Introduction: Genome-wide association studies (GWAS) of brain volumes have identified common genetic variants with modest effect sizes that lie mainly in non-coding regions. We sought to identify novel variants, including rare and low-frequency variants, influencing brain volumes using whole-genome sequence association analyses .

Methods: We analyzed up to 7,607 participants (57% women; 62% European ancestry, 21% African-Americans, 15% Hispanic/Latino, 2% Chinese-American), mean age of 60.5 (16.2), from eight population- or family-based studies (FHS, GENESTAR, CHS, GENOA, ARIC, CARDIA, MESA, and SAFS) from the Trans-Omics for Precision Medicine (TOPMed) Program. We excluded participants with dementia, stroke, presence of large brain infarcts, tumor or low-quality scans. We tested the association of hippocampal (HV), total brain (TBV), lateral ventricular (LVV) and intracranial (ICV) volumes with individual genetic variants passing minor allele counts ≥ 15. Mixed-effect linear regression models were adjusted for age, age2, sex, study and the first 10 principal components. Models for HV, TBV and log(LVV) were additionally adjusted for ICV. We accounted for relatedness and trait variance heterogeneity using an empirical kinship matrix and a random effect for study.

Results: We detected one novel region with low-frequency variants associated with HV (13q14, P=5.8x10-9). The top variant (rs115674829) lies in LINC00598 at 237kb from FOXO1, a member of the forkhead family of transcription factors that have been linked to Alzheimer’s Disease. Additionally, we detected new suggestive associations (P≤10-7) for TBV (16p11) and LVV (1q25, 2q22, 3q13, 5q14, and 10q23), including rare and common variants. Finally, we confirmed the association of common variants in GWAS loci for all traits.

Conclusions: Our whole genome sequence analyses revealed intriguing new loci of low-frequency and common variants, and replicated loci previously associated with brain volumes. Future work will include ancestry-specific and conditional analyses, as well as gene-based and scan tests.

Background: Genome-wide association studies (GWAS) of brain volumes have identified common genetic variants with modest effect sizes that lie mainly in non-coding regions. We sought to identify low frequency and rare variants influencing brain volumes by performing association analyses using whole-genome sequence data from the Trans-Omics for Precision Medicine (TOPMed) Program.

Methods: We analyzed up to 7,607 participants (57% women; 62% European ancestry, 21% African-Americans, 15% Hispanic/Latino, 2% Chinese-American), mean age of 60.5 (16.2), from eight TOPMed population- or family-based studies (FHS, GENESTAR, CHS, GENOA, ARIC, CARDIA, MESA, and SAFS). We excluded participants with dementia, stroke, presence of large brain infarcts, tumor or low-quality scans. We tested the association of hippocampal (HV), total brain (TBV), lateral ventricular (LVV) and intracranial (ICV) volumes to individual genetic variants with minor allele counts ≥ 15 using mixed-effect linear regression models adjusted for age, age2, sex, study and the first 10 principal components. Models for HV, TBV and log(LVV) were adjusted for ICV. We accounted for relatedness using an empirical kinship matrix and trait variance variability by using a random effect for study.

Results: We detected one novel region with low frequency variants associated with HV (13q14, P=5.8x10-9). The top 13q14 variant for HV (rs115674829) minor allele frequency (MAF) was 2% in our pooled sample but was more common in the pan-African 1000 Genomes population (MAF 14%). This variant lies in LINC00598 at 237kb from FOXO1, a member of the forkhead family of transcription factors that has been linked to Alzheimer’s Disease. Additionally, we detected new suggestive associations (P≤10-7) for TBV (16p11) and LVV (1q25, 2q22, 3q13, 5q14, and 10q23), including common variants. Finally, we confirmed the association of common variants in GWAS loci for all traits.

Conclusions: Our whole genome sequence analyses revealed intriguing new loci of low-frequency and common variants, and replicated loci previously associated with brain volumes. Future work will include ancestry-specific and conditional analyses, as well as gene-based and scan tests.

Supported by: AG058589, AG052409, AG054076, NO1-HC-25195, HHSN268201500001I and 75N92019D00031, R01HL131136, P30 AG066546, and K99AG066849.

Background: General cognitive function is an index of performance in multiple cognitive domains. GWAS of general cognitive function, conducted mostly in European ancestry (EA) populations, have identified hundreds of genetic variants. However, investigations of rare (MAF<1%) variants have been limited, especially in non-EA populations.

Methods: We performed whole-genome sequence (WGS) association analysis for general cognitive function in N=12,615 EA, N=4,419 African ancestry (AA), and N=1,140 Hispanic/Latino ethnicity (HIS) participants from the NHLBI TOPMed Program, after exclusion for dementia, clinical stroke, or age <45 years. Linear mixed effects models were used to test for single variant association (MAC>=10), and SKAT-O was used for rare variant aggregation tests (MAF<1%, >5 alternate alleles). Using SKAT-O, we tested coding variants within genes (high-confidence LoF, predicted deleterious missense and protein-altering) and both coding and non-coding variants within genes and their regulatory elements (enhancers and promoters). Associations were evaluated within and across ancestry/ethnicity with adjustment for age, sex, study, and genetic principal components, before and after adjustment for educational attainment.

Results: Single-variant association analysis identified three genome-wide significant loci for general cognitive function: intronic in AMPH (rs17500486, MAF=3.2%, P=3.4x10-8) in EA, proximal to FTLP5 and ATCAY (rs570203641, MAF=0.1%, P=2.7x10-8) in EA, and upstream of RN7SL300P (rs113037723, MAF=8.0%, P=8.8x10-9) in AA. After adjustment for educational attainment, an additional two loci were identified: intronic in SYNPO2 (rs146805942, MAF=2.1%, P=3.4x10-8) in AA and intronic in BRINP3 (rs72729138, MAF=6.6%, P=3.0x10-8) in HIS. Rare variant aggregation tests identified two suggestive associations using coding variants (SH3GL3 in EA (P=3.2x10-6) and P2RY11 in AA (P=9.0x10-6)) and one association using coding and noncoding variants (AC113410.3 in EA, P=9.06x10-6). Of the nine genes identified, all except RN7SL300P and AC113410.3 had previous evidence of association with neuronal development or activity, neurological/cognitive disorders, or cognitive function. No variants or genes were associated with general cognitive function at the genome-wide threshold in the pooled analysis of all ancestry/ethnic groups.

Conclusions: In this WGS association analysis, we identified several new variants associated with general cognitive function, many of which lie in genes that play a role in brain function. Identified loci tended to be specific to ancestry/ethnicity.