure (60%).

Length of stay in the hospital and in the critical care unit and health care costs were double for patients in the ECMO group. There was no difference in quality-of-life markers at 6 months in the survivors.

Conclusions. This pragmatic trial demonstrated that a protocol that includes ECMO could improve survival rates in ARDS.

Of note, the ECMO group got care in regional centers that used protocols. Therefore, in interpreting the results of this trial, we have to consider that being in a center with protocol-specified care for ARDS could drive some of the difference in mortality rates.

Regardless, this trial demonstrated that ECMO is feasible and led to better outcomes than expected. The findings were encouraging, and spurred the use of ECMO in severe ARDS during the 2009 H1N1 pandemic. Two propensity-matched studies and a number of case series reported a survival benefit associated with the use of ECMO in patients with severe ARDS.^27,30

A recent meta-analysis also reported that ECMO might lower the mortality rate in ARDS; however, the patients in the H1N1 pandemic were younger and usually had isolated respiratory failure.³¹

The success of ECMO has opened new possibilities in the management of ARDS. As the technology improves and our experience increases, ECMO will likely gain more acceptance as a treatment for severe ARDS.

Airway pressure release ventilation

The use of airway pressure release ventilation and other ventilator modalities in ARDS is not supported by current evidence, though results of clinical trials may influence our practice in the future.

PHARMACOTHERAPY IN ARDS

The pathogenesis of ARDS includes damage to the alveolar-capillary membrane, with leakage of protein-rich edema fluid into alveoli. This damage is propagated by a complex inflammatory response including but not limited to neutrophil activation, free-radical formation, dysregulation of the coagulation system, and extensive release of inflammatory mediators.^32,33 As a consequence, there are multiple potential targets for pharmacologic therapy in ARDS.

A variety of drugs, including corticosteroids, anti-inflammatory agents, immune-modulating agents, pulmonary vasodilators, antioxidants, and surfactants, have been studied in patients with ARDS.³⁴ But effective pharmacotherapy for ARDS remains extremely limited.

Neuromuscular blockade in early severe ARDS

Mechanical ventilation can result in injurious stretching of the lung parenchyma, either from alveolar overdistention (volutrauma) or from continual recruitment and derecruitment of unstable lung units during the ventilator cycle (atelectrauma).³⁵ Ventilator-induced lung injury can be exacerbated by asynchronous breathing.

In theory, neuromuscular blockers could minimize patient-ventilator asynchrony and provide much better control of tidal volume and pressure in patients with ARDS. This may result in less volutrauma and atelectrauma associated with asynchronous breathing. Data also suggest that cisatracurium (Nimbex), a neuromuscular blocking agent, may have a direct effect on the amount of inflammation in lungs with ARDS.³⁶

The ACURASYS study

The ARDS et Curarisation Systématique (ACURASYS) study³⁷ was a randomized trial in 340 patients undergoing mechanical ventilation for severe ARDS to evaluate the impact of neuromuscular blockade within the first 48 hours in this population.

The primary outcome was the mortality rate before hospital discharge or within 90 days of study entry. Secondary outcomes included the 28-day mortality rate, the rate of intensive care unit-acquired paresis, and the number of ventilator-free days. To be included, patients had to have been mechanically ventilated for less than 48 hours and to meet the AECC criteria for severe ARDS, with a PaO₂/FiO₂ ratio less than 150 mm Hg.

The intervention group received a continuous infusion of cisatracurium for 48 hours, while the control patients received placebo. Muscle strength was evaluated by clinical scoring of strength in different muscle groups.

Findings. The study groups were similar at baseline.

The crude 90-day mortality rate was lower in the cisatracurium group (31.6% vs 40.7%, P = .08). Regression analysis showed an improved 90-day survival rate with the use of this neuromuscular blocker after adjustment for severity of illness and the severity of ARDS (based on degree of hypoxemia and plateau pressures) (hazard ratio for death at 90 days 0.68; 95% CI 0.48–0.98; P = .04). The rate of paresis acquired in the intensive care unit did not differ significantly between the two groups.

Conclusion. In patients with severe ARDS, giving a neuromuscular blocking agent early improved the survival rate and increased the time off the ventilator without increasing muscle weakness.

These data are in line with similar findings from two other studies published by the same group.^38,39 A meta-analysis of 432 patients showed that the use of neuromuscular blockade in early severe ARDS is associated with a statistically significant effect on early mortality (relative risk 0.66, 95% CI 0.50–0.87).⁴⁰ The pooled analysis of these trials did not show any statistically significant critical-illness polyneuropathy.

These results need to be interpreted carefully, as we have inadequate data to see if they generalize to different intensive care units, and the evaluation and categorization of critical-illness polyneuropathy remains to be defined.

Cisatracurium is a promising treatment for moderate to severe ARDS and merits investigation in a large confirmatory randomized controlled trial.

Other pharmacologic agents

A number of other drugs have been studied in ARDS patients, including both inhaled and intravenous beta agonists,^41,42 statins,⁴³ and nutritional supplements.⁴⁴ But as with other drugs previously studied in ARDS such as corticosteroids, N-acetylcysteine, and surfactant,³⁴ these agents showed no effect on outcomes. In fact, a recent trial of intravenous salbutamol in ARDS patients was stopped after an interim analysis because of a higher incidence of arrhythmias and lactic acidosis with this agent.⁴²

These findings reaffirm that pharmacologic therapy needs to be carefully considered, and potential harms associated with these therapies need to be addressed before they are introduced in the care of critically ill patients.

Continued progress in understanding the pathophysiology of acute respiratory distress syndrome (ARDS) is translating into changes in the way we diagnose and manage it. Over the past 20 years, low tidal volume,¹ positive end-expiratory pressure (PEEP),² and fluid restriction³ have become the standard of care. A multidisciplinary approach, including targeted use of sedatives, early mobilization, and protocols for weaning from the ventilator, has also brought about substantial changes in ARDS management and its outcomes.^4–6

In this article, we review the most relevant articles about ARDS in the last 5 years. We include the new definition of ARDS and studies of ventilatory and nonventilatory therapies that have implications in managing patients with ARDS.

A STANDARDIZED APPROACH

ARDS is characterized by damage to the alveolar architecture, severe hypoxemia, and bilateral parenchymal opacities.

The working definition of ARDS developed in 1994 by the American-European Consensus Conference (AECC) was the basis for enrollment in most of the landmark trials and observational studies over the past 20 years.^7,8 However, it was limited in its reliability and validity.

An updated definition

In 2011, the ARDS Definition Task Force, using a novel consensus process, updated the ARDS definition,⁹ focusing on its feasibility, reliability, and validity in predicting response to therapies and outcomes in ARDS. This new “Berlin” definition is not substantially different from the old, but defines the criteria more specifically:

• Bilateral opacities, unexplained by nodules, atelectasis, or effusion, on chest radiography or computed tomography
• New or worsening respiratory symptoms, or a clinical insult associated with ARDS within 7 days of diagnosis
• Objective assessment of cardiac function (eg, with echocardiography) to exclude cardiogenic pulmonary edema
• Hypoxemia, with a partial pressure of arterial oxygen divided by the percentage of inspired oxygen (PaO₂/FiO₂ ratio) of 300 mm Hg or less despite noninvasive or invasive mechanical ventilation with PEEP or continuous positive airway pressure (CPAP) of at least 5 cm H₂O.

In addition, the new definition classifies the severity of disease on the basis of the degree of hypoxemia, ie, the PaO₂/FiO₂ ratio:

• Mild: PaO₂/FiO₂ ratio > 200 and ≤ 300 mm Hg
• Moderate: PaO₂/FiO₂ ratio > 100 and ≤ 200 mm Hg
• Severe: PaO₂/FiO₂ ratio ≤ 100 mm Hg.

The term “acute lung injury” has been eliminated, as has the previous criterion of a pulmonary artery wedge pressure of 18 mm Hg or less.

The panel also evaluated four ancillary variables for predicting outcomes in severe ARDS:

• Compliance of the respiratory system less than or equal to 40 mL/cm H₂O
• Radiographic severity (involvement of three or four quadrants on chest radiography)
• PEEP of 10 cm H₂O or greater
• Corrected expired volume 10 L/min or greater.

The task force evaluated the reliability and validity of this definition in a meta-analysis of 4,400 patients previously enrolled in randomized controlled trials or observational studies.

Findings. The Berlin definition predicted the risk of death better than the AECC definition. The mortality rate increased with the severity of ARDS, from 27% with mild disease to 32% with moderate disease to 45% with severe disease. The four ancillary variables did not contribute to the predictive validity of severe ARDS for mortality and were removed from the definition.

Thille et al¹⁰ retrospectively reviewed autopsy findings from 712 patients and found that the new definition identified a homogeneous group who had severe ARDS.¹⁰

Conclusions. The new definition may overcome some of the limitations of the old one, but it needs to be validated in clinical practice, especially its ability to predict death.

VENTILATORY SUPPORT

Prompt recognition, lung-protective ventilation, and a conservative fluid strategy remain the cornerstones of ARDS management. However, other strategies are being tested.

Prone-position ventilation in severe ARDS: The right therapy in a specific population

Prone-position ventilation was first described almost 30 years ago, but it has been used inconsistently in clinical practice.

Physiologic and observational studies indicated that prone positioning might improve survival in patients with ARDS, but several randomized trials failed to demonstrate any positive effect on outcomes.^11,12 Some trials also reported a higher rate of complications with this intervention.¹³ However, meta-analyses suggested that prone-position ventilation might have a beneficial effect in patients with severe ARDS (defined as a PaO₂/FiO₂ ratio ≤ 100 mm Hg).¹⁴

In view of these findings, investigators conducted a trial of prone-position ventilation exclusively in patients with severe ARDS.

The PROSEVA study

The Proning Severe ARDS Patients (PROSEVA) study was a randomized controlled trial designed to determine whether prone-position ventilation, applied early, would improve outcomes in patients with severe ARDS.¹⁵

In PROSEVA, 466 patients with severe ARDS (defined as a PaO₂/FiO₂ ratio < 150 mm Hg, FiO₂ ≥ 60%, and PEEP ≥ 5 cm H₂O) underwent either at least 16 hours of prone positioning or were left in the supine position after 12 to 24 hours of initial conventional mechanical ventilation. The patients were recruited from centers in France and Spain where prone-position ventilation had been used in daily practice for more than 5 years.

The primary outcome studied was the rate of death at 28 days. The secondary end points were the death rate at day 90, rates of successful extubation, the length of stay in the intensive care unit, and complications.

Findings. At study entry, the patients in the supine group were sicker, more of them required a vasopressor, and fewer of them were receiving neuromuscular blocking agents than those in the prone group. These baseline differences may have influenced the outcomes; the unadjusted 28-day mortality rate was 16.0% in the prone group compared with 32.8% in the supine group (P < .001). However, the hazard ratio for death with prone positioning was 0.39 (95% confidence interval [CI] 0.25–0.63) even after adjusting for severity and the use of vasopressors and neuromuscular blocking agents. Prone-position ventilation was not associated with a higher incidence of complications, and the rate of successful extubation was higher.

Conclusions. In patients with severe ARDS, early use of prolonged prone positioning significantly decreased the 28-day and 90-day mortality rates. This trial has made prone positioning one of the strategies in managing patients with early severe ARDS. To minimize complications such as pressure ulcers and line or tube dislodgement, personnel caring for these patients must follow a protocol and undergo specific training.

These results were corroborated by a meta-analysis by Beitler et al¹⁶ that found a significant decrease in mortality rate with prone-position ventilation even in older studies when lung-protective ventilation strategies were separated from high-tidal-volume ventilation.

High-frequency oscillatory ventilation: No benefit in two trials

Observational data and experimental studies suggested that high-frequency oscillatory ventilation (HFOV) is superior to conventional mechanical ventilation in ARDS patients.^17,18 However, outdated and cumbersome equipment, lack of protocols, and a lack of high-quality evidence led to limited and inconsistent use of HFOV, mainly as a rescue therapy in ARDS.¹⁹

Over the last few years, HFOV has been gaining acceptance, especially earlier in the course of ARDS.²⁰ After preliminary clinical trials reported promising results, two trials conducted in Canada and the United Kingdom compared HFOV vs conventional mechanical ventilation in patients with ARDS.

The OSCAR study

The Oscillation in ARDS (OSCAR) study²¹ was a “pragmatic” trial²² (ie, it had minimal exclusion criteria) of the safety and effectiveness of HFOV as a primary ventilatory strategy for ARDS. It included 795 patients randomized to receive conventional ventilation (n = 397) or HFOV (n = 398). Research centers followed detailed algorithms for HFOV management and adopted their usual practice for conventional ventilation. Medical care was given according to the clinician’s judgment.

The primary outcome studied was survival at 30 days. The secondary outcomes were all-cause mortality in the intensive care unit and the hospital, duration of mechanical ventilation, and use of antimicrobial, sedative, vasoactive, and neuromuscular-blocking drugs.

Findings. The patient baseline characteristics were similar in both groups.

There was no significant difference in intensive care unit mortality rates, hospital mortality rates, or mortality rates at 30 days (41.7% in the HFOV group vs 41.1% in the conventional ventilation group; P = .85, 95% CI 6.1–7.5) even after adjustments for center or severity of illness.

The duration of mechanical ventilation was similar in both groups (14.9 ± 13.3 days in the HFOV group vs 14.1 ± 13.4 days in the conventional ventilation group, P = .41). However, sedatives and neuromuscular-blocking drugs were used more often and longer in the HFOV group than in the conventional ventilation group. There was no difference in the use of vasoactive or antimicrobial medications.

Conclusions. This multicenter randomized control trial did not demonstrate any benefit from using HFOV for routine management of ARDS. Its pragmatic design made it less likely to reach a firm conclusion,²² but it at least made a case against routinely using HFOV in patients with ARDS.

The OSCILLATE study

The Oscillation for Acute Respiratory Distress Syndrome Treated Early (OSCILLATE) study²³ assessed the safety and efficacy of HFOV as a treatment for early-onset moderate-to-severe ARDS.

The inclusion criteria were similar to those in the OSCAR trial except that pulmonary symptoms had to be present less than 2 weeks and ARDS assessment was done under standard ventilator settings. As this was an efficacy trial, it had more exclusion criteria than the OSCAR trial. A total of 548 patients were randomized to receive conventional ventilation (n = 273) or HFOV (n = 275). The baseline characteristics were similar between groups.

Conventional ventilation was given according to a protocol used in an earlier trial² and included recruitment maneuvers. HFOV was given in centers that had experience in this treatment, and there were protocols for ventilation management, hemodynamic optimization, and weaning. All other care was left to the clinician’s choice.

The primary outcome studied was in-hospital mortality. The investigators also evaluated whether there were interactions between the treatment and baseline severity of lung injury and center experience with HFOV.

Findings. The trial was stopped after an interim analysis found that HFOV might be harmful, although the statistical threshold for stopping was not reached. The in-hospital mortality rate was 47% in the HFOV group and 35% in the control group (relative risk of death with HFOV 1.33, 95% CI 1.09–1.64, P = .005). HFOV was worse than conventional ventilation regardless of the severity of disease or center experience. The HFOV group had higher mean airway pressures but similar FiO₂ compared with the conventional ventilation group.

The HFOV group received significantly more vasopressors, sedatives, and neuromuscular blockers. This group’s fluid balance was higher as well, but not significantly so. Refractory hypoxemia (defined as PaO₂ < 60 mm Hg for 1 hour with an FiO₂ of 1.0 and neuromuscular blockade) was more frequent in the conventional ventilation group, but the number of deaths in the subgroup with refractory hypoxemia was similar with either treatment.

Conclusions. This multicenter randomized controlled trial demonstrated that HFOV was harmful when used routinely to manage ARDS. The trial’s protocol was based on the results of a pilot study carried out by the same investigators, which provided the best evidence available regarding the safety of HFOV at that time.

The results of the OSCAR and OSCILLATE trials have quelled enthusiasm for early, routine use of HFOV in ARDS. Although there are concerns that the protocol (ie, the way HFOV was implemented) rather than HFOV itself may have led to worse outcomes, there is no signal to support its routine use. We need further studies to define if it remains a viable rescue therapy.

Extracorporeal membrane oxygenation: Is it a viable option in severe ARDS?

Extracorporeal membrane oxygenation (ECMO) uses cardiopulmonary bypass technology to provide gas exchange. In patients with severe hypoxemia, ECMO can ensure adequate oxygenation and ventilation while ensuring the optimization of lung-protective ventilation. But ECMO was never as successful in adults with ARDS as it was in children and neonates.²⁴

The first two trials of ECMO in ARDS^24,25 reported equal or worse survival rates compared with conventional ventilation, and the overall mortality rate in these studies was staggeringly high. However, these studies were carried out before the era of lung-protective ventilation and at a time when ECMO technology was relatively primitive.

With new technology such as venovenous circuits and smaller cannulas, ECMO has gained more acceptance. It was used in patients with severe or refractory hypoxemia associated with ARDS during the H1N1 pandemic.^26,27

The CESAR trial

The Conventional Ventilatory Support Versus Extracorporeal Membrane Oxygenation for Severe Adult Respiratory Failure (CESAR) trial²⁸ assessed the safety, clinical efficacy, and cost-effectiveness of ECMO in managing severe ARDS. It compared best standard practice vs a protocol that included ECMO. The trial was conducted from 2001 to 2006.

Patients with severe ARDS, as defined by a Murray score²⁹ greater than 3 or uncompensated hypercapnea, were prospectively randomized and recruited from an ECMO center and 148 tertiary intensive care units and referral hospitals in England. This was a pragmatic trial, with minimal exclusion criteria (essentially, mechanical ventilation with high pressures and high FiO₂ for more than 7 days, intracranial bleeding, or contraindication to heparinization).

A total of 180 patients were randomized in a one-to-one ratio to receive ECMO or conventional management. The ventilator management in the conventional treatment group was not done according to a protocol but in general was low-volume and low-pressure. All patients randomized to ECMO were transferred to the ECMO center and treated according to a standardized ventilation protocol. After 12 hours, if predefined goals were not reached, venovenous ECMO was started. Patients assigned to conventional management could not cross over to ECMO.

The primary outcomes were death or severe disability at 6 months after randomization, and cost-effectiveness. The secondary outcomes were hospital resource use (eg, rescue techniques, length of stay, duration of ECMO) and health status after 6 months.

Findings. The groups were similar at baseline. Sixty-eight (75%) of the 90 patients randomized to receive ECMO actually received it. Of the 22 patients who did not receive ECMO, 16 (18% of the 90) improved on conventional therapy, 5 (6%) died during or before transfer, and 1 could not receive heparin.

Two patients had severe complications in the ECMO group: one had an arterial puncture, and one had an oxygen delivery failure during transport. In each case, these events contributed to the death of the patient.

More patients in the ECMO group received lung-protective ventilation, 84 (93%) vs 63 (70%).

The primary outcome, ie, death or severe disability at 6 months, occurred in 33 (37%) of the 90 patients in the ECMO group and in 46 (53%) of the patients in the conventional management group (relative risk 0.69, 95% CI 0.05–0.97, P = .03). More patients in the ECMO group survived, but the difference was not statistically significant (relative risk of death 0.73, 95% CI 0.52–1.03, P = .07). The most common cause of death in the ECMO group was multiorgan failure (42%), whereas in the conventional management group, the most common cause of death was respiratory failure (60%).

Length of stay in the hospital and in the critical care unit and health care costs were double for patients in the ECMO group. There was no difference in quality-of-life markers at 6 months in the survivors.

Conclusions. This pragmatic trial demonstrated that a protocol that includes ECMO could improve survival rates in ARDS.

Of note, the ECMO group got care in regional centers that used protocols. Therefore, in interpreting the results of this trial, we have to consider that being in a center with protocol-specified care for ARDS could drive some of the difference in mortality rates.

Regardless, this trial demonstrated that ECMO is feasible and led to better outcomes than expected. The findings were encouraging, and spurred the use of ECMO in severe ARDS during the 2009 H1N1 pandemic. Two propensity-matched studies and a number of case series reported a survival benefit associated with the use of ECMO in patients with severe ARDS.^27,30

A recent meta-analysis also reported that ECMO might lower the mortality rate in ARDS; however, the patients in the H1N1 pandemic were younger and usually had isolated respiratory failure.³¹

The success of ECMO has opened new possibilities in the management of ARDS. As the technology improves and our experience increases, ECMO will likely gain more acceptance as a treatment for severe ARDS.

Airway pressure release ventilation

The use of airway pressure release ventilation and other ventilator modalities in ARDS is not supported by current evidence, though results of clinical trials may influence our practice in the future.

PHARMACOTHERAPY IN ARDS

The pathogenesis of ARDS includes damage to the alveolar-capillary membrane, with leakage of protein-rich edema fluid into alveoli. This damage is propagated by a complex inflammatory response including but not limited to neutrophil activation, free-radical formation, dysregulation of the coagulation system, and extensive release of inflammatory mediators.^32,33 As a consequence, there are multiple potential targets for pharmacologic therapy in ARDS.

A variety of drugs, including corticosteroids, anti-inflammatory agents, immune-modulating agents, pulmonary vasodilators, antioxidants, and surfactants, have been studied in patients with ARDS.³⁴ But effective pharmacotherapy for ARDS remains extremely limited.

Neuromuscular blockade in early severe ARDS

Mechanical ventilation can result in injurious stretching of the lung parenchyma, either from alveolar overdistention (volutrauma) or from continual recruitment and derecruitment of unstable lung units during the ventilator cycle (atelectrauma).³⁵ Ventilator-induced lung injury can be exacerbated by asynchronous breathing.

In theory, neuromuscular blockers could minimize patient-ventilator asynchrony and provide much better control of tidal volume and pressure in patients with ARDS. This may result in less volutrauma and atelectrauma associated with asynchronous breathing. Data also suggest that cisatracurium (Nimbex), a neuromuscular blocking agent, may have a direct effect on the amount of inflammation in lungs with ARDS.³⁶

The ACURASYS study

The ARDS et Curarisation Systématique (ACURASYS) study³⁷ was a randomized trial in 340 patients undergoing mechanical ventilation for severe ARDS to evaluate the impact of neuromuscular blockade within the first 48 hours in this population.

The primary outcome was the mortality rate before hospital discharge or within 90 days of study entry. Secondary outcomes included the 28-day mortality rate, the rate of intensive care unit-acquired paresis, and the number of ventilator-free days. To be included, patients had to have been mechanically ventilated for less than 48 hours and to meet the AECC criteria for severe ARDS, with a PaO₂/FiO₂ ratio less than 150 mm Hg.

The intervention group received a continuous infusion of cisatracurium for 48 hours, while the control patients received placebo. Muscle strength was evaluated by clinical scoring of strength in different muscle groups.

Findings. The study groups were similar at baseline.

