- Email: [email protected]
Immunotherapy with Mycobacterium vaccae in patients with newly diagnosed pulmonary tuberculosis: a randomised controlled trial
Immunotherapy with Mycobacterium vaccae in patients with newly diagnosed pulmonary tuberculosis: a randomised controlled trial *Durban Immunotherapy Trial Group
Summary Background Mycobacterium vaccae, an environmental saprophyte, has immunogenic properties that enhance the host immune response. Immunotherapy with M vaccae has been suggested to shorten short-course antituberculosis chemotherapy. We tested the hypothesis that the addition of M vaccae to standard short-course antituberculosis chemotherapy would decrease the time to achieve a negative sputum culture. Methods Patients with newly diagnosed tuberculosis were randomly assigned an injection of saline (placebo) or M vaccae on day 8. All patients received antituberculosis chemotherapy with rifampicin, isoniazid, pyrazinamide, and ethambutol. Sputum samples were checked by microscopy and culture every week for the first 8 weeks and monthly until the end of chemotherapy at 6 months. The primary outcome was the time to a negative sputum culture in the first 8 weeks. Intention-to-treat analysis was used and time to sputum clearance was assessed by log-rank test and Cox’s proportional-hazards regression. Findings 172 patients received M vaccae and 175 patients received placebo. At 8 weeks, 70 patients in the M vaccae group and 65 patients in the placebo group had a negative culture; there was no difference between groups in the time to a negative culture (p=0·83). There was no interaction between HIV status and treatment. Interpretation M vaccae immunotherapy has no benefit when added to standard antituberculosis chemotherapy. Lancet 1999; 354: 116–19 See Commentary page xxx
Introduction Modern chemotherapy for tuberculosis (TB) kills most of the rapidly replicating bacilli within days. However, a “persister” population of slowly replicating or dormant organisms need long-term treatment of at least 6 months.1 Thus, there are two phases in the treatment of TB, early bactericidal activity and tissue sterilisation.2 If optimum drug treatment is stopped early—ie, before 6 months— there is a high relapse rate (about 20%) with organisms An sensitive to the original combination.3 immunotherapeutic intervention to shorten the duration of therapy has been hypothesised since by enhancement of the host immune response, “persister” organisms would be killed faster. Most patients with TB have an immune response that seems inappropriate since it results in gross tissue destruction and progression of the disease: in the guineapig this response is the tissue-necrotising Koch phenomenon.4 Tissue necrosis may be due to the additional activity of Thelper (Th) type 2 lymphocytes to those of Th1 *Members listed at end of paper Correspondence to: Dr P C Onyebujoh, National Tuberculosis Research Programme, Medical Research Council, