|
Correction
to
McGuire et al.,
|
| |||||||||||||||||||||||||||||||||||
First, there is increasing evidence that the patients in the control group may not have received the best available platinum-based treatment. McGuire et al. state, "It appears that the relative benefit of paclitaxel in our trial was not due to a poorer-than-anticipated outcome" in the control group. To support their statement, the authors refer to a predecessor to their trial, which had similar results. This previous Gynecologic Oncology Group trial2 compared a standard-dose regimen of cyclophosphamide and cisplatin with an intensive-dose regimen (and did not involve "similar doses of cisplatin and paclitaxel" — this is clearly an error). In the intensive-dose regimen, the dose intensity of both cisplatin and cyclophosphamide was doubled, but the total dose of the two drugs was identical in each group.
A trial by Kaye et al.,3 however, which also compared two cyclophosphamide and cisplatin regimens, one of which involved a doubling of both the dose intensity and the total dose of cisplatin, has shown a survival benefit associated with the regimen involving a higher total dose of cisplatin. There is also strong evidence from meta-analyses that the addition of doxorubicin to cyclophosphamide and cisplatin is a more effective treatment than cyclophosphamide and cisplatin alone.4,5
Second, McGuire et al. estimated that paclitaxel plus cisplatin improved the median survival by 14 months. Nevertheless, there is still uncertainty about the true effect. The approximate 95 percent confidence interval for the range of effects that are still plausible is six months to two years. A difference in survival of two years would be exciting, a difference of six months less so.
Finally, the improvement in median progression-free survival was less than half the observed improvement in median survival. Such an apparent discrepancy would not have been expected and raises the question whether there may be an element of chance associated with the much larger observed difference in overall survival.
Despite the very encouraging results of this trial, there is an urgent need for further large trials comparing the combination of paclitaxel and platinum with the best current standard treatment available. The Third International Collaborative Ovarian Neoplasm Study, which has a maximal accrual target of 1000 patients and 265 patients enrolled to date, is one such trial, and it involves a broader range of women with ovarian cancer.
Mahesh K.B. Parmar, D.Phil.
Josie Sandercock, B.A.
Medical Research Council Cancer Trials Office
Cambridge CB2 2BW, United Kingdom
References
The imbalance of prognostic factors in the two treatment groups (12 percent fewer patients with serous adenocarcinoma and almost double the proportion of grade 1 tumors in the cisplatin–cyclophosphamide group), the lack of mention of differences in outcome among the participating centers (one of the two variables stratified in the randomization process), and the lack of adjustment for imbalances in prognostic factors in the analysis of progression-free and overall survival are some of the technical points that require clarification. The proportion of ineligible patients (5.9 percent) was low, but no mention is made of the distribution of such patients in the two groups or the reasons for their ineligibility (the wrong primary tumor [in 13 patients] and the wrong cell type [in 3] are not clear enough reasons), which is a matter of concern, since there were 9 percent fewer patients in the cisplatin–paclitaxel group than in the cisplatin–cyclophosphamide group.
The fact that 97 percent of the patients had been enrolled by November 1992,1 together with the follow-up range reported (5 to 56 months), shows that the data are at least 30 months old, without any update since mid-1993. The report does not mention the starting and ending dates for enrollment or the rate of enrollment, the cutoff date for the last collection of data is not stated, and the survival curves show no denominators for patients at risk over time. The results of the trial will not change, but the overall shape of the curves and level of significance may change.
The most important issue is that the 5-month difference in the median time to the progression of disease is only about one third the difference in median overall survival (14 months), a counterintuitive and rare finding in clinical trials involving patients with advanced malignant solid tumors. Such a result is probably due to differences in second-line therapy in the two groups. Patients in the cisplatin–paclitaxel group with a recurrence or progression of disease appear to have been treated further with alkylating agents, but the authors do not indicate whether they were treated again with paclitaxel. The authors state that patients in the cisplatin–cyclophosphamide group did not receive paclitaxel as second-line treatment, or received it only very late in the course of the disease, because a limited supply of paclitaxel prevented a crossover study. Either the crossover was specifically not allowed, or there is an ethical problem, since the supply of paclitaxel during the period of the investigation was adequate for its distribution as salvage therapy on a compassionate-use basis to at least 1819 patients with ovarian cancer in the United States.2 This point is worrisome, because the end points of the trial are not clearly stated in the article. The difference in survival between the two groups may thus be due solely to the fact that the patients randomly assigned to the paclitaxel–cisplatin group received three active drugs during their illness, whereas most patients in the control group received only two active drugs.
