Atypical Femur Fractures in Patients Treated with Bisphosphonates: Identification, Management, and Prevention

Osteoporosis is a common condition with significant health care costs. First-line therapy is with bisphosphonates, which have proven anti-fracture efficacy. Around 10 years after the introduction of bisphosphonates reports began to be published of atypical femoral fractures (AFFs) that may be associated with this therapy. These fractures are associated with significant morbidity although lower mortality than the more common osteoporotic neck-of-femur fractures. A case definition has been described to allow identification of this class of fracture. Further work has established a high relative risk of AFFs in patients treated with bisphosphonates, but a low absolute risk in comparison to that of osteoporotic fractures. Proposed pathological mechanisms include low bone turnover states leading to stress/insufficiency fractures. Clinicians should be aware of the risk of AFFs and in particular the high rate of prodromal thigh/groin pain that warrants investigation in a patient receiving a bisphosphonate. If an incomplete fracture is diagnosed then bisphosphonate therapy needs to be stopped and prophylactic surgery may be considered. Due to these rare side effects patients on bisphosphonates require regular review, and this is particularly advised after 5 years of oral or 3 years of intravenous therapy.


BACKGROUND
Osteoporosis is a skeletal disorder characterized by compromised bone strength predisposing to an increased risk of fracture. 1 It is common with an increasing prevalence with age and a greater incidence in post-menopausal women than in men. In 201 0 the prevalence in the 27 countries of the EU was approx imately 27 .6 million. 2 In the United States 1 0 million people ov er the age of 50 hav e osteoporosis. 3 The consequence of osteoporosis is fragility fractures that cause significant morbidity as well as mortality . Worldwide there is a fracture almost every 3 seconds, resulting in 8.9 million fractures annually; 1 in 3 women and 1 in 5 older men will suffer a fragility fracture after the age of 50 with a financial burden that is comparable to other chronic conditions such as asthma, hy pertensiv e heart disease, or rheumatoid arthritis.
The bisphosphonates (BP) are analogues of inorganic pyrophosphate. Bisphosphonates act by inhibiting osteoclast activ ity and thereby inhibit bone resorption. These drugs are commonly prescribed for osteoporosis and hav e been shown in good-quality randomized controlled trials to reduce the risk of fragility fractures at vertebral, non-vertebral, and hip sites (Table 1 ). These studies are mostly of 3-4 y ears' duration and show fracture reduction rates of 40%-7 0% in v ertebral fractures, 1 5%-39% in non-v ertebral fractures, and 20%-50% in hip fractures. [4][5][6][7] On the basis of this evidence, BPs are considered first-line therapy for patients requiring treatment for osteoporosis and are also used in other conditions such as Paget's disease of bone, malignancies involving bone, and tumor -induced hy percalcemia. In addition to the anti-fracture efficacy a reduction in mortality has also been reported in patients treated with intrav enous BP following a hip fracture. 8 The first report of a possible link between BP use and "aty pical fractures" of the femoral shaft was published by Odv ina et al. in 2005. 9 They reported nine patients who had sustained non-traumatic nonv ertebral fractures whilst taking alendronate. Some patients were also taking estrogen and/or glucocorticoids. After this report followed many other case reports of similar fractures. There was often little or no trauma inv olved in these fractures, and ov er 7 0% were preceded by thigh or groin pain. 10 These cases caused great anxiety for patients and clinicians alike, and therefore the American Society of Bone and Mineral Research (ASBMR) convened a taskforce in 2009 to inv estigate this area and ex amine the ev idence in a sy stematic manner. 10 They repeated this in 201 3, publishing an updated report. 11

