the brain-derived neurotrophic factor val66met polymorphism and cerebral white matter...

The Brain-Derived Neurotrophic FactorVAL66MET Polymorphism and Cerebral WhiteMatter Hyperintensities in Late-Life Depression

Warren D. Taylor, M.D., Stephan Zuchner, M.D.,Douglas R. McQuoid, B.S., Martha E. Payne, Ph.D.,

James R. MacFall, Ph.D., David C. Steffens, M.D., M.H.S.,Marcy C. Speer, Ph.D., K. Ranga R. Krishnan, M.D.

Objective: In animal models, brain-derived neurotrophic factor (BDNF) appears toprotect against cerebral ischemia. The authors examined whether the BDNF Val66Metpolymorphism, which affects BDNF distribution, was associated with greater volumes ofhyperintense lesions as detected on magnetic resonance imaging in a cohort of depressedand nondepressed elders. Design: Subjects completed cross-sectional assessments, includ-ing clinical evaluation and a brain magnetic resonance imaging scan, and providedblood samples for Val66Met genotyping. Setting: The study was conducted at a university-based academic hospital. Participants: Participants included 199 depressed and 113nondepressed subjects aged 60 years or older. Measurement: Hyperintensity lesionvolumes were measured using a semiautomated segmentation procedure. Statisticalmodels examined the relationship between genotype and lesion volume while controllingfor depression, presence of hypertension, age, and sex. Results: After controlling forcovariates, Met66 allele carriers exhibited significantly greater white matter hyperinten-sity volumes (F1,311 � 4.09, p � 0.0442). This effect was independent of a diagnosis ofdepression or report of hypertension. Genotype was not significantly related to graymatter hyperintensity volume (F1,311 � 1.14, p � 0.2871). Conclusions: The BDNF Met66allele is associated with greater white matter hyperintensity volumes in older individuals.Further work is needed to determine how this may be associated with other clinicallyrelevant findings in late-life depression. (Am J Geriatr Psychiatry 2008; 16:263–271)

Key Words: genetic polymorphisms, magnetic resonance imaging, depression

Hyperintense lesions are bright areas in the brainparenchyma observed on magnetic resonance

imaging (MRI). These lesions are generally consid-

ered to be ischemic in origin, although pathologicexamination may also reveal perivascular dilation ordemyelination, and ultimately the pathologic exam-

Received May 18, 2007; revised August 10, 2007; accepted August 15, 2007. From the Center for Human Genetics (SZ, MCS), the Neuropsy-chiatric Imaging Research Laboratory (WDT, MEP, JRM, KRRK), and the Departments of Medicine (MCS), Psychiatry (WDT, DRM, MEP, DCS,KRRK), and Radiology (JRM), Duke University Medical Center, Durham, NC; The Miami Institute of Human Genomics (SZ), University of MiamiMiller School of Medicine, Miami, FL; and Duke–NUS Graduate Medical School (KRRK), Singapore. Send correspondence and reprint requests toDr. Warren D. Taylor, Duke University Medical Center, DUMC 3903, Durham, NC 27710. e-mail: [email protected]

© 2008 American Association for Geriatric Psychiatry

Am J Geriatr Psychiatry 16:4, April 2008 263

ination correlates to what is observed on MRI, butmay vary dependent on the brain region being ex-amined.1–3 Hyperintensities are more severe and ofgreater volume in older depressed subjects than incomparison groups4–7 and are more common withinitial development of depression later in life.8–12

This increased hyperintensity severity is associatedwith poorer antidepressant response13–16 and a morechronic course of depression.17,18

A number of clinical factors are associated withhyperintensity severity. Although age19,20 is perhapsthe most strongly associated factor, medical comor-bidity is also associated with hyperintensity severity,particularly vascular risk factors, including but notlimited to hypertension, heart disease, and carotidartery stenosis.21–23 Recent work has moved beyondthese associations, finding that hyperintensity sever-ity may be associated with metabolic24,25 and dietarydifferences,26 and that differences in hyperintensityseverity between depressed and nondepressed sub-jects may be even greater in those without significantvascular risk factors.27 There has also been interest ingenetic polymorphisms that may increase the risk ofdeveloping hyperintensities, including polymor-phisms related to blood pressure28,29 or homocys-teine metabolism.24

Another potential genetic candidate may be theVal66Met polymorphism of the brain-derived neuro-trophic factor (BDNF) gene. This polymorphism re-sults in a valine (Val) to methionine (Met) substitu-tion with the Met allele being associated withabnormal intracellular packaging and altered distri-bution of BDNF.30,31 Although the Met66 allele hasbeen mostly associated with structural and func-tional alterations of the prefrontal cortex and thehippocampus,31–33 this work has primarily beendone in younger adult populations. Meanwhile, an-imal studies of induced cerebral ischemia suggestthat BDNF expression reduces lesion volume andimproves functional recovery.34,35 Hypothetically,the Met66 polymorphism negatively affects BDNFsecretion, which in turn would result in a reduced orslower BDNF response and less protection againstischemia.

