Processing pipeline for extraction of cortical thickness measures used to derive anatomical information from T1-weighted magnetic resonance images (MRIs). Each image is aligned to stereotaxic space, corrected for nonuniformity artifacts, tissue classified, and masked, and inner and outer cortical surfaces are extracted.
Models of white matter tracts measured. A, Left cingulum bundle; B, left inferior longitudinal fasciculus; C, left arcuate fasciculus; D, left uncinate fasciculus; E, left inferior occipitofrontal fasciculus; F, genu of corpus callosum (in red).
Significant interaction of the BDNF Val66Met variant with age at exactly those neural structures and cognitive functions vulnerable at the earliest stages of Alzheimer disease. A, Thickness of entorhinal cortex and inferior temporal gyrus. B, Microstructural integrity (fractional anisotropy) of cingulum bundle, inferior longitudinal fasisculus, and arcuate fasciculus. C, Episodic memory performance. DTI indicates diffusion tensor imaging; FA, fractional anistrophy; Met, methionine; and Val, valine.
Voineskos AN, Lerch JP, Felsky D, Shaikh S, Rajji TK, Miranda D, Lobaugh NJ, Mulsant BH, Pollock BG, Kennedy JL. The Brain-Derived Neurotrophic Factor Val66Met Polymorphism and Prediction of Neural Risk for Alzheimer Disease. Arch Gen Psychiatry. 2011;68(2):198-206. doi:10.1001/archgenpsychiatry.2010.194
The brain-derived neurotrophic factor (BDNF) Val66Met (rs6265) polymorphism may predict the risk of Alzheimer disease (AD). However, genetic association studies of the BDNF gene with AD have produced equivocal results. Imaging-genetics strategies may clarify the manner in which BDNF gene variation predicts the risk of AD via characterization of its effects on at-risk structures or neural networks susceptible in this disorder.
To determine whether the BDNF Val66Met gene variant interacts with age to predict brain and cognitive measures in healthy volunteers across the adult lifespan in an intermediate phenotype pattern related to AD by examining (1) cortical thickness, (2) fractional anisotropy of white matter tracts (ie, white matter integrity), and (3) episodic memory performance.
A cross-sectional study using genetics, high-resolution magnetic resonance imaging, diffusion tensor imaging, and cognitive testing in healthy individuals spanning the adult lifespan.
A total of 69 healthy volunteers ranging from 19 to 82 years of age.
Main Outcome Measures
The BDNF Val66Met genotype, apolipoprotein E genotype, cortical thickness, microstructural integrity of white matter tracts, and episodic memory performance were evaluated.
The BDNF Val66Met polymorphism interacted with age to predict (1) cortical thickness (prominently at the entorhinal cortex and temporal gyri), (2) fractional anisotropy of white matter tracts (prominently at white matter tracts connecting to the medial temporal lobe), and (3) episodic memory performance. For each of these findings, the pattern was similar: valine/valine individuals in late life were susceptible, and in early adult life, methionine allele carriers demonstrated susceptibility.
The BDNF gene confers risk in an age-dependent manner on the brain structures and cognitive functions that are consistent with the neural circuitry vulnerable in the earliest stages of AD. Our novel findings provide convergent evidence in vivo for a BDNF genetic mechanism of susceptibility in an intermediate phenotype related to AD.
Sporadic or late-onset Alzheimer disease (AD) constitutes 90% to 95% of AD cases and is a complex, heterogeneous disorder with increasing prevalence.1 Although some notable examples of genetic risk in AD have been established2 and some promising data from genome-wide studies have emerged,3,4 genetic investigations in this disorder have been fraught with many of the same complexities and conundrums as those of other neuropsychiatric disorders.5 The brain-derived neurotrophic factor (BDNF) gene represents an intriguing potential genetic mechanism for risk of late-onset AD.6 Brain-derived neurotrophic factor is critical for neuronal plasticity and facilitates hippocampal and cortical long-term potentiation,7 processes that are especially important for learning and memory. Learning and memory processes are substantially affected in AD, arising largely from impaired neuronal plasticity.8 In AD patients, BDNF expression is prominently reduced in the hippocampus and the entorhinal cortex,9 and these regions are consistently affected in the earliest stages of the disease.10,11 Variation in the BDNF Val66Met (rs6265, G>A) polymorphism has been shown to be related to episodic memory performance in younger adults via the hippocampal formation, where methionine (Met) allele carriers had poorer episodic memory performance.12 In addition, this polymorphism predicts cognitive performance in elderly individuals13 and may confer risk for AD,14 where valine/valine (Val/Val) individuals in these 2 studies were at risk. Recent animal model findings suggest a compelling potential role for BDNF as a therapeutic agent in AD.15 Taken together, these findings suggest that BDNF gene variation may be a genetic susceptibility mechanism for AD.
