rmm validation: the role of statistics...bact/alert 89/95 47/97 10/56 real microbiological...

RMM Validation:The Role of Statistics

Edwin van den HeuvelProfessor in Statistics

October 2017

PAGE 111-10-2017

Warning: Statistics Presentation

Content:• Measurement Reliability

– Classical– Microbiological– Statistical Theory

• Absence/Presence Tests– Hypothesis testing– Example Data– Simulations

– Process Sampling

• Conclusions

Measurement ReliabilityClassical Theory

PAGE 211-10-2017

Statistical Model:= +• is an arbitrary value• is true value• is measurement error• ~ , is assumed

normally distributed• + is mean value• − is random error

Theory of Errors• is systematic error (bias)• is standard deviation of

random error • Introduced by Carl Friedrich

Gauss (1777-1855)

Measurement ReliabilityClassical Theory

PAGE 311-10-2017Blank

Limit of Quantitation

Upper Quantile

Spike 1 Spike 2 Spike 3

Calibration line provides:• Bias at each spike level• Linearity of measurement method:Spike = Spike• Quantitation (and detection) limit> = Small• Repeatability – from repeated

measurements per spike

Accuracy Experiment:• Use of precise spikes/standards• Measured multiple times

……

….

……

….

……

….

……

….

……

….

Measurement ReliabilityMicrobiological Validation

PAGE 411-10-2017

…………....

…………....

Ideal Microbiological Experiments:• Repeated blank samples for specificity

• Repeated samples with one microorganism for detection limit

• Repeated samples with higher numbers of microorganisms for accuracy

False positive rate

False negative rate

………….... Bias – Spike 1

…………....Bias – Spike Q

uantit

ativ

e T

est

sQ

ualit

ativ

e T

est

s

Traditional statistical analysis methods can be applied to the ideal microbiological experiment

Real Microbiological Experiments:


PAGE 511-10-2017

……

= 1= 0= 1= 1= 2

= 5Spikeuncertain

Dilution(volume )

True number of organisms in test sample is Binomial with and = /

= 2= 4

Qualitative Tests = 4

= 0= 5= 2

Quantitative Tests

Test samples(volume )


PAGE 611-10-2017

Methods 10 CFU 1 CFU 0.1 CFU

Compendial Membrane Filtration 134/138 101/138 20/89

Direct Inoculation 53/56 36/56 10/50

Rapid Milliflex (TSA/SDA) 75/77 66/77 21/70

Rapid Milliflex (SBA) 40/40 39/40 29/33

BACTEC 74/77 38/77 12/57

BacT/Alert 89/95 47/97 10/56

Real Microbiological Experiments:• Spiking and test samples can cause unexpected results• False positives and false negatives can compensate each other• Different organisms may results in different test behavior • This may all lead to non-interpretable results

Parveen S, Kaur S, Wilson David SA, Kenney JL, McCormick WM, Gupta RK, Evaluation of growth based rapid microbiological methods for sterility testing of vaccines and other biological products, Vaccine, 2011, 29:8012-8023


PAGE 711-10-2017

Or even worse…..

“…..You use the (b)(4) method to screen for microbiological contamination in drugs produced entirely at your facility and those manufactured under contract. This (b)(4) screening method (b)(4) for microbiological examination of your liquid drug products is not adequate for its intended use. You attempted to validate your (b)(4) microbial detection method, but were not able to demonstrate that it could reliably and repeatedly determine whether objectionable microorganisms were present in your drugs…..”

https://www.fda.gov/ICECI/EnforcementActions/WarningLetters/2017/ucm568120.htm

Statistical Detection/Enumeration:

Measurement ReliabilityStatistical Theory

PAGE 811-10-2017

…… = 2

= 5Spikeuncertain

Dilution(volume )

= 2= 4

Test samples(volume )

• is true number of organisms in sample ∈ 1,2, … ,

• is the observed result in sample ∈ 1,2, … ,

• Probability mechanisms= = | =– Detection: ∈ 0,1– Enumeration: ∈ 0,1,2, …

• Validation means understanding the probability mechanism

• This is not in conflict with the guidelines EP and USP

PAGE 911-10-2017


Detection Ideal Experiment:• is the true number of organisms• is the detection proportion• is false positive rate• Zero-deflated Binomial (ZDBM)= = 01 − 1 − 1 − > 0• Binomial (BM) has = 0: no false positives

Detection Real Experiment:• is the positive rate over many samples• Zero-deflated Binomial (ZDBM)≡ 1 − 1 − 1 −1 − 1 − exp −• = is the bacterial density• Additionally, spike is imprecise• Binomial (BM) has = 0

11-10-2017


PAGE 10

Enumeration Ideal Experiment:• All test samples had = 40 organisms• is the observed number of organisms• is the detection proportion• Binomial (BM)= | = = 1 −• Observed variance: 1 −

Enumeration Real Experiment:• Test samples had = = 40

organisms on average• Binomial (BM)= = 1 −• Observed variance: 1 −• Additionally, spike is imprecise

Non-inferiority (USP):• is the non-inferiority margin• USP suggests = 0.20: < − vs. : ≥ −• Hypothesis test is based on one-

sided confidence interval

Real Experiments:• Positive rate compendial: • Positive rate alternate:

• and are estimated from data from a real experiment

Absence/Presence TestsHypothesis Testing

PAGE 1111-10-2017

Traditional Hypothesis (EP):• One-sided testing: ≥ vs. : <• Null hypothesis is rejected when

the alternative hypothesis is demonstrated to be likely

• Hypothesis test is based on one-sided confidence interval

−

A

B

C

D

E

0

Non-infe

rior

“Infe

rior”

F

− 0“Non-inferior”

Inferior

11-10-2017

Absence/Presence TestsExample Data

PAGE 12

Compendial Test Method

Sample Result Sample Result

1 1 11 1

2 1 12 1

3 0 13 1

4 0 14 1

5 0 15 1

6 1 16 1

7 0 17 1

8 1 18 0

9 0 19 0

10 0 20 0

Alternative Test Method

Sample Result Sample Result

1 0 11 1

2 0 12 1

3 1 13 1

4 0 14 1

5 1 15 1

6 1 16 1

7 0 17 0

8 0 18 1

9 1 19 0

10 1 20 1

• Non-inferior at = 0.2• : ≥ against : < is not

rejected

• Compendial: = 11/20• Alternate: = 13/20• Difference (90%CI): 0.10 −0.15; 0.35

11-10-2017

Absence/Presence TestsSimulation Study

PAGE 13

CFU/sample20 samples per method 50 samples per method

Non-inferiority Rejected Non-inferiority Rejected

1.0 15.9% 15.7% 29.6% 21.5%

1.5 16.4% 20.5% 27.1% 32.1%

2.0 19.7% 20.0% 32.3% 36.8%

2.5 29.3% 18.9% 44.9% 39.1%

3.0 45.2% 14.9% 61.2% 38.1%

3.5 63.1% 9.5% 78.6% 36.4%

4.0 78.0% 5.1% 91.1% 31.1%

4.5 88.1% 2.4% 97.2% 22.0%

5.0 94.0% 1.0% 99.2% 13.3%

Simulated Situation:• Compendial: = 0.90 and = 0• Alternate: = 0.60 and = 0.10• Alternate is inferior on proportion of detection: − < −0.2• Suggested approach in guideline EP and USP are insufficient

Probability of detecting contamination• False positives will be corrected by identification methods

Absence/Presence TestsProcess Sampling

PAGE 1411-10-2017

……

.Dilution= 5000L

Minimal bacterial density (CFU/L) in a fermenter that can be detected with 95% confidence

Probability Mechanism Density• Perfect 30.2• BM ( = 0.9) 33.4• ZDBM ( = 0.1, = 0.6) 47.0

Fermenter = 0= 0= 0= 0

10 test samples of = 10mL

11-10-2017

Conclusions

PAGE 15

• Really understanding microbiological test methods is understanding the probability mechanism

• Real validation experiments• Are imperfect• Are imprecise on spiked numbers• Introduces sample-to-sample

variation• Complicate the understanding

of probability mechanism

• Dictate a comparison between the alternate and compendial

• Inferior alternate methods may be difficult to identify with too limited test samples

• Thus EP and USP should mention probability mechanisms explicitly

Validation of two rapid sterility test methods

Dr David Roesti, Novartis Pharma Stein AG11th October 2017

NTO BTDM

D. Roesti 11th October 2017

Disclaimer

These slides are intended for educational purposes only and for the personal use of the audience

These slides are not intended for wider distribution outside the intended purpose without presenter approval

The content of this slide deck is accurate to the best of the presenter‘s knowledge at the time of production

Public2


Agenda

Introduction

Validation Celsis Advance® System

Validation Millipore Milliflex® Rapid System

Points to consider for validation

Public3


Agenda

Introduction




Public4


Introduction

The compendial sterility test relies on visual detection of microorganisms and is described in USP <71> and Ph. Eur. 2.6.1

2 media (TSB; FTM) incubated for at least 14 days (20–25 °C TSB; 30–35 °C FTM)

If product renders the medium turbid, portions of the medium are transferred to fresh vessels of the same medium after 14 days + subsequent additional incubation for at least 3-5 days (total of 19 days)

Slow-growing microorganisms may require several days or even weeks until they are visually detectable by the unaided eye

Public5


Guidance Documents RMM validation

A validation should demonstrate that the RMM is suitable for its intended purpose !

