Magnetic resonance image of coronalsections running in the occipitorostral direction. A, The hippocampal body.B, The shape of the hippocampus may be compared with that of a rabbit withthe head directed vertically. C, The amygdala-hippocampal transition area.
Scattergrams of left and righthippocampal white matter showing serotonin transporter polymorphisms in patientswith depression. Bars depict the mean for each group.
Scattergrams of left and righthippocampal gray matter for subjects with the L/L genotype. Patientsand healthy controls did not differ significantly in age, sex, or handedness.The sample size was 16 patients and 13 controls. Bars depict the mean foreach group.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Frodl T, Meisenzahl EM, Zill P, et al. Reduced Hippocampal Volumes Associated With the Long Variant of theSerotonin Transporter Polymorphism in Major Depression. Arch Gen Psychiatry. 2004;61(2):177–183. doi:10.1001/archpsyc.61.2.177
Copyright 2004 American Medical Association. All Rights Reserved.Applicable FARS/DFARS Restrictions Apply to Government Use.2004
Substantial evidence supports a role for dysfunction of the serotonin
transporter in the pathogenesis of major depression. Several studies have
found reciprocal interactions between the serotonergic system and both brain-derived
neurotrophic factor and glutamate, which are known to modulate or affect hippocampal
To examine the influence of a polymorphism (5-HTTLPR) in the promoter region of the serotonin transporter gene on hippocampal
volumes in patients with major depression and healthy controls.
Baseline investigation of a prospective magnetic resonance imaging study
with a 4-year follow-up period.
We examined 40 inpatients with major depression as well as 40 healthy
controls matched for age, sex, and handedness.
Main Outcome Measures
Subjects underwent high-resolution magnetic resonance imaging. Furthermore,
genotyping for the 5-HTTLPR biallelic polymorphism
was performed, which consists of a 44–base pair insertion (L allele) or deletion (S allele).
Patients with the L/L homozygous genotype had
significantly smaller hippocampal gray matter (left hemisphere: P = .003; right hemisphere: P = .01) and white
matter volumes (left hemisphere: P = .001; right
hemisphere: P = .002) than controls with this genotype.
No significant differences were found between patients and controls with the L/S or S/S genotype. Moreover,
patients with the L/L genotype had significantly
smaller hippocampal white matter volumes than those with the L/S or S/S genotype (P =
These findings suggest that homozygosity for the L allele is associated with decreased hippocampal volumes in patients
with major depression but not in healthy controls. A possible explanation
is that the interaction between the serotonergic system and neurotrophic factors
as well as excitatory amino acid neurotransmission may affect hippocampal
One of the major biological substrates in the pathogenesis of majordepression is the serotonergic system.1 Serotoninis a widespread neurotransmitter in the central nervous system. Serotonergicneurons are mainly found in different raphe nuclei, from which they projectto numerous brain regions such as cortical areas, the hippocampus, and thebasal ganglia.2 The fine-tuning of serotoninneurotransmission seems to be critically influenced by the serotonin transporter(5-HTT) via clearing synaptic serotonin.
A polymorphism (5-HTTLPR) in the promoter regionof the 5-HTT gene on chromosome 17q11.2 was identifiedwith a 44–base pair insertion (L allele) ordeletion (S allele).3 Therelationship of this polymorphism to major depression is unequivocal; onestudy found a relationship,4 whereas othersdid not.5,6 Cultured human lymphoblastcell lines homozygous for the L allele were associatedwith nearly 2-fold increased 5-HTT expression and increased serotonin reuptake.3 Moreover, in mice with targeted inactivation (knockout)of the 5-HTT gene, the modulation of serotonergicsystem activity was found to depend on the 5-HTTLPR genotype.7 In vivo studies also support these findings. A significantassociation between the L/L genotype and the depressiveresponse to tryptophan depletion showed that patients homozygous for the L/L genotype who were in remission from a major depressiveepisode had rapid serotonin reuptake combined with decreased brain serotoninavailability.8 Furthermore, in 8 healthy controlshomozygous for the L allele, a significant associationwith increased iodine 123–labeled 2β-carbomethoxy-3β-(4-iodophenyl)tropane ([123I]β-CIT)binding to 5-HTT was seen, which mayindicate a greater amount of 5-HTT.9 However,these findings need clarification because 16 healthy controls did not showa relationship between the 5-HTTLPR polymorphismand [123I]β-CIT binding.10
Interestingly, it has been demonstrated that serotonergic signalingis an important regulator of both early central nervous system development11 and adult neurogenesis.12 Inaddition, because a high density of 5-HTT has been found in the hippocampusby autoradiographic13,14 and immunocytochemicalinvestigations,15 there may be a relationshipbetween serotonergic function and hippocampal morphologic characteristics.
