OR WAIT 15 SECS
Planning for ethnic sensitivity from a scientific and regulatory perspective.
In the three-layer approach proposed in this article, how to interpret data in multi-regional trials into three layers is explored: Layer 1 to first look at overall study results, Layer 2 to identify factors which influence overall results from scientific and regulatory perspectives, and Layer 3 to discuss benefit/risk for a specific country/region.
The International Conference on Harmonization's (ICH) ''E5 Guideline on Ethnic Factors in the Acceptability of Foreign Clinical Data'' (ICH-E5) (R1) was introduced in March 1998.1 Since the ICH-E5 guideline was issued, there has been a lot of discussion on the importance of evaluating the impact of ethnic factors on the ability to extrapolate efficacy and safety data to a new region in order to minimize duplication of clinical trials. Moreover, in recent years, a shift in clinical trials to so-called emerging regions, especially in Eastern European, Latin American, and Asian countries has been noted. This contrasts with traditional clinical trials which have been carried out in relatively affluent countries in North America, Western Europe, and Oceania.2 In other words, in recent years, drug development has become dramatically globalized, and multi-regional trials (MRTs) are now being extensively conducted in both ICH and non-ICH regions.
In these MRTs, it is import to consider carefully potential regional heterogeneity of treatment effect which may be caused by intrinsic or extrinsic ethnic factors. According to ICH-E5,1 "extrinsic ethnic factors are factors associated with the environment and culture in which a person resides," and "intrinsic ethnic factors are factors that help to define and identify a sub-population and may influence the ability to extrapolate clinical data between regions." A number of MRTs in the past, such as Metoprolol CR/XL Randomized Intervention Trial in Heart Failure (MERIT-HF)3 and Platelet Inhibition and Patient Outcomes (PLATO),4 have suggested the possibility of real regional differences in drug effects. For example, both MERIT-HF and PLATO showed an overall statistically significant efficacy. However, the results in North America, and in particular in the United States, showed a trend in the opposite direction. If a regional heterogeneity is found, the cause of the differences should be thoroughly investigated and its findings should be described in the common technical document (CTD) at the time of the NDA.
Historically, in a CTD that is first written in a Western country, a description of results of MRTs focused on the overall study results with a limited amount of investigation of study results from various perspectives, including the effect of intrinsic and extrinsic ethnic factors. Therefore, at the time of NDA submission in countries outside of the West, such as in Japan or other Asian countries, additional analyses comparing the results of Western countries with that of Japan or other Asian countries have often been necessary to satisfy requests from the regulatory authorities in these countries.
The ultimate goal of a globalized drug development that includes MRTs is to provide a "common" technical document that is common not only in the format, but also in its contents, and therefore can be used to support mutual submissions in multiple countries.
With the above information as a backdrop, a new approach to presenting data from MRTs in scientific and effective manner in the "common" technical document is proposed. This article will explore bridging studies with historical background; the three-layer approach with examples of MRTs; and a new proposed approach.
In Japan, data from clinical trials conducted outside of Japan have been accepted in accordance with the ICH-E5 guideline, which proposes that if the health authority in a new region (such as Japan) is presented with a foreign clinical data package that meets the local regulatory requirements, it should ask for only the additional data necessary to evaluate the ability to extrapolate the foreign data to the population of the new region. The additional data can be from a "bridging study," which is often a controlled clinical trial (which could possibly be a dose-response study), carried out in the new region to assess the investigational product's sensitivity to ethnic factors. Typically, a bridging study following completion of Phase II or Phase III studies in the United States or European Union is used to extrapolate the foreign data to construct the full clinical data package in Japan.5 In the bridging strategy, at least one statistically powered study is conducted in both Japan and the foreign countries. Obviously, as each study is sizable, it is possible to compare overall and subgroup analysis results from bridging and the counterpart studies and to evaluate the similarity between the two.
Recently, countries which used the bridging strategy have been shifting to the global simultaneous drug development by designing or participating in MRTs. How to evaluate data from MRTs is different from that in the bridging strategy which consists of independent studies in each country. Therefore, a new approach to evaluating data of MRTs is proposed.
