Dynamic microbiome and metabolome analyses reveal the interaction between gut microbiota and anti-PD-1 based...

Hepatocellular carcinoma (HCC) is a severe disease with high mortality and global incidence. However, the interaction between the gut microbiome and combined immunotherapy for HCC is yet unclear. In this prospective clinical study, patients with unresectable HCC who had not received systemic treatment previously were recruited. Fecal and serum samples were collected at the baseline point and before each subsequent administration as specified. Between 20 October 2019 and 2 February 2021, 61 patients were screened for eligibility, of whom 35 patients were finally included in this study. Alpha diversity of fecal samples from patients who responded to immunotherapy was higher than that of nonresponders at baseline. However, the prominent alpha-diversity between responders and nonresponders became similar as early as week 6 after treatment. The beta diversity of intergroup did not show significant difference at the ninth week after treatment. Alpha-d-Glucose was the only serum metabolite that differed between the responders and nonresponders after 3 months. Responder-enriched Ruminococcus showed a positive correlation with serum galactaric acid, while Klebsiella was positively associated with 3-methylindole and lenticin (all P < .01). The machine learning classifier based on serum metabolites were more able to discriminate HCC patients who potentially benefited from immunotherapy at baseline (AUC 0.793, 95% CI: 0.632-0.954) than the classifier of gut microbiome. In conclusion, gut microbiome biomarkers are associated with the response to anti-PD-1 based immunotherapy in HCC patients. Classifiers based on gut microbiota and serum metabolites are feasible.