Psychological Medicine Depression with atypical neurovegetative

cambridge.org/psm
symptoms shares genetic predisposition
with immuno-metabolic traits and
alcohol consumption
Original Article
Cite this article: Badini I, Coleman JRI, Isabella Badini1, Jonathan R.I. Coleman1,2, Saskia P. Hagenaars1,
Hagenaars SP, Hotopf M, Breen G, Lewis CM, Matthew Hotopf2,3,4, Gerome Breen1,2, Cathryn M. Lewis1,2,5
Fabbri C (2020). Depression with atypical
neurovegetative symptoms shares genetic and Chiara Fabbri1,6
predisposition with immuno-metabolic traits
and alcohol consumption. Psychological 1
Institute of Psychiatry, Psychology and Neuroscience, King’s College London, London, UK; 2National Institute for
Medicine 1–11. https://doi.org/10.1017/
Health Research (NIHR) Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and
S0033291720002342
King’s College London, London, UK; 3Department of Psychological Medicine, Institute of Psychiatry, Psychology
Received: 19 February 2020 and Neuroscience, King’s College London, London, UK; 4South London and Maudsley NHS Foundation Trust,
Revised: 3 June 2020 London, UK; 5Department of Medical and Molecular Genetics, King’s College London, London, UK and
6
Accepted: 12 June 2020 Department of Biomedical and Neuromotor Sciences, University of Bologna, Bologna, Italy

Key words:
Alcohol use; atypical depression; BMI; cardio- Abstract
metabolic diseases; CRP; polygenic risk scores Background. Depression is a highly prevalent and heterogeneous disorder. This study aims to
Author for correspondence:
determine whether depression with atypical features shows different heritability and different
Chiara Fabbri, degree of overlap with polygenic risk for psychiatric and immuno-metabolic traits than other
E-mail: [email protected], chiara.fabbri@ depression subgroups.
yahoo.it Methods. Data included 30 069 European ancestry individuals from the UK Biobank who met
criteria for lifetime major depression. Participants reporting both weight gain and hypersom-
nia were classified as ↑WS depression (N = 1854) and the others as non-↑WS depression
(N = 28 215). Cases with non-↑WS depression were further classified as ↓WS depression
(i.e. weight loss and insomnia; N = 10 142). Polygenic risk scores (PRS) for 22 traits were gen-
erated using genome-wide summary statistics (Bonferroni corrected p = 2.1 × 10−4). Single-
nucleotide polymorphism (SNP)-based heritability of depression subgroups was estimated.
Results. ↑WS depression had a higher polygenic risk for BMI [OR = 1.20 (1.15–1.26),
p = 2.37 × 10−14] and C-reactive protein [OR = 1.11 (1.06–1.17), p = 8.86 × 10−06] v. non-
↑WS depression and ↓WS depression. Leptin PRS was close to the significance threshold
( p = 2.99 × 10−04), but the effect disappeared when considering GWAS summary statistics
of leptin adjusted for BMI. PRS for daily alcohol use was inversely associated with ↑WS
depression [OR = 0.88 (0.83–0.93), p = 1.04 × 10−05] v. non-↑WS depression. SNP-based
heritability was not significantly different between ↑WS depression and ↓WS depression
(14.3% and 12.2%, respectively).
Conclusions. ↑WS depression shows evidence of distinct genetic predisposition to immune-
metabolic traits and alcohol consumption. These genetic signals suggest that biological targets
including immune-cardio-metabolic pathways may be relevant to therapies in individuals with
↑WS depression.

Introduction
Depressive disorders are highly prevalent and a leading cause of global disability and are asso-
ciated with premature mortality (James et al., 2018; Kessler & Bromet, 2013). Twin-based her-
itability of unipolar depression is estimated to be ∼37% (Sullivan, Neale, & Kendler, 2000),
with common single-nucleotide polymorphisms (SNPs) explaining ∼9% of variation in
depression liability (Howard et al., 2019; Wray et al., 2018). Efforts to identify genetic variants
associated with the disease are hindered by the heterogeneity among depressed cases, who can
vary greatly in symptom presentation and severity, clinical course and treatment response
(Fried & Nesse, 2015). Clinical heterogeneity may also reflect different underlying biological
and causal pathways. Increasing evidence suggests that depressive symptoms and subtypes
are differentially associated with genetic risk factors that overlap with other disorders
© The Author(s), 2020. Published by (Beijers, Wardenaar, van Loo, & Schoevers, 2019; Milaneschi et al., 2016, 2017).
Cambridge University Press Investigating the clinical and genetic correlates of more homogenous subtypes may improve
the understanding of specific aetiological mechanisms and the development of potential treat-
ment targets. Similarly, identifying whether genetic risk for other disorders overlaps with cer-
tain characteristics of depression may help to elucidate biological mechanisms underlying
common symptom presentations across multiple disorders.

Downloaded from https://www.cambridge.org/core. Auckland University of Technology, on 14 Jul 2020 at 15:14:46, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291720002342
 2 Isabella Badini et al.

