Pharmacogenetics in mood disorder

Abstract

Purpose of review

Not all patients with mood disorder respond well to drug treatment. Emerging data suggest that genetic mechanisms could be involved. We searched the literature database and highlighted recent molecular genetic studies pertaining to drug response in mood disorders.

Recent findings

Recent pharmacogenetic studies in mood disorders have reported generally positive findings supporting the view that genotyping may improve the confidence of prospectively identifying treatment response and adverse outcomes.

Summary

The evidence documenting genotype-based response to drug treatment is rapidly expanding. Genes encoding target receptors and signal transduction systems may predict the efficacy of drug therapy in mood disorders. Additional predictors of treatment response in bipolar disorder may include the immediate early genes, mitochondrial genes and epigenetic mechanisms, although some of these studies are still preliminary.

Introduction

Recent advances in pharmacological treatment have not eliminated significant problems in the outcomes of mood disorders including mortality [1•,2]. Indeed, antidepressant, antibipolar and antipsychotic drugs are useful first-line or add-on therapeutic agents characterized by widely variable safety and efficacy [3,4,5•].

It is now well recognized that genes can influence both favorable and adverse pharmacodynamic or pharmacokinetic drug effects, making genetic analysis a potentially useful predictor of drug response in mood disorders. In the clinical setting, genotype may be used, perhaps as a panel, to improve the confidence of prospectively identifying treatment response and adverse outcomes. While efforts in this area remain in the experimental realm, the eventual commercial utility of pharmacogenetic assays offers tremendous potential in improving therapeutic outcomes. This simple and elegant idea has stimulated exciting research worldwide with remarkable progress in terms of dissecting the genetic underpinnings of psychotropic drug response [6]. The prospects of more customized drug treatment using molecular genetics are laden with a number of barriers. First, genes are not unlikely to be constrained by gene–environment, gene–gene, and epigenetic mechanisms. Second, mood disorders have an incompletely understood neurobiology. Third, the definition and ascertainment of pharmacogenetic phenotypes are not well refined. As a further consideration, it is important to note that findings from genetic linkage studies pertain to generalizations valid for groups rather than individuals; hence, while specific genes that influence treatment outcome may eventually help tailor pharmacotherapies, patient-specific risk profiling has not yet been established. Lastly, there remains a debate within the field regarding the value of pursuing individual candidate genes for linkage and association studies, relative to genome-wide searches for loci or groups of loci of otherwise unrecognized pharmacotherapeutics.

Antidepressant drugs: pharmacogenetic pharmacokinetic studies

Studies of cytochrome P-450 isoenzymes that appeared in the past year have strengthened the hope of pharmacogenetic-guided therapeutic drug monitoring [7].

Cytochrome P-450 2D6

The cytochrome P-450 2D6 gene is a highly polymorphic enzyme and encodes debrisoquine hydroxylase. Recent studies have focused on comparisons of adverse events or lack of benefit among individuals carrying multiple copies of active 2D6 gene (the ultrarapid metabolizers) or no copies at all (poor metabolizers) versus normal ‘extensive’ metabolizers [8].

Charlier and co-workers [9•] found that standard doses of fluoxetine produced significantly higher plasma levels among poor compared with extensive metabolizers, which suggested that genotyping can help to better predict plasma levels of some antidepressant drugs. Intriguingly, antidepressant drugs that are substrates for cytochrome P-450 2D6 have inconsistently been associated with more side effects among poor metabolizers than extensive metabolizers. For example, Rau and colleagues [10•] documented in a naturalistic study that one in three patients reporting adverse occurrences during antidepressant treatment tested homozygous for the 2D6 null allele. However, recent randomized prospective studies in geriatric and nongeriatric depressed adults have found no more adverse events of antidepressant drugs among poor than extensive metabolizers [11••,12••]. Apparently, poor metabolizers are not inevitably at higher risk of antidepressant drugs’ adverse events despite concomitant treatment with cytochrome P-450 2D6 substrates. In fact, the study by Murphy and co-workers [11••] also found that pharmacodynamic but not pharmacokinetic factors correlated with adverse effects of paroxetine as the C/C genotype of T102C single nucleotide polymorphism in the serotonin type 2A receptor gene was overrepresented among those with severe side effects. Thus, prospective genotyping for cytochrome P-450 2D6 alone may mitigate the cost of treatment, adverse events and lack of benefit [13], but the extent is unclear.

