1

The Analytical Method Transfer Process

PDA AMD-AMV Workshop Baltimore, MD

7-8 October 2013

Stephan O. Krause, Ph.D.

Principal Scientist, Regulatory Science, Development

MedImmune

2

The Analytical Method Transfer Process

Agenda:

Introduction to Analytical Method Transfer (AMT)

Mapping out the Overall Product Development and Tech Transfer Process

Method Types and Acceptance Criteria

Analytical Platform Methods vs. New Methods

AMT Example

Krause/PDA, 2012

3

The Analytical Method Life Cycle

Krause/PDA, 2012

An

aly

tica

l Me

tho

d D

ev

elo

pm

en

t

A

na

lytic

al M

eth

od

V

alid

atio

n

(Po

st-V

alid

atio

n) L

ife

Cy

cle

Ste

ps

Select and Design – Establish Intended Use of Analytical Procedure

Development and Optimization

Performance Review, Qualification

Validation Acceptance Criteria

Validation

Post-Validation Life Cycle Steps

Transfer of Methods

Validation Prerequisites Assessment

IdentityImpurity

LimitImpurity Quantity

Assay / Potency

Tech Transfer

Resource Assessment

Standards and Controls

StabilityVerify Product Specifications

Maintenance Transfer Comparability Study

OOS/Valiation Failures

CQA Development, CMC Changes, Specifications

4

FTIH POC BLA

Tox StudiesPhase 1

Phase 2Phase 3

Clinical ResupplyMfg/Formulation Change(s)

Specifications Revision(s)

Target Quality CriteriaCommercial

Specifications

Negotiations, Final Commercial Specifications

QTPP

Final CQAs & Control Strategy Approval

Potential CQAsProduct & Process Design

Life-CycleManagement

POST-APPROVALCHANGES

PHASE 3PHASE 1/2Pre-IND

CQ

A D

evel

op

men

t(Q

bD

Pro

cess

)S

pec

s L

ife

Cyc

le

Mg

mt

CM

C a

nd

Tec

h

Tra

nsf

er P

roce

ss Analytical

Manufacturing

Strategic or Tactical Changes

Method qualification

Dose change

Delivery Device

PQ lots

Setting of Initial Specifications

Specifications Review/Confirmation

Mfg Transfer

Method validation

Method transfer

Formulation Change

Process Verification

Method Maintenance

Global Supply

Method transfer

5

The Five General AMV/AMT Classes

AMV Class Description Typical Risk /

Uncertainty Level (1=Low, 5=High)

Suggested Prospective AMV Studies

AMV Class No.

Analytical MethodProduct /

Process Sample

A New New 4-5 Full Validation

B New Old (Validated) 3-4(1) Full Validation Plus AMR(2) Studies

CAnalytical Platform

Technology (not validated “as run”)

New 2-3 Partial Validation

D Old (Validated) New 1-2Partial Validation or

Verification

E Compendial New 1-2Verification per USP

<1226>

(1) If a new analytical method (forced method replacement) is needed due to supply reasons, the risk level can be generally considered higherbecause no other option may exist. Unforced test method replacements can be considered to be a lower risk level as more time may be availableto optimize the method performance.

(2) AMR = Analytical Method Replacement. A study to confirm that a new analytical method can perform equally or better than the existing one.

From Krause, PDA/DHI 2007.

6

Risk-Based AMT Protocol Acceptance Criteria

Specifications

Consider Type of

Specifications

Acceptance Criteria

Existing Knowledge

One-Sided Specifications(NMT, NLT, LT)

Two-Sided Specifications

(Range)

Regulatory Requirements

Historical Method

Performance

Historical Data from this

Product and Process

Knowledge from Similar Product and

Process

Krause/PDA, 2012

7

General AMT Strategy

Krause/PDA, 2012

Co-validation/Co-qualification – this may be used early in the life cycle of a test method when appropriate.

Comparative study – AMT study performed concurrently by sending and receiving laboratories. Acceptance criteria determine the equivalence of the two laboratories. Historical and validation data may be used when appropriate for parts of the method transfer study. The sending laboratory typically has collected a significant amount of historical data for test method performance results in addition to test results for the samples to be tested at the receiving laboratory.

