Study Preparation

AutoPilot Toolkit PK Automation projects require an NCA model, which is created in Phoenix before running AutoPilot Toolkit. The model requirements vary depending on the study design selected in the AP Automation object and the NCA model used. Model requirements fall into the following categories:

•Model Variables: Sort, X (time), Y (concentration), and Carry. Any variables that are to be used for stratification or normalization must be mapped to the Carry context in the NCA model's Main Mappings panel.

•Dosing Regimen

•Regression or Lambda Z

•Partial Areas

•Model Options

The NCA model omits a column from the output worksheet if the column has all missing values. For example, if the Lambda Z value was not calculable for any profile in the project, there will be no Lambda Z column in the NCA output. Consequently, AutoPilot Toolkit output will not include any Lambda Z data.

Note:AutoPilot Toolkit cannot use input data that is derived from a sparse dataset.

Study data must be prepared according to the standards defined by Certara as part of the AutoPilot Toolkit description. AutoPilot Toolkit does not modify the original data worksheets.

The topics in this section include:

Study data variables
Required data variables by study type
Support of data stacked by analyte

Study data variables

AutoPilot Toolkit analyses require that study data include specific variables for automation input. Once the study data variables are defined, the user must create a model that defines model settings and parameters for an AutoPilot Toolkit automation project. Specific model requirements are given in the previous section.

The variables fall into the following categories:

•Subject

•Dosing

•Sample Collection Point (SCP)

•Data Collection Point (DCP)

•Demographic

The following tables list the default variables for an AutoPilot Toolkit-ready study, including required variables and additional variables that are often useful. See “Required data variables by study type” for a listing of required and optional variables for each study type.

•Phoenix name: Required column name in Phoenix

•Display name: Default column name used in final AutoPilot Toolkit output.

Note:AutoPilot Toolkit requires that the case of column names match the case in the user’s data. Otherwise, it can create differences in the output. For example, mapping the variable Gender to the user’s data column GENDER may place male data before female data in graphs stratified by gender.

Default subject variables
Phoenix Name	Display Name	Units	Precision	Restrictions	Comments
Subject	Subject	NA	No	Alphanumeric	Patient (subject) identifier

Default dosing variables
Phoenix Name	Display Name	Units	Precision	Restrictions	Comments
Dose	Dose	Yes	dec/0	Numeric	Dose administered

Default Sample Collection Point (SCP) variables
Phoenix Name	Display Name	Units	Precision	Restrictions	Comments
Relative_Nominal_Time	Nominal Time	Yes	dec/2	Numeric	Protocol/nominal time of sampling since last dose. Used for Time, Time-Conc, and Cumul AUC output and possibly X-var in model.
Relative_Actual_Time	Actual Time	Yes	dec/2	Numeric	Actual time of sampling since last dose. Used for Time, Time-Conc output. Often used for the X-var in model.
Relative_Nominal_End_Time	Nominal End Time	Yes	dec/2	Numeric	Two purposes: Observation sheet is nominal interval end time from last dose of sampling for urine (Upper Time in model) and Dosing sheet is nominal end time for infusion.
Relative_Actual_End_Time	Actual End Time	Yes	dec/2	Numeric	Two purposes: Observation sheet is actual interval end time from last dose of sampling for urine (Upper Time in model) and Dosing sheet is actual end time for infusion.
[Matrix]_[AnalyteID]_[Route]_^a	[Matrix] (AnalyteID) Concentration^a	Yes	sig/3	Numeric or identified as Missing	Concentration of sample collected. Every column name using this template is identified as a concentration column, e.g., [Matrix]_(AnalyteID)_RawCONC.
Volume	Volume	Yes	dec/0	Numeric	Sample collection volume (Required for Urine Models 210–212 only)
Midpoint	Midpoint Time	Yes	dec/2	Numeric	Calculated time point that is equidistant between the Lower and Upper collection times of a given urine collection interval.
Rate	Rate	Yes	sig/3	Numeric	Excretion rate for each interval (amount eliminated per unit of time) = (Concentration*Volume) / (Ending time – Starting time).
Amount_Urine	Amount Urine	Yes	sig/3	Volume	Concentration*Volume.

