Mood and anxiety disorders
Table 1 summarises the therapygenetics studies published to date, where clinically diagnosed anxiety or depression are the disorder phenotype of interest. The field is in its infancy and this is reflected in the small, highly heterogeneous and preliminary nature of the existing literature. In particular, sample sizes have ranged from very small pilot studies (e.g. N = 45,
56
) to somewhat larger investigations (e.g. N = 374,
57
). However, modest sample sizes mean the vast majority of studies are underpowered to detect what we expect to be small genetic effects. Presently, this limits our ability to draw strong conclusions regarding the predictive ability of genetic variants for psychological treatment response.
<p>Table 1</p>
Summary of studies investigating associations between genetic variation and response to psychological therapy in mood and anxiety disorders
Authors
Diagnosis
Sample
Treatment
Results
PTSD = post-traumatic stress disorder; PD ±AG = panic disorder with or without agoraphobia; SAD = social anxiety disorder; DEP = depression; SEP = separation anxiety disorder, GAD = generalized anxiety disorder; ANX = all anxiety disorders; OCD = obsessive compulsive disorder.
I
S allele group defined as SASA, SALA, SALG, SGLG, LALG, LGLG and L allele group defined as LALA.
ii Long, high activity risk allele group defined as 3.5, 4 and 5 copy repeat allele males, and 2/4, 3/4, 3.5/4, 4/4, and 4/5 copy repeat allele females. Short, low activity allele group defined as 2 or 3 copy repeat allele males and 2/2, 2/3 and 3/3 females.
5HTTLPR/rs25531
Wang et al.
58
PTSD
N = 35
12wk prolonged exposure therapy or 12 wk Escitalopram (N = 20)
No significant association with treatment response in exposure therapy group
Bryant et al.
56
PTSD
N = 45, Caucasian,
8 × 90 min weekly individual CBT
No significant association at post-treatment. At 6mth follow-up, higher % of S
i
allele than L allele carriers met PTSD criteria and had significantly higher symptom scores
M
age = ~ 43 yrs, 33% female
Lonsdorf et al.
59
PD ± AG
N = 69, Caucasian,
10 × 2 hr weekly group (N = 38) or internet-delivered (N = 31) CBT
No significant association with treatment response
M
age = ~ 35 yrs, 62% female
Furmark et al.
60
SAD
N = 204, M
age = 38 yrs,
9 wk internet delivered CBT or waitlist control
No significant association with treatment response
60% female
Kohen et al.
61
DEP (post- stroke)
N = 61, mixed ethnicity
9 × positive problem solving plus antidepressant vs. usual care plus antidepressant
SS and SL carriers had a significantly greater mean percentage reduction in depression ratings and more likely to be in remission at 9-week follow-up than those in the control group
Sakolsky et al.
62
SEP; GAD; SAD
N = 211, Caucasian,
Sertraline, 14 sessions of CBT, combination therapy or 12 wk placebo
No significant association with treatment response
7-17 yrs,
Eley et al.
10
ANX
N = 359, Caucasian,
10-12 session group or individual CBT or guided self-help
No significant association at post treatment. At follow-up, higher % of SS genotype carriers free of anxiety diagnoses than SL/LL genotype carriers. SS genotype carriers had significantly greater reduction in symptom severity scores
M
age = 9.44 yrs, 49% female
Hedman et al.
63
SAD
N = 126, 98% Caucasian, M
age = ~ 35 yrs, 36% female
15 × 2.5 hr weekly group (N = 62) or internet-delivered (N = 64) CBT
No significant association with treatment response
Bockting et al.
64
Recurrent DEP
N = 180, Caucasian,
Brief CBT vs. treatment as usual
No significant association with time to recurrence
M
age = 45 yrs, 74% female
STin2 VNTR
Kohen et al.
