Neurocognitive and neuropsychiatric presentation identifies a higher-risk clinical phenotype in IDH-wildtype glioblastoma: A retrospective cohort study

Objective In IDH-wildtype glioblastoma, prognosis is conventionally interpreted through established molecular, clinical, surgical and anatomical factors, whereas the prognostic meaning of the mode of onset is less defined. We hypothesized that neurocognitive/neuropsychiatric presentation identifies an adverse clinical phenotype, potentially reflecting diagnostic delay and/or early disruption of distributed cerebral networks. Methods We retrospectively analyzed 165 patients with IDH-wildtype glioblastoma treated with standard chemoradiation at a single institution. The main presenting symptom was reviewed from clinical documentation and dichotomized as neurocognitive/neuropsychiatric versus all other presentations. The primary endpoint was progression-free survival (PFS), and overall survival (OS) was the secondary endpoint. Multivariable Cox models were adjusted for prespecified covariates: MGMT promoter methylation, extent of resection, baseline Karnofsky Performance Status, age, and subventricular zone contact. Results Thirty-one patients (18.8%) had a neurocognitive/neuropsychiatric onset. In unadjusted analysis, PFS showed a non-significant separation (log-rank p = 0.175), and OS curves did not differ (log-rank p = 0.896). After adjustment, neurocognitive/neuropsychiatric onset was associated with shorter PFS (HR 1.77, 95% CI 1.13-2.77; p = 0.012). The OS estimate was concordant and reached statistical significance in the secondary multivariable model (HR 1.71, 95% CI 1.02-2.87; p = 0.041). The PFS association was consistent across sensitivity analyses. Conclusion Neurocognitive/neuropsychiatric onset identified a higher-risk presentation phenotype associated with shorter PFS and, in secondary analyses, shorter OS after adjustment for established prognostic factors. These hypothesis-generating findings suggest that clinical mode of onset may capture prognostic information not fully reflected by conventional variables and warrant prospective multicenter validation.