TABLE 3.
Contribution of groups of metabolites to the association between the Mediterranean diet and cardiovascular disease risk factors: the Fenland Study, 2005–20151
| HDL-C, mmol/L | Cholesterol:HDL-C ratio | Triglycerides, % | HOMA-IR, % | |||||
|---|---|---|---|---|---|---|---|---|
| Difference (95% CI)2 | % Attenuation3 | Difference (95% CI)2 | % Attenuation3 | Difference (95% CI)2 | % Attenuation3 | Difference (95% CI)2 | % Attenuation3 | |
| Reference model3 | 0.02 (0.01, 0.02) | −0.05 (−0.07, −0.02) | −1.99 (−2.99, −0.99) | −3.42 (−4.43, −2.40) | ||||
| Adjusted for metabolite subclasses | ||||||||
| + acylcarnitines | 0.02 (0.01, 0.02) | −2.0 | −0.03 (−0.05, −0.00) | 44.7** | −1.01 (−1.99, −0.01) | 48.3** | −2.99 (−4.03, −1.94) | 11.3*** |
| + amino acids/biogenic amines | 0.01 (0.01, 0.02) | 17.0 | −0.05 (−0.07, −0.02) | 1.9 | −1.98 (−3.03, −0.91) | −4.6 | −2.21 (−3.30, −1.11) | 26.4** |
| + lysophosphatidylcholines | 0.02 (0.01, 0.03) | −12.6 | −0.03 (−0.06, −0.01) | 26.0 | −0.63 (−1.58, 0.32) | 67.1* | −3.34 (−4.40, −2.39) | −1.1 |
| + phosphatidylcholine acyl-alkyls | 0.01 (0.01, 0.02) | 8.2 | −0.04 (−0.07, −0.02) | 7.3 | −1.81 (−2.74, −0.87) | 2.3 | −2.52 (−3.64, −1.39) | 27.0*** |
| + phosphatidylcholine diacyls | 0.02 (0.01, 0.02) | −1.7 | −0.06 (−0.08, −0.03) | −4.3 | −1.40 (−2.37, −0.43) | 37.8* | −2.68 (−3.82, −1.53) | 20.8** |
| + sphingolipids | 0.02 (0.01, 0.02) | 7.7 | −0.03 (−0.05, −0.01) | 31.1** | −1.32 (−2.32, −0.29) | 32.0 | −3.02 (−4.04, −1.98) | 8.4** |
| Adjusted for the metabolite score | ||||||||
| + metabolite score4 | 0.01 (0.00, 0.02) | 20.6 | −0.03 (−0.06, −0.01) | 35.6 | −2.01 (−3.37, −0.63) | 8.9 | −1.76 (−3.12, −0.37) | 37.2* |
Total n = 10,806. *P < 0.05, **P < 0.01, ***P < 0.001 for percentage changes. HDL-C, HDL cholesterol.
Values represent unit or percentage differences in cardiovascular disease risk factors [β coefficients or exp(β) and corresponding 95% CIs]; and percentage change in β (see footnote 3). Regression models were fitted with adjustment for age, sex, test site, education level, income, occupation, medication use, family history of diabetes, objectively measured physical activity, smoking, BMI, and waist circumference. Adherence to the Mediterranean diet was not significantly associated with systolic blood pressure, diastolic blood pressure, and 2-h glucose (P > 0.05) and therefore these phenotypes were not considered as potential mediation (see results in Supplemental Table 2).
Percentage changes in β coefficients were calculated as changes in β coefficients from those of the “reference model” upon statistical adjustment for the metabolite score or metabolite subclasses (mediation analysis). As availability of metabolites varied, sample sizes were different across models: reference model, n = 10,806; + acylcarnitines, n = 10,701; + amino acids/biogenic amines, n = 9224; + lysophosphatidylcholines, n = 10,718; + phosphatidylcholine acyl-alkyls, n = 9868; + phosphatidylcholine diacyls, n = 8946; + sphingomyelins, n = 10,690; + metabolite score, n = 7138.
Metabolite score included all 66 metabolites associated with the Mediterranean diet derived in a random half of the total data set and validated in the second half, weighted by their respective regression coefficients.