建立臺灣人肌酸酐廓清率估計公式及每日肌酸酐排除量的參考值

Abstract

研究背景 依據腎功能調整藥品劑量在腎功能不全的病人是相當重要的事情。收集24小時尿液肌酸酐測得的肌酸酐廓清率（24-hour measured creatinine clearance; 24-hr mClCr）為調整劑量腎功能依據的黃金準則，而臨床上最常用Cockroft-Gault（CG）公式算出的肌酸酐廓清率（estimated creatinine clearance; eClCr）估計24-hr mClCr。 根據本團隊研究，發現CG公式低估24-hr mClCr，會有這樣的差異也許是因為種族的影響，因此想建立一個適用於臺灣人eClCr的公式。另外，也發現利用過去白人所建立的每日尿液肌氨酸酐排除量（urinary creatinine excretion; UCE）來排除收集不全的病人，會排除過多病人，故也想建立一個臺灣人每日肌氨酸酐排除量，以提供之後研究參考。 研究目的 主要目的是：建立臺灣人最適切的eClCr計算公式。次要目的是：建立臺灣人男性與女性各年齡層每日每公斤尿液creatinine排除的參考值。 研究材料與方法 回溯性研究某醫學中心2010年至2013年有24-hr mClCr資料的成年人，納入試驗有684筆24-hr mClCr，將病人以2比1隨機分配為建立組456筆及驗證組228筆。先利用建立組以24-hr mClCr為標準，發展出臺灣人的公式，並比較臺灣人的UCE是否跟西方人不同；再以驗證組的24-hr mClCr比較推導出來的公式以及不同eClCr公式，最主要是比較公式間的均方根誤差（root mean square error; RMSE，平均差值平方再開根號），另外也會看偏差度（bias，eClCr與24-hr mClCr差值中位數）、相對偏差度（relative bias，eClCr與24-hr mClCr相對差值中位數）、精準度（precision，eClCr與24-hr mClCr差值的四分位距）、準確度（accuracy，eClCr在24-hr mClCr ± 30%之內的比例）。 結果與討論 本研究發現臺灣人每天每公斤UCE比西方人還高，可能是相對照的文獻較為久遠，年代的不同造成飲食習慣的差異，進而影響creatinine的排除量。驗證組比較不同公式的表現，CG公式的RMSE為35.92 mL/min、relative bias為-21%，而本研究利用瘦肉體重（lean body weight; LBW）推導出的公式Eq-LBW2是所有eClCr表現最好的公式（RMSE=29.05 mL/min; relative bias=2.56%），而利用真實體重（total body weight; TBW）推導出的公式Eq-TBW2（RMSE=29.77 mL/min; relative bias=-3.04%）與LBW2並沒有顯著的差異。 結論 即使改變了尿液收集不全或過多的條件，CG公式仍然不適合用在臺灣人。新發展出的公式Eq-LBW2的表現最好，但LBW的計算並不易，故要臨床方便使用則用與Eq-LBW2差異不大的Eq-TBW2：男性eClCr=(169-年齡) x體重 x 0.82 (女性)/(75 x serum creatinine)。臺灣人每日尿液排除creatinine的量比西方人過去還多，因此使用西方人過去的UCE來判定尿液收集不全或過多是不恰當的，可考慮參考本研究做出的UCE表格。

Background It’s important to adjust dosage according to residual renal function in chronic kidney disease. Measured 24-hour creatinine clearance (24-hr mClCr) is a gold standard for dosage adjustment. Cockcroft and Gault (CG) equation is most commonly used for creatinine clearance estimation. According to our team’s past research, CG equation tends to underestimate 24-hr mClCr. The discrepancy may be owing to ethnicity. Therefore we wanted to develop an equation for Taiwanese. In the past, we used the daily urinary creatinine excretion (UCE) derived in Western country as a criterion to exclude patients of incomplete urine collection. This may exclude too many patients. Thus, we tempt to develop a creatinine excretion reference table for Taiwanese. Objective The primary endpoint is to derive an equation to estimate creatinine clearance (eClCr) in Taiwanese. The secondary endpoint is to establish a creatinine excretion reference table for Taiwanese. Materials and Methods It is a retrospective study in a medical center . Six hundred and eighty-four 24-hr mClCr from 2010/1/1 to 2013/12/31 were included and randomly assigned training group or validation group at a ratio of 2:1. There were 456 mClCr in training group, 228 mClCr in validation group. These 24-hr mClCr data in the training group were used to develop eClCr equations and creatinine excretion reference table. The performances of developed equations were compared with to other eClCr equations using data in the validation group. Root mean square error (RMSE), bias (median of difference between eClCr and 24-hr mClCr), relative bias (median of relative difference between eClCr and 24-hr mClCr), precision (interquatile of difference between eClCr and 24-hr mClCr), and accuracy (percentage of eClCr in 24-hr mClCr ± 30%) were used as indicators of performance. Results and Discussions We found that Taiwanese’s UCE (mg/kg/day) is higher than Western’s UCE. The reason may be that the corresponding study was old. The dietary habit may have changed with times, which may affect the UCE. The best equation for Taiwanese was Eq-LBW2 (RMSE=29.05 mL/min, relative bias= 2.56%) that developed by lean body weight (LBW) in this study. Eq-TBW2 (RMSE=29.77 mL/min, relative bias= -3.04%) had no significant difference with Eq-LBW2. The performance of CG was not so good as above (RMSE= 35.92 mL/min, relative bias= -21%). Conclusions CG equation may not be suitable for Taiwanese, even though we modified criteria of incomplete-collection/over-collection urine. The equation Eq-LBW2 developed in this study was the best eClCr equation for Taiwanese, but calculation of LBW was not easy. Thus, we recommend Eq-TBW2 that has similar performance to Eq-LBW2 instead in clinical settings. Eq-TBW2: eClCr=(169-age) x weight x 0.82 (if female) / (75 x SCr). Futhermore, because Taiwanese’s UCE (mg/kg/day) is higher than Western’s UCE, it is inappropriate to use Western’ UCE as a criterion of incomplete or over-collection urine in Taiwanese. The reference table developed in this study may be used.