Background: Primary liver cancer has high mortality and morbidity worldwide. However, the characteristic of gut microbiota profile and its correlation with inflammation status in liver cancer patients remains largely unknown, and a gut microbiome-based diagnostic model for liver cancer is still absent.
Methods: Here, we provided a comprehensive analysis based on fecal 16S rRNA sequencing and clinical data in a cohort consisting of 40 healthy volunteers, 143 hepatocellular carcinoma (HCC) patients, and 46 cholangiocarcinoma (CCA) patients.
Results: Our results indicated a distinct shift of gut microbiota composition between two primary liver cancer types and compared with healthy volunteers. Based on the diversity constitute of gut microbiome taxonomy and random forest algorithm, eight genera with mean abundance above 0.1% were selected to construct the classification model with half of the randomly selected cohort. Based on this signature, high diagnostic accuracy in the validation cohort to classify liver cancer types (AUC = 0.989, 0.967, 0.920 for Control, HCC, CCA separately) was achieved. Further analysis showed increased Gram-negative bacteria and elevated inflammatory response markers in CCA group versus HCC group. The correlation analysis between inflammatory response markers and composition of gut microbiome revealed decreased potentially beneficial genus and increased opportunistic pathogens positively correlated with adverse prognostic inflammatory response markers.
Conclusion: Generally, our study established the gut microbiome-based signature for liver cancer prediction and screening and revealed that gut microbiome characteristic in primary liver cancer was correlated with adverse inflammatory response markers in liver cancer.
Keywords: 16 s rRNA sequencing; Cholangiocarcinoma; Diagnostic model; Gut microbiota; Hepatocellular carcinoma; Inflammatory; Noninvasive; Patient screening; Primary liver cancer; Random forest.
© 2021. Asian Pacific Association for the Study of the Liver.