Analytical Similarity and comparability – a point of view of industry.

© PharmaLex

Bruno Boulanger | CSO EFSPI | Working Group

NCS2018 3-5 October 2018

EFSPI Team

Bruno Boulanger, PharmaLex Christophe Agut, Sanofi Armin Boehrer, Boehringer-Ingelheim Mike Denham, GSK Piet Hoogkamer, Abbott Franz Innerbichler, Novartis Martina Kron, Abbvie Beate Krueger, Boehringer-Ingelheim Jens Lamerz, Roche Timothy Mutsvari, Arlenda Christian Seifert, Boehringer-Ingelheim

A bit of history

2015 − April - Invited at ASA-FDA-Industry 2015 meeting to give talk on analytical similarity. − October - EFSPI working group on Analytical Biosimilarity launch

2016 − Talk at NCS2016 by Timothy Mutsvari

2017 − March – Reflection paper published by EMA − April (Re)invited at ASA-FDA-industry meeting − September, – FDA draft Guidance on Analytical Similarity − November, Debate with FDA at IABS-FDA meeting

2018 − March, EFSPI submitted feedback about EMA reflection paper − May, Contributed to the EMA meeting on May 3-4 − June, FDA draft guidance withdrew

© PharmaLex 4

Some news in 2018 …

Reasons for withdrawal… “…..Comments submitted to the docket addressed a range of issues that could impact the cost and efficiency of biosimilar development, including the number of reference product lots the draft guidance would recommend biosimilar developers sample in their evaluation of high similarity and the statistical methods for this evaluation…. “

Unified statistical methodologies ?

Challenge from EMA Reflection Paper:Unified set of recommendations about statistical methodologies for three different questions:

Comparability of processes after a change Small Large Biosimilar product ------ Large Generic product Small -----

What makes them different or common?

Process Large molecule Specifications known Long history − Same assays − Few assays − Clinical data available

Product Small molecule Specifications unknown Short history − New assays − Many assays − No clinical data available

1 - Process or Product ?

When dealing with CMC and Quality Attributes: Is the central question about comparing processes or comparing products ? Patients receive individual batches Individual batches will be released to patients in the future The lots are the experimental units and central to the question − By contrast, in a clinical trial the patients are the experimental units used to estimate

the efficacy/safety of a product.

The same applies to large molecules and small molecules formulation processes

1 - Process or Product ?

When dealing with CMC and Quality Attributes: Should the “acceptance limits” apply − to the Process and individual units ? − to the Product and the means and/or the variances?

How to justify clinically defendable limits for mean or variance of process? Should the decision be made on current (past) batches or on future “capability” to produce lots within “acceptance limits” given observations. The range of the batches is important for the patient safety and efficacy.

2- Specifications and acceptance limits

In pre/post manufacturing change − the specifications are known and constant values.

For biosimilars − specifications are (by definition) unknown − should be established and justified and therefore are random variables.

In pre/post manufacturing change − specifications are about individual batches. − Why should it be different for biosimilars

How to map from specifications on individual batches to acceptance limits on parameters such as mean? For small molecules, there are already several “good practices” fixed limits defined (“80%-125%” rule, CU, 98%-102%, ….)

3 – Long history vs limited history

In pre/post manufacturing changes − Reduced list of Quality Attributes − Many “reference” batches available and few “test” batches usually envisaged − Number of “test” batches not really a matter of debate

In Biosimilars − Large list of Quality Attributes − Several “reference” and several “test” batches are required − Sample size computation of probability of success is a matter of debate

When Biosimilar company evaluates Reference products − not always sure about the independency of batches, age, etc…

4- Same assays or new assays

In pre/post manufacturing changes − the assays used are the same and therefore consistency of results is assumed − Assay variance wrt process variance is “known” − The format of the reportable results is (should) be appropriate

In biosimilars − New assays need to be developed and validated − Head to head assays should be envisaged − What is the contribution of assay component and format

For generics − Mostly physico-chemical procedures whose overall performance are less an issue

5 – Many QAs or limited number of QAs

In pre/post manufacturing changes − the number of Quality Attributes to be evaluated is reduced − The list is less prone to debate given history − The multiplicity issue is limited

In Biosimilars − The number of Quality Attributes to be evaluated is large − The list is a matter of debate and agreement − The multiplicity issue is rather important

Generics − Same as in pre/post manufacturing changes

6 – Clinical data available

If a “reference” product is on the market − it is within specifications − It is clinically acceptable

The range of values obtained for “reference” batches − Are by definition clinically acceptable values and justified − Do applies to the individual batches and are natural “acceptance limits” − How to figure out the real range of values patients are exposed to ?

