- 學位論文
Venetoclax用於急性骨髓性白血病之藥品動態學研究
Pharmacokinetics of Venetoclax in Patients with Acute Myeloid Leukemia
摘要
背景與研究目的 Venetoclax於2018年被核准併用低甲基化劑（hypomethylating agent, HMA）或低劑量cytarabine（low dose cytarabine, LDAC）治療75歲以上、或是無法接受高強度化學治療之初診斷急性骨髓性白血病（acute myeloid leukemia, AML）的病人。Venetoclax在體內主要經由肝臟CYP3A4酵素代謝，與CYP3A抑制劑存在明顯交互作用，然而進行治療AML時，常需使用posaconazole或voriconazole等CYP3A強效抑制劑以預防或治療侵入性黴菌感染，但現有研究中CYP3A強效抑制劑類別、使用情形等和臨床需求不盡相同，且目前亦缺乏大型venetoclax用於AML病人的藥動學研究。此外，於小型藥動學研究結果也發現在相似研究條件下，漢人族群的venetoclax血中濃度較歐美族群高，因此本研究旨在建立臺灣AML病人之venetoclax族群藥動學模型，藉以釐清顯著影響藥物血中濃度變化和個體間藥動參數差異的因子。 研究方法 本研究屬於多中心、前瞻、觀察性研究，研究對象為2020年8月1日至2021年7月31日期間正在使用或預計開始使用venetoclax的AML病人，venetoclax血中濃度採用超高效液相層析串連質譜法分析，也會同時檢測CYP3A抑制劑voriconazole及posaconazole的濃度。族群藥動學模型部分，以Monolix（version 2020R1）軟體來建立模型，參考現有族群藥動文獻來選擇適當的模型架構和藥動參數起始值，再針對分佈體積和清除率分別尋找可能會影響藥動參數的共變項。共變項的選擇過程是藉由逐步選取法（stepwise selection）進行，並以適配性圖（goodness-of-fit plot）和殘差分佈圖（residual scatterplot）評估最終模型是否較基礎模型改善。此外，將使用venetoclax族群藥動模型得出之最低血中濃度（Cmin）、最高血中濃度（Cmax）及AUC0-24和療效、重要血液學不良事件以單變項線性迴歸或羅吉斯迴歸分別分析其關聯性。 研究結果 總共納入36位病人、168筆venetoclax濃度資料進行族群藥動模型建立，平均約8小時左右達到血中濃度高峰。CYP3A強效抑制劑部分，有12位併用posaconazole、16位併用voriconazole。現有文獻中venetoclax族群藥動模型皆以二室模型、一級吸收速率和一級排除速率來模擬其藥動學變化，然而考慮本研究服藥頻率和抽血時間點分佈，及venetoclax之排除半衰期長達26小時的情形下，若採用二室模型可能缺乏足夠的抽血點來描述此族群排除相的血中濃度變化，故採用一室模型、一級吸收速率和一級排除速率為基礎架構，當個體內變異為比例型（proportional）、分佈體積和清除率的個體間差異為指數型（exponential）時，最符合本研究的濃度分佈。經逐步選取法完成共變項篩選後，最終模型的族群分佈體積為47.28 L，且給予的IV輸液體積顯著影響個體分佈體積，當給予0.5 L不限種類的輸液時，該個體的分佈體積會上升84.1%到87.03 L。族群清除率為5.44 L/hr，對於同時併用posaconazole或voriconazole者，其Cmin對於個體清除率具不同程度的影響，當posaconazole Cmin為1 mcg/mL時，該個體的venetoclax清除率會下降67.6%到1.76 L/hr；若voriconazole Cmin為1 mcg/mL時，則venetoclax清除率會下降58.3%變為2.27 L/hr。本研究預測個體分佈體積和清除率之平均值分別為62.81 ± 28.15 L和3.88 ± 2.99 L/hr。針對模型篩選出的顯著共變項進行venetoclax曝露量的比較，顯示有給予及沒有給予IV輸液的病人，venetoclax濃度和AUC沒有顯著差異。併用fluconazole時，venetoclax的平均濃度和AUC上升1.5倍，而併用voriconazole和posaconazole時的dose-normalized Cmin、Cmax和AUC則分別上升6倍、3倍和4倍。 療效方面，有12位（37.5%）病人達到ELN response criteria中定義的完全緩解；血液學不良事件部分，分別有33位（86.8%）、34位（89.5%）、29位（76.3%）和19位（50%）於治療過程中曾發生CTCAE grade 3以上的neutropenia、thrombocytopenia、anemia和febrile neutropenia，僅一位病人於治療過程中發生輕微的tumor lysis syndrome，之後也成功地重新開始venetoclax治療。單變項迴歸分析結果顯示，venetoclax血中濃度或曝露量跟達到完全緩解的機率或發生CTCAE grade 3以上血液學不良事件的機率沒有明顯關聯。經多變項迴歸分析後，性別與達到完全緩解的機率有顯著關聯；此外，本研究中secondary AML病人發生grade 3以上anemia的風險也較高，且當病人開始療程前的absolute neutrophil count較高，發生grade 3以上neutropenia的機率顯著較低。 結論 本研究收納36位AML病人建立之venetoclax族群藥動學模型指出，給予大量IV輸液可能使病人分佈體積顯著上升，然而有給予輸液與未給予者之藥品血中濃度和曝露量沒有顯著差異；而併用posaconazole或voriconazole時，其Cmin越高、個體清除率越低，本研究中併用posaconazole和voriconazole病人族群的venetoclax曝露量也顯著較未併用者高。進一步分析療效、血液學不良事件與venetoclax曝露量之間並無顯著關聯性。本研究除了建立臺灣AML病人族群藥動模型，也提供未來venetoclax在臨床使用上的藥動參考資料。
並列摘要
Background and objective Venetoclax is an oral BCL-2 protein inhibitor approved for use in combination with azacitidine, decitabine, or low-dose cytarabine (LDAC) for front-line treatment of acute myeloid leukemia (AML) in older patients or those unfit for induction chemotherapy. Venetoclax undergoes metabolism via hepatic CYP3A enzymes, and drug interaction may exist when concomitant use with CYP3A inhibitors. The change of venetoclax plasma concentration while patients are taking hepatic CYP enzyme inducers or inhibitors in Taiwanese population remains unclear. This study is designed to examine the plasma concentration of venetoclax in patients with AML, to establish a population pharmacokinetic model of venetoclax in these population, and to analyze the covariates that may have effect on individual pharmacokinetic parameters. Method A multicenter, prospective, observational study was conducted from August 1, 2020 to July 31, 2021. Adult AML patients who initiated or were going to receive venetoclax treatment were included. Venetoclax, voriconazole, and posaconazole plasma concentrations were examined by UHPLC-MS/MS. Venetoclax population pharmacokinetic (PPK) model was built with Monolix (version 2020R1) software, and the appropriate structural model and initial value was selected from PPK literatures. Covariates that may affect volume of distribution and clearance were determined with stepwise selection. For model evaluation, goodness-of-fit