Establishment and evaluation of estimated glomerular filtration rate by serum cystatin C alone and in combination with serum creatinine in patients with chronic kidney disease

Objective To establish equations for estimating glomerular filtration rate (GFR) based on serum Cystatin C (CysC) and creatinine (Cr) concentration in Chinese adult patients with chronic kidney disease (CKD),and evaluate their applicability. Methods Six tertiary hospitals located in different geographic regions of China formed a multicenter nonrandom test research,composed of the Mianyang Central Hospital,the Peking Union Medical College Hospital,the First Affiliated Hospital of Zhengzhou University,the First Bethune Hospital of Jilin University,the Nuclear Industrial 416 Hospital,and the First people′s Hospital of Neijiang.According to K/DOQI guidelines and pre-set exclusion criteria,788 adult patients with CKD were enrolled,from October 2010 to December 2011 (421 males,367 females,age from 19-year-old to 87-year-old),and randomly divided into development group (n=687) and validation group (n=101) with a stratified sampling take into account both gender and disease stages.Two-sample 99mTc-DTPA plasma clearance was determined as the reference GFR (rGFR),meanwhile serum CysC and Cr were also detected.The data of development group was used to fit GFR-estimating equation using nonlinear regression,and the data of validation group to evaluate the applicability of eight equations,including two developed equations,Cockcroft-Gault equation (eGFR1),MDRD equation (eGFR2),two MDRD/CKD-EPI equations (eGFR3, eGFR4) and two Chinese eGFR Investigation Collaboration equations (eGFR5, eGFR6), and compared each other in the consistency with rGFR. Results The mean rGFR,CysC and Cr of development group were 44.19 (3.51-166.00) ml/(min·1.73 m2),1.88 (0.59-8.62) mg/L and 1.73 (0.40-19.77) mg/dl,respectively.The mean rGFR,CysC and Cr of validation group were 47.85 (10.49-148.12) ml/(min·1.73 m2),1.79 (0.66-7.22) mg/L and 1.56 (0.48-23.34) mg/dl, respectively. Two best-fit GFR estimation equations were established using nonlinear regression: eGFR7=173.9×CysC-0.725×Cr-0.184×Age-0.193 (if female×0.89)(R2=0.734,P=0.000),and eGFR8=78.64×CysC-0.964 (R2=0.764,P=0.000). With Bland-Altman plot,Passing & Bablok regression,and Mountain plot analysis,only the 95% agreement limits of eGFR7 and eGFR8,did not exceed the prior acceptable tolerances[<60 ml/(min·1.73 m2)],and they did not show significant differences in proportional differences (the 95% CI of slopes included B=1) and constant differences (the 95% CI of intercepts included A=0) comparied with rGFR,and their Mountain plots almost overlapped each other that indicated their basical agreement in GFR estimating effect.As compared with rGFR,the deviation of eGFR1,eGFR2,eGFR5 or eGFR6 was larger than of eGFR7 or eGFR8,especially eGFR3 and eGFR4 underestimated GFR overall. Conclusions It can provide an equal precision to estimate GFR using serum CysC alone or in combination with serum Cr.The agreement of GFR marker′s measured method must be considered adequately before a literature eGFR equation was chosen.Otherwise,the eGFR value would have a large error,which may lead to misdiagnosis and mistreatment.(Chin J Lab Med,2013,36:352-359) Key words: Renal insufficiency; chronic; Glomerular filtration rate; Cystatin C; Creatinine; Algorithms