Hospital Medicine

In the September issue of the Journal, Dr. Thomas Lansdale discussed the pressures on internists trying to teach and practice medicine in 2008. He concluded that the system doesn’t work for him or his patients, and he now practices internal medicine in an alternative venue. I asked for comments and solutions from our readers. I certainly got them.

You responded to the parts of Dr. Lansdale’s commentary that struck a personal chord. Almost all responders shared his frustration. Many wrote that the American payer system fails to appropriately reward internists and primary care providers and called for restructuring the Medicare and third-party payer systems. Some of you took umbrage at his contention that hospitals are not safe, and that health care delivery systems do not always place quality care above economic imperatives as new programs and “centers of excellence” are implemented. And some of you reacted to the issues of physician satisfaction and difficulties in providing quality care in hospitals regulated by multiple agencies that generate unfunded mandates, while the hospitals already require high numbers of patients in order to survive financially.

I recently did a stint as rheumatology consultant at my hospital, and Dr. Lansdale’s commentary was fresh in my mind. I noticed with satisfaction that the physicians and nurses were using foam antiseptic on their hands. I noted the new checks on verbal orders and a successful emphasis on preventing deep vein thrombosis and bedsores. But I also noted more patient hand-offs between house staff and faculty, and difficulty in finding doctors who actually knew the patient (or doctors that patients recognized as being responsible for their care).

The electronic medical record is legible and available from all over the hospital, and I could tell who signed the notes. But many notes were actually cut-and-pasted from earlier notes, and thus I couldn’t always be sure who actually had said what and when. Technology is not an immediate panacea for the problem of limited physician time!

The house staff “lab” in the hospital with its microscope was closed due to regulatory concerns; thus, there was no easy way to look at a freshly spun urine sample for evidence of glomerulonephritis. This turned out to be a detriment to effective patient care: urine samples sent to the regular laboratory (with the usual transportation delay) rarely if ever reveal cellular casts. But we found creative, if inefficient, ways to deal with this and other problems.

At the end of the day, I realized that I still enjoy my time in the hospital. Patients’ problems can be presented to house staff and students at the bedside and their diagnoses and therapies discussed in real time. Junior physicians can observe how senior physicians talk to patients and families, including the many ways we have learned to say “I don’t know,” and learn to appreciate the value of a well-directed physical examination. There is still a synergy and intellectual satisfaction in being one of a group of senior consultants discussing the care of a shared patient who has complex medical problems.

With rational and caring involvement, individual physicians can alter the trajectory of patient management and remain the primary patient advocates within a health care system that can’t always easily deliver the quality that everyone desires. Caring, patient-focused physicians must remain in charge of health care delivery, lest we pay attention only to the financial and regulatory problems.

Tom, I am older and even more cynical than I was when we roamed the hospital together every third night and never went home on our post-call day until the last laboratory result had been checked and the last transfusion had been given. We inefficiently examined every patient’s urine ourselves (even from those being admitted for cardiac catheterization), and we had to convince patients of the (apparent) need for the urgent 3 AM blood draw to evaluate their 100.5° fever before we prepped the area and drew the blood. We drew blood for sedimentation rates and checked rapid plasma reagins at every admission and checked for urinary light chains in everyone with an elevated creatinine level and anemia, “just to be sure.” We blindly placed Swan-Ganz catheters to monitor many hypotensive patients in the intensive care units, and we aspirated pleural effusions on the basis of our percussive examination. We talked to patients and accepted enormous individual responsibility for their care, but we were also frequently numbed by the overwhelming intensity of the training and the practice.

I am all too aware of the many forces that are eroding physician-patient relationships and that can corrupt patient care in the name of efficiency, financial necessity, marketing advantage, or regulatory compliance. Many of these forces I hope to help change. But I remain a hospital guy because I can still make a difference. I still feel honored that patients entrust their care to me as we attempt to navigate our evolving and, yes, sometimes treacherous medical system. Evading the crocodiles and fighting insurance companies are now in my job description.

In this issue we run two letters in response to Dr. Lansdale’s commentary. In December we will publish more letters, though due to space limitations some will be abridged. We plan to run full text of many of the letters online at www.ccjm.org in December.

In the September issue of the Journal, Dr. Thomas Lansdale discussed the pressures on internists trying to teach and practice medicine in 2008. He concluded that the system doesn’t work for him or his patients, and he now practices internal medicine in an alternative venue. I asked for comments and solutions from our readers. I certainly got them.

You responded to the parts of Dr. Lansdale’s commentary that struck a personal chord. Almost all responders shared his frustration. Many wrote that the American payer system fails to appropriately reward internists and primary care providers and called for restructuring the Medicare and third-party payer systems. Some of you took umbrage at his contention that hospitals are not safe, and that health care delivery systems do not always place quality care above economic imperatives as new programs and “centers of excellence” are implemented. And some of you reacted to the issues of physician satisfaction and difficulties in providing quality care in hospitals regulated by multiple agencies that generate unfunded mandates, while the hospitals already require high numbers of patients in order to survive financially.

I recently did a stint as rheumatology consultant at my hospital, and Dr. Lansdale’s commentary was fresh in my mind. I noticed with satisfaction that the physicians and nurses were using foam antiseptic on their hands. I noted the new checks on verbal orders and a successful emphasis on preventing deep vein thrombosis and bedsores. But I also noted more patient hand-offs between house staff and faculty, and difficulty in finding doctors who actually knew the patient (or doctors that patients recognized as being responsible for their care).

The electronic medical record is legible and available from all over the hospital, and I could tell who signed the notes. But many notes were actually cut-and-pasted from earlier notes, and thus I couldn’t always be sure who actually had said what and when. Technology is not an immediate panacea for the problem of limited physician time!

The house staff “lab” in the hospital with its microscope was closed due to regulatory concerns; thus, there was no easy way to look at a freshly spun urine sample for evidence of glomerulonephritis. This turned out to be a detriment to effective patient care: urine samples sent to the regular laboratory (with the usual transportation delay) rarely if ever reveal cellular casts. But we found creative, if inefficient, ways to deal with this and other problems.

At the end of the day, I realized that I still enjoy my time in the hospital. Patients’ problems can be presented to house staff and students at the bedside and their diagnoses and therapies discussed in real time. Junior physicians can observe how senior physicians talk to patients and families, including the many ways we have learned to say “I don’t know,” and learn to appreciate the value of a well-directed physical examination. There is still a synergy and intellectual satisfaction in being one of a group of senior consultants discussing the care of a shared patient who has complex medical problems.

With rational and caring involvement, individual physicians can alter the trajectory of patient management and remain the primary patient advocates within a health care system that can’t always easily deliver the quality that everyone desires. Caring, patient-focused physicians must remain in charge of health care delivery, lest we pay attention only to the financial and regulatory problems.

Tom, I am older and even more cynical than I was when we roamed the hospital together every third night and never went home on our post-call day until the last laboratory result had been checked and the last transfusion had been given. We inefficiently examined every patient’s urine ourselves (even from those being admitted for cardiac catheterization), and we had to convince patients of the (apparent) need for the urgent 3 AM blood draw to evaluate their 100.5° fever before we prepped the area and drew the blood. We drew blood for sedimentation rates and checked rapid plasma reagins at every admission and checked for urinary light chains in everyone with an elevated creatinine level and anemia, “just to be sure.” We blindly placed Swan-Ganz catheters to monitor many hypotensive patients in the intensive care units, and we aspirated pleural effusions on the basis of our percussive examination. We talked to patients and accepted enormous individual responsibility for their care, but we were also frequently numbed by the overwhelming intensity of the training and the practice.

I am all too aware of the many forces that are eroding physician-patient relationships and that can corrupt patient care in the name of efficiency, financial necessity, marketing advantage, or regulatory compliance. Many of these forces I hope to help change. But I remain a hospital guy because I can still make a difference. I still feel honored that patients entrust their care to me as we attempt to navigate our evolving and, yes, sometimes treacherous medical system. Evading the crocodiles and fighting insurance companies are now in my job description.

In this issue we run two letters in response to Dr. Lansdale’s commentary. In December we will publish more letters, though due to space limitations some will be abridged. We plan to run full text of many of the letters online at www.ccjm.org in December.

In the September issue of the Journal, Dr. Thomas Lansdale discussed the pressures on internists trying to teach and practice medicine in 2008. He concluded that the system doesn’t work for him or his patients, and he now practices internal medicine in an alternative venue. I asked for comments and solutions from our readers. I certainly got them.

You responded to the parts of Dr. Lansdale’s commentary that struck a personal chord. Almost all responders shared his frustration. Many wrote that the American payer system fails to appropriately reward internists and primary care providers and called for restructuring the Medicare and third-party payer systems. Some of you took umbrage at his contention that hospitals are not safe, and that health care delivery systems do not always place quality care above economic imperatives as new programs and “centers of excellence” are implemented. And some of you reacted to the issues of physician satisfaction and difficulties in providing quality care in hospitals regulated by multiple agencies that generate unfunded mandates, while the hospitals already require high numbers of patients in order to survive financially.

I recently did a stint as rheumatology consultant at my hospital, and Dr. Lansdale’s commentary was fresh in my mind. I noticed with satisfaction that the physicians and nurses were using foam antiseptic on their hands. I noted the new checks on verbal orders and a successful emphasis on preventing deep vein thrombosis and bedsores. But I also noted more patient hand-offs between house staff and faculty, and difficulty in finding doctors who actually knew the patient (or doctors that patients recognized as being responsible for their care).

The electronic medical record is legible and available from all over the hospital, and I could tell who signed the notes. But many notes were actually cut-and-pasted from earlier notes, and thus I couldn’t always be sure who actually had said what and when. Technology is not an immediate panacea for the problem of limited physician time!

The house staff “lab” in the hospital with its microscope was closed due to regulatory concerns; thus, there was no easy way to look at a freshly spun urine sample for evidence of glomerulonephritis. This turned out to be a detriment to effective patient care: urine samples sent to the regular laboratory (with the usual transportation delay) rarely if ever reveal cellular casts. But we found creative, if inefficient, ways to deal with this and other problems.

At the end of the day, I realized that I still enjoy my time in the hospital. Patients’ problems can be presented to house staff and students at the bedside and their diagnoses and therapies discussed in real time. Junior physicians can observe how senior physicians talk to patients and families, including the many ways we have learned to say “I don’t know,” and learn to appreciate the value of a well-directed physical examination. There is still a synergy and intellectual satisfaction in being one of a group of senior consultants discussing the care of a shared patient who has complex medical problems.

With rational and caring involvement, individual physicians can alter the trajectory of patient management and remain the primary patient advocates within a health care system that can’t always easily deliver the quality that everyone desires. Caring, patient-focused physicians must remain in charge of health care delivery, lest we pay attention only to the financial and regulatory problems.

Tom, I am older and even more cynical than I was when we roamed the hospital together every third night and never went home on our post-call day until the last laboratory result had been checked and the last transfusion had been given. We inefficiently examined every patient’s urine ourselves (even from those being admitted for cardiac catheterization), and we had to convince patients of the (apparent) need for the urgent 3 AM blood draw to evaluate their 100.5° fever before we prepped the area and drew the blood. We drew blood for sedimentation rates and checked rapid plasma reagins at every admission and checked for urinary light chains in everyone with an elevated creatinine level and anemia, “just to be sure.” We blindly placed Swan-Ganz catheters to monitor many hypotensive patients in the intensive care units, and we aspirated pleural effusions on the basis of our percussive examination. We talked to patients and accepted enormous individual responsibility for their care, but we were also frequently numbed by the overwhelming intensity of the training and the practice.

I am all too aware of the many forces that are eroding physician-patient relationships and that can corrupt patient care in the name of efficiency, financial necessity, marketing advantage, or regulatory compliance. Many of these forces I hope to help change. But I remain a hospital guy because I can still make a difference. I still feel honored that patients entrust their care to me as we attempt to navigate our evolving and, yes, sometimes treacherous medical system. Evading the crocodiles and fighting insurance companies are now in my job description.

In this issue we run two letters in response to Dr. Lansdale’s commentary. In December we will publish more letters, though due to space limitations some will be abridged. We plan to run full text of many of the letters online at www.ccjm.org in December.

A 22-year-old African American man with sickle cell disease comes to the in his joints and chest—a presentation similar to those of his previous sickle cell crises. He is given intravenous fluids for dehydration and morphine sulfate for pain via a peripheral line, and he is admitted to the hospital.

Shortly afterward, the peripheral line becomes infiltrated. After failed attempts at peripheral cannulation, central venous cannulation via an internal jugular site is considered.

The patient alerts us that he has had multiple “neck lines” in the past and that these had been difficult to place. With this in mind, we attempt to place a triple-lumen catheter under ultrasonographic guidance and with the use of sterile precautions and the Seldinger technique. On the first attempt, the guidewire cannot be advanced beyond 4 cm, and the attempt is terminated. On the second attempt, the guidewire advances freely, but as the catheter is advanced, slight resistance is felt at 4 cm and again at 10 cm. This resistance is overcome with slight pressure, and subsequent advancement meets with no further resistance. After confirming nonpulsatile blood return in all three lumens, we suture the catheter at 14 cm from the insertion site. A chest radiograph (Figure 1) is requested to confirm placement.

WHERE IS THE CATHETER TIP?

At first look, the catheter appears to broadly follow an expected trajectory. However, a closer look shows that the catheter is not properly positioned: although it is difficult to see, the tip appears to project beyond the main carina (see arrow), an important landmark to identify catheter tip placement. It appears to go beyond the expected site of the junction of the superior vena cava and the right atrium. Also, at the level of the right main-stem bronchus, the catheter appears to curve with an infero-lateral convexity. To confirm the placement, a lateral view is obtained (Figure 2). As evident in this view, the internal jugular catheter does not terminate at the desirable level, but rather turns posteriorly to extend into the azygos vein (see arrow). The lateral view was required in this patient to ascertain the exact location of the catheter tip.

HAZARDS OF ABERRANT LINE PLACEMENT

Central venous catheters are commonly used to give various infusions (eg, parenteral nutrition), to draw blood, and to monitor the central venous pressure.¹ The internal jugular vein is one of the preferred veins for central venous access.^1,2 Normally, the anatomy of the jugular venous system and the design of the catheter facilitate proper insertion. Occasionally, however, despite proper technique, the tip of the catheter may not terminate at the desired level, resulting in aberrant placement in the internal thoracic vein, superior vena cava, azygos vein, accessory hemiazygos vein, or axillary vein.^1–8

The use of chest radiographs to establish the correct placement of internal jugular and subclavian lines has been advocated and is routinely practiced.^6,7 Obtaining a chest x-ray to confirm line placement is particularly important in patients with prior multiple and difficult catheterizations. The lateral view is seldom obtained when confirming central neck line placement, but when the anteroposterior view is not reassuring, it may be prudent to obtain this alternate view.

In a large retrospective analysis,⁹ cannulation of the azygos arch occurred in about 1.2% of insertions, and the rate was about seven times higher when the left jugular vein was used than when the right one was used. A smaller study gave the frequency of azygos arch cannulation as 0.7%.¹⁰

All procedure-related complications of central line insertion in the neck can also occur with aberrant azygos vein cannulation. These include infection, bacteremia, pneumothorax, hemothorax, arterial puncture, and various other mechanical complications. It should be noted that aberrant cannulation of the azygos arch is particularly hazardous, and that complication rates are typically higher. These complications are mainly due to the smaller vascular lumen and to the direction of blood flow in the azygos venous system.

 

KNOWING THE ANATOMY IS CRUCIAL

Knowledge of venous anatomy and its variants is crucial both for insertion and for ascertaining the correct placement of central venous lines.

The azygos vein has a much smaller lumen than the superior vena cava. Although the lumen size may vary significantly, the maximum diameter of the anterior arch of the azygos vein is about 6 to 7 mm,¹¹ whereas the superior vena cava lumen is typically 1.5 to 2 cm in diameter.¹² In addition, when a catheter is inserted into the superior vena cava, the direction of blood flow and the direction of the infusion are the same, but when the catheter is inserted into the azygos system, the directions of blood flow and infusion are opposite, contributing to local turbulence.

Both these factors increase the chance of puncturing the vein when the azygos arch is aberrantly cannulated for central venous access.⁹ Venous perforation has been reported in as many as 19% of cases in which the azygos arch was inadvertently cannulated. Venous perforation can lead to hemopericardium, hemomediastinum, and hemorrhagic pleural effusions, which can lead to significant morbidity and even death. Perforation, thrombosis, stenosis, and complete occlusion have been reported subsequent to catheter malposition in the azygos vein.¹³

Patients in whom the azygos vein is inadvertently cannulated may tolerate infusions and blood draws, but this does not mean that inadvertent azygos vein cannulation is acceptable, especially given the late complications of vascular perforation that can occur.⁹

The cannulation of the azygos vein in our patient was completely unintentional; nevertheless, the line was kept in and used for a short period for the initial rehydration and pain control and was subsequently removed without any complications.

WHEN IS CANNULATION OF THE AZYGOS VEIN AN OPTION?

In patients with previous multiple central vein cannulations, the rates of thrombosis and of fibrotic changes in these veins are high. In patients with thrombosis of both the superior vena cava and the inferior vena cava, direct insertion of a catheter into the azygos vein has been suggested as an alternate route to obtain access for dialysis.⁸ This approach has also been used successfully to administer total parenteral nutrition for a prolonged time in pediatric patients.¹⁴ In short, the azygos vein is never preferred for central venous access, but it can occasionally serve as an alternate route.^5–9

TAKE-HOME POINTS

The radiographic assessment of an internal jugular or subclavian line may occasionally be deceptive if based solely on the anteroposterior view; confirmation may require a lateral view. We found no guidelines for using the azygos vein for central venous access. The options in cases of aberrant cannulation include leaving the line in, removing and reinserting it at the same or another site under fluoroscopy, and attempting to reposition it after changing the catheter over a guidewire.

The use of central lines found to be in an aberrant position should be driven by clinical judgment based on the urgency of the need of access, the availability or feasibility of other access options, and the intended use. The use of the azygos vein is fraught with procedural complications, as well as postprocedural complications related to vascular perforation. If the position of the catheter tip on the anteroposterior radiographic view is not satisfactory, obtaining a lateral view should be considered.

A 22-year-old African American man with sickle cell disease comes to the in his joints and chest—a presentation similar to those of his previous sickle cell crises. He is given intravenous fluids for dehydration and morphine sulfate for pain via a peripheral line, and he is admitted to the hospital.

Shortly afterward, the peripheral line becomes infiltrated. After failed attempts at peripheral cannulation, central venous cannulation via an internal jugular site is considered.

The patient alerts us that he has had multiple “neck lines” in the past and that these had been difficult to place. With this in mind, we attempt to place a triple-lumen catheter under ultrasonographic guidance and with the use of sterile precautions and the Seldinger technique. On the first attempt, the guidewire cannot be advanced beyond 4 cm, and the attempt is terminated. On the second attempt, the guidewire advances freely, but as the catheter is advanced, slight resistance is felt at 4 cm and again at 10 cm. This resistance is overcome with slight pressure, and subsequent advancement meets with no further resistance. After confirming nonpulsatile blood return in all three lumens, we suture the catheter at 14 cm from the insertion site. A chest radiograph (Figure 1) is requested to confirm placement.

WHERE IS THE CATHETER TIP?

At first look, the catheter appears to broadly follow an expected trajectory. However, a closer look shows that the catheter is not properly positioned: although it is difficult to see, the tip appears to project beyond the main carina (see arrow), an important landmark to identify catheter tip placement. It appears to go beyond the expected site of the junction of the superior vena cava and the right atrium. Also, at the level of the right main-stem bronchus, the catheter appears to curve with an infero-lateral convexity. To confirm the placement, a lateral view is obtained (Figure 2). As evident in this view, the internal jugular catheter does not terminate at the desirable level, but rather turns posteriorly to extend into the azygos vein (see arrow). The lateral view was required in this patient to ascertain the exact location of the catheter tip.

HAZARDS OF ABERRANT LINE PLACEMENT

Central venous catheters are commonly used to give various infusions (eg, parenteral nutrition), to draw blood, and to monitor the central venous pressure.¹ The internal jugular vein is one of the preferred veins for central venous access.^1,2 Normally, the anatomy of the jugular venous system and the design of the catheter facilitate proper insertion. Occasionally, however, despite proper technique, the tip of the catheter may not terminate at the desired level, resulting in aberrant placement in the internal thoracic vein, superior vena cava, azygos vein, accessory hemiazygos vein, or axillary vein.^1–8

The use of chest radiographs to establish the correct placement of internal jugular and subclavian lines has been advocated and is routinely practiced.^6,7 Obtaining a chest x-ray to confirm line placement is particularly important in patients with prior multiple and difficult catheterizations. The lateral view is seldom obtained when confirming central neck line placement, but when the anteroposterior view is not reassuring, it may be prudent to obtain this alternate view.

In a large retrospective analysis,⁹ cannulation of the azygos arch occurred in about 1.2% of insertions, and the rate was about seven times higher when the left jugular vein was used than when the right one was used. A smaller study gave the frequency of azygos arch cannulation as 0.7%.¹⁰

All procedure-related complications of central line insertion in the neck can also occur with aberrant azygos vein cannulation. These include infection, bacteremia, pneumothorax, hemothorax, arterial puncture, and various other mechanical complications. It should be noted that aberrant cannulation of the azygos arch is particularly hazardous, and that complication rates are typically higher. These complications are mainly due to the smaller vascular lumen and to the direction of blood flow in the azygos venous system.

 

KNOWING THE ANATOMY IS CRUCIAL

Knowledge of venous anatomy and its variants is crucial both for insertion and for ascertaining the correct placement of central venous lines.

The azygos vein has a much smaller lumen than the superior vena cava. Although the lumen size may vary significantly, the maximum diameter of the anterior arch of the azygos vein is about 6 to 7 mm,¹¹ whereas the superior vena cava lumen is typically 1.5 to 2 cm in diameter.¹² In addition, when a catheter is inserted into the superior vena cava, the direction of blood flow and the direction of the infusion are the same, but when the catheter is inserted into the azygos system, the directions of blood flow and infusion are opposite, contributing to local turbulence.

Both these factors increase the chance of puncturing the vein when the azygos arch is aberrantly cannulated for central venous access.⁹ Venous perforation has been reported in as many as 19% of cases in which the azygos arch was inadvertently cannulated. Venous perforation can lead to hemopericardium, hemomediastinum, and hemorrhagic pleural effusions, which can lead to significant morbidity and even death. Perforation, thrombosis, stenosis, and complete occlusion have been reported subsequent to catheter malposition in the azygos vein.¹³

Patients in whom the azygos vein is inadvertently cannulated may tolerate infusions and blood draws, but this does not mean that inadvertent azygos vein cannulation is acceptable, especially given the late complications of vascular perforation that can occur.⁹

The cannulation of the azygos vein in our patient was completely unintentional; nevertheless, the line was kept in and used for a short period for the initial rehydration and pain control and was subsequently removed without any complications.

WHEN IS CANNULATION OF THE AZYGOS VEIN AN OPTION?

In patients with previous multiple central vein cannulations, the rates of thrombosis and of fibrotic changes in these veins are high. In patients with thrombosis of both the superior vena cava and the inferior vena cava, direct insertion of a catheter into the azygos vein has been suggested as an alternate route to obtain access for dialysis.⁸ This approach has also been used successfully to administer total parenteral nutrition for a prolonged time in pediatric patients.¹⁴ In short, the azygos vein is never preferred for central venous access, but it can occasionally serve as an alternate route.^5–9

TAKE-HOME POINTS

The radiographic assessment of an internal jugular or subclavian line may occasionally be deceptive if based solely on the anteroposterior view; confirmation may require a lateral view. We found no guidelines for using the azygos vein for central venous access. The options in cases of aberrant cannulation include leaving the line in, removing and reinserting it at the same or another site under fluoroscopy, and attempting to reposition it after changing the catheter over a guidewire.

The use of central lines found to be in an aberrant position should be driven by clinical judgment based on the urgency of the need of access, the availability or feasibility of other access options, and the intended use. The use of the azygos vein is fraught with procedural complications, as well as postprocedural complications related to vascular perforation. If the position of the catheter tip on the anteroposterior radiographic view is not satisfactory, obtaining a lateral view should be considered.

A 22-year-old African American man with sickle cell disease comes to the in his joints and chest—a presentation similar to those of his previous sickle cell crises. He is given intravenous fluids for dehydration and morphine sulfate for pain via a peripheral line, and he is admitted to the hospital.

Shortly afterward, the peripheral line becomes infiltrated. After failed attempts at peripheral cannulation, central venous cannulation via an internal jugular site is considered.

The patient alerts us that he has had multiple “neck lines” in the past and that these had been difficult to place. With this in mind, we attempt to place a triple-lumen catheter under ultrasonographic guidance and with the use of sterile precautions and the Seldinger technique. On the first attempt, the guidewire cannot be advanced beyond 4 cm, and the attempt is terminated. On the second attempt, the guidewire advances freely, but as the catheter is advanced, slight resistance is felt at 4 cm and again at 10 cm. This resistance is overcome with slight pressure, and subsequent advancement meets with no further resistance. After confirming nonpulsatile blood return in all three lumens, we suture the catheter at 14 cm from the insertion site. A chest radiograph (Figure 1) is requested to confirm placement.

WHERE IS THE CATHETER TIP?

At first look, the catheter appears to broadly follow an expected trajectory. However, a closer look shows that the catheter is not properly positioned: although it is difficult to see, the tip appears to project beyond the main carina (see arrow), an important landmark to identify catheter tip placement. It appears to go beyond the expected site of the junction of the superior vena cava and the right atrium. Also, at the level of the right main-stem bronchus, the catheter appears to curve with an infero-lateral convexity. To confirm the placement, a lateral view is obtained (Figure 2). As evident in this view, the internal jugular catheter does not terminate at the desirable level, but rather turns posteriorly to extend into the azygos vein (see arrow). The lateral view was required in this patient to ascertain the exact location of the catheter tip.

HAZARDS OF ABERRANT LINE PLACEMENT

Central venous catheters are commonly used to give various infusions (eg, parenteral nutrition), to draw blood, and to monitor the central venous pressure.¹ The internal jugular vein is one of the preferred veins for central venous access.^1,2 Normally, the anatomy of the jugular venous system and the design of the catheter facilitate proper insertion. Occasionally, however, despite proper technique, the tip of the catheter may not terminate at the desired level, resulting in aberrant placement in the internal thoracic vein, superior vena cava, azygos vein, accessory hemiazygos vein, or axillary vein.^1–8

The use of chest radiographs to establish the correct placement of internal jugular and subclavian lines has been advocated and is routinely practiced.^6,7 Obtaining a chest x-ray to confirm line placement is particularly important in patients with prior multiple and difficult catheterizations. The lateral view is seldom obtained when confirming central neck line placement, but when the anteroposterior view is not reassuring, it may be prudent to obtain this alternate view.

