Pharmaceutics of Oral Anticancer Agents and Stimulants

Pharmaceutics of oral anticancer agents and stimulants

Maikel Herbrink The research in this thesis was performed at the Departments of Pharmacy & Pharma- cology and Medical Oncology & Clinical Pharmacology of the Netherlands Cancer Insti- tute – Antoni van Leeuwenhoek hospital, Amsterdam, the Netherlands.

Printing of this thesis was financially supported by: Netherlands Cancer Institute Chipsoft BV Pfizer BV

ISBN: 978-90-393-6997-5

Design, lay-out and print Gildeprint, Enschede

Farmaceutisch onderzoek naar oraal toegediende antikankermiddelen en stimulantia

(met een samenvatting in het Nederlands)

Proefschrift

ter verkrijging van de graad van doctor aan de Universiteit Utrecht op gezag van de rector magnificus, prof. dr. H.R.B.M. Kummeling, ingevolge het besluit van het college voor promoties in het openbaar te verdedigen op woensdag 27 juni 2018 des middags te 4.15 uur

door

Maikel Herbrink

geboren op 2 februari 1990 te Groningen Promotoren: Prof. dr. J.H. Beijnen Prof. dr. J.H.M. Schellens

Copromotor: Dr. B. Nuijen

Contents

Preface 9 Chapter 1: Introduction 15

1.1 Variability in bioavailability of small molecular tyrosine 17 kinase inhibitors Cancer Treat Rev. 2015 May;41(5):412-422 1.2 Inherent formulation issues of kinase inhibitors 51 J Control Release. 2016 Oct;239:118-127 1.3 High-tech drugs in creaky formulations 81 Pharm Res. 2017 Sep;34(9):1751-1753

Chapter 2: Pharmaceutical analysis of oral anticancer drug substances 89

2.1. Thermal study of pazopanib hydrochloride 91 J Therm Anal Calorim. 2017 Dec;130(3):1491-1499 2.2. Thermal stability study of crystalline and novel spray- 111 dried amorphous nilotinib hydrochloride J Pharm Biomed Anal. 2018 Jan;148:182-188

Chapter 3: Formulation improvement of oral anticancer agents 131

3.1 Solubility and bioavailability improvement of pazopanib 133 hydrochloride Accepted for publication in Int J Pharm. 3.2 Improving the solubility of nilotinib through novel spray- 159 dried solid dispersions Int J Pharm. 2017 Aug;529(1-2):294-302 Chapter 4: Bioanalysis of oral anticancer agents 183

4.1 Quantification of 11 therapeutic kinase inhibitors in 185 human plasma for therapeutic drug monitoring using liquid chromatography coupled with tandem mass spectrometry Ther Drug Monit. 2016 Dec;38(6):649-656 4.2 Development and validation of a liquid chromatography- 205 tandem mass spectrometry analytical method for the therapeutic drug monitoring of eight novel anticancer drugs Biomed Chromatogr. 2018 Apr;32(4)

Chapter 5: Dexamphetamine 227

5.1 Development and validation of a high-performance liquid 229 chromatography-tandem mass spectrometry assay for the quantification of dexamphetamine in human plasma J Pharm Biomed Anal. 2018 Jan;148:259-264 5.2. Pharmacokinetics of sustained-release oral 245 dexamphetamine sulphate in cocaine and heroin dependent patients Accepted for publication in J Clin Psychopharmacol.

Chapter 6: Conclusion & Perspectives 261

Appendix Summary 275 Nederlandse samenvatting 283 Dankwoord 291 List of publications 297 Curriculum vitae 303 Molecular structures 307

Preface

For the past 20 years, the pharmacological therapy of cancer has been undergoing a metamorphosis, rapidly changing from treatment with broad and unspecific cytotoxic agents to personalized and highly specific targeted agents. This new class of drugs largely consists of small molecular compounds that mostly inhibit kinases. Through their disruption of the kinase function, the targeted agents interfere with multiple cancer signaling pathways which are essential to tumor development, survival, growth, and metastasis [1].

Alongside the advancing targeted therapy came the so-called ‘intravenous-to-oral-switch’. This is illustrated by the fact that an increasing number of marketed targeted anticancer drugs are administered orally. Additionally, most of the targeted agents that are being investigated are intended for oral ingestion.

Part I of this thesis addresses the issues that originate from the oral administration of targeted anticancer agents.

Chapter 1 introduces the physicochemical and related pharmacokinetic properties of the orally administered kinase inhibitors. Chapter 1.1 presents the variability in the pharmacokinetics of this class of drugs. Furthermore, it addresses the therapeutic consequences that are associated with this generally unfavorable behavior. Chapter 1.2 discusses the inherent link between the targeted nature of kinase inhibitors and their suboptimal biopharmaceutical properties. Additionally, it shows through literature and patent data that the currently marketed drug formulations are mostly inadequate in overcoming the physicochemical challenges. Chapter 1.3 elaborates on both the previous chapters and takes a strong stance in favor of investing more effort in pharmaceutical formulation development of future oral drugs.

In the process towards a well-designed pharmaceutical formulation, the characterization of the drug substance plays a pivotal role. Chapter 2 describes the thermal preformulation studies of pazopanib hydrochloride and nilotinib hydrochloride. Chapter 2.1 focusses on the thermal degradation profile of crystalline pazopanib hydrochloride and examines future processing possibilities. In Chapter 2.2 we describe the thermal and thermodynamic properties of crystalline and amorphous nilotinib hydrochloride and the stability thereof.

Chapter 3 describes the development of improved pharmaceutical formulations of both pazopanib hydrochloride and nilotinib hydrochloride.Chapter 3.1 reports on the solubility and bioavailability improvement of pazopanib hydrochloride through the formulation of a physical powder mixture formulation. It presents data from in vitro and human in vivo studies. Chapter 3.2 describes the design and development process of a spray-dried formulation that contains amorphous nilotinib hydrochloride.

Suboptimal pharmaceutical formulations that have already been marketed andare routinely used in the clinic often require therapeutic drug monitoring (TDM) to timely detect under- or overdosing. Chapter 4 provides two fully validated LC-MS/MS assays for the quantification of a broad collection of targeted anti-cancer agents inpatient plasma for TDM purposes. The bioanalytical assay described in Chapter 4.1 focusses on the quantification in patient plasma of dasatinib, erlotinib, gefitinib, imatinib, lapatinib, nilotinib, pazopanib, sorafenib, sunitinib, and vemurafenib and the active metabolite, N-desethyl-sunitinib. Chapter 4.2 presents a method for the quantification of afatinib, axitinib, ceritinib, crizotinib, dabrafenib, enzalutamide, regorafenib, and trametinib in patient plasma.

Part II of this thesis focusses on the manufacturing and pharmacokinetic analysis of dexamphetamine SR tablets that were specifically developed for use in cocaine addiction therapy.

Cocaine dependence is a harmful mental disease for which there currently is no pharmacotherapy available. Substitution with the safer and regulated dexamphetamine sustained-release (SR) has been shown in several studies to be a feasible option for patients with this type of addiction [2]. A Dutch study has demonstrated that dexamfetamine SR is an effective and safe pharmacotherapy for comorbid treatment-refractory cocaine dependence in heroin-dependent patients in heroin-assisted treatment [3].

Chapter 5 describes the preparation and execution of a descriptive pharmacokinetic- ex ploration study of dexamphetamine SR tablets. Chapter 5.1 presents a fully validated LC-MS/MS assay for the quantification of dexamphetamine in patient plasma samples which was used in the pharmacokinetic study.Chapter 5.2 describes the results and the discussion of the pharmacokinetic study itself. References

[1] Zhang J, Yang PL, Gray NS. Targeting cancer [3] Nuijten M, Blanken P, van de Wetering B, et al.. with small molecule kinase inhibitors. Nat. Rev. Sustained-release dexamfetamine in the treat- Cancer. 2009;9:28–39. ment of chronic cocaine-dependent patients [2] Castells X, Cunill R, Pérez-Mañá C, et al. Psy- on heroin-assisted treatment: a randomised, chostimulant drugs for cocaine dependence. double-blind, placebo-controlled trial. Lancet. In: Castells X. Cochrane Database Syst. Rev. 2016;387:2226–34. Chichester, UK: John Wiley & Sons, Ltd; 2016.

Chapter 1

Introduction

Chapter 1.1

Variability in bioavailability of small molecular tyrosine kinase inhibitors

M. Herbrink1, B. Nuijen1, J.H.M. Schellens2,3, J.H. Beijnen1,3

1Department of Pharmacy and Pharmacology, The Netherlands Cancer Institute – Antoni van Leeuwenhoek, Amsterdam, The Netherlands. 2Department of Medical Oncology and Clinical Pharmacology, The Netherlands Cancer Institute – Antoni van Leeuwenhoek, Amsterdam, The Netherlands. 3Department of Pharmaceutical Sciences, Utrecht University, Utrecht, The Netherlands. 18 | Chapter 1.1

Abstract

Small molecular tyrosine kinase inhibitors (smTKIs) are in the centre of the very quickly expanding area of personalized chemotherapy and oral applicability thereof. The number of drugs in this class is rapidly growing, with twenty current approvals by both the European Medicines Agency (EMA) and the Food and Drug Administration (FDA). The drugs are, however, generally characterized by a poor oral, and thus variable, bioavailability. This results in significant variation in plasma levels and exposure. The cause is a complex interplay of factors, including poor aqueous solubility, issued permeability, membrane transport and enzymatic metabolism. Additionally, food and drug–drug interactions can play a significant role. The issues related with an impaired bioavailability generally receive little attention. To the best of our knowledge, this article is the first to provide an overview of the factors that determine the bioavailability of the smTKIs. Variability in bioavailability of small molecular tyrosine kinase inhibitors | 19

Introduction 1.1 The development of anticancer drugs is a very quickly expanding area in which two trends are clearly present. In the first place new agents are designed fulfilling the requirements for personalized medicine. The advancement of techniques such as (cell-based) high throughput screening and the diverse possibilities in molecular modeling have led to a therapeutic target-based drug discovery regime [1,2]. Along with the evolution of syn- thetical methods, compounds are found that are highly specific and demonstrate great affinity for molecular targets [3–6]. Individual tumors, and their specific targets, can be genetically characterized and a suitable ‘personalized’ chemotherapy can be appointed depending on the neoplasm’s genotype [7]. The second movement is also referred to as ‘the intravenous to oral switch’. The last decade has shown an increasing number of anticancer drugs that are administered orally [8,9]. Currently, most of the anticancer drugs that are in development or recently approved are destined for oral ingestion. Unlike previous conventions, oral therapy in cancer has proven efficient and less costly [10]. On top of that comes the preference of the patient, especially since oral ingestion can take place in the home setting and is highly convenient compared to intravenous administration [8].

In the middle of these trends stands a promising and growing group of drugs; the tyrosine kinase inhibitors (TKIs). In the past ten years, the size of this group has doubled [11–13]. The TKIs target specific parts of tyrosine kinase receptor proteins that play an important role in the intracellular signaling pathways in tumor cells. Their interference leads to a deregulation of essential cell functions such as proliferation and differentiation [14]. One of the two types of TKIs, the small molecular TKIs (smTKIs) with an intracellular activity, are without exception administered orally. Currently, twenty of these small molecular compounds are approved by both the EMA and the FDA. General information on the drugs is found in table 1 [11,12]. This review will focus on these particular compounds. The other type of TKI is a group of monoclonal antibodies, which possess a larger molecular structure and interfere with signal transduction by binding extracellularly and are administrated intravenously. The small molecular inhibitors have proven useful in the therapy of certain types and lines of cancer [6,11,12,15]. Additionally, smTKIs may be prescribed as alternatives when other therapeutic options have failed or are not appropriate. Although the development of personalized oral chemotherapy is very promising, the nature of the selection process leads to drugs, however, that are hindered by a low and variable bioavailability (F). This aspect and its causes are underexposed 20 | Chapter 1.1

subjects in literature. Indeed, smTKIs may be very potent and suitable for certain tumor types. When they are unable to reach their target in sufficient quantities, the therapy will be suboptimal or even failing. This review will address the bioavailability-determining factors for the smTKIs and presents prerequisites in both the marketed formulations and chemotherapeutical practice to minimize the reduction and variation in oral F. It is important to be aware of and understand the various factors that determine F and its variability of the smTKIs. This will allow for the betterment of their use in chemotherapy.

Critical processes: 1 - Dissolution Dosage 2 - Permeation and metabolism form 3 - Metabolism Portal vein

Liver

3 G Gastro-intestinal tract Systemic circulation

FF FF * FG FF * FG * FH

Figure 1. Schematic overview (simplified) of major bioavailability determinants and corresponding fractions (bottom scale)

Oral bioavailability The fraction of the total ingested drug that reaches the systemic circulation unchanged, and is transported to its therapeutic target, is defined and termed (absolutely) bioavailable (F) [16]. Figure 1 schematically presents the different processes that govern the extent of F. F is the product of the drug fraction that is absorbed (FF), the dose that reaches the hepatic portal vein unchanged (FG) and the fraction of the dose that is not metabolized by enzymes in the liver (FH), as presented in equation 1 [16,17],

(1) F = F# ∗ F% ∗ F& Variability in bioavailability of small molecular tyrosine kinase inhibitors | 21

Table 1. Overview of the general information on tyrosine kinase inhibitors approved for use by the EMA and the FDA by 1 November 2014. 1.1 Drug Trade name Primary indication(s) Oral bioavailability (%) Afatinib Giotrif NSCLC -4 Axitinib Inlyta RCC after failure with sunitinib/cytokines 47-58 Bosutinib Bosulif CML when imatinib, nilotinib and dasatinib - are not appropriate Cabozantinib Cometriq MTC - Crizotinib Xalkori NSCLC 43 Dabrafenib Tafinlar Melanoma 95 Dasatinib Sprycel CML - Erlotinib Tarveca NSCLC, pancreatic cancer 60-76 Gefitinib Iressa NSCLC 57-59 Imatinib Gleevec/Glivec CML, ALL, CEL, HES, MDS/MPD, GIST, DFSP 98 Lapatinib Tyverb HER-2+ breast cancer - Nilotinib Tasigna CML 30 Pazopanib Votrient RCC, STS 14-39 [166] Ponatinib Iclusig CML when imatinib, nilotinib and dasatinib - are not appropriate, ALL1 Regorafenib Stivarga CRC, GIST 69-83 Ruxolitinib Jakavi CIM, PPVM, PETM 955 Sorafenib Nexavar HC, RCC, DTC - Sunitinib Sutent GIST, MRCC, pNET - Vandetanib Caprelsa MTC - Vemurafenib Zelboraf Melanoma - ALL, acute lymphatic leukemia; CEL, Chronic eosinophilic leukemia; CIM, Chronic idiopathic myelofibrosis; CML, Chronic myeloid leukemia; CRC, Colorectal cancer; DFSP, Dermatofibrosarcoma protuberans; DTC, Dif- ferentiated thyroid carcinoma; GIST, Gastrointestinal stromal tumours; HC, Hepatocellular carcinoma; HES, Hypereosinophilic syndrome; MDS, myelodysplastic disease; MPD, myeloproliferative disease; MRCC, Met- astatic renal cell carcinoma; MTC, Medullary thyroid carcinoma; NSCLC, non-small cell lung cancer; PETM, post essential thrombocythaemia myelofibrosis; PPVM, post polycythaemia vera myelofibrosis; pNET, pancreatic neuroendocrine tumours; RCC, Renal cell carcinoma; STS, soft tissue sarcoma.1 Changes in useage are suggested by EMA due to life-threatening vascular events. 2 Approval by EMA was conditional. 3Withdrawn in 2005 due to lack of evidence in prolongation of life.4 -: Data were not available. 5Based on a mass balance study. Source: European Public Assessment Reports (EPARs), Website FDA and Summaries of product characteristics (SmPCs) of the above mentioned smTKIs, accessed at 20th november 2014.

In each of the before mentioned steps, an amount of drug might be lost. Whatever the cause, a low F is associated with an increased intra- and interpatient variability in drug plasma concentration [18]. Registration texts and other studies, as far as could be accessed, show significant inter-individual variation in important pharmacokinetic param- 22 | Chapter 1.1

eters of all smTKIs [19–37]. This may result in possibly dangerous situations for patients that experience extensively low, or high, exposure to the substances. Many anticancer drugs are known to exhibit a small therapeutic window, where the minimum therapeutic dose and the maximum tolerated dose (MTD) are close to each other [38]. The same is true for the smTKIs, with a possible exception of Dabrafenib, Imatinib, Gefitinib and Pa- zopanib [13,19,39–52]. As a consequence of pharmacokinetic variation, inter-individual differences in therapeutic dose and MTD should be taken into account. Under- and overdosing are thus potential hazards of oral chemotherapy. Thus, careful dose titration and adjustments are required to assure an adequate therapy, in both effect and tolerance. Hence, therapeutic drug monitoring (TDM) is upcoming for smTKIs [38].

The human oral F of the smTKIs is largely unknown or inaccessible in the public domain and published values are generally low and the exposure is variable [39,53,54]. The determination of oral F-values requires a comparison between oral and IV-administration. IV-solutions with smTKIs are often difficult to prepare due to the poor water solubility of the drugs (see 3.1). Table 1 presents the currently known values. Low values for oral F may be due to one or more of several factors. It is often the consequence of a complex interplay of both physicochemical and physiological processes. Furthermore, it may also be influenced by concomitant administration of other drugs. Additionally, the intake of food or certain habits of life-style may exert an impact on F.

B(DD)CS-classification The Biopharmaceutics Classification System (BCS) can aid to clarify possible absorption-related causes of an impaired F. Solubility and permeability of a drug are recognized as fundamental parameters in the absorption process [55]. The BCS combines data on the in vitro solubility in the intestinal tract of the drug substance and data on the extent of total permeation through the gut wall and appoints a class to it [55–57]. Figure 2 summarizes the assignment of the classes and presents the classes of the smTKIs [37,58,59]. Classifications may be interpreted as signals for formulation design (class II and IV) or physicochemical modifications (class III and IV) [58].

The newer Biopharmaceutics Drug Disposition Classification system (BDDCS) correlates the passive permeability rate of drug with their metabolic elimination[60–62]. Here, passive permeability is considered ‘good’ when elimination is largely governed by metabolism (>70%)[63]. Figure 2 presents the BDDCS classification between braces where it differs from the BCS classification. The discrepancies between the BCS andBDDCS Variability in bioavailability of small molecular tyrosine kinase inhibitors | 23

classes may be due to the fact that BCS is based on total permeation and BDDCS on the passive permeability rate [64]. The latter does not account for interaction with mem- 1.1 brane transporters.

Figure 2. The Biopharmaceutics Classification System (BCS) as proposed by Amidon et al [56]. and the Bio- pharmaceutics Drug Disposition Classification System (BDDCS) as proposed by Wu et al [61]. B(DD)CS classifications of smTKIs indicated with three-letter abbreviations. 1I/III; in vitro data were inconclusive. 2I/II; Inconclusive solubility data. 3II/IV; conflicting permeability data. Source: EPARs, additional sources indicated where appropriate. 24 | Chapter 1.1

Pharmaceutical factors (FF)

Dissolution The first step in becoming bioavailable is the dissolution of the drug substance into the gastrointestinal fluids. Since only the solute form of the drug can be absorbed, the -re lease from the oral formulation is an important parameter. In fact, the major cause for the different absorption profiles of drugs from various products is dissolution [16]. The dissolution rate of a drug substance can be described using the Noyes-Whitney equation (equation 2) [65,66]:

(2) !" ()*( !# = D ∗ A ∗ + in which dW/dt is the dissolution rate (mg/min), D the diffusion coefficient (cm2/min), A 2 the surface area of the dissolving compound (cm ), Cs the saturation concentration (concentration in the diffusion layer, mg/L), C the concentration in the solvent (mg/L) and h the thickness of the diffusion layer (cm) [66].

The diffusion coefficient and the thickness of the diffusion layer are dependent on highly variable patient parameters, such as gastrointestinal pH and viscosity [67–69]. Certain excipients in formulations may be used to manipulate drug diffusivity [70]. These are not found in smTKI formulations. The following will discuss solubility and surface area, which are both most frequently modified in formulations of smTKIs.

Solubility The degree of a compound’s solubility is dependent on physicochemical factors of the solute and the solvent. The gastrointestinal fluids are aqueous and therefore, the water solubility at various pH-values of a drug is used to predict its apparent solubility in the GI-tract [71]. The small intestine has the greatest of surfaces of the organs in the gastrointestinal tract. It is here where most of the smTKIs are considered to be absorbed. The transit time of materials in the small intestine is approximately 3 to 4 hours [72], in which the absorption must take place. For these reasons, drug solubility in the small intestine is a prerequisite. Variability in bioavailability of small molecular tyrosine kinase inhibitors | 25

Table 2. Overview of physical characteristics of the commercially available dosage forms of the smTKIs. Drug Commercial salt form pK Reference a 1.1 (crystal state1) Tablet formulations Afatinib Di-maleate (A) 5.0, 8.2 Axitinib Free base (XLI) 4.8 [21] Bosutinib Monohydrate (1) n.a. Dasatinib Monohydrate (H1-7) 3.1, 6.8, 10.8 [164] Erlotinib Hydrochloride (A) 5.4 Gefitinib Free base2 5.4, 7.2 Imatinib4 Mesylate (beta / amorphous3) 1.5-8.1 [165] Lapatinib Ditosylate monohydrate2 n.a. Pazopanib Hydrochloride (1) 2.1, 6.4, 10.2 Ponatinib Hydrochloride2 2.8, 7.8 Regorafenib Co-precipitate (amorphous) n.a. Ruxolitinib Phosphate anhydrate2 4.3, 11.8 Sorafenib Tosylate (I) n.a. Vandetanib Free base2 5.2, 9.4 Vemurafenib Co-precipitate (amorphous) 7.9, 11.1 Capsule formulations Cabozantinib L-malate (N-2) n.a. Crizotinib Free base (A) 5.6, 9.4 Dabrafenib Mesylate (1) 2.2, 6.6 Nilotinib Hydrochloride 2.1, 5.4 monohydrate (B) Sunitinib Malate (I) 8.9 [166] 1names of polymorphisms presented as in the texts of EPARs and/or patents. 2Data on polymorphisms not available. 3Generic formulations contain either of both. 4Both tablet and capsule formulations marketed. Source: EPARs and FDA chemistry reviews, additional sources indicated where appropriate. 26 | Chapter 1.1

The pH value in the intestinal lumen rises from pH 2 near the stomach to pH 5-6 in the jejunum, ending at pH 7-8 near the ileocecal valve [73]. The acidity of the GI-fluids greatly influences the solubility of the smTKIs. Most of the smTKIs are weak bases (pKa-values in table 2) which are protonated, and as such most soluble, in acidic environments [74]. The increase in pH-values encountered in the intestines dramatically decreases their solubility [75]. The result of this is a relative high solubility in the stomach, but a significantly lower one in the intestinal lumen. Consequently, the basic compounds classify as B(DD)CS II or IV, due to this issued solubility at higher pH-values. Another explanation for the poor solubility behavior is the general high lipophilicity of the smTKI group [11,76]. Based on structural similarity it is reasonable to assume that the group as a whole has a lipophilic character. Ruxolitinib stands out from the rest; it is relatively water-soluble [11].

Salt formation improves water solubility of compounds, either acidic or basic, because it allows the substances to dissociate in hydrated ions [77]. Salt forms of the marketed formulations are presented in table 2 [11,19]. The choice of the counterion is not only based on solubility but also with regard to the stability of the compound and its tendency to be hygroscopic [78,79].

Many of the smTKIs exhibit crystalline polymorphism [11,19,76]. The particular crystal structure chosen for the pharmaceutical product offers the best combination of stability and solubility. Besides crystal structures, solid compounds can arrange themselves in a more loosely-ordered amorphous state [80]. The advantage of the amorphous form is the higher apparent solubility, owing to the high levels of supersaturation in aqueous media [81]. This form is thermodynamically unstable relative to the crystal and generally has a tendency to revert back to a crystal form [82]. The amorphous form must thus be stabilized by the formulation that contains it. Regorafenib and vemurafenib were found to be poorly soluble in any of the investigated crystal forms. The marketed tablets with these compounds hold solid solutions with amorphous states of the drugs [11]. Table 2 lists the crystal polymorphisms used in the marketed formulations. It has been proposed that hydroxypropyl methylcellulose (HPMC) may inhibit dabrafenib precipitation in vivo [83] by maintaining supersaturation. This could possibly be the case with other capsulated smTKIs as well. Variability in bioavailability of small molecular tyrosine kinase inhibitors | 27

Surface area The surface area of the compound in contact with the solvent is directly proportional 1.1 to the dissolution rate [84]. It can be manipulated by increasing or decreasing particle size. Smaller particle sizes do not only demonstrate a higher dissolution rate due to the enlarged surface area. A more efficient wettability or wetting, the degree with which the drug comes into contact with the solute, also contributes [85]. As far as could be assessed, it was found that size distribution of the drug particles in the commercialized formulations of the smTKIs were optimized in relation to dissolution properties [11]. Further formulation modifications were made to tablets to ensure a fast and complete disintegration, enhancing the in vivo surface area. Five of the smTKIs are marketed as capsules. The capsules contain a powdered formulation that rapidly wettens when it comes in contact with the gastrointestinal fluids [11]. Table 2 summarizes the oral dosage forms of the smTKIs. 28 | Chapter 1.1

Physiological factors (FG & FH)

Permeability and passive transport The degree of permeability describes the extent of (intestinal) membrane penetration of a drug [86]. The passing of molecules occurs through passive or facilitated diffusion, (active) carrier mediated or paracellular transport [87,88]. Lipophile, large and relatively uncharged drugs are moved transcellularly through cell membranes. This passive transport is a or the major route of movement across membranes of many drugs [86,89]. The same is reported for the smTKIs, although this is not expressibly described. Support for this is found in the fact that most smTKIs are classified as class I or II in the BDDCS (figure 2). Considering the lipophile molecular structure of the smTKIs, transcellular transport seems the most probable route for passive movement [11,76].

Carrier-mediated transport One of the physiological factors that can have a great impact on the absorption from the intestinal lumen is the presence of membrane drug transporters [90,91]. The intestinal epithelial lining expresses a large number of transporter proteins [92]. Two major superfamilies, the ATP binding cassette (ABC) transporters and the solute carrier (SLC) transporters, are located in the intestinal wall [93,94]. Transporters can either facilitate uptake or by efflux from the intestinal cells back into the intestinal lumen [90,95].

A subfamily of the SLC transporter, the organic anion transporting proteins (OATPs) is expressed in the small intestine in several isoforms [96,97]. The OATP group of transporters provides mediation for ionic agents in their intestinal absorption, which results in a higher uptake rate than would be expected diffusion-wise. Most of the smTKIs have been reported to be substrates of an isoform of OATP, although clinical relevance has not been established [11,12], table 3.

Of special interest are the ATP binding cassette (ABC) transporters, since these are most abundant and most of the smTKIs are substrates of these proteins. The ABC-transporters can be subdivided into three families that are involved in the transport of anti-cancer drugs in the intestines: permeability-glycoprotein (P-gp), the multidrug resistance-associated proteins (MRP) and the breast cancer resistance protein (BCRP) [98].

P-gp transporters are present in the membranes of cells throughout the entire human body. It is believed to have a function in removing toxins and metabolites out of cells Variability in bioavailability of small molecular tyrosine kinase inhibitors | 29

[95].The presence of P-gp in the membrane on the luminal side of the gut wall can prove troublesome for drug absorption. It functions as an efflux pump with a broad substrate 1.1 range with an overall preference for hydrophobic and amphiphatic molecules. It includes most of the smTKIs, as listed in table 3.

Table 3. Demonstrated substrate specificity of major intestinal transporter proteins for smTKIs and demonstrated enzyme interactions of smTKIs. Drug Transporters Enzymes Reference(s) Afatinib P-gp, BCRP - [167] Axitinib OATP, P-gp, BCRP 3A4/5, 1A2, 2C19, UGT [168,169] Bosutinib P-gp 3A4 [23,170] Cabozantinib -* 3A4, 2C8 [171] Crizotinib OATP, P-gp 3A4/5 [172,173] Dabrafenib P-gp, BCRP 3A4, 2C8 [24] Dasatinib OATP, P-gp, BCRP 3A4 [172,174] Erlotinib P-gp, BCRP 3A4/5, 1A2 [175–177] Gefitinib P-gp, OATP, BCRP 3A4/5, 2D6 [172,176,178] Imatinib P-gp, OATP, BCRP 3A4 [172,179,180] Lapatinib P-gp, OATP, BCRP 3A4/5, 2C19, 2C8 [181,182] Nilotinib P-gp, OATP, BCRP 3A4 [172] Pazopanib P-gp, OATP, BCRP 3A4, 1A2, 2C8, UGT [163,172] Ponatinib P-gp, BCRP 3A4 Regorafenib - 3A4, UGT [13] Ruxolitinib - 3A4 [183] Sorafenib OATP, P-gp, MRP-2, BCRP 3A4, UGT [172,184–187] Sunitinib OATP, P-gp, BCRP 3A4 [172,188,189] Vandetanib OATP, P-gp, BCRP 3A4 [172,190,191] Vemurafenib OATP, P-gp, BCRP 3A4 [172,192,193] *Research ongoing or data not available. Source: EPARs and FDA lable texts, additional sources indicated where appropriate.

The MRP transporter family has one principal member (MRP 2) expressed on the luminal side of the intestinal wall [99]. The substrate range includes, but is not limited to, organic anions. It functions in very much the same fashion as P-gp does, it pumps toxins and foreign drug molecules from the intestinal cells back into the lumen. Table 3 presents the known smTKI substrates of MRP 2. 30 | Chapter 1.1

BRCP transporters align with P-gp and MRP proteins in efflux function [100]. BCRP is present in both the small and the large intestine and its expression does not vary significantly per location [101]. The transport is also involved in the development of resistance to several classes of chemotherapeutics [102]. It is expressed on the luminal side of the intestinal epithelial cells. The range of substrates of BCRP varies greatly and has been shown to include a number of smTKIs, as presented in table 3.

As is the case with many substrates, the substrate specificity for smTKIs of the different efflux transporters overlaps for some substrates [11,12]. This can lead to a synergistic effect of the transporters in limiting drug penetration across the intestinal barrier [90].

Metabolism Metabolism is the major mechanism for elimination of drugs from the body [103–105]. The typical metabolic steps include oxidation, reduction, hydrolysis and conjugation. The first three are often referred to as phase I reactions and the latter as phase II, owing to the order in which the reactions frequently occur. Drugs can be metabolized before they reach the systemic circulation (first-pass metabolism) and after distribution through the body. Metabolism may render drug molecules inactive. On the other hand, metabolites may exhibit lower, higher or equivalent bioactivity compared to the parent compound [106]. The intestinal wall and the liver are responsible for the first-pass metabolism. Respectively, these organs determine the gastrointestinal and the hepatic F. Many of the enzymes involved in phase I and II reactions are present in both the liver and the gut wall, albeit that the liver generally contains larger amounts [107].

Cytochrome P450 (CYP) enzymes are responsible for the majority of phase I metabolism of a enormous variety of exogenous and endogenous compounds [108]. It is estimated that 80% of the oxidative metabolism of drugs can be attributed to CYP enzymes [109]. The isoenzymes of the CYP-superfamily are classified based of structure into families and subfamilies [110]. The liver contains the most diverse collection of CYP enzymes and pos- sesses the highest drug-metabolizing activity. the intestinal wall also contains an array of CYP enzymes and can substantially contribute to the first-pass metabolism [107].In vitro and some in vivo studies have shown that CYP-enzymes play a role in the metabolism of all smTKIs [11,12], except for afatinib that does not seem to be metabolized in the human body to relevant extent. Table 3 presents an overview of the isoenzymes that are primarily involved in the metabolism of smTKIs. In addition to CYP metabolism, glucu- ronidation by UDP-glucuronosyltransferases (UGTs) is an important clearance pathway Variability in bioavailability of small molecular tyrosine kinase inhibitors | 31

[111,112]. This phase II reaction is present in more than 80% of the metabolic pathways of drugs [113]. The enzymes are present in various tissues including intestines and liver. 1.1 Table 3 presents the smTKI substrates for UGT.

Drugs, the smTKIs themselves and other substances present in food and drinks can exhibit an influence on the number and functions of membrane transporters and metabolizing enzymes [114]. Concomitant use of an inducer or inhibitor of these proteins can significantly affect the absorbed dose of substrate smTKIs, ingestion of grapefruit juice is such an example [115–117]. An increasing list of inducers and inhibitors can be found in literature [114,118]. Another variable is genetic polymorphism. Inter- and intrapatient variability may be observed because of different expression patterns of transporters and enzymes. A major factor here is genetic predisposition of patients where ethnic background is important [119–121]. A growing body of evidence suggests that genetic polymorphism may play an important role in determining smTKI plasma concentration [11,12,122]. However, the clinical significance of pharmacogenetics in smTKI therapy still need to be proven in adequately set-up studies [123–125]. 32 | Chapter 1.1

Special populations

Elderly The process of aging involves changes in physiology and the functioning of organ systems. Important hepatic changes when considering drugs are a decrease in liver mass and CYP450 content. Kidneys decrease in mass with age and renal blood flow lessens. Gastric motility slows down and the secretion of gastric juices decreases [126]. Cancer is generally considered to be a disease of the elderly, with 60% of the diagnoses of cancer taking place in patients aged 75 and over [127]. Although age-induced F alterations seem possible, literature and registration texts do not report the necessity of adjusting the dose of smTKIs for the elderly [11–13].

The hepatic impaired The major site of drug metabolism and excretion for smTKIs is the liver. Liver function and hepatic impairment may have a clinically significant influence on the levels of the smTKIs. Liver cirrhosis, a key outcome of hepatic impairment, can be classified using various methods. The FDA wields the Child-Pugh classification (CPC) in their label texts, and so will this review. The CPC applies three classes (A, B and C) in increasing severity and clinical prognoses [128]. Serologic data, e.g. serum albumin and protrombine time are the basis for this classification. Table 4 presents the dose adjustments that need to be taken when hepatic impairment has been diagnosed for different CPC-scores. For most of the smTKIs at most of the different stages of the CPC-scale, no change dosage is recommended. Careful surveillance during therapy is required however, since clinical data upon which dose recommendation are stated are often from small scale studies. Hepatic impairment studies are generally carried out using non-cancer patients. Cancer and chemotherapy may induce additional changes in physiology of the liver, directly or indirectly [129–132]. Both lead to a reasoning where individual patient levels may need to be monitored in cases of (suspected) hepatic impairment.

Variability in bioavailability of small molecular tyrosine kinase inhibitors | 33

Table 4. Dose adjustment advice for the smTKIs in the different stages of the Child-Pugh liver impairment scale. 1.1 Drug A B C Afatinib ↔ ↔ ? Axitinib ↔ ↓ ? Bosutinib ↓ ↓ ↓ Cabozantinib ↔ X X Crizotinib ? ? ? Dabrafenib ? ? ? Dasatinib ↔ ↔ ↔ Erlotinib ↔ ↔ ↔ Gefitinib ↔ ↔ ↔ Imatinib ↔ ↔ ↓ Lapatinib ↔ ↔ ↓ Nilotinib ↓ ↓ ↓ Pazopanib ↔ ↓ X Ponatinib ? ? ? Regorafenib ↔ ↔ X Ruxolitinib ↓ ↓ X Sorafenib ↔ ↔ ? Sunitinib ↔ ↔ ? Vandetanib ↔ X X Vemurafenib ↔ ↔ ↔ ↔, no dose adjustment required; ↓, dose should be reduced; ?, effects not studied; X, do not use. Source: FDA label texts. 34 | Chapter 1.1

Selected interactions

Food The ingestion of food can cause the F of co-ingested drugs to change. The mechanisms by which this happens are not fully understood, but some clarification may be obtained from physiological changes caused by the presence of food [133,134]. The pH-values in the gastric environment may rise from 1-3.5 to 3.0-6.0 due to buffering and dilutant effects of food stuffs [135]. The proximal part of the duodenum does experience lower pH values when gastric emptying takes place [136]. Overall, fluctuations because of food in pH of the intestinal system are reported to be less dramatic and may cause the pH to lower by 1 [137]. Both are subject to inter- and intra-individual variations [138]. The presence of fat in food slows gastric emptying and such prolongs the residence time in the stomach [139,140]. Food triggers the gastro-colonic reflex, which speeds the motility in the terminal ileum and caecum [141]. The third important change is the composition of the gastrointestinal fluids. In the fed-state, the fluids exhibit higher buffer capacities, higher osmolality and higher concentrations of short-chain fatty acids and bile salts [142]. Any of these changes may affect the absorption of drugs from the GI-tract. Fluctuations in pH may change solubility and charging of compounds, in- or decreasing the dissolution rate [143]. Motility directly determines the maximum time available for absorption [144]. The fluid composition might lower the diffusion of compounds due to an increased osmolality or could increase solubility by dissolution of drug in fat or by bile salt emulsification [67].

There are no studies at present that have examined the mechanisms by which the absorption of smTKIs are influenced by food. The effects, however, are known for all the compounds. The smTKIs show a varying change in absorption when fasting and fed-state are compared. Clinical advice on administration timing can be given in situations where adequate evidence is available on food effects. The nature of the advice will depend on the effect of the food on the therapy and whether such an effect is beneficial. An advice to administer a smTKI without food should be interpreted as ingesting a tablet or capsule at least two hours after a meal or one hour before meal consumption [11,12]. Table 5 presents the influence food has on the absorption of smTKIs and lists the advices giving by the EMA and FDA on administration in fed or fasted state. Variability in bioavailability of small molecular tyrosine kinase inhibitors | 35

Table 5. Effect of food and acid regulation drugs on smTKI absorption and administration advices. Food Antacids 1.1 Drug Exposure (AUC) Advice Exposure Advice Reference(s) Afatinib -39% WO ↔ - Axitinib +19% WO/F ↔ - [194] Bosutinib +70% F ↓ A [195] Cabozantinib +57% WO (↓)1 - Crizotinib -14% WO/F (↓) - Dabrafenib -31% WO (↓) -

Dasatinib +14% WO/F ↓ A [196] Erlotinib +109% WO ↓ A Gefitinib -14 - +37% WO/F (↓) A Imatinib ? F2 ↔ - [197,198] Lapatinib +80 - 161% 3 ↓ A [146] Nilotinib +29 - 82% WO ↓ A Pazopanib +100% WO ↓ A [199] Ponatinib -3 - +9% WO/F (↓) A [200] Regorafenib +36 - 48% F (↔) - Ruxolitinib ? WO/F ↔ - Sorafenib -29 - +14% WO (↓) A Sunitinib +12% WO/F ↔ - [201] Vandetanib ? WO/F (↓) - Vemurafenib +200% WO/F4 ? ?4 [147] ↑, increased exposure; ↓, lowered exposure; ↔, no effect; ?, unknown effect; (↑)/(↓)/(↔), possible effect; -, no avoidance needed; A, avoidance needed. WO, without food; F, with food. Source: EPARs and FDA lable texts, additional sources indicated where appropriate.1 No formal studies done. 2Food serves as gastrointestinal protection. 31 hour before or after meal (see text).4 Advice not (yet) given.

Bosutinib is preferably given in the presence of food, an increase in tolerability of bosutinib was observed compared to the fasting-state [145]. Regorafenib should be combined with the intake of a low-fat meal, which increases the exposure of the drug and its active metabolites compared to fasting-state and high-fat meal [12,13]. Lapatinib exhibits a four-time increase in exposure during fed-state in comparison with fasting-state. It is advisable to give lapatinib somewhere in a window of 1 hour before or after a meal, but preferably before [146]. To maintain constant blood levels, this should be standardized for the individual patient. It was chosen to advise the intake of imatinib with food to minimize adverse reactions in stomach as food shields the stomach and esophagus lining 36 | Chapter 1.1

[12]. The food effect on vemurafenib has only very recently been studied. The authori- ties did not issue an advice concerning food intake yet. A conclusive advice needs to be constructed after thorough interpretation of the study results [147].

Stomach pH regulating drugs Drugs that have an effect on the acid secretion and the pH-value of the stomach are abundantly prescribed and bought as over-the-counter (OTC) products [148,149]. Dys- pepsia and related gastrointestinal disorders have been reported as adverse reactions following therapy with smTKIs [11,12,150–152].Treatment options are the use of antacids, histamine H2 receptor antagonists and proton pump inhibitors (PPIs) [149].

The pH increasing drugs neutralize or reduce gastric acid by different mechanisms and are administered in various dosing schemes with PPIs as the most effective and most popular [11,12,153,154].

For most of the smTKIs it has been well established whether they suffer from a change in absorption caused by concomitant therapy with a pH-changing drug. pH-dependent solubility plays a crucial role in these interactions. Absorption of drugs may decrease when the pH is increased in the gastric environment [149,155,156]. When this is the case, the advice to avoid the effects on the pH-value may be needed.

Different avoidance protocols can be used depending on the type of pH-increasing drug. Antacids should not be administered at the same time as the smTKI. Instead, the smTKI should be ingested two hours before or four hours after the antacid. For both H2-antagonists and PPIs, one protocol can be applied; the smTKI must be administered along with the pH-increasing drug, or the acid-regulating drugs should be taken in the morning and the smTKI before the night [11,12]. The concomitant administration avoids pH-effects since the pH-regulator has to be absorbed in order to be active and as such, a window of absorbance is provided for the smTKI.

Cabozantinib, crizotinib, dabrafenib, gefitinib, ponatinib, sorafenib, and vandetanib all exhibit pH-dependent solubility. Interaction studies with the pH-regulating drugs have not been performed with regard to these smTKIs, however. On grounds of their physicochemical properties, such interactions may be probable. This has led to the avoidance advice for ponatinib and sorafenib [11,12]. The protocol is not recommended for the other compounds at this point in time. Nevertheless, caution during concomitant use Variability in bioavailability of small molecular tyrosine kinase inhibitors | 37

is required since interactions are not ruled out. Table 5 lists the effect of concomitant administration of pH regulators and smTKIs and the avoidance advices. 1.1

No statements covering the possible interaction with vemurafenib due to pH-changes are found in literature. The drug is found to be poorly soluble at pH 6.8 in its crystal form. The marketed formulation is comprised of a solid solution that may react differently to changes in pH. The same is true for regorafenib. No substantiated advice can be given for these two drugs, caution, again, is still warranted. 38 | Chapter 1.1

Discussion

The advances in drug discovery have led to a particularly useful niche of chemotherapeutics, the orally applicable smTKIs as key representatives. Although the development of personalized oral chemotherapy is very promising, the nature of the selection process leads to drugs that are hindered by a low and variable F. This low and variable F does not appear to be a hindrance in the wielding of the drugs, but might prove to be troublesome in further expanding the use of these compounds. High or specifically low dosing and frequent multiple dosing regimens may find hurdles here.

Data on the physicochemical properties of the smTKIs is sparsely scattered and largely covered by patents. It is, however, clear that most of the drugs suffer from solubility problems. Marketed formulations only possess the most basic physicochemical adjustments to increase the absorption from tablets and capsules. As has been done with other chemotherapeutics, it may be rewarding to seek novel formulation strategies, such as solid dispersion or self-emulsifying systems [157], to increase solubility. Regorafenib and vemurafenib formulations wield such strategies [11].

In vivo data concerning interactions with intestinal protein transporters and metabolizing enzymes are coming together slowly, but large clinical studies are still to be performed. Interactions with inducers or inhibitors of transporters and enzymes should always be considered during chemotherapy, even though current knowledge on the matter is limited. Once certain enzyme or transporter involvement is deemed clinically relevant, therapy routines or formulations can be combined with an additional inhibitor to decrease enzymatic breakdown or extensive efflux. The causes for the variable and low F for the smTKIs are slowly being uncovered and more studies are coming along to investigate the drugs further. Current knowledge implies that solubility and metabolism, especially concerning CYP450, play a pivotal role.

Interactions with food are profound. The mechanisms underlying the effect of food on F are not understood in detail. It may be beneficial to study these effects more thoroughly to assess the consequence of different amounts of fats in food. This in turn may prove useful in formulation design, especially when lipid-based formulations are considered [158]. Variability in bioavailability of small molecular tyrosine kinase inhibitors | 39

Whatever the value of the F, it is important to keep in mind that it does not give any information about the exposure of the therapeutic target to the drug. A drug that has 1.1 reached the systemic circulation, even in its entirety, still needs to travel to its molecular target in order to be therapeutically active. In the case of the smTKIs, the targets are the kinases that are located inside the tumor cells and often in endothelial cells in the micro-environment. Various obstacles may be encountered that may sharply reduce or even diminish the amount of drug that reaches its target. An example is the blood brain barrier, which has been shown to be a partial blockade to EGFR-smTKIs [159]. This therapeutic availability (FTher) can be described by adding a fraction to equation 1, the fraction of the bioavailable drug that reaches the therapeutic target (FT), to yield equation 3,

(3)

F"#$% = F' ∗ F) ∗ F* ∗ F" Furthermore, patient compliance to therapy is essential in the success of oral cancer treatment. It is estimated that compliance of cancer patients taking oral chemotherapeutics ranges between 20 and 100 % [160]. Adverse events are the main cause for an impaired compliance, especially gastro-intestinal side effects are associated with smT- KIs [161]. Alternating compliance, although not directly influencing the bioavailability, adds another variable to the mix in determining mean plasma concentrations. For this reason, care should also be placed and directed towards the therapeutic infrastructure. Guidance of patients during the experience of adverse events may prove to be helpful in keeping compliance at a high level [162].

Conclusion

The smTKI-group generally suffers from a low and variable F, a problem that is receiving little attention in literature. This review presents the various causes of the low and variable F of the smTKIs and provides possible means to enhance F and to reduce variability. Special attention is directed towards food and pH-dependent interactions, which have consequences for the therapeutic regimen of most of the smTKIs. Physicochemical, physiological and pharmacological data is far from complete. More and larger scale studies on these matters are needed to clarify the individual contribution to the impaired F of each of the factors. This will allow for adequately directed formulative or therapeutic measures for the enhancement of F. 40 | Chapter 1.1

References

1. Kohonen P, Ceder R, Smit I, Hongisto V, Myatt rosine kinase inhibitors - a review on pharma- G, Hardy B, et al. Cancer biology, toxicology cology, metabolism and side effects. Curr Drug and alternative methods development go Metab . 2009 [cited 2014 Sep 1];10(5):470–81. hand-in-hand. Basic Clin Pharmacol Toxicol. 15. Harrison MR. Pharmacotherapy options in 2014;115(1):50–8. advanced renal cell carcinoma: what role 2. Pinne M, Raucy JL. Advantages of cell-based for pazopanib? Clin Med Insights Oncol . high-volume screening assays to assess nucle- 2011;5:349–64. ar receptor activation during drug discovery. 16. Rowland M, Tozer TN. Clinical Pharmacokinet- Expert Opin Drug Discov. 2014;9(6):669–86. ics; Concepts and Applications. 3rd ed. Phila- 3. Maroun CR, Rowlands T. The Met receptor delphia: Lippincott, Williams & Wilkins; 1994. tyrosine kinase: a key player in oncogen- 17. Fan J, de Lannoy IAM. Pharmacokinetics. Bio- esis and drug resistance. Pharmacol Ther. chem Pharmacol . 2014;87(1):93–120. 2014;142(3):316–38. 18. Hellriegel ET, Bjornsson TD, Hauck WW. Inter- 4. Charette N, Vandeputte C, Stärkel P. Ras in patient variability in bioavailability is related digestive oncology: from molecular biolo- to the extent of absorption: implications for gy to clinical implications. Curr Opin Oncol. bioavailability and bioequivalence studies. Clin 2014;26(4):454–61. Pharmacol Ther . 1996;60(6):601–7. 5. Chieffi P. An overview on new anticancer mo- 19. US Food and Drug administration (FDA). Clin- lecular targets in human testicular germ cell ical pharmacology and biopharmaceutics re- tumors. Rend Lincei . 2014;25(2):221–8. views. http//fda.gov, accessed 20 august 2014. 6. Minuti G, D’Incecco A, Landi L, Cappuzzo F. 20. Freiwald M, Schmid U, Fleury A, Wind S, Stop- Protein kinase inhibitors to treat non-small- fer P, Staab A. Population pharmacokinetics of cell lung cancer. Expert Opin Pharmacother. afatinib, an irreversible ErbB family blocker, 2014;15(9):1203–13. in patients with various solid tumors. Cancer 7. Tyner JW. Functional genomics for per- Chemother Pharmacol . 2014 [cited 2014 Dec sonalized cancer therapy. Sci Transl Med . 29];73(4):759–70. 2014;6(243):243fs26. 21. Garrett M, Poland B, Brennan M, Hee B, Pitha- 8. Liu G, Franssen E, Fitch MI, Warner E. Pa- vala YK, Amantea MA. Population pharmacoki- tient preferences for oral versus intrave- netic analysis of axitinib in healthy volunteers. nous palliative chemotherapy. J Clin Oncol. Br J Clin Pharmacol . 2014;77(3):480–92. 1997;15(1):110–5. 22. Hsyu P-H, Mould DR, Upton RN, Amantea M. 9. Benjamin L, Cotté F-E, Philippe C, Mercier F, Pharmacokinetic-pharmacodynamic rela- Bachelot T, Vidal-Trécan G. Physicians’ prefer- tionship of bosutinib in patients with chron- ences for prescribing oral and intravenous an- ic phase chronic myeloid leukemia. Cancer ticancer drugs: a Discrete Choice Experiment. Chemother Pharmacol . 2013 [cited 2014 Dec Eur J Cancer . 2012;48(6):912–20. 29];71(1):209–18. 10. Jibodh RA, Lagas JS, Nuijen B, Beijnen JH, Schel- 23. Hsyu P-H, Mould D, Abbas R, Pearce S, Aman- lens JHM. Taxanes: old drugs, new oral formu- tea M. Abstract A195: A population pharma- lations. Eur J Pharmacol. 2013;717(1–3):40–6. cokinetic (PPK) model of bosutinib (BOS). Mol 11. European Medicines Agency (EMA). European Cancer Ther . 2011 [cited 2014 Dec 29];10(Sup- Public Assessments Reports (EPARs). http// plement 1):A195–A195. www.ema.europa.eu, accessed 20 august 24. Ouellet D, Gibiansky E, Leonowens C, O’Hagan 2014. A, Haney P, Switzky J, et al. Population phar- 12. US Food and Drug administration (FDA). Prod- macokinetics of dabrafenib, a BRAF inhibitor: uct Label Descriptions. http//fda.gov, accessed effect of dose, time, covariates, and relation- 20 august 2014. ship with its metabolites. J Clin Pharmacol . 13. Administration Australian therapeutic goods. 2014;54(6):696–706. Australian Public Assessment Reports for pre- 25. Fukudo M, Ikemi Y, Togashi Y, Masago K, Kim scription medicines (AUSPARs). http//tga.gov. YH, Mio T, et al. Population pharmacokinetics/ au, accessed 20 august 2014. pharmacodynamics of erlotinib and pharma- 14. Hartmann JT, Haap M, Kopp H-G, Lipp H-P. Ty- cogenomic analysis of plasma and cerebro- Variability in bioavailability of small molecular tyrosine kinase inhibitors | 41

spinal fluid drug concentrations in Japanese type on the population pharmacokinetics of patients with non-small cell lung cancer. Clin sunitinib in patients with renal cell carcinoma. Pharmacokinet . 2013;52(7):593–609. Ther Drug Monit . 2014;36(3):310–6. 1.1 26. White-Koning M, Civade E, Geoerger B, 36. Houk BE, Bello CL, Kang D, Amantea M. A Thomas F, Le Deley M-C, Hennebelle I, et al. population pharmacokinetic meta-analysis Population analysis of erlotinib in adults and of sunitinib malate (SU11248) and its prima- children reveals pharmacokinetic character- ry metabolite (SU12662) in healthy volun- istics as the main factor explaining tolerance teers and oncology patients. Clin Cancer Res . particularities in children. Clin Cancer Res. 2009;15(7):2497–506. 2011;17(14):4862–71. 37. Martin P, Oliver S, Kennedy S-J, Partridge E, 27. Thomas F, Rochaix P, White-Koning M, Henne- Hutchison M, Clarke D, et al. Pharmacokinetics belle I, Sarini J, Benlyazid A, et al. Population of vandetanib: three phase I studies in healthy pharmacokinetics of erlotinib and its pharma- subjects. Clin Ther . 2012;34(1):221–37. cokinetic/pharmacodynamic relationships in 38. Yu H, Steeghs N, Nijenhuis CM, Schellens JHM, head and neck squamous cell carcinoma. Eur Beijnen JH, Huitema ADR. Practical guidelines J Cancer . 2009;45(13):2316–23. for therapeutic drug monitoring of anticancer 28. Filppula AM, Neuvonen M, Laitila J, Neuvonen tyrosine kinase inhibitors: focus on the phar- PJ, Backman JT. Autoinhibition of CYP3A4 leads macokinetic targets. Clin Pharmacokinet . to important role of CYP2C8 in imatinib metab- 2014;53(4):305–25. olism: variability in CYP2C8 activity may alter 39. Murakami H, Tamura T, Takahashi T, Nokihara plasma concentrations and response. Drug H, Naito T, Nakamura Y, et al. Phase I study of Metab Dispos . 2013;41(1):50–9. continuous afatinib (BIBW 2992) in patients 29. Yoo C, Ryu M-H, Ryoo B-Y, Beck MY, Chang with advanced non-small cell lung cancer af- H-M, Lee J-L, et al. Changes in imatinib plas- ter prior chemotherapy/erlotinib/gefitinib ma trough level during long-term treatment of (LUX-Lung 4). Cancer Chemother Pharmacol . patients with advanced gastrointestinal stro- 2012;69(4):891–9. mal tumors: correlation between changes in 40. Marshall J, Shapiro GI, Uttenreuther-Fischer covariates and imatinib exposure. Invest New M, Ould-Kaci M, Stopfer P, Gordon MS. Phase I Drugs . 2012;30(4):1703–8. dose-escalation study of afatinib, an ErbB fam- 30. Tanaka C, Yin OQP, Smith T, Sethuraman V, ily blocker, plus docetaxel in patients with ad- Grouss K, Galitz L, et al. Effects of rifampin and vanced cancer. Future Oncol . 2013;9(2):271– ketoconazole on the pharmacokinetics of nilo- 81. tinib in healthy participants. J Clin Pharmacol . 41. Daud AI, Krishnamurthi SS, Saleh MN, Gitlitz 2011;51(1):75–83. BJ, Borad MJ, Gold PJ, et al. Phase I study of bo- 31. Pick AM, Nystrom KK. Pazopanib for the treat- sutinib, a src/abl tyrosine kinase inhibitor, ad- ment of metastatic renal cell carcinoma. Clin ministered to patients with advanced solid tu- Ther . 2012;34(3):511–20. mors. Clin Cancer Res . 2012;18(4):1092–100. 32. Hurwitz HI, Dowlati A, Saini S, Savage S, Suttle 42. Kurzrock R, Sherman SI, Ball DW, Forastiere AB, Gibson DM, et al. Phase I trial of pazopanib AA, Cohen RB, Mehra R, et al. Activity of XL184 in patients with advanced cancer. Clin Cancer (Cabozantinib), an oral tyrosine kinase inhibi- Res . 2009;15(12):4220–7. tor, in patients with medullary thyroid cancer. J 33. Jain L, Woo S, Gardner ER, Dahut WL, Kohn EC, Clin Oncol . 2011;29(19):2660–6. Kummar S, et al. Population pharmacokinetic 43. Mossé YP, Lim MS, Voss SD, Wilner K, Ruffner analysis of sorafenib in patients with solid tu- K, Laliberte J, et al. Safety and activity of crizo- mours. Br J Clin Pharmacol . 2011;72(2):294– tinib for paediatric patients with refractory sol- 305. id tumours or anaplastic large-cell lymphoma: 34. Strumberg D, Clark JW, Awada A, Moore MJ, a Children’s Oncology Group phase 1 consor- Richly H, Hendlisz A, et al. Safety, pharmaco- tium study. Lancet Oncol . 2013;14(6):472–80. kinetics, and preliminary antitumor activity of 44. Menzies AM, Long G V, Murali R. Dabrafenib sorafenib: a review of four phase I trials in pa- and its potential for the treatment of - met tients with advanced refractory solid tumors. astatic melanoma. Drug Des Devel Ther . Oncologist . 2007;12(4):426–37. 2012;6:391–405. 35. Mizuno T, Fukudo M, Fukuda T, Terada T, Dong 45. Falchook GS, Long G V, Kurzrock R, Kim KB, Arke- M, Kamba T, et al. The effect of ABCG2 geno- nau TH, Brown MP, et al. Dabrafenib in patients 42 | Chapter 1.1

with melanoma, untreated brain metastases, 55. Benet LZ. The role of BCS (biopharmaceu- and other solid tumours: a phase 1 dose-es- tics classification system) and BDDCS (bio- calation trial. Lancet . 2012;379(9829):1893– pharmaceutics drug disposition classification 901. system) in drug development. J Pharm Sci. 46. Aplenc R, Blaney SM, Strauss LC, Balis FM, Shus- 2013;102(1):34–42. terman S, Ingle AM, et al. Pediatric phase I trial 56. Amidon GL, Lennernäs H, Shah VP, Crison JR. and pharmacokinetic study of dasatinib: a re- A theoretical basis for a biopharmaceutic drug port from the children’s oncology group phase classification: the correlation of in vitro drug I consortium. J Clin Oncol . 2011;29(7):839–44. product dissolution and in vivo bioavailability. 47. Higgins B, Kolinsky K, Smith M, Beck G, Rashed Pharm Res. 1995;12(3):413–20. M, Adames V, et al. Antitumor activity of- er 57. Lennernäs H, Abrahamsson B. The use of bio- lotinib (OSI-774, Tarceva) alone or in combi- pharmaceutic classification of drugs in drug nation in human non-small cell lung cancer discovery and development: current status tumor xenograft models. Anticancer Drugs [In- and future extension. J Pharm Pharmacol. ternet]. 2004;15(5):503–12. 2005;57(3):273–85. 48. Togashi Y, Masago K, Fujita S, Hatachi Y, Fuku- 58. Bergström CAS, Andersson SBE, Fagerberg JH, hara A, Nagai H, et al. Differences in adverse Ragnarsson G, Lindahl A. Is the full potential events between 250 mg daily gefitinib and of the biopharmaceutics classification system 150 mg daily erlotinib in Japanese patients reached? Eur J Pharm Sci. 2014;57:224–31. with non-small cell lung cancer. Lung Cancer. 59. Australian Therapeutic Goods Administration. 2011;74(1):98–102. Australian Public Assessment Report (AUSPAR) 49. Reardon DA, Desjardins A, Vredenburgh JJ, for Ruxolitinib. http//tga.gov.au, accessed 20 Sathornsumetee S, Rich JN, Quinn JA, et al. august 2014. Safety and pharmacokinetics of dose-intensive 60. Broccatelli F, Cruciani G, Benet LZ, Oprea TI. imatinib mesylate plus temozolomide: phase BDDCS class prediction for new molecular enti- 1 trial in adults with malignant glioma. Neuro ties. Mol Pharm. 2012;9(3):570–80. Oncol. 2008;10(3):330–40. 61. Wu C-Y, Benet LZ. Predicting drug disposition 50. Lin NU, Freedman RA, Ramakrishna N, Younger via application of BCS: transport/absorption/ J, Storniolo AM, Bellon JR, et al. A phase I study elimination interplay and development of a of lapatinib with whole brain radiotherapy in biopharmaceutics drug disposition classifica- patients with Human Epidermal Growth Factor tion system. Pharm Res. 2005;22(1):11–23. Receptor 2 (HER2)-positive breast cancer brain 62. Macheras P, Karalis V, Valsami G. Keeping a metastases. Breast Cancer Res Treat. 2013 [cit- critical eye on the science and the regulation ed 2014 Dec 29];142(2):405–14. of oral drug absorption: a review. J Pharm Sci. 51. Bradeen HA, Eide CA, O’Hare T, Johnson KJ, 2013;102(9):3018–36. Willis SG, Lee FY, et al. Comparison of imati- 63. Benet LZ, Broccatelli F, Oprea TI. BDDCS applied nib mesylate, dasatinib (BMS-354825), and to over 900 drugs. AAPS J. 2011;13(4):519–47. nilotinib (AMN107) in an N-ethyl-N-nitro- 64. Benet LZ. BDDCS - Its impact and application. sourea (ENU)-based mutagenesis screen: 1st MENA Regulatory Conference on bioequiv- high efficacy of drug combinations. Blood. alence, biowaivers, bioanalysis and dissolu- 2006;108(7):2332–8. tion. 2013. 52. Mascarenhas J, Hoffman R. Ruxolitinib: the 65. Noyes AA, Whitney WR. THE RATE OF SOLU- first FDA approved therapy for the treatment TION OF SOLID SUBSTANCES IN THEIR OWN of myelofibrosis. Clin Cancer Res [Internet]. SOLUTIONS. J Am Chem Soc. 1897;19(12):930– 2012;18(11):3008–14. 4. 53. Lankheet NAG, Knapen LM, Schellens JHM, 66. Dressman JB, Amidon GL, Reppas C, Shah VP. Beijnen JH, Steeghs N, Huitema ADR. Plasma Dissolution testing as a prognostic tool for oral concentrations of tyrosine kinase inhibitors drug absorption: immediate release dosage imatinib, erlotinib, and sunitinib in routine clin- forms. Pharm Res. 1998;15(1):11–22. ical outpatient cancer care. Ther Drug Monit. 67. Jambhekar SS, Breen PJ. Drug dissolution: sig- 2014;36(3):326–34. nificance of physicochemical properties and 54. Patson B, B Cohen R, Olszanski AJ. Pharmaco- physiological conditions. Drug Discov Today kinetic evaluation of axitinib. Expert Opin Drug [Internet]. 2013;18(23–24):1173–84. Metab Toxicol. 2012;8(2):259–70. 68. Florence A, Attwood D. Physicochemical princi- Variability in bioavailability of small molecular tyrosine kinase inhibitors | 43

ples of pharmacy. 4rd ed. London: Pharmaceu- 81. Newman A, Knipp G, Zografi G. Assessing the tical Press; 2006. performance of amorphous solid dispersions. J 69. Sjögren E, Westergren J, Grant I, Hanisch G, Pharm Sci. 2012;101(4):1355–77. 1.1 Lindfors L, Lennernäs H, et al. In silico predic- 82. Janssens S, Van den Mooter G. Review: phys- tions of gastrointestinal drug absorption in ical chemistry of solid dispersions. J Pharm pharmaceutical product development: appli- Pharmacol. 2009;61(12):1571–86. cation of the mechanistic absorption model 83. Ouellet D, Grossmann KF, Limentani G, Nebot GI-Sim. Eur J Pharm Sci. 2013;49(4):679–98. N, Lan K, Knowles L, et al. Effects of particle 70. Reppas C, Meyer JH, Sirois PJ, Dressman JB. size, food, and capsule shell composition on Effect of hydroxypropylmethylcellulose on the oral bioavailability of dabrafenib, a BRAF gastrointestinal transit and luminal viscosity in inhibitor, in patients with BRAF mutation-posi- dogs. Gastroenterology [Internet]. 1991;100(5 tive tumors. J Pharm Sci. 2013;102(9):3100–9. Pt 1):1217–23. 84. Sheng JJ, Sirois PJ, Dressman JB, Amidon GL. 71. Augustijns P, Wuyts B, Hens B, Annaert P, But- Particle diffusional layer thickness in a USP ler J, Brouwers J. A review of drug solubility in dissolution apparatus II: a combined function human intestinal fluids: implications for the of particle size and paddle speed. J Pharm Sci. prediction of oral absorption. Eur J Pharm Sci 2008;97(11):4815–29. [Internet]. 2014;57:322–32. 85. Grohganz H, Priemel PA, Löbmann K, Niel- 72. Davis SS, Hardy JG, Fara JW. Transit of pharma- sen LH, Laitinen R, Mullertz A, et al. Refining ceutical dosage forms through the small intes- stability and dissolution rate of amorphous tine. Gut. 1986;27(8):886–92. drug formulations. Expert Opin Drug Deliv. 73. Lennernas H. Modeling Gastrointestinal Drug 2014;11(6):977–89. Absorption Requires More In Vivo Biopharma- 86. Lennernäs H. Human in vivo regional in- ceutical Data: Experience from In Vivo Dissolu- testinal permeability: importance for - phar tion and Permeability Studies in Humans. Curr maceutical drug development. Mol Pharm. Drug Metab. 2007;8(7):645–57. 2014;11(1):12–23. 74. Mitra A, Kesisoglou F. Impaired drug absorp- l parameters for oral delivery and in vitro testing. tion due to high stomach pH: a review of strat- Mol Pharm. 2010;7(5):1388–405. egies for mitigation of such effect to enable 88. Hu Y-J, Wang Y-D, Tan F-Q, Yang W-X. Regula- pharmaceutical product development. Mol tion of paracellular permeability: factors and Pharm. 2013;10(11):3970–9. mechanisms. Mol Biol Rep. 2013;40(11):6123– 75. Psachoulias D, Vertzoni M, Butler J, Busby D, 42. Symillides M, Dressman J, et al. An in vitro 89. Sugano K, Kansy M, Artursson P, Avdeef A, Ben- methodology for forecasting luminal concen- dels S, Di L, et al. Coexistence of passive and trations and precipitation of highly permeable carrier-mediated processes in drug transport. lipophilic weak bases in the fasted upper small Nat Rev Drug Discov [Internet]. 2010 [cited intestine. Pharm Res. 2012;29(12):3486–98. 2014 Aug 20];9(8):597–614. 76. US Food and Drug administration (FDA). Chem- 90. Estudante M, Morais JG, Soveral G, Benet LZ. istry reviews. http//fda.gov, accessed 20 au- Intestinal drug transporters: an overview. Adv gust 2014. Drug Deliv Rev [Internet]. 2013;65(10):1340– 77. Thomas G. Medicinal Chemistry: an introduc- 56. tion. 2nd ed. Chichester: John Wiley and sons; 91. Varma M V, Ambler CM, Ullah M, Rotter CJ, Sun 2007. H, Litchfield J, et al. Targeting intestinal trans- 78. Kumar L, Meena CL, Pawar YB, Wahlang B, porters for optimizing oral drug absorption. Tikoo K, Jain R, et al. Effect of counterions on Curr Drug Metab. 2010;11(9):730–42. physicochemical properties of prazosin salts. 92. Shugarts S, Benet LZ. The role of transporters AAPS PharmSciTech. 2013;14(1):141–50. in the pharmacokinetics of orally administered 79. Stahl P, Wermuth C. Handbook of pharmaceu- drugs. Pharm Res [Internet]. 2009;26(9):2039– tical salts: properties, selection and use. 1st 54. ed. Hoboken: Wiley; 2008. 93. Hediger MA, Romero MF, Peng J-B, Rolfs A, 80. Einfalt T, Planinsek O, Hrovat K. Meth- Takanaga H, Bruford EA. The ABCs of solute ods of amorphization and investigation of carriers: physiological, pathological and ther- theamorphous state. Acta Pharm [Internet]. apeutic implications of human membrane 2013;63:305–34. transport proteinsIntroduction. Pflugers Arch 44 | Chapter 1.1

[Internet]. 2004;447(5):465–8. diated metabolism in the human gut wall. J 94. Vasiliou V, Vasiliou K, Nebert DW. Human Pharm Pharmacol . 2009;61(5):541–58. ATP-binding cassette (ABC) transporter family. 108. Bock KW. Homeostatic control of xeno- and Hum Genomics. 2008;3(3):281. endobiotics in the drug-metabolizing enzyme 95. Zakeri-Milani P, Valizadeh H. Intestinal trans- system. Biochem Pharmacol . 2014 [cited 2014 porters: enhanced absorption through P-glyco- Sep 3];90(1):1–6. protein-related drug interactions. Expert Opin 109. Zhou S-F, Liu J-P, Chowbay B. Polymorphism Drug Metab Toxicol. 2014;10(6):859–71. of human cytochrome P450 enzymes and its 96. Stieger B, Hagenbuch B. Organic anion-trans- clinical impact. Drug Metab Rev . 2009 [cited porting polypeptides. Curr Top Membr. 2014 Sep 3];41(2):89–295. 110. S i m 2014;73:205–32. SC, Ingelman-Sundberg M. The Human Cyto- 97. Liu T, Li Q. Organic anion-transporting poly- chrome P450 (CYP) Allele Nomenclature web- peptides: a novel approach for cancer therapy. site: a peer-reviewed database of CYP variants J Drug Target. 2014 ;22(1):14–22. and their associated effects. Hum Genomics . 98. Noguchi K, Katayama K, Sugimoto Y. Human 2010 [cited 2014 Sep 3];4(4):278. ABC transporter ABCG2/BCRP expression 111. Meech R, Miners JO, Lewis BC, Mackenzie PI. in chemoresistance: basic and clinical per- The glycosidation of xenobiotics and endoge- spectives for molecular cancer therapeutics. nous compounds: versatility and redundancy Pharmgenomics Pers Med. 2014;7:53–64. in the UDP glycosyltransferase superfamily. 99. Liu Y-H, Di Y-M, Zhou Z-W, Mo S-L, Zhou S-F. Pharmacol Ther . 2012;134(2):200–18. Multidrug resistance-associated proteins and 112. Rowland A, Miners JO, Mackenzie PI. The implications in drug development. Clin Exp UDP-glucuronosyltransferases: their role in Pharmacol Physiol. 2010;37(1):115–20. drug metabolism and detoxification. Int J Bio- 100. Jani M, Ambrus C, Magnan R, Jakab KT, Beéry chem Cell Biol . 2013;45(6):1121–32. E, Zolnerciks JK, et al. Structure and function 113. Wu B, Kulkarni K, Basu S, Zhang S, Hu M. First- of BCRP, a broad specificity transporter of pass metabolism via UDP-glucuronosyltrans- xenobiotics and endobiotics. Arch Toxicol. ferase: a barrier to oral bioavailability of phe- 2014;88(6):1205–48. nolics. J Pharm Sci . 2011;100(9):3655–81. 101. Bruyère A, Declèves X, Bouzom F, Ball K, 114. Thomas-Schoemann A, Blanchet B, Bardin C, Marques C, Treton X, et al. Effect of variations in Noé G, Boudou-Rouquette P, Vidal M, et al. the amounts of P-glycoprotein (ABCB1), BCRP Drug interactions with solid tumour-targeted (ABCG2) and CYP3A4 along the human small therapies. Crit Rev Oncol Hematol . 2014 [cited intestine on PBPK models for predicting intesti- 2014 Sep 2];89(1):179–96. nal first pass. Mol Pharm. 2010;7(5):1596–607. 115. Kimura S-I, Kako S, Wada H, Sakamoto K, Ashi- 102. Natarajan K, Xie Y, Baer MR, Ross DD. Role of zawa M, Sato M, et al. Can grapefruit juice de- breast cancer resistance protein (BCRP/ABCG2) crease the cost of imatinib for the treatment in cancer drug resistance. Biochem Pharmacol. of chronic myelogenous leukemia? Leuk Res . 2012;83(8):1084–103. 2011 [cited 2014 Sep 2];35(1):e11-2. 103. Ingelman-Sundberg M. Pharmacogenetics of 116. van Erp NP, Baker SD, Zandvliet AS, Ploeger cytochrome P450 and its applications in drug BA, den Hollander M, Chen Z, et al. Marginal therapy: the past, present and future. Trends increase of sunitinib exposure by grapefruit Pharmacol Sci. 2004;25(4):193–200. juice. Cancer Chemother Pharmacol . 2011 104. Brändén G, Sjögren T, Schnecke V, Xue Y. Struc- [cited 2014 Sep 2];67(3):695–703. ture-based ligand design to overcome CYP in- 117. Deangelo DJ. Managing chronic myeloid leu- hibition in drug discovery projects. Drug Discov kemia patients intolerant to tyrosine kinase Today. 2014;19(7):905–11. inhibitor therapy. Blood Cancer J . 2012 [cited 105. de Groot MJ. Designing better drugs: predict- 2014 Sep 2];2:e95. ing cytochrome P450 metabolism. Drug Discov 118. Francis S, Delgoda R. A patent review on the Today. 2006;11(13–14):601–6. development of human cytochrome P450 106. Brunton L, Chabner B. Goodman and Gilman’s inhibitors. Expert Opin Ther Pat . 2014 [cited The Pharmacological Basis of Therapeutics. 2014 Sep 3];24(6):699–717. 12th ed. New York: McGraw-Hill Professional; 119. Ingelman-Sundberg M, Sim SC, Gomez A, Ro- 2011. 27-28 p. driguez-Antona C. Influence of cytochrome 107. Thelen K, Dressman JB. Cytochrome P450-me- P450 polymorphisms on drug therapies: phar- Variability in bioavailability of small molecular tyrosine kinase inhibitors | 45

macogenetic, pharmacoepigenetic and clinical 131. Syvak LA, Lial’kin SA, Klimanov MI, Hubareva aspects. Pharmacol Ther . 2007 [cited 2014 HO, Askol’s’kyĭ A V, Kasap N V, et al. [Hepato- Sep 3];116(3):496–526. toxicity of cytostatic drugs in treatment of on- 1.1 120. Zanger UM, Schwab M. Cytochrome P450 en- cology patients]. Lik Sprava . 2013 [cited 2014 zymes in drug metabolism: regulation of gene Sep 3];(2):14–24. expression, enzyme activities, and impact of 132. de Mestier L, Manceau G, Neuzillet C, Bachet genetic variation. Pharmacol Ther . 2013 [cited JB, Spano JP, Kianmanesh R, et al. Primary tu- 2014 Sep 3];138(1):103–41. mor resection in colorectal cancer with unre- 121. Martiny VY, Miteva MA. Advances in molecular sectable synchronous metastases: A review. modeling of human cytochrome P450 poly- World J Gastrointest Oncol . 2014 [cited 2014 morphism. J Mol Biol . 2013 [cited 2014 Sep Sep 3];6(6):156–69. 3];425(21):3978–92. 133. Jamei M, Turner D, Yang J, Neuhoff S, Polak S, 122. Nie Q, Yang X-N, An S-J, Zhang X-C, Yang J-J, Rostami-Hodjegan A, et al. Population-based Zhong W-Z, et al. CYP1A1*2A polymorphism mechanistic prediction of oral drug absorption. as a predictor of clinical outcome in advanced AAPS J. 2009;11(2):225–37. lung cancer patients treated with EGFR-TKI and 134. Varum FJO, Hatton GB, Basit AW. Food, its combined effects with EGFR intron 1 (CA)n physiology and drug delivery. Int J Pharm. polymorphism. Eur J Cancer . 2011 [cited 2014 2013;457(2):446–60. Sep 3];47(13):1962–70. 135. Sjögren E, Abrahamsson B, Augustijns P, Becker 123. Klümpen H-J, Samer CF, Mathijssen RHJ, Schel- D, Bolger MB, Brewster M, et al. In vivo meth- lens JHM, Gurney H. Moving towards dose ods for drug absorption - comparative phys- individualization of tyrosine kinase inhibi- iologies, model selection, correlations with tors. Cancer Treat Rev . 2011 [cited 2014 Sep in vitro methods (IVIVC), and applications for 3];37(4):251–60. formulation/API/excipient characterization 124. Pander J, Guchelaar HJ, Gelderblom H. Phar- including food effects. Eur J Pharm Sci . 2014 macogenetics of small-molecule tyrosine ki- [cited 2014 Sep 3];57:99–151. nase inhibitors: Optimizing the magic bullet. 136. Kalantzi L, Goumas K, Kalioras V, Abrahamsson Curr Opin Mol Ther . 2010 [cited 2014 Sep B, Dressman JB, Reppas C. Characterization of 3];12(6):654–61. the human upper gastrointestinal contents un- 125. Nies AT, Schwab M, Keppler D. Interplay of der conditions simulating bioavailability/bio- conjugating enzymes with OATP uptake trans- equivalence studies. Pharm Res . 2006 [cited porters and ABCC/MRP efflux pumps in the 2014 Sep 3];23(1):165–76. elimination of drugs. Expert Opin Drug Metab 137. Persson EM, Gustafsson A-S, Carlsson AS, Nils- Toxicol . 2008 [cited 2014 Sep 3];4(5):545–68. son RG, Knutson L, Forsell P, et al. The effects of 126. Wooten JM. Pharmacotherapy consid- food on the dissolution of poorly soluble drugs erations in elderly adults. South Med J. in human and in model small intestinal fluids. 2012;105(8):437–45. Pharm Res . 2005;22(12):2141–51. 127. Hurria A, Lichtman SM. Clinical pharmacology 138. Bratten J, Jones MP. Prolonged recording of du- of cancer therapies in older adults. Br J Cancer odenal acid exposure in patients with function- . 2008;98(3):517–22. al dyspepsia and controls using a radioteleme- 128. Christensen E. Prognostic models including try pH monitoring system. J Clin Gastroenterol the Child-Pugh, MELD and Mayo risk scores- . 2009;43(6):527–33. -where are we and where should we go? J 139. Karhunen LJ, Juvonen KR, Huotari A, Purhonen Hepatol . 2004;41(2):344–50. AK, Herzig KH. Effect of protein, fat, carbohy- 129. Azim HA, Agbor-Tarh D, Bradbury I, Dinh P, drate and fibre on gastrointestinal peptide Baselga J, Di Cosimo S, et al. Pattern of rash, release in humans. Regul Pept . 2008;149(1– diarrhea, and hepatic toxicities secondary 3):70–8. to lapatinib and their association with age 140. Beglinger C, Degen L. Fat in the intestine as a and response to neoadjuvant therapy: anal- regulator of appetite--role of CCK. Physiol Be- ysis from the NeoALTTO trial. J Clin Oncol . hav . 2004 [cited 2014 Sep 3];83(4):617–21. 2013;31(36):4504–11. Available from: http://www.ncbi.nlm.nih.gov/ 130. Bahirwani R, Reddy KR. Drug-induced liver in- pubmed/15621067 jury due to cancer chemotherapeutic agents. 141. Wilson CG. The transit of dosage forms through Semin Liver Dis . 2014;34(2):162–71. the colon. Int J Pharm . 2010 [cited 2014 Sep 46 | Chapter 1.1

3];395(1–2):17–25. opment. Mol Pharm . 2013 [cited 2014 Sep 142. Cvijić S, Parojčić J, Langguth P. Viscosity-medi- 3];10(11):4055–62. ated negative food effect on oral absorption of 152. Santoni M, Conti A, De Giorgi U, Iacovelli R, poorly-permeable drugs with an absorption Pantano F, Burattini L, et al. Risk of gastrointes- window in the proximal intestine: In vitro ex- tinal events with sorafenib, sunitinib and pazo- perimental simulation and computational veri- panib in patients with solid tumors: a system- fication. Eur J Pharm Sci . 2014 [cited 2014 Sep atic review and meta-analysis of clinical trials. 3];61:40–53. Int J Cancer. 2014;135(4):763–73. 143. Charman WN, Porter CJ, Mithani S, Dressman 153. Shin JM, Vagin O, Munson K, Kidd M, Modlin JB. Physiochemical and physiological mecha- IM, Sachs G. Molecular mechanisms in therapy nisms for the effects of food on drug absorp- of acid-related diseases. Cell Mol Life Sci . 2008 tion: the role of lipids and pH. J Pharm Sci. [cited 2014 Sep 3];65(2):264–81. 1997 [cited 2014 Sep 3];86(3):269–82. 154. Brett S. Science review: The use of proton 144. Dressman JB. Comparison of canine and hu- pump inhibitors for gastric acid suppression in man gastrointestinal physiology. Pharm Res . critical illness. Crit Care . 2005;9(1):45–50. 1986 [cited 2014 Sep 3];3(3):123–31. 155. Zhang L, Wu F, Lee SC, Zhao H. pH-Dependent 145. Abbas R, Hug BA, Leister C, Gaaloul M El, Drug-Drug Interactions for Weak Base Drugs: Chalon S, Sonnichsen D. A phase I ascend- Potential Implications for New Drug Develop- ing single-dose study of the safety, tolera- ment. Clin Pharmacol Ther . 2014;96(2):266– bility, and pharmacokinetics of bosutinib 77. (SKI-606) in healthy adult subjects. Cancer 156. Ogawa R, Echizen H. Clinically significant drug Chemother Pharmacol . 2012 [cited 2014 Sep interactions with antacids: an update. Drugs. 3];69(1):221–7. 2011 ;71(14):1839–64. 146. Devriese LA, Koch KM, Mergui-Roelvink M, 157. Gupta S, Kesarla R, Omri A. Formulation strat- Matthys GM, Ma WW, Robidoux A, et al.- Ef egies to improve the bioavailability of poorly fects of low-fat and high-fat meals on steady- absorbed drugs with special emphasis on state pharmacokinetics of lapatinib in patients self-emulsifying systems. ISRN Pharm . 2013 with advanced solid tumours. Invest New [cited 2014 Aug 16];2013:848043. Drugs . 2014;32(3):481–8. 158. Yao M, Xiao H, McClements DJ. Delivery of 147. Ribas A, Zhang W, Chang I, Shirai K, Ernstoff lipophilic bioactives: assembly, disassembly, MS, Daud A, et al. The effects of a high-fat meal and reassembly of lipid nanoparticles. Annu on single-dose vemurafenib pharmacokinetics. Rev Food Sci Technol . 2014 [cited 2014 Aug J Clin Pharmacol. 2014;54(4):368–74. 16];5:53–81. 148. Fass R. Alternative therapeutic approaches to 159. Jamal-Hanjani M, Spicer J. Epidermal growth chronic proton pump inhibitor treatment. Clin factor receptor tyrosine kinase inhibitors in Gastroenterol Hepatol . 2012 [cited 2014 Sep the treatment of epidermal growth factor 3];10(4):338-45-40. receptor-mutant non-small cell lung cancer 149. Budha NR, Frymoyer A, Smelick GS, Jin JY, Yago metastatic to the brain. Clin Cancer Res . 2012 MR, Dresser MJ, et al. Drug absorption interac- [cited 2014 Sep 3];18(4):938–44. tions between oral targeted anticancer agents 160. Mathes T, Pieper D, Antoine S-L, Eikermann and PPIs: is pH-dependent solubility the Achil- M. Adherence influencing factors in patients les heel of targeted therapy? Clin Pharmacol taking oral anticancer agents: a systematic re- Ther . 2012;92(2):203–13. view. Cancer Epidemiol . 2014 [cited 2014 Sep 150. Joensuu H, Roberts PJ, Sarlomo-Rikala M, An- 18];38(3):214–26. dersson LC, Tervahartiala P, Tuveson D, et al. 161. Loriot Y, Perlemuter G, Malka D, Penault-Llorca Effect of the tyrosine kinase inhibitor STI571 F, Penault-Lorca F, Boige V, et al. Drug insight: in a patient with a metastatic gastrointestinal gastrointestinal and hepatic adverse effects of stromal tumor. N Engl J Med . 2001 [cited 2014 molecular-targeted agents in cancer therapy. Aug 26];344(14):1052–6. Nat Clin Pract Oncol. 2008 ;5(5):268–78. 151. Smelick GS, Heffron TP, Chu L, Dean B, West 162. Schneider SM, Adams DB, Gosselin T. A tai- DA, Duvall SL, et al. Prevalence of acid-re- lored nurse coaching intervention for oral ducing agents (ARA) in cancer populations chemotherapy adherence. J Adv Pract Oncol. and ARA drug-drug interaction potential for 2014;5(3):163–72. molecular targeted agents in clinical devel- 163. Deng Y, Sychterz C, Suttle AB, Dar MM,- Ber Variability in bioavailability of small molecular tyrosine kinase inhibitors | 47

shas D, Negash K, et al. Bioavailability, me- 2010;116(6):1582–91. tabolism and disposition of oral pazopanib in 175. Fukudo M, Ikemi Y, Togashi Y, Masago K, Kim patients with advanced cancer. Xenobiotica. YH, Mio T, et al. Population pharmacokinetics/ 1.1 2013;43(5):443–53. pharmacodynamics of erlotinib and pharma- 164. Roy S, Quiñones R, Matzger AJ. Structural and cogenomic analysis of plasma and cerebro- Physicochemical Aspects of Dasatinib Hydrate spinal fluid drug concentrations in Japanese and Anhydrate phases. Cryst Growth Des . patients with non-small cell lung cancer. 2012 [cited 2014 Sep 2];12(4):2122–6. Clin Pharmacokinet . 2013 [cited 2014 Sep 165. Grillo D, Polla G, Vega D. Conformational 2];52(7):593–609. polymorphism on imatinib mesylate: grind- 176. Li J, Zhao M, He P, Hidalgo M, Baker SD. Differ- ing effects. J Pharm Sci . 2012 [cited 2014 Sep ential metabolism of gefitinib and erlotinib by 2];101(2):541–51. human cytochrome P450 enzymes. Clin Cancer 166. Norges Helsedirektoratet. Scientific Discussion Res . 2007;13(12):3731–7. Sutent. http//www.helsedirektoratet.no, ac- 177. Hamilton M, Wolf JL, Drolet DW, Fettner SH, cessed 20 august 2014. Rakhit AK, Witt K, et al. The effect of rifampi- 167. Stopfer P, Marzin K, Narjes H, Gansser D, cin, a prototypical CYP3A4 inducer, on erlotinib Shahidi M, Uttereuther-Fischer M, et al.- Af pharmacokinetics in healthy subjects. Cancer atinib pharmacokinetics and metabolism Chemother Pharmacol . 2014;73(3):613–21. after oral administration to healthy male 178. Chen Y, Wang M, Zhong W, Zhao J. Pharmacoki- volunteers. Cancer Chemother Pharmacol. netic and pharmacodynamic study of Gefitinib 2012;69(4):1051–61. in a mouse model of non-small-cell lung car- 168. Chen Y, Tortorici MA, Garrett M, Hee B, Klamer- cinoma with brain metastasis. Lung Cancer . us KJ, Pithavala YK. Clinical pharmacology of 2013 [cited 2014 Sep 2];82(2):313–8. axitinib. Clin Pharmacokinet . 2013;52(9):713– 179. Hu S, Franke RM, Filipski KK, Hu C, Orwick SJ, 25. de Bruijn EA, et al. Interaction of imatinib with 169. Wind S, Giessmann T, Jungnik A, Brand T, Mar- human organic ion carriers. Clin Cancer Res . zin K, Bertulis J, et al. Pharmacokinetic drug 2008 [cited 2014 Sep 2];14(10):3141–8. interactions of afatinib with rifampicin and 180. Dutreix C, Peng B, Mehring G, Hayes M, Cap- ritonavir. Clin Drug Investig . 2014 [cited 2014 deville R, Pokorny R, et al. Pharmacokinetic in- Sep 2];34(3):173–82. teraction between ketoconazole and imatinib 170. Abbas R, Hug BA, Leister C, Burns J, Sonnichsen mesylate (Glivec) in healthy subjects. Cancer D. Effect of ketoconazole on the pharmacoki- Chemother Pharmacol . 2004;54(4):290–4. netics of oral bosutinib in healthy subjects. J 181. Castellino S, O’Mara M, Koch K, Borts DJ, Bow- Clin Pharmacol . 2011;51(12):1721–7. ers GD, MacLauchlin C. Human metabolism 171. Karras S, Pontikides N, Krassas GE. Pharma- of lapatinib, a dual kinase inhibitor: implica- cokinetic evaluation of cabozantinib for the tions for hepatotoxicity. Drug Metab Dispos . treatment of thyroid cancer. Expert Opin Drug 2012;40(1):139–50. Metab Toxicol . 2013;9(4):507–15. 182. Chan ECY, New LS, Chua TB, Yap CW, Ho HK, 172. Zimmerman EI, Hu S, Roberts JL, Gibson AA, Nelson SD. Interaction of lapatinib with- cy Orwick SJ, Li L, et al. Contribution of OATP1B1 tochrome P450 3A5. Drug Metab Dispos . and OATP1B3 to the disposition of sorafenib 2012;40(7):1414–22. and sorafenib-glucuronide. Clin Cancer Res . 183. Novartis. Jakavi, Roxulitinib. Prod Descr Novar- 2013 [cited 2014 Sep 2];19(6):1458–66. tis. 2014; 173. Chuan Tang S, Nguyen LN, Sparidans RW, Wa- 184. Gnoth MJ, Sandmann S, Engel K, Radtke M. genaar E, Beijnen JH, Schinkel AH. Increased In vitro to in vivo comparison of the sub- oral availability and brain accumulation of the strate characteristics of sorafenib tosylate ALK inhibitor crizotinib by coadministration of toward P-glycoprotein. Drug Metab Dispos . the P-glycoprotein (ABCB1) and breast cancer 2010;38(8):1341–6. resistance protein (ABCG2) inhibitor elacridar. 185. Shibayama Y, Nakano K, Maeda H. Multidrug Int J Cancer . 2014;134(6):1484–94. resistance protein 2 implicates anticancer 174. Johnson FM, Agrawal S, Burris H, Rosen L, Dhil- drug-resistance to sorafenib. Biol Pharm Bull. lon N, Hong D, et al. Phase 1 pharmacokinetic 2011;34(3):433–5. and drug-interaction study of dasatinib in pa- 186. Lagas JS, van Waterschoot RAB, Sparidans RW, tients with advanced solid tumors. Cancer . Wagenaar E, Beijnen JH, Schinkel AH. Breast 48 | Chapter 1.1

cancer resistance protein and P-glycoprotein Biochem Pharmacol . 2013;85(3):325–34. limit sorafenib brain accumulation. Mol Cancer 194. Escudier B, Gore M. Axitinib for the manage- Ther . 2010;9(2):319–26. ment of metastatic renal cell carcinoma. Drugs 187. Kane RC, Farrell AT, Saber H, Tang S, Williams R D . 2011;11(2):113–26. G, Jee JM, et al. Sorafenib for the treatment of 195. Abbas R, Leister C, Sonnichsen D. A clinical advanced renal cell carcinoma. Clin Cancer Res study to examine the potential effect of lanso- . 2006;12(24):7271–8. prazole on the pharmacokinetics of bosutinib 188. Shukla S, Robey RW, Bates SE, Ambudkar S V. when administered concomitantly to healthy Sunitinib (Sutent, SU11248), a small-molecule subjects. Clin Drug Investig . 2013;33(8):589– receptor tyrosine kinase inhibitor, blocks func- 95. tion of the ATP-binding cassette (ABC) trans- 196. Eley T, Luo FR, Agrawal S, Sanil A, Manning J, porters P-glycoprotein (ABCB1) and ABCG2. Li T, et al. Phase I study of the effect of gastric Drug Metab Dispos . 2009;37(2):359–65. acid pH modulators on the bioavailability of 189. Arakawa-Todo M, Ueyama J, Nomura H, Abe F, oral dasatinib in healthy subjects. J Clin Phar- Tsukiyama I, Matsuura K, et al. Drug interaction macol . 2009;49(6):700–9. between sunitinib and cimetidine and contri- 197. Tawbi HA, Tran AL, Christner SM, Lin Y, John- bution of the efflux transporter ATP-binding son M, Mowrey E, et al. Calcium carbonate cassette C2 to biliary excretion of sunitinib in does not affect nilotinib pharmacokinetics in rats. Anticancer Res . 2013;33(8):3105–11. healthy volunteers. Cancer Chemother Phar- 190. Jovelet C, Deroussent A, Broutin S, Paci A, Fa- macol . 2013;72(5):1143–7. rinotti R, Bidart JM, et al. Influence of the mul- 198. Egorin MJ, Shah DD, Christner SM, Yerk MA, tidrug transporter P-glycoprotein on the intra- Komazec KA, Appleman LR, et al. Effect of cellular pharmacokinetics of vandetanib. Eur J a proton pump inhibitor on the pharmaco- Drug Metab Pharmacokinet . 2013;38(3):149– kinetics of imatinib. Br J Clin Pharmacol . 57. 2009;68(3):370–4. 191. Martin P, Oliver S, Robertson J, Kennedy S-J, 199. Tan AR, Gibbon DG, Stein MN, Lindquist D, Read J, Duvauchelle T. Pharmacokinetic drug Edenfield JW, Martin JC, et al. Effects- ofke interactions with vandetanib during coadmin- toconazole and esomeprazole on the phar- istration with rifampicin or itraconazole. Drugs macokinetics of pazopanib in patients with R D . 2011;11(1):37–51. solid tumors. Cancer Chemother Pharmacol . 192. Durmus S, Sparidans RW, Wagenaar E, Beijnen 2013;71(6):1635–43. JH, Schinkel AH. Oral availability and brain 200. Narasimhan NI, Dorer DJ, Niland K, Haluska F, penetration of the B-RAFV600E inhibitor ve- Sonnichsen D. Effects of food on the pharma- murafenib can be enhanced by the P-GLYCO- cokinetics of ponatinib in healthy subjects. J protein (ABCB1) and breast cancer resistance Clin Pharm Ther . 2013;38(6):440–4. protein (ABCG2) inhibitor elacridar. Mol Pharm 201. Bello CL, Sherman L, Zhou J, Verkh L, Smeraglia . 2012 [cited 2014 Sep 2];9(11):3236–45. J, Mount J, et al. Effect of food on the - phar 193. Wu C-P, Sim H-M, Huang Y-H, Liu Y-C, Hsiao S-H, macokinetics of sunitinib malate (SU11248), a Cheng H-W, et al. Overexpression of ATP-bind- multi-targeted receptor tyrosine kinase inhibi- ing cassette transporter ABCG2 as a potential tor: results from a phase I study in healthy sub- mechanism of acquired resistance to vemu- jects. Anticancer Drugs. 2006;17(3):353–8. rafenib in BRAF(V600E) mutant cancer cells.

Chapter 1.2

Inherent formulation issues of kinase inhibitors

M. Herbrink1, J.H.M. Schellens2,3, J.H. Beijnen1,3, B. Nuijen1

Abstract

The small molecular Kinase Inhibitor (smKI) drug class is very promising and rapidly expanding. All of these drugs are administered orally. The clear relationship between structure and function has led to drugs with a general low intrinsic solubility. The majority of the commercial pharmaceutical formulations of the smKIs are physical mixtures that are limited by the low drug solubility of a salt form. This class of drugs is therefore characterized by an impaired and variable bioavailability rendering them costly and their therapies suboptimal. New formulations are sparingly being reported in literature and patents. The presented data suggests that continued research into formulation design can help to develop more efficient and cost-effective smKI formulation. Moreover, it may also be of help in the future design of the formulations of new smKIs. Inherent formulation issues of kinase inhibitors | 53

Introduction

Small molecular Kinase Inhibitors (smKIs) form a promising and rapidly expanding class of drugs [1,2]. The drugs target specific parts of kinase receptor proteins that play an important part in the intracellular growth signaling pathways in tumor and immune cells 1.2 [3–5]. After the first drug approval by the United States Food and Drug Administration (FDA) of Imatinib in 2001 [6,7], the number of drugs approved by both the FDA and Eu- ropean Medicines Agency (EMA) are now nearing 30, listed in table 1. Many more smKIs are being investigated in clinical trials and are expected to be approved in the coming years. A range of small molecular inhibitors have proven to be useful in the therapy of certain types of cancer. Additionally, smKIs may be prescribed as alternatives when other therapeutic options have failed or are deemed inappropriate [8]. A few compounds are (also) applied in the therapy of immunomodulated diseases [9–12] and may even have a future in the therapy of diseases such as diabetes mellitus [13].

All of the smKIs are without exception administered orally. This has great advantages in terms of patient convenience and cost reduction [14–16]. It presents however serious difficulties for compounds with a low solubility and/or permeability. These drugs are hindered by a reduced and variable bioavailability. This may cause drug plasma concentrations to be ineffectively low or toxically high with all due consequences [17]. Un- derstanding and controlling the parameters of solubility and permeability can therefore have a profound influence on patient plasma drug levels.

The smKIs have been designed using high-throughput screening and combinatorial chemistry from which the intricate structures and inherent solubility issues originate [18]. These means are used in the drug discovery of other drug classes as well, e.g. drugs acting on the central nervous system that also experience a problematic solubility [19].

Drug solubility and the dissolution process are affected by a plethora of factors with clinical implications. A number of these factors are inherent to the smKI structure and function. It follows that, as drug dissolution is most often the primary determinant for the smKI bioavailability, there is an apparent link between the high specificity of the smKIs and their impaired absorption into the systemic circulation.

This article will first present the overall problematic biopharmaceutical properties of the smKI drug group. Secondly, it will discuss the characteristic structural elements that 54 | Chapter 1.2

are responsible for the solubility behavior of the smKIs. It will continue by reviewing the current commercial formulations along with alternative investigational formulations. This article aims to underline that although high specificity can, in many cases, place a challenging strain on drug solubility and bioavailability, different and innovative formulation techniques may present possible solutions to some of these issues. Literature offers several reviews that address (pre)formulation challenges for poorly soluble drugs in general [20–22]. This article focuses on smKIs in particular. To the authors’ knowledge, this review is the first to combine literature and patent research on the solubility and formulation of smKI compounds from a pharmaceutical perspective.

Bioavailability

In the process of reaching the therapeutic target, the first step after oral administration and the disintegration of the dosage form is always the dissolution of the drug substance [75]. The second step, absorption, only takes place with the dissolved portion of the drug. Thus, poor drug solubility can be one of the main causes for a low and variable uptake of a drug into the systemic circulation, i.e. a low and variable bioavailability. This is generally true for the smKIs, as listed in table 1. This reflects back in their BCS (the Bio- pharmaceutical Classification System)-classes of which most are II (solubility hindered bioavailability) or IV (solubility and permeability hindered bioavailability) [76,77].

Factors such as presystemic metabolism and mediated transport by transporter proteins may also play a part in reducing a drug’s bioavailability. The combination of these factors is reviewed elsewhere [78,79]. Inherent formulation issues of kinase inhibitors | 55

[56,57] 1.2 2 72 [50,51] 5 [54,55] 25 95 [46,47] Bioavailability(%) [25,26] 74 [11,12] [31,32] 58 [35,36] 2.9 [39,40] [42,43] 37 [60] 46 [63,64] [67,68] [71,72]

1 IV II/IV IV II II III IV II II I/III BSC II/IV I/III III II/IV Vargatef Zykadia Mekinist Tafinlar Stivarga Xeljanz Bosulif Inlyta Imbruvica Giotrif Tradename Cotellic Tagrisso Lenvima Alecensa Nintedanib Ceritinib Trametinib Dabrafenib Regorafenib Tofacitinib Bosutinib Axitinib Ibrutinib Afatinib Compound Cobimetinib Osimertinib Lenvatinib Alectinib [48,49] 2 [44,45] 2 14-39[52,53] >95 [9,10] <25 98 [23,24] 60 [27,28] 60 [29,30] [33,34] [37,38] [41] 30 [61,62] [65,66] [69,70] [73,74] 43 [58,59] Bioavailability(%) 1 II I II I II II II/IV II IV IV BCS IV II II II IV Votrient Jakavi Tyverb Gleevec Iressa Tarceva Nexavar Sprycel Sutent Tasigna Tradename Zelboraf Caprelsa Iclusig Cometriq Xalkori Approved smKIs (registered trademarks) by the FDA on March 20th 2016. Appointed BCS-classes are taken from registration documents. registration from taken are BCS-classes 20th 2016. Appointed on March the FDA by trademarks) smKIs (registered Approved Pazopanib Ruxolitinib Lapatinib Lapatinib Imatinib Gefitinib Erlotinib Sorafenib Dasatinib Sunitinib Nilotinib Table 1. Table Compound Vemurafenib Vandetanib Ponatinib Cabozantinib on mass balance; Estimated/Based 2. System; Classification 1. BCS, Biopharmaceutical Crizotinib Crizotinib 56 | Chapter 1.2

Physicochemistry

Essential structures In the past few decades the role of signal proteins in the homeostasis of tumors became more and more apparent [80,81]. The advancement of the diverse techniques and possibilities of molecular modelling have led to a therapeutic target-based drug discovery regime [82,83]. With it, structure-activity relations for inhibitory molecules for these proteins were assessed. Key in these relations is the binding of the lead drug molecule to the receptor and the inhibitory action thereon[84]. The latter can be viewed as a dependent of the first but does not necessarily result from the same molecular structure. The independence is illustrated by lenvatinib and sunitinib; they inhibit VEGFR2 by binding to the ATP-binding site with their core structure. Additional binding through a nearby structure in both compounds gives them their difference in residence time without influencing the inhibition itself [85].

The resulting collection of mainly lipophilic structures now forms the backbone of the majority of the KIs. Some molecular structures and scaffolds are found throughout the current marketed collection of smKIs. Even though the drugs inhibit a wide variety of proteins, the necessary structures to do so are similar. Table 2 lists the molecular structures of the free base smKIs ordered by primary target. Bold print indicates the proven binding moieties in the smKIs that are responsible for the inhibitory effect.

The bold printed structures in table 2 have been shown in molecular docking and in vitro crystallization studies to be critical in receptor binding and inhibition[22-53]. These include highly lipophilic moieties such as (substituted) phenyls, aromatic amines, biaryl constructs and heterocyclic aromatics. Using these structures as scaffolds, a great number of studies have designed new smKIs by adding different side groups [117–119]. Com- bining scaffolds to create a new multi-kinase-targeting smKI is also a commonly used strategy [120].

The lipophilic functional moieties and additional scaffold structures make up the larger part of the smKI molecular structure. To improve the overall aqueous solubility of a drug, drug development may include the evaluation of the possible addition of hydrophilic moieties to the core structure [121]. This addition of more hydrophilic structures may not be feasible due to the presence of hydrophobic residues that are essential for binding areas of the inhibitors however [122–125]. Adjustments to the KI chemical structures Inherent formulation issues of kinase inhibitors | 57

to improve the biopharmaceutical behavior can therefore only go so far. This has led to a continuous search for a delicate balance between pharmaceutical and pharmacological properties, most often resulting in the favor of the latter. As a consequence, the aqueous solubility of the free base smKI remains generally low. A recent study shows that this balance may not always lean towards poor solubility and that a combination of high 1.2 inhibitor activity and hydrophilicity can be achieved for some KI structures [126].

Additionally, enhancing aqueous solubility can be achieved by constructing a pro-drug from the drug substance. Here, a solubility-improving side group is attached to the active drug, which is later removed by some form of metabolism in vivo after absorption [127]. Although this has not been used in the marketed smKIs, it is currently being examined for smKIs [128].

Solubility The aqueous solubility of the smKIs is strongly influenced by environmental factors, as described in the following paragraphs. Additionally, the time points of solubility measurements determine whether an equilibrium or a dynamic value is assessed, which may differ significantly [51]. A single value can therefore not be appointed to the aqueous solubility. Reported values should be evaluated carefully, as the circumstances under which it was measured may not be adequately described. pH-dependent solubility The majority of the smKI scaffolds consists of nitrogen-based heterocyclic systems. An important consequence of the nitrogen-containing core inhibitory structures in combination with secondary amine moieties is an overall pH-dependent aqueous solubility.

The majority of the pKa values of the ionizable nitrogen structures are sufficiently low to qualify them as relatively weak bases [129,130]. Table 3 lists the currently known values. The higher pKa values commonly concern nitrogen groups that become neutral when protonated, whilst the lower pKa values correspond to groups that attain a positive charge by protonation. For most compounds this means a relative high solubility in acidic media and a harshly reduced solubility in more alkaline environments [131–133]. Vemurafenib, Regorafenib and Trametinib are exceptions herein with an almost pH-independent solubility. This is due to their pKa values that are either above or below the physiological pH-range. 58 | Chapter 1.2

Table 2. Effective moieties in molecular structures of smKI classes Primary target Molecular structure(s)

BCR-Abl**[86–89] O O NH NH N OH N NH N N N NH S N N N F Cl NH HN N F O N N N F N dasatinib N nilotinib N Cl Cl imatinib O O NH NH N F O F N N F N N O N N N N bosutinib ponatinib

EGFR*[90–92] O S NH O F O N O O N O Cl NH O N O N HN O F N O N O Cl NH N erlotinib lapatinib gefitinib N N O O NH O Cl N NH Br O HN O NH F N N F O NH N N N O N N N vandetanib afatinib osimertinib

VEGFR*, PDG- O N H H N FR*[85,93–102] 2 O N N S N HN O O F O F F NH F N N NH N N Cl NH NH NH O sorafenib O sunitinib pazopanib F O F F O N O Cl NH O HN S HN F O NH O N O NHN O N NH axitinib cabozantinib regorafenib F O HN Cl NH NH O H O O N O NH O N 2 O N N O NH lenvatinib N O nintedanib Inherent formulation issues of kinase inhibitors | 59

Table 2. Continued Primary target Molecular structure(s) B-Raf* [103,104]

F F

Cl O O H2N O F S 1.2 O N NHS NH N F O N N F H N S vemurafenib dabrafenib

Jak*[105–108] N HN N N N N N NH N N N N O ruxolitinib tofacitinib ALK*[109–113] Cl Cl H O O F S NH N N Cl O O N H NH NH N H2N N N O N NH N O N crizotinib ceritinib alectinib

MEK*[114–116] F H OH O N NH I NH N O O O N N F NH NH O F F trametinib cobimetinib I *BCR-Abl, Breakpoint Cluster Region- Abelson murine lukemiaviral oncogene; *EGFR, Endothelial Growth Factor Receptor; VEGFR, Vascular Endothelial Growth Factor Receptor; PDGFR, Platelet Derived Grwoth Factor Receptor; B-raf, v-raf murine sarcoma viral oncogene homolog B1; Jak, Janus Kinase; ALK, Anaplastic Lymphoma Kinase; MEK, Mitogen activated protein Kinase kinase. 60 | Chapter 1.2

Salt and free base polymorphs The presence of ionizable groups in most of the smKI drug structure made the conversion to salts feasible [134]. The commercial formulations of many smKIs contain a salt in a designated stable polymorph form (table 3). This led to a dramatic increase in solubility for some compounds, e.g. Imatinib, Tofacitinib and Afatinib, of which the free base forms are very poorly soluble (BCS II or IV). Their respective mesylate, citrate and dimaleate salts upgraded them to BCS-I or III. For most of the other drugs salt formation did not significantly increase solubility and they remain BCS II or IV. The reasons for using particular salt or free base polymorphs in formulations are, however often not provided in registrations texts or patents. New salt and free base polymorphs with higher solubilities are being patented and reported in literature before and after market authorization is obtained. It seems that, at least in terms of solubility, better options were and are available. Comparisons in dissolution behavior and bioavailability between various salt forms are not yet accessible. The alternative salts may prove useful in future formulation improvements however.

Solvates Solvates are crystalline solids that contain stoichiometric or nonstoichiometric propor- tions of a solvent within their crystal structure. When this solvent is water, the solvate is termed a hydrate. When no solvent or water is present, a compound is termed an anhydrate. Solvates, hydrates and solventless crystals can differ significantly in solubility [135]. Most of the smKIs are formulated as anhydrates. Data on dissolution performance of solvates and hydrates are only rarely published. Dasatinib, Nilotinib, Lapatinib and Bosutinib are present as monohydrate crystals in their respective formulations. After the approval of Sprycel® in 2006, a way to produce Dasatinib anhydrate was found and this form was shown to be 2.4 times more soluble than the monohydrate variant [136]. Trametinib is formulated as a stoichiometric DMSO-solvate. In rats this increased the bioavailability of Trametinib 30-fold compared to the unsolvated form. In aqueous environment, the dissolved Trametinib.DMSO slowly precipitates as the much less soluble unsolvated form [51]. Inherent formulation issues of kinase inhibitors | 61

Formulations

Partly as a consequence of the before-mentioned factors, the solubility of a drug salt polymorph can be described thermodynamically. The following combined equation illustrates this [137]: 1.2

(1) * ∆) 0230 67 + −𝑙𝑙𝑙𝑙𝑙𝑙 𝑋𝑋 = +.-.-/0 1 020 4 + +.-.-/0 (𝛿𝛿: − 𝛿𝛿+)

Where X is the dissolved molar fraction, ΔHf is the latent heat of fusion (heat absorbed during melting), R is the gas constant, T is the temperature, T0 is the melting point of solute, V is the molar volume of liquid solute, Φ is the volume fraction of the solvent, δ is the solubility parameter (expression of cohesion between molecules).

Equation 1 clearly expresses the role of solid state crystallinity in a compound’s solubility through the heat of fusion and the melting point of the drugs. The heat of fusion is the heat necessary to transform a compound from its solid to its liquid state at its melting point. Both the melting point and the heat of fusion are measures of the bond strength in the drug crystal. Stronger bonds between molecules in a crystal structure will increase both parameters and subsequently lower solubility [138]. The solubility parameters in equation 1 denote the significance of the molecular structure and the resulting intermolecular cohesive forces in drug and solvent in the solubility end term. The parameters are descriptors of the interaction between molecules of the drug and the solvent. If such interactions are alike in the separate drug and solvent, a drug is more likely to be soluble in that particular solvent [139]. The last term in equation 1 is low in such a case and the resulting solubility is relatively high.

In a physiological environment, a pharmaceutical solution may be regarded as dilute with a near-to-constant temperature [140,141]. In that situation, the molar volume approaches unity and the volume fraction may be disregarded [137]. This allows for the following simplification to a qualitative equation:

(2) ,-., 4 −𝑙𝑙𝑙𝑙𝑙𝑙 𝑋𝑋 ≈ ∆𝐻𝐻* + ,-, / + (𝛿𝛿3 − 𝛿𝛿4) Equation 2 shows that a low heat of fusion and melting point are beneficial to a compound’s solubility [142,143]. Additionally, the solubility of a drug can be further increased by creating more similarity between the solubility parameters of the drug and the solvent [144,145]. 62 | Chapter 1.2

Commercial physical mixture formulations The commercial immediate release formulations of the smKIs are almost all designed as physical mixtures. They contain a crystal solid form. Since crystallinity is a strong determinant of solubility, the heat of fusion and the melting temperature of crystal polymorphs are closely correlated with it. Although not known for all compounds, the melting points are relatively high [26,36,62,146,147]. Furthermore, the difference between the solubility parameters of the drugs and the physiological solvent is expected to be, based on structure, relatively large [148–150]. The outcome is a very poor solubility, as is presented in table 3. The excipients present in the physical mixtures, such as polymers and surfactants, may have some influence on the composition of the physiological solvent in the direct environment of the dissolving drug. This may lower the difference in the last term of formula 2. In time this effect dissipates due to diffusion of both compounds and the drug will recrystallize [151]. Table 3 lists the excipients present in the commercial formulations. As far as could be assessed, most manufacturing processes entail dry or wet mixing of the formulation ingredients and optimization of the drug particle size [152]. The latter only increases dissolution speed as it does not alter there sultant solubility through any terms in equation 2. The accessibility of solubility data and dissolution curves is limited in literature and patents, so the effect of the formulation on the smKIs dissolution performance is largely unknown in the public domains. The proposed dissolution methods by the FDA for bioequivalence testing of smKI formulations offer clarification to some extent however [153]. The dissolution media are allmore voluminous than the stomach. In many cases the pH is adjusted to highly acidic conditions. And for some drugs surfactants are added. Although these methods are designed for quality control purposes, they provide further evidence of poor drug solubility. The non-bioequivalent dissolution media are designed to enable appropriate drug solubility (>85%) for the comparison of individual drug product batches [153]. When acidification and addition of surfactants to the media are needed to achieve this, drug solubility in bioequivalent media is likely to be significantly low. The formulation of lenvatinib is an exceptional case because it was designed with special focus on its stability. Lenvatinib has the tendency to gelate and decompose under humid and heated storage conditions, unlike the other smKIs. Calcium carbonate and mannitol were chosen as filler/disinte- grant and diluent to keep the overall hygroscopicity of the formulation low. This combination of the carrier and the hypromellose capsule make water uptake by the formulation unlikely and thus prevents gelation and hydrolytic degradation, both of which are detrimental to the bioavailability [71,154]. This capsule, which is additionally kept in a protective blister, can be stored at room temperature and humidity [71]. Inherent formulation issues of kinase inhibitors | 63

Alternative formulations Changing the solid state of the drug may radically alter the first term in equation 2. When a move is made from the highly structured crystal form to the less rigid amorphous form the solubility can increase. This is due to the fact that amorphous materials have a melting range instead of a melting point and lack a definite heat of fusion [173]. Amorphous 1.2 forms of drugs may be prepared in various ways by incorporating them in a polymer matrix in order to retain their amorphous state [174,175]. Such a system is termed a solid dispersion [176]. The commercial formulations of Vemurafenib and Regorafenib (4.2.1. and 4.2.2.) are designed like this. Additional advantages of a solid dispersion are particle size reduction, increased wettability, reduced aggregation and agglomeration and a decrease in the difference in solubility parameters through the polymer changing the physiological solvent. All these factors combined may increase the solubility long enough to improve drug absorption. During clinical studies of three smKIs it became apparent that a simple physical mixture did not suffice in creating an adequate plasma level. These cases are discussed below.

Vemurafenib Crystalline free base vemurafenib is known to exist in several polymorphs and solvates [61,62]. The most thermodynamically stable form II is practically insoluble in water with pH-values ranging from 1.1 to 9. The first clinical study with Vemurafenib was performed with the more soluble crystalline form I in a micronized capsule formulation. Form I transformed to form II over time and the observed bioavailability was low [61,62]. Shah et al. describes the development of the amorphous Vemurafenib formulation that is now used in the marketed Zelboraf® [177]. The amorphous solid dispersion was prepared by a solvent-controlled coprecipitation process as a so-called microprecipitated bulk powder. The amorphous Vemurafenib is herein stabilized by a hypromellose acetate succi- nate matrix to prevent crystallization[178]. Compared to the crystalline formulation, the solid dispersion demonstrated a significantly improved solubility and a five-fold increase in exposure.

Regorafenib Regorafenib as a monohydrate salt is poorly soluble in water at less than 0.1 mg/mL [179]. The possible consequences of the poor solubility for the bioavailability were recognized early on in the formulation development. Therefore, a series of physical mixtures and solid dispersions were tested. A solid dispersion of Regorafenib in PVP 25 with a composition of 1:4 was chosen afterin vitro dissolution screening and the assessment of rat pharmacokinetics[180]. 64 | Chapter 1.2

Table 3. Reported solubilities of smKIs as pure drug substance at given pH values and their types of commercial formulation.

Compound pKa pH Solubility (mg/mL) Dosage Formulation composition form Imatinib Mesylate X £ 5.5 > 1.6m [24] T/C Patent expired Gefitinib FB1 5.4; 7.2 [28] 1.0; 7.0 21m; <0.001 [28] T L, MC, CS, P, SLS, MS Erlotinib Hydrochloride 5.4 [155] 2.0 0.4m [155] T L, MC, SSG, SLS, MS Sorafenib Tosylate X 1.0; 4.5 0.034m; 0.013 [33] T MC, CS, H, SLS, MS

m Dasatinib FB.H2O 3.1; 6.8; 10.8 [156] 2.6; 6.0 18.4 ; 0.008 [37] T L, MC, CS, HPC, MS Sunitinib Malate 9.0 [41] 1.2-6.8 >25m [41] C MN, P, CS, MS

m Nilotinib Hydrochloride.H2O 2.1; 5.4 [157] 1.0; 4.5 0.28 ; <0.1 [45] C L, CP, PX, SC, MS -6 m Lapatinib Ditosilate.H2O 5; 7.2 [158] 1.0; w 10 ; 0.007 [49,159] T MC, P, SSG, MS Pazopanib Hydrochloride 2.1; 6.4; 10.2 [160] 1.1 0.65m [160] T MC, P, SSG, MS Ruxolitinib Phosphate 4.3; 11.8 [161] w X (highly soluble) [161] T L, MC, SSG, MS, SC, HPC, P Crizotinib FB 5.6; 9.4 [59] 1.6; 8.2 >10m; <0.1 [162] C MC, SC, CHP, SSG, MS Vandetanib FB 5.2; 9.4 [147] 6.8; w 0.35m; 0.008 [65,147] T MC, DCP, CP, P, MS Ponatinib Hydrochloride 2.8; 7.8 [70] 1.7; 7.5 7.8m; 0.16*10-3 [70] T L, MC, SSG, SC, MS Cabozantinib Malate X 2; >3 0.11m; x (very low) [163] C CS, SSG, SC, SA, MC Tofacitinib Citrate 5.1 [164] 1.0; w >28m; 2.9 [164] T MC, L, CS, MS

Bosutinib FB.H2O 7.9 [32] <5.0; >5.0 X (high); X (reduced) [32] T MC, CS, PX, P, MS Axitinib FB 4.8 [36] 1.1; >6.0 1.841m; 0.2*10-3 [36] T MC, L, CS, MS Ibrutinib FB 3.8 [40] 1.2; 5.5 2m; 0.003 [40] C MC, CS, MS, SLS Afatinib Dimaleate 5.0; 8.2 [165] <6.0; >7.0 >50m; 0.04 [165,166] T L, MC, CP, SC, MS Dabrafenib Mesylate 1.5; 2.2; 6.6 [47] 1.0; 4-8 X (VSS); X (PI) [47] C MC, MS, SC Trametinib FB.DMSO 0.3 [167] 1.2; 6.8 0.0004; 0.011m [51] T MC, MN, H, CS, MS, SLS, SC Ceritinib FB 4.1; 9.7 [168] 1.0; 6.8 11m; 0.2*10-3 [57] C MC, HPC, SSG, MS, SC Alectinib Hydrochloride 7.1 [169] 1.0; 6.8 0.0013; 0.0279m [60] C L, HPC, SLS, MS, CMC Cobimetinib Fumarate 8.9 [170] 1.0; 6.8 48.21m; 0.78 [64] T L, MC, CS, MS Osimertinib Mesylate 4.4; 9.5 [171] 1.2; 4.5 >3; >11m [68] T MN, H, SSF Lenvatinib Mesylate 5.1 [72] <3.0; 3-7 X (VSS); < 0.096 [71] C MN, MC, CC, H, TC Vemurafenib MPB2 7.9; 11.1 [62] 1.0; 6.8 <0.3*10-3; 0.5*10-3m [62] T Solid solution Regorafenib FB X X X T Solid dispersion Nintedanib Esilate 5.6; 9.4 [172] 1.0; ³ 6.8 5m; 0.011 [55] C Lipophilic suspension 1. FB = Free Base; 2. MPB = MicroPrecipitated Bulk; M = reported maximum solubility; w = in water, pH not specified; x = value not reported; () = description given; VSS, Very slightly soluble; PI, practically insoluble; T, Tablet; C, Capsule; L, lactose; MC; Microcrystalline cellulose; CS, Crosscarmellose sodium; P, Povidone; SLS, Sodium Lauryl Sulphate; MS, Magnesium stearate; SSG. Sodium starch glycolate; H, Hypromellose; HPC, hydroxypropylcellulose; MN, Mannitol; CP, Crospovidone; PX, Poloxamer 188; SC, Silica colloidalis anhydrica; CHP, Calcium hydrogen phosphate; DCP, Dibasic calcium phosphate; SA, Stearic acid; CMC, Carboxymethylcellulose calcium; SSF, Sodium stearyl fumarate; CC, calcium carbonate; TC, Talc. Inherent formulation issues of kinase inhibitors | 65

Table 3. Reported solubilities of smKIs as pure drug substance at given pH values and their types of commercial formulation.

Compound pKa pH Solubility (mg/mL) Dosage Formulation composition form Imatinib Mesylate X £ 5.5 > 1.6m [24] T/C Patent expired Gefitinib FB1 5.4; 7.2 [28] 1.0; 7.0 21m; <0.001 [28] T L, MC, CS, P, SLS, MS 1.2 Erlotinib Hydrochloride 5.4 [155] 2.0 0.4m [155] T L, MC, SSG, SLS, MS Sorafenib Tosylate X 1.0; 4.5 0.034m; 0.013 [33] T MC, CS, H, SLS, MS

m Dasatinib FB.H2O 3.1; 6.8; 10.8 [156] 2.6; 6.0 18.4 ; 0.008 [37] T L, MC, CS, HPC, MS Sunitinib Malate 9.0 [41] 1.2-6.8 >25m [41] C MN, P, CS, MS

Nintedanib The drug compound Nintedanib esilate is suspended in an oily base in its commercial formulation. Nintedanib is suspended in a mixture of medium-chain triglycerides (carrier) and hard fat (thickener) [181,182]. A patent from 2009 describes the development of the formulation [183]. It stated that hydrolytic degradation seems to be problematic for the compound. In combination with the high drug load, a lipophilic carrier suspension in a hydrophilic capsule was deemed appropriate. The formulation had a higher bioavailability than the tested hydrophilic and lipophilic-surfactant systems in rats. Inherent formulation issues of kinase inhibitors | 67

New formulations in literature and patents

Patents and exclusivities are still covering all of the smKIs, except for Imatinib [184,185]. While registered and approved alternative formulations are a long way off, there is a limited body of research published and patented at the time of writing. This section will 1.2 briefly discuss the most frequently reported and patented oral formulation types.

Solid dispersions Producing and characterizing the amorphous form of the smKIs is described throughout the patent body. Using that amorphous form in a pharmaceutical formulation is less frequently reported. This can be due to a too unstable amorphous form, a non-superior solubility or simply because the terrain is still unexplored. Xspray microparticles is one of the very few that give a description of the effect of amorphization on the dissolution of some smKIs [186]. The inventors use supercritical fluid precipitation to produce solid dispersions of Axitinib, Crizotinib, Dasatinib, Erlotinib, Gefitinib, Lapatinib, Nilotinib, Pazopanib, Sorafenib and Vemurafenib with different polymers. The patent presents a significant solubilization of the investigated compounds by the incorporation into -poly meric matrices. Godugu et al. describe a spray dried solid dispersion of Gefitinib that yields a 9-fold increase in rat AUC compared to free base Gefitinib [187]. The group of Truong found that a spray dried formulation of amorphous Sorafenib, a graft polymer and SLS increased the AUC by 1.8-fold in rats [188]. Nanologica patented a nanoporous formulation with loaded amorphous Dasatinib [189]. They report no dissolution data, but state that more Dasatinib is released in Simulated Intestinal Fluid from their formulation than form crystalline drug. Song et al. prepared various solid dispersion of Lapatinib and showed an increased solubility of the products in water with 0.2% SDS [190]. Unsol- vated Trametinib in a spray dried formulation had better dissolution characteristics than the commercial formulation, as patented by Ratiopharm GmBH [191].

Crystalline stabilization A small number of patents describe improved dissolution characteristics for set of smKIs by using excipients that stabilize crystalline polymorphs as solids or as solutes. Stabiliza- tion of an unstable, more soluble polymorph of Erlotinib hydrochloride with a hydrophilic polymer can lead to better dissolution profiles. This was shown by Synthon BV [192]. Liu et al. found that the solubility of Sorafenib can be markedly increased by formulating it with Polyvinylpyrrolidone vinyl acetate copolymer (PVP-VA). The drug-polymer interaction in solution provides a supersaturated state that nearly doubles the AUC in beagle 68 | Chapter 1.2

dogs [193]. The development of sustained release dosage forms of the relatively soluble Tofacitinib citrate was carried out by Pfizer [194]. This may well be an example of the development route of smKIs once the solubility issues are under control.

Cocrystals A cocrystal is defined as a homogenous crystalline material that is made up of two or more molecules in definite stoichiometric amounts held together by non-covalent forces[195]. The physicochemistry of the so-called non-covalent derivative may be very different from a drug salt form that is held together by ionic forces [196]. A patent filed in 2015 showed an increase in solubility of Gefitinib in cocrystalline form with certain carboxylic acids [197]. Basf Se prepared several cocrystals of Dasatinib and showed that especially the combination with methyl gallate increased the aqueous solubility to 42 µg/mL from 0.36 µg/mL of the dasatinib free base monohydrate [198]. The same was done for Nilotinib, which showed that the cocrystal with maleic acid had an almost 6-fold increase in solubility compared to the hydrochloride form [199]. Neither of the two are backed up with bioavailability data. Cocrystalline forms also improve dissolution of Lapatinib. Fabbrica Italiana Sintetici shows that a cocrystal of Lapatinib with Adipic acid has a higher solubility than Lapatinib ditosylate. In rats, however, this did not lead to an improvement of the bioavailability [200].

Discussion The currently approved smKIs have important places in clinical practice for registered indications. A large fraction of these compounds is under investigation for additional indications. The new and upcoming smKIs target many of the same kinases and are designed in a similar fashion as the already marketed ones. Their general physicochemical properties might not differ significantly from those of the older smKIs. A low and variable bioavailability is an enormous problem and may also lead to additional expenses in the patient treatment. High drug loading may be necessary to reach certain plasma levels for drugs with a low bioavailability with due consequences. The narrow therapeutic window of the drugs reflect in serious side effects and reduced activity above and below certain plasma concentration thresholds, respectively [17,201]. A variable bioavailability therefore often requires plasma levels to be monitored over a given time as part of a therapeutic drug monitoring (TDM) regimen which in part takes the advantages of oral therapy away. This is illustrated by the necessity of dose adjustments of smKIs in 25% of the patients treated with these drugs in our own institute. Inherent formulation issues of kinase inhibitors | 69

Poor solubility may also pose a problem for smKIs that do not bind to the active site of the kinase that they inhibit. Such smKIs can bind to a regulatory site of a kinase, e.g. Re- bastinib. The molecular structure of Rebastinib contains similar groups as the presented smKIs in table 2, which are necessary for binding [202]. Solubility data of this drug is not yet available in the public domain. However, based on its structure, rebastinib is expect- 1.2 ed to be poorly soluble.

Plain and relatively uncomplicated formulations may be troublesome and can have very costly consequences during early clinical development. The first clinical trial with vemurafenib is exemplary. During this trial, it was discovered that the formulated crystal polymorph exhibited a low bioavailability. A new formulation with amorphous vemurafenib was produced and the clinical trials had to be repeated [61]. Although such a case is probably an exception, it is worth noticing that this might have been prevented.

With the incidence of cancer on the rise and the value of smKI therapy thoroughly established, the challenges of further improving the therapy are gaining value [203]. With a disease that has such significant consequences for patients, therapeutic uncertainties are all the more undesired. Predicting and addressing these uncertainties, preferably in an early stage of development, seems more than appropriate. Simple and straightforward formulations are surely preferred in terms of cost effectiveness of development and ease-of-production. The experience with the smKI group teaches the most valued lesson that overall cost effectiveness and the ease-of-treatment may not at all be ben- efitted by such choices however. Speed and efficiency are becoming characteristics of drug development [204]. The future challenges will therefore lie in the implementation of thorough screening of formulations and dosage forms into the overall drug development process. To ensure efficiency, a useful translation from in vitro dissolution data to the in vivo setting is needed. Asin vitro-in vivo predictability from simple dissolution set- ups is often troublesome, the adoption of methods such as the gastrointestinal model TIM may prove to be valuable [205].

The solubility-induced variable bioavailability and the accompanying risks and costs may largely benefit from formulations with improved performance. While most patents and exclusivities are still pending and new or bioequivalent formulations are not yet eligible for approval, formulation research increasingly highlights the opportunities for improved drug forms of approved and to-be-approved smKIs. 70 | Chapter 1.2

Conclusions

Due to the very distinct targeting of the smKIs, biocompatible physicochemistry is driven to the edge. This places a strain upon bioavailability and presents challenges to formulation scientists.

