Development and validation of a nomogram to predicting the efficacy of PD-1/PD-L1 inhibitors in patients with nasopharyngeal carcinoma

Abstract

PurposeWith the treatment of nasopharyngeal carcinoma (NPC) by PD-1/PD-L1 inhibitors used widely in clinic, it becomes very necessary to anticipate whether patients would benefit from it. We aimed to develop a nomogram to evaluate the efficacy of anti-PD-1/PD-L1 in NPC patients.MethodsTotally 160 NPC patients were enrolled in the study. Patients were measured before the first PD-1/PD-L1 inhibitors treatment and after 8–12 weeks of immunotherapy by radiological examinations to estimate the effect. The least absolute shrinkage and selection operator (LASSO) logistic regression was used to screen hematological markers and establish a predictive model. The nomogram was internally validated by bootstrap resampling and externally validated. Performance of the model was evaluated using concordance index, calibration curve, decision curve analysis and receiver operation characteristic curve.ResultsPatients involved were randomly split into training cohort ang validation cohort. Based on Lasso logistic regression, systemic immune-inflammation index (SII) and ALT to AST ratio (LSR) were selected to establish a predictive model. The C-index of training cohort and validating cohort was 0.745 and 0.760. The calibration curves and decision curves showed the precise predictive ability of this nomogram. The benefit of the model showed in decision curve was better than TNM stage. The area under the curve (AUC) value of training cohort and validation cohort was 0.745 and 0.878, respectively.ConclusionThe predictive model helped evaluating efficacy with high accuracy in NPC patients treated with PD-1/PD-L1 inhibitors.