Cytochrome P-450 3A4

Pharmacogenetic studies of cytochrome P-450 3A4 are relatively lacking, although the isoenzyme’s crystal structure and chemical diversity were clarified recently [14]. The cytochrome P-450 3A4 isozyme has over 40 polymorphisms and two recent studies are noteworthy. In one study, venlafaxine, fluoxetine and sertraline were not significant inducers or inhibitors of cytochrome P-450 3A4 [15•], whereas the other study suggested that despite the inhibitory effects of protease inhibitors on cytochrome P-450 3A4, concomitant treatment with ritonavir, a protease inhibitor, and escitalopram, a selective serotonin reuptake inhibitor, is probably safe [16••].

Antidepressant drugs: pharmacogenetic pharmacodynamic studies

Within the past year, the serotonin transporter promoter, the norepinephrine transporter and the guanine nucleotide binding (G-) protein genes were investigated with respect to treatment response and age factors.

The serotonin and norepinephrine transporter genes

In the pharmacogenetics of mood disorders, the serotonin transporter promoter insertion/deletion variant has been well studied over the past few years [17]. Studies conducted in Europe and the United States [18–21], but not in Asia [22,23], have found an association between the deletion (or ‘short’) allele and nonresponse to antidepressant drugs. Recently, Serretti and co-workers [24•] used an independent European sample of 128 major depressive and 93 bipolar depressed patients to confirm the previous reports, but the lack of association in Asian studies remains to be clarified. The sum of these findings underscores the potential for ethnic or geographic variation to produce apparently conflicting results if such unrecognized sources of population structure are not taken into consideration. In that report [24•], Serretti et al. also found a marginal trend toward poor antidepressant response among individuals who have the A/A genotype of the A218C variant of the tryptophan hydroxylase gene. Nonetheless, tryptophan hydroxylase remains an excellent candidate gene for treatment response in mood disorders because new preclinical data suggested a key role in brain serotonin synthesis [25••].

Yoshida and colleagues [26•] have explored the association of milnacipran, a dual action serotonin/norepinephrine reuptake inhibitor and serotonin transporter polymorphism in 96 Japanese patients. The ‘short’ allele again failed to predict poorer antidepressant drug response. However, some promising data were reported for the norepinephrine transporter gene as the presence of the T allele of the T-182C polymorphism and the presence of the A/A genotype of the G1287A conferred superior efficacy of milnacipran and slower onset of effects, respectively. The study’s major strengths included the use of patients with first episode of depression and minimal co-morbid substance abuse. However, a large sample size study and a head-to-head comparison of Asian and non-Asian samples are needed.

G-protein systems and novel candidate genes

Initial studies of the C825T polymorphisms in exon 10 of the G-protein β₃ gene have revealed an association of the T/T genotype and excellent antidepressant response [27]. Serretti and colleagues [28•] recently replicated this finding in 490 patients. Some novel candidate genes are also emerging from animal models of anxiety/depression, especially in the hypothalamus/pituitary axis and arginine vasopressin systems [29]. Furthermore, immobilization stress appeared to increase tryptophan hydroxylase messenger RNA expression in rat raphe nuclei [30] as well as phospho-calcium/calmodulin-dependent protein kinase II levels in rat hippocampus [31]. These genes may be potential candidates for pharmacogenetic investigations of depression and or stress-induced psychiatric disorders.

Investigations of gene-by-age interaction

In psychopharmacological research today, the age dependence of antidepressant drug benefits appears to be highly controversial [32]. Joyce and colleagues [33••] have investigated antidepressant drug response and its relationship with age factors, G-protein β₃ and the serotonin transporter insertion/deletion polymorphism among 169 patients randomly assigned to nortriptyline or fluoxetine treatment with 59% reporting an adequate response. Among patients under 25 years of age, the drug, but not serotonin transporter promoter genotype, predicted treatment response. Among patients over 25, the promoter genotype, but not the drug, predicted outcome. Reportedly, nonresponse was 60% more likely for individuals with the short allele than for those without. Interestingly, the G-protein β₃ data revealed a drug-by-genotype interaction in patients aged 25 years or under only with the presence of T alleles conferring a poorer response to nortriptyline compared with fluoxetine. Although the authors noted that the findings may be due to differential brain maturation (i.e. differential gene expression), caution should be used to interpret the study because of the sample size and the arbitrary choice of 25 years for age cut-off.

Mood stabilizers: pharmacogenetic pharmacokinetic studies

In the past year, several well established candidate genes with known functional polymorphisms have been examined with respect to their possible contribution to the pharmacokinetics of mood stabilizing drugs. These may be summarized as follows.

G-proteins and glycogen synthase kinase-3

The regulatory effect of G-proteins on signal transduction and intracellular signaling events is by now well recognized in the case of lithium. In a pilot study, Zill and co-workers [34••] examined lithium response and two polymorphisms of the α-subunit of olfactory G-protein encoded on chromosome 18p. No associations were reported in separate and haplotype analyses. Nowadays, multi-locus gene analyses are becoming increasingly popular in pharmacogenetic studies, although the statistical methods are still evolving.