In this study, we examined the relationship be-tween the BDNF Val66Met polymorphism and vol-umes of both white matter hyperintensities (WMHs)and gray matter hyperintensities (GMHs) in a cohortof older depressed and nondepressed subjects. We

hypothesized that carriers of the Met66 allele (Met66allele homozygotes or heterozygotes) would exhibitlarger hyperintensity volumes than would Val66 al-lele homozygotes.

METHODS

Sample and Clinical Assessments

Subjects were participants in the National Instituteof Mental Health (NIMH) Conte Center for the Neu-roscience of Depression in Late Life located at DukeUniversity Medical Center. Eligibility was limited topatients aged 60 years or older. Depressed subjecteligibility included a diagnosis of major depressionbased on NIMH Diagnostic Interview Schedule36

and clinical evaluation by a geriatric psychiatrist.Exclusion criteria included 1) another major psychi-atric illness, although coexisting anxiety symptomsconsidered to be secondary to the depression diag-nosis were allowed; 2) history of alcohol or drugabuse or dependence; 3) primary neurologic illness,including dementia; and 4) any contraindication toMRI. Subjects were recruited for the study primarilythrough clinical referrals to the study, but alsothrough limited advertising at Duke University Med-ical Center and through self-referral. All depressedsubjects had age of first depression onset assessedthrough self-report, and depression severity wasmeasured using the Montgomery–Asberg Depres-sion Rating Scale.37

Comparison subjects were community-dwellingand recruited from the Aging Center Subject Registryat Duke University. Eligible comparison subjects hada nonfocal neurologic examination, no self-report ofneurologic or psychiatric illness, no evidence of adepression diagnosis based on the Diagnostic Inter-view Schedule, and no contraindication to MRI.

The study protocol was approved by the DukeUniversity Medical Center Institutional ReviewBoard. All subjects provided written informed con-sent before beginning study procedures.

This study included white subjects previously in-cluded in a study examining BDNF Val66Met allelefrequency in geriatric depression.38 However, someof the depressed subjects from that study were notincluded in this study because they did not completeneuroimaging or their neuroimaging data could not

BNDF VAL66MET Polymorphism and Cerebral White Matter Hyperintensities

Am J Geriatr Psychiatry 16:4, April 2008264

be processed. The current study also includes addi-tional nondepressed subjects recruited since the pre-vious study. The majority of these subjects also hadtheir neuroimaging results included in a previousstudy examining hyperintensity volume differencesbetween depressed and nondepressed subjects.22

Subjects were excluded if they had a diagnosis ofdementia or if the study geriatric psychiatrist sus-pected dementia at baseline. The majority of subjectshad Mini-Mental State Examination39 scores above24; some severely depressed individuals had scoresbelow 25. These subjects were followed through anacute three-month treatment phase; if the scores re-mained below 25, they were not included in thisstudy.

Presence of comorbid hypertension, diabetes, andheart disease (coded as “heart trouble”) was assessedthrough a self-report questionnaire. The format wasderived from questions included in NIMH Epidemi-ological Catchment Area Program.40 Only hyperten-sion was included as a primary variable because wehave previously demonstrated that unlike the otherdiagnoses, it is associated with cross-sectional hyper-intensity lesion volumes22 and is more common inindividuals with more severe hyperintensity lesionburden.41 Diabetes and heart disease were includedfor secondary analyses.

Genotyping

Fresh blood samples were obtained from all par-ticipants and DNA was extracted and stored accord-ing to methods and quality checks previouslyreported.42 An aliquot of DNA was used for geno-typing of the BDNF Val66Met polymorphism. DNAsamples were placed in 96-well plates together withno-template controls and four sample duplicates inan asymmetric pattern to avoid unintended plate-switching. DNA was polymerase chain reaction-am-plified applying a Taqman by-design assay (AppliedBiosystems) that recognized the single nucleotidepolymorphism, which defines the Val66Met poly-morphism (rs6265). The samples were examinedwith an ABI7900 DNA analyzer (Applied Biosys-tems) and the genotypes determined with the SDSsoftware package (Applied Biosystems). Greaterthan 95% genotyping efficiency was required beforedata were submitted for further analysis.

Magnetic Resonance Image Acquisition

Subjects were imaged using a 1.5-Tesla whole-body MRI system (Signa; GE Medical Systems, Mil-waukee, WI) using the standard head (volumetric)radiofrequency coil. The scanner alignment light wasused to adjust the head tilt and rotation so that theaxial plane lights passed across the canthomeatal lineand the sagittal lights were aligned with the center ofthe nose. A rapid sagittal localizer scan confirmedthe alignment.

