Table 3.
Metabolites Significantly Associated with Southern Dietary Patterns in the Discovery and Replication Samplesa
| Discovery Sample | Replication Sample | |||
|---|---|---|---|---|
| Dietary Patternb | Metabolite | β (95% CI)c | β (95% CI)c | P-value |
| Meat and fast food | indole-3-propanoic acid | −0.170 (−0.253, −0.087) | −0.143 (−0.274, −0.013) | 0.03 |
| C18:0 LPE B | −0.138 (−0.222, −0.054) | |||
| C22:6 LPC | −0.142 (−0.226, −0.059) | |||
| C24:0 LPC | −0.132 (−0.212, −0.051) | −0.139 (−0.263, −0.014) | 0.02 | |
| C34:2 PE plasmalogen | 0.139 (0.055, 0.224) | 0.218 (0.085, 0.351) | 0.001 | |
| C36:5 PE plasmalogen | 0.127 (0.043, 0.211) | 0.236 (0.106, 0.366) | 0.0004 | |
| C38:5 PE plasmalogen | 0.173 (0.090, 0.257) | 0.311 (0.179, 0.444) | 4.6×10−6 | |
| cotinine | 0.102 (0.043, 0.160) | 0.117 (0.029, 0.205) | 0.01 | |
| creatine | 0.120 (0.044, 0.196) | |||
| diacetylspermine | 0.155 (0.075, 0.236) | |||
| hydroxyproline | 0.167 (0.085, 0.248) | 0.250 (0.127, 0.373) | 7.6×10−5 | |
| N-acetylleucine | 0.139 (0.058, 0.220) | |||
| N-methyl proline | −0.149 (−0.233, −0.066) | −0.143 (−0.274, −0.012) | 0.03 | |
| proline betaine | −0.156 (−0.238, −0.074) | −0.248 (−0.369, −0.126) | 7.5×10−5 | |
| Fish and vegetables | DHA | 0.148 (0.072, 0.224) | 0.217 (0.113, 0.321) | 4.9×10−5 |
| 1,7-dimethyluric acid | −0.233 (−0.302, −0.165) | −0.132 (−0.230, −0.034) | 0.009 | |
| caffeine | −0.239 (−0.308, −0.170) | |||
| C22:6 lysophosphatidylethanolamine (LPE) | 0.137 (0.072, 0.201) | 0.148 (0.050, 0.245) | 0.003 | |
CI, confidence interval
Metabolites that were significantly associated with Southern dietary patterns in the discovery sample using a false discovery rate (FDR) <0.05 to determine statistical significance. The table also presents results for metabolites that were statistically significantly associated with Southern dietary patterns in the replication sample using P<0.05 to determine statistical significance.
No metabolites were statistically significantly associated with the starchy foods dietary pattern.
Multivariable linear regression models, adjusted for age, sex, education, smoking, physical activity, alcohol, estimated glomerular filtration rate, multivitamin use, batch, and body mass index, were used to calculate β coefficients and 95% confidence intervals for the association between metabolites and dietary patterns.