How can “acceptance limits” be built for the mean or variance based on range of individual batches ? − Another arbitrary constant such as 1.5 or 1/8 should then be invoked for biosimilars − For generics, there are already criteria established since a long time (eg 80%125%) that have

a proven relevance

Is it possible to find a unified statistical methodology given those differences ?

What do we mean by similar/comparable?

Demonstrate that proposed new process produces lots of Test product that are (analytically) “equivalent/comparable” to those of the Reference product (both now and in the future).

When are the two distributions equivalent/comparable?

What do we mean by similar/comparable?

Demonstrate that proposed new process produces lots of Test product that are (analytically) “equivalent/comparable” to those of the Reference product (both now and in the future).

When are the two distributions equivalent/comparable?

What do we mean by similar/comparable?

Demonstrate that proposed new process produces lots of Test product that are (analytically) “equivalent/comparable” to those of the Reference product (both now and in the future).

When are the two distributions equivalent/comparable?

A Definition of Biosimilarity

The test product is analytically comparable (for a given attribute) to the reference product if the middle P% of all lots produced by the Test product process lie within the middle P% of the lots produced by the Reference product process. In what follows we will use 99%.

A Definition of Biosimilarity

Combinations of Mean and SD that would be considered Biosimilar

A Decision Procedure for Biosimilarity (1)

Interested in limits defined by central portion of distribution of Reference product lots Mean and variance of Reference estimated with uncertainty The β-content γ-Confidence Tolerance Interval (TI) on Reference is recommended Takes into account the uncertainty on the Mean and the Variance Better statistical properties than Min and Max Both Content and Confidence can be controlled A minimum sample size of Reference is recommended to make β-content γ-Confidence Tolerance Interval (TI) relevant for Similarity limits. (Here we will use 15)

( ){ } γβ =>+<<− RefRefRefRefRefsXskXXskXPPk RefcRefcRefXsXc ,: ,

RefskX cRef ×±

• In what follows we will use 99%-Content 95%-Confidence Tolerance Intervals

A Decision Procedure for Biosimilarity (2)

Test if β-Prediction Interval (PI) of biosimilar is within β-γ-Tolerance Interval (TI) of reference

Equivalent to a 100β% Credible Interval based on Posterior Predictive Distribution of X given the observed data using a Jeffreys Prior More relevant than using an arbitrary c factor (such as 3!) Takes into account the variability of the Test process (between-lots) Takes into account uncertainty on means and variability of new process Demonstrates that Test lots will be within the range of Reference lots with some level of confidence even in the future

( ) ( ) TestTestnTest nstXTest

/111,2/1 +×± −+β

A Decision Procedure for Biosimilarity (3)

Test if β-Prediction Interval is within β-γ-Tolerance Interval

Other Decision Procedures – FDA Tier Approach

Tier 1 – Most Critical (1-2α)100% two-sided Confidence Interval for Difference in Means contained within +/-1.5𝑠𝑅𝑅𝑅

Tier 2 – Moderate Critical Quality Range Method: mean +/- k 𝑠𝑅𝑅𝑅

Tier 3 – Least Critical Raw Data/Graphical Comparison

Compares the means of the two distributions

Compares the central portions of the two distributions

No ‘formal’ assessment of the two distributions

Illustration of FDA Tier approach - Pass

Demonstrate the Operating Characteristics (1)

Simulate or derive the performance of the decision rule for different combinations of the Mean and SD of the Test Product Process E.g. − Assume Reference Mean = 100, Reference SD = 1 − # Reference Lots = 15 − # Test Lots = 5, 10, 15, 20, 25

Decision methods:

FDA Two-Sided 90% Confidence Interval of Mean Difference FDA 90% of Test Lots in Mean +/- 3 SD Proposal β PI within β/γ TI (80% and 98% chosen here)

Demonstrate the Operating Characteristics (3)

Patient Risk

Producer Risk (Good Batch * reject Prob)

The objective: is my process biosimilar?

How to make a decision ?

A

B

What is the probability of obtaining the observed data, if the process is not biosimilar?

What is the probability being biosimilar, given the observed data?

