Bilateral Fractures of the Femur Diaphysis in a Patient With Rheumatoid Arthritis on Long-Term Treatment With Alendronate: Clues to the Mechanism of Increased Bone Fragility

Abstract

Unusual fractures of the femur diaphysis have been reported in patients treated with alendronate and, although no causal relationship has been established, excessive suppression of bone turnover and length of treatment with alendronate have been implicated in their pathogenesis. We report here clinical, biochemical, and radiological findings of a patient with rheumatoid arthritis and multiple risk factors for fractures who was treated with alendronate for 8 yr and developed spontaneous bilateral subtrochanteric/diaphyseal fractures. Bone biopsies obtained form the iliac crest and the femur showed decreased bone formation with histomorphometric evidence of markedly increased bone resorption at the femur. These results show for the first time that an imbalance between bone resorption and bone formation at the affected bone is associated with the occurrence of these atypical femur fractures. The cause of this imbalance is currently unknown, and further studies of the epidemiology and pathogenesis of diaphyseal femur fractures are warranted.

INTRODUCTION

Bisphosphonates are widely used in the management of osteoporosis. They decrease bone resorption and turnover, increase BMD, maintain or improve structural and material properties of bone, and thereby reduce the risk of fractures.⁽¹⁾ It has been suggested that the long-term residence time of bisphosphonates in bone and their ability to decrease bone turnover for prolonged periods may eventually impair the ability of bone to remodel, leading to accumulation of microdamage and increased bone fragility.⁽²⁾ Although no adverse effects on bone strength have been reported in animal and controlled human studies after prolonged exposure to alendronate,^(3,4) a number of case studies have raised concerns regarding a possible deleterious effect of the bisphosphonate on bone fragility.^(5–10) These reports described unusual subtrochanteric/diaphyseal femur fractures in patients treated with alendronate and suggested that excessive suppression of bone turnover and length of treatment may be involved in their pathogenesis, although no causal relationship was shown.

We present here detailed observations of a patient with rheumatoid arthritis on long-term alendronate treatment and bilateral subtrochanteric/diaphyseal femur fractures that help characterize the clinical pattern and potential risk factors for such fractures and show surprising findings regarding their possible pathogenesis.

CASE REPORT

A 76-year-old woman was admitted to the hospital in June 2007 with a 2-mo history of persistent pain of the left thigh without any history of trauma. The pain was worse with standing and did not respond to analgesics, and the patient could hardly walk. Rheumatoid arthritis was diagnosed in 1985 and was originally treated with diclofenac, omeprazol, gold injections, and penicillamine. In March 1996, she was started on prednisone 10 mg/d, which was reduced to 5 mg/d in 1999 concurrently with the start of methotrexate 15 mg once weekly. In 2004, infliximab was added to her treatment (200 mg, IV, every 2 wk for a total of 8 wk); this course was repeated three times. In August 1999, she was also started on oral alendronate initially 10 mg/d and later 70 mg once weekly for prevention of glucocorticoid-induced bone loss. These medications were continued up to the time of admission.

Clinical examination

On admission, she had an inability to raise her left leg, and there was a palpable painful area in the lateral side of the proximal left femur. Knee and hip movements were free and painless, and there were no signs of proximal myopathy. Neurological examination was unremarkable, and there were no abnormal findings in the right leg, no signs of active arthritis, and no physical dysfunction related to rheumatoid arthritis.

Laboratory investigations

Laboratory investigations showed no abnormalities erythrocyte sedimentation rate (ESR) was 10 mm the first hour and serum C-reactive protein (CRP) was 6 mg/liter. Serum calcium was 2.40 mM, phosphate was 1.2 mM, alkaline phosphatase was 45 U/liter, creatinine was 80 μM, 25-hydroxyvitamin D was 42 nM, urinary calcium excretion was 3.1 mmol/24 h, and the urinary hydroxyproline/creatinine ratio was 13.5 μmol/mmol (all results within reference ranges).

Radiographs and bone scintigraphy

Plain radiographs of the left femur showed thick cortices of the femur diaphysis with a discrete thickening of the cortex of the outer side but no evidence of fracture (Fig. 1A). Bone scintigraphy combined with single photon emission CT (SPECT) showed a single area of increased uptake of the radioisotope in the subtrochanteric region of the left femur at the site of the cortical thickening (Figs. 1B and 1C). The appearance of the right femur was normal. These images were compatible with a microfracture (stress fracture) of the left femur.

Figure Figure 1. Radiographs of the left (A, D, and E) and the right (F) femur, bone scintigraphy (B), and SPECT (C) (see text for details).

BMD

BMD measured by DXA (Hologic 4500) was normal: L₂–L₄, 0.985 g/cm² (T-score = −0.65); right femoral neck, 0.740 g/cm² (T-score = −1.0); right total hip, 0.927 g/cm² (T-score = −0.1).

