clostridium septicum vaccines - circabc - welcome · of clostridium septicum vaccines ... the...

Egmond aan Zee, Netherlands15-16 September 2015

ENG

Validation of alternative/3Rs methods for the in-process

quality control of CLOSTRIDIUM

SEPTICUM VACCINES

BSP130 participants workshop report

N. Sinitskaya1, K. Redhead2, A. Daas3, L. Bruckner2 and M.-E. Behr-Gross3

1. Consultant.2. Project leaders.3. European Directorate for the Quality of Medicines & HealthCare, Council of Europe, Strasbourg, France.

Validation of alternative/3Rs methods for the in-process quality control of Clostridium septicum vaccines

BSP130 participants workshop report

Egmond aan Zee, Netherlands 15-16 September 2015

Correspondence address

Dr M.-E. Behr-Gross Department of Biological Standardisation OMCL Network & HealthCare EDQM, Council of Europe CS 30026 F-67081 STRASBOURG FRANCEE-mail: [email protected]

Visiting address

EDQM 7 allée Kastner

F-67000 Strasbourg

Prepared for

Department of Biological Standardisation, OMCL Network & HealthCare European Directorate for the Quality of Medicines & HealthCare (EDQM) Council of Europe 7 allée Kastner, CS 30026 F-67081 STRASBOURGFRANCEWebsite: www.edqm.eu

© Council of Europe, 2016Cover illustration © Fotolia – science photos

Validation of alternative/3Rs methods for the in-process quality control of Clostridium septicum vaccines – BSP130 participants workshop report is published by the European Directorate for the Quality of Medi-cines & HealthCare of the Council of Europe (EDQM).

All rights conferred by virtue of the International Copyright Convention are specifically reserved to the Council of Europe and any reproduction or translation requires the written consent of the Publisher.

Director of the Publication: Dr S. Keitel

Page layout and cover: EDQM

For further information concerning the work of the Council of Europe/EDQM in the area of the Biological Standardisation Programme please consult https://www.edqm.eu/en/BSP-Work-Programme-609.html.

Contents

INTRODUCTION………………………………………………………………………......... 5

KEYWORDS…………………………………………………………………………………. 6

ABBREVIATIONS…………………………………………………………………………… 6

PROGRAMME……………………………………………………………………………..... 7

SUMMARY OF KEY POINTS……………………………………………………………… 8

RECOMMENDATIONS…………………………………………………………………….. 9

ACKNOWLEDGEMENTS………………………………………………………………… 10

SUMMARY OF SESSIONS…………………………………………………………….... 11

Session A: BSP130 Study Outcome

A1. Introduction to clostridial vaccine control and potential for introduction of alternatives to animal testing: the case of Clostridium septicum…………………………………………. 11

Presentation……………………………………………………………….... 13

A2. The BSP130 collaborative study outline…………………………..... 25

Presentation……………………………………………………………….... 27

A3. The BSP130 collaborative study results and statistical analysis…………………………………………………….. 36

Presentation………………………………………………………………… 41

A4. Participant’s experience in the performance of the experiments of BSP130……………………………………………. 68

A5. Round table discussion and conclusions…………………………… 71

Session B: Follow up activity proposals

B1. A cell based assay as compendial test for assessing residual toxicity (MLD) and immunogenicity (TCP): consequences and issues…………………………………………………. 73

Presentation……………………………………………………………….... 74

BSP130 Participants Workshop

3

____________________________________________________________________

B2. Design of a potential follow-up study with full validation of an optimised protocol for the cell-based assays developed in BSP130

B2.1. Study design and participants……………………………... 83

B2.2. Samples and protocol………………………………………. 84

B2.3. Financial and logistical support to the follow-up study….. 85

Presentation………………………………………………………………… 86

B3. Possibility of using same cell culture based assay system for replacing second step of current potency assay in rabbits (i.e. indirect toxin neutralisation assay in mice)…………………………. 89

Presentation………………………………………………………………… 90

B4. Round table discussion – Meeting conclusions and summary…..103

APPENDIX………………………………………………………………………………… 105

REFERENCES……………………………………………………………………………. 106


4

____________________________________________________________________


15-16 September 2015 Hotel/Conference Center Zuiderduin

Egmond aan Zee, Netherlands

Report of the Workshop

INTRODUCTION

The BSP130 workshop took place on 15th and 16th September 2015 under the auspices of the Biological Standardisation Programme (BSP) of the European Directorate for the Quality of Medicines & HealthCare (EDQM) of the Council of Europe (CoE) and the European Partnership for Alternative Approaches to Animal Testing (EPAA). It was organised as a satellite meeting to the IABS Conference ‘3Rs alternatives and consistency testing in vaccine lot release testing' at the Hotel/Conference Center Zuiderduin in Egmond aan Zee, Netherlands. The BSP130 workshop was devoted to the outcome and follow-up activity proposals for an international collaborative study for the validation of cell line assays for use in in-process toxicity and antigenicity control of Clostridium septicum vaccine (BSP130). The workshop involved the study coordination team (2 project leaders, 1 statistician, 1 project coordinator), 8 representatives of the participant laboratories (veterinary pharmaceutical industry and of public sectors official medicines control laboratories) as well as 3 observers. Further to the workshop, the EDQM was charged with coordinating the drafting and the publication of a report for this workshop. This report was prepared under the direction of the project coordinator, the project statistician and the project leaders (see Appendix).


5

____________________________________________________________________

http://www.consistency-congress.org/

http://www.consistency-congress.org/

KEYWORDS

Clostridium septicum, clostridial vaccines, veterinary vaccines, Minimum Lethal Dose (MLD), Residual Toxicity, Total Combined Power (TCP), European Partnership for Alternative Approaches to Animal Testing (EPAA), European Directorate for the Quality of Medicines and HealthCare (EDQM), Biological Standardisation Programme (BSP), European Pharmacopoeia, Council of Europe (CoE)

ABBREVIATIONS

a: Slope of the tolerance curve B: Binding power of the toxoid BRP: Biological Reference Preparation BSP: Biological Standardisation Programme Cl.: Clostridium CSTx: Reference toxin CV: coefficient of variation Di: dilution of the toxoid on row i EDQM: European Directorate for the Quality of Medicines & HealthCare ELISA: enzyme-linked immunosorbent assay EPAA: European Partnership for Alternative Approaches to Animal Testing IS: International Standard IU: International Unit L: dilution of CSTx LD50: Lethal Dose 50 MLD: Minimum Lethal Dose N: Toxin equivalence of CSTx NC3Rs: National Centre for the Replacement Refinement & Reduction of Animals in Research OMCL: Official Medicines Control Laboratory Ph. Eur.: European Pharmacopoeia P: Pre-dilution of mix before plating QC: Quality Control S: Sensitivity of the Vero cells SOP: Standard Operating Procedure TCP: Total Combining Power


6

____________________________________________________________________

PROGRAMME

September 15th 2015

SESSION A: BSP130 Study Outcome

A1. Introduction to clostridial vaccine control and potential for introduction of alternatives to animal testing: the case of Clostridium septicum (Speaker Dr K. Redhead) A2. The BSP130 collaborative study outline (Speaker Dr L. Bruckner) A3. The BSP130 collaborative study results and statistical analysis (Speaker Mr A. Daas) A4. Participants’ experience in the performance of the experiments of BSP130 (All participants) A5. Round table discussion and conclusions (Chairs: Dr M-E. Behr-Gross, Dr L. Bruckner, Dr K. Redhead)

September 16th 2015

SESSION B: Follow up activity proposals

B1. A cell based assay as compendial test for assessing toxicity in residual toxicity (MLD) and immunogenicity (TCP) testing: consequences and issues (Speaker Dr L. Bruckner) B2. Design of potential follow-up study with full validation of an optimised protocol for the cell based-assays developed in BSP130

B2.1. Study design and participants B2.2. Samples and protocol B2.3. Financial and logistical support to follow up study

(Speakers Dr K. Redhead, Dr L. Bruckner, Mr A. Daas) B3. Possibility of using same cell culture based assay system for replacing second step of current potency assay in rabbits (i.e. indirect toxin neutralisation assay in mice) (Speaker Dr M-E. Behr-Gross) B4. Round table discussion – Meeting conclusions and summary (Chairs: Dr M-E. Behr-Gross, Dr L. Bruckner, Dr K. Redhead)


7

____________________________________________________________________

SUMMARY OF KEY POINTS

The following issues were addressed:

• The Clostridial vaccine control and the possibility of replacement of toxicity (Minimum Lethal Dose, MLD) and antigenicity (Total Combining Power, TCP) tests in mice by in vitro Vero cell-based assays.

• Objectives and structure of BSP130 International collaborative study. • The critical points identified for in vitro Vero cell-based assays: the edge effect

observed on the plates and their impact on data output; the storage conditions and stability of the “Detecting toxin”; the potential impact of cell passage number on cell sensitivity.

• Problems linked to data analysis and interpretation in the BSP130 study. • The Maximum Likelihood method which was used to calculate TCP/Binding

Power of toxoids and MLD/Toxin Equivalence. • The in-process toxicity testing and the possible replacement of MLD test by

Toxin Equivalence test. • The possible use of an “assay simulator” to optimise design of follow-up study

protocol. • Objectives and structure of the follow-up study. • Requirements for supply of testing materials for the follow-up study. • The possibility of, and need for running a study on replacing the second step

of the current potency assay in rabbits by a cell-based assay (based on the same principles as TCP/MLD alternatives developed in BSP130).


8

____________________________________________________________________

RECOMMENDATIONS

The following recommendations emerged:

• Vero cell based MLD and TCP assays should be promoted as replacements for the conventional mouse tests for Cl. septicum antigens.

• A follow up study with a protocol optimised for in vitro assays alone should be conducted to take full advantage of the sensitivity and accuracy of the Vero cell based assays.

• The potential for quantifying “Toxin Equivalence” by reference to a standard antitoxin instead of using the classical MLD assay result (expressed as a dilution) to enable the objective assessment of toxicity of different batches of toxin and their comparison should be further investigated.

• The possibility of application of these approaches to other cytotoxic clostridial toxins should be investigated by other groups (e.g. Vac2Vac consortium).

• The Vero cell based TCP (Binding Power) assay should be used to provide more accurate and reproducible assessment of toxoid antigenicity for use in the blending of more consistent and efficacious final vaccines.

• The replacement of the second step of the conventional clostridial vaccine serological potency test (i.e. the assessment of antibody content in rabbit antisera by toxin neutralisation in mice) by a cell line-based cytotoxicity assay should be considered, by those laboratories having not yet implemented and validated an in vitro methodology (e.g. ELISA).


9

____________________________________________________________________

ACKNOWLEDGEMENTS

The drafting of the report was performed by Dr Natalia Sinitskaya (consultant, funded by EPAA) under the supervision of Dr Marie-Emmanuelle Behr-Gross (Scientific Officer, EDQM). The project statistician Mr Arnold Daas, and the project leaders Dr Lukas Bruckner and Dr Keith Redhead are especially acknowledged for their active contribution to the elaboration of the programme for this workshop and for their expert advice in the drafting of this report. The organisers would like to express their sincere thanks to the Department of Biological Standardisation Programme (BSP) of the European Directorate for the Quality of Medicines & HealthCare (EDQM) and to the Clostridial expert working group of the European Partnership for Alternative Approaches to Animal Testing (EPAA) as well as to all participants from Control Authorities and manufacturers for their support to this initiative. Dr Irene Manou (EPAA) and Dr Coenraad Hendriksen (InTraVacc) are gratefully acknowledged for their instrumental support in the preparation and the organisation of the workshop. Ms Sally Woodward (Secretarial Assistant, EDQM) is acknowledged for expert secretarial and organisational support in all steps in the project and in the workshop preparation.

BSP130 Participants Workshop ____________________________________________________________________

10

SUMMARY OF SESSIONS

Session A: BSP130 Study Outcome

A1. Introduction to clostridial vaccine control and potential for introduction of alternatives to animal testing: the case of Clostridium septicum Speaker: Dr K. Redhead

The speaker reviewed background information on clostridial vaccines control and presented the principles of alternatives to animal testing as follows. All clostridial toxoid vaccines are produced according to the similar methods: toxins are produced by microorganisms growing in a liquid medium. Being extracellular, toxins are accumulated in supernatant which, once the cultivation period is finished, is separated from cells by means of centrifugation and/or filtration. At this stage of production, toxicity of toxins (Minimum lethal dose, MLD) is firstly analysed. Then toxins are directly inactivated chemically in supernatant to obtain toxoids which are analysed for their residual toxicity (Minimum lethal dose, MLD) and antigenicity (Total combining power, TCP). After that, toxoids are blended with other antigens and adjuvanted and undergo potency and safety testing. Finally, the vaccine is dispensed into containers.

Importantly, both currently used in-process control MLD (toxin/residual toxoid toxicity) and TCP (toxoid antigenicity) tests are performed in mice and require the use of large numbers of animals. The principle of the MLD test may be defined as: ‘how far might toxin or toxoid be diluted before it is no longer lethal in mice?’ By contrast the TCP test measures how much reference neutralising antitoxin is bound by the toxoid (the amount of active unbound antitoxin remaining is measured on the basis of its ability to neutralise a lethal amount of toxin). However, data provided by these in vivo tests are not fully provided by any of the current in vitro tests. This is why a direct and simple approach was applied for replacement of in-process in vivo tests. Since the principle of these tests is based on cytopathic effects of clostridial toxins where mice are used only as an indicator of toxicity, mice could be replaced by a different indicator, for instance, by cell lines sensitive to toxin. As regards the assay protocols, it was postulated that they could be performed exactly as for the in vivo tests but final toxicity would be determined in cell line instead of mice.

From this approach, replacement of in-process in vivo tests should go through the following steps: identification of a suitable cell line; confirmation of specificity; development of cell line assay including its optimisation; internal validation with regard to repeatability and intermediate precision; correlation with respective mouse test; and, finally, international collaborative study to validate the in vitro assays and assess concordance with in vivo tests.


11

expert

Texte tapé à la machine

expert


expert


expert


expert


According to the proposed outline the preliminary phase of the study was performed in one of the laboratories where Vero cells were first identified as suitable for development of cell line assays on microplates and where good correlation of in-process in vivo and developed in vitro Cl. septicum cell line assays was demonstrated (ρc=0.99 for MLD and ρ=0.99 for TCP). Moreover, it was shown that cell line assays would be more sensitive (16 fold), accurate (5 fold), allow high reproducibility and shorten duration of tests (24 hours instead 4 days) as well as representing advantages in savings in animal and costs.

An international collaborative study for validation of cell line assays for in-process toxicity and antigenicity of Clostridium septicum vaccine antigens was initiated with the support of BSP Steering Committee (EDQM) and the EPAA Clostridial expert working group. The aim was to promote the acceptance of in vitro alternatives to mouse in vivo tests for the in-process control testing of veterinary clostridial vaccine antigens. The international collaborative study aimed at performing in vivo and/or in vitro testing of toxins and toxoids from various sources and of different strengths by the assembled group of participants (manufacturers and OMCLs of different countries), to validate the in vitro assays and to assess concordance with the in vivo tests.


