Technology in Anesthesiology

NIH researchers have developed an AI tool focused on immune checkpoint inhibitors capable of predicting cancer patients’ responses to immunotherapy using routine clinical data, such as blood tests. Traditionally, predictive biomarkers approved by the FDA, like tumor mutational burden and PD-L1, help identify candidates for immunotherapy but are not always accurate. Existing machine-learning models using molecular sequencing data are effective but costly and not routinely available. The new AI model diverges by utilizing five routinely collected clinical features: patient age, cancer type, history of systemic therapy, blood albumin level, and blood neutrophil-to-lymphocyte ratio, along with tumor mutational burden from sequencing panels. This model, evaluated with data from 2,881 patients across 18 solid tumor types, accurately predicts a patient’s likelihood of responding to immune checkpoint inhibitors and their overall survival and disease-free duration. The model also identifies patients with low tumor mutational burden who can still benefit from immunotherapy. Despite the promising results, researchers emphasize the need for larger prospective studies to validate the AI model in clinical settings. The model, named Logistic Regression-Based Immunotherapy-Response Score (LORIS), is publicly accessible.

Source: asamonitor.pub/3yQSL7O