Analytical Similarity Assessment: Practical Challenges and Statistical Perspectives Richard Montes, Ph.D. Hospira, a Pfizer company Biosimilars Pharmaceutical.

Slides:



Advertisements
Similar presentations
Hypothesis Testing Goal: Make statement(s) regarding unknown population parameter values based on sample data Elements of a hypothesis test: Null hypothesis.
Advertisements

A Flexible Two Stage Design in Active Control Non-inferiority Trials Gang Chen, Yong-Cheng Wang, and George Chi † Division of Biometrics I, CDER, FDA Qing.
Significance Tests Hypothesis - Statement Regarding a Characteristic of a Variable or set of variables. Corresponds to population(s) –Majority of registered.
LSU-HSC School of Public Health Biostatistics 1 Statistical Core Didactic Introduction to Biostatistics Donald E. Mercante, PhD.
PSY 307 – Statistics for the Behavioral Sciences Chapter 20 – Tests for Ranked Data, Choosing Statistical Tests.
Sponsored by the Clinical and Translational Science Institute and the Department of Population Health Division of Biostatistics Concepts on the Way from.
BCOR 1020 Business Statistics Lecture 22 – April 10, 2008.
Chapter 11 Multiple Regression.
Lecture 10 Comparison and Evaluation of Alternative System Designs.
Chapter 9 Hypothesis Testing.
BCOR 1020 Business Statistics
DATA QUALITY and ANALYSIS Strategy for Monitoring Post-fire Rehabilitation Treatments Troy Wirth and David Pyke USGS – Biological Resources Division Forest.
Sample Size Determination Ziad Taib March 7, 2014.
Qian H. Li, Lawrence Yu, Donald Schuirmann, Stella Machado, Yi Tsong
Chapter 12 Inferential Statistics Gay, Mills, and Airasian
Difference Two Groups 1. Content Experimental Research Methods: Prospective Randomization, Manipulation Control Research designs Validity Construct Internal.
1 STATISTICAL HYPOTHESES AND THEIR VERIFICATION Kazimieras Pukėnas.
Achieving and Demonstrating “Quality-by-Design” with Respect to Drug Release/dissolution Performance for Conventional or Immediate Release Solid Oral Dosage.
Ch 10 Comparing Two Proportions Target Goal: I can determine the significance of a two sample proportion. 10.1b h.w: pg 623: 15, 17, 21, 23.
The Odds Are Against Auditing Statistical Sampling Plans
T tests comparing two means t tests comparing two means.
CHAPTER 16: Inference in Practice. Chapter 16 Concepts 2  Conditions for Inference in Practice  Cautions About Confidence Intervals  Cautions About.
Notes for Candidates Writing a Practical Report (Unit 2543)
Patterns of Event Causality Suggest More Effective Corrective Actions Abstract: The Occurrence Reporting and Processing System (ORPS) has used a consistent.
EBC course 10 April 2003 Critical Appraisal of the Clinical Literature: The Big Picture Cynthia R. Long, PhD Associate Professor Palmer Center for Chiropractic.
Section Copyright © 2014, 2012, 2010 Pearson Education, Inc. Lecture Slides Elementary Statistics Twelfth Edition and the Triola Statistics Series.
Inference and Inferential Statistics Methods of Educational Research EDU 660.
1 ORALLY INHALED AND NASAL DRUG PRODUCTS FOR LOCAL ACTION Current FDA BA/BE Background and Issues Wallace P. Adams, Ph.D. OPS/CDER/FDA OINDP Subcommittee.
Evaluating Impacts of MSP Grants Hilary Rhodes, PhD Ellen Bobronnikov February 22, 2010 Common Issues and Recommendations.
1 Nonparametric Statistical Techniques Chapter 17.
Educational Research Chapter 13 Inferential Statistics Gay, Mills, and Airasian 10 th Edition.
1 Basis of the Proposed Tactical Plan for a QbD approach for Quality Control and Assurance of Dissolution Rate Ajaz S. Hussain, Ph.D. Deputy Director,
ISECON 2006 The Work System Model as a Tool for Understanding the Problem in an Introductory IS Project Doncho Petkov Eastern Connecticut State University.
COMPARABILITY PROTOCOLUPDATE ADVISORY COMMITTEE FOR PHARMACEUTICAL SCIENCE Manufacturing Subcommittee July 20-21, 2004 Stephen Moore, Ph.D. Chemistry Team.
Guidelines for Critically Reading the Medical Literature John L. Clayton, MPH.
© Copyright McGraw-Hill 2004
Copyright © 2013, 2009, and 2007, Pearson Education, Inc. Chapter 10 Comparing Two Groups Section 10.1 Categorical Response: Comparing Two Proportions.
Aspect 1 Defining the problem - Problem: The design context will normally offer a variety of potential problems to solve. A focused problem and need is.
Educational Research Inferential Statistics Chapter th Chapter 12- 8th Gay and Airasian.
PSY 325 AID Education Expert/psy325aid.com FOR MORE CLASSES VISIT
1 Nonparametric Statistical Techniques Chapter 18.
Chapter 12 REGRESSION DIAGNOSTICS AND CANONICAL CORRELATION.
Chapter 8 Introducing Inferential Statistics.
Regulatory Considerations for Approval: FDA perspective
JMP for Biosimilars: Tools for Analytical Similarity Sept 16, 2015 W
Richard K Burdick Elion Labs MBSW Meetings May 2016
Logic of Hypothesis Testing
Evaluation Requirements for MSP and Characteristics of Designs to Estimate Impacts with Confidence Ellen Bobronnikov March 23, 2011.
Sample Size Determination
JMP Discovery Summit 2016 Janet Alvarado
Research Proposal The Research Proposal must describe a quantitative research project and demonstrate your familiarity with the concepts and techniques.
Lecture Slides Elementary Statistics Twelfth Edition
Statistical Core Didactic
Discussion of presentations Issues 1. imbalanced sample size 2
Carrie O’Reilly, Ph.D., M.S.N., RN Touro University Nevada
Chapter 10: Analysis of Variance: Comparing More Than Two Means
Critical Reading of Clinical Study Results
Testing a Claim About a Mean:  Known
Sue Todd Department of Mathematics and Statistics
New FDA Guidance on Early Alzheimer’s Disease
Figure 1 Overview of biosimilar product development
Psych 231: Research Methods in Psychology
Establishment of acceptance criteria for comparability studies.
What are their purposes? What kinds?
Psych 231: Research Methods in Psychology
Covering Principle to Address Multiplicity in Hypothesis Testing
CHAPTER 16: Inference in Practice
Objective 2 Biomedical Research Methods
GL 51 – Statistical evaluation of stability data
Medical Statistics Exam Technique and Coaching, Part 2 Richard Kay Statistical Consultant RK Statistics Ltd 22/09/2019.
Presentation transcript:

