Leveraging electronic health records to study pleiotropic effects on bipolar disorder and medical comorbidities

Abstract

Patients with bipolar disorder (BD) have a high prevalence of comorbid medical illness. However, the mechanisms underlying these comorbidities with BD are not well known. Certain genetic variants may have pleiotropic effects, increasing the risk of BD and other medical illnesses simultaneously. In this study, we evaluated the association of BD-susceptibility genetic variants with various medical conditions that tend to co-exist with BD, using electronic health records (EHR) data linked to genome-wide single-nucleotide polymorphism (SNP) data. Data from 7316 Caucasian subjects were used to test the association of 19 EHR-derived phenotypes with 34 SNPs that were previously reported to be associated with BD. After Bonferroni multiple testing correction, P<7.7 × 10 ^-5 was considered statistically significant. The top association findings suggested that the BD risk alleles at SNP rs4765913 in CACNA1C gene and rs7042161 in SVEP1 may be associated with increased risk of 'cardiac dysrhythmias' (odds ratio (OR)=1.1, P=3.4 × 10 ^-3 ) and 'essential hypertension' (OR=1.1, P=3.5 × 10^-3 ), respectively. Although these associations are not statistically significant after multiple testing correction, both genes have been previously implicated with cardiovascular phenotypes. Moreover, we present additional evidence supporting these associations, particularly the association of the SVEP1 SNP with hypertension. This study shows the potential for EHR-based analyses of large cohorts to discover pleiotropic effects contributing to complex psychiatric traits and commonly co-occurring medical conditions.

OPEN

Citation: Transl Psychiatry (2016) 6, e870; doi:http://dx.doi.org/10.1038/tp.2016.138

Web End =10.1038/tp.2016.138

http://www.nature.com/tp

Web End =www.nature.com/tp

ORIGINAL ARTICLE

Leveraging electronic health records to study pleiotropic effects on bipolar disorder and medical comorbidities

ML Prieto1,2,8, E Ryu3,8, GD Jenkins3, A Batzler3, MM Nassan1, AB Cuellar-Barboza4, J Pathak5, SL McElroy6,7, MA Frye1 and JM Biernacka1,3

Patients with bipolar disorder (BD) have a high prevalence of comorbid medical illness. However, the mechanisms underlying these comorbidities with BD are not well known. Certain genetic variants may have pleiotropic effects, increasing the risk of BD and other medical illnesses simultaneously. In this study, we evaluated the association of BD-susceptibility genetic variants with various medical conditions that tend to co-exist with BD, using electronic health records (EHR) data linked to genome-wide single-nucleotide polymorphism (SNP) data. Data from 7316 Caucasian subjects were used to test the association of 19 EHR-derived phenotypes with 34 SNPs that were previously reported to be associated with BD. After Bonferroni multiple testing correction, Po7.7 105 was considered statistically signicant. The top association ndings suggested that the BD risk alleles at SNP rs4765913 in CACNA1C gene and rs7042161 in SVEP1 may be associated with increased risk of cardiac dysrhythmias (odds ratio (OR) = 1.1, P = 3.4 103) and essential hypertension (OR = 1.1, P = 3.5 103), respectively. Although these associations arenot statistically signicant after multiple testing correction, both genes have been previously implicated with cardiovascular phenotypes. Moreover, we present additional evidence supporting these associations, particularly the association of the SVEP1 SNP with hypertension. This study shows the potential for EHR-based analyses of large cohorts to discover pleiotropic effects

contributing to complex psychiatric traits and commonly co-occurring medical conditions.

Translational Psychiatry (2016) 6, e870; doi:http://dx.doi.org/10.1038/tp.2016.138

Web End =10.1038/tp.2016.138 ; published online 16 August 2016

INTRODUCTIONPatients with bipolar disorder (BD) have higher early mortality rates compared with the general population.1,2 For example, a study of patients with BD from Scandinavian countries reported that their life expectancy was reduced by an average of 12.7 to19.8 years among men and 11 to 16.2 years among women.3 Even though suicide is a well-known cause of death in this population, increasing evidence has shown a large contribution of medical illness, such as cardiovascular disease,4,5 to premature mortality in patients with BD.1 The prevalence of medical comorbidities in this population is high,6 encompassing cardiovascular, neurovascular, neurologic and metabolic/endocrine diseases,1,710 and it is aggra

vated by the use of mood stabilizers and atypical antipsychotics, which are associated with weight gain, glucose disturbances and diabetes.1113 Therefore, it is imperative to better understand the mechanisms underlying these phenomena to design interventions to treat and prevent these conditions among patients with BD.

