IntroductionThe identification of peripheral biomarkers holds significant promise for uncovering the mechanisms of complex diseases, improving diagnostic accuracy, and enabling prognosis and personalized treatment. One way to identify biomarkers is through genome-wide association studies (GWASs), which have transformed our knowledge of how genetic differences, typically in the form of single nucleotide polymorphisms (SNP), affect phenotypic outcomes1. While most initial GWAS studies focused on identifying genetic variants associated with specific binary traits or disorders1, there has been a subsequent increase in focus towards more quantitative outcomes, such as biological analyte concentrations. The first metabolite GWAS study, conducted in 20082, demonstrated the potential to establish robust causal links between genetic risk factors and target metabolites, and emphasized the importance of identifying relevant molecular intermediaries for a more complete understanding of the mechanistic pathways of disease. This argument has become stronger over the past decade, as the assessment of causal inference of metabolite levels on various diseases has become more accessible due to advancements in Mendelian randomization (MR) methodology and analyte measurement methods.While numerous metabolite GWAS reports have been published since the first study, these have focused primarily on serum and plasma metabolites as opposed to other sample types, such as urine. This trend is exemplified by the fact that a PubMed search for serum and plasma GWAS studies currently returns nearly twenty times more results than for urine-based studies. Nonetheless, since the first urinary metabolite GWAS study was published in 20113, research in this area has been steadily expanding, even though the overall sample sizes have remained relatively similar. In the face of limited large-scale analyses, a potential way to bolster the power of urinary metabolite GWAS studies is through the implementation of meta-analytic methods4. The combination of results from multiple studies could help uncover novel associations and strengthen confidence in previously reported findings.In the present study, we aimed to conduct a comprehensive meta-analysis of existing urinary metabolite GWAS studies to combine all known metabolite-SNP associations into a single resource, assess the consistency of associations between studies, and identify novel genetic associations to urinary metabolite levels. A comprehensive database search was conducted to identify published and proprietary urinary metabolite GWAS data, followed by a sample size-based meta-analysis to identify significant genetic associations.ResultsThe current study consisted of a comprehensive literature review of urinary metabolite GWAS studies carried out to date, followed by extensive data collection and subsequent meta-analysis. Of the identified 191 unique studies of interest, 26 satisfied the inclusion criteria concerning the required urinary metabolite GWAS data and its quality. Selected manuscripts are listed in Supplementary Table S1. Following data enquiries where GWAS summary statistics had not been made publicly available, five studies were included in the final, sample size-based meta-analysis. A summary of the data collection process alongside the list of final included studies is shown in Fig. 1.Fig. 1A summary of the data collection process for the urinary metabolite GWAS meta-analysis.Full size imageDescription of the included studiesA total of five studies were included in the meta-analysis. The studies are summarized in Supplementary Table S2. The first study, by Nicholson et al.5, aimed to identify genetic associations for three urinary analytes in two cohorts, MolTWIN and MolOBB, comprising 211 participants. Urinary metabolites were measured using 1H NMR and genotyping was performed using the Illumina 317 K BeadChip SNP array. In the second study, Raffler et al.6 measured 55 urinary markers using targeted NMR measurements in 3861 participants from the Study of Health in Pomerania (SHIP) and 1691 participants from the Kooperative Gesundheitsforschung in der Region Augsburg (KORA) study. Both analyses used Affymetrix Human SNP Array 6.0 gene chips for genotyping. Sinnott-Armstrong et al.7 conducted a GWAS study of 4 urine analytes measured using clinical tests in 363 228 UK Biobank participants. Genotyping was conducted using the UK Biobank Axiom Array. Calvo-Serra et al.8 performed a GWAS study aiming to identify metabolite quantitative trait loci for 44 urinary metabolites measured using proton NMR. The study included 996 children recruited as a part of the Human Early Life Exposome (HELIX) project who were genotyped using the Infinium Global Screening Array (GSA) MD version 1 (Illumina). Finally, Schlosser et al.9 evaluated genetic associations of 1172 metabolites in 1627 participants from the German Chronic Kidney Disease (GCKD) study and the Study of Health in Pomerania Trend (SHIP-Trend). Metabolites were measured using mass spectrometry, and genotyping was conducted using Illumina Omni2.5Exome BeadChip arrays in the GCKD cohort, and Illumina HumanOmni2.5-Quad in the SHIP-Trend cohort.GWAS data processingThe extracted SNP-analyte association data included 1297 analyte associations to 1297 unique SNPs from the study by Nicholson et al.; 1 799 971 analyte associations to 608 024 unique SNPs from the study by Raffler et al.; 3 113 996 analyte associations with 780 302 unique SNPs from the study by Sinnott-Armstrong et al.; and 12 482 976 analyte associations with 283 704 unique SNPs from the study by Calvo-Serra et al.. Finally, due to the large number of associations in the study by Schlosser et al., only SNP-analyte associations matching the other studies were extracted, which amounted to 1 213 222 unique SNPs and 7 034 023 SNP-analyte associations. Following the removal of associations evaluated only by single studies, 14 179 833 SNP-analyte associations remained in the dataset, corresponding to 1 213 704 unique SNPs and 48 urinary metabolites.Meta-analysis of urinary metabolite GWAS studiesThe sample size-based meta-analysis revealed 1226 SNPs with significant (P