Bridging from Preclinical to Clinical Studies for Tyrosine Kinase Inhibitors Based on Pharmacokinetics/Pharmacodynamics and Toxicokinetics/Toxicodynamics

Drug Metab. Pharmacokinet. 26 (6): 612620 (2011). Copyright © 2011 by the Japanese Society for the Study of Xenobiotics (JSSX) Regular Article Bridging from Preclinical to Clinical Studies for Tyrosine Kinase Inhibitors Based on Pharmacokinetics/Pharmacodynamics and Toxicokinetics/Toxicodynamics

Azusa HOSHINO-YOSHINO1,2,MotohiroKATO2,KohnosukeNAKANO2, Masaki ISHIGAI2,ToshiyukiKUDO1 and Kiyomi ITO1,* 1Research Institute of Pharmaceutical Sciences, Musashino University, Tokyo, Japan 2Pre-clinical Research Department, Chugai Pharmaceutical Co. Ltd., Kanagawa, Japan

Full text of this paper is available at http://www.jstage.jst.go.jp/browse/dmpk

Summary: The purpose of this study was to provide a pharmacokinetics/pharmacodynamics and toxicokinetics/toxicodynamics bridging of kinase inhibitors by identifying the relationship between their clinical and preclinical (rat, dog, and monkey) data on exposure and efﬁcacy/toxicity. For the eight kinase inhibitors approved in Japan (imatinib, geﬁtinib, erlotinib, sorafenib, sunitinib, nilotinib, dasatinib, and lapatinib), the

human unbound area under the concentration-time curve at steady state (AUCss,u) at the clinical dose correlated well with animal AUCss,u at the no-observed-adverse-effect level (NOAEL) or maximum tolerated dose (MTD). The best correlation was observed for rat AUCss,u at the MTD (p < 0.001). Emax model analysis was performed using the efficacy of each drug in xenograft mice, and the efficacy at the human AUC of the clinical dose was evaluated. The predicted efficacy at the human AUC of the clinical dose varied from far

below Emax to around Emax even in the tumor for which use of the drugs had been accepted. These results suggest that rat AUCss,u attheMTD,butnottheefﬁcacy in xenograft mice, may be a useful parameter to estimate the human clinical dose of kinase inhibitors, which seems to be currently determined by toxicity rather than efﬁcacy.

Keywords: kinase inhibitors; PK/PD; TK/TD; prediction; human clinical dose; bridging

determine the human first dose of anticancer drugs from Introduction ¥ animal studies.2 However, it does not distinguish between In recent years, the focus of anticancer drug development conventional cytotoxic agents and molecular-targeted drugs. has moved from cytotoxic agents to molecular-targeted For conventional drugs, the maximum tolerated dose ¤MTD¥ drugs that are designed to target specific molecules which is generally administered as the clinical dose to get the are mutated or overexpressed in tumor cells.1¥ Molecular- maximum efficacy, based on the idea that efficacy is targeted drugs were thought to have broader therapeutic proportional to exposure.3¥ As for predicting human MTD windows than conventional cytotoxic agents and to exhibit in conventional drugs, Fuse et al. reported an excellent less toxicity because of their high selectivity to tumor. correlation between the log area under the concentration- ¤ ¥ ¤ ¥ There are two types of molecular-targeted drugs: tyrosine time curve AUC at the dose lethal for 10% of mice LD10 kinase inhibitors as small molecular drugs and monoclonal and log AUC at MTD for humans for AUC-dependent antibodies as large molecular drugs. Currently, eight oral drugs, but not for time-dependent drugs.4¥ However, the tyrosine kinase inhibitors ¤imatinib, gefitinib, erlotinib, relationship between human and animal AUC for molecular- sorafenib, sunitinib, nilotinib, dasatinib, and lapatinib¥ have targeted drugs has never been reported. The purpose of this been approved in Japan. study was to provide a PK/PD and TK/TD bridging of A guideline published by the International Conference on kinase inhibitors by identifying the relationship between the Harmonization of Technical Requirements for Registration clinical and preclinical data for their AUC and efficacy/ of Pharmaceuticals for Human Use ¤ICH¥ describes how to toxicity.

Received; May 18, 2011, Accepted; August 24, 2011 J-STAGE Advance Published Date: September 6, 2011, doi:10.2133/dmpk.DMPK-11-RG-043 *To whom correspondence should be addressed: Kiyomi ITO, Ph.D., Research Institute of Pharmaceutical Sciences, Musashino University, 1-1-20 Shinmachi, Nishitokyo, Tokyo 202-8585, Japan. Tel/Fax. +81-42-468-9199, E-mail: [email protected]

