Posting: # 19094
In a clinical trial study where multiple doses are done and evaluation of PK concentration is carried out by 2 different bioanalytical method having different cc range (Low and high) and analysis is being done in different lab (Low cc range in lab 1 and high CC range in lab 2). should we conduct cross validation between both method. If yes then how will be the design of the cross validation.
Edit: Please follow the Forum’s Policy. [Helmut]
Posting: # 19095
excellent topic to bring up, actually.
Strictly speaking the residuals in your model will be affected by the bioanalytical method, i.e. they are not necessarily IID within or between.
Having said that, there are so many other assumptions in this game that I believe that if both assays are individually passing the validation criteria then there is no big issue and I don't have worry to the extent that I would think it is in the patient's interest to have additional crossvalidation. But I cannot qualify it further, it is gut feeling only.
Possibly one could specify method as a fixed factor with two levels in the model. This could at least be said to (try to) take bioanalytical accuracy differences into account but not precision differences.
“(...) targeted cancer therapies will benefit fewer than 2 percent of the cancer patients they’re aimed at. That reality is often lost on consumers, who are being fed a steady diet of winning anecdotes about miracle cures.” New York Times (ed.), June 9, 2018.
Posting: # 19098
» In a clinical trial study where multiple doses are done…
Do I get you right – a high accumulation ratio leading to low concentrations after a single dose and high ones in steady state? If not, can you elaborate on the design and its purpose?
Cross validation (e.g., Deming regression, Bland–Altman plot) compares two methods in the same concentration range.
If you really want to go that way, you would have to:
PS: Personally I never came across such a case. In BE studies of widely spread doses (and hence, concentrations) we regularly used different calibration ranges. Was never a regulatory issue to accept the entire submission package.
On the other hand, in dose proportionality studies and SD/MD with high accumulation we always used a method covering the entire range. Was sometimes difficult (limited range of the MS, quenching in fluorescence detection: quadratic model, weighting 1/x², 1/y², or – better – 1/s²y). I would never use different methods and/or different labs in the same study.
The quality of responses received is directly proportional to the quality of the question asked. ☼