Investigating the genetic variation underlying episodicity in major depressive disorder: Suggestive evidence for a bipolar contribution

Journal of Affective Disorders 155 (2014) 81–89

Contents lists available at ScienceDirect

Journal of Affective Disorders journal homepage: www.elsevier.com/locate/jad

Research report

Investigating the genetic variation underlying episodicity in major depressive disorder: Suggestive evidence for a bipolar contribution Panagiotis Ferentinos a,n, Margarita Rivera a,b, Marcus Ising c, Sarah L. Spain d, Sarah Cohen-Woods e, Amy W. Butler a,f, Nicholas Craddock g, Michael J. Owen g, Ania Korszun h, Lisa Jones i, Ian Jones g, Michael Gill j, John P. Rice k, Wolfgang Maier l, Ole Mors m, Marcella Rietschel n, Susanne Lucae c, Elisabeth B. Binder c, Martin Preisig o, Federica Tozzi p, Pierandrea Muglia q, Gerome Breen a,r, Ian W. Craig a, Anne E. Farmer a, Bertram Müller-Myhsok c, Peter McGuffin a, Cathryn M. Lewis a,d a MRC Social Genetic and Developmental Psychiatry Centre, Institute of Psychiatry, King's College London, 16 De Crespigny Park, London SE5 8AF, United Kingdom b Centro de Investigación Biomédica en Red de Salud Mental, CIBERSAM, University of Granada, Spain c Max Planck Institute of Psychiatry, Munich, Germany d Division of Genetics and Molecular Medicine, King's College London School of Medicine, Guy's Hospital, London, United Kingdom e Department of Psychiatry, University of Adelaide, Adelaide, Australia f Department of Psychiatry, University of Hong Kong, Hong Kong, Special Administrative Region, China g MRC Centre for Neuropsychiatric Genetics and Genomics, Neuroscience and Mental Health Research Institute, Cardiff University, Cardiff, United Kingdom h Barts and The London Medical School, Queen Mary University of London, London, United Kingdom i Department of Psychiatry, Neuropharmacology & Neurobiology Section, University of Birmingham, Birmingham, United Kingdom j Department of Psychiatry, Trinity Centre for Health Science, Dublin, Ireland k Department of Psychiatry, Washington University, St. Louis, Missouri, United States l Department of Psychiatry, University of Bonn, Bonn, Germany m Centre for Psychiatric Research, Aarhus University Hospital, Risskov, Denmark n Division of Genetic Epidemiology in Psychiatry, Central Institute of Mental Health, Mannheim, Germany o University Hospital Center and University of Lausanne, Lausanne, Switzerland p Aptuit Center for Drug Discovery & Development, Verona, Italy q Department of Psychiatry, University of Toronto, Toronto, Canada r NIHR Biomedical Research Centre for Mental Health, South London and Maudsley NHS Foundation Trust and Institute of Psychiatry, King's College London, London, United Kingdom

art ic l e i nf o

a b s t r a c t

Article history: Received 29 June 2013 Received in revised form 14 October 2013 Accepted 16 October 2013 Available online 29 October 2013

Background: Highly recurrent major depressive disorder (MDD) has reportedly increased risk of shifting to bipolar disorder; high recurrence frequency has, therefore, featured as evidence of ‘soft bipolarity’. We aimed to investigate the genetic underpinnings of total depressive episode count in recurrent MDD. Methods: Our primary sample included 1966 MDD cases with negative family history of bipolar disorder from the RADIANT studies. Total episode count was adjusted for gender, age, MDD duration, study and center before being tested for association with genotype in two separate genome-wide analyses (GWAS), in the full set and in a subset of 1364 cases with positive family history of MDD (FH þ). We also calculated polygenic scores from the Psychiatric Genomics Consortium MDD and bipolar disorder studies. Results: Episodicity (especially intermediate episode counts) was an independent index of MDD familial aggregation, replicating previous reports. The GWAS produced no genome-wide significant findings. The strongest signals were detected in the full set at MAGI1 (p¼5.1  10  7), previously associated with bipolar disorder, and in the FH þ subset at STIM1 (p¼ 3.9  10  6 after imputation), a calcium channel signaling gene. However, these findings failed to replicate in an independent Munich cohort. In the full set polygenic profile analyses, MDD polygenes predicted episodicity better than bipolar polygenes; however, in the FHþ subset, both polygenic scores performed similarly.

Key words: Bipolar spectrum Major depression Episode count Family history Genome-wide association study Polygenic

n

Corresponding author. Tel.: þ 44 2078480856. E-mail address: [email protected] (P. Ferentinos).

0165-0327/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jad.2013.10.027

82

P. Ferentinos et al. / Journal of Affective Disorders 155 (2014) 81–89

Limitations: Episode count was self-reported and, therefore, subject to recall bias. Conclusions: Our findings lend preliminary support to the hypothesis that highly recurrent MDD with FH þ is part of a ‘soft bipolar spectrum’ but await replication in larger cohorts. & 2013 Elsevier B.V. All rights reserved.

1. Introduction Major depressive disorder (MDD) is a common disease with a lifetime prevalence of 16% (Kessler et al., 2003). It is associated with considerable morbidity, excess mortality and compromise of functioning and quality of life. The etiopathogenesis of MDD remains unclear despite extensive research in the field. MDD tends to run in families; the sibling relative risk λsibling is estimated to be 3 while heritability has been calculated on average at 0.37 (95% CI 0.31–0.42) (Sullivan et al., 2000). Eight genome-wide association studies (GWAS) of MDD have been published (Kohli et al., 2011; Lewis et al., 2010; Muglia et al., 2010; Rietschel et al., 2010; Shi et al., 2011; Shyn et al., 2011; Sullivan et al., 2009; Wray et al., 2012), with one locus of possible genome-wide significance (Kohli et al., 2011). A recently published mega-analysis of GWAS studies in MDD by the Psychiatric Genomics Consortium (PGC) failed to identify any genome-wide significant findings (PGC, 2013). Phenotypic and genetic heterogeneity have been suggested as two main reasons why the genetic architecture of MDD is still elusive. One approach to improve power to detect genetic risk loci would therefore be to reduce phenotypic variation, using samples enriched in clinical subphenotypes of MDD that are associated with increased heritability. Recurrence and early onset have most often featured as clinical indices of higher familial aggregation and heritability (Kendler et al., 1999, 2005, 2007, 1994; McGuffin et al., 1996) while more limited evidence exists for other subphenotypes, such as severity, impairment, number of depressive symptoms endorsed, duration of the longest episode, clinical subtype (atypical, endogenous, psychotic) and comorbidities (Sullivan et al., 2000). For a given sample size, the power to identify genetic variants in GWAS should, therefore, increase when focusing on patients with a family history of depression, who have recurrent, early-onset, clinically ascertained MDD (McGuffin et al., 1996) which is reliably assessed through repeated measurements (Foley et al., 1998; Kendler et al., 1993). Recurrence (episode count of two or higher) is, in fact, a binary/ dichotomous approach to the episodicity subphenotype of MDD. Episodicity is most frequently examined as a binary rather than a quantitative/continuous phenotype in family/genetic studies of MDD. Recurrence is probably the most consistent index of familial aggregation of MDD (Sullivan et al., 2000). The pattern of association of quantitative episodicity (hereafter referred to as just ‘episodicity’) with familiality or heritability of MDD has been explored in a few population-based twin registries or birth cohorts (Kendler et al., 1999, 2007, 1994; Milne et al., 2009) and clinically ascertained samples (Hollon et al., 2006; Nierenberg et al., 2007). Findings have often been inconsistent. Interestingly, the risk for MDD of the co-twin of a twin with MDD was found to have a nonlinear inverted-U shaped association with the number of lifetime episodes reported by the index twin, maximized at 7–9 lifetime episodes (Kendler et al., 1999). Similarly, the proportion of STARnD patients with a positive family history of MDD was highest with intermediate (4–9) numbers of lifetime episodes (Hollon et al., 2006). Episode frequency is a highly familial trait in bipolar disorder (Fisfalen et al., 2005). Highly recurrent MDD cases have reportedly increased risk of shifting to bipolar disorder (Angst et al., 2005). MDD cases with bipolar family history seem to present with more episodes (as well as an earlier age at onset and an atypical pattern) (Akiskal, 2003; Ghaemi et al., 2002; Souery et al., 2012). Moreover, the offspring of probands with bipolar disorder have increased risk

of recurrent MDD (but not of single episodes) compared to offspring of healthy controls (Vandeleur et al., 2012). These observations have spurred the hypothesis that high recurrence frequency is evidence of a ‘soft bipolar’ component in MDD and highly recurrent MDD (often with early onset and a positive family history of bipolar disorder) has been included in ‘bipolar spectrum disorder’ (Akiskal, 2003; Ghaemi et al., 2002; Mitchell et al., 2008; Phelps et al., 2008). The aims of this study were: first, investigate the relationship between episodicity and family history of MDD in a sample of cases with recurrent MDD and negative family history of bipolar disorder; second, perform a GWAS of MDD episodicity (the first of its kind) to identify genetic variants associated with the number of depressive episodes; and, third, to explore contributions of MDD and bipolar disorder polygenes to MDD episodicity through analyses of polygenic scores calculated from the PGC MDD and bipolar disorder studies. For the last two objectives of the study, we also investigated whether family history of MDD had a moderator effect.

2. Subjects and methods 2.1. Samples A total of 1966 MDD cases from the RADIANT studies DeCC (N ¼994), DeNt (N ¼833) and GSK Case-Control study (N ¼ 139) with complete data sets on age, age at onset and episode count were analyzed as a discovery sample. The DeCC (Depression Case Control) study includes cases of recurrent depression fulfilling DSM-IV and/or ICD-10 criteria of at least moderate severity ascertained from three UK clinical sites (London, Cardiff and Birmingham) (Cohen-Woods et al., 2009). The DeNt (Depression Network) affected sibling pair linkage study (Farmer et al., 2004; McGuffin et al., 2005) comprises cases of recurrent depression of at least moderate severity, ascertained from three UK sites (London, Cardiff and Birmingham), four other European sites (Aarhus, Bonn, Dublin and Lausanne) and a site in St. Louis, USA. Only one proband from each family was genotyped and included in the analysis. The GSK Case-Control study included cases of recurrent depression collected in Bonn and Lausanne in collaboration with GSK, using exactly the same protocol as the DeNt study. In all three studies, only adults of European ancestry were recruited. Subjects were excluded if there was a history or family history (in first or second-degree relatives) of schizophrenia, schizoaffective disorder or bipolar disorder, if they had experienced mood incongruent psychotic symptoms, or if mood symptoms were solely related to alcohol or substance misuse or only secondary to medical illness or medication. The replication sample consisted of cases of recurrent depression of at least moderate severity recruited for a case-control study in the Munich area in collaboration with GSK (Muglia et al., 2010; Tozzi et al., 2008). Assessment instruments and inclusion/exclusion criteria were identical to those used in the DeCC and DeNT studies, except that subjects with a family history of bipolar disorder were not excluded. After removing the latter, a total of 372 cases with complete data sets on age, age at onset and episode count were included in the analysis. All cases in both samples were interviewed with the Schedules for Clinical Assessment in Neuropsychiatry (SCAN) (Wing et al., 1990), focusing on their worst and second-worst episodes of

P. Ferentinos et al. / Journal of Affective Disorders 155 (2014) 81–89

depression. Study coordinators for all studies were trained by one investigator (A.E.F.), ensuring homogeneity of clinical assessment methods. Age at onset was recorded in SCAN items 1.016, 1.046– 1.048 and 6.025; episode count was recorded in SCAN items 1.053 and 6.030. Family history of MDD in first-degree relatives was also extracted from SCAN item 1.045. All study participants provided written informed consent and approval was obtained from local ethical committees. 2.2. Aim 1: bivariate and multivariate statistical analyses/ association with family history of MDD We first performed non-parametric bivariate analyses of episode count with gender, age, MDD duration and family history of MDD. For the multivariate analysis including all aforementioned variables, episode count was log transformed due to positive skewness. All statistical analyses were implemented with STATA 11.0 or SPSS 20.0.

83

explanatory potential. Therefore, we fitted a linear mixed model with log(episode count) as the dependent variable, gender, log (age) and log(MDD duration) as fixed effects, and study and center as random effects. Residuals of the fitted models were saved. As their distribution was positively skewed, we performed a rank normalization transformation using Blom's formula (Blom, 1958). In the second step, whole-genome association between ranknormalized adjusted episode count residuals (‘adjusted episodicity’) and SNPs was tested in the discovery sample using linear regression under an additive genetic model; five ancestry-informative PCs were included as covariates. The genomic control λ parameter was calculated to assess inflation in test statistics (Devlin and Roeder, 1999). Analyses were implemented using PLINK v1.07 (Purcell et al., 2007) and R (http://www.r-project.org). Two thresholds of significance were used to interpret association results: genome-wide significance (po5  10–8) and suggestive significance (po5  10–6) (Dudbridge and Gusnanto, 2008). The power of the study to detect association was calculated using QUANTO version 1.2.4 (Gauderman and Morrison, 2006).

2.3. Aim 2: genome-wide association of MDD episodicity 2.3.1. Genotyping Whole-genome genotyping in the RADIANT sample was performed using the Illumina HumanHap610-Quad BeadChip by the Centre National de Génotypage (CNG), France. Genotyping in the GSK-Munich sample was performed on the Illumina HumanHap550K platform at Illumina laboratories (San Diego, CA, USA). Full details of the sample collection and genotyping methods have been published (Lewis et al., 2010; Muglia et al., 2010). 2.3.2. Quality control (QC) In the discovery sample, stringent QC procedures were applied to individual and SNP data (for full details see (Lewis et al., 2010)). Exclusion criteria for individuals included a missing rate 41%, abnormal heterozygosity, a sex assignment that conflicted with phenotypic data, relatedness (up to second degree) with other study subjects, and non-European ancestry. SNPs with minor allele frequency o1% or showing departure from Hardy–Weinberg equilibrium (p o1  10  5) were excluded. A total of 471581 SNPs were finally analyzed in whole-genome association tests. After QC procedures, principal component (PC) analysis was performed using EIGENSTRAT (Price et al., 2006) and five ancestryinformative PCs were included as covariates in the GWAS. In the replication sample, standard QC procedures were applied to individual and SNP data (for full details see Muglia et al., 2010). A total of 511503 SNPs were finally eligible for analysis. No correction for population structure was necessary. The post-QC SNP sets of the discovery and replication samples overlapped substantially (427946 shared SNPs). 2.3.3. Whole-genome association tests Appropriate phenotype delineation was critical for the GWAS of MDD episodicity. We followed a two-step process in the GWAS. In the first step, episode count was adjusted for the confounding effects of gender, age, MDD duration (¼ age age at onset), study, and center. After the effect of all confounders had been removed, the residuals were then tested for association with genotype in a second step. In the first step, inspection of scatterplots showed that the relationship of episode count with age and MDD duration was not linear. Curve estimation tests were used to identify which link function (from linear, logarithmic, power, quadratic, cubic, exponential and logistic models) maximized the proportion of episode count variance explained (R2); the ‘power’ function log(episode count) ¼a þbnlog(age or MDD duration) outperformed the rest in

2.3.4. Imputation For genomic regions where any SNP had an association p-value of 1  10  5 or lower in the discovery sample, genotypes at 1000 Genomes EUR (December 2010) SNPs (Abecasis et al., 2010) within 200-kb regions flanking the most highly associated SNP were imputed using IMPUTE version 2 (Howie et al., 2009). Dosage probabilities of imputed SNPs with an info quality score (R2) of 0.8 or higher and MAF 4 0.01 were tested for association with the phenotype using SNPTEST version 2 (Marchini et al., 2007) so as to identify any SNPs with stronger association signals in the region. Five ancestry-informative PCs were included as covariates in the linear regression model, in a similar way to whole-genome association tests with the genotyped SNPs.

2.3.5. Replication SNPs with p-value o0.0001 in the discovery sample were followed up in the replication sample. Where the associated SNPs were not genotyped in the replication sample, SNPs in high linkage disequilibrium (r2 40.9) were substituted if available. A Bonferroni correction for the number of SNPs analyzed was applied and consistency in the direction of associations across samples was extracted.

