Integrating Efficacy and Toxicity in Preclinical …407166/...In recent decades, an understanding of signalling pathways has emerged, uncovering entirely new drug targets, such as

Sakta men steg för steg

går det framåt

det måste man tro

Märta Tikkanen

List of papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I Haglund C, Åleskog A, Håkansson LD, Jacobson S, Larsson

R, Lindhagen E. The FMCA-GM assays, high throughput non-clonogenic alternatives to CFU-GM in preclinical hematotoxic-ity testing. Toxicology Letters (2010) 194, 102-107

II Haglund C, Åleskog A, Nygren P, Gullbo J, Höglund M, Wickström M, Larsson R, Lindhagen E. In vitro evaluation of clinical activity and toxicity of anticancer drugs using tumor cells from patients and cells representing normal tissues. Sub-mitted manuscript

III Hassan SB, Haglund C, Åleskog A, Larsson R, Lindhagen E. Primary lymphocytes as predictors for species differences in cy-totoxic drug sensitivity. Toxicology In vitro (2007) 21, 1174-81

IV Haglund C, Fryknäs M, Linder S, Larsson R, Rickardson L. Identification of CB3, a novel inhibitor of the ubiquitin-proteasome system. Manuscript

Reprints were made with permission from the respective publishers.

Additional papers

Rickardson L, Fryknäs M, Haglund C, Lövborg H, Nygren P, Gustafsson MG, Isaksson A, Larsson R. Screening of an annotated compound library for drug activity in a resistant myeloma cell line. Cancer Chemotherapy and Pharmacology (2006) 58, 749-758

Wickström M, Haglund C, Lindman H, Nygren N, Larsson R,

Gullbo J. The novel alkylating prodrug J1: diagnosis directed ac-tivity profile ex vivo and combination analyses in vitro. Investiga-tional New Drugs (2008) 26, 195-204

Contents

Introduction...................................................................................................11 Cancer ......................................................................................................11 Cancer drug therapy .................................................................................11

Classic cytotoxic agents.......................................................................12 Targeted drugs .....................................................................................12

Toxicity of cancer drugs...........................................................................15 Preclinical assays for normal cell toxicity ...........................................15 Preclinical assays for prediction of haematological toxicity ...............15 Species differences in drug sensitivity.................................................17

Antitumour activity of cancer drugs.........................................................18 Screening for cancer drugs ..................................................................18 Preclinical models for efficacy assessment .........................................19

Aims..............................................................................................................21

Materials and Method ...................................................................................22 Drugs and reagents ...................................................................................22 Cell models...............................................................................................22

CD34+ stem cells for reflection of bone marrow toxicity....................22 Lymphocytes for reflection of haematological toxicity.......................22 Cells for reflection of liver, epithelial and renal toxicity.....................24 Lymphocytes for studies of species differences ..................................24 Human patient tumour cells.................................................................24 Human tumour cell lines......................................................................25

Measurements of cytotoxicity ..................................................................25 Screening for UPS inhibitors....................................................................26 Activity assay for the 20S proteasome .....................................................26 Connectivity Map.....................................................................................26 Statistics ...................................................................................................27

Results and discussion ..................................................................................28 Normal cell models for toxicity assessment.............................................28

Development of FMCA-GM, a high throughput alternative to CFU-GM (paper I)...............................................................................28 Establishment of a normal cell panel (paper II)...................................30 Development of an in vitro species difference test using lymphocytes (paper III).............................................................................................32

Summary of the methods developed in papers I, II and III .................34 Patient tumour cells for efficacy assessment............................................36

Patient tumour cells used for prediction of diagnosis-specific activity (paper II) ..............................................................................................36

Application of efficacy and toxicity models ............................................39 Identification of a novel UPS inhibitor, CB3 (paper IV).....................39

Conclusions and future perspectives.............................................................42

Svensk populärvetenskaplig sammanfattning...............................................43

Acknowledgements.......................................................................................45

References.....................................................................................................48

Abbreviations

ALL Acute lymphocytic leukemia AML Acute myelocytic leukemia CFU Colony forming units CLL Chronic lymphocytic leukemia Cmap Connectivity Map CML Chronic myelocytic leukaemia DMSO Dimethyl sulphoxide DUB De-ubiquitylating enzymes EFGR Epithelial growth factor receptor FDA Food and Drug Administration FMCA Fluorometric microculture cytotoxicity assay FMCA-E FMCA-erythroid FMCA-GM FMCA-granulocyte macrophage FMCA-Mk FMCA-megacaryocyte HSC Haematopoietic stem cell IC50 Inhibitory concentration 50% IC90 Inhibitory concentration 90% LD10 Lethal dose 10% MTD Maximum tolerated dose mTOR Mammalian target of rapamycin NCI National Cancer Institute NSCLC Non-small cell lung cancer PBMC Peripheral blood mononuclear cell PBS Phosphate–buffered saline SI Survival index UPS Ubiquitin-proteasome system VEGF Vascular endothelial growth factor

11

Introduction

Cancer Every year, more than 50 000 people are diagnosed with cancer in Sweden, and the current estimate is that more than one in three people will develop cancer at some point in their lifetime. Cancer incidence, defined as the num-ber of new cases arising in a period of time, is gender and age specific. Can-cer occurs mostly in older people and two thirds of all cancers affect those over 65 years of age. The incidence of cancer is the same in men and women, but they are affected by different types of cancer. In females, breast cancer is most common and prostate cancer is the most frequent cancer type among men. The mortality of cancer varies according to cancer type, but progress in early diagnosis and treatment of cancer has improved cancer survival rates. Despite this, approximately only half of all cancer patients are cured, and hence there is still a great need for improvement of cancer treat-ment (Airley 2009; Cancerfonden 2011).

The development of cancer is a multi-step process of genetic alterations (Hanahan and Weinberg 2000; Weinberg 2007). These mutations can cause transformation of normal human cells into highly malignant derivates. It takes time to acquire enough mutations to influence the function of the cell, which explains why cancer has an age-dependent incidence. Most cancer cells have mutations that produce oncogenes with gain of function and tu-mour suppressor genes with loss of function. There are a number of essential alterations in cell physiology that together give rise to malignant growth; self-sufficiency in growth signals, insensitivity to growth-inhibitory signals, resistance to cell death, limitless replicative potential, induction of angio-genesis, reprogrammed energy metabolism, evasion of immune destruction, activation of invasion and metastasis. In addition, cancer cells recruit normal cells that form tumour-associated stroma and contribute to the development and progression of the tumour (Hanahan and Weinberg 2000, 2011).

Cancer drug therapy Cancer drug therapy is the main method of treatment for only a few cancers but is often used along with surgery and/or radiotherapy, to achieve and maintain remission. The choice of therapy depends on the tumour type and

12

stage of development. One of the problems with anticancer agents is their toxicity and narrow therapeutic window. Another is the very poor response of some cancers e.g. renal cancer (Weinberg 2007).

Classic cytotoxic agents Many of the anticancer treatments used today were developed prior to 1975, at a time when much of the knowledge of genetic and biochemical mecha-nisms of cancer pathogenesis was yet to be discovered (Chabner and Roberts 2005; Weinberg 2007). The main mechanism of action of these classic cyto-toxic drugs is inhibition of the increased rate of DNA synthesis and replica-tion, or to destroy DNA in tumour cells. Cytotoxic drugs can interact with cells via different mechanisms and are divided into groups accordingly; al-kylating agents (e.g. melphalan), antimetabolites (e.g. cytarabine, fluorouracil), topoisomerase inhibitors (e.g. etoposide, doxorubicin) and microtubule interacting agents (e.g. vincristine, paclitaxel) (Nygren 2001).

Targeted drugs In recent decades, an understanding of signalling pathways has emerged, uncovering entirely new drug targets, such as growth factors, signalling molecules, cell cycle proteins, modulators of apoptosis and molecules that regulate angiogenesis, introducing new classes of anticancer agents (Chabner and Roberts 2005). In contrast to the traditional DNA-targeting cytotoxic agents, these drugs were designed to specifically act on their targets and thereby to be less toxic to normal cells. Traditionally, cancer patients have been dosed with cytotoxic agents in phase I trials up to the maximum toler-ated dose (MTD) since this was assumed to be the most effective dose. How-ever this strategy is shifting since, as targeted drugs are not necessarily most effective at MTD (Millar and Lynch 2003; Yap et al. 2010).

Imatanib (Glivec®) is a successful example of a targeted drug that has im-proved survival rates in chronic myeloid leukaemia (CML) and is generally well tolerated. Imatinib is an inhibitor of the kinase BCR-ABL, the fusion protein product of a chromosomal translocation involved in the pathogenesis of CML. Other examples of tyrosine kinase inhibitors are erlotinib (Tarceva) and gefitinib (Iressa), targeting epithelial growth factor receptor (EGFR). Another target-specific strategy is to inhibit signalling using the larger anti-bodies such as rituximab (Mabthera®) directed against CD20 in lymphoma, trastuzumab (Herceptin®) used against HER2 in breast cancer and bevaci-zumab (Avastin®) that targets vascular endothelial growth factor (VEGF) receptor and inhibits angiogenesis. Other examples of new drugs are, borte-zomib (Velcade®) targeting the proteasome and rapamycin targeting the mammalian target of rapamycin (mTOR) (Airley 2009; Chabner and Roberts

13

2005). It should be noted, however, that these drugs are not always as spe-cific as proposed and are unfortunately associated with adverse effects.

Most of the targeted drugs will be used in combination with other anti-cancer drugs. Unlike BCR-ABL in CML where one main mutation is re-sponsible for malignant progression, most other tumours have several muta-tions that can be targeted. The rational of using cancer drugs in combination therapy is to maximize tumour cell kill using drugs with different modes of action, thus acting on different targets. A benefit of using combinations of drugs may also be a reduction in the risk of adverse events, where lower doses of each agent may be used when selecting agents with dissimilar toxic-ity profiles (Airley 2009; Sharma et al. 2010).

The proteasome as a drug target Malignant cells synthesize large amounts of proteins and thus require a way to degrade misfolded and old proteins. The ubiquitin–proteasome system (UPS) is the main pathway of protein degradation and thereby plays a criti-cal role in the signal transduction pathways important for tumour cell growth and survival (Adams 2004). Also, tumour cells have been shown to be more susceptible to proteasome inhibition compared to normal cells (Adams 2004). The precedent for UPS inhibition in cancer treatment has been estab-lished by bortezomib, which has been successful in the treatment of multiple myeloma and mantel cell lymphoma (Fisher et al. 2006; Orlowski and Kuhn 2008; Richardson et al. 2005). Together, this suggests that inhibition of the UPS is a rational strategy for cancer treatment.

The 26S proteasome is a large multi-enzyme catalytic complex located in the nucleus of eukaryotic cells and consists of a 20S enzymatic core and two regulatory 19S complexes (figure 1) (Adams 2004; Peters et al. 1993). The 20S unit has a barrel-shaped structure consisting of - and -subunits that harbour three distinct catalytically active enzymes characterized as chy-motryptic, tryptic and peptidylglutamyl-like (Adams 2004). Before destruc-tion of the proteasome, substrates are tagged with multiple ubiquitin mole-cules (Ravid and Hochstrasser 2008), which require the sequential action of the E1 ubiquitin-activating enzymes, E2 ubiquitin-conjugating enzymes and E3 ubiquitin-protein ligases. This is a dynamic process that can be reversed by the de-ubiquitylating enzymes (DUBs) that remove ubiquitin from sub-strates (Hoeller et al. 2006).