CASE DEFINITION
The ASBMR task force reviewed and summarized the av ailable ev idence, and this has led to the agreement of a case definition of an atypical femoral fracture (AFF). The aim of this was to identify AFFs as distinct from typical neck-of-femur fractures that are seen with osteoporosis, and femoral shaft fractures that are seen with high-trauma injuries. With a case definition having been established , this has guided ongoing research and aided identification of this cohort of patients.
It is worthy of note that there are cases of AFF in all case series in indiv iduals who hav e not been ex posed to bisphosphonates. 12 The classification of a femoral fracture as an AFF specifically excludes high-trauma fractures, fractures of the neck of femur, intertrochanteric fractures with subtrochanteric spiral ex tension, and pathological fractures associated with primary or secondary metastatic bone disease or other metabolic bone diseases. The case definition was honed in the 201 3 ASBMR report to clarify features that distinguish AFFs from ordinary osteoporotic fractures of the femur. A diagnosis relies on hav ing a least four of fiv e major features, with the major features being: minimal trauma, fracture originates laterally and is transv erse, complete fractures may hav e medial spike, incomplete only involve lateral cortex, the fracture is non-comminuted or minimally comminuted, and the presence of localized periosteal or endosteal thickening of the lateral cortex ("beaking" or "flaring"). There are associated minor features, none of which was required to make a diagnosis, but which hav e been associated with AFFs. Localized periosteal ("beaking" or "flaring") or end-osteal thickening of the lateral cortex at the fracture site was added to the case definition 201 3 as that had recently been reported. See Table 2 for case definition.

EPIDEMIOLOGY
The best-quality epidemiological ev idence comes from studies with radiographic adjudication of the fractures as definite AFFs. Relative risk of BP use in AFF is v ery v ariable in different studies, ranging between 2-and 1 28-fold. 13,14 However, the absolute risk is much lower. Again, this v aries between studies, but ranges between 3.2 and 50 cases per 1 00,000 person y ears. 15,16 A recent sy stematic rev iew has reported an adjusted odds ratio of 2.7 1 for subtrochanteric fractures in patients taking bisphosphonates. 17 Some but not all studies suggest an increase in incidence with duration of therapy-e.g. one study reported an increase in age-adjusted incidence from

ASBMR: Definition of AFF
To satisfy the case definition of AFF, the fracture must be located along the femoral diaphysis from just distal to the lesser trochanter to just proximal to the supracondylar flare In addition, at least four of five major features must be present; none of the minor features is required, but they have sometimes been associated with these fractures 1 .8/1 00,000/y ear with 2 y ears of treatment to 1 1 3/100,000/year with 8-9.9 y ears of treatment. 18 Meier et al. reported an exposure of 5 -9 years being associated with a greater risk of AFF OR of 1 17 .1. 16 Schilcher et al. reported in 201 4 that use of a bisphosphonate for ov er 4 y ears increased the relativ e risk to 1 26, with an absolute risk of 1 1 /1 0,000 person y ears. 19 The 201 0 ASBMR task force report examined all the ev idence and concluded that the incidence of AFF was v ery low, particularly in light of the large numbers of fractures that are prevented by bisphosphonate use, and also that a causal link to bisphosphonates had y et to be proven. Again in 2013 they concluded that there was still no ev idence for a causal relationship, but that the "fairly consistent magnitude of association" is unlikely to be accounted for by confounders that may be currently unknown or unmeasured. Further weight is giv en to potential causality by evidence that stopping treatment with a bisphosphonate reduces the risk going forward of AFFs with a 7 0% reduction in AFFs one y ear after stopping therapy . 19 It is worthy of note that AFFs hav e also been described in patients treated with denosumab (monoclonal antibody against receptor activator of nuclear factor kappa B ligand), which is also a potent inhibitor of bone resorption. 20,21 Incidence of AFF appears to be similar to that seen with bisphosphonate therapy , although no study has made a direct comparison between bisphosphonates and denosumab. Approximately 7 0% hav e a history of prodromal thigh or groin pain, 28% had bilateral fractures and bilateral radiographic abnormalities, and 26% had delay ed healing; 34% had concomitant glucocorticoid use. 10 It has been suggested that the risk of AFFs may be high in indiv iduals taking glucocorticoids, 22 proton pump inhibitors, 23 and in patients with rheumatoid arthritis or diabetes. Other reported potential risk factors include genu v arus, v arus/bowed femur, 24 and collagen disease. 25 The risks appear to be increased in patients of Asian origin. A recent paper showed an 8-fold increased risk in indiv iduals from an Asian background. 26 In addition one study showed a particular increase in AFFs in indiv iduals treated with a bisphosphonate with low bone mass (osteopenia) in comparison with those with osteoporosis. 27 A recent paper has reported two cases of AFF seen in patients where radiographs taken prior to bisphosphonate therapy showed localized cortical thickening, which raises the possibility that AFFs occur at the sites of pre-ex isting stress or insufficiency fractures. 28 Although AFFs are associated with significant morbidity the mortality rates are significantly lower than seen in osteoporotic neck of femur fractures, with a one-y ear reported mortality of 2.4% in comparison to 9.6% among women aged 7 0 -7 5 in a comparable Swedish cohort. 29,30