2.4. Aim 3: common variants polygenic risk profile analysis Two sets of polygenic scores (PGMDD, MDD polygenic scores; PGBP, bipolar disorder polygenic scores) were calculated for the RADIANT sample using PLINK profile scoring procedure on the basis of the PGC MDD and bipolar disorder mega-analyses clumped files, respectively (PGC, 2011, 2013). Each set of scores was calculated on seven different SNP subsets, namely SNPs with association p-values o0.01, o0.05, o0.1, o 0.2, o 0.3, o0.4, o0.5 (Purcell et al., 2009). Scores were weighted by the log of the odds ratio for selected SNPs. For each SNP subset, three linear regression models were fitted using PGMDD, PGBP or both to predict adjusted episodicity; PCs were also included, as in the GWAS study. The joint model (with both PGMDD and PGBP) was compared using likelihood ratio tests to its nested submodels (including either PGMDD or PGBP) to assess the significance of the additional contribution of each polygenic score on top of the other. The proportion of variance of adjusted episodicity (R2) explained by each polygenic score was calculated as the difference in R2 between the compared models.

84

P. Ferentinos et al. / Journal of Affective Disorders 155 (2014) 81–89

duration (rho ¼0.533, p o0.001) but not with gender or FH, even after controlling for gender, (log) age and (log) MDD duration.

3. Results 3.1. Bivariate and multivariate analyses In the RADIANT sample, episode count was significantly correlated with female gender (p¼ 0.008), a longer duration of disease (rho ¼0.213, p o0.001) and positive family history of depression in first-degree relatives (positive FH, FHþ) (p o0.001). In the multiple regression of (log) episode count on gender, log(age), log(MDD duration) and FH, the only significant predictors were log(MDD duration) (po 0.001) and FH (p o0.001); beta coefficient for FH was 0.047 (SE 0.012), meaning that (keeping other covariates constant) FH þ cases had 1.11 (95% CI 1.06–1.17) times more episodes than FH-ones. The proportion of cases with FHþ was highest for intermediate episode counts (4–6 episodes) (Fig. 1). FH was associated with binned episode count (p o0.001) and the association remained significant after correcting for the effect of gender, age and MDD duration (likelihood ratio test, po 0.001). The demographic and clinical characteristics of the discovery and replication samples stratified by FH are presented in Table 1. The replication sample (N ¼372) had fewer cases with FHþ (43.6% vs 69.4%, p o0.001) and fewer females (p ¼0.004) compared to the discovery sample; patients were slightly older, had a much later age at onset, a shorter duration of disease, a greater number and more frequent episodes than RADIANT cases (all po 0.001; see Table 1). Episode count in the GSK-Munich sample was significantly correlated with age (rho¼0.225, p o0.001) and MDD

Fig. 1. Proportion of RADIANT cases (N ¼ 1966) with positive family history of depression (FH þ) by episode count bins (2 episodes, N ¼944; 3 episodes, N ¼ 606; 4–6 episodes, N ¼ 244; 7–17 episodes, N ¼ 129; 20–50 episodes, N ¼ 43).

3.2. Genome-wide association analysis Given the independent significant association of episode count with FH þ in the RADIANT sample (see Section 3.1), we performed two separate genome-wide analyses, one for all MDD cases (N ¼1966), and one for cases with FH þ (N ¼1364); we did not extend our analyses to the FH-subset due to its small sample size. In each genome-wide analysis, we followed a two-step process. In the first step, we adjusted log(episode count) for various confounders (gender, log(age), log(MDD duration), study, center) in two distinct linear mixed models, one for each dataset, saved the residuals and rank-normalized them. In both datasets, log(episode count) was significantly associated with log(MDD duration) (p o0.001) and center (p ¼0.03). In the second step, wholegenome association between rank-normalized adjusted episode count residuals and SNPs was tested using 5 PCs as covariates. No SNP attained genome-wide significance in either RADIANT analysis (all cases and FHþ subset). Genomic control λ values were very low at 1.006 and 1.008, respectively (Fig. S1). In the full RADIANT analysis, two intronic SNPs, in MAGI1 (membrane associated guanylate kinase) (3p14.1) and in ADSS (adenylosuccinate synthase) (1q44), were associated at suggestive-level significance (5.15  10  7 and 4.32  10  6, respectively; Tables 2 and S1, Fig. S2a). There were no SNPs of suggestive significance in the FH þ subset (Fig. S2b); the lowest p-value (5.03  10  6) was observed for an intergenic SNP within a 1.7 Mb gene desert (7p12.1) between COBL and POM121L12. Of note, STIM1 (stromal interaction molecule 1) (11p15.5) harbored 12 intronic top SNPs (p-values o0.0001) in high linkage disequilibrium, including the one with the second lowest p-value in the FHþ subset (Table S2). Genotype imputation was performed in three genomic regions in the all cases RADIANT set and two regions in the FHþ subset, where SNPs with p-values o1  10  5 were located. Analysis of imputed genotypes detected a stronger association signal only in the STIM1 region; 28 imputed SNPs had lower p-values than the top typed SNP in the region and three of them achieved suggestive-level significance (top imputed SNP had a p-value of 3.94  10  6 and an info score of 0.99) (Fig. S3). Power analyses were conducted in the two RADIANT datasets using QUANTO 1.2.4. With a SNP accounting for 1.5% of additive genetic variance, the study had a power of 81.2% at the suggestive level of significance and 50% at genome-wide significance to detect association in the all cases dataset (N ¼ 1966). In the FH þ

Table 1 Demographic and clinical characteristics of the discovery and replication samples stratified by family history of depression in first-degree relatives (FH). RADIANT

Gender (females, %) Age (y) Age at onset (y) MDD duration (y) Episode count Lifetime Episode frequency

GSK-MUNICH

All cases (N ¼1966)

FHþ subset (N ¼1364)

All cases (N ¼372)

FHþ subset (N¼ 162)

72.7 46.1 712.0 (18–85) 22.7 7 11.6 (1–74) 23.3 7 13.5 (1–71) 3.8 7 4.8 (2–50) 0.247 0.28 (0.03–3.0)

74.1 45.4 7 11.9 (18–85) 22.0 7 11.2 (2–74) 23.4 7 13.2 (1–71) 4.0 7 4.9 (2–50) 0.25 7 0.29 (0.03–3.0)

65.3 49.47 13.2 (19–87) 35.07 13.5 (8–73) 14.4 711.3 (1–63) 5.0 7 4.2 (2–30) 0.55 7 0.49 (0.05–3.0)

67.3 46.8 7 12.9 (19–80) 32.17 12.1 (8–73) 14.7 710.5 (1–55) 5.3 7 4.2 (2–30) 0.54 7 0.46 (0.05–.5)

Data presented as % or mean 7SD (range). FHþ ¼ positive family history of depression (first-degree relatives); y ¼ years. a

Comparisons between RADIANT and GSK-MUNICH cases

Mann–Whitney U test.

χ2 ¼ 8.3, p ¼0.004 t¼ 4.49 (df ¼492.9), p o 0.001 z ¼15.8, p o0.001a z ¼12.1, p o 0.001a z ¼12.0, p o 0.001a z ¼17.4, p o 0.001a

P. Ferentinos et al. / Journal of Affective Disorders 155 (2014) 81–89

85

Table 2 Most significant findings (10 SNPs) in the RADIANT sample (all cases, N ¼ 1966 and subset with positive family history of depression in first-degree relatives, N ¼1364) with results from GSK-Munich study also shown; only top SNP for each independent genomic region is presented. CHR

SNP

BP position

A1/A2

Annotationa

RADIANT MAFb

GSK-MUNICH

Betab

p-Value

Rankc

Betab

p-Value

DAd

Ranke

0.042 0.459 0.817 0.688 0.133 0.276 0.922 0.094 0.246 0.733

    þ  þ þ  

3 21 35 29 5 14 41 4 12 31

0.036 0.425 0.919 0.934 0.735 0.455 0.559 0.036

  þ   þ  þ

2 18 47 52 42 22 30 1

0.940

þ

53

All cases 3 rs17295791 1 rs3127462 7 rs791608 13 rs419282 5 rs12523265 13 rs7981459 5 rs10036915 4 rs12374311 2 rs1347310 4 rs1562147

