Bisphosphonate therapy minimizes bone loss and reduces fracture risk by up to 50% in patients with osteoporosis,¹ but it is also associated with increased risks of osteonecrosis of the jaw and atypical femoral fracture. Although atypical femoral fractures are rare, they can have a devastating effect. Patient concern about this complication has contributed to a decrease in bisphosphonate use by about half in the last decade or so,^2,3 and we fear this could result in an increase in hip fracture rates.

In this article, we examine the evidence on bisphosphonate-associated atypical femoral fractures, including risks, pathogenesis, treatment, and prevention.

ATYPICAL FRACTURES INVOLVE THE FEMORAL SHAFT, NOT THE HEAD

An atypical femoral fracture is a transverse fracture of the femoral shaft (diaphysis), defined by both clinical criteria and radiographic appearance.

To be defined as atypical, a femoral fracture must meet 4 of the following 5 criteria⁴:

• Occurs with minimal or no trauma
• Has a predominantly transverse fracture line, originating at the lateral cortex and sometimes becoming oblique as it progresses medially across the femur
• Extends through both cortices and may be associated with a medial spike (complete fractures); or involves only the lateral cortex (incomplete fractures)
• Is noncomminuted or minimally comminuted
• Shows localized periosteal or endosteal thickening (termed “beaking” or “flaring”) of the lateral cortex at the fracture site.

Several minor features are also important but are not required, eg:

• Cortical thickening of the femoral shaft
• Unilateral or bilateral prodromal pain preceding the fracture
• Bilateral incomplete or complete femoral diaphysis fractures
• Delayed fracture healing.

Atypical femoral fracture can occur anywhere along the shaft, from just distal to the lesser trochanter to just proximal to the supracondylar flare. However, most occur in 2 areas, with 1 cluster centered at about 41 mm from the lesser trochanter (more common in relatively younger patients) and the other at 187 mm.⁵

ABSOLUTE RISK IS LOW BUT INCREASES WITH LONGER USE

Atypical femoral fractures are rare. Schilcher et al⁶ reviewed radiographs of 1,234 women who had a subtrochanteric or shaft fracture and found 59 (4.6%) of fractures were atypical. In a systematic review of 14 studies,⁷ the incidence ranged from 3.0 to 9.8 cases per 100,000 patient-years.

Furthermore, not all atypical femoral fractures are in bisphosphonate users: 7.4% were in nonusers in 1 series⁸ and 22% in another.⁹

Nevertheless, most studies show that bis­phosphonate use increases the incidence of atypical femoral fracture, and the incidence increases with duration of use, especially after 3 years.⁷

An international task force of the American Society for Bone and Mineral Research listed the absolute risk as between 3.2 and 50 cases per 100,000 patient-years, with longer use (> 5 years) increasing the risk to about 100 per 100,000 patient-years.⁴ After stopping bis­phosphonate therapy, the risk diminished by 70% per year.⁹

In another study, for 0.1 to 1.9 years of therapy, the age-adjusted atypical fracture rates were 1.78 per 100,000 per year (95% confidence interval [CI] 1.5–2.0), increasing to 113.1 per 100,000 per year (95% CI 69.3–156.8) with exposure from 8 to 9.9 years.¹⁰

A case-control study found that more than 5 years of bisphosphonate use increased the fracture risk by an odds ratio of 2.74 (95% CI 1.25–6.02).¹¹

The incidence of typical femoral fracture was higher in those who adhered better to their oral bisphosphonate regimen in some studies,¹² but the opposite was true in others.¹³

The benefits of bisphosphonate therapy in reducing fracture risk, however, outweigh the risk of atypical fracture.⁴

We do not know whether the rate of atypical femoral fracture is increasing. A review of Kaiser Permanente Northwest records found that the rates of atypical femoral shaft fracture had remained stable from 1996 to 2009. However, 61.9% of patients who met the strict radiographic criteria had taken oral bisphosphonates.¹⁴ These data suggest that bisphosphonate use has not increased the overall population-based risk for subtrochanteric and femoral shaft fractures, but that bisphosphonates and other risk factors may have increased the likelihood that such fractures will exhibit atypical radiographic features.

A population-based study in Denmark¹³ found that alendronate use longer than 10 years was associated with an adjusted 30% lower risk of hip fracture and no increase in the risk of subtrochanteric and femoral shaft fracture. In addition, the risk of subtrochanteric and femoral shaft fracture was lower with high adherence to alendronate treatment (based on medication possession ratio > 80%) compared with low adherence (ratio < 50%) (odds ratio 0.88, 95% CI 0.77–0.99). The risk was not increased in current vs past users.

The Danish study¹³ used the coding of the 10th revision of the International Statistical Classification of Diseases and Related Health Problems (ICD-10) to identify subtrochanteric and femoral shaft fractures without radiologic review for atypical radiographic features. The lack of specific ICD-10 coding for subtrochanteric and femoral shaft fractures with atypical radiographic features has limited our knowledge of their incidence.

After an atypical femoral fracture, patients have a significant risk of fracture on the contralateral side. In a case-control study, 28% of patients with atypical femoral fracture suffered a contralateral fracture, compared with 0.9% of patients presenting with a typical fracture pattern (odds ratio 42.6, 95% CI 12.8–142.4).¹⁵

Contralateral fracture occurs from 1 month to 4 years after the index atypical femoral fracture.¹⁶

There are reports of bisphosphonate-related low-impact fractures in other sites such as the tibia¹⁷ and forearm.¹⁸ However, they may be too rare to warrant screening.

Mortality rates

A Swedish database study found that patients with atypical femoral fractures, whether bisphosphonate users or nonusers, do not have higher mortality rates than patients with ordinary subtrochanteric or femoral shaft fractures.¹⁹ Furthermore, the mortality rates for those with atypical femoral fracture were similar to rates in the general population. In contrast, patients with an ordinary femoral fracture had a higher mortality risk than the general population.¹⁹

Other studies suggest that atypical femoral fracture may be associated with a less favorable prognosis in older patients,²⁰ but this could be due to differences in demographics, treatment adherence, or postfracture care.²¹

In addition, functional outcomes as measured by independent mobility at discharge and at 3 months were comparable between patients with atypical fracture and those with typical fracture.²²

IMAGING STUDIES

If a long-term bisphosphonate user presents with hip, thigh, or groin pain, imaging studies are recommended.

Plain radiography

Radiography is usually the first step and should include a frontal view of the pelvis (Figure 1) and 2 views of the full length of each femur. If radiography is not conclusive, bone scan or magnetic resonance imaging (MRI) should be considered.

A linear cortex transverse fracture pattern and focal lateral cortical thickening are the most sensitive and specific radiographic features.^23,24 Because of the risk of fracture on the contralateral side, radiographic study of that side is recommended as well.

Computed tomography

Computed tomography (CT) is not sensitive for early stress fractures and, given the radiation burden, is not recommended in the workup of atypical fracture.

Bone scanning

Bone scanning using technetium 99m-labeled methylene diphosphonate with a gamma camera shows active bone turnover. Stress fractures and atypical femoral fractures are most easily identified in the third (delayed) phase of the bone scan. Although bone scanning is highly sensitive, the specificity is limited by lack of spatial resolution. Atypical femoral fracture appears as increased activity in the subtrochanteric region with a predilection for the lateral cortex.

Dual-energy x-ray absorptiometry

Conventional dual-energy x-ray absorptiometry (DXA) extends only to 1 to 2 cm below the lesser trochanter and can therefore miss atypical fractures, which usually occur farther down. The overall detection rate for DXA was 61% in a sample of 33 patients.²⁵

Newer scanners can look at the entire femoral shaft.²⁶ In addition, newer software can quantify focal thickening (beaking) of the lateral cortex and screen patients who have no symptoms. The results of serial measurements can be graphed so that the practitioner can view trends to help assess or rule out potential asymptomatic atypical femoral fracture.

A localized reaction (periosteal thickening of the lateral cortex or beaking) often precedes atypical femoral fracture. A 2017 study reported that patients with high localized reaction (mean height 3.3 mm) that was of the pointed type and was accompanied by prodromal pain had an increased risk of complete or incomplete atypical femoral fracture at that site.²⁷ This finding is used by the newer DXA software. The predictive value of beaking on extended femoral DXA may be as high as 83%.²⁶

Magnetic resonance imaging

The MRI characteristics of atypical femoral fracture are similar to those of other stress fractures except that there is a lateral-to-medial pattern rather than a medial pattern. The earliest findings include periosteal reaction about the lateral cortex with a normal marrow signal.

MRI may be of particular benefit in patients with known atypical femoral fracture to screen the contralateral leg. It should image the entire length of both femurs. Contrast enhancement is not needed.

Regardless of whether initial findings were discovered on conventional radiographs or DXA, MRI confirmation is needed. Radio­nuclide bone scanning is currently not recommended because it lacks specificity. Combination imaging is recommended, with either radiography plus MRI or DXA plus MRI.

DIFFERENTIAL DIAGNOSIS

The differential diagnosis of atypical femoral fracture includes stress fracture, pathologic fracture, hypophosphatasia, and osteogenesis imperfecta.²⁸ Hypophosphatemic osteomalacia can cause Looser zones, which can be confused with atypical femoral fractures but usually occur on the medial side.⁴ Stress fracture of the femur can occur below the lesser trochanter but usually begins in the medial, not the lateral, cortex.

Pathologic fractures from underlying osseous lesions can mimic the cortical beaking of bisphosphonate-related fracture, but they usually show the associated underlying lucent lesion and poorly defined margins. A sinus tract along the region of a chronic osteomyelitis may also appear similar.

Hypophosphatasia is an inborn error of metabolism caused by a loss-of-function mutation in the gene encoding alkaline phosphatase, resulting in pyrophosphate accumulation and causing osteomalacia from impaired mineralization. This can result in femoral pseudofracture that is often bilateral and occurs in the subtrochanteric region.²⁹

Patients with atypical femoral fracture are generally a heterogeneous group, but there are risk factors to note other than bisphosphonate exposure.

Asian women had a risk 8 times higher than white women in 1 study.³⁰

Bone geometry. Mahjoub et al⁸ reported that compared with controls, patients with atypical femoral fracture had greater offset of the femoral shaft from the center of rotation of the femoral head, a more acute angle between the femoral neck and shaft, and greater proximal cortical thickness.

Medications. In addition to bisphosphonates, other drugs associated with atypical femoral fracture include RANK-ligand inhibitors such as denosumab (another drug for osteoporosis),³¹ glucocorticoids,^32,33 and proton pump inhibitors.^32,33

Genetics. Three sisters with atypical femoral fracture were found to have 37 rare mutations in 34 genes, including one in the GGPS1 gene, which codes for geranylgeranyl pyrophosphate synthase—an enzyme that bisphosphonates inhibit.³⁴

Medical conditions other than osteoporosis include collagen diseases, chronic pulmonary disease, asthma, rheumatoid arthritis, and diabetes.³⁵

Clinical recommendations

Current recommendations are to reevaluate bisphosphonate use in patients with osteoporosis after 5 or more years of therapy.³⁶

Given that patients with osteoporosis are at increased risk of typical fracture, those at higher risk should be considered for continued bisphosphonate therapy. Factors for high risk include the following:

• History of fracture on therapy
• Hip T score –2.5 or lower
• Older age (≥ 70)
• Other strong risk factors for fracture such as smoking, alcohol use, corticosteroid use, rheumatoid arthritis, and family history
• World Health Organization FRAX fracture risk score above the country-specific threshold.

Those at lower risk should be considered for a 2- to 3-year bisphosphonate holiday with periodic reevaluation of bone density and, possibly, bone markers.³⁶

WHAT IS THE UNDERLYING PATHOPHYSIOLOGY?

The mechanism by which bisphosphonates increase the risk of atypical femoral fracture is not clear. These drugs work by suppressing bone turnover; however, in theory, prolonged use could suppress it too much and increase bone fragility.

One hypothesis is that bisphosphonates impair the toughening of cortical bone, an important barrier to clinical fracture. This is supported by a study that found bisphosphonate users with atypical femoral fracture had deficits in intrinsic and extrinsic bone toughness, perhaps due to treatment-related increases in matrix mineralization.³⁷ Although this study and others showed an increase in matrix mineralization and reduced mineralization heterogeneity with bisphosphonate use,^38,39 it is unclear whether such changes contributed to reduced toughness or to atypical femoral fracture.

Changes in the skeletal geometry of the lower limb such as femoral neck-shaft angle and femoral curvature alter the stresses and strains experienced by the femoral diaphysis with loading. Because the incidence of incomplete atypical femoral fracture is much greater than that of complete fracture, most incomplete atypical femoral fractures heal before the fracture progresses.

Ultimately, all fractures, including atypical femoral fractures, occur when mechanical stress and strain exceed bone strength.

Antiresorptive drugs such as bisphosphonates, estrogen, calcitonin, and RANK ligand inhibitors prevent hip fracture by increasing the strength of the proximal femur—perhaps at the expense of the strength (or toughness) of the subtrochanteric shaft. It is also possible that treatment-related increases in hip strength (and reduced hip fracture rates) promote or sustain the transfer of stress and strain to femoral regions that experience lesser or no increases in strength from treatment, which likely includes the shaft.^40,41

CT studies in Japanese women with osteoporosis have shown that 2 years of zoledronate therapy had greater effects in the hip than in the femoral shaft, with significant increases in cortical thickness and volumetric bone mineral density at the femoral neck and intertrochanteric region compared with baseline.⁴² But zoledronate did not increase femoral shaft cortical thickness and caused only a minor increase in femoral shaft volumetric bone mineral density. Fracture patterns may have depended on damage and effects of bone turnover on mass and structure.

This hypothetical scenario portrays a possible “hip survival bias” mechanism for atypical femoral fracture, with the association with antiresorptive drugs arising from greater stress and strain in cortical regions where these fractures occur rather than from treatment-related reductions in cortical bone strength or toughness.

PRODROMAL PAIN IS COMMON

From 32% to 76% of patients who have incomplete or developing atypical femoral fracture present with a prodrome of groin or hip pain.^4,43 Prodromal pain occurs any time from 2 weeks to several years before the fracture, presenting as pain in the anterior or lateral thigh or in the groin.

Prodromal pain in a patient on antiresorptive therapy should be a signal for the clinician to obtain a radiograph of the hip and to look for contralateral symptoms and fractures. The most common mechanism of injury appears to be a ground-level fall or even a nontraumatic activity such as walking or stepping off a curb.

MEDICAL MANAGEMENT

In bisphosphonate users with radiographic evidence of atypical femoral fracture, the bis­phosphonate should be discontinued and the patient assessed for calcium and vitamin D deficiency, with supplements prescribed if needed.⁴

For patients with incomplete fracture and persistent pain after 3 months of medical management, prophylactic surgical nail fixation is recommended to prevent complete fracture.

Teriparatide, which has been associated with enhanced bone fracture healing, is a possible treatment to promote healing of atypical femoral fracture, either alone or as an adjunct to surgical fixation. A systematic review published in 2015 supported the use of teriparatide for enhancing fracture healing in atypical femoral fracture.⁴⁴ In addition, a 10-patient series⁴⁵ showed that incomplete fractures without radiolucent lines responded to teriparatide alone, whereas those with radiolucent lines needed intramedullary nailing.

These results suggest that teriparatide works best when the fracture site is stable, either inherently or with surgical fixation.

ORTHOPEDIC CARE

Orthopedic care for atypical femoral fracture differs depending on whether the patient experiences pain and whether the fracture is incomplete or complete. Figure 2 shows a treatment algorithm for atypical femoral fracture.

These are difficult fractures to manage, complicated by delayed healing in the elderly, complex displacement patterns, altered bone geometry, and risk of fracture in the opposite limb, all of which raise questions about recommending protected weight-bearing exercise.

Furthermore, atypical femoral fracture is often associated with increased anterolateral bowing of the femur, making it difficult to insert an intramedullary nail: the radius of curvature of the bone is shorter than that of a standard femoral nail. This mismatch can lead to intraoperative complications such as iatrogenic fracture during prophylactic nailing, malunion from excess straightening of the femur (which can itself lead to leg length discrepancy), and gapping of the fracture site, particularly on the medial side.

Intramedullary nailing is the first-line treatment for complete atypical femoral fracture, although the risk of delayed healing and revision surgery may be somewhat higher than with typical femoral fracture.⁴⁶ Prophylactic intramedullary nailing should be considered for a patient with intractable pain.²

A radiograph of the opposite leg should be obtained routinely, looking for an asymptomatic fracture. Bisphosphonates should be discontinued and calcium and vitamin D continued. Teriparatide therapy can be considered as an alternative treatment.

Conservative management for incomplete fracture without pain

Incomplete atypical femoral fracture unaccompanied by pain can be followed conservatively.⁴⁷ In addition to stopping antiresorptive therapy, patients need to avoid high-impact and repetitive-impact activities such as jumping or running. If pain occurs, patients should begin protected weight-bearing exercise.

Treatment is uncertain for incomplete fracture with pain

For patients with incomplete atypical femoral fracture and pain, treatment is controversial. Regimens that include 2 to 3 months of protected weight-bearing exercise, a full metabolic bone workup, calcium and vitamin D supplementation, and anabolic bone agents have produced some success. Some authors have reported poor results from conservative care, with few patients achieving pain relief or signs of complete healing.^48,49 Additionally, if an incomplete fracture is found in the opposite femur, protected weight-bearing of both legs may not be possible.

Patients with incomplete fracture should be monitored regularly with radiography and physical examination. If there is progression of the fracture, escalation of pain, or failure to heal within 2 to 3 months, then surgical treatment is necessary.

Prophylactic placement of an intramedullary nail to prevent completion of the fracture and allow a return to full weight-bearing is generally advised.⁵⁰ A long locking plate can be used if bone deformities make it difficult to place an intramedullary nail; however, nails are preferred because they allow formation of endochondral callus, which can be helpful in these difficult-to-heal fractures.

Results from retrospective reviews have shown that surgically treated patients with bis­phosphonate-associated incomplete atypical femoral fracture were more likely than those treated nonsurgically to be pain-free (81% vs 64%) and have radiographic healing (100% vs 18% at final follow-up).⁴⁶ Results have also been positive for those with complete atypical femoral fracture. At 6 months, 64% of surgically treated patients were pain-free and 98% were radiographically healed.⁵¹

The unusual geometry of the femur in patients with atypical femoral fracture and the presence of intramedullary cortical callus makes the placement of an intramedullary femoral rod more complex than in typical femoral fracture.⁸

Intramedullary nailing of atypical femoral fracture is a challenge for even the most experienced surgeon, and vigilance is imperative to avoid iatrogenic fracture and malunion.

MANY QUESTIONS REMAIN

We need more studies on the pathophysiology of bisphosphonate-associated atypical femoral fracture, the value of periodic screening with DXA, and which factors predict high risk (eg, Asian ethnicity, use of certain medications, femoral geometry). In addition, we need more data on the success of conservative management of incomplete fracture, including use of teriparatide.

Bisphosphonate therapy minimizes bone loss and reduces fracture risk by up to 50% in patients with osteoporosis,¹ but it is also associated with increased risks of osteonecrosis of the jaw and atypical femoral fracture. Although atypical femoral fractures are rare, they can have a devastating effect. Patient concern about this complication has contributed to a decrease in bisphosphonate use by about half in the last decade or so,^2,3 and we fear this could result in an increase in hip fracture rates.

In this article, we examine the evidence on bisphosphonate-associated atypical femoral fractures, including risks, pathogenesis, treatment, and prevention.

ATYPICAL FRACTURES INVOLVE THE FEMORAL SHAFT, NOT THE HEAD

An atypical femoral fracture is a transverse fracture of the femoral shaft (diaphysis), defined by both clinical criteria and radiographic appearance.

To be defined as atypical, a femoral fracture must meet 4 of the following 5 criteria⁴:

• Occurs with minimal or no trauma
• Has a predominantly transverse fracture line, originating at the lateral cortex and sometimes becoming oblique as it progresses medially across the femur
• Extends through both cortices and may be associated with a medial spike (complete fractures); or involves only the lateral cortex (incomplete fractures)
• Is noncomminuted or minimally comminuted
• Shows localized periosteal or endosteal thickening (termed “beaking” or “flaring”) of the lateral cortex at the fracture site.

Several minor features are also important but are not required, eg:

• Cortical thickening of the femoral shaft
• Unilateral or bilateral prodromal pain preceding the fracture
• Bilateral incomplete or complete femoral diaphysis fractures
• Delayed fracture healing.

Atypical femoral fracture can occur anywhere along the shaft, from just distal to the lesser trochanter to just proximal to the supracondylar flare. However, most occur in 2 areas, with 1 cluster centered at about 41 mm from the lesser trochanter (more common in relatively younger patients) and the other at 187 mm.⁵

ABSOLUTE RISK IS LOW BUT INCREASES WITH LONGER USE

Atypical femoral fractures are rare. Schilcher et al⁶ reviewed radiographs of 1,234 women who had a subtrochanteric or shaft fracture and found 59 (4.6%) of fractures were atypical. In a systematic review of 14 studies,⁷ the incidence ranged from 3.0 to 9.8 cases per 100,000 patient-years.

Furthermore, not all atypical femoral fractures are in bisphosphonate users: 7.4% were in nonusers in 1 series⁸ and 22% in another.⁹

Nevertheless, most studies show that bis­phosphonate use increases the incidence of atypical femoral fracture, and the incidence increases with duration of use, especially after 3 years.⁷

An international task force of the American Society for Bone and Mineral Research listed the absolute risk as between 3.2 and 50 cases per 100,000 patient-years, with longer use (> 5 years) increasing the risk to about 100 per 100,000 patient-years.⁴ After stopping bis­phosphonate therapy, the risk diminished by 70% per year.⁹

In another study, for 0.1 to 1.9 years of therapy, the age-adjusted atypical fracture rates were 1.78 per 100,000 per year (95% confidence interval [CI] 1.5–2.0), increasing to 113.1 per 100,000 per year (95% CI 69.3–156.8) with exposure from 8 to 9.9 years.¹⁰

A case-control study found that more than 5 years of bisphosphonate use increased the fracture risk by an odds ratio of 2.74 (95% CI 1.25–6.02).¹¹

The incidence of typical femoral fracture was higher in those who adhered better to their oral bisphosphonate regimen in some studies,¹² but the opposite was true in others.¹³

The benefits of bisphosphonate therapy in reducing fracture risk, however, outweigh the risk of atypical fracture.⁴

We do not know whether the rate of atypical femoral fracture is increasing. A review of Kaiser Permanente Northwest records found that the rates of atypical femoral shaft fracture had remained stable from 1996 to 2009. However, 61.9% of patients who met the strict radiographic criteria had taken oral bisphosphonates.¹⁴ These data suggest that bisphosphonate use has not increased the overall population-based risk for subtrochanteric and femoral shaft fractures, but that bisphosphonates and other risk factors may have increased the likelihood that such fractures will exhibit atypical radiographic features.

A population-based study in Denmark¹³ found that alendronate use longer than 10 years was associated with an adjusted 30% lower risk of hip fracture and no increase in the risk of subtrochanteric and femoral shaft fracture. In addition, the risk of subtrochanteric and femoral shaft fracture was lower with high adherence to alendronate treatment (based on medication possession ratio > 80%) compared with low adherence (ratio < 50%) (odds ratio 0.88, 95% CI 0.77–0.99). The risk was not increased in current vs past users.

The Danish study¹³ used the coding of the 10th revision of the International Statistical Classification of Diseases and Related Health Problems (ICD-10) to identify subtrochanteric and femoral shaft fractures without radiologic review for atypical radiographic features. The lack of specific ICD-10 coding for subtrochanteric and femoral shaft fractures with atypical radiographic features has limited our knowledge of their incidence.

After an atypical femoral fracture, patients have a significant risk of fracture on the contralateral side. In a case-control study, 28% of patients with atypical femoral fracture suffered a contralateral fracture, compared with 0.9% of patients presenting with a typical fracture pattern (odds ratio 42.6, 95% CI 12.8–142.4).¹⁵

Contralateral fracture occurs from 1 month to 4 years after the index atypical femoral fracture.¹⁶

There are reports of bisphosphonate-related low-impact fractures in other sites such as the tibia¹⁷ and forearm.¹⁸ However, they may be too rare to warrant screening.

