2nd DRAFT for Review
1. Analyses and Displays Associated to NonCompartmental Pharmacokinetics – with a focus
on clinical trials
Draft - Version 0.2
Created 30 Jan 2014
A White Paper by the FDA/PhUSE Development of Standard Scripts for Analysis and
Programming Working Group
This white paper does not necessarily reflect the opinion of the institutions of those who
have contributed.
1
2nd DRAFT for Review
2. Table of Contents
Section
Page
1.
Analyses and Displays Associated to NonCompartmental Pharmacokinetics – with a focus on
clinical trials ........................................................................................................................1
2.
Table of Contents ....................................................................................................................2
3.
Revision History ......................................................................................................................4
4.
Purpose ....................................................................................................................................5
5.
Introduction .............................................................................................................................6
6. General Considerations ...........................................................................................................8
6.1. Reporting workflow............................................................................................................8
6.2. CDISC PK datasets creation workflow ..............................................................................9
7. Calculation of PK parameters ................................................................................................12
7.1. Main derived PK parameters ............................................................................................12
7.2. NCA Checklist..................................................................................................................16
7.2.1.
Missing sampling or concentration data ..................................................................17
7.2.2.
Concentration values below the limit of quantification ...........................................17
7.2.3.
Exclusion of outliers or influential data ...................................................................17
7.2.4.
Use of actual v.s. planned sampling timepoints .......................................................17
7.2.5.
Reporting of missing PK parameters .......................................................................18
8. PK Tables, Figures and Listings for Individual Studies ........................................................19
8.1. Standard List of Outputs ...................................................................................................19
8.2. Annotated PK TFLs ............................................................................................................1
8.3. PK TFLs Checklist .............................................................................................................1
8.3.1.
Individual data handling in listings ............................................................................1
8.3.2.
Individual plots ..........................................................................................................1
8.3.3.
Descriptive statistics in tables ....................................................................................2
8.3.3.1.
Statistics in the presence of BQL data .................................................................2
8.3.4.
Individual data handling in summary tables ..............................................................2
8.3.5.
Mean Plots..................................................................................................................3
8.3.6.
Formats for individual data and statistics ..................................................................3
9. Example SAP Language ..........................................................................................................4
9.1. Data to be analysed .............................................................................................................4
9.2. Pharmacokinetic methods ...................................................................................................4
10. References ...............................................................................................................................5
11. Acknowledgements .................................................................................................................6
2
2nd DRAFT for Review
List of Tables
Table 6-1 Symbols and definition of terms used in single and multiple dose NCA ......................12
Table 6-2. Main qualifiers for the determination of PK parameters. .............................................13
Table 6-3 Main formulas for calculation for PK parameters .........................................................14
List of Figures
Figure 6-1 Reporting workflow for pharmacokinetic data ..............................................................8
Figure 6-2 Process map for the creation of SDTM and ADaM PK datasets .................................10
Figure 8-1. Shell for individual PK concentration listing ................................................................1
Figure 8-2. Shell for individual PK concentration listing ................................................................3
Figure 8-3. Shell for overlaying PK concentration-time profiles ....................................................4
Figure 8-4. Shell for overlaying PK concentration-time profiles ....................................................6
Figure 8-5. Shell for overlaying PK concentration-time profiles ....................................................8
Figure 8-6. Shell for summary of PK parameters ..........................................................................10
Figure 8-7. Shell for summary of PK concentration ......................................................................12
3
2nd DRAFT for Review
3. Revision History
Version 1.0 was finalized xx XXXX 201x.
4
2nd DRAFT for Review
4. Purpose
Under CDISC, standards have been defined for data collection (CDASH), tabulation (SDTM),
and analysis (ADaM) datasets. The next step is to develop standard tables, figures and listings.
The Development of Standard Scripts for Analysis and Programming Working Group is leading
an effort to create several white papers providing recommended analyses and displays for
common measurements, and has developed a Script Repository as a place to store shared code.
The purpose of this white paper is to provide advice on displaying, summarizing, and/or
analyzing measures of pharmacokinetic (PK) data in clinical trials. The intent is to begin the
process of developing industry standards with respect to analysis and reporting for PK
concentrations and non-compartmental PK parameters that are common across clinical trials. In
particular, this white paper provides recommended processes for:



