5-HT1A receptor gene C −1019 G polymorphism and amygdala volume in borderline personality disorder

Authors


*T. Zetzsche, Department of Psychiatry and Psychotherapy, Ludwig-Maximilians University, Nussbaumstrasse 7, D-80336 Munich, Germany. E-mail: thomas.zetzsche@med.uni-muenchen.de

Abstract

Alterations of amygdala structure and function have been repeatedly described in patients with borderline personality disorder (BPD). The aim of our study was to determine whether a functional polymorphism of the 5-hydroxytryptamine1A receptor (5-HTR1A) gene C −1019 G (identity number: rs6295 G/C) is associated with structural changes of the amygdala in patients with BPD. Twenty-five right-handed female inpatients with BPD according to DSM IV and 25 healthy controls matched for age, sex, handedness and educational status were enrolled. Brain volumetry of the amygdala was performed with a 1.5-T Magnetom Vision apparatus (Siemens, Erlangen, Germany) and analyzed by the software program ‘brains’. Patients who have the 5-HTR1A gene G allele had significantly smaller amygdala volumes than C/C genotype carriers (= 0.02). While no difference of allelic distribution between patients and controls was detected, the described effect of 5-HTR1A genotype on amygdala volume was found for the whole group of patients, as well as in the subgroup of patients with comorbid major depression (= 0.004) but not in controls. In contrast to these subgroups of BPD patients who had significant amygdala volume differences, the mean amygdala volume of the whole group of BPD patients was not significantly different from that of controls. In summary, our study provides first evidence that 5-HTR1A gene C −1019 G polymorphism is associated with structural changes in the limbic system of BPD patients, a finding that might be disease related and might contribute to explanation of previous discrepant results regarding amygdala volume changes in BPD. Future research is recommended to clarify possible interactions between this functional polymorphism and symptoms, course and treatment responses in this disorder.

Amygdala hyperactivity was detected in patients with borderline personality disorder (BPD) (Donegan et al. 2003; Herpertz et al. 2001) and related to disturbances of affect processing that are supposed to be a key symptom of this disorder (Corrigan et al. 2000; Herpertz 2003). Reduction of amygdala volume was reported for several samples of BPD patients (Driessen et al. 2000; Rüsch et al. 2003; Schmahl et al. 2003; Tebartz van Elst et al. 2003), while other studies did not support these results (Brambilla et al. 2004; New et al. 2007). In comparison, an increase of amygdala volume was detected in first episode of major depression (MD) (Frodl et al. 2002), in young female patients with MD (Lange & Irle 2004), and recently in BPD with comorbid MD (Zetzsche et al. 2006).

Disturbances of serotonin (5-hydroxytryptamine, 5-HT) neurotransmission were associated with symptoms that frequently occur in BPD patients (for review, see Mann 1999; van Praag et al. 2004). A dysfunction of the 5-HT system was related to impulsive and aggressive behavior in BPD patients (Lieb et al. 2004; Schmahl et al. 2002; Soloff et al. 2003), whereas endocrine studies indicated alterations in 5-HT1A receptors (5-HTR1A) in patients with BPD (Coccaro et al. 1990; Hansenne et al. 2002; Rinne et al. 2000). 5-Hydroxytryptamine1A receptors act presynaptically as autoreceptors in the raphe complex, where they reduce the firing rate of serotonergic neurons, and postsynaptically, where they modulate neuronal activity, e.g. in the hippocampus, septum and amygdala (Aghajanian 1995; Blier & Ward 2003; Rainnie 1999).

Neurotrophic effects of 5-HT were reported (Duman et al. 2000; Nestler et al. 2002), which may occur during ontogenesis (Lauder 1993) and in adulthood (Gould 1999). 5-Hydroxytryptamine1A receptors were reported to mediate some of these effects (Duman et al. 2000; Huang & Herbert 2005; Whitaker-Azmitia 1991). It was further suggested that the functional single nucleotide polymorphism (SNP) C −1019 G of the 5-HTR1A promoter (5-HTR1A C −1019 G) might modify the maturation and plasticity of cerebral neuronal networks (Albert & Lemonde 2004; Lesch & Gutknecht 2004). In addition, both stress and trauma were associated with structural hippocampal and amygdala changes in BPD patients (Driessen et al. 2000; Schmahl & Bremner 2006; Schmahl et al. 2003). So far, however, it is not clear whether environmental or genetic factors, or a combination of both, determine structural amygdala alterations in patients with BPD (Schmahl & Bremner 2006).