Aside from the well-documented contribution to learning and memory,the hippocampal formation plays a critical role in the regulation of motivationand emotion.16 The hippocampus is a core regionin the limbic system and has widespread connections to diverse cortical areassuch as the prefrontal cortex, anterior thalamic nuclei, amygdala, basal ganglia,and hypothalamus,17 all of which constitutethe neuroanatomical network of mood regulation.18 Investigationsof patients with recurrent depressive episodes have demonstrated structuralchanges in the hippocampal formation,19-22 whichmay decline with longer illness duration.19 Onestudy investigating first-episode patients and those with multiple episodesfound that only patients with multiple episodes had hippocampal volume reduction,which was associated with the duration of their illness.22 Thesefindings extend those of experimental studies that suppose stress toxicity23 or a lack of neurotrophic factors24 tobe the cause of these well-known structural abnormalities of the hippocampus.However, postmortem analysis of the human hippocampus in patients with majordepression and in glucocorticoid-treated patients did not show any major morphologicchanges associated with neuronal cell death,25 soit remains unclear whether and to what extent hippocampal changes occur inmajor depression.
Recently, studies have begun to address the issue that structural changesmay predispose patients to depression; hippocampal size has been found tobe strongly genetically determined,26 and evenfirst-episode patients with major depression showed a diminished volume of6% in the left side of the hippocampus.27 Anassociation between heritability and brain size is supported by twin studies28 and by studies from our laboratory showing that methioninehomozygosity at codon 129 on the prion protein was associated with white matterreduction as well as enlargement of cerebrospinal fluid volume in healthycontrols,29 and that the allele 2 within thepromoter region of the interleukin 1β gene was related to bifrontal temporalgray matter volume deficits and reduced overall white matter in patients withschizophrenia.30 A genetic influence on brainvolumes was also found in patients with geriatric depression, in whom at least1 APOE ϵ4 allele showed significant declinein the right hippocampus.31
The aim of our study was to investigate the influence of functionalpolymorphisms in the 5-HTT gene on hippocampal morphologiccharacteristics in patients with major depression and healthy controls.
We recruited 40 inpatients with major depression from the Departmentof Psychiatry at Ludwig-Maximilians-University, Munich, Germany (age range,18-65 years; mean ± SD age, 44.4 ± 11.7 years) (Table 1). Psychiatric diagnoses based on DSM-IV criteria and on the Structured Clinical Interview for DSM-IV were determined by a consensus of at least 2 psychiatrists.Mean ± SD illness duration was 7.9 ± 10.1 years, and numberof hospitalizations was 2.2 ± 3.0. Moreover, mean ± SD cumulativeillness duration as described by Sheline et al19 (27.7± 35.2 months) and total previous duration of antidepressant medicationuse (25.5 ± 55.8 months) were obtained retrospectively according tothe anamnesis. Clinical variables were documented using the 21-item HamiltonDepression Rating Scale. Fourteen patients received selective serotonin reuptakeinhibitors (4 sertraline hydrochloride, 7 citalopram, 2 paroxetine, and 1fluvoxamine maleate), 7 were taking tricyclic antidepressants (2 amitriptylinehydrochloride, 2 doxepin hydrochloride, and 1 trimipramine), and 16 receivedother new antidepressants (5 venlafaxine hydrochloride, 4 reboxetine, and7 mirtazapine). Four patients were not taking antidepressant medication atthe time of magnetic resonance imaging.
