Identification of Phenotypic Lipidomic Signatures in Response to Long Chain n-3 Polyunsaturated Fatty Acid Supplementation in Humans

Abstract

Background Supplementation with long chain n-3 polyunsaturated fatty acids is used to reduce total circulating triacylglycerol (TAG) concentrations. However, in about 30% of people, supplementation with long chain n-3 polyunsaturated fatty acids does not result in decreased plasma TAG. Lipidomic analysis may provide insight into this inter-individual variability. Methods Lipidomic analyses using targeted, mass spectrometry were performed on plasma samples obtained from a clinical study in which participants were supplemented with 3 g/day of long chain n-3 in the form of fish oil capsules over a 6-week period. TAG species and cholesteryl esters (CE) were quantified for 130 participants pre- and post-supplementation. Participants were segregated into 3 potential responder phenotypes: (1) positive responder (Rpos; TAG decrease), (2) non-responder (Rnon; lacking TAG change), and (3) negative responder (Rneg; TAG increase) representing 67%, 18%, and 15% of the study participants, respectively. Separation of the 3 phenotypes was attributed to differential responses in TAG with 50 to 54 carbons with 1 to 4 desaturations. Elevated TAG with higher carbon number and desaturation were common to all phenotypes following supplementation. Using the TAG responder phenotype for grouping, decreases in total CE and specific CE occurred in the Rpos phenotype versus the Rneg phenotype with intermediate responses in the Rnon phenotype. CE 20:5, containing eicosapentaenoic acid (20:5n-3), was elevated in all phenotypes. A classifier combining lipidomic and genomic features was built to discriminate triacylglycerol response phenotypes and reached a high predictive performance with a balanced accuracy of 75%. Conclusions These data identify lipidomic signatures, TAG and CE, associated with long chain n-3 response p henotypes and identify a novel phenotype based upon CE changes. Registration URL: https://www.ClinicalTrials.gov; Unique Identifier: NCT01343342.

Keywords: genetics; lipidomics; lipids; mass spectrometry; nutrigenomics.

Conflict of interest statement

None.

Figures

Figure 1. Distribution of triacylglycerol (TAG) response phenotypes and underlying TAG species differences.

The distribution of changes in plasma TAG concentrations following LCn‐3 supplementation for participants (A). Sparse partial least squares discriminant analysis plot demonstrating separation of response phenotypes (B). TAG species and their phenotype groups differences underlying separation along Component 1 of the sparse partial least squares discriminant analysis plot (C). TAG species and their phenotype groups differences underlying separation along component 2 of the sparse partial least squares discriminant analysis plot (D). A scale bar for relative differences for the Loadings plots is provided. Phenotypes: Positive responder (sum TAG decrease < −10%; n=87), non‐responder (sum TAG changes +/‐ 10%; n=24), and negative responder (sum TAG increase >10%; n=19). Rpos indicates positive responder; Rnon, non‐responder; and Rneg, negative responder.

Figure 2. Heatmap analysis of change in triacylglyercol (TAG) species by response phenotype.

The percentage of individuals within a given phenotype (y axis) vs the qualitative change in the specific TAG specie (x axis) is shown. A specific TAG was defined as increased (% change>10%), decreased (% change 10%; n=19). Rpos indicates positive responder; Rnon, non‐responder; and Rneg, negative responder.

Figure 3. Triacylglcyerol (TAG) response phenotype distinguishes plasma cholesterol ester (CE) responses following LCn‐3 supplementation.

Plasma CE responses for total and individual CE were calculated as a percentage change from baseline following long chain n‐3 supplementation. Individual participant data points are provided. Comparisons of phenotypic response for total CE and individual CE were performed by one‐way ANOVA with Tukey post hoc test. Columns without a common superscript letter differ, P≤0.05, with applied false discovery rate. Positive responder (n=87), non‐responder (n=24), and negative responder (n=19). Rpos indicates positive responder; Rnon, non‐responder; and Rneg, negative responder.

Figure 4. Correlations between percent change of triacylglycerol (TAG) and cholesterol ester (CE) species following long chain n‐3 polyunsaturated fatty acid (LCn‐3) supplementation and the genetic risk score.

The percentage change of each TAG and CE species is displayed in the outer layer by the red (percentage decrease) and blue bars (percentage increase). Labels highlighted in red represent those TAG identified as loadings of the factor 1 in the sparse partial least squares discriminant analysis. Significant Pearson correlation coefficients (r) are represented in the inner layer by the green bars. The asterisks represent Pearson correlation P values: *P<0.05, **P<0.01, ***P<0.001.