A genetic variation of the 5-HTR1A, the recently detected 5-HTR1A (C −1019 G) SNP (ID number: rs6295 G/C) was reported to be associated with anxiety, depression and suicidal behavior (for review, see Albert & Lemonde 2004; Lesch & Gutknecht 2004). Carriers of the 5-HTR1A (C −1019 G) G allele had more neurotic traits (Strobel et al. 2003) and had a worse response to treatment with antidepressants (Arias et al. 2005; Hong et al. 2006; Lemonde et al. 2004; Parsey et al. 2006a; Serretti et al. 2004; Yu et al. 2006). Gene imaging studies demonstrated an abnormal amygdala activation pattern in patients expressing the G allele both in panic disorder (Domschke et al. 2005) and in MD (Dannlowski et al. 2006). An inverse relationship between amygdala reactivity and 5-HT1A autoreceptor density was detected (Fisher et al. 2006), and therefore, it is important that the G allele was associated with an increased 5-HT autoreceptor feedback inhibition of raphe neurons (Lemonde et al. 2003).

Because of the described neurotrophic action of 5-HT and 5-HTR1A, we were interested in the potential effects of the (C −1019 G) 5-HTR1A gene polymorphism on structural amygdala changes in patients with BPD. The following hypotheses were tested: first, is the 5-HTR1A C −1019 G SNP associated with amygdala volume differences in BPD patients, and might this SNP therefore contribute to the explanation of previously discrepant results regarding amygdala volume in BPD? Second, is this 5-HTR1A SNP associated with recently described effects of comorbid MD on amygdala volumes in BPD patients? Third, is this potential genotype effect on brain structure exclusively disease related or also found in healthy controls?

Materials and methods

Subjects

Twenty-five patients with BPD were enrolled in the study. All were inpatients of the Ludwig-Maximilians University of Munich (LMU). Structured Clinical Interviews for DSM IV Diagnoses (SCID) II (Wittchen et al. 1997) and the Diagnostic Interview for Borderline Personality Disorder (DIB) (Gunderson & Zanarini 1983) were performed and inpatients who met the criteria of both diagnostic systems were recruited. Twenty-five healthy controls were enrolled in the study and were matched for age and educational status. All patients and control subjects were female and right-handed. Informed written consent was obtained from all patients and controls after complete and extensive description of the study, which was approved by the Ethics Committee of the LMU in accordance with the principles laid down in the Declaration of Helsinki (Rickham 1964). The SCID I and II were employed to validate other axis I and II disorders. Exclusion criteria included bipolar disorder, schizophrenia, schizotypal personality disorder, neurological or severe somatic disorders and head trauma (Zetzsche et al. 2006).

Data from the study sample have been presented in a previous publication (Zetzsche et al. 2006). Two of the patients and six of the controls from the earlier publication were excluded from the current analysis because of lack of genetic data and replaced by matched patients and controls with blood samples available for genetic analysis. The sample characteristics are presented in Table 1. Twenty-one patients (84%) were receiving psychotropic medication (antidepressants, neuroleptics or mood stabilizers) at the time of the study, while 19 patients (76%) had received it previously. SCID examination and clinical assessment indicated that there was a high comorbidity of BPD with MD and anxiety disorders [current major depressive episode (MDE) 64%, lifetime MD 60%, dysthymia 32%, panic disorder 36%, agoraphobia 20% and other anxiety disorders 32%]. In addition, other comorbid diagnoses [post-traumatic stress disorder (PTSD) 32%, lifetime bulimia 24%, somatoform disorder 20% and acute psychotic episode 16%] were detected.

Table 1.  Demographic variables of control subjects and BPD patients, including clinical characteristics in genotype subgroups of BPD patients (5-HTR1A C −1019 G SNP)
 Controls (= 25)Total number of BPD patients (= 25)BPD patients with C/C (= 10)BPD patients with G/C and G/G (= 15)t†,‡ and F§ valuesP values
  • Patients and controls were all females. P values of following statistical procedures are depicted (see Materials and methods).