12

____________________________________________________________________

Cl. perfringens type A Cytopathic toxinCl. perfringens type B Cytopathic toxinCl. perfringens type C Cytopathic toxinCl. perfringens type D Cytopathic toxinCl. novyi type B Cytopathic toxinCl. septicum Cytopathic toxinCl. haemolyticum Cytopathic toxinCl. sordelli Cytopathic toxinCl. difficle Cytopathic toxinCl. tetani NeurotoxinCl. botulinum NeurotoxinCl. chauvoei Toxins + cells?


13

____________________________________________________________________

Grow organism in liquid cultureanalytical

Remove cells (centrifugation &/or filtration)

analytical, toxicity and antigenicity

Chemically inactivate toxins in supernatant

analytical, toxicity and antigenicity

Blend with other antigens and adjuvant

analytical, potency and safety

Dispense vaccineanalytical

In-Process: In vivo

Toxicity of toxin (Minimum lethal dose, MLD)

Toxicity of toxoid (MLD)

Antigenicity of toxoid (total combining power, TCP)


14

____________________________________________________________________

Assessed by the Minimum Lethal Dose (MLD) test using mice

How far can the toxin/toxoid be diluted before it is no longer lethal in mice


15

____________________________________________________________________

Antigenicity of the toxoid is assessed by the Total Combining Power (TCP) test using mice

How much reference neutralising antitoxin is bound by the toxoid

The amount of active unbound antitoxin remaining is measured on the basis of its ability to neutralise a lethal amount of toxin

- assessed in mice


16

____________________________________________________________________

The data provided by these in vivo tests i.e. toxicity and antigenicity of the toxin/toxoid are not fully provided by any of the current in vitro

tests


17

____________________________________________________________________

Test Indicator

MLD Mice

TCP Mice

Mice are used only as an indicator of toxicity

Replace the mice with a different indicator of toxicity –

Toxin-specific cell lines

Treated cell monolayer Control cell monolayer


18

____________________________________________________________________

Seed microplates with suitable cell line

Incubate to form confluent cell layer

Add dilutions of toxin or toxoid or toxin+antitoxin etc

Incubate

Visualise effects (staining of viable cells)

Assess effects (measure staining)

Determine end-points and calculate toxicity or antigenicity

Cell line assay plate


19

____________________________________________________________________

Identification of suitable cell line

Confirmation of specificity

Development of cell line assay (optimisation)

Internal validation (repeatability and intermediate

precision)

Correlation with respective mouse test

International collaborative study to validate the in vitro

assays and assess concordance with the in vivo tests.

In-house correlations of in-process in vitro and in vivo assays


20

____________________________________________________________________


21

____________________________________________________________________

Assay Correlation Linear Regression

MLD 0.99 0.99

L+ 0.87 0.80

TCP 0.99 0.99


22

____________________________________________________________________

Sensitivity (x16)

Accuracy (x5)

Reproducibility

Speed (24 hours vs 4 days)

Ethics

Cost

Assays performed exactly as for the in vivo tests but final toxicity determined by use of cell lines not mice

Relatively easy to demonstrate correlation with existing in vivo tests


23

____________________________________________________________________

Purpose:

To promote the acceptance (within Eur. Pharm. monograph guidelines and by Regulatory Authorities assessing Marketing Authorisation applications) of in vitro alternatives to mouse in vivo tests for the in-process control testing of veterinary clostridial vaccine antigens.

The tests selected for replacement are the Mouse Lethal Dose (MLD) and the Total Combining Power (TCP) and the potential replacement assays identified are cell line based.

Approach:Selection of one species of Clostridium for which the toxin and toxoid can be assessed.

Assemble a group of participants from manufacturing and OMCL backgrounds in different countries able to test this type of toxin and/or toxoid in the in vivo and/or in vitro assays.

Perform an international collaborative study, with toxins and toxoids from various sources and of different strengths, to validate the in vitro assays and assess concordance with the in vivo tests.


24

____________________________________________________________________

A2. The BSP130 collaborative study outline Speaker: Dr L. Bruckner

The speaker briefly recalled that the BSP130 collaborative study aimed to evaluate the transferability and performances of alternative methods to current in vivo mouse tests such as the Minimum Lethal dose (MLD) test used both to measure the toxicity of Cl. septicum toxin and to measure freedom from toxicity of Cl. septicum toxoid as well as the Total Combining Power (TCP) test which is used to measure the antigenicity of Cl. septicum toxoid. According to the proposed protocol, in vitro cell lines tests were performed exactly as for the in vivo tests but after the dilutions necessary for MLD and the toxoid, antitoxin and toxin mixing necessary for TCP, final toxicity was determined in a cell line instead of in mice. The reliability of Vero cell and mouse assays was assessed by means of intra-laboratory variation (as indicator of precision and repeatability of assays) and inter-laboratory variation (as indicator of reproducibility of assays). Finally, the correlation between the in vivo and in vitro assays was assessed. The BSP130 collaborative study was coordinated by Marie-Emmanuelle Behr-Gross (scientific secretariat, EDQM). Statistical analysis was performed by Arnold Daas (statistics, EDQM). Dr Keith Redhead and Dr Lukas Bruckner were designated as project leaders. 13 laboratories committed to participate: both industrial (CEVA, Hungary; MSD AH, UK, USA, New Zealand; CZV, Spain; SYVA, Spain, Zoetis, UK) and non-industrial (Bornova Vet Institute, Turkey; NEBIH, Hungary; PEI, Germany; CVB, USA; EDQM, Europe; IVI, Switzerland). Finally, 11 participants took part in the study and provided data.1

Study materials included: 6 batches of Cl. septicum toxins, 6 batches of Cl. septicum toxoids donated by manufacturers; Cl. septicum (gas gangrene) antitoxin, 3rd international standard (IS) and Cl. septicum reference/detecting toxin as standards and critical reference reagents. In BSP130 it had been intended to use in-house routine methods for in vivo MLD and in vivo TCP assay in mice and Standard Operating Procedures provided in the study protocol for in vitro MLD and in vitro TCP assay in Vero cells. Participants were requested to perform 3 valid assays with 5 three- or five-fold dilutions in 2 mice, and 3 valid assays with 5 three- or five-fold dilutions in 2 rows of cells per dilution.

The experimental part of BSP130 was divided into four parts. Importantly, participants were requested to report to EDQM all results, including those from invalid tests, after completion of each part of the study in the form of electronic data reporting sheets provided with the study protocol. Data were checked and the dilution

1 Due to lack of resources at the time of launching the study, Zoetis and MSD New-Zealand did not participate in the experimental part of the study.


25

factors used were approved prior to allowing the participants to continue the next part of the study.

Part 1, defined as the confirmation of sensitivity of mouse strains and cell lines, consisted of the determination of the initial pre-dilution for the reference toxin that would result in the killing of mice or the Vero cells. The reference toxin was then used at this pre-dilution as the reference toxin on all further relevant MLD assays.

In Part 2, defined as latent toxicity testing of test materials, all 6 toxoids (diluted 1 in 10) and the WHO International Standard Cl. septicum antitoxinxii (reconstituted and diluted to concentration of 5 IU/mL) were tested, at these final concentrations, for toxicity in mice and/or Vero cells, as appropriate, in MLD assays.

In Part 3, defined as preliminary ranging of test materials, the preliminary ranging tests for all 6 test toxins and all 6 test toxoids were conducted in vivo and/or in vitro, as appropriate, in relevant MLD or TCP assays. Centred on the approximate MLD or TCP value supplied for each toxin, participants were asked to perform dilution series.

Part 4, defined as full testing of test materials, consisted of testing of each test toxin and toxoid in the appropriate in vivo or in vitro assay. The testing was repeated on different days until a minimum of 3 valid assays had been completed for each test material in each test that was being assessed.

All participants were acknowledged for taking part in this international collaborative study. Manufacturers who supplied toxins and toxoids were also acknowledged. Arnold Daas (statistics, EDQM) was especially acknowledged for his expert contribution and for performance of the central statistical analysis.


26

05/01/2016

1

Lukas Bruckner ©2015 EDQM, Council of Europe. All rights reserved. 15 September 2015

THE BSP130 COLLABORATIVE

STUDY OUTLINE


AIMS OF THE STUDY

Evaluating transferability and performances of alternative methods to the current in vivo mouse tests used to measure

• the toxicity of Cl. septicum toxin

the Minimum Lethal dose (MLD) test

• the freedom from toxicity of Cl. septicum toxoid

the MLD test

• the antigenicity of Cl. septicum toxoid

the Total Combining Power (TCP) test


27

____________________________________________________________________

05/01/2016

2


EXPECTATIONS OF THE COLLABORATIVE

STUDY

The replacement in vitro assays were expected to be basically the same as the in vivo tests except that after the dilutions necessary for the MLD and the toxoid, antitoxin and toxin mixing and reactions necessary for the TCP the final materials are assessed for indications of toxicity not in mice but on a cell line.


RELIABILITY OF THE ASSAYS

The reliability of the Vero cell assays and of the mouse tests were to be studied by obtaining information on:

• intra-laboratory variation

precision and repeatability

• inter-laboratory variation

reproducibility


28

____________________________________________________________________

05/01/2016

3


RELATIONSHIP BETWEEN ASSAYS

The relationship between the Vero cell assays and of the mouse tests were to be studied by:

• looking for concordance between the relevant in vivo and in vitro assays


PROJECT TEAM

Project Coordinator

Keith Redhead

Lukas Bruckner

Support

Marie-Emmanuelle Behr-Gross

scientific secretariat

Arnold Daas

statistics


29

____________________________________________________________________

05/01/2016

4


PARTICIPANTS

Industrial

CEVA Hungary

MSD AH UK, USA, NZ

Pfizer AH (CZV) UK (Spain)

SYVA Spain

Non-Industrial

Bornova Vet Inst Turkey

NEBIH Hungary

PEI Germany

CVB USA

EDQM Europe

IVI Switzerland


MATERIAL

• Cl. septicum toxins

6 batches of Cl. septicum toxin

• Cl. septicum toxoids

6 batches of Cl. Septicum toxoid

• Standards and critical reference reagents

Standard antitoxin

Cl. septicum (gas gangrene) antitoxin, 3rd international standard (IS)

Reference/detecting toxin

Cl. septicum reference/detecting toxin


30

____________________________________________________________________

05/01/2016

5


METHODS

in vivo MLD assay in mice

in-house routine methods

in vitro Vero cell MLD assay

(SOP) given in the study protocol

in vivo TCP assay


in vitro Vero cell TCP assay

(SOP) given in the study protocol


METHODS

in vivo assays in mice


• MLD assay

• TCP assay

3 valid assays

• 5 three- or five-fold dilutions

• 2 mice per dilution

in vitro Vero cell assays

SOP given in the study protocol

• MLD assay

• TCP assay

3 valid assays

• 5 three- or five-fold dilutions

• 2 rows of cells per dilution


31

____________________________________________________________________

05/01/2016

6


PARTS OF THE STUDY

1) Confirmation of sensitivity of mouse strains and cell lines

2) Latent toxicity testing of test materials

3) Preliminary ranging of test materials

4) Full testing of test materials


1) CONFIRMATION OF SENSITIVITY OF

MOUSE STRAINS AND CELL LINES

Determination of an initial pre-dilution for the reference toxin for use on mice or the Vero cells, that would result in the killing of mice or the Vero cells, respectively.

The reference toxin was then used at this pre-dilution as the reference toxin on all further assays.


32

____________________________________________________________________

05/01/2016

7


2) LATENT TOXICITY TESTING OF

TEST MATERIALS

The standard Cl. septicum antitoxin was applied at a concentration of 5 IU/mL.

Each of the 6 Cl. septicum test toxoids was diluted 1 in 10.

All 6 toxoids and the standard antitoxin were then tested, at these final concentrations, for toxicity in mice and/or Vero cells, as appropriate.


3) PRELIMINARY RANGING OF

TEST MATERIALS

The preliminary ranging tests for all 6 test toxins and all 6 test toxoids were conducted in the in vivo and/or in vitro, as appropriate, MLD or TCP assays, respectively.

Centered on the approximate MLD or TCP value, respectively, supplied, participants were asked to perform dilution series.


33

____________________________________________________________________

05/01/2016

8


4) FULL TESTING OF TEST MATERIALS

For the full collaborative study each of the test toxins and toxoids was tested, in the appropriate in vivo or in vitro assay. The testing was repeated on different days until a minimum of 3 valid assays had been completed for each test material in each test that was being assessed.

All the results, including those from any invalid tests were reported.


DATA REPORTING

• Electronic data reporting sheets were provided

• Results were reported after completion of each part of the study to the EDQM statistician

• Data were checked and the dilution factors used were approved prior to allowing the participant to continue to the following part of the study design


34

____________________________________________________________________

05/01/2016

9


THANK YOU FOR YOUR ATTENTION


35

____________________________________________________________________

A3. The BSP130 collaborative study results and statistical analysis Speaker: Mr A. Daas

The speaker explained the statistical analysis approach used within the BSP130 collaborative study.

MLD

According to the design of the study protocol, the initial intention was that exactly the same doses of diluted toxins were to be applied on mice and on Vero cells in laboratories performing both in vitro and in vivo MLD tests. The intention of the MLD study was to determine the MLD of toxins in mice and on Vero cells, in order to find a correlation and define predictive endpoints. In the real study conditions, this approach turned out to be difficult to apply when laboratories changed pre-dilutions between assays (shift of the endpoint on plates), when there were no useful endpoints on the row corresponding to the mouse MLD or when laboratories only did the in vitro assays. At the same time, taking into account that powerful statistical optimisation techniques are available if responses are expressed as a function of known and unknown parameters, and the fact that some parameters are known from study design (e.g. the volume of pure toxin per well), the endpoints on each row should be consistent and could be combined in one value per plate, and so an alternative way to express the MLD could be applied.

Thus, it was decided to express the endpoint of in vitro MLD in toxin volume units needed to kill the Vero cells. In this case the dose-response function can be modelled to a minimum of 2 unknown parameters: the sensitivity (S) of the Vero cells (LD50 of CSTx) and the toxicity (N) of the test toxin. However, S and N are confounded parameters because higher sensitivity cannot be distinguished from higher toxicity. The number of unknown parameters can be reduced to only one parameter if toxicity is expressed relative to a reference toxin or if sensitivity is known from validation studies. Each plate included a row with the CSTx so it would have been possible to express toxicity of the test toxins relative to the CSTx per plate. However, the CSTx was intended as a control sample and not as a reference sample. It was therefore decided to establish the sensitivity of the cell lines for each laboratory based on the sensitivity tests carried out with CSTx (6 rows per plate). Later this established sensitivity (fixed S per lab) would be used to quantify toxicity (N) of the test samples. Once the sensitivity test with reference toxin (CSTx) was evaluated for mice and Vero cells, 12-fold toxin sensitivity ranges for in vivo assays and 24-fold toxin sensitivity ranges for in vitro assays were observed between laboratories. In that way it was demonstrated that data are quite variable between laboratories.

In the case of the MLD assay for toxins tested in mice, improvement in some, but not in all toxins was obtained when relative values (with respect to CSTx) were applied.


36

____________________________________________________________________

Thus, differences of up to a factor of 14 in absolute values (without correction for sensitivity) were observed between laboratories testing the same toxin, while when expressed as relative values (with respect to CSTx), there was still a difference factor of 7. The interaction between toxicity and cell sensitivity was too large to allow for a satisfactory definition of the predictive endpoint. Therefore it was concluded that the finding of satisfactory correlation for individual assays based on endpoints is difficult. It was hence decided to focus the statistical analysis on repeatability and reproducibility instead of correlation.

