Tuberculin sensitivity in guinea-pigs after vaccination with varying doses of BCG of 12 different strains*

Tuberculin sensitivity in guinea-pigs after vaccination with varying doses of BCG of 12 different strains* A. LADEFOGED,1 K. BUNCH-CHRISTENSEN,2 & J. GULD3 In several previously reported studies a number of BCG strains, including those most widely used in vaccine production, were ranked according to their in vivo activity in various experimental models in rodents and to the local and allergic response that they provoked in children. In this report, 12 strains are ranked in terms of tuberculin conversion in guinea- pigs according to the minimum sensitizing dose. For 10 of these strains, this minimum dose is very low, ranging from S to 50 culturable particles. Thus the traditional practice of tuberculin-testing guinea-pigs that have been vaccinated with a full human dose of BCG (of the order of one million culturable particles) has such a low discriminating power that it is useless as a routine test for currently used vaccines. The ranking obtained in this model was largely but not completely in accordance with similar rankings in other models. The Tokyo strain and, to some extent, the London strain ranked comparatively lower than they do in children. The pronounced lack of virulence of the Prague strain was confirmed. Guinea-pigs are widely used in the laboratory con- trol of the biological activity ofBCG vaccine, mostly because reactions to tuberculin in guinea-pigs are akin to those in man. But there is one great differ- ence between guinea-pigs and human beings in their response to BCG inoculation. Whereas, in man, the tuberculin sensitivity is a function of the dose of BCG, as reported by Edwards et al. in 1953 (6), the tuberculin sensitivity and immunogenic effect in guinea-pigs are independent of the vaccine dose over a wide range, as reported by Jespersen in 1956 (10) and by Tolderlund et al. in 1960 (15). Large doses of BCG induce an earlier but not an ultimately stronger immunity than does even a remarkably small dose. It would seem that BCG is sufficiently virulent in guinea-pigs for it to multiply freely until the immu- nization has reached a maximum. A similar pheno- * From the WHO Collaborating Centre for world-wide reference for BCG seed lots and for co-ordination of control of BCG products, BCG Department, Statens Seruminstitut, Copenhagen, Denmark. 1 Research Assistant, BCG Department, Statens Serum- institut. 2 Chief, BCG Department, Statens Seruminstitut. 3Medical Officer, Tuberculosis, Division of Communi- cable Diseases, World Health Organization, 1211 Geneva 27, Switzerland. menon was observed in experiments of protection in bank voles by Jerspersen in 1954 (9), and confirmed by Jespersen & Weis Bentzon in 1964 (11) and by Ladefoged et al. in 1970 (12). In comparing BCG strains expected to differ only slightly, it would seem justified to make a kind of titration, i.e. to vaccinate with decreasing doses of vaccine in order to estimate the lowest dose that induces tuberculin sensitivity and immunity, rather than to compare the effect of large doses of BCG. We carried out a preliminary study along these lines (unpublished data, 1965) to compare 8 different BCG products, and showed the products to differ more clearly in terms of allergenic potency than of protective power. For some of the products, 10 culturable particles of BCG were enough to induce maximum tuberculin sensitivity. Around 50 cultur- able particles gave almost uniform protection for all products. Based on these preliminary results and aiming at a characterization and ranking of BCG strains widely used in man, a number of further studies have been carried out in the WHO Collaborating Centre for BCG. As in our previously published studies of the effect of BCG in the hamster and the bank vole (3, 4, 12), liquid vaccines from the various strains were 3456 - 435 - BULL. WORLD HEALTH ORGAN., Vol. 53, 1976 A. LADEFOGED ET AL. prepared in this laboratory with a uniform technique aiming at uniform in vitro