Performance Verification - The receiving laboratory may already perform the method for a similar product or for another type of sample for the same product. In this case, a formal method transfer may not be required. Any reduced prospective study considered should be properly justified.

8

Retrospective and Prospective Use of Data for AMV Studies from other Processes Prior to AMV – New Method

Method Qualification

(AMQ)

Method Validation (AMV)

Method Transfer (AMT)

AMQ Studies

ICH Q2(R1) AMV

Studies

PVFTIH BLA

Historical Data - SU

Assay Control

Tech Transfer

Interm. Precision & Reprod.

Historical Data - RU

Assay Control

“Approved” Method

Platform Method

9

Retrospective and Prospective Use of Data for AMV Studies from other Processes Prior to AMV – Platform Method

Method Qualification

(AMQ)

Method Validation (AMV)

Method Transfer (AMT)

(Less)AMQ

Studies

“Verification” Focus on: Accuracy, Specificity

PVFTIH BLA

Historical Data - SU

Assay Control

Tech Transfer

(Less) Interm.

Precision & Reprod.

Historical Data - RU

Assay Control

“Approved” Method

10

“Fixed” vs. “Variable” AMT Execution Plan

A fixed execution plan does not integrate test method result variation and the minimum number of test replicates needed to obtain a desirable statistical confidence in the transfer results. A fixed execution matrix can be more advantageous when transferring multiple products to/from multiple locations.

A variable execution plan does consider test method result variation and may require a larger data sets, especially for test methods with relatively high result variation. For example, a variable execution plan may be advantageous when transferring bioassays with an expected high degree of test result variation.

Krause/PDA, 2012

11

Typical “Fixed” AMT Execution Plan for Late-Stage/Commercial Products

Laboratory Day Analyst Instrument Replicates Sending 1 1 1 3 Sending 1 2 2 3 Sending 2 1 1 3 Sending 2 2 2 3 Sending 3 1 1 3 Sending 3 2 2 3 Receiving 1 1 1 3 Receiving 1 2 2 3 Receiving 2 1 1 3 Receiving 2 2 2 3 Receiving 3 1 1 3 Receiving 3 2 2 3

From Krause, PDA/DHI 2007.

12

Analytical Method Transfer (AMT) Example

A validated analytical method for potency is to be transferred from the original QC laboratory to another QC laboratory to release drug product (DP). The analytical method generates potency (dose) results for lyophilized DP.

The vials are available in three nominal doses between 500 – 2000 IU/vial using an identical formulation. Release testing is performed using three replicate preparations from each of three vials.

Before analysis the content of a vial is reconstituted with 5.0 mL of WFI water and the potency is measured in IU/mL (100 – 400 IU/mL).

The samples and a product-specific reference standard are prepared similarly. The analytical method procedure and statistical evaluation are performed with the parallel-line concept.

The “variable” AMT execution plan is used.

13

AMT Study Design and Acceptance Criteria

Characteristics Evaluated

Accuracy/Matching:The relative difference between lab means should at 90% confidence not be less than -Θ= 10% and not more than +Θ = 10%. The 10% difference limit was set with consideration of product specification.Intermediate Precision:RSD 6 % for all sample types, with appropriate homoscedasticity throughout the potency range (from validation results). This means that any RSD from a sample of n=8 should not exceed 9.43 %

Number of Replicates

Nreplicates = at least 23 independent replicates

The confidence interval for the lab-to-lab difference for N determinations to less than the [10%, +10%]. As above the 10% difference limit was set with consideration of product specification.

Samples to test Nlevel = 3

The range of potency/dosing results is covered by: Lowest dose 500 IU/vial or 100 IU/mLMedium dose 1000 IU/vial or 200 IU/mLHighest dose 2000 IU/vial or 400 IU/mL

Testing design, each sample

Number of operators, n = 2Number of days, n = 2Number of replicates per day per operator, n = 2N = 8 in each lab for each of n= 3 potency levels. Results are converted to “% recoveries vs. expected” to allow pooling Total NTotal = 24 individual observations will be recorded for each laboratory. N=24 individual

observations are needed as N=23 is the minimum number of replicates calculated.