In a large retrospective analysis,⁹ cannulation of the azygos arch occurred in about 1.2% of insertions, and the rate was about seven times higher when the left jugular vein was used than when the right one was used. A smaller study gave the frequency of azygos arch cannulation as 0.7%.¹⁰

All procedure-related complications of central line insertion in the neck can also occur with aberrant azygos vein cannulation. These include infection, bacteremia, pneumothorax, hemothorax, arterial puncture, and various other mechanical complications. It should be noted that aberrant cannulation of the azygos arch is particularly hazardous, and that complication rates are typically higher. These complications are mainly due to the smaller vascular lumen and to the direction of blood flow in the azygos venous system.

 

KNOWING THE ANATOMY IS CRUCIAL

Knowledge of venous anatomy and its variants is crucial both for insertion and for ascertaining the correct placement of central venous lines.

The azygos vein has a much smaller lumen than the superior vena cava. Although the lumen size may vary significantly, the maximum diameter of the anterior arch of the azygos vein is about 6 to 7 mm,¹¹ whereas the superior vena cava lumen is typically 1.5 to 2 cm in diameter.¹² In addition, when a catheter is inserted into the superior vena cava, the direction of blood flow and the direction of the infusion are the same, but when the catheter is inserted into the azygos system, the directions of blood flow and infusion are opposite, contributing to local turbulence.

Both these factors increase the chance of puncturing the vein when the azygos arch is aberrantly cannulated for central venous access.⁹ Venous perforation has been reported in as many as 19% of cases in which the azygos arch was inadvertently cannulated. Venous perforation can lead to hemopericardium, hemomediastinum, and hemorrhagic pleural effusions, which can lead to significant morbidity and even death. Perforation, thrombosis, stenosis, and complete occlusion have been reported subsequent to catheter malposition in the azygos vein.¹³

Patients in whom the azygos vein is inadvertently cannulated may tolerate infusions and blood draws, but this does not mean that inadvertent azygos vein cannulation is acceptable, especially given the late complications of vascular perforation that can occur.⁹

The cannulation of the azygos vein in our patient was completely unintentional; nevertheless, the line was kept in and used for a short period for the initial rehydration and pain control and was subsequently removed without any complications.

WHEN IS CANNULATION OF THE AZYGOS VEIN AN OPTION?

In patients with previous multiple central vein cannulations, the rates of thrombosis and of fibrotic changes in these veins are high. In patients with thrombosis of both the superior vena cava and the inferior vena cava, direct insertion of a catheter into the azygos vein has been suggested as an alternate route to obtain access for dialysis.⁸ This approach has also been used successfully to administer total parenteral nutrition for a prolonged time in pediatric patients.¹⁴ In short, the azygos vein is never preferred for central venous access, but it can occasionally serve as an alternate route.^5–9

TAKE-HOME POINTS

The radiographic assessment of an internal jugular or subclavian line may occasionally be deceptive if based solely on the anteroposterior view; confirmation may require a lateral view. We found no guidelines for using the azygos vein for central venous access. The options in cases of aberrant cannulation include leaving the line in, removing and reinserting it at the same or another site under fluoroscopy, and attempting to reposition it after changing the catheter over a guidewire.

The use of central lines found to be in an aberrant position should be driven by clinical judgment based on the urgency of the need of access, the availability or feasibility of other access options, and the intended use. The use of the azygos vein is fraught with procedural complications, as well as postprocedural complications related to vascular perforation. If the position of the catheter tip on the anteroposterior radiographic view is not satisfactory, obtaining a lateral view should be considered.

Soon, the checklist for internists seeing patients about to undergo surgery may include prescribing one of the lipid-lowering hydroxymethylglutaryl-CoA reductase inhibitors, also called statins.

Statins? Not long ago, we were debating whether patients who take statins should stop taking them before surgery, based on the manufacturers’ recommendations.¹ The discussion, however, has changed to whether patients who have never received a statin should be started on one before surgery to provide immediate prophylaxis against cardiac morbidity, and how much harm long-term statin users face if these drugs are withheld perioperatively.

The evidence is still very preliminary and based mostly on studies in animals and retrospective studies in people. However, an expanding body of indirect evidence suggests that these drugs are beneficial in this situation.

In this review, we discuss the mechanisms by which statins may protect the heart in the short term, drawing on data from animal and human studies of acute myocardial infarction, and we review the current (albeit limited) data from the perioperative setting.

FEW INTERVENTIONS DECREASE RISK

Each year, approximately 50,000 patients suffer a perioperative cardiovascular event; the incidence of myocardial infarction during or after noncardiac surgery is 2% to 3%.² The primary goal of preoperative cardiovascular risk assessment is to predict and avert these events.

But short of canceling surgery, few interventions have been found to reduce a patient’s risk. For example, a landmark study in 2004 cast doubt on the efficacy of preoperative coronary revascularization.³ Similarly, although early studies of beta-blockers were promising^4,5 and although most internists prescribe these drugs before surgery, more recent studies have cast doubt on their efficacy, particularly in patients at low risk undergoing intermediate-risk (rather than vascular) surgery.^6–8

This changing clinical landscape has prompted a search for new strategies for perioperative risk-reduction. Several recent studies have placed statins in the spotlight.

POTENTIAL MECHANISMS OF SHORT-TERM BENEFIT

Statins have been proven to save lives when used long-term, but how could this class of drugs, designed to prevent the accumulation of arterial plaques by lowering low-density lipoprotein cholesterol (LDL-C) levels, have any short-term impact on operative outcomes? Although LDL-C reduction is the principal mechanism of action of statins, not all of the benefit can be ascribed to this mechanism.⁹ The answer may lie in their “pleiotropic” effects—ie, actions other than LDL-C reduction.

The more immediate pleiotropic effects of statins in the proinflammatory and prothrombotic environment of the perioperative period are thought to include improved endothelial function (both antithrombotic function and vasomotor function in response to ischemic stress), enhanced stability of atherosclerotic plaques, decreased oxidative stress, and decreased vascular inflammation.^10–12

EVIDENCE FROM ANIMAL STUDIES

Experiments in animals suggest that statins, given shortly before or after a cardiovascular event, confer benefit before any changes in LDL-C are measurable.

Lefer et al¹³ found that simvastatin (Zocor), given 18 hours before an ischemic episode in rats, blunted the inflammatory response in cardiac reperfusion injury. Not only was reperfusion injury significantly less in the hearts of the rats that received simvastatin than in the saline control group, but the simvastatin-treated hearts also expressed fewer neutrophil adhesion molecules such as P-selectin, and they had more basal release of nitric oxide, the potent endothelial-derived vasodilator with antithrombotic, anti-inflammatory, and antiproliferative effects.¹⁴ These results suggest that statins may improve endothelial function acutely, particularly during ischemic stress.

Osborne et al¹⁵ fed rabbits a cholesterol-rich diet plus either lovastatin (Mevacor) or placebo. After 2 weeks, the rabbits underwent either surgery to induce a myocardial infarction or a sham procedure. Regardless of the pretreatment, biopsies of the aorta did not reveal any atherosclerosis; yet the lovastatin-treated rabbits sustained less myocardial ischemic damage and they had more endothelium-mediated vasodilatation.

Statin therapy also may improve cerebral ischemia outcomes in animal models.^14,16

Sironi et al¹⁶ induced strokes in rats by occluding the middle cerebral artery. The rats received either simvastatin or vehicle for 3 days before the stroke or immediately afterwards. Even though simvastatin did not have enough time to affect the total cholesterol level, rats treated with simvastatin had smaller infarcts (as measured by magnetic resonance imaging) and produced more nitric oxide.

Comment. Taken together, these studies offer tantalizing evidence that statins have short-term, beneficial nonlipid effects and may reduce not only the likelihood of an ischemic event, but—should one occur—the degree of tissue damage that ensues.

 

EFFECTS OF STATINS IN ACUTE CORONARY SYNDROME

The National Registry of Myocardial Infarction¹⁷ is a prospective, observational database of all patients with acute myocardial infarction admitted to 1,230 participating hospitals throughout the United States. In an analysis from this cohort, patients were divided into four groups: those receiving statins before and after admission, those receiving statins only before admission, those receiving statins only after admission, and those who never received statins.

Compared with those who never received statins, fewer patients who received them both before and after admission died while in the hospital (unadjusted odds ratio 0.23, 95% confidence interval [CI] 0.22–0.25), and the odds ratio for those who received statins for the first time was 0.31 (95% CI 0.29–0.33). Patients who stopped receiving a statin on admission were more likely to die than were patients who never received statins (odds ratio 1.09, 95% CI 1.03–1.15). These trends held true even when adjustments were made for potential confounding factors.

Comment. Unmeasured confounding factors (such as the inability to take pills due to altered mental status or the different practice styles of the providers who chose to discontinue statins) might have affected the results. Nevertheless, these results suggest that the protective effects of statins stop almost immediately when these drugs are discontinued, and that there may even be an adverse “rebound” effect when patients who have been taking these drugs for a long time stop taking them temporarily.

The Platelet Receptor Inhibition in Ischemic Syndrome Management trial,¹⁸ in a subgroup analysis, had nearly identical findings. In the main part of this trial, patients with coronary artery disease and chest pain at rest or accelerating pain in the last 24 hours were randomized to receive tirofiban (Aggrastat) or heparin. Complete data on statin use were available for 1,616 (50%) of the 3,232 patients in this trial, and the rate of the primary end point (death, myocardial infarction, or recurrent ischemia) was analyzed on the basis of statin therapy in this subgroup.

The rate of the combined end point was significantly lower at 48 hours for those who had been receiving statins and continued receiving them (2.6%) than in those who never received statins (5.9%) or in those whose statins were discontinued (10.5%). Statins were more helpful if they were started before hospitalization than if they were started at the time of hospitalization.

Comment. Together, these data lead to the conclusion that, when admitted for either acute myocardial infarction or acute coronary syndrome, patients already receiving statins should not have them stopped, and those who had not been receiving statins should receive them immediately. The safety of these medications in the acute setting appears excellent: in the Myocardial Ischemia Reduction With Acute Cholesterol Lowering (MIRACL)¹² and the Pravastatin or Atorvastatin Evaluation and Infection Therapy (PROVE-IT)¹¹ trials, fewer than 5% of statin-treated patients had transient elevations in transaminase levels, and no cases of rhabdomyolysis were reported.

PERIOPERATIVE STATIN STUDIES

The data on perioperative statin use are mostly observational and retrospective and fall into essentially four surgical categories: coronary artery bypass grafting (CABG), carotid endarterectomy,^19,20 noncardiac vascular surgery, and major noncardiac surgery. Two meta-analyses have also evaluated the data.^21,22 The only randomized controlled trial (performed by Durazzo et al²³) was small and was carried out at a single center in vascular surgery patients, and the event rate was low.

Current recommendations from the National Cholesterol Education Program (NCEP)²⁴ say that patients who need CABG, have peripheral arterial disease, have an abdominal aortic aneurysm, or have cerebrovascular disease should already be on a statin to achieve an LDL-C goal level of less than 100 mg/dL, with an optional goal of less than 70 mg/dL, independent of surgery.

Since not all patients who should be on statins are actually on them, questions arise:

• Is it important (and safe) to start statin treatment preoperatively?
• Will patients with cardiovascular risk factors but without known cardiovascular disease benefit from statins perioperatively?

Noncardiac vascular surgery

Multiple retrospective studies have evaluated the effect of statins in patients undergoing major noncardiac vascular surgery.^25–32

Kertai et al²⁵ evaluated 570 patients in Holland who underwent elective open surgery for infrarenal abdominal aortic aneurysms between 1991 and 2001, looking for an association between statin use and the incidence of perioperative death from myocardial infarction. Only 162 of the 570 patients had been on long-term statin therapy before the surgery. The use of statins was only one of many known baseline characteristics that were significantly different between the two groups, including age, body mass index, known coronary artery disease, and use of angiotensin-converting enzyme inhibitors and beta-blockers. In univariate analysis, statins appeared to be protective: 6 (3.7%) of the patients in the statin group died of a myocardial infarction, compared with 45 (11%) of those in the nostatin group. A multivariate analysis yielded similar findings, with an odds ratio of 0.24 (95% CI 0.11–0.54).

Ward et al²⁷ performed a very similar retrospective study, with similar findings. In 446 patients who underwent surgery for infrarenal abdominal aortic aneurysm, statin therapy was associated with a significantly lower incidence of the combined end point of death, myocardial infarction, stroke, and major peripheral vascular complications, with an adjusted odds ratio of 0.36 (95% CI 0.14–0.93).

Poldermans et al²⁶ noted similar findings in a case-control study of noncardiac vascular surgery patients. Statin users had a much lower perioperative risk of death than did nonusers, with an adjusted odds ratio of 0.22 (95% CI 0.10–0.47).

O’Neil-Callahan et al,²⁸ in a cohort study, found that statin users had fewer perioperative cardiac complications, with an adjusted odds ratio of 0.49 (95% CI 0.28–0.84, P = .009).

 

 

Dogma of withdrawing statins before major surgery is challenged

Le Manach et al³³ reviewed the outcomes for all patients of a single hospital in Paris who underwent nonemergency infrarenal aortic procedures between January 2001 and December 2004. In January 2004, the hospital instituted guidelines to ensure that patients on statins continue taking them up to the evening before surgery and that statins be restarted on the first postoperative day (via nasogastric tube if necessary). Before 2004, there had been no specific guidelines, and patients on statins did not receive them for a median of 4 days postoperatively. Types of procedures were similar during the two time periods, as were the rates of beta-blocker use, preoperative revascularization, venous thromboembolism prophylaxis, and perioperative blood pressure control. After surgery, topononin I levels were measured in all patients as surveillance for cardiac events, and were defined as elevated when greater than 0.2 ng/mL.

Compared with patients not on statins at all, those treated with statins continuously throughout the perioperative period (after January 2004) had a lower rate of elevated troponin (relative risk 0.38). In contrast, those who had their statins transiently discontinued perioperatively (prior to 2004) had troponin elevations more often than those who had never been treated (relative risk 2.1). This suggested an over fivefold risk reduction (P < .001) conferred by not discontinuing statins in the immediate postoperative period. This finding was maintained after multivariate adjustment: statin withdrawal was associated with a 2.9-fold (95% CI 1.6–5.5) increase in the risk of cardiac enzyme elevations postoperatively. No fewer deaths were noted, but the study was not powered to detect a mortality difference.

Comment. Although secular trends cannot be entirely discounted as contributing to these findings, the prompt increase in cardiac events after just 4 days of statin withdrawal adds to the growing body of evidence suggesting that statin discontinuation can have harmful acute effects. It also brings up the question: Can starting statins benefit patients in the same time period?

Should statins be started before vascular surgery?

Schouten et al³² evaluated the effects of newly started or continued statin treatment in patients undergoing major elective vascular surgery. Patients were screened before surgery and started on statins if they were not already receiving them and their total cholesterol levels were elevated; new users received the medication for about 40 days before surgery. Of the 981 screened patients, 44 (5%) were newly started on statins and 182 (19%) were continued on their therapy. Perioperative death or myocardial infarction occurred in 22 (8.8%) of the statin users and 111 (14.7%) of the nonusers, a statistically significant difference. Temporary discontinuation (median 1 day) of statins in this study due to the inability to take an oral medication did not appear to affect the likelihood of a myocardial infarction.

Durazzo et al²³ performed a single-center, randomized, prospective, placebo-controlled, double-blind clinical trial of atorvastatin (Lipitor) 20 mg daily vs placebo in 100 patients undergoing noncardiac arterial vascular surgery. Patients were excluded if they had previously used medications to treat dyslipidemia, recently had a cardiovascular event, or had contraindications to statin treatment such as a baseline creatinine level greater than 2.0 mg/dL or severe hepatic disease. The intervention group received atorvastatin starting at least 2 weeks before surgery for a total of 45 days. Patients were then continued or started on a statin after surgery if their LDL-C level was greater than 100 mg/dL. Beta-blocker use was recommended “on the basis of current guidelines.”

One month after surgery, the LDL-C level was statistically significantly lower in the atorvastatin group. Since most patients did not continue or start statin therapy after the 45-day treatment period, the LDL-C levels were not statistically different at 3 and 6 months after surgery.

At 6 months, the rate of the primary end point (death from cardiovascular causes, nonfatal acute myocardial infarction, ischemic stroke, or unstable angina) was 26.0% in the placebo group and 8.0% in the atorvastatin group, a statistically significant difference. Three patients in the atorvastatin group had cardiac events in the first 10 days after surgery, compared with 11 patients in the placebo group. Thirteen of the 17 total cardiac events took place within 10 days after surgery.

One of the atorvastatin patients developed rhabdomyolysis and elevated aminotransferase levels.

Major noncardiac surgery

Lindenauer et al² performed a retrospective cohort study of surgical patients who were at least 18 years old and survived beyond the second hospital day. Patients were divided into a group receiving any form of lipid-lowering treatment (of whom more than 90% were taking statins) and a group that had never never received a lipid-lowering drug or only started one on the third day of the hospitalization or later. The period of study was from January 1, 2000, to December 31, 2001.

In all, 780,591 patients from 329 hospitals throughout the United States were included, of whom only 77,082 (9.9%) received lipid-lowering therapy. Eight percent of the patients underwent vascular surgery. Not surprisingly, the treated patients were more likely to have a history of hypertension, diabetes, ischemic heart disease, or hyperlipidemia. They also were more likely to have a vascular procedure performed, to have two or more cardiac risk factors (high-risk surgery, ischemic heart disease, congestive heart failure, cerebrovascular disease, renal insufficiency, or diabetes mellitus), and to be treated with beta-blockers and angiotensin-converting enzyme inhibitors, but they were less likely to have high-risk and emergency surgery performed.

The primary end point, perioperative death, occurred in 2.13% of the treated patients and 3.05% of the nontreated group. Compared with the rate in a propensity-matched cohort, the odds ratio adjusted for unbalanced covariates was 0.62 (95% CI 0.58–0.67) in favor of lipid treatment. Stratification by cardiac risk index revealed a number needed to treat of 186 for those with no risk factors, 60 for those with two risk factors, and 30 for those with four or more risk factors.

Unfortunately, this analysis was not able to take into account whether and for how long patients were receiving lipid-lowering therapy before hospitalization. It therefore does not answer the questions of whether starting lipid-lowering therapy before surgery is beneficial or whether stopping it is harmful. It also does not shed light on whether perioperative lipid-lowering increases the risk of rhabdomyolysis or liver disease.

 

 

Carotid endarterectomy

Two recent retrospective cohort studies evaluated the outcomes in patients undergoing carotid endarterectomy.^19,20

Kennedy et al¹⁹ found that patients on a statin at the time of admission who had symptomatic carotid disease had lower rates of inhospital death (adjusted odds ratio 0.24, 95% CI 0.06–0.91) and ischemic stroke or death (adjusted odds ratio 0.55, 95% CI 0.31–0.97). However, cardiac outcomes among these symptomatic patients were not significantly improved (odds ratio 0.82, 95% CI 0.45–1.50), nor was there benefit for asymptomatic patients, raising the possibility that the positive findings were due to chance or that patients at lower baseline risk for vascular events may have less benefit.

McGirt et al²⁰ performed a similar study; they did not, however, distinguish whether patients had symptomatic vs asymptomatic carotid disease. The 30-day risk of perioperative stroke was lower in patients treated with a statin, with an odds ratio of 0.41 (95% CI 0.18–0.93); the odds ratio for death was 0.21 (95% CI 0.05–0.96). Cardiac outcomes were not significantly affected.

Coronary artery bypass graft surgery

According to the NCEP recommendations, nearly all patients undergoing CABG should already be on a statin before surgery since they all have known coronary artery disease. Multiple observational studies have offered confirmatory evidence that statins are beneficial in this setting.^34–38

Liakopoulos et al³⁹ evaluated whether the anti-inflammatory effects of statins may, in part, account for their beneficial effect in the perioperative period. The authors prospectively matched 18 patients who were taking statins and were referred for elective CABG with 18 patients who were not prescribed statins previously. The only major measured baseline characteristic that differed between the two groups was a statistically significantly lower LDL-C level in the statin group. The operative characteristics did not differ, and cytokine levels at baseline were similar.

Tumor necrosis factor alpha levels increased significantly in the control group but did not change significantly in the statin group. Interleukin 8 increased in both groups by a similar amount. Interleukin 6 (the major inducer of C-reactive protein) increased from baseline in both groups but did not increase nearly as much in the statin group as in the control group; the intergroup difference was statistically significant. The anti-inflammatory cytokine interleukin 10 increased minimally from baseline in the control group, while the statin group’s levels increased significantly above baseline and those of the control group.

Christenson⁴⁰ also found that inflammatory markers were improved with pre-CABG statin treatment in a small randomized trial in which patients received simvastatin 20 mg 4 weeks prior to CABG surgery vs no statin. Interestingly, far fewer statin-treated patients developed thrombocytosis (platelet count > 400 × 10⁹/L) than did control patients (3% vs 81%, P < .0001).

RISKS OF PERIOPERATIVE STATINS

The risks associated with statin therapy in general appear low, but specific perioperative risks have not been well studied.

Baigent et al,⁴¹ in a meta-analysis of randomized trials of nonperioperative statin therapy, found that rhabdomyolysis occurred in 9 (0.023%) of 39,884 patients receiving statins vs 6 (0.015%) of the 39,817 controls, with a number needed to harm of 12,500. Moreover, the rates of nonvascular death and cancer did not increase. It is plausible that the risk is somewhat greater in the perioperative setting but is likely not enough to outweigh the potential benefits, especially since the risk of ischemic vascular events is particularly high then.

Some of the perioperative studies cited above specifically addressed potential risks. For example, in the study by Schouten et al,³² mild creatine kinase elevations were more common in the statin-treated group, but the incidence of moderate and severe creatine kinase elevations did not differ significantly. No case of rhabdomyolysis occurred, and length of surgery was the only predictor of myopathy. MIRACL and PROVE-IT revealed similar safety profiles; aminotransferase levels normalized when statins were stopped, and no cases of rhabdomyolysis occurred.^11,12 In the vascular surgery study by Durazzo et al,²³ 1 (2%) of the 50 atorvastatin-treated patients developed both rhabdomyolysis and elevated aminotransferase levels that prompted discontinuation of the statin.

Overall, the observational studies do not indicate that statin continuation or treatment is harmful in perioperative patients. However, these studies did not specifically evaluate patients with acute insults from surgery such as sepsis, renal failure, or hepatitis. It is unknown what effect statin therapy would have in those patients and whether statins should be selectively discontinued in patients who develop major hepatic, musculoskeletal, or renal complications after surgery.

 

OUR RECOMMENDATIONS

Before CABG or vascular surgery

Given the NCEP recommendations, existing primary and secondary prevention studies, observational studies of CABG and noncardiac vascular surgery patients, and the one randomized trial of vascular surgery patients, data support the use of statins in nearly all patients undergoing cardiac or vascular surgery. We advocate starting statins in the perioperative period to take advantage of their rapid-acting pleiotropic effects, and continuing them long-term to take advantage of their lipid-lowering effects. This recommendation is in line with the recently released American College of Cardiology/American Heart Association (ACC/AHA) 2007 perioperative guidelines that state “for patients undergoing vascular surgery with or without clinical risk factors, statin use is reasonable.”⁴²

Although the ideal time to start statins is not certain, the study by Durazzo et al²³ suggests that they should be started at least 2 weeks before surgery if possible. Moreover, patients already taking statins should definitely not have their statins discontinued if at all possible.

Before major nonvascular surgery

For patients undergoing major nonvascular (intermediate-risk) surgery, physicians should first ascertain if the patient has an indication for statin therapy based on current nonsurgical lipid level recommendations. However, even if there is no clear indication for statin therapy based on NCEP guidelines, we endorse the recently released ACC/AHA perioperative guidelines that state that statin therapy can be considered in patients with a risk factor who are undergoing intermediate-risk procedures. Moreover, we wholeheartedly support the ACC/AHA’s strongest recommendation that patients who are already receiving statins and are undergoing noncardiac surgery should not have their statins discontinued.

When to discontinue statins?

The risk of harm overall appears to be minimal and certainly less than the likelihood of benefit. It is reasonable to observe patients postoperatively for adverse clinical events that may increase the risk of perioperative statin treatment, such as acute renal failure, hepatic failure, or sepsis, but whether statins should be stopped in patients with these complications remains unknown; we advocate individualizing the decision.

More studies needed

We need more data on whether moderate-risk patients undergoing moderate-risk surgery benefit from perioperative statin therapy, when therapy should be started, whether therapy should be started on the day of surgery if it was not started earlier, which statin and what doses are optimal, how long therapy should be continued, and what degree of risk is associated with perioperative statin therapy.

Fortunately, important data should be forthcoming in the next few years: the Dutch Echocardiographic Cardiac Risk Evaluation Applying Stress Echocardiography (DECREASE-IV) study⁴³ is a 4-year two-by-two factorial placebo-controlled study evaluating the use of fluvastatin (Lescol) and bisoprolol (Zebeta, a beta-blocker) separately and together in patients who are older than 40 years, are undergoing elective noncardiac surgery, have an estimated risk of cardiovascular death of more than 1%, have not used statins previously, and do not have elevated cholesterol.

Soon, the checklist for internists seeing patients about to undergo surgery may include prescribing one of the lipid-lowering hydroxymethylglutaryl-CoA reductase inhibitors, also called statins.

Statins? Not long ago, we were debating whether patients who take statins should stop taking them before surgery, based on the manufacturers’ recommendations.¹ The discussion, however, has changed to whether patients who have never received a statin should be started on one before surgery to provide immediate prophylaxis against cardiac morbidity, and how much harm long-term statin users face if these drugs are withheld perioperatively.

The evidence is still very preliminary and based mostly on studies in animals and retrospective studies in people. However, an expanding body of indirect evidence suggests that these drugs are beneficial in this situation.

In this review, we discuss the mechanisms by which statins may protect the heart in the short term, drawing on data from animal and human studies of acute myocardial infarction, and we review the current (albeit limited) data from the perioperative setting.

FEW INTERVENTIONS DECREASE RISK

Each year, approximately 50,000 patients suffer a perioperative cardiovascular event; the incidence of myocardial infarction during or after noncardiac surgery is 2% to 3%.² The primary goal of preoperative cardiovascular risk assessment is to predict and avert these events.

But short of canceling surgery, few interventions have been found to reduce a patient’s risk. For example, a landmark study in 2004 cast doubt on the efficacy of preoperative coronary revascularization.³ Similarly, although early studies of beta-blockers were promising^4,5 and although most internists prescribe these drugs before surgery, more recent studies have cast doubt on their efficacy, particularly in patients at low risk undergoing intermediate-risk (rather than vascular) surgery.^6–8

This changing clinical landscape has prompted a search for new strategies for perioperative risk-reduction. Several recent studies have placed statins in the spotlight.

POTENTIAL MECHANISMS OF SHORT-TERM BENEFIT

Statins have been proven to save lives when used long-term, but how could this class of drugs, designed to prevent the accumulation of arterial plaques by lowering low-density lipoprotein cholesterol (LDL-C) levels, have any short-term impact on operative outcomes? Although LDL-C reduction is the principal mechanism of action of statins, not all of the benefit can be ascribed to this mechanism.⁹ The answer may lie in their “pleiotropic” effects—ie, actions other than LDL-C reduction.

The more immediate pleiotropic effects of statins in the proinflammatory and prothrombotic environment of the perioperative period are thought to include improved endothelial function (both antithrombotic function and vasomotor function in response to ischemic stress), enhanced stability of atherosclerotic plaques, decreased oxidative stress, and decreased vascular inflammation.^10–12

EVIDENCE FROM ANIMAL STUDIES

Experiments in animals suggest that statins, given shortly before or after a cardiovascular event, confer benefit before any changes in LDL-C are measurable.

Lefer et al¹³ found that simvastatin (Zocor), given 18 hours before an ischemic episode in rats, blunted the inflammatory response in cardiac reperfusion injury. Not only was reperfusion injury significantly less in the hearts of the rats that received simvastatin than in the saline control group, but the simvastatin-treated hearts also expressed fewer neutrophil adhesion molecules such as P-selectin, and they had more basal release of nitric oxide, the potent endothelial-derived vasodilator with antithrombotic, anti-inflammatory, and antiproliferative effects.¹⁴ These results suggest that statins may improve endothelial function acutely, particularly during ischemic stress.

Osborne et al¹⁵ fed rabbits a cholesterol-rich diet plus either lovastatin (Mevacor) or placebo. After 2 weeks, the rabbits underwent either surgery to induce a myocardial infarction or a sham procedure. Regardless of the pretreatment, biopsies of the aorta did not reveal any atherosclerosis; yet the lovastatin-treated rabbits sustained less myocardial ischemic damage and they had more endothelium-mediated vasodilatation.

Statin therapy also may improve cerebral ischemia outcomes in animal models.^14,16

Sironi et al¹⁶ induced strokes in rats by occluding the middle cerebral artery. The rats received either simvastatin or vehicle for 3 days before the stroke or immediately afterwards. Even though simvastatin did not have enough time to affect the total cholesterol level, rats treated with simvastatin had smaller infarcts (as measured by magnetic resonance imaging) and produced more nitric oxide.

Comment. Taken together, these studies offer tantalizing evidence that statins have short-term, beneficial nonlipid effects and may reduce not only the likelihood of an ischemic event, but—should one occur—the degree of tissue damage that ensues.

 

EFFECTS OF STATINS IN ACUTE CORONARY SYNDROME

The National Registry of Myocardial Infarction¹⁷ is a prospective, observational database of all patients with acute myocardial infarction admitted to 1,230 participating hospitals throughout the United States. In an analysis from this cohort, patients were divided into four groups: those receiving statins before and after admission, those receiving statins only before admission, those receiving statins only after admission, and those who never received statins.

Compared with those who never received statins, fewer patients who received them both before and after admission died while in the hospital (unadjusted odds ratio 0.23, 95% confidence interval [CI] 0.22–0.25), and the odds ratio for those who received statins for the first time was 0.31 (95% CI 0.29–0.33). Patients who stopped receiving a statin on admission were more likely to die than were patients who never received statins (odds ratio 1.09, 95% CI 1.03–1.15). These trends held true even when adjustments were made for potential confounding factors.

Comment. Unmeasured confounding factors (such as the inability to take pills due to altered mental status or the different practice styles of the providers who chose to discontinue statins) might have affected the results. Nevertheless, these results suggest that the protective effects of statins stop almost immediately when these drugs are discontinued, and that there may even be an adverse “rebound” effect when patients who have been taking these drugs for a long time stop taking them temporarily.

The Platelet Receptor Inhibition in Ischemic Syndrome Management trial,¹⁸ in a subgroup analysis, had nearly identical findings. In the main part of this trial, patients with coronary artery disease and chest pain at rest or accelerating pain in the last 24 hours were randomized to receive tirofiban (Aggrastat) or heparin. Complete data on statin use were available for 1,616 (50%) of the 3,232 patients in this trial, and the rate of the primary end point (death, myocardial infarction, or recurrent ischemia) was analyzed on the basis of statin therapy in this subgroup.

The rate of the combined end point was significantly lower at 48 hours for those who had been receiving statins and continued receiving them (2.6%) than in those who never received statins (5.9%) or in those whose statins were discontinued (10.5%). Statins were more helpful if they were started before hospitalization than if they were started at the time of hospitalization.

Comment. Together, these data lead to the conclusion that, when admitted for either acute myocardial infarction or acute coronary syndrome, patients already receiving statins should not have them stopped, and those who had not been receiving statins should receive them immediately. The safety of these medications in the acute setting appears excellent: in the Myocardial Ischemia Reduction With Acute Cholesterol Lowering (MIRACL)¹² and the Pravastatin or Atorvastatin Evaluation and Infection Therapy (PROVE-IT)¹¹ trials, fewer than 5% of statin-treated patients had transient elevations in transaminase levels, and no cases of rhabdomyolysis were reported.

PERIOPERATIVE STATIN STUDIES

The data on perioperative statin use are mostly observational and retrospective and fall into essentially four surgical categories: coronary artery bypass grafting (CABG), carotid endarterectomy,^19,20 noncardiac vascular surgery, and major noncardiac surgery. Two meta-analyses have also evaluated the data.^21,22 The only randomized controlled trial (performed by Durazzo et al²³) was small and was carried out at a single center in vascular surgery patients, and the event rate was low.

Current recommendations from the National Cholesterol Education Program (NCEP)²⁴ say that patients who need CABG, have peripheral arterial disease, have an abdominal aortic aneurysm, or have cerebrovascular disease should already be on a statin to achieve an LDL-C goal level of less than 100 mg/dL, with an optional goal of less than 70 mg/dL, independent of surgery.

Since not all patients who should be on statins are actually on them, questions arise:

• Is it important (and safe) to start statin treatment preoperatively?
• Will patients with cardiovascular risk factors but without known cardiovascular disease benefit from statins perioperatively?

Noncardiac vascular surgery

Multiple retrospective studies have evaluated the effect of statins in patients undergoing major noncardiac vascular surgery.^25–32

Kertai et al²⁵ evaluated 570 patients in Holland who underwent elective open surgery for infrarenal abdominal aortic aneurysms between 1991 and 2001, looking for an association between statin use and the incidence of perioperative death from myocardial infarction. Only 162 of the 570 patients had been on long-term statin therapy before the surgery. The use of statins was only one of many known baseline characteristics that were significantly different between the two groups, including age, body mass index, known coronary artery disease, and use of angiotensin-converting enzyme inhibitors and beta-blockers. In univariate analysis, statins appeared to be protective: 6 (3.7%) of the patients in the statin group died of a myocardial infarction, compared with 45 (11%) of those in the nostatin group. A multivariate analysis yielded similar findings, with an odds ratio of 0.24 (95% CI 0.11–0.54).

Ward et al²⁷ performed a very similar retrospective study, with similar findings. In 446 patients who underwent surgery for infrarenal abdominal aortic aneurysm, statin therapy was associated with a significantly lower incidence of the combined end point of death, myocardial infarction, stroke, and major peripheral vascular complications, with an adjusted odds ratio of 0.36 (95% CI 0.14–0.93).

Poldermans et al²⁶ noted similar findings in a case-control study of noncardiac vascular surgery patients. Statin users had a much lower perioperative risk of death than did nonusers, with an adjusted odds ratio of 0.22 (95% CI 0.10–0.47).

O’Neil-Callahan et al,²⁸ in a cohort study, found that statin users had fewer perioperative cardiac complications, with an adjusted odds ratio of 0.49 (95% CI 0.28–0.84, P = .009).

 

 

Dogma of withdrawing statins before major surgery is challenged

Le Manach et al³³ reviewed the outcomes for all patients of a single hospital in Paris who underwent nonemergency infrarenal aortic procedures between January 2001 and December 2004. In January 2004, the hospital instituted guidelines to ensure that patients on statins continue taking them up to the evening before surgery and that statins be restarted on the first postoperative day (via nasogastric tube if necessary). Before 2004, there had been no specific guidelines, and patients on statins did not receive them for a median of 4 days postoperatively. Types of procedures were similar during the two time periods, as were the rates of beta-blocker use, preoperative revascularization, venous thromboembolism prophylaxis, and perioperative blood pressure control. After surgery, topononin I levels were measured in all patients as surveillance for cardiac events, and were defined as elevated when greater than 0.2 ng/mL.

Compared with patients not on statins at all, those treated with statins continuously throughout the perioperative period (after January 2004) had a lower rate of elevated troponin (relative risk 0.38). In contrast, those who had their statins transiently discontinued perioperatively (prior to 2004) had troponin elevations more often than those who had never been treated (relative risk 2.1). This suggested an over fivefold risk reduction (P < .001) conferred by not discontinuing statins in the immediate postoperative period. This finding was maintained after multivariate adjustment: statin withdrawal was associated with a 2.9-fold (95% CI 1.6–5.5) increase in the risk of cardiac enzyme elevations postoperatively. No fewer deaths were noted, but the study was not powered to detect a mortality difference.

Comment. Although secular trends cannot be entirely discounted as contributing to these findings, the prompt increase in cardiac events after just 4 days of statin withdrawal adds to the growing body of evidence suggesting that statin discontinuation can have harmful acute effects. It also brings up the question: Can starting statins benefit patients in the same time period?

Should statins be started before vascular surgery?

Schouten et al³² evaluated the effects of newly started or continued statin treatment in patients undergoing major elective vascular surgery. Patients were screened before surgery and started on statins if they were not already receiving them and their total cholesterol levels were elevated; new users received the medication for about 40 days before surgery. Of the 981 screened patients, 44 (5%) were newly started on statins and 182 (19%) were continued on their therapy. Perioperative death or myocardial infarction occurred in 22 (8.8%) of the statin users and 111 (14.7%) of the nonusers, a statistically significant difference. Temporary discontinuation (median 1 day) of statins in this study due to the inability to take an oral medication did not appear to affect the likelihood of a myocardial infarction.

Durazzo et al²³ performed a single-center, randomized, prospective, placebo-controlled, double-blind clinical trial of atorvastatin (Lipitor) 20 mg daily vs placebo in 100 patients undergoing noncardiac arterial vascular surgery. Patients were excluded if they had previously used medications to treat dyslipidemia, recently had a cardiovascular event, or had contraindications to statin treatment such as a baseline creatinine level greater than 2.0 mg/dL or severe hepatic disease. The intervention group received atorvastatin starting at least 2 weeks before surgery for a total of 45 days. Patients were then continued or started on a statin after surgery if their LDL-C level was greater than 100 mg/dL. Beta-blocker use was recommended “on the basis of current guidelines.”

One month after surgery, the LDL-C level was statistically significantly lower in the atorvastatin group. Since most patients did not continue or start statin therapy after the 45-day treatment period, the LDL-C levels were not statistically different at 3 and 6 months after surgery.

At 6 months, the rate of the primary end point (death from cardiovascular causes, nonfatal acute myocardial infarction, ischemic stroke, or unstable angina) was 26.0% in the placebo group and 8.0% in the atorvastatin group, a statistically significant difference. Three patients in the atorvastatin group had cardiac events in the first 10 days after surgery, compared with 11 patients in the placebo group. Thirteen of the 17 total cardiac events took place within 10 days after surgery.

One of the atorvastatin patients developed rhabdomyolysis and elevated aminotransferase levels.

Major noncardiac surgery

Lindenauer et al² performed a retrospective cohort study of surgical patients who were at least 18 years old and survived beyond the second hospital day. Patients were divided into a group receiving any form of lipid-lowering treatment (of whom more than 90% were taking statins) and a group that had never never received a lipid-lowering drug or only started one on the third day of the hospitalization or later. The period of study was from January 1, 2000, to December 31, 2001.

In all, 780,591 patients from 329 hospitals throughout the United States were included, of whom only 77,082 (9.9%) received lipid-lowering therapy. Eight percent of the patients underwent vascular surgery. Not surprisingly, the treated patients were more likely to have a history of hypertension, diabetes, ischemic heart disease, or hyperlipidemia. They also were more likely to have a vascular procedure performed, to have two or more cardiac risk factors (high-risk surgery, ischemic heart disease, congestive heart failure, cerebrovascular disease, renal insufficiency, or diabetes mellitus), and to be treated with beta-blockers and angiotensin-converting enzyme inhibitors, but they were less likely to have high-risk and emergency surgery performed.

The primary end point, perioperative death, occurred in 2.13% of the treated patients and 3.05% of the nontreated group. Compared with the rate in a propensity-matched cohort, the odds ratio adjusted for unbalanced covariates was 0.62 (95% CI 0.58–0.67) in favor of lipid treatment. Stratification by cardiac risk index revealed a number needed to treat of 186 for those with no risk factors, 60 for those with two risk factors, and 30 for those with four or more risk factors.

Unfortunately, this analysis was not able to take into account whether and for how long patients were receiving lipid-lowering therapy before hospitalization. It therefore does not answer the questions of whether starting lipid-lowering therapy before surgery is beneficial or whether stopping it is harmful. It also does not shed light on whether perioperative lipid-lowering increases the risk of rhabdomyolysis or liver disease.

 

 

Carotid endarterectomy

Two recent retrospective cohort studies evaluated the outcomes in patients undergoing carotid endarterectomy.^19,20

Kennedy et al¹⁹ found that patients on a statin at the time of admission who had symptomatic carotid disease had lower rates of inhospital death (adjusted odds ratio 0.24, 95% CI 0.06–0.91) and ischemic stroke or death (adjusted odds ratio 0.55, 95% CI 0.31–0.97). However, cardiac outcomes among these symptomatic patients were not significantly improved (odds ratio 0.82, 95% CI 0.45–1.50), nor was there benefit for asymptomatic patients, raising the possibility that the positive findings were due to chance or that patients at lower baseline risk for vascular events may have less benefit.

McGirt et al²⁰ performed a similar study; they did not, however, distinguish whether patients had symptomatic vs asymptomatic carotid disease. The 30-day risk of perioperative stroke was lower in patients treated with a statin, with an odds ratio of 0.41 (95% CI 0.18–0.93); the odds ratio for death was 0.21 (95% CI 0.05–0.96). Cardiac outcomes were not significantly affected.

Coronary artery bypass graft surgery

According to the NCEP recommendations, nearly all patients undergoing CABG should already be on a statin before surgery since they all have known coronary artery disease. Multiple observational studies have offered confirmatory evidence that statins are beneficial in this setting.^34–38

Liakopoulos et al³⁹ evaluated whether the anti-inflammatory effects of statins may, in part, account for their beneficial effect in the perioperative period. The authors prospectively matched 18 patients who were taking statins and were referred for elective CABG with 18 patients who were not prescribed statins previously. The only major measured baseline characteristic that differed between the two groups was a statistically significantly lower LDL-C level in the statin group. The operative characteristics did not differ, and cytokine levels at baseline were similar.

Tumor necrosis factor alpha levels increased significantly in the control group but did not change significantly in the statin group. Interleukin 8 increased in both groups by a similar amount. Interleukin 6 (the major inducer of C-reactive protein) increased from baseline in both groups but did not increase nearly as much in the statin group as in the control group; the intergroup difference was statistically significant. The anti-inflammatory cytokine interleukin 10 increased minimally from baseline in the control group, while the statin group’s levels increased significantly above baseline and those of the control group.

Christenson⁴⁰ also found that inflammatory markers were improved with pre-CABG statin treatment in a small randomized trial in which patients received simvastatin 20 mg 4 weeks prior to CABG surgery vs no statin. Interestingly, far fewer statin-treated patients developed thrombocytosis (platelet count > 400 × 10⁹/L) than did control patients (3% vs 81%, P < .0001).

RISKS OF PERIOPERATIVE STATINS

The risks associated with statin therapy in general appear low, but specific perioperative risks have not been well studied.

Baigent et al,⁴¹ in a meta-analysis of randomized trials of nonperioperative statin therapy, found that rhabdomyolysis occurred in 9 (0.023%) of 39,884 patients receiving statins vs 6 (0.015%) of the 39,817 controls, with a number needed to harm of 12,500. Moreover, the rates of nonvascular death and cancer did not increase. It is plausible that the risk is somewhat greater in the perioperative setting but is likely not enough to outweigh the potential benefits, especially since the risk of ischemic vascular events is particularly high then.

Some of the perioperative studies cited above specifically addressed potential risks. For example, in the study by Schouten et al,³² mild creatine kinase elevations were more common in the statin-treated group, but the incidence of moderate and severe creatine kinase elevations did not differ significantly. No case of rhabdomyolysis occurred, and length of surgery was the only predictor of myopathy. MIRACL and PROVE-IT revealed similar safety profiles; aminotransferase levels normalized when statins were stopped, and no cases of rhabdomyolysis occurred.^11,12 In the vascular surgery study by Durazzo et al,²³ 1 (2%) of the 50 atorvastatin-treated patients developed both rhabdomyolysis and elevated aminotransferase levels that prompted discontinuation of the statin.

Overall, the observational studies do not indicate that statin continuation or treatment is harmful in perioperative patients. However, these studies did not specifically evaluate patients with acute insults from surgery such as sepsis, renal failure, or hepatitis. It is unknown what effect statin therapy would have in those patients and whether statins should be selectively discontinued in patients who develop major hepatic, musculoskeletal, or renal complications after surgery.

 

OUR RECOMMENDATIONS

Before CABG or vascular surgery

Given the NCEP recommendations, existing primary and secondary prevention studies, observational studies of CABG and noncardiac vascular surgery patients, and the one randomized trial of vascular surgery patients, data support the use of statins in nearly all patients undergoing cardiac or vascular surgery. We advocate starting statins in the perioperative period to take advantage of their rapid-acting pleiotropic effects, and continuing them long-term to take advantage of their lipid-lowering effects. This recommendation is in line with the recently released American College of Cardiology/American Heart Association (ACC/AHA) 2007 perioperative guidelines that state “for patients undergoing vascular surgery with or without clinical risk factors, statin use is reasonable.”⁴²

Although the ideal time to start statins is not certain, the study by Durazzo et al²³ suggests that they should be started at least 2 weeks before surgery if possible. Moreover, patients already taking statins should definitely not have their statins discontinued if at all possible.

Before major nonvascular surgery

For patients undergoing major nonvascular (intermediate-risk) surgery, physicians should first ascertain if the patient has an indication for statin therapy based on current nonsurgical lipid level recommendations. However, even if there is no clear indication for statin therapy based on NCEP guidelines, we endorse the recently released ACC/AHA perioperative guidelines that state that statin therapy can be considered in patients with a risk factor who are undergoing intermediate-risk procedures. Moreover, we wholeheartedly support the ACC/AHA’s strongest recommendation that patients who are already receiving statins and are undergoing noncardiac surgery should not have their statins discontinued.

When to discontinue statins?

The risk of harm overall appears to be minimal and certainly less than the likelihood of benefit. It is reasonable to observe patients postoperatively for adverse clinical events that may increase the risk of perioperative statin treatment, such as acute renal failure, hepatic failure, or sepsis, but whether statins should be stopped in patients with these complications remains unknown; we advocate individualizing the decision.

More studies needed

We need more data on whether moderate-risk patients undergoing moderate-risk surgery benefit from perioperative statin therapy, when therapy should be started, whether therapy should be started on the day of surgery if it was not started earlier, which statin and what doses are optimal, how long therapy should be continued, and what degree of risk is associated with perioperative statin therapy.

Fortunately, important data should be forthcoming in the next few years: the Dutch Echocardiographic Cardiac Risk Evaluation Applying Stress Echocardiography (DECREASE-IV) study⁴³ is a 4-year two-by-two factorial placebo-controlled study evaluating the use of fluvastatin (Lescol) and bisoprolol (Zebeta, a beta-blocker) separately and together in patients who are older than 40 years, are undergoing elective noncardiac surgery, have an estimated risk of cardiovascular death of more than 1%, have not used statins previously, and do not have elevated cholesterol.

Soon, the checklist for internists seeing patients about to undergo surgery may include prescribing one of the lipid-lowering hydroxymethylglutaryl-CoA reductase inhibitors, also called statins.

Statins? Not long ago, we were debating whether patients who take statins should stop taking them before surgery, based on the manufacturers’ recommendations.¹ The discussion, however, has changed to whether patients who have never received a statin should be started on one before surgery to provide immediate prophylaxis against cardiac morbidity, and how much harm long-term statin users face if these drugs are withheld perioperatively.

The evidence is still very preliminary and based mostly on studies in animals and retrospective studies in people. However, an expanding body of indirect evidence suggests that these drugs are beneficial in this situation.

In this review, we discuss the mechanisms by which statins may protect the heart in the short term, drawing on data from animal and human studies of acute myocardial infarction, and we review the current (albeit limited) data from the perioperative setting.

FEW INTERVENTIONS DECREASE RISK

Each year, approximately 50,000 patients suffer a perioperative cardiovascular event; the incidence of myocardial infarction during or after noncardiac surgery is 2% to 3%.² The primary goal of preoperative cardiovascular risk assessment is to predict and avert these events.

But short of canceling surgery, few interventions have been found to reduce a patient’s risk. For example, a landmark study in 2004 cast doubt on the efficacy of preoperative coronary revascularization.³ Similarly, although early studies of beta-blockers were promising^4,5 and although most internists prescribe these drugs before surgery, more recent studies have cast doubt on their efficacy, particularly in patients at low risk undergoing intermediate-risk (rather than vascular) surgery.^6–8

This changing clinical landscape has prompted a search for new strategies for perioperative risk-reduction. Several recent studies have placed statins in the spotlight.

POTENTIAL MECHANISMS OF SHORT-TERM BENEFIT

Statins have been proven to save lives when used long-term, but how could this class of drugs, designed to prevent the accumulation of arterial plaques by lowering low-density lipoprotein cholesterol (LDL-C) levels, have any short-term impact on operative outcomes? Although LDL-C reduction is the principal mechanism of action of statins, not all of the benefit can be ascribed to this mechanism.⁹ The answer may lie in their “pleiotropic” effects—ie, actions other than LDL-C reduction.

The more immediate pleiotropic effects of statins in the proinflammatory and prothrombotic environment of the perioperative period are thought to include improved endothelial function (both antithrombotic function and vasomotor function in response to ischemic stress), enhanced stability of atherosclerotic plaques, decreased oxidative stress, and decreased vascular inflammation.^10–12

EVIDENCE FROM ANIMAL STUDIES

Experiments in animals suggest that statins, given shortly before or after a cardiovascular event, confer benefit before any changes in LDL-C are measurable.

Lefer et al¹³ found that simvastatin (Zocor), given 18 hours before an ischemic episode in rats, blunted the inflammatory response in cardiac reperfusion injury. Not only was reperfusion injury significantly less in the hearts of the rats that received simvastatin than in the saline control group, but the simvastatin-treated hearts also expressed fewer neutrophil adhesion molecules such as P-selectin, and they had more basal release of nitric oxide, the potent endothelial-derived vasodilator with antithrombotic, anti-inflammatory, and antiproliferative effects.¹⁴ These results suggest that statins may improve endothelial function acutely, particularly during ischemic stress.

Osborne et al¹⁵ fed rabbits a cholesterol-rich diet plus either lovastatin (Mevacor) or placebo. After 2 weeks, the rabbits underwent either surgery to induce a myocardial infarction or a sham procedure. Regardless of the pretreatment, biopsies of the aorta did not reveal any atherosclerosis; yet the lovastatin-treated rabbits sustained less myocardial ischemic damage and they had more endothelium-mediated vasodilatation.

Statin therapy also may improve cerebral ischemia outcomes in animal models.^14,16

Sironi et al¹⁶ induced strokes in rats by occluding the middle cerebral artery. The rats received either simvastatin or vehicle for 3 days before the stroke or immediately afterwards. Even though simvastatin did not have enough time to affect the total cholesterol level, rats treated with simvastatin had smaller infarcts (as measured by magnetic resonance imaging) and produced more nitric oxide.

Comment. Taken together, these studies offer tantalizing evidence that statins have short-term, beneficial nonlipid effects and may reduce not only the likelihood of an ischemic event, but—should one occur—the degree of tissue damage that ensues.

 

EFFECTS OF STATINS IN ACUTE CORONARY SYNDROME

The National Registry of Myocardial Infarction¹⁷ is a prospective, observational database of all patients with acute myocardial infarction admitted to 1,230 participating hospitals throughout the United States. In an analysis from this cohort, patients were divided into four groups: those receiving statins before and after admission, those receiving statins only before admission, those receiving statins only after admission, and those who never received statins.

Compared with those who never received statins, fewer patients who received them both before and after admission died while in the hospital (unadjusted odds ratio 0.23, 95% confidence interval [CI] 0.22–0.25), and the odds ratio for those who received statins for the first time was 0.31 (95% CI 0.29–0.33). Patients who stopped receiving a statin on admission were more likely to die than were patients who never received statins (odds ratio 1.09, 95% CI 1.03–1.15). These trends held true even when adjustments were made for potential confounding factors.

Comment. Unmeasured confounding factors (such as the inability to take pills due to altered mental status or the different practice styles of the providers who chose to discontinue statins) might have affected the results. Nevertheless, these results suggest that the protective effects of statins stop almost immediately when these drugs are discontinued, and that there may even be an adverse “rebound” effect when patients who have been taking these drugs for a long time stop taking them temporarily.