Mortality rates

A Swedish database study found that patients with atypical femoral fractures, whether bisphosphonate users or nonusers, do not have higher mortality rates than patients with ordinary subtrochanteric or femoral shaft fractures.¹⁹ Furthermore, the mortality rates for those with atypical femoral fracture were similar to rates in the general population. In contrast, patients with an ordinary femoral fracture had a higher mortality risk than the general population.¹⁹

Other studies suggest that atypical femoral fracture may be associated with a less favorable prognosis in older patients,²⁰ but this could be due to differences in demographics, treatment adherence, or postfracture care.²¹

In addition, functional outcomes as measured by independent mobility at discharge and at 3 months were comparable between patients with atypical fracture and those with typical fracture.²²

IMAGING STUDIES

If a long-term bisphosphonate user presents with hip, thigh, or groin pain, imaging studies are recommended.

Plain radiography

Radiography is usually the first step and should include a frontal view of the pelvis (Figure 1) and 2 views of the full length of each femur. If radiography is not conclusive, bone scan or magnetic resonance imaging (MRI) should be considered.

A linear cortex transverse fracture pattern and focal lateral cortical thickening are the most sensitive and specific radiographic features.^23,24 Because of the risk of fracture on the contralateral side, radiographic study of that side is recommended as well.

Computed tomography

Computed tomography (CT) is not sensitive for early stress fractures and, given the radiation burden, is not recommended in the workup of atypical fracture.

Bone scanning

Bone scanning using technetium 99m-labeled methylene diphosphonate with a gamma camera shows active bone turnover. Stress fractures and atypical femoral fractures are most easily identified in the third (delayed) phase of the bone scan. Although bone scanning is highly sensitive, the specificity is limited by lack of spatial resolution. Atypical femoral fracture appears as increased activity in the subtrochanteric region with a predilection for the lateral cortex.

Dual-energy x-ray absorptiometry

Conventional dual-energy x-ray absorptiometry (DXA) extends only to 1 to 2 cm below the lesser trochanter and can therefore miss atypical fractures, which usually occur farther down. The overall detection rate for DXA was 61% in a sample of 33 patients.²⁵

Newer scanners can look at the entire femoral shaft.²⁶ In addition, newer software can quantify focal thickening (beaking) of the lateral cortex and screen patients who have no symptoms. The results of serial measurements can be graphed so that the practitioner can view trends to help assess or rule out potential asymptomatic atypical femoral fracture.

A localized reaction (periosteal thickening of the lateral cortex or beaking) often precedes atypical femoral fracture. A 2017 study reported that patients with high localized reaction (mean height 3.3 mm) that was of the pointed type and was accompanied by prodromal pain had an increased risk of complete or incomplete atypical femoral fracture at that site.²⁷ This finding is used by the newer DXA software. The predictive value of beaking on extended femoral DXA may be as high as 83%.²⁶

Magnetic resonance imaging

The MRI characteristics of atypical femoral fracture are similar to those of other stress fractures except that there is a lateral-to-medial pattern rather than a medial pattern. The earliest findings include periosteal reaction about the lateral cortex with a normal marrow signal.

MRI may be of particular benefit in patients with known atypical femoral fracture to screen the contralateral leg. It should image the entire length of both femurs. Contrast enhancement is not needed.

Regardless of whether initial findings were discovered on conventional radiographs or DXA, MRI confirmation is needed. Radio­nuclide bone scanning is currently not recommended because it lacks specificity. Combination imaging is recommended, with either radiography plus MRI or DXA plus MRI.

DIFFERENTIAL DIAGNOSIS

The differential diagnosis of atypical femoral fracture includes stress fracture, pathologic fracture, hypophosphatasia, and osteogenesis imperfecta.²⁸ Hypophosphatemic osteomalacia can cause Looser zones, which can be confused with atypical femoral fractures but usually occur on the medial side.⁴ Stress fracture of the femur can occur below the lesser trochanter but usually begins in the medial, not the lateral, cortex.

Pathologic fractures from underlying osseous lesions can mimic the cortical beaking of bisphosphonate-related fracture, but they usually show the associated underlying lucent lesion and poorly defined margins. A sinus tract along the region of a chronic osteomyelitis may also appear similar.

Hypophosphatasia is an inborn error of metabolism caused by a loss-of-function mutation in the gene encoding alkaline phosphatase, resulting in pyrophosphate accumulation and causing osteomalacia from impaired mineralization. This can result in femoral pseudofracture that is often bilateral and occurs in the subtrochanteric region.²⁹

Patients with atypical femoral fracture are generally a heterogeneous group, but there are risk factors to note other than bisphosphonate exposure.

Asian women had a risk 8 times higher than white women in 1 study.³⁰

Bone geometry. Mahjoub et al⁸ reported that compared with controls, patients with atypical femoral fracture had greater offset of the femoral shaft from the center of rotation of the femoral head, a more acute angle between the femoral neck and shaft, and greater proximal cortical thickness.

Medications. In addition to bisphosphonates, other drugs associated with atypical femoral fracture include RANK-ligand inhibitors such as denosumab (another drug for osteoporosis),³¹ glucocorticoids,^32,33 and proton pump inhibitors.^32,33

Genetics. Three sisters with atypical femoral fracture were found to have 37 rare mutations in 34 genes, including one in the GGPS1 gene, which codes for geranylgeranyl pyrophosphate synthase—an enzyme that bisphosphonates inhibit.³⁴

Medical conditions other than osteoporosis include collagen diseases, chronic pulmonary disease, asthma, rheumatoid arthritis, and diabetes.³⁵

Clinical recommendations

Current recommendations are to reevaluate bisphosphonate use in patients with osteoporosis after 5 or more years of therapy.³⁶

Given that patients with osteoporosis are at increased risk of typical fracture, those at higher risk should be considered for continued bisphosphonate therapy. Factors for high risk include the following:

• History of fracture on therapy
• Hip T score –2.5 or lower
• Older age (≥ 70)
• Other strong risk factors for fracture such as smoking, alcohol use, corticosteroid use, rheumatoid arthritis, and family history
• World Health Organization FRAX fracture risk score above the country-specific threshold.

Those at lower risk should be considered for a 2- to 3-year bisphosphonate holiday with periodic reevaluation of bone density and, possibly, bone markers.³⁶

WHAT IS THE UNDERLYING PATHOPHYSIOLOGY?

The mechanism by which bisphosphonates increase the risk of atypical femoral fracture is not clear. These drugs work by suppressing bone turnover; however, in theory, prolonged use could suppress it too much and increase bone fragility.

One hypothesis is that bisphosphonates impair the toughening of cortical bone, an important barrier to clinical fracture. This is supported by a study that found bisphosphonate users with atypical femoral fracture had deficits in intrinsic and extrinsic bone toughness, perhaps due to treatment-related increases in matrix mineralization.³⁷ Although this study and others showed an increase in matrix mineralization and reduced mineralization heterogeneity with bisphosphonate use,^38,39 it is unclear whether such changes contributed to reduced toughness or to atypical femoral fracture.

Changes in the skeletal geometry of the lower limb such as femoral neck-shaft angle and femoral curvature alter the stresses and strains experienced by the femoral diaphysis with loading. Because the incidence of incomplete atypical femoral fracture is much greater than that of complete fracture, most incomplete atypical femoral fractures heal before the fracture progresses.

Ultimately, all fractures, including atypical femoral fractures, occur when mechanical stress and strain exceed bone strength.

Antiresorptive drugs such as bisphosphonates, estrogen, calcitonin, and RANK ligand inhibitors prevent hip fracture by increasing the strength of the proximal femur—perhaps at the expense of the strength (or toughness) of the subtrochanteric shaft. It is also possible that treatment-related increases in hip strength (and reduced hip fracture rates) promote or sustain the transfer of stress and strain to femoral regions that experience lesser or no increases in strength from treatment, which likely includes the shaft.^40,41

CT studies in Japanese women with osteoporosis have shown that 2 years of zoledronate therapy had greater effects in the hip than in the femoral shaft, with significant increases in cortical thickness and volumetric bone mineral density at the femoral neck and intertrochanteric region compared with baseline.⁴² But zoledronate did not increase femoral shaft cortical thickness and caused only a minor increase in femoral shaft volumetric bone mineral density. Fracture patterns may have depended on damage and effects of bone turnover on mass and structure.

This hypothetical scenario portrays a possible “hip survival bias” mechanism for atypical femoral fracture, with the association with antiresorptive drugs arising from greater stress and strain in cortical regions where these fractures occur rather than from treatment-related reductions in cortical bone strength or toughness.

PRODROMAL PAIN IS COMMON

From 32% to 76% of patients who have incomplete or developing atypical femoral fracture present with a prodrome of groin or hip pain.^4,43 Prodromal pain occurs any time from 2 weeks to several years before the fracture, presenting as pain in the anterior or lateral thigh or in the groin.

Prodromal pain in a patient on antiresorptive therapy should be a signal for the clinician to obtain a radiograph of the hip and to look for contralateral symptoms and fractures. The most common mechanism of injury appears to be a ground-level fall or even a nontraumatic activity such as walking or stepping off a curb.

MEDICAL MANAGEMENT

In bisphosphonate users with radiographic evidence of atypical femoral fracture, the bis­phosphonate should be discontinued and the patient assessed for calcium and vitamin D deficiency, with supplements prescribed if needed.⁴

For patients with incomplete fracture and persistent pain after 3 months of medical management, prophylactic surgical nail fixation is recommended to prevent complete fracture.

Teriparatide, which has been associated with enhanced bone fracture healing, is a possible treatment to promote healing of atypical femoral fracture, either alone or as an adjunct to surgical fixation. A systematic review published in 2015 supported the use of teriparatide for enhancing fracture healing in atypical femoral fracture.⁴⁴ In addition, a 10-patient series⁴⁵ showed that incomplete fractures without radiolucent lines responded to teriparatide alone, whereas those with radiolucent lines needed intramedullary nailing.

These results suggest that teriparatide works best when the fracture site is stable, either inherently or with surgical fixation.

ORTHOPEDIC CARE

Orthopedic care for atypical femoral fracture differs depending on whether the patient experiences pain and whether the fracture is incomplete or complete. Figure 2 shows a treatment algorithm for atypical femoral fracture.

These are difficult fractures to manage, complicated by delayed healing in the elderly, complex displacement patterns, altered bone geometry, and risk of fracture in the opposite limb, all of which raise questions about recommending protected weight-bearing exercise.

Furthermore, atypical femoral fracture is often associated with increased anterolateral bowing of the femur, making it difficult to insert an intramedullary nail: the radius of curvature of the bone is shorter than that of a standard femoral nail. This mismatch can lead to intraoperative complications such as iatrogenic fracture during prophylactic nailing, malunion from excess straightening of the femur (which can itself lead to leg length discrepancy), and gapping of the fracture site, particularly on the medial side.

Intramedullary nailing is the first-line treatment for complete atypical femoral fracture, although the risk of delayed healing and revision surgery may be somewhat higher than with typical femoral fracture.⁴⁶ Prophylactic intramedullary nailing should be considered for a patient with intractable pain.²

A radiograph of the opposite leg should be obtained routinely, looking for an asymptomatic fracture. Bisphosphonates should be discontinued and calcium and vitamin D continued. Teriparatide therapy can be considered as an alternative treatment.

Conservative management for incomplete fracture without pain

Incomplete atypical femoral fracture unaccompanied by pain can be followed conservatively.⁴⁷ In addition to stopping antiresorptive therapy, patients need to avoid high-impact and repetitive-impact activities such as jumping or running. If pain occurs, patients should begin protected weight-bearing exercise.

Treatment is uncertain for incomplete fracture with pain

For patients with incomplete atypical femoral fracture and pain, treatment is controversial. Regimens that include 2 to 3 months of protected weight-bearing exercise, a full metabolic bone workup, calcium and vitamin D supplementation, and anabolic bone agents have produced some success. Some authors have reported poor results from conservative care, with few patients achieving pain relief or signs of complete healing.^48,49 Additionally, if an incomplete fracture is found in the opposite femur, protected weight-bearing of both legs may not be possible.

Patients with incomplete fracture should be monitored regularly with radiography and physical examination. If there is progression of the fracture, escalation of pain, or failure to heal within 2 to 3 months, then surgical treatment is necessary.

Prophylactic placement of an intramedullary nail to prevent completion of the fracture and allow a return to full weight-bearing is generally advised.⁵⁰ A long locking plate can be used if bone deformities make it difficult to place an intramedullary nail; however, nails are preferred because they allow formation of endochondral callus, which can be helpful in these difficult-to-heal fractures.

Results from retrospective reviews have shown that surgically treated patients with bis­phosphonate-associated incomplete atypical femoral fracture were more likely than those treated nonsurgically to be pain-free (81% vs 64%) and have radiographic healing (100% vs 18% at final follow-up).⁴⁶ Results have also been positive for those with complete atypical femoral fracture. At 6 months, 64% of surgically treated patients were pain-free and 98% were radiographically healed.⁵¹

The unusual geometry of the femur in patients with atypical femoral fracture and the presence of intramedullary cortical callus makes the placement of an intramedullary femoral rod more complex than in typical femoral fracture.⁸

Intramedullary nailing of atypical femoral fracture is a challenge for even the most experienced surgeon, and vigilance is imperative to avoid iatrogenic fracture and malunion.

MANY QUESTIONS REMAIN

We need more studies on the pathophysiology of bisphosphonate-associated atypical femoral fracture, the value of periodic screening with DXA, and which factors predict high risk (eg, Asian ethnicity, use of certain medications, femoral geometry). In addition, we need more data on the success of conservative management of incomplete fracture, including use of teriparatide.

Bisphosphonate therapy minimizes bone loss and reduces fracture risk by up to 50% in patients with osteoporosis,¹ but it is also associated with increased risks of osteonecrosis of the jaw and atypical femoral fracture. Although atypical femoral fractures are rare, they can have a devastating effect. Patient concern about this complication has contributed to a decrease in bisphosphonate use by about half in the last decade or so,^2,3 and we fear this could result in an increase in hip fracture rates.

In this article, we examine the evidence on bisphosphonate-associated atypical femoral fractures, including risks, pathogenesis, treatment, and prevention.

ATYPICAL FRACTURES INVOLVE THE FEMORAL SHAFT, NOT THE HEAD

An atypical femoral fracture is a transverse fracture of the femoral shaft (diaphysis), defined by both clinical criteria and radiographic appearance.

To be defined as atypical, a femoral fracture must meet 4 of the following 5 criteria⁴:

• Occurs with minimal or no trauma
• Has a predominantly transverse fracture line, originating at the lateral cortex and sometimes becoming oblique as it progresses medially across the femur
• Extends through both cortices and may be associated with a medial spike (complete fractures); or involves only the lateral cortex (incomplete fractures)
• Is noncomminuted or minimally comminuted
• Shows localized periosteal or endosteal thickening (termed “beaking” or “flaring”) of the lateral cortex at the fracture site.

Several minor features are also important but are not required, eg:

• Cortical thickening of the femoral shaft
• Unilateral or bilateral prodromal pain preceding the fracture
• Bilateral incomplete or complete femoral diaphysis fractures
• Delayed fracture healing.

Atypical femoral fracture can occur anywhere along the shaft, from just distal to the lesser trochanter to just proximal to the supracondylar flare. However, most occur in 2 areas, with 1 cluster centered at about 41 mm from the lesser trochanter (more common in relatively younger patients) and the other at 187 mm.⁵

ABSOLUTE RISK IS LOW BUT INCREASES WITH LONGER USE

Atypical femoral fractures are rare. Schilcher et al⁶ reviewed radiographs of 1,234 women who had a subtrochanteric or shaft fracture and found 59 (4.6%) of fractures were atypical. In a systematic review of 14 studies,⁷ the incidence ranged from 3.0 to 9.8 cases per 100,000 patient-years.

Furthermore, not all atypical femoral fractures are in bisphosphonate users: 7.4% were in nonusers in 1 series⁸ and 22% in another.⁹

Nevertheless, most studies show that bis­phosphonate use increases the incidence of atypical femoral fracture, and the incidence increases with duration of use, especially after 3 years.⁷

An international task force of the American Society for Bone and Mineral Research listed the absolute risk as between 3.2 and 50 cases per 100,000 patient-years, with longer use (> 5 years) increasing the risk to about 100 per 100,000 patient-years.⁴ After stopping bis­phosphonate therapy, the risk diminished by 70% per year.⁹

In another study, for 0.1 to 1.9 years of therapy, the age-adjusted atypical fracture rates were 1.78 per 100,000 per year (95% confidence interval [CI] 1.5–2.0), increasing to 113.1 per 100,000 per year (95% CI 69.3–156.8) with exposure from 8 to 9.9 years.¹⁰

A case-control study found that more than 5 years of bisphosphonate use increased the fracture risk by an odds ratio of 2.74 (95% CI 1.25–6.02).¹¹

The incidence of typical femoral fracture was higher in those who adhered better to their oral bisphosphonate regimen in some studies,¹² but the opposite was true in others.¹³

The benefits of bisphosphonate therapy in reducing fracture risk, however, outweigh the risk of atypical fracture.⁴

We do not know whether the rate of atypical femoral fracture is increasing. A review of Kaiser Permanente Northwest records found that the rates of atypical femoral shaft fracture had remained stable from 1996 to 2009. However, 61.9% of patients who met the strict radiographic criteria had taken oral bisphosphonates.¹⁴ These data suggest that bisphosphonate use has not increased the overall population-based risk for subtrochanteric and femoral shaft fractures, but that bisphosphonates and other risk factors may have increased the likelihood that such fractures will exhibit atypical radiographic features.

A population-based study in Denmark¹³ found that alendronate use longer than 10 years was associated with an adjusted 30% lower risk of hip fracture and no increase in the risk of subtrochanteric and femoral shaft fracture. In addition, the risk of subtrochanteric and femoral shaft fracture was lower with high adherence to alendronate treatment (based on medication possession ratio > 80%) compared with low adherence (ratio < 50%) (odds ratio 0.88, 95% CI 0.77–0.99). The risk was not increased in current vs past users.

The Danish study¹³ used the coding of the 10th revision of the International Statistical Classification of Diseases and Related Health Problems (ICD-10) to identify subtrochanteric and femoral shaft fractures without radiologic review for atypical radiographic features. The lack of specific ICD-10 coding for subtrochanteric and femoral shaft fractures with atypical radiographic features has limited our knowledge of their incidence.

After an atypical femoral fracture, patients have a significant risk of fracture on the contralateral side. In a case-control study, 28% of patients with atypical femoral fracture suffered a contralateral fracture, compared with 0.9% of patients presenting with a typical fracture pattern (odds ratio 42.6, 95% CI 12.8–142.4).¹⁵

Contralateral fracture occurs from 1 month to 4 years after the index atypical femoral fracture.¹⁶

There are reports of bisphosphonate-related low-impact fractures in other sites such as the tibia¹⁷ and forearm.¹⁸ However, they may be too rare to warrant screening.

Mortality rates

A Swedish database study found that patients with atypical femoral fractures, whether bisphosphonate users or nonusers, do not have higher mortality rates than patients with ordinary subtrochanteric or femoral shaft fractures.¹⁹ Furthermore, the mortality rates for those with atypical femoral fracture were similar to rates in the general population. In contrast, patients with an ordinary femoral fracture had a higher mortality risk than the general population.¹⁹

Other studies suggest that atypical femoral fracture may be associated with a less favorable prognosis in older patients,²⁰ but this could be due to differences in demographics, treatment adherence, or postfracture care.²¹

In addition, functional outcomes as measured by independent mobility at discharge and at 3 months were comparable between patients with atypical fracture and those with typical fracture.²²

IMAGING STUDIES

If a long-term bisphosphonate user presents with hip, thigh, or groin pain, imaging studies are recommended.

Plain radiography

Radiography is usually the first step and should include a frontal view of the pelvis (Figure 1) and 2 views of the full length of each femur. If radiography is not conclusive, bone scan or magnetic resonance imaging (MRI) should be considered.

A linear cortex transverse fracture pattern and focal lateral cortical thickening are the most sensitive and specific radiographic features.^23,24 Because of the risk of fracture on the contralateral side, radiographic study of that side is recommended as well.

Computed tomography

Computed tomography (CT) is not sensitive for early stress fractures and, given the radiation burden, is not recommended in the workup of atypical fracture.

Bone scanning

Bone scanning using technetium 99m-labeled methylene diphosphonate with a gamma camera shows active bone turnover. Stress fractures and atypical femoral fractures are most easily identified in the third (delayed) phase of the bone scan. Although bone scanning is highly sensitive, the specificity is limited by lack of spatial resolution. Atypical femoral fracture appears as increased activity in the subtrochanteric region with a predilection for the lateral cortex.

Dual-energy x-ray absorptiometry

Conventional dual-energy x-ray absorptiometry (DXA) extends only to 1 to 2 cm below the lesser trochanter and can therefore miss atypical fractures, which usually occur farther down. The overall detection rate for DXA was 61% in a sample of 33 patients.²⁵

Newer scanners can look at the entire femoral shaft.²⁶ In addition, newer software can quantify focal thickening (beaking) of the lateral cortex and screen patients who have no symptoms. The results of serial measurements can be graphed so that the practitioner can view trends to help assess or rule out potential asymptomatic atypical femoral fracture.

A localized reaction (periosteal thickening of the lateral cortex or beaking) often precedes atypical femoral fracture. A 2017 study reported that patients with high localized reaction (mean height 3.3 mm) that was of the pointed type and was accompanied by prodromal pain had an increased risk of complete or incomplete atypical femoral fracture at that site.²⁷ This finding is used by the newer DXA software. The predictive value of beaking on extended femoral DXA may be as high as 83%.²⁶

Magnetic resonance imaging

The MRI characteristics of atypical femoral fracture are similar to those of other stress fractures except that there is a lateral-to-medial pattern rather than a medial pattern. The earliest findings include periosteal reaction about the lateral cortex with a normal marrow signal.

MRI may be of particular benefit in patients with known atypical femoral fracture to screen the contralateral leg. It should image the entire length of both femurs. Contrast enhancement is not needed.

Regardless of whether initial findings were discovered on conventional radiographs or DXA, MRI confirmation is needed. Radio­nuclide bone scanning is currently not recommended because it lacks specificity. Combination imaging is recommended, with either radiography plus MRI or DXA plus MRI.

DIFFERENTIAL DIAGNOSIS

The differential diagnosis of atypical femoral fracture includes stress fracture, pathologic fracture, hypophosphatasia, and osteogenesis imperfecta.²⁸ Hypophosphatemic osteomalacia can cause Looser zones, which can be confused with atypical femoral fractures but usually occur on the medial side.⁴ Stress fracture of the femur can occur below the lesser trochanter but usually begins in the medial, not the lateral, cortex.

Pathologic fractures from underlying osseous lesions can mimic the cortical beaking of bisphosphonate-related fracture, but they usually show the associated underlying lucent lesion and poorly defined margins. A sinus tract along the region of a chronic osteomyelitis may also appear similar.

Hypophosphatasia is an inborn error of metabolism caused by a loss-of-function mutation in the gene encoding alkaline phosphatase, resulting in pyrophosphate accumulation and causing osteomalacia from impaired mineralization. This can result in femoral pseudofracture that is often bilateral and occurs in the subtrochanteric region.²⁹

Patients with atypical femoral fracture are generally a heterogeneous group, but there are risk factors to note other than bisphosphonate exposure.

Asian women had a risk 8 times higher than white women in 1 study.³⁰

Bone geometry. Mahjoub et al⁸ reported that compared with controls, patients with atypical femoral fracture had greater offset of the femoral shaft from the center of rotation of the femoral head, a more acute angle between the femoral neck and shaft, and greater proximal cortical thickness.

Medications. In addition to bisphosphonates, other drugs associated with atypical femoral fracture include RANK-ligand inhibitors such as denosumab (another drug for osteoporosis),³¹ glucocorticoids,^32,33 and proton pump inhibitors.^32,33

Genetics. Three sisters with atypical femoral fracture were found to have 37 rare mutations in 34 genes, including one in the GGPS1 gene, which codes for geranylgeranyl pyrophosphate synthase—an enzyme that bisphosphonates inhibit.³⁴

Medical conditions other than osteoporosis include collagen diseases, chronic pulmonary disease, asthma, rheumatoid arthritis, and diabetes.³⁵

Clinical recommendations

Current recommendations are to reevaluate bisphosphonate use in patients with osteoporosis after 5 or more years of therapy.³⁶

Given that patients with osteoporosis are at increased risk of typical fracture, those at higher risk should be considered for continued bisphosphonate therapy. Factors for high risk include the following:

• History of fracture on therapy
• Hip T score –2.5 or lower
• Older age (≥ 70)
• Other strong risk factors for fracture such as smoking, alcohol use, corticosteroid use, rheumatoid arthritis, and family history
• World Health Organization FRAX fracture risk score above the country-specific threshold.

