Abstract LB294: Baseline and on-treatment plasma-based genomics as a predictor of outcome in SAVANNAH: Savolitinib + osimertinib in EGFRm MET overexpressed/amplified NSCLC post-osimertinib

Savolitinib, an oral, potent, and highly selective MET-TKI, in combination with osimertinib, a 3rd-generation, irreversible, oral EGFR-TKI, may overcome common acquired MET-driven resistance following osimertinib treatment. The ongoing Ph2 SAVANNAH (NCT03778229) study is investigating this combination in pts with EGFRm NSCLC who have MET overexpression and/or amplification upon progressive disease (PD) post-osimertinib. Using plasma EGFRm as an estimate of tumor burden, we report exploratory analyses of progression-free survival (PFS) by plasma EGFRm clearance during treatment at Day (D) 22. Additionally, agreement between tissue (IHC/FISH) and plasma (NGS) testing for detecting MET overexpression and/or amplification is reported. Following central MET testing on tumor tissue collected upon PD post-osimertinib, eligible pts received oral savolitinib (300 or 600 mg QD, or 300 mg BID) + oral osimertinib 80 mg QD. MET overexpression was assessed by IHC (3+ staining in ≥50% of tumor cells [IHC50+]) and amplification by FISH (MET copy number ≥5 and/or MET:CEP7 ratio ≥2 [FISH5+]). Here, PFS analysis was performed in subgroups identified by exploratory higher biomarker cut-off levels of 3+ staining in ≥90% tumor cells (IHC90+) and/or MET copy number ≥10 (FISH10+). Plasma EGFRm (Ex19del/L858R only) analysis was conducted by ddPCR at D1 and every week thereafter for 4 weeks and at week 6. Plasma EGFRm clearance was defined as undetected EGFRm ctDNA at D22, where it was detected at D1. Baseline (BL) MET amplifications were determined via ctDNA NGS at D1. Agreement analysis was performed (DCO 27Aug2021). This analysis included 192 pts who received savolitinib 300 mg QD + osimertinib and had ≥2 post-BL RECIST scans. Of those with valid tissue FISH and IHC results and evaluable ctDNA at D1 (88% [169/192]), 124 pts had detectable plasma EGFRm. Plasma EGFRm clearance was assessed in 103 pts with detectable D1 plasma EGFRm who also had evaluable ctDNA at D22. Clearance was observed more frequently in pts with IHC90+ or FISH10+ tumors (42% [28/66]) than those with IHC90- and FISH10- tumors (16% [6/37]). Median PFS (95% CI) with vs without clearance was 11.0 (7.4, 13.0) vs 4.1 (2.7, 5.5) months in pts with IHC90+ or FISH10+ tumors. Agreement between tissue FISH and plasma NGS was assessed in 72 evaluable pts. Overall, 21 (29%) pts had MET FISH10+ tumors and 14 (19%) had plasma NGS focal MET amplification, with 11 being positive by both methods (positive percent agreement [P