For comparison, 40 healthy control subjects were matched with respectto age (age range, 19-58 years; mean ± SD age, 41.7 ± 12.1 years),sex, and handedness. A structured interview was used to assess medical history,trauma, and other exclusion criteria. Neither the healthy controls nor theirfirst-degree relatives had a history of neurologic or mental illness. Exclusioncriteria for patients and controls were previous head injury with loss ofconsciousness, cortisol medication use in the medical history, previous alcoholor substance abuse, or neurologic diseases. Comorbidity with other mentalillnesses or personality disorders was also excluded. No subject had receivedprevious electroconvulsive therapy. Hypertension was excluded using bloodpressure measures, and handedness was determined with the Edinburgh inventory.32
After an extensive description of the study to the patients with majordepression and healthy controls, written informed consent was obtained. Thestudy design was approved by the local ethics committee and was prepared inaccordance with the ethical standards in the Declaration of Helsinki.
Magnetic resonance images were obtained (1.5-T Magnetom Vision; Siemens,Erlangen, Germany) using a coronal T2- and proton density–weighted dualecho sequence (repetition time, 3710 milliseconds; echo time, 22/90 ms; totalacquisition time, 9 minutes; number of acquisitions, 1; field of view, 230mm; matrix, 240 × 256; section thickness, 3 mm) and a 3-dimensionalmagnetization prepared rapid acquisition gradient-echo sequence (repetitiontime, 11.6 milliseconds; echo time, 4.9 milliseconds; total acquisition time,9 minutes; number of acquisitions, 1; field of view, 230 mm; matrix, 512 ×512; section thickness, 1.5 mm). A commercial software package was used (Analyze;Biomedical Imaging Resource, Mayo Foundation, Rochester, Minn) for furtherimage processing, with size reduction from 16 to 8 bit and transformationto a uniform matrix of 256 × 256 on 192 sections each 1.0 mm thick.All datasets were realigned and resampled 3-dimensionally for the anteriorcommissure to posterior commissure line according to Talairach coordinateswith the software program BRAINS (Brain Research: Analysis of Images, Networksand Systems; developed by Andreasen et al).33 Thisprogram allowed control of the regions of interest for the sagittal and transversesections simultaneously as well as of the segmentation for calculating theintracranial content and the gray and white matter volumes (cubic centimeters)within the defined region of interest.
We used the definition of the hippocampus (Figure 1) according to Niemann et al.34 Thehippocampal-amygdala border was detected using the description by Convit etal35; in addition, Frodl et al27 providea detailed description of this area. The evaluation staff was blind to eachsubject's status. On each magnetic resonance image, we started with the mostposterior coronal section where the hippocampus was clearly detectable. Theposterior hippocampal body and intralimbic sulcus can be seen in Figure 1A. In Figure 1B, the shape of the hippocampus may be compared with thatof a rabbit with the head directed vertically; the medial ambient cisternis separated from the temporal horn of the lateral ventricles. The amygdala-hippocampaltransition zone appears as a diffuse area of gray matter between the anteriorportion of the hippocampus and the posterior portion of the amygdala. Thisstructure can be identified most reliably in the axial plane. The boundarybetween the hippocampus and amygdala is clearly detectable in the sagittalplane (Figure 1C). The anteriorpart of the hippocampus ends where the cornu inferius of the lateral ventriclebecomes vertically oriented.
For determination of interrater reliability, 10 brains were randomlychosen, and regions of interest were independently determined by 2 raters.The intraclass correlations for the interrater reliability of hippocampalgray matter (r9 = 0.97; P<.001) and hippocampal white matter (r9 = 0.82; P = .008) were high. For the intraraterreliability, 10 brains were obtained 4 weeks apart by 1 rater (T.F.) (hippocampalgray matter: r9 = 0.96; P<.001; hippocampal white matter: r9 = 0.93; P<.001).
