Transcript Document
Using Statistical
Innovation to Impact
Regulatory Thinking
Harry Yang, Ph.D.
MedImmune, LLC
May 20, 2014
How Do We Influence Regulatory
Thinking?
2
04/14/2008 – 6:00pm
An Old Tried and True Method
Throw statisticians at the deep end of regulatory interactions
3
04/14/2008 – 6:00pm
An Old Tried and True Method (Cont’d)
Throw statisticians at the deep end of regulatory interactions
– Low success rate
– Lost potential/opportunities
4
04/14/2008 – 6:00pm
A More Effective Approach to Influencing
Regulatory Thinking
Identify opportunities
Opportunities
Understand our own strengths
Influence thru
collaboration
5
Three Case Examples
Acceptable limits of residual host cell DNA
Risk-based pre-filtration limits
Bridging assays as opposed to clinical studies
6
Acceptable Residual DNA Limits
Biological product contains residual DNA from host cell
Residual DNA could encode or harbor oncogenes and infectious
agents
Mitigate oncogenic and infective risk thru restriction on DNA amount
per dose and size
WHO and FDA guidelines recommend
– Amount ≤ 10 ng/dose
– Size ≤ 200 base pairs (bp)
7
Safety Factor
Safety factor (Pedan, et al., 2006)
– Number of doses taken to induce an oncogenic or infective event
Om
SF
.
mi
I0
E[U ]
M
Om :
I0 :
mi:
M:
E[U]:
Amount of oncogenes to induce an event
Number of oncogenes in host genome
Oncogene sizes
Host genome size
Expected amount of residual hose DNA/dose
Revised Safety Factor (Lewis et al., 2001)
Om
SF
.
m
P * I 0 i E[U ]
M
Om :
I0 :
mi:
M:
E[U]:
P:
Amount of oncogenes to induce an event
Number of oncogenes in host genome
Oncogene sizes
Host genome size
Expected amount of residual hose DNA/dose
Percent of DNA with size ≥ oncogene size
DNA Inactivation
10
Relationship between Enzyme Cutting Efficiency
and Median DNA Size (Yang, et al., 2010)
Probability of enzyme cutting thru two adjacent
nucleotides, p, and median DNA size Med satisfy
p 1 2
1
Med
11
Safety Factor Based on Probabilistic Modeling
(Yang et al., 2010)
I0 :
mi:
M:
Med0:
E[U]:
Number of oncogenes in host genome
Oncogene sizes
Host genome size
Median residual DNA size
Expected amount of residual hose DNA/dose
Method Comparison
Theoretically it can be shown FDA methods either over- or underestimate safety factor (Yang, 2013)
Risk-based Specifications
14
DNA Content and Size Can Be Outside of
Regulatory Limits without Compromising Safety!
15
Establishing Pre-filtration Bioburden Test Limit
16
EMA Guidance (2008): Notes for Guidance on
Manufacture of Finished Dosage Form
17
EMA Guidance (2008): Notes for Guidance on
Manufacture of Finished Dosage Form
18
Risk Associated with Three Different Test Schemes
5%
20 CFU
63 CFU
32 CFU
19
Mitigating Risk of Larger Number of Bioburden thru
Sterial Filtration
20
Sterile Filtration
FDA guidance requires that filters used for the final filtration should
be validated to reproducibly remove microorganisms from a carrier
solution containing bioburden of a high concentration of at least 107
CFU/cm2 of effective filter area (EFA)
21
Upper Bound of Probability p0 for a CFU to Go
Thru Sterile Filter (Yang, et al., 2013)
22
Upper Bound of Probability of Having at least 1
CFU in Final Filtered Solution
It’s a function of batch size S, pre-filtration test volume V, and the
maximum bioburden level D0 of the pre-filtration solution
By choosing the batch size, this probability can be bounded by a
pre-specified small number δ.
23
Maximum Batch Sizes Based on Risks and Prefiltration Test Schemes
24
Bridging Assays as Opposed to Clinical Studies
FFA and TCID50 are different assays but both used for clinical trial
material release (Yang, et al., 2006)
Theoretical mean difference
25
04/14/2008 – 6:00pm
Other Ways to Influence Regulatory Thinking
Serve on committees such as USP Statistics Expert, CMC Working
Groups, Industry Consortiums
Organize joint meetings/conferences/workshops
26
USP Bioassay Guidelines
Originally USP <111> and EP 5.3
<111> was split into two chapters, USP <1032> Design and
Development of Biological Assays and USP <1034> Analysis of
Biological Assays
<1033> Biological Assay Validation added to the suite
“Roadmap” chapter
(to include glossary)
27
27
Parallelism Testing
Significance vs. equivalence test (Hauck et al., 2005)
Feasibility of implementation (Yang et al., 2012)
Method comparison based on ROC analysis (Yang and Zhang, 2012)
Bayesian solution (Novick, Yang, and Peterson, 2012)
28
Testing Assay Linearity
Directly testing linearity
(Novick and Yang, 2013)
Testing linearity over a prespecified range (Yang, Novick,
and LeBlond, 2014)
The method is being
considered to be included in a
new USP chapter on statistical
tools for method validation
29
A Few Additional Thoughts
30
Conduct Innovative Statistical Research on
Regulatory Issues
Solutions based on published
methods are more likely
accepted by regulatory
agencies
31
Take a Good Statistical Lead in Resolving
Regulatory Issues
32
Regularly Communicate with Regulatory
Authorities
33
Conduct Joint Training
34
04/14/2008 – 6:00pm
References
H. Yang, S.J. Novick, and D. LeBlond. (2014). Testing linearity over a pre-specified range. Submitted.
H. Yang, N. Li and S. Chang. (2013). A risk-based approach to setting sterile filtration bioburden limits. PDA J. of Pharm.
Science and Technology. Vol. 67: 601-609
D. LeBlond, C. Tan and H. Yang (2013). Confirmation of analytical method calibration linearity. May – June Issue,
Pharmacopeia Forum. 39(3).
D. LeBlond, C. Tan and H. Yang. (2013). Confirmation of analytical method calibration linearity: practical application.
September - October Issue. Pharmacopeia Forum
S. Novick and H. Yang. (2013). Directly testing the linearity assumption for assay validation. Journal of Chemometrics. DOI:
10.1002/cem.2500
H. Yang. Establishing acceptable limits of residual DNA (2013). PDA J. of Pharm. Sci. and Technol., March – April Issue.
67:155-163
S. Novick, H. Yang and J. Peterson. A Bayesian approach to parallelism testing (2012). Statistics in Biopharmaceutical
Research. Vol. 4, Issue 4, 357-374.
H. Yang, J. Kim, L. Zhang, R. Strouse, M. Schenerman, and X. Jiang. (2012). Parallelism testing of 4-parameter logistic
curves for bioassay. PDA J. of Pharm. Sci. and Technol. May-June Issue, 262-269.
H. Yang and L. Zhang. Evaluations of parallelism test methods using ROC analysis (2012). Statistics in Biopharmaceutical
Research. Volume 4, Issue 2, p 162-173
H. Yang, L. Zhang and M. Galinski. (2010). A probabilistic model for risk assessment of residual host cell DNA in biological
product. Vaccine 28 3308-3311
H. Yang and I. Cho. (2006) Theoretical Relationship between a Direct and Indirect Potency Assays for Biological Product of
Live Virus. Proceedings of 2006 JSM.
35
Q&A
36
04/14/2008 – 6:00pm