In the case of the MLD assay performed for test toxins in vitro, improvement in all toxins was obtained when relative values were applied. It was concluded that, for in vitro MLD assays, best reproducibility and discrimination between toxins were observed when results were corrected for sensitivity. After that, for each laboratory the “best” overall predictive endpoint was defined (i.e. which endpoint on Vero cells was the best predictor of lethality in mice for all toxins simultaneously). This endpoint was then used to predict the in vivo response and to calculate the corresponding MLD. Even if in individual cases not all of the predicted endpoints corresponded to observed endpoints in mice, the predictive power of endpoints on Vero cells, obtained per laboratory and especially for the overall mean, seemed to be good.

After that, relative MLD values obtained for test toxins in vitro were compared to corresponding relative MLD values obtained in vivo. Firstly, similar reproducibility and repeatability were observed. When data for each toxin and for each laboratory were presented graphically, the same overall pattern could be observed in vivo and in vitro but only when whole groups of results were considered. Ranking the test toxins (average per lab) was then done and resulted in some improvement in separation between toxins. Whereas the concordance between relative MLD values in vitro and relative MLD values in vivo for individual assays per laboratory was not conclusive, the same concordance performed for grand means was good (ρc=0.961).

TCP

As for the MLD assay the intention for the TCP assay was to determine the TCP of toxoids in mice and on Vero cells, to find a correlation and define predictive endpoints. But unlike the MLD assay, for TCP assay it turned out to be impossible to link endpoints on Vero cells to endpoints in mice. There are two possible explanations:

i) Strong toxoids are more diluted so the steps between doses are smaller, leaving smaller differences between the final levels of antigenicity. As the detecting toxin at L+ level is highly toxic on Vero cells, the small quantities of antitoxin added leave too much unbound toxin. This results in a less than 2-fold difference between highest and lowest dose and gives practically the same endpoint for all doses (or at most 1 well difference).


37

____________________________________________________________________

ii) Weak toxoids bind almost no antiserum, thus leaving high amounts of antiserum. The detecting toxin might be completely neutralised or remain in only low quantities. A sudden jump will occur when the toxin is neutralised.

It was concluded that for TCP assays the L+ level in mice is not suitable for use on Vero cells because of the high sensitivity. If we want to describe the biological principle of the test in a parametric model we need to include at least 4 unknown parameters:

Expected response =

f(a×ln((0.05×Max(0;1/L-Max(0;2-0.5×B/Di)/N)/P/2j-1)/S))

In this equation:

Known by design are: L = dilution of CSTx (e.g. 170); Di = dilution of the toxoid on row i (e.g. 60) ;P = Pre-dilution of mix before plating (e.g. 16) and j = column. While unknown parameters are: B = Binding power of the toxoid (IU/mL); N = Toxin Equivalence of CSTx (IU/mL); S = Sensitivity of Vero cells (mL CSTx/well) and a = Slope of the tolerance curve.

Therefore, there are too many unknown parameters at play in the Vero cell assays and, in consequence, it is very difficult to determine them from individual plates.

This problem could have the following possible solutions:

• It is possible to use data from this study to “calibrate” the toxin equivalence of CSTx (N) and plug that value back into the equation as a known parameter where toxin equivalence is the amount of antiserum (in IU) needed to neutralise 1mL toxin.

• Once N is known, the sensitivity (S) of the Vero cells can be determined from the toxin-antitoxin tests (in nL/well or µIU/well) assuming that sensitivity is lab-specific and that it is the same in TCP tests as in the toxin-antitoxin tests.

• The exact slope of the tolerance curve (a) does not need to be known with high precision, so it was fixed at an arbitrary but reasonable value of 4.25 (see report for justification). In this case data from the toxin-antitoxin (VI) tests could be used to obtain an estimate of the sensitivity of Vero cells and the toxicity of CSTx in IU/mL (= toxin equivalence) as the VI test was performed in parallel to the TCP. By assigning the toxin equivalence 284 IU/mL to the CSTx, the estimates of sensitivity can be improved and expressed in µIU/well. For some laboratories the calculated sensitivity in the toxin-antitoxin VI test was quite different from that in the MLD test but never more than a factor 3.3. According to what is mentioned above, if the toxin equivalence test had been carried out in a similar way with all other toxins we might have found a much better reproducibility than obtained with the MLD method.


38

____________________________________________________________________

• Further calculations of the Binding power (B) of the toxoids depend on the following (unverified) assumptions:

Firstly, the toxin equivalence (N) was the same in all laboratories and stable during the period in which the tests were performed. (There are indications that the CSTx lost some toxicity during the period of testing but it is not possible to control or correct for this drift with the chosen assay design. It is therefore assumed that N=284 IU/mL for all tests in all laboratories).

Secondly, the sensitivity of the Vero cells (S) in the TCP assays was the same as in the VI test performed in parallel.

With the chosen assay design there is no way to check if these assumptions are correct but the calculation of the binding power seems fairly robust against small deviations from this assumption.

• In summary, with estimates of N and S obtained from the VI test we have reduced the TCP equation to only 1 unknown parameter (B):

Expected response =


Where known parameters are: L = dilution of CSTx (e.g. 170); Di = dilution of the toxoid on row i (e.g. 40, 50, 60, 70, 80); P = Pre-dilution of mix before plating (e.g. 16); N = Toxin Equivalence of CSTx (284 IU/mL); S = Sensitivity of Vero cells in a given Lab (e.g. 115 µL CSTx/well); j = column and a = Slope of tolerance curve (arbitrarily fixed at 4.25).

The unknown parameter is B = Binding power of the toxoid (IU/mL) and it is easier to determine it from individual plates.

For TCP assays performed in mice and on Vero cells, improvement in some, but not in all, results for test toxoids was obtained when TCP values were established in IU/mL. In the same manner, when TCP results for in vivo and in vitro assays were presented graphically, their patterns were similar if looking at the overall level, but at the individual level, inversions sometimes occurred. However, when the test toxoids were ranked (average per lab) there was some improvement in separation between toxoids. Whereas the concordance between in vitro and in vivo TCP values expressed in IU/mL for individual assays per laboratory was not conclusive, the same concordance performed for grand means was good (ρc=0.980).


39

____________________________________________________________________

In conclusion:

• Good concordance between in vivo and in vitro methods was shown for both MLD and TCP but only for the grand means. For individual assays the reproducibility (in vivo and in vitro) is not sufficient to show satisfactory concordance.

• Statistical analysis was difficult because of too many uncontrolled parameters.

• A better plate design is needed to allow for reliable estimates of the toxin equivalence (instead of the MLD) and the binding power (instead of the TCP), from individual assays without having to make unverifiable assumptions.

For the proposed follow-up study design, it was proposed to use an “assay simulator” which could be a helpful tool to try different assay designs and to simulate the expected read-outs. In brief, known-by-design parameters (volumes, dilutions of tested toxin and/or antitoxin, etc.) as well as observed parameters such as true N (toxin equivalence, IU/mL), true S (sensitivity of Vero cells, µIU/well) and true B (binding power, IU/mL) might be used to calculate possible outputs on plates. Once input of known parameters (toxin/toxoid volumes, dilutions, pre-dilution step, etc.) and true parameters (S and N) was done and calculation was accomplished, the read-outs expected were obtained. In comparison with true results, simulations of recovery seemed to be quite good. Based on assay simulator, several proposals for improved plate design were made. Computer simulations showed that a much better reproducibility might be achieved with some simple adaptations in study protocol. However there was yet no agreement on the design for the follow-up study (Phase III of BSP130) and it was decided that the project leaders and the statistician should continue to explore the possible design and that a laboratory trial should be performed prior to the start of phase III.


40

____________________________________________________________________

05/01/2016

1

BSP130Cell line assay for in-process toxicity and

Ag testing of Cl. septicum vaccine Ag

Summary of results

Arnold Daas, 15 September 2015

Arnold Daas ©2015 EDQM, Council of Europe. All rights reserved.

MLD AssayThe intention

• Determine the MLD of toxins in mice and on vero cells

• Find a correlation and define predictive endpoints


2-fold steps

3-foldsteps

Test

Control (Ref.)


41

____________________________________________________________________

05/01/2016

2

The intended approach turned out to be difficult to apply because:

• the endpoint (column number) could shift if laboratories changed pre-dilutions between assays;

• Not all laboratories had useful endpoints on the row corresponding with the mouse MLD;

• It was not clear how the results could be compared with results from labs doing only the in vitro part.




• In this example it is not clear if the endpoint on row C is out of range or not.


2-fold steps

3-foldsteps

Test

Control (Ref.)


42

____________________________________________________________________

05/01/2016

3

It was decided to express the endpoint in toxin volume units needed to kill the vero-cells because:

• The volume of pure toxin per well is known by design;

• The endpoints on each row should be consistent so can be combined in one value per plate;

• Powerful statistical optimization techniques are available if responses are expressed as a function of known and unknown parameters;


MLD AssayAlternative way to express the MLD

The dose-response function can be modelled to a minimum of 2 unknown parameters:

• The sensitivity (S) of the vero cells (LD50 of CSTx);

• The toxicity (N) of the test toxin.

However, S and N are confounded parameters because higher sensitivity cannot be distinguished from higher toxicity.

The number of unknown parameters can be reduced to only 1 if:

• Toxicity is expressed relative to a reference toxin, or

• Sensitivity is known from validation studies




43

____________________________________________________________________

05/01/2016

4

• Each plate included a row with the CSTx so in theory it would have been possible to express toxicity of the test toxins relative to the CSTx per plate.

• However, the CSTx was intended as control sample and not as reference sample.

• It was therefore decided to establish the sensitivity of the cell lines per laboratory based on the sensitivity tests carried out with CSTx (6 rows per plate).

• Use the thus established sensitivity (fixed S per lab) to quantify toxicity (N) of the test samples.



MLD AssayEstablishing the sensitivity of the vero cells

The sensitivity (LD50) in this example is estimated to be 0.76 nL CSTx/well, so MLD=0.1mL/(LD50 × 2½)=1:93 000 (See R script for calculation in annex to report)

Arnold Daas ©2015 EDQM, Council of Europe. All rights reserved.2-fold steps

3-foldsteps CSTx

(nL/well)


44

____________________________________________________________________

05/01/2016

5

MLD AssaySensitivity test (Reference toxin CSTx)

Lab code in vivo in vitro Ratio1 1:1215 - -2 1:1000 1:357 000 14503 1:3000 1:562 000 7604 1: 810 1:175 000 8805 1: 243 1:174 000 29306 1: 468 1: 93 000 8207 - 1:361 000 -8 - 1:300 000 -9 - 1: 42 000 -

10 - 1:529 000 -11 - 1: 23 000 -


Factor 12

Factor 24

MLD AssayResidual toxicity test (“average” plates per Lab)


Lab 2 (1: 357 000) Lab 3 (1: 562 000) Lab 4 (1: 175 000) Lab 5 (1: 174 000) Lab 6 (1: 93 000)

Lab 7 (1: 361 000) Lab 8 (1: 300 000) Lab 9 (1: 42 000) Lab 10 (1: 529 000) Lab 11 (1: 23 000)


45

____________________________________________________________________

05/01/2016

6


MLD AssayThe test toxins (in vivo)

Absolute values (without correction for sensitivity)TxA TxB TxC TxD TxE TxF

GM (N=6) 1:73 1:74 1:5 1:12 1:128 1:110

Inter-Lab GCV 113% 117% 91% 81% 69% 75%

Intra-Lab GCV 25% 47% 33% 33% 25% 28%

Factor Max / Min 14 10 9 5 6 7

TxA TxB TxC TxD TxE TxF

GM (N=6) 0.088 0.089 0.006 0.014 0.153 0.132

Inter-Lab GCV 49% 91% 65% 82% 72% 84%

Intra-Lab GCV 25% 47% 33% 33% 25% 28%


Relative values (with respect to CSTx)

Improvement in some cases but not in all

Absolute MLD Relative MLD


MLD AssayThe test toxins (in vivo)

The dispersion is so large that inversions between laboratories occur, even when corrected for sensitivity


46

____________________________________________________________________

05/01/2016

7

• Differences of up to a factor 14 are observed between laboratories testing the same toxin in mice. When expressed relative to the CSTx it is still up to a factor 7.

• The interaction toxin × sensitivity is too large to allow for a satisfactory definition of the predictive endpoint.

• Finding a satisfactory correlation for individual assays based on endpoints is therefore difficult.

• Focus the statistical analysis on repeatability and reproducibility instead of correlation.


MLD AssayThe problem


MLD AssayThe test toxins (in vitro)

Absolute values (without correction for sensitivity)TxA TxB TxC TxD TxE TxF

GM (N=10) 1:15 901 1:13 292 1:680 1:2694 1:25 201 1:25 617

Inter-Lab GCV 145% 176% 183% 173% 143% 174%

Intra-Lab GCV 29% 25% 50% 28% 24% 26%

Factor Max / Min 27 29 38 45 27 40


GM (N=10) 0.073 0.061 0.003 0.012 0.116 0.118

Inter-Lab GCV 55% 77% 60% 59% 43% 50%

Intra-Lab GCV 29% 25% 50% 28% 24% 26%


Relative values (with respect to CSTx)

Improvement in all cases


47

____________________________________________________________________

05/01/2016

8

Absolute MLD Relative MLD


MLD AssayThe test toxins (in vitro)

Better reproducibility and separation between toxoids when corrected for sensitivity


MLD AssayThe predictive power of endpoints on vero cells

Lab TxA TxB TxC TxD TxE TxF

2 1:104 1:150 1:7 1:25 1:150 1:125

3 1:286 1:262 1:15 1:28 1:315 1:262

4 1:36 1:25 1:3 1:5 1:52 1:36

5 1:21 1:31 1:2 1:6 1:78 1:78

6 1:84 1:106 1:8 1:15 1:153 1:127

Lab TxA TxB TxC TxD TxE TxF

2 1:72 1:150 1:3 1:17 1:125 1:104

3 1:238 1:262 1:22 1:48 1:454 1:454

4 1:12 1:8 1:1 1:4 1:36 1:36

5 1:43 1:15 1:1 1:16 1:65 1:135

6 1:121 1:61 1:5 1:45 1:221 1:265

Observed MLD in mice Predicted MLD in mice

For each laboratory the “best” overall predictive endpoint was defined (i.e. which endpoint on vero cells is the best predictor of lethality in mice for all toxins simultaneously). Use that to predict the in vivo response and calculate the MLD.