612 Preclinical to Clinical PK/PD Bridging for Kinase Inhibitors 613

Table 1. Tyrosine kinase inhibitors approved in Japan

Approval in Japan Drugs ¤ ¥ MTD Clinical dose Dosage form year/month. Trade name Generic name 2001.11 Glivec Imatinib h1000 mg, b.i.d. 400 mg, q.d. Tablet 2002.7 Iressa Geﬁtinib 700 mg, q.d. 250 mg, q.d. Tablet 2007.10 Tarceva Erlotinib 150 mg, q.d. 150 mg, q.d. Tablet 2008.1 Nexavar Sorafenib 600 mg, b.i.d. 400 mg, b.i.d. Tablet 2008.4 Sutent Sunitinib 50 mg, q.d. 50 mg, q.d. Capsule 400 mg, b.i.d. ¤for chronic-phase and accelerated-phase of chronic myelogenous leukemia¥, 2009.1 Tasigna Nilotinib 600 mg, b.i.d. Capsule 300 mg, b.i.d. ¤for newly diagnosed chronic-phase myelogenous leukemia¥ 100 mg, q.d. ¤for chronic phase¥, 2009.1 Sprycel Dasatinib h120 mg b.i.d. Tablet 70 mg, b.i.d. ¤for accelerated phase and blast phase¥ 2009.4 Tykerb Lapatinib 1800 mg, q.d. 1250 mg, q.d. Tablet These data were obtained from interview forms for each drug.5®12¥ MTD, maximum tolerated dose; q.d., once-daily regimen; b.i.d., twice-daily regimen.

Relationship between human AUCss,u at the Methods in vitro clinical dose and IC50 values for target Data collection: For the eight tyrosine kinase kinase inhibition: In vitro IC50 values for target kinase inhibitors approved in Japan ¤Table 1¥, pharmacokinetics, inhibition were obtained from the inhibition studies efficacy, and toxicity data were obtained from application described in the interview form5,12¥ and the litera- documents, interview forms, and the literature. ture.22®26¥ The smallest values were adopted among those fi Relationship between human AUCss,u at the using a variety of puri ed kinases. Then the correlation clinical dose and animal AUCss,u at the NOAEL or between the in vitro IC50 values and human AUCss,u at the ¤ ¥ MTD: AUCss AUC per dose at steady state values at the clinical dose obtained above was evaluated. no-observed-adverse-effect level ¤NOAEL¥ or MTD in Relationship between efficacy in animals and 2-week toxicity studies using rats, dogs, and monkeys were humans: Tumor growth inhibition ¤TGI, %¥ was obtained from the pharmacokinetic analysis described in the calculated using the following equation: ® ¥ application documents.13 20 AUC data from 1-month toxicity ð Þ ð Þ¼ Tumor growth Treated animals studies were adopted if 2-week toxicity studies had not been TGI % 1 ð Þ conducted. Male data were adopted when available. All Tumor growth Control animals ¤ ¥ animal AUCss data were from studies with a once-daily q.d. 100 ð2Þ regimen. For humans, AUCss at the clinical dose was obtained from the application documents13®20¥ describing repeated For all drugs except nilotinib, tumor growth was fi administration studies. Human AUCss data were from studies evaluated at the nal evaluating point of tumor volume with a q.d. regimen except for sorafenib, nilotinib, and obtained from efficacy studies with tumor-bearing mice. As dasatinib, which were given according to a twice-daily for nilotinib, tumor growth was evaluated by measuring the ¤ ¥ fi b.i.d. regimen. For ge tinib and lapatinib, AUCss values luciferase activity in an acute model, in which mice were were calculated by extrapolation from pharmacokinetic injected with murine 32D cells harboring the firefly studies at other doses under linear assumption. Values for luciferase gene and transfected to be dependent upon p210 the unbound fraction ¤fu¥ in plasma or serum were obtained Bcr-Abl. These efficacy data were obtained from the from the application documents13®15,17®20¥ and the literature21¥ efficacy studies described in the application documents13®20¥ 5®12¥ and were multiplied by AUCss to calculate the AUCss,u. The and interview form. Pharmacodynamic analysis using ¤ correlation was evaluated using average fold-error afe, an Emax model was performed to assess the AUC-effect Eq. 1¥ and correlation coefficient ¤r¥: ¤TGI¥ relationship and the E and EðAUC were P max u,50 1 Animal AUCss,u at the NOAEL or MTD calculated based on the following equation using a computer N log afe ¼ 10 Human AUCss,u at the clinical dose ð1Þ program for the non-linear least squares regression method ¤MULTI;27¥¥

Table 2. The data of human AUCss at the clinical dose and animal AUCss at the no-observed-adverse-effect level (NOAEL) and the maximum tolerated dose (MTD)

¤ & ¥ ¤ & ¥ Human AUCss Animal AUCss at the NOAEL ng h/mL Animal AUCss at the MTD ng h/mL Drugs at the clinical dose ¤ng&h/mL¥ Rat Dog Monkey Rat Dog Monkey Imatinib 33200 ¯b ¯b 4150 2 Weeks 4930 2 Weeks 445000 2 Weeks ¯b ¯b 17800 2 Weeks Geﬁtinib 7251.5 4611 1 Month 9700 1 Month ¯b ¯b 22753 1 Month 9700 1 Month ¯b ¯b Erlotinib 42679 4159 1 Month 2682 1 Month ¯b ¯b ¯b ¯b 14334 1 Month ¯b ¯b Sorafenib 36690a ¯b ¯b ¯b ¯b ¯b ¯b 67000 1 Month 52000 1 Month ¯b ¯b Sunitinib 1406 1317 2 Weeks ¯b ¯b 1276 2 Weeks 10190 2 Weeks ¯b ¯b 13174 2 Weeks Nilotinib 19000a 46100c 1 Month 4680 2 Weeks 12040 1 Month 224000 2 Weeks 20370 2 Weeks 12040 1 Month Dasatinib 398.8a 35 2 Weeks ¯b ¯b 181 1 Month 899 2 Weeks ¯b ¯b 634 2 Weeks Lapatinib 33836.5 ¯b ¯b ¯b ¯b ¯b ¯b ¯b ¯b ¯b ¯b ¯b ¯b The AUC per dose at steady state were obtained from application documents for each drug.13®20¥ Human data were from studies with a q.d. regimen except for sorafenib, nilotinib, and dasatinib, which were given according to a b.i.d. regimen. The AUCss values at 400 mg b.i.d. and 70 mg b.i.d. were used for nilotinib and dasatinib, ﬁ respectively. The AUCss of ge tinib and lapatinib were calculated by extrapolation from pharmacokinetic studies at other doses under linear assumption. Animal AUC data were male data after repeated administration in 2-week or 1-month toxicity studies with a q.d. regimen. ab.i.d. regimen. bNot available. cFemale.