properties. A number of these vaccine batches have been used not only for the guinea-pig studies reported here but also for the investigations in man reported by Vallishayee et al. (16). This report, which is the last in a series, covers the results from 8 experiments in guinea-pigs, in which 12 different BCG strains were compared. The studies were undertaken in 1966-70. The guinea- pig investigations were planned to cover both the tuberculin sensitivity induced and the immunogenic power. However, while the results in terms of tuber- culin sensitivity were very clear-cut, the results from the protection tests were much less precise. METHOD AND MATERIALS Design Twelve BCG strains were examined during eight uniformly designed experiments. In experiments 1 and 2, four BCG strains were examined, in experi- ment 3, three BCG strains, and in each of the five remaining experiments, five BCG strains were in- volved. Each strain was examined in two or three experiments (Table 1). For each BCG strain, doses ranging in fourfold steps from 1 to 1000 culturable particles, always in 0.1 ml (estimated from colony counts on solid medium) were inoculated into guinea-pigs by the intradermal route. Five to ten animals (equal numbers of males and females) were used for each Table 1. Strains of BCG used in the eight experiments Experiment Strain of BCG 1 2 3 4 5 6 7 8 Copenhagen 1331 x x x x x x x x Former Danish strain x x Madras 809 x x x Prague 725 x x x Paris 1173 P2 x x Moscow 3522 x x Tokyo 172 x x y x x Rio de Janeiro (Moreau) x x x Gothenburg x x x London F 10 x x x R x x x dose, and 25-50 animals for each unvaccinated placebo group. Twelve weeks after vaccination the vaccinated guinea-pigs and the unvaccinated group given pla- cebo were tested with 10 TU of tuberculin RT 23 with Tween 80. The indurations were read 24 hours later by a specially trained nurse from the Danish Institute for Clinical Epidemiology, Copenhagen. That particular time for tuberculin-testing was chosen because, as shown by Tolderlund (15) with a 1/100 000 dilution of Danish vaccine (about 20 culturable particles), maximum allergy is obtained by the twelfth week. As the studies were designed also to examine the protective effect of the vaccine, the guinea-pigs were challenged by the intraperitoneal route with virulent Mycobacterium tuberculosis one week later. Around 7-12 weeks after challenge, the animals were killed. Autopsy was performed, the spleen and omentum were weighed, and the degree of tuberculosis in the organs was determined on the basis of the macro- scopic findings. BCG strains Strain 1331, which has been used in Copenhagen and Madras since 1966, was included in all eight experiments. The other strains were those routinely used in Prague (725), Paris (1173 P2), Moscow (3522), Tokyo (172), Rio de Janeiro (Moreau), Gothenburg, and London (Glaxo FIO). The former Danish strain and the strain used in Madras (809) until 1966, as well as two experimental strains, designated Y and R, were also examined. Both the latter strains were obtained from the CIBA labora- tories, Basle. Y is derived from the " Phipps " strain that was used by Aronson (1) in a controlled trial in American Indians, and R is derived from a single colony obtained from the Tokyo strain (H. Bloch, personal communication, 1965). Vaccine preparations All strains were grown as surface cultures on Sauton medium and harvested after 10-11 days. Liquid vaccine was prepared according to the rou- tine method used in the BCG Department, Statens Seruminstitut, Copenhagen. The strains from Paris and Madras grew faster than the other strains did and showed pronounced pigmentation. The London strain grew more slowly than the others and the morphology of the surface culture was distinctly different from that of the other strains. Table 2 shows the results of in vitro tests per- 436 TUBERCULIN SENSITIVITY TO DOSES OF BCG 437 Table 2. In vitro characteristics of BCG suspensions Experi- Opacity (suspension Oxygen uptake Germination rate Culturable particles Strain mNeOnt