Summary
The role of integrating genomic scores (GSs) needs to be assessed. Adding a GS to recommended stratification tools does not improve the prediction of very low bone mineral density. However, we noticed that the GS performed equally or above individual risk factors in discrimination.
Purpose
We aimed to investigate whether adding a genomic score (GS) to recommended stratification tools improves the discrimination of participants with very low bone mineral density (BMD).
Methods
BMD was measured in three thoracic vertebrae using CT. All participants provided information on standard osteoporosis risk factors. GSs and FRAX scores were calculated. Participants were grouped according to mean BMD into very low (<80 mg/cm
3
), low (80–120 mg/cm
3
), and normal (>120 mg/cm
3
) and according to
the Bone Health and Osteoporosis Foundation recommendations for BMD testing
into an “indication for BMD testing” and “no indication for BMD testing” group. Different models were assessed using the area under the receiver operating characteristics curves (AUC) and reclassification analyses.
Results
In the total cohort (
n
=1421), the AUC for the GS was 0.57 (95% CI 0.52–0.61) corresponding to AUCs for osteoporosis risk factors. In participants without indication for BMD testing, the AUC was 0.60 (95% CI 0.52–0.69) above or equal to AUCs for osteoporosis risk factors. Adding the GS to a clinical risk factor (CRF) model resulted in AUCs not statistically significant from the CRF model. Using probability cutoff values of 6, 12, and 24%, we found no improved reclassification or risk discrimination using the CRF-GS model compared to the CRF model.
Conclusion
Our results suggest adding a GS to a CRF model does not improve prediction. However, we noticed that the GS performed equally or above individual risk factors in discrimination. Clinical risk factors combined showed superior discrimination to individual risk factors and the GS, underlining the value of combined CRFs in routine clinics as a stratification tool.
Supplementary Information
The online version contains supplementary material available at 10.1007/s00198-023-06857-w.