Leveraging a Novel Phenotype to Identify Complex Traits and Biological Pathways Associated with Resilience to Alzheimer’s Disease

Jared Phillips, Logan Dumitrescu, Derek B. Archer, Alexandra N. Regelson, Shubhabrata Mukherjee, Michael L. Lee, Seo‐Eun Choi, Phoebe Scollard, Emily H. Trittschuh, Walter A. Kukull, Sarah Biber, Jesse Mez, Emily R. Mahoney, Michelle Clifton, Julia B. Libby, Skylar Walters, William S. Bush, Corinne D. Engelman, Qiongshi Lu, David W. Fardo, Keith F. Widaman, Rachel F. Buckley, Elizabeth C. Mormino, Elizabeth A. Sanders, Lindsay R. Clark, Katherine A. Gifford, Badri N. Vardarajan, Michael L. Cuccaro, Margaret A Pericak‐Vance, Lindsay A. Farrer, Li‐San Wang, Gerard D. Schellenberg, Jonathan L. Haines, Angela L. Jefferson, Sterling C. Johnson, Marilyn S. Albert, C. Dirk Keene, Andrew J. Saykin, Shannon L. Risacher, Eric B. Larson, Reisa Sperling, Richard Mayeux, Alison Goate, Alan E. Renton, Edoardo Marcora, Brian Fulton‐Howard, Tulsi Patel, David A. Bennett, Julie A. Schneider, Lisa L. Barnes, Carlos Cruchaga, Jason J. Hassenstab, Michaël E. Belloy, Shea J. Andrews, Susan M. Resnick, Murat Bilgel, Yang An, Lori L. Beason‐Held, Keenan A. Walker, Michael R. Duggan, Brandon S. Klinedinst, Paul K. Crane, Timothy J. Hohman, The Alzheimer’s Disease Sequencing Project (adsp)Alzheimer 2024 · DOI

TL;DR

This paper introduces a novel longitudinal modeling approach to study resilience to Alzheimer's Disease (AD), using a larger and more diverse dataset than previous studies. By incorporating APOE interactions and a "silver" model that doesn't require autopsy data, they identified new biological pathways and genetic associations linked to AD resilience, potentially accelerating genetic discovery.

What Problem Does This Paper Address?

Understanding resilience to Alzheimer's Disease (AD) is crucial for developing effective prevention and treatment strategies. Previous research on AD resilience has often relied on cross-sectional data and limited measures of neuropathology, which can restrict the sample size and statistical power. This approach also may not fully capture the dynamic nature of resilience over time. The authors aim to address this by developing a more robust and inclusive method to identify the genetic underpinnings of AD resilience, thereby increasing participant diversity and statistical power.

Methods

The researchers employed a novel longitudinal modeling approach to analyze AD resilience. They utilized mixed-effects regressions across diverse ancestries to explore the genetics of resilience. A key innovation is the development of two "models" for resilience:

  • "Silver" model: This model leverages longitudinal data and includes APOE and higher-order interaction terms to approximate neuropathological resilience without requiring autopsy data. This allows for a significantly larger sample size.
  • "Gold" model: This model incorporates neuropathology at autopsy, representing a more direct measure of resilience.

Meta-analyses were conducted to maximize statistical power. The models were adjusted for age, sex, APOE ε4 allele count, APOE ε2 allele presence, and all covariate interactions with time. Residual cognitive resilience was quantified from three cognitive domains: memory, executive function, and language. Post-GWAS analyses included gene and pathway tests using MAGMA and genetic correlation with 65 traits using GNOVA.

Datasets

The study leveraged data from 20,513 participants. While specific names of the contributing cohorts are not explicitly listed as distinct datasets in the provided text, the affiliations of the authors indicate participation from various institutions involved in AD research, including the Vanderbilt Memory & Alzheimer's Center, the Alzheimer's Disease Sequencing Project (ADSP), and the National Alzheimer's Coordinating Center. The paper also refers to "silver" and "gold" models, which represent different ways of defining and measuring resilience based on the available data (autopsy vs. no autopsy).

Key Insights and Ideas

  • Longitudinal data is key: The study highlights the power of longitudinal data in capturing the dynamic nature of resilience, moving beyond static cross-sectional snapshots.
  • "Silver" model expands possibilities: The development of a "silver" model that bypasses the need for autopsy data significantly broadens the scope of genetic studies on resilience, allowing for much larger sample sizes and increased statistical power.
  • Resilience is complex and multi-faceted: The identification of multiple biological pathways and complex traits associated with resilience underscores that it's not a single phenomenon but a complex interplay of biological processes.
  • Genetic correlations reveal unexpected links: The finding of negative correlations between resilience and certain neuropsychiatric traits (like Tourette syndrome and schizophrenia) suggests potential shared genetic underpinnings or complex regulatory mechanisms.

Key Results

  • High correlation between models: The "silver" and "gold" resilience metrics were highly correlated (0.77-0.88), validating the "silver" model's ability to approximate neuropathological resilience.
  • Genetic covariance: Positive genetic covariance (0.01-0.13) was observed between the resilience metrics.
  • Pathway associations: Three significant pathway associations with resilience were identified:
  • Metabolism of amino acids and derivatives (silver all participants meta-analysis).
  • Negative regulation of transforming growth factor beta production (silver CU at baseline meta-analysis).
  • The SARS pathway (gold CU at baseline meta-analysis).
  • Genetic locus: A locus on chromosome 17 approached genome-wide significance for resilience, with the top SNP being rs757022 (MAF = 0.18, effect size b = 0.08, P = 1.1×10⁻⁷).
  • Negative correlations: Resilience showed negative correlations with multiple sclerosis and neuropsychiatric traits, including Tourette syndrome and schizophrenia (all P<0.05 FDR corrected).

