In the late 1990s, several chemotherapy regimens were in use in the treatment of advanced ovarian cancer. In the US, both cisplatin (P) alone and cyclophosphamide + cisplatin (CP) combinations were standard, although evidence from a meta-analysis of trials comparing CP with CAP (cyclophosphamide, doxorubicin, cisplatin) showed a survival benefit to the 3-drug combination . The US Gynaecological Oncology Group (GOG) used CP as the control arm in one trial evaluating a paclitaxel-based combination  and P as the control in another . The former trial showed striking benefits to the paclitaxel arm, which were at odds with the results of the European ICON3 trial . ICON3 compared a similar paclitaxel combination with a control group comprising either CAP or single agent carboplatin. This choice followed from the results of the ICON2 trial , which demonstrated the clinical equivalence of adequately dosed carboplatin and CAP. ICON3 did not find a benefit to the paclitaxel combination, and detailed analysis of the data from all the relevant trials identifies the most likely explanation for the conflicting results to be the inadequacy of the CP control arm.
variations which cannot conceivably have a large impact on outcome can and should be accommodated if a trial is to achieve widespread accrual. But can more substantial differences, completely different chemotherapy regimens for example, be accommodated in the design? Perhaps the first thing to consider is the weight of evidence in favour of 'equivalent efficacy.' Is there an opportunity to test this by including a randomization between the competing control arms? If equivalence has alreadybeen established through randomized trials, then is it necessary to distinguish between them at all? Given that the choice between the experimental arm and standard therapy may well be influenced by the very things influencing the choice of standard therapy (toxicity for example), then the answer is yes. If there is no good reason to believe the potential control therapies differ in efficacy, and randomizing between them is (for whatever reason) not appropriate, then it can be helpful to allow a choice of control arms, provided the choice is made before randomization, and randomization is stratified for that choice. One example of such a trial is the ICON3 trial in advanced ovarian cancer, described in Box 4.2. Effectively, this means that the trial can if necessary be considered as two separate trials. Indeed, it is appropriate, if not essential, to estimate the treatment effect separately in the two subgroups, and formally test for heterogeneity (Section 9.4.8) before calculating an overall estimate of the treatment effect.
Was this article helpful?
Learning About 10 Ways Fight Off Cancer Can Have Amazing Benefits For Your Life The Best Tips On How To Keep This Killer At Bay Discovering that you or a loved one has cancer can be utterly terrifying. All the same, once you comprehend the causes of cancer and learn how to reverse those causes, you or your loved one may have more than a fighting chance of beating out cancer.