Background

Atypical femoral fractures (AFF) are low-impact fractures that differ from classic osteoporotic femoral fractures in that they occur in the subtrochanteric region or femoral diaphysis, usually in the setting of prolonged treatment with bisphosphonates (BP).While its physiopathology has not been completely clarified, genetic predisposition appears to be key given that the incidence of AFF is very low in the general population (5.9 per 10,000 person-years) [1], and specific patient populations appear to be at an increased risk of sustaining them including people with Asian ancestry and patients affected with specific monogenic bone disorders [2]. A number of studies have tried to identify the genes involved in the AFF risk suggesting that AFF occur in the setting of a complex and heterogeneous genetic background where each affected individual could have their own genetic burden [3, 4]. In a recent study by Zhou et al. (2023) [5], the authors suggested that genetic variants associated with monogenic bone disorders might play a role in the pathogenesis of AFF. In their study, 37 candidate genes involved in monogenic bone disorders were analyzed from whole-exome sequencing (WES) data in 60 AFF patients, with 95% having received bisphosphonates. Fifteen (25%) of the 60 AFF patients had clinical features of a monogenic bone disorder. In total, nine patients in their AFF cohort (15%) had a (likely) pathogenic variant, eight of whom fulfilled clinical criteria suggestive of monogenic bone disorders.

In a prior study performed by our group [6], we selected genetic variants in genes shared by at least two AFF patients and absent in controls. Hence, those variants only present in one AFF patient were removed from the analysis.

In view of the recent studies [3, 4] where each individual could have their own genetic background and furthermore, genes involved in monogenic bone disorders can play an important role, we decided to contribute with new data on 12 women with AFF and 4 controls without any fracture, all of whom had received bisphosphonates for over 5 years due to a diagnosis of osteoporosis. In the present study, we aimed to recover all rare genetic variants present in our AFF patients and controls from a list of 37 candidate genes proposed by Zhou et al. [5] and other additional genes involved in rare monogenic diseases with bone impairment.

Methods

Participants were previously described in a work published by our group [6]. Briefly, twelve unrelated postmenopausal women with AFF (mean age ± SD of 74.5 ± 6.1) and four postmenopausal women without any fracture (controls) (mean age ± SD of 79 ± 7.2) were recruited in Hospital del Mar (Barcelona, Spain) and Hospital Universitario de La Princesa (Madrid, Spain). All of them received bisphosphonate (BP) treatment for > 5 years due to a diagnosis of osteoporosis. No patient had hypophosphatemia or suspicion of a monogenic disease. Half of the AFF patients had received corticosteroid therapy for more than one year due to polymyositis, rheumatoid arthritis, asthma or chronic bronchitis. None of the controls had receive glucocorticoids.

Whole exome sequencing (WES) was performed at the CNAG facilities (Barcelona, Spain). Capture was performed using Agilent Human All Exon 50 Mb v5 and samples were sequenced at a coverage of 140x on a HiSeq 2000 sequencer. Pipeline of the WES is detailed in Garcia-Giralt et al. (2022) [6].

We filtered for rare variants (MAF < 0.005) with CADD > 20 and the resulting variants were overlapped with our list of candidate genes (Supplemental Table 1).

Results and discussion

A total of 41 genetic variants were detected in genes belonging to the list of candidate genes (Table 1). Only one of the variants identified was rated as pathogenic according to ClinVar or ACMG guidelines (p.Pro582His in DAAM2 [7]), while all others were rated as variants of uncertain significance (VUS) according to ACMG criteria.

Table 1 Genetic variants found in AFF patients and controls detected by WES
Full size table

Interestingly, variants in COL1A1 and COL1A2 were detected in control individuals (with osteoporosis and long-term BP treatment, but without AFF), suggesting their putative involvement in the underlying osteoporosis phenotype. The same could be proposed for the FKBP10 and TNXB genes, which were found mutated in controls and AFF patients. Zhou et al. [5] found likely pathogenic variants and VUS in COL1A1 and COL1A2 genes related to a diagnosis of osteogenesis imperfecta or monogenic osteoporosis in 5 AFF patients, which could suggest a role of these genes in both bone tissue pathology and AFF.

Out of 33 genetic variants identified in women with AFF, eleven (33.3%) were found in genes belonging to the Wnt pathway (LRP5, LRP6, DAAM2, WNT1, and WNT3A). Moreover, the AFF11 woman was homozygous for a mutation in DAAM2. It is difficult to discern whether these genes play a role in the pathophysiology of AFF in addition to their known role in low bone mass. Similarly, LRP5 was also found mutated in 2 AFF patients with a diagnosis of monogenic osteoporosis in the Zhou et al. study [5] suggesting this putative dual role. On the other hand, DAAM2 gene was not assessed in the AFF patients from that study and we cannot know its involvement in bone phenotypes.

Subject AFF1 did not carry any mutations from our list of candidate genes while all other AFF patients were carriers of variants in more than one gene. For example, subject AFF11 accumulated up to 9 rare genetic variants, one of them in homozygosity (DAAM2). Unfortunately, there is no information about this latter variant in the ClinVar database. Interestingly, Daam2 KO mice showed a marked reduction in bone strength, despite minimal changes in bone morphology and mineral content, indicating an abnormal bone composition and structure explained in part by cortical impairment [8].

While none of our patients could be confidently diagnosed of any monogenic bone disorder, like most of the Dutch AFF cohort [5], all shared the common feature of severe postmenopausal osteoporosis that required long-term BP treatment and, in some cases, also denosumab. Noteworthy, 50% of the women with AFF in our cohort and none of the controls had been on long-term glucocorticoid treatment which is a recognized risk factor for AFF [9]. When comparing both cohorts according to the gene candidate list proposed by Zhou et al. [5] (see supplemental Table 1) and considering all AFF patients with or without a clinical suspicion of monogenic bone disorders, a similar number of carriers of variants in Mendelian bone disease genes was detected (46% in Zhou et al. vs. 50% in this cohort).

Conclusion

Encompassing all of our findings we conclude that in our cohort we did not detect a major gene involved in AFF pathophysiology. We speculate that AFF development is probably the result of the sum of genetic variants together with other structural, physiological and environmental factors. If osteoporosis, rare bone diseases, and AFFs share the same genes, identifying unique risk factors could be even more challenging.

These results warrant further studies of genes related to monogenic bone disorders in the setting of severe osteoporosis, in addition to their potential role in AFF pathogenesis.