COMT val158met influence on electroconvulsive therapy response in major depression^†‡

As captured by the “catecholamine hypothesis of depression,” in patients with major depression a deficit in brain norepinephrine and dopamine has been suggested. The two major biological treatment options of depression, namely pharmacological treatment and electroconvulsive therapy (ECT), both have been shown to target these two neurotransmitter systems in differential ways [Mann, 1998]. The catechol-O-methyltransferase (COMT) plays a pivotal role in the degradation of norepinephrine and dopamine [see Mannisto and Kaakkola, 1999] and therefore serves as a promising candidate in the study of the mechanism of antidepressant action.

The COMT gene (NM_000754) mapping to chromosome 22q11.2 contains a functional 472G/A single-nucleotide polymorphism (rs4680) causing an amino acid substitution from valine to methionine in codon 158 (val158met) of the membrane-bound form of the enzyme (codon 108 (val108met) of the soluble form). The valine allele (472G) has initially been reported to result in a 40% or even three- to fourfold higher COMT activity as compared to the methionine allele (472A) [Lachman et al., 1996; Chen et al., 2004]. The COMT val158met polymorphism has repeatedly been investigated for association with Major Depression with contradictory reports of no association [Kunugi et al., 1997; Frisch et al., 1999; Cusin et al., 2002; Serretti et al., 2003], possible association with the valine allele [Funke et al., 2005; Massat et al., 2005] or conversely the methionine allele [Ohara et al., 1998]. Two published studies on the role of the COMT val158met polymorphism in antidepressant treatment response report a tentative negative effect of the COMT 158met/met genotype on mirtazapine and citalopram response in Major Depression [Szegedi et al., 2005; Arias et al., 2006], while another recent study observed a negative influence of the higher activity COMT 158val/val genotype on antidepressant treatment [Baune et al., 2008]. In contrast to pharmacological treatment, little is known about the biological mechanism underlying therapeutic action of ECT as the most powerful antidepressant treatment strategy available today. With respect to the role of COMT val158met in the mediation of response to electroconvulsive therapy (ECT), to the best of our knowledge only one single study has been published to date reporting a significantly better response to ECT as conferred by the more active COMT 158val allele [Anttila et al., 2008].

In the present study, we attempted to further elucidate the effect of the COMT val158met polymorphism on ECT response by investigating a sample of 104 patients with pharmacologically treatment-resistant Major Depression with particular attention to possible gender-specific effects.

A sample of 104 unrelated Caucasian patients with a current Major Depression (mean age: 56.6 ± 13.6; f = 71 (68.3%), m = 33 (31.7%); age of onset: 40.6 ± 14.6; duration of disease in years: 6.8 ± 6.9; number of depressive episodes: 3.7 ± 3.0; number of suicide attempts: 0.7 ± 1.1; HAM-D baseline score at admission: 24.6 ± 8.5) admitted for inpatient treatment and receiving electroconvulsive therapy (ECT) were consecutively recruited at the Department of Psychiatry, University of Muenster, Germany, between 2003 and 2008. Patients with Bipolar Disorder, Schizoaffective Disorders or comorbid Substance Abuse Disorders, mental retardation, neurological or neurodegenerative disorders impairing psychiatric evaluation were not included in this analysis. In order to minimize the risk of ethnic stratification, Caucasian descent was ascertained by Caucasian background of both parents. The ethics committee of the University of Muenster, Muenster, Germany, approved of the study. After complete description of the study to the subjects, written informed consent was obtained. Patients' diagnoses were obtained by the use of a structured clinical interview (SCID-I) according to the criteria of DSM-IV [Wittchen et al., 1997]. Clinical course of depression was assessed by clinically experienced psychiatrists with the Hamilton Depression (HAM-D-21) scale on a weekly basis with a mean pre-ECT HAM-D score of 22.9 ± 8.6 and a mean post-ECT HAM-D score of 9.1 ± 6.9 (mean difference 13.7 ± 11.1, T = 12.6, df = 102, P < 0.0005). Treatment response was measured categorically (response: decrease in HAM-D score > 50%; non-response: decrease in HAM-D score ≤ 50%) and as defined by the intra-individual course of pre- and post-ECT HAM-D scores.