During the investigations and before any therapeutic action, the patient spontaneously fractured her left femur (Fig. 1D). The fracture was fixed with a γ nail, and a biopsy taken at operation showed no evidence of malignancy. Methotrexate was stopped, and the patient continued treatment with prednisone, diclofenac, omeprazol, and alendronate. Radiographs of the femur taken 6 mo after the operation showed delayed union and hardly any evidence of callus formation (Fig. 1E).

In February 2008, the patient spontaneously fractured her right femur (Fig. 1F) without any prodromal symptoms except pain and the inability to stand immediately before the fracture. Alendronate was eventually stopped. Biochemical markers of bone turnover in serum were normal: carboxy-terminal cross-linked telopeptide of type I collagen (CTX), 0.300 ng/ml; procollagen type I N-terminal propeptide (PINP), 50 ng/ml.

Bone biopsies

Bone histomorphometry

Bone biopsies were obtained perioperatively from the iliac crest and the right femur ∼1 cm above the fracture.⁽¹¹⁾ Undecalcified sections were obtained at three different levels separated by 200 μm. Two sections per plane were collected and stained with Goldner's trichrome. Histomorphometric parameters were measured with an automatic (Osteolab; Explora Nova, La Rochelle, France) or semiautomatic (Tablet' Measure; Explora Nova) image analyzer. Bone structure and microarchitecture were assessed by measuring cortical thickness (Ct.Th; μm), cortical porosity (Ct.Po; %), trabecular thickness (Tb.Th; μm), trabecular number (Tb.N; /mm), and trabecular separation (Tb.Sp; μm). Bone formation was evaluated by osteoid volume/bone volume (OV/BV; %) and osteoid surface/bone surface (OS/BS; %) and bone resorption by eroded surfaces (ES/BS; %) and number of osteoclasts per unit of bone surface (OcN/BS; /mm).

Histomorphometric indices of cancellous bone from the iliac crest and cortical bone from the femur are shown in Table 1. Cancellous bone parameters of structure and microarchitecture were normal compared with age- and sex-matched controls. The bone formation parameters in the iliac crest were highly decreased but were within the previously described range in patients with postmenopausal and glucocorticoid-induced osteoporosis treated with alendronate.^(12,13) In contrast, the eroded surfaces were markedly increased: these were ∼3-fold higher than controls and 6.5- to13-fold higher than after alendronate treatment of glucocorticoid-induced and postmenopausal osteoporosis, respectively. The number of osteoclasts was also increased, being ∼4-fold higher than that previously reported after alendronate treatment. Despite the absence of control values for femoral cortex, the eroded surfaces appeared extended with numerous osteoclasts. The osteoclast number was six times higher in the femoral cortex than in iliac cancellous bone. They had all the morphological characteristics of active osteoclasts,⁽¹⁴⁾ contained several nuclei, and were in contact with the crenated bone surfaces (Figs. 2A and 2B). Only a few of them contained >15 nuclei (Figs. 2C and 2D). These osteoclasts were different from the giant apoptotic osteoclasts reported after treatment with alendronate,⁽¹⁵⁾ which have a lightly stained cytoplasm, contain a high number of pyknotic nuclei, and are detached from a shallow resorption cavity. An example of such an osteoclast is depicted in Fig. 3 for comparison. As in cancellous bone, the static bone formation parameters were low.

Table Table 1.. Bone Histomorphometry

Figure Figure 2. Bone biopsy of the right femur; undecalcified section stained with Goldner's trichrome: (A and B) magnification, ×200 and (C and D) isolated osteoclasts.

Figure Figure 3. Giant hypernucleated osteoclast from a woman with osteoporosis after treatment with alendronate.

Mineralization of bone tissue

The degree of mineralization of cortical bone from the femur was measured by quantitative microradiography on 100-μm-thick sections. Microradiographs were performed, and the mean degree of mineralization of bone measured on microradiographs was derived from the quantitative evaluation of the absorption of X-rays, as previously described.^(16,17) Additional evaluations of mineralization index, mineral maturity, and crystallinity index were performed by Fourier transform infrared microspectroscopy (FTIRM).

The mean degree of mineralization of bone tissue in the femur was 1.13 g/cm³. A comparison with values measured with the same technique in iliac crest specimens of control elderly subjects (1.11 ± 0.08 g/cm³), osteoporotic women (1.08 ± 0.08 g/cm³), or women treated with alendronate (1.185 ± 0.016 g/cm³) showed no evidence of “hypermineralized” bone.⁽¹⁶⁾ The mineralization index in new bone of our patient, measured by FTIRM, was higher than that of controls (2.389 versus 1.988) but similar to that of a patient treated with alendronate for 3 yr (2.380). The values of the treated patients were very similar to those of old bone of control subjects (2.638). Mineral maturity of new bone was higher in our patient (2.519) than that of controls (1.738) and of a patient treated with alendronate for 3 yr (2.128). Crystallinity index was very similar in the three groups (controls, 25.166; alendronate for 3 yr, 25.934; our patient, 25.231). These results indicate that there is an increase in the mineral maturity after alendronate treatment without any modifications in the crystallinity index.