61
DEP (post- stroke)
N = 64, mixed ethnicity
9 × positive problem solving plus antidepressant vs. usual care plus antidepressant
9/12 and 12/12 genotype carriers in intervention group had a significantly greater mean percentage reduction in depression scores and greater likelihood of remission at 9 wk follow-up than controls
Sakolsky et al.
62
SEP; GAD; SAD
N = 211, Caucasian,
Sertraline, 14 sessions of CBT, combination therapy or 12 wk placebo
At 12-week assessment STin2 12-copy variant carriers showed significantly greater improvement.
7-17 yrs
HTR2A rs7997012
Kotte et al.
65
DEP
N = 58, 100% male
16-wk group CBT
G allele predicted significantly larger reduction in BDI scores across treatment compared to A allele carriers
TPH2 G-703T
Furmark et al.
60
SAD
N = 204, M
age = 38 yrs,
9 wk internet delivered CBT or waitlist control
In the CBT group, a better treatment response was observed in TPH2 GG homozygotes relative to T allele carriers
60% female
MAOA-u VNTR
Reif et al.
66
PD + AG
N = 288, Caucasian
12 × twice weekly CBT
Carriers of the long, higher activity alleleii had significantly worse treatment outcome, elevated heart rate, greater fear and panic attacks during a behavioral avoidance task and failure to habituate during repetitive exposure
COMT val158met
Lonsdorf et al.
59
PD ± AG
N = 69, Caucasian,
10 × 2 hr weekly group (N = 38) or internet-delivered (N = 31) CBT
No significant effect of COMTval158met genotype on change in anxiety or depression scores across cognitive modules (weeks 1-3). met/met genotype carriers had significantly smaller reduction in anxiety scores across exposure modules (weeks 4-9) compared to val-carriers
M
age = ~ 35 yrs, 62% female
Hedman et al.
63
SAD
N = 126, 98% Caucasian, M
age = ~ 35 yrs, 36% female
15 × 2.5 hr weekly group (N = 62) or internet-delivered (N = 64) CBT
No significant association with treatment response
NGF rs6330
Lester et al.
57
ANX
N =374, Caucasian,
10-12 session group or individual CBT or guided self-help
No significant association with treatment response at post treatment. At follow-up, children with one or more copies of T allele of NGF rs6330 were significantly more likely to be free of anxiety diagnosis
M
age = 9.46 yrs, 49% female
BDNF val66met; rs7934165; rs1519480; rs11030104
Sakolsky et al.
67
SEP; GAD; SAD
N = 211, Caucasian,
Sertraline, 14 sessions of CBT, combination therapy or 12 wk placebo
No significant association with treatment response
7-17 yrs
Lester et al.
57
ANX
N = 374, Caucasian,
10-12 session group or individual CBT or guided self-help
No significant association with treatment response
M
age = 9.46 yrs, 49% female
Hedman et al.
63
SAD
N = 126, 98% Caucasian, M
age = ~ 35 yrs, 36% female
15 × 2.5 hr weekly group (N = 62) or internet-delivered (N = 64) CBT
No significant association with treatment response
Fullana et al.
68
OCD
N = 106, Caucasian,
20 × 45 min weekly exposure based CBT plus SSRI
Met allele carriers significantly less likely to respond to treatment than non-met allele carriers. Genotype predicted response only and not change in severity scores
M
age = 33 yrs, 50% female
GRIN2B rs1019385
Sakolsky et al.
67
SEP; GAD; SAD
N = 211, Caucasian,
Sertraline, 14 sessions of CBT, combination therapy or 12 wk placebo
No significant association with treatment response
7-17 yrs
GRIK4 rs1954787
Sakolsky et al.
