In the 2 GVAX Ph2 trials, the high dose subgroup had a median survival in asymptomatic AIPC of 34.9 and 35 months. GVAX is being compared to Taxotere. In the pivotal Ph3 Taxotere trial, the asymptomatic subgroup had a median survival of 23 months. The number of events (deaths) required for the interim look should be about 200 to 220. CEGE has stated that they expect this to occur in 1H08. The trial began enrollment in July 2004. Median refers to the point at which 50% of a group reaches a defined point.Since as in all randomized clinical trials there is an experimental arm and a control arm, if you assume a positive treatment effect, the overall median for both trial arms will be greater than the median for the control arm and less than the experimental arm median. Thus, if the final number of events required to open either of the 600 patient Vital or Vital 2 trials is 400 (the number of events required to end the DNDN Ph3 9902b trial is 360) and half of that is required to open the interim, twice the number of patients required to open the interim (2X200?) should be enrolled at least 23 months prior to the interim at a minimum, but actually even for a longer period to account for a slower event rate in 50% of enrollees who are in the experimental arm. Fast foward to 6/08, and that would suggest that in June of 06 or a bit earlier, 400 of the 600 patients in Vital 1 would have been enrolled at a point approximately two thirds through the period of total enrollment (7/04 to 7/07), which seems a bit optimistic on the enrollment front and a bit conservative on the treatment arm front. Then again, CEGE has the actual enrollment/treatment rate and we don't.If the PH2 high dose subgroup results are repeated in Vital 1, it would suggest a Hazard Ratio of over 2, which would make GVAX statistically significant (p=<0.05) at the interim, which depending on the allocated alpha (p=0.01?) possibly a basis for opening the data and filing a BLA. As a reference, DNDN's 225 patients in their 9901 and 9902a trials had 164 events with an integrated increase in median survival of 4.3 months, a Hazard Ratio of 1.5 and a Kaplan Meier p value of 0.011. A greater treatment effect combined with a greater number of events in Vital 1 could push the p value below 0.01 and increase the treatment effect Hazard Ratio above 2. DNDN is projecting that their 9902b interim will be reached midway through 2H08.
Interestingly, both CEGE and DNDN consider the allocated alpha (p value)at their interim points proprietary. This could be a very competitive situation for both companies. The general alpha allocation rule in Ph3 trials is to keep an interim look alpha extremely low so as to maximize the chance of success at the final point when more data will have matured, increasing the power of the trial to report a p value lower than what remains of the alpha after subtracting that allocated for the interim. In this case, there is a significant likelihood that the two competitive immunotherapies will be statistically significant (p=<0.05) at their interim points. However assume for discussion that both the GVAX Vital 1 and Provenge 9902b have a p value of 0.02 at the interim and less than 0.01 at their final opening. The company that allocated a p value greater than 0.02 at the interim would have a significant lead in receiving earlier FDA approval as compared to a competitor taking a much lower conservative allocation. It seems to me that this situation, especially wrt two survival trials, does not put the interests of terminal AIPC patients first. Possible mitigation of this situation could occur if the independent Data Safety Monitoring Boards in each trial were to end the trial beyond the interim but earlier than the projected final if maturing patient data made achieving the remaining alpha a virtual certainty. All JMHO.
This was posted on a CEGE MB in response to some comments there, but is of obvious relevance to DNDN as well.JMHO
