Alendronate sodium (alendronate) is a bisphosphonate that has been extensively and successfully used for the treatment of osteoporosis [1–3]. The mechanism of action involves the induction of osteoclast apoptosis, thereby reducing bone resorption [4]. Recently, several studies have reported an increased prevalence of low-energy subtrochanteric femoral fractures in patients receiving long-term alendronate therapy [5–13]. Studies have also reported an association between the use of bisphosphonates and osteonecrosis of the jaw [14, 15]. The pathogenesis of these complications is not fully understood but may be related to osteoclast inhibition and the resulting suppression of bone turnover and bone remodeling, leading to atypical skeletal fragility [16–20]. The significance of recognizing this type of fracture is paramount. In our experience, the number of these types of fractures continues to increase and the clinical implications and importance of fractures reported in women receiving long-term bisphosphonates cannot be understated. This study focuses on the imaging findings in a series of patients receiving long-term alendronate therapy who experienced proximal femoral fractures.

Institutional review board approval was obtained, and informed consent was waived for this retrospective HIPAA-compliant study.

A retrospective review of the imaging findings in 34 proximal femoral fractures in 22 patients receiving long-term alendronate therapy was performed. The fractures occurred in 22 patients, 12 of whom suffered bilateral fractures. Twenty of these patients were seen at our orthopedic department, and two patients were referred for imaging during a 4-year period. All patients were women (age range, 50–81 years; mean age, 64 years). The patients had received alendronate therapy for a minimum of 4 years and up to 14 years (mean, 6 years). All patients reported no or minor trauma. The imaging studies included radiography (n = 34), bone scintigraphy (n = 4), CT (n = 4), and MRI (n = 5). For the complete fractures, the bone scintigraphy and the cross-sectional studies were performed to exclude underlying pathologic processes, in view of the lack of significant trauma; for the incomplete fractures, the studies were performed to further assess the cause for hip pain. For one patient with incomplete fracture, MRI was performed to exclude an osteoid osteoma, suspected on radiographs; for the other patients, MRI was performed to evaluate the cause of continued pain.

Two musculoskeletal radiologists, one with 20 years and one with 1 year of musculoskeletal radiology experience, reviewed the studies in consensus. The following fracture characteristics were recorded: complete or incomplete fracture, distance from lesser trochanter, fracture orientation, alignment, presence of associated lateral cortical thickening, and time to complete healing. When additional imaging techniques were used, the reviewers also reviewed those studies in consensus, with knowledge of the radiographic findings. In the case of CT scans, the same fracture characteristics described for radiographs were evaluated. When an MR study was performed (incomplete fractures only), the presence of cortical thickening, endosteal or periosteal edema, as well as a fracture line was noted. Finally, when a bone scan was obtained, the degree and pattern of uptake were noted.

A complete fracture was defined as a fracture that extended through the entire femoral cortex. An incomplete fracture was defined as partial involvement of the femoral cortex. In cases of incomplete fractures, the location of the fracture along the femoral shaft (medial or lateral) was documented. In cases of complete fractures, the displacement pattern of the distal fracture fragment and the presence of a medial beak were noted. A medial beak was defined as a spiked appearance of the medial femoral cortex at the fracture site. The distance from the lesser trochanter was determined by measuring the distance from the inferior edge of the lesser trochanter to the level of the most superior portion of the fracture site. Fracture orientation referred to the direction of the major fracture component (transverse, vertical, or oblique). The presence of associated cortical thickening was noted when there was a conspicuous difference in the thickness of the femoral cortex in the region of the fracture. Also, evolution of healing and time to complete healing, defined as complete or almost complete resolution of the fracture line, were determined on follow-up radiographs for 21 (62%) of 34 fractures.

All of the patients reported either no trauma history or a history of minor trauma, such as a fall from a standing height. In one patient with bilateral fractures (one side complete and the other side incomplete), the incomplete side was relatively asymptomatic. All fractures were transverse and were located 0.5–18.3 cm away from the lesser trochanter (mean, 4.8 cm). Most of the fractures (27 [79%] of 34) occurred ≤ 5 cm distal to the lesser trochanter.

Twenty (59%) of the 34 fractures were complete, and 17 (85%) of those factures had a medial beak with superior displacement and varus angulation of the distal fracture fragment (Fig. 1). In all 20 patients with complete fractures, the presence of a “skirt” of focal buttressing and thickening at the opposing lateral cortices could be seen (Fig. 2), suggesting chronicity as well as indicating that the fractures originate laterally.

In three of the 20 complete fractures, the fracture occurred much more distal to the lesser trochanter and there was no varus angulation at the fracture sites (Fig. 3A). In three of the 20 complete fractures, no medial beak was present (Fig. 3B).

In 14 (41%) of 34 cases, the fractures were incomplete. For all 14 incomplete fractures, radiographs showed focal thickening at the lateral aspect of the femoral cortex (indicative of chronicity), with (n = 10) (Fig. 4) or without (n = 4) (Fig. 5) a visible transverse cortical fracture line. One incomplete fracture was followed up, and it progressed to a nearly complete fracture over the course of 2 years (Fig. 6A, 6B).

For one patient with complete fracture, bone scintigraphy studies showed increased uptake at the fracture site in all three phases, as would be expected in the setting of acute fracture. No additional sites of uptake were detected to indicate metastatic disease. bone scintigraphy of three incomplete fractures showed increased uptake along the lateral cortex, corresponding to the site of cortical thickening (Fig. 7A, 7B).

Three of the four CT studies were performed for patients with complete fractures (Fig. 8) and revealed the typical transverse fracture pattern, skirt-like focal cortical thickening with a medial beak and fragment displacement, and no evidence for underlying lesion. The MRI studies for the incomplete fractures showed focal lateral cortical thickening and endosteal edema compatible with an incomplete fracture (Fig. 9). None of the studies revealed an underlying process. For one patient with a complete fracture (Figs. 10A and 10B), a stress fracture involving the lateral cortex was diagnosed according to an MRI performed elsewhere, 1 month before the complete fracture developed (Fig. 10C).

All (100%) of the complete fractures and seven (50%) of the 14 incomplete fractures were treated with cephalomedullary nail fixation. Follow-up for 21 (62%) of 34 fractures revealed complete healing in 3–8 months (mean, 5 months). The remaining five complete and eight incomplete fractures were either lost to follow-up or occurred so recently that assessment of healing could not be completed before this article was written.

Alendronate was the first oral bisphosphonate available in the United States for the treatment of osteoporosis. It has been proven in randomized controlled studies to significantly reduce the risk of osteoporotic fractures and to increase bone mineral density [1–3]. Thus, the paradoxical association of low-energy subtrochanteric femoral fractures associated with long-term (defined in this series as ≥ 4 years) alendronate therapy is an unexpected and only recently recognized phenomenon.

Alendendronate induces osteoclast apoptosis, thereby inhibiting osteoclast-mediated bone resorption. Although osteoclast inhibition will increase bone mineral density, in the long term, it can lead to abnormal bone remodeling and repair, allowing bone microdamage and atypical skeletal fragility [16–20]. This process will increase the fracture risk in some patients. Cheung et al. [10] described the findings on bone biopsy for a patient who sustained a low-energy alendronate-related femoral fracture. The histologic specimen depicted depressed bone formation and marked reduction in osteoid thickness and volume. Although 40% of the trabecular surface was covered with osteoid, the osteoid was very thin [10]. A recent study on skeletal fragility related to bisphosphonate therapy by Visekruna et al. [16] speculates that some patients have “physiologically vulnerable” osteoclasts, with less tolerance for bisphosphonate effects, thereby increasing the fracture risk.

The characteristic imaging features for our patients included transverse orientation of the proximal femoral fracture line, medial beak, superior displacement of the distal fragment, and varus angulation at the fracture site. In addition, the fractures were 0.5–18.3 cm away from the lesser trochanter (mean, 4.8 cm). A skirt of lateral cortical thickening and buttressing was also noted in all cases of incomplete fractures, indicating chronicity and a lateral origin of the fracture. In cases of incomplete fractures, the only finding that may be evident on radiographs is subtle focal cortical thickening of the lateral femoral cortex, although in some cases (10 of 14 fractures in our series), a faint transversely oriented fracture line can be appreciated. Our observations corroborated the findings reported in four prior series published in the orthopedic literature [5–7, 13].

Proximal femoral fractures are typically seen in the setting of major trauma, such as motor vehicle accidents, and tend to be spiral or oblique. Conversely, subtrochanteric femoral fractures associated with long-term alendronate therapy present with minimal or no history of trauma and, when incomplete, may be missed clinically or radiographically. When fractures are incomplete, their typical transverse orientation and their origin along the lateral femoral cortex resemble stress or insufficiency fractures. Other stress fractures with a transverse orientation are commonly recognized as being related to either insufficiency or fatigue, such as those within the sacrum, fibula, distal tibia, tarsal navicular, and metatarsals [21, 22]. In contrast to the alendronate-associated fractures seen in our series, most of these fractures are recognized before progression to a displaced fracture and present as areas of sclerosis with associated callus formation. Some of these fractures are also noted to occur in characteristic locations within certain bones, such as the medial femoral neck, anterior tibial shaft, or dorsal aspect of the tarsal navicular bone [22].

