切除可能な早期非小細胞肺癌におけるPD-L1発現を予測する臨床病理学的CTモデル
Yaoyao Zhuo1,2,3, Qingle Wang1,2,3, Yi Zhan4
1Department of Radiology, Zhongshan Hospital, Fudan University, Shanghai, China.
Background:
The expression of programmed death-ligand 1 (PD-L1) has an impact on survival outcomes in non-small cell lung cancer (NSCLC) patients, but preoperative diagnosis is challenging. This study aimed to construct and validate a non-invasive model for predicting PD-L1 expression in early-stage resected NSCLC based on computed tomography (CT) features and clinicopathological characteristics.
Methods:
In this retrospective study, the clinical, pathological, and CT data were obtained from consecutive NSCLC patients who had undergone resection from January 2016 to March 2018. The clinicopathologic, CT, and clinicopathologic-CT models were constructed after univariate and multivariate logistic regression analyses. The Kaplan-Meier analysis and log-rank test were used for survival analysis.
Results:
A total of 679 consecutive patients with 695 early-stage NSCLC nodules were included, and there were 243 {median age 57 [interquartile range (IQR), 48-63] years; 152 females} in the positive PD-L1 group and 452 [median age 58 (IQR, 50-65) years; 315 females] in the negative PD-L1 group. Smoking history, spread through air spaces (STAS), average CT value, lobulation, and cytokeratin 7 (CK7) were independent predictors of positive PD-L1 NSCLC. In validation set, the area under the curve (AUC) value of clinicopathologic model, CT model and clinicopathologic-CT model were 0.630 [95% confidence interval (CI): 0.621-0.702; sensitivity =0.701; specificity =0.542], 0.629 (95% CI: 0.574-0.638; sensitivity =0.624; specificity =0.615), and 0.819 (95% CI: 0.740-0.837; sensitivity =0.763; specificity =0.760), respectively. The clinicopathologic-CT model had higher predictive performance than the other two models by DeLong test, both in the training and validation sets.
Conclusions:
Smoking history, STAS, average CT value, lobulation, and CK7 might be helpful in the diagnosis of PD-L1 expression in patients with early-stage NSCLC. The clinicopathologic-CT model had higher predictive performance than the clinicopathologic and CT models.