‘Classical’ physical mixtures may work to achieve a relatively high bioavailability for some compounds, namely Imatinib, Ruxolitinib and Dabrafenib. This is certainly not the case for the majority of smKIs however. Bearing patents and exclusivities in mind, the past experiences can lead to new and innovative formulations that may provide further improvement of the efficacy of anticancer treatment. Inherent formulation issues of kinase inhibitors | 71

References

[1] M. Hojjat-Farsangi, Targeting non-receptor tient preferences for oral versus intravenous tyrosine kinases using small molecule inhib- palliative chemotherapy, J. Clin. Oncol. 15 itors: an overview of recent advances, J. Drug (1997) 110–115. Target. 24 (2016) 192–211. [17] N. Eckstein, L. Röper, B. Haas, H. Potthast, U. 1.2 [2] A.K.A. Gaumann, F. Kiefer, J. Alfer, S.A. Lang, Hermes, C. Unkrig, F. Naumann-Winter, H. E.K. Geissler, G. Breier, Receptor tyrosine ki- Enzmann, Clinical pharmacology of tyrosine nase inhibitors: Are they real tumor killers?, kinase inhibitors becoming generic drugs: the Int. J. Cancer. 138 (2016) 540–554. regulatory perspective, J. Exp. Clin. Cancer [3] G. Mirone, A. Shukla, G. Marfe, Signaling Res. 33 (2014) 15. mechanisms of resistance to EGFR- and An- [18] W.P. Janzen, Screening Technologies for Small ti-Angiogenic Inhibitors cancer, Crit. Rev. On- Molecule Discovery: The State of the Art, col. Hematol. 97 (2016) 85–95. Chem. Biol. 21 (2014) 1162–1170. [4] R.R. Kudchadkar, K.S.M. Smalley, L.F. Glass, [19] K. Nikolic, L. Mavridis, T. Djikic, J. Vucicevic, D. J.S. Trimble, V.K. Sondak, Targeted therapy in Agbaba, K. Yelekci, J.B.O. Mitchell, Drug De- melanoma., Clin. Dermatol. 31 (n.d.) 200–8. sign for CNS Diseases: Polypharmacological [5] S.W. Tas, C.X. Maracle, E. Balogh, Z. Szekan- Profiling of Compounds Using Cheminformat- ecz, Targeting of proangiogenic signalling ic, 3D-QSAR and Virtual Screening Methodol- pathways in chronic inflammation, Nat. Rev. ogies, Front. Neurosci. 10 (2016). Rheumatol. 12 (2015) 111–122. [20] S. Stegemann, F. Leveiller, D. Franchi, H. de [6] US Food and Drug administration (FDA), Jong, H. Lindén, When poor solubility be- CDER 2001 Report to the Nation, (2001). comes an issue: From early stage to proof of [7] US Food and Drug administration (FDA), FDA concept, Eur. J. Pharm. Sci. 31 (2007) 249– Prescribing Information Gleevec, (2001). 261. [8] P. Wu, T.E. Nielsen, M.H. Clausen, Small-mol- [21] Y. Kawabata, K. Wada, M. Nakatani, S. Yama- ecule kinase inhibitors: an analysis of FDA-ap- da, S. Onoue, Formulation design for poorly proved drugs, Drug Discov. Today. 21 (2016) water-soluble drugs based on biopharmaceu- 5–10. tics classification system: Basic approaches [9] European Medicines Agency (EMA), Assess- and practical applications, Int. J. Pharm. 420 ment report Jakavi (EPAR), (2012). (2011) 1–10. [10] US Food and Drug administration (FDA), Clin- [22] A. Dahan, A. Beig, D. Lindley, J.M. Miller, The ical Pharmacology and Biopharmaceutics Re- solubility–permeability interplay and oral view Jakafi, (2011). drug formulation design: Two heads are bet- [11] US Food and Drug administration (FDA), Clin- ter than one, Adv. Drug Deliv. Rev. 101 (2016) ical Pharmacology and Biopharmaceutics Re- 99–107. view Xeljanz, (2012). [23] European Medicines Agency (EMA), Assess- [12] European Medicines Agency (EMA), Assess- ment report Glivec (EPAR), (2004). ment report Xeljanz (EPAR), (2013). [24] US Food and Drug administration (FDA), Clin- [13] A. Fountas, L.-N. Diamantopoulos, A. Tsatsou- ical Pharmacology and Biopharmaceutics Re- lis, Tyrosine Kinase Inhibitors and Diabetes: A view Gleevec, (2001). Novel Treatment Paradigm?, Trends Endocri- [25] US Food and Drug administration (FDA), Clin- nol. Metab. 26 (2015) 643–656. ical Pharmacology and Biopharmaceutics Re- [14] R.A. Jibodh, J.S. Lagas, B. Nuijen, J.H. Beijnen, view Stivarga, (2012). J.H.M. Schellens, Taxanes: old drugs, new oral [26] European Medicines Agency (EMA), Assess- formulations., Eur. J. Pharmacol. 717 (2013) ment report Stivarga (EPAR), (2014). 40–6. [27] European Medicines Agency (EMA), Assess- [15] L. Benjamin, F.-E. Cotté, C. Philippe, F. -Mer ment report Iressa (EPAR), (2009). cier, T. Bachelot, G. Vidal-Trécan, Physicians’ [28] US Food and Drug administration (FDA), Clin- preferences for prescribing oral and intrave- ical Pharmacology and Biopharmaceutics Re- nous anticancer drugs: a Discrete Choice Ex- view Iressa, (2003). periment., Eur. J. Cancer. 48 (2012) 912–20. [29] European Medicines Agency (EMA), Assess- [16] G. Liu, E. Franssen, M.I. Fitch, E. Warner, Pa- ment report Tarceva (EPAR), (2005). 72 | Chapter 1.2

[30] US Food and Drug administration (FDA), Clin- ical Pharmacology and Biopharmaceutics Re- ical Pharmacology and Biopharmaceutics Re- view Mekinist, (2013). view Tarceva, (2004). [52] European Medicines Agency (EMA), Assess- [31] European Medicines Agency (EMA), Assess- ment report Votrient (EPAR), (2011). ment report Bosulif (EPAR), (2014). [53] US Food and Drug administration (FDA), Clin- [32] US Food and Drug administration (FDA), Clin- ical Pharmacology and Biopharmaceutics Re- ical Pharmacology and Biopharmaceutics Re- view Votrient, (2009). view Bosulif, (2012). [54] European Medicines Agency (EMA), Assess- [33] European Medicines Agency (EMA), Assess- ment report Vargatef (EPAR), (2015). ment report Nexavar (EPAR), (2006). [55] US Food and Drug administration (FDA), Clin- [34] US Food and Drug administration (FDA), Clin- ical Pharmacology and Biopharmaceutics Re- ical Pharmacology and Biopharmaceutics Re- view Vargatef, (2014). view Nexavar, (2005). [56] European Medicines Agency (EMA), Assess- [35] European Medicines Agency (EMA), Assess- ment report Zykadia (EPAR), (2015). ment report Inlyta (EPAR), (2012). [57] US Food and Drug administration (FDA), Clin- [36] US Food and Drug administration (FDA), Clin- ical Pharmacology and Biopharmaceutics Re- ical Pharmacology and Biopharmaceutics Re- view Zykadia, (2014). view Inlyta, (2012). [58] European Medicines Agency (EMA), Assess- [37] European Medicines Agency (EMA), Assess- ment report Xalkori (EPAR), (2013). ment report Sprycel (EPAR), (2006). [59] US Food and Drug administration (FDA), Clin- [38] US Food and Drug administration (FDA), Clin- ical Pharmacology and Biopharmaceutics Re- ical Pharmacology and Biopharmaceutics Re- view Xalkori, (2011). view Sprycel, (2006). [60] US Food and Drug administration (FDA), Clin- [39] European Medicines Agency (EMA), Assess- ical Pharmacology and Biopharmaceutics Re- ment report Imbruvica (EPAR), (2014). view Alecensa, (2015). [40] US Food and Drug administration (FDA), Clin- [61] European Medicines Agency (EMA), Assess- ical Pharmacology and Biopharmaceutics Re- ment report Zelboraf (EPAR), (2011). view Imbruvica, (2013). [62] US Food and Drug administration (FDA), Clin- [41] US Food and Drug administration (FDA), Clin- ical Pharmacology and Biopharmaceutics Re- ical Pharmacology and Biopharmaceutics Re- view Zelboraf, (2011). view Sutent, (2006). [63] European Medicines Agency (EMA), Assess- [42] European Medicines Agency (EMA), Assess- ment report Cotellic (EPAR), (2015). ment report Giotrif (EPAR), (2013). [64] US Food and Drug administration (FDA), Clin- [43] US Food and Drug administration (FDA), Clin- ical Pharmacology and Biopharmaceutics Re- ical Pharmacology and Biopharmaceutics Re- view Cotellic, (2015). view Giotrif, (2013). [65] European Medicines Agency (EMA), Assess- [44] European Medicines Agency (EMA), Assess- ment report Caprelsa (EPAR), (2013). ment report Tasigna (EPAR), (2009). [66] US Food and Drug administration (FDA), Clin- [45] US Food and Drug administration (FDA), Clin- ical Pharmacology and Biopharmaceutics Re- ical Pharmacology and Biopharmaceutics Re- view Caprelsa, (2011). view Tasigna, (2007). [67] European Medicines Agency (EMA), Assess- [46] European Medicines Agency (EMA), Assess- ment report Tagrisso (EPAR), (2016). ment report Tafinlar (EPAR), (2013). [68] US Food and Drug administration (FDA), Clin- [47] US Food and Drug administration (FDA), Clin- ical Pharmacology and Biopharmaceutics Re- ical Pharmacology and Biopharmaceutics Re- view Tagrisso, (2015). view Tafinlar, (2013). [69] European Medicines Agency (EMA), Assess- [48] European Medicines Agency (EMA), Assess- ment report Iclusig (EPAR), (2013). ment report Tyverb (EPAR), (2009). [70] US Food and Drug administration (FDA), Clin- [49] US Food and Drug administration (FDA), Clin- ical Pharmacology and Biopharmaceutics Re- ical Pharmacology and Biopharmaceutics Re- view Iclusig, (2012). view Tykerb, (2007). [71] European Medicines Agency (EMA), Assess- [50] European Medicines Agency (EMA), Assess- ment report Lenvima (EPAR), (2015). ment report Mekinist (EPAR), (2014). [72] US Food and Drug administration (FDA), Clin- [51] US Food and Drug administration (FDA), Clin- ical Pharmacology and Biopharmaceutics Re- Inherent formulation issues of kinase inhibitors | 73

view Lenvima, (2015). Tsuruoka, A. Inoue, J. Matsui, Distinct Binding [73] European Medicines Agency (EMA), Assess- Mode of Multikinase Inhibitor Lenvatinib Re- ment report Cometriq (EPAR), (2014). vealed by Biochemical Characterization, ACS [74] US Food and Drug administration (FDA), Clin- Med. Chem. Lett. 6 (2015) 89–94. ical Pharmacology and Biopharmaceutics Re- [86] L. Hu, Y. Zheng, Z. Li, Y. Wang, Y. Lv, X. Qin, view Cometriq, (2012). C. Zeng, Design, synthesis, and biological ac- [75] M. Rowland, T.N. Tozer, Clinical Pharmacoki- tivity of phenyl-pyrazole derivatives as BCR– 1.2 netics; Concepts and Applications, 3rd ed., ABL kinase inhibitors, Bioorg. Med. Chem. 23 Lippincott, Williams & Wilkins, Philadelphia, (2015) 3147–3152. 1994. [87] N.M. Levinson, S.G. Boxer, Structural and [76] L.Z. Benet, The role of BCS (biopharmaceutics Spectroscopic Analysis of the Kinase Inhibitor classification system) and BDDCS (biophar- Bosutinib and an Isomer of Bosutinib Binding maceutics drug disposition classification sys- to the Abl Tyrosine Kinase Domain, PLoS One. tem) in drug development., J. Pharm. Sci. 102 7 (2012) e29828. (2013) 34–42. [88] J.S. Tokarski, The Structure of Dasatinib (BMS- [77] M. Herbrink, B. Nuijen, J.H.M. Schellens, J.H. 354825) Bound to Activated ABL Kinase Do- Beijnen, Variability in bioavailability of small main Elucidates Its Inhibitory Activity against molecular tyrosine kinase inhibitors, Cancer Imatinib-Resistant ABL Mutants, Cancer Res. Treat. Rev. 41 (2015) 412–422. 66 (2006) 5790–5797. [78] M. Estudante, J.G. Morais, G. Soveral, L.Z. [89] A. Tse, G.M. Verkhivker, Molecular Determi- Benet, Intestinal drug transporters: an over- nants Underlying Binding Specificities of the view., Adv. Drug Deliv. Rev. 65 (2013) 1340– ABL Kinase Inhibitors: Combining Alanine 56. Scanning of Binding Hot Spots with Network [79] I. Pereira de Sousa, A. Bernkop-Schnürch, Analysis of Residue Interactions and Coevolu- Pre-systemic metabolism of orally adminis- tion, PLoS One. 10 (2015) e0130203. tered drugs and strategies to overcome it, J. [90] R. Roskoski, ErbB/HER protein-tyrosine kinas- Control. Release. 192 (2014) 301–309. es: Structures and small molecule inhibitors, [80] C.J. Tynan, V. Lo Schiavo, L. Zanet- Pharmacol. Res. 87 (2014) 42–59. ti-Domingues, S.R. Needham, S.K. Roberts, [91] Y. Yosaatmadja, S. Silva, J.M. Dickson, A. V. M. Hirsch, D.J. Rolfe, D. Korovesis, D.T. Clarke, Patterson, J.B. Smaill, J.U. Flanagan, M.J. M.L. Martin-Fernandez, A tale of the epider- McKeage, C.J. Squire, Binding mode of the mal growth factor receptor: The quest for breakthrough inhibitor AZD9291 to epi- structural resolution on cells, Methods. 95 dermal growth factor receptor revealed, J. (2016) 86–93. Struct. Biol. 192 (2015) 539–544. [81] P. Khanna, P.J. Chua, B.H. Bay, G.H. Baeg, The [92] S. Ravez, O. Castillo-Aguilera, P. Depreux, L. JAK/STAT signaling cascade in gastric carcino- Goossens, Quinazoline derivatives as antican- ma (Review)., Int. J. Oncol. 47 (2015) 1617– cer drugs: a patent review (2011 – present), 26. Expert Opin. Ther. Pat. 25 (2015) 789–804. [82] M. Pinne, J.L. Raucy, Advantages of cell- [93] Z. Zhan, J. Ai, Q. Liu, Y. Ji, T. Chen, Y. Xu, M. based high-volume screening assays to as- Geng, W. Duan, Discovery of Anilinopyrim- sess nuclear receptor activation during drug idines as Dual Inhibitors of c-Met and VEG- discovery., Expert Opin. Drug Discov. 9 (2014) FR-2: Synthesis, SAR, and Cellular Activity, 669–86. ACS Med. Chem. Lett. 5 (2014) 673–678. [83] P. Kohonen, R. Ceder, I. Smit, V. Hongisto, G. [94] G.J. Roth, R. Binder, F. Colbatzky, C. Dallinger, Myatt, B. Hardy, O. Spjuth, R. Grafström, Can- R. Schlenker-Herceg, F. Hilberg, S.-L. Wollin, cer biology, toxicology and alternative meth- R. Kaiser, Nintedanib: From Discovery to the ods development go hand-in-hand., Basic Clinic, J. Med. Chem. 58 (2015) 1053–1063. Clin. Pharmacol. Toxicol. 115 (2014) 50–8. [95] J. Li, N. Zhou, K. Luo, W. Zhang, X. Li, C. Wu, [84] R. Roskoski, Classification of small molecule J. Bao, In Silico Discovery of Potential VEG- protein kinase inhibitors based upon the FR-2 Inhibitors from Natural Derivatives structures of their drug-enzyme complexes, for Anti-Angiogenesis Therapy, Int. J. Mol. Pharmacol. Res. 103 (2016) 26–48. Sci. 15 (2014) 15994–16011. doi:10.3390/ [85] K. Okamoto, M. Ikemori-Kawada, A. Jestel, ijms150915994. K. von König, Y. Funahashi, T. Matsushima, A. [96] W.M. Eldehna, M. Fares, H.S. Ibrahim, M.H. 74 | Chapter 1.2

Aly, S. Zada, M.M. Ali, S.M. Abou-Seri, H.A. (TG101348), Leukemia. 28 (2014) 404–407. Abdel-Aziz, D.A. Abou El Ella, Indoline ureas [106] M.K. Kim, O. Bae, Y. Chong, Design, Synthe- as potential anti-hepatocellular carcinoma sis, and Molecular Docking Study of Flavonol agents targeting VEGFR-2: Synthesis, in vitro Derivatives as Selective JAK1 Inhibitors, Bull. biological evaluation and molecular docking, Korean Chem. Soc. 35 (2014) 2581–2584. Eur. J. Med. Chem. 100 (2015) 89–97. [107] Y. Duan, L. Chen, Y. Chen, X. Fan, c-Src Binds [97] Y. Shan, C. Wang, L. Zhang, J. Wang, M. Wang, to the Cancer Drug Ruxolitinib with an Active Y. Dong, Expanding the structural diversity of Conformation, PLoS One. 9 (2014) e106225. diarylureas as multi-target tyrosine kinase [108] N.K. Williams, R.S. Bamert, O. Patel, C. Wang, inhibitors, Bioorg. Med. Chem. 24 (2016) P.M. Walden, A.F. Wilks, E. Fantino, J. Ross- 750–758. doi:10.1016/j.bmc.2015.12.038. john, I.S. Lucet, Dissecting Specificity in the [98] F. Meng, Molecular Dynamics Simulation of Janus Kinases: The Structures of JAK-Specific VEGFR2 with Sorafenib and Other Urea-Sub- Inhibitors Complexed to the JAK1 and JAK2 stituted Aryloxy Compounds, J. Theor. Chem. Protein Tyrosine Kinase Domains, J. Mol. Biol. 2013 (2013) 1–7. 387 (2009) 219–232. [99] T. Pemovska, E. Johnson, M. Kontro, G.A. Re- [109] A. Kumar, V. Shanthi, K. Ramanathan, Com- pasky, J. Chen, P. Wells, C.N. Cronin, M. Mc- putational Investigation and Experimental Tigue, O. Kallioniemi, K. Porkka, B.W. Murray, Validation of Crizotinib Resistance Conferred K. Wennerberg, Axitinib effectively inhibits by C1156Y Mutant Anaplastic Lymphoma Ki- BCR-ABL1(T315I) with a distinct binding con- nase, Mol. Inform. 34 (2015) 105–114. formation, Nature. 519 (2015) 102–105. [110] J.J. Cui, M. Tran-Dube, H. Shen, M. Nambu, [100] M. McTigue, B.W. Murray, J.H. Chen, Y.-L. P.-P. Kung, M. Pairish, L. Jia, J. Meng, L. Funk, I. Deng, J. Solowiej, R.S. Kania, Molecular con- Botrous, M. McTigue, N. Grodsky, K. Ryan, E. formations, interactions, and properties as- Padrique, G. Alton, S. Timofeevski, S. Yamaza- sociated with drug efficiency and clinical per- ki, Q. Li, H. Zou, J. Christensen, B. Mroczkow- formance among VEGFR TK inhibitors, Proc. ski, S. Bender, R.S. Kania, M.P. Edwards, Natl. Acad. Sci. 109 (2012) 18281–18289. Structure Based Drug Design of Crizotinib [101] Y. Jia, J. Zhang, J. Feng, F. Xu, H. Pan, W. Xu, (PF-02341066), a Potent and Selective Dual Design, Synthesis and Biological Evaluation of Inhibitor of Mesenchymal–Epithelial Tran- Pazopanib Derivatives as Antitumor Agents, sition Factor (c-MET) Kinase and Anaplastic Chem. Biol. Drug Des. 83 (2014) 306–316. Lymphoma Kinase (ALK), J. Med. Chem. 54 [102] J. Zhang, X. Jiang, Y. Jiang, M. Guo, S. Zhang, (2011) 6342–6363. J. Li, J. He, J. Liu, J. Wang, L. Ouyang, Recent [111] Z. Ni, T.-C. Zhang, Computationally unraveling advances in the development of dual VEGFR how ceritinib overcomes drug-resistance mu- and c-Met small molecule inhibitors as anti- tations in ALK-rearranged lung cancer, J. Mol. cancer drugs, Eur. J. Med. Chem. 108 (2016) Model. 21 (2015) 175. 495–504. [112] X. Jiang, J. Zhou, J. Ai, Z. Song, X. Peng, L. [103] M. Pellowka, D. Merk, Advances in personal Xing, Y. Xi, J. Guo, Q. Yao, J. Ding, M. Geng, medicine - medicinal chemistry and pharma- A. Zhang, Novel tetracyclic benzo[b]carbazo- cology of vemurafenib and ivacaftor, Pharma- lones as highly potent and orally bioavailable zie. 68 (2013) 484–491. ALK inhibitors: Design, synthesis, and struc- [104] L. Ren, K.A. Ahrendt, J. Grina, E.R. Laird, A.J. ture—activity relationship study, Eur. J. Med. Buckmelter, J.D. Hansen, B. Newhouse, D. Chem. 105 (2015) 39–56. Moreno, S. Wenglowsky, V. Dinkel, S.L. Gloor, [113] Z. Song, M. Wang, A. Zhang, Alectinib: a G. Hastings, S. Rana, K. Rasor, T. Risom, H.L. novel second generation anaplastic lympho- Sturgis, W.C. Voegtli, S. Mathieu, The dis- ma kinase (ALK) inhibitor for overcoming covery of potent and selective pyridopyrim- clinically-acquired resistance, Acta Pharm. idin-7-one based inhibitors of B-RafV600E Sin. B. 5 (2015) 34–37. doi:10.1016/j. kinase, Bioorg. Med. Chem. Lett. 22 (2012) apsb.2014.12.007. 3387–3391. [114] H. Yari, M.R. Ganjalikhany, H. Sadegh, In [105] T. Zhou, S. Georgeon, R. Moser, D.J. Moore, silico investigation of new binding pocket A. Caflisch, O. Hantschel, Specificity and for mitogen activated kinase kinase (MEK): mechanism-of-action of the JAK2 tyrosine Development of new promising inhibitors, kinase inhibitors ruxolitinib and SAR302503 Comput. Biol. Chem. 59 (2015) 185–198. Inherent formulation issues of kinase inhibitors | 75

doi:10.1016/j.compbiolchem.2015.09.013. and free energy calculation studies of kinase [115] I. V. Hartung, S. Hammer, M. Hitchcock, R. inhibitors binding to active and inactive con- Neuhaus, A. Scholz, G. Siemeister, R. Bohl- formations of VEGFR-2, J. Mol. Graph. Model. mann, R.C. Hillig, F. Pühler, Optimization of 56 (2015) 103–112. allosteric MEK inhibitors. Part 2: Taming the [124] E. Weisberg, H.G. Choi, A. Ray, R. Barrett, J. sulfamide group balances compound distri- Zhang, T. Sim, W. Zhou, M. Seeliger, M. Cam- bution properties, Bioorg. Med. Chem. Lett. eron, M. Azam, J.A. Fletcher, M. Debiec-Rych- 1.2 26 (2016) 186–193. ter, M. Mayeda, D. Moreno, A.L. Kung, P.A. [116] K.D. Robarge, W. Lee, C. Eigenbrot, M. Ultsch, Janne, R. Khosravi-Far, J. V Melo, P.W. Manley, C. Wiesmann, R. Heald, S. Price, J. Hewitt, P. S. Adamia, C. Wu, N. Gray, J.D. Griffin, Dis- Jackson, P. Savy, B. Burton, E.F. Choo, J. Pang, covery of a small-molecule type II inhibitor J. Boggs, A. Yang, X. Yang, M. Baumgardner, of wild-type and gatekeeper mutants of BCR- Structure based design of novel 6,5 heter- ABL, PDGFRalpha, Kit, and Src kinases: nov- obicyclic mitogen-activated protein kinase el type II inhibitor of gatekeeper mutants., kinase (MEK) inhibitors leading to the discov- Blood. 115 (2010) 4206–16. ery of imidazo[1,5-a] pyrazine G-479, Bioorg. [125] E.P. Reddy, A.K. Aggarwal, The Ins and Outs Med. Chem. Lett. 24 (2014) 4714–4723. of Bcr-Abl Inhibition, Genes Cancer. 3 (2012) [117] T. Yogo, H. Nagamiya, M. Seto, S. Sasaki, H. 447–454. Shih-Chung, Y. Ohba, N. Tokunaga, G.N. Lee, [126] C. Fraser, J.C. Dawson, R. Dowling, D.R. C.Y. Rhim, C.H. Yoon, S.Y. Cho, R. Skene, S. Houston, J.T. Weiss, A.F. Munro, M. Muir, L. Yamamoto, Y. Satou, M. Kuno, T. Miyazaki, Harrington, S.P. Webster, M.C. Frame, V.G. H. Nakagawa, A. Okabe, S. Marui, K. Aso, M. Brunton, E.E. Patton, N.O. Carragher, A. Un- Yoshida, Structure-Based Design and Synthe- citi-Broceta, Rapid Discovery and Structure– sis of 3-Amino-1,5-dihydro-4 H -pyrazolopy- Activity Relationships of Pyrazolopyrimidines ridin-4-one Derivatives as Tyrosine Kinase 2 That Potently Suppress Breast Cancer Cell Inhibitors, J. Med. Chem. 59 (2016) 733–749. Growth via SRC Kinase Inhibition with - Ex [118] V. Simov, S. V. Deshmukh, C.J. Dinsmore, F. ceptional Selectivity over ABL Kinase, J. Med. Elwood, R.B. Fernandez, Y. Garcia, C. Gibeau, Chem. 59 (2016) 4697–4710. H. Gunaydin, J. Jung, J.D. Katz, B. Kraybill, B. [127] J. Rautio, H. Kumpulainen, T. Heimbach, R. Lapointe, S.B. Patel, T. Siu, H. Su, J.R. Young, Oliyai, D. Oh, T. Järvinen, J. Savolainen, Pro- Structure-based design and development of drugs: design and clinical applications, Nat. (benz)imidazole pyridones as JAK1-selective Rev. Drug Discov. 7 (2008) 255–270. kinase inhibitors, Bioorg. Med. Chem. Lett. [128] J.D. Oslob, S.A. Heumann, C.H. Yu, D.A. Allen, 26 (2016) 1803–1808. S. Baskaran, M. Bui, E. Delarosa, A.D. Fung, [119] M.A. Aziz, R.A.T. Serya, D.S. Lasheen, A.K. A. Hashash, J. Hau, S. Ivy, J.W. Jacobs, W. Lew, Abdel-Aziz, A. Esmat, A.M. Mansour, A.N.B. J. Maung, R.S. McDowell, S. Ritchie, M.J. Ro- Singab, K.A.M. Abouzid, Discovery of Potent manowski, J.A. Silverman, W. Yang, M. Zhong, VEGFR-2 Inhibitors based on Furopyrimidine T. Fuchs-Knotts, Water-soluble prodrugs of an and Thienopyrimidne Scaffolds as Cancer Tar- Aurora kinase inhibitor, Bioorg. Med. Chem. geting Agents, Sci. Rep. 6 (2016) 24460. Lett. 19 (2009) 1409–1412. [120] J. Han, S.J. Kaspersen, S. Nervik, K.G. Nørsett, [129] P.S. Charifson, W.P. Walters, Acidic and Basic E. Sundby, B.H. Hoff, Chiral 6-aryl-furo[2,3-d] Drugs in Medicinal Chemistry: A Perspective, pyrimidin-4-amines as EGFR inhibitors, Eur. J. J. Med. Chem. 57 (2014) 9701–9717. Med. Chem. 119 (2016) 278–299. [130] G. Thomas, Medicinal Chemistry: An Intro- [121] G. Thomas, Medicinal Chemistry: an introduction, Second Edition, in: Wiley (Ed.), 2nd duction, 2nd ed., John Wiley and sons, Chich- ed., West Sussex, 2007: pp. 49–51. ester, 2007. [131] B.J. Krieg, S.M. Taghavi, G.L. Amidon, G.E. [122] W. Xing, J. Ai, S. Jin, Z. Shi, X. Peng, L. Wang, Y. Amidon, In Vivo Predictive Dissolution: Com- Ji, D. Lu, Y. Liu, M. Geng, Y. Hu, Enhancing the paring the Effect of Bicarbonate and Phos- cellular anti-proliferation activity of pyridazi- phate Buffer on the Dissolution of Weak -Ac nones as c-met inhibitors using docking anal- ids and Weak Bases, J. Pharm. Sci. 104 (2015) ysis., Eur. J. Med. Chem. 95 (2015) 302–12. 2894–2904. [123] X. Wu, S. Wan, G. Wang, H. Jin, Z. Li, Y. Tian, Z. [132] L. Zhang, F. Wu, S.C. Lee, H. Zhao, pH-Depen- Zhu, J. Zhang, Molecular dynamics simulation dent Drug-Drug Interactions for Weak Base 76 | Chapter 1.2

Drugs: Potential Implications for New Drug metinib dimethyl sulfoxide and methods of Development., Clin. Pharmacol. Ther. 96 making and using thereof, US9181243B2. (2014) 266–77. (2013). [133] N.R. Budha, A. Frymoyer, G.S. Smelick, J.Y. [147] Australian Therapeutic Goods Administra- Jin, M.R. Yago, M.J. Dresser, S.N. Holden, tion, Australian Public Asessment Report L.Z. Benet, J.A. Ware, Drug absorption in- (AusPAR): Vandetanib, (2013). teractions between oral targeted anticancer [148] T.A. Albahri, Accurate prediction of the sol- agents and PPIs: is pH-dependent solubility ubility parameter of pure compounds from the Achilles heel of targeted therapy?, Clin. their molecular structures, Fluid Phase Pharmacol. Ther. 92 (2012) 203–13. Equilib. 379 (2014) 96–103. [134] G.A. Stephenson, A. Aburub, T.A. Woods, [149] F. Gharagheizi, A. Eslamimanesh, F. Farjood, Physical Stability of Salts of Weak Bases in the A.H. Mohammadi, D. Richon, Solubility Pa- Solid-State, J. Pharm. Sci. 100 (2011) 1607– rameters of Nonelectrolyte Organic Com- 1617. pounds: Determination Using Quantitative [135] R. Censi, P. Di Martino, Polymorph Impact Structure–Property Relationship Strategy, on the Bioavailability and Stability of Poorly Ind. Eng. Chem. Res. 50 (2011) 11382–11395. Soluble Drugs, Molecules. 20 (2015) 18759– [150] C.M. Hansen, Hansen Solubility Parameters: 18776. A User’s Handbook, 1st ed., CRC Press, Boca [136] S. Roy, R. Quiñones, A.J. Matzger, Structural Raton, Florida, 1999. and Physicochemical Aspects of Dasatinib Hy- [151] J. Goole, D.J. Lindley, W. Roth, S.M. Carl, K. drate and Anhydrate phases., Cryst. Growth Amighi, J.-M. Kauffmann, G.T. Knipp, The ef- Des. 12 (2012) 2122–2126. fects of excipients on transporter mediated [137] A. Martin, Non ideal solutions, in: Phys. absorption, Int. J. Pharm. 393 (2010) 17–31. Pharm., 4th ed., Lippincott, Williams & [152] European Medicines Agency (EMA), Europe- Wilkins, Baltimore, 1993: pp. 223–225. an Public Assessment Report (EPAR), (n.d.). [138] R. Pinal, Effect of molecular symmetry on [153] US Food and Drug administration (FDA), Dis- melting temperature and solubility., Org. Bio- solution methods, (n.d.). www.accessdata. mol. Chem. 2 (2004) 2692–9. fda.gov/scripts/cder/dissolution/ (accessed [139] C.M. Hansen, 50 Years with solubility param- March 1, 2016). eters—past and future, Prog. Org. Coatings. [154] E. R&D, Pharmaceutical composition of Len- 51 (2004) 77–84. vatinib, US8969379, 2005. [140] I.B. Mekjavic, Contribution of thermal and [155] US Food and Drug administration (FDA), nonthermal factors to the regulation of body Chemistry Review Tarceva, (2004). temperature in humans, J. Appl. Physiol. 100 [156] Australian Therapeutic Goods Administra- (2006) 2065–2072. tion, Australian Public Assessment Report [141] G. Kelly, Body temperature variability (Part (AUSPAR) Dasatinib, (2011). 1): a review of the history of body tempera- [157] US Food and Drug administration (FDA), ture and its variability due to site selection, Chemistry Review Tasigna, (2007). biological rhythms, fitness, and aging., Al- [158] US Food and Drug administration (FDA), tern. Med. Rev. 11 (2006) 278–93. Chemistry Review Tykerb, (2007). [142] F.S. Mortimer, MELTING POINT, LATENT HEAT [159] Australian Therapeutic Goods Administra- OF FUSION AND SOLUBILITY, J. Am. Chem. tion, Australian Public Assessment Report Soc. 44 (1922) 1416–1429. (AUSPAR) Lapatinib, (2012). [143] E. Thomas, J. Rubino, Solubility, melting point [160] Australian Therapeutic Goods Administra- and salting-out relationships in a group of tion, Australian Public Assessment Report secondary amine hydrochloride salts, Int. J. (AUSPAR) Pazopanib, (2013). Pharm. 130 (1996) 179–185. [161] Australian Therapeutic Goods Administra- [144] B. Hancock, The use of solubility parameters tion, Australian Public Assessment Report in pharmaceutical dosage form design, Int. J. (AUSPAR) Ruxolitinib, (2014). Pharm. 148 (1997) 1–21. [162] US Food and Drug administration (FDA), Pre- [145] C.M. Hansen, The three dimensional solubil- scribing information Xalkori, (2011). ity parameter and solvent diffusion coeffi- [163] M. Laberge, J. Grenier, D. Armas, Evaluation cient, Danmarks Tekniske Hojskole, 1967. of the effect of food or administration ofa [146] H.P. Pharmaceutical, Solvate form M of tra- proton pump inhibitor on the single-dose Inherent formulation issues of kinase inhibitors | 77

plasma pharmacokinetics of cabozantinib WO2010114928. (2010). in healthy adults subjects, in: n.d.: p. AAPS– [179] Bayer, Coated pharmaceutical composition M1344. containing regorafenib, US 20140065212, [164] Australian Therapeutic Goods Administra- 2013. tion, Australian Public Assessment Report [180] Bayer, New pharmaceutical compositions (AUSPAR) Tofacitinib, (2015). comprising 4-(4-(3-(4-chloro-3-trifluo- [165] Australian Therapeutic Goods Administra- romethyl-phenyl)-ureido)-3-fluoro-phe- 1.2 tion, Australian Public Assessment Report noxy)-pyridine-2-carboxylic acid for the treat- (AUSPAR) Afatinib, (2014). ment of hyper-proliferative disorders, WO [166] Australian Therapeutic Goods Administra- 2006026500 A1, 2005. tion, Product information Afatinib, (2014). [181] B. Ingelheim, Pharmaceutical combination, [167] Australian Therapeutic Goods Administra- US20110178099A1, 2009. tion, Australian Public Assessment Report [182] A. Larsen, Method for treating colorectal can- (AUSPAR) Trametinib, (2014). cer, WO2010081817A1, 2010. [168] US Food and Drug administration (FDA), Pre- [183] B. Ingelheim, Capsule pharmaceutical dosage scribing information Zykadia, (2014). from comprising a suspension formulation [169] US Food and Drug administration (FDA), Pre- of an indoline derivative, wo2009147212A1, scribing information Alecensa, (2014). 2008. [170] US Food and Drug administration (FDA), [184] US Food and Drug administration (FDA), FDA Chemistry Review Cotellic, (2015). Orange Book Imatinib, (2016). [171] US Food and Drug administration (FDA), [185] European Medicines Agency (EMA), Euro- Chemistry Review Tagrisso, (2015). pean Public Assessment Report Imatinib, [172] Boehringer-Ingelheim, Product monograph (2016). Ofev, (2015). [186] Xspray microparticles Ab, A method for pro- [173] S. Gaisford, M. Saunders, Essentials of Phar- ducing stable, amorphous nanoparticles maceutical Preformulation, 1st ed., Wi- comprising at least one protein kinase inhibi- ley-Blackwell, Chichester, 2013. tor and at least one polymeric stabilizing and [174] D.K. Mishra, V. Dhote, A. Bhargava, D.K. Jain, matrix forming component, WO 2013105894 P.K. Mishra, Amorphous solid dispersion A1, 2012. technique for improved drug delivery: basics [187] C. Godugu, R. Doddapaneni, A.R. Patel, R. to clinical applications, Drug Deliv. Transl. Singh, R. Mercer, M. Singh, Novel Gefitinib Res. 5 (2015) 552–565. doi:10.1007/s13346- Formulation with Improved Oral Bioavail- 015-0256-9. ability in Treatment of A431 Skin Carcinoma., [175] K.T. Savjani, A.K. Gajjar, J.K. Savjani, Drug Sol- Pharm. Res. 33 (2016) 137–54. doi:10.1007/ ubility: Importance and Enhancement Tech- s11095-015-1771-6. niques, ISRN Pharm. 2012 (2012) 1–10. [188] D.H. Truong, T.H. Tran, T. Ramasamy, J.Y. Choi, [176] W.L. Chiou, S. Riegelman, Pharmaceutical H.-G. Choi, C.S. Yong, J.O. Kim, Preparation applications of solid dispersion systems., J. and characterization of solid dispersion us- Pharm. Sci. 60 (1971) 1281–302. ing a novel amphiphilic copolymer to en- [177] N. Shah, R.M. Iyer, H.-J. Mair, D. Choi, H. Tian, hance dissolution and oral bioavailability of R. Diodone, K. Fahnrich, A. Pabst-Ravot, K. sorafenib, Powder Technol. 283 (2015) 260– Tang, E. Scheubel, J.F. Grippo, S.A. Moreira, Z. 265. doi:10.1016/j.powtec.2015.04.044. Go, J. Mouskountakis, T. Louie, P.N. Ibrahim, [189] Nanologica Ab, Super-saturated delivery H. Sandhu, L. Rubia, H. Chokshi, D. Singhal, vehicles for poorly water-soluble pharma- W. Malick, Improved Human Bioavailability ceutical and cosmetic active ingredients, EP of Vemurafenib, a Practically Insoluble Drug, 2616050 A1, 2011. Using an Amorphous Polymer-Stabilized Solid [190] Y. Song, X. Yang, X. Chen, H. Nie, S. Byrn, J.W. Dispersion Prepared by a Solvent-Controlled Lubach, Investigation of drug-excipient inter- Coprecipitation Process, J. Pharm. Sci. 102 actions in lapatinib amorphous solid disper- (2013) 967–981. sions using solid-state NMR spectroscopy., [178] L. Roche, Propane- i-sulfonic acid {3- [5- (4 Mol. Pharm. 12 (2015) 857–66. -chloro-phenyl) -1h-pyrrolo [2, 3-b] pyr- [191] Ratiopharm GmBH, Pharmaceutical composi- idine-3-carbonyl] -2, 4-difluoro-pheny l tion comprising trametinib, EP 2913048 A1, } -amide compositions and uses thereof, 2014. 78 | Chapter 1.2

[192] Synthon BV, Pharmaceutical composition WO 2015162007, 2015. comprising Erlotinib hydrochloride, WO [201] A. Arora, E.M. Scholar, Role of tyrosine kinase 2014118112, 2014. inhibitors in cancer therapy., J. Pharmacol. [193] C. Liu, Z. Chen, Y. Chen, J. Lu, Y. Li, S. Wang, Exp. Ther. 315 (2005) 971–9. G. Wu, F. Qian, Improving Oral Bioavailability [202] W.W. Chan, S.C. Wise, M.D. Kaufman, Y.M. of Sorafenib by Optimizing the “Spring” and Ahn, C.L. Ensinger, T. Haack, M.M. Hood, J. “Parachute” Based on Molecular Interaction Jones, J.W. Lord, W.P. Lu, D. Miller, W.C. Patt, Mechanisms, Mol. Pharm. 13 (2016) 599– B.D. Smith, P.A. Petillo, T.J. Rutkoski, H. Te- 608. likepalli, L. Vogeti, T. Yao, L. Chun, R. Clark, [194] Pfizer Inc, Tofacitinib oral sustained release P. Evangelista, L.C. Gavrilescu, K. Lazarides, dosage forms, WO 20140271842 A1, 2014. V.M. Zaleskas, L.J. Stewart, R.A. Van Etten, [195] C.B. Aakeröy, M.E. Fasulo, J. Desper, Cocrystal D.L. Flynn, Conformational Control Inhibition or salt: does it really matter?, Mol. Pharm. 4 of the BCR-ABL1 Tyrosine Kinase, Includ- (n.d.) 317–22. ing the Gatekeeper T315I Mutant, by the [196] E. Stoler, J. Warner, Non-Covalent Deriva- Switch-Control Inhibitor DCC-2036, Cancer tives: Cocrystals and Eutectics, Molecules. 20 Cell. 19 (2011) 556–568. (2015) 14833–14848. [203] WHO, Cancer Factsheet no 297, 2015. [197] Council of Scientific & Indistrial research, [204] US Food and Drug administration (FDA), Fast Pharmaceutical cocrystals of gefitinib, WO Track, Breakthrough, Therapy, Accelerated 2015170345 A1, 2015. Approval, Priority review, (2015). fda.gov/for- [198] Basf SE, Multicomponent crystals comprising patients/approvals/fast/ucm20041766.htm dasatinib and selected cocrystal formers, WO (accessed August 19, 2016). 2013186726 A2, 2013. [205] M. Minekus, The TNO Gastro-Intestinal [199] Basf SE, Multicomponent crystalline system Model (TIM), in: Impact Food Bioact. Heal., comprising Nilotinib and selected co-crystal Springer International Publishing, Cham, formers, WO 2014060449 A1, 2013. 2015: pp. 37–46. [200] FIS, Co-crystals of Lapatinib monoacid salts,

Chapter 1.3

High-tech drugs in creaky formulations

M. Herbrink1, B. Nuijen1, J.H.M. Schellens2,3, J.H. Beijnen1,3

Abstract

Recent literature reviews and registration documents covering novel Signal Transduction Inhibitors in the treatment of cancer paint a picture of inefficiency and variability, where formulation improvements could be valuable. In this article, we discuss apparent drug design flaws as we impose the current standard formulation practice. High-tech drugs in creaky formulations | 83

Combining personalized chemotherapy with oral drug ingestion has demonstrated to be an almost golden duo. The success of this combination has been illustrated by the registration of 28 oral signal transduction inhibitors (STIs) in the past years. This group of therapeutics has proven its value in the clinic and a large number of the member drugs have become the standard care for an increasing amount of tumor types. Despite the efficient and specific nature of the drugs, the full potential of this inhibitor class has yet to be realized. In harsh contrast to the pharmacological fine tuning done for this group 1.3 stands the nascent development of their pharmaceutical formulations. At least 16 of these highly specific drugs, with sometimes life lengthening properties, suffer from low bioavailability and high variability, which in many cases could be improved with more optimal formulations [1].

Recent analyses of the pharmacokinetic properties of STI formulations present a trou- bling picture. As defining properties, we are confronted with low, fluctuating and highly susceptible exposures due to restricting absorption and/or first pass clearance [2,3]. The results of these studies expose a situation where an apparent imbalance exists between the development of the drug substance and that of the formulation. Biopharmaceutical issues and considerations are seemingly not prioritized by drug innovators where absorption is a limiting factor. As a large majority of STIs are classified as either unfavorable Biopharmaceutical Classification System (BCS) class II or IV(further explained and -pre sented in figure 1), additional attention to formulation design was especially warranted here.

Inevitably, this leads to situations where some drugs are being wasteful. Firstly, in the most literal sense, since large parts of the drug substances are not absorbed into the systemic circulation. These parts are subsequently discarded. An exemplary case is pazopanib of which only 21% is absorbed into the systemic circulation [1]. Secondly, further consideration of the pricing of personalized chemotherapy makes these matters even more deplorable. In the dire patient’s experience of cancer, every additional uncertainty is one too many and inadequate medication forms should not be added to this. Most worrying of all, is the imaginable loss of life quantity and quality that patients might suffer due to insufficient drug efficacy over time. 84 | Chapter 1.3

Figure 1. Currently marketed STIs in their respective BCS-class. Class 1, drugs with good solubility and permeability; Class II, drugs with poor solubility and good permeability; Class III, drugs with good solubility and poor permeability; Class IV, drugs with both poor solubility and permeability. 1-2, Imatinib, Ruxolitinib; 3-4, Afatinib, Cobimetinib; 5, Osimertinib; 6-17, Erlotinib, Gefitinib, Dasatinib, Lapatinib, Pazopanib, Vande- tanib, Ponatinib, Cabozantinib, Regorafenib, Axitinib, Ibrutinib, Dabrafenib; 18-21, Sorafenib, Nintedanib, Alectinib, Lenvatinib; 22-28, Sunitinib, Nilotinib, Crizotinib, Vemurafenib, Bosutinib, Trametinib, Ceritinib. [1,9] Changing the formulation of a class II or IV compound may shift its absorption behavior towards class I and III, respectively.

The variability in exposure has drawbacks that stretch beyond the obvious impact on drug efficacy and possible toxicity. In order to avoid as much (intrapatient) variability as possible, patients are requested to combine or avoid intake with food and certain co- medication. To restrain the interpatient variability, patients are subjected to therapeutic drug monitoring with increasing frequency [4]. In fact, in our institute dose corrections are warranted in 25% of the cases for patients treated with STIs due to inadequate exposure. This current practice entails a costly, high maintenance therapy.

The fact that at least 50% of the STI formulations are presently performing below what is reasonably achievable is further highlighted by the effect that food and comedication High-tech drugs in creaky formulations | 85

can have on drug exposure [1,5]. Ingesting food concomitantly with an STI generally can have a profound effect on the exposure of that drug [1,3]. A clear example of effect is the significant increase in exposure up to 4.7-fold of pazopanib, vemurafenib and lapatinib when they are taken with food [6]. Such effects reveal solubility alterations in different environments and are often good indicators of the poor performance of the pharmaceutical formulation. The same is true for pH effects; acidification of the GI-tract has been shown to lead to additional exposure to Erlotinib, whilst acid-reducing drugs can sharply 1.3 lower the bioavailability of a large group of STIs [1,7].

Therapeutic specificity is this drug class’ greatest asset. The molecular structures are tai- lor made and well balanced out to bind and inhibit the proteins that have gone haywire in various tumor types.

Yet these structures also turn out to be complicating factors that should be recognized and taken into account when developing STI formulations. Appropriate binding and inhibiting moieties often contain relatively lipophilic structures that make pure drug dissolution increasingly difficult. This is especially true for the STI group. Data from pharmacokinetic studies show a strong correlation between drug solubility, bioavailability and pharmacokinetic variability thereof [8]. This combination of facts should have sparked ample response in formulative development. For the STIs, at least, the efforts and -at tempts at improving biopharmaceutical properties are only marginally available.

In the present collection of STI drugs appropriate effort put into the formulations has ac- tually led to drastic solubility and bioavailability improvement. The solubility of five compounds (Imatinib, Ruxolitinib, Dabrafenib, Cobimetinib and Osimertinib) was accommo- dated by the selection of a relatively soluble salt form [9]. A statement could be made in favor of formulating these drugs as tablets or capsules from simple powder mixtures. Yet for the vast remaining majority this is certainly not the case. Contrary to the desir- ability of an intricate formulation test cycle, all but three drugs are marketed as physical mixtures. Perhaps the most infamous of the three is Vemurafenib. Vemurafenib’s first clinical trial formulation was swiftly altered to Micro Precipitated Bulk which increased the solubility 30-fold, but only after the previous Phase I trial revealed a very poor bioavailability due to solubility issues. This belated solubility-improving alteration in the formulation also increased patient exposure by a 5-fold [10]. This alteration is a typical case study that can be added to the before mentioned wasteful state of affairs. 86 | Chapter 1.3

An increasing number of studies illuminate an array of methods that can improve drug solubility and subsequently show a related boost in exposure. Techniques that produce solid dispersions, cocrystals and nanoformulations have all proven to be beneficial for STIs. Even simple changes, such as a change in excipient composition, can result in a better performance [11]. Inclusion of such techniques into early drug development stages may prove to be valuable ventures and lead to more efficient drug formulations.

The rate of drug discovery and development has sped up in the past years. Along with accelerated authority registration it has brought new therapeutic options to patient -fast er than ever before. It seems, however, as if the hastened pace of personalized cancer drug development has a dubious side as well. In this field where time is of the essence, simple and time-saving formulation development is favored. These formulations are designed to roughly deliver the drug exposure that is needed for a therapeutic effect, albeit with large deviations. Through the recent decades we have learned that a development course like this may save the pharmaceutical industry from certain effort and costs. In -re ality, however, the efforts and costs are merely postponed until they are left to be dealt with by clinicians and eventually, the patients.

The realization that the virtue of cancer therapy is not just the sum, but the product of all the actions and decisions taken from drug discovery to patient guidance, is eminent in this matter. Hence, it follows that the drug manufacturers play a significant role therein. Drug manufacturers should hold themselves to a self-explanatory intrinsic ethical commitment to optimize a drug’s mode of administration. When such a commitment is well organized in an early stage of development, commercial and financial interests are unlikely to be compromised.

With the incidence of cancer still on the rise and the molecular exploration of its causes still ongoing, it is very likely that more STIs are to appear in the foreseeable future. As long as the largest portion of cancer care’s weight is still pulled by chemotherapy and the shift towards personalized drugs continues, the issue of drug performance will remain a crucial one. To ensure a future where cancer drugs approach optimal capability, in both financial and therapeutic sense, the pharmaceutical field must increase investments in chemotherapy formulation. High-tech drugs in creaky formulations | 87

References

[1] US Food and Drug administration (FDA), Clin- van der Holt, J.G. Aerts, T. van Gelder, R.H.J. ical Pharmacology and Biopharmaceutics Re- Mathijssen, Influence of the Acidic Beverage views, (n.d.). Cola on the Absorption of Erlotinib in Patients [2] B. Wulkersdorfer, M. Zeitlinger, M. Schmid, With Non-Small-Cell Lung Cancer., J. Clin. On- Pharmacokinetic Aspects of Vascular Endothe- col. 34 (2016) 1309–14. lial Growth Factor Tyrosine Kinase Inhibitors., [8] M. Sato, M. Narukawa, Factors affecting in- Clin. Pharmacokinet. 55 (2016) 47–77. trasubject variability of PK exposure: absolute [3] M. Herbrink, B. Nuijen, J.H.M. Schellens, J.H. oral bioavailability and acidic nature of drugs., 1.3 Beijnen, Variability in bioavailability of small Int. J. Clin. Pharmacol. Ther. 53 (2015) 955–62. molecular tyrosine kinase inhibitors, Cancer [9] European Medicines Agency (EMA), European Treat. Rev. 41 (2015) 412–422. Public Assessment Report (EPAR), (n.d.). [4] J.H. Beumer, Without therapeutic drug mon- [10] N. Shah, R.M. Iyer, H.-J. Mair, D. Choi, H. Tian, itoring, there is no personalized cancer care., R. Diodone, K. Fahnrich, A. Pabst-Ravot, K. Clin. Pharmacol. Ther. 93 (2013) 228–30. Tang, E. Scheubel, J.F. Grippo, S.A. Moreira, Z. [5] R.H.J. Mathijssen, A. Sparreboom, J. Verweij, Go, J. Mouskountakis, T. Louie, P.N. Ibrahim, Determining the optimal dose in the develop- H. Sandhu, L. Rubia, H. Chokshi, D. Singhal, W. ment of anticancer agents., Nat. Rev. Clin. On- Malick, Improved Human Bioavailability of Ve- col. 11 (2014) 272–81. murafenib, a Practically Insoluble Drug, Using [6] A.E.C.A.B. Willemsen, F.J.E. Lubberman, J. Tol, an Amorphous Polymer-Stabilized Solid Dis- W.R. Gerritsen, C.M.L. van Herpen, N.P. van persion Prepared by a Solvent-Controlled Co- Erp, Effect of food and acid-reducing agents precipitation Process, J. Pharm. Sci. 102 (2013) on the absorption of oral targeted therapies 967–981. in solid tumors., Drug Discov. Today. 21 (2016) [11] M. Herbrink, J.H.M. Schellens, J.H. Beijnen, B. 962–76. Nuijen, Inherent formulation issues of kinase [7] R.W.F. van Leeuwen, R. Peric, K.G.A.M. Hus- inhibitors., J. Control. Release. 239 (2016) 118– saarts, E. Kienhuis, N.S. IJzerman, P. de Bruijn, 27. C. van der Leest, H. Codrington, J.S. Kloover, B.

Chapter 2

Pharmaceutical analysis of oral anticancer drug substances

Chapter 2.1

Thermal study of pazopanib hydrochloride

M. Herbrink1, J.H.M. Schellens2,3, J.H. Beijnen1,3, B. Nuijen1

Abstract

The currently marketed formulation of pazopanib hydrochloride has a poor bioavailability and pharmacokinetic profile. An alternative formulation of the drug might have an improved performance. Knowledge about the thermal behavior of the drug may be instrumental in the development process of such a formulation. The aim of this study was to thermally characterize raw bulk drug material of the multiple kinase inhibitor pazopanib hydrochloride using different analytical techniques. Solid state characterization was carried out with Fourier transform infrared (FT-IR) spectroscopy and X-ray diffraction (XRD). Thermal characterization was performed with thermogravimetric analysis (TG) and differential scanning calorimetry (DSC). Thermal and thermodynamic parameters were assessed using the Ozawa-Flynn-Wall, Friedman and Kissinger-Akahira-Sunose isoconversional methods. Thermal study of pazopanib hydrochloride | 93

Introduction

Pazopanib, i.e. 5-({4-[(2,3-Dimethyl-2H-indazol-6-yl)methylamino]pyrimidin-2-yl}amino)-2-methylbenzenesulfonamide, figure 1, is a fully synthetic multiple kinase inhibitor that is currently trademarked as Votrient®. It is indicated for the treatment of advanced renal cell carcinoma and advanced tissue cell carcinoma [1,2]. In the commercial formulation, pazopanib is processed as the monohydrochloride salt (figure 1). The marketed tablet formulation ,however, has a low and variable bioavailability [3,4]. Although pazopanib is very potent and suitable for certain tumor types it may not consistently reach its target in sufficient quantities. As a results, the therapy may therefore at times 2.1 be suboptimal [5]. Pazopanib can be classified as a class II (bioavailability impaired by solubility) drug according to the Biopharmaceutical Classification System [3,6]. Changing the formulation of pazopanib and improving the bioavailability through solubility improvement may be beneficial for the treatment of patients eligible for pazopanib therapy. Pazopanib has been taken up in the Priority New Monograph list of the FDA [7]. To the best of the authors’ knowledge, a thorough description of thermal behavior of pazopanib hydrochloride is not yet presented in literature. Formulation development that follows Quality by Design (QbD) principles should include a full assessment of drug characteristics[8]. Such an approach will allow for critical drug properties to be identified timely. This may benefit future formulation studies where new methods of production may be designed based on and in agreement with the critical properties such as thermal and thermodynamic parameters [9–12]. Examples thereof may be the determination of the processing conditions of techniques such as hot-melt extrusion and spray drying and the more accurate prediction of storage conditions and shelf half-life [13–15]. No information is found in the literature concerning thermal and thermodynamic parameters of pazopanib hydrochloride.

This study seeks to provide the thermal properties of pazopanib hydrochloride which may be relevant for formulation studies and quality control purposes. This article provides a description of relevant methods for the solid state characterization and thermal analysis of pazopanib hydrochloride. It continues to present the results from X-ray diffraction, Infrared spectroscopic, thermogravimetric and calorimetric analyses of pazopanib hydrochloride. In the last part of this paper the evaluation of the thermal analysis of pazopanib hydrochloride will be discussed. 94 | Chapter 2.1

O NH2 S O

N HN C lH N

N N N

Figure 1. Molecular structure of pazopanib hydrochloride

Materials and methods

Materials Pazopanib hydrochloride was purchased from Avachem Scientific (San Antonio, TX, USA) and was supplied with a certificate of analysis (CoA) with >99% purity. The reference standards of pazopanib and pazopanib hydrochloride were acquired from AlsaChim (Illkirch, France) and SelleckChem (Houston, TX, USA) both with CoAs. Each analysis was performed of the hydrochloride was performed with both batches to exclude any possible interference by batch-to-batch irregularities.

Powder X-ray diffraction analysis (XRD) X-ray diffraction of powder samples was performed with an X’pert pro diffractometer equipped with an X-celerator (PANanalytical, Almelo, The Netherlands). Samples were placed in a 0.5 mm deep metal sample holder. The analyses were performed at 20 oC. Samples were scanned at a current of 30 mA and a tension of 40 kV. The scanning grange was 10 – 60 °2-θ, with a step size of 0.020 °2-θ and a scanning speed of 0.002 °2-θ per second. The obtained spectra were processed by using the HighScore software v4.2 (PA- Nanalytical). The analysis was performed in triplicate and the spectra were averaged. Thermal study of pazopanib hydrochloride | 95

Fourier Transform Infrared (FTIR) spectroscopy FTIR spectra were recorded from 650 to 4000 cm-1 with a resolution of 2 cm-1 with a FT-IR 8400S Spectrophotometer equipped with a golden gate® which functions as attenuated total reflectance (ATR) attachment (Shimadzu, ‘s-Hertogenbosch, the Netherlands). A total of 64 scans were averaged into one spectrum using IR Solution software v1.40 (Shi- madzu). Background spectra were recorded and subtracted during each analysis. Three samples were collected and analyzed at different locations of the samples to verify uniformity.

Thermogravimetric analysis (TG) 2.1 TG and Differential Thermal Gravimetry (DTG) were performed on a Q50 thermogravimetric analyzer (TA Instruments, New Castle, DE, USA) under a nitrogen flow of 50 mL min-1. The sample (approximately 10 mg) was weighted into a platinum sample pan (TA Instruments). Samples were heated in a temperature range of 25-1000 oC. Temperature calibration was carried using a high purity magnetic reference (nickel) for Curie temperature determination. The analysis was performed in triplicate and the spectra were averaged. The data were analyzed with the Advantage software v5.5 (TA Instruments). Non-isothermal data were investigated using the model free Ozawa-Flynn-Wall (OFW), Friedman (F) and Kissinger-Akahira-Sunose (KAS) isoconversional kinetic methods [16– 18]. The use of model free methods offers the advantage of analyzing thermal degradation data without the consideration of the reaction model. By using the temperature dependence of the activation energy the methods are capable of distinguishing in com- plexity of the degradation process into single and multi-step degradation kinetics [19]. The methods require the analysis of the fraction of degradation (α) at different heating rates in order to give an estimation of the activation energy and thermodynamic parameters. This is recommended by the International Confederation for Thermal Analysis and Calorimetry [20].

The heating rates were 2.5, 5.0, 10, 20 and 40o C min-1 in the before mentioned temperature range. The kinetic methods assume that the isothermal degradation rate, dα/dt, is linearly related to the temperature-dependent constant, k(T), and a function of degradation, f(α), as in equation 1 [21,22].

(1) !" 12 !# = 𝑘𝑘(𝑇𝑇)𝑓𝑓(𝛼𝛼) = 𝐴𝐴exp /− 345 𝑓𝑓(𝛼𝛼) 96 | Chapter 2.1

In conditions that are not isothermal, equation 1 turns into equation 2.

(2) !" ' -./ #(") = ( exp ( 01 )d𝑇𝑇 o -1 -1 Where β is the heating rate ( C min ), A is the pre-exponential factor (min ), Ea, is the activation energy (J mol-1) and R is the gas constant (J K-1mol-1).

The OZW, F and KAS methods may be used to calculate the activation energy under the assumption that f(α) does not change with α in one degradation phase. Equation 2 can be rewritten as equation 3, 4 and 5 for the OZW, F and KAS methods respectively [23].

The value of Ea can be estimated from the slope of lnβ-1/T (OZW), ln(β(da/dt))-1/T (F) and ln(β/T2)-1/T (KAS) curves [24–27].

(3) %&' &' ln 𝛽𝛽 = ln ( − 5.331 − 1.052 (0 (4) %& ./ ln(𝛽𝛽 %') = ln 𝐴𝐴𝐴𝐴(𝛼𝛼) − 0' (5) $ )* 2, & ln #% ' = ln #+,-(/)' − 3% The estimation of the pre-exponential factor was carried out in the assumption that the kinetics of the decomposition process can be described by the phase-boundary model (R2), with kinetic exponent n = 2 and a four degree ‘rational approximation’ [28,29].

In the extent of the estimated pre-exponential factor, the change of entropy (S≠), enthalpy (H≠) and Gibbs free energy (G≠) for the activated complex were calculated using equations 6-8 [30,31]:

(6) # () 𝛥𝛥𝛥𝛥 = 𝑅𝑅 ln *+,- (7) # 𝛥𝛥𝛥𝛥 = 𝐸𝐸& − 𝑅𝑅𝑅𝑅 (8) # # # 𝛥𝛥𝛥𝛥 = 𝛥𝛥𝐻𝐻 − 𝑇𝑇𝑇𝑇𝑆𝑆 Where H is the Planck’s constant (J s), e is the Neper (Euler’s) number (-), χ is the transition factor (-) which is unity for monomolecular reactions [32], k is the Boltzmann constant (J K-1) and T is the peak temperature of the DTG curve (K). Thermal study of pazopanib hydrochloride | 97

Differential scanning calorimetry (DSC) DSC was carried out with a Discovery DSC (TA Instruments, New Castle, DE, USA) equipped with a refrigerating device that was suitable for direct heat capacity measurements. Tem- perature scale and heat flow were calibrated with indium reference discs. The sample masses of 3 mg (± 0.3 mg) were placed in Tzero aluminum pans (TA instruments). An empty pan of the same type was used as a reference. The heating rate was set at 10o C min-1 in the range of 25-300 oC. Analysis of the results was carried out with Trios discovery evaluation software version v4.0.2.30774 (TA Instruments). Each analysis was done in triplicate for the different types of samples where after the obtained data was averaged. 2.1 Karl Fischer titration Moisture may have adhered to the drug substance during storage, shipment and measurement. To quantify this and to be to interpret and correct the TG and DSC data, Karl Fischer measurements were carried out using a Metrohm 758 KFD Titrino (Herisau, Switzerland). The samples were dissolved in methanol and titrated with a titrant (Hy- dranal, composite 2, Honeywell, Seelze, Germany). The end-point of the titration was determined biamperometrically using a 915 KF Ti-Touch titration apparatus (Metrohm, Herisau, Switzerland). Measurements were performed in triplicate. 98 | Chapter 2.1

Results and discussion

X-ray diffraction analysis (XRD) The commercial tablet formulation of pazopanib hydrochloride, Votrient®, contains crystal form I [33]. An XRD spectrum of this polymorph is not found in literature, but is presented in a patent text [34]. Table 1 lists the principal peaks of the Form I from this patent text. Figure 2 shows the recorded XRD spectrum of pazopanib hydrochloride (this study). The spectra of the pazopanib hydrochloride samples that were used in this study were found to correspond to the peaks mentioned in the patent text of Form I as shown in table 1.

Fourier Transform Infrared (FTIR) spectroscopy The core structure of pazopanib consists of heterocyclic systems with amine bridges. The only homocyclic ring is attached to a sulfonamide group. These structures provide a variety of resonance frequencies. This makes it possible to identify 8 regions with FT- IR as inherent characteristics of pazopanib hydrochloride. Table 2 lists the characteristic absorption bands and the corresponding vibrations and figure 3 presents the FT-IR spectrum of pazopanib hydrochloride bulk drug. Figure 3 also shows the FT-IR spectrum of pazopanib free base. The fingerprint region of the spectrum, 1700-600 cm-1, shows similar peaks as the hydrochloride salt. In the region between 3100 and 2300 cm-1, more absorbance is seen in the spectrum of the hydrochloride salt. Additionally, the free base lacks a twin set of peaks at approximately 3430 cm-1. These peaks are indicative of the N-H stretching, in this case following protonation by hydrochloride. The asymmetric N-H stretching peaks between 3400 and 3200 cm-1 also shift and change intensity moving from free base to hydrochloride salt. These spectral differences are important to notice; it allows a relatively quick evaluation of the salt species in the bulk substance. Thermal study of pazopanib hydrochloride | 99

Table 1. XRD data obtained from the patent text [34] and study measurements Patent data (°2θ) Obtained data (°2θ) 10.2 10.209 10.5 10.510 11.4 11.429 13.5 13.484 14.2 14.320 15.3 15.306 16.1 16.108 16.5 16.509 2.1 17.7 17.712 19.1 19.099 19.5 19.500 20.1 20.102 21.0 21.074 21.9 21.916 23.3 23.293 23.8 23.778 24.0 24.029 25.0 25.015 25.7 25.700 26.1 25.981 26.9 26.904 27.6 27.589 27.9 27.906 28.6 28.608 28.9 28.909 30.3 30.295 32.5 32.502 33.2 33.204 34.2 34.223 100 | Chapter 2.1 Intensity (counts)

10 20 30 Angle (o2q) Figure 2. XRD spectrum of pazopanib hydrochloride

Table 2. FT-IR signals of pazopanib hydrochloride Wavenumber (cm-1) Proposed assignment

3400-3300 N-H stretch (-NH2)

1640-1620 N-H (-NH2) scissoring 1500-1480 N-H bend (-NH-) 1380-1360 S=O stretch 1340-1300 C-N stretch (aromatic) 1160-1150 S=O stretch 1140-1120 C-N stretch (-NH-) 800-780 N-H wag

Thermogravimetric analysis (TG) The first loss of mass during heating occurs at 40 oC with a mass reduction of 3.7% as shown in figure 4D. This corresponds to the water mass percentage found with Karl Fischer titration (3.71%). Dried pazopanib hydrochloride (5 min. at 120 oC) does not experience this loss of mass, figure 4D. For the thermal analysis of the decomposition of pazopanib hydrochloride, dried bulk drug was used. The TG curves of pazopanib hydrochloride from different heating rates are presented in figure 4A. Figure 4B shows the DTG curve at a heating rate of 40 oC min-1. The first mass losses of the dried drug take place between 200 oC and 305 oC (Δm = 7.4%). This can be attributed stoichiometrically to the loss of hydrochloride. Figure 4C shows the partial TG curve of pazopanib free base in which the first mass loss does not occur. The loss of hydrochloride seems to happen Thermal study of pazopanib hydrochloride | 101

in two phases. The two phases may correspond to the loss of hydrochloride from the surface and the interior of the powder aggregates.

