Re PDD Dose titration in OLE
Conclusion: Dose must be titrated up in order to establish a delta due to blinding.
Supporting summary:
Assume a 14 week trial. Assume a distribution of participants: one-third, one-third, one-third, with each distribution group receiving placebo (no active compound), 10mg active dose, and 20mg active dose, respectively.
After completion of the 14 week trial (which participants from each distribution group complete at different times), assume a rollover to some compassionate use “holding tank” is in use, per Polarbear’s research. As the overseer of such “holding tank”, in order to continue use for the active users and initiate use for the placebo group, I must give each participant some dose of active drug. However, I do not know what each participant had been taking prior to entering my jurisdiction (no unblinding at this point to preserve integrity of data).
I draw a marble randomly from one of 3 containers holding marbles each receiving different doses of “marble S1r agonist”.
What do I do?
First, as a patient enters, I do not know what dose he/she was on. Agreed? Therefore, since everyone MUST receive the drug, I begin administering a dose to each participant. Thus, the guarantee of receiving an active drug is fulfilled. At what dose do I begin?
My choices are 10mg or 20mg. Since I do not know what any patient was receiving before, I cannot begin the dosing lower than 20mg because if a participant from the 20mg bag were reduced to 10mg, that would be a downward adjustment and could lessen that person’s efficacy.
Assume that you start everyone on 20mg. This is reasonable (and only reasonable due to not wanting to downshift the 20mg patients for a possible year. 100% volunteer might see some (one-third) drop out rate if their dose were suddenly halved). This is also reasonable since one -third have already been exposed to 20mg for 14 weeks. Therefore, out of sheer ignorance of each patient’s dose, administer 20mg to everyone in the rollover period.
Now, you have: placebo participants receiving 20mg (highest dose and showing most improvement delta, if effective and neglecting placebo effect for now), 10mg cohort receiving a higher dose and showing an improved delta correlating to dose, and 20mg cohort showing no change.
This would appear on a graph as though the highest dose did not improve and the weakest/no dose improved the most. This is a dose correlation but is upside down.
During OLE, in order to get the correlation graph back to the correct meaningful accurate dose response, that would necessitate titrating the 20mg recipients, at least, to a higher dose. But, at this point, all participants would have been on 20mg for varying lengths of time and are unblinded so there would be no way to show effectiveness of drug from baseline unless you are able to titrate everyone up to higher ranges, and then emphasize less the statistical change (comparison of outcomes from groups which vary) and instead emphasize the effective difference (comparison of change for all from a change in input over time, i.e., increase in dose correlates to decrease in symptoms for all).
Thank you to Investor and all for input for the thought exercise. It could all come out in the wash or we just lose another sock in the dryer. (Life)
Making brains work on a Saturday??? That’s just not right! Time for a beer -
Biostock
(Just another opinion on the logic based speculation, fwiw)
