Using the Expected Survival to Explain Differences Between the Results of Randomized Trials: A Case in Advanced Ovarian Cancer

Buyse, Marc; Burzykowski, Tomasz; Parmar, Mahesh; Torri, Valter; Omura, George; Colombo, Nicoletta; Williams, Chris; Conte, Pierfranco; Vermorken, Jan B.

doi:10.1200/jco.2003.04.088

Cited by 10 publications

(4 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“… Our goal has never been to provide an exhaustive comparison of the four trials; instead, we wanted to highlight some differences between them that were not fully addressed in the two papers where MKB Parmar himself provided indirect comparisons of these same four trials! ( 1,2 ) We do, however, share his view that these comparisons can be misleading and are unlikely to shed light on the issue. Contrary to the authors' statement, information on how many patients were given paclitaxel at relapse is given in table 3, and the number of patients who progressed or died is clearly mentioned at the bottom of figures 1 and 2. In line with the authors' suggestion, we agree that re‐analysis, by an independent body, of the original data from all four trials would probably be more informative than indirect cross‐trial comparisons. It would be interesting to see how the authors now explain the results of their ICON4 trial: in this trial, which enrolled 802 patients with relapsed ovarian cancer and a platinum‐free interval of at least 6 months, paclitaxel plus platinum chemotherapy was found to improve survival and progression‐free survival compared with conventional platinum‐based chemotherapy ( 3 ) . We know of no other clinical example of a drug (here, paclitaxel) being inactive in a primary disease setting and showing activity in a salvage disease one. We profoundly disagree with the authors' conclusion that carboplatin alone can be considered as reasonable first‐line treatment of women with advanced ovarian cancer.…”

supporting

confidence: 51%

mentioning

confidence: 92%

See 1 more Smart Citation

Response to the Letter

McGuire¹,

Piccart²

2004

Int J Gynecol Cancer

View full text Add to dashboard Cite

There has been, and there continues to be, a deep sense of frustration among trialists working in the ovarian cancer field: what started as a clean story (ie, two median size randomized clinical trials showing short-and long-term superiority of cisplatinpaclitaxel over cisplatin-cyclophosphamide) finished in a state of confusion, with two other trials, the very large ICON3 study and the smaller GOG 132 trial, not confirming a benefit from the incorporation of paclitaxel into front-line platinum-based chemotherapy.In response to the letter of AM Swart and MKB Parmar, we would like to make the following points:1. Our goal has never been to provide an exhaustive comparison of the four trials; instead, we wanted to highlight some differences between them that were not fully addressed in the two papers where MKB Parmar himself provided indirect comparisons of these same four trials! (1,2) We do, however, share his view that these comparisons can be misleading and are unlikely to shed light on the issue. 2. Contrary to the authors' statement, information on how many patients were given paclitaxel at relapse is given in table 3, and the number of patients who progressed or died is clearly mentioned at the bottom of figures 1 and 2. 3. In line with the authors' suggestion, we agree that re-analysis, by an independent body, of the original data from all four trials would probably be more informative than indirect cross-trial comparisons. 4. It would be interesting to see how the authors now explain the results of their ICON4 trial: in this trial, which enrolled 802 patients with relapsed ovarian cancer and a platinum-free interval of at least 6 months, paclitaxel plus platinum chemotherapy was found to improve survival and progressionfree survival compared with conventional platinumbased chemotherapy (3) . We know of no other clinical example of a drug (here, paclitaxel) being inactive in a primary disease setting and showing activity in a salvage disease one. 5. We profoundly disagree with the authors' conclusion that carboplatin alone can be considered as reasonable first-line treatment of women with advanced ovarian cancer. In our view, the longterm results of a not very large trial conducted in a rather homogeneous patient population and with tightly controlled protocol compliance, along with the strikingly convergent long-term results from a similar trial in the US and the recent results from ICON4, which conflict with those of ICON3, make it very difficult not to consider paclitaxel as an important drug for the upfront management of women with advanced ovarian cancer.It is time to put this sterile discussion to rest and to focus our energies on new treatment strategies for our patients, because their long-term prognosis remains largely unsatisfactory. References1 Buyse M, Burzykowski T, Parmar M et al. Using the expected survival to explain differences between the results of randomized trials: a case in advanced ovarian cancer. J Clin Oncol 2003;21:1682-7. 2 Sandercock J, Parmar MKB, Torri V, Qian W. First-line trea...

show abstract

supporting

confidence: 51%

mentioning

confidence: 92%

Response to the Letter

McGuire¹,

Piccart²

2004

Int J Gynecol Cancer

View full text Add to dashboard Cite

show abstract

“…Also, could cyclophosphamide adversely affect results by (a) prompting patients to drop out sooner because of toxicity and invoking occurrence of progression earlier and (b) affecting the response to subsequent therapy? Support for an adverse effect of cyclophorphamide has recently been published (34) .…”

Section: Comparing Results Across Decadesmentioning

confidence: 99%

Sequential single agents as first-line chemotherapy for ovarian cancer: a strategy derived from the results of GOG-132

Muggia¹

2003

Int J Gynecol Cancer

View full text Add to dashboard Cite

First-line chemotherapy for ovarian cancer during the past decade has evolved toward the use of carboplatin and paclitaxel combinations. This has been based on randomized trials showing that combinations of these two drugs lead to a outcome similar to that obtained using cisplatin and paclitaxel (that had, in turn, proven superior in progression-free survival and overall survival to cisplatin and cyclophosphamide) but with less toxicity. Surprisingly, taxane-platinum combinations were not superior to control arms in two studies (ICON3 and GOG-132) utilizing carboplatin or cisplatin as the comparators. This has renewed interest in the role of single agents in first-line chemotherapy - a concept also supported by a number of prior clinical trials with single-agent platinum compounds yielding results not inferior to combinations. Early 'pre-emptive' crossover (prior to the stipulated clinical progression) to paclitaxel or a paclitaxel-containing regimen, however, occurred in 24% of patients initially treated with cisplatin on GOG-132. This has led to the interpretation of this trial as a combination versus sequential design. Although not subscribing to this interpretation, the results of GOG-132 and ICON3 not only raise doubts over a clear superiority of combinations over single agents but also lead to a consideration of sequential treatment designs for first line. Advantages of such designs are: (a) ability to provide 'dose-dense' platinums followed by 'dose-dense' paclitaxel and, perhaps, other drugs; and (b) the potential of acquiring biological data linked to the antitumor effects of a specific drug. Mathematical modeling and recent positive results in breast cancer adjuvant therapy support the use of 'dose-dense' strategies, and these should be considered in the design of future trials in ovarian cancer.

show abstract

“…This difference in the background treatment could explain the difference between the results of the trial and of the meta-analysis. The availability of individual patient's data in both a meta-analysis and in an associated large trial may help to explain apparent discordant results: if the discordance is related to different patient characteristics, the discordance ought to disappear after controlling the treatment comparison for the effect of these characteristics [14].…”

mentioning

confidence: 99%