Table 3. The unbound fractions in plasma or serum (fu) used ð Þ¼ Emax AUCu ð Þ in the analysis TGI % þ 3 E AUCu;50 AUCu Unbound fraction in plasma or serum ¤fu¥ fi ¤ ¥ where Emax represents maximum ef cacy TGI in mice, ð Drugs Human Mouse Rat Dog Monkey E AUCu,50 represents AUCu at 50% of Emax in mice, and Imatinib 0.074a,b 0.024a 0.055a 0.19a 0.101a AUCu represents the AUC values at the dose of evaluating fi ¯d TGI, extrapolated from single-dose pharmacokinetic studies Ge tinib 0.089 0.06 0.125 0.081 in mice under linear assumption, multiplied by fu. There- Erlotinib 0.084 0.055 0.085 0.149 ¯d Sorafenib 0.005 0.004 0.005 0.009 ¯d after, predicted TGI in xenograft mice at the human AUCss,u of the clinical dose was calculated using Eq. 3. Sunitinib 0.047c 0.059c 0.022c 0.05c 0.051c Nilotinib 0.016 0.026 0.009 0.018 0.01 Results Dasatinib 0.0365 0.082 0.026 0.042 0.031 Relationship between human AUCss,u at the Lapatinib 0.011 0.011 0.011 0.011 ¯d clinical dose and animal AUCss,u at the NOAEL or All data were obtained from application documents13®15,17®20¥ except for imatinib. ¥ MTD: The data for AUCss and fu are shown in Tables 2 aFrom Kretz et al.21 and 3, respectively. Figure 1 shows the relationship bAverage of values at 300®500, 5000, 12000, and 26000 ng/mL. cAverage of values at 0.25, 1, and 10 µM. between human AUCss,u at the clinical dose and animal d fi Not available. AUCss,u at the NOAEL or MTD. As shown in the gures, human AUCss,u at the clinical dose correlated with AUCss,u at the NOAEL or MTD of rats, dogs, and monkeys in each monkey h rat h dog for the correlation between human case. For most drugs, the human AUCss,u at the clinical dose AUCss,u at the clinical dose and animal AUCss,u at the h h was larger than the animal AUCss,u at the NOAEL, whereas NOAEL and rat dog monkey for the correlation it was smaller than that at the MTD. The parameters for the between human AUCss,u at the clinical dose and animal correlation between human AUCss,u at the clinical dose and AUCss,u at the MTD. The correlation was statistically fi ¤ g ¥ animal AUCss,u at the NOAEL or MTD are summarized in signi cant p 0.05 when AUCss,u at the NOAEL of Table 4. The afe values and Figure 1 suggest that human monkeys or that at the MTD of rats and dogs were used. AUCss,u values at the clinical dose are approximately 2.4- to Relationship between human AUCss,u at the in vitro 3.1-fold higher than those at animal NOAEL, and 1.2- to clinical dose and IC50 values for target 3.7-fold lower than those at animal MTD. The afe values kinase inhibition: Table 5 shows in vitro IC50 values were in the order rat h dog h monkey for the correlation for target kinase inhibition and their relationship to human between human AUCss,u at the clinical dose and animal AUCss,u at the clinical dose is shown in Figure 2. The h h AUCss,u at the NOAEL, and rat monkey dog for the human AUCss,u at the clinical dose correlated with in vitro correlation between human AUCss,u at the clinical dose and IC50 values for target kinase inhibition, except for erlotinib ¤ ¥ animal AUCss,u at the MTD. The r values were in the order Fig. 2 .

a Rat Dog Monkey 100000 100000 100000 Erlotinib Erlotinib Imatinib 10000 10000 10000 Nilotinib Nilotinib 1000 Sunitinib 1000 1000 Imatinib Sunitinib at the clinical

Dasatinib at the clinical Dasatinib 100 Gefitinib 100 Gefitinib at the clinical 100 ss,u

Nilotinib ss,u 10 10 ss,u 10 dose (ng*h/mL) dose (ng*h/mL)

1 1 dose (ng*h/mL) 1

0.1 0.1 0.1 Human AUC Human AUC 0.1 10 1000 100000 0.1 10 1000 100000 Human AUC 0.1 10 1000 100000

