Noncompartmental analysis (NCA) of the pharmacokinetic (PK) characteristics of a drug is a common method used during drug development to investigate what happens to the drug in the body after dose administration.

The release of the ADNCA dataset, designed to follow the ADaM Basic Data Structure (BDS), will help close a critical gap in clinical data standards and drive readiness for NCA analysis. ADNCA dataset will ensure that data used for calculating PK parameters will be easier to review by regulatory authorities, contributing to a faster and smoother submission and review process.

Learn the answers to some key questions surrounding this new dataset from Senior Principal CDISC Consultant Kristin Kelly and PK Scientist Michael Clifton:

Q: Does P21E recognize ADNCA and perform conformance checks (validate)?

A: Yes, the CDISC Conformance Rules specific to ADNCA are implemented in P21E with the 2405.0 engine released in August 2024. Other standard ADaM checks will also be run.

Q: Is it “nice to have” or “necessary” to submit ADNCA dataset for regulatory submission?

A: No regulatory agency currently requires ADNCA but since the standard is available, it is highly recommended to review and implement it. This ensures standardization across the industry.

Q: If we have ADPC in a CRT package, do we need to have ADNCA in it? Or can we have either ADPC or ADNCA in the package?

A: ADNCA is a standard data structure that would replace ADPC. Only one or the other should be submitted. Also, the dataset does not need to be named ADNCA as long as the convention for naming ADaM datasets is followed, e.g. starts with 'AD' and is not longer than 8 characters. The dataset and variable metadata should be fully described in the define.xml.

Q: What is the difference between ADPC and ADNCA?

A: ADNCA is a standard data structure for submitting drug concentration data over time and contains much of the same data that would be contained in ADPC. Also, the dataset does not need to be named ADNCA as long as the convention for naming ADaM datasets is followed, e.g. starts with 'AD' and is not longer than 8 characters. A user can follow the ADaM-IG NCA guidance and is free to still name the dataset, ADPC.

Q: What's difference between ADNCA and ADPP (PP -> ADPP)?

A: ADNCA is a standard data structure for submitting drug concentration data over time and contains the input needed to calculate PK parameters contained in either PP or ADPP.

Q: What is the difference between ADNCA and ADADA ADaM datasets?

A: ADNCA provides the data structure for data needed for PK analysis. The dataset would be input into a tool such as WinNonLin to calculate the PK parameters. ADADA would contain results data from assays for anti-drug antibody analyses and not PK.

Q: Is it recommended ADPC is created to support PK TFL generation or can the ADNCA dataset be used for both NCA and PK TFLs?

A: The ADNCA dataset can used for NCA and PK TFLs thus, it would replace ADPC.

Q: Is the ADNCA file mandatory as part of the dataset package?

A: This webinar was an overview of the CDISC ADaM standard for NCA input data. It is currently not 'required' for any regulatory agency but it is encouraged to start submitting it.

Q: In ADNCA, how we can accommodate the imputed BLQ values in AVAL when the imputations rules are different for NCA analysis and PK Summaries (eg. For NCA analysis, BLQs before first measurable are imputed to 0 and the rest to missing. But for Summaries all BLQ's are imputed to 0.)?

A: In the ADaMIG-NCA, it is recommended to copy over the PCSTRESC variable from SDTM. This can be used to flag records in ADNCA that are BLQ, regardless of how the imputation is done for AVAL.

Q: Do you usually give an exclusion flag for the following:

1. For BQL values of PCSTRESN

2. For rescreened subjects

A: It would depend on how these situations are to be handled for a given study and outlined in the SAP on excluding records from the analysis.

Q: Is ADNCA acts like an input dataset to create SDTM PP?

A: Yes. ADNCA is the data structure that contains the input for SDTM PP and ADPP in ADaM.

Q: We may require weight, height & BMI at every dose. In this case do we need to create additional variables for this?

A: Yes, any baseline characteristics needed for the analysis should be brought over from ADSL. Due to the needs of analysis tools such as Phoenix WinNonLin, the units for each should be stored in a separate variable and not as part of the label.

Q: There could be some PK samples collected outside the window period. For these records, can we keep NCA exclusion flag as Y?

A: Yes, it would depend on the analysis rules for exclusion of records from the protocol/SAP.

Q: How should NERRLT and AERRLT populated for AUC(0-24) parameter?

A: NERRLT and AERRLT are variables in the ADNCA dataset. ADNCA is the input data needed into to calculate the PK parameters. There should be no PK parameter data in ADNCA. PK parameter results would be found in SDTM PP and/or ADPP in ADaM.

