Circulating proteins paint a powerful picture of health

To get a comprehensive picture of health status, doctors today must rely on a battery of expensive and time-consuming tests. A recent study¹ published in Nature Medicine suggests that patient-specific patterns in the plasma proteome could provide a single simultaneous and holistic insight into multiple health parameters.

Researchers profiled ~7,000 proteins in nearly 17,000 patients (>85 million total datapoints). Based on plasma protein patterns shared among patients with the same health conditions, the researchers developed 11 different health indicator models. In the future, this work could be expanded to include hundreds of different models for different diseases and conditions, resulting in a comprehensive liquid health check from a single blood test.

From proteins to the proteome

Personalized medicine to date has focused primarily on genomics, although multiple factors beyond the genome influence health status. Indeed, the genome is a static measure, but the proteome is ever-changing, reflecting not only the basic genetic makeup of an individual, but also changes in gene expression over time in response to environment, diet, etc. Unfortunately, prior to the SomaScan® Assay, our technological ability to measure more than a handful of the 30,000 or so fundamental human proteins in a single assay has lagged behind that of genome sequencing.

The SomaScan Assay enables scientists to measure as many as 7,000 human proteins in a single sample simultaneously. Although not the total human proteome, the current list of measurable proteins is comprised largely of those that are already known to play important roles in human health.

Measuring 7,000 proteins from a single blood draw

The authors of the Nature Medicine paper set out to determine if differing expression levels of the 7,000 proteins measured by SomaLogic’s current SomaScan Assay could be a proxy for a variety of important health conditions. Using machine-learning techniques, they looked for patterns in protein expression that aligned closely with health metrics gathered from more traditional health-measurement methods like exercise tests and ultrasounds. They translated the complex language of proteins into the language of health by developing predictive models that linked specific protein patterns to specific health statuses.

Predicting health from proteins

Essentially, the researchers asked the following questions: Can we use protein scanning from a single blood draw to measure multiple important health factors? What about predicting health risks and health-related behaviors?

The short answers are “yes and yes”. By measuring 7,000 proteins circulating in plasma, the researchers identified biomarkers for current health states such as body fat composition and heart health, risky behaviors such as smoking, and even predicted the future development of health complications.

Significant findings

• Characterized 6 current health states: liver fat, kidney function, body fat, lean body mass, visceral fat, and cardiopulmonary fitness.
• Predicted 3 health-related behaviors: alcohol consumption, weekly physical activity, and cigarette smoking
• Forecasted two disease risks: conversion from prediabetes to diabetes in 10 years and probability of a cardiovascular event in 5 years

Using a single platform to replace multiple tests

This study demonstrated an important proof-of-concept: it is possible to use a single protein assay to gather multiple health-related metrics. The researchers estimated that to get the same answers provided by this one assay, patients would have to visit a doctor approximately nine times. Those visits would have to include lab testing, exercise testing, and imaging, amounting to a substantial healthcare cost.

Importantly, the researchers also saw relatively little overlap between the protein biomarkers correlated with each condition, despite the fact that the conditions analyzed in this study all relate to metabolic and cardiovascular health. This suggests that the protein signatures they identified are truly specific health markers (like fatty liver or predisposition to diabetes) and not an artifact of more general health conditions (like obesity).

This significant study blazes a new trail for the field of precision medicine, underlining the critical role for proteomics in the future of individual (and population) health screening and maintenance. The 11 specific health condition models identified in this study just scratch the surface of what comprehensive protein measurement can bring to health care.