The atypical subtype of major depression is specified in the substance use and other psychiatric traits compared to depression
DSM-5 by the presence of at least two of the following symptoms: without ↑WS.
hypersomnia, increased appetite and/or weight gain, leaden par-
alysis, and interpersonal rejection sensitivity (American
Psychiatric Association, 2013). Reversed neurovegetative symp- Materials and methods
toms (hypersomnia, increased appetite and/or weight gain), in Sample
particular, have been found to be highly specific and predictive
of clinically defined atypical MDD, as they identify patients Individuals who met lifetime criteria for major depression were
with similar sociodemographic and clinical correlates as those drawn from the UKB. UKB is a prospective population-based
reported for classification based on the full DSM criteria study of ∼500 000 individuals recruited across the UK, aged
(Benazzi, 2002). Classification of atypical depression based on between 40 and 69 at baseline (UK Biobank, 2019). A total of
reversed neurovegetative symptoms alone is more feasible in epi- 157 387 participants completed an online Mental Health
demiological studies, many of which have adopted this criterion Questionnaire (MHQ) assessing self-reported psychiatric symp-
instead of the full DSM-5 criteria (Lee, Ng, & Tsang, 2009; toms corresponding to clinical diagnostic criteria and self-
Matza, Revicki, Davidson, & Stewart, 2003). reported professional diagnoses (Davis et al., 2020).
Epidemiological studies have shown differences between atyp- Genome-wide genetic data have been collected on all UKB parti-
ical and non-atypical depression in sociodemographic factors, cipants (Bycroft et al., 2018), as detailed below. All participants
clinical features, lifestyle factors and comorbidities. Atypical provided written informed consent and all procedures contribut-
depression has been associated with earlier age of onset, female ing to this work comply with the ethical standards of the relevant
gender, more severe and recurrent depressive episodes (Agosti national and institutional committees on human experimentation
& Stewart, 2001; Blanco et al., 2012; Brailean, Curtis, Davis, and with the Helsinki Declaration of 1975, as revised in 2008.
Dregan, & Hotopf, 2020). Findings from UK Biobank (UKB)
showed higher rates of smoking, social isolation, loneliness, Measures
greater exposure to adverse life events and lower rates of moderate
physical activity among atypical cases compared to non-atypical Lifetime psychiatric diagnoses were assessed in the MHQ using
cases (Brailean et al., 2020). Atypical depression has also been the Composite International Diagnostic Interview Short Form
associated with higher rates of bipolar disorder and psychiatric (CIDI-SF) (Kessler, Andrews, Mroczek, Ustun, & Wittchen,
comorbidity such as anxiety disorders, binge eating disorder 1998). Criteria for lifetime major depressive episode were in
and substance abuse (Agosti & Stewart, 2001; Blanco et al., accordance with DSM-V. The full CIDI is a validated measure
2012; Brailean et al., 2020; Lee et al., 2009; Łojko, Buzuk, of depression, demonstrated to have good concordance with dir-
Owecki, Ruchała, & Rybakowski, 2015). Physical health ect clinical assessment (Haro et al., 2006).
comorbidities more strongly associated with atypical cases include Cases reporting both weight gain (data field 20 536) and
higher body mass index (BMI), inflammation, metabolic syn- hypersomnia (data field 20 534) were classified as ↑WS depression
drome and cardiovascular disease (CAD) (Brailean et al., 2020; (N = 1854), and the remaining cases classified as depression with-
Lasserre et al., 2014; Milaneschi et al., 2017). Specifically, evidence out ↑WS (N = 28 215). From this group, depression with both
suggests stronger links between atypical features of increased weight loss and decreased sleep (data fields 20 533 and 20 535)
appetite and/or weight and immuno-metabolic dysregulations, was defined as ↓WS depression (N = 10 142). These definitions
such as BMI, C-reactive protein (CRP) and leptin (Milaneschi used the same coding conventions as those used in a previous
et al., 2017). study in the UKB (Brailean et al., 2020); however, we preferred
There is increasing evidence to suggest partially distinct gen- to avoid the terms atypical and typical depression to avoid confu-
etic profiles among depressive subtypes. SNP-based heritability sion with the standard nosological classification. We did not con-
(h 2SNP) was found to vary across individual depressive symptoms sider variations in appetite to distinguish the subgroups, since the
(h 2SNP range from 6% to 9%), and patterns of SNP associations available measure (data field 20 511) did not differentiate between
and genetic correlations differed across symptoms (Thorp et al., hypophagia and hyperphagia.
2019). A study that classified cases into subtypes according to
change in neurovegetative symptoms [i.e. no change, increased
Genotyping and quality control
or decreased appetite and/or weight (A/W)] found similar
SNP-heritability of 10–11% in all groups (Milaneschi et al., Genetic data came from the full release of the UKB data (N = 488
2017). Polygenic risk scoring analyses confirmed that the 377; Bycroft et al., 2018). Genotyping was performed using two
increased A/W subtype had higher polygenic risk for CRP, highly-overlapping arrays covering ∼800 000 markers (UK
BMI, leptin and triglycerides levels than the decreased A/W sub- Biobank Axiom Array Content Summary). Autosomal genotype
type (Milaneschi et al., 2016, 2017). Overlapping genetic aetiology data underwent centralised quality control to adjust for possible
between depression and CAD was demonstrated (Hagenaars et al., array effects, batch effects, plate effects and departures from
2019), but no studies investigated if depression with atypical fea- Hardy–Weinberg equilibrium (HWE; Bycroft et al., 2018).
tures may have a greater genetic overlap with CAD compared to Variants for this analysis were limited to common variants
other depression subtypes. (minor allele frequency >0.01) that were directly genotyped.
In the present study, we examined the genetic overlap between SNPs were further excluded based on missingness (>0.02) and
depression with atypical features (increased weight and sleepiness: on HWE ( p < 10–8). Individuals were removed for high levels
↑WS) and a range of traits and disorders using polygenic risk scores of missingness (>0.05) or abnormal heterozygosity (as defined
(PRS). Based on previous findings, we hypothesised that ↑WS during centralised quality control), relatedness of up to third-
depression would show similar heritability to depression without degree kinship (KING r < 0.044; Manichaikul et al., 2010), or
↑WS and higher polygenic risk for immune-cardio-metabolic, phenotypic and genotypic gender discordance (phenotypic

Downloaded from https://www.cambridge.org/core. Auckland University of Technology, on 14 Jul 2020 at 15:14:46, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291720002342
 Psychological Medicine 3

males with FX < 0.9, phenotypic females with FX > 0.6). Population relationship matrix-restricted maximum likelihood (GREML)
structure within the UKB cohort was assessed using principal com- methods (Lee, Wray, Goddard, & Visscher, 2011) were used to
ponent analysis, with European ancestry defined by four-means estimate the variance in liability attributable to the additive effects
clustering on the first two genetic principal components (Warren of all SNPs (h 2SNP) for ↑WS depression, ↓WS depression and
et al., 2017). Among respondents to the MHQ, 95% were of depression not falling in these two groups. A random subset of
European ancestry and therefore individuals from other ancestries 8000 healthy controls were selected for the analyses among
were excluded from further analyses to maximise statistical power. those who completed the MHQ. This number of controls was
After quality control, the final sample of respondents to the MHQ selected because it provided adequate power (at least 80%) to esti-
consisted of 126 522 individuals with genotype data. mate heritability considering an expected heritability of 0.10
(Milaneschi et al., 2017), using GCTA-GREML Power
Calculator (Visscher et al., 2014). The genetic relationship matrix
Statistical analysis (GRM) was adjusted for incomplete tagging of causal SNPs and
Polygenic risk scores we excluded related individuals using a GRM-cut-off of 0.05.
PRS were calculated based on GWAS summary statistics for 22 traits We calculated the genetic correlation between depression sub-
reflecting the hypothesis formulated in the Introduction, including groups using bivariate GREML with independent subsets of
major psychiatric disorders, personality traits, substance use-related 8000 healthy controls for each depression subgroup.
traits, cardio-metabolic traits and CRP (online Supplementary We also calculated h 2SNP using GCTB (Genome-wide Complex
Table S1). There was no overlap between the samples included in Trait Bayesian analysis) Bayes S method, which estimates polyge-
these GWASs and the sample included in this study, except a nicity from the data (i.e. the proportion of SNPs with non-zero
very marginal overlap with the GWAS of anorexia nervosa [349/ effects, π). GCTB also calculates the relationship between effect
16 992 (2%) of cases included in Watson et al. (2019)]. size and MAF (S) which can be used to detect signatures of nat-
PRS were calculated using PRSice v.2 (Choi & O’Reilly, 2019; ural selection (Zeng et al., 2018). We used the standard settings of
Euesden, Lewis, & O’Reilly, 2015). PRSice computes scores in an 21 000 simulations with the first 1000 as burn-in and standard
independent (target) sample by calculating the weighted sum of initial S and π values.
trait-associated alleles using summary data from GWAS discovery GCTA-GREML and GCTB Bayes S analyses were adjusted for
samples. SNPs in linkage disequilibrium [r 2 ⩾ 0.1 (250-kb win- the same covariates included in the PRS analysis, but we also
dow)] were removed using the clumping procedure. We used adjusted for BMI because there was a significant difference in
the default average option that calculates the ratio between the BMI between cases with ↑WS depression (30.55 ± 5.78), ↓WS
PRS and the number of alleles included in each individual and depression (25.93 ± 4.41) and healthy controls (26.45 ± 4.14)
PRS were standardised (mean = 0, S.D. = 1). PRS were calculated ( p = 1.51 × 10−221 and p = 1.25 × 10−154, respectively); heritability
at 11 p value thresholds PT (5 × 10−8, 1 × 10−5, 1 × 10−3, 0.01, estimates would be confounded by this variable which was
0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 1) and the most predictive PT was demonstrated to have a heritability of about 28% (Zeng et al.,
selected. Logistic regression models were used to estimate associa- 2018).
tions between ↑WS depression v. non-↑WS depression and each Possible differences among h 2SNP of depression subgroups were
PRS adjusting for covariates of six genetic ancestry principal com- compared by a Z-test as explained for PRS results. h 2SNP estimates
ponents, centre and batch effects. We estimated the proportion of were transformed to a liability scale (Lee et al., 2012) and we
variance explained by PRS on the observed and liability scale (Lee, reported heritability considering a range of plausible prevalence
Goddard, Wray, & Visscher, 2012), considering a range of pos- values (Levitan et al., 1997; Lim et al., 2018; Łojko &
sible values of prevalence among depressed cases (Levitan, Rybakowski, 2017). The code used for these analyses is available
Lesage, Parikh, Goering, & Kennedy, 1997; Łojko & as online Supplementary material (code_used_for_analyses).
Rybakowski, 2017). For PRS of traits associated with ↑WS depres-
sion, we calculated: (1) the OR of ↑WS depression v. depression Sensitivity analyses
without ↑WS for each decile of the PRS, taking the first decile Analyses were repeated (1) excluding cases with probable bipolar
as reference and (2) if the effect was comparable when consider- disorder or missing information for this variable [wider bipolar
ing each of the symptoms separately (↑weight and ↑ sleep v. no disorder definition, as described in Brailean et al. (2020), n =
increase in weight or sleep). We estimated differences between 1747] and schizophrenia or missing information for this variable
PRS results of different comparisons by comparing their estimates (n = 61); (2) comparing ↑WS depression with ↓WS depression
(E1 and E2) and SE (SE1 
 and SE2) using a Z-test (rather than non-↑WS depression); and (3) comparing both



















