LBA4 - Effect of post-study immunotherapy (IO) on overall survival (OS) outcome in patients with metastatic (m) NSCLC treated with first-line durvalumab (D) vs chemotherapy (CT) in the Phase 3 MYSTIC study

Abstract

Background

In MYSTIC (NCT02453282), an open-label, Phase 3 study of first-line D (anti-PD-L1) ± tremelimumab vs platinum-based CT in mNSCLC, while not statistically significant, a clinically meaningful improvement in OS was seen with D vs CT in pts with tumour cell PD-L1 expression ≥25% (PD-L1 TC ≥25%; HR 0.76 [97.54% CI 0.56–1.02], p=0.036). Here we describe subsequent treatment patterns and explore the effect of subsequent IO on the OS outcome with D vs CT.

Methods

IO/CT-naïve mNSCLC pts were randomised to D (20 mg/kg i.v. q4w until disease progression) or CT (up to 6 cycles; pemetrexed maintenance permitted). In-study crossover from CT to D was not allowed. For D vs CT, the primary endpoint was OS in pts with PD-L1 TC ≥25%. Three statistical models were employed in exploratory analyses to evaluate the effect of subsequent (post-study) IO on the OS data: the rank preserving structural failure time (RPSFT) method, the inverse probability of censoring weighting (IPCW) method, and a 2-stage method.

Results

163 and 162 pts with PD-L1 TC ≥25% were randomised to D and CT, respectively. At data cut-off (04 Oct 2018), 44.8% of pts in the D arm and 58.6% of pts in the CT arm had received subsequent treatment (Table). Most pts started subsequent treatment within 2 mos of discontinuing study treatment. Among pts who received subsequent treatment, IO was administered to 10/73 (13.7%) pts in the D arm and 64/95 (67.4%) pts in the CT arm; most commonly nivolumab. Using the 2-stage method, which was the most appropriate for evaluating the effect of subsequent IO, OS was improved with D vs CT (HR 0.66 [95% CI 0.51, 0.86]).Table: LBA4