Curr Osteoporos Rep (2014) 12:65–73 DOI 10.1007/s11914-014-0198-z
EPIDEMIOLOGY AND PATHOPHYSIOLOGY (PR EBELING AND EF ERIKSEN, SECTION EDITORS)
Review: Epidemiology and Pathophysiology of Atypical Femur Fractures Alvin C. Ng & Meng Ai Png & David T. Chua & Joyce S. B. Koh & Tet Sen Howe
Published online: 14 February 2014 # Springer Science+Business Media New York 2014
Abstract The recent recognition of the clinical phenomenon of atypical femoral fractures has garnered significant scientific interest. In this review, we will discuss and summarize the salient developments in the current understanding of the epidemiology, pathophysiology, and radiology of atypical femoral fractures. Keywords Atypical femoral fractures . Stress . Insufficiency . Subtrochanteric . Femoral . Diaphyseal osteoporosis . Bisphosphonates . Epidemiology . Pathophysiology . Radiology . Bone . Pain . Suppression of bone turnover
sufficiently rare that it is unlikely any randomized prospective clinical trial will ever detect them. Thus, the onus is on the clinicians to remain vigilant and report sporadic cases so that eventually a clearer picture emerges in the diagnosis and treatment of these rare and baffling conditions. We review atypical femoral fractures (AFFs) in this article, fully cognizant of the fact that although these fractures were first reported in the literature in 2005, even today we have much to learn about them.
Epidemiology Introduction Bisphosphonates (BPs) are the most commonly used drug in the treatment of osteoporosis and have an excellent safety profile. Most early randomized prospectively clinical trials did not show any major adverse effects. However, with more widespread use, cases of osteonecrosis of the jaw and atypical femoral fractures have been increasingly reported. These are
A. C. Ng (*) Department of Endocrinology, Singapore General Hospital, Outram Road, Singapore 169608, Singapore e-mail: [email protected] M. A. Png Department of Radiology, Singapore General Hospital, Singapore, Singapore D. T. Chua Department of Orthopedic Surgery, Changi General Hospital, Singapore, Singapore J. S. B. Koh : T. S. Howe Department of Orthopedic Surgery, Singapore General Hospital, Singapore, Singapore
There is an evolutionary trend in epidemiological studies published for AFFs from its first recognition till the present time. A comprehensive summary of the early epidemiological studies contributing to our knowledge of AFF was published by Nieves and Cosman in 2010 [1]. They classified them into case reports, case series, and epidemiologic studies. These reports predate the first ASBMR Task Force Report of 2010 [2••]. Hence, they have largely been referenced in the Task Force’s Report and were contributory to the original case definition criteria.
Epidemiologic Studies Pre-Dating the 1st ASBMR Task Force Report Case Reports and Series The earliest case report was published as early as 1997 describing a subtrochanteric stress fracture of the femur following total knee arthroplasty [3]. The clinical entity remained hardly recognized till the mid-2000s as subsequent case reports and series only appeared between 2005 [4] and 2009 [5–7]. Odvina et al [8] published the first case series that
66
Curr Osteoporos Rep (2014) 12:65–73
included AFFs, in which five of the ten reported fracture cases were femoral. The subsequent series of femoral fractures by Goh et al [9] stressed the unusual location of this fracture in the subtrochanteric region. This region (defined as being from the lesser trochanter to the junction of the proximal and middle third of the femoral shaft) is notably resilient to traumatic injuries and is largely broken only in high energy trauma or pathological fracture from metastatic disease [10]. The predilection for this anatomic region was also demonstrated in a series by Neviaser et al where 50 of 70 low energy femoral shaft fractures were subtrochanteric [11] as well as other subsequent series [12, 13]. Features described in these case reports and series differentiating AFFs from the common low energy osteoporosisrelated hip fractures included & & & & & & & & &
Prodromal pain described as discomfort, weakness, or actual pain involving the thigh or lower limb for weeks or months preceding the fracture [3, 9, 14–17]; Use of another antiresorptive or steroid therapy, in addition to the bisphosphonate [5, 8, 9, 17, 18]; Lack of trauma precipitating a fracture [3, 9, 15, 17, 19, 20]; Bilaterality (either simultaneous or sequential) [4, 5, 9, 17, 19]; Transverse fractures [3, 9]; Cortical hypertrophy or thickness [16, 17]; Stress reaction on the affected and/or unaffected side [3, 4, 15, 17, 18]; Poor fracture healing [7, 8, 17]; and Normal or low bone mass but not osteoporosis in the hip region [5, 8, 9, 13].
