davidal, there are several aspects to your question.
1) 1st 33%, 2nd 50% and unblinding 100% of deaths of the TARGETED deaths
In the beginning 582 patients and about 67% of deaths was assumed as the # of TARGETED deaths for final unblinding. Since last CC we know that when PPHM reached 90% of the enrolled patients they had sufficient patients enrolled to finish the trial. The FDA will have set a # with PPHM which will have been based on the REQUIRED # patients by the FDA, even if PPHM decided to enrol some more.
We could therefore say that we don't take 67% of 582 but 67% of 90% of 582 as the # TARGETED deaths for unblinding. That is about 351 deaths to unblind. So 33% is 115 deaths and 50% is 175 deaths.
Randomizing was 1:1 so 291 Control arm patients and 291 Bavi 3mg/Kg patients that can deliver those deaths. From Herbst et al. 2010, the Docetaxel control arm that is closest to SUNRISE's type of ECOG enrolment, we learn that Docetaxel alone patients live about slightly less then 10 months.
PS: This is a sh..t text to write as my thoughts are with Krakonos and possible other board members of this board that have Cancer and must read this quite cold analytic approach were death is mentioned as a mathematical quantity as if it doesn't matter any further. It does!
2) Look-In goals and unblinding
The company (PPHM) was clear on the subject. 1st look-in now adjusted and planned for Q1/2016 with the goal to check for futility (is there any merit in continuing the clinical trial) and safety (didn't the clinical trial do more harm then good).
The 2nd look-in, while of course always also for safety, is about efficacy and is expected around mid-2016. A trial must reach end-points and it must be statistically significant, in layman's terms but not 100% correct but close - not be obtained by chance. Stat.sig is about whether this result can be consistently reproduced under the same conditions or not and the FDA requires a p-value of p>0.02 (in mathematics we use p>0.05 most often). This can NOT linearly be translated in a percentage (p=0.018 is NOT 100-1.8=98.2% sure, it is roughly about 92% from memory).
Final unblinding, if no early halt (see below) is expected for end 2016, officially December 2016 as per clinicaltrials.gov.
3) Early HALT
PPHM does NOT expect an early halt of the clinical trial for the 1st look-in (not for safety or futility problems but neither for efficacy).
They do not expect, but explicitly did not exclude, the possibility that an early halt for efficacy would be possible for the 2nd look-in.
Keep in mind that the IDMC only gives an advice and that PPHM is under NO obligation to early stop the clinical trial for efficacy. One motivation to NOT do this could be to work towards a stronger result and become SOC in 2nd ln NSCLC that is harder to beat. Opdivo (BMY) for instance runs alone against Docetaxel (no combo) and had better results but for a very small foot-print of patients (about 30%). Bavituximab has the ability to increase that foot-print but the SUNRISE results will very probably influence at what extend competitors will be/have to be motivated to run PD-1/PD-L1 combo's with Bavituximab to become SOC. AstraZenaca (Durvalumab) is the Bavi combo leader in the PD-L1 arena for now.
So an early halt at 2nd look-in is possible but there is no guarantee that PPHM will stop the trial before DEC 2016. Waiting 6 extra months could make them A LOT of money.
4) Study End
To stop a study early a p-value of p>0.02 may not cut it. The earlier you look the more severe p-value the FDA will require. We have NO information on what the requirements are. 115, 175 or 351 deaths on 524 to 582 patients doesn't seem difficult to achieve (involves 262 to 291 control arm Docetaxel alone patients). But since we have no idea when exactly each patient was enrolled we cannot apply Herbst et al, 2010 in a very reliable way to simulate (not that several members of this board didn't try). The results are always 3 to 6 month windows.
Conclusion
IMO it is best to just expect end of trial in DEC 2016 (that is in 11 months from now) in the worst case. All the rest is bonus. Each look-in will take away remaining risk of this trial now that the main risk for a small cap biotech (the ability to run and finance the trial within planned or acceptable limits) is gone because all needed patients are enrolled already. Futility, Safety and Efficacy are the 3 remaining risks (result/performance based). Nobody worries about safety or efficacy (all past trials on more the 500 patients were safe and our 2nd ln NSCLC PII was statistical significant and obtained 113% improvement on SOC while suffering a dose switching incident that was 25% in the disadvantage of Bavituximab. This happened at our 3rd party CRO, CSM in Fargo, ND because it was a registrational double blinded clinical trial too).
So theoretically stat. sig. efficacy shouldn't be a problem either BUT Bavituximab has NEVER BEFORE been tested on such a large sample as with the 582 SUNRISE patients. Therefore the SUNRISE results are the final peace of the puzzle that PPHM absolutely needs to take away all possible doubts.
Market Data 
Markets 
AI