the calculation of PK parameters using non-compartmental analysis (NCA),
the production of PK listings, tables and figures for inclusion in clinical study reports,
and
the definition of statistical analysis plans (SAP) for PK data
Separate white papers address other types of data.
Model-based PK analyses are considered out-of-scope for this white paper.
This advice can be used when developing the analysis plan for individual clinical trials in which
PK data are of interest. Although the focus of this white paper pertains to clinical trials where
intense PK sampling is made, some of the content may apply to trials where only sparse samples
are collected. Similarly, although the focus of this white paper pertains to clinical trials, some of
the content may apply to pre-clinical studies where PK is being assessed.
Development of standard Tables, Figures, and Listings (TFLs) and associated analyses will lead
to improved standardization from collection through data storage. (You need to know how you
want to analyze and report results before finalizing how to collect and store data.) The
development of standard TFLs will also lead to improved product lifecycle management by
ensuring reviewers receive the desired analyses for the consistent and efficient evaluation of
patient safety and drug exposure. Although having standard TFLs is an ultimate goal, this white
paper reflects recommendations only and should not be interpreted as “required” by any
regulatory agency.
Detailed specifications for TFL development are in the scope of this white paper. The hope is
that code (utilizing SDTM and ADaM data structures) will be developed consistent with the
concepts outlined in this white paper, and placed in the publicly available FDA/PhUSE Standard
Scripts Repository.
5
2nd DRAFT for Review
5. Introduction
Industry standards have evolved over time for data collection (CDASH), observed data
(SDTM), and analysis datasets (ADaM). There is now recognition that the next step would be to
develop standard TFLs for common measurements across clinical trials and across therapeutic
areas. Some could argue that perhaps the industry should have started with creating standard
TFLs prior to creating standards for collection and data storage (consistent with end-in-mind
philosophy), however, having industry standards for data collection and analysis datasets
provides a good basis for creating standard TFLs.
The beginning of the effort leading to this white paper came from the FDA computational
statistics group (CBER and CDER). The FDA identified key priorities and teamed up with the
Pharmaceuticals Users Software Exhange (PhUSE) to tackle various challenges using
collaboration, crowd sourcing, and innovation (Rosario, et. al. 2012). The FDA and PhUSE
created several working groups to address a number of these challenges. The working group
titled “Development of Standard Scripts for Analysis and Programming” has led the
development of this white paper, along with the development of a platform for storing shared
code. Most contributors and reviewers of this white paper are industry statisticians, with input
from non-industry statisticians (e.g., FDA and academia) and industry and non-industry
clinicians. Hopefully additional input (e.g., other regulatory agencies) will be received for future
versions of this white paper.
There are several existing documents that contain suggested TFLs for PK measurements.
However, many of the documents are now relatively outdated, and generally lack sufficient
detail to be used as support for the entire standardization effort. Nevertheless, these documents
were used as a starting point in the development of this white paper. The documents include:












ICH E3: Structure and Content of Clinical Study Reports
ICH E7, Studies in Support of Special Populations: Geriatrics
US: Guideline for Industry: Structure and Content of Clinical Study Reports
US: Study and Evaluation of Gender Differences in the Clinical Evaluation of Drugs
US: General Considerations for Pediatric Pharmacokinetic Studies for Drugs and
Biological Products (draft)
US: Pharmacokinetics in Patients with Impaired Renal Function: Study Design, Data
Analysis and Impact on Dosing and Labeling
US: Pharmacokinetics in Patients with Hepatic Insufficiency: Study Design, Data
Analysis and Impact on Dosing and Labeling (draft)
US: In Vivo Metabolism/Drug Interactions Studies: Study Design, Data Analysis and
Recommendations for Dosing and Labeling (draft)
US: Population Pharmacokinetics
US: Exposure-Response Relationships: Study Design, Data Analysis, and Regulatory
Applications
Japan: Clinical Pharmacokinetic Studies of Pharmaceuticals
EU: Pharmacokinetic Studies in man
6
2nd DRAFT for Review



EU: Questions & Answers: Positions on specific questions addressed to the EWP
therapeutic subgroup on Pharmacokinetics
EU: Clinical Investigation of the Pharmacokinetics of Therapeutic Proteins
EU: Points to Consider on Pharmacokinetics and Pharmacodynamics in the Development
of Antibacterial Medicinal Products
These guidance documents present high-level requirements for the collection, analysis and
presentation of PK results in a variety of clinical trials. They do not provide, however, detailed
information that would enable to standardize the presentation of PK results. This white paper
tries to fill this gap and provides a set of standard rules and checklists to standardize the
production of PK TFLs in clinical trials.
7
2nd DRAFT for Review
6. General Considerations
6.1. Reporting workflow
The general workflow for the analysis and reporting of PK data in clinical trials involves two
major steps, as outlined in Figure 6-1:
1. Calculation of pharmacokinetic parameters
2. Production of PK TFLs
For each step, we shall define in subsequent sections, a checklist of standard rules that need to be
followed. The SDTM to ADaM mapping for PK concentrations (PC) and parameters (PP) will
also be discussed.
PK analysis checklist
PK Analysis
PK
Concentration
PK Parameters
PK Datasets
CDISC standard
SDTM PC/PP domains => ADaM
PK TFL checklist
Figure 6-1 Reporting workflow for pharmacokinetic data
8
2nd DRAFT for Review
6.2. CDISC PK datasets creation workflow
According the recommended CDISC process, SDTM Pharmacokinetic Concentration (PC) data,
SDTM Pharmacokinetic Parameter (PP) data, ADaM Pharmacokinetic Concentration (ADPC)
data and ADaM Pharmacokinetic Parameters (ADPP) are created based on SDTM/ADaM
structure data, clinical data, bioanalytical data and the derived PK parameters calculated by
scientists. Then, all the related listings, tables and figures can be generated based on ADPC and
ADPP data sets.
The general process for creating SDTM and ADaM PK-related datasets is summarized in Figure
6-2.
9
2nd DRAFT for Review
Structure
Clinical
Bioanalysis
Structure
PC
ADSL
Keys
Raw/CDASH
ADPC
Statistics
SDTM
NCA
PC TFLs
ADaM
Parameters
Analysis
PP
ADPP
Statistics
Output
PP TFLs
Figure 6-2 Process map for the creation of SDTM and ADaM PK datasets
It works as follows:
1. First, SDTM PC dataset is created based on SDTM structure dataset, clinical datasets,
and bioanalytical dataset.
2. Second, based on the ADaM structure dataset, SDTM PC dataset and ADaM
ADSL(Subject Level Analysis Dataset) are merged to create ADaM ADPC dataset.
ADaM ADPC dataset supports PK parameters calculation. It also provides information to
create PK concentration tables and figures.
3. Third, using specific software for non-compartmental analysis such as SAS or
WinNonlin, PK parameters are calculated from ADaM ADPC. A derived dataset is
created including all these calculated PK parameter information, and SDTM PP dataset is
created based on this dataset.
10
2nd DRAFT for Review
4. Fourth, ADaM ADPP dataset is created based ADaM structure dataset and SDTM PP.
ADPP dataset is the PK analysis dataset which is used for producing summary tables,
statistical tables, and any other PK analysis.
11
2nd DRAFT for Review
7. Calculation of PK parameters
7.1. Main derived PK parameters
In Table 7-1, we present the main PK parameters and terms used for the non-compartmental
analysis (NCA).
Table 7-1 Symbols and definition of terms used in single and multiple dose NCA
Symbol
Aa
Ae
Ae(t1-t2)
Ae
Af
Af(t1-t2)
At
AUC
AUC(0-t)
AUCextr
AUC(t1-t2)
AUC
AUMC
BLQ
Cav
Clast
CL
CL/F
CLCR
CLfm/F
CLNR
CLR
CLss/F
Cmax
Cmin
C(t)
Ctrough
D
F
Frel
fe
Definition
Total amount of drug excreted in expired air
Total amount of drug excreted in urine
Amount of drug excreted in urine from t1 to t2
Amount of drug excreted in urine over a dosing interval
Total amount of drug excreted in feces
Amount of drug excreted in feces from t1 to t2
Total amount of drug excreted in expired air, feces and urine
Area Under the Curve from 0 to infinity
Area under the curve from 0 to the time of the last quantifiable concentration
Extrapolated AUC
Partial Area Under the Curve between t1 and t2
Area Under the Curve over a dosing interval
Area Under the first Moment Curve from 0 to infinity
Below Limit of Quantification
Average concentration over a dosing interval
Last observed (quantifiable) concentration
Total body clearance
Apparent total body clearance
Creatinine clearance
Apparent Formation clearance of a metabolite
Non-Renal Clearance
Renal Clearance
Apparent Total body clearance at steady state
Maximum concentration
Minimum concentration over a dosing interval
Drug concentration at any time t
Measured concentration at the end of a dosing interval at steady state
Dose
Absolute bioavailability. F= fD x fA x fI x fH where fD, fA, fI and fH
represent the fraction dissolved, the fraction absorbed, the fraction
escaping intestinal first pass and the fraction escaping liver first
pass respectively
Relative bioavailability
Fraction of the dose excreted (urine by default, add qualifier
for other fluids)
12
2nd DRAFT for Review
z
LF
LOQ
MRT
PD
PK
PTF
R
Swing

Tinf
tlag
tmax
t½
Vss
Vur
Vz
Vz/F
CLHD
CLD
CLUF
E
Fr
QP
QUF
First order terminal elimination rate constant or
Apparent first order terminal elimination rate constant, for
compounds presenting release/absorption as limiting steps
Linearity Factor
Limit of Quantification
Mean Residence Time
Pharmacodynamic(s)
Pharmacokinetic(s)
Peak to trough fluctuation
Accumulation ratio
Percentage of swing
Dosing interval
Infusion duration
Time delay between drug administration and the first measurable
(quantifiable) concentration.
Time of Cmax
Terminal elimination half-life or
Apparent terminal elimination half-life, for compounds presenting
release/absorption as limiting steps
Volume of distribution at steady-state
Volume of urine
Volume of distribution
Apparent volume of distribution
Hemodialysis clearance
Dialysis clearance or dialysance
Ultrafiltration clearance
Extraction coefficient
Fractional removal
Plasma flow through the dialyzer
Ultrafiltration flow rate
Additional qualifier may be used when parameters need to be further defined for clarification
purpose. They are often inserted as subscript. By default, the matrix will in general be plasma. A
non-exhaustive list of qualifiers is presented in Table 7-2.
Table 7-2. Main qualifiers for the determination of PK parameters.
Matrices
bl
Blood
csf
Cerebrospinal fluid
fcs
Feces
mlk
Breast milk
p
Plasma
rbc
Red Blood Cells
sal
Saliva
ser
Serum
ur
Urine
Routes of administration
im
Intra-muscular
nas
Intra-nasal
iv
Intravenous
po
Per os
rec
Rectal
sc
Subcutaneous
sbl
Sublingual
top
Topical
13
2nd DRAFT for Review
Dosing regimen
ss
Steady-state
dayx
If multiple administration, this qualifier can be used to specify the day at
which the parameter is calculated
Binding
b
Bound
u
Unbound
The formulas used for the calculation of the main PK parameters by NCA are presented in Table
7-3, below.
Table 7-3 Main formulas for calculation for PK parameters
Parameters
Ae
Ae(t1-t2)
Determination
Single Dose
Ae 
C
Multiple Dose (if different)
ur
* Vur
Ae( t1  t 2 ) 
C
t2
ur
* Vur
t1
AUC
AUC(t1-t2)
AUC  AUC(0  t ) 
AUC( t 1  t 2 ) 
t2

t1
AUMC
AUMC(t1-t2)
(C( t 1 )  C( t 2 ))
* (t 2  t 1 )
2
AUMC  AUMC0  t  
t2
AUMC ( t 1  t 2 )  
t1
Cav
CL
C last
z
AUC

