Abstract

Background: Although gastric cancer outcomes vary by race/ethnicity, a molecular basis for these differences has not been fully elucidated. To address this gap, we used the Tempus Lens Platform (Tempus AI, Inc., Chicago, IL) to query the multimodal de-identified database in order to establish and subsequently analyze a cohort of patients with gastric adenocarcinoma who underwent germline and tumor testing using xT (DNA) and xR (RNA).

Methods: Patients were categorized by R/E. RNA-seq data were normalized as transcripts per million (TPM) and reported as log2(TPM+1). Single-sample Gene Set Enrichment Analysis (ssGSEA) using Hallmark gene sets was used to compare cell signaling pathway activity. Real-world overall survival (rwOS) was defined as the time from biopsy to death or last known follow-up. rwOS survival curves were estimated using the Kaplan Meier (KM) approach and differences tested using log-rank tests. Hazard ratios (HR) were calculated using Cox proportional hazard models.

Results: Among 1,842 patients with reported race/ethnicity data, 1,186 (64%) identified as White, 327 (17.7%) as Hispanic/Latino (HL), 217 (11.8%) as Black or African American (AA) and 112 (6.1%) as Asian. In the multivariable Cox model, compared to White patients, rwOS was longer for HL (HR 0.65, 95% CI 0.48-0.88, p=0.005) and Asian (HR 0.64, 95% CI 0.41-0.98, p=0.042) patients. Asian patients had longer median rwOS compared to White patients (19 vs 10.9 months, log-rank p<0.001). Compared to White patients, HL patients were younger at diagnosis (median 57 vs 67 years, p<0.001) and a greater proportion were female (45% vs 28%, p<0.001). Tumor grade, microsatellite instability, tumor mutational burden, and CLDN18.2 expression differed significantly across groups (all p<0.01), whereas HER2 and PD-L1 status did not. At the genomic level, the frequency of somatic alterations in TP53 (57% HL vs 75% White), CDKN2A (11% Asian vs 21% White), CDH1 (6.9% AA vs 16% HL) differed between groups (all p,q<0.001). Pathogenic germline variants were identified in 10.1% of AA, 10.7% of Asian, 10.7% of HL, and 15.8% of White patients, most commonly involving HDAC2, ATR, ATM, and MUTYH, respectively. Transcriptomic profiling via ssGSEA revealed an inflammatory pathway activation among HL and Asian patients, with a significant increase in IL6/JAK/STAT3 and IFN-γ signaling vs White patients (all q<0.001). HL patients also demonstrated greater enrichment of the IL2/STAT5, TNFα and complement response pathways vs White patients (q<0.001).

Conclusion: This study represents the largest and most diverse clinico-genomic analysis of gastric cancer. Collectively, these findings highlight distinct clinical presentations, molecular alterations and survival outcomes by race/ethnicity, promoting the critical need for broad molecular testing to inform precision cancer care.

VIEW THE PUBLICATION