DISCUSSION

The patient described here had all the features previously described in patients with subtrochanteric/diaphyseal femur fractures treated with alendronate: prodromal pain complaints in the thigh, progress to spontaneous transverse fracture without trauma, and radiographic appearances consisting of thickening of the diaphyseal cortices with beaking of the cortex on one side. As shown here, bone scintigraphy can assist in the early diagnosis of the condition. Importantly, at the time of the investigation, the other femur was not affected. Our patient had numerous risk factors for fractures, the most important being rheumatoid arthritis and use of glucocorticoids.⁽¹⁸⁾ Long-term use of omeprazol has also been reported to increase the risk of fractures, particularly those of the hip,⁽¹⁹⁾ whereas methotrexate treatment has been associated with increased fracture risk, and cases with a presentation similar to that of our patient and insufficiency fractures of the lower extremities have been described during methotrexate treatment.^(20,21) It is also important to note that three of five cases of postmenopausal women with diaphyseal fractures associated with alendronate treatment and described in detail were on long-term prednisone treatment.^(5,6) Strikingly, in all these cases, as in the patient reported here, BMD was either normal or osteopenic.

The association of subtrochanteric/diaphyseal fractures with alendronate treatment is intriguing. Alendronate decreases the risk of vertebral and nonvertebral fractures,⁽²²⁾ and greater reductions of bone turnover are associated with better antifracture efficacy at the hip.⁽²³⁾ Moreover, in control studies of postmenopausal women with osteoporosis extending to 10 yr, there was no evidence of an increased incidence of nonvertebral fractures.^(3,4) Thus, although, in theory, suppression of bone turnover, particularly of cortical bone with already low remodeling rates, may impair the ability of bone to remove microdamage, resulting in its accumulation and consequently in impairment of its strength, evidence for that is lacking. However, in two studies of bone biopsies taken from the iliac crest from patients with diaphyseal fractures on alendronate treatment, excessively decreased bone turnover was shown. It should be noted that similar rates of bone turnover have been previously reported in glucocorticoid-induced osteoporosis, healthy premenopausal women, and patients with osteoporosis treated with alendronate and no diaphyseal fractures. Moreover, in all previous reports, including ours, biochemical markers of bone turnover representing total skeletal turnover were not excessively decreased. The question is whether these fractures represent part of the spectrum of osteoporotic fractures or whether they are a separate entity representing an uncommon side effect of alendronate treatment. Subtrochanteric/diaphyseal fractures account for ∼10% of femur fractures, and a recent large case-control study did not find any difference in their incidence among patients with a history of fractures treated or not with alendronate.⁽²⁴⁾ The contribution of other risk factors, as in the patient described here, may be important.

We had the opportunity to obtain biopsies from the iliac crest and the affected femur. This is the first time that such biopsy has been obtained from a patient with diaphyseal fractures on alendronate and analyzed histomorphometrically. Consistent with previous reports, static parameters of bone formation were markedly decreased in both specimens, being, however, within the range observed in patients with osteoporosis treated with alendronate, as well as in patients treated with glucocorticoids that primarily decrease bone formation. However, this decrease in bone formation was not coupled to a decrease in bone resorption as expected by antiresorptive therapy. Bone resorption parameters (eroded surfaces and osteoclast number) were higher than those measured in osteoporotic patients treated with alendronate and were markedly increased in the femur specimen. The increase in osteoclast numbers in the femur was not caused by the action of the bisphosphonate, because these cells were attached to the bone surface and were morphologically active. The possibility that these were caused by the fracture itself is also unlikely because the biopsy was taken 1 cm above the fracture within 12 h of the event, and there were no prodromal symptoms suggestive of a stress reaction. In our patient, there was, therefore, a strong imbalance between bone formation and bone resorption at the tissue level in favor of the latter. Elucidation of the cause of this imbalance, particularly at the femur, can help in the identification of patients at risk for such fractures. In addition to histomorphometry, we examined, also for the first time, intrinsic material properties of the bone tissue that may affect bone strength and can be modified by bisphosphonate treatment. Our results showed no evidence of hypermineralized bone and no change in the size and perfection of the crystals. Crystals were more mature than in control subjects, consistent with the effect of alendronate on bone turnover and secondary mineralization.⁽¹⁷⁾

In conclusion, our study provides for the first time evidence that an imbalance between bone resorption and bone formation at the affected femur rather than excessive suppression of bone turnover may be the mechanism underlying the occurrence of subtrochanteric/diaphyseal femur fractures in patients treated with alendronate. The cause of this imbalance is currently unknown and, until the epidemiology and pathogenesis of these fractures is elucidated, it is advisable to discontinue alendronate in patients with normal BMD values on long-term glucocorticoid treatment.

Acknowledgements

We thank Audrey Doublier and Yohann Bala for expert technical assistance.