The combination of neuroimaging and genetics (ie, imaging-genetics) offers the potential to characterize the effects of BDNF risk variants on at-risk neural structures relevant to AD via the intermediate phenotype approach. Such an approach may evince greater penetrance of the effects of the gene on the vulnerable neural structure or function and is not subject to confounds present in disease populations.16 In AD, a prominent at-risk neural feature in gray matter is reduced cortical thickness in temporal lobe structures, demonstrated via structural magnetic resonance imaging. Reduced thickness is most prominent in the entorhinal cortex,17,18 a finding present at the earliest stages of disease, which aligns directly with neuropathologic studies that show that the earliest and greatest neurodegenerative changes occur in the entorhinal cortex and then in the hippocampus.11 However, structural brain changes in AD are not limited to gray matter; more recently, white matter abnormalities have become a focus of investigation.19,20
Diffusion tensor imaging (DTI) is a powerful tool that can differentiate between normal and abnormal white matter.21 In patients with AD, DTI has demonstrated disruption of white matter fibers in AD in corticocortical association fiber tracts.22 Recent work23 has identified that disruption of the cingulum bundle is highly correlated with hippocampal atrophy, represents the source of disconnection between the hippocampus and the posterior cingulate cortex, and is the primary factor in posterior cingulate cortex hypometabolism, a characteristic feature of this disorder.24 White matter findings in AD align with neuropathologic studies, in which individuals with AD exhibit more severe oligodendroglial loss and myelin breakdown,25 as well as axonal loss,26 compared with matched control individuals. Brain-derived neurotrophic factor plays a role in mediating myelination,27 provides trophic support for oligodendrocytes, and influences levels of myelin basic protein,28 the major protein in the myelin sheath.
We conducted a study in healthy volunteers spanning the adult lifespan to assess the effect of the BDNF gene and age on neural structures and cognitive functions that are disrupted in AD. We hypothesized that the BDNF Val66Met polymorphism would interact with age to predict variation in (1) cortical thickness in temporal lobe structures, (2) microstructural integrity of white matter tracts that connect to the medial temporal lobe, and (3) episodic memory performance.
Sixty-nine healthy volunteers (44 men and 25 women; mean [SD] age, 46  years; age range, 19-82 years) met the inclusion criteria (age between 18 and 85 years; right handedness) and none of the exclusion criteria (any history of a mental disorder, including dementia; current substance abuse or a history
of substance dependence; a positive urine toxicologic screen result; a history of head trauma with loss of consciousness, seizure, or another neurologic disorder; or a first-degree relative with a history of psychotic mental disorder). The ethnic distribution was 67 whites and 2 Asians. All participants were assessed with the Edinburgh handedness inventory,29 were interviewed by a psychiatrist, and completed the Structured Clinical Interview for DSM-IV Disorders30 and the Mini-Mental State Examination.31 They also completed a urine toxicology screen. Participants were characterized using the following instruments (Table): the Wechsler Test for Adult Reading, the Hollingshead index,32 the Cumulative Illness Rating Scale for Geriatrics,33 and body mass index (calculated as weight in kilograms divided by height in meters squared) and blood pressure measurement. The study was approved by the Research Ethics Board of the Centre for Addiction and Mental Health (Toronto, Ontario, Canada), and all participants provided informed, written consent.
