利用測量的肌氨酸酐廓清率比較公式計算之肌氨酸酐廓清率及腎絲球過濾率

A comparison of measured creatinine clearance (mClCr) versus estimated creatinine clearance (eClCr) and estimated glomerular filtration rate (eGFR)

吳玉惠(Yu-Hui Wu)

指導教授 ： 林慧玲

國立臺灣大學/醫學院/臨床藥學研究所/碩士(2013年)

https://doi.org/10.6342/NTU.2013.10339

摘要

研究背景
臺灣地區慢性腎臟疾病人口眾多。雖然利用血中肌氨酸酐（creatinine；SCr）評估腎功能有諸多影響因子，但由此評估方式已廣被接受。常用於劑量調整的Cockcroft-Gault（CG）公式計算之肌氨酸酐廓清率（estimated creatinine clearance；eClCr）和用於慢性腎臟疾病分級的Modification of Diet in Renal Disease（MDRD）公式估計的腎絲球過濾率（estimated glomerular filtration rate；eGFR）不僅單位不同，兩個計算結果在不同人種及個體上也沒有一定關係。收集尿液測量的肌氨酸酐廓清率（measured ClCr；mClCr）為臨床可執行的測量腎功能方法，而腎功能評估公式計算數值與mClCr間關係和差值也尚未清楚瞭解。了解eClCr 和eGFR公式與mClCr間關係和差值對於腎功能評估數值的解讀有很大的幫助，找出影響其間差值的因素也可提供更多臨床評估腎功能時的參考。此外，臨床上仍有爭議的問題例如在過重病人使用CG公式估計eClCr時是否用理想體重帶入、Salazar公式是否適合應用於過重病人；在SCr<0.8 mg/dL病人在估算eGFR和eClCr時是否應以0.8 mg/dL帶入計算；以及在計算CG公式時女性校正參數是否為0.85 。

研究目的
1. 主要目的：
利用臨床收集尿液測量之mClCr，找出各eClCr和eGFR公式計算值與mClCr數值差異，研究差異是否因性別、體型、年齡、SCr數據等病而變化，並研究特殊族群（過重、SCr過低病人）是否須調整帶入公式之體重或SCr。
2. 次要目的：
利用有測量SCr的所有病人資料，試圖找出年齡、體重、SCr數值對eGFR和eClCr公式計算數值及其間比值、差值的影響。

研究材料與方法
使用2012年臺大醫院電子病歷資料庫有測量SCr病人族群，排除SCr不在檢驗線性範圍、身高、體重紀錄不完整或極限值、急性腎損傷、特殊疾病的病人後得到有測量SCr的所有病人大資料庫；再由其中找出記錄24小時尿量並測量尿中creatinine（urine creatinine ; UCr）之病人。利用Siersbaek-Nielsen研究所建立依照年齡每天由尿液排除的creatinine量數據，確認24小時尿液收集的完整性，得到有mClCr病人。
1. 有mClCr病人分析
由尿量、UCr、SCr計算並利用體重、BSA校正單位得到的mClCr，做為參考值，與各eClCr（CG、體重或BSA校正的CG、Salazar）和eGFR（MDRD、CKD-EPI、Taiwanese MDRD）公式計算值做偏差（bias，公式計算值和mClCr差值）、精確度（precision，公式計算值和mClCr分佈集中程度）、相對偏差（relative bias，公式計算值和measured ClCr相比高估或低估百分比）及準確度（accuracy）。利用性別、體型、SCr數值、年齡、CKD stage、有無利尿劑處方做分層分析研究不同病人臨床特性是否造成差異。
2. 有測量SCr的所有病人大型資料庫分析
有測量SCr的所有病人資料利用迴歸係數（regression coefficient）、判定係數（R2）、散佈圖（scatter plot），試圖找出年齡、體重、SCr數值對eGFR（MDRD、CKD-EPI、Taiwanese MDRD）和eClCr（CG、體重跟BSA校正的CG）公式計算數值及其間比值、差值的影響。

