In a recent article published in The Lancet Psychiatry, researchers analyzed the relationships between polygenic liability for bipolar disorder, schizophrenia, and major depressive disorder (MDD) to stratify heterogeneous, multidimensional symptoms of bipolar disorder.
Study: Specificity of polygenic signatures across symptom dimensions in bipolar disorder: an analysis of UK Bipolar Disorder Research Network data. Image Credit: DraganaGordic/Shutterstock.com
Genome-wide association studies (GWAS) suggest that bipolar disorder is highly polygenic, and common alleles shared with schizophrenia or MDD could explain the observed genetic variance of this disorder.
Due to these shared symptom subdomains, many researchers have postulated that their genetic and phenotypic overlaps are correlated in a cause-and-effect pattern, giving rise to the causal heterogeneity hypothesis. Considering this hypothesis is true, these polygenic overlaps could help stratify bipolar disorder.
Another intriguing observation comes from family studies suggesting that there is independent transmission of the core symptom dimensions of bipolar disorder, MDD, and psychosis. Thus, despite their clinical concordance, they seem to be etiologically distinct.
Another unmet need is the lack of discovery of a proper diagnosis for bipolar disorder, a disease with no pathognomonic symptoms and biomarkers. So, currently, psychiatrists diagnose it using a heuristically established list of core criteria.
Psychosis frequently accompanies bipolar disorder but is not part of the diagnostic criteria. An individual has to pass a threshold count of symptoms, which results in substantial heterogeneity in the diagnosis of bipolar disorder.
Overall, heterogeneity in its current definitions and clinical symptoms makes it challenging to understand the pathophysiology of bipolar disorder, obstructing etiological research and hindering drug discovery. Therefore, the stratification of bipolar disorder is a high research priority.
About the study
In the present study, researchers extensively searched multiple databases, viz., MEDLINE, and PsychINFO, to name a few, and systematically identified all relevant literature published in English between January 1, 2006, and September 15, 2022, that used contemporary approaches for genomic stratification of mood disorders.
Several papers suggested clinical heterogeneity in bipolar disorder but presented weak evidence of its biologically validated symptoms. Many studies found that clinical heterogeneity of bipolar disorder indexed differential underlying genetics.
Nevertheless, since symptoms of bipolar disorder and MDD/schizophrenia are highly correlated, this could impede its clinically relevant stratification.
The authors used the criteria of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition—DSM-IV, to identify people with bipolar disorder from the UK Bipolar Disorder Research Network (BDRN) database.
Inclusion criteria for enrolment into this study were that individuals were 18 years or older at the time of the interview, were of European ancestry (EA), lived in the United Kingdom (UK), and registered with the BDRN.
They assessed the data using the Operational Checklist for Psychotic Disorders (OPCRIT) completed by trained research psychologists or psychiatrists. Then, the researchers performed supervision and inter-rater reliability assessments to ensure consistency and accuracy in OPCRIT ratings.
Schedules for Clinical Assessment in Neuropsychiatry (SCAN) interviews and psychiatric case notes served as sources of information, and their corroboration helped endorse each OPCRIT item. The team collected sex data via genetic testing, and research psychiatrists also collected blood samples during the SCAN interview for genotyping patient data.
They used a high dimensional Bayesian regression framework called PRS-CS software that adjusted single nucleotide polymorphisms (SNPs) effect sizes via a continuous shrinkage
(tuning) informed by multivariate modeling of the underlying genetic structure.
In other words, the study model helped stratify the bipolar disorder phenotype into symptom subdomains in a genetically informed sample.
The multiple indicator multiple causes (MIMIC) model allowed researchers to explore the patterns of complex associations between polygenic liability indexed by disease symptoms, their dimensionality, and polygenic risk scores (PRS).
The team examined 59 OPCRIT items related to bipolar disorder psychopathology during the exploratory factor analysis (EFA), and their calibration subsample comprising 2648 of 4148 participants helped them explore relationships among OPCRIT items and identify strongly correlated symptoms, referred to as common factors.
Though OPCRIT includes a broad range of psychopathology, in this analysis, the team used its five most reliable indicators that provided good coverage for the identification of symptom dimensions for depression, mania, and psychosis dimensions.
Likewise, the validation dataset of the study helped the researchers test the reproducibility of the identified common factors structure during the confirmatory factor analysis (CFA).
In total, 4,198 individuals were eligible for inclusion in this study; however, only 4,148 individuals of EA living in the UK with a median age of 45 years at the SCAN interviews constituted the analysis sample set, of which 2,804 and 1,344 were female and male, respectively.
The authors found correlations between the PRSs for bipolar disorder, schizophrenia, and MDD in the study sample, of which the PRS for schizophrenia was the only statistically significant predictor of the psychosis factor.
Likewise, the PRS for MDD and bipolar disorder were only statistically significant predictors of depressive and maniac factors, respectively.
The study findings were consistent with the clinical heterogeneity hypothesis and suggested that the shared symptom dimensions of bipolar disorder, at least in part, had distinct causal components. Also, each was genetically unique and had a polygenic liability signature.
The study results challenged current diagnostic systems and highlighted that dimensional representations of bipolar disorder could accommodate problems related to poorly explained mixed psychiatric states with the same (or similar) symptom presentation.
It could inform new approaches to bipolar disorder stratification, a necessary step towards enhancing the understanding of causal mechanisms of many psychiatric disorders.
MIMIC, a unitary analytical approach used in this study, confirmed that while the identified symptom dimensions could index multiple domains spanning psychiatric disorders pathophysiologies; however, each had varying prominence in different diagnostic categories.
Thus, these stratifiers for bipolar disorder provided genetically validated phenotypic biomarkers to explore the mechanisms governing this disorder.
MIMIC simultaneously and consistently estimated relationships between symptoms, common factors, and PRSs. Importantly, it reduced potential bias due to measurement error and multiple testing.
In the future, as large GWAS on bipolar disorder become available, PRS prediction power will also increase, which would facilitate the construction of more homogeneous subsamples, such as bipolar disorder type 1 (higher specificity), which, in turn, would help in the discovery of more specific causal components in psychiatry.
Thus, precision psychiatry might use symptom dimensions for psychiatric diagnosis in the future, irrespective of the primary diagnosis.
Allardyce, J., Cardno, A.G., Gordon-Smith, K., Jones, L., Di Florio, A., Walters, J.T.R., Holmans, P.A., Craddock, N.J., Jones, I., Owen, M.J., Escott-Price, V. & O'Donovan, M.C. (2023) Specificity of polygenic signatures across symptom dimensions in bipolar disorder: an analysis of UK Bipolar Disorder Research Network data. The Lancet Psychiatry. doi: 10.1016/S2215-0366(23)00186-4. https://www.thelancet.com/journals/lanpsy/article/PIIS2215-0366(23)00186-4/fulltext