Association of Etiological Factors for Hypomanic Symptoms, Bipolar Disorder, and Other Severe Mental Illnesses

This cohort study examines the etiology of hypomanic symptoms in young adults and its association with that of bipolar disorder and other serious mental illnesses, combined with polygenic risk score analyses.

Key Points

Question

What is the specific and shared genetic and environmental architecture of subsyndromal hypomania and bipolar disorder (BD), schizophrenia, and major depressive disorder in young adults?

Findings

In this cohort study of 8568 twin pairs, higher heritability estimates for hypomania were found for male individuals compared with female individuals. Moderate genetic and nonshared environmental correlations between hypomania and BD were detected, and hypomania was significantly associated with the polygenic risk scores for schizophrenia and major depressive disorder but not for BD.

Meaning

The etiology of subsyndromal hypomania overlaps with BD, major depressive disorder, and schizophrenia, indicating that it may be a continuous trait for psychiatric disorders reflected at its extreme.

Abstract

Importance

Subsyndromal hypomanic symptoms are relatively common in the general population and are linked to the onset of bipolar disorder. Little is known about their etiology and whether this is shared with the etiology of bipolar disorder or other mental illnesses.

Objective

To examine the genetic and environmental architecture of hypomanic symptoms in a nonclinical youth sample and compare estimates at varying severity levels and their association with diagnosed bipolar disorder.

Design, Setting, and Participants

This cohort study used phenotypic and genetic data from the Child and Adolescent Twin Study in Sweden and included individuals with International Statistical Classification of Diseases and Related Health Problems, Tenth Revision diagnosis of psychiatric disorders from national registries for residents of Sweden. Associations between hypomania and polygenic risk scores for bipolar disorder, major depressive disorder and schizophrenia were also investigated. Analysis began November 2018 and ended October 2021.

Main Outcomes and Measures

Hypomanic symptoms were assessed using the parent-rated Mood Disorders Questionnaire when the twins were aged 18 years. Bipolar disorder diagnosis and/or lithium prescription were ascertained from national registries for residents of Sweden. Polygenic risk scores for psychiatric disorders were calculated using independent discovery genetic data.

Results

A total of 8568 twin pairs aged 18 years (9381 [54.7%] female) were included in the study. The hypomania heritability estimate was 59% (95% CI, 52%-64%) for male individuals and 29% (95% CI, 16%-44%) for female individuals. Unique environmental factors accounted for 41% (95% CI, 36%-47%) of the hypomania variance in male individuals and 45% (95% CI, 40%-50%) in female individuals. Shared environmental factors were only detected for female individuals and explained 26% (95% CI, 13%-38%) of the variance. The heritability estimates were fairly consistent across different hypomania severity groups. Moderate genetic (0.40; 95% CI, 0.21-0.58) and shared environmental (0.41; 95% CI, 0.03-0.75) correlations between hypomania and diagnosed bipolar disorder were found. Hypomania was significantly associated with the polygenic risk scores for schizophrenia (β = 0.08; SE = 0.026; P = .002) and major depressive disorder (β = 0.09; SE = 0.027; P = .001) but not bipolar disorder (β = 0.017; SE = 0.03; P = 0.57) (bipolar disorder I [β = 0.014; SE = 0.029; P = .64] or bipolar disorder II [β = 0.045; SE = 0.027; P = .10]).

Conclusions and Relevance

Higher heritability for hypomania was found for male compared with female individuals. The results highlight the shared etiologies between hypomanic symptoms, bipolar disorder, major depression, and schizophrenia in youths. Future research should focus on identifying specific shared genetic and environmental factors. These findings support a possible dimensional model of bipolar disorder, with hypomania representing a continuous trait underlying the disorder.

