Integrating immunohistochemical biomarkers into patient-specific reaction-diffusion models of glioblastoma growth

Objective.Glioblastoma exhibits highly heterogeneous growth and infiltration patterns, limiting the predictive value of routine imaging and population-averaged tumor growth models. This study aimed to develop a clinically feasible, biomarker-informed reaction-diffusion framework that integrates standard magnetic resonance imaging with routinely available immunohistochemical (IHC) markers to support patient-specific modeling of glioblastoma growth.Approach.A reaction-diffusion model was formulated using standard clinical MRI sequences, including contrast-enhanced T1-weighted and T2-weighted images, together with IHC biomarkers, particularly the antigen protein (Ki-67) and Isocitrate dehydrogenase 1 (IDH-1) status. Baseline diffusion and proliferation parameters were estimated from MRI-derived growth metrics. A pathological coefficient was introduced as a biologically interpretable modifier of the proliferation term to incorporate microscopic tumor aggressiveness into the macroscopic growth model. The coefficient was identified through constrained inverse parameter estimation and subsequently interpreted using subgroup-specific representative values. Short-term agreement between simulated tumor extent and postoperative follow-up MRI was assessed within the available cohort.Main results.The proposed framework improved agreement between simulated and observed tumor extent within the studied cohort and demonstrated biologically coherent behavior across molecular subgroups. The pathology-informed proliferation term allowed the model to reflect differences in tumor aggressiveness associated with Ki-67 expression and IDH-1 status. In addition, the model generated spatial representations of tumor infiltration extending beyond the visible margins detected by routine MRI, highlighting regions of potential microscopic spread that are not directly observable on standard imaging.Significance.This study presents an exploratory step toward linking routinely acquired pathological biomarkers with physics-based glioblastoma growth modeling, by using standard-of-care MRI and IHC data, the proposed framework provides a clinically accessible tumor-growth prior that may support future treatment-response modeling, survival analysis, and individualized planning. Further validation in larger independent cohorts and extension to three-dimensional modeling are required before prospective clinical application.