High-resolution magnetic resonance images were acquired as part of a multimodal imaging protocol using an 8-channel head coil on a 1.5-T GE Echospeed system (General Electric Medical Systems, Milwaukee, Wisconsin), which permits maximum gradient amplitudes of 40 mT/m. Axial inversion recovery–prepared spoiled gradient recall images were acquired: echo time, 5.3; repetition time, 12.3; time to inversion, 300.0; flip angle, 20°; and number of excitations, 1 (for a total of 124 contiguous images, 1.5-mm thickness). For DTI, a single-shot spin echo planar sequence was used with diffusion gradients applied in 23 noncollinear directions and B = 1000 s/mm2. Two B = 0 images were obtained. Fifty-seven sections were acquired for whole brain coverage oblique to the axial plane. Section thickness was 2.6 mm, and voxels were isotropic. The field of view was 330 mm, and the size of the acquisition matrix was 128 × 128 mm, with an echo time of 85.5 milliseconds and a repetition time of 15 000 milliseconds. The entire sequence was repeated 3 times to improve signal to noise ratio.
All magnetic resonance images were submitted to the CIVET processing pipeline (version 1.1.9; Montreal Neurological Institute at McGill University, Montreal, Quebec, Canada). T1-weighted images were registered to the ICBM152 nonlinear sixth-generation template with a 9-parameter linear transformation, inhomogeneity corrected34 and tissue classified.35,36 Deformable models were then used to create white and gray matter surfaces for each hemisphere separately, resulting in 4 surfaces of 40 962 vertices each.37,38 From these surfaces, the t-link metric was derived for determining the distance between the white and gray surfaces.39 The thickness data were subsequently blurred using a 20-mm surface–based diffusion blurring kernel in preparation for statistical analyses. Unnormalized, native-space thickness values were used in all analyses owing to the poor correlation between cortical thickness and brain volume. Normalizing for global brain size when it has little pertinence to cortical thickness risks introducing noise and reducing power40 (Figure 1).
The 3 repetitions were coregistered to the first B = 0 image in the first repetition using the Functional Magnetic Resonance Imaging of the Brain Software Library (version 4.0; Functional Magnetic Resonance Imaging of the Brain Centre, University of Oxford, Department of Clinical Neurology, Oxford, England; www.fmrib.ox.ac.uk) to produce a new averaged image, with gradients reoriented using a weighted least squares approach. Registration corrects eddy current distortions and subject motion, important artifacts that can affect the data, and averaging improves the signal to noise ratio. A brain mask was then generated. Points were seeded throughout each voxel of the brain. Whole-brain tractography was performed with a deterministic (streamline) approach (Runge-Kutta order 2 tractography with a fixed step size of 0.5 mm). More detailed descriptions of our tractography approach and our clustering segmentation algorithm have been recently published41,42 and are summarized here. Threshold parameters for tractography were based on the linear anisotropy measure CL, which provides specific advantages compared with thresholding using fractional anisotrophy.43,44 The parameters chosen for this study were as follows: Tseed, CL = 0.3; Tstop, 0.15; and Tlength, 20 mm. Tractography and creation of white matter fiber tracts were performed using the 3D Slicer (www.slicer.org) and MATLAB 7.0 (The Mathworks Inc, Natwick, Massachusetts; www.mathworks.com).
A pairwise fiber trajectory similarity was quantified and the directed distances between fibers A and B were converted to a symmetric pairwise fiber distance. A spectral embedding of fibers was then created based on the eigenvectors of the fiber affinity matrix, and shape similarity information for each fiber was calculated using a k -way normalized cuts clustering alogorithm.41
Once the whole brain cluster model was produced, a trained operator (A.N.V.) combined clusters corresponding to a given fiber tract. Left and right association fiber tracts connecting to the temporal lobe were selected42: uncinate fasciculus, inferior occipitofrontal fasciculus, cingulum bundle, inferior longitudinal fasciculus, and arcuate fasciculus. The genu of the corpus callosum was selected for comparative purposes because this structure is highly susceptible to age-related fractional anistrophy (FA) change in healthy aging populations45 and is not preferentially disrupted at the earliest stages of AD19,46 (Figure 2) (although it may be affected in later stages of AD25,47). As reported elsewhere,42 excellent spatial and quantitative reliability using this clustering method (ie, voxel overlap and scalar measures of the tensor showed high agreement) has been demonstrated. For each white matter tract, MATLAB (version 7.0) was used to calculate a mean FA48 value along the selected tract.
