Skip to main content
. 2022 Feb 2;12(2):139. doi: 10.3390/metabo12020139

Figure 6.

Figure 6

Inter-peak correlations and abundance by study arm and timepoint of peaks changed by CSM from baseline to month 6. (A) Heatmap of correlations between PQN-normalised peak intensity values for 14 selected metabolites. Metabolites were selected on the basis of being identified as strongly changed by CSM from baseline to month 6 by multivariate PLS models. Additional metabolites were manually selected on the basis of belonging to similar functional groupings to those metabolites identified by PLS models. Heatmap rows and columns depict individual metabolites, with Pearson r correlation values presented in the heatmap cells. Functional groupings of metabolites are identified by the row and column annotations as follows: blue = metabolites reflective of changes in gut microbial metabolism or gut transport after RYGB; grey = metabolites reflective of BCAA catabolism; brown = aromatic amino acids. Gaps in heatmap rows and columns demarcate clusters of metabolites based on correlation coefficients. (B) PQN-normalised abundance by study arm and timepoint of the 14 metabolites outlined in the heatmap in panel (a). Number of samples: CSM0, n = 26; CSM6, n = 19; MTA0, n = 28; MTA6, n = 24. The top 2 rows (blue panels) indicate metabolites reflective of changes in gut microbial metabolism or gut transport after RYGB. The bottom 2 rows indicate aromatic amino acids (brown panels) and metabolites reflective of BCAA catabolism (grey panels). p-values derived from unpaired t-tests of baseline and month 6 differences in peak intensity in the CSM and MTA arms are presented. p-values were multiplicity-corrected for the number of comparisons using the Benjamini–Hochberg method. BCAA, branched-chain amino acid; CSM, combined metabolic surgery plus medical therapy; CSM0, baseline samples from CSM group; CSM6, month 6 samples from CSM group; MTA, medical therapy alone; MTA0, baseline samples from MTA group; MTA6, month 6 samples from MTA group; PLS, partial least squares; PQN, probabilistic quotient normalization; RYGB, Roux-en-Y gastric bypass surgery.