Psychoneuroendocrinology (2014) 39, 94—103
Available online at www.sciencedirect.com
ScienceDirect j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / p s y n e u e n
Effect of BDNF Val66Met polymorphism on regional white matter hyperintensities and cognitive function in elderly males without dementia Chu-Chung Huang a,b, Mu-En Liu c, Kun-Hsien Chou b,i, Albert C. Yang d,e,f, Chia-Chun Hung b,g,h, Chen-Jee Hong d,e,g, Shih-Jen Tsai d,e,*, Ching-Po Lin a,b,g,i,** a
Department of Biomedical Imaging and Radiological Sciences, National Yang-Ming University, Taipei, Taiwan Brain Connectivity Lab, Institute of Neuroscience, National Yang-Ming University, Taipei, Taiwan c Department of Psychiatry, Kaohsiung Veterans General Hospital, Kaohsiung, Taiwan d Department of Psychiatry, Taipei Veterans General Hospital, Taipei, Taiwan e School of Medicine, National Yang-Ming University, Taipei, Taiwan f Center for Dynamical Biomarkers and Translational Medicine, National Central University, Chungli, Taiwan g Institute of Brain Science, National Yang-Ming University, Taipei, Taiwan h Department of Psychiatry, Taichung Veterans General Hospital, Taichung, Taiwan i Brain Research Center, National Yang-Ming University, Taiwan b
Received 19 March 2013; received in revised form 30 September 2013; accepted 30 September 2013
KEYWORDS Brain-derived neurotrophic receptor; White matter hyperintensities; Cognition; Aged; Polymorphism
Summary White matter lesions, also termed White Matter Hyperintensities (WMH), on T2weighted MR images, are common in the elderly population. Of note, their presence is often accompanied with cognitive decline and the risk of dementia. Even though previous brain ischemia and WM lesion studies have been conducted and indicated that brain-derived neurotrophic factor (BDNF) might protect against neuronal cell death, the interaction between regional WMH volume and the BDNF Val66Met polymorphism on the cognitive performance of healthy elderly population remains unclear. To investigate the genetic effect of BDNF on cognitive function and regional WMH in the healthy elderly population, 90 elderly men, without dementia, with a mean age of 80.6 5.6 y/o were recruited to undergo cognitive tests, structural magnetic
* Corresponding author at: Department of Psychiatry, Taipei Veterans General Hospital, No. 201, Shih-Pai Road, Sec. 2, 11217 Taipei, Taiwan. Tel.: +886 2 2875 7027x276; fax: +886 2 2872 5643. ** Corresponding author at: Institute of Neuroscience, National Yang-Ming University, 155, Li-Nong Street, 112, Taipei, Taiwan. Tel.: +886 2 2826 7338; fax: +886 2 2826 2285. E-mail addresses: [email protected] (S.-J. Tsai), [email protected], [email protected], [email protected] (C.-P. Lin). 0306-4530/$ — see front matter # 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.psyneuen.2013.09.027
BDNF Val66Met effects on WMH and cognition in healthy elderly
95
resonance imaging (MRI) scans, and genotyping of BDNF alleles. Compared with Met homozygotes, Val homozygotes showed significantly inferior short-term memory (STM) performance (P = .001). A tendency toward dose-dependent effects of the Val allele on WMH volume was found, and Val homozygotes showed larger WMH volume in the temporal (P = .035), the occipital (P = .006), and the global WMH volume (P = .025) than others. Significant interaction effects of BDNF genotypes with temporal WMH volume on STM performance was observed (F1,89 = 4.306, P = .041). Val homozygotes presented steeper negative correlation compared to Met carriers. Mediation analysis also demonstrated that WMH in temporal, limbic, and subcortical regions might mediate the relationship between BDNF’s genetic effect and STM performance. Our findings supported the hypothesis that the BDNF Val66Met polymorphism may affect susceptibility to regional WMH volume and such genotype-by-WMH interaction effect is correlated with cognitive decline in non-demented elderly males, in which the Met allele plays a protective role. # 2013 Elsevier Ltd. All rights reserved.
