Wednesday, October 06, 2010

German CLL8 - part 6

OK. So what about secondary endpoints? The important one for this trial is the difference in overall survival between chemotherapy and immunochemotherapy.

As you can see the curves are separating quite widely so that by 5 years follow-up of those who received FC 62% are still alive, whereas, of those who received FCR 75% are still alive. Unfortunately, there are very few patients who have been followed up for 5 years, so down at the right hand end of the graph the figures are not statistically significant - they could have occurred by chance. We are only going to see how big the difference between the two treatments as the trial matures. Indeed, if we go out even further, the lines actually cross, but the numbers are so small at this level, that the crossing is meaningless. The statisticians are only satisfied to report the position at a median of 3 years follow-up. Overall survival 3 years post randomization: FCR: 87.2% FC: 82.5% n=817, HR 0.664, p=0.012. These are the figures that Chris has been giving in his postings. It is a less than 5% difference, to be sure, but the point is that we are sure that it is a real difference, not one that could have happened by chance. Just by eyeballing the graph, we can be pretty sure that the difference will grow as time passes. But statisticians are creatures of exactitude and they won't allow us to say so until it is mathematically proven.

Let's just emphasize this point. For the first time with CLL a treatment has been shown to improve overall survival. Prior to this trial one could easily say that a treatment might give longer remissions, but if you are happy with less severe treatments given more frequently, then it doesn't matter which treatment you receive first. Three 12 month remissions are as good as two 18 month remissions. Now it is not possible to say that. If you go with FC rather than FCR as first line treatment then on average you waon't live as long.

Of course, patients want to know a great deal more than that. If for example you were absolutely sure that you would live longer with FCR than FC, but only 5 days longer, then I guess there wouldn't be many takers for a more aggressive treatment. So we do need to know some more details.

Here's one: The time to 25% of patients dying with FCR was 62.5 months, but with FC it was only 46.8 months. That is the quarter of patients who actually do worst with either of these regimes get an extra 15 months with FCR - and again that is statistically significant. So despite what Chris has been saying, this trial is holding out hope for improvement for a lot of people.

11 comments:

Unknown said...

Are we allowed to extrapolate and say that oral FCR is equally effective?

Terry Hamblin said...

I think so. The evidence from LRF CLL4 is that oral FC is equivalent to iv FC.

Burke said...

Doc,
I notice that both graphs flatlined at about 56 or 57 mos, except for the "meaningless" dropoff for FCR. Is this significant in any way?

Terry Hamblin said...

Not yet.

Anonymous said...

Just wondering if the results of this trial are stratified by chomosomal mutations (in particular trisomy 12),or other markers (high CD38 etc).

Terry Hamblin said...

Yes. Still more information to come.

~chris said...

>> So despite what Chris has been saying, this trial is holding out hope for improvement for a lot of people.


I have never stated this...FCR certainly give some hope.

~chris

Terry Hamblin said...

But you have been playing down the benefits - only a few percentage points difference in OS. Likewise Scott has been over-emphasizing the demerits of FCR. I was equally opposed to universal FCR before this paper, which has satisfied at least some of my doubts.

~chris said...

I would like to remind you of two statements from the study.

"These results suggest that a molecularly guided treatment approach could be useful for patients with chronic lymphocytic leukaemia; those with del(17p) might not benefit substantially from chemoimmunotherapy with fludarabine, cyclophosphamide, and rituximab."

and

" However, the patients’ median age of 61 years was much younger than that of the average population of patients with chronic lymphocytic leukaemia (with a median age at disease onset of about 70 years).

Therefore, the population in this trial represents a selection of fairly young and physically fit patients. As a consequence, conclusions from this trial should not be generalised to physically unfit, elderly patients with chronic lymphocytic leukaemia."

How universal does this make it?

Terry Hamblin said...

You anticipate my next blog which will be a review of Stephan Stilgenbauer's article in Haematologica, which makes the same points.

~chris said...

Sorry what article?

Biological diversity and risk-adapted treatment of CLL
Thorsten Zenz, Daniel Mertens, Stephan Stilgenbauer