Guidance on the validation of RMMs: USP <1223> - Validation of alternative microbiological methods

Ph. Eur. 5.1.6. - Alternative methods for control of microbiological quality

PDA Technical Report No. 33 - Evaluation, validation and implementation of alternative and rapid microbiological methods

The parameters differ between qualitative and quantitative methods as well as between the documents themselves

Public6


RMM Validation Parameters

Public7

Validation parameter Qualitative RMM Quantitative RMM

Robustness Yes Yes

Ruggedness Yes Yes

Repeatability Yes Yes

Specificity Yes Yes

Limit of Detection Yes Yes

Accuracy & Precision No Yes

Limit of Quantification No Yes

Linearity No Yes

Range No Yes

Non-inferiority in routineoperation

Yes Yes

Combination of parameters from USP <1223>, Ph. Eur. 5.1.6., PDA Technical Report #33


Agenda

Introduction




Public8


Validation RST-DICelsis Advance® System (Charles River)

Features

• Incubation time reduced from 14-19 days to 7 days

• Detection through ATP-bioluminescence

• Non filterable samples may be tested

• Growth in TSB and FTM

• Detects up to 1 CFU after incubation

• End point reading

• Aliquots of samples are transferred in Celsis cuvettes at the end of the incubation time

• Staining and reading with the Celsis Multitube Luminometer

• Microbial growth based on a ATP bioluminescence signal translated into relative light units (RLU)

• A sample is declared as positive (microbial growth detected) if the RLU number exceeds a pre-defined threshold

• Validated method only applicable for products with low ATP background

• Micro-ID possible from original sample culture

Public9


Validation RST-DI

After incubation, sample pre-treatment to detect molds Some microorganisms (especially molds) form aggregates during growth

which may lead to false negative results

Shearing and even distribution of fungal aggregates/mycelia by stirring for 30 min. with a magnetic stir bar and glass beads

Centrifugation step to concentrate slow growing microorganisms In pre-studies, the slow growing bacterium M. radiotolerans could not be

detected robustly under the test conditions

After disaggregation, step, small samples (1.5 mL) are concentrated 7.5 x

Public10

Without pre-treatmentWith pre-treatment


Validation RST-DILimit of detection and non-inferiority in routine operation

The Limit of Detection is defined as the lowest number of microorganisms that can be detected, but not necessarily enumerated, under the stated experimental conditions

The Limit of Detection must be appropriate for the intended application

The Limit of Detection is often assessed in comparison to a reference method

Public11

Non-inferiority in routine operation should demonstrate results with the alternative method at least as good as the compendial method.

Parallel testing of exemplary product / real test sample with both the alternative method and the reference method

If exemplary test material does not harbor microorganisms at a sufficient frequency (e.g. sterile manufactured drug product), the sample has to be artificially inoculated with an appropriate number of microorganisms



Limit of Detection assessed for 4 stressed in-house slow growing isolates (including fungi, endospores and anaerobic microorganisms)

Direct comparison to compendial sterility test

Serial dilution to extinction: 5 CFU, 0.5 CFU and 0.05 CFU

4 test runs with 10 replicates per concentration and microorganism.

Public12

For non-inferiority, RST-DI experiments were also conducted with presence of exemplary product types (e.g. microparticles, syringes, ampoules and vials) in order to verify that these do not negatively interfere with the test method. The expectation was that the RST-DI performs non-inferior to the traditional method irrespective of the presence of the different product types.



Statistical evaluation: The 95 % confidence interval of the most probable number according

to a ten-replicate MPN-table obtained for RST-DI with or without product must overlap with the respective confidence interval of the compendial test

No significant difference with Repeated Measures ANOVA between the obtained MPN values of all test strains, product types and runs

Fisher’s exact test modified for demonstrating one-tailed equivalence of success rates on pooled presence/absence data of all test strains, product types and runs

Public13


Validation Validation RST-DIMPN method example

Public14

Most Probable Number

RST Compendial ST

MPN/g

Low 95 % Cl

High 95 % Cl

MPN/g

Low 95 % Cl

High 95 % Cl

5.9 2.5 12.0 6.2 2.6 14.0

https://www.fda.gov/food/foodscienceresearch/laboratorymethods/ucm109656.htm

RST Compendial ST

Positive tubes Positive tubes

5 CFU

0.5 CFU

0.05 CFU

5 CFU

0.5 CFU

0.05 CFU

10 4 1 10 5 0


Validation RST-DIMPN method example

Public15

Microorganism 2

RST Compendial ST

MPN/g

Low 95 % Cl

High 95 % Cl

MPN/g

Low 95 % Cl

High 95 % Cl

4.4 2.0 9.1 0.72 0.31 1.5

Microorganism 1

RST Compendial ST

MPN/g

Low 95 % Cl

High 95 % Cl

MPN/g

Low 95 % Cl

High 95 % Cl

1.7 0.77 3.4 1.9 0.9 3.9


Validation RST-DILimit of detection and non-inferiority in routine operationMPN method results

Public16


Validation RST-DILimit of detection and non-inferiority in routine operationAnova results

Public17

Individual MPN values as well as the means are shown (ns, no significant difference with RepeatedMeasures ANOVA for the log-transformed MPN values)


Method B inferior to method A (70% boundary)

Method B non-inferior to method A

(70% boundary)

Validation RST-DINon-inferiority test

The modified Fisher’s test asks if there is non-inferiority to a defined acceptable boundary. The null hypothesis is tested against the one-sided alternative hypothesis HA

HA: pRapid_Method > pTraditional_Method – delta. The null hypothesis is rejected and the alternative hypothesis is accepted if the p-value < 0.05

Delta is a correction factor which allows the method under evaluation to be minimally worse”. In the present case, delta was set to ≤ 0.3 (or 30 %), which was based on the requirement of 70 % recovery for microbiological tests in USP <1227>

Public18


Validation RST-DILimit of detection and non-inferiority in routine operationNon inferiority test results

The pooled presence/absence data of all test microorganisms, product types and runs were evaluated using Fisher’s exact test modified for demonstrating one-tailed equivalence of success rates at a confidence level of 95 %.)

Public19

Results Limit of Detection and non inferiority (Pooled Data)


Validation RST-DIRobustness

Robustness describes the reliability of the method in routine use. The application of “small but deliberate variations in method parameters” must not lead to significantly different results

Robustness determination is best suited to demonstration by the supplier of the method

Supplier data should be reviewed and method parameters specific to the intended application identified and validated

Examples:

Incubation time

Duration of a pre-treatment

Reagent contact time

Public20


Validation RST-DIRobustness

The supplier data was reviewed and three additional robustness parameters identified. These were the incubation time, stirring time and the concentration factor. The RMM method with the worst case parameters was compared with the compendial sterility test method

Two slow-growing and an aggregate forming microorganism were tested with an inoculum of ≤ 5 CFU

2 test runs with 10 replicates per microorganism and test method.

Public21

Fisher’s exact test modified for demonstrating one-tailed equivalence of success rates on pooled data

RST-DI worst case setupGrowth / No Growth

Compendial STGrowth / No Growth

Results 58 / 2 55 / 5

p-value 0.00096

Interpretation of results Statistical proof of equivalence (non-inferiority)


Validation RST-DISpecificity

The specificity of a method is defined as the potential to detect the intended range of microorganisms

Inoculation of ≤ 5 CFU of 18 different microorganisms including gram-negative rods, gram-positive rods, gram-positive sporulating bacteria, gram-positive cocci and yeasts/molds, ATCC strains and in-house isolates.

