Transcript Discovery San Diego Presentationx

JMP for Biosimilars: Tools for
Analytical Similarity
Sept 16, 2015
W. Heath Rushing, Adsurgo LLC
Andrew Karl, Adsurgo LLC
Richard Burdick, Elion Labs
Outline
• Introduction
• Current FDA Thinking/Approach
• Demonstration: Statistical Equivalence using Sample Size and
Variance Adjustment Method
• Future Considerations
2
INTRODUCTION
3
What is a Generic?
• “A drug product that is comparable to a brand/reference
listed drug product in dosage form, strength, route of
administration, quality and performance characteristics,
and intended use.” – Center for Drug Evaluation and
Research (CDER)
• A generic drug can be licensed when a brand drug is off
patent; must contain same active product ingredients
and dosage form.
• Hatch-Waxman Act of 1984.
• Common approach is to show statistical equivalence for
quality and performance characteristics:
– Potency, purity, and stability.
Reference: http://www.fda.gov/downloads/Drugs/DevelopmentApprovalProcess/SmallBusinessAssistance/ucm127615.pdf
4
What is a Biosimilar?
Biological products are drugs that are developed from
living organisms.
• Were patented by companies starting in the 1982.
• Contain proteins or other complex structures that are
at risk of degradation throughout the manufacturing
process. The additional structural complexity in
biologic products creates additional challenges when
defining targets for potency, purity, and stability in
these products.
Feroz, J.; Hershenson, S.; Khan, M. A.; Martin-Moe, S. (2015) Quality by Design for Biopharmaceutical Drug Product Development.
Springer: New York
5
What is a Biosimilar?
• The Biologics Price Competition and Innovation Act (BPCI)
of 2009 created an abbreviated licensure pathway for
biological products shown to be similar or “biosimilar” to a
pre-existing FDA-licensed product.
• The BPCI Act was an amendment [section 351(k)] to the
Public Health Service (PHS) Act.
• Section 351(k) of the PHS Act defines biosimilar as “highly
similar to the reference product notwithstanding minor
differences between the biological product and the
reference product in terms of safety, purity, and potency of
the product.”
Reference: FDA (2015) Scientific Considerations in Demonstrating Biosimilarity to a Reference Product, Guidance for Industry, United
States FDA, Silver Spring, Maryland, USA.
6
First FDA-approved Biosimilar
• Sandoz, Inc. (Novartis) received approval for Zarxio,
biosimilar to Amgen’s Neupogen, on Mar 6, 2015.
• Amgen originally filed the biologics licensing
agreement (BLA) for Neupogen on Feb 20, 1991.
• Zarxio received approval for all five indications
associated with Neupogen.
Reference: http://www.fda.gov/NewsEvents/Newsroom/PressAnnouncements/ucm436648.htm
7
Biosimilars
http://www.biopharma-reporter.com/Markets-Regulations/Where-are-tomorrow-s-biosimilar-hotspots
8
Analytical Similarity
Although there is only one FDA-approved biosimilar, it
is estimated there are at least 280 biosimilars in the
pipeline and clinical trials are growing at a rate of 20%
per year.
• Describe the current FDA thinking/approach:
– Sample size and variance adjusted margin.
• Demonstrate the sample size and variance adjusted
margin method using a JMP script.
• Outline future considerations for the current FDA
approach.
Reference: http://www.biopharma-reporter.com/Markets-Regulations/Where-are-tomorrow-s-biosimilar-hotspots
9
CURRENT FDA THINKING/APPROACH
10
Guidance Documents
In 2012, the FDA issued three (draft) guidance documents on the
subject of biosimilars:
• FDA (2012a) Scientific considerations in demonstrating
biosimilarity to a reference product. FDA, Silver Spring,
Maryland, USA.
• FDA (2012b) Quality considerations in demonstrating
biosimilarity to a reference protein product. FDA, Silver
Spring, Maryland, USA.
• FDA (2012c) Biosimilars: Questions and Answers Regarding
Implementation of the Biologics Price Competition and
Innovation Act of 2009. FDA, Silver Spring, Maryland, USA.
11
Stepwise Approach
• To demonstrate biosimilarity, the FDA recommends
applicants use a stepwise approach for comparison of the
biosimilar to the reference product; referred to as the
“totality of evidence” approach.
• Starts with extensive evaluation of CQAs associated with
the structural and functional characterization of both the
biosimilar and reference product.
• This extensive evaluation serves as the foundation of a
biosimilar development program; known as ‘analytical
similarity.’
• Analytical similarity requires the biosimilar to be “highly
similar to the reference product notwithstanding minor
differences in clinical inactive components.”
Reference: FDA (2015) Scientific Considerations in Demonstrating Biosimilarity to a Reference Product, Guidance for Industry, United
States FDA, Silver Spring, Maryland, USA.
Reference: Tsong, Yi and OB CMC Analytical Biosimilar Method Development Team (Meiyu Shen, Cassie Xiaoyu Dong).
Development of Statistical Approaches for Analytical Biosimilarity Evaluation [Powerpoint]. DIA/FDA Statistics Forum 2015. 12
Analytical Similarity
Analytical similarity is an extensive evaluation of CQAs
associated with the structural and functional
characterization of both the biosimilar and reference
product.
1. Identify CQAs.
2. Classify CQAs into three tiers of criticality.
3. Assess analytical similarity for CQAs based upon the
different tiers. Different tiers require different levels
of statistical rigor.
Reference: Chow SC (2014) On Assessment of Analytical Similarity in Biosimilar Studies. Drug Des 3: 119. doi:10.4172/21690138.1000e124.
13
Three Tier Approach
• Tier 1: Most critical.
– Statistical equivalence test.
• Tier 2: Mild to moderate critical/less critical
– Quality range method.
• Tier 3: Least critical to clinical outcomes
– Graphical.
Reference: Tsong, Yi and OB CMC Analytical Biosimilar Method Development Team (Meiyu Shen, Cassie Xiaoyu Dong).
Development of Statistical Approaches for Analytical Biosimilarity Evaluation [Powerpoint]. DIA/FDA Statistics Forum 2015. 14
Tier 1 – Statistical Equivalence
• Since tier 1 is the most critical, it requires the most
statistical rigor.
• Relies on the use of a statistical equivalence test.
– Generate 90% confidence interval for difference in means between
reference product and biosimilar.
– Compare to equivalence criteria (δ).
Selection of this equivalence criteria (δ) is the key to the outcome of
analytical similarity.
Reference: Tsong, Yi and OB CMC Analytical Biosimilar Method Development Team (Meiyu Shen, Cassie Xiaoyu Dong).
Development of Statistical Approaches for Analytical Biosimilarity Evaluation [Powerpoint]. DIA/FDA Statistics Forum 2015. 15
Tier 1 – Statistical Equivalence
Selection of this margin (equivalence criteria) is the key
to the outcome of analytical similarity.
• Current methods used:
– Fixed Margin
– Limentani Approach.
• Proposed FDA method is the sample size and
variance adjusted margin.
– Focuses on power of the test for low sample sizes.
Reference: Tsong, Yi and OB CMC Analytical Biosimilar Method Development Team (Meiyu Shen, Cassie Xiaoyu Dong).
Development of Statistical Approaches for Analytical Biosimilarity Evaluation [Powerpoint]. DIA/FDA Statistics Forum 2015. 16
Statistical Equivalence using Sample Size
and Variance Adjustment Method
• Step 1: Determine variability of reference product
using the standard deviation.
• Step 2: Calculate statistical equivalence acceptance
criteria as ± c * standard deviation of the reference
product, where c = 1.5.
• Step 3: Compute 90% confidence interval on the
difference between product means. Compare against
acceptance criteria.
Note: A test size of α will construct a (1 – 2α) confidence interval.
17
Statistical Equivalence using Sample Size
and Variance Adjustment Method
Method depends on the number of reference versus
the number biosimilar lots used in the comparison:
• The total number of reference lots are less than or equal to the total
number of biosimilar lots.
• The total number of reference lots are greater than the total number of
biosimilar lots.
18
Statistical Equivalence using Sample Size
and Variance Adjustment Method
The total number of reference lots is less than or equal to the number of biosimilar lots.
1.
2.
3.
4.
5.
6.
Let Nr denote the total number of reference lots and Nb denote the number of biosimilar
lots where Nr ≤ Nb.
All reference lots (Nr) and all biosimilar lots (Nb) will be used to calculate the difference
between mean attribute measurements between the two products, and the associated 90%
confidence interval.
All reference lots (Nr) are used to estimate the standard deviation (s) of the reference
material. The test assumes that the attribute variance is constant across products.
Calculate the acceptance criterion as ± 1.5 * s.
Calculate a 90% confidence interval for a difference in means using the data from step 2.
Compare the 90% confidence interval to the acceptance criteria. The null hypothesis
(products differ with respect to the attribute) is rejected if the endpoints of the confidence
interval lie within the acceptance margin.
Calculate the power for a given difference (s/8) and test size (α).
Reference: Dong, Xiaoyu (January 13, 2015) Statistical Revew and Evaluation of BLA No. 125553, Center for Drug Evaluation
and Research, United States FDA, Silver Spring, Maryland, USA.
19
DEMONSTRATION - STATISTICAL
EQUIVALENCE USING SAMPLE SIZE AND
VARIANCE ADJUSTMENT METHOD
WHERE NR ≤ NB
20
Statistical Equivalence using Sample Size
and Variance Adjustment Method
The total number of reference lots is greater than the number of biosimilar lots.
