Partial lumbar spine response in hip non-responders to romosozumab: Baseline bone mineral density as the principal determinant in a multi-center real-world Japanese cohort (n = 398).

Abstract

Romosozumab produces substantially greater bone mineral density (BMD) gains at the lumbar spine than at the total hip, with approximately half of treated patients failing to achieve a least significant change of +3% at the hip - the priority site in contemporary treat-to-target frameworks. The determinants of this site-discordant response and its clinical interpretation remain incompletely characterized in real-world Japanese practice. We retrospectively analyzed 398 Japanese patients (mean age 76.7 ± 8.7 years; 94.5% female) who completed 12 monthly doses of romosozumab at Mutsu General Hospital (n = 305) and Towada City Hospital (n = 93) between April 2019 and April 2025. The ±3% least significant change threshold was applied separately to the lumbar spine and total hip BMD changes to define four response patterns. Three parallel multivariable logistic regression analyses identified independent predictors of (i) non-response, (ii) full versus partial response, and (iii) partial versus bilateral non-response. Twenty-four-month outcomes were assessed in 202 patients. Hip non-response occurred in 196 patients (49.2%); among them, 169 (86.2%) demonstrated a partial response with concomitant lumbar spine gain ≥ + 3%, while only 27 patients (6.8% of the full cohort) exhibited bilateral non-response. In the multivariable analyses, the principal independent predictors of hip non-response were baseline total hip BMD (odds ratio 1.04 per %YAM; 95% CI 1.02-1.07; p = 0.001), baseline lumbar spine BMD (OR 1.02; p = 0.020), age (OR 1.04; p = 0.020), and body mass index (OR 0.93; p = 0.033), collectively consistent with a baseline-BMD ceiling effect. Baseline 25-hydroxyvitamin D levels and prior denosumab exposure were not independent predictors (p = 0.218 and p = 0.391, respectively). Among 85 patients with 12-month partial response who completed the 24-month follow-up, hip BMD recovered to ≥ + 3% in 39 (45.9%). In this real-world, multi-center Japanese cohort, the principal determinants of hip non-response to romosozumab were baseline BMD, age, and BMI, consistent with a ceiling effect, rather than prior anti-resorptive exposure. Most hip non-responders nonetheless gained BMD at the lumbar spine, and approximately half subsequently recovered at the hip by 24 months, supporting the interpretation of hip non-response as predominantly delayed rather than absent.