Background

Depression affects more than 40 million Americans at some time during their lives and represents one of the most debilitating medical conditions. A substantial body of evidence drawn from a range of methods, proband groups, and diagnostic criteria has established that a familial phenotype is present in patients with major depression [1–7]. Despite their limitations, molecular methods such as candidate gene association studies also support the concept of a genetic influence in vulnerability to depression [8, 9]. Although genetic studies promise to improve our understanding of the pathophysiological and genetic aspects of the depressive syndrome, such understanding is limited by the genetic complexity and lack of discrete, etio-pathologically related, and homogeneous "depression" phenotypes.

Treatment resistant depression (TRD) is a characteristic of some depression phenotypes. It can range from not being able to respond to one medication, to not clinically responding to electroconvulsive therapy[10]. TRD, or non-response can be viewed as a spectrum of severity of major depressive disorder, as the amount of suffering is greater for those with TRD, than those without[11]. Whether or not TRD is a subtype of depression with a distinct underlying biology is still unknown and under investigation. It is possible that to some extent TRD is a result of hyperfunctioning enzymes that metabolize typical antidepressant medications [12]. In this case, the TRD subjects would have a genetic background that leads to a pharmacokinetic difference from the depressive group which is capable of antidepressant treatment response. This genetic discrepancy between two diagnostically similar groups (though phenotypically difference in terms of response profiles) may be detectable, and would provide insight to the biology of treatment resistant depression.

To explore genetic differences within depression endophenotypes, we genotyped Caucasian subjects with unipolar major depression from two distinct treatment response profiles: those able to respond to treatment, and those not able to respond. The genes being studied encode for receptors, transporters, synthetic or degradation enzymes, transcription factors, and neurotrophic factors related to monoamine function. We hypothesized that the depressive group as a whole would be genetically distinct from the healthy controls. It was predicted that there would be a higher frequency of "risk" genotypes (excessive or deficient in functioning depending on the case) in the depressive groups when compared to the controls.

We report one two single gene findings and one polygenic discovery. A difference in genotypic frequency between the controls and depressed subjects was detected in the serotonin transporter intron 2 VNTR polymorphism. A surprising difference was detected between the responder and non-responders in the dopamine D4 receptor exon III VNTR. Furthermore, we demonstrate that depressed subjects have more risk genotypes that non-depressed subjects, and that non-responders have the most risk genotypes of all our phenotypic groups analyzed.

Methods

Subjects

Ninety-seven subjects, aged 18 to 85 years and diagnosed with unipolar major depression (DSM-IV)[13] participated in the study (see Table 1). Fifty-eight of these subjects had participated in a research trial for treatment resistant depression (TRD); selection criteria for this group included clinical diagnosis of chronic or recurrent depression and failure to achieve remission after at least two adequate antidepressant trials as documented by the Antidepressant Treatment History Form (ATHF) [14–20] These subjects are a subgroup of a vagal nerve stimulation (VNS) trial, and express a highly debilitating course of major depressive disorder, and nearly 90% failed to respond to selective serotonin re-uptake inhibitors (SSRIs). Further details about the subjects has been previously described[21]. Thirty-nine others were selected because they had recovered from a major depressive episode (diagnosed according to DSM-III-R criteria), facilitated by the capability to respond to standard antidepressant medication. The response to the antidepressant medication was indicated by a score of ≤14 on the 25 item – Hamilton Depression Rating Scale[22]. The control group consisted of 85 volunteer students from the University of Arizona, who denied a personal history of mental illness based on the mood disorders section of the Structured Questionnaire Interview for DSM-IV-R (SCID) questionnaire[23]. All participants were of European descent.

Table 1 Clinical description and demographics of subjects groups
Full size table

Genotyping

The study was approved by each of the participating institutions. All subjects gave written informed consent for genotyping; samples for DNA analysis were obtained from whole blood or cheek cells. PCR-based genotyping was performed at the Laboratory of Molecular Psychiatry of The University of Arizona for 21 genetic polymorphisms. The details of the analyzed polymorphisms can be found in Table 2. Primer sequences, methodology details and allele description can be found in TABLE 1S provided as supplemental information (See additional file 1). Genotypic frequencies did not deviate from expected Hardy-Weinberg frequencies.