Those at lower risk should be considered for a 2- to 3-year bisphosphonate holiday with periodic reevaluation of bone density and, possibly, bone markers.³⁶

WHAT IS THE UNDERLYING PATHOPHYSIOLOGY?

The mechanism by which bisphosphonates increase the risk of atypical femoral fracture is not clear. These drugs work by suppressing bone turnover; however, in theory, prolonged use could suppress it too much and increase bone fragility.

One hypothesis is that bisphosphonates impair the toughening of cortical bone, an important barrier to clinical fracture. This is supported by a study that found bisphosphonate users with atypical femoral fracture had deficits in intrinsic and extrinsic bone toughness, perhaps due to treatment-related increases in matrix mineralization.³⁷ Although this study and others showed an increase in matrix mineralization and reduced mineralization heterogeneity with bisphosphonate use,^38,39 it is unclear whether such changes contributed to reduced toughness or to atypical femoral fracture.

Changes in the skeletal geometry of the lower limb such as femoral neck-shaft angle and femoral curvature alter the stresses and strains experienced by the femoral diaphysis with loading. Because the incidence of incomplete atypical femoral fracture is much greater than that of complete fracture, most incomplete atypical femoral fractures heal before the fracture progresses.

Ultimately, all fractures, including atypical femoral fractures, occur when mechanical stress and strain exceed bone strength.

Antiresorptive drugs such as bisphosphonates, estrogen, calcitonin, and RANK ligand inhibitors prevent hip fracture by increasing the strength of the proximal femur—perhaps at the expense of the strength (or toughness) of the subtrochanteric shaft. It is also possible that treatment-related increases in hip strength (and reduced hip fracture rates) promote or sustain the transfer of stress and strain to femoral regions that experience lesser or no increases in strength from treatment, which likely includes the shaft.^40,41

CT studies in Japanese women with osteoporosis have shown that 2 years of zoledronate therapy had greater effects in the hip than in the femoral shaft, with significant increases in cortical thickness and volumetric bone mineral density at the femoral neck and intertrochanteric region compared with baseline.⁴² But zoledronate did not increase femoral shaft cortical thickness and caused only a minor increase in femoral shaft volumetric bone mineral density. Fracture patterns may have depended on damage and effects of bone turnover on mass and structure.

This hypothetical scenario portrays a possible “hip survival bias” mechanism for atypical femoral fracture, with the association with antiresorptive drugs arising from greater stress and strain in cortical regions where these fractures occur rather than from treatment-related reductions in cortical bone strength or toughness.

PRODROMAL PAIN IS COMMON

From 32% to 76% of patients who have incomplete or developing atypical femoral fracture present with a prodrome of groin or hip pain.^4,43 Prodromal pain occurs any time from 2 weeks to several years before the fracture, presenting as pain in the anterior or lateral thigh or in the groin.

Prodromal pain in a patient on antiresorptive therapy should be a signal for the clinician to obtain a radiograph of the hip and to look for contralateral symptoms and fractures. The most common mechanism of injury appears to be a ground-level fall or even a nontraumatic activity such as walking or stepping off a curb.

MEDICAL MANAGEMENT

In bisphosphonate users with radiographic evidence of atypical femoral fracture, the bis­phosphonate should be discontinued and the patient assessed for calcium and vitamin D deficiency, with supplements prescribed if needed.⁴

For patients with incomplete fracture and persistent pain after 3 months of medical management, prophylactic surgical nail fixation is recommended to prevent complete fracture.

Teriparatide, which has been associated with enhanced bone fracture healing, is a possible treatment to promote healing of atypical femoral fracture, either alone or as an adjunct to surgical fixation. A systematic review published in 2015 supported the use of teriparatide for enhancing fracture healing in atypical femoral fracture.⁴⁴ In addition, a 10-patient series⁴⁵ showed that incomplete fractures without radiolucent lines responded to teriparatide alone, whereas those with radiolucent lines needed intramedullary nailing.

These results suggest that teriparatide works best when the fracture site is stable, either inherently or with surgical fixation.

ORTHOPEDIC CARE

Orthopedic care for atypical femoral fracture differs depending on whether the patient experiences pain and whether the fracture is incomplete or complete. Figure 2 shows a treatment algorithm for atypical femoral fracture.

These are difficult fractures to manage, complicated by delayed healing in the elderly, complex displacement patterns, altered bone geometry, and risk of fracture in the opposite limb, all of which raise questions about recommending protected weight-bearing exercise.

Furthermore, atypical femoral fracture is often associated with increased anterolateral bowing of the femur, making it difficult to insert an intramedullary nail: the radius of curvature of the bone is shorter than that of a standard femoral nail. This mismatch can lead to intraoperative complications such as iatrogenic fracture during prophylactic nailing, malunion from excess straightening of the femur (which can itself lead to leg length discrepancy), and gapping of the fracture site, particularly on the medial side.

Intramedullary nailing is the first-line treatment for complete atypical femoral fracture, although the risk of delayed healing and revision surgery may be somewhat higher than with typical femoral fracture.⁴⁶ Prophylactic intramedullary nailing should be considered for a patient with intractable pain.²

A radiograph of the opposite leg should be obtained routinely, looking for an asymptomatic fracture. Bisphosphonates should be discontinued and calcium and vitamin D continued. Teriparatide therapy can be considered as an alternative treatment.

Conservative management for incomplete fracture without pain

Incomplete atypical femoral fracture unaccompanied by pain can be followed conservatively.⁴⁷ In addition to stopping antiresorptive therapy, patients need to avoid high-impact and repetitive-impact activities such as jumping or running. If pain occurs, patients should begin protected weight-bearing exercise.

Treatment is uncertain for incomplete fracture with pain

For patients with incomplete atypical femoral fracture and pain, treatment is controversial. Regimens that include 2 to 3 months of protected weight-bearing exercise, a full metabolic bone workup, calcium and vitamin D supplementation, and anabolic bone agents have produced some success. Some authors have reported poor results from conservative care, with few patients achieving pain relief or signs of complete healing.^48,49 Additionally, if an incomplete fracture is found in the opposite femur, protected weight-bearing of both legs may not be possible.

Patients with incomplete fracture should be monitored regularly with radiography and physical examination. If there is progression of the fracture, escalation of pain, or failure to heal within 2 to 3 months, then surgical treatment is necessary.

Prophylactic placement of an intramedullary nail to prevent completion of the fracture and allow a return to full weight-bearing is generally advised.⁵⁰ A long locking plate can be used if bone deformities make it difficult to place an intramedullary nail; however, nails are preferred because they allow formation of endochondral callus, which can be helpful in these difficult-to-heal fractures.

Results from retrospective reviews have shown that surgically treated patients with bis­phosphonate-associated incomplete atypical femoral fracture were more likely than those treated nonsurgically to be pain-free (81% vs 64%) and have radiographic healing (100% vs 18% at final follow-up).⁴⁶ Results have also been positive for those with complete atypical femoral fracture. At 6 months, 64% of surgically treated patients were pain-free and 98% were radiographically healed.⁵¹

The unusual geometry of the femur in patients with atypical femoral fracture and the presence of intramedullary cortical callus makes the placement of an intramedullary femoral rod more complex than in typical femoral fracture.⁸

Intramedullary nailing of atypical femoral fracture is a challenge for even the most experienced surgeon, and vigilance is imperative to avoid iatrogenic fracture and malunion.

MANY QUESTIONS REMAIN

We need more studies on the pathophysiology of bisphosphonate-associated atypical femoral fracture, the value of periodic screening with DXA, and which factors predict high risk (eg, Asian ethnicity, use of certain medications, femoral geometry). In addition, we need more data on the success of conservative management of incomplete fracture, including use of teriparatide.

Several months ago, we invited readers to submit short personalized commentaries on articles that changed the way they approach a specific clinical problem and the way they take care of patients. In this issue of the Journal, addiction specialist Charles Reznikoff, MD, discusses 3 observational studies that focused on how prescribing opioids for acute pain can lead to chronic opioid use and addiction, and how these studies have influenced his practice.

Although observational studies rank lower on the level-of-evidence scale than randomized controlled trials, they can intellectually stimulate and inform us in ways that lead us to modify how we deliver clinical care.

The initial prescribing of pain medications and the management of patients with chronic pain are currently under intense scrutiny, and are the topic of much discussion in the United States. The opioid epidemic has spilled over into all aspects of daily life, far beyond the medical community. But since we physicians are the only legal and regulated source of narcotics and other pain medications, we are under the microscope—and rightly so.

We, our patients, the pharmaceutical industry, legislators, and the law enforcement community struggle to navigate a complex maze, one with moving walls. Not long ago, physicians were told that we were not attentive enough to our patients’ suffering and needed to do better at relieving it. “Pain” became a vital sign and a recorded metric of quality care. Some excellent changes evolved from this focus, such as increased emphasis on postoperative regional and local pain control. But pain measurements continue to be recorded at every outpatient visit, an almost mindless requirement.

Recently, a patient with lupus nephritis whom I was seeing for blood pressure management reported a pain level of 8 on a scale of 10. I confess that I usually don’t even look at these metrics, but for whatever reason I saw her answer. I asked her about it. She had burned her finger while cooking and said, “I had no idea what number to pick. I picked 8. It’s no big deal.”

But the ongoing emphasis on this metric may lead some patients to expect total pain relief, a problematic expectation in those with chronic pain syndromes such as fibromyalgia. As Dr. Reznikoff points out, a large proportion of patients report they have chronic pain, and many (but clearly not all) suffer from recognized or masked chronic anxiety and depression disorders¹ that may well influence how they use pain medications.

Thus, while physicians indeed are on the front lines of offering initial prescriptions for pain medications, we remain betwixt and between in the challenges of responding to the immediate needs of our patients while trying to predict the long-term effects of our prescription on the individual patient and of our prescribing patterns on society in general.

I again welcome your submissions describing how individual publications have affected your personal approach to managing patients and specific diseases. We will publish selected contributions in print and online.

Several months ago, we invited readers to submit short personalized commentaries on articles that changed the way they approach a specific clinical problem and the way they take care of patients. In this issue of the Journal, addiction specialist Charles Reznikoff, MD, discusses 3 observational studies that focused on how prescribing opioids for acute pain can lead to chronic opioid use and addiction, and how these studies have influenced his practice.

Although observational studies rank lower on the level-of-evidence scale than randomized controlled trials, they can intellectually stimulate and inform us in ways that lead us to modify how we deliver clinical care.

The initial prescribing of pain medications and the management of patients with chronic pain are currently under intense scrutiny, and are the topic of much discussion in the United States. The opioid epidemic has spilled over into all aspects of daily life, far beyond the medical community. But since we physicians are the only legal and regulated source of narcotics and other pain medications, we are under the microscope—and rightly so.

We, our patients, the pharmaceutical industry, legislators, and the law enforcement community struggle to navigate a complex maze, one with moving walls. Not long ago, physicians were told that we were not attentive enough to our patients’ suffering and needed to do better at relieving it. “Pain” became a vital sign and a recorded metric of quality care. Some excellent changes evolved from this focus, such as increased emphasis on postoperative regional and local pain control. But pain measurements continue to be recorded at every outpatient visit, an almost mindless requirement.

Recently, a patient with lupus nephritis whom I was seeing for blood pressure management reported a pain level of 8 on a scale of 10. I confess that I usually don’t even look at these metrics, but for whatever reason I saw her answer. I asked her about it. She had burned her finger while cooking and said, “I had no idea what number to pick. I picked 8. It’s no big deal.”

But the ongoing emphasis on this metric may lead some patients to expect total pain relief, a problematic expectation in those with chronic pain syndromes such as fibromyalgia. As Dr. Reznikoff points out, a large proportion of patients report they have chronic pain, and many (but clearly not all) suffer from recognized or masked chronic anxiety and depression disorders¹ that may well influence how they use pain medications.

Thus, while physicians indeed are on the front lines of offering initial prescriptions for pain medications, we remain betwixt and between in the challenges of responding to the immediate needs of our patients while trying to predict the long-term effects of our prescription on the individual patient and of our prescribing patterns on society in general.

I again welcome your submissions describing how individual publications have affected your personal approach to managing patients and specific diseases. We will publish selected contributions in print and online.

Several months ago, we invited readers to submit short personalized commentaries on articles that changed the way they approach a specific clinical problem and the way they take care of patients. In this issue of the Journal, addiction specialist Charles Reznikoff, MD, discusses 3 observational studies that focused on how prescribing opioids for acute pain can lead to chronic opioid use and addiction, and how these studies have influenced his practice.

Although observational studies rank lower on the level-of-evidence scale than randomized controlled trials, they can intellectually stimulate and inform us in ways that lead us to modify how we deliver clinical care.

The initial prescribing of pain medications and the management of patients with chronic pain are currently under intense scrutiny, and are the topic of much discussion in the United States. The opioid epidemic has spilled over into all aspects of daily life, far beyond the medical community. But since we physicians are the only legal and regulated source of narcotics and other pain medications, we are under the microscope—and rightly so.

We, our patients, the pharmaceutical industry, legislators, and the law enforcement community struggle to navigate a complex maze, one with moving walls. Not long ago, physicians were told that we were not attentive enough to our patients’ suffering and needed to do better at relieving it. “Pain” became a vital sign and a recorded metric of quality care. Some excellent changes evolved from this focus, such as increased emphasis on postoperative regional and local pain control. But pain measurements continue to be recorded at every outpatient visit, an almost mindless requirement.

Recently, a patient with lupus nephritis whom I was seeing for blood pressure management reported a pain level of 8 on a scale of 10. I confess that I usually don’t even look at these metrics, but for whatever reason I saw her answer. I asked her about it. She had burned her finger while cooking and said, “I had no idea what number to pick. I picked 8. It’s no big deal.”

But the ongoing emphasis on this metric may lead some patients to expect total pain relief, a problematic expectation in those with chronic pain syndromes such as fibromyalgia. As Dr. Reznikoff points out, a large proportion of patients report they have chronic pain, and many (but clearly not all) suffer from recognized or masked chronic anxiety and depression disorders¹ that may well influence how they use pain medications.

Thus, while physicians indeed are on the front lines of offering initial prescriptions for pain medications, we remain betwixt and between in the challenges of responding to the immediate needs of our patients while trying to predict the long-term effects of our prescription on the individual patient and of our prescribing patterns on society in general.

I again welcome your submissions describing how individual publications have affected your personal approach to managing patients and specific diseases. We will publish selected contributions in print and online.

A 49-year-old man developed nephrotic-range proteinuria (urine protein–creatinine ratio 4.1 g/g), and primary membranous nephropathy was diagnosed by kidney biopsy. He declined therapy apart from angiotensin receptor blockade.

Five months after undergoing the biopsy, he presented to the emergency room with marked dyspnea, cough, and epigastric discomfort. His blood pressure was 160/100 mm Hg, heart rate 95 beats/minute, and oxygen saturation by pulse oximetry 97% at rest on ambient air, decreasing to 92% with ambulation.

Initial laboratory testing results were as follows:

• Sodium 135 mmol/L (reference range 136–144)
• Potassium 3.9 mmol/L (3.7–5.1)
• Chloride 104 mmol/L (97–105)
• Bicarbonate 21 mmol/L (22–30)
• Blood urea nitrogen 14 mg/dL (9–24)
• Serum creatinine 1.1 mg/dL (0.73–1.22)
• Albumin 2.1 g/dL (3.4–4.9).

Urinalysis revealed the following:

• 5 red blood cells per high-power field, compared with 1 to 2 previously
• 3+ proteinuria
• Urine protein–creatinine ratio 11 g/g
• No glucosuria.

Electrocardiography revealed normal sinus rhythm without ischemic changes. Chest radiography did not show consolidation.

At 7 months after the thrombotic event, there was no evidence of residual renal vein thrombosis on magnetic resonance venography, and at 14 months his serum creatinine level was 0.9 mg/dL, albumin 4.0 g/dL, and urine protein–creatinine ratio 0.8 g/g.

RENAL VEIN THROMBOSIS: RISK FACTORS AND CLINICAL FEATURES

Severe hypoalbuminemia in the setting of nephrotic syndrome due to membranous neph­ropathy is associated with the highest risk of venous thromboembolic events, with renal vein thrombus being the classic complication.¹ Venous thromboembolic events also occur in other nephrotic syndromes, albeit at a lower frequency.²

Venous thromboembolic events are estimated to occur in 7% to 33% of patients with membranous glomerulopathy, with albumin levels less than 2.8 g/dL considered a notable risk factor.^1,2

While often a chronic complication, acute renal vein thrombosis may present with flank pain and hematuria.³ In our patient, the dramatic increase in proteinuria and possibly the increase in hematuria suggested renal vein thrombosis. Proximal tubular dysfunction, such as glucosuria, can be seen on occasion.

DIAGNOSIS AND TREATMENT

Screening asymptomatic patients for renal vein thrombosis is not recommended, and the decision to start prophylactic anticoagulation must be individualized.⁴

Although renal venography historically was the gold standard test to diagnose renal vein thrombosis, it has been replaced by noninvasive imaging such as computed tomography and magnetic resonance venography.

While anticoagulation remains the treatment of choice, catheter-directed thrombectomy or surgical thrombectomy can be considered for some patients with acute renal vein thrombosis.⁵

A 49-year-old man developed nephrotic-range proteinuria (urine protein–creatinine ratio 4.1 g/g), and primary membranous nephropathy was diagnosed by kidney biopsy. He declined therapy apart from angiotensin receptor blockade.

Five months after undergoing the biopsy, he presented to the emergency room with marked dyspnea, cough, and epigastric discomfort. His blood pressure was 160/100 mm Hg, heart rate 95 beats/minute, and oxygen saturation by pulse oximetry 97% at rest on ambient air, decreasing to 92% with ambulation.

Initial laboratory testing results were as follows:

• Sodium 135 mmol/L (reference range 136–144)
• Potassium 3.9 mmol/L (3.7–5.1)
• Chloride 104 mmol/L (97–105)
• Bicarbonate 21 mmol/L (22–30)
• Blood urea nitrogen 14 mg/dL (9–24)
• Serum creatinine 1.1 mg/dL (0.73–1.22)
• Albumin 2.1 g/dL (3.4–4.9).

Urinalysis revealed the following:

• 5 red blood cells per high-power field, compared with 1 to 2 previously
• 3+ proteinuria
• Urine protein–creatinine ratio 11 g/g
• No glucosuria.

Electrocardiography revealed normal sinus rhythm without ischemic changes. Chest radiography did not show consolidation.

At 7 months after the thrombotic event, there was no evidence of residual renal vein thrombosis on magnetic resonance venography, and at 14 months his serum creatinine level was 0.9 mg/dL, albumin 4.0 g/dL, and urine protein–creatinine ratio 0.8 g/g.

RENAL VEIN THROMBOSIS: RISK FACTORS AND CLINICAL FEATURES

Severe hypoalbuminemia in the setting of nephrotic syndrome due to membranous neph­ropathy is associated with the highest risk of venous thromboembolic events, with renal vein thrombus being the classic complication.¹ Venous thromboembolic events also occur in other nephrotic syndromes, albeit at a lower frequency.²

Venous thromboembolic events are estimated to occur in 7% to 33% of patients with membranous glomerulopathy, with albumin levels less than 2.8 g/dL considered a notable risk factor.^1,2

While often a chronic complication, acute renal vein thrombosis may present with flank pain and hematuria.³ In our patient, the dramatic increase in proteinuria and possibly the increase in hematuria suggested renal vein thrombosis. Proximal tubular dysfunction, such as glucosuria, can be seen on occasion.

DIAGNOSIS AND TREATMENT

Screening asymptomatic patients for renal vein thrombosis is not recommended, and the decision to start prophylactic anticoagulation must be individualized.⁴

Although renal venography historically was the gold standard test to diagnose renal vein thrombosis, it has been replaced by noninvasive imaging such as computed tomography and magnetic resonance venography.

While anticoagulation remains the treatment of choice, catheter-directed thrombectomy or surgical thrombectomy can be considered for some patients with acute renal vein thrombosis.⁵

A 49-year-old man developed nephrotic-range proteinuria (urine protein–creatinine ratio 4.1 g/g), and primary membranous nephropathy was diagnosed by kidney biopsy. He declined therapy apart from angiotensin receptor blockade.

Five months after undergoing the biopsy, he presented to the emergency room with marked dyspnea, cough, and epigastric discomfort. His blood pressure was 160/100 mm Hg, heart rate 95 beats/minute, and oxygen saturation by pulse oximetry 97% at rest on ambient air, decreasing to 92% with ambulation.

Initial laboratory testing results were as follows:

• Sodium 135 mmol/L (reference range 136–144)
• Potassium 3.9 mmol/L (3.7–5.1)
• Chloride 104 mmol/L (97–105)
• Bicarbonate 21 mmol/L (22–30)
• Blood urea nitrogen 14 mg/dL (9–24)
• Serum creatinine 1.1 mg/dL (0.73–1.22)
• Albumin 2.1 g/dL (3.4–4.9).

Urinalysis revealed the following:

• 5 red blood cells per high-power field, compared with 1 to 2 previously
• 3+ proteinuria
• Urine protein–creatinine ratio 11 g/g
• No glucosuria.

Electrocardiography revealed normal sinus rhythm without ischemic changes. Chest radiography did not show consolidation.

At 7 months after the thrombotic event, there was no evidence of residual renal vein thrombosis on magnetic resonance venography, and at 14 months his serum creatinine level was 0.9 mg/dL, albumin 4.0 g/dL, and urine protein–creatinine ratio 0.8 g/g.

RENAL VEIN THROMBOSIS: RISK FACTORS AND CLINICAL FEATURES

Severe hypoalbuminemia in the setting of nephrotic syndrome due to membranous neph­ropathy is associated with the highest risk of venous thromboembolic events, with renal vein thrombus being the classic complication.¹ Venous thromboembolic events also occur in other nephrotic syndromes, albeit at a lower frequency.²

Venous thromboembolic events are estimated to occur in 7% to 33% of patients with membranous glomerulopathy, with albumin levels less than 2.8 g/dL considered a notable risk factor.^1,2

While often a chronic complication, acute renal vein thrombosis may present with flank pain and hematuria.³ In our patient, the dramatic increase in proteinuria and possibly the increase in hematuria suggested renal vein thrombosis. Proximal tubular dysfunction, such as glucosuria, can be seen on occasion.

DIAGNOSIS AND TREATMENT

Screening asymptomatic patients for renal vein thrombosis is not recommended, and the decision to start prophylactic anticoagulation must be individualized.⁴

Although renal venography historically was the gold standard test to diagnose renal vein thrombosis, it has been replaced by noninvasive imaging such as computed tomography and magnetic resonance venography.

While anticoagulation remains the treatment of choice, catheter-directed thrombectomy or surgical thrombectomy can be considered for some patients with acute renal vein thrombosis.⁵

A 63-year-old woman presented with an acute exacerbation of chronic back pain after a fall. She was taking warfarin because of a history of factor V Leiden, deep vein thrombosis, and pulmonary embolism, for which a temporary inferior vena cava (IVC) filter had been placed 8 years ago. Her physicians had subsequently tried to remove the filter, without success. Some time after that, 1 of the filter struts had been removed after it migrated through her abdominal wall.

Laboratory testing revealed a supratherapeutic international normalized ratio of 8.5.

The patient underwent endovascular aneurysm repair with adequate placement of a vascular graft. She was discharged on therapeutic anticoagulation, and her back pain had notably improved.

