I believe, it’s on the 9th: at least, this is what the Merck Webex says when you click “Join WebEx meeting”.
Good points of discussion!
Looking forward to the discussion.
Want to add one more point: when the PK analysis is performed in Phoenix, is there a need for ADPC and ADPP if the FDA agrees to accept the Phoenix project file as part of the CDISC package.
My views are outlined below.
How one can use SDTM (tabulations) while doing the analysis? In general, SDTM is for the listings, and ADaM is for the tables and, e.g., inferential stats analysis. SDTM does not contain information about administered treatment, unless it’s part of the PCGRPID, which is not necessarily the case even with the same Sponsor, imagine how unpredictable it becomes when dealing with the DM at many different Sponsors. And what if the additional grouping is required (imagine grouping by SEX, which is not part of the SDTM PC). As a data model, ADaM is designed specifically to be used for the purpose of the analysis. Hence, my vote goes for ADaM, both ADPC and ADPP.
Flags are a big topic. I believe, you are referring to the PKFL that is similar to ITTFL and SAFFL. These flags identify whether or not the subject is included in the specified population. The PK population is defined in the SAP, hence, it’s available prior to DB lock. This flag can be included in ADSL based on the SAP, same as the other population indicator variables, and ADSL does not need to wait for the availability of PP. In a very simple case we are dealing with two flags, PKFL that identifies the PK population (technically, a subset of FASFL or ITTFL) and then with the record-level ANL01FL, which either identifies the concentrations used for PK parameters calculations in ADPC, or records with PK parameters in ADPP that are used in the tables (descriptive stats) and inferential stats. This is really a simple case. More complex cases will require more complex flags, most of which, btw, can be still coded in a binary ANLzzFL.
PP is part of the SDTM portion of the CDISC package. All available data are tabulated in the SDTM datasets. How else would FDA know what are the values of the estimated PK parameters? Same as LB, EX, VS, etc., it has “collected” values, even though they are derived in WNL, Phoenix or other PK tools.
Same considerations are applicable to ADPC and ADPP: how else would FDA know that a particular AUCinf, even though tabulated in the PP, is not used in the analysis (tabulations or inferential stats) in cases when Rsq_adj was, say, 0.6, or %AUCExtrap > 20 (or whatever your SOP had defined as a threshold)? And, again, the grouping factors and covariates (weight groups, age groups, etc.) that are used in the further analysis are only available in the analysis datasets.
The analysis datasets should be created by whoever performs the analysis: if it is the Sponsor, then PC, ADPC, PP and ADPP are created by the Sponsor. If the Sponsor only provides PC (either non-complete, clinical, or full, merged with the concentration data) and a specialized company is analyzing the data (e.g., calculates and tabulates the PK parameters and performs the inferential stats), then creation of the PK/PD-related items of the CDISC submission package should be a responsibility of that specialized company.