Romosozumab Achieves Consistent Lumbar Spine BMD Gains Across All BMI Categories: The Apparent Enhanced Response in Underweight Patients is a Mathematical Artifact of Percentage Change.

Body mass index (BMI) influences bone metabolism; however, its impact on romosozumab treatment response remains unclear. We investigated whether BMI independently predicts bone mineral density (BMD) response to romosozumab and re-examined the role of the outcome metric (absolute vs. percentage change) in generating the apparent BMI signal. This retrospective study analyzed 304 patients (97.4% female, mean age 76.8 years) with high-risk osteoporosis who completed 12-month romosozumab treatment (2019-2025). The primary outcome was the absolute change in lumbar spine (LS) BMD (post-treatment minus baseline, expressed in %YAM points); the percentage change was reported in parallel for comparison with the prior literature. We employed multivariate regression, mediation analysis, stratified analyses, bone turnover marker assessment including baseline TRACP-5b mediation analysis, geriatric nutritional risk index evaluation, BMI × prior treatment interaction testing, and 2-year longitudinal analysis (n = 145). When the response was expressed as an absolute change in LS-BMD, BMI showed no significant univariate association (Pearson's r = - 0.064, p = 0.267; ANOVA across five BMI categories: F = 0.54, p = 0.708), and the mean absolute gain was similar across all BMI categories (+ 7.8 to + 9.6%YAM points). Multivariate regression confirmed that BMI was non-significant (β = - 0.103, p = 0.396), while baseline LS-BMD (β = - 0.070, p = 0.020) and prior treatment (β = - 4.106, p < 0.001) were the only independent predictors. The TRACP-5b mediation analysis showed the same pattern: in the full model adjusting for both baseline LS-BMD and baseline TRACP-5b, BMI remained non-significant (β = + 0.019, p = 0.877), while both baseline LS-BMD (β = - 0.088, p = 0.005) and baseline TRACP-5b (β = + 0.008, p < 0.001) were independent predictors. Bone turnover marker changes were BMI-independent (P1NP: p = 0.831; TRACP-5b: p = 0.766). The GNRI also showed no significant association (unadjusted p = 0.198; adjusted p = 0.427). The interaction between BMI and prior treatment was not significant (p = 0.948). At 2 years, BMI remained non-predictive of cumulative absolute BMD gains (β = - 0.228, p = 0.163) and BMD maintenance during sequential therapy (β = - 0.044, p = 0.680). In contrast, when the response was expressed as a percentage change, BMI appeared to be inversely correlated with LS-BMD response (r = - 0.135, p = 0.019); this association was a mathematical artifact of the percentage-change metric, which inflates relative gains at lower baseline BMD. The apparent enhanced romosozumab response in underweight patients reflects the mathematical structure of the percentage change at lower baseline values rather than a genuine BMI-dependent pharmacological effect. Absolute BMD gains were similar across all BMI categories, and convergent evidence from multivariate regression, bone turnover marker analysis, nutritional assessment, and longitudinal follow-up confirmed the BMI-independent efficacy. Mechanistically, larger percentage gains at lower baseline BMD likely reflect greater anabolic reserve and higher baseline bone remodeling in less-mineralized bone rather than differential drug action by body habitus.