Therapeutic Drug Monitoring of Targeted Anticancer Protein Kinase Inhibitors in Routine Clinical Use: A Critical Review

✅ 全文

靶向抗癌蛋白激酶抑制剂在常规临床使用中的治疗药物监测：一项批判性综述

作者 Evelina Cardoso; Monia Guidi; Benoı̂t Blanchet; Marie Paule Schneider; Laurent A. Décosterd; Thierry Buclin; Chantal Csajka; Nicolas Widmer 期刊 Therapeutic Drug Monitoring 发表日期 2019 ISSN 0163-4356 DOI 10.1097/ftd.0000000000000699 类型原创研究 (Original Research)

📄 中文摘要 Chinese Abstract

中文

口服靶向抗癌蛋白激酶抑制剂（PKIs）的治疗反应在患者之间差异显著，部分病例疗效不足，另一些病例则出现不可接受的药物不良反应。这种异质性源于多种原因，包括影响血药浓度的药代动力学（PK）变异、用药依从性波动以及癌细胞先天性或获得性耐药。对接受PKI治疗的肿瘤患者进行合理管理，需要各方共同努力以优化这些药物的使用，而这些药物的潜力可能尚未完全发挥。在此背景下，治疗药物监测（TDM）尤为适合通过基于循环药物浓度测定的剂量方案调整来优化PKIs的治疗活性。这种PK监测可防止患者暴露于无效或过度毒性的药物浓度水平。

📋 英文结构化总结 English Structured Summary

全文整理

Background:

Therapeutic response to oral targeted anticancer protein kinase inhibitors (PKIs) varies widely between patients, with insufficient efficacy in some cases and unacceptable adverse reactions in others. This heterogeneity arises from several causes, including pharmacokinetic (PK) variability affecting blood concentrations, fluctuating medication adherence, and constitutional or acquired drug resistance of cancer cells. The appropriate management of oncology patients with PKI treatments requires concerted efforts to optimize the utilization of these drug agents, which have probably not yet revealed their full potential. In this context, therapeutic drug monitoring (TDM) is ideally suited to optimize the therapeutic activity of PKIs through an adaptation of the dosage regimen based on the measurement of circulating drug concentration. This PK monitoring could prevent patients from being exposed to ineffective or unnecessarily toxic drug levels.

Methods:

An extensive literature review was performed on MEDLINE (up to April 2019) using MeSH terms such as "antineoplastic agents", "protein kinase inhibitors", "protein-tyrosine kinases", "protein-serine-threonine kinases", "pharmacology", "pharmacokinetics", and "drug monitoring", as well as various targeted anticancer names. The references listed in the relevant articles were also examined. The review addresses the limitations of current PKI utilization, the benefits of TDM, the prerequisites for TDM application, and practical aspects and pitfalls of TDM for PKIs.

Results:

The article provides criteria for determining PKIs suitable for TDM, including the availability of analytical methods, observational pharmacokinetic studies, PK-PD relationship analysis, and randomized controlled studies. It reviews the major characteristics and limitations of PKIs, such as the questionable selection of standard dosages based on the maximum tolerated dose (MTD) approach, which may cause severe toxicities. Most PKIs exhibit very different plasma concentrations among patients (interpatient PK variability) due to factors affecting absorption, distribution, metabolism, and excretion (ADME). The detection of low drug concentrations under a standard dosage indicates that TDM might prevent prolonged exposure to insufficient levels before it translates into cancer escape and drug resistance. Monitoring plasma concentrations allows precocious identification of overexposure, enabling timely dose reduction before adverse events occur.

Data Summary:

For lenvatinib, adverse events of grade 3 or higher occurred in 75.9% of patients under the recommended regimen, with dose reduction and discontinuation occurring in 68% and 14% of patients, respectively. The absorption of afatinib decreases by 39% with a high-fat meal, while those of ceritinib and vemurafenib increase by 73% and approximately 5-fold, respectively. A mutation in T790 of the EGFR kinase was found in approximately 50% of erlotinib, gefitinib, or afatinib resistant patients. Smokers present a 50% decrease in the erlotinib Cmin compared to non-smokers. The I-COME trial for imatinib included a small patient number (n = 56).

Conclusions:

Adaptation of PKIs dosage regimens at the individual patient level, through a rational TDM approach, could prevent oncology patients from being exposed to ineffective or unnecessarily toxic drug concentrations in the era of personalized medicine. TDM should constitute an additional approach to favor more effective, tolerable, and sustainable oral regimens, ultimately maximizing clinical benefit. However, routine monitoring of PKI concentrations is not yet advised due to the frequent lack of information about exposure-efficacy/toxicity relationships, the absence of therapeutic targets, and a lack of clear demonstration of its clinical benefit for most PKIs.

Practical Significance:

TDM might be particularly useful in situations such as poor treatment efficacy, severe or unexpected toxicities, recent co-administration of a drug potentially influencing PKI pharmacokinetics, suspected non-adherence to treatment, or altered patients' pathophysiological status (e.g., hepatic or renal impairments, advanced age, poor general status, and sarcopenia). An early concentration measurement needs to be performed shortly after treatment initiation to ensure optimal exposure, as the efficacy of PKIs is generally assessed only two to three months after initiation.

📋 中文结构化总结 Chinese Structured Summary

中文

背景:

方法:

在MEDLINE数据库中进行了广泛的文献检索（截至2019年4月），使用的MeSH主题词包括"抗肿瘤药物"、"蛋白激酶抑制剂"、"蛋白-酪氨酸激酶"、"蛋白-丝氨酸-苏氨酸激酶"、"药理学"、"药代动力学"和"药物监测"，以及各种靶向抗癌药物名称。同时查阅了相关文章中列出的参考文献。本综述探讨了当前PKI应用的局限性、TDM的益处、TDM应用的前提条件以及PKIs治疗药物监测的实际操作要点与潜在陷阱。

结果:

本文提供了确定适合进行TDM的PKIs的标准，包括分析方法的可获得性、观察性药代动力学研究、PK-PD关系分析以及随机对照研究。本文综述了PKIs的主要特征和局限性，例如基于最大耐受剂量（MTD）方法选择标准剂量的合理性存疑，该方法可能导致严重毒性反应。大多数PKIs在患者间表现出显著不同的血浆浓度（患者间PK变异），这是由影响吸收、分布、代谢和排泄（ADME）的因素所致。在标准剂量下检测到低药物浓度表明，TDM可能有助于防止患者在癌症逃逸和耐药发生之前长期暴露于不足浓度水平。监测血浆浓度可早期识别药物过度暴露，从而在不良事件发生前及时减少剂量。

数据总结:

对于仑伐替尼，在推荐方案下75.9%的患者发生了3级或更高级别的不良事件，分别有68%和14%的患者进行了剂量降低和停药。高脂饮食使阿法替尼的吸收降低39%，而使色瑞替尼和维莫非尼的吸收分别增加73%和约5倍。在约50%对厄洛替尼、吉非替尼或阿法替尼耐药的患者中发现了EGFR激酶T790位点突变。吸烟者的厄洛替尼Cmin较非吸烟者降低50%。伊马替克的I-COME试验纳入的患者数量较少（n=56）。

结论:

在精准医学时代，通过合理的TDM方法在个体患者水平调整PKIs的剂量方案，可防止肿瘤患者暴露于无效或过度毒性的药物浓度。TDM应作为一种补充手段，有助于实现更有效、更可耐受和更可持续的口服治疗方案，最终最大化临床获益。然而，由于多数PKIs缺乏暴露-疗效/毒性关系的相关信息、缺乏治疗靶点，且尚未明确证实其临床获益，目前尚不推荐常规监测PKI浓度。

实践意义:

TDM在以下情况下可能特别有用：治疗无效、严重或意外毒性、近期合并使用可能影响PKI药代动力学的药物、疑似治疗依从性不佳，或患者病理生理状态改变（例如肝肾功能不全、高龄、一般状况差和肌少症）。需要在治疗开始后不久尽早进行浓度监测以确保最佳暴露，因为PKIs的疗效通常仅在开始治疗后两到三个月才进行评估。

📖 英文全文 English Full Text

Therapeutic Drug Monitoring Publish Ahead of Print

DOI: 10.1097/FTD.0000000000000699 1

Therapeutic Drug Monitoring of Targeted Anticancer Protein Kinase Inhibitors in Routine

Clinical Use: A Critical Review

Evelina Cardoso, PharmD1,2; Monia Guidi, PhD1,2; Benoît Blanchet, PhD3,4; Marie Paule

Schneider, PhD2; Laurent A. Decosterd, PhD1; Thierry Buclin, MD1; Chantal Csajka, PhD1,2*;

Nicolas Widmer, PhD1,2,5,* 1Service of Clinical Pharmacology, Lausanne University Hospital and University of

Lausanne, Switzerland 2School of Pharmaceutical Sciences, University of Geneva, University of Lausanne,

Switzerland 3Department of Pharmacokinetics and Pharmacochemistry, Cochin Hospital, Assistance

Publique-Hôpitaux de Paris, Paris, France.

4UMR8638 CNRS, Pharmacy UFR, University of Paris Descartes, PRES Sorbonne Paris

Cité, France 5Pharmacy of Eastern Vaud Hospitals, Vevey, Switzerland

* Contributed equally – joint senior author

Address for correspondence:

Nicolas Widmer Service de Pharmacologie Clinique Centre Hospitalier Universitaire Vaudois

Rue du Bugnon 17 1011 Lausanne Switzerland ACCEPTED

Tel: +41 21 314 42 60 Fax: +41 21 314 42 66

Disclosure of Conflict of Interests Support for this work was solely provided by internal funds. TB, LAD, and CC are co- founders of Sotalya Inc., a company aiming to develop software tools to support TDM. MPS and CC have received a funding from the Accentus and Swiss Cancer Research foundations to perform research on the optimization of targeted anti-cancer therapies (HSR-4077-11- 2016). NW is a member of the Roche and Pfizer Swiss advisory board for hospital pharmacists, not related to the topic of this review and has received travel grants from Roche in 2018.

Word count: 6345 words Character count: 1’662/2000 characters

Abstract Purpose: Therapeutic response to oral targeted anticancer protein kinase inhibitors (PKIs) varies widely between patients, with insufficient efficacy of some of them and unacceptable adverse reactions of others. There are several possible causes for this heterogeneity, such as pharmacokinetic variability affecting blood concentrations, fluctuating medication adherence, and constitutional or acquired drug resistance of cancer cells. The appropriate management of oncology patients with PKI treatments thus requires concerted efforts to optimize the utilization of these drug agents, which have probably not yet revealed their full potential.

Methods: An extensive literature review was performed on MEDLINE on the pharmacokinetics, pharmacodynamics and therapeutic drug monitoring (TDM) of PKIs (up to

April 2019).

Results: This article provides the criteria for determining PKIs suitable candidates for TDM (e.g. availability of analytical methods, observational pharmacokinetic studies, PK-PD relationship analysis and randomized controlled studies). It reviews the major characteristics and limitations of PKIs, the expected benefits of TDM for cancer patients receiving them, as well as the prerequisites for the appropriate utilization of TDM. Finally, the paper discusses various important practical aspects and pitfalls of TDM for supporting better implementation in the field of cancer treatment.

Conclusion: Adaptation of PKIs dosage regimens at the individual patient level, through a rational TDM approach could prevent oncology patients from being exposed to ineffective or unnecessarily toxic drug concentrations in the era of personalized medicine.

Keywords: molecular targeted therapies, pharmacokinetics, variability, drug monitoring

Introduction In the last decade, considerable efforts have been invested in understanding the mechanisms driving cancer development.1, 2 Information related to proliferation, survival, differentiation, and migration in the cell is communicated through intracellular signaling pathways, many of which involve protein kinases. The overexpression or mutations of genes for these protein kinases may lead to constitutive activation of the corresponding pathways resulting in cancer development.3, 4 The detailed understanding of signaling pathways and the identification of dysregulated proteins involved in cancer pathophysiology have facilitated the development of rationally designed targeted drugs: protein kinase inhibitors (PKIs).5 These small molecules have dramatically changed the treatment and prognosis of certain cancers. They are currently

the mainstay of systemic treatment for hematological malignancies, such as chronic myelogenous leukemia (CML) or Philadelphia-positive acute lymphoblastic leukemia (Phi+LLA), and for several metastatic solid tumors including hepatocellular carcinoma (HCC), renal cell carcinoma (RCC), malignant melanoma, gastrointestinal stromal tumor (GIST), and defined molecular subgroups of lung cancer (activating mutations of the epidermal growth factor receptor [EGFR]). Moreover, considering their specificity for mutated kinase variants, certain PKIs are increasingly being experimented in cancers not controlled by conventional treatments, when genetic profiling reveals mutations in signaling pathways possibly amenable to targeted pharmacological interventions. This facilitates the large applicability of PKIs; however, adequate formalization and validation through appropriate studies is required.

Because of their oral administration, PKIs offer a greater autonomy and simpler outpatient care than cytotoxic chemotherapies and immunotherapies, thus improving the quality of life for patients.6 Nevertheless, the therapeutic response to PKIs varies widely between patients, with insufficient efficacy in some cases and unacceptable adverse reactions in others. There are several possible causes for this heterogeneity, such as pharmacokinetic (PK) variability, fluctuating medication adherence, and constitutive or acquired drug resistance of cancer cells.

Moreover, these oral-targeted therapies are highly expensive, which represents a serious burden for public healthcare systems.7 Therefore, the appropriate use of PKIs for the management of oncology patients requires definite developments in the personalization of indications, in the individualization of dosages, and in the precise piloting of therapies. All efforts must now be concentrated on strategies aiming at optimizing the treatment, i.e., decisions about drug choice and dosage regimen with regular monitoring and reevaluation to meet the specific needs of each cancer patient as far as possible. In this context, therapeutic drug monitoring (TDM) is ideally suited to optimize the therapeutic activity of PKIs through

an adaptation of the dosage regimen based on the measurement of circulating drug concentration. This PK monitoring could prevent patients from being exposed to ineffective or unnecessarily toxic drug levels.

This review first addresses the question: “Why do targeted anticancer PKIs require TDM?” by presenting some limitations of the current utilization of PKIs and the benefits of TDM for cancer patients treated with PKIs based on current scientific evidence. Second, the prerequisites of the TDM application will be reviewed. Finally, various practical aspects and pitfalls of TDM will be described focusing on the protein kinase inhibitors. The subjects addressed will be illustrated by examples of recently approved PKIs in cancer treatment.

This article focuses on the practical aspects of TDM in clinical practice, highlighting for instance the pitfalls related to sampling time, as well as to the protein binding and medication adherence issues. Further, the paper describes a formal three-step process for interpreting a

TDM output and for the subsequent dose individualization.