Dried %)

e( Bulk nc ta mit s

an Free base 2.1 Tr

Free base %) e( nc ta smit an

Tr HCl

3000 2800 2600 2400 Wavenumber (cm-1)

3800 3600 3400 3200 3000 1500 1000 Wavenumber (cm-1) Wavenumber (cm-1)

Figure 3. FTIR spectra of pazopanib-free base and hydrochloride: bulk drug and dried drug Insert: Spectral differences between hydrochloride and free base drug

A plateau separates the second from the third mass loss, which occurs between 340 oC and 375 oC (Δm = 18.0 %). The third mass loss overlaps with the fourth (375 oC- 430 oC, Δm = +/- 13.1 %), which in turn overlaps with the fifth (430 oC- 550 oC, Δm = +/- 16.0 %). The latter three mass losses are structural degradation events of pazopanib’s core structure. The mass loss of 18.0 % may correspond to the thermolysis of the sulfonamide group (Mw = 80 Da (18.2%)).

The latter two degradation steps more strongly overlap. This makes interpretation and description of the mechanism of thermolysis in these regions difficult. To elucidate what takes place in temperatures above 375 oC, analysis of the gaseous degradation products by spectroscopy might prove useful.

The thermal decomposition was studied at various heating rates; figure 4A. Increasing the heating rate shifted the decompositions to higher temperatures. The DTG curve in figure 4B reveals 5 distinct peaks. The thermal decomposition of pazopanib hydrochloride was analyzed accordingly in 5 phases. The degradation speed above 550 oC did not exceed 0.05 mg min-1 and was not included in the analyses. 102 | Chapter 2.1 )

-1 1.5 B

in III gm

A (m 1.0 100 ed pe ns

itio 0.5 IV II V ) I 0.0 Decompos

/w 0 500 80 Temperature (oC) ,w 102 C ) (% /w

,w Free base -1 100 40 Kmin (% 20 Kmin-1 HCl -1 entage 98 entage 60 10 Kmin 5 Kmin-1 perc -1 96 perc 2.5 Kmin Mass

200 220 240 260 280 Temperature (oC) Mass 102 D 40 ) /w

,w Dried 100 I II III IV V (%

entage 98 200 400 600 800 Bulk o perc Temperature ( C) 96 Mass

20 40 60 80 100 Temperature (oC) Figure 4. Thermogravimetric results: A. Mass-loss curves of hydrochloride salt from different heating rates, where the dotted lines indicate the start of the assigned degradation phase; B. DTG curve for the heating rate of 40 oC min-1 , numerals indicate DTG-peaks of the assigned degradation phases; C. Curves from dried hydrochloride salt and free base at a heating rate of 40o C min-1; D. Curves from bulk drug and dried drug at a heating rate of 40 oC min-1.

The α-T curves for the non-isothermal decomposition at different heating rates are shown in figure 5A. Figure 5B shows the lnβ-1/T curves. By fitting these plots using the

OZW method, the activation energies Ea and pre-exponential factors A were calculated (table 3). The activation energies of the various degradation steps are constant within their degradation phase and increase with the individual phases as shown in figure 6. The activation energies were also calculated using the F and KAS methods from ln(β(da/ dt))-1/T and ln(β/T2)-1/T curves and are presented in table 3 and figure 6. The consistency of the Ea values in each phase suggests that the degradation in those phases takes place in one step. A variation is observed between E values from the three methods. This may be due to the different temperature integral approximations of the methods. For the calculation of A, the degradation was assumed to behave according to a R2 (phase boundary controlled reaction) kinetic reaction model. Thermal study of pazopanib hydrochloride | 103

A 4 B 4 Phase I (HCl) Phase II (HCl) ) ) 3 3 -1 -1 in in Cm Cm 2

2 o o

0.6 n n (l (l b b ln ln 1 1

0 0 0.00180 0.00190 0.00200 0.00210 0.00220 0.00160 0.00170 0.00180 0.00190 0.00200 1/T (K-1) 1/T (K-1)

0.4 4 4 Phase III Phase IV ) ) 3 3 -1 -1 a in in Cm Cm 2 2 o o n n (l o -1 (l b 40 C min b ln ln 1 1 20 oC min-1 o -1 10 C min 0 0 0.2 5 oC min-1 0.00155 0.00160 0.00165 0.00170 0.00140 0.00150 0.00160 1/T (K-1) 1/T (K-1) 2.1 2.5 oC min-1 4 Phase V

) 3 -1 in

Cm 2 o n (l b 0.0 ln 1

200 400 600 800 0 o 0.00130 0.00140 0.00150 Temperature ( C) 1/T (K-1)

Figure 5. Results from non-isothermal analysis methods: A. α-T curves for the non-isothermal decomposition; B. lnβ-1/T curves.

300 150 OZW 140 130 F 250 120

l) 110 KAS

100 mo 200 1.00 0.98 0.96 0.94 0.92 (kJ/ a E 150

100 0.8 0.6 a

Figure 6. Activation energy (Ea) as a function of the conversion (α) as calculated through the OZW, F and KAS methods. The dotted lines separate the five phases of degradation.

The values of the thermodynamic parameters ΔS≠, ΔH≠ and ΔG≠ for the activated complex (transition state) were calculated using the calculated A and the averaged activation energy obtained from the isoconversional methods. 104 | Chapter 2.1

Thermal degradation can additionally be studied by using isothermal methods. In this case such an approach is not feasible in this particular case because of the multiple degradation phases that may interfere with one-another’s signals during isothermal heating.

Table 3. Thermodynamic parameters (averaged) of the thermal decomposition of pazopanib hydrochloride Temperature range (oC) DTG (oC) E A (s-1) ΔS≠ ΔH≠ ΔG≠ (kJ.mol-1) (kJ.mol-1) (J.mol-1.K-1) (kJ.mol-1) 200-250 (I) 230 Ozawa-Flynn-Wall 119.6 9.4*1010 -47.5 115 134 Friedman 116.9 Kissinger-Akahira-Sunose 116.9 250-305 (II) 280 Ozawa-Flynn-Wall 130.9 1.6*1011 -44.1 126 151 Friedman 129.2 Kissinger-Akahira-Sunose 129.9 340-375 (III) 360 Ozawa-Flynn-Wall 167.1 6.9*1012 -13.6 165 173 Friedman 170.7 Kissinger-Akahira-Sunose 170.3 375-430 (IV) 395 Ozawa-Flynn-Wall 192.1 5.2*1014 21.8 186 172 Friedman 192.5 Kissinger-Akahira-Sunose 191.5 430-550 (V) 460 Ozawa-Flynn-Wall 223.0 6.3*1015 40.5 228 198 Friedman 235.4 Kissinger-Akahira-Sunose 233.6

Differential Scanning Calorimetry (DCS) The DSC results are shown in figure 7. The bulk drug curve is the result of the DSC ramp measurement with the pure drug substance. A broad endothermic peak is observed between 48 and 125 oC followed by an exothermic peak at 238 oC where after sharp endothermic peaks begin at 280 oC. The first endothermic peak was suspected to be an evaporation event that was not related to the drug structure. To investigate this, new samples were subjected to a 5 min. period of preheating at 120o C. A FT-IR spectrum was taken and a new ramp was recorded thereafter which resembled the spectrum of the dried drug. The first endothermic peak is absent while the rest of the curve is similar to that of the bulk drug. In combination with the TG and the Karl Fisher titrations it can be concluded that this peak corresponds to loss of water and that the drug substance may be hygroscopic. The exothermic peak corresponds to the loss of HCl which is also seen in the TG results in that particular temperature range. A similar peak was seen during Thermal study of pazopanib hydrochloride | 105

the HCl loss of Terazosin during thermal analysis [35]. The sharp endothermic peaks after 280 oC are clear signs of structural degradation [36]. It is interesting to note that no melting peak is observed, this supports the observation in melting point determinations where pazopanib hydrochloride shows no liquid phase before degradation and the lack of a reported value in literature.

Dried 2.1

Bulk drug (W/g ) flow Heat

≠Exo

100 Temperature (oC) 200 300

Figure 7. DSC results for the pazopanib hydrochloride bulk drug and the dried pazopanib hydrochloride 106 | Chapter 2.1

Conclusions

This study describes the thermal analysis of pazopanib hydrochloride. XRD showed that the pazopanib hydrochloride that was used in this study was in polymorph Form I, the same polymorph that is used in the commercial formulation. Thermal characterization revealed adherence of water and the initiation of degradation at 200 oC and step wise solid phase degradation. By using three different model-free isoconversional methods for the analysis of degradation at multiple heating rates the activation energy and- ther modynamic parameters of the degradation were calculated. The activation energy values for the five degradation phases were found to be relatively constant within each phase and differ between phases, supporting the five-stage degradation conclusion. Calorimetry showed agreement with thermogravimetric results in the degradation of the hydrochloride and demonstrates the lack of a melting point of pazopanib hydrochloride. In conclusion, pazopanib hydrochloride is a suitable candidate for new formulation studies were thermal production methods are utilized such as dispersion through hot- melt extrusion and spray-drying. This could possibly yield formulations with an improved bioavailability. Thermal study of pazopanib hydrochloride | 107

References

[1] US Food and Drug administration (FDA). Highlights of Polym. Symp. 2007;6:183–95. prescribing information Votrient. 2012; [18] Kissinger HE. Reaction Kinetics in Differential Thermal [2] European Medicines Agency (EMA). Summary of Analysis. Anal. Chem. 1957;29:1702–6. product characteristics Votrient. 2016; [19] Vyazovkin S, Wight CA. Kinetics in solids. Annu. Rev. [3] US Food and Drug administration (FDA). Clinical Phar- Phys. Chem. 1997;48:125–49. macology and Biopharmaceutics Review Votrient. [20] Vyazovkin S, Burnham AK, Criado JM, Pérez-Maqueda 2009; LA, Popescu C, Sbirrazzuoli N. ICTAC Kinetics Commit- [4] Deng Y, Sychterz C, Suttle AB, Dar MM, Bershas D, tee recommendations for performing kinetic compu- Negash K, et al. Bioavailability, metabolism and dis- tations on thermal analysis data. Thermochim. Acta position of oral pazopanib in patients with advanced 2011;520:1–19. cancer. Xenobiotica. 2013;43:443–53. [21] Atanassov A, Genieva S, Vlaev L. Study on the Ther- [5] Herbrink M, Nuijen B, Schellens JHM, Beijnen JH. Vari- mooxidative Degradation Kinetics of Tetrafluoroeth- 2.1 ability in bioavailability of small molecular tyrosine ki- ylene-Ethylene Copolymer Filled with Rice Husks Ash. nase inhibitors. Cancer Treat. Rev. 2015;41:412–22. Polym. Plast. Technol. Eng. 2010;49:541–54. [6] US Food and Drug administration (FDA). Chemistry [22] Fandaruff C, Araya-Sibaja AM, Pereira RN, Hoffmeis- Review Votrient. 2009; ter CRD, Rocha HVA, Silva MAS. Thermal behavior and [7] US Pharmacopeial Convention. Priority New Mono- decomposition kinetics of efavirenz under isothermal graph List. 2016; and non-isothermal conditions. J. Therm. Anal. Calo- [8] Yu LX, Amidon G, Khan MA, Hoag SW, Polli J, Raju GK, rim. 2014;115:2351–6. et al. Understanding pharmaceutical quality by de- [23] Jankovic B, Kinetic analysis of the nonisothermal sign. AAPS J. 2014 [cited 2014 Sep 18];16:771–83. decomposition of potassium metabisulfite using the [9] Giron D. Contribution of thermal methods and relat- model-fitting and isoconversional (model-free) meth- ed techniques to the rational development of phar- ods. Chem. Eng. J. 2008;139:128–35. maceuticals—Part 1. Pharm. Sci. Technolo. Today. [24] Kissinger HE. Variation of peak temperature with 1998;1:191–9. heating rate in differential thermal analysis. J. Res. [10] Salama NN, Mohammad MA, Fattah TA. Thermal Natl. Bur. Stand. 1957;57:217–21. behavior study and decomposition kinetics of ami- [25] Alshareef AH, Azyat K, Tykwinski RR, Gray MR. Mea- sulpride under non-isothermal and isothermal condi- surement of Cracking Kinetics of Pure Model Com- tions. J. Therm. Anal. Calorim. 2015;120:953–8. pounds by Thermogravimetric Analysis. Energy & [11] Perpétuo GL, Gálico DA, Fugita RA, Castro RAE, Eu- Fuels 2010;24:3998–4004. sébio MES, Treu-Filho O, et al. Thermal behavior [26] Muraleedharan K, Kannan MP, Ganga Devi T. Ther- of some antihistamines. J. Therm. Anal. Calorim. mal decomposition kinetics of potassium iodate. J. 2013;111:2019–28. Therm. Anal. Calorim. 2011;103:943–55. [12] Tita B, Jurca T, Tita D. Thermal stability of pentoxi- [27] Mothé CG, de Miranda IC. Study of kinetic parame- fylline: active substance and tablets. J. Therm. Anal. ters of thermal decomposition of bagasse and sugar- Calorim. 2013;113:291–9. cane straw using Friedman and Ozawa–Flynn–Wall [13] Howell BA. Utility of kinetic analysis in the determina- isoconversional methods. J. Therm. Anal. Calorim. tion of reaction mechanism. J. Therm. Anal. Calorim. 2013;113:497–505. 2006;85:165–7. [28] Flynn JH. The “Temperature Integral” — Its use and [14] Doca SC, Albu P, Ceban I, Anghel A, Vlase G, Vlase abuse. Thermochim. Acta 1997;300:83–92. T. Sodium alendronate used in bone treatment. J. [29] Senum GI, Yang RT. Rational approximations of the Therm. Anal. Calorim. 2016;126:189–94. integral of the Arrhenius function. J. Therm. Anal. Cal- [15] Ledeţi I, Vlase G, Vlase T, Fuliaş A. Kinetic analysis of orim. 1977;11:446. solid-state degradation of pure pravastatin versus [30] Vlaev L, Nedelchev N, Gyurova K, Zagorcheva M. A pharmaceutical formulation. J. Therm. Anal. Calorim. comparative study of non-isothermal kinetics of de- 2015;121:1103–10. composition of calcium oxalate monohydrate. J. Anal. [16] Flynn J, Wall L. Thermal Methods in Polymer Analysis. Appl. Pyrolysis 2008;81:253–62. Phys. Chem. 1966;70:487–523. [31] Boonchom B, Thongkam M. Kinetics and Thermody- [17] Friedman HL. Kinetics of thermal degradation of namics of the Formation of MnFeP 4 O 12. J. Chem. char-forming plastics from thermogravimetry. Ap- Eng. Data 2010;55:211–6. plication to a phenolic plastic. J. Polym. Sci. PartC [32] Georgieva V, Zvezdova D, Vlaev L. Non-isothermal 108 | Chapter 2.1

kinetics of thermal degradation of chitosan. Chem. [35] Attia A, Abdel-Moety M. Thermoanalytical Investi- Cent. J. 2012;6:1–10. gation of Terazosin Hydrochloride. Adv. Pharm. Bull. [33] European Medicines Agency (EMA). Assessment re- 2013;3:147–52. port Votrient (EPAR). 2011; [36] Bibi S, Bremner DH, Macdougall-Heasman M, Reid R, [34] Chava S, Gorantla S, Indukuri V. An improved process Simpson K, Tough A, et al. A preliminary investigation for the preparation of pazopanib or a pharmaceuti- to group disparate batches of licit and illicit diazepam cally acceptable salt thereof. Laurus Labs Private Ltd; tablets using differential scanning calorimetry. Anal. 2014. Methods 2015;7:8597–604.

Chapter 2.2

Thermal stability study of crystalline and novel spray-dried amorphous nilotinib hydrochloride

M. Herbrink1, H. Vromans2, J.H.M. Schellens2,3, J.H. Beijnen1,3, B. Nuijen1

Abstract

The thermal characteristics and the thermal degradation of crystalline and amorphous nilotinib hydrochloride (NH) were studied. The spray drying technique was successfully utilized for the amorphization of NH and was evaluated by spectroscopic techniques and differential scanning calorimetry (DSC). The ethanolic spray drying process yielded amor- o phous NH with a glass transition temperature (Tg) of 147 C. Thermal characterization of the amorphous phase was performed by heat capacity measurements using modulated DSC (mDSC). Thermal degradation was studied by thermogravimetric analysis (TGA). The derived thermodynamic properties of the amorphous NH indicate fragile behaviour and a low crystallization tendency. NH was found to be molecularly stable up to 193o C. After which, the thermal degradation displayed two phases. The values of the thermal degradation parameters were estimated using the Ozawa-Flynn-Wall and Friedman non-isothermal, model-free, isoconversional methods. The results indicate the two phases to be single-step reactions. The examination of the physical stability of amorphous NH during storage and at elevated temperatures showed stability at 180 oC for at least 5 hours and at 20-25 oC/60% RH for at least 6 months. During these periods, no crystallization was observed. This study is the first to report the thermal characteristics of NH. Additionally, it is also the first to describe the full thermal analysis of a spray-dried amorphous drug. The thermal data may be used in the projection of future production processes and storage conditions of amorphous NH. Thermal stability study of crystalline and novel spray-dried amorphous nilotinib hydrochloride | 113

Introduction

The tyrosine kinase inhibitor nilotinib hydrochloride (NH) (figure 1) is registered for the treatment of newly diagnosed adults with Philadelphia chromosome positive myeloid leukemia (Ph+ CML) in the chronic phase. It is also indicated for the treatment of chronic and accelerated phase Ph+ CML in adult patients that are resistant or intolerant to prior therapy that included imatinib [1].

N 2.2 O F NH N NH F F N

Figure 1. Chemical structure of nilotinib (C28H22F3N7O; Mw 529.52 Da)

The oral dosage form of NH (Tasigna®) contains the crystalline polymorph B and is associated with poor solubility and permeability that hinders its bioavailability [2]. Sol- ubility and bioavailability improvement of drugs may be achieved by formulating the amorphous form. Although the amorphous form is often less stable than the crystalline polymorphs, it tends to exhibit improved dissolution characteristics [3].

A nanoparticle formulation containing amorphous NH have been shown to markedly increase the solubility of the drug [4]. Furthermore, another study with this formulation demonstrated an increase in the human bioavailability and shows the advantages of utilizing the amorphous form [5].

Amorphous drug can be manufactured from crystalline material by usage of a large variety of techniques. The spray drying technique was chosen for this study because it 114 | Chapter 2.2

enables relatively short production times, large- and small-scale productions and due to its setting flexibility [6]. Additionally, spray drying has successfully been applied in the development and production of various formulations that contain amorphous drugs [7]. The use of the spray-drying technique for the production of amorphous NH has not been reported previously in literature.

The application of thermal analysis methods is important in the pharmaceutical industry, especially in cases where amorphous forms are involved [8]. The methods can be used in studies of polymorph stability, crystallization, stability, compatibility and kinetic parameters. The thermal properties may determine the choice of production methods, storage conditions, bioavailability and by extension, therapeutic efficiency [9]. Additionally,- for mulation development that follows Quality by Design (QbD) principles should include a full assessment of drug characteristics [10]. This approach will allow for critical drug properties to be identified timely.

For these reasons, it is of importance to study the thermal characteristics and thermal stability of the amorphous form and the crystalline polymorph from which it is produced. No information is available in literature about the thermal behavior of either crystalline or amorphous NH. Furthermore, a full thermal and thermodynamic analysis of a spray-dried amorphous drug has also not been reported before. This study is the first to perform these analyses and establish these characteristics.

The spray-drying process was evaluated by 1H- and 13C-Nuclear Magnetic Resonance spectroscopy (NMR), Fourier transform infrared spectroscopy (FTIR), powder X-ray diffraction (XRD) and differential scanning calorimetry (DSC). The thermal characteristics of the amorphous state of NH were assessed using modulated temperature DSC (MTDSC). The thermal stability and degradation of NH were studied with thermogravimetric analysis and were evaluated by the non-isothermal, isoconversional Ozawa-Flynn-Wall (OFW) and Friedman (F) methods as is recommended by the International Confederation for Thermal Analysis and Calorimetry [11].

The results from this study may be used in the quality control of both crystalline and amorphous and may play a role in the design of production processes and the choice of storage conditions. Thermal stability study of crystalline and novel spray-dried amorphous nilotinib hydrochloride | 115

Materials and methods

Materials NH monohydrate was purchased from Avachem Scientific (San Antonio, TX, USA) and was supplied with a certificate of analysis. All analyses were compared to a reference standard to exclude any batch-to-batch differences. The reference standard of NH (monohydrate) was acquired from AlsaChim (Illkirch, France). All other chemicals were of analytical grade. Residual solvents (<0.5%) were detected by gas chromatography. Moisture content was determined by Karl Fisher titration.

Amorphous phase preparation The amorphous phase of the compound was prepared by the spray drying technique using a Büchi MiniSpray Dryer B-290, Inert Loop B-295, High performance cyclone, 1.5 2.2 mm nozzle cap and 0.7 mm nozzle tip (Büchi). Spray dry system settings were: Spray feed 20%; N2 atomization flow 40 mm; aspirator flow 100%; inlet temperature 120 oC; outlet temperature 75 oC; inert loop temperature -20oC. NH monohydrate (10 g/L) was dissolved in 100% ethanol and the solution was stirred using a magnetic stirrer at 20-25 oC.

1H-, 13C-Nuclear magnetic resonance spectroscopy (NMR) NMR spectra were recorded with an Avance Ultrashield instrument (Bruker Corporation, Billerica, MA, USA). The 1H-NMR and 13C-NMR spectra were recorded at 300 MHz and 75 MHZ, respectively. The samples were dissolved in DMSO (highest peak in multiplet at 2.50 ppm). In the 13C-spectrum, the center peak of the DMSO signal was at 38.5 ppm.

Fourier transform infrared spectroscopy (FTIR) FT-IR spectra were recorded from 650 to 4000 cm-1 with a resolution of 2 cm-1 with a FT-IR 8400S Spectrophotometer equipped with a golden gate® (Shimadzu, ‘s-Hertogen- bosch, the Netherlands). A total of 64 scans were averaged into one spectrum. Data analysis was performed with IR Solution software V1.4 (Shimadzu).

Powder X-ray diffraction (XRD) X-ray diffraction of powder samples was performed with an X’pert pro diffractometer equipped with an X-celerator (PANanalytical, Almelo, The Netherlands). Samples were placed in a 0.5 mm deep metal sample holder. Samples were scanned at a current of 30 mA and a tension of 40 kV. The scanning grange was 10 – 60 degrees 2-θ, with a step size 116 | Chapter 2.2

of 0.020 degrees 2-θ and a scanning speed of 0.002 degrees 2-θ per second. The spectra were processed using HighScore software v4.5.

(modulated) differential scanning calorimetry ((m)DSC) mDSC and DSC measurements were performed with a Discovery DSC (TA Instruments) equipped with a refrigerating device and suitable for direct heat capacity measurements. Temperature scale and heat flow were calibrated with indium reference disks. Drug samples of approximately 3-5 mg were weighed in Tzero aluminium pans (TA Instruments), compacted, sealed and placed in the autosampler. Each sample was equilibrated at 20 oC, after which the samples were heated with a speed of 10o C / min to the previous with TGA determined safe temperature. An empty sample pan was weighed and used as reference to the heat capacity of the pans.

The mDSC instrument settings for the heat capacity measurements were as follows: a modulation period of 100 s, a modulation amplitude of +/- 0.5 oK and an underlying o heat rate of 1 K / min. The heat capacity constant (KCp) was calibrated using a sapphire disc weighing approximately 25 mg. Heat capacity measurements for the crystalline and amorphous compound were performed over a temperature range of 25 oC (absolute) to 20 oC above the melting point of NH. Analysis of the (m)DSC results was carried out with the Trios discovery evaluation software version v4.0.2.30774 (TA instruments).

Thermogravimetric analysis Thermogravimetric analysis (TGA) was performed on a Q50 thermogravimetric analyzer (TA Instruments, New Castle, DE, USA). The weight change is measured as a function of time and/or temperature. A typical sample (+/- 10 mg) was placed on a platina sample holder and heated from room temperature to 1000 oC at heating rates of 2.5, 5, 10, 20 and 40 oK / min. Advantage software (TA Instruments) v5.5.22 was used for the analysis of the TGA data.

Theory and calculations Thermodynamics of the amorphous state

The thermodynamic fragility FT of amorphous NH was evaluated from the heat capacity change at the glass transition temperature (Tg) using equation 1 [12]:

&'() (1) $% 𝐹𝐹" = ∆$% Thermal stability study of crystalline and novel spray-dried amorphous nilotinib hydrochloride | 117

conf Where Cp is the heat capacity difference between the crystalline and the amorphous o drug (J/mol/ K), the configurational heat capacity andΔC p is the change in heat capacity of the amorphous drug at Tg. The Kauzmann temperature (Tk) was calculated using the different equations 2a, 2b and 2c:

) (2a) %&'( 𝑇𝑇"# ≈ &* (2b)

𝑇𝑇"# = 𝑇𝑇% − 50 (2c) ! ! *+& "#$ = "& '1 + , - The K-value (J/mol) in equation 2c was determined using the relationship expressed in equation 3: 2.2

(3) %&'( 𝐾𝐾 = 𝐶𝐶$ ∗ 𝑇𝑇 conf Temperature-independency of K was verified by plotting Cp *T versus T. The dynamic fragility of amorphous NH was analysed by two different methods. The first method utilizes the Vogel-Tammann-Fulcher relation to calculate the strength parameter D1 (-) through equation 4:

(4) ()*+ 𝛤𝛤 = 𝛤𝛤#𝑒𝑒𝑒𝑒𝑒𝑒 '*,*+-

Where Γ is the molecular relaxation time (s),Γ 0 is the vibrational lifetime of approximate- -14 ly 10 seconds (s) and T0 is the constant Vogel temperature. Using this method assumes that Tk from equation 2c andT o have roughly the same value and also the value of Γ at Tg is approximately 100 s.

The strength parameter was subsequently used to calculate the fragility parameter m1 (-) using equation 5 [13]:

(5) $%&' /' 0 " ( 𝑚𝑚 = & (*+",)("./()

The second method uses the different values of Tg that are measured at various heating rates β (oK/min). Through the linear relationship between the inverse values of the measured Tg’s and the natural logarithm of the heating rate, the glass transition activation energy,ΔE Tg (kJ/mol) can be calculated using equation 6 [13]: 118 | Chapter 2.2

(6) ∆#$% ( )* + − & = (,-//%0 o Where R is the universal gas constant (J/ K/mol). The fragility parameter, m2, and the strength parameter, D2, can then be determined by using equations 7 and 8:

(7) ∆%&' 𝑚𝑚" = () *+∗-∗.' (8) , ".%&%'()*+- 𝐷𝐷" = (,.()*+

Where mmin is the logarithm of the ratio of the mean relaxation time at Tg and the vibrational lifetime for which the value is 16 [13]. These fragility parameters allow the calculation of the Vogel temperature using equation 9:

(9) ()*+ 𝑇𝑇" = 𝑇𝑇$ %1 − (, -

T0 was reintroduced in equation 4 to calculate relaxation times at different temperatures. o The enthalpic driving force Hc (kJ/mol) and the configurational entropy Sc (J/mol/ K), were calculated with equations 10 and 11:

(10) / "*+, 𝐻𝐻" = 𝛥𝛥𝛥𝛥% + ∫/0 𝐶𝐶) 𝑑𝑑𝑑𝑑 *+,- (11) . () 𝑆𝑆" = 𝛥𝛥𝛥𝛥% + ∫.1 . 𝑑𝑑𝑑𝑑 where:

(12) 𝛥𝛥𝛥𝛥# 𝛥𝛥𝛥𝛥# = 𝑇𝑇#

ΔHm and ΔSm are the enthalpy and entropy of melting as determined by DSC (J/mol and o o o J/mol/ K), T is temperature ( K), Tm is melting temperature ( K) [14]. The overall force for crystallisation (Gibbs free energy) Gc (kJ/mol) was derived from equation 13:

(13)

𝐺𝐺" = 𝐻𝐻" − (𝑆𝑆" ∗ 𝑇𝑇) Thermal decomposition parameters The results from the TGA analyses at different heating rates were evaluated with both Thermal stability study of crystalline and novel spray-dried amorphous nilotinib hydrochloride | 119

the Ozawa-Flynn-Wall (OFW) and Friedman (F) isoconversional methods. These kinetic methods propose that the isothermal rate of conversion, dα/dt, is a linear function of the temperature, the rate constant, k(T), and a temperature-independent conversion function, f(α), equation 14:

(14) !" ( -.//12 #(") = ')* ∗ 𝑒𝑒 𝑑𝑑𝑑𝑑 o Where β is the heating rate in K/min, A, the pre-exponential factor andE a the activation energy [11]. The OFW and the F methods can be utilized to calculateE a. The methods are based on the assumption that for every value ofα , f(α) is not dependent on the heating rate. It is then a requirement to measure α at different heating rates. Equation 14 under these conditions can be rewritten as equation 15 for OFW: 2.2 (15) %&' &' ln 𝛽𝛽 = ln ( − 5.331 − 1.052 (0 and as equation 16 for F [15]:

(16) %& ./ ln(𝛽𝛽 %') = ln 𝐴𝐴𝐴𝐴(𝛼𝛼) − 0' When plots are constructed of lnβ and ln(βdα/dt) against 1/T a linear relationship should be found. The slope of the linearity can then be used to calculate Ea and A. The estimation of the pre-exponential factor was carried out in the assumption that the kinetics of the decomposition process can be described by the phase-boundary model (R2), with kinetic exponent n=2 and a four degree ‘rational approximation’ [16]. The change of entropy, H≠ (kJ/mol), enthalpy, S≠ (J/mol/oK), and Gibbs free energy, G≠ (kJ/ mol), for the formation of the activated complex were calculated using equations 17-19:

(17) # () 𝛥𝛥𝛥𝛥 = 𝑅𝑅 ln *+,-. (18) # 𝛥𝛥𝛥𝛥 = 𝐸𝐸 − 𝑅𝑅𝑇𝑇) (19) # 𝛥𝛥𝛥𝛥 = 𝛥𝛥𝛥𝛥 − 𝑇𝑇(𝛥𝛥𝛥𝛥 Where h is the Planck’s constant (J·s), e is the Neper (Euler’s) number (-), χ is the transi- -1 tion factor (-), k is the Boltzmann constant (J·K ) and Tp is the peak temperature of the DTG curve (oK). 120 | Chapter 2.2

Results and discussion

Nuclear Magnetic Resonance Spectroscopy (NMR) 1H- and 13C-NMR spectra were recorded of NH monohydrate, before and after spray drying. Table 1 presents the proposed assignments of the NMR-signals for both spectra. The NMR-spectra showed no significant change after spray drying, suggesting conservation of the NH molecular structure.

Table 1. Spectroscopy results for NH Analytical Results Structure numbering method

1 H-NMR 2.37 (m, 6H, C15-CH3 & C7-CH3); 7.47 (s, 1H, H14);

7.44 (s, 1H, H13); 7.50 (s, 1H, H19); 7.93 (s, 1H,

NH); 7.87 (s, 1H, H23); 8.36 (s, 2H, H1 & H3); 8.01 (s, 1H, H ); 8.56 (s, 1H, H ); 8.57 (s, 1H, H ); 11 22 16 F F 8.65 (s, 1H, H ); 8.66 (s, 1H, H ); 9.21 (s, 1H, 5 24 F H8); 9.30 (s, 1H, H9); 9.60 (s, 1H, H20); 10.99 (s, 8 3 4 1H, NH) N N 2 5 13C-NMR 9.91 (C -CH ); 18.16 (C -CH ); 107.921 (Ar-C ); 7 7 3 15 3 1 9 1 6 a 113.54 (Ar-C3); 116.80 (Ar-C11); 116.88 (Ar-C9); NH 117.18 (Ar-C5 & Ar-C19); 121.50 (Ar-C15); 123.64 25 (Ar-C12); 124.01 (Ar-C7); 124.34 (Ar-C13); 125.11 O

(Ar-C14); 130.35 (Ar-C4); 130.65 (C4-CF3); 131.30 12 13 (Ar-C ); 131.41 (Ar-C ); 132.28 (Ar-C ); 22 21 23 11 14 134.91 (Ar-C21); 136.01 (Ar-C8); 136.64 (Ar-C6); 10 15 138.06 (Ar-C10); 141.53 (Ar-C2); 147.38 (Ar-C24); ß 150.64 (Ar-C20); 159.51 (Ar-C18); 160.90 (Ar- HN

C16); 161.26 (Ar-C17); 165.69 (C25ONH) 17 N 22 23 18 21 24 FT-IR 3300-3200 cm-1: N-H stretch & water O-H N stretch; 3100-3050 cm-1: amide overtone II; 16 19 20 N 2900-2800 cm-1: 2400-2200 cm-1: water combination stretch; 1620-1590 cm-1: pyridine stretch; 1600-1570 cm-1: C=O stretch; 1540- 1500: C-N stretch; 1500-1480 cm-1: N-H bend (-NH-); 1120-1080 cm-1: C-F stretch a; all peaks with relative abundances over 10% are listed.

Fourier transform infrared spectroscopy (FTIR) FTIR spectra were recorded of NH monohydrate, NH and spray-dried NH. The spectra are presented in figure 2a The obtained results for all drug forms show similarity in the characteristic bands as listed in table 1. The similarity in the spectra of NH monohydrate and NH is indicative of resembling molecular interaction between the drug molecules with and without the presence of water of crystallization. Thermal stability study of crystalline and novel spray-dried amorphous nilotinib hydrochloride | 121

Fourier transform infrared spectroscopy (FTIR) FTIR spectra were recorded of NH monohydrate, NH and spray-dried NH. The spectra are presentedThe spectrum in figure of the 2a spray-dried The obtained NH resultsdisplays for a blueall drug shift forms and anshow increase similarity in transmit in the- chartanceacteristic for the bands bands as corresponding listed in table 1.to The the similarity C-F stretch in the(1085, spectra 1118, of 1228,NH monohydrate 1261 and 1292 and NH is indicative of resembling molecular interaction between the drug molecules with and -1 -1 -1 withoutcm ), the the C=Opresence stretch of water (1585 of cm crystallization.) and the C-N stretch (1540 cm ) relative to the NH Thespectrum. spectrum Additionally, of the spray- dried the NH bands displays corresponding a blue shift and to an N-Hincrease stretches in transmittance (2964, 3055, for 3209 -1 thecm bands-1) demonstrate corresponding less tointensity the C-F andstretch broadening. (1085, 1118, These 1228, observations 1261 and 1292 are cm suggestive), the C=O of stretch (1585 cm-1) and the C-N stretch (1540 cm-1) relative to the NH spectrum. Additionally,amorphization the bandsof NH. corresponding Furthermore, to the N- Hspectra stretches reveal (2964, that 3055, the 3209mentioned cm-1) demonstrate bonds are of lessimportance intensity inand maintaining broadening. the These crystal observations structure ofare NH suggestive and in the of amorphizationrecrystallization of fromNH. Furthermore,the amorphous the form.spectra reveal that the mentioned bonds are of importance in maintaining the crystal structure of NH and in the recrystallization from the amorphous form.

2.2

FigureFigure 2. Characterization results: results: a). a). FTIR FTIR spectra: spectra I. crystalline: I. crystalline NH NHmonohydrate; monohydrate; II. crystalline II. crystalline NH; III. amorNH;- phous NH; b).III XRD. amorphous patterns of NH; crystalline b). XRD NH patterns and amorphous of crystalline NH; NH and amorphous NH;

Powder X-ray diffraction (XRD) Powder X-ray diffraction (XRD) FigureFigure 2 2bb presentspresents thethe XRDXRD spectraspectra ofof NH monohydrate andand spray dried material. The spectrumspectrum of of NH monohydratemonohydrate displaysdisplays a asharp sharp peak peak pattern pattern that that is characteristicis characteristic of crysof- crystallinetalline material material and and corresponds corresponds well well to theto thespectrum spectrum of Form of Form B [17]. B [17] This. crystallineThis crystalline pat- pattern is not observed in the spectrum of the spray dried material. This further indicates thetern successful is not observed amorphization in the ofspectrum NH by spray of the drying. spray dried material. This further indicates the successful amorphization of NH by spray drying. (Modulated temperature) differential scanning calorimetry ((MT)DSC) Figure 3a represents the DSC curves of NH monohydrate and NH. Both curves show an (Modulated temperature) differential scanning calorimetry ((MT)DSC)o o endothermic peak that corresponds to the melting event (Tpeak = 199.1 C; Tonset = 194.2 C). TheFigure melting 3a represents enthalpy a ndthe entropy DSC curves are listed of NH in monohydratetable 2 and were and found NH. Bothto be curves similar showfor both an forms.endothermic The DSC peak curve that of NH corresponds monohydrate to theexhibits melting a second event endothermic (T = 199.1 peak oC; (TT peak = 194.2133.4 oC; T = 97.81 oC) that corresponds to the loss of water of peakcrystallization. Theseonset results oC). onsetThe melting enthalpy and entropy are listed in table 2 and were found to be similar show that the melting event is preceded by the transformation of the monohydrate to the anhydratefor both forms. salt form, The which DSC curveis consistent of NH monohydratewith literature exhibitsdata [17] a. second endothermic peak o o (Tpeak = 133.4 C; Tonset = 97.81 C) that corresponds to the loss of water of crystallization. These results show that the melting event is preceded by the transformation of the monohydrate to the anhydrate salt form, which is consistent with literature data [17]. 94

122 | Chapter 2.2

A B NH 20 ) ) 10 /g /g (W (W NH.H2O 5 3 flow flow 1 Heat Heat

50 100 150 200 135 140 145 150 155 Temperature (oC) Temperature (oC)

Figure 3. DSC results: a). DSC curves of NH and NH monohydrate obtained at a heating rate of 10 oC/min; b). DSC curves of amorphous NH at heating rates of 1, 3, 5, 10 and 20 oC/min (indicated in figure), zoomed

in on Tg.

The DSC curves of spray-dried NH recorded at different heating rates of 1, 3, 5, 10 and 20 oK/min are presented in figure 3b. The curves show an endothermic shift character- o o istic of glass transition (Tmidmean = 147.0 C; Tonsetmean = 145.2 C). The measurements were repeated with in situ amorphisized (quench cooled) NH. The observed Tg was similar to o o that of spray-dried material (Tmidmean = 146.7 C; Tonsetmean = 145.0 C). This indicates that spray-dried material is not likely to contain significant plasticizing amounts of ethanol or water. This was confirmed by Karl Fischer titrations and gas chromatography.

The heat capacities of NH and spray-dried NH were measured by mDSC from 25o C to 220 oC. Figure 4a shows the difference between these capacities, the configurational heat conf conf o capacity (Cp ). The ΔCp during glass transition was found to be 161 J/mol/ K. This rel- conf atively large increase in ΔCp signifies fast relaxation during glass transition (i.e. fragile behaviour) of amorphous material/glass. This is in line with the generalized notion that pharmaceutical solids characterize as fragile glasses [18].

Thermodynamics of the amorphous state Table 1 lists the calculated thermodynamic parameters of the amorphous state of NH.

The parameters FT and m1, D1, m2 and D2 are all below or above the respective reference values for strong glass behaviour (FT > 1.5; D > 30; m < 40). These results indicate that amorphous NH is likely to be a fragile system. This is in agreement with the observed conf ΔCp . Thermal stability study of crystalline and novel spray-dried amorphous nilotinib hydrochloride | 123

A B 200 2 0 ) 150 /K C conf -2

D p (s)) -4 og /mol (l

(J 100

G -6 g conf lo p -8 C 50 -10 0 -12 130 140 150 160 0.00.5 1.0 Temperature oC Tg/T

35 C 80 DE3

70 K) l) 30 l) 2 ∞ 2.2 / mo mo ol 60 (kJ/ (kJ/ (J/m c c c H 25 G 1 S 50

20 40 0 0.00.5 1.0 0.00.5 1.0 0.00.5 1.0

T-Tg/Tm-Tg T-Tg/Tm-Tg T-Tg/Tm-Tg

conf Figure 4. mDSC results: a). Cp as a function of the temperature of amorphous NH. Indicated is the deter-

mination of ΔCp; b). Calculated values for Γ up to Tg for amorphous NH. The dotted line represents the ideal strong behaviour, deviations from linearity signify fragility; c-e). Configurational thermodynamic properties of

amorphous NH as a function of the reduced temperature, where 0 equals Tg and 1 equals Tm: c. Hc; d. Sc; e. Gc.

o The Tk values display a range of approximately 40 C, which is consistent with observations in literature [19]. The Tk values, as indicators for negligible molecular mobility, imply amorphous stability at typical storage temperatures (20-25 oC).

Γ was found to be near 100 s at Tg and found to increase exponentially as temperature decreases (figure 4b). This corresponds to literature data on fragile amorphous systems [13].

Figure 4c-e presents Hc, Sc, and Gc in relation to reduced temperature. The observed trends and values of the parameters are thought to correspond to a reduced tendency to crystallize [20]. The results suggest amorphous NH to be a stable, fragile system. These are desirable characteristics from a pharmaceutical point of view. 124 | Chapter 2.2

Thermogravimetric analysis Figure 5a presents the TGA curve at heating rates of 2.5, 5, 10, 20 and 40 °K/min and the DTG curve. The figure shows that the thermolytic degradation takes place in two distinct phases. Table 2 lists the thermal decomposition parameters of both phases at 10

°K/min: the extrapolated onset temperature of degradation (Tonset); the temperature at which 0.5% mass loss occurs (T0.5%); the temperature at maximum weight loss rate (Tmax); the extrapolated temperature at which the degradation process end (Tend); the amount of residue at Tend (res%).

A Deco 100 TGA 250 B 40 DTG

20 mp

w/w) 200 0.0 osi (% l) ti on 150 2.5 10 mo 50 sp ee (kJ/

d( 100 a E mg/min) percentage -0.1 50 OFW F 5 Mass 0 0 200 400 600 800 0.00 0.05 20 30 40 50 60 70 o Temperature ( C) a

Figure 5. TGA results: a). TGA (solid) and DTG (dotted; 10 oC/min) curves of crystalline NH obtained at heat- o ing rates of 2.5, 5, 10, 20 and 40 C/min (indicated in figure); b). The mean values ofE a plotted to α calculated by OFW and F.

The thermograms show a trend of movement towards higher temperatures with increasing heating rate. This can be attributed to a delay in thermal degradation and an increase in thermal lag.

o At a heating rate of 10 °K/min, NH is stable up to 193 C (T0.5%). This is below the Tm of 199 oC, which indicates that melting is accompanied by partial drug degradation. In the first phase a weight loss of 6.4 % (±0.5 %) occurs. This This can be attributed stoichiometrically to the loss of hydrochloride (6.4% (w/w)). The second phase involves a weight loss of 68.9 % (w/w). This is likely to resemble the degradation of a large part of the aromatic backbone of the drug structure. Thermal stability study of crystalline and novel spray-dried amorphous nilotinib hydrochloride | 125

Table 2. Summary of the thermal properties of crystalline and amorphous NH Parameter Value Parameter Value Parameter Value

o Tm ( C) 199 Phase I Phase I Phase II E (kJ/mol) ΔH (kJ/mol) 32.5 T (oC) 216 a 103 205 m onset Ozawa-Flynn-Wall o 91.4 197 ΔSm (J/mol/°K) 70.6 T0.5% ( C) 193 Friedman o Tmax ( C) 248 o o -1 6 11 Tg ( C) 147 Tend ( C) 294 A (s ) 5.6*10 4.7*10 6 11 ΔC (J/mol/°K) 161 res (%) 93.6 Ozawa-Flynn-Wall 1.6*10 3.3*10 p Friedman

FT (-) 1.31 Phase II o o ≠ Tk1 ( C) 108 Tonset ( C) 325 H (kJ/mol) 98.4 198 T (oC) 97 T (oC) 434 Ozawa-Flynn-Wall 87.1 190 k2 max Friedman o o Tk3 ( C) 65 Tend ( C) 546 ≠ Tm/Tg (-) 1.10 res (%) 20.4 S (J/mol/°K) -129 -36.9 D (-) 9.07 Ozawa-Flynn-Wall -139 -40.5 1 Friedman 2.2

m1 (-) 80.9 ≠ ΔETg (kJ/mol) 558 G (kJ/mol) 165 226 D (-) 11.0 Ozawa-Flynn-Wall 159 221 2 Friedman

m2 (-) 69.4 o T0 ( C) 50.2

Γ at Tg (s) 89.2

Hc at Tg (kJ/mol) 24.0 o Sc at Tg (J/mol/ K) 51.2

Gc at Tg (kJ/mol) 2.82

Thermal decomposition parameters The calculated thermal degradation parameters are listed in table 2. The isoconversional plots of the two degradation phases based on the OFW model and the isoconversional plots according to the F model are presented in figure 6.

The Ea values from both methods are shown as a function ofα in figure 5b. The resulting

Ea values vary between the two methods. This may be due to the different temperature integral approximations of the methods. The obtained Ea values from methods showed consistency within the phases of degradation. This suggests that the non-isothermal degradation phases are single-step processes.

The values of ΔS≠, ΔH≠ and ΔG≠ are dependent on the choice of the reaction model, which is in turn dependent on the particle shape and diffusion characteristics. This is beyond the scope of this analysis and should be the subject of future studies. 126 | Chapter 2.2

Thermal degradation can additionally be studied by using isothermal methods. In this case, such an approach is not feasible because of the two degradation phases that may interfere during isothermal heating.

4 A Phase I -4 B Phase I 99 99

) 98 -5 98 -1 3

97 T) 97

/d -6 2 96 a 96 ln(K*min b *d 95 ( -7 95 b )( ( 1 ln ln 94 -8 94

0 -9 0.0010 0.0012 0.0014 0.0016 0.0010 0.0012 0.0014 0.0016 -1 1/T (K ) 1/T (K-1)

A Phase II 4 -2 B Phase II 80 80

) 70 -3 70 -1 3

60 T) 60

/d -4 2 50 a 50 ln(K*min b *d 40 ( -5 40 b )( ( 1 ln ln 30 -6 30

0 -7 0.0009 0.0010 0.0011 0.0012 0.0010 0.0011 0.0012 -1 1/T (K ) 1/T (K-1)

Figure 6. Isoconversional plots of the two degradation phases derived by using the model-free methods: a, Ozawa-Flynn-Wall plots; b, Friedman plots.

Physical stability To experimentally assess the physical stability of amorphous NH, amorphous material was analysed isothermically by DSC at temperatures between Tg and Tm. Drug material was kept at 150, 155, 160, 165, 170, 175 and 180 oC for 5 hours. During these periods, no exothermic event was observed that could indicate crystallization. Additionally,T gs were still present during reanalysis. These observations are in agreement with the thermodynamic parameters that indicate amorphous stability.

To study any physical changes in the bulk amorphous material under storage conditions, the material was kept in the dark at 20-25 oC and a relative humidity of 60%. Analyses were performed with Karl Fisher titration, XRD and DSC at regular time intervals. Table 3 presents the results up to 6 months of storage. The amorphous material attracts water up to 0.54% over time but does not seem to destabilize because of it, at least up to 6 months. Thermal stability study of crystalline and novel spray-dried amorphous nilotinib hydrochloride | 127

Table 3. Physical stability results of amorphous NH

o Time point Water content (%, (w/w)) XRD results Tg ( C) 1 month 0.28 Amorphous 147 2 months 0.34 Amorphous 146 4 months 0.42 Amorphous 147 6 months 0.54 Amorphous 146

Conclusions

This study presents the thermal stability analysis of crystalline NH monohydrate, NH and spray-dried amorphous NH. The resulting thermal characteristics of NH can be used to evaluate the pharmaceutical quality and to further understand the response of the 2.2 drug to thermal stress during processing and storage. The spray drying amorphization procedure was monitored by NMR, FTIR and XRD techniques. The method was shown to produce amorphous NH without drug degradation.

The amorphous phase of NH has been characterized with respect to thermodynamic and molecular mobility both above and below Tg. The thermodynamic properties indicate a fragile amorphous phase and low crystallization tendency.

The thermal degradation of NH takes places in two phases. The initial degradation -tem o o perature was found to 193 C, below the Tm of 199 C. The obtained results from TGA were evaluated using non-isothermal, model-free isoconversional methods. The derived

Ea values are consistent with single-step degradation processes for both phases.

The examination of the physical stability of amorphous NH during storage and at elevated temperatures showed stability at 180 oC for at least 5 hours and at 20-25 oC/60% RH for at least 6 months.

The results from this study demonstrate that crystalline and amorphous NH may be regarded as relatively stable drug phases. In conclusion, NH is a suitable candidate for new formulation studies where thermal production methods are utilized such as hot-melt extrusion and spray drying.

Future studies should focus on the stability of newly developed formulations with amorphous NH. 128 | Chapter 2.2

References

[1] US Food and Drug administration (FDA), Prescribing [10] L.X. Yu, G. Amidon, M.A. Khan, S.W. Hoag, J. Polli, G.K. information Tasigna(R), 2010. Raju, J. Woodcock, Understanding pharmaceutical [2] F.X. Mahon, S. Hayette, V. Lagarde, F. Belloc, B. Turcq, quality by design., AAPS J. 16 (2014) 771–83. F. Nicolini, C. Belanger, P.W. Manley, C. Leroy, G. Eti- [11] S. Vyazovkin, A.K. Burnham, J.M. Criado, L.A. enne, S. Roche, J.-M. Pasquet, Evidence that Resis- Pérez-Maqueda, C. Popescu, N. Sbirrazzuoli, ICTAC tance to Nilotinib May Be Due to BCR-ABL, Pgp, or Kinetics Committee recommendations for performing Src Kinase Overexpression, Cancer Res. 68 (2008) kinetic computations on thermal analysis data, Ther- 9809–9816. mochim. Acta. 520 (2011) 1–19. [3] B.C. Hancock, G. Zografi, Characteristics and signifi- [12] D. Huang, G.B. McKenna, New insights into the fragili- cance of the amorphous state in pharmaceutical sys- ty dilemma in liquids, J. Chem. Phys. 114 (2001) 5621. tems., J. Pharm. Sci. 86 (1997) 1–12. [13] K.J. Crowley, G. Zografi, The use of thermal methods [4] G. Jesson, M. Brisander, P. Andersson, M. Demirbüker, for predicting glass-former fragility, Thermochim. H. Derand, H. Lennernäs, M. Malmsten, Carbon Diox- Acta. 380 (2001) 79–93. ide-Mediated Generation of Hybrid Nanoparticles for [14] K.A. Graeser, J.E. Patterson, J.A. Zeitler, K.C. Gordon, Improved Bioavailability of Protein Kinase Inhibitors, T. Rades, Correlating thermodynamic and kinetic pa- Pharm. Res. 31 (2014) 694–705. rameters with amorphous stability, Eur. J. Pharm. Sci. [5] P. Andersson, M. Von Euler, M. Beckert, Compara- 37 (2009) 492–498. ble pharmacokineties of 85 mg RightSize nilotinib [15] H.L. Friedman, Kinetics of thermal degradation of (XS003) and 150 mg Tasigna in healthy volunteers us- char-forming plastics from thermogravimetry. Ap- ing a hybrid nariopartick-based formulation platform plication to a phenolic plastic, J. Polym. Sci. PartC for protein kinase inhibitors, in: 2014 ASCO Annu. Polym. Symp. 6 (2007) 183–195. Meet., 2014. [16] J.H. Flynn, The “Temperature Integral” — Its use and [6] S. Gupta, R. Kesarla, A. Omri, Formulation strategies abuse, Thermochim. Acta. 300 (1997) 83–92. to improve the bioavailability of poorly absorbed [17] P.W. Manley, W.-C. Shieh, P.A. Sutton, P.H. Karpins- drugs with special emphasis on self-emulsifying sys- ki, Crystalline forms of 4-methyl-n-[3-(4-meth- tems., ISRN Pharm. 2013 (2013) 848043. yl-imidazol-1-yl)-5-trifluoromethyl-phenyl]-3-(4- [7] J. Li, D. Patel, G. Wang, Use of Spray-Dried Dispersions pyridin-3-yl-pyrimidin-2-ylamino)-benzamide, in Early Pharmaceutical Development: Theoretical WO2007015870A2, 2005. and Practical Challenges, AAPS J. 19 (2017) 321–333. [18] C.A. Angell, Entropy and fragility in supercooling liq- [8] D. Giron, Contribution of thermal methods and relat- uids, J. Res. Natl. Inst. Stand. Technol. 102 (1997) 171. ed techniques to the rational development of phar- [19] A. Kaushal, A. Bansal, Thermodynamic behavior of maceuticals—Part 1, Pharm. Sci. Technolo. Today. 1 glassy state of structurally related compounds, Eur. J. (1998) 191–199. Pharm. Biopharm. 69 (2008) 1067–1076. [9] S. Bajaj, D. Singla, N. Sakhuja, Stability Testing of [20] D. Zhou, G.G.Z. Zhang, D. Law, et al., Physical stability Pharmaceutical Products, J. Appl. Pharm. Sci. 2 (2012) of amorphous pharmaceuticals: Importance of con- 129–138. figurational thermodynamic quantities and molecular mobility., J. Pharm. Sci. 91 (2002) 1863–72.

Chapter 3

Formulation improvement of oral anticancer agents

Chapter 3.1

Solubility and bioavailability improvement of pazopanib hydrochloride

M. Herbrink1, S. Groenland1, A.D.R. Huitema1,2, J.H.M. Schellens3,4, J.H. Beijnen1,4, N. Steeghs3, B. Nuijen1

1Department of Pharmacy and Pharmacology, The Netherlands Cancer Institute – Antoni van Leeuwenhoek, Amsterdam, The Netherlands. 2Department of Clinical Pharmacy, Utrecht University Medical Center, Utrecht, The Netherlands 3Department of Medical Oncology and Clinical Pharmacology, The Netherlands Cancer Institute – Antoni van Leeuwenhoek, Amsterdam, The Netherlands. 4Department of Pharmaceutical Sciences, Utrecht University, Utrecht, The Netherlands. 134 | Chapter 3.1

Abstract

The anticancer drug pazopanib hydrochloride (PZH) has a very low aqueous solubility and a variable oral bioavailability. A new pharmaceutical formulation with an improved solubility may enhance the bioavailability and reduce the variability. A broad selection of polymer excipients was tested for their compatibility and solubilizing properties by conventional microscopic, thermal and spectrometric techniques. A wet milling and mixing technique was used to produce homogenous powder mixtures. The dissolution properties of the formulation were tested by a pH-switch dissolution model. The final formulation was tested in vivo in cancer patient following a dose escalation design. Of the tested mixture formulations, the one containing the co-block polymer Soluplus® in a 8:1 ratio with PZH performed best in terms of in vitro dissolution properties. The in vivo results indicated that 300 mg of the developed formulation yields similar exposure and a lower variability (379 μg/mL*h (36.7% CV)) than previously reported values for the standard PZH formulation (Votrient®) at the approved dose of 800 mg. Furthermore, the expected plasma-Cthrough levels (27.2 μg/mL) exceeds the defined therapeutic efficacy threshold of 20 μg/mL. Solubility and bioavailability improvement of pazopanib hydrochloride | 135

Introduction

Pazopanib (figure 1) is a multiple kinase inhibitor that is currently trademarked as Vot- rient®. It is indicated for the treatment of advanced renal cell carcinoma and advanced soft tissue sarcoma [1,2]. The recommended dose is 800 mg once daily taken without food.

pKa 6.4 N N N NH

pKa 2.1 O N S N O NH2 3.1

Figure 1. Molecular structure of pazopanib with pKa values and associated nitrogen atoms.

Pazopanib hydrochloride (PZH) was found to exhibit a low aqueous solubility [3]. The intestinal permeability of PZH is considered to be high [4]. Therefore, the drug is classified as a class II compound in the biopharmaceutics classification system (BCS). This implies that its absorption and bioavailability is primarily hindered by solubility.

The commercial formulation of PZH consists of a physical mixture of microcrystalline cellulose, povidone K30, magnesium stearate and sodium starch glycolate in the form of an immediate release tablet. The oral bioavailability of 800 mg of the commercial formulation was found to have a median of 21% [5]. Pazopanib exhibits a relatively large inter-patient variability in both exposure and plasma concentrations [6]. The intra-patient variability in these pharmacokinetic parameters was found to be similarly large [7]. This large variability may be related to the variable absorption process of pazopanib [8].

Since bioavailability and drug exposure are tightly linked, pharmacokinetic parameters are influenced by poor solubility and permeability. Variability in drug exposure may cause drug plasma levels to be inadequately low or toxically high [9]. Enhancing the solubility of PZH through improvement of its formulation may increase the bioavailability and 136 | Chapter 3.1

possibly reduce pharmacokinetic variability. An open-label study has been reported that compared crushed Votrient® tablets and a pazopanib suspension to regular Votrient® and that showed the interventions to significantly increase both exposure and plasma concentrations [10]. Co-administration with food also increases the pazopanib exposure by two-fold, further supporting the possible impact of solubility enhancement [11].

As of yet, no clinical study has been performed with an improved-solubility solid oral formulation containing pazopanib hydrochloride, as far as we know.

Improvement of drug solubility may be achieved through the use of various techniques [12–14]. Amongst the possible methods, the preparation of tablets or capsules from physical powder mixtures with solubility-enhancing excipients is most straightforward. Excipients like these are able to undergo a physical interaction with dissolved drug molecules, thereby stabilizing the dissolved system at higher drug concentrations than would occur in an excipient-free drug solution. As the preparation of physical mixtures most -of ten does not require relatively expensive equipment or extensive product pre-treatment steps, it can be considered as the most economical formulation approach. Furthermore, the manufacture of powder mixtures does not involve thermal processing, as is the case with techniques such as spray-drying and hot-melt extrusion.

The aim of this study was to develop such an oral solid dosage form of PZH with an increased solubility and to characterize this formulation both in vitro and in vivo. This study is first and unique in several aspects concerning the PZH and the formulation design approach. Firstly, it takes into account the excipient compatibilities of PZH. Second- ly, excipient selection and formulation composition fine-tuning was performed through the utilization of model pH-switch system to account for different environments in the gastro-intestinal system. Additionally, the nature of the solubility improvement of the best-performing excipient is explored. Finally, the optimal formulation was testedin vivo in cancer patients.

The above constitutes an efficient and economical procedure that resulted in new data, especially for the drug PZH and the excipient Soluplus®, that may be of value to future formulation projects. Solubility and bioavailability improvement of pazopanib hydrochloride | 137

Materials and methods

Materials PZH was purchased from Avachem Scientific (San Antonio, TX, USA). Kollidon 12PF, 30. VA64, 90F, polyethylene glycol (PEG) 6,000 (6K) and 35,000 (35K), Lutrol F68, and Soluplus® were kindly supplied by BASF (Ludwigshafen, Germany). Lactose monohydrate was obtained from DFE Pharma (Goch, Germany). Crystalline Cellulose PH102 and gela- tine capsules size 1 were purchased from Spruyt Hillen (Ijsselstein, The Netherlands). All other reagents were purchased in analytical grade.

Fourier transform infrared spectroscopy (FT-IR) Infrared spectra were recorded from 600 – 4000 cm-1 with a resolution of 2 cm-1 with a FT-IR 8400S Spectrophotometer equipped with a Golden Gate® (Shimadzu, ‘s-Hertogen- bosch, The Netherlands). A total of 64 scans were averaged into one spectrum.

Thermogravimetric analysis (TGA) 3.1 TGA was performed on a Q50 thermogravimetric analyzer (TA Instruments, New Castle, DE, USA) under a nitrogen flow of 50 mL min-1. The sample (approximately 10 mg) was weighed into a platinum sample pan (TA Instruments). Samples were heated in a temperature range of 25-1000 oC. Temperature calibration was carried out using a high purity magnetic reference (nickel) for Curie temperature determination. The analysis was performed in triplicate and the spectra were averaged. The data were analyzed with the Advantage software v5.5 (TA Instruments).

Preparation of PZH formulations Physical mixtures of PZH and polymers were prepared with mortar and pestle in the presence of methyl tert-butyl ether (MTBE) as a wetting agent. Batch sizes were between 900 and 9,000 mg with a total mixing and grinding time of 10 min. Geometric dilution was applied during the mixing process. The best performing physical mixture was filled out in cellulose capsules size 1. Content uniformity of screening powder mixtures was determined on 3 samples from each batch. Content uniformity of clinical batches was performed according to the European Pharmacopoeia 9th ed. All samples were dissolved in eluent and were subsequently analyzed using a validated stability-indicating High performance liquid chromatography (HPLC)-UV method: injection volume of 10 μL on a gemini C18 column (50 x 2.0 mm, 4.6 μm; Phenomenex, Torrance, CA, USA) at 55 oC, isocratic elution at 0.4 mL/min with 55%B (A: 10 mM 4NH OH in water; B, 1 mM NH4OH in methanol) and UV-detection at 268 nm.

Dissolution testing The pH-solubility profile of PZH was assessed by measuring dissolved drug in NaCl/HCl and phosphate buffers at different pH values. An excess amount of PZH was placed in a known volume of buffer and allowed to saturate for 1 hour after which a sample was analysed after filtration through an 0.20 μm filter by HPLC-UV.

A collection of different excipients were efficiently screened on their potential solubility improving properties by suspending an excess of PZH in 10% polymer solutions with a pH of 6.8 for 2 and 24 hours.

The dissolution behavior of various formulations was determined with a small-scale dissolution test. Briefly, an amount of powder equivalent to 10 mg PZH was added to a 25 mL beaker containing 10 mL medium. The temperature was kept at 37 oC and the medium was stirred at 200 rpm. The commercial formulation (Votrient®) and the capsulated investigational formulations were tested in a larger dissolution volume of 200 mL. Disso- lution tests were performed in single pH environment (simulated intestinal fluid without pancreatin (SIFsp), pH 6.8) and a pH-switch system. In this system, the 25 mL beaker contained 10 mL of simulated gastric fluid without pepsin (SGFsp), pH 1.6. The sample was kept in this environment for 30 min after which the pH and composition were changed into that of SIFsp in a step-wise manner by briefly adding phosphate or carbonate buffer while monitoring the pH during the dissolution. The switch of pH and the composition adjustment was typically completed within 32 min with a total endo volume of 20.72 mL Solubility and bioavailability improvement of pazopanib hydrochloride | 139

for the phosphate system and 22.10 mL for the carbonate system. Samples were taken at designated time points, filtered using a 0.20 μm filter (Millipore, Burlington, MA, USA) and diluted 2-fold with methanol. All samples were subsequently analyzed using the validated stability-indicating (HPLC)-UV method. Each dissolution experiment was carried out in triplicate.

Micelle size measurements The presence and size of micelles formed by Soluplus® were detected and measured through dynamic light scattering (DLS) by using a Zetasizer Nano S90 particle size analyzer (Malvern Instruments, Worcestershire, UK). Samples were taken from the dissolution media at different time points for micelles, placed in polystyrene cuvettes (Malvern In- struments) and analyzed in threefold at 10 analyses per measurement.

Electron microscopy The morphology and particle size of the bulk and physical mixture powders were studied using scanning electron microscopy (SEM). Samples were placed on conducting double 3.1 sided adhesive tape and on an aluminium holder. Samples were sputter-coated with gold using a SCD 040 sputter coater (Oerlikon, Balzers, Liechtenstein). Imaging was performed through back-scattering with a Sigma Field Emission Gemini system (Zeiss, Oberkochen, Germany). Each microscopy analysis was applied on three separate samples.