These reports served to highlight unusual features observed in the clinical and radiological presentations of these fractures, thus, bringing the attention of the medical community to them as a distinct entity.
Early Epidemiologic Studies Before the definition of major and minor criteria for identifying AFFs by the ASBMR Task Force [2••], there were epidemiologic studies attempting to define potential associations of AFFs with BP therapy, and to explore the risks of these fractures compared with the risk of osteoporotic fractures. These studies have demonstrated somewhat contradictory results. This could be due to a lack of proper case definition criteria and also the use of different denominators. Almost all cohort studies predating the first ASBMR Task Force Report on atypical femur fractures [2••] failed to show any association between BP intake and AFFs. These early registry-based cohort studies from Denmark [21, 22], Taiwan [23],
Pennsylvania, and New Jersey [24] based their denominators on the presence or absence of BP intake. They could reliably identify BP and other concomitant drug therapy but lacked radiographic adjudication in differentiating the usual osteoporotic femur fractures from “nontypical fractures”. Subtrochanteric and diaphyseal femur fractures were only identified from disease coding of national hospital discharge information [21, 22] or national health insurance or utilization databases [23, 24], and there was no way to verify the exact radiological characteristics of these fractures. Because of heterogeneity of fracture configurations in the subtrochanteric and diaphyseal regions, incorporating all such fractures into a single arm for analysis would logically not be specific enough to demonstrate potential associations between BP intake and AFFs. However, AFFs are fractures with specific features forming part of the total pool of subtrochanteric/ diaphyseal femoral fractures. Hence, the potential usefulness of these studies lies in defining the upper risk limit for AFFs. Overall, all such studies show a far lower incidence of subtrochanteric/diaphyseal fractures compared with hip fractures. Even so, an interesting trend of a sustained incidence of subtrochanteric and diaphyseal fractures seemed evident against a decreasing trend of hip fractures in populations exposed to BP therapy between 1996 and 2006 [25]. This could be interpreted as a lack of protection against subtrochanteric/diaphyseal fractures by BPs [21] or indeed an actual increase in AFFs, which now offset the decline in typical osteoporotic diaphyseal femur fractures affected by antiresorptive therapy. The latter is plausible as shown by another study involving a large-scale nationwide survey of hip fractures and medication use in the elderly US population. Wang et al [26] found a decrease in age-adjusted rates for typical hip fractures by 31.6 % among women and 20.5 % among men from 1996 to 2007. This contrasted with an increase of 20.4 % in subtrochanteric fragility fractures among women but no such trend in men over the same period. This corresponded to an increasing trend in the national use of BPs over the same period. It is also notable that women were the predominant users of BPs and sustained the vast majority of subtrochanteric fractures. However, in contrast to using BP intake as the variable, studies using fractures with the specific radiological features of AFFs as inclusion criteria tended to show a different picture. Lenart and co-workers found a significantly higher intake of BP among this group compared with controls [27]. Similarly, Girgis et al found an atypical fracture pattern 96.7 % specific to patients receiving BPs, although other risk factors such as history of low-energy fracture, the use of glucocorticoid therapy for more than 6 months, active rheumatoid arthritis, and a level of serum 25-hydroxyvitamin D of less than 16 ng per milliliter were also identified [28]. Giusti et al identified a very low incidence of AFF fracture patterns (only
Curr Osteoporos Rep (2014) 12:65–73
ten of 63 low energy subtrochanteric/diaphyseal femur fractures were considered atypical) but a high odds ratio (OR, 17.0; 95 % CI, 2.6–113.3) in this group for BP and glucocorticoid use [29].
