Prospective investigation of the blood metabolome to identify risk biomarkers for biliary tract cancer

Biliary tract cancer (BTC) is a group of rare and aggressive malignancies; most (~80%) patients are diagnosed with metastatic disease, which has a five-year survival rate of 3%. A better understanding of BTC etiology and biology is critical for designing cost-effective prevention strategies. Many known risk factors for BTC, such as obesity, diabetes, and chronic inflammation, are related to metabolic disturbance, suggesting a significant role of metabolic perturbation in BTC pathogenesis. Thus, a systemic investigation of circulating metabolites could provide valuable information regarding biological mechanisms of BTC etiology and identify potential risk biomarkers.We conducted an individually matched case-control study nested in eight prospective cohort studies from Australia, China, and the United States, and evaluated associations of circulating metabolites in pre-diagnostic blood samples with BTC risk. Metabolites were measured in two batches using Metabolon's Global Discovery Panel assay . A total of 1,997 biochemicals were detected, of which 1,593 were detected in ≥1 study. We identified 146 principal components that account for 99.5% of the metabolomic variance in 1,593 metabolites in our data. Thus, a p-value of 3.42x10 -04 (i.e., 0.05 /146) was set as the significance threshold for multiple comparison adjustment. Of the 1,593, 1,466 (1,202 known and 264 unknown) were present in ≥70% of participants and were analyzed herein. Within each cohort, metabolite levels were trimmed to three standard deviations, undetectable values were imputed with the minimum, log-transformed (x+1.0), and z-scored. We conducted conditional logistic regression by assay batch and meta-analyzed results together with fixed-effects meta-analysis.This study included 1,383 total participants, comprising 836 controls matched to 547 BTC cases: 33.1% gallbladder cancer, 15.0% intrahepatic cholangiocarcinoma (CCA), 22.7% extrahepatic CCA, and 29.3% Ampulla of Vater and unspecified BTC cases. Study participants had a median baseline age of 61 [interquartile range (IQR) 12] years old, 57.1% were female, and had a median [IQR] 9.6 [4.1] years from blood collection to BTC diagnosis. We found multiple metabolites associated with BTC risk, including those in the glutamate, sphingomyelin, and bile acid pathways, after adjusting for multiple comparisons. Selected significant associations (OR (95% CI) for every 1-SD increase, p-value) with BTC are glutamate: 1.32 (1.17-1.49), p=6.3x10 -06 ; sphingomyelin (10): 0.76 (0.67-0.86), p=3.3x10 -05 ; and glycodeoxycholate 3-sulfate: 1.23 (1.10-1.38), p=2.72x10 -04 .This is the first prospective study to systematically search the blood metabolome for biomarkers of BTC risk. We provide strong evidence that bile acid and other metabolic pathways play critical roles in BTC etiology. We are expanding the analysis with data from additional cohort studies to further validate the findings.