Abstract
BACKGROUND: Elderly high-grade glioma (HGG) or glioblastoma (GBM) patients face challenging treatment conditions due to comorbidities and age-related factors. The age-adjusted Charlson Comorbidity Index (ACCI) accounts for age and comorbidities and serves as a tool for predicting survival rates in various clinical scenarios. This study examined its prognostic value in elderly HGG patients undergoing radiotherapy (RT) and concurrent chemoradiotherapy (CRT). METHODS: We retrospectively analyzed 163 elderly HGG patients (≥ 60 years) treated with radiotherapy (RT) or chemo-RT (CRT) at TUM University Hospital (2001-2021). Kaplan-Meier analysis estimated median overall survival (OS) by ACCI group (≤ 5 vs. ≥6). Multivariate Cox regressions assessed OS and progression-free-survival (PFS) based on fractionation and treatment strategies. Further Cox models evaluated ACCI scores, age, comorbidities, and mortality. A random survival forest (RSF) identified key survival predictors, using permutation importance with bootstrapped confidence intervals. RESULTS: Among the 163 HGG patients, those with greater comorbidities (ACCI ≥ 6) had a shorter median OS (14.8 months) than did those with ACCI ≤ 5 (22.6 months) (log-rank p = 0.463). In the ACCI ≤ 5 subgroup, hypofractionated RT (hRT) alone was significantly associated with worse OS than Stupp was (HR = 85.7, 95% CI: 7.1-914.3, p = 0.0004), whereas no significant differences were detected in the ACCI ≥ 6 subgroup. Hypofractionated RT was associated with improved PFS in patients with an ACCI ≥ 6 (HR = 0.47, 95% CI: 0.24-0.92, p = 0.027), and MGMT methylation better predicted OS (HR = 0.31, p = 0.0039) and PFS (HR = 0.32, p = 0.0059). Diabetes without complications independently predicted worse OS (HR = 2.91 (95% CI: 1.63-5.18, p < 0.001)) and PFS (HR = 2.59 (95% CI: 1.43-4.70, p = 0.002), with a significant interaction between diabetes and the ACCI (HR = 0.26, 95% CI: 0.07-0.91, p = 0.03). RSF models identified age as the key predictor of OS and MGMT methylation as the main predictor of PFS, while ACCI ≥ 6 contributed only modestly (mean drop for OS: 0.025; and PFS: 0.019). CONCLUSIONS: The ACCI showed limited and inconsistent prognostic value in elderly glioblastoma patients, while diabetes emerged as the only consistent comorbidity predictor of OS and PFS. These findings suggest that comorbidity burden may influence outcomes but underscore the need for larger studies to clarify the role of the ACCI in treatment stratification.