The shape and size of micelles were visualized with Transmission Electron Microsco- py (TEM). Samples from dissolution media were diluted 100 times and were applied on Agar1 formvar/carbon coated copper grids (van Loenen instruments, Zaandam, The Netherlands). Negative staining with 2% (w/w) phosphotungstic acid was applied with subsequent drying for 10 min prior to the microscopy. Recording was performed with a Tecnai T12 G2 Spirit Biotwin set-up (FEI company, Hillsboro, OR, USA).

X-ray powder diffraction (XRD) X-ray diffraction of powder samples was performed with a X’pert pro diffractometer equipped with an X-celerator (PANanalytical, Almelo, The Netherlands). Samples were placed in a 0.5 mm deep metal sample holder. Samples were scanned at a current of 30 mA and a tension of 40 kV. The scanning range was 10 – 60 degrees 2-θ, with a step size of 0.020 degrees 2-θ and a scanning speed of 0.002 degrees 2-θ per second. Analyses were performed in triplicate. 140 | Chapter 3.1

Residual solvents Residual MTBE was determined with capillary gas chromatography (GC) analysis. Sam- ples of approximately 20 mg of powder were dissolved or suspended in 1.0 mL water and shaken for 3 hours. Aliquots were subsequently filtered and transferred to autosampler vials. Analyses were performed using 6890N GC system (Agilent, Santa Clara, CA, USA) equipped with a Flame-ionization detector (Agilent) and an RTX-1301 capillary column (3.0 µm film, 30 m x 0.53 mm; Restek Corporation, Bellefonte, PA, USA). Analyses were performed in triplicate.

Clinical study design The clinical study was designed as an open-label, proof-of-concept study. The pharmacokinetic parameters of pazopanib after administration of formulation-filled capsules were determined in cohorts consisting of 3 patients per dose level. The capsules were administered in combination with approximately 150 mL tap water after an overnight fast. The study objective was to find the dose of the formulation that could provide similar PK parameters as 800 mg of Votrient®. The dose level of the first cohort was set on 100 mg (4 capsules of 25 mg). Doses were increased in-between cohorts with a cap at 200% of the previous dose. This continued for a maximum of 3 cohorts, after which an expansion of 3 patients at the highest dose cohort was planned for statistical purposes.

A complete physical examination along with a review of the medical history and concomitant medication was performed before inclusion. During the study, vital signs, WHO performance status, weight, hematology, blood chemistry, and adverse events were monitored. The study protocol was approved by the medical ethics committee of the Netherlands Cancer Institute; all patients had to give written informed consent prior to the start of the study. The study was registered on ClinicalTrials.gov under NCT02768441 and at EudraCT under 2016-001105-16.

Pharmacokinetics and bioanalysis Blood samples were drawn in potassium EDTA tubes at baseline and at 0.5, 1, 2, 3, 4, 5, 6, 8, 10, 12 and 24 hours after pazopanib intake. Samples were immediately centrifuged at 2000 g for 10 min at 4 oC. Plasma was stored at or below -20 oC until analysis. Pazopanib was quantified in plasma by use of HPLC with tandem mass spectrometric detection (LC- MS/MS) as described previously [15]. Solubility and bioavailability improvement of pazopanib hydrochloride | 141

Data analysis For all quantitative analyses, results were pooled per sample type and means and standard deviations (SD) were calculated.

Individual pharmacokinetic parameters were analyzed using descriptive pharmacokinetic methods and validated R scripts (R version 3.3.2). The areas under the plasma concentration-time curves up to 24 hours (AUC0-24) were estimated by the linear trapezoidal rule for the absorption phase and the logarithmic trapezoidal rule for the elimination phase.

The observed maximum plasma concentration (Cmax), minimum plasma concentration

(Cmin) and the time to the maximum plasma concentration (tmax) were reported.

3.1 142 | Chapter 3.1

Results

Powder mixing Dry mixing of PZH and excipients did not yield homogenous mixtures. Physical mixtures were therefore produced by wet milling and mixing. For the wetting, several organic solvents were tested for their possible undesirable ability to dissolve (>0.5 μg/mL) one or both components of the physical mixture. Only MTBE was found to not dissolve or react with any of the components. Additionally, MTBE evaporates relatively quickly, leaving less than 100 ppm after 10 min as determined by GC analysis. Additionally, it was found that powder mixtures had appropriate uniformity of content (content >98% and SD <2%) after a mixing time of 10 min.

Excipient compatibility The excipient compatibility study was carried out for 6 months at 30 oC in the dark. The results are listed in table 1. Of the examined excipient-PZH combinations, the mixtures with lactose monohydrate, PEG 6K and PEG 35K showed signs of interaction in FTIR. DSC analysis showed shifts in melting temperatures for these excipients. Furthermore, TGA revealed a shift in degradation temperatures from 190o C for PZH for the PEG 6K and the PEG 35K combinations. A reduced content was found by HPLC-UV (for each of the three excipients suggesting chemical reactivity. The formulation development study was continued with the excipients that displayed no interaction with PZH.

Dissolution Figure 2a presents the dissolution curves of both Votrient® tablets, the pure PZH drug material and the pH solubility curve of PZH. The initial 30 min of the dissolution curve represents the stomach (pH 1.6) in which the percentage of dissolved PZH from Votri- ent® accumulates up to approximately 47%. At time points 35, 40 and 45 min a relatively small increase to 52% in the dissolved drug was observed when the pH rises from 1.6 to 3.2. From 50 min on the dissolved amount decreases to less than 0.5% when the pH is raised above 5.2 and is subsequently adjusted to 6.8. The same behavior is seen with the pure PZH of which the pH-solubility profile in a series of buffers is shown in figure 2b. Solubility and bioavailability improvement of pazopanib hydrochloride | 143

Table 1. Data (mean (SD)) from the compatibility study of PZH and excipient:PZH (1:1) physical mixtures after 6 months of storage at 30oC and 60%RH. Powder DSC TGA FTIR HPLC-UV

o o Type Temperature ( C) Tonset ( C) Indication of Content (%) chemical interaction*: PZH - - 190 (3.2) - - MCC 160 (0.5) 260 (4.1) T None 99.8 (2.2) MCC:PZH g 158 (1.0) 185 (3.0) LM 221 (1.7) 301 (2.9) T Yes 94.7 (4.2) LM:PZH m 208 (3.1) 187 (2.4) SLP 69 (0.6) 248 (3.4) T None 101.2 (3.2) SLP:PZH g 70 (1.1) 192 (2.7) 12PF 87 (0.4) 183 (5.1) T None 99.6 (0.5) 12PF:PZH g 86 (0.8) 185 (1.7) K30 147 (0.2) 175 (2.8) T None 99.3 (1.6) K30:PZH g 147 (0.3) 177 (3.5) VA64 97 (0.8) 203 (4.6) T None 98.9 (4.8) VA64:PZH g 96 (0.7) 189 (2.0) 90F 153 (0.6) 180 (3.9) T None 99.7 (1.4) 90F:PZH g 151 (0.5) 182 (3.0) 3.1 6K 61 (2.1) 197 (2.1) T None 96.7 (5.2) 6K:PZH m 55 (2.5) 134 (4.5) 35K 70 (3.2) 201 (1.9) T None 94.6 (4.5) 35K:PZH m 65 (2.4) 147 (2.1) F68 53 (1.1) 203 (2.0) T None 100.4 (1.8) F68:PZH m 53 (2.8) 195 (2.8) MCC, Microcrystalline Cellulose; LM, Lactose monohydrate; SLP, Soluplus®; 12PF, Kollidon 12PF; K30, Kol- lidon 30; VA64, Kollidon VA64; 90F, Kollidon 90F; 6K, PEG 6,000; 35K, PEG 35,000; F68, Lutrol 68F. *Indi- cated by deviation from overlap spectrum (t=0) of more than 1%.

60 AB1500

Votrient(R) ) ) mL (% g/

40 m 1000

ZH PZH H( dP PZ

20 500 Dissolve Dissolved 0 0 0204060 02468 Time (min) pH

Figure 2. A. Dissolution curves of Votrient® tablets and PZH bulk powder; B. pH-solubility curve of PZH. The dotted line indicates the start of the pH-switch. 144 | Chapter 3.1

The excipient screening experiments were performed for both 2 and 24h (figure 3a). PZH has a solubility of approximately 0.1 μg/mL in this buffer system after both periods. The tests show that the addition of Kollidon VA64 , Soluplus®, Kollidon 30 and Kollidon 90F to the formulation provided the highest PZH concentrations after both 2h and 24h.

B A 100 1000

) 42 min 100 mL %) g/ (

m 30 min ( 10 ZH dP ZH 50 ve dP ol

ve 1 Diss 60 min ssol

Di 0.1

0.01 0 F 0204060 H 64 us 30 PZ A F68 90F V 12P upl l ol on on Time (min) on on So tr llid Lu o llid llid llid K o o o K K K

Figure 3. A. Excipient screening results: PZH solubility after 2 (left bars) and 24 h (right bars); B. Dissolution curve of PZH:Kollidon VA64 mixture (1:1, w/w) with dissolution points referring to table 2.

Several physical mixtures with different ratios between drug and excipient were produced (Table 2) and their dissolution behavior was tested using the pH-switch system. The results are shown in figure 3b and table 2. After the initial solubility increase upon elevating the pH, the solubility of PZH in all of the Kollidon powder mixes collapsed to > 1% before the pH reaches 6.8. A slight solubility increase beyond the pH-switch was seen for the 1/5 Soluplus® mixture, but this collapsed around pH 6.0. The 1/10 Soluplus® mixture proved to be able to maintain the solubility of PZH. Furthermore, it increased solubility to 100% during the pH increase to 6.8. Additional physical mixtures of PZH and Soluplus® were prepared and subjected to dissolution tests to further optimize the composition between the two components. The results are presented in figure 4a. In- creasing the amount of Soluplus® relative to PZH increased solubility up to 8 equivalent parts (figure 4b). This composition was chosen as the best performing formulation and was designated PZHSol001 and filled out in cellulose capsules size 1 with a potency of 25 mg per capsule. Solubility and bioavailability improvement of pazopanib hydrochloride | 145

AB 1/10 100 1/8 100

1/7 ) ) (% (% ZH ZH dP dP 50 50 ve ve ol ol Diss Diss 1/6 1/5 1/3 0 1/1 0 0204060 0510 Time (min) Equivalent parts Soluplus“ (-)

Figure 4. A. Dissolution curves of PZH:Soluplus® with different compositions as indicated on the right. The dotted line indicates the start of the pH-switch; B. Solubility of PZH from physical mixtures with Soluplus® as a function of the equivalent parts of Soluplus®.

Characterization of the final formulation

PZHSol001 shows an increase in solubility upon the transformation from SGFsp to SIFsp. It was observed that this increase was accompanied by the medium becoming more clear. 3.1 Initially, in SGFsp, the solution was opaque. Analysis of medium samples by TEM clearly showed the presence of uniformly shaped micelles at both stomach and intestinal pH values (figure 5).

At different time points during the dissolution in both phosphate and carbonate systems, samples were taken to measure the micelle diameters, figure 5a and 5b. In both systems, the micelle diameter initially increases upon pH increase from 91 ± 4.2 nm in SGFsp to 194 ± 10 nm at pH 1.8 in the carbonate system and to 169 ± 8.2 nm at pH 2.6 in the phosphate system. In both systems, the micelle diameter decreased at higher pH values up to 6.8 to 80 ± 2.1 nm (carbonate) and 101 ± 5.0 nm (phosphate). The micelle size of Soluplus® in absence of PZH showed no significant increases in either system. Figure 5c shows TEM micrographs of micelles from the physical mixture of PZH and Soluplus® 1:8 in SGFsp and SIFsp. The micrographs confirm the similarity of sizes in SGFsp and SIFsp and the magnitude of the micelle diameters as measured by DLS. 146 | Chapter 3.1

AB PZH: Soluplus“ . 1:8 Soluplus “ m) (n e 100 erag av z

246 246 C pH pH

200 nm Figure 5. Results from the micelle analyses. Mean micelle diameters as a function of the pH of the dissolu-

tion medium: A. Phosphate SIFsp; B. Carbonate SIFsp; C. TEM micrograph of PZH:Soluplus® micelles from a 1:8

physical mixture in SIFsp (Magnification: 120000, high voltage: 100 kV), left: in SGFsp, right: in phosphate SIFsp.

Figure 6 presents the SEM micrographs of PZH and Soluplus® bulk powder and PZH- Sol001 powder. The images illustrate the size difference between the drug and excipient particles. Additionally, it shows that the wet mixing and milling method reduces the overall particle size of the formulation. The size of the Soluplus® particles is especially decreased, which is beneficial to the mixture homogeneity.

Both bulk powder and final product were analyzed using XRD (figure 6). The pattern of Soluplus® did not display characteristic ‘crystal’ reflections and can, therefore, be considered amorphous. The obtained pattern of PZHSol001 exhibited the same reflections as the pattern of PZH, albeit reduced in intensity due to the presence of Soluplus®.

After 12 months of storage in open containers at 20-25 oC and 60% relative humidity (RH) and after 6 months in open containers at 40 oC and 75% RH the formulation was subjected to dissolution and assay tests. In this period, no significant changes occurred in drug content and dissolution properties. The moisture content did show an increase of 2.7% (60% RH) and 4.5% (75% RH). Solubility and bioavailability improvement of pazopanib hydrochloride | 147

3.1

Figure 6.6. XRD patterns ofof PZHPZH bulk,bulk, Soluplus®Soluplus® bulk and PZHSol001 (physical (physical mixture PZH:Soluplus®, PZH:Soluplus®, 1:8). Inserts1:8). Insertsshow SEM show micrographs SEM micrographs (extra high (extra tension: high 2.00 tension: kV) of 2.00 the correspondingkV) of the corresponding materials (magnifications: materials 500x, 50x and 250x respectively).(magnifications: 500x, 50x and 250x respectively).

Both bulk powder and final product were analyzed using XRD (figure 6). The pattern of Soluplus®Human in did vivo not clinical display study characteristic ‘crystal’ reflections and can, therefore, be considered amorphous.Twelve evaluable The obtained patients pattern were includedof PZHSol001 in the exhibited study, relevant the same patient reflections characteristics as the patternare listed of inPZH, table albeit 3. Thereduced cohort in intensity doses were due toas thefollows: presence cohort of Soluplus®. 1: 100 mg (n=3), cohort 2: After 12 months of storage in open containers at 20-25 oC and 60% relative humidity (RH) and200 aftermg (n=3), 6 months cohort in open 3: 300 containers mg (n=3). at 40 Cohort oC and 4 75% (n=3) RH was the formulationan extension was of subjected cohort 3. to All dissolutionpatients received and assay PZHSol001 tests. In this25 period,mg capsules no significant at the designated changes occurred cohort indose drug level. content Figure and7 shows dissolution the mean properties. concentration-time The moisture content curves did of show pazopanib an increase after of 2.7%administration (60% RH) andfor each 4.5% (75% RH). dose level. The relevant pharmacokinetic parameters (Tmax, Cmax, Cmin, and AUC0-24) and the variations thereof are shown in table 4. for the cohorts of PZHSol001.

116

148 | Chapter 3.1

Table 2. Components, weight ratios, content uniformity (mean (SD)) (n=3) and essential dissolution time points (means (SD)) (n=3) of PZH physical mixture formulations. Formulation Components Weight ratio Content Dissolution (%)* (w/w) uniformity (%) 30 min 42 min 62 min A PZH - - 28 (2.2) 38 (4.5) 0.3 (0.0) B PZH, Kollidon VA64 1/1 98.4 (1.2) 40 (2.4) 50 (2.0) 0.1 (0.0) C PZH, Kollidon VA64 1/5 99.1 (1.0) 69 (5.4) 81 (4.1) 0.1 (0.0) D PZH, Kollidon VA64 1/10 100.2 (0.5) 99 (3.0) 100 (4.0) 0.3 (0.1) E PZH, Soluplus® 1/1 99.4 (0.6) 40 (2.5) 50 (2.1) 0.1 (0.0) F PZH, Soluplus® 1/5 99.2 (0.8) 56 (2.7) 70 (2.6) 2.5 (0.2) G PZH, Soluplus® 1/10 99.5 (0.3) 83 (2.3) 98 (3.1) 100 (2.3) H PZH, Kollidon 30 1/1 98.7 (1.1) 40 (3.6) 49 (2.4) 0.1 (0.1) I PZH, Kollidon 30 1/5 98.8 (1.8) 67 (2.3) 81 (2.8) 0.2 (0.1) J PZH, Kollidon 30 1/10 99.2 (1.4) 97 (3.5) 100 (8.7) 0.5 (0.1) K PZH, Kollidon 90F 1/1 99.0 (0.4) 40 (2.1) 48 (3.2) 0.0 (0.0) L PZH, Kollidon 90F 1/5 99.5 (0.7) 55 (2.5) 66 (3.6) 0.0 (0.0) M PZH, Kollidon 90F 1/10 99.3 (0.9) 77 (5.1) 93 (1.2) 0.0 (0.0) * time points correspond to the dissolution time points as indicated in figure 3

) 40 mL g/ m (

on 30 rati nt ce

on 20 ac sm la 300 mg bp 10 ni

pa 200 mg zo 100 mg Pa 0 04812 16 20 24 Time (hours)

Figure 7. Mean plasma concentration-time curves of pazopanib after oral administration of single PZHSol001 (physical mixture PZH:Soluplus®, 1:8) doses as 25 mg capsules; doses are indicated on the right, 100 mg (n=3), 200 mg (n=3) and 300 mg (n=6).

The single dose administrations of PZHSol001 were well tolerated. Adverse events that were possibly related to the administration of PZHSol001, were: bleeding gums, head- ache, productive cough (all CTCAE grade 1; n=1; dose 300 mg). Solubility and bioavailability improvement of pazopanib hydrochloride | 149

Table 3. Patient characteristics of the clinical study per cohort. Parameter N Cohort 1 2 3 & 4 Dose 100 mg 200 mg 300 mg Sex Male 1 2 5 Female 2 1 1 Age (years) Median 63 56 61 Range 45-73 40-60 47-74 Ethnic background Caucasian 3 3 4 African 2 WHO performance status 0 2 3 6 1 1 Pathological diagnosis Soft Tissue 1 2 2 Kidney 1 Skin 1 1 Breast 1 3.1 Gastric 2 Colorectal 1 Number of prior treatments (chemotherapy) 0 1 1 2 1 1 1 4 2 1 1 150 | Chapter 3.1

Table 4. PK parameters from the patients of the clinical study (means (CV%)) and PK parameters from the Votrient ® phase I trial [6]. Dose (mg) Parameter PZHSol001 Votrient® 100 n 3 3

tmax (h) 2.65 (23.5) 4.0*

Cmax (μg/mL) 4.52 (39.2) 6.5 (25.7)

Cmin (μg/mL) 2.31 (38.9) 2.8 (11.2)

AUC0-24 (μg/mL*h) 70.3 (40.3) 96.9 (17.6) 200 n 3 3

tmax (h) 4.01 (0.66) 3.0*

Cmax (μg/mL) 15.4 (28.4) 7.5 (54.5)

Cmin (μg/mL) 5.9 (12.1) 2.9 (53.6)

AUC0-24 (μg/mL*h) 192 (18.1) 104.5 (54.6) 300 n 6 -

tmax (h) 4.02 (1.11) -

Cmax (μg/mL) 28.4 (33.1) -

Cmin (μg/mL) 11.3 (41.2) -

AUC0-24 (μg/mL*h) 379 (36.7) - 800 n - 10

tmax (h) - 3.5*

Cmax (μg/mL) - 19.4 (176)

Cmin (μg/mL) - 9.4 (240)

AUC0-24 (μg/mL*h) - 275.1 (203) *No CV values were reported Solubility and bioavailability improvement of pazopanib hydrochloride | 151

Discussion

Powder mixing Dry mixing of the powder components did not yield homogenous mixtures, mainly due to the electrostatic charging of PZH. The size initial difference between excipient and drug may also account for this as shown in the SEM micrographs. For this reason, the powder components were wetted prior to mixing and left to dry in a desiccator thereafter. This approach allowed for the deagglomeration of PZH particles and sharply reduced particle sizes of both PZH and the excipients, especially for Soluplus®, and offers an explanation for the more homogenous powder mixtures. Particle size may also influence mixing efficiency. The results from the content uniformity show that the mixing and grinding approach is sufficient in minimizing the effect of the particle size on the homogeneity of the powder mixtures. Additionally, this was also found to be reproducible.

Excipient compatibility Drug-excipient interactions have been shown to negatively affect the pharmaceutical 3.1 properties of drug formulations. The interaction with lactose monohydrate may be due to the secondary amine and/or sulfonamide structures of PZH undergoing the Maillard reaction [16,17]. The mechanism behind the interaction with PEG 6K and PEG 35K is unknown but a similar drug-excipient interaction has been described before [18].

Dissolution The pH-solubility profile of PZH displays an expected low solubility at higher pH levels that seems to be in line with its pKa values of 2.1 and 6.4 (US Food and Drug administration (FDA), 2009). The maximum solubility was observed around a pH of 3 below which the solubility decreases. The cause of this may be a common ion effect with Cl- ions. These may form apolar ion-pairs with the positively charged pazopanib in solution which causes crystallization [20].

As it is important to take the different regions of the gastrointestinal tract into consideration, the formulation needed to increase and maintain the solubility of PZH at both stomach and intestinal pH-values and compositions. To evaluate this during the in vitro testing of experimental formulations, a pH-switch was included after 30 min during the dissolution experiments. This was done to simulate the worst-case scenario of short residence time in a fasted stomach [21]. 152 | Chapter 3.1

The formulation components in Votrient® seem to only increase PZH solubility by a factor 1.5 at a low pH. No improvement was seen at higher pH levels The observed curves were in accordance with the pH solubility curve of PZH. According to the dissolution of Votrient® and pure PZH results, the bottleneck for the dissolution of PZH lies within its poor solubility at higher pH values. Therefore, the starting point of this study was to increase the solubility of PZH at pH 6.8. To simulate a worst-case-scenario where the crystalline drug is solubilized by an excipient at a pH in which it is very poorly soluble, the initial excipient-screening was performed at SIFsp with pH 6.8. The observed solubility increases in this set-up may be the result of non-complexing hydrophilic and/or hydrophobic interaction between the excipients and the drug molecules in solution [22,23]. The increase in solubility by Soluplus® between 2 and 24 h can possibly be caused by the relatively high viscosity of the solution, in comparison to the other solutions. This slows the process of wetting and dissolution to continue after the initial 2h.

The Kollidon excipients all increase the solubility of PZH in SGFsp, this was not seen for any of the Soluplus® formulations. This may be due to the time effect, as was seen in figure 3. The increase in PZH solubility in SGFsp could only be maintained into SIFsp if a ‘parachute’ effect of the excipients were to occur. After the initial solubility increase upon elevating the pH, the solubility of PZH in all of the Kollidon powder mixes collapsed. None of the Kollidon types were capable of providing the ‘parachute’ effect. This parachute effect was observed for the Soluplus® mixtures of 5 equivalent parts and up.

From the investigation of the influence of the ratio between PZH and Soluplus® equivalent parts in the formulation, it is clear that reducing this ratio below 8 equivalent parts strongly reduces the solubility of PZH The composition 1/8 was therefore chosen as the best performing formulation. Although this makes for a bulky dosage form, possible bioavailability improvement may correct by necessitating less intake of PZH, and hence, the formulation.

Characterization of the final formulation

With the Soluplus® formulation, initially an opalescent solution in SGFsp was observed, which is in accordance with the observations that maximally approximately 50% of PZH was dissolved. In the relatively short period of time at stomach pH, the Soluplus® does not seem to solubilize PZH to a significant extent above that of pure PZH at the studied concentrations. Thus, Soluplus® seems to be unable to overcome the common-ion effect. The formation of micelles by Soluplus® has been described in literature. As it may Solubility and bioavailability improvement of pazopanib hydrochloride | 153

provide an explanation for the solubilizing effect on PZH, micelles were studies by TEM and their size was determined.

An explanation for the behavior of the PZH and Soluplus® combination may be found in the pH-solubility relationship and the pKa values of PZH. At pH values below 2.1, the dissolved part of PZH is diprotonated and is likely not to be included in the micelles of Soluplus® because it can sustain itself in solution, this is supported by the fact that the micelle size in this region of pH is relatively small.

As the pH increases to 3.0, the solubility of unsolubilized PZH reaches a maximum. Addi- tionally, in this pH region between 2.1 and 6.4, PZH molecules are largely monoprotonated. It may be hypothesized that the PZH molecules in such a state interact very differently with the Soluplus® micelles, as was seen with the drug nilotinib [24]. The PZH may then partly be located in the outer, more hydrophilic ranges of the micelles as they need solubilization for the non-protonated part of the molecule which is then located in the more hydrophobic parts of the micelle. This may, in turn, change the overall size of the micelles. 3.1 As the solution equilibrium of PZH molecules is directed towards the Soluplus® micelles just above pH 3.0, more ‘room’ is made available in the medium for undissolved PZH to dissolve. This hypothesis is supported by the fact that additional PZH dissolved above pH 3.0 and that no new precipitation of crystalline PZH is observed in this pH region.

At pH levels above 6.4, PZH is largely unprotonated. Without solubilization, the drug molecules will precipitate virtually instantly from the dissolution medium. In the presence of Soluplus® PZH is fully incorporated into the hydrophobic parts of the micelles. To test this hypothesis, spectroscopic studies (with labeled PZH) should be performed in the future to localize PZH in the micelles. Additionally, possible release and/or transfer of PZH from and between micelles should be studied in the future to offer more clarity about the nature of the solubility increase. The clinical relevance of these phenomena is still unclear as the micelles seem to minimize to their stomach size at pH > 6. Therefore, in vivo investigation was crucial.

The filtration prior to HPLC analysis was performed using filters with a pore size of 200 nm. The increase in micelle diameter upon a change in pH is therefore not reflected in a change in the concentration of dissolved PZH. An increase in micelle diameter may prove troublesome for the uptake in vivo, however. Micelles with an increased diameter may be prone to agglomerate and precipitate at a much faster rate. 154 | Chapter 3.1

This increase in micelle diameter was observed to be more profound in the formulations with a smaller number of equivalent parts of Soluplus®. It may in part explain the observed loss of solubility at higher pH values of these formulations. To test whether the pH-dependent increase in micellar diameter was not related to the dissolution medium, a pH-switch system was designed that leads to a SIFsp with a carbonate buffer. The same trend in micelle size was observed in this system as well.

The XRD experiments showed that the pattern of PZH is consistent with patent data on the anhydrous polymorph Form I [25] and that the obtained pattern of PZHSol001 exhibits the same peaks as the pattern of PZH, albeit reduced in intensity due to the presence of Soluplus®. This indicates that no solid phase transitions have occurred during the production of PZHSol001.

Human in vivo clinical study The PK parameters of PZH after a single administration of 100 mg PZHSol001 in the first cohort were found to be similar to the reported parameters of 100 mg Votrient®. Al- though mean values differ between the two formulations, the CV-ranges overlap and drug absorption may, therefore, be comparable. It could be that the absolute amount of Soluplus® present from 100 mg is not sufficient to solubilize PZH in the volumes of the gastrointestinal tract. As an increased dose also increases the absolute amount of Soluplus®, the 2nd cohort dose was set at the maximum of 200 mg.

A more-than-linear increase in the PK parameters was observed for the 200 mg dose. An apparent difference is observed between the PK parameters of a single 200 mg dose of the formulations. 200 mg of PZHSol001 yields higher plasma concentrations (Cmax and

Cmin) and exposure than a similar dose of Votrient®. A strong relationship between CSS- min>20 μg/mL and tumor shrinkage and progression-free survival has been established [26]. Based on the results of the 200 mg cohort and an accumulation index of 2.41 (assuming a half-life of 31h and a dosing interval of 24h), the observed Cmin is not sufficient rd for clinical application (theoreticalSSmin C of 14.2). Therefore, the dose for the 3 cohort was increased to 300 mg. The PK parameters roughly doubled after the dose increase from 200 to 300 mg. Furthermore, the parameters were slightly higher than earlier data on Votrient® 800 mg. The estimated SSminC for 300 mg PZHSol001 was 27.2 μg/mL, which is well above the defined efficacy threshold. Solubility and bioavailability improvement of pazopanib hydrochloride | 155

The non-linear increase in PK parameters in relation to the dose might be caused by the concentration of Soluplus® present in the immediate vicinity of PZH in the gastrointestinal system. As was observed in vitro, the solubilizing effect of Soluplus® on PZH occurs in an apparent exponential fashion. Assuming that the stomach and intestinal dissolution volume in vivo were approximately constant in the study population, the linearly increasing concentration of Soluplus® facilitated a more-than-linear solubilizing effect. This means that although the composition of Soluplus® and PZH remained constant throughout the cohorts, the absolute amount of Soluplus® per volume may have been an important factor. This might have had a similar effect on the absorbed amount of PZH and consequently, the PK parameters. Furthermore, the formulation itself as en- capsulated loose powder may have contributed to an increased absorption by increasing the overall surface area for dissolution. This was also seen in a study that compared the PK of whole Votrient® tablets to crushed tablets [10].

An additional observation was the low variability in PK parameters of PZHSol001 compared to the parameters of Votrient®, albeit that the numbers are small. 3.1

Based on these results, 300 mg PZHSol001 might be a usable alternative to 800 mg Vot- rient®. Future studies need to be directed to the scale-up, tableting possibility (to reduce the bulk volume) and stability of the formulation. Special attention should there be directed to minimizing the formulation volume with regard to patient comfort and compliance.

Conclusions

In this study, we proved that the solubility of PZH can be significantly improved by formulation of the drug with the excipient Soluplus®. The optimal composition that produced full solubility in SIFsp was 1/8 PZH/Soluplus® (PZHSol001). In a clinical-proof-of-concept study, we demonstrated that the improved-solubility formulation PZHSol001 25 mg capsule in a dose of 300 mg produces a clinically useful pharmacokinetic profile. 156 | Chapter 3.1

References

[1] European Medicines Agency (EMA), Assess- [11] A.M. Pick, K.K. Nystrom, Pazopanib for the ment report Votrient (EPAR), (2011). treatment of metastatic renal cell carcinoma., [2] US Food and Drug administration (FDA), High- Clin. Ther. 34 (2012) 511–20. doi:10.1016/j. lights of prescribing information Votrient, clinthera.2012.01.014. (2012). [12] A.C.F. Rumondor, S.S. Dhareshwar, F. Kes- [3] US Food and Drug administration (FDA), Clin- isoglou, Amorphous Solid Dispersions or Prod- ical Pharmacology and Biopharmaceutics Re- rugs: Complementary Strategies to Increase view Votrient, (2009). Drug Absorption, J. Pharm. Sci. 105 (2016) [4] Australian Therapeutic Goods Administration, 2498–2508. doi:10.1016/j.xphs.2015.11.004. Australian Public Assessment Report (AUSPAR) [13] D.J. Berry, J.W. Steed, Pharmaceutical cocrys- Pazopanib, (2010). tals, salts and multicomponent systems; in- [5] Y. Deng, C. Sychterz, A.B. Suttle, M.M. Dar, termolecular interactions and property based D. Bershas, K. Negash, Y. Qian, E.P. Chen, P.D. design, Adv. Drug Deliv. Rev. 117 (2017) 3–24. Gorycki, M.Y.K. Ho, Bioavailability, metabolism doi:10.1016/j.addr.2017.03.003. and disposition of oral pazopanib in patients [14] B.J. Aungst, Optimizing Oral Bioavailability in with advanced cancer., Xenobiotica. 43 (2013) Drug Discovery: An Overview of Design and 443–53. doi:10.3109/00498254.2012.734642. Testing Strategies and Formulation Options, J. [6] H.I. Hurwitz, A. Dowlati, S. Saini, S. Savage, A.B. Pharm. Sci. 106 (2017) 921–929. doi:10.1016/j. Suttle, D.M. Gibson, J.P. Hodge, E.M. Merkle, L. xphs.2016.12.002. Pandite, Phase I trial of pazopanib in patients [15] M. Herbrink, N. de Vries, H. Rosing, A.D.R. Hu- with advanced cancer., Clin. Cancer Res. 15 itema, B. Nuijen, J.H.M. Schellens, J.H. Beijnen, (2009) 4220–7. doi:10.1158/1078-0432.CCR- Quantification of 11 Therapeutic Kinase Inhib- 08-2740. itors in Human Plasma for Therapeutic Drug [7] D. de Wit, N.P. van Erp, J. den Hartigh, R. Monitoring Using Liquid Chromatography Cou- Wolterbeek, M. den Hollander-van Deurs- pled With Tandem Mass Spectrometry., Ther. en, M. Labots, H.-J. Guchelaar, H.M. Verheul, Drug Monit. 38 (2016) 649–656. doi:10.1097/ H. Gelderblom, Therapeutic drug monitor- FTD.0000000000000349. ing to individualize the dosing of pazopanib: [16] D.D. Wirth, S.W. Baertschi, R.A. Johnson, S.R. a pharmacokinetic feasibility study., Ther. Maple, M.S. Miller, D.K. Hallenbeck, S.M. Drug Monit. 37 (2015) 331–8. doi:10.1097/ Gregg, Maillard reaction of lactose and fluox- FTD.0000000000000141. etine hydrochloride, a secondary amine., J. [8] H. Yu, N. van Erp, S. Bins, R.H.J. Mathijssen, Pharm. Sci. 87 (1998) 31–9. doi:10.1021/ J.H.M. Schellens, J.H. Beijnen, N. Steeghs, js9702067. A.D.R. Huitema, Development of a Pharma- [17] H.B. Sheth, V.A. Yaylayan, N.H. Low, M.E. Stiles, cokinetic Model to Describe the Complex Phar- P. Sporns, Reaction of reducing sugars with macokinetics of Pazopanib in Cancer Patients., sulfathiazole and importance of this reaction Clin. Pharmacokinet. 56 (2017) 293–303. to sulfonamide residue analysis using chro- doi:10.1007/s40262-016-0443-y. matographic, colorimetric, microbiological, or [9] M. Herbrink, B. Nuijen, J.H.M. Schellens, J.H. ELISA methods, J. Agric. Food Chem. 38 (1990) Beijnen, Variability in bioavailability of small 1125–1130. doi:10.1021/jf00094a047. molecular tyrosine kinase inhibitors, Cancer [18] C.E. Malan, M.M. de Villiers, A.P. Lötter, Ap- Treat. Rev. 41 (2015) 412–422. doi:10.1016/j. plication of differential scanning calorimetry ctrv.2015.03.005. and high performance liquid chromatography [10] E.I. Heath, K. Forman, L. Malburg, S. Gainer, to determine the effects of mixture composi- A.B. Suttle, L. Adams, H. Ball, P. LoRusso, A tion and preparation during the evaluation of phase I pharmacokinetic and safety evalua- niclosamide-excipient compatibility., J. Pharm. tion of oral pazopanib dosing administered Biomed. Anal. 15 (1997) 549–57. http://www. as crushed tablet or oral suspension in pa- ncbi.nlm.nih.gov/pubmed/8953499. tients with advanced solid tumors., Invest. [19] US Food and Drug administration (FDA), Chem- New Drugs. 30 (2012) 1566–74. doi:10.1007/ istry Review Votrient, (2009). s10637-011-9725-2. [20] S. Miyazaki, H. Inouse, T. Nadai, T. Arita, M. Na- Solubility and bioavailability improvement of pazopanib hydrochloride | 157

kano, Solubility characteristics of weak bases matoporphyrin, J. Phys. Chem. B. 116 (2012) and their hydrochloride salts in hydrochloric 7334–7341. doi:10.1021/jp300301z. acid solutions., Chem. Pharm. Bull. (Tokyo). 27 [24] M. Herbrink, J.H.M. Schellens, J.H. Beijnen, (1979) 1441–1447. doi:10.1248/cpb.27.1441. B. Nuijen, Improving the solubility of nilo- [21] D.M. Mudie, G.L. Amidon, G.E. Amidon, Phys- tinib through novel spray-dried solid disper- iological parameters for oral delivery and in sions, Int. J. Pharm. 529 (2017). doi:10.1016/j. vitro testing., Mol. Pharm. 7 (2010) 1388–405. ijpharm.2017.07.010. doi:10.1021/mp100149j. [25] S. Chava, S. Gorantla, V. Indukuri, An improved [22] S.S. Bansal, A.M. Kaushal, A.K. Bansal, Molec- process for the preparation of pazopanib or ular and Thermodynamic Aspects of Solubility a pharmaceutically acceptable salt thereof, Advantage from Solid Dispersions, Mol. Pharm. WO2015068175A2, 2014. 4 (2007) 794–802. doi:10.1021/mp7000796. [26] R.B. Verheijen, J.H. Beijnen, J.H.M. Schellens, [23] Y.-C. Li, S. Rissanen, M. Stepniewski, O. Cramar- A.D.R. Huitema, N. Steeghs, Clinical Phar- iuc, T. Róg, S. Mirza, H. Xhaard, M. Wytrwal, macokinetics and Pharmacodynamics of Pa- M. Kepczynski, A. Bunker, Study of Interac- zopanib: Towards Optimized Dosing., Clin. tion Between PEG Carrier and Three Relevant Pharmacokinet. (2017). doi:10.1007/s40262- Drug Molecules: Piroxicam, Paclitaxel, and He- 017-0510-z

3.1

Chapter 3.2

Improving the solubility of nilotinib through novel spray-dried solid dispersions

M. Herbrink1, J.H.M. Schellens2,3, J.H. Beijnen1,3, B. Nuijen1

Abstract

The tyrosine kinase inhibitor nilotinib has a very low aqueous solubility and a low and variable oral bioavailability. A pharmaceutical formulation with an improved solubility may enhance the bioavailability and reduce the variability thereof and of the pharmacokinetics. The aim of this study was to enhance the solubility of nilotinib by developing a spray dried solid dispersion. A broad selection of polymer excipients were tested for solubilizing properties. The spray drying technique was used to produce solid dispersions of nilotinib hydrochloride (NH) in matrices of the best performing polymers. Both the dissolution and physicochemical characteristics of the formulations were studied using a pH-switch dissolution model and conventional microscopic, thermal and spectrometric techniques. Of the tested spray dried solid dispersions, the ones containing the co-block polymer Soluplus® performed best in terms of in vitro dissolution properties. Further testing led to an optimized weight ratio of 1:7 (NH:Soluplus®) that improved the solubility up to 630-fold compared to crystalline NH (1.5 μg/mL) in simulated intestinal fluid. This effect can be attributed to the amorphization of NH and the solubilization of the drug due to micelle formation. A spray dried solid dispersion formulation of NH with Soluplus® in a ratio of 1:7 was developed that showed a significant increase in solubility. Improving the solubility of nilotinib through novel spray-dried solid dispersions | 161

Introduction

The tyrosine kinase inhibitor nilotinib (figure 1) is registered for the treatment of newly diagnosed adults with Philadelphia chromosome positive myeloid leukemia (Ph+ CML) in the chronic phase [1,2]. It is also indicated for the treatment of chronic and accelerated phase Ph+ CML in adult patients that are resistant or intolerant to prior therapy that included imatinib. The drug is marketed as Tasigna® since it received authorization in 2007 by both the European Medicines Agency (EMA) and the US Food and Drug Admin- istration (FDA). The recommended dose ranges between 300 and 400 mg twice daily and is administered orally [3].

O NH - + Cl HN N NH N F N F F 3.2

Figure 1. Chemical structure of nilotinib hydrochloride.

Nilotinib hydrochloride (NH) monohydrate is deemed poorly to moderately soluble [1]. Furthermore, the permeability of nilotinib is considered to be low in part due to efflux by P-glycoprotein (Pgp)-transporters in the gut wall [4]. The drug is thus classified in the biopharmaceutics classification system (BCS) as class IV [1]. This indicates that its absorption and bioavailability is hindered by solubility and permeability. NH is currently formulated as physical mixture with lactose, crospovidone and poloxamer 188 in an immediate release capsule. No study into the absolute oral bioavailability of the Tasigna® formulation has been performed hitherto. Based on mass balance studies, however, the bioavailability is estimated to be approximately 30% [2]. The intra- and interpatient variability in the pharmacokinetic parameters Cmax and area under the plasma-time curve (AUC) range between 32 and 72% [2,5]. Since bioavailability and drug exposure are tightly linked, Cmax and AUC are also influenced by poor solubility and permeability. Variation 162 | Chapter 3.2

in drug exposure may cause drug plasma levels to be inadequately low or toxically high [6]. Enhancing the solubility of NH monohydrate through improvement of its formulation may increase the bioavailability and possibly reduce pharmacokinetic variability.

Drug solubility enhancement can be achieved through the use of various techniques [7–12]. One of which is the application of the amorphous form of the drug in a formulation [13,14]. Amorphous drug material has higher Gibbs free energy than crystalline material, giving it a higher apparent solubility [15].

The amorphous form of a drug often has the tendency to revert back over time to the thermodynamically stable crystal form because of its excess entropy, possibly under influence of present impurities or moisture [16,17]. To stabilize such a system, the amorphous form is often dispersed through a polymer matrix, a so-called solid dispersion [18]. The amorphous form may either be molecularly dispersed or be scattered as non-molecular precipitates through the matrix, with a preference for the first as it favors the stability of the amorphous form [14].

Jesson et al. demonstrated that through enhancing the solubility of NH with a formulation in which amorphous nilotinib was non-molecularly dispersed through a matrix of hydroxypropyl methylcellulose phthalate (HP55) it was indeed possible to achieve an increased drug exposure (+730%) in male beagle dogs [19]. This research group produced nanoparticles of NH using the super critical fluid precipitation (SCP) method. Andersson et al. showed that this increase in bioavailability also takes place in humans (+76%) [20]. Colombo et al. used SCP to produce a nilotinib formulation with a pH-responsive polymer of which they extensively studied the dissolution properties [21].

The goal of this study was to develop an oral solid dosage form preferably containing a molecular dispersion, or solid solution, of NH in a matrix with a high solubility by fast, easy and economic means. A diverse selection of polymer excipients was screened for solubility-improving capabilities. The spray drying technique was used with the best performing excipients to produce dispersions of NH in polymer matrices. The physicochemical characteristics and the dissolution properties of the experimental formulations were examined. As it is important to take the different regions of the gastrointestinal tract into consideration, the formulation needed to retain the solubility of NH at both stomach and intestinal pH-values and compositions. To evaluate this during the in vitro testing of experimental formulations, a pH-switch was included after 30 min. during the dissolution Improving the solubility of nilotinib through novel spray-dried solid dispersions | 163

experiments. This was done to simulate the worst-case scenario of short residence time in a fasted stomach [22]. The micelle formation of Soluplus® was also investigated. The best performing formulation was also evaluated for long-term stability.

Materials and methods

Materials NH monohydrate was purchased from Avachem Scientific (San Antonio, TX, USA). Kolli- don VA64, Kollidon 12PF and polyethylene glycol (PEG), Lutrol F68, and Soluplus® were kindly supplied by BASF (Ludwigshafen, Germany). All other reagents were purchased in analytical grade.

Solubility testing and solvent selection for spray drying The solubility of NH was determined in water and the following cosolvent systems: 25, 50, 75, 100% (v/v) acetonitrile, ethanol, methanol, dimethylformamide (DMF), dimethylsulfoxide (DMSO), N-methylpyrrolidone (NMP); ethyl acetate; dichloromethane and methyl tert-butyl ether (MTBE). An excess of bulk drug was placed in a known vol- 3.2 ume of each system in a closed, light protected vial and allowed to saturate for 72 hours after which a sample was analyzed after filtration through an 0.45 um filter by HPLC. Subsequently, the usability of solvents with a high NH solvability (> 10 g/L) for spray drying was evaluated.

High performance liquid chromatography (HPLC) Nilotinib concentrations were determined using an Agilent 1100 series HPLC system consisting of a binary pump, Model G1312A, an autosampler, Model G1367A, a column oven, Model G1316A and a UV-detector Model G1314A (all from Agilent technologies, Amstelveen, The Netherlands). A reversed phase Gemini C18 analytical column (50 x 2.0 mm ID, 4 µm particle size, Phenomenex, Torrance, CA, USA) was used which was kept at 40 oC. Absorbance was measured at 265 nm. Injection of 10 µL was followed by a gradient from 60% to 90% methanol with 1 mM NH4OH. The flow rate was 0.4 mL/ min. Chromatograms were processed using Chromeleon software (Dionex Corporation, Sunnyvale, CA, USA). Stability indicating capability was established. 164 | Chapter 3.2

Preparation of formulations Physical mixtures (PM) of NH and polymers were prepared with mortar and pestle. To prepare solid dispersions (SD), NH and various polymers were dissolved in ethanol (HPLC grade) to ±10 g/L NH. The solutions were spray dried using a Büchi B-290 mini spray dryer equipped with an Inert Loop B-295, High Performance Cyclone, 1.5 mm nozzle cap and 0.7 mm nozzle tip (BUCHI Labortechnik AG, Flawil, Switzerland). Spray dry system settings: Spray feed: 10 mL/min;2 N atomization flow 40 mm (bottom of the ball); aspirator flow: 35 3m /h; inlet temperature 120 oC; outlet temperature 85 oC; inert loop temperature -20 oC.

Dissolution testing A collection of different excipients were efficiently screened on their possible solubility improving properties by suspending an excess of NH monohydrate in 10% polymer solutions with a pH of 6.8 for 2 and 48 hours. The dissolution behavior of various formulations was determined with a small-scale dissolution test. Briefly, an amount of powder equivalent to 10 mg NH was added to a 25 mL beaker containing 10 mL of medium. Temperature was kept at 37 oC and the medium was stirred at 200 rpm. The commercial formulation (Tasigna®) and the capsulated investigational formulations were tested in a larger dissolution volume of 200 mL. Dissolution tests were performed in single pH environment (simulated intestinal fluid (SGF), pH 6.8) and a pH-switch system. In this system, the 25 mL beaker contained 10 mL of simulated gastric fluid (SGF), pH 1.6. The sample was kept in this environment for 30 min. after which the pH and composition was changed into that of SIF in a step-wise manner by briefly adding phosphate or carbonate buffer. The switch of pH and the composition adjustment was typically completed within 32 min. Samples were taken at designated time points, filtrated using a 0.45 um filter and diluted 1/1 with methanol. All samples were subsequently analyzed using HPLC.

Micelle size measurements The presence and size of micelles was detected and measured using a Zetasizer Nano S90 particle size analyser (Malvern Instruments, Worcestershire, UK). Samples were taken from the dissolution media, placed in polystyrene cuvettes (Malvern Instruments) and analysed in threefold at 10 analyses per measurement.

Electron microscopy (EM) The morphology of the bulk and spray-dried powders were studied using Scanning Electron Microscopy (SEM). Samples were placed on conducting double sided adhesive tape and Improving the solubility of nilotinib through novel spray-dried solid dispersions | 165

on an aluminium holder. Samples were sputter-coated with gold using a SCD 040 sputter coater (Oerlikon, Balzers, Liechtenstein). Imaging was performed through back-scattering with a Sigma Field Emission Gemini system (Zeiss, Oberkochen, Germany).

The shape and size of micelles were visualized with Transmission Electron Microsco- py (TEM). Samples from dissolution media were diluted 100 times and were applied on Agar® formvar/carbon coated copper grids (van Loenen instruments, Zaandam, The Netherlands). The samples were negatively stained with 2% (w/w) phosphotungstic acid and were dried for 10 min. prior to the microscopy. The micrographs were recorded using a Tecnai T12 G2 Spirit Biotwin set-up (FEI company, Hillsboro, OR, USA).

Modulated differential scanning calorimetry (MDSC) MDSC measurements were performed with a discovery differential scanning calorimeter 3.2 (DSC) (TA Instruments, New Castle, DE, USA). Temperature scale and heat flow were calibrated with Indium. Samples of approximately 3 mg powder were weighed into Tzero aluminium pans (TA Instruments), sealed and placed in the autosampler. Each sample was equilibrated at 20 oC for 3 min, after which the sample was heated to 225 oC at a speed of 3 oC/min with a modulated temperature amplitude of 1 oC per period of 60 seconds.

X-ray powder diffraction (XRD) X-ray diffraction of powder samples was performed with an X’pert pro diffractometer equipped with an X-celerator (PANanalytical, Almelo, The Netherlands). Samples were placed in a 0.5 mm deep metal sample holder. Samples were scanned at a current of 30 mA and a tension of 40 kV. The scanning range was 10 – 60 degrees 2-θ, with a step size of 0.020 degrees 2-θ and a scanning speed of 0.002 degrees 2-θ per second.

Residual solvents Residual ethanol was determined with capillary gas chromatography (GC) analysis. Sam- ples of approximately 50 mg of powder were dissolved or suspended in 5.0 mL water and shaken for 3 hours. Aliquots were subsequently filtered and transferred to autosampler vials. Analyses were performed using 6890N GC system (Agilent, Santa Clara, CA, USA) 166 | Chapter 3.2

equipped with a Flame-ionization detector (Agilent) and a RTX-1301 capillary column (3.0 µm film, 30 m x 0.53 mm; Restek Corporation, Bellefonte, PA, USA).

Results and discussion

Solubility and spray dry settings In order to design an efficient spray drying protocol for NH, the solubility of the compound was assessed in various cosolvent systems. Figure 2 presents the relationship between solubility and the cosolvent percentage. The applicability of the cosolvent systems with the highest solubility (>10 g/L) in the spray drying process was investigated. Table 1. lists the minimum required settings for the various systems. The spray drying process proved to be inefficient when carried out with the high-boiling point solvents DMSO, DMF and NMP. The process with methanol and ethanol was found to be similar in terms of spray dryer settings. Ethanol was selected because it is the least toxic of the two solvents. The spray drying process was further optimized for ethanol so that residual ethanol was below 100 ppm. Bulk and spray-dried Soluplus® were compared (dissolution, FTIR and micelle size) and no significant differences were found that would indicate changed physicochemical properties. The spray-dried formulations were analysed by HPLC to investigate possible degradation of NH during spray drying or due to interaction with Soluplus®. No evidence of this was observed in any of the spray-dried compositions.

3 10 Ethanol Methanol

) 102 Acetonitrile Dimethylsulfoxide

(g/L Dimethylformamide 101 NMP NH

ed 0

lv 10

-1 Disso 10

10-2 050 100 Cosolvent % (v/v)

Figure 2. Equilibrium solubility of NH in various cosolvent systems. Improving the solubility of nilotinib through novel spray-dried solid dispersions | 167

Table 1: Spray dryer settings during applicability investigation of various cosolvent systems. Solvent system Inlet Outlet Limiting factor temperature temperature (oC) (oC) 75% DMF 155 140 Residual solvent present o Tout < 140 C: Deposition of solvent o Tout > 130 C: Polymer residue in drying column outlet 75% NMP 160 145 Residual solvent present Ethanol 95 55 None Methanol 90 50 Relatively toxic

o DMSO/DMF/NMP >160 >140 Tout < 150 C: Deposition of solvent o Tout > 130 C: Polymer residue in drying column outlet Required low feed flow rate

Dissolution behaviour and formulation selection The dissolution profile of the marketed formulation Tasigna® was investigated. Figure 3A presents the dissolution curves (n = 3) of both the Tasigna® capsule and its decapsulated powder. The first 30 min. of the curve represent the stomach (pH 1.6) in which the percentage of dissolved NH adds up to approximately 26%. At time points 35, 40 and 45 3.2 min. a relatively small decrease to 23% in dissolved drug is seen when the pH increases from 1.6 to 3.2. From 50 min. on the dissolved amount decreases to less than 1% when the pH is elevated above 5.2 and is subsequently adjusted to 6.8. This behaviour agrees with the pH-solubility curve (figure 3B) and the observations by Jesson et al [19]. The solubility of NH at higher pH values can be considered its bottleneck. Therefore, improving the drug’s solubility at pH 6.8 was taken as a starting point in this study.

30 AB3000

Powder ) ) mL (% g/

20 m 2000

Capsule ( NH NH ed

10 lv 1000 Dissolved Disso

0 0 0204060 02468 Time (min) pH

Figure 3. A. Dissolution curves of Tasigna® capsules and decapsulated Tasigna® powder; B. pH-solubility curve of NH. The dotted line indicates the start of the pH-switch. 168 | Chapter 3.2

Figure 4 shows the results from the excipient screening experiments after both 2 and 24 hours. This manner of solubility testing is based on the worst-case-scenario where crystalline drug is solubilized by an excipient at a pH in which it is very poorly soluble. NH monohydrate has a solubility of approximately 1.5 μg/mL in this buffer system. The tests show that Kollidon VA64, 12PF, Soluplus® and Lutrol F68 display the highest maintaining increase in solubility. For this reason, these excipients were selected for further formulation development. The increase in solubility may be due to non-complexing hydrophilic and/or hydrophobic interaction between the excipients and the drug[23,24].

1000 ) mL

g/ 100 m ( NH ved

ol 10 ss Di

1 F s 64 NH A F68 90F V 12P uplu 6000 l ol 35000 on on on So tr Lu PEG llid llid llid PEG o o o K K K

Figure 4. Excipient screening results: NH solubility after 2 and 24 hours.

The increase in solubility by Soluplus® between 2 and 24 hours is possibly the result of the relative high viscosity of the solution, in comparison to the other solutions. This slows the process of wetting and dissolution to continue after the initial 2 hours.

With the above four excipients both PM and SD formulations with different ratios between drug and excipient were produced. Table 2 lists these formulations. The dissolution behaviour of the formulations was tested using the pH-switch system. The results are presented in figure 5. Figure 5A shows the dissolution curves of crystalline and amorphous NH. Both forms have very different solubilities at stomach pH (38% against 83%) but show a very similar decrease in solubility when the pH is increased towards 6.8. The amorphous drug structure has an increased initial stomach solubility but still needs a ‘parachute’ in the intestinal medium to prevent recrystallization. The PMs (figure 5B-E) all exhibited similar dissolution properties as the crystalline drug in SGF, the resulting Improving the solubility of nilotinib through novel spray-dried solid dispersions | 169

polymer concentrations in the dissolution medium of the tested PMs are likely too low to aid the dissolution and increase the solubility of NH in SGF. The desired ‘parachute’ effect was only observed with the Soluplus® formulations.

PM5 PM 5 100 A B SD 5 C SD 5 SD NH 100 100 PM10 PM 10 NH.H2O SD10 SD 10 %) %) %) H( H( H( dN dN dN 50 50 50 ve ve ve ol ol ol ss ss ss Di Di Di

0 0 0 0204060 0204060 0204060 Time (min) Time (min) Time (min)

D PM 5 E 100 100 SD 5 SD 5 PM 10 SD 10 SD 10 PM 10 PM 5 %) %) H( H( dN dN 50 50 ve ve ol ol ss ss Di Di

0 0 0204060 0204060 Time (min) Time (min) 3.2 Figure 5. A. Dissolution curve of crystalline NH monohydrate and spray dried NH. Dissolution curves of the physical mixtures and spray dried formulations with B. Kollidon VA64; C. Kollidon 12PF; D. Soluplus®; E. Lutrol 68F. The dotted line indicates the start of the pH-switch.

The dissolution curves of the SD formulations show an increased solubility in SGF, indicating the enhanced apparent solubility of the amorphous drug with 100% (1 mg/mL) solubility for both Soluplus® and Kollidon VA64. During the transition from pH 1.2 to 6.8, Lutrol F68 and Kollidon 12PF were unable to maintain an increased solubility in both compositions. Soluplus® SD 1/5 and 1/10 and Kollidon VA64 1/10 did offer a ‘parachute’ effect of which Soluplus® 1/10 showed the best performance. Based on these results, Soluplus® was chosen for further formulation development.

SDs with 5 different NH/Soluplus® compositions were prepared and subjected to dissolution experiments to further investigate the influence of Soluplus® on the solubility of NH. The results are presented in figure 6A. Figure 6B shows the percentage dissolved NH as a function of the Soluplus® equivalent parts. From this apparent relationship, it can be concluded that a higher number of equivalent parts of Soluplus® within pharmaceutical relevant volumes would not further increase the solubility of NH at pH 6.8. The composi- 170 | Chapter 3.2

tion 1/7 was selected as the best performing formulation as it offers the highest solubility at pH 6.8 with the lowest total formulation volume. The 1/7 composition may make for a bulky dosage form when the same dose of NH is needed. An in vivo study should point out whether the formulation increases the bioavailability of NH. In that case, less of the formulation is required to reach adequate drug plasma levels and administration of large amounts are not needed.

Table 2: Components, weight ratios, and preparation methods of NH formulations

Formulation Components Weight ratio (w/w) Formulation method

A Crystalline NH monohydrate 1/0 Pure drug

B Amorphous NH 1/0 Spray dying

C Crystalline nilotinib*, Kollidon VA64 1/5 Physical mixing

D Crystalline nilotinib*, Kollidon VA64 1/5 Spray dying

E Crystalline nilotinib*, Kollidon 12PF 1/5 Physical mixing

F Crystalline nilotinib*, Kollidon 12PF 1/5 Spray dying

G Crystalline nilotinib*, Soluplus® 1/5 Physical mixing

H Crystalline nilotinib*, Soluplus® 1/5 Spray dying

I Crystalline nilotinib*, Lutrol F68 1/5 Physical mixing

J Crystalline nilotinib*, Lutrol F68 1/5 Spray dying

K Crystalline nilotinib*, Kollidon VA64 1/10 Physical mixing

L Crystalline nilotinib*, Kollidon VA64 1/10 Spray dying

M Crystalline nilotinib*, Kollidon 12PF 1/10 Physical mixing

N Crystalline nilotinib*, Kollidon 12PF 1/10 Spray dying

O Crystalline nilotinib*, Soluplus® 1/10 Physical mixing

P Crystalline nilotinib*, Soluplus® 1/10 Spray dying

Q Crystalline nilotinib*, Lutrol F68 1/10 Physical mixing

R Crystalline nilotinib*, Lutrol F68 1/10 Spray dying

*as hydrochloride monohydrate Improving the solubility of nilotinib through novel spray-dried solid dispersions | 171

AB100 100 W V U 80 %) %) H( H( 60 dN dN 50 T ve ve ol ol 40 ss ss Di Di 20

0 S 0 0204060 0510 Time (min) Equivalent parts SLP (-)

Figure 6. Dissolution results of NH:Soluplus® formulations: A. Dissolution curves of NH:Soluplus® SDs with 5 compositions: S, T, U, V and W; B. Dissolved NH at pH 6.8 as a function of the Soluplus® equivalent parts.

Micelle size measurements As figure 6A demonstrates, the solubility of NH permanently sharply collapses for formulation S during the pH increase. The other formulations all show a decrease in solubility during the switch that is related to the Soluplus® equivalent parts, after which the solubility increases to a seemingly constant level. The drop in solubility coincided with the solutions turning opaque. 3.2

At different time points during the dissolution of formulations S to W, samples were taken to measure the micelle diameters of which figures 7A and 7B present the results. The observed opacity during the solubility drop can be linked to the sudden increase of the micelle diameter. The large micelles (>450 nm) are filtered out and the incorporated nilotinib within these micelles is subsequently not quantified by HPLC. This clarifies the apparent ‘drop’ in solubility. 172 | Chapter 3.2

ABPhosphate Carbonate S T

m) 1000 U (n

e V

erag W av z SLP 100

246 246 pH pH C D

200 nm 200 nm

Figure 7. Micelle diameters as a function of the pH of the dissolution medium: A. Phosphate SIFsp; B. Car-

bonate SIFsp. C. TEM micrograph of a representative NH:Soluplus® micelle in SGFsp (Magnification: 120000

x, high voltage: 100 kV); D. TEM micrograph of NH:Soluplus® micelles in SIFsp (Magnification: 120000 x, high voltage: 100 kV).

To test whether the phenomenon is not related to the dissolution medium, a pH-switch system was designed that leads to a SIFsp with a carbonate buffer. The same pattern was observed as with the phosphate system although the micelle diameters in the carbonate system were generally larger. This distinction is important, since the intestinal medium in vivo is a carbonate buffered system instead of a phosphate buffered one. Despite this difference in size, the micelle for formulations U,V and W regained a size of < 100 nm at pH > 6. Figures 7A and B also present the size of Soluplus® micelles without NH where the pH-dependent increase does not occur. An explanation for the behaviour of the NH/

Soluplus® micelles might come from the pKa values of NH, which are 2.1 and 5.4 [25,26] (dotted lines in figure 7). At pH values below 2.1 NH is diprotonated enabling a relatively high solubility. A relative large fraction of NH molecules in such an environment remain in solution, instead of being incorporated into Soluplus® micelles. This keeps the micelles small. In the pH region between 2.1 and 5.4, NH molecules are largely monoprotonated. It may be hypothesized that the NH molecules in such a state interact very differently with the Soluplus® micelles. The NH may then in part be located in the outer, more hydrophilic ranges of the micelles but still need solubilisation for the non-protonated part Improving the solubility of nilotinib through novel spray-dried solid dispersions | 173

of the molecule. This may in turn change the overall size of the micelles. At pH levels above 5.4, NH is largely unprotonated. Without solubilisation the NH molecules will precipitate from the dissolution medium. In the presence of Soluplus® NH is incorporated into the micelles. To test this hypothesis, spectroscopic studies should be performed in the future to localize NH in the micelles.

The size of the resulting micelles is also strongly dependent on the equivalent parts of Soluplus® that are present, as this determines the initial number of micelles over which the NH molecules can divide. The chloride counter-ion from NH may also have an impact on the micelle size through its influence on the ionic strength of the solution [27].

The clinical relevance of this phenomenon is still unclear as the micelles seem to minimize to their stomach size at pH > 6. Therefore, the formulation should be studied in vivo.

Various other parameters also influence micelle size, i.e. drug molecular size, partition coefficient and the presence of lipids and bile salts. In certain combinations witha pK ’s of around 2 and 6-8 may cause other drugs to behave similarly. 3.2

The dissolution and micelle size results from this study highlight unexplored possible characteristics of Soluplus®-containing formulations. Previous studies with spray-dried Soluplus® formulations did not employ pH-switches during dissolution and thus make no mention of micelle size changes [28,29]. The results presented here indicate that the application of a pH-switch may prove useful in uncovering possible unexpected dissolution behaviour of new formulations.

Electron microscopy (EM)

Figure 8A and 8B present the SEM micrographs of NH.H2O bulk powder and formulation V, respectively. Figure 8A clearly show the presence of large crystalline particles in the bulk powder. Additionally, figure 8B shows that the spray-dried formulation consists of smaller particles than the bulk powder, which likely plays an important role in the improved dissolution performance of formulation V.

To study the in vitro assembled micelles during the dissolution of the Soluplus® containing formulations, TEM was performed. Figure 7 C and D show a micrograph of NH/

Soluplus® micelles of formulation in SGFsp and SIFsp, respectively. The micelles are have 174 | Chapter 3.2

maintained their integrity, during the transition from SGFsp to SIFsp. Figure 7 C and D also show that the size in SGFsp and SIFsp are similar. The micelles in the pH region 3-5 could not be successfully be visualized in this TEM set-up. The size of the micelles changed after addition of the phosphotungstic acid, likely due to the limited buffer capacity in this pH-region.

Fourier transform infrared spectroscopy Crystalline NH monohydrate, amorphous NH and the spray dried formulations S through W (table 3) with Soluplus® were physically examined using FTIR (figure 8C). The molecular structure of nilotinib contains heterocyclic systems with a trifluormethyl moiety that interconnect through amine bridges resulting in characteristic absorption bands. The spectrum of formulation S (1:1 ratio) did not show any peak changes nor did it show any appearance or disappearance of characteristic peaks. This suggests that NH and Soluplus® do not suffer from detrimental intermolecular forces. The spray drying process removed the water of crystallization and produced anhydrous amorphous NH. This is evident from the FTIR spectra which lack signals corresponding to water at around 2000 cm-1 and 3200 cm-1. Karl Fischer titration results are consistent with these results and show < 0.3% water content initially.

