My point regarding the median OS or PFS. The point I am trying to make is the difference between median and mean, particularly as it relates to the blockade inhibitor trials and the DCVax-L trial where there might be great efficacy, but only for less than half of the patients.
My understanding is that these trials usually specify median efficacy thresholds for approval. While mean is the average, median is the center number within the group of patients. Ie, if there were 401 patients, and you placed them in order of observed efficacy, the efficacy observed for the 201th patient would represent the median efficacy for the entire group. The actual efficacy for the other 400 patients would be of no bearing on the median.
One of the reasons that median is usually used to measure efficacy is because it could take 50 years to learn the mean, or average efficacy. As long as one patient is still alive, for OS, or has not progressed, for PFS, then those mean numbers continue to increase. Mean also unfairly weights any outlier patients.
But the median efficacy has it's own flaw. Take a group of 401 patients. If you list them in order of OS, then the 201th patient's OS would be the median OS. Now increase the OS for the top 200 patients by 10 fold. This would have no effect on the median OS. For that reason it is a horrible means of measuring efficacy, for the Blockade Inhibitors and possibly for DCVax-L if it too shows little or no efficacy for more than half of patients.
The clinicaltrials.gov listing for DCVax-L does not specify whether PFS and OS are to be measured using median or mean measures. Maybe the actual trial protocol does. I think Reefrad got hold of that protocol.
Clearly the FDA did not use median efficacy to evaluate the approved Blockade Inhibitors, and hopefully, if DCVax-L proves very effective for less than half of the patients, the FDA will also not use the median efficacy to determine pass or fail.