F*D
CL 
AUC
C av 
Clast * t last
z

Clast
2z
( t 1 * (C( t 1 )  t 2 * C( t 2 ))
* (t 2  t 1 )
2
AUC
C av 

F* D
CLss 
AUC
NB: after iv, F=1
CL/F
CLfm/F
D
AUC
Aemetabolite
D
/F
 fe *
AUC
AUC
CL / F 
CL fm
CLss / F 
D
AUC
Clast
Directly obtained from the observed concentration vs. time curves.
Cmax
Directly obtained from the observed concentration vs. time curves
Cmin
Directly obtained from the observed concentration vs. time curves.
CLNR
CL NR  CL  CL R
calculated only after iv or if F is known or if F is
explicitly assumed to be 1
14
2nd DRAFT for Review
CLR
Ctrough
F
Frel
Ae
AUC 
Directly obtained from the observed concentration vs. time curves.
AUCpo * D iv
AUC  po * D iv
F
F
AUCiv * D po
AUC  iv * D po
CL R 
Ae
AUC
Frel 
AUC po * Dref
CL R 
AUCref * D po
fe
fe 
LF
AUC 
AUC dose1
Estimated slope of the linear regression of ln concentration vs. time.
After single dose oral or iv bolus:
After multiple po or iv bolus:
z
MRT
Ae
D
LF 
MRT 
AUMC
AUC
MRT 
AUMC    (AUC  AUC)
AUC
PTF 
C max  C min
Cav
After infusion
MRT 
AUMC T inf

AUC
2
PTF
R
For mono-compartmental model:
1
R
1  e  Z * at steady-state
After multiple dose administration:
C max, ss
R max 
C max, dose1
1  e  n z 
R
1  e  z  at nth dose
R min 
Swing
R AUC 
Swing 
C min, ss
C min, dose1
AUC ss
AUC dose1
C max  C min
C min
tlag
Directly obtained from the observed concentrations
tmax
Directly obtained from the observed concentrations. If two identical values
are recorded for Cmax, the first one will be considered for tmax.
t½
t1/ 2 
Vss
Vss  MRT * CL 
ln 2
Z
F * D * AUMC
AUC2
15
2nd DRAFT for Review
Vz
Vz/F
CL
Z
CL / F
VZ / F 
Z
VZ 
NB: after iv, F=1
CLHD
CLD
CLUF
E
CL HD  CL D  CL UF
Ci  Co
where Ci is
Ci
the concentration entering the
dialyzer and Co is the concentration
getting out of the dialyzer
C
CL UF  Q UF o where Ci is the
Ci
concentration entering the dialyzer
and Co is the concentration getting
out of the dialyzer
CL D  Q PL
E
Ci  Co
Ci
where Ci is the
concentration entering the dialyzer
and Co is the concentration getting
out of the dialyzer
Fr
Fr 100 *
t1/ 2 (1)  t1/ 2 ( 2)
t1/ 2 (1)
 (1  e
  Z( 2 ) *t
)
where (1) refers to the period before
the start of the dialysis and (2) to the
period during the dialysis.
QPL
Q PL  blood flow * (1  hematocrit )
QUF
Ws  We  Wfl
where Ws
dialysis time
and We are the body weight at start
and end of dialysis, and Wfl is the
weight of the dialysate
Q UF 
7.2. NCA Checklist
A set of standard rules needs to be defined for the management of particular source data points in
NCA. These include the following:
•
Management of missing sampling or concentration data
•
Management of concentration values below the lower limit of quantification
16
2nd DRAFT for Review
•
Exclusion of outliers
•
Use of actual v.s. planned sampling timepoints
•
Reporting of missing PK parameters
We detail these rules in the following sub-sections.
7.2.1. Missing sampling or concentration data
Unless otherwise specified below, missing sampling or concentration values should not be
imputed and left missing in the calculation of derived PK parameters.
If the actual sampling time is missing but a valid concentration value has been measured, the
protocol time is generally used for the calculation of derived PK parameters.
A missing pre-dose value for single-dose study is usually replaced by 0 for the PK calculations.
7.2.2. Concentration values below the limit of quantification
For plasma concentrations, all BLQ (Below the Limit of Quantification (LOQ)) values occurring
prior to Cmax are replaced by “0” (i.e., for lag-time characterization), except for embedded BLQ
values (between two measurable data points) which are treated as missing. Post-Cmax BLQ values
are treated as missing.
For urine, when calculating individual amounts and cumulative amounts excreted, BLQ urine
levels are set to zero.
7.2.3. Exclusion of outliers
On a case by case basis, it may be necessary to exclude individual PK concentration values for
the calculation of derived PK parameters, because they are abnormal. Any excluded data should
be flagged in the individual data listings. If known, the reason for exclusion should also be
documented.
For chemical entities, it may be necessary to exclude a subject from all pharmacokinetic
evaluations if the pre-dose concentration is significantly non-null (a value larger than 5% of the
subject’s Cmax may be used as a threshold). For biological entities, predose concentration can be
included in all pharmacokinetic measurements and calculations.
7.2.4. Use of actual v.s. planned sampling timepoints
If possible, actual post-dose time should be used in calculation of PK parameters and in the
generation of individual concentration-time profiles.
Planned sampling times may be used for pre single-dose values and as a replacement for
unknown or missing actual times.
17
2nd DRAFT for Review
7.2.5. Reporting of missing PK parameters
The percentage of extrapolated AUC should not exceed 20 % for each individual profile. If the
percentage of extrapolated AUC is more than 20 %, the individual result should be flagged for
exclusion in the report, as well as the parameters depending on AUC.
Terminal half-life should be determined over a time interval equal to at least 2 x t½, using at
least 3 data points and with Adj_RSq2 should be greater or equal to 0.85. If at least one of these
three conditions is not fulfilled, the terminal half-life should be flagged for exclusion and
mentioned in the report, as well as the parameters depending on t½.
18
2nd DRAFT for Review
8. PK Tables, Figures and Listings for Individual Studies
8.1. Standard List of Outputs
In individual studies where PK data is collected, the following list of PK outputs are commonly
produced:
•
Listing of individual PK concentrations
•
Listing of individual PK parameters
•
Summary table of PK concentrations
•
Summary table of PK parameters
•
Figures for PK concentration-time profiles:
– Individual plots (separate and/or overlaying)
– Mean plots with or without error bars
In addition, statistical TFLs are created in trials where a statistical analysis of PK data is planned.
Section 8.2 provides illustrative shells for the main types of PK TFLs and Section 8.3 presents a
set of standard rules for the reporting of PK data in TFLs.
The proposed standard PK TFLs contain 3 parts: the title, body and footnote, that can be adapted
to match individual company standard or study-specific requirements.
In our standard template, the title part contains the following pieces of information:
1. Sponsor/Protocol/Product information, such as the name of the company, the protocol
numbar and or the compound name/code.
2. Listing/Table/Figure label to identify the type of output
3. The output number according to ICH E3 guidance document.
4. The output title
5. A page numbering indicator for the page number and the total number of pages.
6. The analysis population
The body part is broken up into two parts:
1. An optional headline defining the information displayed on any particular page (the bylines).
2. The actual output content presented in a tabular grid.
The footnote contains the following pieces of information:
1. The definition of all abbreviations
2. Annotations for flagged data values. Usually, flags are used for exclusion of individual
data.
19
2nd DRAFT for Review
3.
4.
5.
6.
A description of how BQL values are reported (optional).
Information about source dataset, program and output path
Information about data and program status (development/test/production)
Production date and time.
20
2nd DRAFT for Review
8.2. Annotated PK TFLs
SPONSOR/PROTOCOL/PRODUCT INFO
(page x of x)
Listing 16.2.5-x.x Individual PK parameters
by compound, matrix, analyte and [actual/randomised] [treatments/group]
Analysis Set : All subjects
Compound: XXX, Matrix: YYY, Analyte: ZZZ
[Actual/Randomised] [treatment/group] [sequence]: AAAAAA
Country/
Site/
Subject
Age/
Sex/
Race
CNTR/
ST1/
XXXXX
YY/
M/
Ca
Period
Profile
day
Parameter (unit)
1
1
AUCinf (hr*ng/mL)
xxx
AUClast (hr*ng/mL)
Tmax (hr)
xxx
Value
*
xx.x
- Value * was not considered for summary and inferential procedures.
- Age/Sex/Race: M=Male, F=Female, Ca=Caucasian, …
PATH DATA/PROGRAM/OUTPUT
PRODUCTION STATUS/RUN DMMMYYYY: HHMM
Figure 8-1. Shell for individual PK concentration listing
Annotations:


If the actual treatment is known and if it differs from the randomised treatment, it is recommended to report according to actual
treatment. Otherwise, the randomised treatment is reported.
In some particular multi-arm trials, all groups receive the same treatment but they differ by other characteristics (such as
gender, disease status or stage, administration with food, etc…). In these cases, data are reported according to the group and
not the treatment.
1
2nd DRAFT for Review


In sequential and cross-over trials, the sequence and period are reported. In parallel trials, these data are usually skipped. In
multi-part trials, the part is displayed either in the title, headline or column.
A footnote indicates how BQL values are reported.
2
2nd DRAFT for Review
SPONSOR/PROTOCOL/PRODUCT INFO
(page x of x)
Listing 16.2.5-x.x Pharmacokinetic concentration
by compound, matrix, analyte and [actual/randomised] [treatments/group]
Analysis Set : PK analysis set
Compound: XXX, Matrix: YYY, Analyte: ZZZ
[Actual/Randomised] [treatment/group] [sequence]: AAAAAA
Country/
Site/
Subject
Age/
Sex/
Race
CNTR/
ST1/
XXXXX
YY/
M/
Ca
Period
Profile
day
1
1
Scheduled
Sampling
Time
(uom)
0.5
Date/Time of
collection
Elapsed
Time
(uom)
Concentration
(uom)
2000-02-12Txx:xx
xx.x
xxx.xx *
1.0
2000-02-12Txx:xx
1.5
2000-02-12Txx:xx
- Value * was not considered for summary and inferential procedures.
- Values <LLOQ were considered as [zero;missing;LLOQ; LLOQ/Z;…].
- Age/Sex/Race: M=Male, F=Female, Ca=Caucasian, …
PATH DATA/PROGRAM/OUTPUT
PRODUCTION STATUS/RUN DMMMYYYY: HHMM
xx.x
xx.x
xxx.xx
xxx.xx
Figure 8-2. Shell for individual PK concentration listing
Annotations:




If the actual treatment is known and if it differs from the randomised treatment, it is recommended to report according to actual
treatment. Otherwise, the randomised treatment is reported.
In some particular multi-arm trials, all groups receive the same treatment but they differ by other characteristics (such as
gender, disease status or stage, administration with food, etc…). In these cases, data are reported according to the groups and
not the treatment.
In sequential and cross-over trials, the sequence and period are reported. In parallel trials, these data are skipped. In multi-part
trials, the part is displayed either in the title, headline or column.
A footnote indicates how BQL values are reported.
3
2nd DRAFT for Review
SPONSOR/PROTOCOL/PRODUCT INFO
(page x of x)
Figure 16.2.5-x.x Overlaying individual concentration-time profiles
by compound, matrix, analyte and [actual/randomised] [treatments/group]
Analysis Set : PK analysis set
Compound: XXX, Matrix: YYY, Analyte: ZZZ
[Actual/Randomised] [treatment/group] : AAAAAA
LLOQ
- Values <LLOQ were considered as [zero;missing;LLOQ; LLOQ/Z;…].
PATH DATA/PROGRAM/OUTPUT
PRODUCTION STATUS/RUN DMMMYYYY: HHMM
Figure 8-3. Shell for overlaying PK concentration-time profiles
Annotations:

If the actual treatment is known and if it differs from the randomised treatment, it is recommended to report according to actual
treatment. Otherwise, the randomised treatment is reported.
4
2nd DRAFT for Review


In some particular multi-arm trials, all groups receive the same treatment but they differ by other characteristics (such as
gender, disease status or stage, administration with food, etc…). In these cases, data are reported according to the groups and
not the treatment.
A footnote indicates how BQL values are reported.
5
2nd DRAFT for Review
SPONSOR/PROTOCOL/PRODUCT INFO
(page x of x)
Figure 16.2.5-x.x Individual concentration-time profiles
by compound, matrix, analyte and [actual/randomised] [treatments/group]
Analysis Set : PK analysis set
Compound: XXX, Matrix: YYY, Analyte: ZZZ
Country/Site/Subject: CNTR/ST1/XXXXX
LLOQ
- Values <LLOQ were considered as [zero;missing;LLOQ; LLOQ/Z;…].
PATH DATA/PROGRAM/OUTPUT
PRODUCTION STATUS/RUN DMMMYYYY: HHMM
Figure 8-4. Shell for overlaying PK concentration-time profiles
Annotations:

If the actual treatment is known and if it differs from the randomised treatment, it is recommended to report according to actual
treatment. Otherwise, the randomised treatment is reported.
6
2nd DRAFT for Review


In some particular multi-arm trials, all groups receive the same treatment but they differ by other characteristics (such as
gender, disease status or stage, administration with food, etc…). In these cases, data are reported according to the groups and
not the treatment.
A footnote indicates how BQL values are reported.
Notes to programmer:




Plot against actual time since last dosing, if possible. Otherwise, use protocol times.
For multiple dose trials, display the entire time course or split into different panels by dosing occasion, as most appropriate
For multi-period trials, display overlay treatments separately for each analyte
Scale of Y axis may be either identical across all subjects or subject-specific, as most relevant.
7
2nd DRAFT for Review
SPONSOR/PROTOCOL/PRODUCT INFO
(page x of x)
Figure 14.2-x.x [Arithmetic/Geometric] mean (SD) concentration-time plot per treatment (overlaying)
and analyte (separately)
Analysis Set : PK analysis set
LLOQ
- Values <LLOQ were considered as [zero;missing;LLOQ; LLOQ/Z;…].
PATH DATA/PROGRAM/OUTPUT
PRODUCTION STATUS/RUN DMMMYYYY: HHMM
Figure 8-5. Shell for overlaying PK concentration-time profiles
Annotations:

It is customary to produce arithmetic mean plots for PK data. Given the
8
2nd DRAFT for Review


In some particular multi-arm trials, all groups receive the same treatment but they differ by other characteristics (such as
gender, disease status or stage, administration with food, etc…). In these cases, data are reported according to the groups and
not the treatment.
A footnote indicates how BQL values are managed for the computation of the statistics.
Notes to programmer:


Plot against protocol time since last dosing. Individual data may have been flagged for exclusion if actual time differs
significantly from scheduled ones.
One- or two-sided error bars may be used. For one-sided, either use the same side or chose the side depending on the mean
values (upward for higher and downward for smaller).
9
2nd DRAFT for Review
SPONSOR/PROTOCOL/PRODUCT INFO
(page x of x)
Table 14.2-x.x Summary statistics for PK parameters
by compound, matrix, analyte and [actual/randomised] [treatments/group]
Analysis Set : PK analysis set
Compound: XXX, Matrix: YYY, Analyte: ZZZ
Actual
Period
treatment
day
Statistic
1
n
TRTA
Mean (SD)
CV% mean
Geo-mean
CV% geo-mean
Median
[Min; Max]
<Parameter 1>
<Parameter 2>
<Parameter 3>
<unit>
<unit>
<unit>
xx
xxx (xxx)
xx.x
xxx
xx.x
xxx
[xxx;xxx]
xx
xxx (xxx)
xx.x
xxx
xx.x
xxx
[xxx;xxx]
xx
xxx (xxx)
xx.x
xxx
xx.x
xxx
[xxx;xxx]
CV% = coefficient of variation (%)=sd/mean*100;
CV% geo-mean=(sqrt (exp (variance for log transformed data)-1))*100
Geo-mean: Geometric mean.
Geo-mean and CV% geo-mean not presented when the minimum concentration is zero at
respective timepoint.
PATH DATA/PROGRAM/OUTPUT
PRODUCTION STATUS/RUN DMMMYYYY: HHMM
Figure 8-6. Shell for summary of PK parameters
Annotations:
10
2nd DRAFT for Review