A dual-echo fast spin-echo acquisition was ob-tained in the axial plane for morphometric analysisof lesion volumes. The pulse sequence parametersare relaxation time � 4000 msec, excitation time � 30,135 msec, 32 kHz (�16 KHz) full imaging bandwidth,echo train length � 16, a 256 � 256 matrix, 3-mm sec-tion thickness, 1 excitation, and a 20-cm field of view.The images were acquired in two separate acquisitionswith a 3-mm gap between sections for each acquisition.The second acquisition was offset by 3 mm from thefirst so that the resulting data set consisted of contigu-ous sections with no gap.

Magnetic Resonance Image Analysis

The segmentation protocol has been previouslydescribed43,44 and uses a modified version of MrXsoftware created by GE Corporate Research and De-velopment (Schenectady, NY) originally modified byBrigham and Women’s Hospital (Boston) for imagesegmentation.45 This semiautomated method usesthe multiple MRI contrasts to identify different tissueclassifications through a “seeding” process in whicha trained analyst manually selected pixels in eachtissue type to be identified (such as gray matter,white matter, cerebrospinal fluid, lesions, and back-ground). Lesion areas were selected based on a set ofexplicit rules developed from neuroanatomic guide-lines, consultation with a neuroradiologist, andknowledge of the neuropathology of lesions. Bothperiventricular and deep white matter lesions werecombined to provide a measure of WMHs on thesegmented image. GMHs were those occurring insubcortical gray matter structures.

All technicians received extensive training by ex-perienced volumetric analysts. Reliability was estab-lished by repeated measurements on 16 MRI scansby each rater before raters were approved to process

Taylor et al.


study data. Intraclass correlation coefficients were:left cerebral gray matter lesions � 0.995, right cere-bral gray matter lesions � 0.996, left cerebral whitematter lesions � 0.988, and right cerebral white mat-ter lesions � 0.994.

Statistical Analysis

Tests for deviations from Hardy-Weinberg equilib-rium were conducted in unrelated cases and controlsusing the exact test from Genetic Data Analysis soft-ware.46,47 All other analyses were conducted using SAS8.02 (Cary, NC). Because the allele frequency for Met66is low, homozygotes are rare. Thus, we dichotomizedsubjects into those who had no copies of Met66 (Val66homozygotes) and those who carried at least one Met66allele. Univariate analyses examined for differences indemographic variables and hyperintensity volumes be-tween depressed and nondepressed subjects, but alsobetween Val66 allele homozygotes and Met66 allelecarriers. These tests used chi-square models for cate-gorical variables and two-tailed t tests for continuousvariables. The Satterthwaite t test was used for contin-uous variables with unequal variances. These initialanalyses were extended by developing general linearmodels, in which demographic measures or hyperin-tensity volumes were dependent variables, whereasdepression diagnosis and Val66Met genotype were in-dependent variables.

General linear models were created next, examin-ing first WMH volume and then GMH volume as the

dependent variable. Independent variables includedVal66Met genotype, presence or absence of depres-sion, age, sex, and presence or absence of hyperten-sion. They also included a gene-by-diagnosis inter-action term, which was removed if it did not reachstatistical significance. Secondary analyses addition-ally included presence or absence of diabetes andheart disease; these were not included in primaryanalyses because we have not previously found across-sectional relationship between these measuresand hyperintensity volume.22

RESULTS

This study included 312 white subjects, 199 of whomhad a diagnosis of depression, and 113 were nonde-pressed control subjects. Minority subjects were notincluded, because we have previously identified adifference in allele frequency between minority sub-jects and white subjects.38 There were no significantdifferences in sex representation or age betweenthose with and without depression (Table 1); how-ever, the nondepressed group overall was more ed-ucated. As we have presented previously, the de-pressed group exhibited greater volumes of WMHlesions and GMH lesions.22 The depressed groupexhibited a mean age of depression onset of 45.2years (standard deviation: 20.9 years; range: 4–86years) and a mean Montgomery–Asberg Depression

TABLE 1. Univariate Group Differences by Diagnosis and BDNF Val66Met Genotype

Depressed (N � 199) Nondepressed (N � 113) df Test Statistic p Value

Age 70.0 (7.8) 69.9 (5.6) 294 t � 0.06 0.9533Sex (percent female) 65.3% (130/199) 72.6% (82/113) 1 �2 � 1.73 0.1878Education 13.7 (2.8) 15.6 (1.6) 310 t � 7.57 �0.0001WMH 7.0 (11.0) 4.8 (6.4) 294 t � 2.12 0.0351GMH hyperintensities 0.28 (0.50) 0.16 (0.23) 280 t � 2.77 0.0059

BDNF Genotype

Val/Val (N � 203) Met Carrier (N � 109) df Test Statistic p ValuePercent depressed 59.1% (120/203) 72.5% (79/109) 1 �2 � 5.48 0.0192Age 70.5 (7.1) 68.5 (6.8) 310 t � 1.82 0.0696Sex (percent female) 66.5% (135/203) 70.6% (77/109) 1 �2 � 0.56 0.4550Education 14.4 (2.6) 14.3 (2.7) 310 t � 0.30 0.7651WMH 5.5 (6.4) 7.3 (13.6) 130 t � 1.32 0.1902GMH 0.21 (0.39) 0.26 (0.47) 184 t � 0.75 0.4513

Note: Age and education are in years.BDNF: brain-derived neurotrophic factor; WMH: white matter hyperintensity volume; GMH: gray matter hyperintensity volume. Both of these

volume measures are in milliliters.



Rating Scale score of 26.7 (standard deviation: 8.1;range: 16–54).

Two hundred three subjects were Val66 allele ho-mozygotes, 11 were Met66 allele homozygotes, andthe remaining 98 were heterozygotes. Tests forHardy-Weinberg equilibrium deviations were calcu-lated in affected and unaffected individuals. We foundno evidence of deviation from Hardy-Weinberg equi-librium for rs6265 (Val66Met). As previously reported,there were significantly more Met66 allele carriers inthe depressed cohort (Table 1).38 To better understandthe relationship between depression diagnosis andVal66Met genotype, we tested for differences betweenhyperintensity volumes and demographic variables us-ing general linear models in which diagnosis and ge-notype were independent variables (Table 2).

In the initial univariate analyses (Table 1), therewere no significant differences in demographic mea-sures between Val66 homozygotes and Met66 allelecarriers, although age was significantly different be-tween genotype groups when controlling for depres-sion (Table 2). Initially, it appeared that Met66 allelecarriers exhibited larger white matter and gray mat-ter hyperintense lesion volumes, although these dif-ferences were not statistically significant (Table 1).When the relationship between hyperintensity vol-ume and genotype was examined by depression di-agnosis (Table 2), the Met66 allele appeared consis-tently related to WMH volume in both diagnosticgroups, but not GMH. Given previous workassociating advanced age with greater lesion vol-

umes,19,20,22,48 which could confound the results, weelected to proceed with more complete statisticalmodels. We hypothesized the lower age in the Met66allele group may have influenced the results of theunivariate analyses of lesion volumes.

The first model examined WMH volume as thedependent variable while including the independentvariables Val66Met genotype, presence or absence ofdepression, and hypertension, age, and sex. The ini-tial model included a depression-by-genotype inter-action term; this term was not statistically significant,so the model was rerun without it. In the subsequentmodel, Met66 allele carriers exhibited significantlygreater WMH volumes (F1,311 � 4.09, p � 0.0442).Age (F1,311 � 45.48, p �0.0001) and the presence ofhypertension (F1,311 � 8.91, p � 0.0031) were alsosignificantly associated with WMH volume. Neithersex (F1,311 � 0.21, p � 0.6487) nor depression(F1,311 � 1.30, p � 0.2552) was significantly associ-ated with WMH volume.

The next model used a similar structure to examineGMH volume as the dependent variable. The depres-sion-by-genotype interaction term again did not reacha level of statistical significance and so was removedfrom the model and the model was rerun. BDNF ge-notype was not significantly associated with GMH vol-ume (F1,311 � 1.14, p�0.2871). Depressed subjects ex-hibited significantly higher GMH volumes(F1,311 � 4.33, p � 0.0384), and age was positively asso-ciated with GMH volume (F1,311 � 31.41, p�0.0001).Neither hypertension (F1,311 � 1.08, p � 0.2992) nor sex

TABLE 2. Group Differences by Both BDNF Val66Met Genotype and Diagnosis

Depressed Nondepressed

Val/Val(N � 120)

Met carrier(N � 79)

Val/Val(N � 83)

Met Carrier(N � 30)

IndependentVariable F Value p Value

Age 70.3 (7.7) 68.8 (6.8) 70.1 (5.8) 67.8 (5.3) Diagnosis 0.07 0.7914Genotype 4.60 0.0327

Sex (percent female) 76/120 (63.3%) 54/79 (68.4%) 59/83 (71.1%) 23/30 (76.7%) Diagnosis 3.14 0.0774Genotype 0.64 0.4258

Education 13.7 (2.7) 13.8 (2.8) 15.5 (1.7) 15.6 (2.0) Diagnosis 30.88 �0.0001Genotype 0.00 0.9628

WMH 6.1 (7.3) 8.6 (15.7) 4.4 (5.1) 5.4 (8.8) Diagnosis 3.44 0.0648Genotype 2.58 0.1095

GMH 0.25 (0.48) 0.32 (0.55) 0.16 (0.22) 0.14 (0.25) Diagnosis 5.45 0.0202Genotype 0.65 0.4209

Note: Continuous variables presented as mean (standard deviation). Age and education are in years.BDNF: brain-derived neurotrophic factor; WMH: white matter hyperintensity volume; GMH: gray matter hyperintensity volume. Both of these

volume measures are in milliliters. Test results are of analyses in which diagnosis and genotype were included as independent variables of ageneral linear model examining demographics or hyperintensity volume as the dependent variable. Each independent variable had 1,311 degreesof freedom.

Taylor et al.


(F1,311 � 0.07, p � 0.7974) was significantly associatedwith GMH volume.

