Transgenic mice as an alternative to monkeys for neurovirulence testing of live oral poliovirus vaccine: validation by a WHO collaborative study.

Transgenic mice as an alternative to monkeys for neurovirulence testing of live oral poliovirus vaccine: validation by a WHO collaborative study Eugenia Dragunsky,1 Tatsuji Nomura,2 Kazimir Karpinski,3 John Furesz,4 David J. Wood,5 Yuri Pervikov,6 Shinobu Abe,7 Takeshi Kurata,8 Olivier Vanloocke,9 Galina Karganova,10 Rolf Taffs,11 Alan Heath,12 Anna Ivshina,13 & Inessa Levenbook14 Objective Extensive WHO collaborative studies were performed to evaluate the suitability of transgenic mice susceptible to poliovirus (TgPVR mice, strain 21, bred and provided by the Central Institute for Experimental Animals, Japan) as an alternative to monkeys in the neurovirulence test (NVT) of oral poliovirus vaccine (OPV). Methods Nine laboratories participated in the collaborative study on testing neurovirulence of 94 preparations of OPV and vaccine derivatives of all three serotypes in TgPVR21 mice. Findings Statistical analysis of the data demonstrated that the TgPVR21 mouse NVT was of comparable sensitivity and reproducibility to the conventional WHO NVT in simians. A statistical model for acceptance/rejection of OPV lots in the mouse test was developed, validated, and shown to be suitable for all three vaccine types. The assessment of the transgenic mouse NVT is based on clinical evaluation of paralysed mice. Unlike the monkey NVT, histological examination of central nervous system tissue of each mouse offered no advantage over careful and detailed clinical observation. Conclusions Based on data from the collaborative studies the WHO Expert Committee for Biological Standardization approved the mouse NVT as an alternative to the monkey test for all three OPV types and defined a standard implementation process for laboratories that wish to use the test. This represents the first successful introduction of transgenic animals into control of biologicals. Keywords Poliovirus vaccine, Oral/toxicity; Mice, Transgenic/physiology; Macaca mulatta; Nervous system/virology; Virulence; Sensitivity and specificity; Reproducibility of results; World Health Organization; Comparative study; Validation studies (source: MeSH, NLM ). Mots cle´s Vaccin antipoliomye´litique Sabin/toxicite´; Souris transge´niques/physiologie; Macaca mulatta; Syste`me nerveux/virologie; Virulence; Sensibilite´ et spe´cificite´ (Epide´miologie); Reproductibilite´ des re´sultats; Organisation mondiale de la Sante´; Etude comparative; Etude validation (source: MeSH, INSERM). Palabras clave Vacuna antipolio oral/toxicidad; Ratones transge´nicos/fisiologı´a; Macaca mulatta; Sistema nervioso/virologı´a; Virulencia; Sensibilidad y especificidad; Reproducibilidad de resultados; Organizacio´n Mundial de la Salud; Estudio comparativo; Estudios de validacio´n (fuente: DeCS, BIREME ). Bulletin of the World Health Organization 2003;81:251-260. Voir page 259 le re´sume´ en franc¸ais. En la pa´gina 259 figura un resumen en espan˜ol. Introduction The neurovirulence test (NVT) for oral poliovirus vaccine (OPV) is a key test for monitoring the consistency of vaccine production (1), and following WHO guidelines is required for each monovalent bulk lot of OPV produced. The WHONVT (2) is a standardized procedure. If consecutive lots of monovalent bulks consistently meet the specifications of the 1 Biologist, Center for Biologics Evaluation and Research, Food and Drug Administration, 1401 Rockville Pike, Rockville, MD 20852-1448, USA (email: dragunsky@cber.fda.gov). Correspondence should be addressed to this author. 2 Director, Central Institute for Experimental Animals, Miyamae, Kawasaki, Japan. 3 Statistical Consultant, Orleans, Ontario, Canada. 4 Consultant, Ottawa, Ontario, Canada. 5 Previously: Principal Scientist, National Institute for Biological Standards and Control, Potters Bar, Hertfordshire, England. Currently: Scientist, Vaccines and Biologicals, World Health Organization, Geneva, Switzerland. 6 Medical Officer, Vaccines and Biologicals, World Health Organization, Geneva, Switzerland. 7 Assistant Director, Japanese Poliomyelitis Research Institute, Tokyo, Japan. 8 Deputy Director-General, National Institute of Infectious Diseases, Tokyo, Japan. 9 Manager, In Vivo QC, GlaxoSmithKline Biologicals, Rixensart, Belgium. 10 Laboratory Chief, Institute of Poliomyelitis and Viral Encephalitides, Moscow Region, Russian Federation. 11 Microbiologist, Center for Biologics Evaluation and Research, Food and Drug Administration, Rockville, MD, USA. 12 Statistician, National Institute for Biological Standards and Control, Potters Bar, Hertfordshire, England. 13 Research Fellow, Center for Biologics Evaluation and Research, Food and Drug Administration, Rockville, MD, USA. 14 WHO Advisor, Northbrook, IL, USA. Ref. No. 02-0101 251Bulletin of the World Health Organization 2003, 81 (4) WHO test, there is a high level of assurance that the vaccines will be safe when used for human immunizations (3, 4). So far, the test for neurovirulence safety of OPV has been performed using monkeys, because only primates are naturally susceptible to poliovirus. In 1990–91, two laboratories (5, 6) with the support of WHO, established lines of transgenic mice carrying a human receptor to poliovirus. In 1992, WHO recommended that a comparison be made of the sensitivity of TgPVR mice (7) with that of monkeys for type-3 poliovirus strains with different degrees of neurovirulence, and an evaluation of TgPVR mice as a possible alternative to monkeys for the neurovirulence testing of OPV (8). Initial experiments were performed in Japan and the USA that were aimed at selecting the most suitable TgPVRmouse line and route of inoculation, developing basic test methodology, and accumulating initial data. The results obtained with TgPVR21 mice (9, 10) indicated the capacity of the test to discriminate between acceptable batches of OPV and preparations of high neurovirulence. A collaborative study was therefore launched by WHO in 1993 (11) to investigate in more detail the suitability of the method for batch release of bulk OPV. Investigators at the Central Institute for Experimental Animals (CIEA, Japan) succeeded in developing TgPVR21mice from a limited research tool into a reliable supply of standard animals available in large numbers (12, 13). Eleven institutions from Asia, Europe, and the USA participated in the study.a The study started with type-3 OPV, the least stable strain in terms of its neurovirulence, and was completed for all three serotypes inOctober 2000. The results of the collaborative study up to 1999 have recently been published (14). The present paper presents the final results of the collaborative study and validation of the mouse NVT. A statistical model was developed for acceptance or rejection of OPV batches in the mouse test. It has previously been shown that the WHO monkey NVT was a reproducible and sensitive assay, ensuring the safety of OPV (3, 4). Numerous data, obtained in this collaborative study, have proven that the mouse NVT is as reliable as the WHOmonkey NVT for OPV. In 1999 the WHO Expert Committee on Biological Standardization therefore approved the mouse NVT as an alternative to themonkey test for poliovirus type-3 (15) and in 2000 for poliovirus type-1 and type-2 (16). Materials and methods Vaccines The type-1, type-2, and type-3 OPV virus samples used in the study had been tested previously using the monkey NVT according to the WHO requirements for OPV (2) by six manufacturers and three national control authorities. Vaccines of all three types produced in each of the three currently permissible cell substrates (primary monkey kidney, Vero monkey kidney, and human diploid cells) were obtained from nine manufacturers, including six UNICEF suppliers. In all, 75 commercial samples and one experimental sample that passed the monkey NVT were evaluated in mice. In addition, the following vaccine virus samples that failed the monkey NVT were used in this study: nine type-3 commercially produced vaccines and three samples of each serotype that were either experimental vaccines or derivatives of commer- cially produced vaccines additionally passaged in African green monkey kidney (AGMK) or Vero cells at 37–38 oC (a temperature favouring reversion to neurovirulence). The experimental samples of type-1 generated by passage of vaccine lots at elevated temperature were used as surrogates for commercially produced vaccines that consistently failed the monkey NVT, samples of which could not be located despite intensive worldwide searches. This is a limitation of the study design. Experimental samples of type-3 increased the number of preparations that failed the monkey NVT. Mice Two of several mouse lines, TgPVR1 and TgPVR21, derived in Dr A. Nomoto’s laboratory (6, 7), were evaluated in the investigative stage of the study. TgPVR1mice contained more copies of the poliovirus receptor (PVR) and were more sensitive to poliovirus, whereas TgPVR21 mice with a lower PVR copy number were less sensitive. After initial experiments (17, 18), the TgPVR21 mouse line and the intraspinal route of inoculation were selected as the most suitable combination for evaluation of all three poliovirus serotypes. TgPVR21 mice were monitored at the CIEA for freedom from 22 specified pathogens and for generational stability of genetic background and the introduced gene. Maintenance, containment, and transport of mice were conducted in accordance with recommendations of the WHO Memorandum on transgenic mice susceptible to human viruses (19). Each laboratory animal facility that participated was approved by the CIEA before entering the study. Inoculation procedure Sixteen 6-to-7-week-old mice of each gender in each dose group were inoculated with the test vaccine and the same number of animals with the reference vaccine, resulting in 128 mice per test. A technique for intraspinal inoculation of mice described previously (17, 18) was scrupulously optimized, standardized (20), and used in the study. The US Food and Drug Administration (FDA) developed a multi-step system to train investigators in the technique of intraspinal inoculation of mice and in evaluation of clinical signs. All the participants received training at FDA or the Japanese Poliomyelitis Research Institute. The mouse test methodology is fully described and illustrated in a standard operating procedure (SOP) available from WHO.b Statistical methodology The key components of the statistical design and analysis are outlined below. 1. A test vaccine was tested concurrently with the WHO reference vaccine in a randomized experiment. 2. The test vaccine and the concurrently tested reference vaccine were tested at two doses: 3.5 and 4.5 log10TCID50/ 5 ml (5.8 and 6.8 log10 TCID50/ml) for type-3, 1.75 and 2.75 log10TCID50/5 ml (4.05 and 5.05 log10 TCID50/ml) for type-1 and 5.0 and 6.0 log10TCID50/5 ml (7.3 and 8.3 log10 TCID50/ml) for type-2. The need to use different doses for different virus types was determined in the investigative phase of the study (see below). 3. Each dose was inoculated intraspinally into 16 male and 16 female mice. a Names of participating investigators and institutions are given in the Annex 1, part III (see online version at: www.who.int/bulletin). b Available from Dr E. Griffiths, Coordinator, QSB, World Health Organization, 1211 Geneva 27, Switzerland (email: griffithse@who.int). 252 Bulletin of the World Health Organization 2003, 81 (4) Research 4. Mice were randomized to cages, doses, and vaccines. Randomization, which protects against possible inadvertent biases, was also applied to cage location and order of inoculation. 5. Clinical observations of mice and recording of specific neurological signs, such as paresis and paralysis, were performed daily. Paralysis was taken as the primary indicator of degree of neurovirulence and the log odds ratio (LOR) was used as a measure of the neurovirulence of the test vaccine relative to that of the reference vaccine. 6. Estimates and tests of significance were based on logistic regression analysis of the proportions of paralysed mice. 7. Validity criteria that were applied to ensure that each experiment has adequate power to differentiate between Table 1. Summary of results of the WHO collaborative study of TgPVR21 mice with type-3 oral poliovirus vaccine lot 93/636a Laboratory Test Vaccine No. of Proportion paralysed Statistical mice/doseb analysis (P) Dose (log10 TCID50) c 1 2 3 4 5 1.5 2.5 3.5 4.5 5.5 C 1 WHO/IIId 5, 6 0 0.2 0.5 0.4 1 ND ND ND ND ND NA 93/636 5, 6 0 0.333 0.667 1 1 ND ND ND ND ND 0.029e 2 WHO/III 10 ND ND ND ND ND ND 0.1 0.1 0.6 ND NA 93/636 10 ND ND ND ND ND ND 0.1 0.6 0.9 ND 0.004 e 3 WHO/III 15 ND ND ND ND ND ND ND 0.4 0.733 ND NA 93/636 15 ND ND ND ND ND ND ND 0.2 0.867 ND 0.380f 4 WHO/III 15 ND ND ND ND ND ND ND 0.2 0.467 ND NA 93/636 15 ND ND ND ND ND ND ND 0.6 0.867 ND <0.001e 5 WHO/III 15 ND ND ND ND ND ND ND 0.067 0.4 ND NA 93/636 15 ND ND ND ND ND ND ND 0.467 0.933 ND <0.001e 6 WHO/III 15 ND ND ND ND ND ND ND 0.143 0.467 ND NA 93/636 15 ND ND ND ND ND ND ND 0.933 0.933 ND <0.001e 7 WHO/III 15 ND ND ND ND ND ND ND 0.067 0.6 ND NA 93/636 15 ND ND ND ND ND ND ND 0.467 0.8 ND <0.001e A 1 WHO/III 10 ND 0 0.6 0.8 0.9 ND ND ND ND ND NA 93/636 5–10 ND 0.444 1 1 0.9 ND ND ND ND ND 0.004e 2 WHO/III 9 ND 0 0.1 ND ND ND ND ND ND ND NA 93/636 10 ND 0.222 0.556 ND ND ND ND ND ND ND 0.003e B 1 WHO/III 10 ND 0 0.1 0.2 1 ND ND ND ND ND NA 93/636 10 ND 0 0.5 0.9 0.9 ND ND ND ND ND <0.001e 2 WHO/III 12 ND 0 0.1 0 0.7 ND ND ND ND ND NA 93/636 12 ND 0 0 0.833 1 ND ND ND ND ND <0.001e D 1 WHO/III 15 ND ND ND ND ND 0 0 0 0.333 0.733 NA 93/636 15 ND ND ND ND ND 0.67 0.53 0.867 0.933 0.867 <0.001e 2 WHO/III 10 ND ND ND ND ND ND 0 0.2 0.3 0.9 NA 93/636 10 ND ND ND ND ND ND 0.2 0.9 0.9 1 <0.001e E 1 WHO/III 15 ND ND ND ND ND ND ND 0.467 0.467 ND NA 93/636 15 ND ND ND ND ND ND ND 0.8 1 ND <0.001e I 1 WHO/III 16 ND ND ND ND ND ND ND 0.375 ND ND NA 93/636 14 ND ND ND ND ND ND ND 0.929 ND ND <0.001e F 1 WHO/III 16 ND ND ND ND ND ND ND 0.25 0.813 ND NA 93/636 16 ND ND ND ND ND ND ND 0.938 1000 ND <0.001e G 1 WHO/III 15 ND ND ND ND ND ND ND 0.4 0.667 ND NA 93/636 14, 15 ND ND ND ND ND ND ND 0.714 1000 ND <0.001e H 1 WHO/III 7–10 ND ND ND ND ND 0 0 0.5 0.6 ND NA 93/636 10 ND ND ND ND ND 0 0.1 0.5 0.9 ND 0.112f 2 WHO/III 10 ND ND ND ND ND 0 0 0.1 0.3 ND NA 93/636 9, 10 ND ND ND ND ND 0 0 0.8 0.6 ND <0.001e a Lot 93/636 failed monkey neurovirulence test and contains 3% 472-C revertants. b Equal numbers of mice were given at each dose unless otherwise stated. c TCID = tissue culture infectious dose; NA= not applicable; ND = not determined. d WHO/III = reference vaccine for type-3. e Statistically significant: more neurovirulent than reference vaccine (P<0.05). f Statistically insignificant: not more neurovirulent than reference vaccine (P>0.05). 253Bulletin of the World Health Organization 2003, 81 (4) Transgenic mice for neurovirulence testing of live oral poliovirus vaccine good and bad vaccines included the following: (a) the combined (male plus female) paralysis rates for the reference vaccine must be40.95 at the high dose and 50.05 at the low dose; (b) dose effect must be significant; if it is not significant, the vaccine effect must be significant; and (c) no significant vaccine-by-dose interaction. The decision rule, i.e. the specific criteria for accepting or rejecting a vaccine lot, requires comparison of the LOR with limits, L1 and L2, derived from historical data for the reference vaccine. A test vaccine passes if the LOR 4 L1. L1 was calculated so that a test vaccine equivalent to the reference vaccinewould have a 95%probability of passing. A test vaccine fails if the LOR5L2 and hence L2was calculated so that a test vaccine equivalent to the reference vaccine would have a 1% probability of failing. The statistical decision model for acceptance/rejection of a test vaccine is presented in more detail in Annex 1, part I (see online version at: www.who.int/bulletin). It has been applied and successfully validated in the last phases of the study. Results Investigative stage The studies began with a comparison of the suitability of TgPVR1 and TgPVR21 mouse strains for OPV neuroviru- Table 2. Summary of results of the WHO collaborative study of TgPVR21 mice with type-3 oral poliovirus vaccine lot 95/526a Laboratory Test Vaccine No. of mice/doseb Paralysis rate Statistical analysis (P ) 3.5 log10 TCID50 c 4.5 log10 TCID50 c C 1 WHO/IIId 20 0.4 0.7 NAe 95/526 20 0.3 0.85 0.400f 2 WHO/III 20 0.15 0.6 NA 95/526 20 0.45 0.9 <0.001g 3 WHO/III 30 0.033 0.467 NA 95/526 30 0.533 0.933 <0.001g 4 WHO/III 30 0.333 0.633 NA 95/526 30 0.6 0.967 <0.001g 5 WHO/III 30 0.133 0.267 NA 95/526 30 0.167 0.767 <0.001g D 1 WHO/III 30, 29 0.233 0.379 NA 95/526 21, 24 0.381 0.75 0.001g A 1 WHO/III 30 0.133 0.633 NA 95/526 30 0.2 0.9 0.010g 2 WHO/III 30 0.333 0.533 NA 95/526 30 0.467 0.967 <0.001g E+I 1 WHO/III 30 0.567 0.767 NA 95/526 29, 30 0.483 1 0.162f F 1 WHO/III 28, 30 0.429 0.967 NA 95/526 30 0.8 0.967 0.003g 2 WHO/III 30 0.4 0.833 NA 95/526 30 0.667 0.967 0.002g G 1 WHO/III 31, 30 0.226 0.567 NA 95/526 30 0.452 0.839 0.007g 2 WHO/III 29, 28 0.241 0.607 NA 95/526 29, 30 0.552 0.833 0.007g B 1 WHO/III 29 0.276 0.483 NA 95/526 28, 30 0.321 0.7 0.061f 2 WHO/III 29, 30 0.172 0.567 NA 95/526 30, 29 0.433 0.828 0.008g 3 WHO/III 26, 30 0.231 0.533 NA 95/526 30 0.4 0.8 0.005g a Lot 95/526 failed monkey neurovirulence test and contains 1.7% 472-C revertants. b Equal numbers of mice were given at each dose unless otherwise stated. c TCID = tissue culture infectious dose. d WHO/III = reference vaccine for type-3. e NA = not applicable. f Statistically insignificant: not more neurovirulent than reference vaccine (P>0.05). g Statistically significant: more neurovirulent than reference vaccine (P<0.05). 254 Bulletin of the World Health Organization 2003, 81 (4) Research lence testing. Although TgPVR1 mice discriminated between wild-type poliovirus and a vaccine strain, they did not distinguish between vaccine lots that passed or failed the monkey NVT (9). Therefore, the TgPVR21 mouse line was selected for further studies (17, 18, 20). Preliminary experi- ments were conducted using these mice with all three types of polioviruses. TgPVR21 mice were able to discriminate OPV samples that passed from those that failed the monkey NVT. The data generated allowed selection of the appropriate dose range for inoculation, duration of clinical observation, identification of paresis/paralysis as the most important clinical sign for assessment of neurovirulence, and develop- ment of criteria for statistical decision-making model. These preliminary studies also suggested that the mouse model could be based on paralysis scores, in contrast to the simian model, which is based on lesion scores from histopathological examination. A special study was therefore performed to investigate the added value of histopathological examination of the mouse central nervous system for pass/fail decisions. Unlike the situation with the monkey NVT, histological examination of the mouse central nervous system offered no advantage for discriminating vaccine batches over clinical observation alone (21). The mouse test thus can be completed more rapidly than the monkey test. WHO collaborative study TheWHOcollaborative study underwent five phases. The first three phases were focused on type-3 poliovirus as this is agreed to be genetically the least stable strain of OPV. Type-1 and type-2 poliovirus vaccine samples were studied in phases 4 and 5. WHO vaccine references of all three types for the monkey NVT were used in all mouse tests. The choice of type-3 vaccine viruses was based on previous results from the monkey test and from the mutant Table 3. Results of the WHO collaborative study of TgPVR21 mice with type-3 oral poliovirus vaccine Laboratory Test Vaccine Monkey NVTa Mouse NVTa Resultsb No. of Paralysis rate mice/dosec 3.5 log10 4.5 log10 TCID50 d TCID50 C 1 WHO/IIIe Reference 32 0.188 0.469 – 96/568 Failed 32 0.438 0.75 Failed 93/664 Passed 32 0 0.156 Passed 2 WHO/III Reference 32 0.344 0.75 – 96/568 Failed 32 0.688 0.938 Failed 93/644 Passed 32 0 0.062 Passed A 1 WHO/III Reference 32, 30 0.219 0.7 – 95/526 Failed 31, 32 0.419 0.967 Failed 93/664 Passed 31 0.032 0.129 Passed 2 WHO/III Reference 32, 31 0.188 0.742 – 96/568 Failed 31, 32 0.548 1 Failed 93/644 Passed 32, 31 0.062 0.065 Passed 3 WHO/III Reference 32, 30 0.344 0.733 – 96/568 Failed 32 0.594 0.906 Failed 93/664 Passed 32 0.031 0.156 Passed F 1 WHO/III Reference 32 0.188 0.562 – 96/568 Failed 32 0.531 0.875 Failed 93/664 Passed 32 0. 062 0.031 Passed 2 WHO/III Reference 32, 31 0.188 0.742 – 96/568 Failed 32 0.344 0.969 Failed 93/644 Passed 32, 31 0.031 0.032 Passed G 1 WHO/III Reference 31, 30 0.226 0.567 – 95/526 Failed 31 0.452 0.839 Failed 93/658 Passed 30, 32 0.367 0.625 Passed 2 WHO/III Reference 32 0.281 0.594 – 96/568 Failed 32 0.469 0.812 Failed 93/644 Passed 30, 32 0.067 0.094 Passed 3 WHO/III Reference 30, 31 0.267 0.581 – 96/568 Failed 31, 32 0.419 0.688 Passedf 93/664 Passed 29, 30 0.103 0.167 Passed a NVT = neurovirulence test. b Results were analysed using the decision model. c Equal numbers of mice were given each dose unless otherwise stated. d TCID = tissue culture infectious dose. e WHO/III = reference vaccine for type-3. f Invalid test. 255Bulletin of the World Health Organization 2003, 81 (4) Transgenic mice for neurovirulence testing of live oral poliovirus vaccine analysis by polymerase chain reaction and restriction enzyme cleavage (MAPREC) test. Initial evaluations in mice used vaccine samples that failed the monkey test by a large margin and contained unusually high amounts (>3%) of neurovirulent 472-C revertants (23, 24) (Table 1). The evaluation was continued with vaccine lots that failed the monkey test and which contained only slightly increased amounts (>1%) of 472-C mutants (Table 2). These first two phases of the study provided data for development of a statistical model to define pass/fail decisions. Phase 3 was designed to validate the statistical decision model in tests when previously passed and failed vaccine lots were tested simultaneously (Table 3). Altogether 43 vaccine samples were tested in 10 laboratories in 114 mouse tests (Table 4). Thirty-one commercial OPV lots that passed the monkey NVT also passed the mouse NVT. Nine vaccine lots that failed the monkey NVT also failed the mouseNVT. To increase the number of samples that failed the monkey NVT, commercially produced vaccine viruses were passaged at 37–38 oC in AGMK or Vero cells to increase their neurovirulence for monkeys. Three such vaccine derivatives were prepared by thismethod and they also failed bothmonkey and mouse NVTs. The applicability for type-1 strain of the statistical pass/ fail decisionmodel developed for type-3OPVwas evaluated in two series of tests. Initially doses of 1.5 and 2.5 log10 TCID50 were used (Table 5) but were subsequently increased to 1.75 and 2.75 log10 TCID50 to achieve paralysis rates of the reference vaccine at the low dose of above 0.05 (Table 6), as required by the statistical decision model. Since there were no commercial vaccine lots available that had failed themonkey test, original vaccineswere passaged in AGMK or Vero cells at 37–38 oC. Four such experimental samples, one that passed and three that failed the monkey NVT, were tested in mice. A total of 20 type-1 commercial vaccine lots that had passed themonkeyNVTwere included in the study. In 39 mouse NVTs performed in five laboratories there was complete correlation of results between mice and monkeys (Table 4). A total of 27 type-2 vaccine samples were tested in six laboratories in 53 mouse tests (Table 4). Twenty-three commercial vaccine batches that had passed the monkey test also passed the mouse test. One vaccine batch that passed the monkey test gave variable results in the mouse test, passing five times and failing twice. This suggests that the batch concerned may have been on the borderline between pass and fail in the mouse test. It is not known what the results would have been had the monkey test been repeated one or more times, and thus whether this batch was also on the borderline between pass and fail in the monkey test. This batch was the only one that gave anomalous results in the mouse and monkey tests for any of the serotypes. Three experimental batches obtained from one manufacturer, who at that time was investigating potential changes to the production process, were the only available lots that failed the monkey NVT. All three samples failed the mouse NVT in all laboratories. Thus there was close agreement between mice and monkeys in the study with serotype 2. The results obtained demonstrated that the statistical pass/fail decision model developed for type-3 polioviruses was applicable and valid for type-2 OPV (Table 7). Conclusion A WHO Collaborative Study on transgenic mice as an alternative to the monkey NVT has been completed with all three OPV serotypes. Eighty-four commercial vaccine batches and ten experimental vaccine samples of type-1, type-2, and type-3 were tested in 206 mouse NVTs. A limitation to the study design was the unavailability of commercially produced type-1 batches that consistently failed the monkey NVT. Table 4. Summary of results of the WHO collaborative study of TgPVR21 mouse neurovirulence test with oral poliovirus vaccine Type Samples No. of samples Monkey NVTa Mouse NVT Tests (n) Laboratories (n) Results Pass Fail Type-3 Commercial batches 31 Pass 54 8 54 0 Commercial batches 9 Fail 56b 10 0 53 Experimental samplesc 3 Fail 4 2 0 4 Total 43 NA 114 20 54 57 Type-1 Commercial batches 20d Pass 32e 5 29 0 Experimental samplec 1 Pass 1 1 1 0 Experimental samplesc 3 Fail 6 2 0 6 Total 24 NA 39 8 30 6 Type-2 Commercial batches 24 Pass 39f 6 36 2 Experimental batchesg 3 Fail 14 6 0 14 Total 27 NA 53 12 36 16 a NVT = neurovirulence test; NA, not applicable. b Three tests in one laboratory were invalid; improvement in technique was required. c Passages of original vaccine in AGMK or Vero cells at 37–38 oC. d Batches evaluated in ‘‘in-house’’ tests against national references are not included. e Two tests were repeated because L1<LOR<L2, where LOR is log odds ratio; one test was repeated because paralysis rate at a low dose was <0.05. f One test was repeated because L1<LOR<L2. g Obtained from one manufacturer experimenting with production process. 