of 0.75 mg/cm3) (I per 120 mg (percentage at 24 h) a (millions/mg)No. ~~~~~~~~perhour) Copenhagen 1331 1 0.34 118 75 21.2 2 0.38 145 75 29.2 3 0.32 171 75 19.0 4 0.40 167 75-90 21.6 5 0.34 150 75 11.4 6 0.30 150 75-(90) a 12.8 7 0.40 173 75 21.2 8 0.36 166 75 15.5 Former Danish strain 1 0.42 149 75 28.3 2 0.43 141 75 27.9 Madras 809 1 0.44 123 75-90 25.5 2 0.34 138 75-90 23.9 3 0.38 192 75-90 22.2 Prague 725 1 0.42 125 75-90 15.3 2 0.43 129 50-75 17.6 3 0.33 130 75 15.6 Paris 1173 P2 4 0.29 149 75-90 17.3 5 0.37 110 75 12.9 Moscow 3522 4 0.39 208 75-90 34.6 5 0.35 209 75 24.4 Tokyo 172 4 0.38 195 75 79.7 5 0.35 194 75-90 73.2 Y 4 0.42 143 75 16.3 5 0.38 134 75 15.3 Rio de Janeiro (Moreau) 6 0.32 183 75-90 17.4 7 0.41 218 75-90 28.7 8 0.41 155 75 22.9 Gothenburg 6 0.30 182 75 39.0 7 0.40 166 75 48.0 8 0.37 188 75 29.7 London F 10 6 0.32 90 75 5.0 7 0.36 84 50 7.1 8 0.36 83 50-75 6.2 R 6 0.33 190 75 25.6 7 0.34 174 75 43.4 8 0.40 161 75 25.3 a The result is considered to be closer to the figure not in parenthesis. 438 A. LADEFOGED ET AL. formed on the vaccines from the various strains. The strain from Tokyo yielded vaccines with a very high colony count and also the strain from Gothenburg showed a rather high colony count. The London strain yielded vaccines with a lower proportion of live organisms, as shown by the oxygen uptake, germina- tion rate, and colony count. Otherwise, the vaccines from the various strains were much alike in a number of in vitro tests. Animals The guinea-pigs were 9-11 weeks old at the timne of vaccination. The animals were, separately for each experiment and each sex, randomly allocated to the different strains and doses ofBCG and to a placebo. Three guinea-pigs of the same sex were placed in each cage, and the cages were placed in the animal house in a formally randomized order. Follow-up and analysis Of the 3100 guinea-pigs included in the experi- ments, 2626 were vaccinated and 474 were given placebo only. Between vaccination and tuberculin- testing, 78 guinea-pigs died and have been excluded from the analysis: 62 in the vaccinated groups and 16 in the placebo groups. The tuberculin reactions for the individual animals are given in the Appendix tables a. RESULTS The results are illustrated in Fig. 1-8 in terms of mean tuberculin reactions as a function of vaccine dose in nanograms of semidry weight. (Dosages in culturable particles can be derived from Table 2.) With very low dosages, around 0.5 ng (5-20 cultur- able particles), the strains are seen to differ. For dosages around 5 ng (50-200 culturable particles) the curves tend to level off, showing maximum reactions, except for the Prague strain (725), which shows only a slight increase in the tuberculin reactions even with a dose as high as 30-60 ng (around 500-1000 culturable particles). The consistency of the experiments can be exam- ined by comparing the graphs (Fig. 1-8). In experi- ments 1 and 2, dealing with the same 4 strains, and in experiment 3, in which 3 of these strains are included, the ranking of the strains is consistent, the a The Appendix tables have been deposited in the WHO Library, and single copies may be obtained on request to: Chief Librarian, World Health Organization, 1211 Geneva 27, Switzerland. Madras strain (809) being the strongest and the Prague strain (725) the weakest. Strain 1331 and the former Danish strain appear to be slightly weaker than Madras 809 and to an equal extent. Of the 5 strains tested in experiments 4 and 5, those from Moscow, Paris, and Copenhagen seem to be almost equal in strength; the Tokyo strain is weaker, and strain Y the weakest. In experiments 6, 7, and 8, the Rio de Janeiro, Gothenburg, and Copenhagen strains are the strongest. The Glaxo strain shows the weakest reac- tions and the R strain gives conflicting results. In two of the experiments-6 and 8-the curves for the R strain are similar to those obtained for the Tokyo strain in experiments 4 and 5. From experiment 7 it is obvious that a much larger dose (at least 8 times as large) of the R strain is needed for obtaining maxi- mum reactions. Similar inconsistency in the R strain was observed in our experiments in the bank vole (12). There is no obvious explanation of the erratic behaviour of this strain. Results in terms ofprotection The protection tests gave much more variable results than the tuberculin tests. In some cases the animals were killed too early, and the tuberculous disease in the control animals was only slight. In other cases the animals were killed too late, and the disease was too severe. The strain used for challenge was apparently attenuated and the challenge dosage was not under sufficient control. The results (not shown), while very imprecise, did not contradict those reported here: guinea-pigs with increased tuberculin reactions after vaccination all showed some degree of protection compared with unvac- cinated guinea-pigs. In an examination of four different strains in the same model, Jespersen & Bentzon (19) found an association between tuberculin sensitivity and immu- nogenic potency, and concluded that the former can be used as an indication of the latter in this model. DISCUSSION The guinea-pig in quality control ofBCG The minimum sensitizing dose varied, among the strains examined, from 10 to 1000 culturable par- ticles-even disregarding the weakest of the strains examined. Above this minimum dose, the response in guinea-pigs is dose-independent, in contrast to man, in whom (as already mentioned) the response to BCG is quantitative and dose-dependent. It is a Fig. 1. Tuberculin reactions by strain and dose of BCG (experiment 1) DOSE OF BCG (ng) Fig. 2. Tuberculin reactions by strain and dose of BCG (experiment 2) v -.. COPENHAGEN 1337 0----O MADRAS 809 - - -- PRAGUE 725 , / U __ UN~~VACCINA TED ANIMA LS I I I lI I 0.03 0.12 0.48 1.9 7.5 30 120 DOSE OFBCG (ng) WBO M304 Fig. 3. Tuberculin reactions by strain and dose of BCG (experiment 3) Fig. 4. Tuberculin reactions by strain and dose of BCG (experiment 4) 14 I14 -E12 0 U Z1o 6) 4 U~ 0 LU Nz Z12 Z U 10 N LUt 0 CC 6IQ N 4 Lz 2 f* - * COPENHAGEN 1331 .-e-.-e PARIS 1173P2 *- -4MOSCOW 3522 +i-1+-iIH+44-4 TOKYO172 V_r-tAWWI_Y UNVACC/NA TED ANIMAALS I I-I 1- I 1 2)3 0.12 0.48 1.9 7.5 30 1200.( DOSE OFBCG (ng) IF110 76.11 14 12 QC cc 1 L 8N 6 ; U 2 0.03 0.12 0.48 1.9 7.5 DOSE OF BCG (ng) Fig. 5. Tuberculin reactions by strain and dose of BCG (experiment 5) - - COPENHAGEN 1331 -ee-o-oRoo/RIODE JANEIRO (MOREAU) 14 a---O GOTHENBL'RG -&- LONDON FIO C< U- UN VACCINATED ANIMAIALS 21 0.03 0.12 0-48 1.9 7.5 30 120 DOSE OF BCG (ng) WHO 76312 Fig. 6. Tuberculin reactions by strain and dose of BCG (experiment 6) 141- -12 c 10 N K8 K N4 O Q~ 2 C u 2 0.03 0.12 0.48 1.9 7.5 DOSE OF BCG (ng) 30 120 14110 761rI Fig. 8. Tuberculin reactions by strain and dose of BCG (experiment 8) * w COPENHAGEN 1331 e Oe O a PARIS1173 P2 * .- - * M41OSCOI 3522 i+i+41wel+i TOKYOI1;2 .,,, 0-- UNVACC/IA TED ANIMA L S 30 120 ?WHO 7631 LCOPENHIAGEN 1331 a- a-e--e- RIOODEJ1\NEIRO(MORLA' O-- -E COTHENBURG - 0. - 0 ~~~~~~-I.. / ^ - / 0 UNV ACCINA TED ANIMALS Fig. 7. Tuberculin reactions by strain and dose of BCG (experiment 7) _ . w TUBERCULIN SENSITIVITY TO DOSES OF BCG reasonable inference from the findings that, if the initial dose is sufficient to start multiplication at all, BCG multiplies freely in the guinea-pig (more so than in man) until the maximum immune effect is reached. In many laboratories, a " test of skin reactivity in guinea-pigs" is carried out for every batch of vac- cine, including the use of a human intradermal dose of BCG (18). In some laboratories, the same animals (or even animals given 50 human doses each, for the purpose of the safety test) are further tested with tuberculin after an interval so as to demonstrate the allergenic potency of the vaccine. It appears from the present results, as well as from earlier publications, that as a routine such a tuberculin test is meaningless because of the high BCG dosage. One could imagine instead a test of the " smallest effective allergenic dose ", but to carry out a test for