Peak inflammatory stress versus baseline neutrophil-to-lymphocyte ratio (NLR) for prediction of in-hospital mortality in critically ill glioblastoma patients: MIMIC-IV cohort analysis.

Abstract

e14103 Background: Systemic inflammation is a recognized driver of disease progression in Glioblastoma (GBM). While the Neutrophil-to-Lymphocyte Ratio (NLR) is an established oncologic biomarker. It is still ambiguous whether acute adverse outcomes are driven by a patient’s baseline inflammatory state or the magnitude of their physiological response to acute stress (surgery, clinical decline). We analyzed whether the "peak inflammatory surge" within the first 24 hours of ICU admission is a prognostic marker of in-hospital mortality compared to baseline values. Methods: we conducted a retrospective cohort study utilizing the MIMIC-IV(v3.1) database. Adult GBM patients admitted to the ICU were included. To capture the dynamic range of acute inflammation, we derived two distinct metrics from complete blood counts obtained within the first 24 hours of admission: "Baseline NLR" (minimum physiological stress: min neutrophils/max lymphocytes) and "Peak NLR" (maximum physiological stress: max neutrophils/min lymphocytes). The primary endpoint was in-hospital mortality. Due to the right-skewed distribution of inflammatory biomarkers, NLR values were natural log-transformed (ln[NLR]) for multivariate logistic regression analysis, adjusted for age. Results: A total of 242 patients met the inclusion-exclusion criteria. The overall in-hospital mortality rate was 9.9% (n=24). Baseline NLR did not differ significantly between survivors and non-survivors (median 7.71 vs 8.39; P=0.203), suggesting comparable initial physiological states. In contrast, Peak NLR was significantly high in nonsurvivors compared to survivors (median 10.28 vs 8.74; P=0.039). In univariate analysis, Peak NLR was significantly associated with mortality (OR 1.77; 95% CI 1.08–2.88; P=0.023). Upon multivariate adjustment for age, Peak NLR remained an independent predictor of mortality (OR 1.69; 95% CI 1.03–2.76; P=0.038). Conclusions: Mortality in critical GBM cases is independently associated with the magnitude of acute inflammatory surge rather than baseline status. Peak NLR may serve as a valuable prognostic and cost-effective marker for risk stratification in neuro-oncology critical care. Statistically significant clinical and laboratory predictors of in-hospital mortality. Predictor Variable Survivors (N=218) Non-Survivors (N=24) Adjusted / Unadjusted Odds Ratio (95% CI) P-value Age, median (IQR), years 61.0 (51.0–69.0) 71.5 (57.8–79.3) 1.04 (1.00–1.07) 0.018 Peak NLR, median (IQR) 8.74 (3.54–16.37) 10.28 (6.61–23.53) — 0.039 Log (Peak NLR), univariate — — 1.77 (1.08–2.88) 0.023 Log (Peak NLR), multivariate* — — 1.69 (1.03–2.76) 0.038 Age, multivariate* — — 1.03 (1.00–1.06) 0.055 Mortality rate by stress tier (%) 8.8 13.1 — 0.330 CI: Confidence Interval, IQR: Interquartile Range.