If the actual treatment is known and if it differs from the randomised treatment, it is recommended to report according to actual
treatment. Otherwise, the randomised treatment is reported.
In some particular multi-arm trials, all groups receive the same treatment but they differ by other characteristics (such as
gender, disease status or stage, administration with food, etc…). In these cases, data are reported according to the groups and
not the treatment.
In sequential and cross-over trials, the sequence and period are reported. In parallel trials, these data are skipped. In multi-part
trials, the part is displayed either in the title, headline or column.
A footnote indicates how BQL values are reported.
11
2nd DRAFT for Review
SPONSOR/PROTOCOL/PRODUCT INFO
(page x of x)
Table 14.2-x.x Summary statistics for PK concentration
by compound, matrix, analyte and [actual/randomised] [treatments/group]
Analysis Set : PK analysis set
Compound: XXX, Matrix: YYY, Analyte: ZZZ, Unit : uom
Dose
Scheduled
Profile
reference
time point
day
id
(hrs)
1
0.0
1
Statistic
n
Mean (SD)
CV% mean
Geo-mean
CV% geo-mean
Median
[Min; Max]
TRTA
xx
xxx (xxx)
xx.x
xxx
xx.x
xxx
[xxx;xxx]
TRTB
xx
xxx (xxx)
xx.x
xxx
xx.x
xxx
[xxx;xxx]
CV% = coefficient of variation (%)=sd/mean*100;
CV% geo-mean=(sqrt (exp (variance for log transformed data)-1))*100
Geo-mean: Geometric mean.
Geo-mean and CV% geo-mean not presented when the minimum concentration is zero at
respective timepoint.
BLQ Values considered as zero in descriptive statistics calculation.
PATH DATA/PROGRAM/OUTPUT
PRODUCTION STATUS/RUN DMMMYYYY: HHMM
Figure 8-7. Shell for summary of PK concentration
Annotations:
12
2nd DRAFT for Review