The Platelet Receptor Inhibition in Ischemic Syndrome Management trial,¹⁸ in a subgroup analysis, had nearly identical findings. In the main part of this trial, patients with coronary artery disease and chest pain at rest or accelerating pain in the last 24 hours were randomized to receive tirofiban (Aggrastat) or heparin. Complete data on statin use were available for 1,616 (50%) of the 3,232 patients in this trial, and the rate of the primary end point (death, myocardial infarction, or recurrent ischemia) was analyzed on the basis of statin therapy in this subgroup.

The rate of the combined end point was significantly lower at 48 hours for those who had been receiving statins and continued receiving them (2.6%) than in those who never received statins (5.9%) or in those whose statins were discontinued (10.5%). Statins were more helpful if they were started before hospitalization than if they were started at the time of hospitalization.

Comment. Together, these data lead to the conclusion that, when admitted for either acute myocardial infarction or acute coronary syndrome, patients already receiving statins should not have them stopped, and those who had not been receiving statins should receive them immediately. The safety of these medications in the acute setting appears excellent: in the Myocardial Ischemia Reduction With Acute Cholesterol Lowering (MIRACL)¹² and the Pravastatin or Atorvastatin Evaluation and Infection Therapy (PROVE-IT)¹¹ trials, fewer than 5% of statin-treated patients had transient elevations in transaminase levels, and no cases of rhabdomyolysis were reported.

PERIOPERATIVE STATIN STUDIES

The data on perioperative statin use are mostly observational and retrospective and fall into essentially four surgical categories: coronary artery bypass grafting (CABG), carotid endarterectomy,^19,20 noncardiac vascular surgery, and major noncardiac surgery. Two meta-analyses have also evaluated the data.^21,22 The only randomized controlled trial (performed by Durazzo et al²³) was small and was carried out at a single center in vascular surgery patients, and the event rate was low.

Current recommendations from the National Cholesterol Education Program (NCEP)²⁴ say that patients who need CABG, have peripheral arterial disease, have an abdominal aortic aneurysm, or have cerebrovascular disease should already be on a statin to achieve an LDL-C goal level of less than 100 mg/dL, with an optional goal of less than 70 mg/dL, independent of surgery.

Since not all patients who should be on statins are actually on them, questions arise:

• Is it important (and safe) to start statin treatment preoperatively?
• Will patients with cardiovascular risk factors but without known cardiovascular disease benefit from statins perioperatively?

Noncardiac vascular surgery

Multiple retrospective studies have evaluated the effect of statins in patients undergoing major noncardiac vascular surgery.^25–32

Kertai et al²⁵ evaluated 570 patients in Holland who underwent elective open surgery for infrarenal abdominal aortic aneurysms between 1991 and 2001, looking for an association between statin use and the incidence of perioperative death from myocardial infarction. Only 162 of the 570 patients had been on long-term statin therapy before the surgery. The use of statins was only one of many known baseline characteristics that were significantly different between the two groups, including age, body mass index, known coronary artery disease, and use of angiotensin-converting enzyme inhibitors and beta-blockers. In univariate analysis, statins appeared to be protective: 6 (3.7%) of the patients in the statin group died of a myocardial infarction, compared with 45 (11%) of those in the nostatin group. A multivariate analysis yielded similar findings, with an odds ratio of 0.24 (95% CI 0.11–0.54).

Ward et al²⁷ performed a very similar retrospective study, with similar findings. In 446 patients who underwent surgery for infrarenal abdominal aortic aneurysm, statin therapy was associated with a significantly lower incidence of the combined end point of death, myocardial infarction, stroke, and major peripheral vascular complications, with an adjusted odds ratio of 0.36 (95% CI 0.14–0.93).

Poldermans et al²⁶ noted similar findings in a case-control study of noncardiac vascular surgery patients. Statin users had a much lower perioperative risk of death than did nonusers, with an adjusted odds ratio of 0.22 (95% CI 0.10–0.47).

O’Neil-Callahan et al,²⁸ in a cohort study, found that statin users had fewer perioperative cardiac complications, with an adjusted odds ratio of 0.49 (95% CI 0.28–0.84, P = .009).

 

 

Dogma of withdrawing statins before major surgery is challenged

Le Manach et al³³ reviewed the outcomes for all patients of a single hospital in Paris who underwent nonemergency infrarenal aortic procedures between January 2001 and December 2004. In January 2004, the hospital instituted guidelines to ensure that patients on statins continue taking them up to the evening before surgery and that statins be restarted on the first postoperative day (via nasogastric tube if necessary). Before 2004, there had been no specific guidelines, and patients on statins did not receive them for a median of 4 days postoperatively. Types of procedures were similar during the two time periods, as were the rates of beta-blocker use, preoperative revascularization, venous thromboembolism prophylaxis, and perioperative blood pressure control. After surgery, topononin I levels were measured in all patients as surveillance for cardiac events, and were defined as elevated when greater than 0.2 ng/mL.

Compared with patients not on statins at all, those treated with statins continuously throughout the perioperative period (after January 2004) had a lower rate of elevated troponin (relative risk 0.38). In contrast, those who had their statins transiently discontinued perioperatively (prior to 2004) had troponin elevations more often than those who had never been treated (relative risk 2.1). This suggested an over fivefold risk reduction (P < .001) conferred by not discontinuing statins in the immediate postoperative period. This finding was maintained after multivariate adjustment: statin withdrawal was associated with a 2.9-fold (95% CI 1.6–5.5) increase in the risk of cardiac enzyme elevations postoperatively. No fewer deaths were noted, but the study was not powered to detect a mortality difference.

Comment. Although secular trends cannot be entirely discounted as contributing to these findings, the prompt increase in cardiac events after just 4 days of statin withdrawal adds to the growing body of evidence suggesting that statin discontinuation can have harmful acute effects. It also brings up the question: Can starting statins benefit patients in the same time period?

Should statins be started before vascular surgery?

Schouten et al³² evaluated the effects of newly started or continued statin treatment in patients undergoing major elective vascular surgery. Patients were screened before surgery and started on statins if they were not already receiving them and their total cholesterol levels were elevated; new users received the medication for about 40 days before surgery. Of the 981 screened patients, 44 (5%) were newly started on statins and 182 (19%) were continued on their therapy. Perioperative death or myocardial infarction occurred in 22 (8.8%) of the statin users and 111 (14.7%) of the nonusers, a statistically significant difference. Temporary discontinuation (median 1 day) of statins in this study due to the inability to take an oral medication did not appear to affect the likelihood of a myocardial infarction.

Durazzo et al²³ performed a single-center, randomized, prospective, placebo-controlled, double-blind clinical trial of atorvastatin (Lipitor) 20 mg daily vs placebo in 100 patients undergoing noncardiac arterial vascular surgery. Patients were excluded if they had previously used medications to treat dyslipidemia, recently had a cardiovascular event, or had contraindications to statin treatment such as a baseline creatinine level greater than 2.0 mg/dL or severe hepatic disease. The intervention group received atorvastatin starting at least 2 weeks before surgery for a total of 45 days. Patients were then continued or started on a statin after surgery if their LDL-C level was greater than 100 mg/dL. Beta-blocker use was recommended “on the basis of current guidelines.”

One month after surgery, the LDL-C level was statistically significantly lower in the atorvastatin group. Since most patients did not continue or start statin therapy after the 45-day treatment period, the LDL-C levels were not statistically different at 3 and 6 months after surgery.

At 6 months, the rate of the primary end point (death from cardiovascular causes, nonfatal acute myocardial infarction, ischemic stroke, or unstable angina) was 26.0% in the placebo group and 8.0% in the atorvastatin group, a statistically significant difference. Three patients in the atorvastatin group had cardiac events in the first 10 days after surgery, compared with 11 patients in the placebo group. Thirteen of the 17 total cardiac events took place within 10 days after surgery.

One of the atorvastatin patients developed rhabdomyolysis and elevated aminotransferase levels.

Major noncardiac surgery

Lindenauer et al² performed a retrospective cohort study of surgical patients who were at least 18 years old and survived beyond the second hospital day. Patients were divided into a group receiving any form of lipid-lowering treatment (of whom more than 90% were taking statins) and a group that had never never received a lipid-lowering drug or only started one on the third day of the hospitalization or later. The period of study was from January 1, 2000, to December 31, 2001.

In all, 780,591 patients from 329 hospitals throughout the United States were included, of whom only 77,082 (9.9%) received lipid-lowering therapy. Eight percent of the patients underwent vascular surgery. Not surprisingly, the treated patients were more likely to have a history of hypertension, diabetes, ischemic heart disease, or hyperlipidemia. They also were more likely to have a vascular procedure performed, to have two or more cardiac risk factors (high-risk surgery, ischemic heart disease, congestive heart failure, cerebrovascular disease, renal insufficiency, or diabetes mellitus), and to be treated with beta-blockers and angiotensin-converting enzyme inhibitors, but they were less likely to have high-risk and emergency surgery performed.

The primary end point, perioperative death, occurred in 2.13% of the treated patients and 3.05% of the nontreated group. Compared with the rate in a propensity-matched cohort, the odds ratio adjusted for unbalanced covariates was 0.62 (95% CI 0.58–0.67) in favor of lipid treatment. Stratification by cardiac risk index revealed a number needed to treat of 186 for those with no risk factors, 60 for those with two risk factors, and 30 for those with four or more risk factors.

Unfortunately, this analysis was not able to take into account whether and for how long patients were receiving lipid-lowering therapy before hospitalization. It therefore does not answer the questions of whether starting lipid-lowering therapy before surgery is beneficial or whether stopping it is harmful. It also does not shed light on whether perioperative lipid-lowering increases the risk of rhabdomyolysis or liver disease.

 

 

Carotid endarterectomy

Two recent retrospective cohort studies evaluated the outcomes in patients undergoing carotid endarterectomy.^19,20

Kennedy et al¹⁹ found that patients on a statin at the time of admission who had symptomatic carotid disease had lower rates of inhospital death (adjusted odds ratio 0.24, 95% CI 0.06–0.91) and ischemic stroke or death (adjusted odds ratio 0.55, 95% CI 0.31–0.97). However, cardiac outcomes among these symptomatic patients were not significantly improved (odds ratio 0.82, 95% CI 0.45–1.50), nor was there benefit for asymptomatic patients, raising the possibility that the positive findings were due to chance or that patients at lower baseline risk for vascular events may have less benefit.

McGirt et al²⁰ performed a similar study; they did not, however, distinguish whether patients had symptomatic vs asymptomatic carotid disease. The 30-day risk of perioperative stroke was lower in patients treated with a statin, with an odds ratio of 0.41 (95% CI 0.18–0.93); the odds ratio for death was 0.21 (95% CI 0.05–0.96). Cardiac outcomes were not significantly affected.

Coronary artery bypass graft surgery

According to the NCEP recommendations, nearly all patients undergoing CABG should already be on a statin before surgery since they all have known coronary artery disease. Multiple observational studies have offered confirmatory evidence that statins are beneficial in this setting.^34–38

Liakopoulos et al³⁹ evaluated whether the anti-inflammatory effects of statins may, in part, account for their beneficial effect in the perioperative period. The authors prospectively matched 18 patients who were taking statins and were referred for elective CABG with 18 patients who were not prescribed statins previously. The only major measured baseline characteristic that differed between the two groups was a statistically significantly lower LDL-C level in the statin group. The operative characteristics did not differ, and cytokine levels at baseline were similar.

Tumor necrosis factor alpha levels increased significantly in the control group but did not change significantly in the statin group. Interleukin 8 increased in both groups by a similar amount. Interleukin 6 (the major inducer of C-reactive protein) increased from baseline in both groups but did not increase nearly as much in the statin group as in the control group; the intergroup difference was statistically significant. The anti-inflammatory cytokine interleukin 10 increased minimally from baseline in the control group, while the statin group’s levels increased significantly above baseline and those of the control group.

Christenson⁴⁰ also found that inflammatory markers were improved with pre-CABG statin treatment in a small randomized trial in which patients received simvastatin 20 mg 4 weeks prior to CABG surgery vs no statin. Interestingly, far fewer statin-treated patients developed thrombocytosis (platelet count > 400 × 10⁹/L) than did control patients (3% vs 81%, P < .0001).

RISKS OF PERIOPERATIVE STATINS

The risks associated with statin therapy in general appear low, but specific perioperative risks have not been well studied.

Baigent et al,⁴¹ in a meta-analysis of randomized trials of nonperioperative statin therapy, found that rhabdomyolysis occurred in 9 (0.023%) of 39,884 patients receiving statins vs 6 (0.015%) of the 39,817 controls, with a number needed to harm of 12,500. Moreover, the rates of nonvascular death and cancer did not increase. It is plausible that the risk is somewhat greater in the perioperative setting but is likely not enough to outweigh the potential benefits, especially since the risk of ischemic vascular events is particularly high then.

Some of the perioperative studies cited above specifically addressed potential risks. For example, in the study by Schouten et al,³² mild creatine kinase elevations were more common in the statin-treated group, but the incidence of moderate and severe creatine kinase elevations did not differ significantly. No case of rhabdomyolysis occurred, and length of surgery was the only predictor of myopathy. MIRACL and PROVE-IT revealed similar safety profiles; aminotransferase levels normalized when statins were stopped, and no cases of rhabdomyolysis occurred.^11,12 In the vascular surgery study by Durazzo et al,²³ 1 (2%) of the 50 atorvastatin-treated patients developed both rhabdomyolysis and elevated aminotransferase levels that prompted discontinuation of the statin.

Overall, the observational studies do not indicate that statin continuation or treatment is harmful in perioperative patients. However, these studies did not specifically evaluate patients with acute insults from surgery such as sepsis, renal failure, or hepatitis. It is unknown what effect statin therapy would have in those patients and whether statins should be selectively discontinued in patients who develop major hepatic, musculoskeletal, or renal complications after surgery.

 

OUR RECOMMENDATIONS

Before CABG or vascular surgery

Given the NCEP recommendations, existing primary and secondary prevention studies, observational studies of CABG and noncardiac vascular surgery patients, and the one randomized trial of vascular surgery patients, data support the use of statins in nearly all patients undergoing cardiac or vascular surgery. We advocate starting statins in the perioperative period to take advantage of their rapid-acting pleiotropic effects, and continuing them long-term to take advantage of their lipid-lowering effects. This recommendation is in line with the recently released American College of Cardiology/American Heart Association (ACC/AHA) 2007 perioperative guidelines that state “for patients undergoing vascular surgery with or without clinical risk factors, statin use is reasonable.”⁴²

Although the ideal time to start statins is not certain, the study by Durazzo et al²³ suggests that they should be started at least 2 weeks before surgery if possible. Moreover, patients already taking statins should definitely not have their statins discontinued if at all possible.

Before major nonvascular surgery

For patients undergoing major nonvascular (intermediate-risk) surgery, physicians should first ascertain if the patient has an indication for statin therapy based on current nonsurgical lipid level recommendations. However, even if there is no clear indication for statin therapy based on NCEP guidelines, we endorse the recently released ACC/AHA perioperative guidelines that state that statin therapy can be considered in patients with a risk factor who are undergoing intermediate-risk procedures. Moreover, we wholeheartedly support the ACC/AHA’s strongest recommendation that patients who are already receiving statins and are undergoing noncardiac surgery should not have their statins discontinued.

When to discontinue statins?

The risk of harm overall appears to be minimal and certainly less than the likelihood of benefit. It is reasonable to observe patients postoperatively for adverse clinical events that may increase the risk of perioperative statin treatment, such as acute renal failure, hepatic failure, or sepsis, but whether statins should be stopped in patients with these complications remains unknown; we advocate individualizing the decision.

More studies needed

We need more data on whether moderate-risk patients undergoing moderate-risk surgery benefit from perioperative statin therapy, when therapy should be started, whether therapy should be started on the day of surgery if it was not started earlier, which statin and what doses are optimal, how long therapy should be continued, and what degree of risk is associated with perioperative statin therapy.

Fortunately, important data should be forthcoming in the next few years: the Dutch Echocardiographic Cardiac Risk Evaluation Applying Stress Echocardiography (DECREASE-IV) study⁴³ is a 4-year two-by-two factorial placebo-controlled study evaluating the use of fluvastatin (Lescol) and bisoprolol (Zebeta, a beta-blocker) separately and together in patients who are older than 40 years, are undergoing elective noncardiac surgery, have an estimated risk of cardiovascular death of more than 1%, have not used statins previously, and do not have elevated cholesterol.

A 44-year-old man comes to the emergency room because of light-headedness and fatigue. He says he has had several similar but milder episodes in the last several months. He also mentions that he thinks he has been coughing up blood. He says he has no major medical or surgical problems of which he is aware, but he appears confused and unable to give an accurate history. No family members can be contacted for further history at the moment.

Physical examination reveals nothing remarkable, but the patient does cough up some blood during the examination. His hemoglobin level is 6.0 g/dL (reference range 13.5–17.5).

What imaging tests would be helpful in this patient’s evaluation?

HEMOPTYSIS HAS MANY CAUSES

Hemoptysis is defined as the expectoration of blood originating from the tracheobronchial tree or the pulmonary parenchyma.

Most cases of hemoptysis are benign and self-limited; life-threatening hemoptysis is rare.^1–3 However, hemoptysis can be a sign of serious tracheopulmonary disease.

The bleeding can be from the large (Table 1) or the small (Table 2) pulmonary vessels. Bleeding from the small vessels is known as diffuse alveolar hemorrhage, and it characteristically presents as alveolar infiltrates on chest radiography. In these cases, further imaging studies provide little benefit.⁴ This paper will focus on the imaging of and radiographic interventions for large-vessel bleeding.

The causes of hemoptysis are numerous; common causes of bleeding from the large vessels nowadays include bronchiectasis, fungal infections, tuberculosis, and cancer.^1,5,6 Still, no cause is identified in 15% to 30% of all cases,^1,2,5 even after extensive evaluation.

Definition of ‘massive’ hemoptysis can vary

Various definitions of the severity of hemoptysis have been proposed. The threshold of “massive” hemoptysis has been defined as as low as 100 mL/24 hours and as high as 1 L/24 hours; the most common definition is 300 mL, or about 1 cup.^2,3,5–10

However, the patient’s cardiorespiratory status must also be considered.^5,6,9 If the patient cannot maintain his or her airway, a small amount of bleeding could be life-threatening and should be considered significant or massive. Thus, we define massive hemoptysis as more than 300 mL of blood within 24 hours or any amount of blood with concurrent cardiorespiratory compromise.

It is important to recognize massive hemoptysis quickly, because without urgent treatment, up to 80% of patients may die.^5,6,11 This can sometimes pose a challenge, as the history may not always be helpful and the patient’s perception of massive hemoptysis may differ from the clinically accepted definition. For example, in a patient without respiratory compromise, we would not consider bloodtinged sputum or small amounts of blood that add up to 1 to 2 teaspoons (5–10 mL) to be massive, although the patient might. On the other hand, hemoptysis with cardiorespiratory compromise must be considered significant (and very possibly massive) until proven otherwise, even if the amount of blood is small.

Massive hemoptysis is usually the result of erosion of systemic (rather than pulmonary) arteries by bronchial neoplasm, active tuberculosis, or aspergilloma.^6,9,12,13 Arteriovenous malformations and pulmonary artery aneurysms are much less common causes.^5,11,13

IMAGING AND DIAGNOSTIC OPTIONS

Most cases of hemoptysis have an identifiable source and cause of the bleeding at the time of initial diagnosis.¹⁴ Currently, there is no consensus on what is the best workup for hemoptysis. Still, a complete evaluation includes patient history, physical examination, bronchoscopy, laboratory tests, and imaging studies (Figure 1). Imaging studies that can be helpful include chest radiography, conventional computed tomography (CT), multi-detector CT angiography, and conventional angiography.

Chest radiography

Chest radiography is an excellent initial imaging test for evaluating hemoptysis. It is quick and inexpensive and can provide insight into acute chest problems. As mentioned above, in cases of alveolar hemorrhage, radiography typically reveals alveolar infiltrates.⁴ In cases of hemoptysis due to large-vessel bleeding, radiography can reveal a variety of pertinent findings, such as a mass, pneumonia, chronic lung disease, atelectasis, or a cavitary lesion (FIGURE 2). Even if the findings are nonspecific (such as in pneumonia), radiography can narrow the location of the problem to a single lobe or at least to a single lung, and this information can guide further evaluation by bronchoscopy.^4,9

In as many as 40% of cases of hemoptysis, however, the findings on chest radiography are normal or do not reveal the source of the bleeding.^15,16 Approximately 5% to 6% of patients with hemoptysis and normal results on radiography are eventually found to have lung cancer.¹⁴ Thus, while a localizing finding on radiography is helpful, a normal or nonlocalizing finding warrants further evaluation by other means, including conventional CT, multidetector CT angiography, or bronchoscopy.

 

Computed tomography

Both conventional CT and multidetector CT angiography are quick and noninvasive ways to locate the site of bleeding, determine the cause of bleeding (Figure 3, Figure 4), and create a map to guide further therapy.^5,6,11,13

CT is superior to fiberoptic bronchoscopy in finding a cause of hemoptysis, its main advantage being its ability to show distal airways beyond the reach of the bronchoscope, and the lung parenchyma surrounding these distal airways.^5,15,16 In locating the site of bleeding, CT performs about as well as fiberoptic bronchoscopy.⁵

However, while CT imaging is extremely useful in evaluating bleeding from larger vessels, it adds little information beyond that obtained by chest radiography in cases of diffuse alveolar hemorrhage.⁴

Multidetector CT angiography is the optimal CT study for evaluating hemoptysis. In addition to showing the lung parenchyma and airways, it allows one to evaluate the integrity of pulmonary, bronchial, and nonbronchial systemic arteries within the chest. It is at least as good as (and, with multiplanar reformatted images, possibly even better than) conventional angiography in evaluating bronchial and nonbronchial systemic arteries. Multidetector CT angiography is recommended before bronchial artery embolization to help one plan the procedure and shorten the procedure time, if the patient is stable enough that this imaging study can be done first.^6,12,13

The iodinated contrast material used in CT angiography can cause contrast nephropathy in patients with renal failure. At Cleveland Clinic, we avoid using contrast if the patient’s serum creatinine level is 2.0 mg/dL or greater or if it is rapidly rising, even if it is in the normal range or only slightly elevated; a rapid rise would indicate acute renal failure (eg, in glomerulonephritis). In these cases, we recommend CT without contrast.

CT of the chest has revealed malignancies in cases of hemoptysis in which radiography and bronchoscopy did not.^15,17 Although CT is more than 90% sensitive in detecting endobronchial lesions, it has limitations: a blood clot within the bronchus can look like a tumor, and acute bleeding can obscure an endobronchial lesion.⁵ Thus, bronchoscopy remains an important, complementary diagnostic tool in the evaluation of acute hemoptysis.

Bronchoscopy

Bronchoscopy is overall much less sensitive than CT in detecting the cause of the bleeding,^15,16,18 but, if performed early it as useful as CT in finding the site of bleeding,^5,9 information that can be helpful in planning further therapy.¹⁹ It may be more useful than CT in evaluating endobronchial lesions during acute hemoptysis, as active bleeding can obscure an endobronchial lesion on CT.⁵ However, the distal airways are often filled with blood, making them difficult to evaluate via bronchoscopy.

In approximately 10% of cases of massive hemoptysis, rigid bronchoscopy is preferred over fiberoptic bronchoscopy, and it is often used in a perioperative setting. However, its use is not usually possible in unstable patients receiving intensive care. Instead, flexible fiberoptic bronchoscopy can be used in patients whose condition is too unstable to allow them to leave the intensive care unit to undergo CT. Flexible fiberoptic bronchoscopy does not require an operating room or anesthesia,¹⁹ and can be done in the intensive care unit itself.

Not only can bronchoscopy accurately locate the site of bleeding, it can also aid in controlling the airway in patients with catastrophic hemorrhage and temporarily control bleeding through Fogarty balloon tamponade, direct application of a mixture of epinephrine and cold saline, or topical hemostatic tamponade therapy with a solution of thrombin or fibrinogen and thrombin.^2,3,19 It also provides complementary information about endobronchial lesions and is valuable in providing samples for tissue diagnosis and microbial cultures.

Diagnostic angiography has limitations

Although it is possible to bypass radiography, CT, and bronchoscopy in a case of massive hemoptysis and to rush the patient to the angiography suite for combined diagnostic angiography and therapeutic bronchial artery embolization, this approach has limitations. Diagnostic angiography does not identify the source of bleeding as well as CT does.⁶ It is important to locate the bleeding site first via CT, multidetector CT angiography, or bronchoscopy. Diagnostic angiography can be time-consuming. The procedure time can be significantly shorter if CT, bronchoscopy, or both are done first to ascertain the site of bleeding before bronchial artery embolization.^1,6 Another reason that performing CT first is important is that it can rule out situations in which surgery would be preferred over bronchial artery embolization.⁶

In more than 90% of cases of hemoptysis requiring embolization or surgery, the bleeding is from the bronchial arteries.^{5,6,9,11–13} However, bronchoscopy before bronchial artery embolization is unnecessary in patients with hemoptysis of known cause if the site of bleeding can be determined from radiography or CT and if no bronchoscopic airway management is needed.¹⁸

 

BRONCHIAL ARTERY EMBOLIZATION: AN ALTERNATIVE TO SURGERY

After a cause of the hemoptysis has been established by radiography, CT, or bronchoscopy, bronchial artery embolization is an effective first-line therapy to control massive, life-threatening bleeding.⁶ It is an alternative in patients who cannot undergo surgery because of bilateral or extensive disease that renders them unable to tolerate life after a lobectomy.^6,12,18

Indications for bronchial artery embolization include failure of conservative management, massive hemoptysis, recurrent hemoptysis, and poor surgical risk. It is also done to control bleeding temporarily before surgery.¹

Another indication for this therapy is peripheral pulmonary artery pseudoaneurysm, which is found in up to 11% of patients undergoing bronchial angiography for hemoptysis. These patients typically present with recurrent hemoptysis (sometimes massive) and occasionally with both hemoptysis and clubbing. Most of these patients have either chronic active pulmonary tuberculosis or a mycetoma complicating sarcoidosis or tuberculosis. Occlusion of the pulmonary artery pseudoaneurysm may require embolization of bronchial arteries, nonbronchial systemic arteries, or pulmonary artery branches.²⁰

Surgery, however, is still the definitive treatment of choice for thoracic vascular injury, bronchial adenoma, aspergilloma resistant to other therapies, and hydatid cyst.⁶ A cardiothoracic surgeon should be consulted in these cases.