結果
有測量SCr的病人共有80542人，男性佔49.4%；SCr<0.8 mg/dL有30679人（佔全體人數38.1%）其中女性佔87.3%。有收集24小時尿液且經確認完整性共268人，男性佔53.7%，SCr<0.8 mg/dL有80人（佔全體人數29.9%）其中女性佔有80%。有measured ClCr的病人比起全體有測量SCr的病人，腎功能較差。
各eClCr和eGFR公式計算值對應校正單位後的mClCr相關性很好（R2達0.8以上），不過在SCr<0.8 mg/dL族群R2數值降低。在79.1%（Salazar）~93.3%（Taiwanese MDRD）病人族群中eClCr和eGFR公式計算數值會低估mClCr，雖然不同公式低估的bias略有不同，不過relative bias在每個公式約為20%；且腎功能公式計算結果落在measured ClCr數值正負10%內比例很低，最高的出現在Salazar公式，不過也只有26%落在measured ClCr數值正負10%內。
分層分析，女性相較男性低估較多且準確度普遍也較差。過重族群比起非過重族群bias較多且accuracy普遍也較差；而以理想體重帶入CG公式會使得relative bias由原本的17.7%變成37.2%。Salazar公式在過重的病人低估24.0%，而在非過重的病人只低估14.6%。SCr<0.8 mg/dL病人雖然bias、precision相對SCr≥0.8 mg/dL較差，但因mClCr數值較大，relative bias及accuracy反而較好。在SCr<0.8 mg/dL病人以0.8 mg/dL帶入公式計算，會使bias、precision、relative bias、accuracy都變差。不同CKD group雖然可見到bias、precision隨著腎功能變差明顯變好，但relative bias、accuracy卻沒有相同趨勢。而以mClCr（mL/min/1.73 m2）及校正單位的eClCr和eGFR將病人分到CKD stage 1~5，依據mClCr與不同公式算出值在不同CKD組別中人數百分比差異，在CKD stage 1最大，然後隨著腎功能變差而漸減少。此外，有利尿劑處方的族群比起沒有的年齡較大、尿量較多、但UCr較低、mClCr顯著較差，而bias、precision、relative bias、accuracy在有利尿劑族群結果較好。
由所有測量SCr的所有病人資料做eGFR （MDRD、CKD-EPI、Taiwanese MDRD）和eClCr（CG、體重跟BSA校正的CG）公式相關性研究，MDRD公式得到的值和CG公式得到的值R2＝0.64為最差，其餘eGFR與eClCr的R2都大於0.7，顯示eClCr和eGFR公式關係良好，對腎功能評估趨勢相同。當年齡增加一歲，CG公式算出之eClCr會降低1.41 mL/min，體重校正後數值會降低1.57 mL/min/72 kg，BSA校正後數值會降低1.38 mL/min/1.73 m2，而在eGFR公式是每增加一歲，約略降低1 mL/min/1.73 m2。體重每增加一公斤，CG公式計算值會增加1 mL/min，而校正單位後的CG公式及其他eGFR公式幾乎不受影響。由eGFR和eClCr公式計算數值比值和差值分佈圖可發現，隨著年齡的增加，eGFR與CG公式及體重、BSA校正的CG公式所得eClCr差值由負變正；而比值方面在60多歲有明顯上升趨勢（eGFR數值較大）。而隨著體重的增加eGFR公式與CG公式及BSA校正的CG公式差值由正變負，比值也由>1（eGFR計算結果較大）變為<1（eGFR計算結果較小），但這現像在eGFR公式與體重校正過後的CG公式看不到。

結論
所有eClCr和eGFR公式計算數值結果與mClCr對於腎功能評估有共同趨勢，但所有eClCr和eGFR公式都會低估mClCr大約20%，且各公式計算結果與mClCr相比，準確度不佳。若以mClCr為腎功能標準，過重及SCr<0.8 mg/dL族群校正帶入eClCr和eGFR公式的數值會使低估的情況變得更嚴重。Salazar公式結果在過重的族群反而比非過重的族群對mClCr有更多偏差更不準確且在過重病人估算eClCr較CG公式有更多偏差更不準確。分析有測量SCr的所有病人資料eClCr和eGFR公式關係很好，年齡對所有公式計算數值都有影響、體重對CG公式影響特別大。eGFR公式大於eClCr公式計算結果出現在年紀大、體重輕的族群。對於有測量SCr的所有病人資料可以依性別、年齡、體重、SCr更進一步做分層分析找出在哪些族群eGFR和eClCr公式間差異特別大或小。

