Presentation Title (26pt, Arial Bold)
Download
Report
Transcript Presentation Title (26pt, Arial Bold)
When to Use CDISC Standards A practical experience gained from a
US submission with Pharmacokinetic data
BASAS, July 5 2011
Kristie Kooken, Scott Bahlavooni, Amy Klopman
Objectives for today’s talk
Page 2
To better understand CDISC guidance for
Pharmacokinetic (PK) data
To hear first-hand experience using these guidance
on a complex filing
To learn about possible challenges with CDISC
models for PK data and solutions we implemented
STATISTICAL PROGRAMMING AND ANALYSIS
Project Background
Page 3
New Molecular Entity with large PK piece – filing for New Drug Application
(NDA)
Small Molecule (taken orally)
PK piece is comprised of two types of studies:
Healthy Volunteers (HV)
Patient Studies
Retrospectively mapped to SDTM structure
PK piece includes 7 studies
4 are fully-outsourced
3 are in-house by Genentech
PK & PD analytes for each study
PK analyses cover:
PK characteristics of the drug
Food effects - label enabling
Exposure-response analyses for efficacy and safety
QTc – label enabling
Population PK modeling to determine covariates that
could potentially alter dose-concentration relationships
STATISTICAL PROGRAMMING AND ANALYSIS
Our Goal
Is To
Develop
Support
Datasets
Data and Data Handling
Page 4
PK data (any type of assay data) is complex
Time by drug concentrations is unit of analysis
Concentration data receipt is not aligned with other clinical data
PK analysis (i.e. calculating PK parameters), typically occurs
late in the clinical trial
At Genentech, a specialty programming group oversees all PK
deliverables from Statistical Programming and Analysis
Allows for a deeper understanding of this data type, the purpose
of various analyses and greater contribution to analyses
STATISTICAL PROGRAMMING AND ANALYSIS
PK Data: time vs. concentration plot
Page 5
100
Serum anti-IL13 (µg/mL)
1 mg/kg SC
10
1
0.1
0.01
0
14
28
42
Time (day)
STATISTICAL PROGRAMMING AND ANALYSIS
56
70
84
98
CDISC Overview
Page 6
Mission Statement:
To develop and support global, platform-independent data standards that enable
information system interoperability to improve medical research and related areas of
healthcare
Study Data Tabulation Model (SDTM):
[To] define a standard structure for study data tabulations that are to be submitted
as part of a product application to a regulatory authority
Analysis Data Model (ADaM):
To provide a framework that enables analysis of the data, while at the same time
allowing reviewers and other recipients of the data to have a clear understanding of
the data’s lineage from collection to analysis to results
STATISTICAL PROGRAMMING AND ANALYSIS
CDISC Data Flow
STATISTICAL PROGRAMMING AND ANALYSIS
Page 7
SDTM PK Domains
Page 8
Pharmacokinetic Concentrations (PC)
Data collected about concentrations of analytes from blood samples
as function of time after dosing
One record per concentration or sample characteristic
Pharmacokinetic Parameters (PP)
Data describing the parameters of the time-concentration curve for
PC data, such as Tmax, HL, AUC, Cmax
One record per PK parameter per time-concentration profile
It is recognized the PP is a derived dataset and may be produced
from an analysis dataset
STATISTICAL PROGRAMMING AND ANALYSIS
SDTM PK Domain Ambiguity
Page 9
At DIA/FDA CDER/CBER Computational Science Annual
Meeting, Mar 2011 - Helena Sviglin, CDER, stated the FDA is
working to better understand PK domains
Should laboratory/PD assessments be included in PC?
Examples: Albumin, urine volume, time-point based glucose
How to relate records in PC to PP?
PPRFTDTC
Analysis metadata
RELREC
How to implement controlled terminology?
PK CT released in 2010
STATISTICAL PROGRAMMING AND ANALYSIS
Practical Experience: SDTM & PK
Page 10
PC and PP domains mapped retrospectively for 4 of 7 studies
Pharmacologists consulted to align reported parameter names with CDISC
PK parameter controlled terminology
Selecting the appropriate method to relate PC and PP
Consulted with internal and external subject matter experts including
SDTM experts, leading industry experts, Pharmacologists
Differing opinions for defining reference time point
Protocol defined collection schedule vs. analysis time points
Decision:
Protocol defined collection schedule is represented in PC timing
variables. PPRFTDTC is defined as the analysis reference date.
PPGRPID is defined to indicated the concentrations used to calculate
the parameter.
The specific linking between dosing datetime and PK concentration
datetime are described in the analysis metadata
STATISTICAL PROGRAMMING AND ANALYSIS
Considerations for SDTM & PK (1)
Page 11
Sponsor
Define and document a methodology for defining PK SDTM timing
variables and relating PC and PP domains
Document (e.g. a datastore) mapping reported parameter names to
CDISC PK terminology
Study
Prospectively map PC and PP domains
Obtain "test data" early in the clinical study life-cycle
Ensure time point reference dates are clearly defined and collected
Complicated derivations should not be required to assign time point
reference dates
Collaborate with the Pharmacologist to clearly document the
concentrations used to calculate each PK parameter
Confirm with the Pharmacologist what PK parameters are reported in
the Clinical Study Report(s)
STATISTICAL PROGRAMMING AND ANALYSIS
Considerations for SDTM & PK (2)
Page 12
CDISC
Reassess PK domains
Does the SDTM IG provide clear guidance to sponsors?
Do the SDTM IG examples sufficiently document various PK data
analyses?
Are the examples in the SDTM IG aligned with the controlled
terminology?
Can the relationship between the domains be clearly defined?