  • *

    Statistically significant results (< 0.05).

  • t-Test (two sided) to examine the potential effects on depicted variables by following groups: control subjects vs. total number of BPD patients (df = 2,48)/BPD patients with C/C vs. BPD patients with GC and GG (df = 2,23).

  • t-Test (two sided) potential effects on variables by following groups: BPD patients with C/C vs. BPD patients with GC and GG (df = 2,23).

  • §

    ancova analysis for effects of childhood abuse, first hospitalization and suicide attempts on amygdala volumes within all BPD patients. Total brain volume and age were used as covariates (ancova, df = 4,21, see also Materials and methods).

Age (years)27.6 ± 6.226.7 ± 6.825.8 ± 5.427.3 ± 7.70.48/−0.550.63/0.59
Height (cm)171.0 ± 6.1167.4 ± 6.5168.1 ± 6.7166.9 ± 6.62.03/0.430.048*/0.67
Weight (kg)64.6 ± 8.068.8 ± 16.271.1 ± 19.067.3 ± 14.6−1.19/0.560.24/0.58
School education (years)11.7 ± 1.611.4 ± 1.611.1 ± 1.711.5 ± 1.60.78/−0.640.44/0.53
Total brain volume (ccm)1226.67 ± 81.681226.95 ± 128.331231.77 ± 86.241223.73 ± 153.04−0.01/0.150.99/0.88
HAMD23.8 ± 10.724.3 ± 10.023.4 ± 11.40.200.84
BGLHA scale5.6 ± 5.14.2 ± 4.36.5 ± 5.5−1.090.29
DIB total27.0 ± 4.325.1 ± 3.928.3 ± 4.3−1.880.074
DIB scaled8.2 ± 1.28.0 ± 1.28.3 ± 1.3−0.520.61
First hospitalization7 (28%)4 (40%)3 (20%)0.18§0.68§
Childhood abuse13 (52%)6 (60%)7 (47%)0.05§0.62§
Suicide attempts19 (76%)9 (90%)10 (67%)0.001§0.97§

Magnetic resonance imaging collection and processing

1.5-T Magnetom Vision (Siemens, Erlangen, Germany) was used to obtain three-dimensional magnetization-prepared rapid acquisition gradient echo (3D-MPRAGE), proton density (PD) and T2 sequences of magnetic resonance images. The following specifications were applied: coronal T2- and PD-weighted dual-echo sequence: repetition time (TR) = 3710 ms, echo time (TE) = 22 ms (T2), 90 ms (PD), field of view (FOV) = 230 mm, matrix size = 256 × 256 pixels and slice thickness = 3 mm, and 3D-MPRAGE sequence: TR = 11.6 ms, TE = 4.9 ms, FOV = 230 mm, matrix size = 512 × 512 pixels and slice thickness = 1.5 mm. The commercial software package analyze 1.0 (Biomedical Imaging Resource, Mayo Foundation, Rochester, MN, USA) was used for further image processing (Zetzsche et al. 2006). Segmentation was performed using the software program brains, 1 (v56.119) which was developed by Andreasen et al. (1993).

Definition of amygdala borders

The description of Convit et al. (1999) was used to define the amygdala borders. A region of interest approach was used. The procedures have been described in detail in earlier publications (Frodl et al. 2002; Zetzsche et al. 2006). Good intraclass correlations were obtained for the interrater (= 0.93) and intrarater reliability (= 0.91) of amygdala volume measurements.

Diagnostic instruments and questionnaires

The Hamilton Depression Scale (HAMD) (21-items version) was used to assess depressive symptoms. The Brown–Goodwin Assessment of Lifetime History of Aggression (BGLHA) scale and the Barratt Impulsiveness Scale (BIS-11) were used to determine aggressive and impulsive behaviors (for references, see Zetzsche et al. 2006).