67
SEP; GAD; SAD
N = 211, Caucasian,
Sertraline, 14 sessions of CBT, combination therapy or 12 wk placebo
Significant association with treatment response at 12 week assessment
7-17 yrs
Studies thus far have relied on a candidate gene approach, testing the effect of a single marker or limited number of genetic polymorphisms within a gene on psychological treatment response. The most widely assessed polymorphism has been the 5HTTLPR. In the following section, we briefly outline the rationale for each genetic marker being a plausible candidate for involvement in psychological treatment response. We then review the findings of therapygenetics studies to date, grouped by the markers that have been considered. In light of the infancy of the therapygenetics field, we then draw attention to several methodological issues for consideration in future research.
Serotonin transporter linked polymorphic region (5HTTLPR)
Serotonergic neurotransmission is associated with anxiety and mood disorders and is implicated in the treatment of both types of disorder. By far the most widely studied polymorphism has been the 5HTTLPR, which is a 43 bp insertion deletion functional polymorphism in the 5’ promoter region of the SLC6A4 gene, resulting in short and long alleles, which differ in 5-HTT expression and function. The short allele variant is associated with approximately 50% less 5-HTT expression and function leading to higher concentrations of serotonin in the synaptic cleft
69
. The 5HTTLPR is often studied in conjunction with the A/G single nucleotide polymorphism (SNP) rs25531, which has been shown to differentially impact the L allele function. The minor G allele is commonly in phase with the long allele of the 5HTTLPR with the LG allele reported to have transcriptional activity similar to the S allele
70
. Participants are often classified on the basis of presumed efficacy of serotonergic neurotransmission into low (SS, SLG, or LGLG), intermediate (SLA or LALG) and high (LALA).
The evidence for an association between the 5HTTLPR and mood and anxiety disorders has been somewhat mixed. The low expression allele has most often been associated with greater affective and anxiety-related traits (e.g.
71
72
73
) and related intermediate phenotypes (e.g. attentional bias to emotional stimuli
74
; fear conditioning
75
76
; amygdala reactivity
77
). Consistent with the differential susceptibility hypothesis, a recent meta-analysis showed that the low expression allele was associated with poorer outcomes following stress (
25
, but see also
27
for nonsignificant effects) but also appears to be associated with better outcomes under low stress or positive environments
36
. This led to the hypothesis that the low expression S allele would be associated with enhanced response to psychological therapies
10
.
Nine studies have tested for an association between the 5HTTLPR and response to psychological therapy with three studies reporting evidence of a significant association. With respect to the five negative studies, there was no significant association between the 5HTTLPR and relapse rates following a course of CBT in remitted recurrently depressed adults (N = 180,
64
); improvement in symptoms in response to internet based CBT for socially anxious adults (N = 204,
60
; N = 126,
63
); improvement in anxiety symptom scores in response to exposure based CBT in panic disorder patients (N = 69,
59
) and improvement in symptoms in response to multimodal treatment in child anxiety disorders (N = 211,
62
). All of these studies have small samples and are therefore underpowered to detect what are very likely to be small genetic effect sizes.
With regard to positive studies, anxiety-disordered children (N = 359) carrying the low expression (SS) genotype were significantly more likely than the intermediate (SL)/high (LL) expression genotype carriers to be free of their anxiety disorder diagnosis (78.4% vs. 58.4%) and to have a larger reduction in symptom severity scores after a course of CBT or guided self-help
10
. In the second positive study (N = 61), the low expression (SS) genotype was associated with a greater percentage change in depression scores in clinically depressed older adults after ischemic stroke and in response to a brief problem solving plus antidepressant intervention
61
. These findings appear consistent with a differential susceptibility explanation in which the same genetic variant, the low expression S allele, which in prior studies has been associated with increased risk for anxiety and depression, is also associated with the most improvement in response to a positive psychological treatment environment.
