Summarized & reviewed by The Peptide Dispatch Editorial Team · Last reviewed July 20, 2026
Almost every blood panel run in the United States reports a number called eGFR, the estimated glomerular filtration rate. It is the standard measure of kidney function, it comes back as a tidy two-digit figure, and it is calculated from a single input: serum creatinine. What almost nobody is told is that there is a second way to estimate the same thing, using a protein called cystatin C, and that…
This dispatch covers Your Kidney Function Has Two Different Answers, and the Gap Between Them Predicts Mortality in the metabolic research category, authored by The Peptide Dispatch Editorial Team. Estimated reading time: 9 minutes. The Peptide Dispatch curates peer-reviewed peptide research for self-directed learners. All summaries are presented for Research Use Only and do not constitute medical advice.
Almost every blood panel run in the United States reports a number called eGFR, the estimated glomerular filtration rate. It is the standard measure of kidney function, it comes back as a tidy two-digit figure, and it is calculated from a single input: serum creatinine.
What almost nobody is told is that there is a second way to estimate the same thing, using a protein called cystatin C, and that the two methods frequently disagree. Not by rounding-error amounts. By thirty percent or more, in roughly one in nine ordinary outpatients.
The interesting part is what happens next. Over the past three years the research has converged on a finding that is genuinely counterintuitive: the disagreement between the two tests carries prognostic information that neither number carries on its own. The gap is the signal.
Creatinine is a waste product of muscle metabolism. Your muscles produce it at a fairly steady rate, your kidneys filter it out, and the concentration left in your blood is used as a proxy for how well the filtering is going. High creatinine implies poor filtration.
The obvious weakness is that creatinine production depends on how much muscle you have and what you eat. More muscle mass means more creatinine generated at any given level of kidney function. Less muscle means less. Meat intake raises it. Some medications block its tubular secretion without touching filtration at all. The equations used to convert creatinine into eGFR try to correct for this with age and sex terms, but those are population averages applied to individuals, and individuals vary.
Cystatin C is a small protein produced at a relatively constant rate by essentially all nucleated cells in the body. It is freely filtered by the glomerulus and then almost entirely reabsorbed and broken down in the tubules. Critically, its production is far less dependent on muscle mass and diet. It is not perfectly independent of everything, and we will get to that, but on the specific axis where creatinine is weakest, cystatin C is stronger.
So the two tests measure the same physiological quantity through different windows. When they agree, you have reasonable confidence. When they disagree substantially, something is going on.
According to research indexed in PubMed, the largest analysis of this question to date was published in JAMA in December 2025 by Estrella and colleagues, working from the Chronic Kidney Disease Prognosis Consortium (DOI). They pooled individual-level data on 821,327 outpatients across 23 cohorts, with a mean follow-up of eleven years, plus a further 39,639 hospitalized patients.
They defined a large negative discordance as a cystatin C-based eGFR at least 30 percent lower than the creatinine-based eGFR. Among outpatients, 11 percent met that definition. Among hospitalized patients, 35 percent did.
The outcomes in that discordant group, compared with people whose two estimates agreed within 30 percent in either direction:
Note the ordering. The discordance predicted death and cardiovascular events more strongly than it predicted kidney failure. That is a strange result for something billed as a kidney test, and it is the central puzzle in this literature.
A UK Biobank analysis published in Clinical Chemistry in 2025 by Liu and colleagues found the same pattern in a general, largely healthy middle-aged population (DOI). Across 325,356 participants followed for a median of 13.7 years, 15.5 percent had a discordantly lower cystatin C-based estimate, and that group carried a 53 percent higher mortality risk (HR 1.53, 1.48 to 1.57). In the subgroup where the cystatin C estimate was below 60 percent of the creatinine estimate, mortality risk roughly doubled (HR 2.25, 2.04 to 2.47).
The mirror-image finding is just as notable. Participants whose cystatin C estimate was discordantly higher than their creatinine estimate had a 30 percent lower mortality risk (HR 0.70, 0.66 to 0.75). The gap runs in both directions and carries information in both directions.
The intuitive story goes like this. Cystatin C is the honest test. Creatinine is inflated or deflated by muscle mass. If your cystatin C estimate is much worse than your creatinine estimate, it is because you have low muscle mass, low muscle mass means frailty and sarcopenia, and frailty is what is actually killing people. The discordance is just a muscle-mass detector wearing a lab coat.
