The pharmacokinetics (PK) and pharmacodynamics (PD) of most drugs can vary widely even if the same doses are administered. Many factors influence this variability.1-3 Among these clinical pharmacoethnic differences arising from functional variants in implicated genes have been reported for in 5-fluorouracil,4-6 doxorubicin,7 cyclophosphamide,8 vincristine9 and irinotecan hydrochloride.10,11 Pharmacogenomic difference in metabolizing enzymes and transporters can influence the toxicities of anticancer drugs differently across ethnic groups. Single nucleotide polymorphisms (SNPs) in UDP glucuronosyl transferase 1A1 and dihydropyrimidine dehydrogenase, which are germ line variations, have been shown to be associated with the hematological toxicities of irinotecan hydrochlolide and fluorinated pyrimidines, respectively.4,11 On the other hand, antitumor activity is mainly influenced by somatic changes which reflect the pharmacogenomics of target molecules in tumors. Even prior to epidermal growth factor receptor tyrosine-kinase inhibitors (EGFR-TKIs) development (Fig. 1), ethnic differences were reported in regard to survival outcomes between Asians and Caucasians with advanced non-small cell lung cancer (NSCLC) treated with systemic chemotherapy.12 Similarly, treatment response to EGFR-TKIs which are active against driver mutation positive NSCLC exhibited clear ethnic differences in Caucasians and East Asians due to the difference in EGFR mutation rates.13 In trials of antiEGFR and anti-vascular endothelial growth factor antibodies against lung and stomach cancer, respectively, as well as sorafenib against liver cancer, pharmacoethnic differences induced by differences in environmental and local medical practices have also been noted in addition to differences in pharmacogenomics (Table 1). Accordingly, recent phase I study on tivantinib, a MET-TKI inhibitor, showed the presence of CYP2C19 polymorphisms in 30% of Japanese patients, which was 10 times higher than that in Caucasians. In the present review, the development of protocols for global trials and interpretation of the results of such trials will be discussed from the perspective of pharmacoethnicity.

In previous common arm trials cooperatively conducted together with SWOG, differences in the effectiveness and toxicities of paclitaxel and carboplatin in NSCLC14 and of etoposide or irinotecan plus cisplatin in small cell lung cancer (SCLC) were analyzed.15 Prior to the initiation of the study, whether studies in one country could be directly extrapolated to patients within another country was unclear. The purpose of the trial was to demonstrate similarities and differences in the patient characteristics and outcomes in Japan and the United States under the same treatment regimen. The trials also sought to provide a basis for the standardization of future study design, for which to better conduct future trials in both countries.

For NSCLC, three trials were performed. Two of the studies were performed in Japan [Four Arm Cooperative Study (FACS), n=14516 and JMTO LC-00-03 study, n=19717], and one study was performed in the USA (SWOG S0003 study, n=186).18 All three studies used identical eligibility criteria, staging criteria and treatment protocols (Table 2). Paclitaxel and carboplatin were considered to be a common arm. The dose of carboplatin administered in all of the studies was an area under the curve (AUC) 6 ug/mL×min, and while the administered dose of paclitaxel was 225 mg/m² in the JMTO and SWOG trials and 200 mg/m² in the FACS trial. This regimen was administered every 3-4 weeks. The patient characteristics such as age, sex, proportion with stage IIIB/IV, and histology were evenly distributed in all three studies. Treatment compliance was also the same in all three studies. The response rates were 32%, 31% and 37% in the FACS, LC00-03 and S0003 studies, respectively. The median survival and the one year survival rates were higher in Japanese patients (Table 3). In addition the frequency of hematological toxicities such as grade IV neutropenia and febrile neutropenia were significantly higher among the Japanese patients (Table 4). Genotype profiles in Japanese and American patients showed differences in frequencies of variants in CYP3A4, CYP3A5, CYP2C8 and ERCC2. Among them CYP3A4 and ERCC2 were correlated with progression free survival (PFS) and treatment response, respectively. These results suggest that global clinical trials, especially those conducted internationally, should be carefully designed and conducted to account for potential genetic differences in the respective patient population.

