OUP user menu

DNA sequence variants of the FKBP5 gene are associated with unipolar depression

Astrid Zobel, Anna Schuhmacher, Frank Jessen, Susanne Höfels, Olrik von Widdern, Martin Metten, Ute Pfeiffer, Claudia Hanses, Tim Becker, Marcella Rietschel, Lukas Scheef, Wolfgang Block, Hans H. Schild, Wolfgang Maier, Sibylle G. Schwab
DOI: http://dx.doi.org/10.1017/S1461145709991155 649-660 First published online: 1 June 2010


FKBP5 is a glucocorticoid receptor-regulating co-chaperone of hsp-90 and, therefore, is suggested to play a role in the regulation of the hypothalamic–pituitary–adrenocortical system and the pathophysiology of depression. Previously, three studies identified single nucleotide polymorphisms (SNPs) in the FKBP5 gene associated with response to antidepressants, and one study found an association with diagnosis of depression. We selected five markers from the region of interest. A case-control sample comprising 268 German in-patients with recurrent unipolar depression, and 284 German controls recruited from the general population were available. Association of the selected FKBP5 sequence variants with clinical depression were analysed. In addition, we explored association with treatment response by (a) the Hamilton Depression Rating Scale and (b) the dexamethasone/corticotrophin-releasing hormone (Dex/CRH) test, as well as association with hippocampal volumes in a subpopulation of 110 patients. For three of the five investigated SNPs we were able to show association with the diagnosis of depression. In the subpopulation of 110 patients, diagnosis-related alleles were also associated with the reduction of cortisol secretion in the Dex/CRH test during a 4-wk treatment period, while psychopathological changes were not associated. Furthermore, diagnosis-related alleles were associated with reduction of the hippocampal volume. This study extends the replicated association of a promoter SNP with antidepressant response on a biological level by demonstrating normalization of the cortisol response under Dex/CRH stimulation during treatment. Furthermore, several of the investigated SNPs were associated with the disease status and the intermediate phenotype of hippocampal volume.

Key words
  • Depression-related phenotypes
  • FKBP5
  • hippocampus
  • HPA
  • unipolar depression


FKBP5 belongs to the family of immunophilin class of proteins. It is a co-chaperone of hsp-90 and, as such, is involved in the regulation of the hypothalamic–pituitary–adrenal (HPA) system through adaptive changes in the glucocorticoid receptor (GR). The HPA system is regarded as a playing a major role in response to stress (Park et al. 2007). It is dysregulated in stress-associated disorders as during depressive episodes, and there is strong evidence that normalization of HPA activity is a prerequisite for sustaining remission (Holsboer-Trachsler et al. 1991; Zobel et al. 2001). Consequently, factors involved in regulating the HPA system are also thought to play a major role in development and relief of depressive episodes. Two lines of evidence were used to implicate FKBP5 in response to antidepressants: (a) consistent allelic association with response to antidepressant treatment in two independent samples, and (b) functional effects on gene expression for specific implicated alleles (Binder et al. 2004). In a Spanish sample of depressive patients, association of one of these single nucleotide polymorphisms (SNPs) with treatment response using selective serotonin reuptake inhibitors (SSRIs) was shown, but with the opposite allele (Papiol et al. 2007). More recently, association of FKBP5 variants with treatment response was replicated in an independent sample of German depressed patients (Kirchheiner et al. 2008).

Binder and colleagues also demonstrated that specific alleles in the FKBP5 gene contribute to the inter-individual variation of recurrence of depressive episodes (Binder et al. 2004). However, Binder and colleagues were unable to support an association with the diagnosis of affective disorders in 294 hospitalized in-patients compared to 339 matched controls after controlling for multiple testing among 12 selected markers and their combinations.

Gawlik et al. (2006) selected three markers from the Binder study and investigated association with major affective disorders (more than 70% of the cases with bipolar disorder) in a German sample. Gawlik and colleagues were unable to show association, even though one rare haplotype displayed a strong difference between both groups and another haplotype was associated with longer duration of the disorder.