Secondary models were run including presence ofdiabetes and heart disease as dependent variables.Neither measure was associated with either WMH orGMH volume. Inclusion of these variables did notsubstantially change our findings.

DISCUSSION

This report demonstrates an association between theBDNF Val66Met polymorphism and hyperintense le-sion volume in older individuals, expanding our un-derstanding of the influence of this genetic locus onbrain structure beyond what has been observed inyounger adult populations.31–33 Our finding adds tothe broader literature of other genetic risk factors forhyperintensities; other polymorphisms associatedwith hyperintensity severity or ischemic disease in-clude polymorphisms affecting homocysteine metab-olism24 and the renin–angiotensin–aldosterone sys-tem.28,29,49 Although we have found that the Met66allele is more common in depressed older subjects,and we now associate the Met66 allele with greaterWMH volume, we did not find a gene by diagnosisinteraction. Thus, the Met66 allele is associated withgreater hyperintensity severity in both depressedand nondepressed elders.

Studies reporting an association between theMet66 allele and alterations in structure or functionof the prefrontal cortex and hippocampus have ob-served that these regions show abundant BDNF ex-pression.32,50 They propose that the relationship theyobserved may be secondary to fixed changes of syn-aptic and cellular plasticity, presumably related toaltered BDNF secretion seen with Met66 alleles,which in turn could affect cortical morphology.31 Ingeneral, studies examining hippocampal neuronssupport a role for BDNF in neuronal growth andsurvival.51,52

The mechanism for how BDNF is related to hyper-intensity lesion volume may hinge on its role inneuronal survival. Although hyperintensities mayrepresent various pathologic findings, cerebral isch-emia is thought to be a common contributor to theirdevelopment.1,2 BDNF may be protective againstischemia, because animal models have shown that

BDNF infusion will reduce cerebral damage causedby transient ischemia as well as improve functionaloutcomes.34,35,53 Thus, the Met66 allele, which is as-sociated with altered BDNF secretion, may be relatedto the severity of cerebral damage seen with isch-emia.

This relationship may become particularly impor-tant in individuals with comorbid medical condi-tions that increase the risk of developing hyperin-tense lesions or cerebral ischemia. Hypertension isassociated with hyperintensity lesion severity,whereas diabetes is associated with greater progres-sion of lesion severity.22,54 Additionally, hyperten-sion and elevated glucose levels appear to down-regulate BDNF expression.55,56 Thus, the presence ofcomorbid vascular risk factors, in conjunction withthe Met66 allele, may predispose individuals to asituation in which they cannot adequately upregu-late BDNF secretion in response to ischemia, result-ing in more severe injury to the brain. Clearly, thishypothesis needs further investigation.

Our previous work38 supports that the BDNFMet66 allele occurs more frequently in depressedelders; we now report that the Met66 allele is alsoassociated with greater WMH severity but indepen-dent of diagnosis. In multivariate models, the differ-ence in WMH between depressed and nondepressedsubjects was not statistically significant after controllingfor BDNF genotype; this polymorphism apparently ac-counts for some of the differences in WMH volumeobserved between the diagnostic groups. However,other genetic polymorphism affecting the renin–angio-tensin system or homocysteine metabolism also affectWMH severity28,29 and have been associated with de-pression.57,58 Thus, individuals who have more severeWMH disease, which is associated with depression,may have more of these polymorphisms, although any-one regardless of a depression diagnosis with one ormore of these polymorphisms may have greater WMHseverity than someone without any. Thus, gene effectsare not necessarily specific to depression, but someonewhose depression may be related to WMH disease mayhave more of these polymorphisms.

It is possible that the BDNF Met66 allele mayincrease the risk for depression independently of itsrelationship with WMH severity. Presumably by af-fecting BDNF’s role in neurogenesis, the Met66 allelehas been associated with the structure and cognitivefunction of the hippocampus33,59 and dorsolateral



prefrontal cortex,32 two regions critically involved inmood regulation. These changes in turn could pre-dispose individuals to depressive responses to ad-versity as recent studies have associated BDNF withsocial stress adaptation60 and how individuals re-spond to aversive social experiences.61,62

We demonstrate an association between this poly-morphism and greater severity of hyperintense le-sions. However, given that this study is only cross-sectional, it limits our ability to make conclusions onthe mechanism behind this relationship. The studyhas other limitations, including use of self-report forthe evaluation of hypertension, heart disease, anddiabetes. Although these measures have been used inother studies, there is the possibility of false-negativereports in which medical illness is present but notrecognized. In addition, other risk factors for cerebro-vascular disorders were not included in the study suchas hyperlipidemia or tobacco use; moreover, clinicalmeasures of vital signs may be preferable to obtainingonly a clinical history. A particular strength of thestudy is the use of measures of hyperintensity volumerather than visual rating scales as seen in other studiesexamining this issue in late-life depression.