3 test runs with 3 replicates per microorganism

Public22


Validation RST-DISpecificity

Public23

Growth within 7 days of incubation was tested positive with the Celsis Advance for all test microorganisms. Every single bottle which was negative with the CelsisAdvance did not show visually detectable growth after 14 days (due to low inoculum there is probability that some bottles remain unspiked)


Agenda

Introduction




Public24


Validation MX RapidMillipore Milliflex® Rapid System

Features

• Incubation time reduced from 14 days to 5 days

• Detection through ATP-bioluminescence

• Only filterable samples are tested

• Filtration membrane deposited on modified Schaedler blood agar in isolator

• Incubation of membranes: aerobic at 20-25 °C, aerobic at 30-35 °C, anaerobic at 30-35 °C

• Detects up to 1 CFU after incubation

• End point reading

• End of incubation: separate the membrane from nutrient medium, transfer of membrane onto AutoSpray Station inside Laminar Flow

• Spraying of ATP-releasing agent followed by spraying of the bioluminescence reagent

• The emitted light from microbial colonies is detected using a highly sensitive CCD (charged coupled device) camera.

• A sample is declared as positive (microbial growth detected) if at least one colony is detected

• Validated method only applicable for products with low ATP background

• Micro-ID possible from colonies after ATP treatment following regrowth procedure

Public25


Validation MX RapidLimit of detection and non-inferiority in routine operation

Limit of Detection assessed 5 stressed in-house slow growing isolates (including fungi, endospores and anaerobic microorganisms)

Direct comparison to compendial sterility test

Statistical evaluation: The 95 % confidence interval of the most probable number according to a ten-

replicate MPN-table obtained for MX Rapid must overlap with the respective confidence interval of the compendial test

Fisher’s exact test modified for demonstrating one-tailed equivalence of success rates on pooled presence/absence data of all test strains and runs

Public26

For non-inferiority in routine operations, RST-DI experiments were also conducted with presence of 6 different products that were inoculated with 1-5 CFU of 3 different microorganisms (E. coli, S. aureus and stressed P. acnes). The expectation was that the RST-DI performs non-inferior to the traditional method irrespective of the presence of the different products.


Validation MX RapidLimit of detection and non-inferiority in routine operation

Public27

95 % confidence intervals overlapped in all runs all microorganisms and test conditions with one exception

For one test microorganisms the recovery was actually worse for the compendial test as compared to the MX Rapid

Statistical proof of equivalence (non-inferiority) was demonstrated with the Fisher’s exact test modified for demonstrating one-tailed equivalence of success rates


Validation MX RapidSpecificity

Inoculation of ≤ 5 CFU of 22 different microorganisms including gram-negative rods, gram-positive rods, gram-positive sporulating bacteria, gram-positive cocci and yeasts/molds, ATCC strains and in-house isolates.

10 replicates per microorganism

Direct comparison to the compendial sterility test

Statistical evaluation performed strain-specifically as well as over the whole tested spectrum: Fisher‘s exact test

Fisher’s exact test modified for demonstrating one-tailed equivalence of success rates

Chi-Square test

Public28


Validation MX RapidSpecificity

Public29

• Fisher‘s exact test (individual test strains: All test strains no significant difference. Only one strain with significant better results for RST compared to compendial

• Fisher’s exact test modified (all strains): Statistical proof of equivalence (non-inferiority) • Chi square test (all strains) : No significant difference

Strain Rapid Sterility TestUnsterile/Sterile

Compendial Sterility Test Unsterile/Sterile

C. afermentans 9/1 3/7Penicillium spez. 10/0 10/0Cl. Sporogenes 6/4 6/4B. pumilus 7/3 10/0B. clausii 3/7 5/5B. licheniformis 10/0 10/0B. sphaericus 10/0 10/0B. idriensis 10/0 10/0B. subtilis 10/0 10/0P. acnes 6/4 4/6C. albicans 5/5 3/7Kocuria spez. 10/0 10/0S. warneri 10/0 10/0S. capitis 10/0 8/2S. epidermidis 10/0 10/0A. brasiliensis 5/5 4/6K. rhizophila 8/2 6/4A. lwoffii 3/7 2/8S. aureus 9/1 10/0P. aeruginosa 10/0 9/1E. coli 10/0 10/0S. maltophilia 9/1 10/0All strains 180/40 170/50


Validation MX RapidRobustness

Robustness towards different incubation times & membrane transfer time after reagent spraying

Microorganisms: 10-100 CFU fast-grower and worst-case microorganism (E. coli and stressed P. acnes)

3 test runs with 3 replicates per microorganism

Comparison of (micro-)colonies counts

Statistical evaluation: One-way ANOVA

Public30


Validation MX RapidRobustness

No statistically significant difference (p-values ≥ 0.05, corresponding to a confidence level of 95 %) between the incubation times otmembrane transfer times

Public31


Validation MX RapidProduct-specific Method suitability test

Sample effects study Verification that product does not generate a bioluminescent background. Unspiked

product sample is filtered. Membranes are evaluated using the Millipore Milliflex® Rapid system. The Millipore® Milliflex Rapid system must not report detection of any microbial colonies.

Verification that drug product does not inhibit bioluminescence reaction. Confirmed if P.acnes can be detected in the B&F test.

Bacteriostasis and Fungistasis test Spiking nutrient media containing product with ≤ 100 CFU USP <71> compendial test

strains, P.acnes and + 1 or 2 in-house isolates.

Control is nutrient medium without product.

At least 70 % recovery of colony count as compared to the control.

Parallel testing Requested by health authority during filing.

Testing of 3 different lots of product in parallel with the Millipore® Milliflex Rapid and compendial sterility test.

Result must be same for both methods (e.g. absence of growth vs absence of growth).

Public32


Validation conclusion

Public33

The Millipore Milliflex® Rapid System was successfully validatedwith 5 days incubation as as an alternative sterility test for filterabledrug products

The RST-DI on basis of the Celsis Advance system (using the AMPiScreen™ reagent kit) was successfully validated with 7 days incubation as alternative method for materials or non-filterbale drug products tested for sterility by the direct inoculation method.


Agenda

Introduction




Public34



Thoroughly evaluate your assay early in the project and before validation starts to limit unplanned “surprises” during validation phase

Public35



Rationally evaluate which validation parameters are relevant or critical for the application of interest

Public36

Validation parameter Qualitative RMM

Quantitative RMM

Robustness Yes Yes

Ruggedness Yes Yes

Repeatability Yes Yes

Specificity Yes Yes

Limit of Detection Yes Yes

Accuracy & Precision No Yes

Limit of Quantification No Yes

Linearity No Yes

Range No Yes

Non-inferiority in routineoperation

Yes Yes



Do not underestimate the intrinsic variability of microbiological assays „Classical“ microbiological aspects like nutrient media or growth

conditions may have a much higher impact on assay performance as compared to the technical equipment which allows earlier detection of microbial growth

Public37



Do not uncritically adopt acceptance criteria suggested in the RMM validation guidance documents they may not apply in your particular case

Define well-thought-out acceptance criteria which also take into consideration the process behind the assay For a manufacturing process in-process test, Is it really critical to have

an equivalent LOD as compared to the currently used method, if the time of result is faster?

Can viable particle count signals really be correlated to CFU counts?

Public38



Use a wide variety of potentially relevant microbial species

Using a very limited number of compendial test strains is not really challenging the system and would not reflect realistic testing conditions

Use microbial inocula which are appropriate for the assay of interest (e.g. less than 5 CFU for sterility)

Public39

For growth based methods use of only fast growing microorganisms for validation is nonsense !

