Association of computed tomography (CT)-derived muscle area and density at multiple sites with 10-year fracture risk in older men

Loss of muscle quantity and quality with age has been related to fracture risk independent of areal BMD (aBMD) assessed by DXA. Automated approaches to assess muscle on CT images make it more feasible to assess associations of muscle quantity (cross-sectional area) and quality (density) of multiple anatomical regions on fracture risk. We investigated whether automated analysis of muscle area and density at the trunk (L1 and L3) and proximal thigh (right and left) predicted 10-yr fracture risk (any clinical, clinical spine, and non-spine) independent of DXA aBMD as well as CT vBMD and muscle function in older men from the Osteoporotic Fractures in Men (MrOS) study. Men with CT imaging and complete covariate measures were included (<i>n</i> = 3237, 73.7 ± 5.9 yr). Fractures were centrally adjudicated. Proportional hazards models assessed relationships, adjusting for DXA FN aBMD, CT FN vBMD, CT LS vBMD, and muscle function. For any fracture, a 10%-25% and 8%-12% risk reduction was observed per 1-SD increment in muscle area and density, respectively, across anatomical locations, though HR estimates only reached significance for muscle area at the thigh and muscle density at the Trunk-L3 (<i>p</i> > .05), despite overlapping 95% CIs across anatomical sites. Similar reductions were seen for non-spine fractures. For spine fracture, only Trunk-L3 muscle density had a significant 31% risk reduction. For hip fracture, muscle area at the right (40%) and left thigh (35%) and muscle density at Trunk-L3 (25%) were associated with reduced risk (<i>p</i> < .05). Results remained unchanged or slightly attenuated after adjusting for muscle function and BMD, respectively. Automated measures of muscle area and density on CT images may enhance fracture risk assessment in older men that does not vary substantially across different anatomic locations. The relative importance of muscle area and density for fracture risk prediction varies across fracture types.