Pre-Diagnostic Targeted Metabolomic Profile Of Hepatocellular Carcinoma Risk In A Nested Case-Control Study Within EPIC Cohort

Magdalena STEPIEN, International Agency for Research on Cancer (IARC-WHO), France
ACHAINTRE D. 1 , DUARTE-SALLES T. 1 , PERRIER F. 1 , ASSI N. 1 , FEDIRKO V. 2 , ROMIEU I. 1 , SCALBERT A. 1 , JENAB M. 1

1 International Agency for Research on Cancer (IARC-WHO), Lyon, France
2 Rollins School of Public Health, Winship Cancer Institute, Emory University, Atlanta, USA

 
Purpose: Hepatocellular carcinoma (HCC) is highly malignant with increasing incidence rates in Western countries. We aimed to identify metabolite alterations associated with HCC development in the EPIC cohort.
Methods: In a nested case-control study with 147 HCC cases matched to 1 control each, 141 metabolites were measured in prospectively collected sera using Biocrates AbsoluteIDQ p180Kit. Metabolite levels were log-transformed and standardized, and odds ratios and 95% confidence intervals were calculated with multivariable conditional logistic regression. Bonferroni correction for multiple testing was applied. In order to identify the most consistently predictive metabolites, we subsequently applied and compared four additional methods for variable selection: (i) PCA; (ii) variable importance in projection in PLSDA; (iii) stepwise selection in logistic regression and (iv) elastic net analysis. Receiver operating characteristic (ROC) curves were constructed to estimate discriminatory accuracy of the metabolites that appeared as significant in at least 4 methods.
Results: Bonferroni correction showed significant HCC risk associations for 31 metabolites: 3 amino acids (glutamine, glutamate and tyrosine), 20 glycerophospholipids (8 lyso phosphatidylcholin(PC): c16:0, c17:0, c18:0, c18:1, c18:2, c20:3, c20:4, c28:1; 8 diacyl PC: c28:1, c32:3, c34:4, c36:4, c36:6, c38:5, c28:6, c42:5; 4 acyl-alkyl PC: c30:2, c36:1, c38:0, c40:1), 4 sphingomyelins (c16:1, c18:0, c18:1, c20:2) and 4 hydroxysphingomyelins (c14:1, c16:1, c22:1, c22:2). In all 5 statistical analyses glutamate, diacyl PC c42:5, lysoPCs c17:0 and c20:4 appeared as predictors of HCC development, while PC diacyl c32:3 and acyl-alkyl c38:0 and sphingomyelin c18:1 appeared in 4 out of 5 analyses. Based on ROC, discriminatory power of these 7 metabolites was 89%.
Conclusions: Our findings indicate that metabolic alterations, especially those related to glycerophospholipids and glutamate metabolism, are involved in HCC development. To identify additional novel metabolic pathways in HCC development, we are currently conducting an untargeted metabolomic analysis of the same samples.
Funding source:French National Cancer Institute