Proteomics: Urinary HPGD Signals Liver Cancer Relapse Early
After curative resection of hepatocellular carcinoma (HCC), surveillance relies on serum AFP, DCP, and periodic MRI, all of which tend to catch relapse late. Could urine do better? Xing and colleagues tracked 378 patients with urinary proteomics, pairing data-independent acquisition (DIA) discovery with parallel reaction monitoring (PRM) verification, and built a score that flagged recurrence a median of 4.1 months before imaging.
Key Takeaways
Urinary proteomics can track liver cancer tumor burden after surgery, complementing serum AFP testing and imaging.
Urine HPGD combined with serum AFP, DCP, and GGT predicted HCC recurrence months before it appeared on MRI.
The single-center model did not account for kidney function, so multicenter validation must precede clinical use.
Tracking Tumor Burden With Urinary Proteomics
The design hinges on an unusual group: 26 patients who had a resection, relapsed, and returned for a second operation. Urine from five fixed time points (130 samples) let each patient serve as their own control. Samples were run by DIA mass spectrometry and normalized after excluding the 20 most abundant proteins, such as uromodulin, a choice that probably matters more than the authors emphasize.
Key Findings
Urine beat serum on depth: about 4,100 proteins per urine sample, against roughly 300 per serum sample.
Two surgeries, one shared signal: 217 proteins shifted after both resections, yielding six leading candidates for PRM.
HPGD held up in validation: in 46 patients it fell after each resection (p < 0.001) and climbed before relapse, while AFP and DCP did not rebound.
Earlier warning from a four-marker score: in 306 postoperative patients, HPGD plus AFP, DCP, and GGT reached an AUC of 0.86, against 0.73 for AFP alone.
Figure 1. Study design. Discovery (n = 26) and validation (n = 46) patients gave urine and serum at five time points across two surgeries. DIA-MS candidates were confirmed by PRM-MS and combined with serum AFP, PIVKA-II, and GGT into a tumor burden monitoring model, then tested for early recurrence prediction in a 306-patient exploration cohort. Adapted from Xing et al. (2026), Molecular & Cellular Proteomics.
Why HPGD and Not the Familiar Liver Markers
HPGD encodes 15-hydroxyprostaglandin dehydrogenase, a prostaglandin E2-degrading enzyme and reported tumor suppressor. Mechanism was not tested. More telling is what failed: GPC3, GP73, and HSP70 appeared in urine but did not track tumor burden over follow-up. Peptide quantification used heavy isotope-labeled standards, the PRM approach Peterson and colleagues (2012) introduced on quadrupole-Orbitrap instruments.
Clinical Performance and Its Limits
In multivariate Cox models, the combined score independently predicted recurrence-free survival (HR 2.81) and overall survival (HR 2.49). One number deserves scrutiny. Adding HPGD to a serum-only AFP, DCP, and GGT model raised the validation AUC from 0.82 to 0.86, a real but modest gain, and false negatives clustered in patients with cirrhosis or hepatitis who relapsed early.
In Practice: From DIA Discovery to PRM Assays
Only HPGD, of six DIA candidates, came through PRM verification cleanly enough to enter the final model, which is why that step shouldn't be skipped. We run the same two-stage design in our lab at Dalton for urinary and plasma proteomics studies, and locking normalization before validation samples arrive matters more than the instrument. For multi-omics biomarker discovery in serially sampled cohorts, we'd add creatinine or eGFR covariates from day one.
Frequently Asked Questions
What is urinary proteomics used for in cancer research?
Urinary proteomics is large-scale protein profiling of urine, usually by LC-MS, to find or monitor tumor markers without a blood draw. Urine carries fewer dominant proteins than plasma, so faint signals are easier to detect.
Can a urine test detect liver cancer recurrence before an MRI?
In this study, urinary HPGD combined with serum AFP, DCP, and GGT rose a median of 4.1 months before imaging confirmed recurrence. The finding comes from one center and needs multicenter validation.
How many samples does a proteomics biomarker study need?
It depends on the design. Many programs run untargeted discovery in a small, well-phenotyped set, then targeted verification in a larger cohort; here, 130 discovery samples led to PRM verification in 230 and clinical testing in 612.
Conclusion
It's now reasonable to treat urine as a serious matrix for postoperative HCC surveillance, provided collection is scheduled and normalization is careful. What isn't proven is whether HPGD holds up across hospitals, liver disease etiologies, and levels of kidney function. Study designers should copy the sampling plan, which did much of the statistical work here.
Related Reading
See how we run these analyses in one lab: Dalton's multi-omics CRO services.
Citation
Xing, X., Hu, E., Song, L., Zheng, Z., Cai, L., Ouyang, J., Li, J., Pan, X., Liu, M., Wang, Y., & Liu, X. (2026). Urinary proteomics enables noninvasive and longitudinal monitoring of tumor burden dynamics and early recurrence in hepatocellular carcinoma. Molecular & Cellular Proteomics, 25(9), 101635. https://doi.org/10.1016/j.mcpro.2026.101635
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.