every batch would in most cases be prohibitively expensive; for an otherwise well-examined strain it may indeed be merely a particularly expensive way of examining the viability of the batch. It is therefore proposed to use the model of the smallest effective dose only, as at present, for comparisons of strains-that is, seed- lots. Ranking of strains In 1956 Dubos (5) advanced the hypothesis that "invasiveness -i.e., virulence as expressed by the ability of a BCG strain to multiply in the organs of the mouse-was associated with protective effect against virulent challenge. This hypothesis was con- firmed in 1961 by Villis & Vandiviere (17) by a retrospective comparison of the effect of various strains in animal models, with controlled trials in human beings of three of these strains. A further discussion of this evidence was presented by Guld in 1971 (8). On the basis of all the data from our experiments in four different models (see also 3, 4, 12, 16), it should now be possible to rank the strains according to in vivo activity, or virulence, in mammalian hosts. A strict ranking is not entirely possible because only one strain (1331) has been used consistently through- out: the choice of the other strains for the various experiments rather reflects the changing emphasis on the importance and promise held out by one or another strain-an emphasis that changed with time as experimental results accumulated, but also with developments in production practices. We have not attempted a strictly quantitative biometrical analysis, but should like to offer the following scoring and ranking, based on an inspection of the graphic presentations only. Rio de Janeiro (Moreau) Paris (1173 P2) Copenhagen (1331) Moscow Gothenburg Madras (former strain) Copenhagen (former strain) Tokyo R y London (Glaxo) Prague Ham- Bank Guinea- Chil- Mean sters voles pigs dren score 5 4 4 3 4 5 5 4 4 5 4 4 4 4 3 5 4 4 5 3 4.8 4.3 4.0 4.0 3.8 1 5 5 3 3.5 3 3 3 2 2 1 2 1-4 4 2 2 4 2 1-4 1 2 0 3 5 3 2 1 (3.3) 3.0 (2.7) 2.5 2.0 1.0 The mutual ranking of about half of these strains (those listed first), might very likely shift if yet another model were introduced, and that of strains close in ranking might even shift if the experiments were repeated. For three of the more active strains, Paris, Copen- hagen, and Moscow, there has been wide practical experience over a number of years. For the Moreau strain, scoring uniformly high in all models, there has been unfortunately only scanty experience in man with vaccines of high viability. The Paris and Copenhagen strains have given rise from time to time to complaints about complications (regional lymphadenitis) nearly always in the new-born and associated with an unintentionally high dosage. While suppurative lymphadenitis after BCG is en- tirely benign and self-healing, a sudden increase in its frequency tends to upset both the public and the medical profession and thus may endanger the acceptability of a programme. On the other hand, its occurrence may be taken as yet another piece of evidence, in yet another " in vivo model ", that these are active, " invasive " strains in the sense meant by Dubos. The Gothenburg strain has been used in the new- born and in infants in Sweden and Finland for many years, with an obvious protective effect, as reported by Bjartveit & Waaler (2). But in recent years there have also been reports of complicating osteomyelitis (G. Dahlstrbm, unpublished observations, 1976) that is not entirely benign and needs treatment. Although rare, this complication is not negligible when vac- cination of this age group is universal and infant tuberculosis is rare. Whether it is specific for the 441 A. LADEFOGED ET AL. Gothenburg strain is not known, since no other strain has been used as widely in the new-born in an area with well developed health services. The strain formerly used in Copenhagen is of no particular interest compared with 1331, which was frozen as a seed-lot earlier. The strain formerly used in Madras, which scores low in hamsters and high in bank