Thank you

Acknowledgements:

Alexandra StaerkOliver GordonJennifer Isken

PCR-based Adventitious Agents Testing

International Microbiology Symposium October 10-11, 2017, EDQM, Strasbourg, France

Dr. Sven M. Deutschmann, Roche Diagnostics GmbH, Director QC Pharma Biotech PenzbergHead of gASAT “Adventitious Agents Testing & Alternative Microbiological Methods”

Background

Vision

2

PCR-Based Adventitious Agents Testing

Biological Products: Testing for Adventitious Agents• Global Requirements for Biotechnological / Biological Products: “Contaminants in a product include all adventitiously

introduced materials not intended to be part of the manufacturing process, such as chemical and biochemical materials (e.g., microbial proteases), and/or microbial species (note: e.g. contaminants of viral, bacterial, fungal origin). Contaminants should be strictly avoided and/or suitably controlled with appropriate in-process acceptance criteria […]”*

• Current Test Procedures: The current indicator cell-based and growth-based tests for the detection of adventitious agents assays require (i) incubation / testing times up to 28 days, (ii) manual handling, (iii) specialized equipment and laboratories, (iv) high level of expertise for the result interpretation and are (v) characterized by low sensitivity and limited specificity.

• Vision: Replace all indicator cell-based and growth-based adventitious agents tests with PCR-based detection methods using the same “state-of the art”, fully automated technology platform

* see e.g. ICH Q6B “Specifications: Test Procedures and Acceptance Criteria for Biotechnological / Biological Products”, CPMP/ICH/365/96 3


Target Organisms

Leptospira:

• … is a genus of spirochaete bacteria, including a small number of pathogenic and saprophytic species

• L. licerasiae contaminations occurred in the Biotechnology Industry

Mollicutes:

• … are a class of bacteria distinguished by the absence of a cell wall

• … are parasites of various animals and plants, living on or in the host cells. Many cause diseases in humans (species of Mycoplasma and Ureaplasma in the respiratory or urogenital tracts). Acholeplasma laidlawii may contaminate bovine serum but also occurs in serum-free cell culture media products. Phytoplasma and Spiroplasma are plant pathogens associated with insect vectors.

MMV (or MVM – Minute Virus of Mice):

• … is the prototype virus of the Protoparvovirus genus within the family Parvoviridae

• … is a Biotechnology Industry threat has been seen to come from animal and non-animal-derived components4


Target Organism: Leptospira (1)

Q14.: Can Leptospira species penetrate sterilizing-grade filters? If so, what should manufacturers keep in mind in their ongoing lifecycle risk management efforts to assure microbial control ?

See hyperlink: http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm124782.htm#14

5


Target Organism: Leptospira (2)

Q&A on cGMP - Production and Process Controls (cont.)

• “Compendial microbiological test methods […] are not capable of detecting this type of bacteria.”

• “[…] manufacturer should use a sound risk management and be aware of unusual microbiota […].”

• “[…] risk mitigation procedures and practices for this microorganism should include at least the following:1. Use of molecular or nonconventional microbial monitoring methods at appropriate intervals

Examples include - use of specialized media such as EMJH- use of validated PCR methods [...] special stain techniques or other

means to identify the presence of Leptospira.

2. Use of conventional approaches. Firms should continue to properly employ basic, standard microbiological practices […] the laboratory should ensure that microscopic examination is part of its routine cell culture process control program […].”

6


Target Organism: Mollicutes

Problems Associated with Mycoplasma Contaminations:

• Altered levels of protein, RNA, and DNA synthesis

• Alteration of cellular metabolism and proliferation characteristics (growth, viability)

• Induction of chromosomal aberrations (numerical and structural alterations)

• Alteration of cellular morphology, such as change in cell membrane composition (surface antigen and receptor expression)

• Induction (or inhibition) of lymphocyte activation

• Induction (or suppression) of cytokine expression

• Interference with various biochemical and biological assays

• Influence on signal transduction

• Promotion of cellular transformation

• Total culture degeneration and loss

7

MMV as an Industry Threat:

MMV …

• … contamination has been seen to come from animal and non-animal-derived components.

• … is an environmental contaminant.

• … is difficult to inactivate.

• … contamination is continues to be reported as a contaminant.

• … may not be detected in the current in vitro general virus assay nor MMV cell based detection method.


Target Organism: MMV

8


Target Organisms: Regulatory Requirements

Tests for Adventitious Agents are required in various regulatory documents:

• Ph. Eur., see– various General Monographs,

• Monoclonal Antibodies for Human Use• Vaccines for Human Use

– Biological Tests• Ph. Eur. 2.6.16: Tests for Extraneous Agents in Viral Vaccines for Human Use

• ICH-Documents– Q5A – Q5E-Series: Quality of Biotechnological Products – Q7: Good Manufacturing Practice of Guide for Active Pharmaceutical Ingredients

• CFR– 21 CFR Part 610: General Biological Products Standard– 21 CFR Part 864, Subpart C: Cell and Tissue Culture Products– 9 CFR Part 147, Subpart A: Blood Testing Procedures– 9 CFR 113.300: General Requirements for live virus vaccines

• Points to Consider (FDA)– Points to Consider in the Characterization of Cell Lines Used to Produce Biologicals

• EMEA/CHMP/BWP/398498/2005: Guideline on Virus Safety Evaluation of Biotechnological IMP’s

… and many, many more !!! 9


Target Organism: Detection Methods

The current detection methods are traditional methods, i.e.

• Growth-based detection methods

Leptospira detection method

Mycoplasma culture method

• Indicator cell-based detection methods using indicator cell cultures

Mycoplasma indicator cell culture method

MMV 324K-assay

10

Leptospira:

The sample to be tested is inoculated into liquid EMJH-Medium, incubated in the dark at 28 – 30°C for up to 28 days.

A Leptospira contamination renders the enrichment medium turbid.

see: Ellinghausen, Jr., H.C., and W.G. McCullough. 1965. Am. J. Vet. Research. 26:45-51. & Johnson, R., and V.G. Harris, 1967. J. Bacteriol. 94:27-31.

Mollicutes – Culture Method:

The sample to be tested is inoculated into liquid Hayflick media, incubated under aerobic and

microaerophilic conditions at 35 – 38° C for 21 days. At specified intervals subcultures on solid media are prepared using the same incubation conditions.

The agar cultures are observed after 7 – 14 days of incubation for so-called “fried-egg-colonies”.


Growth-Based Detection Methods

11

Mollicutes – Indicator Cell Culture Method:

The sample to be tested is co-incubated with a suitable indicator cell culture and Mollicutes present in the sample are enriched. After three days a subculture on cover slips is prepared. This subculture is fixed after 3 - 5 days and stained with a suitable DNA stain.

Mycoplasmas produce extranuclear fluorescence (pinpoints or filaments over the indicator cell cytoplasm and the intracellular space)

MMV – 324K-Assay:

The sample to be tested is co-incubated with a human kidney-cell line (324K) for 21 days at 35 – 37°C.

At specified intervals the indicator cell culture is observed for normal appearance, cell distribution and confluence indicating cytopathic effects - CPE. Cultures that shows CPE’s are recorded as positive (see lower picture).


Cell-Based Detection Methods

12


Problem Statement: Issues, Pains & OpportunitiesTraditional Assays - Problem Statement:

13

Issue / Pain Missed Opportunities

• Some cytotoxic New Molecular Entities interfere with in vitro cell culture

assays

• Low sensitivity (due to dilution needs – see bullet above)

• Limited Specificity

• Different assay-platforms

• High effort due to manual handling

• Complex sample workflow

• Specialized equipment and dedicated laboratories

• Lengthy testing times period

• Result interpretation require high level of expertise

• Safety, Health and Environment issues, e.g. “REACH-Verordnung” in EU

(EG 1907/2006)

• Non „state of the art”-assays

• One common platform

• Simplification

• Agility, flexibility

• Cost saving

• Sample processing automation

• High throughput testing

• Early detection of contaminations

Background

Vision

14

Current Situation: Detection of adventitious agents are on different detection platforms!

• Rodent in-vitro virus detection: Cell based (324K), cannot detect all current industry threats of

rodent virus, prone to false positive.

• Legacy Genentech PCR rodent virus detection (hold-step): Not up to current Quality business

efficiency standards, uses toxic chemical for operator.

• Mycoplasma detection via Culture and Indicator Cell Culture Method: Insensitive and one

month duration.

• Legacy Roche Mycoplasma PCR Detection: All manual intensive operator work, uses outdated

readout technology.

• Leptospira-PCR: “state-of the art”, fully automated sample preparation (Roche MagNAPure®) and electronic readout (Roche LightCycler® 480 real-time PCR System).