BD is one of the most heritable psychiatric diseases, demonstrating 85% heritability in a large twin study.14 Genome-wide association studies (GWAS) have identied several genetic variants associated with BD risk,1518 including genes such as CACNA1C, ODZ4, ITIH3ITIH4, ANK3, SYNE1 and TRANK1. Moreover, many medical diseases frequently occurring in patients with BD are also complex heritable conditions.19 Thus, one potential mechanism underlying the increased prevalence of certain medical comorbidities in patients with BD may be pleiotropic effects of genetic

variants associated with both BD and these medical conditions. For example, BD has been reported to be associated with the interaction between body mass index and variants in the TCF7L2 gene, 20 which is a well-established type 2 diabetes risk gene.21,22 Similarly, myocardial infarction risk has been reported to be associated with a genetic variant in the ITIH3ITIH4 genes,23 which has also been implicated in the risk for BD.16 Therefore, it is plausible that other common genetic variants may have pleiotropic effects, contributing to the risk for BD and its common medical comorbidities.24

Electronic health records (EHRs) are being increasingly utilized in research to identify patients with phenotypes of interest. When genetic data are available on large cohorts with EHR data, EHR-based phenotyping allows for testing for association of genetic variants with a wide range of medical conditions. The major advantage of this approach, which has been referred to as a phenome-wide association study, or PheWAS, is that it provides a cost-effective approach to testing for association of genetic variants with multiple phenotypes.2527 A recent study based on over 13 000 subjects reported that 66% of previously reported GWAS associations were replicated by this PheWAS approach, and several potential pleiotropic effects were identied.25 Following this EHR-based PheWAS approach, but focusing on previously reported BD risk single-nucleotide polymorphisms (SNPs) and selected key BD medical comorbidities (metabolic/endocrine, cardiovascular and neurological diseases), the current study aimed

1Department of Psychiatry and Psychology, Mayo Clinic College of Medicine, Rochester, MN, USA; 2Universidad de los Andes, Facultad de Medicina, Departamento de Psiquiatra, Santiago, Chile; 3Department of Health Sciences Research, Mayo Clinic College of Medicine, Rochester, MN, USA; 4Department of Psychiatry, Universidad Autnoma de Nuevo Len, Nuevo Len, Mexico; 5Division of Health Informatics, Weill Cornell Medical College, Cornell University, New York, NY, USA; 6Lindner Center of HOPE, Mason, OH, USA and

7Department of Psychiatry and Behavioral Neuroscience, University of Cincinnati College of Medicine, Cincinnati, OH, USA. Correspondence: Dr JM Biernacka, Department of Health Sciences Research, Mayo Clinic, 200 First Street SW, Rochester, MN 55905, USA.

E-mail: mailto:[email protected]

Web End [email protected]

8These authors contributed equally to this work.

Received 5 November 2015; revised 13 May 2016; accepted 15 June 2016

Pleiotropy of bipolar disorder and comorbidities

ML Prieto et al

2

to test whether specic BD risk alleles are also associated with these medical illnesses.

MATERIALS AND METHODS Study subjects

This study used the data from 7316 self-reported white subjects that are part of the Mayo Genomic Consortium (MayoGC).28 MayoGC is a cohort of patients from Mayo Clinic with longitudinal EHR data and existing genetic data from prior GWAS of various medical conditions, including peripheral arterial disease, venous thromboembolism, pancreatic cancer, brain cancer, lung cancer, ovarian cancer and chronic lymphocytic leukemia. For the cancer studies, disease cases are not included in the MayoGC.28 All subjects with data in the MayoGC gave general research consent to share genetic data with other researchers and to link the GWAS data with their EHR. This study was approved by the Mayo Clinic Institutional Review Board.

Genotype dataThe MayoGC participants were genotyped using various genotyping platforms as part of the original contributing studies, including the Illumina 660, 610, 550, and 370 (San Diego, CA, USA) and Affymetrix 6.0 (Santa Clara, CA, USA).28 For each study, standard quality control was performed. Briey, SNPs with call rate o95% were removed, and subjects with missing genotype rates 45% were excluded. Subjects genotyped in more than one contributing study were identied and GWAS data with the highest call rate was retained for the analysis. All GWAS data was mapped to the forward strand and genome-wide imputation was performed using IMPUTE2 after pre-phasing the genotypes with the SHAPEIT program.29

The 1000 genomes cosmopolitan population was used as a reference panel for imputation, and imputed genotypes with imputation quality score (r2) of 0.3 or higher were included in the analysis.

