Polygenic risk score adds to a clinical risk score in the prediction of cardiovascular disease in a clinical setting

Abstract

Background and aims: A cardiovascular disease polygenic risk score (CVD-PRS) can stratify individuals into different categories of cardiovascular risk, but whether the addition of a CVD-PRS to clinical risk scores improves the identification of individuals at increased risk in a real-world clinical setting is unknown.

Methods: The Genetics and the Vascular Health Check Study (GENVASC) was embedded within the UK National Health Service Health Check (NHSHC) programme which invites individuals between 40-74 years of age without known CVD to attend an assessment in a UK general practice where CVD risk factors are measured and a CVD risk score (QRISK2) is calculated. Between 2012-2020, 44,141 individuals (55.7% females, 15.8% non-white) who attended an NHSHC in 147 participating practices across two counties in England were recruited and followed. When 195 individuals (cases) had suffered a major CVD event (CVD death, myocardial infarction or acute coronary syndrome, coronary revascularisation, stroke), 396 propensity-matched controls with a similar risk profile were identified, and a nested case-control genetic study undertaken to see if the addition of a CVD-PRS to QRISK2 in the form of an integrated risk tool (IRT) combined with QRISK2 would have identified more individuals at the time of their NHSHC as at high risk (QRISK2 10-year CVD risk of ≥10%), compared with QRISK2 alone.

Results: The distribution of the standardised CVD-PRS was significantly different in cases compared with controls (cases mean score .32; controls, -.18, P = 8.28×10-9). QRISK2 identified 61.5% (95% confidence interval [CI]: 54.3%-68.4%) of individuals who subsequently developed a major CVD event as being at high risk at their NHSHC, while the combination of QRISK2 and IRT identified 68.7% (95% CI: 61.7%-75.2%), a relative increase of 11.7% (P = 1×10-4). The odds ratio (OR) of being up-classified was 2.41 (95% CI: 1.03-5.64, P = .031) for cases compared with controls. In individuals aged 40-54 years, QRISK2 identified 26.0% (95% CI: 16.5%-37.6%) of those who developed a major CVD event, while the combination of QRISK2 and IRT identified 38.4% (95% CI: 27.2%-50.5%), indicating a stronger relative increase of 47.7% in the younger age group (P = .001). The combination of QRISK2 and IRT increased the proportion of additional cases identified similarly in women as in men, and in non-white ethnicities compared with white ethnicity. The findings were similar when the CVD-PRS was added to the atherosclerotic cardiovascular disease pooled cohort equations (ASCVD-PCE) or SCORE2 clinical scores.

Conclusions: In a clinical setting, the addition of genetic information to clinical risk assessment significantly improved the identification of individuals who went on to have a major CVD event as being at high risk, especially among younger individuals. The findings provide important real-world evidence of the potential value of implementing a CVD-PRS into health systems.

Keywords: ASCVD-PCE; CVD risk assessment; Polygenic risk scores; QRISK2; SCORE2.

Structured Graphical Abstract

GENVASC design and key findings: The GENVASC study was set up to evaluate whether the addition of a polygenic risk score for cardiovascular disease (CVD-PRS) to clinical risk assessment at an NHS Health Check would increase the number of individuals who went to have a major CVD event as being at high risk at their health check. 44 141 individuals free of CVD were recruited at their NHS Health Check and when 195 CVD events had occurred during follow-up, a retrospective case-control study was undertaken to see how many additional individuals would have been identified as high risk at their health check by the addition of the CVD-PRS in the form of an integrated risk tool (IRT) to assessment by the clinical risk score (QRISK2) alone.

Figure 1

Distribution of standardised cardiovascular disease polygenic risk score by case status. (A) Distribution of cardiovascular disease polygenic risk score in all major cardiovascular disease event cases (blue) and controls (pink); (B) Distribution of cardiovascular disease polygenic risk score in major cardiovascular disease event cases (blue) and controls (pink) partitioned by gender; (C) Distribution of cardiovascular disease polygenic risk score in major cardiovascular disease event cases (blue) and controls (pink) partitioned by ethnicity. (D) Distribution of cardiovascular disease polygenic risk score in major cardiovascular disease event cases (blue) and controls (pink) partitioned by age group. Vertical lines indicate mean polygenic risk score value in each category

Figure 2

Enrichment of cases in the top percentiles of the cardiovascular disease polygenic risk score distribution. Bars represent the top 20% (blue), top 10% (purple), top 5% (orange), top 3% (pink) and top 1% (green) of the polygenic risk score distribution. Y-axis represents the proportion of major cardiovascular disease event cases captured in individuals at or above the stated polygenic risk score percentile. Yellow lines indicate no-enrichment values (% of cases matches % of polygenic risk score distribution)

Figure 3

Percentage of cases identified as at high risk at the time of the National Health Service Health Check by QRISK2, Atherosclerotic Cardiovascular Disease Pooled Cohort Equations or SCORE2 alone or a combination of QRISK2, Atherosclerotic Cardiovascular Disease Pooled Cohort Equations or SCORE2 and integrated risk tool. Y-axis represents the percentage of major cardiovascular disease event cases identified as at high risk by QRISK2 (A), atherosclerotic cardiovascular disease pooled cohort equations (B) or SCORE2 (C) (purple) or a combination of each risk tool and their respective integrated risk tool (blue) at the time of their National Health Service Health Check for all cases and for those aged 40–54 years of age at the time of their assessment. ** indicates that the P-value for difference is <.01. *** indicates that the P-value for difference is <.001