Originally published as JCO Early Release 10.1200/JCO.2008.20.8611 on February 23 2009

This Article Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Rights & Permissions Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Sargent, D. Search for Related Content PubMed PubMed Citation Articles by Sargent, D. Related Articles Related Article Social Bookmarking                            What's this?

Early Stopping for Benefit in National Cancer Institute–Sponsored Randomized Phase III Trials: The System Is Working

Division of Biomedical Statistics and Informatics, Mayo Clinic, Rochester, MN

The act of consent for a patient to enter a randomized, clinical trial requires a fundamental pact between physician and patient. For the patient, the choice to allow chance to determine his treatment rests on the assumption that his physician and the medical community at large believe that there is clinical equipoise between the treatments under investigation. This equipoise allows ethical randomization.¹ In addition to the need for equipoise at the time of trial entry, a further expectation for patient participation in clinical trials requires that if any relevant information becomes available during the course of the trial, this information will be made available to trial participants. This need for constant reinforcement of the ethics of randomization raises the issue of when, and through what mechanisms, data from ongoing trials should be examined to determine if the trial's hypothesis has been resolved and, therefore, that the result is known and should be communicated to the scientific and patient communities.

Multiple valid reasons exist to consider premature termination of a clinical trial, several of which cannot be argued against (excessive toxicity, changes in standard of care, poor accrual and, somewhat more controversially, stopping for futility because the trial is highly unlikely to ever attain a statistically significant result²). The reason for early stopping within an ongoing trial that generates the greatest controversy, however, is terminating a trial early due to the apparent efficacy of the experimental arm versus the control. Recently, considerable attention has been focused on issues surrounding the early release of efficacy data from ongoing clinical trials.^3–7 Clearly, the premature termination of a clinical trial can have highly problematic consequences. Conclusions based on immature data may change with further follow-up or provide inadequate evidence to convince the broader scientific community. In addition, data from the initial patients enrolled on a trial may not be representative of data that would be obtained from the remainder of the trial, as the early patients (or early enrolling sites) may not be representative of the general population of patients or sites. Release of early data may also prohibit the collection of meaningful data on long-term benefits or adverse events from therapy if the data release results in a change in care for patients currently on the clinical trial (for example, with cross-over treatment). Finally, the estimation of the treatment effect has been shown to be biased away from the null hypothesis (and thus be overly optimistic) if the clinical trial is stopped early, even if stopped at a planned interim analysis,⁸ although recent work in oncology suggests that if the trial is stopped after at least 50% of the planned events have been observed, this bias is modest.⁹

Clearly, these considerations result in a complex set of factors that must be considered as clinical trials are ongoing, in particular if the interim data from the trial is in some sense promising. The need for regular, careful, and deliberate monitoring of data from ongoing clinical trials has led to the now widely established model of the independent data monitoring committee (IDMC; also often known as a Data Safety Monitoring Board); the need to confine data access in ongoing trials to such a committee has been clearly articulated.¹⁰ The IDMC is generally comprised of three to eight individuals with expertise in the field of the disease under study, who are not affiliated in any way with the ongoing trial that the IDMC is charged to monitor. These IDMCs typically consist of clinicians, statisticians, and often patient representatives or ethicists. The prominence (and sophistication) of IDMCs has become such that a book focused exclusively on this topic has recently been published.¹¹ The US National Cancer Institute (NCI) has required IDMC monitoring of phase III clinical trials conducted through its cooperative group program since at least 1996.

In this issue of Journal of Clinical Oncology, Korn et al¹² present a thorough review of the outcomes associated with clinical trials stopped or reported early that have been sponsored by the NCI through the cooperative group program. Based on a comprehensive review, Korn et al identified 27 phase III clinical trials from 1990 to 2005 in which either the accrual was stopped early or outcome data were released before attaining the protocol-specified number of events. As expected, the trials had a wide variety of primary end points; in 70% of the trials, accrual was complete at the time of closing, which has been recommended by some as an important criterion before trial reporting should be considered.⁶ In only one case were data released before at least 25% of the protocol specified number required for the primary analysis had been observed, consistent with recommendations that trial termination based on very early and incomplete data should be avoided.³ The most important finding by Korn et al was that in 89% of trials (24 of 27) stopped or reported early, the conclusions based on the final analysis were consistent with those made at the time of the early trial reporting. It should be noted that in four of these 24 trials, the estimated treatment effect at the final analysis was more moderate than that observed at the trial stopping point, providing some data to support the concept that the observed treatment effect when a trial is stopped for success at an interim time point may be overly optimistic. In one trial, the estimated treatment effect became considerably smaller at the final analysis compared with the interim (although still statistically significant per the protocol). In two (7%) of 27 trials, the significant results present at the analysis that led to trial termination were not maintained in a later analysis; in both trials, examination of the survival curves suggested an early benefit of therapy that was not maintained at a statistically significant level in the long term. These reversals highlight the need for continued long-term follow-up of patients on clinical trials, particularly when the trial is reported before its planned final analysis.

How concerned should we be with regard to these two trials whose conclusions changed from significant to nonsignificant with further follow-up? Recent experience has demonstrated that the reversal may go both ways: in the Multicenter International Study of Oxaliplatin/5FU-LV in the Adjuvant Treatment of Colon Cancer trial, no statistically significant benefit was observed for the secondary end point of overall survival at the protocol-specified analysis time point of 5 years,¹³ but subsequent analysis after 6 years' follow-up did demonstrate a significant survival benefit in stage III patients from the addition of oxaliplatin to fluorouracil and leucovorin.¹⁴ In the two cases from the NCI trials discussed by Korn et al,¹² while the statistical significance did vanish with further follow-up, in neither case did the survival curves reverse at the later analysis, indicating that while scientific information may have been diminished, no direct patient harm resulted from the early stopping.

Does appropriate early trial reporting result in patient benefit? I believe the answer is clearly yes. In the case of a clearly convincing superiority of one treatment over another, if both therapies are available, the importance of allowing an appropriate change in clinical practice is clear. Even if the superior intervention is not generally available, cross-over to the superior arm may be allowed for patients on the other arm(s), providing all enrolled patients the opportunity to receive the therapy with greater benefit. For unapproved agents, appropriate early reporting allows more rapid regulatory filings to allow a therapy to become widely approved, a necessary step to widespread patient benefit from any new therapy. Appropriate early reporting also leads to scientific benefit; from the scientific perspective, once a result is known, it should be released to speed scientific progress. For example, the early reporting of the efficacy of trastuzumab as adjuvant therapy for human epidermal growth factor receptor 2–positive breast cancer¹⁵ allowed the more rapid development of a subsequent trial to compare trastuzumab with lapatinib in that setting (the ongoing Adjuvant Lapatinib and/or Trastuzumab Treatment Optimisation trial). Finally, the reporting of definitive early results fulfills the aforementioned pact between the physician and the patient: a clinical trial is only ethical if equipoise is maintained.

The recent literature on early stopping and/or reporting of clinical trials for efficacy, supplemented by the new report of Korn et al,¹² allows several conclusions. First, interim monitoring of clinical trials is critical to ensure appropriate patient protection. Second, this monitoring should be performed by experienced, independent monitoring committees, who report directly to the study sponsor. In my opinion, interim analyses, reviewed by an IDMC for safety, efficacy, and futility, are all appropriate, and should be the default in randomized phase III trials. Third, when this interim monitoring is performed by well-informed committees, guided by appropriate statistical stopping rules to protect against the dangers of multiple comparisons,¹⁶ the vast majority of the time, positive results from interim analyses are maintained. Fourth, when trials that are stopped early are reported, the critical details must be given of when, why, under what guidelines, and by whom the trial was stopped.⁴ Fifth, continued follow-up of patients on all study arms, to allow future confirmatory analyses, is critical. Finally, and most importantly, the procedures established and continually improved on by the NCI and the cooperative groups—in which data review is performed by independent boards that are guided but not constrained by statistical boundaries—remain appropriate to protect the critical trust that patients place in their providers and in the clinical research community.

AUTHOR'S DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Although all authors completed the disclosure declaration, the following author(s) indicated a financial or other interest that is relevant to the subject matter under consideration in this article. Certain relationships marked with a "U" are those for which no compensation was received; those relationships marked with a "C" were compensated. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.

Employment or Leadership Position: None Consultant or Advisory Role: Daniel Sargent, Sanofi aventis (C), Genentech (C), Pfizer (C), GlaxoSmithKline (C), Roche (C) Stock Ownership: None Honoraria: Daniel Sargent, Sanofi aventis, Genentech, Pfizer, GlaxoSmithKline, Roche Research Funding: None Expert Testimony: None Other Remuneration: None

REFERENCES

1. Freedman B: Equipoise and the ethics of clinical research. N Engl J Med 317:141–145, 1987.[Abstract]

2. DeMets DL: Futility approaches to interim monitoring by data monitoring committees. Clinical Trials 3:522–529, 2006.[CrossRef][Medline]

3. Wheatley K, Clayton D: Be skeptical about unexpected large apparent treatment effects: The case of an MRC AML12 randomization. Control Clin Trials 24:66–70, 2003.[CrossRef][Medline]

4. Montori VM, Devereaux PJ, Adhikari NKJ, et al: Randomized trials stopped early for benefit: A systematic review. JAMA 294:2203–2209, 2005.[Abstract/Free Full Text]

5. Mueller PS, Montori VM, Bassler D, et al: Ethical issues in stopping randomized trials early because of apparent benefit. Ann Intern Med 146:878–881, 2007.[Abstract/Free Full Text]

6. Bassler D, Montori VM, Briel M, et al: Early stopping of randomized clinical trials for overt efficacy is problematic. J Clin Epidemiol 61:241–246, 2008.[CrossRef][Medline]

7. Wilcox RA, Djulbegovic B, Guyatt GH, et al: Randomized trials in oncology stopped early for benefit. J Clin Oncol 26:18–19, 2008.[Free Full Text]

8. Pocock S, White I: Trials stopped early: Too good to be true? Lancet 353:943–944, 1999.[CrossRef][Medline]

9. Freidlin B, Korn EL: Stopping clinical trials early for benefit: Impact on estimation. Clinical Trials in press.

10. Piantadosi S: Rigor in monitoring clinical trials is ethical. J Clin Oncol 26:683–685, 2008.[Free Full Text]

11. Ellenberg S, Fleming T, DeMets D. Data Monitoring in Clinical Trials: A Practical Perspective. Hoboken, NJ: John Wiley & Sons, Inc, 2002.

12. Korn EL, Freidlin B, Mooney M: Stopping or reporting early for positive results in randomized clinical trials: The National Cancer Institute Cooperative Group Experience from 1990 to 2005. J Clin Oncol 27:1712–1721, 2009.[Abstract/Free Full Text]

13. de Gramont A, Boni C, Navarro M, et al: Oxaliplatin/5FU/LV in the adjuvant treatment of stage II and stage III colon cancer: Efficacy results with a median follow-up of 4 years. J Clin Oncol 23:246s; 2005 (suppl) abstr 3501.[CrossRef]

14. de Gramont A, Boni C, Navarro M, et al: Oxaliplatin/5U/LV in adjuvant colon cancer: Updated efficacy results of the MOSAIC trial, including survival, with a median follow-up of six years. J Clin Oncol 25:165s; 2007 (suppl) abstr 4007.

15. Romond EH, Perez EA, Bryant J, et al: Trastuzumab plus adjuvant chemotherapy for operable HER2-positive breast cancer. N Engl J Med 353:1673–1684, 2005.[Abstract/Free Full Text]

16. O'Brien PC, Fleming TR: A multiple testing procedure for clinical trials. Biometrics 35:549–556, 1979.[CrossRef][Medline]

CiteULike    Complore    Connotea    Del.icio.us    Digg    Facebook    Reddit    Technorati    Twitter    What's this?

Related Article

• Stopping or Reporting Early for Positive Results in Randomized Clinical Trials: The National Cancer Institute Cooperative Group Experience From 1990 to 2005
  Edward L. Korn, Boris Freidlin, and Margaret Mooney
  JCO 2009 27: 1712-1721 [Abstract] [Full Text]

This Article Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Rights & Permissions Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Sargent, D. Search for Related Content PubMed PubMed Citation Articles by Sargent, D. Related Articles Related Article Social Bookmarking                            What's this?