(BCG) in Mice - NCBI - NIH

INFECTION AND IMMUNITY, Nov. 1973, P. 736-742 Copyright 0 1973 American Society for Microbiology

Vol. 8, No. 5 Printed in U.S.A.

Virulence of Six Strains of Mycobacterium bovis (BCG) in Mice NEAL A. SHER, SOTIROS D. CHAPARAS, JACK PEARSON, AND MICHAEL CHIRIGOS Division of Bacterial Products, Bureau of Biologics, Food and Drug Administration, Rockville, Maryland 20852, and National Cancer Institute, Bethesda, Maryland 20014 Received for publication 27 June 1973

The abilities of six strains of Mycobacterium bovis strain BCG to replicate in vivo and to cause spleen enlargement in the CDF, male mouse were compared. The strains were grown in synthetic medium and harvested soon after complete pellicles were produced on liquid medium. In addition, two freeze-dried commercially available preparations were tested. The comparisons were made on the basis of injecting equal numbers of colony-forming units. Strain differences were marked. The Brazil strain was most effective in stimulating spleen enlargement, and the Japan strain was the least effective. This correlated with the number of colony-forming units isolated from the spleen.

Bacillus Calmette-Guerin (BCG) has been established as one of the most effective stimulators of the reticuloendothelial system (RES). In this regard, it has shown considerable promise in the treatment of experimental tumors in animals (17, 25) and in man (5, 8, 12, 14, 15). The antigenic content of BCG vaccines produced in different laboratories throughout the world has been reported to be remarkably stable (3), but differences have appeared in cultural, morphological, biochemical, and immunological characteristics, as well as in ability to induce resistance to tuberculous infection (19, 21). These differences in protection to mycobacterial infection could not be correlated with either skin sensitivity or antigenic content of the BCG. Virulence may be one of the more important differences with regard to the ability of BCG strains to induce RES stimulation. It is this type of stimulation of cellular immunity which provides the bulk of protection against tuberculosis and neoplasms. In the past, various parameters to measure virulence have been used. Jensen (9) in 1946 used an intracutaneous injection (in milligrams) of four doses of vaccines on the abdomen of guinea pigs. The size of induration and the length of reaction time were recorded. Levy (13) compiled a method permitting correction for the effect of killed BCG by comparing skin reactions of viable BCG to that of contralateral reactions using only killed bacteria. Panisset and Marois (16) measured the weight of guinea pig omentum after an intraperitoneal dose of BCG. Other investigators looked for macro-

scopic lesions on autopsy after injection by various routes (2, 6). Utilizing the fact that the golden hamster Mesocrisetus auratus is more sensitive to BCG infection than most laboratory animals, Jespersen and others have used the survival of this animal after infection as an indicator of virulence (7, 10). The "organ culture" methods modified for use in our studies were first developed by Pierce and Dubos (18, 19). They found that, after suitable doses of BCG were injected into groups of mice, bacteria in the various organs could be enumerated. The rate of multiplication can be measured as well as the persistence of the live organisms in the tissues. The newer applications of BCG vaccines as nonspecific stimulators of the RES require means of evaluating and comparing efficacy of the various strains. More information is required on identically produced vaccines which are compared on the basis of inoculation with equal numbers of viable organisms. The lack of agreement in many of the recent reports concerning the use of BCG in tumor immunotherapy may have been due to strain differences or to different preparation and application of the organisms (5, 14). In the present investigations, the virulence of six strains of BCG that have been used in various parts of the world for the prevention of tuberculosis and for tumor immunotherapy was assessed under identical conditions.

736

MATERIALS AND METHODS Animals. Adult CDF, male mice, 6 to 8 weeks old,

VOL. 8, 1973

obtained from the National Institutes of Health (NIH) Animal Production Section. The animals were housed in plastic cages and fed Purina Laboratory Chow with water ad libitum. Animals weighed between 23 and 28 g before being given 1, 3-bis (2-chloroethyl)-1-nitrosourea (BCNU). Drug. BCNU was kindly supplied by the Drug Development Branch, National Cancer Institute, Bethesda, Md. The alkylating agent was dissolved in a steroid-suspending vehicle and administered in the scapular area subcutaneously (s.c.) in a dose of 30 mg/kg. Preparation of BCG vaccines. The Brazil, Japan, Pasteur, Phipps, Tice, and Glaxo strains were obtained from the Trudeau Institute, Saranac Lake, N.Y. They are described in the Department of Health, Education, and Welfare Publication no. (NIH) N-289. The cultures were transferred three times in order to build up seed material. They were then grown in Long synthetic medium at 35 C. Soon after pellicle formation (about 4 weeks), cultures were dispensed into single-cell suspensions in Youman modified media with 0.05% Tween 80 by gentle grinding with a Teflon grinder on an electric motor. The organisms were dispensed into vials, slowly frozen, and stored at 0 to 70 C. Two commercially available freeze-dried vaccines were also tested. The Tice vaccine, lot no. T94(S)74(S)10, was obtained from the Research Foundation and University of Illinois Medical Center, Chicago, Ill., in the freeze-dried form. The Glaxo vaccine, lot no. B77, was obtained from Glaxo Laboratories, Ltd., Greenford, Middlesex, United Kingdom. These will be referred to as Tice or Glaxo (commercial) to distinguish them from similar strains grown in our laboratory. Determination of colony-forming units. All colony counts were performed in triplicate and rechecked every 3 months by the method of Rosenthal et al. (22). Total bacterial counts were performed in a PetroffHausser Counter (C. A. Hausser and Sons, Philadelphia, Pa.) by using phase microscopy. Inoculation of organisms. Desired concentrations of BCG were given in 0.1 or 0.2 ml of phosphate buffered saline, intravenously (i.v.) in the tail with a no. 26-gauge needle in a 1-ml tuberculin syringe. Intradermal (i.d.) inoculations were administered in the flank. Determination of spleen CFU. Mice were sacrificed by cervical dislocation. Spleens were removed aseptically, trimmed of fat, weighed, and ground at a constant speed in 5 ml of 1.9% sodium glutamate for 50 s in a Teflon tissue grinder on an electric motor. Total BCG cell counts in spleens were performed by a tube dilution technique with the spleen suspensions using modified Youman media with 0.5% bovine albumin (22). After the dilutions were made, 0.5 ml of 1.5% liquid agar was injected into each tube to semisolidify the media. The tubes were read 3 to 4 weeks after incubation at 35 C, and total spleen colony-forming units (CFU) and CFU per milligram of spleen were determined. Unless otherwise indicated, all values represent the mean of four to five animals. Statistics. The means were arithmetic. The P were

737

M. BOVIS VIRULENCE IN MICE

values were based on Student's t test (one sided), and a value of P > 0.05 was considered non-significant.

RESULTS Number of organisms in stock vaccines. The range of CFU was 1.2 x 10 to 3.1 x 108/ml (Table 1). The total number of organisms was generally one to two logs higher in number than the viable organisms. The ratios are listed in Table 1. The concentrations of the commercial Tice and Glaxo strains were 2.0 x 108 and 6.7 x 106 CFU/ml, respectively. Effect of grinding time on CFU recovery. It is most probable that the organisms in the spleen are in clumps and that the liberation of single organisms depends directly on the grinding time. To demonstrate this, 107 CFU of Brazil BCG was injected i.v. in 0.1 ml into CDF, male mice. At day 13 after injection, weighed portions of the spleens of these animals were ground for 15, 50, or 300 s, and the CFU per milligram of spleen were determined. A linear relationship was observed for times up to 300 s. All CFU determinations in this paper were values obtained after 50 s of grinding at constant revolutions per minute (see Fig. 1). Host response to BCG inoculation. The Brazil strain at a dose of 2 x 106 CFU/0.1 ml was injected i.v. on day zero into 60 mice. The diluent (phosphate-buffered saline [PBS], 0.1 ml) was injected i.v. as a control. At various intervals groups of five mice were sacrificed, and spleen weight and CFU per milligram of spleen were determined on groups of five mice (Fig. 2). The number of CFU reached a peak at around day 12 and then had a rapid fall and a leveling off. The spleen weight reached a peak at 3 to 4 weeks, fell somewhat, but remained elevated for up to 4 months, at which time the study was terminated. A similar experiment was performed with the Phipps strain of BCG (Fig. 3). Similar curves were produced. However, the absolute numbers of CFU were lower. Two million heat-killed TABLE 1. Total and viable organisms per milliliter-BCG stock vaccine. Strain

