Whole-genome sequence genome-wide association study in All of Us identifies a novel glaucoma risk locus in African ancestry individuals

This study leverages whole-genome sequencing data from the diverse All of Us cohort to identify a novel glaucoma risk locus near *CYP2A7* specifically in African ancestry individuals, demonstrating the critical value of genetic diversity and refined phenotypic definitions in uncovering new genetic insights for primary open-angle glaucoma.

The Gist

Imagine the human genome as a massive, ancient library containing the instruction manual for building and maintaining a human body. For decades, scientists trying to find the "glaucoma" section of this manual have mostly been reading copies written in one specific dialect: European.

While they found some useful clues in that dialect, they missed huge chunks of the story because they didn't have enough books written in other dialects, like African or Latino. Furthermore, the tools they used to read these books were like low-resolution photocopies—they could only see the most common words, missing the rare but critical typos that cause disease.

This paper is like a team of detectives finally getting access to a brand-new, high-definition library (the All of Us Research Program) that contains millions of books in many different dialects. They used a super-powerful scanner (Whole Genome Sequencing) to read every single letter, not just the common ones.

Here is what they discovered, broken down simply:

1. The "Strict" vs. "Relaxed" Detective Work

The researchers had to figure out who actually had glaucoma and who didn't.

• The Relaxed Approach: They looked at medical records and asked, "Does this person have a code in their file that says 'Glaucoma'?" It's like checking a guest list at a party.
• The Strict Approach: They went a step further. They asked, "Does this person not only have the code, but are they also taking medicine or having surgery for it?" This is like checking the guest list and seeing if the person is actually dancing on the dance floor.

The Lesson: Being stricter about who counts as a "case" helped them find the genetic clues more clearly, especially in people of African ancestry.

2. Finding a New "Smoking Gun" in the African Ancestry Group

When they scanned the European books, they found the same old clues they already knew about (like finding a familiar street sign). But when they scanned the African ancestry books using the "Strict" method, they found a brand new clue they had never seen before.

• The Clue: A tiny typo (a genetic variant) near a gene called CYP2A7.
• The Metaphor: Imagine the genome is a city. The researchers found a broken traffic light (the genetic variant) in a specific neighborhood (near the CYP2A7 gene) that seems to cause traffic jams (glaucoma) specifically in that neighborhood.
• Why it matters: This clue was hiding in plain sight in previous studies because those studies didn't have enough African participants to notice it. It's like trying to find a specific accent in a crowd of 100 people; if you only have 5 people with that accent, you might miss it. But if you have 1,000, you can hear it clearly.

3. The "Oxygen Sensor" Connection

The researchers didn't just stop at finding the clue; they asked, "What does this gene actually do?"

• They found that the nearby gene, EGLN2, acts like an oxygen sensor in the body. It helps cells know when there is too little oxygen.
• The Analogy: Think of the eye's nerve cells (retinal ganglion cells) as delicate flowers. If the "oxygen sensor" is broken, the flowers might think they are in a drought when they aren't, or vice versa, causing them to wither and die. This dying of nerve cells is exactly what happens in glaucoma.
• Interestingly, this specific genetic variation might have been helpful in the past for fighting diseases like malaria (which is common in African history), but now, in a modern context, it might accidentally increase the risk of glaucoma. It's a case of "evolutionary trade-off."

4. Why This Matters for Everyone

• Diversity is Key: If we only study one group of people, we miss the unique genetic risks that affect others. This study proves that including diverse groups isn't just about fairness; it's about finding more answers for everyone.
• Better Tools: Using the "Whole Genome Sequencing" (reading the whole book) instead of "Genotyping Arrays" (reading a summary) allowed them to see the rare typos that cause disease.
• Future Hope: By finding this new "broken traffic light," scientists can now design better tests to see who is at risk and potentially develop new drugs that fix the oxygen-sensing mechanism in the eye.

The Bottom Line

This paper is a victory lap for diversity and precision. It shows that when we look at the whole picture with high-quality tools and include people from all backgrounds, we can find the hidden keys to unlocking complex diseases like glaucoma. It's a reminder that the instruction manual for human health is written in many dialects, and we need to read them all to understand the full story.

Technical Summary

1. Problem Statement

Primary Open-Angle Glaucoma (POAG) is a complex, heritable disease that disproportionately affects individuals of African and Latino ancestry, who often present with earlier onset and more severe disease compared to those of European ancestry. However, prior genetic discovery has been limited by two major factors:

• Lack of Ancestral Diversity: Most Genome-Wide Association Studies (GWASs) have been conducted in populations of European (EUR) ancestry, leaving a significant gap in understanding the genetic architecture of POAG in African (AFR) and Latino/Admixed American (AMR) populations.
• Methodological Limitations: Previous studies largely relied on genotyping arrays, which assay only ~2% of genomic variants. The remaining variants are imputed, a process that suffers from reduced accuracy in non-European populations due to smaller reference panels and differing linkage disequilibrium (LD) patterns.
• Phenotypic Heterogeneity: Many large-scale biobank studies rely solely on International Classification of Diseases (ICD) codes for case definition, which can lack specificity and reduce statistical power.

2. Methodology

The study leveraged the National Institutes of Health (NIH) All of Us (AoU) Research Program v8 data release, which includes Whole Genome Sequencing (WGS) data for over 414,000 participants, nearly half of whom are from historically underrepresented groups.