14

Figure 1. A schematic view shows how poly-ubiquitinated proteins are degraded by the proteasome. (Courtesy of Dr. Jenny Felth, Uppsala, Sweden)

Owing to the success of bortezomib, several second generation proteasome inhibitors are currently in clinical development e.g. carfilzomib, CEP18770, NPI-0052 and MLN9708. Like bortezomib, several of these compounds inhibit the chymotrypsin-like sites of the 20S subunit but differ in enzyme kinetics. Whether the binding is reversible or irreversible is an important feature with regard to activity and safety and might result in differences in tissue distribution (Bedford et al. 2011).

Apart from the proteasome itself, the enzymes involved in the UPS cas-cade represent novel drug targets. For example, the E1-activating enzymes (Xu et al. 2010; Yang et al. 2007b) and E3 ligases (Yang et al. 2005) have been suggested as novel therapeutic targets. Another strategy is to target the DUBs and ubiquitin like proteins (UBLs) (Bedford et al. 2011). Recently, the compound, WP1130 was shown to inhibit DUBs, leading to the accumu-lation of ubiquitinated proteins, without inhibiting the 20S enzymatic activi-ties (Kapuria et al. 2010). WP1130 has previously been shown to be active against CML in vivo (Bartholomeusz et al. 2007). Thus, compounds acting on new targets in the UPS have been shown to be effective in vivo without targeting the enzymatic activity of the proteasome.

20S Core Particle

19S Regulatory Particle

Ubiquitin

Peptide PolyUbiquitinated Protein Conjugate

Peptide Fragments

15

Toxicity of cancer drugs The toxicity of classic anticancer agents usually affects cells with a rapid turnover, such as blood forming cells and epithelial cells, but also tissues with metabolic functions and a substantial blood flow such as the liver and kidney (Kintzel 2001; Lee 2003). In drug development, it is therefore impor-tant to find novel anticancer drugs with both good clinical effects and low toxicity (Parent-Massin et al. 2010; Valeri et al. 2010).

Preclinical assays for normal cell toxicity

Efforts have been made to develop in vitro assays for predicting toxicity profiles and therapeutic index. Normal cells are often less proliferative and more fragile than tumour cells, making in vitro culturing something of a challenge. Ideally, primary cells should be used, which is possible for blood cells, but for most other tissues other solutions must be found. Primary kid-ney cells of high quality are commercially available, can be kept in culture and reflect the in vivo renal cell situation (Li et al. 2003; Li et al. 2006). Genetically altered normal cell lines may also be used as in vitro models for normal tissue. For example, the epithelial cell line, hTERT-RPE1, is immor-talized by the introduction of human telomerase reverse transcriptase (Airley 2009).

The most common method for measuring drug induced hepatoxicity is to use isolated rat hepatocytes. These cells have retained many of the functions of the in vivo hepatocyte, but are cumbersome to prepare and culture and have the disadvantage of being of non-human origin (Martignoni et al. 2006). Hepatotoxicity has also been measured using the liver cell line HepG2, which is easy to culture and gives access to an unlimited cell num-ber, but with the disadvantage of being of cancer origin and with some nor-mal functions are lost or decreased (Guillouzo 1998).

Preclinical assays for prediction of haematological toxicity One of the most common dose limiting adverse effects in cancer treatment is myelosuppression which describes adverse effects on the blood forming system (Parent-Massin et al. 2010). The rapid turnover of blood cells sensi-tizes them as targets for conventional chemotherapeutics, and this toxicity may lead to acute neutropenia, thrombocytopenia and/or anaemia (Ferlini et al. 2001).

16

Haematopoiesis Haematopoiesis is the formation of blood cells, derived from the haemato-poietic stem cell (HSC). The bone marrow is the primary site of HSC sur-vival, growth and development. The pluripotent HSC is capable of generat-ing virtually all of the cell types in the immune system and one stem cell is capable of producing one million mature blood cells after 20 cell divisions (Hoffbrand 2007; Metcalf 2010). The stem cells have the capability of self-renewal so that marrow cellularity remains constant in a normal healthy steady state. The term, self-renewal, means that the stem cell divides, in which one cell replaces the stem cell and the other is committed to differen-tiation. The HSC can generate lymphoid and myeloid progenitors which are also capable of self-renewal (Flores-Guzman et al. 2002; Kondo et al. 2003; Pessina et al. 2003). These early committed progenitors express low levels of transcription factors that commit them to a specific lineage, while growth factor stimulation determines the type of blood cells that are produced. The growth factors are glycoprotein hormones that regulate proliferation and differentiation of progenitor cells, affect the function of mature cells and prevent apoptosis. Growth factors are produced by the stromal cells with the exception of erythropoietin which is mainly synthesized in the kidney and thrombopoietin which is made largely in the liver (Hoffbrand 2007).

Colony forming unit assays The most frequently used in vitro method to predict drug induced myelosup-pression is colony forming unit (CFU) assays such as those for granulocyte-macrophage (CFU-GM), erythroid (CFU-E), and megakaryocyte (CFU-Mk) (Olaharski et al. 2009; Pessina et al. 2001; Pessina et al. 2005; Pessina et al. 2009). In CFU assays with human haematopoietic progenitors, cells are cul-tured in an immobilizing semi-solid medium in the presence of a growth factor cytokine cocktail, stimulating the cells to proliferate, differentiate and form colonies which are manually enumerated after 14 days of culturing (Gribaldo 2002; Masubuchi et al. 2004; Metcalf 2010; Pessina et al. 2001; Volpe and Warren 2003). The CFU assays offer the possibility to distinguish colonies from different lineages, but drawbacks include low throughput, subjective colony count and no quantification of colony which precludes its use in high throughput screening (Greenwalt et al. 2001; Rich and Hall 2005).

Liquid culture assays Undifferentiated haematopoietic cells are capable of giving rise to specific committed progenitors in liquid culture if appropriate cytokine stimulation is provided (Dal Negro et al. 2006; Ferlini et al. 2001; Greenwalt et al. 2001; Leglise et al. 1996). Thus, liquid culture assays have been used to evaluate myelotoxicity of antitumour agents. In these assays, inhibition of cell growth

17

or cell death have been quantified with dimethylthiazol diphenyltetrazolium bromide (Leglise et al. 1996), flow cytometry (Dal Negro et al. 2006), or by manual counting of viable cells after trypan blue staining (Ferlini et al. 2001). In the liquid culture assays described in the literature, the cells were cultured in 24- or 96-well plates which enable use in larger scale drug screens compared to the CFU assays. There is also a methylcellulose based assay known as the Hemotoxicity Assay via Luminescence Output (HALO®) that detects clonal expansion of cells by measuring luminescence of intracellular ATP in a 96-well plate (Reems et al. 2008; Rich and Hall 2005).

Species differences in drug sensitivity Preclinical testing of anticancer drugs largely rests on animal experiments. However, there are differences between animals used for preclinical testing and humans in sensitivity for anticancer drugs. The reason for species dis-crepancy can be either differences in pharmacokinetics and/or pharmacody-namics (Lindhagen et al. 2004). For example, mouse bone marrow is less sensitive to many cytotoxic agents compared to human bone marrow, allow-ing blood levels in preclinical efficacy testing that cannot be achieved in patients (Kurtzberg et al. 2009).

The starting dose of a new anticancer compound in clinical trials is based on the MTD in the most sensitive species. It is usually one tenth of the dose that causes severe toxicity in rodents (LD10), provided that it is tolerated in a non-rodent species (DeGeorge et al. 1998; Price et al. 2008). Generally, this extrapolation leads to reasonable starting doses, but occasionally there are major differences between the rodent LD10 and the human MTD which may cause great problems when entering clinical trials. If the MTD is underesti-mated, the phase I trial can be lengthy and include many patients on sub-therapeutic doses, but if the MTD is higher than predicted, then patient safety is at risk (Lindhagen et al. 2004).

In vitro tests to investigate species differences in drug sensitivity can aid in the animal/human extrapolation and provide a better basis for calculating clinical dosages and estimations of the levels of drug exposure tolerated by humans. The most common way to investigate species differences in vitro is by using the CFU-GM assay, which has proven useful when comparing the sensitivity of bone marrow between species (Bagley et al. 2009; Erickson-Miller et al. 1997; Kurtzberg et al. 2008; Kurtzberg et al. 2009; Pessina et al. 2003; Roda et al.). It has been suggested that by using the ratio of mouse/human CFU-GM IC90 values when calculating the human MTD, ad-justments for interspecies differences can be made and thereby reduce the risk an unexpected species difference in drug sensitivity (Pessina et al. 2003).

18

Antitumour activity of cancer drugs Screening for cancer drugs Screening for novel compounds with cytotoxicity or affinity for certain tar-gets is well established in anticancer drug development and can be applied to different collections of compounds (Chabner and Roberts 2005). For exam-ple, screening for cytotoxic activity can be performed among natural product extracts, in annotated compound libraries or drugs isolated through rational drug design i.e. substances synthesized to act against specific molecular tar-gets (Damia and D'Incalci 2009; Holbeck et al. 2010; Rickardson et al. 2006; Sharma et al. 2010). The initial stage of drug screening often involves a large number of compounds and is usually carried out using tumour cell lines (Shoemaker 2006). If a line of research is designed to target a particular cancer site or genetically specific subgroup, drug screens may be carried out using a panel of tumour cell lines corresponding to that tumour type (Sharma et al. 2010). To obtain more specific information from a screen and study single cells in detail, it is possible to perform image based screening which has proven useful in the identification of compounds with proteosomal activ-ity (Rickardson et al. 2007). Screening a large number of compounds can be very labour intensive. Some of the desirable qualities of a drug screening assay therefore include simplicity, low costs and reproducibility, and the assays must allow screening of a large number of compounds as well as pro-vide a reasonably accurate assessment of drug sensitivity (Sharma et al. 2010).

Early characterization of mechanism of action The National Cancer Institute has a panel of 60 cell lines (NCI60) that has been used to discover and develop novel anticancer drugs. The NCI60 panel has also been used together with the COMPARE algorithm to reveal the mechanism of action of new drugs, since it has been shown that drugs with similar mechanisms of action tend to have similar patterns of growth inhibi-tion in the NCI60 screen (Holbeck et al. 2010). Another tool for the identifi-cation of mechanism of action is to use drug induced gene expression profil-ing in the publicly available Connectivity Map (Cmap) database. In the Cmap, gene expression profiles have been generated from a well character-ized cancer cell line treated with small molecule compounds, creating a li-brary of signatures of drug responses. The signature (list of genes obtained) from one experiment can be used to find matches, assessing the degree of similarity between gene lists (Lamb et al. 2006).

19

Preclinical models for efficacy assessment There is a need for improved methods in the transition from promising pre-clinical results to choosing the cancer diagnoses that could most benefit from the treatment. The advancement from preclinical models to clinical trials is based on both in vitro and in vivo investigations. However, prediction of clinical outcome from these models is difficult and the disease-specific ac-tivity is still most frequently determined by empirical findings in clinical trials. With this approach there is a risk of giving a new agent to a large number of patients for whom the drug is ineffective, which is problematic from both ethical and financial perspectives. Thus, despite advanced tech-niques to identify anticancer drug candidates, a large gap remains between preclinical models and the optimal strategy for clinical trials (Damia and D'Incalci 2009; Peterson and Houghton 2004; Voskoglou-Nomikos et al. 2003).

