Abstract
There is substantial inter-individual variation in response and adverse reactions to antidepressants, and genetic variation may, in part, explain these differences. GNB3 encodes the β3 subunit of the G protein complex, which is involved in the downstream signalling cascade following monoamine receptor activation. A functional polymorphism in this gene (C825T) has been associated with response to antidepressants. Several lines of evidence suggest that GNB3 moderates improvement in the neurovegetative symptoms of depression (such as sleep and appetite) and related adverse reactions independently of change in core mood symptoms. We here report analysis of data from GENDEP, a part-randomized pharmacogenomic trial, on the outcome of 811 subjects with major depression undergoing treatment with either escitalopram or nortriptyline in which the C825T SNP and three further SNPs in GNB3 were genotyped. The TT genotype was significantly associated with a superior response to nortriptyline and these effects were specific to improvements in neurovegetative symptoms. In addition, the same genotype predicted fewer incidents of treatment-emergent insomnia and greater weight gain on the same drug. Our results are consistent with previous associations with GNB3 and emphasize the importance of signalling genes in antidepressant response.
Introduction
More than half of all patients fail to respond to the first antidepressant they are prescribed and adverse drug reactions (ADRs) frequently interfere with treatment, often resulting in non-compliance and discontinuation (Demyttenaere et al., 1998; Uher et al., 2009a). Genetic variation may, in part, explain the inter-individual variability in response and therefore be used to predict treatment outcomes.
Membrane-associated heterotrimeric guanine nucleotide binding proteins (G-proteins) regulate secondary messenger pathways for a variety of membrane-bound receptors implicated in antidepressant action, including serotonin and noradrenaline receptors. The G-protein complex is composed of three subunits, which, following receptor activation, dissociate to influence a range of intracellular processes. Given the role of G-proteins in relaying signals caused by neurotransmitter binding into intracellular post-synaptic responses, it has been suggested that variation in the genes encoding the three subunits of G-proteins [alpha (α), beta (β) and gamma (γ)] may moderate response to antidepressants (Zill et al., 2000).
G-proteins are considerably diverse and several isoforms of each subunit have been identified. Siffert et al. (1998) identified a functional polymorphism (C825T or rs5443) in exon 10 of the gene encoding the third isoform of the beta subunit (GNB3). While the polymorphism does not result in an amino acid change in the resulting protein, it associated with the alternative splicing of GNB3 mRNA to produce the splice variant GNB3-s. GNB3-s comprises 41 fewer amino acids than the wild-type GNB3 protein, which results in increased signal transduction (Siffert et al., 1998) and an increased risk of major depression (Zill et al., 2000). A number of studies have also reported an association between the TT genotype at this single-nucleotide polymorphism (SNP) and good response to a variety of antidepressants (Joyce et al., 2003; Lee et al., 2004; Serretti et al., 2003; Wilkie et al., 2007; Zill et al., 2000). Contradictory findings have, however, been reported (Kang et al., 2007; Kato et al., 2008) and a recent meta-analysis concluded there was no effect of the C825T variant on antidepressant response (Kato and Serretti, 2010). The underlying cause of these discrepant findings is unclear; however, one possibility is that the effects of genotype are symptom specific.
Major depressive disorder is a heterogeneous construct defined by a wide variety of symptoms, which may respond differentially to treatment. Improvements in neurovegetative symptoms (such as sleep disturbance and changes in weight) have, for example, been shown to occur relatively independently of changes in mood symptoms (Katz et al., 1996), and polymorphisms in the serotonin transporter gene (SLC6A4) or circadian Locomotor Output Cycles Kaput (CLOCK) gene have been shown to affect either core mood symptoms or sleep symptoms respectively (Huezo-Diaz et al., 2009; Serretti et al., 2005, 2007).
