神經外科急重症病人之藥品動態學研究—以萬古黴素為例

Abstract

研究背景與目的 萬古黴素在臨床上最重要地位在於methicillin-resistant Staphylococcus aureus （MRSA）感染的治療。為了避免治療失敗或是引起其他抗藥性菌種產生，治療過程維持萬古黴素的血清濃度高於目標值是很重要的。 過去的回溯性研究顯示台大醫院神經外科急重症病人具顯著較高的萬古黴素廓清率（Clv）。此研究導出之模型可預測此類病人的藥品動態學參數。此模型以Cockcroft-Gault公式估計之肌肝酸廓清率（ClCr）以及尿量/腦脊髓液引流量為變數，預測Clv；另以年齡和體重預測萬古黴素之分布體積（Vd）。 因此，本前瞻性研究欲驗證先前研究之藥品動態學模型，並描述此類病人萬古黴素排除的途徑。本研究也會以體外細胞模型評估高張力甘露醇（Mannitol）對於萬古黴素穿透運輸的影響。 研究方法 本研究針對神經外科急重症病人收案，共有16位使用萬古黴素治療且進行藥品血中濃度監測（TDM）的病人參與。我們收集病人的尿液檢體，而病人若有使用腦室外引流則同時收集腦脊髓液引流液檢體。我們建立了一個高效液相層析（HPLC）分析法，用以測量病人檢體之萬古黴素濃度。之後，可從TDM資訊和檢體濃度資訊計算出Clv和Vd的觀察值，並且使用迴歸分析及迴歸診斷檢驗其與Clv和Vd預測值之相關性。 另外，將Caco-2細胞種於transwell系統中使其形成cell monolayer。再以不同濃度之mannitol評估對萬古黴素穿透運輸的concentration-dependent影響。 研究結果 此研究收案之病人平均年齡66 ± 13歲，入院體重平均為61.6 ± 6.1公斤，TDM當日平均尿量4013 ± 1710 mL/day（65.12 ± 26.82 mL/kg/day）。TDM當日之平均血液肌肝酸（SCr）濃度為0.58 ± 0.21 mg/dL，且測量之肌肝酸廓清率（ClCr,measured）平均值為161.6 ± 73.7 mL/min。 24小時內經尿液排除之萬古黴素約佔100 %之總劑量。另外，萬古黴素之腎廓清率（Clv,renal）與萬古黴素之總廓清率（Clv,total）比值為0.99 ± 0.10，且兩組數值具高相關性（Pearson’s r = 0.9577）。因此後段的分析使用Clv,renal為標竿，進行先前研究之Clv模型驗證。初步的迴歸分析顯示低相關性（Pearson’s r皆低於0.5）；然而將一個極值排除後，觀察到整體相關性提升，而迴歸分析則顯示以ClCr模型預測之Clv具最佳的預測性（Pearson’s r = 0.8344）。另外，模型預測之Vd與觀察之Vd具低相關性（Pearson’s r = 0.2912）。 我們也針對萬古黴素之排除路徑進行研究。24小時內經腦脊髓液引流液所排除的萬古黴素量極低（與藥品劑量比值為1.16 x 10-4），且萬古黴素具低腦脊髓穿透度（18.6 %）。腎排除為主要途徑，我們認為其主要機轉是經由腎絲球過濾排除，因為ClCr,measured可解釋高達70 % 之 Clv,renal變異。 至於細胞實驗的結果，我們觀察到高張mannitol會造成Caco-2 cell monolayer完整性的破壞，進而增加萬古黴素之穿透。 結論 本研究收案之病人具高尿量、高腎廓清率之特性，且這樣的特性與高Clv具高度關連性。此研究發現神經外科急重症病人之Clv可藉由ClCr預測。再者，我們可以使用預測之Clv計算更適當的給藥劑量，以達更好的治療效果。

Background and Objective Vancomycin plays an important role in the treatment of methicillin-resistant Staphylococcus aureus (MRSA). To avoid treatment failure and development of resistant strains, it is prudent to keep vancomycin serum levels above the target trough level during therapy. In the past, a retrospective study had shown neurosurgical intensive care (NSICU) patients of our institution had significantly elevated vancomycin drug clearance (Clv). The same study derived pharmacokinetic models using Cockcroft Gault estimated creatinine clearance (ClCr) and urine output (UO) /cerebrospinal fluid (CSF) drainage to predict Clv; also using age and body weight to predict vancomycin volume of distribution (Vd) for this population of patients. Therefore, it is the intent of this prospective study to validate the preliminary pharmacokinetic models; and to describe the route of vancomycin clearance. We are also interested in assessing the effect of hypertonic mannitol on vancomycin transport using an in vitro cellular model. Methods In this study, we prospectively recruited 16 NSICU patients receiving vancomycin therapy, and undergoing therapeutic drug monitoring (TDM). We also collected patient’s urine samples, and CSF sample if the patient received external CSF drainage during vancomycin therapy. We established a high-performance liquid chromatography (HPLC) assay to analyze urine and CSF vancomycin concentrations. Thereafter, we calculated observed Clv and Vd from TDM data and patient sample concentration data; by comparing observed and predicted Clv and Vd, we could validate preliminary models. A Caco-2 cell transwell system was used to assess the concentration-dependent effect of mannitol on vancomycin transport across epithelial cell. Results The NSICU patients recruited in this study had a mean age of 66 ± 13 years old, mean body weight on admission was 61.6 ± 6.1 kg, the mean urine output on TDM day was 4013 ± 1710 mL/day (65.12 ± 26.82 mL/kg/day). The average SCr on TDM day was 0.58 ± 0.21 mg/dL, with corresponding measured creatinine clearance (ClCr,measured) of 161.6 ± 73.7 mL/min. Vancomycin excreted from urine over 24 hours was found to contribute to about 100 % of given dose. Additionally, vancomycin renal clearance (Clv,renal) ratio to vancomycin total clearance (Clv,total) was 0.99 ± 0.10, with high correlation (Pearson’s r = 0.9577). Therefore, Clv,renal was used as a gold standard to validate the Clv model. Initial validation found low correlation (Pearson’s r less than 0.5); however, after exclusion of an outlier, the ClCr model showed best prediction of Clv (Pearson’s r = 0.8344). On the other hand, the Vd model from preliminary study showed suboptimal predictability (Pearson’s r = 0.2912). We also investigated the pathway of vancomycin excretion from the body. External CSF drainage had very little contribution to Clv (ratio to dose 1.16 x 10-4), and low CSF penetration was observed (18.6 %). Renal clearance was found to be the major route of excretion; and about 70 % of Clv variability could be explained by ClCr,measured, indicating glomerular filtration as the primary excretion pathway. From the Caco-2 cell in vitro data, we found that hypertonic mannitol could disrupt the integrity of Caco-2 monolayer, resulting in up to 9-fold increase of vancomycin transport from donor to acceptor compartment. Conclusion The NSICU patients of this study were presented with high urine output and enhanced renal clearance, which was associated with elevated Clv. The Clv of NSICU patients in our institution was best predicted by ClCr. We could then use the predicted Clv to calculate appropriate daily maintenance dose for NSICU patients.