Laboratory methods

Genomic DNA was isolated from whole blood according to standard procedures. The genotyping of the C −1019 G polymorphism in the 5-HTR1A was performed by the fluorescence resonance energy transfer method using the Light Cycler System (Roche Diagnostics, Mannheim, Germany). The following conditions were applied. Forward primer: 5′-CCG TTT TGT TGT TGT TGT CG-3′, reverse primer: 5′-CCA GCA AAA CTG GGG TTG3′, donor hybridization probe: 5′-TTT AAA AAC GAA GAC ACA CTC-fluorescein-3′ and acceptor hybridization probe: 5′-LCRed640-CTT CTT CCA TCA ATT AGC AAT AAT TGG GAG-3′. Polymerase chain reaction (PCR) was performed with 50 ng DNA in a total volume of 20 μl containing 2 μl reaction mix, 0.5 μm of each primer, 0.2 μm of each hybridization probe and 2 μm MgCl2 according to the manufacturer’s instructions for 35 cycles of denaturation (95°C, 0 seconds, ramp rate 20°C/seconds), annealing (55°C, 10 seconds, ramp rate 20°C/seconds) and extension (72°C, 10 seconds, ramp rate 20°C/seconds). After amplification, a melting curve was generated by holding the reaction at 40°C for 20 seconds and then heating slowly to 95°C with a ramp rate of 0.2°C/seconds. The fluorescence signal was plotted against temperature to give melting curves for each sample. Peaks were obtained at 57°C for the C allele and at 46°C for the G allele.

PCR products were separated on a 3% agarose and visualized by ethidium bromide staining. All laboratory procedures were carried out blind to diagnosis.

Nucleotide sequences

The C −1019 G polymorphism in the 5-HTR1A gene has the SNP ID number rs6295 according to the ID numbers from the SNP database (http://www.ncbi.nlm.nih.gov/SNP/). The SNP is documented in the GenBank contig: accession number NT_006713.14 (http://www.ncbi.nlm.nih.gov/).

Statistical analyses

Morphometric measurements from both groups were tested for normal distribution and homogeneity of variance. All statistical tests were performed two sided and considered to be statistically significant at < 0.05. A P value ≥0.05 and ≤0.07 was defined as a statistical trend. We used Student’s t-tests and analyses of variance to test for differences in demographic variables between groups and genotypes. Fisher’s exact tests were applied to compare genotype frequencies between patients and controls and between BPD patients with and without MDE.

To test our first hypothesis, amygdala volumes of patients with BPD underwent analysis of covariance (ancova) to assess main and interaction effects of within-subject factor hemisphere (left or right) and between-subject factor 5-HTR1A genotype (C/C or C/G and G/G), using total intracranial volume and age as cofactors. To examine our second hypothesis of a possible interaction between genotype and depression on amygdala volume in BPD, comorbidity with MDE (in BPD patients) was introduced as second between-subject factor in this ancova analysis. We tested our third hypothesis of a possible interaction between genotype and BPD diagnosis on amygdala volume in the combined group of BPD patients and controls and performed a similar ancova analysis with BPD diagnosis as first between-subject factor and by introducing 5-HTR1A genotype (C/C or C/G and G/G) as second between-subject factor, using total intracranial volume and age as cofactors. Then, univariate ancovas with 5-HTR1A genotype as between-subject factor and age and total intracranial volume as covariates were used for confirmatory analyses of potential effects of genotypes on amygdala volumes in control subjects and patient groups (Table 2).

Table 2.  Effects of 5-HTR1A C −1019 G SNP genotypes on amygdala volumes in left and right hemisphere of control subjects and BPD patient groups
5-HTR1A C −1019 G SNPTotalC/CC/G and G/GF values (df)P values
Amygdala volumeccmccmccm
  • Comparisons of genotype distribution (Fisher’s exact test, see Materials and methods) revealed no significant differences between groups: controls vs. whole BPD patient group (= 0.36); BPD patients with MDE vs. BPD patients without MDE (= 1.00). Amygdala volumes are given as mean (SD).

  • *

    Statistically significant results (< 0.05).

  • F values of potential effects of 5-HTR1A SNP genotype (C/C vs. C/G and G/G) on left and right amygdala volumes were tested by ancovas in controls and in each patient group separately (in this Table, results of confirmatory ancova analyses are shown, see Materials and methods).

  • P values of potential effects of 5-HTR1A SNP genotype (C/C vs. C/G and G/G) on left and right amygdala volumes were tested by ancovas in controls and in each patient group separately (in this Table, results of confirmatory ancova analyses are shown, see Materials and methods).