A more recent analysis of three SNPs selected from the Binder et al. (2004) study was conducted in the STAR*D sample comprising 1809 cases with unipolar depression and 739 DNA controls (from a cell repository) (Lekman et al. 2008). Significant group differences in the frequency distribution of genotypes were detected for two of the selected markers yet the effect sizes were very modest. One of these variants revealed the same association with clinical response under treatment with citalopram as reported by Binder et al. (2004) This relationship is primarily evident on the level of remission after 14 wk. The reported association was most prominent in the white non-Hispanic population.

Genetic associations to other disorders and psychosocial conditions were also reported: to post-traumatic stress disorder by Binder et al. (2008) and to psychosocial stress in healthy volunteers by Ising et al. (2008). Recently, Willour et al. (2009) reported a family-based association of FKBP5 variants with bipolar disorder, thereby corroborating the view that FKBP5 impacts on the risk for affective disorders.

The present study exclusively focuses on recurrent unipolar depression, and explores the genetic association with variants in the FKBP5 gene. For this purpose we selected five markers from the region revealing association with treatment response in the study by Binder et al. (2004) (Fig. 1). It has been suggested that heritable biological correlates of the psychopathologically defined diagnoses may be a more suitable phenotype for genetic association studies. Those associated phenotypes have also been identified for unipolar depression (Zobel & Maier, 2004, 2009). The hippocampal volume is of particular interest in this respect as (a) a reduction in volume is consistently found to be associated with unipolar depression (McKinnon et al. 2009; Savitz & Drevets, 2009), (b) it is a heritable trait (Frodl et al. 2008; Savitz & Drevets, 2009), and (c) many known depression-related pathophysiological pathways are located in or linked to the hippocampus (Balu & Lucki, 2009; Savitz & Drevets, 2009). The hippocampal volume was therefore included in our analysis.

Fig. 1

Genomic structure of the FKBP5 gene.

Another associated phenotype of depression is HPA activity, which can be most specifically measured by the combined dexamethasone/corticotrophin-releasing hormone (Dex/CRH) test (Ising et al. 2007). Therefore, association of FKBP5 markers with treatment response defined by using psychopathology measures (Hamilton Rating Scale for Depression; HAMD; Hamilton, 1960) and HPA activity (Dex/CRH test) is also explored in a subsample of the investigated cases.



Two hundred and sixty-eight consecutively admitted in-patients with recurrent unipolar depression (according to DSM-IV criteria) aged 18–60 yr were recruited (Department of Psychiatry at the University of Bonn). A structured clinical interview (SCID; Wittchen et al. 1997) was administered by experienced and well trained clinicians. Reliability for the diagnosis of recurrent major depression was 0.95. Diagnosis was made using a best-estimate procedure based on the clinical diagnosis and the SCID evaluation by an independent senior consultant. We aimed at a diagnostically homogeneous sample; thus, exclusion criteria were psychotic disorders, bipolar I and II disorder, antisocial or borderline personality disorder or substance dependence and abuse, as well as other serious medical conditions.

Two hundred and eighty-four controls were systematically recruited from the general population living in the same region (Bonn) using the register of inhabitants. The selection was stratified by age and gender in order to match to the patient group (Table 1).

View this table:
Table 1

All patients and controls were German Caucasians. All patients and controls gave written informed consent after the study protocol was fully explained. The study was approved by the Medical Ethical Committee of the University of Bonn.

Treatment (subsample of 110 patients)

One hundred and fifty-six patients of the original sample received a standardized antidepressant routine treatment after tapering off the previous antidepressant medication within a period of up to 4 d immediately after admission. A wash-out interval was not planned in order to guarantee the naturalistic nature of the treatment with citalopram. Treatment followed an internal quality assurance programme, conducted according to evidence-based clinical guidelines. The following procedure was applied:

  1. citalopram (precondition: no severe agitation, no previous treatment resistance to citalopram),

  2. stepwise increase of dosage to 20, 30 or 40 mg citalopram from day 1 to day 4 (dosage by clinical decision),

  3. fixed dosage day 4 to day 36,

  4. exclusively lorazepam up to 3 mg (fixed dosage from day 4 to day 36) was allowed as additional psychotropic medication.