The BDNF Val66Met polymorphism (rs6265) was genotyped in each study participant. This polymorphism lies in the 5′ region of the BDNF gene and affects intracellular packaging and secretion of BDNF.12 Genotyping of this polymorphism was performed using a standard (Applied Biosystems Inc, Foster City,
California) 5′ nuclease TaqMan assay-on-demand protocol in a total volume of 10 μL. Postamplification products were analyzed on the ABI 7500 Sequence Detection System (Applied Biosystems), and genotype calls were performed manually. Results were verified independently by 2 laboratory personnel masked to demographic and phenotypic information. Quality control analysis was performed on 10.0% of the sample. All participants also underwent genotyping at the apolipoprotein E (APOE) gene to determine APOE4 allele status. Apolipoprotein E was genotyped by combining allelic results from 2 single-nucleotide polymorphism assays (also assay-on-demand protocol) for rs429358 (T/C) and rs7412 (T/C). The combination of these 2 polymorphisms result in cysteine-to-arginine amino acid substitutions in APOE at positions 130 and 176. The E2 allele is represented by the Cys-Cys combination, E3 by the Cys130-Arg176 combination, and E4 by the Arg-Arg combination.
Sixty-five of the study participants completed cognitive testing that included the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS). Verbal episodic memory performance and visuospatial episodic memory performance were measured using the list recall and figure recall tests of the RBANS, respectively.
Three separate analyses were performed according to the general linear model to examine the effects of the BDNF gene and age on (1) cortical thickness, (2) white matter tract integrity, and (3) cognitive performance. Two genotypic groups were created: Met allele carriers and Val/Val individuals. Genotypic group served as the between-group factor in each model.
The first model examined an analysis of covariance (ANCOVA) relating BDNF genotype and age to cortical thickness. Statistical thresholds were determined by application of a 5% false discovery rate correction, where q <0.05 was considered significant.49
The second model used a repeated-measures ANCOVA with BDNF genotype group as the between-group factor and age as the covariate to examine white matter tract FA (all tract FA values were within-group measures) of association fiber tracts and of the genu of the corpus callosum. For episodic memory performance, a repeated-measures ANCOVA was conducted with BDNF genotype group as the between-group factor and age as the covariate. Scores on the list recall and figure recall tests of the RBANS were the 2 within-group measures. Because recent evidence suggests that risk conferred by the BDNF Val66Met for AD may be dependent on sex,50 we conducted a separate analysis on brain measures and cognitive performance stratified for sex.
The 2 genotypic groups did not differ in terms of age, sex, IQ, years of education, ethnicity, socioeconomic status, systolic blood pressure, diastolic blood pressure, or body mass index (Table). Of the 69 healthy volunteers, there were 28 Met allele carriers (including 5 Met homozygotes), and 41 individuals who were Val/Val homozygotes (χ21 = 0.34; P = .56). Of the Met allele carriers, 3 were APOE4 allele carriers, and of the Val/Val individuals, 9 were APOE4 allele carriers. A 100% genotyping success rate was achieved. The sample did not deviate from Hardy-Weinberg equilibrium (χ21 = 0.487, 2-tailed P = .48). No individual had 2 APOE4 alleles, and 12 individuals were carriers of 1 APOE4 allele.
A BDNF genotype by age interaction predicted cortical thickness at several regions in the temporal lobe, with large effect sizes, prominently at the entorhinal cortex (F1,65 = 12.5, q = 0.03, partial η2 = 0.15) and inferior temporal gyrus (F1,65 = 13.9, q = 0.016, partial η2 = 0.18) after false discovery rate correction (Figure 3A). Cortical thickness at the middle temporal gyrus and superior temporal gyrus, along with the parietooccipital sulcus, also met the false discovery rate threshold for the BDNF genotype by age interaction (eAppendix and eTable 1). No main effects of the BDNF genotype but significant effects of age were present (all q < 0.05, except for right inferior temporal gyrus).