Introduction

Understanding the early manifestations of bipolar disorder (BD) are critical for the development of effective prevention and intervention. One approach to tackle this issue is to study subsyndromal hypomania (symptoms that do not meet diagnostic criteria for hypomanic/manic episodes) in community samples during adolescence and early adulthood. This approach holds great promise because hypomanic and manic episodes are BD’s defining feature.¹ Moreover, subsyndromal hypomanic symptoms are linked to manic and hypomanic episodes and BD onset.² Focusing on this developmental period is particularly informative for understanding BD’s cause because its onset typically occurs between age 15 and 24 years.³ Questions concerning the relationship between BD and hypomanic symptoms in nonclinical populations remain,⁴ such as do hypomania and BD lie on the same phenotypic continuum, with the disorder representing the extreme end? Understanding the etiological relationship between these phenotypes will help address this and related questions.

Heritability estimates within a similar range have been reported for BD (57%-85%)^5,6 and pediatric BD phenotypes (75%),⁷ but no studies on subsyndromal hypomania in youths exist, to our knowledge. Heritability estimates vary by BD type and severity, with a higher estimate of 57% for BD I (diagnosed with a lifetime manic episode) compared with 46% for BD II⁸ (diagnosed with both hypomanic and major depressive episodes¹), which may be more closely related to subsyndromal hypomania. There is evidence that overlapping and distinct genetic factors influence different BD subtypes; thus, there is value in examining them separately.⁸ However, these results do not reveal whether there is any genetic overlap between subsyndromal hypomania and BD.

Preliminary evidence indicates that common genetic variants for BD captured using polygenic risk scores (PRS) are not significantly associated with hypomanic symptoms,^9,10,11 although weak associations have been detected with more severe hypomania symptom profiles.⁹ These initial findings need to be replicated. Given the phenotypic and genetic overlap between BD and other psychiatric disorders, including major depressive disorder (MDD) and schizophrenia,¹² it would be useful to consider whether these cross-phenotypic relationships are mirrored for hypomanic symptoms. Significant associations between BD and the schizophrenia¹³ and MDD¹⁴ (particularly for BD II) PRS have been reported previously. However, there are no studies examining the association between hypomania in nonclinical youth samples and schizophrenia and MDD PRS warranting further investigation.

PRS only capture additive effects of common genetic variants, which account for 15.6% to 18.6% of BD’s genetic variance on the liability scale.¹⁵ The heritability parameter estimated using twin data covers the entire genetic variance (common and rare genetic factors) and can be used to examine genetic covariance between 2 or more phenotypes.¹⁶

To our knowledge, this is the first twin study to examine the etiology of hypomanic symptoms in young adults and its association with that of BD and other serious mental illnesses, combined with PRS analyses. Specifically, this study will examine the genetic and environmental architecture of hypomanic symptoms at age 18 years, focusing on a continuous hypomania measure and subgroups based on extreme symptom levels (top 10%, 5%, and 1%). Second, the extent to which the genetic and environmental risk factors for BD are associated with hypomania will be investigated. Finally, the polygenic overlap between hypomania and BD, schizophrenia, and MDD will be explored.

Methods

Participants

Each year since 2004, families of Swedish twins aged 9 or 12 years are invited to participate in the Child and Adolescent Twin Study in Sweden (CATSS)¹⁷ and are followed up at age 15 and 18 years.¹⁷ A total of 8568 twin pairs participated in CATSS at age 18 years with available parent-rated hypomania data (eFigure in the Supplement). A higher proportion of female individuals with fewer neurodevelopmental/psychiatric diagnoses and parents with higher educational qualifications completed follow-up at age 18 years compared with those who declined to participate. Parents and twins provided consent prior to participation. CATSS was approved by the Regional Ethical Review Board in Stockholm (2016/2135-31).

