Narrative review suggests multi-omics AI models improve glioma radiotherapy response prediction over single-modality approaches.

Radiotherapy remains a cornerstone in glioma treatment, yet its efficacy is significantly hindered by tumor heterogeneity and molecularly driven radioresistance. This review systematically delineates molecular biomarkers that influence radiotherapy outcomes, categorizing them into radiosensitivity (e.g., IDH1 mutations, MGMT promoter methylation, TIM-3) and radioresistance (e.g., CD133, CD44, PRMT1, CSF-1R,RAD51,HMGB2). Mechanistically, radiosensitivity is governed by DNA repair fidelity (MGMT), ferroptosis suppression (PRMT1), and immune modulation (TIM-3/TAMs). Radioresistance arises from cancer stem cell maintenance (CD133/HMGB2), TAM polarization (CSF-1R/CD44), and enhanced homologous recombination (RAD51). Integrating molecular stratification into radiotherapy paradigms demonstrates clinical utility: MGMT methylation permits radiation dose de-escalation (52–54 Gy vs. 60 Gy) without compromising survival (32 vs. 25 months), while TIM-3 expression predicts responsiveness to combinatorial immunotherapy. A multi-omics AI model combining radiomics, dosiomics, and clinical data to predict radiotherapy response in glioma. Using a support vector machine trained on 176 patients, the fused model achieved an AUC of 0.728(95% CI:0.717–0.739) in validation, outperforming single-modality approaches. These advances underscore the transformative potential of biomarker-guided precision radiotherapy, enabling tailored interventions that counteract resistance mechanisms and synergize with immunotherapies. By bridging molecular insights with clinical innovation, this paradigm shift promises to redefine glioma management, offering renewed hope for overcoming therapeutic recalcitrance in this devastating malignancy.