Heterogeneity in biomarkers for advanced gastric cancer, gastroesophageal junction cancer, and esophageal adenocarcinoma (GC/GEJC/EAC) across regions: Insights from CheckMate-649.

4056 Background: In CheckMate 649, first-line nivolumab plus chemotherapy showed improved clinical efficacy vs chemotherapy in patients (pts) with advanced GC/GEJC/EAC, with varied benefit across regions in a subgroup analysis. To identify biomarkers predictive of differential efficacy outcomes, we did post hoc exploratory biomarker analyses by region. Methods: Post hoc exploratory analyses were done with whole-exome sequencing (WES) and RNA-seq of baseline tumor tissue. Pts were grouped by region: China (CN), Asia (including CN), United States + Canada + European Union (UCE), Latin America (LA). For WES-evaluable pts, gene alterations were profiled and pts grouped into genomic subtypes: chromosomal instability (CIN), genomically stable (GS), hypermutated (H), and Epstein–Barr virus (EBV)-positive. Gene expression signatures (GES) related to mitogen-activated protein kinase (MAPK) pathway, stroma, and angiogenesis were assessed by RNA-seq, and pts were stratified into tertiles based on GES score: high, medium, or low. Results: Of 1512 pts randomized, 647 were WES-evaluable (79, 141, 351, 155 in CN, Asia, UCE, and LA, respectively) and 769 RNA-evaluable (50, 106, 423, 240, respectively). A numerically lower proportion of CIN subtype and higher proportion of EBV subtype was seen in LA vs other regions (Table). Percentage (%) of select gene alterations was generally comparable across regions: of note, there was numerically higher % of altered FGFR2 (10%) in CN, altered TP53 (58%) , KRAS (16%) , CDKN2A (11%), and CDH1 (7%) in UCE, and altered PIK3CA (10%) and ERBB2 (22%) in LA. By contrast, there was numerically lower % of altered EGFR (4%) in Asia and altered ARID1A (8%) in LA. GES profile was broadly similar across regions, with numerically higher % of MAPK-high subgroup in UCE, stroma-medium/low subgroup in LA, and angiogenesis-low subgroup in CN (Table). Conclusions: While the biomarker profile was largely comparable across regions, differential features were noted for certain subgroups; clinical utility of these biomarkers requires validation in future trials. Clinical trial information: NCT02872116 . Regions CN Asia(including CN) UCE LA Genomic subtypes CIN 53 (67) 86 (61) 223 (64) 81 (52) GS 22 (28) 47 (33) 99 (28) 49 (32) H 4 (5) 6 (4) 18 (5) 6 (4) EBV 0 2 (1) 11 (3) 19 (12) GES 5-gene MAPK HighMediumLow 15 (30)13 (26)22 (44) 31 (29)29 (27)46 (43) 145 (34)153 (36)125 (30) 74 (31)79 (33)87 (36) 15-gene stroma HighMediumLow 20 (40)11 (22)19 (38) 38 (36)34 (32)34 (32) 151 (36)138 (33)134 (32) 70 (29)79 (33)91 (38) 5-gene angiogenesis HighMediumLow 15 (30)14 (28)21 (42) 35 (33)34 (32)37 (35) 137 (32)138 (33)148 (35) 81 (34)89 (37)70 (29) Data are n (%). % was calculated as events divided by number of WES- or RNA-seq–evaluable pts.