48

____________________________________________________________________

05/01/2016

9

1

2

4

8

16

32

64

128

256

512

1 2 4 8 16 32 64 128 256 512

Pred

icte

d M

LD in

mic

eObserved MLD in mice



MLD AssayThe predictive power of endpoints on vero cells

1

2

4

8

16

32

64

128

256

512

1 2 4 8 16 32 64 128 256 512

Pred

icte

d M

LD in

mic

e

Observed MLD in mice


Per laboratory Overall mean (N=5)


MLD AssayThe test toxins (in vitro versus in vivo)

Relative values in vivo


GM (N=10) 0.073 0.061 0.003 0.012 0.116 0.118

Inter-Lab GCV 55% 77% 60% 59% 43% 50%

Intra-Lab GCV 29% 25% 50% 28% 24% 26%


Relative values in vitro


GM (N=6) 0.088 0.089 0.006 0.014 0.153 0.132

Inter-Lab GCV 49% 91% 65% 82% 72% 84%

Intra-Lab GCV 25% 47% 33% 33% 25% 28%


Reproducibility and repeatability are similar


49

____________________________________________________________________

05/01/2016

10

Relative MLD in vivo Relative MLD in vitro


MLD AssayThe test toxins

The same overall pattern can be observed in vivo and in vitro but only when whole groups of results are considered.


MLD AssayRanking the test toxins (Average per lab)


50

____________________________________________________________________

05/01/2016

11


MLD AssayRanking the test toxins (Average per lab)


0,001

0,010

0,100

1,000

0,001 0,010 0,100 1,000

In v

itro

(MLD

ratio

) (N

=5)

In vivo (MLD ratio) (N=5)


MLD AssayIn vivo vs In vitro (rel. to CSTx)

Per laboratory Grand mean (N=5 vs N=10)

Concordance: ρc=0.961

TxATxB

TxC

TxD

TxETxF

0.001

0.010

0.100

1.000

0.001 0.010 0.100 1.000

In vi

tro (M

LD ra

tio to

CSTx

)

In vivo (MLD ratio to CSTx)

Figure 13: Concordance plot of the average MLD (in vitro versus in vivo)


51

____________________________________________________________________

05/01/2016

12

TCP AssayThe intention

• Determine the TCP of toxoids in mice and on vero cells

• Find a correlation and define predictive endpoints


2-fold steps

Steps of 20 TCP

units

Test

No Control!

TCP AssayThe reality

• It turned out to be impossible to link endpoints on verocells to endpoints in mice


Example 1 Example 2

? ?


52

____________________________________________________________________

05/01/2016

13


• Strong toxoids are more diluted so the steps between doses are smaller, leaving smaller differences between the final levels of toxicity.

• The detecting toxin at L+ level is highly toxic on vero cells, so the small quantities of antitoxin leave too much unbound toxin.

TCP AssayThe problem

Less than 2-fold difference between highest and lowest dose, gives for all doses practically the same endpoint (or at most 1 well difference)(B=200 IU/mL, L+=1/170 mL)

+ → + → 100%

+ → + → 85%

+ → + → 73%

+ → + → 63%

+ → + → 55%

Tx

Td At At Tx Tx

Td At At Tx

Td At At Tx Tx

Td At At Tx Tx

Td At At Tx Tx


• Weak toxoids bind less antiserum, thus leaving higher amounts of antiserum.

• The detecting toxin might be completely neutralized or remain in only low quantities.


Difference between highest and lowest dose is only 3-fold giving only 1 or 2 wells difference(B=60 IU/mL, L+=1/143 mL)

+ → + → 100%

+ → + → 100%

+ → + → 60%

+ → + → 42%

+ → + → 33%Td At At Tx Tx

Td At At Tx Tx

Td At At Tx Tx

Td At 0 Tx Tx

Td At 0 Tx Tx


53

____________________________________________________________________

05/01/2016

14


• Weak toxoids bind almost no antiserum, thus leaving high amounts of antiserum.

• The detecting toxin might be completely neutralized or remain in only low quantities.


A sudden jump will occur when all toxin is neutralized(B=62 IU/mL, L+=1/260mL)

+ → + → 100%

+ → + → 100%

+ → + → 30%

+ → + → 0%

+ → + → 0%Td At At Tx 0

Td At At Tx 0

Td At At Tx Tx

Td At 0 Tx Tx

Td At 0 Tx Tx

• The L+ level in mice is not suitable for use on verocells because of the high sensitivity

• There are too many unknown parameters at play in the vero cell assays:

• Unknown sensitivity of the vero cells

• Unknown toxicity of the detecting toxin

• Finding a satisfactory correlation for individual assays based on endpoints is therefore hopeless.

• So… what can we do with the data???




54

____________________________________________________________________

05/01/2016

15

• If we want to describe the biological principle of the test in a parametric model we need to include at least 4 unknown parameters:

Expected response =


In this equation the green parameters are known by design and the red parameters are unknown



L = dilution of CSTx (e.g. 170)Di = dilution of the toxoid on row i (e.g. 60)P = Pre-dilution of mix before plating (e.g. 16)

B = Binding power of the toxoid (IU/mL)N = Toxin equivalence of CSTx (IU/mL)S = Sensitivity of vero cells (mL CSTx/well)a = Slope of the tolerance curve



Finding 4 unknown parameters from individual plates is a perilous operation.

It is like walking to the top of a mountain via a narrow ridge.


55

____________________________________________________________________

05/01/2016

16


TCP AssayA possible solution

Use data from this study to “calibrate” the toxin equivalence of CSTx (N) and plug that value back into the equation as a known parameter. Toxin equivalence is the amount of antiserum (in IU) needed to neutralize 1mL toxin.

Once N is known, the sensitivity (S) of the verocells can be determined from the toxin-antitoxin tests (in nL/well or µIU/well). Assume that sensitivity is lab-specific and that it is the same in TCP tests as in the toxin-antitoxin tests.

The exact slope of the tolerance curve does not need to be known with high precision, so it was fixed at an arbitrary but reasonable value of 4.25 (see report for justification).

Data from the toxin-antitoxin (VI) tests could be used to obtain an estimate of the sensitivity of the vero cells and the toxicity of CSTx in IU/mL (= toxin equivalence).


TCP AssayA possible solution

Antitoxin

1.50 IU1.25 IU1.00 IU0.75 IU0.50 IU

+ CSTx

+ 1/170 mL+ 1/170 mL+ 1/170 mL+ 1/170 mL+ 1/170 mL

The difference between endpoints is characteristic for the toxin equivalence.


56

____________________________________________________________________

05/01/2016

17

Lab S (nL/well) N (IU/mL)2 0.348 295

3a 0.154 291

3b 0.384 307

4 0.520 282

5 0.174 274

6 0.258 258

7 0.492 278

8 0.340 313

9 0.750 279

10 0.497 296

11 1.506 246

Mean - 284

GCV - 7%

Factor Max / Min

- 1.3


TCP AssayToxin Equivalence of CSTx in the VI test

The toxin equivalence method appears to be much more reproducible than the MLD method!

Toxin GCV Max/Min

TxA 55% 5

TxB 77% 7

TxC 60% 6

TxD 59% 7

TxE 43% 4

TxF 50% 5

CSTx (N) 7% 1.3

MLD


If the toxin equivalence test had been carried out in a similar way with all other toxins we might have found a much better reproducibility than obtained with the MLD method

TCP AssayVariability of Toxin Equivalence compared to MLD


57

____________________________________________________________________

05/01/2016

18

Lab S (µIU/well) N (IU/mL)

2 89 284

3a 41 284

3b 91 284

4 151 284

5 51 284

6 86 284

7 149 284

8 76 284

9 226 284

10 127 284

11 682 284


TCP AssaySensitivity of the vero cells

By assigning 284 IU/mL to the CSTx, the estimates of sensitivity can be improved and expressed in µIU/well

Lab S (µIU/well) Difference

2 56 1.6

3 36 1.12.5

4 115 1.3

5 115 2.3

6 215 2.5

7 56 2.7

8 67 1.1

9 481 2.1

10 38 3.3

11 866 1.3

In VI test In MLD test

For some laboratories the calculated sensitivity in the VI test is quite different from that in the MLD test but never more than a factor 3.3

Further calculations of the Binding power (B) of the toxoids depend on the following (unverified) assumptions:

1. The toxin equivalence (N) is the same in all laboratories and stable during the period in which the tests were performed

• There are indications that the CSTx lost some toxicity during the period of testing but it is not possible to control or correct for this drift with the chosen assay design. It is therefore assumed that N=284 IU/mL for all tests in all laboratories.

2. The sensitivity of the vero cells (S) in the TCP assays is the same as in the VI test performed in parallel

• With the chosen assay design there is no way to check if this assumption is correct but the calculation of the binding power seems fairly robust against small deviations from this assumption.


TCP AssayBinding power of the toxoids


58

____________________________________________________________________

05/01/2016

19

• With the estimates of N and S obtained from the VI test we have reduced the TCP equation to only 1 unknown parameter:

Expected response =



TCP AssayBinding power of the toxoids

L = dilution of CSTx (e.g. 170)Di = dilution of the toxoid on row i (e.g. 40, 50, 60, 70, 80)P = Pre-dilution of mix before plating (e.g. 16)N = Toxin equivalence of CSTx (284 IU/mL)S = Sensitivity of vero cells in a given Lab (e.g. 115 µL CSTx/well)a = Slope of tolerance curve (arbitrarily fixed at 4.25)B = Binding power of the toxoid (IU/mL)



Finding only 1 unknown parameter is much easier.

It is like going to the top of the mountain in a ski-lift.


59

____________________________________________________________________

05/01/2016

20


TCP AssayThe test toxoids

In mice (IU/mL)TdG TdH TdJ TdK TdL TdM

GM (N=5) 142 48 21 178 69 71

Inter-Lab GCV 33% 42% 51% 31% 73% 110%

Intra-Lab GCV 15% 20% 0% 10% 21% 31%

Factor Max / Min 2.1 2.9 4.2 1.9 5.1 8.8

TdG TdH TdJ TdK TdL TdM

GM (N=10) 142 52 30 157 76 63

Inter-Lab GCV 41% 40% 42% 25% 30% 46%

Intra-Lab GCV 24% 20% 10% 15% 17% 28%

Factor Max / Min 4.1 3.7 4.8 2.7 2.9 4.2

On vero cells (IU/mL)

Improvement in some cases but not in all

in vivo in vitro


TCP AssayThe test toxoids

The patterns are similar when looking at the overall level, but at the individual level, inversions occur.


60

____________________________________________________________________

05/01/2016

21


TCP AssayRanking the test toxoids (Average per lab)


TCP AssayRanking the test toxoids (Average per lab)


61

____________________________________________________________________

05/01/2016

22


0

50

100

150

200

250

0 50 100 150 200 250

In v

itro

(IU/m

L) (N

=5)

In vivo (IU/mL) (N=5)

TdG TdH TdJ TdK TdL TdM

TCP AssayIn vivo vs In vitro

Per laboratory Grand mean (N=5 vs N=10)

TdG

TdH

TdJ

TdK

TdLTdM

0

50

100

150

200

0 50 100 150 200In

vitro

(IU/

mL)

In vivo (IU/mL)

Figure 14: Concordance plot of the average TCP (in vitro versus in vivo)

Concordance: ρc=0.980

Assay simulator


Ucon Uvol Bdil Bvol Ldil Lvol Wpre Wvol WstpIU/mL mL 1/… mL 1/… mL 1/… mL 1/… A T0 T1

A B 0 IU/mLB 1.5 1 1 0 170 1 16 0.1 2 1.5 0 0 N 284 IU/mLC 1.25 1 1 0 170 1 16 0.1 2 1.25 0 0 S 142 µIU/wellD 1 1 1 0 170 1 16 0.1 2 1 0 0E 0.75 1 1 0 170 1 16 0.1 2 0.75 0 0F 0.5 1 1 0 170 1 16 0.1 2 0.5 0 0GH

µIU/well1 2 3 4 5 6 7 8 9 10 11 12

A 0 0 0 0 0 0 0 0 0 0B 533 267 133 66.6 33.3 16.7 8.33 4.16 2.08 1.04C 1314 657 329 164 82.1 41.1 20.5 10.3 5.13 2.57D 2096 1048 524 262 131 65.5 32.7 16.4 8.19 4.09E 2877 1438 719 360 180 89.9 45 22.5 11.2 5.62F 3658 1829 915 457 229 114 57.2 28.6 14.3 7.14G 0 0 0 0 0 0 0 0 0 0H 0 0 0 0 0 0 0 0 0 0

This can be a helpful tool to try different assay designs and simulate the expected read-outs.

The yellow grid defines the parameters known by design (volumes, dilutions, etc.)

The purple grid defines the true (unknown) parameters.

Check in R how parameters are reproduced:Dil<-c(1.50,1.25,1.00,0.75,0.50)Obs<-c(2,4,4,5,5)ThisAssay<-VIassay(Dil,170,16,Obs)fOptim(ThisAssay,’NS’)

Result: N=277.5 IU/mLS=0.500 nL/well (=138.6 µIU/well)


62

____________________________________________________________________

05/01/2016

23

Conclusions and recommendations

• Good concordance between in vivo and in vitro method was shown for MLD and TCP assay…

• … but only at the level of the grand mean. For individual assays the reproducibility (in vivo and in vitro) is not sufficient to show satisfactory concordance.

• Statistical analysis was difficult because of too many uncontrolled parameters. A better plate design is needed to allow for reliable estimates of the toxin equivalence (instead of the MLD) and the binding power (instead of the TCP), from individual assays without having to make unverifiable assumptions.

• Several proposals for improved plate design are included in the study report. Computer simulations show that a much better reproducibility might be achieved with some simple adaptations.


These examples are just intended to serve as inspiration to explore possible assay designs.

They are not necessarily presented as the proposed design for the follow up study.


Example designs for discussion on Day 2


63

____________________________________________________________________

05/01/2016

24

Example design toxin equivalence



A B 0 IU/mLB 0 1 1 0 5 1 2 0.1 2 0 0.2 0.005 N 284 IU/mLC 0 1 1 0 25 1 2 0.1 2 0 0.04 0.001 S 142 µIU/wellD 0 1 1 0 125 1 2 0.1 2 0 0.008 0.0002E 0 1 1 0 625 1 2 0.1 2 0 0.0016 4E-05F 0 1 1 0 3125 1 2 0.1 2 0 0.0003 8E-06G 0 1 1 0 15625 1 2 0.1 2 0 6E-05 2E-06H

µIU/well1 2 3 4 5 6 7 8 9 10 11 12

A 0 0 0 0 0 0 0 0 0 0B 1E+06 7E+05 4E+05 2E+05 88750 44375 22188 11094 5547 2773.4C 3E+05 1E+05 71000 35500 17750 8875 4438 2219 1109 554.69D 56800 28400 14200 7100 3550 1775 887.5 443.8 221.9 110.94E 11360 5680 2840 1420 710 355 177.5 88.75 44.38 22.188F 2272 1136 568 284 142 71 35.5 17.75 8.875 4.4375G 454.4 227.2 113.6 56.8 28.4 14.2 7.1 3.55 1.775 0.8875H 0 0 0 0 0 0 0 0 0 0

From Annex: This design uses a 5-fold progression of the toxin dilutions:1/5; 1/25; 1/125; 1/625; 1/3125; 1/15625

Without knowledge about the sensitivity S of the vero-cells it is impossible to expressed toxicity in IU/mL.

At best one can express toxicity as MLD similar to the mouse assay but due to the high sensitivity of the verocells it is not clear how this relates to the MLD in mice.

But…..




