Circulating immune protein biomarkers of early response and resistance in patients with metastatic HER-2–negative esophagogastric cancer (mEGC) treated with pembrolizumab and chemotherapy.

4024 Background: Immune checkpoint inhibition (ICI) in combination with chemotherapy is standard of care for first line HER2- mEGC. Tissue-based biomarkers can improve patient selection for ICI therapy, but these biomarkers require invasive biopsy, and the optimal cutoff for treatment benefit is controversial. Longitudinal plasma samples were collected from patients (pts) enrolled in the KeyLargo trial (NCT03342937), a single-arm phase II study of pembrolizumab (pembro) combined with oxaliplatin and capecitabine in first line HER2- mEGC, to identify novel prognostic and predictive circulating protein biomarkers. Methods: We employed large scale Next generation multiplex proteomics (Olink Explore) to evaluate 1536 functional protein biomarkers in 36 pts, including markers known to drive tumor biology and immune response. Samples at baseline (n=36), one week of pembro monotherapy (C1D8) (n=28), C2D1 (n=25), first restaging (n=26), and progression (n=20) were analyzed. Cox models assessed univariable associations between each marker and outcome at baseline. Markers that changed from baseline to C1D8 with effect size >0.5, and to other time points with effect size >1, were deemed most important, all with p<0.05 (paired t-test). Results: At baseline, high PREB, EPHA10, GHRHR, BAMBI and GDF15 levels were associated with worse progression-free survival (PFS) (p<0.001); high SPINK1, S100P, LDLR, GDF15 and PM20D1 levels were associated with worse overall survival (OS) (p<0.003). On-treatment analyses revealed that circulating PDL1 levels were significantly elevated in the presence of pembro at all time points (p<10 -16 ). After one week of pembro monotherapy (C1D8), 6 immune-related biomarkers (TNFRSF8, CXCL9, CCL19, IFNγ, IL10, CXCL10) exhibited significant change (effect size >0.5). At C2D1 and first restaging, the addition of chemotherapy to IO revealed a shift as key metabolic and hormone-regulating proteins (KRT19, GCG, ARG2) were most significantly altered (effect size >1). However, at progression an immune-related phenotype re-emerged as IL10, IL17A and GPA33 significantly changed (effect size >1). Subsequent pathway analyses aligned with the individual markers, highlighting the impact of immune cell signaling after IO monotherapy while cell signaling pathways, hormone-controlled pathways, and extracellular matrix degradation were activated at later times. Conclusions: Multiple prognostic biomarkers were identified that implicate proteins related to hormone metabolism and TGFβ biology in mEGC, including GDF15, a key molecule regulating obesity, tumor metastasis, immunomodulation, and cachexia. Significant changes in immune-related cytokines were seen early (C1D8) as well as at progression, highlighting a potential role for circulating immune protein biomarkers in mEGC.