Forty-six patients could not follow the complete study protocol because of (multiple selections were possible) deterioration of clinical symptomatology (n=3), side-effects of citalopram medication (n=9) and/or non-compliance, rejection of the follow-up examination or withdrawal of consent (n=37). Seven patients dropped out during the first 4 d (increase of citalopram dosage and tapering off the previous medication), five patients dropped out between days 4 and 8 (t1), and 34 subsequently, mainly because of rejection of the follow-up examination or withdrawal of consent. As such, we analysed the completed treatment results of 110 patients (32.7% males, 67.3% females). The mean age was 46.5±12.4 yr (mean±s.d.) (males: 47.3±12.9 yr; females: 46.0±12.2 yr). The mean age of onset was 36.1±12.3 yr, the mean total duration of the disease 9.8±10.6 yr, the number of previous episodes 2.6±1.7, and the duration of the longest episode 34.4±22.1 wk.

‘Completers’ (n=110) and ‘drop-outs’ (n=46) did not differ significantly according to age (p=0.14), age of onset of the disease (p=0.61), total duration of the disease (p=0.12) or duration of the longest episode (p=0.72). Drop-outs showed a higher number of previous episodes (3.9±4.7, p=0.002), corresponding with a higher treatment resistance in this group, often accounting for withdrawal from the study protocol.

Within the first 4 d a dosage that was able to be fixed over the subsequent 4 wk was reached. The mean maintenance citalopram dose administered was 31.2±7.4 mg/d. The following 3 d were needed to achieve a stable plasma level under the fixed dosage regime. As expected, the intra-individual correlation of citalopram plasma level was high (Spearman correlation coefficient, r=0.90).

Fifty-seven percent of the patients received lorazepam with a mean dose of 1.3±1.0 mg. Non-psychotropic medications were prescribed in 32% of the patients (e.g. anti-hypertensiva) and remained unchanged during the whole study protocol.

Severity of depression (HAMD) was measured at admission, on day 8 (t1) and day 36 (t2). We aimed to measure citalopram-specific effects. Therefore, ‘baseline’ for this study was defined as day 8 in order to avoid confounding effects by admission, previous medication or change of medication. For exploration of treatment-induced changes only measurements of days 8 and 36 are used in the analysis. The mean baseline HAMD (t1) score of 23.9±5.8 points indicated a severe to moderate level of disease. HAMD scores at admission and t1 did not differ significantly (p=0.27).

Phenotypic measurements in the treatment subsample (110 patients)

Repeated Dex/CRH test

In the treatment subsample, combined Dex/CRH tests were performed twice, at t1 and t2, under fixed medication according to a previously used protocol (Heuser et al. 1996; Zobel et al. 2001). After pretreatment with an oral dose of 1.5 mg dexamethasone at 23:00 hours the previous evening, test blood samples were collected at 15:00, 15:30, 15:45, 16:00 and 16:15 hours through an intravenous catheter. The HPA system was challenged with infusion of 100 µg human CRH (Schering, Germany) at 15:02 hours, i.e. immediately after taking the first blood sample. During the whole test procedure patients rested supine but were not allowed to sleep.

The data analysis mainly relies on the area under the curve (AUC) of cortisol (area under the concentration time-curve minus linear background, determined by the trapezoidal rule), which is considered to be the most sensitive indicator (Heuser et al. 1994, 1996).

Hormone assays

Commercially available radioimmunoassay (RIA) kits (Immulite 2000, Germany) were used for cortisol measurements with a detection limit of 0.2 µg/dl plasma, and intra- and interassay coefficients of variation of 8.1% and 3.65%, respectively.

Phenotypic measurements in a second subsample of 110 patients: magnetic resonance imaging and volumetry of the hippocampus

These patients were mostly the same individuals as in the treatment subsample (97%). Magnetic resonance imaging (MRI) scanning was performed on a Philips 1.5 T Achieva whole body system with a 3D FFE sequence (TE/TR/FLIP: 15/3.6 ms/30°), with 140 slices and a resolution of 1×1×1 mm3. Data were converted to the analyse format and manual volumetry was performed with Analyse 7.0.