• Study Design: Ancestry-stratified GWASs were performed for EUR, AFR, and AMR groups, followed by cross-ancestry meta-analyses.
• Phenotype Definitions: Two distinct case/control definitions were applied to test the impact of phenotypic stringency:
  1. Relaxed Phenotype: Cases defined by ICD-9/10 codes for POAG (excluding secondary/angle-closure glaucoma). Controls were age >65 with no ICD codes for glaucoma or ocular hypertension.
  2. Stringent Phenotype: Cases met the relaxed criteria plus evidence of glaucoma treatment (topical drops, laser, or surgery) in the Electronic Health Record (EHR). Controls had no evidence of treatment.
• Data Processing:
  • QC: Standard WGS quality control was applied (coverage ≥30×, contamination ≤1%, etc.). Variants were filtered for call rate, Hardy-Weinberg equilibrium, and minor allele frequency.
  • Analysis: Firth logistic regression was used within ancestry groups (covariates: age, sex, top 10 principal components).
  • Meta-analysis: Fixed-effects, inverse variance-weighted meta-analyses were conducted across ancestries using METAL. Heterogeneity was assessed via $I^2$.
• Validation & Annotation:
  • Replication: A novel locus was tested in an independent dataset (GGLAD-Duke) of 701 AFR cases and 606 controls.
  • Functional Annotation: Variants were queried in GTEx and FIVEx databases for expression Quantitative Trait Loci (eQTL) effects, using neuronal tissues as surrogates for ocular tissue. Single-cell data from the Broad Institute was used to verify gene expression in ocular cell types.
  • Correlation: Effect sizes from AoU were compared against prior large-scale meta-analyses (IGGC and MTAG) to validate data quality.

3. Key Contributions

• First WGS-based Cross-Ancestry POAG GWAS: This is the first study to utilize WGS data (rather than imputed array data) for a cross-ancestry meta-analysis of POAG, providing a more comprehensive view of common and rare variants.
• Phenotypic Refinement: The study systematically demonstrates how varying phenotype stringency (ICD codes vs. ICD + treatment evidence) influences genetic discovery power.
• Discovery of a Novel Locus: Identification of a genome-wide significant risk locus specifically in African ancestry individuals, a population historically underrepresented in glaucoma genetics.

4. Key Results

• Sample Sizes:
  • Relaxed Phenotype: 3,193 cases / 110,787 controls.
  • Stringent Phenotype: 2,640 cases / 103,020 controls.
• Replication of Known Loci:
  • Known risk loci (TMCO1, CDKN2B-AS1, GMDS) reached genome-wide significance ($P < 5 \times 10^{-8}$) in EUR and cross-ancestry meta-analyses.
  • Effect sizes in AoU strongly correlated with prior IGGC and MTAG meta-analyses (Pearson $r = 0.75–0.84$, $P < 10^{-5}$), validating the dataset's utility.
• Novel Discovery (AFR Specific):
  • A novel risk locus near CYP2A7 (variant rs76935404[T]) was identified in the AFR GWAS using the stringent phenotype ($OR = 1.35$, $P = 1.2 \times 10^{-8}$).
  • Replication: The variant replicated in the independent GGLAD-Duke cohort ($OR = 1.34$, $P = 5.8 \times 10^{-3}$). Meta-analysis of AoU and GGLAD-Duke confirmed genome-wide significance ($P = 2.3 \times 10^{-10}$).
  • Ancestry Specificity: While the variant is polymorphic across all ancestries, the effect size was significantly stronger in AFR individuals ($OR \approx 1.35$) compared to EUR and AMR, where it did not reach significance in the cross-ancestry meta-analysis.
• Functional Insights:
  • eQTL Analysis: rs76935404 is a significant eQTL for CYP2A7 (in hypothalamus, basal ganglia, hippocampus) and EGLN2 (in sensory neuronal tissue).
  • Biological Plausibility:
    • CYP2A7: Involved in drug metabolism and steroid/lipid biosynthesis; expressed in optic nerve astrocytes and retinal ganglion cells.
    • EGLN2: A key enzyme in oxygen sensing (regulates HIF-1$\alpha$ degradation). Hypoxia-induced retinal ganglion cell death is a known mechanism in glaucoma. EGLN2 is also a target for HIF prolyl hydroxylase inhibitors.
  • Evolutionary Context: The authors hypothesize that the strong association in AFR populations may reflect evolutionary selection pressures (e.g., malaria resistance) that favored variants enhancing detoxification or immune response, which may have pleiotropic effects on glaucoma risk.

5. Significance

• Addressing Health Disparities: The study highlights the critical need for diverse genetic datasets. A locus significant in African ancestry populations was missed in European-only analyses, demonstrating that excluding diverse populations leads to incomplete genetic maps of disease.
• Methodological Validation: The strong correlation of effect sizes with prior studies confirms that WGS data from the All of Us program is robust for complex trait discovery, even with smaller sample sizes compared to array-based meta-analyses, provided phenotyping is rigorous.
• Therapeutic Implications: The identification of EGLN2 and the hypoxia-sensing pathway as a potential driver of POAG in specific ancestries opens new avenues for targeted therapies (e.g., repurposing HIF prolyl hydroxylase inhibitors).
• Future Directions: The findings underscore that "one-size-fits-all" genetic models are insufficient. Future research must prioritize diverse WGS cohorts and refined phenotyping (incorporating treatment data) to uncover ancestry-specific risk factors and causal mechanisms.