In contrast to the direction of effect observed in the two previous studies
10
61
, a third study found that the low expression genotype was associated with more severe PTSD symptoms and therefore fewer treatment gains at 6-months in patients undergoing exposure-based CBT for PTSD (N = 45;
56
). In this study, the authors suggest that the low expression allele may limit the capacity to gain control over the anxiety elicited through CBT as a consequence of greater amygdala reactivity, stronger fear conditioning and deficits in extinction learning
56
. These findings are also consistent with past (albeit mixed) research reporting an association between the low expression genotype and poorer response to SSRIs
78
. In a further study, PTSD patients received a 12-week trial of citalopram (N = 20) or prolonged exposure therapy (N = 15,
58
). The low expression SS genotype was again associated with a poorer treatment response in both treatment groups with the effect only attaining statistical significance in the pharmacotherapy group
58
. The contradictory findings observed for 5HTTLPR may reflect not only small sample sizes and varying clinical phenotypes, but also the role of medication. Further adequately powered studies are required to clarify the association between 5HTTLPR and response to psychological treatment.
Serotonin transporter intron 2 variable number tandem repeat (STin2 VNTR)
The STin2 VNTR of the 5-HTT gene has multiple repeated copies of a 16-17bp element
79
. Nine (STin2.9), ten (STin2.10) and 12 (STin2.12) repeat alleles have been identified. Different copy number variations of the VNTR have been shown to have differential regulatory effects on transcription
80
81
. The short variant (9 or 10 repeats) is associated with lower levels of 5-HTT mRNA and has been linked with unipolar depression
82
and anxiety
83
. However, others have reported an association between STin2.12 alleles and neuroticism
84
and anxiety disorders
85
.
Variation in STin2 has been examined in two therapygenetics studies. In the first, post-stroke depressed patients with 9/12 or 12/12 repeat alleles showed a larger percentage reduction in depression scores to a problem-solving intervention (N = 64,
61
). Similar significant associations with treatment response were reported in a second study of children with a diagnosis of separation anxiety disorder, generalized anxiety disorder or social phobia receiving either sertraline medication, CBT or a combination therapy (N = 211,
62
). Significantly greater improvement defined by Clinical Global Impression-Improvement scores of very much or much improved at post-treatment was observed in carriers of the 12-copy variant after controlling for age and treatment condition. The findings from these two studies are in contrast to those reported in some studies for response to SSRIs. For example, poorer remission rates from major depression were reported in carriers of the 12/12 genotype in the Sequenced Treatment Alternatives to Relieve Depression study (STAR*D
86
).
5-hydroxytryptamine (serotonin) receptor 2A gene (HTR2A)
The HTR2A gene encodes for 5-hydroxytryptamine 2A receptors, which have been implicated in animal and human models of depression and are thought to play an important role in antidepressant drug action
87
. Several polymorphisms in this gene have been studied in relation to mood and anxiety disorders although associations have tended to be small and findings have been mixed
88
89
. The most widely studied polymorphisms have been A-1438G (rs6311); C102T (rs6313) and His452Tyr (rs6314). There is some evidence that genetic variants in the HTR2A gene may predict differences in response to SSRI treatment
87
90
. For example, a further SNP, rs7997012, located in intron 2 of the HTR2A gene has been associated with treatment response to antidepressants in the STAR*D sample in both a discovery and replication sample. Being homozygote for the A allele was associated with a 16-18% reduction in risk of non-response compared to GG homozygote carriers
87
90
.
A single therapygenetics study has tested for an association between the HTR2A rs7997012 polymorphism and response to a 16-week course of CBT in a sample of 58 male veterans experiencing unipolar depression
65
. In contrast to the STAR*D studies investigating associations with antidepressant response, preliminary findings showed that carriers of the G allele had a significantly larger reduction in depression symptom scores in response to CBT treatment than AA homozygotes.
Tryptophan hydroxylase 2 gene (TPH2)
The neuronal isoform of the TPH2 gene is expressed exclusively within the nervous system and codes for the rate-limiting enzyme for 5-HT biosynthesis in midbrain and serotonergic neurons
91
. TPH2 has previously been associated with depression
92
and anxiety disorders
93
, although findings have not always replicated. The SNP G-703T (rs4570625) is located in the promoter region of the TPH2 gene. It is thought to play a functional role in regulating 5HT signaling and has a significant effect on gene expression with the T allele associated with reduced transcriptional activity
94
and increased amygdala reactivity to emotional stimuli
95
96
.