1. Methods An extensive literature review was performed on MEDLINE (up to April 2019) using especially the following MeSH terms to identify relevant studies and articles: “antineoplastic agents”, “protein kinase inhibitors”, “protein-tyrosine kinases”, “protein-serine-threonine kinases”, “pharmacology”, “pharmacokinetics” and “drug monitoring”, as well as various targeted anticancer names. The references listed in the relevant articles were also examined.

2. Why do targeted anticancer PKIs require TDM?

Limitations affecting the clinical development of PKIs

Selection of standard dosages Recommended doses for traditional chemotherapeutic drugs are usually determined by dose- escalation during phase I trials conducted in cancer patients; the standard dose generally corresponds to the maximum tolerated dose (MTD), i.e., the highest dose still associated with an acceptable level of toxicity. Such an approach of dose selection is appropriate for cytotoxic chemotherapies, with the assumption that higher doses would achieve higher effects on growing tumor cells.8 On the contrary, for new oral anticancer drugs that have specific targets and are administered daily, several authors argue that the MTD approach is no longer appropriate.9-11 The inhibition of cellular signal transduction by PKIs follows a classical sigmoid log-concentration-response curve that reaches the maximum value beyond a certain level of exposure. Dosage regimens based on MTD may thus indicate higher doses than that required for maximum efficacy, which may cause severe toxicities with consequent dose reductions and/or treatment interruptions.9-11 Most approved PKIs have been marketed using the MTD approach for the selection of the standard dose, without any defined concentration- response relationship.12 This is, for example, the case of lenvatinib approved by the FDA as treatment for locally recurrent or metastatic, radioactive iodine-refractory differentiated thyroid carcinoma (DTC). The dose of 24 mg once daily has been determined as the MTD in a phase I dose-escalation study in solid cancer patients.13 However, the adverse events of grade 3 or higher occurred in 75.9% of patients under the recommended regimen in a phase

III study. Dose reduction and discontinuation of treatment for adverse events occurred in 68% and 14% of patients, respectively.14 These results cast doubt on the selection of the optimal dosage of lenvatinib, and postmarketing trials are still required to investigate whether lower doses could provide a better safety profile while maintaining efficacy11, 15 and reducing costs.

Thus, there is a definite need to optimize dose selection before entering the drug onto the market by considering alternative dose-finding strategies.9, 11, 16, 17 Efficacy rather than safety should govern dosing decisions, with a focus on exposure-response modeling.18

Appraisal of interpatient PK variability Most PKIs exhibit very different plasma concentrations among patients receiving the standard dosage, corresponding to interpatient PK variability. Such variability results from several factors (e.g., patho-physiological, pharmacogenetic, and environmental) affecting the

PK processes of absorption, distribution, metabolism, and excretion (ADME).17, 19

Absorption and metabolism are both processes mostly affected by variability in the disposition of PKIs.

The shift to oral anticancer treatment requires investigating the influence of food intake on the bioavailability of PKIs; this impact is usually assessed under three different food conditions: fasted state, light meals, or high fat meals. The absorption of PKIs can be either unaltered by concomitant food intake (e.g., cobimetinib, ponatinib, or ribociclib), or significantly affected by a high-fat meal. For example, the area under the concentration-time curve (AUC) of afatinib decreases by 39%, and, inversely, those of ceritinib20 and vemurafenib21 increase by 73% and approximately 5-fold, respectively, compared to the values in the fasted state. It is worth noting that the current recommendations on vemurafenib intake with respect to food differ between the Swiss Agency for Therapeutic Products (Swissmedic22) and American (FDA23) or European (EMA24) authorities. Swissmedic recommends taking vemurafenib in the fasted state to avoid overexposure, whereas the latter agencies suggest an intake independent of food, with the warning to avoid taking the drug systematically without food to reduce the risk of underexposure. The absorption of other

PKIs (e.g., bosutinib, gefitinib, and pazopanib) is modulated by gastric pH, which affects the

solubility of the molecule.25, 26 Indeed, an increase in the gastric pH due to co-administration of antacid therapy—frequently used by cancer patients27—can significantly decrease the solubility of some PKIs, which leads to reduced absorption and potential treatment failure. A large study on the non-small-cell lung cancer (NSCLC) patients treated with erlotinib demonstrated poor outcomes in patients receiving concomitant acid-reducing agents, which was attributed to a decrease in erlotinib bioavailability.28 Further, a significant decrease in treatment efficacy was observed when pazopanib was associated with antacid therapy.29

Almost all PKIs are primarily metabolized by liver enzymes of the cytochrome P450 (CYP) family, mostly by the CYP3A4 isoenzyme, whose activity can be strongly influenced by co- administration of inhibitors or inducers. This results in clinically relevant variations in PKIs plasma concentrations. 30, 31 Besides the dose of the CYP inhibitor or inducer, the extent to which such drug-drug interactions affect the exposure to PKIs depends on their potency and the number of pathways contributing to the metabolism of PKIs. For example, lenvatinib is metabolized through multiple pathways, including oxidation mediated by CYP3A4 and aldehyde oxidase, and glutathione conjugation.22 The co-administration of the potent CYP3A inhibitor ketoconazole had only a weak impact on lenvatinib exposure (AUC increase by

15%),32 confirming that alternative pathways can compensate for the reduced CYP3A4 activity. Any increase in the number of medications prescribed, for example in elderly cancer patients with comorbidities, increases the risk of drug interactions.33, 34 In addition to polymedication, complementary and alternative medicines (e.g., herbal products, minerals, vitamins, and antioxidants) and other over-the-counter drugs are being widely used by cancer patients.35, 36 Oncologists are often unaware of such use, probably because patients are reluctant to disclose such consumptions and tend to perceive these “natural products” as safe, ignoring that these substances can significantly affect PKI concentrations through an impact on CYP activity.37-39

The restrictive selection of patients limits clinical trials from assessing interpatient PK variability during drug development properly. Therefore, a larger variability in circulating concentration is expected in “real-world” cancer patients, who are often elderly, with worse general status and more comorbidities and concomitant medications than patients included in the stringent frame of clinical trials.40-42 According to Talarico et al.,43 adults older than 75 years represent less than 10% of patients included in published trials, while 26% of cancers are diagnosed after this age.44 Fortunately, this issue starts to be considered in current drug development programs.45

Altogether, a better appreciation of the variability of PKIs disposition is a prerequisite to limit both their toxicity and their lack of efficacy (potentially favoring the selection of resistant cell clones) as described further in the following sections.46

Management of adverse events Because of their relatively high target specificity, PKIs were expected to present less toxicity than traditional cytotoxic chemotherapies. However, PKIs are by no means devoid of the adverse effects of variable incidence and severity. Some undesirable reactions result from the inhibition of physiological functions endorsed by the targeted kinases in normal cells, which makes them theoretically predictable.47, 48 Skin rashes, in particular acneiform eruptions, represent a very common, class-specific toxicity induced by EGFR inhibitors 49 that are used in the treatment of NSCLC (erlotinib, gefitinib, afatinib, and osimertinib), breast cancer (lapatinib), and medullary thyroid cancer (vandetanib). Because EGFR is over-expressed and/or over-activated in various tumors, it is a rational target for therapeutic interventions.

However, this kinase is also present in normal epithelial tissues, and the inhibition of EGFR- mediated signal pathways lead to cutaneous inflammation and development of skin lesions, such as papules and pustules, known to have a major effect on the quality of life of the

patient.50 Interestingly, skin rashes seem to correlate positively with EGFR inhibitor efficacy, and they have been therefore suggested as a surrogate marker to monitor therapeutic response.51, 52 Another well-known specific toxicity is hypertension with proteinuria induced by the inhibition of the vascular endothelial growth factor (VEGF) signaling pathway by anti- angiogenic agents, such as sorafenib, sunitinib, axitinib, and lenvatinib.53 A further example of a class-wide adverse reaction is the hematological toxicity (neutropenia, leukopenia, anemia, and thrombocytopenia) induced by recently developed specific cyclin-dependent kinase 4 and 6 (CDK4/6) inhibitors (palbociclib and ribociclib), used against metastatic breast cancer.54 Given that most of these adverse effects are dose-related, they can be managed clinically by dose reduction or temporary treatment interruption. However, the administration of incriminated PKIs is often discontinued—definitely in cases of severe and unbearable adverse reactions55—as prescribing oncologists are reluctant to decrease drug dosages, which could result in insufficient concentration exposure.

Coping with resistance The administration of PKIs is not infrequently associated with a lack of therapeutic response in a fraction of treated patients, presumed to result from a resistance of cancer cells to these molecules. The resistance is called “primary” or “constitutive” if the patient shows no response from the outset of treatment, and “secondary” or “acquired” when he or she is found escaping to cancer control after an initial phase of response. Several mechanisms of resistance to PKIs56, 57 have been described: a) Resistance induced by genetic alterations (mutations or gene amplification) of the targeted kinase.58 The proliferation of tumor cells under the selective pressure of PKIs provides an evolutionary advantage to cells whose kinase has acquired additional genetic modifications enabling them to escape from pharmacological inhibition. The

most common alterations are mutations affecting the binding domain of the targeted kinase, leading to a reduced affinity of the PKI for its target. For example, a mutation in T790—a key position in the binding site of EGFR kinase—was found in approximately 50% of erlotinib, gefitinib, or afatinib resistant patients, while it is rarely detected in treatment-naïve patients.57, 59, 60 b) Resistance induced by upregulation of alternative kinase pathways (bypass tracts).

The tumor cells can acquire resistance through the activation of downstream signaling pathways, making the PKI progressively ineffective.56 c) Resistance induced by drug transport proteins. Some PKIs are substrates of one or several ATP-binding cassette (ABC) transporters, drug efflux pumps such as the P- glycoprotein (P-gp), and the breast cancer resistance protein (BCRP). They can also be carried by solute liquid carrier (SLC) transporters, such as organic anion (OATs,

OATPs) or cation transporters (OCTs). Resistance against PKIs can result from the overexpression of these transporters in cancer cells,56, 61, 62 causing a decrease in cellular drug levels, and therefore, in the PKIs anticancer efficacy.

The development of a resistance against PKIs is problematic since it leads to the progression of the tumor and to a restriction of the panel of effective treatments. It is therefore important to prevent and overcome resistant mutants with different strategies: a) Ensuring appropriate systemic exposure to PKIs. The occurrence of resistance- inducing mutations is favored by exposure to sub-therapeutic concentrations or intermittent drug-free periods and insufficient concentration to effectively block cell replication, which favors mutation-selection cycles in the tumor.63 This phenomenon, amply demonstrated with all types of chemotherapies, targeted from bacteria to pest animals, calls for optimizing the level and the regularity of drug concentration exposure throughout the treatment.

b) Resorting to next-generation PKIs to overcome acquired resistance. Following the success of the second and third generation agents (nilotinib, dasatinib, bosutinib, and ponatinib) in CML patients harboring imatinib-resistant BCR-ABL mutations, a similar approach was adopted for EGFR-mutant NSCLC, which lead to the development of next-generation EGFR inhibitors such as osimertinib and rociletinib.64 c) Using PKI combinations that target simultaneously several kinases. This approach appears particularly successful in melanoma treatment with the association of BRAF and

MEK inhibitors (vemurafenib/cobimetinib, dabrafenib/trametinib, and encorafenib/binimetinib) targeting two different kinases in the Ras-Raf-MEK-ERK signaling pathway.65, 66 This paradigm will be extended to others cancers, e.g., lenvatinib with everolimus in metastatic RCC.67

Prospect of clinical benefits from TDM Several of the above-mentioned limitations in the utilization of PKIs can be overcome by optimizing dosage regimens at the individual patient level through a rational TDM approach.

These drugs feature various characteristics that represent classical criteria for the implementation of a TDM program. To date, most if not all available PKIs reached the market with fixed standard dose recommendations. The labelled dosage selected for certain

PKIs is questionable; however, the generalized absence of consideration for their important interpatient PK variability, which results in a striking heterogeneity in plasma concentration profiles achieved by a standard dosage, is even more questionable.

The detection of low drug concentrations under a standard dosage of PKIs indicates that

TDM might prevent prolonged exposure to insufficient concentration levels, along with an adherence-enhancement program, before it translates into cancer escape from control and the occurrence of drug resistance.

Further, adverse effects are major contributors to treatment discontinuation, and as there are only a few therapeutic options, a timely management of toxicity is crucial for the patient to ensure long-term PKIs self-management.68 Monitoring the plasma concentrations of patients allows a precocious identification of overexposure, and a timely and fact-based dose reduction can be undertaken before the occurrence of adverse drug events, which helps improve persistence on treatment and preventing early switches to other molecules and decreasing the anxiety of the patient.

For all these reasons, in addition to the fact that cancer is a life-threatening disease as well as to the economic pressure of such treatments on public health systems, TDM should constitute one additional approach to favor more effective, tolerable, and sustainable oral regimens.

Ultimately, this should lead to maximize the clinical benefit for the patient.

3. What are the prerequisites for TDM application to PKIs?

The appropriate TDM application requires standardized PKI measurement methods and prior knowledge on the PK properties, PK/PD relationships, therapeutic targets, as well as the level of evidence of the benefit provided by this approach for the various PKIs.

Analytical methods The access to a reliable analytical method for the quantification of PKIs in biological fluids is the first prerequisite for TDM. Currently, it is possible to develop quantification techniques for molecules with low molecular weights such as that of PKIs. The most commonly used method is liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS),69-71 which demonstrates excellent performances compared to other available bioanalytical methods. The main benefits of LC-MS/MS are its speed, selectivity, and sensitivity. By reducing the analysis time, in particular with ultra-performance liquid chromatography (UPLC), this technology allows an easy routine application of a real-time TDM.72 It also

covers a large clinically relevant range of plasma concentrations to support routine TDM and

PK/PD studies.

Owing to the selectivity provided by tandem mass spectrometry, the simultaneous measurement of multiple structurally unrelated PKIs is possible in a short analytical time.

The multiplex approach allows for the optimization of laboratory resources and turn-around time, with a unique sample extraction procedure and a single chromatographic run. This approach is particularly relevant when patients are treated with a PKIs combination, such as vemurafenib/cobimetinib or dabrafenib/trametinib, which are prescribed for unresectable or metastatic BRAF-mutated melanoma.73 For example, several LC-MS/MS multiplex methods for PKIs measurements have been recently developed and validated.74, 75 One limitation to such combined analyses are the major differences in concentration levels, as for pazopanib and vemurafenib, which can thus be included with difficulty in the same analytical run.