  • Bold values indicate amygdala values and P-values.

Controls
 Left1.70 (0.30)1.70 (0.37)1.70 (0.28)0.020.89
 Right1.71 (0.34)1.64 (0.25)1.73 (0.36)0.090.76
 n25619(4,21) 
Whole BPD patient group
 Left1.70 (0.40)1.88 (0.37)1.58 (0.39)5.520.029*
 Right1.76 (0.41)1.93 (0.33)1.66 (0.43)4.970.037*
 n251015(4,21) 
BPD patients with MDE
 Left1.80 (0.44)2.06 (0.32)1.65 (0.45)12.400.004*
 Right1.88 (0.42)2.06 (0.32)1.78 (0.46)8.090.015*
 n16610(4,12) 
BPD patients without MDE
 Left1.51 (0.23)1.61 (0.25)1.43 (0.19)0.090.78
 Right1.55 (0.31)1.73 (0.29)1.41 (0.26)1.680.25
 n945(4,5) 

Additional ancova analyses were performed to evaluate the possible effects of comorbid diagnoses and medication. Amygdala volume was used as the dependent factor, presence of comorbid diagnoses (or medication) as the between-subject factor, side (left, right) as the within-subject factor and total intracranial volume and age as cofactors. In subsequent analyses, history of childhood abuse or suicide attempts and first hospital stay were tested as between-subject factors for possible effects. Potential associations between amygdala volumes and both the HAMD score and the ratings for impulsive and aggressive behavior were evaluated by calculating Spearman nonparametric correlation coefficients.

Results

All 25 females included in the study were inpatients (age range 18–42 years). Seven of them were hospitalized for the first time. Eighteen patients with multiple admissions had a mean of 5 ± 3 hospitalizations (range 2–12). Sixteen of the 25 BPD patients (64%) met the DSM IV criteria for current MDE. A large number of our patients reported histories of childhood sexual or physical abuse (= 13, 52%). The majority of these patients (= 19, 76%) had a history of suicide attempts. Further details are shown in Table 1.

As presented in Table 1, while BPD patients and controls did not differ significantly in most of their demographic variables, control subjects were found to be significantly taller. There was no significant differences of total brain volumes between BPD patients and controls. Furthermore, there was no significant differences between BPD genotype subgroups [5-HTR1A (C −1019 G) SNP] with respect to age, height, weight, school education, depressive symptoms (HAMD), lifetime aggression (BGLHA), severity of BPD symptoms (DIB) and total brain volume.

Amygdala volume and polymorphisms of 5-HTR1A (C −1019 G)

The distributions of the 5-HTR1A (C −1019 G) C/C or C/G and G/G genotypes in BPD patients (with and without comorbid MDE) and control subjects are presented in Table 2. Statistical analysis revealed no significant difference in the 5-HTR1A C −1019 G genotype distribution across groups (Table 2). In Table 2, amygdala volumes of both BPD patients and control subjects in their respective whole groups and in their respective two separate 5-HTR1A genotype groups (C/C vs. C/G and G/G) are shown.

Evaluation of potential interactions between 5-HTR1A (C −1019 G) SNP and amygdala volume within BPD subjects

When testing our first hypothesis, ancova analysis in the whole group of BPD patients revealed a significantly larger amygdala volume in patients carrying the 5-HTR1A C/C genotype compared with those carrying the G allele (C/G and G/G) (= 6.10, df = 4,21, = 0.02). This significant result in BPD patients was confirmed by univariate ancovas for effects of genotype on left and right amygdala volumes in the whole group of BPD patients (for results, see Table 2). In the next step, an ancova analysis was performed to test the second hypothesis of a possible interaction between MDE comorbidity and genotype, which indicates a significant effect of genotype, a highly significant effect of MDE, and a trend for an interaction between genotype and MDE on amygdala volumes [ancova, df 6,19, 5-HTR1A C −1019 G genotype: = 6.55, = 0.02; MDE: = 10.61, = 0.004; 5-HTR1A genotype × MDE: = 3.64, = 0.07 (trend)]. No significant effects of side (left, right) was found: = 0.48, = 0.50. An interaction between both between-subject factors (5-HTR1A genotype and MDE) was supported by a highly significant effect of genotype on amygdala volume in the group of BPD patients with current MDE [= 12.94, df = 4,12, = 0.004]. Results of univariate ancovas of genotype effects in this group of BPD patients with MDE confirmed this significant result (see Table 2). Largest amygdala volumes were found in BPD patients with MDE and C/C genotype and lowest volumes in patients without MDE expressing the G allele (Table 2).

