Genetics and Brain Morphology

Electronic Supplementary Material 3-5

Genetics and brain morphology

Lachlan T. Strike1*, Baptiste Couvy-Duchesne1, Narelle K. Hansell1, Gabriel Cuellar-Partida, Sarah E. Medland, Margaret J. Wright

Lachlan T. Strike, Baptiste Couvy-Duchesne

Neuroimaging Genetics, QIMR Berghofer Medical Research Institute, Brisbane QLD 4006, Australia

School of Psychology, University of Queensland, Brisbane QLD 4072, Australia

Centre for Advanced Imaging, University of Queensland, Brisbane QLD 4072, Australia

Narelle K. Hansell

Neuroimaging Genetics, QIMR Berghofer Medical Research Institute, Brisbane QLD 4006, Australia

Gabriel Cuellar-Partida

Statistical Genetics, QIMR Berghofer Medical Research Institute, Brisbane QLD 4006, Australia

Sarah E. Medland

Quantitative Genetics, QIMR Berghofer Medical Research Institute, Brisbane QLD 4006, Australia

Margaret J. Wright

Neuroimaging Genetics, QIMR Berghofer Medical Research Institute, Brisbane QLD 4006, Australia

School of Psychology, University of Queensland, Brisbane QLD 4072, Australia

1 LS, BCD and NH contributed equally to this work.

* Corresponding author

Lachlan T. Strike

en

Online Resource 3.APOEGenotype Effects on Neuroimaging Structural Phenotypes: Overview of Studies January 2014 to July 2014