voles and guinea-pigs, is a striking example of disagreement between models. That the London strain scores rather low through- out these studies is not an unambiguous finding. It was prepared for these experiments with a technique (surface growth) to which it is not adapted, and its viability was obviously lower (perhaps by 50 %) than that which would have been obtained with the deep culture technique normally used for it. But even allowing for this the strain still appears to be rather low in rank. The Tokyo strain is perhaps more interesting. While second to no strain in inducing delayed hypersensitivity in children, it is definitely weak in bank voles and guinea-pigs. Actually, the response in guinea-pigs would have appeared even weaker had it been presented in terms of culturable particles and not moist weight. Whereas, for the Paris strain, the lowest dose used-approximately 7 culturable particles-gave an average tuberculin reaction size of 23 mm, it took approximately 70 culturable particles of the Tokyo strain to give an average of 18 mm. The London and the Tokyo strains have both been widely used over the past decade, because 15 years ago they happened to be available as the first freeze-dried vaccines of high stability. The Y and R strains are of historical interest only. The former is a laboratory strain derived from a vaccine that gave high protection in man 40 years ago, in Aronson's trial in American Indians (1), but there is no evidence that the strain has not become less active over the years. The R strain gave highly variable results. While derived from a single colony, it appears not to be genetically homogeneous. The Prague strain is an interesting example of low allergenicity combined with low immune effect. A similar association was reported by Engbaek et al. (7) and by Mackaness (13). Admittedly, the lack of complete agreement be- tween different animal models raises the question of the validity of each model for man. However, taken together and as far as they agree, they can scarcely be completely disregarded in selecting a BCG strain for production. No doubt better animal models are desirable. An elegant model has quite recently been proposed by Fok et al. (14), using respiratory challenge with so few viable units that not all lung lobes are involved in the primary infection. By counting the number of infected lobes after a suitable interval, the prevention of haematogenous spread to other lobes (which is the direct effect of BCG, also in man) can be directly observed. The use of this technique in an animal with approximately the same susceptibility as man to M. tuberculosis and to BCG (e.g., the white mouse and, if possible, also a non-rodent) would intuitively appear more valid than any other protection test in a particular animal and, it is to be hoped, also more relevant for man. RItSUMIt SENSIBILITE DES COBAYES A LA TUBERCULINE APRES VACCINATION PAR DIVERSES DOSES DE BCG PROVENANT DE 12 SOUCHES DIFFERENTES Dans plusieurs dtudes rapportees anterieurement, un certain nombre de souches de BCG, et notamment celles qui sont le plus couramment utilisees pour la production de vaccin, ont ete classees en fonction de leur activite in vivo chez les rongeurs dans differents modeles exp6rimentaux, et en fonction de la reponse locale et allergique chez les enfants. Dans le present travail, 12 souches sont classees sur la base du virage de la reaction tuberculinique chez le cobaye, selon la dose sensibilisante minimale. La souche 1331, employee A Copenhague et A Madras depuis 1966, a figure dans toutes les experiences. Les autres souches etaient celles qui sont couramment utilisees A Prague (725), A Paris (1173 P2), A Moscou (3522), A Tokyo (172), A Rio de Janeiro (Moreau), A Gothenbourg et A Londres (Glaxo FIO). L'ancienne souche danoise et la souche utilisee A Madras (809) jusqu'en 1966 ont aussi ete examinees ainsi que deux souches experimen- tales designees par Y et R, ces deux dernieres provenant des laboratoires CIBA de BAle. Toutes les souches ont et6 cultivees en