In vivo models for assessment of diagnosis-specific activity Finding the ideal preclinical model for prediction of diagnosis-specific activ-ity is difficult. In vivo models such as human xenografts are commonly used and have the benefit of taking pharmacokinetics and active metabolites into consideration (Martignoni et al. 2006). The drawback is that similarities with clinical tumours in factors such as tumour size, tumour growth rate and de-gree of vascularization is probably limited and the risk of problems arising from species differences is ever present (Damia and D'Incalci 2009; Price et al. 2008). It should also be noted that in vivo studies for prediction of diag-nosis-specific activity require a large number of animals. Moreover, in stud-ies investigating the predictive value of xenografts, the correlation between activity in xenografts with the same human cancer histology was found to be weak (Johnson et al. 2001).

In vitro models for assessment of diagnosis-specific activity In a previous report, a correlation between the in vitro activity of anticancer drugs in primary cultures of tumour cells from patients and phase II activity (Fridborg et al. 1999) was shown. Compared to cell lines, primary tumour cells are more difficult to work with, as they tend to be non-proliferative and demand good laboratory skills for preparations. However, in tumour cells from patients, the disease-specific phenotype is largely intact and may there-fore be better predictors of clinical activity than cell lines (Fridborg et al. 1999). In addition, where a line of research requires a cell culture model that closely resembles the characteristics of a specific tumour type, for example from a particular cancer subtype where available cell lines are limited or from a cohort of patients in a clinical study, primary cells may be more ap-propriate (Airley 2009). The use of cell line panels has some utility in pre-dicting diagnosis-specific activity but has proven to be most useful in pre-

20

dicting mode of action (Holbeck et al. 2010). However, Sharma et al (Sharma et al. 2010) recently showed that the use of large cell line panels (approximately 1200 cell lines) could reflect the genomic heterogeneity of human cancer, and thereby provide diagnosis-specific information but due to their magnitude, using these kind of panels are labour intensive and accessi-ble to very few labs.

21

Aims

The aim of this thesis was to establish and evaluate preclinical methods use-ful in the development of new anticancer agents with favourable efficacy and toxicity profiles. The specific aims of the studies were: To develop a simple, robust and high-throughput alternative to the

CFU-GM assay for preclinical haematotoxicity studies, using CD34+ human umbilical cord blood cells

To establish a normal cell panel for preliminary evaluation of drug toxic-ity profile and assessment of therapeutic index

To establish a panel of primary lymphocytes for detection of species differences in drug sensitivity using cells from human, dog, rat and mouse

To evaluate the benefit of using primary tumour cells from patients for prediction of cancer diagnosis-specific activity of anticancer drugs

To apply the efficacy and toxicity models for experimental follow-up of a novel inhibitor for the UPS

22

Materials and Method

Drugs and reagents All drugs included in the studies (table 1) were obtained from commercial sources and dissolved according to manufactures’ instructions. Details are indicated in the papers. 384-well microplates (Nunclon Surface, NUNC Brand Products, Denmark) were prepared with 5 µl of drug solution in du-plicates at 10 times the desired final drug concentrations using the Biomek 2000 Pipetting Station (Beckman Coulter Inc, Fullerton, CA, USA). Serially diluted drug solutions in duplicate were used, with the exception of screen-ing of the Chembridge compound library, for which fixed single concentra-tions were used. The plates were stored at -70°C until further use.

Cell models CD34+ stem cells for reflection of bone marrow toxicity In FMCA-GM (Haglund et al. 2010), cryopreserved human umbilical cord blood CD34+ stem cells were purchased from 3H Biomedical AB (Uppsala, Sweden) and were rapidly thawed and suspended in a cell thawing medium. The cells were cultured in cell culture medium containing a cytokine cocktail for differentiation toward the granulocytopoetic lineage. Cell plating density was 250-500 cells/well in drug prepared 384-well plates. Viability was ana-lyzed after seven days of drug exposure.

Lymphocytes for reflection of haematological toxicity Peripheral blood mononuclear cells (PBMC) from healthy donors were iso-lated from peripheral blood using density gradient centrifugation on Ficoll-Paque. PBMC isolated as described consist of approximately 95% lympho-cytes and are therefore referred to as lymphocytes throughout this work. The cells were handled as described below for patient tumour samples, before plating at 37 500 cells/well in 384-well plates prepared with drug samples. Viability was analyzed after three days.

23

Table 1. Compounds investigated in papers I-IV.

Drugs included in the thesis Paper I Paper II Paper III Paper IV

acyclovir X amsacrine X arsenic trioxide X 2-amino-6-mercaptopurine X bortezomib X X X X busulfan X CB3 (Phosphoric acid, 2,3-dihydro-1,1-dioxido-3-thienyl diphenyl ester) X chembridge compound library * X carboplatin X chloramphenicol X chlorpromazine X CHS 828 X cisplatin X X X clozapine X cytarabine X X X docetaxel X doxorubicin X X X epirubicin X erlotinib X etoposide X X X fludarabine X X 5-fluorouracil X X X gefitinib X X gemcitabine X X hydroxyurea X imatinib X X irinotecan X lomustine X melphalan X X X X PKC412 X prednisolone X rapamycin X sorafenib X sunitinib X taxol X X 6-thioguanine X topotecan X vincristine X * A library of 10 000 compounds

24

Cells for reflection of liver, epithelial and renal toxicity

For studies on the toxic effects drugs on epithelial tissues, human telomerase reverse transcriptase subunit-retinal pigment epithelial cells (hTERT-RPE1) (Clontech Laboratories Palo Alto, CA, USA) were used. HepG2 (Sigma-Aldrich, ECACC, Salisbury, UK) is a liver cell line that was used for studies on liver toxicity. HepG2 and hTERT-RPE1 were sub-cultured twice a week. For studies on renal toxicity, Human Renal Proximal Tubular Epithelial Cells (HRPTEpiC) (ScienCell research laboratories, San Diego, CA, USA) were used. HRPTEpiC were cultured according to the manufacturer’s in-structions. HepG2, hTERT-RPE1 and HRPTEpiC were plated at 2 500 cells/well in 384-well plates prepared with drug samples and cell survival was analyzed after three days.

Lymphocytes for studies of species differences Peripheral blood lymphocytes from adult human volunteers, mixed race dogs, male DA rats and male BALB/c mice were used for prediction of spe-cies difference in drug sensitivity. The cells were commercially available at 3H Biomedical AB. Cell preparation was performed on three independent occasions, with pooled samples for rats and mice and from one individual each time for human and dog. Cell plating density was 50 000 cells/well and cytotoxicity of the 17 anticancer drugs was measured after three days.

Human patient tumour cells Primary tumour cells from patients were used for prediction of diagnosis-specific activity. Tumour samples were obtained by routine surgery, diag-nostic biopsy, or bone marrow/peripheral blood sampling after approval by the ethical committee at Uppsala University (Dnr 21/93 and 2007-237). The samples were obtained from patients with acute lymphocytic leukaemia (ALL), acute myelocytic leukaemia (AML), chronic lymphocytic leukaemia (CLL), chronic myelocytic leukaemia (CML), lymphoma, multiple mye-loma, breast cancer, colon carcinoma, non-small cell lung cancer, pseudo-myxoma peritonei, ovarian cancer and renal cancer. Leukemic cells were isolated from bone marrow or peripheral blood using density gradient cen-trifugation in Ficoll-Paque (1.007 g/ml). Tissue from solid tumour samples was processed by mincing with scissors and tumour cells were then isolated by collagenase dispersion followed by purification using density gradient centrifugation on Percoll. Cryopreserved cells were used in paper II with the exception of myeloma cells which were freshly prepared. Freshly prepared cells were used in paper IV. Cell viability was determined by trypan blue exclusion, and the proportion of tumor cells (>70%) was assessed by inspec-

25

tion of May-Grünwald-Giemsa-stained cytospin preparations. All cells were cultured in complete medium under standard culturing conditions. Cell plat-ing density was 5 000 cells/well for solid tumours and 12 to 40 000 cells/well for leukemic cells. Cell survival was analyzed after three days of drug exposure.

Human tumour cell lines The cell lines used for characterization of CB3 were the myeloma cell lines NCI-H929, RPMI 8226, 8226/Dox40 (selected for P-gp-mediated doxorubi-cin resistance), the T-cell leukemia cell line, CCRF-CEM, the acute myeloid leukaemia cell lines, Kasumi-1 and HL-60, and the colon cell line, HTC 116. Cells were plated at a density of 5 000 cells/well and cell survival was meas-ured after three days of drug exposure. The stably transfected HEK 293 Zs Green Proteasome Sensor cell line expressing the ZsProSensor-1 protein was used for screening and identification of proteasome inhibitors. The cell line, MelJuSo UbG76V-YFP, a human melanoma cell line expressing ubiquitin coupled to yellow fluorescent protein (YFP), was used to reflect activity of the proteasome system. All cell lines were cultured in complete medium and were subcultured twice a week.

Measurements of cytotoxicity Cell survival was measured using the non-clonogenic fluorometric microcul-ture cytotoxicity assay (FMCA), described previously in detail (Larsson et al. 1992; Lindhagen et al. 2008). Briefly, this method is based on the ability of cells with intact cell membranes to convert non-fluorescent fluorescein diacetate to fluorescent fluorescein. After three to seven days of incubation, the FMCA was performed using the automated Optimized Robot for Chemi-cal Analysis (Orca, Beckman Coulter) programmed with SAMI software (Beckman Coulter). Medium and drugs were aspirated, the cells were washed twice with PBS and fluorescein diacetate was added. After 50 to 70 min of incubation, fluorescence was measured at 485/520 nm in a fluorome-ter (Fluostar Optima, BMG Technologies, Germany). The fluorescence is proportional to the number of living cells. Cell survival is presented as sur-vival index, defined as fluorescence in experimental wells as a percentage of fluorescence in control wells, with blank values subtracted. Quality criteria: The fluorescence in the mean controls had to be five times higher than in mean blank values and the coefficient of variation less than 30%.

26

Screening for UPS inhibitors The screening for UPS inhibitors was performed as described (Rickardson et al. 2007) using the HEK 293 ZsGreen Proteasome Sensor Cell Line, which has been engineered to express an ornithine decarboxylase (ODC)-fusion green-fluorescent protein degraded by the proteasome without the need for prior ubiquitination.. The cells were plated in black optically clear bottom ViewPlates (PerkinElmer, Waltham, MA, USA), left to attach and then treated with 5 µM of compounds for 16 h. The nuclei were then stained with Hoechst 33342 and the plates were analysed in an ArrayScan II instrument (Thermo Fisher Scientific, Waltham, MA, USA). The criterion for a hit compound was a generated fluorescence of three standard deviations (SD) above the background fluorescence.

Activity assay for the 20S proteasome Quantification of the chymotrypsin activity of the 20S proteasome in the presence of CB3 was determined using the cell free 20S Proteasome Assay Kit (BostonBiochem, Cambridge, Massachusetts, USA) according to the manufacturer’s instructions. Briefly, 20S activity was measured by adding the CB3 and a known inhibitor (bortezomib) to 20S erythrocytes in reaction and activation buffer (25 mM HEPES, 0.5 mM EDTA, pH 7.6 and 0.03% SDS). The compounds were incubated with 20S enzyme solution for 15 min at 37°C before adding substrate solution (10µM Suc-LLVY-AMC). The increase in fluorescence was measured every third minute using excitation and emission wavelengths of 340 and 450 nm, respectively. Monitoring the increase in fluorescence over time allowed quantification of the enzymatic activity.