Study / Samplea Size
(Age) / Structural
Phenotypeb / Finding
Babies and Toddlers
Dean et al. (2014) / N=59
(6-25 months) / voxel-based gray matter volume (GMV) and white matter (WM) myelin water fraction (MWF) / ɛ4 carriers relative to non-carriers had lower GMV and MWF (an indicator of myelin sheath integrity) in brain regions preferentially affected by AD, as well asgreater GMV and MWF in extensive frontal regions. No GMV results survived correction for multiple comparisons. However, multiple MWF measures survived correction for multiple comparisons (decreased in corticospinal tract (p≤4.0x10-4), but increased in frontal (p=2.0x10-5), parietal (p=1.0x10-5), superior corona radiata (p≤2.0x10-5), inferior fronto-occiptal fasciculus (p=3.0x10-4), and genu of the corpus callosum (p=5.0x104), in ε4 carriers compared to non-carriers).
Knickmeyer et al. (2014) / N=272
(neonates) / automated region of interest (ROI) GMV / Reduced volumes found in temporal regions known to be vulnerable to atrophy in the elderly, but also regions of increased volume (in parietal, frontal, and occipital cortices) in ɛ4 carriers compared to non-carriers(multiple ROIs retained significance after correction for multiple testing with p≤1.0x10-4).
Korja et al. (2013) / N=322
(neonates) / unspecified brain pathology / Among this sample of very preterm and/or very low birthweight babies, no genotype effects were found to be associated with brain pathology
Healthy Young & Middle-Aged Adults
Dowell et al. (2013) / N=41
(18-30 years) / multiple WM measures / ɛ4 allele was associated with higher WM volume ratio (p=0.022) and increases in axial diffusivity(p=0.011) and mode of anisotropy (p˂0.05).
Goveas et al. (2013) / N=46
(44-65 years) / voxel-based GMV / No genotype differences found.
Patel et al. (2013) / N=36
(M=45 years) / WM fractional anisotropy / No genotype differences found.
Healthy Elderly Adults
Akoudad et al. (2013) / N=4493
(M=63.9 yrs) / multiple WM measures / The presence of lobar microbleeds was associated with loss of WM structural integrity in ɛ4 carriers, but not in non-carriers (p=0.022 (fractional anisotropy); p=0.002 (mean diffusivity)).
de Jong et al. (2014) / N=361
(~71-99 yrs) / temporal lobe & basal ganglia volume / Longitudinal Study:ɛ4 carriers relative to non-carriers showed greater medial temporal lobe (MTL) atrophy over 2.4 years (p˂1.0x10-4). Vascular risk factors independently influenced MTL atrophy.
Desikan et al. (2013) / N=107
(M=75.9 yrs) / entorhinal cortex volume / Longitudinal Study: ɛ4 status was associated with amyloid-beta deposition, but not with amyloid-beta associated volume loss, which only occurred in the presence of phosphor-tau.
Salminen et al. (2013) / N=64
(aged 50+) / WM fiber bundles / ɛ4 carriersfound to have shorter WM fiber bundles in left uncinate fasciculus compared to non-carriers (p=0.038).
Zade et al. (2013) / N=1969
(M=61 years) / multiple brain volumes / Greater central obesity was associated with reduced frontal lobe volume (p=0.047), greater WM hyperintensity volume (p=0.033) and worse cognitive function in ɛ4 carriers, but not in non-carriers.
Zhu et al. (2013) / N=323
(M=73.3 yrs) / brain infarcts / In ɛ4 carriers, but not non-carriers, the activity of serum amyloid-beta degrading proteases was associated with small vessel, but not large vessel, infarcts (p˂1.0x10-4).
Mild Cognitive Impairment (MCI) and Alzheimer’s Disease (AD) Patients
Goni et al. (2013) / N=65 MCI
(M=73.4 yrs) / voxel-wise
GMV / ɛ4 allele was associated with regions of reduced volume with notable right hemisphere vulnerability (p’s<0.005).
Hostage et al. (2014) / N=237 MCI
(M=79.9 yrs) / GMV regions associated with AD / Longitudinal Study:13 of 15 regions showed higher atrophy over 12-48 months in ɛ4 carriers relative to non-carriers, with largest effects found in amygdalae (baseline atrophy rate = 1.51% loss in volume per annum; additional effect of ɛ4 = 2.40% loss pa; p<0.001) and hippocampi (baseline atrophy rate = 2.76% loss in volume per annum; additional effect of ɛ4 = 1.53% loss pa;p<0.001) – p values survive correction for multiple testing.
Kerchner et al. (2014) / N=39 AD, MCI, controls