Since second-line therapeutic choices in ovarian cancer result in good palliation and the prolongation of life, differences in overall survival are likely to be significantly influenced by second-line treatment. If McGuire et al. denied such treatment to one or both groups of patients, we believe this approach is questionable as an ethical therapeutic design.
Esteban Cvitkovic, M.D.
Jean-Louis Misset, M.D.
Hôpital Paul Brousse
94804 Villejuif, France
References
It is also important that the two groups of patients did not differ in terms of the number of complete responses confirmed by second-look operations. We think a longer follow-up is needed to determine whether the difference in survival is maintained.
This limitation of the study, the high cost of the treatment, and the difficulties of the schedule proposed by the authors (a 24-hour infusion, which suggests an added cost) lead us to conclude that paclitaxel and cisplatin should not be adopted in routine clinical practice until more studies have confirmed that this regimen increases the cure rate or long-term progression-free survival.
Angel J. Lacave, M.D.
Ignacio Peláez, M.D.
Isabel Palacio, M.D.
Hospital General de Asturias
33006 Oviedo, Spain
To the Editor: We agree that clinical trials in women with advanced ovarian carcinoma that focus only on survival can be misleading. Since it is not clear how tumor debulking during reassessment surgery and the timing of salvage chemotherapy affect survival, we are continuing to assess clinical responses in patients with measurable disease.
Parmar and Sandercock and Lacave et al. suggest that the benefit we found in overall and progression-free survival among patients treated with cisplatin and paclitaxel is due to an inappropriate standard treatment. Parmar and Sandercock suggest that we should have used a more intensive regimen of cisplatin, whereas Lacave et al. note that the toxic effects of the standard regimen required delays between courses of treatment and therefore resulted in a diminished dose intensity. The doses in our study were a reasonable compromise between intensive regimens of cisplatin and cyclophosphamide, which cause substantial toxic effects, and less intensive regimens, which prolong the treatment in order to achieve the total doses planned.1
The results of a meta-analysis suggest that adding doxorubicin to regimens based on an alkylating agent or platinum may decrease the death rate by 15 percent.2 Parmar and Sandercock note that this result is similar to an estimate from a meta-analysis of randomized trials comparing platinum and nonplatinum regimens.3 Most of these trials, however, did not restrict the use of platinum in salvage therapy. Any advantage due to platinum given in the initial treatment was probably obscured, since the survival comparisons reflect early versus late treatment with platinum.3 Because the evidence for doxorubicin is not compelling, considering it equivalent to platinum overstates its benefit.
Cvitkovic and Misset suggest that the observed survival benefit associated with paclitaxel is due to the 3 percent difference in grade 1 tumors. This small difference, however, favors the cisplatin–cyclophosphamide regimen rather than the cisplatin–paclitaxel regimen. In addition, previous studies have indicated that with the exception of the clear-cell and mucinous types, the cell type is not an important prognostic factor in women with suboptimally debulked ovarian cancer.
Although the manuscript we submitted indicated that patients were enrolled in this study between April 1990 and March 1992, this information was deleted in the editorial process. The results presented summarize the data on survival as of April 1995. The tick marks in Figure 1 and Figure 2 of the article indicate the censored survival times. The follow-up of these patients is ongoing.
W.P. McGuire, M.D.
Emory University
Atlanta, GA 30322
W.J. Hoskins, M.D.
Memorial Sloan-Kettering Cancer Center
New York, NY 10021
M.F. Brady, B.S.
Roswell Park Cancer Institute
Buffalo, NY 14263
References
| |||||||||||||||||||||||||||||||||||
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