Epidemiologic Studies after the ASBMR Task Force Report The ASBMR Task Force first consensus report on atypical fractures published in November 2010 [2••] defined major and minor criteria for AFFs, thus refining the methodology for case finding in subsequent studies. Studies that followed were now largely cohort studies using fracture patterns as the primary variable. In a cohort analysis of 12,777 women 55 years of age or older with femur fractures, Schilcher et al isolated 59 AFFs out of 1271 subtrochanteric/shaft fractures [30]. The ageadjusted relative risk of atypical fracture was 47.3 (95 % confidence interval [CI], 25.6 to 87.3), far higher than the relative risk of 1.27 (95 % CI, 0.85–1.90) for any other type of fracture associated with BP use. The risk of an atypical fracture rose with an increasing duration of BP use, with an odds ratio of 1.3 (95 % CI, 1.1–1.6) per 100 prescribed daily doses. This risk was approximately ten times as high as a normal level of fracture risk within the first 2 years of use and 50 times as high thereafter. However, in view of the low absolute risk of 50 cases per 100,000 patients-years (95 % CI, 4–7), the authors argued that the benefits of fracture prevention with BP use in the presence of appropriate indications greatly outweighed the risk of AFFs. Other epidemiologic studies with similar methodologies also yielded similar findings. Thompson et al from the United Kingdom [31], Feldstein et al [32], and Lo et al [33] from the Kaiser Permanente Northwest region of California, and Dell et al [34] from the Kaiser Permanente Southwest region isolated and analyzed cohorts of fractures from a pool of subtrochanteric/shaft fractures derived from medical records of patients admitted for femoral fractures. The incidence of AFFs among all forms of femoral fractures was uniformly low (22/3515 in Thompson et al [31], 75/5034 in Feldstein et al [32], 38/3078 in Lo et al [33], and 142/11466 in Dell et al [34]). AFFs also formed a low percentage of subtrochanteric/ shaft fractures in Meier et al (39 AFFs of 477 subtrochanteric/ shaft fractures) [35] and Warren et al (6/528) [36]. Shkolnikova et al (20/62) has reported, by far, the highest incidence (30 %) of AFFs among subtrochanteric/shaft fractures [37]. Using more specific case definition criteria, the proportion of patients (exposed to BP) with fracture appearances fulfilling criteria for AFFs was also higher than those with nonatypical patterns (62 % vs 16 % in Feldstein et al [32], 97 % vs 42 % in Lo et al [33], and 82 % vs 6 % in Meier et al
67
[35]. Increased incidences of AFFs with increasing duration of BP exposure has been implied in Dell et al (1.8 per 100,000 cases per year for 0.1–1.9 years of use to 113.1 per 100,000 cases per year for 8.0–9.0 years of use) [33] and Meier et al (BP exposure of 5–9 years had OR of 117.1, 95 % CI, 34.2– 401.7 compared with shorter durations) [35]. Besides a significant association with BP usage, AFFs also were found to be associated with: & & &
A younger cohort (74 vs 81 years) [32] and (73 +/- 10 vs 80 +/-12) [37], Asian populations (50 % vs 2 %) [33], A previous stress fracture of the contralateral femur (40 % vs 2 %) [33]; (28 % vs 0.9 %) [35], and glucocorticoid use [34, 36].
Overall, epidemiologic estimates point to a very low incidence and absolute risk for AFFs. The isolation of fractures exhibiting specific clinical and radiological criteria as defined by the ASBMR Task has led to identification of positive associations with BP therapy including longer duration of use. Together with evolving epidemiological characteristics including gender and ethnic predilection, these should be considered in elucidating the pathophysiological mechanisms behind AFFs. This is discussed in the following section.
