Dynamic Peripheral Blood Immune Cell Markers for Predicting the Response of Patients with Metastatic Cancer to Immune Checkpoint Inhibitors

Chen Wei,Mengyu Wang,Quanli Gao,Shasha Yuan,Wenying Deng,Liangyu Bie,Yijie Ma,Chi Zhang,Shuyi Li,Suxia Luo,Ning Li
DOI: https://doi.org/10.1007/s00262-022-03221-5
IF: 6.63
2023-01-01
Cancer Immunology Immunotherapy
Abstract:Immune checkpoint inhibitors (ICIs) have shown durable responses in various malignancies. However, the response to ICI therapy is unpredictable, and investigation of predictive biomarkers needs to be improved.In total, 120 patients receiving ICI therapy and 40 patients receiving non-ICI therapy were enrolled. Peripheral blood immune cell markers (PBIMs), as liquid biopsy biomarkers, were analyzed by flow cytometry before ICI therapy, and before the first evaluation. In the ICI cohort, patients were randomly divided into training (n = 91) and validation (n = 29) cohorts. Machine learning algorithms were applied to construct the prognostic and predictive immune-related models.Using the training cohort, a peripheral blood immune cell-based signature (BICS) based on four hub PBIMs was developed. In both the training and the validation cohorts, and the whole cohort, the BICS achieved a high accuracy for predicting overall survival (OS) benefit. The high-BICS group had significantly shorter progression-free survival and OS than the low-BICS group. The BICS demonstrated the predictive ability of patients to achieve durable clinical outcomes. By integrating these PBIMs, we further constructed and validated the support vector machine-recursive and feature elimination classifier model, which robustly predicts patients who will achieve optimal clinical benefit.Dynamic PBIM-based monitoring as a noninvasive, cost-effective, highly specific and sensitive biomarker has broad potential for prognostic and predictive utility in patients receiving ICI therapy.
What problem does this paper attempt to address?