^a[Matrix] is replaced with the matrix value from the study data. (AnalyteID) is replaced with the Analyte ID from the study data. An administrator can configure the concentration columns regarding name, use, and order of Matrix and AnalyteID.

When running stacked data, Analyte is not to be included in the concentration template.

Default demographic variables
Phoenix Name	Display Name	Units	Precision	Restrictions	Comments
Discrete Demographic
Sequence ^a	Sequence	NA	NA	Discrete, alphanumeric	Sequence of treatments received (randomized crossover studies only)
Gender	Gender	NA	NA	Discrete, values: male, female	Subject sex
Race	Race	NA	NA	Discrete, alpha. e.g., “Caucasian”	Subject ethnicity
Smoke	Smoke	NA	NA	Discrete, values: yes/no	Subject smoking status
Genotype	Genotype	NA	NA	Discrete, alphanumeric, e.g., “CYP2D6 Extensive”	Subject Baseline genotype status
Child_Pugh	Child Pugh	NA	NA	Discrete variable	Subject Child Pugh classification
Alcohol	Alcohol	NA	NA	Discrete, values: yes/no	Subject status: consumes alcohol or not
Continuous Demographic
Age	Age	year	dec/0	Continuous, numeric	Subject age
Wgt	Weight	kg	dec/1	Continuous, numeric	Subject body weight at screening
Height	Height	cm	dec/0	Continuous, numeric	Subject height
BMI	BMI	kg/m2	dec/1	Continuous, numeric	Subject Body Mass Index
LBW	LBW	kg	dec/1	Continuous, numeric	Subject Lean Body Weight
BSA	BSA	m2	dec/2	Continuous, numeric	Subject Body Surface Area
CrCL	CrCL	mL/min	sig/3	Continuous, numeric	Subject Baseline Creatinine Clearance

^aRequired for randomized crossover study designs to conduct inferential statistics.

An Administrator can add variables by providing the name, category, and data restrictions and map them to the AutoPilot Toolkit system variables. Associated units from these added variables is taken directly from the column headers in the study data.

Required data variables by study type

The following table identifies the required versus optional data variables, by study type and matrix (plasma or urine).

Req: Required

Opt: Optional

Variable requirements for plasma data (NCA Models 200 –202) and urine data (NCA Models 210 –212)
Variable	Study Design
Variable	*RCT	**NRCT	Replicated	Parallel	Trough (All)
Analyte ^a	Req	Req	Req	Req	Req
Subject	Req	Req	Req	Req	Req
Treatment_Description	Req	Req	Req	Req	Req
Period	Req	Opt	Req	Opt	Req for replicated, else Opt
Day	Req	Req	Req	Req	Req
RAT	Opt	Opt	Opt	Opt	Opt
RNT	Req	Req	Req	Req	Req
RAET	Opt	Opt	Opt	Opt	Opt
RNET	Opt (Plasma) Req (Urine)	Opt (Plasma) Req (Urine)	Opt (Plasma) Req (Urine)	Opt (Plasma) Req (Urine)	Opt (Plasma) Req (Urine)
XX_Dose	Req	Req	Req	Req	Req
Concentration variable ^b	Req	Req	Req	Req	Req
Volume	Req	Req	Req	Req	Req
Sequence	Req	Opt	Opt	Opt	Opt
Demogs (Categorical)	Opt	Opt	Opt	Opt	Opt
Demogs (Continuous)	Opt	Opt	Opt	Opt	Opt
Midpoint_Time	Opt	Opt	Opt	Opt	Opt

^aAnalyte required for stacked data only.

^bEither XX_RawCONC or XX_PKCONC is required. What exactly is required depends on the base of the concentration template.

* RCT: Randomized Crossover Trial

** NRCT: Non-randomized Crossover Trial

Support of data stacked by analyte

Stacked data theory
Other requirements and supported options

Stacked data theory

The simplest datasets include data from a single concentration assayed from a single source, or matrix, and resulting from the drug being administered by a single route. Such data might have only time and concentration fields.