Pathophysiology: Proposed Mechanisms The true pathogenesis of AFFs remains largely unknown. However, its epidemiologic association with BP therapy warrants attention. BPs can affect the mechanical properties of bone in a variety of ways. Animal studies inform us about how they can alter the collagen within the bone organic matrix, affect bone mineralization, and bone remodeling. BPs alter collagen maturity and cross-linking, as indicated by the increase in pyridinoline (PYD)/deoxypyridinoline (DPD) ratio and this has been shown to increase the strength (which is the stress needed to break bone) and stiffness of bone [38, 39] (also known as the resistance to bending, measured by Young's modulus of elasticity as the ratio of stress to strain). However, reducing bone turnover also increases pentosidine, which interacts with collagen through oxidative nonenzymatic cross-linkage leading to the accumulation of AGEs in bone, which are in turn associated with reduced toughness. Besides its effects on the organic matrix, the major effect of BP therapy is the reduction of bone remodeling. This process seems to be associated with a narrowing of the bone mineral density distribution, particularly at cortical sites, including the proximal femur. The clinical implication of a having a more homogeneously mineralized bone tissue is that it may be less effective in preventing fracture crack initiation
68
and propagation [40]. Another potential issue with a reduction in remodeling is the accumulation of microdamage in bone tissue since the mechanism for removing cracks through osteoclastic resorption might have become impaired. While animal studies have confirmed that BP use can result in the accumulation of microdamage [41], the data in humans are less conclusive [42, 43]. This issue of microdamage accumulation may be further compounded by the preferential localization of BPs at sites of high bone remodeling, including sites of stress fractures, areas of microdamage, and completed fractures as a result of injury-associated hypervascularity [44]. Therefore, by suppressing remodeling at these local sites, in AFFs BPs can potentially affect the intracortical repair of a developing stress fracture, allowing the initial crack to progress to a full fracture. There are also suggestions that BPs may be antiangiogenic, further impairing the process of a stress fracture repair [45], echoing some of the proposed pathogenic mechanism in antiresorptive drug-related osteonecrosis of the jaw (ARONJ). Our group has also reported a novel recent observation made with regards to the presence of bacterial biofilms at the site of an AFF, although this has been documented only in a single case, of which the pathologic significance remains uncertain [46]. A peculiar pharmacological feature of BPs is their extremely high affinity for, and consequent deposition into, bone relative to other tissues. After bone uptake, BPs are liberated again only when the bone in which they are deposited is resorbed and therefore, the skeletal half-life of BPs can be very long [47]. It is, therefore, not surprising that concerns have surfaced regarding accumulation of the drug and its continued and persistent suppression of bone remodeling. Most of the AFF case reports which have included bone histomorphometric information were derived from iliac crest bone biopsies. These showed reduced bone turnover, which were expected with BP treatment, including the loss of double tetracycline labels, which is a common and expected finding in BP-treated subjects [48, 49], and may even be seen in some untreated postmenopausal women [50, 51]. While bone biopsies located at the subtrochanteric region might yield a better understanding of the local environment, there is the potential pitfall of confounding by the injury-associated rise in remodeling. Most reported cases with biopsies obtained close to the fracture concur with a process associated with reduction in bone remodeling. However, one particular case report noted the contrary ie, a process consistent with an increased in bone resorption, in which the biopsy taken close to the fracture site showed an increase in eroded surface and osteoclast numbers [5]. In the little information available on bone turnover markers in AFFs, they are usually within the normal premenopausal range, occasionally elevated and suppressed in a minority of cases [2••]. Of note, these measurements when obtained after a fracture may reflect fracture healing rather than the rate of bone remodeling throughout the skeleton.