Two lines of evidence suggest that the effects of C825T may be specific to change in neurovegetative symptoms. First, the T allele of this SNP has been associated with a number of phenotypes that overlap with the neurovegetative symptoms of depression, including disturbances in circadian rhythm (Lee et al., 2007), seasonal affective disorder (Willeit et al., 2003) and obesity (Siffert et al., 1999). T allele carriers also show more pronounced hunger and mental discomfort when fasting (Michalsen et al., 2009), gain more weight during antipsychotic treatment (Tsai et al., 2004) and benefit less from the appetite suppressant subitramine (Hauner et al., 2003). Second, a recent transcriptomic study in the rat revealed that GNB3 expression is nearly 100-fold greater in the pineal gland than other brain regions including the neocortex, cerebellum, hippocampus, hypothalamus and amygdala (Bailey et al., 2009). The pineal gland, which synthesizes melatonin in 24-h rhythm, is a central component of the mammalian circadian clock that not only regulates the sleep–wake cycle but also has been linked to metabolism, appetite (Laposky et al., 2008) and sexual behaviour (Drago and Busa, 2000).
The Genome-based Therapeutic Drugs for Depression study (GENDEP; http://gendep.iop.kcl.ac.uk/results.php) was designed to investigate both genetic and clinical predictors of response and ADRs to two antidepressants with different modes of action: escitalopram (a selective serotonin reuptake inhibitor, SSRI) and nortriptyline (a tricyclic antidepressant, TCA) (Uher et al., 2009b). In this report, we investigated the effects of GNB3 on response to both antidepressants. Given the apparent link between GNB3 and neurovegetative symptoms, we hypothesized that variation in this gene would predict changes in this symptom domain specifically but not in the core symptoms of depression.
As GNB3 is associated with both sleep and appetite, a secondary aim was to investigate the effects of GNB3 on two common ADRs to antidepressant treatment: weight gain and insomnia.
Methods
Design and sample
The data used in the current report were collected as part of the clinical trial component of GENDEP, a multi-centre part-randomized pharmacogenomics study (Uher et al., 2009b); 811 participants aged between 18 and 75 were recruited from nine treatment centres in eight European countries. Participants were of white European parentage and diagnosed with moderate to severe major depression using the Schedules for Clinical Assessment in Neuropsychiatry (SCAN version 2.1; WHO, 1994). Those without contraindications were randomly assigned to receive either escitalopram or nortriptyline and those with contraindications (such as a previous history of an adverse reaction) were non-randomly assigned to receive the alternative. Participants received flexible doses of their allocated drug daily for 12 weeks.
Measures
Antidepressant response was measured longitudinally as change in depression severity over the 12-week trial. Three established measures of depression severity were used: the Montgomery–Åsberg Depression Rating Scale (MADRS) (Montgomery and Asberg, 1979), the Hamilton Rating Scale for Depression (HRSD-17) (Hamilton, 1967) and the Beck Depression Inventory (BDI) (Beck et al., 1961). Previous psychometric analyses using the GENDEP data suggested that the MADRS had the greatest internal consistency of the three measures (Uher et al., 2008). As this scale was also designed specifically to be sensitive to change over time (Montgomery and Asberg, 1979), in the current report, participants’ weekly score on the MADRS was used as the primary measure of their response to both antidepressants. Using categorical item factor analysis, we previously reported that depressive symptoms in the GENDEP study could be described in more detail by three symptom dimensions: observed mood, cognitive and neurovegetative (somatic) symptoms (Uher et al., 2008). Weight was measured longitudinally at baseline and weeks 6, 8 and 12, and height was taken at baseline, in order to calculate body mass index (BMI).
Treatment-emergent insomnia was also measured longitudinally using the Antidepressant Side-Effect Checklist (ASEC), a self-report questionnaire designed to measure the occurrence and severity of 21 ADRs, which is provided in full as an Appendix in the published paper (Uher et al., 2009a). Participants rated the presence and severity for insomnia on a 4-point Likert scale (0 = absent, 1 = mild, 2 = moderate, 3 = severe), and also indicated whether they considered the symptom to be a side-effect of the allocated medication (yes or no). Participants completed the ASEC at baseline and each week during the 12 weeks of treatment.
Genetic analyses
The precise mechanism by which the C825T polymorphism results in the alternative splicing of GNB3 remains unknown. However, it has been suggested that in addition to C825T, other polymorphisms in linkage disequilibrium (LD) with the marker may further contribute to the alternate splicing of GNB3 by affecting mRNA folding or further splice acceptance sites (Rosskopf et al., 2002). Therefore, in addition to C825T, we genotyped three further SNPs (rs5441, rs5442 and rs5446) in GNB3. These were selected from a larger panel of eight SNPs previously genotyped in 100 control subjects of European ancestry (supplementary Table S1). Selection criteria were that the markers were polymorphic, in Hardy–Weinberg equilibrium, had a minor allele frequency of more than 5% and were not in complete linkage disequilibrium with C825T.