Table 2 Polymorphism Descriptions
Full size table

Statistical analysis

Initially, four comparisons were made: between subjects with a history of major depression vs. controls, non-responders vs. controls, responders vs. controls, and non-responders vs. responders. The data presented in this study is a result of a chi-square analysis was used to test for differences in the frequency of genotype polymorphisms between the depressive subjects and healthy controls as well as between the responders and non-responders. The remaining two comparisons (responders vs. controls, non-responders vs. controls) is available upon request. All significance tests (aside from QVALUE) were done with SPSS version 14. Genotypic frequency differences were analyzed by Chi-Square tests. False discovery rate analysis was used to control for multiple comparisons between groups and number of polymorphisms. The q-values[24] are reported with all p values ≤0.05. A q-value is an indication of the percent of time a Type I error would be committed if a corresponding p-value was considered as statistically significant. For example, we accepted a false discovery rate of 15%, thus based on the value distribution of the present dataset, a value of q ≤0.137 was considered statistically significant, however, this still implies that 13.7% of the time when we accept a corresponding p-value as statistically significant, an error would be made. The statistical testing for each polymorphism was between a risk genotype determined a priori to testing (see Table 2), and the remaining genotypic groups. The mean number of risk genotypes among the phenotypic groups was analyzed using independent samples t-tests. We chose this approach because we had hypotheses that the history positive group would have more risk genotypes that the history negative, and likewise, that the non-responders would have more risk genotypes that the non-responders.

Results

Table 3 provides the p-value, and corresponding Q-value, for all the polymorphisms tested. A multiple comparison correction (Q-value) was used to decrease the number of false discoveries in the analysis. Genotype frequency differences were statistically significant for the serotonin transporter gene intron-2 VNTR (STin2) between subjects with a history of depression and controls (p = 0.004). The disparate genotypic distribution exists in the heterozygote and 12-repeat homozygote groups. Over 50% of the controls fall into the 12/12 genotypic group, whereas just less than 30% of the depressed individuals do. Likewise, over half of the depressed subjects fall in the heterozygote group, whereas just over 30% of the controls do. The comparison between the depressed and non-depressed subjects also revealed that the DRD3 and two COMT polymorphisms were close to being statistically significant but did not hold up to multiple comparison correction. A statistically significant difference was detected in the genotypic frequency distributions between the non-responders and the responders of the DRD4 exon 3 VNTR (p = 0.009). The genotypic group with zero-7-repeat alleles was largely over-represented in the non-responders, compared to 38% of the responders. Likewise, the homozygous 7-repeat genotypic group had twice the amount of responders compared to non-responders.

Table 3 P and Q-values for trending and statistically significant (after multiple comparison correction *) polymorphisms
Full size table

Figure 1 shows the distribution of the number of risk genotypes for all phenotypic groups. The mean number of risk genotypes (see Figure 1) is statistically significantly different between the controls and depressed subjects (p = 0.017) as well as between controls and non-responders (p = 0.005).

Figure 1
figure 1

Distribution of risk genotypes for all phenotypic groups (mean # risk genotypes ± SEM).

Full size image

Discussion

The genetic susceptibility to major depression was tested using genotype frequency comparisons between subjects with a history of depression and controls. We detected three polymorphisms (SLC6A4 intron 2 VNTR, 2 COMT polymorphisms) which may contribute to the genetic susceptibility to major depression, however, only one (SLC6A4) which held up to multiple comparison corrections.

The comparison between the non-responders and responders is a method to detect the genetic basis for antidepressant response, in general. Out of the 21 polymorphisms tested, only one (DRD4 exon 3 VNTR) was statistically significant. This is compared to the 3–4 polymorphisms which demonstrate a genetic susceptibility to depression. It was unpredicted that the genetic basis for antidepressant response may differ from the genetic basis of susceptibility to major depression.