Curr Osteoporos Rep (2014) 12:65–73
It is interesting to note that the AFFs which are phenotypically indistinguishable from the cases of BP-associated AFFs may occur in BP-naive individuals suggesting that the pathophysiology cannot be entirely explained by BPs [52]. Genetic factors have been proposed as possible mechanisms why certain individuals may become susceptible to AFFs. For example, patients with X-linked hypophosphatemia (XLH) and hypophosphatasia can have pseudofractures that resemble AFFs, leading to the hypothesis that carriers of mutations or polymorphisms of the gene encoding the tissue nonspecific (bone) isoenzyme of alkaline phosphatase (TNSALP) may be one such predisposed group [53]. It is uncertain the use of an additional antiresorptive agent (eg, estrogen, raloxifene, or calcitonin) has any clinical significance, although there are several case reports of AFFs of patients on dual antiresorptives [2••].There have been suggestions that vitamin D deficiency also might be an important risk factor [2••, 28]. Thus far, only one series have examined the association between Vitamin D levels and atypical femoral fractures [28]. In this study, serum 25-hydroxyvitamin D (25OHD)
EPIDEMIOLOGY AND PATHOPHYSIOLOGY (PR EBELING AND EF ERIKSEN, SECTION EDITORS)
Review: Epidemiology and Pathophysiology of Atypical Femur Fractures Alvin C. Ng & Meng Ai Png & David T. Chua & Joyce S. B. Koh & Tet Sen Howe
Published online: 14 February 2014 # Springer Science+Business Media New York 2014
Abstract The recent recognition of the clinical phenomenon of atypical femoral fractures has garnered significant scientific interest. In this review, we will discuss and summarize the salient developments in the current understanding of the epidemiology, pathophysiology, and radiology of atypical femoral fractures. Keywords Atypical femoral fractures . Stress . Insufficiency . Subtrochanteric . Femoral . Diaphyseal osteoporosis . Bisphosphonates . Epidemiology . Pathophysiology . Radiology . Bone . Pain . Suppression of bone turnover
sufficiently rare that it is unlikely any randomized prospective clinical trial will ever detect them. Thus, the onus is on the clinicians to remain vigilant and report sporadic cases so that eventually a clearer picture emerges in the diagnosis and treatment of these rare and baffling conditions. We review atypical femoral fractures (AFFs) in this article, fully cognizant of the fact that although these fractures were first reported in the literature in 2005, even today we have much to learn about them.
Epidemiology Introduction Bisphosphonates (BPs) are the most commonly used drug in the treatment of osteoporosis and have an excellent safety profile. Most early randomized prospectively clinical trials did not show any major adverse effects. However, with more widespread use, cases of osteonecrosis of the jaw and atypical femoral fractures have been increasingly reported. These are
A. C. Ng (*) Department of Endocrinology, Singapore General Hospital, Outram Road, Singapore 169608, Singapore e-mail: [email protected] M. A. Png Department of Radiology, Singapore General Hospital, Singapore, Singapore D. T. Chua Department of Orthopedic Surgery, Changi General Hospital, Singapore, Singapore J. S. B. Koh : T. S. Howe Department of Orthopedic Surgery, Singapore General Hospital, Singapore, Singapore
There is an evolutionary trend in epidemiological studies published for AFFs from its first recognition till the present time. A comprehensive summary of the early epidemiological studies contributing to our knowledge of AFF was published by Nieves and Cosman in 2010 [1]. They classified them into case reports, case series, and epidemiologic studies. These reports predate the first ASBMR Task Force Report of 2010 [2••]. Hence, they have largely been referenced in the Task Force’s Report and were contributory to the original case definition criteria.
Epidemiologic Studies Pre-Dating the 1st ASBMR Task Force Report Case Reports and Series The earliest case report was published as early as 1997 describing a subtrochanteric stress fracture of the femur following total knee arthroplasty [3]. The clinical entity remained hardly recognized till the mid-2000s as subsequent case reports and series only appeared between 2005 [4] and 2009 [5–7]. Odvina et al [8] published the first case series that
66
Curr Osteoporos Rep (2014) 12:65–73
included AFFs, in which five of the ten reported fracture cases were femoral. The subsequent series of femoral fractures by Goh et al [9] stressed the unusual location of this fracture in the subtrochanteric region. This region (defined as being from the lesser trochanter to the junction of the proximal and middle third of the femoral shaft) is notably resilient to traumatic injuries and is largely broken only in high energy trauma or pathological fracture from metastatic disease [10]. The predilection for this anatomic region was also demonstrated in a series by Neviaser et al where 50 of 70 low energy femoral shaft fractures were subtrochanteric [11] as well as other subsequent series [12, 13]. Features described in these case reports and series differentiating AFFs from the common low energy osteoporosisrelated hip fractures included & & & & & & & & &
Prodromal pain described as discomfort, weakness, or actual pain involving the thigh or lower limb for weeks or months preceding the fracture [3, 9, 14–17]; Use of another antiresorptive or steroid therapy, in addition to the bisphosphonate [5, 8, 9, 17, 18]; Lack of trauma precipitating a fracture [3, 9, 15, 17, 19, 20]; Bilaterality (either simultaneous or sequential) [4, 5, 9, 17, 19]; Transverse fractures [3, 9]; Cortical hypertrophy or thickness [16, 17]; Stress reaction on the affected and/or unaffected side [3, 4, 15, 17, 18]; Poor fracture healing [7, 8, 17]; and Normal or low bone mass but not osteoporosis in the hip region [5, 8, 9, 13].
