Now showing 1 - 4 of 4
ItemNo Preview AvailableBrain Aging in Major Depressive Disorder: Results From the ENIGMA MDD ConsortiumSchmaal, L ; Han, L ; Dinga, R ; Thompson, P ; Veltman, D ; Penninx, B (ELSEVIER SCIENCE INC, 2018-05-01)Background: Major Depressive Disorder has been associated with accelerated biological aging. From a brain perspective, normal aging is associated with significant loss of grey matter and depression may have an accelerating effect on age-related brain atrophy. Here, data on brain aging in MDD from the ENIGMA MDD Working Group will be presented. Methods: A normative model of brain-based age was devel- oped in 4708 healthy controls by applying a Gaussian Process Regression analysis with 10-fold cross-validation to estimate chronological age from structural MRI scans, separately for males and females. This model was then applied to 2924 MDD individuals to predict their brain-based age. Accelerated brain aging was measured as the difference between predicted brain-based age and actual chronological age (brain age gap). Results: The brain age model explained 92% and 93% of the age variance in female and male healthy controls, respectively. The mean absolute error (MAE) was 6.79 years in females and 6.60 in males. Application of the model to MDD patients showed a mean brain age gap of 0.75 years in females (MAE¼6.82) and 0.64 in males (MAE¼6.68), which were significantly lower than brain age gap estimates in healthy controls in both females (F(1,4379)¼6.10,P¼0.01) and males (F(1,3166)¼4.07,P¼0.04). Our preliminary analysis also showed greater brain age gap associations with various clinical characteristics. Conclusions: We found preliminary evidence for accelerated brain aging in MDD, however, the brains of patients were estimated to be only <1 years older than healthy controls. The impact of different methods, feature selection and potential confounding effects will also be discussed.
ItemNo Preview AvailableMETHYLOME-WIDE ASSOCIATION STUDIES FOR MAJOR DEPRESSIVE DISORDER IN BLOOD OVERLAP WITH METHYLATION RESULTS FROM BRAIN AND LARGE-SCALE GWASAberg, K ; Dean, B ; Shabalin, A ; Zhao, M ; Chan, R ; Hattab, M ; van Grootheest, G ; Han, L ; Aghajani, M ; Milaneschi, Y ; Jansen, R ; Xie, L ; Clark, S ; Penninx, B ; van den Oord, E (ELSEVIER SCIENCE BV, 2019-01-01)Background: Epigenetic modifications such as DNA methy- lation provide stability and diversity to the cellular phenotype and aberrant methylation has been implicated in processes underlying psychiatric disorders. Therefore, studies combining DNA methylation and genotype information provide a promis- ing approach to study disorders where genotype information alone has failed to reveal the full etiology. Methods: We applied an optimized MBD-seq protocol to assay the complete CpG methylome in cases with Major Depressive Disorder (MDD) and controls using blood samples (N=1,132) from Netherlands Study of Depression and Anxiety and brain samples (N=64) from the Victorian Brain Bank Network. Data were analyzed with RaMWAS, a novel Biocon- ductor package specifically designed for Methylome-Wide Association Studies (MWAS). To study the overlap between top MWAS findings in blood and brain, we used a permutation based enrichment test (shiftR) that accounted for the depen- dency between adjacent CpG sites. Furthermore, we utilized the methylation data in combination with existing genotype information from the same individuals in a MWAS of CpGs created or destroyed by SNPs. Next, we tested whether top results from this CpG-SNP MWAS overlapped with recent large- scale GWAS to identify robust associations with genomic loci of importance for MDD etiology Results: The MWAS in blood identified five methylome- wide significant sites (P o 5 10-8) from three distinct loci and 472 nominally significant (P o 1 10-5) CpG sites. To study the robustness of the overall MWAS signal, we used an “in-sample” replication based on k-fold cross validation. Results showed that the findings replicated (P = 4.0 10- 10). When we compared blood and brain we found that top blood MWAS findings were significantly enriched for top CpGs in the brain MWAS (P = 5.4 10-3). The MWAS of CpG-SNPs identified 32 nominally significant sites and in- sample replication showed that the signal replicated (P = 2.2 10-8). Finally, the top CpG-SNP MWAS showed a consistent trend towards enrichment in all tested large- scale GWAS, with the most significant enrichment observed for the 23andMe study (P = 4.9 10-3). This overlap involved 55 genes that were overrepresented (P o 0.01) in 12 level-5 gene ontology terms, of which a major portion was related to neuronal regulation, function and development. Discussion: This work involves the largest MWAS for MDD performed to date. Our integrated analysis with brain tissue, genotype information and GWAS results highlighted biological functions of potential value for MDD etiology. Part of the associated methylation marks in blood overlapped with MWAS finding in brain. As blood can easily be collected in a clinical setting, these loci may be of direct value as potential diagnostic biomarkers for MDD.