1. Introduction Cerebral white matter hyperintensities (WMH) are areas of increased intensity observed on T2-weighted MR images and common in the non-demented elderly population. They are potentially a result from ischemia due to chronic microvascular disease or hypoperfusion and imply white matter damage (Fernando et al., 2006; Pantoni and Garcia, 1997). Many studies have indicated that WMH may be involved in the normal aging process, coupled with cognitive decline (Sachdev et al., 2007, 2008; Smith et al., 2011). They are even a potential risk factor for dementia (Brun and Englund, 1986; Smith et al., 2008). Moreover, early studies support the hypothesis that increased regional WMH volume in deep or periventricular areas, may be associated with emotion or cognitive function in the elderly (Kim et al., 2008; Kohama et al., 2011; Ota et al., 2009). Clinical-anatomical correlation further indicates that by affecting the connection of cortical and subcortical structures, the location of regional WMHs may relate to specific executive and episodic memory dysfunction (Smith et al., 2011). Thus, when it comes to the impact of WMH on cognitive impairment during aging process, locations of increased WMH volume should be the first sign to be noted and examined further. The prevalence of WMH is associated with several risk factors. In addition to age and vascular diseases, genetic effect has also been considered as a potential risk factor for increased WMH volume. One of the genes implicated is brainderived neurotrophic factor (BDNF) (Taylor et al., 2008). BDNF is a homodimeric neurotrophic factor that is crucial for neuronal differentiation and survival during embryonic development, and for the maintenance of neuron viability in adulthood, in both the central and peripheral nervous systems (Jones and Reichardt, 1990; Leibrock et al., 1989). In humans, the BDNF gene maps to chromosome 11p13 and is composed of six 50 exons that are differentially spliced to a single 30 terminal exon that encodes the entire sequence of mature BDNF. A common single nucleotide polymorphism (SNP), consisting of a missense change (G196A), has been identified in the coding exon of the BDNF gene at position 66 (Val66Met, rs6265), and produces a valine to methionine substitution (Ventriglia et al., 2002). This SNP is thought to disrupt cellular processing, trafficking, and activity-dependent secretion of BDNF (Egan et al., 2003). In turn, this polymorphism of BDNF has been associated with the
modulation of activity-dependent synaptic plasticity, neurogenesis, learning, and memory processing in both in vitro and in vivo studies (Yamada et al., 2002). In addition, expression of BDNF was also suggested to have a neuroprotective role in cerebral ischemia and WM lesion volume in animal studies (Ferrer et al., 2001; Nomura et al., 2005). Studies of BDNF genotype on cognitive performance in the elderly, especially on memory function, are becoming increasingly common. (Egan et al., 2003; Harris et al., 2006). Despite the fact that both BDNF and WMH play important roles in the normal brain aging process, and are related to executive and memory functions, few studies have attempted to identify an association between BDNF, cognitive performance and regional WMH volume. Only one study has linked the effect of the BDNF Val66Met polymorphism to WMH volume in late-life depression (Taylor et al., 2008). However, the interaction between WMH volume and the effect of the BDNF Val66Met polymorphism on cognitive decline in healthy elderly participants has not been specifically evaluated. In this study, we have recruited healthy elderly males to verify the effect of the BDNF Val66Met polymorphism on cognitive performance, regional WMH volume and their interaction. To minimize the confounding effects of gender, a homogenous group was constructed, composed entirely of healthy elderly Han Chinese males. As mentioned before, BDNF is thought to exert beneficial effects on neuronal growth and repair, which may be associated with WMH volume, and may affect cognitive performances in healthy elders. In this study, we therefore seek to assess (1) the correlation between BDNF genotypes and neurocognitive performance, (2) the effect of BDNF genotype on regional WMH, and (3) the hypothesis that BDNF Val66Met polymorphism might interact with regional WMH volume and be related to cognitive decline in aged males without dementia.
2. Materials and methods 2.1. Participants This study involved one hundred and four elderly ethnic Chinese male participants, aged from 65 to 92 years old. They were recruited from a total of 857 residents in the veteran community home in northern Taiwan, the majority of the target population in this community being male.
96 Experiments were conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Taipei Veterans General Hospital. Written informed consent was obtained from all participants ensuring adequate understanding of the study. Any participants with the following psychological and physical conditions were excluded: (1) presence of any diagnosis on Axis I of DSM-IV, such as mood disorders or psychotic disorders; (2) presence of neurobiological disorders, such as dementia, head injury, stroke, or Parkinson’s disease; (3) presence of cerebrovascular risk factors, such as hypertension, diabetes, hyperlipidemia or coronary heart disease; (4) severe medical illness, such as malignancy, heart failure, or renal failure; (5) illiteracy; (6) presence of ferromagnetic foreign bodies or implants that were electrically, magnetically, or mechanically activated anywhere in the body.
2.2. Clinical assessments All participants possessed sufficient visual and auditory acuity for cognitive testing after MRI scanning. Their daily activities and cognitive functions were evaluated mainly with the Clinical Dementia Rating Scale (CDR) (Hughes et al., 1982). The Cognitive Abilities Screening Instrument (CASI2.0, Chinese version, Liu et al., 2002) and the Wechsler Digit Span Task test (forward and backward) were also conducted. The CASI test is a 100-point cognitive assessment that contains 9 domains of cognitive functions including long-term memory, short-term memory (STM), attention, concentration/mental manipulation, orientation, abstraction and judgment, language, visual construction, and list-generating fluency. The CASI cognitive test was designed for crosscultural studies and was adapted in Chinese for individuals with limited or no formal education. A study by Liu et al. (1994) found that sensitivity was 0.88 and specificity was 0.94 when using a cut-off of CASI score less than or equal to 50 for dementia. This exclusion criterion was set up for patients with a risk of dementia. Thus, any subjects with a CDR score higher than 0.5 or a CASI score less than 50 were excluded to avoid a possible confound from dementia.