Connectivity Map Data from Connectivity Map (www.broadinstitute.org/cmap/) were downloaded and visualized using TMV4 (Saeed et al. 2003). The 30 most up- and down-regulated genes after CB3 treatment were selected from the two instances of CB3 (5155877, ChemBridge ID) treated MCF-7 cells. The expression values for the selected genes after treatment with the UPS inhibi-tors MG-132, MG-262, thiostrepton, 15-delta prostaglandin J2 (15d-PGJ2) and withaferin A were retrieved.

27

Statistics For tumor cells and epithelial, liver, lymphocyte and cell models, IC50 values (drug concentration resulting in 50% cell survival compared with controls) were calculated using non-linear curve fit with constraints set to a minimum value of 0 and a maximum value of 100 in the Hills Equation, using Graph-PadPrism software (GraphPad Software, Inc. San Diego, CA). For the FMCA-GM, data were processed in Microsoft Excel using a log linear inter-polation to calculate the log IC50 values. In cases where an IC50 could not be determined, the IC50 value was reported as the highest or lowest concentra-tion tested. Correlation was performed with Pearson’s correlation.

28

Results and discussion

Normal cell models for toxicity assessment The aim of papers I to III was to develop normal cell models suitable for a high throughput setting. Paper I describes the development of FMCA-GM, a high throughput alternative to the clonogenic assays using CD34+ cells in liquid culture. In paper II, FMCA-GM was complemented with additional normal cell models and a normal cell panel representing haematological, epithelial, renal and liver tissue was established. In paper III, we set out to investigate whether primary lymphocytes from human, dog, rat and mouse could be used to detect species differences in drug sensitivity.

Development of FMCA-GM, a high throughput alternative to CFU-GM (paper I) One of the most common dose limiting adverse events in cancer treatment is myelotoxicity. The aim of paper I was to develop an in vitro method for measuring the potential myelotoxic properties of drug candidates in a high throughput setting. In parallel, we investigated the possibility of replacing the golden standard of haematotoxicity assessment, the colony forming unit (CFU) assay, with the non-clonogenic FMCA.

Method development FMCA-GM was performed with CD34+ umbilical cord blood cells in a liq-uid culture, supplemented with cytokines to induce differentiation toward the granulocytic lineage. Method validation was performed using 24 drugs with known myelotoxic properties. Two types of assay set ups were evaluated; in FMCA-GM7, cells were exposed to drugs directly after thawing and cell survival was measured on day 7. In FMCA-GM14, on the other hand, the cells were cultured for seven days prior to plating and analyzed for viability on day 14 of differentiation. See flowchart in figure 2.

To be able to predict possible lineage specificity, efforts were made to es-tablish assays for the megakaryocyte/thrombopoietic (FMCA-Mk) and erythroid (FMCA-E) lineages. Under the experimental settings used, how-ever, the differentiation of the FMCA-E and FMCA-Mk lineages was poor, which excluded them from the study.

29

Drug exposure384-well plates, 250 cells/well

Culturing inU-shaped 96-well plates

Day 7 Day 14FMCA-GM14

Day 0

Drug exposure384-well plates, 500 cells/well

Day 7FMCA-GM7

Day 0

Figure 2. Flowchart of experimental design, in FMCA-GM7 cells are exposed to drugs directly after thawing. In FMCA-GM14 cells are cultured for 7 days prior to plating and drug exposure.

FMCA-GM14 is best suited to a high throughput setting One of the differences between the two assays set ups is that with FMCA-GM7, a higher proportion of the cells are primitive when exposed to drugs as with the CFU-GM assay, and in the FMCA-GM14, more mature cells are exposed. Consequently, FMCA-GM14 could potentially miss or underesti-mate the potential stem cell toxicity of a drug. When comparing the IC50 values from FMCA-GM7 and FMCA-GM14, however, they corresponded well and no obvious trends between the two assays or between cytotoxic and non-cytotoxic drugs were observed. The great advantage with FMCA-GM14 is that much less progenitor cells are required, since a small number of pro-genitor cells will proliferate greatly during one week of culturing in cytokine supplemented medium (Flores-Guzman et al. 2002). CD34+ umbilical cord blood cells are very expensive and this makes it preferable to use FMCA-GM14 in larger studies and for screening purposes. The labour intensity of the two assays is comparable, with either handling of very few cells in FMCA-GM7, or culturing for one extra week in FMCA-GM14.

IC50 values from CFU-GM correlates with those of FMCA-GM When the 50% inhibitory concentrations (IC50) from the FMCA-GM7 and FMCA-GM14 were correlated with IC50 from the CFU-GM assay obtained from literature, a high correlation was seen (r = 0.81 and r = 0.82, respec-tively). A difference between the assays that could affect outcome is the drug exposure time, which is 7 days for FMCA-GM and 14 days for human cells in CFU-GM.

FMCA-GM is a high throughput alternative to CFU-GM The CFU-GM has been used in assessment of haemtotoxicity for decades (Bradley and Metcalf 1966; Pessina et al. 2001). CFU-GM offers the possi-bility to distinguish colonies from different lineages, but drawbacks with these assays include low throughput, subjective colony count and no quanti-fication of colony (Rich and Hall 2005). The great benefit with FMCA-GM

30

is that it is performed in 384-well microplates with automated, objective cytotoxicity quantification. This makes it possible to investigate the mye-lotoxic potential of hundreds of compounds in 7 to 14 days with a robust assay. To conclude, in the present study we have demonstrated that FMCA-GM is a potential alternative to the CFU-GM assay for early evaluation of haematotoxicity. By replacing the CFU, time and labour will be saved, mak-ing FMCA-GM a possible solution for screening of large compound librar-ies.

Establishment of a normal cell panel (paper II) To complement FMCA-GM with additional normal cell models, a normal cell panel, including cells to reflect epithelial, liver, renal and haematological toxicity, was established (table 2).

Table 2. Display of normal cell models.

Normal cell models

Epithelial cells Human telomerase reverse transcriptase subunit pigment epithelial cells

Liver cells HepG2 cells, human liver cancer cell line

Lymphocytes Human peripheral blood mononuclear cells

Renal cells Human renal proximal tubular epithelial cells

FMCA-GM Human umbilical blood CD34+ stem cells

For method validation, the cytotoxic effects of 14 anticancer drugs were tested in the normal cell panel and the drug sensitivity was compared with known clinical toxicity profiles. For estimation of therapeutic index, the sensitivity of the normal cell panel was compared with a panel of primary tumour cells of both haematological and solid tumour origin. FMCA-GM differs from the other normal cell assays, in that cell survival is analyzed after seven days of drug exposure instead of three days and that the prolifera-tive capacity of the progenitor cells is higher. Effects of anti-proliferative drugs might thus have a larger impact on FMCA-GM. For this reason, FMCA-GM14 was presented separately in this study.

Normal cell models roughly predict toxicity profiles The normal cell panel was able to roughly predict the clinical toxicity profile when comparing in vitro toxicity with established side effects. The response

31

of the normal cells varied considerably between drugs. In agreement with clinical experience, epithelial, liver and renal cells were relatively resistant to drugs such as amsacrine, cytarabine and melphalan while lymphocytes were more sensitive (Ramström 2009). For bortezomib, myeloma cells were more sensitive than all normal cell types, which is in agreement with its rela-tively broad therapeutic window in myeloma treatment. Platinum agents, such as cisplatin, are well known for their renal toxicity (Kintzel 2001; Li et al. 2006) and normal renal cells were the most sensitive cell type in vitro, (figure 3). The high tolerability of imatinib was detected and CML cells were clearly more sensitive than any of the normal cells (figure 3) (Wolf and Rumpold 2009). For some of the other drugs tested the concordance between in vitro toxicity and known clinical adverse events was less evident and the normal cell panel rather gave a rough indication of therapeutic index.

Figure 3. Cytotoxic activity of cisplatin and imatinib in the normal cell panel to-gether with the median IC50 of the most sensitive tumour type for each drug.

FMCA-GM14 reflects the therapeutic window To investigate whether FMCA-GM14 can be used to rank compounds in terms of in vitro therapeutic index, the GM14 index was established. The GM14 index, was defined as the ratio of IC50 for FMCA-GM14 and median IC50 for the most sensitive diagnosis from the tumour cells tested and is dis-played in figure 4. The highest GM14 index was seen for the targeted drugs imatinib, bortezomib and gefitinib, reflecting a broad therapeutic window with respect to myelotoxicity. A lower GM14 index was seen for conven-tional cytotoxic agents and the immunosuppressant, rapamycin. With the exception of 5-fluorouracil and rapamycin, this is in reasonable agreement with clinical tolerance.

32

Figure 4. GM14 index of the drugs, defined as the ratio between the mean IC50 from the FMCA-GM14 and the median IC50 for the most sensitive tumor type in vitro. The most sensi-tive tumor type for each drug was: amsacrine-ALL, arsenic trioxide-CML, bortezomib-myeloma, cisplatin-ovarian, cytarabine-lymphoma, doxorubicin-lymphoma, etoposide-ALL, 5-fluorouracil-NSCLC, gefitinib-AML, imatinib-CML, melphalan-ALL, PKC412-CLL, rapamycin-ALL and vincristine-CLL.

Development of an in vitro species difference test using lymphocytes (paper III) The aim of paper III was to investigate whether lymphocytes from peripheral blood from human, dog, rat and mouse could be used to detect species dif-ferences in cellular drug sensitivity using the FMCA.

Method validation For method validation, the cytotoxic effects of 17 anticancer drugs from different mechanistic classes were tested on lymphocytes from the four se-lected species in 384-well microplates, in a high throughput setting. This study was planned to include peripheral neutrophils in parallel to the lym-phocytes, but viability problems in the three day assay excluded them from the study.

Dog lymphocytes showed similar sensitivity to the human cells The estimated IC50 for each drug was used to calculate a ratio of human/dog cells and human/rodent cells, expressing the human drug sensitivity com-

33

pared to the three other species. For most of the drugs tested, dog lympho-cytes showed similar sensitivity to the human cells. Rat and mouse cells were similarly drug sensitive and were generally more drug resistant. This is consistent with previous findings (Price et al. 2008).

Species differences in drug sensitivity were detected A species difference was seen for three of five antimetabolites tested; for gemcitabine, fludarabine and cytarabine, the human cells were more sensi-tive, but this phenomenon was not seen for 6-thioguanine and 5-fluorouracil. Gemcitabine, fludarabine and cytarabine all need to be activated by deoxy-cytidine kinase (Bagley et al. 2009; Rooseboom et al. 2004), while 6-thioguanine and 5-fluorouracil are not dependent on this enzyme. Humans have higher levels of deoxycytidine kinase compared to mice, which can lead to a higher exposure (Bagley et al. 2009).

For the camptothecins, it is well known that rodents tolerate a higher drug exposure than humans (Erickson-Miller et al. 1997; Kurtzberg et al. 2008; Masubuchi et al. 2004; Thompson et al. 1998). This has also been shown in the CFU-GM assay (Erickson-Miller et al. 1997; Masubuchi et al. 2004). In our study, a similar trend was detected for the camptothecin topotecan.

For the anthracyclins, human cells were slightly more sensitive than ro-dent cells. This has not been reported from the CFU-GM assays or from in vivo data (Fuse et al. 1994; Pessina et al. 2003) and the reason for this dis-crepancy is not known. However, for several of the other mechanistic groups tested e.g. alkylating agents, platinum agents and microtubule stabilisers, the drug effects were similar on lymphocytes from all four species, which is in agreement with previous findings (Clark et al. 1999; Fuse et al. 1994).