Brazil Glaxo Japan Pasteur Phipps Tice

Total organisms/ml 6.0 6.2 8.2 1.0 8.3 3.7

x x x x x x

1010 1010 1010

10" 1010 1010

Viable (CFU)

Ratio (total/viable)

x x x x x x

194 36 68 238 69 100

3.1 1.7 1.2 4.2 1.2 3.7

108 109 109

108

10'9

10'

738

SHER ET AL.

strains of BCG, doses of 103 to 107 CFU in 0.1 ml were given i.v. On days 11 to 13, groups of five mice were sacrificed, and spleen weights determined. Figure 4 summarizes these results and shows the differences among these strains. The Brazil BCG caused the greatest increase in spleen weight, and the difference was most marked at the higher doses. The Japan strain was the least effective and barely increased the weight over that of a saline control. The Pasteur and Tice strains were very similar in response. There were no deaths, even at the higher dose

4 0 z

INFECT. IMMUNITY

3

LJ a-

(n) E 2 "I

levels. Correlation of spleen weight and CFU. In another experiment, 2 x 105 CFU in 0.2 ml of , , , 300 PBS was injected into groups of mice (Table 2). 240 180 120 60 0 At Day 13, the mice receiving the Brazil strain GRINDING TIME (seconds) FiX 1. Effect of grinding times on recoverable were the only group with significantly enlarged colonty-forming units (CFO from the spleen stan- spleens (P < 0.05) (Table 2). However, there was a wide variation in the number of CFU per dard deviation. spleen. The number of CFU per milligram of SPLEEN WEIGHT vs DAYS AFTER spleen ranged from 4,538 in the Brazil strain to 2.0 x 06 BRAZIL IV. DAY 0 307 in the Japan strain. 00 _ w W In another group of CDF1 male mice, doses of E 8 x 106 CFU in 0.2 ml of PBS were injected i.v. Do _ / ~+ x 22x106 BCG \ 0 or i.d. BCNU, an alkylating agent, was used in S some groups in a dosage of 30 mg/kg, s.c., 6 days oo _ ,[ j + 95 before BCG inoculation. Skin transplantation z and other data have shown that BCNU is imoo / -J 0. (I, munosuppressive, especially to cellular immuSALINE CONTROL oo $ - -even at 6 days after administration to the nity, CDF, mouse in the dose we were utilizing (M. , Chirigos, unpublished data). BCNU has been COLONY-FORMING UNITS/mg SPLEEN vs DAYS used in our laboratory to control LSTRA, a transplantable Moloney virus-induced leuISO(00 _ // Xkemia used to test various immunotherapy z w 120i°o regimens (17) (Table 3). There was a diminu-J tion in spleen weight after BCNU therapy in the a 0. 90(00 control groups when the animals were sacrificed E at Day 13 (Table 3). The spleen weight inU. 601 00 r creases in the BCG-treated groups were also almost by 60% in some cases. As seen reduced, 301oo 00 --- -~ ~ previously in the groups without drugs, the , , ,y, Brazil strain induced the largest increase in 115 0 10 20 30 40 50 60 70 spleen weight, and the Japan strain induced the B

,

nitst

DAYS

inject FiI G. 2. Effect of 2 x 106 CFUof Brazil BCG, i.v., on splee?n weight (- standard deviation) and on recoverable CFU per milligram of spleen. Injected on day 0. Each determination is the mean offive animals.