For white matter tract integrity, a significant BDNF genotype by age interaction (F1,65 = 14.0, P < .001) and main effects of genotype and age (F1,65 = 9.0, P = .004, and F1,65 = 53.7, P < .001, respectively) were seen. Because the overall model for white matter integrity was statistically significant, follow-up univariate ANCOVAs were used with a Bonferroni corrected threshold P value for 11 comparisons at P = .004. The interaction was most notable, with large effect sizes, at the left cingulum bundle (F1,65 = 10.8, P = .002, partial η2 = 0.14) (Figure 3B) and left inferior longitudinal fasciculus (F1,65 = 10.2, P = .002, partial η2 = 0.14), which are white matter tracts connecting to the medial temporal lobe, and the left arcuate fasciculus (F1,65 = 10.0, P = .002, partial η2 = 0.13), a white matter tract with temporoparietal and temporofrontal fibers. There was no significant interaction between BDNF genotype and the integrity of any of the other white matter tracts studied. In particular, there was no interaction for the genu of the corpus callosum, the white matter tract typically most vulnerable in healthy aging studies (F1,65 = 4.0, P = .05) (eTable 2).
Finally, for episodic memory performance, a BDNF genotype by age interaction (F1,61 = 6.2, P = .02) (Figure 3C) and main effects of BDNF genotype and age (F1,61 = 4.5, P = .04, and F1,61 = 18.6, P < .001, respectively) were also present. Because the overall model for episodic memory was statistically significant, follow-up univariate ANCOVAs were used with a Bonferroni-corrected P value for 2 comparisons, at threshold P = .02, to investigate each episodic memory task separately. The BDNF genotype by age interaction revealed only small to modest effect sizes for visuospatial episodic memory performance (F1,61 = 4.7, P = .03, partial η2 = 0.07) and verbal episodic memory performance (F1,61 = 3.2, P = .08, partial η2 = 0.05) (eTable 2). Cortical thickness, white matter tract integrity, and episodic memory performance results remained significant after reanalysis of the data without the 2 participants of Asian ethnicity or the 12 APOE4 carriers.
After stratification of our analyses for sex, similar patterns in men and women, as in the overall analysis, were observed for Met allele carriers and for Val/Val individuals, in which Met allele carriers were at risk in earlier life and Val/Val individuals in later life for reduced cortical thickness, white matter integrity, and episodic memory performance.
We found that the BDNF Val66Met polymorphism interacts with age in a biologically convergent manner to predict variation in at-risk neural structures and cognitive functions of AD in healthy humans. Our findings support BDNF as a genetic susceptibility mechanism in an intermediate phenotype related to AD via its effect on thickness of temporal lobe structures, including the entorhinal cortex,10,18 white matter integrity of association fiber tracts connecting to the medial temporal lobe,19,20,51 and episodic memory formation.52
Multiple lines of evidence implicate BDNF in the AD process: BDNF expression is reduced in the hippocampus and the entorhinal cortex in AD9; neurons containing neurofibrillary tangles, the hallmark finding of AD, do not have detectable levels of BDNF immunoreactive material, whereas neurons more intensely labeled with BDNF -specific antibodies are free of tangles53; and altered levels of BDNF in serum and cerebrospinal fluid have been found in AD in vivo54,55 and have been associated with disease severity and episodic memory performance. Furthermore, recent data suggest potentially substantial effects of BDNF as a therapeutic agent: hippocampal neural stem cell transplantation restored spatial learning and memory deficits in aged triple transgenic mice, expressing pathogenic forms of amyloid precursor protein, presenilin, and tau, without altering Aβ or tau pathologic findings,56 but rather mediated via BDNF. In another study,15BDNF gene delivery to the entorhinal cortex in amyloid transgenic mice reversed synaptic loss, improved cell signaling, and restored learning and memory without altering amyloid plaque load. Therefore, BDNF can exert substantial protective effects on crucial neuronal circuitry in AD by acting through amyloid-independent mechanisms.