Outcomes of embolization

Images courtesy of Abraham Levitin, MD.

Aside from the cases in which surgery is indicated, bronchial artery embolization (Figure 5) is a very successful minimally invasive therapy that controls bleeding immediately in 66% to 90% of patients.^1,7,21 It is the preferred emergency treatment for massive hemoptysis, as the death rate is 7.1% to 18.2%, which, though high, is considerably less than the 40% seen in emergency surgery for massive hemoptysis.⁶

If a patient with massive hemoptysis undergoes successful bronchial artery embolization but the bleeding recurs 1 to 6 months later, the cause is likely an undetected nonbronchial systemic arterial supply and incomplete embolization.^1,22 Late rebleeding (6–12 months after the procedure) occurs in 20% to 40% of patients and is likely to be from disease progression.^1,7

Common complications of bronchial artery embolization are transient chest pain and dysphagia. Very rare complications include subintimal dissection and spinal cord ischemia due to inadvertent occlusion of the spinal arteries.⁶ Another complication in patients with renal failure is contrast nephropathy, the risk of which must be weighed against the possible consequences—including death—of not performing bronchial artery embolization in a patient who cannot undergo surgery.

CASE REVISITED: CLINICAL COURSE

In the patient described at the beginning of this article, a chest radiograph obtained in the emergency room showed an area of nonspecific consolidation in the left upper lung. Conventional chest CT was then ordered (Figure 4), and it revealed a cavitary lesion in the left upper lobe, consistent with aspergilloma. Bronchoscopy was then performed, and it too indicated that the bleeding was coming from the left upper lobe. Samples obtained during the procedure were sent to the laboratory for bacterial and fungal cultures.

In the meantime, family members were contacted, and they revealed that the patient had a history of sarcoidosis.

The patient went on to develop massive hemoptysis. Although the treatment of choice for mycetoma is primary resection, our patient’s respiratory status was poor as a result of extensive pulmonary sarcoidosis, and he was not considered a candidate for emergency surgery at that time. He was rushed to the angiography suite and successfully underwent emergency bronchial artery embolization.

A 44-year-old man comes to the emergency room because of light-headedness and fatigue. He says he has had several similar but milder episodes in the last several months. He also mentions that he thinks he has been coughing up blood. He says he has no major medical or surgical problems of which he is aware, but he appears confused and unable to give an accurate history. No family members can be contacted for further history at the moment.

Physical examination reveals nothing remarkable, but the patient does cough up some blood during the examination. His hemoglobin level is 6.0 g/dL (reference range 13.5–17.5).

What imaging tests would be helpful in this patient’s evaluation?

HEMOPTYSIS HAS MANY CAUSES

Hemoptysis is defined as the expectoration of blood originating from the tracheobronchial tree or the pulmonary parenchyma.

Most cases of hemoptysis are benign and self-limited; life-threatening hemoptysis is rare.^1–3 However, hemoptysis can be a sign of serious tracheopulmonary disease.

The bleeding can be from the large (Table 1) or the small (Table 2) pulmonary vessels. Bleeding from the small vessels is known as diffuse alveolar hemorrhage, and it characteristically presents as alveolar infiltrates on chest radiography. In these cases, further imaging studies provide little benefit.⁴ This paper will focus on the imaging of and radiographic interventions for large-vessel bleeding.

The causes of hemoptysis are numerous; common causes of bleeding from the large vessels nowadays include bronchiectasis, fungal infections, tuberculosis, and cancer.^1,5,6 Still, no cause is identified in 15% to 30% of all cases,^1,2,5 even after extensive evaluation.

Definition of ‘massive’ hemoptysis can vary

Various definitions of the severity of hemoptysis have been proposed. The threshold of “massive” hemoptysis has been defined as as low as 100 mL/24 hours and as high as 1 L/24 hours; the most common definition is 300 mL, or about 1 cup.^2,3,5–10

However, the patient’s cardiorespiratory status must also be considered.^5,6,9 If the patient cannot maintain his or her airway, a small amount of bleeding could be life-threatening and should be considered significant or massive. Thus, we define massive hemoptysis as more than 300 mL of blood within 24 hours or any amount of blood with concurrent cardiorespiratory compromise.

It is important to recognize massive hemoptysis quickly, because without urgent treatment, up to 80% of patients may die.^5,6,11 This can sometimes pose a challenge, as the history may not always be helpful and the patient’s perception of massive hemoptysis may differ from the clinically accepted definition. For example, in a patient without respiratory compromise, we would not consider bloodtinged sputum or small amounts of blood that add up to 1 to 2 teaspoons (5–10 mL) to be massive, although the patient might. On the other hand, hemoptysis with cardiorespiratory compromise must be considered significant (and very possibly massive) until proven otherwise, even if the amount of blood is small.

Massive hemoptysis is usually the result of erosion of systemic (rather than pulmonary) arteries by bronchial neoplasm, active tuberculosis, or aspergilloma.^6,9,12,13 Arteriovenous malformations and pulmonary artery aneurysms are much less common causes.^5,11,13

IMAGING AND DIAGNOSTIC OPTIONS

Most cases of hemoptysis have an identifiable source and cause of the bleeding at the time of initial diagnosis.¹⁴ Currently, there is no consensus on what is the best workup for hemoptysis. Still, a complete evaluation includes patient history, physical examination, bronchoscopy, laboratory tests, and imaging studies (Figure 1). Imaging studies that can be helpful include chest radiography, conventional computed tomography (CT), multi-detector CT angiography, and conventional angiography.

Chest radiography

Chest radiography is an excellent initial imaging test for evaluating hemoptysis. It is quick and inexpensive and can provide insight into acute chest problems. As mentioned above, in cases of alveolar hemorrhage, radiography typically reveals alveolar infiltrates.⁴ In cases of hemoptysis due to large-vessel bleeding, radiography can reveal a variety of pertinent findings, such as a mass, pneumonia, chronic lung disease, atelectasis, or a cavitary lesion (FIGURE 2). Even if the findings are nonspecific (such as in pneumonia), radiography can narrow the location of the problem to a single lobe or at least to a single lung, and this information can guide further evaluation by bronchoscopy.^4,9

In as many as 40% of cases of hemoptysis, however, the findings on chest radiography are normal or do not reveal the source of the bleeding.^15,16 Approximately 5% to 6% of patients with hemoptysis and normal results on radiography are eventually found to have lung cancer.¹⁴ Thus, while a localizing finding on radiography is helpful, a normal or nonlocalizing finding warrants further evaluation by other means, including conventional CT, multidetector CT angiography, or bronchoscopy.

 

Computed tomography

Both conventional CT and multidetector CT angiography are quick and noninvasive ways to locate the site of bleeding, determine the cause of bleeding (Figure 3, Figure 4), and create a map to guide further therapy.^5,6,11,13

CT is superior to fiberoptic bronchoscopy in finding a cause of hemoptysis, its main advantage being its ability to show distal airways beyond the reach of the bronchoscope, and the lung parenchyma surrounding these distal airways.^5,15,16 In locating the site of bleeding, CT performs about as well as fiberoptic bronchoscopy.⁵

However, while CT imaging is extremely useful in evaluating bleeding from larger vessels, it adds little information beyond that obtained by chest radiography in cases of diffuse alveolar hemorrhage.⁴

Multidetector CT angiography is the optimal CT study for evaluating hemoptysis. In addition to showing the lung parenchyma and airways, it allows one to evaluate the integrity of pulmonary, bronchial, and nonbronchial systemic arteries within the chest. It is at least as good as (and, with multiplanar reformatted images, possibly even better than) conventional angiography in evaluating bronchial and nonbronchial systemic arteries. Multidetector CT angiography is recommended before bronchial artery embolization to help one plan the procedure and shorten the procedure time, if the patient is stable enough that this imaging study can be done first.^6,12,13

The iodinated contrast material used in CT angiography can cause contrast nephropathy in patients with renal failure. At Cleveland Clinic, we avoid using contrast if the patient’s serum creatinine level is 2.0 mg/dL or greater or if it is rapidly rising, even if it is in the normal range or only slightly elevated; a rapid rise would indicate acute renal failure (eg, in glomerulonephritis). In these cases, we recommend CT without contrast.

CT of the chest has revealed malignancies in cases of hemoptysis in which radiography and bronchoscopy did not.^15,17 Although CT is more than 90% sensitive in detecting endobronchial lesions, it has limitations: a blood clot within the bronchus can look like a tumor, and acute bleeding can obscure an endobronchial lesion.⁵ Thus, bronchoscopy remains an important, complementary diagnostic tool in the evaluation of acute hemoptysis.

Bronchoscopy

Bronchoscopy is overall much less sensitive than CT in detecting the cause of the bleeding,^15,16,18 but, if performed early it as useful as CT in finding the site of bleeding,^5,9 information that can be helpful in planning further therapy.¹⁹ It may be more useful than CT in evaluating endobronchial lesions during acute hemoptysis, as active bleeding can obscure an endobronchial lesion on CT.⁵ However, the distal airways are often filled with blood, making them difficult to evaluate via bronchoscopy.

In approximately 10% of cases of massive hemoptysis, rigid bronchoscopy is preferred over fiberoptic bronchoscopy, and it is often used in a perioperative setting. However, its use is not usually possible in unstable patients receiving intensive care. Instead, flexible fiberoptic bronchoscopy can be used in patients whose condition is too unstable to allow them to leave the intensive care unit to undergo CT. Flexible fiberoptic bronchoscopy does not require an operating room or anesthesia,¹⁹ and can be done in the intensive care unit itself.

Not only can bronchoscopy accurately locate the site of bleeding, it can also aid in controlling the airway in patients with catastrophic hemorrhage and temporarily control bleeding through Fogarty balloon tamponade, direct application of a mixture of epinephrine and cold saline, or topical hemostatic tamponade therapy with a solution of thrombin or fibrinogen and thrombin.^2,3,19 It also provides complementary information about endobronchial lesions and is valuable in providing samples for tissue diagnosis and microbial cultures.

Diagnostic angiography has limitations

Although it is possible to bypass radiography, CT, and bronchoscopy in a case of massive hemoptysis and to rush the patient to the angiography suite for combined diagnostic angiography and therapeutic bronchial artery embolization, this approach has limitations. Diagnostic angiography does not identify the source of bleeding as well as CT does.⁶ It is important to locate the bleeding site first via CT, multidetector CT angiography, or bronchoscopy. Diagnostic angiography can be time-consuming. The procedure time can be significantly shorter if CT, bronchoscopy, or both are done first to ascertain the site of bleeding before bronchial artery embolization.^1,6 Another reason that performing CT first is important is that it can rule out situations in which surgery would be preferred over bronchial artery embolization.⁶

In more than 90% of cases of hemoptysis requiring embolization or surgery, the bleeding is from the bronchial arteries.^{5,6,9,11–13} However, bronchoscopy before bronchial artery embolization is unnecessary in patients with hemoptysis of known cause if the site of bleeding can be determined from radiography or CT and if no bronchoscopic airway management is needed.¹⁸

 

BRONCHIAL ARTERY EMBOLIZATION: AN ALTERNATIVE TO SURGERY

After a cause of the hemoptysis has been established by radiography, CT, or bronchoscopy, bronchial artery embolization is an effective first-line therapy to control massive, life-threatening bleeding.⁶ It is an alternative in patients who cannot undergo surgery because of bilateral or extensive disease that renders them unable to tolerate life after a lobectomy.^6,12,18

Indications for bronchial artery embolization include failure of conservative management, massive hemoptysis, recurrent hemoptysis, and poor surgical risk. It is also done to control bleeding temporarily before surgery.¹

Another indication for this therapy is peripheral pulmonary artery pseudoaneurysm, which is found in up to 11% of patients undergoing bronchial angiography for hemoptysis. These patients typically present with recurrent hemoptysis (sometimes massive) and occasionally with both hemoptysis and clubbing. Most of these patients have either chronic active pulmonary tuberculosis or a mycetoma complicating sarcoidosis or tuberculosis. Occlusion of the pulmonary artery pseudoaneurysm may require embolization of bronchial arteries, nonbronchial systemic arteries, or pulmonary artery branches.²⁰

Surgery, however, is still the definitive treatment of choice for thoracic vascular injury, bronchial adenoma, aspergilloma resistant to other therapies, and hydatid cyst.⁶ A cardiothoracic surgeon should be consulted in these cases.

Outcomes of embolization

Images courtesy of Abraham Levitin, MD.

Aside from the cases in which surgery is indicated, bronchial artery embolization (Figure 5) is a very successful minimally invasive therapy that controls bleeding immediately in 66% to 90% of patients.^1,7,21 It is the preferred emergency treatment for massive hemoptysis, as the death rate is 7.1% to 18.2%, which, though high, is considerably less than the 40% seen in emergency surgery for massive hemoptysis.⁶

If a patient with massive hemoptysis undergoes successful bronchial artery embolization but the bleeding recurs 1 to 6 months later, the cause is likely an undetected nonbronchial systemic arterial supply and incomplete embolization.^1,22 Late rebleeding (6–12 months after the procedure) occurs in 20% to 40% of patients and is likely to be from disease progression.^1,7

Common complications of bronchial artery embolization are transient chest pain and dysphagia. Very rare complications include subintimal dissection and spinal cord ischemia due to inadvertent occlusion of the spinal arteries.⁶ Another complication in patients with renal failure is contrast nephropathy, the risk of which must be weighed against the possible consequences—including death—of not performing bronchial artery embolization in a patient who cannot undergo surgery.

CASE REVISITED: CLINICAL COURSE

In the patient described at the beginning of this article, a chest radiograph obtained in the emergency room showed an area of nonspecific consolidation in the left upper lung. Conventional chest CT was then ordered (Figure 4), and it revealed a cavitary lesion in the left upper lobe, consistent with aspergilloma. Bronchoscopy was then performed, and it too indicated that the bleeding was coming from the left upper lobe. Samples obtained during the procedure were sent to the laboratory for bacterial and fungal cultures.

In the meantime, family members were contacted, and they revealed that the patient had a history of sarcoidosis.

The patient went on to develop massive hemoptysis. Although the treatment of choice for mycetoma is primary resection, our patient’s respiratory status was poor as a result of extensive pulmonary sarcoidosis, and he was not considered a candidate for emergency surgery at that time. He was rushed to the angiography suite and successfully underwent emergency bronchial artery embolization.

A 44-year-old man comes to the emergency room because of light-headedness and fatigue. He says he has had several similar but milder episodes in the last several months. He also mentions that he thinks he has been coughing up blood. He says he has no major medical or surgical problems of which he is aware, but he appears confused and unable to give an accurate history. No family members can be contacted for further history at the moment.

Physical examination reveals nothing remarkable, but the patient does cough up some blood during the examination. His hemoglobin level is 6.0 g/dL (reference range 13.5–17.5).

What imaging tests would be helpful in this patient’s evaluation?

HEMOPTYSIS HAS MANY CAUSES

Hemoptysis is defined as the expectoration of blood originating from the tracheobronchial tree or the pulmonary parenchyma.

Most cases of hemoptysis are benign and self-limited; life-threatening hemoptysis is rare.^1–3 However, hemoptysis can be a sign of serious tracheopulmonary disease.

The bleeding can be from the large (Table 1) or the small (Table 2) pulmonary vessels. Bleeding from the small vessels is known as diffuse alveolar hemorrhage, and it characteristically presents as alveolar infiltrates on chest radiography. In these cases, further imaging studies provide little benefit.⁴ This paper will focus on the imaging of and radiographic interventions for large-vessel bleeding.

The causes of hemoptysis are numerous; common causes of bleeding from the large vessels nowadays include bronchiectasis, fungal infections, tuberculosis, and cancer.^1,5,6 Still, no cause is identified in 15% to 30% of all cases,^1,2,5 even after extensive evaluation.

Definition of ‘massive’ hemoptysis can vary

Various definitions of the severity of hemoptysis have been proposed. The threshold of “massive” hemoptysis has been defined as as low as 100 mL/24 hours and as high as 1 L/24 hours; the most common definition is 300 mL, or about 1 cup.^2,3,5–10

However, the patient’s cardiorespiratory status must also be considered.^5,6,9 If the patient cannot maintain his or her airway, a small amount of bleeding could be life-threatening and should be considered significant or massive. Thus, we define massive hemoptysis as more than 300 mL of blood within 24 hours or any amount of blood with concurrent cardiorespiratory compromise.

It is important to recognize massive hemoptysis quickly, because without urgent treatment, up to 80% of patients may die.^5,6,11 This can sometimes pose a challenge, as the history may not always be helpful and the patient’s perception of massive hemoptysis may differ from the clinically accepted definition. For example, in a patient without respiratory compromise, we would not consider bloodtinged sputum or small amounts of blood that add up to 1 to 2 teaspoons (5–10 mL) to be massive, although the patient might. On the other hand, hemoptysis with cardiorespiratory compromise must be considered significant (and very possibly massive) until proven otherwise, even if the amount of blood is small.

Massive hemoptysis is usually the result of erosion of systemic (rather than pulmonary) arteries by bronchial neoplasm, active tuberculosis, or aspergilloma.^6,9,12,13 Arteriovenous malformations and pulmonary artery aneurysms are much less common causes.^5,11,13

IMAGING AND DIAGNOSTIC OPTIONS

Most cases of hemoptysis have an identifiable source and cause of the bleeding at the time of initial diagnosis.¹⁴ Currently, there is no consensus on what is the best workup for hemoptysis. Still, a complete evaluation includes patient history, physical examination, bronchoscopy, laboratory tests, and imaging studies (Figure 1). Imaging studies that can be helpful include chest radiography, conventional computed tomography (CT), multi-detector CT angiography, and conventional angiography.

Chest radiography

Chest radiography is an excellent initial imaging test for evaluating hemoptysis. It is quick and inexpensive and can provide insight into acute chest problems. As mentioned above, in cases of alveolar hemorrhage, radiography typically reveals alveolar infiltrates.⁴ In cases of hemoptysis due to large-vessel bleeding, radiography can reveal a variety of pertinent findings, such as a mass, pneumonia, chronic lung disease, atelectasis, or a cavitary lesion (FIGURE 2). Even if the findings are nonspecific (such as in pneumonia), radiography can narrow the location of the problem to a single lobe or at least to a single lung, and this information can guide further evaluation by bronchoscopy.^4,9

In as many as 40% of cases of hemoptysis, however, the findings on chest radiography are normal or do not reveal the source of the bleeding.^15,16 Approximately 5% to 6% of patients with hemoptysis and normal results on radiography are eventually found to have lung cancer.¹⁴ Thus, while a localizing finding on radiography is helpful, a normal or nonlocalizing finding warrants further evaluation by other means, including conventional CT, multidetector CT angiography, or bronchoscopy.

 

Computed tomography

Both conventional CT and multidetector CT angiography are quick and noninvasive ways to locate the site of bleeding, determine the cause of bleeding (Figure 3, Figure 4), and create a map to guide further therapy.^5,6,11,13

CT is superior to fiberoptic bronchoscopy in finding a cause of hemoptysis, its main advantage being its ability to show distal airways beyond the reach of the bronchoscope, and the lung parenchyma surrounding these distal airways.^5,15,16 In locating the site of bleeding, CT performs about as well as fiberoptic bronchoscopy.⁵

However, while CT imaging is extremely useful in evaluating bleeding from larger vessels, it adds little information beyond that obtained by chest radiography in cases of diffuse alveolar hemorrhage.⁴

Multidetector CT angiography is the optimal CT study for evaluating hemoptysis. In addition to showing the lung parenchyma and airways, it allows one to evaluate the integrity of pulmonary, bronchial, and nonbronchial systemic arteries within the chest. It is at least as good as (and, with multiplanar reformatted images, possibly even better than) conventional angiography in evaluating bronchial and nonbronchial systemic arteries. Multidetector CT angiography is recommended before bronchial artery embolization to help one plan the procedure and shorten the procedure time, if the patient is stable enough that this imaging study can be done first.^6,12,13

The iodinated contrast material used in CT angiography can cause contrast nephropathy in patients with renal failure. At Cleveland Clinic, we avoid using contrast if the patient’s serum creatinine level is 2.0 mg/dL or greater or if it is rapidly rising, even if it is in the normal range or only slightly elevated; a rapid rise would indicate acute renal failure (eg, in glomerulonephritis). In these cases, we recommend CT without contrast.

CT of the chest has revealed malignancies in cases of hemoptysis in which radiography and bronchoscopy did not.^15,17 Although CT is more than 90% sensitive in detecting endobronchial lesions, it has limitations: a blood clot within the bronchus can look like a tumor, and acute bleeding can obscure an endobronchial lesion.⁵ Thus, bronchoscopy remains an important, complementary diagnostic tool in the evaluation of acute hemoptysis.

Bronchoscopy

Bronchoscopy is overall much less sensitive than CT in detecting the cause of the bleeding,^15,16,18 but, if performed early it as useful as CT in finding the site of bleeding,^5,9 information that can be helpful in planning further therapy.¹⁹ It may be more useful than CT in evaluating endobronchial lesions during acute hemoptysis, as active bleeding can obscure an endobronchial lesion on CT.⁵ However, the distal airways are often filled with blood, making them difficult to evaluate via bronchoscopy.

In approximately 10% of cases of massive hemoptysis, rigid bronchoscopy is preferred over fiberoptic bronchoscopy, and it is often used in a perioperative setting. However, its use is not usually possible in unstable patients receiving intensive care. Instead, flexible fiberoptic bronchoscopy can be used in patients whose condition is too unstable to allow them to leave the intensive care unit to undergo CT. Flexible fiberoptic bronchoscopy does not require an operating room or anesthesia,¹⁹ and can be done in the intensive care unit itself.

Not only can bronchoscopy accurately locate the site of bleeding, it can also aid in controlling the airway in patients with catastrophic hemorrhage and temporarily control bleeding through Fogarty balloon tamponade, direct application of a mixture of epinephrine and cold saline, or topical hemostatic tamponade therapy with a solution of thrombin or fibrinogen and thrombin.^2,3,19 It also provides complementary information about endobronchial lesions and is valuable in providing samples for tissue diagnosis and microbial cultures.

Diagnostic angiography has limitations

Although it is possible to bypass radiography, CT, and bronchoscopy in a case of massive hemoptysis and to rush the patient to the angiography suite for combined diagnostic angiography and therapeutic bronchial artery embolization, this approach has limitations. Diagnostic angiography does not identify the source of bleeding as well as CT does.⁶ It is important to locate the bleeding site first via CT, multidetector CT angiography, or bronchoscopy. Diagnostic angiography can be time-consuming. The procedure time can be significantly shorter if CT, bronchoscopy, or both are done first to ascertain the site of bleeding before bronchial artery embolization.^1,6 Another reason that performing CT first is important is that it can rule out situations in which surgery would be preferred over bronchial artery embolization.⁶

In more than 90% of cases of hemoptysis requiring embolization or surgery, the bleeding is from the bronchial arteries.^{5,6,9,11–13} However, bronchoscopy before bronchial artery embolization is unnecessary in patients with hemoptysis of known cause if the site of bleeding can be determined from radiography or CT and if no bronchoscopic airway management is needed.¹⁸

 

BRONCHIAL ARTERY EMBOLIZATION: AN ALTERNATIVE TO SURGERY

After a cause of the hemoptysis has been established by radiography, CT, or bronchoscopy, bronchial artery embolization is an effective first-line therapy to control massive, life-threatening bleeding.⁶ It is an alternative in patients who cannot undergo surgery because of bilateral or extensive disease that renders them unable to tolerate life after a lobectomy.^6,12,18

Indications for bronchial artery embolization include failure of conservative management, massive hemoptysis, recurrent hemoptysis, and poor surgical risk. It is also done to control bleeding temporarily before surgery.¹

Another indication for this therapy is peripheral pulmonary artery pseudoaneurysm, which is found in up to 11% of patients undergoing bronchial angiography for hemoptysis. These patients typically present with recurrent hemoptysis (sometimes massive) and occasionally with both hemoptysis and clubbing. Most of these patients have either chronic active pulmonary tuberculosis or a mycetoma complicating sarcoidosis or tuberculosis. Occlusion of the pulmonary artery pseudoaneurysm may require embolization of bronchial arteries, nonbronchial systemic arteries, or pulmonary artery branches.²⁰

Surgery, however, is still the definitive treatment of choice for thoracic vascular injury, bronchial adenoma, aspergilloma resistant to other therapies, and hydatid cyst.⁶ A cardiothoracic surgeon should be consulted in these cases.

Outcomes of embolization

Images courtesy of Abraham Levitin, MD.

Aside from the cases in which surgery is indicated, bronchial artery embolization (Figure 5) is a very successful minimally invasive therapy that controls bleeding immediately in 66% to 90% of patients.^1,7,21 It is the preferred emergency treatment for massive hemoptysis, as the death rate is 7.1% to 18.2%, which, though high, is considerably less than the 40% seen in emergency surgery for massive hemoptysis.⁶

If a patient with massive hemoptysis undergoes successful bronchial artery embolization but the bleeding recurs 1 to 6 months later, the cause is likely an undetected nonbronchial systemic arterial supply and incomplete embolization.^1,22 Late rebleeding (6–12 months after the procedure) occurs in 20% to 40% of patients and is likely to be from disease progression.^1,7

Common complications of bronchial artery embolization are transient chest pain and dysphagia. Very rare complications include subintimal dissection and spinal cord ischemia due to inadvertent occlusion of the spinal arteries.⁶ Another complication in patients with renal failure is contrast nephropathy, the risk of which must be weighed against the possible consequences—including death—of not performing bronchial artery embolization in a patient who cannot undergo surgery.

CASE REVISITED: CLINICAL COURSE

In the patient described at the beginning of this article, a chest radiograph obtained in the emergency room showed an area of nonspecific consolidation in the left upper lung. Conventional chest CT was then ordered (Figure 4), and it revealed a cavitary lesion in the left upper lobe, consistent with aspergilloma. Bronchoscopy was then performed, and it too indicated that the bleeding was coming from the left upper lobe. Samples obtained during the procedure were sent to the laboratory for bacterial and fungal cultures.

In the meantime, family members were contacted, and they revealed that the patient had a history of sarcoidosis.

The patient went on to develop massive hemoptysis. Although the treatment of choice for mycetoma is primary resection, our patient’s respiratory status was poor as a result of extensive pulmonary sarcoidosis, and he was not considered a candidate for emergency surgery at that time. He was rushed to the angiography suite and successfully underwent emergency bronchial artery embolization.

Staphylococcus aureus is rearing its ugly head in new and interesting ways, both in the hospital and in the community.

Rates of invasive infections with methicillin-resistant S aureus (MRSA) have been increasing both in the hospital and in the community, a trend that has attracted considerable interest in the lay media. Curiously, the most common community-associated MRSA strain, which up to now has been distinct from hospital-associated MRSA strains, is invading our hospitals. Alarmingly, vancomycin (Vancocin), the drug of last resort for MRSA infections for the past 40 years, does not seem to be as effective as it used to be.

This paper summarizes the changing epidemiology of S aureus, particularly the emergence of MRSA outside of the hospital; reviews the difficulties associated with S aureus bacteremia and its treatment in view of; some changes in vancomycin susceptibility; and appraises the old and new treatment options.

MRSA IS ON THE RISE IN THE HOSPITAL

S aureus, a gram-positive, coagulase-positive bacterium, is one of the leading nosocomial bloodstream pathogens, second only to coagulase-negative staphylococci.¹ And the incidence of S aureus infections is increasing. MRSA in particular is increasingly causing infections throughout hospitals, including intensive care units. As of 2004, nearly two-thirds of isolates of S aureus from intensive care units were MRSA.²

MRSA infections are worse than methicillin-susceptible S aureus (MSSA) infections in terms of the rates of death and other undesirable outcomes.³ Several factors may be responsible: MRSA infection may be a marker of severity of illness (sicker patients may be more likely to have MRSA), our treatment for MRSA may not be as effective as it is for MSSA, and the organism may be inherently more virulent.

METHICILLIN RESISTANCE IS ALSO ON THE RISE IN THE COMMUNITY

Community-associated MRSA began emerging clinically about 10 years ago. It was first described in a cohort of children with necrotizing pneumonia in Minnesota, but soon other populations at risk began to emerge, such as residents of correctional facilities, men who had sex with men, competitive athletes (eg, fencers, wrestlers, and football players), and Alaskan natives and other native populations. A common factor in all these groups was close proximity of the members to each other. Later, it began to spread beyond these traditional risk groups into the community at large.

Community-associated MRSA strains have a characteristic pattern of antimicrobial susceptibility (see below). In the laboratory, they grow somewhat faster than health-care-associated MRSA strains, but not as fast as MSSA. They have a strong association with skin and soft-tissue infections: when you see a skin or soft-tissue infection, be it in an outpatient or an inpatient, think about MRSA. Their virulence varies, but rapid onset and progression of illness are quite common. Their most common strain in the United States at present is USA 300.

Case 1: A young woman with necrotizing fasciitis

A 21-year-old college student presented to our service in May 2004 with high fever and severe arm pain, which had been worsening for several days. She had been previously healthy, had not had any contact with the health care system, and had not received any antibiotics.

Her arm was red and extremely painful (Figure 1). Pain out of proportion to findings is a cardinal sign of necrotizing fasciitis, and this is indeed what she had.

Her blood cultures were positive for MRSA, as were cultures of the deep tissue of the deltoid muscle and fascia when she underwent emergency surgical debridement. The infection required several additional surgical debridements and removal of one head of her deltoid muscle, but she was fortunate: in the past, some patients with this problem might have undergone radical amputation of the arm or even more extensive surgery. This patient continued to have positive blood cultures 4 days postoperatively, but she ultimately recovered, completing 28 days of daptomycin (Cubicin) therapy at a dose of 6 mg/kg every 24 hours. The last 10 days of daptomycin therapy were given at home via a percutaneous intravenous central catheter.

Comment. The epidemiology of MRSA infections is changing. More patients who have no traditional risk factors, specifically health care contact, are getting MRSA infections. A recent report from the US Centers for Disease Control and Prevention (CDC) indicates that the proportion of patients with invasive disease due to MRSA has doubled since 2001–2002.⁴ Part of the reason undoubtedly is that MRSA, particularly community-associated MRSA, often carries specific virulence factors that make it more invasive. The CDC estimated that in 2005 there were nearly 100,000 cases of invasive MRSA infection in the United States, and nearly a fifth of these infections resulted in death.

 

 

Resistance and virulence factors in community-associated MRSA

Most community-associated MRSA strains carry a mobile genetic element called type IV SCCmec (staphylococcal chromosomal cassettemec) that enhances its antimicrobial resistance. This genetic component was probably borrowed from coagulase-negative staphylococci, in which it is quite common but does not cause as much of a problem. It is now present in a wide range of S aureus strains. Most of the S aureus strains that carry type IV SCCmec are MRSA, but a few MSSA strains do carry it as well.

The potent toxin Panton-Valentine leukocidin is an extracellular product that is detected in fewer than 5% of hospital strains but is more common in community-associated strains. It kills leukocytes by forming pores in the cell membrane and causing skin necrosis in cutaneous infections. It is associated with skin abscesses and rapidly progressive necrotizing pneumonia in MSSA or MRSA.

Epidemiologic differences between community- and health-care-associated MRSA

Patients with community-associated MRSA infections tend to be younger than those who traditionally get health-care-associated MRSA infections: in a study from Naimi et al in 2003, the mean ages were 23 vs 68 years.⁵ A greater proportion of patients with community-associated MRSA strains are nonwhite.⁴,⁵

Most community-associated MRSA infections are of the skin and soft tissue (75% in the series from Naimi et al⁵), but this pathogen causes other infections as well. Bacteremia of unknown origin has been seen, as has necrotizing pneumonia. Most of the skin and soft-tissue infections are relatively superficial, such as folliculitis or furunculosis, but deeper tissue infections such as necrotizing fasciitis and pyomyositis have also been seen.⁶

The incidence of community-associated MRSA infections varies greatly by geographic region.⁷ The northeastern United States has so far been relatively spared, but in Atlanta, Houston, and Los Angeles up to 80% of cases of characteristic skin or soft-tissue infections seen in emergency or outpatient departments are due to community-associated MRSA. Physicians at the Texas Children’s Hospital in Houston assume that all skin or soft-tissue infections are due to community-associated MRSA unless proven otherwise.⁸

Differences in antibiotic susceptibility

Community-associated MRSA is more susceptible to various antibiotics than health-care-associated MRSA,⁵ but not by much. Strains are usually susceptible to vancomycin, tetracyclines, trimethoprim-sulfamethoxazole (Bactrim, Septra), and rifampin (Rifadin). Unlike hospital strains, a fair number of community-acquired strains are susceptible to clindamycin (Cleocin) in the laboratory, but with a caveat: some of these clindamycin-susceptible strains actually may harbor the tools for inducible resistance. In fact, they can become resistant to clindamycin even without being exposed to it.

The laboratory test for inducible clindamycin resistance is called the D test. After coating an agar plate with S aureus, the technician places erythromycin and clindamycin disks. If the erythromycin induces clindamycin resistance, the plate is clear of growth around the clindamycin disk except for the portion nearest the erythromycin disk, leaving a characteristic D-shaped area of lucency.

Risk factors for MRSA

Moran et al⁷ analyzed the risk factors for community-associated MRSA in patients with skin or soft-tissue infections seen in the emergency department. The infection was more likely to be due to community-associated MRSA if the patient was black, had used any antibiotic in the past month, had a history of MRSA infection, or had close contact with a person with a similar infection. Many patients interpreted the infections as spider bites because the lesions tended to have a dark center surrounded by a tender area. These infections were not associated with underlying illness. In some cases, community-associated MRSA skin infections have been associated with tattooing and even manicuring.

However, it is very difficult to distinguish between community-associated MRSA and MSSA skin and soft-tissue infections on the basis of clinical and epidemiologic characteristics. Miller et al⁹ studied a large group of patients in Los Angeles who were hospitalized with community-associated skin and soft-tissue S aureus infections. All the patients were followed up for 30 days after hospital discharge. Regardless of whether they had MRSA or MSSA, they had similar outcomes. Close contacts of the patients also tended to develop infection.

A key point from this and many other studies: patients were more likely to remain infected if they did not undergo incision and drainage. This key intervention is indicated for any patient who has a skin and soft-tissue infection with an undrained focus of infection.

 

COMMUNITY-ASSOCIATED MRSA IS INVADING THE HOSPITAL

In a new development, community-associated MRSA strains are now appearing in the hospital. This is not only because patients are bacteremic when they come in: patients in the hospital are getting nosocomial infections due to community-associated MRSA strains.

Seybold et al¹⁰ analyzed 116 cases of MRSA bloodstream infections in Atlanta, GA. In 9 (8%) of the cases the patient had not had any contact with the health care system within the past year, and these cases were classified as truly community-associated. Of the remaining 107 cases, 49 (42%) were nosocomial, and the USA 300 strain—the predominant community-associated MRSA strain—accounted for 10 (20%) of the nosocomial cases.

In the recent CDC study of invasive MRSA infections, Klevens et al⁴ reported that nearly a third of cases of bacteremia were due to community-associated MRSA, and these strains accounted for a greater proportion of cases of cellulitis and endocarditis than did health-care-associated strains.

In a study of hospital-associated MRSA, Maree et al¹¹ found that the percentage of cases in which the bacteria carried the SCCmec type IV marker had increased from less than 20% in 1999 to more than 50% in 2004.

Comment. Suffice it to say that we are surrounded by MRSA. Community-associated MRSA is here to stay. It is even invading our hospitals, and we need to consider this very carefully when choosing antimicrobial therapy.

NAGGING QUESTIONS ABOUT VANCOMYCIN

Case 2: Vancomycin-intermediate S aureus (VISA) bacteremia and endocarditis

In December 2006 we saw a very ill 60-year-old woman who was hospitalized with MRSA bacteremia, pacemaker endocarditis, and superior vena cava thrombosis. Although she was treated with vancomycin and rifampin, her condition worsened, she had a stroke, and she developed renal failure. In a difficult operation, the pacemaker was removed, but the bacteremia persisted. In early February 2007 she underwent another difficult operation in which the superior vena cava clot was debrided, a right atrial clot was removed, and her mitral valve was replaced. Less than 2 weeks later, and despite ongoing vancomycin and rifampin therapy, the MRSA bacteremia recurred.

During the approximately 6 weeks that the patient had been receiving these antibiotics, the minimal inhibitory concentration (MIC) of rifampin against the S aureus isolate increased from less than 1 μg/mL (susceptible) to 2 μg/mL (resistant). The MIC of vancomycin went from 2 μg/mL (susceptible) to 4 μg/mL (intermediately susceptible). Vancomycin and rifampin were discontinued, and daptomycin and gentamicin (Garamycin) therapy were started. (Her daptomycin MIC was 0.5 μg/mL). The patient’s condition stabilized, and she was discharged to a long-term nursing facility. She had no relapse of MRSA bacteremia, but she died in early April of that year.

Is vancomycin becoming less effective? Degrees of vancomycin resistance

Vancomycin has been our stalwart for treating MRSA infections for more than 40 years but it is not working as well as it used to, at least in certain situations.

Liu C, Chambers HF. Staphylococcus aureus with heterogeneous resistance to vancomycin: epidemiology, clinical significance, and critical assessment of diagnostic methods. Antimicrob Agents Chemother 2003; 47:3040–3045

Various terms describe the degrees of susceptibility or resistance to vancomycin (Figure 2).¹²

VRSA (vancomycin-resistant S aureus) is rare. These fully resistant strains probably acquired a resistance mechanism (the vanA operon) from vancomycin-resistant enterococci. Infections tend to occur in patients simultaneously infected with both S aureus and vancomycin-resistant enterococci, giving the bacteria an opportunity to exchange genetic material.

VISA (vancomycin-intermediate S aureus) infections tend to occur in patients like the one described above who have had long-term vancomycin therapy. VISA strains appear to overproduce a matrix that captures vancomycin and keeps it from entering the cell. On electron microscopy, these bacteria have a very thick cell wall.¹³

Vancomycin tolerance is a state in which the bacteria are “stunned” or kept in check but not killed by vancomycin. That is manifested in the laboratory by a ratio of minimum bactericidal concentration to MIC greater than 32.

hVISA (heteroresistant VISA) is new and worrisome. These organisms have an overall MIC in the susceptible range, but within that population are individual isolates with an MIC that is much higher—in the intermediate or perhaps even in the resistant range.¹⁴

Reported rates of hVISA vary from less than 2% to as high as 76%, because the methods for detecting it are still very poorly standardized. The usual automated laboratory tests do not detect hVISA.

hVISA is probably clinically relevant, as evidence is emerging both in vitro and in vivo that the higher the MIC for vancomycin, the worse the clinical outcome.¹⁵ hVISA has been associated with failures of therapy in several situations, usually in cases of severe invasive or deep infection, endocarditis, and bacteremia with vertebral osteomyelitis where vancomycin concentrations at the site of infection may be suboptimal.^16–19 While most hVISA strains that have been described were resistant to methicillin, some were susceptible.

The E test is emerging as the standard test for hVISA. This test uses a plastic strip that contains gradually increasing concentrations of vancomycin along its length. Placed in the culture dish, the strip inhibits growth of the organism at its high-concentration end but not at its low-concentration end. If the sample contains hVISA, the cutoff is not well defined, with a few colonies growing at higher concentrations.

 

 

New definition of vancomycin susceptibility

Recognizing that the MICs for vancomycin have been rising in the last few years, the Clinical and Laboratory Standards Institute last year changed the break points between susceptibility and resistance. The new definitions are:

• Susceptible—an MIC of 2.0 μg/mL or less (formerly 4.0 μg/mL or less)
• Intermediate—4.0 to 8.0 μg/mL (formerly 8.0 to 16 μg/mL)
• Resistant—16 μg/mL or greater (formerly 32 μg/mL or greater).

One should pay attention to the MIC numbers on the laboratory reports, not just to the words “susceptible” or “not susceptible.” If the number is, say, 0.5 μg/mL or less, the organism should really be susceptible. If the number is 1 or 2, it is still in the susceptible range, but those are the organisms that may cause problems later on.

Further, even if the vancomycin MIC is in the susceptible range, higher MICs may affect outcomes. The average duration of MRSA bacteremia on therapy is 8 to 9 days, vs 3 to 4 days with MSSA bacteremia.²⁰,²¹ But Sakoulas et al¹⁵ found that, in MRSA bacteremia, the success rate with vancomycin therapy was 56% if the MIC was 0.5 or lower, compared with 10% if the MIC was 1.0 to 2.0 μg/mL. Examined in another way, the success rate was 50% if the logarithm of killing was 6.27 colony-forming units per mL or greater, 23% if 4.71 to 6.26, and zero if less than 4.71.

Case 3: Prolonged MRSA bacteremia

In the summer of 2006, a 66-year-old woman with a history of gastric bypass and cirrhosis underwent a long stay in the surgical intensive care unit because of a recurrent enterocutaneous fistula and chronic renal insufficiency. On November 5th, she had a positive blood culture for MRSA, which was treated appropriately with vancomycin for 4 weeks. She was discharged to subacute care but came back 2 days later, again with MRSA bacteremia. At that time her Hickman catheter, which had been inserted for total parenteral nutrition because of the enterocutaneous fistula, was removed.

Transthoracic echocardiography revealed no vegetations, but her bacteremia persisted. Her mental status was poor this entire time: she was mute and could barely be awakened. We looked for clots and infected clots; duplex ultrasonographic examinations of all four extremities were negative. Finally, magnetic resonance imaging of her back—performed empirically because of the persistent bacteremia—revealed vertebral osteomyelitis at level T12-L1. We also noticed on serial evaluations that the vancomycin MIC for her organism increased from 0.5 to 2.0 μg/mL, so therapy was changed from vancomycin to daptomycin.

Her bacteremia cleared. Follow-up echocardiography was negative, but she had two subsequent relapses of MRSA bacteremia, one in April 2007 and one before she died in the summer of 2007.

Prolonged bacteremia: Is it vancomycin resistance, or something else?

The MRSA isolates that cause prolonged bacteremia seem to have certain characteristics.²² Higher MICs are probably associated with longer periods of bacteremia. But some genetic components within some strains of S aureus give them a survival advantage. They have less susceptibility to the body’s thrombin-induced platelet microbicidal protein. These isolates are not only associated with prolonged bacteremia: they are also associated with osteomyelitis, deep abscesses, endocarditis, recurrent infection, and increased death rate.²² Clinical laboratories do not test for these genetic components. One wonders whether our patient may have had an isolate with these mutations that gave it a survival advantage.

Do not use vancomycin for MSSA

Avoid using vancomycin for MSSA infections. It has been shown time and time again that MSSA infections do not respond as well to vancomycin as they do to beta-lactam antibiotics, specifically to the semisynthetic penicillins such as oxacillin and nafcillin, and even some of the first-generation cephalosporins. Chang et al²³ found that patients with MSSA bacteremia had higher rates of persistent infections, relapse, and bacteriologic failure if they received vancomycin than if they received nafcillin.

Do vancomycin trough levels affect toxicity?

The vancomycin trough levels that we aimed for in the past (5 to 10 μg/mL) were probably too low. Today, we aim for trough levels of 15 to 20 μg/mL, and many physicians are aiming for 20 to 25 μg/mL. Part of the reason is that vancomycin MICs are higher than they used to be: in order to keep the vancomycin level above the MIC for a longer period of time, the vancomycin trough level needs to be higher. In theory, keeping the vancomycin levels above the MIC for longer periods should improve outcomes. Yet Fowler et al²² found that vancomycin trough levels among patients who had persistent MRSA bacteremia were actually higher than trough levels among those in whom the bacteremia resolved, although the difference was not statistically significant.

We measure the vancomycin trough level to make sure it is high enough (and give larger doses if it is not); among adults, peak levels need not be monitored on a routine basis because of the predictable pharmacokinetics of vancomycin.

Vancomycin toxicity can be either idiosyncratic or synergistic. Idiosyncratic toxicity occurs when a patient who has been on vancomycin for a long time develops a fixed rash, not associated with infusion. This is an immunologic phenomenon. It is a rare and very serious situation and may require steroid therapy.

Synergistic toxicity occurs when vancomycin is given with other nephrotoxic agents, notably gentamicin. Vancomycin plus gentamicin equals nephrotoxicity. Vancomycin alone is usually not nephrotoxic, but close monitoring of renal function parameters is warranted with the use of higher doses.²⁴

 

IN UNEXPLAINED BACTEREMIA, LOOK FOR ENDOCARDITIS

In blood cultures from patients with bacteremia, S aureus is never a contaminant. Even if just one blood culture is positive for S aureus, believe that S aureus is the culprit.

Reports in the 1950s suggested that at least half of patients who had S aureus bacteremia had endocarditis,²⁵ leading to recommendations that all patients with S aureus bacteremia without an obvious primary source of infection should be evaluated for endocarditis. Subsequent estimates were lower, in the range of 15% to 25%.²⁶,²⁷ However, throughout the world S aureus endocarditis continues to have a very high mortality rate: at least a third of patients die.²⁸

Clinical criteria (community acquisition, no primary focus, and metastatic sequelae) were developed to try to predict the risk of endocarditis in bacteremic patients.²⁶ However, these criteria did not work very well. The clinical definition of endocarditis has evolved. The criteria of von Reyn et al²⁹ from 1981 did not use echocardiography as part of the definition, but the 1994 Duke criteria,³⁰ which were refined³¹ in 2000, use both clinical and echocardiographic parameters.

Stratton et al³² performed transthoracic echocardiography in 14 patients with bacteremia and found 1 patient with cryptic tricuspid infective endocarditis. Bayer et al³³ subsequently reported that of 72 patients with bacteremia, 6 (18%) of those who had no clinical findings suggestive of infectious endocarditis had findings on echocardiography that led to changes in their regimen. Adding echocardiography to three clinical risk factors increased the sensitivity of diagnosing endocarditis from 70% to 85% with a specificity of 100% and predictive value of 96%.

The Duke criteria call for transesophageal echocardiography, which is not feasible in some patients, eg, those with cirrhosis and esophageal varices.

S aureus endocarditis has changed over the years as our patient population has changed, and MRSA endocarditis tends to hit some of our most vulnerable patients. In a study by Miro et al³⁴ in 2005, MRSA was the leading pathogen in patients who were diagnosed with S aureus endocarditis in 1990 or later. We will only see these numbers go up. Patients with diabetes tend to have more MRSA, and of diabetic patients with MRSA endocarditis, 30% to 40% die in the hospital.

Indications for surgery

Certain conditions are indications for surgery among patients with endocarditis, and no antibiotic will cure the endocarditis if the patient has one of these conditions, eg:

• Persistent bacteremia during antibiotic therapy
• Recurrent emboli
• Heart failure that cannot be controlled
• Perivalvular or myocardial abscesses
• Large vegetations
• Early prosthetic valve infection
• Certain arrhythmias.

How long should S aureus bacteremia be treated?

In cases of bacteremia in which endocarditis has been ruled out and removable foci of infection (eg, intravascular catheters) have been removed, some evidence indicates that treatment for 2 weeks would be as effective as the 4 to 6 weeks that we would use for endocarditis or other severe or invasive infections.³⁵ The issue is controversial. If the patient has had frequent hospitalizations or a chronic medical condition I would hesitate to treat for less than 4 weeks, even if the infection appears to be associated with a removable focus.

Treatment of endocarditis

In the guidelines for treatment of endocarditis from the American Heart Association and Infectious Diseases Society of America,³⁶ all the recommendations are relatively old and many of them are somewhat empiric—they are not based on evidence from randomized clinical trials. Rather, they are best opinions based on clinical experience and some observational studies over the years.

For MSSA. In cases of native-valve endocarditis, oxacillin (Bactocill), nafcillin (Unipen), or another semisynthetic beta-lactam antibiotic is recommended. For penicillin-allergic patients, we have other options, such as cefazolin (Ancef, Kefzol).

Combination therapy is frequently recommended for native valve endocarditis as well as for prosthetic valve endocarditis, with either rifampin or gentamicin along with a primary agent. There is some evidence that one can clear staphylococcal bacteremia a day or two more quickly by use of combination therapy with nafcillin plus an aminoglycoside than with nafcillin alone.³⁷,³⁸ For MSSA-associated endocarditis, vancomycin does not work as well as beta-lactam antibiotics.³⁹,⁴⁰

Korzeniowski and Sande³⁷ and Chambers et al³⁸ reported that the mean duration of bacteremia was 3.4 days for patients treated with nafcillin alone and 2.9 days for those treated with nafcillin plus an aminoglycoside. These studies led to consideration of a short course of gentamicin to clear the bacteremia quickly.

With MRSA, bacteremia often requires a week or more to clear. Levine et al²¹ reported a study in 42 patients, mostly injection-drug users, with right-sided native-valve endocarditis. The median duration of bacteremia was 7 days in patients who received vancomycin alone vs 9 days in those who received vancomycin plus rifampin; however, some patients were bacteremic for up to 27 days. Fever persisted for a median of 7 days, probably partly due to septic pulmonary emboli. Three patients died, and three required valve replacement.

 

NEW ANTIBIOTICS

Several new antibiotics are active against gram-positive cocci.^41–44 However, the majority of them have not been prospectively studied for treating bacteremia or endocarditis.

Quinupristin/dalfopristin (Synercid) has not been formally studied for treatment of MRSA bacteremia or endocarditis. There are a few case reports of its use in these conditions.⁴⁵ Quinupristin/dalfopristin is bacteriostatic, and its use may be associated with phlebitis, myalgias, and arthralgias.⁴⁶

Linezolid (Zyvox) is approved for treatment of complicated skin and soft-tissue infections and for hospital-acquired pneumonia. There have been no specific studies of linezolid in the treatment of S aureus bacteremia or endocarditis. However, Shorr et al⁴⁷ retrospectively looked at the bacteremic patients in five previous studies of linezolid vs vancomycin and found 144 cases of S aureus bacteremia, half of which were due to MRSA. Of 53 assessable patients with MRSA bacteremia, the primary infection was cured in 14 (56%) of the linezolid patients and 13 (46%) of the vancomycin patients.

The oral form is 100% bioavailable. One should avoid concomitant use of serotonin-reuptake inhibitors because of the risk of serotonin syndrome. Adverse effects include altered taste sensation and peripheral neuropathy. There are other potential toxicities, including hematologic changes (thrombocytopenia, leukopenia) and metabolic effects (lactic acidosis), so clinical and laboratory monitoring is important.⁴⁸ The role of linezolid in the treatment of patients with S aureus bacteremia or endocarditis remains to be defined.

Daptomycin is indicated for complicated skin and soft-tissue infections, bacteremia, and right-sided endocarditis due to S aureus. Fowler et al²⁰ found that daptomycin was not inferior to beta-lactam antibiotics for treatment of MSSA bacteremia and right-sided endocarditis, and for MRSA infections it outperformed vancomycin, but the difference was not statistically significant.

The dosing interval should be increased from once every 24 hours to every 48 hours if the creatinine clearance is 30 mL/minute or less. Adverse effects include myalgia, rhabdomyolysis (rare), and elevations in creatine phosphokinase. Reports of rising MICs during daptomycin therapy, in some cases associated with persistent infection,⁴⁹ suggest that careful attention be paid to dosing and clinical monitoring.

Tigecycline (Tygacil) is indicated for complicated skin and soft-tissue infections and complicated intra-abdominal infections due to susceptible organisms. It is active against both MSSA and MRSA, but clinical experience with its use in invasive infections is somewhat limited.⁵⁰ The dose of tigecycline should be reduced in advanced cirrhosis. Adverse effects include nausea and vomiting.

Telavancin, dalbavancin, and oritavancin, investigational parenteral antibiotics that are derivatives of vancomycin, are in clinical trials. The pharmacokinetic activity of these agents is of interest: telavancin is being studied with a once-daily dosing interval and dalbavancin’s half-life allows once-weekly dosing. In a limited trial, dalbavancin was found to be safe and effective in the treatment of catheter-related bloodstream infections.⁵¹ None of the antibiotics in this group has been studied for treatment of S aureus endocarditis. Telavancin therapy has been associated with rash, hypokalemia, QT prolongation, and creatinine elevations. Gastrointestinal symptoms have been reported with the use of dalbavancin.

Ceftobiprole, another investigational agent, is the only cephalosporin antibiotic that is active against MRSA. It is given every 12 hours. Adverse effects include nausea and taste disturbance.

Iclaprim is a novel diaminopyrimidine and a dihydrofolate reductase inhibitor. In vitro, it is active against gram-positive bacteria, including MRSA, VISA, and VRSA; clinical investigations at this point are limited to the treatment of skin and soft-tissue infections.

