Most of the analysis I have seen is using 17 as the denominator - not the 20 number expected to be randomized into Arm D. The 17 are those receiving 1 or more treatments.
The drop rate in Arm D was much higher than in Prove 1 by the way, 4/17 vs. 11% overall.
It also appears there may be some failures with SVRs judging from the response at 12 weeks measured on an ITT-basis vs. LOC-basis. The LOC basis is significantly higher and represents using the last measurement for droppers.
Market Data 
Markets 
AI
