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 Fosså, S. D. Articles by Skovlund, E. Search for Related Content PubMed PubMed Citation Articles by Fosså, S. D. Articles by Skovlund, E.

Interim Analyses in Clinical Trials: Why Do We Plan Them?

Department of Medical Oncology and Radiotherapy The Norwegian Radium Hospital Oslo, Norway

THERAPEUTIC CLINICAL trials are scientific experiments in human beings. The performance of a clinical trial is only justified if the clinical investigators in advance consider ethical aspects and if an external ethical committee has approved the conduct of the study according to a defined protocol. Ethical aspects include the obligation that only the minimum number of patients should be entered onto the trial, which is necessary to achieve the study’s primary objective given previous experience with the specific treatment. If the medical literature or previous experience indicates a high variability of the efficacy of treatment to be tested or if the statistical assumptions upon which the trial is based seem questionable, close monitoring of the trial conduct and its results is highly recommended. Such awareness is of particular importance if the trial treatment leads to considerable toxicity.

In such a situation, investigators may plan an interim analysis, preferably conducted by an independent data monitoring committee (IDMC). The task of the IDMC is to assess whether the ongoing trial realistically can be expected to answer its primary question, taking into account the following aspects:

1. Is the inclusion rate of patients acceptable and as expected?
   
2. Is there an unexpectedly high rate of severe or life-threatening adverse events, which may indicate the premature closure of the trial?
   
3. Is the outcome of the trial treatment comparable with that of the previous experience upon which the specific trial is based? The answer to this question is of particular importance in early phase III multicenter trials, which are based on results from smaller phase II trials performed at one or few institutions. It is well known that the results of such limited phase II studies are often superior to the outcome of multicenter studies because of different patient selection.
   
4. Do the results of the interim analysis prove statistically significant differences between the trial treatments that exceed the differences defined by the statistical guidelines of the trial? This would warrant closure of the study. This policy led, for example, to premature closure of two large trials testing the role of metoprolol in chronic heart disease¹ or simvastatin in hypercholesterolemia.²
   

Recognizing the possible difficulties with early stage III trials in clinical oncology, some cooperative groups start a trial as a randomized phase II study that may or may not lead to a subsequent phase III trial, dependent on predefined outcome criteria of an interim analysis and the advice of an IDMC. Patients of such a randomized phase II study can subsequently be included in the phase III trial if the study continues. This policy of a randomized phase II study preceding a phase III trial has been followed several times by the Genitourinary Group of the European Organization for Research and Treatment of Cancer, resulting in early discontinuation³ or continuation⁴ of the trial. Importantly, no patient was entered onto the planned phase III trial before the interim analysis had been performed and the IDMC had given its advice. Extension to the phase III trial was felt unjustifiable before the recommendation of the IDMC was known.

The article by Negrier et al⁵ gives an example of misconduct of a clinical trial in this respect as the investigators disregarded their own study design and the role of the predefined interim analysis. Based on favorable though preliminary phase II trial results with selected chemoimmunotherapy in metastatic renal cell carcinoma,⁵ the combination of subcutaneous interleukin-2 (IL-2), interferon alfa (IFN), and fluorouracil was compared with the combination of IL-2 and IFN. This comparative study was planned before the final phase II results⁶ were known, which proved the chosen chemoimmunotherapy to be ineffective in this malignancy. When the phase III trial was designed, it was evident from the medical literature that only few patients with metastatic renal cell carcinoma would benefit from immunotherapy: those with a good performance status and minimal metastatic disease, preferably in lung and lymph nodes. Oncologists treating these patients knew that the majority of the patients would experience significant constitutional toxicity from IL-2/IFN–based immunotherapy without tumor response or prolongation of life. Because of this uncertainty, the principal investigators wisely planned a randomized phase II design with 21 patients in each arm and a subsequent interim analysis. The protocol contained clear stopping rules to be based on the results of the interim analysis: if a 10% response rate was obtained in either trial arm and a difference in response rates of greater than 15% between the two alternatives were observed, then the trial would be discontinued. These criteria may be challenged, as they represent guidelines for selection of the best treatment in a randomized phase II single-drug trial.⁷ Identification of the best treatment is, however, not the goal of a randomized phase II trial preceding a phase III trial.

