John1045, Thanks for the reminder.
This explains why NWBO designed their phase 3 trial the way they did. Only a 1/100,000 chance of the prior PFS results being random. What is also impressive is that there were 3 phase 1 trials upon which to gather this data. Moreover, since that time, NWBO designed their current phase 3 trial to exclude patients (placed in compassionate arm instead) that had progression after chemo/radiation.
Moreover, even though there was only a 4/10,000 chance the OS statistics were random, the crossover arm basically removes control group OS as an endpoint -- this wisely places all trial emphasis on PFS (only a 1/100,000 chance of previous results being random) where it should be (for scientific, ethical and reasonable development timeline concerns). The European regulatory authorities were historically slightly ahead of the U.S. in accepting PFS as a primary endpoint for cancer trials, but the FDA is now behind this metric as well. Finally, in the phase 3 trial, all patients with prior recurrent GBM were also excluded (placed in another study or compassionate program instead).
I still believe that while the data looks great, an investor must strive to understand the science behind it. This then might allow one to understand why somewhat similar companies with decent phase 1 results performed somewhat disappointedly on their later stage trials. While I understand the need to leave the science to the scientists, I stand by the premise that investing means delving into unchartered waters to find the pearls. For instance, if one strove to understand the scientific difference between ICT-107 and the DCVAX science, trial design and development strategy, one may have been drawn initially to one company more than another. IMUC may be a worthy investment in the long run, but based upon their science and previous data, they set themselves up for a setback. NWBO is still a risky investment, but from all outside appearances, they designed their current trial appropriately based upon scientific understanding (albeit incomplete) and prior data.
