以液相層析質譜技術精進藥物療劑監測和生物指標開發

Abstract

近年來，液相層析質譜技術（LC-MS）顯示出巨大的潛力以提升精準醫療於各種疾病領域。透過液相層析質譜技術精進藥物療劑監測（therapeutic drug monitoring; TDM）和生物標誌物的開發，使得藥物的定量能更加準確並有望找出具有疾病特異性的生物標誌物。然而，液相層析質譜技術依然存在一些問題，其中包括全血基質的基質效應會引起的分析偏差，以及無法取得穩定同位素標記內標（SIL-IS）的問題。因此，目前仍需要創新的分析策略來解決這些可能影響藥物療劑監測準確度的問題。此外，雖然體學（omics）科學已廣泛應用於尋找生物標誌物，新興的脂質體學領域額外提供了脂質組成和結構的詳細資訊，這些資訊仍需要更多研究需來進一步探討脂質在疾病中的作用。因此，本論文旨在透過探索這些問題來促進精準醫療的發展。 本論文分為三個部分。前兩部分專注於藥物療劑監測，而第三部分旨在尋找脂質類的生物標誌物。第一部分致力於提升乾血點（DBS）的定量準確性以利於藥物療劑監測的應用。乾血點樣本的其中一個分析挑戰是血比容（Hct）效應，不同血比容可能影響乾血點的定量準確性。以往的研究常常忽視與血比容相關的分佈偏差。本研究旨在提供一套完善流程以評估血比容效應，並選擇了voriconazole和posaconazole這兩個藥物作為示範來評估血比容效應。通過研究三個重要參數，我們證明了添加固態的分析物到全血、在製備不同血比容樣品之前預先添加分析物，以及給予充分的平衡時間，將可以更良好的評估血比容效應。我們後續透過比對71對乾血點和血漿樣本，驗證了此流程有效性。本研究提供了一種可靠的策略來評估血比容效應，為未來的乾血點研究奠定基礎。 第二部分旨在開發一種液相層析質譜方式用於測量一新型抗癌治療藥物，抗體-藥物複合物（ADCs）的小分子藥物載體（payload）。然而，這些小分子藥物載體的同位素內標準品取得困難，造成準確定量上的挑戰。為此，我們開發了一種液相層析質譜方法並使用管柱後注入內標（PCI-IS）的策略來定量trastuzumab emtansine、trastuzumab deruxtecan，以及sacituzumab govitecan中的小分子藥物載體及其衍生物。我們選擇了結構相似物當成管柱後注入內標的候選分析物，並發現他們的校正效果得以使結構相對應的藥物擁有良好的定量準確性。該方法成功應用於定量乳癌病人臨床樣品當中的小分子藥物載體。此外，我們使用此液相層析質譜方法進行了trastuzumab deruxtecan的藥物療劑監測，發現其小分子藥物載體DXd濃度與血小板低下之間存在正相關關係。因此，建議未來對DXd濃度超過4.4 ng/mL的臨床患者，應常規監測血小板計數以及早發現副作用的發生。 第三部分，由於不同中風病因需要不同的預防策略，因此我們利用脂質體學來探索能夠區分中風病因的潛在臨床生物標誌物。我們使用實驗室內部的標的脂質體學的液相層析質譜方法，研究了包括甘油脂質（glycerolipids）、磷脂質（glycerophospholipids）和神經磷脂質（sphingolipids）的不同脂質。並比較不同的中風病因之間的脂質差異，包括心因性（cardioembolism; CE）、血管性（large artery atherosclerosis; LAA）或癌症。在606種發現的脂質種類中，我們觀察到只有血栓樣本可以顯著區分心因性和血管性組別之間以及心因性和癌症組別之間的不同。尤其是部分二酸甘油脂（DG）和多不飽和三酸甘油脂（TG）在心因性患者中比血管性患者中含量更多，而大多數神經醯胺（ceramides）和許多磷脂質在癌症組別中顯著升高。此外，我們分別確定了26種和116種脂質在區分血管性和癌症相對於心因性患者時，具有至少80%的準確性。這些發現強調了血栓脂質在區分中風病因中的重要性，並有助於我們對中風病理生理學的理解，以期能提供臨床端更多資訊來調整病人後續的治療方針。 總體而言，本論文展現了液相層析質譜技術在各種臨床應用的潛力。前兩部分透過解決藥物定量中的分析問題，以提升藥物療劑監測準確度，並展現藥物療劑監測對個人化醫療中的劑量優化的重要性。我們透過開發一套完善流程以評估血比容效應和創新的管柱後注入內標的定量策略，不僅有效提升了液相層析質譜技術的分析可靠性，還拓展了其在精準醫療中的應用。此外，通過探索中風病因分類中的脂質生物標誌物，我們展示了基於液相層析質譜技術的脂質體學在提供可能的病理機制的潛力。鑑定新穎的生物標誌物不僅增加了我們的診斷方法，亦提供了對疾病病理生理學的重要資訊，為標靶治療和精準醫療奠定了基礎。若未來能更廣泛將液相層析質譜技術整合到臨床常規治療流程中，將能對病患的個人化照護產生巨大的影響力。