These reports served to highlight unusual features observed in the clinical and radiological presentations of these fractures, thus, bringing the attention of the medical community to them as a distinct entity.
Early Epidemiologic Studies Before the definition of major and minor criteria for identifying AFFs by the ASBMR Task Force [2••], there were epidemiologic studies attempting to define potential associations of AFFs with BP therapy, and to explore the risks of these fractures compared with the risk of osteoporotic fractures. These studies have demonstrated somewhat contradictory results. This could be due to a lack of proper case definition criteria and also the use of different denominators. Almost all cohort studies predating the first ASBMR Task Force Report on atypical femur fractures [2••] failed to show any association between BP intake and AFFs. These early registry-based cohort studies from Denmark [21, 22], Taiwan [23],
Pennsylvania, and New Jersey [24] based their denominators on the presence or absence of BP intake. They could reliably identify BP and other concomitant drug therapy but lacked radiographic adjudication in differentiating the usual osteoporotic femur fractures from “nontypical fractures”. Subtrochanteric and diaphyseal femur fractures were only identified from disease coding of national hospital discharge information [21, 22] or national health insurance or utilization databases [23, 24], and there was no way to verify the exact radiological characteristics of these fractures. Because of heterogeneity of fracture configurations in the subtrochanteric and diaphyseal regions, incorporating all such fractures into a single arm for analysis would logically not be specific enough to demonstrate potential associations between BP intake and AFFs. However, AFFs are fractures with specific features forming part of the total pool of subtrochanteric/ diaphyseal femoral fractures. Hence, the potential usefulness of these studies lies in defining the upper risk limit for AFFs. Overall, all such studies show a far lower incidence of subtrochanteric/diaphyseal fractures compared with hip fractures. Even so, an interesting trend of a sustained incidence of subtrochanteric and diaphyseal fractures seemed evident against a decreasing trend of hip fractures in populations exposed to BP therapy between 1996 and 2006 [25]. This could be interpreted as a lack of protection against subtrochanteric/diaphyseal fractures by BPs [21] or indeed an actual increase in AFFs, which now offset the decline in typical osteoporotic diaphyseal femur fractures affected by antiresorptive therapy. The latter is plausible as shown by another study involving a large-scale nationwide survey of hip fractures and medication use in the elderly US population. Wang et al [26] found a decrease in age-adjusted rates for typical hip fractures by 31.6 % among women and 20.5 % among men from 1996 to 2007. This contrasted with an increase of 20.4 % in subtrochanteric fragility fractures among women but no such trend in men over the same period. This corresponded to an increasing trend in the national use of BPs over the same period. It is also notable that women were the predominant users of BPs and sustained the vast majority of subtrochanteric fractures. However, in contrast to using BP intake as the variable, studies using fractures with the specific radiological features of AFFs as inclusion criteria tended to show a different picture. Lenart and co-workers found a significantly higher intake of BP among this group compared with controls [27]. Similarly, Girgis et al found an atypical fracture pattern 96.7 % specific to patients receiving BPs, although other risk factors such as history of low-energy fracture, the use of glucocorticoid therapy for more than 6 months, active rheumatoid arthritis, and a level of serum 25-hydroxyvitamin D of less than 16 ng per milliliter were also identified [28]. Giusti et al identified a very low incidence of AFF fracture patterns (only
Curr Osteoporos Rep (2014) 12:65–73
ten of 63 low energy subtrochanteric/diaphyseal femur fractures were considered atypical) but a high odds ratio (OR, 17.0; 95 % CI, 2.6–113.3) in this group for BP and glucocorticoid use [29].