Z = (E1 − E2)/ (SE1)2 + (SE2)2 . The code used for these ↑WS depression and ↓WS depression with healthy controls who
completed the MHQ (n = 64 604) (Fig. 1).
analyses is available as online Supplementary material
(code_used_for_analyses).
A Bonferroni correction was applied to account for multiple Results
testing, providing a required significance level of p < 2.1 × 10−4
[the PRS of 17 traits were analysed at 11 PT, while for five traits Sample characteristics
there were no SNPs with p < 5 × 10−8 and 10 thresholds were tested The total sample comprised of 30 069 participants who met cri-
(0.05/(11 × 17 + 10 × 5) = 2.1 × 10−4)]. The use of Bonferroni cor- teria for lifetime major depressive episode, including 1854 cases
rection is conservative, since the different PT are highly correlated. classified as ↑WS depression and 28 215 classified as depression
without ↑WS. Non-↑WS cases were further classified as ↓WS
SNP-based heritability depression (N = 10 142). The ↑WS group was 75% female and
Using genome-wide complex trait analysis software v.1.93.1beta had a mean age of 59.99 (S.D. = 7.09) years, compared to the
(GCTA) (Yang, Lee, Goddard, & Visscher, 2011), genetic non-↑WS group which was 68% female and had a mean age of

Downloaded from https://www.cambridge.org/core. Auckland University of Technology, on 14 Jul 2020 at 15:14:46, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291720002342
 4 Isabella Badini et al.

Fig. 1. Flow chart of the performed analyses. ↑WS,

depression with increased weight and sleepiness; ↓WS
depression, depression with decreased weight and
insomnia.

Table 1. Polygenic risk scores associated with depression with atypical features (↑WS depression) compared with depression without atypical features

Polygenic risk score Reference Sample size PT p OR (95% CI) R2

BMI (Locke et al., 2015) 322 154 0.001 2.37 × 10−14* 1.20 (1.15–1.26) 0.00520
−04
Leptin (Kilpeläinen et al., 2016) 31 816 0.001 2.99 × 10 1.09 (1.04–1.14) 0.00117
−06
CRP (Ligthart et al., 2018) 204 402 0.4 8.86 × 10 * 1.11 (1.06–1.17) 0.00177
Daily alcohol use (Schumann et al., 2016) 70 460 0.4 1.04 × 10−05* 0.88 (0.83–0.93) 0.00174
−04
Major depressive disorder (Wray et al., 2018) 143 265 0.3 1.14 × 10 * 1.10 (1.05–1.15) 0.00133
Results are shown for the PT threshold attaining the lowest p value.
R 2 estimated on the observed scale as the results are referred to case-only comparisons.
*Significant associations ( p < 2.1 × 10−4).

62.50 (7.53) years. The ↓WS depression group was 75% female inversely associated with ↓WS depression v. healthy controls
and had a mean age of 62.55 (7.49). Further description of the (p = 1.24 × 10−04) while leptin, CRP and alcohol daily use PRS
clinical-demographic features of the subtypes can be found in a had no effect (online Supplementary Table S5).
previous paper in UKB that used the same classification criteria The association between ↑WS v. depression without ↑WS and
(Brailean et al., 2020). leptin PRS was close to the significance threshold [OR = 1.09
(1.04–1.14), p = 2.99 × 10−04] but disappeared when considering
GWAS summary statistics of leptin adjusted for BMI. Nominal
Polygenic risk analysis significant associations ( p < 0.05) included PRS for type 2 dia-
Traits where the PRS were significantly associated with ↑WS betes, coronary artery disease, triglycerides, total HDL cholesterol
depression compared to non-↑WS depression included BMI and ischaemic stroke (Fig. 3; online Supplementary Table S2). The
[OR = 1.20 (1.15–1.26), p = 2.37 × 10−14], CRP [OR = 1.11 (1.06– PRS for coronary artery disease reached statistical significance for
1.17), p = 8.86 × 10−06], daily alcohol use [OR = 0.88 (0.83–0.93), association with both ↑WS depression and ↓WS depression when
p = 1.04 × 10−05] and MDD [OR = 1.10 (1.05–1.15), p = 1.14 × the comparator group was healthy controls, in addition to trigly-
10−04], as shown in Table 1 (see online Supplementary Table S2 cerides PRS only for ↑WS depression v. controls. Case–healthy
for all results). The OR for ↑WS depression was 2.03 (1.62– controls analyses showed expected associations with the PRS of
2.55), 1.36 (1.10–1.68), 0.62 (0.49–0.79) and 1.30 (1.04–1.61) in psychiatric traits and contributed to clarifying the effect of alcohol
the 10th PRS decile v. the 1st PRS decile for BMI, CRP, daily alco- use-related traits PRS: higher PRS for alcohol dependence but not
hol use and MDD, respectively (Fig. 2). Results were consistent daily alcohol use increased the risk of ↓WS depression, while ↑WS
between different PT (online Supplementary Fig. S1). The same depression was still associated with lower daily alcohol use PRS
direction of effect was observed when comparing cases with (online Supplementary Tables S4–S5, Fig. 4). Sensitivity analysis
weight gain v. cases without weight gain, with similar effect size revealed very similar results when excluding cases with probable
for alcohol daily use, MDD, leptin and CRP PRS, while larger bipolar disorder and schizophrenia or missing information for
effect size for BMI PRS (z = 2.12, p = 0.03). These PRS showed these variables (n = 196 and n = 12 among patients with ↑WS
the same direction of effect though not significant when consid- depression; n = 1551 and n = 49 among those with non-↑WS
ering depression with hypersomnia compared to depression with- depression; online Supplementary Figs S2–S4).
out this symptom (online Supplementary Table S2C). The results
were consistent when comparing ↑WS depression with ↓WS
SNP-based heritability
depression and ↑WS depression with healthy controls (online
Supplementary Tables S3 and S4). Interestingly, the effect of SNP-based heritability (h 2SNP) on the liability scale for ↑WS
BMI PRS was very close to the significance threshold for being depression was estimated to be 0.14 (95% CI 0.03–0.25) using

