In oncology randomized controlled trials, offering patients the opportunity to cross over to treatment from the other arm at disease progression is a routine practice to address ethical issues (associated with placebo controls, for instance, or even when comparing two active treatments, such as many trials with target-therapy in 1st-line treatment of NSCLC). In this situation, it is common to justify that the intervention has a PFS benefit but not an OS benefit due to the crossover. On the other hand, some colleges say that “true good” treatments can impact in OS despite crossover and minor gain in PFS should not be considered relevant. How should we interpret data form these studies? How to avoid underestimate a real benefit? How to avoid overestimate it?
It's interesting question... I think 3 possible scenario exist. First - situation when time of experimental drug application matter, i.e. some patients in less effective control group died early due to progression and couldn't receive experimental drug after crossover. In this situation despite crossover we observe and PFS and OS benefit. Example - trastuzumab in breast cancer, BRAF inhibitors in melanoma etc. Second situation - when experimental drug more efficacious, but "the sum didn't change due to relocation of drugs", i.e. patients receive new drug benefit irrespectively of it application time (before or after control treatment). In this situation we observe only PFS, but not OS benefit. BUT... In this scenario OS usually much better, then in 'historic control' - i.e new drug extend OS, but we couldn't see it in crossover trial. Possible example - EGFR inhibitors in NSCLC. Third situation - new drug really extend only PFS, but didn't extend duration of patient's life (OS). Example - bevacizumab in breast cancer. Usually we can't clear distinguish second and third scenario. But it's fate, because usually trials without crossover are not ethical.
Hi Allan,
The following article by Nick Latimer could be usefull for you.
"Adjusting for the Confounding Effects of Treatment Switching-The BREAK-3 Trial: Dabrafenib Versus Dacarbazine". Oncologist 2015
http://www.ncbi.nlm.nih.gov/pubmed/26040620
It is a case-study example of a RCT in oncology where performing analyses with or without taking into account cross-over (treatment switch). It also includes all the usefull methodological references.
Hope it helps!
Kind regards,
Béranger
Béranger, thanks for mentioning our article. This is a big issue at the moment and lots of papers are being written about it. Some more useful references:
http://www.ncbi.nlm.nih.gov/pubmed/24449433
http://www.ncbi.nlm.nih.gov/pubmed/24595585
http://www.ncbi.nlm.nih.gov/pubmed/25893990
The last one is a short editorial which goes over the main issues and gives plenty of examples, mainly from an HTA perspective, although this is clearly also an issue for regulatory agencies.
Nick
In clinical research terms, a cross-over design is adopted to rule out any possible 'period effect' the trial medications may have on the clinical outcome. It is also expected to reduce inter-subject variability. A cross-over trial may be useful if the study employs subjective parameters so as to minimize any bias in assessment. This may not be so if objective measurements are used to assess the clinical outcome.
Further, cross-over trials may not be suitable in the case of acute conditions such as myocardial infarction or infections because the 'disease state' may not be present for the second treatment phase. They also may not be suitable for chronic conditions such as arthritis, bronchial asthma etc where the disease will still be present and in a similar state for both treatments!
A D Paul, thanks for your comment. This highlights an important issue: it's becoming more common to refer to the type of 'crossover' referred to by Allan as 'treatment switching', to avoid confusion with 'crossover trials', where typically each patient acts as their own control. Treatment switching as referred to here is (usually) where patients randomised to the control group are permitted to switch onto the experimental treatment at some point during follow-up - perhaps after an interim analysis that has suggested that the new treatment is beneficial.
I totally agree to N. Lattimer, and especially to the article interpretation of B. Lueza.
- Badges
- Science topic