A three-protein serum risk score for predicting immunotherapy response and prognosis in non-small cell lung cancer

Immune checkpoint inhibitors (ICIs) have extended survival in patients with non-small cell lung cancer (NSCLC) but their therapeutic benefit is limited to a proportion of patients. Predictive biomarkers based on tissue of origin of the tumor have their limitations, and thus there is a need for solid and minimally invasive predictive biomarkers. Our aim was to investigate serum proteomics via liquid biopsy for biomarker discovery.

In this retrospective extension of the TD-FOREKNOW trial, deep proteomic profiling was undertaken on pre-treatment serum samples of 72 patients with NSCLC receiving neoadjuvant therapy. Further quantitation of proteins in serum was performed by data independent acquisition mass spectrometry to obtain candidates associated with treatment outcome. Statistical regression was also used to screen for proteins related to ICI efficacy and a risk score composite model was set up to predict treatment response and prognosis.

From the 1,802 analyzed serum proteins, 59 serum proteins were differentially expressed in patients receiving immunotherapy plus chemotherapy. Using univariate logistic regression followed by least absolute shrinkage and selection operator (LASSO) regression, three factors, SERPINE2, DAZAP1, and MGAT4B, were identified whose baseline expression was correlated with the response to ICI therapy. The risk score model using the three proteins was an effective biomarker in predicting ICI response with an area under the curve (AUC) of 0.946 (95% CI: 0.874-1.000). Its predictive value for ICI response was validated in further survival analysis, showing that patients with a low risk score had significantly longer progression-free survival (HR = 0.13, 95% CI: 0.04-0.46, P = 0.002) and overall survival (HR = 0.14, 95% CI: 0.03-0.62, P = 0.033) than those with a high risk score.

A pre-treatment serum-based risk score that effectively predicts response to ICI therapy in patients with NSCLC was developed and validated. Our findings reveal the great prospect of human serum proteomics as a powerful liquid biopsy platform for biomarker discovery and construction of clinical prognostic models.