Cancer drug development: for populations or for individuals?

Abstract

Drug approval always involves a degree of extrapolation. We approve drugs for indicated populations based on results from a sample of patients who were willing and able to participate in clinical trials. This process of extrapolation becomes more challenging when drug development becomes a global process. Drug approval in the United States and many other countries is often based, at least in part, on evidence from foreign studies. Multinational studies commonly support cancer drug approval in the United States, and questions frequently arise whether evidence from one region of the world will support use in a different region. The use of foreign data has been addressed in a guidance document from the International Conference on Harmonization—Ethnic Factors in the Acceptability of Foreign Clinical Data (ICH E5). This document acknowledges potential differences between the uses of medicines in different regions of the world. These ethnic differences are categorized as intrinsic (based on genetic or physiologic differences) or extrinsic (based on cultural or environmental characteristics). ICH E5 assumes that the entire drug development program need not be repeated in each region of the world. When the regulatory authority or drug sponsor is concerned that differences in ethnic factors could alter the efficacy or safety of the medicine in the new region, ICH E5 describes the generation of “bridging” data that allow extrapolation to a new region. Drugs of particular concern would include drugs cleared by enzymes showing genetic polymorphism and with steep dose-response curves. Such medicines are termed “ethnically sensitive.” On the other hand, often conditions exist where bridging data are not needed, and the medicine is said to be “ethnically insensitive,” when medical practice and trial conduct in the regions are generally similar. In this issue of the Journal of Clinical Oncology, Shirao et al report results of a bridging study. They describe a comparison of uracil/tegafur (UFT) plus oral leucovorin (LV) in Japanese and American patients with advanced colorectal cancer. The purpose of this comparison was to determine whether the results of phase III studies performed in the West could be extrapolated to Japanese patients. The authors assumed that if the study “showed equality of efficacy, safety, and pharmacokinetics in both Japanese patients and American patients, then the results of the Western phase III trials could be extrapolated to Japanese patients.” Presumably, regulatory authorities in Japan will determine whether concerns exist regarding ethnic differences between the Japanese and American study populations, and whether bridging data satisfy those concerns. The UFT studies described by Shirao et al are two identical phase II and pharmacokinetic (PK) studies in advanced colon cancer patients, one performed in the United States and one in Japan. Patients received identical 300 mg/m UFT daily doses plus LV. PK parameters (sampled up to 8 hours), response rates, and toxicities were compared. This study design has limitations because it compares two groups sampled from different populations. Even though identical entry criteria were applied, the samples may reflect different patterns of patient selection; therefore, a statistically significant difference between these samples does not necessarily reflect a statistically significant or clinically relevant difference between the Japanese and US cancer populations. The PK results are not definitive. First, the study as originally designed had only limited statistical power to evaluate similarity of PK parameters between the Japanese and American patients. Second, even for this limited goal, only a subgroup of patients received comparable doses for the PK study. PK parameters are best compared in the 18 American patients and 26 Japanese patients with similar body surface area (1.5 to 1.83 m). No formal test for similarity between these groups is provided, but the authors note that the Japanese patients’ data are distributed within the range of the American patients’ data. The striking findJOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 22 NUMBER 17 SEPTEMBER 1 2004