Phenotypes based on EHRsAll available EHR data from January 1994 to January 2015 were searched for all International Classication of Diseases Ninth Revision (ICD-9) codes. The ICD-9 codes were then classied into disease categories based on the Clinical Classications Software (https://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp

Web End =https://www.hcup-us.ahrq.gov/toolssoft https://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp

Web End =ware/ccs/ccs.jsp ). Disease categories with strong epidemiological support of comorbidity with BD, excluding other mental disorders, (that is, metabolic/endocrine, cardiovascular and neurological diseases) were selected from all the CCS categories.3037 Nonspecic disorders (codes that include other, such as other circulatory disease) were excluded, owing to difculty in interpreting association results for these phenotypes. Subjects with at least one ICD-9 code in the category were classied as cases, whereas remaining subjects were classied as controls. Finally, only phenotypes with prevalence of at least 5% in our sample were included in the analysis (Table 1).

Selection of BD susceptibility SNPsThe National Human Genome Research Institute (NHGRI) catalog of published GWAS ndings38 was used to identify 308 SNPs that were previously reported as associated with BD (accessed on July 2013). These SNPs were ltered using results from a mega-analysis of 7481 BD cases and 9250 controls performed by the Psychiatric Genomics Consortium (PGC),15

selecting only SNPs with P-values o104 in the mega-analysis. Of those SNPs supported by the PGC data, pairwise SNP linkage disequilibrium was calculated to identify SNP clusters in high linkage disequilibrium (r2 0.9).

For each cluster, one SNP having the strongest association with BD risk in the PGC data was retained. This ltering strategy provided 34 SNPs with strong support for association with BD (Supplementary Table 1).

Statistical analysisAdjusting for contributing study and sex, logistic regression models were used to test the association between each SNP and each phenotype assuming log-additive allele effects. Odds ratios (ORs) for the effect of the minor allele on risk of the phenotype, and the corresponding nominal P-values are presented. After Bonferroni multiple testing correction, Po7.7 105 ( = 0.05/(34 19)) was considered statistically signicant.

All analyses were performed using R software (https://cran.r-project.org/

Web End =https://cran.r-project.org/).

ReplicationAlthough a direct replication cohort was not available, SNP-phenotype combinations corresponding to the top association signals were further examined using the PheWAS catalog (https://phewas.mc.vanderbilt.edu/

Web End =https://phewas.mc.vanderbilt.edu/), which represents a large collection of PheWAS results from the eMERGE network.25,39 In addition, phenotypic and genetic data from the Mayo Clinic Bipolar Disorder Biobank40,41 was used to examine the top SNP-

phenotype combinations. Replication analyses using the Bipolar Biobank were limited to phenotypes available for BD Biobank subjects that were assessed using the Cumulative Illness Rating Scale (CIRS), which collects self-reported severity of various medical illness categories on an ordinal scale.42 The relevant phenotypes were dichotomized (none or no impairment vs at least mild problem), and logistic regression was used to test for association of the phenotype with the additively coded genotype, adjusting for sex.

RESULTSThe median age of the 7316 MayoGC subjects was 72 years old at the time of EHR-based phenotyping (or age at death), and 53% were male (Table 1). Among the 19 phenotypes included in the analysis, disorders of lipid metabolism was the most prevalent (75%), followed by essential hypertension (65%).

The top three SNP-phenotype association results are shown in Table 2. Among all combinations between the 19 phenotypes and34 SNPs, the strongest association was observed between rs4765913 in the CACNA1C gene and the phenotype cardiac dysrhythmias (Table 2). The minor allele A of the SNP (minor allele frequency (MAF) = 0.21), which was the risk allele for BD in the PGC analysis (OR = 1.15, P = 1.4 106), showed an increased risk effect for cardiac dysrhythmias (OR = 1.13, P = 3.4 103). Similar

Table 1. Demographic characteristics of study subjects and frequencies of 19 phenotypes

Demographic characteristics and phenotypes Study subjects (n = 7316)

Age, median (25th75th percentiles) 72 (6380) Sex, male N (%) 3843 (53%)

Phenotypes, N (%)Metabolic/endocrine

Disorders of lipid metabolism 5468 (75%) Diabetes mellitus 3941 (46%) Thyroid disorders 1916 (26%)

CardiovascularEssential hypertension 2564 (65%) Cardiac dysrhythmias 3509 (52%) Coronary atherosclerosis and otherheart disease

2947 (40%)

Peripheral visceral atherosclerosis 2907 (40%) Heart valve disorders 2519 (34%) Conduction disorders 1937 (26%) Occlusion or stenosis of precerebralarteries

1144 (16%)

Aortic, peripheral and visceral artery aneurysms

1078 (15%)

Congestive heart failure; nonhypertensive

848 (12%)

Acute cerebrovascular disease 649 (8.9%) Peri-, endo- and myocarditis,cardiomyopathy

614 (8.4%)