Despite their clearly defined stopping rules, Negrier et al⁵ did not follow their own design: patient inclusion was continued during the period of interim analysis until the trialists themselves required the premature end of the trial because of an unexpectedly low response rate. At that time, 131 of the 182 planned patients had been entered, whereas the results of the interim analysis would have led to closure of the trial after 42 patients. The continuation of the trial is even more unexplainable, as the disappointing results of the preceding phase II trial should have been suspected at the time when the interim analysis was due.

Proper interim analysis after 42 patients and the results of their own previous phase II study⁶ would also have led to the consideration of another problem with the study by Negrier et al⁵: because of the very rapid inclusion rate in this phase III study, despite its being a multicenter effort, the principal investigators should have suspected an inadequate selection of patients. In their final report, the authors correctly discuss this fact as a reason for the low response rate. This problem could, however, have been largely avoided by a proper interim analysis and discussion of inclusion rate with the trialists during an investigator meeting.

The investigators’ choice to continue the trial before the results of the interim analysis were made known to an IDMC, thus neglecting their own stopping rules, cannot be justified, even though many patients would probably have received immunotherapy in any case. Clinical care of an individual patient has ethical rules different from the strict regulations of a clinical trial. If an interim analysis is initially planned to monitor treatment effects and the safety of a trial (based on appropriate statistics), and if stopping rules are stated in the protocol, then trialists have to stick to their own rules and await the results of the interim analysis before continuation of the trial. If these elementary rules for clinical research are not adhered to, then ethical committees might become reluctant to permit the conduct of combined phase II/III studies in the future.

REFERENCES

1. Effect of metoprolol CR/XL in chronic heart failure: Metoprolol CR/XL Randomised Intervention Trial in Congestive Heart Failure (MERIT-HF) [see comments]. Lancet 353:2001-2007, 1999

2. Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease: The Scandinavian Simvastatin Survival Study (4S) [see comments]. Lancet 344:1383-1389, 1994

3. Albrecht W, Horenblas S, Marechal JM, et al: Randomized phase II trial assessing estramustine and vinblastine combination chemotherapy vs estramustine alone in patients with hormone escaped progressive metastatic prostate cancer. J Urology 161: 298, 2000 (abstr 1276)

4. Fosså S, Mikisch GHJ, Mulder PM, et al: Interferon-alfa 2A (IFN) with or without 13-cis-retinoid acid (RA) in metastatic renal cell carcinoma (MRCC): Toxicity analysis of an EORTC randomised phase II study. Proc Am Soc Clin Oncol 17: 331, 1998

5. Negrier S, Caty A, Lesimple T, et al: Treatment of patients with metastatic renal carcinoma with a combination of subcutaneous interleukin-2 and interferon alfa with or without fluorouracil. J Clin Oncol 18: 4009-4015, 2000[Abstract/Free Full Text]

6. Ravaud A, Audhuy B, Gomez F, et al: Subcutaneous interleukin-2, interferon alfa-2a, and continuous infusion of fluorouracil in metastatic renal cell carcinoma: A multicenter phase II trial—Groupe Francais d’Immunotherapie. J Clin Oncol 16: 2728-2732, 1998[Abstract]

7. Simon R, Wittes RE, Ellenberg SS: Randomized phase II clinical trials. Cancer Treat Rep 69: 1375-1381, 1985[Medline]

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 Fosså, S. D. Articles by Skovlund, E. Search for Related Content PubMed PubMed Citation Articles by Fosså, S. D. Articles by Skovlund, E.