With its high sensitivity, the LC-MS/MS is also a potential method for detecting extremely low drug levels. For example, low concentrations of PKIs are expected in the cerebrospinal fluid (CSF) because of their limited diffusion across the blood-brain-barrier (BBB) and their high plasma protein binding. Since many reports provide evidence on the efficacy of these treatments on the development of central nervous system metastases commonly disseminated from melanoma, breast, and lung cancers, the quantification of such low concentration levels in profound compartments is of clinical relevance.76-80 The quantification of drugs in the CSF might also be of interest to better understand the BBB permeability and to calculate drug penetration rates.81, 82 As plasma and CSF are different complex biological matrices, it may be appropriate to develop a specific method for the quantification of drugs in small CSF volumes.83-85 However, stable isotopically labeled internal standards of target PKIs enable circumventing the differences in biological matrices when, for practical or ethical reasons, the

actual blank matrix is not available for the calibration sample preparation, and thus, there is need to use proxy matrices instead.

Dried blood spot (DBS) is another recent sample collection technique successfully applied in clinical practice for PKIs TDM.86-88 The DBS sampling 89 requires only a small volume disposed onto a paper card, which is easily compatible with the analytical sensitivity of the

LC-MS/MS method. Moreover, this approach offers the advantages of being less invasive than conventional venous sampling for cancer patients with difficult to access peripheral vessels caused by frequent blood sampling and/or chemotherapy administration. However, this sampling method raises several well-described bioanalytical challenges.90, 91

Observational pharmacokinetic studies The pharmacokinetics of PKIs in cancer patients is investigated in the early stages of drug development. Phase I clinical studies aim at providing PK profiles and parameters for different doses, but these studies have the limitation of being conducted with small groups of homogenous cancer patients who are not representative of the diversity of “real-world” cancer populations. Hence, in modern clinical development and in postmarketing surveillance, the characterization of PKIs concentrations over time based on a population pharmacokinetic (popPK) approach is being increasingly used (e.g., with afatinib,92 ceritinib,93 and cobimetinib94).

The popPK studies that use non-linear mixed effects regression models are clearly superior to traditional PK studies for different reasons.95 PopPK can analyze data gathered from a large number of patients, and blood samples may be collected at various times following drug administration (sparse sampling) from patients taking different doses in routine clinical care (observational analysis). The absence of a restrictive design facilitates the participation of outpatients with cancer. PopPK modeling aims to quantify the PK parameters in the study population, as well as the variability in plasma drug levels within (intra-) and between (inter-)

individuals. It also aims at identifying and assessing the contribution of various potential factors affecting drug concentrations. Moreover, a popPK model can be used to simulate percentile curves describing the expected concentration-time profile of drugs under several scenarios (e.g., standard or alternative dose regimen, suboptimal adherence patterns or presence of drug-related interaction). This model is essential for Bayesian therapeutic monitoring.

A popPK analysis found higher lenvatinib exposure in patients with low body weight [BW] (induced by a modification of its distribution and elimination) and it indicated the early dose reduction or discontinuation due to toxicity in 74% of patients with advanced hepatocellular cancer (HCC).96, 97 A dose adaptation is thus formally recommended according to BW in this population (12 mg once daily for patients with a BW ≥ 60 kg and 8 mg once daily for patients with a BW < 60 kg). 98

PK meta-analyses In the field of clinical pharmacokinetics, meta-analyses approaches can be used to combine the results of multiple PK studies.99 Indeed, when sufficient quantitative PK parameters are available from a systematic literature search, even if not uniformly reported, a meta-analysis can derive average PK parameters from a larger set of patients than individual studies. In the era of big data, with the growing number of PK analyses published in the literature, this approach could represents an innovative approach for a rational TDM development of

PKIs.100, 101 Ultimately, the developed “meta-models” would be integrated in TDM software and can guide the choice of the TDM model and strengthen the dosage regimen adaptation procedure.102

PK-PD relationship analysis The application of TDM is mostly effective if the plasma concentrations are related to the therapeutic response of the drug (efficacy or toxicity).

The PK parameters most commonly correlated with the therapeutic response are the trough concentration (Cmin), and less frequently, the AUC.103 Moreover, some PKIs have pharmacologically active metabolites, which contribute substantially to the oncological effect. In such cases, the concentrations of metabolites must be considered for the PK-PD analysis. For example, recent analyses of sunitinib,104 osimertinib105, and dabrafenib106 integrated their respective metabolites (SU12662, AZ5104 and hydroxy-dabrafenib), both in the popPK and in the population PK-PD models.

A PK-PD analysis can be performed by population models,107 considering potential factors influencing the therapeutic response to the drug. In the PK-PD model of palbociclib investigating the relationship between its concentration and neutropenia, the gender of the patient and the baseline albumin level were integrated in the model as significant contributors to the variability of absolute neutrophil count at the baseline.108 Numerous other studies have investigated the PK-PD relationships of PKIs; however, for the most recent drugs e.g., ceritinib, cobimetinib, dabrafenib, lenvatinib, osimertinib, and ribociclib), data are lacking with regard to TDM application. A review recently reported the exposure-efficacy/toxicity relationships of more than twenty PKIs approved by the FDA.109 However, the results of the

PK-PD relationship studies often vary for a given PKI, which prevent from the robust identification of therapeutic targets. As described by Kim et al.,110 such differences could be explained by the complexities of these analyses, by the choice of the PK/PD outcomes, and by a potential tumor-specific PK-PD relationship, which is possibly also dependent on the kinase mutational status. For the specific TDM of PKIs, therapeutic windows are not formally established, and therefore, the common recommendation is to reach a minimal PKI concentration threshold to improve treatment efficacy.

Randomized controlled studies Currently, TDM is applied to PKIs based on the safety and clinical feasibility of the approach111-113 despite a definite lack of clinical evidence for most PKIs, except for imatinib,114, 115 as explained below. The clinical benefit of routine TDM should indeed be formally demonstrated by prospective, randomized controlled clinical trials, comparing predefined clinical efficacy and tolerability endpoints between a group of patients receiving the recommended standard dose and another group receiving an individually adjusted dose by

TDM.116 Large-scale, evidence-based assessment of TDM of life-saving therapies such as PKIs using randomized controlled designs is warranted.117 However, in a real clinical setting, these types of studies are often perceived as unethical by patients and practitioners because of the denied access to potentially life-saving strategies in the control group. Therefore, alternative study designs are required to assess TDM benefits.118 For instance, the Imatinib COncentration

Monitoring Evaluation (I-COME) trial was conducted to prospectively establish scientific evidence for the routine application of TDM to prevent unfavorable outcomes.114 The I- COME trial could not formally demonstrate the benefit of “routine TDM” for imatinib, especially due to the small patient number (n = 56) and limited prescriber adherence to dosage recommendations. However, a trend was found indicating that the patients for whom advised dosage adaptations were actually implemented by physicians more often met not only the target concentrations, but also a predefined composite score of combined outcomes (efficacy, tolerance, and persistence).

However, the sample size of studies quantifying the clinical benefit of TDM is often small, and therefore, there is an urgent need for a larger number of patients to reach sufficient statistical power. For example, a prospective study aiming at optimizing the pazopanib dose through TDM was terminated prematurely due to lack of patient recruitment.119

4. How to implement TDM of PKIs in clinical practice?

The routine monitoring of PKI concentrations is not yet advised due to the frequent lack of information about exposure-efficacy/toxicity relationships, the absence of therapeutic targets, as well as a lack of clear demonstration of its clinical benefit.109 However, it might be particularly useful in several situations such as poor treatment efficacy, or severe or unexpected toxicities; recent co-administration of a drug potentially influencing the PKI pharmacokinetics; suspected non-adherence to treatment; altered patients’ pathophysiological status (e.g., hepatic or renal impairments); or vulnerability (e.g., advanced age, poor general status, and sarcopenia). Considering that the efficacy of PKIs is generally assessed only two to three months after its initiation, an early concentration measurement needs to be performed shortly after treatment initiation; the steady state is achieved for ensuring an optimal exposure (i.e., plasma levels in the expected range, based on available PK data). This approach could be combined with the pharmacodynamic monitoring of PKIs, based on the measurement of biomarkers, which is a methodology under development as well.120

Time of sampling Ideally, the sample should be collected once the steady state is achieved (i.e., 5 half-lives).

This implies a variable waiting time, e.g., for more than a week for ceritinib (half-life of 41 h121) but only for 2 days for ibrutinib (half-life of 4-6 h122).

The time of the last PKI intake and the time of blood sampling are essential information to record the interpretation of the result. Trough concentration (Cmin) is the usual target used in clinical practice to assess the adequacy of drug exposure.109 However, since oral targeted anticancer drugs are mostly used by outpatients, blood samples are not always drawn at trough, but rather at defined clinical point of care, i.e., at unselected times after PKI intake.

Therefore, TDM results cannot often be directly compared with the Cmin reference target; however, they require a PK extrapolation to the trough level. Wang et al.123 proposed an

algorithm to extrapolate imatinib Cmin based on the elimination constant of the drug. This linear extrapolation approach is often used for other PKIs124, 125 based on the assumption that they exhibit linear elimination (i.e., no profound compartment) and that samples have been collected during the drug elimination phase. However, because of intra-patient variability, such assumptions might not be valid, precluding the method robustness.

An alternative approach is the Bayesian maximum a posteriori (MAP) estimation. Based on popPK models, measured plasma concentration(s), and relevant patients’ characteristics, this approach allows the estimation of individual PK parameters and the prediction of drug concentrations at any time point of the dosing interval. With such a method, AUC can also be derived from a single measured concentration, which might be of interest in some situations.126 With its good predictive performance in terms of bias and precision, the

Bayesian optimization method easily assists the TDM of anticancer drugs,116, 127, 128 and it can be further developed for numerous current and upcoming PKI generations. These extrapolation approaches offer a flexibility in blood sampling times, thus facilitating the translation of TDM in clinical practice.103

TDM interpretation and dose individualization The interpretation of a TDM output and the subsequent dose individualization should follow a three-step established process71 (Figure 1).

The first step implies the evaluation of the normality of the measured concentration by considering the drug dosage regimen and defined patient’s characteristics. The

“expectedness” of this concentration is assessed by comparison either with population-based reference percentile curves or with the average exposure derived from traditional PK analyses. If the measured concentration is lower or higher than what is expected, a comprehensive search for the origin of this variability (e.g., drug-interaction, adherence issue) has to be initiated. In presence of a known factor influencing the PKI

pharmacokinetics, the knowledge of the magnitude of its impact is useful for the interpretation and subsequent dose adjustment. For example, because smokers present a 50%- decrease in the erlotinib Cmin compared to non-smokers, a 2-fold maximum increase of the dose regimen could be carefully considered in smoker patients.129 The DDI predictor is a web tool that can be used in clinical practice to quantitatively predict the drug-drug interaction (http://www.ddi-predictor.org130).

The second step consists on evaluating the “suitability” of the concentrations, meaning whether the measured concentration is adequate considering the known therapeutic target. In absence of an evidence-based therapeutic target, the concentration predicted to inhibit the targets (IC) based on preclinical data could be used as references for the interpretation, as proposed by Verheijen et al. for some PKIs.109 The mean or median population exposure at the recommended dose can also serve as an alternative for an effective PKIs concentration.

Indeed, a recent analysis emphasized that the 81.7% TDM-based concentration targets matched with the average population PKI exposure.109 Furthermore, if the measured concentration is appropriate according to the therapeutic target, but the patient presents a lack of clinical efficacy, it is important to search for PKI resistance. In such a case, a switch of treatment has to be considered.

The third step is the adjustment of the current drug dosage, whenever necessary, to reach the therapeutic target (a posteriori adaptation). This can be performed with some manual strategies, such as the simple rule of three between dose and concentration, or by using TDM- guided software (see below). Ideally, the dose individualization of PKIs should also be carried out before the treatment starts (a priori adaptation). Indeed, in order to estimate the initial dose necessary to reach the therapeutic target, Bayesian approaches based on a popPK model could be used to estimate the individual PK parameters, considering the patients’ characteristics known to affect the PKI pharmacokinetics. However, because of unexplained

residual variability in each model, it is essential to closely monitor subsequent concentrations at steady-state and adjust the dose to reach the therapeutic target in a stepwise approach.116

Plasma protein binding pitfall Except for a few exceptions, PKIs are highly bound to circulating plasma proteins (usually at least 90%), mostly to the albumin and the α-acid glycoprotein (AGP).131 However, due to technical and cost limitations, routine analytical methods usually measure the total plasma concentration of PKIs, which includes the (major) plasma protein-bound fraction, in equilibrium with (generally low) free PKIs. Nevertheless, only the latter fraction is likely to penetrate the cell and thus exert its pharmacological action.

Moreover, the level of the plasma protein binding could have a significant impact on the distribution and clearance of PKIs, leading to drug concentration variability, as observed for imatinib132 and erlotinib, 133 as well as for recently developed PKIs, such as axitinib134 and osimertinib.105 This additional level of complexity has to be considered to correctly interpret

TDM outputs based on total plasma concentration measured in patients with fluctuating protein levels; for instance, a high level of AGP in the case of an acute phase of cancer 135 or in hypoalbuminemia in case of hepatic impairment or malnutrition. For a PKI such as imatinib with a high AGP binding and a rather low hepatic extraction, an increase of the AGP level may lead to an increase in total imatinib plasma concentration, without change in the free concentration, thus resulting in potential inadequate interpretation and unsound dosage reduction. 132

The quality of TDM interpretation would therefore be improved by assessing the free PKI concentrations in situations where protein bindings are severely altered. Free PKI concentrations could be measured with analytical methods (i.e., generally ultrafiltration136) or they could be extrapolated with mathematical models based on total concentrations.137

Medication adherence issue Medication adherence is defined as the process by which patients take their medications as prescribed. It is characterized by three components: the initiation, implementation, and discontinuation of the treatment. The implementation is defined as the extent to which a patient takes the right drug at the right dosage with the right timing day-by-day, and persistence is delimited by the time period from treatment initiation until complete discontinuation.138

Medication adherence is a major determinant in the therapeutic success of oral PKIs.139, 140

Although cancer is life-threatening, long-term adherence to oral targeted anticancer therapy can be suboptimal.141-143 Several complex, dynamic, and interrelated determinants affect medication adherence to PKIs,141, 144 such as treatment-related adverse events or complexity, alternate treatment beliefs of the patient, anxiety, depression, lack of social support, or economic burden. Several medication adherence programs currently explore interventions supporting patients in optimal PKI self-management.145-150

Medication adherence is an important patient behavior to consider in TDM interpretation, as it represents a potential source of bias. Indeed, a deviation from a prescribed regimen is a source of variation in drug exposure (under- or over-exposure), potentially leading to an inappropriate dose adaptation. Sound adherence monitoring would eliminate doubts about poor adherence in case of underexposure and about over-adherence before planned medical visits (“white coat adherence”) in the case of overexposure.

