Metabolomic profiles predict individual multidisease outcomes

Abstract

Risk stratification is critical for the early identification of high-risk individuals and disease prevention. Here we explored the potential of nuclear magnetic resonance (NMR) spectroscopy-derived metabolomic profiles to inform on multidisease risk beyond conventional clinical predictors for the onset of 24 common conditions, including metabolic, vascular, respiratory, musculoskeletal and neurological diseases and cancers. Specifically, we trained a neural network to learn disease-specific metabolomic states from 168 circulating metabolic markers measured in 117,981 participants with ~1.4 million person-years of follow-up from the UK Biobank and validated the model in four independent cohorts. We found metabolomic states to be associated with incident event rates in all the investigated conditions, except breast cancer. For 10-year outcome prediction for 15 endpoints, with and without established metabolic contribution, a combination of age and sex and the metabolomic state equaled or outperformed established predictors. Moreover, metabolomic state added predictive information over comprehensive clinical variables for eight common diseases, including type 2 diabetes, dementia and heart failure. Decision curve analyses showed that predictive improvements translated into clinical utility for a wide range of potential decision thresholds. Taken together, our study demonstrates both the potential and limitations of NMR-derived metabolomic profiles as a multidisease assay to inform on the risk of many common diseases simultaneously.

Robust and High-Throughput Analytical Flow Proteomics Analysis of Cynomolgus Monkey and Human Matrices with Zeno SWATH Data Independent Acquisition

“…With omics technologies becoming more mature, they are being increasingly used in epidemiological studies generating high content quality datasets for disease risk prediction and prevention as well as disease mechanism investigation. Focusing on circulating blood markers, metabolomics and proteomics are the two most popular approaches and successfully generated predictive models for a wide range of disorders including metabolic, cardiovascular and neurological diseases [35][36][37][38]. Even though proteomics has been successful in capturing important aspects of human disease pathophysiology, adding information from complementary omics layers has improved sensitivity and specificity of predictive models in studies of complex diseases (i.e.…”

A metabolomic-based biomarker discovery study for predicting phototherapy duration for neonatal hyperbilirubinemia

“…Both biological and environmental factors infect metabolites, and metabolome is thought to be most predictive of phenotype ( 37 ). It has been used in the prediction of most diseases and is superior to conventional clinical predictors ( 38 ). Localizing metabolic perturbations in patients is also crucial to diagnosing and addressing diseases ( 39 ), but species determination is still challenging in metabolomics, the structure and function of numbers of unknown metabolites are waiting for an investigation.…”

Lipoprotein profile and metabolic fine-mapping of genetic lipid risk loci

“…Phenome-wide association data. Buergel et al trained a neural network to learn diseasespecific metabolomic states from the 168 UKB NMR traits to predict individual multi-disease outcomes 68 . The authors provide the ranked variable importance for the NMR traits to predict 24 conditions, including common metabolic, vascular, respiratory, musculoskeletal and neurological disorders and cancers (Supplementary Table 12 in Buergel et al).…”