The selected SNPs (rs5441, rs5442, rs5443 and rs5446) were genotyped using multiplex SNaPshot following the manufacturer’s protocol. In brief, the PCR extension reactions were carried out for 30 cycles at 96°C for 10 s, 50°C for 5 s and 60°C for 30 s in a ABI9700 instrument (Applied Biosystems, Foster City, CA, USA) and the products were digested with Shrimp alkaline phosphatase to dephosphorylate unincorporated fluorescent ddNTPs. The extension primer sequences are listed in the supplementary Table S2.
The SNaPshot reaction products were analysed in 3130xl DNA Analyzer (Applied Biosystems) by mixing 1 µl of SNaPshot product with 0.5 µl of GS120 Liz Size Standard and 9.5 µl of Hi-Di Formamide (Applied Biosystems) for each sample and denaturing the injection mix at 95°C for 5 min prior to the analysis using POP7 polymer (Applied Biosystems) in E5 dyeset. Subsequent automated genotype analysis of the data was performed with GeneMapper 4.0 (Applied Biosystems).
Genotyping accuracy was determined by re-analysing 10% of the total sample with different technical approaches, and the concordance rate between genotype callings was >99.7%.
Statistical analyses
Linear mixed models were used to investigate the effects of genotype on depression severity scores, symptom dimension scores and BMI. Non-linear mixed models were used to investigate the effects of genotype on the reporting of insomnia (coded as presence or absence) over the 12-week trial.
By modelling the relatedness of repeated measures within the same individual as random effects, mixed models allow response to be assessed longitudinally and, unlike traditional end-point analyses, do not depend on retention of the whole population to measure outcome, or imputation methods, which can introduce biases (Gueorguieva and Krystal, 2004). In previous analyses of the GENDEP data, we reported that treatment centre had a moderate effect on antidepressant response (Uher et al., 2008); therefore, in addition to the random effects of subject, we modelled recruitment centre as a higher order random effect (subjects within centres) in all analyses. All models also included the covariates of allocation status (0 = random, 1 = non-random), gender (0 = male, 1 = female), age (as a continuous variable centred at the mean), standardized baseline depression severity (MADRS) and the linear and quadratic effects of time. Models assessing the effects of genotype on insomnia and weight gain also included the weekly depression severity score (MADRS) and weekly dose.
There is evidence to suggest that C825T genotypes operate in either a recessive or additive genetic model; therefore, the fixed effects of genotype assuming both models were tested. Total MADRS score was used to assess the effects of genotype on depressive symptoms overall and to compare the effects on genotype on change in specific symptoms, analyses were repeated using the symptom dimension scores (mood, cognitive, neurovegetative) as outcome variables. Similar models were used to explore the effects of genotype on change in BMI and self-reported insomnia. However, as insomnia was treated as a binary outcome (measured as weekly as presence or absence), a non-linear or logistic mixed model was used to analyse the effects of genotype on insomnia. Both models included the covariates as previously described and baseline BMI or insomnia score. To detect drug-specific effects, analyses were repeated separately in escitalopram- and nortriptyline-treated subjects.
In order to test the effects of C825T in the context of other polymorphisms in GNB3, haplotypes including all of the genotyped SNPs were estimated in Phase, version 2.1 (Stephens and Donnelly, 2003; Stephens et al., 2001) and used as alternative predictors in each of the above models. All reported p-values are two tailed. Statistical analyses were conducted in STATA version 10.0.
Results
Of the 811 participants recruited in GENDEP, 795 (292 male and 503 female) had complete data for all four of the genotyped SNPs (mean age 42.3, SD = 11. 8). There was no significant difference in the age, gender, age of onset or baseline depression severity of those included versus those not included in the study owing to missing genotype data. Of the included participants, 460 were treated with escitalopram (224 randomly allocated; 236 non-randomly allocated) and 358 were treated with nortriptyline (244 randomly allocated; 114 non-randomly allocated).