COMPLICATIONS OF IVC FILTERS

In the United States, the use of IVC filters has increased significantly over the last decade, with placement rates ranging from 12% to 17% in patients with venous thromboembolism.¹

The American Heart Association recommends filter placement for patients with venous thromboembolism for whom anticoagulation has failed or is contraindicated, patients unable to withstand pulmonary embolism, and patients who are hemodynamically unstable.² While indications vary in the guidelines released by different societies, filters are most often placed in patients who have an acute bleed, significant surgery after admission for venous thromboembolism, metastatic cancer, and severe illness.³

Complications can occur during and after insertion and during removal. They are more frequent with temporary than with permanent filters, and include filter movement and fracture as well as occlusion and penetration.^4,5

In our patient, we believe that the 3 remaining filter struts likely penetrated the wall of the IVC to the extent that they encountered adjacent structures (aorta, duodenum, kidney).

Of cases of IVC filter penetration reported to a US Food and Drug Administration database, 13.1% involved small bowel perforation, 6.5% involved aortic perforation, and 4.2% involved retroperitoneal bleeding. Symptoms such as abdominal and back pain were present in 38.3% of cases involving IVC penetration.⁵

Therefore, the differential diagnosis for patients with a history of IVC filter placement presenting with these symptoms should address filter complications, including occlusion,  incorrect placement, fracture, migration, and penetration of the filter.⁴ If complications occur, treatment options include anticoagulation, endovascular repair, and surgical intervention.

A 63-year-old woman presented with an acute exacerbation of chronic back pain after a fall. She was taking warfarin because of a history of factor V Leiden, deep vein thrombosis, and pulmonary embolism, for which a temporary inferior vena cava (IVC) filter had been placed 8 years ago. Her physicians had subsequently tried to remove the filter, without success. Some time after that, 1 of the filter struts had been removed after it migrated through her abdominal wall.

Laboratory testing revealed a supratherapeutic international normalized ratio of 8.5.

The patient underwent endovascular aneurysm repair with adequate placement of a vascular graft. She was discharged on therapeutic anticoagulation, and her back pain had notably improved.

COMPLICATIONS OF IVC FILTERS

In the United States, the use of IVC filters has increased significantly over the last decade, with placement rates ranging from 12% to 17% in patients with venous thromboembolism.¹

The American Heart Association recommends filter placement for patients with venous thromboembolism for whom anticoagulation has failed or is contraindicated, patients unable to withstand pulmonary embolism, and patients who are hemodynamically unstable.² While indications vary in the guidelines released by different societies, filters are most often placed in patients who have an acute bleed, significant surgery after admission for venous thromboembolism, metastatic cancer, and severe illness.³

Complications can occur during and after insertion and during removal. They are more frequent with temporary than with permanent filters, and include filter movement and fracture as well as occlusion and penetration.^4,5

In our patient, we believe that the 3 remaining filter struts likely penetrated the wall of the IVC to the extent that they encountered adjacent structures (aorta, duodenum, kidney).

Of cases of IVC filter penetration reported to a US Food and Drug Administration database, 13.1% involved small bowel perforation, 6.5% involved aortic perforation, and 4.2% involved retroperitoneal bleeding. Symptoms such as abdominal and back pain were present in 38.3% of cases involving IVC penetration.⁵

Therefore, the differential diagnosis for patients with a history of IVC filter placement presenting with these symptoms should address filter complications, including occlusion,  incorrect placement, fracture, migration, and penetration of the filter.⁴ If complications occur, treatment options include anticoagulation, endovascular repair, and surgical intervention.

A 63-year-old woman presented with an acute exacerbation of chronic back pain after a fall. She was taking warfarin because of a history of factor V Leiden, deep vein thrombosis, and pulmonary embolism, for which a temporary inferior vena cava (IVC) filter had been placed 8 years ago. Her physicians had subsequently tried to remove the filter, without success. Some time after that, 1 of the filter struts had been removed after it migrated through her abdominal wall.

Laboratory testing revealed a supratherapeutic international normalized ratio of 8.5.

The patient underwent endovascular aneurysm repair with adequate placement of a vascular graft. She was discharged on therapeutic anticoagulation, and her back pain had notably improved.

COMPLICATIONS OF IVC FILTERS

In the United States, the use of IVC filters has increased significantly over the last decade, with placement rates ranging from 12% to 17% in patients with venous thromboembolism.¹

The American Heart Association recommends filter placement for patients with venous thromboembolism for whom anticoagulation has failed or is contraindicated, patients unable to withstand pulmonary embolism, and patients who are hemodynamically unstable.² While indications vary in the guidelines released by different societies, filters are most often placed in patients who have an acute bleed, significant surgery after admission for venous thromboembolism, metastatic cancer, and severe illness.³

Complications can occur during and after insertion and during removal. They are more frequent with temporary than with permanent filters, and include filter movement and fracture as well as occlusion and penetration.^4,5

In our patient, we believe that the 3 remaining filter struts likely penetrated the wall of the IVC to the extent that they encountered adjacent structures (aorta, duodenum, kidney).

Of cases of IVC filter penetration reported to a US Food and Drug Administration database, 13.1% involved small bowel perforation, 6.5% involved aortic perforation, and 4.2% involved retroperitoneal bleeding. Symptoms such as abdominal and back pain were present in 38.3% of cases involving IVC penetration.⁵

Therefore, the differential diagnosis for patients with a history of IVC filter placement presenting with these symptoms should address filter complications, including occlusion,  incorrect placement, fracture, migration, and penetration of the filter.⁴ If complications occur, treatment options include anticoagulation, endovascular repair, and surgical intervention.

Most patients admitted with pulmonary embolism (PE) do not need transthoracic echocardiography (TTE); it should be performed in hemodynamically unstable patients, as well as in hemodynamically stable patients with specific elevated cardiac biomarkers and imaging features.

The decision to perform TTE should be based on clinical presentation, PE burden, and imaging findings (eg, computed tomographic angiography). TTE helps to stratify risk, guide management, monitor response to therapy, and give prognostic information for a subset of patients at increased risk for PE-related adverse events.

RISK STRATIFICATION IN PULMONARY EMBOLISM

PE has a spectrum of presentations ranging from no symptoms to shock. Based on the clinical presentation, PE can be categorized as high, intermediate, or low risk.

High-risk PE, often referred to as “massive” PE, is defined in current American Heart Association guidelines as acute PE with sustained hypotension (systolic blood pressure < 90 mm Hg for at least 15 minutes or requiring inotropic support), persistent profound bradycardia (heart rate < 40 beats per minute with signs or symptoms of shock), syncope, or cardiac arrest.¹

Intermediate-risk or “submassive” PE is more challenging to identify because patients are more hemodynamically stable, yet have evidence on electrocardiography, TTE, computed tomography, or cardiac biomarker testing—ie, N-terminal pro-B-type natriuretic peptide (NT-proBNP) or troponin—that indicates myocardial injury or volume overload.¹

Low-risk PE is acute PE in the absence of clinical markers of adverse prognosis that define massive or submassive PE.¹

ECHOCARDIOGRAPHIC FEATURES OF HIGH-RISK PULMONARY EMBOLISM

Certain TTE findings suggest increased risk of a poor outcome and may warrant therapy that is more invasive and aggressive. High-risk features include the following:

• Impaired right ventricular function
• Interventricular septum bulging into the left ventricle (“D-shaped” septum)
• Dilated proximal pulmonary arteries
• Increased severity of tricuspid regurgitation
• Elevated right atrial pressure
• Elevated pulmonary artery pressure
• Free-floating right ventricular thrombi, which are associated with a mortality rate of up to 45% and can be detected in 7% to 18% of patients⁶
• Tricuspid annular plane systolic excursion, an echocardiographic measure of right ventricular function¹; a value less than 17 mm suggests impaired right ventricular systolic function⁷
• The McConnell sign, a feature of acute massive PE: akinesia of the mid-free wall of the right ventricle and hypercontractility of the apex.

These TTE findings often lead to treatment with thrombolysis, transfer to the intensive care unit, and activation of the interventional team for catheter-based therapies.^1,8 Free-floating right heart thrombi or thrombus straddling the interatrial septum (“thrombus in transit”) through a patent foramen ovale may require surgical embolectomy.⁸

PATIENT SELECTION AND INDICATIONS FOR ECHOCARDIOGRAPHY

Most patients admitted with pulmonary embolism (PE) do not need transthoracic echocardiography (TTE); it should be performed in hemodynamically unstable patients, as well as in hemodynamically stable patients with specific elevated cardiac biomarkers and imaging features.

The decision to perform TTE should be based on clinical presentation, PE burden, and imaging findings (eg, computed tomographic angiography). TTE helps to stratify risk, guide management, monitor response to therapy, and give prognostic information for a subset of patients at increased risk for PE-related adverse events.

RISK STRATIFICATION IN PULMONARY EMBOLISM

PE has a spectrum of presentations ranging from no symptoms to shock. Based on the clinical presentation, PE can be categorized as high, intermediate, or low risk.

High-risk PE, often referred to as “massive” PE, is defined in current American Heart Association guidelines as acute PE with sustained hypotension (systolic blood pressure < 90 mm Hg for at least 15 minutes or requiring inotropic support), persistent profound bradycardia (heart rate < 40 beats per minute with signs or symptoms of shock), syncope, or cardiac arrest.¹

Intermediate-risk or “submassive” PE is more challenging to identify because patients are more hemodynamically stable, yet have evidence on electrocardiography, TTE, computed tomography, or cardiac biomarker testing—ie, N-terminal pro-B-type natriuretic peptide (NT-proBNP) or troponin—that indicates myocardial injury or volume overload.¹

Low-risk PE is acute PE in the absence of clinical markers of adverse prognosis that define massive or submassive PE.¹

ECHOCARDIOGRAPHIC FEATURES OF HIGH-RISK PULMONARY EMBOLISM

Certain TTE findings suggest increased risk of a poor outcome and may warrant therapy that is more invasive and aggressive. High-risk features include the following:

• Impaired right ventricular function
• Interventricular septum bulging into the left ventricle (“D-shaped” septum)
• Dilated proximal pulmonary arteries
• Increased severity of tricuspid regurgitation
• Elevated right atrial pressure
• Elevated pulmonary artery pressure
• Free-floating right ventricular thrombi, which are associated with a mortality rate of up to 45% and can be detected in 7% to 18% of patients⁶
• Tricuspid annular plane systolic excursion, an echocardiographic measure of right ventricular function¹; a value less than 17 mm suggests impaired right ventricular systolic function⁷
• The McConnell sign, a feature of acute massive PE: akinesia of the mid-free wall of the right ventricle and hypercontractility of the apex.

These TTE findings often lead to treatment with thrombolysis, transfer to the intensive care unit, and activation of the interventional team for catheter-based therapies.^1,8 Free-floating right heart thrombi or thrombus straddling the interatrial septum (“thrombus in transit”) through a patent foramen ovale may require surgical embolectomy.⁸

PATIENT SELECTION AND INDICATIONS FOR ECHOCARDIOGRAPHY

Most patients admitted with pulmonary embolism (PE) do not need transthoracic echocardiography (TTE); it should be performed in hemodynamically unstable patients, as well as in hemodynamically stable patients with specific elevated cardiac biomarkers and imaging features.

The decision to perform TTE should be based on clinical presentation, PE burden, and imaging findings (eg, computed tomographic angiography). TTE helps to stratify risk, guide management, monitor response to therapy, and give prognostic information for a subset of patients at increased risk for PE-related adverse events.

RISK STRATIFICATION IN PULMONARY EMBOLISM

PE has a spectrum of presentations ranging from no symptoms to shock. Based on the clinical presentation, PE can be categorized as high, intermediate, or low risk.

High-risk PE, often referred to as “massive” PE, is defined in current American Heart Association guidelines as acute PE with sustained hypotension (systolic blood pressure < 90 mm Hg for at least 15 minutes or requiring inotropic support), persistent profound bradycardia (heart rate < 40 beats per minute with signs or symptoms of shock), syncope, or cardiac arrest.¹

Intermediate-risk or “submassive” PE is more challenging to identify because patients are more hemodynamically stable, yet have evidence on electrocardiography, TTE, computed tomography, or cardiac biomarker testing—ie, N-terminal pro-B-type natriuretic peptide (NT-proBNP) or troponin—that indicates myocardial injury or volume overload.¹

Low-risk PE is acute PE in the absence of clinical markers of adverse prognosis that define massive or submassive PE.¹

ECHOCARDIOGRAPHIC FEATURES OF HIGH-RISK PULMONARY EMBOLISM

Certain TTE findings suggest increased risk of a poor outcome and may warrant therapy that is more invasive and aggressive. High-risk features include the following:

• Impaired right ventricular function
• Interventricular septum bulging into the left ventricle (“D-shaped” septum)
• Dilated proximal pulmonary arteries
• Increased severity of tricuspid regurgitation
• Elevated right atrial pressure
• Elevated pulmonary artery pressure
• Free-floating right ventricular thrombi, which are associated with a mortality rate of up to 45% and can be detected in 7% to 18% of patients⁶
• Tricuspid annular plane systolic excursion, an echocardiographic measure of right ventricular function¹; a value less than 17 mm suggests impaired right ventricular systolic function⁷
• The McConnell sign, a feature of acute massive PE: akinesia of the mid-free wall of the right ventricle and hypercontractility of the apex.

These TTE findings often lead to treatment with thrombolysis, transfer to the intensive care unit, and activation of the interventional team for catheter-based therapies.^1,8 Free-floating right heart thrombi or thrombus straddling the interatrial septum (“thrombus in transit”) through a patent foramen ovale may require surgical embolectomy.⁸

PATIENT SELECTION AND INDICATIONS FOR ECHOCARDIOGRAPHY

A 76-year-old man whose history included abdominal aortic aneurysm repair, bilateral femoral artery bypass for popliteal artery aneurysm, hypertension, and peptic ulcer disease was admitted to a community hospital with pleuritic chest pain and shortness of breath. Two days earlier, he had undergone repair of a ventral hernia.

At the time of that admission, he reported no fever, chills, night sweats, cough, or history of heart or lung disease. His vital signs were normal, and physical examination had revealed no apparent respiratory distress, no jugular venous distention, normal heart sounds, and no pedal edema; however, decreased air entry was noted in the right lung base. Initial serum levels of troponin and N-terminal pro-B-type natriuretic peptide were normal.

At that time, computed tomographic angiography of the chest showed segmental pulmonary emboli in the left upper and right lower lobes of the lungs and right pleural effusion. Transthoracic echocardiography showed normal atrial and ventricular sizes with no right or left ventricular systolic dysfunction and a left ventricular ejection fraction of 59%.

Treatment with intravenous heparin was started, and the patient was transferred to our hospital.

PLEURAL EFFUSION AND PULMONARY EMBOLISM

1. Which of the following is true about pleural effusion?

• It is rarely, if ever, associated with pulmonary embolism
• Most patients with pleural effusion due to pulmonary embolism do not have pleuritic chest pain
• Pulmonary embolism should be excluded in all cases of pleural effusion without a clear cause

Pulmonary embolism should be excluded in all cases of pleural effusion that do not have a clear cause. As for the other answer choices:

• Pulmonary embolism is the fourth leading cause of pleural effusion in the United States, after heart failure, pneumonia, and malignancy.¹
• About 75% of patients who develop pleural effusion in the setting of pulmonary embolism complain of pleuritic chest pain on the side of the effusion.² Most effusions are unilateral, small, and usually exudative.³

EVALUATION BEGINS: RESULTS OF THORACENTESIS

Our patient continued to receive intravenous heparin.

He underwent thoracentesis on hospital day 3, and 1,000 mL of turbid sanguineous pleural fluid was removed. Analysis of the fluid showed pH 7.27, white blood cell count 3.797 × 10⁹/L with 80% neutrophils, and lactate dehydrogenase (LDH) concentration 736 U/L (a ratio of pleural fluid LDH to a concurrent serum LDH > 0.6 is suggestive of an exudate); the fluid was also sent for culture and cytology. Thoracentesis was terminated early due to cough, and follow-up chest radiography showed a moderate-sized pneumothorax.

Computed tomography (CT) of the chest at this time showed a small wedge-shaped area of lung consolidation in the right lower lobe (also seen on CT done 1 day before admission to our hospital), with an intrinsic air-fluid level suggesting a focal infarct or lung abscess, now obscured by adjacent consolidation and atelectasis. In the interval since the previous CT, the multiloculated right pleural effusion had increased in size (Figure 1).

THE NEXT STEP

2. What is the most appropriate next step for this patient?

• Consult an interventional radiologist for chest tube placement
• Start empiric antibiotic therapy and ask an interventional radiologist to place a chest tube
• Start empiric antibiotic therapy, withhold anticoagulation, and consult a thoracic surgeon
• Start empiric antibiotic therapy and consult a thoracic surgeon while continuing anticoagulation

The most appropriate next step is to start empiric antibiotic therapy and consult a thoracic surgeon while continuing anticoagulation.

In this patient, it is appropriate to initiate antibiotics empirically on the basis of his significant pleural loculations, a wedge-shaped consolidation, and 80% neutrophils in the pleural fluid, all of which suggest infection. The unmasking of a wedge-shaped consolidation after thoracentesis, with a previously noted air-fluid level and an interval increase in multiloculated pleural fluid, raises suspicion of a necrotic infection that may have ruptured into the pleural space, a possible lung infarct, or a malignancy. Hence, simply placing a chest tube may not be enough.

Blood in the pleural fluid does not necessitate withholding anticoagulation unless the bleeding is heavy. A pleural fluid hematocrit greater than 50% of the peripheral blood hematocrit suggests hemothorax and is an indication to withhold anticoagulation.¹ Our patient’s pleural fluid was qualitatively sanguineous but not frankly bloody, and therefore we judged that it was not necessary to stop his heparin.

HOW DOES PULMONARY INFARCTION PRESENT CLINICALLY?

3. Which of the following statements about pulmonary infarction is incorrect?

• Cavitation and infarction are more common with larger emboli
• Cavitation occurs in fewer than 10% of pulmonary infarctions
• Lung abscess develops in more than 50% of pulmonary infarctions
• Pulmonary thromboembolism is the most common cause of pulmonary infarction

Lung abscess develops in far fewer than 50% of cases of pulmonary infarction. The rest of the statements are correct.

Cavitation complicates about 4% to 7% of infarctions and is more common when the infarction is 4 cm or greater in diameter.⁴ These cavities are usually single and predominantly on the right side in the apical or posterior segment of the upper lobe or the apical segment of the right lower lobe, as in our patient.^5–8 CT demonstrating scalloped inner margins and cross-cavity band shadows suggests a cavitary pulmonary infarction.^9,10

Infection and abscess in pulmonary infarction are poorly understood but have been linked to larger infarctions, coexistent congestion or atelectasis, and dental or oropharyngeal infection. In an early series of 550 cases of pulmonary infarction, 23 patients (4.2%) developed lung abscess and 6 (1.1%) developed empyema.¹¹ The mean time to cavitation for an infected pulmonary infarction has been reported to be 18 days.¹²

A reversed halo sign, generally described as a focal, rounded area of ground-glass opacity surrounded by a nearly complete ring of consolidation, has been reported to be more frequent with pulmonary infarction than with other diseases, especially when in the lower lobes.¹³

CASE CONTINUED: THORACOSCOPY

A cardiothoracic surgeon was consulted, intravenous heparin was discontinued, an inferior vena cava filter was placed, and the patient underwent video-assisted thoracoscopy.

Purulent fluid was noted on the lateral aspect of right lower lobe; this appeared to be the ruptured cavitary lesion functioning like an uncontrolled bronchopleural fistula. Two chest tubes, sizes 32F and 28F, were placed after decortication, resection of the lung abscess, and closure of the bronchopleural fistula. No significant air leak was noted after resection of this segment of lung.

Pathologic study showed acute organizing pneumonia with abscess formation; no malignant cells or granulomas were seen (Figure 2). Pleural fluid cultures grew Streptococcus intermedius, while the tissue culture was negative for any growth, including acid-fast bacilli and fungi.

On 3 different occasions, both chest tubes were shortened, backed out 2 cm, and resecured with sutures and pins, and Heimlich valves were applied before the patient was discharged.

Intravenous piperacillin-tazobactam was started on the fifth hospital day. On discharge, the patient was advised to continue this treatment for 3 weeks at home.

The patient was receiving enoxaparin subcutaneously in prophylactic doses; 72 hours after the thorascopic procedure this was increased to therapeutic doses, continuing after discharge. Bridging to warfarin was not advised in view of his chest tubes.

Our patient appeared to have developed a right lower lobe infarction that cavitated and ruptured into the pleural space, causing a bronchopleural fistula with empyema after a recent pulmonary embolism. Other reported causes of pulmonary infarction in pulmonary embolism are malignancy and heavy clot burden,⁶ but these have not been confirmed in subsequent studies.⁵ Malignancy was ruled out by biopsy of the resected portion of the lung, and our patient did not have a history of heart failure. A clear cavity was not noted (because it ruptured into the pleura), but an air-fluid level was described in a wedge-shaped consolidation, suggesting infarction.

How common is pulmonary infarction after pulmonary embolism?

Pulmonary infarction occurs in few patients with pulmonary embolism.¹³ Since the lungs receive oxygen from the airways and have a dual blood supply from the pulmonary and bronchial arteries, they are not particularly vulnerable to ischemia. However, the reported incidence of pulmonary infarction in patients with pulmonary embolism has ranged from 10% to higher than 30%.^5,14,15

The reasons behind pulmonary infarction with complications after pulmonary embolism have varied in different case series in different eras. CT, biopsy, or autopsy studies reveal pulmonary infarction after pulmonary embolism to be more common than suspected by clinical symptoms.

In a Mayo Clinic series of 43 cases of pulmonary infarction diagnosed over a 6-year period by surgical lung biopsy, 18 (42%) of the patients had underlying pulmonary thromboembolism, which was the most common cause.¹⁶

4. Which statement about risk factors for pulmonary infarction in pulmonary embolism is incorrect?

• Heart failure may be a risk factor for pulmonary infarction
• Pulmonary hemorrhage is a risk factor for pulmonary infarction
• Pulmonary infarction is more common with more proximal sites of pulmonary embolism
• Collateral circulation may protect against pulmonary infarction

Infarction is more common with emboli that are distal rather than proximal.

Dalen et al¹⁵ suggested that after pulmonary embolism, pulmonary hemorrhage is an important contributor to the development of pulmonary infarction independent of the presence or absence of associated cardiac or pulmonary disease, but that the effect depends on the site of obstruction.

This idea was first proposed in 1913, when Karsner and Ghoreyeb¹⁷ showed that when pulmonary arteries are completely obstructed, the bronchial arteries take over, except when the embolism is present in a small branch of the pulmonary artery. This is because the physiologic anastomosis between the pulmonary artery and the bronchial arteries is located at the precapillary level of the pulmonary artery, and the bronchial circulation does not take over until the pulmonary arterial pressure in the area of the embolism drops to zero.

Using CT data, Kirchner et al⁵ confirmed that the risk of pulmonary infarction is higher if the obstruction is peripheral, ie, distal.

Using autopsy data, Tsao et al¹⁸ reported a higher risk of pulmonary infarction in embolic occlusion of pulmonary vessels less than 3 mm in diameter.

Collateral circulation has been shown to protect against pulmonary infarction. For example, Miniati et al¹⁴ showed that healthy young patients with pulmonary embolism were more prone to develop pulmonary infarction, probably because they had less efficient collateral systems in the peripheral lung fields. In lung transplant recipients, it has been shown that the risk of infarction decreased with development of collateral circulation.¹⁹

Dalen et al,¹⁵ however, attributed delayed resolution of pulmonary hemorrhage (as measured by resolution of infiltrate on chest radiography) to higher underlying pulmonary venous pressure in patients with heart failure and consequent pulmonary infarction. In comparison, healthy patients without cardiac or pulmonary disease have faster resolution of pulmonary hemorrhage when present, and less likelihood of pulmonary infarction (and death in submassive pulmonary embolism).

Data on the management of infected pulmonary infarction are limited. Mortality rates have been as high as 41% with noninfected and 73% with infected cavitary infarctions.⁴ Some authors have advocated early surgical resection in view of high rates of failure of medical treatment due to lack of blood supply within the cavity and continued risk of infection.

KEY POINTS

In patients with a recently diagnosed pulmonary embolism and concurrent symptoms of bacterial pneumonia, a diagnosis of cavitary pulmonary infarction should be considered.