Measures

Hypomanic symptoms were assessed by the parent-rated Mood Disorder Questionnaire (MDQ)¹⁸ when the twins were age 18 years. The MDQ’s 13 yes/no items relate to the presence of symptoms based on DSM-IV criteria for a hypomanic or manic episode.^18,19 Additional items inquire whether symptoms occur during the same period (episode) and affect functioning (impairment). The parent-rated MDQ has good sensitivity (0.72) and specificity (0.81) in identifying adolescent BD.^18,20 Individuals were categorized as being at high risk of BD if at least 7 MDQ items were endorsed, clustered in the same period, and caused moderate/severe impairment.²¹

BD cases were identified using 2 approaches. First, via BD diagnosis (International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes F30-F31) up to age 24 years, identified using the Swedish National Patient Register,²² which documents all specialist inpatient and outpatient care delivered to residents of Sweden. Second, through lithium prescriptions using the Prescribed Drug Register,²³ which records all medication prescriptions for residents of Sweden since 2005, the Swedish National Patient Register was also used to identify participants with an MDD (F32-F34) or schizophrenia (F20) diagnosis.

Genetic Information

DNA was extracted from saliva samples provided at study enrollment and genotyped using PsychChip (Illumina). Standard quality control and imputation procedures were applied (details provided elsewhere²⁴). A total of 11 081 samples passed quality control assessment; 2495 monozygotic (MZ) twins were then imputed based on their genotyped co-twin. A total of 13 456 participants were included in genetic analyses, following imputation, quality control, and exclusions (eFigure in the Supplement).

PRS for 5 disorders were calculated for each participant. PRS were derived from publicly available genome-wise association study summary statistics for BD, BD I, BD II,¹⁵ MDD,²⁵ and schizophrenia,²⁶ independent of CATSS. The PRS–continuous shrinkage approach²⁷ was used for PRS scoring (eAppendix in the Supplement). PRS were calculated using PLINK by scoring the number of the risk alleles for the respective illness weighted by their effect size for each individual. Scores were standardized using z score transformations. Principal component analysis was used to derive covariates to account for population stratification.

Statistical Analyses

The MDQ was positively skewed and therefore log-transformed. Participants were split into high-risk and low-risk groups based on MDQ cutoff described above.²¹ Logistic regression models within a generalized estimating equations (GEE) framework were undertaken to compare the rates of BD diagnosis among those in the high-risk group with the remaining sample, controlling for sex and birth year. Differences in hypomania score between those with and without a BD diagnosis was tested using a linear regression model with GEE framework controlling for sex and birth year. GEEs were used to account for the use of twins, to calculate robust standard errors, and were implemented in the drgee package of R version 3.5.3 (R Foundation).²⁸ A 1-sided P < .05 was considered to test statistical significance. Analysis began November 2018 and ended October 2021.

The classic twin method was used to examine the degree genetic and environmental factors influence individual phenotypes (hypomania and BD diagnosis) as well as the degree to which they are shared between them. Phenotypic variance can be partitioned into additive genetic (A), shared (C; common to both twins), and unique (E; differ across twins and measurement error) environmental influences.¹⁶ This is based on the assumption that identical or MZ twins share 100% of their segregating DNA code compared with nonidentical or dizygotic (DZ) twins, who share approximately half. Higher twin pair correlations among MZ compared with DZ twins suggests genetic influences on a trait. The general principles of the twin design are described in detail elsewhere.¹⁶

DeFries-Fulker extreme analyses was used to contrast the degree to which the genetic and environmental factors influence differing hypomania severity by focusing on the cause of the mean difference of extreme scores and the whole population.²⁹ A significant group heritability estimate suggests that there is a genetic link between the extreme groups and full sample. A joint categorical-continuous bivariate model was used to estimate the genetic correlation between BD and hypomania, and the degree to which overlapping genetic and environmental influences explained their association.¹⁰

To test associations between hypomania (symptoms and high-risk group) and each of the PRS, GEEs were performed with robust SEs based on clustering related individuals; 10 principal components were used as covariates in these analyses (eAppendix in the Supplement).

