Cardiovascular disease (CVD) is a catchall term for any medical condition affecting the heart or the blood vessels. Serious cardiovascular health issues include heart attacks, strokes, and heart failure, events often preceded by hypertension (high blood pressure) and/or atherosclerosis (plaque in the arteries). Even with medical monitoring and a large array of cardiovascular drugs, CVD is still the leading annual cause of death in the United States with roughly 700,000 fatalities per year. One of the reasons that deaths remain high is the lack of effective predictive algorithms that are personalized for each patient. For example, the Framingham Risk Score is a widely used predictive tool that looks at six factors (age, gender, cholesterol levels, blood pressure, diabetes, and smoking status). This tool gives a general assessment of your risk for a serious cardiovascular event based on population averages. What it fails to do is give a personalized assessment that takes into account your individual genetics, biochemistry, and metabolism. Two people with the same Framingham score could actually have vastly different CVD risks due to factors not accounted for in this calculator. This makes Framingham and similar approaches fairly blunt tools. Having a predictive tool based on accurate measurements of our detailed personal characteristics (our biomarkers) would provide a more accurate risk assessment which would allow treatments to be tailored for each individual. It would also allow treatment efficacy to be assessed and tweaked based on measured changes in the biomarkers which would enhance the therapeutic benefit to the patient.
A new study in Science Translational Medicine used a proteomics approach to investigate plasma proteins as potential biomarkers for CVD. Proteomics is the large-scale study of many proteins simultaneously, and in this application, the goal was to correlate levels of plasma proteins with CVD risk. The researchers evaluated 5000 proteins in 32,130 patient samples from 22,849 individuals. Using machine learning, an unbiased algorithm for CVD risk was developed. The algorithm identified a panel of 27 proteins whose levels in plasma collectively created a patient profile that correlated with low, medium, or high 4-year risk for heart attack, stroke, heart failure, or death. To validate the algorithm it was tested against a different set of 11,609 patients and used to predict their risk of a cardiovascular event. The proteomic approach was twice as accurate as conventional risk scores for predicting which individuals would have a cardiovascular event within the next four years. Larger scale trials are already underway, and this proteomics-based approach has great potential for providing personalized cardiac care based on a simple blood test. While it is likely that there will be much refinement and improvement of this proteomics approach in the coming years, this is an exciting step on the pathway to highly personalized medicine.
In addition to direct patient benefit, this protein panel could be used to facilitate clinical trials and evaluations of new drugs and therapies. In current clinical trials, patients must be followed for many years to see if cardiovascular events are reduced in the test population versus the control group. Consequently, it can take years to know if a new drug is effective or not. This makes drug trials long and expensive activities which greatly slows the introduction of new drugs into clinical use. Using a protein biomarker panel as a surrogate indicator for CVD, you could simply look for changes in the panel that indicated improved cardiovascular outcomes. As such changes would likely occur much more rapidly than the actual disease events, drugs without any clinical benefit could be quickly identified and discarded while promising drugs would be fast-tracked for further study. I’m looking forward to the implementation of these new advances that speed up drug discovery, reduce drug development costs, and improve individualized patient care.