Sunday, October 03, 2010

German CLL8 trial 2

What sort of trial is the German CLL8 trial?

It is an open label, randomized phase III trial. First it is 'open label'. This means that both the patients and the doctors knew who was getting which regimen. The alternative would have been to control for this with placebo. Since one arm of the trial was getting rituximab and the other wasn't, it would be pretty hard to disguise this. They could have made up a dummy fluid that looked like rituximab and infused that, but rituximab tends to have side effects and the patients would soon have tumbled to which one they were having, and in any case the doctors would need to know who was having rituximab so that they could apply specific remedies for the side effects.

Second, it was randomized. Randomization is teh great discovery of clinical trials. Ideally patients in either arm of a trial should be very well matched. You can try and select patients that are well-matched, but unfortunately, you don't know in advance all the criteria that they should be matched under. For example, until 10 years ago you would never have known that each arm should have roughly equal numbers of cases with mutated and unmutated IGHV genes. However, if you have large enough numbers, and you allocate patients randomly to either arm, you usually end up with well-matched groups. You should check afterwards to see that this is so.

Third it was a phase III trial. This means that you aren't just looking for whether the disease responds to the particular drug combination, but for whether there is a difference in outcome for the patient. There have been previous phase II trials of FCR, as we know and we also knew that there have been pretty impressive responses, but the only way that we could look at outcome for those patients was to see what had happened in phase II trials of other combinations. Sure enough at MDACC, FCR seemed to out-perform FC or F alone, but over the years other things might have changed. The doctors may have got better, the supportive care might have got better and the type of patient might have changed, with richer, fitter patients following the reputation of the Institute. Randomization sorts out these sorts of biases.

Fourth, it was a multi-center study. Many hospitals in Germany were involved, but also centers in Austria, Belgium, Israel, France, Italy, the Czech Republic, New Zealand and Australia. This means that it covered the generality of cases across a wide area.

No comments: