Neuropathology-based GWAS for Alzheimer’s disease reveals novel susceptibility loci and highlights sex-specific pathways

Yin Jin, Apostolia Topaloudi, Sudhanshu Shekhar, Alicia Nicole Scott, Bryce David Colon, Petros Drineas, Chris Rochet, Peristera PaschoumedRxiv 2023 · DOI

TL;DR

By shifting the focus from clinical symptoms to "gold standard" neuropathological data, researchers identified novel genetic drivers of Alzheimer’s disease (AD) that traditional studies missed. The study highlights significant sex-specific differences, including a link between AD pathology in women and genes expressed in the ovary, while identifying potential blood-based biomarkers like ACSBG1 and TAC3 that may causally influence disease progression.

What Problem Does This Paper Address?

Alzheimer’s disease is notoriously heterogeneous. Clinical diagnosis—relying on cognitive tests and behavioral observations—carries a misdiagnosis rate of approximately 24%. Many patients diagnosed with AD actually have mixed dementias, while others in "control" groups may have pre-clinical pathology but no symptoms yet. This "noise" in the data makes it difficult for standard Genome-Wide Association Studies (GWAS) to find the genetic roots of the disease.

Furthermore, women are significantly more likely to develop AD and experience faster cognitive decline, yet most genetic studies treat sex as a mere covariate rather than a primary variable. This paper aims to cut through the clinical noise by using post-mortem neuropathological measurements (like tau tangles and amyloid plaques) as the primary phenotypes and conducting sex-stratified analyses to uncover the unique biological pathways driving the disease in men versus women.

Methods

The researchers employed a multi-omic integrative approach:

  • Neuropathology-based GWAS: Instead of "AD vs. Control," they used three specific phenotypes: neuropathologically-confirmed AD (ncAD), Braak stage (measuring tau tangles), and Neuritic Plaque (NP) scores.
  • Transcriptome-Wide Association Study (TWAS): They integrated GWAS results with gene expression data from 14 brain tissues to identify genes whose expression (not just their DNA sequence) is linked to AD.
  • Mendelian Randomization (MR): To move beyond correlation, they used MR to investigate whether circulating blood proteins have a causal relationship with AD pathology.
  • Multi-omics Integration: They cross-referenced their findings with proteomic and transcriptomic data from the ROSMAP and Mayo Clinic studies to identify robust biomarkers.

Datasets

The study integrated 13 different genomic and neuropathology datasets, totaling 6,960 individuals. Key datasets included:

  • ROSMAP (Religious Orders Study and Memory and Aging Project): Provided individual-level genetic, transcriptomic, and proteomic data.
  • Mayo Clinic AD Research Center: Provided RNA-seq data from the temporal lobe.
  • ADC (Alzheimer’s Disease Center): Contributed a large sample of neuropathological records.
  • INTERVAL Study: Used for plasma protein GWAS summary statistics in the MR analysis.
  • Fenland Study: Used as an independent cohort for replication of biomarker findings.

Key Insights and Ideas

  • Pathology Reveals What Symptoms Hide: The study confirmed the well-known APOE and BIN1 loci but also uncovered genes like OPCML and CDH4 that are specifically associated with the physical buildup of plaques and tangles, rather than just clinical symptoms.
  • The Sex-Specific "Aha" Moment: The most striking finding was in females. A tissue enrichment analysis linked NP scores in women to genes expressed in the ovary. This provides strong genetic evidence for the long-suspected connection between sex hormones (like estrogen) and AD risk.
  • BIN1 is a Female-Dominant Risk Factor: While BIN1 is a known AD risk gene, this study showed its effect is significantly more pronounced in women, specifically relating to the progression of tau pathology.
  • Causal Biomarkers: The researchers identified ACSBG1 as having a positive causal relationship with Braak stage (tau progression) and TAC3 as having a negative association with plaque scores in females, suggesting these proteins could be targets for future "in-vivo" (live patient) diagnostic tools.

Key Results

  • Novel Loci: Identified OPCML (involved in synaptogenesis) and CDH4 (involved in cell adhesion and neuron development) as significant hits for neuritic plaque scores.
  • TWAS Success: Identified 8 novel genes (including ST8SIA1, EIF2B2, and NEUROD2) whose expression levels are significantly associated with AD pathology but had never been flagged in traditional clinical GWAS.
  • Sex-Specific Loci: In women, the researchers found significant associations at 2q14 (BIN1), 4q27 (QRFPR), and the SGCZ gene. No additional significant loci beyond the APOE region were found in the male-only analysis.
  • Biomarker Causal Links: Mendelian Randomization showed that higher blood levels of ACSBG1 are causally linked to higher Braak stages (Beta: 0.065, P = 0.024).

Limitations and Future Work

The authors acknowledge several hurdles:

  • Sample Size: While 6,960 is large for a pathology study, it is small compared to clinical GWAS (which often exceed 100,000). This limits the power to detect rare variants.
  • Blood vs. Brain: The protein biomarkers were measured in blood, which may not perfectly reflect the environment of the brain where AD pathology occurs.
  • Replication: Some findings, such as the TAC3 association, showed a consistent directional trend in independent cohorts but did not reach full statistical significance, requiring further validation in even larger datasets.

Future work will likely focus on exploring the "ovary-brain" axis more deeply and testing whether the identified causal proteins can be developed into reliable blood tests for early AD detection.

Summary updated

Abstract

Abstract Dementia refers to an umbrella phenotype of many different underlying pathologies with Alzheimer’s disease (AD) being the most common type. While Genome-wide Association Studies (GWAS) of clinically defined AD have identified multiple AD susceptibility variants, a significant portion of the heritability remains unexplained highlighting the phenotypic and genetic heterogeneity of the clinically defined entity. Furthermore, despite women’s increased susceptibility to dementia there is a lack of sex-specific genetic studies and understanding of sex-specific background for the disorder. Here, we aim to tackle the heterogeneity of AD by specifically concentrating on neuropathological features which are the gold standard for diagnosis and pursuing sex-specific analysis. We bring together 13 different genomic and neuropathology datasets (6,960 individuals) and we integrate our GWAS findings with transcriptomic and proteomic data (ROSMAP and Mayo studies) aiming to also identify in vivo biomarkers for AD progression. We uncover novel genetic associations to AD neuropathology, including BIN1, OPCML, and CDH4. Our sex-specific analysis points to a role for BIN1 specifically in women as well as novel AD loci including QRFPR and SGCZ. Tissue-specific associations in females, particularly in the ovary issue, suggest a connection between sex hormones and AD. Finally, we pursue the identification of in-vivo biomarkers associated with AD neuropathology, with a goal to offer insights into disease progression and potential therapeutic targets. Our findings contribute to unraveling the molecular basis of AD, emphasizing the importance of sex-specific analyses and multi-omics approaches. Further research is needed to validate and explore the clinical utility of these findings.
BibTeX
@misc{jin2023neuropathology,
  author = {Yin Jin and Apostolia Topaloudi and Sudhanshu Shekhar and Alicia Nicole Scott and Bryce David Colon and Petros Drineas and Chris Rochet and Peristera Paschou},
  title = {{Neuropathology-based GWAS for Alzheimer’s disease reveals novel susceptibility loci and highlights sex-specific pathways}},
  year = {2023},
  doi = {10.1101/2023.12.17.23300106},
  howpublished = {medRxiv},
}
Figure