The crude 90-day mortality rate was lower in the cisatracurium group (31.6% vs 40.7%, P = .08). Regression analysis showed an improved 90-day survival rate with the use of this neuromuscular blocker after adjustment for severity of illness and the severity of ARDS (based on degree of hypoxemia and plateau pressures) (hazard ratio for death at 90 days 0.68; 95% CI 0.48–0.98; P = .04). The rate of paresis acquired in the intensive care unit did not differ significantly between the two groups.

Conclusion. In patients with severe ARDS, giving a neuromuscular blocking agent early improved the survival rate and increased the time off the ventilator without increasing muscle weakness.

These data are in line with similar findings from two other studies published by the same group.^38,39 A meta-analysis of 432 patients showed that the use of neuromuscular blockade in early severe ARDS is associated with a statistically significant effect on early mortality (relative risk 0.66, 95% CI 0.50–0.87).⁴⁰ The pooled analysis of these trials did not show any statistically significant critical-illness polyneuropathy.

These results need to be interpreted carefully, as we have inadequate data to see if they generalize to different intensive care units, and the evaluation and categorization of critical-illness polyneuropathy remains to be defined.

Cisatracurium is a promising treatment for moderate to severe ARDS and merits investigation in a large confirmatory randomized controlled trial.

Other pharmacologic agents

A number of other drugs have been studied in ARDS patients, including both inhaled and intravenous beta agonists,^41,42 statins,⁴³ and nutritional supplements.⁴⁴ But as with other drugs previously studied in ARDS such as corticosteroids, N-acetylcysteine, and surfactant,³⁴ these agents showed no effect on outcomes. In fact, a recent trial of intravenous salbutamol in ARDS patients was stopped after an interim analysis because of a higher incidence of arrhythmias and lactic acidosis with this agent.⁴²

These findings reaffirm that pharmacologic therapy needs to be carefully considered, and potential harms associated with these therapies need to be addressed before they are introduced in the care of critically ill patients.

Continued progress in understanding the pathophysiology of acute respiratory distress syndrome (ARDS) is translating into changes in the way we diagnose and manage it. Over the past 20 years, low tidal volume,¹ positive end-expiratory pressure (PEEP),² and fluid restriction³ have become the standard of care. A multidisciplinary approach, including targeted use of sedatives, early mobilization, and protocols for weaning from the ventilator, has also brought about substantial changes in ARDS management and its outcomes.^4–6

In this article, we review the most relevant articles about ARDS in the last 5 years. We include the new definition of ARDS and studies of ventilatory and nonventilatory therapies that have implications in managing patients with ARDS.

A STANDARDIZED APPROACH

ARDS is characterized by damage to the alveolar architecture, severe hypoxemia, and bilateral parenchymal opacities.

The working definition of ARDS developed in 1994 by the American-European Consensus Conference (AECC) was the basis for enrollment in most of the landmark trials and observational studies over the past 20 years.^7,8 However, it was limited in its reliability and validity.

An updated definition

In 2011, the ARDS Definition Task Force, using a novel consensus process, updated the ARDS definition,⁹ focusing on its feasibility, reliability, and validity in predicting response to therapies and outcomes in ARDS. This new “Berlin” definition is not substantially different from the old, but defines the criteria more specifically:

• Bilateral opacities, unexplained by nodules, atelectasis, or effusion, on chest radiography or computed tomography
• New or worsening respiratory symptoms, or a clinical insult associated with ARDS within 7 days of diagnosis
• Objective assessment of cardiac function (eg, with echocardiography) to exclude cardiogenic pulmonary edema
• Hypoxemia, with a partial pressure of arterial oxygen divided by the percentage of inspired oxygen (PaO₂/FiO₂ ratio) of 300 mm Hg or less despite noninvasive or invasive mechanical ventilation with PEEP or continuous positive airway pressure (CPAP) of at least 5 cm H₂O.

In addition, the new definition classifies the severity of disease on the basis of the degree of hypoxemia, ie, the PaO₂/FiO₂ ratio:

• Mild: PaO₂/FiO₂ ratio > 200 and ≤ 300 mm Hg
• Moderate: PaO₂/FiO₂ ratio > 100 and ≤ 200 mm Hg
• Severe: PaO₂/FiO₂ ratio ≤ 100 mm Hg.

The term “acute lung injury” has been eliminated, as has the previous criterion of a pulmonary artery wedge pressure of 18 mm Hg or less.

The panel also evaluated four ancillary variables for predicting outcomes in severe ARDS:

• Compliance of the respiratory system less than or equal to 40 mL/cm H₂O
• Radiographic severity (involvement of three or four quadrants on chest radiography)
• PEEP of 10 cm H₂O or greater
• Corrected expired volume 10 L/min or greater.

The task force evaluated the reliability and validity of this definition in a meta-analysis of 4,400 patients previously enrolled in randomized controlled trials or observational studies.

Findings. The Berlin definition predicted the risk of death better than the AECC definition. The mortality rate increased with the severity of ARDS, from 27% with mild disease to 32% with moderate disease to 45% with severe disease. The four ancillary variables did not contribute to the predictive validity of severe ARDS for mortality and were removed from the definition.

Thille et al¹⁰ retrospectively reviewed autopsy findings from 712 patients and found that the new definition identified a homogeneous group who had severe ARDS.¹⁰

Conclusions. The new definition may overcome some of the limitations of the old one, but it needs to be validated in clinical practice, especially its ability to predict death.

VENTILATORY SUPPORT

Prompt recognition, lung-protective ventilation, and a conservative fluid strategy remain the cornerstones of ARDS management. However, other strategies are being tested.

Prone-position ventilation in severe ARDS: The right therapy in a specific population

Prone-position ventilation was first described almost 30 years ago, but it has been used inconsistently in clinical practice.

Physiologic and observational studies indicated that prone positioning might improve survival in patients with ARDS, but several randomized trials failed to demonstrate any positive effect on outcomes.^11,12 Some trials also reported a higher rate of complications with this intervention.¹³ However, meta-analyses suggested that prone-position ventilation might have a beneficial effect in patients with severe ARDS (defined as a PaO₂/FiO₂ ratio ≤ 100 mm Hg).¹⁴

In view of these findings, investigators conducted a trial of prone-position ventilation exclusively in patients with severe ARDS.

The PROSEVA study

The Proning Severe ARDS Patients (PROSEVA) study was a randomized controlled trial designed to determine whether prone-position ventilation, applied early, would improve outcomes in patients with severe ARDS.¹⁵

In PROSEVA, 466 patients with severe ARDS (defined as a PaO₂/FiO₂ ratio < 150 mm Hg, FiO₂ ≥ 60%, and PEEP ≥ 5 cm H₂O) underwent either at least 16 hours of prone positioning or were left in the supine position after 12 to 24 hours of initial conventional mechanical ventilation. The patients were recruited from centers in France and Spain where prone-position ventilation had been used in daily practice for more than 5 years.

The primary outcome studied was the rate of death at 28 days. The secondary end points were the death rate at day 90, rates of successful extubation, the length of stay in the intensive care unit, and complications.

Findings. At study entry, the patients in the supine group were sicker, more of them required a vasopressor, and fewer of them were receiving neuromuscular blocking agents than those in the prone group. These baseline differences may have influenced the outcomes; the unadjusted 28-day mortality rate was 16.0% in the prone group compared with 32.8% in the supine group (P < .001). However, the hazard ratio for death with prone positioning was 0.39 (95% confidence interval [CI] 0.25–0.63) even after adjusting for severity and the use of vasopressors and neuromuscular blocking agents. Prone-position ventilation was not associated with a higher incidence of complications, and the rate of successful extubation was higher.

Conclusions. In patients with severe ARDS, early use of prolonged prone positioning significantly decreased the 28-day and 90-day mortality rates. This trial has made prone positioning one of the strategies in managing patients with early severe ARDS. To minimize complications such as pressure ulcers and line or tube dislodgement, personnel caring for these patients must follow a protocol and undergo specific training.

These results were corroborated by a meta-analysis by Beitler et al¹⁶ that found a significant decrease in mortality rate with prone-position ventilation even in older studies when lung-protective ventilation strategies were separated from high-tidal-volume ventilation.

High-frequency oscillatory ventilation: No benefit in two trials

Observational data and experimental studies suggested that high-frequency oscillatory ventilation (HFOV) is superior to conventional mechanical ventilation in ARDS patients.^17,18 However, outdated and cumbersome equipment, lack of protocols, and a lack of high-quality evidence led to limited and inconsistent use of HFOV, mainly as a rescue therapy in ARDS.¹⁹

Over the last few years, HFOV has been gaining acceptance, especially earlier in the course of ARDS.²⁰ After preliminary clinical trials reported promising results, two trials conducted in Canada and the United Kingdom compared HFOV vs conventional mechanical ventilation in patients with ARDS.

The OSCAR study

The Oscillation in ARDS (OSCAR) study²¹ was a “pragmatic” trial²² (ie, it had minimal exclusion criteria) of the safety and effectiveness of HFOV as a primary ventilatory strategy for ARDS. It included 795 patients randomized to receive conventional ventilation (n = 397) or HFOV (n = 398). Research centers followed detailed algorithms for HFOV management and adopted their usual practice for conventional ventilation. Medical care was given according to the clinician’s judgment.

The primary outcome studied was survival at 30 days. The secondary outcomes were all-cause mortality in the intensive care unit and the hospital, duration of mechanical ventilation, and use of antimicrobial, sedative, vasoactive, and neuromuscular-blocking drugs.

Findings. The patient baseline characteristics were similar in both groups.

There was no significant difference in intensive care unit mortality rates, hospital mortality rates, or mortality rates at 30 days (41.7% in the HFOV group vs 41.1% in the conventional ventilation group; P = .85, 95% CI 6.1–7.5) even after adjustments for center or severity of illness.

The duration of mechanical ventilation was similar in both groups (14.9 ± 13.3 days in the HFOV group vs 14.1 ± 13.4 days in the conventional ventilation group, P = .41). However, sedatives and neuromuscular-blocking drugs were used more often and longer in the HFOV group than in the conventional ventilation group. There was no difference in the use of vasoactive or antimicrobial medications.

Conclusions. This multicenter randomized control trial did not demonstrate any benefit from using HFOV for routine management of ARDS. Its pragmatic design made it less likely to reach a firm conclusion,²² but it at least made a case against routinely using HFOV in patients with ARDS.

The OSCILLATE study

The Oscillation for Acute Respiratory Distress Syndrome Treated Early (OSCILLATE) study²³ assessed the safety and efficacy of HFOV as a treatment for early-onset moderate-to-severe ARDS.

The inclusion criteria were similar to those in the OSCAR trial except that pulmonary symptoms had to be present less than 2 weeks and ARDS assessment was done under standard ventilator settings. As this was an efficacy trial, it had more exclusion criteria than the OSCAR trial. A total of 548 patients were randomized to receive conventional ventilation (n = 273) or HFOV (n = 275). The baseline characteristics were similar between groups.

Conventional ventilation was given according to a protocol used in an earlier trial² and included recruitment maneuvers. HFOV was given in centers that had experience in this treatment, and there were protocols for ventilation management, hemodynamic optimization, and weaning. All other care was left to the clinician’s choice.

The primary outcome studied was in-hospital mortality. The investigators also evaluated whether there were interactions between the treatment and baseline severity of lung injury and center experience with HFOV.

Findings. The trial was stopped after an interim analysis found that HFOV might be harmful, although the statistical threshold for stopping was not reached. The in-hospital mortality rate was 47% in the HFOV group and 35% in the control group (relative risk of death with HFOV 1.33, 95% CI 1.09–1.64, P = .005). HFOV was worse than conventional ventilation regardless of the severity of disease or center experience. The HFOV group had higher mean airway pressures but similar FiO₂ compared with the conventional ventilation group.

The HFOV group received significantly more vasopressors, sedatives, and neuromuscular blockers. This group’s fluid balance was higher as well, but not significantly so. Refractory hypoxemia (defined as PaO₂ < 60 mm Hg for 1 hour with an FiO₂ of 1.0 and neuromuscular blockade) was more frequent in the conventional ventilation group, but the number of deaths in the subgroup with refractory hypoxemia was similar with either treatment.

Conclusions. This multicenter randomized controlled trial demonstrated that HFOV was harmful when used routinely to manage ARDS. The trial’s protocol was based on the results of a pilot study carried out by the same investigators, which provided the best evidence available regarding the safety of HFOV at that time.

The results of the OSCAR and OSCILLATE trials have quelled enthusiasm for early, routine use of HFOV in ARDS. Although there are concerns that the protocol (ie, the way HFOV was implemented) rather than HFOV itself may have led to worse outcomes, there is no signal to support its routine use. We need further studies to define if it remains a viable rescue therapy.

Extracorporeal membrane oxygenation: Is it a viable option in severe ARDS?

Extracorporeal membrane oxygenation (ECMO) uses cardiopulmonary bypass technology to provide gas exchange. In patients with severe hypoxemia, ECMO can ensure adequate oxygenation and ventilation while ensuring the optimization of lung-protective ventilation. But ECMO was never as successful in adults with ARDS as it was in children and neonates.²⁴

The first two trials of ECMO in ARDS^24,25 reported equal or worse survival rates compared with conventional ventilation, and the overall mortality rate in these studies was staggeringly high. However, these studies were carried out before the era of lung-protective ventilation and at a time when ECMO technology was relatively primitive.

With new technology such as venovenous circuits and smaller cannulas, ECMO has gained more acceptance. It was used in patients with severe or refractory hypoxemia associated with ARDS during the H1N1 pandemic.^26,27

The CESAR trial

The Conventional Ventilatory Support Versus Extracorporeal Membrane Oxygenation for Severe Adult Respiratory Failure (CESAR) trial²⁸ assessed the safety, clinical efficacy, and cost-effectiveness of ECMO in managing severe ARDS. It compared best standard practice vs a protocol that included ECMO. The trial was conducted from 2001 to 2006.

Patients with severe ARDS, as defined by a Murray score²⁹ greater than 3 or uncompensated hypercapnea, were prospectively randomized and recruited from an ECMO center and 148 tertiary intensive care units and referral hospitals in England. This was a pragmatic trial, with minimal exclusion criteria (essentially, mechanical ventilation with high pressures and high FiO₂ for more than 7 days, intracranial bleeding, or contraindication to heparinization).

A total of 180 patients were randomized in a one-to-one ratio to receive ECMO or conventional management. The ventilator management in the conventional treatment group was not done according to a protocol but in general was low-volume and low-pressure. All patients randomized to ECMO were transferred to the ECMO center and treated according to a standardized ventilation protocol. After 12 hours, if predefined goals were not reached, venovenous ECMO was started. Patients assigned to conventional management could not cross over to ECMO.

The primary outcomes were death or severe disability at 6 months after randomization, and cost-effectiveness. The secondary outcomes were hospital resource use (eg, rescue techniques, length of stay, duration of ECMO) and health status after 6 months.

Findings. The groups were similar at baseline. Sixty-eight (75%) of the 90 patients randomized to receive ECMO actually received it. Of the 22 patients who did not receive ECMO, 16 (18% of the 90) improved on conventional therapy, 5 (6%) died during or before transfer, and 1 could not receive heparin.

Two patients had severe complications in the ECMO group: one had an arterial puncture, and one had an oxygen delivery failure during transport. In each case, these events contributed to the death of the patient.

More patients in the ECMO group received lung-protective ventilation, 84 (93%) vs 63 (70%).

The primary outcome, ie, death or severe disability at 6 months, occurred in 33 (37%) of the 90 patients in the ECMO group and in 46 (53%) of the patients in the conventional management group (relative risk 0.69, 95% CI 0.05–0.97, P = .03). More patients in the ECMO group survived, but the difference was not statistically significant (relative risk of death 0.73, 95% CI 0.52–1.03, P = .07). The most common cause of death in the ECMO group was multiorgan failure (42%), whereas in the conventional management group, the most common cause of death was respiratory failure (60%).

Length of stay in the hospital and in the critical care unit and health care costs were double for patients in the ECMO group. There was no difference in quality-of-life markers at 6 months in the survivors.

Conclusions. This pragmatic trial demonstrated that a protocol that includes ECMO could improve survival rates in ARDS.

Of note, the ECMO group got care in regional centers that used protocols. Therefore, in interpreting the results of this trial, we have to consider that being in a center with protocol-specified care for ARDS could drive some of the difference in mortality rates.

Regardless, this trial demonstrated that ECMO is feasible and led to better outcomes than expected. The findings were encouraging, and spurred the use of ECMO in severe ARDS during the 2009 H1N1 pandemic. Two propensity-matched studies and a number of case series reported a survival benefit associated with the use of ECMO in patients with severe ARDS.^27,30

A recent meta-analysis also reported that ECMO might lower the mortality rate in ARDS; however, the patients in the H1N1 pandemic were younger and usually had isolated respiratory failure.³¹

The success of ECMO has opened new possibilities in the management of ARDS. As the technology improves and our experience increases, ECMO will likely gain more acceptance as a treatment for severe ARDS.

Airway pressure release ventilation

The use of airway pressure release ventilation and other ventilator modalities in ARDS is not supported by current evidence, though results of clinical trials may influence our practice in the future.

PHARMACOTHERAPY IN ARDS

The pathogenesis of ARDS includes damage to the alveolar-capillary membrane, with leakage of protein-rich edema fluid into alveoli. This damage is propagated by a complex inflammatory response including but not limited to neutrophil activation, free-radical formation, dysregulation of the coagulation system, and extensive release of inflammatory mediators.^32,33 As a consequence, there are multiple potential targets for pharmacologic therapy in ARDS.

A variety of drugs, including corticosteroids, anti-inflammatory agents, immune-modulating agents, pulmonary vasodilators, antioxidants, and surfactants, have been studied in patients with ARDS.³⁴ But effective pharmacotherapy for ARDS remains extremely limited.

Neuromuscular blockade in early severe ARDS

Mechanical ventilation can result in injurious stretching of the lung parenchyma, either from alveolar overdistention (volutrauma) or from continual recruitment and derecruitment of unstable lung units during the ventilator cycle (atelectrauma).³⁵ Ventilator-induced lung injury can be exacerbated by asynchronous breathing.

In theory, neuromuscular blockers could minimize patient-ventilator asynchrony and provide much better control of tidal volume and pressure in patients with ARDS. This may result in less volutrauma and atelectrauma associated with asynchronous breathing. Data also suggest that cisatracurium (Nimbex), a neuromuscular blocking agent, may have a direct effect on the amount of inflammation in lungs with ARDS.³⁶

The ACURASYS study

The ARDS et Curarisation Systématique (ACURASYS) study³⁷ was a randomized trial in 340 patients undergoing mechanical ventilation for severe ARDS to evaluate the impact of neuromuscular blockade within the first 48 hours in this population.

The primary outcome was the mortality rate before hospital discharge or within 90 days of study entry. Secondary outcomes included the 28-day mortality rate, the rate of intensive care unit-acquired paresis, and the number of ventilator-free days. To be included, patients had to have been mechanically ventilated for less than 48 hours and to meet the AECC criteria for severe ARDS, with a PaO₂/FiO₂ ratio less than 150 mm Hg.

The intervention group received a continuous infusion of cisatracurium for 48 hours, while the control patients received placebo. Muscle strength was evaluated by clinical scoring of strength in different muscle groups.

Findings. The study groups were similar at baseline.

The crude 90-day mortality rate was lower in the cisatracurium group (31.6% vs 40.7%, P = .08). Regression analysis showed an improved 90-day survival rate with the use of this neuromuscular blocker after adjustment for severity of illness and the severity of ARDS (based on degree of hypoxemia and plateau pressures) (hazard ratio for death at 90 days 0.68; 95% CI 0.48–0.98; P = .04). The rate of paresis acquired in the intensive care unit did not differ significantly between the two groups.

Conclusion. In patients with severe ARDS, giving a neuromuscular blocking agent early improved the survival rate and increased the time off the ventilator without increasing muscle weakness.

These data are in line with similar findings from two other studies published by the same group.^38,39 A meta-analysis of 432 patients showed that the use of neuromuscular blockade in early severe ARDS is associated with a statistically significant effect on early mortality (relative risk 0.66, 95% CI 0.50–0.87).⁴⁰ The pooled analysis of these trials did not show any statistically significant critical-illness polyneuropathy.

These results need to be interpreted carefully, as we have inadequate data to see if they generalize to different intensive care units, and the evaluation and categorization of critical-illness polyneuropathy remains to be defined.

Cisatracurium is a promising treatment for moderate to severe ARDS and merits investigation in a large confirmatory randomized controlled trial.

Other pharmacologic agents

A number of other drugs have been studied in ARDS patients, including both inhaled and intravenous beta agonists,^41,42 statins,⁴³ and nutritional supplements.⁴⁴ But as with other drugs previously studied in ARDS such as corticosteroids, N-acetylcysteine, and surfactant,³⁴ these agents showed no effect on outcomes. In fact, a recent trial of intravenous salbutamol in ARDS patients was stopped after an interim analysis because of a higher incidence of arrhythmias and lactic acidosis with this agent.⁴²

These findings reaffirm that pharmacologic therapy needs to be carefully considered, and potential harms associated with these therapies need to be addressed before they are introduced in the care of critically ill patients.

A 61-year-old man with type 2 diabetes mellitus on glimepiride therapy presented with somnolence and slurred speech. His capillary glucose level was 17 mg/dL and his serum glucose level was 28 mg/dL. He was treated with intravenous dextrose, and his glucose level promptly returned to normal.

He had been adherent to his medication regimen and denied overmedicating or accidental overdosing. Over the past 7 months, he had noted redness on his palms, a rash on his legs, intermittent moderate to severe headaches, weight loss, and decreased appetite. In addition, his blood pressure had been labile, which his physicians had attributed to autonomic instability. He had continued on the same dose of glimepiride despite losing weight.

His history included multivessel coronary artery disease treated with angioplasty and placement of multiple coronary stents; ischemic cardiomyopathy with a left ventricular ejection fraction of 28%; implantation of a cardioverter-defibrillator for secondary prevention of ventricular arrhythmia; an ischemic stroke; and multiple sclerosis complicated by bilateral blindness, with optic nerve involvement and autonomic instability, present for over a year and manifested by labile blood pressure. He was a long-time tobacco user. His daily medications included ticagrelor 90 mg, aspirin 81 mg, metoprolol 50 mg, ramipril 10 mg, simvastatin 20 mg, glimepiride 2 mg, and esomeprazole 40 mg. He needed help taking his medications.

At the time of hospital admission, his heart rate was 69 beats per minute with a regular rhythm, blood pressure 115/73 mm Hg, respiratory rate 11 breaths per minute with an oxygen saturation of 99% on room air, and oral temperature 34.7°C (94.5°F). He appeared to be in no distress.