65776646 242664515 127660228 68170158 178017822 104799655 63899435 161444004 79444480 85579009

C/T T/C T/C C/A A/G A/G T/C G/A A/G C/T

MAGI1 (intronic) ADSS (intronic) SND1|LEP (intergenic) LOC730236|KLHL1 (intergenic) CLK4|ZNF354A (intergenic) SLC10A2|DAOA (intergenic) RGS7BP (intronic) RAPGEF2|FSTL5 (intergenic) REG3A|CTNNA2 (intergenic) LOC152845|NKX6-1 (intergenic)

N ¼1966 0.07 0.315 0.22 0.175 0.07  0.281 0.49 0.140 0.46  0.136 0.41  0.137 0.17  0.177 0.31  0.141 0.31 0.138 0.07  0.255

5.15E  07 4.32E  06 5.41E  06 1.18E  05 2.12E  05 2.51E  05 2.62E  05 3.32E  05 4.08E  05 4.14E  05

1 2 3 5 6 7 8 10 11 12

N ¼372  0.330  0.063 0.038  0.030  0.108 0.083  0.011  0.131  0.091 0.044

FH þSubsets 7 rs2716764 11 rs10767811 13 rs5027696 5 rs12523265 8 rs13270766 8 rs7822331 6 rs1553116 5 rs4566837 5 rs7732660f 13 rs7336551

52063111 4022974 33949536 178017822 120037914 40782946 155992324 2136797 177946729 102074154

A/C A/G C/T A/G A/G T/C G/T G/A G/A A/G

LOC100131871|POM121L12 (intergenic) STIM1 (intronic) RFC3|NBEA (intergenic) CLK4|ZNF354A (intergenic) TNFRSF11B|COLEC10 (intergenic) ZMAT4 (intronic) NOX3|ARID1B (intergenic) IRX4|IRX2 (intergenic) COL23A1 (intronic) TPP2 (intronic)

N ¼1364 0.43 0.175 0.32 0.183 0.36 0.175 0.45  0.164 0.10 0.273 0.44 0.164 0.25  0.185 0.33  0.176 0.46  0.158 0.39  0.163

5.03E  06 9.00E  06 1.28E  05 1.47E  05 1.48E  05 1.92E  05 2.47E  05 2.66E  05 2.99E  05 3.13E  05

1 2 6 8 9 13 14 15 16 17

N ¼162  0.264  0.093 0.011 0.009  0.071 0.081 0.075  0.253 f  0.009

FHþ ¼ positive family history of depression (first-degree relatives); A1 ¼minor allele. a

Annotation based on Illumina 610-Quad annotation file. For A1 allele. Rank based on p-value. d DA ¼Direction of association (as evident through beta signs) across samples ( þ ¼consistent,  ¼inconsistent). e Rank based on replication p-value (top RADIANT SNPs followed-up in GSK-Munich). f SNP not genotyped in GSK-Munich sample. b c

subset (N ¼1364), power was 75.3% at the suggestive level of significance and 42% at genome-wide significance for a SNP accounting for 2% of additive genetic variance. 3.2.1. Replication SNPs with p-value o 0.0001 in RADIANT (41 SNPs in the all cases dataset, 56 SNPs in the FHþ subset) were followed-up into the respective GSK-Munich analyses (Tables S1 and S2). No SNP was associated at the Bonferroni-corrected p-value thresholds (0.00122 and 0.00089, respectively). Beta values were not in the same direction as the discovery sample more often than expected by chance (sign tests, p ¼0.76 and 0.098, respectively). 3.2.2. Family history of depression specificity We tested whether GWAS results were specific to FH subsets. To this end, we fitted a model with an interaction between FH and additive genetic effects. Table S3 presents top SNPs in the FHþ subset followed-up into the all cases set as well as interaction term p-values, which reflect specificity of association in FHþ cases. Out of all top FHþ subset SNPs, an intergenic SNP downstream of ARMC1 (8q13.1) had the lowest interaction p-value (4.91  10  6, also lowest genome-wide).

0.007 **

0.006

*

0.004 0.003 0.002

**

*

0.005 * *

*

*

*

*

*

0.001 0

<0.01 <0.05 <0.1

<0.2

<0.3

<0.4

<0.5

0.007 *

0.006

*

0.005 0.004

*

* *

*

* *

*

*

PG-MDD PG-BP

0.003 0.002

* p<0.05

3.3. Polygenic profile analysis 0.001

Fig. 2 and Table S4 present the proportion of variance (R2) of adjusted episodicity explained by MDD or bipolar disorder polygenic scores and related significance p-values in the two RADIANT sets. In the all cases dataset, MDD polygenic scores predicted episodicity better than bipolar disorder polygenic scores. However, in the FHþ subset, both polygenic scores performed similarly

0

<0.01 <0.05 <0.1

<0.2

<0.3

<0.4

<0.5

Fig. 2. Proportion of variance of adjusted episodicity (R2) explained by major depression (PG-MDD) and bipolar disorder (PG-BP) polygenic scores in RADIANT all cases dataset and FHþ subset, by SNP p-value threshold. (a) All cases dataset (N ¼ 1966) and (b) FHþsubset (N ¼1364).

86

P. Ferentinos et al. / Journal of Affective Disorders 155 (2014) 81–89

while bipolar scores outperformed MDD scores at some SNP subsets (respective p-values 0.004 vs 0.022 for the po 0.2 SNP subset). In order to formally test whether contributions of MDD and bipolar polygenic scores differed by FH subsets, we fitted additional models with an interaction term between the polygenes and FH. Interaction effects for bipolar polygenes were stronger than for MDD polygenes but were non-significant (Table S5; for the p o0.3 SNP subset, bipolar p ¼0.065 and MDD p ¼0.83).

4. Discussion This study was one of very few GWAS on MDD subphenotypes conducted to date (Edwards et al., 2012; PGC, 2013; Power et al., 2012; Schosser et al., 2011; Shi et al., 2011). The analysis of subphenotypes is important for two reasons. First, the course, illness burden, and long-term prognosis of MDD may be moderated by clinical subphenotypes, such as age at onset (Korten et al., 2011; Zisook et al., 2007), recurrence (Kocsis, 2006), comorbidities (RoyByrne et al., 2000), suicidality (Overholser et al., 1987), psychotic symptoms (Coryell et al., 1996) and depressive symptoms pattern (Wardenaar et al., 2012); therefore, research into their biology is of potential clinical benefit. Second, better understanding MDD subphenotypes might help elucidate the pathogenesis of MDD itself, given that it is phenotypically and genetically heterogeneous. This is the first GWAS, and one of only a few genetic studies, on episodicity in MDD as a continuous phenotype (Ho et al., 2000; Hollon et al., 2006; Kendler et al., 1999, 2007, 1994; Milne et al., 2009; Nierenberg et al., 2007). The main reasons why binary episodicity (recurrence) has prevailed in MDD genetic research are: first, recurrence is a cornerstone of MDD classification in current taxonomy (DSM-IV-TR, ICD-10); second, the risk of lifetime recurrence is believed to peak and level off after a second depressive episode (Judd, 1997); and third, episode count is less reliably assessed, especially in retrospect. However, considerable variability of lifetime episode count has been recorded in clinical samples and the risk of recurrence during a 10-year prospective follow-up period increased by 16% with each successive recurrence (Solomon et al., 2000). Therefore, studying both episodicity and recurrence might prove helpful in MDD genetic research. Positive family history of depression had a small but significant independent effect on episodicity in the RADIANT sample, suggesting that both episodicity and recurrence are indexes of familial aggregation of MDD. The proportion of FHþ cases was highest for intermediate episode counts, replicating previous reports of an ‘inverted U’ relationship (Hollon et al., 2006; Kendler et al., 1999). The proportion of FHþ cases in RADIANT was highest for an episode count of 4–6, which is closer to the one reported in Hollon et al. (2006) clinical sample (4–9) than in Kendler et al. (1999) twin registry (7–9). The profile of RADIANT cases (Table 1) was much closer to the former (means: age 41.2, age at onset 22.4 years) than to the latter (means: age 36.1, age at onset 25.9 years). Female gender, which has often been associated in the literature with higher episode frequencies (Smith et al., 2008), was a univariate predictor of episodicity in our sample but did not survive multivariate adjustment. Episodicity was also significantly predicted by MDD duration in both bivariate and multivariate analyses; this finding might be considered as obvious (the longer the duration of depression the greater is the number of episodes a patient is expected to have) but might also imply that patients with an earlier onset (and therefore a more familial form) of disease tend to have higher episode counts (Zisook et al., 2007). Previous studies have shown that identifying genome-wide associations with MDD or its subphenotypes is challenging, and we found no genome-wide significant SNPs in both all cases

