Optimizing adjuvant treatment options for patients with glioblastoma

• DOI: 10.3389/fneur.2024.1326591
• Publication Date: 2024-02-21T05:52:05Z
• Authors: Enzhao Zhu; Jiayi Wang; Weizhong Shi; Qi Jing; Pu Ai; Dan Shan; Zisheng Ai
• Keywords: Chemoradiotherapy

Background This study focused on minimizing the costs and toxic effects associated with unnecessary chemotherapy. We sought to optimize adjuvant therapy strategy, choosing between radiotherapy (RT) chemoradiotherapy (CRT), for patients based their specific characteristics. selection process utilized an innovative deep learning method. Methods trained six machine (ML) models advise most suitable treatment glioblastoma (GBM) patients. To assess protective efficacy of these ML models, we employed various metrics: hazards ratio (HR), inverse probability weighting (IPTW)-adjusted HR (HR a ), difference in restricted mean survival time (dRMST), number needed treat (NNT). Results The Balanced Individual Treatment Effect Survival data (BITES) model emerged as effective, demonstrating significant benefits (HR: 0.53, 95% CI, 0.48–0.60; IPTW-adjusted HR: 0.65, 0.55–0.78; dRMST: 7.92, 7.81–8.15; NNT: 1.67, 1.24–2.41). Patients whose aligned BITES recommendations exhibited notably better rates compared those who received different treatments, both before after IPTW adjustment. In CRT-recommended group, advantage was observed when CRT over RT ( p < 0.001). However, this not case RT-recommended group = 0.06). Males, older patients, tumor invasion is confined ventricular system were more frequently advised undergo RT. Conclusion Our suggests that can effectively identify GBM likely benefit from CRT. These show promise transforming complex heterogeneity real-world clinical practice into precise, personalized recommendations.