2.3. Genotyping After MRI scanning, we used the polymerase chain reaction restriction and fragment length polymorphism (PCR-RFLP) method to genotype for the BDNF Val66Met polymorphism, from participant blood samples. Genomic DNA was extracted from leukocytes by standard DNA extraction procedures. The PCR amplifications were established with the primer pairs (50 ACTCTGGAGAGCGTGAAT-30 and 50 -ATACTGTCACACACGCTC30 ). PCR products were subsequently digested by the NlaIII restriction enzyme on 3% EtBr agarose gels. Fragments from RFLP can identify relevant genotypes. Contamination and partial digestion were controlled for by a blank and positive control in each group of experiments. Genotyping was processed blind to clinical data. Apolipoprotein E (APOE) was also genotyped to identify any possible effect of the APOE e4 allele.
2.4. MRI acquisition All MR scanning was performed on a 3.0T Siemens MRI scanner (Siemens Magnetom Tim Trio, Erlangen, Germany) with a
C.-C. Huang et al. 12-channel head coil, at National Yang-Ming University, Taiwan. For image registration, calculation of brain volumes, and brain mask generation, high-resolution structural T1-weighted MR images (T1w) were acquired with a 3D magnetizationprepared rapid gradient echo sequence (3D-MPRAGE; TR/TE = 2530/3.5 ms, TI = 1100 ms, FOV = 256 mm, flip angle = 78, matrix size = 256 256, 192 sagittal slices, voxel size = 1.0 mm 1.0 mm 1.0 mm, no gap). The T2-weighted fluid-attenuated inversion recovery (T2-FLAIR) images were acquired with multi-shot Turbo Spin Echo (TSE) sequences (2DBLADE; TR/TE = 9000/134 ms, TI = 2500 ms, FOV = 225, flip angle = 1408, matrix size = 320 320, 42 transversal slices, slice thickness = 3 mm, ETL = 35) for WMH volume calculation. All images were acquired parallel to the anterior commissureposterior commissure line. Each participant’s head was immobilized with cushions inside the coil to minimize motion artifacts generated during image acquisition.
2.5. Image analysis To increase the accuracy of WMH segmentation in T2-FLAIR images, skull stripping was included in following preprocessing steps. First, the Brain Extraction Tool (Smith, 2002) was used on T1w images to remove any non-brain components and to generate brain masks in native space. The brain masks were subsequently applied to co-registered T2-FLAIR images for skull and non-brain tissue stripping. Each skull-stripped T2-FLAIR image was registered to a corresponding T1w image with a linear registration approach using FMRIB’s Linear Image Registration Tool (Jenkinson and Smith, 2001). The co-registered and skull-stripped T2-FLAIR images were then used as initial input for the subsequent automatic WMH segmentation procedure. To optimize WMH segmentation on T2-FLAIR images, an intensity non-uniformity correction was applied. We adopted the non-parametric non-uniform intensity normalization (N3) approach (Sled et al., 1998) (iterations = 150, stop threshold = 0.0001, distance = 55 mm) on co-registered and skullstripped T2-FLAIR images. The FLAIR lesion segmentation toolbox (FLEX) was then used to segment WMH automatically with same protocols described in the previous WMH segmentation study (Gibson et al., 2010). Four steps were included in this automated protocol, including noise-reduction filtering, removal of clearly hyperintense voxels, two-class fuzzy Cmeans clustering, and thresholding to segment probable WMH. After segmentation, all WMH segmented images appeared in the same dimension as the native space T1w images. To investigate the correlation between BDNF genotype, regional WMH volume, and cognitive performance, we divided the whole brain, except the cerebellum, into 6 regions (bilateral frontal, parietal, limbic, subcortical, temporal, and occipital lobe) using WFU PickAtlas (Maldjian et al., 2003). To parcellate individual T1w images into various sub-divisions in native space, each inverse transformation matrix from MNI standard space to native space was calculated with the linear registration approach (FMRIB’s Linear Image Registration Tool v5.5; FLIRT). Each transformation matrix was applied to the parcellation atlas, to generate atlases in individual native space. These individual brain atlases were subsequently applied to corresponding WMH
BDNF Val66Met effects on WMH and cognition in healthy elderly
97
Figure 1 The flow diagram of current WMH preprocessing protocol. Individual T2-FLAIR images (B) were used to conduct linear registration with their T1 images (A). The co-registered T2-FLAIR images were used to perform WMH segmentation by the FLEX toolbox in individual T1 native space. An inverse transformation matrix derived from the registration between native T1 image and MNI space T1 template was used to transform the brain atlas (C) (L/R frontal area: violet/blue-violet; L/R parietal area: green/light-green; L/R temporal area: orange/red; L/R occipital area: cyan/blue-green; L/R limbic area: blue/light-blue; L/R subcortical area: yellow/ yellow-orange) from MNI standard space into individual native space (E). Individual brain atlases were finally used to perform the WMH localization (F). L, Left; R, right. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
segmented images to localize the WMH loci in native space. Global and regional WMH volumes in various sub-regions were subsequently extracted and calculated. To further control for the effects of brain size on global and regional WMH volume, the total intracranial volume (TIV) for each participant was also calculated using the VBM8 toolbox (http://dbm.neuro.uni-jena.de/) with default settings, running under SPM8 (http://www.fil.ion.ucl.ac.uk/ spm/software/spm8/). The automated WMH segmentation and localizing procedures are illustrated as a flow diagram in Fig. 1.