Limitations and Future Work

  • Approximation of neuropathology: While the "silver" model is a powerful tool, it is an approximation of neuropathological resilience. Future work could aim to further refine these models or integrate them with emerging biomarkers.
  • Genome-wide significance: The identified locus on chromosome 17 approached, but did not reach, genome-wide significance. Further studies with larger sample sizes or different cohorts may be needed to confirm this finding.
  • Accelerating discovery: The authors suggest that their "silver" standard modeling approaches can complement current "gold" standard methods to accelerate genetic discovery for AD resilience. This implies future work will focus on applying and refining these methods to uncover more genetic associations and biological insights.

Summary updated

Abstract

Abstract Background Previous models of resilience to Alzheimer’s Disease (AD) have relied on cross‐sectional designs and inclusion of measures of neuropathology. Here, we present a novel modeling approach incorporating longitudinal data and the use of APOE and higher order interaction terms to approximate neuropathological resilience, vastly increasing participant diversity and statistical power. We validate this approach and report novel genetic associations with neuropathological resilience. Method Leveraging 20,513 participants (Table 1), we explored the genetics of resilience metrics using mixed‐effects regressions across ancestries. We conducted GWAS on four resilience metrics: two in a “silver” model, which benefit from a larger sample size and did not include neuropathology at autopsy, and two in a “gold” model, which incorporate neuropathology at autopsy. These models were built in all diagnoses and separately in cognitively unimpaired (CU) at baseline participants. Silver‐gold meta‐analyses were conducted to maximize statistical power. Models adjusted for age, sex, APOE ε4 allele count, presence of the APOE ε2 allele and all covariate interactions with interval (years from baseline). Residual cognitive resilience was quantified from residuals in three cognitive domains (memory, executive function, and language). Post‐GWAS analyses included gene and pathway tests using MAGMA and genetic correlation with 65 traits using GNOVA. Result Silver and gold resilience metrics were highly correlated (0.77‐0.88) and showed positive genetic covariance (0.01‐0.13). We identified negative correlations of resilience with multiple sclerosis and neuropsychiatric traits, including Tourette syndrome and schizophrenia (all P FDR &lt;2.6×10 −2 ; Figure 1). Furthermore, we observed three significant pathway associations with resilience: metabolism of amino acids and derivatives in the silver all participants meta‐analysis, negative regulation of transforming growth factor beta production in the silver CU at baseline meta‐analysis, and the SARS pathway in the gold CU at baseline meta‐analysis (all P FDR &lt;5.0×10 −2 ). Finally, we identified a locus on chromosome 17 associated with resilience that approached genome‐wide significance (top SNP: rs757022, MAF = 0.18, b = 0.08, P = 1.1×10 −7 ; Figure 2). Conclusion Using novel, longitudinal modeling approaches, we identified multiple biological pathways and complex traits associated with resilience to AD as well as genetic loci approaching genome‐wide significance. Our findings indicate that silver standard modeling approaches can complement current gold standard methods to accelerate genetic discovery.
BibTeX
@inproceedings{phillips2024leveraging,
  author = {Jared Phillips and Logan Dumitrescu and Derek B. Archer and Alexandra N. Regelson and Shubhabrata Mukherjee and Michael L. Lee and Seo‐Eun Choi and Phoebe Scollard and Emily H. Trittschuh and Walter A. Kukull and Sarah Biber and Jesse Mez and Emily R. Mahoney and Michelle Clifton and Julia B. Libby and Skylar Walters and William S. Bush and Corinne D. Engelman and Qiongshi Lu and David W. Fardo and Keith F. Widaman and Rachel F. Buckley and Elizabeth C. Mormino and Elizabeth A. Sanders and Lindsay R. Clark and Katherine A. Gifford and Badri N. Vardarajan and Michael L. Cuccaro and Margaret A Pericak‐Vance and Lindsay A. Farrer and Li‐San Wang and Gerard D. Schellenberg and Jonathan L. Haines and Angela L. Jefferson and Sterling C. Johnson and Marilyn S. Albert and C. Dirk Keene and Andrew J. Saykin and Shannon L. Risacher and Eric B. Larson and Reisa Sperling and Richard Mayeux and Alison Goate and Alan E. Renton and Edoardo Marcora and Brian Fulton‐Howard and Tulsi Patel and David A. Bennett and Julie A. Schneider and Lisa L. Barnes and Carlos Cruchaga and Jason J. Hassenstab and Michaël E. Belloy and Shea J. Andrews and Susan M. Resnick and Murat Bilgel and Yang An and Lori L. Beason‐Held and Keenan A. Walker and Michael R. Duggan and Brandon S. Klinedinst and Paul K. Crane and Timothy J. Hohman and The Alzheimer’s Disease Sequencing Project (ADSP)},
  title = {{Leveraging a Novel Phenotype to Identify Complex Traits and Biological Pathways Associated with Resilience to Alzheimer’s Disease}},
  booktitle = {Alzheimer},
  year = {2024},
  doi = {10.1002/alz.095693},
}
Figure