(M=70.6 yrs) / hippocampal subfields / Dose-dependent association of ɛ4 allele with greater thinning of the CA1 apical neuropil – a hippocampal subregion vulnerable to AD (p=0.014) and worse episodic memory – potentially causally related.
Pievani et al. (2013) / N=72 AD, controls
(M=69.7 yrs) / striatum GMV and surface area / APOEgenotype was found to have little or only small effects on striatal atrophy. However, sample sizes for APOE subgroups were small.
Yoon et al. (2013) / N=1472 MCI
(M=70.0 yrs: Korean) / WM hyperintensities (WMH) / Greater severity of WMH was associated with more rapid decline in cognitive function in ɛ4 carriers compared to non-carriers.
Conversion from MCI to AD
Erten-Lyons et al. (2013) / N=71:
20 converted
(M=94.7 at death) / volume trajectories (ventricular, total brain, hippocampal) / Longitudinal Study: Brain scans were collected within 36 months of death and again at autopsy. Ventricular volume trajectory, which correlated positively with ɛ4 status (p<0.00l), was found to be a more sensitive marker of accruing AD than total brain or hippocampal volume trajectories
Morgen et al. (2013) / N=82:
34 converted
(M=69.3 yrs) / voxel-based GMV / Longitudinal Study:APOE genotype was associated with distinct patterns of gray matter atrophy over ~2 years in those converting: more pronounced occipital atrophy predicting conversion in ɛ4 carriers, and frontoparietal atrophy (+ worse executive function) predicting conversion in non-carriers (corrected p-values ranged 0.04-0.001).
Pereira et al. (2014) / N=1147 AD,
MCI, controls
(M=75.1 yrs) / medial temporal lobe volume / APOE genotype can influence the classification of medial temporal lobe atrophy (used as a marker of AD) and must be taken into account to avoid misdiagnosis.
Other Patient Groups
Chung et al. (2013) / N=100
Parkinson’s
(M=68.9 yrs; Korean) / voxel-based
GMV / There are suggestions that ɛ4 status may influence risk of MCI and AD in Parkinson’s patients, with ɛ4 carriers found to have more severe cortical atrophy in the left parahippocampal gyrus compared to non-carriers (uncorrected cluster threshold of P<0.001).
Fukazawa et al. (2013) / N=175
diabetes
(M=76.8 yrs; Japanese) / medial temporal lobe volume and markers of cerebrovascular disease / Among AD patients with diabetes, a subgroup appear to be better characterized by diabetes-related metabolic abnormalities (with clinical features including lower ɛ4 allele frequency and less severe medial temporal lobe atrophy) and may require alternative therapeutic approaches.
Hoare et al. (2013) / N=45 HIV
(18-35 years; African) / WM integrity of corpus callosum / ɛ4 carriers had greater memory impairment and white matter damage among stage III HIV-positive individuals than non-carriers (survived correction for multiple comparisons, p≤0.007).
McDonald et al. (2013) / N=79
breast cancer
(M=49.9 yrs) / voxel-based GMV / Longitudinal study: Gray matter density was examined before, and again one month after, chemotherapy in breast cancer patients. Changes did not differ for APOE genotype.
Methodological
Hua et al. (2013) / N=468 AD, MCI, controls
(M=75.3 yrs) / Temporal lobe ROI volumes (cumulative atrophy) / Longitudinal Study:Studyassessed biases that can arise when tracing brain changes in longitudinal imaging and showed that smaller sample sizes are needed to track AD progression if data is stratified by APOE ɛ4 status. A faster rate of atrophy was observed in ɛ4 carriers relative to non-carriers among AD, MCI, and healthy control groups.
Shi et al. (2013) / N=843 AD, MCI, controls
(55-90 years) / hippocampal volume / A newly developed automated registration system is introduced. It outperformed two other systems (FIRST, SPHARM) in assessing the hippocampus, and successfully showed APOE effects (regions of smaller volume in ɛ4 carriers versus non-carriers) in individuals with MCI and in healthy controls (p=0.0014).
Gene-Gene Interaction
Goni et al. (2013) / N=65 MCI
(M=73.4 yrs) / voxel-wise
GMV / APOE-MAPT: ɛ4 carriers who were also MAPT H1/H1 had smaller frontal and parietal regions compared to other genotype combinations (p<0.005).

