It is 10 years since Levey et al. published the Modification of Diet in Renal Disease (MDRD) Study equation for estimating glomerular filtration rate (GFR). The MDRD equation has gained widespread acceptance, despite its limitations; in particular underestimation of GFR at higher values. In this recent publication, Levey et al. develop a new equation for estimated GFR: the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation.

Data from 10 studies that compared serum creatinine concentration with iothalamate clearance were pooled and used to develop the equation (n = 5504). The equation was subsequently externally validated against data pooled from 16 other studies (n = 3896). The CKD-EPI equation is a two-slope linear spline with gender-specific knots at 62 μmol/L in women and 80 μmol/L in men. The introduction of the knots allows the equation to have a reduced slope at lower creatinine concentrations, leading to higher, more accurate estimates of GFR in this range. The equation also uses gender, race and age as in the MDRD equation, but on the natural rather than log scale.

The new equation, while improved, still has significant limitations. The pooled population used to derive this equation is unlikely to reflect the general population. There were few patients older than 70 years and few participants were of racial minorities other than black, for whom a separate equation was derived.

The CKD-EPI equation appears to be more accurate than the MDRD equation, especially at higher GFRs. It yields a lower estimated prevalence of CKD than the MDRD equation (11.5% versus 13.1%), primarily because of a lower estimated prevalence of stage 3 disease. For example, it is estimated that the prevalence of CKD in the USA would be reduced by 3 million using the new equation. The authors suggest the CKD-EPI equation should replace the MDRD equation in general clinical use.

Boudville et al. have independently assessed creatinine-based equations in predicting GFR in orthotopic liver transplant patients. The incidence of CKD in this patient population is high and partially attributable to the use of nephrotoxic calcineurin inhibitors such as tacrolimus.

Analysis of creatinine in these patients overestimates kidney function due to poor muscle bulk, impaired creatinine biosynthesis and the well-described negative interference of hyperbilirubinaemia in Jaffe creatinine methods. Hence equations that use creatinine to estimate GFR are prone to error. Cystatin C has been advocated as a marker of GFR in many different populations. Here Boudville et al. compare ⁵¹Cr ethylenediaminetetraacetic acid (EDTA) clearance with both creatinine- (MDRD and Cockcroft-Gault) and cystatin C- (Hoek, Filler, Larsson and Le Bricon) based GFR equations in 41 liver transplant recipients.

The MDRD and Le Bricon equations showed the least bias and greatest precision when compared with ⁵¹Cr EDTA clearance. However, both equations were inaccurate with only 22% and 27%, respectively, estimating the GFR within 10% of the ⁵¹Cr EDTA GFR results. The accuracy of the equations was better at GFR results <60 mL/min/1.73 m²; however, it is the early detection of CKD in these patients which is important so that interventions can be initiated promptly. The degree of inaccuracy places doubt over the clinical utility of these equations in orthotopic liver transplant patients, and there was no evidence of benefit from the use of cystatin C in this population. It remains to be seen whether a two-slope linear spline equation may be of benefit in liver transplant patients. Interestingly, introducing organ transplantation as a variable in the CKD-EPI equation did not substantially improve its performance.