plot and residual scatterplot were used to evaluate whether the final model is better than the basic model. In addition, relationship between plasma concentration, exposure and therapeutic effectiveness as well as hematological adverse events will be tested with simple linear regression or simple logistic regression individually. Result A total of 36 patients and 168 venetoclax concentration data were included. Peak plasma concentration was reached in about 8 hours on average. Among all the patients, 12 patients had concomitant use with posaconazole and 16 patients with voriconazole. Two-compartment model with first-order absorption and first-order elimination was the most common structural model in previous studies. However, there may not be enough blood sampling points in this study to describe plasma concentration changes during the elimination phase in our study. Therefore, a one-compartment model with first-order absorption and first-order elimination was selected. The intra-individual variability and inter-individual variability were best described by proportional model and exponential model, respectively. The final model showed that population volume of distribution was 47.28 L and volume of IV infusion fluid, as a continuous covariate, increased individual volume of distribution. When 0.5 L of IV infusion fluid was given, the patient’s individual volume of distribution would rise to 87.03 L. The population clearance was 5.44 L/hr. For patients who had concomitant use of posaconazole or voriconazole, its minimum concentration (Cmin) was explanatory for the value of individual clearance. For instance, if posaconazole Cmin was 1 mcg/mL, the individual clearance would drop to 1.76 L/hr. If voriconazole Cmin was 1 mcg/mL, the individual clearance would become 2.27 L/hr. Mean individual volume of distribution and clearance, estimated with our final model, were 62.81 ± 28.15 L and 3.88 ± 2.99 L/hr, respectively. Venetoclax concentrations and AUC were compared between patients with and without receiving IV infusion fluid, which showed that there was no significant difference. Comparing with patients that didn’t use any CYP3A inhibitor, venetoclax mean dose-normalized Cmin, Cmax and AUC were approximately 1.5-fold higher while concomitant with fluconazole. Furthermore, mean dose-normalized Cmin, Cmax and AUC were about 6-, 3-, and 4-fold higher while concomitant with voriconazole and posaconazole. In terms of treatment outcome, 12 (37.5%) patients achieved complete remission during the observation period in our study. Hematological adverse events were common under venetoclax treatment. Greater than CTCAE grade 3 neutropenia, thrombocytopenia, anemia, and febrile neutropenia occurred in 33 (86.8%), 34 (89.5%), 29 (76.3%) and 19 (50%) patients, respectively. Only one patient developed minor tumor lysis syndrome, and re-initiated venetoclax treatment soon after TLS recovery. Simple logistic regression analysis showed that venetoclax exposure had no relationship with the probability of complete remission or hematological adverse events statistically. After multivariable logistic regression, female patients were associated with increasing probability of complete remission. On the other hand, baseline absolute neutrophil count and AML type were also found to be associated with probability of ≥ grade 3 neutropenia and anemia, respectively. Conclusion The venetoclax population pharmacokinetic model established in this study illustrated that administration of a large amount of IV infusion fluid may increase volume of distribution and further decrease plasma concentration. When concomitant with posaconazole or voriconazole, its Cmin had significant impact on venetoclax individual clearance. Our study also demonstrated that venetoclax exposure were significantly higher in these patients. These covariates were important to AML patients under clinical treatment and monitoring. This study not only clarified the influence of these factors on venetoclax individual parameters, but also provided information of pharmacokinetic characteristics in clinical situations.