Accuracy of glomerular filtration rate (GFR) estimation using creatinine and cystatin C and albuminuria for monitoring disease progression in patients with stage 3 chronic kidney disease: a prospective longitudinal study in a multiethnic population

Design. 1) A prospective longitudinal cohort study to estimate accuracy in eGFR and changes in eGFR. 1300 people will be followed for 3 years with reference (measured) GFR and test (estimated GFR [eGFR] and ACR) measurements at baseline and 3 years. Test measurements will also be undertaken every 6 months, as recommended in NICE guidance. People will be recruited over 6 centres and will include a representative sample of Asians and African-Caribbeans, and people with diabetes and proteinuria. 2) A sub-study of patterns of disease progression of 300 people (100 each of Caucasian, Asian and African-Caribbean origin; in each case containing subjects at high [proteinuria/diabetes] and low risk of kidney disease progression) Additional reference GFR measurements will be undertaken after 1 and 2 years to enable a model of disease progression to be built. 3) A study of biological variability to establish reference change values (RCVs) for reference GFR measurement and its estimated surrogates. 4) A modelling study of the performance of monitoring strategies on detecting progression, utilising estimates of accuracy, patterns of disease progression and estimates of measurement error from studies 1), 2) and 3). 5) A model based cost-effectiveness analysis of alternative monitoring strategies. Setting. Primary and secondary care. Target population. Adults with stage 3 CKD (GFR 30-59 mL/min/1.73 m2) proportionally enriched to include people more likely to have progressive kidney disease (i.e. those with proteinuria and/or diabetes) and including Asians and African-Caribbeans. Health technologies being assessed. Estimated GFR using simplified ID-MS traceable version of the MDRD equation and three Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equations, using either creatinine or cystatin C or a combination of both, in addition to urinary ACR. Creatinine and cystatin C will be measured using methods traceable to the international reference method. We will assess the performance of monitoring rules utilising these tests at various time intervals. Measurement of cost and outcomes. Iohexol measured GFR will be the reference measure of GFR against which each GFR estimating equation will be compared. The performance of the GFR estimating equations will be evaluated by assessing bias, precision and accuracy (expressed as P30, the measure of error variability commonly used in this context and being the percentage of eGFR values within 30% of ‘true’ GFR). We will also consider error in the change in eGFR, across multiple time points. Large error will be accepted as greater than 3 mL/min/1.73 m2/year or a greater than 5%/year difference in slope between measured and eGFR. Data will be modelled as a linear function of time utilising all available (maximum 7) eGFR time points compared to the difference between final and baseline reference GFR values. The percentage of participants demonstrating large error with the respective GFR estimating equations will be compared. Predictive value will be compared by dividing subjects based on whether or not they show progressive decline in kidney function over the 3 years and the sensitivity and specificity of each of the eGFRs to identify progressive disease will be tested using receiver-operator curve analysis. ACR (in addition to the various eGFRs) will be tested as a predictor of progression. The rate of decline in reference GFR measured every year in a substudy will be modelled over time using a longitudinal linear or nonlinear (exponential decline) random coefficients regression model to estimate disease progression and error. The health economic component will use primary data collection to develop a comprehensive cost database for each diagnostic approach enabling cost-effectiveness modelling of the optimal strategy. Sample size. 1300 (see ‘additional information’), including at least 165 African-Caribbean and 165 South Asians. Sample size justification: We estimated the minimal clinically significant difference between P30 values to be 5% (e.g. a change from 81% to 86%). A sample size of 1300 (including allowance for a dropout rate of 15%-20%) provides 90% power to detect a clinically significant difference in P30 allowing identification of the most accurate GFR estimating equation. Confidence intervals (95%) on estimates of P30 in the ethnic subgroups will be less than 10 percentage points wide. Project timetable (including recruitment rate). The study will commence on 1st April 2013. 0-6 months: regulatory approvals, staff appointments, staff training. 0-12 months: sub-study of biological variation. 7-24 months: Longitudinal study patient recruitment and baseline reference GFR and associated tests (220 participants per centre with research nurses undertaking 1-2 baseline reference GFRs at each site per working day and collecting contemporaneous samples for eGFR and ACR). 13-54 months: 6 monthly blood and urine tests. 43-60 months: 3 year/final follow-up tests. 55-60 months; health economics and model building. 61-66 months: study closure, detailed health economic analysis, report writing, dissemination. Additional points. GFR equations: the MDRD equation is used routinely to estimate GFR but is negatively biased and imprecise. A revised equation, the CKD-EPIcreatinine is thought to partially address these issues [1, 2]. Observational data suggests cystatin C and ACR are superior to creatinine-based eGFR in identifying people most likely to have progressive or clinically significant disease [3] but this has not been tested against a reference measure of GFR. Newer GFR-estimating equations are available which include cystatin C, either alone or in combination with creatinine [4]. Standard clearance of inulin, including urine collection, remains the "gold-standard" method for GFR measurement but few studies use this. We have chosen iohexol clearance as the reference measure of GFR for our study because it is equivalent to inulin clearance, is non-radioactive, can be measured accurately and precisely, and is relatively cheap [5]. NICE guidance recommends that eGFR is checked every 6 months in stage 3 CKD [6]. [1] Levey, Ann Int Med 2009;150:604-12 [2] Earley, Ann Int Med 2012;156:785-95 [3] Peralta, JAMA 2011;305:1545-52 [4] Inker, NEJM 2012;367:20-9 [5] Schwartz, Clin J Am Soc Nephrol 2009;4:1832-43 [6] NICE, 2008, http://publications.nice.org.uk/chronic-kidney-disease-cg73