We further demonstrated that the number of risk genotypes is not consistent across phenotypes. The non-responders have the most risk genotypes, followed by (in decreasing order) the depressed group as a whole, the responders, and the controls. This supports the notion that Major Depressive Disorder is polygenic, and that the number of risk genotypes may be an indication of the susceptibility to Major Depressive Disorder as well as the severity of the disorder.

In retrospect, we recognize that much of our current understanding of the pathophysiology of depressive disorders has been inferred from the prevailing hypotheses of mechanisms for antidepressant action. Accordingly, candidate genes selected for their function are commonly associated with monoamine function or their putative intracellular responses to neurotransmitter activation. The neurobiology of treatment resistance is not well understood; it may represent an extreme phenotype along a unique pathological continuum of depression. This interpretation may be supported by the fact that the usual monoamine-based treatment interventions, by definition, provide little or no benefit to patients with treatment resistant depression.

The Hamilton Depression Rating Scale has been for decades the gold standard of depression symptoms quantification and has been utilized in >95% of antidepressant treatment studies. There have been lately a number of concerns primarily about the scale's ability to reliably quantify antidepressant responses given the multi-dimensionality of the scale, which may non-specifically determine response in individuals treated with sedating agents such as TCA's. In the other hand, given the number of non-melancholic specific items, it may be less sensitive at detecting an antidepressant response. It is possible that alternative tools such as the Bech Rafelsen or the Montgomery Asberg Depression Rating Scale among others may more accurately reflect antidepressant responses leading to a more reliable detection of therapeutic effects.

This pilot study provides a proof of concept that two depressive phenotypes (antidepressant response and non-response) may be subtypes of Major Depressive Disorder. In terms of medication response, these patients are clearly distinct from each other, thus it is logical to propose their genetic constitution may be distinct. The polymorphisms analyzed in this study is by no means an exhaustive list of polymorphisms which play a role in Major Depressive Disorder, however it is a fairly representative list for the monoamine related polymorphisms thought to confer susceptibility to major depression. It is true that the mean age of the controls is younger compared to the depressed groups, thus they have not necessarily passed the age of onset typical of major depressive disorder. As a result, we may be committing a type I error in the analyses, and be missing or underestimating genetic differences that exist between the controls and depressed subjects, however, we are confident in the effect that we did detect. A large number of comparisons between groups for each polymorphism were initially performed in this study. To correct for false discovery, a q-value was determined, which provided an indication of statistical significance. We accepted a q-value less than 15%, which indicates that 15% of the time we considered out findings to be significant, we would be incorrect. Given that this is a pilot study with relatively small sample size, our single-polymorphism findings may be false positives, and thus necessitate replication in an independent and larger sample. Polymorphisms in alternative neurotransmitter systems, neurotrophic pathways, neurosteroids and antidepressant metabolic pathways (e.g. cytochrome p450 factors) should be the focus of further research.

Conclusion

Despite the modest sample size, these data support the existence of a genetic basis to the susceptibility to major depression. Furthermore, depending on the phenotypic definition used in testing, different associations may be detected. Given that we utilized phenotypic definitions based on patters of treatment response, these findings further suggest that the genetic basis for the capacity to respond to monoamine-based antidepressants is different from that of susceptibility to major depression. These data provide a proof of concept that major depressive disorder is a polygenic disorder, and that the number of risk genotypes may be an indication of susceptibility to the disorder and the severity.

Previous presentations

These data have been presented in abstract form at the Annual Meeting of the American College of Neuropsychopharmacology, December, 2004, the Society of Biological Psychiatry, May, 2005 (poster), and West Coast College of Biological Psychiatry, April, 2006 (talk).

Table 4 Genotype frequencies for polymorphisms in Table 3 for all phenotypic groups. Risk genotype indiciated by an *.
Full size table