Downloaded from https://www.cambridge.org/core. Auckland University of Technology, on 14 Jul 2020 at 15:14:46, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291720002342
 Psychological Medicine 5

Fig. 2. Decile plots showing the odds ratio of ↑WS depression

(depression with atypical features) v. depression without ↑WS
for BMI PRS (a), C-reactive protein (CRP) PRS (b), daily alcohol
use PRS (c) and major depressive disorder (MDD) PRS (d ). *p
< 0.05; **p < 1 × 10−05.

GCTB-Bayes S and 0.21 (95% CI 0.07–0.35) using gain (z = 2.36, p = 0.018), which however did not survive
GCTA-GREML (Table 2). h 2SNP was 0.12 (95% CI 0.08–0.16) Bonferroni correction for five tests (online Supplementary
and 0.16 (95% CI 0.10–0.22) for ↓WS depression using the two Fig. S5).
methods, respectively. GCTA-GREML h 2SNP was not significantly Genetic correlations were 0.54 (S.E. = 0.14) for ↑WS depression
different comparing ↑WS depression with ↓WS depression (z = and ↓WS depression, 1.05 (S.E. = 0.15) for ↑WS depression and
0.57, p = 0.56), depression with weight gain (z = 1.09, p = 0.27), depression without ↑WS or ↓WS, and 0.74 (S.E. = 0.14) between
depression with weight loss (z = 0.75, p = 0.46) or depression ↓WS depression and depression without ↑WS or ↓WS.
without ↑WS or ↓WS (z = 1.25, p = 0.21). GCTB-Bayes S esti-
mates were not significantly different compared with
Discussion
GCTA-GREM estimates for any depression subgroup (Table 2).
The estimated proportion of SNPs contributing to h 2SNP ranged This study examined the genetic overlap between depression with
from 2% to 4%; S estimation was negative for all traits but not sig- atypical features (↑WS) and a range of traits and disorders using
nificantly different from zero, except for depression with weight PRS in 30 069 cases with major depression from the UKB. The

Downloaded from https://www.cambridge.org/core. Auckland University of Technology, on 14 Jul 2020 at 15:14:46, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291720002342
 6 Isabella Badini et al.

Fig. 2. Continued.

findings showed that persons with higher BMI PRS, CRP PRS and alcohol daily use and MDD; however, the direction of the effect
MDD PRS are more likely to have ↑WS depression v. non-↑WS on weight and sleep increase was the same for all these PRS, sug-
depression and ↓WS depression, while those with lower PRS for gesting that the symptoms of weight gain and sleepiness in
alcohol daily use were more likely to have non-↑WS depression depression have shared rather than divergent genetics. The PRS
rather than ↑WS depression. Associations with these PRS were of coronary artery disease and triglycerides was associated with
consistent among different PT. The effect of BMI PRS and CRP ↑WS depression v. healthy controls, in line with nominal associa-
PRS on the risk of ↑WS depression was similar when taking tions between the PRS of other cardio-metabolic traits (e.g. type 2
healthy controls as comparator group, while it was in the opposite diabetes) and the risk of ↑WS depression compared to other
direction or absent, respectively, when we compared ↓WS depres- depression subgroups as well as healthy controls. The PRS for
sion v. healthy controls. The analyses of individual symptoms alcohol dependence was associated with the risk of ↓WS depres-
(weight changes and sleep changes) showed that BMI, non-BMI sion compared to healthy controls but not for ↑WS depression
adjusted leptin and CRP PRS had a higher effect size on depres- compared to healthy controls, though the direction of the effect
sion with weight increase than sleep increase, but not the PRS of was the same. Interestingly, the PRS of alcohol daily use was

Downloaded from https://www.cambridge.org/core. Auckland University of Technology, on 14 Jul 2020 at 15:14:46, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291720002342
 Psychological Medicine 7

Fig. 3. PRS odds ratio (OR) and 95% confidence intervals of ↑WS depression v. depression without ↑WS. Colors indicate different groups of traits (substance related
disorders, major psychiatricdisorders, personality traits, immune metabolic traits). BP, bipolar disorder; SCZ, schizophrenia; ANX, anxiety disorders; PTSD, post-
traumatic stress disorder; AN, anorexia nervosa; ALCDEP, alcohol dependence; ALCUSE, daily alcohol use; N_CIGARETTES, n cigarettes per day; CANN, cannabis
use lifetime; EXTR, extraversion; NEU, neuroticism; DM2, type 2 diabetes mellitus; CAD, coronary artery disease; ISCH_STROKE, ischaemic stroke; TG, triglycerides;
LDL, total LDL cholesterol; HDL, total HDL cholesterol; BMI, body max index; BMI_adjust_leptin, leptin adjusted for BMI; CRP, C-reactive protein.

