CT-Based Lesion Volume as an Independent Predictor of Surgical Recurrence in Medication-Related Osteonecrosis of the Jaw: A Multi-Center Study of 1007 Patients

Background/Objectives: To determine whether preoperative CT-measured lesion volume independently predicts surgical recurrence of MRONJ, and whether surgical approach influences recurrence after adjusting for lesion volume and clinical confounders. Methods: We conducted a retrospective multi-center cohort study of 1007 patients with single-lesion MRONJ who underwent surgery at five tertiary institutions in the Republic of Korea (2015–2023), with a minimum two-year follow-up. Preoperative lesion volume was measured by CT segmentation (Osteomyelitis Analyzer SW, version 2.0.6; Aemasue, Seoul, Republic of Korea) with cross-institutional consensus validation. Surgical recurrence was defined as re-fulfillment of 2022 AAOMS criteria after confirmed mucosal healing (Hayashida H1). Three logistic regression models of increasing adjustment were constructed a priori; reporting adhered to STROBE. Results: Recurrence occurred in 108 patients (10.7%) and increased in a dose–response pattern across volume quartiles (Q1 5.6%, Q2 9.2%, Q3 12.0%, Q4 16.0%; Cochran–Armitage trend Z = 3.89, p < 0.001), with a 2.9-fold relative gradient (absolute difference 10.4 percentage points). Surgeons preferentially selected aggressive approaches for larger lesions (OR = 1.651; 95% CI 1.338–2.037; p < 0.001). After adjustment, surgical approach was not an independent predictor in any model (fully adjusted OR = 0.735; 95% CI 0.381–1.419; p = 0.360), whereas lesion volume remained independent throughout (OR = 1.543–1.625 per log-unit; all p < 0.001; c-statistic 0.625–0.699). Conclusions: CT-measured preoperative lesion volume is an independent predictor of MRONJ surgical recurrence; we did not detect an independent surgical-approach effect, although the sample could not exclude a clinically meaningful one. Among measured factors, lesion volume—rather than surgical approach—was the principal predictor of recurrence risk, supporting consideration of routine CT volumetry in preoperative assessment, pending prospective validation.