323P - Development of a model to predict PD-L1 expression in pulmonary squamous cell carcinoma based on CT imaging features

Background

An immune checkpoint inhibitor that targets programmed cell death protein 1 (PD-1) or programmed death ligand 1 (PD-L1) has recently been discovered, and it demonstrates superior treatment efficacy and improves prognosis over existing chemotherapeutic agents in non-small cell lung cancer. Several studies have demonstrated correlations between computed tomography (CT) imaging features and PD-L1 expression in lung adenocarcinoma (ADC). However, there have been few studies on the association between lung squamous cell carcinoma (SCC) and CT findings. Thus, we aimed to identify the associations between CT findings and PD-L1 expression in SCC and to develop models to predict the expression of PD-L1 in lung SCC using CT.

Methods

This retrospective study included 99 patients diagnosed with SCC and pretreatment CT images and PD-L1 expression assay results. We performed CT feature analysis of the tumors using pretreatment CT images. We then constructed multiple logistic regression models to predict PD-L1 positivity in the total patient group and in 41 advanced-stage (≥ stage IIIB) patients. To identify predictive efficiency, areas under the receiver operating characteristic curves (AUCs) of each model were calculated.

Results

Among the 99 patients, 68 (68.7%) were PD-L1-positive and among the 41 advanced patients, 30 (73.2%) were PD-L1-positive. For the total patient group, the AUC of the ‘total significant features model’ (tumor stage, primary tumor size, pleural nodularity and lung metastasis) was 0.637, and that of the ‘selected feature model’ (pleural nodularity) was 0.556. For advanced-stage patients, the AUC of the selected feature model (primary tumor size, central distribution, pleural nodularity, and pulmonary oligometastases) was 0.882. Among these factors, the presence of pleural nodularity and pulmonary oligometastases had high odds ratios (4.11 and 9.60, respectively).

Conclusions

Our model was able to predict PD-L1 expression in patients with advanced-stage SCC of the lung, and the presence of pleural nodularity and pulmonary oligometastases were notable predictive CT features of PD-L1. A predictive model based on CT features may facilitate non-invasive assessment of PD-L1 expression.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

This study was supported by a VHS Medical Center Research Grant, Republic of Korea (grant number: VHSMC 21007).

Disclosure

All authors have declared no conflicts of interest.