The DNA was extracted from of a 5-mL blood sample using a kit (QIAampBlood Isolation Kit; QIAGEN GmbH, Hilden, Germany) following the prescriptionof the supplier. Genotyping was carried out applying polymerase chain reactionamplification using the primers and methods described earlier by Heils etal.36 The final volume was 25 µL consistingof 50 ng of DNA, l µmol/L of each primer, 200µM of deoxynucleotidetriphosphate, 100µM of 7-deaza–guanosine triphosphate, 5% dimethylsulfoxide, 10mM of Tris hydrochloride (pH, 8.3), 50mM of potassium chloride,1.5mM of magnesium chloride, and 2.5 U of DNA polymerase (AmpliTaq Gold; PerkinElmer,Langen, Germany). The polymerase chain reaction products were separated ona 3% agarose gel (FMC NuSieve, 3:1; Biozym Diagnostic GmbH, Oldendorf, Germany)and visualized with ethidium bromide staining.
Morphometric measurements in both groups were tested for normal distributionand homogeneity of variance. All statistical tests were considered to be significantat P<.05. We used t testsand analysis of variance to test for differences in demographic variablesbetween patients and controls as well as between genotypes. We used χ2 tests to compare the genotype frequencies between patients and controls.Hippocampal volumes underwent analysis of covariance (ANCOVA) assessing themain and interaction effects of the within-subject factor of hemisphere (leftor right) and the between-subject factors of diagnosis (patients with depressionor controls) and genotype (L/L, L/S, or S/S) using total cranial volume asthe cofactor. Brain lobe volumes also underwent ANCOVA with the within-subjectfactors of hemisphere (left or right) and region (frontal, temporal, parietal,or occipital) and the between-subject factors of diagnosis (patients withdepression or controls) and genotype (L/L, L/S, or S/S). Furthermore, hippocampalvolumes of individuals homozygous for the L/L genotypewere compared with those carrying the L/S or S/S genotype in each diagnostic group by ANCOVA design.Posthoc analyses were carried out with ANCOVA and t teststo test hippocampal volumes for differences between genotypes. Effect sizesregarding the association between genotype and hippocampal volumes were presentedas Pearson correlation coefficients according to a method described by Rosenthal.37
Patients and controls did not differ with regard to demographic variables(Table 1), and these variableswere not different between genotypes (L/L, L/S, and S/S). Illness duration(F2,39 = 0.15; P = .86), age at onset(F2,39 = 0.40; P = .67), number of hospitalizations(F2,39 = 0.98; P = .39), and depressionseverity measured with the Hamilton Depression Rating Scale (F2,39 =0.27; P = .77) did not differ between genotypes (L/L, L/S, and S/S). Furthermore, antidepressant medication use (selective serotoninreuptake inhibitors, tricyclic antidepressants, and other antidepressants; χ238 = 3.4; P = .50) and total previousduration of antidepressant treatment (F2,39 = 0.68; P = .51) were not significantly different between genotypes. The alleleand genotype frequencies are indicated in Table 1; they did not differ between patients and controls (χ239 = 0.56; P = .76).
The analysis of variance for frontal, temporal, parietal, and occipitallobes did not show significant effects. There were neither significant maineffects for diagnosis and genotype nor significant interactions of diagnosisand genotype; genotype and region; genotype, diagnosis, and region; or genotype,diagnosis, hemisphere, and region.
The results of ANCOVA for the gray and white matter of the hippocampusappear in Table 2. A significantmain diagnosis effect was found, indicating smaller hippocampal gray and whitematter volumes in patients with major depression as compared with healthycontrols. Moreover, there was a significant association between genotype andhippocampal volumes, with a significant diagnosis and genotype interactionfor hippocampal white matter (F2,73 = 3.0; P = .05) but not hippocampal gray matter (F2,73 = 0.88; P = .42).