NOTE: PubMed search:((Brain) OR ("white matter") OR ("gray matter") OR ("grey matter")) AND (*MRI OR DTI OR FA) AND (gene OR genes) AND (APOE) AND (“2013/01/01"[Date - Publication] : "2014/07/15"[Date - Publication]) NOT (genome-wide OR “genome wide” OR GWA OR GWAS). The search was further filtered for human samples and English language and only candidate gene studies were retained.

aAs most samples are Caucasian (or predominantly Caucasian), ancestry is only noted when it is known to differ.

bROI = Region of interest, GMV = gray matter volume; WM = white matter; AD = Alzheimer’s disease, MCI = mild cognitive impairment

Online Resource 4.BDNF Val66Met Genotype Effects on Neuroimaging Structural Phenotypes: Overview of Studies January 2014 to July 2014

Study / Samplea Size
(Age) / Structural
Phenotypeb / Finding
Healthy Individuals
Brooks et al. (2014) / N=345
(70-75 years) / voxel-based GM, WM and CSF volumes / Elderly Met66 carriers had better working memory performance compared to Val66 homozygotes, as well as larger gray matter volumes in memory, decision-making, and motor regions (e.g. prefrontal cortex,p<0.001), with reduced volumes in somatorsensory/arousal regions (e.g. occipito-temporal lobe,p=0.001), leading the authors to propose that their better performance may be due to greater cognitive control and reduced arousal interference.
Forde et al. (2014) / N=60
(18-55 years) / multiple voxel-based GM and WM measures / The met-dose effect in relation to brain morphology was examined and an “inverted-U” shaped profile rather than the expected linear effect was found for gray matter volume (including the cerebellum: val/met > val/val (p=0.006) and met/met (p=0.09)) and fractional anisotropy (right dorsal cingulum bundle: val/met > met/met(p=0.02) and val/val (p=0.26)). **bonferonni corrected**
Harrisberger et al. (2014) / 1. N=5298
2. N=643
(M=22.9 yrs) / hippocampalvolume / 1. Meta-analysis: Data from 32 samples were analysed and a weak effect showing Met66 carriers having smaller volumes relative to non-carriers was identified (p=0.04,g=0.09 (Hedges’s g values above 0.2, 0.5, and 0.8 correspond to small medium, and large effect sizes respectively)). However, evidence of a publication bias among studies using manual tracing was observed, and after excluding these studies, no genotype effect was found.
2. No genotype effect observed.
Huang et al. (2014) / N=90
(65-92 years;
Chinese) / WM hyperintensity (WMH) volume / Val66 homozygotes had inferior memory performance and larger WMH volumes than Met66 carriers (temporal, P=0.035; occipital, P=0.006; global WMH, P=0.025).
Knickmeyer et al. (2014) / N=272
(neonates) / automated region of interest (ROI) GMV / Val158 homozygotes relative to Met158 carriers had multiple clusters of increased and/or reduced volumes in regions including occipital and primary motor and somatosensory areas(multiple ROIs retained significance after correction for multiple testing with p≤1.0x10-4).
Lim et al. (2013) / N=165
(M=71.4 yrs) / hippocampal volume / Longitudinal (3 year follow-up): APOE genotype interacted with amyloid-beta. Among individuals with high amyloid-beta (but not low amyloid-beta), Val66 homozygotes showed reduced cognitive decline and less hippocampal atrophy over 3 years than Met66 carriers (cohen’s d = 0.73 (95% CIs 0.08, 1.35)). Also, among Met66 carriers, atrophy was reduced in those with low amyloid-beta (cohen’s d = 1.11 (95% CIs 0.46, 1.73)).
Tost et al. (2013) / N=85
(M=33.5 yrs) / multiple WM measures / Decreases in fractional anisotropy and widespread increases in radial diffusivity, particularly in prefrontal and occipital pathways were found in Val66 homozygotes compared to Met66 carriers (P<0.05). Authors proposed that increased axonal branching, as shown in Met/Met mice compared to wild type Val/Val (Cao et al. 2007),may underlie diminished radial diffusivity in human Met66 carriers.
C. Wang et al. (2014) / N=280
(18-30 years; Han Chinese) / cortical surface area / Met66 was dose-related to larger surface area for the anterior insular cortex (p<0.001).
Alzheimer’s Disease (AD), Mild Cognitive Impairment (MCI), Major Depressive Disorder (MDD), and Other Patient Groups
Cardoner et al. (2013) / N=37 MDD
(44-75 years) / regional GM volumes / Met66 carriers had regions of reduced volume in the left hippocampus (P=0.003) but larger orbitofrontal cortex volumes(P0.0005) compared to Val66 homozygotes.
Hayashi et al. (2014) / N=30 MDD
(M=44 years) / multiple WM measures / Val66Met did not differentiate between temporal lobe white matter abnormalities found in patients with first episode, treatment-naïve, MDDand controls.
Honea et al. (2013) / N=645 AD, MCI, controls
