Patients with type 2 diabetes (T2DM) are at elevated—but varying—risk of atherosclerotic cardiovascular disease (ASCVD). Several recent guidelines for selecting pharmacotherapies emphasize the importance of matching the intensity of preventive efforts to an individual’s risk profile.
Diabetes-specific tools for predicting ASCVD risk are available but have either been derived from noncontemporary data sets or designed to predict broad composite outcomes. For example, “cardiovascular death” is often included in composite outcomes but may be attributable to heart failure, pulmonary embolism, or other events unrelated to ASCVD.
Using contemporary data sets reflecting current clinical practices, researchers at Brigham and Women’s Hospital developed the TIMI Risk Score for Atherothrombosis in Diabetes, which uses information readily available from the medical chart to predict atherothrombotic complications in patients with T2DM.
David D. Berg, MD, MPH, a member of the Thrombolysis in Myocardial Infarction (TIMI) Study Group in the Division of Cardiovascular Medicine, Marc S. Sabatine, MD, MPH, Lewis Dexter, MD distinguished chair in Cardiovascular Medicine and chair of the TIMI Study Group, and colleagues describe the development and validation of the risk model in the Journal of the American College of Cardiology.
Methods
The effort made use of data on 42,181 patients with T2DM from four contemporary, multinational, randomized, placebo-controlled trials:
- Derivation cohort—all 11,284 patients enrolled in SAVOR–TIMI 53 and 12,359 patients enrolled in DECLARE–TIMI 58
- Validation cohort—all 8,542 patients from FOURIER, all 6,793 patients from CAMELLIA–TIMI 61, and 3,203 patients from DECLARE–TIMI 58 who had no established ASCVD
The primary outcome for this analysis was the composite of fatal or nonfatal myocardial infarction (MI) or ischemic stroke.
Risk Score Development
16 routinely assessed clinical variables proved to be independent predictors of MI or ischemic stroke in the derivation cohort:
- Demographics and anthropometrics: Older age, male sex, greater waist circumference
- Atherosclerosis risk factors: Higher low-density lipoprotein cholesterol, higher systolic blood pressure, tobacco use
- Diabetes-related factors: Insulin use, higher hemoglobin A1c
- Kidney disease–related factors: Lower estimated glomerular filtration rate, higher urine albumin-to-creatinine ratio (UACR)
- Existing ASCVD: Coronary artery disease, prior MI, prior percutaneous coronary intervention, prior coronary artery bypass grafting, prior ischemic stroke, peripheral artery disease
Overall Risk Score Performance
The primary model identified more than an eight-fold gradient of risk for MI or ischemic stroke:
- Derivation cohort: 3-year Kaplan–Meier event rate, 14.4% in the top quintile of predicted risk vs. 1.7% (P<0.0001) in the bottom quintile; Harrell’s C-index, 0.704 (95% CI, 0.692–0.717)
- Validation cohort: 3-year Kaplan–Meier event rate, 14.9% vs. 1.4% (P<0.0001); Harrell’s C-index, 0.706 (95% CI, 0.691–0.721)
Risk Score Performance by ASCVD Status
Discrimination was consistent in patients with and without established ASCVD:
Patients with established ASCVD
- Derivation cohort: 3-year Kaplan–Meier event rate, 16.7% in the top quintile of predicted risk vs. 3.4% in the bottom quintile (P<0.0001)
- Validation cohort: 16.8% vs. 4.2% (P<0.0001)
Patients without established ASCVD
- Derivation cohort: 6.2% vs. 1.1% (P<0.0001)
- Validation cohort: 5.3% vs. 1.0% (P<0.0001)
Clinical Application of the Risk Score
To evaluate the potential clinical application of the new risk score to guide the selection of therapies, the researchers tested for heterogeneity in the treatment effect of dapagliflozin versus placebo in DECLARE-TIMI 58 and of evolocumab versus placebo in FOURIER according to baseline risk as predicted by the new risk score.
The absolute reduction in MI or ischemic stroke rates tended to be greater in patients with higher baseline predicted risk for both dapagliflozin (absolute risk reduction, 2.1% vs. 0.2%) and evolocumab (ARR, 3.2% vs. 1.0%).
Conclusion
Because the TIMI Risk Score for Atherothrombosis in Diabetes performs well in patients with and without established ASCVD, it will be valuable in both primary and secondary prevention. Combining the new model with the TIMI Risk Score for Heart Failure in Diabetes will allow clinicians to provide a comprehensive, individualized cardiovascular risk profile to guide shared decision-making.
Online calculators for these risk scores can be found at https://timibiostat.shinyapps.io/calculators/.