Because of the minimal trauma reported by patients with alendronate-associated femoral fractures, underlying pathologic processes, such as a bone tumor or metastasis, are often suspected clinically. In our study, seven cases required additional cross-sectional studies (CT or MRI) to arrive at the correct diagnosis or to exclude a pathologic fracture. Also, four patients underwent bone scanning, in one case to exclude metastatic disease (in the setting of a complete fracture) and in three cases to confirm the presence of an incomplete fracture.

Unlike a few reports in the literature [16], our patients had a fairly typical healing course not significantly different from that of other patients with traumatic proximal femoral fractures, with all but one of the fixated fractures healing within 6 months. However, in one case of incomplete fracture that was monitored for 2 years before fixation, the fracture did not heal with time and protected weightbearing.

Given the relatively new phenomenon of proximal femoral fractures associated with alendronate therapy, no globally accepted treatment has been established, but in our experience and in other case reports [8, 12], these fractures are typically treated with intramedullary nail fixation, as would conventional subtrochanteric traumatic fractures. At our institution, the orthopedic surgeons generally recommend prophylactic operative treatment of incomplete fractures, to avoid progression to complete fractures. In addition, given the high number of patients who suffer bilateral factures, radiographs of the contralateral femur are recommended for patients receiving long-term alendronate therapy who present with a subtrochanteric or diaphyseal femur fracture. If a contralateral stress fracture is found, prophylactic fixation of the femur should also be considered [13]. Furthermore, our clinical colleagues at our institution think that alendronate therapy should be discontinued, at least temporarily, in patients with proximal femoral fractures that are characteristic of those associated with alendronate. Therefore, familiarity with the characteristic clinical and imaging features of these fractures, whether complete or incomplete, is crucial for appropriate clinical management. In patients receiving long-term alendronate therapy who present with hip or femoral pain, imaging, primarily with radiography and perhaps with MR, should certainly be performed.

The association of proximal femoral fractures with long-term alendronate therapy requires future extensive study. At present, there is no consensus on the appropriate duration of alendronate therapy in osteoporotic patients and, more specifically, in patients with associated proximal femoral fractures. Also, it is unknown whether the association with low-energy subtrochanteric fractures holds true for all classes of bisphosphonates or is limited solely to alendronate therapy, because most studies to date have focused on alendronate. Two case reports have been published that describe similar fractures in patients receiving zoledronic acid (Zometa, Novartis) and risedronate sodium (Actonel, Procter & Gamble Pharmaceuticals) [23, 24]. The question whether IV infusion is a more potent form of bisphosphonate, which may lead to an increased prevalence of fractures as compared with the oral bisphosphonates, is also unknown. This is the case with osteonecrosis of the jaw. Furthermore, it is unclear which category of patients is more susceptible to the development of proximal femoral fractures and what clinical markers may herald the onset of these fractures. Regardless, the association of alendronate with femoral fractures indicates that greater caution is required in the administration and monitoring of bisphosphonate therapy. Familiarity with the characteristic clinical and imaging features of these fractures, whether complete or incomplete, is also crucial for appropriate clinical management.

The major limitations of our study are its retrospective nature and small sample size. Also, we had limited access to the clinical information on the patients. It would have been useful to correlate our findings with metabolic indexes, such as serum calcium level, phosphorus and alkaline phosphatase levels, bone mineral density, and parathyroid hormone levels. Also, the use of other drugs that may influence bone stock, such as hormone replacement therapy or steroids, was not known. Future prospective studies may be useful to assess the prevalence of these fractures in patients receiving alendronate therapy.

In conclusion, subtrochanteric femoral fractures associated with alendronate therapy have typical clinical and radiographic patterns. In complete fractures, these include minor or no trauma, initial involvement of the lateral proximal femoral cortex, transverse orientation, medial beak, skirt-like focal thickening at the opposing lateral cortical surfaces, superior displacement of the distal fragment, and varus angulation at the fracture site. Incomplete fractures show focal lateral cortical thickening with or without an incomplete transverse fracture line. Familiarity with these features will expedite detection and treatment and alert the clinician to reconsider the risks and benefits of alendronate therapy.

Address correspondence to S. S. Chan ([email protected]).