Moreover, because conditions surrounding PKI intake may vary on a daily basis, which can influence measured concentration, systematic information should be collected from the patient during blood samplings, such as the date and hour of the last PKI intake, the time and type of food, as well as the actual day of the treatment for PKIs with a cyclic dosing schedule (e.g., sunitinib, regorafenib, cobimetinib, palbociclib, and ribociclib).151. It is noteworthy that

the knowledge on the impact of adherence to PKIs on plasma concentration in routine TDM programs is still missing. On the other side, pharmacokinetically-based estimations of the adherence profiles of patients are under investigations for imatinib and other anticancer drugs from the perspective of estimating individual adherence from a single TDM sample.152, 153

Computer assistance TDM is perceived as a complex approach in clinical practice due to the critical amount of information to gather and the complexity of the calculations necessary to perform state-of- the-art dosage adaptations. Owing to a recent growing interest, several software applications have been developed to facilitate the daily practice of TDM154 by supporting clinicians in the interpretation of a drug concentration and by guiding dose adjustment decisions.

MwPharm++, InsightRX, and more recently Tucuxi155 developed by our group in Lausanne, are examples of recent TDM-guided software. By integrating popPK data and PK/PD parameters, this software can both evaluate the expectedness and the suitability of the measured concentration considering drug and patient specific characteristics, and it can propose a dosage adaptation through Bayesian calculations. Such software has the possibility to interface with hospital information systems, such as electronic medical records (e.g., patient clinical history and laboratory information). Moreover, they should be intuitive, robust, and user-friendly, with interactive graphical displays.

5. Conclusion A new era of cancer therapy has emerged in the last decade, with oral targeted anticancer drugs, especially PKIs, directed against cancer-specific molecules and signaling pathways.

Nevertheless, drug resistance, persistence of cancer stem cells, and adverse drug effects still limit their ability to stabilize or cure malignant diseases in the long term. Despite a significant

inter-patient PK variability, PKIs are essentially licensed at fixed doses. However, they are suitable candidates for dose individualization through a TDM program because of their large pharmacokinetic variability across patients and their sensitive exposure balance from efficacy to toxicity.

Oncology patients, as well as health systems, certainly deserve further optimization and individualization in the piloting of these drugs in the context of emerging personalized medicine efforts.156 In particular, prospective randomized controlled trials to assess the level of evidence of the TDM of new PKIs are eagerly warranted, as the development of intelligent, user-friendly computer systems is promising. Such computer systems will assist clinicians in their TDM practice for overcoming known practical issues, such as the random measurement of drug concentration, real-time patient adherence to PKIs, and plasma protein binding simulation.

Appropriate TDM clinical research and practice should thus be included in modern personalized cancer care,157 together with pharmacodynamic monitoring and pharmacogenetic approaches. As “every oncology patient is different,” such optimization should emphasize that the administration of “the right anticancer drug at the right dosage to the right patient” could be made possible in routine clinical settings.

6. Acknowledgments Support for this work was solely provided by internal funds. TB, LAD, and CC are co- founders of Sotalya Inc., a company aiming to develop software tools to support TDM. MPS and CC have received a funding from the Accentus and Swiss Cancer Research foundations to perform research on the optimization of targeted anti-cancer therapies (HSR-4077-11- 2016). NW is a member of the Roche and Pfizer Swiss advisory board for hospital

pharmacists, not related to the topic of this review and has received travel grants from Roche in 2018.

References 1. Hanahan D, Weinberg RA. The hallmarks of cancer. Cell. 2000;100(1):57-70.

2. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell.

2011;144(5):646-74.

3. Giamas G, Man YL, Hirner H, et al. Kinases as targets in the treatment of solid tumors.

Cellular signalling. 2010;22(7):984-1002.

4. Ardito F, Giuliani M, Perrone D, et al. The crucial role of protein phosphorylation in cell signaling and its use as targeted therapy. Int J Mol Med. 2017 Aug 1;40(2):271-80.

5. Zhang J, Yang PL, Gray NS. Targeting cancer with small molecule kinase inhibitors. Nat

Rev Cancer. 2009;9(1):28.

6. Ruddy K, Mayer E, Partridge A. Patient adherence and persistence with oral anticancer treatment. CA: A Cancer J Clin. 2009;59:56-66.

7. Newcomer LN. Insurers and 'targeted biologics' for cancer: a conversation with Lee N.

Newcomer. Interview by Barbara J. Culliton Health affairs (Project Hope).

2008;27:w41-51.

8. Salzberg M. First-in-human phase 1 studies in oncology: The new challenge for investigative sites. Rambam Maimonides Med J. 2012;3:e0007.

9. Nie L, Rubin EH, Mehrotra N, et al. Rendering the 3 + 3 design to rest: More efficient approaches to oncology dose-finding trials in the era of targeted therapy. Clin Cancer

Res. 2016;22:2623-2629.

10. Postel-Vinay S, Arkenau HT, Olmos D, et al. Clinical benefit in phase-I trials of novel molecularly targeted agents: does dose matter? Br J Cancer. 2009;100:1373.

11. Janne PA, Kim G, Shaw AT, et al. Dose finding of small-molecule oncology drugs:

Optimization throughout the development life cycle. Clin Cancer Res. 2016;22:2613- 2617.

12. Bullock JM, Rahman A, Liu Q. Lessons learned: Dose selection of small molecule- targeted oncology drugs. Clin Cancer Res. 2016;22:2630-2638.

13. Boss DS, Glen H, Beijnen JH, et al. A phase I study of E7080, a multitargeted tyrosine kinase inhibitor, in patients with advanced solid tumours. Br J Cancer. 2012;106:1598- 1604.

14. Schlumberger M, Tahara M, Wirth LJ, et al. Lenvatinib versus placebo in radioiodine- refractory thyroid cancer. The New Eng J Med. 2015;372:621-630.

15. Nair A, Lemery SJ, Yang J, et al. FDA approval summary: Lenvatinib for progressive, radio-iodine-refractory differentiated thyroid cancer. Clin Cancer Res. 2015;21:5205- 5208.

16. Parulekar WR and Eisenhauer EA. Phase I trial design for solid tumor studies of targeted, non-cytotoxic agents: theory and practice. J Nat Cancer Inst. 2004;96:990-997.

17. Mathijssen RH, Sparreboom A, Verweij J Determining the optimal dose in the development of anticancer agents. Nature Rev Clin Oncol. 2014;11:272-281.

18. Wang Y, Booth B, Rahman A, et al. Toward greater insights on pharmacokinetics and exposure-response relationships for therapeutic biologics in oncology drug development.

Clin Pharmacol Ther. 2017;101:582-584.

19. Rowland A, van Dyk M, Mangoni AA, et al. Kinase inhibitor pharmacokinetics: comprehensive summary and roadmap for addressing inter-individual variability in exposure. Expert Opin Drug Metab Toxicol. 2017;13:31-49.

20. Lau YY, Gu W, Lin T, et al. Effects of meal type on the oral bioavailability of the ALK inhibitor ceritinib in healthy adult subjects. J Clin Pharmacol. 2016;56:559-566.

21. Ribas A, Zhang W, Chang I, et al. The effects of a high-fat meal on single-dose vemurafenib pharmacokinetics. J Clin Pharmacol. 2014;54:368-374.

22. Information sur le médicament.

Swissmedic.

Available at: http://www.swissmedicinfo.ch/. Accessed 15.12.2018.

23. FDA (US FOOD and Drug Administration). Full Prescribing information of Zelboraf.

Available at: https://www.accessdata.fda.gov/drugsatfda_docs/label/2017/202429s012lbl.pdf.

Accessed 15.12.2018.

24. European Medicines Agency. Science Medicines Health. Summary of product characteristics of

Zelboraf.

Available at: https://www.ema.europa.eu/en/documents/product-information/zelboraf-epar-product- information_en.pdf. Accessed 15.12.2018.

25. Budha NR, Frymoyer A, Smelick GS, et al. Drug absorption interactions between oral targeted anticancer agents and PPIs: is pH-dependent solubility the Achilles heel of targeted therapy? Clin Pharmacol Ther. 2012;92:203-213.

26. van Leeuwen RWF, Jansman FGA, Hunfeld NG, et al. Tyrosine kinase inhibitors and proton pump inhibitors: An evaluation of treatment options. Clin Pharmacokinet.

2017;56:683-688.

27. Smelick GS, Heffron TP, Chu L, et al. Prevalence of acid-reducing agents (ARA) in cancer populations and ARA drug-drug interaction potential for molecular targeted agents in clinical development. Mol Pharmaceut. 2013;10:4055-4062.

28. Chu MP, Ghosh S, Chambers CR, et al. Gastric Acid suppression is associated with decreased erlotinib efficacy in non-small-cell lung cancer. Clin Lung Cancer.

2015;16:33-39.

29. Mir O, Touati N, Lia M, et al. A impact of concomitant administration of gastric acid- suppressive agents and pazopanib on outcomes in soft-tissue sarcoma patients treated within the EORTC 62043/62072 trials. Clin Cancer Res. 2019;25:1479-1485.

30. Teo YL, Ho HK, Chan A. Metabolism-related pharmacokinetic drug-drug interactions with tyrosine kinase inhibitors: current understanding, challenges, and recommendations.

Br J Clin Pharmacol. 2015;79:241-253.

31. van Leeuwen RW, van Gelder T, Mathijssen RH, et al. Drug-drug interactions with tyrosine-kinase inhibitors: a clinical perspective. The Lancet Oncol. 2014;15:e315-326.

32. Shumaker R, Aluri J, Fan J, et al. Effects of ketoconazole on the pharmacokinetics of lenvatinib (E7080) in healthy participants. Clin Pharmacol Drug Develop. 2015;4:155- 160.

33. Lees J, Chan A. Polypharmacy in elderly patients with cancer: clinical implications and management. The Lancet Oncol. 2011;12:1249-1257.

34. Bowlin SJ, Xia F, Wang W, et al. Twelve-month frequency of drug-metabolizing enzyme and transporter-based drug-drug interaction potential in patients receiving oral enzyme-targeted kinase inhibitor antineoplastic agents. Mayo Clinic Proc. 2013;88:139- 148.

35. de Jong FA, Sparreboom A, Verweij J, et al. Lifestyle habits as a contributor to anti- cancer treatment failure. Eur J Cancer. 2008;44:374-382.

36. Loquai C, Dechent D, Garzarolli M, et al. Use of complementary and alternative medicine: A multicenter cross-sectional study in 1089 melanoma patients. Eur J Cancer.

2017;71:70-79.

37. Tascilar M, de Jong FA, Verweij J, et al. Complementary and alternative medicine during cancer treatment: beyond innocence. The Oncologist. 2006;11:732-741.

38. Davis EL, Oh B, Butow PN, et al. Cancer patient disclosure and patient-doctor communication of complementary and alternative medicine use: a systematic review.

The Oncologist. 2012;17:1475-1481.

39. Jermini M, Dubois J, Rodondi PY, et al. Complementary medicine use during cancer treatment and potential herb-drug interactions from a cross-sectional study in an academic centre. Sci Rep. 2019;9:5078.

40. Hurria A. Clinical trials in older adults with cancer: past and future. Oncol. 2007;21:351- 358; discussion 363-354, 367.

41. Feliu J, Heredia-Soto V, Girones R, et al. Can we avoid the toxicity of chemotherapy in elderly cancer patients? Critical Rev Oncol/Hematol. 2018;131:16-23.

42. Sarfati D, Koczwara B, Jackson C. The impact of comorbidity on cancer and its treatment. CA: A Cancer J Clin. 2016;66:337-350.

43. Talarico L, Chen G, Pazdur R. Enrollment of elderly patients in clinical trials for cancer drug registration: a 7-year experience by the US food and drug administration. J Clin

Oncol. 2004;22:4626-4631.

44. Noone AM HN, Krapcho M, Miller D, et al. SEER Cancer Statistics Review, 1975-2015,

National Cancer Institute. Available at: https://seer.cancer.gov/csr/1975_2015/. Accessed

19.12.2018.

45. Singh H, Hurria A, Klepin HD. Progress through collaboration: An ASCO and U.S. Food and Drug Administration workshop to improve the evidence base for treating older adults with cancer. Am Soc Clin Oncol Educational Book. 2018:392-399.

46. de Wit D, Guchelaar HJ, den Hartigh J, et al. Individualized dosing of tyrosine kinase inhibitors: are we there yet? Drug Disc Today. 2015;20:18-36.

47. Klumpen HJ, Samer CF, Mathijssen RH, et al. Moving towards dose individualization of tyrosine kinase inhibitors. Cancer Treat Rev. 2011;37:251-260.

48. Keller DA, Brennan, R. J., Leach, K. L. title. Antitargets and Drug Safety. 2015.

49. Hu JC, Sadeghi P, Pinter-Brown LC, et al. Cutaneous side effects of epidermal growth factor receptor inhibitors: clinical presentation, pathogenesis, and management J Am

Acad Dermatol. 2007;56:317-326.

50. Fabbrocini G, Panariello L, Caro G, et al. Acneiform rash induced by EGFR inhibitors:

Review of the literature and new insights. Skin Appendage Disorders. 2015;1:31-37.

51. Wacker B, Nagrani T, Weinberg J, et al. Correlation between development of rash and efficacy in patients treated with the epidermal growth factor receptor tyrosine kinase inhibitor erlotinib in two large phase III studies. Clin Cancer Res. 2007;13:3913-3921.

52. Perez-Soler R. Rash as a surrogate marker for efficacy of epidermal growth factor receptor inhibitors in lung cancer. Clin Lung Cancer. 2006;8 Suppl 1:S7-14.

53. Launay-Vacher V, Deray G. Hypertension and proteinuria: A class-effect of antiangiogenic therapies. Anti-cancer Drugs. 2009;20:81-82.

54. Kassem L, Shohdy KS, Lasheen S, et al. Hematological adverse effects in breast cancer patients treated with cyclin-dependent kinase 4 and 6 inhibitors: A systematic review and meta-analysis. Breast Cancer. 2018;25:17-27.

55. Hehlmann R, Cortes JE, Zyczynski T, et al. Tyrosine kinase inhibitor interruptions, discontinuations and switching in patients with chronic-phase chronic myeloid leukemia in routine clinical practice: SIMPLICITY. Am J Hematol. 2019;94:46-54.

56. Chen YF, Fu LW. Mechanisms of acquired resistance to tyrosine kinase inhibitors. Acta

Pharmaceutica Sinica B. 2011;1:197-207.

57. Gainor JF, Shaw AT. Emerging paradigms in the development of resistance to tyrosine kinase inhibitors in lung cancer. J Clin Oncol. 2013;31:3987-3996.

58. Gillis NK, McLeod HL. The pharmacogenomics of drug resistance to protein kinase inhibitors. Drug Resistance Updates. 2016;28:28-42.

59. Sharma SV, Bell DW, Settleman J, et al. Epidermal growth factor receptor mutations in lung cancer. Nat Rev Cancer. 2007;7:169-181.

60. Gao J, Li HR, Jin C, et al. Strategies to overcome acquired resistance to EGFR TKI in the treatment of non-small cell lung cancer. Clin Translational Oncol. 2019.