Modulated differential scanning calorimetry (MDSC) MDSC was used to analyse the solid state characteristics of NH, Soluplus® and the formulations S to W. Improving the solubility of nilotinib through novel spray-dried solid dispersions | 175

3.2

Figure 8. Physicochemical characterization of NH (monohydrate) and formulations: A. SEM micrograph of Figure 8. Physicochemical characterization of NH (monohydrate) and formulations: A. SEM NH.H O bulk powder (Magnification: 500x, extra high tension: 2.00 kV); B. SEM micrograph of formulation V 2micrograph of NH.H O bulk powder (Magnification: 500x, extra high tension: 2.00 kV); B. SEM (Magnification: 500x, extra2 high tension: 2.00 kV); C. FTIR spectra of NH monohydrate , NH SD, formulations micrograph of formulation V (Magnification: 500x, extra high tension: 2.00 kV); C. FTIR spectra of NH S, T, U, V and W and Soluplus®; D. DSC results of NH and NH monohydrate; E. MDSC results of NH:Soluplus® monohydrate , NH SD, formulations S, T, U, V and W and Soluplus®; D. DSC results of NH and NH SD formulations S, T, U, V and W; F. XRD patterns of NH monohydrate, NH:Soluplus® physical mixture 1:7, NH SD monohydrate;and NH:Soluplus® E. MDSC formulation results V,of the NH:Soluplus® peaks around SD 43° formulations 2-θ originate S, fromT, U, theV and sample W; F. holder. XRD patterns of NH monohydrate, NH:Soluplus® physical mixture 1:7, NH SD and NH:Soluplus® formulation V, the peaks around 43° 2-θ originate from the sample holder.

138

176 | Chapter 3.2

The MDSC thermogram of crystalline NH monohydrate in figure 8D shows two clear endothermic events at 133 oC and 201 oC. The first indicates the loss of water of crystallisation and the transition to the anhydrate form. This conversion to the anhydrate before melting is described in a patent text [30]. The latter event corresponds to the melting point observed during analysis in the melting apparatus. As a comparison, figure 8D also shows a thermogram of crystalline NH anhydrate where the water of crystallization peak is not present. The MDSC spectrum in figure 8E of spray dried NH shows a glass o transition temperature (Tg) at 146 C and no other thermal events up to degradation at 230 oC. This supports the observations from the FTIR experiments. Pure and spray dried o o Soluplus® have a Tg as the only thermal event at 68 C that corresponds to the 70 C reported by BASF [31]. The spray dried formulations all show a single Tg and no melting event. This indicates that NH is molecularly dispersed in the spray dried formulations.

The trend in Tg’s is in line with the compositions of the formulations as described by formula 1 and listed in table 3. These results aid in the conclusion that NH is molecularly dispersed through the matrix of Soluplus®.

(1) ! %& %( "# = "#& + "#( Where wx is the weight fraction of component x and Tgx is the glass transition temperature of component x.

Table 3. Weight ratios and Tg’s of spray dried soluplus®-NH formulations o Formulation NH monohydrate : Soluplus® Tg ( C) S 1 : 1 95.3 T 1 : 3 80.5 U 1 : 5 75.9 V 1 : 7 74.2 W 1 : 10 73.7

Powder X-ray diffraction XRD patterns were recorded to establish the solid state of the prepared spray dried dispersions. Figure 8F shows the XRD patterns for crystalline NH monohydrate, NH:So- luplus® physical mixture (1:7) and spray dried NH and formulation V. The XRD pattern of crystalline NH monohydrate shows the characteristic sharp refraction peaks of crystallinity which have completely disappeared in the pattern of spray dried NH. The peaks of crystalline material are also present in the physical mixtures of NH monohydrate and Improving the solubility of nilotinib through novel spray-dried solid dispersions | 177

Soluplus®. The XRD patterns of the spray dried formulations lacked the sharp peaks as well. From the results of the physical characterization experiments it can be concluded that the raw NH monohydrate is crystalline and that it amorphicizes through the process of spray-drying and that the prepared solid dispersion also contains amorphous NH. This corresponds with the findings during the MDSC experiments, where no melting event was observed and only a single Tg was recorded for the formulations S to W.

Stability Quality control testing of the 1:7 Soluplus® formulation showed near-complete dissolution in both USP SIFsp and SGFsp. SIFsp was chosen for stability tests because a decline in any of the formulations functional parameters (amorphicity, purity, etc.) is most likely to be noticed first in this medium because of the low intrinsic solubility of NH at SIFsp pH. During 6 months of storage in open containers at 20-25 oC and 60% relative humidity (RH) and after 1 month in open containers at 40o C and 75% RH the formulation was subjected again to dissolution and assay tests (table 4). In this period, no significant changes occurred in chemical or physical properties of the formulation. The moisture content did show an increase, this did not prove detrimental to the performance of the formulation. This indicative stability study should be extended in the future to map a conclusive sta- 3.2 bility profile of the formulation.

Table 4. Stability results of NH : Soluplus® (1:7) spray dried powder Initial 6 months at 20-25 1 month at 40 oC/60% RH, dark, oC/75% RH, dark, open container open container Nilotinib peak purity (%) 100 99.6 99.7 Nilotinib dissolved at t = 45 min (%)a 95.2 (3.7) 94.0 (5.6) 94.1 (3.2) Nilotinib dissolved at t = 120 min (%)a,b 94.9 (4.1) 93.8 (3.1) 94.0 (6.8) Nilotinib dissolved at t = 240 min (%)a,b 94.6 (2.4) 93.9 (4.0) 93.5 (5.7) XRD spectrum Fully amorphous Fully amorphous Fully amorphous Moisture (%) w/w 0.28 0.77 0.54 aValues are means and coefficients of variation. bTime points were included to detect possible recrystallization of nilotinib HCl from the solution. 178 | Chapter 3.2

Conclusion

This study shows that spray dried solid dispersions of NH are very effective in improving the low solubility of the NH in SGFsp and SIFsp compared to the marketed Tasigna® formulation. SDs containing Kollidon VA64 and Soluplus® were also effective in maintaining an increased solubility after the simulated pH switch from SGFsp to SIFsp. Of the polymer excipients tested, Soluplus® exhibited the best performance. The ratio 1:7 (NH:Soluplus®) was the lowest ratio that showed the highest performance and was thus developed further. Stability of the latter formulation was established for at least 6 months. This study further presents characteristic micelle behavior of the NH:Soluplus® combinations when exposed to aqueous environments at different pH levels that may be relevant for other compounds in combination with Soluplus®. Hence, Soluplus® in a ratio of 7:1 with NH demonstrates a great increase in solubility of NH. Whether this leads to a higher bioavailability and a possibly reduced variability needs to be tested in vivo. Although a 1:7 ratio would make for a bulky tablet or capsule, the increase in bioavailability might enable smaller doses to be administered which could reduce the final formulation size. Improving the solubility of nilotinib through novel spray-dried solid dispersions | 179

References

[1] US Food and Drug administration (FDA), Clin- niques, ISRN Pharm. 2012 (2012) 1–10. [12] ical pharmacology and biopharmaceutics re- J.M. Keen, J.W. McGinity, R.O. Williams, En- view Tasigna, (2007). hancing bioavailability through thermal pro- [2] European Medicines Agency (EMA), European cessing., Int. J. Pharm. 450 (2013) 185–96. public asessment report Tasigna, (2007). [13] A. Newman, G. Knipp, G. Zografi, Assessing the [3] US Food and Drug administration (FDA), Pre- performance of amorphous solid dispersions., scribing information Tasigna(R), 2010. J. Pharm. Sci. 101 (2012) 1355–77. [4] S. Shukla, A.P. Skoumbourdis, M.J. Walsh, [14] S. Janssens, G. Van den Mooter, Review: phys- A.M.S. Hartz, K.L. Fung, C.-P. Wu, M.M. Gottes- ical chemistry of solid dispersions., J. Pharm. man, B. Bauer, C.J. Thomas, S. V Ambudkar, Pharmacol. 61 (2009) 1571–86. Synthesis and characterization of a BODIPY [15] T. Einfalt, O. Planinsek, K. Hrovat, Methods of conjugate of the BCR-ABL kinase inhibitor amorphization and investigation of theamor- Tasigna (nilotinib): evidence for transport phous state, Acta Pharm. 63 (2013) 305–334. of Tasigna and its fluorescent derivative by doi:102478/acph-2013-0026. ABC drug transporters., Mol. Pharm. 8 (2011) [16] L. Yu, Amorphous pharmaceutical solids: 1292–302. preparation, characterization and stabilization, [5] C. Tanaka, O.Q.P. Yin, V. Sethuraman, T. Smith, Adv. Drug Deliv. Rev. 48 (2001) 27–42. X. Wang, K. Grouss, H. Kantarjian, F. Giles, O.G. [17] Q. Viel, C. Brandel, Y. Cartigny, M.E.S. Eusébio, Ottmann, L. Galitz, H. Schran, Clinical- Phar J. Canotilho, V. Dupray, E. Dargent, G. Coquer- macokinetics of the BCR–ABL Tyrosine Kinase el, S. Petit, Crystallization from the Amorphous Inhibitor Nilotinib, Clin. Pharmacol. Ther. 87 State of a Pharmaceutical Compound: Impact (2010) 197–203. doi:10.1038/clpt.2009.208. of Chirality and Chemical Purity, Cryst. Growth [6] M. Herbrink, B. Nuijen, J.H.M. Schellens, J.H. Des. 17 (2017) 337–346. doi:10.1021/acs. Beijnen, Variability in bioavailability of small cgd.6b01566. molecular tyrosine kinase inhibitors, Cancer [18] W.L. Chiou, S. Riegelman, Pharmaceutical 3.2 Treat. Rev. 41 (2015) 412–422. applications of solid dispersion systems., J. [7] B. Singh, S. Bandopadhyay, R. Kapil, R. Singh, O. Pharm. Sci. 60 (1971) 1281–302. Katare, Self-emulsifying drug delivery systems [19] G. Jesson, M. Brisander, P. Andersson, M. (SEDDS): formulation development, character- Demirbüker, H. Derand, H. Lennernäs, M. ization, and applications., Crit. Rev. Ther. Drug Malmsten, Carbon Dioxide-Mediated Genera- Carrier Syst. 26 (2009) 427–521. tion of Hybrid Nanoparticles for Improved Bio- [8] S. Lohani, H. Cooper, X. Jin, B.P. Nissley, K. availability of Protein Kinase Inhibitors, Pharm. Manser, L.H. Rakes, J.J. Cummings, S.E. Fau- Res. 31 (2014) 694–705. ty, A. Bak, Physicochemical Properties, Form, [20] P. Andersson, M. Von Euler, M. Beckert, Com- and Formulation Selection Strategy for a Bio- parable pharmacokineties of 85 mg RightSize pharmaceutical Classification System Class II nilotinib (XS003) and 150 mg Tasigna in healthy Preclinical Drug Candidate., J. Pharm. Sci. 103 volunteers using a hybrid nariopartick-based (2014) 3007–21. formulation platform for protein kinase inhib- [9] M.A. Rahman, R. Harwansh, M.A. Mirza, S. itors, in: 2014 ASCO Annu. Meet., 2014. Hussain, A. Hussain, Oral lipid based drug de- [21] S. Colombo, M. Brisander, J. Haglöf, P. Sjövall, livery system (LBDDS): formulation, character- P. Andersson, J. Østergaard, M. Malmsten, ization and application: a review., Curr. Drug Matrix effects in nilotinib formulations with Deliv. 8 (2011) 330–45. pH-responsive polymer produced by car- [10] Y. Kawabata, K. Wada, M. Nakatani, S. Yamada, bon dioxide-mediated precipitation., Int. J. S. Onoue, Formulation design for poorly wa- Pharm. 494 (2015) 205–17. doi:10.1016/j. ter-soluble drugs based on biopharmaceutics ijpharm.2015.08.031. classification system: Basic approaches and [22] D.M. Mudie, G.L. Amidon, G.E. Amidon, Phys- practical applications, Int. J. Pharm. 420 (2011) iological parameters for oral delivery and in 1–10. vitro testing., Mol. Pharm. 7 (2010) 1388–405. [11] K.T. Savjani, A.K. Gajjar, J.K. Savjani, Drug Sol- doi:10.1021/mp100149j. ubility: Importance and Enhancement Tech- [23] S.S. Bansal, A.M. Kaushal, A.K. Bansal, Molec- 180 | Chapter 3.2

ular and Thermodynamic Aspects of Solubility Dev. Technol. (2016) 1–14. Advantage from Solid Dispersions, Mol. Pharm. [28] Z.M.M. Lavra, D. Pereira de Santana, M.I. Ré, 4 (2007) 794–802. Solubility and dissolution performances of [24] Y.-C. Li, S. Rissanen, M. Stepniewski, O. Cramar- spray-dried solid dispersion of Efavirenz in iuc, T. Róg, S. Mirza, H. Xhaard, M. Wytrwal, Soluplus, Drug Dev. Ind. Pharm. 43 (2017) 42– M. Kepczynski, A. Bunker, Study of Interac- 54. doi:10.1080/03639045.2016.1205598. tion Between PEG Carrier and Three Relevant [29] E.-S. Ha, I. Baek, W. Cho, S.-J. Hwang, M.-S. Kim, Drug Molecules: Piroxicam, Paclitaxel, and He- Preparation and evaluation of solid dispersion matoporphyrin, J. Phys. Chem. B. 116 (2012) of atorvastatin calcium with Soluplus® by spray 7334–7341. drying technique., Chem. Pharm. Bull. (Tokyo). [25] European Medicines Agency (EMA), Assess- 62 (2014) 545–51. ment report Tasigna (EPAR), (2009). [30] P.W. Manley, W.-C. Shieh, P.A. Sutton, P.H. [26] US Food and Drug administration (FDA), Chem- Karpinski, Crystalline forms of 4-methyl-n-[3- istry Review Tasigna, (2007). (4-methyl-imidazol-1-yl)-5-trifluoromethyl- [27] N.-Q. Shi, H.-W. Lai, Y. Zhang, B. Feng, X. Xiao, phenyl]-3-(4-pyridin-3-yl-pyrimidin-2-ylami- H.-M. Zhang, Z.-Q. Li, X.-R. Qi, On the inherent no)-benzamide, WO2007015870A2, 2005. properties of Soluplus and its application in [31] K. Kolter, M. Karl, A. Gryczke, Hot-Melt Extru- ibuprofen solid dispersions generated by mi- sion with BASF Pharma Polymers, 2nd ed., crowave-quench cooling technology, Pharm. BASF SE, Ludwigshafen, 2012.

Chapter 4

Bioanalysis of oral anticancer agents

Chapter 4.1

Quantification of 11 therapeutic kinase inhibitors in human plasma for therapeutic drug monitoring using liquid chromatography coupled with tandem mass spectrometry

M. Herbrink1, N. de Vries1, H. Rosing1, A.D.R. Huitema1,2, B. Nuijen1, J.H.M. Schellens3,4, J.H. Beijnen1,4

Abstract

A liquid chromatography/tandem mass spectrometry assay was developed to facilitate therapeutic drug monitoring (TDM) for 9 anticancer compounds (dasatinib,- er lotinib, gefitinib, imatinib, lapatinib, nilotinib, pazopanib, sorafenib, sunitinib,- andve murafenib) and the active metabolite, N-desethyl-sunitinib. The TDM assay is based on reversed-phase chromatography coupled with tandem mass spectrometry in the positive ion mode using multiple-reaction monitoring for analyte quantification. Stable isotopically labeled compounds were used as internal standards. The sample pretreatment consisted of protein precipitation with acetonitrile using a small plasma volume of 50 µL. The validation procedures were based on the guidelines on bioanalytical methods issued by the US Food and Drug Administration (FDA), and were modified to fit the requirements of the clinical TDM environment. The method was validated over a linear range of 5.00-100 ng/mL for dasatinib, sunitinib, and N-desethyl-sunitinib; 50.0-1000 ng/mL for gefitinib and lapatinib; 125-2,500 ng/mL for erlotinib, imatinib, and nilotinib; and 500-10,000 ng/mL for pazopanib, sorafenib, and vemurafenib. The results of the validation study demonstrated good intra- and inter-assay accuracy (bias<6.0%) and precision (12.2%) for all analytes. This newly validated method met the criteria for TDM and has successfully been applied to routine TDM service for tyrosine kinase inhibitors. Bioanalysis of 11 kinase inhibitors | 187

Introduction

Therapeutic drug monitoring (TDM) of anticancer drugs is becoming an important tool in the targeted treatment of patients with cancer, and tyrosine kinase inhibitors (TKIs) are among the class of targeted drug therapies. Almost all drugs in this rapidly growing group are subject to high pharmacokinetic variability. TDM has been increasingly useful in managing the considerable interindividual variability that is observed in the pharmacokinetics of these drugs [1–7]. Variations in absorption, distribution, metabolism, and excretion in and between individuals may be because of food intake, concomitant medication, underlying disease, age, genetics, and other factors [8].Consequently, drug levels vary considerably and this may lead to insufficient efficacy or substantial toxicity. TDM aims to individualize drug dosing by focusing on balancing the therapeutic efficacy and the avoidance of drug toxicity [9,10]. This is achieved by quantifying drug concentrations in patient blood plasma or serum and by comparing the results with predetermined guidelines and target levels.

Guidelines for the TDM of TKIs are sparse in the literature; however, previous literature reviews by various authors have provided information on pharmacokinetic targets that may be used in the TDM practice [11–13]. A rapidly performed, robust, and adequately ranged TDM assay for a large variety of frequently used TKIs is not currently available in routine clinical practice. Therefore, to meet this need, we developed and validated a bio- 4.1 analytical assay devoted to the TDM of dasatinib, erlotinib, gefitinib, imatinib, lapatinib, nilotinib, pazopanib, sorafenib, sunitinib, N-desethyl-sunitinib, and vemurafenib. The validation procedures were designed according to the US Food and Drug Administration (FDA) Guidelines [14,15], with the following modifications. 1) Four non-zero calibrators were used in the calibration curve instead of six to eight. 2) The quality control low concentration (QCL) sample was replaced with a QC lower limit of quantification (LLOQ) sample for the determination of the inaccuracy and imprecision. This allowed for the development of a method that is routinely useable in the time-limited clinical practice while still offering a bioanalytical validation approach. This article describes the validation and application of the quantitative analysis of 10 TKIs and a metabolite in human plasma for TDM purposes. After protein precipitation, the samples were analyzed using a ultra-performance liquid chromatography (UPLC) system equipped with a high-performance liquid chromatography (HPLC) column with tandem mass spectrometry (MS/MS) detection. 188 | Chapter 4.1

This method was based on an assay that was developed by our laboratory for the quantitative analysis of dasatinib, erlotinib, gefitinib, imatinib, lapatinib, nilotinib, sorafenib, and sunitinib in human plasma [16], and was adapted to quantify additional analytes. Additional changes were made to allow for a shorter analysis time as well as its use in a TDM setting. The validated ranges of the method were chosen to include the mean observed Ctrough levels of the analytes, which Yu et al. found to be higher than the suggested

Cthrough levels for most TKIs [11]. This assured a proper range for both measurement and clinical interpretation in the future. Subsequently, the applicability of the method in analyzing clinical samples was demonstrated. Bioanalysis of 11 kinase inhibitors | 189

Materials and methods

Chemicals The analytes and internal standards were purchased from the following indicated sources. Dasatinib, erlotinib hydrochloride, gefitinib, imatinib mesylate, lapatinib ditosylate, nilotinib, sunitinib maleate, sorafenib tosylate, and vemurafenib were from Sequoia 13 2 Research Products (Pangbourne, UK). Erlotinib- C6 hydrochloride, gefitinib- H3, imati- 13 2 13 2 13 2 2 nib- C and H3, lapatinib- C and H7, pazopanib- C and H3 hydrochloride, sunitinib- H10, 13 2 13 sorafenib- C and H3, and vemurafenib- C6 were from AlsaChim (Illkirch, France). Pa- zopanib hydrochloride was from GlaxoSmithKline (London, UK) while the N-desethyl 2 2 2 sunitinib, dasatinib- H8, nilotinib- H3, and N-desethyl-sunitinib- H5 were from Toronto Research Chemicals (Toronto, ON, Canada). The chemical formulas and masses of the analytes are presented in table 1. Acetonitrile, isopropanol, and methanol were purchased from Biosolve Ltd., (Amsterdam, the Netherlands). Ammonium, formic acid, and LiChrosol water for HPLC were obtained from Merck (Darmstadt, Germany). Small volumes of control drug-free human plasma obtained from the Slotervaart Hospital (Am- sterdam, the Netherlands) were pooled and used for the validation.

Table 1. Molecular formulas and masses of analytes Generic drug name Molecular formula Molecular Mass (Da) Dasatinib C H Cl N O S 488.0 22 26 1 7 2 1 4.1

Erlotinib C22H23N3O4 393.4

Gefitinib C22H24ClFN4O3 446.9

Imatinib C29H31N7O 493.6

Lapatinib C29H26ClFN4O4S 581.1

N-desethyl-sunitinib C20H23FN4O2 370.4

Nilotinib C29H22F3N7O 529.5

Pazopanib C21H23N7O2S 437.5

Sorafenib C21H16N4ClF3O3 465.1

Sunitinib C22H27FN4O2 398.5

Vemurafenib C23H18ClF2N3O3S 489.9 190 | Chapter 4.1

Chromatographic equipment and conditions A Waters Acquity I class ultra-performance liquid chromatography (UPLC) system, consisting of a binary pump, column oven, on-line degasser, and an autosampler (Waters, Milford, MA, USA) was used. The mobile phase A consisted of a 10 mmol/L mixture of ammonium hydroxide (NH4OH) in water and mobile phase B was 1 mmol/L NH4OH in methanol. The UPLC system was equipped with a Gemini C18 HPLC column (50 × 2.0 mm I.D., 5.0-µm, Phenomenex, Torrance, CA, USA) and the flow rate was set to 0.25 mL/min. The analytical column was protected in-line by a Gemini C18 column (4 × 2.0 mm I.D., 5.0-µm, Phenomenex). The column was kept at 40oC, the autosampler was set at 8oC, and the injection volume was 1 µL. The autosampler needle was rinsed with a mixture of acetonitrile/methanol/isopropanol/water (1/1/1/1, v/v/v/v) containing 0.1% formic acid. During the first 0.5 min, the eluate was directed to the waste container using a divert valve to prevent contamination of the mass spectrometer. A rapid gradient program was used to achieve the separation. The initial conditions were 55% mobile phase B, which was maintained for 0.5 min. In the following 1.5 min, the mobile phase was increased to 80% B. After 1.5 min and at 80% B, the initial conditions were restored with a total run time of 5.5 min.

MS equipment and conditions An API 5500 triple quadrupole mass spectrometer equipped with a turbo ion spray (TIS) source (Sciex, Foster City, CA, USA) operating in the positive ion mode was used as the detector. For the quantification, multiple-reaction monitoring (MRM) chromatograms were acquired and processed using the Analyst® software version 1.6 (Sciex, Foster City, CA, USA). The quadrupoles were operated at unit resolution (0.7 Da). The operating parameters of the system and mass transitions are listed in table 2.

Preparation of calibrators, quality control samples, and internal standard solutions Two separate stock solutions of all the analytes (1 or 2 mg/mL) and internal standards (0.5 or 1 mg/mL) were prepared. Exactly 1 mg was weighed and dissolved in 1 mL of methanol. For the analytes, one stock solution was used to prepare the calibrators and the other stock for the preparation of the QC standards. The preparation of the two stock solutions for each compound was checked and in all cases, deviations were < ±5%. The stock solutions were further diluted with methanol to obtain separate working solutions each containing all of the analytes at a 20-fold concentration of the corresponding plasma samples. Bioanalysis of 11 kinase inhibitors | 191

Table 2. Tandem mass spectrometry (MS/MS) operating parameters

Setting Parameter Run duration 5.50 min Ion spray voltage 4500 V Collision gas 9 psi Curtain gas 20 psi Turbo gas 40 psi Ionization temperature 500 oC

Specific parameters Parent mass Product Collision Declustering Collision exit Analyte (m/z) mass (m/z) energy (V) potential (V) potential (V) Dasatinib 488 401 2 45 236 42 Dasatinib- H8 496 406 Erlotinib 394 278 13 45 236 34 Erlotinib- C6 400 284 Gefitinib 447 128 2 13 131 20 Gefitinib- H3 455 136 Imatinib 494 394 13 2 43 46 42 Imatinib- C, H3 498 394 Lapatinib 581 365 13 2 57 111 36 Lapatinib- C, H7 589 365 N-desethyl-sunitinib 371 283 2 27 126 40 N-desethyl-sunitinib- H5 376 283 Nilotinib 530 289 2 47 36 10 4.1 Nilotinib- H3 533 289 Pazopanib 438 357 13 2 11 136 14 Pazopanib- C, H3 442 361 Sorafenib 465 252 13 2 49 176 28 Sorafenib- C, H3 469 256 Sunitinib 399 326 2 31 106 44 Sunitinib- H10 409 326 Vemurafenib 490 383 13 45 196 38 Vemurafenib- C6 496 389

The calibrators were freshly prepared for every validation run by spiking a volume of 50 µL working solution to 1.0 mL of control human plasma. The QC samples were prepared in batches and stored at -20 oC until analysis. A mixture of the internal standard stock solutions was prepared and diluted with methanol to obtain a working solution that was used for sample preparation (table 3 lists the concentrations). 192 | Chapter 4.1

Table 3. Concentrations of stock and calibrators Analyte DSN ELN GFN IMN LPN mSNN NLN PZN SRN SNN VMN Solution Analyte stock (mg/mL) 1 2 2 2 2 1 2 2 2 1 2 IS stock (mg/mL) 1 0.5 1 1 0.5 0.5 0.5 1 1 0.5 1 CAL4/ULOQ (ng/mL) 100 2,500 1,000 2,500 1,000 100 2,500 10,000 10,000 100 10,000 CAL3/MID (ng/mL) 50 1,250 500 1,250 500 50 1,250 5,000 5,000 50 5,000 CAL2 (ng/mL) 10 250 100 250 100 10 250 1,000 1,000 10 1,000 CAL1/LLOQ (ng/mL) 5 125 50 125 50 5 125 500 500 5 500 IS working (ng/mL) 50 1,250 500 1,250 500 50 1,250 5,000 5,000 50 5,000 IS, internal standard; DSN, dasatinib; ELN, erlotinib; GFN, gefitinib; IMA, imatinib; LPN, lapatinib; NLN,- ni lotinib; PZN, pazopanib; SRN, sorafenib; SNN, sunitinib; mSNN, N-desethyl-sunitinib; VMN, vemurafenib; ULOQ, upper limit of quantification; MID, mid range level; LLOQ, lower limit of quantification.

Sample preparation Acetonitrile was used for protein precipitation in the sample pre-treatment. To 50 µL each of plasma and the internal standard working solution, 1000 µL acetonitrile was added. The mixture was vortexed for 10 s, and then centrifuged for 10 min at 11,300 g at 20-25 oC. A 500-µL aliquot of the mixture was transferred to an autosampler vial and

500 µL 100 mmol/L NH4OH in water was added. The vials were capped, vortexed for 10 seconds, and then stored at 2-8 oC until analysis.

Validation procedures The assay was validated according to the FDA Guidelines, and adjustments were made for the process to fit TDM purposes [14,15].

Regression models Four non-zero calibrators were prepared in duplicate for each run and analyzed in three separate independent runs. The linearity was evaluated using back-calculated concentrations of the calibrators. For all the analytes, the reciprocal of the squared concentration (1/x2) was used as the weighting factor. The deviations from the mean calculated concentrations over three runs were expected to be within ± 15% of the nominal concentrations. At the LLOQ, a deviation of ± 20% was permitted and the response of the analyte was expected to be at least five times higher than that of the blank sample.

Inaccuracy and imprecision Intra- and inter-assay inaccuracies and imprecisions of the method were determined by quantifying five replicates of each of the QC samples at the LLOQ, midrange level Bioanalysis of 11 kinase inhibitors | 193

(MID) and upper limit of quantification (ULOQ) concentration level in three separate runs. The concentration of each QC sample was calculated using the calibrators that were analyzed in duplicate in the same run. The difference between the nominal and the measured concentration were used to calculate the inaccuracies. The inaccuracy (bias) was expected to be ≤ 15% for MID and ULOQ, and ≤ 20% for the LLOQ. The imprecision (% coefficient of variation, CV) was expected to be ≤ 15% for the MID and ULOQ level and ≤ 20% at the LLOQ level. The ability to dilute samples with an analyte concentration that was originally above the ULOQ was demonstrated for pazopanib and vemurafenib by analyzing QC samples containing 10 times the concentration of the high QC sample. 4.1 These samples were prepared as 5-fold concentrations and were analyzed after a 10-fold dilution with the control human plasma.

Carry-over The carry-over was determined by injecting two processed control human plasma samples after a ULOQ sample. The peak areas in the blank processed sample were expected to be ≤20% of the peak area of the LLOQ sample.

Specificity and selectivity Six individual batches of control human plasma were used to assess the specificity and selectivity of the method. To determine whether any endogenous constituents inter- fered with the assay, a double blank and a sample spiked at the LLOQ of each batch were processed. The samples were then analyzed using the procedures described above. The areas of the peaks co-eluting with the analytes were expected to be ≤ 20% of the peak area of the LLOQ sample in each of the six batches. The areas of peaks in the double 194 | Chapter 4.1

blanks co-eluting with the internal standards were expected to be ≤ 5% of the peak area of the mean internal standard response. For the LLOQ, sample inaccuracies were expected to be within ± 20% of the nominal concentration in four of the six samples.

Stability The stability of the analytes was investigated in the stock solutions at an ambient temperature (2 h) and at -20 oC (1 month). The stability was also tested in human plasma for 48 h at 20-25 oC, for 1 month at -20 oC, and after three freeze-thaw cycles (-20o C to 20-25 oC) with a minimum interval of 24 h at two concentrations. The processed sample stability was investigated after 8 days of storage at 2-8 oC at two concentrations. The stability samples were quantified using freshly prepared calibrators. Reinjection reproducibility was determined in the processed sample after 24 h at approximately 2-8 oC at three concentrations. All the stability experiments were performed in triplicate. The analytes were considered stable in the stock solutions when 95-105% of the original concentration was detected. The analytes were considered stable in the plasma or processed sample when 85-115% of the initial concentration was recovered. The stability of the labeled internal standards was assumed to be equal to that of the corresponding unlabeled analytes. The isotopic purity of the internal standards was investigated during each analytical run by spiking control human plasma with the internal standard working solution. The peak area in the unlabeled isotope analyte window was expected to be < 20% of the peak area of the analytes at their LLOQ level.

Applicability of the method for TDM The applicability of the assay was demonstrated by analyzing the steady-state plasma concentrations of patients receiving the regular prescribed doses for the registered indications of the TKIs. Additionally, steady-state plasma concentrations of N-desethyl-sunitinib were measured in the plasma of patients receiving sunitinib. Bioanalysis of 11 kinase inhibitors | 195

Results

Regression models All the calibration curves were constructed using a weighting factor of 1/x2 and fitted linearly. The assay was linear for the validated concentration ranges of 5.00–100 ng/ mL for dasatinib, sunitinib, and N-desethyl-sunitinib; 50.0–1,000 ng/mL for gefitinib and lapatinib; 125–2,500 ng/mL for erlotinib, imatinib, and nilotinib; and 500–10,000 ng/mL for pazopanib, sorafenib, and vemurafenib. The correlation coefficients (r2) were at least 0.995 and the accuracies were within 85-115% in all cases. At all concentration levels, the CVs were ≤ 15% for the analytes.

Inaccuracy and imprecision The assay performance (intra-assay and inter-assay inaccuracy and imprecision) of all the analyzed compounds is summarized in table 4. The intra- and inter-assay biases were ≤ 15% for the MID and ULOQ levels and ≤ 20% for the LLOQ levels. The imprecisions were ≤ 15% for the MID and ULOQ levels and ≤ 20% for the LLOQ levels.

Carry-over The areas of the peaks in the first blank processed sample were ≤20% of the peak area of the LLOQ sample for all the analytes except for lapatinib, which required an extra blank sample. 4.1

Specificity and selectivity The MRM chromatograms of six batches of the control human plasma contained no co-eluting peaks that were > 20% of the area at the LLOQ level for all analytes and no co-eluting peaks > 5% of the area of all the internal standards. Figure 1 shows representative chromatograms of both LLOQ and blank samples. The influence of different control human plasma batches on the accuracy and precision at the LLOQ level was also investigated. The inaccuracies of the analytes at the LLOQ level were within ±20% of the nominal concentrations for all six batches of the control human plasma. 196 | Chapter 4.1

Table 4. Assay performance data Intra-assay Inter-assay Analyte Nominal concentration Bias CVa Bias CV (ng/mL) (%) (%) (%) (%) Dasatinib 4.97 4.1 15.9 3.0 11.5 49.7 10.2 7.2 -0.5 9.5 99.4 6.8 7.7 1.6 7.9 Erlotinib 128 -4.7 2.7 -1.7 3.4 1280 2.5 1.1 2.1 1.1 2550 -3.5 1.8 -1.6 2.0 Gefitinib 48.6 -12.1 11.5 -6.0 10.6 486 10.6 5.7 5.5 7.0 972 -5.6 5.2 -0.9 5.0 Imatinib 130 6.5 4.3 3.2 3.6 1300 1.4 2.3 0.1 1.8 2600 -2.3 2.5 -1.7 2.1 Lapatinib 51.0 -8.7 5.2 -3.9 5.6 510 4.2 2.8 1.0 3.4 1020 2.2 1.6 0.1 2.0 N-desethyl-sunitinib 5.11 -14.1 12.2 -5.1 12.2 51.1 -7.4 12.0 -1.7 10.3 102 -3.1 8.8 -1.3 6.3 Nilotinib 125.0 -4.8 4.6 -1.2 4.9 1250 2.9 1.5 2.3 1.2 2500 -4.8 2.3 -3.7 2.0 Pazopanib 502.0 8.8 9.1 4.4 8.7 5020 10.1 4.6 4.8 5.0 10000 -7.7 4.2 -0.9 5.9 Sunitinib 4.87 -5.2 14.9 -1.0 9.6 48.7 3.9 6.5 1.9 4.8 97.5 4.4 5.7 -0.8 5.6 Sorafenib 497 -3.8 2.2 -1.3 2.6 4970 4.1 1.0 3.0 1.1 9940 -6.7 1.3 -5.6 1.4 Vemurafenib 480 2.4 2.4 0.5 3.2 4800 2.5 2.3 1.5 1.9 9600 -2.0 3.5 -0.5 2.4 aCV, coefficient of variation (n = 5 per run, n = 15 in total) Bioanalysis of 11 kinase inhibitors | 197

DSN 488 à 401

Cl NH

N NH

N N N OH

Time , min Time , min

ELN O 394 à 278

O N

O N O

Time , min Time , min

GFN N 447 à 128 N NH

O O Cl F

Time , min Time , min

O IMN NH 494 à 394 4.1

NH N N N N N

Time , min Time , min LPN N N 581 à 365 NH

Cl O O

O F S O

Time , min Time , min

Figure 1. Part 1. Chromatograms of spiked plasma samples at lower limit of quantification (LLOQ, bracketed concentrations [ng/mL], left) with proposed fragmentation pathways and blank samples (right) TKIs: DSN, dasatinib (5.00); ELN, erlotinib (125); GFN, gefitinib (50.0); IMN, imatinib (125); LPN, lapatinib (50.0). 198 | Chapter 4.1

H mSNN N à O 371 283 H F N

O HN

Time , min Time , min NLN 530 à 289 O NH

N N NH N F N F F

Time , min Time , min

PZN O NH2 438 à 357 S O

N NH

N N N

Time , min Time , min

SNN N 399 à 326

O NH F

N O H

Time , min Time , min F F F SRN Cl 465 à 252 NH

HN O

Time , min Time , min

Figure 1. Part 2. Chromatograms of spiked plasma samples at lower limit of quantification (LLOQ, bracketed concentrations [ng/mL], left) with proposed fragmentation pathways and blank samples (right) TKIs: mSNN, N-desethyl-sunitinib (5.00); NLN, nilotinib (125); PZN, pazopanib (500); SNN, sunitinib (5.00); SRN, sorafenib (500). Bioanalysis of 11 kinase inhibitors | 199

VMNK 490 à 383 O O S NH

F F NH O

Time, min Time, min Figure 1. Part 3. Chromatogram of spiked plasma sample at lower limit of quantification (LLOQ, bracketed concentration [ng/mL], left) with proposed fragmentation pathway and blank sample (right) TKIs: VMN, vemurafenib (500).

Stability The stability of the analytes in various solutions and concentrations was investigated. The stock solutions were stable for at least 2 h at 20-25 oC. Plasma samples were stable for at least up to 48 h at 20-25 oC and for one month at -20 oC. The calibrators and QC samples were aliquoted, stored at -20 oC, and thawed directly before processing to minimize the number of freeze-thaw cycles. The stability in plasma was guaranteed for at least three freeze (-20 oC)-thaw cycles. The processed samples were stable for at least 8 days at 2-8 oC. The reinjection reproducibility after 24 h at approximately 2-8 oC was ensured. After storing the working solution for 6 months at -20o C, the peak areas of the analytes in the processed control human plasma sample spiked with the internal stan- 4.1 dard working solution were ≤ 20% of the peak areas of the analytes in an LLOQ sample.

Method applicability for TDM The results of the analyses of 10 plasma concentrations per inhibitor and N-desethyl sunitinib are presently reported here. The mean measured concentrations are shown in table 5 along with the lowest and highest measured plasma concentrations. Figure 2 shows typical chromatograms of a patient sample for each analyte. 200 | Chapter 4.1

DSN ELN GFN IMN Signal (cps) Signal (cps) Signal (cps) Signal Signal (cps) Signal

Time (min) Time (min) Time (min) Time (min)

mSNN LPN NLN PZN Signal (cps) Signal Signal (cps) Signal Signal (cps) Signal (cps) Signal

Time (min) Time (min) Time (min) Time (min)

SNN SRN VMN Signal (cps) Signal Signal (cps) Signal Signal (cps) Signal

Time (min) Time (min) Time (min)

Figure 2. Chromatograms of analyzed patient samples (bracketed concentrations [ng/mL]) TKIs: DSN, dasatinib (57.2); ELN, erlotinib (1,790); GFN, gefitinib (498); IMN, imatinib (1,140); LPN, lapatinib (883); mSNN, N-desethyl-sunitinib (27); NLN, nilotinib (1,730); PZN, pazopanib (28,900); SNN, sunitinib (67.5); SRN, sorafenib (2,190); and VMN, vemurafenib (48,000).

Table 5. Analyte concentrations in plasma collected from patients receiving standard therapy Drug Mean plasma concentration (ng/mL) Min.-Max. concentrations (ng/mL) Dasatinib 22.0 8.76–45.6 Erlotinib 993 259–1770 Gefitinib 305 89–761 Imatinib 1522 581–2170 Lapatinib 221 65.9–828 N-desethyl-sunitinib 18.3 7.94–32.3 Nilotinib 1615 603–2280 Pazopanib 40800* 13500*–82900* Sunitinib 49.1 5.51–94.2 Sorafenib 2998 1120–5530 Vemurafenib 42911* 12300*–90000* *Plasma samples were diluted ten times before sample preparation, 10 patients/analyte Bioanalysis of 11 kinase inhibitors | 201

Discussion A previously developed assay for the analysis of multiple TKIs was improvized with regard to the number of analytes that can be analyzed per run, the overall time of analysis, and its applicability to TDM routines [6]. The previous assay was developed for the analysis of eight compounds, and pazopanib, N-desethyl-sunitinib, and vemurafenib were added in the improved version. The sample preparation and the concentration range were altered along with these changes. Additionally, the chromatographic conditions were adjusted and, thereby, the total analysis time could be reduced from 10 to 5.5 min. The gradient profile was modified from a three- to a two-step profile while maintaining a flow of 0.25 mL/min. Typical chromatograms of samples at the LLOQ are depicted in figure 1. To develop a method more suitable for an efficient daily TDM assay, the bioanalytical set-up was changed. The previous assay had a larger calibration range (ULOQ = 400 x LLOQ), which was scaled down in the new assay (ULOQ = 20 x LLOQ). The number of calibrators was reduced to four, which provided a full calibration curve coverage with a shortened preparation time. The QC was carried out at three different levels (LLOQ, MID, and ULOQ) instead of the four levels (LLOQ, LOW, MID, and HIGH) used previously. The developed assay was suited to the demands of clinical practice where accurate results are needed in a relatively short time.

All the validated parameters complied with the FDA Guidelines that were adjusted to suit TDM use, except for the carry-over of lapatinib. Therefore, lapatinib samples should 4.1 be evenly distributed across the entire batch of samples. The stability results demonstrated that the developed assay had a workable set-up. The stock stability at 20-25 oC was sufficient for the preparation of the working solutions and subsequent spiking of the plasma. The patient samples were adequately stable but may require reanalysis, which would result in additional freeze-thaw cycles. The validated three-cycle process ensures this possibility. When needed, a full analytical run consisting of calibrators, QC samples, and patient samples can be reinjected up to 24 h after the first run. The measured patient plasma concentrations of all the analytes were within the validated range of the developed assay. This demonstrates that the developed method is suitable for the use in TDM practice. 202 | Chapter 4.1

Conclusions

A LC-MS/MS method was developed and validated for the simultaneous analysis of the TKIs: dasatinib, erlotinib, gefitinib, imatinib, lapatinib, nilotinib, pazopanib, sunitinib, sorafenib, and vemurafenib, as well as the metabolite N-desethyl-sunitinib in human plasma for TDM. The TKI- and metabolite-spiked human plasma was pre-treated by protein precipitation with acetonitrile and the addition of stable isotope labeled internal standards. The developed assay is appropriate for use in TDM services. Bioanalysis of 11 kinase inhibitors | 203

References

1. Terada T, Noda S, Inui K. Management of dose 2015;41:412–422. variability and side effects for individualized 9. Bardin C, Veal G, Paci A, et al. Therapeutic drug cancer pharmacotherapy with tyrosine kinase monitoring in cancer – Are we missing a trick? inhibitors. Pharmacol Ther. 2015;152:125–134. Eur J Cancer. 2014;50:2005–2009. 2. Petit-Jean E, Buclin T, Guidi M, et al. Erlotinib. 10. Bach DM, Straseski JA, Clarke W. Therapeutic Ther Drug Monit. 2015;37:2–21. drug monitoring in cancer chemotherapy. Bio- 3. Gao B, Yeap S, Clements A, et al. Evidence for analysis. 2010;2:863–879. therapeutic drug monitoring of targeted anti- 11. Yu H, Steeghs N, Nijenhuis CM, et al. Practical cancer therapies. J Clin Oncol. 2012;30:4017– guidelines for therapeutic drug monitoring of 4025. anticancer tyrosine kinase inhibitors: focus on 4. Decosterd L, Dahmane E, Neeman M, et al. the pharmacokinetic targets. Clin Pharmacoki- Therapeutic drug monitoring of targeted an- net. 2014;53:305–325. ticancer therapy. Tyrosine kinase inhibitors 12. de Wit D, Guchelaar H-J, den Hartigh J, et al. and selective estrogen receptor modulators: A Individualized dosing of tyrosine kinase inhib- clinical pharmacology laboratory perspective. itors: are we there yet? Drug Discov Today. In: Xu QA, Madden TL, eds. LC-MS in Drug Bio- 2015;20:18–36. analysis. Boston: Springer, 2012:197–250. 13. Josephs DH, Fisher DS, Spicer J, et al. Clinical 5. Widmer N, Bardin C, Chatelut E, et al. Review pharmacokinetics of tyrosine kinase inhibitors: of therapeutic drug monitoring of anticancer implications for therapeutic drug monitoring. drugs part two – Targeted therapies. Eur J Can- Ther Drug Monit. 2013;35:562–587. cer. 2014;50:2020–2036. 14. US Food and Drug administration (FDA). Guid- 6. Lankheet NAG, Knapen LM, Schellens JHM, et ance for Industry: Bioanalytical Method Valida- al. Plasma concentrations of tyrosine kinase tion. 2001;4–10. inhibitors imatinib, erlotinib, and sunitinib in 15. US Food and Drug administration (FDA). Guid- routine clinical outpatient cancer care. Ther ance for Industry: Bioanalytical Method Valida- Drug Monit. 2014;36:326–334. tion, draft guidance. 2013;4–24. 7. Lankheet NAG, Kloth JSL, Gadellaa-van 16. Lankheet NAG, Hillebrand MJX, Rosing H, et Hooijdonk CGM, et al. Pharmacokinetically al. Method development and validation for guided sunitinib dosing: a feasibility study in the quantification of dasatinib, erlotinib, ge- 4.1 patients with advanced solid tumours. Br J fitinib, imatinib, lapatinib, nilotinib, sorafenib Cancer. 2014;110:2441–2449. and sunitinib in human plasma by liquid chro- 8. Herbrink M, Nuijen B, Schellens JHM, et al. matography coupled with tandem mass spec- Variability in bioavailability of small molecular trometry. Biomed Chromatogr. 2013;27:466– tyrosine kinase inhibitors. Cancer Treat Rev. 476.

Chapter 4.2

Development and validation of a liquid chromatography-tandem mass spectrometry analytical method for the therapeutic drug monitoring of eight novel anticancer drugs

M. Herbrink1, N. de Vries1, H. Rosing1, A.D.R. Huitema1,2, B. Nuijen1, J.H.M. Schellens3,4, J.H. Beijnen1,4

Abstract

To support therapeutic drug monitoring (TDM) of patients with cancer, a fast and accurate method for simultaneous quantification of the registered anticancer drugs afatinib, axitinib, ceritinib, crizotinib, dabrafenib, enzalutamide, regorafenib and trametinib in human plasma using liquid chromatography tandem mass spectrometry was developed and validated.

Human plasma samples were collected from treated patients and stored at -20 oC. An- alytes and internal standards (stable isotopically labeled analytes) were extracted with acetonitrile. An equal amount of 10 mM NH4CO3 was added to the supernatant to yield the final extract. Two μL of this extract was injected onto a C18-column, gradient elution was applied, and triple-quadrupole mass spectrometry in positive-ion mode was used for detection.

All results were within the acceptance criteria of the latest US FDA guidance and EMA guidelines on method validation, except for the carry-over of ceritinib and crizotinib. These are corrected for by the injection order of samples. Additional stability tests were carried out for axitinib and dabrafenib in relation to their reported photostability.

In conclusion, the described method to simultaneously quantify the eight selected anticancer drugs in human plasma was successfully validated and applied for therapeutic drug monitoring in cancer patients treated with these drugs. Bioanalysis of 8 anticancer drugs | 207

Introduction

The group of personalized anticancer drugs is an expanding one [1–3]. Most of the recently developed and approved anticancer drugs are destined for oral ingestion. The bioavailability of this drug class is often low and variable leading, usually in combination with other complicating factors, to a variation in drug levels and exposure [4–8]. This may lead to insufficient efficacy or substantial toxicity [9,10].

For a large number of compounds, a relation between plasma concentration and efficacy of therapy has been established. For others, this remains to be demonstrated. Where such a relation is proven for drugs used in the clinic, the need arises for the integration of this knowledge into clinical practice. With this purpose, therapeutic drug monitoring (TDM) is a useful tool in the therapy of patients with cancer [11].

TDM aims to individualize drug dosing by focusing on balancing the therapeutic efficacy and the avoidance of drug toxicity. This is accomplished by quantifying drug concentrations in patient blood plasma or serum and by comparing the results with predetermined guidelines and target levels [11].

Our laboratory previously reported a TDM assay for the quantification of 11 oral anticancer compounds for which there was a demand from the clinic [12]. The clinical requirement for TDM analyses has grown to entail a larger variety of anticancer compounds. To suit these needs, an additional TDM assay was developed and validated for the quantification of the promising oral drugs afatinib, axitinib, ceritinib, crizotinib, dabrafenib, 4.2 enzalutamide, regorafenib and trametinib in human plasma. These targeted drugs are indicated for a variety of tumor types (table 1) and all exhibit variability in pharmacokinetics. The drugs are frequently used and a rapid, robust, and adequately ranged combination assay has not been reported in literature, as far as we know.

Guidelines and target levels for the TDM of the 8 compounds are few in number; however, information on pharmacokinetic targets that may be used in the clinical practice is provided by clinical studies, literature reviews and approval documents [13–22].

The concentration ranges of the assay were chosen to enable a short analysis-evaluation time and to ensure a swift feedback to the clinic whilst taking the plasma trough levels into account. 208 | Chapter 4.2

Table 1. Trivial names, molecular formulas and molecular masses of the assay analytes Trivial name Primary Primary Molecular Molecular indication(s) target(s) formula mass (Da)

Afatinib NSCLC EGFR C24H25ClFN5O3 485.9

Axitinib RCC VEGFR, PDGFR C22H18N4OS 386.5

Ceritinib NSCLC after treatment MEK C28H36ClN5O3S 558.1 with crizotinib

Crizotinib NSCLC MEK C21H22Cl2FN5O 450.3

Dabrafenib Melanoma B-Raf C23H20F3N5O2S2 519.6

Enzalutamide PC Androgen receptor C21H16F4N4O2S 464.4

Regorafenib CRC, GIST VEGFR, PDGFR C21H17ClF4N4O4 482.8

Trametinib Melanoma, NSCLC MEK C16H23FIN4O4 615.4 B-raf, v-raf murine sarcoma viral oncogene homolog B1; EGFR, Endothelial Growth Factor Receptor; GIST, gastro-instestinal stromal tumors; CRC, colorectal cancer; MEK, Mitogen activated protein Kinase kinase; NSCLC, non-small cell lung cancer; PC, prostate carcinoma; PDGFR, Platelet Derived Grwoth Factor Recep- tor; RCC, Renal cell carcinoma; VEGFR, Vascular Endothelial Growth Factor Receptor.

The validation parameters were tested according to the US Food and Drug Administra- tion (FDA) and European Medicines Agency (EMA) Guidelines [23,24], with the following modifications. (1) Four non-zero calibration standards were used in the calibration curve instead of 6 to 8. (2) The quality control low concentration sample was replaced with a QC lower limit of quantification (LLOQ) sample for the determination of the inaccuracy and imprecision. These adaptations made it possible to develop and validate a method that is applicable in time-limited daily clinical routine.

This article describes the development, validation, and application of the quantitative analysis of 8 anticancer compounds in human plasma for TDM purposes and is the first to do so for these compounds. Special attention was directed to the light sensitivity of axitinib and dabrafenib as reported earlier [25,26]. Plasma samples were subjected to protein precipitation and were subsequently analyzed using a high-performance liquid chromatography (HPLC) system coupled to a tandem mass spectrometry detection. The method is now routinely used in our laboratory for TDM of patients treated with any of these drugs. Plasma levels are directly reported to hospital pharmacists who relate the levels to pharmacokinetic targets. A case-to-case dosage advice is subsequently discussed with the treating oncologist. Bioanalysis of 8 anticancer drugs | 209

Experimental

Chemicals and reagents Afatinib, axitinib, ceritinib, crizotinib, dabrafenib, enzalutamide, regorafenib,- trame 13 13 2 2 13 2 2 tinib, afatinib- C6, axitinib- C, H3, ceritinib- H7, crizotinib- C2, H5, dabrafenib- H9, en- 2 13 2 13 zalutamide- H6, regorafenib- C, H3 and trametinib- C6 were purchased from AlsaChim (Illkirch, France). Acetonitrile and water (HPLC grade), used to prepare mobile phases, along with methanol for stocks and working solutions were obtained from Biosolve Ltd (Valkenswaard, The Netherlands). Dimethylsulfoxide (DMSO) for stocks and ammonium hydrogen carbonate, used for the mobile phases were obtained from Merck (Darmstadt,

Germany). K2EDTA plasma was obtained from the Medical Center Slotervaart (Amster- dam, the Netherlands).

Stock solutions, calibrations standards and quality control samples Stock solutions containing axitinib and dabrafenib were stored in amber-colored containers. The stock of axitinib was prepared in methanol in which it is less soluble than in DMSO, but more stable [26,27]. Separate stock solutions for calibration standards and quality control samples were prepared in methanol or DMSO, see table 2. Stock solutions of the internal standards were prepared in the same medium and concentration as their analyte equivalents. The IS working solution was prepared in methanol at 13 13 2 2 a concentration of 125 ng/mL for afatinib- C6, axitinib- C, H3, dabrafenib- H9 and tra- 13 2 13 2 metinib- C6 and of 3,125 ng/mL for ceritinib- H7, crizotinib- C2, enzalutamide- H6 and regorafenib-13C,2H . All stock and working solutions were stored at -20o C. 3 4.2 Working solutions were prepared to spike the calibration and quality control plasma samples containing the highest concentration of analytes. This was done to avoid adsorption and stability issues that may occur with axitinib and dabrafenib at lower concentrations in organic solvents. The calibration standard working solution and the quality control sample working solution both contained 4,000 ng/mL afatinib, axitinib, dabrafenib and trametinib and 100,000 ng/mL ceritinib, crizotinib, enzalutamide and regorafenib.

The calibration standard containing the highest concentration of analytes was spiked directly after the preparation of the working solutions. A volume of 50 μL of calibration standard working solution was added to 950 μL control human K2EDTA plasma. The highest standard was subsequently diluted in control human K2EDTA plasma to obtain the target calibration concentrations as listed in table 2. The quality control (QC) plasma 210 | Chapter 4.2

sample containing the highest concentration of analytes was spiked using a separately prepared working solution. This QC sample (QC high) was further diluted in control human K2EDTA plasma to obtain QC mid and QC low (table 2). Both the calibration standards and quality control samples were stored in 50 μL aliquots at -20 oC.

Table 2. Concentrations of analytes in stock solutions, calibration and quality control plasma samples. Analyte Stock (mg/mL) Calibration (ng/mL) Quality control (ng/mL) Afatinib 1.00 (DMSO) 2; 10; 100; 200 2; 100; 200 Axitinib 0.02 (methanol) 2; 10; 100; 200 2; 100; 200 Ceritinib 1.00 (DMSO) 50; 250; 2500; 5000 50; 2500; 5000 Crizotinib 1.00 (DMSO) 50; 250; 2500; 5000 50; 2500; 5000 Dabrafenib 1.00 (DMSO) 2; 10; 100; 200 2; 100; 200 Enzalutamide 1.00 (DMSO) 50; 250; 2500; 5000 50; 2500; 5000 Regorafenib 1.00 (DMSO) 50; 250; 2500; 5000 50; 2500; 5000 Trametinib 0.10 (DMSO) 2; 10; 100; 200 2; 100; 200

Sample preparation Directly after sample collection by venipuncture in the clinic, the blood samples were centrifuged for 10 min at 2,000 g at 4 oC. After centrifugation, the plasma was isolated and stored in amber containers at -20 oC pending analysis. Prior to processing the samples were thawed at 20-25 oC and a 50 μL aliquot was transferred to an amber colored container of 1.5 mL. A volume of 20 μL of IS working solution was added to the plasma sample. To precipitate plasma proteins and to extract the analytes from the biomatrix, a volume of 150 μL of acetonitrile was added. The samples were then vortex-mixed and centrifuged (10 min at 20 oC and 23,100 g). A 100 μL aliquot was subsequently added to an autosampler vial that contained 100 μL 10 mM ammonium bicarbonate in water. The dilution factor of sample to final extract is 8.8, yielding 0.114 mL sample per mL of final extract. The final extract was vortex-mixed and stored at 2-8o C until analysis.

Liquid chromatography-tandem mass spectrometry Chromatographic separation was carried out using an UPLC Acquity I Class binary pump with integrated degasser, an UPLC Acquity FTN autosampler and an UPLC Acquity Column CH-A heater (all: Waters, Milford, MS, USA). The autosampler temperature was kept at 4 oC and the column heater at 40 oC. The mobile phase A consisted of 10 mM ammonium bicarbonate in water and mobile phase B was 10 mM ammonium bicarbonate in meth- Bioanalysis of 8 anticancer drugs | 211

anol-water (1:9, v/v). A gradient program was used at a flow of 0.250 mL/min through a Gemini C18 column (5.0 μm, 50 x 2.0 mm I.D.) (Phenomenex, Torrance, CA, USA) with an additional Gemini C18 guard column (4x2.0 mm I.D.). The gradient program was as follows: 40% B, 250 μL/min (0-0.1 min); 40 to 100% B, 250 μL/min (0.1-2.0 min); 100% B, 250 μL/min (2.0-4.0 min); 100 to 40% B, 250 μL/min (4.0-4.01 min); 40% B, 500 μL/min (4.01-4.5 min); 40% B, 250 μL/min (4.5-5.0 min). The divert valve directed the flow to the mass spectrometer between 2.0 and 5.0 min; during the remainder of the run the eluent was directed to the waste container. Table 3 lists the typical retention times.

The drugs were analyzed using an API5500 quadrupole mass spectrometer (MS) (Sciex, Thornhill, ON, Canada). This instrument is equipped with a turbo ion spray interface, operating in positive ion mode and configured in multiple reaction monitoring (MRM) mode. The LC-MS/MS data were acquired and processed with Analyst™ software version 1.5.2 (Sciex). Table 3 summarizes the MS operating parameters.

Validation procedure The assay was fully validated for calibration model, accuracy and precision, lower limit of quantification, sensitivity and selectivity, dilution integrity, carry-over, matrix effect and stability. The validation was based on the FDA and latest EMA guidelines on bioanalytical method validation. The accuracy is expressed as the bias and the following equations were used:

Intra-assay bias (%)= (1) 100% · (mean measured conc. per run-nominal conc.)/(nominal conc.) 4.2

Inter-assay bias (%)= (2) 100% · (overall mean measured conc.-nominal conc.)/(nominal conc.)

The precision is expressed as the coefficient of variation (CV) and the following equations were used:

Intra assay CV (%)= (3) 100% · (SD of the measured conc. per run)/(mean measured conc. per run)

(4) 29 /((; =⋯=; ?<)?@(; ?<)29=⋯=A(; ?<)29BC) 9 9 3456788 < 7 < < 7 7 (; ?<)2 =⋯=(;7?<)2 1 D?E < < 7F 7?< ;<=⋯=;7?7 0 ; ( ) 7 Inter assay CV (%) = GHIJ KL MNJO 212 | Chapter 4.2

Table 3. Mass spectrometric settings for the analytes and their internal standards and the proposed mass fragmentation pathways. Parameter Setting Run duration 8.0 min Ion spray voltage 5500 kV Collision gas 6.0 psi Curtain gas 30 psi Turbo gas 30 psi Temperature 500 oC

Analyte Setting Proposed fragmentation Analyte Setting Proposed fragmentation Parameter pathway Parameter pathway Afatinib 486.1 ➔ 371 Dabrafenib 520.2 ➔ 307.3 13 2 307 Afatinib- C 492.3 ➔ 377.1 O Dabrafenib- H 529.2 ➔ 316.3 H N 6 O N 9 2 N Retention time 4.7 min Retention time 3.9 min F Collision energy 37 V N Collision energy 44 V N HN O Declustering potential 96 V HN Declustering potential 116 V S O NH Collision exit potential 26 V Collision exit potential 20 V F O S F N F N 371 Cl

Axitinib 387.0 ➔ 327.9 Enzalutamide 465.0 ➔ 209.0 13 2 356 2 209 Axitinib- C, H3 391.3 ➔ 356.1 Enzalutamide- H6 471.2 ➔ 215.2 Retention time 2.3 min NH O Retention time 4.1 min H N O Collision energy 17 V H Collision energy 25 V N F S N N F Declustering potential 50 V Declustering potential 131 V O N Collision exit potential 28 V Collision exit potential 14 V F S F

Ceritinib 558.2 ➔ 516.1 NH Regorafenib 483.1 ➔ 270.1 Cl 2 13 2 F Ceritinib- H7 565.4 ➔ 517.3 Cl Regorafenib- C, H3 487.0 ➔ 274.1 N F NH Retention time 5.6 min Retention time 4.8 min F F 516 Collision energy 25 V NH N NH Collision energy 30 V O NH O Declustering potential 116 V S O Declustering potential 136 V O O 270 Collision exit potential 14 V Collision exit potential 18 V NH N O

Crizotinib 450.1 ➔ 177.0 H Trametinib 615.9 ➔ 490.9 13 2 N 13 Crizotinib- C2, H5 457.3 ➔ 177.1 Trametinib- C6 622.2 ➔ 497.0 O O N O 491 Retention time 4.4 min Retention time 4.6 min F HN N NH Collision energy 35 V N Collision energy 45 V Cl H N N Declustering potential 81 V O Declustering potential 90 V I Collision exit potential 20 V Collision exit potential 12 V O N H N Cl 2 F 177 Bioanalysis of 8 anticancer drugs | 213

Ceritinib 558.2 ➔ 516.1 NH Regorafenib 483.1 ➔ 270.1 Cl 4.2 2 13 2 F Ceritinib- H7 565.4 ➔ 517.3 Cl Regorafenib- C, H3 487.0 ➔ 274.1 N F NH Retention time 5.6 min Retention time 4.8 min F F 516 Collision energy 25 V NH N NH Collision energy 30 V O NH O Declustering potential 116 V S O Declustering potential 136 V O O 270 Collision exit potential 14 V Collision exit potential 18 V NH N O

Where s is the standard deviation, n is the number of replicates and a is the number of runs. The matrix factor (MF) was determined by using the following equation:

MF (-) = (5) (area of processed blank sample spiked with neat solution (matrix present))/(area of neat solution(matrix absent))

Resolution (-) = (difference in retention time)/0.5*(sum of base widths) (6)

Photodegradation of axitinib and dabrafenib Special attention was given to the light-sensitivity of axitinib and dabrafenib. Sparidans et al. showed that axitinib exhibits light-mediated trans to cis isomerization [26]. Nijen- huis et al. reported light-instability of dabrafenib in organic solvents [25]. The stability of these two analytes in plasma in both light and dark environments was investigated.

Human K2EDTA control plasma was spiked at both QCLow and QCHigh levels and was stored at ambient temperature in both light and dark environments for 48 consecutive hours to simulate possible light exposure during transport from the clinic to the laboratory. Samples from both methods were analyzed to investigate whether isomerization or degradation had occurred.

Clinical application The purpose of this assay is to facilitate TDM of afatinib, axitinib, ceritinib, crizotinib, dabrafenib, enzalutamide, regorafenib, trametinib in treated patients with cancer. As part of routine clinical care, 2K EDTA blood samples (4 mL) were collected from patients who were treated with any of the drugs at the Antoni van Leeuwenhoek – the Nether- lands Cancer Institute. The plasma samples were collected and processed as described in this report. Bioanalysis of 8 anticancer drugs | 215

Results and discussion

Development Mass spectrometry Mass spectrometric parameters were optimized for each analyte by performing direct infusion of the analytes. General settings were chosen to maximize the response of trametinib, which was the lowest responder and due to the fact that it shows the lowest target concentrations. The molecular ions (M+H+) that were observed and used to gen- erate product ion spectra are listed in table 3. To achieve high selectivity and sensitivity, MRM scan mode was utilized to monitor and select the mass transition for each analyte that yielded the product ion with the highest abundance (table 3). The optimal settings, however, led to the non-linearity of the calibration model for ceritinib, dabrafenib and enzalutamide due to saturation of the detector. To prevent this over-response, the collision energy (CE) of the analytes was lowered from 35 to 25, 47 to 44 and 41 to 25 for ceritinib, dabrafenib and enzalutamide, respectively. Figure 1 presents the chromatograms of 8 analytes in a single sample, indicating the capability of simultaneous detection.

56234 871 100 ) (% 50

Signal 4.2

0 234 Time (min)

Figure 1. Representative normalized MRM chromatograms of spiked human control K2-EDTA plasma at QC- mid concentrations: afatinib (1; 100 ng/mL), axitinib (2; 100 ng/mL), ceritinib (3; 2,500 ng/mL), crizotinib (4; 2,500 ng/mL), dabrafenib (5; 100 ng/mL), enzalutamide (6; 2,500 ng/mL), regorafenib (7; 2,500 ng/mL) and of trametinib (8; 100 ng/mL). 216 | Chapter 4.2

HPLC The chromatographic set-up that was used in a previously reported TDM-assay was taken as starting point for this assay. This system used a Gemini C18 column and a gradient program that consisted of 10 mM NH4OH in water and 1 mM NH4OH in methanol. Both afatinib, axitinib and trametinib responded poorly under these conditions and the chromatographic peaks of crizotinib and regorafenib in the chromatogram were broad and tailing was observed. The poor response of trametinib could be improved by increasing the injection volume to 2 μL. The increase in response of afatinib and axitinib was not satisfactory. The mobile phases were subsequently replaced by the bicarbonate system as mentioned in the materials section. This resulted in improved peak shapes and sensitivity for afatinib and axitinib.