Linkage disequilibrium (as r2 and D′) between the SNPs genotyped in GNB3, with their minor allele frequencies (MAF), and Hardy–Weinberg equilibrium (HWE) p-values
Distribution of GNB3 genotypes by allocated drug
Age, age of onset, gender, number of depressive episodes, duration of illness or current episode, baseline depression severity (MADRS), neurovegetative, cognitive and mood dimension scores were unrelated to treatment arm in those randomly allocated to a drug. Results were similar in those non-randomly assigned to a drug. However, those allocated non-randomly to nortriptyline had higher baseline neurovegetative scores (t = 2.01, p = 0.02) and a younger age of onset (t = −2.23, p = 0.01) and more than those non-randomly allocated to escitalopram. In addition, more of the non-randomly allocated patients were treated with nortriptyline. These and other unmeasured baseline differences may confound combined analyses of both randomly and non-randomly allocated patients. We therefore conducted sensitivity analyses using only those randomly allocated to either antidepressant.
Antidepressant response
Change in MADRS scores. The results from linear mixed models assessing response using total MADRS score showed that C825T moderated response in the expected direction. Specifically, individuals with the TT genotype at this SNP showed an improved response to both escitalopram and nortriptyline. The effects of genotype were marginally significant (β = −0.12, 95% CI −0.24 to −0.01, p = 0.07). In a sensitivity analysis conducted using only randomly allocated participants the size and significance of the effect increased (β = −0.23, 95% CI −0.39 to −0.06, p = 0.01).
Change in symptom dimensions. Consistent with our hypothesis, C825T genotype predicted significant improvements in neurovegetative symptoms and analyses by treatment arm suggested that these effects were specific to patients treated with nortriptyline (Table 3, Figure 1). In analyses of patients randomly allocated to nortriptyline, the effects of genotype remained significant for nortriptyline (β = −0.41, 95% CI −0.62 to −0.21, p < 0.001) and non-significant for those treated with escitalopram (β = −0.13, 95% CI −0.35 to 0.10, p > 0.1). Similar results were also found when examining only patients who were medication free at baseline.
Mean neurovegetative symptom severity score by genotype over the 12-week trial (adjusted for baseline). A and B represent patients treated with escitalopram and nortriptyline, respectively. Mixed-model analyses of the effects of C825T on neurovegetative symptoms, insomnia and weight gain over the 12-week trial in patients treated with either escitalopram or nortriptyline Neurovegetative symptom score and weight (measured as BMI) were treated as a continuous outcome in linear mixed models. Insomnia (presence or absence) was treated as binary outcome in non-linear mixed models. All models included random effects both at the level of the subject and recruitment centre (subject within centre). MADRS: Montgomery–Åsberg Depression Rating Scale; BMI: body mass index.
Genotype did not predict improvement in mood (β = −0.05, 95% CI −0.18 to 0.06, p > 0.1) or cognitive symptoms (β = −0.08, 95% CI −0.21 to 0.05, p > 0.1). A sensitivity analysis suggested that these effects did not differ in those randomly or non-randomly allocated to either treatment arm (all p > 0.1).
Weight gain and treatment-emergent insomnia. C825T moderated weight gain to nortriptyline in the expected direction. Patients with the TT genotype experienced significantly more weight gain during treatment with nortriptyline over the 12-week trial. However, no effect of genotype was detected in the escitalopram-treated patients (Table 3, Figure 2). The effects of genotype were similar when analysing only those randomly allocated to nortriptyline (β = 0.68 95% CI 0.045 to 1.6, p = 0.045) or escitalopram (β = −0.099 95% CI −0.352 to −0.155, p > 0.1)
Mean percentage weight change over the 12-week trial by genotype. (A) and (B) represent patients treated with nortriptyline and escitalopram, respectively. The numbers of subjects within each genotype group are indicated in parentheses.
The same polymorphism also moderated the occurrence of treatment-emergent insomnia during nortriptyline treatment and these effects remained significant after controlling for depression severity (MADRS) and dose. Those with the TT allele reported fewer incidences of insomnia after commencing treatment over the 12-week trial. Again, no effects of genotype were detected in the escitalopram-treated patients (Table 3). Analyses conducted on the randomly allocated patients remained significant for nortriptyline (β = −1.51, 95% CI −2.73 to −0.40, p = 0.009) and non-significant in escitalopram (β = −0.33, 95% CI −1.26 to 0.60, p > 0.1).