關鍵字

肌氨酸酐 ； 肌氨酸酐廓清率 ； 腎絲球過濾率 ； 慢性腎功能不全

並列摘要

Background
The prevalence of chronic kidney disease (CKD) is high in Taiwan. Thought there are many factors that may affect the measurement of serum creatinine, it is a well-accepted marker for renal function assessment. The estimated creatinine clearance (eClCr) by Cockroft-Gault (CG) equation is commonly used as a reference for dosage adjustment; while the estimated glomerular filtration rate (eGFR) by Modification of Diet in Renal Disease (MDRD) equation is used in CKD staging. eClCr and eGFR not only have different units, but the results of estimation from the two varied among races and individuals. Collecting 24-hour urine to calculate measured createnine clearance (measured ClCr; mClCr) is a clinical applicable measurement of renal function. However, the relationship and difference between renal function estimation and mClCr remains unknown. Understanding the correlation between mClCr and renal function assessment equations and the factors that may influence the difference between eClCr and eGFR can is very helpful during clinical renal function assessment. Furthermore, there are some clinical controversial questions, such as whether to use ideal boday weight (IBW) to replace actual body weight (ABW) in CG equation for obese patiens or should Salazar equation be used instead; whether to use SCr=0.8 mg/dL in eClCr and eGFR equations when true SCr<0.8 mg/dL; and whether 0.85 should be used as correcting factor for female in CG equation.

Objective
1. Primary endpoint:
Using database that containing patients whose mClCr was available (mClCr database) to explore the correlation and difference between mClCr and eGFR or eClCr; to investigate the influence of gender, body size, age or SCr on the difference; and to evaluate the need to adjust weight for the obese patients and the need to adjust SCr for the low-SCr patients.
2. Secondary endpoint:
Using the database containing all patients whose SCr was available (SCr database) to investigate the impact of age, weight, SCr on the eGFR and eClCr estimation; and to evaluate the difference and ratio among these estimations.

Materials and Methods
In National Taiwan University Hospital electronic database, patients whose SCr was available in 2012 were identified. Exclude those patients with SCr beyond linear test range, lack of weight or height, with extreme weight or height, with acute kidney injuiry or specific diagnosis to form the SCr database. We identified the patients who had a record of 24 hour urine and urine creatinine (UCr) in the SCr database then used the urine cretinine excretion amount in different age from Siersbaek-Nielsen results to ensure the completeness of urine collection to form a database for patients who had mClCr (mClCr database).
1. Analysis using mClCr database
Using mClCr derived from urine amount, UCr, SCr and body weight or BSA adjustment as a reference, we calculated the bias (difference between equation estimation and mClCr), precision (standard deviation of bias), relative bias (% underestimate or % overestimate of equation estimation), and accuracy of eClCr (by CG, weight and BSA adjusted CG, or Salazar equation) or eGFR (by MDRD, CKD-EPI, Taiwanese MDRD equation). Stratified analysis between patients with different gender, body size, SCr, age, CKD stage and the use of diuretics or not to see if these characteristics caused different results.
2. Analysis using SCr database
In the SCr database, we used regression coefficient, R2 and scatter plot to find the influence of age, weight and SCr on the correlations and differences among eGFR (by MDRD, CKD-EPI, or Taiwanese MDRD equation) and eClCr (by CG, weight or BSA adjusted CG equation) from different equations.