Validation procedures

Calibration model Calibration standards were prepared and analyzed in duplicate in three analytical runs. The linear regression of peak area ratios versus the concentration (x) was weighed (1/ x2) to obtain the lowest absolute total bias and the most constant bias across the range. The calibration range for afatinib, axitinib, dabrafenib and trametinib was 2.0 – 200 ng/ mL and for ceritinib, crizotinib, enzalutamide and regorafenib was 50 – 5,000 ng/mL. Cal- ibration curves were accepted if 75% of the non-zero calibration standards, including a LLOQ and a upper limit of quantification standard (ULOQ), had a bias within ±15% of the nominal concentration (±20% for the LLOQ). All calibration curves (n = 3) of the analytes met these criteria and correlation coefficients (r2) were 0.996 or better.

Accuracy and precision To assess the accuracy and precision of the assay, five replicates of QC low (= LLOQ), QC- mid and QC high in plasma were analyzed in three analytical runs. Table 4 summarizes the intra- and inter-assay accuracies and precisions of the assay. The biases and CVs were within the acceptance criteria (within ±20% and ≤20%, respectively, at the low level and within ±15% and ≤15% at the other QC levels). Bioanalysis of 8 anticancer drugs | 217

Table 4. Assay performance data for the analysis of 8 anti-cancer drugs in human plasma Intra-assay Inter-assay Analyte Nom. Conc. (ng/mL) Biasa (%) CVa (%) Bias (%) CV (%) Matrix factor (-) (CV%) Afatinib 2 ±10 ≤9.1 3.6 3.9 1.03 (7.9) 100 ±3.5 ≤5.7 -1.8 2.1 200 ±3.7 ≤4.3 1.8 1.8 0.85 (5.3) Axitinib 2 ±4.4 ≤10.0 -4.2 -* 1.01 (7.1) 100 ±2.6 ≤2.5 -2.1 -* 200 ±5.4 ≤2.0 1.0 3.7 1.05 (8.7) Ceritinib 50 ±16 ≤7.5 -3.1 11.4 1.02 (13.3) 2500 ±4.1 ≤4.6 0.1 3.7 5000 ±4.3 ≤2.6 4.1 -* 1.07 (4.5) Crizotinib 50.3 ±7.8 ≤2.2 -3.8 3.6 1.04 (2.7) 2520 ±3.7 ≤1.2 1.8 2.1 5030 ±1.8 ≤2.1 -0.6 1.6 0.93 (4.1) Dabrafenib 2 ±10.9 ≤11.6 -7.1 -* 0.97 (14.5) 100 ±1.0 ≤4.9 0.4 -* 200 ±4.4 ≤3.8 2.4 1.7 1.12 (10.7) Enzalutamide 49.9 ±6.8 ≤5.5 -2.2 5.4 1.13 (8.6) 2500 ±6.5 ≤4.9 -3.1 2.7 4990 ±1.0 ≤2.9 0.2 -* 1.00 (13.5) Regorafenib 50.2 ±3.9 ≤2.3 2.0 2.8 1.05 (8.6) 2510 ±1.3 ≤2.0 -0.6 0.9 5020 ±0.9 ≤1.5 0.5 -* 0.98 (8.3) Trametinib 2 ±6.7 ≤7.5 1.7 3.4 0.95 (7.7) 100 ±1.0 ≤2.0 0.2 -* 200 ±6.0 ≤2.3 2.1 3.2 0.99 (11.5) * The inter-run precision could not be calculated because there is no significant additional variation due to the performance of the assay in different runs 4.2 Lower limit of quantification The analyte responses at the low were at least 5 times the response compared to a blank response in three validation runs. The lowest signal to noise ratio was 7 for ceritinib. Figure 2 shows representative ion chromatograms of all 8 analytes in QC low samples and double blank samples. 218 | Chapter 4.2

Figure 2, part 1. MRM Chromatograms of a blank sample (A-series) and spiked QC sample at the LLOQ (B-se- Figureries): afatinib 2, part (1;1. MRM2 ng/mL), Chromatograms axitinib (2; 2 ofng/mL), a blank ceritinib sample (3; (A 50-series) ng/mL), and crizotinib spiked (4; QC 50sample ng/mL). at the LLOQ (B-series): afatinib (1; 2 ng/mL), axitinib (2; 2 ng/mL), ceritinib (3; 50 ng/mL), crizotinib (4; 50 ng/mL).

173

Bioanalysis of 8 anticancer drugs | 219

4.2

Figure 2,2, partpart 2.2. MRMMRM Chromatograms Chromatograms of ofa blanka blank sample sample (A-series) (A-series) and and spiked spiked QC sampleQC sample at the at LLOQthe LLOQ(B-series): (B-series): dabrafenib dabrafenib (5; 2 ng/mL), (5; 2 ng/mL), enzalutamide enzalutamide (6; 50 ng/mL), (6; 50 regorafenib ng/mL), regorafenib (7; 50 ng/mL) (7; and 50 of ng/mL) trametinib and (8; 2 ng/mL). of trametinib (8; 2 ng/mL).

174

220 | Chapter 4.2

Selectivity and sensitivity

Six different batches of control human 2K EDTA plasma were spiked at the LLOQ level to investigate the selectivity and the sensitivity. Single determinations were performed. The mean deviations from the nominal concentrations from the assay were all within ±20% with CV values ≤20%. There were no peaks observed with areas ≥20% of the LLOQ Selectivityin double and blank sensitivity samples of these batches for any of the analytes and also no interfer- Sixences different were batchesdetected of atcontrol the retention human K 2timesEDTA plasma of the internalwere spiked standards at the forLLOQ both level assays. to investigate the selectivity and the sensitivity. Single determinations were performed. The Selectivity was therefore considered acceptable. mean deviations from the nominal concentrations from the assay were all within ±20% with CV values ≤20%. There were no peaks observed with areas ≥20% of the LLOQ in double blank samplesCross-analyte of these and batches internal for anystandard of the interferenceanalytes and wasalso testedno interferences by spiking were control detected human at theplasma retention at ULOQ times level of withthe internalall analytes standards and IS atfor nominal both assays. concentrations Selectivity separately.was therefo There considered acceptable. Crosscross-analyte analyte and and internal internal standard standard interferenceinterference wasat the tested retention by spiking times control of all humananalytes plasmawas less at thanULOQ 20% level of withthe peak all analytes area of theand LLOQIS at levelnominal of the concentrations respective analytes.separately. For The the crossinternal analyte standard and internal the interference standard wasinterference less than at 5%. the For retention all analytes times it wasof all found analytes that wasthe less than 20% of the peak area of the LLOQ level of the respective analytes. For the internal cross analyte and internal standard interference was acceptable. standard the interference was less than 5%. For all analytes it was found that the cross analyte and internal standard interference was acceptable. InIn orderorder to investigateinvestigate thethe photo-stabilityphoto-stability ofof axitinib, axitinib, thethe separation separation betweenbetween thethe -ste stereoisomersreoisomers of of axitinib axitinib waswas tested.tested. FigureFigure 3 3present present a acharacteristic characteristic chromatogram chromatogram of a of light-exposed axitinib methanolic sample with a resolution of 6. The assay was thus capable ofa separating light-exposed the twoaxitinib stereoisomers. methanolic sample with a resolution of 6. The assay was thus capable of separating the two stereoisomers.

Figure 3. MRM ChromatogramFigure 3. MRM of Chromatogram an axitinib sample of anprocessed axitinib from sample a methanolic processed solution exposed to light in a transparent container for 2 hours: the E-isomer (t = 2.3 min) and the formed Z-isomer (t = 3.2 min). from a methanolic solution exposedr to light in a transparent r container for 2 hours: the E-isomer (tr = 2.3 min) and the formed Z-isomer (tr = 3.2 min).

175

Bioanalysis of 8 anticancer drugs | 221

Dilution integrity Plasma samples of enzalutamide have to be diluted ten-fold to ensure quantification within the calibration range. Five replicate plasma samples of enzalutamide at QC>ULOQ level were diluted ten-fold with control human plasma (10 μL was added to 90 μL control human matrix). The intra-assay bias and CVs were -4.0% and 4.2%, respectively. The bias and CV were within ±15% and ≤15% which indicates that the study samples can be diluted 10 times in control matrix with acceptable accuracy and precision.

Carry-over Carry-over was investigated by injecting two double blank samples after a ULOQ sample in three analytical runs. Eluting peaks with areas >20% of the LLOQ were only observed in blank samples injected after ULOQ samples of ceritinib and crizotinib. However, -car ry-over into the second double blank samples was ≤20% of the LLOQ level. Therefore, samples containing these analytes should not be grouped. In this way, the carry-over will not have an impact on the integrity of the data.

Matrix factor The internal standard-normalized matrix factor was determined in six plasma batches, at QC low and QC high concentrations. Single determinations were performed. Processed blank samples were spiked with working solutions and compared to matrix free solutions. The CV of the internal standard-normalized matrix factor was ≤15% for all analytes at both concentration levels (table 4). The acceptance criteria were thus met. The aver- age values of the internal standard-normalized matrix factors are listed in table 4 and are around 1. These results indicate that the stable isotopes as internal standards are most 4.2 effective to compensate for matrix effects.

Stability All analytes were found to be stable in plasma at -20 oC for at least 1 month and after 3 freeze (-20 oC)-and-thaw (20-25 oC) cycles. Stability was also demonstrated for all analytes in plasma for at least 48 hours at 20-25 oC in amber-colored containers. In final extracts, stability was demonstrated for at least 48 hours at 4 oC. Reinjection reproducibility results showed that an entire analytical run can be reanalyzed after 48 hours when kept at 2-8 oC. Stock solution are stable for at least 5 months at -20 oC and stability tests are still ongoing. 222 | Chapter 4.2

Photodegradation of axitinib and dabrafenib The isomerization under influence of light of axitinib was monitored using the chromatographic system described above. Axitinib and dabrafenib were found to be light-stable in plasma at 20-25 oC for at least 48 hours in transparent containers. Axitinib showed isomerization from the therapeutically active E-isomer to the Z-isomer under the influence of light in methanolic stock and working solutions (data not shown) as described by Spar- idans et al [26]. The isomerization was not observed in methanolic solutions that were kept protected from light in amber colored containers. During the sample preparation as presented here, no degradation of dabrafenib was observed.

Clinical application The recommended doses for the drugs are listed in table 5. To test the applicability of this assay the concentrations were determined in 10 patient samples for each drug of patients treated with the drugs. The results are listed in table 5. All values are within the validated range of 2.0 to 200 ng/mL for afatinib, axitinib, dabrafenib and trametinib and 50.0 to 5,000 ng/mL for ceritinib, crizotinib, enzalutamide and regorafenib. These results demonstrate the applicability of this assay for TDM purposes. In clinical practice, plasma level and drug administration data are used in comparison with pharmacokinetic targets to provide an adequate dosage advice to treating oncologists. By doing so, insufficient or excessive exposure is effectively recognized and managed.

Table 5. Recommended doses and plasma concentrations of the analytes in patient samples of patients treated with the drugs (n = 10). Drug Recommended Mean plasma Minimum – maximum Calibration range dose (mg) [28] concentration (ng/mL) concentration (ng/mL) (ng/mL) Afatinib 40 od 43.3 10.7 - 120 2 – 200 Axitinib 5 bd 19.0 7.57 - 34.9 2 – 200 Ceritinib 750 od 1,195 333 – 2,630 50 – 5,000 Crizotinib 250 bd 227 139 – 458 50 – 5,000 Dabrafenib 150 bd 102 14.1 – 191 2 – 200 Enzalutamide* 160 od 10,293 7,140 – 12,200 50 – 5,000 Regorafenib 160 od 1,118 178 – 2,030 50 – 5,000 Trametinib 2 od 8.40 5.77 – 11.6 2 – 200 od, once daily; bd, bi-daily. *Enzalutamide patient samples are diluted 10-fold prior to analysis Bioanalysis of 8 anticancer drugs | 223

Conclusion

A sensitive new LC-MS/MS assay for the quantification of afatinib, axitinib, ceritinib, crizotinib, dabrafenib, enzalutamide, regorafenib and trametinib in human plasma was developed and validated. The validated assay ranges from 2.0 to 200 ng/mL for afatinib, axitinib, dabrafenib and trametinib and from 50.0 to 5,000 ng/mL for ceritinib, crizotinib, enzalutamide and regorafenib were linear and correlation coefficients (r2) of 0.996 or better were obtained. Dilution integrity experiments show that samples can be diluted 10-fold for enzalutamide prior to analysis. Both axitinib and dabrafenib are prone to degrade when exposed to light in stock and working solutions but these analytes were shown to be stable in plasma for at least one week at 20-25 oC.

4.2 224 | Chapter 4.2

References

[1] US Food and Drug administration (FDA), CDER the pharmacokinetic targets., Clin. Pharmaco- 2014 report to the Nation, (2014). kinet. 53 (2014) 305–25. doi:10.1007/s40262- [2] US Food and Drug administration (FDA), CDER 014-0137-2. 2015 report to the Nation, (2015). [12] M. Herbrink, N. de Vries, H. Rosing, A.D.R. [3] US Food and Drug administration (FDA), CDER Huitema, B. Nuijen, J.H.M. Schellens, J.H. Bei- 2016 Report to the Nation, (2016). jnen, Quantification of 11 Therapeutic Kinase [4] M. Herbrink, B. Nuijen, J.H.M. Schellens, Inhibitors in Human Plasma for Therapeutic J.H. Beijnen, Variability in bioavailability of Drug Monitoring Using Liquid Chromatogra- small molecular tyrosine kinase inhibitors, phy Coupled With Tandem Mass Spectrom- Cancer Treat. Rev. (2015). doi:10.1016/j. etry., Ther. Drug Monit. 38 (2016) 649–656. ctrv.2015.03.005. doi:10.1097/FTD.0000000000000349. [5] T. Terada, S. Noda, K. Inui, Management of dose [13] H.I. Scher, T.M. Beer, C.S. Higano, A. Anand, M.- variability and side effects for individualized E. Taplin, E. Efstathiou, D. Rathkopf, J. Shelkey, cancer pharmacotherapy with tyrosine kinase E.Y. Yu, J. Alumkal, D. Hung, M. Hirmand, L. inhibitors, Pharmacol. Ther. 152 (2015) 125– Seely, M.J. Morris, D.C. Danila, J. Humm, S. Lar- 134. doi:10.1016/j.pharmthera.2015.05.009. son, M. Fleisher, C.L. Sawyers, Prostate Cancer [6] B. Gao, S. Yeap, A. Clements, B. Balakrishnar, Foundation/Department of Defense Prostate M. Wong, H. Gurney, Evidence for Therapeu- Cancer Clinical Trials Consortium, Antitumour tic Drug Monitoring of Targeted Anticancer activity of MDV3100 in castration-resistant Therapies, J. Clin. Oncol. 30 (2012) 4017–4025. prostate cancer: a phase 1-2 study., Lan- doi:10.1200/JCO.2012.43.5362. cet (London, England). 375 (2010) 1437–46. [7] N. Widmer, C. Bardin, E. Chatelut, A. Paci, J. doi:10.1016/S0140-6736(10)60172-9. Beijnen, D. Levêque, G. Veal, A. Astier, Review [14] US Food and Drug administration (FDA), Clin- of therapeutic drug monitoring of anticancer ical Pharmacology and Biopharmaceutics Re- drugs part two – Targeted therapies, Eur. J. view Giotrif, (2013). Cancer. 50 (2014) 2020–2036. doi:10.1016/j. [15] L.D. Locati, L. Licitra, L. Agate, S.-H.I. Ou, A. ejca.2014.04.015. Boucher, B. Jarzab, S. Qin, M.A. Kane, L.J. [8] N.A.G. Lankheet, L.M. Knapen, J.H.M. Schel- Wirth, C. Chen, S. Kim, A. Ingrosso, Y.K. Pitha- lens, J.H. Beijnen, N. Steeghs, A.D.R. Huitema, vala, P. Bycott, E.E.W. Cohen, Treatment of ad- Plasma Concentrations of Tyrosine Kinase In- vanced thyroid cancer with axitinib: Phase 2 hibitors Imatinib, Erlotinib, and Sunitinib in study with pharmacokinetic/pharmacodynam- Routine Clinical Outpatient Cancer Care, Ther. ic and quality-of-life assessments, Cancer. 120 Drug Monit. 36 (2014) 326–334. doi:10.1097/ (2014) 2694–2703. doi:10.1002/cncr.28766. FTD.0000000000000004. [16] US Food and Drug administration (FDA), Clin- [9] H. Kato, N. Sassa, M. Miyazaki, M. Takeuchi, M. ical Pharmacology and Biopharmaceutics Re- Asai, A. Iwai, Y. Noda, M. Gotoh, K. Yamada, view Xalkori, (2011). Association of axitinib plasma exposure and [17] D. Ouellet, E. Gibiansky, C. Leonowens, A. genetic polymorphisms of ABC transporters O’Hagan, P. Haney, J. Switzky, V.L. Goodman, with axitinib-induced toxicities in patients with Population pharmacokinetics of dabrafenib, a renal cell carcinoma., Cancer Chemother. Phar- BRAF inhibitor: effect of dose, time, covariates, macol. 78 (2016) 855–62. doi:10.1007/s00280- and relationship with its metabolites., J. Clin. 016-3145-0. Pharmacol. 54 (2014) 696–706. doi:10.1002/ [10] R.B. Verheijen, H. Yu, J.H.M. Schellens, J.H. jcph.263. Beijnen, N. Steeghs, A.D.R. Huitema, Practical [18] G.S. Falchook, G. V. Long, R. Kurzrock, K.B. Recommendations for Therapeutic Drug Mon- Kim, H.-T. Arkenau, M.P. Brown, O. Hamid, J.R. itoring of Kinase Inhibitors in Oncology, Clin. Infante, M. Millward, A. Pavlick, M.T. Chin, S.J. Pharmacol. Ther. (2017). doi:10.1002/cpt.787. O’Day, S.C. Blackman, C.M. Curtis, P. Lebowitz, [11] H. Yu, N. Steeghs, C.M. Nijenhuis, J.H.M. Schel- B. Ma, D. Ouellet, R.F. Kefford, Dose Selection, lens, J.H. Beijnen, A.D.R. Huitema, Practical Pharmacokinetics, and Pharmacodynamics guidelines for therapeutic drug monitoring of of BRAF Inhibitor Dabrafenib (GSK2118436), anticancer tyrosine kinase inhibitors: focus on Clin. Cancer Res. 20 (2014) 4449–4458. Bioanalysis of 8 anticancer drugs | 225

doi:10.1158/1078-0432.CCR-14-0887. s00280-016-2993-y. [19] B.I. Rini, B. Melichar, M.N. Fishman, M. Oya, [23] US Food and Drug administration (FDA), Guid- Y.K. Pithavala, Y. Chen, A.H. Bair, V. Grünwald, ance for Industry: Bioanalytical Method Valida- Axitinib dose titration: analyses of exposure, tion, draft guidance, (2013) 4–24. blood pressure and clinical response from a [24] European Medicines Agency (EMA), C. for M.P. randomized phase II study in metastatic renal for H. Use, Guideline on bioanalytical method cell carcinoma., Ann. Oncol. Off. J. Eur. Soc. validation, (2011). Med. Oncol. 26 (2015) 1372–7. doi:10.1093/ [25] C.M. Nijenhuis, H. Haverkate, H. Rosing, J.H.M. annonc/mdv103. Schellens, J.H. Beijnen, Simultaneous quan- [20] European Medicines Agency (EMA), Assess- tification of dabrafenib and trametinib in hu- ment report Zykadia (EPAR), (2015). man plasma using high-performance liquid [21] J.R. Infante, L.A. Fecher, G.S. Falchook, S. Nal- chromatography–tandem mass spectrometry, lapareddy, M.S. Gordon, C. Becerra, D.J. DeMa- J. Pharm. Biomed. Anal. 125 (2016) 270–279. rini, D.S. Cox, Y. Xu, S.R. Morris, V.G. Peddared- doi:10.1016/j.jpba.2016.03.049. digari, N.T. Le, L. Hart, J.C. Bendell, G. Eckhardt, [26] R.W. Sparidans, D. Iusuf, A.H. Schinkel, J.H.M. R. Kurzrock, K. Flaherty, H.A. Burris, W.A. Schellens, J.H. Beijnen, Liquid chromatog- Messersmith, Safety, pharmacokinetic, phar- raphy-tandem mass spectrometric assay macodynamic, and efficacy data for the oral for the light sensitive tyrosine kinase inhib- MEK inhibitor trametinib: a phase 1 dose-esca- itor axitinib in human plasma, J. Chromato- lation trial, Lancet Oncol. 13 (2012) 773–781. gr. B. 877 (2009) 4090–4096. doi:10.1016/j. doi:10.1016/S1470-2045(12)70270-X. jchromb.2009.10.024. [22] D. Ouellet, N. Kassir, J. Chiu, M.-S. Mouksassi, [27] Pfizer Inc, Typically Polymorphic Form Has a C. Leonowens, D. Cox, D.J. DeMarini, O. Gard- Subtle Effect on Photoreactivity, in: n.d.: pp. ner, W. Crist, K. Patel, Population pharmacoki- 1–37. netics and exposure-response of trametinib, a [28] US Food and Drug administration (FDA), Pre- MEK inhibitor, in patients with BRAF V600 mu- scribing information for the respective drugs, tation-positive melanoma., Cancer Chemoth- (n.d.). er. Pharmacol. 77 (2016) 807–17. doi:10.1007/

4.2

Chapter 5

Dexamphetamine

Chapter 5.1

Development and validation of a high-performance liquid chromatography-tandem mass spectrometry assay for the quantification of Dexamphetamine in human plasma

M. Herbrink1, B. Thijssen1, M.J.X. Hillebrand1, H. Rosing1, J.H.M. Schellens2,3, B. Nuijen1, J.H. Beijnen1,3

Abstract

Dexamphetamine is registered for the treatment of attention deficit hyperactivity disorder and narcolepsy. Current research has highlighted the possible application of dexamphetamine in the treatment of cocaine addiction. To support clinical pharmacologic trials a new simple, fast, and sensitive assay for the quantification of dexamphetamine in human plasma using liquid chromatography tandem mass spectrometry (LC-MS/MS) was developed. Additionally, it is the first reported LC-MS assay with these advantages to be fully validated according to current US FDA and EMA guidelines. Human plasma samples were collected on an outpatient basis and stored at nominally -20 oC. The analyte and the internal standard (stable isotopically labeled dexamphetamine) were extracted using double liquid-liquid extraction (plasma-organic and organic-water) combined with snap-freezing. The aqueous extract was filtered and 2 μL was injected on a C18-column with isocratic elution and analyzed with triple quadrupole mass spectrometry in positive ion mode. The validated concentration range was from 2.5-250 ng/mL and the calibration model was linear. A weighting factor of 1 over the squared concentration was applied and correlation coefficients of 0.997 or better were obtained. At all concentrations the bias was within ±15% of the nominal concentrations and imprecision was ≤15%. All results were within the acceptance criteria of the latest US FDA guidance and EMA guidelines on method validation. In conclusion, the developed method to quantify dexamphetamine in human plasma was fit to support a clinical study with slow-release dexamphetamine. Bioanalysis of dexamphetamine | 231

Introduction

Dexamphetamine (dextroamphetamine, S-amphetamine) (figure 1) is a powerful central nervous system stimulant [1]. It exerts its pharmacological action through the rapid increase of dopamine levels in the striatum and noradrenaline levels in the prefrontal cortex [2].

119.1 119.1 91.0 65.1 91.0

NH2 cps) y( 65.1 nsit te In 51.2 63.1 136.2

40 60 80 100 120 140 m/z (amu) Figure 1. Molecular structure with suggested fragmentation pathways, product ion mass spectrum of dexamphetamine and parent ion (MH+) signal at m/z 136.2.

In Europe, dexamphetamine is used in the therapy of attention-deficit hyperactivity disorder (ADHD) for children between the ages of 6 and 17 years where methylphenidate and atomoxetine were not effective [3,4]. In the United States, in addition to ADHD, dexamphetamine is also licensed to be prescribed to patients that suffer from narcolepsy [5]. Increasing attention is directed towards the possible application of dexamphetamine in the treatment of cocaine dependence, for which no medication is yet approved. Sev- 5.1 eral reviews and studies suggest the promising effectiveness of dexamphetamine in the agonist pharmacotherapy of cocaine dependence [6,7].

A study where the acceptance, safety, and efficacy of dexamphetamine in the therapy of chronic cocaine-dependent patients on heroin-assisted treatment were investigated was recently carried out in the Netherlands [8]. For this trial, a new sustained release formulation of dexamphetamine was developed. To support further pharmacokinetic clinical studies of this formulation, a fast and accurate bioanalytical assay for the quantification of dexamphetamine in human plasma was developed. 232 | Chapter 5.1

Several analytical assays for dexamphetamine have been described in literature [9–23]. The published assays for the quantification of dexamphetamine in plasma cover several analytical methods, such as gas and high-performance liquid chromatography coupled to mass spectrometry (HPLC-MS).

The HPLC-MS methods for human plasma require double liquid-liquid extraction with toxic chloroform, expensive solid-phase extraction, long analysis time or a relatively large injection volume [9,16,18,22]. Additionally, no bioanalytical HPLC-MS method with an isotopically labeled internal standard, which has been fully validated for human plasma according to current bioanalytical guidelines, has been published as far as we know, however. Our goal was to develop a sensitive LC-MS method with a simple, fast and low-cost sample pre-treatment. Previously published methods fall short in one or more of these factors.

This article describes a simple, fast and sensitive method for the quantification of dexamphetamine in human plasma. The assay was fully validated according to the latest US Food and Drug Administration (FDA) and European Medicines Agency (EMA) guidelines on bioanalytical method validation [24,25]. Additionally, the applicability of the method was tested and demonstrated in clinical study sample analysis with concentrations of dexamphetamine. Bioanalysis of dexamphetamine | 233

Experimental

Chemicals

Dexamphetamine sulfate ((C9H13N)2SO4 was purchased from Fagron (Waregem, Belgium). 2 Deuterated amphetamine ( H6), used as an internal standard (IS) for the assay was manufactured by and obtained from LGC Standards (Teddington, UK). Methanol and water (UPLC-MS grade) were purchased from Biosolve (Amsterdam, The Netherlands). Methyl tert-butyl ether (Lichrosolv grade), 25% ammonia (Emsure grade) and 99% formic acid (Emsure grade) were obtained from Merck (Darmstadt, Germany).

Stock solutions, calibration standards, and quality control samples Separate stock solutions for calibration standards and quality control samples (QC samples) of 0.1 mg/mL were prepared in water. These stock solutions were further diluted in water to obtain working solutions. Stock solutions of the IS were prepared at a concentration of 0.1 mg/mL in water. The IS working solution contained 250 ng/mL. All stock and working solutions were stored at -20o C.

Calibration standards were spiked directly after the preparation of the working solutions. A volume of 50 μL of calibration working solution was added to 950 μL control human K2EDTA plasma to obtain concentrations of 2.5, 5, 10, 25, 50, 100, 200 and 250 ng/mL. The QC samples were prepared by adding 100 μL QC working solution to 1900 μL control human K2EDTA plasma. Final concentrations at the lower limit of quantification (LLOQ), QC low, QC mid, QC high and QC above upper limit of quantification (>ULOQ) were 2.5, 7.5, 25, 200 and 2000, respectively.

Sample preparation 5.1 A volume of 20 µL internal standard working solution and 30 µL of 1 M NH4OH were added to a 200 µL plasma aliquot. The mixture was vortexed for 10 seconds. 1000 µL of methyl tert-butyl ether (MTBE) was added, after which the volume was vortexed for 10 seconds and mixed in an automatic shaker for 10 min at 1250 rpm. Next, the samples were centrifuged at 20,000 g for 5 min. The aqueous layer was snap frozen on an ethanolic dry ice bath. The organic layer was then transferred to a 1.5 mL Eppendorf tube. 200 µL of 0.01 M formic acid was introduced to the organic layer and the mixture was vortexed for 10 seconds and mixed on an automatic shaker for 10 min. at 1250 rpm. After centrifuging at 20,000 g for 5 min, the aqueous layer was snap frozen and the organic layer discarded. Next, the aqueous fraction was thawed and filtered over a 0.2 234 | Chapter 5.1

µm spin-filter (10,000 g, 5 min), transferred to an autosampler vial and stored at 2-8oC until analysis.

Liquid chromatography The chromatographic separation was carried out using an HPLC-system consisting of an 1100 series binary pump, column oven, on-line degasser and well plate autosampler (Agilent technologies, Palo Alto, CA, USA). Mobile phase A was prepared by adding 0.5 mL 99% formic acid to 1000 mL of water. Mobile phase B was prepared in the same fashion, with methanol instead of water. Mobile phase A and B were isocratically (60% B) pumped through a Kinetex C18 100 Å column (150 x 2.1 mm I.D., 2.6 µm; Phenomenex, Torrance, CA, USA) at a flow rate of 0.15 mL/min. The analytical column was protected by an inline filter (0.2 µm, Upchurch Scientific, Oak Harbor, WA, USA). The column was maintained at 40oC during analysis. Volumes of 2 µL were injected using the autosampler at a temperature of 4oC. The total run time was 5 min.. The autosampler needle was rinsed before each injection with methanol. During the first minute, the eluate was directed to waste using a divert valve to prevent the introduction of endogenous compounds into the mass spectrometer.

Mass spectrometer An API 3000 triple quadrupole mass spectrometer equipped with a Turbo Ion Spray In- terface (TIS) source (AB Sciex, Foster City, CA, USA) operating in the positive ion mode was used as a detector. For quantification, multiple reaction monitoring (MRM) chromatograms were acquired and processed using Analyst® software (AB Sciex). The qua- druples were operating at unit resolution (0.7 Da). TIS-MS/MS operating parameters and mass transitions are presented in table 1. Figure 1 shows the product ion spectrum of dexamphetamine.

Validation procedures The assay validation was performed in accordance with the OECD principles of Good Laboratory Practice (GLP) [26]. Calibration model, accuracy, and precision, selectivity, dilution integrity, lower limit of quantitation, matrix effect, carry-over and stability under various conditions were determined according to the latest FDA and EMA guidelines covering bioanalytical method validation [24,25]. Bioanalysis of dexamphetamine | 235

Table 1. ESI-MS/MS operating parameters General settings Parameter Setting Run duration 5 min Ion spray voltage 5500 V Collision gas 2 AU Curtain gas 8 AU Nebulizer gas 5 AU Turbo gas temperature 500oC Turbo gas flow 7 L/min

Dexamphetamine Amphetamine-D6 Mass transition 136.2 -> 91.0 m/z 142.1 -> 93.2 m/z Dwell time 150 msec Collision energy 19 Declustering potential 16 Collision exit potential 4 Focussing potential 90 Entrance potential 10 Retention time 2.3 min

The inaccuracy is expressed as the bias and the following equations were used:

Intra-assay bias (%)= (1) 100% · (mean measured conc. per run-nominal conc.)/(nominal conc.)

Inter-assay bias (%)= (2) 100% · (overall mean measured conc.-nominal conc.)/(nominal conc.) 5.1

The precision is expressed as the coefficient of variation (CV) and analysis of variance (ANOVA) was used to calculate the precision:

Intra-assay CV (%)= (3) 100% · (SD of the measured conc. per run)/(mean measured conc. per run)

3: /((< >⋯>< @=)@A(< @=)3:>⋯>B(< @=)3: CD) : : (4) 4567899 = 8 = = 8 8 (< @=)3 >⋯>(<8@=)3 2 E@F = = 8G 8@= <=>⋯><8@8 1 < ( ) 8

Inter − assay CV (%) = HIJK LM NOKP 236 | Chapter 5.1

Where s is the standard deviation, n is the number of replicates and a is the number of runs. The matrix factor (MF) was determined by using the following equation:

MF= (5) (area of processed blank sample spiked with neat solution (matrix present))/(area of neat solution(matrix absent))

Results and discussion

Development An easy-to-use and fast double liquid-liquid extraction was developed for the plasma sample preparation. It was found that the robustness and reproducibility of the assay were improved by performing an acid-base extraction. In alkaline solution, dexamphetamine is uncharged and readily partitions into the organic MTBE phase. After acidification of the organic phase (charging dexamphetamine), re-extraction into the aqueous phase was carried out, leaving impurities behind.

MTBE was selected as the organic extraction phase because of its low toxicity and low melting point (-109 oC [27]) which is required for snap-freezing.

The mass spectrometric parameters were optimized for dexamphetamine by performing direct infusion (analyte in 70% methanol). The observed molecular ion and the product ions as shown in figure 1 were consistent with data found in literature [13,17]. Analyte response and signal-to-noise ratios (S/N) were investigated for three mass transitions (136 ➔ 65.1, 91.0 and 119.1). It was found that although the transition to 119.1m/z yielded the highest signal, it also showed the lowest S/N however. Low S/N values were also found for 65.1 m/z, where these were caused by a relatively low analyte response. The S/N values and the signal strength were optimal for 91.0m/z . It was observed that baseline levels and S/N ratios for all mass transition increased and decreased, respectively from day-to-day, when other drug assays were run on the system prior to a dexamphetamine run. This may be due to the ubiquity of the low mass ions during the ionization process originating from analytes, solvents, and contaminants present in the mass spectrometer. Such a situation is likely to occur in a high-throughput LC-MS/MS setting. The baseline was efficiently reduced to <300 cps before each run by injecting a series of blank samples (10) into the system. Bioanalysis of dexamphetamine | 237

Regression model Eight non-zero plasma calibrators were prepared freshly and analyzed in duplicate in five separate analytical runs. The linear regression of the ratio of the areas of the analyte and the IS peaks versus the concentration were weighted with weighting factors of 1/ x2 (where x = concentration). The linearity was evaluated by means of back-calculated concentrations of the calibrators. The assay was linear over the validated concentration range from 2.50 to 250 ng/mL. Calibration curves were accepted if 75% of the non-zero calibration standards, including an LLOQ and a ULOQ, had a bias within ±15% of the nominal concentration (±20% for the LLOQ). All calibration curves (n=5) fulfilled these criteria and correlation coefficients were at least 0.997.

Accuracy and precision Intra- and inter-assay bias and precision (CV%) of the analysis were determined by assay- ing five replicates of each of the QC samples at LLOQ, low, mid and high concentration level in three separate runs. The concentration of each QC sample was calculated using the calibration standards that were analyzed in duplicate in the same analytical run. Table 2 summarizes the results. The difference between the calculated and the nominal concentrations was used to determine the bias. The bias was within the acceptance criteria of ±15% for low, mid and high concentration levels and within ±20% for the LLOQ level. The imprecision was below the acceptance criteria of 15% of the coefficient of variation (CV), and below 20% for the LLOQ.

Table 2. Assay performance data for the analysis of dexamphetamine in human plasma Nominal Intra-assay Inter-assay concentration Biasa (%) CVa (%) Bias (%) CV (%) (ng/mL) 2.50 4.4 - 5.4 4.8 – 9.7 4.8 -* 7.51 -0.5 - 11.6 3.4 – 7.8 4.4 5.4 5.1 25.0 -0.5 - 3.5 4.6 – 8.8 1.5 -* 200 -2.7 - 5.4 1.2 – 3.9 2.4 4.2 aIf multiple validation runs were performed, the range of bias’ and imprecisions are listed. *The inter-assay precision could not be calculated because there is no significant additional variation due to the performance of the assay in different runs.

Lower limit of quantification The analyte response at the LLOQ level was at least 5 times the response compared to a blank response in five validation runs. Figure 2 shows representative MRM chromatograms of a QC LLOQ sample and a double blank sample. 238 | Chapter 5.1

Figure 2. Representative MRM Chromatograms. Blank human K EDTA plasma (A1. Dexamphetamine, B1. Figure 2. Representative MRM Chromatograms. Blank human K2 EDTA plasma (A1. Dexamphetamine, D -amphetamine); Spiked human K EDTA plasma at LLOQ concentration2 (A2. Dexamphetamine (2.51 ng/ B1.6 D -amphetamine); Spiked human2 K EDTA plasma at LLOQ concentration (A2. Dexamphetamine mL), B2.6 D -amphetamine (25 ng/mL)); Patient2 (t=0 hours) sample (A3. Dexamphetamine (0 ng/mL), B3. (2.51 ng/mL),6 B2. D -amphetamine (25 ng/mL)); Patient (t=0 hours) sample (A3. Dexamphetamine (0 D -amphetamine (256 ng/mL)); Patient (t=8 hours) sample (A4. Dexamphetamine (34 ng/mL), B4. D -Am- 6ng/mL), B3. D -amphetamine (25 ng/mL)); Patient (t=8 hours) sample (A4. Dexamphetamine (346 phetamine (25 ng/mL)).6 ng/mL), B4. D6-Amphetamine (25 ng/mL)).

189

Bioanalysis of dexamphetamine | 239

Cross-analyte and IS interference were tested by spiking control human plasma at ULOQ level with the analyte and IS at nominal concentration separately. The cross-analyte and IS interference at the retention time of the analyte and IS were less than 20% of the peak area of the LLOQ level. For the IS the interference was less than 5%. The cross-analyte and IS interference was considered acceptable.

Dilution integrity Five replicate plasma samples at QC>ULOQ level were diluted 10-fold with control human K2EDTA plasma (50 μL sample and 450 μL control plasma). The intra-assay bias and CV were -1.9% and 1.5%, respectively. The bias and CV were within ±15% and ≤15% which indicates that the study samples can be diluted 10 times in control human matrix with acceptable accuracy and precision. The extended concentration range for dexamphetamine is therefore 2.5-2500 ng/mL.

Carry-over Carry-over was determined by analyzing two processed control human plasma samples after a ULOQ sample in three separate runs. Eluting peaks with areas >20% of the LLOQ were not observed in the blank samples injected directly after ULOQ samples, and therefore, the criteria for carry-over were met.

Matrix factor and recovery The matrix factor (MF) was determined in six plasma batches (pooled), at QC low and QC high concentration levels. Single determinations were performed. Processed blank samples were spiked with working solutions and compared to matrix free neat solutions. In addition to the MF, the internal standard-normalized MF was calculated by dividing the MF of the analyte by the MF of the internal standard. The maximal CV of the internal standard-normalized MF calculated from the six plasma batches for the low and high 5.1 concentrations were 5.1% and 1.5%, respectively. With this, the assay fulfilled the criteria (<15%). The matrix factor ranged from 0.951 to 1.11. These results (MF around 1) indicate that the use of the stable isotope-labeled standard is effectively minimizing the influence of matrix effects.

Stability The stability data are summarized in table 3. dexamphetamine was stable at -20 oC in plasma and in water for at least 16 and 17 months, respectively. Stability experiments in plasma were performed with QC low and QC high samples. In plasma dexamphetamine 240 | Chapter 5.1

was stable for at least 24 hours at ambient temperature, indicating that no specific stability precautions were required during sample handling at the clinical site. Final extract stability shows that the extracts can be injected up to 17 days after sample preparation.

Table 3. Stability data for dexamphetamine in stock solution and plasma. All experiments performed in plasma were performed in triplicate in QC low and QC high samples (n=3). Matrix Conditions Nominal Measured Deviation C.V. conc. conc. (ng/ (%) (%) (ng/mL) mL) Water -20 oC, 17 months 42.9* 42.3* -1.3 2.8 Plasma 3 freeze (-20 oC) / thaw cycles 7.51 7.75 3.2 7.8 200 193 -3.5 0.9 Plasma Ambient, 24 hours 7.51 7.62 1.5 5.6 200 192 -4.0 2.3 Plasma -20 oC, 16 months 7.51 8.23 9.6 5.0 200 207 3.7 3.7 Final extract 2-8 oC, 17 days 7.51 7.42 -1.2 8.0 200 195 -2.7 1.7 Final extract 2-8 oC, 30 hours (reinjection reproducibility) 7.51 7.39 -1.6 1.3 200 190 -5.2 2.4 *Presented are normalized response values, where corrections were made for reference drug weights and solvent volumes.

Application in a clinical study During a clinical study with a slow-release formulation of dexamphetamine, patients received doses of 60 or 30 mg once daily. To test the applicability of this assay, dexamphetamine concentrations were measured in 130 patient samples of patients treated with dexamphetamine in this study. A typical patient curve is shown in figure 3. All values are within the validated range of 2.5-250 ng/mL. These results demonstrate the applicability of this assay for clinical pharmacokinetic studies. Bioanalysis of dexamphetamine | 241

Conclusions

A simple, fast, and sensitive LC-MS/MS assay for the quantification of dexamphetamine in human plasma was developed and validated to support clinical studies. The presented assay was validated according to the current US FDA and latest EMA guidelines. The assay is able to quantify dexamphetamine over a range of 2.5 to 250 ng/mL in 200 μL samples, with the possibility to dilute the samples containing higher concentrations 10- fold with control plasma prior to analysis. In conclusion, the assay is considered well suited for its purpose and is now used to support clinical studies with dexamphetamine.

60 ) n mL io

(ng/ 40 tr at ne en mi nc ta co he

ma 20 mp as xa Pl de

0 0510 15 20 25 Time (hours) Figure 3. Typical patients plasma curve after intake of a single dose of sustained-release dexamphetamine (orally, 60 mg).

5.1 242 | Chapter 5.1

References

[1] E. Sinita, D. Coghill, The use of stimulant med- Ν.S. Τhomaidis, Simultaneous determination ications for non-core aspects of ADHD and of 148 pharmaceuticals and illicit drugs in in other disorders, Neuropharmacology. 87 sewage sludge based on ultrasound-assisted (2014) 161–172. extraction and liquid chromatography-tandem [2] D.J. Heal, S.L. Smith, R.S. Kulkarni, H.L. Rowley, mass spectrometry., Anal. Bioanal. Chem. 407 New perspectives from microdialysis studies (2015) 4287–97. in freely-moving, spontaneously hypertensive [12] L. Guo, Z. Lin, Z. Huang, H. Liang, Y. Jiang, Y. Ye, rats on the pharmacology of drugs for the Z. Wu, R. Zhang, Y. Zhang, Y. Rao, Simple and treatment of ADHD., Pharmacol. Biochem. Be- rapid analysis of four amphetamines in human hav. 90 (2008) 184–97. whole blood and urine using liquid–liquid ex- [3] P. Hodgkins, M. Shaw, S. McCarthy, F.R. Sallee, traction without evaporation/derivatization The Pharmacology and Clinical Outcomes of and gas chromatography–mass spectrometry, Amphetamines to Treat ADHD, CNS Drugs. 26 Forensic Toxicol. 33 (2015) 104–111. (2012) 245–268. [13] H.J. Leis, G. Fauler, W. Windischhofer, Enanti- [4] S. King, S. Griffin, Z. Hodges, H. Weatherly, C. oselective quantitative analysis of ampheta- Asseburg, G. Richardson, S. Golder, E. Taylor, mine in human plasma by liquid chromatogra- M. Drummond, R. Riemsma, A systematic re- phy/high-resolution mass spectrometry., Anal. view and economic model of the effectiveness Bioanal. Chem. 406 (2014) 4473–80. and cost-effectiveness of methylphenidate, [14] J. Wang, Z. Yang, J. Lechago, Rapid and simul- dexamfetamine and atomoxetine for the treat- taneous determination of multiple classes of ment of attention deficit hyperactivity disorder abused drugs and metabolites in human urine in children and adolescents., Health Technol. by a robust LC-MS/MS method - application to Assess. 10 (2006) iii–iv, xiii-146. urine drug testing in pain clinics, Biomed. Chro- [5] US Food and Drug Administration (FDA), Pre- matogr. 27 (2013) 1463–1480. doi:10.1002/ scribing information Adderall XR, (2013). bmc.2945. [6] X. Castells, M. Casas, C. Pérez-Mañá, C. Ronce- [15] M.K. Bjørk, K.W. Simonsen, D.W. Andersen, P.W. ro, X. Vidal, D. Capellà, Efficacy of psychostimu- Dalsgaard, S.R. Sigurðardóttir, K. Linnet, B.S. lant drugs for cocaine dependence., Cochrane Rasmussen, Quantification of 31 illicit and me- Database Syst. Rev. (2010) CD007380. dicinal drugs and metabolites in whole blood [7] D. V Herin, C.R. Rush, J. Grabowski, Agonist-like by fully automated solid-phase extraction and pharmacotherapy for stimulant dependence: ultra-performance liquid chromatography-tan- preclinical, human laboratory, and clinical dem mass spectrometry., Anal. Bioanal. Chem. studies., Ann. N. Y. Acad. Sci. 1187 (2010) 76– 405 (2013) 2607–17. 100. [16] S. Napoletano, C. Montesano, D. Compagnone, [8] M. Nuijten, P. Blanken, B. van de Wetering, B. R. Curini, G. D’ascenzo, C. Roccia, M. Sergi, De- Nuijen, W. van den Brink, V.M. Hendriks, Sus- termination of Illicit Drugs in Urine and Plasma tained-release dexamfetamine in the treat- by Micro-SPE Followed by HPLC–MS/MS, Chro- ment of chronic cocaine-dependent patients matographia. 75 (2012) 55–63. on heroin-assisted treatment: a randomised, [17] A.A.S. Marais, J.B. Laurens, Rapid GC-MS con- double-blind, placebo-controlled trial, Lancet. firmation of amphetamines in urine by extrac- 387 (2016) 2226–2234. tive acylation., Forensic Sci. Int. 183 (2009) [9] L. Martinsson, X. Yang, O. Beck, N.G. Wahlgren, 78–86. doi:10.1016/j.forsciint.2008.10.021. S. Eksborg, Pharmacokinetics of dexampheta- [18] M. Sergi, E. Bafile, D. Compagnone, R. Curini, mine in acute stroke., Clin. Neuropharmacol. G. D’Ascenzo, F.S. Romolo, Multiclass analy- 26 270–6. sis of illicit drugs in plasma and oral fluids by [10] T.H. Boles, M.J.M. Wells, Analysis of amphet- LC-MS/MS., Anal. Bioanal. Chem. 393 (2009) amine and methamphetamine in municipal 709–18. wastewater influent and effluent using weak [19] K. Kudo, T. Ishida, K. Hara, S. Kashimura, A. Tsu- cation-exchange SPE and LC-MS/MS., Electro- ji, N. Ikeda, Simultaneous determination of 13 phoresis. 37 (2016) 3101–3108. amphetamine related drugs in human whole [11] P. Gago-Ferrero, V. Borova, M.E. Dasenaki, blood using an enhanced polymer column and Bioanalysis of dexamphetamine | 243

gas chromatography-mass spectrometry., J. the quantitation of amphetamines in human Chromatogr. B. Analyt. Technol. Biomed. Life plasma and oral fluid by LC-MS-MS., J. Anal. Sci. 855 (2007) 115–20. Toxicol. 27 (2003) 78–87. [20] C. Wang, G. Fan, M. Lin, Y. Chen, W. Zhao, Y. [23] H.H. Maurer, T. Kraemer, C. Kratzsch, F.T. Pe- Wu, Development and validation of a liquid ters, A.A. Weber, Negative ion chemical ioniza- chromatography/tandem mass spectrometry tion gas chromatography-mass spectrometry assay for the simultaneous determination of and atmospheric pressure chemical ionization D-amphetamine and diphenhydramine in bea- liquid chromatography-mass spectrometry of gle dog plasma and its application to a phar- low-dosed and/or polar drugs in plasma., Ther. macokinetic study., J. Chromatogr. B. Analyt. Drug Monit. 24 (2002) 117–24. Technol. Biomed. Life Sci. 854 (2007) 48–56. [24] US Food and Drug administration (FDA), Guid- [21] M. Concheiro, A. de Castro, O. Quintela, M. ance for Industry: Bioanalytical Method Valida- López-Rivadulla, A. Cruz, Determination of tion, draft guidance, (2013) 4–24. drugs of abuse and their metabolites in human [25] European Medicines Agency (EMA), C. for M.P. plasma by liquid chromatography-mass spec- for H. Use, Guideline on bioanalytical method trometry. An application to 156 road fatalities., validation, (2011). J. Chromatogr. B. Analyt. Technol. Biomed. Life [26] Organisation for economic co-operation and Sci. 832 (2006) 81–9. development(OECD), OECD Principles of Good [22] M. Wood, G. De Boeck, N. Samyn, M. Morris, Laboratory Practice, (1998). D.P. Cooper, R.A.A. Maes, E.A. De Bruijn, De- [27] D.R. Lide, CRC Handbook of Chemistry and velopment of a rapid and sensitive method for Physics, 78th ed., CRC Press, 1997.

5.1

Chapter 5.2

Pharmacokinetics of sustained-release oral dexamphetamine sulphate in cocaine and heroin dependent patients

M. Herbrink1, M. Nuijten2, B. Nuijen1, A.D.R. Huitema1,3, J.H. Beijnen1,4, V.M. Hendriks2,6, P. Blanken2, A. Janmohamed7, W. van den Brink8

1Department of Pharmacy and Pharmacology, The Netherlands Cancer Institute – Antoni van Leeuwenhoek Hospital, Amster- dam, The Netherlands. 2Parnassia Addiction Research Center (PARC), Brijder Addiction Care, The Hague, The Netherlands. 3Department of Clinical Pharmacy, University Medical Center Utrecht, Utrecht, The Netherlands. 4Department of Pharmacoepidemiology and Clinical Pharmacology, Science Faculty, Utrecht Institute for Pharmaceutical Sci- ences, Utrecht University, Utrecht, The Netherlands. 6Curium, Leiden University Medical Centre, Department of Child and Adolescent Psychiatry, Leiden University, Leiden, The Netherlands 7CHEOS (Centre for Health Evaluation and Outcome Sciences) Providence Health Care Research Institute, Vancouver, Canada 8Academic Medical Center, Department of Psychiatry, Amsterdam, The Netherlands 246 | Chapter 5.2

Abstract

Research has shown that sustained-release (SR) dexamphetamine is a promising agonist treatment for cocaine dependence. However, little is known about the pharmacokinetics of SR oral dexamphetamine. This study examined the pharmacokinetics of a new SR dexamphetamine formulation in cocaine plus heroin dependent patients currently in heroin-assisted treatment. The study was designed as an open-label pharmacokinetic study in two cohorts: n=5 with once daily 60 mg and n=7 with once daily 30 mg SR oral dexamphetamine. Five days of blood plasma dexamphetamine concentrations measured with liquid chromatography-mass spectrometry with pharmacokinetic parameter estimates using non-compartmental analysis and non-linear mixed effects modelling. Twelve cocaine plus heroin dependent patients in heroin-assisted treatment were included. The initial cohort 1 dose of 60 mg once daily was adjusted to 30 mg following mild to moderate adverse events. Following oral administration, maxt values (CV%) were

6.0 (17.0%) and 6.3 (16.3%) hours and t1/2 were 11 (24.6%) and 12 (25.4%) hours for 60 mg and 30 mg SR dexamphetamine, respectively. After three days, CSSmax values were reached at 90.2 (17.0%) ng/mL and 46.7 (16.7%) ng/mL, whereas CSSmin values were 37.9 (7.53%) ng/mL and 20.3 (9.85%) ng/mL for 60 mg and 30 mg, respectively. The investigated SR formulation of dexamphetamine showed favorable slow-release characteristics in cocaine and heroin dependent patients. A dose-proportional steady-state concentration was achieved within three days. These findings support the suitability of the SR formulation in the treatment of cocaine dependence. Pharmacokinetics of SR dexamphetamine 247

Introduction

Dexamphetamine (dextroamphetamine, S-amphetamine) is a central nervous system stimulant [1]. It exerts its pharmacological action through the rapid increase of dopamine levels in the striatum and noradrenaline levels in the prefrontal cortex [2,3].

In Europe, immediate-release dexamphetamine is used in the treatment of attention deficit hyperactivity disorder (ADHD) for children between the ages of 6 and 17 years if methylphenidate and atomoxetine are not effective [4,5]. In the United States, in addition to ADHD, dexamphetamine is also licensed for the treatment of narcolepsy [6].

After oral administration, immediate-release (IR) dexamphetamine displays a peak plasma concentration between 1 and 4 hours [7]. Mono-amino oxidase (MAO), CYP2C-iso- zymes and CYP2D6 are involved in the metabolism of dexamphetamine [8]. Interactions between dexamphetamine and inhibitors of MAO have been described in the literature and have been linked with symptoms of toxicity [9]. There is also a potential for interactions between dexamphetamine and CYP2D6 inhibitors [10]. Elimination of unchanged dexamphetamine and metabolites mainly occurs via the kidneys [11]. The elimination half-life of dexamphetamine is reported to be between 7 and 12 hours when subject urine is acidic. When the urine is alkaline, the half-life may increase to 18-34 hours [12]. There is an increasing interest for sustained-release (SR) dexamphetamine as a potential treatment of cocaine dependence, for which no medication is yet approved [13]. Several reviews and studies support the efficacy of SR dexamphetamine as an agonist treatment for cocaine dependence [13–18]. In these patients, a SR formulation is preferred over the IR formulation as it allows drug release (and effect) through-out the day after a single dose that increases medication compliance, where an IR formulation would require multiple doses at several time points during the day [19]. Additionally, a SR formulation minimizes the potential for abuse of the medication [14]. 5.2 Finally, studies suggest that SR dexamphetamine is more effective in reducing cocaine use than IR formulations [16,17,20].

We developed a new SR formulation of dexamphetamine sulphate for the treatment of cocaine dependence. These dexamphetamine sulphate 30 mg SR tablets were designed and produced with an in vitro dissolution characteristic that is similar to Dexedrine® spansules. The similarity was assessed using the well described f2 similarity factor. This 248 | Chapter 5.2

single-value outcome factor uses mathematical indices to define similarity between dissolution curves [21,22]. Based on these results, a single daily dose was deemed adequate in providing day-long exposure.

In a recent study, we have shown that our SR dexamphetamine formulation was well accepted, effective and safe in treatment-refractory cocaine and heroin dependent patients currently in heroin-assisted treatment [18].

However, the pharmacokinetics (PKs) of the new formulation are not yet studied. This is important because this provides information on the mean level and duration of the dexamphetamine plasma concentrations and their inter- and intra-patient variabilities. The effect of dexamphetamine on the reduction in (crack-)cocaine use is mediated through a plasma-concentration-linked dose-dependent increase of dopamine levels from dopaminergic neurons in the mesolimbic system [23–25]. This response has been shown to be highly variable between individuals [26]. Given this dose-dependent dopamine elevation, knowledge about (variablility in) plasma levels in the target patient population may help in the evidence-based application of this formulation in the treatment of cocaine dependence, including the safety profile of the formulation.

Additionally, chronic use of cocaine and heroin can have profound effects on organ and tissue systems (e.g. liver, kidney and fat) that can alter the pharmacokinetics of a medicinal product [27–30]. Therefore, a representative PK study is preferred to be performed in patients from the target population.

This single-center, open-label descriptive PK study is the first that aims to investigate the PK properties of an SR dexamphetamine formulation in the target population of cocaine and heroin dependent patients currently in heroin-assisted treatment. Pharmacokinetics of SR dexamphetamine 249

Materials and methods

Study design This was a single center, open-label descriptive exploratory phase I study in crack-cocaine dependent patients that previously participated in a placebo-controlled randomized clinical trial (RCT) with our SR dexamphetamine formulation [18]. Patients for that study were cocaine dependent patients receiving pharmaceutical grade heroin along with oral methadone for their concurrent heroin dependence in a supervised heroin-assisted treatment program in Amsterdam, The Netherlands (for details see supplemen- tary material).

Patients 25 years or older that previously participated in the SR dexamphetamine arm of the previous RCT were eligible. The inclusion criteria of the previous study, and consequently this study, were: daily heroin use, and poor physical, mental or social functioning, a DSM-5 diagnosis of cocaine dependence in the past year and the previous five years, cocaine use on at least 8 days in the previous month, using cocaine primarily by means of basing, at least two earlier failed treatments aimed to reduce or abstain from cocaine use (treatment-refractory).Exclusion criteria were: severe medical problems (e.g. cardiovascular disease, hyperthyroidism, hypersensitivity to sympathico-mi- metic amines and hyperexcitability), pregnancy or breastfeeding, pharmacotherapy with a drug that could potentially influence the pharmacodynamics and/or pharmacokinetics of dexamphetamine (e.g. non-selective MAO-inhibitors, cobicistat, ritonavir, acetazol- amide, ammonium chloride, alcohol and disulfiram), insufficient command of the Dutch language, current participation in another addiction treatment trial, and intake of glu- tamic acid, ascorbic acid or juices.

Study treatment consisted of a once-daily single oral dose of SR dexamphetamine at least 1 hour before a standardized breakfast at the treatment unit for four consecutive days. Study medication was dispensed during the morning visit at the heroin-assisted 5.2 treatment facility and had to be taken under supervision. Standardized breakfast consisted of a cheese sandwich and juice.

As this population is quite vulnerable, certain precautions had to be taken in the design of the study. Continuation of the heroin-assisted treatment required flexibility of the patient sampling scheme and the number of blood samples was therefore minimized. The first cohort started with the same daily dose as was prescribed in the previous RCT, 250 | Chapter 5.2

i.e. 60 mg once daily (two tablets of 30 mg). The second cohort received 30 mg once daily. Dexamphetamine sulphate in polymer matrix SR tablets 30 mg were developed and produced by the Pharmacy of the Antoni van Leeuwenhoek Hospital and MC Slotervaart (Amsterdam, The Netherlands). The tablets consisted of dexamphetamine sulphate and an SR excipient and exhibited slow release dissolution characteristics in vitro.

Urine samples were collected at t = 12 and t = 72 hours after the first intake of dexamphetamine and were analyzed to determine the pH value as the renal excretion of dexamphetamine is dependent on the acidity of the urine. Additional assessments included blood sampling, medical monitoring of heart rate, blood pressure, body weight, co-medication, drug abuse, and daily registration of supervised medication adherence. Participants received a maximum of € 80 as remuneration of travel expenses. The insti- tutional medical ethics committee of the Academic Medical Centre approved the study protocol. Written informed consent was obtained from all patients. The study was conducted in accordance with the ethical principles of the Declaration of Helsinki and Good Clinical Practice.

Pharmacokinetic and bioanalysis The PK sampling scheme was divided into five critical time points (t= 0, 24, 48, 72 and 96 hours after first dose), where a missed PK sample meant patient exclusion, and eight lenient time points (t= 0.5, 1, 2, 4, 6, 8, 10, and 30 hours after first dose (±1 hour)), where a missed sample could be drawn within 1 hour prior to or 1 hour after the intended time point. This measure was taken to avoid additional patient stress and to reduce the impact of possible low patient adherence to the schedule. Delayed and/or incomplete sampling is inherent to this scheme. Certain time points were adjusted to fit the daily schedule of the heroin-assisted treatment of each individual patient. The blood samples were taken at the integrated treatment facility.

Blood samples (4 mL) were drawn in K2-EDTA tubes through intravenous catheters or venipuncture, on the patients’ preference, at baseline and at the designated time points. All blood samples were immediately centrifuged at 3000 g for 10 min. at 20-25 oC. Plas- ma was stored at or below -20 oC until analysis. Dexamphetamine plasma levels were quantified by use of validated high-performance liquid chromatography with tandem mass spectrometry detection (LC-MS/MS) with a lower limit of quantification (LLOQ) of 2.5 ng/mL, intra- and inter-assay bias below 15%, and a sample stability of at least 17 months at -20 oC [31]. Pharmacokinetics of SR dexamphetamine 251

To correct for the possible effect of the urine acidity on the excretion of dexamphetamine, the urine pH was measured at two different time points in the PK sampling scheme.

Pharmacokinetic endpoints The primary PK endpoints were time at which the highest plasma concentration occurs after administration of a dose (tmax) and terminal elimination half-life (t1/2) as to assess the suspected SR character of the formulation. Secondary PK endpoints included the minimum concentration of drug in plasma at steady state (CSS,min), and observed maximum plasma concentration (Cmax) at the first and second day.

Data analysis Dexamphetamine plasma concentrations and relative time values were used to calculate the PK parameters using non-compartmental analysis and individual modelling. The calculation of PK parameters was performed using a validated platform (R version 3.3.2).

Elimination half-life (t1/2) was calculated from t1/2 = ln(2)/ke, where ke, the elimination constant, was derived from the last 2 or 3 time points of the linear terminal phase of the logC-time plotted data.

Descriptive statistics were calculated for baseline demographic and clinical data to check for any group differences. PK parameters were expressed by geometric means and coefficients of variation (%CV) by dose.

Power considerations Sample size for this study was not based on explicit power and type I error assumptions, but rather on empirical considerations and on the experiences in previous studies; thus no formal sample size calculation was performed. Given the vulnerability and special characteristics of the study population, a limited but still significant number of 10 to 12 5.2 patients was expected to be included.

Safety All signs and symptoms observed from the time of first dose were recorded as Adverse Events (AEs) and were graded according to the Common Terminology Criteria for Ad- verse Events (CTCAE) version 4.0 [32]. All AEs were recorded until 7 days after the last administration of study treatment. After this follow-up period, only new or ongoing treat- 252 | Chapter 5.2

ment-related AEs were recorded, except for ongoing serious AEs, which were assessed until resolution or stabilization, regardless of relationship to study treatment.

Results

Study participants Between August 2016 and May 2017, 14 patients were screened for eligibility of which 12 were enrolled: five in the first cohort and seven in the second cohort.

Included patients were of different ethnic background and mostly male (table 1). Mean age and body weight were similar in the two cohorts. Renal function based on creatinine clearance was well above 60 mL/min for all patients. Liver enzyme values aspartate aminotransferase (AST) and alanine aminotransferase (ALT) were all below 30 IU/L and similar in both cohorts. Mean body mass index (BMI) were also found to be similar. All patients participated in heroin-assisted treatment with inhalable heroin for at least one year and smoked crack-cocaine.

In the first cohort, two of the five patients dropped out. However, all five patients did complete the first PK sampling day, which provided the most crucial data for the primary PK endpoints. The first drop-out was caused by a grade 1-2 (mild to moderate) insomnia and related fatigue and the second drop-out was caused by a failure to show up for the third critical sampling time point at 48 hours.

The dose of 60 mg once daily was lowered to 30 mg once daily for the second cohort in consideration of the experienced adverse events (AEs). AEs did not lead to a premature end of study in cohort 2, although one of the seven patients from this cohort dropped out. This drop-out was caused by the failure to appear at the third critical sampling time point and the supervised medication administration at 48 hours.

Pharmacokinetics A total number of 55 plasma samples (out of 65 scheduled samples; 85%) were obtained from the 5 patients in cohort 1 and 88 plasma samples (out of 91 scheduled samples; 97%) were obtained from the 7 patients of cohort 2. Samples were missed due to patient drop-out. Pharmacokinetics of SR dexamphetamine 253

Plasma dexamphetamine concentrations over the first 24 hours are displayed in figure 1. The two dose levels (60 and 30 mg) displayed similar uptake and elimination.

Table 1. Baseline patient characteristics and treatment characteristics Cohort 1 Cohort 2 Patient characteristics Daily dose SR dexamphetamine, mg 60 30 Number of patients, n 5 7 Male/female, n 5/0 5/2 Age, years, mean (SD) 52 (7.5) 51 (10) Ethnicity, n Caucasian 4 3 Asian 0 1 African 1 3 Body weight, kg, mean (SD) 72 (17) 65 (20) Height, cm, mean (SD) 177 (10.5) 171 (11.0) Body mass index (BMI), kg/m2, mean (SD) 21.7 (3.16) 21.3 (3.92) Aspartate aminotransferase (AST), IU/L, mean (SD) 27 (5.6) 29 (8.5) Alanine aminotransferase (ALT), IU/L, mean (SD) 24 (7.0) 27 (5.0) Creatinine clearance, mL/min, mean (SD) 78 (13) 77 (13)

Table 2 presents the pharmacokinetic single dose and estimated multiple dose parameters as calculated by non-compartmental analysis. The tmax values were taken from observed data. The two dose levels exhibit similar absorption characteristics with tmax at

6.0 and 6.3 hours, respectively. Elimination half-life (t1/2) was also similar at 11 h and 12 h for 60 and 30 mg, respectively. The Cmax and Cmin values after the first administration were found to be positively correlated with the dose as well as the values after reaching steady-state, CSSmax, CSSmin and AUCSS.