2 / 18

Two different ways to make a decision based on

A Pr 𝐨𝐨𝐨𝐨𝐨𝐨𝐨𝐨 𝐨𝐝𝐝𝐝 𝐧𝐨𝐝 𝐨𝐛𝐨𝐨𝐛𝐛𝐛𝐛𝐝𝐨 )

Better known as the p-value concept

Used in the null hypothesis test (or decision)

This is the likelihood of the data assuming an hypothetical explanation (e.g. the “null hypothesis”)

Classical statistics perspective (Frequentist)

B Pr 𝐨𝐛𝐨𝐨𝐛𝐛𝐛𝐛𝐝𝐨 𝐨𝐨𝐨𝐨𝐨𝐨𝐨𝐨 𝐨𝐝𝐝𝐝 )

Bayesian perspective

It is the probability of efficacy given the data

3 / 18

Bayesian principle

After having observed the data of the study, the prior distribution of the treatment effect is updated to obtain the posterior distribution Instead of having a point estimate (+/- standard deviation), we have a complete distribution for any parameter of interest

P(treatment effect > 5.5)= P(success)

0 2 4 6 8 10

0.0

0.1

0.2

0.3

0.4

0.5

0 2 4 6 8 10

0.0

0.2

0.4

0.6

0.8

1.0

1.2

PRIOR distribution STUDY data POSTERIOR distribution

∝ +

Difference Simulations/Predictions

Simulations the “new observations” are drawn from distribution “centered” on estimated location and dispersion parameters (treated as “true values”).

Predictions the uncertainty of parameter estimates (location and dispersion) is taken into account before drawing “new observations” from relevant distribution

Why Bayesian for Biosimilarity

What is the question? − P( Biosimilar | Data ) vs P( Data | assuming not biosimilar) − what is the predictive probability having future lots within the limits given available data.

The Bayesian approach provides naturally the predictive distribution of future observation: − function of parameters and their uncertainty.

The parameter estimation is an intermediate steps to figure out future lots The Bayesian approach can easily handle multivariate problems Simplification occur drastically when moving from comparison of interval to predictive probability: − the number of dimension doesn’t matter anymore.

PI/TI - Integrating under the predictive dist.

Lower_TI Upper_TI

Predictive Distribution

Pr[Lower_TI ≤ Predictive Dist. ≤ Upper_TI] ≥ 0.8

PI/TI - Integrating under the predictive dist

By Integrating

PI/TI - Integrating under the predictive dist

By Integrating

Prior Distributions

Assume a conjugate joint prior distribution for 𝑃(𝜇𝑇,𝜇𝑅 ,𝜎𝑇 ,𝜎𝑅)

𝑃 𝜇𝑇 , 𝜇𝑅,𝜎𝑇 ,𝜎𝑅 = 𝑃 𝜇𝑇 ,𝜎𝑇 × 𝑃 𝜇𝑅,𝜎𝑅 = 𝑃 𝜇𝑇|𝜎𝑇 𝑃 𝜎𝑇 × 𝑃 𝜇𝑅|𝜎𝑅 𝑃 𝜎𝑇

Prior Distributions

Assume a conjugate joint prior distribution for 𝑃(𝜇𝑇,𝜇𝑅 ,𝜎𝑇 ,𝜎𝑅)

𝑃 𝜇𝑇 , 𝜇𝑅,𝜎𝑇 ,𝜎𝑅 = 𝑃 𝜇𝑇 ,𝜎𝑇 × 𝑃 𝜇𝑅,𝜎𝑅 = 𝑃 𝜇𝑇|𝜎𝑇 𝑃 𝜎𝑇 × 𝑃 𝜇𝑅|𝜎𝑅 𝑃 𝜎𝑇

Normal

Prior Distributions

Assume a conjugate joint prior distribution for 𝑃(𝜇𝑇,𝜇𝑅 ,𝜎𝑇 ,𝜎𝑅)

𝑃 𝜇𝑇 , 𝜇𝑅,𝜎𝑇 ,𝜎𝑅 = 𝑃 𝜇𝑇 ,𝜎𝑇 × 𝑃 𝜇𝑅,𝜎𝑅 = 𝑃 𝜇𝑇|𝜎𝑇 𝑃 𝜎𝑇 × 𝑃 𝜇𝑅|𝜎𝑅 𝑃 𝜎𝑅

Normal I-Gamma

Next steps and challenges

PI/TI seems robust in presence of outliers Developing a PI/TI approach for other distributions such as bi-modal, ordinal, rates/beta (defects, purity). Reference may not be independent. How to asses and evaluate in those conditions? Integrating under the PI instead of using the Interval

Take away

Revisit the question – It is not about the “averages” Similarity should be proven whatever the future lots PI/TI seems robust in presence of outliers Problem underestimated: variability of analytical methods and bioassays Draft publication for EFSPI WG is in progress Waiting for next steps from Regulatory – EMA or FDA

When SIMILAR is not the SAME!

From Timothy Mutsvari, NCS2016

All content is based on the current science and best practices and is subject to change.This document is property of the PharmaLex Group and is intended for the client only. Unauthorized use, disclosure, or copying of the information contained in this document is prohibited.

44

Bruno Boulanger Chief Scientific Officer +32 476 813936 Bruno.boulanger@arlenda.com PharmaLex Belgiim 5 Rue Edouard Belin 1435 Mont-Saint-Guibert Belgium

© PharmaLex