Genetic Risk Scores Fail To Predict Survival After Diagnosis

Trending 1 month ago

New investigation crossed 7 world biobanks shows that nan DNA driving illness onset does not find survival; instead, lifespan-linked genes and cross-trait scores clasp nan existent clues to prognosis.

flasks for humor analysisStudy: Limited overlap betwixt familial effects connected illness susceptibility and illness survival. Image credit: Natalia Kirsanova/Shutterstock.com

In a caller study published successful Nature Genetics, researchers tested whether familial determinants of illness consequence besides foretell post-diagnosis endurance crossed 9 diseases, and compared susceptibility versus longevity polygenic scores (PGSs) for prognosis.

Background 

Two neighbors tin stock nan aforesaid test yet unrecorded for dramatically different lengths of time. Genetics that nudge a personification toward a illness whitethorn not beryllium nan aforesaid genetics that style what happens aft nan first session visit. For years, genome-wide relation studies (GWASs) person mapped thousands of variants for who gets a disease, but acold less for really accelerated it progresses aliases whether it proves fatal.

Clinicians and families attraction astir nan second because it guides curen intensity, follow-up, and planning. Emerging biobanks and physics wellness records make endurance analyses imaginable astatine scale, yet signals look sparse. More investigation is needed to understand which familial factors genuinely foretell prognosis.

About nan study

Researchers pooled 7 biobanks (primary analysis) and registry-linked cohorts to study 9 high-mortality conditions: Alzheimer’s disease, bosom cancer, colorectal cancer, coronary artery disease, type 2 glucosuria mellitus, chronic kidney disease, bosom failure, prostate cancer, and stroke. Disease definitions and causes of decease were standardized utilizing nan International Classification of Diseases, Tenth Revision (ICD-10). The main endpoint was disease-specific mortality, pinch all-cause mortality successful sensitivity analyses.

Within-patient GWASs of disease-specific mortality utilized Cox proportional hazards models implemented successful Genome-wide Analysis of Time-to-Event (GATE) aliases Saddlepoint Approximation Cox (SPACox), adjusting for property astatine diagnosis, commencement year, sex, main components (PCs), and study covariates. Eligible patients required ≥3 months follow-up. Summary statistic passed value power (imputation accusation (INFO) people > 0.7; insignificant allele count ≥ 20), were aligned to quality genome build 38 (hg38) via LiftOver, meta-analyzed pinch fixed-effect models successful Meta-Analysis Helper (METAL), and assessed for heterogeneity pinch Cochran’s Q.

PGSs were constructed pinch Mega Polygenic Risk Score (MegaPRS) nether Baseline Linkage Disequilibrium-Linkage Disequilibrium Adjusted Kinships (BLD-LDAK) assumptions; a wide longevity PGS utilized nan Linkage Disequilibrium Adjusted Kinships-Thin (LDAK-Thin) model. Associations pinch test and post-diagnosis endurance were tested via logistic aliases Cox models. Sensitivity analyses addressed subsister bias, follow-up truncation (2/5/10 years), age-at-diagnosis strata, and relatedness. For type 2 glucosuria mellitus, macrovascular and microvascular complication endpoints were analyzed, pinch matched GWASs successful non-diabetic populations to probe shared architecture.                  

Study results

Across 9 diseases, only 1 genome-wide important locus for disease-specific mortality emerged: rs7360523 adjacent Sulfatase 2 (SULF2) for bosom nonaccomplishment mortality. Notably, that locus did not show a comparable effect connected bosom nonaccomplishment susceptibility and moreover had nan other guidance of effect successful susceptibility analyses. When nan squad compared 804 lead susceptibility variants against mortality, nary remained important aft multiple-testing correction; astir half shared nan aforesaid effect direction, nary much than expected by chance.

These patterns matched little heritability estimates for mortality versus susceptibility. When researchers equalized sample sizes and methods successful a down-sampling test, susceptibility GWASs still uncovered galore much loci than mortality GWASs, suggesting nan deficiency of mortality signals was not simply a powerfulness issue.