Analytical Similarity Assessment: Practical Challenges and Statistical Perspectives Richard Montes, Ph.D. Hospira, a Pfizer company Biosimilars Pharmaceutical Sciences (Statistics)

2 Presentation Outline Background on demonstration of biosimilarity Tier 1 (Equivalence Testing) Issues in setting equivalence margins Unequal variances Imbalanced samples sizes Tier 2 (Quality Range Assessment) Proposed algorithm to derive k multiplier

3 FDA recommends a stepwise approach to demonstrate biosimilarity Scientific Considerations in Demonstrating Biosimilarity to a Reference Product (2015) “Totality of Evidence”

4 Structural and Functional Characterization M 5 methods M 6 methods M 4 methods M 3 methods M 2 methods M 1 methods II. Primary Structure III. Higher Order Structure I. Functional Activity VI. Drug Product Characteristics V. Product Related Substances and Impurities IV. Post- Translational Modifications

5 Determination of Critical Quality Attributes (CQA) and Tiered Ranking Tiered ranking where K is the number of criteria (e.g., attribute impact to biologic activity, relative abundance, etc.) Relevance to mechanism of action (MOA) Amenability to quantitative statistical analyses

6 FDA Framework for Tiered Statistical Analysis of Attributes Used in Biosimilarity Assessment TierDescriptionStatistical Treatment 1 MOST RELEVANT to mechanism of action (MOA) function of product clinical effects Formal equivalence testing 2 POTENTIALLY RELEVANT to mechanism of action function of product clinical effects Evaluation versus reference product quality ranges 3 LEAST RELEVANT to mechanism of action (MOA) function of product clinical effects or not amenable to quantitative comparisons Raw data and graphical comparison Chow, Drug Des 2014, 3:3; On Assessment of Analytical Similarity in Biosimilar Studies.

7 Tier 1: Equivalence Testing T= Biosimilar R = Reference TT TT RR RR LL UU HYPOTHESISLOWER LIMITUPPER LIMIT Null (inequivalence) Alternate (equivalence) Conclude EQUIVALENCE if both null hypotheses rejected CHALLENGE: How to select equivalence margins [  L,  U ]? A. Two One-Sided Tests (TOST) Conclude EQUIVALENCE if 90% CI of mean difference is within [, ] B. Confidence Interval approach

8 Tier 2: Quality Range Assessment Quality assessment based on control charting concept. 2. If a large percentage (e.g., 90%) of Biosimilar lots are within Quality Range, conclude as HIGHLY SIMILAR CHALLENGE: How to select k?