Results

Parent-rated hypomanic symptoms using the MDQ were available for 8568 twin pairs (9381 [54.7%] female; Table 1). All participants were aged 18 years at the time of hypomania assessment. Female individuals reported significantly more hypomanic symptoms compared with male individuals (β = 0.11; SE = 0.04; P = .01). A total of 64 participants (0.8%) were categorized as high risk for BD using published MDQ cutoffs,²¹ and 54 (0.3%) received a BD diagnosis or lithium prescription. Individuals with BD had a significantly higher hypomania score compared with the rest of the sample (β = 3.01; SE = 0.73; P < .001). A significantly greater proportion of the high-risk group (9 [14%]) had a BD diagnosis compared with the remaining sample (45 [0.27%]) (odds ratio, 1.32; 95% CI, 1.13-1.55; P < .001).

Table 1. Sample Description.

Characteristic	No. (%)
	Overall	Male	Female
No. of individuals	17 136	7755 (45.3)	9381 (54.7)
Twins
MZ	5162 (30.1)	2201 (28.4)	2961 (31.6)
DZ	5917 (34.5)	2705 (34.9)	3212 (34.2)
DZ opposite sex	6057 (35.3)	2849 (36.7)	3208 (34.2)
Maternal educational level
Compulsory education	639 (4.4)	283 (4.2)	356 (4.5)
Upper secondary	6382 (43.8)	2915 (43.7)	3467 (43.9)
College or university	7142 (49.0)	3285 (49.3)	3857 (48.8)
Paternal educational level
Compulsory education	1355 (10.2)	613 (10.0)	742 (10.4)
Upper secondary	6586 (49.7)	3018 (49.3)	3568 (50.1)
College or university	4925 (37.2)	2320 (37.9)	2605 (36.6)
MDQ
Mean (SD)	0.79 (1.75)	0.73 (1.71)	0.84 (1.79)
High risk	64 (0.8)	25 (0.7)	39 (0.9)
Severity groups, MDQ
10%	838 (11.0)	358 (10.2)	480 (11.7)
5%	363 (4.8)	152 (4.3)	211 (5.1)
1%	104 (1.4)	45 (1.3)	59 (1.4)
Bipolar disorder and/or lithium prescription	54 (0.3)	15 (0.2)	39 (0.4)
Bipolar disorder diagnosis	47 (0.3)	13 (0.2)	34 (0.4)
Lithium prescription	26 (0.2)	9 (0.1)	17 (0.2)

Abbreviations: DZ, dizygotic; MDQ, Mood Disorder Questionnaire; MZ, monozygotic.

Twin Analyses

Table 2 presents twin correlations for hypomanic symptoms and BD diagnosis. Because there were sex differences for hypomania twin correlations, the parameters were estimated separately for male and female individuals in subsequent analyses. MZ correlations were higher than DZ correlations, suggesting genetic contributions to hypomania and BD. MZ correlations were less than 1, indicating nonshared environmental influences.

Table 2. Phenotypic, Twin, and Cross-Trait Cross-Twin Correlations for Hypomania and Bipolar Disorder.

Variable	Correlation coefficient (95% CI)
	r ph	MZ	DZ	MZF	DZF	MZM	DZM	DZOS
Hypomania at age 18 y				0.55 (0.50 to 0.60)	0.41 (0.35 to 0.48)	0.61 (0.56 to 0.66)	0.19 (0.12 to 0.25)	0.25 (0.20 to 0.30)
Bipolar disorder		0.88 (0.71 to 0.96)	0.35 (−0.02 to 0.62)
Cross-trait	0.39 (0.30 to 0.47)	0.31 (0.17 to 0.43)	0.07 (−0.05 to 0.19)

Abbreviations: DZ, dizygotic twin pairs; DZF, dizygotic female twin pairs; DZM, dizygotic male twin pairs; DZOS, dizygotic opposite sex twin pairs; MZ, monozygotic twin pairs; MZF, monozygotic female twin pairs; MZM, monozygotic male twin pairs; r_ph, phenotypic correlation.

eTables 1 and 2 in the Supplement present the fit statistics for the univariate twin models. ACE models were chosen as best fitting, with sex differences observed for hypomania. BD was highly heritable with nonshared environmental factors explaining the remaining variance (Figure). Genetic influences explained most of the hypomania variance in male individuals (59% [95% CI, 52%-64%]), with nonshared environmental factors accounting for the rest (42% [95% CI, 36%-47%]). In contrast, for female individuals, nonshared environmental factors accounted for the majority of the hypomania variance (45% [95% CI, 40%-50%]) with genetic and shared environmental factors explaining 29% (95% CI, 16%-44%) and 26% (95% CI, 13%-38%), respectively.

