Multi-Omics: Urine Protein Panel Flags Kidney Cancer
Most kidney cancers are found by accident. Clear cell renal cell carcinoma (ccRCC) makes up roughly 80 percent of renal malignancy, yet has no approved liquid biopsy, so diagnosis waits on an abdominal scan ordered for something else. A urinary multi-omics study in EMBO Molecular Medicine went after that gap, running proteomics, lipidomics, and untargeted metabolomics on the same urine from 22 ccRCC patients, 9 with non-clear-cell tumors, and 12 matched controls, then distilling the protein hits into a three-marker assay they call the UrineScore.
Key Takeaways
Urine suits kidney cancer discovery because tumors arising in the proximal tubule shed protein straight into the compartment being excreted.
A three-protein urinary panel (SAA1, haptoglobin, LCN15) separated ccRCC from controls at AUROC 0.96 in discovery and 0.95 in validation.
Multi-omics discovery does not obligate a multi-omics assay. Lipid features added almost nothing once the three proteins were in the model.
With 22 and 10 ccRCC cases across two cohorts, this is a proof of concept, not a clinic-ready test.
What Multi-Omics Profiling Found in ccRCC Urine
Proteins elevated in ccRCC urine supernatant enriched for fatty acid transport and high-density lipoprotein particle remodeling; the sediment proteome lit up instead for mitochondrial respiration. Untargeted metabolomics ran the other way, with 36 annotated metabolites depressed in patient urine across glutamate, glyoxylate, glutathione, and carnitine pathways. Lipid profiling closed the loop: 3 of 9 detected lipid classes were higher in ccRCC. One workflow detail carried real weight, since the two-step acetone then chloroform:methanol precipitation largely set how many peptides survived to be identified.
Key Findings
Lipids rise while metabolite pools fall: coenzyme Q10, phosphatidylcholines, and phosphatidylethanolamines were all elevated, with PC 38:3 and PC 38:6 reaching AUROCs of 0.91 and 0.90 alone.
Three proteins cleared a deliberately strict filter: SAA1, haptoglobin, and LCN15 each showed log2 fold change above 2 versus controls, stayed flat in non-clear-cell tumors, and were not sex-specific.
The composite score validated externally: UrineScore reached AUROC 0.96 in Vienna and 0.95 in a Madrid biobank cohort, with negative predictive value above 99 percent at 1 percent prevalence.
Integration helped the biology more than the diagnosis: a seven-feature protein-plus-lipid model scored AUC 0.883 against 0.879 for proteins alone (DeLong P = 0.957).
Figure 1. Multi-omics profiling of ccRCC patient urine points to metabolic dysregulation. Panel A schematizes how urogenital analytes collect before voiding. Panels B and C give gene ontology enrichment for proteins upregulated in ccRCC urine supernatant and sediment. Panel D shows pathway enrichment of the downregulated metabolites against KEGG and SMPDB. Panel E is a fold-change heatmap across lipid families, Panel F plots the three families that differed significantly, and Panel G summarizes the proposed metabolic network. Adapted from Jonsson et al. (2026), EMBO Molecular Medicine.
From Discovery Proteomics to a Targeted PRM Readout
Bulk discovery proteomics is the wrong instrument for screening: long gradients, heavy data processing, a per-sample cost no health system absorbs. The three candidates were moved onto parallel reaction monitoring, the approach described by Peterson and colleagues (2012), keeping only peptides with clean elution profiles. Each marker was scored as summed peptide area over that of three normalization proteins, UMOD, KLK1, and APOD, chosen for appearing in every sample with low across-group variability.
Where the UrineScore Holds and Where It Slips
Scoring is deliberately blunt: one point per protein exceeding the upper bound of the control 95 percent confidence interval, an integer from 0 to 3. That crudeness is probably why it transferred to a second hospital without retuning. Separation from other renal carcinoma subtypes was weaker, at AUROC 0.78, and that is the number that matters clinically, since a patient with a mass on imaging needs to know which tumor they have. Urine has been productive for other kidney phenotypes too, including work tying TRIM28 to acute kidney injury signals.
One Sample, Three Layers, One Injection
Running each analyte class as its own workflow protected per-class sensitivity, but it left an integrated model combining seven features across 34 samples, where it couldn't show an advantage. Pulling proteins, lipids, and polar metabolites from a single injection, which the Omni-MS platform at Dalton was built around, puts every feature behind the same aliquot and drops the split-sample tradeoff that quietly weakens integrated models in small cohorts. That question recurs in multi-omics profiling of low-volume biofluids like urine, and it is one of the design decisions covered in our guide to multi-omics biomarker discovery.
Frequently Asked Questions
What is multi-omics analysis used for?
Multi-omics analysis measures two or more molecular layers, such as proteins, lipids, or metabolites, from the same samples. Teams use it to find biomarkers or test whether a signal in one layer is corroborated in another. Cross-omics analysis of urine here showed lipid accumulation and metabolite depletion converging on one picture.
Can a urine test detect clear cell renal cell carcinoma?
Not yet in routine care, though this study shows it is feasible. A three-protein score measured by targeted mass spectrometry separated patients from controls at 96 and 95 percent accuracy across two hospitals. Cohorts including benign renal masses are the next step.
How much sample does multi-omics biomarker discovery need?
Less than most groups assume. This work used standard voided urine with no special collection hardware, and targeted PRM assays need far less material than discovery runs. Volume binds mainly when each omics layer is run from its own aliquot.
Conclusion
Urinary SAA1, haptoglobin, and LCN15 look like genuine ccRCC signals, corroborated at the transcript level in TCGA and the protein level in CPTAC. What isn't established is performance in the population where such a test would be used: patients with incidental renal masses of uncertain type, not confirmed cancers against healthy volunteers. The transferable lesson is the handoff: discovery across three layers, then a lean targeted assay carrying only what earned its place.
Related Reading
See how we run these analyses in one lab: Dalton's multi-omics CRO services.
Citation
Jonsson, G., Oliveira, T., Hofmann, M., Lemberger, U., Stejskal, K., Krssakova, G., Sakic, I., Novatchkova, M., Mereiter, S., Grabmann, G., Kocher, T., Koglgruber, R., Kikic, Z., Camano Paez, S., Luna Sanchez, B., Diez Nicolas, V., Rechberger, G. N., Zullig, T., Hagelkruys, A., ... Penninger, J. M. (2026). Urinary multi-omics reveal non-invasive diagnostic biomarkers in clear cell renal cell carcinoma. EMBO Molecular Medicine, 18(9), 3649-3676. https://doi.org/10.1038/s44321-026-00498-2
Note
This blog post summarizes findings from the above-cited research. Figures are adapted from the original publication. For full details, please refer to the source article.
By Seungjun Yeo, CEO at Dalton Bioanalytics. Specializing in multi-omics mass spectrometry for drug discovery and biomarker research.