Evaluation of potential interactions between 5-HTR1A (C −1019 G) SNP and amygdala volume within the combined group of control subjects and BPD patients

Finally, the third hypothesis was tested by an ancova analysis, which evaluated the potential effects of the between-subject factors 5-HTR1A (C −1019 G) SNP genotypes (C/C or C/G and G/G) and diagnosis (BPD patients vs. controls) on amygdala volume in the combined group of control subjects and BPD patients. This analysis revealed no significant effects of genotype in the combined group [= 1.04, df = 6,44, = 0.31] and no effect of diagnosis [= 0.63, df = 6,44, = 0.43], but there was a statistical trend for an interaction between diagnosis and genotype [= 3.42, df = 6,44, = 0.07]. An interaction between diagnosis and genotype was supported by the fact that, in contrast to the significant results in BPD patients, a separate ancova analysis of the group of control subjects showed no significant effect of 5-HTR1A (C −1019 G) genotype on amygdala volumes (Table 2).

5-HTR1A (C −1019 G) SNP and clinical symptoms

No significant differences was detected between patients with different 5-HTR1A genotypes regarding either depressive symptoms (as measured by HAMD total scores) or intensity of aggressive or impulsive symptoms (BGLHA and BIS-11).

Amygdala volume, depression and aggressive behavior

Amygdala volume was larger in those BPD patients with comorbid MDE (= 16) compared to those without (= 9) [= 5.15, df = 4,21, = 0.03]. There was a significant positive correlation between left, but not right amygdala volume, and depressive symptoms assessed by the HAMD (= 0.43, = 0.03). No significant correlations between amygdala volumes and trait measures for aggressive or impulsive behavior (BGLHA and BIS-11) was detected.

Potential influence of psychiatric comorbidity, medication, suicide attempts and history of childhood abuse on amygdala volume

In contrast to current MDE, no significant effects of other comorbid psychiatric diagnoses [PTSD: = 0.97, df = 4,21, = 0.34; dysthymia: = 0.76, df = 4,21, = 0.42 and panic disorder: = 0.76, df = 4,21, = 0.39] on amygdala volumes was found. Moreover, no significant amygdala volume differences between those patients with and those without a history of suicide attempts [= 0.001, df = 4,21, = 0.97] or between BPD patients with and without a history of childhood abuse (for values, see Table 1) was noted. Finally, no significant interaction between type of concurrent psychotropic medication [antidepressants: = 2.63, df = 4,21, = 0.12; neuroleptics: = 0.04, df = 4,21, = 0.85 or mood stabilizers = 0.07, df = 4,21, = 0.80] and amygdala volumes was found.

Discussion

The aim of our study was to investigate the potential interactions between amygdala volume and the functional 5-HTR1A C −1019 G polymorphism in subjects with BPD. Our first hypothesis was confirmed as BPD patients showed a smaller amygdala volume in the presence of the G allele. Second, evidence was provided that effects of MDE on amygdala volume in BPD might be modified by this 5-HTR1A polymorphism. Third, an association between 5-HTR1A genotype and amygdala volume was present only in the patient group and not in controls. In contrast to subgroups of BPD patients that showed significant amygdala volume differences, the mean amygdala volume of the whole group of BPD patients was not significantly different from that of controls. In conclusion, our results support the assumption that both MDE (Zetzsche et al. 2006) and 5-HTR1A C −1019 G SNP are moderating factors that might affect amygdala volume in BPD and may therefore contribute to discrepant findings regarding structural amygdala alterations in previous studies of this disorder.