Figure. Genetic and Environmental Univariate Estimates and Bivariate Correlations for Hypomanic Symptoms at Age 18 Years and Bipolar Disorder (Diagnosis and Lithium Prescription) at Age 24 Years.

Genetic and environmental influences on hypomania at age 18 for male (A) and female (B) individuals. C, Genetic and environmental influences for bipolar disorder and/or lithium prescription up to age 24 years. D, Joint-continuous model estimates for hypomania at 18 years and bipolar disorder/lithium prescription up to age 24 years. A indicates additive genetic influences; C, shared environmental factors; E, nonshared environmental influences; rA, genetic correlation; rE, nonshared environmental correlation.

Hypomania and BD Etiological Overlap

There was a moderate phenotypic correlation between hypomania and BD (r = 0.38; 95% CI, 0.29-0.47). The limited number of patients with BD (n = 54) meant joint models with hypomania were not able to consider male and female individuals separately. Cross-trait cross-twin correlations are shown in Table 2 and are higher for MZ compared with DZ twins, suggesting a genetic contribution to the covariance between hypomania and BD. The Figure shows the moderate genetic (0.40; 95% CI, 0.21-0.58) and nonshared environmental (0.41; 95% CI, 0.03-0.75) correlation between hypomania and BD using an AE model. The hypomania-BD phenotypic correlation was found to be mainly explained by genetic factors (72%; 95% CI, 41%-98%) with a smaller contribution from nonshared environmental factors (28%; 95% CI, 2%-59%). Results were similar when an ACE model was applied (eTable 3 in the Supplement).

The hypomania-MDD correlation was small (r = 0.12; 95% CI, 0.04-0.20) and only 30 individuals with schizophrenia were identified, which prohibited the exploration of the associations between these disorders and hypomania using twin methods.

Hypomania Extreme Analyses

AE models showed the best fit for the DeFries-Fulker analyses. Significant group heritability was found in the DeFries-Fulker analyses, which suggests a genetic link between severe levels of hypomania and variation in hypomanic symptoms in the whole sample (Table 3). Similar heritability estimates were found across each of the hypomania severity groups, although slightly lower in the top 1% group.

Table 3. Extremes Analyses for Hypomania Using DeFries-Fulker Analysis.

Model and parameters	Severity groups
	>10%	>5%	>1%	High riska
Ab	0.53 (0.47-0.60)	0.47 (0.40-0.55)	0.31 (0.19-0.44)	0.37 (0.25-0.48)
E (residual)c	0.47 (0.40-0.53)	0.53 (0.45-0.60)	0.69 (0.56-0.81)	0.63 (0.52-0.75)

^a Mood Disorders Questionnaire score of at least 7 with symptoms clustered together in the same period and moderate to severe problems (eg, work and legal problems).

^b Additive genetic environmental influences.

^c Additive unique environmental influences.

PRS Analyses

The schizophrenia and MDD PRS were significantly associated with hypomanic symptoms but not when the high-risk group was considered (Table 4). No significant associations were detected between any of the BD PRS and hypomania.

Table 4. Association Between Hypomanic Symptoms and Polygenic Risk Scores for Severe Mental Illnesses.

