Multikingdom and functional gut microbiota markers for cholangiocarcinoma

While the association between gut microbiota and cholangiocarcinoma (CCA) has been explored, existing research has been limited to bacterial components. The alterations in non-bacterial members (fungi, archaea, and viruses) and their functional characteristics in CCA remain unclear. This study aimed to comprehensively characterize the multi-kingdom gut microbiota and functional profiles in CCA patients and develop a non-invasive diagnostic model.

We performed metagenomic sequencing on 700 fecal samples collected from three distinct regions in China, comprising 289 CCA patients, 312 healthy controls, and 99 hepatocellular carcinoma (HCC) patients. Multi-kingdom microbial composition and functional profiles were analyzed. We integrated multiple machine learning algorithms to generate 117 distinct modeling schemes for constructing and screening the optimal CCA diagnostic model. The model was validated across three independent cohorts and tested for specificity against HCC. We also benchmarked the model against established diagnostic approaches by reviewing 15 CCA clinical practice guidelines/consensus statements and their cited evidence and by evaluating model detection in CA19-9-positive and CA19-9-negative patients.

Extensive alterations were identified in CCA patients across multi-kingdom and functional layers, including 198 archaeal, 154 bacterial, 78 fungal, and 120 viral species, as well as 278 genes, 99 pathways, and 151 carbohydrate-active enzymes (CAZymes). A panel of 21 multi-kingdom and functional markers demonstrated superior diagnostic accuracy with an average AUC of 91% across three validation cohorts, outperforming any single-kingdom model. The model exhibited high specificity in distinguishing CCA from HCC, achieving an AUC of 99.74% with a false positive rate of 11.11%. Benchmarking supported a complementary role as a noninvasive screening aid, the model showed pooled sensitivity and specificity of 96.2%, and 78.5%, respectively, and detected 97.6% of CA19-9-negative CCA cases.

This study represents the first comprehensive characterization of multi-kingdom gut microbiome and functional markers in cholangiocarcinoma. Gut multi-kingdom and functional microbial signatures represent a promising noninvasive tool that may aid the screening and early identification of cholangiocarcinoma in high-risk populations, complementing existing diagnostic approaches.