All patients undergoing electroconvulsive treatment suffered from pharmacologically treatment-resistant depression after at least two unsuccessful cycles of pharmacological antidepressant treatment.

Brief pulse ECT was administered three times a week with a 2-day interval (Monday–Wednesday–Friday) in all patients using the Thymatron System-IV™ system (Thymatron, Somatics Inc., Lake Bluff, IL) with a mean total number of single ECT treatments of 14.2 ± 6.0. Initial stimulus intensity was determined. Restimulation including dosage elevation in 5–10% steps was carried out during subsequent treatments in cases of short EEG (<25 sec) seizure activity. In case of increased seizure threshold throughout the course of ECT, stimulus intensity was adapted in the same manner. For anesthesia, trapanal or etomidate were used. For muscle relaxation, succinylcholine was administered. Physiological monitoring included a two-lead EEG, single-lead EMG, ECG and blood pressure monitoring. All patients received right unilateral ECT. In six patients, treatment was switched to bilateral ECT because of insufficient response to unilateral treatment.

Seventy-four patients (71.2%) received concomitant antidepressant pharmacological treatment during ECT, with 39 patients being treated with a combination of two to three antidepressants (see Table I).

Genotyping for the COMT val158met polymorphism was performed according to published protocols [e.g., Domschke et al., 2004] and yielded a completion rate of 100% for all included patients. Genotypes were determined by investigators blinded for clinical diagnoses. For statistical analyses, genotypes were grouped according to functionality with carriers of at least one more active 472G allele (AG/GG) (N = 82) versus homozygous carriers of the A allele (N = 22) [cf. Szegedi et al., 2005; Arias et al., 2006; Domschke et al., 2008].

Categorical association analyses (e.g., genotype × gender; genotype × concomitant medication; genotype × categorical response) were performed using Pearson's Chi-square-test (see Tables I and II). Further statistical comparisons between genotype and continuous measures such as age, age of onset, HAM-D baseline score and duration of ECT (see Table I) were made using multivariate ANOVA. Pre–post-ECT HAM-D score difference was normally distributed as assessed by Kolmogorov–Smirnov test. Homoschedasticity was verified by Levene's test for homogeneity of variances. COMT val158met genotype effects on individual absolute HAM-D score changes over the course of ECT were assessed using an overall ANOVA (genotype as fixed factor, gender and HAM-D baseline score as covariates (since these two variables influenced treatment outcome) and pre–post-ECT HAM-D score difference as independent variable) (SPSS, Version 15.0, SPSS Inc., Chicago, IL). In addition, stratified analyses for gender as well as three dimensions underlying the HAM-D scale “core depression” (items 1, 2, 7, 8, 13), “anxiety” (9, 10, 11, 15, 16) and “insomnia” (items 4, 5, 6) [cf. Fleck et al., 1995] were performed. False discovery rate (FDR) [Benjamini and Hochberg, 1995] was applied to control for multiple testing and prevent from type I error. FDR was calculated for the number of analyses carried out in relation to treatment response. The resulting FDR has a corrected P-value of P ≤ 0.034. With these parameters, for continuous measurements, our sample had a high power (99%) to detect a difference of at least 6.6 points (Cohen's d = 0.73; effect size r = 0.35) on the HAM-D scale between two genotypes (type I error probability for a two sided test α 0.05). Genotypes did not deviate from Hardy–Weinberg equilibrium (AA: 22, AG: 58, GG: 24; P = 0.24) as examined using the program Finetti provided as an online source (http://ihg.gsf.de/cgi-bin/hw/hwa1.pl; Wienker TF and Strom TM).