Aortic and peripheral arterial embolism or thrombosis

603 (8.2%)

Acute myocardial infarction 520 (7.1%) Transient cerebral ischemia 485 (6.6%)

Neurological

Headache including migraine 1810 (25%) Epilepsy; convulsions 711 (9.7%)

Translational Psychiatry (2016), 1 5

Pleiotropy of bipolar disorder and comorbidities ML Prieto et al

3

Table 2. Top three most signicant associations between BD-associated SNPs and EHR-derived phenotypes

SNP Gene chr Position Minor allele MAF OR for BD riska P-value for BD riska Phenotype ORb P-valueb

rs4765913 CACNA1C 12 2290157 A 0.21 1.15 1.4 106 Cardiac dysrhythmias 1.13 3.4 103 rs7042161 SVEP1 9 112237084 T 0.34 0.90 7.3 105 Essential hypertension 0.90 3.5 103 rs10896135 C11orf80 11 66307578 C 0.26 0.88 1.8 106 Headache including migraine 1.12 8.5 103

Abbreviations: BD, bipolar disorder; EHR, electronic health record; MAF, minor allele frequency; OR, odds ratio; SNP, single-nucleotide polymorphism. aOR and P-value for the minor allele from the Psychiatric Genomics Consortium BD GWAS.15 bOR and P-value for the association between the SNP minor allele and the phenotype.

associations were observed between a protective BD allele (T) at SNP rs7042161 in the SVEP1 gene and essential hypertension (OR = 0.90; P = 3.5 103). The SNP rs10896135 (MAF = 0.26) in C11orf80 gene also showed some evidence of association with risk of headache including migraine; however, the protective BD allele of rs10896135 showed an increased risk effect for headache including migraine, which suggests that this association is likely a false-positive. These associations are not signicant after Bonferroni correction for multiple testing (threshold P = 7.7 105).

The PheWAS catalog (https://phewas.mc.vanderbilt.edu/

Web End =https://phewas.mc.vanderbilt.edu/) was queried in an attempt to obtain additional evidence for the top observed SNP-phenotype associations. The SVEP1 SNP with suggestive evidence of association with essential hypertension in our study is included in the online PheWAS catalog, with the results supporting our ndings (P = 0.0075 for association with hypertension; P = 0.0086 for association with essential hypertension). The CACNA1C SNP with suggestive evidence of association with cardiac dysrhythmias in our study (rs4765913) was not included in the analyses performed by the eMERGE network that are documented in the PheWAS catalog. In fact, rs216013 is the only CACNA1C SNP in the PheWAS catalog results, and the catalog reports nominally signicant evidence for association of this SNP with cardiac dysrhythmias (P = 0.034). Because this SNP is in very low linkage disequilibrium with rs4765913 (r2 = 0.012), the potential association of rs216013 with cardiac dysrhythmias does not constitute replication of the result in the MayoGC sample, but it may provide additional evidence for a role of CACNA1C in this phenotype. The PheWAS catalog did not include any association results for SNPs in C11orf80 gene; thus association of variation in this gene with headache/migraine could not be investigated using this resource.

Analysis of medical comorbidity phenotypes in the Mayo Clinic Bipolar Disorder Biobank precluded the replication of the associations between CACNA1C and cardiac dysrhythmias and between C11orf80 and headache including migraine, as these phenotypes were not assessed in the biobank cohort. However, the CIRS was used to collect information on hypertension, with 644 of the genotyped BD subjects reporting none or no impairment related to hypertension, and 245 of the subjects reporting at least a mild problem related to hypertension. Comparison of these groups of patients provided independent marginal evidence for association of the SVEP1 SNP rs7042161 with hypertension (OR = 0.80, P = 0.054 without sex adjustment; OR = 0.82, P = 0.082 after correction for sex).

DISCUSSIONThis study applied a focused EHR-based PheWAS approach to explore genetic pleiotropic effects, focusing on genetic variants associated with BD risk and phenotypes that have been observed to co-occur with BD. The EHR-based PheWAS approach has been successfully applied in several large-scale studies, for example, by the Electronic Medical Records and Genomes (eMERGE) Network.25 A recent study demonstrated that an EHR-based approach produced nearly identical association signals to analysis of curated casecontrol data on hypothyroidism.27 Another study

based on multiple populations in the National Health and Nutrition Examination Surveys identied 13 SNPs with evidence of pleiotropy.43 As more institutions are establishing biobanks for the purposes of biomedical research, and as the sample sizes of these collections increase, these types of studies will become more feasible and more powerful, providing the opportunity to investigate the genetic underpinnings of a variety of comorbid conditions. To our knowledge, this study represents the rst application of the PheWAS approach, specically in the context of investigating potential pleiotropic effects on BD risk and various comorbid medical conditions. We note that although recent efforts to study pleiotropy have often focused on determining total genome-wide genetic overlap between diseases,4446 for

example using polygenic risk score methods, the approach applied here is complementary to such efforts, as it attempts to identify specic SNPs that contribute to multiple diseases, in this case BD and comorbid medical conditions.