Post hoc ANCOVA revealed a significant effect of genotype for hippocampalwhite matter in patients (F2,36 = 3.8; P =.03) (Figure 2), whereas no significanteffect was found for healthy controls (F2,36 = 0.58; P = .56). Furthermore, significantly smaller right hippocampal whitematter volumes (F1,39 = 8.0; P = .008; r39 = 0.42) and a trend toward smaller lefthippocampal white matter volumes (F1,39 = 3.7; P = .06; r39 = 0.30) were foundin patients with the L/L genotype as compared withthose who carried the L/S or S/S genotype.
Post hoc t tests showed significantly smallerright hippocampal white matter volumes in patients homozygous for the L/L genotype compared with the L/S (t30 = 2.6; P = .02;effect size: r29 = 0.41) and S/S (t22 = 2.7; P = .01; r21 = 0.43) genotypes,as well as smaller left hippocampal white matter volumes compared with the S/S genotype (t22 =2.3; P = .03; effect size: r21 = 0.36) but not compared with the L/S genotype(t30 = 1.5; P =.15; effect size: r29 = 0.24).
The results of ANCOVA to assess differences between patients and controlsin each genotype group are presented in Table 3. Significantly smaller hippocampal gray matter (Figure 3) and white matter volumes were detectedin patients homozygous for the L/L genotype as comparedwith controls who carried this genotype. On the other hand, no significantdifferences were observed between patients heterozygous for the L/S genotype and controls with this genotype or between those homozygousfor the S/S genotype and controls who carried that genotype.
There were no significant correlations between hippocampal volumes andillness duration (left hippocampal gray matter: r29 = −0.17; P = .30; right hippocampalgray matter: r29 = −0.17; P = .29; left hippocampal white matter: r29 = −0.18; P = .26; righthippocampal white matter: r29 = 0.06; P = .73). There was also no significant correlation betweenseverity of depression as measured by the Hamilton Depression Rating Scaleand hippocampal volumes (left hippocampal gray matter: r29 = −0.17; P = .30; righthippocampal gray matter: r29 = −0.17; P = .29; left hippocampal white matter: r29 = −0.18; P = .26; righthippocampal white matter: r29 = 0.06; P = .73). Furthermore, including illness duration as acofactor in the ANCOVA did not change the effect of genotype on hippocampalvolumes.
When we excluded from analysis the patient and control subject who appearedto be outliers in the scattergrams, these results did not change. Because1 study reported brain abnormalities in psychotic depression as compared withnonpsychotic depression,38 the analysis wasalso conducted without the 5 patients who had psychotic features, but againthe results did not change.
Our study investigates the contribution of functional genetics to hippocampalvolumes in patients with major depression and healthy controls. In line withprevious results,20,25-27 hippocampalvolumes were found to be significantly smaller in patients as compared withhealthy controls. The main new finding of this study was that the L/L genotype was associated with reduced gray and white matter of thehippocampal formation in patients vs healthy controls.
To our knowledge, this is the first study of the relationship betweenthe 5-HTTLPR polymorphism and hippocampal volumesin major depression. Therefore, our findings deserve discussion, particularlywith respect to limitations. The study was not designed to test an associationof the 5-HTTLPR polymorphism with major depressionbecause the sample size was too small to detect modest contributions. Nevertheless,this result was similar to that in recent investigations with larger samples.5,6
Case-control associations are known to be sensitive to population stratification.Patients who have recurrent depression may have a longer duration of exposureto antidepressant medication as compared with those with a first episode.However, because patients and controls with the 5-HTTLPR genotype did not differ with respect to age, sex, illness duration,age at onset, medication use, or national origin, the population stratificationin this study was likely not a problem.