Consolidations that are pleural-based with sharp, rounded margins and with focal areas of central hyperlucencies representing hemorrhage on the mediastinal windows on CT are more likely to represent a pulmonary infarct.²⁰

A 76-year-old man whose history included abdominal aortic aneurysm repair, bilateral femoral artery bypass for popliteal artery aneurysm, hypertension, and peptic ulcer disease was admitted to a community hospital with pleuritic chest pain and shortness of breath. Two days earlier, he had undergone repair of a ventral hernia.

At the time of that admission, he reported no fever, chills, night sweats, cough, or history of heart or lung disease. His vital signs were normal, and physical examination had revealed no apparent respiratory distress, no jugular venous distention, normal heart sounds, and no pedal edema; however, decreased air entry was noted in the right lung base. Initial serum levels of troponin and N-terminal pro-B-type natriuretic peptide were normal.

At that time, computed tomographic angiography of the chest showed segmental pulmonary emboli in the left upper and right lower lobes of the lungs and right pleural effusion. Transthoracic echocardiography showed normal atrial and ventricular sizes with no right or left ventricular systolic dysfunction and a left ventricular ejection fraction of 59%.

Treatment with intravenous heparin was started, and the patient was transferred to our hospital.

PLEURAL EFFUSION AND PULMONARY EMBOLISM

1. Which of the following is true about pleural effusion?

• It is rarely, if ever, associated with pulmonary embolism
• Most patients with pleural effusion due to pulmonary embolism do not have pleuritic chest pain
• Pulmonary embolism should be excluded in all cases of pleural effusion without a clear cause

Pulmonary embolism should be excluded in all cases of pleural effusion that do not have a clear cause. As for the other answer choices:

• Pulmonary embolism is the fourth leading cause of pleural effusion in the United States, after heart failure, pneumonia, and malignancy.¹
• About 75% of patients who develop pleural effusion in the setting of pulmonary embolism complain of pleuritic chest pain on the side of the effusion.² Most effusions are unilateral, small, and usually exudative.³

EVALUATION BEGINS: RESULTS OF THORACENTESIS

Our patient continued to receive intravenous heparin.

He underwent thoracentesis on hospital day 3, and 1,000 mL of turbid sanguineous pleural fluid was removed. Analysis of the fluid showed pH 7.27, white blood cell count 3.797 × 10⁹/L with 80% neutrophils, and lactate dehydrogenase (LDH) concentration 736 U/L (a ratio of pleural fluid LDH to a concurrent serum LDH > 0.6 is suggestive of an exudate); the fluid was also sent for culture and cytology. Thoracentesis was terminated early due to cough, and follow-up chest radiography showed a moderate-sized pneumothorax.

Computed tomography (CT) of the chest at this time showed a small wedge-shaped area of lung consolidation in the right lower lobe (also seen on CT done 1 day before admission to our hospital), with an intrinsic air-fluid level suggesting a focal infarct or lung abscess, now obscured by adjacent consolidation and atelectasis. In the interval since the previous CT, the multiloculated right pleural effusion had increased in size (Figure 1).

THE NEXT STEP

2. What is the most appropriate next step for this patient?

• Consult an interventional radiologist for chest tube placement
• Start empiric antibiotic therapy and ask an interventional radiologist to place a chest tube
• Start empiric antibiotic therapy, withhold anticoagulation, and consult a thoracic surgeon
• Start empiric antibiotic therapy and consult a thoracic surgeon while continuing anticoagulation

The most appropriate next step is to start empiric antibiotic therapy and consult a thoracic surgeon while continuing anticoagulation.

In this patient, it is appropriate to initiate antibiotics empirically on the basis of his significant pleural loculations, a wedge-shaped consolidation, and 80% neutrophils in the pleural fluid, all of which suggest infection. The unmasking of a wedge-shaped consolidation after thoracentesis, with a previously noted air-fluid level and an interval increase in multiloculated pleural fluid, raises suspicion of a necrotic infection that may have ruptured into the pleural space, a possible lung infarct, or a malignancy. Hence, simply placing a chest tube may not be enough.

Blood in the pleural fluid does not necessitate withholding anticoagulation unless the bleeding is heavy. A pleural fluid hematocrit greater than 50% of the peripheral blood hematocrit suggests hemothorax and is an indication to withhold anticoagulation.¹ Our patient’s pleural fluid was qualitatively sanguineous but not frankly bloody, and therefore we judged that it was not necessary to stop his heparin.

HOW DOES PULMONARY INFARCTION PRESENT CLINICALLY?

3. Which of the following statements about pulmonary infarction is incorrect?

• Cavitation and infarction are more common with larger emboli
• Cavitation occurs in fewer than 10% of pulmonary infarctions
• Lung abscess develops in more than 50% of pulmonary infarctions
• Pulmonary thromboembolism is the most common cause of pulmonary infarction

Lung abscess develops in far fewer than 50% of cases of pulmonary infarction. The rest of the statements are correct.

Cavitation complicates about 4% to 7% of infarctions and is more common when the infarction is 4 cm or greater in diameter.⁴ These cavities are usually single and predominantly on the right side in the apical or posterior segment of the upper lobe or the apical segment of the right lower lobe, as in our patient.^5–8 CT demonstrating scalloped inner margins and cross-cavity band shadows suggests a cavitary pulmonary infarction.^9,10

Infection and abscess in pulmonary infarction are poorly understood but have been linked to larger infarctions, coexistent congestion or atelectasis, and dental or oropharyngeal infection. In an early series of 550 cases of pulmonary infarction, 23 patients (4.2%) developed lung abscess and 6 (1.1%) developed empyema.¹¹ The mean time to cavitation for an infected pulmonary infarction has been reported to be 18 days.¹²

A reversed halo sign, generally described as a focal, rounded area of ground-glass opacity surrounded by a nearly complete ring of consolidation, has been reported to be more frequent with pulmonary infarction than with other diseases, especially when in the lower lobes.¹³

CASE CONTINUED: THORACOSCOPY

A cardiothoracic surgeon was consulted, intravenous heparin was discontinued, an inferior vena cava filter was placed, and the patient underwent video-assisted thoracoscopy.

Purulent fluid was noted on the lateral aspect of right lower lobe; this appeared to be the ruptured cavitary lesion functioning like an uncontrolled bronchopleural fistula. Two chest tubes, sizes 32F and 28F, were placed after decortication, resection of the lung abscess, and closure of the bronchopleural fistula. No significant air leak was noted after resection of this segment of lung.

Pathologic study showed acute organizing pneumonia with abscess formation; no malignant cells or granulomas were seen (Figure 2). Pleural fluid cultures grew Streptococcus intermedius, while the tissue culture was negative for any growth, including acid-fast bacilli and fungi.

On 3 different occasions, both chest tubes were shortened, backed out 2 cm, and resecured with sutures and pins, and Heimlich valves were applied before the patient was discharged.

Intravenous piperacillin-tazobactam was started on the fifth hospital day. On discharge, the patient was advised to continue this treatment for 3 weeks at home.

The patient was receiving enoxaparin subcutaneously in prophylactic doses; 72 hours after the thorascopic procedure this was increased to therapeutic doses, continuing after discharge. Bridging to warfarin was not advised in view of his chest tubes.

Our patient appeared to have developed a right lower lobe infarction that cavitated and ruptured into the pleural space, causing a bronchopleural fistula with empyema after a recent pulmonary embolism. Other reported causes of pulmonary infarction in pulmonary embolism are malignancy and heavy clot burden,⁶ but these have not been confirmed in subsequent studies.⁵ Malignancy was ruled out by biopsy of the resected portion of the lung, and our patient did not have a history of heart failure. A clear cavity was not noted (because it ruptured into the pleura), but an air-fluid level was described in a wedge-shaped consolidation, suggesting infarction.

How common is pulmonary infarction after pulmonary embolism?

Pulmonary infarction occurs in few patients with pulmonary embolism.¹³ Since the lungs receive oxygen from the airways and have a dual blood supply from the pulmonary and bronchial arteries, they are not particularly vulnerable to ischemia. However, the reported incidence of pulmonary infarction in patients with pulmonary embolism has ranged from 10% to higher than 30%.^5,14,15

The reasons behind pulmonary infarction with complications after pulmonary embolism have varied in different case series in different eras. CT, biopsy, or autopsy studies reveal pulmonary infarction after pulmonary embolism to be more common than suspected by clinical symptoms.

In a Mayo Clinic series of 43 cases of pulmonary infarction diagnosed over a 6-year period by surgical lung biopsy, 18 (42%) of the patients had underlying pulmonary thromboembolism, which was the most common cause.¹⁶

4. Which statement about risk factors for pulmonary infarction in pulmonary embolism is incorrect?

• Heart failure may be a risk factor for pulmonary infarction
• Pulmonary hemorrhage is a risk factor for pulmonary infarction
• Pulmonary infarction is more common with more proximal sites of pulmonary embolism
• Collateral circulation may protect against pulmonary infarction

Infarction is more common with emboli that are distal rather than proximal.

Dalen et al¹⁵ suggested that after pulmonary embolism, pulmonary hemorrhage is an important contributor to the development of pulmonary infarction independent of the presence or absence of associated cardiac or pulmonary disease, but that the effect depends on the site of obstruction.

This idea was first proposed in 1913, when Karsner and Ghoreyeb¹⁷ showed that when pulmonary arteries are completely obstructed, the bronchial arteries take over, except when the embolism is present in a small branch of the pulmonary artery. This is because the physiologic anastomosis between the pulmonary artery and the bronchial arteries is located at the precapillary level of the pulmonary artery, and the bronchial circulation does not take over until the pulmonary arterial pressure in the area of the embolism drops to zero.

Using CT data, Kirchner et al⁵ confirmed that the risk of pulmonary infarction is higher if the obstruction is peripheral, ie, distal.

Using autopsy data, Tsao et al¹⁸ reported a higher risk of pulmonary infarction in embolic occlusion of pulmonary vessels less than 3 mm in diameter.

Collateral circulation has been shown to protect against pulmonary infarction. For example, Miniati et al¹⁴ showed that healthy young patients with pulmonary embolism were more prone to develop pulmonary infarction, probably because they had less efficient collateral systems in the peripheral lung fields. In lung transplant recipients, it has been shown that the risk of infarction decreased with development of collateral circulation.¹⁹

Dalen et al,¹⁵ however, attributed delayed resolution of pulmonary hemorrhage (as measured by resolution of infiltrate on chest radiography) to higher underlying pulmonary venous pressure in patients with heart failure and consequent pulmonary infarction. In comparison, healthy patients without cardiac or pulmonary disease have faster resolution of pulmonary hemorrhage when present, and less likelihood of pulmonary infarction (and death in submassive pulmonary embolism).

Data on the management of infected pulmonary infarction are limited. Mortality rates have been as high as 41% with noninfected and 73% with infected cavitary infarctions.⁴ Some authors have advocated early surgical resection in view of high rates of failure of medical treatment due to lack of blood supply within the cavity and continued risk of infection.

KEY POINTS

In patients with a recently diagnosed pulmonary embolism and concurrent symptoms of bacterial pneumonia, a diagnosis of cavitary pulmonary infarction should be considered.

Consolidations that are pleural-based with sharp, rounded margins and with focal areas of central hyperlucencies representing hemorrhage on the mediastinal windows on CT are more likely to represent a pulmonary infarct.²⁰

A 76-year-old man whose history included abdominal aortic aneurysm repair, bilateral femoral artery bypass for popliteal artery aneurysm, hypertension, and peptic ulcer disease was admitted to a community hospital with pleuritic chest pain and shortness of breath. Two days earlier, he had undergone repair of a ventral hernia.

At the time of that admission, he reported no fever, chills, night sweats, cough, or history of heart or lung disease. His vital signs were normal, and physical examination had revealed no apparent respiratory distress, no jugular venous distention, normal heart sounds, and no pedal edema; however, decreased air entry was noted in the right lung base. Initial serum levels of troponin and N-terminal pro-B-type natriuretic peptide were normal.

At that time, computed tomographic angiography of the chest showed segmental pulmonary emboli in the left upper and right lower lobes of the lungs and right pleural effusion. Transthoracic echocardiography showed normal atrial and ventricular sizes with no right or left ventricular systolic dysfunction and a left ventricular ejection fraction of 59%.

Treatment with intravenous heparin was started, and the patient was transferred to our hospital.

PLEURAL EFFUSION AND PULMONARY EMBOLISM

1. Which of the following is true about pleural effusion?

• It is rarely, if ever, associated with pulmonary embolism
• Most patients with pleural effusion due to pulmonary embolism do not have pleuritic chest pain
• Pulmonary embolism should be excluded in all cases of pleural effusion without a clear cause

Pulmonary embolism should be excluded in all cases of pleural effusion that do not have a clear cause. As for the other answer choices:

• Pulmonary embolism is the fourth leading cause of pleural effusion in the United States, after heart failure, pneumonia, and malignancy.¹
• About 75% of patients who develop pleural effusion in the setting of pulmonary embolism complain of pleuritic chest pain on the side of the effusion.² Most effusions are unilateral, small, and usually exudative.³

EVALUATION BEGINS: RESULTS OF THORACENTESIS

Our patient continued to receive intravenous heparin.

He underwent thoracentesis on hospital day 3, and 1,000 mL of turbid sanguineous pleural fluid was removed. Analysis of the fluid showed pH 7.27, white blood cell count 3.797 × 10⁹/L with 80% neutrophils, and lactate dehydrogenase (LDH) concentration 736 U/L (a ratio of pleural fluid LDH to a concurrent serum LDH > 0.6 is suggestive of an exudate); the fluid was also sent for culture and cytology. Thoracentesis was terminated early due to cough, and follow-up chest radiography showed a moderate-sized pneumothorax.

Computed tomography (CT) of the chest at this time showed a small wedge-shaped area of lung consolidation in the right lower lobe (also seen on CT done 1 day before admission to our hospital), with an intrinsic air-fluid level suggesting a focal infarct or lung abscess, now obscured by adjacent consolidation and atelectasis. In the interval since the previous CT, the multiloculated right pleural effusion had increased in size (Figure 1).

THE NEXT STEP

2. What is the most appropriate next step for this patient?

• Consult an interventional radiologist for chest tube placement
• Start empiric antibiotic therapy and ask an interventional radiologist to place a chest tube
• Start empiric antibiotic therapy, withhold anticoagulation, and consult a thoracic surgeon
• Start empiric antibiotic therapy and consult a thoracic surgeon while continuing anticoagulation

The most appropriate next step is to start empiric antibiotic therapy and consult a thoracic surgeon while continuing anticoagulation.

In this patient, it is appropriate to initiate antibiotics empirically on the basis of his significant pleural loculations, a wedge-shaped consolidation, and 80% neutrophils in the pleural fluid, all of which suggest infection. The unmasking of a wedge-shaped consolidation after thoracentesis, with a previously noted air-fluid level and an interval increase in multiloculated pleural fluid, raises suspicion of a necrotic infection that may have ruptured into the pleural space, a possible lung infarct, or a malignancy. Hence, simply placing a chest tube may not be enough.

Blood in the pleural fluid does not necessitate withholding anticoagulation unless the bleeding is heavy. A pleural fluid hematocrit greater than 50% of the peripheral blood hematocrit suggests hemothorax and is an indication to withhold anticoagulation.¹ Our patient’s pleural fluid was qualitatively sanguineous but not frankly bloody, and therefore we judged that it was not necessary to stop his heparin.

HOW DOES PULMONARY INFARCTION PRESENT CLINICALLY?

3. Which of the following statements about pulmonary infarction is incorrect?

• Cavitation and infarction are more common with larger emboli
• Cavitation occurs in fewer than 10% of pulmonary infarctions
• Lung abscess develops in more than 50% of pulmonary infarctions
• Pulmonary thromboembolism is the most common cause of pulmonary infarction

Lung abscess develops in far fewer than 50% of cases of pulmonary infarction. The rest of the statements are correct.

Cavitation complicates about 4% to 7% of infarctions and is more common when the infarction is 4 cm or greater in diameter.⁴ These cavities are usually single and predominantly on the right side in the apical or posterior segment of the upper lobe or the apical segment of the right lower lobe, as in our patient.^5–8 CT demonstrating scalloped inner margins and cross-cavity band shadows suggests a cavitary pulmonary infarction.^9,10

Infection and abscess in pulmonary infarction are poorly understood but have been linked to larger infarctions, coexistent congestion or atelectasis, and dental or oropharyngeal infection. In an early series of 550 cases of pulmonary infarction, 23 patients (4.2%) developed lung abscess and 6 (1.1%) developed empyema.¹¹ The mean time to cavitation for an infected pulmonary infarction has been reported to be 18 days.¹²

A reversed halo sign, generally described as a focal, rounded area of ground-glass opacity surrounded by a nearly complete ring of consolidation, has been reported to be more frequent with pulmonary infarction than with other diseases, especially when in the lower lobes.¹³

CASE CONTINUED: THORACOSCOPY

A cardiothoracic surgeon was consulted, intravenous heparin was discontinued, an inferior vena cava filter was placed, and the patient underwent video-assisted thoracoscopy.

Purulent fluid was noted on the lateral aspect of right lower lobe; this appeared to be the ruptured cavitary lesion functioning like an uncontrolled bronchopleural fistula. Two chest tubes, sizes 32F and 28F, were placed after decortication, resection of the lung abscess, and closure of the bronchopleural fistula. No significant air leak was noted after resection of this segment of lung.

Pathologic study showed acute organizing pneumonia with abscess formation; no malignant cells or granulomas were seen (Figure 2). Pleural fluid cultures grew Streptococcus intermedius, while the tissue culture was negative for any growth, including acid-fast bacilli and fungi.

On 3 different occasions, both chest tubes were shortened, backed out 2 cm, and resecured with sutures and pins, and Heimlich valves were applied before the patient was discharged.

Intravenous piperacillin-tazobactam was started on the fifth hospital day. On discharge, the patient was advised to continue this treatment for 3 weeks at home.

The patient was receiving enoxaparin subcutaneously in prophylactic doses; 72 hours after the thorascopic procedure this was increased to therapeutic doses, continuing after discharge. Bridging to warfarin was not advised in view of his chest tubes.

Our patient appeared to have developed a right lower lobe infarction that cavitated and ruptured into the pleural space, causing a bronchopleural fistula with empyema after a recent pulmonary embolism. Other reported causes of pulmonary infarction in pulmonary embolism are malignancy and heavy clot burden,⁶ but these have not been confirmed in subsequent studies.⁵ Malignancy was ruled out by biopsy of the resected portion of the lung, and our patient did not have a history of heart failure. A clear cavity was not noted (because it ruptured into the pleura), but an air-fluid level was described in a wedge-shaped consolidation, suggesting infarction.

How common is pulmonary infarction after pulmonary embolism?

Pulmonary infarction occurs in few patients with pulmonary embolism.¹³ Since the lungs receive oxygen from the airways and have a dual blood supply from the pulmonary and bronchial arteries, they are not particularly vulnerable to ischemia. However, the reported incidence of pulmonary infarction in patients with pulmonary embolism has ranged from 10% to higher than 30%.^5,14,15

The reasons behind pulmonary infarction with complications after pulmonary embolism have varied in different case series in different eras. CT, biopsy, or autopsy studies reveal pulmonary infarction after pulmonary embolism to be more common than suspected by clinical symptoms.

In a Mayo Clinic series of 43 cases of pulmonary infarction diagnosed over a 6-year period by surgical lung biopsy, 18 (42%) of the patients had underlying pulmonary thromboembolism, which was the most common cause.¹⁶

4. Which statement about risk factors for pulmonary infarction in pulmonary embolism is incorrect?

• Heart failure may be a risk factor for pulmonary infarction
• Pulmonary hemorrhage is a risk factor for pulmonary infarction
• Pulmonary infarction is more common with more proximal sites of pulmonary embolism
• Collateral circulation may protect against pulmonary infarction

Infarction is more common with emboli that are distal rather than proximal.

Dalen et al¹⁵ suggested that after pulmonary embolism, pulmonary hemorrhage is an important contributor to the development of pulmonary infarction independent of the presence or absence of associated cardiac or pulmonary disease, but that the effect depends on the site of obstruction.

This idea was first proposed in 1913, when Karsner and Ghoreyeb¹⁷ showed that when pulmonary arteries are completely obstructed, the bronchial arteries take over, except when the embolism is present in a small branch of the pulmonary artery. This is because the physiologic anastomosis between the pulmonary artery and the bronchial arteries is located at the precapillary level of the pulmonary artery, and the bronchial circulation does not take over until the pulmonary arterial pressure in the area of the embolism drops to zero.

Using CT data, Kirchner et al⁵ confirmed that the risk of pulmonary infarction is higher if the obstruction is peripheral, ie, distal.

Using autopsy data, Tsao et al¹⁸ reported a higher risk of pulmonary infarction in embolic occlusion of pulmonary vessels less than 3 mm in diameter.

Collateral circulation has been shown to protect against pulmonary infarction. For example, Miniati et al¹⁴ showed that healthy young patients with pulmonary embolism were more prone to develop pulmonary infarction, probably because they had less efficient collateral systems in the peripheral lung fields. In lung transplant recipients, it has been shown that the risk of infarction decreased with development of collateral circulation.¹⁹

Dalen et al,¹⁵ however, attributed delayed resolution of pulmonary hemorrhage (as measured by resolution of infiltrate on chest radiography) to higher underlying pulmonary venous pressure in patients with heart failure and consequent pulmonary infarction. In comparison, healthy patients without cardiac or pulmonary disease have faster resolution of pulmonary hemorrhage when present, and less likelihood of pulmonary infarction (and death in submassive pulmonary embolism).

Data on the management of infected pulmonary infarction are limited. Mortality rates have been as high as 41% with noninfected and 73% with infected cavitary infarctions.⁴ Some authors have advocated early surgical resection in view of high rates of failure of medical treatment due to lack of blood supply within the cavity and continued risk of infection.

KEY POINTS

In patients with a recently diagnosed pulmonary embolism and concurrent symptoms of bacterial pneumonia, a diagnosis of cavitary pulmonary infarction should be considered.

Consolidations that are pleural-based with sharp, rounded margins and with focal areas of central hyperlucencies representing hemorrhage on the mediastinal windows on CT are more likely to represent a pulmonary infarct.²⁰

Mary, age 38, was hospitalized for acute cholecystitis requiring laparoscopic surgery. Her hospital course was uneventful. At the time of discharge, I, her inpatient doctor, prescribed 15 hydrocodone tablets for postoperative pain. I never saw her again. Did she struggle to stop taking the hydrocodone I prescribed?

Heather is a 50-year-old patient in my addiction medicine clinic who developed opioid use disorder while being treated for chronic pain. After much hardship and to her credit, she is now in long-term remission. Did her opioid use disorder start with an opioid prescription for an accepted indication?

The issues Mary and Heather face seem unrelated, but these 2 patients may be at different time points in the progression of the same disease. As a hospitalist, I want to optimize the chances that patients taking opioids for acute pain will be able to stop taking them.

CHRONIC USE VS OPIOID USE DISORDER

There is a distinction between chronic use of opioids and opioid use disorder. The latter is also known as addiction.

Patients who take opioids daily do not necessarily have opioid use disorder, even if they have physiologic dependence on them. Physiologic opioid dependence is commonly confused with opioid use disorder, but it is the expected result of regularly taking these drugs.

Opioid use disorder is a chronic disease of the brain characterized by loss of control over opioid use, resulting in harm. The Diagnostic and Statistical Manual, fifth edition, excludes physiologic dependence on opioids (tolerance and withdrawal) from its criteria for opioid use disorder if the patient is taking opioids solely under medical supervision.¹ To be diagnosed with opioid use disorder, patients need to do only 2 of the following within 12 months:

• Take more of the drug than intended
• Want or try to cut down without success
• Spend a lot of time in getting, using, or recovering from the drug
• Crave the drug
• Fail to meet commitments due to the drug
• Continue to use the drug, even though it causes social or relationship problems
• Give up or reduce other activities because of the drug
• Use the drug even when it isn’t safe
• Continue to use even when it causes physical or psychological problems
• Develop tolerance (but, as noted, not if taking the drug as directed under a doctor’s supervision)
• Experience withdrawal (again, but not if taking the drug under medical supervision).

WHY DO SOME PATIENTS STRUGGLE TO STOP TAKING OPIOIDS?

Studying opioid use disorder as an outcome in large groups of patients is complicated by imperfect medical documentation. However, using pharmacy claims data, researchers can accurately describe opioid prescription patterns in large groups of patients over time. This means we can count how many patients keep taking prescribed opioids but not how many become addicted.