dataset and FH þ subset of the discovery sample. The study was underpowered to detect weak genetic effects (additive genetic variance of 1.5% or lower). The RADIANT SNPs with strongest evidence for association failed to replicate in the GSK-Munich sample. Potential reasons for this are genetic and phenotypic heterogeneity of MDD, reflected in our replication sample having a different demographic and clinical profile from the primary sample (fewer females, fewer FHþ cases, later age at onset, briefer disease duration). These mismatches may have unpredictably compromised inference, e.g. earlier age at onset may be associated with higher heritability (Kendler et al., 2005) while longer disease duration might inflate episode count recall bias (Patten et al., 2012). Family studies have shown that relatives of MDD probands have equally increased risk of MDD (RR 3.6) and bipolar disorder (RR 3.5) compared to those of controls (Merikangas and Yu, 2002). MDD and bipolar disorder are genetically correlated and share a polygenic component (McGuffin et al., 2003; Schulze et al., in press; Smoller et al., 2013); genetic susceptibility loci common to both disorders were recently identified (Green et al., 2009; Huang et al., 2010; McMahon et al., 2010). Furthermore, it has been suggested that latent bipolar liability might also manifest as specific MDD subphenotypes; in this context, high depressive episode frequency is an index of ‘soft bipolarity’ (Akiskal, 2003) and highly recurrent MDD forms part of the so-called ‘bipolar spectrum’ (Akiskal, 2003; Phelps et al., 2008). Our polygenic profile analysis showed that both MDD and bipolar polygenic scores independently contributed to episodicity in the RADIANT sample. In the all cases set, MDD polygenes had a stronger contribution to episodicity than bipolar polygenes; however, in FHþ cases bipolar polygenic scores explained as much of the variability in the episodicity phenotype as MDD polygenic scores. These findings provide for the first time suggestive evidence that episodicity in MDD is under partial genetic control by a bipolar polygenic component, especially in FH þ cases; they also lend preliminary support to the hypothesis that highly recurrent MDD with FHþ is part of a ‘soft bipolar spectrum’. We looked back at our GWAS results for evidence in support of the polygenic profile analysis findings. Many top SNPs in the analyses on the discovery sample were located in genomic regions previously linked to bipolar disorder (1q44, 8q13.1, 11p15.5) (Serretti and Mandelli, 2008). Furthermore, copy number variation in MAG1, our top gene in the all cases dataset (encoding a scaffolding protein at cell–cell junctions) was recently associated with bipolar disorder (Karlsson et al., 2012). Calcium channel signaling genes have been associated with bipolar disorder (mainly CACNA1C) (Ferreira et al., 2008) but also seem to have cross-disorder pleiotropic effects (Green et al., 2009; Smoller et al., 2013). Interestingly, STIM1, our top gene (after imputation) in the FHþ subset, which has not been associated with psychiatric disorders, encodes an endoplasmic reticulum transmembrane protein that mediates calcium influx after depletion of intracellular calcium stores by gating store-operated calcium channels (Soboloff et al., 2012). Finally, identified genes of potential functional significance also include ADSS (encoding the enzyme adenylosuccinate synthetase which catalyzes the first step in the conversion of inosine monophosphate to adenosine monophosphate) and ARMC1 (of yet unknown function). This study had several limitations. First, our primary sample had limited power to detect SNPs of small effect size, thus explaining the lack of genome-wide significant results. Second, the replication sample had an even smaller size and a demographic and clinical profile very different from the primary sample's (as outlined above), which have reduced power to replicate findings. Third, episode count was retrospectively selfreported during the SCAN interview without validation from

P. Ferentinos et al. / Journal of Affective Disorders 155 (2014) 81–89

external sources; it is therefore expected to be subject to recall bias and be influenced by patient's current mood state or personality characteristics (neuroticism, histrionic traits). Finally, family history of bipolar disorder (up to second-degree relatives) was an exclusion criterion in the RADIANT sample; one could reasonably suppose that the contribution of the bipolar polygene to MDD episodicity might be stronger in unselected clinical samples, where bipolar family history is reported by up to 20% of cases (Benazzi, 2003). In conclusion, this study aimed to investigate the genetic underpinnings of a continuous episodicity phenotype in recurrent MDD using three different strategies (family history of MDD, GWAS, polygenic profile analysis). Episodicity (especially intermediate episode counts) was a significant independent index of MDD familial aggregation in our sample, replicating previous reports. No genome-wide significant findings were identified both in the all cases dataset and in a subset of cases with positive family history of depression; we failed to replicate top findings in an independent sample. Finally, our polygenic profile analysis showed that both MDD and bipolar disorder polygenic scores independently contributed to MDD episodicity but the bipolar polygenic component tended to provide the strongest contribution in cases with positive family history of depression. These findings provide preliminary support to the hypothesis that highly recurrent MDD with positive family history of depression is part of a ‘soft bipolar spectrum’ but await replication in larger cohorts and across a wider range of mood disorders.

Role of funding source This work was funded by a joint Grant from the Medical Research Council, UK and GlaxoSmithKline [G0701420], and by financial support from the NIHR Biomedical Research Centre for Mental Health at the South London and Maudsley NHS Foundation Trust and Institute of Psychiatry, King's College London. This work was also supported by the German Federal Ministry of Education and Research within the context of the German National Genome Research Network (NGFN-2 and NGFN-plus). These funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. GlaxoSmithKline funded the collection of the DeNt cohort of depression cases, the genotyping of all RADIANT cases (with the MRC), and both the collection and genotyping of the GSK-Munich cohort of depression cases.

Conflict of interest Farmer and McGuffin have received consultancy fees and honoraria for participating in expert panels for pharmaceutical companies, including GlaxoSmithKline. Muglia and Tozzi were employees of GlaxoSmithKline when the research was performed. Ising has received consultancy honoraria from MSD Merck. Binder has received grant support from PharmaNeuroboost. There are no patents, products in development or marketed products to declare. All other authors (Ferentinos, Rivera, Spain, Cohen-Woods, Butler, Craddock, Owen, Korszun, Jones I, Jones L, Gill, Rice, Maier, Mors, Rietschel, Lucae, Preisig, Breen, Craig, MüllerMyhsok, and Lewis) declare no conflicts of interest.

Acknowledgments We would like to thank all patients who participated in this study.

Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.jad.2013.10.027. References Abecasis, G.R., Altshuler, D., Auton, A., Brooks, L.D., Durbin, R.M., Gibbs, R.A., Hurles, M.E., McVean, G.A., 2010. A map of human genome variation from population-scale sequencing. Nature 467, 1061–1073. Akiskal, H.S., 2003. Validating ‘hard’ and ‘soft’ phenotypes within the bipolar spectrum: continuity or discontinuity? J. Affect. Disord 73, 1–5.

