Nuclear Eccentricity as a Potential Biomarker for Tumor Biology and Survival in Glioblastoma

Purpose: To investigate whether quantitative nuclear morphology from routine H&E images can be used as a biomarker for glioblastoma (GBM) biology and prognosis. Methods: Nuclear eccentricity, circularity, density, and heterogeneity were measured from H&E whole-slide images in the TCGA-GBM cohort and independently validated in the CPTAC-GBM cohort. A multivariable Cox model was used to test the survival associations in both cohorts, and transcriptome-wide and gene set enrichment analyses were conducted in TCGA RNA data and validated using CPTAC RNA and proteomic data. Results: In TCGA-GBM, greater nuclear eccentricity was associated with better survival (HR, 0.85; 95% CI, 0.71-1.02; p = 0.089), with the same favorable association independently observed in CPTAC-GBM (HR, 0.74; 95% CI, 0.60-0.90; p = 0.003). No individual gene association with eccentricity was statistically significant in the TCGA cohort, but pathway analysis identified its association with epithelial-mesenchymal transition (EMT), inflammatory, hypoxic, and stress-related programs. Across the TCGA RNA, CPTAC RNA and proteomic datasets, greater eccentricity was consistently associated with increased NF-κB signaling, collagen biosynthesis, and ER-to-Golgi transport and decreased L1CAM-related and neuronal-projection programs, with EMT showing the strongest cross-cohort consistency. Conclusions: Nuclear eccentricity from patient H&E slide images is a potential biomarker associated with survival and transcriptional/translational programs in GBM. Integrating quantitative pathology with RNA and proteomic data may provide an accessible approach for GBM prognosis and tumor biology characterization.