Rat AUCss,u at the NOAEL (ng*h/mL) Dog AUCss,u at the NOAEL (ng*h/mL) Monkey AUCss,u at the NOAEL (ng*h/mL)

b Rat Dog Monkey 100000 100000 100000 Erlotinib 10000 10000 10000 Imatinib

1000 1000 Nilotinib 1000 Nilotinib

Imatinib at the clinical at the clinical at the clinical Gefitinib ss,u ss,u Gefitinib 100 Nilotinib 100 ss,u 100 Sorafenib Sorafenib dose (ng*h/mL) dose (ng*h/mL) Sunitinib Sunitinib dose (ng*h/mL) 10 10 10 Dasatinib Dasatinib 1 Human AUC Human AUC 1 1 Human AUC 1 100 10000 1 100 10000 1 100 10000 Rat AUC at the MTD (ng*h/mL) Dog AUC at the MTD (ng*h/mL) ss,u ss,u Monkey AUCss,u at the MTD (ng*h/mL)

Fig. 1. (a) Relationship between human AUCss,u at the clinical dose and animal AUCss,u at the NOAEL. (b) Relationship between human AUCss,u at the clinical dose and animal AUCss,u at the MTD The solid lines represent a 1:1 correspondence and the dotted lines represent a variance of 0.33–3.

Table 4. Statistical data for the correlation between human Table 5. In vitro IC50 values for target kinase inhibition AUCss,u at the clinical dose and animal AUCss,u at the NOAEL ¤ ¥ or MTD Drugs Target kinase IC50 ng/mL Imatinib Bcr-Abl 12.3a AUCss,u at the NOAEL AUCss,u at the MTD Geﬁtinib EGFR 12.1b Rat Dog Monkey Rat Dog Monkey Erlotinib EGFR 0.787c n 544 644 Sorafenib C-Raf 2.79d afe 3.14 3.08 2.44 3.73 1.22 1.39 Sunitinib VEGFR-1 ¤FLT1¥ 0.797e Animal AUC/Human 0.571 0.476 0.488 4.513 1.393 3.148 Nilotinib ¯ f ¯ f AUCa g r 0.875 0.610 0.959 0.986 0.959 0.778 Dasatinib FYN 0.0976 h p g0.1 h0.1 g0.05 g0.001 g0.05 h0.1 Lapatinib HER2 6.02 P Bcr-Abl, breakpoint cluster region-Abelson leukemia; EGFR, epidermal growth 1 Animal AUCss,u at the NOAEL or MTD aAnimal AUC/Human AUC © factor receptor; VEGFR, vascular endothelial growth factor receptor; FLT, FMS- N Human AUC at the clinical dose ss,u like tyrosine kinase; HER, human epidermal growth factor receptor. aFrom Druker et al.22¥ b 25¥ 10000 From Wakeling et al. cFrom Moyer et al.23¥ dFrom Wilhelm et al.26¥ e 12¥ 1000 From interview form of Sunitinib. fNot available. gFrom interview form of Dasatinib.5¥ hFrom Rusnak et al.24¥ 100 at the clinical dose (ng*h/mL) ss,u 10 Relationship between efﬁcacy in animals and humans: The examples of Emax model analysis are shown 1 in Figure 3. In the case of imatinib against chronic Human AUC 0.01 1 100 10000 myelogenous leukemia ¤CML¥, the predicted TGI at the IC values for target kinase inhibition (ng/mL) 50 human AUCss,u of the clinical dose was 107%, which is near the E . However, in the case of sorafenib against renal cell Fig. 2. Relationship between human AUCss,u at the clinical dose max and in vitro IC50 values for target kinase inhibition carcinoma, the predicted TGI at the human AUCss,u of the The dotted line represents the regression line excluding erlotinib. clinical dose was 22.8%, which is far below the Emax. The

a b c 120 120 100 100 100 80 Predicted TGI 107% Predicted TGI 96.4% 80 80 60 60

60 TGI (%) Predicted TGI 48.7% TGI (%) 40 40

TGI (%) 40 20 20 20

0 0 0 0 1000 2000 3000 0 2000 4000 6000 8000 0 2000 4000 6000 8000

-20 AUCu (ng*h/mL) AUCu (ng*h/mL) AUCu (ng*h/mL)

d e f 120 120 120

100 100 100 Predicted inhibition 97.4% 80 80 80

60 60 TGI (%) 60 TGI (%) Predicted TGI 49.3% 40 40 40

20 Predicted TGI 22.8% 20 20 Photon emission inhibition (%) 0 0 0 0 1000 2000 3000 4000 5000 0 200 400 600 800 1000 0 500 1000 1500

AUCu (ng*h/mL) AUCu (ng*h/mL) AUCu (ng*h/mL)

g 120 h 140 Predicted TGI 118% 100 120

80 100

80 60 TGI (%)

TGI (%) 60 40 40

20 20

0 0 0 50 100 150 200 0 200 400 600 AUCu (ng*h/mL) AUCu (ng*h/mL)

Fig. 3. Examples of Emax model analysis The values of tumor growth inhibition (TGI) at various doses were obtained from the results of efﬁcacy studies in mice. The AUC values at each dose were calculated by extrapolation from single-dose pharmacokinetic studies in mice under linear assumption, and multiplied by fu to obtain

