基于血肌酐和胱抑素C的四个CKD—EPI方程对我国慢性肾脏病患者的适用性研究

Validation study of chronic kidney disease epidemiology collaboration predictive equations based on serum creatinine and cystatin C in Chinese chronic kidney disease patients

目的探讨基于血肌酐（SCr）和胱抑素C（CysC）建立的4个慢性肾脏病流行病学合作研究组（CKD—EPI）肾小球滤过率评估方程对我国CKD患者的适用性，为我国CKD群体肾小球滤过率（GFR）评估方程的选择提供依据。方法选取2007年9月至2010年12月就诊于中国境内不同区域的4家三级甲等综合医院的CKD患者共254例。以双血浆法99mTc-二乙三胺五乙酸（99mTc-DTPA）血浆清除率作为254例CKD患者的rGFR，测定患者SCr以及CysC浓度。同时应用CKD-EPI SCr方程、CysC方程、CysC方程（年龄、性别、种族校正）及SCr／CysC联合方程计算eGFR（分别标记为eGFR1、eGFR2、eGFR3、eGFRg），通过Wilcoxon符号秩检验、组内相关系数（ICC）及Spearman相关分析，比较4个eGFR与rGFR的差异、相关性、偏差、精密度。应用Bland—Altman图比较各eGFR与rGFR偏离程度。计算各eGFR的15％、25％、30％准确性（P15、P25、P30），以及对CKD各期的分期一致性。结果254例CKD患者以FR水平为[48．07（26．19～92．97）]ml·min-1·（1．73m2）-1。各eGFR与rGFR的Spearman相关系数在0．873～0．896之间，组内相关系数（ICC）为0．920—0．942，eGFR4的相关性最佳。绝对偏差以及偏差精密度结果eGFR4〈eGFR3〈eGFR2〈eGFR1。Bland—Altman图显示，4个eGFR的偏差回归线95％预测区间宽度分别为92．5、87．3、83，0、76．1ml·min-1·（1．73m2）-1，以eGFR4精确度最佳。P30结果为eGFR4〉eGFR3〉eGFR2〉eGFR1，但χ2检验显示差异无统计学意义（χ2=6．448，P=0．092）。4个eGFR对CKD分期的总一致率在48．4％～57．5％，以eGFRg一致性最高，但均不理想（Kappa值分别为0．405、0．348、0．366、0．463）。结论与CKD—EPISCr方程相比，CKD-EPI CysC方程优势并不明显。年龄、性别、种族校正的CysC方程在偏差、精密度、相关性、准确性等方面略优于CysC方程。CKD—EPISCr／CysC联合方程在偏差、精密度、相关性、准确性、分期一致性等方面均优于其他方程，但对我国CKD群体的适用情况仍不十分理想。建议建立适于我国CKD人群，并综合考虑年龄、性别等因素的SCr／CysC联合方程。

Objective To validate the four chronic kidney disease epidemiology collaboration （CKD-EPI） predictive equations based on serum creatinine （SCr） and cystatin C （Cys C） in Chinese CKD patients, and try to develop the GFR predictive equations for Chinese CKD patients. Methods 254 CKD patients were randomly selected from four Grade ⅢA hospitals in different regions in China from September 2007 to December 2010. Clearance of dual plasma sampling 99mTc-DTPA was used to measure glomerular filtration rate （rGFR） in 254 CKD patients. The serum concentration of Cr and Cys C were measured. CKD- EPI SCr equation, Cys C equation, Cys C equation adjusted for age, sex and race, SCr/Cys C combinated equation adjusted for age, sex and race were used to estimate GRF （ labeled as eGFR1, eGFR2, eGFR3 and eGFR4, respectively）. The correlation, bias and precision of eGFRs were compared with rGFR by Wilcoxon signed rank test, intraclass correlation coefficient （ICC） and Spearman correlation analysis. The deviation degree between rGFR and different eGFRs was compared via Bland-Altman graph. The accuracy within 15%, 25% , 30% （P15, P25, P30 ） and the staging correctness of eGFR against CKD at different stages was calculated. Results The rGFR in 254 CKD participants was [48.07 （26. 19 -92. 97）] ml·min-1 （1.73 m2） -1 The Spearman correlation coefficients （CC） of eGFR and rGFR varied within the range of 0. 873 - 0. 896 （ P = 0. 000 ）. The intra-class CC （ICC） varied within the range of 0. 920 - 0. 942. The correlation of eGFR4 was the best. The absolute deviations of 4 eGFRs and deviation precision were eGFR4 〈 eGFR3 〈 eGFR2 〈 eGFR1. The 95% confidence intervals for the regression line of 4 eGFRs shown by Bland- Altman graphs were 92. 5, 87.3, 83.0 and 76. 1 ml·min- 1 ·（ 1.73 m2 ） -1, respectively, with the best result of eGFR4. For P30, the correctness of 4 eGFRs were eGFR4 〉 eGFR3 〉 eGFR2 〉 eGFR1, but no significant difference was found by Chi square test （χ2 =6. 448 ,P =0. 092）. The overall correctness rate in 4 eGFRs against CKD stages were 48.4% -57. 5%, with the highest consistency of eGFR4, but their staging correctness rate were not ideal （Kappa values were 0. 405,0. 348, 0. 366 and 0. 463, respectively）. Conclusions Compared with CKD-EPI SCr equation, no advantage was found in CKD-EPI Cys C equation. The Cys C equation adjusted by age and sex shows a little advantages over CKD-EPI Cys C equation in bias, precision, correlation and accuracy. The CKD-EPI SCr/Cys C combinated equation adjusted by age, sex and race has advantage over other three equations not only in bias, precision, correlation and accuracy, but also in staging correctness. However, the validation of this equation is still not fairly ideal for Chinese CKD patients. Based on these findings, it is essential for the Chinese CKD patients to develop SCr/Cys C combined predictive equation which adjusted by age, sex or other factors.