similar between ↓WS depression and healthy controls, but signifi- disorder and schizophrenia. GCTA-GREML h 2SNP was similar to
cantly lower in ↑WS depression, suggesting that the pathogenesis a previous study of depression with weight/appetite gain [h 2SNP
of alcohol use disorders (AUD) may involve different mechan- = 0.11 (S.E. = 0.03)] and weight/appetite loss [h 2SNP = 0.10 (S.E. =
isms in ↑WS compared to ↓WS depression. A recent GWAS 0.02)] (z = 1.24, p = 0.22; z = 1.72, p = 0.09, respectively)
demonstrated that alcohol consumption and AUD show signifi- (Milaneschi et al., 2017). However, this study considered both
cant genetic differences, with the genetics of AUD being more weight and appetite changes in the definition of the subgroups,
closely related to other psychiatric disorders, and the genetics of while we used only weight changes because of the lack of informa-
alcohol consumption to that of some positive health outcomes, tion on the direction of appetite changes in UKB. Genetic corre-
such as reduced risk of CAD, and lower BMI, in line with our lations suggested that ↑WS depression genetics is highly shared
findings (Kranzler et al., 2019). with depression without ↑WS or ↓WS, and the genetic correlation
Taken together, these results suggest partially distinct genetic between these subgroups was significantly higher than the genetic
pathways between depression with atypical and typical neurovege- correlation between ↑WS depression and ↓WS depression (z =
tative symptoms and that this divergence may be attributable to 2.49, p = 0.013). In line with previous evidence, major depression
distinct genetic predisposition to immune-metabolic traits. subgroups did not show evidence of negative natural selection
Although there is no convincing evidence that these depression based on GCTB-Bayes S estimates (Zeng et al., 2018).
subgroups may respond differently to conventional antidepressant Observational studies have shown differential pathophysio-
treatments, drugs acting on the specific biological mechanisms logical correlates among subtypes of depression, with atypical
implicated in ↑WS depression may have clinical benefits. For depression more strongly associated with obesity and metabolic
example, peroxisome proliferator-activated receptor (PPAR)-γ dysregulations than other depression groups (Brailean et al.,
agonists target insulin resistance and the related oxidative and 2020; Lasserre et al., 2014; Milaneschi et al., 2017). Consistent
pro-inflammatory changes (which are also involved in the patho- with previous research, the present findings suggest that the
genesis of depressive symptoms); they were demonstrated to have phenotypic association between depression with atypical features
antidepressant effects in patients with treatment-resistant bipolar and cardio-metabolic traits may partly result from shared genetic
depression and concomitant insulin resistance (Kemp et al., and biological mechanisms (Milaneschi et al., 2016, 2017). For
2014). example, pleiotropy may occur when shared genetic variants
The SNP-based heritability was estimated to be similar influence obesity-related traits and ↑WS depression through com-
between ↑WS depression and other subtypes and it did not sig- mon pathways such as inflammation and/or leptin system dysre-
nificantly change after excluding cases with probable bipolar gulation (Guo & Lu, 2014; Ring & Zeltser, 2010). A previous study

Downloaded from https://www.cambridge.org/core. Auckland University of Technology, on 14 Jul 2020 at 15:14:46, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291720002342
 8 Isabella Badini et al.

Fig. 4. PRS odds ratio (OR) and 95% confidence intervals of ↑WS depression and ↓WS depression compared with healthy controls. BP, bipolar disorder; SCZ, schizo-
phrenia; ANX, anxiety disorders; PTSD, posttraumatic stress disorder; AN, anorexia nervosa; ALCDEP, alcohol dependence; ALCUSE, daily alcohol use;
N_CIGARETTES, n cigarettes per day; CANN, cannabis use lifetime; EXTR, extraversion; NEU, neuroticism; DM2, type 2 diabetes mellitus; CAD, coronary artery dis-
ease; ISCH_STROKE, ischaemic stroke; TG, triglycerides; LDL, total LDL cholesterol; HDL, total HDL cholesterol; BMI, body max index; BMI_adjust_leptin, leptin
adjusted for BMI; CRP, C-reactive protein.

(Milaneschi et al., 2017) reported that the genetic correlation depression (Blanco et al., 2012; Brailean et al., 2020). There are
between leptin and ↑WS depression was decreased but not absent no previous studies which examined the genetic overlap between
when considering leptin adjusted for BMI, while we found no alcohol use and atypical depression, although a positive genetic
association between PRS for BMI-adjusted leptin levels and correlation between MDD and alcohol dependence has been
↑WS depression or depression with weight increase. Leptin reported (Andersen et al., 2017; Kranzler et al., 2019). Our results
reduces food intake and obesity is associated with increased leptin suggest that genetic risk for daily alcohol use is lower in ↑WS
levels, through the induction of leptin resistance (Myers, Leibel, depression v. other groups, in contrast with epidemiological
Seeley, & Schwartz, 2010). The adjustment of leptin levels for observations of higher risk of AUD in atypical cases. However,
BMI eliminates an important source in leptin inter-individual as previously noted, the genetics of alcohol daily use only partially
variability, and separates the genetic factors regulating BMI overlaps with the genetics of alcohol dependence. In addition, the
from those that regulate only leptin production. Our results indi- increased rate of alcohol use among cases with ↑WS depression
cate that the genetic factors predisposing to increased BMI and may result from secondary or environmental risk factors rather
increased leptin levels in ↑WS depression are highly correlated, than from a genetic aetiology. In line with this hypothesis, depres-
in line with the hypothesis that increased leptin levels are gener- sion with atypical features in UKB show longer, more severe and
ally a consequence of increased BMI rather than a cause (at least recurrent disease episodes, increased risk of comorbidities, and
in common forms of obesity), although different pathogenetic more lifetime deprivation and adversity (Brailean et al., 2020),
mechanisms may lead to similar phenotypes (Myers et al., and these factors may be responsible for increased risk of AUD
2010). This discrepancy may also relate to the different criteria (Yang et al., 2018).
used to define depression subgroups and sample-specific Several limitations should be considered when interpreting the
characteristics. results from the present study. First, the UKB is not representative
The current findings also showed that PRS for alcohol daily of the general population, with respondents more likely to be
use was inversely associated with risk for ↑WS depression both older, female, healthier, of a higher socioeconomic background
in case only and case–control analyses, while it showed no differ- and better educated (Fry et al., 2017). Compared to the overall
ence between ↓WS depression and healthy controls. In case–con- UKB cohort, respondents of the MHQ have on average higher
trol comparisons, the PRS of alcohol dependence was associated educational and occupational attainment, are less likely to
with ↓WS depression and at the nominal level with ↑WS depres- smoke or report longstanding illness/disability (Davis & Hotopf,
sion. Epidemiological studies have shown increased alcohol con- 2019). This is likely to impact on prevalence and severity of
sumption and/or risk of AUD among persons with atypical depression within the sample, with persons with severe

Downloaded from https://www.cambridge.org/core. Auckland University of Technology, on 14 Jul 2020 at 15:14:46, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291720002342
 Psychological Medicine 9

depression less likely to have completed the MHQ. Second, mea-

Heritability and S.E. in parenthesis are reported on the liability scale according to different possible values of prevalence (K) of each subtype in the population. The most credible values of population prevalence are in bold and a Z-test was performed to
Table 2. SNP heritability (SNP-h2) of depression with weight gain and hypersomnia (↑WS depression), depression with weight loss and reduced sleep (↓WS depression), major depression not falling in any of these two

sures for depression were based on recall of the worst episode of

Z-test GCTA-GCTB depression and assessed using self-reported symptoms, rather

Z = 0.71, p = 0.50

Z = 0.85, p = 0.39

Z = 1.09, p = 0.27

Z = 1.09, p = 0.27

Z = 0.77, p = 0.44
than clinical diagnoses. This raises the possibility that self-
results

reported symptoms may be affected by recall bias or other medical

conditions, and we are not able to exclude that participants had
other depressive episodes with different neurovegetative symp-
toms compared to the one reported in the MHQ. Third, our
measurement of depression with atypical features did not capture
all DSM-5 symptoms but only the neurovegetative component.
SNP-h2 K = 0.07

0.1633 (0.0317)
0.1177 (0.0272)
0.1165 (0.0262)
0.0903 (0.0217)
Our results were in line with previous studies and contributed
to characterise the genetics of these symptoms, but future research
using the full atypical spectrum should be considered and dimen-
sional classifications should be considered as an alternative to bin-
ary categorisation. Fourth, due to the small numbers of
0.1115 (0.0251) individuals of non-European groups in UKB, and their systematic
0.0864 (0.0207)
SNP-h2 K = 0.06

