Is Progression-Free Survival a Good Clinical Trial Endpoint?
In my last post, I mentioned that Synta had shown a doubling of progression-free survival in their STA-4783 phase 2b trial on melanomas. I thought I’d take a break from analyzing Synta and figure out if progression-free survival is really the best endpoint for them to be using (and is a doubling really all that great). I’m not a clinician nor do I have a background in running clinical trials, so if there’s anyone out there that wants to chime in, feel free to do so in the comments.
Here’s what InteliHealth reports as the definition of progression-free survival:
This term defines the length of time during and after treatment that the cancer does not grow. Progression-free survival includes the amount of time patients have experienced a complete response or a partial response, as well as the amount of time patients have experienced stable disease.
So basically progression-free survival measures how long it takes for the tumor (in this case) to start growing again. It seems like overall survival would be a better measure of a drugs effectiveness since curing the disease is the ultimate goal of medicine, but from the patients perspective, I’m not sure it’s all that bad of a measure. As I mentioned in the last post, the median survival for the late stage metastatic cancers is 6-9 months, so a doubling in progression-free survival means an extra few month of life.
This is what a graph of PFS looks like (this is from Synta’s phase 2 trial of STA-4783). The color circles are mine:
The X-axis is obviously days in the study. Progression-free Survival Probability could be labeled as “% of people in the study whose tumors started growing again” since that’s what it really is. The diamonds therefore represent the % of people in the study on any given day whose tumors hadn’t started growing. As you can see, all the people in the study eventually had tumors that started growing again. I circled the patients who seemed to have responded to STA-4783 in blue. Compare that part of the line to the part of the Paclitaxel alone I circled in red. The median progression free survival was 4.4 months for patients who received STA-4783 compared to 1.8 months with for those who didn’t. That’s a 58% increase, so that results in a hazard ratio of 0.42 (1-0.58=0.42). The P-value is a statistical test (2-sided log-rank test) which says that their is a 1.7% chance of this data occurring due to chance alone. We’d like to see that number below 5% (although the lower, the better), so it’s most likely that the difference is really due to the patients receiving the drug.
So, in my non-clinical, non-statistical expert opinion, it seems that PFS is a decent measure of a clinical trials end point. And Synta seems to have met that goal pretty well in their clinical trial.
Filed under: Evaluating Clinical Trials
[…] If you haven’t read it yet, I have a post on whether Progression-free survival is a good end point to measure for clinical trials. […]
good stuff
rwpq3fzgcpg15u8y
Hi, I found this blog post while was searching for online shopping related information on bing and found it very good article, thanks for sharing, best regards and cheers!
This article gave us a major Brainstorm session of all the probabilities we can make use of on our blog.
Apcalis sx….
Apcalis sx….
some truly grand work on behalf of the owner of this website , perfectly outstanding articles .
as soon as I found this site I went on reddit to share some of the love with them.
Dude where did you get your information? I need to know more about this, Great Stuff:}
Excellent analysis. Really not playing gotcha, but I’ve been trying to figure this out. You say, “so a doubling in progression-free survival means an extra few month of life.” Does making the tumor stop growing for a while necessarily extend life? Logically, I can imagine a scenario where the tumor stops growing because all the cells are metastisizing and killing the patient. Or, maybe it stops for a while, but when it starts back, it actually grows faster and kills the person. But, in reality, anyone know if stopping growth for a while always, or almost always, extends life?
Pretty good post. I just stumbled upon your blog and wanted to say that I
have really enjoyed reading your blog posts. Any way I will be subscribing to your
feed and I hope you post again soon. Thanks!
There are some fundamental misunderstandings here. PFS does not necessarily mean a life extension. Think of it this way: a drug may control a disease for a while, but then the patient go rapidly downhill and die. The patients not having the trial drug may have slowly growing disease throughout. So the two groups of patients may end up living for the same amount of time. Clearly having controlled disease is preferable - and you could argue that quality of life may be better. Except that all drugs have side effects and these can be pretty bad with cancer treatments, for example. So, when trying to assess benefit for a whole group of patients, or worse, trying to decide whether a treatment should be nationally funded, PFS may tell us very little. We could end up as a society paying a fortune for drugs that provide very little, if any, benefit. In order to state definitively that one end point is equivalent to (or in the jargon, a surrogate for) another, the basic research proving that needs to be done. And it hasn’t.