A B 0 IU/mLB 0.1 1 1 0 5 1 2 0.1 2 0.1 0.1996 0.005 N 284 IU/mLC 0.1 1 1 0 25 1 2 0.1 2 0.1 0.0396 0.001 S 142 µIU/wellD 0.1 1 1 0 125 1 2 0.1 2 0.1 0.0076 0.0002E 0.1 1 1 0 625 1 2 0.1 2 0.1 0.0012 3E-05F 0.1 1 1 0 3125 1 2 0.1 2 0.1 0 0G 0.1 1 1 0 15625 1 2 0.1 2 0.1 0 0H

µIU/well1 2 3 4 5 6 7 8 9 10 11 12

A 0 0 0 0 0 0 0 0 0 0B 1417500 708750 354375 177188 88594 44297 22148 11074 5537 2768.6C 281500 140750 70375 35188 17594 8797 4398 2199 1100 549.8D 54300 27150 13575 6787.5 3394 1697 848.4 424.2 212.1 106.05E 8860 4430 2215 1107.5 553.8 276.9 138.4 69.22 34.61 17.305F 0 0 0 0 0 0 0 0 0 0G 0 0 0 0 0 0 0 0 0 0H 0 0 0 0 0 0 0 0 0 0

But….. If a 2nd series of tubes is prepared with 1mL antitoxin at a concentration of 0.1 IU/mL instead of 1mL buffer it is possible to obtain an estimate of the toxicity expressed in IU/mL (=toxin equivalence)

Simulations of recovery:50 IU/mL → 34 IU/mL (S= 95 µIU/well)100 IU/mL → 100 IU/mL (S=135 µIU/well)150 IU/mL → 132 IU/mL (S=124 µIU/well)200 IU/mL → 243 IU/mL (S=154 µIU/well)250 IU/mL → 167 IU/mL (S= 91 µIU/well)300 IU/mL → 316 IU/mL (S=144 µIU/well)350 IU/mL → 373 IU/mL (S=171 µIU/well)400 IU/mL → 403 IU/mL (S=128 µIU/well)


64

____________________________________________________________________

05/01/2016

25



The 5-fold range is suitable for preliminary ranging. The range can be narrowed by centering smaller steps around the preliminary value found.

Suppose preliminary value found was 85 IU/mL so we place 1/850 on row E. A possible dilution series would be

1/650; 1/650; 1/750; 1/850; 1/950; 1/1050

Row B is the same as Row C but without antitoxin. All other rows are done with 0.1 IU of antitoxin

1 2 3 4 5 6 7 8 9 10 11 12A 0 0 0 0 0 0 0 0 0 0B 6154 3077 1538 769.2 384.6 192.3 96.15 48.08 24.04 12.019C 1154 576.9 288.5 144.2 72.12 36.06 18.03 9.014 4.507 2.2536D 333.3 166.7 83.33 41.67 20.83 10.42 5.208 2.604 1.302 0.651E 0 0 0 0 0 0 0 0 0 0F 0 0 0 0 0 0 0 0 0 0G 0 0 0 0 0 0 0 0 0 0H 0 0 0 0 0 0 0 0 0 0

If the read-outs are as shown here, the ML method givesN=81 IU/mL and S=168 µIU/well.

This is close to the true values of N=80 IU/mL and S=142 µIU/well.



Another approach is to adapt the design from the study to smaller antitoxin levels and include one row without antitoxin.Antitoxin at 0.0 – 0.2 – 0.4 – 0.6 – 0.8 – 1.0 IU and toxin at 0.5 IU on all rows (e.g. if approximate value of toxin is 250 IU/mL then dilute 1/500). This design allows for quantification between 125 and 500 IU/mL.

Recovery:125 IU/mL → N=133 IU/mL (S=147 µIU/well)150 IU/mL → N=200 IU/mL (S=221 µIU/well)200 IU/mL → N=200 IU/mL (S=110 µIU/well)250 IU/mL → N=274 IU/mL (S=168 µIU/well)300 IU/mL → N=300 IU/mL (S=127 µIU/well)350 IU/mL → N=385 IU/mL (S=190 µIU/well)400 IU/mL → N=400 IU/mL (S=122 µIU/well)450 IU/mL → N=471 IU/mL (S=164 µIU/well)500 IU/mL → N=500 IU/mL (S=125 µIU/well)


A B 0 IU/mLB 0.000 1 1 0 500 1 1 0.1 2 0 0.002 0.0001 N 250 IU/mLC 0.200 1 1 0 500 1 1 0.1 2 0.2 0.0012 6E-05 S 150 µIU/wellD 0.400 1 1 0 500 1 1 0.1 2 0.4 0.0004 2E-05E 0.600 1 1 0 500 1 1 0.1 2 0.6 0 0F 0.800 1 1 0 500 1 1 0.1 2 0.8 0 0G 1.000 1 1 0 500 1 1 0.1 2 1 0 0H

µIU/well1 2 3 4 5 6 7 8 9 10 11 12

A 0 0 0 0 0 0 0 0 0 0B 25000 12500 6250 3125 1563 781.3 390.6 195.3 97.66 48.828C 15000 7500 3750 1875 937.5 468.8 234.4 117.2 58.59 29.297D 5000 2500 1250 625 312.5 156.3 78.13 39.06 19.53 9.7656E 0 0 0 0 0 0 0 0 0 0F 0 0 0 0 0 0 0 0 0 0G 0 0 0 0 0 0 0 0 0 0H 0 0 0 0 0 0 0 0 0 0


65

____________________________________________________________________

05/01/2016

26



Same example as previous slide with correct assumptions about sensitivity (S=150 µIU/mL)

Recovery:125 IU/mL → N=137 IU/mL150 IU/mL → N=183 IU/mL200 IU/mL → N=200 IU/mL250 IU/mL → N=271 IU/mL300 IU/mL → N=300 IU/mL350 IU/mL → N=369 IU/mL400 IU/mL → N=400 IU/mL450 IU/mL → N=467 IU/mL500 IU/mL → N=500 IU/mL



µIU/well1 2 3 4 5 6 7 8 9 10 11 12

A 0 0 0 0 0 0 0 0 0 0B 25000 12500 6250 3125 1563 781.3 390.6 195.3 97.66 48.828C 15000 7500 3750 1875 937.5 468.8 234.4 117.2 58.59 29.297D 5000 2500 1250 625 312.5 156.3 78.13 39.06 19.53 9.7656E 0 0 0 0 0 0 0 0 0 0F 0 0 0 0 0 0 0 0 0 0G 0 0 0 0 0 0 0 0 0 0H 0 0 0 0 0 0 0 0 0 0



Same example as previous slide but with incorrect assumptions about sensitivity (S= 75 µIU/mL and S= 300 µIU/mL when true sensitivity is S= 150µIU/mL)

Recovery: S= 75 IU/mL S= 300 IU/mL125 IU/mL → N=116 IU/mL N= 201 IU/mL150 IU/mL → N=129 IU/mL N= 201 IU/mL200 IU/mL → N=151 IU/mL N= 201 IU/mL250 IU/mL → N=229 IU/mL N= 301 IU/mL300 IU/mL → N=255 IU/mL N= 301 IU/mL350 IU/mL → N=328 IU/mL N= 401 IU/mL400 IU/mL → N=354 IU/mL N= 401 IU/mL450 IU/mL → N=427 IU/mL N= 501IU/mL500 IU/mL → N=453 IU/mL N= 501 IU/mL

So: Calculations without assumptions on sensitivity are not that bad and the estimate of S can be used as validity criterion.



µIU/well1 2 3 4 5 6 7 8 9 10 11 12

A 0 0 0 0 0 0 0 0 0 0B 25000 12500 6250 3125 1563 781.3 390.6 195.3 97.66 48.828C 15000 7500 3750 1875 937.5 468.8 234.4 117.2 58.59 29.297D 5000 2500 1250 625 312.5 156.3 78.13 39.06 19.53 9.7656E 0 0 0 0 0 0 0 0 0 0F 0 0 0 0 0 0 0 0 0 0G 0 0 0 0 0 0 0 0 0 0H 0 0 0 0 0 0 0 0 0 0


66

____________________________________________________________________

05/01/2016

27

Example Improved Design TCP assay



A B 180 IU/mLB 4 0.5 10 0.5 170 1 2 0.1 2 0 0.0059 0.0001 N 284 IU/mLC 4 0.5 20 0.5 170 1 2 0.1 2 0 0.0059 0.0001 S 142 µIU/wellD 4 0.5 40 0.5 170 1 2 0.1 2 0 0.0059 0.0001E 4 0.5 80 0.5 170 1 2 0.1 2 0.875 0.0028 7E-05F 4 0.5 160 0.5 170 1 2 0.1 2 1.4375 0.0008 2E-05GH

µIU/well1 2 3 4 5 6 7 8 9 10 11 12

A 0 0 0 0 0 0 0 0 0 0B 41765 20882 10441 5221 2610 1305 652.6 326.3 163.1 81.572C 41765 20882 10441 5221 2610 1305 652.6 326.3 163.1 81.572D 41765 20882 10441 5221 2610 1305 652.6 326.3 163.1 81.572E 19890 9945 4972 2486 1243 621.6 310.8 155.4 77.69 38.847F 5827 2914 1457 728.4 364.2 182.1 91.05 45.53 22.76 11.381G 0 0 0 0 0 0 0 0 0 0H 0 0 0 0 0 0 0 0 0 0

From Annex: This design uses a geometric progression of the toxoid dilutions for preliminary ranging:1/10; 1/20; 1/40; 1/80; 1/160

Simulations show a fairly good recovery of the true binding power B (without assumptions about N and S):

15 IU/mL → 47 IU/mL45 IU/mL → 48 IU/mL75 IU/mL → 68 IU/mL105 IU/mL → 95 IU/mL135 IU/mL → 93 IU/mL165 IU/mL → 128 IU/mL195 IU/mL → 190 IU/mL

Example Improved Design TCP assay



A B 105 IU/mLB 4 0.5 10 0.5 170 1 2 0.1 2 0 0.0059 0.0001 N 284 IU/mLC 4 0.5 20 0.5 170 1 2 0.1 2 0 0.0059 0.0001 S 142 µIU/wellD 4 0.5 40 0.5 170 1 2 0.1 2 0.6875 0.0035 9E-05E 4 0.5 80 0.5 170 1 2 0.1 2 1.3438 0.0012 3E-05F 4 0.5 160 0.5 170 1 2 0.1 2 1.6719 0 0GH

µIU/well1 2 3 4 5 6 7 8 9 10 11 12

A 0 0 0 0 0 0 0 0 0 0B 41765 20882 10441 5221 2610 1305 652.6 326.3 163.1 81.572C 41765 20882 10441 5221 2610 1305 652.6 326.3 163.1 81.572D 24577 12289 6144 3072 1536 768 384 192 96 48.002E 8171 4085 2043 1021 510.7 255.3 127.7 63.84 31.92 15.959F 0 0 0 0 0 0 0 0 0 0G 0 0 0 0 0 0 0 0 0 0H 0 0 0 0 0 0 0 0 0 0

If the TCP assay is performed in parallel with the toxin/antitoxin test, the estimates can be further improved:

Plate 1: Toxoid + Antitoxin + Det. ToxinPlate 2: Antitoxin + Det. ToxinPlate 3: Det. Toxin

15 IU/mL → 16 IU/mL45 IU/mL → 48 IU/mL75 IU/mL → 73 IU/mL105 IU/mL → 95 IU/mL135 IU/mL → 141 IU/mL165 IU/mL → 169 IU/mL195 IU/mL → 189 IU/mL


67

____________________________________________________________________

A4. Participant’s experience in the performance of the experiments of BSP130

Participants were asked to comment on performance of the experiments of BSP130 in their respective laboratories. In particular, to summarise their own experience each participant was requested to give the total number of each type of test performed (MLD in vivo, MLD in vitro, TCP in vivo, TCP in vitro) as well as the number of successful valid assays (MLD in vivo, MLD in vitro, TCP in vivo, TCP in vitro) and, if known, probable reasons for any unsuccessful or invalid assays. Additionally any insufficiencies in materials provided as well as any problems (for example, loss of activity) with materials met during the study were to be reported. Any problems with the clarity of the protocol or data recording sheets and comments on any aspects of BSP130 were also to be reported. Finally, participants were requested to provide suggestions for improvements of the protocol to be used in the follow-up study.

Among the six participants present at the meeting, three laboratories performed only in vitro methods while three others used both in vitro and in vivo methods. In general, in all these laboratories the overall number of successful valid assays in comparison with the unsuccessful or invalid ones was reported to be very high: 58 from 67 for MLD in vivo (87%); 182 from 194 for MLD in vitro (94%); 54 from 67 for TCP in vivo (81%); 183 from 195 for TCP in vitro (94%).

One participant gave a general appreciation of the in vitro testing in Vero cells performed by all laboratories in accordance with the methodologies described in the BSP130 study protocol, commenting that in vitro assays were easy to perform. Two participants out of six reported no unsuccessful or invalid assays in both in vitro MLD and TCP tests. Four other participants who had 24 unsuccessful or invalid in vitro MLD and TCP assays (which represented 6% of performed assays) came up with the following reasons and comments:

Firstly, four participants with unsuccessful or invalid in vitro MLD and TCP assays remarked that it was mainly due to the high coefficient of variation (CV) of the negative control wells which was greater than 15% and that resulted in 14 invalid tests (58% of invalid assays) in all these laboratories. It was presumed that an edge effect observed in the plates may be the cause of this kind of failure and it was suggested that the negative control row, when possible, should be put in the middle of the plate. In this context, problems with cell staining were considered. Indeed, cell staining higher than that which was expected according to study protocol was reported by three out of six laboratories. To avoid potential impact of such an excessive cell staining on CV readings, it was suggested that an additional washing step after cell staining should be included in the protocol of the follow-up study.

In addition, two laboratories reported that non-confluence of cell culture led to significant increase of incubation time with toxins or mixtures of reference serum/detecting toxin/toxoids (more than 24 hours, instead of the 24 hours stipulated


68

____________________________________________________________________

in the study protocol). After brief discussion of possible reasons for cell culture problems encountered by these participants, it was assumed that besides cell origin, cell passage number can also influence cell growth and, probably, cells’ sensitivity to toxins. It was suggested that recording of the cell passage number for each assay in data reporting sheets should be improved so as to provide more information about cell culture conditions. Some other minor issues linked with cell culture were then addressed: inter alia, poor cell attachment in spite of confluent cell monolayer, which resulted in 1 invalid assay in one laboratory. In this case it was proposed to include, in the protocol of the follow-up study, the request for a training period and to provide images of typical cell layers.

Neither cell culture, nor methodologies of BSP130 study could be implicated for the 9 other unsuccessful or invalid in vitro MLD and TCP assays reported by two participants. Reasons for these failures were more likely linked with some toxin supply and storage issues. Indeed, one participant observed loss of CSTx detecting toxin activity in thawed aliquots stored at 4°C for more than the two week period which was stipulated by the BSP130 study protocol. Another participant who didn’t observe loss of CSTx detecting toxin activity (maybe owing to rapid use of this toxin), reported however that 3 in vitro TCP invalid assays were caused by non-conformance of CSTx detecting toxin results. Even if these observations do not fit with the known substantial stability of toxins (which was reported to be at least one year in the laboratory of one of the project leaders), with a view to improving the follow up study it was proposed to review and to refine storage conditions and storage period of toxins at 4°C. In particular, it was proposed to store CSTx detecting toxin at -20°C in smaller aliquots and to use it, once thawed, immediately or after no longer than 2 weeks.