Problem Statement: Misaligned Assays

15

Sampling

• Pre-harvest cell culture fluid (PHCCF)

NA-Purification

• Automated DNA purification using MagNA Pure® 96 System

• Volume: 1 mL of PHCCF (cells and medium, no pretreatment such as centrifugation or clarification)

Amplification and Detection

• LightCycler® 480 real-time PCR System

• Qualitative real-time PCR


Vision

Outlook:All adventitious agents tests on the same “state-of the art”, fully automated platform.

16


Automation Tools

MagNA Pure 96 System:

• Automated isolation of nucleic acids (DNA, total RNA, total viral nucleic acids).

• Different kinds of biological sample material.

LightCycler® 480 II Real-Time PCR System:

• Compact benchtop instrument with interchangeable thermal block cycler for 96- and/or 384-multiwell plates.

• High-throughput Real-Time PCR.

17

18


MagNA Pure 96 Isolation Technology

Elution

in low-salt buffer

WashingCell disruption

lysis buffer, proteinase K

Binding

to MGP

SeparationSample

18


LightCycler® 480 System – General Overview

LightCycler® 480 Instrument II:

• Xenon or LED

• Detection formats:

Intercalating green dyes (e.g., SYBR Green I)

Hydrolysis/Universal ProbeLibrary (UPL) probes

HybProbe Probes

• Exchangeable Block Cycler Unit:

96 well

384 well

picture placeholder

19


Hydrolysis probes - TaqMan® Probes

Detection Principle:

• sequence-specific probe. Probes consist of a fluorophore (5’-end of the oligonucleotide probe) and a quencher (3’-end of the oligonucleotide probe)

• quencher molecule quenches the fluorescence emitted by the fluorophore via FRET (fluorescence resonance energy transfer) as long as both are in proximity.

• the probe that has annealed to the template is hydrolyzed by 5’ exonuclease activity of Taq Polymerase, separating reporter dye from quencher. This degradation of the probe releases the fluorophore from it and breaks the close proximity to the quencher, thus relieving the quenching effect and allowing fluorescence of the fluorophore.

• fluorescence detected real-time in the PCR thermal cycler is directly proportional to the fluorophore released and the amount of DNA template present in the PCR

Measurement once per cycle

20

21

Test Design


Probe-based detection format for high specificity

• Target PCR with 4 replicates enhance sensitivity.

– Mycoplasma and rodent parvovirus: 4x 20 μL of the eluate is PCR tested

– Leptospira: 4x 13 μL of the eluate is PCR tested

• Internal Control (IC) controls for DNA recovery and detection of PCR inhibition spiked into sample:

– Mycoplasma and Leptospira: internal control plasmid at calibrated concentration

– Rodent parvorirus: viable M13K07 phage at calibrated concentration

• Target PCR and Internal Control PCR are performed

– Mycoplasma and Leptospira: in separate PCR reactions

– Rodent parvovirus: as duplex PCR reaction

• Positive Control (PC) PCR

– Positive control plasmid at low-copy numbers must be amplified (note: it controls the activity of the enzyme and the proper function of all PCR components)

• Negative Control (NC) controls PCR

– Buffer without template DNA must be negative (note: it controls PCR reagents for DNA contaminations)

21


Validation Requirements acc. to Ph. Eur. (1)

22

Mycoplasma NAT Validation - Requirements per E. P. 2.6.7:(see subchapter Guideline for Mycoplasma NAT Validation)

• 3 Parameters should be evaluated:

Specificity

Sensitivity / Detection Limit

Robustness / Precision

• In addition, a comparison of the alternative method and official methods must be performed, if NAT is intended to replace the compendial methods:

Comparison of the respective detection limits. Specificity (mycoplasma panel detected, putative false positive results) should also be considered.

Acceptance criteria for the detection limit:o replacement of the culture method: the NAT system must be shown to detect 10 CFU/mL for each mycoplasma

test species described in Ph. Eur. 2.6.7, paragraph 2-2; o replacement of the indicator cell culture method: the NAT system must be shown to detect 100 CFU/mL for

each mycoplasma test species described in in Ph. Eur. 2.6.7 paragraph 2-2.


Validation Requirements acc. to Ph. Eur. (2)

23

NAT Validation - Requirements per E. P. 2.6.21: (see subchapter Guideline for Validation of nucleic acid amplification techniques (NAT) for the detection of hepatitis C virus(HCV) RNA in plasma pools)

• 3 Parameters should be evaluated:

Specificity

Sensitivity / Detection Limit

Robustness

• Ph. Eur. 2.6.21 provides information on QUALITY ASSURANCE, e.g. technical and operator qualification.

• RECOMMENDATION: In cases where the NAT-based detection method intends to replace an existing assay – e.g. rodent parvovirus PCR as an replacement of the indicator cell line MMV-detection assay (324K-assay) - it is highly recommended to perform a comparison of the two methods. Parameter to be compared at least sensitivity and specificity.


Validation Status

24

Target: Leptospira(1) Mollicutes(2) Rodent Parvovirus(1)

Activ

ity

Generic Method Validation:

Sensitivity < 100 cells/mL < 10 CFU/mL < 0.5 TCID50/mL

Specificity passed slight gap passed

Robustness passed passed passed

Compara-bility:Sensitvity n. a. comparable PCR is superior

Specificity n. a. PCR is superior unknown for 324K-assay

Product-specific Method Validationpassed

(47 products)passed

(15 products)passed

(9 products)

Method Validation: Performed in acc. with

(1) Ph. Eur. 2.6.21 „Nucleic Acid Amplification Techniques“

(2) Ph. Eur. 2.6.7 „Mycoplasmas“

25

Thank you for your attention !

?? ??

?

?

? ??Questions ?

?


Q & A

LIGHTCYCLER, MAGNA PURE, MYCOTOOL and VIROTOOL are trademarks of Roche. 26

Doing now what patients need next

Thierry BONNEVAY, Sanofi Pasteur FrancePh. Microbiology Expert (Group 1)

Cré

dits

photo

s :

©

Development, Validation and Implementation of an alternative NAT

Mycoplasma detection method in replacement of the compendial

methods in viral vaccine for human use

11 OCTOBER 2017 STRASBOURG

A. laidlawii PG-8A, CBiB 2010

Agenda

• Introduction• Mycoplasma• Regulatory Environment

• Development of a NAT alternative method

• Validation of the NAT alternative Method

• Implementation of the final NAT method as release test • Case study for a new vaccine through CTD submission• Case study for a commercialized vaccine through a variation dossier

• Conclusion and Perspective

10 - 11 October 2017

T. Bonnevay - International Microbiology Symposium EDQM 2

INTRODUCTION

INTRODUCTION - Mycoplasma

Trivial name for the Mollicutes class

including (but not limited to) Mycoplasma, Acholeplasma, Ureaplasma, Spiroplasma,

Anaeroplasma, and Asteroleplasma

4 Orders, 5 Families, 9 Genera (more than 200 Mollicutes sp.)

Smallest bacteria (0.15-0.30µm) and lack of cell wall

Fried egg appearance on solid media

Frequently found as contaminants

of cell substrates and other materials used for the manufacture of

cell-derived biological products

May be introduced into the process as contamination by an operator : M. fermentans, M. orale, M. salivarium, M. hominis…

and/or contamination of raw materials : Bovine sera (M. arginini) ;

Porcine trypsin (M. hyorhinis) ; Materials of plant or insect origin

(Spiroplasma sp.)

10 - 11 October 2017


Medical Microbiology Chapter 37, Mycoplasmas 4ème édition. Galveston (TX), 1996

Can pass through filters that exclude bacteria (0.22µm)

Changes in cell metabolism, membrane antigenicity, protein expression and chromosomal aberrations have been observed

Fastidious to grow

Contaminated cell cultures usually do not show overt signs of microbial contamination

Why the Mycoplasma detection is challenging?

INTRODUCTION - Mycoplasma

10 - 11 October 2017 T. Bonnevay - International Microbiology Symposium EDQM 5

Rosengarden et al, 2014


INTRODUCTION – Regulatory Environment● Regulatory requirements to ensure the quality and safety of biologics

•

10 - 11 October 2017


European Pharmacopoeia (Ph.