AUCu.Emax model analysis was performed to estimate Emax and E_AUCu,50 and then the TGIs in xenograft mice at the human AUCss,u of the clinical dose were predicted. (a) Imatinib against chronic myelogenous leukemia (CML), (b) Geﬁtinib against non-small-cell lung cancer (NSCLC), (c) Erlotinib against vulval cancer, (d) Sorafenib against renal cell carcinoma (RCC), (e) Sunitinib against squamous cell carcinoma, (f) Nilotinib against mouse CML, (g) Dasatinib against CML, (h) Lapatinib against NSCLC. Diamonds, observed values; solid curves, regression lines; dotted lines, human AUCss,u at the clinical dose.

results of Emax model analysis for TGI are summarized in clinical dose at the early drug discovery phase to raise Table 6 and the ratios of the predicted TGI in humans to the success rate and to shorten the period of the clinical Emax in mice are shown in Figure 4. The ratios varied from phase study. 18.7% to 96.4%, even in tumors for which the use of the According to the guideline for nonclinical evaluation of drugs had been accepted. anticancer pharmaceuticals published by ICH, it is recommended that a common approach for many small molecules Discussion is to set the start dose at either 1/10 the severely toxic dose The success rate of drug candidates from ﬁrst-in-man to in 10% of the animals ¤STD 10¥ in rodents or 1/6 the registration in the oncology area is reported to be about highest non-severely toxic dose ¤HNSTD¥ in non-rodents ¤if 5%,28¥ which is lower than that for other therapeutic the non-rodent is the most appropriate species¥.2¥ The areas. It is very helpful to be able to estimate the human rationale behind the guideline is that the AUC at LD10 in

Table 6. Summary of the results of Emax model analysis for tumor growth inhibition

EðAUC Predicted TGI Ratios to E Drugs Tumor Cell line E ¤%¥ SD u,50 SD max max ¤ng&h/mL¥ ¤%¥ ¤%¥ Imatinib CMLa p210bcr-abl 111 33 102 85 107 96.3 NSCLCa A549 82.7 13.5 450 120 48.7 58.9 Geﬁtinib Vulval cancer A431 138 35 1244 421 47.2 34.2 Prostate cancer Du145 99.7 29.1 870 421 42.5 42.6 Vulval cancer A431 115 17 699 217 96.4 83.7 Erlotinib Head and neck cancer HN5 121 37.7 1192 1068 91.0 75.0 RCCa 786-O 122 36 800 816 22.8 18.7 HCCa PLC/PRF/5 149 15 591 171 35.2 23.7 Sorafenib Mouse RCCa RENCA 110 11.2 705 170 22.8 20.6 Melanoma LOX IMVI 1080 2530 45200 88700 4.4 0.4 NSCLC NCI-H460 209 59 5530 2100 6.7 3.2 Squamous cell carcinoma A431 113 4 85 12 49.3 43.8 Rat glioblastoma C6 116 15 206 60 28.3 24.3 Sunitinib Rectal cancer COLO205 122 15 124 57 42.6 34.8 SCLC NCI-H526 82.7 12.3 45 41 49.4 59.7 RCCa 786-O 100 ¯¯ ¯ ¯ ¯ Nilotinib Mouse CMLa 32D 101 0 11.4 0.4 97.4 96.4 Dasatinib CMLa K562 100 ¯¯ ¯ ¯ ¯ Lapatinib NSCLC NCI-H322 150 7 100 14 118 78.8

These data were obtained by calculation according to Eq. 3. The SD values shown in this table were obtained from Emax model analysis by the damping Gauss-Newton method using MULTI. ﬁ ¤ ¥ ð Emax, maximum ef cacy TGI in mice; E AUCu,50, AUCu at 50% of Emax in mice; TGI, tumor growth inhibition; CML, chronic myelogenous leukemia; NSCLC, non-small- cell lung cancer; RCC, renal cell carcinoma; HCC, hepatocellular carcinoma; SCLC, small-cell lung cancer. aTumor for which use of the drug is accepted.

mice correlates with the AUC at MTD in humans for some to estimate the human AUCss,u at the clinical dose from the conventional cytotoxic agents4¥ and that MTD has been results of short-term toxicity studies in rats. As the success generally used as the clinical recommended dose, based on of drug development depends in large part on clinical the idea that efficacy is proportional to exposure.3¥ On studies, estimating the optimum clinical dose is important fi the other hand, for the eight kinase inhibitors investigated, for achieving an ef cient clinical study. The animal AUCss,u the relationship between efficacy and exposure in clinical values used in this study are those at the NOAEL or MTD studies18,29®33¥ or preclinical studies5,6,8,22,34¥ has been from 2-week or 1-month toxicity studies. It is an important fi reported; however, no reports have been published about nding of this study that the animal AUCss,u obtained from the relationship between efficacy or toxicity in clinical the early phase of drug discovery correlated with human studies and preclinical studies. AUCss,u at the clinical dose. Meanwhile, an investigation of The present study compared the human AUCss,u at the the application documents of the eight kinase inhibitors clinical dose and animal AUCss,u at the NOAEL or MTD. As studied suggested that the clinical dose was determined by shown in Figure 1, human AUCss,u at the clinical dose was toxicity in most cases. For example, dose limiting toxicity ¤ ¥ larger than AUCss,u at the NOAEL, but smaller than that at DLT or grade 3 toxicity appeared at the clinical dose for the MTD in rats, dogs, and monkeys for most drugs; four kinase inhibitors ¤erlotinib, nilotinib, gefitinib, and however, there was some correlation. The afe values suggest dasatinib¥ and at a dose only one step higher than the clinical that the animal AUCss,u was closest to human AUCss,u at the dose for the other four. Taking these data together, it is clinical dose for NOAEL in monkeys and MTD in dogs, reasonable to consider that AUCss,u at the clinical dose fl although the species difference was small. The largest r value re ects the AUCss,u at the toxicity dose. It is also suggested with statistical significance ¤p g 0.001¥ was obtained for the from Figure 1 that the inhibitory activity of these drugs MTD in rats. These findings suggest that it may be possible against toxicity-related enzymes in humans correlates with