1.
2.
3.
4.
5.
6.
Let Nr denote the total number of reference lots, and Nb denote the number of biosimilar
lots where Nb < Nr. The difference between the lot sizes will be represented by mr = Nr – Nb.
A random subset (of size Nb) of the total number of reference lots (Nr) and all Nb biosimilar
lots will be used to calculate the difference between mean attribute measurements
between the two products, and the associated 90% confidence interval.
Use the remaining number of reference lots (mr) to estimate the standard deviation (s) of
the reference material. The test assumes that the attribute variance is constant across
products.
Calculate the acceptance criterion as ± 1.5 * s.
Calculate a 90% confidence interval for a difference in means using the data from step 2.
Compare the 90% confidence interval to the acceptance criteria. The null hypothesis
(products differ with respect to the attribute) is rejected if the endpoints of the confidence
interval lie within the acceptance margin.
Calculate the power for a given difference (s/8) and test size (α).
Reference: Chow SC (2014) On Assessment of Analytical Similarity in Biosimilar Studies. Drug Des 3: 119. doi:10.4172/216921
0138.1000e124.
DEMONSTRATION - STATISTICAL
EQUIVALENCE USING SAMPLE SIZE AND
VARIANCE ADJUSTMENT METHOD
WHERE NR > NB
22
FUTURE CONSIDERATIONS
23
Future Considerations
Yi Tsong & OB CMC Analytical Biosimilar Method Development Team:
• Since the margin depends on s, more appropriate test (than a t-test)?
• Develop multiple approaches be based on number of reference lots?
• Adjust the acceptance criteria (margin) based upon a constant shift?
• How to determine margin when equivalence uses ratio (versus
difference)?
Rick Burdick and Jose Ramirez (Amgen):
• Use all reference lots for 90% confidence interval if acceptance
criteria set using scientific knowledge.
• Develop equivalence margin based on effect size:
– Use a confidence interval on the effect size.
– Correlated reference lots will require the value of c to be increased for
desired power.
Reference: Tsong, Yi and OB CMC Analytical Biosimilar Method Development Team (Meiyu Shen, Cassie Xiaoyu Dong).
Development of Statistical Approaches for Analytical Biosimilarity Evaluation [Powerpoint]. DIA/FDA Statistics Forum 2015.
Reference: Burdick, Richard K. and José G. Ramirez. Statistical Issues in Biosimilar Analytical Assessment: Perspectives on FDA
24
ODAC Analysis [Powerpoint]. DIA/FDA Statistics Forum 2015.
REFERENCES
25
References
1.
Burdick, Richard K. and José G. Ramirez. Statistical Issues in Biosimilar Analytical Assessment: Perspectives on FDA ODAC
Analysis [Powerpoint]. DIA/FDA Statistics Forum 2015.
2.
Chow SC (2014) On Assessment of Analytical Similarity in Biosimilar Studies. Drug Des 3: 119. doi:10.4172/21690138.1000e124.
3.
Dong, Xiaoyu (January 13, 2015) Statistical Revew and Evaluation of BLA No. 125553, Center for Drug Evaluation and
Research, United States FDA, Silver Spring, Maryland, USA.
4.
FDA (2012a) Scientific considerations in demonstrating biosimilarity to a reference product. United States FDA, Silver Spring,
Maryland, USA.
5.
FDA (2012b) Quality considerations in demonstrating biosimilarity to a reference protein product. United States FDA, Silver
Spring, Maryland, USA.
6.
FDA (2012c) Biosimilars: Questions and Answers Regarding Implementation of the Biologics Price Competition and Innovation
Act of 2009. United States FDA, Silver Spring, Maryland, USA.
7.
FDA (2015) Scientific Considerations in Demonstrating Biosimilarity to a Reference Product, Guidance for Industry, United
States FDA, Silver Spring, Maryland, USA.
8.
Feroz, J.; Hershenson, S.; Khan, M. A.; Martin-Moe, S. (2015) Quality by Design for Biopharmaceutical Drug Product
Development. Springer: New York.
9.
Tsong, Yi and OB CMC Analytical Biosimilar Method Development Team (Meiyu Shen, Cassie Xiaoyu Dong). Development of
Statistical Approaches for Analytical Biosimilarity Evaluation [Powerpoint]. DIA/FDA Statistics Forum 2015.
Websites:
1. http://www.biopharma-reporter.com/Markets-Regulations/Where-are-tomorrow-s-biosimilar-hotspots
2. http://www.fda.gov/downloads/Drugs/DevelopmentApprovalProcess/SmallBusinessAssistance/ucm127615.pdf [Powerpoint].
26
3. http://www.fda.gov/NewsEvents/Newsroom/PressAnnouncements/ucm436648.htm