The data satisfied the Criterion #2. To satisfy the fourth , a cox model that includes both treatment identifier (provenge vs placebo), and the continuous variable which is the cumulative CD54 upregulation ratio. If the p-value associated with treatment become insignificant, then #4 would have been met.
If all four criteria are met, then argument for CD54 upregulation ratio as a surrogate endpoint will be very strong.
What's more important then is to actually submit CD54 data from 9902b (which is already available and summarized as box whisker plot by Provost), to use as supportive efficacy evidence, and to predict the overall survival expected in 9902b. That way, if there is any doubt in FDA's mind about the current efficacy package in the absence of 9902b TTP, FDA could look at the CD54 data, and say hey you guys did prove that CD54 upregulation ratio is a surrogate endpoint for survival, and since the 9902b upregulation is similar to that observed for 9901 and 9902a, then hey that's one less doubt.
I am so excited!