OMICs Signatures Linking Persistent Organic Pollutants to Cardiovascular Disease in the Swedish Mammography Cohort.

Cardiovascular disease (CVD) development may be linked to persistent organic pollutants (POPs), including organochlorine compounds (OCs) and perfluoroalkyl and polyfluoroalkyl substances (PFAS). To explore underlying mechanisms, we investigated metabolites, proteins, and genes linking POPs with CVD risk. We used data from a nested case-control study on myocardial infarction (MI) and stroke from the Swedish Mammography Cohort - Clinical (n = 657 subjects). OCs, PFAS, and multiomics (9511 liquid chromatography-mass spectrometry (LC-MS) metabolite features; 248 proteins; 8110 gene variants) were measured in baseline plasma. POP-related omics features were selected using random forest followed by Spearman correlation adjusted for confounders. From these, CVD-related omics features were selected using conditional logistic regression. Finally, 29 (for OCs) and 12 (for PFAS) unique features associated with POPs and CVD. One omics subpattern, driven by lipids and inflammatory proteins, associated with MI (OR = 2.03; 95% CI = 1.47; 2.79), OCs, age, and BMI, and correlated negatively with PFAS. Another subpattern, driven by carnitines, associated with stroke (OR = 1.55; 95% CI = 1.16; 2.09), OCs, and age, but not with PFAS. This may imply that OCs and PFAS associate with different omics patterns with opposite effects on CVD risk, but more research is needed to disentangle potential modifications by other factors.

The association between plasma metabolites and future risk of all-cause mortality.

BACKGROUND: Metabolite profiles provide snapshots of the overall effect of numerous exposures accumulated over life courses, which may lead to health outcomes in the future. OBJECTIVE: We hypothesized that the risk of all-cause mortality is linked to alterations in metabolism earlier in life, which are reflected in plasma metabolite profiles. We aimed to identify plasma metabolites associated with future risk of all-cause mortality. METHODS: Through metabolomics, 110 metabolites were measured in 3833 individuals from the Malmö Diet and Cancer-Cardiovascular Cohort (MDC-CC). A total of 1574 deaths occurred within an average follow-up time of 22.2 years. Metabolites that were significantly associated with all-cause mortality in MDC-CC were replicated in 1500 individuals from Malmö Preventive Project re-examination (MPP), among whom 715 deaths occurred within an average follow-up time of 11.3 years. RESULTS: Twenty two metabolites were significantly associated with all-cause mortality in MDC-CC, of which 13 were replicated in MPP. Levels of trigonelline, glutamate, dimethylglycine, C18-1-carnitine, C16-1-carnitine, C14-1-carnitine, and 1-methyladenosine were associated with an increased risk, while levels of valine, tryptophan, lysine, leucine, histidine, and 2-aminoisobutyrate were associated with a decreased risk of all-cause mortality. CONCLUSION: We used metabolomics in two Swedish prospective cohorts and identified replicable associations between 13 metabolites and future risk of all-cause mortality. Novel associations between five metabolites-C18-1-carnitine, C16-1-carnitine, C14-1-carnitine, trigonelline, and 2-aminoisobutyrate-and all-cause mortality were discovered. These findings suggest potential new biomarkers for the prediction of mortality and provide insights for understanding the biochemical pathways that lead to mortality.

Adjusting for covariates and assessing modeling fitness in machine learning using MUVR2.

Motivation: Machine learning (ML) methods are frequently used in Omics research to examine associations between molecular data and for example exposures and health conditions. ML is also used for feature selection to facilitate biological interpretation. Our previous MUVR algorithm was shown to generate predictions and variable selections at state-of-the-art performance. However, a general framework for assessing modeling fitness is still lacking. In addition, enabling to adjust for covariates is a highly desired, but largely lacking trait in ML. We aimed to address these issues in the new MUVR2 framework. Results: The MUVR2 algorithm was developed to include the regularized regression framework elastic net in addition to partial least squares and random forest modeling. Compared with other cross-validation strategies, MUVR2 consistently showed state-of-the-art performance, including variable selection, while minimizing overfitting. Testing on simulated and real-world data, we also showed that MUVR2 allows for the adjustment for covariates using elastic net modeling, but not using partial least squares or random forest. Availability and implementation: Algorithms, data, scripts, and a tutorial are open source under GPL-3 license and available in the MUVR2 R package at https://github.com/MetaboComp/MUVR2.