作者
Magdalena Stępień,Pekka Keski‐Rahkonen,Agnèta Kiss,Nivonirina Robinot,Talita Duarte‐Salles,Neil Murphy,Gabriel Perlemuter,Vivian Viallon,Anne Tjønneland,Agnetha Linn Rostgaard‐Hansen,Christina C. Dahm,Kim Overvad,Marie‐Christine Boutron‐Ruault,Francesca Romana Mancini,Yahya Mahamat‐Saleh,Krasimira Aleksandrova,Rudolf Kaaks,Tilman Kühn,A. Trichopoulou,Anna Karakatsani,Salvatore Panico,Rosario Tumino,Domenico Palli,Giovanna Tagliabue,Alessio Naccarati,Roel Vermeulen,H. B. Bueno-de-Mesquita,Elisabete Weiderpass,Guri Skeie,J. Ramón Quirós,Eva Ardanáz,Olatz Mokoroa,Núria Sala,María‐José Sánchez,José María Huerta,Anna Winkvist,Sophia Harlid,Bodil Ohlsson,Klas Sjöberg,Julie A. Schmidt,Nicholas J. Wareham,Kay‐Tee Khaw,Pietro Ferrari,Joseph A. Rothwell,Marc J. Gunter,Elio Ríboli,Augustin Scalbert,Mazda Jenab
摘要
Hepatocellular carcinoma (HCC) development entails changes in liver metabolism. Current knowledge on metabolic perturbations in HCC is derived mostly from case-control designs, with sparse information from prospective cohorts. Our objective was to apply comprehensive metabolite profiling to detect metabolites whose serum concentrations are associated with HCC development, using biological samples from within the prospective European Prospective Investigation into Cancer and Nutrition (EPIC) cohort (>520 000 participants), where we identified 129 HCC cases matched 1:1 to controls. We conducted high-resolution untargeted liquid chromatography-mass spectrometry-based metabolomics on serum samples collected at recruitment prior to cancer diagnosis. Multivariable conditional logistic regression was applied controlling for dietary habits, alcohol consumption, smoking, body size, hepatitis infection and liver dysfunction. Corrections for multiple comparisons were applied. Of 9206 molecular features detected, 220 discriminated HCC cases from controls. Detailed feature annotation revealed 92 metabolites associated with HCC risk, of which 14 were unambiguously identified using pure reference standards. Positive HCC-risk associations were observed for N1-acetylspermidine, isatin, p-hydroxyphenyllactic acid, tyrosine, sphingosine, l,l-cyclo(leucylprolyl), glycochenodeoxycholic acid, glycocholic acid and 7-methylguanine. Inverse risk associations were observed for retinol, dehydroepiandrosterone sulfate, glycerophosphocholine, γ-carboxyethyl hydroxychroman and creatine. Discernible differences for these metabolites were observed between cases and controls up to 10 years prior to diagnosis. Our observations highlight the diversity of metabolic perturbations involved in HCC development and replicate previous observations (metabolism of bile acids, amino acids and phospholipids) made in Asian and Scandinavian populations. These findings emphasize the role of metabolic pathways associated with steroid metabolism and immunity and specific dietary and environmental exposures in HCC development.