Microbiome associations with chemotherapy-induced nausea and vomiting: A metagenomic shotgun sequencing study.

12110 Background: Chemotherapy-induced nausea and vomiting (CINV) is common, with four out of every ten patients reporting moderate to severe nausea even after treatment with antiemetics. Existing risk prediction models have focused on demographic, clinical, and patient-reported factors, but predictive accuracy remains low. A missing predictor may be the gut microbiome, or gut microorganisms that engage in reciprocal signaling with the brain via many of the same pathways known to cause nausea. Methods: In this original research study, we conducted whole-metagenomic shotgun sequencing of the gut microbiome to evaluate its association with patient-reported CINV and improve existing risk prediction algorithms by employing a random forest approach. Nausea and vomiting were reported 24 hours after infusion (acute) and 5 days after infusion (delayed) in patients treated with moderately- or highly-emetogenic chemotherapy using the MASCC Antiemesis Tool. Participants were recruited from five ORIEN centers: City of Hope, Moffitt Cancer Center, Ohio State University, Rutgers University, and University of Colorado-Denver. Fecal samples were collected prior to chemotherapy using at-home fecal occult blood test (FOBT) cards. Results: Participants (n=233) were 57 years of age on average (SD=12), and most were female (80%), White (89%), with breast cancer (55%). More than half of participants (65%) had early stage cancer. Any acute and delayed self-reported nausea was reported by 40% and 67% of participants, respectively. Any acute and delayed vomiting was reported by 3% and 12% of participants, respectively; the low rate of acute vomiting precluded additional analyses. No significant difference was observed in microbial diversity and the abundance of dominant bacteria at the phylum level between patients who reported nausea or vomiting and those who did not. Bacillota and Pseudomonadota were two phyla identified as the most significantly enriched in patients who reported nausea or vomiting, at either timepoint (FDR-adjusted p < 0.05). The predictive performance of risk models based on established factors were improved by incorporating the top 15 differentially abundant bacterial species for acute nausea (AUC=69% vs. 58%), delayed nausea (AUC=68% vs. 53%), and delayed vomiting (AUC=77% vs. 54%). Conclusions: Gut microbial diversity did not differ between patients who self-reported acute or delayed nausea and those who did not, but risk prediction models for acute nausea, delayed nausea, and delayed vomiting were all improved by incorporating differentially abundant bacterial species. Results suggest that the gut microbiome is an important factor associated with the development of CINV.