PRS	Hypomanic symptoms			High risk for bipolar disorder
	β (SE)	P value	Effect size	Control mean (SD)	Risk mean (SD)	OR (95% CI)	P value
Bipolar disorder	0.017 (0.03)	.57	−1.10 ×10−3	−0.01 (0.99)	−0.01 (0.89)	0.99 (0.78-1.26)	.92
Bipolar disorder I	−0.014 (0.029)	.64	−1.13 ×10−3	−0.03 (0.98)	0.02 (0.88)	1.04 (0.82-1.32)	.73
Bipolar disorder II	0.045 (0.027)	.10	−5.76 ×10−4	−0.01 (1.00)	0.01 (0.89)	1.00 (0.78-1.30)	.97
Schizophrenia	0.08 (0.026)	.002	7.76 ×10−4	−0.01 (0.99)	0.15 (0.98)	1.19 (0.92-1.53)	.19
Major depressive disorder	0.09 (0.027)	.001	1.17 ×10−3	−0.03 (1.01)	0.14 (0.90)	1.19 (0.89-1.59)	.24

Abbreviation: OR, odds ratio; PRS, polygenic risk scores.

Discussion

To our knowledge, this is the first twin study to examine the specific and shared etiology of hypomanic symptoms, BD, MDD, and schizophrenia in a nonclinical youth sample, combined with a polygenic approach. We found higher heritability for hypomania among male individuals (59%) compared with female individuals (29%). Common environmental factors were only found to influence hypomania for female individuals (26%), but unique environmental influences accounted for a similar degree of the variance for male (41%) and female individuals (45%). Significant group heritability was detected when groups with more severe levels of hypomania were examined; this suggests that similar genetic factors influence low and more severe levels of hypomania. Moderate genetic and nonshared environmental correlations between hypomania and BD were found. The small correlation between hypomania and MDD and the restricted number of schizophrenia diagnoses prohibited using twin analyses to assess the shared etiology between these phenotypes. The PRS for schizophrenia and MDD but not BD were significantly associated with subsyndromal hypomania.

This investigation provides a novel contribution by examining the etiology of subsyndromal hypomania in a community sample of young adults. The heritability estimate we found of 59% for male individuals parallels the results for BD in adults (range, 59%³⁰ to 85%⁶) but is higher than what is reported for adult BD II (46%).⁸ Our results for female individuals were more curious where environmental influences accounted for the majority of the hypomania variance and genetic factors explained 29%. Various environmental factors have been implicated in the cause of BD and hypomania, including childhood maltreatment.^31,32,33 Research shows that female individuals are at greater risk of experiencing such trauma compared with male individuals,³⁴ which could explain the diminished role of genetic factors at this developmental stage. Future studies are needed to replicate our findings and assess any developmental changes in the hypomania etiology.

Phenotypic continuums between psychiatric disorders (eg, MDD) and their respective subthreshold symptoms (eg, depressive symptoms) that are more frequently observed in the general population are used to understand the early manifestations, developmental trajectories, and etiology of such disorders.^35,36 Such research is critical for the development of effective prevention and intervention.³⁷ Research in this space concerning BD is emerging and gaining momentum.

Our findings partly address this issue showing a moderate genetic (0.40) and nonshared environmental (0.41) correlation between subsyndromal hypomania and BD for the first time. The genetic correlation is lower but not that dissimilar to those reported for depressive symptoms and MDD in the same sample (0.53-0.58).¹⁰ Our work adds to the validity of a hypomania continuum of BD when combined with research showing links between hypomania and bipolar disorder–related risk factors (eg, childhood maltreatment).³³

Caution should be taken when applying the hypomania continuum to BD given that the disorder is not characterized by 1 symptom dimension, unlike MDD. For instance, patients with BD commonly experience psychosis³⁸ and depressive episodes, with the latter included in the BD II diagnostic criteria.¹ Thus, a BD quantitative trait model may be best explained as the intersection of extremes of multiple symptom dimensions. Our results suggest that hypomania would be one such symptom dimension. Depressive symptoms and psychoticlike experiences may represent other dimensions to consider but need to be empirically tested.