One nding of potential interest is the association between CACNA1C genetic variation and the cardiac dysrhythmias phenotype. SNPs in the CACNA1C gene are among the most replicated associations in GWAS of BD.15 Besides its role in the expression of calcium channels in the brain, the CACNA1C gene encodes voltage-gated calcium channels (CaV1.2) expressed in the cardiac myocites.47 Genetic variations in this gene have been reported to be associated with inherited arrhythmic disorders, involving shortened QTc interval, idiopathic ventricular brillation, early repolarization syndrome, J-wave syndromes, Brugada and Timothy syndromes and sudden cardiac death.47,48 Mutations in CACNA1C gene have also been associated with long QTc interval syndrome, hypertrophic cardiomyopathy, congenital heart defects and sudden cardiac death.49 In Japanese patients mutations in this gene were associated with Brugada syndrome.50

It is therefore plausible that patients with BD that have alterations in the CACNA1C gene, may also express abnormal calcium channels in the heart, predisposing them to severe arrhythmias, particularly when exposed to mood stabilizers and antipsychotics which may impact the QTc interval; these calcium channel alterations may affect cardiac electrophysiology and may increase the risk for sudden cardiac death.51 If CACNA1C genetic variations are further conrmed to contribute to both BD and cardiovascular disease risk, it may impact treatment strategies, suggesting that perhaps calcium channel blockers should be investigated as mood stabilizers with cardiac protective effects for patients with BD.

The association between SVEP1 (Sushi, Von Willebrand Factor Type A, EGF And Pentraxin Domain Containing 1) gene variants and essential hypertension also warrants attention. SVEP1 protein is composed of multiple domains: Sushi domain, von Willebrand factor type A domain, epidermal growth factor domain, hyaline repeat domain, and pentraxin domains.52,53 The protein encoded

by this gene has a role in several inammatory mechanisms and cell adhesion processes; the sushi domain belongs to the family of complement control proteins and SVEP1 protein is similar to the family of selectin adhesion molecules. Although this gene is not believed to have a direct role in blood pressure control, it interacts with the ICAM1 (intercellular adhesion molecule 1) gene, which

Translational Psychiatry (2016), 1 5

Pleiotropy of bipolar disorder and comorbidities ML Prieto et al

4

has been implicated in several cardiovascular phenotypes, encompassing atherosclerosis, coronary artery disease and hyper-tension owing to its role in inammatory processes.54 In particular, one study demonstrated an association between circulating levels of ICAM1 and blood pressure reduction in patients who underwent renal sympathetic denervation for refractory hypertension.55

Because the association between BD and hypertension has been well demonstrated,10,56 this potential pleiotropic effect of a variant in SVEP1 should be further investigated.

These results should be interpreted in light of the studys limitations. First, as in typical PheWAS, phenotyping was solely based on EHR, and subjects with at least one ICD9 code in a given category were considered as cases, whereas remaining subjects made up the control group. Underreporting of cases can be expected in such EHR-based studies as the lack of a given ICD9 code in a health record does not necessarily indicate absence of the medical condition. Furthermore, because the EHR-derived codes include billing codes used for screening purposes, some false-positive case assignments are expected. However, we performed sensitivity analyses requiring at least two ICD9 codes to dene a case, and the results were very similar (data not shown). Second, despite the fairly large overall sample, owing to the study design the number of cases for some of the phenotypes was relatively small (for example, 520 cases of acute myocardial infarction), providing limited power for some of the analyses. Third, because the study did not include patients with BD, we were not able to test for pleiotropic effects in patients with both BD and the medical diseases we studied. Finally, the lack of a direct replication sample is a limitation, and further replication will be important to conrm the observed associations. Nevertheless, the additional results from the PheWAS catalog and the Mayo Clinic Bipolar Disorder Biobank lend some support for the top two observed associations.

In conclusion, this study illustrates the application of an EHR-based genetic PheWAS approach to the investigation of the pleiotropic effects contributing to medical comorbidities of psychiatric illness. With the increasing availability of genetic data, such studies are beginning to have a prominent role in biomedical research, and may make a signicant contribution to the understanding of complex genetic diseases.

CONFLICT OF INTEREST

The authors declare no conict of interest.

ACKNOWLEDGMENTS

This study was funded by the Marriot Foundation and the Mayo Clinic Center for Individualized Medicine.