Q: Do we have any variable which gives information on Reference ID corresponding to a dose? This helps to pick the correct dose record from exposure to be linked to PC.

A: There is the --REFID variable that could be used in SDTM if that variable was populated. It could be copied over directly to ADNCA from the SDTM datasets.

Q: Will this dataset be used for NONMEM (Nonlinear Mixed Model analysis)?

A: NONMEM is for population PK modeling and ADNCA provides the data structure for input needed for NCA to calculate PK parameters. CDISC recently published a standard for population PK, ADaMIG-Population PK.

Q: Is my understanding correct that ADNCA is an ADaM dataset designed for NCA (calculation of PK parameters) and is not designed for creating tables/figures for CSR (although it is possible to create tables/figures for CSR)? Therefore, it is "Intermediate data" for SDTM.PP. Additionally, is it correct that ADNCA might be included in the e-Data submission package, but it is not mandatory?

A: Yes, that is correct. It is possible to create TLFs for PK concentration data but not for PK parameter data as ADNCA is the input data needed for the analysis. It is currently not 'required' for any regulatory agency, but it is encouraged to start submitting it.

Q: Are certain variables required in both numeric and character format, for example TRTA and TRTAN, APERIOD and APERIODC? Or how do you decide which variable to include?

A: There are several paired variables in the guidance such as NCAEXFL and NCAXFN, COHORT and COHORTN, etc. Please refer to the ADaMIG and the ADaMIG-NCA for further details.

Q: What is the preferred AVAL value for BLQ concentrations?

A: Typically, AVAL would be set to '0' for BLQ concentrations but would dedepend onhe information contained in the protocol and/or SAP.

Q: DOSEA and DOSEP are subject level dosing variables. How is the dosing handled if the dose of the study drug changes over time e.g. in a dose escalation or dose finding study?

A: The information on handling dosing in dose escalation or dose finding studies would be specific to that particular study and how it is handled would be outlined in the protocol and/or SAP.

Q: Should AVISIT in ADNCA match AVISIT in other ADaM for a study?

A: It would depend on the time window for the analysis. From the guidance:

'AVISIT should be unique for a given analysis visit window. In the event a record does not fall within any predefined analysis time-point window, AVISIT can be populated in any way that the producer chooses to indicate this fact (i.e., blank or “Not Windowed”). The way that AVISIT is calculated, including the variables used in its derivation, should be indicated in the variable metadata for AVISIT. The values and the rules for deriving AVISIT may be different for different parameters within the same dataset. Values of AVISIT are producer-defined and are often directly usable in clinical study report (CSR) displays.'

Q: How can one distinguish non-compartmental PK analysis and compartmental PK analysis looking at source data?

A: NCA vs compartmental analysis is being done would be outlined in the study's SAP. Since ADNCA provides the input data needed for NCA and not the results of the analysis, a user would not be able to readily determine what analysis was done. The PK parameter results are found in SDTM PP and/or ADPP in ADaM.

Q: Is there any example where DOSEA is not same as EX.EXDOSE?

A: Since EXDOSE/EXDOSU should be the units specified in the protocol, the numeric dose may need to be modified for the analysis in DOSEA and DOSEU in ADNCA.

Q: What about AVALC?

A: There's no mention of AVALC in the ADNCA document, and concentrations are typically quantitative measurements. Instead of putting BLQ values into AVALC, it is suggested to copy over PCSTRESC instead. That avoids using AVALC for parameters that are inherently numeric.

Q: Any recommendation on how to handle 24h post-dose samples collected and used both as 24h post-dose and pre-dose of Day 2? Should we duplicate the record? And how should it be flagged in ADNCA?

A: The situation you describe is provided as an example in the ADaMIG-NCA under the 'Duplicated records for Analysis' section (page 19). The record is duplicated, the timepoints updated to reflect that the sample is pre-dose on Day 2, and the duplicated record is flagged in DTYPE = COPY.

Q: There has been some chatter about removing PP from SDTM. Any updates?

A: Yes, it is currently planned to deprecate PP for SDTM in the next version of the SDTMIG v4.0. The ADaM team is currently working on a more streamlined solution to support PK parameter results only being submitted in ADaM. However, PP will be kept for SEND because ADaM is not used in nonclinical.

Q: Could data from ADaM datasets like ADEX be used as source for ADNCA? Would that be preferable?

A: It could if perhaps ADEX has analysis variables already derived (that are not found in SDTM EX) that are needed for ADNCA. Recommend to clarify the source for each variable in the define.xml.

