DOI: 10.1002/alz.078351 ISSN: 1552-5260

Association of low‐frequency and rare variants with cognition in genetic frontotemporal dementia: Results from GENFI

Saira S. Mirza, Andrew D. Patterson, Carmela Tartaglia, Sara Berman Mitchell, Sandra E. Black, Morris Freedman, David F. Tang‐Wai, Ekaterina Rogaeva, David M Cash, Martina Bocchetta, John C. van Swieten, Robert Laforce, Fabrizio Tagliavini, Barbara Borroni, Daniela Galimberti, James B. Rowe, Caroline Graff, Elizabeth Finger, Sandro Sorbi, Alexandre de Mendonça, Christopher R. Butler, Alexander Gerhard, Raquel Sanchez‐Valle, Fermin Moreno, Matthis Synofzik, Rik Vandenberghe, Simon Ducharme, Johannes Levin, Adrian Danek, Markus Otto, Isabel Santana, Jonathan D. Rohrer, Mario Masellis
  • Psychiatry and Mental health
  • Cellular and Molecular Neuroscience
  • Geriatrics and Gerontology
  • Neurology (clinical)
  • Developmental Neuroscience
  • Health Policy
  • Epidemiology



Frontotemporal dementia (FTD) is a neurodegenerative disorder characterized by heterogeneous clinical, pathological, and genetic features. Mutations in three genes account for the majority of autosomal dominant FTD: GRN, MAPT, and C9orf72. We tested whether gene‐based aggregate burden of genome‐wide low frequency variants contribute to variation in executive function, memory, and language performance in the GENetic Frontotemporal dementia Initiative (GENFI), after controlling for effects of causative mutations.


GENFI recruits symptomatic and presymptomatic participants from families segregating genetic FTD. We included 565 participants with genotype (Neurochip; imputed against TOPMed), and neuropsychological data. Gene‐based burden tests that aggregate the number of rare alleles by gene, were used to examine the association of low‐frequency variants (minor allele frequency: 0.000001 to <0.05) with factor scores of executive function, memory, and language, controlling for age, sex, education, mutation status, population stratification, and family membership (kinship matrix) in multiple linear regression models. Cognitive factor scores were derived from Confirmatory Factor Analyses. In two separate analyses, we used the following set of annotations to account for (i) loss of function mutations (LOF): start gain, stop loss, start loss, essential splice site, stop gain, normal splice site, and non‐synonymous, (ii) insertions, deletions, and frameshift mutations (indel‐fs).


Of the 565 participants (316 women), 307 were mutation carriers (symptomatic = 108). For LOF mutations, aggregate burden of variants in FAM183A reached statistical significance (741 genes tested; significance threshold: p = 0.05/731 = 6.7×10−5) for executive function (standardized β:4.2; p = 6.0×10−5), whereas IPO13 (standardized β: ‐1.7; p = 3.1×10−5) and FAAHP1 (standardized β: ‐1.7; p = 1.8×10−5) were significant for memory. For the indel‐fs mutations, HMGB4 was associated with worse memory function (standardized β: ‐0.95; p = 5.9×10−6). No genes were significantly associated with language. [Gene info: (i) FAM183A‐codes a ciliary‐base protein; location1p34.2; associated with autosomal recessive intellectual disability. (ii) IPO13‐a nuclear transporter gene; location 1p34.1; associated with Partington syndrome and Agenesis corpus callosum with abnormal genitalia. (iii) HMGB4‐a transcription regulator gene; location 1p35.1].


Identification of low frequency variants contributing to disease phenotypes may help identify genetic modifiers of familial FTD. Replication of results followed by functional studies are required.

More from our Archive