Phip ps BCG cells (100 C, 30 min) had no effect s pleen weight. There was no morbidity or morttality in both the Brazil and Phipps BCG stud ies. Ef!ect of six BCG strains on spleen weight. To ccompare the relative virulence of the six

on

smallest. The differences cant.

were

highly signifi-

The number of CFU per milligram of spleen correlated with the spleen weight. The Brazil strain again had the most recoverable units at day 13 and had an almost two log10 increase over the Japan strain. The i.d. route, although considered less effective in bringing organisms to the spleen, showed a significantly higher number of CFU after inoculation with the Brazil strain as compared to the Japan strain (P < 0.01). As expected, the organisms multiplied more

739

M. BOVIS VIRULENCE IN MICE

VOL. 8, 1973

SPLEEN WEIGHT vs DAYS AFTER 2.0 x 106 PHIPPS BCG I.V., DAY 0

0%

300

E -

x 106 viable BCG

200

3 z

w

aLO

Control 100

2 x16heat killed BCG

2000r 1600

K

1200

-u

z

w w J

cn

cp

E

,q

COLONY-FORMING UNITS/mg SPLEEN vs DAYS

/

"w, U_

uu- 8001_ 1-U

400'

0

10

20

30

50 40 DAYS

60

90

100

110

inec FIG. 3. Effect of 2 x 106 CFU of Phipps BCG, i.v., on spleen weight (± standard deviation) and on recoverable CFUper milligram of spleen. Injected on day 0. Each determination is the mean of five animals.

rapidly in the animals pretreated with BCNU. There was an increase in all the strains studied, but the weakly virulent Japan strain had the largest increase from 250 to 37,000 CFU/mg spleen weight. The only observed toxicity from the administration of the drug alone was some weight loss (less than 3 g). This was usually regained in 2 to 3 weeks.

DISCUSSION Enumeration of BCG organisms in the spleen after intravenous inoculation provided a quantitative, reproducible method of assessing virulence of different strains for the CDF, mouse. Pierce et al. has compared virulence of strains by this method and has found constant differences over intervals of years (18). Jespersen (11), in an extensive monograph on BCG, has discussed drawbacks in the organ

culture method of determining virulence. He correctly pointed out that, after a BCG suspension is injected intravenously, the majority of the bacteria were removed rapidly from the blood and were taken up in the lungs, spleen, and other organs containing RES components. Mechanical disruption of the organ would not liberate all the bacteria from the cells or separate all the clumps into single bacterium. However, if vaccines produced under identical conditions are used, although the grinding time of 50 s did not release the maximal amount of viable organisms from the spleen, the linear relationship (Fig. 1) made the comparison at this lower level valid. On the basis of these studies, one could divide the strains into a high, intermediate, and low category of virulence. The Brazil strain and possibly the Phipps were classified in the highvirulence group. The Japan strain was in the

740

SHER ET AdL.

INFECT. IMMUNITY

500 -*- Brazil Phipps -*- Glaxo Tice

400

---O--- Pasteur

Japan Control

E

M-I 3 300 z

-J

a-

(n

200

100

3

4

6 5 LOG DOSE IN 0.1 cc IV

7

FIG. 4. Effect of six BCG strains on spleen weight. Control animals received 0.1 ml of saline. TABLE 2. Effect of five different BCG strains on spleen weight and colony-forming units in BCG strain

Control Brazil Pasteur Glaxo (commercial) Tice (commercial) Japan

CDF, micea

Total spleen CFU + SEb

CFU/mg spleen

(mg)

± SEb

Total spleen CFU/ CFU injected

108 4 3.8 208 i 22.1 110 4 3.4 120 i 4.1 107 ± 3.7 107 i 6.5

950,000 i 160,000 234,000 i 28,213 166,000 ± 23,151 148,000 ± 17,435 31,600 i 2,400

