Design and development of a urinary cell pellet mRNA PCR-based assay for progressive kidney disease: Nephro-Dx

Vinod Kumar, Gabriel Barsotti, Zhengzi Yi, Zeguo Sun, Anand Reghuvaran, E. M. Tanvir, John Pell, Hongmei Shi, Sudhir Perincheri, Melissa Shaw, Candice Kent, Daris Javed, P. H. dos S. Leite, Deepa Jayaram, Jeffrey M. Turner, Kristin Meliambro, Randy L. Luciano, John Cijiang He, Dennis G. Moledina, Perry Wilson, Weijia Zhang, Madhav C. MenonmedRxiv 2025 · DOI

TL;DR

Researchers have developed Nephro-Dx, a non-invasive, 10-gene urinary mRNA assay that acts as a "liquid biopsy" for Chronic Kidney Disease (CKD). By measuring specific genetic markers shed into the urine, the test can differentiate diseased kidneys from healthy ones, correlate with specific tissue damage seen on a physical biopsy, and predict future kidney function decline.

What Problem Does This Paper Address?

Chronic Kidney Disease (CKD) is a massive public health burden, but our current diagnostic tools are surprisingly blunt. We typically rely on the eGFR (estimated glomerular filtration rate), which measures how well the kidneys filter blood, and proteinuria (protein in the urine). While useful, these are "global" markers—they tell you the kidney is struggling, but not why or where the damage is happening.

The gold standard for specific diagnosis is a kidney biopsy. However, biopsies are invasive, carry risks like bleeding, and are nearly impossible to perform repeatedly to monitor disease progression. There is a critical "knowledge gap" for a repeatable, non-invasive test that can provide the same level of cell-specific detail as a biopsy.

Methods

The authors developed a multiplexed qPCR-based assay called Nephro-Dx. The core innovation lies in the parsimonious selection of 10 target genes. Rather than a "shotgun" approach, they used existing single-cell RNA sequencing (scRNA-seq) data to identify genes that are highly specific to certain kidney compartments or pathological processes:

  • Glomerular (Filter) Health: NPHS1 and NPHS2 (Podocyte markers).
  • Tubular Injury: KIM-1 (acute injury) and AQP2 (collecting duct health).
  • Progression & Fibrosis: TGFB1 (fibrogenesis), VCAM1 (failed repair), and SHROOM3 (progressive injury).
  • Inflammation: CD68 (macrophage marker).
  • Controls: 18s rRNA (total cell content) and UPK1A (urothelial shedding).

The team optimized the extraction of mRNA from urinary cell pellets and used an automated digital pathology pipeline to quantify features from actual kidney biopsies, allowing them to compare the "urine signal" directly against the "tissue reality."

Datasets

The study utilized a real-world cohort from the Yale Biorepository:

  • Cases: 48 patients with diverse, biopsy-confirmed kidney diseases (e.g., FSGS, Diabetic Kidney Disease, Lupus Nephritis).
  • Controls: 19-20 healthy individuals, including living kidney donors and volunteers.
  • Validation: The gene selection was cross-referenced against published scRNA-seq datasets from patients with FSGS and AKI to ensure the markers were detectable in urine.

Key Insights and Ideas

  • Urine as a Transcriptomic Mirror: The study confirms that urinary shedding of specific cells is proportional to the injury of those cells within the kidney. The mRNA in the urine isn't just "noise"; it's a high-fidelity message reflecting the state of the nephron.
  • Beyond Global Metrics: Interestingly, the urinary mRNA levels did not correlate with eGFR on the day of the biopsy. This suggests that Nephro-Dx provides orthogonal information—it sees the cellular injury that blood tests might miss until the damage is much further along.
  • Digital Pathology Synergy: By using AI-driven digital pathology to score biopsies, the researchers removed human inter-observer bias, providing a more rigorous "ground truth" to validate their urine test.
  • Predictive Power: The assay doesn't just look backward at current damage; it looks forward. Higher levels of markers like VCAM1 and SHROOM3 were strong predictors of which patients would lose >20% of their kidney function over the following six months.

Key Results

  • Diagnostic Accuracy: All 10 genes were significantly elevated in CKD cases compared to controls. Logistic regression models showed excellent discrimination, with Area Under the Curve (AUC) values ranging from 0.71 to 0.88.
  • Compartment Specificity:
    • Urinary NPHS1 correlated with the number of normal glomeruli (R=0.47).
  • Urinary VCAM1 strongly correlated with the percentage of atrophic (dysfunctional) tubules (R=0.41) and interstitial fibrosis.
  • Prognostic Value: Patients who experienced a significant eGFR decline (>20%) within 6 months had significantly higher baseline urinary levels of NPHS2, AQP2, KIM1, CD68, SHROOM3, and VCAM1.

Limitations and Future Work

The authors acknowledge several hurdles:

  • Sample Size: This was a pilot study with 48 cases. Larger, multi-center cohorts are needed to validate the predictive models.
  • Technical Exclusions: About 11% of patient samples had to be excluded due to low RNA quality or "pink pellets" (urate crystals that interfere with processing), highlighting a need for even more robust stabilization methods.
  • Customization: While the 10-gene panel is a strong start, the authors suggest the platform could be expanded to include specific markers for T-cell or B-cell infiltration to further refine the "liquid biopsy" for specific diseases like transplant rejection.

Summary updated

Abstract

Progressive chronic kidney disease (CKD) is a major source of public health spending. Current non-invasive tests estimate CKD but provide a minimal understanding of cell- or compartment-specific injury. The gold standard for CKD diagnosis is a kidney biopsy, which affords risks and is impractical to repeat multiple times. Hence, repeatable, non-invasive tests to estimate pathologic kidney injury for diagnoses, prognosis and follow-up of CKD represent a knowledge gap. We hypothesized that urinary shedding of specific cells is proportional to injury of those cells on biopsy, and that tracking cell-specific urine mRNA will correlate with ongoing injury. Informed by apriori biopsy and urine single cell RNA studies, we developed a targeted 10-gene urine mRNA assay to estimate kidney injury non-invasively (Nephro-Dx). In a pilot study of 48 patients with diverse kidney pathology on biopsy and 20 controls, we confirmed our assay's utility in differentiating any kidney disease from controls. Within biopsied cases, we confirmed correlations of cell-specific urinary gene expression with corresponding compartment injury on biopsy using a validated quantitative digital pathology platform. We show that the gene signatures including individual genes associate with subsequent loss of kidney function within our cases providing an advantage over existing non-invasive tests. Our parsimonious set of gene signatures in Nephro-Dx shows advantages in early diagnosis, monitoring, and prognosis to impact this public health problem.
BibTeX
@misc{kumar2025design,
  author = {Vinod Kumar and Gabriel Barsotti and Zhengzi Yi and Zeguo Sun and Anand Reghuvaran and E. M. Tanvir and John Pell and Hongmei Shi and Sudhir Perincheri and Melissa Shaw and Candice Kent and Daris Javed and P. H. dos S. Leite and Deepa Jayaram and Jeffrey M. Turner and Kristin Meliambro and Randy L. Luciano and John Cijiang He and Dennis G. Moledina and Perry Wilson and Weijia Zhang and Madhav C. Menon},
  title = {{Design and development of a urinary cell pellet mRNA PCR-based assay for progressive kidney disease: Nephro-Dx}},
  year = {2025},
  doi = {10.1101/2025.10.31.25338655},
  howpublished = {medRxiv},
}
Figure