0.1803 (0.0335)
0.1357 (0.0232)
0.1566 (0.0305)
0.1127 (0.0261)

under-representation in GWAS studies used to derive PRS, our

sample is limited to individuals of European descent, limiting
the generalisability of our findings. Last, we had no statistical
power to perform a genome-wide association study of ↑WS
depression or Mendelian randomisation to assess the causal rela-
tionship between immune-metabolic traits and ↑WS depression.
0.1490 (0.0291)
0.1071 (0.0249)
SNP-h2 K = 0.05

In conclusion, this study showed specific genetic overlap of

0.1358 (0.0408)
0.0930 (0.0286)
0.1718 (0.0321)
0.1290 (0.0222)

0.1060 (0.0239)
0.0820 (0.0197)

↑WS and ↓WS depression with immune-metabolic and alcohol-

related traits. Our findings suggest that depression with typical
and atypical neurovegetative symptoms may represent relatively
homogenous subtypes characterised by partially distinct genetic
liabilities. Understanding the shared genetic aetiology between
immune-metabolic traits and depression subtypes could play an
0.1620 (0.0305)
0.1215 (0.0210)
SNP-h2 K = 0.04

0.2457 (0.0852)
0.1701 (0.0662)
0.1278 (0.0387)
0.0874 (0.0270)

important role in the prevention/treatment of depressive episodes

and the development of tailored treatments.
Supplementary material. The supplementary material for this article can
be found at https://doi.org/10.1017/S0033291720002342.

Acknowledgements. The authors acknowledge the use of the research com-

0.1185 (0.0362)
0.0808 (0.0252)
SNP-h2 K = 0.03

0.2287 (0.0796)
0.1578 (0.0616)

puting facility at King’s College London, Rosalind (https://rosalind.kcl.ac.uk),

which is delivered in partnership with the National Institute for Health
Research (NIHR) Biomedical Research Centres at South London &
Maudsley and Guy’s & St. Thomas’ NHS Foundation Trusts, and part-funded
by capital equipment grants from the Maudsley Charity (award 980) and
Guy’s & St. Thomas’ Charity (TR130505). UK Biobank data were accessed
under application 18177 ‘Multi-trait GWAS analyses in the UK Biobank’.
0.2073 (0.0726)
0.1425 (0.0560)
SNP-h2 K = 0.02

We thank the CHARGE Inflammation Working Group (CIWG), the

groups (no ↑WS or ↓WS), depression with weight gain or weight loss

Psychiatric Genomics Consortium (PGC), the International Cannabis

compare the estimate and S.E. of GCTA-GREML and GCTB Bayes S results.

Consortium (ICC), the Genetics of Personality Consortium (GPC), the

DIAbetes Genetics Replication And Meta-analysis (DIAGRAM) Consortium,
the Global Lipids Genetics Consortium (GLGC), The ADIPOGen Consortium,
The AGEN-BMI Working Group, The CARDIOGRAMplusC4D Consortium,
The CKDGen Consortium, The ICBP, The MAGIC Investigators, The
Method

MuTHER Consortium, The MIGen Consortium, The PAGE Consortium, The

GCTB

GCTB

GCTB

GCTB

GCTB
GCTA

GCTA

GCTA

GCTA

GCTA

ReproGen Consortium, The GENIE Consortium, The International Endogene

Consortium, AFGen Consortium, Cohorts for Heart and Aging Research in
Genomic Epidemiology (CHARGE) Consortium, International Genomics of
Blood Pressure (iGEN-BP) Consortium, INVENT Consortium, STARNET,
COMPASS Consortium, EPIC-CVD Consortium, EPIC-InterAct Consortium,
Weight gain + increased sleep

Weight loss + reduced sleep

International Stroke Genetics Consortium (ISGC), METASTROKE

Consortium, Neurology Working Group of the CHARGE Consortium, NINDS
Stroke Genetics Network (SiGN), UK Young Lacunar DNA Study,
MEGASTROKE Consortium and all the authors that contributed to generate
Depression type

No ↑WS or ↓WS

the summary statistics used for the analyses of this study.

Weight gain

Weight loss

Financial support. This research was supported by the UK Medical

(↑WS)

(↓WS)

Research Council (MR/N015746 and MR/S0151132). This paper represents

independent research part-funded by the National Institute for Health

Downloaded from https://www.cambridge.org/core. Auckland University of Technology, on 14 Jul 2020 at 15:14:46, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291720002342
 10 Isabella Badini et al.