That story is partly right and mostly insufficient, and the study that shows it is worth reading carefully.
McCoy and colleagues published an analysis of the Chronic Renal Insufficiency Cohort in the American Journal of Kidney Diseases in 2025 (DOI). This cohort is unusual because it has directly measured GFR using iothalamate clearance, not just estimates, along with 24-hour urine creatinine collections as a muscle-mass proxy and plasma measurements of larger molecules such as beta-2 microglobulin.
Two findings matter. First, when they adjusted the association between discordance and outcomes for muscle mass, protein intake, obesity, inflammation, and middle-molecule clearance, the associations with death and heart failure hospitalization were essentially unchanged. Second, all of those putative causes together explained only 36 percent of the variance in the gap between the two estimates. Nearly two thirds of it remains unaccounted for.
So the honest position is this: we know the discordance predicts hard outcomes, we know it is not simply a proxy for muscle mass, and we do not yet know what most of it represents. Candidate explanations include impaired clearance of larger molecules that creatinine cannot see, subclinical inflammation, and cystatin C picking up some component of general physiological reserve. These are hypotheses under investigation, not established mechanism.
Anyone who tells you confidently what the gap means is ahead of the data.
There is one application where the stakes are concrete and near-term. Renally cleared drugs are dosed off eGFR. If your creatinine-based eGFR overstates your true filtration, you get dosed as though your kidneys clear the drug faster than they do, and the drug accumulates.
Hanna and colleagues examined this in 1,869 cancer patients at two Boston academic centers, published in JAMA Network Open in 2023 (DOI). Twenty-nine percent had a cystatin C estimate more than 30 percent below their creatinine estimate. That group had markedly more supratherapeutic drug levels, including vancomycin troughs above 30 micrograms per milliliter in 24 percent versus 9 percent of concordant patients, with an adjusted odds ratio of 2.59. Their adjusted 30-day mortality hazard ratio was 1.98.
That is a specific, mechanistic, actionable harm. It is also the reason cystatin C testing has moved fastest in oncology and inpatient medicine rather than in wellness settings.
This is not fringe. KDIGO guidance has for years recommended confirmatory cystatin C testing in adults whose creatinine-based eGFR falls in the 45 to 59 range without other markers of kidney damage, precisely because that band is where misclassification is most consequential. The CKD-EPI 2021 equations include a validated combined creatinine-plus-cystatin C formula, which generally outperforms either marker alone against measured GFR.
What has changed recently is the weight of evidence that the two-marker comparison is useful in people who are not suspected of having kidney disease at all.
Cystatin C is not a clean marker either. Its levels are influenced by thyroid dysfunction in both directions, by corticosteroid use, by obesity, by smoking, and by systemic inflammation. Some of what looks like superior prognostic performance may reflect cystatin C partially functioning as an inflammation marker rather than a pure filtration marker. That possibility does not make the risk prediction less real, but it does complicate the claim that cystatin C is simply the more accurate kidney test.
Assay standardization has improved substantially but historically lagged behind creatinine. The test costs more than creatinine, though it remains inexpensive in absolute terms and is widely available through major reference labs.
Most importantly: there is no randomized trial showing that measuring cystatin C, discovering a discordance, and then doing something about it improves survival. The evidence is prognostic. It tells you the gap identifies higher-risk people. It does not yet tell you that any intervention closes that risk, because the intervention question has not been answered.
Two accepted methods exist for estimating kidney function, they disagree by 30 percent or more in roughly 11 percent of outpatients and 35 percent of inpatients, and that disagreement is associated with a 53 to 69 percent higher risk of death across two very large independent datasets. The disagreement predicts cardiovascular events and mortality more strongly than it predicts kidney failure. Low muscle mass is a real contributor but explains only a minority of the effect. In renally dosed drug therapy the discordance produces measurable overdosing and measurable harm.
What does not survive: the idea that a single eGFR number derived from creatinine is a settled answer about your kidney function, and equally, the idea that cystatin C is a clean replacement that resolves the question. The useful information is in the comparison, and the comparison only exists if both tests get ordered.
Most panels order one.
This article is educational and is not medical advice. Laboratory interpretation belongs in a clinical relationship with a licensed clinician who knows your history. Research citations retrieved via PubMed.
Educational content — not medical advice. Effects described are drawn from cited research in study subjects.