Considerable evidence implicates BDNF in AD. However, results of genetic association studies examining the BDNF gene with AD have not been consistently replicated. Early genetic studies of the BDNF Val66Met polymorphism demonstrated that the Val/Val genotype was associated with AD.14 Prospective data from the large Lothian Birth Cohort demonstrated that BDNF Val/Val individuals in late life experience a greater age-related decline in reasoning skills than Met carriers.13 However, these findings have not been consistently replicated.57,58 Such difficulties in genetic association studies of complex disorders have been well characterized, and a number of explanations have been put forward.59,60 One challenge may be that the rate-limiting step in gene identification in complex behavioral disorders can be the effect size of the risk allele on phenotypic variance.16
Imaging genetics offers an alternative strategy to conventional genetic association studies by delineating neural systems that are affected by genetic variation via the intermediate phenotype strategy.16 Genotype to brain phenotype associations can be shown in carriers of risk alleles even if the carriers do not exhibit the clinical phenotype. Our findings are most robust at the level of brain structure and least robust at the level of observable behavior (ie, cognition), consistent with the intermediate phenotype concept. Importantly, BDNF variation is not related in our study to structures prominently affected in healthy aging, namely, frontal gray matter61 or white matter tracts that are frontally based, such as the genu of the corpus callosum.45 Rather, the structures affected by BDNF in our healthy study participants are the structures affected in the preclinical and earliest clinical stages of AD. In gray matter, medial and then lateral temporal areas are affected first, before extending to cingulate cortex and temporoparietal regions62; in white matter, corticocortical association pathways
(eg, cingulum bundle, inferior longitudinal fasciculus, and arcuate fasciculus), the latest-myelinating fiber pathways in the brain, are affected earliest in AD22; cognitively, episodic memory performance is also affected in preclinical and the earliest clinical stages of AD. Intermediate phenotypes in other neuropsychiatric disorders, such as schizophrenia63 and depression,64 have been previously characterized using similar approaches.
Unlike the APOE gene, for which APOE4 allele carriers are at a disadvantage even in early adult life,65 the direction of the effects of the BDNF Val66Met on brain structure and cognitive function found in our study differ in an age-dependent manner. Previous investigations in young healthy individuals suggest that BDNF Met allele carriers demonstrate reduced hippocampal/parahippocampal complex volumes, function, and episodic memory performance.12,66 One explanation for Met allele vulnerability in early adult life is based on findings that the Met allele may fail to localize BDNF to secretory granules or synapses,12 altering activity-dependent processes of cortical development and plasticity in the process. Our results support these findings and add new evidence based on cortical thickness and tractography measures: young BDNF Met allele carriers are more likely to have reduced cortical thickness of medial and lateral temporal lobe structures and reduced microstructural integrity of white matter tracts connecting to medial and lateral temporal lobe regions.
In contrast to our findings in early adult life, Val/Val individuals in late adult life had diminished entorhinal cortex thickness, white matter tract integrity, and episodic memory performance. Here, our findings also align with previous findings, where others have shown that Val/Val individuals are at increased risk in later life for poor cognitive performance13 and AD.14 The substantial literature implicating BDNF in the pathophysiology of mood disorders provides an intriguing genetic mechanism for an overlapping clinical picture with AD. First, a history of depressive mood episodes is a risk factor for subsequent AD.67 Second, the BDNF Val/66Met polymorphism has been associated with risk for mood disorders68 and for neuroticism,69 although the manner in which risk is determined (ie, Met carrier vs Val/Val) is under debate. A recent investigation highlighted the complexity of risk determined by the BDNF Val66Met on physiologic measures of depression and anxiety.70 It is possible, therefore, that there is a lifetime burden with a Val/Val genotype, whereby effects of mood vulnerability, highly sensitive plasticity (eg, high stress sensitivity), or reduced resilience contribute to the intermediate phenotype found in our study.