It should be noted that there is a mean differenceof 1.8 mL of WMH volume between those with andwithout the Met66 allele, in which Met66 allele car-riers exhibit a 32.7% greater mean WMH volume.Although 1.8 mL is not a large amount when totalbrain volume in this population is approximately

1000 mL,7 an increase of one third of the lesionvolume may be clinically relevant when consideredthat increased hyperintense lesion volumes are asso-ciated not only with psychiatric comorbidity, butalso cognitive deficits,20,63 gait instability,64 and in-continence.65

It is not clear how the Val66Met polymorphism isrelated to other critical issues of late-life depressionsuch as cognitive deficits, treatment outcomes, ormortality. Future studies examining these issuesshould test for potential interactions between thispolymorphism and other genetic polymorphismsassociated with hyperintense lesions as well asexamine for a relationship with severity of comor-bid medical conditions such as hypertension anddiabetes. Identification of genetic polymorphismsrelated to WMH may have clinical implications,because such individuals may require more ag-gressive control of comorbid conditions to preventor slow WMH development and improve clinicaloutcomes.

This study was supported by NIMH grants K23MH65939, R01 MH54846, and P50 MH60451 and Na-tional Institute of Environmental Health Sciences grantES11961.

Preliminary data were presented at the 2007 AnnualMeeting of the American Association for Geriatric Psy-chiatry, New Orleans, LA, March 2–4, 2007.

References

1. Thomas AJ, O’Brien JT, Davis S, et al: Ischemic basis for deepwhite matter hyperintensities in major depression. Arch GenPsychiatry 2002; 59:785–792

2. Thomas AJ, Perry R, Barber R, et al: Pathologies and pathologicalmechanisms for white matter hyperintensities in depression. AnnNY Acad Sci 2002; 977:333–339

3. Tomimoto H, Lin J, Ihara M, et al: Subinsular vascular lesions: Ananalysis of 119 consecutive autopsied brains. Eur J Neurol 2007;14:95–101

4. Fujikawa T, Yamawaki S, Touhouda Y: Incidence of silent cere-bral infarction in patients with major depression. Stroke 1993;24:1631–1634

5. Krishnan KRR, Hays JC, Blazer DG: MRI-defined vascular depres-sion. Am J Psychiatry 1997; 154:497–501

6. Krishnan KRR, McDonald WM, Doraiswamy PM, et al: Neuroana-tomical substrates of depression in the elderly. Eur Arch Psychi-atry Clin Neurosci 1993; 243:41–46

7. Kumar A, Bilker W, Jin Z, et al: Atrophy and high intensitylesions: complementary neurobiological mechanisms in late-life depression. Neuropsychopharmacology 2000; 22:264–274

8. de Groot JC, de Leeuw F, Oudkerk M, et al: Cerebral white matterlesions and depressive symptoms in elderly adults. Arch GenPsychiatry 2000; 57:1071–1076

9. Figiel GS, Krishnan KRR, Doraiswamy PM, et al: Subcortical hy-perintensities on brain magnetic resonance imaging: a compari-son between late age onset and early onset elderly depressedsubjects. Neurobiol Aging 1991; 12:245–247

10. Lavretsky H, Lesser IM, Wohl M, et al: Relationship of age, age atonset, and sex to depression in older adults. Am J Geriatr Psychi-atry 1998; 6:248–256

11. Salloway S, Malloy P, Kohn R, et al: MRI and neuropsychologicaldifferences in early- and late-life-onset geriatric depression. Neu-rology 1996; 46:1567–1574

12. Lesser IM, Hill-Gutierrez E, Miller BL, et al: Late-onset depressionwith white matter lesions. Psychosomatics 1993; 34:364–367

13. Hickie I, Scott E, Mitchell P, et al: Subcortical hyperintensities onmagnetic resonance imaging: Clinical correlates and prognosticsignificance in patients with severe depression. Biol Psychiatry1995; 37:151–160

14. O’Brien J, Ames D, Chiu E, et al: Severe deep white matter lesions

Taylor et al.


and outcome in elderly patients with major depressive disorder:follow up study. BMJ 1998; 317:982–984

15. Simpson S, Talbot PR, Snowden JS, et al: Subcortical vasculardisease in elderly patients with treatment resistant depression.J Neurol Neurosurg Psychiatry 1997; 62:196–197

16. Iosifescu DV, Renshaw PF, Lyoo IK, et al: Brain white-matterhyperintensities and treatment outcome in major depressive dis-order. Br J Psychiatry 2006; 188:180–185

17. Lavretsky H, Lesser IM, Wohl M, et al: Clinical and neuroradio-logic features associated with chronicity in late-life depression.Am J Geriatr Psychiatry 1999; 7:309–316

18. Taylor WD, Steffens DC, MacFall JR, et al: White matter hyperin-tensity progression and late-life depression outcomes. Arch GenPsychiatry 2003; 60:1090–1096

19. Awad IA, Spetzler RF, Hodak JA, et al: Incidental subcorticallesions identified on magnetic resonance imaging in the elderly.I. Correlation with age and cerebrovascular risk factors. Stroke1986; 17:1084–1089

20. Longstreth WTJ, Manolio TA, Arnold A, et al: Clinical correlates ofwhite matter findings on cranial magnetic resonance imaging of3301 elderly people: The Cardiovascular Health Study. Stroke1996; 27:1274–1282

21. Fazekas F, Niederkor K, Schmidt R, et al: White matter signalabnormalities in normal individuals: correlation with carotid ul-trasonography, cerebral blood flow measurements, and cerebro-vascular risk factors. Stroke 1988; 19:1285–1288

22. Taylor WD, MacFall JR, Payne ME, et al: Greater MRI lesionvolumes in elderly depressed subjects than in control subjects.Psychiatry Res 2005; 139:1–7

23. Patankar T, Widjaja E, Chant H, et al: Relationship of deep whitematter hyperintensities and cerebral blood flow in severe carotidartery stenosis. Eur J Neurol 2006; 13:10–16

24. Hassan A, Hunt BJ, O’Sullivan M, et al: Homocysteine is a riskfactor for cerebral small vessel disease, acting via endothelialdysfunction. Brain 2004; 127:212–219

25. Iosifescu DV, Papakostas GI, Lyoo IK, et al: Brain MRI hyperin-tensities and one-carbon cycle metabolism in non-geriatric out-patients with major depressive disorder (part I). Psychiatry Res2005; 140:291–299

26. Payne ME, Haines PS, Chambless LE, et al: Food group intake andbrain lesions in late-life vascular depression. Int Psychogeriatr2007; 19:295–305

27. Firbank MJ, Lloyd AJ, Ferrier N, et al: A volumetric study of MRIsignal hyperintensities in late-life depression. Am J Geriatr Psy-chiatry 2004; 12:606–612

28. Takami S, Imai Y, Katsuya T, et al: Gene polymorphism of therenin–angiotensin system associates with risk for lacunar infarc-tion. The Ohasama Study. Am J Hypertens 2000; 13:121–127

29. Schmidt H, Fazekas F, Kostner GM, et al: Angiotensin gene pro-moter haplotype and microangiopathy-related cerebral damage.Results of the Austrian Stroke Prevention Study. Stroke 2001;32:405–412

30. Chen ZY, Patel PD, Sant G, et al: Variant brain-derived neurotro-phic factor (BDNF) (Met66) alters the intracellular trafficking andactivity-dependent secretion of wild-type BDNF in neurosecre-tory cells and cortical neurons. J Neurosci 2004; 24:4401–4411

31. Egan MF, Kojima M, Callicott JH, et al: The BDNF val66metpolymorphism affects activity-dependent secretion of BDNF andhuman memory and hippocampal function. Cell 2003; 112:257–269

32. Pezawas L, Verchinski BA, Mattay VS, et al: The brain-derivedneurotrophic factor val66met polymorphism and variation inhuman cortical morphology. J Neurosci 2004; 24:10099–10102

33. Szesko PR, Lipsky R, Mentschel C, et al: Brain-derived neurotro-phic factor val66met polymorphism and volume of the hip-pocampal formation. Mol Psychiatry 2005; 10:631–636

34. Nomura T, Honmou OHK, Houkin K, et al: I.V. infusion ofbrain-derived neurotrophic factor gene-modified human mesen-chymal stem cells protects against injury in a cerebral ischemiamodel in adult rat. Neuroscience 2005; 136:161–169

35. Yanamoto H, Xue JH, Miyamoto S, et al: Spreading depressioninduces long-lasting brain protection against infarcted lesion de-velopment via BDNF gene-dependent mechanism. Brain Res2004; 1019:178–188

36. Robins LN, Helzer JE, Croughan J, et al: National Institute ofMental Health Diagnostic Interview Schedule. Its history, charac-teristics, and validity. Arch Gen Psychiatry 1981; 38:381–389

37. Montgomery SA, Asberg M: A new depression scale designed tobe sensitive to change. Br J Psychiatry 1979; 134:382–389

38. Taylor WD, Zuchner S, McQuoid DR, et al: Allelic differences inthe BDNF val66met polymorphism in late-life depression. Am JGeriatr Psychiatry 2007; 15:850–857

39. Folstein MF, Folstein SE, McHugh PR: ‘Mini-mental state’: a prac-tical method for grading the cognitive state of patients for theclinician. J Psychiatr Res 1975; 12:189–198

40. Regier DA, Myers JK, Kramer M, et al: The NIMH EpidemiologicCatchment Area program: historical context, major objectives,and study population characteristics. Arch Gen Psychiatry 1984;41:934–941

41. Taylor WD, McQuoid DR, Krishnan KR: Medical comorbidity inlate-life depression. Int J Geriatr Psychiatry 2004; 19:935–943

42. Rimmler J, McDowell JG, Slotterback BD, et al: Development of adata coordinating center (DCC): data quality control for complexdisease studies. Am J Hum Genet 1998; SupplementSuppl 63:A240

