ASPREE-iPGS
Polygenic Risk Identifies Older Adults Who May Benefit From Aspirin for the Primary Prevention of Ischemic Stroke
Bottom Line
In older adults without prior CVD, individuals in the highest iPGS quintile experienced a 51% relative reduction in ischemic stroke with aspirin (HR 0.49) without increased major bleeding, while no benefit was seen overall or in lower-risk quintiles — supporting polygenic risk stratification as a potential tool to target aspirin for stroke primary prevention.
Major Points
- Post hoc genetic analysis of 12,031 European-ancestry ASPREE participants ≥70 years using an updated integrative PGS (22 stroke-related GWAS traits, 1.2 million variants).
- Each 1-SD increase in iPGS raised ischemic stroke risk (HR 1.39 [1.20–1.62]); Q5 vs Q1 HR 2.62 [1.60–4.28].
- Significant aspirin×iPGS interaction for ischemic stroke (P=0.037 continuous; P=0.045 Q5 vs Q1) but not for major bleeding.
- Aspirin reduced ischemic stroke by 51% only in Q5 (HR 0.49 [0.28–0.85], P=0.011); no benefit in overall cohort (HR 0.76, P=0.063) or lower quintiles.
- Bleeding harm was concentrated in Q1 (HR 1.72 [1.11–2.65], P=0.015); Q5 bleeding was not increased (HR 1.15 [0.71–1.88]).
- Net benefit–harm per 1000 person-years: +2.6 in Q5 (favorable), −1.0 overall, −5.2 in Q1 (harmful).
- Genetic risk scores captured susceptibility not reflected in clinical risk equations (Framingham, SCORE2-OP), which failed to identify aspirin-responsive subgroups in ASPREE.
- Hypothesis-generating: post hoc design, European-ancestry only, no formal multiple-comparison adjustment; requires external multiancestry validation.
Design
Study Type: Post hoc genetic sub-analysis of a randomized, double-blind, placebo-controlled trial (ASPREE)
Randomization: 1
Blinding: Double-blind (parent ASPREE trial)
Enrollment Period: 2010–2017
Follow-up Duration: Median 4.6 years (IQR 3.5–5.6) during the randomized controlled period
Countries: Australia, United States
Sample Size: 12031
Analysis: Cox proportional hazards regression on the randomized treatment phase; aspirin×iPGS interaction terms; two nested models (model 1: age, sex, top 10 genetic PCs; model 2: + smoking, alcohol, BMI, SBP, DBP, HDL-C, non-HDL-C, hemoglobin, creatinine, diabetes, antihypertensive and statin use, family history of stroke). STROBE guidelines followed.
Inclusion Criteria
- Community-dwelling adults aged ≥70 years (≥65 years for US minorities) at ASPREE enrollment
- No prior diagnosed atherothrombotic cardiovascular disease events
- No dementia, physical disability, or other illness likely to limit survival to <5 years
- European ancestry (per genetic principal component analysis with 1000 Genomes phase 3 reference)
- Contributed biospecimens to the ASPREE Healthy Ageing Biobank and successfully genotyped
- Randomized during the ASPREE controlled treatment period
- Postimputation genotype quality control passed (R² ≥0.3)
Exclusion Criteria
- Uncontrolled hypertension (SBP ≥180 mm Hg or DBP ≥105 mm Hg)
- Prior cardiovascular disease events, dementia, or disability at baseline
- Non-European ancestry (excluded from this sub-analysis due to insufficient power for ancestry-specific PGS)
- Age <70 years at enrollment (post hoc restriction)
- Close biological relatives (coefficient of relatedness >0.05)
- Imputation quality score R² <0.3
Arms
| Field | Aspirin | Control |
|---|---|---|
| Intervention | Enteric-coated aspirin 100 mg orally once daily | Matching placebo once daily |
| Duration | Median 4.6 years (randomized ASPREE treatment phase) | Median 4.6 years |
Outcomes
| Outcome | Type | Control | Intervention | HR / OR / RR | P-value |
|---|---|---|---|---|---|
| Incident ischemic stroke (WHO criteria; acute focal neurological deficit of vascular origin ≥24 h or fatal; imaging- or autopsy-confirmed) during the randomized controlled period, analyzed for aspirin×iPGS interaction. | Primary | Placebo: overall event rate; Q5 iPGS 7.6/1000 person-years | Aspirin: Q5 iPGS 4.0/1000 person-years | 0.49 (Q5 aspirin vs placebo); 0.76 (overall cohort aspirin vs placebo) | 0.011 (Q5); 0.063 (overall); Pinteraction=0.037 (continuous iPGS×aspirin), 0.045 (Q5 vs Q1) |