2.6. Statistical analysis Statistical analysis was performed by the Statistical Package for Social Sciences (SPSS) software package (SPSS 18 for Windows, Chicago, IL, USA). Demographic data were compared among three groups (Val/Val, Val/Met, Met/Met) with a chi-squared test (for categorical variables) or analysis of variance (ANOVA) (for continuous variables) to evaluate group differences. Because TIV was associated with the volume of WMH, we normalized each WMH volume by this variable to evaluate the effect of brain size (Smith et al., 2008) in the following statistical models.
Analysis of covariance (ANCOVA) was used to show the group differences in normalized WMH volume and neuropsychological performance, controlling for age as a nuisance variable. Furthermore, to evaluate the effect of BDNF genotype on the correlation between regional WMH volume and neuropsychological performance, univariate analysis using a general linear model was performed to examine if neuropsychological performance showed a linear interaction effect, predicted by the main effect of BDNF genotype (Met/Met, Val/Met, Val/Val), regional WMH volume and WMH-by-genotype interaction. To further elucidate the causal path of interaction models, mediation analyses were applied following the procedure of Baron and Kenny (1986). Firstly, simple correlations between BDNF genetic effect and cognitive performances, between BDNF and regional WMH volume, and between WMH volume and cognitive performances were examined. Finally, the relationship between BDNF genotype and cognitive performance, after controlling for WMH volume, was assessed. After the above steps, the indirect coefficients can be calculated to identify whether there is a significant directional relationship between the 3 variables. If BDNF and WMH volume both significantly predict cognitive performance, the results would support a partial mediation effect. All 2-tailed corrected P-values of 0.05 in post hoc analysis were adjusted for multiple comparisons by the Bonferroni correction in all models.
98 Table 1
C.-C. Huang et al. Demographic characteristics and cognitive assessments among BDNF Val66Met groups.
Demographics Mean (SD)
Met/Met (N = 24)
Val/Met (N = 46)
Val/Val (N = 20)
F or X 2
P-Value
Age (years) TIV (liter) Handedness (L/R) APOE (e4+/e4) CDR
79.9 (1.14) 1.50 (0.02) 0L/24R 2/22 0.14 (0.05)
80.9 (0.83) 1.47 (0.01) 3L/43R 8/38 0.22 (0.04)
80.7 (0.96) 1.48 (0.02) 0L/20R 3/17 0.30 (0.06)
0.300 1.435 5.005 1.054 2.159
0.742 0.244 0.287 0.591 0.122
Digit span task Forward Backward
12.29 (0.62) 4.25 (0.68)
12.32 (0.42) 3.58 (0.39)
11.05 (0.67) 3.25 (0.59)
1.412 0.178
0.249 0.838
CASI CASI total scores Long-term memory Short-term memory Attention Concentration/mental manipulation Orientation Abstraction and judgment Language Visual construction List-generating fluency
90.29 (1.63) 9.50 (0.25) 10.87 (0.21) 7.04 (0.19) 7.63 (0.42) 17.41 (0.43) 10.00 (0.47) 9.79 (0.13) 9.08 (0.36) 8.63 (0.45)
87.02 (1.42) 8.78 (0.21) 11.00 (0.17) 6.41 (0.18) 6.98 (0.39) 17.59 (0.18) 9.08 (0.31) 9.32 (0.18) 8.65 (0.30) 8.91 (0.25)
82.60 (2.19) 9.40 (0.26) 9.45 (0.52) a 6.15 (0.25) 6.25 (0.72) 17.70 (0.22) 8.65 (0.53) 9.00 (0.32) 8.00 (0.44) 7.90 (0.50)
2.875 2.685 7.935 3.017 0.786 0.195 1.390 1.761 0.997 1.759
0.062 0.074 0.001 0.054 0.459 0.824 0.255 0.178 0.373 0.178
All of the demographics were compared between groups with ANOVA tests and Chi-squared tests. Values are mean (SE). Abbreviations: TIV, total intracranial volume; CDR, Clinical Dementia Rating; CASI, cognitive abilities screening instrument; Met, methionine; Val, valine. a Represents significantly lower mean value than the Met/Met and Val/Met groups in post hoc test with Bonferroni correction.
3. Results 3.1. Demographics and neuropsychological performance The final study cohort included 90 male participants, after excluding 14 participants because of head motion or artifact problems in MR imaging. Genotyping data showed that 20 participants were Val-homozygotes, 46 participants were Val/Met heterozygotes, and 24 participants were Met-homozygotes. BDNF genotype distributions did not differ from the Hardy-Weinberg equilibrium (P = .818). Table 1 shows the demographics and neuropsychological scores of the study participants. No significant differences were observed in age (F 2,87 = .300, P = .742), TIV (F 2,87 = 1.435, P = .244), and handedness (F 2,87 = 5.005, P = .287) among the three genotype groups. Out of the cognition tests, only STM scores revealed significant differences among the three groups (F 2,87 = 8.283, P = .001). Post hoc analysis identified significant differences in STM performance between the Val/Val group and the other two groups (Met/Met, P = .006; Val/Met, P = .000). A dominant model of Met allele inheritance revealed that Met-carriers possessed significantly better STM performance than Val-homozygotes (P
Available online at www.sciencedirect.com
ScienceDirect j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / p s y n e u e n
Effect of BDNF Val66Met polymorphism on regional white matter hyperintensities and cognitive function in elderly males without dementia Chu-Chung Huang a,b, Mu-En Liu c, Kun-Hsien Chou b,i, Albert C. Yang d,e,f, Chia-Chun Hung b,g,h, Chen-Jee Hong d,e,g, Shih-Jen Tsai d,e,*, Ching-Po Lin a,b,g,i,** a
Department of Biomedical Imaging and Radiological Sciences, National Yang-Ming University, Taipei, Taiwan Brain Connectivity Lab, Institute of Neuroscience, National Yang-Ming University, Taipei, Taiwan c Department of Psychiatry, Kaohsiung Veterans General Hospital, Kaohsiung, Taiwan d Department of Psychiatry, Taipei Veterans General Hospital, Taipei, Taiwan e School of Medicine, National Yang-Ming University, Taipei, Taiwan f Center for Dynamical Biomarkers and Translational Medicine, National Central University, Chungli, Taiwan g Institute of Brain Science, National Yang-Ming University, Taipei, Taiwan h Department of Psychiatry, Taichung Veterans General Hospital, Taichung, Taiwan i Brain Research Center, National Yang-Ming University, Taiwan b
Received 19 March 2013; received in revised form 30 September 2013; accepted 30 September 2013
KEYWORDS Brain-derived neurotrophic receptor; White matter hyperintensities; Cognition; Aged; Polymorphism
Summary White matter lesions, also termed White Matter Hyperintensities (WMH), on T2weighted MR images, are common in the elderly population. Of note, their presence is often accompanied with cognitive decline and the risk of dementia. Even though previous brain ischemia and WM lesion studies have been conducted and indicated that brain-derived neurotrophic factor (BDNF) might protect against neuronal cell death, the interaction between regional WMH volume and the BDNF Val66Met polymorphism on the cognitive performance of healthy elderly population remains unclear. To investigate the genetic effect of BDNF on cognitive function and regional WMH in the healthy elderly population, 90 elderly men, without dementia, with a mean age of 80.6 5.6 y/o were recruited to undergo cognitive tests, structural magnetic
* Corresponding author at: Department of Psychiatry, Taipei Veterans General Hospital, No. 201, Shih-Pai Road, Sec. 2, 11217 Taipei, Taiwan. Tel.: +886 2 2875 7027x276; fax: +886 2 2872 5643. ** Corresponding author at: Institute of Neuroscience, National Yang-Ming University, 155, Li-Nong Street, 112, Taipei, Taiwan. Tel.: +886 2 2826 7338; fax: +886 2 2826 2285. E-mail addresses: [email protected] (S.-J. Tsai), [email protected], [email protected], [email protected] (C.-P. Lin). 0306-4530/$ — see front matter # 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.psyneuen.2013.09.027
BDNF Val66Met effects on WMH and cognition in healthy elderly
95
resonance imaging (MRI) scans, and genotyping of BDNF alleles. Compared with Met homozygotes, Val homozygotes showed significantly inferior short-term memory (STM) performance (P = .001). A tendency toward dose-dependent effects of the Val allele on WMH volume was found, and Val homozygotes showed larger WMH volume in the temporal (P = .035), the occipital (P = .006), and the global WMH volume (P = .025) than others. Significant interaction effects of BDNF genotypes with temporal WMH volume on STM performance was observed (F1,89 = 4.306, P = .041). Val homozygotes presented steeper negative correlation compared to Met carriers. Mediation analysis also demonstrated that WMH in temporal, limbic, and subcortical regions might mediate the relationship between BDNF’s genetic effect and STM performance. Our findings supported the hypothesis that the BDNF Val66Met polymorphism may affect susceptibility to regional WMH volume and such genotype-by-WMH interaction effect is correlated with cognitive decline in non-demented elderly males, in which the Met allele plays a protective role. # 2013 Elsevier Ltd. All rights reserved.
