In silico clinical trial to predict the efficacy of alendronate for preventing hip fractures

Hip fractures associated with osteoporosis are a major healthcare concern. Existing drugs have limited efficacy in reducing hip fractures. However, clinical trials require large cohorts and lengthy follow-up. Computational models could potentially improve the development of more effective treatments. The aim of this study was to validate an In Silico Trial (BoneStrength) by reproducing a published clinical trial on the efficacy of alendronate. The predicted number of fractures was compared to the clinical data.

A statistical atlas was used to generate a virtual cohort (N = 1050), with baseline characteristics replicating the reference trial. Treatment with alendronate was simulated by increasing bone mineral density over time. Fracture incidence was predicted using a Markov Chain process. The impact force associated with each fall was estimated with a multiscale stochastic model. Finite Element models were used to predict femur strength. A patient was considered fractured when the impact force exceeded femur strength.

In the placebo group, virtual patients (N = 1050) experienced 15 ± 4 hip fractures in four years, whereas in the reference trial 24 occurred for 2218 patients (11 for 1050 patients). In the alendronate arm, fractures were reduced to 10 ± 3 in the virtual cohort, while 19 were observed in 2214 patients (9 for 1050 patients).

The distribution of hip fracture incidence predicted by the model included the clinical data for both groups. This In Silico trial can be applied in the future to improve clinical trial design and drug development, enabling a virtual pathway to the efficacy assessment of bone drugs.