43. Byrum CE, MacFall JR, Charles HC, et al: Accuracy and reproduc-ibility of brain and tissue volumes using a magnetic resonancesegmentation method. Psychiatry Res 1996; 67:215–234

44. Payne ME, Fetzer DL, MacFall JR, et al: Development of a semi-automated method for quantification of MRI gray and whitematter lesions in geriatric subjects. Psychiatry Res 2002; 115:63–77

45. Kikinis R, Shenton ME, Gerig G, et al: Routine quantitative anal-ysis of brain and cerebrospinal fluid spaces with MR imaging.J Magn Reson Imaging 1992; 2:619–629

46. Lewis PO, Zaykin D: Genetic Data Analysis 2000. Available at:http://hydrodictyon.eeb.uconn.edu/people/plewis/software.php

47. Weir BS: Genetic Data Analysis II: Methods for Discrete Popula-tion Genetic Data. Sunderland, MA, Sinauer Associates, 1996

48. Kumar A, Bilker W, Jin Z, et al: Age of onset of depression andquantitative neuroanatomic measures: absence of specific corre-lates. Psychiatry Res 1999; 91:101–110

49. Verpillat P, Alperovitch A, Cambien F, et al: Aldosterone synthase(CYP11B2) gene polymorphism and cerebral white matter hyper-intensities. Neurology 2001; 56:673–675

50. Webster MJ, Weickert CS, Herman MM, et al: BDNF mRNA ex-pression during postnatal development, maturation and aging ofthe human prefrontal cortex. Brain Res Dev Brain Res 2002;139:139–150

51. Shetty AK, Turner DA: In vitro survival and differentiation ofneurons derived from epidermal growth factor-responsive post-natal hippocampal stem cells: inducing effects of brain-derivedneurotrophic factor. J Neurobiol 1998; 35:395–425

52. Lowenstein DH, Arsenault L: The effects of growth factors on thesurvival and differentiation of cultured dentate gyrus neurons.J Neurosci 1996; 16:759–769



53. Schwartz ED, Shumsky JS, Wehrli S, et al: Ex vivo MR determinedapparent diffusion coefficients correlate with motor recoverymediated by intraspinal transplants of fibroblasts genetically mod-ified to express BDNF. Exp Neurol 2003; 182:49–63

54. Taylor WD, MacFall JR, Provenzale JM, et al: Serial MR imaging ofhyperintense white matter lesion volumes in elderly subjects:correlation with vascular risk factors. AJR Am J Roentgenol 2003;181:571–576

55. Krabbe KS, Nielsen AR, Krogh-Madsen R, et al: Brain-derivedneurotrophic factor (BDNF) and type 2 diabetes. Diabetologia2007; 50:431–438

56. Lee TH, Yang JT, Kato H, et al: Hypertension downregulates theexpression of brain-derived neurotrophic factor in the ischemia-vulnerable hippocampal CA1 and cortical areas after carotid ar-tery occlusion. Brain Res 2006; 1116:31–38

57. Baghai TC, Binder EB, Schule C, et al: Polymorphisms in theangiotensin-converting enzyme gene are associated with unipolardepression, ACE activity and hypercortisolism. Mol Psychiatry2006; 11:1003–1015

58. Bjelland I, Tell GS, Vollset SE, et al: Folate, vitamin B12, homo-cysteine, and the MTHFR 677C-�T polymorphism in anxiety anddepression: The Hordaland Homocysteine Study. Arch Gen Psy-chiatry 2003; 60:618–626

59. Hariri AR, Goldberg TE, Mattay VS, et al: Brain-derived neurotro-

phic factor val66met polymorphism affects human memory-re-lated hippocampal activity and predicts memory performance.J Neurosci 2003; 23:6690–6694

60. Surtees PG, Wainwright NW, Willis-Owen SA, et al: The brain-derived neurotrophic factor Val66Met polymorphism is asso-ciated with a sense of coherence in a non-clinical communitysample of 7335 adults. J Psychiatr Res 2006; Epub July 25

61. Berton O, McClung CA, DiLeone RJ, et al: Essential role of BDNFin the mesolimbic dopamine pathway in social defeat stress.Science 2006; 311:864–868

62. Kaufman J, Yang BZ, Douglas-Palumberi H, et al: Brain-derivedneurotrophic factor-5-HTTLPR interactions and environmentalmodifiers of depression in children. Biol Psychiatry 2006; 59:673–680

63. O’Brien JT, Wiseman R, Burton EJ, et al: Cognitive associations ofsubcortical white matter lesions in older people. Ann N Y AcadSci 2002; 977:436–444

64. Whitman GT: A prospective study of cerebral white matter ab-normalities in older people with gait dysfunction. Neurology2001; 57:990–994

65. Sakakibara R, Hattori T, Uchiyama T, et al: Urinary function inelderly people with and without leukoaraiosis: relation to cogni-tive and gait function. J Neurol Neurosurg Psychiatry 1999; 67:658–660

Taylor et al.


the brain-derived neurotrophic factor val66met polymorphism and cerebral white matter...

Documents