Cardiovascular examination revealed no murmurs or gallops; there was mild nonpitting edema of the lower extremities. Pulmonary, abdominal, and neurologic examinations were unrevealing except for bilateral blindness. Vascular examination revealed no bruits. Results of a complete blood cell count and metabolic panel were normal except for a hemoglobin level of 9.9 g/dL (reference range 13.5–17.5) and a platelet count of 477 × 10⁹/L (150–450).

Although he continued to receive the same medications he had been taking at home, his blood pressure fluctuated. On the second hospital day, it reached 186/135 mm Hg, at which time he also had palpitations, dyspnea, and crackles in the lower lobes of both lungs. Volume resuscitation on admission was suspected to have played a role, and he received furosemide, which improved his symptoms. But several hours later, his blood pressure rose again, and he became diaphoretic. Despite aggressive treatment with different antihypertensive agents, his blood pressure remained high and his symptoms persisted. Chest radiography showed no evidence of pulmonary edema. Because of his progressive dyspnea, the diagnosis of pulmonary embolism was entertained.

CAUSES OF RESISTANT HYPERTENSION

1. What could explain this patient’s high blood pressure?

• A drug effect
• Renovascular disease
• Excess circulating catecholamines
• Obstructive sleep apnea
• Primary aldosteronism

Sympathomimetic drugs such as epinephrine, norepinephrine, dopamine, and vasopressin, which are used when hemodynamic support is required, can raise both systolic and diastolic blood pressure. Nonsteroidal anti-inflammatory drugs and nasal decongestants are common culprits in the community. However, our patient was using none of these drugs.

Renovascular disease is one of many causes of resistant hypertension, accounting for 8% of all cases.^1,2 Despite fluctuations, the blood pressure often remains chronically elevated, its changes are less paroxysmal than in our patient, and a precipitating factor such as a dietary indiscretion is sometimes identified.¹

Excess circulating catecholamines can be a result of stress, exogenous administration, or endogenous oversecretion. Our patient’s clinical presentation is highly suspicious for a high-catecholamine state, and this should be further evaluated.

Obstructive sleep apnea is common in patients with resistant hypertension, with an estimated prevalence as high as 60% in this group.^3,4

Primary aldosteronism has an estimated prevalence of about 20% in patients evaluated for resistant hypertension.⁵

AN ADRENAL MASS IS INCIDENTALLY DISCOVERED

Computed tomographic angiography of the chest revealed no evidence of pulmonary emboli. There was mild dilation of the central pulmonary arteries and an incidental, incompletely imaged 4.7-by-3.4-cm mass of mixed attenuation in the right adrenal gland, with macroscopic fat within the lesion.

Computed tomography (CT) of the abdomen with dedicated cuts through the adrenal glands revealed a 4.7-cm heterogeneous right adrenal mass with a density of 34 Hounsfield units (HU). The left adrenal gland appeared diffusely enlarged without a discretely seen mass, consistent with hyperplasticity (Figure 1).

2. Based on the patient’s clinical presentation and findings on CT, what would be the most likely diagnosis for this incidentally found adrenal mass?

• Adrenocortical adenoma
• Adrenocortical carcinoma
• Metastatic mass
• Pheochromocytoma

Adrenocortical adenoma can present as a small homogeneous mass of variable size, with smooth margins, and rarely containing hemorrhagic tissue or calcifications. The typical density on nonenhanced CT is less than 10 HU. On enhanced CT, it is nonvascular. T2-weighted magnetic resonance imaging (MRI) shows a lesion of the same intensity as liver tissue.⁶

Adrenocortical adenoma is not classically associated with autologous activity and thus is less likely to explain our patient’s symptoms.

Adrenocortical carcinoma can present as a large heterogeneous mass, usually greater than 4 cm in diameter, with irregular margins and areas of necrosis, hemorrhage, or calcification. The typical density on nonenhanced CT is greater than 10 HU. On enhanced CT, the mass is usually vascular, and T2-weighted MRI will show a lesion more intense than liver tissue.⁶

Adrenocortical carcinoma is also not classically associated with autologous activity, and so is not likely to explain our patient’s symptoms.⁶

Metastatic disease can present with masses of variable size, often bilaterally, and occasionally with cysts or areas of hemorrhage. The typical density of metastatic lesions on nonenhanced CT is greater than 10 HU. On enhanced CT, they are usually vascular, and on T2-weighted MRI they are hyperintense.⁶ The characteristics of the mass and the absence of a primary malignancy on CT of the chest and abdomen do not support the diagnosis of metastatic disease.

Pheochromocytoma is a neuroendocrine tumor of the adrenal medulla that can present as a large heterogeneous mass, greater than 3 cm in diameter, with clear margins and cysts or areas of hemorrhage. Extra-adrenal neuroendocrine tumors are typically called paragangliomas and have features similar to those of pheochromocytoma. The typical density of pheochromocytoma on nonenhanced CT is greater than 10 HU. On enhanced CT, it is usually vascular, and T2-weighted MRI shows a hyperintense lesion. Pheochromocytoma can be biochemically active and thus can cause signs and symptoms that will lead to the diagnosis.⁶

Other imaging tests may play a role in the evaluation of adrenal masses but are not required for the diagnosis of pheochromocytoma. Functional positron emission tomography using metaiodobenzylguanidine labeled with iodine 123 or-iodine 131 or using the glucose analogue F-18 fluorodeoxyglucose has been used in the initial assessment of pheochromocytoma, with good sensitivity and specificity.^7,8

Our patient’s pacemaker-defibrillator precluded him from undergoing MRI.

DIAGNOSIS: PHEOCHROMOCYTOMA

Pheochromocytoma was highly suspected on the basis of the patient’s clinical presentation, and metoprolol was immediately discontinued. He was started on the calcium channel blocker verapamil and the alpha-blocker phenoxybenzamine.

Serum samples were obtained to measure metanephrines, dehydroepiandrosterone, aldosterone, and cortisol, and a 24-hour urine collection was obtained to measure creatinine, dopamine, epinephrine, norepinephrine, cortisol, and metanephrines. Based on the results (Table 1) and on the findings on imaging, the patient was diagnosed with pheochromocytoma. A surgical consultation was obtained, and surgery was recommended.

WHEN DOES PHEOCHROMOCYTOMA CALL FOR SURGERY?

3. Which criterion is most important when determining the need for surgery for pheochromocytoma?

• Findings on fine-needle aspiration biopsy
• Biochemical activity
• Size of the mass
• Bilateral masses

Fine-needle aspiration biopsy can be done when a mass is found incidentally and no evidence of biochemical activity is detected, although it is not an essential part of the diagnostic workup.⁹ In most cases, the sampling from fine-needle aspiration is not sufficient to achieve a diagnosis.

Biochemical activity is the most important factor when determining the need for prompt surgical intervention. The excess circulating catecholamines have been associated with increased risk of cardiovascular morbidity and death independent of the morbidity associated with hypertension alone.¹⁰ Biochemical activity can be independent of the size of the mass, but larger masses typically present with symptoms.

Bilateral masses have been associated with metastatic disease.¹¹ In retrospect, the patient’s history of hypertension and cerebrovascular accident could be associated with the development of a catecholamine-releasing tumor.

A GOOD OUTCOME FROM SURGERY

The patient was continued on phenoxybenzamine for 7 days and responded well to this therapy.

After this preoperative preparation, he underwent laparoscopic right adrenalectomy with excision of a retroperitoneal adrenal mass. His postoperative course was complicated by transient hypotension requiring low-dose vasopressin support for less than 24 hours. He was then restarted on his previous dosage of metoprolol and was discharged home on postoperative day 5 with stable blood pressure. Follow-up 24-hour urine collection 1 month after he was discharged showed normalization of metanephrine, normetanephrine, epinephrine, and norepinephrine levels.

Despite low suspicion for an underlying genetic syndrome, he was referred for genetic testing and was scheduled to have a repeat 24-hour urine collection and imaging in 6 months to follow his enlarged left adrenal gland, which did not appear to be metabolically hyperactive.

4. What is the most common perioperative complication of pheochromocytoma excision?

• Hypoglycemia
• Hypotension
• Hypocortisolism
• Hypertension
• Tachycardia

Hypoglycemia has been observed after removal of pheochromocytoma, as levels of catecholamines (which normally inhibit pancreatic beta cells) decrease and insulin secretion consequently increases.¹² Our patient developed hypoglycemia before surgery, not after, and it was likely due to the combination of his antidiabetic therapy, weight loss, and decreased oral intake.

Hypotension is the most common complication in the perioperative period. It is associated with excessive loss of catecholamine secretion. It is usually short-lived but may require aggressive administration of intravenous fluids and use of sympathomimetic agents.

Hypocortisolism is unlikely in patients with pheochromocytoma, but it is likely after removal of adrenocortical adenoma.

Hypertension and tachycardia affect up to 40% of pheochromocytoma patients in some case series.¹²

PHEOCHROMOCYTOMA: A CATECHOLAMINE-SECRETING TUMOR

The pathophysiology of pheochromocytoma is complex. It is characterized by accelerated growth of cells producing catecholamines, which may produce symptoms when secreted into the bloodstream. The classic triad of symptoms is headache, hypertension, and hyperglycemia, although our patient had very low blood sugar levels. Other common symptoms are nausea, orthostasis, and tremor, although not all symptoms are invariably seen.

Genetic testing recommended

Genetic associations have been described and are thought to be responsible for 20% to 30% of cases of pheochromocytoma. All associated germline mutations are autosomal dominant, some with variable penetrance. These include:

• Succinate dehydrogenase subunit B, C, and D mutations
• von Hippel-Lindau syndrome
• Multiple endocrine neoplasia type 1 and type 2 syndromes
• Neurofibromatosis type 1.^13,14

The succinate dehydrogenase subunit mutations have been associated with, but not limited to, extra-adrenal adenomas (paragangliomas) and carry a worse prognosis.

Some experts recommend genetic testing in all cases of pheochromocytoma, sporadic or familial, and this testing should be followed by counseling if a mutation is found.¹⁵ Others recommend genetic testing based on the patient’s age (under age 50), history, imaging, and biochemical features of the tumor (metanephrines predominate in multiple endocrine neoplasia syndromes, and normetanephrines in von Hippel-Lindau syndrome).¹³

Serious consequences

A thorough evaluation is recommended, since pheochromocytoma has been associated with increased cardiovascular morbidity. In a retrospective series, Stolk et al¹⁰ reported that patients with pheochromocytoma had a higher incidence of myocardial infarction, angina, and stroke in the years preceding the diagnosis than did patients with essential hypertension (13.8% vs 1.1%, P < .001).¹⁰

Catecholamine cardiomyopathy has been described and shares clinical features with Takotsubo or stress cardiomyopathy, with global left ventricular systolic and diastolic dysfunction that improve or resolve after the adrenergic insult is removed.¹⁶

Conditions that warrant further evaluation or that may suggest pheochromocytoma are malignant hypertension, hypertensive encephalopathy, ischemic stroke, subarachnoid hemorrhage, acute pulmonary edema, angina pectoris, myocardial infarction, aortic dissection, and kidney injury.

When to suspect pheochromocytoma

Pheochromocytoma should be suspected in a patient with resistant hypertension, family history, or imaging findings that suggest an adrenal mass with a heterogeneous appearance. The diagnostic algorithm follows the same pathway as for the evaluation of an incidentally found adrenal mass, with determination of its dimension and characteristics by CT or MRI, and with biochemical testing of urine catecholamines, plasma free metanephrines, renin, aldosterone, and cortisol.

The diagnosis of pheochromocytoma is established by obtaining fractionated metanephrines and catecholamines in a 24-hour urine collection (sensitivity 90%, specificity 98%). Analysis of plasma metanephrines has a higher sensitivity (97%) but lower specificity (85%).¹⁷ The combination of typical signs, symptoms, and laboratory findings makes the diagnosis likely, especially in combination with a unilateral adrenal mass.

Laparoscopic surgery after medical preparation for active tumors

If the mass appears benign and not biochemically hyperactive, then follow-up at 1 year is recommended, with repeat testing. Surgical evaluation and intervention is recommended for lesions that appear malignant or that are biochemically active and clinically symptomatic.⁹

Preoperative hemodynamic control is essential in the management of pheochromocytoma to prevent or minimize hemodynamic changes that can be driven by increased catecholamines. Control is typically achieved with initial alpha-blockade and then beta-blockade to avoid worsening hypertension and to prevent an acute hypertensive crisis during surgical intervention. Phenoxybenzamine, the mainstay of therapy, is a nonselective alpha-blocker with a long duration of action that requires titration over several days up to 3 weeks.

A selective alpha-1-blocker such as doxazosin can be used to control postoperative hypotension, as it has a shorter half-life than phenoxybenzamine. Alternative strategies include calcium channel blockers, centrally acting sympathetic blockers, and magnesium.¹⁸

Laparoscopic adrenalectomy by an experienced surgeon after excellent medical preparation is often considered the treatment of choice, but for larger or malignant masses, an open procedure is recommended. The risk of perioperative morbidity and death can be reduced by adequate medical management. With successful surgical resection, the long-term prognosis is favorable.

A 61-year-old man with type 2 diabetes mellitus on glimepiride therapy presented with somnolence and slurred speech. His capillary glucose level was 17 mg/dL and his serum glucose level was 28 mg/dL. He was treated with intravenous dextrose, and his glucose level promptly returned to normal.

He had been adherent to his medication regimen and denied overmedicating or accidental overdosing. Over the past 7 months, he had noted redness on his palms, a rash on his legs, intermittent moderate to severe headaches, weight loss, and decreased appetite. In addition, his blood pressure had been labile, which his physicians had attributed to autonomic instability. He had continued on the same dose of glimepiride despite losing weight.

His history included multivessel coronary artery disease treated with angioplasty and placement of multiple coronary stents; ischemic cardiomyopathy with a left ventricular ejection fraction of 28%; implantation of a cardioverter-defibrillator for secondary prevention of ventricular arrhythmia; an ischemic stroke; and multiple sclerosis complicated by bilateral blindness, with optic nerve involvement and autonomic instability, present for over a year and manifested by labile blood pressure. He was a long-time tobacco user. His daily medications included ticagrelor 90 mg, aspirin 81 mg, metoprolol 50 mg, ramipril 10 mg, simvastatin 20 mg, glimepiride 2 mg, and esomeprazole 40 mg. He needed help taking his medications.

At the time of hospital admission, his heart rate was 69 beats per minute with a regular rhythm, blood pressure 115/73 mm Hg, respiratory rate 11 breaths per minute with an oxygen saturation of 99% on room air, and oral temperature 34.7°C (94.5°F). He appeared to be in no distress.

Cardiovascular examination revealed no murmurs or gallops; there was mild nonpitting edema of the lower extremities. Pulmonary, abdominal, and neurologic examinations were unrevealing except for bilateral blindness. Vascular examination revealed no bruits. Results of a complete blood cell count and metabolic panel were normal except for a hemoglobin level of 9.9 g/dL (reference range 13.5–17.5) and a platelet count of 477 × 10⁹/L (150–450).

Although he continued to receive the same medications he had been taking at home, his blood pressure fluctuated. On the second hospital day, it reached 186/135 mm Hg, at which time he also had palpitations, dyspnea, and crackles in the lower lobes of both lungs. Volume resuscitation on admission was suspected to have played a role, and he received furosemide, which improved his symptoms. But several hours later, his blood pressure rose again, and he became diaphoretic. Despite aggressive treatment with different antihypertensive agents, his blood pressure remained high and his symptoms persisted. Chest radiography showed no evidence of pulmonary edema. Because of his progressive dyspnea, the diagnosis of pulmonary embolism was entertained.

CAUSES OF RESISTANT HYPERTENSION

1. What could explain this patient’s high blood pressure?

• A drug effect
• Renovascular disease
• Excess circulating catecholamines
• Obstructive sleep apnea
• Primary aldosteronism

Sympathomimetic drugs such as epinephrine, norepinephrine, dopamine, and vasopressin, which are used when hemodynamic support is required, can raise both systolic and diastolic blood pressure. Nonsteroidal anti-inflammatory drugs and nasal decongestants are common culprits in the community. However, our patient was using none of these drugs.

Renovascular disease is one of many causes of resistant hypertension, accounting for 8% of all cases.^1,2 Despite fluctuations, the blood pressure often remains chronically elevated, its changes are less paroxysmal than in our patient, and a precipitating factor such as a dietary indiscretion is sometimes identified.¹

Excess circulating catecholamines can be a result of stress, exogenous administration, or endogenous oversecretion. Our patient’s clinical presentation is highly suspicious for a high-catecholamine state, and this should be further evaluated.

Obstructive sleep apnea is common in patients with resistant hypertension, with an estimated prevalence as high as 60% in this group.^3,4

Primary aldosteronism has an estimated prevalence of about 20% in patients evaluated for resistant hypertension.⁵

AN ADRENAL MASS IS INCIDENTALLY DISCOVERED

Computed tomographic angiography of the chest revealed no evidence of pulmonary emboli. There was mild dilation of the central pulmonary arteries and an incidental, incompletely imaged 4.7-by-3.4-cm mass of mixed attenuation in the right adrenal gland, with macroscopic fat within the lesion.

Computed tomography (CT) of the abdomen with dedicated cuts through the adrenal glands revealed a 4.7-cm heterogeneous right adrenal mass with a density of 34 Hounsfield units (HU). The left adrenal gland appeared diffusely enlarged without a discretely seen mass, consistent with hyperplasticity (Figure 1).

2. Based on the patient’s clinical presentation and findings on CT, what would be the most likely diagnosis for this incidentally found adrenal mass?

• Adrenocortical adenoma
• Adrenocortical carcinoma
• Metastatic mass
• Pheochromocytoma

Adrenocortical adenoma can present as a small homogeneous mass of variable size, with smooth margins, and rarely containing hemorrhagic tissue or calcifications. The typical density on nonenhanced CT is less than 10 HU. On enhanced CT, it is nonvascular. T2-weighted magnetic resonance imaging (MRI) shows a lesion of the same intensity as liver tissue.⁶

Adrenocortical adenoma is not classically associated with autologous activity and thus is less likely to explain our patient’s symptoms.

Adrenocortical carcinoma can present as a large heterogeneous mass, usually greater than 4 cm in diameter, with irregular margins and areas of necrosis, hemorrhage, or calcification. The typical density on nonenhanced CT is greater than 10 HU. On enhanced CT, the mass is usually vascular, and T2-weighted MRI will show a lesion more intense than liver tissue.⁶

Adrenocortical carcinoma is also not classically associated with autologous activity, and so is not likely to explain our patient’s symptoms.⁶

Metastatic disease can present with masses of variable size, often bilaterally, and occasionally with cysts or areas of hemorrhage. The typical density of metastatic lesions on nonenhanced CT is greater than 10 HU. On enhanced CT, they are usually vascular, and on T2-weighted MRI they are hyperintense.⁶ The characteristics of the mass and the absence of a primary malignancy on CT of the chest and abdomen do not support the diagnosis of metastatic disease.

Pheochromocytoma is a neuroendocrine tumor of the adrenal medulla that can present as a large heterogeneous mass, greater than 3 cm in diameter, with clear margins and cysts or areas of hemorrhage. Extra-adrenal neuroendocrine tumors are typically called paragangliomas and have features similar to those of pheochromocytoma. The typical density of pheochromocytoma on nonenhanced CT is greater than 10 HU. On enhanced CT, it is usually vascular, and T2-weighted MRI shows a hyperintense lesion. Pheochromocytoma can be biochemically active and thus can cause signs and symptoms that will lead to the diagnosis.⁶

Other imaging tests may play a role in the evaluation of adrenal masses but are not required for the diagnosis of pheochromocytoma. Functional positron emission tomography using metaiodobenzylguanidine labeled with iodine 123 or-iodine 131 or using the glucose analogue F-18 fluorodeoxyglucose has been used in the initial assessment of pheochromocytoma, with good sensitivity and specificity.^7,8

Our patient’s pacemaker-defibrillator precluded him from undergoing MRI.

DIAGNOSIS: PHEOCHROMOCYTOMA

Pheochromocytoma was highly suspected on the basis of the patient’s clinical presentation, and metoprolol was immediately discontinued. He was started on the calcium channel blocker verapamil and the alpha-blocker phenoxybenzamine.

Serum samples were obtained to measure metanephrines, dehydroepiandrosterone, aldosterone, and cortisol, and a 24-hour urine collection was obtained to measure creatinine, dopamine, epinephrine, norepinephrine, cortisol, and metanephrines. Based on the results (Table 1) and on the findings on imaging, the patient was diagnosed with pheochromocytoma. A surgical consultation was obtained, and surgery was recommended.

WHEN DOES PHEOCHROMOCYTOMA CALL FOR SURGERY?

3. Which criterion is most important when determining the need for surgery for pheochromocytoma?

• Findings on fine-needle aspiration biopsy
• Biochemical activity
• Size of the mass
• Bilateral masses

Fine-needle aspiration biopsy can be done when a mass is found incidentally and no evidence of biochemical activity is detected, although it is not an essential part of the diagnostic workup.⁹ In most cases, the sampling from fine-needle aspiration is not sufficient to achieve a diagnosis.

Biochemical activity is the most important factor when determining the need for prompt surgical intervention. The excess circulating catecholamines have been associated with increased risk of cardiovascular morbidity and death independent of the morbidity associated with hypertension alone.¹⁰ Biochemical activity can be independent of the size of the mass, but larger masses typically present with symptoms.

Bilateral masses have been associated with metastatic disease.¹¹ In retrospect, the patient’s history of hypertension and cerebrovascular accident could be associated with the development of a catecholamine-releasing tumor.

A GOOD OUTCOME FROM SURGERY

The patient was continued on phenoxybenzamine for 7 days and responded well to this therapy.

After this preoperative preparation, he underwent laparoscopic right adrenalectomy with excision of a retroperitoneal adrenal mass. His postoperative course was complicated by transient hypotension requiring low-dose vasopressin support for less than 24 hours. He was then restarted on his previous dosage of metoprolol and was discharged home on postoperative day 5 with stable blood pressure. Follow-up 24-hour urine collection 1 month after he was discharged showed normalization of metanephrine, normetanephrine, epinephrine, and norepinephrine levels.

Despite low suspicion for an underlying genetic syndrome, he was referred for genetic testing and was scheduled to have a repeat 24-hour urine collection and imaging in 6 months to follow his enlarged left adrenal gland, which did not appear to be metabolically hyperactive.

4. What is the most common perioperative complication of pheochromocytoma excision?

• Hypoglycemia
• Hypotension
• Hypocortisolism
• Hypertension
• Tachycardia

Hypoglycemia has been observed after removal of pheochromocytoma, as levels of catecholamines (which normally inhibit pancreatic beta cells) decrease and insulin secretion consequently increases.¹² Our patient developed hypoglycemia before surgery, not after, and it was likely due to the combination of his antidiabetic therapy, weight loss, and decreased oral intake.