In a country where nearly 40% of adults are prescribed an opioid annually, the question is not why people start taking opioids, but why some have to struggle to stop.² Several recent studies used pharmacy claims data to identify factors that may predict chronic opioid use in patients prescribed opioids for acute pain. The findings suggest that we can better treat acute pain to prevent chronic opioid use.

We don’t yet know how to protect patients like Mary from opioid use disorder, but the following 3 studies have already changed my practice.

HIGHER TOTAL DOSE MEANS HIGHER RISK

[Shah A, Hayes CJ, Martin BC. Characteristics of initial prescription episodes and likelihood of long-term opioid use—United States, 2006–2015. MMWR Morb Mortal Wkly Rep 2017; 66(10):265–269.]

Shah et al³ reported a study of nearly 1.3 million opioid-naive patients who received opioid prescriptions. Of those prescribed at least 1 day of opioids, 6% were still taking them 1 year later, and 2.9% were still taking them 3 years later.

Opioid exposure in acute pain was measured in total “morphine milligram equivalents” (MME), ie, the cumulative amount of opioids prescribed in the treatment episode, standardized across different types of opioids. We usually think of exposure in terms of how many milligrams a patient takes per day, which correlates with mortality in chronic opioid use.⁴ But this study showed a linear relationship between total MME prescribed for acute pain and ongoing opioid use in opioid-naive patients. By itself, the difference between daily and total MME made the article revelatory.

But the study went further, asking how much is too much: ie, What is the cutoff MME above which the patient is at risk of chronic opioid use? The relationship between acute opioid dose and chronic use is linear and starts early. Shah et al suggested that a total threshold of 700 MME predicts chronic opioid use—140 hydrocodone tablets, or 1 month of regular use.³

Many doctors worry that specific opioids such as oxycodone, hydromorphone, and fentanyl may be more habit-forming. Surprisingly, this study showed that these drugs were associated with rates of chronic use similar to those of other opioids when they controlled for potency.

Bottom line. Total opioid use in acute pain was the best predictor of chronic opioid use, and it showed that chronicity begins earlier than thought.

[Barnett ML, Olenski AR, Jena AB. Opioid-prescribing patterns of emergency physicians and risk of long-term use. N Engl J Med 2017; 376(7):663–673.]

Barnett et al⁵ analyzed opioid prescribing for acute pain in the emergency department, using Medicare pharmacy data from 377,629 previously opioid-naive patients. They categorized the emergency providers into quartiles based on the frequency of opioid prescribing.

The relative risk of ongoing opioid use 1 year after being treated by a “high-intensity” prescriber (ie, one in the top quartile) was 30% greater than in similar patients seen by a low-intensity prescriber (ie, one in the bottom quartile). In addition, those who were treated by high-intensity prescribers were more likely to have a serious fall.

In designing the study, the authors assumed that patients visiting an emergency department had their doctor assigned randomly. They controlled for many patient variables that might have confounded the results, such as age, sex, race, depression, medical comorbidities, and geographic region. Were the higher rates of ongoing opioid use in the high-intensity-prescriber group due to the higher prescribing rates of their emergency providers, or did the providers counsel patients differently? This is not known.

Bottom line. Different doctors manage similar patients differently when it comes to pain, and those who prescribe more opioids for acute pain put their patients at risk of chronic opioid use and falls. I don’t want to be a high-intensity opioid prescriber.

SURGERY AND CHRONIC OPIOID USE

[Brummett CM, Waljee JF, Goesling J, et al. New persistent opioid use after minor and major surgical procedures in US adults. JAMA Surg 2017; 152(6):e170504.]

Brummett et al⁶ examined ongoing opioid use after surgery in 36,177 opioid-naive patients and in a nonsurgical control group. After 3 months, 6% of the patients who underwent surgery remained on opioids, compared with only 0.4% of the nonsurgical controls. Whether the surgery was major or minor did not affect the rate of postoperative opioid use.

Risk factors for ongoing opioid use were preexisting addiction to anything (including tobacco), mood disorders, and preoperative pain disorders. These risk factors have previously been reported in nonsurgical patients.⁷

Brummett et al speculated that patients are counseled about postoperative opioids in a way that leads them to overestimate the safety and efficacy of these drugs for treating other common pain conditions.⁶ 

Bottom line. Patients with mental health comorbidities have a hard time stopping opioids. The remarkable finding in this study was the similarity between major and minor surgery in terms of chronic opioid use. If postoperative opioids treat only the pain caused by the surgery, major surgery should be associated with greater opioid use. The similarity suggests that a mechanism other than postoperative pain confers risk of chronic opioid use.

THINKING ABOUT OPIOIDS

Collectively, these articles describe elements of acute pain treatment that correlate with chronic ongoing opioid use: a higher cumulative dose,³ being seen by a physician who prescribes a lot of opioids,⁵ undergoing surgery,⁶ and psychiatric comorbidity.⁶ They made me wonder if opioid use for acute pain acts as an inoculation, analogous to inoculating a Petri dish with bacteria.  The likelihood of chronic opioid use arises from the inoculum dose, the host response, and the context of inoculation. 

These articles do not show how patients taking opioids chronically for pain become addicted. Stumbo et al⁸ interviewed 283 opioid-dependent patients and identified 5 pathways to opioid use disorder, 3 of which were related to pain control: inadequately controlled chronic pain, exposure to opioids during acute pain episodes, and chronic pain in patients who already had substance use disorders. Brat et al⁹ recently estimated the risk of opioid use disorder after receiving opioids postoperatively to be less than 1%, but it increased dramatically with duration of opioid treatment.

A patient who fills an opioid prescription does not necessarily have chronic pain. Nor do all patients with chronic pain require an opioid prescription. These studies did not establish whether the patients had a pain syndrome. In practice, we call our patients who chronically take opioids our “chronic pain patients.” But 40% of Americans have chronic pain, while only 5% take opioids daily for pain.^11,12

We assume that those taking opioids have the most severe pain. But Brummett et al suggested that continued opioid use is predicted less by pain and more by psychiatric comorbidity.⁶ More than half of the opioid prescriptions in the United States are written for patients with serious mental illness, who represent one-sixth of that population.¹¹ Maybe chronic opioid use for pain has more to do with vulnerability to opioids and less to do with a pain syndrome.

I now think about daily opioid use in much the same way as I think about daily prednisone use. Patients on daily prednisone have a characteristic set of medical risks from the prednisone itself, regardless of its indication. Yet we do not consider these patients addicted to prednisone. Opioid use may be similar.

Like most doctors, I am troubled by the continued rise in the opioid overdose rate.¹³ Yet addiction and death from overdose are not the only risks that patients on chronic opioids face; they also have higher rates of falls, cardiovascular death, pneumonia, death from chronic obstructive pulmonary disease, and motor vehicle crashes.^14–17 Patients on chronic opioids for pain have greater mental health comorbidity and worse function.¹⁸

Most concerning, chronic opioid treatment for pain lacks proof of benefit. In fact, a recent study disproved the benefit of opioids for chronic pain compared with nonopioid options.¹⁹ When I meet with patients who are taking chronic opioids for pain, I often can’t identify why the drugs were started or ought to be continued, and I anticipate a bad outcome. Yet the patient is afraid to stop the drug. For these reasons, chronic opioid use for pain strikes me as worth considering separately from opioid use disorder.

The studies described above have had a powerful effect on my clinical care as a hospitalist.

I now talk to all patients starting opioids about how hard it can be to stop. Some patients are defensive at first, believing this does not apply to them. But I politely continue.

People with depression and anxiety can have a harder time stopping opioids. Addiction is both a risk with ongoing opioid use and a possible outcome of acute opioid use.⁸ But one can struggle to stop opioids without being addicted or depressed. Even the healthiest person may wish to continue opioids past the point of benefit.

I am careful not to invalidate the patient’s experience of pain. It is challenging for patients to find the balance between current discomfort and a possible future adverse effect. In these conversations, I imagine how I would want a loved one counseled on their pain control. This centers me as I choose my words and my tone.

I now monitor the total amount of opioid I prescribe for acute pain in addition to the daily dose. I give my patients as few opioids as reasonable, and advise them to take the minimum dose required for tolerable comfort. I offer nonopioid options as the preferred choice, presenting them as effective and safe. I do this irrespective of the indication for opioids.

I limit opioids in all patients, not just those with comorbidities. I include in my shared decision-making process the risk of chronic opioid use when I prescribe opioids for acute pain, carefully distinguishing it from opioid use disorder. Instead of excess opioids, I give patients my office phone number to call in case they struggle. I rarely get calls. But I find patients would rather have access to a doctor than extra pills. And offering them my contact information lets me limit opioids while letting them know that I am committed to their comfort and health.

As an addiction medicine doctor, I consult on patients not taking their opioids as prescribed. Caring for these patients is intellectually and emotionally draining; they suffer daily, and the opioids they take provide a modicum of relief at a high cost. The publications I have discussed here provide insight into how a troubled relationship with opioids begins. I remind myself that these patients have an iatrogenic condition. Their behaviors that we label “aberrant” may reflect an adverse reaction to medications prescribed to them for acute pain.

Mary, my patient with postoperative pain after cholecystectomy, may over time develop opioid use disorder as Heather did. That progression may have begun with the hydrocodone I prescribed and the counseling I gave her, and it may proceed to chronic opioid use and then opioid use disorder.

I am looking closely at the care I give for acute pain in light of these innovative studies. But even more so, they have increased the compassion with which I care for patients like Heather, those harmed by prescribed opioids.

Mary, age 38, was hospitalized for acute cholecystitis requiring laparoscopic surgery. Her hospital course was uneventful. At the time of discharge, I, her inpatient doctor, prescribed 15 hydrocodone tablets for postoperative pain. I never saw her again. Did she struggle to stop taking the hydrocodone I prescribed?

Heather is a 50-year-old patient in my addiction medicine clinic who developed opioid use disorder while being treated for chronic pain. After much hardship and to her credit, she is now in long-term remission. Did her opioid use disorder start with an opioid prescription for an accepted indication?

The issues Mary and Heather face seem unrelated, but these 2 patients may be at different time points in the progression of the same disease. As a hospitalist, I want to optimize the chances that patients taking opioids for acute pain will be able to stop taking them.

CHRONIC USE VS OPIOID USE DISORDER

There is a distinction between chronic use of opioids and opioid use disorder. The latter is also known as addiction.

Patients who take opioids daily do not necessarily have opioid use disorder, even if they have physiologic dependence on them. Physiologic opioid dependence is commonly confused with opioid use disorder, but it is the expected result of regularly taking these drugs.

Opioid use disorder is a chronic disease of the brain characterized by loss of control over opioid use, resulting in harm. The Diagnostic and Statistical Manual, fifth edition, excludes physiologic dependence on opioids (tolerance and withdrawal) from its criteria for opioid use disorder if the patient is taking opioids solely under medical supervision.¹ To be diagnosed with opioid use disorder, patients need to do only 2 of the following within 12 months:

• Take more of the drug than intended
• Want or try to cut down without success
• Spend a lot of time in getting, using, or recovering from the drug
• Crave the drug
• Fail to meet commitments due to the drug
• Continue to use the drug, even though it causes social or relationship problems
• Give up or reduce other activities because of the drug
• Use the drug even when it isn’t safe
• Continue to use even when it causes physical or psychological problems
• Develop tolerance (but, as noted, not if taking the drug as directed under a doctor’s supervision)
• Experience withdrawal (again, but not if taking the drug under medical supervision).

WHY DO SOME PATIENTS STRUGGLE TO STOP TAKING OPIOIDS?

Studying opioid use disorder as an outcome in large groups of patients is complicated by imperfect medical documentation. However, using pharmacy claims data, researchers can accurately describe opioid prescription patterns in large groups of patients over time. This means we can count how many patients keep taking prescribed opioids but not how many become addicted.

In a country where nearly 40% of adults are prescribed an opioid annually, the question is not why people start taking opioids, but why some have to struggle to stop.² Several recent studies used pharmacy claims data to identify factors that may predict chronic opioid use in patients prescribed opioids for acute pain. The findings suggest that we can better treat acute pain to prevent chronic opioid use.

We don’t yet know how to protect patients like Mary from opioid use disorder, but the following 3 studies have already changed my practice.

HIGHER TOTAL DOSE MEANS HIGHER RISK

[Shah A, Hayes CJ, Martin BC. Characteristics of initial prescription episodes and likelihood of long-term opioid use—United States, 2006–2015. MMWR Morb Mortal Wkly Rep 2017; 66(10):265–269.]

Shah et al³ reported a study of nearly 1.3 million opioid-naive patients who received opioid prescriptions. Of those prescribed at least 1 day of opioids, 6% were still taking them 1 year later, and 2.9% were still taking them 3 years later.

Opioid exposure in acute pain was measured in total “morphine milligram equivalents” (MME), ie, the cumulative amount of opioids prescribed in the treatment episode, standardized across different types of opioids. We usually think of exposure in terms of how many milligrams a patient takes per day, which correlates with mortality in chronic opioid use.⁴ But this study showed a linear relationship between total MME prescribed for acute pain and ongoing opioid use in opioid-naive patients. By itself, the difference between daily and total MME made the article revelatory.

But the study went further, asking how much is too much: ie, What is the cutoff MME above which the patient is at risk of chronic opioid use? The relationship between acute opioid dose and chronic use is linear and starts early. Shah et al suggested that a total threshold of 700 MME predicts chronic opioid use—140 hydrocodone tablets, or 1 month of regular use.³

Many doctors worry that specific opioids such as oxycodone, hydromorphone, and fentanyl may be more habit-forming. Surprisingly, this study showed that these drugs were associated with rates of chronic use similar to those of other opioids when they controlled for potency.

Bottom line. Total opioid use in acute pain was the best predictor of chronic opioid use, and it showed that chronicity begins earlier than thought.

[Barnett ML, Olenski AR, Jena AB. Opioid-prescribing patterns of emergency physicians and risk of long-term use. N Engl J Med 2017; 376(7):663–673.]

Barnett et al⁵ analyzed opioid prescribing for acute pain in the emergency department, using Medicare pharmacy data from 377,629 previously opioid-naive patients. They categorized the emergency providers into quartiles based on the frequency of opioid prescribing.

The relative risk of ongoing opioid use 1 year after being treated by a “high-intensity” prescriber (ie, one in the top quartile) was 30% greater than in similar patients seen by a low-intensity prescriber (ie, one in the bottom quartile). In addition, those who were treated by high-intensity prescribers were more likely to have a serious fall.

In designing the study, the authors assumed that patients visiting an emergency department had their doctor assigned randomly. They controlled for many patient variables that might have confounded the results, such as age, sex, race, depression, medical comorbidities, and geographic region. Were the higher rates of ongoing opioid use in the high-intensity-prescriber group due to the higher prescribing rates of their emergency providers, or did the providers counsel patients differently? This is not known.

Bottom line. Different doctors manage similar patients differently when it comes to pain, and those who prescribe more opioids for acute pain put their patients at risk of chronic opioid use and falls. I don’t want to be a high-intensity opioid prescriber.

SURGERY AND CHRONIC OPIOID USE

[Brummett CM, Waljee JF, Goesling J, et al. New persistent opioid use after minor and major surgical procedures in US adults. JAMA Surg 2017; 152(6):e170504.]

Brummett et al⁶ examined ongoing opioid use after surgery in 36,177 opioid-naive patients and in a nonsurgical control group. After 3 months, 6% of the patients who underwent surgery remained on opioids, compared with only 0.4% of the nonsurgical controls. Whether the surgery was major or minor did not affect the rate of postoperative opioid use.

Risk factors for ongoing opioid use were preexisting addiction to anything (including tobacco), mood disorders, and preoperative pain disorders. These risk factors have previously been reported in nonsurgical patients.⁷

Brummett et al speculated that patients are counseled about postoperative opioids in a way that leads them to overestimate the safety and efficacy of these drugs for treating other common pain conditions.⁶ 

Bottom line. Patients with mental health comorbidities have a hard time stopping opioids. The remarkable finding in this study was the similarity between major and minor surgery in terms of chronic opioid use. If postoperative opioids treat only the pain caused by the surgery, major surgery should be associated with greater opioid use. The similarity suggests that a mechanism other than postoperative pain confers risk of chronic opioid use.

THINKING ABOUT OPIOIDS

Collectively, these articles describe elements of acute pain treatment that correlate with chronic ongoing opioid use: a higher cumulative dose,³ being seen by a physician who prescribes a lot of opioids,⁵ undergoing surgery,⁶ and psychiatric comorbidity.⁶ They made me wonder if opioid use for acute pain acts as an inoculation, analogous to inoculating a Petri dish with bacteria.  The likelihood of chronic opioid use arises from the inoculum dose, the host response, and the context of inoculation. 

These articles do not show how patients taking opioids chronically for pain become addicted. Stumbo et al⁸ interviewed 283 opioid-dependent patients and identified 5 pathways to opioid use disorder, 3 of which were related to pain control: inadequately controlled chronic pain, exposure to opioids during acute pain episodes, and chronic pain in patients who already had substance use disorders. Brat et al⁹ recently estimated the risk of opioid use disorder after receiving opioids postoperatively to be less than 1%, but it increased dramatically with duration of opioid treatment.

A patient who fills an opioid prescription does not necessarily have chronic pain. Nor do all patients with chronic pain require an opioid prescription. These studies did not establish whether the patients had a pain syndrome. In practice, we call our patients who chronically take opioids our “chronic pain patients.” But 40% of Americans have chronic pain, while only 5% take opioids daily for pain.^11,12

We assume that those taking opioids have the most severe pain. But Brummett et al suggested that continued opioid use is predicted less by pain and more by psychiatric comorbidity.⁶ More than half of the opioid prescriptions in the United States are written for patients with serious mental illness, who represent one-sixth of that population.¹¹ Maybe chronic opioid use for pain has more to do with vulnerability to opioids and less to do with a pain syndrome.

I now think about daily opioid use in much the same way as I think about daily prednisone use. Patients on daily prednisone have a characteristic set of medical risks from the prednisone itself, regardless of its indication. Yet we do not consider these patients addicted to prednisone. Opioid use may be similar.

Like most doctors, I am troubled by the continued rise in the opioid overdose rate.¹³ Yet addiction and death from overdose are not the only risks that patients on chronic opioids face; they also have higher rates of falls, cardiovascular death, pneumonia, death from chronic obstructive pulmonary disease, and motor vehicle crashes.^14–17 Patients on chronic opioids for pain have greater mental health comorbidity and worse function.¹⁸

Most concerning, chronic opioid treatment for pain lacks proof of benefit. In fact, a recent study disproved the benefit of opioids for chronic pain compared with nonopioid options.¹⁹ When I meet with patients who are taking chronic opioids for pain, I often can’t identify why the drugs were started or ought to be continued, and I anticipate a bad outcome. Yet the patient is afraid to stop the drug. For these reasons, chronic opioid use for pain strikes me as worth considering separately from opioid use disorder.

The studies described above have had a powerful effect on my clinical care as a hospitalist.

I now talk to all patients starting opioids about how hard it can be to stop. Some patients are defensive at first, believing this does not apply to them. But I politely continue.

People with depression and anxiety can have a harder time stopping opioids. Addiction is both a risk with ongoing opioid use and a possible outcome of acute opioid use.⁸ But one can struggle to stop opioids without being addicted or depressed. Even the healthiest person may wish to continue opioids past the point of benefit.

I am careful not to invalidate the patient’s experience of pain. It is challenging for patients to find the balance between current discomfort and a possible future adverse effect. In these conversations, I imagine how I would want a loved one counseled on their pain control. This centers me as I choose my words and my tone.

I now monitor the total amount of opioid I prescribe for acute pain in addition to the daily dose. I give my patients as few opioids as reasonable, and advise them to take the minimum dose required for tolerable comfort. I offer nonopioid options as the preferred choice, presenting them as effective and safe. I do this irrespective of the indication for opioids.

I limit opioids in all patients, not just those with comorbidities. I include in my shared decision-making process the risk of chronic opioid use when I prescribe opioids for acute pain, carefully distinguishing it from opioid use disorder. Instead of excess opioids, I give patients my office phone number to call in case they struggle. I rarely get calls. But I find patients would rather have access to a doctor than extra pills. And offering them my contact information lets me limit opioids while letting them know that I am committed to their comfort and health.

As an addiction medicine doctor, I consult on patients not taking their opioids as prescribed. Caring for these patients is intellectually and emotionally draining; they suffer daily, and the opioids they take provide a modicum of relief at a high cost. The publications I have discussed here provide insight into how a troubled relationship with opioids begins. I remind myself that these patients have an iatrogenic condition. Their behaviors that we label “aberrant” may reflect an adverse reaction to medications prescribed to them for acute pain.

Mary, my patient with postoperative pain after cholecystectomy, may over time develop opioid use disorder as Heather did. That progression may have begun with the hydrocodone I prescribed and the counseling I gave her, and it may proceed to chronic opioid use and then opioid use disorder.

I am looking closely at the care I give for acute pain in light of these innovative studies. But even more so, they have increased the compassion with which I care for patients like Heather, those harmed by prescribed opioids.

Mary, age 38, was hospitalized for acute cholecystitis requiring laparoscopic surgery. Her hospital course was uneventful. At the time of discharge, I, her inpatient doctor, prescribed 15 hydrocodone tablets for postoperative pain. I never saw her again. Did she struggle to stop taking the hydrocodone I prescribed?

Heather is a 50-year-old patient in my addiction medicine clinic who developed opioid use disorder while being treated for chronic pain. After much hardship and to her credit, she is now in long-term remission. Did her opioid use disorder start with an opioid prescription for an accepted indication?

The issues Mary and Heather face seem unrelated, but these 2 patients may be at different time points in the progression of the same disease. As a hospitalist, I want to optimize the chances that patients taking opioids for acute pain will be able to stop taking them.

CHRONIC USE VS OPIOID USE DISORDER

There is a distinction between chronic use of opioids and opioid use disorder. The latter is also known as addiction.

Patients who take opioids daily do not necessarily have opioid use disorder, even if they have physiologic dependence on them. Physiologic opioid dependence is commonly confused with opioid use disorder, but it is the expected result of regularly taking these drugs.

Opioid use disorder is a chronic disease of the brain characterized by loss of control over opioid use, resulting in harm. The Diagnostic and Statistical Manual, fifth edition, excludes physiologic dependence on opioids (tolerance and withdrawal) from its criteria for opioid use disorder if the patient is taking opioids solely under medical supervision.¹ To be diagnosed with opioid use disorder, patients need to do only 2 of the following within 12 months:

• Take more of the drug than intended
• Want or try to cut down without success
• Spend a lot of time in getting, using, or recovering from the drug
• Crave the drug
• Fail to meet commitments due to the drug
• Continue to use the drug, even though it causes social or relationship problems
• Give up or reduce other activities because of the drug
• Use the drug even when it isn’t safe
• Continue to use even when it causes physical or psychological problems
• Develop tolerance (but, as noted, not if taking the drug as directed under a doctor’s supervision)
• Experience withdrawal (again, but not if taking the drug under medical supervision).

WHY DO SOME PATIENTS STRUGGLE TO STOP TAKING OPIOIDS?

Studying opioid use disorder as an outcome in large groups of patients is complicated by imperfect medical documentation. However, using pharmacy claims data, researchers can accurately describe opioid prescription patterns in large groups of patients over time. This means we can count how many patients keep taking prescribed opioids but not how many become addicted.

In a country where nearly 40% of adults are prescribed an opioid annually, the question is not why people start taking opioids, but why some have to struggle to stop.² Several recent studies used pharmacy claims data to identify factors that may predict chronic opioid use in patients prescribed opioids for acute pain. The findings suggest that we can better treat acute pain to prevent chronic opioid use.

We don’t yet know how to protect patients like Mary from opioid use disorder, but the following 3 studies have already changed my practice.

HIGHER TOTAL DOSE MEANS HIGHER RISK

[Shah A, Hayes CJ, Martin BC. Characteristics of initial prescription episodes and likelihood of long-term opioid use—United States, 2006–2015. MMWR Morb Mortal Wkly Rep 2017; 66(10):265–269.]

Shah et al³ reported a study of nearly 1.3 million opioid-naive patients who received opioid prescriptions. Of those prescribed at least 1 day of opioids, 6% were still taking them 1 year later, and 2.9% were still taking them 3 years later.