A single therapygenetics study has tested for an association between the TPH2 G-703T polymorphism and response to a 9-week internet-based CBT intervention for social anxiety disorder (N = 204,
60
). A significantly greater improvement in self-report social anxiety symptom scores was observed in GG homozygote individuals compared to T allele carriers at post-treatment. Interestingly, in an earlier paper the GG genotype significantly predicted greater improvement in social anxiety symptoms in response to a placebo drug condition
95
. Statistical analyses were consistent with the improved treatment response in GG genotype carriers being mediated by its effect on attenuating amygdala activity. This suggests a possible biological mechanism linking genetically controlled serotonergic modulation of amygdala reactivity and treatment response. One might anticipate a similar mediation mechanism linking TPH2 polymorphism and CBT response given that placebo treatments are thought to act on essentially the same neural pathways as those influenced by active treatments
97
98
. This association with placebo response also raises the interesting possibility that the G-703T GG genotype may not be specifically associated with therapeutic responsiveness but rather a more general tendency to recover irrespective of treatment. This can be best clarified further by the inclusion of a no treatment control group alongside active and placebo treatment conditions.
Monoamine oxidase-A variable number tandem repeat (MAOA-u VNTR)
MAOA is a logical candidate gene for involvement in psychiatric disorders due to its role in metabolizing serotonin and norepinephrine
99
. A 30 bp VNTR polymorphism called MAOA-uVNTR in the promoter region of the gene has recently been studied with regard to psychological treatment response. This VNTR has a long allele (3.5, 4 and 5 repeats) and short allele (2 and 3 repeats). The long alleles demonstrate increased transcriptional activity
100
101
with levels of 5-hydrocyindoleacetic acid (5-HIAA), the main metabolite of serotonin influenced by this polymorphism
102
. The MAOA-uVNTR is often referred to as the warrior-worrier gene
103
due to short-allele repeats being associated with impulsive-aggressive behavior in males
21
104
105
. Long repeat alleles have been associated with panic disorder in females
100
106
and heightened sensitivity to aversive experiences
107
.
Only one therapygenetics study has explored variation in the MAOA-u VNTR
66
. In a sample of 288 patients diagnosed with panic disorder with agoraphobia and receiving exposure-based CBT, carriers of the long repeat higher activity risk allele had a significantly worse treatment response (46% responders) than homozygote carriers of the short repeat allele (67% responders)
66
. In response to a behavioural avoidance test involving entering a small dark chamber, L allele carriers experienced significantly higher heart-rate during anticipation of the task, as well as during exposure and recovery phases, more intense anxiety during exposure, a greater frequency of panic attacks and a failure to habituate. The authors argue that the poorer outcome seen in L allele carriers is consistent with a possible effect of MAOA on the efficiency of extinction learning via decreased noradrenergic availability
66
. This would be consistent with evidence that adrenergic enhancers facilitate exposure based CBT for anxiety disorders (e.g.
108
).
Catechol-O-methyltransferase gene (COMT)
Catechol-O-methyltransferase is a methylation enzyme encoded by the COMT gene that acts to degrade monoaminergic neurotransmitters (e.g. dopamine, epinephrine, norepinephrine)
109
. An A/G SNP, COMTval158met (rs4680) codes for a valine (val) to methionine (met) exchange at position 158. The val allele is associated with a three to four-fold increase in functional activity of the methylation enzyme compared to the met allele, which impacts on the degradation of dopamine and leads to an excess of synaptic dopamine in met carriers
110
111
. Associations have been reported between the COMTval158met polymorphism and fear extinction processes (for a review see
19
) and psychopathology, although the direction of effects has not proven entirely consistent. While COMT genotype was not found to be associated with indices of fear conditioning
75
112
, there was some evidence that the met allele was associated with stronger and extinction-resistant fear memories using a 24-hour delayed extinction paradigm
75
. However, there was no significant association with extinction learning in a further study using an immediate extinction procedure
112
. The val/val genotype has also been associated with enhanced fear reacquisition under certain experimental conditions
76
. With regard to psychopathology, the met allele has also been associated with severity of PTSD
113
, and phobic avoidance and panic attacks in adolescent females
114
. However, the val allele has also been associated with panic disorder
115
.
Two therapygenetics studies have tested for an association between the COMT gene and psychological treatment response. Building on experimental studies of fear conditioning and extinction, the COMTval158met was considered for association with efficacy of group or internet-delivered exposure-based CBT in a sample of 69 panic disorder patients
59
. No significant association with genotype was observed for change in depression symptom scores across treatment or for change in anxiety symptom scores across the cognitive treatment modules, which occurred early in treatment. However, met/met genotype carriers reported a significantly smaller reduction in anxiety symptoms across the exposure treatment modules compared to val carriers. This effect remained significant even after controlling for demographic and clinical variables and 5HTTLPR/rs25531 genotype. The authors suggest that impaired top-down cognitive control over emotional reactions and a related failure to extinguish fear reactions may represent one pathway via which the met/met group gains less benefit from psychological therapy
59
. In the only other therapygenetics study to investigate the COMTval158met, no significant association with response to group or internet delivered CBT was observed in a sample of 126 patients diagnosed with social anxiety disorder
63
. It is possible that differences in disorder phenotype (e.g. PTSD vs. SAD) may account for the discrepancy in findings or that COMT genotype may be of greater relevance when treatment modality is predominantly exposure based rather than cognitive based. Further research is needed to test these hypotheses.
Nerve growth factor (NGF)
A number of therapygenetics studies have focused on candidate genes that are purportedly implicated in the pathophysiology of learning with the rationale being that CBT, and exposure based CBT especially rely on the basic principles of extinction learning
116
. NGF is a neurotrophic gene that is thought to have a key role in neuronal survival, activity-dependent neuroplasticity and learning, is implicated in the orchestration of response to stress and is widely expressed in limbic areas of the central nervous system involved in mood and cognition
117
. In NGF, rs6330 has been investigated as a predictor of response to psychological therapy. This non-synonymous SNP produces an alanine to valine substitution at amino acid position 35 and is thought to have a possible functional role on intracellular processing and secretion of NGF
118
. Allelic variation at this locus has shown associations with anxiety-related traits and affective disorders
119
120
.
Variation in NGF rs6330 has been examined in a single therapygenetics study to date. Likelihood of remission increased with each extra T allele of the NGF rs6330 marker in a sample of anxiety-disordered children after a course of CBT or guided self-help (N = 384,
57
). Fifty-three percent of CC genotype carriers were free of their primary anxiety diagnosis compared to 63.5% and 76.7% for CT and TT genotypes respectively. However, the effect of genotype was observed only at follow-up assessment and not immediately post-treatment. One possibility is that the NGF T allele may influence capacity for continued benefit from CBT. The authors suggest that children with the T allele may show subtle differences in neurotrophic signalling that influence the extent to which environmental influences, in this instance therapeutic interventions bring about neuroplastic modifications, which in turn modulate mood
57
.
Brain-derived neurotrophic factor gene (BDNF)
BDNF, like NGF is also a neurotrophic gene. BDNF secretion is activity dependent with decreases associated with stress, mood disorders and anxiety-related behaviour and increases with antidepressant medication
121
. The most frequently studied polymorphism in the BDNF gene has been the functional val66met (rs6265) polymorphism. In val66met, the more common G allele encodes for valine (val) with the A allele encoding for methionine (met). The met allele is associated with diminished activity-dependent secretion of BDNF
122
123
, abnormalities in brain structure and function in limbic regions
124
125
126
. The val66met polymorphism has also been studied in conjunction with associative fear learning, fear generalisation and extinction processes, albeit with mixed findings (for a review see
19
).