Staphylococcus aureus is rearing its ugly head in new and interesting ways, both in the hospital and in the community.

Rates of invasive infections with methicillin-resistant S aureus (MRSA) have been increasing both in the hospital and in the community, a trend that has attracted considerable interest in the lay media. Curiously, the most common community-associated MRSA strain, which up to now has been distinct from hospital-associated MRSA strains, is invading our hospitals. Alarmingly, vancomycin (Vancocin), the drug of last resort for MRSA infections for the past 40 years, does not seem to be as effective as it used to be.

This paper summarizes the changing epidemiology of S aureus, particularly the emergence of MRSA outside of the hospital; reviews the difficulties associated with S aureus bacteremia and its treatment in view of; some changes in vancomycin susceptibility; and appraises the old and new treatment options.

MRSA IS ON THE RISE IN THE HOSPITAL

S aureus, a gram-positive, coagulase-positive bacterium, is one of the leading nosocomial bloodstream pathogens, second only to coagulase-negative staphylococci.¹ And the incidence of S aureus infections is increasing. MRSA in particular is increasingly causing infections throughout hospitals, including intensive care units. As of 2004, nearly two-thirds of isolates of S aureus from intensive care units were MRSA.²

MRSA infections are worse than methicillin-susceptible S aureus (MSSA) infections in terms of the rates of death and other undesirable outcomes.³ Several factors may be responsible: MRSA infection may be a marker of severity of illness (sicker patients may be more likely to have MRSA), our treatment for MRSA may not be as effective as it is for MSSA, and the organism may be inherently more virulent.

METHICILLIN RESISTANCE IS ALSO ON THE RISE IN THE COMMUNITY

Community-associated MRSA began emerging clinically about 10 years ago. It was first described in a cohort of children with necrotizing pneumonia in Minnesota, but soon other populations at risk began to emerge, such as residents of correctional facilities, men who had sex with men, competitive athletes (eg, fencers, wrestlers, and football players), and Alaskan natives and other native populations. A common factor in all these groups was close proximity of the members to each other. Later, it began to spread beyond these traditional risk groups into the community at large.

Community-associated MRSA strains have a characteristic pattern of antimicrobial susceptibility (see below). In the laboratory, they grow somewhat faster than health-care-associated MRSA strains, but not as fast as MSSA. They have a strong association with skin and soft-tissue infections: when you see a skin or soft-tissue infection, be it in an outpatient or an inpatient, think about MRSA. Their virulence varies, but rapid onset and progression of illness are quite common. Their most common strain in the United States at present is USA 300.

Case 1: A young woman with necrotizing fasciitis

A 21-year-old college student presented to our service in May 2004 with high fever and severe arm pain, which had been worsening for several days. She had been previously healthy, had not had any contact with the health care system, and had not received any antibiotics.

Her arm was red and extremely painful (Figure 1). Pain out of proportion to findings is a cardinal sign of necrotizing fasciitis, and this is indeed what she had.

Her blood cultures were positive for MRSA, as were cultures of the deep tissue of the deltoid muscle and fascia when she underwent emergency surgical debridement. The infection required several additional surgical debridements and removal of one head of her deltoid muscle, but she was fortunate: in the past, some patients with this problem might have undergone radical amputation of the arm or even more extensive surgery. This patient continued to have positive blood cultures 4 days postoperatively, but she ultimately recovered, completing 28 days of daptomycin (Cubicin) therapy at a dose of 6 mg/kg every 24 hours. The last 10 days of daptomycin therapy were given at home via a percutaneous intravenous central catheter.

Comment. The epidemiology of MRSA infections is changing. More patients who have no traditional risk factors, specifically health care contact, are getting MRSA infections. A recent report from the US Centers for Disease Control and Prevention (CDC) indicates that the proportion of patients with invasive disease due to MRSA has doubled since 2001–2002.⁴ Part of the reason undoubtedly is that MRSA, particularly community-associated MRSA, often carries specific virulence factors that make it more invasive. The CDC estimated that in 2005 there were nearly 100,000 cases of invasive MRSA infection in the United States, and nearly a fifth of these infections resulted in death.

 

 

Resistance and virulence factors in community-associated MRSA

Most community-associated MRSA strains carry a mobile genetic element called type IV SCCmec (staphylococcal chromosomal cassettemec) that enhances its antimicrobial resistance. This genetic component was probably borrowed from coagulase-negative staphylococci, in which it is quite common but does not cause as much of a problem. It is now present in a wide range of S aureus strains. Most of the S aureus strains that carry type IV SCCmec are MRSA, but a few MSSA strains do carry it as well.

The potent toxin Panton-Valentine leukocidin is an extracellular product that is detected in fewer than 5% of hospital strains but is more common in community-associated strains. It kills leukocytes by forming pores in the cell membrane and causing skin necrosis in cutaneous infections. It is associated with skin abscesses and rapidly progressive necrotizing pneumonia in MSSA or MRSA.

Epidemiologic differences between community- and health-care-associated MRSA

Patients with community-associated MRSA infections tend to be younger than those who traditionally get health-care-associated MRSA infections: in a study from Naimi et al in 2003, the mean ages were 23 vs 68 years.⁵ A greater proportion of patients with community-associated MRSA strains are nonwhite.⁴,⁵

Most community-associated MRSA infections are of the skin and soft tissue (75% in the series from Naimi et al⁵), but this pathogen causes other infections as well. Bacteremia of unknown origin has been seen, as has necrotizing pneumonia. Most of the skin and soft-tissue infections are relatively superficial, such as folliculitis or furunculosis, but deeper tissue infections such as necrotizing fasciitis and pyomyositis have also been seen.⁶

The incidence of community-associated MRSA infections varies greatly by geographic region.⁷ The northeastern United States has so far been relatively spared, but in Atlanta, Houston, and Los Angeles up to 80% of cases of characteristic skin or soft-tissue infections seen in emergency or outpatient departments are due to community-associated MRSA. Physicians at the Texas Children’s Hospital in Houston assume that all skin or soft-tissue infections are due to community-associated MRSA unless proven otherwise.⁸

Differences in antibiotic susceptibility

Community-associated MRSA is more susceptible to various antibiotics than health-care-associated MRSA,⁵ but not by much. Strains are usually susceptible to vancomycin, tetracyclines, trimethoprim-sulfamethoxazole (Bactrim, Septra), and rifampin (Rifadin). Unlike hospital strains, a fair number of community-acquired strains are susceptible to clindamycin (Cleocin) in the laboratory, but with a caveat: some of these clindamycin-susceptible strains actually may harbor the tools for inducible resistance. In fact, they can become resistant to clindamycin even without being exposed to it.

The laboratory test for inducible clindamycin resistance is called the D test. After coating an agar plate with S aureus, the technician places erythromycin and clindamycin disks. If the erythromycin induces clindamycin resistance, the plate is clear of growth around the clindamycin disk except for the portion nearest the erythromycin disk, leaving a characteristic D-shaped area of lucency.

Risk factors for MRSA

Moran et al⁷ analyzed the risk factors for community-associated MRSA in patients with skin or soft-tissue infections seen in the emergency department. The infection was more likely to be due to community-associated MRSA if the patient was black, had used any antibiotic in the past month, had a history of MRSA infection, or had close contact with a person with a similar infection. Many patients interpreted the infections as spider bites because the lesions tended to have a dark center surrounded by a tender area. These infections were not associated with underlying illness. In some cases, community-associated MRSA skin infections have been associated with tattooing and even manicuring.

However, it is very difficult to distinguish between community-associated MRSA and MSSA skin and soft-tissue infections on the basis of clinical and epidemiologic characteristics. Miller et al⁹ studied a large group of patients in Los Angeles who were hospitalized with community-associated skin and soft-tissue S aureus infections. All the patients were followed up for 30 days after hospital discharge. Regardless of whether they had MRSA or MSSA, they had similar outcomes. Close contacts of the patients also tended to develop infection.

A key point from this and many other studies: patients were more likely to remain infected if they did not undergo incision and drainage. This key intervention is indicated for any patient who has a skin and soft-tissue infection with an undrained focus of infection.

 

COMMUNITY-ASSOCIATED MRSA IS INVADING THE HOSPITAL

In a new development, community-associated MRSA strains are now appearing in the hospital. This is not only because patients are bacteremic when they come in: patients in the hospital are getting nosocomial infections due to community-associated MRSA strains.

Seybold et al¹⁰ analyzed 116 cases of MRSA bloodstream infections in Atlanta, GA. In 9 (8%) of the cases the patient had not had any contact with the health care system within the past year, and these cases were classified as truly community-associated. Of the remaining 107 cases, 49 (42%) were nosocomial, and the USA 300 strain—the predominant community-associated MRSA strain—accounted for 10 (20%) of the nosocomial cases.

In the recent CDC study of invasive MRSA infections, Klevens et al⁴ reported that nearly a third of cases of bacteremia were due to community-associated MRSA, and these strains accounted for a greater proportion of cases of cellulitis and endocarditis than did health-care-associated strains.

In a study of hospital-associated MRSA, Maree et al¹¹ found that the percentage of cases in which the bacteria carried the SCCmec type IV marker had increased from less than 20% in 1999 to more than 50% in 2004.

Comment. Suffice it to say that we are surrounded by MRSA. Community-associated MRSA is here to stay. It is even invading our hospitals, and we need to consider this very carefully when choosing antimicrobial therapy.

NAGGING QUESTIONS ABOUT VANCOMYCIN

Case 2: Vancomycin-intermediate S aureus (VISA) bacteremia and endocarditis

In December 2006 we saw a very ill 60-year-old woman who was hospitalized with MRSA bacteremia, pacemaker endocarditis, and superior vena cava thrombosis. Although she was treated with vancomycin and rifampin, her condition worsened, she had a stroke, and she developed renal failure. In a difficult operation, the pacemaker was removed, but the bacteremia persisted. In early February 2007 she underwent another difficult operation in which the superior vena cava clot was debrided, a right atrial clot was removed, and her mitral valve was replaced. Less than 2 weeks later, and despite ongoing vancomycin and rifampin therapy, the MRSA bacteremia recurred.

During the approximately 6 weeks that the patient had been receiving these antibiotics, the minimal inhibitory concentration (MIC) of rifampin against the S aureus isolate increased from less than 1 μg/mL (susceptible) to 2 μg/mL (resistant). The MIC of vancomycin went from 2 μg/mL (susceptible) to 4 μg/mL (intermediately susceptible). Vancomycin and rifampin were discontinued, and daptomycin and gentamicin (Garamycin) therapy were started. (Her daptomycin MIC was 0.5 μg/mL). The patient’s condition stabilized, and she was discharged to a long-term nursing facility. She had no relapse of MRSA bacteremia, but she died in early April of that year.

Is vancomycin becoming less effective? Degrees of vancomycin resistance

Vancomycin has been our stalwart for treating MRSA infections for more than 40 years but it is not working as well as it used to, at least in certain situations.

Liu C, Chambers HF. Staphylococcus aureus with heterogeneous resistance to vancomycin: epidemiology, clinical significance, and critical assessment of diagnostic methods. Antimicrob Agents Chemother 2003; 47:3040–3045

Various terms describe the degrees of susceptibility or resistance to vancomycin (Figure 2).¹²

VRSA (vancomycin-resistant S aureus) is rare. These fully resistant strains probably acquired a resistance mechanism (the vanA operon) from vancomycin-resistant enterococci. Infections tend to occur in patients simultaneously infected with both S aureus and vancomycin-resistant enterococci, giving the bacteria an opportunity to exchange genetic material.

VISA (vancomycin-intermediate S aureus) infections tend to occur in patients like the one described above who have had long-term vancomycin therapy. VISA strains appear to overproduce a matrix that captures vancomycin and keeps it from entering the cell. On electron microscopy, these bacteria have a very thick cell wall.¹³

Vancomycin tolerance is a state in which the bacteria are “stunned” or kept in check but not killed by vancomycin. That is manifested in the laboratory by a ratio of minimum bactericidal concentration to MIC greater than 32.

hVISA (heteroresistant VISA) is new and worrisome. These organisms have an overall MIC in the susceptible range, but within that population are individual isolates with an MIC that is much higher—in the intermediate or perhaps even in the resistant range.¹⁴

Reported rates of hVISA vary from less than 2% to as high as 76%, because the methods for detecting it are still very poorly standardized. The usual automated laboratory tests do not detect hVISA.

hVISA is probably clinically relevant, as evidence is emerging both in vitro and in vivo that the higher the MIC for vancomycin, the worse the clinical outcome.¹⁵ hVISA has been associated with failures of therapy in several situations, usually in cases of severe invasive or deep infection, endocarditis, and bacteremia with vertebral osteomyelitis where vancomycin concentrations at the site of infection may be suboptimal.^16–19 While most hVISA strains that have been described were resistant to methicillin, some were susceptible.

The E test is emerging as the standard test for hVISA. This test uses a plastic strip that contains gradually increasing concentrations of vancomycin along its length. Placed in the culture dish, the strip inhibits growth of the organism at its high-concentration end but not at its low-concentration end. If the sample contains hVISA, the cutoff is not well defined, with a few colonies growing at higher concentrations.

 

 

New definition of vancomycin susceptibility

Recognizing that the MICs for vancomycin have been rising in the last few years, the Clinical and Laboratory Standards Institute last year changed the break points between susceptibility and resistance. The new definitions are:

• Susceptible—an MIC of 2.0 μg/mL or less (formerly 4.0 μg/mL or less)
• Intermediate—4.0 to 8.0 μg/mL (formerly 8.0 to 16 μg/mL)
• Resistant—16 μg/mL or greater (formerly 32 μg/mL or greater).

One should pay attention to the MIC numbers on the laboratory reports, not just to the words “susceptible” or “not susceptible.” If the number is, say, 0.5 μg/mL or less, the organism should really be susceptible. If the number is 1 or 2, it is still in the susceptible range, but those are the organisms that may cause problems later on.

Further, even if the vancomycin MIC is in the susceptible range, higher MICs may affect outcomes. The average duration of MRSA bacteremia on therapy is 8 to 9 days, vs 3 to 4 days with MSSA bacteremia.²⁰,²¹ But Sakoulas et al¹⁵ found that, in MRSA bacteremia, the success rate with vancomycin therapy was 56% if the MIC was 0.5 or lower, compared with 10% if the MIC was 1.0 to 2.0 μg/mL. Examined in another way, the success rate was 50% if the logarithm of killing was 6.27 colony-forming units per mL or greater, 23% if 4.71 to 6.26, and zero if less than 4.71.

Case 3: Prolonged MRSA bacteremia

In the summer of 2006, a 66-year-old woman with a history of gastric bypass and cirrhosis underwent a long stay in the surgical intensive care unit because of a recurrent enterocutaneous fistula and chronic renal insufficiency. On November 5th, she had a positive blood culture for MRSA, which was treated appropriately with vancomycin for 4 weeks. She was discharged to subacute care but came back 2 days later, again with MRSA bacteremia. At that time her Hickman catheter, which had been inserted for total parenteral nutrition because of the enterocutaneous fistula, was removed.

Transthoracic echocardiography revealed no vegetations, but her bacteremia persisted. Her mental status was poor this entire time: she was mute and could barely be awakened. We looked for clots and infected clots; duplex ultrasonographic examinations of all four extremities were negative. Finally, magnetic resonance imaging of her back—performed empirically because of the persistent bacteremia—revealed vertebral osteomyelitis at level T12-L1. We also noticed on serial evaluations that the vancomycin MIC for her organism increased from 0.5 to 2.0 μg/mL, so therapy was changed from vancomycin to daptomycin.

Her bacteremia cleared. Follow-up echocardiography was negative, but she had two subsequent relapses of MRSA bacteremia, one in April 2007 and one before she died in the summer of 2007.

Prolonged bacteremia: Is it vancomycin resistance, or something else?

The MRSA isolates that cause prolonged bacteremia seem to have certain characteristics.²² Higher MICs are probably associated with longer periods of bacteremia. But some genetic components within some strains of S aureus give them a survival advantage. They have less susceptibility to the body’s thrombin-induced platelet microbicidal protein. These isolates are not only associated with prolonged bacteremia: they are also associated with osteomyelitis, deep abscesses, endocarditis, recurrent infection, and increased death rate.²² Clinical laboratories do not test for these genetic components. One wonders whether our patient may have had an isolate with these mutations that gave it a survival advantage.

Do not use vancomycin for MSSA

Avoid using vancomycin for MSSA infections. It has been shown time and time again that MSSA infections do not respond as well to vancomycin as they do to beta-lactam antibiotics, specifically to the semisynthetic penicillins such as oxacillin and nafcillin, and even some of the first-generation cephalosporins. Chang et al²³ found that patients with MSSA bacteremia had higher rates of persistent infections, relapse, and bacteriologic failure if they received vancomycin than if they received nafcillin.

Do vancomycin trough levels affect toxicity?

The vancomycin trough levels that we aimed for in the past (5 to 10 μg/mL) were probably too low. Today, we aim for trough levels of 15 to 20 μg/mL, and many physicians are aiming for 20 to 25 μg/mL. Part of the reason is that vancomycin MICs are higher than they used to be: in order to keep the vancomycin level above the MIC for a longer period of time, the vancomycin trough level needs to be higher. In theory, keeping the vancomycin levels above the MIC for longer periods should improve outcomes. Yet Fowler et al²² found that vancomycin trough levels among patients who had persistent MRSA bacteremia were actually higher than trough levels among those in whom the bacteremia resolved, although the difference was not statistically significant.

We measure the vancomycin trough level to make sure it is high enough (and give larger doses if it is not); among adults, peak levels need not be monitored on a routine basis because of the predictable pharmacokinetics of vancomycin.

Vancomycin toxicity can be either idiosyncratic or synergistic. Idiosyncratic toxicity occurs when a patient who has been on vancomycin for a long time develops a fixed rash, not associated with infusion. This is an immunologic phenomenon. It is a rare and very serious situation and may require steroid therapy.

Synergistic toxicity occurs when vancomycin is given with other nephrotoxic agents, notably gentamicin. Vancomycin plus gentamicin equals nephrotoxicity. Vancomycin alone is usually not nephrotoxic, but close monitoring of renal function parameters is warranted with the use of higher doses.²⁴

 

IN UNEXPLAINED BACTEREMIA, LOOK FOR ENDOCARDITIS

In blood cultures from patients with bacteremia, S aureus is never a contaminant. Even if just one blood culture is positive for S aureus, believe that S aureus is the culprit.

Reports in the 1950s suggested that at least half of patients who had S aureus bacteremia had endocarditis,²⁵ leading to recommendations that all patients with S aureus bacteremia without an obvious primary source of infection should be evaluated for endocarditis. Subsequent estimates were lower, in the range of 15% to 25%.²⁶,²⁷ However, throughout the world S aureus endocarditis continues to have a very high mortality rate: at least a third of patients die.²⁸

Clinical criteria (community acquisition, no primary focus, and metastatic sequelae) were developed to try to predict the risk of endocarditis in bacteremic patients.²⁶ However, these criteria did not work very well. The clinical definition of endocarditis has evolved. The criteria of von Reyn et al²⁹ from 1981 did not use echocardiography as part of the definition, but the 1994 Duke criteria,³⁰ which were refined³¹ in 2000, use both clinical and echocardiographic parameters.

Stratton et al³² performed transthoracic echocardiography in 14 patients with bacteremia and found 1 patient with cryptic tricuspid infective endocarditis. Bayer et al³³ subsequently reported that of 72 patients with bacteremia, 6 (18%) of those who had no clinical findings suggestive of infectious endocarditis had findings on echocardiography that led to changes in their regimen. Adding echocardiography to three clinical risk factors increased the sensitivity of diagnosing endocarditis from 70% to 85% with a specificity of 100% and predictive value of 96%.

The Duke criteria call for transesophageal echocardiography, which is not feasible in some patients, eg, those with cirrhosis and esophageal varices.

S aureus endocarditis has changed over the years as our patient population has changed, and MRSA endocarditis tends to hit some of our most vulnerable patients. In a study by Miro et al³⁴ in 2005, MRSA was the leading pathogen in patients who were diagnosed with S aureus endocarditis in 1990 or later. We will only see these numbers go up. Patients with diabetes tend to have more MRSA, and of diabetic patients with MRSA endocarditis, 30% to 40% die in the hospital.

Indications for surgery

Certain conditions are indications for surgery among patients with endocarditis, and no antibiotic will cure the endocarditis if the patient has one of these conditions, eg:

• Persistent bacteremia during antibiotic therapy
• Recurrent emboli
• Heart failure that cannot be controlled
• Perivalvular or myocardial abscesses
• Large vegetations
• Early prosthetic valve infection
• Certain arrhythmias.

How long should S aureus bacteremia be treated?

In cases of bacteremia in which endocarditis has been ruled out and removable foci of infection (eg, intravascular catheters) have been removed, some evidence indicates that treatment for 2 weeks would be as effective as the 4 to 6 weeks that we would use for endocarditis or other severe or invasive infections.³⁵ The issue is controversial. If the patient has had frequent hospitalizations or a chronic medical condition I would hesitate to treat for less than 4 weeks, even if the infection appears to be associated with a removable focus.

Treatment of endocarditis

In the guidelines for treatment of endocarditis from the American Heart Association and Infectious Diseases Society of America,³⁶ all the recommendations are relatively old and many of them are somewhat empiric—they are not based on evidence from randomized clinical trials. Rather, they are best opinions based on clinical experience and some observational studies over the years.

For MSSA. In cases of native-valve endocarditis, oxacillin (Bactocill), nafcillin (Unipen), or another semisynthetic beta-lactam antibiotic is recommended. For penicillin-allergic patients, we have other options, such as cefazolin (Ancef, Kefzol).

Combination therapy is frequently recommended for native valve endocarditis as well as for prosthetic valve endocarditis, with either rifampin or gentamicin along with a primary agent. There is some evidence that one can clear staphylococcal bacteremia a day or two more quickly by use of combination therapy with nafcillin plus an aminoglycoside than with nafcillin alone.³⁷,³⁸ For MSSA-associated endocarditis, vancomycin does not work as well as beta-lactam antibiotics.³⁹,⁴⁰

Korzeniowski and Sande³⁷ and Chambers et al³⁸ reported that the mean duration of bacteremia was 3.4 days for patients treated with nafcillin alone and 2.9 days for those treated with nafcillin plus an aminoglycoside. These studies led to consideration of a short course of gentamicin to clear the bacteremia quickly.

With MRSA, bacteremia often requires a week or more to clear. Levine et al²¹ reported a study in 42 patients, mostly injection-drug users, with right-sided native-valve endocarditis. The median duration of bacteremia was 7 days in patients who received vancomycin alone vs 9 days in those who received vancomycin plus rifampin; however, some patients were bacteremic for up to 27 days. Fever persisted for a median of 7 days, probably partly due to septic pulmonary emboli. Three patients died, and three required valve replacement.

 

NEW ANTIBIOTICS

Several new antibiotics are active against gram-positive cocci.^41–44 However, the majority of them have not been prospectively studied for treating bacteremia or endocarditis.

Quinupristin/dalfopristin (Synercid) has not been formally studied for treatment of MRSA bacteremia or endocarditis. There are a few case reports of its use in these conditions.⁴⁵ Quinupristin/dalfopristin is bacteriostatic, and its use may be associated with phlebitis, myalgias, and arthralgias.⁴⁶

Linezolid (Zyvox) is approved for treatment of complicated skin and soft-tissue infections and for hospital-acquired pneumonia. There have been no specific studies of linezolid in the treatment of S aureus bacteremia or endocarditis. However, Shorr et al⁴⁷ retrospectively looked at the bacteremic patients in five previous studies of linezolid vs vancomycin and found 144 cases of S aureus bacteremia, half of which were due to MRSA. Of 53 assessable patients with MRSA bacteremia, the primary infection was cured in 14 (56%) of the linezolid patients and 13 (46%) of the vancomycin patients.

The oral form is 100% bioavailable. One should avoid concomitant use of serotonin-reuptake inhibitors because of the risk of serotonin syndrome. Adverse effects include altered taste sensation and peripheral neuropathy. There are other potential toxicities, including hematologic changes (thrombocytopenia, leukopenia) and metabolic effects (lactic acidosis), so clinical and laboratory monitoring is important.⁴⁸ The role of linezolid in the treatment of patients with S aureus bacteremia or endocarditis remains to be defined.

Daptomycin is indicated for complicated skin and soft-tissue infections, bacteremia, and right-sided endocarditis due to S aureus. Fowler et al²⁰ found that daptomycin was not inferior to beta-lactam antibiotics for treatment of MSSA bacteremia and right-sided endocarditis, and for MRSA infections it outperformed vancomycin, but the difference was not statistically significant.

The dosing interval should be increased from once every 24 hours to every 48 hours if the creatinine clearance is 30 mL/minute or less. Adverse effects include myalgia, rhabdomyolysis (rare), and elevations in creatine phosphokinase. Reports of rising MICs during daptomycin therapy, in some cases associated with persistent infection,⁴⁹ suggest that careful attention be paid to dosing and clinical monitoring.

Tigecycline (Tygacil) is indicated for complicated skin and soft-tissue infections and complicated intra-abdominal infections due to susceptible organisms. It is active against both MSSA and MRSA, but clinical experience with its use in invasive infections is somewhat limited.⁵⁰ The dose of tigecycline should be reduced in advanced cirrhosis. Adverse effects include nausea and vomiting.

Telavancin, dalbavancin, and oritavancin, investigational parenteral antibiotics that are derivatives of vancomycin, are in clinical trials. The pharmacokinetic activity of these agents is of interest: telavancin is being studied with a once-daily dosing interval and dalbavancin’s half-life allows once-weekly dosing. In a limited trial, dalbavancin was found to be safe and effective in the treatment of catheter-related bloodstream infections.⁵¹ None of the antibiotics in this group has been studied for treatment of S aureus endocarditis. Telavancin therapy has been associated with rash, hypokalemia, QT prolongation, and creatinine elevations. Gastrointestinal symptoms have been reported with the use of dalbavancin.

Ceftobiprole, another investigational agent, is the only cephalosporin antibiotic that is active against MRSA. It is given every 12 hours. Adverse effects include nausea and taste disturbance.

Iclaprim is a novel diaminopyrimidine and a dihydrofolate reductase inhibitor. In vitro, it is active against gram-positive bacteria, including MRSA, VISA, and VRSA; clinical investigations at this point are limited to the treatment of skin and soft-tissue infections.