61. van Hoppe S, Sparidans RW, Wagenaar E, et al. Breast cancer resistance protein (BCRP/ABCG2) and P-glycoprotein (P-gp/ABCB1) transport afatinib and restrict its oral availability and brain accumulation. Pharmacol Res. 2017;120:43-50.

62. Wu CP, Ambudkar SV. The pharmacological impact of ATP-binding cassette drug transporters on vemurafenib-based therapy. Acta Pharmaceutica Sinica B. 2014;4:105- 111.

63. Lee Y, Choi YR, Kim KY, et al. The impact of intermittent versus continuous exposure to EGFR tyrosine kinase inhibitor on selection of EGFR T790M-mutant drug-resistant clones in a lung cancer cell line carrying activating EGFR mutation. Oncotarget.

2016;7:43315-43323.

64. Kazaz SN, Oztop I. Treatment after first-generation epidermal growth factor receptor tyrosine kinase inhibitor resistance in non-small-cell lung cancer. Turkish Thoracic J.

2017;18:66-71.

65. Richman J, Martin-Liberal J, Diem S, et al. BRAF and MEK inhibition for the treatment of advanced BRAF mutant melanoma. Expert Opin Pharmacother. 2015;16:1285-1297.

66. Sun J, Zager JS, Eroglu Z. Encorafenib/binimetinib for the treatment of BRAF-mutant advanced, unresectable, or metastatic melanoma: design, development, and potential place in therapy. Onco Targets and Ther. 2018;11:9081-9089.

67. De Lisi D, De Giorgi U, Lolli C, et al. Lenvatinib in the management of metastatic renal cell carcinoma: A promising combination therapy? Expert Opin Drug Metab Toxicol.

2018;14:461-467.

68. Tahara M, Brose MS, Wirth LJ, et al. Impact of dose interruption on the efficacy of lenvatinib in a phase 3 study in patients with radioiodine-refractory differentiated thyroid cancer. Eur J Cancer. 2019;106:61-68.

69. Rood JJM, Schellens JHM, Beijnen JH, et al. Recent developments in the chromatographic bioanalysis of approved kinase inhibitor drugs in oncology. J

Pharmaceut Biomed Anal. 2016;130:244-263.

70. Wong AL, Xiang X, Ong PS, et al. A review on liquid chromatography-tandem mass spectrometry methods for rapid quantification of oncology drugs. Pharmaceut. 2018;10.

71. Decosterd LA, Widmer N, André P, et al. The emerging role of multiplex tandem mass spectrometry analysis for therapeutic drug monitoring and personalized medicine. TrAC

Trends Analytical Chem. 2016;84:5-13.

72. Nováková L, Matysová L, Solich P. Advantages of application of UPLC in pharmaceutical analysis. Talanta. 2006;68:908-918.

73. Rousset M, Titier K, Bouchet S, et al. An UPLC-MS/MS method for the quantification of BRAF inhibitors (vemurafenib, dabrafenib) and MEK inhibitors (cobimetinib, trametinib, binimetinib) in human plasma. Clinica Chimica Acta. 2017;470:8-13.

74. Merienne C, Rousset M, Ducint D, et al. High throughput routine determination of 17 tyrosine kinase inhibitors by LC-MS/MS. J Pharmaceut Biomed Anal. 2018;150:112- 120.

75. Cardoso E, Mercier T, Wagner AD, et al. Quantification of the next-generation oral anti- tumor drugs dabrafenib, trametinib, vemurafenib, cobimetinib, pazopanib, regorafenib

and two metabolites in human plasma by liquid chromatography-tandem mass spectrometry. J Chromatography B. 2018;1083:124-136.

76. Masuda T, Hattori N, Hamada A, et al. Erlotinib efficacy and cerebrospinal fluid concentration in patients with lung adenocarcinoma developing leptomeningeal metastases during gefitinib therapy. Cancer Chemother Pharmacol. 2011;67:1465-1469.

77. Bernard S, Goldwirt L, Amorim S, et al. Activity of ibrutinib in mantle cell lymphoma patients with central nervous system relapse. Blood. 2015;126:1695-1698.

78. McArthur GA, Maio M, Arance A, et al. Vemurafenib in metastatic melanoma patients with brain metastases: an open-label, single-arm, phase 2, multicentre study. Annals of

Oncol. 2017;28:634-641.

79. Steeg PS, Camphausen KA, Smith QR. Brain metastases as preventive and therapeutic targets Nature Rev Cancer. 2011;11:352-363.

80. Namba Y, Kijima T, Yokota S, et al. Gefitinib in patients with brain metastases from non-small-cell lung cancer: review of 15 clinical cases. Clin Lung Cancer. 2004;6:123- 128.

81. Deng Y, Feng W, Wu J, et al. The concentration of erlotinib in the cerebrospinal fluid of patients with brain metastasis from non-small-cell lung cancer. Mol Clin Oncol.

2014;2:116-120.

82. Zhao J, Chen M, Zhong W, et al. Cerebrospinal fluid concentrations of gefitinib in patients with lung adenocarcinoma. Clin Lung Cancer. 2013;14:188-193.

83. Beauvais D, Goossens JF, Boyle E, et al. Development and validation of an UHPLC- MS/MS method for simultaneous quantification of ibrutinib and its dihydrodiol- metabolite in human cerebrospinal fluid. J Chromatography B. 2018;1093-1094:158- 166.

84. Bai F, Johnson J, Wang F, Yang L, Broniscer A and Stewart CF Determination of vandetanib in human plasma and cerebrospinal fluid by liquid chromatography electrospray ionization tandem mass spectrometry (LC-ESI-MS/MS). J Chromatography

B. 2011;879:2561-2566.

85. Hooshfar S, Basiri B, Bartlett MG. Development of a surrogate matrix for cerebral spinal fluid for liquid chromatography/mass spectrometry based analytical methods. Rapid

Comm Mass Spect. 2016;30:854-858.

86. Kralj E, Trontelj J, Pajic T, et al. Simultaneous measurement of imatinib, nilotinib and dasatinib in dried blood spot by ultra high performance liquid chromatography tandem mass spectrometry. J Chromatography B. 2012;903:150-156.

87. Nijenhuis CM, Huitema AD, Marchetti S, et al. The use of dried blood spots for pharmacokinetic monitoring of vemurafenib treatment in melanoma patients. J Clin

Pharmacol. 2016;56:1307-1312.

88. Verheijen RB, Bins S, Thijssen B, et al. Development and clinical validation of an LC- MS/MS method for the quantification of pazopanib in DBS. Bioanalysis. 2016;8:123- 134.

89. Wilhelm AJ, den Burger JC, Swart EL. Therapeutic drug monitoring by dried blood spot: progress to date and future directions. Clin Pharmacokinet. 2014;53:961-973.

90. Edelbroek PM, van der Heijden J, Stolk LM. Dried blood spot methods in therapeutic drug monitoring: methods, assays, and pitfalls. Ther Drug Monit. 2009;31:327-336.

91. Timmerman P, White S, Globig S, et al. EBF recommendation on the validation of bioanalytical methods for dried blood spots. Bioanalysis. 2011;3:1567-1575.

92. Freiwald M, Schmid U, Fleury A, et al. Population pharmacokinetics of afatinib, an irreversible ErbB family blocker, in patients with various solid tumors. Cancer

Chemother Pharmacol. 2014;73:759-770.

93. Hong Y, Passos VQ, Huang PH, et al. Population pharmacokinetics of Ceritinib in adult patients with tumors characterized by genetic abnormalities in anaplastic lymphoma kinase. J Clin Pharmacol. 2017;57:652-662.

94. Han K, Jin JY, Marchand M, et al. Population pharmacokinetics and dosing implications for cobimetinib in patients with solid tumors. Cancer Chemother Pharmacol.

2015;76(5):917-24.

95. Ette E, Williams PJ. Population pharmacokinetics I: background, concepts, and models.

Annals of Pharmacother. 2004;38:1702-1706.

96. Ikeda K, Kudo M, Kawazoe S, et al. Phase 2 study of lenvatinib in patients with advanced hepatocellular carcinoma. J Gastroenterol. 2017;52:512-519.

97. Tamai T, Hayato S, Hojo S, et al. Dose finding of Lenvatinib in subjects with advanced hepatocellular carcinoma based on population pharmacokinetic and exposure-response analyses. J Clin Pharmacol. 2017;57:1138-1147.

98. FDA (US FOOD and Drug Administration). Full Prescribing information of Lenvima.

Available at: https://www.accessdata.fda.gov/drugsatfda_docs/label/2018/206947s007lbl.pdf.

Accessed 11.12.2018.

99. Mann C. Meta-analysis in the breech. Sci. 1990;249:476-480.

100. Gotta V, Buclin T, Csajka C, et al. Systematic review of population pharmacokinetic analyses of imatinib and relationships with treatment outcomes. Ther Drug Monit.

2013;35:150-167.

101. Petit-Jean E, Buclin T, Guidi M, et al. Erlotinib: Another candidate for the therapeutic drug monitoring of targeted therapy of cancer?

A pharmacokinetic and pharmacodynamic systematic review of literature. Ther Drug Monit. 2015;37:2-21.

102. Gotta V. Application and evaluation of population pharmacokinetics for imatinib dosage

Individualization. Geneva: University of Geneva. PhD Thesis, 2013: 153 p. Available at: https://archive-ouverte.unige.ch/unige:29042.

103. Ward MB, Reuter SE, Martin JH. How 'optimal' are optimal sampling times for tyrosine kinase inhibitors in cancer? practical considerations. Clin Pharmacokinet. 2016;55:1171- 1177.

104. Diekstra MH, Fritsch A, Kanefendt F, et al. Population modeling integrating pharmacokinetics, pharmacodynamics, pharmacogenetics, and clinical outcome in patients with sunitinib-treated cancer. CPT: Pharmacomet Sys Pharmacol. 2017;6:604- 613.

105. Brown K, Comisar C, Witjes H, et al. Population pharmacokinetics and exposure- response of osimertinib in patients with non-small cell lung cancer. Br J Clin Pharmacol.

2017;83:1216-1226.

106. Puszkiel A, Noe G, Bellesoeur A, et al. Clinical pharmacokinetics and pharmacodynamics of dabrafenib. Clin Pharmacokinet. 2019;58:451-467.

107. Csajka C, Verotta D. Pharmacokinetic-pharmacodynamic modelling: History and perspectives. J Pharmacokinet Pharmacodynam. 2006;33:227-279.

108. Sun W, O'Dwyer PJ, Finn RS, et al. Characterization of neutropenia in advanced cancer patients following palbociclib treatment using a population pharmacokinetic- pharmacodynamic modeling and simulation approach. J Clin Pharmacol. 2017;57:1159- 1173.

109. Verheijen RB, Yu H, Schellens JHM, et al. Practical recommendations for therapeutic drug monitoring of kinase inhibitors in oncology. Clin Pharmacol Ther. 2017;102:765- 776.

110. Kim HY, Martin JH, Mclachlan AJ, et al. Precision dosing of targeted anticancer drugs—challenges in the real world. Trans Cancer Res. 2017:S1500-S1511.

111. Lankheet NAG, Desar IME, Mulder SF, et al. Optimizing the dose in cancer patients treated with imatinib, sunitinib and pazopanib. Br J Clin Pharmacol. 2017;83:2195- 2204.

112. Goulooze SC, Galettis P, Boddy AV, et al. Monte Carlo simulations of the clinical benefits from therapeutic drug monitoring of sunitinib in patients with gastrointestinal stromal tumours. Cancer Chemother Pharmacol. 2016;78:209-216.

113. Verheijen RB, Bins S, Mathijssen RH, et al. Individualized Pazopanib dosing: A prospective feasibility study in cancer patients. Clin Cancer Res. 2016;22:5738-5746.

114. Gotta V, Widmer N, Decosterd LA, et al. Clinical usefulness of therapeutic concentration monitoring for imatinib dosage individualization: Results from a randomized controlled trial. Cancer Chemother Pharmacol. 2014;74:1307-1319.

115. Rousselot P, Johnson-Ansah H, Huguet F, et al. Personalized daily doses of imatinib by therapeutic drug monitoring increase the rates of molecular responses in patients with chronic myeloid leukemia. Final results of the randomized OPTIM Imatinib study.

Blood. 2015;126:133-133.

116. Rousseau A, Marquet P. Application of pharmacokinetic modelling to the routine therapeutic drug monitoring of anticancer drugs. Fundam Clin Pharmacol. 2002;16:253- 262.

117. Buclin T, Widmer N, Biollaz J, et al. Who is in charge of assessing therapeutic drug monitoring? The case of imatinib. Lancet Oncol. 2011;12:9-11.

118. Buclin T, Gotta V, Fuchs A, et al. Monitoring drug therapy. Br J Clin Pharmacol.

2012;73:917-923.

119. ClinicalTrials. Optimizing pazopanib exposure in RCC patients (OPERA). Available at: www.clinicaltrials.gov/ct2/show/NCT02089802. Accessed 18.01.2019.

120. Veal GJ, Amankwatia EB, Paludetto MN, et al. Pharmacodynamic therapeutic drug monitoring for cancer: challenges, advances, and future opportunities. Ther Drug Monit.

2019;41:142-159.

121. FDA (US FOOD and Drug Administration). Full Prescribing information of Zykadia.

Available at: https://www.accessdata.fda.gov/drugsatfda_docs/label/2017/205755s011lbl.pdf.

Accessed 20.12.2018.

122. FDA (US FOOD and Drug Administration). Full Prescribing information of Imbruvica.

Available at: https://www.accessdata.fda.gov/drugsatfda_docs/label/2018/210563s000lbl.pdf.

Accessed 20.12.2018.

123. Wang Y, Chia YL, Nedelman J, et al. A therapeutic drug monitoring algorithm for refining the imatinib trough level obtained at different sampling times. Ther Drug Monit.

2009;31:579-584.

124. Crombag MBS, van Doremalen JGC, Janssen JM, et al. Therapeutic drug monitoring of small molecule kinase inhibitors in oncology in a real-world cohort study: does age matter? Br J Clin Pharmacol. 2018;84:2770-2778.

125. Lankheet NA, Knapen LM, Schellens JH, et al. Plasma concentrations of tyrosine kinase inhibitors imatinib, erlotinib, and sunitinib in routine clinical outpatient cancer care. Ther

Drug Monit. 2014;36:326-334.

126. Widmer N, Decosterd LA, Leyvraz S, et al. Relationship of imatinib-free plasma levels and target genotype with efficacy and tolerability. Br J Cancer. 2008;98:1633-1640.

127. Rousseau A, Sabot C, Delepine N, et al. Bayesian estimation of methotrexate pharmacokinetic parameters and area under the curve in children and young adults with localised osteosarcoma. Clin pharmacokinet. 2002;41:1095-1104.

