Recent Publications

"I'm always looking for the creative spark. Always." - Jimmy Page


BioRxiv (Dec 2019)

Transcriptomic analysis of brain hypometabolism links the cholinergic and gingipain hypotheses of Alzheimer's disease pathogenesis

Patel S, Howard D, Man A, Schwartz D, Jee J, Felsky D, Pausova Z, Paus T, French L

Alzheimer's disease (AD) starts decades before clinical symptoms appear. Low glucose utilization in regions of the cerebral cortex mark early AD and is informative for clinical diagnosis. To identify these specific regions, we conducted a meta-analysis of positron emission tomography studies that compared AD patients with healthy controls. Using the Allen Human Brain Atlas, we then identified genes with expression patterns associated with this hypometabolism map. Of the six brains in the Atlas, one demonstrated a strong spatial association with the hypometabolism pattern. Within this brain, genes encoding cytosolic ribosome proteins are highly expressed in the hypometabolic regions. Analyses of human and mouse data show that expression of these genes increases across AD-associated states of microglial activation, is high in acetylcholine-rich brain regions and neurons, and is up-regulated in inflamed gingival tissue. Taken together, our molecular characterization of cortical hypometabolism links the cholinergic and gingipain hypotheses of AD.


Biological Psychiatry (March 2019)

Polygenic Risk and Neural Substrates of Attention-Deficit/Hyperactivity Disorder Symptoms in Youths With a History of Mild Traumatic Brain Injury

Stojanovski S, Felsky D, Viviano J, Shahab S, Bangali R, Burton C, Dalsgaard S, Szatmari P, Chakravarty MM, Ameis S, Schachar R, Voinesekos AN, Wheeler AL

BACKGROUND: Attention-deficit/hyperactivity disorder (ADHD) is a major sequela of traumatic brain injury (TBI) in youths. The objective of this study was to examine whether ADHD symptoms are differentially associated with genetic risk and brain structure in youths with and without a history of TBI. METHODS: Medical history, ADHD symptoms, genetic data, and neuroimaging data were obtained from a community sample of youths. ADHD symptom severity was compared between those with and without TBI (TBI n = 418, no TBI n = 3193). The relationship of TBI history, genetic vulnerability, brain structure, and ADHD symptoms was examined by assessing 1) ADHD polygenic score (discovery sample ADHD n = 19,099, control sample n = 34,194), 2) basal ganglia volumes, and 3) fractional anisotropy in the corpus callosum and corona radiata. RESULTS: Youths with TBI reported greater ADHD symptom severity compared with those without TBI. Polygenic score was positively associated with ADHD symptoms in youths without TBI but not in youths with TBI. The negative association between the caudate volume and ADHD symptoms was not moderated by a history of TBI. However, the relationship between ADHD symptoms and structure of the genu of the corpus callosum was negative in youths with TBI and positive in youths without TBI. CONCLUSIONS: The identification of distinct ADHD etiology in youths with TBI provides neurobiological insight into the clinical heterogeneity in the disorder. Results indicate that genetic predisposition to ADHD does not increase the risk for ADHD symptoms associated with TBI. ADHD symptoms associated with TBI may be a result of a mechanical insult rather than neurodevelopmental factors.


Neurobiology of Aging (Feb 2018)

Genetic influence of plasma homocysteine on Alzheimer's disease

Roostaei T, Felsky D, Nazeri A, Voineskos AN, for the International Genomics of Alzheimer’s Project (IGAP), Alzheimer’s Disease Neuroimaging Initiative (ADNI), Alzheimer’s Disease Genomics Consortium (ADGC), Genotype-Phenotype Alzheimer's disease Associations (GenADA), and Religious Orders Study and Rush Memory and Aging Project (ROS/MAP) Study Groups

Observational studies have consistently reported elevated plasma homocysteine as a risk factor for Alzheimer's disease (AD). However, results from clinical trials of homocysteine-lowering treatments are inconsistent. This discrepancy may be explained by a lack of causal association between homocysteine and AD. Mendelian randomization studies have the potential to provide insight into the causality of this association through studying the effect of genetic predisposition to high homocysteine on AD. Our analyses using summarized (n = 54,162) and individual participant (n = 6987) data from Caucasian participants did not show an effect of plasma homocysteine genetic risk on susceptibility to AD. Although with smaller sample sizes, further subanalyses also did not support an effect of genetically determined plasma homocysteine on cognitive impairment and decline, beta-amyloid and tau pathology and gray matter atrophy in AD. However, we found associations with tau tangle burden (n = 251) and gray matter atrophy (n = 605) in cognitively normal elderly. Our results do not support a causal association between elevated homocysteine and risk, severity, and progression of AD. However, the relationship between genetically determined homocysteine and brain pathology in cognitively normal elderly requires further exploration.