Eur.), 9th Edition, Chapter 2.6.7.,

Mycoplasmas (01/2008)Appear 70’s for

Microbiological cultureCompleted in 2000’s

with Indicator cellmethod

And with NAT method in 2007

FDA/CBER, (PTC) Points to consider in the characterization of cell lines used to produce biologicals, Attachment 2 (1993)

FDA/CBER, Guidance for industry: Characterization and qualification of cell substrates and other biological materials used in the production of viral vaccines for infectious disease indications (2010)

United States Pharmacopoeia (USP), Chapter <63>, Mycoplasma Tests (2010)

FDA, CFR Title 21, Part 610 – General Biological Products Standards, Subpart D -Mycoplasma, Section 610.30, Test for Mycoplasma (1998) – No evolution since 1970

Japanese Pharmacopoeia (JP)

XV, Section 14. Mycoplasma Testing for Cell Substrates used for

the Production of Biotechnological/

Biological Products (2011)

INTRODUCTION – Ph. Eur. § 2.6.7 Mycoplasma

• European Pharmacopoeia – 1st edition 1977• Test for Mycoplasma gallisepticum and Mycoplasma synoviae: test in vitro (solid and liquid

media) and test in vivo (inoculate 30 chickens)

• European Pharmacopoeia – 3rd edition 1996• § 2.6.7. Test used for avian vaccines (M. gallisepticum and M. synoviae) and test for non-avian

vaccines Mycoplasmas and Ureaplasmas (M. hyopneumoniae and U. urealyticum): liquid and solid media

• European Pharmacopoeia – 4TH edition – 2002• 6 strains – culture method and indicator cell culture method

• EDQM Symposium 5-6 May 2003 in Copenhagen : MICROBIOLOGICAL CONTROL METHODS IN THE EUROPEAN PHARMACOPOEIA: PRESENT AND FUTURE• EDQM establish a Working Group to introduce NAT methods in Chapter 2.6.7 : Peter Castle led

this group with first meeting 30 Sept 2004 – Now Mycoplasma is a Dormant Working Party • Supplement 5.8 to the 5th Edition, published 12 Dec 2006 (implementation date

07/2007): Nucleic acid amplification technique: a new section has been introduced followed by recommendations on the validation of such tests

10 - 11 October 2017


1977

2006

INTRODUCTION – Compendial Testing

• Compendial Culture Method comprises three steps in accordance with Ph. Eur. 2.6.7.● Direct inoculation of solid agar media with the test product.

● Preliminary enrichment: a liquid or semi-solid medium is inoculated with the test product to promote the replication of mycoplasmas.

● Subculture: the culture obtained by enrichment is inoculated over the surface of a solid medium to obtain identifiable colonies

10 - 11 October 2017


INTRODUCTION – Compendial Testing

• Compendial Indicator Cell Culture Method (Epifluorescence) in accordance with Ph. Eur. 2.6.7.● Developped to allow detection of “non cultivable” strain on culture media like M.

hyorhinis α● Direct Inoculation of 1 mL sample on culture cells

● Cell culture coloration with fluorescent dye like Hoëchst that linked to DNA

10 - 11 October 2017

9

International Microbiology Symposium EDQM


INTRODUCTION – Compendial NAT Mycoplasma

• Since 2007, introduction of NAT method to detect Mycoplasma in regulatory documents● Changes introduced in international compendia pharmacopeia or guidelines indicating that

NAT-based methods could be substituted for either compendia method, after undergoing suitable validation

● 2007 – Ph. Eur. § 2.6.7 Mycoplasma with detailed guidelines for specific mycoplasma NAT approaches, validation and guidelines for the validation (specificity, LOD, robustness, comparability study) and cross reference with § 2.6.21

● 2010 – USP <63> Mycoplasma: no detail or validation but in the introduction “a validated NAT based method may be used to detect Mycoplasma, provided such a method is shown to be comparable to both methods A and B. Alternative methods must be suitably validated.”

● 2011 - WHO revised TRS on cell substrates (adopted Oct 2010, published July 2011) WHO TRS on Yellow fever and Dengue and cross refer to WHO revised TRS on cell substrates with introduction of NAT as an alternative method and WHO TRS on cell substrates cross refers to Ph. Eur. § 2.6.7

● 2011 – JP 14. Mycoplasma testing for cell substrate. Method C . Polymerase Chain reaction Detection Method : detail and Example Procedure, description of Nested PCR and positive and negative control

10 - 11 October 2017


INTRODUCTION – Compendial NAT Mycoplasma

• Since 2008, introduction of NAT methods to detect Mycoplasma in Scientific documents / articles

● 2010 – PDA Technical Report N° 50 – Alternative Methods for Mycoplasma Testing – 41 pages – very informative and complete

● 2010 – Y. Zhi, A. Mayhew, N. Seng, G.B. Takle, Validation of a PCR method for the detection of mycoplasmas according to European pharmacopoeia section 2.6.7,Biologicals 38 (2) (2010) 232–237.

● 2010 - S.M. Deutschmann, H. Kavermann, Y. Knack, Validation of a NAT-based mycoplasma assay according European pharmacopoiea, Biologicals 38 (2) (2010) 238–248

● 2017 – E. Abachin, M. Marius, S. Falque, J. Arnaud, V. Detrez, S. Imbert, L. Mallet, T. Bonnevay – Validation of a PCR coupled to a microarray method for detection of mycoplasma in vaccines – Biologicals – 23 sept 2017 (Article in press)

10 - 11 October 2017


To summarize, since 2010 many regulatory and scientific documents supporting the implementation of NAT for the detection of Mycoplasma

Development of the NAT Method

Drivers for rapid mycoplasma method implementation

• Overcome disadvantages of compendial methods

10 - 11 October 2017


• Our goals• Improve and secure quality and

safety of Sanofi Pasteur’s vaccines

• Improve Time To Market (TTM) and secure supply

Develop a robust and cGMP NAT method for all Mycoplasma detection

Long testingtime

Cumbersome

No identification

Samplehandling

Narrow detection

range

Assessment of new technologies

• Since 2001: evaluation of alternative Nucleic Acid Amplification Technique based methods for mycoplasma testing • In a QC molecular biology laboratory

under GMP environment

• Since 2008: use of the CytoInspect™ method (supplier Greiner Bio-One) based on PCR/microarray to overcome specificity issues encountered

10 - 11 October 2017


DEVELOPMENT(2001- 2008)

Technology selectedCytoInspect™ - The PCR + microarray technique

• Workflow: QIAamp extraction kit + CytoInspect™ kit

• 10 mL test sample volume, as compendial culture method

• 39 Mollicutes species-specific probes and 1 universal probe that tests for the presence of mycoplasma sp.

10 - 11 October 2017



Technology selectedCytoInspect™ - the PCR + microarray technique

• Main advantages:

10 - 11 October 2017



Validation of the NAT Method

VALIDATION STRATEGY

First step

• Full validation using a generic matrix (applicable for all products)• Specificity • Limit of detection (LOD) determined for 10 Mycoplasma species according to

European Pharmacopeia § 2.6.7.• Data on method robustness available• Comparison of performance with the two compendial culture-based methods

Second step

• Specific matrix verification (one per product/stage)• Specificity: ability to detect M. orale spiked at 10 Colony Forming Units

(CFU)/mL in matrix and to not detect mycoplasma in unspiked matrix• Absence of interference and/or inhibition: ability to detect 10 CFU/mL maximum

using at least 3 mycoplasma species based on the manufacturing process and percent of frequency of occurrence in pharmaceutical industry (based on recent literature)

10 - 11 October 2017


Stock preparation and characterization (1/2)

Stocks

• from American Type Culture Collection (ATCC) or European Directorate for the Quality of Medicines (EDQM), low passage strains as frozen high titer preparations and grown in liquid Mycoplasma medium • Harvested in exponential

phase, aliquoted , titrated• Characterized

• CFU/mL, ng/mL by PicoGreen converted to Genome Copy (GC)/mL,

• purity and identity by CytoInspect and

• GC/CFU ratio)• Stored at ≤-60°C

Titration

• before and after freezing. Stability over time performed – up to 5 years• Acceptance criterion:

0.5log10CFU/mL

GC/CFU ratio

• determined using the genome length of each individual strains• Acceptance criterion:

<100, but preferably <10

10 - 11 October 2017


Sanofi Pasteur actively participated to two collaborative studies organized by FDA and WHO

- FDA - Collaborative study report: evaluation of the ATCC experimental mycoplasma reference strains panel prepared for comparison of NAT-based and conventional mycoplasma detection methods. Dabrazhynetskaya A et al. (2013)