Age (P = 0.66), age of onset (P = 0.14), the duration of disease in years (P = 0.17), the number of depressive episodes (P = 0.29), number of suicide attempts (P = 0.21), number of ECT treatments (P = 0.17) or the concomitant use of antidepressants (P = 0.52), antipsychotics (P = 0.09), or anxiolytics (P = 0.19) did not affect treatment response. However, female patients responded significantly better to ECT than male patients (females: mean HAM-D score change: 15.6 ± 10.6, males: mean HAM-D score change: 9.7 ± 11.2; P = 0.004). HAM-D baseline score also influenced treatment outcome (P < 0.01). There was no association of COMT val158met genotype with age, age of onset, HAM-D baseline score, the number of ECT treatments and the concomitant use of antidepressants, antipsychotics or anxiolytics (see Table I).

However, COMT val158met genotype (AA vs. AG/GG) was significantly associated with pre-ECT severity of depression (AG/GG: mean HAM-D score: 23.8 ± 8.8, AA: mean HAM-D score: 19.2 ± 6.8; P = 0.024), particularly in the female subgroup of patients (P = 0.017), with carriers of at least one G allele being rendered more severely affected after at least two cycles of unsuccessful antidepressant pharmacological treatments.

These more severely affected carriers of at least one G allele responded significantly better to ECT as compared to AA homozygotes over the course of treatment when using a categorical analysis for COMT val158met influence on treatment response as defined by a >50% decrease in HAM-D scores (P = 0.009; see Table II) as well as when analyzing the intra-individual decrease of HAM-D scores over the course of ECT (AG/GG: mean absolute HAM-D score change: 15.1 ± 11.2, AA: mean HAM-D score change: 8.5 ± 9.3; P = 0.012; with gender (P = 0.028) and HAM-D baseline score (P < 0.01) as covariates); after correction for outliers as implemented in SPSS (N = 3): P = 0.005). Sub-analyses for the dimensions “core depression,” “anxiety,” and “insomnia,” which have been found to underlie the HAM-D scale in factor-analytical studies, revealed that this effect mainly resulted from changes in core depression symptoms (P = 0.006) as well as sleep-related symptoms (P = 0.034), but did not relate to anxiety symptoms (P = 0.128). Stratification of the sample for gender yielded a significantly better treatment response to ECT as conferred by the more active G allele in the female subgroup of patients (AG/GG: mean HAM-D score change: 17.0 ± 10.3, AA: mean HAM-D score change: 9.0 ± 9.8; P = 0.016), but not in the male group of patients (AG/GG: mean HAM-D score change: 10.4 ± 12.1, AA: mean HAM-D score change: 7.9 ± 9.3; P = 0.560).

In the present study, a significant association of the higher active COMT 158val (472G) allele with (1) higher pre-ECT severity of depression as well as (2) better treatment response to ECT—particularly regarding the core symptoms of depression as well as sleep-related symptoms—after at least two unsuccessful cycles of pharmacological antidepressant therapy was observed, which also held true after correction for multiple testing by FDR. The first result supports related findings from association studies reporting the COMT 158val allele to be a risk factor for Major Depression [Funke et al., 2005; Massat et al., 2005]. The second observation for the first time replicates a very recent finding by Anttila et al. 2008 suggesting the COMT 158val allele to be associated with a superior response to ECT. How could these two findings be embedded in an integrative pathophysiological model of antidepressant treatment response? As suggested by a previous study, COMT 158val allele carriers seem to exhibit a significantly worse treatment response to pharmacological antidepressant agents possibly by increased catechol-O-methyltransferase activity resulting in a decreased availability of dopamine and norepinephrine and therefore impairing the pharmacological efficacy of serotonergic and noradrenergic antidepressants [Baune et al., 2008]. Thus, after at least two cycles of unsuccessful pharmacological therapy, COMT 158val allele carriers might indeed be rendered with more severe depression as expressed by higher HAM-D scores prior to ECT in comparison to 158met allele carriers. In turn, given the dopaminergic effects of ECT [e.g., Yoshida et al., 1998; Ishihara and Sasa, 1999; Andrade et al., 2002], particularly carriers of the more active COMT 158val allele conferring a decreased potentially prefrontal availability of dopamine might benefit from partly dopaminergic electroconvulsive therapy as possibly captured by the present finding. However, since ECT has also been shown to alter adrenergic and serotonergic transmission as well as several subtypes of 5-HT-receptors and neurotrophic factors in the central nervous system [as reviewed by Wahlund and von Rosen, 2003], genetic influence on ECT response has to be interpreted within a very complex framework of neurotransmission and neural networks—particularly, since patients were treated with a combination of ECT and antidepressant pharmacotherapy. Thus, further research is warranted to conclusively evaluate the impact of COMT gene variation—preferably COMT tagging SNPs applying haplotype analysis—on antidepressant treatment response with respect to pharmacotherapy, electroconvulsive therapy or a combination of both, respectively.