Alzheimer's and Dementia (Oct 2017)

Genetic Epistasis Regulates Amyloid Deposition in Resilient Aging

Felsky D, Xu J, Chibnik, L, Knight, J, Schneider JA, Kennedy JL, the Alzheimer’s Disease Neuroimaging Initiative, Bennett DA, De Jager PL

The brain-derived neurotrophic factor (BDNF) interacts with important genetic Alzheimer’s disease (AD) risk factors. Specifically, variants within the SORL1 gene determine BDNF’s ability to reduce Aβ in vitro. We sought to test whether functional BDNF variation interacts with SORL1 genotypes to influence expression and downstream AD-related processes in humans.We analyzed postmortem brain RNA-sequencing and neuropathological data for 441 subjects from the Religious Orders Study/Memory and Aging Project, and molecular and structural neuroimaging data for 1 285 subjects from the Alzheimer’s Disease Neuroimaging Initiative.We found one SORL1 RNA transcript strongly regulated by SORL1-BDNF interactions in elderly without pathological AD, and showing stronger associations with diffuse than neuritic Aβ plaques. The same SORL1-BDNF interactions also significantly influenced Aβ load as measured with [18F]Florbetapir PET.Our results bridge the gap between risk and resilience factors for AD, demonstrating interdependent roles of established SORL1 and BDNF functional genotypes.

Using Transcriptomic Hidden Variables to Infer Context-Specific Genotype Effects in the Brain1-s2_edited.jpg

American Journal of Human Genetics (Aug 2019)

Using Transcriptomic Hidden Variables to Infer Context-Specific Genotype Effects in the Brain

Ng B, Casazza W, Patrick E, Tasaki S, Novakovsky G, Felsky D, Ma Y, Bennett DA, Gaiteri C, De Jager PL, Mostafavi S

Deciphering the environmental contexts at which genetic effects are most prominent is central for making full use of GWAS results in follow-up experiment design and treatment development. However, measuring a large number of environmental factors at high granularity might not always be feasible. Instead, here we propose extracting cellular embedding of environmental factors from gene expression data by using latent variable (LV) analysis and taking these LVs as environmental proxies in detecting gene-by-environment (GxE) interaction effects on gene expression, i.e., GxE expression quantitative trait loci (eQTLs). Applying this approach to two largest brain eQTL datasets (n = 1,100), we show that LVs and GxE eQTLs in one dataset replicate well in the other dataset. Combining the two samples via meta-analysis, 895 GxE eQTLs are identified. On average, GxE effect explains an additional ∼4% variation in expression of each gene that displays a GxE effect. Ten of these 52 genes are associated with cell-type-specific eQTLs, and the remaining genes are multi-functional. Furthermore, after substituting LVs with expression of transcription factors (TF), we found 91 TF-specific eQTLs, which demonstrates an important use of our brain GxE eQTLs.


Scientific Data (Aug 2018)

A multi-omic atlas of the human frontal cortex for aging and Alzheimer's disease research

De Jager PL, Ma Y, McCabe C, Xu J, Vardarajan BN, Felsky D, Klein HU, White CC, Petters MA, Lodgson B, Nejad P, Tang, A, Mangravite LM, Yu L, Gaiteri, C, Mostafavi S, Schneider JA, Bennett, DA

We initiated the systematic profiling of the dorsolateral prefrontal cortex obtained from a subset of autopsied individuals enrolled in the Religious Orders Study (ROS) or the Rush Memory and Aging Project (MAP), which are jointly designed prospective studies of aging and dementia with detailed, longitudinal cognitive phenotyping during life and a quantitative, structured neuropathologic examination after death. They include over 3,322 subjects. Here, we outline the first generation of data including genome-wide genotypes (n=2,090), whole genome sequencing (n=1,179), DNA methylation (n=740), chromatin immunoprecipitation with sequencing using an anti-Histone 3 Lysine 9 acetylation (H3K9Ac) antibody (n=712), RNA sequencing (n=638), and miRNA profile (n=702). Generation of other omic data including ATACseq, proteomic and metabolomics profiles is ongoing. Thanks to its prospective design and recruitment of older, non-demented individuals, these data can be repurposed to investigate a large number of syndromic and quantitative neuroscience phenotypes. The many subjects that are cognitively non-impaired at death also offer insights into the biology of the human brain in older non-impaired individuals.