Traditionally, multinational trials had been conducted in limited locations such as in North America, Western Europe, and Oceania as a natural extension of multi-center trials. ICH-E5 Question and Answer No.11 introduced MRTs in 2006, foreseeing the current paradigm of global simultaneous development.6 A shift from multinational trials to MRTs implied an expansion of participating countries in emerging regions to MRT, especially in Asian and Eastern European countries, whose selection is motivated primarily by the ability to reduce operational costs while recruiting a large number of patients.2 Increasing the number of regions in clinical trials results in ethnic and cultural diversity of patients, which in turn leads to a potential for increased variability in treatment response among patients. Therefore, in comparison with traditional multinational trials, data of other regions in MRTs may cause additional issues in terms of ethnic sensitivity/insensitivity.
A successful MRT requires scientifically valid results from a population with diverse ethnic backgrounds. To assess ethnic sensitivity or insensitivity, the design of an MRT needs to be considered from various perspectives such as medical, statistical, regulatory, commercial, operational, and marketing, encompassing the entire development strategy. Moreover, to generate the data necessary to achieve the trial objectives, the sample size in each country and region should be fully taken into consideration.
In addition, before conducting an MRT, intrinsic and extrinsic ethnic factors that may affect efficacy and safety of drugs should be assessed and then incorporated into the study design and the analysis plan of MRTs.
As stated above, how to design and evaluate an MRT is critical to its success.
Current approach to investigation of clinical data in multi-regional trials. As shown in Figure 1, in MRTs the overall results of efficacy and safety are investigated as the first step. As the second step, the current approach tends to give only a cursory examination of data and bypass detailed investigations which may include subgroup analysis from various points of view such as baseline prognostic factors in evaluation of ethnic insensitivity, and proceeds directly to a step where results from subgroup analyses in each country/region are conducted to assess the benefits/risks profile of a drug, when the new drug application is filed in a specific country.7 It is natural for regulatory authorities to be interested in the differences among regions as well as the data for their own people—they have accountability for the benefit/risk profile of the drug which they approve in their respective countries. However, from a scientific point of view, subgroup analyses by demographic and baseline characteristics for overall data should be conducted at the same level as those by country/region factor in the second step to further investigate characteristics of data. Therefore, the term "three-layer approach" is proposed.
When interpreting results from MRTs, it is natural that one should look at the overall results, and then investigate the subgroup results to identify factor(s) which might influence the interpretation of the overall results.
Ethnic insensitivity has been evaluated in a similar manner with the bridging strategy as shown in Figure 1. However, it is noted that one should not rush into subgroup analysis within each country/region. Figure 2 schematize the proposed three-layer approach for interpreting the study results from MRTs.
Looking at overall results (Layer 1). In the first step, one looks at the overall results of a study without considering potential differences among subgroups.
Identifying factors that may influence the overall results (Layer 2). Subgroup analyses are typically performed to investigate whether a specific subgroup behaves differently from other subgroups or the overall population, or whether any specific factors exert an influence on the overall results. Such factors may include gender, age, body weight, race, baseline characteristics, and other intrinsic ethnic factors. We refer to the perceptual approach to these factors as a scientific perspective. In contrast, looking at a country or geographical region is a regulatory perspective rather than a scientific one. Each regulatory authority has an accountability to explain the benefit-risk profile of the drug to its public. Therefore, all regulatory authorities are interested in the results from their respective country or region. We note that extrinsic ethnic factors may be confounding the regulatory perspective, and that extrinsic ethnic factors are often not collected in the case report forms and cannot be incorporated into data analyses.
Observed inconsistencies in responses among the regions/countries, if any, will be investigated from a scientific perspective by examining the distribution of ethnic factors. For example, if a genetic polymorphism in a metabolic enzyme is the most influential factor on the efficacy/safety endpoint, and if the distribution of the polymorphism is different among countries/regions, the observed difference in responses among countries/regions may be attributable to the unequal distribution of the polymorphism.