5-Hydroxytryptamine1A receptors have been implicated in the pathogenesis and treatment of depression (Drevets et al. 1999), as well as in anxiety disorders (Neumeister et al. 2004), aggression (Parsey et al. 2002) and BPD (see introductory paragraphs). Our result of an association of the G allele with a reduced amygdala volume is in line with studies describing correlations between genetic variations of different candidate genes and brain morphology (for review of ‘imaging genetics’, see Hariri et al. 2006) in healthy subjects (Pezawas et al. 2005), in MD (e.g. Frodl et al. 2004; Taylor et al. 2005), in schizophrenia (Meisenzahl et al. 2001; Szeszko et al. 2005) and now in patients with BPD.

Our finding of a reduced amygdala volume in the presence of the 5-HTR1A C −1019 G SNP G allele in BPD patients with comorbid MD corresponds to the results of a recent study, which detected abnormal amygdala activation in depressed patients expressing the G allele of this SNP (Dannlowski et al. 2006). In addition, recent positron emission tomography studies have found a change of 5-HTR1A binding in the presence of the 5-HTR1A G allele that was significant in depressed but not in nondepressed subjects (David et al. 2005; Parsey et al. 2006b), which is in line with our finding of SNP effects on amygdala volume to be present only in patients but not in controls. As mentioned in the introductory paragraphs, an inverse relationship between 5-HT1A autoreceptor density and amygdala reactivity was described (Fisher et al. 2006), and this relationship is in accordance with the hypothesis that the 5-HTR1A G allele is associated with an altered feedback inhibition of the raphe neurons that might influence both amygdala activity and structure. These points will be discussed in more detail later.

The number of patients and controls is comparable to previous gene imaging studies (Dannlowski et al. 2006; Frodl et al. 2004; Szeszko et al. 2005). Although the 5-HTR1A genotype distribution did not differ between patients and controls, it cannot be ruled out with a small sample size as used in our study that the genotype frequency found in our patients may differ from that in the general population or in a larger number of BPD subjects. An important role of 5-HT in neurogenesis and differentiation during brain development and in the adult is well documented (Duman et al. 2000; Nestler et al. 2002, also see introductory paragraphs). Therefore, neurotrophic 5-HT effects may underlie the association between HTR1A polymorphisms and amygdala volume in BPD patients. The amygdala is densely innervated by 5-HT axons, which arise in the raphe nuclei (Törk 1990) and which might represent the structural basis of 5-HT release and subsequent trophic action of 5-HT in this region.

We found that BPD patients carrying the 5-HTR1A G allele have smaller amygdala volumes compared with those with the C/C genotype, which supports the hypothesis that the 5-HTR1A C −1019 G SNP influences amygdala morphology. Albert and Lemonde (2004) have proposed a mechanism by which the G allele of the 5-HTR1A C −1019 G polymorphism might reduce the firing rate of raphe neurons through an increased negative feedback. This decline could result in the release of less 5-HT in terminal regions, such as the amygdala. The resulting reduced neurotrophic action of 5-HT might explain the association between the G allele of the 5-HTR1A C −1019 G SNP and the reduction of amygdala volume. The reduced availability of 5-HT in terminal regions predicted by this model might also explain the otherwise surprising fact that the G allele, which was described as being associated with an increased 5-HT1A activity (Parsey et al. 2006b), is not associated with an increased neurotrophic effect. However, our results correspond with the fact that in postsynaptic regions the G allele is predicted to decrease (rather than increase as in raphe neurons) 5-HT1A expression because the transcription factor Deaf-1 displays enhancer rather than repressor activity in postsynaptic cell models (Czesak et al. 2006). In addition, in postsynaptic terminal regions, the potential neurotrophic effects of 5-HT might be mediated not only by 5-HTR1As (Gould 1999) but also by other types of 5-HT receptors (for review, see Lesch 2001).

Previous functional magnetic resonance imaging studies in control subjects reported a correlation between amygdala activation patterns and a polymorphism of the 5-HT transporter gene linked polymorphic region (5-HTTLPR) (Hariri et al. 2005). Taken together, the findings of Dannlowski et al. (2006), Domschke et al. (2005), Hariri et al. (2005) and Pezawas et al. (2005) and the results of this study support the assumption that functionally relevant polymorphisms of the serotonin system (e.g. 5-HTR1A and 5-HT transporter) have influences on amygdala structure or function, whereas the effects of 5-HTR1A SNP on amygdala volume were restricted to BPD patients and not found in controls. The fact that correlation between 5-HTR1A polymorphism and amygdala volume was only found in BPD patients suggests that additional disease-related factors are required for the effects of genotype on brain structure. As BPD patients often report stressful life events, including traumata (Lieb et al. 2004; Zanarini et al. 1997), gene–environment interaction, including gene–stress interactions, might be important for the potential effects of the 5-HTR1A C −1019 G polymorphism on amygdala volume.

