It has nothing to do with the selection criterion being "nefarious". For a randomized trial that is perfectly proper and always will be fine tuned for the treatment.
What does not follow is that one can compare such selections to other trials that are different. Even LL said that recently when she talked about matched controls for historical comparisons. But the DE/UK docs to not say matched controls, they say contemporary trials. And this board keep comparing to non-matched controls.
Your assertion that they are on par with others is between questionable an BS, depending on the trial. And then you always have the huge elephant in the Flaskworks lab. When they pulled rPD, were they more carefull about differentiating that from psPD? Long time bull, Regina, said they were. And the methylated numbers might support her. Add that to the intent for near total resection and steroid selection and you now can explain things.
Do these in reality explain the difference? I do not claim to know. But the burden of proof is not on me.
Market Data 
Markets 
AI
