ABSTRACT
In a multiregional clinical trial (MRCT), besides issues that would typically be addressed by data monitoring, regional aspects of the MRCT will also be monitored. General principles and recommendations are made by some authors (Gallo et al., 2011), especially when an independent Data Monitoring Committee (DMC) is in place to review the accumulated, unblinded data (Wittes, 2016). Group sequential designs that allow early stopping of a trial due to convincing efficacy at unblinded interim analyses and safety monitoring, which may lead to trial modification or termination if there are undue risks to patients, are widely used in large-scale MRCTs. It has been recommended that due to the limited amount of information within individual regions at interim analyses, the overall treatment effect estimate should usually be interpreted as the best estimate for all individual regions during DMC evaluation (Wedel et al., 2001), and that a DMC early stopping
CONTENTS
13.1 Introduction .............................................................................................. 167 13.2 Different Data Patterns by Region ......................................................... 168 13.3 Enrollment and Data Contribution by Region ..................................... 169 13.4 Imbalance in Risk Factors by Region .................................................... 170
13.4.1 EVEREST Trial ............................................................................ 171 13.4.2 PURSUIT Trial ............................................................................ 171 13.4.3 PLATO Trial ................................................................................ 171
13.5 Adaptation to Address Imbalance in Prognostic/Predictive Factors ........................................................................................................ 172 13.5.1 Adaptation Plan .......................................................................... 174
13.6 Concluding Remarks ............................................................................... 174 References ............................................................................................................. 175
decision based on efficacy should primarily be based upon the overall signal, with appropriate consideration of supporting evidence of consistency across regions or other subgroups (O’Shea and DeMets, 2001). That was the rationale when Metoprolol CR/XL Randomized Intervention Trial in Congestive Heart Failure was stopped early due to convincing efficacy based on overall data. Subsequently, there was a debate on the appeared difference of treatment effect in patients from the United States (US). This raised the issue if a decision at an interim analysis should also take into account the treatment effect in certain markets that are considered important, medically or commercially. If a discrepancy from overall results is suggested for such a region, extra efforts and additional exploration may be appropriate before a decision is made to stop the trial, with additional challenges due to the smaller amount of data available. For such a reason, some authors suggest that at the interim analysis, always report US alone and perform other subgroup and regional analyses (Wittes, 2013). In a recent outcome study, part of the criteria in “early efficacy stopping rules” included a check on whether results for US-based patients are in the right direction. In fact, when developing the study design and assessing the design characteristics, the simulation model considered the impact of the stopping rules, including the “in the right direction” requirement on the probability of early stopping, either correctly or incorrectly. The criteria were mutually agreed by the sponsor and the DMC. Similar considerations apply to stopping an MRCT early due to futility or lack of efficacy. If the early data which the interim decision is based upon come primarily from one or a few countries, care is needed to ensure that these early data are representative of the global data.