Hypotension is the most common complication in the perioperative period. It is associated with excessive loss of catecholamine secretion. It is usually short-lived but may require aggressive administration of intravenous fluids and use of sympathomimetic agents.

Hypocortisolism is unlikely in patients with pheochromocytoma, but it is likely after removal of adrenocortical adenoma.

Hypertension and tachycardia affect up to 40% of pheochromocytoma patients in some case series.¹²

PHEOCHROMOCYTOMA: A CATECHOLAMINE-SECRETING TUMOR

The pathophysiology of pheochromocytoma is complex. It is characterized by accelerated growth of cells producing catecholamines, which may produce symptoms when secreted into the bloodstream. The classic triad of symptoms is headache, hypertension, and hyperglycemia, although our patient had very low blood sugar levels. Other common symptoms are nausea, orthostasis, and tremor, although not all symptoms are invariably seen.

Genetic testing recommended

Genetic associations have been described and are thought to be responsible for 20% to 30% of cases of pheochromocytoma. All associated germline mutations are autosomal dominant, some with variable penetrance. These include:

• Succinate dehydrogenase subunit B, C, and D mutations
• von Hippel-Lindau syndrome
• Multiple endocrine neoplasia type 1 and type 2 syndromes
• Neurofibromatosis type 1.^13,14

The succinate dehydrogenase subunit mutations have been associated with, but not limited to, extra-adrenal adenomas (paragangliomas) and carry a worse prognosis.

Some experts recommend genetic testing in all cases of pheochromocytoma, sporadic or familial, and this testing should be followed by counseling if a mutation is found.¹⁵ Others recommend genetic testing based on the patient’s age (under age 50), history, imaging, and biochemical features of the tumor (metanephrines predominate in multiple endocrine neoplasia syndromes, and normetanephrines in von Hippel-Lindau syndrome).¹³

Serious consequences

A thorough evaluation is recommended, since pheochromocytoma has been associated with increased cardiovascular morbidity. In a retrospective series, Stolk et al¹⁰ reported that patients with pheochromocytoma had a higher incidence of myocardial infarction, angina, and stroke in the years preceding the diagnosis than did patients with essential hypertension (13.8% vs 1.1%, P < .001).¹⁰

Catecholamine cardiomyopathy has been described and shares clinical features with Takotsubo or stress cardiomyopathy, with global left ventricular systolic and diastolic dysfunction that improve or resolve after the adrenergic insult is removed.¹⁶

Conditions that warrant further evaluation or that may suggest pheochromocytoma are malignant hypertension, hypertensive encephalopathy, ischemic stroke, subarachnoid hemorrhage, acute pulmonary edema, angina pectoris, myocardial infarction, aortic dissection, and kidney injury.

When to suspect pheochromocytoma

Pheochromocytoma should be suspected in a patient with resistant hypertension, family history, or imaging findings that suggest an adrenal mass with a heterogeneous appearance. The diagnostic algorithm follows the same pathway as for the evaluation of an incidentally found adrenal mass, with determination of its dimension and characteristics by CT or MRI, and with biochemical testing of urine catecholamines, plasma free metanephrines, renin, aldosterone, and cortisol.

The diagnosis of pheochromocytoma is established by obtaining fractionated metanephrines and catecholamines in a 24-hour urine collection (sensitivity 90%, specificity 98%). Analysis of plasma metanephrines has a higher sensitivity (97%) but lower specificity (85%).¹⁷ The combination of typical signs, symptoms, and laboratory findings makes the diagnosis likely, especially in combination with a unilateral adrenal mass.

Laparoscopic surgery after medical preparation for active tumors

If the mass appears benign and not biochemically hyperactive, then follow-up at 1 year is recommended, with repeat testing. Surgical evaluation and intervention is recommended for lesions that appear malignant or that are biochemically active and clinically symptomatic.⁹

Preoperative hemodynamic control is essential in the management of pheochromocytoma to prevent or minimize hemodynamic changes that can be driven by increased catecholamines. Control is typically achieved with initial alpha-blockade and then beta-blockade to avoid worsening hypertension and to prevent an acute hypertensive crisis during surgical intervention. Phenoxybenzamine, the mainstay of therapy, is a nonselective alpha-blocker with a long duration of action that requires titration over several days up to 3 weeks.

A selective alpha-1-blocker such as doxazosin can be used to control postoperative hypotension, as it has a shorter half-life than phenoxybenzamine. Alternative strategies include calcium channel blockers, centrally acting sympathetic blockers, and magnesium.¹⁸

Laparoscopic adrenalectomy by an experienced surgeon after excellent medical preparation is often considered the treatment of choice, but for larger or malignant masses, an open procedure is recommended. The risk of perioperative morbidity and death can be reduced by adequate medical management. With successful surgical resection, the long-term prognosis is favorable.

A 61-year-old man with type 2 diabetes mellitus on glimepiride therapy presented with somnolence and slurred speech. His capillary glucose level was 17 mg/dL and his serum glucose level was 28 mg/dL. He was treated with intravenous dextrose, and his glucose level promptly returned to normal.

He had been adherent to his medication regimen and denied overmedicating or accidental overdosing. Over the past 7 months, he had noted redness on his palms, a rash on his legs, intermittent moderate to severe headaches, weight loss, and decreased appetite. In addition, his blood pressure had been labile, which his physicians had attributed to autonomic instability. He had continued on the same dose of glimepiride despite losing weight.

His history included multivessel coronary artery disease treated with angioplasty and placement of multiple coronary stents; ischemic cardiomyopathy with a left ventricular ejection fraction of 28%; implantation of a cardioverter-defibrillator for secondary prevention of ventricular arrhythmia; an ischemic stroke; and multiple sclerosis complicated by bilateral blindness, with optic nerve involvement and autonomic instability, present for over a year and manifested by labile blood pressure. He was a long-time tobacco user. His daily medications included ticagrelor 90 mg, aspirin 81 mg, metoprolol 50 mg, ramipril 10 mg, simvastatin 20 mg, glimepiride 2 mg, and esomeprazole 40 mg. He needed help taking his medications.

At the time of hospital admission, his heart rate was 69 beats per minute with a regular rhythm, blood pressure 115/73 mm Hg, respiratory rate 11 breaths per minute with an oxygen saturation of 99% on room air, and oral temperature 34.7°C (94.5°F). He appeared to be in no distress.

Cardiovascular examination revealed no murmurs or gallops; there was mild nonpitting edema of the lower extremities. Pulmonary, abdominal, and neurologic examinations were unrevealing except for bilateral blindness. Vascular examination revealed no bruits. Results of a complete blood cell count and metabolic panel were normal except for a hemoglobin level of 9.9 g/dL (reference range 13.5–17.5) and a platelet count of 477 × 10⁹/L (150–450).

Although he continued to receive the same medications he had been taking at home, his blood pressure fluctuated. On the second hospital day, it reached 186/135 mm Hg, at which time he also had palpitations, dyspnea, and crackles in the lower lobes of both lungs. Volume resuscitation on admission was suspected to have played a role, and he received furosemide, which improved his symptoms. But several hours later, his blood pressure rose again, and he became diaphoretic. Despite aggressive treatment with different antihypertensive agents, his blood pressure remained high and his symptoms persisted. Chest radiography showed no evidence of pulmonary edema. Because of his progressive dyspnea, the diagnosis of pulmonary embolism was entertained.

CAUSES OF RESISTANT HYPERTENSION

1. What could explain this patient’s high blood pressure?

• A drug effect
• Renovascular disease
• Excess circulating catecholamines
• Obstructive sleep apnea
• Primary aldosteronism

Sympathomimetic drugs such as epinephrine, norepinephrine, dopamine, and vasopressin, which are used when hemodynamic support is required, can raise both systolic and diastolic blood pressure. Nonsteroidal anti-inflammatory drugs and nasal decongestants are common culprits in the community. However, our patient was using none of these drugs.

Renovascular disease is one of many causes of resistant hypertension, accounting for 8% of all cases.^1,2 Despite fluctuations, the blood pressure often remains chronically elevated, its changes are less paroxysmal than in our patient, and a precipitating factor such as a dietary indiscretion is sometimes identified.¹

Excess circulating catecholamines can be a result of stress, exogenous administration, or endogenous oversecretion. Our patient’s clinical presentation is highly suspicious for a high-catecholamine state, and this should be further evaluated.

Obstructive sleep apnea is common in patients with resistant hypertension, with an estimated prevalence as high as 60% in this group.^3,4

Primary aldosteronism has an estimated prevalence of about 20% in patients evaluated for resistant hypertension.⁵

AN ADRENAL MASS IS INCIDENTALLY DISCOVERED

Computed tomographic angiography of the chest revealed no evidence of pulmonary emboli. There was mild dilation of the central pulmonary arteries and an incidental, incompletely imaged 4.7-by-3.4-cm mass of mixed attenuation in the right adrenal gland, with macroscopic fat within the lesion.

Computed tomography (CT) of the abdomen with dedicated cuts through the adrenal glands revealed a 4.7-cm heterogeneous right adrenal mass with a density of 34 Hounsfield units (HU). The left adrenal gland appeared diffusely enlarged without a discretely seen mass, consistent with hyperplasticity (Figure 1).

2. Based on the patient’s clinical presentation and findings on CT, what would be the most likely diagnosis for this incidentally found adrenal mass?

• Adrenocortical adenoma
• Adrenocortical carcinoma
• Metastatic mass
• Pheochromocytoma

Adrenocortical adenoma can present as a small homogeneous mass of variable size, with smooth margins, and rarely containing hemorrhagic tissue or calcifications. The typical density on nonenhanced CT is less than 10 HU. On enhanced CT, it is nonvascular. T2-weighted magnetic resonance imaging (MRI) shows a lesion of the same intensity as liver tissue.⁶

Adrenocortical adenoma is not classically associated with autologous activity and thus is less likely to explain our patient’s symptoms.

Adrenocortical carcinoma can present as a large heterogeneous mass, usually greater than 4 cm in diameter, with irregular margins and areas of necrosis, hemorrhage, or calcification. The typical density on nonenhanced CT is greater than 10 HU. On enhanced CT, the mass is usually vascular, and T2-weighted MRI will show a lesion more intense than liver tissue.⁶

Adrenocortical carcinoma is also not classically associated with autologous activity, and so is not likely to explain our patient’s symptoms.⁶

Metastatic disease can present with masses of variable size, often bilaterally, and occasionally with cysts or areas of hemorrhage. The typical density of metastatic lesions on nonenhanced CT is greater than 10 HU. On enhanced CT, they are usually vascular, and on T2-weighted MRI they are hyperintense.⁶ The characteristics of the mass and the absence of a primary malignancy on CT of the chest and abdomen do not support the diagnosis of metastatic disease.

Pheochromocytoma is a neuroendocrine tumor of the adrenal medulla that can present as a large heterogeneous mass, greater than 3 cm in diameter, with clear margins and cysts or areas of hemorrhage. Extra-adrenal neuroendocrine tumors are typically called paragangliomas and have features similar to those of pheochromocytoma. The typical density of pheochromocytoma on nonenhanced CT is greater than 10 HU. On enhanced CT, it is usually vascular, and T2-weighted MRI shows a hyperintense lesion. Pheochromocytoma can be biochemically active and thus can cause signs and symptoms that will lead to the diagnosis.⁶

Other imaging tests may play a role in the evaluation of adrenal masses but are not required for the diagnosis of pheochromocytoma. Functional positron emission tomography using metaiodobenzylguanidine labeled with iodine 123 or-iodine 131 or using the glucose analogue F-18 fluorodeoxyglucose has been used in the initial assessment of pheochromocytoma, with good sensitivity and specificity.^7,8

Our patient’s pacemaker-defibrillator precluded him from undergoing MRI.

DIAGNOSIS: PHEOCHROMOCYTOMA

Pheochromocytoma was highly suspected on the basis of the patient’s clinical presentation, and metoprolol was immediately discontinued. He was started on the calcium channel blocker verapamil and the alpha-blocker phenoxybenzamine.

Serum samples were obtained to measure metanephrines, dehydroepiandrosterone, aldosterone, and cortisol, and a 24-hour urine collection was obtained to measure creatinine, dopamine, epinephrine, norepinephrine, cortisol, and metanephrines. Based on the results (Table 1) and on the findings on imaging, the patient was diagnosed with pheochromocytoma. A surgical consultation was obtained, and surgery was recommended.

WHEN DOES PHEOCHROMOCYTOMA CALL FOR SURGERY?

3. Which criterion is most important when determining the need for surgery for pheochromocytoma?

• Findings on fine-needle aspiration biopsy
• Biochemical activity
• Size of the mass
• Bilateral masses

Fine-needle aspiration biopsy can be done when a mass is found incidentally and no evidence of biochemical activity is detected, although it is not an essential part of the diagnostic workup.⁹ In most cases, the sampling from fine-needle aspiration is not sufficient to achieve a diagnosis.

Biochemical activity is the most important factor when determining the need for prompt surgical intervention. The excess circulating catecholamines have been associated with increased risk of cardiovascular morbidity and death independent of the morbidity associated with hypertension alone.¹⁰ Biochemical activity can be independent of the size of the mass, but larger masses typically present with symptoms.

Bilateral masses have been associated with metastatic disease.¹¹ In retrospect, the patient’s history of hypertension and cerebrovascular accident could be associated with the development of a catecholamine-releasing tumor.

A GOOD OUTCOME FROM SURGERY

The patient was continued on phenoxybenzamine for 7 days and responded well to this therapy.

After this preoperative preparation, he underwent laparoscopic right adrenalectomy with excision of a retroperitoneal adrenal mass. His postoperative course was complicated by transient hypotension requiring low-dose vasopressin support for less than 24 hours. He was then restarted on his previous dosage of metoprolol and was discharged home on postoperative day 5 with stable blood pressure. Follow-up 24-hour urine collection 1 month after he was discharged showed normalization of metanephrine, normetanephrine, epinephrine, and norepinephrine levels.

Despite low suspicion for an underlying genetic syndrome, he was referred for genetic testing and was scheduled to have a repeat 24-hour urine collection and imaging in 6 months to follow his enlarged left adrenal gland, which did not appear to be metabolically hyperactive.

4. What is the most common perioperative complication of pheochromocytoma excision?

• Hypoglycemia
• Hypotension
• Hypocortisolism
• Hypertension
• Tachycardia

Hypoglycemia has been observed after removal of pheochromocytoma, as levels of catecholamines (which normally inhibit pancreatic beta cells) decrease and insulin secretion consequently increases.¹² Our patient developed hypoglycemia before surgery, not after, and it was likely due to the combination of his antidiabetic therapy, weight loss, and decreased oral intake.

Hypotension is the most common complication in the perioperative period. It is associated with excessive loss of catecholamine secretion. It is usually short-lived but may require aggressive administration of intravenous fluids and use of sympathomimetic agents.

Hypocortisolism is unlikely in patients with pheochromocytoma, but it is likely after removal of adrenocortical adenoma.

Hypertension and tachycardia affect up to 40% of pheochromocytoma patients in some case series.¹²

PHEOCHROMOCYTOMA: A CATECHOLAMINE-SECRETING TUMOR

The pathophysiology of pheochromocytoma is complex. It is characterized by accelerated growth of cells producing catecholamines, which may produce symptoms when secreted into the bloodstream. The classic triad of symptoms is headache, hypertension, and hyperglycemia, although our patient had very low blood sugar levels. Other common symptoms are nausea, orthostasis, and tremor, although not all symptoms are invariably seen.

Genetic testing recommended

Genetic associations have been described and are thought to be responsible for 20% to 30% of cases of pheochromocytoma. All associated germline mutations are autosomal dominant, some with variable penetrance. These include:

• Succinate dehydrogenase subunit B, C, and D mutations
• von Hippel-Lindau syndrome
• Multiple endocrine neoplasia type 1 and type 2 syndromes
• Neurofibromatosis type 1.^13,14

The succinate dehydrogenase subunit mutations have been associated with, but not limited to, extra-adrenal adenomas (paragangliomas) and carry a worse prognosis.

Some experts recommend genetic testing in all cases of pheochromocytoma, sporadic or familial, and this testing should be followed by counseling if a mutation is found.¹⁵ Others recommend genetic testing based on the patient’s age (under age 50), history, imaging, and biochemical features of the tumor (metanephrines predominate in multiple endocrine neoplasia syndromes, and normetanephrines in von Hippel-Lindau syndrome).¹³

Serious consequences

A thorough evaluation is recommended, since pheochromocytoma has been associated with increased cardiovascular morbidity. In a retrospective series, Stolk et al¹⁰ reported that patients with pheochromocytoma had a higher incidence of myocardial infarction, angina, and stroke in the years preceding the diagnosis than did patients with essential hypertension (13.8% vs 1.1%, P < .001).¹⁰

Catecholamine cardiomyopathy has been described and shares clinical features with Takotsubo or stress cardiomyopathy, with global left ventricular systolic and diastolic dysfunction that improve or resolve after the adrenergic insult is removed.¹⁶

Conditions that warrant further evaluation or that may suggest pheochromocytoma are malignant hypertension, hypertensive encephalopathy, ischemic stroke, subarachnoid hemorrhage, acute pulmonary edema, angina pectoris, myocardial infarction, aortic dissection, and kidney injury.

When to suspect pheochromocytoma

Pheochromocytoma should be suspected in a patient with resistant hypertension, family history, or imaging findings that suggest an adrenal mass with a heterogeneous appearance. The diagnostic algorithm follows the same pathway as for the evaluation of an incidentally found adrenal mass, with determination of its dimension and characteristics by CT or MRI, and with biochemical testing of urine catecholamines, plasma free metanephrines, renin, aldosterone, and cortisol.

The diagnosis of pheochromocytoma is established by obtaining fractionated metanephrines and catecholamines in a 24-hour urine collection (sensitivity 90%, specificity 98%). Analysis of plasma metanephrines has a higher sensitivity (97%) but lower specificity (85%).¹⁷ The combination of typical signs, symptoms, and laboratory findings makes the diagnosis likely, especially in combination with a unilateral adrenal mass.

Laparoscopic surgery after medical preparation for active tumors

If the mass appears benign and not biochemically hyperactive, then follow-up at 1 year is recommended, with repeat testing. Surgical evaluation and intervention is recommended for lesions that appear malignant or that are biochemically active and clinically symptomatic.⁹

Preoperative hemodynamic control is essential in the management of pheochromocytoma to prevent or minimize hemodynamic changes that can be driven by increased catecholamines. Control is typically achieved with initial alpha-blockade and then beta-blockade to avoid worsening hypertension and to prevent an acute hypertensive crisis during surgical intervention. Phenoxybenzamine, the mainstay of therapy, is a nonselective alpha-blocker with a long duration of action that requires titration over several days up to 3 weeks.

A selective alpha-1-blocker such as doxazosin can be used to control postoperative hypotension, as it has a shorter half-life than phenoxybenzamine. Alternative strategies include calcium channel blockers, centrally acting sympathetic blockers, and magnesium.¹⁸

Laparoscopic adrenalectomy by an experienced surgeon after excellent medical preparation is often considered the treatment of choice, but for larger or malignant masses, an open procedure is recommended. The risk of perioperative morbidity and death can be reduced by adequate medical management. With successful surgical resection, the long-term prognosis is favorable.

A 55-year-old lawyer with hypertension well controlled on lisinopril and amlodipine is scheduled for elective hernia repair under general anesthesia. His surgeon has referred him for a preoperative evaluation. He has never smoked, runs 4 miles on days off from work, and enjoys long hiking trips. On examination, his body mass index is 26 kg/m² and his blood pressure is 130/78 mm Hg; his cardiac examination and the rest of the clinical examination are unremarkable. He asks if he should have an electrocardiogram (ECG) as a part of his workup.

A preoperative ECG is not routinely recommended in all asymptomatic patients undergoing noncardiac surgery.

Consider obtaining an ECG in patients planning to undergo a high-risk surgical procedure, especially if they have one or more clinical risk factors for coronary artery disease, and in patients undergoing elevated-cardiac-risk surgery who are known to have coronary artery disease, chronic heart failure, peripheral arterial disease, or cerebrovascular disease. However, a preoperative ECG is not routinely recommended for patients perceived to be at low cardiac risk who are planning to undergo low-risk surgery. In those patients it could delay surgery unnecessarily, cause further unnecessary testing, drive up costs, and increase patient anxiety.

Here we discuss the perioperative cardiac risk based on type of surgery and patient characteristics and summarize the current guidelines and recommendations on obtaining a preoperative 12-lead ECG in patients undergoing noncardiac surgery.

RISK DEPENDS ON TYPE OF SURGERY AND PATIENT FACTORS

Physicians, including primary care physicians, hospitalists, cardiologists, and anesthesiologists, are routinely asked to evaluate patients before surgical procedures. The purpose of the preoperative evaluation is to optimize existing medical conditions, to identify undiagnosed conditions that can increase risk of perioperative morbidity and death, and to suggest strategies to mitigate risk.^1,2

Cardiac risk is multifactorial, and risk factors for postoperative adverse cardiac events include the type of surgery and patient factors.^1,3

Cardiac risk based on type of surgery

Low-risk procedures are those in which the risk of a perioperative major adverse cardiac event is less than 1%.^1,4 Examples:

• Ambulatory surgery
• Breast or plastic surgery
• Cataract surgery
• Endoscopic procedures.

Elevated-risk procedures are those in which the risk is 1% or higher. Examples:

• Intraperitoneal surgery
• Intrathoracic surgery
• Carotid endarterectomy
• Head and neck surgery
• Orthopedic surgery
• Prostate surgery
• Aortic surgery
• Major vascular surgery
• Peripheral arterial surgery.

Cardiac risk based on patient factors

The 2014 American College of Cardiology and American Heart Association (ACC/AHA) perioperative guidelines list a number of clinical risk factors for perioperative cardiac morbidity and death.¹ These include coronary artery disease, chronic heart failure, clinically suspected moderate or greater degrees of valvular heart disease, arrhythmias, conduction disorders, pulmonary vascular disease, and adult congenital heart disease.

Patients with these conditions and patients with unstable coronary syndromes warrant preoperative ECGs and sometimes even urgent interventions before any nonemergency surgery, provided such interventions would affect decision-making and perioperative care.¹

The risk of perioperative cardiac morbidity and death can be calculated using either the Revised Cardiac Risk Index scoring system or the American College of Surgeons National Surgical Quality Improvement Program calculator.¹⁵⁷ The former is fairly simple, validated, and accepted, while the latter requires use of online calculators (eg, www.surgicalriskcalculator.com/miorcardiacarrest, www.riskcalculator.facs.org).

