Association Between Very High PD-L1 Expression versus High PD-L1 Expression in Patients Receiving Pembrolizumab as First-line Treatment for Advanced Non-small Cell Lung Cancer

Background: A prior single-center study found that among advanced non-small cell lung cancer (aNSCLC) patients with PD-L1 expression (PD-L1+e) of 50%—100% receiving immunotherapy as monotherapy, a very high level of PD-L1+e (≥90%) was associated with longer survival.

Objectives: We sought to validate this finding using independent real-world data from community oncology practices across the U.S.

Methods: We conducted a retrospective cohort study of aNSCLC patients who initiated pembrolizumab monotherapy for 1st line treatment and had a PDL1+e ≥ 50% using a nationwide, de-identified longitudinal electronic health record—derived database (Flatiron Health). We excluded patients with incomplete treatment data, those harboring alterations in EGFR, ALK, and ROS1 genes, and patients with a gap of >90 days between diagnosis and first visit or medication order. The primary outcome was overall survival, measured from initiation of pembrolizumab to date of death or last confirmed activity. Multiple imputation was used to impute missing covariates. Propensity score-based inverse probability weighting (IPW) was used to address confounding in Kaplan-Meier curves and Cox proportional hazard regression models. Baseline characteristics included in IPW were age at therapy initiation, race, sex, smoking history, tumor histology, presence of KRAS or BRAF mutation, and ECOG performance status.

Results: The cohort included1952 aNSCLC patients receiving pembrolizumab monotherapy. 50% of cohort members were female, median age at 1st line therapy initiation was 73 years IQR [65-80], 71% had non-squamous histology, 94% had a history of smoking, 24% had a KRAS mutation, and 46% had very high PDL1+e. IPW successfully balanced baseline covariates. Median overall survival in the weighted sample from the index date was 15.84 months [very high -PDL1+e (≥90%)] vs 12.72 months [high -PDL1+e (50-89%)], respectively. Having a very high PDL1+e was associated with lower mortality (IPW hazard ratio 0.79, 95% CI 0.69-0.91).

Conclusions: After adjusting for potential confounders, patients with PDL1+e very high (≥90%) aNSCLC receiving first-line pembrolizumab experienced prolonged median survival compared to those with high PDL1+ (50-89%) in a multicenter real-world setting. Early-phase aNSCLC immunotherapy trials with fewer than 100 patients per treatment arm should consider utilizing PDL1+e stratified randomization (50-89% and ≥90%) and report outcomes accordingly given the differential survival benefit seen with immunotherapy alone. This would ensure that randomized groups are evenly balanced with respect to very high and high PDL1+e levels leading to treatment effect estimates that are reliable and valid.