Disease-specific PGSs powerfully predicted who developed each illness (hazard ratios per modular deviation from ~1.17 to ~1.90), yet they were anemic predictors of disease-specific mortality aft diagnosis. In bosom failure, nan susceptibility PGS had only a humble relation pinch bosom nonaccomplishment mortality, while successful chronic kidney illness and prostate cancer, susceptibility PGSs moreover trended toward protective effects connected mortality.

In contrast, a wide longevity PGS, derived from lifespan GWAS, was importantly associated pinch disease-specific mortality successful 7 of nan 9 diseases and outperformed susceptibility PGSs successful astir settings. Notably, nan longevity PGS hit susceptibility PGSs successful 7 of 9 diseases. At nan aforesaid time, successful FinnGen, a composite mortality PGS edged retired longevity for coronary artery illness and type 2 glucosuria mellitus, highlighting nan worth of cross-trait information.

Because mortality whitethorn beryllium an imprecise proxy for progression successful immoderate diseases, investigators examined type 2 glucosuria mellitus complications. A locus connected chromosome 9 achieved genome-wide value for macrovascular complications among individuals pinch type 2 glucosuria mellitus, but was not associated pinch type 2 glucosuria mellitus susceptibility. Prior cardiovascular illness was excluded erstwhile defining macrovascular complications to guarantee a cleaner phenotype. In akin GWASs of cardiovascular traits successful group without diabetes, nan aforesaid awesome appeared: it was stronger successful nan wide organization but weaker successful those pinch diabetes, suggesting communal biology shaped by disease-specific modifiers.

Moreover, nan PGS for coronary artery illness predicted macrovascular complications successful type 2 glucosuria mellitus acold amended than nan type 2 glucosuria mellitus susceptibility PGS; for microvascular outcomes, only nan age-related macular degeneration PGS showed a mini nominal association, while nan chronic kidney illness PGS did not.

Age astatine test besides mattered: for Alzheimer’s disease, nan susceptibility PGS showed a stronger relation pinch mortality successful younger patients but not successful older ones. Simulations nether a liability-threshold model showed that conditioning connected cases tin induce index-event bias. Still, bias correction changed small successful a discourse wherever progression heritability appears debased and mortality is highly heterogeneous.

Together, nan information connote that biologic mechanisms governing who gets a illness and who dies from it overlap only modestly. Cross-trait information, for illustration lifespan aliases cardiovascular risk, tin amended seizure endurance consequence aft test than disease-specific susceptibility genetics alone.

Conclusions 

This ample multi-biobank study finds constricted overlap betwixt familial effects connected illness susceptibility and disease-specific mortality. Lead susceptibility variants seldom power survival, susceptibility PGSs execute poorly for prognosis, and a wide longevity PGS amended stratifies post-diagnosis mortality crossed galore diseases.

Clinically, this cautions against utilizing illness susceptibility scores to counsel patients astir endurance and highlights nan imaginable of cross-trait aliases longevity-informed models for consequence discussions and proceedings enrichment.

Methodologically, much power, refined progression phenotypes, and integration of related general-population traits are needed to uncover progression biology and actionable targets, particularly wherever attraction entree and treatments powerfully style outcomes.

Download your PDF transcript now!

Journal reference:

  • Yang, Z., Pajuste, F.-D., Zguro, K., Cheng, Y., Kurant, D. E., Eoli, A., Wanner, J., Jermy, B., Rämö, J., FinnGen, Kanoni, S., van Heel, D. A., Genes & Health Research Team, Hayward, C., Marioni, R. E., McCartney, D. L., Renieri, A., Furini, S., INTERVENE consortium, Mägi, R., Gusev, A., Drineas, P., Paschou, P., Heyne, H., Ripatti, S., Mars, N., & Ganna, A. (2025). Limited overlap betwixt familial effects connected illness susceptibility and illness survival. Nat Genet. DOI: 10.1038/s41588-025-02342-8. https://www.nature.com/articles/s41588-025-02342-8

More