The Revised Cardiac Risk Index has six clinical predictors of major perioperative cardiac events:

• History of cerebrovascular disease
• Prior or current compensated congestive heart failure
• History of coronary artery disease
• Insulin-dependent diabetes mellitus
• Renal insufficiency, defined as a serum creatinine level of 2 mg/dL or higher
• Patient undergoing suprainguinal vascular, intraperitoneal, or intrathoracic surgery.

A patient who has 0 or 1 of these predictors would have a low risk of a major adverse cardiac event, whereas a patient with 2 or more would have elevated risk. These risk factors must be taken into consideration to determine the need, if any, for a preoperative ECG.

What an ECG can tell us

Abnormalities such as left ventricular hypertrophy, ST-segment depression, and pathologic Q waves on a preoperative ECG in a patient undergoing an elevated-risk surgical procedure may predict adverse perioperative cardiac events.^3,6

In a retrospective study of 23,036 patients, Noordzij et al⁷ found that in patients undergoing elevated-risk surgery, those with an abnormal preoperative ECG had a higher incidence of cardiovascular death than those with a normal ECG. However, a preoperative ECG was obtained only in patients with established coronary artery disease or risk factors for cardiovascular disease. Hence, although an abnormal ECG in such patients undergoing elevated-risk surgery was predictive of adverse postoperative cardiac outcomes, we cannot say that the same would apply to patients without these characteristics undergoing elevated-risk surgery.

In a prospective observational study of patients with known coronary artery disease undergoing major noncardiac surgery, a preoperative ECG was found to contain prognostic information and was predictive of long-term outcome independent of clinical findings and perioperative ischemia.⁸

CURRENT GUIDELINES AND RECOMMENDATIONS

Several guidelines address whether to order a preoperative ECG but are mostly based on low-level evidence and expert opinion.^1,2,6,9

Current guidelines recommend obtaining a preoperative ECG in patients with known coronary, peripheral arterial, or cerebrovascular disease.^1,6,9

Obesity and associated comorbidities such as coronary artery disease, heart failure, systemic hypertension, and sleep apnea can predispose to increased perioperative complications. A preoperative 12-lead ECG is reasonable in morbidly obese patients (body mass index ≥ 40 kg/m²) and in obese patients (body mass index ≥ 30 kg/m²) with at least one risk factor for coronary artery disease or poor exercise tolerance, or both.¹⁰

Liu et al¹¹ looked at the predictive value of a preoperative 12-lead ECG in 513 elderly patients (age ≥ 70) undergoing noncardiac surgery and found that electrocardiographic abnormalities were not predictive of adverse cardiac outcomes. In this study, although electrocardiographic abnormalities were common (noted in 75% of the patients), they were nonspecific and less useful in predicting postoperative cardiac complications than was the presence of comorbidities.¹¹ Age alone as a cutoff for obtaining a preoperative ECG is not predictive of postoperative outcomes and a preoperative ECG is not warranted in all elderly patients. This is also reflected in current ACC/AHA guidelines on perioperative cardiovascular evaluation¹ and is a change from prior ACC/AHA guidelines when age was used as a criterion for preoperative ECGs.¹²

Current guidelines do not recommend getting a preoperative ECG in asymptomatic patients undergoing low-cardiac-risk surgery.^1,4,9

Although the ideal time for ordering an ECG before a planned surgery is unknown, obtaining one within 90 days before the surgery is considered adequate in stable patients in whom an ECG is indicated.¹

BACK TO OUR PATIENT

On the basis of current evidence, our patient does not need a preoperative ECG, as it is unlikely to alter his perioperative management and instead may delay his surgery unnecessarily if any nonspecific changes prompt further cardiac workup.

CLINICAL BOTTOM LINE

Although frequently ordered in clinical practice, preoperative electrocardiography has a limited role in predicting postoperative outcome and should be ordered only in the appropriate clinical setting.¹ Moreover, there is little evidence that outcomes are better if we obtain an ECG before surgery. The clinician should consider patient factors and the type of surgery before ordering diagnostic tests, including electrocardiography.

In asymptomatic patients undergoing nonemergent surgery:

• It is reasonable to obtain a preoperative ECG in patients with known coronary artery disease, significant arrhythmia, peripheral arterial disease, cerebrovascular disease, chronic heart failure, or other significant structural heart disease undergoing elevated-cardiac-risk surgery.
• Do not order a preoperative ECG in asymptomatic patients undergoing low-risk surgery.
• Obtaining a preoperative ECG is reasonable in morbidly obese patients and in obese patients with one or more risk factors for coronary artery disease, or poor exercise tolerance, undergoing high-risk surgery.

A 55-year-old lawyer with hypertension well controlled on lisinopril and amlodipine is scheduled for elective hernia repair under general anesthesia. His surgeon has referred him for a preoperative evaluation. He has never smoked, runs 4 miles on days off from work, and enjoys long hiking trips. On examination, his body mass index is 26 kg/m² and his blood pressure is 130/78 mm Hg; his cardiac examination and the rest of the clinical examination are unremarkable. He asks if he should have an electrocardiogram (ECG) as a part of his workup.

A preoperative ECG is not routinely recommended in all asymptomatic patients undergoing noncardiac surgery.

Consider obtaining an ECG in patients planning to undergo a high-risk surgical procedure, especially if they have one or more clinical risk factors for coronary artery disease, and in patients undergoing elevated-cardiac-risk surgery who are known to have coronary artery disease, chronic heart failure, peripheral arterial disease, or cerebrovascular disease. However, a preoperative ECG is not routinely recommended for patients perceived to be at low cardiac risk who are planning to undergo low-risk surgery. In those patients it could delay surgery unnecessarily, cause further unnecessary testing, drive up costs, and increase patient anxiety.

Here we discuss the perioperative cardiac risk based on type of surgery and patient characteristics and summarize the current guidelines and recommendations on obtaining a preoperative 12-lead ECG in patients undergoing noncardiac surgery.

RISK DEPENDS ON TYPE OF SURGERY AND PATIENT FACTORS

Physicians, including primary care physicians, hospitalists, cardiologists, and anesthesiologists, are routinely asked to evaluate patients before surgical procedures. The purpose of the preoperative evaluation is to optimize existing medical conditions, to identify undiagnosed conditions that can increase risk of perioperative morbidity and death, and to suggest strategies to mitigate risk.^1,2

Cardiac risk is multifactorial, and risk factors for postoperative adverse cardiac events include the type of surgery and patient factors.^1,3

Cardiac risk based on type of surgery

Low-risk procedures are those in which the risk of a perioperative major adverse cardiac event is less than 1%.^1,4 Examples:

• Ambulatory surgery
• Breast or plastic surgery
• Cataract surgery
• Endoscopic procedures.

Elevated-risk procedures are those in which the risk is 1% or higher. Examples:

• Intraperitoneal surgery
• Intrathoracic surgery
• Carotid endarterectomy
• Head and neck surgery
• Orthopedic surgery
• Prostate surgery
• Aortic surgery
• Major vascular surgery
• Peripheral arterial surgery.

Cardiac risk based on patient factors

The 2014 American College of Cardiology and American Heart Association (ACC/AHA) perioperative guidelines list a number of clinical risk factors for perioperative cardiac morbidity and death.¹ These include coronary artery disease, chronic heart failure, clinically suspected moderate or greater degrees of valvular heart disease, arrhythmias, conduction disorders, pulmonary vascular disease, and adult congenital heart disease.

Patients with these conditions and patients with unstable coronary syndromes warrant preoperative ECGs and sometimes even urgent interventions before any nonemergency surgery, provided such interventions would affect decision-making and perioperative care.¹

The risk of perioperative cardiac morbidity and death can be calculated using either the Revised Cardiac Risk Index scoring system or the American College of Surgeons National Surgical Quality Improvement Program calculator.¹⁵⁷ The former is fairly simple, validated, and accepted, while the latter requires use of online calculators (eg, www.surgicalriskcalculator.com/miorcardiacarrest, www.riskcalculator.facs.org).

The Revised Cardiac Risk Index has six clinical predictors of major perioperative cardiac events:

• History of cerebrovascular disease
• Prior or current compensated congestive heart failure
• History of coronary artery disease
• Insulin-dependent diabetes mellitus
• Renal insufficiency, defined as a serum creatinine level of 2 mg/dL or higher
• Patient undergoing suprainguinal vascular, intraperitoneal, or intrathoracic surgery.

A patient who has 0 or 1 of these predictors would have a low risk of a major adverse cardiac event, whereas a patient with 2 or more would have elevated risk. These risk factors must be taken into consideration to determine the need, if any, for a preoperative ECG.

What an ECG can tell us

Abnormalities such as left ventricular hypertrophy, ST-segment depression, and pathologic Q waves on a preoperative ECG in a patient undergoing an elevated-risk surgical procedure may predict adverse perioperative cardiac events.^3,6

In a retrospective study of 23,036 patients, Noordzij et al⁷ found that in patients undergoing elevated-risk surgery, those with an abnormal preoperative ECG had a higher incidence of cardiovascular death than those with a normal ECG. However, a preoperative ECG was obtained only in patients with established coronary artery disease or risk factors for cardiovascular disease. Hence, although an abnormal ECG in such patients undergoing elevated-risk surgery was predictive of adverse postoperative cardiac outcomes, we cannot say that the same would apply to patients without these characteristics undergoing elevated-risk surgery.

In a prospective observational study of patients with known coronary artery disease undergoing major noncardiac surgery, a preoperative ECG was found to contain prognostic information and was predictive of long-term outcome independent of clinical findings and perioperative ischemia.⁸

CURRENT GUIDELINES AND RECOMMENDATIONS

Several guidelines address whether to order a preoperative ECG but are mostly based on low-level evidence and expert opinion.^1,2,6,9

Current guidelines recommend obtaining a preoperative ECG in patients with known coronary, peripheral arterial, or cerebrovascular disease.^1,6,9

Obesity and associated comorbidities such as coronary artery disease, heart failure, systemic hypertension, and sleep apnea can predispose to increased perioperative complications. A preoperative 12-lead ECG is reasonable in morbidly obese patients (body mass index ≥ 40 kg/m²) and in obese patients (body mass index ≥ 30 kg/m²) with at least one risk factor for coronary artery disease or poor exercise tolerance, or both.¹⁰

Liu et al¹¹ looked at the predictive value of a preoperative 12-lead ECG in 513 elderly patients (age ≥ 70) undergoing noncardiac surgery and found that electrocardiographic abnormalities were not predictive of adverse cardiac outcomes. In this study, although electrocardiographic abnormalities were common (noted in 75% of the patients), they were nonspecific and less useful in predicting postoperative cardiac complications than was the presence of comorbidities.¹¹ Age alone as a cutoff for obtaining a preoperative ECG is not predictive of postoperative outcomes and a preoperative ECG is not warranted in all elderly patients. This is also reflected in current ACC/AHA guidelines on perioperative cardiovascular evaluation¹ and is a change from prior ACC/AHA guidelines when age was used as a criterion for preoperative ECGs.¹²

Current guidelines do not recommend getting a preoperative ECG in asymptomatic patients undergoing low-cardiac-risk surgery.^1,4,9

Although the ideal time for ordering an ECG before a planned surgery is unknown, obtaining one within 90 days before the surgery is considered adequate in stable patients in whom an ECG is indicated.¹

BACK TO OUR PATIENT

On the basis of current evidence, our patient does not need a preoperative ECG, as it is unlikely to alter his perioperative management and instead may delay his surgery unnecessarily if any nonspecific changes prompt further cardiac workup.

CLINICAL BOTTOM LINE

Although frequently ordered in clinical practice, preoperative electrocardiography has a limited role in predicting postoperative outcome and should be ordered only in the appropriate clinical setting.¹ Moreover, there is little evidence that outcomes are better if we obtain an ECG before surgery. The clinician should consider patient factors and the type of surgery before ordering diagnostic tests, including electrocardiography.

In asymptomatic patients undergoing nonemergent surgery:

• It is reasonable to obtain a preoperative ECG in patients with known coronary artery disease, significant arrhythmia, peripheral arterial disease, cerebrovascular disease, chronic heart failure, or other significant structural heart disease undergoing elevated-cardiac-risk surgery.
• Do not order a preoperative ECG in asymptomatic patients undergoing low-risk surgery.
• Obtaining a preoperative ECG is reasonable in morbidly obese patients and in obese patients with one or more risk factors for coronary artery disease, or poor exercise tolerance, undergoing high-risk surgery.

A 55-year-old lawyer with hypertension well controlled on lisinopril and amlodipine is scheduled for elective hernia repair under general anesthesia. His surgeon has referred him for a preoperative evaluation. He has never smoked, runs 4 miles on days off from work, and enjoys long hiking trips. On examination, his body mass index is 26 kg/m² and his blood pressure is 130/78 mm Hg; his cardiac examination and the rest of the clinical examination are unremarkable. He asks if he should have an electrocardiogram (ECG) as a part of his workup.

A preoperative ECG is not routinely recommended in all asymptomatic patients undergoing noncardiac surgery.

Consider obtaining an ECG in patients planning to undergo a high-risk surgical procedure, especially if they have one or more clinical risk factors for coronary artery disease, and in patients undergoing elevated-cardiac-risk surgery who are known to have coronary artery disease, chronic heart failure, peripheral arterial disease, or cerebrovascular disease. However, a preoperative ECG is not routinely recommended for patients perceived to be at low cardiac risk who are planning to undergo low-risk surgery. In those patients it could delay surgery unnecessarily, cause further unnecessary testing, drive up costs, and increase patient anxiety.

Here we discuss the perioperative cardiac risk based on type of surgery and patient characteristics and summarize the current guidelines and recommendations on obtaining a preoperative 12-lead ECG in patients undergoing noncardiac surgery.

RISK DEPENDS ON TYPE OF SURGERY AND PATIENT FACTORS

Physicians, including primary care physicians, hospitalists, cardiologists, and anesthesiologists, are routinely asked to evaluate patients before surgical procedures. The purpose of the preoperative evaluation is to optimize existing medical conditions, to identify undiagnosed conditions that can increase risk of perioperative morbidity and death, and to suggest strategies to mitigate risk.^1,2

Cardiac risk is multifactorial, and risk factors for postoperative adverse cardiac events include the type of surgery and patient factors.^1,3

Cardiac risk based on type of surgery

Low-risk procedures are those in which the risk of a perioperative major adverse cardiac event is less than 1%.^1,4 Examples:

• Ambulatory surgery
• Breast or plastic surgery
• Cataract surgery
• Endoscopic procedures.

Elevated-risk procedures are those in which the risk is 1% or higher. Examples:

• Intraperitoneal surgery
• Intrathoracic surgery
• Carotid endarterectomy
• Head and neck surgery
• Orthopedic surgery
• Prostate surgery
• Aortic surgery
• Major vascular surgery
• Peripheral arterial surgery.

Cardiac risk based on patient factors

The 2014 American College of Cardiology and American Heart Association (ACC/AHA) perioperative guidelines list a number of clinical risk factors for perioperative cardiac morbidity and death.¹ These include coronary artery disease, chronic heart failure, clinically suspected moderate or greater degrees of valvular heart disease, arrhythmias, conduction disorders, pulmonary vascular disease, and adult congenital heart disease.

Patients with these conditions and patients with unstable coronary syndromes warrant preoperative ECGs and sometimes even urgent interventions before any nonemergency surgery, provided such interventions would affect decision-making and perioperative care.¹

The risk of perioperative cardiac morbidity and death can be calculated using either the Revised Cardiac Risk Index scoring system or the American College of Surgeons National Surgical Quality Improvement Program calculator.¹⁵⁷ The former is fairly simple, validated, and accepted, while the latter requires use of online calculators (eg, www.surgicalriskcalculator.com/miorcardiacarrest, www.riskcalculator.facs.org).

The Revised Cardiac Risk Index has six clinical predictors of major perioperative cardiac events:

• History of cerebrovascular disease
• Prior or current compensated congestive heart failure
• History of coronary artery disease
• Insulin-dependent diabetes mellitus
• Renal insufficiency, defined as a serum creatinine level of 2 mg/dL or higher
• Patient undergoing suprainguinal vascular, intraperitoneal, or intrathoracic surgery.

A patient who has 0 or 1 of these predictors would have a low risk of a major adverse cardiac event, whereas a patient with 2 or more would have elevated risk. These risk factors must be taken into consideration to determine the need, if any, for a preoperative ECG.

What an ECG can tell us

Abnormalities such as left ventricular hypertrophy, ST-segment depression, and pathologic Q waves on a preoperative ECG in a patient undergoing an elevated-risk surgical procedure may predict adverse perioperative cardiac events.^3,6

In a retrospective study of 23,036 patients, Noordzij et al⁷ found that in patients undergoing elevated-risk surgery, those with an abnormal preoperative ECG had a higher incidence of cardiovascular death than those with a normal ECG. However, a preoperative ECG was obtained only in patients with established coronary artery disease or risk factors for cardiovascular disease. Hence, although an abnormal ECG in such patients undergoing elevated-risk surgery was predictive of adverse postoperative cardiac outcomes, we cannot say that the same would apply to patients without these characteristics undergoing elevated-risk surgery.

In a prospective observational study of patients with known coronary artery disease undergoing major noncardiac surgery, a preoperative ECG was found to contain prognostic information and was predictive of long-term outcome independent of clinical findings and perioperative ischemia.⁸

CURRENT GUIDELINES AND RECOMMENDATIONS

Several guidelines address whether to order a preoperative ECG but are mostly based on low-level evidence and expert opinion.^1,2,6,9

Current guidelines recommend obtaining a preoperative ECG in patients with known coronary, peripheral arterial, or cerebrovascular disease.^1,6,9

Obesity and associated comorbidities such as coronary artery disease, heart failure, systemic hypertension, and sleep apnea can predispose to increased perioperative complications. A preoperative 12-lead ECG is reasonable in morbidly obese patients (body mass index ≥ 40 kg/m²) and in obese patients (body mass index ≥ 30 kg/m²) with at least one risk factor for coronary artery disease or poor exercise tolerance, or both.¹⁰

Liu et al¹¹ looked at the predictive value of a preoperative 12-lead ECG in 513 elderly patients (age ≥ 70) undergoing noncardiac surgery and found that electrocardiographic abnormalities were not predictive of adverse cardiac outcomes. In this study, although electrocardiographic abnormalities were common (noted in 75% of the patients), they were nonspecific and less useful in predicting postoperative cardiac complications than was the presence of comorbidities.¹¹ Age alone as a cutoff for obtaining a preoperative ECG is not predictive of postoperative outcomes and a preoperative ECG is not warranted in all elderly patients. This is also reflected in current ACC/AHA guidelines on perioperative cardiovascular evaluation¹ and is a change from prior ACC/AHA guidelines when age was used as a criterion for preoperative ECGs.¹²

Current guidelines do not recommend getting a preoperative ECG in asymptomatic patients undergoing low-cardiac-risk surgery.^1,4,9

Although the ideal time for ordering an ECG before a planned surgery is unknown, obtaining one within 90 days before the surgery is considered adequate in stable patients in whom an ECG is indicated.¹

BACK TO OUR PATIENT

On the basis of current evidence, our patient does not need a preoperative ECG, as it is unlikely to alter his perioperative management and instead may delay his surgery unnecessarily if any nonspecific changes prompt further cardiac workup.

CLINICAL BOTTOM LINE

Although frequently ordered in clinical practice, preoperative electrocardiography has a limited role in predicting postoperative outcome and should be ordered only in the appropriate clinical setting.¹ Moreover, there is little evidence that outcomes are better if we obtain an ECG before surgery. The clinician should consider patient factors and the type of surgery before ordering diagnostic tests, including electrocardiography.

In asymptomatic patients undergoing nonemergent surgery:

• It is reasonable to obtain a preoperative ECG in patients with known coronary artery disease, significant arrhythmia, peripheral arterial disease, cerebrovascular disease, chronic heart failure, or other significant structural heart disease undergoing elevated-cardiac-risk surgery.
• Do not order a preoperative ECG in asymptomatic patients undergoing low-risk surgery.
• Obtaining a preoperative ECG is reasonable in morbidly obese patients and in obese patients with one or more risk factors for coronary artery disease, or poor exercise tolerance, undergoing high-risk surgery.

More then 2.5 million patients in the United States are on long-term anticoagulation therapy for atrial fibrillation, venous thromboembolic disease, or mechanical heart valves,¹ and the number is expected to rise as the population ages. Each year, about 10% of these patients undergo an invasive procedure or surgery that requires temporary interruption of anticoagulation.²

Most physicians are familiar with the perioperative management of warfarin, a vitamin K antagonist, since for decades it has been the sole oral anticoagulant available. However, many physicians lack experience with the three target-specific oral anticoagulants (TSOACs; also known as “novel” oral anticoagulants) approved so far: the direct thrombin inhibitor dabigatran (Pradaxa) and the direct factor Xa inhibitors rivaroxaban (Xarelto) and apixaban (Eliquis).

With their rapid onset of action, predictable pharmacokinetics, relatively short half-lives, and fewer drug-drug interactions than warfarin, TSOACs overcome many of the limitations of the older oral anticoagulant warfarin. In many ways, these qualities simplify the perioperative management of anticoagulation. At the same time, these new drugs also bring new challenges: caution is needed in patients with renal impairment; the level of anticoagulation is difficult to assess; and there is no specific antidote or standardized procedure to reverse their anticoagulant effect. While various periprocedural protocols for TSOAC therapy have been proposed, evidence-based guidelines are still to come.

This article first discusses the pharmacology of dabigatran, rivaroxaban, and apixaban that is pertinent to the perioperative period. It then briefly reviews the general principles of perioperative management of anticoagulation. The final section provides specific recommendations for the perioperative management of TSOACs.

PHARMACOLOGY OF TARGET-SPECIFIC ORAL ANTICOAGULANTS

Dabigatran, a factor IIa inhibitor

Information from references 3, 4, 14, and 23.

Dabigatran is an oral direct thrombin (factor IIa) inhibitor. It exerts its anticoagulant effect by blocking the generation of fibrin, inhibiting platelet aggregation, and dampening the activity of factors V, VIII, and XI (Figure 1).^3,4 From its introduction in October 2010 through August 2012, nearly 3.7 million prescriptions were dispensed to 725,000 patients in the United States.⁵

Indications for dabigatran. Dabigatran is approved in the United States and Canada for preventing stroke in nonvalvular atrial fibrillation (Table 1).⁶ More recently, it received US approval for treating deep vein thrombosis or pulmonary embolism after 5 to 10 days of a parenteral anticoagulant.^7,8 It is also approved in Europe and Canada for preventing venous thromboembolism (VTE) after total hip replacement and knee arthroplasty.^9,10

Dabigatran is contraindicated in patients with a mechanical heart valve, based on a phase 2 study in which it conferred a higher risk of thromboembolism and bleeding than warfarin.^3,11

Pharmacokinetics of dabigatran. Dabigatran is formulated as a prodrug, dabigatran etexilate, in a capsule containing multiple small pellets.¹² The capsules should not be crushed, as this significantly increases oral bioavailability. The prodrug is absorbed across the gastric mucosa and is then rapidly converted to the active form (Table 2).

Plasma concentrations peak within 2 hours of ingestion, which means that therapeutic anticoagulation is achieved shortly after taking the drug.

Only 35% of dabigatran is protein-bound, which allows it to be removed by hemodialysis. Nearly 85% of the drug is eliminated in the urine. It has a half-life of 13 to 15 hours in patients with normal renal function.³ However, its half-life increases to about 27 hours in patients whose creatinine clearance is less than 30 mL/min. As a result, the dose must be reduced in patients with renal impairment (Table 1).

Dabigatran is not metabolized by the cytochrome P450 enzymes, but it is a substrate for P-glycoprotein, so it still has the potential for drug-drug interactions.³ Practitioners should be familiar with these potential interactions (Table 3), as they can result in higher- or lower-than-expected plasma concentrations of dabigatran in the perioperative period.¹³

Rivaroxaban, a factor Xa inhibitor

Rivaroxaban is an oral direct factor Xa inhibitor. It has been approved by the US Food and Drug Administration (FDA) for the prevention of stroke in nonvalvular atrial fibrillation, for VTE treatment, and for VTE prophylaxis after hip replacement or knee replacement (Table 1).^14–20 It has not yet been studied in patients with hip fracture.

Pharmacokinetics of rivaroxaban. Rivaroxaban is manufactured as a tablet that is best absorbed in the stomach (Table 2).¹⁴ In contrast to dabigatran, it can be crushed and, for example, mixed with applesauce for patients who have trouble swallowing. It can also be mixed with water and given via nasogastric tube; however, postpyloric administration should be avoided.

Plasma concentrations peak within a few hours after ingestion. Rivaroxaban is highly protein-bound, so it cannot be eliminated by hemodialysis.

The drug relies on renal elimination to a smaller degree than dabigatran, with one-third of the dose eliminated unchanged in the urine, one-third eliminated in the urine as inactive metabolite, and the remaining one-third eliminated in the feces. However, enough parent compound is cleared through the kidneys that the half-life of rivaroxaban increases from 8.3 hours in healthy individuals to 9.5 hours in patients whose creatinine clearance is less than 30 mL/min.²¹ As with dabigatran, the dose must be adjusted for renal impairment (Table 1).

Rivaroxaban has significant liver metabolism, specifically through the cytochrome P450 3A4 enzyme, and it is also a substrate of P-glycoprotein. Therefore, potential drug-drug interactions must be taken into account, as they may lead to important alterations in plasma concentrations (Table 3).

Apixaban, a factor Xa inhibitor

Apixaban is also an oral direct factor Xa inhibitor. It is the newest of the oral anticoagulants to be approved in the United States, specifically for preventing stroke in nonvalvular atrial fibrillation (Table 1).²²

Pharmacokinetics of apixaban. Apixaban is produced as a tablet that is absorbed slowly through the gastrointestinal tract, mainly the distal small bowel and ascending colon (Table  2).²³

Peak plasma concentrations are reached a few hours after ingestion. Like rivaroxaban, apixaban is highly protein-bound, so it cannot be removed by hemodialysis.

Apixaban is similar to rivaroxaban in that 27% of the parent compound is cleared through the kidneys, it undergoes significant hepatic metabolism through cytochrome P450 3A4, and it is a substrate for P-glycoprotein.

Drug-drug interactions must be considered as a potential source of altered drug exposure and clearance (Table 3).

Unlike dabigatran and rivaroxaban, dose reduction is not based on the calculated creatinine clearance. Instead, a reduced dose is required if the patient meets two of the following three criteria:

• Serum creatinine level ≥ 1.5 mg/dL
• Age ≥ 80
• Weight ≤ 60 kg (Table 1).

The American Heart Association/American Stroke Association guidelines further recommend against using apixaban in patients with a creatinine clearance less than 25 mL/min.²⁴

Edoxaban, a factor Xa inhibitor in development

Edoxaban (Savaysa), another factor Xa inhibitor, is available in Japan and has been submitted for approval in the United States for treating VTE and for preventing stroke in patients with

PERIOPERATIVE CONSIDERATIONS IN ANTICOAGULATION

Before addressing the perioperative management of TSOACs, let us review the evidence guiding the perioperative management of any chronic anticoagulant.

In fact, no large prospective randomized trial has clearly defined the risks and benefits of using or withholding a bridging anticoagulation strategy around surgery and other procedures, though the PERIOP 2 and BRIDGE trials are currently ongoing.^25,26 There are some data regarding continuing anticoagulation without interruption, but they have mainly been derived from specific groups (eg, patients on warfarin undergoing cardiac pacemaker or defibrillator placement) and in procedures that pose a very low risk of bleeding complications (eg, minor dental extractions, cataract surgery, dermatologic procedures).^2,27 Recommendations are, therefore, necessarily based on small perioperative trials and data gleaned from cohort review and from studies that did not involve surgical patients.

Ultimately, the decisions whether to discontinue oral anticoagulants and whether to employ bridging anticoagulation are based on assumptions about the risks of bleeding and the risk of thrombotic events, with similar assumptions regarding the effects of anticoagulants on both outcomes. In addition, the relative acceptance of bleeding vs thrombotic risks implicitly guides these complex decisions.

Perioperative bleeding risk

Many risk factors specific to the patient and to the type of surgery affect the rates and severity of perioperative bleeding.²⁸

As for patient-specific risk factors, a small retrospective cohort analysis revealed that a HAS-BLED score of 3 or higher was highly discriminating in predicting perioperative bleeding in atrial fibrillation patients receiving anticoagulation.²⁹ (The HAS-BLED score is based on hypertension, abnormal renal or liver function, stroke, bleeding, labile international normalized ratio [INR], elderly [age > 65] and drug therapy.³⁰) However, there are no widely validated tools that incorporate patient-specific factors to accurately predict bleeding risk in an individual patient.

Therefore, the American College of Chest Physicians (ACCP) guidelines suggest coarsely categorizing bleeding risk as either low or high solely on the basis of the type of procedure.² Procedures considered “high-risk” have a risk greater than 1.5% to 2% and include urologic surgery involving the prostate or kidney, colonic polyp resections, surgeries involving highly vascular organs such as the liver or spleen, joint replacements, cancer surgeries, and cardiac or neurosurgical procedures.

Perioperative thrombotic risk

The ACCP guidelines² place patients with atrial fibrillation, VTE, or mechanical heart valves in three risk groups for perioperative thromboembolism without anticoagulation, based on their annual risk of a thrombotic event:

• High risk—annual risk of a thrombotic event > 10%
• Moderate risk—5% to 10%
• Low risk—< 5%.

Comparing the risks calculated by these methods with the real-world risk of perioperative thrombosis highlights the problem of applying nonperioperative risk calculations: the perioperative period exposes patients to a higher risk than these models would predict.³¹ Nonetheless, these risk categorizations likely have some validity in stratifying patients into risk groups, even if the absolute risks are inaccurate.

Perioperative bridging for patients taking warfarin

Many patients with atrial fibrillation, VTE, or a mechanical heart valve need to interrupt their warfarin therapy because of the bleeding risk of an upcoming procedure.

The perioperative management of warfarin and other vitamin K antagonists is challenging because of the pharmacokinetics and pharmacodynamics of these drugs. Because it has a long half-life, warfarin usually must be stopped 4 to 5 days before a procedure in order to allow not only adequate clearance of the drug itself, but also restoration of functional clotting factors to normal or near-normal levels.¹² Warfarin can generally be resumed 12 to 24 hours after surgery, assuming adequate hemostasis has been achieved, and it will again take several days for the INR to reach the therapeutic range.

The ACCP guidelines recommend using the perioperative risk of thromboembolism to make decisions about the need for bridging anticoagulation during warfarin interruption.² They suggest that patients at high risk of thrombosis receive bridging with an alternative anticoagulant such as low-molecular-weight heparin or unfractionated heparin, because of the prolonged duration of subtherapeutic anticoagulation.

There has been clinical interest in using a TSOAC instead of low-molecular-weight or unfractionated heparin for bridging in the perioperative setting. Although this approach may be attractive from a cost and convenience perspective, it cannot be endorsed as yet because of the lack of information on the pros and cons of such an approach.

Patients at low thrombotic risk do not require bridging. In patients at moderate risk, the decision to bridge or not to bridge is based on careful consideration of patient-specific and surgery-specific factors.

PERIOPERATIVE MANAGEMENT OF TARGET-SPECIFIC ORAL ANTICOAGULANTS

As summarized above, the perioperative management strategy for chronic anticoagulation is based on limited evidence, even for drugs as well established as warfarin.

The most recent ACCP guidelines on the perioperative management of antithrombotic therapy do not mention TSOACs.² For now, the management strategy must be based on the pharmacokinetics of the drugs, package inserts from the manufacturers, and expert recommendations.^{3,14,23,32–34} Fortunately, because TSOACs have a more favorable pharmacokinetic profile than that of warfarin, their perioperative uses should be more streamlined. As always, the goal is to minimize the risk of both periprocedural bleeding and thromboembolism.

Timing of cessation of anticoagulation

The timing of cessation of TSOACs before an elective procedure depends primarily on two factors: the bleeding risk of the procedure and the patient’s renal function. Complete clearance of the medication is not necessary in all circumstances.

TSOACs should be stopped four to five half-lives before a procedure with a high bleeding risk, so that there is no or only minimal residual anticoagulant effect. The drug can be stopped two to three half-lives before a procedure with a low bleeding risk. Remember: the half-life increases as creatinine clearance decreases.

Specific recommendations may vary across institutions, but a suggested strategy is shown in Table 4.^{3,4,21,23,32–35} For the small subset of patients on P-glycoprotein or cytochrome P450 inhibitors or inducers, further adjustment in the time of discontinuation may be required.

Therapy does not need to be interrupted for procedures with a very low bleeding risk, as defined above.^33,34 There is also preliminary evidence that TSOACs, similar to warfarin, may be continued during cardiac pacemaker or defibrillator placement.³⁶

Evidence from clinical trials of perioperative TSOAC management

While the above recommendations are logical, studies are needed to prospectively evaluate perioperative management strategies.

The RE-LY trial (Randomized Evaluation of Long-Term Anticoagulation Therapy), which compared the effects of dabigatran and warfarin in preventing stroke in patients with atrial fibrillation, is one of the few clinical trials that also looked at periprocedural bleeding.³⁷ About a quarter of the RE-LY participants required interruption of anticoagulation for a procedure.

Warfarin was managed according to local practices. For most of the study, the protocol required that dabigatran be discontinued 24 hours before a procedure, regardless of renal function or procedure type. The protocol was later amended and closely mirrored the management plan outlined in Table 4.

With either protocol, there was no statistically significant difference between dabigatran and warfarin in the rates of bleeding and thrombotic complications in the 7 days before or 30 days after the procedure.

A major limitation of the study was that most patients underwent a procedure with a low bleeding risk, so the analysis was likely underpowered to evaluate rates of bleeding in higher-risk procedures.

The ROCKET-AF trial (Rivaroxaban Once-daily Oral Direct Factor Xa Inhibition Compared With Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation) also shed light on periprocedural bleeding.¹⁵ About 15% of the participants required temporary interruption of anticoagulation for a surgical or invasive procedure.³⁸

The study protocol called for discontinuing rivaroxaban 2 days before any procedure. Warfarin was to be held for 4 days to achieve a goal INR of 1.5 or less.¹⁵

Rates of major and nonmajor clinically significant bleeding at 30 days were similar with rivaroxaban and with warfarin.³⁸ As with the RE-LY trial, the retrospective analysis was probably underpowered for assessing rates of bleeding in procedures with higher risk.

Perioperative bridging

While stopping a TSOAC in the perioperative period decreases the risk of bleeding, it naturally increases the risk of thromboembolism. However, patients on TSOACs should not routinely require perioperative bridging with an alternative anticoagulant, regardless of thrombotic risk.

Of note, dabigatran, rivaroxaban, and apixaban carry black-box warnings that discontinuation places patients at higher risk of thrombotic events.^3,14,23 These warnings further state that coverage with an alternative anticoagulant should be strongly considered during interruption of therapy for reasons other than pathologic bleeding.

However, it does not necessarily follow that perioperative bridging is required. For example, the warning for rivaroxaban is based on the finding in the ROCKET-AF trial that patients in the rivaroxaban group had higher rates of stroke than those in the warfarin group after the study drugs were stopped at the end of the trial.³⁹ While there was initial concern that this could represent a prothrombotic rebound effect, the authors subsequently showed that patients in the rivaroxaban group were more likely to have had a subtherapeutic INR when transitioning to open-label vitamin-K-antagonist therapy.^39,40 There was no difference in the rate of stroke or systemic embolism between the rivaroxaban and warfarin groups when anticoagulation was temporarily interrupted for a procedure.³⁸

The risks and benefits of perioperative bridging with TSOACs are difficult to evaluate, given the dearth of trial data. In the RE-LY trial, only 17% of patients on dabigatran and 28% of patients on warfarin underwent periprocedural bridging.³⁷ The selection criteria and protocol for bridging were not reported. In the ROCKET-AF trial, only 9% of patients received bridging therapy despite a mean CHADS₂ score of 3.4.³⁸ (The CHADS₂ score is calculated as 1 point each for congestive heart failure, hypertension, age ≥ 75, and diabetes; 2 points for stroke or transient ischemic attack.) The decision to bridge or not was left to the individual investigator. As a result, the literature offers diverse opinions about the appropriateness of transitioning to an alternative anticoagulant.^41–43

Bridging does not make sense in most instances, since anticoagulants such as low-molecular-weight heparin have pharmacokinetics similar to those of the available TSOACs and also depend on renal clearance.⁴¹ However, there may be situations in which patients must be switched to a parenteral anticoagulant such as unfractionated or low-molecular-weight heparin. For example, if a TSOAC has to be held, the patient has acute renal failure, and a needed procedure is still several days away, it would be reasonable to start a heparin drip for an inpatient at increased thrombotic risk.

In patients with normal renal function, these alternative anticoagulants should be started at the time the next TSOAC dose would have been due.^3,14,23 In patients with reduced renal function, initiation of an alternative anticoagulant may need to be delayed 12 to 48 hours depending on which TSOAC is being used, as well as on the degree of renal dysfunction. This delay would help ensure that the onset of anticoagulation with the alternative anticoagulant is timed with the offset of therapeutic anticoagulation with the TSOAC.

Although limited, information from available coagulation assays may assist with the timing of initiation of an alternative anticoagulant (see the following section on laboratory monitoring). Serial testing with appropriate coagulation assays may help identify when most of a TSOAC has been cleared from a patient.

Laboratory monitoring

Inevitably, some patients on TSOACs require urgent or emergency surgery. In certain situations, such as before an orthopedic spine procedure, in which the complications of bleeding could be devastating, it may be necessary to know if any residual anticoagulant effect is present.

Monitoring dabigatran. As one might expect, direct thrombin inhibitors such as dabigatran can prolong the prothrombin time and activated partial thromboplastin time (aPTT).^44–47 However, the prothrombin time is not recommended for assessing the level of anticoagulation from dabigatran. Many institutions may be using a normal aPTT to rule out therapeutic concentrations of dabigatran, based on results from early in vitro and ex vivo studies.⁴⁶ While appealing from a practical standpoint, practitioners should exercise caution when relying on the aPTT to assess the risk of perioperative bleeding. A more recent investigation in patients treated with dabigatran found that up to 35% of patients with a normal aPTT still had a plasma concentration in the therapeutic range.⁴⁸

The thrombin time and ecarin clotting time are more sensitive tests for dabigatran. A normal thrombin time or ecarin clotting time indicates that no or only minimal dabigatran is present.⁴⁸ Unfortunately, these two tests often are either unavailable or are associated with long turnaround times, which limits their usefulness in the perioperative setting.

Monitoring rivaroxaban and apixaban. Factor Xa inhibitors such as rivaroxaban and apixaban can also influence the prothrombin time and aPTT (Figure 1).^{44–47,49,50} The aPTT is relatively insensitive to these drugs at low concentrations. It has been suggested that a normal prothrombin time can reasonably exclude therapeutic concentrations of rivaroxaban.^45,46 However, the effects on the prothrombin time are highly variable, changing with the reagent used.^49,50 In addition, apixaban appears to have less impact on the prothrombin time overall. The INR is not recommended for monitoring the effect of factor Xa inhibitors.

Anti-factor Xa assays likely represent the best option to provide true quantitative information on the level of anticoagulation with either rivaroxaban or apixaban. However, the assays must be specifically calibrated for each drug for results to be useful. (Anti-factor Xa assays cannot be used for heparin or low-molecular-weight heparin.) Further, most institutions do not yet have this capability. When appropriately calibrated, normal anti-factor Xa levels would exclude any effect of rivaroxaban or apixaban.

Reversal of anticoagulation

If patients on TSOACs require emergency surgery or present with significant bleeding in the setting of persistent anticoagulation, it may be necessary to try to reverse the anticoagulation.

Unlike warfarin or heparin, TSOACS do not have specific reversal agents, though specific antidotes are being developed. For example, researchers are evaluating antibodies capable of neutralizing dabigatran, as well as recombinant thrombin and factor Xa molecules that could antagonize dabigatran and rivaroxaban, respectively.^51–53

Reversal can be attempted by neutralizing or removing the offending drug. Activated charcoal may be able to reduce absorption of TSOACs that were recently ingested,⁴⁴ and dabigatran can be removed by hemodialysis.

However, certain practical considerations may limit the use of dialysis in the perioperative period. Insertion of a temporary dialysis line in an anticoagulated patient poses additional bleeding risks. A standard 4-hour hemodialysis session may remove only about 70% of dabigatran from the plasma, which may not be enough to prevent perioperative bleeding.⁵⁴ Dabigatran also tends to redistribute from adipose tissue back into plasma after each dialysis session.⁵⁵ Serial sessions of high-flux intermittent hemodialysis or continuous renal replacement therapy may therefore be needed to counteract rebound elevations in the dabigatran concentration.

Reversal can also be attempted through activation of the coagulation cascade via other mechanisms. Fresh-frozen plasma is unlikely to be a practical solution for reversal.⁴⁴ Although it can readily replace the clotting factors depleted by vitamin K antagonists, large volumes of fresh-frozen plasma would be needed to overwhelm thrombin or factor Xa inhibition by TSOACs.

There are limited data on the use of prothrombin complex concentrates or recombinant activated factor VIIa in patients on TSOACs, though their use can be considered.⁵⁶ In a trial in 12 healthy participants, a nonactivated four-factor prothrombin complex concentrate containing factors II, VII, IX, and X immediately and completely reversed the anticoagulant effect of rivaroxaban but had no effect on dabigatran.⁵⁷ Before 2013, there were no nonactivated four-factor prothrombin complex concentrates available in the United States. The FDA has since approved Kcentra for the urgent reversal of vitamin K antagonists, meaning that the reversal of TSOACs in major bleeding events would still be off-label.⁵⁸ Giving any of the clotting factors carries a risk of thromboembolism.

Resumption of anticoagulation

TSOACs have a rapid onset of action, and therapeutic levels are reached within a few hours of administration.

Extrapolating from the ACCP guidelines, TSOACs can generally be restarted at therapeutic doses 24 hours after low-bleeding-risk procedures.² Therapeutic dosing should be delayed 48 to 72 hours after a procedure with a high bleeding risk, assuming adequate hemostasis has been achieved. Prophylactic unfractionated heparin or low-molecular-weight heparin therapy can be given in the interim if deemed safe. Alternatively, for orthopedic patients ultimately transitioning back to therapeutic rivaroxaban after hip or knee arthroplasty, prophylactic rivaroxaban doses can be started 6 to 10 hours after surgery.¹⁴

There are numerous reasons why the resumption of TSOACs may have to be delayed after surgery, including nothing-by-mouth status, postoperative nausea and vomiting, ileus, gastric or bowel resection, and the anticipated need for future procedures. Since dabigatran capsules cannot be crushed, they cannot be given via nasogastric tube in patients with postoperative dysphagia. Parenteral anticoagulants should be used until these issues resolve.

Unfractionated heparin is still the preferred anticoagulant in unstable or potentially unstable patients, given its ease of monitoring, quick offset of action, and reversibility. When patients have stabilized, TSOACs can be resumed when the next dose of low-molecular-weight heparin would have been due or when the unfractionated heparin drip is discontinued.^3,14,23

UNTIL EVIDENCE-BASED GUIDELINES ARE DEVELOPED

The development of TSOACs has ushered in an exciting new era for anticoagulant therapy. Providers involved in perioperative medicine will increasingly encounter patients on dabigatran, rivaroxaban, and apixaban. However, until evidence-based guidelines are developed for these new anticoagulants, clinicians will have to apply their knowledge of pharmacology and critically evaluate expert recommendations in order to manage patients safely throughout the perioperative period.

More then 2.5 million patients in the United States are on long-term anticoagulation therapy for atrial fibrillation, venous thromboembolic disease, or mechanical heart valves,¹ and the number is expected to rise as the population ages. Each year, about 10% of these patients undergo an invasive procedure or surgery that requires temporary interruption of anticoagulation.²

Most physicians are familiar with the perioperative management of warfarin, a vitamin K antagonist, since for decades it has been the sole oral anticoagulant available. However, many physicians lack experience with the three target-specific oral anticoagulants (TSOACs; also known as “novel” oral anticoagulants) approved so far: the direct thrombin inhibitor dabigatran (Pradaxa) and the direct factor Xa inhibitors rivaroxaban (Xarelto) and apixaban (Eliquis).

With their rapid onset of action, predictable pharmacokinetics, relatively short half-lives, and fewer drug-drug interactions than warfarin, TSOACs overcome many of the limitations of the older oral anticoagulant warfarin. In many ways, these qualities simplify the perioperative management of anticoagulation. At the same time, these new drugs also bring new challenges: caution is needed in patients with renal impairment; the level of anticoagulation is difficult to assess; and there is no specific antidote or standardized procedure to reverse their anticoagulant effect. While various periprocedural protocols for TSOAC therapy have been proposed, evidence-based guidelines are still to come.

This article first discusses the pharmacology of dabigatran, rivaroxaban, and apixaban that is pertinent to the perioperative period. It then briefly reviews the general principles of perioperative management of anticoagulation. The final section provides specific recommendations for the perioperative management of TSOACs.

PHARMACOLOGY OF TARGET-SPECIFIC ORAL ANTICOAGULANTS

Dabigatran, a factor IIa inhibitor

Information from references 3, 4, 14, and 23.

