Prediagnostic Plasma Nutrimetabolomics and Prostate Cancer Risk: A Nested Case-Control Analysis Within the EPIC Study

Abstract

Background and Objective: Nutrimetabolomics may reveal novel insights into early metabolic alterations and the role of dietary exposures on prostate cancer (PCa) risk. We aimed to prospectively investigate the associations between plasma metabolite concentrations and PCa risk, including clinically relevant tumor subtypes. Methods: We used a targeted and large-scale metabolomics approach to analyze plasma samples of 851 matched PCa case-control pairs from the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. Associations between metabolite concentrations and PCa risk were estimated by multivariate conditional logistic regression analysis. False discovery rate (FDR) was used to control for multiple testing correction. Results: Thirty-one metabolites (predominately derivatives of food intake and microbial metabolism) were associated with overall PCa risk and its clinical subtypes (p < 0.05), but none of the associations exceeded the FDR threshold. The strongest positive and negative associations were for dimethylglycine (OR = 2.13; 95% CI 1.16-3.91) with advanced PCa risk (n = 157) and indole-3-lactic acid (OR = 0.28; 95% CI 0.09-0.87) with fatal PCa risk (n = 57), respectively; however, these associations did not survive correction for multiple testing. Conclusions: The results from the current nutrimetabolomics study suggest that apart from early metabolic deregulations, some biomarkers of food intake might be related to PCa risk, especially advanced and fatal PCa. Further independent and larger studies are needed to validate our results.

Keywords: EPIC; nested case–control; nutrimetabolomics; prostate cancer.

Figure 1

Volcano plot showing the magnitude of associations (odds ratio [OR]) and statistical significance (−log10 p values) for plasma metabolites and overall prostate cancer risk in the nested case–control within the EPIC cohort. Significant associations (above the dotted horizontal line) were set at −log10 p-values ≥ −1.30 (p ≤ 0.05).

Figure 2

Volcano forest plot showing odds ratios and 95% CI for the associations between plasma metabolites and five relevant clinical tumor subtypes of prostate cancer in the nested case–control study within the EPIC cohort. Only significant associations at p < 0.05 (not FDR adjusted) are shown. Abbreviations: 3-MPPAG = 3-(3′-methoxyphenyl)propanoic acid-4′-glucuronide; 3-MPPAS = 3-(3′-methoxyphenyl)propanoic acid-4′-sulfate; 4-HHPPA = (R/S)-2-hydroxy-3-(4′-hydroxyphenyl)propanoic acid; 6-AMMU = 6-amino-5-(N-methylformylamino)-1-methyluracil; CMPF = 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid; DHPPA = 3-(3′,4′-dihydroxyphenyl)propanoic acid; DHSBA = 3,5-dihydroxy-4-(sulfooxy)benzoic acid; HPAA sulfate = N-(2-hydroxyphenyl)-acetamide sulfate; TMAO = trimethylamine N-oxide.

Figure 3

Heatmap showing significant (p < 0.05) Spearman’s correlations between plasma concentrations of 31 metabolites and the habitual intake of selected foods and food groups in the nested prostate cancer case–control study within the EPIC cohort. Full data on rho coefficients and statistical significance for each pair of correlations are shown in Supplementary Table S2. Abbreviations: 3-MPPAG = 3-(3′-methoxyphenyl)propanoic acid-4′-glucuronide; 3-MPPAS = 3-(3′-methoxyphenyl)propanoic acid-4′-sulfate; 4-HHPPA = (R/S)-2-hydroxy-3-(4′-hydroxyphenyl)propanoic acid; 6-AMMU = 6-amino-5-(N-methylformylamino)-1-methyluracil; CMPF = 3-carboxy-4-methyl-5-propyl-2-furanpropionic acid; DHPPA = 3-(3′,4′-dihydroxyphenyl)propanoic acid; DHSBA = 3,5-dihydroxy-4-(sulfooxy)benzoic acid; HPAA sulfate = N-(2-hydroxyphenyl)-acetamide sulfate; TMAO = trimethylamine N-oxide.