For SCLC, the JCOG951119 and SWOG-021420 trials were conducted. Both studies compared etoposide/cisplatin vs. irinotecan/cisplatin. The dose and administration schedules of the drugs were exactly the same. The patient demographics showed that 86% and 57% of the patients in the JCOG 9511 and SWOG-0124 trials were male, respectively, and the performance status (PS) of the patients was better in the SWOG trial. Treatment delivery and dose intensities were similar for both trials. The response rates to both regimens were significantly higher in the JCOG9511 trial. The overall survival and one year survival rates were similar between the two trials for the cisplatin/etoposide regimen, however, these outcomes were significantly better for the cisplatin/irinotecan regimen in the JCOG 9511 trial (overall survival (OS): 12.8 vs. 6.8 months, one-year survival: 58% vs. 41%). On the other hand, the frequencies of hematological toxicities (>grade 3) such as neutropenia, leukopenia and anemia were significantly higher for both regimens in the JCOG 9511 trial than those in the SWOG-0124 trial. These results suggested the need for 1) consideration of different patient characteristics and outcomes amongst patients receiving identical therapy. 2) The utilization of a common arm model in prospective trials and 3) the inclusion of pharmacogenomics correlates in cancer trials where pharmacoethnic differences in drug disposition are expected.

The molecular target of EGFR-TKI was thought to be wild type EGFR during the initial development of EGFR-TKIs. During a phase I trial conducted in Japan, 5 patients with heavily treated NSCLC experienced a partial response21 despite numerous reports of a lack of a response in other Phase 1 studies except for Ranson's report.22-24 Phase II studies of gefitinib such as the Ideal 1 and 2 trials reported response rates for Japanese and Non-Japanese patients of 30 and 10%, respectively.25,26 The response rate of another EGFR-TKI, erlotinib, was also about 30% in Japanese population.27 Various clinical trials have also suggested that EGFR-TKIs are active in non-smoking East-Asian, female with adenocarcinoma whereas the BR-21 study demonstrated the completely opposite results (Fig. 2).28,29 The majority of clinical trials against unselected populations produced negative results. In first line NSCLC chemotherapy studies such as the Intact-1/-2, Talent and Tribute trials which included more than 1000 patients each, the addition of gefitinib or erlotinib did not improve overall survival.30-33 In second line trials performed in unselected patients, the adenocarISEL trial using gefitinib produced negative results34 while the BR-21 trial using erlotinib produced positive results.29 If the ISEL population was subdivided into Asians and Caucasians, gefitinib was shown to only be effective against Asian patients (Fig. 3).34 On the other hands in the BR-21 study, erlotinib was more effective against squamous carcinoma, Caucasians, and male smokers29 although recent trials examining erlotinib have reported conflicting results. (Fig. 4). EGFR mutations were first discovered in 2004, and somatic mutations in tyrosine kinase (TK) domain of EGFR were shown to be strongly associated with responses to treatment with EGFR-TKIs.35,36 Moreover the mutation rate of EGFR in patients with NSCLC was shown to be remarkably different between East Asians (35-40%) and Caucasians (<10%) (Fig. 5).37,38 Seven phase II studies on the use of EGFR-TKIs against patients with EGFR mutation demonstrated high response rates ranging from 63% to 91% across Asia. The Iressa Pan-Aisa Study (IPASS) trial, which compared gefitinib vs. carboplatin/paclitaxel in clinically selected patients, included patients from 9 Asian countries including China, Japan, Hong Kong, the Philippines, Singapore, Taiwan and Thailand, accruing 1217 patients within 18 months.39 Patient selection in the IPASS trial was, however, performed based solely on clinical factors such as adenocarcinoma and never/light smokers in addition to an East Asian ethnicity. The overall data was difficult to explain because the PFS curves crossed at 6 months. If the patients were subdivided according to the EGFR mutational status, the PFS of the gefitinib was better in the mutation positive patients, while that of the carboplatin/paclitaxel group was better in the mutation negative patients. In their study, however, EGFR mutation status was only analyzed in 36% of the patients. Substantiating their findings, the Korean First Signal Study obtained similar results. At the same time two randomized controlled trials (RCTs) using gefitinib to treat EGFR mutation positive patients were conducted by two groups in Japan, North East Japan (NEJ) 00240 and West Japan Thoracic Oncology Group (WJTOG) 3405.41 Later two RCTs using erlotinib were conducted in China (OPTIMAL)42 and Europe (EURTAC).43 All four of these trials produced similar results (Table 5), suggesting that the biological character of EGFR mutated NSCLC was similar in East Asians and Caucasians despite differences in mutation frequencies. Presently, clinical trials of next generation EGFR-TKIs are on-going. The majority of patients accrued to those trials are East Asian patients, since all of these trials required pharmacogenomic selection; for example 70% of patients in the LUX-LUNG3 global trial for afatinib are East Asian (Fig. 6).44 Fig. 7 shows the ethnic differences in the molecular classification of adenocarcinoma of the lung based on data from the National Cancer Center Research Institute of Japan and the Memorial Sloan Kettering Cancer Center-in the United States (Fig. 7). Careful selection is becoming mandatory to ensure that clinical trials are completed very quickly and yield scientifically sound data.

