Artificial neural network identified a 20-gene panel with promising value in predicting immunotherapy response and survival benefits after anti-PD1/PD-L1 treatment in GBM patients

<b>Question: Which biomarker can effectively predict the immunotherapy response and prognostic benefits of GBM patients, and how can potential responders to immune checkpoint inhibitors (ICIs) be discerned?</b><b>Findings: In this retrospective, multiple-cohort study, we used two groups of ICI-treated GBM patients with large differences in survival benefits to define nonresponders and responders. Then, an artificial neural network was developed and showed excellent performance in predicting the immunotherapy response and prognostic benefits of GBM patients.</b><b>Meaning: The 20-gene panel could guide oncologists to accurately select potential responders for the preferential use of ICIs in GBM.</b><br>