For the proteasome inhibitor, bortezomib, a clear species difference was shown, where the rodent cells were 50 to 300 times less sensitive than dog and human cells. This is consistent with the fact that there was no haemato-logical toxicity in animal studies, while thrombocytopenia is a well known adverse effect in humans (Adams et al. 1999).

Comparing usage of primary lymphocytes with CFU-GM The IC50 ratios from our study were compared with published data (Pessina et al. 2003). Data from the nine drugs used in both studies were included in the comparison. The human/mouse IC90 ratios from the CFU-GM assay and in vivo MTD/LD10 ratios were correlated with the IC50 ratios from our non-clonogenic assay and had correlation coefficients of 0.64 and 0.60, respec-tively. For example, the results differed for the anthracyclins, where human and dog cells were more sensitive than in our study. This was not seen in the CFU-GM assay or in vivo studies.

34

Primary lymphocytes can be used to detect species differences. Using lymphocytes from peripheral blood to compare cytotoxic effects be-tween species is methodologically simple and straightforward. The great benefit with the non-clonogenic FMCA as a tool for measuring cytotoxicity is that it is high-throughput compared to the labour intensive CFU-GM as-say. The drawback is that the sensitivity of haematopoietic stem cells, which are believed to be the most important target for haematological toxicity, is less likely to be reflected in this assay. Compared to using the mononuclear bone marrow cells in the CFU-GM, however, lymphocytes are easy to han-dle, cheap and easily accessible. To conclude, these current data indicate that by testing lymphocytes from different species, it is possible to predict in vivo species differences in drug sensitivity.

Summary of the methods developed in papers I, II and III A persistent challenge in drug discovery is to develop in vitro assays that can accurately predict potential toxicological risk. The aim of papers I to III was to develop toxicity models suitable for a high throughput setting, and this had to be considered when choosing cells for the assays.

Ideally when measuring toxicity on normal tissue, primary cells from each organ would be used, but the availability and possibility to culture pri-mary cells are limited. For example, among blood cells, both lymphocytes and neutrophils are easily accessible but neutrophils are only viable for a few hours in culture (Nagami et al. 2002). For cells originating from other tis-sues, other solutions such as using immortalized cell lines must be found.

Establishing a model reflecting liver toxicity was especially difficult. The golden standard is to use primary rat hepatocytes, but the cells are difficult to prepare and are of non-human origin. The human liver cell line, HepG2, used in this panel is easy to culture, but has the obvious drawback of being of cancer origin.

Hence, there are difficult issues to consider when choosing a cell model that ideally should be easily accessible, robust and easy to culture but still be as close as possible to the original phenotype.

Application of the normal cell models The in vitro normal cell models established in papers I to III are useful in contributing knowledge of potential toxicity profiles in preclinical drug de-velopment. Some of the tissue specific toxicities were well predicted using the normal cell panel but it will probably be most useful in giving a rough prediction of therapeutic index rather than information on each specific or-gan. A possible application of the normal cell panel is when choosing among drug candidate analogues with similar tumour activity, and selecting the candidates with most beneficial therapeutic index. Another application is in

35

early characterization of a lead candidate. To complement the normal cell panel, there is also the possibility to make a new effort to develop the FMCA-Mk and FMCA-E, establishing myelotoxicity measurements on the different lineages. Although these in vitro systems cannot fully mirror the complex interactions in vivo and replace in vivo studies, they provide impor-tant predictive information, and may allow a reduction in animal testing. The species differences assay could be used for selection of an appropriate in vivo model, interpretation of in vivo results or at the latest as an aid in the selection of phase I starting doses before entering clinical trials.

36

Patient tumour cells for efficacy assessment In paper II, a panel of patient tumour cells of both haematological and solid tumour origin was established for efficacy assessment of 14 anticancer drugs.

Patient tumour cells used for prediction of diagnosis-specific activity (paper II) In a previous report from our group, a correlation between the in vitro activ-ity of anticancer drugs in primary patients tumour cells and phase II activity was demonstrated (Fridborg et al. 1999). Since then, the FMCA has been upgraded from a 96- to a 384-well format, allowing testing of full concentra-tion-effect curves in a highly automated setting instead of single concentra-tions. The aim of this study was to investigate the benefit of using primary patient samples in this upgraded system. Efforts were made to collect over one hundred tumour samples from a variety of diagnoses over several years. This study was performed and planned in the beginning of this century, which is reflected in the choice of drugs. For example, the targeted drugs included were novel at the time and are now clinically established. Another aspect is that the different subgroups of diagnoses were not genetically char-acterized, which has become more common in recent years

Tumour cells used The antitumour activity of 14 anticancer drugs, classical cytotoxic agents as well as targeted drugs, was tested in primary tumour samples from patients with solid or haematological malignancies (table 3). Tumour samples were obtained by routine surgery, diagnostic biopsy, or bone marrow/peripheral blood sampling after approval by the ethical committee at Uppsala Univer-sity. For method evaluation the in vitro activity of the drugs was compared with indications approved by the Food and Drug Administration (FDA 2010).

37

Table 3. Display of diagnoses included.

Diagnoses included in the study

Acute lymphocytic leukaemia (ALL) Acute myelocytic leukaemia (AML) Chronic lymphocytic leukaemia (CLL) Chronic myelocytic leukaemia (CML) Lymphoma Multiple myloma Breast cancer Colon cancer Non-small cell lung cancer Ovarian cancer Renal cancer

In vitro drug activity in tumour cells is in accordance with clinical use of the drugs In general, in vitro drug activity in patient tumour cells reflected the known clinical activity of the drugs investigated. The activity in leukemias for typi-cal leukemic drugs such as amsacrine, arsenic trioxide, bortezomib, cytara-bine, melphalan and vincristine was well reflected in vitro including the very specific effect of imatinib on CML. In addition, the solid tumour effects of cisplatin and 5-fluorouracil was well reflected.

In detail, myeloma samples were sensitive to bortezomib and relatively sensitive to melphalan, which are important drugs in the treatment of mye-loma. Ovarian cancer samples were clearly more sensitive to cisplatin than cells from other diagnoses, which is consistent with the main clinical use of the drug. Doxorubicin is approved for use in both solid and haematological malignancies, which corresponds with its in vitro activity. Etoposide had in vitro activity in leukaemias, which are not approved indications by the FDA. In Sweden, however, AML and lymphoma, as well as lung and testicular cancer, are approved indications (FASS 2010).

Data evaluation using approved indications by the FDA Using approved indications by the FDA to measure clinical activity of drugs, as performed in this study is simple and robust, but is limited by a high off-label use of the older drugs. Another option would have been to use phase II data, but it was difficult to obtain comparable data for the 14 drugs tested in the 11 diagnoses. For example, in most cases, treatment protocols are based on the combination of many different drugs.

Application of primary tumour cells It should be emphasized that although in vitro drug activity in a specific tumour type reflects clinical efficacy fairly well, the correspondence is far

38

from perfect since clinical efficacy depends on a number of factors not in-cluded in the assay, such as tumour cell interactions with stroma, the extra-cellular milieu of tumours, achievable plasma drug concentrations and dis-tribution to tumour tissue (Airley 2009; Damia and D'Incalci 2009). The advantage with using tumour cells from patients is that the disease-specific phenotype is largely intact with respect to drug sensitivity (Fridborg et al. 1999). The use of patient samples can be refined further by using genetically characterized tumour samples (Hallbook et al. 2005). Hence, there is the possibility to test very specific subtypes of diagnoses, even if no established cell lines or xenograft models are available.

In conclusion, the results support the hypothesis that tumour cells from patient samples can be used for the prediction of cancer diagnosis-specific activity in the preclinical stage of drug development and may aid in choosing diagnoses for clinical trials.

Future studies Conducting a study of this proportion requires cooperation among interested clinicians, logistics and laboratory skills and to perform a similar study again would need input and collaborations between different professional groups. Cause for optimism lies in the U-CAN project, a collaboration between aca-demic institutions and regional hospitals. The U-CAN project aims at col-lecting and storing patient tissue and blood in a biobank together with sam-ple-related information, patient clinical data and digital images. Currently, samples are being collected from prostate and colorectal cancer patients, as well as patients with haematological, neuroendocrine and brain malignancies at Uppsala University Hospital (U-CAN 2011).

39

Application of efficacy and toxicity models In paper IV, the screening for a novel inhibitor of the UPS in a compound library resulted in one hit, CB3 (figure 5). In the characterization of CB3, several of the models described in this thesis were applied and turned out to be useful. A preclinical assessment of the activity and toxicity profile of CB3 was made, as well as mechanistic studies. Studies of species differences have not yet been conducted but are important if CB3 is taken further in the drug development process.

Figure 5. Structural formula of CB3 (Phosphoric acid, 2,3-dihydro-1,1-dioxido-3-thienyl diphenyl ester). (Illustrated by Dr. Jenny Felth, Uppsala, Sweden)

Identification of a novel UPS inhibitor, CB3 (paper IV) In recent years, the ubiquitin proteasome system (UPS) has emerged as an interesting target for the development of new anticancer compounds. The clinically approved drug bortezomib (Velcade®) with good antitumour activ-ity and an acceptable therapeutic index (Fisher et al. 2006; Richardson et al. 2003) is a successful example. In this work, we therefore set out to find novel UPS inhibitors with favourable toxicity profiles.

Screening for a novel UPS inhibitor A library of 10 000 compounds was screened for the ability to induce cyto-toxicity in HCT 116 colon carcinoma cells using the FMCA assay as a read-out. 382 compounds showed cytotoxic activity and when screening the 382 bioactive compounds for UPS activity, CB3 (Phosphoric acid, 2,3-dihydro-1,1-dioxido-3-thienyl diphenyl ester) was the only hit compound (figure 6). Our hypothesis was strengthened when searching the cmap, since the changes in gene expression induced by CB3-treatment (compound 5155877 in cmap) was similar to other compounds previously shown to act as UPS inhibitors, such as the experimental proteasome inhibitors MG-132 and MG-262 as well as withaferin A and 15d-PGJ2 (Shibata et al. 2003; Yang et al. 2007a). To study the UPS activity of CB3 in more detail, we investigated the ability of the compound to accumulate ubiquitin in the melanoma cell line MelJuSo UbG76V-YFP, which was confirmed.

40

Screening of 10 000 compounds at 25 µM

in HCT 116

382 compounds showed cytotoxic activity

Screening for inhibitors of the ubiquitin-proteasome system (UPS)

One compound was identified, CB3

Figure 6. Flowchart of the screening procedure and identification of CB3 as a hit compound.

CB3 does not inhibit the proteasomal chymotrypsin activity Since accumulation of ubiquitin is non-specific and would be the result of most UPS-inhibitors, we continued to examine the effect of CB3 on the 20S enzymatic core of the proteasome. The chymotrypsin activity in the 20S subunit was measured in a cell-free assay, but CB3 did not inhibit the enzy-matic activity at all. These results suggest that unlike bortezomib, CB3 does not block 20S activity but does cause accumulation of ubiquitin by acting on other parts of the UPS.