REFERENCES

1 Crump C, Sundquist K, Winkleby MA, Sundquist J. Comorbidities and mortality in bipolar disorder: a Swedish national cohort study. JAMA Psychiatry 2013; 70: 931939.

2 Osby U, Brandt L, Correia N, Ekbom A, Sparen P. Excess mortality in bipolar and unipolar disorder in Sweden. Arch Gen Psychiatry 2001; 58: 844850.

3 Laursen TM, Wahlbeck K, Hllgren J, Westman J, sby U, Alinaghizadeh H et al. Life expectancy and death by diseases of the circulatory system in patients with bipolar disorder or schizophrenia in the Nordic countries. PloS one 2013; 8: e67133.

4 Westman J, Hallgren J, Wahlbeck K, Erlinge D, Alfredsson L, Osby U. Cardiovascular mortality in bipolar disorder: a population-based cohort study in Sweden. BMJ Open 2013; 3: e002373.

5 Prieto ML, Cuellar-Barboza AB, Bobo WV, Roger VL, Bellivier F, Leboyer M et al. Risk of myocardial infarction and stroke in bipolar disorder: a systematic review and exploratory meta-analysis. Acta Psychiatr Scand 2014; 130: 342353.

6 Krishnan KR. Psychiatric and medical comorbidities of bipolar disorder. Psychosom Med 2005; 67: 18.

7 Fornaro M, Stubbs B. A meta-analysis investigating the prevalence and moderators of migraines among people with bipolar disorder. J Affect Disord 2015; 178: 8897.

8 Prieto ML, Cuellar-Barboza AB, Bobo WV, Roger VL, Bellivier F, Leboyer M et al. Risk of myocardial infarction and stroke in bipolar disorder: a systematic review and exploratory meta-analysis. Acta Psychiatr Scand 2014; 130: 342353.

9 Weber NS, Fisher JA, Cowan DN, Niebuhr DW. Psychiatric and general medical conditions comorbid with bipolar disorder in the National Hospital Discharge Survey. Psychiatr Serv 2011; 62: 11521158.

10 Goldstein BI, Fagiolini A, Houck P, Kupfer DJ. Cardiovascular disease and hyper-tension among adults with bipolar I disorder in the United States. Bipolar Disord 2009; 11: 657662.

11 Bobo WV, Cooper WO, Stein C, Olfson M, Graham D, Daugherty J et al. Anti-psychotics and the risk of type 2 diabetes mellitus in children and youth. JAMA Psychiatry 2013; 70: 10671075.

12 Guo JJ, Keck PE Jr., Corey-Lisle PK, Li H, Jiang D, Jang R et al. Risk of diabetes mellitus associated with atypical antipsychotic use among Medicaid patients with bipolar disorder: a nested case-control study. Pharmacotherapy 2007; 27: 2735.

13 Kessing LV, Thomsen AF, Mogensen UB, Andersen PK. Treatment with anti-psychotics and the risk of diabetes in clinical practice. Br J Psychiatry 2010; 197: 266271.

14 McGufn P, Rijsdijk F, Andrew M, Sham P, Katz R, Cardno A. The heritability of bipolar affective disorder and the genetic relationship to unipolar depression. Arch Gen Psychiatry 2003; 60: 497502.

15 Psychiatric GCBDWG. Large-scale genome-wide association analysis of bipolar disorder identies a new susceptibility locus near ODZ4. Nat Genet 2011; 43: 977983.

16 Ripke S, Sanders AR, Kendler KS, Levinson DF, Sklar P, Holmans PA et al. Genome-wide association study identies ve new schizophrenia loci. Nat Genet 2011; 43: 969U977.

17 Muhleisen TW, Leber M, Schulze TG, Strohmaier J, Degenhardt F, Treutlein J et al. Genome-wide association study reveals two new risk loci for bipolar disorder. Nat Commun 2014; 5: 3339.

18 Cichon S, Muehleisen TW, Degenhardt FA, Mattheisen M, Miro X, Strohmaier J et al. Genome-wide association study identies genetic variation in neurocan as a susceptibility factor for bipolar disorder. Am J Hum Genet 2011; 88: 396.

19 Sun X, Yu W, Hu C. Genetics of type 2 diabetes: insights into the pathogenesis and its clinical application. Bio Med Res Int 2014; 2014: 926713.

20 Winham SJ, Cuellar-Barboza AB, Oliveros A, McElroy SL, Crow S, Colby C et al. Genome-wide association study of bipolar disorder accounting for effect of body mass index identies a new risk allele in TCF7L2. Mol Psychiatry 2013; 130: 342353.

21 Grant SF, Thorleifsson G, Reynisdottir I, Benediktsson R, Manolescu A, Sainz J et al. Variant of transcription factor 7-like 2 (TCF7L2) gene confers risk of type 2 diabetes. Nat Genet 2006; 38: 320323.

22 Saxena R, Elbers CC, Guo Y, Peter I, Gaunt TR, Mega JL et al. Large-scale gene-centric meta-analysis across 39 studies identies type 2 diabetes loci. Am J Hum Genet 2012; 90: 410425.

23 Ebana Y, Ozaki K, Inoue K, Sato H, Iida A, Lwin H et al. A functional SNP in ITIH3 is associated with susceptibility to myocardial infarction. J Hum Genet 2007; 52: 220229.

24 Solovieff N, Cotsapas C, Lee PH, Purcell SM, Smoller JW. Pleiotropy in complex traits: challenges and strategies. Nat Rev Genet 2013; 14: 483495.

25 Denny JC, Bastarache L, Ritchie MD, Carroll RJ, Zink R, Mosley JD et al. Systematic comparison of phenome-wide association study of electronic medical record data and genome-wide association study data. Nat Biotechnol 2013; 31: 11021110.

26 Ritchie MD, Denny JC, Zuvich RL, Crawford DC, Schildcrout JS, Bastarache L et al. Genome- and phenome-wide analyses of cardiac conduction identies markers of arrhythmia risk. Circulation 2013; 127: 13771385.

27 Denny JC, Crawford DC, Ritchie MD, Bielinski SJ, Basford MA, Bradford Y et al. Variants near FOXE1 are associated with hypothyroidism and other thyroid conditions: using electronic medical records for genome- and phenome-wide studies. Am J Hum Genet 2011; 89: 529542.

28 Bielinski SJ, Chai HS, Pathak J, Talwalkar JA, Limburg PJ, Gullerud RE et al. Mayo Genome Consortia: a genotype-phenotype resource for genome-wide association studies with an application to the analysis of circulating bilirubin levels. Mayo Clin Proc 2011; 86: 606614.

29 Hancock DB, Levy JL, Gaddis NC, Bierut LJ, Saccone NL, Page GP et al. Assessment of genotype imputation performance using 1000 Genomes in African American studies. PLoS ONE 2012; 7: e50610.

30 Evans-Lacko SE, Zeber JE, Gonzalez JM, Olvera RL. Medical comorbidity among youth diagnosed with bipolar disorder in the United States. J Clin Psychiatry 2009; 70: 14611466.

Translational Psychiatry (2016), 1 5

Pleiotropy of bipolar disorder and comorbidities ML Prieto et al

5

31 Jette N, Patten S, Williams J, Becker W, Wiebe S. Comorbidity of migraine and psychiatric disorders--a national population-based study. Headache 2008; 48: 501516.

32 Kilbourne AM, Cornelius JR, Han X, Pincus HA, Shad M, Salloum I et al. Burden of general medical conditions among individuals with bipolar disorder. Bipolar Disord 2004; 6: 368373.

33 Magalhaes PV, Kapczinski F, Nierenberg AA, Deckersbach T, Weisinger D, Dodd S et al. Illness burden and medical comorbidity in the systematic treatment enhancement program for bipolar disorder. Acta Psychiatr Scand 2012; 125: 303308.

34 Ortiz A, Cervantes P, Zlotnik G, van de Velde C, Slaney C, Garnham J et al. Cross-prevalence of migraine and bipolar disorder. Bipolar Disord 2010; 12: 397403.

35 Perron BE, Howard MO, Nienhuis JK, Bauer MS, Woodward AT, Kilbourne AM. Prevalence and burden of general medical conditions among adults with bipolar I disorder: results from the National Epidemiologic Survey on Alcohol and Related Conditions. J Clin Psychiatry 2009; 70: 14071415.

36 Sicras A, Rejas J, Navarro R, Serrat J, Blanca M. Metabolic syndrome in bipolar disorder: a cross-sectional assessment of a Health Management Organization database. Bipolar Disord 2008; 10: 607616.

37 Soreca I, Fagiolini A, Frank E, Houck PR, Thompson WK, Kupfer DJ. Relationship of general medical burden, duration of illness and age in patients with bipolar I disorder. J Psychiatr Res 2008; 42: 956961.

38 Welter D, MacArthur J, Morales J, Burdett T, Hall P, Junkins H et al. The NHGRI GWAS Catalog, a curated resource of SNP-trait associations. Nucleic Acids Res 2014; 42(Database issue): D1001D1006.

39 Denny JC, Ritchie MD, Basford MA, Pulley JM, Bastarache L, Brown-Gentry K et al. PheWAS: demonstrating the feasibility of a phenome-wide scan to discover gene-disease associations. Bioinformatics 2010; 26: 12051210.