Results
The study included 80542 patients who had SCr data among them 49.4% were male, and 30679 (38.1%) patients had a SCr<0.8 mg/dL. Among the 30679 patients, 87.3% were female. In the 268 patients who had mClCr, 53.7% is male, while 80 (29.9%) patients had a SCr<0.8 mg/dL. Among those with SCr<0.8 mg/dL, 80% is female. Compared to SCr database, patients in mClCr database had poorer renal function.
Good correlation existed between the estimation by different eClCr and eGFR equations and unit-adjusted mClCr, but R2 decrease in patients with SCr<0.8 mg/dL. In 79.1% (Salazar) to 93.3% (Taiwanese MDRD) patient’s eClCr and eGFR underestimated mClCr. Though different eClCr and eGFR equations caused different bias, but the relative bias remained around 20% for all equations. The 10% accuracy was poor for all equations. Estimation from Salazar equation had the best accuracy, but still, only 26% patients were within 10% difference of mClCr.
In stratified analysis, female underestimated more than male, and had poorer precision. Obese patients had worse bias, precision and accuracy than nonobese patients. Substituting ABW by IBW in CG equation in obese patients resulted in an increase in relative bias from 17.7% to 37.2%. Salazar equation underestimated eClCr by 24.0% in obese patients, while only underestimated by 14.6% in non-obese patients. Patients with SCr<0.8 mg/dL had poorer bias and precision when compared to those with SCr≥0.8 mg/dL. However, because patients with SCr<0.8 mg/dL had a better mClCr, relative bias and accuracy were better. Using SCr=0.8 mg/dL to substitute the real SCr in patients with SCr<0.8 mg/dL to calculate eClCr and eGFR resulted in greater bias and inaccuracy. The bias and precision improved as renal function deteriorated, while relative bias and accuracy did not show the same trend. Using mClCr and estimations from different equations (adjusted to mL/min/1.73 m2) to group patients into CKD stage1 to stage 5, the differences in percentage of patients in each stage grouped by mClCr and different equations decreased as renal function deteriorated. Furthermore, patients in diuretics group had statistically significantly greater age and urine amount, less UCr and mClCr; but had less bias, relative bias with better precision and accuracy when compared eClCr or eGFR with mClCr.
The correlation between eGFR (by MDRD, CKD-EPI, or Taiwanese MDRD equation) and eClCr (by CG, weight or BSA adjusted CG equation) in SCr database was good, with most R2> 0.7. The worst R2 was seen between estimations by MDRD equation and CG equation. When age increased by 1 year, the eClCr by CG equation decreased by 1.41 mL/min, the weight-adjusted eClCr decreased by 1.57 mL/min/72 kg, the BSA-adjusted eClCr decreased by 1.38 mL/min/1.73 m2. Among the eGFR equations, when age increased by 1 year, eGFR decreased by about 1 mL/min/1.73 m2. When body weight increased by 1 kg, eClCr by CG equation increased abound 1 mL/min, while eClCr and eGFR by other equations were almost unchanged. The scatter plot of ratio and difference between estimations by eGFR and eClCr equations revealed that the difference between eGFR and eClCr increased as age increased, with the ratio of eGFR and eClCr had a sharp increase at around the age of 60s. As weight increased, the difference between eGFR and eClCr (by CG or BSA-adjusted CG equations) decreased and the ratio also decreased from >1 (eGFR> eClCr) to <1 (eGFR< eClCr). However, this trend was not seen in eGFR and weight-adjusted CG equation.

Conclusions
The estimations from all eClCr and eGFR equations showed good correlation with mClCr. However, all eClCr and eGFR equations underestimated mClCr by about 20% and had poor accuracy. Using mClCr as a reference, adjusting SCr for patients with low SCr and adjusting weight for obese patients worsen the underestimation. Salazar equation appears to be less accurate in the obese patients than non-obese and had greater bias and was less accurate than CG equation in estimating eClCr. Analysis in SCr database revealed good correlation between eGFR and eClCr equations. Age affected all equations, while body weight had greater impact on CG equations. After normalized eClCr to mL/min/1.73 m2, eGFR was greater than eClCr among old and slim patients. Further study is needed to stratify patients in SCr database to explore the influence of gender, age, weight and SCr on the difference between eGFR and eClCr.

並列關鍵字

Creatinine ； creatinie clearance ； glomerular filtration rate ； renal insufficiency, chronic

參考文獻

1. 行政院衛生署國民健康局. 2007 年台灣地區高血壓、高血糖、 高血脂之追蹤調查研究.

2. Wen CP, Cheng TYD, Tsai MK, et al. All-cause mortality attributable to chronic kidney disease: a prospective cohort study based on 462 293 adults in Taiwan. The Lancet 2008;371:2173-82.

3. JT D, RL T, GC Y, al e, editors. Pharmacotherapy: a pathophysiologic approach. 8th edition2011.

4. Verbeeck RK, Musuamba FT. Pharmacokinetics and dosage adjustment in patients with renal dysfunction. European journal of clinical pharmacology 2009;65:757-73.

5. Tett SE, Kirkpatrick CMJ, Gross AS, McLachlan AJ. Principles and Clinical Application of Assessing Alterations in Renal Elimination Pathways. Clin Pharmacokinet 2003;42:1193-211.