Research (NIHR) Biomedical Research Centre at South London and Maudsley population. American Journal of Epidemiology, 186(9), 1026–1034. https://
NHS Foundation Trust and King’s College London. The views expressed are doi.org/10.1093/aje/kwx246
those of the author(s) and not necessarily those of the NHS, the NIHR or Guo, M., & Lu, X.-Y. (2014). Leptin receptor deficiency confers resistance to
the Department of Health and Social Care. Chiara Fabbri is supported by behavioral effects of fluoxetine and desipramine via separable substrates.
Fondazione Umberto Veronesi (https://www.fondazioneveronesi.it). Translational Psychiatry, 4(12), e486. https://doi.org/10.1038/tp.2014.126
Hagenaars, S., Coleman, J., Choi, S., Gaspar, H., Adams, M., Howard, D., …
Conflict of interest. Cathryn Lewis is a member of the Scientific Advisory Lewis, C. (2019). Genetic comorbidity between major depression and
Board of Myriad Neurosciences. Matthew Hotopf is the principal investigator cardio-metabolic disease, stratified by age at onset of major depression
of RADAR-CNS (funded by European Commission, Janssen, Biogen, UCB, [Preprint]. bioRxiv. https://doi.org/10.1101/645077
MSD and Lundbeck). The other authors declare no potential conflict of Haro, J. M., Arbabzadeh-Bouchez, S., Brugha, T. S., De Girolamo, G., Guyer,
interest. M. E., Jin, R., … Kessler, R. C. (2006). Concordance of the Composite
International Diagnostic Interview Version 3.0 (CIDI 3.0) with standar-
dized clinical assessments in the WHO World Mental Health Surveys.
International Journal of Methods in Psychiatric Research, 15(4), 167–180.
https://doi.org/10.1002/mpr.196
References
Howard, D. M., Adams, M. J., Clarke, T.-K., Hafferty, J. D., Gibson, J., Shirali,
Agosti, V., & Stewart, J. W. (2001). Atypical and non-atypical subtypes of M., … McIntosh, A. M. (2019). Genome-wide meta-analysis of depression
depression: Comparison of social functioning, symptoms, course of illness, identifies 102 independent variants and highlights the importance of the
co-morbidity and demographic features. Journal of Affective Disorders, 65 prefrontal brain regions. Nature Neuroscience, 22(3), 343–352. https://doi.
(1), 75–79. https://doi.org/10.1016/S0165-0327(00)00251-2 org/10.1038/s41593-018-0326-7
American Psychiatric Association. (2013). Diagnostic and statistical manual of James, S. L., Abate, D., Abate, K. H., Abay, S. M., Abbafati, C., Abbasi, N., …
mental disorders (5th Edn). Washington DC: American Psychiatric Murray, C. J. L. (2018). Global, regional, and national incidence, prevalence,
Association Publishing. https://doi.org/10.1176/appi.books.9780890425596. and years lived with disability for 354 diseases and injuries for 195 countries
Andersen, A. M., Pietrzak, R. H., Kranzler, H. R., Ma, L., Zhou, H., Liu, X., … and territories, 1990–2017: A systematic analysis for the Global Burden of
Han, S. (2017). Polygenic scores for major depressive disorder and risk of Disease Study 2017. The Lancet, 392(10159), 1789–1858. https://doi.org/10.
alcohol dependence. JAMA Psychiatry, 74(11), 1153. https://doi.org/10. 1016/S0140-6736(18)32279-7
1001/jamapsychiatry.2017.2269 Kemp, D. E., Schinagle, M., Gao, K., Conroy, C., Ganocy, S. J., Ismail-Beigi, F.,
Beijers, L., Wardenaar, K. J., van Loo, H. M., & Schoevers, R. A. (2019). & Calabrese, J. R. (2014). PPAR-γ agonism as a modulator of mood:
Data-driven biological subtypes of depression: Systematic review of bio- Proof-of-concept for pioglitazone in bipolar depression. CNS Drugs, 28
logical approaches to depression subtyping. Molecular Psychiatry, 24(6), (6), 571–581. https://doi.org/10.1007/s40263-014-0158-2
888–900. https://doi.org/10.1038/s41380-019-0385-5 Kessler, R. C., Andrews, G., Mroczek, D., Ustun, B., & Wittchen, H.-U. (1998).
Benazzi, F. (2002). Can only reversed vegetative symptoms define atypical The World Health Organization Composite International Diagnostic
depression? European Archives of Psychiatry and Clinical Neuroscience, Interview short-form (CIDI-SF). International Journal of Methods in
252(6), 288–293. https://doi.org/10.1007/s00406-002-0395-0 Psychiatric Research, 7(4), 171–185. https://doi.org/10.1002/mpr.47
Blanco, C., Vesga-López, O., Stewart, J. W., Liu, S.-M., Grant, B. F., & Hasin, D. Kessler, R. C., & Bromet, E. J. (2013). The epidemiology of depression across
S. (2012). Epidemiology of major depression with atypical features: Results cultures. Annual Review of Public Health, 34(1), 119–138. https://doi.org/10.
from the National Epidemiologic Survey on Alcohol and Related Conditions 1146/annurev-publhealth-031912-114409
(NESARC). The Journal of Clinical Psychiatry, 73(02), 224–232. https://doi. Kilpeläinen, T. O., Carli, J. F. M., Skowronski, A. A., Sun, Q., Kriebel, J.,
org/10.4088/JCP.10m06227 Feitosa, M. F., … Loos, R. J. F. (2016). Genome-wide meta-analysis
Brailean, A., Curtis, J., Davis, K., Dregan, A., & Hotopf, M. (2020). uncovers novel loci influencing circulating leptin levels. Nature
Characteristics, comorbidities, and correlates of atypical depression: Communications, 7, 10494. https://doi.org/10.1038/ncomms10494
Evidence from the UK Biobank Mental Health Survey. Psychological Kranzler, H. R., Zhou, H., Kember, R. L., Vickers Smith, R., Justice, A. C.,
Medicine, 50(7), 1129–1138. https://doi.org/10.1017/S0033291719001004 Damrauer, S., … Gelernter, J. (2019). Genome-wide association study of
Bycroft, C., Freeman, C., Petkova, D., Band, G., Elliott, L. T., Sharp, K., … alcohol consumption and use disorder in 274424 individuals from multiple
Marchini, J. (2018). The UK Biobank resource with deep phenotyping populations. Nature Communications, 10(1), 1499. https://doi.org/10.1038/
and genomic data. Nature, 562(7726), 203–209. https://doi.org/10.1038/ s41467-019-09480-8.
s41586-018-0579-z Lasserre, A. M., Glaus, J., Vandeleur, C. L., Marques-Vidal, P., Vaucher, J.,
Choi, S. W., & O’Reilly, P. F. (2019). PRSice-2: Polygenic Risk Score software for Bastardot, F., … Preisig, M. (2014). Depression with atypical features and
biobank-scale data. GigaScience, 8(7), giz082. https://doi.org/10.1093/giga- increase in obesity, body mass index, waist circumference, and fat mass:
science/giz082. A prospective, population-based study. JAMA Psychiatry, 71(8), 880.
Davis, K. A. S., Coleman, J. R. I., Adams, M., Allen, N., Breen, G., Cullen, B., … https://doi.org/10.1001/jamapsychiatry.2014.411
Hotopf, M. (2020). Mental health in UK Biobank – development, imple- Lee, S. H., Goddard, M. E., Wray, N. R., & Visscher, P. M. (2012). A better
mentation and results from an online questionnaire completed by 157 coefficient of determination for genetic profile analysis: A better coefficient
366 participants: A reanalysis. BJPsych Open, 6(2), e18. https://doi.org/10. of determination. Genetic Epidemiology, 36(3), 214–224. https://doi.org/10.
1192/bjo.2019.100. 1002/gepi.21614
Davis, K., & Hotopf, M. (2019). Mental health phenotyping in UK Biobank. Lee, S., Ng, K. L., & Tsang, A. (2009). Prevalence and correlates of depression
Progress in Neurology and Psychiatry, 23(1), 4–7. https://doi.org/10.1002/ with atypical symptoms in Hong Kong. Australian & New Zealand Journal
pnp.522 of Psychiatry, 43(12), 1147–1154. https://doi.org/10.3109/00048670903279895
Euesden, J., Lewis, C. M., & O’Reilly, P. F. (2015). PRSice: Polygenic Risk Score Lee, S. H., Wray, N. R., Goddard, M. E., & Visscher, P. M. (2011). Estimating
software. Bioinformatics, 31(9), 1466–1468. https://doi.org/10.1093/bio missing heritability for disease from genome-wide association studies. The
informatics/btu848 American Journal of Human Genetics, 88(3), 294–305. https://doi.org/10.
Fried, E. I., & Nesse, R. M. (2015). Depression is not a consistent syndrome: 1016/j.ajhg.2011.02.002
An investigation of unique symptom patterns in the STAR*D study. Levitan, R. D., Lesage, A., Parikh, S. V., Goering, P., & Kennedy, S. H. (1997).
Journal of Affective Disorders, 172, 96–102. https://doi.org/10.1016/j.jad. Reversed neurovegetative symptoms of depression: A community study of
2014.10.010 Ontario. The American Journal of Psychiatry, 154(7), 934–940. https://doi.
Fry, A., Littlejohns, T. J., Sudlow, C., Doherty, N., Adamska, L., Sprosen, T., … org/10.1176/ajp.154.7.934
Allen, N. E. (2017). Comparison of sociodemographic and health-related Ligthart, S., Vaez, A., Võsa, U., Stathopoulou, M. G., de Vries, P. S., Prins, B.
characteristics of UK Biobank participants with those of the general P., … Alizadeh, B. Z. (2018). Genome analyses of >200000 individuals

