Project Summary
Over the last decade, scientists have accelerated their efforts to understand Alzheimer’s
disease (AD). This has led to unprecedented knowledge of the genetic and biological
bases of AD risk, and vast stores of valuable data for further mining. Understanding the
genetic and biological risk states for AD is, in itself, extraordinarily valuable for guiding
mechanistic studies, developing better diagnostics, and formulating therapeutics. But an
understanding of risk states also has the benefit of allowing research on resilience to
AD. Research on the genetic and biological bases of resilience necessarily lags behind
the discovery of risk factors. Now, as the risk architecture of AD is coming into view, it is
feasible to study resilience to AD in individuals who are cognitively normal despite being
at elevated risk for the disease. The approach we have devised for identifying resilience
factors is straightforward yet, to our knowledge, unprecedented. We identify unaffected
individuals at the highest levels of multivariate risk, match them to affected individuals at
equivalent levels of risk, and contrast these two subgroups to find residual variation
associated with the absence of disease. In this project, we will capitalize on the wealth of
existing high-throughput AD risk-factor results and data, and our involvement in many of
the world’s largest AD consortia, to efficiently map resilience to AD at three levels
(genetics, transcriptomics, and neuroimaging), and to integrate across these levels. In
Aim 1, we will identify genetic variation associated with resilience to AD in the presence
of elevated genetic risk conferred by APOE e4 alleles, an elevated AD polygenic risk
score, or an elevated AD polygenic hazard score. In Aim 2, we will mega-analyze all
available transcriptomic data from studies of postmortem hippocampal tissue and of
peripheral blood in AD to identify transcriptomic risk scores and machine-learning
algorithms that maximally distinguish AD from cognitively normal control subjects, and
scores and algorithms that then identify residual transcriptomic variation that offsets the
transcriptomic risk in resilient controls. In Aim 3, we will identify an MRI-based structural
brain signature that is associated with resilience to AD in the presence of an AD-
associated cortical risk signature. Lastly, in our exploratory Aim 4, we will integrate
genetic, transcriptomic, brain structural, and clinical data to identify biological
relationships across Aims, and novel phenotypes of resilience. Collectively, these Aims
will identify multivariate, genetic, transcriptomic, and brain-structural profiles of resilience
to AD, as well as molecular, neurobiological, and clinical phenotypes stemming from AD-
resilience genotypes.