CB3 displayed tumour activity of in cell lines and primary tumour samples When investigating the cytotoxic activity of CB3, myeloma cell lines and leukaemia cell lines were the most sensitive, although the colon cancer cell line included in the study was surprisingly sensitive. The cytotoxic activity of CB3 was also investigated in patient tumour cells from ALL, CLL, colon cancer and pseudomyxoma peritonei. No obvious difference in activity of CB3 was detected between solid and haematological tumours, indicating a potential usefulness in the treatment of solid tumours. Ideally, a larger num-ber and a wider range of diagnoses would have been included in the study, with multiple myeloma of particular interest, but the access to clinical pa-tient samples was limited.

Assessment of toxicity profile of CB3 The toxicological properties of CB3 were investigated in the normal cell panel (paper II) together with two myeloma cell lines and primary CLL samples to provide information on therapeutic index. CB3 displayed an over-

41

all beneficial therapeutic profile. Normal cells of epithelial and renal origin were less sensitive to CB3 than tumour cells, and tumour cell selectivity was seen when comparing lymphocytes with their malignant counterpart, CLL. FMCA-GM14 was slightly more or equally sensitive compared to tumour cells.

CB3 is a UPS inhibitor of interest In summary, treatment with CB3 causes cellular accumulation of ubiquiti-nated proteins and the gene expression profile induced by CB3 is similar to the gene profiles of other known UPS inhibitors. Unlike bortezomib, CB3 does not inhibit chymotrypsin activity. CB3 is cytotoxic to both haemato-logical and solid tumour cells and has a promising toxicity profile. To con-clude, CB3 is a novel inhibitor of the UPS with a promising toxicity profile but the exact mechanism of action remains to be elucidated.

42

Conclusions and future perspectives

The results of this thesis may be summarized as follows: FMCA-GM assay is useful for prediction of the myelotoxic potential of

anticancer drug candidates and can be used as an alternative to CFU-GM assays

The normal cell panel is suitable for a high throughput setting and can be useful in predicting overall therapeutic index, although information of specific tissues was not as evident

A panel of lymphocytes from human, dog, rat and mouse can detect species differences in anticancer drug sensitivity and can be used as an alternative to the CFU-GM when studying species differences

Primary patient tumour cells used in a high throughout, automated set up can roughly detect diagnosis-specific activity of anticancer drugs and can aid in the selection of diagnosis in early clinical trials

The efficacy and toxicity models can be used for experimental follow-up of a novel drug candidate. The novel inhibitor of the UPS, CB3 dis-played antitumour activity and had a promising toxicity profile, although the exact mechanism remains to be elucidated

It has been estimated that a novel compound entering phase I testing has only an 8% chance of reaching the market and that the probability is even lower for anticancer drugs (Tan et al. 2009). The reason for this failure is non tolerable toxicity and/or an unsatisfactory clinical effect. Therefore, it is important, with improved characterization of pharmacological properties in the early stages of drug development, to reduce the risk of a late stage rejec-tion.

The main aim of this thesis was to establish informative in vitro assays for prediction of both antitumour activity and toxicity. These models are not yet sufficiently predictive to replace animal studies but they showed results in concordance with clinical experience. It was concluded that if both tumour and normal cell panels are used together, important information regarding therapeutic index can be retrieved which can be useful in the selection of drug candidates or in characterization of lead compounds. The methods pre-sented in this thesis can also contribute with knowledge of species differ-ences and diagnosis-specific activity in a preclinical phase of anticancer drug development.

43

Svensk populärvetenskaplig sammanfattning

Utvecklingen av nya cancerläkemedel har hjälpt många patienter till för-längd överlevnad, men tyvärr är det långt ifrån alla som botas. Många can-cerformer svarar inte på läkemedelsbehandling och resistensutveckling är ett vanligt förekommande problem. Dessutom ger de flesta cancerläkemedel biverkningar. Särskilt utsatta är vävnader med snabbt delande celler t.ex. benmärg samt vävnader med stort blodgenomflöde som njure och lever. Det är därför viktigt att hitta nya läkemedel som har effekt på tumörceller men inte är alltför giftiga för normala celler.

När man tar fram nya cancerläkemedel ligger fokus oftast på experiment för att undersöka läkemedlens effekt på tumörceller, men intresset för att även försöka förutsäga biverkningar med hjälp av laborativa metoder ökar. Målet med denna avhandling var att utveckla och testa metoder för att förut-säga både biverkningar och tumöreffekter av nya läkemedelssubstanser.

Skador på de blodbildande cellerna i benmärgen är oftast begränsande för hur höga cytostatikadoser som kan ges. I delarbete I utvecklade vi en ny metod, FMCA-GM, för att mäta hematologisk toxicitet med hjälp av odlade benmärgsceller (CD34+-stamceller). För utvärdering av FMCA-GM använ-des cancerläkemedel med redan känd biverkningsprofil och vår metod visade sig överensstämma väl med den etablerade metoden CFU-GM i förmågan att förutsäga hematologisk toxicitet. I delarbete II kompletterades FMCA-GM med ytterligare celltyper från bl.a. lever och njure, och en hel toxicitetspanel etablerades. Några olika läkemedels toxiska effekt i toxicitetspanelen jäm-fördes sedan med välkända kliniska biverkningar. Toxicitetspanelen kunde prediktera terapeutiskt fönster, rangordna läkemedel med avseende på hur mycket biverkningar de ger och i viss mån spegla toxicitet hos enskilda cell-typer. Ett användningsområde för denna typ av panel kan vara att enkelt få preliminär toxikologisk information om nya substanser vilket kan vara till hjälp i urvalet och karakteriseringen av potentiella läkemedelskandidater.

För att bestämma startdosen när en substans prövas första gången i män-niska används för det mesta maximal tolererad dos från djurstudier, oftast från mus eller råtta men även ibland hund. Många gånger ger den extrapole-ringen rimliga startdoser. Men för vissa läkemedel är skillnaden i känslighet mellan djur och människa stor vilket kan leda till såväl oväntad toxicitet som utebliven effekt. En för hög startdos vara kan vara farligt för patienten och en för låg startdos gör prövningen långdragen och kostsam. Delarbete III beskriver hur man med hjälp av vita blodkroppar framrenade från blodprover

44

från människa, hund, råtta och mus kan förutsäga skillnader i läkmedels-känslighet mellan olika arter. Denna metod kan t.ex. användas för att välja den bäst lämpade djurmodellen, tolka resultat från djurstudier eller för att optimera valet av startdos i kliniska studier.

Innan ett läkemedel testas i kliniska studier är det viktigt att veta vilka tumörtyper som kan ha nytta av behandlingen. Både ur ett etiskt och kost-nadsmässigt perspektiv är det olämpligt om läkemedel ges till patientgrupper där det inte har någon effekt. I delarbete II undersöktes nyttan av att använda tumörcellsprover från patienter för att förutsäga diagnos-specifik aktivitet. Effekten av ett antal etablerade cancerläkemedel testades i över hundra tu-mörprover från både solida och hematologiska diagnoser och dessa resultat jämfördes sedan med hur dessa läkemedel används i kliniken. Den kliniska aktiviteten speglades generellt bra, vilket innebär att primära tumörceller från patienter kan vara användbara i valet av patientgrupper till kliniska stu-dier.

Tumörceller producerar stora mängder proteiner eftersom de delar sig så ofta och behöver ett fungerande sätt att bryta ner dessa. Denna nedbrytning av proteiner sker i huvudsak i cellorganellen proteasomen som på så vis hål-ler cellen i balans. Det är känt sedan tidigare att cancerceller är känsligare för proteasomhämning än normalceller och det finns ett godkänt läkemedel som verkar genom att hämma proteasomen, bortezomib (Velcade®). Borte-zomib används kliniskt bl.a. i diagnosen multipelt myelom. Eftersom protea-somen är ett intressant mål för cancerbehandling försökte vi hitta nya sub-stanser med proteasomhämmande aktivet. I screening av ett bibliotek bestå-ende av 10 000 substanser fann vi CB3, en substans som hämmar proteaso-men. I delarbete IV testades den redan beskrivna toxicitetspanelen och primära tumörceller för att karakterisera CB3. I denna preliminära undersök-ning hade CB3 både god effekt på tumörceller och relativ lite toxicitet på normala celler. Exakt hur CB3 utövar sin effekt på proteasomen är dock inte känt. Då CB3 är en intressant substans som visat lovande resultat kommer den att studeras vidare.

Sammanfattningsvis kan de metoder som beskrivs i denna avhandling an-vändas i olika steg av läkemedelsutvecklingen och därmed bidra i karakteri-seringen av nya cancerläkemedel.

45

Acknowledgements

There are a number of people to whom I would like to express my gratitude for their support during my work on this thesis:

Dr. Elin Lindhagen, my supervisor, for all the valuable advice and guidance, encouragement and for your enthusiasm to answer all of my everyday-questions.

Dr. Anna Åleskog, my co-supervisor, for your support, for always being optimistic and for generously sharing your knowledge and experience.

Professor Rolf Larsson, my co-supervisor, for providing me with the oppor-tunity to work in your group, for being full of ideas and for sharing your extensive scientific knowledge.

Dr. Martin Höglund, for support and contributing with your vast expertise.

Dr. Linda Rickardson, for giving me an excellent introduction to research during my exam work, and for all the help, support and encouragement, I value this very much. Also, for being a true friend and for all the great mo-ments hanging out.

Dr. Malin Wickström, for the help and encouragement during these years, your support has meant very much to me, not least our stimulating discus-sions on a great variety of topics, and for our friendship.

Anna Eriksson, for all the support, positive energy, and for sharing your knowledge, but above all, for being a great friend.

Dr. Mårten Fryknäs, for great company and for the interesting discussions and collaborations.

Sara Laitinen, for being a friend and a very nice contribution to our group.

46

My co-authors, Dr. Saadia Bashir Hassan, Dr. Lena Douhan Håkansson, Dr. Joachim Gullbo, Professor Stig Linder and Professor Peter Nygren for your constructive help with the manuscripts and for contributing with your exper-tise.

Christina Leek, Lena Lenhammar and Nasrin Najafi for all the skilful help in the lab and work with the cell lines. A special thanks to Lena for all your help and support during my experiments.

Kristin Blom, Annika Jonasson, Anna-Karin Lannergård, Nadja Lundström and David Munro for skilful help with patient samples and other work in the lab. Lena Fredriksson and Gunilla Frenne for good advice.

Dr. Jenny Felth, for support, good company and collaborations. Thanks also for providing me with the nice illustrations for my thesis.

Dr. Claes Andersson, for all the help with computer and mathematically related issues.

Maria Axetun, for skilful assistance during your exam-work.

Mao Mao, for being a good companion during teaching.

Sofie Schwan, for the nice chats during breaks.

Other colleagues in the “Cancer Pharmacology and Informatics Group”, Professor Mats Gustafsson, Hanna Göransson Kultima, Anna Haukkala, Dr. Anders Isaksson, Dr. Daniel Laryea, Dr. Erika Nordberg, Markus Rasmus-sen, Maria Rydåker and Dr. Helena Strömbergsson for good advice and col-laborations.

Kristin Bryon, Elisabet Harryson and Eva Prado for all the help with admin-istrative issues.

All other colleagues at the Department for Clinical Pharmacology for con-tributing to a friendly atmosphere.

Dr. Simon Miles and Professor Jenefer Blackwell for generously hosting us during our visits to Perth and Wiluna. You contributed to making the trip “Down under” a great experience.

Professor Peter Klinken and Dr. Louise Winteringham at the Western Aus-tralian Institute for Medical Research in Perth for hosting me at your lab.

47

All other friends outside the lab, especially Caroline Banemark for being a great friend.