Furthermore, three participants who performed both in vitro and in vivo MLD and TCP assays pointed out low titres of TxC toxin and, probably, of TxD toxin that resulted in both low mouse sensitivity and correspondingly in insufficient quantities of these toxins for some in vitro tests, as was mentioned by one of participants. In the same manner, all these participants noted also that the supply of Cl. septicum standard antitoxin (IV) may be insufficient for performance of the in vivo tests. Therefore additional vials had to be ordered to allow the completion of the study experiments by some of the laboratories. In addition, one of participants who had a sufficient amount of toxins and Cl. septicum standard antitoxin (IV) to perform necessary tests, noted that the number of CSTx detecting toxin and Cl. septicum standard antitoxin (IV) vials provided, would not have been sufficient to perform retesting in case of an invalid assay.

Among general issues addressed in relation to the study protocol, questions of pre-dilutions and dilutions especially for in vitro assays were mentioned by three participants. One of these participants reported that the description of dilutions was difficult to understand, as was anticipation if adjustment was needed from one assay to the next one (link between TCP units/end point units and dilutions to be carried


69

____________________________________________________________________

out). Another participant wondered if the pre-dilution in the in vitro test should be part of a two-fold series, while the last participant questioned which pre-dilutions/dilutions to use. Therefore, it was suggested that the follow-up study protocol should include additional examples or recommendations of serial dilutions to be performed. Finally, two participants suggested including comments about invalid assays in reporting sheets.

Concerning in vivo methods, apart from the remark expressed by one participant that benefits and drawbacks of including in-house methods should be considered, few comments were made by the three participants who performed in vivo MLD and TCP assays in mice, according their own methodologies, for these tests. In general, one participant reported absence of unsuccessful or invalid assays in both in vivo MLD and TCP tests, two other participants had 22 unsuccessful or invalid in vivo MLD and TCP tests (16% of performed in vivo assays). They proposed the following explanations and comments: in 6 invalid TCP in vivo tests and in 1 invalid MLD in vivo test, absence of endpoint was reported by both laboratories; in 6 other invalid TCP in vivo tests, non-conformance of CSTx detecting toxin results was reported by one of participants; in 6 invalid MLD in vivo tests an invalid Cl. septicum standard antitoxin (IV) control was observed by the same participant; finally, in 1 invalid TCP in vivo test and 2 invalid MLD in vivo tests, inconsistencies between results of two consecutive assays were reported by another participant.

To summarise the participants’ experience in performing cell line assays, the session chairs expressed overall satisfaction with the results obtained by participants, by noting that 6% of invalid in vitro Vero cell assays represents a very low level of invalidity for the completely new methodological approach which was proposed in the BSP130 collaborative study. The vast majority of failed in vitro assays were due to high coefficient of variation (CV) of the negative control wells. It was underlined that for the follow up study, such solutions as the relocation of negative control wells on column 11 of the plate, and the doubling of pre-dilutions as well as a possible additional washing step would be considered. Furthermore, it was remarked that the cause of the major part of other unsuccessful or invalid assays was short supply in Toxin C and Cl. septicum standard antitoxin (IV) and, as a result, unequal sharing of materials between in vitro and in vivo assays. Thus, in some laboratories Toxin C and Cl. septicum standard antitoxin (IV) were mainly used for in vivo tests and in vitro assays could not be completed. Insufficiencies in toxin and antitoxin should not occur in the follow up study which intends to request performance of only in vitro assays, which require only limited amounts of test toxins/toxoids.


70

____________________________________________________________________

A5. Round table discussion and conclusions Chairs: Dr M-E. Behr-Gross, Dr L. Bruckner, Dr K. Redhead

The session chairs introduced the discussion by noting that the final results of the BSP130 collaborative study, its statistical analysis and conclusions were included in the revised BSP130 study report which was distributed to participants before the meeting. It was added that it is planned to prepare publication on the basis of this report to allow other manufacturers access to study details. Then there was a reminder that there would be a follow up study with an optimised protocol only for the in vitro assays so as to take advantage of the sensitivity and accuracy of Vero cell methods. To achieve this purpose, proposals for a follow-up study protocol should be based not only on the study results but also on participants’ experience in the performance of the experiments in a standardised way. Moreover, it was announced that in the follow-up study it was planned to request participants to perform the statistical analysis of the data generated in their own laboratory. Participants were asked to comment on the conclusions and the statistical approach presented in the study report and then to express their concerns and suggestions toward the optimisation of the study protocol in particular.

A participant commented on maximum likelihood methods employed for statistical analysis. Because of the novelty of the Vero cells assays and their much greater sensitivity to use, it was decided, after completion of the experimental part of the study, to use an elaborated statistical analysis which could not have been designed before the beginning of the study due to several unknown parameters. For the optimisation of the follow-up study (Phase III), critical parameters should be identified and fixed. This would allow good accuracy and reproducibility of results for in vitro assays to be obtained, and allow the use of more classical statistical models for the analysis of individual laboratories’ results.

In this context it was also remarked that the plate layout applied for assay simulator did not take into account the edge effect observed during this study. When possible, negative control rows should be placed inside of the plate, for instance, one control row should be placed on column 11. However, this would involve loss of the second control row and would have impact on the CV of the negative control wells. It was assumed that the establishment of additional criteria for CV limits would be needed: inter alia, acceptance of 20 % for CV might be considered. There was a reminder that some comments on in vitro tests, such as high CV and short supply of antitoxin and some toxins, had already been mentioned. However, some others points important for the elaboration of new protocol emerged from the discussion and need to be clarified. In particular, the potential impact of cell passage number on cell sensitivity to toxins was mentioned. This, in part, could explain the 24-fold toxin sensitivity range observed between laboratories. It was remarked that to understand this observation it would be interesting to know the complete history of cell lines used by participants,


71

____________________________________________________________________

intervals between passaging and seeding as well as the age of these cells before seeding. But in the first instance, it was decided to focus on the cell passage number which appears to be crucial for cell sensitivity to toxins. Therefore, participants were requested to give more precise feedback information such as practical details on the source of cells and exact cell passage number for the cell line used in the study. This should allow standardising of the cell culture conditions in the protocol of the follow- up study and allow advice to be given the inexperienced participants in terms of cell passages and, especially, cell origin which also seems to have impact on cell sensitivity.

Some practical issues were also addressed. In particular, the number of plates needed to perform the study according to the proposed design. The speaker responded that for TCP assay it would be 3 plates (6 taking into account duplicates): one with Detecting toxin, one with Antitoxin + Detecting toxin and one with three components: Toxoid + Antitoxin + Detecting toxin. Whereas for Toxin/Antitoxin test, 2 plates would be needed (4 taking into account duplicates): one with Toxin and another with both Toxin and Antitoxin.

In addition, the current situation for the replacement of the conventional mouse tests for Cl. septicum antigens by Vero cell assays was discussed. It was reported that this kind of in vitro assay is already used in production lines of one manufacturer where in vitro testing not only results in significant savings in animal usage, but also allows more accurate in-process toxicity and antigenicity testing of Cl. septicum vaccine antigens. Moreover, it was reported that one manufacturer had developed a cell line assay for Cl. septicum type D, and that this type of assay had been accepted by one national regulatory authority.

Finally, one of the participants expressed the general feeling that additional time is required for reflection and comprehension of the elaborate BSP130 report. After brief debate it was proposed that participants would return with their final proposals to the project coordinators and a new deadline was fixed to allow participants to submit comments on the study report and suggestions for the follow-up study. All the participants were acknowledged for taking part in the discussion.


72

____________________________________________________________________

Session B: Follow up activity proposals

Speaker: Dr L. Bruckner

B1. A cell based assay as compendial test for assessing residual toxicity (MLD) and immunogenicity (TCP): consequences and issues

The speaker went through the European Pharmacopeia Monograph for Clostridium septicum vaccinei indicating the main points where animal tests are stipulated. In brief, according to Ph. Eur. monograph Clostridium septicum vaccine for veterinary use (0364), the vaccine must be satisfactory with respect to safety (5.2.6) and efficacy (5.2.7) for the animals for which it is intended. To meet these requirements, the Ph. Eur. monograph provides descriptions of the tests for safety, residual toxicity and batch potency. The safety test (section 2-2-1) has to be done on a batch of vaccine using the target species and requires the use of a certain number of animals. The test for residual toxicity (section 2-3-1 and also section 3-3) is performed in mice. This is one of the most severe tests in veterinary vaccines quality control which, in addition, uses a large number of animals. The batch potency test (section 2-3-2) involves vaccination of rabbits and subsequent quantification of antibodies, for which it also prescribes the replacement of the mouse (challenge) test by the use of suitable alternative methods such as an immunochemical method (2.7.1) or neutralisation in cell cultures.

The in vivo test for assessing residual toxicity (MLD test) is based on cytopathic effects of clostridial toxins where mice are used only as an indicator of toxicity. In this case mice could be replaced by a different indicator such as cell lines. However, sensitivity of these in vitro and in vivo tests is quite different and could depend on unknown factors such as which laboratory performs the test, technician, mouse strains and/or cell culture passage number. Therefore, it was noted that to establish product specifications for the alternative assay, both the current (in vivo) and alternative method(s) would need to be initially run in parallel. The alternative methodology would be used to establish suitable specifications for the new test with the lots defined as compliant with the in vivo method and the other release criteria. Importantly, since the TCP test is not mentioned in the Ph. Eur. Monograph, this in-process test is under the responsibility of the manufacturer. But if manufacturers wanted to apply the replacement of the TCP in vivo test, the same approach as for the MLD test should be applied, i.e. to run in vivo and in vitro TCP assays in parallel and to establish correlation. In conclusion, the speaker emphasised his personal conviction of the importance of finding alternatives to animal testing.


73

____________________________________________________________________

Lukas BrucknerBSP130 Participants Workshop, Egmond aan Zee, 16 September 2015

A CELL BASED ASSAY AS COMPENDIALTEST FOR ASSESSINGTOXICITY IN RESIDUAL TOXICITY (MLD) ANDIMMUNOGENICITY (TCP) TESTING:CONSEQUENCES AND ISSUES

Lukas Bruckner


74

____________________________________________________________________


THE PH.EUR. MONOGRAPHCLOSTRIDIUM SEPTICUM VACCINE1. DEFINITIONClostridium septicum vaccine for veterinary use is prepared from a liquid culture of a suitable strain of Clostridium septicum.

The whole culture or its filtrate or a mixture of the two is inactivated to eliminate its toxicity while maintaining adequate immunogenic properties. This monograph applies to vaccines intended for active immunisation of animals and/or to protect passively their progeny against disease caused by C. septicum.


75

____________________________________________________________________


THE PH.EUR. MONOGRAPHCLOSTRIDIUM SEPTICUM VACCINE2. PRODUCTION2-1. PREPARATION OF THE VACCINEC. septicum used for production is grown in an appropriate liquid medium. Toxoid and/or inactivated cultures may be treated with a suitable adjuvant.

2-2. CHOICE OF VACCINE COMPOSITIONThe vaccine is shown to be satisfactory with respect to safety (5.2.6) and efficacy (5.2.7) for the animals for which it is intended. For the latter, it shall be demonstrated that for each target species the vaccine, when administered according to the schedule to be recommended, stimulates an immune response (for example, induction of antibodies) consistent with the claims made for the product.

The following test for safety (section 2-2-1) may be used during the demonstration of safety.


THE PH.EUR. MONOGRAPHCLOSTRIDIUM SEPTICUM VACCINE2-2-1 SafetyCarry out the tests for each route and method of administration to be recommended for vaccination and where applicable, in animals of each category for which the vaccine is intended, using in each case animals not older than the minimum age to be recommended for vaccination. Use a batch of vaccine containing not less than the maximum potency that may be expected in a batch of vaccine. For each test, use not fewer than 8 animals that do not have antibodies against C. septicum. Administer to each animal 1 dose of the vaccine. If the schedule to be recommended requires a 2nd dose, administer another dose after an interval of at least 14 days. Observe the animals at least daily until 14 days after the last administration.

The vaccine complies with the test if no animal shows abnormal local or systemic reactions or dies from causes attributable to the vaccine. If the test is carried out in pregnant animals, no adverse effets on gestation or the offspring are noted.


76

____________________________________________________________________


THE PH.EUR. MONOGRAPHCLOSTRIDIUM SEPTICUM VACCINE2-3. MANUFACTURER’S TESTS2-3-1 Residual toxicityA test for detoxification is carried out immediately after the detoxification process and, when there is risk of reversion, a 2nd test is carried out at as late a stage as possible during the production process. The test for residual toxicity (section 3-3) may be omitted by the manufacturer.


THE PH.EUR. MONOGRAPHCLOSTRIDIUM SEPTICUM VACCINE2-3-2 Batch potency test It is not necessary to carry out the potency test (section 3-4) for each batch of vaccine if it has been carried out using a batch of vaccine with a minimum potency. Where the test is not carried out, an alternative validated method is used, the criteria for acceptance being set with reference to a batch of vaccine that has given satisfactory results in the test described under Potency and that has been shown to be satisfactory with respect to immunogenicity in the target species. The following test may be used after a satisfactory correlation with the test under Potency (section 3-4) has been established.

Vaccinate rabbits as described under Potency and prepare sera. Determine the level of antibodies against the toxin of C. septicum in the individual sera by a suitable method such as an immunochemical method (2.7.1) or neutralisation in cell cultures. Use a homologous reference serum calibrated in International Units of C. septicum antitoxin. Clostridia (multicomponent) rabbit antiserum BRP is suitable for use as a reference serum. The vaccine complies with the test if the level of antibodies is not less than that found for a batch of vaccine that has given satisfactory results in the test described under Potency and that has been shown to be satisfactory with respect to immunogenicity in the target species.


77

____________________________________________________________________


THE PH.EUR. MONOGRAPHCLOSTRIDIUM SEPTICUM VACCINE3. BATCH TESTS3-1 Identification When injected into animals that do not have C. septicum antitoxin, the vaccine stimulates the formation of such antitoxin.

3-2 Bacteria and fungi The vaccine, including where applicable the diluent supplied for reconstitution, complies with the test for sterility prescribed in the monograph Vaccines for veterinary use (0062).

3-3 Residual toxicityInject 0.5 mL of the vaccine by the subcutaneous route into each of 5 mice, each weighing 17-22 g. Observe the mice at least daily for 7 days. The vaccine complies with the test if no animal shows notable signs of disease or dies from causes attributable to the vaccine.


THE PH.EUR. MONOGRAPHCLOSTRIDIUM SEPTICUM VACCINE3-4 Potency

Use for the test not fewer than 10 healthy rabbits, 3-6 months old. Administer to each rabbit by the subcutaneous route a quantity of vaccine not greater than the minimum dose stated on the label as the 1st dose. After 21-28 days, administer to the same animals a quantity of the vaccine not greater than the minimum dose stated on the label as the 2nd dose. 10-14 days after the 2nd injection, bleed the rabbits and pool the sera.

The vaccine complies with the test if the potency of the pooled sera is not less than 2.5 IU/mL.

The International Unit is the specific neutralising activity for C. septicum toxin contained in a stated amount of the International Standard, which consists of a quantity of dried immune horse serum. The equivalence in International Units of the International Standard is stated by the World Health Organization.