Staphylococcus aureus is rearing its ugly head in new and interesting ways, both in the hospital and in the community.

Rates of invasive infections with methicillin-resistant S aureus (MRSA) have been increasing both in the hospital and in the community, a trend that has attracted considerable interest in the lay media. Curiously, the most common community-associated MRSA strain, which up to now has been distinct from hospital-associated MRSA strains, is invading our hospitals. Alarmingly, vancomycin (Vancocin), the drug of last resort for MRSA infections for the past 40 years, does not seem to be as effective as it used to be.

This paper summarizes the changing epidemiology of S aureus, particularly the emergence of MRSA outside of the hospital; reviews the difficulties associated with S aureus bacteremia and its treatment in view of; some changes in vancomycin susceptibility; and appraises the old and new treatment options.

MRSA IS ON THE RISE IN THE HOSPITAL

S aureus, a gram-positive, coagulase-positive bacterium, is one of the leading nosocomial bloodstream pathogens, second only to coagulase-negative staphylococci.¹ And the incidence of S aureus infections is increasing. MRSA in particular is increasingly causing infections throughout hospitals, including intensive care units. As of 2004, nearly two-thirds of isolates of S aureus from intensive care units were MRSA.²

MRSA infections are worse than methicillin-susceptible S aureus (MSSA) infections in terms of the rates of death and other undesirable outcomes.³ Several factors may be responsible: MRSA infection may be a marker of severity of illness (sicker patients may be more likely to have MRSA), our treatment for MRSA may not be as effective as it is for MSSA, and the organism may be inherently more virulent.

METHICILLIN RESISTANCE IS ALSO ON THE RISE IN THE COMMUNITY

Community-associated MRSA began emerging clinically about 10 years ago. It was first described in a cohort of children with necrotizing pneumonia in Minnesota, but soon other populations at risk began to emerge, such as residents of correctional facilities, men who had sex with men, competitive athletes (eg, fencers, wrestlers, and football players), and Alaskan natives and other native populations. A common factor in all these groups was close proximity of the members to each other. Later, it began to spread beyond these traditional risk groups into the community at large.

Community-associated MRSA strains have a characteristic pattern of antimicrobial susceptibility (see below). In the laboratory, they grow somewhat faster than health-care-associated MRSA strains, but not as fast as MSSA. They have a strong association with skin and soft-tissue infections: when you see a skin or soft-tissue infection, be it in an outpatient or an inpatient, think about MRSA. Their virulence varies, but rapid onset and progression of illness are quite common. Their most common strain in the United States at present is USA 300.

Case 1: A young woman with necrotizing fasciitis

A 21-year-old college student presented to our service in May 2004 with high fever and severe arm pain, which had been worsening for several days. She had been previously healthy, had not had any contact with the health care system, and had not received any antibiotics.

Her arm was red and extremely painful (Figure 1). Pain out of proportion to findings is a cardinal sign of necrotizing fasciitis, and this is indeed what she had.

Her blood cultures were positive for MRSA, as were cultures of the deep tissue of the deltoid muscle and fascia when she underwent emergency surgical debridement. The infection required several additional surgical debridements and removal of one head of her deltoid muscle, but she was fortunate: in the past, some patients with this problem might have undergone radical amputation of the arm or even more extensive surgery. This patient continued to have positive blood cultures 4 days postoperatively, but she ultimately recovered, completing 28 days of daptomycin (Cubicin) therapy at a dose of 6 mg/kg every 24 hours. The last 10 days of daptomycin therapy were given at home via a percutaneous intravenous central catheter.

Comment. The epidemiology of MRSA infections is changing. More patients who have no traditional risk factors, specifically health care contact, are getting MRSA infections. A recent report from the US Centers for Disease Control and Prevention (CDC) indicates that the proportion of patients with invasive disease due to MRSA has doubled since 2001–2002.⁴ Part of the reason undoubtedly is that MRSA, particularly community-associated MRSA, often carries specific virulence factors that make it more invasive. The CDC estimated that in 2005 there were nearly 100,000 cases of invasive MRSA infection in the United States, and nearly a fifth of these infections resulted in death.

 

 

Resistance and virulence factors in community-associated MRSA

Most community-associated MRSA strains carry a mobile genetic element called type IV SCCmec (staphylococcal chromosomal cassettemec) that enhances its antimicrobial resistance. This genetic component was probably borrowed from coagulase-negative staphylococci, in which it is quite common but does not cause as much of a problem. It is now present in a wide range of S aureus strains. Most of the S aureus strains that carry type IV SCCmec are MRSA, but a few MSSA strains do carry it as well.

The potent toxin Panton-Valentine leukocidin is an extracellular product that is detected in fewer than 5% of hospital strains but is more common in community-associated strains. It kills leukocytes by forming pores in the cell membrane and causing skin necrosis in cutaneous infections. It is associated with skin abscesses and rapidly progressive necrotizing pneumonia in MSSA or MRSA.

Epidemiologic differences between community- and health-care-associated MRSA

Patients with community-associated MRSA infections tend to be younger than those who traditionally get health-care-associated MRSA infections: in a study from Naimi et al in 2003, the mean ages were 23 vs 68 years.⁵ A greater proportion of patients with community-associated MRSA strains are nonwhite.⁴,⁵

Most community-associated MRSA infections are of the skin and soft tissue (75% in the series from Naimi et al⁵), but this pathogen causes other infections as well. Bacteremia of unknown origin has been seen, as has necrotizing pneumonia. Most of the skin and soft-tissue infections are relatively superficial, such as folliculitis or furunculosis, but deeper tissue infections such as necrotizing fasciitis and pyomyositis have also been seen.⁶

The incidence of community-associated MRSA infections varies greatly by geographic region.⁷ The northeastern United States has so far been relatively spared, but in Atlanta, Houston, and Los Angeles up to 80% of cases of characteristic skin or soft-tissue infections seen in emergency or outpatient departments are due to community-associated MRSA. Physicians at the Texas Children’s Hospital in Houston assume that all skin or soft-tissue infections are due to community-associated MRSA unless proven otherwise.⁸

Differences in antibiotic susceptibility

Community-associated MRSA is more susceptible to various antibiotics than health-care-associated MRSA,⁵ but not by much. Strains are usually susceptible to vancomycin, tetracyclines, trimethoprim-sulfamethoxazole (Bactrim, Septra), and rifampin (Rifadin). Unlike hospital strains, a fair number of community-acquired strains are susceptible to clindamycin (Cleocin) in the laboratory, but with a caveat: some of these clindamycin-susceptible strains actually may harbor the tools for inducible resistance. In fact, they can become resistant to clindamycin even without being exposed to it.

The laboratory test for inducible clindamycin resistance is called the D test. After coating an agar plate with S aureus, the technician places erythromycin and clindamycin disks. If the erythromycin induces clindamycin resistance, the plate is clear of growth around the clindamycin disk except for the portion nearest the erythromycin disk, leaving a characteristic D-shaped area of lucency.

Risk factors for MRSA

Moran et al⁷ analyzed the risk factors for community-associated MRSA in patients with skin or soft-tissue infections seen in the emergency department. The infection was more likely to be due to community-associated MRSA if the patient was black, had used any antibiotic in the past month, had a history of MRSA infection, or had close contact with a person with a similar infection. Many patients interpreted the infections as spider bites because the lesions tended to have a dark center surrounded by a tender area. These infections were not associated with underlying illness. In some cases, community-associated MRSA skin infections have been associated with tattooing and even manicuring.

However, it is very difficult to distinguish between community-associated MRSA and MSSA skin and soft-tissue infections on the basis of clinical and epidemiologic characteristics. Miller et al⁹ studied a large group of patients in Los Angeles who were hospitalized with community-associated skin and soft-tissue S aureus infections. All the patients were followed up for 30 days after hospital discharge. Regardless of whether they had MRSA or MSSA, they had similar outcomes. Close contacts of the patients also tended to develop infection.

A key point from this and many other studies: patients were more likely to remain infected if they did not undergo incision and drainage. This key intervention is indicated for any patient who has a skin and soft-tissue infection with an undrained focus of infection.

 

COMMUNITY-ASSOCIATED MRSA IS INVADING THE HOSPITAL

In a new development, community-associated MRSA strains are now appearing in the hospital. This is not only because patients are bacteremic when they come in: patients in the hospital are getting nosocomial infections due to community-associated MRSA strains.

Seybold et al¹⁰ analyzed 116 cases of MRSA bloodstream infections in Atlanta, GA. In 9 (8%) of the cases the patient had not had any contact with the health care system within the past year, and these cases were classified as truly community-associated. Of the remaining 107 cases, 49 (42%) were nosocomial, and the USA 300 strain—the predominant community-associated MRSA strain—accounted for 10 (20%) of the nosocomial cases.

In the recent CDC study of invasive MRSA infections, Klevens et al⁴ reported that nearly a third of cases of bacteremia were due to community-associated MRSA, and these strains accounted for a greater proportion of cases of cellulitis and endocarditis than did health-care-associated strains.

In a study of hospital-associated MRSA, Maree et al¹¹ found that the percentage of cases in which the bacteria carried the SCCmec type IV marker had increased from less than 20% in 1999 to more than 50% in 2004.

Comment. Suffice it to say that we are surrounded by MRSA. Community-associated MRSA is here to stay. It is even invading our hospitals, and we need to consider this very carefully when choosing antimicrobial therapy.

NAGGING QUESTIONS ABOUT VANCOMYCIN

Case 2: Vancomycin-intermediate S aureus (VISA) bacteremia and endocarditis

In December 2006 we saw a very ill 60-year-old woman who was hospitalized with MRSA bacteremia, pacemaker endocarditis, and superior vena cava thrombosis. Although she was treated with vancomycin and rifampin, her condition worsened, she had a stroke, and she developed renal failure. In a difficult operation, the pacemaker was removed, but the bacteremia persisted. In early February 2007 she underwent another difficult operation in which the superior vena cava clot was debrided, a right atrial clot was removed, and her mitral valve was replaced. Less than 2 weeks later, and despite ongoing vancomycin and rifampin therapy, the MRSA bacteremia recurred.

During the approximately 6 weeks that the patient had been receiving these antibiotics, the minimal inhibitory concentration (MIC) of rifampin against the S aureus isolate increased from less than 1 μg/mL (susceptible) to 2 μg/mL (resistant). The MIC of vancomycin went from 2 μg/mL (susceptible) to 4 μg/mL (intermediately susceptible). Vancomycin and rifampin were discontinued, and daptomycin and gentamicin (Garamycin) therapy were started. (Her daptomycin MIC was 0.5 μg/mL). The patient’s condition stabilized, and she was discharged to a long-term nursing facility. She had no relapse of MRSA bacteremia, but she died in early April of that year.

Is vancomycin becoming less effective? Degrees of vancomycin resistance

Vancomycin has been our stalwart for treating MRSA infections for more than 40 years but it is not working as well as it used to, at least in certain situations.

Liu C, Chambers HF. Staphylococcus aureus with heterogeneous resistance to vancomycin: epidemiology, clinical significance, and critical assessment of diagnostic methods. Antimicrob Agents Chemother 2003; 47:3040–3045

Various terms describe the degrees of susceptibility or resistance to vancomycin (Figure 2).¹²

VRSA (vancomycin-resistant S aureus) is rare. These fully resistant strains probably acquired a resistance mechanism (the vanA operon) from vancomycin-resistant enterococci. Infections tend to occur in patients simultaneously infected with both S aureus and vancomycin-resistant enterococci, giving the bacteria an opportunity to exchange genetic material.

VISA (vancomycin-intermediate S aureus) infections tend to occur in patients like the one described above who have had long-term vancomycin therapy. VISA strains appear to overproduce a matrix that captures vancomycin and keeps it from entering the cell. On electron microscopy, these bacteria have a very thick cell wall.¹³

Vancomycin tolerance is a state in which the bacteria are “stunned” or kept in check but not killed by vancomycin. That is manifested in the laboratory by a ratio of minimum bactericidal concentration to MIC greater than 32.

hVISA (heteroresistant VISA) is new and worrisome. These organisms have an overall MIC in the susceptible range, but within that population are individual isolates with an MIC that is much higher—in the intermediate or perhaps even in the resistant range.¹⁴

Reported rates of hVISA vary from less than 2% to as high as 76%, because the methods for detecting it are still very poorly standardized. The usual automated laboratory tests do not detect hVISA.

hVISA is probably clinically relevant, as evidence is emerging both in vitro and in vivo that the higher the MIC for vancomycin, the worse the clinical outcome.¹⁵ hVISA has been associated with failures of therapy in several situations, usually in cases of severe invasive or deep infection, endocarditis, and bacteremia with vertebral osteomyelitis where vancomycin concentrations at the site of infection may be suboptimal.^16–19 While most hVISA strains that have been described were resistant to methicillin, some were susceptible.

The E test is emerging as the standard test for hVISA. This test uses a plastic strip that contains gradually increasing concentrations of vancomycin along its length. Placed in the culture dish, the strip inhibits growth of the organism at its high-concentration end but not at its low-concentration end. If the sample contains hVISA, the cutoff is not well defined, with a few colonies growing at higher concentrations.

 

 

New definition of vancomycin susceptibility

Recognizing that the MICs for vancomycin have been rising in the last few years, the Clinical and Laboratory Standards Institute last year changed the break points between susceptibility and resistance. The new definitions are:

• Susceptible—an MIC of 2.0 μg/mL or less (formerly 4.0 μg/mL or less)
• Intermediate—4.0 to 8.0 μg/mL (formerly 8.0 to 16 μg/mL)
• Resistant—16 μg/mL or greater (formerly 32 μg/mL or greater).

One should pay attention to the MIC numbers on the laboratory reports, not just to the words “susceptible” or “not susceptible.” If the number is, say, 0.5 μg/mL or less, the organism should really be susceptible. If the number is 1 or 2, it is still in the susceptible range, but those are the organisms that may cause problems later on.

Further, even if the vancomycin MIC is in the susceptible range, higher MICs may affect outcomes. The average duration of MRSA bacteremia on therapy is 8 to 9 days, vs 3 to 4 days with MSSA bacteremia.²⁰,²¹ But Sakoulas et al¹⁵ found that, in MRSA bacteremia, the success rate with vancomycin therapy was 56% if the MIC was 0.5 or lower, compared with 10% if the MIC was 1.0 to 2.0 μg/mL. Examined in another way, the success rate was 50% if the logarithm of killing was 6.27 colony-forming units per mL or greater, 23% if 4.71 to 6.26, and zero if less than 4.71.

Case 3: Prolonged MRSA bacteremia

In the summer of 2006, a 66-year-old woman with a history of gastric bypass and cirrhosis underwent a long stay in the surgical intensive care unit because of a recurrent enterocutaneous fistula and chronic renal insufficiency. On November 5th, she had a positive blood culture for MRSA, which was treated appropriately with vancomycin for 4 weeks. She was discharged to subacute care but came back 2 days later, again with MRSA bacteremia. At that time her Hickman catheter, which had been inserted for total parenteral nutrition because of the enterocutaneous fistula, was removed.

Transthoracic echocardiography revealed no vegetations, but her bacteremia persisted. Her mental status was poor this entire time: she was mute and could barely be awakened. We looked for clots and infected clots; duplex ultrasonographic examinations of all four extremities were negative. Finally, magnetic resonance imaging of her back—performed empirically because of the persistent bacteremia—revealed vertebral osteomyelitis at level T12-L1. We also noticed on serial evaluations that the vancomycin MIC for her organism increased from 0.5 to 2.0 μg/mL, so therapy was changed from vancomycin to daptomycin.

Her bacteremia cleared. Follow-up echocardiography was negative, but she had two subsequent relapses of MRSA bacteremia, one in April 2007 and one before she died in the summer of 2007.

Prolonged bacteremia: Is it vancomycin resistance, or something else?

The MRSA isolates that cause prolonged bacteremia seem to have certain characteristics.²² Higher MICs are probably associated with longer periods of bacteremia. But some genetic components within some strains of S aureus give them a survival advantage. They have less susceptibility to the body’s thrombin-induced platelet microbicidal protein. These isolates are not only associated with prolonged bacteremia: they are also associated with osteomyelitis, deep abscesses, endocarditis, recurrent infection, and increased death rate.²² Clinical laboratories do not test for these genetic components. One wonders whether our patient may have had an isolate with these mutations that gave it a survival advantage.

Do not use vancomycin for MSSA

Avoid using vancomycin for MSSA infections. It has been shown time and time again that MSSA infections do not respond as well to vancomycin as they do to beta-lactam antibiotics, specifically to the semisynthetic penicillins such as oxacillin and nafcillin, and even some of the first-generation cephalosporins. Chang et al²³ found that patients with MSSA bacteremia had higher rates of persistent infections, relapse, and bacteriologic failure if they received vancomycin than if they received nafcillin.

Do vancomycin trough levels affect toxicity?

The vancomycin trough levels that we aimed for in the past (5 to 10 μg/mL) were probably too low. Today, we aim for trough levels of 15 to 20 μg/mL, and many physicians are aiming for 20 to 25 μg/mL. Part of the reason is that vancomycin MICs are higher than they used to be: in order to keep the vancomycin level above the MIC for a longer period of time, the vancomycin trough level needs to be higher. In theory, keeping the vancomycin levels above the MIC for longer periods should improve outcomes. Yet Fowler et al²² found that vancomycin trough levels among patients who had persistent MRSA bacteremia were actually higher than trough levels among those in whom the bacteremia resolved, although the difference was not statistically significant.

We measure the vancomycin trough level to make sure it is high enough (and give larger doses if it is not); among adults, peak levels need not be monitored on a routine basis because of the predictable pharmacokinetics of vancomycin.

Vancomycin toxicity can be either idiosyncratic or synergistic. Idiosyncratic toxicity occurs when a patient who has been on vancomycin for a long time develops a fixed rash, not associated with infusion. This is an immunologic phenomenon. It is a rare and very serious situation and may require steroid therapy.

Synergistic toxicity occurs when vancomycin is given with other nephrotoxic agents, notably gentamicin. Vancomycin plus gentamicin equals nephrotoxicity. Vancomycin alone is usually not nephrotoxic, but close monitoring of renal function parameters is warranted with the use of higher doses.²⁴

 

IN UNEXPLAINED BACTEREMIA, LOOK FOR ENDOCARDITIS

In blood cultures from patients with bacteremia, S aureus is never a contaminant. Even if just one blood culture is positive for S aureus, believe that S aureus is the culprit.

Reports in the 1950s suggested that at least half of patients who had S aureus bacteremia had endocarditis,²⁵ leading to recommendations that all patients with S aureus bacteremia without an obvious primary source of infection should be evaluated for endocarditis. Subsequent estimates were lower, in the range of 15% to 25%.²⁶,²⁷ However, throughout the world S aureus endocarditis continues to have a very high mortality rate: at least a third of patients die.²⁸

Clinical criteria (community acquisition, no primary focus, and metastatic sequelae) were developed to try to predict the risk of endocarditis in bacteremic patients.²⁶ However, these criteria did not work very well. The clinical definition of endocarditis has evolved. The criteria of von Reyn et al²⁹ from 1981 did not use echocardiography as part of the definition, but the 1994 Duke criteria,³⁰ which were refined³¹ in 2000, use both clinical and echocardiographic parameters.

Stratton et al³² performed transthoracic echocardiography in 14 patients with bacteremia and found 1 patient with cryptic tricuspid infective endocarditis. Bayer et al³³ subsequently reported that of 72 patients with bacteremia, 6 (18%) of those who had no clinical findings suggestive of infectious endocarditis had findings on echocardiography that led to changes in their regimen. Adding echocardiography to three clinical risk factors increased the sensitivity of diagnosing endocarditis from 70% to 85% with a specificity of 100% and predictive value of 96%.

The Duke criteria call for transesophageal echocardiography, which is not feasible in some patients, eg, those with cirrhosis and esophageal varices.

S aureus endocarditis has changed over the years as our patient population has changed, and MRSA endocarditis tends to hit some of our most vulnerable patients. In a study by Miro et al³⁴ in 2005, MRSA was the leading pathogen in patients who were diagnosed with S aureus endocarditis in 1990 or later. We will only see these numbers go up. Patients with diabetes tend to have more MRSA, and of diabetic patients with MRSA endocarditis, 30% to 40% die in the hospital.

Indications for surgery

Certain conditions are indications for surgery among patients with endocarditis, and no antibiotic will cure the endocarditis if the patient has one of these conditions, eg:

• Persistent bacteremia during antibiotic therapy
• Recurrent emboli
• Heart failure that cannot be controlled
• Perivalvular or myocardial abscesses
• Large vegetations
• Early prosthetic valve infection
• Certain arrhythmias.

How long should S aureus bacteremia be treated?

In cases of bacteremia in which endocarditis has been ruled out and removable foci of infection (eg, intravascular catheters) have been removed, some evidence indicates that treatment for 2 weeks would be as effective as the 4 to 6 weeks that we would use for endocarditis or other severe or invasive infections.³⁵ The issue is controversial. If the patient has had frequent hospitalizations or a chronic medical condition I would hesitate to treat for less than 4 weeks, even if the infection appears to be associated with a removable focus.

Treatment of endocarditis

In the guidelines for treatment of endocarditis from the American Heart Association and Infectious Diseases Society of America,³⁶ all the recommendations are relatively old and many of them are somewhat empiric—they are not based on evidence from randomized clinical trials. Rather, they are best opinions based on clinical experience and some observational studies over the years.

For MSSA. In cases of native-valve endocarditis, oxacillin (Bactocill), nafcillin (Unipen), or another semisynthetic beta-lactam antibiotic is recommended. For penicillin-allergic patients, we have other options, such as cefazolin (Ancef, Kefzol).

Combination therapy is frequently recommended for native valve endocarditis as well as for prosthetic valve endocarditis, with either rifampin or gentamicin along with a primary agent. There is some evidence that one can clear staphylococcal bacteremia a day or two more quickly by use of combination therapy with nafcillin plus an aminoglycoside than with nafcillin alone.³⁷,³⁸ For MSSA-associated endocarditis, vancomycin does not work as well as beta-lactam antibiotics.³⁹,⁴⁰

Korzeniowski and Sande³⁷ and Chambers et al³⁸ reported that the mean duration of bacteremia was 3.4 days for patients treated with nafcillin alone and 2.9 days for those treated with nafcillin plus an aminoglycoside. These studies led to consideration of a short course of gentamicin to clear the bacteremia quickly.

With MRSA, bacteremia often requires a week or more to clear. Levine et al²¹ reported a study in 42 patients, mostly injection-drug users, with right-sided native-valve endocarditis. The median duration of bacteremia was 7 days in patients who received vancomycin alone vs 9 days in those who received vancomycin plus rifampin; however, some patients were bacteremic for up to 27 days. Fever persisted for a median of 7 days, probably partly due to septic pulmonary emboli. Three patients died, and three required valve replacement.

 

NEW ANTIBIOTICS

Several new antibiotics are active against gram-positive cocci.^41–44 However, the majority of them have not been prospectively studied for treating bacteremia or endocarditis.

Quinupristin/dalfopristin (Synercid) has not been formally studied for treatment of MRSA bacteremia or endocarditis. There are a few case reports of its use in these conditions.⁴⁵ Quinupristin/dalfopristin is bacteriostatic, and its use may be associated with phlebitis, myalgias, and arthralgias.⁴⁶

Linezolid (Zyvox) is approved for treatment of complicated skin and soft-tissue infections and for hospital-acquired pneumonia. There have been no specific studies of linezolid in the treatment of S aureus bacteremia or endocarditis. However, Shorr et al⁴⁷ retrospectively looked at the bacteremic patients in five previous studies of linezolid vs vancomycin and found 144 cases of S aureus bacteremia, half of which were due to MRSA. Of 53 assessable patients with MRSA bacteremia, the primary infection was cured in 14 (56%) of the linezolid patients and 13 (46%) of the vancomycin patients.

The oral form is 100% bioavailable. One should avoid concomitant use of serotonin-reuptake inhibitors because of the risk of serotonin syndrome. Adverse effects include altered taste sensation and peripheral neuropathy. There are other potential toxicities, including hematologic changes (thrombocytopenia, leukopenia) and metabolic effects (lactic acidosis), so clinical and laboratory monitoring is important.⁴⁸ The role of linezolid in the treatment of patients with S aureus bacteremia or endocarditis remains to be defined.

Daptomycin is indicated for complicated skin and soft-tissue infections, bacteremia, and right-sided endocarditis due to S aureus. Fowler et al²⁰ found that daptomycin was not inferior to beta-lactam antibiotics for treatment of MSSA bacteremia and right-sided endocarditis, and for MRSA infections it outperformed vancomycin, but the difference was not statistically significant.

The dosing interval should be increased from once every 24 hours to every 48 hours if the creatinine clearance is 30 mL/minute or less. Adverse effects include myalgia, rhabdomyolysis (rare), and elevations in creatine phosphokinase. Reports of rising MICs during daptomycin therapy, in some cases associated with persistent infection,⁴⁹ suggest that careful attention be paid to dosing and clinical monitoring.

Tigecycline (Tygacil) is indicated for complicated skin and soft-tissue infections and complicated intra-abdominal infections due to susceptible organisms. It is active against both MSSA and MRSA, but clinical experience with its use in invasive infections is somewhat limited.⁵⁰ The dose of tigecycline should be reduced in advanced cirrhosis. Adverse effects include nausea and vomiting.

Telavancin, dalbavancin, and oritavancin, investigational parenteral antibiotics that are derivatives of vancomycin, are in clinical trials. The pharmacokinetic activity of these agents is of interest: telavancin is being studied with a once-daily dosing interval and dalbavancin’s half-life allows once-weekly dosing. In a limited trial, dalbavancin was found to be safe and effective in the treatment of catheter-related bloodstream infections.⁵¹ None of the antibiotics in this group has been studied for treatment of S aureus endocarditis. Telavancin therapy has been associated with rash, hypokalemia, QT prolongation, and creatinine elevations. Gastrointestinal symptoms have been reported with the use of dalbavancin.

Ceftobiprole, another investigational agent, is the only cephalosporin antibiotic that is active against MRSA. It is given every 12 hours. Adverse effects include nausea and taste disturbance.

Iclaprim is a novel diaminopyrimidine and a dihydrofolate reductase inhibitor. In vitro, it is active against gram-positive bacteria, including MRSA, VISA, and VRSA; clinical investigations at this point are limited to the treatment of skin and soft-tissue infections.