Diagnostic value of cystatin C-based prediction equations for glomerular filtration rate estimation / 中华核医学与分子影像杂志

ABSTRACT

Objective To investigate whether cystatin C-based prediction equations for GFR estimation are superior to SCr-based prediction equations.Methods One hundred and ninety-eight consecutive patients (85 males,113 females,average age 66.5 years) who underwent GFR measurement with 99TcmDTPA and serum cystatin C and SCr tests were included in this retrospective study.GFR,serum cystatin C and SCr concentrations were determined by the Gates method (measured GFR),the particle-enhanced turbidimetric immunoassay,and the Jaffe method,respectively.Eight different equations (6 equations based on the serum cystatin C,and the other 2 based on SCr) were used to estimate GFR values,and the results were compared with that of the Gates method.Patients were divided into different groups according to the measured GFR (normalized to body surface area,1.73 m-2)normal renal function,mild,moderate or severe renal impairment groups.One-way analysis of variance and the least significant difference t test were used to compare the estimated GFR,andx2 test was used to compare the diagnostic efficiencies of different GFR estimation equations.Results Among 198 patients,159 cases were with renal impairment (78 mild,58 moderate,23 severe),and the other 39 cases were with normal renal function.For patients with moderate or severe renal impairment,the estimated GFR calculated by the Tan formula was not different from the measured GFR (severe(20.7±7.4) ml · min-1 vs (19.9±8.2) ml · min-1; F=6.75,t＜1.05; moderate(42.1±14.4) ml· min-1 vs (46.8±9.2) ml· min-1; F=10.49,t＜1.63; both P＞0.05),and it had the least error compared with the measured GFR (severe(12.3±7.0) ％ ; moderate(17.9± 13.0) ％).For the patients with mild renal impairment and normal renal function,the estimated GFR calculated by the Tan formula was not valuable.For the diagnosis of renal impairment,the sensitivity and accuracy of the modification of diet in renal disease (MDRD) formula were 66.0％(105/159) and 71.2％(141/198),respectively,and those of the chronic kidney disease-epidemiology collaboration (CKD-EPI) formula were 70.4％ (112/ 159) and 73.7％(146/198),respectively.The sensitivities and accuracies of the cystatin C-based formulas (≥83.6％ (133/159) and ≥79.3％(157/198),respectively) were higher than those of MDRD formula and CKD-EPI formula (x2 ≥23.50,all P＜0.01).For the diagnosis of chronic kidney disease (including 81 patients with moderate and severe renal impairment),the sensitivities of cystatin C-based prediction equations (≥ 86.4％ (70/81)) were higher than those of the MDRD formula and the CKD-EPI formula (76.5％ (62/81),79.0％ (64/81)),but the accuracies were slightly lower (Tan formula80.3％ (159/198),x2≥ 56.42,all P＜0.05).Conclusion The Tan formula may be more suitable for the GFR estimation than the MDRD formula and CKD-EPI formula in the patients with severe or moderate renal impairment (serum cystatin C≥ 1.55 mg/L),but it may not be reliable for the patients with mild renal impairment and normal renal function.