Recently, liquid chromatography–mass spectrometry (LC–MS) has shown great promise in enhancing precision medicine across various disease areas. By advancing therapeutic drug monitoring (TDM) and biomarker discovery, LC–MS has enabled more accurate drug quantification and identification of disease-specific biomarkers. However, challenges remain, particularly matrix effects that cause analytical biases from whole blood matrices and issues with unavailable stable isotope-labeled internal standards (SIL-IS). Innovative analytical strategies are needed to address these TDM issues. Additionally, while omics sciences have been extensively utilized for biomarker discovery, the emerging field of lipidomics offers detailed insights into lipid composition and structure, necessitating further investigation into the role of lipids in diseases. Therefore, this dissertation aims to advance precision medicine by exploring these critical areas. This dissertation is divided into three sections. The first two sections focus on TDM, while the third section aims to discover lipid biomarkers. The first section seeks to improve the quantification accuracy of dried blood spots (DBS), a minimally invasive sampling technique in TDM. One major analytical challenge of DBS is the hematocrit (Hct) effect, which affects DBS quantification accuracy. Previous studies often overlooked the Hct-related distribution bias. We propose a new DBS preparation protocol to evaluate the Hct effect through the demonstration of voriconazole and posaconazole. By investigating three critical parameters, we demonstrated that a solid-state target analyte, pre-spiking target analytes before preparing different Hct levels, and allowing sufficient equilibrium time provide a more comprehensive Hct effect evaluation. The validity of the new protocol was confirmed by conversion factors from 71 paired DBS and plasma samples, offering a reliable strategy to assess the Hct effect for further DBS studies. The second section develops an LC–MS method for TDM of payloads in antibody–drug conjugates (ADCs), a novel class of anticancer therapeutics. A major issue of lacking SIL-IS for ADC payloads hinders accurate quantification. We developed an LC‒MS/MS method using a postcolumn-infused internal standard (PCI-IS) strategy for quantifying payloads and their derivatives in trastuzumab emtansine, trastuzumab deruxtecan, and sacituzumab govitecan. Structural analogs were selected as PCI-IS candidates, showing excellent accuracy for the respective target. This method was successfully applied to quantify payloads in clinical samples from breast cancer patients. Additionally, we conducted TDM of trastuzumab deruxtecan with the developed LC–MS method, revealing a positive association found between DXd concentrations and the development of thrombocytopenia. It is suggested that routinely monitoring platelet counts for patients with DXd concentrations over 4.4 ng/mL could be valuable in clinical practice for early detection of the potential site effect. The third section uses lipidomics to explore potential clinical biomarkers for differentiating stroke etiologies, which require distinct preventive strategies. Using in-house targeted lipidomics LC-MS/MS methods, we investigated various lipid species, including glycerolipids, glycerophospholipids, and sphingolipids. Patients were compared among different stroke etiologies, including cardioembolism (CE), large artery atherosclerosis (LAA), or Cancer. Among 606 identified lipid species, considerably significant differences were only found in thrombi samples between CE and LAA groups, and between CE and cancer groups. Notably, certain diacylglycerols (DG) and polyunsaturated triacylglycerols (TG) were enriched in CE patients compared to LAA patients, while most ceramides and many glycerophospholipids were elevated in the Cancer group. Additionally, we identified 26 and 116 lipids with at least 80% accuracy in distinguishing LAA and cancer from CE, respectively. These findings highlight the importance of thrombus lipids in differentiating stroke etiologies and contribute to our understanding of stroke pathophysiology, potentially informing future clinical strategies. In summary, this dissertation highlights the impact of LC–MS technology in various clinical applications. Addressing key methodological challenges in drug quantification, the first two sections underscore the importance of accurate TDM for dose optimization in personalized medicine. Developing robust preparation protocols and innovative quantification strategies enhances the reliability of LC–MS-based analyses and broadens its application in precision medicine. Additionally, exploring lipid biomarkers in stroke etiology classification demonstrates the potential of LC–MS-based lipidomics in providing mechanistic insights. Identifying novel biomarkers not only enhances our diagnostic approaches but also offers valuable insights into disease pathophysiology, paving the way for targeted therapies and precision medicine. Enhancing the integration of LC–MS-based approaches into routine clinical workflows holds great promise for revolutionizing personalized patient care.