Four therapygenetics studies have tested for an association between the BDNF gene and psychological treatment response. The most frequently studied polymorphism has been the val66met (rs6265) polymorphism, however other polymorphisms within this gene have also received attention (rs7934165, rs1519480, rs11030104). Three studies failed to find any association between BDNF variants and response to psychotherapy for child anxiety disorders (N = 374,
57
); response to multimodal treatment for child anxiety disorders (N = 211,
67
) or adult social anxiety disorder treated with group or internet-based CBT (N = 126,
63
). However, a single study reported an association between the val66met polymorphism and response to exposure-based CBT in a sample of 106 OCD patients who had previously shown partial or non-response to a 12-week pharmacological trial
68
. Thirty-six percent of met allele carriers compared to 60% of val carriers responded to treatment defined as a 35% or greater decrease in Yale-Brown Obsessive Compulsive Scale scores. Of note, all participants also received concurrent SSRI treatment in this study and the treatment modality was predominantly exposure based rather than cognitive based. While warranting further investigation, these methodological differences may in part account for the discrepancy with other studies.
Glutamate receptor, ionotropic N-methyl D-asparate 2B gene (GRIN2B)
The GRIN2B gene encodes the NR2 subunit of the N-methyl-D-aspartate (NMDA) glutamate receptors, which are important for fear acquisition and fear consolidation
127
and learning and memory more broadly
128
. The SNP, rs1019385 located in NMDA 2 beta subunit is a promoter region variant in which the G allele leads to reduced transcriptional activity which in turn may affect glutamate neurotransmission
129
. GRIN2B has been associated with childhood OCD
130
, and related biological traits including reduced anterior cingulate glutamatergic concentration
131
. However, in the only therapygenetics study to type this marker there was no significant association between GRIN2B rs1019385 and response to CBT, pharmacotherapy or combined treatment defined by improvement scores in a sample of children with anxiety diagnoses (N = 213,
67
). Note, however, that no children with a primary OCD diagnosis were included in this sample.
Glutamate receptor, ionotropic kainite 4 gene (GRIK4)
The GRIK4 gene encodes for a kainate preferring ionotropic glutamate receptor subtype involved in modulating neurotransmitter release and excitatory neurotransmission. The SNP, rs1954787 located in the 3’ end of the first intron is not thought to alter protein sequence but may have functional relevance via regulation of gene expression
90
. A recent study reported an association between the C allele of this marker and improved treatment response to antidepressants in the STAR*D cohort
90
but this effect did not replicate in the Munich Antidepressant Response Signature sample after correction for the number of SNPs tested
132
. This marker has been investigated in one therapygenetics study to date. A significant association between the rs1954787 SNP and treatment response defined as a rating of very much or much improved on the Clinician Global Impression – Improvement scale was observed after controlling for age, treatment condition and other genetic polymorphisms investigated in a sample of children with anxiety diagnoses (N = 213,
67
). However, it is unclear whether GRIK4 genotype interacts with treatment type as some children received medication, others CBT and some a combined therapy.
Methodological considerations
A small number of studies have provided preliminary but provocative evidence in support of an association between genetic polymorphisms and response to psychological therapy. Independent replication of these early findings in adequately powered samples should be a priority. Several other studies have failed to find any evidence for a significant association or have produced conflicting results. In this section we outline some of the methodological issues and directions for future research, which we believe should be considered in future therapygenetics research.