Anita, Göran and Johan for all the support and encouragement and all the other Wickströms for good company.

My family for their encouragement, help and believing in me: My grandpar-ents, Eva and Sven Eric for all the good times, my sister Anna for always sharing good advice and for being a wonderful sister, Kalle and Nils for great company, my parents Elisabeth and Lars for their never ending love and support.

Magnus, for your love and for always believing in me. Also for keeping my mind on something else. Life with you is never boring.

48

References

Adams, J. (2004). The proteasome: a suitable antineoplastic target. Nature reviews 4, 349-60.

Adams, J., Palombella, V. J., Sausville, E. A., Johnson, J., Destree, A., Lazarus, D. D., Maas, J., Pien, C. S., Prakash, S., and Elliott, P. J. (1999). Proteasome in-hibitors: a novel class of potent and effective antitumor agents. Cancer research 59, 2615-22.

Airley, R., ed. (2009). Cancer Chemotheraphy. John Wiley & Sons LTD, West Sussex, UK.

Bagley, R. G., Roth, S., Kurtzberg, L. S., Rouleau, C., Yao, M., Crawford, J., Krumbholz, R., Lovett, D., Schmid, S., and Teicher, B. A. (2009). Bone marrow CFU-GM and human tumor xenograft efficacy of three antitumor nucleoside analogs. International journal of oncology 34, 1329-40.

Bartholomeusz, G. A., Talpaz, M., Kapuria, V., Kong, L. Y., Wang, S., Estrov, Z., Priebe, W., Wu, J., and Donato, N. J. (2007). Activation of a novel Bcr/Abl de-struction pathway by WP1130 induces apoptosis of chronic myelogenous leu-kemia cells. Blood 109, 3470-8.

Bedford, L., Lowe, J., Dick, L. R., Mayer, R. J., and Brownell, J. E. (2011). Ubiq-uitin-like protein conjugation and the ubiquitin-proteasome system as drug tar-gets. Nat Rev Drug Discov 10, 29-46.

Bradley, T. R., and Metcalf, D. (1966). The growth of mouse bone marrow cells in vitro. Aust J Exp Biol Med Sci 44, 287-99.

Cancerfonden (2011). www.cancerfonden.se. Chabner, B. A., and Roberts, T. G., Jr. (2005). Timeline: Chemotherapy and the war

on cancer. Nature reviews 5, 65-72. Clark, D. L., Andrews, P. A., Smith, D. D., DeGeorge, J. J., Justice, R. L., and Beitz,

J. G. (1999). Predictive value of preclinical toxicology studies for platinum anti-cancer drugs. Clin Cancer Res 5, 1161-7.

Dal Negro, G., Vandin, L., Bonato, M., and Sciuscio, D. (2006). Toward refinement of the colony-forming unit-granulocyte/macrophage clonogenic assay: inclusion of a metabolic system. Toxicol In Vitro 20, 743-9.

Damia, G., and D'Incalci, M. (2009). Contemporary pre-clinical development of anticancer agents--what are the optimal preclinical models? Eur J Cancer 45, 2768-81.

DeGeorge, J. J., Ahn, C. H., Andrews, P. A., Brower, M. E., Giorgio, D. W., Go-heer, M. A., Lee-Ham, D. Y., McGuinn, W. D., Schmidt, W., Sun, C. J., and Tripathi, S. C. (1998). Regulatory considerations for preclinical development of anticancer drugs. Cancer chemotherapy and pharmacology 41, 173-85.

Erickson-Miller, C. L., May, R. D., Tomaszewski, J., Osborn, B., Murphy, M. J., Page, J. G., and Parchment, R. E. (1997). Differential toxicity of camptothecin, topotecan and 9-aminocamptothecin to human, canine, and murine myeloid pro-genitors (CFU-GM) in vitro. Cancer chemotherapy and pharmacology 39, 467-72.

49

FASS, ed. (2010). FASS. The Swedish Drug Compendium. Läkemedelsindustrifö-renningen AB.

FDA (2010). http://dailymed.nlm.nih.gov/dailymed/about.cfm, Food and Drug Ad-ministration.

Ferlini, C., Distefano, M., Pierelli, L., Bonanno, G., Fattorossi, A., Battaglia, A., Mancuso, S., and Scambia, G. (2001). A new method to evaluate in vitro mye-lotoxicity of antitumour agents in the first steps of drug development. Pharma-col Toxicol 89, 231-6.

Fisher, R. I., Bernstein, S. H., Kahl, B. S., Djulbegovic, B., Robertson, M. J., de Vos, S., Epner, E., Krishnan, A., Leonard, J. P., Lonial, S., Stadtmauer, E. A., O'Connor, O. A., Shi, H., Boral, A. L., and Goy, A. (2006). Multicenter phase II study of bortezomib in patients with relapsed or refractory mantle cell lym-phoma. J Clin Oncol 24, 4867-74.

Flores-Guzman, P., Gutierrez-Rodriguez, M., and Mayani, H. (2002). In vitro prolif-eration, expansion, and differentiation of a CD34+ cell-enriched hematopoietic cell population from human umbilical cord blood in response to recombinant cy-tokines. Arch Med Res 33, 107-14.

Fridborg, H., Jonsson, E., Nygren, P., and Larsson, R. (1999). Relationship between diagnosis-specific activity of cytotoxic drugs in fresh human tumour cells ex vivo and in the clinic. Eur J Cancer 35, 424-32.

Fuse, E., Kobayashi, S., Inaba, M., Suzuki, H., and Sugiyama, Y. (1994). Applica-tion of pharmacokinetically guided dose escalation with respect to cell cycle phase specificity. J Natl Cancer Inst 86, 989-96.

Greenwalt, D. E., Szabo, J., and Manchel, I. (2001). High throughput cell-based assay of hematopoietic progenitor differentiation. J Biomol Screen 6, 383-92.

Gribaldo, L. (2002). Haematotoxicology: scientific basis and regulatory aspects. Altern Lab Anim 30 Suppl 2, 111-3.

Guillouzo, A. (1998). Liver cell models in in vitro toxicology. Environ Health Per-spect 106 Suppl 2, 511-32.

Haglund, C., Aleskog, A., Hakansson, L. D., Hoglund, M., Jacobsson, S., Larsson, R., and Lindhagen, E. (2010). The FMCA-GM assays, high throughput non-clonogenic alternatives to CFU-GM in preclinical hematotoxicity testing. Toxi-cology letters 194, 102-7.

Hallbook, H., Barbany, G., Aleskog, A., Bjornberg, A., Larsson, R., Sundstrom, C., and Lindhagen, E. (2005). In vitro activity of imatinib in cells from patients with adult acute lymphoblastic leukemia. Anticancer Drugs 16, 631-4.

Hanahan, D., and Weinberg, R. A. (2000). The hallmarks of cancer. Cell 100, 57-70. Hanahan, D., and Weinberg, R. A. (2011). Hallmarks of cancer: the next generation.

Cell 144, 646-74. Hoeller, D., Hecker, C. M., and Dikic, I. (2006). Ubiquitin and ubiquitin-like pro-

teins in cancer pathogenesis. Nature reviews 6, 776-88. Hoffbrand (2007). Essential Haematology. Blackwell. Holbeck, S. L., Collins, J. M., and Doroshow, J. H. (2010). Analysis of Food and

Drug Administration-approved anticancer agents in the NCI60 panel of human tumor cell lines. Molecular cancer therapeutics 9, 1451-60.

Johnson, J. I., Decker, S., Zaharevitz, D., Rubinstein, L. V., Venditti, J. M., Schepartz, S., Kalyandrug, S., Christian, M., Arbuck, S., Hollingshead, M., and Sausville, E. A. (2001). Relationships between drug activity in NCI preclinical in vitro and in vivo models and early clinical trials. Br J Cancer 84, 1424-31.

Kapuria, V., Peterson, L. F., Fang, D., Bornmann, W. G., Talpaz, M., and Donato, N. J. (2010). Deubiquitinase inhibition by small-molecule WP1130 triggers ag-gresome formation and tumor cell apoptosis. Cancer research 70, 9265-76.

50

Kintzel, P. E. (2001). Anticancer drug-induced kidney disorders. Drug Saf 24, 19-38.

Kondo, M., Wagers, A. J., Manz, M. G., Prohaska, S. S., Scherer, D. C., Beilhack, G. F., Shizuru, J. A., and Weissman, I. L. (2003). Biology of hematopoietic stem cells and progenitors: implications for clinical application. Annu Rev Im-munol 21, 759-806.

Kurtzberg, L. S., Battle, T., Rouleau, C., Bagley, R. G., Agata, N., Yao, M., Schmid, S., Roth, S., Crawford, J., Krumbholz, R., Ewesuedo, R., Yu, X. J., Wang, F., Lavoie, E. J., and Teicher, B. A. (2008). Bone marrow and tumor cell colony-forming units and human tumor xenograft efficacy of noncamptothecin and camptothecin topoisomerase I inhibitors. Molecular cancer therapeutics 7, 3212-22.

Kurtzberg, L. S., Roth, S. D., Bagley, R. G., Rouleau, C., Yao, M., Crawford, J. L., Krumbholz, R. D., Schmid, S. M., and Teicher, B. A. (2009). Bone marrow CFU-GM and human tumor xenograft efficacy of three tubulin binding agents. Cancer chemotherapy and pharmacology 64, 1029-38.

Lamb, J., Crawford, E. D., Peck, D., Modell, J. W., Blat, I. C., Wrobel, M. J., Lerner, J., Brunet, J. P., Subramanian, A., Ross, K. N., Reich, M., Hieronymus, H., Wei, G., Armstrong, S. A., Haggarty, S. J., Clemons, P. A., Wei, R., Carr, S. A., Lander, E. S., and Golub, T. R. (2006). The Connectivity Map: using gene-expression signatures to connect small molecules, genes, and disease. Science (New York, N.Y 313, 1929-35.

Larsson, R., Kristensen, J., Sandberg, C., and Nygren, P. (1992). Laboratory deter-mination of chemotherapeutic drug resistance in tumor cells from patients with leukemia, using a fluorometric microculture cytotoxicity assay (FMCA). Inter-national journal of cancer 50, 177-85.

Lee, W. M. (2003). Drug-induced hepatotoxicity. The New England journal of medicine 349, 474-85.

Leglise, M. C., Darodes de Tailly, P., Vignot, J. L., Le Bot, M. A., Le Roux, A. M., and Riche, C. (1996). A cellular model for drug interactions on hematopoiesis: the use of human umbilical cord blood progenitors as a model for the study of drug-related myelosuppression of normal hematopoiesis. Cell Biol Toxicol 12, 39-53.

Li, W., Choy, D. F., Lam, M. S., Morgan, T., Sullivan, M. E., and Post, J. M. (2003). Use of cultured cells of kidney origin to assess specific cytotoxic effects of nephrotoxins. Toxicol In Vitro 17, 107-13.

Li, W., Lam, M., Choy, D., Birkeland, A., Sullivan, M. E., and Post, J. M. (2006). Human primary renal cells as a model for toxicity assessment of chemo-therapeutic drugs. Toxicol In Vitro 20, 669-76.

Lindhagen, E., Nygren, P., and Larsson, R. (2008). The fluorometric microculture cytotoxicity assay. Nat Protoc 3, 1364-9.

Lindhagen, E., Vig Hjarnaa, P., Friberg, L., Scilla, L., and Larsson, R. (2004). Pharmacodynamic differences between species exemplified by the novel anti-cancer agent CHS 828 Drug developement research Volume 61, (p 218-226).

