Most studies are small and based on patient datasets (e.g. COPD) | Large-scale, population-based studies (including replication) needed |
Limited number of studies on intermediate biomarkers linking exposure and disease | Implementation of the meet-in-the-middle (MITM) and/or other causal mediation approaches |
Heterogeneities in metabolomics assessment | Uniform analytical platform and metabolite annotation protocols |
Cross-sectional data | Exposure and metabolomic profiles assessed before disease occurrence (e.g. in longitudinal studies) |
Difficult to evaluate latency and long-term effects | Longitudinal studies with long-term follow-up |
Lack of confounding control | A complete exposome concept taking all relevant exposures into account |
Susceptibility factors ignored | Inclusion of genetics (e.g. polygenic risk scores) and other host factors in the analyses |
Lack of integration with other omics data | Apply multi-omics models (including e.g. epigenetics, transcriptomics, proteomics) |