The effects of genotype on neurovegetative symptoms, weight gain and insomnia did not differ when analysing only those who were medication free (n = 538) at baseline.
Haplotype analysis
In order to test the effects of C825T in the context of the other genotyped SNPs, we used estimated haplotypes using all of the genotyped SNPs as alterative predictors of outcome. Although predicted haplotypes containing the T allele were associated with improved neurovegetative symptoms, increased weight gain and reduced insomnia when treated with nortriptyline, the inclusion of haplotypes containing rs5443 SNP provided no advantage in the tested models over using rs5443 (C825T) as a single marker.
Discussion
The C825T polymorphism in GNB3 predicted response to nortriptyline in the GENDEP sample and these effects were specific to change in neurovegetative symptoms. The same polymorphism also predicted weight gain and treatment-emergent insomnia in nortriptyline-treated patients.
Antidepressant response, weight gain and treatment-emergent insomnia
In line with our hypothesis, the TT genotype of C825T moderated change in neurovegetative symptoms during nortriptyline treatment, but had no effect on the mood or cognitive symptoms of depression. Although no previous studies have examined the effects of GNB3 on change in specific symptom dimensions, our findings are consistent with those reporting that the TT genotype was associated with a greater improvement in overall depression rating scores (Lee et al., 2004; Serretti et al., 2003; Wilkie et al., 2007; Zill et al., 2000).
In addition to its effects on response, C825T also moderated weight gain and treatment-emergent insomnia in those treated with nortriptyline. Those with the TT genotype gained significantly more weight and reported significantly less insomnia over the 12-week trial. These effects were independent of dose and depression score and remained significant when analysing only those randomized to treatment and those medication free at baseline.
Although previous studies have not found an association between the occurrence of ADRs to antidepressants and GNB3 genotype, none has examined insomnia and weight gain specifically (Bishop et al., 2006; Kato et al., 2008). In line with our results, T allele carriers are more likely to gain weight during treatment with the antipsychotic clozapine (Tsai et al., 2004), are more likely to be obese (Siffert et al., 1999) and gain more weight during pregnancy (Dishy et al., 2003). Studies have suggested that the increases in weight seen in T allele carriers may be the result of decreased lipolysis (Hauner et al., 2002) or increased mental discomfort during hunger (Michalsen et al., 2009) in this genotype group; however, further research on this important candidate is required before any conclusions can be drawn regarding this mechanism. Also in line with our results, C825T has been shown to moderate diurnal preference in healthy volunteers (Lee et al., 2007) and the T allele has been associated with seasonal affective disorder (Willeit et al., 2003).
The primary mode of action of nortriptyline is to increase available noradrenaline and to a lesser extent serotonin in the synaptic cleft. As the T allele of C825T results in increased G-protein activation following neurotransmitter binding (Siffert et al., 1998), it may be hypothesized that the downstream processes responsible for a clinical response to antidepressants are accelerated in T allele carriers.
While the mechanisms by which antidepressants bring about change in a wide range of symptoms are largely unknown, it is likely that multiple pathways exist, which are moderated by both overlapping and distinct genetic factors. Indeed, three studies have reported symptom-specific effects of polymorphisms in both the serotonin transporter gene and CLOCK gene (Huezo-Diaz et al., 2009; Serretti et al., 2005, 2007). Given the high relative expression of GNB3 in the pineal gland (Bailey et al., 2009), it is possible that the moderating effects of C825T are specific to antidepressant action within this brain region, moderating symptoms related to circadian function.
Drug-specific effects
The effect of C825T on response was marked in those treated with the TCA nortriptyline but not significant in patients treated with the SSRI escitalopram. Previous studies have been based on samples treated with a variety of antidepressants, the majority of which have been underpowered to evaluate drug-specific effects. Interestingly, however, one study that did report drug-specific effects suggested C825T moderated response to nortriptyline but not the SSRI fluoxetine (Joyce et al., 2003).