- WHO - World Health Organization International Standard To Harmonize Assays for Detection of Mycoplasma DNA. Nübling CM et al. (2015)

Stock preparation and characterization (2/2)

• Summary of characterization data for all mycoplasma stocks used during validation and matrix verification

10 - 11 October 2017


Mycoplasma strains Original source

Titre (CFU/mL) DNA quantification (ng/mL)

GC/CFU ratio

M. hyorhinis (cultivable) EDQM – BRP 1.15E+08 3.29E+08 3

M. hyorhinis α (non cultivable) ATCC 29052 < 5000 EqCFU/mLA N/Ap N/Ap

M. orale EDQM – BRP 2.40E+08 9.73E+08 6

M. arginini ATCC 23838 1.20E+09 1.28E+09 2

A. laidlawii EDQM – BRP 2.65E+07 6.70E+08 17

M. fermentans EDQM – BRP 1.13E+08 1.21E+09 11

M. gallisepticum ATCC 19610 6.14E+08 1.82E+08 3

M. pneumoniae ATCC 15531 7.87E+06 5.33E+08 83

M. synoviae EDQM – BRP 1.78E+06 3.57E+07 25

S. citri ATCC 27556 6.05E+06B 9.31E+07 8

A EqCFU/mL: equivalent CFU/mLB Determined by Mycoplasma Biosafety Services

Generic validation using a generic matrix (1/2)

10 - 11 October 2017


Specificity designSpikes with 10 CFU/mL of M. orale to assess internal cross-reactivity

Spikes with phylogenetically-related bacteria (Streptococcus agalactiae, Lactobacillus rhamnosus and Clostridium sporogenes) to assess cross-reactivity

Unspiked matrices

Specificity Results: the method is specific to mycoplasma

Generic validation using a generic matrix (2/2)

10 - 11 October 2017


LOD design

Minimum number of CFU per volume of sample that can be detected in 95% of test runs

Positive cut-off determined by 2 analysts, in 6 independent dilution series with 5 test replicates for each dilution giving a total number of 30 test results

LOD result: LOD < 10CFU/mL for all Mycoplasma

* EqCFU: Equivalent CFU/mL

Specific matrix verification strategy

• Testing on 3 different lots per stages (if available)

• Evaluation of specificity by spiking 10 CFU/mL of M. orale in product matrix • 1 replicate per lot

• Evaluation of matrix interference by spiking 10 CFU/mL of at least 3 mycoplasma species based on the manufacturing process and % of frequency of occurrence in pharmaceutical industry (based on recent literature)

•5 replicates per lot•Implementation of a risk analysis protocol/tool to properly select the appropriate species

• Stages to be tested• Depend on the product and the Pharmacopeia product specific monograph but mainly Crude

harvest of the production lot and seed lot

10 - 11 October 2017


Validation strategy: Comparability study

• Comparison of specificity and LOD results between the 3 methods• Alternative method at least as sensitive as the

compendial methods

Comparison of the performance of the NAT method using previously

obtained data from compendial method

validation

• Comparison of unspiked lots by the 3 methods• No mycoplasma detected by the 3 methods

• Comparison of 3 spiked lots tested by the 3 methods• Mycoplasma detected by the 3 methods in the

matrices

Comparison of the performance of the NAT method using previously

obtained data from compendial method validation for matrix

verification

10 - 11 October 2017


Implementation of the NAT MethodAs release testing Case study with a New

Vaccine Through CTD file

New Vaccine in the CTD

• The new vaccine CTD file has been submitted to over 30 countries to date and including stringent regulatory authorities

• All the validation data of PCR+Microarray with the generic matrix as well as the specific matrices at two stages are developed in the Section “3.2.S.2.4 Controls of Critical Steps and Intermediates” in the CTD

10 - 11 October 2017



•Specifications and testing strategy detailed in the CTD: •The PCR + microarrays method is the release method for Mycoplasmas testing. However, compendial methods for Mycoplasmas testing can be used as alternative methods in case of failure of the PCR + microarrays equipment and/or in case of major impossibility to obtain material or reagents to perform the test

•So the compendial methods are validated at the two stages, and are back-up methods as explained above

10 - 11 October 2017



• Of the total questions received on the analytical component (in total 103 - in mid 2017), only one about mycoplasma NAT test• One question from Brazil regulatory Health Authority:

“Send a technical justification for the replacement of the tests for Mycoplasma by microbiological culture and epifluorescence methods, at the stage of control cells and crude harvest, by PCR + microarrays method. Send the validation of this new method for Mycoplasma”

The answer was based on the advantages of the NAT method, as well as the limitations of compendial methods and all the data for the generic and specific validation were already present in the CTD

• Since 2015, the two stages of this new commercialized vaccine :

• control cells and crude harvest are tested in routine with the CytoInspect method only:

10 - 11 October 2017


All results negative

Implementation of the NAT MethodAs release testing Case study with a

commercialized vaccine through variation

Commercialized virus vaccine: Validation strategy

• Comparison of the performance of the NAT method using previously obtained data from compendial method validation • Comparison of specificity and LOD results between the 3 methods

• Alternative method at least as sensitive as the compendial methods

• Comparison of the performance of the NAT method using previously obtained data from compendial method validation for matrix verification • Comparison of 3 unspiked lots tested by the methods:

- No mycoplasma detected by the 3 methods

• Comparison of spiked lots by the 3 methods

- Mycoplasma detected by the 3 methods in matrices

10 - 11 October 2017


Variation: Regulatory Affairs

• Objective: Replacement of the compendial methods by the CytoInspect™ method. PCR and microarray on single harvest of the WSL and single harvest (intermediate) before the Drug Substance of a legacy product.

• Strategy: Compendial methods kept as alternative methods in case of failure of the PCR and microarray equipment and/or in case of major impossibility to obtain material or reagents to perform the test

• Countries: Europe, Canada and International area10 - 11 October

2017T. Bonnevay - International Microbiology Symposium EDQM 31

Variation: Regulatory Affairs

• Evaluation of the data: Define R&R and deliverable to write the dossier. Define prerequisites for evaluation (check list: products, stage… changes)

10 - 11 October 2017


Variation dossier: preparation and submission

10 - 11 October 2017


13pagesDossier sent on July 2016 to UK

Variation dossier: preparation and submission

10 - 11 October 2017


Questions received on 12 Sep 2016

Response to HA Questions

10 - 11 October 2017


•M. gallisepticum: absence of avian material during production and vaccine not intended for use in poultry

•M. pneumoniae: extremely low probability to be encountered as contaminants of cell cultures as per recent publications of Uphoff and Drexler (2002), Timenetsky et al. (2006) and Nikfarjam and Farzaneh(2012)

Assays performed to confirm the detection of M. arginini spiked at 10CFU/mL in vaccine matrices

Cell

Cell


10 - 11 October 2017


Submission of the comparability report showing absence of impact from matrices

on mycoplasma detection using compendial and PCR + Microarray

methods

Performance of the NAT-based method demonstrated to be at least equivalent to

both compendial methods

Submission of the stability report providing data showing calibration,

quantification and stability for preparations as well as number of

passages (<15)


10 - 11 October 2017


NAT performed routinely as official release test and compendial methods kept as alternative methods Does not correspond to a replacement of method

Therefore, as there is no specific variation category for alternative method in the Variation Guideline, modification of the variation category to: - Type II, variation, B.I.b.2. z) Addition of a test procedure, registered methods being kept as

alternative methods

Application form updated accordingly

Conclusion and summary

10 - 11 October 2017


ConclusionAnd Perspectives

Conclusion

• Compendial Mycoplasma detection is a long and huge QC process

• Since 2007, regulatory documents, pharmacopeia and scientific articles give guideline and authorize / encourage to perform “state of the art” method to detect Mycoplasma by NAT method

• The implementation of the NAT method requires consistent validation work and QC design laboratories for molecular biology (as described in § 2.6.21. of Ph. Eur.)

• One of the Key factors is the production and characterization of Mycoplasma stocks especially the GC/CFU content

• Implementation in CTD file and especially variation dossier and submission for legacy products is a long process

• But this is worth it!