When clinical drug studies are run, researchers often gather a multitude of data from analyzing both blood and urine samples. The drug is often given in pill form during one part of the study, and it is given intravenously in another. The study may only analyze blood and urine samples for concentration amounts of the drug itself. Typically, however, concentrations of other chemical entities that exist in the body must be analyzed as well. Therefore, instead of a simple concentration being examined from one matrix as a result of a single route of administration, the data can consist of concentrations from a variety of analytes, found in multiple matrices, and resulting from a variety of administration routes.

The layout of such data varies. The data layout can be described as stacked, unstacked, or partially stacked. The levels of data stacking are listed under “Variable assignments”. In AutoPilot Toolkit, there are three fields that determine how a dataset is stacked.

•Matrix: plasma or urine

•Analyte: drug given or other chemical entities

•Route: pill or intravenous

Using stacked data in AutoPilot Toolkit

In AutoPilot Toolkit, all of the analytes can be processed during a single run if the data is in a format such that all analyte concentration data is found in a single column. A new column, called Analyte, must contain the text names of the analytes themselves. These text values are similar to previous individual column headers in an unstacked dataset.

Refer to “Concentration Variable Template Selection tab” for additional details on how to setup the Administrator Tool to support stacked analyte data.

Variable assignments

The tables below detail the NCA model requirements for plasma and urine matrices, respectively, for different PK Automation study designs. The Sort Variables in the model must be ordered as they are presented in the table below. For example, Subject then Treatment_Description.

Note:Trough analyses do not require an NCA model.

NCA requirements for plasma data (Models 200–202) and
urine data models 210–212)
Variable	Automation Study Design
	Crossover			Parallel
	RCT	non-RCT	Replicated	Parallel
Sort Variables
Analyte ^a	X	X	X	X
Subject	X	X	X	X
Treatment_Description	X	X
Period ^b			X
Day ^c	X	X	X	X
Lower Times
Relative_Actual_Time or Relative_Nominal_Time	X	X	X	X
Upper Times
Relative_Actual_End_Time or Relative_Nominal_End_Time	X	X	X	X
Volume
Volume	X	X	X	X
Concentration
Concentration variable	X	X	X	X
Carry-Alongs
Sequence ^d	X
Period^b	X
Treatment_Description			X	X

^aRequired for stacked data only.

^bThe variable Period is required as a Sort Variable for a Crossover-Replicated study design or as a Carry-Along variable for a Crossover-Randomized study that includes inferential statistics. Failure to include Period as a variable in your study data will result in the use of a parallel model for the calculation of the inferential statistics in the case of these Crossover-Randomized studies.

^cThe variable Day is needed as a sort variable only when the study has multiple full-profile days. If Day is not selected as a Sort Variable or Carry-Along for single or multiple dose studies, then AutoPilot Toolkit automatically creates a Day column and sets all day values to “1”. For this reason, if the study has Day values other than “1”, Day must be included as a Carry-Along if it is not a Sort key.

^dThe variable Sequence is needed as a Carry-Along variable only for a Randomized Crossover that includes inferential statistics.

Note:If there is a missing or blank Day value in a dataset, it is automatically set to “1”. This can result in discrepancies between the number of subjects in each sequence in InText PK Parameter and Lambda Z tables because the subject that has a missing or blank Day value, the subject will be counted in Day 1 of a study but not in the next Day of a study.

Other requirements and supported options

Dosing Regimen: AutoPilot Toolkit requires that, for stacked input data, dose data be specified in the NCA model.
Lambda Z: Turning off curve stripping is supported with stacked data. This option appears in the Lambda Z Ranges dialog (select Lambda Z Ranges in the NCA Model menu). When Disable curve stripping is selected, no Lambda Z selections are assigned. In this case, Phoenix does not calculate any PK Parameters that use the Lambda Z.
Partial AUCs: Partial AUC intervals can be defined in the NCA model for stacked data. AutoPilot Toolkit allows up to three partial AUC calculations per profile to be included in the output.