256 Bulletin of the World Health Organization 2003, 81 (4) Research Table 5. Summary of results of the WHO collaborative study of TgPVR21 mice with type-1 oral poliovirus vaccine (initial doses of inoculum) Labora- Test Vaccine Paralysis rate (n) LORa Results tory Females Males Dose (log10 TCID50) b Mouse NVTc Monkey NVT 1.5 2.5 1.5 2.5 A 1 I-2 0.063 (16)d 0.5 (16) 0 (16) 0.467 (15) 0.291 Re-teste Pass I-5 0 (16) 0.5 (16) 0.063 (16) 0.4 (15) 0.174 Re-teste Pass WHO/If 0 (16) 0.5 (16) 0 (16) 0.375 (16) – Reference Reference 2 I-1 0 (16) 0.333 (15) 0 (16) 0.813 (16) –1.437 Pass Pass I-7 0.063 (16) 0.286 (14) 0 (16) 0.267 (15) –0.517 Pass Pass WHO/I 0.133 (15) 0.688 (16) 0.063 (16) 0.563 (16) – Reference Reference 3 I-4 0.063 (16) 0.6 (15) 0.313 (16) 0.733 (15) 0.162 Pass Pass I-6 0.125 (16) 0.625 (16) 0.125 (16) 0.875 (16) 0.202 Pass Pass WHO/I 0.25 (16) 0.563 (16) 0.067 (15) 0.733 (15) – Reference Reference E 1 I-10 0.2 (15) 0.313 (16) 0.067 (15) 0.25 (16) –1.939 Pass Pass I-3 0 (16) 0.063 (16) 0.063 (16) 0.188 (16) –0.686 Pass Pass WHO/I 0.063 (16) 0.625 (16) 0.133 (15) 0.438 (16) – Reference Reference F 1 I-3 0 (16) 0.063 (16) 0.063 (16) 0.063 (16) –2.44 Pass Pass I-6 0.188 (16) 0.813 (16) 0.063 (16) 0.625 (16) 0.816 Re-testg Pass WHO/I 0.125 (16) 0.5 (16) 0 (16) 0.563 (16) – Reference Reference 2 I-4 0 (16) 0.563 (16) 0.063 (16) 0.4 (15) –0.111 Pass Pass I-9 0.125 (16) 0.625 (16) 0.063 (16) 0.313 (16) 0.039 Pass Pass WHO/I 0.188 (16) 0.333 (15) 0 (15) 0.563 (16) – Reference Reference C 1 I-2 0.063 (16) 0.25 (16) 0.063 (16) 0.563 (16) –1.17 Pass Pass I-8 0.125 (16) 0.75 (16) 0.125 (16) 0.5 (16) –0.096 Pass Pass WHO/I 0 (16) 0.375 (16) 0.25 (16) 0.938 (16) – Reference Reference 2 289 0.187 (16) 0.562 (16) 0.063 (16) 0.562 (16) –0.089 Pass Pass SID 38/4h 0.313 (16) 0.875 (16) 0.25 (16) 0.938 (16) 1.469 Fail Fail WHO/I 0 (16) 0.562 (16) 0.25 (16) 0.623 (16) – Reference Reference a LOR = log odds ratio; limit values for LOR used in the analysis: L1 = 0.718; L2 = 1.016. b TCID = tissue culture infectious dose. c NVT = neurovirulence test. d Figures in parentheses indicate the number of mice. e Re-test because reference paralysis rate at the 1.5 dose is 0.0. f WHO/I = reference vaccine for type-1. g Re-test because L1<LOR<L2. h Fourth passage of original vaccine in Vero cells at 38 oC. Table 6. Summary of results of theWHO collaborative study of TgPVR21micewith type-1 oral poliovirus vaccine (final doses of inoculum) Labora- Test Vaccine Paralysis rate (n) LORa Results tory Females Males Dose (log10 TCID50) b Mouse NVTc Monkey NVT 1.75 2.75 1.75 2.75 A 1 I-10 0.063 (16)d 0.625 (16) 0.2 (15) 0.75 (16) –0.696 Pass Pass I-2 0.125 (16) 0.313 (16) 0.063 (16) 0.333 (15) –1.95 Pass Pass WHO/Ie 0.125 (16) 0.875 (16) 0.133 (15) 0.875 (16) – Reference Reference 2 99-I-1 0.062 (16) 0.625 (16) 0 (16) 0.5 (16) –0.225 Pass Pass 99-I-2 0.375 (16) 0.467 (15) 0.25 (16) 0.467 (15) –0.518 Pass Pass WHO/I 0.25 (16) 0.938 (16) 0 (16) 0.562 (16) – Reference Reference 3 99-I-2 0.125 (16) 0.8 (15) 0.133 (15) 0.733 (15) 0.603 Pass Pass WHO/I 0.063 (16) 0.625 (16) 0.083 (12) 0.714 (14) – Reference Reference E 1 I-10 0.125 (16) 0.438 (16) 0.063 (16) 0.688 (16) 0.107 Pass Pass I-11 0 (16) 0.375 (16) 0.133 (15) 0.875 (16) 0.163 Pass Pass WHO/I 0.063 (16) 0.667 (15) 0.125 (16) 0.438 (16) – Reference Reference J 1 J/I/1 0.25 (16) 0.875 (16) 0.25 (16) 0.875 (16) –0.202 Pass Pass J/I/2 0.313 (16) 1 (16) 0.438 (16) 0.875 (16) 0.409 Pass Pass WHO/I 0.313 (16) 0.813 (16) 0.313 (16) 0.938 (16) – Reference Reference a LOR = log odds ratio; limit values for LOR used in the analysis: L1= 0.734, L2 = 1.037. b, c, d, e See corresponding footnotes, Table 5. 257Bulletin of the World Health Organization 2003, 81 (4) Transgenic mice for neurovirulence testing of live oral poliovirus vaccine However vaccine derivatives were used as surrogates. There was good correlation between the results of the monkey and mouse NVTs for all three OPV types. Statistical analysis of the data demonstrated that the TgPVR21mouse test is as sensitive and reliable as the monkey NVT. A statistical model for acceptance/rejection of OPV lots tested in the mouse test has been validated and proved to be suitable for all three types.Our results demonstrate the first successful introduction of transgenic animals into control of biologicals. The special line of mice with defined genetic and microbiological quality standards yielded highly uniform results, and a significantly shorter time was required for the test—2weeks for themouse test instead of 1.5–2 months for the monkey test. The transgenic mouse NVT is more attractive than the monkey NVT for ethical and practical considerations since it reduces use of primates and eliminates hazards to personnel working with primates. The WHO Ethical Committee on Biological Safety has approved the mouse NVT as an alternative to the Table 7. Summary of results of the WHO collaborative study of TgPVR21 mice with type-2 oral poliovirus vaccine Labora- Test Vaccine Paralysis rate (n) LORa Results tory Females Males Dose (log10 TCID50) b Mouse NVTc Monkey NVT 5 6 5 6 A 1 98/690 0 (16)d 0.75 (16) 0.375 (16) 0.938 (16) 0.118 Pass Pass 98/702 0.625 (16) 1 (16) 0.938 (16) 1 (16) 3.434 Fail Fail WHO/IIe 0.063 (16) 0.625 (16) 0.375 (16) 0.938 (16) – Reference Reference 2 99-II-3 0.063 (16) 0.313 (16) 0.125 (16) 0.563 (16) –0.47 Pass Pass 99-II-5 0.063 (16) 0.188 (16) 0 (16) 0.313 (16) –1.375 Pass Pass WHO/II 0.188 (16) 0.438 (16) 0.125 (16) 0.625 (16) – Reference Reference C 1 98/688 0.188 (16) 0.5 (16) 0.625 (16) 0.813 (16) 0.353 Pass Pass 99/II/4 0.125 (16) 0.563 (16) 0.25 (16) 0.875 (16) –0.092 Pass Pass WHO/II 0.063 (16) 0.375 (16) 0.563 (16) 0.875 (16) – Reference Reference 2 99-II-1 0 (16) 0.188 (16) 0.25 (16) 0.625 (16) –0.756 Pass Pass 99-II-2 0.062 (16) 0.25 (16) 0.312 (16) 0.733 (15) –0.29 Pass Pass WHO/II 0.062 (16) 0.375 (16) 0.438 (16) 0.688 (16) – Reference Reference 3 M-2-4 0.063 (16) 0.125 (16) 0.5 (16) 0.75 (16) 0 Pass Pass WHO/II 0.188 (16) 0.25 (16) 0.313 (16) 0.688 (16) – Reference Reference E 1 98/690 0.125 (16) 0.313 (16) 0.375 (16) 0.533 (15) –1.106 Pass Pass 98/702 0.563 (16) 0.875 (16) 0.938 (16) 0.867 (15) 1.287 Fail Fail WHO/II 0.4 (15) 0.533 (15) 0.5 (16) 0.875 (16) – Reference Reference F 1 98/688 0 (16) 0.25 (16) 0.125 (16) 0.5 (16) –0.625 Pass Pass 99-II-6 0 (16) 0.375 (16) 0.125 (16) 0.438 (16) –0.495 Pass Pass WHO/II 0.062 (16) 0.562 (16) 0.125 (16) 0.5 (16) – Reference Reference 2 M-2-4 0.313 (16) 0.688 (16) 0.313 (16) 0.938 (16) 0.403 Pass Pass WHO/II 0.063 (16) 0.5 (16) 0.438 (16) 1 (16) – Reference Reference 3 98/688 0.063 (16) 0.333 (15) 0.5 (16) 0.75 (16) 0.959 Fail Pass 98/702 0.688 (16) 0.6 (15) 0.938 (16) 0.938 (16) 3.001 Fail Fail WHO/II 0 (16) 0.188 (16) 0.313 (16) 0.5 (16) – Reference Reference 4 98/688 0.063 (16) 0.438 (16) 0.5 (16) 0.688 (16) 0.984 Fail Pass 98/702 0.5 (16) 0.875 (16) 0.625 (16) 0.938 (16) 2.929 Fail Fail WHO/II 0 (16) 0.4 (15) 0.062 (16) 0.563 (16) – Reference Reference J 1 98/702 0.875 (16) 1 (16) 0.813 (16) 1 (15) 2.433 Fail Fail WHO/II 0.188 (16) 0.75 (16) 0.625 (16) 1 (16) – Reference Reference 2 J/II/2 0.25 (16) 0.667 (15) 0.188 (16) 0.75 (16) 0.226 Pass Pass J/II/3 0.125 (16) 0.563 (16) 0.4 (15) 0.75 (16) 0.184 Pass Pass WHO/II 0.125 (16) 0.688 (16) 0.25 (16) 0.625 (16) – Reference Reference 3 J/II/4 0.063 (16) 0.75 (16) 0.375 (16) 1 (16) 0.774 Re-testf Pass J/II/5 0.063 (16) 0.467 (15) 0.375 (16) 0.813 (16) –0.049 Pass Pass WHO/II 0.188 (16) 0.375 (16) 0.188 (16) 1 (16) – Reference Reference 4 J/II/4 0.375 (16) 1 (16) 0.25 (16) 0.688 (16) –0.851 Pass Pass J/II/1 0.125 (16) 0.875 (16) 0.625 (16) 0.933 (15) –0.588 Pass Pass WHO/II 0.312 (16) 0.875 (16) 0.688 (16) 1 (16) – Reference Reference a LOR = log odds ratio; limit values for LOR used in the analysis: L1 = 0.665; L2 = 0.940. b, c, d See corresponding footnotes, Table 5. e WHO/II = reference vaccine for type-2. f Re-test because L1<LOR<L2. 258 Bulletin of the World Health Organization 2003, 81 (4) Research monkey NVT for all three types of OPV (16). To avoid confusion however the Committee also confirmed that the test in simians remains the gold standard for evaluating the neurovirulence of OPV, and should be used to validate new virus seed lots or changes in the manufacturing process. Laboratories cannot simply switch from using monkeys tomice. Although the transgenic mouse NVTwas successfully introduced into most laboratories, others had some metho- dological difficulties. For example, the test requires the very precise positioning of the inoculum into the mouse spinal cord — a very small target area. Operators are therefore required to acquire this special skill during the training period. A standard implementation process has been developed by WHO to facilitate introduction of the new technique. Also, WHO has recommended that to qualify as competent to perform the mouse test, laboratories should complete the standard implementation process (see Annex 1, part II at www.who.int/bulletin) and satisfy their national control authority that they have gained sufficient experience in the test. Once qualified as competent, each laboratory should continue to monitor its continued competence to perform the test (22). In order to ensure the supply of such mice, two breeding stations of TgPVR21 mice have been established in Asia and Europe. Both these stations are provided with frozen embryos from the CIEA and will conduct consistent controls of the quality of animals as prescribed by the CIEA. n Re´sume´ Des souris transge´niques en remplacement des singes pour l’e´preuve de neurovirulence applique´e au vaccin antipoliomye´litique oral vivant : validation par une e´tude collective de l’OMS Objectif De´terminer si des souris transge´niques sensibles au poliovirus (souris TgPVR ligne´e 21, e´leve´es et fournies par le Central Institute for Experimental Animals (Japon)) peuvent eˆtre utilise´es en remplacement des singes dans l’e´preuve de neurovirulence applique´e au vaccin antipoliomye´litique oral (VPO). Me´thodes Les donne´es de 9 laboratoires ont e´te´ utilise´es pour e´valuer la neurovirulence de 94 pre´parations de VPO ou de´rive´s du vaccin contre les trois se´rotypes de poliovirus teste´es sur la souris TgPVR21 lors d’une vaste e´tude collective mene´e par l’OMS. Re´sultats L’analyse statistique des donne´es a montre´ que l’e´preuve de neurovirulence sur la souris TgPVR21 e´tait de sensibilite´ et de reproductibilite´ comparables a` celles de l’e´preuve classique OMS sur le singe. Un mode`le statistique d’acceptation ou de rejet des lots de VPO d’apre`s les re´sultats de l’e´preuve chez la souris a e´te´ de´veloppe´ et valide´, et s’est re´ve´le´ convenir pour les trois types de vaccin. L’e´valuation de l’e´preuve de neurovirulence chez la souris transge´nique reposait sur l’examen clinique des souris paralyse´es. Contrairement a` ce qui se passe dans l’e´preuve de neurovirulence chez le singe, l’examen histologique du syste`me nerveux central de chacune des souris ne pre´sentait pas d’avantage sur un examen clinique approfondi. Conclusion A partir de ces donne´es, le Comite´ OMS d’experts de la standardisation biologique a approuve´ l’e´preuve de neuroviru- lence chez la souris en remplacement de l’e´preuve sur le singe pour les trois types de vaccin VPO et a de´fini une proce´dure normalise´e de mise en œuvre a` l’intention des laboratoires qui souhaitent appliquer cette nouvelle e´preuve. Il s’agit la` de la premie`re utilisation re´ussie d’animaux transge´niques dans le domaine du controˆle des produits biologiques. Resumen Ratones transge´nicos como alternativa a los monos para la prueba de neurovirulencia de la vacuna oral viva contra el poliovirus: validacio´n en un estudio en colaboracio´n de la OMS Objetivo Determinar la idoneidad del uso de ratones transge´nicos sensibles al poliovirus (ratones TgPVR, cepa 21, criados y proporcionados por el Instituto Central para Animales de Laboratorio, Japo´n) como alternativa a los monos en la prueba de neurovirulencia (PNV) para la vacuna oral viva contra el poliovirus (OPV). Me´todos Se utilizaron los datos de nueve laboratorios para evaluar la neurovirulencia de 94 preparados de OPV o derivados vacunales contra los tres serotipos en ratones TgPVR21 en un amplio estudio en colaboracio´n de la OMS. Resultados El ana´lisis estadı´stico de los datos demostro´ que la PNV aplicada a los ratones TgPVR21 era comparable, en cuanto a sensibilidad y reproducibilidad, a la prueba convencional de la OMS con monos. Se desarrollo´ y valido´ un modelo estadı´stico para aceptar o rechazar los lotes de OPV en la prueba con ratones, modelo que resulto´ adecuado para los tres tipos de vacuna. La evaluacio´n de la PNV en ratones transge´nicos se basa en la observacio´n clı´nica de los ratones con para´lisis. A diferencia de la PNV con monos, el examen histolo´gico del sistema nervioso central de los ratones no reporto´ ninguna ventaja adicional en comparacio´n con una observacio´n clı´nica cuidadosa y detallada. Conclusio´n Teniendo en cuenta estos datos, el Comite´ de Expertos de la OMS en Patrones Biolo´gicos aprobo´ la PNV en ratones como una alternativa va´lida a la prueba en monos para los tres tipos de OPV, y describio´ un procedimiento de aplicacio´n normalizado para los laboratorios que deseen utilizarla. Es la primera vez que se logra utilizar con e´xito animales transge´nicos para controlar productos biolo´gicos. 