Epidemiologic Studies after the ASBMR Task Force Report The ASBMR Task Force first consensus report on atypical fractures published in November 2010 [2••] defined major and minor criteria for AFFs, thus refining the methodology for case finding in subsequent studies. Studies that followed were now largely cohort studies using fracture patterns as the primary variable. In a cohort analysis of 12,777 women 55 years of age or older with femur fractures, Schilcher et al isolated 59 AFFs out of 1271 subtrochanteric/shaft fractures [30]. The ageadjusted relative risk of atypical fracture was 47.3 (95 % confidence interval [CI], 25.6 to 87.3), far higher than the relative risk of 1.27 (95 % CI, 0.85–1.90) for any other type of fracture associated with BP use. The risk of an atypical fracture rose with an increasing duration of BP use, with an odds ratio of 1.3 (95 % CI, 1.1–1.6) per 100 prescribed daily doses. This risk was approximately ten times as high as a normal level of fracture risk within the first 2 years of use and 50 times as high thereafter. However, in view of the low absolute risk of 50 cases per 100,000 patients-years (95 % CI, 4–7), the authors argued that the benefits of fracture prevention with BP use in the presence of appropriate indications greatly outweighed the risk of AFFs. Other epidemiologic studies with similar methodologies also yielded similar findings. Thompson et al from the United Kingdom [31], Feldstein et al [32], and Lo et al [33] from the Kaiser Permanente Northwest region of California, and Dell et al [34] from the Kaiser Permanente Southwest region isolated and analyzed cohorts of fractures from a pool of subtrochanteric/shaft fractures derived from medical records of patients admitted for femoral fractures. The incidence of AFFs among all forms of femoral fractures was uniformly low (22/3515 in Thompson et al [31], 75/5034 in Feldstein et al [32], 38/3078 in Lo et al [33], and 142/11466 in Dell et al [34]). AFFs also formed a low percentage of subtrochanteric/ shaft fractures in Meier et al (39 AFFs of 477 subtrochanteric/ shaft fractures) [35] and Warren et al (6/528) [36]. Shkolnikova et al (20/62) has reported, by far, the highest incidence (30 %) of AFFs among subtrochanteric/shaft fractures [37]. Using more specific case definition criteria, the proportion of patients (exposed to BP) with fracture appearances fulfilling criteria for AFFs was also higher than those with nonatypical patterns (62 % vs 16 % in Feldstein et al [32], 97 % vs 42 % in Lo et al [33], and 82 % vs 6 % in Meier et al
67
[35]. Increased incidences of AFFs with increasing duration of BP exposure has been implied in Dell et al (1.8 per 100,000 cases per year for 0.1–1.9 years of use to 113.1 per 100,000 cases per year for 8.0–9.0 years of use) [33] and Meier et al (BP exposure of 5–9 years had OR of 117.1, 95 % CI, 34.2– 401.7 compared with shorter durations) [35]. Besides a significant association with BP usage, AFFs also were found to be associated with: & & &
A younger cohort (74 vs 81 years) [32] and (73 +/- 10 vs 80 +/-12) [37], Asian populations (50 % vs 2 %) [33], A previous stress fracture of the contralateral femur (40 % vs 2 %) [33]; (28 % vs 0.9 %) [35], and glucocorticoid use [34, 36].
Overall, epidemiologic estimates point to a very low incidence and absolute risk for AFFs. The isolation of fractures exhibiting specific clinical and radiological criteria as defined by the ASBMR Task has led to identification of positive associations with BP therapy including longer duration of use. Together with evolving epidemiological characteristics including gender and ethnic predilection, these should be considered in elucidating the pathophysiological mechanisms behind AFFs. This is discussed in the following section.