| Ischemic stroke per 1-SD iPGS increase (fully adjusted) | Secondary | — | — | 1.39 | <0.001 |
| Ischemic stroke Q5 vs Q1 (fully adjusted) | Secondary | Q1 reference | Q5 | 2.62 | <0.001 |
| Major bleeding — overall cohort (aspirin vs placebo) | Secondary | Placebo | Aspirin | 1.3 | 0.014 |
| Major bleeding — Q5 (aspirin vs placebo) | Secondary | Placebo | Aspirin | 1.15 | ns |
| Major bleeding — Q1 (aspirin vs placebo) | Secondary | Placebo | Aspirin | 1.72 | 0.015 |
| Aspirin×iPGS interaction — major bleeding (continuous) | Secondary | — | — | 0.245 (model 2); all quintile Pinteraction >0.05 | |
| Net benefit–harm (Q5, per 1000 person-years) | Secondary | Placebo | Aspirin | descriptive | |
| Net benefit–harm (overall cohort, per 1000 person-years) | Secondary | Placebo | Aspirin | descriptive | |
| Net benefit–harm (Q1, per 1000 person-years) | Secondary | Placebo | Aspirin | descriptive | |
| Total major bleeds | Adverse | 373 events (101 intracranial; 46 hemorrhagic strokes) | |||
| Major bleeding aspirin (overall) | Adverse | HR 1.30 [1.05–1.61]; P=0.014 vs placebo | |||
| Major bleeding Q1 (aspirin vs placebo) | Adverse | 10.7 vs 6.2 per 1000 person-years; HR 1.72 [1.11–2.65]; P=0.015 | |||
| Major bleeding Q5 (aspirin vs placebo) | Adverse | 7.0 vs ~8.2 per 1000 person-years; HR 1.15 [0.71–1.88] (ns) | |||
| Notable pattern | Adverse | Bleeding rates on aspirin declined across increasing iPGS quintiles (10.7 in Q1 to 7.0 in Q5); higher polygenic stroke risk did not confer higher bleeding risk | |||
Subgroup Analysis
Analyses stratified by iPGS quintiles Q1–Q5. Prior Lp(a) and CAD PGSs tested as alternate explanations — neither showed significant aspirin×PGS interaction for ischemic stroke (all P>0.05; Table S4), supporting specificity of the iPGS finding. Event counts were too low to permit stroke-subtype (large-vessel/small-vessel/cardioembolic) or bleeding-subtype (intracranial vs extracranial) analyses.
Criticisms
- Post hoc, hypothesis-generating design without prespecified aspirin×iPGS interaction analysis; no formal adjustment for multiple comparisons.
- Restricted to European-ancestry participants; generalizability to non-European populations is unknown and requires multiancestry validation.
- ASPREE contributed to the GIGASTROKE consortium used to derive the iPGS (<1% of European samples) — a minor theoretical overfitting risk, though unlikely to bias interaction estimates.
- Insufficient events for clinically important subtype analyses (disabling vs nondisabling stroke; intracranial vs gastrointestinal bleeding).
- Simplified 1:1 net benefit–harm calculation weighting ischemic strokes and major bleeds equally oversimplifies real clinical utility (severity, disability, mortality differ).
- Only genotyped ASPREE participants included (12,031 of 19,114); genotyping selection could introduce bias, though baseline characteristics were similar to the overall cohort (Table S2).
- Findings pertain to primary prevention in older adults (≥70) without prior CVD and cannot be extrapolated to younger patients or secondary prevention.
- Real-world clinical utility, cost-effectiveness, and infrastructure for delivering polygenic risk stratification remain unproven — implementation studies required.
Funding
ASPREE parent trial funding: US National Institute on Aging (NIA), National Cancer Institute, National Health and Medical Research Council (NHMRC) of Australia, Monash University, and the Victorian Cancer Agency. Full funding disclosures in the source publication.
Based on: ASPREE-iPGS (Stroke, 2026)
Authors: Yu C, Hussain SM, Fransquet PD, ..., senior author)
Citation: Stroke. 2026;57(7). Originally Published 8 May 2026. doi:10.1161/STROKEAHA.125.054979
Content summarized and formatted by NeuroTrials.ai.