128. Gotta V, Widmer N, Montemurro M, et al. Therapeutic drug monitoring of imatinib:

Bayesian and alternative methods to predict trough levels. Clin Pharmacokinet.

2012;51:187-201.

129. FDA (US FOOD and Drug Administration). Full Prescribing information of Tarceva.

Available at: https://www.accessdata.fda.gov/drugsatfda_docs/label/2010/021743s14s16lbl.pdf.

Accessed 21.12.2018.

130. Group GIW. DDI-predictor. Available at: http://www.ddi-predictor.org/. Accessed

21.12.2018.

131. Josephs DH, Fisher DS, Spicer J, et al. Clinical pharmacokinetics of tyrosine kinase inhibitors: implications for therapeutic drug monitoring. Ther Drug Monit. 2013;35:562- 587.

132. Widmer N, Decosterd LA, Csajka C, et al. Population pharmacokinetics of imatinib and the role of alpha-acid glycoprotein. Br J Clin Pharmacol. 2006;62:97-112.

133. Lu JF, Eppler SM, Wolf J, et al. Clinical pharmacokinetics of erlotinib in patients with solid tumors and exposure-safety relationship in patients with non-small cell lung cancer.

Clin Pharmacol Ther. 2006;80:136-145.

134. Sorich MJ, Mutlib F, van Dyk M, Et al. Use of physiologically based pharmacokinetic modeling to identify physiological and molecular characteristics driving variability in axitinib exposure: A fresh approach to precision dosing in oncology. J Clin Pharmacol.

2019;59:872-879.

135. Smith SA, Waters NJ. Pharmacokinetic and pharmacodynamic considerations for drugs binding to alpha-1-acid glycoprotein. Pharmaceut Res. 2018;36:30.

136. Fayet A, Beguin A, de Tejada BM, et al. Determination of unbound antiretroviral drug concentrations by a modified ultrafiltration method reveals high variability in the free fraction. Ther. Drug Monit. 2008;30:511-522.

137. Haouala A, Widmer N, Guidi M, et al. Prediction of free imatinib concentrations based on total plasma concentrations in patients with gastrointestinal stromal tumours. Br J

Clin Pharmacol. 2013;75:1007-1018.

138. Vrijens B, De Geest S, Hughes DA, et al. A new taxonomy for describing and defining adherence to medications. Br J Clin Pharmacol. 2012;73:691-705.

139. Marin D, Bazeos A, Mahon FX, et al. Adherence is the critical factor for achieving molecular responses in patients with chronic myeloid leukemia who achieve complete cytogenetic responses on imatinib. J Clin Oncol. 2010;28:2381-2388.

140. Ibrahim AR, Eliasson L, Apperley JF, et al. Poor adherence is the main reason for loss of

CCyR and imatinib failure for chronic myeloid leukemia patients on long-term therapy.

Blood. 2011;117:3733-3736.

141. Breccia M, Efficace F, Sica S, et al. Adherence and future discontinuation of tyrosine kinase inhibitors in chronic phase chronic myeloid leukemia. A patient-based survey on

1133 patients. Leukemia Res. 2015;39:1055-1059.

142. Noens L, van Lierde MA, De Bock R, et al. Prevalence, determinants, and outcomes of nonadherence to imatinib therapy in patients with chronic myeloid leukemia: the

ADAGIO study Blood. 2009;113:5401-5411.

143. Karuturi MS, Holmes HM, Lei X, et al. Potentially inappropriate medication use in older patients with breast and colorectal cancer. Cancer. 2018;124:3000-3007.

144. Verbrugghe M, Duprez V, Beeckman D, et al. Factors influencing adherence in cancer patients taking oral tyrosine kinase inhibitors: A qualitative study. Cancer Nurs.

2016;39:153-162.

145. Mathes T, Antoine SL, Pieper D, et al. Adherence enhancing interventions for oral anticancer agents: a systematic review. Cancer Treat Rev. 2014;40:102-108.

146. Colombo LRP, Aguiar PM, Lima TM, et al. The effects of pharmacist interventions on adult outpatients with cancer: A systematic review. J Clin Pharmacy Ther. 2017;42:414- 424.

147. Occhipinti S, Petit-Jean E, Pinguet F, et al. Pharmacist involvement in supporting care in patients receiving oral anticancer therapies: A situation report in French cancer centers.

Bulletin du Cancer. 2017;104:727-734.

148. Brunot A, Le Roy F, Le Sourd S, et al. Implementation of a nurse-driven educational program improves management of sorafenib's toxicities in hepatocellular carcinoma.

Cancer Nurs. 2018;41:418-423.

149. Schneider SM, Hess K, Gosselin T. Interventions to promote adherence with oral agents.

Seminars Oncol Nurs. 2011;27:133-141.

150. Schneider MP, Achtari JL, Chevaux B, et al. A novel approach to better characterize medication adherence in oral anticancer treatments. Frontiers Pharmacol. 2019;9.

151. Cardoso E, Csajka C, Schneider MP, et al.

Effect of adherence on pharmacokinetic/pharmacodynamic relationships of oral targeted anticancer drugs. Clin

Pharmacokinet. 2018;57:1-6.

152. Barriere O, Li J, Nekka F. A Bayesian approach for the estimation of patient compliance based on the last sampling information. J Pharmacokinet Pharmacodynam. 2011;38:333- 351.

153. Henin E, Tod M, Trillet-Lenoir V, et al. Pharmacokinetically based estimation of patient compliance with oral anticancer chemotherapies: in silico evaluation. Clin

Pharmacokinet. 2009;48:359-369.

154. Fuchs A, Csajka C, Thoma Y, et al. Benchmarking therapeutic drug monitoring software: a review of available computer tools. Clin Pharmacokinet. 2013;52:9-22.

155. Dubovitskaya A, Buclin T, Schumacher M, et al. Tucuxi: An intelligent system for personalized medicine from individualization of treatments to research databases and back. In: Proc 8th ACM Int Conf Bioinformat, Comput Biol, Health Informat. 2017; 223- 232.

156. Decosterd LA, Widmer N, Zaman K, et al. Therapeutic drug monitoring of targeted anticancer therapy. Biomarkers Med. 2015;9:887-893.

157. Bardin C, Veal G, Paci A, et al. Therapeutic drug monitoring in cancer--are we missing a trick? Eur J Cancer. 2014;50:2005-2009.

Figure legend Figure 1. Steps of a TDM interpretation. Schematic of the steps involved in the interpretation of an imatinib TDM result of 620 µg/L, measured 9 h after last dose intake under a regimen of 400 mg q.d. in a 50-year, 60 kg female patient. This result is slightly above the percentile 10% of expected levels (1, brown bands). The recommended target concentration at the trough is 1000 µg/L (2, red cross). A Bayesian maximum likelihood individual curve is computed, which indicates an extrapolated trough level of 575 µg/L (3a, green dashed curve). A simple rule of three shows that a dosage of 800 mg q.d. would have high probability to ensure a trough level of 1150 µg/L, thus reaching the target appropriately (3b, blue dashed-dotted curve).

Figure drawn by T. Buclin.

📖 中文全文 Chinese Full Text

中文

# 治疗药物监测

## 靶向抗癌蛋白激酶抑制剂在常规临床应用中的治疗药物监测：一项批判性综述

**Evelina Cardoso, PharmD¹,²; Monia Guidi, PhD¹,²; Benoît Blanchet, PhD³,⁴; Marie Paule Schneider, PhD²; Laurent A. Decosterd, PhD¹; Thierry Buclin, MD¹; Chantal Csajka, PhD¹,²*; Nicolas Widmer, PhD¹,²,⁵***

¹ 瑞士洛桑大学医院及洛桑大学临床药理学服务部 ² 瑞士日内瓦大学及洛桑大学药学院 ³ 法国巴黎公立医院集团-科钦医院药代动力学与药化学系 ④ 法国巴黎笛卡尔大学药学系，UMR8638 CNRS，PRES索邦巴黎西岱大学 ⁵ 瑞士沃州东部医院药房

* 共同贡献——共同通讯作者

---

**摘要**

**目的：** 口服靶向抗癌蛋白激酶抑制剂（PKIs）的治疗反应在患者间差异显著，部分患者疗效不足，另一些患者则出现不可接受的药物不良反应。这种异质性可能由多种原因引起，如影响血药浓度的药代动力学变异性、波动的用药依从性，以及癌细胞的先天性或获得性耐药性。因此，肿瘤患者接受PKI治疗时需要协同努力以优化这些药物的使用，这些药物可能尚未发挥其全部潜力。

**方法：** 在MEDLINE数据库中对PKIs的药代动力学、药效学及治疗药物监测（TDM）相关文献进行了广泛检索（截至2019年4月）。

**结果：** 本文提供了确定适合进行TDM的PKIs的标准（如分析方法的可获得性、观察性药代动力学研究、PK-PD关系分析及随机对照研究）。综述了PKIs的主要特征与局限性、TDM对接受PKIs治疗的癌症患者的预期益处，以及合理应用TDM的前提条件。最后，论文讨论了TDM在支持癌症治疗更好实施中的各种重要实践方面和潜在陷阱。

**结论：** 在个体化医疗时代，通过合理的TDM方法在个体患者水平调整PKIs的给药方案，可以防止肿瘤患者暴露于无效或不必要的毒性药物浓度。

**关键词：** 分子靶向治疗；药代动力学；变异性；药物监测

---

**引言**

在过去十年中，人们在理解驱动癌症发展的机制方面投入了大量努力。¹,² 细胞增殖、存活、分化和迁移的相关信息通过细胞内信号通路传递，其中许多通路涉及蛋白激酶。这些蛋白激酶的基因过表达或突变可能导致相应通路的组成性激活，从而引发癌症。³,⁴ 对信号通路的深入理解以及参与癌症病理生理的失调蛋白的鉴定，促进了合理设计的靶向药物——蛋白激酶抑制剂（PKIs）的开发。⁵ 这些小分子药物显著改变了某些癌症的治疗和预后。它们目前是血液系统恶性肿瘤（如慢性髓性白血病[CML]或费城染色体阳性急性淋巴细胞白血病[Phi+ALL]）以及多种转移性实体瘤（包括肝细胞癌[HCC]、肾细胞癌[RCC]、恶性黑色素瘤、胃肠道间质瘤[GIST]和特定分子亚型肺癌[表皮生长因子受体[EGFR]激活突变]）系统治疗的支柱。此外，鉴于某些PKIs对突变激酶变体的特异性，当基因分析揭示信号通路中存在可能适合靶向药物干预的突变时，这些PKIs正越来越多地应用于传统治疗无法控制的癌症的试验中。这促进了PKIs的广泛应用；然而，需要通过适当的研究进行充分的规范化和验证。

由于口服给药，PKIs比细胞毒性化疗和免疫治疗提供了更大的自主性和更简单的门诊护理，从而改善了患者的生活质量。⁶ 然而，PKIs的治疗反应在患者间差异显著，部分病例疗效不足，另一些病例则出现不可接受的药物不良反应。这种异质性可能由多种原因引起，如药代动力学（PK）变异性、波动的用药依从性，以及癌细胞的先天性或获得性耐药性。此外，这些口服靶向治疗费用高昂，对公共医疗系统构成沉重负担。⁷ 因此，在肿瘤患者管理中合理使用PKIs需要在适应症个体化、剂量个体化和治疗精确引导方面取得明确进展。现在必须将所有努力集中于旨在优化治疗的策略，即关于药物选择和给药方案的决策，通过定期监测和重新评估，尽可能满足每位癌症患者的特定需求。在此背景下，治疗药物监测（TDM）非常适合通过基于循环药物浓度测量的给药方案调整来优化PKIs的治疗活性。这种PK监测可以防止患者暴露于无效或不必要的毒性药物水平。

本综述首先通过介绍PKIs当前使用的一些局限性以及基于当前科学证据的TDM对接受PKIs治疗的癌症患者的益处，来回答"为什么靶向抗癌PKIs需要TDM？"这一问题。其次，将综述TDM应用的前提条件。最后，将描述TDM的各种实践方面和潜在陷阱，重点关注蛋白激酶抑制剂。所讨论的主题将以近期获批的PKIs在癌症治疗中的实例加以说明。

本文重点关注TDM在临床实践中的实际方面，例如强调与采样时间、蛋白结合和用药依从性相关的问题。此外，本文描述了一个正式的TDM结果解读和后续剂量个体化的三步流程。

---

**1. 方法**

在MEDLINE数据库中进行了广泛的文献检索（截至2019年4月），特别使用以下MeSH术语来识别相关研究和文章："抗肿瘤药物"、"蛋白激酶抑制剂"、"蛋白酪氨酸激酶"、"蛋白丝氨酸-苏氨酸激酶"、"药理学"、"药代动力学"和"药物监测"，以及各种靶向抗癌药物名称。同时检查了相关文章中列出的参考文献。

---

**2. 为什么靶向抗癌PKIs需要TDM？**

**影响PKIs临床开发的局限性**

**标准剂量的选择**

传统化疗药物的推荐剂量通常通过在癌症患者中进行的I期试验的剂量递增来确定；标准剂量通常对应于最大耐受剂量（MTD），即仍与可接受毒性水平相关的最高剂量。这种剂量选择方法适用于细胞毒性化疗，其假设是更高剂量将对生长的肿瘤细胞产生更强的效果。⁸ 相反，对于具有特定靶点并每日给药的新型口服抗癌药物，多位作者认为MTD方法不再适用。⁹⁻¹¹ PKIs对细胞信号转导的抑制遵循经典的S形对数浓度-反应曲线，在超过一定暴露水平后达到最大值。因此，基于MTD的给药方案可能指示高于最大疗效所需的剂量，这可能导致严重毒性，进而需要减量和/或治疗中断。⁹⁻¹¹ 大多数获批的PKIs在标准剂量选择时采用了MTD方法，没有明确的浓度-反应关系。¹² 例如，仑伐替尼被FDA批准用于治疗局部复发或转移性、放射性碘难治性分化型甲状腺癌（DTC）。每日24 mg的剂量是在实体癌患者的I期剂量递增研究中确定的MTD。¹³ 然而，在III期研究中，推荐方案下75.9%的患者发生了3级或更高级别的不良事件。因不良事件而减量和停药的患者比例分别为68%和14%。¹⁴ 这些结果对仑伐替尼最佳剂量的选择提出了质疑，上市后试验仍需研究较低剂量是否能在维持疗效的同时提供更好的安全性特征¹¹,¹⁵ 并降低成本。

因此，在进入市场之前，需要通过考虑替代剂量寻找策略来优化剂量选择。⁹,¹¹,¹⁶,¹⁷ 疗效而非安全性应主导剂量决策，重点放在暴露-反应建模上。¹⁸

**患者间PK变异性的评估**

大多数PKIs在接受标准剂量的患者中表现出非常不同的血浆浓度，即患者间PK变异性。这种变异性源于影响PKIs吸收、分布、代谢和排泄（ADME）PK过程的多种因素（如病理生理、遗传药理学和环境因素）。¹⁷,¹⁹ 吸收和代谢都是受PKIs处置变异性影响最大的过程。

向口服抗癌治疗的转变需要研究食物摄入对PKIs生物利用度的影响；这种影响通常在三种不同食物条件下进行评估：空腹状态、清淡饮食或高脂饮食。PKIs的吸收可能不受伴随食物摄入的影响（如考比替尼、泊那替尼或瑞博西尼），也可能受到高脂餐的显著影响。例如，与空腹状态相比，阿法替尼的浓度-时间曲线下面积（AUC）降低39%，而色瑞替尼²⁰ 和维莫非尼²¹ 的AUC分别增加73%和约5倍。值得注意的是，目前关于维莫非尼与食物同服的推荐意见在瑞士治疗产品管理局（Swissmedic²²）与美国（FDA²³）或欧洲（EMA²⁴）监管机构之间存在差异。Swissmedic建议空腹服用维莫非尼以避免过度暴露，而后两个机构建议与食物无关地服用，同时警告避免系统性地空腹服药以减少暴露不足的风险。其他PKIs（如博舒替尼、吉非替尼和帕唑帕尼）的吸收受胃pH值调节，这影响分子的溶解度。²⁵,²⁶ 事实上，由于癌症患者常用的抗酸治疗²⁷ 导致的胃pH值升高可显著降低某些PKIs的溶解度，从而导致吸收减少和潜在的治疗失败。一项关于接受厄洛替尼治疗的非小细胞肺癌（NSCLC）患者的大型研究显示，同时接受降酸药物的患者预后不良，这归因于厄洛替尼生物利用度的降低。²⁸ 此外，当帕唑帕尼与抗酸治疗联合使用时，观察到治疗疗效显著降低。²⁹

几乎所有PKIs都主要通过肝脏细胞色素P450（CYP）酶家族代谢，主要是CYP3A4同工酶，其活性可受到抑制剂或诱导剂联合给药的强烈影响。这导致PKIs血浆浓度的临床相关变化。³⁰,³¹ 除了CYP抑制剂或诱导剂的剂量外，此类药物相互作用对PKIs暴露的影响程度取决于其效力和参与PKIs代谢的途径数量。例如，仑伐替尼通过多种途径代谢，包括CYP3A4介导的氧化和醛氧化酶代谢，以及谷胱甘肽结合。²² 强效CYP3A4抑制剂酮康唑的联合给药仅对仑伐替尼暴露产生微弱影响（AUC增加15%），³² 证实替代途径可以补偿CYP3A4活性的降低。任何处方药物数量的增加，例如患有合并症的老年癌症患者，都会增加药物相互作用的风险。³³,³⁴ 除了多重用药外，癌症患者还广泛使用补充和替代药物（如草药产品、矿物质、维生素和抗氧化剂）以及其他非处方药。³⁵,³⁶ 肿瘤科医生通常不了解这些使用情况，可能是因为患者不愿透露此类消费，并倾向于认为这些"天然产品"是安全的，忽略了这些物质可能通过对CYP活性的影响而显著改变PKI浓度。³⁷⁻³⁹

患者选择的限制性限制了在药物开发期间正确评估患者间PK变异性。因此，在"真实世界"癌症患者中预期会有更大的循环浓度变异性，这些患者通常年龄较大，一般状况较差，合并症和伴随药物比纳入临床试验严格框架的患者更多。⁴⁰⁻⁴² 根据Talarico等人的研究，⁴³ 75岁以上的成人占已发表试验中纳入患者的不到10%，而26%的癌症在此年龄之后被诊断。⁴⁴ 幸运的是，这一问题在当前药物开发项目中开始得到考虑。⁴⁵

总之，更好地了解PKIs处置的变异性是限制其毒性和疗效不足（可能有利于耐药细胞克隆选择）的先决条件，如下文各节所述。⁴⁶

**不良事件的管理**

由于其相对较高的靶点特异性，PKIs预期比传统细胞毒性化疗具有更低的毒性。然而，PKIs绝非没有发生率和严重程度不等的不良反应。一些不良反应源于靶向激酶在正常细胞中执行的生理功能被抑制，这在理论上是可以预测的。⁴⁷,⁴⁸ 皮疹，特别是痤疮样皮疹，是EGFR抑制剂诱导的非常常见的类特异性毒性，⁴⁹ 用于治疗NSCLC（厄洛替尼、吉非替尼、阿法替尼和奥希替尼）、乳腺癌（拉帕替尼）和甲状腺髓样癌（凡德他尼）。由于EGFR在各种肿瘤中过表达和/或过度激活，它是治疗干预的合理靶点。然而，这种激酶也存在于正常上皮组织中，EGFR介导的信号通路被抑制会导致皮肤炎症和皮肤病变的发展，如丘疹和脓疱，已知对患者的生活质量有重大影响。⁵⁰ 有趣的是，皮疹似乎与EGFR抑制剂的疗效正相关，因此已被建议作为监测治疗反应的替代标志物。⁵¹,⁵² 另一个众所周知的特异性毒性是抗血管生成药物（如索拉非尼、舒尼替尼、阿昔替尼和仑伐替尼）抑制血管内皮生长因子（VEGF）信号通路引起的高血压伴蛋白尿。⁵³ 类广泛不良反应的另一个例子是最近开发的特定细胞周期蛋白依赖性激酶4和6（CDK4/6）抑制剂（帕博西尼和瑞博西尼）诱导的血液学毒性（中性粒细胞减少、白细胞减少、贫血和血小板减少），用于治疗转移性乳腺癌。⁵⁴ 鉴于这些不良作用大多与剂量相关，临床上可通过减量或暂时中断治疗来管理。然而，涉事PKIs的给药通常被停止——在严重和无法耐受的不良反应情况下则永久停药⁵⁵——因为处方肿瘤科医生不愿意降低药物剂量，这可能导致浓度暴露不足。

**应对耐药性**

PKIs的给药并非不常见地与部分患者缺乏治疗反应相关，推测是由于癌细胞对这些分子的耐药性所致。如果患者在治疗开始时就无反应，则称为"原发性"或"先天性"耐药；如果在初始反应期后发现癌症失控，则称为"继发性"或"获得性"耐药。已描述了多种PKIs耐药机制⁵⁶,⁵⁷：

a) 由靶向激酶的基因改变（突变或基因扩增）诱导的耐药。⁵⁸ 在PKIs的选择性压力下，肿瘤细胞的增殖为那些激酶获得额外基因修饰从而逃避药物抑制的细胞提供了进化优势。最常见的改变是影响靶向激酶结合域的突变，导致PKI对其靶点的亲和力降低。例如，在约50%的厄洛替尼、吉非替尼或阿法替尼耐药患者中发现了T790位点突变——EGFR激酶结合位点的关键位置——而在初治患者中很少检测到。⁵⁷,⁵⁹,⁶⁰

b) 由替代激酶通路上调（旁路途径）诱导的耐药。肿瘤细胞可通过下游信号通路的激活获得耐药性，使PKI逐渐失效。⁵⁶

c) 由药物转运蛋白诱导的耐药。某些PKIs是一种或多种ATP结合盒（ABC）转运蛋白的底物，如P-糖蛋白（P-gp）和乳腺癌耐药蛋白（BCRP）等药物外排泵。它们也可被溶质载体（SLC）转运蛋白携带，如有机阴离子（OATs、OATPs）或阳离子转运蛋白（OCTs）。PKIs的耐药可能源于这些转运蛋白在癌细胞中的过度表达，⁵⁶,⁶¹,⁶² 导致细胞内药物水平降低，从而降低PKIs的抗癌疗效。

PKIs耐药的发展是有问题的，因为它会导致肿瘤进展和有效治疗选择的限制。因此，重要的是通过不同策略预防和克服耐药突变体：

a) 确保适当的PKIs全身暴露。耐药诱导突变的发生因暴露于亚治疗浓度或间歇性无药期以及不足以有效阻断细胞复制的浓度而有利于肿瘤中的突变-选择周期。⁶³ 这一现象已在从细菌到害虫的各类化疗药物中得到充分证明，需要优化整个治疗过程中药物浓度暴露的水平和规律性。

b) 采用下一代PKIs来克服获得性耐药。继第二代和第三代药物（尼洛替尼、达沙替尼、博舒替尼和泊那替尼）在携带伊马替尼耐药BCR-ABL突变的CML患者中取得成功之后，类似方法被用于EGFR突变NSCLC，从而开发了奥希替尼和罗西替尼等下一代EGFR抑制剂。⁶⁴

c) 使用同时靶向多种激酶的PKI联合方案。这种方法在黑色素瘤治疗中似乎特别成功，联合使用BRAF和MEK抑制剂（维莫非尼/考比替尼、达拉非尼/曲美替尼和恩考芬尼/比美替尼），靶向Ras-Raf-MEK-ERK信号通路中的两种不同激酶。⁶⁵,⁶⁶ 这一范式将扩展到其他癌症，例如仑伐替尼联合依维莫司治疗转移性RCC。⁶⁷

**TDM的临床获益前景**

上述PKIs使用中的多项局限性可以通过合理的TDM方法在个体患者水平优化给药方案来克服。这些药物具有多种特征，代表了实施TDM项目的经典标准。迄今为止，大多数（如果不是全部）现有PKIs以固定标准剂量推荐上市。某些PKIs选择的标示剂量值得商榷；然而，更值得商榷的是对其重要的患者间PK变异性完全缺乏考虑，这导致标准剂量下实现的血浆浓度谱存在显著异质性。

在PKIs标准剂量下检测到低药物浓度表明，TDM可能防止长期暴露于浓度不足的水平，同时结合依从性增强计划，在转化为癌症失控和耐药发生之前进行干预。

此外，不良作用是导致治疗中断的主要因素，由于治疗选择有限，及时管理毒性对于患者确保长期PKI自我管理至关重要。⁶⁸ 监测患者的血浆浓度可以早期识别过度暴露，可以在药物不良事件发生之前及时、基于事实地进行减量，这有助于改善治疗持续性，防止早期换用其他药物，并减轻患者的焦虑。

基于所有这些原因，再加上癌症是危及生命的疾病以及此类治疗对公共医疗系统的经济压力，TDM应构成支持更有效、可耐受和可持续口服方案的额外方法。最终，这应使患者的临床获益最大化。

---

**3. PKIs的TDM应用有哪些前提条件？**

适当的TDM应用需要标准化的PKI测量方法，以及关于PK特性、PK/PD关系、治疗靶点以及该方法为各种PKIs提供益处的证据水平的先验知识。

**分析方法**

获得可靠的PKIs生物样本定量分析方法是TDM的首要前提条件。目前，可以为PKIs等低分子量分子开发定量技术。最常用的方法是液相色谱-串联质谱联用（LC-MS/MS），⁶⁹⁻⁷¹ 与其他可用生物分析方法相比表现出优异性能。LC/MS/MS的主要优势是速度、选择性和灵敏度。通过减少分析时间，特别是采用超高效液相色谱（UPLC），该技术允许轻松常规应用实时TDM。⁷² 它还覆盖了支持常规TDM和PK/PD研究的宽临床相关血浆浓度范围。

凭借串联质谱提供的选择性，可以在短分析时间内同时测量多种结构不相关的PKIs。多重方法允许优化实验室资源和周转时间，采用独特的样品提取程序和单次色谱运行。当患者接受PKI联合治疗时，这种方法特别相关，如维莫非尼/考比替尼或达拉非尼/曲美替尼，用于不可切除或转移性BRAF突变黑色素瘤。⁷³ 例如，最近已开发并验证了多种PKIs测量的LC-MS/MS多重方法。⁷⁴,⁷⁵ 此类联合分析的一个限制是浓度水平的显著差异，如帕唑帕尼和维莫非尼，因此难以纳入同一分析运行。

凭借高灵敏度，LC-MS/MS也是检测极低药物水平的潜在方法。例如，由于PKIs的血脑屏障（BBB）穿透性有限和高血浆蛋白结合率，预期脑脊液（CSF）中的PKI浓度较低。由于许多报告提供了这些治疗对常见于黑色素瘤、乳腺癌和肺癌的中枢神经系统转移发展有效的证据，因此在深部隔室中定量此类低浓度水平具有临床相关性。⁷⁶⁻⁸⁰ 药物在CSF中的定量也可能有助于更好地理解BBB渗透性并计算药物穿透率。⁸¹,⁸² 由于血浆和CSF是不同的复杂生物基质，可能需要开发特定方法用于小体积CSF中的药物定量。⁸³⁻⁸⁵ 然而，当由于实际或伦理原因无法获得实际空白基质用于校准样品制备时，目标PKIs的稳定同位素标记内标可以克服生物基质差异，从而需要使用替代基质。

干血斑（DBS）是另一种最近成功应用于PKIs TDM临床实践的样本采集技术。⁸⁶⁻⁸⁸ DBS采样⁸⁹ 仅需将少量体积分配到纸卡上，与LC-MS/MS方法的分析灵敏度轻松兼容。此外，对于因频繁采血和/或化疗给药导致外周血管难以触及的癌症患者，这种方法比常规静脉采血侵入性更小的优势。然而，这种采样方法带来了多项已充分描述的生物分析挑战。⁹⁰,⁹¹

**观察性药代动力学研究**

PKIs在癌症患者中的药代动力学在药物开发的早期阶段进行研究。I期临床研究旨在提供不同剂量的PK谱和参数，但这些研究的局限性在于由少量同质癌症患者组成，不能代表"真实世界"癌症人群的多样性。因此，在现代临床开发和上市后监测中，基于群体药代动力学（popPK）方法表征PKIs浓度随时间的变化正越来越多地被使用（如阿法替尼⁹²、色瑞替尼⁹³和考比替尼⁹⁴）。

使用非线性混合效应回归模型的popPK研究明显优于传统PK研究，原因有多种。⁹⁵ PopPK可以分析从大量患者收集的数据，血液样本可以在给药后的不同时间（稀疏采样）从接受不同剂量的门诊癌症患者中采集（观察性分析）。非限制性设计有利于癌症门诊患者的参与。PopPK建模旨在量化研究人群中的PK参数，以及个体内（组内）和个体间（组间）血浆药物水平的变异性。它还旨在识别和评估影响药物浓度的各种潜在因素的贡献。此外，popPK模型可用于模拟描述多种情景下药物预期浓度-时间曲线的百分位数曲线（如标准或替代给药方案、次优依从模式或存在药物相关相互作用）。该模型对于贝叶斯治疗监测至关重要。

一项popPK分析发现，低体重患者（由于分布和消除的改变诱导）的仑伐替尼暴露较高，并指出74%的晚期肝细胞癌（HCC）患者因毒性而早期减量或停药。⁹⁶,⁹⁷ 因此，正式推荐根据该人群的体重进行剂量调整（体重≥60 kg的患者每日12 mg，体重<60 kg的患者每日8 mg）。⁹⁸

**PK荟萃分析**

在临床药代动力学领域，荟萃分析方法可用于合并多项PK研究的结果。⁹⁹ 事实上，当系统文献检索中获得足够的定量PK参数时（即使报告不统一），荟萃分析可以从比单项研究更大的患者集中得出平均PK参数。在大数据时代，随着文献中发表的PK分析数量不断增加，这种方法可能代表了PKIs合理TDM开发的创新方法。¹⁰⁰,¹⁰¹ 最终，开发的"元模型"将被整合到TDM软件中，可以指导TDM模型的选择并加强给药方案调整程序。¹⁰²

**PK-PD关系分析**

如果血浆浓度与药物的治疗反应（疗效或毒性）相关，则TDM的应用最为有效。

与治疗反应最常相关的PK参数是谷浓度（Cmin），较少见的是AUC。¹⁰³ 此外，一些PKIs具有药理活性代谢物，对肿瘤学效应有实质性贡献。在这种情况下，PK-PD分析必须考虑代谢物的浓度。例如，最近对舒尼替尼¹⁰⁴、奥希替尼¹⁰⁵和达拉非尼¹⁰⁶的分析将其各自的代谢物（SU12662、AZ5104和羟基达拉非尼）整合到了popPK和群体PK-PD模型中。

PK-PD分析可以通过群体模型进行，¹⁰⁷ 考虑影响药物治疗反应的潜在因素。在帕博西尼的PK-PD模型中研究其浓度与中性粒细胞减少的关系时，患者性别和基线白蛋白水平被整合为基线绝对中性粒细胞计数变异性的显著贡献因素。¹⁰⁸ 许多其他研究已经调查了PKIs的PK-PD关系；然而，对于最新的药物（如色瑞替尼、考比替尼、达拉非尼、仑伐替尼、奥希替尼和瑞博西尼），缺乏关于TDM应用的数据。一篇综述最近报告了FDA批准的二十多种PKIs的暴露-疗效/毒性关系。¹⁰⁹ 然而，给定PKI的PK-PD关系研究结果常常不同，这阻碍了治疗靶点的稳健识别。如Kim等人所述，¹¹⁰ 这些差异可能由这些分析的复杂性、PK/PD终点的选择以及潜在的肿瘤特异性PK-PD关系来解释，这也可能取决于激酶突变状态。对于PKIs的特异性TDM，治疗窗尚未正式建立，因此常见建议是达到最低PKI浓度阈值以改善治疗疗效。

**随机对照研究**

目前，TDM应用于PKIs是基于该方法的安全性和临床可行性¹¹¹⁻¹¹³，尽管大多数PKIs缺乏临床证据（伊马替尼除外¹¹⁴,¹¹⁵，如下所述）。TDM的临床益处确实应通过前瞻性、随机对照临床试验正式证明，比较接受推荐标准剂量的患者组与通过TDM接受个体化调整剂量的患者组之间预定义的临床疗效和耐受性终点。¹¹⁶

需要使用随机对照设计对PKIs等挽救生命的治疗进行大规模、循证评估。¹¹⁷ 然而，在真实临床环境中，这些类型的研究通常被患者和从业者视为不道德，因为对照组被剥夺了可能挽救生命的策略。因此，需要替代研究设计来评估TDM益处。¹¹⁸ 例如，伊马替尼浓度监测评估（I-COME）试验旨在前瞻性地建立常规应用TDM预防不良结果的科学证据。¹¹⁴ I-COME试验未能正式证明伊马替尼"常规TDM"的益处，特别是由于患者数量少（n=56）和处方医生对剂量建议的依从性有限。然而，发现了一个趋势，表明医生实际实施建议剂量调整的患者不仅更常达到目标浓度，还更常达到预定的综合结局评分（疗效、耐受性和持续性的组合）。

然而，量化TDM临床益处的研究样本量通常较小，因此迫切需要更多患者以达到足够的统计效力。例如，一项旨在通过TDM优化帕唑帕尼剂量的前瞻性研究因缺乏患者招募而提前终止。¹¹⁹

---

**4. 如何在临床实践中实施PKIs的TDM？**

由于缺乏暴露-疗效/毒性关系的信息、缺乏治疗靶点以及缺乏临床益处的明确证据，目前尚未建议常规监测PKI浓度。¹⁰⁹ 然而，在多种情况下TDM可能特别有用，如治疗效果差或严重/意外毒性；近期联合给药可能影响PKI药代动力的药物；疑似治疗依从性差；患者病理生理状态改变（如肝或肾功能损害）；或脆弱性（如高龄、一般状况差和肌少症）。鉴于PKIs的疗效通常在开始后两到三个月才评估，需要在治疗开始后尽早进行浓度测量；达到稳态以确保最佳暴露（即基于可用PK数据的预期范围内的血浆水平）。这种方法可以与PKIs的药效学监测相结合，基于生物标志物测量，这也是一种正在发展的方法。¹²⁰

**采样时间**

理想情况下，样本应在达到稳态后采集（即5个半衰期）。这意味着不同的等待时间，例如色瑞替尼需要超过一周（半衰期41小时¹²¹），而依鲁替尼仅需2天（半衰期4-6小时¹²²）。

最后一次PKI摄入时间和采血时间是记录结果解读的基本信息。谷浓度（Cmin）是临床实践中用于评估药物暴露充分性的常用靶点。¹⁰⁹ 然而，由于口服靶向抗癌药物主要由门诊患者使用，血液样本并不总是在谷值时采集，而是在确定的临床护理点采集，即PKI摄入后的非选择时间。因此，TDM结果通常不能直接与Cmin参考靶点比较；而是需要向谷值水平进行PK外推。Wang等人¹²³ 提出了一种基于药物消除常数外推伊马替尼Cmin的算法。这种线性外推方法常用于其他PKIs¹²⁴,¹²⁵，基于它们表现出线性消除（即无深部隔室）的假设以及样本在药物消除期采集的假设。然而，由于患者内变异性，这些假设可能不成立，从而影响方法的稳健性。

另一种方法是贝叶斯最大后验（MAP）估计。基于popPK模型、测得的血浆浓度和相关的患者特征，该方法允许估计个体PK参数并预测给药间隔内任何时间点的药物浓度。通过这种方法，还可以从单个测得的浓度推导AUC，这在某些情况下可能很有意义。¹²⁶ 凭借其在偏差和精度的良好预测性能，贝叶斯优化方法轻松辅助抗癌药物的TDM，¹¹⁶,¹²⁷,¹²⁸ 并可进一步开发用于众多当前和即将问世的PKI代次。这些外推方法提供了采血时间的灵活性，从而促进了TDM在临床实践中的转化。¹⁰³

**TDM解读和剂量个体化**

TDM结果的解读和后续剂量个体化应遵循一个已建立的三步流程⁷¹（图1）。

第一步包括通过考虑药物给药方案和确定的患者特征来评估测得浓度的正常性。通过与基于人群的参考百分位数曲线或从传统PK分析得出的平均暴露进行比较，评估该浓度的"预期性"。如果测得的浓度低于或高于预期，必须全面搜索这种变异性的来源（如药物相互作用、依从性问题）。在存在已知影响PKI药代动力的因素的情况下，了解其影响程度对解读和后续剂量调整很有用。例如，由于吸烟者的厄洛替尼Cmin比非吸烟者低50%，在吸烟患者中可谨慎考虑最大2倍的给药方案增加。¹²⁹ DDI预测器是一个可在临床实践中用于定量预测药物相互作用的网络工具（http://www.ddi-predictor.org¹³⁰）。

第二步包括评估浓度的"适宜性"，即考虑已知治疗靶点，测得的浓度是否充分。在缺乏循证治疗靶点的情况下，可基于临床前数据使用预测抑制靶点的浓度（IC）作为解读参考，如Verheijen等人为某些PKIs所建议的。¹⁰⁹ 推荐剂量下的平均或中位群体暴露也可作为有效PKI浓度的替代指标。事实上，最近一项分析强调，81.7%的基于TDM的浓度靶点与平均群体PKI暴露相匹配。¹⁰⁹ 此外，如果根据治疗靶点测得的浓度是适当的，但患者缺乏临床疗效，重要的是寻找PKI耐药性。在这种情况下，必须考虑换药。

第三步是在必要时调整当前药物剂量以达到治疗靶点（后验调整）。这可以通过一些手动策略进行，如剂量与浓度之间的简单三法则，或使用TDM引导软件（见下文）。理想情况下，PKIs的剂量个体化也应在治疗开始前进行（先验调整）。事实上，为了估计达到治疗靶点所需的初始剂量，可以使用基于popPK模型的贝叶斯方法来估计个体PK参数，考虑已知影响PKI药代动力的患者特征。然而，由于每个模型中存在无法解释的残余变异性，必须在稳态下密切监测后续浓度，并逐步调整剂量以达到治疗靶点。¹¹⁶

**血浆蛋白结合的陷阱**

除少数例外，PKIs与循环血浆蛋白高度结合（通常至少90%），主要与白蛋白和α-酸性糖蛋白（AGP）结合。¹³¹ 然而，由于技术和成本限制，常规分析方法通常测量PKIs的总血浆浓度，包括（主要的）血浆蛋白结合部分与（通常较低的）游离PKI之间的平衡。然而，只有后者可能穿透细胞并发挥其药理作用。

此外，血浆蛋白结合水平可能对PKIs的分布和清除产生显著影响，导致药物浓度变异性，如伊马替尼¹³²和厄洛替尼¹³³以及最近开发的PKIs如阿昔替尼¹³⁴和奥希替尼¹⁰⁵所观察到的。在蛋白水平波动的患者中，基于总血浆浓度测量的TDM结果解读必须考虑这一额外的复杂性水平；例如，在癌症急性期AGP水平升高¹³⁵或在肝功能损害或营养不良情况下出现低白蛋白血症。对于具有高AGP结合和相当低肝脏提取率的伊马替尼等PKI，AGP水平升高可能导致伊马替尼总血浆浓度增加，而游离浓度不变，从而导致潜在的解读不当和不合理的减量。¹³²

因此，在蛋白结合严重改变的情况下评估游离PKI浓度将提高TDM解读的质量。游离PKI浓度可以通过分析方法测量（即通常的超滤¹³⁶），或基于总浓度通过数学模型外推。¹³⁷

**用药依从性问题**

用药依从性定义为患者按处方服药的过程。它由三个特征组成：治疗的启动、实施和停药。实施定义为患者逐日按正确剂量、正确时间服用正确药物的程度，持续性由治疗启动至完全停药的时间段界定。¹³⁸

用药依从性是口服PKI治疗成功的主要决定因素。¹³⁹,¹⁴⁰ 尽管癌症危及生命，但长期口服靶向抗癌治疗的依从性可能不理想。¹⁴¹⁻¹⁴³ 影响PKI用药依从性的多种复杂、动态和相互关联的决定因素¹⁴¹,¹⁴⁴，如治疗相关不良作用或复杂性、患者的替代治疗信念、焦虑、抑郁、缺乏社会支持或经济负担。多项用药依从性计划目前探索支持患者优化PKI自我管理的干预措施。¹⁴⁵⁻¹⁵⁰

用药依从性是TDM解读中需要考虑的重要患者行为，因为它代表潜在的偏倚来源。事实上，偏离处方方案是药物暴露变异（暴露不足或过度）的来源，可能导致不适当的剂量调整。良好的依从性监测将消除暴露不足时对依从性差的疑虑，以及过度暴露时计划就诊前的过度依从（"白大褂依从性"）的疑虑。

此外，由于PKI摄入周围的条件可能逐日变化，这可能影响测得的浓度，因此在采血期间应系统收集患者信息，如最后一次PKI摄入的日期和时间、食物摄入的时间和类型，以及具有周期性给药方案的PKIs的实际治疗日（如舒尼替尼、瑞戈非尼、考比替尼、帕博西尼和瑞博西尼）。¹⁵¹ 值得注意的是，常规TDM项目中关于PKI依从性对血浆浓度影响的知识仍然缺乏。另一方面，从药代动力学角度估计患者依从性特征的研究正在探索中，旨在从单个TDM样本估计个体对伊马替尼和其他抗癌药物的依从性。¹⁵²,¹⁵³

**计算机辅助**

TDM在临床实践中被认为是一种复杂的方法，因为需要收集大量信息以及进行最先进的剂量调整所需的复杂计算。由于近期兴趣日益增长，已开发多种软件应用来促进TDM的日常实践¹⁵⁴，通过支持临床医生解读药物浓度并指导剂量调整决策。

MwPharm++、InsightRX以及最近由我们洛桑团队开发的Tucuxi¹⁵⁵ 是近期TDM引导软件的例子。通过整合popPK数据和PK/PD参数，该软件可以评估考虑药物和患者特定特征后测得浓度的预期性和适宜性，并可通过贝叶斯计算提出剂量调整方案。此类软件可以与医院信息系统（如电子病历）接口（如患者临床历史和实验室信息）。此外，它们应直观、稳健且用户友好，具有交互式图形显示。

---

**5. 结论**

在过去十年中，随着口服靶向抗癌药物（特别是PKIs）的出现，癌症治疗进入了一个新时代，这些药物针对癌症特异性分子和信号通路。然而，耐药性、癌症干细胞持续存在和药物不良作用仍然限制其长期稳定或治愈恶性疾病的能力。尽管存在显著的患者间PK变异性，PKIs基本上以固定剂量获批。然而，由于其跨患者的广泛药代动力学变异性和从疗效到毒性的敏感暴露平衡，它们是通过TDM项目进行剂量个体化的合适候选药物。

在个体化医疗努力不断发展的背景下，肿瘤患者以及医疗系统当然值得在引导这些药物方面进一步优化和个体化。¹⁵⁶ 特别是，迫切需要前瞻性随机对照试验来评估新PKIs的TDM证据水平，因为智能、用户友好的计算机系统的发展前景广阔。此类计算机系统将协助临床医生进行TDM实践，克服已知的实际问题，如药物浓度的随机测量、患者对PKIs的实时依从性以及血浆蛋白结合模拟。

因此，适当的TDM临床研究和实践应纳入现代个体化癌症护理¹⁵⁷，与药效学监测和药物遗传学方法相结合。由于"每位肿瘤患者都不同"，这种优化应强调"以正确的剂量将正确的抗癌药物给予正确的患者"在常规临床环境中成为可能。

---

**6. 致谢**

本工作仅由内部资金支持。TB、LAD和CC是Sotalya Inc.的联合创始人，该公司旨在开发支持TDM的软件工具。MPS和CC获得了Accentus和瑞士癌症研究基金会的资助，用于开展优化靶向抗癌治疗的研究（HSR-4077-11-2016）。NW是罗氏和辉瑞瑞士医院药师顾问委员会成员，与本综述主题无关，并于2018年获得罗氏的差旅资助。