Our patients were currently receiving medication. We are aware of nostudies on the influence of antidepressants on brain structures in depressedpatients, so the medications' effects are unknown. A preliminary investigationin 10 pediatric patients with obsessive-compulsive disorder showed enlargedthalamic volumes before treatment and a decrease in thalamic volumes after12 weeks of treatment with paroxetine hydrochloride; it is unclear if thiseffect is due to the medication or to changes in symptoms during treatment.39 In neural cell lines, tricyclic antidepressants andselective serotonin reuptake inhibitors were found to be neurotoxic.40 In the future, it would be interesting to study patientswith no previous medication use to overcome the bias that hippocampal volumesare influenced by medication.
Because the L allele is more common in thepopulation than the S allele, it could be arguedthat an equal number of individuals for each genetic subtype should have beensampled. However, this would have meant including an extremely large numberof patients and controls.
The results of our study indicate that hippocampal volume is geneticallydetermined.19Concerning the functional importanceof the neurotransmitter serotonin on brain morphologic characteristics, ithas been demonstrated that serotonergic signaling is an important regulatorof early central nervous system development.11 Hippocampalvolumes were already found to be reduced early in the disease20;thus, reduced hippocampal volumes in patients with the L/L genotype might be a risk factor rather than a consequence of majordepression.
The effects we saw of genotype on the hippocampus could also be linkedto central nervous system development, modulation of certain effects of theillness, different vulnerability to stress, or different response to antidepressanttreatment. If the higher reuptake of serotonin in subjects with the L/L genotype modulates the course of the disease, the hippocampalvolumes might be affected as a result of depression or stress-related neurotoxicprocesses. The L/L genotype might cause a distinctvulnerability to the stress reaction. A hypothesis is that stress and increasedglucocorticoids may contribute to hippocampal volume loss via glutamatergictoxic effects.29 Stress decreases the expressionof brain-derived neurotrophic factor in the hippocampus,41 whichwas found to have trophic effects on serotonergic neurons.42 Brain-derivedneurotrophic factor and the serotonergic system may regulate reciprocal functions;the 5-HTT function is modulated by brain-derived neurotrophic factor,43 which in turn was found to be elevated in the hippocampusand frontal cortex after antidepressant treatment.42 However,our study did not show a significant correlation between the hippocampus andillness duration in contrast to that by Sheline et al,19 whoinvestigated elderly patients. Thus, a hippocampal decline during the courseof depression must be further evaluated in prospective studies.
Furthermore, the short variant of the 5-HTTLPR polymorphismwas associated with a poor response to antidepressant treatment.44 Therefore,particular patients with the L/L genotype may benefitfrom antidepressant drugs. Adaptation of the cyclic adenosine monophosphatesecond-messenger pathway, including up-regulation of the cyclic adenosinemonophosphate response element binding protein as a result of therapy, increasesthe expression of brain-derived neurotrophic factor,45 whichhas neurotrophic and possibly protective effects on hippocampal neurons.
In summary, this is the first study that investigates genetic contributionsof the 5-HTTLPR polymorphism to hippocampal volumereduction in major depression. Patients with the L/L genotypemay have a higher vulnerability to hippocampal changes. It is unclear whetherthese changes occur before the beginning of the disease or are triggered froma variety of factors, such as stress or emotional trauma during the depressiveepisode. We hope that our study may stimulate further research into the influenceof functional genetics on brain structure in major depression.
Corresponding author: Eva M. Meisenzahl, MD, head, imaging laboratory,Department of Psychiatry, Ludwig-Maximilians-University of Munich, Nussbaumstr7, Munich 80336, Germany (e-mail: firstname.lastname@example.org)
Submitted for publication July 24, 2002; final revision received July14, 2003; accepted July 21, 2003.
This study was supported by the German Federal Research Ministry (Berlin)within the German Research Networks in Medicine promotion as part of the GermanResearch Network on Depression project.
We thank Nancy C. Andreasen, MD, PhD, and her staff, who provided generoussupport with the BRAINS segmentation program, and Anton Strauss, MD, and BernhardBurgermeister, who provided technical support.
Create a personal account or sign in to: