Author + information
- Sanjay Kaul, MD⁎ (, )
- George A. Diamond, MD, FACC and
- William S. Weintraub, MD, FACC
- ↵⁎Division of Cardiology, Cedars-Sinai Medical Center, 8700 Beverly Boulevard, Los Angeles, California 90048
We appreciate the opportunity to clarify several of the statistical issues raised by the SPORTIF studies. Our analysis (1) was submitted for presentation at the American College of Cardiology (ACC) scientific sessions in September 2004, a month before the Food and Drug Administration’s (FDA) October 2004 review (2), and a preliminary analysis indicating a high probability of ximelagatran being worse than warfarin was e-mailed to the principal investigator (J.H.) on November 13, 2003.
1. The noninferiority margin. The statement by the investigators that “There is more focus on the predefined delta than on the results” ignores the International Conference on Harmonization guidance (3) to the effect that the operative margin should be pre-specified, founded on both “clinical judgment” and “statistical reasoning,” and be “suitably conservative.” Instead, SPORTIF’s margin appears to be chosen entirely on the basis of expert consensus and was judged by the FDA as too liberal (2). We stand by the more objective meta-analytic estimate in our study (1).
2. The expected warfarin rate. SPORTIF’s rate of 3.1% is inconsistent with 5 previous studies (1.9%) and the meta-analysis described in the trial design (1.4%).
3. The relative noninferiority margin. Based on a conservative estimate of the margin (1.44 relative risk [RR]), noninferiority would have been established only for SPORTIF III, but not for SPORTIF V.
4. Neglecting SPORTIF III. The warfarin event rate was twice as high in SPORTIF III (open-label) compared to SPORTIF V (double-blind) despite similar ximelagatran rates. We, therefore, agree with the FDA that only the latter should be considered pivotal to a judgment of efficacy (2). We did, however, incorporate prior information from SPORTIF III in our Bayesian meta-analysis.
5. Deleterious effects. Despite ximelagatran’s advantage in convenience and pharmacokinetics, it was associated with increased hepatotoxicity, intolerability, and cost without clear bleeding advantage. Analysis of noninferiority should ideally be founded on 3 pre-requisite judgments—that the new treatment 1) exhibits “therapeutic noninferiority” to the standard treatment, 2) would exhibit “therapeutic efficacy” in a placebo-controlled trial, and 3) offers ancillary “nonefficacy benefits” in safety, tolerability, convenience, or cost. We hereby propose a composite score by which each of these attributes is graded on a 0 (unestablished) to 1 (established) scale. A score of 3 out of 3 thereby supports a judgment of so-called virtual superiority to justify consideration of the new over the standard treatment. As summarized in Table 1,virtual superiority is not established for either of the SPORTIF trials or for their combined analysis.
We agree that “balancing risks and benefits” should be integral to the interpretation of clinical trials to avoid introduction of suboptimal (and potentially harmful) treatments into routine clinical practice. We hope that the suggestions outlined here represent a small step in that direction. Of note, on February 14, 2006, AstraZeneca announced it had decided to withdraw ximelagatran from the worldwide market and terminate its development (4). Res ipsa loquitur!
- American College of Cardiology Foundation
- Kaul S.,
- Diamond G.A.,
- Weintraub W.S.
- ↵Lawrence J, Hung J, Mahjoob K, et al. Statistical review and evaluation, clinical studies, NDA 21- 686 (2004). FDA Web site. Available at: http://www.fda.gov/ohrms/dockets/ac/04/briefing/2004-4069B1_07_FDABackgrounder-C-R-stat%20Review.pdf. Accessed October 10, 2004.
- ↵International Conference on Harmonization. Statistical principles for clinical trials (ICH E 9) (1998); International Conference on Harmonization. Guidance on choice of control group and related design and conduct issues in clinical trials (ICH E 10) (2000). Food and Drug Administration, Department of Health and Human Services.
- ↵AstraZeneca Decides to Withdraw Exanta™. Available at: http://www.exanta.com. Accessed February 20, 2006.