Cross-Disciplinary Biomarkers Research: Lessons Learned by the CKD Biomarkers Consortium

Abstract

Significant advances are needed to improve the diagnosis, prognosis, and management of persons with CKD. Discovery of new biomarkers and improvements in currently available biomarkers for CKD hold great promise to achieve these necessary advances. Interest in identification and evaluation of biomarkers for CKD has increased substantially over the past decade. In 2009, the National Institute of Diabetes and Digestive and Kidney Diseases established the CKD Biomarkers Consortium (http://www.ckdbiomarkersconsortium.org/), a multidisciplinary, collaborative study group located at over a dozen academic medical centers. The main objective of the consortium was to evaluate new biomarkers for purposes related to CKD in established prospective cohorts, including those enriched for CKD. During the first 5 years of the consortium, many insights into collaborative biomarker research were gained that may be useful to other investigators involved in biomarkers research. These lessons learned are outlined in this Special Feature and include a wide range of issues related to biospecimen collection, storage, and retrieval, and the internal and external quality assessment of laboratories that performed the assays. The authors propose that investigations involving biomarker discovery and validation are greatly enhanced by establishing and following explicit quality control metrics, including the use of blind replicate and proficiency samples, by carefully considering the conditions under which specimens are collected, handled, and stored, and by conducting pilot and feasibility studies when there are concerns about the condition of the specimens or the accuracy or reproducibility of the assays.

Keywords: CKD; biomarkers; epidemiology; outcomes; quality control; risk factors.

Figure 1.

Evaluation of sample labeling error. Comparison of urine creatinine concentration results in urine samples from one cohort assayed before 2009 and again in 2013 (left panel). Comparison of urine creatinine concentrations in samples from the same cohort from assays performed by a Biomarkers Consortium laboratory in 2013 to those from assays performed in the same laboratory in 2012 (right panel). Data points falling near the line of identity are shown in the rectangular box.

Figure 2.

Data templates on two Luminex instruments. The left panel shows sample order assigned horizontally and the right panel shows sample order assigned vertically.

Figure 3.

Scatter plot of urine ACR measured at the time of sample collection (old ACR) and after long-term frozen storage (new ACR). The line of identity is shown. ACR, albumin/creatinine ratio.

Figure 4.

Evaluation of sample desiccation. Scatter plots of urine albumin concentration measured at the time of sample collection (Old UALB) and after long-term frozen storage (New UALB) before (left panel) and after (right panel) correcting for water loss in the sample during storage. Water loss was estimated by computing the ratio of urine creatinine concentration measured at the time of sample collection and the concentration measured after long-term storage. The urine albumin concentration measured after storage was then multiplied by this ratio. The lines of identity are shown. UALB, urine albumin.

Figure 5.

Evaluation of incomplete mixing. Scatter plots of urine NAG concentration measured in the parent sample (NAG 1, vertical axis) or blind replicate sample (NAG 2, horizontal axis) in samples aliquoted in two large batches (batch 1, left panel; batch 2, right panel). The Pearson correlation and Deming regression for each batch are shown. The lines of identity are shown. 95% CI, 95% confidence interval; NAG, N-acetyl-β-d-glucosaminidase.