DENVER – More than a decade since its inception, the Alzheimer's Disease Sequencing Project (ADSP) continues to forge ahead, producing more datasets while emphasizing sample diversity, data harmonization, and treatment discovery, according to updates presented by project researchers here at the American Society of Human Genetics annual meeting on Thursday.
The international collaborative plans to release whole-genome sequencing data for roughly 62,000 participant samples later this year, representing diverse patient backgrounds collected from more than 50 study cohorts spanning the world.
With existing data for more than 36,000 individuals, the upcoming data release puts ADSP on track to achieve its near-term goal to study and analyze more than 100,000 participants with Alzheimer's disease and related dementias (ADRD).
Commenced by the National Institute on Aging in 2012, ADSP was established to help understand the molecular underpinning of ADRD and accelerate the development of treatments for the diseases. As the project has progressed, ADSP launched an initiative in 2021 to expand its sample set to represent more diverse patient populations for ADRD, hoping to better understand the etiology of ADRD among patients of different racial and ethnic backgrounds.
Margaret Pericak-Vance, director of the John P. Hussman Institute for Human Genomics at the University of Miami and a collaborator for the ADSP project, said that historically, Alzheimer's disease studies, including the early phase of ADSP, were predominantly skewed for samples from patients of European descendants, but that is changing in terms of the study participant makeup.
She noted that the third data release for ADSP, which was in 2021, consisted of samples from 3,100 patients of African descent, 22 Asian participants, 3,153 Latino and Hispanic participants, and 10,715 non-Hispanic White participants. But sample diversity has since improved, and a subsequent data release in fall 2022 comprised data for 5,295 patients of African descent, 2,704 Asian patients, 11,071 Latino and Hispanic patients, and 16,616 non-Hispanic White participants.
And the trend is continuing, Pericak-Vance noted, for upcoming release number five. That dataset will include 10,457 patients of African descent, 5,422 Asian patients, and 19,320 Latino and Hispanic patients, as well as 31,663 non-Hispanic White participants.
As part of its effort to foster data exchange and sample diversity, ADSP has also teamed up with other ADRD research projects across the globe, Pericak-Vance said. These include the Gwangju Alzheimer's & Related Dementias (GARD) study in South Korea, the Aspirin in Reducing Events in the Elderly (ASPREE) trial cohort in Australia, the Longitudinal Aging Study in India-Diagnostic Assessment of Dementia (LASI-DA), and the ADRD Prevalence in American Samoa project.
In addition to creating a more inclusive database, the project also established a data harmonization consortium to synergize the vast endophenotype data associated with the sequenced participants to help achieve more streamlined and efficient data analysis across the board.
In the last decade, there has been a "huge explosion" of endophenotype data for Alzheimer's disease, and how to harmonize them among different cohorts presents a challenge, said Logan Dumitrescu, a researcher from Vanderbilt University Medical Center who is involved in the phenotype harmonization consortium.
Taking cognition, an important phenotype for ADRD, as an example: There are "big differences" in how it is measured among studies, ranging for a six-item screener to multiple-hour cognitive batteries, Dumitrescu pointed out. "Imagine you have 10 studies, and all 10 studies measure memory differently," Dumitrescu said. "So, we really needed a strategy to pull this data across studies not only for analysis but also for replication, as well."
However, data harmonization can be a complicated process, Dumitrescu noted, and it often leverages advanced statistical approaches. One strategy, for instance, is to identity anchor items — metrics that are more or less equivalent across tests — to help normalize the rest of the data among different cohorts, Dumitrescu said.
In the long run, Dumitrescu said the goal for the consortium is to work with other ADSP workgroups to streamline the endophenotype data for the ADRD samples and generate datasets that can be perpetually curated and shared through the NIAGADS Data Sharing Service, a national data repository developed to facilitate data dissemination for ADSP and other NIA-funded Alzheimer’s Disease and Related Dementias genomic studies.
Along the way, the consortium also hopes to educate the research community on available harmonized resources and best practices, Dumitrescu noted.
Beyond data generation and harmonization, ADSP collaborators have also started to leverage the project data to inform ADRD treatment discovery and development. Alexandra Munch, a researcher from Icahn School of Medicine at Mount Sinai and a ADSP collaborator, presented efforts on this front from her lab.
By integrating human genetics with cell-type-specific genomic data, Munch and her colleagues are working to identity causal variants, genes, cells, and pathways implicated in ADRD. These potential molecular targets for disease treatment are then validated in vitro using engineered human stem cells as well as in vivo using engineered mice and xenotransplantation mouse models, Munch said.
So far, her team has identified genetically driven targets, such as mutations in the MS4A4A and MS4A6A genes and are continuing to study and validate these targets for potential therapy development.
"The goal of ADSP is to use the lens of genetics to better inform drug discovery," Munch said. "All of this data should allow us to gain a better mechanistic understanding of AD pathogenesis and ultimately design disease-modifying therapeutics."