It is worth noting that driver mutations are likely to become the most important and promising molecular target for future anticancer drugs.45 Accordingly, possible ethnic difference in driver mutations must be kept in mind when performing preclinical and early clinical trials of molecular targeted drugs.

Interstitial lung disease (ILD) affects the lung parenchyma and alveoli of the lungs. Drug-induced ILD presents with acute diffuse alveolar damage and sometimes involves a fatal outcomes. Unfortunately no specific treatment for ILD exists. Palliative therapies with oxygen inhalation and corticosteroid administration have been attempted, but some patients require assisted ventilation and die. ILD can reportedly be induced by cytotoxic chemotherapy and radiotherapy, the prevalence of which is higher among Japanese than among Caucasians.46 EGFR-TKI-induced ILD is considered to be relatively rare47 and appears to be limited to patients in Japan.48 A WJTOG retrospective study of 1976 patients reported the frequencies of ILD and ILD-induced mortality by gefitinib as 3.2 and 1.3% respectively. The risk factors thereof were a male gender, a history of smoking and a history of pulmonary fibrosis.28 AstraZeneca performed a prospective study of 3322 patients receiving gefitinib. ILD and ILD-induced death were observed in 5.8 and 2.5% of these patients, respectively. The reported risk factors were a history of smoking, a poor PS, a history of pulmonary fibrosis and prior chemotherapy. In a nested case controlled study performed by AstrasZeneca, the overall odds ratio for ILD due to gefitinib treatment, compared with chemotherapy, was 3.2.49 This specific trial showed that, among the patients who died as a result of ILD, 31.6% had received gefitinib and 27.9% undergone chemotherapy, respectively. Based on the AstraZeneca global Drug Safety Database, the reported rate of ILD-type events in patients receiving gefitinib was only 0.23% throughout the rest of the world, excluding Japan, according to the records of more than 215000 patients treated with gefitinib. Even in other East Asian countries such as Korea and Taiwan the rate was only 0.17%. On the other hand the percentages of patients involving a fatal outcome after developing ILD were similar, 37% in Japan and 31% in other countries. Within the case-controlled study performed by AstraZeneca, analyses were conducted to identify genetic and proteomic predictors for ILD, unfortunately however, no specific biomarkers capable of predicting ILD have been identified.

Moreover, the prevalence of erlotinib-induced ILD has not been intensively analyzed. Limited data suggest a similar prevalence (4/108: 3.7%) to that of other drugs.50-52

Ethnic diversity is recognized as an important factor accounting for inter-individual variations in anticancer drug responsiveness and toxicity. However, similar doses of anticancer drugs have been prescribed to different ethnic populations without consideration of the potential differences in pharmacokinetics and pharmacodynamics, both of which are essentially influenced by pharmacogenomics. Although pharmacoethnicity is determined by genetic and non-genetic factors, the latter factors have not yet been well identified. The possible determinants of ethnicity could include 1) environmental factors that influence bioavailability and metabolism, such as the frequency of smoking, alcohol drinking, herbal medicine use and local dietary varieties, 2) local medical care preferences, 3) ethnic specific drug-drug interactions influenced by drug lag in a specific region, 4) variability of genetic polymorphisms in drug metabolizing enzymes and transporters, and 5) the prevalence of an ethnically restricted mutations in a drug's receptor/target that may cause a particular sensitivity or resistance to that drug.1

