Microbiological profile and prognostic significance of infections following radiotherapy in glioblastoma

Infections arising after radiotherapy for glioblastoma are insufficiently characterized. We aimed primarily to describe the microbiological profile and antimicrobial resistance patterns of documented culture-positive infections and secondarily to explore the time-updated association between microbiological ascertainment and overall survival (OS). This retrospective cohort included 152 adults with glioblastoma who completed radiotherapy. For descriptive analyses, patients were categorized during the 12 months after radiotherapy as having a culture-positive infection, a culture-negative evaluation, or no culture obtained. OS was measured from radiotherapy completion. In a secondary exploratory analysis, microbiological ascertainment state was modeled as a three-level time-dependent categorical covariate, with sampled patients transitioning at the index date. Baseline variables associated with OS at p < 0.05 in univariable Cox models were entered into the multivariable model. A 30-day lagged sensitivity analysis delayed state transitions to evaluate whether the association was driven by deaths occurring shortly after culture collection. Of 152 patients, 48 (31.6%) had a documented culture-positive infection, 22 (14.5%) had a culture-negative evaluation, and 82 (53.9%) had no culture obtained. Respiratory ( n = 16) and urinary tract infections ( n = 15) were most frequent, and 41 of 60 isolates were Gram-negative. In the multivariable time-dependent Cox model, culture-positive infection was associated with higher subsequent mortality than the no-culture-yet state (adjusted hazard ratio [aHR] 6.60, 95% confidence interval [CI] 4.38–9.93; p < 0.001), whereas culture-negative evaluation was not (aHR 1.65, 95% CI 0.96–2.83; p = 0.071). Mortality was also higher in the culture-positive than in the culture-negative state (aHR 4.01, 95% CI 2.22–7.22; p < 0.001). The culture-positive association remained significant after a 30-day lag (aHR 3.36, 95% CI 2.09–5.39; p < 0.001). The descriptive microbiological findings, characterized by predominantly Gram-negative, respiratory or urinary, and frequently healthcare-associated infections, represent the most robust contribution of this study. In a secondary time-dependent analysis, documented culture-positive infection was a marker of increased subsequent mortality. Because sampling was indication-driven, with residual confounding and reverse causation still possible, the survival association should be considered hypothesis-generating rather than causal.