For many years, cardiovascular risk prediction has relied heavily on traditional algorithms like the Framingham and QRISK models. These multivariable equations, derived from large cohorts and survival regression analysis, aim to estimate a person's absolute cardiovascular risk over a span of 5 to 30 years. They consider factors such as age, sex, blood pressure, cholesterol levels, smoking habits, and diabetes status.
Despite widespread use and significant contributions to public health, these models present a notable problem. A substantial portion of cardiovascular events, as demonstrated in the Prospective Urban Rural Epidemiology (PURE) study, occurs in individuals deemed low-risk by these algorithms. Specifically, 41% of first major cardiovascular events were reported in this cohort among those classified as low risk. Such findings suggest that many individuals are excluded from preventative therapies based on predicted risks, highlighting a critical limitation in current predictive models and the concept of the "low-risk paradox".
Addressing the limitations of traditional methods requires a fresh approach—broadening the predictors used in cardiovascular risk assessment. Historical analyses within the cardiovascular–kidney–metabolic framework have shown only marginal performance gains with the inclusion of individual clinical measures in predictive models, often reflected in minimal increases in the C-index, typically less than 0.01.
To truly escape this predictive ceiling, there is a clear need to redefine both what factors are assessed and the outcomes being targeted in cardiovascular risk. The omics revolution is fundamentally shaping this evolution. Genome-wide polygenic risk scores highlight lifelong susceptibilities but are static and cannot adequately reflect the dynamic state of illness over time.
On the other hand, the plasma proteome characterizes real-time interactions among genetic, environmental, and disease-related factors. Currently, advancements in assay technologies allow for the simultaneous measurement of hundreds, if not thousands, of proteins from a single blood sample, effectively paving the way for enhanced utility in risk prediction.
The outcome targets for cardiovascular risk assessment are expanding along the disease progression continuum. Traditionally focused on major clinical events, prediction models are now shifting attention towards subclinical atherosclerosis. This earlier-stage target is particularly relevant as its progression is often predictive of future events.
Coronary artery calcium (CAC) scoring exemplifies how identifying subclinical conditions can lead to more informed risk reclassification, although challenges in accessibility and cost remain. Moreover, new algorithms are also beginning to address heart failure as a target outcome, recognizing its increasing incidence and potential prevention opportunities. Notably, models like PREVENT now estimate both heart failure and atherosclerotic cardiovascular disease risks, demonstrating this dual focus.
The transition from statistical insights to clinically actionable tools presents its own set of challenges. While integrating molecular predictors with traditional factors tends to yield only modest improvements in model discrimination, such as the examples presented in recent research, their true utility may lie in their potential to enhance net reclassification.
For instance, some recent studies show that adding new protein risk scores improves predictive power by a small margin over existing models like China-PAR, while also returning some intermediate-risk individuals back to lower-risk categories. These findings raise important questions about the generalizability of these scores across varied populations and conditions.
Ultimately, as cardiovascular risk prediction methodologies evolve, the incorporation of dynamic monitoring systems such as electronic health records and wearable device data will be paramount. Continuous monitoring can contribute to risk assessments that are not static but rather evolve with the patient’s condition over time.
In conclusion, the landscape of cardiovascular risk prediction is shifting significantly. As researchers delve deeper into the integration of molecular signatures into predictive frameworks, the potential for early intervention strategies grows, marking a crucial step towards managing cardiovascular disease trajectories more effectively.