78

____________________________________________________________________


THE PH.EUR. MONOGRAPHCLOSTRIDIUM SEPTICUM VACCINEThe potency of the pooled sera obtained from the rabbits is determined by comparing the quantity necessary to protect mice or other suitable animals against the toxic effects of a dose of C. septicum toxin with the quantity of a reference preparation of Clostridium septicum antitoxin, calibrated in International Units, necessary to give the same protection. For this comparison, a suitable preparation of C. septicum toxin for use as a test toxin is required. The dose of the test toxin is determined in relation to the reference preparation; the potency of the serum to be examined is determined in relation to the reference preparation using the test toxin.

3-4-1 Preparation of test toxin

3-4-2 Determination of test dose of toxin

3-4-3 Determination of the potency of the serum obtained from rabbits

4. LABELLING


3-3 RESIDUAL TOXICITY

Inject 0.5 mL of the vaccine by the s.c. route into each of 5 mice, each weighing 17-22 g. Observe the mice at least daily for 7 days. The vaccine complies with the test if no animal shows notable signs of disease or dies from causes attributable to the vaccine.


79

____________________________________________________________________


PRO’S AND CON’S TO REPLACE THEIN VIVO TEST BY THE IN VITRO TEST

PRO’S• Both, in vivo test and

in vitro test are functional tests based on the same principle: toxicity

CON’S• Different sensitivity of

in vivo test and in vitro test• Sensitivity of in vivo test

and in vitro test depends on factors not known

• lab? • technician?• mouse strain?• cell culture passage number?


A POSSIBLE WAY TO GO

To establish product specifications for the alternative assay, both the current (in vivo) and alternative method(s) would need to be initially run in parallel. The alternative methodology would be used to establish suitable specifications for the new test with the lots defined as compliant with in vivo method and the other release criteria.


80

____________________________________________________________________


3-3 RESIDUAL TOXICITY

This test is not performed for routine purposes. After establishment of a suitable correlation with an alternative test using cell cultures as indicator of toxicity, the alternative test is used. Suitable correlation may be established by testing not less than 10 batches simultaneously in the mouse test and the alternative test.

Inject 0.5 mL of the vaccine by the s.c. route into each of 5 mice, each weighing 17-22 g. Observe the mice at least daily for 7 days. The vaccine complies with the test if no animal shows notable signs of disease or dies from causes attributable to the vaccine.


THE TCP TEST

Not mentioned in Ph.Eur. monograph! in process test: responsibility of the

manufacturer

same approach as for MLD test in vivo and in vitro would need to be run in

parallel to establish correlation


81

____________________________________________________________________

THANK YOU FOR YOUR ATTENTION


82

____________________________________________________________________

B2. Design of a potential follow-up study with full validation of an optimised protocol for the cell-based assays developed in BSP130 Speakers: Dr K. Redhead, Dr L. Bruckner, Mr A. Daas B2.1. Study design and participants

In the discussion on the study design, the choice of clostridial species for follow-up study was debated first. It was decided to not extend the follow-up study to other clostridial species and to focus on Cl. septicum only. Concerning the methods to be used, it was noted that comparison between in vivo and in vitro tests would not be necessary due to the good concordance obtained in the previous study. Consequently, it was agreed that only in vitro MLD and TCP assays would be evaluated in the next study. Regarding toxins and toxoids to be used, it was decided to maintain their number at six of each.

The next question was the minimum number of participants required for the follow-up study. It was agreed that, in the follow-up study, there should be at least 10 participants and the maximum number of participants was fixed at 15 (taking into account that an increase in participant numbers would result in an unjustified increase in the amount of data to analyse). This would include laboratories already participating and a few new laboratories which were familiar with the cell culture test methods. It was emphasised that, with the aim of global harmonisation, it was intended to perform a worldwide study where not only European manufacturers but also laboratories from the Americas should be included. Therefore, laboratories willing to participate in the follow-up study were invited to contact the project coordinators.

Concerning the level of standardisation of the cell lines and media to be used, there was a split in opinion. On one hand, it was suggested that participants be provided with reference numbers of a centrally reserved (eg ATCC) Vero cells batch and of a commercial cell culture medium. This would avoid variations in results caused, in particular, by different cell sensitivity. But on the other hand, it was recalled that, in line with the usual Pharmacopeia methods, the standard operating procedure provided for the study should not be unreasonably restrictive and should allow the use of Vero cells lines of different origins. Consequently, it was decided that participants would be requested to use their own Vero cell banks and, where these were not available, it would be possible to advise participants on suitable sources for Vero cells and culture medium.

Regarding the study design, the number of independent assays performed for each study method was considered. It was decided that 3 independent assays per laboratory were enough for the follow-up study. As additional independent assays were still needed for internal method validation purposes, participants may decide to increase the number of repeats, if they have enough reagents and test samples.


83

____________________________________________________________________

In the latter case the participants would be advised to report the results of the first three valid assays to the EDQM only. Indeed no resources were available at the EDQM to treat additional data which, though useful for individual participants’ in-house method validation demonstration, were not needed for the purpose of the collaborative study.

B2.2. Samples and protocol

Concerning the final MLD assays, two possibilities were discussed. The first one was to maintain the MLD test which is based on cell death from diluted toxin and to optimise its conditions, for example by means of finer dilution steps. The second one was to use standard antitoxin (VI), i.e. an approach based on the amount of a standard antitoxin required to neutralise cell death. In general, participants were in favour of standard antitoxin (VI) use. It was assumed that this would greatly reduce the variability of results linked with different sensitivity of cells and so allow their comparison between laboratories. Moreover, some participants reported that they already used standard antitoxin (VI) in their tests and were confident of the results. After brief debate it was proposed to opt for the Toxin/Antitoxin approach (“Toxin Equivalence” by reference to a standard antitoxin).

Another question was linked to the use and the supply of standard antitoxin (WHO IS VI)xii. In particular, it has to be confirmed whether NIBSC has sufficient stocks to allow the provision of VI for the study and for its subsequent use in routine in-process controls of Cl. Septicum vaccines by manufacturers. It was assumed that 1 vial containing 500 IU of Cl. septicum standard antitoxin (VI) per laboratory would be enough for the collaborative study, Nevertheless it was concluded that the amount of standard antitoxin (VI) needed for the follow up study had to be calculated with a view to ensuring sufficiency for the planned number of MLD and TCP assays, as some participants reported that they had run out of their stock in the previous study.

Concerning the final TCP assays, it was asked whether these assays should be performed as suggested by the statistician in the annex to the study report, or if participants would have to start with large dilution steps followed by finer tuning and, in this case, the initial and finer dilution steps would need to be predetermined or decided by the participants themselves. After brief discussion, it was decided to choose the second option, i.e. participants would perform their own dilutions using a pre-established list of recommendations.

The next question related to toxins and toxoids supply. The session chairs pointed out that the study samples might be either typical or experimental batches with no need for toxins and toxoids to be obtained in the same production. One of manufacturers present at the meeting expressed his willingness to supply toxins and toxoids. Some others manufacturers promised to return to the project coordinators with their final decisions on toxins and/or toxoids supply as soon as possible. It was decided to collect at least 6 toxins and 6 toxoids with a volume of at least 1 litre (1L)


84

____________________________________________________________________

http://www.nibsc.org/documents/ifu/VI.pdf


of each one. As for their final concentration, dilution of some of the toxins and toxoids provided might be considered, after preliminary testing in the dispatch centre, to obtain a larger range of test materials. The detecting toxin would be chosen from among the supplied toxins after characterisation by the dispatch centre. Thus the established detecting toxin would have to have the highest possible titre to ensure that it was available in amounts necessary to fulfil the needs of the study.

B2.3. Financial and logistical support to the follow-up study

It was added that to ensure that financial support to the follow up study is granted by EPAA (EPFIA and by the European Commission), the project needed to have an industrial champion, i.e. a manufacturer which would be actively involved in the study management. So the choice of dispatch centre was discussed and a potential applicant volunteered. Official confirmation will be sought by the EDQM after the workshop. Also the possibility of involving ECVAM in the study management will be explored by the EDQM. Additional ways to allow the start, completion and dissemination of the study results and promotion of regulatory acceptance of the established alternative methods will also be sought.


85

____________________________________________________________________

Study to be based round:

Cl. septicum only?

Another clostridial species?

More than one clostridial species?

In vitro only assays to be employed for MLD and TCP?

Number of toxins and toxoids to be used:

Six as previously?

Five (one at 3 concentrations and two from other manufacturers)?


86

____________________________________________________________________

Minimum number of participants required?

How do we allow for drop outs?

Participants to use their own cell lines, media etc?

How many tests on each sample:

Three duplicate repeats as previously?

Increase the number if only in vitro test will be used?

Final MLD assays to be performed as:

Doubling dilutions as previously?

Finer dilution steps, to be determined?

Based on cell death from diluted toxin?

Based on the amount of a standard antitoxin required to neutralise cell death?

If Cl. septicum is to be used:

Do we use the same standard antitoxin (VI)?

Does NIBSC have sufficient VI for the study and subsequent use?


87

____________________________________________________________________

Final TCP assays to be performed as:

Suggested by Arnold in report annex?

Start with large dilution steps followed by finer tuning?

Initial and finer dilution steps to be predetermined or decided by participant?

Possibility of replacement of vaccine potency MNT:

Do we want to do this?

Will this require mouse MNTs to be performed?

Can we use sera and MNT values produced as part of manufacturers routine QC testing?

Which facilities will be able to:

Participate in the study?

Supply toxins and/or toxoids?

Supply retained vaccine potency rabbit sera and test values?

Act as a centre for material receipt, preparation and dispatch?


88

____________________________________________________________________

B3. Possibility of using same cell culture based assay system for replacing second step of current potency assay in rabbits (i.e. indirect toxin neutralisation assay in mice)

Speaker: Dr M-E. Behr-Gross

The speaker quickly described the in vivo potency tests stipulated in Ph. Eur. monographs on clostridial vaccines. These challenge tests in laboratory animals (direct or indirect) were allowed but raised many concerns due to their potential for generating distress, harm, suffering and death of large numbers of laboratory animals. Thus, the replacement by alternatives was prescribed with a view to implementation of the policy of the 3Rs. A number of BSP projects supporting replacement/refinement/reduction of in vivo tests for quality control of clostridial vaccines had been run successfully by the EDQM since 1997. The validated methods were serological methods (Cl. perfringens, Cl. tetani) including ELISA (Cl. perfringens, Cl. tetani) and ToBI (Cl. tetani). The BRPs established were guinea pig and rabbit Cl. tetanus antisera (C2424500 and C2424600) and Cl. multicomponent rabbit antiserum BRP (C2424400). The latter BRP was developed to allow the validation and use of C. multicomponent in vitro and/or cell based assays. However titration in mice against this calibrated antiserum was also permitted and some manufacturers still ran this test routinely.

The potency testing for the Cl. septicum vaccines could be improved by using a cell culture-based seroneutralisation model to replace the 2nd step in indirect challenge. That is, it was proposed to replace indirect challenge in batch potency testing by serology in rabbits followed by in-vitro evaluation of antibody titres. This proposal was based on studies by Salvarani, Lobato and colleagues ix, x, xi where a similar approach was used. In the latter study, for Cl. septicum toxin (produced in-house) the MLD test was performed in three models: in mice, guinea pig and on a Vero cells (assay was similar to the one used in BSP130). The toxin was « retrotitrated » using a homologous antitoxin (NIBSC). Rabbits were immunised with commercial vaccines, and neutralising antibodies in rabbit antisera were titrated against toxin in mouse, guinea pig and Vero cell cultures. Good correlation between titres of neutralising antibodies in rabbit sera obtained in mice/guinea pig and Vero cell cultures was demonstrated. The same team performed similar experiments on Cl. perfringens vaccines and demonstrated that BHK cells culture seroneutralisation correlates also to mouse seroneutralisation. To sum up, cell culture seroneutralisation may be applied for titration of rabbit antisera and therefore used for batch potency assay of clostridial multi-component vaccines.


89

____________________________________________________________________

15-16 September 2015

Hotel/Conference Centre Zuiderduin

Egmond aan Zee

Possibility of using a cell culture based assay system

for replacement of the second step of indirect challenge test

in laboratory animals performed in the

potency assay of clostridial vaccines

Marie-Emmanuelle Behr-Gross

EDQM, Council of Europe


90

____________________________________________________________________

Ph. Eur. monographs on clostridial vaccines

Clostridium chauvoei vaccine for veterinary use

Clostridium novyi type B vaccine for veterinary use

Clostridium perfringens vaccine for veterinary use

Clostridium septicum vaccine for veterinary use

Tetanus vaccine for veterinary use

Methodology for potency assay of clostridial vaccines by Ph. Eur

Challenge test in laboratory animals

(direct or indirect) allowed

but implementation of policy on 3Rs strongly supported:

Replacement by alternatives is prescribed


91

____________________________________________________________________

slide adapted from Keith Redhead

Concept of the direct challenge test-step 1

Mice are vaccinated with vaccine sample (diluted)


Concept of the direct challenge test-step2

Toxin is used to challenge the vaccinated mice

Death/survival treshhold is recorded and the test vaccine is calibrated by comparison with a reference vaccine


92

____________________________________________________________________


Concept of the indirect challenge test-1

Sampling of rabbit antisera is done after immunisation

Rabbit are vaccinated with vaccine sample (diluted)


toxin

Reference antitoxin

antitoxin + toxin

Concept of the indirect challenge test-2

Rabbit antisera

or

Death/survival treshhold is recorded and used to calculate the level of antibodies generated by test vaccine. This level is compared to the specification


93

____________________________________________________________________

Consensus

Due to their potential for generating use and death of large numbers as well as

distress, harm and suffering of laboratory animals

challenge tests should be replaced

ICAVAM/ECVAM/national and international 3Rs and animal protection platforms and consortia/Pharmacopoeias/Regulatory authorities/EU

organs and agencies/CoE/manufacturers associations/IABS/…

BSP Projects supporting replacing/refining/reducing

in vivo tests for QC of Clostridial vaccines*

* For more information see Pharmeuropa BIO

Study Topic (vaccine)Method

validated BRP established Start Completion Ph. Eur. Revised

BSP130 C. septicumCell based TCP

and MLDnone 2013/2014 ongoing

BSP053 C. perfringens Serology (ELISA) none 2002 stopped 0363

BSP024 C. tetani (vet use)Serology (ToBI

and ELISA)

guinea pig and rabbit antisera (C2424500 and

C2424600)

1997 2000 0697

BSP022 C. multicomponent Nonerabbit antiserum

(C2424400 )1997 1999

0362, 0363, 0364, 0697


94

____________________________________________________________________

Status for potency assay methodologyof Clostridial vaccines

Implementation of neither animal friendly nor animal free test (full in vitro test) as replacement of direct/indirect challenge is yet completely achieved for most products.

How to rapidly improve the situation ?

