The intestinal microbiota of patients with PSC are different from healthy controls and patients with ulcerative colitis across geographical regions

Question:

Primary sclerosing cholangitis (PSC) is a cholestatic liver disease of unknown origin which is highly associated to inflammatory bowel disease (IBD). Single centre studies found characteristic changes in the faecal microbiome of patients with PSC. However, it is unknown whether PSC is associated with a common microbial signature across geographical regions.

Methods:

Stool samples were collected from two independent cohorts from Norway and Germany. The pooled patient population included 388 patients at a median age of 47 (range, 15 – 78), 137 patients with PSC (75 with concurrent IBD), 88 patients with ulcerative colitis (UC) and 133 healthy controls (HC). The microbiomes were profiled by 16S ribosomal RNA gene sequencing targeting the V1-V2 region, biodiversity analysis, machine learning classification and receiver operating characteristic (ROC) analysis.

Results:

On the regional level as well as in the pooled cohort, the beta diversity measured by the Bray-Curtis dissimilarity was significantly different between PSC and HC (PERMANOVA p < 0.001) as well as between PSC and UC (PERMANOVA p < 0.001). In each test, the disease state had the strongest impact on microbial beta diversity after adjusting for age, gender, BMI and smoking status. No differences in beta diversity were observed between PSC with and without colitis. OTUs of the genera Lactobacillus and Propionibacterium were enriched in PSC compared to both HC and UC. Regarding the alpha diversity, there were no differences between PSC and HC in the German cohort, but a decreased alpha diversity in the Norwegian cohort in patients with PSC (Chao1 p = 0.02; Shannon index p = 0.01). We next tried to assess, whether the faecal microbial profile can be used to differentiate between healthy and diseased individuals. A random forest classifier trained on the microbial signature of 20 genera in the German cohort yielded an area under the curve (AUC) of 0.81 classifying the Norwegian population in PSC versus HC. Additionally, we trained random forests on stratified randomly split samples from the whole population including Norwegian and German patients. The classifiers achieved an AUC = 0.87 classifying between PSC versus HC (using 20 genera) and an AUC = 0.82 classifying PSC versus UC (using 8 genera) in the respective validation samples.

Conclusion:

The faecal microbiome community-structure of patients with PSC is distinct from both HC and UC and allows a diagnostic classification which is accurate across different geographical areas. The faecal microbiome is similar between PSC patients with or without associated colitis, strongly suggesting that the liver disease drives the observed differences.