87

Angst, J., Sellaro, R., Stassen, H.H., Gamma, A., 2005. Diagnostic conversion from depression to bipolar disorders: results of a long-term prospective study of hospital admissions. J. Affect. Disord. 84, 149–157. Benazzi, F., 2003. Unipolar depression with bipolar family history: links with the bipolar spectrum. Psychiatry Clin. Neurosci. 57, 497–503. Blom, G., 1958. Statistical Estimates and Transformed Beta Variables. John Wiley and Sons, New York, NY. Cohen-Woods, S., Gaysina, D., Craddock, N., Farmer, A., Gray, J., Gunasinghe, C., Hoda, F., Jones, L., Knight, J., Korszun, A., Owen, M.J., Sterne, A., Craig, I.W., McGuffin, P., 2009. Depression Case Control (DeCC) Study fails to support involvement of the muscarinic acetylcholine receptor M2 (CHRM2) gene in recurrent major depressive disorder. Hum. Mol. Genet. 18, 1504–1509. Coryell, W., Leon, A., Winokur, G., Endicott, J., Keller, M., Akiskal, H., Solomon, D., 1996. Importance of psychotic features to long-term course in major depressive disorder. Am. J. Psychiatry 153, 483–489. Devlin, B., Roeder, K., 1999. Genomic control for association studies. Biometrics 55, 997–1004. Dudbridge, F., Gusnanto, A., 2008. Estimation of significance thresholds for genomewide association scans. Genet. Epidemiol. 32, 227–234. Edwards, A.C., Aliev, F., Bierut, L.J., Bucholz, K.K., Edenberg, H., Hesselbrock, V., Kramer, J., Kuperman, S., Nurnberger Jr., J.I., Schuckit, M.A., Porjesz, B., Dick, D. M., 2012. Genome-wide association study of comorbid depressive syndrome and alcohol dependence. Psychiatr. Genet. 22, 31–41. Farmer, A., Breen, G., Brewster, S., Craddock, N., Gill, M., Korszun, A., Maier, W., Middleton, L., Mors, O., Owen, M., Perry, J., Preisig, M., Rietschel, M., Reich, T., Jones, L., Jones, I., McGuffin, P., 2004. The Depression Network (DeNT) Study: methodology and sociodemographic characteristics of the first 470 affected sibling pairs from a large multi-site linkage genetic study. BMC Psychiatry 4, 42. Ferreira, M.A., O’Donovan, M.C., Meng, Y.A., Jones, I.R., Ruderfer, D.M., Jones, L., Fan, J., Kirov, G., Perlis, R.H., Green, E.K., Smoller, J.W., Grozeva, D., Stone, J., Nikolov, I., Chambert, K., Hamshere, M.L., Nimgaonkar, V.L., Moskvina, V., Thase, M.E., Caesar, S., Sachs, G.S., Franklin, J., Gordon-Smith, K., Ardlie, K.G., Gabriel, S.B., Fraser, C., Blumenstiel, B., Defelice, M., Breen, G., Gill, M., Morris, D.W., Elkin, A., Muir, W.J., McGhee, K.A., Williamson, R., MacIntyre, D.J., MacLean, A.W., St, C.D., Robinson, M., Van Beck, M., Pereira, A.C., Kandaswamy, R., McQuillin, A., Collier, D.A., Bass, N.J., Young, A.H., Lawrence, J., Ferrier, I.N., Anjorin, A., Farmer, A., Curtis, D., Scolnick, E.M., McGuffin, P., Daly, M.J., Corvin, A.P., Holmans, P.A., Blackwood, D.H., Gurling, H.M., Owen, M.J., Purcell, S.M., Sklar, P., Craddock, N., 2008. Collaborative genome-wide association analysis supports a role for ANK3 and CACNA1C in bipolar disorder. Nat. Genet. 40, 1056–1058. Fisfalen, M.E., Schulze, T.G., DePaulo Jr., J.R., DeGroot, L.J., Badner, J.A., McMahon, F.J., 2005. Familial variation in episode frequency in bipolar affective disorder. Am. J. Psychiatry 162, 1266–1272. Foley, D.L., Neale, M.C., Kendler, K.S., 1998. Reliability of a lifetime history of major depression: implications for heritability and co-morbidity. Psychol. Med. 28, 857–870. Gauderman, W., Morrison, J., 2006. QUANTO 1.1: A Computer Program for Power and Sample Size Calculations for Genetic-Epidemiology Studies, 〈http://hydra. usc.edu/gxe〉. Ghaemi, S.N., Ko, J.Y., Goodwin, F.K., 2002. “Cade's disease” and beyond: misdiagnosis, antidepressant use, and a proposed definition for bipolar spectrum disorder. Can. J. Psychiatry 47, 125–134. Green, E.K., Grozeva, D., Jones, I., Jones, L., Kirov, G., Caesar, S., Gordon-Smith, K., Fraser, C., Forty, L., Russell, E., Hamshere, M.L., Moskvina, V., Nikolov, I., Farmer, A., McGuffin, P., Holmans, P.A., Owen, M.J., O’Donovan, M.C., Craddock, N., 2009. The bipolar disorder risk allele at CACNA1C also confers risk of recurrent major depression and of schizophrenia. Mol. Psychiatry 15, 1016–1022. Ho, L.W., Furlong, R.A., Rubinsztein, J.S., Walsh, C., Paykel, E.S., Rubinsztein, D.C., 2000. Genetic associations with clinical characteristics in bipolar affective disorder and recurrent unipolar depressive disorder. Am. J. Med. Genet. 96, 36–42. Hollon, S.D., Shelton, R.C., Wisniewski, S., Warden, D., Biggs, M.M., Friedman, E.S., Husain, M., Kupfer, D.J., Nierenberg, A.A., Petersen, T.J., Shores-Wilson, K., Rush, A.J., 2006. Presenting characteristics of depressed outpatients as a function of recurrence: preliminary findings from the STARnD clinical trial. J. Psychiatr. Res. 40, 59–69. Howie, B.N., Donnelly, P., Marchini, J., 2009. A flexible and accurate genotype imputation method for the next generation of genome-wide association studies. PLoS Genet. 5, e1000529. Huang, J., Perlis, R.H., Lee, P.H., Rush, A.J., Fava, M., Sachs, G.S., Lieberman, J., Hamilton, S.P., Sullivan, P., Sklar, P., Purcell, S., Smoller, J.W., 2010. Crossdisorder genomewide analysis of schizophrenia, bipolar disorder, and depression. Am. J. Psychiatry 167, 1254–1263. Judd, L.L., 1997. The clinical course of unipolar major depressive disorders. Arch. Gen. Psychiatry 54, 989–991. Karlsson, R., Graae, L., Lekman, M., Wang, D., Favis, R., Axelsson, T., Galter, D., Belin, A.C., Paddock, S., 2012. MAGI1 copy number variation in bipolar affective disorder and schizophrenia. Biol. Psychiatry 71, 922–930. Kendler, K.S., Gardner, C.O., Prescott, C.A., 1999. Clinical characteristics of major depression that predict risk of depression in relatives. Arch. Gen. Psychiatry 56, 322–327. Kendler, K.S., Gatz, M., Gardner, C.O., Pedersen, N.L., 2005. Age at onset and familial risk for major depression in a Swedish national twin sample. Psychol. Med. 35, 1573–1579.