Dabigatran is an oral direct thrombin (factor IIa) inhibitor. It exerts its anticoagulant effect by blocking the generation of fibrin, inhibiting platelet aggregation, and dampening the activity of factors V, VIII, and XI (Figure 1).^3,4 From its introduction in October 2010 through August 2012, nearly 3.7 million prescriptions were dispensed to 725,000 patients in the United States.⁵

Indications for dabigatran. Dabigatran is approved in the United States and Canada for preventing stroke in nonvalvular atrial fibrillation (Table 1).⁶ More recently, it received US approval for treating deep vein thrombosis or pulmonary embolism after 5 to 10 days of a parenteral anticoagulant.^7,8 It is also approved in Europe and Canada for preventing venous thromboembolism (VTE) after total hip replacement and knee arthroplasty.^9,10

Dabigatran is contraindicated in patients with a mechanical heart valve, based on a phase 2 study in which it conferred a higher risk of thromboembolism and bleeding than warfarin.^3,11

Pharmacokinetics of dabigatran. Dabigatran is formulated as a prodrug, dabigatran etexilate, in a capsule containing multiple small pellets.¹² The capsules should not be crushed, as this significantly increases oral bioavailability. The prodrug is absorbed across the gastric mucosa and is then rapidly converted to the active form (Table 2).

Plasma concentrations peak within 2 hours of ingestion, which means that therapeutic anticoagulation is achieved shortly after taking the drug.

Only 35% of dabigatran is protein-bound, which allows it to be removed by hemodialysis. Nearly 85% of the drug is eliminated in the urine. It has a half-life of 13 to 15 hours in patients with normal renal function.³ However, its half-life increases to about 27 hours in patients whose creatinine clearance is less than 30 mL/min. As a result, the dose must be reduced in patients with renal impairment (Table 1).

Dabigatran is not metabolized by the cytochrome P450 enzymes, but it is a substrate for P-glycoprotein, so it still has the potential for drug-drug interactions.³ Practitioners should be familiar with these potential interactions (Table 3), as they can result in higher- or lower-than-expected plasma concentrations of dabigatran in the perioperative period.¹³

Rivaroxaban, a factor Xa inhibitor

Rivaroxaban is an oral direct factor Xa inhibitor. It has been approved by the US Food and Drug Administration (FDA) for the prevention of stroke in nonvalvular atrial fibrillation, for VTE treatment, and for VTE prophylaxis after hip replacement or knee replacement (Table 1).^14–20 It has not yet been studied in patients with hip fracture.

Pharmacokinetics of rivaroxaban. Rivaroxaban is manufactured as a tablet that is best absorbed in the stomach (Table 2).¹⁴ In contrast to dabigatran, it can be crushed and, for example, mixed with applesauce for patients who have trouble swallowing. It can also be mixed with water and given via nasogastric tube; however, postpyloric administration should be avoided.

Plasma concentrations peak within a few hours after ingestion. Rivaroxaban is highly protein-bound, so it cannot be eliminated by hemodialysis.

The drug relies on renal elimination to a smaller degree than dabigatran, with one-third of the dose eliminated unchanged in the urine, one-third eliminated in the urine as inactive metabolite, and the remaining one-third eliminated in the feces. However, enough parent compound is cleared through the kidneys that the half-life of rivaroxaban increases from 8.3 hours in healthy individuals to 9.5 hours in patients whose creatinine clearance is less than 30 mL/min.²¹ As with dabigatran, the dose must be adjusted for renal impairment (Table 1).

Rivaroxaban has significant liver metabolism, specifically through the cytochrome P450 3A4 enzyme, and it is also a substrate of P-glycoprotein. Therefore, potential drug-drug interactions must be taken into account, as they may lead to important alterations in plasma concentrations (Table 3).

Apixaban, a factor Xa inhibitor

Apixaban is also an oral direct factor Xa inhibitor. It is the newest of the oral anticoagulants to be approved in the United States, specifically for preventing stroke in nonvalvular atrial fibrillation (Table 1).²²

Pharmacokinetics of apixaban. Apixaban is produced as a tablet that is absorbed slowly through the gastrointestinal tract, mainly the distal small bowel and ascending colon (Table  2).²³

Peak plasma concentrations are reached a few hours after ingestion. Like rivaroxaban, apixaban is highly protein-bound, so it cannot be removed by hemodialysis.

Apixaban is similar to rivaroxaban in that 27% of the parent compound is cleared through the kidneys, it undergoes significant hepatic metabolism through cytochrome P450 3A4, and it is a substrate for P-glycoprotein.

Drug-drug interactions must be considered as a potential source of altered drug exposure and clearance (Table 3).

Unlike dabigatran and rivaroxaban, dose reduction is not based on the calculated creatinine clearance. Instead, a reduced dose is required if the patient meets two of the following three criteria:

• Serum creatinine level ≥ 1.5 mg/dL
• Age ≥ 80
• Weight ≤ 60 kg (Table 1).

The American Heart Association/American Stroke Association guidelines further recommend against using apixaban in patients with a creatinine clearance less than 25 mL/min.²⁴

Edoxaban, a factor Xa inhibitor in development

Edoxaban (Savaysa), another factor Xa inhibitor, is available in Japan and has been submitted for approval in the United States for treating VTE and for preventing stroke in patients with

PERIOPERATIVE CONSIDERATIONS IN ANTICOAGULATION

Before addressing the perioperative management of TSOACs, let us review the evidence guiding the perioperative management of any chronic anticoagulant.

In fact, no large prospective randomized trial has clearly defined the risks and benefits of using or withholding a bridging anticoagulation strategy around surgery and other procedures, though the PERIOP 2 and BRIDGE trials are currently ongoing.^25,26 There are some data regarding continuing anticoagulation without interruption, but they have mainly been derived from specific groups (eg, patients on warfarin undergoing cardiac pacemaker or defibrillator placement) and in procedures that pose a very low risk of bleeding complications (eg, minor dental extractions, cataract surgery, dermatologic procedures).^2,27 Recommendations are, therefore, necessarily based on small perioperative trials and data gleaned from cohort review and from studies that did not involve surgical patients.

Ultimately, the decisions whether to discontinue oral anticoagulants and whether to employ bridging anticoagulation are based on assumptions about the risks of bleeding and the risk of thrombotic events, with similar assumptions regarding the effects of anticoagulants on both outcomes. In addition, the relative acceptance of bleeding vs thrombotic risks implicitly guides these complex decisions.

Perioperative bleeding risk

Many risk factors specific to the patient and to the type of surgery affect the rates and severity of perioperative bleeding.²⁸

As for patient-specific risk factors, a small retrospective cohort analysis revealed that a HAS-BLED score of 3 or higher was highly discriminating in predicting perioperative bleeding in atrial fibrillation patients receiving anticoagulation.²⁹ (The HAS-BLED score is based on hypertension, abnormal renal or liver function, stroke, bleeding, labile international normalized ratio [INR], elderly [age > 65] and drug therapy.³⁰) However, there are no widely validated tools that incorporate patient-specific factors to accurately predict bleeding risk in an individual patient.

Therefore, the American College of Chest Physicians (ACCP) guidelines suggest coarsely categorizing bleeding risk as either low or high solely on the basis of the type of procedure.² Procedures considered “high-risk” have a risk greater than 1.5% to 2% and include urologic surgery involving the prostate or kidney, colonic polyp resections, surgeries involving highly vascular organs such as the liver or spleen, joint replacements, cancer surgeries, and cardiac or neurosurgical procedures.

Perioperative thrombotic risk

The ACCP guidelines² place patients with atrial fibrillation, VTE, or mechanical heart valves in three risk groups for perioperative thromboembolism without anticoagulation, based on their annual risk of a thrombotic event:

• High risk—annual risk of a thrombotic event > 10%
• Moderate risk—5% to 10%
• Low risk—< 5%.

Comparing the risks calculated by these methods with the real-world risk of perioperative thrombosis highlights the problem of applying nonperioperative risk calculations: the perioperative period exposes patients to a higher risk than these models would predict.³¹ Nonetheless, these risk categorizations likely have some validity in stratifying patients into risk groups, even if the absolute risks are inaccurate.

Perioperative bridging for patients taking warfarin

Many patients with atrial fibrillation, VTE, or a mechanical heart valve need to interrupt their warfarin therapy because of the bleeding risk of an upcoming procedure.

The perioperative management of warfarin and other vitamin K antagonists is challenging because of the pharmacokinetics and pharmacodynamics of these drugs. Because it has a long half-life, warfarin usually must be stopped 4 to 5 days before a procedure in order to allow not only adequate clearance of the drug itself, but also restoration of functional clotting factors to normal or near-normal levels.¹² Warfarin can generally be resumed 12 to 24 hours after surgery, assuming adequate hemostasis has been achieved, and it will again take several days for the INR to reach the therapeutic range.

The ACCP guidelines recommend using the perioperative risk of thromboembolism to make decisions about the need for bridging anticoagulation during warfarin interruption.² They suggest that patients at high risk of thrombosis receive bridging with an alternative anticoagulant such as low-molecular-weight heparin or unfractionated heparin, because of the prolonged duration of subtherapeutic anticoagulation.

There has been clinical interest in using a TSOAC instead of low-molecular-weight or unfractionated heparin for bridging in the perioperative setting. Although this approach may be attractive from a cost and convenience perspective, it cannot be endorsed as yet because of the lack of information on the pros and cons of such an approach.

Patients at low thrombotic risk do not require bridging. In patients at moderate risk, the decision to bridge or not to bridge is based on careful consideration of patient-specific and surgery-specific factors.

PERIOPERATIVE MANAGEMENT OF TARGET-SPECIFIC ORAL ANTICOAGULANTS

As summarized above, the perioperative management strategy for chronic anticoagulation is based on limited evidence, even for drugs as well established as warfarin.

The most recent ACCP guidelines on the perioperative management of antithrombotic therapy do not mention TSOACs.² For now, the management strategy must be based on the pharmacokinetics of the drugs, package inserts from the manufacturers, and expert recommendations.^{3,14,23,32–34} Fortunately, because TSOACs have a more favorable pharmacokinetic profile than that of warfarin, their perioperative uses should be more streamlined. As always, the goal is to minimize the risk of both periprocedural bleeding and thromboembolism.

Timing of cessation of anticoagulation

The timing of cessation of TSOACs before an elective procedure depends primarily on two factors: the bleeding risk of the procedure and the patient’s renal function. Complete clearance of the medication is not necessary in all circumstances.

TSOACs should be stopped four to five half-lives before a procedure with a high bleeding risk, so that there is no or only minimal residual anticoagulant effect. The drug can be stopped two to three half-lives before a procedure with a low bleeding risk. Remember: the half-life increases as creatinine clearance decreases.

Specific recommendations may vary across institutions, but a suggested strategy is shown in Table 4.^{3,4,21,23,32–35} For the small subset of patients on P-glycoprotein or cytochrome P450 inhibitors or inducers, further adjustment in the time of discontinuation may be required.

Therapy does not need to be interrupted for procedures with a very low bleeding risk, as defined above.^33,34 There is also preliminary evidence that TSOACs, similar to warfarin, may be continued during cardiac pacemaker or defibrillator placement.³⁶

Evidence from clinical trials of perioperative TSOAC management

While the above recommendations are logical, studies are needed to prospectively evaluate perioperative management strategies.

The RE-LY trial (Randomized Evaluation of Long-Term Anticoagulation Therapy), which compared the effects of dabigatran and warfarin in preventing stroke in patients with atrial fibrillation, is one of the few clinical trials that also looked at periprocedural bleeding.³⁷ About a quarter of the RE-LY participants required interruption of anticoagulation for a procedure.

Warfarin was managed according to local practices. For most of the study, the protocol required that dabigatran be discontinued 24 hours before a procedure, regardless of renal function or procedure type. The protocol was later amended and closely mirrored the management plan outlined in Table 4.

With either protocol, there was no statistically significant difference between dabigatran and warfarin in the rates of bleeding and thrombotic complications in the 7 days before or 30 days after the procedure.

A major limitation of the study was that most patients underwent a procedure with a low bleeding risk, so the analysis was likely underpowered to evaluate rates of bleeding in higher-risk procedures.

The ROCKET-AF trial (Rivaroxaban Once-daily Oral Direct Factor Xa Inhibition Compared With Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation) also shed light on periprocedural bleeding.¹⁵ About 15% of the participants required temporary interruption of anticoagulation for a surgical or invasive procedure.³⁸

The study protocol called for discontinuing rivaroxaban 2 days before any procedure. Warfarin was to be held for 4 days to achieve a goal INR of 1.5 or less.¹⁵

Rates of major and nonmajor clinically significant bleeding at 30 days were similar with rivaroxaban and with warfarin.³⁸ As with the RE-LY trial, the retrospective analysis was probably underpowered for assessing rates of bleeding in procedures with higher risk.

Perioperative bridging

While stopping a TSOAC in the perioperative period decreases the risk of bleeding, it naturally increases the risk of thromboembolism. However, patients on TSOACs should not routinely require perioperative bridging with an alternative anticoagulant, regardless of thrombotic risk.

Of note, dabigatran, rivaroxaban, and apixaban carry black-box warnings that discontinuation places patients at higher risk of thrombotic events.^3,14,23 These warnings further state that coverage with an alternative anticoagulant should be strongly considered during interruption of therapy for reasons other than pathologic bleeding.

However, it does not necessarily follow that perioperative bridging is required. For example, the warning for rivaroxaban is based on the finding in the ROCKET-AF trial that patients in the rivaroxaban group had higher rates of stroke than those in the warfarin group after the study drugs were stopped at the end of the trial.³⁹ While there was initial concern that this could represent a prothrombotic rebound effect, the authors subsequently showed that patients in the rivaroxaban group were more likely to have had a subtherapeutic INR when transitioning to open-label vitamin-K-antagonist therapy.^39,40 There was no difference in the rate of stroke or systemic embolism between the rivaroxaban and warfarin groups when anticoagulation was temporarily interrupted for a procedure.³⁸

The risks and benefits of perioperative bridging with TSOACs are difficult to evaluate, given the dearth of trial data. In the RE-LY trial, only 17% of patients on dabigatran and 28% of patients on warfarin underwent periprocedural bridging.³⁷ The selection criteria and protocol for bridging were not reported. In the ROCKET-AF trial, only 9% of patients received bridging therapy despite a mean CHADS₂ score of 3.4.³⁸ (The CHADS₂ score is calculated as 1 point each for congestive heart failure, hypertension, age ≥ 75, and diabetes; 2 points for stroke or transient ischemic attack.) The decision to bridge or not was left to the individual investigator. As a result, the literature offers diverse opinions about the appropriateness of transitioning to an alternative anticoagulant.^41–43

Bridging does not make sense in most instances, since anticoagulants such as low-molecular-weight heparin have pharmacokinetics similar to those of the available TSOACs and also depend on renal clearance.⁴¹ However, there may be situations in which patients must be switched to a parenteral anticoagulant such as unfractionated or low-molecular-weight heparin. For example, if a TSOAC has to be held, the patient has acute renal failure, and a needed procedure is still several days away, it would be reasonable to start a heparin drip for an inpatient at increased thrombotic risk.

In patients with normal renal function, these alternative anticoagulants should be started at the time the next TSOAC dose would have been due.^3,14,23 In patients with reduced renal function, initiation of an alternative anticoagulant may need to be delayed 12 to 48 hours depending on which TSOAC is being used, as well as on the degree of renal dysfunction. This delay would help ensure that the onset of anticoagulation with the alternative anticoagulant is timed with the offset of therapeutic anticoagulation with the TSOAC.

Although limited, information from available coagulation assays may assist with the timing of initiation of an alternative anticoagulant (see the following section on laboratory monitoring). Serial testing with appropriate coagulation assays may help identify when most of a TSOAC has been cleared from a patient.

Laboratory monitoring

Inevitably, some patients on TSOACs require urgent or emergency surgery. In certain situations, such as before an orthopedic spine procedure, in which the complications of bleeding could be devastating, it may be necessary to know if any residual anticoagulant effect is present.

Monitoring dabigatran. As one might expect, direct thrombin inhibitors such as dabigatran can prolong the prothrombin time and activated partial thromboplastin time (aPTT).^44–47 However, the prothrombin time is not recommended for assessing the level of anticoagulation from dabigatran. Many institutions may be using a normal aPTT to rule out therapeutic concentrations of dabigatran, based on results from early in vitro and ex vivo studies.⁴⁶ While appealing from a practical standpoint, practitioners should exercise caution when relying on the aPTT to assess the risk of perioperative bleeding. A more recent investigation in patients treated with dabigatran found that up to 35% of patients with a normal aPTT still had a plasma concentration in the therapeutic range.⁴⁸

The thrombin time and ecarin clotting time are more sensitive tests for dabigatran. A normal thrombin time or ecarin clotting time indicates that no or only minimal dabigatran is present.⁴⁸ Unfortunately, these two tests often are either unavailable or are associated with long turnaround times, which limits their usefulness in the perioperative setting.

Monitoring rivaroxaban and apixaban. Factor Xa inhibitors such as rivaroxaban and apixaban can also influence the prothrombin time and aPTT (Figure 1).^{44–47,49,50} The aPTT is relatively insensitive to these drugs at low concentrations. It has been suggested that a normal prothrombin time can reasonably exclude therapeutic concentrations of rivaroxaban.^45,46 However, the effects on the prothrombin time are highly variable, changing with the reagent used.^49,50 In addition, apixaban appears to have less impact on the prothrombin time overall. The INR is not recommended for monitoring the effect of factor Xa inhibitors.

Anti-factor Xa assays likely represent the best option to provide true quantitative information on the level of anticoagulation with either rivaroxaban or apixaban. However, the assays must be specifically calibrated for each drug for results to be useful. (Anti-factor Xa assays cannot be used for heparin or low-molecular-weight heparin.) Further, most institutions do not yet have this capability. When appropriately calibrated, normal anti-factor Xa levels would exclude any effect of rivaroxaban or apixaban.

Reversal of anticoagulation

If patients on TSOACs require emergency surgery or present with significant bleeding in the setting of persistent anticoagulation, it may be necessary to try to reverse the anticoagulation.

Unlike warfarin or heparin, TSOACS do not have specific reversal agents, though specific antidotes are being developed. For example, researchers are evaluating antibodies capable of neutralizing dabigatran, as well as recombinant thrombin and factor Xa molecules that could antagonize dabigatran and rivaroxaban, respectively.^51–53

Reversal can be attempted by neutralizing or removing the offending drug. Activated charcoal may be able to reduce absorption of TSOACs that were recently ingested,⁴⁴ and dabigatran can be removed by hemodialysis.

However, certain practical considerations may limit the use of dialysis in the perioperative period. Insertion of a temporary dialysis line in an anticoagulated patient poses additional bleeding risks. A standard 4-hour hemodialysis session may remove only about 70% of dabigatran from the plasma, which may not be enough to prevent perioperative bleeding.⁵⁴ Dabigatran also tends to redistribute from adipose tissue back into plasma after each dialysis session.⁵⁵ Serial sessions of high-flux intermittent hemodialysis or continuous renal replacement therapy may therefore be needed to counteract rebound elevations in the dabigatran concentration.

Reversal can also be attempted through activation of the coagulation cascade via other mechanisms. Fresh-frozen plasma is unlikely to be a practical solution for reversal.⁴⁴ Although it can readily replace the clotting factors depleted by vitamin K antagonists, large volumes of fresh-frozen plasma would be needed to overwhelm thrombin or factor Xa inhibition by TSOACs.

There are limited data on the use of prothrombin complex concentrates or recombinant activated factor VIIa in patients on TSOACs, though their use can be considered.⁵⁶ In a trial in 12 healthy participants, a nonactivated four-factor prothrombin complex concentrate containing factors II, VII, IX, and X immediately and completely reversed the anticoagulant effect of rivaroxaban but had no effect on dabigatran.⁵⁷ Before 2013, there were no nonactivated four-factor prothrombin complex concentrates available in the United States. The FDA has since approved Kcentra for the urgent reversal of vitamin K antagonists, meaning that the reversal of TSOACs in major bleeding events would still be off-label.⁵⁸ Giving any of the clotting factors carries a risk of thromboembolism.

Resumption of anticoagulation

TSOACs have a rapid onset of action, and therapeutic levels are reached within a few hours of administration.

Extrapolating from the ACCP guidelines, TSOACs can generally be restarted at therapeutic doses 24 hours after low-bleeding-risk procedures.² Therapeutic dosing should be delayed 48 to 72 hours after a procedure with a high bleeding risk, assuming adequate hemostasis has been achieved. Prophylactic unfractionated heparin or low-molecular-weight heparin therapy can be given in the interim if deemed safe. Alternatively, for orthopedic patients ultimately transitioning back to therapeutic rivaroxaban after hip or knee arthroplasty, prophylactic rivaroxaban doses can be started 6 to 10 hours after surgery.¹⁴

There are numerous reasons why the resumption of TSOACs may have to be delayed after surgery, including nothing-by-mouth status, postoperative nausea and vomiting, ileus, gastric or bowel resection, and the anticipated need for future procedures. Since dabigatran capsules cannot be crushed, they cannot be given via nasogastric tube in patients with postoperative dysphagia. Parenteral anticoagulants should be used until these issues resolve.

Unfractionated heparin is still the preferred anticoagulant in unstable or potentially unstable patients, given its ease of monitoring, quick offset of action, and reversibility. When patients have stabilized, TSOACs can be resumed when the next dose of low-molecular-weight heparin would have been due or when the unfractionated heparin drip is discontinued.^3,14,23

UNTIL EVIDENCE-BASED GUIDELINES ARE DEVELOPED

The development of TSOACs has ushered in an exciting new era for anticoagulant therapy. Providers involved in perioperative medicine will increasingly encounter patients on dabigatran, rivaroxaban, and apixaban. However, until evidence-based guidelines are developed for these new anticoagulants, clinicians will have to apply their knowledge of pharmacology and critically evaluate expert recommendations in order to manage patients safely throughout the perioperative period.

More then 2.5 million patients in the United States are on long-term anticoagulation therapy for atrial fibrillation, venous thromboembolic disease, or mechanical heart valves,¹ and the number is expected to rise as the population ages. Each year, about 10% of these patients undergo an invasive procedure or surgery that requires temporary interruption of anticoagulation.²

Most physicians are familiar with the perioperative management of warfarin, a vitamin K antagonist, since for decades it has been the sole oral anticoagulant available. However, many physicians lack experience with the three target-specific oral anticoagulants (TSOACs; also known as “novel” oral anticoagulants) approved so far: the direct thrombin inhibitor dabigatran (Pradaxa) and the direct factor Xa inhibitors rivaroxaban (Xarelto) and apixaban (Eliquis).

