Development of an Algorithm to Predict Appendicular Lean Mass Index From Regional Spine and Hip Dxa Scans

Abstract

Introduction: Sarcopenia is characterized by progressive muscle loss with reduced physical function and/or reduced muscle strength. Operational definitions of sarcopenia include a measurement of muscle mass, most often from dual-energy X-ray absorptiometry (DXA)-derived appendicular lean mass. Appendicular lean mass can be derived from whole-body dual-DXA scans; however, these scans are performed less commonly than hip and spine scans as part of clinical care. The objective of our study was to develop an algorithm to predict appendicular lean mass index (ALMI) from regional spine and hip dual-energy X-ray absorptiometry (DXA) scans.

Methods: We performed a retrospective cross-sectional study using a subset of patients from the Manitoba Bone Mineral Density Registry who had hip, spine, and whole-body DXA scans at the same visit. We developed the algorithm using the following candidate covariates: age, sex, height, weight, DXA-derived spine and hip fat fraction, DXA-derived spine and hip tissue thickness. We internally validated the algorithm using the bootstrap method. Mean bootstrap parameter estimates were used as the final equation.

Results: DXA scans from 676 patients were included in the analytic dataset. Mean ALMI was 6.73 (SD 1.43) kg/m². The final predictive model included sex, age, height, weight, spine fat fraction and hip fat fraction. Sex also acted as an interaction term on weight and hip fat fraction. After bootstrap validation, model adjusted R2 was 0.863, root mean square error was 0.529 kg/m², and AUROC to predict low ALMI per the European Working Group on Sarcopenia version 2 was 0.88.

Conclusion: Hip and spine DXA scans can be used to predict appendicular lean mass index. Future studies should test whether these predictions can be used to assess relationships between sarcopenia and other clinical conditions.