Pathophysiology: Proposed Mechanisms The true pathogenesis of AFFs remains largely unknown. However, its epidemiologic association with BP therapy warrants attention. BPs can affect the mechanical properties of bone in a variety of ways. Animal studies inform us about how they can alter the collagen within the bone organic matrix, affect bone mineralization, and bone remodeling. BPs alter collagen maturity and cross-linking, as indicated by the increase in pyridinoline (PYD)/deoxypyridinoline (DPD) ratio and this has been shown to increase the strength (which is the stress needed to break bone) and stiffness of bone [38, 39] (also known as the resistance to bending, measured by Young's modulus of elasticity as the ratio of stress to strain). However, reducing bone turnover also increases pentosidine, which interacts with collagen through oxidative nonenzymatic cross-linkage leading to the accumulation of AGEs in bone, which are in turn associated with reduced toughness. Besides its effects on the organic matrix, the major effect of BP therapy is the reduction of bone remodeling. This process seems to be associated with a narrowing of the bone mineral density distribution, particularly at cortical sites, including the proximal femur. The clinical implication of a having a more homogeneously mineralized bone tissue is that it may be less effective in preventing fracture crack initiation
68
and propagation [40]. Another potential issue with a reduction in remodeling is the accumulation of microdamage in bone tissue since the mechanism for removing cracks through osteoclastic resorption might have become impaired. While animal studies have confirmed that BP use can result in the accumulation of microdamage [41], the data in humans are less conclusive [42, 43]. This issue of microdamage accumulation may be further compounded by the preferential localization of BPs at sites of high bone remodeling, including sites of stress fractures, areas of microdamage, and completed fractures as a result of injury-associated hypervascularity [44]. Therefore, by suppressing remodeling at these local sites, in AFFs BPs can potentially affect the intracortical repair of a developing stress fracture, allowing the initial crack to progress to a full fracture. There are also suggestions that BPs may be antiangiogenic, further impairing the process of a stress fracture repair [45], echoing some of the proposed pathogenic mechanism in antiresorptive drug-related osteonecrosis of the jaw (ARONJ). Our group has also reported a novel recent observation made with regards to the presence of bacterial biofilms at the site of an AFF, although this has been documented only in a single case, of which the pathologic significance remains uncertain [46]. A peculiar pharmacological feature of BPs is their extremely high affinity for, and consequent deposition into, bone relative to other tissues. After bone uptake, BPs are liberated again only when the bone in which they are deposited is resorbed and therefore, the skeletal half-life of BPs can be very long [47]. It is, therefore, not surprising that concerns have surfaced regarding accumulation of the drug and its continued and persistent suppression of bone remodeling. Most of the AFF case reports which have included bone histomorphometric information were derived from iliac crest bone biopsies. These showed reduced bone turnover, which were expected with BP treatment, including the loss of double tetracycline labels, which is a common and expected finding in BP-treated subjects [48, 49], and may even be seen in some untreated postmenopausal women [50, 51]. While bone biopsies located at the subtrochanteric region might yield a better understanding of the local environment, there is the potential pitfall of confounding by the injury-associated rise in remodeling. Most reported cases with biopsies obtained close to the fracture concur with a process associated with reduction in bone remodeling. However, one particular case report noted the contrary ie, a process consistent with an increased in bone resorption, in which the biopsy taken close to the fracture site showed an increase in eroded surface and osteoclast numbers [5]. In the little information available on bone turnover markers in AFFs, they are usually within the normal premenopausal range, occasionally elevated and suppressed in a minority of cases [2••]. Of note, these measurements when obtained after a fracture may reflect fracture healing rather than the rate of bone remodeling throughout the skeleton.
Curr Osteoporos Rep (2014) 12:65–73
It is interesting to note that the AFFs which are phenotypically indistinguishable from the cases of BP-associated AFFs may occur in BP-naive individuals suggesting that the pathophysiology cannot be entirely explained by BPs [52]. Genetic factors have been proposed as possible mechanisms why certain individuals may become susceptible to AFFs. For example, patients with X-linked hypophosphatemia (XLH) and hypophosphatasia can have pseudofractures that resemble AFFs, leading to the hypothesis that carriers of mutations or polymorphisms of the gene encoding the tissue nonspecific (bone) isoenzyme of alkaline phosphatase (TNSALP) may be one such predisposed group [53]. It is uncertain the use of an additional antiresorptive agent (eg, estrogen, raloxifene, or calcitonin) has any clinical significance, although there are several case reports of AFFs of patients on dual antiresorptives [2••].There have been suggestions that vitamin D deficiency also might be an important risk factor [2••, 28]. Thus far, only one series have examined the association between Vitamin D levels and atypical femoral fractures [28]. In this study, serum 25-hydroxyvitamin D (25OHD)