• Increased safety of biological products and Reduced TTM

10 - 11 October 2017


Perspectives

• Implementation and expansion of the strategy for New Vaccines and all viral other commercialized vaccines

• Implementation of the NAT mycoplasma method for US market

• Discussion already engaged with CBER to accept our current strategy of validation and implementation

• Lessons learned

• Variation for this kind of change should be Type II, variation, B.I.b.2. z) Addition of a test procedure

• Comparison data between compendial and NAT methods should be directly submitted as well as stability data of mycoplasma references

• Reflexion about other specified bacteria with NAT detection

• Mycobacteria

• Others

10 - 11 October 2017


THANK YOUSANOFI PASTEUR : MARINE MARIUS, ERIC ABACHIN, NATHALIE FAURE-CHANEL, ERIC SARCEY, VALERIE DETREZ, JULIEN ARNAUD, STEPHANIE

FALQUE, BENEDICTE MOUTERDE, SANDY IMBERT, LAURENT MALLET*

QUESTIONS ?

RAPID DETECTION OF MICRO-ORGANISMS BY DIRECT DETECTION: A SOLUTION FOR TIME CRITICAL IN-PROCESS MONITORING

Marja Claassen-Willemse

October 2017

Content

1. Rapid Micro Methods (RMM) deployment at MSD

2. Case study: a business need for rapid results

3. Comparison studies of RMM and compendial: spiked and unspikedsamples

4. Conclusions & next steps

5. Acknowledgments

2

Initial focus of rapid method deployment

Focus of last 10 years:

• Growth based technologies

Easy adoption

• In process Tests

Limited regulatory risks compared to finished product testing

• Biotech processing:

Strong business case

• High risk; process supporting growth of contaminants

• High benefit, e.g. to prevent downstream contamination (e.g. costs of columns can be ~$200K

3

Examples of rapid method deployment in MSD

• BacT/Alert

– Principle is CO2 detection of growing microorganisms

– Absence of contaminants of samples of bioculture of biopharmaceuticals

• Milliflex Quantum

– Principle is conversion of substrate into fluorochrome by living cells

– Rapid bioburden testing of in process samples of Ab purification

• Rapid Hybrid Mycoplasma test

– Principle is PCR detection after growth step

– Absence of mycoplasma in Cell Harvest of bioculture

4

Enrichment step (7 days) DNA isolation (45 min) qPCR (3 hours)

RMM Vision and Strategy at MSD

In-line/existing productsMigration from Conventional way of testing to Rapid Micro Methods New products/new processes Develop a guidance document & deployment strategies for early integration

Develop Business Case Deliver proposal to the Sites/Business UnitsImplement in coordination with the Sites/Business Units

Engage Merck Research Labs, Process Development groups, External ManufacturingSelect Technology depending on product type and strategy for adoption

Transfer RMMs to Merck internal & external sites/QC labs5

Many internal hurdles Build in business process

Development of non-growth based technologies

• BioTrak

– Real time detection of living cells in air by autofluorescence

– Application in isolators

BPOG Workstream of Fill Finish: Alternative Rapid Micro Methods

• MuScan

– Direct detection of living microorganisms by enzymetic activity

– Rapid bioburden testing of column samples before loading product

In collaboration with supplier

6

Increase microbiological control by direct detection

Consequences of non-growth based methods

• Pioneering

• Moving away from the golden standard:

Colony Forming Unit or “CFU”

becomes

Fluorescence Forming Unit or “FFU”

• FFU ≠ CFU

– Validation challenge

• Statistical models required to analyze

– Correlation with compendial method

– False positive & false negative rate

– Calculate new meaningful limits

7

What are false positives, do we know?

8

False positives

VBNC

Stressed

CFU Viable and culturable: will grow out to colonies

Damaged microorganism: need time to switch to growth

Viable but NOT culturable (VBNC): will never grow on TSA/R2A

Fluorescent artefacts: will be mainly removed by software and manual check

• Ratio’s are unknown and will probably differ direct test method and sample type• Nobody knows the relevance of VBNC

Unkn

ow

n

non-

CF

U c

ounts

FFU are composed of CFU counts and non-CFU counts:

Case study

9

MuScan:– Filtration using a micro sieve followed by staining of living organisms using enzymatic

conversion– Automatically detection of FFU’s within ~1 hour

Filtration Staining Read Out

Biotechnology production process

10

Downstream purification process is a low bioburden process

In-process bioburden testing with MuScan

11

• Problem: Expensive purification columns in purificaton of Biologics are currently sanitized using aggressive agents to prevent microbial growth. These agents shorten the column life span ($).

• Solution: Use a milder sanitization program with 0.1 M Sodium Hydroxide and use the heavy sanitization based on an alert/action limit strategy

Theoretical graphs of CFU and FFU

12

Stringent sanitization using 0.5 M NaOH

Time

FFU’s higher than CFU’s but following same pattern

FFU’s not following the same pattern. VBNC’s may predict growth

CFU’s growing under light regime

New limit

Traditional limit

Counts

Light sanitization

Proof of Concept studies

• Comparison experiments MuScan vs Compendial

1. Spiking studies with several ATCC strains relevant for TAMC

2. Testing samples with bioburden: tapwater

PoC spiking study

• 12 days

• 2 methods: MuScan, compendial TAMC method

• blank and 6 spike concentrations (0, 2, 4, 8, 16, 32, 64 CFU/100 mL)

• 3 samples per day, method, and concentration

• 1 technician, buffer (PBS), sample volume 100 mL

Executed for 3 MO (Bioballs): E. coli, P. aeruginosa and S. aureus

Statistical model of spiking study

• Observed count ~ μ• Mean μ depends on method parametersμ ∙ spike η

– Detection proportion – probability to detect an organism

– False positive count η– ~ 0, σ represents day-day variability additional to Poisson

variability

• May or may not depend on the method

E. coli results______

MuScanTSA

ηMuScan 0.90 0.25TSA 0.85 0

S. aureus results______

MuScanTSA

ηMuScan 0.85 0.22TSA 0.85 0

P. aeruginosa results______

MuScanTSA

ηMuScan 0.89 0.26TSA 0.84 0

MuScan distribution results: E. coli

Spike = 0

Spike = 2 CFU/sample




Spike =32 CFU/sample


Similar plot for compendial with the diffence that no “false positives” were obtained

Calculation of limits using the model

Number of samples n = 1

Number of samples n = 5

• Overlapping of FFU distribution curves • Higher discriminative effect by increasing sample size

n=1

n=2

n=3n=4

n=5n=10

95%

5%

0.9 1.5

Bioburden CFU/100 mL5.03.31.9 2.1

Power to distinguish bioburden from 0

n alert limit(FFU/n samples)

1 12 13 2/34 3/45 3/510 5/10

Conclusions PoC spiking study

• Results were similar for the 3 micro-organisms

• MuScan had similar or slightly higher counts than compendial

• MuScan had on average 0.2 to 0.3 false positives, compendial has no false positives

• Day-day variability was reasonably similar and very small for both methods

• Limits can be calculated and depend on model, but also on sample size, sample volume

PoC study: testing samples with bioburden

23

• 6 days

• 2 methods: MuScan, compendial TAMC method (TSA and R2A)

• 3 dilutions of tap water (undiluted, 1:10, 1:100)

• 3 consecutive samples of a single tap point per day, method, and concentration

• Sample volume 10 mL

Results of testing tapwater dilutions

24

Observations:• FFU count a factor ~50 times higher compared to CFU’s on TSA• FFU count a factor ~14-23 times higher compared to CFU’s on R2A• Non-CFU counts will be investigated

Mean and SD of 6 days (n=18)

Next steps

• Pre-validation study using the actual sample matrix (buffer)

• Validation using the Ph.Eur 5.1.6 and USP <1223>

– Phase 1: testing equivalency with compendial regarding validation parameters:

• Specificity, Precision, Accuracy, Linearity, Limit of Quantification

– Phase 2: testing comparability with compendial method using real samples for a certain period of time.

• Calculation of limits

25

ACKNOWLEDGEMENTSMuScan team

26

Daan de Gouw

Project Lead

Quirine Hartman

Technician

Geert Verdonk

Director Micro CoE

Pieta IJzerman

Principal Statistician

Edwin vd Heuvel

Professor in Statistics

TU Eindhoven

Ronald van Doorn

Manager R&D Innosieve Diagnostics

Michel Klerks

CEO Innosieve

Innosieve Diagnostics

THANK YOU