Massive Proteogenomic Study Maps How 3,000 Proteins Drive Disease Risk Across the Human Body
A 54-cohort analysis links thousands of genetic variants to circulating proteins, creating a roadmap for disease mechanisms and drug targets.
Summary
Researchers analyzed genetic data alongside blood protein measurements from over 100,000 people across 54 study cohorts to identify how specific DNA variants influence protein levels — and how those proteins connect to disease. Using a technique called Mendelian randomization, they identified causal relationships between proteins and hundreds of conditions, from heart disease and diabetes to neurodegeneration and cancer. The study catalogued over 600,000 protein-gene associations, validated findings across independent datasets, and highlighted proteins that may serve as drug targets. This is one of the most comprehensive maps of how genetics shapes the blood proteome and how those changes translate into disease risk — a crucial resource for longevity medicine and therapeutic development.
Detailed Summary
Why this matters: The circulating proteome — all proteins detectable in blood — is increasingly recognized as a window into aging biology and disease risk. Proteins are both biomarkers of biological age and direct mediators of pathology. Understanding which genetic variants control protein levels, and which proteins causally influence disease, is foundational to identifying new drug targets and designing precision interventions to extend healthspan.
What was studied: This landmark study, published in Cell, integrated genome-wide association studies (GWAS) of plasma protein levels — called pQTL studies — across 54 cohorts totaling more than 100,000 participants of predominantly European ancestry. Proteins were measured using the Olink Proximity Extension Assay and SomaScan platforms, covering approximately 3,000 unique proteins. The consortium — named SCALLOP (Systematic and Combined Analysis of OLink Proteins) — performed meta-analyses to identify protein quantitative trait loci (pQTLs), genetic variants robustly associated with protein abundance.
Key results: The study identified over 600,000 pQTL associations across the proteome, with thousands of cis-pQTLs (variants near the gene encoding the protein) and trans-pQTLs (variants acting at a distance). Notably, the dataset included 16,263 independent protein-locus associations after fine-mapping. A substantial proportion of pQTLs colocalized with disease GWAS signals, providing genetic evidence that protein changes mediate disease risk. Using two-sample Mendelian randomization (MR), the authors identified over 1,000 statistically robust protein–disease causal relationships spanning cardiovascular, metabolic, neurological, and immune conditions. Specific proteins with causal links included well-validated targets like IL-6 (interleukin-6) in cardiovascular and inflammatory disease, as well as novel candidates. The study also demonstrated that cis-pQTLs are substantially more reliable instruments for MR than trans-pQTLs due to lower horizontal pleiotropy risk, a key methodological insight.
Implications for aging and longevity: The diseasome-wide MR analysis revealed that many proteins with causal disease effects are encoded by druggable genes, making this dataset a practical roadmap for drug repurposing and target prioritization. Several proteins implicated in neurodegeneration (e.g., GFAP, NEFL) and cardiometabolic disease showed strong genetic support for causal roles. The study's scale and multi-platform validation approach means findings are far more robust than any single-cohort pQTL study. For longevity medicine, proteins that causally influence multiple age-related diseases simultaneously are of particular interest — these represent high-value intervention points.
Caveats: The cohorts are predominantly of European ancestry, limiting generalizability to other populations. Protein measurement platforms (Olink and SomaScan) differ in epitope binding and coverage, introducing platform-specific effects that were partially but not fully resolved through cross-platform harmonization. MR assumptions (relevance, independence, exclusion restriction) may be violated in some protein–disease pairs, particularly for trans-pQTLs. The study is observational-genetic in design; randomized validation of the causal protein–disease pathways identified here will require clinical trials targeting these specific proteins.
Key Findings
- Over 600,000 protein-quantitative trait loci (pQTL) associations identified across ~3,000 plasma proteins in 54 cohorts (N>100,000 participants)
- 16,263 independent protein-locus associations identified after fine-mapping, providing the most comprehensive human plasma pQTL atlas to date
- More than 1,000 statistically robust protein–disease causal relationships identified via Mendelian randomization across cardiovascular, metabolic, neurological, and immune disease categories
- Cis-pQTLs showed substantially stronger colocalization with disease GWAS signals than trans-pQTLs, with cis instruments flagged as more reliable for causal inference
- A large proportion of proteins with causal disease links are encoded by known druggable genes, directly supporting drug target prioritization pipelines
- Proteins associated with neurodegeneration (including GFAP and NEFL) and cardiometabolic conditions received strong multi-cohort genetic support for causal disease roles
- Cross-platform validation (Olink vs. SomaScan) confirmed a core set of pQTLs replicable across both proximity extension assay and aptamer-based technologies
Methodology
This was a large-scale meta-analytic study integrating GWAS of plasma protein levels (pQTL studies) from 54 international cohorts comprising over 100,000 participants, predominantly of European ancestry. Proteins were measured using two platforms — Olink Proximity Extension Assay and SomaScan — covering approximately 3,000 unique proteins. Genetic associations were meta-analyzed, fine-mapped, and subjected to colocalization analyses with disease GWAS. Two-sample Mendelian randomization (MR) was used to infer causal protein–disease relationships, with sensitivity analyses distinguishing cis from trans instruments and testing for horizontal pleiotropy.
Study Limitations
The study population is predominantly of European ancestry, limiting the applicability of findings to non-European populations and potentially missing ancestry-specific pQTLs. Platform differences between Olink and SomaScan (including aptamer binding artifacts and epitope variability) introduce measurement heterogeneity that cross-platform harmonization only partially addresses. Mendelian randomization cannot fully exclude horizontal pleiotropy — particularly for trans-pQTLs — and the causal protein–disease relationships identified require experimental and clinical validation before therapeutic application. Several authors are affiliated with pharmaceutical companies (including Pfizer), representing a potential conflict of interest in the prioritization and framing of druggable targets.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe: