Urinary Vesicles: Are They Ready for Real-World Use?

Physicians have analyzed urine to make inferences about health for over six millennia. Sadly, this practice was not always supported by a robust evidence base, leading uroscopists in the late seventeenth century to be derided as “pisse-procrastinators.”¹ So how are we faring in the era of molecular medicine? Can we return to using urine as a source of biomarkers without inciting accusations of “uromancy?”

Urinary extracellular vesicles (uEVs) carry a multimolecular cargo derived from kidney cells and therefore, hold promise as a source of clinical and research biomarkers.^2,3 However, two significant roadblocks must be passed before uEV-derived biomarkers can find applications in routine clinical practice: (1) lack of consensus about the best method of uEV quantification and (2) uncertainty regarding how to normalize uEV content.

International consensus guidelines have been enormously helpful in garnering opinion on these issues but leave open questions, in particular regarding the normalization of uEV-derived assay results.⁴ Normalization is required for any biomarker measured in spot urine samples because urine flow rate varies so widely. A widespread approach is to divide the concentration of the substance of interest (in this case, uEV particle number) by the concentration of creatinine: a reference substance chosen because it is easy to measure and is excreted at a uniform rate in most circumstances.^5,6

In this issue of JASN, Blijdorp et al. ⁷ conduct a comprehensive comparison of uEV quantification methods and examine the performance of urine creatinine concentration as a normalizer. They tested three methods used to quantify particle number in uEV preparations: nanoparticle tracking analysis (NTA), a plate-based fluorescence immunoassay (CD9 time-resolved fluorescence immunoassay), and a novel dye-based microscopy assay (EVQuant).

Their primary observation was that these three methods gave different results. Although this is not surprising, Blijdorp et al. ⁷ also provide insight into the mechanism underlying this discrepancy. In NTA, uEVs are detected in a video recording as particles moving by Brownian motion in solution. In EVQuant, uEVs are detected by confocal microscopy as fluorescent particles immobilized in a gel.^8(preprint) The lower limits of detection are around 70 and 35 nm, respectively—thresholds that lie well within the range of the uEV size distribution. This relative insensitivity accounts for the lower uEV abundance reported by NTA assays.

We cannot apply a simple “conversion factor” to equate results obtained by different methods because the magnitude of this discrepancy varied between samples, being less pronounced in dilute urine. This phenomenon was explained by a shift to larger uEV particle sizes in dilute urine. The authors probed the underlying mechanism for this shift in a series of well-designed in vivo and ex vivo experiments, concluding that there are contributions from osmotic uEV swelling and from uromodulin interfering with the NTA assay. This serves as a reminder that all of these methods of uEV quantification report a proxy measure—particle number—and are inherently vulnerable to confounding by a variety of factors relating to the experimental subject, urine collection, and sample processing. Although not explored in their paper, Blijdorp et al. ⁷ point out that proteinuria has been shown to affect NTA assay results. This important work adds to our existing understanding of the relative merits of different approaches to uEV quantification (Table 1).

Table 1.

Comparison of methods of uEV particle quantification

Quantification Method	Notes
NTAa	Use of specific label optional
	Sensitivity is size dependent; confounded by urine concentration7
	Requires video capture apparatus and processing software
Time-resolved fluorescence immunoassaya	Requires specific label
	Gives relative (not absolute) quantification
EVQuanta	Use of specific label optional
	Requires spinning disk confocal microscope
Transmission electron microscopy	Requires electron microscope
	Laborious; not high throughput
Total protein abundance	Confounded by copurification of nonextracellular-vesicle proteins4
Total lipid abundance	Dye-based methods insensitive4
	Spectroscopy methods require expensive equipment4
Total RNA abundance	Confounded by copurification of RNA-binding proteins4
Marker protein abundance	Uncertainty over appropriate marker proteins

^aMethods studied by Blijdorp et al.⁷

Blijdorp et al. ⁷ also addressed a practical issue that has puzzled investigators in this field for some time. In kidney research, most cell-specific antibodies are directed against intracellular epitopes. This should—in theory—limit their utility in detecting uEVs in which membrane orientation is thought to mirror that of the cell (i.e., “outside out”). Nevertheless, antibodies directed against intracellular epitopes appear to be able to bind at least a subpopulation of uEVs.⁹ Blijdorp et al. ⁷ showed that 0.01% SDS can enhance detection of intracellular epitopes by immunofluorescence antibodies, presumably by partially permeabilizing the vesicle membrane.

