SAN DIEGO, California — Based on distinct urinary proteomic patterns, a chronic kidney disease (CKD) risk classifier could predict microalbuminuria 4 years before it develops in patients with type 2 diabetes. Speaking during a poster session here at Kidney Week 2012: American Society of Nephrology 45th Annual Meeting, Morten Lindhardt, MD, from the Steno Diabetes Center in Gentofte, Denmark, said such a risk classifier may allow earlier intervention for at-risk patients.
The CKD risk classifier ("CKD273") is based on a specific pattern of 273 urinary peptides generated by capillary electrophoresis–mass spectrometry from previous cross-sectional case-control studies. A Cox regression model has been built to predict the progression of albuminuria.
The present study was conducted post hoc on 740 patients with type 2 diabetes and normoalbuminuria at baseline in the Effect of Candesartan on Progression and Regression of Retinopathy in Type 2 Diabetes Trial (DIRECT-Protect 2), a randomized controlled clinical trial of candesartan for slowing the progression of retinopathy.
The purpose of the present study was to confirm that urinary proteomics could predict microalbuminuria using a large cohort of patients with diabetes and normoalbuminuria. The primary endpoint was the development of persistent microalbuminuria, defined as albumin excretion greater than 20 μg/min in 3 of 4 overnight urine samples. The study had a mean follow-up of 4.7 years.
At baseline, the patients had an average age was 57 years, an 8.9-year duration of diabetes, and blood pressure of 134/74 mm Hg; 67% were being treated for hypertension. Their median urinary albumin excretion rate (UAER) was 5 μg/min.
Baseline Risk Score Predicts Development of Microalbuminuria
"With this classifier, we were able to identify 10% of the patients [n = 74] being at high risk with the classifier, and of these patients, 26% [n = 19] of them developed the endpoint.... [As] compared to the low-risk patients identified with the classifier [n = 666], only 11% of these patients [n = 72] developed the endpoint," Dr. Lindhardt said. "So we were actually able from baseline urine samples to do the classifier and from that to stratify those as high risk or low risk."
The risk prediction was independent of treatment (candesartan vs placebo), age, sex, diabetes duration, or baseline values of systolic blood pressure, UAER, estimated glomerular filtration rate (eGFR), and HbA1c (adjusted Cox regression model hazard ratio [HR], 2.1; 95% confidence interval [CI], 1.18 - 3.65; P = .012). In an unadjusted model, the prediction of development of microalbuminuria was statistically significant for the placebo group (HR, 2.40; 95% CI, 1.21 - 4.80; P = .013) but not for the candesartan one (HR, 1.76; 95% CI, 0.57 - 3.82; P = .16).
By Kaplan-Meier analysis, more patients in the predicted high-risk group developed microalbuminuria sooner than those in the low-risk group for all time points past 1 year (P < .001).
Dr. Lindhardt concluded that CKD273 was an independent predictor of microalbuminuria in an unselected group of patients with type 2 diabetes who had normal albumin excretion at baseline. He noted that the prediction tool "overclassified" many patients (ie, those who did not develop microalbuminuria).
He said a limitation of the study is that many of the hypertensive patients were treated with angiotensin converting enzyme inhibitors or angiotensin receptor blockers, "and that of course interferes with the endpoint, so we have to do some censoring before we can safely assert how we actually interpret these data." He said the investigators already have enough patient data and a pool of more than 5000 proteins that can be detected, so the next step is not to gather more patient data but to refine the model. And a prospective trial is in the works.
Dr. Lindhardt and colleagues are also testing the tool and model on patients with type 1 diabetes.
Uwe Andag, PhD, a project leader at Evotek biotechnology company of Hamburg, Germany, who was viewing the poster, commented to Medscape Medical News, "I think that's kind of a new way to look into panels to identify biomarkers — not only a single biomarker but a huge panel, which might be more predictive."
John Arthur, MD, PhD, DCI Endowed Chair in Nephrology, professor of medicine, and the director of the Southeastern Kidney Disease Consortium and the Nephrology Proteomics Facility at the Medical University of South Carolina, in Charleston, suggested to Dr. Lindhardt that he investigate the slopes of the rise in eGFR and not only microalbuminuria.
"The problem with looking at albuminuria is that albuminuria is what we use as a predictor for loss of renal function [and] all the bad outcomes that are associated with diabetes, but it has a lot of problems itself," Dr. Arthur commented to Medscape Medical News. "So this is sort of predicting a predictor. But loss of GFR...would come, I think, closer to looking at what we truly want to find — what's going to predict who's going to lose renal function — hopefully, who's eventually going to end up on dialysis, and [we would] be able to treat those people."
Previous research has shown that CKD273 can predict microalbuminuria as much as 4 years in advance of its development. "You could potentially alter [patients'] treatments and hopefully prevent them from having these adverse outcomes. So I think that's a really tremendous possibility," Dr. Arthur said.
The study received support from AstraZeneca and Takeda. Dr. Lindhardt works at the Steno Diabetes Center, which is owned by Novo Nordisk. Dr. Andag and Dr. Arthur were not involved in the study and disclosed no relevant financial relationships.
Kidney Week 2012: American Society of Nephrology 45th Annual Meeting. Abstract TH-PO507. Presented November 1, 2012.
Medscape Medical News © 2012 WebMD, LLC
Send comments and news tips to email@example.com.
Cite this: Urinary Proteomics Test Predicts Diabetic Nephropathy - Medscape - Nov 04, 2012.