Two urine samples were taken and analyzed from every participant during the cohorts. All urine samples were found to have pH values between 5 and 6. 5.2

Safety No serious adverse events occurred. The reported adverse events after single and multiple doses of dexamphetamine 60 mg were insomnia (2 cases; 1 grade 2, 1 grade 1) and insomnia in combination with fatigue (1 case; grade 2 insomnia and grade 1 fatigue). The AEs were probably related to the study medication since all AEs were reported in the first 24 hours after the first dose of SR dexamphetamine. It should be noted, however, that 254 | Chapter 5.2

in each of these cases, illicit crack-cocaine was also used during the study treatment. No adverse events were reported after single or multiple doses of 30 mg dexamphetamine. L) L) A C

ng/m 100 ng/m 100

e( 1 e( 1 in in

am 80 2 am 2 phet phet 60 3 3 am am

ex 4 ex 10 4

nd 40 nd io 5 io 5 at at 20 concentr concentr 0 1 ma ma 0510 15 20 25 0510 15 20 25 Plas Plas Time (h) Time (h) L) L) B D

ng/m 100 ng/m 100

e( 1 e( 1 in in

am 80 2 am 2 phet phet 60 3 3 am am

ex 4 ex 10 4

nd 40 nd io 5 io 5 at at 6 6 20 7 7 concentr concentr 0 1 ma ma 0510 15 20 25 0510 15 20 25 Plas Plas Time (h) Time (h)

Figure 1. Individual concentration-time curves of plasma dexamphetamine during the first dosing interval, 0-24 hours: A, of the first cohort, linear (60 mg); B, the second cohort, linear (30 mg); C, the first cohort, logarithmic; D, the second cohort, logarithmic.

Table 2. Non-compartmental pharmacokinetic parameters for SR dexamphetamine, means (CV%). Parameter 60 mg 30 mg (n=5) (n=7)

tmax, h 6.0 (17.0) 6.3 (16.3)

t1/2, h 11.3 (24.6) 12.4 (25.4)

Cmax, ng/mL 76.7 (20.1) 42.5 (13.6)

Cmin, ng/mL 25.6 (41.9) 15.2 (32.6)

CSSmax, ng/mL 100 (27.5) 58.4 (14.4)

CSSmin, ng/mL 39.5 (38.9) 21.77 (19.8)

AUCSS, ng*h/mL 1613 (38.9) 936.5 (31.7) Pharmacokinetics of SR dexamphetamine 255

Discussion

Although marketed SR formulations of dexamphetamine exist, none of these were available in Europe for treatment of this patient population. Therefore, a new SR formulation was developed and produced with a dose unit of 30 mg. This formulation has been used in a previously reported RCT that demonstrated the efficacy of SR dexamphetamine (60 mg/day) compared with placebo in reducing crack-cocaine use in heroin dependent patients currently in heroin-assisted treatment with concurrent treatment-refractory cocaine dependence [13]. However, so far no pharmacokinetic data were available for this formulation.

In this pharmacokinetic study among 12 patients from the previous RCT, 3 of the 5 patients from cohort 1 reported insomnia with 60 mg/day SR dexamphetamine and, therefore, the dose was reduced in the 7 patients of cohort 2 to 30 mg/day. The dose was not expected to have a profound influence on the maxt and t1/2 parameters, which meant that the study could be continued with a reduced 30 mg dose.

The lack of tolerance for the 60 mg/day schedule as compared to the previous effect study is important to address. Previous effect studies with higher daily doses were all set up as double-blind randomized clinical trials. The study described here is the first to have an open label design. The conscious intake of dexamphetamine may have come with an expectancy bias that was less profoundly present in the effect studies. This may have clinical implications. Higher doses than 30 mg/day may still be possible, however, but may require a dose titration approach.

All patients continued regular treatment that included administration of dose-personalized pharmaceutical grade heroin during this study. Literature suggests that there are no direct pharmacokinetic interactions between methadone, heroin or cocaine with dexamphetamine [33,34]. Therefore, the effect of methadone, heroin and cocaine use of the 5.2 patients on the study outcome are not likely.

The assessed clinical parameters of the participating patients for nutritional status (BMI), renal function (creatinine clearance) and liver function (AST and ALT) are all within the accepted normal values for the general population [35–37]. From this, significant effects on the pharmacokinetics of dexamphetamine due to altered tissue or organ systems were not expected. 256 | Chapter 5.2

The studied SR formulation had a maxt of about six hours, which was about twice the tmax of approximately 3 hours of dexamphetamine immediate release formulations that are described in literature [38,39]. Importantly, the observed elimination half-life (t1/2) of our dexamphetamine SR formulation was 11-12 hours, which is consistent with the other long-acting formulations. However, a shorter t1/2 of 8.8 h has been reported for a short-release formulation [38] indicating that the terminal half-life is shorter and may have been estimated too long in the present and other studies using long-acting formulations. The reason for this is likely the small number of samples taken during the elimination phase, for some patients this was below 3 and thus suboptimal for the calculation of t1/2. Additionally, the half-life could have been overestimated as it was influenced by the slow-release formulation through so-called flip flop kinetics [40]. As the excretion of dexamphetamine is pH dependent, the urine pH of the participants was checked [41,42]. We found all pH values to be between 5 and 6. Therefore, urine acidity dependent excretion-linked variability in dexamphetamine PK is unlikely.

The estimated absorption rate parameterk a translates into an absorption half-life of 7.96 hours. Consequently, it would take approximately 3.3 half-life periods (i.e. 26 hours) to absorb 90% of the dexamphetamine. This means that probably not the full dose is absorbed and, hence that the bioavailability may be less than 1.

Steady-state concentrations of dexamphetamine were achieved within three days after the first administration, which is in agreement with the elimination half-life [43]. Patient adherence is, therefore, crucial since the relatively short half-life means that a missed once daily dose sharply reduces the plasma dexamphetamine concentration over the course of 48 hours. This could mean that the therapeutic effect diminishes during such a period.

Toxicity of dexamphetamine has been reported at plasma levels above 200 ng/mL [44].

The predicted CSSmax levels of 100 ng/mL for 60 mg/day and 58.4 ng/mL for 30 mg/day are sufficiently below these earlier reported toxic levels.

Insomnia and fatigue have also been reported previously by Nuijten et al. in 34% of the patients taking 60 mg SR dexamphetamine, compared to 9% in the placebo arm of that study [18]. Most of these AEs were resolved before the end of the study. The incidence of sleep problems in this PK study was seemingly higher, albeit with a limited number of patients. Although there may be a relationship between insomnia and fatigue and the Pharmacokinetics of SR dexamphetamine 257

study medication, this relationship may be compromised by patients lifestyle, infrequent concomitant use of crack-cocaine and other stimulants and the intensive setting of this PK study.

A limitation of this study is the small sample size together with the patients lost to follow up, the missed blood samples and the split cohorts across doses. This means that the results need to be interpreted with caution. Despite these drawbacks, the exploratory and flexible design of the study does allow confirmation of the important sustained-release characteristics of the investigated formulation, and these pharmacokinetics support the previous efficacy study, demonstrating the superiority of SR dexamphetamine over placebo in reducing crack-cocaine use. Furthermore, this study brought to light a previously unidentified factor that may be important for the implementation of this treatment, namely the expectancy bias of patients taking open label dexamphetamine in this setting. An additional unforeseen benefit of the split into two doses is the information obtained from both dose levels, albeit limited in quantity. Overall, although the study has included a small number of patients, the data was sufficient to provide an answer to its objective and added to the still limited experience with this new treatment.

This study is the first to describe the pharmacokinetics of SR dexamphetamine in a patient population with co-morbid crack-cocaine and heroin dependence currently in heroin-assisted treatment. The results of this study show that the investigated formulation of dexamphetamine exhibits SR characteristics in vivo.

Additionally, this study shows the mean steady-state plasma levels at two different dose levels. This data demonstrates a dose proportional PK characteristics. This may prove important at a later stage in the development of therapies with dexamphetamine SR where dose-effect titrations may be necessary.

Further investigation is required into the relationship between plasma levels and effica- 5.2 cy. This might warrant dose-titration protocols and dosing guidance in the near future. In conclusion, the investigated SR dexamphetamine formulation is suitable for use in the heroin-assisted treatment setting providing a once daily dosing option resulting in an adequate accumulation rate to steady-state levels. 258 | Chapter 5.2

Acknowledgements

The authors thank the participating patients, committed medical staff, pharmacists, and other collaborators of the contributing hospitals and GGD Amsterdam for their contribu- tions to the execution of the study.

References

[1] E. Sinita, D. Coghill, The use of stimulant med- [10] J.S. Markowitz, K.S. Patrick, Pharmacokinetic ications for non-core aspects of ADHD and in and Pharmacodynamic Drug Interactions in other disorders, Neuropharmacology 2014; the Treatment of Attention-Deficit Hyperac- 87: 161–172. tivity Disorder, Clin Pharmacokinet 2001; 40: [2] R. Kuczenski, D.S. Segal, Effects of methylphe- 753–772. nidate on extracellular dopamine, serotonin, [11] L.G. Dring, R.L. Smith, R.T. Williams, The met- and norepinephrine: comparison with am- abolic fate of amphetamine in man and other phetamine., J. Neurochem 1997; 68: 2032–7. species., Biochem J 1970; 116: 425–35. [3] D.J. Heal, S.L. Smith, R.S. Kulkarni et al., New [12] E. Anggård, L.E. Jönsson, A.L. Hogmark et al., Am- perspectives from microdialysis studies in free- phetamine metabolism in amphetamine psycho- ly-moving, spontaneously hypertensive rats on sis, Clin Pharmacol Ther 1973; 14: 870–80. the pharmacology of drugs for the treatment [13] X. Castells, R. Cunill, C. Pérez-Mañá et al., Psy- of ADHD., Pharmacol Biochem Behav 2008; 90: chostimulant drugs for cocaine dependence, 184–97. Cochrane Database Syst Rev 2016; 9. [4] P. Hodgkins, M. Shaw, S. McCarthy et al., The [14] D. V Herin, C.R. Rush, J. Grabowski, Agonist-like Pharmacology and Clinical Outcomes of Am- pharmacotherapy for stimulant dependence: phetamines to Treat ADHD, CNS Drugs 2012; preclinical, human laboratory, and clinical 26: 245–268. studies., Ann N Y Acad Sci 2010; 1187: 76–100. [5] S. King, S. Griffin, Z. Hodges et al., A systematic [15] J. Grabowski, H. Rhoades, J. Schmitz et al., review and economic model of the effective- Dextroamphetamine for cocaine-dependence ness and cost-effectiveness of methylpheni- treatment: a double-blind randomized clinical date, dexamfetamine and atomoxetine for the trial., J Clin Psychopharmacol 2001; 21: 522–6. treatment of attention deficit hyperactivity [16] J. Grabowski, H. Rhoades, A. Stotts et al., Ag- disorder in children and adolescents, Health onist-like or antagonist-like treatment for co- Technol Assess 2006;10: iii–iv, xiii-146. caine dependence with methadone for heroin [6] M.J. Thorpy, Y. Dauvilliers, Clinical and practical dependence: two double-blind randomized considerations in the pharmacologic manage- clinical trials., Neuropsychopharmacology ment of narcolepsy, Sleep Med 2015; 16: 9–18. 2004; 29: 969–81. [7] B. Angrist, J. Corwin, B. Bartlik et al., Early [17] J. Shearer, A. Wodak, I. van Beek et al., Pilot pharmacokinetics and clinical effects of oral randomized double blind placebo-controlled D-amphetamine in normal subjects., Biol Psy- study of dexamphetamine for cocaine depen- chiatry 1987; 22: 1357–68. dence., Addiction 2003; 98: 1137–41. [8] T. Kraemer, H.H. Maurer, Toxicokinetics of [18] M. Nuijten, P. Blanken, B. van de Wetering amphetamines: metabolism and toxicokinetic et al., Sustained-release dexamfetamine in data of designer drugs, amphetamine, meth- the treatment of chronic cocaine-dependent amphetamine, and their N-alkyl derivatives., patients on heroin-assisted treatment: a ran- Ther Drug Monit 2002; 24: 277–89. domised, double-blind, placebo-controlled tri- [9] M. Wimbiscus, O. Kostenko, D. Malone, MAO al, Lancet 2016; 387: 2226–2234. inhibitors: risks, benefits, and lore., Cleve Clin J [19] S.J. Tulloch, Y. Zhang, A. McLean et al., SLI381 Med 2010; 77: 859–82. (Adderall XR), a Two-Component, Extend- Pharmacokinetics of SR dexamphetamine 259

ed-Release Formulation of Mixed Amphet- Pharm Biomed Anal 2018; 148: 259–264. amine Salts: Bioavailability of Three Test For- [32] Common Terminology Criteria for Adverse mulations and Comparison of Fasted, Fed, and Events (CTCAE) version 4.03. Rockville, MD: Sprinkled Administration, Pharmacotherapy National Cancer Institute; 2009. Updated June 2002; 22: 1405–1415. 2010. [20] M.E. Mooney, D. V. Herin, S. Specker et al., Pilot [33] W.T. Lindsey, D. Stewart, D. Childress, Drug In- study of the effects of lisdexamfetamine on co- teractions between Common Illicit Drugs and caine use: A randomized, double-blind, place- Prescription Therapies, Am J Drug Alcohol bo-controlled trial, Drug Alcohol Depend 2015; Abuse 2012; 38: 334–343. 153: 94–103. [34] E.F. McCance-Katz, L.E. Sullivan, S. Nallani, Drug [21] J.W. Moore, H.H. Flanner, Mathematical Com- Interactions of Clinical Importance among the parison of Curves with an Emphasis on in Vitro Opioids, Methadone and Buprenorphine, and Dissolution Profiles, Pharm Technol 1996; 20: Other Frequently Prescribed Medications: A 64–74. Review, Am J Addict 2010; 19: 4–16. [22] V.P. Shah, Y. Tsong, P. Sathe et al., In vitro disso- [35] F.Q. Nuttall, Body Mass Index, Nutr Today lution profile comparison--statistics and analy- 2015; 50: 117–128. sis of the similarity factor, f2., Pharm Res 1998; [36] P. Delanaye, E. Schaeffner, N. Ebert et al., Nor- 15: 889–96. mal reference values for glomerular filtration [23] R.B. Rothman, B.E. Blough, M.H. Baumann, rate: what do we really know?, Nephrol Dial Dopamine/serotonin releasers as medications Transplant 2012; 27: 2664–2672. for stimulant addictions., Prog Brain Res 2008; [37] E.G. Giannini, R. Testa, V. Savarino, Liver en- 172: 385–406. zyme alteration: a guide for clinicians., CMAJ [24] M.J. Ferris, E.S. Calipari, J.H. Rose et al., A Sin- 2005; 172: 367–79. gle Amphetamine Infusion Reverses Deficits in [38] P.C. Dolder, P. Strajhar, P. Vizeli et al., Phar- Dopamine Nerve-Terminal Function Caused by macokinetics and Pharmacodynamics of Lis- a History of Cocaine Self-Administration, Neu- dexamfetamine Compared with D-Amphet- ropsychopharmacology 2015; 40: 1826–1836. amine in Healthy Subjects, Front Pharmacol [25] W.P. Melega, A.E. Williams, D.A. Schmitz et 2017; 8: 617. al., Pharmacokinetic and pharmacodynamic [39] Y.N. Wong, L. Wang, L. Hartman et al., Compar- analysis of the actions of D-amphetamine and ison of the single-dose pharmacokinetics and D-methamphetamine on the dopamine termi- tolerability of modafinil and dextroamphet- nal., J Pharmacol Exp Ther 1995; 274: 90–6. amine administered alone or in combination [26] C.T. Smith, L.C. Dang, R.L. Cowan et al., Variabil- in healthy male volunteers., J Clin Pharmacol ity in paralimbic dopamine signaling correlates 1998; 38: 971–8. with subjective responses to d-amphetamine, [40] J.A. Yáñez, C.M. Remsberg, C.L. Sayre et al., Neuropharmacology 2016; 108: 394–402. Flip-flop pharmacokinetics – delivering- are [27] I. Riezzo, C. Fiore, D. De Carlo et al., Side ef- versal of disposition: challenges and oppor- fects of cocaine abuse: multiorgan toxicity and tunities during drug development, Ther Deliv pathological consequences., Curr Med Chem 2011; 2: 643–672. 2012; 19: 5624–46. [41] E. Anggard, L.-M. Gunne, L.-E. Jonsson et al., [28] L. Karila, A. Petit, W. Lowenstein et al., Diag- Pharmacokinetic and clinical studies on am- nosis and consequences of cocaine addiction., phetamine dependent subjects, Eur J Clin Curr Med Chem 2012; 19: 5612–8. Pharmacol 1970; 3: 3–11. [29] A.V. Crowe, Substance abuse and the kidney, [42] A.H. Beckett, M. Rowland, Urinary excretion ki- QJM 2000; 93: 147–152. netics of amphetamine in man., J Pharm Phar- 5.2 [30] K.D. Ersche, J. Stochl, J.M. Woodward et al., macol 1965; 17: 628–39. The skinny on cocaine: insights into eating be- [43] M. Rowland, T.N. Tozer. Extravascular Dose. In: havior and body weight in cocaine-dependent Clinical Pharmacokinetics; Concepts and Ap- men., Appetite 2013; 71: 75–80. plications. Philadelphia: Lippincott, Williams & [31] M. Herbrink, B. Thijssen, M.J.X. Hillebrand et Wilkins, 1994; 34:3. al., Development and validation of a high-per- [44] M. Schulz, S. Iwersen-Bergmann, H. Andresen formance liquid chromatography-tandem et al., Therapeutic and toxic blood concentra- mass spectrometry assay for the quantifica- tions of nearly 1,000 drugs and other xenobi- tion of Dexamphetamine in human plasma., J. otics, Crit Care 2012; 16: R136.

Chapter 6

Conclusions and perspectives

Conclusion & Perspectives | 263

Conclusions and perspectives

The studies and the results described in this thesis are the resultants of two very different lines of research. Even though these studies share similar features, they stem from vastly dissimilar therapeutic fields and backgrounds. For that reason this thesis is divided in two parts; the first presenting the studies regarding anticancer agents and the second describing the studies concerning the stimulant dexamphetamine.

Part I

For the past 20 years, the pharmacological therapy of cancer has been undergoing a metamorphosis, rapidly changing from treatment with broad and unspecific cytotoxic agents to personalized and highly specific targeted agents. This new class of drugs largely consists of small molecular compounds that mostly inhibit kinases. Alongside the advancing targeted therapy of the small molecular kinase inhibitors (smKIs) came the so-called ‘intravenous-to-oral-switch’. This is illustrated by the fact that an increasing number of marketed targeted anticancer drugs are administered orally. Additionally, most of the targeted agents that are being investigated are intended for oral ingestion.

Oral targeted cancer therapy The first chapter of this thesis shows that although the therapy of cancer is increasingly becoming more sophisticated in terms of specificity and effectiveness, the clinical and therapeutic application has not yet been fully refined.

Chapter 1.1 presents an inventory and an analysis of the therapeutic variability that is observed in the clinical application of the smKI drug class and the various factors that have a crucial influence thereon. The overview shows that the kinase smKIs in general suffers from a low and highly variable bioavailability. The underlying cause is the poor permeability in the gastrointestinal tract, but even more so the poor solubility of these anticancer agents. This is highlighted by the fact that the majority of these drugs are marked with Biopharmaceutics Classification System (BCS) class II or IV [1]. Various phys- 6 iological factors can contribute further to increase the variability, such as membrane transporters and metabolism. Furthermore, the initial high variability can be higher in certain therapeutic situations. Special patient populations such as the hepatically impaired and the elderly may be at an additional risk. Another important factor that may 264 | Chapter 6

contribute to the observed variability is the co-administration of food and particular co-medication. This chapter shows that most orally administered kinase inhibitors are susceptible to variability from different sources, but also that large amounts of necessary descriptive data are still lacking to accurately describe the overall situation. Initial evidence is presented that therapeutic, but most notably formulative, measures may be valuable tools to increase drug bioavailability and possibly reduce pharmacokinetic variability, however. This may not only be true for the currently marketed drugs, but also for the upcoming generation and should therefore receive sufficient attention during pharmaceutical development.

The fundamental causes of the low and variable bioavailability are addressed in more detail in chapter 1.2. Here, the molecular structures and scaffolds of the kinase inhibitors are the main focus. Each of the kinase inhibitors is specifically designed on a molecular level to bind to a protein and inhibit its signaling function and thereby preventing the execution of its role in tumor growth. This has resulted in a collection of highly lipophilic structures that play critical roles in target binding. As combinations of these lipophilic structures make up the larger part of the kinase inhibitor molecules, the overall structure is subsequently very poorly water soluble with often a high dependence on the pH level. For the kinase inhibitor drug class, salt and solvate selection prior to marketing has not led to solubility improvement beyond BCS class II or IV for most of the anticancer agents. Furthermore, research into forms other than crystals has rarely led to clinical formulations. The marketed formulations are therefore all, with the exception of vemurafenib, regorafenib and nintedanib, developed as ‘classical’ physical mixtures with these crystalline salts and solvates that are not capable of fully dealing with the inherent poor solubility of the drugs’ inhibiting structure.

Published literature and patent texts hint at various ways that may be used to improve drug formulations of kinase inhibitors, although substantial in vivo results are still lacking.

With the support of literature and patent data and the analyses from the previous chapters, chapter 1.3 continues the discussion by taking a strong stance in favor of more thorough formulation development of future clinical drug products. It does so by high- lighting practical cases and situations in which already marketed kinase inhibitors not only suffer from a low solubility, but also from a low and highly susceptible bioavailability. Through an analysis of the previous decades and a extrapolation to the foreseeable future, the chapter predicts that a relatively large number of new kinase inhibitors will Conclusion & Perspectives | 265

be introduced before long. In order to facilitate the clinical practice and the financial bedding of the therapy with all these new drugs, the pharmaceutical potentials should be optimized. In conclusion, this should translate to more investments by innovators in the formulations of chemotherapy.

To further investigate the hypothesis that an improved drug solubility indeed leads to an increased and less variable bioavailability and to further support previous observations, a new research line was set up. Here, the main objective was to develop new and solubility-enhanced oral formulations of kinase inhibitors from a firm basis of adequate and extensive (bio)pharmaceutical profiling. Based on their unfavorable pharmaceutical properties and therapeutic variability, two kinase inhibitors, namely pazopanib and nilotinib, were selected as candidates for formulation improvement.

Pharmaceutical analysis of oral anticancer drug substances The first step in the design of a new formulation should always be the adequate pharmaceutical characterization of the drug substance [2]. Both pazopanib and nilotinib were extensively studied in terms of their thermal and thermodynamic properties in relation to possible pharmaceutical processing [3]. The results of these studies are presented in chapter 2.

The thermal stability and the thermal degradation of pazopanib hydrochloride as polymorph form I is resolved and discussed in chapter 2.1. It was shown that Fourier transform infra-red spectroscopy (FT-IR) can be used to differentiate between the -hy drated and anhydrate forms of pazopanib hydrochloride and that the presence of the hydrochloride in the drug substance is also observable. This indicates that FT-IR may be used during thermal processing to identify drug substance degradation. Pazopanib hydrochloride subjected to thermogravimetric analysis (TGA) revealed thermal stability up to 200 oC and a five-stage degradation process. An important observation came from calorimetry analyses. Here, it was established that pazopanib hydrochloride form I does not melt prior to thermal decomposition. This has an important implication for the choice of future processing methods as molecular dispersions cannot be obtained through melting techniques. Overall, the results from this study support the conclusion 6 that the high thermal stability makes pazopanib hydrochloride a suitable candidate for new formulation studies where thermal production methods, excluding melting-based techniques, are utilized. 266 | Chapter 6

The preparation of crystal-free amorphous phase has been examined for many drugs as an approach to increase the dissolution properties and the solubility. The amorphous form lacks the particulate structure of crystalline forms which often sharply increases the overall surface area and drastically increases the dissolution speed. Furthermore, the relatively high free energy of the amorphous phase generally causes an increased initial apparent solubility. This heightened free energy may also lead to increased tendency to collapse and recrystallize however. This means that besides a well-designed stability study of the crystalline form, the assessment of the stability of the amorphous phase should also be part of formulation development. Chapter 2.2 applies these principles to nilotinib hydrochloride polymorph form B in a pharmaceutical characterization study. The monohydrated hydrochloride salt of nilotinib was successfully subjected to spray-drying to yield anhydrous and amorphous nilotinib hydrochloride. The results of the study show that FT-IR can also be utilized to indicate the presence of water and the amorphicity of the spray-dried powder. Thermal analysis by TGA of the crystalline drug demonstrated stability during heating up to 193 oC. Modulated temperature differential scanning calorimetry (MTDSC) was employed to thoroughly study and uncover various stability parameters of the crystalline and the spray-dried amorphous phase. Loss of water of crystallization from the crystalline drug occurred with a peak of 133 oC after which melting was observed with a peak temperature of 199 oC. The mean glass transition temperature of the amorphous phase was found to be 147 oC, indicating a stable spray-dried powder. Heat capacities of both forms were used to calculate several thermodynamic parameters. These parameters indicated that the amorphous phase had a worst-case-scenario Kauzmann temperature (where molecular mobility is negligible) of 65 oC and the characteristics of a fragile amorphous ‘glass’ system. Along with a physical stability of at least 6 months at 20-25 oC and a relative humidity of 60%, this signifies a highly stable amorphous phase at common storage conditions with a favorable stability profile at elevated temperatures. It was therefore concluded that amorphous nilotinib hydrochloride is a suitable candidate for formulation development.

Both studies revealed and highlighted important drug substance properties that may contribute to the efficient and accurate formulation development.

Formulation improvement of oral anticancer agents A large number of techniques are available to a pharmaceutical formulation scientist with which the solubility of a poorly soluble drug can be improved. This can be done by either altering the physical attributes of the drug itself or through the combination of the drug with solubility-enhancing excipients. Conclusion & Perspectives | 267

The two drugs of choice, pazopanib hydrochloride and nilotinib hydrochloride, have both been marketed in pharmaceutical formulations that consist of simple powder mixtures. The compositions and manufacturing procedures seem to match appropriate sta- bilities and dissolution properties, not to maximize the biopharmaceutical potential of both drugs.

Chapter 3 presents and discusses the results from the studies that did seek to optimize these potentials for pazopanib hydrochloride in chapter 3.1 and for nilotinib hydrochloride in chapter 3.2.

To accurately and to properly evaluate the solubility improvement of different formulations with these drugs, a (small-scale) dissolution system was developed that was capable of simulating the changes in composition and pH as these are encountered in the gastrointestinal system. Scrutiny was applied to the marketed formulations Votri- ent®(pazopanib hydrochloride) and Tasigna®(nilotinib hydrochloride) in this new dissolution system. It showed that the maximum solubility in the stomach environment was up to approximately 48% for Votrient® and 26% for Tasigna®. Upon transition to the intestinal environment, the solubility of both drugs dropped below 1% at final intestinal pH values. The latter is critical as most of the drug uptake is likely to take place in the small intestine for which drug in solution is necessary. Both studies therefore focused on enhancing the solubility of the drugs in the stomach environment and on preventing the solubility loss in the intestinal environment. An extensive number of excipients in combination with both drugs at different compositions were examined in terms of compatibility, solubility improvement and dissolution characteristics.

The solubility of pazopanib hydrochloride in the stomach environment was increased slightly to a maximum of 54% by the addition of certain excipients in a binary powder mixture. The solubility of pazopanib hydrochloride in the intestinal environment could only be increased and maintained by the excipient Soluplus® through an unexpected solubility increase upon transition from stomach to intestinal conditions. At a composition of 1/8 (w/w) (pazopanib hydrochloride/Soluplus®) and after mixing of the components with a mixing and grinding technique in the presence of methyl-tert-butyl ether as a wet- 6 ting agent, the solubility could even be increased to 100% at a pH of 6.8 through incorporation of pazopanib hydrochloride in Soluplus® micelles. A solubility increase with a factor of >300 was obtained with the newly developed formulation. The powder mixture formulation was brought up to volume with microcrystalline cellulose and was filled out 268 | Chapter 6

in gelatin capsules at 25 mg of pazopanib hydrochloride per capsule. This new dosage form was named PazSol001 (PZHSol001) for which a clinical study was designed to determine its pharmacokinetic profile. Doses of 100, 200 and 300 mg were administered to cancer patients and the plasma concentrations after the single dose were measured. In a comparison with literature data of the marketed Votrient®, the first dose level of 100 mg did not significantly improve patient exposure. The second and third dose levels, however, did show an increase in exposure compared to Votrient®. The dose level of 200 mg exhibited an exposure increase of 84% relative to the same dose of Votrient®. The 300 mg dose of PazSol001 was found to reach an exposure of 379 μg/mL*h, that was similar to the exposure of 800 mg of Votrient® at 275 μg/mL*h. Additionally, the exposures of PazSol001 exhibited less variability than those of Votrient®.

The results from this study described chapter 3.1 clearly show the additional benefit of an extensive and physiologically-based pharmaceutical formulation development process. The results from this proof-of-concept study are promising and may form the basis for additional investigation. This research should be directed to further optimize the- for mulation and the production process. The initial hypothesis that an increased solubility leads to an increased and less variable bioavailability is supported by this study, however. In contrast to the sudden and unexpected solubility increase that was observed for the combination of pazopanib hydrochloride and Soluplus®, the solubility of nilotinib hydrochloride could only be maintained upon transition from the stomach to the intestine. An additional step was needed to further improve the solubility in the stomach environment. In chapter 3.2 it was shown that by spray drying nilotinib hydrochloride from ethanol, the initial solubility in the stomach environment could be significantly increased. This is most likely caused by the higher apparent solubility of the amorphous from and the altered interaction between the drug molecules and the excipients. Excipients were selected based on the solubility improvement and attainment in the stomach and intestinal environments, respectively. The results indicated that Soluplus® in a composition of 1/7 (w/w) (nilotinib hydrochloride/Soluplus®) was able to not only increase the stomach solubility to 100%, but also maintain that solubility at intestinal pH levels. Hence, this formulation is deemed a suitable candidate for evaluationin vivo. This will further clarify if the principle of the initial hypothesis is also applicable to nilotinib hydrochloride and may perhaps even be more generalizable. Conclusion & Perspectives | 269

Bioanalysis of oral anticancer agents The variable pharmacokinetic nature of the current kinase inhibitors and other personalized chemotherapy are likely to influence the effectiveness of their therapy. This is further supported by the increasing insight in the relationship between certain pharmacokinetic targets and the effect on disease outcome and patient survival [4]. Therapeutic drug monitoring (TDM) provides the clinicians with a valuable way of evaluating the plasma levels of cancer patients and may use the information to adjust dosing schemes to improve therapy outcome. To accommodate the TDM for frequently used kinase inhibitors and targeted agents in the clinic, two bioanalytical quantitative methods were developed. The first method, as presented inchapter 4.1 was designed for the determination of plasma concentrations for 10 kinase inhibitors and 1 metabolite. Chapter 4.2 discusses the second method, which is developed to quantify 7 kinase inhibitors and 1 antihormone. Both methods were set-up to match the fast and high through-put clinical setting.

The measurements of plasma concentrations by both assays has since the validation been adopted into the clinical routine. The bioanalysis assists in not only the monitoring of plasma concentrations and the feedback to the clinicians, but also in providing information about the overall performance of the analyzed drugs.

In line with the continuing emergence of new drugs and the replacement of old drugs in certain arms of oncological therapy, the agents in the TDM methods may over time become obsolete. In practice that means that methods may need to be adjusted to incor- porate the quantification of other drugs or assays may be abandoned altogether. What- ever direction is chosen, the fundaments of potential new TDM assays and the logistics and infrastructure that are needed to develop and validate them may find relevant and useful support in the work presented in chapter 4.

6 270 | Chapter 6

Part II

Dexamphetamine Cocaine dependence is classified as a psychiatric disorder that has a significant impact on the lives of patients and their surroundings. As there is currently no available pharmacotherapy, patients are faced with a chronic condition that majorly disrupts everyday life and is most often accompanied by other mental and physical afflictions [5].

Several drugs have been under investigation for the treatment of cocaine dependence. Amongst these, sustained release (SR) dexamphetamine has been shown to be the most effective in lowering the use of cocaine [6]. For the investigation of the effectiveness of SR dexamphetamine in the therapy of patients with cocaine dependence in the Neth- erlands, a new SR formulation was developed at the Netherlands Cancer Institute and the MC Slotervaart. The study that was carried out with this new formulation showed that SR dexamphetamine is a safe and effective pharmacotherapy for the treatment of comorbid treatment-refractory cocaine dependence in heroin-dependent patients in heroin-assisted treatment [7].

A second study was designed to examine the pharmacokinetics of the new SR dexamphetamine. The first part of this study, the development and validation of a plasma assay, is described in chapter 5.1. This bioanalytical assay was designed to measure dexamphetamine plasma concentrations over a broad range and may in the future be extended to measure in dried-blood spots and urine and may so be instrumental in measuring patient therapy adherence. The second part of the study involved the execution of a clinical trial in which the pharmacokinetics of two different dose levels (30 and 60 mg) of SR dexamphetamine were analyzed during 5 days of therapy, as detailed in chapter 5.2. The results from this study demonstrate the SR character of the formulation and support the once daily dosing scheme as used in the effectiveness study. Surprisingly, the participating patients experienced more side effects at the 60 mg dose level than in the previous. This warrants further investigation.

Now that the effectiveness, safety and pharmacokinetics have been explored in this small and specific patient population, the application of the SR formulation of dexamphetamine may also be investigated in other patient populations. Conclusion & Perspectives | 271

References

[1] G.L. Amidon, H. Lennernäs, V.P. Shah, J.R. [5] X. Castells, R. Cunill, C. Pérez-Mañá, X. Vidal, Crison, A theoretical basis for a biopharmaceu- D. Capellà, Psychostimulant drugs for cocaine tic drug classification: the correlation of in vitro dependence, in: X. Castells (Ed.), Cochrane drug product dissolution and in vivo bioavail- Database Syst. Rev., John Wiley & Sons, Ltd, ability., Pharm. Res. 12 (1995) 413–20. Chichester, UK, 2016. [2] L.X. Yu, Pharmaceutical Quality by Design: [6] D. V Herin, C.R. Rush, J. Grabowski, Agonist-like Product and Process Development, Under- pharmacotherapy for stimulant dependence: standing, and Control, Pharm. Res. 25 (2008) preclinical, human laboratory, and clinical 781–791. studies., Ann. N. Y. Acad. Sci. 1187 (2010) 76– [3] J.M. Keen, J.W. McGinity, R.O. Williams, En- 100. hancing bioavailability through thermal pro- [7] M. Nuijten, P. Blanken, B. van de Wetering, B. cessing., Int. J. Pharm. 450 (2013) 185–96. Nuijen, W. van den Brink, V.M. Hendriks, Sus- doi:10.1016/j.ijpharm.2013.04.042. tained-release dexamfetamine in the treat- [4] R.B. Verheijen, H. Yu, J.H.M. Schellens, J.H. ment of chronic cocaine-dependent patients Beijnen, N. Steeghs, A.D.R. Huitema, Practical on heroin-assisted treatment: a randomised, Recommendations for Therapeutic Drug Mon- double-blind, placebo-controlled trial, Lancet. itoring of Kinase Inhibitors in Oncology, Clin. 387 (2016) 2226–2234. Pharmacol. Ther. (2017).

Appendix

Summary Nederlandse samenvatting Dankwoord List of publications Curriculum Vitae Molecular structures

Appendix I

Summary

Summary | 277

Summary

For the past 20 years, the pharmacological therapy of cancer has been undergoing a metamorphosis, rapidly changing from treatment with broad and unspecific cytotoxic agents to personalized and highly specific targeted agents. This new class of drugs largely consists of small molecular compounds that mostly inhibit kinases. By way of their disruption of the kinase function, the targeted agents interfere with multiple cancer signaling pathways which are essential to tumor development, survival, growth, and metastasis.

Part I of this thesis discusses the issues that originate from the oral administration of targeted anticancer agents.

The oral administration of the targeted and highly specific small molecular kinase inhibitors (smKIs) results in a clinical setting where a poor bioavailability and a large variability in exposure and possibly effectiveness is common.Chapter 1 reviews these matters and addresses both their origins and clinical implications.

Chapter 1.1 provides an overview of the most important sources of the pharmacokinetic variability. It links the (bio)pharmaceutical characteristics of the kinase inhibitors to clinically relevant situations, such as co-medication and food interactions. By doing so, a clear sketch of the current state of affairs is presented. Additionally, the chapter draws attention to the possibilities of reducing the high variability in the patient exposure of the kinase inhibitors.

The root cause of the issues of poor drug uptake and variable pharmacokinetics of the drug class is the focus of chapter 1.2. Here, it is demonstrated that the same molecular structures that grant the kinase inhibitors their high specificity are also responsible for drugs’ low aqueous solubility. Furthermore, the measures that were taken in the formulation development of the marketed dosage forms are summarized along with the A general poor biopharmaceutical performance associated with these efforts. The chapter 278 | Appendix

concludes with a synopsis of the various formulation improvement techniques that have been applied in literature and patents to enhance the biopharmaceutics of the kinase inhibitors and illustrates this with clear examples.

The observations and the discussions from the previous chapters are revisited inchapter 1.3, where a firm stance is adopted in support of more thorough biopharmaceutical -re search and development prior to the market release of new chemotherapeutics.

Two of the most poorly pharmaceutically performing smKIs were selected as suitable candidates for formulation improvement, namely pazopanib and nilotinib. The topic of Chapter 2 is the first step in the proper development of a new formulation, namely the pharmaceutical characterization of these anticancer agents. In both chapters, thermal analytical methods were used to reveal previously unknown physicochemical features. Chapter 2.1 presents a thermal profile of pazopanib hydrochloride as crystal polymorph form I, which is the form that is used in the marketed formulation Votrient®. The study involved the characterization of the drug substance by X-ray diffraction analysis and Fou- rier transform infrared spectroscopy (FTIR). Through thermogravimetric analysis and differential scanning calorimetry, this chapter also demonstrates the thermal stability of the crystalline form and the stages of the thermal degradation process.

Both the thermal analyses of the crystalline (polymorphs form B) and amorphous forms of nilotinib hydrochloride are discussed in chapter 2.2. Both forms were subjected to thorough analysis by nuclear magnetic resonance spectroscopy, FTIR and X-ray diffraction analysis. Thermogravimetric analysis was applied to the crystalline form to yield data on the thermal stability and the phases of the degradation process. The study also included the thermodynamic analysis of the calorimetric data that was obtained for both forms of the drug. The chapter concludes with an extensive stability profile of both crystalline and amorphous nilotinib hydrochloride.

Chapter 3 presents the results and the discussion of the formulation development studies for pazopanib hydrochloride and nilotinib hydrochloride. The studies reflect back on the pharmaceutical characterization and implemented a physiology-based dissolution system to ascertain a rational formulation design.

The solubility of pazopanib hydrochloride was improved in Chapter 3.1. After an excipient-compatibility study and an initial excipient-solubility screening, the testing of binary Summary | 279

physical mixtures yielded a final formulation that exhibited a solubility increase with a factor of more than 300 in an intestinal environment. Furthermore, the mechanism of the solubility increase of the best-performing formulation was meticulously studied by two types of electron microscopy, X-ray diffraction and dynamic light scattering. The selected final formulation was subsequently tested in vivo in patients from which pharmacokinetic samples were collected. Single dose pharmacokinetics of three different doses were determined and compared to Votrient®. The comparison showed that 300 mg of the newly designed formulation produced a similar patient exposure as Votrient® 800 mg with a reduced variability.

Chapter 3.2 describes the study of the formulation improvement of nilotinib hydrochloride. A variety of excipients was tested for the solubility improving effect on the drug, both after physical mixing and spray drying. The powder formulations that were manufactured with the spray drying technique proved to be superior in dissolution characteristics compared to physical mixtures. The optimal drug-excipient combination was selected and analyzed with different spectroscopic, microscopic and thermal techniques.

The variability that stems from the marketed pharmaceutical smKI formulations may result in plasma concentrations that are ineffectively low or toxically high. To timely detect under- or overdosing, the clinic often relies on therapeutic drug monitoring (TDM). Chapter 4 presents two bioanalytical assays that were specifically designed to simultaneously quantify multiple targeted chemotherapeutics. The suitability of the methods for the purpose of TDM was improved by certain adjustments to the general recommendations for bioanalytical method validation.

Chapter 4.1 outlines the fully validated liquid chromatography–tandem mass spectrometry (LC-MS/MS) method for the quantification of dasatinib, erlotinib, gefitinib, imatinib, lapatinib, nilotinib, pazopanib, sorafenib, sunitinib, vemurafenib and the active metabolite, N-desethyl-sunitinib in patient plasma. The applicability of the method was additionally proven by the analysis of plasma samples of patients with cancer that were under treatment with one of the analyzed drugs.

The development and validation of the bioanalytical LC-MS/MS assay for the quantification of afatinib, axitinib, ceritinib, crizotinib, dabrafenib, enzalutamide, regorafenib and trametinib in plasma is presented in Chapter 4.2. The development directed particular A attention to the light instability of the drugs axitinib and dabrafenib with regard to sam- 280 | Appendix

ple handling in the clinical practice. As with the previous chapter, the clinical applicability of this method was also confirmed through the analysis of treated patient plasma samples.

In summary, this thesis shows the current situation of highly specific anticancer agents of which the clinical performance is significantly hindered by a low and variable bioavailability. In addition, it clearly shows that by way of a well-designed formulation, this clinical performance can be greatly improved.

Part II of this thesis addresses the manufacture and the pharmacokinetic analysis of dexamphetamine sustained-release (SR) tablets. These tablets were specifically designed for use in the therapy of cocaine dependence. Cocaine dependence is a harmful mental disease for which there currently is no pharmacotherapy available.

Chapter 5 describes the preparation and execution of a descriptive pharmacokinetic- ex ploration study of dexamphetamine SR tablets. This study was facilitated by the development and validation of a LC-MS/MS assay for the quantification of dexamphetamine in patient plasma samples. The method is presented inchapter 5.1. Herein, the situation of the quantification of a small molecule analyte in a high-throughput environment and the associated issues are also addressed. Likewise, the suitability of the assay for use in the pharmacokinetic study was established.

Chapter 5.2 presents the results of the pharmacokinetic study itself. Two different doses of the SR dexamphetamine were administered to chronic crack-cocaine-dependent patients with comorbid heroin dependence, currently on heroin-assisted treatment. The plasma samples were analyzed by the LC-MS/MS assay and the pharmacokinetics were evaluated through non-compartmental analysis. The study clearly shows the SR pharmacokinetics of the formulation at both dose levels, 30 and 60 mg.

Part II provides further evidence in favor of the implementation of SR dexamphetamine in the therapy of cocaine dependence.

Appendix II

Nederlandse samenvatting

Nederlandse samenvatting| 285

Nederlandse samenvatting

Gedurende de afgelopen 20 jaar is de farmacologische therapie van kanker een meta- morfose ondergaan waarin het zich veranderde van de behandeling met brede en aspe- cifieke cytotoxische middelen richting sterk specifieke doelgerichte stoffen. Deze nieuwe groep van geneesmiddelen bestaat voor het grootste gedeelte uit laagmoleculaire ver- bindingen die met name eiwitkinases remmen. Door het verstoren van de kinasefunctie, zijn deze gerichte middelen in staat om de vele kankersignaaltransductiewegen te on- derbreken. Deze zijn essentieel voor tumorontwikkeling, overleving, groei en uitzaaiing. Tegelijk met het opkomen van de gerichte therapie verscheen de zogenaamde ‘intrave- neus-naar-oraal-wisseling’. Illustratief hiervoor is het grote en toenemende aantal gerichte middelen dat oraal moeten worden toegediend. Daarnaast is het merendeel van de huidig in onderzoek zijnde gerichte middelen bestemd voor orale toediening.

Deel I van dit proefschrift behandelt de problemen die voortkomen vanuit de orale toediening van de gerichte antikankermiddelen.

De orale toediening van de gerichte en sterk specifieke laagmoleculaire kinaseremmers (small molecular Kinase Inhibitors, smKIs) resulteert in een klinische situatie waar een lage biologische beschikbaarheid en een hoge variabiliteit in blootstelling en mogelijk effectiviteit veelvoorkomend zijn. Hoofdstuk 1 beschouwt deze zaken en bespreekt de oorzaken en de klinische implicaties.

Hoofdstuk 1.1 geeft een overzicht van de belangrijkste bronnen van de farmacokinetische variabiliteit. Het verbindt de (bio)farmaceutische karakteristieken van dekina- seremmers aan klinisch relevante situaties, zoals comedicatie en voedselinteracties. Hierdoor wordt een duidelijk beeld gegeven van de huidige stand van zaken. Daarnaast schenkt het hoofdstuk aandacht aan de beschikbare mogelijkheden om de hoge variabiliteit in patiëntblootstelling van de smKIs te verlagen.

De kernoorzaak van de problemen met de opname en de variabele farmacokinetiek van deze groep geneesmiddelen wordt verder besproken in hoofdstuk 1.2. Hier wordt aan- getoond dat het de gelijkende moleculaire structuren van de gerichte middelen zijn die maken dat de oplosbaarheid in water bijzonder laag is. Deze structuren zijn gelijkend vanwege de sterk overeenkomende werkingsmechanismen van de stoffen. Bovendien A 286 | Appendix

wordt er een samenvatting gegeven van de wijzen waarop hiermee is omgegaan in de huidige handelsformuleringen en de hieruit volgende slechte biofarmaceutische prestaties.

Het hoofdstuk sluit af met een overzicht van verschillende formuleringstechnieken die zijn gebruikt in gepubliceerde of gepatenteerde onderzoeken om de biofarmaceutische aspecten van de smKIs te verbeteren en licht dit toe met duidelijke voorbeelden.

De observaties en discussies uit de voorgaande hoofdstukken worden opnieuw aan- gehaald in hoofdstuk 1.3. Daaropvolgend wordt gepleit voor meer diepgaand biofar- maceutisch onderzoek en ontwikkeling voorafgaand aan het op de markt brengen van nieuwe antikankermiddelen.

Twee van de slechtspresenterende smKIs werden uitgekozen als geschikte kandidaten voor het ontwikkeling van een verbeterde formulering; pazopanib en nilotinib. Het on- derwerp van hoofdstuk 2 is de eerste stap in de fatsoenlijke ontwikkeling van een nieuwe formulering, namelijk het farmaceutisch karakteriseren van de actieve stof. In beide hoofdstukken (2.1 en 2.2) werden thermoanalytische methoden gebruikt om voorheen onbekende fysisch-chemische eigenschappen van de twee stoffen te verkennen.

Hoofdstuk 2.1 beschrijft het thermische profiel van pazopanibhydrochloride in de kristallijne polymorf I. Dit is tevens de vorm is zoals die wordt gebruikt in de handelsformu- lering Votrient®. In deze studie wordt de stof gekarakteriseerd door middel van Röntgen- diffractieanalyse (XRD) en Fourier-Transform-Infraroodspectrometrie (FTIR). Met behulp van thermogravimetrische analyse en dynamische differentiecalorimetrie, geeft het hoofdstuk een duidelijk beeld van de thermische stabiliteit van pazopanibhydrochloride en de fasen van thermische degradatie.

Zowel de thermische analyses van de kristallijne (polymorf B) als van de amorfe vorm van nilotinibhydrochloride worden besproken in hoofdstuk 2.2. Beide vormen werden onderworpen aan een grondige analyse door middel van kernspinresonantiespectrosco- pie (NMR), FTIR en XRD. Thermogravimetrische analyse werd toegepast op de kristallijne vorm om een beeld te krijgen van de thermische stabiliteit en de fasen van het thermische degradatieproces. Er werd tevens een thermodynamische analyse van calorimetri- sche data van beide vormen van de stof uitgevoerd in de studie. Het hoofdstuk eindigt met uitgebreide stabiliteitsprofielen van kristallijn en amorf nilotinibhydrochloride. Nederlandse samenvatting| 287

Hoofdstuk 3 beschrijft de resultaten en de discussie van de formuleringsontwikkelings- studies die werden uitgevoerd voor de middelen pazopanibhydrochloride en nilotinibhydrochloride. De studies nemen de farmaceutische karakterisering in acht en maken gebruik van een op de fysiologie gebaseerd dissolutiesysteem om te komen tot een rati- onele ontwikkeling van de nieuwe formuleringen.

De oplosbaarheid van pazopanibhydrochloride werd verbeterd in het onderzoek dat staat beschreven in hoofdstuk 3.1. Volgend op een hulpstofcompatibiliteitsstudie en een initiële hulpstofoplosbaarheidsscreening leidde het testen van binaire -poeder mengsels tot een finale formulering die de oplosbaarheid in intestinale condities met een factor van meer dan 300 kon verhogen. Daarnaast werd het mechanisme achter de oplosbaarheidsverhoging gedetailleerd onderzocht met behulp van twee vormen van electronmicroscopie, XRD en dynamische lichtverstrooiing. De best presterende formulering werd uiteindelijk in vivo in patiënten getest, waarvan farmacokinetische monsters werden verzameld. De farmacokinetiek van een eenmalige dosis van drie verschillende doses werd vergeleken met die van Votrient®. De vergelijking liet zien dat 300 mg van de nieuwe formulering een gelijkende blootstelling tot gevolg had als 800 mg Votrient®. Daarnaast heeft de nieuwe formulering minder farmacokinetische variabiliteit.

Hoofdstuk 3.2 behandelt de resultaten en de discussie van de formuleringsverbeterings- studie van nilotinibhydrochloride. Een verzameling verschillende hulpstoffen werd -ge test op oplosbaarheidsverbeterend vermogen, zowel na fysiek mengen als na sproeidrogen. De poederformuleringen die werden gemaakt met behulp van sproeidrogen bleken superieur te zijn qua dissolutiegedrag. De optimale geneesmiddel-hulpstofcombinatie werd geselecteerd en geanalyseerd met verschillende spectroscopische, microscopische en thermische technieken.

De variabiliteit die voorkomt vanuit de op de markt zijnde farmaceutische smKI-formuleringen kan tot gevolg hebben dat resulterende plasmaconcentraties ineffectief laag of toxisch hoog kunnen zijn. Om dergelijke onder- en overdoseringen tijdig te detecteren vertrouwt de kliniek vaak op therapeutische geneesmiddelcontrole (Therapeutic Drug Monitoring, TDM). Hoofdstuk 4 beschrijft twee bioanalytische methoden die nadruk- kelijk ontworpen zijn om gelijktijdig meerdere gerichte antikankermiddelen te kunnen kwantificeren. De toepasbaarheid op klinische monsters met het oog op TDM werd bij deze methoden verbeterd door bepaalde aanpassingen op de algemene richtlijnen voor A de validatie van bioanalytische methoden. 288 | Appendix

Hoofdstuk 4.1 rapporteert over de ontwikkeling en de volledige validatie van een mas- saspectrometergekoppelde vloeistofchromatografische methode (LC-MS/MS) die is ontworpen voor het kwantificeren van dasatinib, erlotinib, gefitinib, imatinib, lapatinib, nilotinib, pazopanib, sorafenib, sunitinib, vemurafenib en de actieve metaboliet, N-desethyl-sunitinib in patiëntplasma. De klinische toepasbaarheid voor TDM werd onderzocht en ondersteunt door de analyse van plasmamonsters van patiënten met kanker die werden behandeld met een van de middelen uit de methode.

De ontwikkeling en de validatie van de bioanalytische LC-MS/MS-methode voor het kwantificeren van afatinib, axitinib, ceritinib, crizotinib, dabrafenib, enzalutamide, regorafenib and trametinib wordt uiteengezet inhoofdstuk 4.2. Tijdens de ontwikkeling ging speciale aandacht uit naar de invloed van licht op de stabiliteit van axitinib en dabrafenib gelet op de praktische handelingen met monsters in de kliniek. Zoals in het voorgaande hoofdstuk, werd ook hier de klinische toepasbaarheid bekrachtigd door de analyse van patiëntenmonsters.

Samenvattend laat dit proefschrift de huidige situatie zien van de sterk specifieke antikankermiddelen waarvan de klinische prestaties significant gehinderd worden door een lage en variabele biologische beschikbaarheid. Bovendien laat het beschreven onderzoek duidelijk zien dat een goed ontworpen farmaceutische formulering deze klinische prestatie sterk kan verbeteren.

Deel II van dit proefschrift is gewijd aan de productie en de farmacokinetische analyse van dexamfetaminetabletten met vertraagde afgifte (VA). Deze tabletten werden speci- fiek ontworpen voor gebruik in de therapie van cocaïneverslaving. De afhankelijkheid van cocaïne is een schadelijke geestelijke aandoening waarvoor er tot op heden geen farmacotherapie beschikbaar is.

Hoofdstuk 5 gaat over de voorbereiding en de uitvoering van een beschrijvende explo- ratieve farmacokinetische studie van dexamfetaminetabletten VA. De studie werd ondersteunt door de ontwikkeling en validatie van een LC-MS/MS methode die bestemd was voor het kwantificeren van dexamfetamine in patiëntplasma. Deze methode wordt beschreven in hoofdstuk 5.1. Hier wordt ook de situatie van de analyse van een laagmo- leculair analiet in een setting besproken waar er veel afwisseling van analysemethoden op analyseapparatuur plaatsvindt. Ook hier wordt de klinische toepasbaarheid, ditmaal voor een farmacokinetische studie, vastgesteld. Nederlandse samenvatting| 289

Hoofdstuk 5.2. geeft een overzicht van de resultaten van de farmacokinetische studie. Twee verschillende doses werden toegediend aan chronisch crack-cocaïneverslaafde patiënten met een comorbide heroïneverslaving, die ten tijde van de studie een behandeling ondergingen met medicinale heroïne. De plasmamonsters werden geanalyseerd met de LC-MS/MS methode en de farmacokinetiek werd geëvalueerd door middel van non-compartimentele analyse. De studie liet duidelijk de VA-farmacokinetiek vande twee doseringen, 30 en 60 mg, zien.

Deel II voorziet in verder bewijs dat de implementatie van VA dexamfetamine in de therapie van cocaïneverslaving ondersteunt.

Appendix III

Dankwoord

Dankwoord | 293

Dankwoord

Het fundament waarop het beschreven onderzoek in dit proefschrift staat is mogelijk gemaakt door de waardevolle en onmisbare bijdragen van vele mensen.

Mijn meest diepe erkenning van dank gaat uit naar die mensen voor wie het onderzoek is opgezet; de patiënten. Zij die veelal in de laatste fase van het leven bereid waren om kostbare uren en dagen ter beschikking te stellen ben ik uitzonderlijke dank verschul- digd.

Zonder herhaaldelijke wijze raad, nuchter advies en inzichtgevende consultatie had het onderzoek in dit proefschrift niet kunnen plaatsvinden. Gelukkig kon ik alle drie altijd vinden bij mijn copromotor. Beste Bastiaan, een betere farmaceutische coach had ik me niet kunnen wensen!

Gedegen multidisciplinair onderzoek komt alleen tot bloei wanneer alle omstandighe- den precies juist zijn. Mijn promotoren, Jos en Jan, hebben mij de kans geboden om onderzoek te doen op het snijvlak van de farmacie en de oncologie. De vele mogelijkheden die ik heb gekregen en talrijke mooie projecten waar ik deel van was, maken dat ik hen erg dankbaar ben.

Het ontwerpen, opzetten en uitvoeren van analysemethoden, productieprocessen en klinische studies kan een lange, en op punten eindeloos lijkende, exercitie zijn. Met en dankzij de hulp en ondersteuning van de verschillende afdelingen van het AvL-NKI zijn de projecten in dit proefschrift toch tot een goed einde gekomen.

De steun en het vertrouwen van de apotheekmedewerkers van de productieafdeling waren mij zeer kostbaar. De met ongebreidelde enthousiasme gebrachte wijsheden en praktische tips hebben een groot deel van het pad naar dit proefschrift verlicht.

Speciale dank gaat uit naar alle mensen van het Bioanalytisch Laboratorium waar ik vele onvergetelijke uren, dagen en maanden heb mogen doorbrengen onder het nimmer af- latende doch altijd inspirerende GLP-regime.

Ook de CRU afdeling wil ik veel dank schenken voor mijn korte patiëntenrijke periode A waarin ik gastvrij werd ontvangen en daadkrachtig werd gegidst. 294 | Appendix

Daar een deel van het onderzoek zich buitengewoon extramuraal afspeelde wil ik zeker een woord van dank richten aan de medewerkers van het Academisch Psychiatrisch Cen- trum in het Academisch Medisch Centrum. Hier voor mij onlosmakelijk aan verbonden gaat ook veel dankbaarheid uit het team van de Gemeentelijke Gezondheidsdienst aan de Flierbosdreef.

Waarom het U-gebouw (SLZ) liefkozend de Keet werd genoemd liet zich bij de allereer- ste aankomst niet raden. De harde, koude en spaarzaam bedekte werkelijkheid bleek slechts een uiterlijke vertoning. De vele collega’s en stagiairs die ik heb zien gaan en komen hebben het verblijf aldaar en in zowel het O- als H-gebouw tot een warm bad gemaakt waar geen U-gebouw (AvL) tegen op zou kunnen.

Bepaald geen sinecure te noemen, maar er is een leven naast promoveren. Naast vele andere dingen waren er reizen naar tropische en wat minder tropische oorden, frequen- te nachtopvang met vele koffers en talloze sociale situaties die ook de afgelopen vier jaar voor de meest plezierige onderbrekingen hebben gezorgd. Lieve paranimfen, Abigaël en Rianne, wat ontzettend mooi dat ik daar deel van mocht uitmaken.

Na lange en zware dagen van onderzoek plegen is naar huis gaan plezierig, maar thuis komen het fijnste wat er is. Lieve Milan, ik mag van het meest onwaarschijnlijke geluk spreken dat ik met jou het merendeel van dit proefschrift heb mogen zien ontstaan. Dank je wel voor alle emotionele en inhoudelijke steun, jouw eindeloze geduld en de ruimte die je me gaf in de krapheid aan de Conradstraat.

Mijn dankwoord eindigt met hen waarmee het allemaal begon. Vanaf mijn eerste eigen- wijze melkweguitspraken tot aan het publiceren van mijn meest recente artikel heb ik altijd de steun en ondersteuning gehad van mijn familie. Pappa, Mamma, Renee en Kelly, het vertrouwen en de houvast door de jaren heen hebben mij de solide basis gegeven waarvoor woorden geen recht doen aan mijn dankbaarheid.

Appendix IV

List of publications

List of publications | 299

List of publications

• Variability in bioavailability of small molecular tyrosine kinase inhibitors Herbrink M, Nuijen B, Schellens JHM, Beijnen JH Cancer Treat Rev. 2015 May;41(5):412-422 • Inherent formulation issues of kinase inhibitors Herbrink M, Schellens JHM, Beijnen JH, Nuijen B J Control Release. 2016 Oct;239:118-127 • Quantification of 11 therapeutic kinase inhibitors in human plasma for therapeutic drug monitoring using liquid chromatography coupled with tandem mass spectrometry Herbrink M, de Vries N, Rosing H, Huitema ADR, Nuijen B, Schellens JHM, Beijnen JH Ther Drug Monit. 2016 Dec;38(6):649-656 • Improving the solubility of nilotinib through novel spray-dried solid dispersions Herbrink M, Schellens JHM, Beijnen JH, Nuijen B Int J Pharm. 2017 Aug;529(1-2):294-302 • High-tech drugs in creaky formulations Herbrink M, Nuijen B, Schellens JHM, Beijnen JH Pharm Res. 2017 Sep;34(9):1751-1753 • Thermal study of pazopanib hydrochloride Herbrink M, Nuijen B, Schellens JHM, Beijnen JH J Therm Anal Calorim. 2017 Dec;130(3):1491-1499 • Thermal stability study of crystalline and novel spray-dried amorphous nilotinib hydrochloride Herbrink M, Vromans H, Schellens JHM, Beijnen JH, Nuijen B J Pharm Biomed Anal. 2018 Jan;148:182-188 • Development and validation of a high-performance liquid chromatography-tandem mass spectrometry assay for the quantification of dexamphetamine in human plasma Herbrink M, Thijssen B, Hillebrand MJX, Rosing H, Schellens JHM, Nuijen B, Beijnen JH J Pharm Biomed Anal. 2018 Jan;148:259-264 • Development and validation of a liquid chromatography-tandem mass spectrometry analytical method for the therapeutic drug monitoring of eight novel anticancer drugs Herbrink M, de Vries N, Rosing H, Huitema ADR, Nuijen B, Schellens JHM, Beijnen JH Biomed Chromatogr. 2018 Apr;32(4) A 300 | Appendix

• Pharmacokinetics of sustained-release oral dexamphetamine sulphate in cocaine and heroin dependent patients Herbrink M, Nuijten M, Nuijen B, Huitema ADR, Beijnen JH, Hendriks VM, Blanken P, Janmohamed A, van den Brink W Accepted for publication in J Clin Psychopharmacol. • Solubility and bioavailability improvement of pazopanib hydrochloride Herbrink M, Groenland SL, Huitema ADR, Schellens JHM, Beijnen JH, Steeghs N, Nuijen B Accepted for publication in Int J Pharm

Patents

• Shielding device, WO2017072279A1 Schreuder N, Swiers G, Herbrink M

Appendix V

Curriculum vitae

Curriculum vitae | 305

Curriculum vitae

Maikel Herbrink was born on February 2nd 1990 in Groningen and grew up in Assen. After graduating from pre-university education (VWO) at the Werkmancollege in Groningen, he chose to study Pharmacy at the University of Groningen. There he received both his Bachelor and Master of Science degrees, cum laude. He conducted his scientific intern- ship at the department of Pharmaceutical Technology and Biopharmacy in collaboration with Harro Höfliger GmbH in Allmersbach im Tall, Germany under the supervision of prof. dr. H.W. Frijlink and dr. A.H. de Boer. The research focused on the influence of powder filling techniques on the characteristics of powders for inhalation. Subsequently, he started his PhD research at the department of Pharmacy and Pharma- cology of the Netherlands Cancer Institute –Antoni van Leeuwenhoek and MC Sloter- vaart supervised by dr. B. Nuijen, prof. dr. J.H. Beijnen and prof. dr. J.H.M. Schellens. His PhD research resulted in this thesis, which he will defend at the University of Utrecht on June 27th 2018.

Appendix VI

Molecular structures

Molecular structures | 309

Molecular structures

H N 2 O N S N O

NH HCl N N N

Pazopanib, formulated as Votrient® with the hydrochloride salt

N F F N N F N HCl .H2O NH N NH O

Nilotinib, formulated as Tasigna® with the monohydrated hydrochloride salt

A 310 | Appendix

O NH

N N S

H O N NH N 2 N

Dasatinib, formulated as Sprycel® with the monohydrated free base

O N O N O HCl HN

Erlotinib, formulated as Tarceva® with the hydrochloride salt

Cl F

O NH N O N

O N

Gefitinib, formulated as Iressa® with the free base Molecular structures | 311

N N N O NH O N S OH NH O N

Imatinib, initially formulated as Glivec® with the mesylate salt

O S NH O O OH S O O

O OH S O F N O

Cl NH N Lapatinib, formulated as Tyverb® with the ditosylate salt

A 312 | Appendix

NH NH OH O O HO OH F O O N H Sunitinib, formulated as Sutent® with the malate salt

NH NH O

F O N H N-desethyl Sunitinib, an active metabolite of sunitinib Molecular structures | 313

NH Cl O F O N F OH S O F NH O O NH

Sorafenib, formulated as Nexavar® with the tosylate salt

NH S F O Cl O

N N H Vemurafenib, formulated as Zelboraf® with the free base

N Cl F O O HO OH O HN HN O O N HO OH O O N Afatinib, formulated as Giotrif® with the dimaleate salt A 314 | Appendix

N N S H O NH Axitinib, formulated as Inlyta® with the free base

O O S O NH N NH

N Cl NH

Ceritinib, formulated as Zykadia® with the free base

Cl H N O N

N Cl H2N F Crizotinib, formulated as Xalkori® with the free base Molecular structures | 315

F N F O O NH S S S OH O O N F N H2N Dabrafenib, formulated as Tafinlar® with the mesylate salt

F O F F N N N

N S H O F Enzalutamide, formulated as Xtandi with the free base

NH Cl O F N F O F H O NH 2 O NH

F Regorafenib, formulated as Stivarga® with the monohydrated free base A 316 | Appendix

O O N O F O NH N NH S N I O Trametinib, formulated as Mekinist® with the dimethylsulfoxide-solvated free base

NH2

OH O S OH NH2 O

Dexamphetamine, often formulated as the sulphate salt