As an example of how to investigate data in Layer 2, four patterns of interpretation for results are shown in Figure 3: First row. There is no factor that affects the overall results from either scientific or regulatory perspectives. In this case the overall results would be very robust.
Second row. There is no factor from regulatory perspectives, but there are some from scientific perspectives. This means intrinsic ethnic factors such as age or polymorphism exist that would have common effect in any countries/regions, but there is no difference among countries/regions in terms of the outcome.
Third row. There is no factor from scientific perspectives, but there is one from the regulatory perspectives. In this case, the difference in outcome among countries/regions might be caused by those factors that cannot be explained with scientific perspectives, such as extrinsic ethnic factors or operational factors. It is likely that causes of the differences are difficult to pinpoint and the results should be interpreted with caution. When such differences are found, we should consider whether they occurred by chance based on statistical calculation with the number of patients for each subgroup, variance, effect size, and so on to evaluate risk and benefit in a given country/region. Causes of the differences might also be difficult to show only based on the trial data. Especially, smaller sample size of the country in question would make such interpretations difficult. Other information or knowledge that is internal or external to the trial including administrative information may help the interpretation of the study results, e.g., possible causes of higher placebo response in some countries or issues with the quality of data or safety reporting in some countries. These explanations may not always be sufficient for some regulatory authorities and their acceptability would be the subject of determination in the Layer 3 discussion with each regulatory authority.
Fourth row. There are such factors from both scientific and regulatory perspectives. For example, if a baseline value has an effect on the efficacy outcome, the difference among countries/regions could be explained to some extent by the difference in distributions of the baseline values (i.e., scientific perspective), but not all. The cause of any residual country/region differences may be difficult to explain conclusively, as described in the third case above.
In order to identify factors that may explain subgroup differences, further analyses could also be conducted in Layer 2. In these analyses, the important way of thinking is that the more clues from various assessments of efficacy and safety pointing the same direction, the greater the likelihood the finding is real and not a false signal. Meanwhile, if one uses hypothetical testing in Layer 2, he/she may consider multiplicity issues not to be deluded by the false signals.
Because Layer 2 mostly has no pre-specified hypothesis and may have insufficient power to detect any inconsistency, it is important to integrate any available information such as findings from other studies or knowledge of other drugs in the same drug class and to not rely solely on the study results.
Discussing the benefit/risk for a specific country or region (Layer 3). When the new drug application is filed in a specific country/region, benefit-risk balance for that country/region would be explained based on the discussion in Layer 2. No additional extensive discussion on a specific country/region subgroup versus overall population would be needed if analyses in Layer 2 can cover the discussion. If they cannot cover the discussion, in Layer 3, subgroup analyses by baseline prognostic factors in each country or region (such as Japan) would be conducted to explain benefit-risk balance to the regulatory authorities.
Subgroup analyses. Analyses which are conducted in Layer 2 and Layer 3 could be pre-specified in the statistical analysis plans if some specific factors are known in advance to have influence on the overall results. However, in most cases, subgroup analyses in Layer 2 and Layer 3 would be determined by what is found in the initial planned analyses for Layer 2 and Layer 3; post-hoc analyses would be performed after results from analyses in Layer 2 and Layer 3 are provided.
The three-layer approach emphasizes the importance of analyses in Layer 2. Discussions in Layer 1 and Layer 2 are not limited to a specific country/region and all countries/regions should be included in the discussion. Such a discussion could be shared in the CTD, which any countries/regions could use when writing the CTD for the countries/regions.