Solicit sponsor and (FDA) reviewer feedback
Do PK domains meet sponsor tabulation and analytical needs?
Do PK domains meet (FDA) review needs?
STATISTICAL PROGRAMMING AND ANALYSIS
CDISC Guidance for ADaM
Page 13
ADSL
The ADSL dataset contains one record per subject
Population flags, planned and actual treatment variables for each
period, demographic information, stratification and subgrouping
variables, etc.
ADSL and its related metadata are required in a CDISC-based
submission of data from a clinical trial even if no other analysis
datasets are submitted
Basic Data Structure (BDS)
Dataset with one or more records per subject, per analysis parameter,
per analysis time point
Specific features to map for PK data:
Central set of variables represent the data being analyzed
PARAM – Description of the value being analyzed
AVAL – Actual value being analyzed
Timing variables – ADT (date), ATM (time)
STATISTICAL PROGRAMMING AND ANALYSIS
ADaM BDS Ambiguity
Page 14
ADaM Implementation Guide states:
Though the BDS supports most statistical analyses, it does not support
all statistical analyses... it does not support simultaneous analysis of
multiple dependent (response/outcome) variables or a correlation
analysis across a range of response variables. The BDS was not
designed to support analysis of incidence of adverse events or other
occurrence data
This version of the implementation guide does not fully cover dose
escalation trials or integration of multiple studies.
Use ADaM for PK data where it fits…
How did we determine when and where to use BDS?
ADaM BDS works well
Concentration analysis dataset to support summary outputs
ADaMIG is not clear
Datasets used to support modeling or explore PKPD
relationships
STATISTICAL PROGRAMMING AND ANALYSIS
Practical Experience: ADaM & PK (1)
Page 15
Gathered advice from subject matter experts within & outside of Statistical
Programming and Analysis:
Met with Genentech Global Data Standards team
Developed basic mapping for ADaM domains
Clinical Pharmacology group met with Modeling & Simulation expert - Marc
Gastonguay (Metrum Research Group)
Determined analysis format for Pharmacometric Reviewer
Developed a systematic data plan for each analysis:
PART 1:
Clinical Study Reports – 6 studies
ADaM structured datasets to support TLG generation
ADPC – Pharmacokinetic concentrations supporting summary
outputs
Mapped assay data in support of TLG (inclusion of PD data
collected as Safety Labs)
ADPP – Pharmacokinetic parameters supporting summary outputs
STATISTICAL PROGRAMMING AND ANALYSIS
Practical Experience: ADaM & PK (2)
Page 16
PART 2:
QTc Analysis
Description: ECG and PK concentrations at matching timepoints
Purpose: QTc changes over drug concentration level could be potentially
dangerous (cardiotoxicity)
Significant increase from baseline is a clinical concern
Design: Analysis dataset structured to best fit analysis
ADaM-like dataset, all ADSL variables were present
PKPD analyses
Description: Exposure-response analyses with Efficacy / Safety
Purpose: Explore the relationship between drug exposure and
pharmacodynamic measures
Design: Structured for NONMEM or S-Plus software
Population PK (5 studies)
Description: Drug concentration data formatted for statistical modeling
Purpose: Estimate the population PK and explore relationships to study
covariates (Biomarkers, age, race, baseline disease severity, etc.)
Design: Structured for NONMEM software
STATISTICAL PROGRAMMING AND ANALYSIS
Considerations for ADaM & PK
Page 17
Current guidelines for CDISC ADaM limit PK data to the BDS structure
Works well for “Analysis-Ready” PK datasets supporting TLGs
Otherwise, determine best analysis format
Consider analysis tools FDA will use
Whatever analysis format to use, source to SDTM
Focus on “Fit for analysis”
CDISC Analysis Data Model Version 2.1:
The focus of any analysis format is to
Facilitate clear and unambiguous communication
Be readily useable by commonly available software tools
Provide traceability between the analysis data and its source data
(ultimately SDTM)
STATISTICAL PROGRAMMING AND ANALYSIS
Other Considerations
Page 18
Outsourcing
Four of the 7 studies with a PK data piece were outsourced
Create a plan for how to incorporate outsourced data into
your filing
Does outsourced SDTM / ADaM match Sponsor SDTM /
ADaM?
Do you have in-house all the pieces you need to create
analysis deliverables?
Can be difficult to assess if data is only used for
population PK analysis
STATISTICAL PROGRAMMING AND ANALYSIS
Questions
STATISTICAL PROGRAMMING AND ANALYSIS
Page 19
Additional Information
Pharmacology:
LABELING
http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm065010.htm
BIOPHARMACEUTICS
http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm064964.htm
CLINICAL PHARMACOLOGY
http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm064982.htm
MODEL/DATA FORMAT
http://www.fda.gov/AboutFDA/CentersOffices/CDER/ucm180482.htm
SDTM / Define xml:
FDA (data standards)
http://www.fda.gov/ForIndustry/DataStandards/default.htm
CDISC (standards)
http://www.cdisc.org/standards
OpenCDISC (SDTM validator)
http://www.opencdisc.org/projects/validator
SAS (clinical standards toolkit)
http://support.sas.com/rnd/base/cdisc/cst/index.html
FDA Standards for Electronic Submissions:
SAS
http://www.sas.com/industry/government/fda/faq.html
STATISTICAL PROGRAMMING AND ANALYSIS
Page 20
Thank you to:
Bei Wang
Michael Ward
Laura Harris
Rick Graham
Jin Jin
Yongcun Zhang
Patty Gerend
STATISTICAL PROGRAMMING AND ANALYSIS
Page 21