Glycogen synthase kinase 3, an important mediator in neuronal cell death after the withdrawal of trophic support, is a strong candidate gene in mood stabilizer response. In several recent clinical or preclinical studies, the T-50C single nucleotide promoter polymorphism in the glycogen synthase kinase 3 gene was implicated in circadian rhythm mechanisms, age of onset of bipolar disorder [35], as well as therapeutic [36] and antitumor effects of lithium [37]. As a result, pharmacogenetic investigations of G-proteins and glycogen synthase kinase should be further pursued in the coming years.

Gene expression

Mood stabilizers appear to affect gene expression. Recent preclinical studies suggested common biochemical pathways for lithium and valproate, including the extracellular signal-regulated kinase systems [38,39•]. Lithium stimulates antiapoptotic Bcl-2 expression [40] and blocks transcription factor c-Jun expression. Also, Hongisto and colleagues [41••] suggested that lithium probably acted on c-Jun via inhibition of glycogen synthase kinase 3. Post-mortem studies have revealed no differences in the levels of lithium inhibitable enzymes, the inositol mono-phosphatase and the glycogen synthase kinase 3 in normal controls and bipolar patients [42]. A further point of consideration relevant to gene expression studies involves the extent to which genetic material obtained from non-central nervous system (CNS) tissue represents a reasonable proxy for CNS-obtained DNA samples with regard to gene expression. One of the most common examples of this question concerns the linked polymorphic region of the serotonin transporter as derived from peripheral (i.e. leukocytes) versus central (i.e. brain) sources of DNA. In the past year, Hranilovic and colleagues [43•] found no significant differences in serotonin transporter messenger RNA levels and homozygous ‘short’ (i.e. low expressing) or long (i.e. high expressing) genotypes based on samples obtained from the peripheral lymphoblast cell lines. It is unknown whether or not concordance exists in the CNS between this polymorphic locus and its resultant messenger RNA.

Mitochondrial genes

Alterations in energy metabolism occur in mood disorders implicating mitochondrial dysfunction [44,45]. Indeed, mitochondria are important arbiters of cell fate via energy homeostasis and calcium metabolism. The mitochondria have a circular double-stranded DNA molecule suggesting a common ancestry with prokaryotes. Thirty-seven genes are found in human mitochondria, 13 of which encode the enzymes of the electron transport chain.

In a very preliminary study of mitochondrial polymorphisms, Washizuka and co-workers [46•] found a possible role of the NADH : ubiquinone dehydrogenase (mitochondrial complex I) gene variant in lithium response. The 10398A functional polymorphism was overrepresented in lithium responders compared with nonresponders. More studies are needed to dissect the pharmacogenetics of mitochondrial genes in mood disorders.

Additional candidate genes that were investigated in the past year have included the serotonin transporter promoter variant. Earlier, Serretti and co-workers [47] had found a highly significant association between the homozygous short form of the transporter polymorphism and inadequate lithium prophylaxis. More recently, the same group could not confirm this finding in an independent sample given lithium prophylaxis for a period of 3 years [48••].

The genetics of antidepressant-induced mania

Individuals with bipolar diatheses are at risk for antidepressant-induced mania particularly with a family history of bipolar illness, previous antidepressant-induced manias, and past exposure to multiple drug trials [5•]. Unfortunately, the mechanism of antidepressant-induced mania is poorly understood.

In the past year, the pharmacogenetics of antidepressant-induced mania was investigated by Serretti and coworkers [49•]. They used a retrospective analysis of data from 65 patients diagnosed with new onset acute mania who had ‘switched’ to mania while on antidepressant drugs with no mood stabilizer therapy relative to another group of 117 ‘never switched’ patients with bipolar disorder who had never been previously diagnosed with antidepressant-induced mania. The comparison group consisted of 133 randomly recruited patients with major depressive disorder who never had manic symptoms. No group differences were found in the frequencies of the polymorphisms of the serotonin transporter gene in its upstream regulatory region, nor of tryptophan hydroxylase, G-protein β₃ subunit, monoamine oxidase A, catechol-O-methyltransferase, serotonin receptor type 2A, dopamine receptors D2 or D4 gene variants. Important guidelines for antidepressant treatment of bipolar depression may emerge from pharmacogenetic studies of antidepressant-induced mania. This study alongside a 2003 report by Rousseva and colleagues [50] both fail to replicate a previously described positive relationship between antidepressant-induced mania and the serotonin transporter gene promoter region ‘short’ allele [51]. However, increased intraindividual risk for this clinical phenomenon is suggested by observations that antidepressant-associated switches often recur [5•]. Conflicting results may stem partly from the retrospective assessment of antidepressant-associated manias, variability in its operational definition, and heterogeneity across studies in the use of adjunctive mood stabilizers, which could represent an epigenetic factor.