1. Introduction Cerebral white matter hyperintensities (WMH) are areas of increased intensity observed on T2-weighted MR images and common in the non-demented elderly population. They are potentially a result from ischemia due to chronic microvascular disease or hypoperfusion and imply white matter damage (Fernando et al., 2006; Pantoni and Garcia, 1997). Many studies have indicated that WMH may be involved in the normal aging process, coupled with cognitive decline (Sachdev et al., 2007, 2008; Smith et al., 2011). They are even a potential risk factor for dementia (Brun and Englund, 1986; Smith et al., 2008). Moreover, early studies support the hypothesis that increased regional WMH volume in deep or periventricular areas, may be associated with emotion or cognitive function in the elderly (Kim et al., 2008; Kohama et al., 2011; Ota et al., 2009). Clinical-anatomical correlation further indicates that by affecting the connection of cortical and subcortical structures, the location of regional WMHs may relate to specific executive and episodic memory dysfunction (Smith et al., 2011). Thus, when it comes to the impact of WMH on cognitive impairment during aging process, locations of increased WMH volume should be the first sign to be noted and examined further. The prevalence of WMH is associated with several risk factors. In addition to age and vascular diseases, genetic effect has also been considered as a potential risk factor for increased WMH volume. One of the genes implicated is brainderived neurotrophic factor (BDNF) (Taylor et al., 2008). BDNF is a homodimeric neurotrophic factor that is crucial for neuronal differentiation and survival during embryonic development, and for the maintenance of neuron viability in adulthood, in both the central and peripheral nervous systems (Jones and Reichardt, 1990; Leibrock et al., 1989). In humans, the BDNF gene maps to chromosome 11p13 and is composed of six 50 exons that are differentially spliced to a single 30 terminal exon that encodes the entire sequence of mature BDNF. A common single nucleotide polymorphism (SNP), consisting of a missense change (G196A), has been identified in the coding exon of the BDNF gene at position 66 (Val66Met, rs6265), and produces a valine to methionine substitution (Ventriglia et al., 2002). This SNP is thought to disrupt cellular processing, trafficking, and activity-dependent secretion of BDNF (Egan et al., 2003). In turn, this polymorphism of BDNF has been associated with the
modulation of activity-dependent synaptic plasticity, neurogenesis, learning, and memory processing in both in vitro and in vivo studies (Yamada et al., 2002). In addition, expression of BDNF was also suggested to have a neuroprotective role in cerebral ischemia and WM lesion volume in animal studies (Ferrer et al., 2001; Nomura et al., 2005). Studies of BDNF genotype on cognitive performance in the elderly, especially on memory function, are becoming increasingly common. (Egan et al., 2003; Harris et al., 2006). Despite the fact that both BDNF and WMH play important roles in the normal brain aging process, and are related to executive and memory functions, few studies have attempted to identify an association between BDNF, cognitive performance and regional WMH volume. Only one study has linked the effect of the BDNF Val66Met polymorphism to WMH volume in late-life depression (Taylor et al., 2008). However, the interaction between WMH volume and the effect of the BDNF Val66Met polymorphism on cognitive decline in healthy elderly participants has not been specifically evaluated. In this study, we have recruited healthy elderly males to verify the effect of the BDNF Val66Met polymorphism on cognitive performance, regional WMH volume and their interaction. To minimize the confounding effects of gender, a homogenous group was constructed, composed entirely of healthy elderly Han Chinese males. As mentioned before, BDNF is thought to exert beneficial effects on neuronal growth and repair, which may be associated with WMH volume, and may affect cognitive performances in healthy elders. In this study, we therefore seek to assess (1) the correlation between BDNF genotypes and neurocognitive performance, (2) the effect of BDNF genotype on regional WMH, and (3) the hypothesis that BDNF Val66Met polymorphism might interact with regional WMH volume and be related to cognitive decline in aged males without dementia.
2. Materials and methods 2.1. Participants This study involved one hundred and four elderly ethnic Chinese male participants, aged from 65 to 92 years old. They were recruited from a total of 857 residents in the veteran community home in northern Taiwan, the majority of the target population in this community being male.
96 Experiments were conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Taipei Veterans General Hospital. Written informed consent was obtained from all participants ensuring adequate understanding of the study. Any participants with the following psychological and physical conditions were excluded: (1) presence of any diagnosis on Axis I of DSM-IV, such as mood disorders or psychotic disorders; (2) presence of neurobiological disorders, such as dementia, head injury, stroke, or Parkinson’s disease; (3) presence of cerebrovascular risk factors, such as hypertension, diabetes, hyperlipidemia or coronary heart disease; (4) severe medical illness, such as malignancy, heart failure, or renal failure; (5) illiteracy; (6) presence of ferromagnetic foreign bodies or implants that were electrically, magnetically, or mechanically activated anywhere in the body.
2.2. Clinical assessments All participants possessed sufficient visual and auditory acuity for cognitive testing after MRI scanning. Their daily activities and cognitive functions were evaluated mainly with the Clinical Dementia Rating Scale (CDR) (Hughes et al., 1982). The Cognitive Abilities Screening Instrument (CASI2.0, Chinese version, Liu et al., 2002) and the Wechsler Digit Span Task test (forward and backward) were also conducted. The CASI test is a 100-point cognitive assessment that contains 9 domains of cognitive functions including long-term memory, short-term memory (STM), attention, concentration/mental manipulation, orientation, abstraction and judgment, language, visual construction, and list-generating fluency. The CASI cognitive test was designed for crosscultural studies and was adapted in Chinese for individuals with limited or no formal education. A study by Liu et al. (1994) found that sensitivity was 0.88 and specificity was 0.94 when using a cut-off of CASI score less than or equal to 50 for dementia. This exclusion criterion was set up for patients with a risk of dementia. Thus, any subjects with a CDR score higher than 0.5 or a CASI score less than 50 were excluded to avoid a possible confound from dementia.
