Fig. 2: Heterogeneity in metabolism and determination of the dominant metabolic subphenotype.
a, Detailed illustration of the four metabolic measures for our cohort sorted by the HbA1c. b, Muscle IR is categorized on the basis of SSPG (mg dl−1) values (IS when SSPG ≤ 120 and IR when SSPG > 120). c, β-cell function is categorized according to disposition index (DI) ((pmol dl)/(kg ml)): normal when DI > 2.2, intermediate when 1.2 ≤ DI ≤ 2.2, dysfunction when DI < 1.2. d, Incretin effect is categorized on the basis of the measured incretin effect (IE, %): normal when IE > 64, intermediate when 39 ≤ IE ≤ 64, and dysfunction when IE < 39. e, Hepatic IR is categorized on the basis of the hepatic-IR-index: IS when hepatic-IR-index > 3.95, intermediate when 3.95 ≤ hepatic-IR-index ≤ 4.8, and IR when hepatic-IR-index > 4.8. f, Heat map showing the standardized deviance score of each metabolic measure for each participant (Methods). β-cell function and incretin effect values have been reversed so that higher positive values represent a greater abnormality. High positive deviance (darker red) indicates greater deviance from study population average in the abnormal direction (for example, higher IR or lower incretin effect), while high negative deviance (lighter yellow) indicates a healthier metabolic phenotype than the average population. g, Pairwise correlation network showing the association between the standardized deviance of the four metabolic measures. Pearson correlation coefficient (−1<r < 1) was used to determine the strength of each relationship, where r = 1 ‘dark orange’ means strong positive correlation, and r = −1 ‘dark cyan’ means strong negative correlation. The asterisk symbol on edges means that the correlation is statistically significant (P < 0.05) between the two corresponding metabolic subphenotypes. h, T2D polygenic risk score (PRS) for all individuals in our cohort sorted in descending order and coloured by the glycaemic status measured using HbA1c. P values in g and h were calculated using t-test for Pearson correlations (evaluating the null hypothesis that the correlation is zero).