The right and left hippocampi were manually traced on sagittal MRI slices, starting on a midsagittal slice throughout the hippocampus. The anterior border was defined by the alveus which separated the hippocampus from the amygdala. The lateral ventricle served as the posterior border. The fimbria, as well as the lateral ventricle and the alveus defined the superior border. Inferiorly, the uncus separated the hippocampus from the parahippocampal gyrus. On further medial slices the hippocampus was divided into two separate structures by the thalamic nuclei. The head of the hippocampus bordered the temporal horn of the lateral ventricle from ventral, while the tail bordered the same structure from dorsal. The intra-rater reliability of the protocol was 0.97.

The intracranial volumes were obtained by automated tissue segmentation with SPM5 (Wellcome Department of Cognitive Neurology, UK) using tissue probability maps. The volumes of hippocampi and amygdala were divided by the total intracranial volume to adjust for differences in head size.

Marker selection/genotyping

The selection of markers for investigation was based on the report of Binder et al. (2004). Five polymorphic loci revealing statistically significant allelic association with antidepressant treatment response in this report were selected and examined in the present study (rs3800373, rs755658, rs1360780, rs1334894, rs4713916). These loci are in linkage disequilibrium (LD), with D′ values ranging from 1.0 to 0.75 (Table 2). Thus, all markers under study define a single haplotype block.

View this table:
Table 2

For genotyping, DNA was isolated from whole blood using a Qiagen Blood and Cell Culture kit (Qiagen, Germany). Our primary aim was the replication of the study by Binder et al. (2004).

All five SNPs were genotyped using fluorescence-based allelic discrimination techniques (TaqMan SNP genotyping assays, Applied Biosystems, USA) as previously described (Schwab et al. 2005). For quality control purposes, around 10% of the sample was genotyped in duplicate. No differences between duplicated samples were observed. Genotyping completion rates (>96% completion for all markers and samples) and goodness-of-fit for Hardy–Weinberg equilibrium using χ2 tests were assessed as further quality control criteria (Table 2).

Statistical analysis of the case-control sample by allele and haplotype frequencies

Differences in single-marker allele and genotype frequencies between cases and controls were evaluated with the χ2 test of independence (1 and 2 degrees of freedom, respectively).

The SNP spectral decomposition method described by Nyholt (2004) was used to determine the effective number of independent marker loci to control for multiple testing at an α value of 0.05.

Haplotype frequencies were estimated using the famhap program (Becker & Knapp, 2004). This program provides global statistical tests for each marker locus and all marker combinations as well as odds ratios for all alleles and allelic combinations. We compared haplotype frequencies between cases and controls using the respective association test (option ‘hapcc') implemented in famhap (Becker et al. 2005). We tested all marker combinations for association with the disease and calculated a global p value, which was corrected for multiple testing via Monte Carlo simulations, thereby taking LD into account (Becker et al. 2005). p=0.05 was selected as the threshold of significance for the global test.

Power analysis: comparison between cases and controls by allelic distribution

The Genetic Power Calculator (Purcell et al. 2003) was used to calculate power to detect association in the sample of 284 controls and 268 patients for the discrete phenotype of depression. The following parameters were assumed: risk-allele frequency of 30%, penetrance of 17% for homozygotes of the risk allele, 11% for heterozygotes and 7% for homozygotes of the non-risk allele, lifetime disease prevalence of 10% (α-error of 5%). The parameters were chosen to match the marker allele frequencies selected for this study. Under these conditions, the sample under study provided a power of 94% to detect association, assuming a relative risk of 2.25 for the homozygotes of the risk allele, and 1.5 for the heterozygotes (co-dominant transmission). Even for the female subsample (170 cases, 169 controls), the power to detect an effect with a genotypic relative risk of 1.5 was still 80%. However, for the male subsample (98 cases, 115 controls) the power dropped to 60% assuming the above parameters. Moreover, our power was reduced for lower marker allele frequencies (<10%).