2.3. Genotyping After MRI scanning, we used the polymerase chain reaction restriction and fragment length polymorphism (PCR-RFLP) method to genotype for the BDNF Val66Met polymorphism, from participant blood samples. Genomic DNA was extracted from leukocytes by standard DNA extraction procedures. The PCR amplifications were established with the primer pairs (50 ACTCTGGAGAGCGTGAAT-30 and 50 -ATACTGTCACACACGCTC30 ). PCR products were subsequently digested by the NlaIII restriction enzyme on 3% EtBr agarose gels. Fragments from RFLP can identify relevant genotypes. Contamination and partial digestion were controlled for by a blank and positive control in each group of experiments. Genotyping was processed blind to clinical data. Apolipoprotein E (APOE) was also genotyped to identify any possible effect of the APOE e4 allele.
2.4. MRI acquisition All MR scanning was performed on a 3.0T Siemens MRI scanner (Siemens Magnetom Tim Trio, Erlangen, Germany) with a
C.-C. Huang et al. 12-channel head coil, at National Yang-Ming University, Taiwan. For image registration, calculation of brain volumes, and brain mask generation, high-resolution structural T1-weighted MR images (T1w) were acquired with a 3D magnetizationprepared rapid gradient echo sequence (3D-MPRAGE; TR/TE = 2530/3.5 ms, TI = 1100 ms, FOV = 256 mm, flip angle = 78, matrix size = 256 256, 192 sagittal slices, voxel size = 1.0 mm 1.0 mm 1.0 mm, no gap). The T2-weighted fluid-attenuated inversion recovery (T2-FLAIR) images were acquired with multi-shot Turbo Spin Echo (TSE) sequences (2DBLADE; TR/TE = 9000/134 ms, TI = 2500 ms, FOV = 225, flip angle = 1408, matrix size = 320 320, 42 transversal slices, slice thickness = 3 mm, ETL = 35) for WMH volume calculation. All images were acquired parallel to the anterior commissureposterior commissure line. Each participant’s head was immobilized with cushions inside the coil to minimize motion artifacts generated during image acquisition.
2.5. Image analysis To increase the accuracy of WMH segmentation in T2-FLAIR images, skull stripping was included in following preprocessing steps. First, the Brain Extraction Tool (Smith, 2002) was used on T1w images to remove any non-brain components and to generate brain masks in native space. The brain masks were subsequently applied to co-registered T2-FLAIR images for skull and non-brain tissue stripping. Each skull-stripped T2-FLAIR image was registered to a corresponding T1w image with a linear registration approach using FMRIB’s Linear Image Registration Tool (Jenkinson and Smith, 2001). The co-registered and skull-stripped T2-FLAIR images were then used as initial input for the subsequent automatic WMH segmentation procedure. To optimize WMH segmentation on T2-FLAIR images, an intensity non-uniformity correction was applied. We adopted the non-parametric non-uniform intensity normalization (N3) approach (Sled et al., 1998) (iterations = 150, stop threshold = 0.0001, distance = 55 mm) on co-registered and skullstripped T2-FLAIR images. The FLAIR lesion segmentation toolbox (FLEX) was then used to segment WMH automatically with same protocols described in the previous WMH segmentation study (Gibson et al., 2010). Four steps were included in this automated protocol, including noise-reduction filtering, removal of clearly hyperintense voxels, two-class fuzzy Cmeans clustering, and thresholding to segment probable WMH. After segmentation, all WMH segmented images appeared in the same dimension as the native space T1w images. To investigate the correlation between BDNF genotype, regional WMH volume, and cognitive performance, we divided the whole brain, except the cerebellum, into 6 regions (bilateral frontal, parietal, limbic, subcortical, temporal, and occipital lobe) using WFU PickAtlas (Maldjian et al., 2003). To parcellate individual T1w images into various sub-divisions in native space, each inverse transformation matrix from MNI standard space to native space was calculated with the linear registration approach (FMRIB’s Linear Image Registration Tool v5.5; FLIRT). Each transformation matrix was applied to the parcellation atlas, to generate atlases in individual native space. These individual brain atlases were subsequently applied to corresponding WMH
BDNF Val66Met effects on WMH and cognition in healthy elderly
97
Figure 1 The flow diagram of current WMH preprocessing protocol. Individual T2-FLAIR images (B) were used to conduct linear registration with their T1 images (A). The co-registered T2-FLAIR images were used to perform WMH segmentation by the FLEX toolbox in individual T1 native space. An inverse transformation matrix derived from the registration between native T1 image and MNI space T1 template was used to transform the brain atlas (C) (L/R frontal area: violet/blue-violet; L/R parietal area: green/light-green; L/R temporal area: orange/red; L/R occipital area: cyan/blue-green; L/R limbic area: blue/light-blue; L/R subcortical area: yellow/ yellow-orange) from MNI standard space into individual native space (E). Individual brain atlases were finally used to perform the WMH localization (F). L, Left; R, right. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
segmented images to localize the WMH loci in native space. Global and regional WMH volumes in various sub-regions were subsequently extracted and calculated. To further control for the effects of brain size on global and regional WMH volume, the total intracranial volume (TIV) for each participant was also calculated using the VBM8 toolbox (http://dbm.neuro.uni-jena.de/) with default settings, running under SPM8 (http://www.fil.ion.ucl.ac.uk/ spm/software/spm8/). The automated WMH segmentation and localizing procedures are illustrated as a flow diagram in Fig. 1.
