PD-L1 expression in primary tumour vs metastatic samples in the phase III MYSTIC study in first-line metastatic (m) NSCLC

Abstract

Background: PD-L1 expression determined by immunohistochemistry (IHC) can be a useful biomarker to assess the likelihood of benefit with anti-PD-1/PD-L1 therapies in pts withmNSCLC. Understanding the impact of sample type (primary tumour or metastatic) on predictivity of benefit will inform the suitability of such samples for clinical testing. MYSTIC (NCT02453282) was an open-label, phase III study of durvalumab (D)±tremelimumab vs chemotherapy (CT) as first-line treatment for mNSCLC; stratification factors for randomisation included tumour cell (TC) PD-L1 expression (≥25% vs<25%). While not statistically significant, D showed a clinically meaningful improvement in OS compared with CT (HR 0.76 [97.54% CI 0.56-1.02], p=0.036) in pts with PD-L1 TC≥25%. We investigated whether the use of a primary tumour or metastatic site biopsy to determine PD-L1 status impacted prevalence of TC≥25% or clinical benefit. Methods: All pts enrolled in MYSTIC were assessed centrally for TC staining for PD-L1 from tissue samples acquired<3 months prior to randomisation from either the site of the primary tumour or a distant metastasis. Testing was performed using the VENTANA PD-L1 (SP263) IHC assay. A post-hoc analysis evaluated prevalence, OS, ORR and duration of response (DoR) in pts with PD-L1 TC ≥25% as determined using either a primary tumour or a distant metastatic sample. Results: Of 1118 pts randomised, 716 (64.0%) provided a primary tumour sample and 402 (36.0%) provided a metastatic sample. The prevalence of PD-L1 TC≥25% assessed using primary vs metastatic samples was 43.0% vs 44.8% (p=0.569). Outcomes in pts with TC≥25% determined using samples of each type are shown. Conclusions: In MYSTIC, PD-L1 TC ≥25% prevalence was similar using primary or metastatic samples. Favourable HRs for OS with D vs CT were seen in pts with TC≥25% expression as determined in either the primary tumour or a metastatic site. Results should be interpreted with caution given the retrospective nature of the analysis.