PLOS Medicine (April 2017)

Identification of genes associated with dissociation of cognitive performance and neuropathological burden: Multistep analysis of genetic, epigenetic, and transcriptional data

White C, Yang HS, Yu L, Chibnik L, Dawe R, Yang, J, Klein HU, Felsky D, Afanakis K, Honer W, Sperling R, Schneider J, Bennett DA, De Jager PL

The molecular underpinnings of the dissociation of cognitive performance and neuropathological burden are poorly understood, and there are currently no known genetic or epigenetic determinants of the dissociation.“Residual cognition” was quantified by regressing out the effects of cerebral pathologies and demographic characteristics on global cognitive performance proximate to death. To identify genes influencing residual cognition, we leveraged neuropathological, genetic, epigenetic, and transcriptional data available for deceased participants of the Religious Orders Study (n = 492) and the Rush Memory and Aging Project (n = 487). Given that our sample size was underpowered to detect genome-wide significance, we applied a multistep approach to identify genes influencing residual cognition, based on our prior observation that independent genetic and epigenetic risk factors can converge on the same locus. In the first step (n = 979), we performed a genome-wide association study with a predefined suggestive p < 10−5, and nine independent loci met this threshold in eight distinct chromosomal regions. Three of the six genes within 100 kb of the lead SNP are expressed in the dorsolateral prefrontal cortex (DLPFC): UNC5C, ENC1, and TMEM106B. In the second step, in the subset of participants with DLPFC DNA methylation data (n = 648), we found that residual cognition was related to differential DNA methylation of UNC5C and ENC1 (false discovery rate < 0.05). In the third step, in the subset of participants with DLPFC RNA sequencing data (n = 469), brain transcription levels of UNC5C and ENC1 were evaluated for their association with residual cognition: RNA levels of both UNC5C (estimated effect = −0.40, 95% CI −0.69 to −0.10, p = 0.0089) and ENC1 (estimated effect = 0.0064, 95% CI 0.0033 to 0.0096, p = 5.7 × 10−5) were associated with residual cognition. In secondary analyses, we explored the mechanism of these associations and found that ENC1 may be related to the previously documented effect of depression on cognitive decline, while UNC5C may alter the composition of presynaptic terminals. Of note, the TMEM106B allele identified in the first step as being associated with better residual cognition is in strong linkage disequilibrium with rs1990622A (r2 = 0.66), a previously identified protective allele for TDP-43 proteinopathy. Limitations include the small sample size for the genetic analysis, which was underpowered to detect genome-wide significance, the evaluation being limited to a single cortical region for epigenetic and transcriptomic data, and the use of categorical measures for certain non-amyloid-plaque, non-neurofibrillary-tangle neuropathologies.Through a multistep analysis of cognitive, neuropathological, genomic, epigenomic, and transcriptomic data, we identified ENC1 and UNC5C as genes with convergent genetic, epigenetic, and transcriptomic evidence supporting a potential role in the dissociation of cognition and neuropathology in an aging population, and we expanded our understanding of the TMEM106B haplotype that is protective against TDP-43 proteinopathy.


Nature Communications (Jan 2019)

Neuropathological correlates and genetic architecture of microglial activation in elderly human brain

Felsky D, Roostaei T, Nho K, Risacher SL, Bradshaw EM, Petyuk V, Schneider JA, Saykin A, Bennett DA, De Jager PL

Microglia, the resident immune cells of the brain, have important roles in brain health. However, little is known about the regulation and consequences of microglial activation in the aging human brain. Here we report that the proportion of morphologically activated microglia (PAM) in postmortem cortical tissue is strongly associated with β-amyloid, tau-related neuropathology, and the rate of cognitive decline. Effect sizes for PAM measures are substantial, comparable to that of APOE ε4, the strongest genetic risk factor for Alzheimer's disease, and mediation models support an upstream role for microglial activation in Alzheimer's disease via accumulation of tau. Further, we identify a common variant (rs2997325) influencing PAM that also affects in vivo microglial activation measured by [11C]-PBR28 PET in an independent cohort. Thus, our analyses begin to uncover pathways regulating resident neuroinflammation and identify overlaps of PAM's genetic architecture with those of Alzheimer's disease and several other traits.


