Table 4.
Multivariable models. Association between principal components loads, individual metabolites with high reliability in bootstrap resampling and adverse cardiac outcomes. Left side—regression models built with PC loads and clinical covariates only, reported are the variables included in the final regression model. Right side—regression models built with PC loads and clinical covariates (if statistically significant) and the highly reliable metabolites.
| PC Only | PC + Highly Reliable Metabolites | |||||
|---|---|---|---|---|---|---|
| Outcomes | PC Loads | Clinical Covariates | QIC (PC) | PC Loads | Significant Metabolites | QIC (mets) |
| O2 saturation < 95% | 4 | SV, SD | 95.6 | ADMA/Arg, serotonin/tryptophan, GCDCA, Betaine | 84.6 | |
| Cardiac dysfunction | 4 | None | 119.6 | 2,5 | Citrulline, Cystine, 1-Methylhistidine | 103.4 |
| RV dysfunction | 4 | SRV, dyslipidemia | 79.9 | 4 | DHEAS | 76.4 |
| Arrhythmia history | 2,4 | SRV, SV, age | 179.3 | 2 | TCA, Orn/Arg | 175.4 |
| Shortness of breath | 1,4 | None | 143.0 | Ornithine, SDMA/Arg, Homo-L-arginine | 120.9 | |
ADMA—asymmetric dimethyl arginine; Arg—arginine; DHEAS—dehydroepiandrosterone; GCDCA—glycochenodeoxycholic acid; Orn—ornithine; SDMA—symmetric dimethyl arginine; TCA—Taurocholic acid.