259Bulletin of the World Health Organization 2003, 81 (4) Transgenic mice for neurovirulence testing of live oral poliovirus vaccine References 1. Wood DJ, Macadam A. Laboratory tests for live attenuated poliovirus vaccines. Biologicals 1997;25:3-15. 2. Requirements for poliomyelitis vaccine (oral). Geneva: World Health Organization; 1990. p. 46-9. WHO Technical Report Series, No.800. 3. Contreras G, Furesz J, Karpinski K, Grinwich K, Gardell C. Experience in Canada with the new revised monkey neurovirulence test for oral poliovirus vaccine. Journal of Biological Standards 1988;16:195-205. 4. Furesz J, Contreras G. Some aspects of the monkey neurovirulence test used for the assessment of live poliovirus vaccines. In: Brown F, Lewis BP, editors. Poliovirus attenuation: molecular mechanisms and practical aspects. Work- shop, Bethesda, MD, December 1991. Developments in Biological Standards 1993;78:61-70. 5. Ren R, Costantini F, Gorgacz EJ, Lee JJ, Racaniello VR. Transgenic mice expressing a human poliovirus receptor: a new model for poliomyelitis. Cell 1990;63:353-62. 6. Koike S, Taya C, Kurata T, Abe S, Ise I, Yonekawa H, et al. Transgenic mice susceptible to poliovirus. Proceedings of the National Academy of Sciences of the United States of America 1991;88:951-5. 7. Koike S, Taya C, Aoki J, Matsuda Y, Ise I, Takeda H, et al. Characterization of three transgenic mouse lines that carry human poliovirus receptor gene — influence of the transgene expression on pathogenesis. Archives of Virology 1994;139:351-63. 8. New approaches to poliovirus diagnosis using laboratory technique: Memorandum from a WHO meeting. Bulletin of the World Health Organization 1992;70:27-33. 9. Dragunsky E, Gardner D, Taffs R, Levenbook I. Transgenic PVR Tg-1 mice for testing of poliovirus type 3 neurovirulence: comparison with monkey test. Biologicals 1993;21:233-7. 10. Horie H, Koike S, Kurata T, Sato-Yoshida Y, Ise I, Ota Y, et al. Transgenic mice carrying the human poliovirus receptor: new animal model for study of poliovirus neurovirulence. Journal of Virology 1994;68:681-8. 11. Meeting on transgenic mice susceptible to poliovirus as a model for neurovirulence tests of live poliomyelitis vaccine (oral). Geneva: World Health Organization; 1993 WHO document WHO/MIM/PVD/94.1. 12. Hioki K, Katoh H, Endo S, Saito M, Nomura T. [Planned production of poliovirus- sensitive transgenic mice] Experimental Animals 1993;42:300-3. In Japanese. 13. Nomura T. Practical development of genetically engineered animals as human disease models. Laboratory Animal Science 1997;47:113-7. 14. Levenbook I, Dragunsky E, Pervikov Y. Development of a transgenic mouse neurovirulence test for oral poliovirus vaccine: international collaborative study 1993-1999. Vaccine 2000;19:163-6. 15. Wood DJ, Padilla A, Griffiths E. WHO Expert Committee on Biological Standardization: highlights of the 50th meeting, October 1999. Biologicals 2000;28:199-206. 16. Quality assurance and safety of biologicals. WHO Drug Information 2000;14:218-21. 17. Abe S, Ota Y, Koike S, Kurata T, Horie H, Nomura T, et al. Neurovirulence test for oral live poliovaccines using poliovirus-sensitive transgenic mice. Virology 1995;206:1075-83. 18. Abe S, Ota Y, Doi Y, Nomoto A, Nomura T, Chumakov KM, et al. Studies on neurovirulence in poliovirus-sensitive transgenic mice and cynomolgus monkeys for the different temperature-sensitive viruses derived from the Sabin type 3 virus. Virology 1995;210:160-6. 19. Maintenance and distribution of transgenic mice susceptible to human viruses: Memorandum from a WHO Meeting. Bulletin of the World Health Organization 1993;71:497-502. 20. Dragunsky E, Taffs R, Chernokhvostova Y, Nomura T, Hioki K, Gardner D, et al. A poliovirus-susceptible transgenic mouse model as a possible replacement of the monkey neurovirulence test of oral poliovirus vaccine. Biologicals 1996;24:77-86. 21. Dragunsky E, Chernokhvostova Y, Taffs R, Chumakov K, Gardner D, Asher D, et al. TgPVR21 mice for testing type-3 oral poliovirus vaccines: role of clinical observation and histological examination. Vaccine 1997;15:1863-6. 22. Recommendations for the production and control of poliomyelitis vaccine (oral). In: WHO Expert Committee on Biological Standardization. Fiftieth Report. Geneva: World Health Organization; 2002. p. 31-93. WHO Technical Report Series, No. 904. 23. Chumakov K, Powers L, Noonan K, Roninson I, Levenbook I. Correlation between amount of virus with altered nucleotide sequence and the monkey test for acceptability of oral poliovirus vaccine. Proceedings of the National Academy of Sciences of USA 1991;88:199-203. 24. Chumakov K, Norwood L, Parker M, Dragunsky E, Taffs R, Ran Y, et al. Assessment of the viral RNA sequence heterogeneity for control of OPV neurovirulence. Developments in Biological Standardization 1992;78:77-86. 25. Manual of laboratory methods for testing the potency of final vaccines used in the WHO Expanded Programme on Immunization. Part II. 9. Live oral poliomyelitis vaccine 1990. WHO document WHO/BLG/95.1. p. 67-74. 260 Bulletin of the World Health Organization 2003, 81 (4) Research Annex 1 Part I. Statistical decision model for acceptance or rejection of a test vaccine The decisionmodel for acceptance or rejection of a test vaccine can be applied by using the six steps described below. Step 1: Check paralysis proportions for the reference vaccine. The combined (male and female) paralysis rates for the reference vaccine must be40.95 at the high dose and50.05 at the low dose. Samples that do not meet this criterion require re-testing. Step 2: Check estimability. If the maximum likelihood procedure fails to converge, then the pass/fail criteria are applied at each dose. The test vaccine is accepted if the vaccine passes at both doses (LOR<L1). The test vaccine fails if the vaccine fails at both doses (LOR>L2). (LOR is log odds ratio and L1 and L2 are limits 1 and 2, respectively, indicating the lower and upper limits applied in the statistical decision model).All other outcomes will require a re-test. Step 3: Check vaccine 6 dose interaction by applying the maximum likelihood procedure to the logistic regression model. If the interaction is significant, the pass/fail criteria are applied to each dose. The test vaccine is accepted if the vaccine passes at both doses (LOR<L1 at each dose) and fails if LOR>L2 at both doses. If paralysis