PAT HOGENESIS
A number of possible mechanisms hav e been put forward to ex plain the pathogenesis of these fractures. Currently this area remains largely unclear. There are commonalities between AFFs and stress fractures. A stress fracture is caused by abnormal loading of normal bone, whereas an insufficiency fracture is caused by normal loading of an abnormal bone. Both these ty pes of fractures are most common in the lower limbs due to the increased loading. In a stress fracture microcracks appear that ov er time coalesce and without repair progress to a fracture. The current consensus of the ASBMR task force is that AFFs are stress or insufficiency fractures that progress ov er time. 11 There hav e been long-standing concerns that by reducing bone turnover the bisphosphonates may lead to brittle bones. The features of a transv erse fracture and the lack of comminution are features of fractures in brittle bones. Almost all biopsy studies in AFFs rev eal reduced or absent osteoblasts and osteoclasts and little or no tetracy cline doublelabeling, which would indicate suppression of bone remodeling. 10 This may be ex pected in patients taking a bisphosphonate, and not all biopsies were from the fracture site, with some being iliac crest biopsies. Howev er, the suggestion is that bone turnover is suppressed, potentially causing insufficiency fractures under normal loading conditions. Further theories hav e been put forward that include the effects of bisphosphonates on the collagen in the organic matrix of bone, 31 effects on angiogenesis, 32 the healing of fatigue fractures, and a reported influence of the lower limb geometry on the stress generated on the lateral aspect of the femoral cortex . 33,34

MANAGEMENT OF AFFS
It appears that these AFFs are stress fractures that dev elop ov er time. 35 Initially one can identify a cortical bump that is thought to represent early periosteal thickening, then a transv erse cortical lucency develops, which may progress to become a complete fracture.
More than half of the patients experience prodromal thigh/groin pain. 9 Health professionals need to be aware of AFF as a potential complication of BP use and the potential significance of thigh pain. Patients on BP (or other anti-resorptiv e drugs) should be routinely asked about thigh/groin pain, and patients should also be educated to be aware of these sy mptoms. If sy mptoms are reported then a plain radiograph of the femur should be obtained ( Figure 1 ). If the radiograph has features to suggest incomplete fragility fractures, such as cortical thickening, then further imaging should be obtained. Ideally magnetic resonance imaging (MRI) should be performed, which allows assessment of a cortical fracture line, associated bone marrow edema, or hy peremia suggestive of a stress fracture (Figure 2). A cortical lucency would suggest an incomplete AFF, whereas purely edema would be more in keeping with a stress reaction. If MRI is not possible computerized tomography (CT) can also help and would detect fracture lines and new bone formation (Figure 3). In addition, incomplete fractures will be metabolically activ e on isotope bone scanning, although this will not demonstrate a fracture line.
It is essential that imaging of the contralateral femur, ev en if asy mptomatic, is also carried out, as these fractures are often bilateral.
If an incomplete fracture is detected the BP should be stopped immediately . Patients should receive adequate calcium and v itamin D. If there is pain, then orthopedic interv ention with intramedullary nailing is adv isable to prevent progres-   sion to a complete fracture. If there is no pain, then management can be conserv ativ e with reduced weight-bearing for 2-3 months. If there is no improv ement, then prophylactic nailing should be considered. 11 Teriparatide (1 -34 parathy roid hormone analogue, TPD) has been used as a treatment to improv e fracture healing in cases of AFF. This is the only currently av ailable anabolic agent for bone. Sy stematic review has shown positive outcomes for fracture healing with the use of TPD after AFF. 36 Histology shows increased bone formation in patients treated with TPD for AFFs. 37 Although no randomized controlled trial data are currently av ailable for TPD it seems reasonable to try in patients who fail to heal with c onservative therapy. There is an ongoing randomized controlled trial to ev aluate this treatment formally .
There are two reported cases of strontium ranelate being used and improving fracture healing. 38,39 Recent warnings about cardiov ascular risk may preclude this, but short-term use may be helpful in some indiv iduals, particularly in health economies where it may be difficult to get access to TPD. There is also one reported case in the literature of TPD being used on a weekly rather than daily basis and improv ing fracture healing, which may be more cost-effective. 40 In addition one case has been reported of the use of low-intensity pulsed ultrasound, which may prove helpful in patients in whom TPD is contraindicated (e.g. those with malignancy ). 41 After an AFF the options for treatment of osteoporosis are TPD or ralox efine (selectiv e estrogen receptor modulator); these have not been associated with AFF and are not anti-resorptiv e in mode of action.