With regard to normalization, Blijdorp et al. ⁷ argue that urinary creatinine (uCr) concentration makes an attractive normalizer because its excretion is largely unaffected by water intake and because it is highly correlated with uEV particle number. These properties mean that uCr is indeed likely to be a good normalizer for comparing different samples within the same individual subjected to physiologic perturbations. However, as when uCr is used to normalize protein excretion in a protein-creatinine ratio, one must bear in mind the dietary, pharmacologic, and physiologic determinants of creatinine excretion, particularly when drawing comparisons between individuals with different muscle bulk or in the context of AKI.

This study gave a broad insight into the optimal normalization strategy because it included healthy controls, patients with polycystic kidney disease, and subjects of a physiologic intervention (water loading). In all of these groups, uCr correlated well with uEV excretion. The correlation was less in men than in women; the authors speculate that variable excretion of prostate-derived extracellular vesicles might account for this, and it would be interesting to test that hypothesis.

Parameters other than creatinine have been used to normalize uEV abundance; they each have strengths and weaknesses, and therefore, the “best” normalizer for any given study will depend on the aims and design of that study (Table 2).

Table 2.

Comparison of methods of uEV particle normalization

Normalization Method	Notes
Urine flow (i.e., calculate excretion rate)	Requires timed urine collection
	Uncertainty over how uEV excretion is regulated; effects of urine flow rate, etc.
Creatinine concentration	Excretion influenced by muscle mass, acute changes in GFR, drugs affecting tubular transport
	[Creatinine] correlated with uEV particle number7
Osmolality	Affected by both water and osmolar load7
Uromodulin (THP)	Abundance correlates with uEV particle number11
	Wide intersubject variation in excretion
uEV marker proteins	CD9/CD63 only expressed in distal nephron segments7

THP, Tamm-Horsfall protein.

uEVs are now well established as research tools. There are many reported associations of physiologic and disease phenotypes with uEV-derived biomarkers.¹⁰ The trove of molecular treasures in their payload undoubtedly provides a source of useful biologic information, which we are slowly learning to unlock. However, uEV-derived biomarkers have yet to find a place in routine clinical practice, largely because we lack analytic methods that are rapid, cheap, and reliable. For example, none of the quantification methods tested by Blijdorp et al. ⁷ are suitable for rapid, high-throughput testing and use in resource-poor settings. This is arguably now the main challenge facing this field.

Finding simple ways of normalizing results to facilitate meaningful intra- and intersubject comparisons will also be important: hence, the attractiveness of using uCr. The work published in this issue suggests that using uCr to normalize uEV abundance in spot urine samples is a reasonable approach in many settings.

Disclosures

J.W. Dear reports consultancy agreements with CymaBay Therapeutics but has not actually performed any work or received any payment; research funding as chief investigator on an investigator-led pharma-funded phase 1 trial POP trial (PP100-01, Calmangafodipir, for Overdose of Paracetamol Trial) funded by PledPharma AB; and scientific advisor or membership with the European Medicines Agency Expert Advisory Committees on Paracetamol and as an scientific advisory board member for EU IMI (European Union Innovative Medicines Initiative) TransBioLine Consortium and Vetsina Diagnostics. R.W. Hunter reports research funding from Wellcome Trust.

Funding

R.W. Hunter is supported by a fellowship from the Wellcome Trust (209562/Z/17/Z).