Purpose: Immune checkpoint inhibitors (ICIs) for non-small-cell lung cancer (NSCLC) are on the rise, but unfortunately, only a small percentage of patients benefit from them in the long term. Thus, it is crucial to identify biomarkers that can forecast the efficacy of immunotherapy.
Methods: We retrospectively studied 224 patients with NSCLC who underwent anti-PD-1 therapy. The role of biomarkers and clinical characteristics were assessed in a prognostic model.
Results: Only 14.3% of patients had both programmed death ligand 1 (PD-L1) and tumor mutational burden (TMB) outcomes, highlighting the need to investigate more available biomarkers. Our analysis found a correlation between histological PD-L1 TPS and hematological PD-1 expression. Analysis of hematological biomarkers revealed that elevated expression of CD4/CD8 and LYM% are positively associated with effective immunotherapy, while PD-1+ on T cells, NLR, and MLR have a negative impact. Moreover, high level of ΔCEA%, CYFRA21-1 and LDH may suggest ineffective ICIs. We also observed that disparate immunotherapy drugs didn't significantly impact prognosis. Lastly, by comparing squamous carcinoma and adenocarcinoma cohorts, ΔCEA%, CD3+PD-1+, CD4+PD-1+, and CD4/CD8 are more important in predicting the prognosis of adenocarcinoma patients, while age is more significant for squamous carcinoma patients.
Conclusion: Our research has yielded encouraging results in identifying a correlation between immunotherapy's response and clinical characteristics, peripheral immune cell subsets, and biochemical and immunological biomarkers. The screened hematological detection panel could be used to forecast an NSCLC patient's response to anti-PD-1 immunotherapy with an accuracy rate of 76.3%, which could help customize suitable therapeutic decision-making.
Keywords: Hematological biomarkers; Immunotherapy; Machine learning; Non-small-cell lung cancer.
© 2024. The Author(s).