3-2
Modification of the MDRD Study equation and development of a new equation for Japanese patients with CKD.
Masaru Horio, Enyu Imai, Kosaku Nitta, Kunihiro Yamagata, Kunitoshi Iseki, Yusuke Tsukamoto, Sadayoshi Ito, Hirofumi Makino, Akira Hishida and Seiichi Matsuo for the Japan Chronic Kidney Disease Initiatives

Background:
Recently, the re-expressed MDRD Study equation with a standardized serum creatinine was reported (Ann Intern Med 2006, 145:247). In the equation, creatinine was measured by the enzymatic method using IDMS traceable reference material. In Japan, most laboratories are using the enzymatic method to measure creatinine. Application of MDRD equation without calibration of serum creatinine is quite useful. The MDRD equation may accurately estimate the white and black population but it is obscure whether the MDRD equation is appropriate for Japanese population. The purpose of this study was to validate the MDRD equation by conferring the coefficient and to develop a new equation for Japanese patients with CKD.

Methods:
Glomerular filtration rate was measured by the gold standard, inulin renal clearance. Inulin clearance of 116 and 131 Japanese with CKD patients were obtained from the clinical trial of the safety of inulin. S (samples were measured at central laboratory) and Tokyo Women's University hospital, respectively. Serum creatinine values were measured in the two laboratories by the enzymatic method. We validated the difference of the serum creatinine measurements between laboratories using a calibration panel composed of 42 serum samples, whose creatinine values were assigned at the Cleveland Clinic. 214 patients with Cin <90 mL/min/1.73 m2 were used for modification of the MDRD Study equation and development of a new equation. We calculated a suitable value of the multiplication factor to minimize the root mean squared error. New equation was developed by using multiple linear regression. The regression was performed on log-transformed data (GFR, age, serum creatinine, gender). We are now accumulating the serum samples and inulin clearance data of CKD patients from 82 hospitals in Japan. Inulin and serum creatinine are measured in one laboratory, the central laboratory. We are planning to use these data for validation and development of a more accurate Japanese GFR equation.

Results:
The coefficient of variation (CV) for creatinine values from the three laboratories (Cleveland Clinic, central laboratory, Tokyo Women's University hospital) was 1.59%. Bland and Altman plot for difference of serum creatinine value indicated a small proportional error (Fig.1). There was only 2-3% difference between laboratories in Japan and Cleveland Clinic Laboratory indicating that inter-laboratory difference in creatinine measurement by the enzymatic method was very small and the value was allowable for use in calculation of the MDRD Study equation.
The MDRD Study equation overestimated over the entire range of GFR and did not perform well for estimating GFR in Japanese patients with CKD (Fig.2). A Japanese coefficient 0.741 and a new equation by multiple linear regression improved the accuracy of the MDRD Study equation in subjects with GFR less than 90 mL/min/1.73m2. There was no significant difference in accuracy and bias between the modified MDRD Study equation by Japanese coefficient and the new equation.

Conclusion:
Use of either MDRD Study equation with the Japanese coefficient or the new equation for estimating GFR significantly improved the estimation of GFR when less than 90 mL/min/1.73m2. In clinical practice, the use of either of equations will be encouraged for assessment of CKD in Japanese patients.