rates for the test vaccine doses are either at 0.0 or 1.0, the following decision process can be applied: if both doses of the test vaccine produce 0.0 paralysis, the vaccine is accepted; if 0.0 paralysis occurs for the low dose and at high dose the paralysis rate is lower than the corresponding reference result, use step 5; if the test vaccine has paralysis rates of 1.0 at both doses, the vaccine fails; if the test vaccine has 1.0 paralysis rate only at the high dose, then the decision process is applied to the combined results for both doses and also to the LOR for the low dose — the vaccine is required to pass the decision criteria for both the combined estimate of the LOR and the estimate at the low dose. Step 4: Check for a significant dose effect. If the dose effect is significant then proceed to Step 5: if it is not significant, test the vaccine effect. If the vaccine effect is also not significant, the experiment must be repeated. Otherwise, proceed to Step 5. Step 5: Calculate the LOR: if LOR4L1, the vaccine passes; if L1<LOR<L2, retesting is required (go to Step 6); and if LOR5L2, the vaccine fails. Step 6: If a pass/fail decision is not reached in Steps 1–5 and a repeat experiment is required, the decision process is applied either to pooled data from the two experiments or the data from the second experiment alone. If the re-test was initiated as a result of a technical problem in the first test or because of a lack of validity of the reference profile, steps 1–5 must be repeated using the data from the second experiment alone. If the re-test was initiated because LOR was between L1 and L2, or because of a problemwith the test profile, Steps 1–5must be repeated using pooled data from both experiments. If the experiment involved testing with more than one test vaccine, individual analysis of the data must be carried out comparing each vaccine with the concurrent WHO reference vaccine. Details of the procedures used to calculate the limits, L1 and L2, are available in the standard operating procedure available from WHO. Part II. Implementation process The implementation process consists of three main compo- nents: training, evaluation, and implementation. Training: Intraspinal inoculation of mice assumes that the inoculum will be delivered to a small target area, the anterior horns of the spinal cord lumbar segment. The laboratory is required to acquire a special skill during the training period. (a) Initial training in the intraspinal inoculation, clinical assessment of mice, and statistical analysis procedure at the Food and Drug Administration or Japanese Poliomye- litis Research; (b) Practising the intraspinal inoculation of conventional mice with India ink; (c) Performing two tests on TgPVR mice with vaccine samples of each type. The samples are provided with known titres and themonkey andmouseNVTdata (passed or failed vaccine). Evaluation of precision of virus titration: The laboratory is required to use the standardWHO poliovirus titration method (1) and is provided with a titration reference reagent. The laboratory should perform several assays to obtain a precision for the confidence limits for the mean 40.3 log10 (tissue culture infectious dose) (TCID)50. The mean value obtained is compared with the assigned value for the reference to normalize titration values for the vaccine samples. Implementation procedure: (a) Vaccines are provided byWHO to the laboratory as a panel of coded samples. (b) A minimum of three valid tests (from a total of no more than four tests) are required to complete the implementa- tion process. Test results are submitted to WHO. (c) On the basis of the obtained results, WHO will assess whether the laboratory has successfully implemented the mouse test. (d) The implementation procedure shall be performed for each oral poliovirus vaccine type. During the implementation process the laboratory accumu- lates data from five valid tests to determine its own L1 and L2 limits and will then use these for batch release purposes after the implementation has been successfully completed. Part III. Participating investigators and institutions Principal investigators: Dr T. Nomura, Central Institute of Experimental Animals (CIEA), Japan; Dr S. Abe, Japan Poliomyelitis Research Institute (JPRI), Japan; Dr T. Kurata, National Institute of Infectious Diseases (NIID), Japan; Dr A. Schmeel, Chiron Behring (CB), Germany; Dr Guo Ren, Central Institute ofMedical Biology, China; Dr A. Deatly, Wyeth-Lederle Vaccines and Pediatrics (WLVP), USA; Dr E. Dragunsky, US Food and Drug Administration (FDA); Dr G. Karganova, Institute of Poliomyelitis and Viral Encephalitides (IPVE), Russian Federation; Dr E. Evreinova, LA Tarassevich State Research Institute for Standardization and Control of Medical and Biological Preparations, Russian Federation; Dr O. Vanloocke, GlaxoSmithKline Biologicals (GSKB), Belgium; Dr D. J. Wood, National Institute for Biological Standards and Control (NIBSC), England. Transgenic mice for neurovirulence testing of live oral poliovirus vaccine ABulletin of the World Health Organization 2003, 81 (4) Statistical analysis of the data was performed, and the statistical model for acceptance or rejection of oral poliovirus vaccine lots in the mouse NVT was developed by Dr K. Karpinski (Canada), with participation of Dr R. Taffs, Dr A. Ivshina, Dr H. Hsu (FDA), and Dr A. Heath (NIBSC). Standard operating procedures were prepared by Dr E. Dragunsky, Dr R. Taffs, Dr D. Asher, Dr I. Leven- book (FDA), Dr K. Karpinski (Canada), and revised by Dr D. J. Wood (NIBSC). Histological examination was performed by Dr E. Dragunsky (FDA) and Mrs S. Marsden (NIBSC). The study was initiated and supported by WHO (Dr Y. Ghendon, Dr Y. Pervikov, Dr E. Griffiths) and coordinated by Dr I. Levenbook (FDA, later WHO). The consultant for the study was Dr J. Furesz (Canada). The collaborative study would not have been a success without the dedication and hard work of Mr M. Saito, Mr K. Hioki (CIEA); Dr H. Ota (JPRI); Dr N. Nagata, Dr Y. Horiuchi, Dr K. Konishi, Mr. I. Hatano, Ms A. Harashima (NIID); Dr M. Fibi (CB); Dr Carolyn Weeks-Levy, Ms Toya McWilliams, M.G. McMullen (WLVP); Dr D. Gardner, Dr K. Chumakov, Dr G. Rezapkin, Ms J. Enterline (FDA); Dr A. Rumyant- sev (IPVE); Ms A. Millecamps, P. Beaufort, A-F. Macq, Dr A. Van-den-Bossche, Dr D. Gustin, Mr I. Hotelet (GSKB); Dr R. Hull, Mr G. Crossland, Ms G. Dunn, Ms S. Marsden (NIBSC); and support of Dr S. Hashizume and Dr Y. Doi (JPRI), Dr D. Asher (FDA), Dr V. Grachev (IPVE), and Dr P. Minor (NIBSC). Research B Bulletin of the World Health Organization 2003, 81 (4)