Recently, more and more clinical trials have been conducted globally. The new approach, the three-layer approach, can be used to analyze and interpret data of MRTs under this new paradigm of drug development. In the three-layer approach, how to interpret data in MRTs into three layers is categorized: Layer 1 to first look through overall study results, Layer 2 to identify factors which influence overall results from scientific and regulatory perspectives, and Layer 3 to discuss benefits/risks for a specific country or region according to the requirements of its regulatory authorities. Discussions in Layer 1 and Layer 2 is not limited to a specific country/region. The importance of the discussions in Layer 2 to be utilized in the new drug application in any country included in the MRTs is emphasized. Such discussions could be used commonly by any country/region, resulting in a more scientific and structured presentation that is more effective in providing evidence to be used commonly and globally. The three-layer approach is also useful from a regulatory point of view in that Layer 3 allows benefits/risks of a drug to be assessed in each country/region. Although subgroup analyses in each country/region may have limitations due to smaller number of subjects in each subgroup, regulatory authorities have often requested subgroup analyses of their own country/region. In the three-layer approach, whether or not there are factors which affect benefits/risks of a drug and to what extent these factors affect results can be firstly evaluated in Layer 2 and then might proceed to Layer 3 to assess the benefits/risks of the drug in each country/region. The important point of the three-layer approach is that no additional extensive discussion on a specific country/region subgroup versus overall population would be needed if analyses in Layer 2 can cover the discussion.
Moreover, in Layer 2, to investigate factors which would affect efficacy and safety, it would be useful to conduct subgroup analyses. In the case that there are such baseline factors which affect efficacy and safety, in Layer 3, assessing similarity of distribution in such baseline factors among regions or countries could be supportive to explain similarity/difference of regions or countries. However, the limitations of subgroup analysis by baseline factors for specific countries/regions should be taken into consideration.8
The proposed three-layer approach may be applied not only to one MRT but also to a set of clinical trials, whose components are conducted in different countries or regions. In this case, as one example, Layer 1 (overall results) could use results from the integrated summary of clinical efficacy and Layer 2 (consistency assessments from various angles) could be evaluated based on the integrated dataset.
In summary, when the three-layer approach is widely taken into consideration, data from MRTs could be used effectively for global drug development from the perspective of planning development strategies and preparing the CTD. In order to create a CTD following the three-layer approach, sponsors in pharmaceutical industries should prepare a plan to create the CTD which includes globally-shared Layer 2. Especially, it is critical to consider clinical development strategy early in the product development, create a study design, and conduct the study while taking into consideration the evaluation of data by the three-layer approach.
Authors' Note: The authors would like to thank Yoichi Ii, Christy Chuang-Stein, Zhen Luo, Norisuke Kawai, Peng Qu, and Satoru Mogami for their help with this project.
Osamu Komiyama, Shintaro Hiro, Naoki Isogawa, Shigeyuki Toyoizumi, Nobushige Matsuoka, Satoshi Hashigaki, Tamotsu Yoshiyama, and Nami Maruyama*are all Clinical Statisticians at Pfizer Japan Inc., Shinjuku Bunka Quint Building, Shibuya-Ku, Tokyo, Japan, e-mail: [email protected].
*To whom all correspondence should be addressed.
(1) International Conference on Harmonization, "Ethnic Factors in the Acceptability of Foreign Clinical Data E5 (R1)," (1998), http://bit.ly/Lckcof.
(2) F. A. Thiers, et al., "Trends in the Globalization of Clinical Trials," Nature Reviews Drug Discovery, 7, 13-14 (2008).
(3) C. Anello et al., "Multicentre Trials: A US Regulatory Perspective," Statistical Methods in Medical Research, 14 (3) 303–318 (2005).
(4) L. Wallentin et al., "Ticagrelor Versus Clopidogrel in Patients With Acute Coronary Syndromes," New England Journal of Medicine, 361, (11) 1045–1057 (2009).
(5) M. Tanaka and T. Nagata, "Characterization of Clinical Data Packages Using Foreign Data in New Drug Applications in Japan," Clinical Pharmacology & Therapeutics, 84 (3) 340–346 (2008).
(6) International Conference on Harmonization, "E5 Implementation Working Group, Questions & Answers (R1)," (2006), http://bit.ly/173XM7n.
(7) M. Ohishi, "Potential Factors Influencing Regional Differences and Similarities in Multiregional Clinical Trials," Drug Information Journal, 46, 565–572 (2012).
(8) R. Wand et al., "Statistics in Medicine—Reporting of Subgroup Analyses in Clinical Trials," New England Journal of Medicine, 357 (21) 2189–2194 (2007).
Related Content:Trial Design