Friday, May 29, 2015

ASCO: CR30 as PFS Surragate for FL Trials

I had every intention of using this past week to review some ASCO abstracts, and the week just got away from me. My oldest child is graduating from high school this weekend, my youngest is graduating from middle school, and work is no less crazy than it was in January. Life just gets in the way.

So let's take a look at an ASCO abstract, and find something to be hopeful about.

This first one is called "Evaluation of complete response rate at 30 months (CR30) as a surrogate for progression-free survival (PFS) in first-line follicular lymphoma (FL) studies: Results from the prospectively specified Follicular Lymphoma Analysis of Surrogacy Hypothesis (FLASH) analysis with individual patient data (IPD) of 3,837 patients (pts)."

It's kind of an odd one to start with, but it's a hopeful one.

Right now, if a clinical trial is looking at how well patients are responding to a treatment, the most common measure is PFS: Progression Free Survival. If we want to know how good a treatment is in comparison to other treatments, we compare their PFS -- how long it takes for the disease to return after a treatment has wiped it out (if only temporarily, which is usually the way it is for Follicular Lymphoma). Because many forms of Follicular Lymphoma are indolent, and grow slowly, PFS can take a long time to measure in some patients.

Waiting around for FL patients to hit PFS is a good thing --  it means that their disease hasn't returned. I'm sure it can be frustrating, too, because it means the data is incomplete. Perhaps there is some way to tell, statistically, if someone is going to reach a long PFS before they actually reach it?

So a very large number of smart people from cancer research centers all over the world looked at data from 13 different trials involving Follicular Lymphoma patients. Some of the trials involved Rituxan, and some didn't. Some looked at Maintenance. Some were first treatments. In other words, a lot of different situations. Overall, they looked at data from 3,837 patients -- a very large number.

Then they worked some statistical magic.They looked at all of the data from those 13 trials. In their words, they relied on "correlation of CR30 odds ratio (OR) with PFS hazard ratio (HR), using both linear regression (R2WLS) and copula bivariate (R2Copula) models. Prespecified criteria for CR30 surrogacy required either R2WLS or R2Copula ≥ 0.80 with a lower bound of the 95% confidence interval (CI) > 0.60, with neither estimate < 0.70. The minimum CR30 difference to predict significant PFS difference was calculated."

And the result? The data showed that "treatment effects on CR30 predict effects on PFS in pts with previously untreated FL. Multiple sensitivity and IPD surrogacy analyses supported the robustness of the primary analysis. A minimum 10% absolute improvement in CR30 over a control CR30 of 50% predicted significant improvement in PFS."

Now, what does all of that mean?

I have no idea. I'm a cancer nerd, not a statistics nerd.

But I sure understand the conclusion of the study. Statistically, counting the Complete Responses after 30 months can help predict Progression Free Survival for a trial.

The good news is this -- researchers can argue that a large number of patients in a trial with a CR at 30 months means the treatment is working, and maybe that means approval for that treatment can come faster than it otherwise would have. faster approvals means more potential treatments available for patients to try. And that matters when there are lots of treatments in the pipeline.

But let's all be very clear about what it does NOT mean -- a Complete Response for 30 months means nothing for an individual patient, other than that the patient has had a CR for 30 months. In other words, 30 months gives lots of reason to celebrate, but it also means that the disease can return in 30 months and 1 day. Just as with any statistical measure, it can predict a lot about large groups, but predict nothing about the individuals within that group. Treatment X might have a median effectiveness of 72 months. For some patients, it will go beyond that. For others, it won't work at all. Same thing applies here. Don't think a 30 month CR puts you in the clear.

Still, there is something worth celebrating here. This statistical magic just might get some treatments to us a little sooner than before. If it happens in a stage III trial, it means there have already been enough patients recruited and treated.

We'll have to see how this plays out with the people who decide these things.

No comments: