Microbiome-derived bile acids and association with dietary fiber intake to anti-tumor responses in patients with melanoma receiving immune checkpoint blockade (ICB).

9575 Background: Multiple studies have now linked high fiber diet (HFD) and outcomes with ICBs, however, the clinical actionability and mechanistic basis of this association remain incompletely defined. We investigated whether microbiome-derived metabolites may function as potentially actionable mediators of this interaction. Methods: Blood samples from two studies in patients (pts) with melanoma starting ICB were utilized: (1) longitudinal samples from the prospective Phase II DIET clinical trial (NCT04645680), in which pts with melanoma starting ICBs were randomized (2:1) to a HFD up to 50 g fiber/day (d) or a healthy control diet (HCD, 20 g/d) and (2) baseline samples from an observational prospective diet/microbiome ICB-treated melanoma cohort. Untargeted and targeted metabolome analysis of bile acids and tryptophan-metabolites were performed via liquid chromatography (LC)- and short-chain fatty acids (SCFAs) via gas chromatography (GC)-mass spectrometry. Circulating metabolite levels were compared between pre- and on-treatment samples in DIET trial by arm and between pts with response (R, assessed by RECIST 1.1) and no response (NR) in the observational cohort. Results: A total of 110 pts were included: 45 DIET study pts randomized to HFDI (n=30) vs HCD (n=15), and 65 pts in the observational cohort. As previously reported, response rates were numerically improved in HFD (77% vs 29% in HCD (p=0.06), in the DIET study. Untargeted metabolomics yielded 38 metabolites with statistically significant differences (nominal p5.0). Targeted metabolomics showed that HFD reduced secondary-to-primary bile acid ratios (deoxycholic acid DCA to cholic acid CA, p=0.02) as well as glycine- and taurine-conjugated to unconjugated bile acid ratios (glycocholic acid GCA to CA, taurocholic acid TCA to CA, both p=0.03). In the observational cohort, GCA/CA and TCA/CA ratios were significantly lower in R vs NR (p=0.03 and p=0.04, respectively) whereas no difference was seen in DCA/CA ratios. SCFAs and tryptophan metabolites showed neither temporal changes in DIET trial nor response associations in the observational cohort. Conclusions: We observed that bile acid deconjugation, a key microbiome-mediated step, was reduced by an HFD intervention and inversely associated with response to ICB in an observational cohort. Building on prior work demonstrating that additional HFD-mediated secondary bile acids can impair T cell function, microbial bile acid metabolism emerges as a potential determinant of anti-tumor response targetable through HFD. Clinical trial information: NCT04645680 .