Comparative dosimetric analysis of the whole brain, dural venous sinuses and skull bone marrow as candidate organs at risk for treatment-related lymphopenia in glioblastoma

Abstract Treatment-related lymphopenia is an established prognostic factor in glioblastoma, yet the anatomical origin remains unclear. This retrospective single-center study compared the whole brain, dural venous sinus and cranial bone marrow as organs at risk for predicting lymphocyte depletion and evaluated the prognostic significance of nadir absolute lymphocyte count (ALC). The cohort comprised 155 patients with IDH-wildtype glioblastoma treated with concurrent chemoradiotherapy. V20 Gy was extracted for each structure. Cox proportional hazards regression evaluated associations between nadir ALC and overall survival. Multiple linear regression predicted nadir ALC from dosimetric parameters adjusted for clinical factors and regimen. Nadir ALC independently predicted overall survival (HR 0.897 per 100 cells/μl; P = 0.004), and severe lymphopenia (ALC < 500 cells/μl) was associated with shorter median overall survival (15.1 vs 21.1 months; P = 0.012). Whole brain and skull V20 Gy were each significantly associated with nadir ALC (both P < 0.01), whereas dural venous sinuses V20 Gy was not (P = 0.194). Brain and skull models showed equivalent predictive performance (AIC 185 vs 184, R2 0.351 vs 0.353). The two doses were strongly correlated (Spearman ρ = 0.925), and neither remained significant when entered simultaneously, so the responsible structure cannot be determined. In regimen-stratified analyses, both structures remained significant in the 40 Gy cohort, but not in the 60 Gy cohort. Cranial bone marrow dose predicts nadir ALC comparably to, but inseparably from, whole brain dose, and nadir ALC is independently prognostic for survival. These hypothesis-generating findings require validation before adopting skull-sparing strategies in glioblastoma radiotherapy.