Association analyses using treatment response and brain volumetric measures as phenotypes were performed in two distinct subsamples of n=110 (volumetric measures) and n=110 (treatment response). Assuming the same risk allele characteristics and an α-error of 5% we could detect an effect size of 0.27 with 80% power.

Statistical analysis of associated phenotypes in the case samples

The investigation of associated phenotypes was restricted to those FKBP5 sequence variants which differed significantly between cases and controls (allelic variants and haplotypes) on an allelic as well as genotypic level (see Table 3). Cases carrying the minor allele (once or twice) were compared to cases homozygous for the major allele. Mean values of the associated phenotype measure were compared between both comparison groups by t tests.

View this table:
Table 3


The genotypic distribution for all five markers did not significantly deviate from Hardy–Weinberg equilibrium in cases and controls (Table 3).

Allelic associations with diagnosis

Five markers were used to explore an association with the diagnosis of unipolar depression, and three markers revealed significant allelic frequency differences (rs3800373, p=0.014; rs1360780, p=0.035; rs4713916, p=0.013). The major (i.e. more frequent) alleles were more frequent among cases at all these three loci. The three diagnosis-associated marker loci (on an individual level) span the whole FKBP5 region under investigation, the two non-associated other markers lie in between. It should be noted that at the loci with no allelic association, the frequency of the minor alleles was <30% in controls, leading to insufficient power.

Using the method described by Nyholt (2004), we calculated that a p value of 0.017 was required for experiment-wide significance. This method of correction for multiple testing was evaluated by Salyakina et al. (2005) and it was found to be very conservative in the presence of haplotype block structure. Two markers (rs3800373, p=0.014; rs4713916, p=0.013) were significantly associated with the diagnosis of unipolar depression at this threshold. In both cases the major allele is more common among patients than controls (Table 3).

Splitting the sample by gender revealed a sex-specific pattern: the smaller male population (98 cases vs. 115 controls) did not reveal a single trend for a difference of allelic frequencies (all p values ⩾0.10); the frequency distributions of the major alleles in males at the two loci which revealed the most robust allelic associations in the total sample were 71% for cases vs. 68% for controls at rs3800373, and 70% vs. 67% at rs4713916. These differences were not significant. In contrast, the observed associations were stronger in females than in the whole population: the major alleles of rs3800373 and rs4713916, respectively, occurred in 76% (74%) of cases vs. 68% (65%) of controls (p=0.008, p=0.015, respectively).

Genotypic associations with diagnosis

For SNPs rs3800373 and rs4713916 where we were able to reveal allelic association, we were also able to show genotypic association (p=0.05, p=0.01, respectively). However, for marker rs1360780, which showed significant allelic association, only a trend for significance at the genotypic level was observed (see Table 3).

Haplotype analysis

LD analysis (see Table 2) revealed that all tested SNPs are part of a single haplotype block (all pairwise D′ values ⩾0.75) in our sample. We first considered the most informative single haplotype combinations encompassing all five loci under study (broadest window). The individual frequencies and p values for the most comprehensive single 5-marker haplotype combination with frequencies ⩾5% among controls are reported in Table 4. The global p value, taking all marker combinations into account revealed borderline significance (p=0.052). This global p value accounts for the multiple testing problem introduced by analysis of haplotypes in a case-control dataset in an appropriate way, while avoiding being overly conservative (Becker et al. 2005).

View this table:
Table 4

Due to high LD in this region, a relatively high frequency for the most common 5-marker haplotype was observed in controls (61%) (Table 4). This single 5-marker haplotype is significantly more frequent in cases than in controls (p=0.005), and increases the risk for depression (OR 1.37).

Treatment response among cases

Biological treatment response is measured by the change in cortisol response following Dex/CRH stimulation (t2 – t1). Negative values (Table 5) indicate reduction of cortisol secretion (i.e. reduction of HPA activity). In order to maximize power, we restricted the statistical analysis to the comparison of two groups at each locus (homozygote carriers of the major allele vs. heterozygote and homozygote carriers of the minor allele). For two markers (rs3800373, rs4713916) the homozygote carriers of the major allele, which is associated with the clinical diagnosis, showed less reduction of cortisol secretion in the Dex/CRH test after 4 wk of standardized citalopram treatment [p=0.08* and p=0.04* respectively; (*) covariate: baseline].

