A universal risk prediction model for MACE in persons with and without ASCVD

In a cohort of persons from a US community-based study, a universal risk prediction model for MACE incorporating established risk factors and biomarkers showed good performance in persons with and without ASCVD.

This summary is based on the publication of Mok Y, Dardari Z, Sang Y, et al. - Universal Risk Prediction for Individuals With and Without Atherosclerotic Cardiovascular Disease. J Am Coll Cardiol. 2024 Feb 6;83(5):562-573. doi: 10.1016/j.jacc.2023.11.028.

Introduction and methods

Background

Different risk prediction tools are used in primary and secondary CVD prevention [1-7]. However, these different tools use similar risk predictors, raising the question whether a universal risk prediction system can be used to assess CVD risk in persons with and without ASCVD.

Aim of the study

The aim of the study was to: (1) evaluate whether established and other CVD risk predictors are similarly associated with MACE in persons with and without ASCVD, and (2) develop and validate a novel universal risk prediction model for persons with and without ASCVD.

Methods

The authors used data of the ARIC (Atherosclerosis Risk In Communities) study to evaluate the performance of predictors. ARIC was an observational cohort study with 15,792 individuals who were between 45 and 64 years old at visit 1 (1987-1989) from 4 US communities. Participants were invited to follow-up visits in 1990-1992 (visit 2), 1993-1995 (visit 3) and 1996-1998 (visit 4), and participated in annual health status follow-up via telephone calls. For the current study, data from visit 4 was used as baseline (n=609 with a history of ASCVD, and n=8529 without a history of ASCVD). Participants were followed until Dec 31, 2020, the date of outcome of interest, or loss to follow-up, whichever occurred first. Median follow-up period was 18.9 (IQI: 10.8-22.9) years.

The authors implemented several models to assess the interaction between predictors and MACE. The base model included established predictors of the pooled cohort equation (PCE) and dichotomous secondary risk classification system from the 2018 ACC/AHA Cholesterol Guideline. Extended models included the predictors of the base model plus nonlaboratory-based predictors (extended model 1; family history, BMI, and heart rate), plus laboratory-based predictors (extended model 2; hs-CRP, Lp(a), triglyceride, and ApoB), plus 2 major cardiac biomarkers (NT-proBNP and hs-cTnT) in clinical practice (extended model 3), or plus all the added predictors of extended model 1 to 3 (extended model 4).

Predictors for the new universal prediction model were identified by using least absolute shrinkage and selection operation regression and bootstrapping.

Outcomes

The primary outcome was MACE (a composite of MI, stroke, and HF, including fatal and nonfatal events).

Main results

CVD candidate predictors

• A total of 3209 participants developed MACE, of which 2797 persons had no history of ASCVD and 412 persons had a history of ASCVD. The incidence rate per 1000 person-years was 21.3 for MACE, 12.6 for MI/stroke, and 13.8 for HF.
• Most predictors of the base model had similar associations with MACE in persons with and without ASCVD at baseline. There were a few exceptions:
  • weaker associations were detected for age, systolic blood pressure and total cholesterol in persons with ASCVD compared with persons with no ASCVD (P for interaction= 0.002, 0.045, and 0.027, respectively);
  • HDL-c was more strongly associated with MACE in persons with ASCVD compared with persons with no ASCVD (P for interaction=0.040).
• The predictors of the extended 1 to 3 models showed similar associations with MACE in persons with and without ASCVD.
• The C-statistic of the base model was 0.675 and 0.725 for persons with and without ASCVD at baseline, respectively. The inclusion of all predictors in the model (extended 4 model) led to the best improvements in discrimination in persons with and without ASCVD (∆C-statistic: 0.039; 95%CI: 0.007-0.071, and ∆C-statistic: 0.035; 95%CI: 0.021-0.047 respectively).
• Extended model 1 did not show any improvements compared with the base model, whereas extended model 2 and 3 showed small improvements in persons without ASCVD (extended model 2, ∆C-statistic: 0.007; and extended model 3, ∆C-statistic: 0.027) but not in persons with ASCVD.

The universal prediction model

• 10 variables were identified for the universal prediction model, which were age, interaction age x history of ASCVD, diabetes, systolic blood pressure, antihypertensive medication, interaction total cholesterol x history of ASCVD, current smoker, log hs-CRP, log NT-proBNP, and log hs-cTnT.
• The universal prediction model showed good discrimination for MACE in persons without ASCVD (C-statistic: 0.748; 95%CI: 0.726-0.770) and in persons with ASCVD (C-statistic: 0.692; 95%CI: 0.650-0.735).
• The 5-year predicted risk according to the universal predication model was comparable to the 5-year observed risk (calibration slope was 1.05 [95%CI: 0.93-1.18] in persons without ASCVD and 0.92 [95%CI: 0.69-1.15] in persons with ASCVD).
• In general, the risk of MACE was lower in persons without ASCVD compared with persons with ASCVD according to the universal prediction model. However, the 5-year observed risk in the highest quintile of predicted risk in persons without ASCVD was higher than that of the lowest 2 quintiles of predicted risk in persons with ASCVD.
• The universal prediction model was validated in an external validation cohort using data from MESA (Multi-Ethnic Study of Atherosclerosis).

Conclusion

Using data from the observational ARIC cohort, it was shown that most of the established predictors showed similar associations with MACE in persons with and without a history of ASCVD. The inclusion of additional predictors on top of the established predictors in the base model improved discrimination performance for MACE, irrespective of ASCVD history.

The authors identified 10 predictors for the universal risk prediction model for MACE, which were age, interaction age x history of ASCVD, diabetes, systolic blood pressure, antihypertensive medication, interaction total cholesterol x history of ASCVD, current smoker, log hs-CRP, log NT-proBNP, and log hs-cTnT. The novel universal risk prediction approach demonstrated good discrimination and calibration in persons with and without ASCVD. “This approach may [..] contribute to more efficient risk communication between providers and patients in the transition from primary prevention to secondary prevention after the occurrence of initial CVD”, according to the authors.

Find this article online at JACC

References

1. Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. Executive Summary of The Third Report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, And Treatment of High Blood Cholesterol In Adults (Adult Treatment Panel III). JAMA. 2001;285: 2486–2497.

2. Goff DC Jr, Lloyd-Jones DM, Bennett G, et al. 2013 ACC/AHA guideline on the assessment of cardiovascular risk: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol. 2014;63:2935–2959.

3. Grundy SM, Stone NJ, Bailey AL, et al. 2018 AHA/ACC/AACVPR/AAPA/ABC/ACPM/ADA/AGS/ APhA/ASPC/NLA/PCNA guideline on the manage ment of blood cholesterol: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. J Am Coll Cardiol. 2019;73:e285–e350.

4. Arnett DK, Blumenthal RS, Albert MA, et al. 2019 ACC/AHA guideline on the primary prevention of cardiovascular disease: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guide lines. J Am Coll Cardiol. 2019;74:e177–e232.

5. Whelton PK, Carey RM, Aronow WS, et al. 2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA guideline for the prevention, detection,evaluation, and management of high blood pressure in adults: a report of the American College of Car diology/American Heart Association Task Force on Clinical Practice Guidelines. J Am Coll Cardiol. 2018;71:e127–e248.

6. Antman EM, Cohen M, Bernink PJ, et al. The TIMI risk score for unstable angina/non-ST elevation MI: A method for prognostication and therapeutic decision making. JAMA. 2000;284: 835–842.

7. Granger CB, Goldberg RJ, Dabbous O, et al. Predictors of hospital mortality in the global registry of acute coronary events. Arch Intern Med. 2003;163:2345-2353.