Early circulating tumor (ct) DNA dynamics and efficacy of lorlatinib: Analysis from the CROWN study.

9011 Background: Lorlatinib, a third-generation ALK tyrosine kinase inhibitor, significantly improved progression-free survival (PFS) and overall/intracranial responses vs crizotinib in patients (pts) with previously untreated ALK-positive advanced non-small cell lung cancer (NSCLC) in the ongoing randomized Phase 3 CROWN study (NCT03052608). To identify whether additional molecular biomarker analysis correlated with efficacy, we evaluated early ctDNA dynamics compared with clinical outcomes. Methods: Plasma samples were prospectively collected at screening (SC), week 4 (cycle 2, day 1 [C2D1]), week 24 (C7D1), and end of treatment for ctDNA analysis. ctDNA was analyzed using Guardant360 CDx (Guardant Health, Inc., Redwood City, CA, USA). Mean variant allele fraction (VAF) of ALK alterations (fusions and/or mutations) and overall detected alterations at each time point and longitudinal mean change (dVAF) as (VAF C2D1 ) – (VAF SC ) were calculated; dVAF <0 indicated decreased ctDNA at week 4. Objective tumor response and PFS were evaluated according to dVAF. These analyses were repeated vs ctDNA results at week 24. Additional correlation analyses between depth of molecular response and/or ctDNA clearance and clinical outcomes are ongoing. Results: Paired samples were available at SC and week 4 from 232 of 255 pts included in the ctDNA analysis: 118/130 (90.8%) in the lorlatinib arm and 114/125 (91.2%) in the crizotinib arm. ALK alterations were detected in 122/232 (52.6%) pts at SC (62/118 [52.5%] from the lorlatinib arm) but only 19/232 (8.2%) at week 4 (8/118 [6.8%] from the lorlatinib arm). Mean VAF of ALK alterations at week 4 was significantly decreased compared with SC in both treatment arms (lorlatinib -1.54, crizotinib -1.25; both P<0.0001; P=0.4239 between arms). In the lorlatinib arm, mean VAF at week 4 was significantly decreased compared with SC in pts with a complete or partial response (dVAF -1.53; n=47; P<0.0001), or stable disease (dVAF -1.37; n=12; P=0.0304). Similar results were observed in the crizotinib arm. In pts with dVAF <0 for ALK alterations, mean percent change from screening in tumor size was -40.8% with lorlatinib (n=59) and -38.7% with crizotinib (n=58). Only 2 pts had dVAF ≥0, both from the crizotinib arm. Median PFS for pts with dVAF <0 for ALK alterations was not reached in the lorlatinib arm (n=62), and was 7.4 months (95% CI, 7.2–9.3) in the crizotinib arm (n=58). Similar response and PFS data were observed in the analysis of dVAF for ALK alterations at week 24. Conclusions: Early ctDNA dynamics may predict lorlatinib efficacy in pts with previously untreated ALK-positive NSCLC. The magnitude of reduction in ctDNA at 4 weeks may be associated with better responses and potentially longer PFS. These findings further support the utility of dynamic ctDNA monitoring in ALK-positive NSCLC. Reference: Shaw AT, et al. N Engl J Med. 2020;383:2018-2029. Clinical trial information: NCT03052608.