View this table:
Table 5

There were no significant results for psychopathological measurements (HAMD score). This result remained unchanged when (a) including as covariates the number of depressive episodes, age and age at onset additionally to baseline HAMD (t1) and baseline cortisol AUC (t1) (Table 5) and (b) LOCF was applied instead of ‘completer analysis’ (data not given).

Hippocampal volume among cases

Volumes of left and right hippocampus were compared between the homozygote carriers of the major allele and all other cases. Two of the three alleles associated with depression at an individual level also showed a systematic, significant or borderline relationship with the right hippocampal volume in a subsample of 110 cases (Table 6): rs1360780 (p=0.04) and rs3800373 (p=0.06). In both cases the homozygote carriers of the major allele revealed a smaller mean volume of the right hippocampus which is characteristic for depressed patients vs. controls. Consistent with this finding, the homozygote carriers of the major allele at these two sites are more frequent among depressed patients than controls.

View this table:
Table 6

No statistically significant results were obtained for the left hippocampal volume.


In our study we demonstrated that DNA sequence variants in the FKBP5 gene are associated with vulnerability to unipolar depression. The most discriminative markers are rs3800373 and rs4713916. These diagnosis-associated markers are identical with the FKBP5 markers which were shown previously to predict response to antidepressants (Binder et al. 2004). The depression-associated alleles at both loci are also specifically related to hippocampal volume and HPA response to antidepressive treatment.

Association with depression

The observed allelic and genotypic associations with the diagnosis of unipolar depression are located in an interval of 115 kb between markers rs3800373 and rs4713916 in the FKBP5 gene. The two markers associated with depression rs3800373 and rs4713916 are in LD (D′=0.75); therefore, both associations are likely to reflect the impact of a single functional variant with pathogenic effect.

LD is strong between all five markers studied, forming a single block of LD. Markers showing no association with depression in this haplotype block might have had reduced power due to their low minor allele frequencies. Thus, it is possible that the pathogenic variant is located somewhere between rs3800373 and rs4713916, or is in LD with variants at these loci. In this context it is remarkable that the promoter SNP rs4713916 is located in a putatively functional region (Lekman et al. 2008).

Our finding of an allelic association with the diagnosis of unipolar depression at FKBP5 gene loci should be seen in context of other case-control studies on the FKBP5 gene. Lekman et al. (2008) analysed three markers (rs3800373, rs1360780, rs6713916) in the STAR*D samples of cases and controls. They reported a significant genotypic association with marker rs1360780 after correcting for multiple testing. No allelic association was found in their study. We have included the same three markers in our study. In agreement with Lekman and colleagues we detected a diagnostic association at marker rs1360780, but only at a nominal allelic level. However, these two significant associations are not fully consistent: the major allele at this bi-allelic locus is more common among cases in our study, but less common in the STAR*D sample of cases. It should be mentioned that minor allele frequencies for this marker vary across populations and range from as low as 20% in Asian populations up to 40% in African American populations. Therefore, even minor differences in the study population can contribute to different outcomes.

The power of our sample compared to the STAR*D sample is of course reduced. We can not exclude the possibility that the observed association might be due a type I error. However, we point out that even for a relative risk of 1.4, our sample has a power of 84% to detect association. In addition, major differences in modes of recruitment of patients (outpatients in non-university settings in STAR*D vs. hospitalized in-patients in a university setting in our study), differences in recruitment of controls (volunteers for blood donation vs. general population controls) and possible differences in the population background (white non-Hispanic US population vs. German population with exclusion of immigrants and their children from outside Germany in our study) may explain the divergent results. Differences in allelic frequencies for both case and control samples support the possibility that the studies are dealing with non-identical populations, with different LD architectures in the region under study.