Opioid exposure in acute pain was measured in total “morphine milligram equivalents” (MME), ie, the cumulative amount of opioids prescribed in the treatment episode, standardized across different types of opioids. We usually think of exposure in terms of how many milligrams a patient takes per day, which correlates with mortality in chronic opioid use.⁴ But this study showed a linear relationship between total MME prescribed for acute pain and ongoing opioid use in opioid-naive patients. By itself, the difference between daily and total MME made the article revelatory.

But the study went further, asking how much is too much: ie, What is the cutoff MME above which the patient is at risk of chronic opioid use? The relationship between acute opioid dose and chronic use is linear and starts early. Shah et al suggested that a total threshold of 700 MME predicts chronic opioid use—140 hydrocodone tablets, or 1 month of regular use.³

Many doctors worry that specific opioids such as oxycodone, hydromorphone, and fentanyl may be more habit-forming. Surprisingly, this study showed that these drugs were associated with rates of chronic use similar to those of other opioids when they controlled for potency.

Bottom line. Total opioid use in acute pain was the best predictor of chronic opioid use, and it showed that chronicity begins earlier than thought.

[Barnett ML, Olenski AR, Jena AB. Opioid-prescribing patterns of emergency physicians and risk of long-term use. N Engl J Med 2017; 376(7):663–673.]

Barnett et al⁵ analyzed opioid prescribing for acute pain in the emergency department, using Medicare pharmacy data from 377,629 previously opioid-naive patients. They categorized the emergency providers into quartiles based on the frequency of opioid prescribing.

The relative risk of ongoing opioid use 1 year after being treated by a “high-intensity” prescriber (ie, one in the top quartile) was 30% greater than in similar patients seen by a low-intensity prescriber (ie, one in the bottom quartile). In addition, those who were treated by high-intensity prescribers were more likely to have a serious fall.

In designing the study, the authors assumed that patients visiting an emergency department had their doctor assigned randomly. They controlled for many patient variables that might have confounded the results, such as age, sex, race, depression, medical comorbidities, and geographic region. Were the higher rates of ongoing opioid use in the high-intensity-prescriber group due to the higher prescribing rates of their emergency providers, or did the providers counsel patients differently? This is not known.

Bottom line. Different doctors manage similar patients differently when it comes to pain, and those who prescribe more opioids for acute pain put their patients at risk of chronic opioid use and falls. I don’t want to be a high-intensity opioid prescriber.

SURGERY AND CHRONIC OPIOID USE

[Brummett CM, Waljee JF, Goesling J, et al. New persistent opioid use after minor and major surgical procedures in US adults. JAMA Surg 2017; 152(6):e170504.]

Brummett et al⁶ examined ongoing opioid use after surgery in 36,177 opioid-naive patients and in a nonsurgical control group. After 3 months, 6% of the patients who underwent surgery remained on opioids, compared with only 0.4% of the nonsurgical controls. Whether the surgery was major or minor did not affect the rate of postoperative opioid use.

Risk factors for ongoing opioid use were preexisting addiction to anything (including tobacco), mood disorders, and preoperative pain disorders. These risk factors have previously been reported in nonsurgical patients.⁷

Brummett et al speculated that patients are counseled about postoperative opioids in a way that leads them to overestimate the safety and efficacy of these drugs for treating other common pain conditions.⁶ 

Bottom line. Patients with mental health comorbidities have a hard time stopping opioids. The remarkable finding in this study was the similarity between major and minor surgery in terms of chronic opioid use. If postoperative opioids treat only the pain caused by the surgery, major surgery should be associated with greater opioid use. The similarity suggests that a mechanism other than postoperative pain confers risk of chronic opioid use.

THINKING ABOUT OPIOIDS

Collectively, these articles describe elements of acute pain treatment that correlate with chronic ongoing opioid use: a higher cumulative dose,³ being seen by a physician who prescribes a lot of opioids,⁵ undergoing surgery,⁶ and psychiatric comorbidity.⁶ They made me wonder if opioid use for acute pain acts as an inoculation, analogous to inoculating a Petri dish with bacteria.  The likelihood of chronic opioid use arises from the inoculum dose, the host response, and the context of inoculation. 

These articles do not show how patients taking opioids chronically for pain become addicted. Stumbo et al⁸ interviewed 283 opioid-dependent patients and identified 5 pathways to opioid use disorder, 3 of which were related to pain control: inadequately controlled chronic pain, exposure to opioids during acute pain episodes, and chronic pain in patients who already had substance use disorders. Brat et al⁹ recently estimated the risk of opioid use disorder after receiving opioids postoperatively to be less than 1%, but it increased dramatically with duration of opioid treatment.

A patient who fills an opioid prescription does not necessarily have chronic pain. Nor do all patients with chronic pain require an opioid prescription. These studies did not establish whether the patients had a pain syndrome. In practice, we call our patients who chronically take opioids our “chronic pain patients.” But 40% of Americans have chronic pain, while only 5% take opioids daily for pain.^11,12

We assume that those taking opioids have the most severe pain. But Brummett et al suggested that continued opioid use is predicted less by pain and more by psychiatric comorbidity.⁶ More than half of the opioid prescriptions in the United States are written for patients with serious mental illness, who represent one-sixth of that population.¹¹ Maybe chronic opioid use for pain has more to do with vulnerability to opioids and less to do with a pain syndrome.

I now think about daily opioid use in much the same way as I think about daily prednisone use. Patients on daily prednisone have a characteristic set of medical risks from the prednisone itself, regardless of its indication. Yet we do not consider these patients addicted to prednisone. Opioid use may be similar.

Like most doctors, I am troubled by the continued rise in the opioid overdose rate.¹³ Yet addiction and death from overdose are not the only risks that patients on chronic opioids face; they also have higher rates of falls, cardiovascular death, pneumonia, death from chronic obstructive pulmonary disease, and motor vehicle crashes.^14–17 Patients on chronic opioids for pain have greater mental health comorbidity and worse function.¹⁸

Most concerning, chronic opioid treatment for pain lacks proof of benefit. In fact, a recent study disproved the benefit of opioids for chronic pain compared with nonopioid options.¹⁹ When I meet with patients who are taking chronic opioids for pain, I often can’t identify why the drugs were started or ought to be continued, and I anticipate a bad outcome. Yet the patient is afraid to stop the drug. For these reasons, chronic opioid use for pain strikes me as worth considering separately from opioid use disorder.

The studies described above have had a powerful effect on my clinical care as a hospitalist.

I now talk to all patients starting opioids about how hard it can be to stop. Some patients are defensive at first, believing this does not apply to them. But I politely continue.

People with depression and anxiety can have a harder time stopping opioids. Addiction is both a risk with ongoing opioid use and a possible outcome of acute opioid use.⁸ But one can struggle to stop opioids without being addicted or depressed. Even the healthiest person may wish to continue opioids past the point of benefit.

I am careful not to invalidate the patient’s experience of pain. It is challenging for patients to find the balance between current discomfort and a possible future adverse effect. In these conversations, I imagine how I would want a loved one counseled on their pain control. This centers me as I choose my words and my tone.

I now monitor the total amount of opioid I prescribe for acute pain in addition to the daily dose. I give my patients as few opioids as reasonable, and advise them to take the minimum dose required for tolerable comfort. I offer nonopioid options as the preferred choice, presenting them as effective and safe. I do this irrespective of the indication for opioids.

I limit opioids in all patients, not just those with comorbidities. I include in my shared decision-making process the risk of chronic opioid use when I prescribe opioids for acute pain, carefully distinguishing it from opioid use disorder. Instead of excess opioids, I give patients my office phone number to call in case they struggle. I rarely get calls. But I find patients would rather have access to a doctor than extra pills. And offering them my contact information lets me limit opioids while letting them know that I am committed to their comfort and health.

As an addiction medicine doctor, I consult on patients not taking their opioids as prescribed. Caring for these patients is intellectually and emotionally draining; they suffer daily, and the opioids they take provide a modicum of relief at a high cost. The publications I have discussed here provide insight into how a troubled relationship with opioids begins. I remind myself that these patients have an iatrogenic condition. Their behaviors that we label “aberrant” may reflect an adverse reaction to medications prescribed to them for acute pain.

Mary, my patient with postoperative pain after cholecystectomy, may over time develop opioid use disorder as Heather did. That progression may have begun with the hydrocodone I prescribed and the counseling I gave her, and it may proceed to chronic opioid use and then opioid use disorder.

I am looking closely at the care I give for acute pain in light of these innovative studies. But even more so, they have increased the compassion with which I care for patients like Heather, those harmed by prescribed opioids.

My urologic career began in the late 1980s, just before prostate-specific antigen (PSA) testing was introduced. Ever since, a busy prostate cancer practice has given me a frontline view of the benefits and possible harms of PSA screening.

See related article

In the pre-PSA era, about half of men with newly diagnosed prostate cancer presented with incurable disease, either locally advanced or metastatic. The most common treatment was bilateral orchiectomy, which was the only safe form of androgen deprivation available.

Fast-forward a few years to the mid-1990s. Within 5 years after the introduction of PSA testing, the rate of incurable disease at diagnosis fell to just 5%, and treatment for localized disease skyrocketed, including radical prostatectomy, external beam radiation, and brachytherapy. As a result of earlier diagnosis and improved treatments, the death rate from prostate cancer in US men has fallen more than 30% since 1990.

The first-hand experience of seeing this massive stage migration to curable disease has forever convinced me that PSA screening is beneficial. Robust statistical models lend credence to this belief, with estimates that screening is responsible for 45% to 70% of this decline in mortality.¹

Fast-forward again to 2012, when the US Preventive Services Task Force (USPSTF) published a strong recommendation against screening. The recommendation had so much force that as recently as 2014, only 11% of men at highest risk of prostate cancer in the Cleveland Clinic system were screened for it,² mirroring national trends.

What happened? Colored by the experience in the era before PSA, when men presented frequently with painful metastatic disease and had an average life expectancy of 18 to 24 months, it was widely believed that all detected prostate cancer required treatment. What was not appreciated was that while PSA detects lots of prostate cancer, the most common reason for PSA levels to reach a range worrisome enough to trigger biopsy was actually benign prostatic hypertrophy.

The resulting increase in the number of biopsies resulted in the detection of a substantial number of low-grade cancers that were never destined to cause clinical harm but that got treated anyway, based on the fear that all cancers had metastatic potential. The USPSTF based its recommendation against screening on the harms caused by this overdetection and overtreatment of nonlethal disease, focusing on risks of biopsy such as sepsis, and on treatment-related adverse effects such as changes in urinary, bowel, and sexual function.

RANDOMIZED TRIALS SHOW A BENEFIT FROM SCREENING

As a result of this controversy, several large randomized trials designed to test whether PSA screening was beneficial were organized and begun in the 1990s, with one in the United States and another in Europe.^3,4 Mature data from both trials have now established that there is indeed benefit to population-level screening.

The US Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO), was initially reported to show no difference in prostate cancer-specific mortality rates in those screened vs not screened, but because more than 90% of the men in the no-screening arm were screened anyway, that conclusion is erroneous.³

With 13-year follow-up and far less PSA contamination in the unscreened arm, the European Randomized Study of Screening for Prostate Cancer (ERSPC) in men ages 55 to 69 demonstrated a 27% reduction in the rate of death and a 35% reduction in the need for palliative treatments (androgen deprivation or radiation, or both) for metastatic disease in those screened vs not screened, clearly establishing substantial clinical benefit to PSA screening.⁴

A recent analysis of both trials that controlled for PSA drop-ins (comparing those actually screened with those actually not screened) concluded that the benefit of screening in terms of mortality reduction (estimated at about 30%) are equal in both trials.⁵ A large cohort study from Kaiser Permanente with 16-year follow-up has suggested that PSA screening has both a prostate cancer-specific benefit and an overall mortality benefit.⁶

In parallel with the design and completion of these trials, there was a significant effort to better identify and manage patients initially overdiagnosed with nonlethal cancers by developing active surveillance regimens.

This management strategy recognizes that most low-grade cancers pose no short-term risk to the patient’s health or longevity, that definitive therapy can be deferred, and that with regular monitoring by digital rectal examination, PSA measurement, and repeat biopsy, cancers that progress can still be cured. The result of this strategy is a marked reduction in the harms caused by overtreatment (ie, the aforementioned adverse effects), as well as the avoidance of unnecessary treatment in many patients.

A randomized trial and 2 large prospective cohort studies have confirmed the long-term safety of this approach,^7–9 and the development of commercially available, biopsy-based gene expression profiling tools promises to further improve risk stratification at diagnosis and during follow-up for individual patients.¹⁰

NEW USPSTF RECOMMENDATIONS: AN INDIVIDUAL, INFORMED DECISION

Based on the results of the ERSPC and the widespread adoption and safety of active surveillance, which together show benefit to screening and fewer harms in overdetection and overtreatment, in 2018 the USPSTF recast its recommendations. In upgrading the recommendation from “D” to “C,” the recommendation now states that for men ages 55 to 69, PSA screening should be an individual decision after a discussion with an informed provider, although men over 70 are still advised not to undergo screening at all.¹¹

Some may think that this recommendation has arrived just in time, or that it should be  made even stronger to actually recommend screening, as recent data from 2 national registries—the Surveillance, Epidemiology, and End Results program and the National Cancer Database—show that the fall in screening after the 2012 USPSTF guidelines has resulted in an increase in men presenting with advanced stage disease.^12,13 (All of you Back to the Future fans, please return to the mid to late 1980s to see how that plays out.)

So the pendulum has now swung back in favor of screening, largely supported by solid data showing meaningful clinical benefit, better understanding of PSA and prostate cancer biology, and adoption of active surveillance.

AN IDEAL SCREENING PROGRAM

An ideal screening program would detect only biologically significant cancers, thus eliminating overdetection and overtreatment. There is reason for optimism on this front.

Second-generation PSA tests have better diagnostic accuracy for high-grade disease than earlier tests. Two such tests, the Prostate Health Index (Beckman Coulter) and the 4K-score (Opko Health), are commercially available though not usually covered by commercial insurers.¹⁴ A third test, IsoPSA (Cleveland Diagnostics), is under development. Most hospital laboratories will be able to be run this test with no need for a central laboratory.¹⁵ All 3 tests have been shown to reduce unnecessary biopsies (because of a low probability of finding a biologically significant cancer) by 30% to 45% and will help reduce overdetection.

Moreover, multiparametric magnetic resonance imaging of the prostate has been shown to improve detection of high-grade cancers,¹⁶ and a randomized trial has suggested that its incorporation into a screening strategy is cost-effective and could be better than PSA testing plus transrectal ultrasonography alone (the current standard of care).¹⁷

Several risk scores based on germline genomics also hold promise for better identifying those at risk and for helping to de-intensify screening for those unlikely to have high-grade cancer.¹⁸

Screening for prostate cancer reduces mortality rates and the burden of metastatic disease, and the paradigm continues to evolve. Men at risk by virtue of age (55 to 69, and healthy men > 70), family history, race, and newly identified factors (germline genetics) all deserve an informed discussion on the benefits and risks of screening

My urologic career began in the late 1980s, just before prostate-specific antigen (PSA) testing was introduced. Ever since, a busy prostate cancer practice has given me a frontline view of the benefits and possible harms of PSA screening.

See related article

In the pre-PSA era, about half of men with newly diagnosed prostate cancer presented with incurable disease, either locally advanced or metastatic. The most common treatment was bilateral orchiectomy, which was the only safe form of androgen deprivation available.

Fast-forward a few years to the mid-1990s. Within 5 years after the introduction of PSA testing, the rate of incurable disease at diagnosis fell to just 5%, and treatment for localized disease skyrocketed, including radical prostatectomy, external beam radiation, and brachytherapy. As a result of earlier diagnosis and improved treatments, the death rate from prostate cancer in US men has fallen more than 30% since 1990.

The first-hand experience of seeing this massive stage migration to curable disease has forever convinced me that PSA screening is beneficial. Robust statistical models lend credence to this belief, with estimates that screening is responsible for 45% to 70% of this decline in mortality.¹

Fast-forward again to 2012, when the US Preventive Services Task Force (USPSTF) published a strong recommendation against screening. The recommendation had so much force that as recently as 2014, only 11% of men at highest risk of prostate cancer in the Cleveland Clinic system were screened for it,² mirroring national trends.

What happened? Colored by the experience in the era before PSA, when men presented frequently with painful metastatic disease and had an average life expectancy of 18 to 24 months, it was widely believed that all detected prostate cancer required treatment. What was not appreciated was that while PSA detects lots of prostate cancer, the most common reason for PSA levels to reach a range worrisome enough to trigger biopsy was actually benign prostatic hypertrophy.

The resulting increase in the number of biopsies resulted in the detection of a substantial number of low-grade cancers that were never destined to cause clinical harm but that got treated anyway, based on the fear that all cancers had metastatic potential. The USPSTF based its recommendation against screening on the harms caused by this overdetection and overtreatment of nonlethal disease, focusing on risks of biopsy such as sepsis, and on treatment-related adverse effects such as changes in urinary, bowel, and sexual function.

RANDOMIZED TRIALS SHOW A BENEFIT FROM SCREENING

As a result of this controversy, several large randomized trials designed to test whether PSA screening was beneficial were organized and begun in the 1990s, with one in the United States and another in Europe.^3,4 Mature data from both trials have now established that there is indeed benefit to population-level screening.

The US Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO), was initially reported to show no difference in prostate cancer-specific mortality rates in those screened vs not screened, but because more than 90% of the men in the no-screening arm were screened anyway, that conclusion is erroneous.³

With 13-year follow-up and far less PSA contamination in the unscreened arm, the European Randomized Study of Screening for Prostate Cancer (ERSPC) in men ages 55 to 69 demonstrated a 27% reduction in the rate of death and a 35% reduction in the need for palliative treatments (androgen deprivation or radiation, or both) for metastatic disease in those screened vs not screened, clearly establishing substantial clinical benefit to PSA screening.⁴

A recent analysis of both trials that controlled for PSA drop-ins (comparing those actually screened with those actually not screened) concluded that the benefit of screening in terms of mortality reduction (estimated at about 30%) are equal in both trials.⁵ A large cohort study from Kaiser Permanente with 16-year follow-up has suggested that PSA screening has both a prostate cancer-specific benefit and an overall mortality benefit.⁶

In parallel with the design and completion of these trials, there was a significant effort to better identify and manage patients initially overdiagnosed with nonlethal cancers by developing active surveillance regimens.

This management strategy recognizes that most low-grade cancers pose no short-term risk to the patient’s health or longevity, that definitive therapy can be deferred, and that with regular monitoring by digital rectal examination, PSA measurement, and repeat biopsy, cancers that progress can still be cured. The result of this strategy is a marked reduction in the harms caused by overtreatment (ie, the aforementioned adverse effects), as well as the avoidance of unnecessary treatment in many patients.

A randomized trial and 2 large prospective cohort studies have confirmed the long-term safety of this approach,^7–9 and the development of commercially available, biopsy-based gene expression profiling tools promises to further improve risk stratification at diagnosis and during follow-up for individual patients.¹⁰

NEW USPSTF RECOMMENDATIONS: AN INDIVIDUAL, INFORMED DECISION

Based on the results of the ERSPC and the widespread adoption and safety of active surveillance, which together show benefit to screening and fewer harms in overdetection and overtreatment, in 2018 the USPSTF recast its recommendations. In upgrading the recommendation from “D” to “C,” the recommendation now states that for men ages 55 to 69, PSA screening should be an individual decision after a discussion with an informed provider, although men over 70 are still advised not to undergo screening at all.¹¹

Some may think that this recommendation has arrived just in time, or that it should be  made even stronger to actually recommend screening, as recent data from 2 national registries—the Surveillance, Epidemiology, and End Results program and the National Cancer Database—show that the fall in screening after the 2012 USPSTF guidelines has resulted in an increase in men presenting with advanced stage disease.^12,13 (All of you Back to the Future fans, please return to the mid to late 1980s to see how that plays out.)

So the pendulum has now swung back in favor of screening, largely supported by solid data showing meaningful clinical benefit, better understanding of PSA and prostate cancer biology, and adoption of active surveillance.

AN IDEAL SCREENING PROGRAM

An ideal screening program would detect only biologically significant cancers, thus eliminating overdetection and overtreatment. There is reason for optimism on this front.

Second-generation PSA tests have better diagnostic accuracy for high-grade disease than earlier tests. Two such tests, the Prostate Health Index (Beckman Coulter) and the 4K-score (Opko Health), are commercially available though not usually covered by commercial insurers.¹⁴ A third test, IsoPSA (Cleveland Diagnostics), is under development. Most hospital laboratories will be able to be run this test with no need for a central laboratory.¹⁵ All 3 tests have been shown to reduce unnecessary biopsies (because of a low probability of finding a biologically significant cancer) by 30% to 45% and will help reduce overdetection.

Moreover, multiparametric magnetic resonance imaging of the prostate has been shown to improve detection of high-grade cancers,¹⁶ and a randomized trial has suggested that its incorporation into a screening strategy is cost-effective and could be better than PSA testing plus transrectal ultrasonography alone (the current standard of care).¹⁷

Several risk scores based on germline genomics also hold promise for better identifying those at risk and for helping to de-intensify screening for those unlikely to have high-grade cancer.¹⁸

Screening for prostate cancer reduces mortality rates and the burden of metastatic disease, and the paradigm continues to evolve. Men at risk by virtue of age (55 to 69, and healthy men > 70), family history, race, and newly identified factors (germline genetics) all deserve an informed discussion on the benefits and risks of screening

My urologic career began in the late 1980s, just before prostate-specific antigen (PSA) testing was introduced. Ever since, a busy prostate cancer practice has given me a frontline view of the benefits and possible harms of PSA screening.

See related article

In the pre-PSA era, about half of men with newly diagnosed prostate cancer presented with incurable disease, either locally advanced or metastatic. The most common treatment was bilateral orchiectomy, which was the only safe form of androgen deprivation available.

Fast-forward a few years to the mid-1990s. Within 5 years after the introduction of PSA testing, the rate of incurable disease at diagnosis fell to just 5%, and treatment for localized disease skyrocketed, including radical prostatectomy, external beam radiation, and brachytherapy. As a result of earlier diagnosis and improved treatments, the death rate from prostate cancer in US men has fallen more than 30% since 1990.

The first-hand experience of seeing this massive stage migration to curable disease has forever convinced me that PSA screening is beneficial. Robust statistical models lend credence to this belief, with estimates that screening is responsible for 45% to 70% of this decline in mortality.¹

Fast-forward again to 2012, when the US Preventive Services Task Force (USPSTF) published a strong recommendation against screening. The recommendation had so much force that as recently as 2014, only 11% of men at highest risk of prostate cancer in the Cleveland Clinic system were screened for it,² mirroring national trends.

What happened? Colored by the experience in the era before PSA, when men presented frequently with painful metastatic disease and had an average life expectancy of 18 to 24 months, it was widely believed that all detected prostate cancer required treatment. What was not appreciated was that while PSA detects lots of prostate cancer, the most common reason for PSA levels to reach a range worrisome enough to trigger biopsy was actually benign prostatic hypertrophy.

The resulting increase in the number of biopsies resulted in the detection of a substantial number of low-grade cancers that were never destined to cause clinical harm but that got treated anyway, based on the fear that all cancers had metastatic potential. The USPSTF based its recommendation against screening on the harms caused by this overdetection and overtreatment of nonlethal disease, focusing on risks of biopsy such as sepsis, and on treatment-related adverse effects such as changes in urinary, bowel, and sexual function.

RANDOMIZED TRIALS SHOW A BENEFIT FROM SCREENING

As a result of this controversy, several large randomized trials designed to test whether PSA screening was beneficial were organized and begun in the 1990s, with one in the United States and another in Europe.^3,4 Mature data from both trials have now established that there is indeed benefit to population-level screening.