40 Frye MA, McElroy SL, Fuentes M, Sutor B, Schak KM, Galardy CW et al. Development of a bipolar disorder biobank: differential phenotyping for subsequent biomarker analyses. Int J Bipolar Disord 2015; 3: 30.

41 Frye MA, Doederlein A, Koenig B, McElroy SL, Nassan M, Seymour LR et al. National survey and community advisory board development for a bipolar disorder biobank. Bipolar Disord 2015.

42 Prieto ML, McElroy SL, Hayes SN, Sutor B, Kung S, Bobo WV et al. Association between history of psychosis and cardiovascular disease in bipolar disorder. Bipolar Disord 2015; 17: 518527.

43 Hall MA, Verma A, Brown-Gentry KD, Goodloe R, Boston J, Wilson S et al. Detection of pleiotropy through a Phenome-wide association study (PheWAS) of epidemiologic data as part of the Environmental Architecture for Genes Linked to Environment (EAGLE) study. PLoS Genet 2014; 10: e1004678.

44 Krapohl E, Euesden J, Zabaneh D, Pingault JB, Rimfeld K, von Stumm S et al. Phenome-wide analysis of genome-wide polygenic scores. Mol Psychiatry; e-pub ahead of print 25 August 2015; doi: 10.1038/mp.2015.126.

45 Bulik-Sullivan B, Finucane HK, Anttila V, Gusev A, Day FR, Loh PR et al. An atlas of genetic correlations across human diseases and traits. Nat Genet 2015; 47: 12361241.

46 Li C, Yang C, Gelernter J, Zhao H. Improving genetic risk prediction by leveraging pleiotropy. Hum Genet 2014; 133: 639650.

47 Liao P, Soong TW. CaV1.2 channelopathies: from arrhythmias to autism, bipolar disorder, and immunodeciency. Pugers Arch 2010; 460: 353359.

48 Burashnikov E, Pfeiffer R, Barajas-Martinez H, Delpon E, Hu D, Desai M et al.

Mutations in the cardiac L-type calcium channel associated with inherited J-wave syndromes and sudden cardiac death. Heart Rhythm 2010; 7: 18721882.

49 Boczek NJ, Ye D, Jin F, Tester DJ, Huseby A, Bos JM et al. Identication and functional characterization of a Novel CACNA1C-mediated cardiac disorder characterized by prolonged qt intervals with hypertrophic cardiomyopathy, congenital heart defects, and sudden cardiac death. Circ Arrhythm Electrophysiol 2015; 8: 11221132.

50 Fukuyama M, Ohno S, Wang Q, Kimura H, Makiyama T, Itoh H et al. L-type calcium channel mutations in japanese patients with inherited arrhythmias. Circ J 2013; 77: 17991806.

51 Glassman AH, Bigger JT Jr. Antipsychotic drugs: prolonged QTc interval, torsade de pointes, and sudden death. Am J Psychiatry 2001; 158: 17741782.

52 Schwanzer-Pfeiffer D, Rossmanith E, Schildberger A, Falkenhagen D. Characterization of SVEP1, KIAA, and SRPX2 in an in vitro cell culture model of endotoxemia. Cell Immunol 2010; 263: 6570.

53 Gilges D, Vinit MA, Callebaut I, Coulombel L, Cacheux V, Romeo PH et al.

Polydom: a secreted protein with pentraxin, complement control protein, epidermal growth factor and von Willebrand factor A domains. Biochem J 2000; 352 (Pt 1): 4959.

54 Androulakis E, Tousoulis D, Papageorgiou N, Latsios G, Siasos G, Tsious C et al.

Inammation in hypertension: current therapeutic approaches. Curr Pharm Des 2011; 17: 41214131.

55 Dorr O, Liebetrau C, Mollmann H, Gaede L, Troidl C, Rixe J et al. Soluble fms-like tyrosine kinase-1 and endothelial adhesion molecules (intercellular cell adhesion molecule-1 and vascular cell adhesion molecule-1) as predictive markers for blood pressure reduction after renal sympathetic denervation. Hypertension 2014; 63: 984990.

56 Johannessen L, Strudsholm U, Foldager L, Munk-Jorgensen P. Increased risk of hypertension in patients with bipolar disorder and patients with anxiety compared to background population and patients with schizophrenia. J Affect Disord 2006; 95: 1317.

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/ http://creativecommons.org/licenses/by/4.0/

Web End =by/4.0/

The Author(s) 2016

Supplementary Information accompanies the paper on the Translational Psychiatry website (http://www.nature.com/tp)

Translational Psychiatry (2016), 1 5