100

80 in mice (%)

max 70

0 Ratios of the predicted TGI in human to E Ratios of the predicted

Fig. 4. Ratios of the predicted TGI at the human AUCss,u of the clinical dose to Emax in mice Solid columns are data for cell lines derived from the carcinoma for which the drugs are accepted for use. TGI values at the human AUCss,u of the clinical dose were predicted as shown in Figure 3. HCC, hepatocellular carcinoma; SCLC, small cell lung cancer. that in animals, although such inhibition data for animal not necessarily been determined to show the maximum enzymes are unavailable. efficacy against each carcinoma. In other words, as far as the As shown in Figure 2, a correlation was observed kinase inhibitors so far approved are concerned, it is possible between human AUCss,u at the clinical dose and in vitro IC50 that their dose could not be further increased because of values for target kinase inhibition except for erlotinib. their toxicity. In fact, it was confirmed by this study that the Although the reason for the poor correlation for erlotinib clinical doses of recently developed kinase inhibitors were in is unclear, it might be possible that its clinical dose has been most cases determined by MTD or DLT, the same approach set too high for its pharmacological potential, as discussed as for cytotoxic agents ¤Table 1¥. Also, it has become clear later. that molecular-targeted drugs exhibit various toxicities such To find another approach with regard to the efficacy, as nausea,30¥ myalgias,30¥ fluid retention,35¥ rash31,33,36,37¥ fi 33,36®39¥ 31,39¥ 35¥ prediction of ef cacy at the clinical AUCss,u was attempted diarrhea, fatigue, hemorrhage, and hand-foot fi 38¥ by an Emax model analysis using ef cacy data from an in vivo syndrome at clinical doses. It is reasonable, therefore, that fi xenograft model. As shown in Figures 3 and 4, the AUCss,u the method for determining the human rst dose is not at the clinical dose was close to that showing maximum differentiated between conventional cytotoxic agents and efficacy in mice for some drugs such as imatinib, but was molecular-targeted drugs in the ICH guideline. According to considerably smaller than that showing Emax for others such our results, it is possible that the clinical doses of some as sorafenib. The ratios of predicted TGI in humans to Emax kinase inhibitors have been set unnecessarily high because in mice varied from 18.7% to 96.4% even in the tumors for toxicity was not apparent, which could explain the poor which the drugs had been accepted for use. The reasons for correlation for erlotinib shown in Figure 2. It would be this variability in predicted response might include the interesting to perform similar PK/PD analyses on future possibility that the antitumor activity of a drug is associated drugs, which are expected to show sufficient efficacy at a with the inhibition of multiple kinases, among which the dose far below the toxic dose, to investigate the possibility of target enzymes, which showed a good correlation in estimating the clinical dose from the efficacy in xenografted Figure 2, are related to toxicity rather than efficacy. The animals. low ratio for sorafenib against renal cell carcinoma might be In this study, we showed that AUCss,u at the NOAEL or related to its minimum response ¤MR¥ in clinical studies, MTD in short-term toxicity studies may be a useful compared with complete response ¤CR¥ or partial response parameter to estimate the human clinical dose of kinase ¤PR¥ for other drugs such as imatinib, gefitinib, nilotinib, and inhibitors, which seems to be currently determined by lapatinib.7®10¥ These results suggest that the clinical dose has toxicity rather than efficacy.

