Assumptions:
1. 50% of patients in TIVO-3 (161 patients) had received first dose of treatment by January 31, 2017. I think this is reasonable in light of (1) safety monitor review was conducted in January, 2017, (2) AVEO announced it was enrolling two months ahead of schedule in January 2017, and (3) futility study press release disclosed that 128 PFS events occurred on May 29, 2017.
Accordingly, I’m also assuming 50% of the patients would have been dosed after January 30, 2017.
2. At least 33% of patients taking TIVO will have a PFS event within the first nine months of treatment. For TIVO-1 in 1L RCC, it was about 40% by month nine, and so I think that assumption is safe.
3. Sorafenib’s performance will track the Dovitivinib study, with mPFS of 4 months and 5% PFS at 9 months.
4. There is a two month data reporting lag (demonstrated by the fact Aveo was alerted to the 128 PFS events for the futility study in early August 2017, but subsequent analysis showed the data cut off date was May 29, 2017 (nine week lag).
5. Final dosing occurred no later than July, 2017 (supported by June 2017 announcement of full trial enrollment). Accordingly, there would be at least 9 months of data for any patient dosed in July, 2017 (11 months to May 2018, minus the two month data lag discussed above).
Sorafenib Performance
Of the 80 patients dosed on or before January 30, 2017, the Dovitinib study tells me all 80 would have had PFS events by now.
Of the 81 patients dosed after January 30, 2017, I have assumed a PFS rate of 5%, meaning 77 have had PFS events.
So, with sorafenib, the PFS event total is 157.
Tivozanib Performance
I am assuming a 33% PFS response rate before month 9, which means of the 80 patients dosed on or before January 30, 2017, at least 27 would be PFS responders, and of the 81 dosed after January 30, 2017 but before July 30, 2017, at least 27 would be PFS responders.
That gives us a total of 54 PFS responders to add to the sorafenib total (157), for an aggregate amount of 211.
Analysis
Remember that we know 255 events have not occurred as of April 2018 (2 month lag), and it is 14 months from January 31, 2017 to April 30 2018.
If we assume none of the remaining 54 post-January 2017 dosers have not had a PFS event (the least favorable assumption to AVEO), that would mean at least 43 of the pre-January 2017 dosers have not had a PFS event, because 44 PFS events in that group would equal 255 (211 plus 44).
That would mean 43 of the original 80 pre-February 2017 dosers (53.75%) have mPFS of greater than 14 months, which means, for that 80 person group, mPFS is 14 mths or more. For every PFS event in the post-January 2017 dosers, the total of pre-February 2017 dosers must rise, further increasing the percentage of that group that have mPFS of 14 months.
If we assume two 80 person trials would have similar results, then over 50% of the TIVO arm has mPFS of 14 months.
