Abstract
BACKGROUND: Immune checkpoint inhibitor (ICI) therapy has been used in various tumors. The biomarkers predictive of a response to ICI treatment remain unclear, and additional and combined biomarkers are urgently needed. Secreted factors related to the tumor microenvironment (TME) have been evaluated to identify novel noninvasive predictive biomarkers. METHODS: We analyzed 85 patients undergoing ICI therapy as the primary cohort. The associations between ICI response and all biomarkers were evaluated. A prediction model and a nomogram were developed and validated based on the above factors. RESULTS: Seventy-seven patients were enrolled in the validation cohort. In the primary cohort, the baseline serum levels of H3Cit, IL-8 and CRP were significantly higher in nonresponder patients. A model based on these three factors was developed, and the "risk score" of an ICI response was calculated with the formula: "risk score" = 3.4591×H3Cit + 2.5808×IL8 + 2.0045 ×CRP- 11.3844. The cutoff point of the "risk score" was 0.528, and patients with a "risk score" lower than 0.528 were more likely to benefit from ICI treatment (AUC: 0.937, 95% CI: 0.886-0.988, with sensitivity 80.60%, specificity 91.40%). The AUC was 0.719 (95% CI: 0.600-0.837, P = 0.001), with a sensitivity of 70.00% and specificity of 65.20% in the validation cohort. CONCLUSIONS: A model incorporating H3Cit, IL-8 and CRP has an excellent prediction ability for ICI response; thus, patients with a lower "risk score" selectively benefit from ICI treatment, which may have significant clinical implications for the early detection of an ICI response.