|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2002 American Society for Clinical Oncology
Are All Hypotheses Generated Before Data Analysis Prospective?Earle A. Chiles Research Institute, Providence Portland Medical Center, Portland, OR, Data Management Services, Inc, National Cancer Institute-Frederick Cancer Research and Development Center, Frederick, MD THE SURVIVAL OF patients with metastatic melanoma has improved little over the past quarter of a century.1 The introduction of high-dose interleukin-2 (IL-2) and biochemotherapy regimens produced long-term disease-free survival in a small percentage of affected patients but was associated with severe toxicity and available only at specialized treatment centers.2,3 Despite its shortcomings, high-dose IL-2 was approved by the Food and Drug Administration (FDA) because durable complete remissions occurred in approximately 5% of treated patients. In 2000, a new, IL-2–based regimen, IL-2 plus histamine, was presented to the FDA with claims of reduced toxicity and improved survival for one of the worst subgroups of metastatic melanoma, patients with liver metastases (LM). The sponsor’s proposal, which was reviewed by the Oncologic Drugs Advisory Committee (ODAC) in June 2000, was not recommended for approval. In the January 1, 2002 issue of the Journal of Clinical Oncology,4 the data from the randomized trial that formed the basis for the FDA submission were published.
The trial was a multicenter, randomized phase III study comparing combined treatment with histamine dihydrochloride and IL-2 to treatment with IL-2 alone in patients with metastatic melanoma. Preclinical data show that histamine enhances the antitumor effects of IL-2, presumably by decreasing the local production of reactive oxygen species by tumor-infiltrating phagocytes, which can inhibit the function of NK and T cells, the presumed ultimate effectors of IL-2–mediated tumor cell destruction. Furthermore, two small phase II clinical trials in which histamine was combined with IL-2 and interferon-alfa (IFN- What can we conclude from this trial? The data presented indicate that the addition of histamine to this IL-2 regimen is safe. However, based on an analysis of the intent-to-treat overall population (ITT-OA), the addition of histamine adds nothing to the minimal antitumor effects of IL-2 administered at this dose and according to this schedule. Overall response rates were only 3% and 4% in the IL-2 and IL-2 plus histamine arms, respectively. There were only three complete responses (1%) among the more than 300 treated patients, confirming the lack of efficacy of this IL-2 regimen. Granting the authors’ contention that response rate may not be an accurate representation of the antitumor activity of this regimen, we move on to an analysis of the primary end point, survival. There was no significant difference in median survival between the two treatment groups, 8.2 months for the IL-2 group and 9.0 months for the IL-2 plus histamine group, (P = .125). Although there was no survival benefit for the ITT-OA population, after a subset analysis, the authors discovered a statistically significant difference in median survival between the two treatment groups in patients with LM (ITT-LM), 5.1 months for IL-2 alone and 9.4 months for IL-2 plus histamine (P = .008). Does this represent a clinically important finding that would warrant a change in the current standard of care? A major consideration in the interpretation of this study is whether the authors identified the ITT-LM subset prospectively as a study group for primary analysis or whether they performed a post hoc subset analysis. In the former case, careful consideration should be given to the potential benefit of the regimen to a group of patients who have a fatal disease and limited treatment options. In the latter case, we should recall the International Conference on Harmonization guidelines which state the following about subgroup analysis: "These analyses are not intended to ‘salvage’ an otherwise nonsupportive study, but may suggest hypotheses worth examining in other studies or be helpful in refining labeling information, patient selection, dose selection, and so on." The authors claim that the ITT-LM subgroup was selected for analysis prospectively. This is described in the statistical analysis section where they seem to consider how many patients with liver metastases would be required to demonstrate the superiority of histamine and IL-2. However, a detailed reading reveals that the authors never declared the expected benefit or the number of patients required to detect that benefit in the ITT-LM population, as they did for the other primary end point, survival of the entire ITT-OA population. The question then arises, if the population were chosen prospectively, why did the authors fail to stratify on LM versus non-LM groups in their randomization of treatments? In fact, the authors did not stratify for any prognostic factors in this study. This led, at a minimum, to a significant imbalance in the ITT-LM group between the two arms, the subset of benefit according to the authors.
