A 20-Year Proteome Map: Departures From Personal Protein Baselines Are Associated With Mortality Risk
Some Blood Proteins Have Measurable Personal Baselines
In a study published on 10 August in Nature Health, the authors tracked the levels of 10 776 protein markers in 1 298 participants. Measurements were collected in four waves: in 2002, 2007, 2012 and 2022.
The reference ranges on a laboratory report show how a result compares with values from a large group of people. For some proteins, an individual’s level remains different from the population average for years and is characteristic of that person. A personal baseline can reveal a change that would remain hidden within the population reference range.
The authors first adjusted the data for the overall shift associated with age and for systematic differences between measurement waves. For each protein, they then compared measurements from the same individuals across two visits. This produced the protein homeostasis index (PHI): the more consistently a person with a higher protein level at one visit remained above others at the next visit, the higher the PHI and the more stable that person’s individual level.
A separate laboratory measure produced the same ranking. Proteins had a high PHI when their levels varied less within an individual than they differed between individuals. This allowed the researchers to identify, among thousands of markers, those for which a person’s multiyear history could serve as a useful baseline.
The analysis of departures from this baseline included 856 participants with samples from 2002, 2007 and 2012. Using the first two measurements, the researchers calculated the protein level expected in 2012. They then examined whether the difference between the expected and observed values was associated with all-cause mortality after 2012. Among 571 proteins with a high PHI, deviations in 21 remained associated with mortality after statistical adjustment for hundreds of simultaneous tests. When the same procedure was applied to 26 repeatedly measured laboratory markers, the association remained only for high-sensitivity C-reactive protein, a marker of inflammation.
In their response to reviewers, the authors state the principle as follows: “When developing protein biomarkers for precision medicine, temporal stability should be considered alongside effect size and the strength of the association with disease.”
The authors also tested this approach in a version of a protein-based aging clock, using PHI as a filter when selecting proteins for the panel. A single measurement shows where a result stands today. Repeated measurements address a second question: how far has it departed from the individual’s own trajectory?