4,538 i 967 2,082 ± 283 1,390 i 220 1,370 ± 290 307 ± 29

4.76 1.18 0.83 0.74 0.16

Spleen wt. + SE

All groups received 2 x 105 CFU in 0.2 ml of PBS on day 0. Control received 0.2 ml of PBS alone. For both total spleen CFU and CFU per milligram of spleen, Brazil versus Glaxo, Glaxo or Tice versus Japan were significant (P < .01). Brazil versus Pasteur was significant (P < .05). Pasteur versus Tice was not significant. Abbreviation: SE, standard error. a "

low category. The other strains could not be methods, some workers have shown the Brazil differentiated from each other by these methods strain to be of low virulence (18, 23, 24). To the and fall into a middle range. The ranking of contrary, Jespersen (11) classified the same these strains was in the same order when spleen Brazil strain as one of the high-virulence weight assays were performed in the Fischer rat strains. In the spleen, the multiplication rates and the and the AKR mouse (N. A. Sher, unpublished data). The literature shows a bewildering array cellular reactions may be explained in a number of conflicting reports which failed to reach a of ways. The low-virulence strains may be consensus. For example, utilizing a variety of inhibited in their multiplication because they

741

M. BOVIS VIRULENCE IN MICE

VOL. 8, 1973

TABLE 3. Effect of different BCG strains on spleen weights and viable organisms from spleens in the CDF, mouse with or without pretreatment with BCNUa Group

BCG strain

1 2 3 4 5 6 7 8 9 10

Control Control Brazil Brazil Pasteur Pasteur Japan

Japan Brazil Japan

BCNU6, 30 mg/kg

Route (BCG)c

Spleen Wt.

i.v. Saline i.v. Saline i.v. i.v. i.v. i.v. i.v. i.v. i.d. i.d.

108 ± 8.5 78 ± 4.8 544 ± 35.0 313 ± 30.5 508 46.0 293 13.0 126 ± 32.0 89 ± 8.0 109 ± 15.4 89± 5.8

Total spleen CFU ±i- SEe

SE (Mg)d Dy7 S..(a±) No Yes No Yes No Yes No Yes Yes Yes

8.45 x 5.75 x 1.46 x 3.14 x 2.71 x 1.97 x 1.20 x 2.90 x

106 ± 1.36 x 106 106 ± 8.53 x 105 106 2.02 x 105 106 ± 6.0 x 105 104 ± 1.73 x 103 106 ± 2.76 x 105 104 ± 1.08 X 103 103 ± 5.56 x 102

p

NS

p < 05 P < 01 . P O 0.05). 30 mg/kg, s.c. on day 0. Dose: 8 x 106 CFU/0.2 ml. d Without BCNU, spleen weight: Brazil versus Pasteur, Brazil versus Japan, Pasteur versus Japan, P < 0.01. 'Total spleen CFU and CFU per milligram without BCNU: Brazil versus Pasteur, Brazil versus Japan, and Pasteur versus Japan, all significant P < 0.01. a

b BCNU,

c

produced a cellular response in the host which destroyed the viable mycobacteria. These less virulent mycobacteria may induce a decreased cellular response which was observed by the smaller spleen weights compared to those induced by the more virulent strains. The strains of greater virulence may more closely approximate the course of events observed with virulent strains of Mycobacterium tuberculosis. The multiplication rate may be higher and, although greater cellular reactions may be produced, the organisms may also be difficult to kill intracellularly (20). The increase in spleen weight was not due to the weight of the bacteria. There was some evidence that the increased spleen weight was due, in part, to cellular infiltration and cellular replication. These mechanisms are currently under study, but preliminary histology of various enlarged spleens in the BCG inoculated animals at 2 weeks after injection revealed a diffuse increase in cellularity of all elements of the spleen, including the erythroid and reticular elements, as well as macrophages and megakaryocytes. No differences due to BCG strain could be detected in macro or microscopic examination of equally enlarged spleens. In these studies, the more virulent strain induced a greater