If the actual treatment is known and if it differs from the randomised treatment, it is recommended to report according to actual
treatment. Otherwise, the randomised treatment is reported.
In some particular multi-arm trials, all groups receive the same treatment but they differ by other characteristics (such as
gender, disease status or stage, administration with food, etc…). In these cases, data are reported according to the groups and
not the treatment.
In sequential and cross-over trials, the sequence and period are reported. In parallel trials, these data are skipped. In multi-part
trials, the part is displayed either in the title, headline or column.
A footnote indicates how BQL values are reported.
13
2nd DRAFT for Review
8.3. PK TFLs Checklist
8.3.1. Individual data handling in listings
Concentration data below the lower limit of quantification (BLQ) should be labelled as such in
the data listings. Several flagging options are possible for BLQ data, including:
 the actual numerical data with a flag (eg, ”*”),
 a missing value,
 an imputed value such as zero, the LLOQ, the LLOQ/2, etc…
 the labels “BQL” or “<X” where X is the numerical value of the LOQ established by the
laboratory.
It is useful to add a footnote to the listing in order to indicate how BLQ data were reported.
Missing values should also be labelled as such in the data listings. Label such as “NV” (no value)
or “.” (dot) may be used.
Any missing sampling or concentration data that was imputed should be flagged in the
concentration data listing.
Any individual data excluded from NCA or statistical analysis should be flagged in the listings.
8.3.2. Individual plots
Depending on the aim of the study (crossover, parallel design), individual graphs can be
presented per treatment (spaghetti plots) and/or by subject.
Plasma concentration vs. time profiles are often reported both on linear scale and on semilogarithmic scale.
For urine, amount excreted, cumulative or not, are usually presented in linear scale.
The actual times are most often reported in individual plots. The protocol time may be used as an
alternative when actual times are missing or if that presentation is more relevant.The label of the
X-axis should match the time scale that was used to avoid any confusion.
Regarding scales, the axes can be optimized per treatment, for the entire study, per subject or per
occasion, as deemed appropriate.
Most often, individual plots present all available data. Datapoint flagged for exclusion or that
have been imputed may be identified using different symbols and/or colors in the plots. For BQL
values, a footnote is often added that details how these data were imputed or managed in the
plots. An horizontal reference line at the BQL numerical value may also be added to indicate the
threshold in the plots.
1
2nd DRAFT for Review
8.3.3. Descriptive statistics in tables
Otherwise specify in the protocol or SAP, the following descriptive statistics are often calculated
for PK concentrations and PK parameters: N, arithmetic mean, SD, CV%, minimum, median,
maximum, geometric mean and geometric CV%. The geometric CV% is computed as:
CV% geo-mean=(sqrt (exp (variance for log transformed data)-1))*100.
For Tmax and Tlag, it is customary to report only the N, median, minimum and maximum
statistics.
Additional statistics such as the number of missing observations, quartiles (Q1, Q3), specific
percentiles, the standard error, %SEM and confidence intervals are less frequently reported.
The CV% will be reported as missing if the arithmetic mean is zero.
If the non-missing data values are not all positive, then the geometric mean and CV% cannot be
calculated and should be reported as missing.
The measures of precision (SD, CV%, geometric CV%, etc…) are not reported when there is
only one non-missing data.
8.3.3.1. Statistics in the presence of BQL data
In case of values below the LLOQ or above the upper limit of quantification (ULOQ), the
frequency (n, %) of values below the LLOQ and above the ULOQ, respectively, may be
reported.
It may not be relevant to report standard empirical statistics when the total number of BLQ
values is large (eg, when it exceeds 1/3rd of the total). Instead, the summary statistics (mean,
standard deviation) may be adapted to the presence of censored values (values below the LLOQ
and/or values above the ULOQ), by reporting the maximum likelihood estimates from a
parametric model for data that can be right censored and left censored (e.g., using SAS PROC
LIFEREG). In the case of censoring, the empirical median may not be reported. Likewise, the
empirical minimum (maximum) may not be reported if there are values below the LLOQ (above
the ULOQ).
8.3.4. Individual data handling in summary tables
When the actual sampling time differs significantly from the protocol time (eg., when the
deviation is greater than 10%), then the concentration should be excluded from descriptive
statistics calculation but kept in the PK parameters determination. A flag and a footnote should
be presented in the table.
The method used to handle BQL data prior to the calculation of summary statistics should be
presented in a footnote on the summary tables. If BQL data have been imputed, the imputation
2
2nd DRAFT for Review
value (eg, 0, LLOQ or LLOQ/2) should be indicated. If data has been left missing, if the actual
values BQL or if a censoring method has been used, this should be indicated in the footnote.
In urine, only descriptive statistics on amount excreted/fraction or cumulative amount excreted
over time are usually computed (no descriptive statistics on concentration or volume). When
these amounts are not estimable they should be considered as missing.
In specific situations, it may not be possible to calculate PK parameters, either because some
required input data are not available or for some justifiable calculation reason. If the proportion
of missing PK parameter is large (eg, larger than 1/3 of all data), descriptive statistics may not be
calculated. In that case, the specific reason for not reporting the statistics should be indicated in
the table.
8.3.5. Mean Plots
When generating the mean concentration-time plots from the average dataset, replace the early
(pre-Cmax) not calculated values by zero in order to capture the lag-time, if any.
Protocol times are used for generating the mean concentration-time data.
Considering the inherently log-normal distribution of concentrations, plots of geometric mean
concentration versus time may be generated in addition to or in replacement of the arithmetic
mean plots. Linear and log-linear displays are often produced side by side to clearly delineate the
concentration-time profiles.
Error bars are usually added to the mean display in the linear-linear scale to characterise the data
distribution in the population. In that case the standard deviation (SD) is often reported. Error
bars may also be used in particualr situations to characterize the precision on the mean. Then,
either standard errors (SE) or confidence intervals (CI) are reported. In general two-sided error
bars are reported, unless the figures becomes too busy. In that case, one-sided bars may be
considered. Different sides (upper/lower) may also be considered in multi-line plots to improve
rendering of the graph. For instance, use a lower bar for groups having low concentrations and
vice-versa.
8.3.6. Formats for individual data and statistics
If possible, the individual PK concentrations and parameters should be formatted to 3 significant
figures in the individual data listings. Other formats or rounding presentations may be considered
as deemed appropriate.
For statistical tables, the descriptive statistics are often rounded to one additional digit (eg, 4
significant figures) for the mean, and median, to two additional digits (eg, 5 significant figures)
for the SD, and to the same number of digits (eg, 3 significant figures) for the minimum and
maximum values. The CV values are often reported in percent unit and using one decimal place.
3
2nd DRAFT for Review
9. Example SAP Language
9.1. Data to be analysed
All subjects with evaluable PK parameter data and no major protocol deviation with an impact
on PK data will be included in the PK analysis set.
9.2. Pharmacokinetic methods
All patients within the PK analysis set will be included in the pharmacokinetic data analysis.
Individual PK data for all randomised subjects will be listed.
Biofluid concentrations will be expressed in [UOM]. All concentrations below the limit of
quantification (LLOQ) or missing data will be labeled as such in the concentration data listings.
Concentrations below the LLOQ will be treated as [zero or LLOQ or LLOQ/2] in summary
statistics for concentration data only. They will not be considered for calculation of PK
parameters (with the exception of the pre-dose samples). PK concentration profiles will be
summarized by treatment and over time in tabular and graphical formats. Arithmetic mean (+/SD) concentration-time plots will be produced.
The following pharmacokinetic parameters will be determined using non-compartmental
methods:

Primary PK parameters: AUC, AUC(0-t), Cmax.

Secondary PK parameters: tmax, t½.
Descriptive statistics of pharmacokinetic parameters and concentrations will include mean, SD,
and CV, min and max. When a geometric mean will be presented it will be stated as such. Since
Tmax is generally evaluated by a nonparametric method, median values and ranges will be given
for this parameter.
4
2nd DRAFT for Review
10. References
TO BE COMPLETED
5
2nd DRAFT for Review
11. Acknowledgements
The key contributors include: Francois Vandenhende, Ingrid Burton, Sascha Ahrweiler, and
Vincent Buchheit.
Additional contributors and members of the white paper project within the FDA/PhUSE
Development of Standard Scripts for Analysis and Programming Working Group include: …
Acknowledgement to others who provided text for various sections, review comments, and/or
participated in discussions related to methodology: …
6