Studies on pharmacoethnicity face many challenging issues. However, preclinical and clinical studies should be carefully designed to account for these problems. 1) Clinical trials on pharmacoethnicity require diverse populations. International collaboration, including global trials, is mandatory and repeated trials using the same treatment strategies should be conducted in multiple countries. 2) Chemotherapy-related effects are likely to be under multi-gene control. Unbiased genome-wide models are needed and such models should include multiple mechanism-related and metabolism-related pathways. 3) Potentially important polymorphisms or mutations are generally uncommon. The sample size of clinical trials should be sufficient to enable an appropriate statistical power. Clinical trials in ethnic populations with a specific phenotype of interest may sometimes be necessary. To discover rare variants, new generation of sequencing methodology should be introduced. 4) The detection of genome-wide associations requires multiple SNP testing, which may generate several false-positive SNPs. Accordingly, test and validation sets are mandatory. In addition, both of preclinical and clinical validations, as well as the application of a rigorous statistical methodology is needed. Furthermore, as 5) anti-cancer drugs cannot be administered to healthy volunteers, in vitro cell based models are needed1 and pharmacogenomical selection of patients is recommended by appropriate biomarkers.

The clinical trials included in the present review were large enough to identify ethnic differences, despite the numerous hidden factors that remain unknown. As shown by common arm trials between USA and Japan, Asian patients experience more frequent and profound neutropenia despite receiving the same treatment doses, schedules and same pharmacokinetics. A difference in sensitivity at receptor site has been suggested to be the main cause of these observations, but a concrete mechanism, thereof has not been clarified. On the other hand, the survival period of NSCLC patients was significantly longer among Asian patients than among Caucasians even before EGFR-TKIs became available. In a common arm trial, the OS and one-year survival were both significantly better in Asian patients although response rates were exactly the same. Differences in sensitivity to EGFR-TKI have been clearly explained by EGFR mutation in both Asians and Caucasians. However, why the frequency of EGFR mutation is higher among Asian patients remains unknown. Some germ line and environmental factors may influence this mutation rate. However, all previous genome-wide and proteome analyses that have been performed in association with prospective clinical trials failed to pick up on relevant factors. There is also no clear explanation as to why EGFR-TKI-induced ILD is observed so frequently in Japanese patients only. To date, only clinical characteristics indicating a susceptibility to ILD have been identified. The data of three trials the FLEX, AVAGAST, and Sorafenib after TACE trials, suggested that disease status and local medical care preference strongly influenced the OS of the included patients.

Recently the Shizuoka Cancer Center reported interesting results on Phase I study of ARQ197, a selective, non-ATP competitive inhibitor of c-MET, a receptor tyrosine kinase involved in tumor migration, invasion and proliferation (Fig. 9). Therein, the ratios of poor metabolizers (PM), who exhibited a single nucleotide polymorphism in CYP2C9, a major metabolizing enzyme for ARQ197, among Caucasians and Asians have been reported to be 3 and 20%, respectively (Table 11). Recommended phase II dose of ARQ197 for subjects of western countries has been decided to be a single dose of 360 mg bid. On the other hand, the study in Japan demonstrated that CYP2C19 genotype clearly affected exposures such as AUC and C-max of ARQ, which led to the designation of two different recommended doses for phase II trials, 360 mg bid for extensive metabolizer patients and 240 mg bid for PM patients (Fig. 10). Clear ethnic differences mandates the necessity of different protocols for phase II studies of Asians and Caucasians.58

In summary basic pharmacogenomic differences could only be identified from the prospective analyses of patients of a homogeneous background. The clarification of pharmacoethnic differences will be crucially important to the future development of new anticancer drugs.