Genetic Diagnostics in Routine Osteological Assessment of Adult Low Bone Mass Disorders

Abstract

Context: Many different inherited and acquired conditions can result in premature bone fragility/low bone mass disorders (LBMDs).

Objective: We aimed to elucidate the impact of genetic testing on differential diagnosis of adult LBMDs and at defining clinical criteria for predicting monogenic forms.

Methods: Four clinical centers broadly recruited a cohort of 394 unrelated adult women before menopause and men younger than 55 years with a bone mineral density (BMD) Z-score < -2.0 and/or pathological fractures. After exclusion of secondary causes or unequivocal clinical/biochemical hallmarks of monogenic LBMDs, all participants were genotyped by targeted next-generation sequencing.

Results: In total, 20.8% of the participants carried rare disease-causing variants (DCVs) in genes known to cause osteogenesis imperfecta (COL1A1, COL1A2), hypophosphatasia (ALPL), and early-onset osteoporosis (LRP5, PLS3, and WNT1). In addition, we identified rare DCVs in ENPP1, LMNA, NOTCH2, and ZNF469. Three individuals had autosomal recessive, 75 autosomal dominant, and 4 X-linked disorders. A total of 9.7% of the participants harbored variants of unknown significance. A regression analysis revealed that the likelihood of detecting a DCV correlated with a positive family history of osteoporosis, peripheral fractures (> 2), and a high normal body mass index (BMI). In contrast, mutation frequencies did not correlate with age, prevalent vertebral fractures, BMD, or biochemical parameters. In individuals without monogenic disease-causing rare variants, common variants predisposing for low BMD (eg, in LRP5) were overrepresented.

Conclusion: The overlapping spectra of monogenic adult LBMD can be easily disentangled by genetic testing and the proposed clinical criteria can help to maximize the diagnostic yield.

Keywords: genetic risk score; genotype-phenotype correlation; low bone mass disorder; monogenic disorder; osteoporosis; rare genetic variant.

Figure 1.

Statistical analysis of clinical thresholds predicting monogenic forms of adult LBMDs. Regression tree analysis of the performed classification showing clinical thresholds determining DCV frequency, starting from the entire cohort of n = 394. Clinical criteria and P values are shown in circles; boxes show percentages and total numbers of LBMD individuals with a DCV falling under the respective criteria. Thresholds are peripheral fractures > 10, family history for fractures/osteoporosis, peripheral fractures > 2. Further subdivision followed according to BMI. BMI, body mass index; DCV, disease-causing variant; LBMD, adult low bone mass disorder.

Figure 2.

Role of common BMD-associated and rare deleterious variants in LBMD individuals without monogenic DCV. (A) Comparison of BMD predicted by a GRS relative to healthy controls between LBMD without DCV (LBMD noDCV: 312), the LBMD with DCV (LBMD DCV: n = 82), the 1KG EUR control cohort (1KG EUR: n = 503), and individuals with high BMD (high BMD: n = 56). (B) ROC curves of GRSs of common variants associated with BMD. The GRS of individuals with LBMD without DCV (LBMD noDCV) was compared with 1KG EUR (red) and high BMD (blue). An AUC of 0.5 (dashed line) shows a random partitioning. Full partitioning using GRS would be achieved at an AUC of 1.0. (C) The frequency of one of the strongest BMD-influencing common variants, SNP rs4988321, in LRP5 is significantly increased in LBMD noDCV compared with controls (*P < 0.05, **P < 0.01). 1KG EUR, 1000 Genomes Project; AUC, area under the curve; BMD, bone mineral density; DCV, disease-causing variant; GRS, genetic risk score; LBMD, low bone mass disorder; ROC, receiving operating characteristic.