The study by Caspi et al. provided strong evidence that the interaction between stressful life experiences and genetic variants of the 5-HT system, in this case the 5-HTTLPR, is important for the later development of psychiatric symptoms (Caspi et al. 2003). Previous studies have shown interactions between 5-HTR1A and stress regulation (Lopez et al. 1998; van Praag et al. 2004). In addition, animal experiments demonstrated that 5-HTR1A expression in the brain might be affected by social stress in adults (Flügge 1995) and by environmental influences during development, such as neonatal handling (Garoflos et al. 2005) and separation of pups from parent animals (Ziabreva et al. 2003). Social isolation was shown to change 5-HTR1A densities both in the amygdala and in the raphe nuclei (Schiller et al. 2006b), and there are indications for a genetic (strain differences) modification of these effects (Schiller et al. 2006a). As mentioned, the 5-HTR1A G allele was recently associated with changes of ligand binding to 5-HTR1A (Parsey et al. 2006b). Future studies should elucidate potential interactions of polymorphism and stress on 5-HTR1A expression and clarify if these mechanisms have consequences for neurotrophic effects of 5-HT (Duman et al. 2000; Huang & Herbert 2005) or if these interactions modify stress effects (e.g. via steroids) on limbic brain structures (Sala et al. 2004).

Limitations of our study have to be mentioned. First, a homogeneous sample of patients with respect to gender and handedness was recruited to exclude the influence of these confounding factors. Only right-handed and female patients were enrolled and thus the results cannot be generalized for male BPD subjects. Second, the statistical power of our study might have been insufficient to demonstrate a correlation of the 5-HTR1A C −1019 G polymorphism with clinical symptoms. Finally, most of our patients had received psychotropic medication. While no statistical evidence for a medication effect on amygdala volume was detected, we cannot rule out that the effects of the 5-HTR1A C −1019 G SNP on amygdala volume were implemented via a modification of psychotropic medication effects (Tebartz van Elst et al. 2004). Clinical effects of antidepressants have been reported to be reduced in the presence of the G allele of the 5-HTR1A C −1019 G SNP (for review, see Lesch & Gutknecht 2004 and introductory paragraphs, see also Reynolds et al. 2006). Consequently, future studies should examine whether effects of this SNP on brain structures are realized by modifying drug-mediated neurotrophic effects (Nestler et al. 2002). If this hypothesis holds true, no effects of this SNP on amygdala volume would be predicted for never-treated (drug-naïve) BPD patients. Otherwise, alternative mechanisms have to be considered: some authors mentioned that the effects of 5-HTR1A C-1019 G SNP might take place during sensitive periods in early developmental stages (Albert & Lemonde 2004; Lesch & Gutknecht 2004) (see also Gross et al. 2002), which would suggest a neurotrophic effect of the 5-HT1A polymorphism on neuronal networks long before the application of psychotropic medication in BPD patients.

Our study provides first evidence that 5-HTR1A C −1019 G polymorphism is associated with structural changes in the limbic system of BPD patients, which might indicate an interaction between genetics and neuroimaging abnormalities in these patients (for recent review, see Lis et al. 2007). The fact that significant correlations were found in patients only suggests that additional disease-related factors are necessary for the receptor polymorphism to have effects on brain structure. Future research on this topic is recommended to clarify possible interactions between this functional polymorphism of the 5-HTR1A and clinical symptoms, course and treatment responses of BPD patients.

Acknowledgments

We thank Nancy C. Andreasen, MD, PhD, and her staff, who provided generous support with the brains segmentation program; Jacqueline Klesing for English language review and Anton Strauss, MD, Bernhard Burgermeister and Sylvia de Jonge, who provided technical support.

Ancillary