NT-proBNP, the blood test cardiologists lean on to track heart failure, can drift up or down by about a third in the same stable person from one month to the next, with nothing about their heart having actually changed. That is the headline number from a systematic review and meta-analysis published online June 11 in Clinical Chemistry ↗, and it has a direct bearing on how to read any single patient's result.
The work comes from the European biological-variation group that maintains the field's reference database, led by Pilar Fernández-Calle and Jorge Díaz-Garzón with Sverre Sandberg and Aasne Aarsand. They pooled 21 studies that had measured how much cardiovascular blood markers wobble within healthy people over time, and graded each study against a formal quality checklist.
What "biological variation" means
Every blood marker has a baseline jitter. Draw the same person's blood twice, days or weeks apart, with no change in their health, and the two numbers will not match exactly. Part of that gap is the lab instrument. The rest is the body itself, the natural rise and fall of a molecule around each person's own set point. That second part is the within-subject biological variation, written as CVI and quoted as a percentage.
The CVI is not a footnote. It sets the size of change a doctor needs to see in a repeat test before the change is real rather than noise. If a marker naturally swings by 10 percent on its own, a 12 percent rise on a repeat draw means little. If it swings by 34 percent, even a sizeable jump can be background.
The number depends on the gap between draws
For NT-proBNP, the inactive fragment that labs measure as a stand-in for the heart's own B-type natriuretic peptide, the meta-analysis found the jitter is not one fixed value. It grows with the time between blood draws. Across samples taken less than 15 days apart, the within-person variation was 9.9 percent (95 percent confidence interval 8.8 to 18.0). Across samples more than 15 days apart, it more than tripled, to 34.3 percent (95 percent confidence interval 25.0 to 60.0).
That split is the practical finding. Two NT-proBNP results a week apart sit on a fairly tight band, so a clear move between them is more likely to mean something. Two results months apart sit on a band wide enough that a third of the apparent change could be the molecule drifting around its own set point. A clinic comparing today's value to one from a visit half a year ago is working with a much noisier yardstick than one rechecking a hospitalized patient over days.
Most of the underlying data is weak
The quieter finding is about the evidence itself. Of the 21 studies the authors pooled, only 3 earned an A grade on the appraisal checklist, meaning full compliance with how a biological-variation study should be run and reported. For the other markers in the review, the data is thinner still: brain natriuretic peptide, the N-terminal fragment of A-type natriuretic peptide, galectin, copeptin, soluble ST2, and growth differentiation factor 15 each had only one to five eligible studies behind them. These are the markers heart-failure research keeps proposing as add-ons to NT-proBNP, and the basic question of how much each one drifts on its own has barely been answered to standard.
The platform angle
The marker at the center of this is a peptide. B-type natriuretic peptide is a 32-amino-acid hormone the heart's ventricles release when their walls are stretched by pressure or volume, and it pushes the body to widen blood vessels and shed salt and water. Labs do not usually measure the active hormone directly; they measure NT-proBNP, the inert piece snipped off when the precursor is processed, because it lingers longer in blood and is easier to assay. The active form has its own drug incarnation: a lab-made copy of human BNP is sold as nesiritide ↗, once given by infusion to relieve acute heart-failure crises. The same peptide family that the body uses as a stress gauge is both the assay target and, in recombinant form, a therapy.
For anyone tracking heart failure with serial blood tests, the message is narrow and useful. A repeat NT-proBNP is only informative against its own noise floor, and that floor is wider the longer you wait between draws. The review puts a graded number on a quantity clinicians have long treated as a single textbook constant, and shows that it is neither single nor well-measured.