In extension of the study by Anttila et al. 2008, in the present study a possible female-specific effect of COMT val158met in ECT response was detected. This is in line with a previous report on COMT val158met influence on antidepressant treatment response restricted to female patients [Baune et al., 2008], suggesting a sexually dimorphic pattern of genetic susceptibility to treatment response potentially in part being conferred by COMT gene variation. Female-specific effects of COMT gene variation also fit well with findings of sexually dimorphic COMT activity [e.g., Boudikova et al., 1990], which might be due to the fact that estrogen can regulate COMT transcription by binding to estrogen response elements in the promoter region of the COMT gene [Xie et al., 1999]. However, in the present sample females responded significantly better to ECT in general and also COMT val158met was significantly associated with pre-ECT severity of depression particularly in the female subgroup of patients. Additionally, the male subsample was relatively small (n = 33) and a recent study reported a male-specific effect of COMT val158met on clinical response to fluoxetine in major depressive patients [Tsai et al., 2009]. Further male-specific association findings of COMT val158met have been reported for obsessive-compulsive disorder as well as regarding cognitive function in children [as reviewed by Harrison and Tunbridge, 2008]. Thus, given a most probably underpowered sample size in the present study, particularly in the male subsample, replication studies in larger samples are required to draw a firm conclusion regarding a potential gender-specific impact of COMT gene variation on electroconvulsive therapy response as suggested here.

The following limitations have to be considered while interpreting the present results: Applying a hypothesis-driven approach, we only investigated the functionally relevant COMT val158met variant, but not a set of tagging SNPs covering the entire genomic region, which limits the presently provided information on COMT gene influence on ECT response. Also, the present study does not allow for evaluation of a molecular pattern with a set of variations in a set of molecularly correlated enzymes related to COMT. Additionally, since 71.2% of the patients concomitantly received antidepressant pharmacotherapy as invariably done in daily clinical practice, the present effect—although not statistically significant in our sample—could be influenced by concomitant antidepressant pharmacotherapy [Baghai et al., 2006] and might therefore not be specific for ECT. Furthermore, the relatively small sample sizes of patients in the diagnostic subgroups (HAM-D sub-factors, gender) might have increased the risk for a false-negative or false-positive result. There was no information on stressful life events, marital status, concomitant somatic diseases or other psychiatric disorders such as anxiety or personality disorders, which may have impacted the outcome [Ross et al., 2004] and therefore possibly have confounded the present results.

In conclusion, the present finding supports a significant impact of COMT gene variation on electroconvulsive therapy response, potentially in a female-specific manner. Given robust identification of the COMT val158met variant as a clinical predictor for treatment response to antidepressant pharmacological as well as electroconvulsive treatment in further replication studies, in the future a-priori identification of non-responders to psychopharmacotherapy based on COMT val158met genotype might potentially aid in more efficient, economical and time saving treatment decisions in favor of electroconvulsive therapy shortening the patients' suffering as well as lowering the economic burden of healthcare costs for the treatment of depression.