Machine learning contributes to advances in surrogate end points for cardiovascular clinical trials

One of the great challenges in improving health in old age is measuring how well someone is ageing. An individual who is “aging well” will be less likely to get ill than someone of the same age who is “aging badly”. UK SPINE wants to develop new medicines that will help people age well and make them less likely to get ill. But how do you measure this?

In a recently published Science Translational Medicine article, Williams et al. describe a new test, developed by SomaLogic, which will predict how likely an individual is to experience certain health conditions within the next four years. This could help shorten cardiovascular disease clinical trials, enhance the cost-effectiveness of drug prescribing and improve outcomes for patients.

As also reported in The Guardian, the blood plasma test uses machine learning to produce a signature of 27 proteins which can be used to predict the four-year likelihood of heart attack, stroke, heart failure or death.

While currently risk scores are based on age, sex, race, medical history, cholesterol and blood pressure, the SomaLogic test, as validated in 11,609 individuals, is twice as accurate at predicting risk of cardiovascular issues. The results from the test indicate allows categorisation of people from high risk to low risk, with the generation of a percentage likelihood of suffering an adverse cardiovascular event in the subsequent four years.

The test contributes to advancing personalised medical practices, where patients are treated in line with their risk factors, thereby contributing to cost effectiveness, as well as providing a method by which the efficacy of a drug can be assessed during treatment protocol. In addition, there is potential for the test to be used as a surrogate endpoint in clinical trials, thereby reducing the duration of drug development.

UK SPINE is aiming to develop new medicines that can improve health in old age. We are currently working on a similar blood test that would tell us if an existing class of drugs called bisphosphonates can reduce the chance of people getting ill as they age. We are comparing blood from patients who have taken bisphosphonates to those who haven’t to see if we can find a protein signature that may indicate improved health.

In our quest to discover and bring to market therapeutics to treat and prevent multiple long term conditions, we recognise a major barrier to be the availability of clinically relevant endpoints which can facilitate trials in this area. We therefore welcome this announcement and are excited about the potential role tests such as this can play in the drug development process.