88

P. Ferentinos et al. / Journal of Affective Disorders 155 (2014) 81–89

Kendler, K.S., Gatz, M., Gardner, C.O., Pedersen, N.L., 2007. Clinical indices of familial depression in the Swedish Twin Registry. Acta Psychiatr. Scand. 115, 214–220. Kendler, K.S., Neale, M.C., Kessler, R.C., Heath, A.C., Eaves, L.J., 1993. The lifetime history of major depression in women. Reliability of diagnosis and heritability. Arch. Gen. Psychiatry 50, 863–870. Kendler, K.S., Neale, M.C., Kessler, R.C., Heath, A.C., Eaves, L.J., 1994. The clinical characteristics of major depression as indices of the familial risk to illness. Br. J. Psychiatry 165, 66–72. Kessler, R.C., Berglund, P., Demler, O., Jin, R., Koretz, D., Merikangas, K.R., Rush, A.J., Walters, E.E., Wang, P.S., 2003. The epidemiology of major depressive disorder: results from the National Comorbidity Survey Replication (NCS-R). J. Am. Med. Assoc. 289, 3095–3105. Kocsis, J.H., 2006. Recurrent depression: patient characteristics, clinical course, and current recommendations for management. CNS Spectr. 11, 6–11. Kohli, M.A., Lucae, S., Saemann, P.G., Schmidt, M.V., Demirkan, A., Hek, K., Czamara, D., Alexander, M., Salyakina, D., Ripke, S., Hoehn, D., Specht, M., Menke, A., Hennings, J., Heck, A., Wolf, C., Ising, M., Schreiber, S., Czisch, M., Muller, M.B., Uhr, M., Bettecken, T., Becker, A., Schramm, J., Rietschel, M., Maier, W., Bradley, B., Ressler, K.J., Nothen, M.M., Cichon, S., Craig, I.W., Breen, G., Lewis, C.M., Hofman, A., Tiemeier, H., van Duijn, C.M., Holsboer, F., Muller-Myhsok, B., Binder, E.B., 2011. The neuronal transporter gene SLC6A15 confers risk to major depression. Neuron 70, 252–265. Korten, N.C., Comijs, H.C., Lamers, F., Penninx, B.W., 2011. Early and late onset depression in young and middle aged adults: differential symptomatology, characteristics and risk factors? J. Affect. Disord 138, 259–267. Lewis, C.M., Ng, M.Y., Butler, A.W., Cohen-Woods, S., Uher, R., Pirlo, K., Weale, M.E., Schosser, A., Paredes, U.M., Rivera, M., Craddock, N., Owen, M.J., Jones, L., Jones, I., Korszun, A., Aitchison, K.J., Shi, J., Quinn, J.P., Mackenzie, A., Vollenweider, P., Waeber, G., Heath, S., Lathrop, M., Muglia, P., Barnes, M.R., Whittaker, J.C., Tozzi, F., Holsboer, F., Preisig, M., Farmer, A.E., Breen, G., Craig, I.W., McGuffin, P., 2010. Genome-wide association study of major recurrent depression in the U.K. population. Am. J. Psychiatry 167, 949–957. Marchini, J., Howie, B., Myers, S., McVean, G., Donnelly, P., 2007. A new multipoint method for genome-wide association studies by imputation of genotypes. Nat. Genet. 39, 906–913. McGuffin, P., Katz, R., Watkins, S., Rutherford, J., 1996. A hospital-based twin register of the heritability of DSM-IV unipolar depression. Arch. Gen. Psychiatry 53, 129–136. McGuffin, P., Knight, J., Breen, G., Brewster, S., Boyd, P.R., Craddock, N., Gill, M., Korszun, A., Maier, W., Middleton, L., Mors, O., Owen, M.J., Perry, J., Preisig, M., Reich, T., Rice, J., Rietschel, M., Jones, L., Sham, P., Farmer, A.E., 2005. Whole genome linkage scan of recurrent depressive disorder from the depression network study. Hum. Mol. Genet. 14, 3337–3345. McGuffin, P., Rijsdijk, F., Andrew, M., Sham, P., Katz, R., Cardno, A., 2003. The heritability of bipolar affective disorder and the genetic relationship to unipolar depression. Arch. Gen. Psychiatry 60, 497–502. McMahon, F.J., Akula, N., Schulze, T.G., Muglia, P., Tozzi, F., Detera-Wadleigh, S.D., Steele, C.J., Breuer, R., Strohmaier, J., Wendland, J.R., Mattheisen, M., Muhleisen, T.W., Maier, W., Nothen, M.M., Cichon, S., Farmer, A., Vincent, J.B., Holsboer, F., Preisig, M., Rietschel, M., 2010. Meta-analysis of genome-wide association data identifies a risk locus for major mood disorders on 3p21.1. Nat. Genet. 42, 128–131. Merikangas, K., Yu, K., 2002. Genetic epidemiology of bipolar disorder. Clin. Neurosci. Res. 2, 127–141. Milne, B.J., Caspi, A., Harrington, H., Poulton, R., Rutter, M., Moffitt, T.E., 2009. Predictive value of family history on severity of illness: the case for depression, anxiety, alcohol dependence, and drug dependence. Arch. Gen. Psychiatry 66, 738–747. Mitchell, P.B., Goodwin, G.M., Johnson, G.F., Hirschfeld, R.M., 2008. Diagnostic guidelines for bipolar depression: a probabilistic approach. Bipolar Disord. 10, 144–152. Muglia, P., Tozzi, F., Galwey, N.W., Francks, C., Upmanyu, R., Kong, X.Q., Antoniades, A., Domenici, E., Perry, J., Rothen, S., Vandeleur, C.L., Mooser, V., Waeber, G., Vollenweider, P., Preisig, M., Lucae, S., Muller-Myhsok, B., Holsboer, F., Middleton, L.T., Roses, A.D., 2010. Genome-wide association study of recurrent major depressive disorder in two European case-control cohorts. Mol. Psychiatry 15, 589–601. Nierenberg, A.A., Trivedi, M.H., Fava, M., Biggs, M.M., Shores-Wilson, K., Wisniewski, S.R., Balasubramani, G.K., Rush, A.J., 2007. Family history of mood disorder and characteristics of major depressive disorder: a STARnD (sequenced treatment alternatives to relieve depression) study. J. Psychiatr. Res. 41, 214–221. Overholser, J.C., Miller, I.W., Norman, W.H., 1987. The course of depressive symptoms in suicidal vs. nonsuicidal depressed inpatients. J. Nerv. Ment. Dis. 175, 450–456. Patten, S.B., Williams, J.V., Lavorato, D.H., Bulloch, A.G., D’Arcy, C., Streiner, D.L., 2012. Recall of recent and more remote depressive episodes in a prospective cohort study. Soc. Psychiatry Psychiatr. Epidemiol. 47, 691–696. PGC (Psychiatric GWAS Consortium Bipolar Disorder Working Group), 2011. Largescale genome-wide association analysis of bipolar disorder identifies a new susceptibility locus near ODZ4. Nat. Genet. 43, 977–983. PGC (Major Depressive Disorder Working Group of the Psychiatric GWAS Consortium), 2013. A mega-analysis of genome-wide association studies for major depressive disorder. Mol. Psychiatry 18, 497–511. Phelps, J., Angst, J., Katzow, J., Sadler, J., 2008. Validity and utility of bipolar spectrum models. Bipolar Disord. 10, 179–193.