The number of viable organisms reached a peak around 2 weeks and has already begun to decrease before the maximal spleen weight is reached with its augmented cellular components. Blanden et al. (1) has confirmed that BCG multiplication in the mouse spleen after intravenous injection reached a peak at about day 12 and fell off. Nonspecific RES stimulation, as measured by resistance to Listeria monocytogenes, peaked at around day 12 as well and also fell off gradually. This may be accounted for by a probable peak RES stimulation at a time when the number of viable organisms reached a peak and then declined in the face of augmented immunity. The data suggested that the continued elevated spleen weights may have been stimulated by a residual number of organisms which persisted at decreased levels in the spleen for long periods of time (months). Collins (4) has suggested that high levels of acquired resistance persisted only as long as the infecting organisms survived in vivo. It is not certain whether spleen weights would decrease upon chemotherapy to eliminate residual organisms. BCNU has been used in the treatment of various forms of neoplasia. In addition to destroying tumor cells, BCNU also has a supprescellular response. Ample evidence exists that sive effect on thymic derived lymphocytes (Tsuch cellular responses were effective in stimu- cells). T-cells are responsible for a variety of lating nonspecific immunity (1). It has also functions, including specific and nonspecific been shown that the more efficiently the BCG immune destruction of tumor cells and many organisms multiply in vivo, the more rapidly types of microorganisms. One of the hopes of they elicit tuberculous immunity in the vac- BCG vaccines in the treatment of neoplasia lies cinated host. Furthermore, the longer they per- in its ability to overcome immune suppression sisted in the tissues, the more lasting was the due to drugs and tumor and to stimulate the immunity to which they give rise (19). appropriate lymphoid elements which result in

742

SHER ET AL.

cellular immunity to specific tumor antigens. The multiplication of the Japan strain in the drug-treated mice was much higher than nondrug-treated mice (Table 3). This may indicate that the Japan strain was the strain least able to induce sufficient immunity in conjunction with BCNU to overcome its replication. On the other hand, the Brazil strain multiplied at the same rate in the drug-treated and non-drug-treated groups. The Pasteur strain was intermediate. Although the total CFU per spleen is higher in the more virulent strains, the ability to develop a good immune response may depend highly on its ability to multiply. Of course, too much virulence may totally overcome the immune defenses and lead to anergy and disseminated infection. An increased incidence of local and disseminated BCG infection in man has been reported in cancer patients who are undergoing therapy with BCG (C. W. Aungst and J. E. Sokal, Cancer Res. 14:108, 1973). BCG strain selection should be evaluated from the standpoint of ability to magnify cellular immunity without overwhelming the host's ability to cope with the vaccinating organisms. Investigations into the effects of these strain differences and the effects of the dead organisms in the vaccine in immunotherapy are in progress. These differences might explain the conflicting results in the plethora of recent reports concerning the use of BCG in human cancer. At the very least, if strain differences were not important in terms of the host's ability to destroy tumor cells, it would be important to choose strains carefully in terms of virulence for humans as this might lower the risk of localized or disseminated mycobacterial infection. ACKNOWLEDGMENTS We wish to thank Amy Steere for her excellent technical assistance. The assistance of H. Young, Sally Hedrick, Frank Malik, and Judith Torgeson is gratefully acknowledged. LITERATURE CITED 1. Blanden, R. V., M. J. Lefford, and G. B. Mackaness. 1969. The host response to Calmette-Guerin bacillus infection in mice. J. Exp. Med. 129:1079. 2. Chang, S. S. 1958. Studies on three substrains of BCG. I. Intensity and extent of lesions produced in guinea pigs by intraperitoneal injection of BCG. J. Formosan Med. Ass. (T'ai-Wan Hsueh Hui Tsa Chih) 57:453-463. 3. Chaparas, S. D., and S. R. Hedrick. 1973. Comparisons of strains of BCG I. Antigenic analysis and tuberculin reactivity. Infect. Immunity 7:777-780. 4. Collins, F. M. 1971. Mechanisms in antimicrobial immunity. J. Reticuloendothel. Soc. 10:58-99.

