The following is a summary of “Optimizing adjuvant treatment options for patients with glioblastoma,” published in the February 2024 issue of Neurology by Zhu et al.
Researchers started a retrospective study using a novel deep-learning method to personalize the choice between adjuvant radiotherapy (RT) and chemoradiotherapy (CRT) for patients based on their unique characteristics.
They utilized six machine learning (ML) models to recommend optimal treatment for patients diagnosed with glioblastoma (GBM). To evaluate the effectiveness of these ML models, several metrics were utilized HR, inverse probability treatment weighting (IPTW)-adjusted HR (HRa), difference in restricted mean survival time (dRMST), and number needed to treat (NNT).
The results showed Balanced Individual Treatment Effect for Survival data (BITES) model proved most effective, demonstrating notable protective benefits (HR: 0.53, 95% CI, 0.48–0.60; IPTW-adjusted HR: 0.65, 95% CI, 0.55–0.78; dRMST: 7.92, 95% CI, 7.81–8.15; NNT: 1.67, 95% CI, 1.24–2.41). Patients adhering to BITES recommendations experienced significantly improved survival rates than those receiving alternative treatments, pre- and post-IPTW adjustment. In the CRT-recommended cohort, selecting CRT over RT conferred a notable survival advantage (P<0.001). However, this discrepancy was not observed in the RT-recommended group (P=0.06). Males, older individuals, and patients with tumors confined to the ventricular system were more frequently advised to undergo RT.
They concluded that the GBM study suggests BITES identified patients for CRT, showing promise for ML-driven personalized treatment.
Source: frontiersin.org/journals/neurology/articles/10.3389/fneur.2024.1326591/full