© 2005 American Thoracic Society
Poor Choice of Primary Outcome in a Clinical Trial of Pirfenidone in Patients with IPFFrom the Authors:Several of Mathai and Polito's and Levitt and Gould's concerns about our paper have already been addressed in the article itself and in the editorial that accompanied it (1, 2). While we agree that the study would have revealed further insights regarding the safety and efficacy of pirfenidone had it not been aborted, the ethical issues raised by the Data Safety Monitoring Board's (DSMB) recommendation for the early cessation of the trial should be appreciated. The patients and clinicians in Japan are sensitive to the diagnosis of acute exacerbation (AE) of idiopathic pulmonary fibrosis (IPF) and are well aware of this clinical scenario ever since it was first described in Japanese patients with IPF (3). Clinicians are increasingly becoming aware of AE as a fatal complication in the clinical course of IPF. In fact, an acute clinical deterioration preceded death in 47% of patients who died with progressive IPF (4, 5). Thus, recognition of the occurrence of AE in patients with IPF has important implications and is clinically relevant. However, the criteria for definition of AE in IPF need to be established by an international consensus among experts. Since the efficacy of a treatment regimen for IPF in reducing the occurrence/frequency of AE and consequent improvement in survival is yet to be determined, we were cautious in reporting our findings as encouraging and never concluded that pirfenidone was "superior" as alleged. In our study, a modified version of the criteria for AE used by Kondoh and coworkers (3) was defined and prespecified. Thus the DSMB's recommendation to stop the trial is far from "dubious." Mathai and Polito have misinterpreted the data regarding the randomization of the eligible population. Levitt and Gould were also concerned regarding this. An enrollment center determined and ascertained that all subjects enrolled in the study met the prespecified eligibility requirements before randomization occurred. The eligibility criteria did not depend on the subject's ability to complete the 6-min exercise test (6MET). Thus, all eligible patients (including the 27 patients in question) were entered in this study, randomized, and analyzed as the full analysis set. Based on prespecifications, we amended our design at the beginning of the study and proactively decided to perform the analysis in this subset of patients. We have emphasized this in the article and the data for all of the eligible population have been presented as primary data. The power calculation in determining the sample size was based on simulations; the sample size was prespecified to be 90 patients. This minimum number of patients provided statistical power of approximately 0.9 to detect assumed efficacy at the significance level of 0.025 and maintained the power greater than 0.8 assuming some patients may drop out. A decrease in SpO2 during exertion is a characteristic clinical feature of IPF. In this regard, oxygen desaturation using SpO2 measurements during the 6-min walk test (6MWT) and a modified version, the timed walk test, have been associated with survival in the IPF population in two recent studies (6, 7). Our decision in choosing the primary endpoint was based on the encouraging, but unpublished, results of an ongoing study (8). In this prospective study, end exercise SpO2, change in SpO2 during exercise, walk distance and walk velocity were strongly correlated with survival in patients with IPF followed for five years (7). While the recently published study by Eaton and coworkers (9) raises potential concerns in utilizing and assessing SpO2 measurements obtained during the routinely used 6MWT, a steady state exercise study (the 6MET) was used in the pirfenidone study. Hence, Eaton and coworkers' results cannot be extrapolated to the findings with the 6MET used in our study. Anticipating reproducibility problems in the 6MWT in the Japanese patients, we opted for the steady state 6MET (10). The protocol was followed per prespecifications. Since the speed chosen by the individual patient at baseline for the 6MET was kept the same during the follow-up 6MET, the physical burden associated with the 6MET was standardized for each patient. To investigate if the walking speed affected the lowest SpO2, we have evaluated the mean changes of the lowest SpO2 from baseline to the 6th and 9th month among the groups defined by the treadmill speed with one-way ANOVA. Equality was not rejected for the values observed at 6th and 9th month follow-up (p = 0.2114 and p = 0.4732, respectively). There was no significant difference in the distribution of the treadmill speed between pirfenidone and placebo groups (p value of the Wilcoxon rank-sum test was 0.9683). Nevertheless, the limitation of utilizing this endpoint has been acknowledged and discussed (1, 2). It must be emphasized that conventionally used measures of the functional status in patients with IPF were not omitted as they were assessed as secondary endpoints in this study. The issue regarding the disclosure of the membership of the DSMB is an appropriate one. The following were the members of the DSMB:
It should be reiterated that this was a phase II clinical study and not a definitive, phase III clinical trial. The signals and lessons from this study have provided useful clinical information for better design and choice of appropriate endpoints for IPF clinical trials. The "call" for a well-designed phase III clinical trial is thus valid. The difficulty and feasibility of conducting a real placebo-controlled trial (i.e., a control population without any treatment) is independent of the results of this study and a subject for a different debate. A placebo-controlled phase III clinical trial using pirfenidone in Japanese patients is already well underway. It is hoped that the results of this ongoing study will clarify the concerns raised in the phase II study.
University of Washington, Seattle, Washington Acknowledgments This letter is submitted on behalf of the Japanese IPF study group and my coinvestigators. FOOTNOTES Conflict of Interest Statement: G.R. received $1,000 in 2002 from Shionogi, $500 in 2003 from Intermune as a consultant, and $2,500 in 2004 from Intermune as an Advisor for pirfenidone studies in idiopathic pulmonary fibrosis. REFERENCES
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
