The Creation of ADaM Datasets for Pharmacokinetic (PK) Analysis [Video]

In this recorded presentation a member of the Quanticate's Statistical Programming Team explores the creation of two ADaM datasets; ADPC and ADPP for Pharmacokinetic (PK) Analysis.

Video Transcript

"Hello everyone! Welcome to the presentation on creation of ADPC and ADPP datasets, the two ADaM datasets of PK analysis. I'm going to summarize my presentation as below. To start with, I'm going to give you a brief introduction to the two PK datasets PC and PP. Next, I'm going to take you through the PK Process Flow implemented at various stages of PK submission.

Then we will take a look at the creation of ADPC dataset followed by ADPP dataset. I have outlined some of the specific variables in each of these datasets which are very vital for the PK analysis. As you all know, FDA encourages the use of CDISC standards for the regulatory submission of clinical trial data. In CDISC, the SDTM domains are the source of raw datasets and the ADaM datasets of the analysis ready datasets derived from the SDTM domains. For the submission of PK data, we have two SDTM domains, the PC Domains for pharmacokinetic concentration, and the PP Domain for the pharmacokinetic parameters data. After the PK concentration data is collected in clinical database, it is converted to SDTM PP domain and the pharmacokinetic parameters derived by the pharmacokineticist are converted to SDTM PP Domain.

The analysis datasets ADPC is derived from SDTM PC and ADPP is derived from SDTM PP. The ADaM datasets contain both, collected data, as well as derived data to enable PK analysis, and for the creation of Tables, Listings, and Figures under PK results. These ADaM datasets are based on the basic data structure, the BDS structure of CDISC ADaM. Coming to the PK Process Flow, as you can see in the slide, the sample is drawn at the site, and from the collected sample, the concentration is measured and sent in the form of a CSV file. So we have the concentration data in the form of a CSV file, and on the other hand, we have the collected data in the form of the dataset. We merge both of them to create SDTM PC Domain.

The ADPC dataset is then derived from the SDTM PC Domain. When deriving ADPC dataset, the ADSL dataset which is nothing, but the subject level analysis dataset is merged to retrieve treatment group information and demographic information. This ADPC dataset can be directly used to create some of the concentration Tables, Listings, and Figures, and also, this dataset is sent as a CSV file for WNL PK Parameter derivation.

The summarized PK Parameters from pharmacokineticists are fed back to us, and this converted to SDTM PP Domain. The ADPP dataset is then derived from SDTM PP Domain with additional information from ADSL in the same way as it was done with ADPC dataset. Next, we will take a look in creation of ADPC dataset. CDISC ADaM do not have guidelines for pharmacokinetics so far, however the ADaM basic data structure, variables provide sufficient flexibility to support PK analysis. By working according to the ADaM rules for BDS, we can create a standardized PK dataset.

In order to create ADPC dataset, we merge PC Domain and EX domain which is nothing, but the exposure dataset to retrieve treatment data and time data, and also ADSL dataset is integrated with ADPC dataset to import general subject level information. Additionally, some derived variables are added to ADPC which has its own purpose like, handling of missing data, handling of values below threshold, and other sponsor specific data handling.

Now let us look into the structure of ADPC dataset. ADPC is used for the submission of all concentration results, and has the structure of one record per subject, per analyte, and per time point. Hence, concentration time provides us store data as multiple records. ADPC typically contains all relevant variables from SDTM PC like user's ID, VISIT information, and each time point within a particular VISIT.

The primary subject level data like treatment information from ADSL are imported to ADPC dataset, and some additional variable is required for the PK analysis, like the relative time variable, which is the relative time to the start-up dosing for each profile are added to the ADPC dataset. The actual concentration measured from PC Domain are rederived at ADPC with AVALC variable being the character value of concentration measure, and AVAL being the numeric equivalent. As an example here, the missing value of concentration measure is imputed to zero in AVAL variable. The qualifying records for analysis are flagged with the analysis flag variable ANL0FL.