(55-90 years) / hippocampal and whole brain volume / Longitudinal (381 with 2 year follow-up): The influence of 8 BDNF SNPs was examined. Six were associated with hippocampal and/or whole brain atrophy over two years. Met66 carriers showed greater right hippocampal atrophy than Val66 homozygotes (only found in controls, N=127, p=0.027). However, cross-sectional analyses suggested Val66 homozygotes had greater whole brain atrophy with increasing age than Met66 carriers (N=578, p<0.005).
Mueller et al. (2013) / N=102
anxiety, controls
(M=13.3 yrs) / voxel-based GM volumes / Among adolescents with an anxiety disorder (N=39), Met66 carriers had smaller anterior insula (p=0.004, Cohen’s d = 0.79) and posterior insula volumes (p=0.008, Cohen’s d = 0.57) compared to Val66 homozygotes, while no genotype-related differences were found in healthy controls (N=63).
Mon et al. (2013) / N=62 alcohol dependent
(M=50.8 yrs) / GM, WM, and CSF volumes / Longitudinal (41 with 5 week follow-up during abstinence): This is thefirst study examining Val66Met effects on recovery of brain tissue volume during short-term abstinence. Genotype was associated with different recovery patterns. Val66 homozygotes showed increases in gray matter volumes (e.g. for frontal GM % change = 0.30 for Val/Met, p=ns; % change = 1.20 for Val/Val, p<0.001). However, Met66 carriers showed increases predominantly in white matter volumes(e.g. for frontal WM % change = 1.49 for Val/Met, p=0.020; % change = -0.07 for Val/Val, p=ns). Notably, only gray matter increases were associated with improvements in neurocognitive measures. No genotype effects were seen in cross-sectional analyses.
Suarez-Pinilla et al. (2013) / N=123
psychosis, controls
(M=29.6 yrs) / total brain tissue, plus GM, WM, ventricles, & CSF volumes / Longitudinal (3 year follow-up): No genotype effects were found for brain volume changes in first episode non-affective psychosis patients and controls.
Taylor et al. (2013) / N=62 MDD,
controls
(aged 60+) / WM lesion ratio / A trend for greater lesion ratio in the frontal corpus callosum in Met66 carriers compared to Val66 homozygotes was identified (P=0.0375).
Interaction with Environmental Factors
Healthy Individuals
Brown et al. (2014) / N=114
(60+ years) / ROI brain volumes / Higher levels of physical activity were associated with larger hippocampal and temporal lobe volumes in Val66 homozygotes, but smaller temporal lobe volume in Met66 carriers (hippocampus: p=0.02, standardised β = -0.22; temporal lobe: p=0.003, standardised β = -0.28 (significance threshold adjusted for multiple comparisonsp≤0.0083)).
Rabl et al. (2014) / N=153
(18-45 years) / hippocampal volume / The association between number of stressful life events and hippocampal volume in healthy individuals was positive in Met66 carriers, but negative in Val66 homozygotes (p=0.003, corrected for multiple comparisons).
Patients (psychosis, depression)
Aas et al. (2013) / N=106
Psychosis
(N=32.7 yrs) / ventricular and hippocampal volumes / Childhood abuse was negatively associated with cognitive performanceand right hippocampal volumein Met66 carriers, but not Val homozygotes(p=0.0050.008 respectively, significant after correction for multiple testing).
Carballedo et al. (2013) and Frodl et al. (2014) / N=133 MDD,
controls
(18-65 years) / hippocampal volume / Among those with no history of childhood adversity (CA), Met66 carriers (both MDD and controls) had larger left hippocampal volumes than Val66 homozygotes (p=0.042, corrected for multiple comparisons). However, having a history of CA, relative to none, was associated with reduced left hippocampal volume in Met66 carriers (p=0.028), but not in Val66homozygotes.
(Hernaus et al. (2014)) / N=184
psychosis, controls
(M=28.9) / hippocampal volume / A history of childhood trauma wastentatively associated with increased hippocampal volume in Met66 carriers relative to Val66 homozygotes among patients, but the opposite (decreased volume) among their non-psychotic siblings. However, substance abuse was found to be confounding factor and no genotype effects were found after controlling for substance abuse (compared to patients, alcohol consumption was higher, but cannabis and other drug use was lower, in healthy siblings).
Gene-Gene Interactions
Healthy Individuals
Rabl et al. (2014) / N=153
(18-45 years) / hippocampal volume / COMT-BDNF-SLC6A4: No gene-gene interactions were observed. However, a cumulative risk score (reflecting the number of COMT Met158 alleles, SLC6A4 5-HTTLPR Short alleles, and BDNF Val66 homozygotes) was highly predictive of hippocampal volume in interaction with number of adverse life events (p1.0x10-5, β=-0.41).