References 23¥ Moyer, J. D., Barbacci, E. G., Iwata, K. K., Arnold, L., Boman, B., Cunningham, A., DiOrio, C., Doty, J., Morin, M. J., Moyer, 1¥ Gibbs, J. B.: Mechanism-based target identification and drug M. P., Neveu, M., Pollack, V. A., Pustilnik, L. R., Reynolds, discovery in cancer research. Science, 287: 1969®1973 ¤2000¥. M. M., Sloan, D., Theleman, A. and Miller, P.: Induction of 2¥ ICH. Harmonized Tripartite Guideline: nonclinical evaluation for apoptosis and cell cycle arrest by CP-358,774, an inhibitor of anticancer pharmaceuticals S9 current step 4 version dated 29 epidermal growth factor receptor tyrosine kinase. Cancer Res., 57: October 2009. http://www.ich.org/LOB/media/MEDIA5785. 4838®4848 ¤1997¥. pdf. Accessed 25 June 2010. 24¥ Rusnak, D. W., Lackey, K., Affleck, K., Wood, E. R., Alligood, 3¥ Evans, W. E. and Relling, M. V.: Clinical pharmacokinetics- K. J., Rhodes, N., Keith, B. R., Murray, D. M., Knight, W. B., pharmacodynamics of anticancer drugs. Clin. Pharmacokinet., 16: Mullin, R. J. and Gilmer, T. M.: The effects of the novel, 327®336 ¤1989¥. reversible epidermal growth factor receptor/ErbB-2 tyrosine kinase 4¥ Fuse, E., Kobayashi, S., Inaba, M., Suzuki, H. and Sugiyama, Y.: inhibitor, GW2016, on the growth of human normal and tumor- Application of pharmacokinetically guided dose escalation with derived cell lines in vitro and in vivo. Mol. Cancer Ther., 1:85®94 respect to cell cycle phase specificity. J. Natl. Cancer Inst., 86: 989® ¤2001¥. 996 ¤1994¥. 25¥ Wakeling, A. E., Guy, S. P., Woodburn, J. R., Ashton, S. E., 5¥ Interview form of Dasatinib. http://www.sprycel.jp/pdf/ Curry, B. J., Barker, A. J. and Gibson, K. H.: ZD1839 ¤Iressa¥:an spðif101001.pdf. Accessed 28 June 2010. orally active inhibitor of epidermal growth factor signaling with 6¥ Interview form of Erlotinib. http://chugai-pharm.jp/hc/ss/pr/ potential for cancer therapy. Cancer Res., 62: 5749®5754 ¤2002¥. onco/drug/tarðfil0150/if/index.html. Accessed 28 June 2010. 26¥ Wilhelm, S. M., Carter, C., Tang, L., Wilkie, D., McNabola, A., 7¥ Interview form of Gefitinib. http://med.astrazeneca.co.jp/ Rong, H., Chen, C., Zhang, X., Vincent, P., McHugh, M., Cao, product/IF/IREðIF.pdf. Accessed 28 June 2010. Y., Shujath, J., Gawlak, S., Eveleigh, D., Rowley, B., Liu, L., 8¥ Interview form of Imatinib. http://www.novartis.co.jp/product/ Adnane, L., Lynch, M., Auclair, D., Taylor, I., Gedrich, R., gli/if/ifðgli0909.pdf. Accessed 28 June 2010. Voznesensky, A., Riedl, B., Post, L. E., Bollag, G. and Trail, 9¥ Interview form of Lapatinib. http://glaxosmithkline.co.jp/ P. A.: BAY 43-9006 exhibits broad spectrum oral antitumor clubðGSK/if/tykerb.html. Accessed 28 June 2010. activity and targets the RAF/MEK/ERK pathway and receptor 10¥ Interview form of Nilotinib. http://www.novartis.co.jp/product/ tyrosine kinases involved in tumor progression and angiogenesis. tas/if/ifðtas1002.pdf. Accessed 28 June 2010. Cancer Res., 64: 7099®7109 ¤2004¥. 11¥ Interview form of Sorafenib. http://www.bayer-hv.jp/hv/ 27¥ Yamaoka, K., Tanigawara, Y., Nakagawa, T. and Uno, T.: products/if/nex.pdf. Accessed 28 June 2010. A pharmacokinetic analysis program ¤multi¥ for microcomputer. 12¥ Interview form of Sunitinib. https://pfizerpro.jp/documents/if/ J. Pharmacobiodyn., 4: 879®885 ¤1981¥. sut/sut01ðif.htm. Accessed 28 June 2010. 28¥ Kola, I. and Landis, J.: Can the pharmaceutical industry reduce 13¥ Pharmaceuticals and Medical Devices Agency ¤PMDA¥ Drug label attrition rates? Nat. Rev. Drug Discov., 3: 711®715 ¤2004¥. Dasatinib approved on 01/2009. http://www.info.pmda.go.jp/ 29¥ Druker, B. J., Sawyers, C. L., Kantarjian, H., Resta, D. J., Reese, approvalSrch/ApprovalSrch. Accessed 30 June 2010. S. F., Ford, J. M., Capdeville, R. and Talpaz, M.: Activity of a 14¥ Pharmaceuticals and Medical Devices Agency ¤PMDA¥ Drug label specific inhibitor of the BCR-ABL tyrosine kinase in the blast crisis Erlotinib approved on 10/2007. http://www.info.pmda.go.jp/ of chronic myeloid leukemia and acute lymphoblastic leukemia with approvalSrch/ApprovalSrch. Accessed 30 June 2010. the Philadelphia chromosome. N. Engl. J. Med., 344: 1038®1042 15¥ Pharmaceuticals and Medical Devices Agency ¤PMDA¥ Drug label ¤2001¥. Gefitinib approved on 07/2002. http://www.info.pmda.go.jp/ 30¥ Druker, B. J., Talpaz, M., Resta, D. J., Peng, B., Buchdunger, approvalSrch/ApprovalSrch. Accessed 30 June 2010. E., Ford, J. M., Lydon, N. B., Kantarjian, H., Capdeville, R., 16¥ Pharmaceuticals and Medical Devices Agency ¤PMDA¥ Drug label Ohno-Jones, S. and Sawyers, C. L.: Efficacy and safety of a specific Imatinib approved on 11/2001. http://www.info.pmda.go.jp/ inhibitor of the BCR-ABL tyrosine kinase in chronic myeloid approvalSrch/ApprovalSrch. Accessed 30 June 2010. leukemia. N. Engl. J. Med., 344: 1031®1037 ¤2001¥. 17¥ Pharmaceuticals and Medical Devices Agency ¤PMDA¥ Drug label 31¥ Kantarjian, H., Giles, F., Wunderle, L., Bhalla, K., OöBrien, S., Lapatinib approved on 04/2009. http://www.info.pmda.go.jp/ Wassmann, B., Tanaka, C., Manley, P., Rae, P., Mietlowski, W., approvalSrch/ApprovalSrch. Accessed 30 June 2010. Bochinski, K., Hochhaus, A., Griffin, J. D., Hoelzer, D., Albitar, 18¥ Pharmaceuticals and Medical Devices Agency ¤PMDA¥ Drug label M., Dugan, M., Cortes, J., Alland, L. and Ottmann, O. G.: Nilotinib approved on 01/2009. http://www.info.pmda.go.jp/ Nilotinib in imatinib-resistant CML and Philadelphia chromosome- approvalSrch/ApprovalSrch. Accessed 30 June 2010. positive ALL. N. Engl. J. Med., 354: 2542®2551 ¤2006¥. 19¥ Pharmaceuticals and Medical Devices Agency ¤PMDA¥ Drug label 32¥ Peng, B., Lloyd, P. and Schran, H.: Clinical pharmacokinetics of Sorafenib approved on 01/2008. http://www.info.pmda.go.jp/ imatinib. Clin. Pharmacokinet., 44: 879®894 ¤2005¥. approvalSrch/ApprovalSrch. Accessed 30 June 2010. 33¥ Ranson, M., Hammond, L. A., Ferry, D., Kris, M., Tullo, A., 20¥ Pharmaceuticals and Medical Devices Agency ¤PMDA¥ Drug label Murray, P. I., Miller, V., Averbuch, S., Ochs, J., Morris, C., Sunitinib approved on 04/2008. http://www.info.pmda.go.jp/ Feyereislova, A., Swaisland, H. and Rowinsky, E. K.: ZD1839, a approvalSrch/ApprovalSrch. Accessed 30 June 2010. selective oral epidermal growth factor receptor-tyrosine kinase 21¥ Kretz, O., Weiss, H. M., Schumacher, M. M. and Gross, G.: inhibitor, is well tolerated and active in patients with solid, In vitro blood distribution and plasma protein binding of the malignant tumors: results of a phase I trial. J. Clin. Oncol., 20: tyrosine kinase inhibitor imatinib and its active metabolite, 2240®2250 ¤2002¥. CGP74588, in rat, mouse, dog, monkey, healthy humans and 34¥ Kamath, A. V., Wang, J., Lee, F. Y. and Marathe, P. H.: patients with acute lymphatic leukaemia. Br. J. Clin. Pharmacol., 58: Preclinical pharmacokinetics and in vitro metabolism of dasatinib 212®216 ¤2004¥. ¤BMS-354825¥: a potent oral multi-targeted kinase inhibitor against 22¥ Druker, B. J., Tamura, S., Buchdunger, E., Ohno, S., Segal, SRC and BCR-ABL. Cancer Chemother. Pharmacol., 61: 365®376 G. M., Fanning, S., Zimmermann, J. and Lydon, N. B.: Effects ¤2008¥. of a selective inhibitor of the Abl tyrosine kinase on the growth of 35¥ Keam, S. J.: Dasatinib: in chronic myeloid leukemia and Bcr-Abl positive cells. Nat. Med., 2: 561®566 ¤1996¥. Philadelphia chromosome-positive acute lymphoblastic leukemia.

