Urinary Peptidomics: New Biomarkers for Kidney Disease
Urinary Peptides & Kidney Biomarkers
273 peptides
The CKD273 classifier combines 273 urinary peptide fragments to predict kidney disease progression with AUC values of 0.95 to 1.00 across nine clinical centers.
Siwy et al., Nephrology Dialysis Transplantation, 2014
Siwy et al., Nephrology Dialysis Transplantation, 2014
View as imageYour kidneys filter roughly 180 liters of blood every day, and the small peptide fragments that pass into urine carry a molecular record of what is happening inside your nephrons. Urinary peptidomics, the large-scale analysis of these peptide fragments using mass spectrometry, is building a new class of kidney disease biomarkers that can flag damage years before creatinine or albumin levels shift. This article covers the technology behind peptidomic profiling, the evidence for specific classifiers like CKD273, and where the field still falls short. For a broader overview of what urinary peptides reveal about the kidney, see our pillar guide to urinary peptides and kidney health.
Key Takeaways
- The CKD273 urinary peptide classifier detected progression to macroalbuminuria 1.5 years earlier than standard albumin testing, with an AUC of 0.93 in a 5-year follow-up of diabetic patients (Pontillo et al., 2017)
- Multicentre validation across 9 institutions and 165 type 2 diabetes patients showed CKD273 AUC values from 0.95 to 1.00, independent of age (16-89 years) and sex (Siwy et al., 2014)
- None of the CKD patients with a CKD273 score below 0.55 required dialysis or died during a 3.6-year follow-up, while all 15 who reached an endpoint scored above 0.55 (Argiles et al., 2013)
- A plasma peptidomics model using 14 peptide markers distinguished CKD from non-CKD patients with an AUC of 0.87 and correlated with CKD stage at r = 0.83 (Gajjala et al., 2019)
- Multi-omics combining peptidomics and metabolomics correctly classified 89.9% of healthy controls, 75.5% of type 2 diabetes, and 69.6% of early diabetic kidney disease patients in an external validation cohort of 766 participants (Jiang et al., 2023)
What Is Peptidomics and Why Does It Matter for Kidneys?
Peptidomics is the systematic study of all peptides in a biological sample. Unlike proteomics, which analyzes intact proteins, peptidomics focuses on the naturally occurring fragments, typically 0.8 to 15 kilodaltons, that result from protein processing and degradation throughout the body.[1]
Urine is an ideal sample for peptidomics because the kidney acts as a natural filter. Peptide fragments from blood filtration, tubular secretion, and local kidney tissue breakdown all end up in urine. A single urine sample contains thousands of distinct peptides, and shifts in this peptide landscape reflect changes in kidney structure and function before conventional markers respond.[2]
Serum creatinine and urinary albumin, the current clinical standards, have well-documented blind spots. Creatinine does not rise meaningfully until roughly half of kidney function is already lost. Albumin excretion can fluctuate with exercise, infection, and hydration. Both markers lag behind the actual tissue damage that peptidomics aims to catch early.
The most commonly identified peptides in urinary peptidomic studies come from collagen alpha-1(I), collagen alpha-1(III), alpha-1-antitrypsin, and uromodulin. Collagen fragments in particular reflect extracellular matrix turnover in the kidney, a process directly tied to fibrosis, the scarring that drives chronic kidney disease (CKD) progression.[1]
For a comparison of single-peptide biomarkers like cystatin C, peptidomics offers a fundamentally different approach: instead of measuring one molecule, it reads the collective signature of hundreds of fragments simultaneously.
How Mass Spectrometry Reads the Urinary Peptidome
Two primary analytical platforms dominate urinary peptidomics research: capillary electrophoresis coupled to mass spectrometry (CE-MS) and liquid chromatography coupled to mass spectrometry (LC-MS).
CE-MS separates peptides by their charge-to-size ratio using an electric field, then identifies them by mass. This approach generated the Human Urinary Proteome Database, which catalogs over 5,010 unique urinary peptides across more than 60,000 individual measurements.[3] CE-MS is the platform behind the CKD273 classifier and has been standardized across multiple clinical centers.
LC-MS separates peptides based on their chemical properties (hydrophobicity, charge) using a chromatographic column before mass analysis. Label-free quantitative LC-MS can profile approximately 6,000 polypeptide ions in undigested urine specimens, as demonstrated in kidney transplant research where Quintana et al. profiled 39 patients with chronic allograft dysfunction and 32 controls.[4]
MALDI-TOF MS (matrix-assisted laser desorption/ionization time-of-flight) offers another approach, particularly useful when combined with surface-based peptide capture. Jiang et al. (2023) used MALDI-TOF MS with porous silicon microparticles to capture urinary peptides from 766 volunteers across five disease stages, uncovering six peptides that changed in a stepwise pattern across diabetic kidney disease progression.[5]
Each platform makes different tradeoffs. CE-MS offers high reproducibility and is best validated clinically. LC-MS provides broader peptide coverage but requires more complex sample preparation. MALDI-TOF is faster per sample but less comprehensive. No single platform captures the entire urinary peptidome.