2.6. Statistical analysis Statistical analysis was performed by the Statistical Package for Social Sciences (SPSS) software package (SPSS 18 for Windows, Chicago, IL, USA). Demographic data were compared among three groups (Val/Val, Val/Met, Met/Met) with a chi-squared test (for categorical variables) or analysis of variance (ANOVA) (for continuous variables) to evaluate group differences. Because TIV was associated with the volume of WMH, we normalized each WMH volume by this variable to evaluate the effect of brain size (Smith et al., 2008) in the following statistical models.
Analysis of covariance (ANCOVA) was used to show the group differences in normalized WMH volume and neuropsychological performance, controlling for age as a nuisance variable. Furthermore, to evaluate the effect of BDNF genotype on the correlation between regional WMH volume and neuropsychological performance, univariate analysis using a general linear model was performed to examine if neuropsychological performance showed a linear interaction effect, predicted by the main effect of BDNF genotype (Met/Met, Val/Met, Val/Val), regional WMH volume and WMH-by-genotype interaction. To further elucidate the causal path of interaction models, mediation analyses were applied following the procedure of Baron and Kenny (1986). Firstly, simple correlations between BDNF genetic effect and cognitive performances, between BDNF and regional WMH volume, and between WMH volume and cognitive performances were examined. Finally, the relationship between BDNF genotype and cognitive performance, after controlling for WMH volume, was assessed. After the above steps, the indirect coefficients can be calculated to identify whether there is a significant directional relationship between the 3 variables. If BDNF and WMH volume both significantly predict cognitive performance, the results would support a partial mediation effect. All 2-tailed corrected P-values of 0.05 in post hoc analysis were adjusted for multiple comparisons by the Bonferroni correction in all models.
98 Table 1
C.-C. Huang et al. Demographic characteristics and cognitive assessments among BDNF Val66Met groups.
Demographics Mean (SD)
Met/Met (N = 24)
Val/Met (N = 46)
Val/Val (N = 20)
F or X 2
P-Value
Age (years) TIV (liter) Handedness (L/R) APOE (e4+/e4) CDR
79.9 (1.14) 1.50 (0.02) 0L/24R 2/22 0.14 (0.05)
80.9 (0.83) 1.47 (0.01) 3L/43R 8/38 0.22 (0.04)
80.7 (0.96) 1.48 (0.02) 0L/20R 3/17 0.30 (0.06)
0.300 1.435 5.005 1.054 2.159
0.742 0.244 0.287 0.591 0.122
Digit span task Forward Backward
12.29 (0.62) 4.25 (0.68)
12.32 (0.42) 3.58 (0.39)
11.05 (0.67) 3.25 (0.59)
1.412 0.178
0.249 0.838
CASI CASI total scores Long-term memory Short-term memory Attention Concentration/mental manipulation Orientation Abstraction and judgment Language Visual construction List-generating fluency
90.29 (1.63) 9.50 (0.25) 10.87 (0.21) 7.04 (0.19) 7.63 (0.42) 17.41 (0.43) 10.00 (0.47) 9.79 (0.13) 9.08 (0.36) 8.63 (0.45)
87.02 (1.42) 8.78 (0.21) 11.00 (0.17) 6.41 (0.18) 6.98 (0.39) 17.59 (0.18) 9.08 (0.31) 9.32 (0.18) 8.65 (0.30) 8.91 (0.25)
82.60 (2.19) 9.40 (0.26) 9.45 (0.52) a 6.15 (0.25) 6.25 (0.72) 17.70 (0.22) 8.65 (0.53) 9.00 (0.32) 8.00 (0.44) 7.90 (0.50)
2.875 2.685 7.935 3.017 0.786 0.195 1.390 1.761 0.997 1.759
0.062 0.074 0.001 0.054 0.459 0.824 0.255 0.178 0.373 0.178
All of the demographics were compared between groups with ANOVA tests and Chi-squared tests. Values are mean (SE). Abbreviations: TIV, total intracranial volume; CDR, Clinical Dementia Rating; CASI, cognitive abilities screening instrument; Met, methionine; Val, valine. a Represents significantly lower mean value than the Met/Met and Val/Met groups in post hoc test with Bonferroni correction.
3. Results 3.1. Demographics and neuropsychological performance The final study cohort included 90 male participants, after excluding 14 participants because of head motion or artifact problems in MR imaging. Genotyping data showed that 20 participants were Val-homozygotes, 46 participants were Val/Met heterozygotes, and 24 participants were Met-homozygotes. BDNF genotype distributions did not differ from the Hardy-Weinberg equilibrium (P = .818). Table 1 shows the demographics and neuropsychological scores of the study participants. No significant differences were observed in age (F 2,87 = .300, P = .742), TIV (F 2,87 = 1.435, P = .244), and handedness (F 2,87 = 5.005, P = .287) among the three genotype groups. Out of the cognition tests, only STM scores revealed significant differences among the three groups (F 2,87 = 8.283, P = .001). Post hoc analysis identified significant differences in STM performance between the Val/Val group and the other two groups (Met/Met, P = .006; Val/Met, P = .000). A dominant model of Met allele inheritance revealed that Met-carriers possessed significantly better STM performance than Val-homozygotes (P
