Investigating subregional PD-L1 expression within primary tumors to predict clinical outcomes in advanced NSCLC patients who received ICB-based therapy

TL;DR

This study shows that PD-L1 expression in the deep part of lung tumors better predicts treatment success in advanced lung cancer patients receiving immunotherapy.

Contribution

The study demonstrates that PD-L1 expression in the deep tumor region is a more reliable predictor of immunotherapy response than superficial regions.

Findings

PD-L1 expression in the deep tumor region (PTdeep) strongly correlates with higher response rates to immunotherapy.

PTdeep PD-L1 TPS ≥50% predicts significantly better progression-free survival compared to lower TPS.

Deep region PD-L1 outperforms superficial region and random PD-L1 in predicting treatment outcomes.

Abstract

Programmed cell death-ligand 1 (PD-L1) immunohistochemical expression currently is the only approved useful biomarker associated with the PD-1/PD-L1 immune checkpoint blockade (ICB) efficacy for non-small cell lung carcinoma (NSCLC) patients. However, different tumor biopsy strategies could reflect the substantial heterogeneity of PD-L1 within the same tumor (spatial heterogeneity). Therefore, we aimed to explore the impact of spatial heterogeneity on the predictive value of PD-L1 expression in NSCLC patients on the ICB treatment after two cycles.

All consecutive subjects with NSCLC receiving first-line ICB-based therapy for at least two cycles between January 2020 and March 2024 were enrolled and classified according to the biopsy strategies. Transbronchial lung biopsy (TBLB) or transbronchial mucosal biopsy was performed to obtain samples from the primary tumor superficial (PTsup) region. Moreover, endobronchial ultrasound-transbronchial needle aspiration (EBUS-TBNA) or percutaneous cutting needle biopsy (PCNB) was performed to get the primary tumor deep region (PTdeep). The predictive capacity of PD-L1 TPS to ORR from these two sites was assessed and compared by logistic regression analysis and ROC curve analysis. The prognostic value of PTdeep- and PTsup-related PD-L1 TPS to PFS was also expanded by performing propensity score matching as well as stratified analysis.

Among NSCLC receiving ICB therapy, PTsup-related PD-L1 TPS ≥50% was not associated with higher ORR (15.8% vs. 26.1%, P = 0.197) by stratified analysis. Instead, PTdeep-related PD-L1 TPS ≥50% could bring substantially a higher ORR than those with TPS <50% (52.4% vs. 17.4%, P = 0.025). Furthermore, cross analysis displayed that the PD-L1 TPS <50% from the superficial or deep subregion reached relatively similar ORRs (15.8% vs. 17.4%, P = 0.861), whereas patients with PTdeep-related PD-L1 TPS ≥50% manifested a higher ORR than those with PTsup TPS ≥50% (52.4% vs. 26.1%, P = 0.036). Moreover, PTdeep-related PD-L1 yielded the best performance in area under the curve (AUC) to predict the ORR (AUC = 0.699, P = 0.032) than random PD-L1 TPS (AUC=0.627, P=0.022) and PTsup-related PD-L1 TPS (AUC = 0.589, P = 0.204). As for the PFS, patients with PTdeep-related PD-L1 TPS ≥50% had a significantly superior PFS (mPFS 19.4 vs. 10.8 months; P = 0.006) compared with patients with PTdeep-related PD-L1 TPS <50%. After conducting matched and stratified analysis to control for potential confounding factors including immunotherapy agents and gender, PTdeep-related remained the most stable predictor for PFS.

PD-L1 from the deep subregion is a more solid predictive biomarker of both short- and long-time efficacies of ICB-based therapy, and optimizing the assessment of spatial heterogeneity provides a new perspective for clinicians to screen advanced NSCLC patients who can benefit from ICB-based therapy.

Figures