surface sur milieu de Sauton et recoltees au bout de 10 A 11 jours. Le vaccin liquide a ete prepare selon la methode habi- tuellement utilisee au Departement du BCG, Statens Seruminstitut, Copenhague. Les souches de Paris et de Madras ont pousse plus rapidement que les autres AA'2 TUBERCULIN SENSITIVITY TO DOSES OF BCG 443 et presentaient une pigmentation prononcee. La souche de Londres a pousse moins vite que les autres, et la morphologie de la culture superficielle etait nettement differente de celle des autres souches. Le tableau 2 montre les r6sultats d'un certain nombre d'epreuves in vitro effectu&es sur les vaccins provenant des differentes souches. Pour chaque souche de BCG, on a inocul6 A des cobayes, par voie intradermique, des doses croissant de 4 fois en 4 fois, i partir de 1 A 1000 particules culti- vables, sous le volume constant de 0,1 ml (les estimations etaient faites d'aprbs les num6rations des colonies sur milieu solide). On a utilise 5 A 10 animaux (nombre egal de males et de femelles) pour chaque dose et 25 A 50 animaux pour chaque groupe non vaccine recevant un placebo. Douze semaines apres la vaccination, les cobayes vaccines et ceux qui avaient recu le placebo ont ete 6prouv6s par 10 unites de tuberculine RT 23 additionnee de Tween 80. Les resultats des diverses experiences sont representes dans les figures 1 a 8. Pour 10 des 12 souches etudiees, la dose minimale sensibilisante est extremement faible, allant de 5 a 50 particules cultivables. Ainsi, la methode traditionnelle consistant a eprouver a la tuberculine les cobayes vaccines avec la totalite d'une dose destin6e A l'homme (qui est de l'ordre de 1 million de particules cultivables) a un pouvoir de discrimination tellement faible qu'elle est inutile en tant qu'epreuve de routine pour les vaccins utilises actuellement. Le classement obtenu avec ce modele concordait en grande partie, mais pas totalement, avec des classe- ments identiques etablis avec d'autres modeles. Le rang de la souche de Tokyo, et, dans une certaine mesure, celui de la souche de Londres, ont ete nettement plus bas d'apres ce modele que d'apres leur effet chez l'enfant. L'avirulence prononcee de la souche de Prague a ete confirmee. REFERENCES 1. ARONSON, J. D. ET AL. Archives ofinternal medicine, 101: 881 (1958). 2. BJARTVEIT, K. & WAALER, H. Bulletin of the World Health Organization, 33: 289 (1965). 3. BUNCH-CHRISTENSEN, K. ET AL. Bulletin ofthe World Health Organization, 39: 821 (1968). 4. BUNCH-CHRISTENSEN, K. ET AL. Bulletin ofthe World Health Organization, 43: 65 (1970). 5. DUBOS, R. J. & PIERCE, C. H. American review of tuberculosis and pulmonary diseases, 74: 699 (1956). 6. EDWARDS, L. B. ET AL. BCG vaccination. Geneva, World Health Organization, 1953, pp. 125-128 (Monograph series, No. 12). 7. ENGBEK, H. C. ET AL. Acta tuberculosea et pneumo- logica Scandinavica, Suppl. LVI: 114-117 (1964). 8. GULD, J. In: Immunization in tuberculosis. Bethesda, U.S. Department of Health Education & Welfare, 1971, pp. 149-156 (DAEW publication No. (NIH) 72-68). 9. JESPERSEN, A. Acta pathologica et microbiologica Scandinavica, 35: 396 (1954). 10. JESPERSEN, A. Acta pathologica et microbiologica Scandinavica, 38: 203 (1956). 11. JESPERSEN, A. & BENTZON, M. W. Acta tuberculosea et pneumologica Scandinavica, 44: 253, 276 (1964). 12. LADEFOGED, A. ET AL. Bulletin of the World Health Organization, 43: 71 (1970). 13. MACKANESS, G. B. American review of respiratory diseases, 97: 337-344 (1967). 14. FOK, J. S. ET AL. Journal of infectious diseases, 133: 137-144 (1976). 15. TOLDERLUND, K. ET AL. Bulletin of the World Health Organization, 22: 185 (1960). 16. VALLISHAYEE, R. S. ET AL. Bulletin ofthe WorldHealth Organization, 51: 489 (1974). 17. WILLIS, S. & VANDIVIERE, M. American review of respiratory diseases, 84: 288 (1961). 18. WHO EXPERT COMMrIEE ON BIOLOGICAL STANDAR- DIZATION. Eighteenth report. Geneva, 1966, p. 42 WHO Technical report series, No. 329). 19. JESPERSEN, A. & BENZTON, M. W. Acta pathologica et microbiologica scandinavica, 71: 114-131 (1967).