Future directions
At present, genetic predictors do not have sufficient predictive power to warrant their use as a clinical biomarker. Moving toward analytic methods that aggregate across multiple polymorphisms and/or genes into a single parameter in combination with machine learning methods may be the most powerful means of achieving clinically significant prediction and building toward a stratified medicine approach for mood and anxiety disorders
134
139
. Combining genetic information with clinical and demographic predictors
140
or perhaps with neuroimaging biomarkers
141
within a single algorithm may also prove to be a particularly powerful approach. More broadly, the inclusion of clinical covariates within models designed to test association between specific SNPs and psychological treatment outcome is important. Clinical predictors such as baseline symptom severity, the presence of comorbid disorders and familial history of illness are themselves likely to be under genetic influence. It is therefore important to show that any predictive effect of a genetic variant exists above and beyond the effect of any significant clinical predictors.
For an individual patient, the key clinical decision is not whether to treat or not but instead which treatment to select to maximize the chance of recovery
7
. In these circumstances, using genetic markers to predict response to a particular treatment is of limited clinical utility. Instead, as a medium term aim, therapygenetics studies need to move away from predicting response to a single psychological treatment and instead should aim to identify differential predictors of response to alternative treatment modalities. For example being able to estimate on the basis of genetic information, the relative likelihood of response to pharmacotherapy versus psychotherapy would be of considerably greater clinical value than identifying that a particular patient is more or less likely to respond positively to SSRIs. However, it may not be realistic to expect to find a genetic variant associated with response to pharmacological treatment per se given the different biological pathways via which different classes of drugs exert their influence. One way to progress might be to try and identify genetic variants that differentially predict response to a specific psychological therapy e.g. CBT compared to response to a specific class of drugs e.g. SSRIs or MAOIs. Unreplicated research suggests an interesting potential double dissociation between 5HTTLPR genotype and response to SSRIs and CBT treatment modalities. In adult depression, response to SSRIs has been shown to better in LL carriers of the 5HTTLPR
47
, while some therapygenetics studies have observed better response to CBT in SS carriers
10
61
. However, it is important to note that several studies failed to find any association between 5HTTLPR and response to psychological therapy
58
59
60
62
63
64
, while one study reported that SS carriers did in fact have fewer positive treatment gains
56
, which would argue against a dissociation relative to pharmacological findings. Studies are required in which patients are randomly assigned to pharmacotherapy or CBT to fully test this hypothesis. One possibility is that interactions with environmental variables and differing etiological pathways into and out of mood and anxiety disorders may partly account for these differential findings.
To date, no studies have explored gene-environment interplay (e.g. with SLEs and early developmental experiences such as maltreatment) in predicting treatment outcome for psychological therapies. This would require large samples due to testing mutually dependent influences of multiple factors and accurate and in-depth characterization of environmental exposure history, ideally both negative and positive environmental exposures, treatment outcomes and the collection of genetic material
7
. Despite these methodological challenges, we believe that this would be a promising line of research, although it is unclear at this stage whether G×E interactions will have sufficient predictive value to guide personalized treatment decisions. Two recent studies identified that carriers of the SS genotype of the 5HTTLPR responded poorly to SSRI treatment but this association was only observed in those who experienced a stressful life event prior to treatment
142
143
. These findings are consistent with a combination of stress and greater genetic risk to stress predicting poorer response. However, this finding did not replicate in a third study
144
. While as yet untested and in contrast to findings from antidepressant studies, one possibility is that increased genetic vulnerability to stress (e.g. presence of the 5HTTLPR SS genotype) in combination with the presence of a stressful life event may in fact be associated with a better response to psychological therapies. In the antidepressant literature, significant G×E interactions have been reported between exposure to SLEs and SNPs in FKBP5 (rs1360780) and CRHR1 (rs110402) (see
18
for a review). However, here a combination of exposure to life stress and presence of the putative risk allele predicted a better treatment response. As suggested by others, disentangling the etiological pathways to mood and anxiety disorders may make it possible to differentially predict which individuals are at an increased likelihood of responding to pharmacological or psychotherapeutic interventions
18
.