CKD273: The 273-Peptide Classifier That Predicts Kidney Decline
CKD273 is the most clinically advanced urinary peptidomics biomarker. Developed using CE-MS data, it integrates the abundance of 273 specific urinary peptide fragments into a single score that reflects kidney disease risk.[6]
Diagnostic accuracy
In the original validation, CKD273 distinguished CKD patients from healthy individuals with a sensitivity of 85% and specificity of 100% (AUC 0.96) in an independent test set of 110 CKD patients and 34 controls. The classifier separated patients according to their renal function level, and among the 273 component peptides, serum-derived fragments increased while collagen-derived fragments decreased as kidney function worsened.[6]
Prognostic power
The prognostic data is what sets CKD273 apart from diagnostic-only biomarkers. During a 3.6-year follow-up, none of the patients with a CKD273 score below 0.55 required dialysis or died, while all 15 patients who reached a renal endpoint scored above 0.55.[6] In diabetic patients, CKD273 detected progression to macroalbuminuria 1.5 years earlier than standard albumin testing, with an AUC of 0.93 compared to albumin's AUC of 0.67.[3]
Multicentre validation
Siwy et al. (2014) tested CKD273 prospectively across nine clinical centers in Europe, analyzing samples from 165 type 2 diabetes patients. The classifier showed AUC values ranging from 0.95 to 1.00 across all centers. It performed independently of age (tested range: 16-89 years), sex, and the type of urine storage container used. Fragments of blood-derived and extracellular matrix proteins were the most consistently detected components across centers.[7]
The PRIORITY trial
The largest clinical test of CKD273 was the PRIORITY trial (Proteomic prediction and Renin angiotensin aldosterone system Inhibition prevention Of early diabetic nephRopathy In TYpe 2 diabetic patients with normoalbuminuria). This prospective study enrolled 1,775 type 2 diabetes patients with normal albumin excretion across 15 European centers. Of these, 216 (12%) had a high-risk urinary proteomic pattern by CKD273. Over a median 2.5-year follow-up, the high-risk CKD273 score was associated with increased progression to microalbuminuria, independent of clinical characteristics. The embedded randomized trial tested whether spironolactone could prevent progression in high-risk patients; it did not.
Evidence appraisal
A systematic review by Critselis and Lambers Heerspink (2016) evaluated CKD273 against Oxford Evidence-Based Medicine and SORT guidelines. The classifier attained evidence level 1b (individual prospective cohort study with good follow-up), though SORT scores were lower (ranges 1-4) due to the lack of studies demonstrating added value beyond existing clinical risk predictors and cost-effectiveness data.[8]
CKD273 is commercially available as an in vitro diagnostic test and has received a letter of support from the US FDA. It is not yet part of standard clinical guidelines for CKD management.
Beyond CKD273: Plasma Peptidomics and Multi-Omics Approaches
Urinary peptidomics is the most developed approach, but it is not the only one. Plasma peptidomics and multi-omics combinations are opening additional windows into kidney disease biology.
Plasma peptide profiling
Gajjala et al. (2019) performed untargeted plasma peptidomics using LC-ESI-MS and MALDI-MS/MS on 106 CKD patients and 66 non-CKD controls. A logistic regression model selected 14 peptide markers that distinguished CKD from non-CKD with an AUC of 0.87. The predicted peptidomic score correlated strongly with CKD stage (Spearman r = 0.83, concordance index 0.899). The most consistently associated clinical variables were C-reactive protein, sodium, and uric acid, connections that had not been extensively studied in the CKD peptidome context.[9]
Plasma peptidomics captures different information than urinary peptidomics. Plasma peptides are more likely to reflect systemic inflammation and vascular damage, while urinary peptides are more kidney-specific. Neither replaces the other.
Multi-omics integration
Jiang et al. (2023) combined urinary peptidomics with metabolomics in a cohort of 766 volunteers spanning healthy controls, type 2 diabetes, early diabetic kidney disease (DKD), and overt DKD. They identified ten metabolites and six peptides that changed in a stepwise pattern across disease stages. Combined with machine learning, the multi-omics model correctly classified 89.9% of healthy controls, 75.5% of type 2 diabetes patients, 69.6% of early DKD patients, and 75.7% of overt DKD patients in an external validation cohort. It also correctly identified 87.5% of "grey-zone" patients (those with abnormal albumin but unclear diagnosis).[5]
The altered pathways concentrated in amino acid metabolism and cellular protein processing, findings that aligned with kidney tissue transcriptomics data. This cross-validation between urinary peptidome and tissue-level gene expression strengthens confidence that the peptide changes reflect real biological processes rather than analytical artifacts.