INFEcr. IMMUNfTY

5. Fairley, G. H. 1970. Immunotherapy of acute lymphoblastic leukemia, p. 278. XIII International Congress of Hematology. J. H. Lehrmann, Munich. 6. Grumbach, F. 1953. Pathogenicity of BCG strains for the mouse. Bull. Int. Un. Tuberc. 27:174. 7. Hauduroy, P., and W. Rosset. 1963. Douze ans d'experimentation sur l'infection des hamsters par le BCG. Ann. Inst. Pasteur 104:131-132. 8. Henderson, E. S., and B. G. Leventhal. 1972. Immunotherapy of leukemia from cancer chemotherapy. II, p. 327. The Twenty-second Hahnemann Symposium. Grune and Stratton, Inc. 9. Jensen, K. A. 1946. Practice of the Calmette vaccination. Acta Tuberc. Scand. 20:1-45. 10. Jesperson, A., and M. W. Bentzon. 1964a. The virulence of various strains determined on the golden hamster. Acta Tuberc. Scand. 44:222-249. 11. Jespersen, Andreas. 1971. The potency of BCG vaccines determined on animals. (monograph) Copenhagen, Denmark. 12. Krementz, E. T., M. S. Samuels, J. H. Wallace, and E. N. Benes. 1971. Clinical experiences in immunotherapy of cancer. Surg. Gynecol. Obstet. 133:209-217. 13. Levy, F. M., G. Conge, C. Fillastre, and E. Orssaud. 1963. Perspectives de la standardisation du BCG. Rev. Hyg. Med. Soc. 11:371-379. 14. Mathe, G., J. L. Amiel, L. Schwarzenberg, M. Schneider, A. Cattan, J. R. Schlumberger, M. Hayat, F. De Vassal. 1969. Active immunotherapy for acute lymphoblastic leukemia. Lancet 1:697. 15. Nadler, S. H., and G. E. Moore. 1969. Immunotherapy of malignant disease. Arch. Surg. 99:376. 16. Panisset, M., and P. Marois. 1953. Response de lEpiploon du cobaye a l'inoculation intraperitoneale de M. tuberculosis var BCG, interpretation quantitative des resultats. Rev. Pathol. Gen. Comp. 53:424-429. 17. Pearson, J. W., G. R. Pearson, W. T. Gibson, J. C. Chermann, and M. A. Chirigos. 1972. Combined chemoimmunostimulation therapy against murine leukemia. Cancer Res. 32:904-907. 18. Pierce, C. H., R. J. Dubos, and W. B. Schaefer. 1953. Multiplication and survival of tubercle bacilli in the organs of mice. J. Exp. Med. 97: 189-206. 19. Pierce, C. H., R. J. Dubos, and W. B. Shaefer. 1956. Differential characteristics in vitro and in vivo of several substrains of BCG. Amer. Rev. Tuberc. 74:655-717. 20. Rich, A. 1951. The pathogenesis of tuberculosis, 2nd ed., p. 108. Charles C. Thomas, Springfield, Ill. 21. Rosenthal, S. R., 1957. BCG Vaccination against tuberculosis, p. 389. Little, Brown and Co., Boston, Mass. 22. Rosenthal, S. R., A. H. Kenniewbrew, and N. Raisys. 1962. A modified semi-solid medium for determining viability of BCG. Acta Tuberc. Pneumol. Scand. 42:167-174. 23. Shikanai, K., K. Sato, and K. Fukushi. 1961. Comparative studies on the biological characteristics of Japanese strain and Moreau strain of BCG. Kosankinbyo Kenkyo Zasshi 15:167. 24. Togounova, A. I., and M. L. Khatenever. 1965. Material sur l'etude comparative des sousouches BCG. Probl. Tuberc. 43:61-66. 25. Zbar, B., I. Bernstein, T. Tanaka, H. J. Rapp. 1970. Tumor immunity produced by the intradermal inoculation of living tumor cells and living mycobacterium bovis (strain BCG). Science 170:1217-1218.