A two-sample Mendelian randomization study identified genetically predicted associations between several urinary metabolites and psychiatric disorders, highlighting potential biomarker candidates that may eventually aid differential diagnosis following further validation.
Researchers analyzed genome-wide association study (GWAS) data for 163 urinary analytes and paired those data with psychiatric disorder GWAS data from the Psychiatric Genomics Consortium. Mendelian randomization uses inherited genetic variants associated with a biological exposure—in this case urinary metabolites—to assess whether those exposures may be linked to disease risk while reducing some confounding present in observational studies.
The analysis included attention-deficit/hyperactivity disorder (ADHD), anorexia nervosa, autism spectrum condition, bipolar disorder, major depressive disorder, schizophrenia, and Tourette syndrome. Most patients represented in the underlying data sets were of European ancestry.
Following multiple-comparison correction and assumption validation, researchers identified 67 metabolite-disorder associations involving 44 unique analytes. Researchers applied Benjamini-Hochberg false discovery rate correction rather than the more conservative Bonferroni approach, reasoning that the tested metabolites and psychiatric disorders likely shared underlying genetic architecture that could make Bonferroni correction overly conservative. No statistically significant associations were identified for autism spectrum condition, major depressive disorder, or Tourette syndrome.
Of the significant associations, 21 were exclusive to a single disorder: 5 with schizophrenia, 15 with bipolar disorder, and 1 with ADHD. The researchers described the inverse association between genetically predicted urinary tyrosine levels and schizophrenia risk as the most robust finding because it was supported across multiple studies and sensitivity analyses. An inverse association between genetically predicted urinary creatine levels and bipolar disorder risk was also supported across multiple studies.
The schizophrenia-specific analytes appeared to cluster into two biologically relevant pathways. Tyrosine and beta-citrylglutamate—a glutamate derivative—were linked to dopaminergic and glutamatergic signaling pathways implicated in schizophrenia pathophysiology, whereas methylsuccinate, ethylmalonate, and 2-methylmalonylcarnitine were associated with mitochondrial metabolic pathways. The researchers noted that methylsuccinate and ethylmalonate are linked to ethylmalonic encephalopathy, a mitochondrial disorder, suggesting a potential connection between mitochondrial dysfunction and schizophrenia risk.
For bipolar disorder, disorder-specific associations involved glycine-related compounds, sulfates, cholates, N-acetyl-aspartyl-glutamate, N-methylpipecolate, and methylsuccinylcarnitine. The researchers noted that creatine has established links to brain energy metabolism and previously reported reductions in several brain regions among patients with bipolar disorder. Prior studies have also associated creatine metabolism with depressive symptom severity in bipolar disorder.
The largest overlap in metabolite associations occurred between bipolar disorder and schizophrenia, with 22 shared analytes demonstrating similar directionality. Many of the overlapping metabolites were N-acetylated compounds.
The researchers suggested this overlap may have clinical relevance because bipolar disorder and schizophrenia are frequently misdiagnosed with each other. The paper cited prior estimates that approximately 24% of patients with schizophrenia may initially receive a diagnosis of bipolar disorder, whereas approximately one-third of patients with bipolar disorder may initially receive a diagnosis of schizophrenia.
Researchers also identified an inverse association between N,N-dimethylglycine and ADHD. Pyridoxal and ferulic acid 4-sulfate were positively associated with anorexia nervosa, although those analytes also overlapped with bipolar disorder and schizophrenia with opposite directionality.
Sensitivity analyses did not identify evidence of horizontal pleiotropy among retained findings. However, many metabolite associations relied on single-nucleotide polymorphism instruments and therefore required Wald ratio estimates, limiting the ability to perform more extensive sensitivity analyses such as Mendelian Randomization Pleiotropy RESidual Sum and Outlier analyses.
The researchers emphasized that the findings should not yet be interpreted as clinically validated diagnostic biomarkers. Urinary analyte levels may vary substantially with hydration, diet, and activity levels. Additional limitations included the inability to perform multivariable Mendelian randomization because most metabolites had only one associated genetic instrument, as well as potential residual confounding because the underlying GWAS data sets were derived from observational cohorts. Limited statistical power for some psychiatric disorder data sets may also have contributed to the absence of statistically significant findings for major depressive disorder, autism spectrum condition, and Tourette syndrome.
The researchers also noted that outcome data sets compared patients with healthy controls rather than with patients with other psychiatric disorders, meaning that some disorder-specific findings may overlap in real-world patient populations. The findings may not generalize beyond predominantly European ancestry populations.
“The results identify candidate urinary biomarkers and demonstrate the utility of genetic instruments for biomarker discovery, warranting experimental validation in independent cohorts,” wrote lead study author Jihan K. Zaki, of the University of Cambridge, and colleagues.
The researchers stated that additional validation steps would be required before clinical application, including targeted metabolomics studies in independent case-control urine cohorts, replication across multiple populations, and assessment of the predictive performance of combined biomarker panels for differentiating bipolar disorder and schizophrenia.
Disclosures: The study was supported by the Stanley Medical Research Institute and the Oskar Huttunen Foundation. Jakub Tomasik reported a patent pending related to bipolar disorder biomarkers. Sabine Bahn reported grants from the Stanley Medical Research Institute and Psyomics during the study; being a founder and shareholder in Psyomics; serving as Director of Psynova Neurotech; and a patent pending related to bipolar disorder biomarkers. Other researchers reported no competing interests.
Source: BMC Psychiatry