Molecular Neurodegeneration (July 2018)

Polygenic analysis of inflammatory disease variants and effects on microglia in the aging brain

Felsky D, Patrick E, Schneider JA, Mostafavi S, Gatieri, C, Patsopoulos N, Bennett DA, De Jager PL

BACKGROUND: The role of the innate immune system in Alzheimer's disease (AD) and neurodegenerative disease susceptibility has recently been highlighted in genetic studies. However, we do not know whether risk for inflammatory disease predisposes unaffected individuals to late-life cognitive deficits or AD-related neuropathology. We investigated whether genetic risk scores for seven immune diseases and central nervous system traits were related to cognitive decline (nmax = 1601), classical AD neuropathology (nmax = 985), or microglial density (nmax = 184).

METHODS: Longitudinal cognitive decline, postmortem amyloid and tau neuropathology, microglial density, and gene module expression from bulk brain tissue were all measured in participants from two large cohorts (the Rush Religious Orders Study and Memory and Aging Project; ROS/MAP) of elderly subjects (mean age at entry 78 +/- 8.7 years). We analyzed data primarily using robust regression methods. Neuropathologists were blind to clinical data.

RESULTS: The AD genetic risk scores, including and excluding APOE effects, were strongly associated with cognitive decline in all domains (min Puncor = 3.2 × 10- 29). Multiple sclerosis (MS), Parkinson's disease, and schizophrenia risk did not influence cognitive decline in older age, but the rheumatoid arthritis (RA) risk score alone was significantly associated with microglial density after correction (t146 = - 3.88, Puncor = 1.6 × 10- 4). Post-hoc tests found significant effects of the RA genetic risk score in multiple regions and stages of microglial activation (min Puncor = 1.5 × 10- 6). However, these associations were driven by only one or two variants, rather than cumulative polygenicity. Further, individual MS (Pone-sided < 8.4 × 10- 4) and RA (Pone-sided = 3 × 10- 4) variants associated with higher microglial density were also associated with increased expression of brain immune gene modules.

CONCLUSIONS: Our results demonstrate that global risk of inflammatory disease does not strongly influence aging-related cognitive decline but that susceptibility variants that influence peripheral immune function also alter microglial density and immune gene expression in the aging brain, opening a new perspective on the control of microglial and immune responses within the central nervous system. Further study on the molecular mechanisms of peripheral immune disease risk influencing glial cell activation will be required to identify key regulators of these pathways.


Cerebral Cortex (May 2018)

BDNF-Dependent Effects on Amygdala–Cortical Circuitry and Depression Risk in Children and Youth

Wheeler AL, Felsky D, Viviano JD, Ameis SH, Szatmari P, Lerch JP, Chakravarty MM, Voineskos AN

The brain-derived neurotrophic factor (BDNF) is critical for brain development, and the functional BDNF Val66Met polymorphism is implicated in risk for mood disorders. The objective of this study was to determine how the Val66Met polymorphism influences amygdala–cortical connectivity during neurodevelopment and assess the relevance for mood disorders. Age- and sex-specific effects of the BDNF Val66Met polymorphism on amygdala–cortical connectivity were assessed by examining covariance of amygdala volumes with thickness throughout the cortex in a sample of Caucasian youths ages 8–22 that were part of the Philadelphia Neurodevelopmental Cohort (n = 339). Follow-up analyses assessed corresponding BDNF genotype effects on resting-state functional connectivity (n = 186) and the association between BDNF genotype and major depressive disorder (MDD) (n = 2749). In adolescents, amygdala–cortical covariance was significantly stronger in Met allele carriers compared with Val/Val homozygotes in amygdala–cortical networks implicated in depression; these differences were driven by females. In follow-up analyses, the Met allele was also associated with stronger resting-state functional connectivity in adolescents and increased likelihood of MDD in adolescent females. The BDNF Val66Met polymorphism may confer risk for mood disorders in females through effects on amygdala–cortical connectivity during adolescence, coinciding with a period in the lifespan when onset of depression often occurs, more commonly in females.

©2020 by Whole Person and Population Modelling. Proudly created with