One limitation of our study is its cross-sectional design. Specifically, we are only able to conclude that Val/Val individuals in late life and Met allele carriers in early adult life may be at a disadvantage, given the phenotypic measures used. A longitudinal study would have allowed us to examine progression across adult life of our phenotypic measures according to the BDNF genotype. However, such a study design carries its own set of challenges, including technical limitations of repeated imaging measures, attrition, and cost. Despite the cross-sectional nature of our sample, our finding is unlikely to be due to a sampling bias or cohort effect because our elderly individuals were not different from our younger individuals for IQ or educational levels. Furthermore, conclusions regarding AD severity, outcome, or treatment response, in relation to potential effects of the BDNF Val66Met polymorphism on brain and cognitive measures, cannot be drawn because AD patients were not included in the present study. Although we screened for dementia using the Mini-Mental State Examination, it is possible, given low scores on episodic memory testing, that 2 individuals in our study had mild cognitive impairment, and this might be considered a limitation of our study. Another limitation is that we did not include the fornix of the hippocampus, a commissural white matter tract, for study in our sample because of challenges in achieving high reliability42 using streamline tractography for the fornix. Investigation of this fiber tract in relation to the BDNF genotype would be useful because the fornix is an important part of the hippocampal system, may be involved in learning and memory, and may be disrupted in AD.51 Finally, although mean systolic and diastolic blood pressure results for our sample indicated that our participants were not characterized by high blood pressure as a group, 7 individuals had blood pressure results that fall within the range of stage I hypertension (as defined by the American Heart Association71). This could be considered a limitation of our study because hypertension has been associated with lower white matter integrity.72
Although others have investigated the effects of the BDNF gene on brain structure in healthy individuals across the adult lifespan,73,74 none, to our knowledge, has investigated the effects of BDNF on cortical thickness or association fiber tracts as intermediate phenotype measures. The convergent pattern of our findings across gray matter, white matter, and cognitive performance provide a more convincing picture of the effect of the BDNF Val66Met polymorphism on an intermediate phenotype related to AD than any one of these findings alone. Our findings suggest that the BDNF gene may be a susceptibility mechanism for AD and highlight a critical alternative pathway in this neurodegenerative disorder.
Correspondence: Aristotle N. Voineskos, MD, PhD, FRCPC, Department of Psychiatry, Geriatric Mental Health Program, Centre for Addiction and Mental Health, 250 College St, Toronto, Ontario, Canada M5T 1R8 (email@example.com).
Submitted for Publication: April 14, 2010; final revision received July 19, 2010; accepted August 10, 2010.
Financial Disclosure: Dr Pollock receives research support from the National Institutes of Health and the Canadian Institutes of Health Research. Within the past 2 years, he has been a member of the advisory board of Lundbeck Canada (final meeting was May 15, 2009) and has served 1 time as a consultant for Wyeth Pharmaceuticals (October 4-5, 2008). He is currently a faculty member of the Lundbeck International Neuroscience Foundation. Dr Mulsant currently receives research support from the US National Institute of Mental Health, the Canadian Institutes of Health Research, Bristol-Myers Squibb, and Wyeth Pharmaceuticals. During the past 5 years, he has also received research support or honoraria from AstraZeneca Pharmaceuticals, Eli Lilly and Company, Forest Laboratories Inc, GlaxoSmithKline PLC, Janssen, H. Lundbeck A/S, and Pfizer Inc.
Funding/Support: This work was supported by the Canadian Institutes of Health Research Clinician Scientist Award (Dr Voineskos), American Psychiatric Association/American Psychiatric Institute for Research and Education AstraZeneca Young Minds in Psychiatry Award (Dr Voineskos), Canadian Institutes of Health Research Fellowship (Dr Rajji), the Sandra A. Rotman Program of the Rotman Research Institute (Dr Pollock), and the Centre for Addiction and Mental Health.
Additional Information: Drs Voineskos and Lerch contributed equally to the manuscript. Drs Pollock and Kennedy are senior coauthors.
Additional Contributions: Tiffany Chow, MD, and Sandra Moses, PhD, provided thoughtful comments and suggestions in the writing of the manuscript and Faranak Farzan, PhD, contributed help.