The US Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO), was initially reported to show no difference in prostate cancer-specific mortality rates in those screened vs not screened, but because more than 90% of the men in the no-screening arm were screened anyway, that conclusion is erroneous.³

With 13-year follow-up and far less PSA contamination in the unscreened arm, the European Randomized Study of Screening for Prostate Cancer (ERSPC) in men ages 55 to 69 demonstrated a 27% reduction in the rate of death and a 35% reduction in the need for palliative treatments (androgen deprivation or radiation, or both) for metastatic disease in those screened vs not screened, clearly establishing substantial clinical benefit to PSA screening.⁴

A recent analysis of both trials that controlled for PSA drop-ins (comparing those actually screened with those actually not screened) concluded that the benefit of screening in terms of mortality reduction (estimated at about 30%) are equal in both trials.⁵ A large cohort study from Kaiser Permanente with 16-year follow-up has suggested that PSA screening has both a prostate cancer-specific benefit and an overall mortality benefit.⁶

In parallel with the design and completion of these trials, there was a significant effort to better identify and manage patients initially overdiagnosed with nonlethal cancers by developing active surveillance regimens.

This management strategy recognizes that most low-grade cancers pose no short-term risk to the patient’s health or longevity, that definitive therapy can be deferred, and that with regular monitoring by digital rectal examination, PSA measurement, and repeat biopsy, cancers that progress can still be cured. The result of this strategy is a marked reduction in the harms caused by overtreatment (ie, the aforementioned adverse effects), as well as the avoidance of unnecessary treatment in many patients.

A randomized trial and 2 large prospective cohort studies have confirmed the long-term safety of this approach,^7–9 and the development of commercially available, biopsy-based gene expression profiling tools promises to further improve risk stratification at diagnosis and during follow-up for individual patients.¹⁰

NEW USPSTF RECOMMENDATIONS: AN INDIVIDUAL, INFORMED DECISION

Based on the results of the ERSPC and the widespread adoption and safety of active surveillance, which together show benefit to screening and fewer harms in overdetection and overtreatment, in 2018 the USPSTF recast its recommendations. In upgrading the recommendation from “D” to “C,” the recommendation now states that for men ages 55 to 69, PSA screening should be an individual decision after a discussion with an informed provider, although men over 70 are still advised not to undergo screening at all.¹¹

Some may think that this recommendation has arrived just in time, or that it should be  made even stronger to actually recommend screening, as recent data from 2 national registries—the Surveillance, Epidemiology, and End Results program and the National Cancer Database—show that the fall in screening after the 2012 USPSTF guidelines has resulted in an increase in men presenting with advanced stage disease.^12,13 (All of you Back to the Future fans, please return to the mid to late 1980s to see how that plays out.)

So the pendulum has now swung back in favor of screening, largely supported by solid data showing meaningful clinical benefit, better understanding of PSA and prostate cancer biology, and adoption of active surveillance.

AN IDEAL SCREENING PROGRAM

An ideal screening program would detect only biologically significant cancers, thus eliminating overdetection and overtreatment. There is reason for optimism on this front.

Second-generation PSA tests have better diagnostic accuracy for high-grade disease than earlier tests. Two such tests, the Prostate Health Index (Beckman Coulter) and the 4K-score (Opko Health), are commercially available though not usually covered by commercial insurers.¹⁴ A third test, IsoPSA (Cleveland Diagnostics), is under development. Most hospital laboratories will be able to be run this test with no need for a central laboratory.¹⁵ All 3 tests have been shown to reduce unnecessary biopsies (because of a low probability of finding a biologically significant cancer) by 30% to 45% and will help reduce overdetection.

Moreover, multiparametric magnetic resonance imaging of the prostate has been shown to improve detection of high-grade cancers,¹⁶ and a randomized trial has suggested that its incorporation into a screening strategy is cost-effective and could be better than PSA testing plus transrectal ultrasonography alone (the current standard of care).¹⁷

Several risk scores based on germline genomics also hold promise for better identifying those at risk and for helping to de-intensify screening for those unlikely to have high-grade cancer.¹⁸

Screening for prostate cancer reduces mortality rates and the burden of metastatic disease, and the paradigm continues to evolve. Men at risk by virtue of age (55 to 69, and healthy men > 70), family history, race, and newly identified factors (germline genetics) all deserve an informed discussion on the benefits and risks of screening

Taurine, also known as 2-aminoethanesulfonic acid, is a naturally occurring beta-amino acid (which has a sulphonic acid group instead of carboxylic acid, differentiating it from other amino acids) yielded by methionine and cysteine metabolism in the liver.^1,2 An important free beta-amino acid in mammals, it is often the free amino acid present in the greatest concentrations in several cell types in humans.^1,2 Dietary intake of taurine also plays an important role in maintaining the body’s taurine levels because of mammals’ limited ability to synthesize it.¹

olavs/Thinkstock

Notably in terms of dermatologic treatment options, the combination product taurine bromamine is known to impart antioxidant, anti-inflammatory, and antibacterial activities.³ And taurine itself is associated with antioxidant, anti-inflammatory, antifibrotic, and immunomodulatory characteristics,^1,4 and is noted for conferring antiaging benefits.⁵

Acne and other inflammatory conditions

The use of topical taurine bromamine, the physiological product of hypobromous acid and taurine, is one of the new emerging approaches to treating acne.^6,7

In response to the problem of evolving antibiotic resistance, Marcinkiewicz reported in 2009 on the then-new therapeutic option of topical taurine bromamine for the treatment of inflammatory skin disorders such as acne. The author pointed out that Propionibacterium acnes is particularly sensitive to taurine bromamine, with the substance now known to suppress H₂O₂ production by activated neutrophils, likely contributing to moderating the severity and lowering the number of inflammatory acne lesions. In a 6-week double-blind pilot clinical study, Marcinkiewicz and his team compared the efficacy of 0.5% taurine bromamine cream with 1% clindamycin gel in 40 patients with mild to moderate acne. Treatments, which were randomly assigned, occurred twice daily through the study. Amelioration of acne symptoms was comparable in the two groups, with more than 90% of patients improving clinically and experiencing similar decreases in acne lesions (65% in the taurine bromamine group and 68% in the clindamycin group). Marcinkiewicz concluded that these results indicate the viability of taurine bromamine as an option for inflammatory acne therapy, particularly for patients who have shown antibiotic resistance.³

Wide-ranging protection potential

In 2003, Janeke et al. conducted analyses that showed that taurine accumulation defended cultured human keratinocytes from osmotically- and UV-induced apoptosis, suggesting the importance of taurine as an epidermal osmolyte necessary for maintaining keratinocyte hydration in a dry environment.²

Three years later, Collin et al. demonstrated the dynamic protective effects of taurine on the human hair follicle in an in vitro study in which taurine promoted hair survival and protected against TGF-beta1-induced damage.¹

Taurine has also been found to stabilize and protect the catalytic activity of the hemoprotein cytochrome P450 3A4, which is a key enzyme responsible for metabolizing various endogenous as well as foreign substances, including drugs.⁸
 

Penetration enhancement

In 2016, Mueller et al. studied the effects of urea and taurine as hydrophilic penetration enhancers on stratum corneum lipid models as both substances are known to exert such effects. With inconclusive results as to the roots of such activity, they speculated that both entities enhance penetration through the introduction of copious water into the corneocytes, resulting from the robust water-binding capacity of urea and the consequent osmotic pressure related to taurine.⁹

Possible skin whitening and anti-aging roles and other promising lab results

Based on their previous work demonstrating that azelaic acid, a saturated dicarboxylic acid found naturally in wheat, rye, and barley, suppressed melanogenesis, Yu and Kim investigated the antimelanogenic activity of azelaic acid and taurine in B16F10 mouse melanoma cells in 2010. They found that the combination of the two substances exhibited a greater inhibitory effect in melanocytes than azelaic acid alone, with melanin production and tyrosinase activity suppressed without inducing cytotoxicity. The investigators concluded the combination of azelaic acid and taurine may be an effective approach for treating hyperpigmentation.¹⁰

In 2015, Ito et al. investigated the possible anti-aging role of taurine using a taurine transporter knockout mouse model. They noted that aging-related disorders affecting the skin, heart, skeletal muscle, and liver and resulting in a shorter lifespan have been correlated with tissue taurine depletion. The researchers proposed that proper protein folding allows endogenous taurine to perform as an antiaging molecule.⁵

Also in 2015, Kim et al. investigated potential mechanisms of the antiproliferative activity of taurine on murine B16F10 melanoma cells via the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) and neutral red assays and microscopic analysis. They found that taurine prevented cell proliferation and engendered apoptosis in B16F10 cells, concluding that taurine may have a role to play as a chemotherapeutic agent for skin cancer.¹¹

In 2014, Ashkani-Esfahani et al. studied the impact of taurine on cutaneous leishmaniasis wounds in a mouse model. Investigators induced 18 mice with wounds using L. major promastigotes, and divided them into a taurine injection group, taurine gel group, and no treatment group, performing treatments every 24 hours over 21 days. The taurine treatment groups exhibited significantly greater numerical fibroblast density, collagen bundle volume density, and vessel length densities compared with the nontreatment group. The taurine injection group displayed higher fibroblast numerical density than did the taurine gel group. The researchers concluded that taurine has the capacity to enhance wound healing and tissue regeneration but showed no direct anti-leishmaniasis effect.⁴

Conclusion

Taurine has been found over the last few decades to impart salutary effects for human health. This beta-amino acid that occurs naturally in humans and other mammals also appears to hold promising potential in the dermatologic realm, particularly for its anti-inflammatory and antioxidant effects. More research is needed to ascertain just how pivotal this compound can be for skin health.

Dr. Baumann is a private practice dermatologist, researcher, author and entrepreneur who practices in Miami. She founded the Cosmetic Dermatology Center at the University of Miami in 1997. Dr. Baumann wrote two textbooks: “Cosmetic Dermatology: Principles and Practice” (New York: McGraw-Hill, 2002), and “Cosmeceuticals and Cosmetic Ingredients,” (New York: McGraw-Hill, 2014), and a New York Times Best Sellers book for consumers, “The Skin Type Solution” (New York: Bantam Dell, 2006). Dr. Baumann has received funding for advisory boards and/or clinical research trials from Allergan, Evolus, Galderma, and Revance. She is the founder and CEO of Skin Type Solutions Franchise Systems LLC. Write to her at [email protected].

References

1. Int J Cosmet Sci. 2006 Aug;28(4):289-98.

2. J Invest Dermatol. 2003 Aug;121(2):354-61.

3. Pol Arch Med Wewn. 2009 Oct;119(10):673-6.

4. Adv Biomed Res. 2014 Oct 7;3:204.

5. Adv Exp Med Biol. 2015;803:481-7.

6. Am J Clin Dermatol. 2012 Dec 1;13(6):357-64.

7. Eur J Dermatol. 2008 Jul-Aug;18(4):433-9.

8. Biochemistry (Mosc). 2015 Mar;80(3):366-73.

9. Biochim Biophys Acta. 2016 Sep;1858(9):2006-18.

10. J Biomed Sci. 2010 Aug 24;17 Suppl 1:S45.

11. Adv Exp Med Biol. 2015;803:167-77.

Taurine, also known as 2-aminoethanesulfonic acid, is a naturally occurring beta-amino acid (which has a sulphonic acid group instead of carboxylic acid, differentiating it from other amino acids) yielded by methionine and cysteine metabolism in the liver.^1,2 An important free beta-amino acid in mammals, it is often the free amino acid present in the greatest concentrations in several cell types in humans.^1,2 Dietary intake of taurine also plays an important role in maintaining the body’s taurine levels because of mammals’ limited ability to synthesize it.¹

olavs/Thinkstock

Notably in terms of dermatologic treatment options, the combination product taurine bromamine is known to impart antioxidant, anti-inflammatory, and antibacterial activities.³ And taurine itself is associated with antioxidant, anti-inflammatory, antifibrotic, and immunomodulatory characteristics,^1,4 and is noted for conferring antiaging benefits.⁵

Acne and other inflammatory conditions

The use of topical taurine bromamine, the physiological product of hypobromous acid and taurine, is one of the new emerging approaches to treating acne.^6,7

In response to the problem of evolving antibiotic resistance, Marcinkiewicz reported in 2009 on the then-new therapeutic option of topical taurine bromamine for the treatment of inflammatory skin disorders such as acne. The author pointed out that Propionibacterium acnes is particularly sensitive to taurine bromamine, with the substance now known to suppress H₂O₂ production by activated neutrophils, likely contributing to moderating the severity and lowering the number of inflammatory acne lesions. In a 6-week double-blind pilot clinical study, Marcinkiewicz and his team compared the efficacy of 0.5% taurine bromamine cream with 1% clindamycin gel in 40 patients with mild to moderate acne. Treatments, which were randomly assigned, occurred twice daily through the study. Amelioration of acne symptoms was comparable in the two groups, with more than 90% of patients improving clinically and experiencing similar decreases in acne lesions (65% in the taurine bromamine group and 68% in the clindamycin group). Marcinkiewicz concluded that these results indicate the viability of taurine bromamine as an option for inflammatory acne therapy, particularly for patients who have shown antibiotic resistance.³

Wide-ranging protection potential

In 2003, Janeke et al. conducted analyses that showed that taurine accumulation defended cultured human keratinocytes from osmotically- and UV-induced apoptosis, suggesting the importance of taurine as an epidermal osmolyte necessary for maintaining keratinocyte hydration in a dry environment.²

Three years later, Collin et al. demonstrated the dynamic protective effects of taurine on the human hair follicle in an in vitro study in which taurine promoted hair survival and protected against TGF-beta1-induced damage.¹

Taurine has also been found to stabilize and protect the catalytic activity of the hemoprotein cytochrome P450 3A4, which is a key enzyme responsible for metabolizing various endogenous as well as foreign substances, including drugs.⁸
 

Penetration enhancement

In 2016, Mueller et al. studied the effects of urea and taurine as hydrophilic penetration enhancers on stratum corneum lipid models as both substances are known to exert such effects. With inconclusive results as to the roots of such activity, they speculated that both entities enhance penetration through the introduction of copious water into the corneocytes, resulting from the robust water-binding capacity of urea and the consequent osmotic pressure related to taurine.⁹

Possible skin whitening and anti-aging roles and other promising lab results

Based on their previous work demonstrating that azelaic acid, a saturated dicarboxylic acid found naturally in wheat, rye, and barley, suppressed melanogenesis, Yu and Kim investigated the antimelanogenic activity of azelaic acid and taurine in B16F10 mouse melanoma cells in 2010. They found that the combination of the two substances exhibited a greater inhibitory effect in melanocytes than azelaic acid alone, with melanin production and tyrosinase activity suppressed without inducing cytotoxicity. The investigators concluded the combination of azelaic acid and taurine may be an effective approach for treating hyperpigmentation.¹⁰

In 2015, Ito et al. investigated the possible anti-aging role of taurine using a taurine transporter knockout mouse model. They noted that aging-related disorders affecting the skin, heart, skeletal muscle, and liver and resulting in a shorter lifespan have been correlated with tissue taurine depletion. The researchers proposed that proper protein folding allows endogenous taurine to perform as an antiaging molecule.⁵

Also in 2015, Kim et al. investigated potential mechanisms of the antiproliferative activity of taurine on murine B16F10 melanoma cells via the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) and neutral red assays and microscopic analysis. They found that taurine prevented cell proliferation and engendered apoptosis in B16F10 cells, concluding that taurine may have a role to play as a chemotherapeutic agent for skin cancer.¹¹

In 2014, Ashkani-Esfahani et al. studied the impact of taurine on cutaneous leishmaniasis wounds in a mouse model. Investigators induced 18 mice with wounds using L. major promastigotes, and divided them into a taurine injection group, taurine gel group, and no treatment group, performing treatments every 24 hours over 21 days. The taurine treatment groups exhibited significantly greater numerical fibroblast density, collagen bundle volume density, and vessel length densities compared with the nontreatment group. The taurine injection group displayed higher fibroblast numerical density than did the taurine gel group. The researchers concluded that taurine has the capacity to enhance wound healing and tissue regeneration but showed no direct anti-leishmaniasis effect.⁴

Conclusion

Taurine has been found over the last few decades to impart salutary effects for human health. This beta-amino acid that occurs naturally in humans and other mammals also appears to hold promising potential in the dermatologic realm, particularly for its anti-inflammatory and antioxidant effects. More research is needed to ascertain just how pivotal this compound can be for skin health.

Dr. Baumann is a private practice dermatologist, researcher, author and entrepreneur who practices in Miami. She founded the Cosmetic Dermatology Center at the University of Miami in 1997. Dr. Baumann wrote two textbooks: “Cosmetic Dermatology: Principles and Practice” (New York: McGraw-Hill, 2002), and “Cosmeceuticals and Cosmetic Ingredients,” (New York: McGraw-Hill, 2014), and a New York Times Best Sellers book for consumers, “The Skin Type Solution” (New York: Bantam Dell, 2006). Dr. Baumann has received funding for advisory boards and/or clinical research trials from Allergan, Evolus, Galderma, and Revance. She is the founder and CEO of Skin Type Solutions Franchise Systems LLC. Write to her at [email protected].

References

1. Int J Cosmet Sci. 2006 Aug;28(4):289-98.

2. J Invest Dermatol. 2003 Aug;121(2):354-61.

3. Pol Arch Med Wewn. 2009 Oct;119(10):673-6.

4. Adv Biomed Res. 2014 Oct 7;3:204.

5. Adv Exp Med Biol. 2015;803:481-7.

6. Am J Clin Dermatol. 2012 Dec 1;13(6):357-64.

7. Eur J Dermatol. 2008 Jul-Aug;18(4):433-9.

8. Biochemistry (Mosc). 2015 Mar;80(3):366-73.

9. Biochim Biophys Acta. 2016 Sep;1858(9):2006-18.

10. J Biomed Sci. 2010 Aug 24;17 Suppl 1:S45.

11. Adv Exp Med Biol. 2015;803:167-77.

Taurine, also known as 2-aminoethanesulfonic acid, is a naturally occurring beta-amino acid (which has a sulphonic acid group instead of carboxylic acid, differentiating it from other amino acids) yielded by methionine and cysteine metabolism in the liver.^1,2 An important free beta-amino acid in mammals, it is often the free amino acid present in the greatest concentrations in several cell types in humans.^1,2 Dietary intake of taurine also plays an important role in maintaining the body’s taurine levels because of mammals’ limited ability to synthesize it.¹

olavs/Thinkstock

Notably in terms of dermatologic treatment options, the combination product taurine bromamine is known to impart antioxidant, anti-inflammatory, and antibacterial activities.³ And taurine itself is associated with antioxidant, anti-inflammatory, antifibrotic, and immunomodulatory characteristics,^1,4 and is noted for conferring antiaging benefits.⁵

Acne and other inflammatory conditions

The use of topical taurine bromamine, the physiological product of hypobromous acid and taurine, is one of the new emerging approaches to treating acne.^6,7

In response to the problem of evolving antibiotic resistance, Marcinkiewicz reported in 2009 on the then-new therapeutic option of topical taurine bromamine for the treatment of inflammatory skin disorders such as acne. The author pointed out that Propionibacterium acnes is particularly sensitive to taurine bromamine, with the substance now known to suppress H₂O₂ production by activated neutrophils, likely contributing to moderating the severity and lowering the number of inflammatory acne lesions. In a 6-week double-blind pilot clinical study, Marcinkiewicz and his team compared the efficacy of 0.5% taurine bromamine cream with 1% clindamycin gel in 40 patients with mild to moderate acne. Treatments, which were randomly assigned, occurred twice daily through the study. Amelioration of acne symptoms was comparable in the two groups, with more than 90% of patients improving clinically and experiencing similar decreases in acne lesions (65% in the taurine bromamine group and 68% in the clindamycin group). Marcinkiewicz concluded that these results indicate the viability of taurine bromamine as an option for inflammatory acne therapy, particularly for patients who have shown antibiotic resistance.³

Wide-ranging protection potential

In 2003, Janeke et al. conducted analyses that showed that taurine accumulation defended cultured human keratinocytes from osmotically- and UV-induced apoptosis, suggesting the importance of taurine as an epidermal osmolyte necessary for maintaining keratinocyte hydration in a dry environment.²

Three years later, Collin et al. demonstrated the dynamic protective effects of taurine on the human hair follicle in an in vitro study in which taurine promoted hair survival and protected against TGF-beta1-induced damage.¹

Taurine has also been found to stabilize and protect the catalytic activity of the hemoprotein cytochrome P450 3A4, which is a key enzyme responsible for metabolizing various endogenous as well as foreign substances, including drugs.⁸
 

Penetration enhancement

In 2016, Mueller et al. studied the effects of urea and taurine as hydrophilic penetration enhancers on stratum corneum lipid models as both substances are known to exert such effects. With inconclusive results as to the roots of such activity, they speculated that both entities enhance penetration through the introduction of copious water into the corneocytes, resulting from the robust water-binding capacity of urea and the consequent osmotic pressure related to taurine.⁹

Possible skin whitening and anti-aging roles and other promising lab results

Based on their previous work demonstrating that azelaic acid, a saturated dicarboxylic acid found naturally in wheat, rye, and barley, suppressed melanogenesis, Yu and Kim investigated the antimelanogenic activity of azelaic acid and taurine in B16F10 mouse melanoma cells in 2010. They found that the combination of the two substances exhibited a greater inhibitory effect in melanocytes than azelaic acid alone, with melanin production and tyrosinase activity suppressed without inducing cytotoxicity. The investigators concluded the combination of azelaic acid and taurine may be an effective approach for treating hyperpigmentation.¹⁰

In 2015, Ito et al. investigated the possible anti-aging role of taurine using a taurine transporter knockout mouse model. They noted that aging-related disorders affecting the skin, heart, skeletal muscle, and liver and resulting in a shorter lifespan have been correlated with tissue taurine depletion. The researchers proposed that proper protein folding allows endogenous taurine to perform as an antiaging molecule.⁵

Also in 2015, Kim et al. investigated potential mechanisms of the antiproliferative activity of taurine on murine B16F10 melanoma cells via the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) and neutral red assays and microscopic analysis. They found that taurine prevented cell proliferation and engendered apoptosis in B16F10 cells, concluding that taurine may have a role to play as a chemotherapeutic agent for skin cancer.¹¹

In 2014, Ashkani-Esfahani et al. studied the impact of taurine on cutaneous leishmaniasis wounds in a mouse model. Investigators induced 18 mice with wounds using L. major promastigotes, and divided them into a taurine injection group, taurine gel group, and no treatment group, performing treatments every 24 hours over 21 days. The taurine treatment groups exhibited significantly greater numerical fibroblast density, collagen bundle volume density, and vessel length densities compared with the nontreatment group. The taurine injection group displayed higher fibroblast numerical density than did the taurine gel group. The researchers concluded that taurine has the capacity to enhance wound healing and tissue regeneration but showed no direct anti-leishmaniasis effect.⁴

Conclusion

Taurine has been found over the last few decades to impart salutary effects for human health. This beta-amino acid that occurs naturally in humans and other mammals also appears to hold promising potential in the dermatologic realm, particularly for its anti-inflammatory and antioxidant effects. More research is needed to ascertain just how pivotal this compound can be for skin health.

Dr. Baumann is a private practice dermatologist, researcher, author and entrepreneur who practices in Miami. She founded the Cosmetic Dermatology Center at the University of Miami in 1997. Dr. Baumann wrote two textbooks: “Cosmetic Dermatology: Principles and Practice” (New York: McGraw-Hill, 2002), and “Cosmeceuticals and Cosmetic Ingredients,” (New York: McGraw-Hill, 2014), and a New York Times Best Sellers book for consumers, “The Skin Type Solution” (New York: Bantam Dell, 2006). Dr. Baumann has received funding for advisory boards and/or clinical research trials from Allergan, Evolus, Galderma, and Revance. She is the founder and CEO of Skin Type Solutions Franchise Systems LLC. Write to her at [email protected].

References

1. Int J Cosmet Sci. 2006 Aug;28(4):289-98.

2. J Invest Dermatol. 2003 Aug;121(2):354-61.

3. Pol Arch Med Wewn. 2009 Oct;119(10):673-6.

4. Adv Biomed Res. 2014 Oct 7;3:204.

5. Adv Exp Med Biol. 2015;803:481-7.

6. Am J Clin Dermatol. 2012 Dec 1;13(6):357-64.

7. Eur J Dermatol. 2008 Jul-Aug;18(4):433-9.

8. Biochemistry (Mosc). 2015 Mar;80(3):366-73.

9. Biochim Biophys Acta. 2016 Sep;1858(9):2006-18.

10. J Biomed Sci. 2010 Aug 24;17 Suppl 1:S45.

11. Adv Exp Med Biol. 2015;803:167-77.