ItemNo Preview AvailableDeviations From Normative Age-Brain Associations in Over 3,000 Individuals With Major Depressive DisorderSchmaal, L ; Han, L ; Bayer, J ; Marquand, A ; Dinga, R ; Cole, J ; Hahn, T ; Penninx, B ; Veltman, D ; Thompson, P (ELSEVIER SCIENCE INC, 2019-05-15)Background: Major depressive disorder (MDD) is a complex heterogeneous disorder. Identifying brain alterations as indi- vidual deviations from normative patterns of brain-age asso- ciations, instead of patient group mean differences, can provide important insights into heterogeneous patterns of brain abnormalities observed in MDD. Methods: We estimated normative models of (1) age pre- dicting individual structural brain measures, and (2) structural brain measures predicting age (Brain Age model) using ma- chine learning in healthy individuals (N¼2,515) from the ENIGMA MDD consortium. We applied model parameters to independent samples of healthy individuals (N¼2,513) and MDD patients (N¼3,433) to obtain predicted values of brain structure (model 1) and age (model 2). Z-scores quantifying differences between predictive and true values were calcu- lated, representing individual deviations from the normative range. Results: The estimated normative models showed good model fit in the training sample; e.g. a correlation of R¼0.86 between actual and predicted age for the Brain Age Model, and good generalization to independent healthy and MDD samples. We identified heterogeneous patterns of brain deviations in MDD patients (model 1). Patients with more extreme deviations showed different clinical characteristics compared to patients residing within the normative range. Additionally, patients were estimated on average w1 year older than controls (model 2), but we also observed large between-person variation in brain age gaps. Further ana- lyses showed associations between brain age gap and clinical symptoms. Conclusions: Our work shows substantial heterogeneity in deviations from normal age-related variation in brain structure in individuals with MDD. The impact of and solutions for con- founding effects of scan site will also be discussed.
ItemNo Preview AvailableMeta-Analysis of Hippocampal Subfields: Results From the ENIGMA-MDD Working GroupSaemann, P ; Czisch, M ; Jahanshad, N ; Whelan, CD ; van Velzen, L ; Hibar, D ; Han, L ; Veer, IM ; Walter, H ; Veltman, D ; Schmaal, L (ELSEVIER SCIENCE INC, 2019-05-15)Background Hippocampal volume reductions in major depressive disorder (MDD) represent a robust finding in retrospective meta-analyses. Subregional specificity of this finding has been suspected from several smaller previous studies. Given the complex role of the hippocampus both for stress response regulation and its vulnerability to chronic disease, we aim at finer mapping of this result using FreeSurfer based, automated subfield segmentation. Methods Twenty-three centers with MDD/control samples contributed. Results reported here stem from 2522 patients and 4244 controls. After segmentation and standardized QC, local statistical were run for 25 models in total. Key models were: Cases vs. controls (covarying for age, age squared, sex-by-age, sex-by-age-squared, ICV and scanner/site); recurrent vs. controls, first episode vs. controls, early onset (EO, <22 years) vs. controls, late onset (LO) vs. controls. Eventually, inverse variance-weighted random-effect meta-analysis model in R (metafor package) with FDR correction for 14 phenotypes was performed. Results Regional specificity of volume deficits were detected in MDD as a whole (2522 patients, 4244 controls) (CA3>whole>CA1>GC.ML.DG>CA4>molecular layer). No robust effects were found in first episode patients (743 patients, 3812 controls) except for nominal effects. In recurrent MDD, only CA1 effects were robust. EO depression showed unexpectedly strong effects (836 patients, 3472 controls). Similarly, patients with current AD treatment showed strong effects, similarly distributed as in MDD except for CA1. No correlation with depression severity was detected. Conclusions Hippocampal structural changes in MDD show subregion specificity. While first episode status seems less critical and first/recurrent episode patients are similar, early onset appears as key predictor of structural abnormalities.