Q: When will the ADNCA conformance rules be available in P21 Community?

A: The validation engine that contains the rules specific to ADNCA, 2405.0, is already available in P21E. The next release of P21C has not yet been confirmed but ADNCA rules would be included in that release.

Q: Can I use ADNCA for compartmental analysis?

A: There are data elements from the model that could be leveraged for compartmental analysis.

Q: Is it allowed to have derived records in ADNCA?

A: Since it is allowed in ADaM datasets, records could be derived based on the analysis need. There is an example of records being duplicated for the analysis need.

Q: So, we create ADaM (e.x., ADNCA) to be able to create SDTM PP later using external soft like Phoenix? In other words, we create SDTM PC, then ADNCA (and ADSL as well), then get back to SDTM PP, then work with ADPP. Seems like a bad circle?

A: Some of the data contained in ADNCA is coming from the SDTM datasets, typically PC and EX so there is traceability from SDTM to ADaM. ADNCA can be used to create both PP and ADPP, though it is convoluted to put the derived PK parameter data back into SDTM. There is discussion on deprecation of SDTM PP for the next version of the SDTMIG v4.0 and to only submit the PK parameter data in ADaM. Several sponsors have adopted this approach already.

Q: What is the use of AVALU? As far as I know in previous versions, units for AVAL had to be included in PARAM and no AVALU was to be created.

A: In the ADaMIG-NCA, AVALU has a 'Core' of 'Required' meaning it must be present in the ADaM dataset. Because this dataset is used as the input into PK analysis tools, one requirement of the tool is to have units in separate variables. Units can be part of the PARAM value but AVALU should also be present.

Q: Does the ADNCA file replace the PK merge file ie, must the PKist use the ADNCA file to calculate the parameters?

A: PK merge could be used as input for creating ADNCA and then used to calculate the PK parameters.

Q: If the dataset is not named ADNCA are the validation rules applied on a dataset with the NCA subclass?

A: Currently, the validation of the dataset is based on the ADNCA naming convention but it is planned in a near future release of the P21 engine that it will be based on subclass instead.

Q: It looks like ADNCA includes most information from Pkmerge or prewinnle. Would it be recommended to generate ADNCA from PKmerge?

A: Sounds like an intermediate dataset is being used to bring in all the relevant data for NCA from the SDTM datasets. This would be acceptable if it makes it easier to create ADNCA.

Q: On Slide 15, there is a typo in variable name PKSUMXF in the first column. It should be PKSUMXFL.

A: The variable name in the ADaMIG-NCA is 'PKSUMXF' and not 'PKSUMXFL', so the slide matches what is in the guide. If it is a typo, I would suggest reporting this to the CDISC ADaM team.

Q: Is it permissible to implement the same exclusion category system for PKSUM variables as for NCA variable (e.g., have PKSUMXRSN and PKSUMXRS)?

A: It is permissible to create additional variables for reasons for exclusion from the PK summary.

Q: Is AVALU required when we have units in PARAM?

A: In the ADaMIG-NCA, AVALU has a 'Core' of 'Required' meaning it must be present in the ADaM dataset. Because this dataset is used as the input into PK analysis tools, one requirement of the tool is to have units in separate variables. Units can be part of the PARAM value but AVALU should also be present.

Q: Today's presentation did not cover the section 'Duplicated Records for Analysis' in the guidance document. Will there be any P21 checks for that?

A: There are no duplicates records checks for ADaM in P21 since duplicating records for analysis is common practice.

Q: Is there an example that supports PD data from the LB Domain that may be included in the PK NCA Analysis?

A: In the ADaMIG-NCA, there is no specific example for PD data but it is stated that ADNCA can be leveraged to use for PD:

'The NCA dataset structure can also be leveraged for pharmacodynamics (PD) analysis where the primary dataset source is not PC. In that case it is assumed that the relevant traceability variables and domain references will be included in place of PC.'

Q: The biggest concern is that Biostats and Programming care about CDISC but many PK analysts do not. They could refuse to use ADNCA or manipulate the data in ways that make the implied input of ADNCA different from the actual NCA. How should Biostats and programming work with the Clinical Pharmacology groups to ensure they are both using ADNCA and setting expectations?

A: I would suggest meeting with them to align on this new standard and set expectations.

Q: Why is PCREFDTC (reference exposure datetime) required? What about studies that do not collect PCREFDTC? My understanding is last dose datetime can be derived using exposure.