Martignoni, M., Groothuis, G. M., and de Kanter, R. (2006). Species differences between mouse, rat, dog, monkey and human CYP-mediated drug metabolism, inhibition and induction. Expert opinion on drug metabolism & toxicology 2, 875-94.

Masubuchi, N., May, R. D., and Atsumi, R. (2004). A predictive model of human myelotoxicity using five camptothecin derivatives and the in vitro colony-forming unit granulocyte/macrophage assay. Clin Cancer Res 10, 6722-31.

51

Metcalf, D. (2010). The colony-stimulating factors and cancer. Nature reviews 10, 425-34.

Millar, A. W., and Lynch, K. P. (2003). Rethinking clinical trials for cytostatic drugs. Nature reviews 3, 540-5.

Nagami, K., Kawashima, Y., Kuno, H., Kemi, M., and Matsumoto, H. (2002). In vitro cytotoxicity assay to screen compounds for apoptosis-inducing potential on lymphocytes and neutrophils. J Toxicol Sci 27, 191-203.

Nygren, P. (2001). What is cancer chemotherapy? Acta Oncol 40, 166-74. Olaharski, A. J., Uppal, H., Cooper, M., Platz, S., Zabka, T. S., and Kolaja, K. L.

(2009). In vitro to in vivo concordance of a high throughput assay of bone mar-row toxicity across a diverse set of drug candidates. Toxicology letters 188, 98-103.

Orlowski, R. Z., and Kuhn, D. J. (2008). Proteasome inhibitors in cancer therapy: lessons from the first decade. Clin Cancer Res 14, 1649-57.

Parent-Massin, D., Hymery, N., and Sibiril, Y. (2010). Stem cells in myelotoxicity. Toxicology 267, 112-7.

Pessina, A., Albella, B., Bayo, M., Bueren, J., Brantom, P., Casati, S., Croera, C., Gagliardi, G., Foti, P., Parchment, R., Parent-Massin, D., Schoeters, G., Sibiril, Y., Van Den Heuvel, R., and Gribaldo, L. (2003). Application of the CFU-GM assay to predict acute drug-induced neutropenia: an international blind trial to validate a prediction model for the maximum tolerated dose (MTD) of myelo-suppressive xenobiotics. Toxicol Sci 75, 355-67.

Pessina, A., Albella, B., Bueren, J., Brantom, P., Casati, S., Gribaldo, L., Croera, C., Gagliardi, G., Foti, P., Parchment, R., Parent-Massin, D., Sibiril, Y., and Van Den Heuvel, R. (2001). Prevalidation of a model for predicting acute neutro-penia by colony forming unit granulocyte/macrophage (CFU-GM) assay. Toxi-col In Vitro 15, 729-40.

Pessina, A., Malerba, I., and Gribaldo, L. (2005). Hematotoxicity testing by cell clonogenic assay in drug development and preclinical trials. Curr Pharm Des 11, 1055-65.

Pessina, A., Parent-Massin, D., Albella, B., Van Den Heuvel, R., Casati, S., Croera, C., Malerba, I., Sibiril, Y., Gomez, S., de Smedt, A., and Gribaldo, L. (2009). Application of human CFU-Mk assay to predict potential thrombocytotoxicity of drugs. Toxicol In Vitro 23, 194-200.

Peters, J. M., Cejka, Z., Harris, J. R., Kleinschmidt, J. A., and Baumeister, W. (1993). Structural features of the 26 S proteasome complex. Journal of molecu-lar biology 234, 932-7.

Peterson, J. K., and Houghton, P. J. (2004). Integrating pharmacology and in vivo cancer models in preclinical and clinical drug development. Eur J Cancer 40, 837-44.

Price, P. S., Keenan, R. E., and Swartout, J. C. (2008). Characterizing interspecies uncertainty using data from studies of anti-neoplastic agents in animals and hu-mans. Toxicology and applied pharmacology 233, 64-70.

Ramström, H., ed. (2009). Läkemedelsboken. Apoteket AB, Stockholm, Sweden. Ravid, T., and Hochstrasser, M. (2008). Diversity of degradation signals in the ubiq-

uitin-proteasome system. Nat Rev Mol Cell Biol 9, 679-90. Reems, J. A., Hall, K. M., Gebru, L. H., Taber, G., and Rich, I. N. (2008). Devel-

opment of a novel assay to evaluate the functional potential of umbilical cord blood progenitors. Transfusion 48, 620-8.

Rich, I. N., and Hall, K. M. (2005). Validation and development of a predictive paradigm for hemotoxicology using a multifunctional bioluminescence colony-forming proliferation assay. Toxicol Sci 87, 427-41.

52

Richardson, P. G., Barlogie, B., Berenson, J., Singhal, S., Jagannath, S., Irwin, D., Rajkumar, S. V., Srkalovic, G., Alsina, M., Alexanian, R., Siegel, D., Orlowski, R. Z., Kuter, D., Limentani, S. A., Lee, S., Hideshima, T., Esseltine, D. L., Kauffman, M., Adams, J., Schenkein, D. P., and Anderson, K. C. (2003). A phase 2 study of bortezomib in relapsed, refractory myeloma. The New England journal of medicine 348, 2609-17.

Richardson, P. G., Sonneveld, P., Schuster, M. W., Irwin, D., Stadtmauer, E. A., Facon, T., Harousseau, J. L., Ben-Yehuda, D., Lonial, S., Goldschmidt, H., Reece, D., San-Miguel, J. F., Blade, J., Boccadoro, M., Cavenagh, J., Dalton, W. S., Boral, A. L., Esseltine, D. L., Porter, J. B., Schenkein, D., and Anderson, K. C. (2005). Bortezomib or high-dose dexamethasone for relapsed multiple myeloma. The New England journal of medicine 352, 2487-98.

Rickardson, L., Fryknas, M., Haglund, C., Lovborg, H., Nygren, P., Gustafsson, M. G., Isaksson, A., and Larsson, R. (2006). Screening of an annotated compound library for drug activity in a resistant myeloma cell line. Cancer chemotherapy and pharmacology 58, 749-58.

Rickardson, L., Wickstrom, M., Larsson, R., and Lovborg, H. (2007). Image-based screening for the identification of novel proteasome inhibitors. J Biomol Screen 12, 203-10.

Roda, E., Coccini, T., Acerbi, D., Castoldi, A. F., and Manzo, L. Comparative in vitro and ex-vivo myelotoxicity of aflatoxins B1 and M1 on haematopoietic progenitors (BFU-E, CFU-E, and CFU-GM): species-related susceptibility. Toxicol In Vitro 24, 217-23.

Rooseboom, M., Commandeur, J. N., and Vermeulen, N. P. (2004). Enzyme-catalyzed activation of anticancer prodrugs. Pharmacological reviews 56, 53-102.

Saeed, A. I., Sharov, V., White, J., Li, J., Liang, W., Bhagabati, N., Braisted, J., Klapa, M., Currier, T., Thiagarajan, M., Sturn, A., Snuffin, M., Rezantsev, A., Popov, D., Ryltsov, A., Kostukovich, E., Borisovsky, I., Liu, Z., Vinsavich, A., Trush, V., and Quackenbush, J. (2003). TM4: a free, open-source system for microarray data management and analysis. BioTechniques 34, 374-8.

Sharma, S. V., Haber, D. A., and Settleman, J. (2010). Cell line-based platforms to evaluate the therapeutic efficacy of candidate anticancer agents. Nature reviews 10, 241-53.

Shibata, T., Yamada, T., Kondo, M., Tanahashi, N., Tanaka, K., Nakamura, H., Masutani, H., Yodoi, J., and Uchida, K. (2003). An endogenous electrophile that modulates the regulatory mechanism of protein turnover: inhibitory effects of 15-deoxy-Delta 12,14-prostaglandin J2 on proteasome. Biochemistry 42, 13960-8.

Shoemaker, R. H. (2006). The NCI60 human tumour cell line anticancer drug screen. Nature reviews 6, 813-23.

Tan, D. S., Thomas, G. V., Garrett, M. D., Banerji, U., de Bono, J. S., Kaye, S. B., and Workman, P. (2009). Biomarker-driven early clinical trials in oncology: a paradigm shift in drug development. Cancer J 15, 406-20.

Thompson, J., Stewart, C. F., and Houghton, P. J. (1998). Animal models for study-ing the action of topoisomerase I targeted drugs. Biochimica et biophysica acta 1400, 301-19.

U-CAN (2011). http://www.u-can.uu.se/. Valeri, A., Alonso-Ferrero, M. E., Cerrato, L., Martinez, S., Bueren, J. A., and Al-

bella, B. (2010). Development of an in vitro model for the simultaneous study of the efficacy and hematotoxicity of antileukemic compounds. Toxicology letters 199, 317-22.

53

Weinberg, ed. (2007). The biology of cancer. Garland Science Taylor and Francis Group. LLC, New York.

Wolf, D., and Rumpold, H. (2009). A benefit-risk assessment of imatinib in chronic myeloid leukaemia and gastrointestinal stromal tumours. Drug Saf 32, 1001-15.

Volpe, D. A., and Warren, M. K. (2003). Myeloid clonogenic assays for comparison of the in vitro toxicity of alkylating agents. Toxicol In Vitro 17, 271-7.

Voskoglou-Nomikos, T., Pater, J. L., and Seymour, L. (2003). Clinical predictive value of the in vitro cell line, human xenograft, and mouse allograft preclinical cancer models. Clin Cancer Res 9, 4227-39.

Xu, G. W., Ali, M., Wood, T. E., Wong, D., Maclean, N., Wang, X., Gronda, M., Skrtic, M., Li, X., Hurren, R., Mao, X., Venkatesan, M., Beheshti Zavareh, R., Ketela, T., Reed, J. C., Rose, D., Moffat, J., Batey, R. A., Dhe-Paganon, S., and Schimmer, A. D. (2010). The ubiquitin-activating enzyme E1 as a therapeutic target for the treatment of leukemia and multiple myeloma. Blood 115, 2251-9.

Yang, H., Shi, G., and Dou, Q. P. (2007a). The tumor proteasome is a primary target for the natural anticancer compound Withaferin A isolated from "Indian winter cherry". Molecular pharmacology 71, 426-37.

Yang, Y., Kitagaki, J., Dai, R. M., Tsai, Y. C., Lorick, K. L., Ludwig, R. L., Pierre, S. A., Jensen, J. P., Davydov, I. V., Oberoi, P., Li, C. C., Kenten, J. H., Beutler, J. A., Vousden, K. H., and Weissman, A. M. (2007b). Inhibitors of ubiquitin-activating enzyme (E1), a new class of potential cancer therapeutics. Cancer re-search 67, 9472-81.

Yang, Y., Ludwig, R. L., Jensen, J. P., Pierre, S. A., Medaglia, M. V., Davydov, I. V., Safiran, Y. J., Oberoi, P., Kenten, J. H., Phillips, A. C., Weissman, A. M., and Vousden, K. H. (2005). Small molecule inhibitors of HDM2 ubiquitin ligase activity stabilize and activate p53 in cells. Cancer cell 7, 547-59.

Yap, T. A., Sandhu, S. K., Workman, P., and de Bono, J. S. (2010). Envisioning the future of early anticancer drug development. Nature reviews 10, 514-23.

Documents

Integrating Efficacy and Toxicity in Preclinical …407166/...In recent decades, an understanding of signalling pathways has emerged, uncovering entirely new drug targets, such as