Related Experiment Video
Updated: Jun 30, 2026
Using 22C3 Anti-PD-L1 Antibody Concentrate on Biopsy and Cytology Samples from Non-small Cell Lung Cancer Patients
Published on: September 25, 2018
Risk stratification of PD-L1 expression in non-small cell lung cancer: a predictive model for prognosis
Longfei Jia1,2, Yi Yang3, Xinye Xia4
1Department of Thoracic Surgery, Nanfang Hospital, Southern Medical University, Guangzhou, China.
Background:
Immune checkpoint inhibitors (ICIs) have become the standard of care for non-small cell lung cancer (NSCLC) patients with programmed cell death ligand-1 (PD-L1) expression ≥1%. However, not all patients benefit equally from this treatment. This study develops a predictive model for risk stratification in NSCLC patients, integrating clinical, radiological, genomic, and PD-L1 expression data to enhance immunotherapy outcomes.
Methods:
We conducted a multicenter, retrospective cohort study of NSCLC patients treated with immunotherapy from January 2018 to June 2024. The primary endpoint was overall survival (OS), and secondary endpoints included progression-free survival (PFS). We constructed a prognostic nomogram using Kaplan-Meier survival analysis and multivariable Cox regression to identify key prognostic factors. Model performance was rigorously validated internally via C-index, time-dependent receiver operating characteristic (time-ROC) curves, calibration plots, 1,000-replicate bootstrap resampling, and decision curve analysis (DCA)-and patients were stratified into distinct risk groups.
Results:
PD-L1 expression ≥1% was associated with significantly improved OS (563 vs. 421 days, P=0.004). The multivariable Cox regression model included sex, histologic subtype, stage, Eastern Cooperative Oncology Group (ECOG) performance status (PS), and mutations in BRAF, KRAS, and TP53. The model demonstrated good discrimination with area under the ROC curve values of 0.723 [95% confidence interval (CI): 0.671-0.775] for 1-year OS, 0.789 (95% CI: 0.716-0.862) for 3-year OS, and 0.688 (95% CI: 0.586-0.790) for 5-year OS predictions, respectively. Calibration plots indicated good agreement between predicted and observed 1-year OS probabilities. DCA demonstrated that the model provides clinical benefit at a range of decision thresholds for 1-year OS.
Conclusions:
The prognostic model constructed in this study has excellent short-term predictive efficacy and clinical practicability. It can effectively stratify the prognosis risks of NSCLC patients after ICIs treatment, providing a convenient quantitative reference for clinical decision-making.