A: PCRFTDTM is an ADaM variable and is required for ADNCA. PCRFTDTC in SDTM is collected data and does come from the EX domain. It needs to be merged with the concentration data to be mapped to PC so that the concentrations are associated with the relevant dose for the time-concentration curve.

Q: We have a variable NCAwXRS (reason for PK NCA exclusion) in ADNCAig. Do you think PKSMwXRS or equivalent for PK summary exclusion reason would be needed?

A: If there is a reason that the record is excluded from the PK summary that is different from the reason it is excluded from the NCA summary, an additional variable for PK summary exclusion reason can be added.

Q: I hear dataset names need not match the ADNCA. Would the same apply for Variable names that are not 'Required'?

A: I would suggest using the standard variable names if they are applicable so as to create consistency across studies.

Q: So, we could create ADNCA before Winnonlin and/or it could be created from Winnolin(with exclusion flag added)?

A: PK Submit only creates an ADNCA dataset after an analysis, where you will have all the flags needed to create the dataset - not before the analysis. However, if you have an ADNCA dataset available, you could use it to perform an analysis using PK Submit.

Q: Can WinNonlin PKSubmit create CDISC compliant SDTM.PP?

A: Yes.

Q: WinNonLin is not able to handle/read numeric SAS dates and times (date9. time5. or datetime formats) in SAS transport files or datasets. Is there is a specific solution to this issue?

A: PK Submit will read the data in as Text and format it to DateTime when needed. When PK Submit generates PC/PP domains, it puts it into the right format.

Q: WinNonLin exported datasets (not using PK submit) do not come with variable labels. Is there a solution to this issue?

A: When PK Submit generates the xpt files for the CDISC datasets it creates, it includes the labels for the variables.

Q: Is PK submit to be used by programmers? Or for PK Scientists only?

A: It is moslty for anyone familiar with Phoenix WinNonlin, which includes PK scientists but if a programmer is familiar with Phoenix, they may be able to use it as well.

Q: What is the difference between NONMEM software and WinNonLin?

A: NONMEM is for population modeling, which is unrelated to WinNonLin. It can be compared to PHX NLME, but NONMEM does not have a UI like NLME, so users must write scripts.

Q: Will PKsubmit be used for creation of ADPC and ADPP?

A: Yes, it can be used to create ADPC and ADPP.

Q: Is ADNCA generated by PK Submit used as input for WiNonlin-NCA analysis? I would like to know data flow from PC, PP and ADNCA.

A: It could be but for a separate analysis. The ADNCA dataset that PK Submit creates is done at the end of the analysis workflow.

Q: So is it that now PK Submit software will create the ADAM datasets ADPP, ADPC AND ADNCA?

A: Correct. When a customer uses PK Submit to generate results for an NCA analysis, there is an option to generate CDISC domains based on those results. When the user selects SDTM as the CDISC standard, it will generate both standard SDTM datasets (PC, PP, etc.) plus ADaM datasets ADPC, ADPP, and ADNCA.

Q: Does the PK Submit tool contain an audit trail for which data points have been modified (to provide traceability)?

A: PK Submit generates a log that tracks what actions the user performed and when they were changed, such as what BLQ rules were used, exclusions applied, variables created, etc.

Q: Is 'PK submit' a feature within WinNonlin?

A: PK Submit is a plugin for the Phoenix platform that works with WinNonlin. The user does have to purchase a license to use this plugin.

Q: Does PK Submit create ADNCA or use ADNCA as an input?

A: Both. If a user is provided an ADNCA dataset for their analysis, that user could use it to map to the data model in PK Submit to perform the NCA. PK Submit can also generate an ADNCA data set from an analysis the user performed. The dataset gets generated at the end of the analysis workflow, once final concentration results with flags have been generated. This is when the user performs the CDISC step to generate all pk-related CDISC domains for either SEND or SDTM/ADaM.

Q: How do we create ADNCA? Through programming, or using Phoenix WinNonlin?

A: Through Phoenix WinNonLin by utilizing PK Submit to map/merge source data and modify that data to apply approriate flags. Once that is done, it will create a Master Concentration worksheet that can be used to create the ADNCA dataset.

Q: Does the process of using the toolkit require a finalized ADSL as input? Implementing ADPC/ADPP usually identifies additional covariates you would want to add to ADSL.

A: It’s not required in PK Submit, but if baseline characteristics are needed for your NCA and you need those in your ADNCA dataset, then its best to have a completed ADSL and use that as a source file for PK Submit, mapping it to the data model and merging it with the rest of your source data.

Want more information on using Phoenix WinNonLin or PK Submit? Contact Customer Success to be routed to the right contact.