Now let us look into some of the important variables which we created at the ADPC level. Those variables are highlighted here PARAM, PARAMN, PARAM CD, ADTM, ASTDTM, AENDTM, AVISIT, AVISITN, ATPT, ATPTN, ARELTM, ARELTMU, ANLzzFL, AVAL, AVALC, CRITy, CRITyFL, CRITyFN. Now let us look into each variables in detail. The PARAM variables contains the description of the analytes being analyzed in the pharmacokinetic concentration, and the abbreviation of the analyte is stored in PARAM CD.

PARAMN is the numeric counterpart of each analyte. The date, time variable, ADTM, ASTDTM, and AAMDTM, the date and time associated with the analysis value AVAL is stored in ADTM variable, and it's numeric version of the PCDTC variable from the PC Domain. ALDTTM and AENDTM are associated with start and end time of an analysis interval, example for urine collection.

The VISIT variables, AVISIT and its numeric counterpart, a AVISITN are derived from the variables VISIT and VISITNUM from PC Domain. All PK concentrations that refer to the same exposure will have the same AVISITN value. The plan time points are represented in ADPT and ATPTN variable. Only for pre-do values they differ from PCTPT, and PCTPTNUM variables for the PC Domain.

The value of PCTPTNUM which is in general negative for pre-do samples is put to zero in ATPTN variables in ADPC dataset. If the variable of PCTPTNUM from PC Domain contains the plan, turn points and minutes, it can be converted into hours in ATPTN variable in ADPC dataset. The relative time variables needed for the PK analysis is stored in the variable AAVAL time. If a relative time is calculated based upon the referenced time, EXSDDTC which is the start date and time of treatment and exposure domain and with the PC date and time, PCDTC.

This value will be negative for a pre-dose values and the unit of the relative time is stored in the variable ARL time view. The analysis record flax can be used to select that set of records for one or more analysis. ZZ represents an index for a record selection algorithm which can range from 01-09.

The analysis value in standard unit is recorded in AVAL variable, in most cases it's equal to PCSGRESN variabl from PC Domain. The character counterpart is recorded in AVALC variable. The AVAL variable is adjusted to handle various data handling techniques as described in protocol or rap. For example, values that are below the lower limit of quantification can be imputed to zero in the AVAL variable. The analysis criteria are evaluated in the CRITy flag varying wide ranges from zero to nine.

For an example, if more than 10% of the sample is drawn too late or too early from the planned time point, such values can be flagged with CRIT variables and can be excluded from the analysis. The outcome of the analysis criteria such as sample excluded yes or no, or presented in the CRITyFL and CRITyFN variables. The ADPC dataset can then be directly fed into the Phoenix WinNonlin software or alternative PK software packages to derive pharmacokinetic parameters, such as C-MAX, T-MAX, AUC, half-life etc. These parameters are subsequently used to populate SDTM PP Domain which is in turn merged with the ADSL dataset to create ADPP dataset.

The ADPP dataset is used in the statistical analysis of the PK data. The ADPP dataset has a structure of one record per subject per parameter. Similar to ADPC, we return relevant information from SDTM PP like USUBJID . The PK parameter variables like PP Tests, CDPP Tests and the result of each parameter in PPSTRESN and PPSTRESC variables along with its units in PPSTRESU variable.

All subject level data is imported from ADSL dataset. To be in accordance with CDISC ADaM guidelines, we have PARAM CD, PARAM variables, and AVAL, AVALC variables. Additionally, rederrived in ADPP dataset which holds the PK parameter description and its results. The below are some of the important variables within ADPP dataset.

The PARAM, PARAM, and PARAMCD, ANLZZFL, ANLZZFD, AVAL, AVALC, CRITy, CRITyFL, and CRITyFN. The value of PARAM is the PK parameter with its unit, for example CMAX, with its unit NGYNY, and the abbreviation of the PK parameter is stored in the PARAM CD variable. The numeric counterpart is presented in PARAMN variable. The qualifying records for analysis can be flagged with analysis flag variable, ANLZZFL where the zip can range from 0, 1 to 99.

The analysis value equal to PPSTRESN, which is the numerical result in standard unit from PP Domain is stored in the AVAL variable, and AVALC is the character counterpart. All PK parameters, any PK parameters can be included or excluded from the analysis based on the criteria as specified in the rap of protocol by using the CRIT flags, CRITy, CRITyFL and CRITyFN the y ranges from zero to nine.

Thank you for your attention. Thank you."