Construction and validation of a prognostic prediction model for advanced non-small cell lung cancer with first-line immune checkpoint inhibitors
- VernacularTitle:ICIs一线治疗晚期NSCLC的预后预测模型构建与验证
- Author:
Hongxia CUI
1
;
Su LI
1
;
Yu LIANG
2
;
Xin LIU
2
Author Information
1. Dept.of Pharmacy，Cancer Hospital of Dalian University of Technology/Liaoning Cancer Hospital & Institute，Shenyang 110042，China
2. Dept. of Colorectal Surgery，Cancer Hospital of Dalian University of Technology/Liaoning Cancer Hospital & Institute，Shenyang 110042，China
- Publication Type:Journal Article
- Keywords:
non-small cell lung cancer;
immune checkpoint inhibitors;
prognosis;
Prediction model;
LASSO regression
- From:
China Pharmacy
2026;37(18):2395-2401
- CountryChina
- Language:Chinese
-
Abstract:
OBJECTIVE To construct and validate a prognostic prediction model for first-line immune checkpoint inhibitors （ICIs） therapy for advanced non-small cell lung cancer （NSCLC）.METHODS The clinicopathological data and peripheral blood parameters from 211 patients with advanced NSCLC who received first-line ICIs therapy at our hospital from January 2020 to December 2023 were retrospective collected. The patients were randomly divided into a training set （151 cases） and a validation set（60 cases） at a ratio of 7∶3. Using progression-free survival （PFS） as the outcome measure， LASSO regression was used to preliminarily screen predictive variables in the training set，and the preliminarily screened predictive variables were entered into a multivariable Cox regression model to construct Model 1. Meanwhile，restricted cubic splinec analysis was performed on the continuous variables preliminarily screened by LASSO regression. The variables with nonlinear relationships （ P ＜0.1） were analyzed by multivariable Cox regression in RCS form to construct Model 2. The concordance index，receiver operating characteristic curves，calibration curves，and decision curve analysis were used to compare the predictive performance of the two models； the optimal model was selected to construct a nomogram.Sensitivity analysis was performed by reconstructing the multivariable Cox regression model after excluding cases with unknown programmed death-ligand 1（PD-L1）status. Kaplan-Meier method was used to plot survival curves to analyze the survival differences among patients with different risk strata.RESULTS The variables used to construct Model 1 were sex， smoking history， PD-L1 expression status， bone metastasis， and carcinoembryonic antigen （CEA）；the variables used to construct Model 2 were sex， smoking history， PD-L1 expression status， bone metastasis， CEA， neutrophil（Neu）， and albumin（ALB）. The concordance indices of Model 2 were 0.710 in the training set and 0.672 in the validation set，both superior to those of Model 1 （0.649，0.634）. The areas under the curve （AUC） of Model 2 for 12， 18， and 24 months PFS were 0.737，0.782 and 0.788 in the training set，and 0.709，0.630 and 0.663 in the validation set，respectively，all outperforming Model 1 （0.653， 0.704， 0.697 in the training set；0.688， 0.605， 0.651 in the validation set）.The calibration curves of both models showed good consistency between the predicted probability and actual probability，and the clinical net benefit of Model 2 was higher than that of Model 1. Sensitivity analysis indicated that the prediction model was robust. The median PFS of patients in the low-risk group was significantly longer than that in the high-risk group（ P ＜0.05）.CONCLUSIONS Sex， smoking history， PD-L1 expression status， bone metastasis， Neu， ALB， and CEA were independent influencing factors for PFS in advanced NSCLC patients receiving first-line ICIs. The prediction model constructed based on the above factors has good predictive performance and can be used to predict the prognosis of advanced NSCLC patients.