9 Tier 3: Qualitative assessment graphical and tabular descriptive statistics raw analytical measurement outputs e.g., chromatograms no formal quantitative statistical analyses

10 Limit Tier 1 attributes to the most relevant to mechanism of action (MOA) Pairwise correlation matrix of various Functional Assays In Vivo Biopotency is most relevant to MOA Some functional assays are overlapping (orthogonal). Rank as Tier 1: In Vivo Biopotency In Vitro Specific Activity = [In Vitro Biopotency / Protein] Rank as Tier 2 or 3: all others

11 Practical challenges / statistical considerations – Basis for setting Equivalence Margins

12 Effect Size* as an alternative to Mean Difference HYPOTHESISLOWER LIMITUPPER LIMIT Null (inequivalence) Alternate (equivalence) HYPOTHESISLOWER LIMITUPPER LIMIT Null (inequivalence) Alternate (equivalence) MEAN DIFFERENCE EFFECT SIZE* * Burdick et al. (DIA Statistics Forum 2015) In practice,

13 Power Type I Error Case: Equal Sample Sizes, Equal Variances

14 Practical challenges / statistical considerations – Unequal Product Variances

15 T = Biosimilar R = Originator a post-approval commitment to implement control strategies to improve biosimilar variability may address this

16 (* ) splitting – use sample subset to establish margins, use remainder for TOST

17 Power Type I Error

18 Power Type I Error Using imbalanced sample sizes can unduly influence the similarity conclusion.

19 Use n R* =min(1.5n T,n R ) Power Type I Error

20 Summary on Tier 1 Assessment

21 FDA requires justification of selected k Cited assay variability, nature and criticality of attribute as determinants of k but no prescriptive approach Tsong 1 et al., recommended: k can be chosen from 2~3 based on the targeted coverage. (k =2, for 95%; k = 2.5 for 99%; and k = 3 for 99.7%) Currently requires 90% of biosimilar lots covered by the quality range ( 1 ) Tsong et al., “Development of Statistical Approaches for Analytical Biosimilarity Evaluation”. DIA Statistics Forum 2015

22 Issues with the Tier 2 framework

23 Proposed customized algorithm to derive k Part I Part II

24 Part I: Simulation of lots and Tier 2 assessment repeated 10,000 times Continue to Part II

25 Continued from Part I

26

27 Selected k for different mean shifts for 99% confidence, 90% proportion Indexed against multiplier for a 99% confidence / 90% proportion tolerance interval

28 Selected k for different mean shifts for 95% confidence, 90% proportion Indexed against multiplier for a 95% confidence / 90% proportion tolerance interval

29 Proposed algorithm is more flexible than a traditional tolerance interval

30 Algorithm adapted to attribute criticality and variances inequality (if justifiable) VariancesConfidence (%)k More critical attribute uses 95% confidence (lower k).

31 Link Tier 1 and Tier 2 assessments (k=3 vs k=custom at 99% confidence) Varies from 4.5 to 2.4

32 Link Tier 1 and Tier 2 assessments (k=3 vs k=custom at 95% confidence) Varies from 3.4 to 2.1

33 If Biosimilar product is assessed to have <90% of lots within Quality Range, it can be deduced:

34 Practical utility of the algorithm Serves as grid to vet whether the underlying assumptions are met or not Points to the sponsors what areas to further investigate If biosimilar product variance is larger than reference product, what is the root cause and how can it be addressed? If mean shift is larger than allowable, what is the root cause? Is the mean shift practically relevant? What process changes can be implemented?

35 Summary of Tier 2 Assessment No prescriptive approach for setting k multiplier Current Control Charting concept paradigm only tests whether Biosimilar lots are from the same population as Reference products treats problem as single population instead of two distinct population comparison Proposed algorithm to set k accounts for mean shifts and product variances inequality (if justified), sampling variability, and criticality of the attribute

36 References Burdick, R.K. and Ramirez, J.G. (2015), “Statistical Issues in Biosimilar Analytical Assessment: Perspectives on FDA ODAC Analysis”. Presentation at DIA Statistics Forum. April North Bethesda, MD Chow, S.C. (2014). “On Assessment of Analytical Similarity in Biosimilar Studies”. Drug Des 3: 119. doi: / e124 Tsong, Y., Shen, M., Dong, C. (2015), “Development of Statistical Approaches for Analytical Biosimilarity Evaluation”. Presentation at DIA Statistics Forum. April North Bethesda, MD

37 Acknowledgments Richard Burdick (Elion Labs) Effect Size concept and SAS code Aili Cheng, Pfizer PharmSci & PGS Statistics Imbalanced sample size discussion