Prediabetes risk stratification
The newest application extends peptidomics upstream of kidney disease entirely. Schork et al. (2025) measured urinary peptidome classifiers in 249 individuals across six prediabetes clusters. CKD273, HF2 (heart failure), and CAD238 (coronary artery disease) classifiers all differed across clusters, with elevated values in high-risk cluster 6 (associated with nephropathy and all-cause mortality). Machine learning identified 112 urinary peptides that distinguished low-risk from high-risk prediabetes clusters with an AUC of 0.868.[10]
This work suggests peptidomic profiling could identify kidney disease risk before diabetes itself develops, a shift from early detection to true prevention.
Peptidomics for Kidney Transplant Monitoring
Kidney transplant recipients face a specific challenge: detecting chronic allograft dysfunction (CAD) early enough to adjust immunosuppressive therapy. Current monitoring relies on creatinine and protocol biopsies, both imperfect tools.
Quintana et al. (2009) applied label-free quantitative LC-MS to profile the urinary peptidome of 39 CAD patients and 32 controls. Among roughly 6,000 polypeptide ions detected, peptides from uromodulin and kininogen were markedly reduced in CAD patients. Specific ions at m/z 645.59 and m/z 642.61 distinguished different forms of CAD with 90% sensitivity and specificity in a training set and approximately 70% in an independent validation set. Combining low uromodulin expression at m/z 638.03 with high expression of m/z 642.61 identified CAD in virtually all cases.[4]
The drop in uromodulin-derived peptides is biologically coherent: uromodulin is produced by healthy tubular cells, and its decrease reflects tubular damage. This connection between the peptidome signal and the underlying biology is what makes peptidomic biomarkers mechanistically interpretable rather than purely statistical.
For context on other peptide biomarkers used to detect kidney injury, see our article on NGAL and KIM-1, which covers injury-specific markers that complement the broader peptidome-level view described here.
What Peptidomics Cannot Do Yet
The gap between peptidomic research and routine clinical use remains wide for several reasons.
Standardization challenges. While CE-MS-based CKD273 shows good reproducibility across centers, the broader peptidomics field lacks universal sample processing protocols. Pre-analytical variables (collection timing, storage temperature, freeze-thaw cycles) affect peptide profiles. Different MS platforms produce different peptide inventories from the same sample.
Clinical utility is unproven for most classifiers. CKD273 has the strongest clinical evidence but still lacks data showing it changes patient outcomes when incorporated into treatment decisions. The PRIORITY trial showed CKD273 identified high-risk patients, but the tested intervention (spironolactone) did not prevent progression. A biomarker that identifies risk without enabling better treatment has limited clinical value.
Cost and infrastructure barriers. CE-MS analysis requires specialized equipment and expertise not available in standard clinical laboratories. The cost per sample, estimated at several hundred euros, limits population-level screening. Until costs drop or the test demonstrably saves money by preventing dialysis, widespread adoption is unlikely.
Validation cohorts are still small. The largest peptidomics study in kidney disease (PRIORITY) enrolled 1,775 patients, which is large for proteomics but small by clinical biomarker standards. Drug approval biomarkers typically require validation in tens of thousands of patients across diverse populations. Most peptidomic kidney studies have been conducted in European populations, leaving questions about generalizability.
Overlap with simpler biomarkers. Whether peptidomics adds predictive value beyond established biomarkers like cystatin C, NGAL, and KIM-1, combined with clinical risk scores, remains uncertain. A systematic review of CKD273 noted this gap: the classifier's added value "beyond currently available clinical risk predictors" has not been clearly established.[8]
These are solvable problems, and the field is working on each of them. But they explain why peptidomics remains a research tool with one commercially available test rather than a standard part of nephrology practice. For researchers interested in the therapeutic side of kidney peptide biology, see our overview of peptide therapeutics for chronic kidney disease.
The Bottom Line
Urinary peptidomics has produced the most clinically advanced multi-peptide biomarker in nephrology: CKD273, a 273-peptide classifier that detects kidney disease progression years before albumin testing and performs consistently across clinical centers. Plasma peptidomics and multi-omics approaches are expanding the toolkit, with evidence showing utility from prediabetes risk stratification to transplant monitoring. The technology is real and the biological signals are robust, but the path from validated biomarker to standard clinical practice requires larger diverse validation cohorts, demonstrated cost-effectiveness, and proof that peptidomic testing leads to better patient outcomes, not just earlier diagnosis.