The suprachiasmatic nucleus (SCN), which contains the central oscillator of the circadian clock, activates the 24-h rhythmic synthesis of melatonin in the pineal gland by the release of noradrenaline (Moller and Baeres, 2002). A recent study of the pineal transcriptome suggested that the majority of gene expression changes required for melatonin synthesis were dependent on the noradrenergic-cyclic AMP signalling pathway, of which heterotrimeric G-proteins are an integral component (Bailey et al., 2009). Melatonin has been shown to be effective in the treatment of both sleep disorders (Kayumov et al., 2001) and depression (Bourin and Prica, 2009). Taken together these results suggest that nortriptyline may alleviate the neurovegetative symptoms of depression by increasing melatonin synthesis. Our findings add further support to this hypothesis in that GNB3, an integral component of the same pathway, may moderate these effects.
Limitations
Our findings should be interpreted in the light of the following limitations. Commonly reported ADRs such as weight gain and insomnia show considerable overlap with the symptoms of depression. In order to dissect the boundary between treatment-emergent insomnia and the normal symptoms of depression, we covaried for baseline insomnia scores in our analyses and only considered cases of insomnia that were considered, by the patient, to be related to the drug they were taking. Similarly, we also covaried for BMI at baseline in the weight gain analyses. Despite this, it is possible that the moderation of these ADRs by genotype was at least in part the result of a superior response of TT patients to nortriptyline treatment. Indeed, decreased insomnia and increased weight may constitute part of the normal clinical response to antidepressant treatment (especially with the use of TCAs).
GNB3 genotype explained a relatively small proportion of the variance in neurovegetative symptoms, treatment-emergent insomnia and, in particular, weight gain over the 12-week trial. Although GENDEP is the largest comparative pharmacogenetic trial to study the effects of GNB3 on response to antidepressants, after stratifying the sample by treatment arm, we were able to detect only small to moderate genetic effects. Therefore, while the associations detected satisfy the nominal significance thresholds, they must be interpreted with caution and require replication in other pharmacogenetic samples.
Conclusion
Our results suggest that GNB3 genotype moderates response to the TCA nortriptyline. These effects are specific to changes in neurovegetative symptoms and the occurrence of two closely related ADRs. It is therefore possible that the inconsistent findings reported thus far for GNB3 and antidepressant response may be at least partly explained by previously unexplored drug- and symptom-specific effects.
Footnotes
Disclosures,acknowledgements and funding
The GENDEP clinical trial was funded as part of an Integrated Project by a European Commission Framework 6 grant, EC Contract Ref.: LSHB-CT-2003-503428. Lundbeck provided both nortriptyline and escitalopram free of charge for GENDEP. GlaxoSmithKline, the Medical Research Council, and the Biomedical Research Centre for Mental Health at the Institute of Psychiatry and South London and Maudsley NHS Foundation Trust (funded by the United Kingdom National Institute for Health Research of the Department of Health) latterly contributed to the funding of the sample collection at the Institute of Psychiatry, London, through add-on projects or staff funding. The sponsors had no role in the design and conduct of the study, in data collection, analysis, interpretation or writing of the report.
Keers, Scassellati, Bonvicini, Uher, Placentino, Giovannini, Rietschel, Kozel, Mors, Maier, Hauser, Schmäl, Zobel, Larsen, Mendlewicz, Szczepankiewicz, Kovakic, Elkin, Craig and Gennarelli report no competing interests. Henigsberg and Souery participated in clinical trials sponsored by pharmaceutical companies including GlaxoSmithKline and Lundbeck. Aitchison, Farmer and McGuffin have received consultancy fees and honoraria for participating in expert panels from pharmaceutical companies including Lundbeck and GlaxoSmithKline.
We would like to acknowledge the contribution of the following collaborators: Helen Dean, Bhanu Gupta, Joanna Gray, Cerisse Gunasinghe, Desmond Campbell, Richard J Williamson, Thomas Schulze, Jana Strohmaier, Susanne Höfels, Anna Schuhmacher, Ute Pfeiffer, Sandra Weber, Anne Schinkel Stamp, Dejan Kozel, Alenka Tancic, Jerneja Sveticic, Zrnka Kovacic, Paweł Kapelski, Maria Skibińska, Aleksandra Rajewska, Aleksandra Szczepankiewicz, and Elzbieta Cegielska. We would like especially to acknowledge the contribution of Jorge Perez, who was the Lead Investigator at Brescia, Italy, and who passed away in October 2007, and the late Professor Andrej Marusic, who was the Lead Investigator at Ljubljana, and who passed away in June 2008.