Downloaded from https://www.cambridge.org/core. Auckland University of Technology, on 14 Jul 2020 at 15:14:46, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291720002342
 identify 58 loci for chronic inflammation and highlight pathways that link Schumann, G., Liu, C., O’Reilly, P., Gao, H., Song, P., Xu, B., … Elliott, P.
inflammation and complex disorders. American Journal of Human (2016). KLB Is associated with alcohol drinking, and its gene product
Genetics, 103(5), 691–706. https://doi.org/10.1016/j.ajhg.2018.09.009 β-Klotho is necessary for FGF21 regulation of alcohol preference.
Lim, G. Y., Tam, W. W., Lu, Y., Ho, C. S., Zhang, M. W., & Ho, R. C. (2018). Proceedings of the National Academy of Sciences of the USA, 113(50),
Prevalence of depression in the community from 30 countries between 1994 14372–14377. https://doi.org/10.1073/pnas.1611243113
and 2014. Scientific Reports, 8(1), 2861. https://doi.org/10.1038/s41598-018- Sullivan, P. F., Neale, M. C., & Kendler, K. S. (2000). Genetic epidemiology of
21243-x major depression: Review and meta-analysis. The American Journal of
Locke, A. E., Kahali, B., Berndt, S. I., Justice, A. E., Pers, T. H., Day, F. R., … Psychiatry, 157(10), 1552–1562. https://doi.org/10.1176/appi.ajp.157.10.1552
Speliotes, E. K. (2015). Genetic studies of body mass index yield new Thorp, J. G., Marees, A. T., Ong, J.-S., An, J., MacGregor, S., & Derks, E. M.
insights for obesity biology. Nature, 518(7538), 197–206. https://doi.org/ (2019). Genetic heterogeneity in self-reported depressive symptoms identi-
10.1038/nature14177 fied through genetic analyses of the PHQ-9. Psychological Medicine, 1–12.
Łojko, D., Buzuk, G., Owecki, M., Ruchała, M., & Rybakowski, J. K. (2015). https://doi.org/10.1017/S0033291719002526
Atypical features in depression: Association with obesity and bipolar dis- UK Biobank (2019). UK Biobank: Improving the health of future generations.
order. Journal of Affective Disorders, 185, 76–80. https://doi.org/10.1016/j. https://www.ukbiobank.ac.uk
jad.2015.06.020 Visscher, P. M., Hemani, G., Vinkhuyzen, A. A. E., Chen, G.-B., Lee, S. H.,
Łojko, D., & Rybakowski, J. (2017). Atypical depression: Current perspectives. Wray, N. R., … Yang, J. (2014). Statistical power to detect genetic (co)vari-
Neuropsychiatric Disease and Treatment, Volume, 13, 2447–2456. https:// ance of complex traits using SNP data in unrelated samples. PLoS Genetics,
doi.org/10.2147/NDT.S147317 10(4), e1004269. https://doi.org/10.1371/journal.pgen.1004269
Manichaikul, A., Mychaleckyj, J. C., Rich, S. S., Daly, K., Sale, M., & Chen, Warren, H. R., Evangelou, E., Cabrera, C. P., Gao, H., Ren, M., Mifsud, B., …
W.-M. (2010). Robust relationship inference in genome-wide association UK Biobank CardioMetabolic Consortium BP working group. (2017).
studies. Bioinformatics (Oxford, England), 26(22), 2867–2873. https://doi. Genome-wide association analysis identifies novel blood pressure loci and
org/10.1093/bioinformatics/btq559 offers biological insights into cardiovascular risk. Nature Genetics, 49(3),
Matza, L. S., Revicki, D. A., Davidson, J. R., & Stewart, J. W. (2003). Depression 403–415. https://doi.org/10.1038/ng.3768
with atypical features in the National Comorbidity Survey: Classification, Watson, H. J., Yilmaz, Z., Thornton, L. M., Hübel, C., Coleman, J. R. I.,
description, and consequences. Archives of General Psychiatry, 60(8), 817. Gaspar, H. A., … Bulik, C. M. (2019). Genome-wide association study iden-
https://doi.org/10.1001/archpsyc.60.8.817 tifies eight risk loci and implicates metabo-psychiatric origins for anorexia
Milaneschi, Y., Lamers, F., Peyrot, W. J., Abdellaoui, A., Willemsen, G., nervosa. Nature Genetics, 51(8), 1207–1214. https://doi.org/10.1038/s41588-
Hottenga, … Penninx, B. W. J. H. (2016). Polygenic dissection of major 019-0439-2
depression clinical heterogeneity. Molecular Psychiatry, 21(4), 516–522. Wray, N. R., Ripke, S., Mattheisen, M., Trzaskowski, M., Byrne, E. M., Abdellaoui,
https://doi.org/10.1038/mp.2015.86 A., … Major Depressive Disorder Working Group of the Psychiatric
Milaneschi, Y., Lamers, F., Peyrot, W. J., Baune, B. T., Breen, G., Dehghan, A., Genomics Consortium. (2018). Genome-wide association analyses identify
… CHARGE Inflammation Working Group and the Major Depressive 44 risk variants and refine the genetic architecture of major depression.
Disorder Working Group of the Psychiatric Genomics Consortium. Nature Genetics, 50(5), 668–681. https://doi.org/10.1038/s41588-018-0090-3
(2017). Genetic association of major depression with atypical features and Yang, J., Lee, S. H., Goddard, M. E., & Visscher, P. M. (2011). GCTA: A tool
obesity-related immunometabolic dysregulations. JAMA Psychiatry, 74 for genome-wide complex trait analysis. American Journal of Human
(12), 1214–1225. https://doi.org/10.1001/jamapsychiatry.2017.3016 Genetics, 88(1), 76–82. https://doi.org/10.1016/j.ajhg.2010.11.011
Myers, M. G., Leibel, R. L., Seeley, R. J., & Schwartz, M. W. (2010). Obesity and Yang, P., Tao, R., He, C., Liu, S., Wang, Y., & Zhang, X. (2018). The risk factors
leptin resistance: Distinguishing cause from effect. Trends in Endocrinology of the alcohol use disorders – through review of its comorbidities. Frontiers
& Metabolism, 21(11), 643–651. https://doi.org/10.1016/j.tem.2010.08.002 in Neuroscience, 12, 303. https://doi.org/10.3389/fnins.2018.00303.
Ring, L. E., & Zeltser, L. M. (2010). Disruption of hypothalamic leptin signal- Zeng, J., de Vlaming, R., Wu, Y., Robinson, M. R., Lloyd-Jones, L. R., Yengo,
ing in mice leads to early-onset obesity, but physiological adaptations in L., … Yang, J. (2018). Signatures of negative selection in the genetic archi-
mature animals stabilize adiposity levels. Journal of Clinical Investigation, tecture of human complex traits. Nature Genetics, 50(5), 746–753. https://
120(8), 2931–2941. https://doi.org/10.1172/JCI41985 doi.org/10.1038/s41588-018-0101-4

Downloaded from https://www.cambridge.org/core. Auckland University of Technology, on 14 Jul 2020 at 15:14:46, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291720002342