Phenotypic diversity: the roles of genetic and epigenetic mechanisms

The coding sequences (exons) maintain the fidelity of genetic instructions and alterations can produce protein diversity. Conversely, the ‘noncoding’ introns are generally not transcribed because they are silenced by DNA methylation, although exceptions exist in transcriptional or post-transcriptional intron reorganization [52]. In a post-mortem brain tissue study, Kan and co-workers [53•] recently provided some evidence of epigenetic dysfunction among patients with major psychiatric disorders. Transcriptional activity with the production of unusual CNS proteins possibly occurs in epigenetic abnormalities and should be of interest in pharmacogenetic research. Also, RNA interference is now recognized to silence some genes either at the transcriptional level or at the post-transcriptional level [54]. Clearly, novel applications of gene silencing may result in the functional characterization of every gene with enormous pharmacogenetic implications.

Conclusions from the past year

Reviewing pharmacogenetic studies in the past year, genetics appears to be an important factor in inter-individual drug response and toxicity. Genetic factors can inactivate metabolic enzymes with the result that administered substrates for the enzyme including antidepressant drugs may build up in the blood. Some recent pharmacogenetic studies of antidepressant drugs, however, found no significant adverse effects among poor metabolizers, probably due to the drug’s safety profile or polymorphic pharmacodynamic factors.

Recent pharmacogenetic studies in mood disorders have reported generally positive findings. The association studies of the short allele of serotonin transporter promoter gene and poor response to antidepressant drugs was well supported by earlier studies conducted in Europe and the United States, although the studies in Asia suggested just the opposite.

Further research on the pharmacogenetics of serotonin transporter promoter polymorphism is warranted particularly because vulnerability to affective disorders has been associated with the short allele of the serotonin transporter [55••].

In the past year, pharmacogenetic studies of mood stabilizers were notably insightful. Glycogen synthase kinase 3 and mitochondrial genes are emerging as candidates for predicting lithium response. However, valproate, carbamazepine and newer putative mood stabilizers remain relatively understudied in pharmacogenetic laboratories.

Finally, the pharmacogenetic contributors to the adverse reactions of medication in mood disorders are being increasingly recognized. Extensive efforts are needed in this area, however, as exemplified in the case of antidepressant-induced mania where two retrospective studies in the past 2 years have failed to replicate a previously reported positive association.

Future directions

Pharmacogenetics is a promising tool for clinical identification of treatment responders. However, gene identification in pharmacogenetics must still overcome a number of barriers related to phenotype definition, disease neurobiology and cost-effectiveness to make large-scale application possible at the bedside.

Also, it is advantageous that pharmacogenetic laboratories continue to collaborate [56]. In pharmacogenetic studies, the use of ‘hidden’ traits (endophenotypes) such as event-related potentials, brain imaging and neurocognitive profile may improve case definition. For instance, Potkin and co-workers [57] have demonstrated that patients who responded to clozapine treatment were likely to have positron emission tomographic scan changes that highly correlated with dopamine D1 receptor alleles. Therefore, molecular genetic studies of drug response should benefit from full characterization of the neurobiology of psychiatric disorders and the mechanism of action of psychotropic drugs.

Ethnic stratification hampers case-control association studies and could be eliminated with family-based approaches. However, genome wide scans are prohibitively expensive for routine pharmacogenetic studies and have case ascertainment issues to contend with. An alternative strategy to deal with ethnic stratification in association studies is ‘genomic control’ [1•]. The feasibility of implementing large-scale linkage and association studies, in part, will depend on the extent to which strategies are developed for overcoming unrecognized characteristics of the population that could lead to spurious positive findings.

Recently, the contribution of haplotypes to drug response has come to the fore and statistical approaches are evolving. In addition, molecular genetic laboratories are investigating epigenetic mechanisms in phenotypic diversity. Before long, epigenetic studies may be complementing designs that involve genome-wide scanning, case-control approaches, and gene expression studies in psychiatric genetics. Important advances in RNA interference are also under way, with widespread enthusiasm regarding the functional definition of every gene and therapeutic applications [58].

Conclusion

Recent progress in pharmacogenetic studies should provide clinicians with a simple bedside tool for a priori judgment of medication response and adverse events. The challenge of pharmacogenetics in the twenty-first century is to manage the rapidly accumulating data, and examine emerging theories using different frameworks.

Abbreviations

CNS

central nervous system

G-protein

guanine nucleotide-binding protein