14

AMT Results from Sending and Receiving Labs Theoretical

Potency Level in IU/mL

Operator Day Replicate

Sending lab Receiving lab

Experimental Potency in IU/mL

%Recovery vs. Theoretical

Potency

Experimental Potency in IU/mL

%Recovery vs. Theoretical

Potency

100 1 1 1 103 103.0 95 95.0

100 1 1 2 104 104.0 99 99.0

100 1 2 1 108 108.0 104 104.0

100 1 2 2 101 101.0 103 103.0

100 2 1 1 94 94.0 93 93.0

100 2 1 2 99 99.0 96 96.0

100 2 2 1 102 102.0 92 92.0

100 2 2 2 104 104.0 100 100.0

200 1 1 1 212 106.0 208 104.0

200 1 1 2 208 104.0 192 96.0

200 1 2 1 191 95.5 199 99.5

200 1 2 2 201 100.5 195 97.5

200 2 1 1 204 102.0 208 104.0

200 2 1 2 206 103.0 211 105.5

200 2 2 1 198 99.0 203 101.5

200 2 2 2 200 100.0 183 91.5

400 1 1 1 375 93.8 383 95.8

400 1 1 2 401 100.3 401 100.3

400 1 2 1 408 102.0 389 97.3

400 1 2 2 388 97.0 391 97.8

400 2 1 1 402 100.5 408 102.0

400 2 1 2 415 103.8 421 105.3

400 2 2 1 406 101.5 415 103.8

400 2 2 2 410 102.5 403 100.8

15

AMT Result Summary from Sending and Receiving Labs

(1) Raw data was used unrounded. Upper and lower 90% CIs were calculated using equation below .

(2) N=24 data points are not independent.

Separate and Pooled Potency Levels Evaluated

Sending lab Receiving lab

Statistical Parameters

%Recovery vs. Theoretical

Potency

Statistical Parameters

%Recovery vs. Theoretical

Potency

TOST with acceptance criteria [-10%, +10%] (1)

N1 24(2) N2 24

Mean1 101.1 Mean2 99.3

SD 3.5(2) SD 4.2

RSD 3.4 RSD 4.2

Pooled SD(2) 3.9

Mean1-Mean2 1.8

t-value 1.679

Upper 90% CI limit(1) 4 (3.6)

Lower 90% CI limit(1) 0 (-0.1)

Transfer Acceptance Conclusion Pass

212,2121

11)(

21 nnstxx pnn

16

Graphical Representation of Potency Results Per Potency Level Between Laboratories

The boxes represent the 25th – 75th percentile distribution of the results for the two laboratories. Medians (line in the box) and means (cross in the box) are approximately centered while the medians are equidistant from the box hinges, providing a visual indication for a normal data distribution(s) among data points within each laboratory set.

One potential outlier (lower open circle outside of the whiskers) is observed in the sending lab, however, this does not change the overall interpretation for the demonstration of lab-to-lab equivalence.

The variation in the test results (wider 25th – 75th percentile boxes) appears to be higher in the receiving laboratory which may be attributed to less test method execution experience.

80

90

100

110

120

Sending lab Receiving lab

%R

eco

very

ver

sus

theo

reti

cal

po

ten

cy (

in%

)

17

Graphical Representation of the Combined Percent Recoveries Between Laboratories for all Concentration Levels

18

Summary

AMTs may occur at different product development stages.

AMT study execution matrix and the acceptance criteria should be developed based on risk(s).

Many thanks to:• Rashmi Rawat (Product Quality Reviewer, CDER)

• Pat Cash (Sr. Director, Analytical Biochemistry, MedImmune)

• Mark Schenerman (VP, Analytical Biochemistry, MedImmune)

• Martin Van Trieste (SVP, Quality, Amgen)

• Rich Levy (SVP, PDA)

• Pierre Douette, Ph.D., Eurogentec S.A., Belgium

• Michael Warncke, Ph.D., Bayer HealthCare, USA

• Earl K. Zablackis, Ph.D., Sanofi Pasteur, USA

Krause/PDA, 2012