It is critical that prognostic factors that might impact on survival be taken into account in the randomization process to ensure that the distribution of patients with respect to these factors is equal. An unequal distribution of an important prognostic factor could lead to spurious results. For example, in E1684, a trial of adjuvant IFN treatment in patients with melanoma, the subgroup of patients with melanomas thicker than 4 mm and no lymph node involvement actually seemed to do worse as a result of IFN treatment.5 The adverse effect of adjuvant IFN was explained away as an imbalance of patients with ulceration, a poor prognostic factor in the IFN group. Data presented in Table 2 of the Agarwala et al4 study show that among the 20 LM patients with only one site of metastasis (ie, liver only disease), 65% (n = 13) received the histamine + IL-2 treatment and 35% (n = 7) received IL-2 alone. However, for the 109 patients with metastases at two or more sites, only 39% (n = 42) received IL-2 + histamine and 61% (n = 67) received IL-2 alone. This difference in the relative proportions of patients allocated to the two treatment conditions with either one or more than one affected organ sites is significant (P = .028, by a Further support for this conclusion can be found in the public record of the FDA report to the ODAC (www.fda.gov/cder/audiences/acspage/oncologicmeetings1.htm#2000). The minutes of the afternoon session of the ODAC meeting indicate that the final decision to include ITT-LM in the primary analysis was made after patient enrollment onto the study was completed. According to Dr Chiao of the FDA, "The null hypothesis testing for the liver subgroup was introduced in the final statistical analysis plan, dated November 18, 1999, which was about a month after the study finished accrual and five months before the data was analyzed. It is our view that survival analysis in the liver subgroup is not a prospectively defined primary analysis." There are other statistical problems with the study that can be reviewed at the ODAC Web site. Does IL-2 with histamine really prolong survival in patients with LM from malignant melanoma? We don’t think so, but we will leave it to the readers to decide for themselves. However, even a perfectly performed single study of a treatment with this level of benefit would require confirmation before it warranted FDA approval and altered the standard of care. Furthermore, even if all hypotheses were generated prospectively, the study patients appropriately stratified and appropriate statistical analysis performed, the low probability of a positive result in this patient population raises a significant chance that the findings are a false-positive result.6 We believe ODAC was correct in recommending against the approval of this regimen based on this study, and that, unfortunately, the addition of histamine to IL-2 does not represent a significant clinical advance for patients with metastatic melanoma, even those with liver metastases. REFERENCES 1. Barth A, Warek LA, Morton DL: Prognostic factors in 1,521 melanoma patients with distant metastases. J Am Coll Surg 181: 193-201, 1995[Medline]
2.
Atkins MB, Lotze MT, Dutcher JP, et al: High-dose recombinant interleukin-2 therapy for patients with metastatic melanoma: analysis of 270 patients treated between 1985 and 1993. J Clin Oncol 17: 2105-2116, 1999 3. Legha SS, Ring S, Eton O, et al: Development and results of biochemotherapy in metastatic melanoma: The University of Texas M. D. Anderson Cancer Center experience. Cancer J Sci Am 3: S9-S15, 1997 (suppl 1)
4.
Agarwala SS, Glaspy J, O’Day SJ, et al: Results from a randomized phase III study comparing combined treatment with histamine dihydrochloride plus interleukin-2 versus interleukine-2 alone in patients with metastatic melanoma. J Clin Oncol 20: 125-133, 2002 5. Kirkwood JM, Strawderman MH, Ernstoff MS, et al: Interferon alpha-2b adjuvant therapy of high-risk resected cutaneous melanoma: The Eastern Cooperative Oncology Group Trial EST 1684. J Clin Oncol 14: 7-17, 1996[Abstract]
6.
Tannock IF: Commentary on cytoreduction nephrectomy in metastatic renal cancer: The results of Southwest Oncology Group trial 8949. J Clin Oncol 18: 39s-42s, 2000
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2002 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
) yielded promising results and encouraged the authors to proceed with the phase III trial. Inexplicably, the authors chose a dose and schedule of IL-2 that had never been tested before and elected not to include IFN-
2 test). One must ask, if patients with LM treated with histamine plus IL-2 survive longer than those treated with IL-2 alone, is it because they received histamine plus IL-2 or because they had metastases at fewer organ sites and, therefore, a smaller tumor burden at study entry? This imbalance in the number of patients with metastases at only one site that were allocated to the IL-2 plus histamine treatment condition could account for the difference in survival between the two treatment groups in the LM subgroup. Had the authors stratified on the number of organ sites affected (one v > one site), this potential bias could have been avoided. Because no similar survival benefit was found in the ITT-OA population as a whole, because an imbalance in prognostic subsets could explain the findings, and because no scientific hypothesis based on other preclinical data or clinical data predicts that patients with LM should experience a better response to IL-2 plus histamine, we consider the findings of a significant difference in the ITT-LM group to be unreliable. 
