Can the Urinary Peptidome Outperform Creatinine and Albumin to Predict Renal Function Decline?

CKD represents an increasingly prevalent diagnosis affecting >10% of adults in the United States, Europe, Australia, and Asia.¹ The current system to classify CKD is a population–based diagnostic tool established by the National Kidney Foundation Kidney Disease Outcomes Quality Initiative and initially built on estimations of GFRs from measured serum creatinine. The insufficient predictive capacity of serum creatinine to establish likelihood of renal function decline has been augmented by urine albumin, and more markers have been proposed.^2–5 Given the array of clinical factors (notably diabetes and hypertension) and sociodemographic factors (e.g., aging, ethnicity, environment, and education) that can modify renal function and susceptibility to progressive loss of renal function,² it is not surprising that serum creatinine alone is insufficient to build a model that accurately allows for risk prediction and early diagnosis of renal function decline for the individual.⁶ Corresponding it should be no surprise that one could hypothesize that an array of markers are needed for a better classification of individuals at risk for early renal function decline in the presence or absence of microalbuminuria or proteinuria.

Proteomics encompasses a set of tools that can be used to help gain insight into human renal health and disease.⁷ At its heart, this tool set relies on the use of mass spectrometers (MSs) to study lower molecular weight proteins and peptides.⁸ By and large, the paradigm is one of expression or functional proteomics and follows a bottom-up approach, wherein proteins are digested ex vivo using a protease, such as trypsin, before MS analysis. Some laboratories have used an alternative approach termed peptidomics, in which the protein fragments/peptides that are studied are generated in vivo.^9–11 The peptidomics approach is a feature-based approach, where MS data on large numbers of clinical samples are collected and compared. It is not as easily applied to understanding the biology of proteins, which one might desire to discuss biomarkers of CKD.¹² However, these discriminatory MS features can be tabulated, be used as a diagnostic tool, and for purposes of understanding biology, undergo tandem MS experiments to gain information on biologic identity.

In this issue of JASN, Schanstra et al.¹³ articulate a model for the diagnosis and prediction of progressive CKD on the basis of the observations of large numbers (n=273) of specific urinary peptides: a peptide panel termed CKD273.^10,13 The work extends a peptidomic CKD marker panel previously developed by the group over the last decade in large, cross-sectional cohorts of patients. These peptides, fragments of extracellular matrix, sera, and urinary proteins were identified during a comparison of healthy control patients (n=379) and patients with various biopsy–proven renal diseases (n=230), including vasculitis, SLE, membranous GN, minimal change disease, IgA nephropathy, FSGS, and diabetic nephropathy. In an independent blinded cohort of 34 normal patients and 110 patients with renal disease, this panel of peptides allowed for collective identification of all normal patients and 94 of 110 patients with CKD.

Given that these markers discriminated individuals that had reduced renal function, the goal of this work is attractive in that it addresses the goal of CKD diagnosis and risk stratification for progression. The large multicenter effort is admirable in that Schanstra et al.¹³ evaluated a large cross-sectional cohort of patients (n=1990), including 83% with a mean baseline eGFR of 78±21 and 17% with a mean eGFR of 29±11. The CKD273 panel was better correlated to eGFR than urinary albumin (Rho=−0.437 versus Rho=−0.339). A subset of these patients (n=522) had five follow-up visits over 3 years with serum creatinine values. Again, the CKD273 panel was better correlated to percentage change in eGFR than percentage change in albuminuria (Rho=−0.395 versus Rho=−0.293). For a subset of these patients, albumin-to-creatinine measurements were available, and on evaluation, the CKD273 panel was better correlated to eGFR slope than to albumin-to-creatinine (Rho=−0.509 versus Rho=−0.401). First to assess identification of rapid progressors, the cohort was categorized into two groups using a demarcation of >−5%/year. The CKD273 panel identified 75% of the rapid progressors, whereas urinary albumin identified 65% of the rapid progressors. An analysis by the net reclassification index suggested that the urinary CKD273 panel allowed for an improved detection of 30% more rapid progressors (0.303±−0.065) over the use of urinary albumin alone.

In addition to discrimination of risk prediction, data from the work by Schanstra et al.¹³ also provide a framework for moving forward to hypotheses for a mechanism on the basis of the identities of protein fragments. Using a dichotomous cutoff approach (presence or absence of CKD), fragments of specific proteins could be associated with baseline GFR and progression of CKD. Regarding baseline GFR, fragments of serum proteins were negatively correlated to baseline eGFR (β2-microglobulin, apo A-I, α1-antitrypsin, and serum albumin), and fragments of extracellular matrix [collagen-α1(I) and -α1(III)] were positively correlated to baseline eGFR. Extracellular matrix protein fragments were correlated with early and moderate CKD, but fragments of apo A-I and β2-microglobulin emerged as correlated to moderate CKD. Regarding progression in CKD, fragments of two blood-derived proteins, α1-antitrypsin and albumin, were negatively correlated to change in eGFR.

These data do present some important and intriguing findings, but at present these results are insufficient to create a new framework for a paradigm of renal function decline.¹³ Clearly, these data suggest that there are biologic events that contribute to the degradation of proteins that occurs in concert with loss of renal function. What this means in a larger context is, at present, unknown. Several challenges and questions remain. How do these markers perform if one considers the aggregate work of Krolewski et al.¹⁴ and Macisaac and Jerums¹⁵ defining a complex diabetic landscape of variable rates of progression and remission as well as early progression all in the presence of normoalbuminuria, microalbuminruia, and/or proteinuria?^14–16 What is the performance of these CKD markers considering urinary peptide changes as a function of age?¹¹ We need high-quality longitudinal data with multiple measurements and substantial follow-up time to observe hard end points (ESRD or 50% eGFR reduction). How will these tools be assay validated for use in reference laboratories? The progression of CKD has been illustrated in a conceptual framework as an orderly transition between stages. The work of Schanstra et al.¹³ moves the field significantly ahead toward the important goal of early CKD diagnosis and risk stratification by using a urinary peptide marker panel, but it falls short of a conclusive answer given the constraints of the cross-sectional cohort used and the limited mechanistic information that this approach has yielded. When we are able to assemble longitudinal cohorts of patient samples within stages of normal GFR for such -omics analyses, we will see much greater progress on discovery of risk markers for CKD progression.

Disclosures

None.