📅 7 July 2026 🏷️ Alzheimer's Genetics ⏱️ 7 min read 👩‍🔬 Linda Osaghale

Why Your Gut Feeling About Alzheimer's Genetics Might Be Wrong

Have you ever read one study and thought, "This is it! The answer!" And then you read another study that contradicts it?

That's the problem with single studies. They can be wrong. They can be biased. They can be just plain lucky.

But when you combine multiple studies—well, that's when you start to see the truth.

Meta-Analysis: Combining evidence from multiple studies for stronger conclusions in Alzheimer's genetics research
Figure 1: Meta-analysis combines results from multiple studies to produce a summary estimate. Based on Osaghale et al. (2026).

What Is Meta-Analysis?

Meta-analysis is a statistical method for combining results from multiple studies:

StepWhat It Does
1Collects results from multiple studies
2Weighs each study by its precision
3Combines the results
4Produces a summary estimate

Meta-analysis gives you the big picture.

Why We Used Meta-Analysis

The Problem with Single Studies

IssueWhy It Matters
Limited sample sizeSingle studies may be underpowered
Population differencesWhat's true in one population may not be true in another
Random variationSome findings are due to chance
Publication biasPositive results are more likely to be published

The Solution: Meta-Analysis

SolutionHow It Works
Increased powerCombining studies increases sample size
GeneralizabilityMultiple populations increase generalizability
Reduced random variationCombining reduces chance findings
Publication bias assessmentCan detect publication bias

How We Performed Meta-Analysis

Step 1: Collect Results

We extracted results from three cohorts:

CohortNDescription
FinnGen211,678Discovery cohort
IEU488,285Replication cohort 1
EBI85,934Replication cohort 2

Step 2: Harmonize Effect Sizes

We used:

Step 3: Choose the Model

We used:

Step 4: Assess Heterogeneity

We used:

Our Meta-Analysis Results

Variant 1: rs429358

CohortORP-value
FinnGen4.585.20 × 10⁻¹⁹⁷
IEU1.0042.80 × 10⁻¹⁷⁵
Combined1.004< 0.001
HeterogeneityValue
99.9%
Q893.84
P-het< 0.001

Finding: Significant combined effect. Extreme heterogeneity.

Variant 2: rs3178166

CohortORP-value
FinnGen0.801.53 × 10⁻¹¹
IEU0.99973.90 × 10⁻³
EBI0.9671.47 × 10⁻⁴
Combined1.0000.0007
HeterogeneityValue
96.7%
Q59.77
P-het< 0.001

Finding: Significant combined effect. Substantial heterogeneity.

Variant 3: rs111371860

Issue: Opposite effect direction in FinnGen vs. replication cohorts.

Solution: Meta-analysis restricted to replication cohorts.

CohortBetaP-value
FinnGen+0.36603.91 × 10⁻¹⁰
IEU-0.001134.30 × 10⁻⁹
EBI-0.27511.37 × 10⁻⁴⁶
CohortORP-value
IEU0.9994.30 × 10⁻⁹
EBI0.7591.37 × 10⁻⁴⁶
Combined0.9992.95 × 10⁻⁹
HeterogeneityValue
99.5%
Q203.59
P-het< 0.001

Finding: Significant combined effect. Extreme heterogeneity.

What the Heterogeneity Means

Why I² > 95%?

ReasonExplanation
Different phenotype definitionsClinical vs. algorithm-defined
Different populationsFinnish vs. UK vs. European
Different covariate adjustmentsDifferent variables adjusted for
Different quality controlDifferent QC procedures

What This Tells Us

ImplicationDetail
Replication is essentialMeta-analysis alone isn't enough
Population differences existEffect sizes vary across populations
Phenotype mattersHow you define Alzheimer's affects results
Standardization is neededConsistent methods would reduce heterogeneity

The Bottom Line

Meta-analysis is powerful, but it's not a substitute for replication.

Key Takeaways

FindingImplication
Combined effects significantStrong evidence overall
Heterogeneity is highPopulation differences exist
Replication is essentialDon't rely on meta-analysis alone
Standardization neededConsistent methods would help

What do you think?

👍 😂 ❤️ 😮 😡 😢

0 Responses

Osaghale L, Beshiru A, Subhan U. (2026). Replication-guided functional genomic prioritization of regulatory risk variants in Alzheimer's disease. Gene Reports. 44: 102551.

DOI: https://doi.org/10.1016/j.genrep.2026.102551


Next post: "The Hidden Networks: How Alzheimer's Genes Connect to Biological Pathways" — Coming soon!

← Back to Blog Home