EGFR mutation status affects intra-tumoural heterogeneity of PD-L1 expression but not agreement between assays in resectable non-small cell lung cancer

Background

The predictive value of PD-L1 to select patients for immunotherapy in resectable NSCLC remains imprecise, confounded by different assays used across trials and intra-tumoural heterogeneity (ITH). We sought to compare the concordance between 3 PD-L1 antibodies stratified by EGFR mutation status, evaluate ITH and implications on survival outcomes.

Methods

Tissue microarrays were constructed from stage IA-IIIA NSCLC with 3 tumour cores per patient. Tumour proportion score (TPS) was evaluated by 3 pathologists for SP263, SP142, 22C3 and analysed in tertiles of < 1 %, 1–49 % and ≥ 50 %. ITH was defined as discordant TPS in ≥ 2/3 tumour cores. Cohen’s kappa test was used to assess agreement. Survival outcomes were estimated using Kaplan-Meier.

Results

A total of 561 patients were included, 59.5% (334/561) were EGFR-mutant. Stage IA comprised 45.5%(255/561), IB 24.1%(135/561), IIA 12.7%(71/561), IIB 4.5%(25/561) and IIIA 13.4%(75/561).

Across 1683 tumour cores, SP263 and 22C3 had the highest concordance (Kappa = 0.689), followed by 22C3 and SP142 (Kappa = 0.354), then SP263 and SP142 (Kappa = 0.284), similar between EGFR-mutant and EGFR-wildtype. Agreement between pathologists was almost perfect.

ITH by SP263 was observed in 14.1 % of EGFR-mutant versus 24.2 % in EGFR-wildtype(p = 0.002). Discordance was highest among TPS 1–49 % at 92.6 % (88/95) followed by ≥ 50 % at 37.8 % (14/37) and least among < 1 % at 0 % (0/429) (p < 0.001). For tumour cores scored 1–49 %, 63 %/70 % of adjacent cores were scored < 1 % for EGFR-wildtype/mutant respectively. Histological grade was the only independent predictor of PD-L1 ITH on multivariable analysis. PD-L1 ITH was not associated with survival on multivariable analysis.

Conclusion

PD-L1 scoring by SP263 and 22C3 are interchangeable but not SP142 regardless of EGFR status. PD-L1 ITH was more common in EGFR-wildtype versus EGFR-mutant tumours. Extra care should be taken to select the most representative tumour core for tumours with high histological grade or TPS 1–49% as this may influence peri-operative treatment decisions.