BioDrugs, 22:59®69 ¤2008¥. cancer. N. Engl. J. Med., 353: 123®132 ¤2005¥. 36¥ Escudier, B., Eisen, T., Stadler, W. M., Szczylik, C., Oudard, S., 38¥ Geyer, C. E., Forster, J., Lindquist, D., Chan, S., Romieu, C. G., Siebels, M., Negrier, S., Chevreau, C., Solska, E., Desai, A. A., Pienkowski, T., Jagiello-Gruszfeld, A., Crown, J., Chan, A., Rolland, F., Demkow, T., Hutson, T. E., Gore, M., Freeman, S., Kaufman, B., Skarlos, D., Campone, M., Davidson, N., Berger, Schwartz, B., Shan, M., Simantov, R. and Bukowski, R. M.: M., Oliva, C., Rubin, S. D., Stein, S. and Cameron, D.: Lapatinib Sorafenib in advanced clear-cell renal-cell carcinoma. N. Engl. J. plus capecitabine for HER2-positive advanced breast cancer. N. Med., 356: 125®134 ¤2007¥. Engl. J. Med., 355: 2733®2743 ¤2006¥. 37¥ Shepherd, F. A., Rodrigues Pereira, J., Ciuleanu, T., Tan, E. H., 39¥ Motzer, R. J., Hutson, T. E., Tomczak, P., Michaelson, M. D., Hirsh, V., Thongprasert, S., Campos, D., Maoleekoonpiroj, S., Bukowski, R. M., Rixe, O., Oudard, S., Negrier, S., Szczylik, C., Smylie, M., Martins, R., van Kooten, M., Dediu, M., Findlay, B., Kim, S. T., Chen, I., Bycott, P. W., Baum, C. M. and Figlin, Tu, D., Johnston, D., Bezjak, A., Clark, G., Santabarbara, P. and R. A.: Sunitinib versus interferon alfa in metastatic renal-cell Seymour, L.: Erlotinib in previously treated non-small-cell lung carcinoma. N. Engl. J. Med., 356: 115®124 ¤2007¥.