Papiol et al. (2007) described another attempt searching for a diagnostic association by using only a single marker (rs1360780) in a smaller sample. These authors failed to detect association with unipolar depression. This result is in accord with our study. Unfortunately, other FKBP5 markers were not explored by Papiol et al. (2007).

Association with disease status was not shown by Binder et al. (2004). Binder et al. included patients with bipolar as well as unipolar depression, whereas we exclusively focused on unipolar depression. However, both studies are based on clinical samples which are susceptible to overrepresentation of treatment-resistant patients. Thus, it remains a possibility that we are testing ‘treatment resistance’ to antidepressants when comparing depressive in-patients and controls. This suggestion is supported by the observation that the same alleles at rs3800373 and rs4713916 which predict non-response in the Binder et al. (2004) study are associated with unipolar depression in our study.

Relationship to treatment response

We were not able to replicate association of SNPs in the FKBP5 gene and antidepressive response as measured by the HAMD scale (Binder et al. 2004). In comparison to reports by Lekman et al. (2008) and Kirchheiner et al. (2008), it must be conceded that we only applied one follow-up measure at week 4. It is particularly evident from the analysis by Lekman et al. (2008) that the strength of genotypic effect is dependent on the definition of outcome using psychopathological measures (e.g. remission compared to responder analysis). In addition, our relatively small sample size for treatment response might have had a negative impact on the power for detecting a true effect. The limited sensitivity of scale-based assessment of antidepressant response might also explain other negative reports (e.g. by Tsai et al. 2007); however, biological indicators might overcome the problems of measurement and timing of assessments.

Therefore, we used the Dex/CRH test in addition for measurement of the biological treatment effect (Holsboer-Trachsler et al. 1991; Zobel et al. 2001). It has been well documented by previous research that changes in the Dex/CRH test indicate ‘true’ response earlier than psychopathological measures. Using this biological tool we were able to show that subjects who were homozygous for the major allele at marker rs4713916 displayed a smaller reduction of cortisol secretion during the 4-wk treatment compared to carriers of the minor allele. A similar trend was found for marker rs3800373. This association did not withstand correction for multiple testing, but is consistent with Binder et al. (2004), where homozygosity for major alleles at these two loci was overrepresented among non-responders during a 5-wk treatment period. This was also confirmed by Lekman et al. (2008) for rs4713916 in the 14-wk STAR*D study, particularly for remitters, while Kirchheiner et al. (2008) described a trend for a higher chance of responding to therapy for carriers of the C allele of rs3800373. It should be noted that lack of reduction of HPA hyperactivity during antidepressive treatment defines the biological basis for early detection of clinical non-response or instability of clinical response (Holsboer-Trachsler et al. 1991; Zobel et al. 2001). This supports the observation by Binder et al. (2004) that allelic variants at the FKBP5 gene predict response to antidepressant treatment among patients with depression.

Relationship to hippocampal volume

Diagnoses-related associations are further validated by volumetric hippocampus measures for marker rs3800373. Homozygotes for the major allele display the lowest volumetric measures (right); the same genotype is significantly more frequent among depressed patients. This is in line with the observation that reduced hippocampal volumes characterize unipolar depression.


We were able to demonstrate the impact of allelic variants of the FKBP5 gene on diagnosis, treatment response and other associated phenotypes of unipolar depression. This study is the first to extend the replicated association of the putatively functionally relevant promoter SNP rs4713916 in the FKBP5 gene to response to antidepressants by adding a biological basis: homozygosity for the common allele at this site is characterized by a reduced degree of normalization of the Dex/CRH test through citalopram.

We also propose a genetic association of rs4713916 and of two other markers (rs3800373, rs1360780) with the disease status of unipolar depression. These diagnostic associations are partly supported by our finding using hippocampal volume as a phenotype. Depression-associated alleles at two of these loci are also associated with reduced hippocampal volumes in a subset of cases.

Since correction for multiple testing did not include correction for all phenotypes tested, further studies are warranted.


We thank Llewellyn Saggers-Gray for assistance in editing the manuscript.

Statement of Interest



View Abstract