PREVENT ION OF AFFS
These rare complications and others such as osteonecrosis of the jaw hav e led to a change in practice with regard to long-term BP therapy . It is now adv ised that BP therapy should be regularly reviewed with the aim of stopping treatment in patients who no longer require it and thereby reducing the risk of complications of long-term treatment. The optimal duration of treatment with a BP is not known.
Bisphosphonates strongly adhere to hy drox yapatite on the bone surface and therefore hav e a persistent effect even when therapy is stopped. Two published studies have evaluated the long-term use of BPs. The FLEX study was a 5 -y ear ex tension of the piv otal fracture prevention study with alendronate and showed a persev erance of fracture risk reduction for 2 y ears after a 5 -y ear course of treatment. 4 An ex tension of the HORIZON study compared the use of 3 y ears of intrav enous zoledronic acid and stopping treatment, with 6 y ears of continuous therapy . 42 It showed that 3 y ears of treatment gives fracture risk reduction for a further 3 y ears. This evidence for an ongoing effect of treatment, ev en when it is stopped, and the risks of longterm side effects hav e led to the concept of a "drug holiday ." The UK National Osteoporosis Guideline Group (NOGG) adv ises a treatment review after 5 years of treatment with oral BPs and 3 y ears of treatment with zoledronic acid. 43 They recommend continued treatment in those at high risk, for which they give the ex amples of: aged 7 5 or more, prev ious hip/ v ertebral fracture, glucocorticoids at a dose of ≥7 .5mg daily . The ASBMR has also published a taskforce report on managing osteoporosis patients after long-term BP therapy and recommend that high-risk older women with a low hip T-score or high fracture risk should continue on treatment. 44 The UK NOGG also recommend a re v iew of treatment in those who sustain one or more fractures when receiving therapy (after checking >80% compliance) and also recommend treatment is continued in patients where the total hip or femoral neck T-score is below -2.5.
The ASBMR suggest that for those women not at high fracture risk after 3-5 years of treatment a drug holiday of 2-3 y ears could be considered with periodic reassessment, and NOGG state that if treatment is stopped fracture risk should be reassessed if a patient has a new fracture or after 2 y ears if no fracture occurs. In the case of zoledronic acid 3 y ears of treatment should be sufficient for the majority of patients.
The United States Food and Drug Administration hav e added to the "warning and precautions" regarding bisphosphonates and AFFs, advising that patients on bisphosphonates should be reassessed after 3-5 y ears of therapy .
The ASBMR long-term BP taskforce report makes the v alid point that it is unlikely that there will be further future studies to address the question of long-term BP use and therefore we are unlikely to hav e further data av ailable to make definitiv e recommendations. 44 Ev idence suggests that stopping a BP will hav e an influence on the risk of AFF. Schilcher et al. reported on a Swedish cohort of 45 women with confirmed AFFs who then stopped bisphosphonates. 19 They reported that the risk fell by 7 0% each y ear after stopping therapy with the most dramatic risk reduction in the first y ear. Howev er, it is worth noting that these analy ses were based on 46 AFF ev ents with only four occurring after the first y ear. There was only short-term follow-up, which may lead to the deriv ed estimates being overestimated.
Data from the Kaiser Database suggests that if BP is stopped soon after an AFF then only 20% will fracture the contralateral femur as opposed to 50% if the BP is continued for 3 y ears after AFF. 45

CONCLUSIONS
Aty pical femoral fractures ( Figure 4) are rare, but the ev idence when taken as a whole does suggest that their incidence is significantly increased in patients treated with BP therapy . The relativ e risk may be high, but the absolute risk is low. For the v ast majority of patients the reduction in risk of osteoporotic fracture with treatment v astly outweighs the risks of rare side effects such as AFFs.
Howev er, BP therapy should be regularly reviewed, particularly after 5 y ears, and only continued in those patients deemed to be at high risk. Patients who complain of thigh or groin pain must be inv estigated for potential incomplete fractures and managed appropriately, in particular hav ing their anti-resorptive therapy stopped. Health professionals need to be aware of risks, but it is important that the comparative risks of treatment v ersus lack of treatment are conv ey ed to patients.