NOTE: PubMed search:((Brain) OR ("white matter") OR ("gray matter") OR ("grey matter")) AND (*MRI OR DTI OR FA) AND (gene OR genes) AND (BDNF) AND (“2013/01/01"[Date - Publication] : "2014/07/15"[Date - Publication]) NOT (genome-wide OR “genome wide” OR GWA OR GWAS). The search was further filtered for human samples and English language and only candidate gene studies were retained.

aAs most samples are Caucasian (or predominantly Caucasian), ancestry is only noted when it is known to differ.

bROI = region of interest, GM = gray matter; WM = white matter; CSF = cerebrospinal fluid, AD = Alzheimer’s disease, MCI = mild cognitive impairment

Online Resource 5.COMT Val158Met Genotype Effects on Neuroimaging Structural Phenotypes: Overview of Studies January 2014 to July 2014

Study / Samplea Size
(Age) / Structural
Phenotypeb / Finding
Healthy Individuals
Knickmeyer et al. (2014) / N=272
(neonates) / automated region of interest (ROI) GMV / Val158 homozygotes relative to Met158 carriers had multiple clusters of increased and/or reduced volumes in regions including temporal, parietal, occipital and supplementary motor areas(multiple ROIs retained significance after correction for multiple testing with p≤1.0x10-4).
Y. Wang et al. (2013) / N=320
(19-21 years, Chinese) / hippocampal volume / Both working memory performance and hippocampal volume were greater in Val158 homozygotes, with Met158 alleles having a negative load effect (right hippocampal volume P=0.019, β=-0.118)
Major Depressive Disorder (MDD) and Other Patient Groups
Hayashi et al. (2014) / N=30 MDD, controls
(M=44 years) / multiple WM measures / MDD was associated with white matter changes, with decreased temporal lobe fractional anisotropy and axial diffusivity in Met158 carriers with MDD compared with sex- and age-matched controls (P<0.05).
Kim et al. (2013) / N=52 panic disorder, controls
(18-60 years, Korean) / multiple WM measures / Increased fractional anisotropy was found in right hemisphere clusters in Met158 carriers compared to Val158 homozygotes, for patients (corrected p<0.05), but not for matched controls.
Seok et al. (2013) / N=148 MDD,
controls
(M=43.6 yrs, Korean) / multiple WM measures / Reduced fractional anisotropy (temporal, frontal, cingulum) was found for MDD Val158 homozygotes compared to Met158 carriers but not controls (e.g. bilateral frontal WM and genu of corpus callosum, p<0.001). Threshold free cluster enhancement corrected p value (family-wise error corrected p<0.01).
Zhang et al. (2013) / N=274 substance abuse, controls
(M=32.6 yrs) / multiple WM measures / In the context of addiction, Met158 homozygotes showed greater vulnerability to drug-related changes in prefrontal fractional anisotropy (e.g.p=0.009, corrected for family-wise error).
Gene-Gene Interaction
Healthy Individuals
Brauns et al. (2013) / N=94
(18-60 years) / cortical thickness / COMT-GAD1: Cortical thickness (left parahippocampal gyrus) was found to be reduced by 7.5% inGAD1 G allele homozygotes, relative to A allele carriers (p=0.008), but only in the presence of the COMT Val158 allele (p=0.006), reflecting and interaction between the GABAergic and dopaminergic systems. Results corrected for multiple comparisons.
Radua et al. (2014) / N=91
(M=33 years) / fronto-limbic GM volumes / COMT-SLC6A4: Hippocampal and amygdalar volumes were smaller in individuals with both COMT Met158 and 5HTTLPR Short alleles, or COMT Val158 and r-HTTLPR Long alleles (e.g. left parahippocampal gyrus: p<0.001), compared to other allelic combinations. Bonferroni-Holm correction for multiple comparisons.
Rabl et al. (2014) – see Gene-Gene Interaction, COMT-BDNF-SLC6A4, Table 5

NOTE: PubMed search:((Brain) OR ("white matter") OR ("gray matter") OR ("grey matter")) AND (*MRI OR DTI OR FA) AND (gene OR genes) AND (COMT) AND (“2013/01/01"[Date - Publication] : "2014/07/15"[Date - Publication]) NOT (genome-wide OR “genome wide” OR GWA OR GWAS). The search was further filtered for human samples and English language and only candidate gene studies were retained.