Proposal : Where no alternative has yet been developped consideration could be given to use a cell culture-based seroneutralisation model to replace the 2nd step in indirect challenge

Example of the potency assay for Clostridium septicum vaccines

(Ph. Eur. 8th ed. 0364)

Principle: Indirect challenge

but

Batch potency testing may de done by serology in rabbits followed by

in-vitro evaluation of antibody titres


95

____________________________________________________________________

Previous experience in replacing the mice by cells as indicator of toxicity in the 2nd step

of indirect challenge for C. septicum vaccines

Work by Salvarani, Lobato and colleagues(publications distributed as working documents)

Briefly, the authors determined for a C. Septicum toxin(produced in–house)

- MLD in mice, guinea pig and on a Vero cell-based assaysimilar to the one developped at MSD (recording of cytopathic effect of toxin/toxoid)

- « Retrotitrated » the toxin using a homologous antitoxin(NIBSC)

Salvarani, Lobato and colleagues

-immunised rabbits with commercial vaccines and titratedthe neutralising antibodies in rabbit antisera against toxin in mouse, guinea pig and Vero cell cultures


96

____________________________________________________________________

Salvarani, Lobato and colleagues

-showed correlation between titres of neutralising antibodies in rabbit sera obtained in mice/guinea pig and Vero cell cultures

Previous experience in replacing the mice as indicator of toxicity in the 2nd step in indirect

challenge

The same team performed similar experiments on C. perfringens vaccines and demontrated BHK cells culture seroneutralisation correlates to mouse seroneutralisation

ConclusionsCell culture seroneutralisation may be applied for titration of rabbit antisera and therefore used for batch potency assay of clostridial multi-component vaccines


97

____________________________________________________________________

Discussion

Feedback from WS participants on proposal

« Where no alternative has yet been developped consideration should be given to use a cell culture-based seroneutralisation model to replace 2nd step in indirect challenge »

(for an interim period as long as no purely in vitro test is available)

Discussion

Need for validation ?

- only in-house (at manufacturers and/or OMCL)?

- by international collaborative study ?

If yes: Through BSP130 follow up study ?


98

____________________________________________________________________

Potential for replacing in vivo tests forcell culture-based seroneutralisation model

Toxicity of toxin (Minimum lethal dose, MLD)

Toxicity of toxoid (MLD)

Antigenicity of toxoid (total combining power, TCP)

Batch Potency of vaccine

BSP130Follow-up

study

Discussion

Additional samples and reagents needed ?

- rabbit serum pools from routine QC

Donators ?

- C2424400 Clostridia (multi-component) rabbit antiserum(for vaccines-vet.use) BRP at 7.5 IU/vial C. septicum antitoxin

Provided by EDQM


99

____________________________________________________________________

Discussion

Consequences for

study design ?

workload ?

schedule ?

AoB

Thank you for your attention and participation!


100

____________________________________________________________________

Potency assay for Clostridium septicum vaccines (Ph. Eur. 8th ed. 0364)

Principle: Indirect challenge

but

Batch potency may de done by serology in rabbits followed by in-vitro evaluation of

antibody titres

3-4. Potency. Use for the test not fewer than 10 healthy rabbits, 3-6 months old. Administer to each rabbit by the subcutaneous route a quantity of vaccine not greater than the minimum dose stated on the label as the 1st dose. After 21-28 days, administer to the same animals a quantity of the vaccine not greater than the minimum dose stated on the label as the 2nd dose. 10-14 days after the 2nd injection, bleed the rabbits and pool the sera.

The vaccine complies with the test if the potency of the pooled sera is not less than 2.5 IU/mL.

The International Unit is the specific neutralising activity for C. septicum toxin contained in a stated amount of the International Standard, which consists of a quantity of dried immune horse serum. The equivalence in International Units of the International Standard is stated by the World Health Organization.

The potency of the pooled sera obtained from the rabbits is determined by comparing the quantity necessary to protect mice or other suitable animals against the toxic effects of a dose of C. septicum toxin with the quantity of a reference preparation of Clostridium septicum antitoxin, calibrated in International Units, necessary to give the same protection. For this comparison, a suitable preparation of C. septicum toxin for use as a test toxin is required. The dose of the test toxin is determined in relation to the reference preparation; the potency of the serum to be examined is determined in relation to the reference preparation using the test toxin.


101

____________________________________________________________________

Batch potency test. It is not necessary to carry out the potency test (section 3-4) for each batch of vaccine if it has been carried out using a batch of vaccine with a minimum potency. Where the test is not carried out, an alternative validated method is used, the criteria for acceptance being set with reference to a batch of vaccine that has given satisfactory results in the test described under Potency and that has been shown to be satisfactory with respect to immunogenicity in the target species. The following test may be used after a satisfactory correlation with the test under Potency (section 3-4) has been established.

Vaccinate rabbits as described under Potency and prepare sera. Determine the level of antibodies against the toxin of C. septicum in the individual sera by a suitable method such as an immunochemical method (2.7.1) or neutralisationin cell cultures. Use a homologous reference serum calibrated in International Units of C. septicum antitoxin. Clostridia (multicomponent) rabbit antiserum BRP is suitable for use as a reference serum. The vaccine complies with the test if the level of antibodies is not less than that found for a batch of vaccine that has given satisfactory results in the test described under Potency and that has been shown to be satisfactory with respect to immunogenicity in the target species.


102

____________________________________________________________________

B4. Round table discussion – Meeting conclusions and summary Chairs: Dr M-E. Behr-Gross, Dr L. Bruckner, Dr K. Redhead

In the discussion which followed, several issues were addressed.

Many participants commented on safety and residual toxicity tests stipulated in the Ph. Eur. monographi for batch testing and the similar in-process residual toxicity testing also performed by manufacturers. With respect to residual toxicity testing, it was firstly noted that there was some confusion between testing of vaccine with adjuvant (that may have a toxic effect) and toxoid in-process control testing. The session chair posed the central question of what is measured in the residual toxicity test in vaccine: residual toxicity itself or a combined effect linked with adjuvant which can be also toxic? The reversion to toxicity of clostridial septicum toxoids seems to be very rare. Thus, the place of residual toxicity testing in the manufacturing process needs to be clarified. At the same time it was recalled that, according to the Ph. Eur. Monograph, residual toxicity testing was performed on the final product, i.e. on vaccine. But within the manufacturing process, residual toxicity was tested on different steps of production: in the earliest stage, i.e. on bulks immediately after inactivation of toxin (in-process control) and later immediately before addition of adjuvant (batch testing). It was mentioned that the first test was an important in-process check allowing quality control of the detoxification procedure and so it should not be omitted. Hence this in-process residual toxicity test should be replaced by in vitro Vero cell assays.

A participant expressed his concerns about the outcome of replacement of qualitative in vivo testing by quantitative in vitro testing and his position with regards to the regulatory authorities. In particular he commented on acceptance criteria for replacement tests. For instance, if for in vivo safety tests acceptance criteria were “Yes/No”, what would be the acceptance criteria for the alternative in vitro assays? Taking into account the high sensitivity of in vitro assays, what response level should be accepted for a test performed in vitro. It was suggested that, in the case of in vivo test replacement, acceptance criteria for alternative in vitro testing should be established a posteriori by means of data accumulation, application of a consistent approach and establishment of trends in parallel with in vivo tests. The resulting correlation would allow, as necessary, acceptance by the regulatory authorities.

Regarding the project of replacement of the 2nd step in indirect challenge by a cell culture-based seroneutralisation model, it was noted that this might be interesting for participants who have not yet developed alternatives for potency testing in animals. Even if most participants reported that they already used alternative methods, some participants currently using challenge tests in animals were invited to consider applying this approach. The possibility was raised of running these experiments in parallel with BSP130 due to the similarity of approach. However, despite the evident


103

____________________________________________________________________

advantages of an international collaborative study which would have facilitated the introduction of these alternative assays and, as a result, allowed the complete removal of mouse assays for MLD, TCP and serological potency assay, simultaneous running of both studies might complicate the protocol and could increase the risk of failure. For these reasons it was decided to opt for in-house validation only for the proposed replacement method. For this purpose EDQM would supply, upon request, C2424400 Clostridia (multi-component) rabbit antiserum (for vaccines-vet.use) BRP at 7.5 IU/vial Cl. septicum antitoxin.

At the closure of the meeting, it was reported that it was planned to complete the report of the BSP130 collaborative study and to produce a report of the present workshop. It was also intended to prepare a publication recording the outcomes of BSP130 and to submit it to “Pharmeuropa Bio and Scientific notes” where it would be publicly available. For the follow-up study, it was planned to submit the project extension to the BSP Steering Committee and to prepare an application to be granted EPAA support. In anticipation of positive feedback from both Steering Committee and EPAA, the work on the project would be started. The project leaders would be recruited and, in collaboration with project leaders, the draft protocol would be elaborated. Participants would be invited and they would be asked to comment on the draft protocol. One of the laboratories would be chosen as repository and testing centre where test material donations would be centralised, the proposed optimised protocol would be checked and preliminary testing of test materials and study samples would be performed. Finally, the collaborative study would be started in 2016.


104

____________________________________________________________________

APPENDIX: BSP130 Workshop Participants

Management team:

• Marie-Emmanuelle Behr-Gross (project coordinator, EDQM)

• Arnold Daas (statistician, EDQM)

Project leaders:

• Keith Redhead (previously MSD, UK)

• Lukas Bruckner (previously IVI, CH)

Participants:

Public sector representatives:

• Sabahatin ICIN (Bornova Veterinary Control Institute, Turkey)

• Elisabeth BALKS (Paul Ehrlich Institute, Germany)

• Sylvie JORAJURIA (European Directorate for the Quality of Medicines & HelthCare, Council of Europe)

Industry representatives:

• Balazs DALMADI (CEVA-Phylaxia Veterinary Biologicals Co. Ltd., Hungary)

• Botond SIKLODI (CEVA-Phylaxia Veterinary Biologicals Co. Ltd., Hungary)

• Conchita FERNANDEZ (CZ Veterinaria, Spain)

• Goncalo LEITE (MSD Animal Heath, UK)

• Oscar MARTINEZ (SYVA, Spain)

Observers:

• Eriko TERAO (EDQM)

• Nadia ROMERO GALLEGOS (Boehringer Inhelheim Vetmedica, SA de CV, Mexico)

• Rudiger RAUE (Zoetis, Belgium)

Scientific writer:

• Natalia Sinitskaya


105

____________________________________________________________________

REFERENCES

i. Clostridium septicum vaccine for veterinary use. Monograph No 0364 European Pharmacopoeia, 8th Edition (2013).

ii. Ebert E, Kusch M, Öppling V, Werner E, Cussler K. An in vitro alternative to the mouse neutralisation assay for potency testing of betatoxoid containing Clostridium perfringens type B and type C vaccines for veterinary use. ALTEX. 1996;13(2):68-75.

iii. Lensing HH, Esposito-Farese ME, Daas A, Spieser JM. Collaborative study for the establishment of two European Pharmacopoeia Biological Reference Preparations for serological potency testing of tetanus vaccines for veterinary use. Pharmeuropa Spec Issue Biol. 2000-2 Feb; 2001:45-54.

iv. Lucken R, Daas A, Esposito-Farese ME. Collaborative study for the establishment of a European Pharmacopoeia Biological Reference Preparation for Clostridia antiserum for serological potency testing of clostridial vaccines for veterinary use. Pharmeuropa Spec Issue Biol. 2000-2 Feb; 2001:65-87.

v. Ph. Eur. Reference Standard – leaflet. Clostridia (multi-component) rabbit antiserum (for vaccines for veterinary use) BRP batch 1. https://crs.edqm.eu/db/4DCGI/View=C2424400 consulted on 16 December 2015.

vi. Ph. Eur. Reference Standard – leaflet. Clostridium tetani guinea pig antiserum (for vaccines for human use) BRP batch 1. https://crs.edqm.eu/db/4DCGI/View=C2424500 consulted on 16 December 2015.

vii. Ph. Eur. Reference Standard – leaflet. Clostridium tetani rabbit antiserum (for vaccines for veterinary use) BRP batch 1. https://crs.edqm.eu/db/4DCGI/View=C2425600 consulted on 16 December 2015.

viii. Rosskopf-Streicher U, Volkers P, Werner E. Control of Clostridium perfringens vaccines using an indirect competitive ELISA for the epsilon toxin component - examination of the assay by a collaborative study. Pharmeuropa Bio. 2004 Jan; 2003(2):91-6.

ix. Salvarani FM, Lobato ZIP, Assis RA, Lima CGRD, Silva ROS, Pires PS, Lobato FCF. In vitro evaluation of Clostridium septicum alpha toxoid. Arq. Bras. Med. Vet. Zootec. 2010 Aug; 62(4):778-783.

x. Salvarini FM, Lobato ZIP, Pires PS, Silva ROS, Alves GG, Pereira PLL, Lobato FCF. In vitro potency test for evaluation of Clostridium perfringens type D epsilon toxoid. Arq. Inst. Biol. São Paulo, 2013; 80(4):450-452.


106

____________________________________________________________________

https://crs.edqm.eu/db/4DCGI/View=C2424400



xi. Souza Júnior MF, Lobato ZIP, Pires PS, Silva ROS, Salvarani FM, Assis RAde, Lobato FCF. Standardization of the titration of the epsilon toxin of Clostridium perfringens type D in cell line as an alternative to animal bioassay Cienc. Rural, Santa Maria Mar. 2010 Feb; 40(3):600-603.

xii. WHO International Standard Clostridium septicum (Gas Gangrene) Antitoxin Equine, 3rd International Standard NIBSC code: VI Instructions for use (Version 7.0, Dated 24/01/2014) http://www.nibsc.org/documents/ifu/VI.pdf consulted on 16 December 2015.


107

____________________________________________________________________


The Council of Europe is the continent’s leading human rights organisation. It comprises 47 member states, 28 of which are members of the European Union. The European Directorate for the Quality of Medicines & HealthCare (EDQM) is a directorate of the Council of Europe. Its mission is to contribute to the basic human right of access to good quality medicines and healthcare and to promote and protect public health.

www.edqm.eu

ENG

Following the 3Rs concept: Replacement, reduction and refinement of animal assays as proposed by Russell and Burch in 1959, the Council of Europe, a pioneer in the field of 3Rs, created in 1986 the first European legally binding instrument by opening for signature the International European Treaty (ETS No123) “Council of Europe Convention for the protection of Vertebrate Animals used for Experimental and Scientific purposes”.

In line with this policy, the European Pharmacopoeia has been actively involved in the replacement, reduction and refinement of animal assays for the quality control of medicines and its secretariat, the European Directorate for the Quality of Medicines and Healthcare (EDQM) also contributes to a technical Platform for 3Rs of the European Partnership for Alternative Approaches to animal Testing (EPAA), a partnership between the European Commission (EC) and industry.

Within this framework, the Quality Control methods for several vaccine categories were considered from the 3Rs perspective and potential possibilities for improvement were evaluated. Due to the large numbers of animals currently used in Quality Control testing of clostridial vaccines, work in this field was given the highest priority leading, in 2013, to the start of a study aimed at supporting the concept of using an alternative (cell line model) to the mouse model as toxicity indicator for clostridial vaccines in-process testing. The study, performed under the aegis of the EDQM/Council of Europe Biological Standardisation Programme with the support of EPAA, allowed demonstrating overall concordance between in vitro (cell line-based) and in vivo methods. A workshop, during which results and follow up activities were discussed, took place in Egmond aan Zee on 15&16 September 2015 and its report is presented herein.

clostridium septicum vaccines - circabc - welcome · of clostridium septicum vaccines ... the...

Documents