The source of the genetic overlap between BD and hypomanic symptoms is unclear given that we and others have not found an association between hypomania and the BD PRS.^9,10,11 Several reasons may explain this result. First, the hypomania-BD genetic overlap may not be due to common genetic variation captured in PRS but rare genetic factors included in the twin heritability parameter³⁹ used here. Second, the BD PRS is calculated using adults with BD who experience more severe symptomatology,¹⁵ and we focused on subsyndromal hypomania in youths. Finally, the BD PRS explains 15.6% to 18.6% of the BD genetic variance on the liability scale.¹⁵ Hence, the lack of association between the BD PRS and hypomania provides a limited picture of their genetic overlap.

Previous studies have reported a significant association between schizophrenia¹³ and MDD¹⁴ PRS and BD diagnosis as well as hypomanic/manic episodes (schizophrenia PRS only). We extend these findings by showing a link with hypomania at the subsyndromal level. The divergent associations between the BD PRS (15.6%-18.6%)¹⁵ and the schizophrenia (24%)²⁶ and MDD (8.7%)²⁵ PRS with hypomania do not seem to be attributable to the degree to which they explain the genetic variance for their respective disorders on the liability scale.

Our findings have important research and clinical implications. First, our work underscores the importance of studying subsyndromal hypomania in its own right given its association with the diagnosis and etiology of several psychiatric disorders. Other studies have found that hypomania in youths is associated with adverse outcomes, including suicidality and other psychopathological symptoms.^40,41,42 Hypomania prevention and intervention strategies need to be developed to avoid progression to clinical disorders. Second, the hypomania-BD association reported here highlights the potential for using hypomania to identify high-risk BD groups that may benefit most from prevention and intervention efforts.

Strengths and Limitations

The strengths of this investigation include use of a large, longitudinal, genetically informative sample, national registries, and a multimethod approach, but there are some limitations. First, a limited number of participants received a BD and schizophrenia diagnosis, which means the hypomania-BD results should be interpreted cautiously and the hypomania-schizophrenia association could not be examined using the twin method. This also prevented the examination of the sensitivity and specificity of subsyndromal hypomania in predicting later psychiatric diagnoses. Future studies should consider exploring such relationships in a sample of individuals who have passed through the typical age of BD and schizophrenia onset. Second, hypomania was assessed with the parent-rated MDQ because studies have shown that parent report is more accurate in assessing adolescent hypomania than self- and teacher ratings.⁴³ Also, the MDQ is one of the best validated and discriminating BD instruments for youths.⁴³ However, late adolescence/early adulthood marks a transitional period into independence where parents are not fully aware of their children’s behavior and emotions. Future studies should use self-report instruments particularly in older patients. The MDQ does not enquire about symptom duration, a key component of the diagnostic criteria for hypomanic/manic episodes; this would be useful to incorporate in our high-risk group classification. Finally, BD cases were identified using official diagnoses and lithium prescription. Lithium is predominantly used to treat BD but can be used for treatment-resistant depression and hypomania that does not meet BD diagnostic criteria.⁴⁴ Therefore, some BD cases may have been misclassified here.

Conclusions

To our knowledge, this is the first twin study in a nonclinical sample of young adults to investigate the specific and shared etiology of subsyndromal hypomania and BD, MDD, and schizophrenia, which also uses a PRS approach. Higher heritability estimates for hypomania were observed for male compared with female individuals. Moderate genetic and nonshared environmental correlations between hypomania and BD were found. We found this genetic overlap was not explained by common genetic factors using the BD PRS. Hypomania was significantly associated with schizophrenia and MDD PRS. These results suggest a shared cause between subsyndromal hypomania and BD, MDD, and schizophrenia. The findings are also relevant to exploring a BD dimensional model, providing preliminary support for hypomania representing a quantitative trait underlying this disorder. However, more research is needed for firm conclusions to be drawn.

Supplement.

eAppendix.

eFigure. Strobe flow diagram of CATSS participation

eTable 1. Univariate assumptions testing

eTable 2. Twin model fit statistics

eTable 3. ACE joint categorical-continuous bivariate model between hypomania and bipolar disorder

eReferences.