With their rapid onset of action, predictable pharmacokinetics, relatively short half-lives, and fewer drug-drug interactions than warfarin, TSOACs overcome many of the limitations of the older oral anticoagulant warfarin. In many ways, these qualities simplify the perioperative management of anticoagulation. At the same time, these new drugs also bring new challenges: caution is needed in patients with renal impairment; the level of anticoagulation is difficult to assess; and there is no specific antidote or standardized procedure to reverse their anticoagulant effect. While various periprocedural protocols for TSOAC therapy have been proposed, evidence-based guidelines are still to come.

This article first discusses the pharmacology of dabigatran, rivaroxaban, and apixaban that is pertinent to the perioperative period. It then briefly reviews the general principles of perioperative management of anticoagulation. The final section provides specific recommendations for the perioperative management of TSOACs.

PHARMACOLOGY OF TARGET-SPECIFIC ORAL ANTICOAGULANTS

Dabigatran, a factor IIa inhibitor

Information from references 3, 4, 14, and 23.

Dabigatran is an oral direct thrombin (factor IIa) inhibitor. It exerts its anticoagulant effect by blocking the generation of fibrin, inhibiting platelet aggregation, and dampening the activity of factors V, VIII, and XI (Figure 1).^3,4 From its introduction in October 2010 through August 2012, nearly 3.7 million prescriptions were dispensed to 725,000 patients in the United States.⁵

Indications for dabigatran. Dabigatran is approved in the United States and Canada for preventing stroke in nonvalvular atrial fibrillation (Table 1).⁶ More recently, it received US approval for treating deep vein thrombosis or pulmonary embolism after 5 to 10 days of a parenteral anticoagulant.^7,8 It is also approved in Europe and Canada for preventing venous thromboembolism (VTE) after total hip replacement and knee arthroplasty.^9,10

Dabigatran is contraindicated in patients with a mechanical heart valve, based on a phase 2 study in which it conferred a higher risk of thromboembolism and bleeding than warfarin.^3,11

Pharmacokinetics of dabigatran. Dabigatran is formulated as a prodrug, dabigatran etexilate, in a capsule containing multiple small pellets.¹² The capsules should not be crushed, as this significantly increases oral bioavailability. The prodrug is absorbed across the gastric mucosa and is then rapidly converted to the active form (Table 2).

Plasma concentrations peak within 2 hours of ingestion, which means that therapeutic anticoagulation is achieved shortly after taking the drug.

Only 35% of dabigatran is protein-bound, which allows it to be removed by hemodialysis. Nearly 85% of the drug is eliminated in the urine. It has a half-life of 13 to 15 hours in patients with normal renal function.³ However, its half-life increases to about 27 hours in patients whose creatinine clearance is less than 30 mL/min. As a result, the dose must be reduced in patients with renal impairment (Table 1).

Dabigatran is not metabolized by the cytochrome P450 enzymes, but it is a substrate for P-glycoprotein, so it still has the potential for drug-drug interactions.³ Practitioners should be familiar with these potential interactions (Table 3), as they can result in higher- or lower-than-expected plasma concentrations of dabigatran in the perioperative period.¹³

Rivaroxaban, a factor Xa inhibitor

Rivaroxaban is an oral direct factor Xa inhibitor. It has been approved by the US Food and Drug Administration (FDA) for the prevention of stroke in nonvalvular atrial fibrillation, for VTE treatment, and for VTE prophylaxis after hip replacement or knee replacement (Table 1).^14–20 It has not yet been studied in patients with hip fracture.

Pharmacokinetics of rivaroxaban. Rivaroxaban is manufactured as a tablet that is best absorbed in the stomach (Table 2).¹⁴ In contrast to dabigatran, it can be crushed and, for example, mixed with applesauce for patients who have trouble swallowing. It can also be mixed with water and given via nasogastric tube; however, postpyloric administration should be avoided.

Plasma concentrations peak within a few hours after ingestion. Rivaroxaban is highly protein-bound, so it cannot be eliminated by hemodialysis.

The drug relies on renal elimination to a smaller degree than dabigatran, with one-third of the dose eliminated unchanged in the urine, one-third eliminated in the urine as inactive metabolite, and the remaining one-third eliminated in the feces. However, enough parent compound is cleared through the kidneys that the half-life of rivaroxaban increases from 8.3 hours in healthy individuals to 9.5 hours in patients whose creatinine clearance is less than 30 mL/min.²¹ As with dabigatran, the dose must be adjusted for renal impairment (Table 1).

Rivaroxaban has significant liver metabolism, specifically through the cytochrome P450 3A4 enzyme, and it is also a substrate of P-glycoprotein. Therefore, potential drug-drug interactions must be taken into account, as they may lead to important alterations in plasma concentrations (Table 3).

Apixaban, a factor Xa inhibitor

Apixaban is also an oral direct factor Xa inhibitor. It is the newest of the oral anticoagulants to be approved in the United States, specifically for preventing stroke in nonvalvular atrial fibrillation (Table 1).²²

Pharmacokinetics of apixaban. Apixaban is produced as a tablet that is absorbed slowly through the gastrointestinal tract, mainly the distal small bowel and ascending colon (Table  2).²³

Peak plasma concentrations are reached a few hours after ingestion. Like rivaroxaban, apixaban is highly protein-bound, so it cannot be removed by hemodialysis.

Apixaban is similar to rivaroxaban in that 27% of the parent compound is cleared through the kidneys, it undergoes significant hepatic metabolism through cytochrome P450 3A4, and it is a substrate for P-glycoprotein.

Drug-drug interactions must be considered as a potential source of altered drug exposure and clearance (Table 3).

Unlike dabigatran and rivaroxaban, dose reduction is not based on the calculated creatinine clearance. Instead, a reduced dose is required if the patient meets two of the following three criteria:

• Serum creatinine level ≥ 1.5 mg/dL
• Age ≥ 80
• Weight ≤ 60 kg (Table 1).

The American Heart Association/American Stroke Association guidelines further recommend against using apixaban in patients with a creatinine clearance less than 25 mL/min.²⁴

Edoxaban, a factor Xa inhibitor in development

Edoxaban (Savaysa), another factor Xa inhibitor, is available in Japan and has been submitted for approval in the United States for treating VTE and for preventing stroke in patients with

PERIOPERATIVE CONSIDERATIONS IN ANTICOAGULATION

Before addressing the perioperative management of TSOACs, let us review the evidence guiding the perioperative management of any chronic anticoagulant.

In fact, no large prospective randomized trial has clearly defined the risks and benefits of using or withholding a bridging anticoagulation strategy around surgery and other procedures, though the PERIOP 2 and BRIDGE trials are currently ongoing.^25,26 There are some data regarding continuing anticoagulation without interruption, but they have mainly been derived from specific groups (eg, patients on warfarin undergoing cardiac pacemaker or defibrillator placement) and in procedures that pose a very low risk of bleeding complications (eg, minor dental extractions, cataract surgery, dermatologic procedures).^2,27 Recommendations are, therefore, necessarily based on small perioperative trials and data gleaned from cohort review and from studies that did not involve surgical patients.

Ultimately, the decisions whether to discontinue oral anticoagulants and whether to employ bridging anticoagulation are based on assumptions about the risks of bleeding and the risk of thrombotic events, with similar assumptions regarding the effects of anticoagulants on both outcomes. In addition, the relative acceptance of bleeding vs thrombotic risks implicitly guides these complex decisions.

Perioperative bleeding risk

Many risk factors specific to the patient and to the type of surgery affect the rates and severity of perioperative bleeding.²⁸

As for patient-specific risk factors, a small retrospective cohort analysis revealed that a HAS-BLED score of 3 or higher was highly discriminating in predicting perioperative bleeding in atrial fibrillation patients receiving anticoagulation.²⁹ (The HAS-BLED score is based on hypertension, abnormal renal or liver function, stroke, bleeding, labile international normalized ratio [INR], elderly [age > 65] and drug therapy.³⁰) However, there are no widely validated tools that incorporate patient-specific factors to accurately predict bleeding risk in an individual patient.

Therefore, the American College of Chest Physicians (ACCP) guidelines suggest coarsely categorizing bleeding risk as either low or high solely on the basis of the type of procedure.² Procedures considered “high-risk” have a risk greater than 1.5% to 2% and include urologic surgery involving the prostate or kidney, colonic polyp resections, surgeries involving highly vascular organs such as the liver or spleen, joint replacements, cancer surgeries, and cardiac or neurosurgical procedures.

Perioperative thrombotic risk

The ACCP guidelines² place patients with atrial fibrillation, VTE, or mechanical heart valves in three risk groups for perioperative thromboembolism without anticoagulation, based on their annual risk of a thrombotic event:

• High risk—annual risk of a thrombotic event > 10%
• Moderate risk—5% to 10%
• Low risk—< 5%.

Comparing the risks calculated by these methods with the real-world risk of perioperative thrombosis highlights the problem of applying nonperioperative risk calculations: the perioperative period exposes patients to a higher risk than these models would predict.³¹ Nonetheless, these risk categorizations likely have some validity in stratifying patients into risk groups, even if the absolute risks are inaccurate.

Perioperative bridging for patients taking warfarin

Many patients with atrial fibrillation, VTE, or a mechanical heart valve need to interrupt their warfarin therapy because of the bleeding risk of an upcoming procedure.

The perioperative management of warfarin and other vitamin K antagonists is challenging because of the pharmacokinetics and pharmacodynamics of these drugs. Because it has a long half-life, warfarin usually must be stopped 4 to 5 days before a procedure in order to allow not only adequate clearance of the drug itself, but also restoration of functional clotting factors to normal or near-normal levels.¹² Warfarin can generally be resumed 12 to 24 hours after surgery, assuming adequate hemostasis has been achieved, and it will again take several days for the INR to reach the therapeutic range.

The ACCP guidelines recommend using the perioperative risk of thromboembolism to make decisions about the need for bridging anticoagulation during warfarin interruption.² They suggest that patients at high risk of thrombosis receive bridging with an alternative anticoagulant such as low-molecular-weight heparin or unfractionated heparin, because of the prolonged duration of subtherapeutic anticoagulation.

There has been clinical interest in using a TSOAC instead of low-molecular-weight or unfractionated heparin for bridging in the perioperative setting. Although this approach may be attractive from a cost and convenience perspective, it cannot be endorsed as yet because of the lack of information on the pros and cons of such an approach.

Patients at low thrombotic risk do not require bridging. In patients at moderate risk, the decision to bridge or not to bridge is based on careful consideration of patient-specific and surgery-specific factors.

PERIOPERATIVE MANAGEMENT OF TARGET-SPECIFIC ORAL ANTICOAGULANTS

As summarized above, the perioperative management strategy for chronic anticoagulation is based on limited evidence, even for drugs as well established as warfarin.

The most recent ACCP guidelines on the perioperative management of antithrombotic therapy do not mention TSOACs.² For now, the management strategy must be based on the pharmacokinetics of the drugs, package inserts from the manufacturers, and expert recommendations.^{3,14,23,32–34} Fortunately, because TSOACs have a more favorable pharmacokinetic profile than that of warfarin, their perioperative uses should be more streamlined. As always, the goal is to minimize the risk of both periprocedural bleeding and thromboembolism.

Timing of cessation of anticoagulation

The timing of cessation of TSOACs before an elective procedure depends primarily on two factors: the bleeding risk of the procedure and the patient’s renal function. Complete clearance of the medication is not necessary in all circumstances.

TSOACs should be stopped four to five half-lives before a procedure with a high bleeding risk, so that there is no or only minimal residual anticoagulant effect. The drug can be stopped two to three half-lives before a procedure with a low bleeding risk. Remember: the half-life increases as creatinine clearance decreases.

Specific recommendations may vary across institutions, but a suggested strategy is shown in Table 4.^{3,4,21,23,32–35} For the small subset of patients on P-glycoprotein or cytochrome P450 inhibitors or inducers, further adjustment in the time of discontinuation may be required.

Therapy does not need to be interrupted for procedures with a very low bleeding risk, as defined above.^33,34 There is also preliminary evidence that TSOACs, similar to warfarin, may be continued during cardiac pacemaker or defibrillator placement.³⁶

Evidence from clinical trials of perioperative TSOAC management

While the above recommendations are logical, studies are needed to prospectively evaluate perioperative management strategies.

The RE-LY trial (Randomized Evaluation of Long-Term Anticoagulation Therapy), which compared the effects of dabigatran and warfarin in preventing stroke in patients with atrial fibrillation, is one of the few clinical trials that also looked at periprocedural bleeding.³⁷ About a quarter of the RE-LY participants required interruption of anticoagulation for a procedure.

Warfarin was managed according to local practices. For most of the study, the protocol required that dabigatran be discontinued 24 hours before a procedure, regardless of renal function or procedure type. The protocol was later amended and closely mirrored the management plan outlined in Table 4.

With either protocol, there was no statistically significant difference between dabigatran and warfarin in the rates of bleeding and thrombotic complications in the 7 days before or 30 days after the procedure.

A major limitation of the study was that most patients underwent a procedure with a low bleeding risk, so the analysis was likely underpowered to evaluate rates of bleeding in higher-risk procedures.

The ROCKET-AF trial (Rivaroxaban Once-daily Oral Direct Factor Xa Inhibition Compared With Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation) also shed light on periprocedural bleeding.¹⁵ About 15% of the participants required temporary interruption of anticoagulation for a surgical or invasive procedure.³⁸

The study protocol called for discontinuing rivaroxaban 2 days before any procedure. Warfarin was to be held for 4 days to achieve a goal INR of 1.5 or less.¹⁵

Rates of major and nonmajor clinically significant bleeding at 30 days were similar with rivaroxaban and with warfarin.³⁸ As with the RE-LY trial, the retrospective analysis was probably underpowered for assessing rates of bleeding in procedures with higher risk.

Perioperative bridging

While stopping a TSOAC in the perioperative period decreases the risk of bleeding, it naturally increases the risk of thromboembolism. However, patients on TSOACs should not routinely require perioperative bridging with an alternative anticoagulant, regardless of thrombotic risk.

Of note, dabigatran, rivaroxaban, and apixaban carry black-box warnings that discontinuation places patients at higher risk of thrombotic events.^3,14,23 These warnings further state that coverage with an alternative anticoagulant should be strongly considered during interruption of therapy for reasons other than pathologic bleeding.

However, it does not necessarily follow that perioperative bridging is required. For example, the warning for rivaroxaban is based on the finding in the ROCKET-AF trial that patients in the rivaroxaban group had higher rates of stroke than those in the warfarin group after the study drugs were stopped at the end of the trial.³⁹ While there was initial concern that this could represent a prothrombotic rebound effect, the authors subsequently showed that patients in the rivaroxaban group were more likely to have had a subtherapeutic INR when transitioning to open-label vitamin-K-antagonist therapy.^39,40 There was no difference in the rate of stroke or systemic embolism between the rivaroxaban and warfarin groups when anticoagulation was temporarily interrupted for a procedure.³⁸

The risks and benefits of perioperative bridging with TSOACs are difficult to evaluate, given the dearth of trial data. In the RE-LY trial, only 17% of patients on dabigatran and 28% of patients on warfarin underwent periprocedural bridging.³⁷ The selection criteria and protocol for bridging were not reported. In the ROCKET-AF trial, only 9% of patients received bridging therapy despite a mean CHADS₂ score of 3.4.³⁸ (The CHADS₂ score is calculated as 1 point each for congestive heart failure, hypertension, age ≥ 75, and diabetes; 2 points for stroke or transient ischemic attack.) The decision to bridge or not was left to the individual investigator. As a result, the literature offers diverse opinions about the appropriateness of transitioning to an alternative anticoagulant.^41–43

Bridging does not make sense in most instances, since anticoagulants such as low-molecular-weight heparin have pharmacokinetics similar to those of the available TSOACs and also depend on renal clearance.⁴¹ However, there may be situations in which patients must be switched to a parenteral anticoagulant such as unfractionated or low-molecular-weight heparin. For example, if a TSOAC has to be held, the patient has acute renal failure, and a needed procedure is still several days away, it would be reasonable to start a heparin drip for an inpatient at increased thrombotic risk.

In patients with normal renal function, these alternative anticoagulants should be started at the time the next TSOAC dose would have been due.^3,14,23 In patients with reduced renal function, initiation of an alternative anticoagulant may need to be delayed 12 to 48 hours depending on which TSOAC is being used, as well as on the degree of renal dysfunction. This delay would help ensure that the onset of anticoagulation with the alternative anticoagulant is timed with the offset of therapeutic anticoagulation with the TSOAC.

Although limited, information from available coagulation assays may assist with the timing of initiation of an alternative anticoagulant (see the following section on laboratory monitoring). Serial testing with appropriate coagulation assays may help identify when most of a TSOAC has been cleared from a patient.

Laboratory monitoring

Inevitably, some patients on TSOACs require urgent or emergency surgery. In certain situations, such as before an orthopedic spine procedure, in which the complications of bleeding could be devastating, it may be necessary to know if any residual anticoagulant effect is present.

Monitoring dabigatran. As one might expect, direct thrombin inhibitors such as dabigatran can prolong the prothrombin time and activated partial thromboplastin time (aPTT).^44–47 However, the prothrombin time is not recommended for assessing the level of anticoagulation from dabigatran. Many institutions may be using a normal aPTT to rule out therapeutic concentrations of dabigatran, based on results from early in vitro and ex vivo studies.⁴⁶ While appealing from a practical standpoint, practitioners should exercise caution when relying on the aPTT to assess the risk of perioperative bleeding. A more recent investigation in patients treated with dabigatran found that up to 35% of patients with a normal aPTT still had a plasma concentration in the therapeutic range.⁴⁸

The thrombin time and ecarin clotting time are more sensitive tests for dabigatran. A normal thrombin time or ecarin clotting time indicates that no or only minimal dabigatran is present.⁴⁸ Unfortunately, these two tests often are either unavailable or are associated with long turnaround times, which limits their usefulness in the perioperative setting.

Monitoring rivaroxaban and apixaban. Factor Xa inhibitors such as rivaroxaban and apixaban can also influence the prothrombin time and aPTT (Figure 1).^{44–47,49,50} The aPTT is relatively insensitive to these drugs at low concentrations. It has been suggested that a normal prothrombin time can reasonably exclude therapeutic concentrations of rivaroxaban.^45,46 However, the effects on the prothrombin time are highly variable, changing with the reagent used.^49,50 In addition, apixaban appears to have less impact on the prothrombin time overall. The INR is not recommended for monitoring the effect of factor Xa inhibitors.

Anti-factor Xa assays likely represent the best option to provide true quantitative information on the level of anticoagulation with either rivaroxaban or apixaban. However, the assays must be specifically calibrated for each drug for results to be useful. (Anti-factor Xa assays cannot be used for heparin or low-molecular-weight heparin.) Further, most institutions do not yet have this capability. When appropriately calibrated, normal anti-factor Xa levels would exclude any effect of rivaroxaban or apixaban.

Reversal of anticoagulation

If patients on TSOACs require emergency surgery or present with significant bleeding in the setting of persistent anticoagulation, it may be necessary to try to reverse the anticoagulation.

Unlike warfarin or heparin, TSOACS do not have specific reversal agents, though specific antidotes are being developed. For example, researchers are evaluating antibodies capable of neutralizing dabigatran, as well as recombinant thrombin and factor Xa molecules that could antagonize dabigatran and rivaroxaban, respectively.^51–53

Reversal can be attempted by neutralizing or removing the offending drug. Activated charcoal may be able to reduce absorption of TSOACs that were recently ingested,⁴⁴ and dabigatran can be removed by hemodialysis.

However, certain practical considerations may limit the use of dialysis in the perioperative period. Insertion of a temporary dialysis line in an anticoagulated patient poses additional bleeding risks. A standard 4-hour hemodialysis session may remove only about 70% of dabigatran from the plasma, which may not be enough to prevent perioperative bleeding.⁵⁴ Dabigatran also tends to redistribute from adipose tissue back into plasma after each dialysis session.⁵⁵ Serial sessions of high-flux intermittent hemodialysis or continuous renal replacement therapy may therefore be needed to counteract rebound elevations in the dabigatran concentration.

Reversal can also be attempted through activation of the coagulation cascade via other mechanisms. Fresh-frozen plasma is unlikely to be a practical solution for reversal.⁴⁴ Although it can readily replace the clotting factors depleted by vitamin K antagonists, large volumes of fresh-frozen plasma would be needed to overwhelm thrombin or factor Xa inhibition by TSOACs.

There are limited data on the use of prothrombin complex concentrates or recombinant activated factor VIIa in patients on TSOACs, though their use can be considered.⁵⁶ In a trial in 12 healthy participants, a nonactivated four-factor prothrombin complex concentrate containing factors II, VII, IX, and X immediately and completely reversed the anticoagulant effect of rivaroxaban but had no effect on dabigatran.⁵⁷ Before 2013, there were no nonactivated four-factor prothrombin complex concentrates available in the United States. The FDA has since approved Kcentra for the urgent reversal of vitamin K antagonists, meaning that the reversal of TSOACs in major bleeding events would still be off-label.⁵⁸ Giving any of the clotting factors carries a risk of thromboembolism.

Resumption of anticoagulation

TSOACs have a rapid onset of action, and therapeutic levels are reached within a few hours of administration.

Extrapolating from the ACCP guidelines, TSOACs can generally be restarted at therapeutic doses 24 hours after low-bleeding-risk procedures.² Therapeutic dosing should be delayed 48 to 72 hours after a procedure with a high bleeding risk, assuming adequate hemostasis has been achieved. Prophylactic unfractionated heparin or low-molecular-weight heparin therapy can be given in the interim if deemed safe. Alternatively, for orthopedic patients ultimately transitioning back to therapeutic rivaroxaban after hip or knee arthroplasty, prophylactic rivaroxaban doses can be started 6 to 10 hours after surgery.¹⁴

There are numerous reasons why the resumption of TSOACs may have to be delayed after surgery, including nothing-by-mouth status, postoperative nausea and vomiting, ileus, gastric or bowel resection, and the anticipated need for future procedures. Since dabigatran capsules cannot be crushed, they cannot be given via nasogastric tube in patients with postoperative dysphagia. Parenteral anticoagulants should be used until these issues resolve.

Unfractionated heparin is still the preferred anticoagulant in unstable or potentially unstable patients, given its ease of monitoring, quick offset of action, and reversibility. When patients have stabilized, TSOACs can be resumed when the next dose of low-molecular-weight heparin would have been due or when the unfractionated heparin drip is discontinued.^3,14,23

UNTIL EVIDENCE-BASED GUIDELINES ARE DEVELOPED

The development of TSOACs has ushered in an exciting new era for anticoagulant therapy. Providers involved in perioperative medicine will increasingly encounter patients on dabigatran, rivaroxaban, and apixaban. However, until evidence-based guidelines are developed for these new anticoagulants, clinicians will have to apply their knowledge of pharmacology and critically evaluate expert recommendations in order to manage patients safely throughout the perioperative period.