Power, R.A., Keers, R., Ng, M.Y., Butler, A.W., Uher, R., Cohen-Woods, S., Ising, M., Craddock, N., Owen, M.J., Korszun, A., Jones, L., Jones, I., Gill, M., Rice, J.P., Hauser, J., Henigsberg, N., Maier, W., Zobel, A., Mors, O., Placentino, A.S., Rietschel, M., Souery, D., Kozel, D., Preisig, M., Lucae, S., Binder, E.B., Aitchison, K.J., Tozzi, F., Muglia, P., Breen, G., Craig, I.W., Farmer, A.E., Muller-Myhsok, B., McGuffin, P., Lewis, C.M., 2012. Dissecting the genetic heterogeneity of depression through age at onset. Am. J. Med. Genet. B Neuropsychiatr. Genet. 159B, 859–868. Price, A.L., Patterson, N.J., Plenge, R.M., Weinblatt, M.E., Shadick, N.A., Reich, D., 2006. Principal components analysis corrects for stratification in genome-wide association studies. Nat. Genet. 38, 904–909. Purcell, S., Neale, B., Todd-Brown, K., Thomas, L., Ferreira, M.A., Bender, D., Maller, J., Sklar, P., de Bakker, P.I., Daly, M.J., Sham, P.C., 2007. PLINK: a tool set for wholegenome association and population-based linkage analyses. Am. J. Hum. Genet. 81, 559–575. Purcell, S.M., Wray, N.R., Stone, J.L., Visscher, P.M., O’Donovan, M.C., Sullivan, P.F., Sklar, P., 2009. Common polygenic variation contributes to risk of schizophrenia and bipolar disorder. Nature 460, 748–752. Rietschel, M., Mattheisen, M., Frank, J., Treutlein, J., Degenhardt, F., Breuer, R., Steffens, M., Mier, D., Esslinger, C., Walter, H., Kirsch, P., Erk, S., Schnell, K., Herms, S., Wichmann, H.E., Schreiber, S., Jockel, K.H., Strohmaier, J., Roeske, D., Haenisch, B., Gross, M., Hoefels, S., Lucae, S., Binder, E.B., Wienker, T.F., Schulze, T.G., Schmal, C., Zimmer, A., Juraeva, D., Brors, B., Bettecken, T., Meyer-Lindenberg, A., MullerMyhsok, B., Maier, W., Nothen, M.M., Cichon, S., 2010. Genome-wide association-, replication-, and neuroimaging study implicates HOMER1 in the etiology of major depression. Biol. Psychiatry 68, 578–585. Roy-Byrne, P.P., Stang, P., Wittchen, H.U., Ustun, B., Walters, E.E., Kessler, R.C., 2000. Lifetime panic-depression comorbidity in the National Comorbidity Survey: association with symptoms, impairment, course and help-seeking. Br. J. Psychiatry 176, 229–235. Schosser, A., Butler, A.W., Ising, M., Perroud, N., Uher, R., Ng, M.Y., Cohen-Woods, S., Craddock, N., Owen, M.J., Korszun, A., Jones, L., Jones, I., Gill, M., Rice, J.P., Maier, W., Mors, O., Rietschel, M., Lucae, S., Binder, E.B., Preisig, M., Perry, J., Tozzi, F., Muglia, P., Aitchison, K.J., Breen, G., Craig, I.W., Farmer, A.E., Muller-Myhsok, B., McGuffin, P., Lewis, C.M., 2011. Genomewide association scan of suicidal thoughts and behaviour in major depression. PLoS One 6, e20690. Schulze, T.G., Akula, N., Breuer, R., Steele, J., Nalls, M.A., Singleton, A.B., Degenhardt, F.A., Nothen, M.M., Cichon, S., Rietschel, M., The bipolar genome study and McMahon, F. J., 2013. molecular genetic overlap in bipolar disorder, schizophrenia, and major depressive disorder. World J. Biol. Psychiatry, http://dx.doi.org/10.3109/ 15622975.2012, 662282, (in press). Serretti, A., Mandelli, L., 2008. The genetics of bipolar disorder: genome ‘hot regions,’ genes, new potential candidates and future directions. Mol. Psychiatry 13, 742–771. Shi, J., Potash, J.B., Knowles, J.A., Weissman, M.M., Coryell, W., Scheftner, W.A., Lawson, W.B., DePaulo Jr., J.R., Gejman, P.V., Sanders, A.R., Johnson, J.K., Adams, P., Chaudhury, S., Jancic, D., Evgrafov, O., Zvinyatskovskiy, A., Ertman, N., Gladis, M., Neimanas, K., Goodell, M., Hale, N., Ney, N., Verma, R., Mirel, D., Holmans, P., Levinson, D.F., 2011. Genome-wide association study of recurrent early-onset major depressive disorder. Mol. Psychiatry 16, 193–201. Shyn, S.I., Shi, J., Kraft, J.B., Potash, J.B., Knowles, J.A., Weissman, M.M., Garriock, H.A., Yokoyama, J.S., McGrath, P.J., Peters, E.J., Scheftner, W.A., Coryell, W., Lawson, W.B., Jancic, D., Gejman, P.V., Sanders, A.R., Holmans, P., Slager, S.L., Levinson, D.F., Hamilton, S.P., 2011. Novel loci for major depression identified by genome-wide association study of Sequenced Treatment Alternatives to Relieve Depression and meta-analysis of three studies. Mol. Psychiatry 16, 202–215. Smith, D.J., Kyle, S., Forty, L., Cooper, C., Walters, J., Russell, E., Caesar, S., Farmer, A., McGuffin, P., Jones, I., Jones, L., Craddock, N., 2008. Differences in depressive symptom profile between males and females. J. Affect. Disord. 108, 279–284. Smoller, J.W., Craddock, N., Kendler, K., Lee, P.H., Neale, B.M., Nurnberger, J.I., Ripke, S., Santangelo, S., Sullivan, P.F., 2013. Identification of risk loci with shared effects on five major psychiatric disorders: a genome-wide analysis. Lancet 381, 1371–1379. Soboloff, J., Rothberg, B.S., Madesh, M., Gill, D.L., 2012. STIM proteins: dynamic calcium signal transducers. Nat. Rev. Mol. Cell Biol. 13, 549–565. Solomon, D.A., Keller, M.B., Leon, A.C., Mueller, T.I., Lavori, P.W., Shea, M.T., Coryell, W., Warshaw, M., Turvey, C., Maser, J.D., Endicott, J., 2000. Multiple recurrences of major depressive disorder. Am. J. Psychiatry 157, 229–233. Souery, D., Zaninotto, L., Calati, R., Linotte, S., Mendlewicz, J., Sentissi, O., Serretti, A., 2012. Depression across mood disorders: review and analysis in a clinical sample. Compr. Psychiatry 53, 24–38. Sullivan, P.F., de Geus, E.J., Willemsen, G., James, M.R., Smit, J.H., Zandbelt, T., Arolt, V., Baune, B.T., Blackwood, D., Cichon, S., Coventry, W.L., Domschke, K., Farmer, A., Fava, M., Gordon, S.D., He, Q., Heath, A.C., Heutink, P., Holsboer, F., Hoogendijk, W.J., Hottenga, J.J., Hu, Y., Kohli, M., Lin, D., Lucae, S., Macintyre, D.J., Maier, W., McGhee, K.A., McGuffin, P., Montgomery, G.W., Muir, W.J., Nolen, W.A., Nothen, M.M., Perlis, R.H., Pirlo, K., Posthuma, D., Rietschel, M., Rizzu, P., Schosser, A., Smit, A.B., Smoller, J.W., Tzeng, J.Y., van Dyck, R., Verhage, M., Zitman, F.G., Martin, N.G., Wray, N.R., Boomsma, D.I., Penninx, B.W., 2009. Genome-wide association for major depressive disorder: a possible role for the presynaptic protein piccolo. Mol. Psychiatry 14, 359–375. Sullivan, P.F., Neale, M.C., Kendler, K.S., 2000. Genetic epidemiology of major depression: review and meta-analysis. Am. J. Psychiatry 157, 1552–1562. Tozzi, F., Prokopenko, I., Perry, J.D., Kennedy, J.L., McCarthy, A.D., Holsboer, F., Berrettini, W., Middleton, L.T., Chilcoat, H.D., Muglia, P., 2008. Family history of depression is associated with younger age of onset in patients with recurrent depression. Psychol. Med. 38, 641–649.

P. Ferentinos et al. / Journal of Affective Disorders 155 (2014) 81–89

Vandeleur, C., Rothen, S., Gholam-Rezaee, M., Castelao, E., Vidal, S., Favre, S., Ferrero, F., Halfon, O., Fumeaux, P., Merikangas, K.R., Aubry, J.M., Burstein, M., Preisig, M., 2012. Mental disorders in offspring of parents with bipolar and major depressive disorders. Bipolar Disord. 14, 641–653. Wardenaar, K.J., Giltay, E.J., van Veen, T., Zitman, F.G., Penninx, B.W., 2012. Symptom dimensions as predictors of the two-year course of depressive and anxiety disorders. J. Affect. Disord. 136, 1198–1203. Wing, J.K., Babor, T., Brugha, T., Burke, J., Cooper, J.E., Giel, R., Jablenski, A., Regier, D., Sartorius, N., 1990. SCAN: Schedules for Clinical Assessment in Neuropsychiatry. Arch. Gen. Psychiatry 47, 589–593. Wray, N.R., Pergadia, M.L., Blackwood, D.H., Penninx, B.W., Gordon, S.D., Nyholt, D.R., Ripke, S., MacIntyre, D.J., McGhee, K.A., Maclean, A.W., Smit, J.H., Hottenga, J.J.,

89

Willemsen, E.J., Middeldorp, C.M., de Geus, E.J., Lewis, C.M., McGuffin, P., Hickie, I. B., van den Oord, E.J., Liu, J.Z., Macgregor, S., McEvoy, B.P., Byrne, E.M., Medland, S. E., Statham, D.J., Henders, A.K., Heath, A.C., Montgomery, G.W., Martin, N.G., Boomsma, D.I., Madden, P.A., Sullivan, P.F., 2012. Genome-wide association study of major depressive disorder: new results, meta-analysis, and lessons learned. Mol. Psychiatry 17, 36–48. Zisook, S., Lesser, I., Stewart, J.W., Wisniewski, S.R., Balasubramani, G.K., Fava, M., Gilmer, W.S., Dresselhaus, T.R., Thase, M.E., Nierenberg, A.A., Trivedi, M.H., Rush, A.J., 2007. Effect of age at onset on the course of major depressive disorder. Am. J. Psychiatry 164, 1539–1546.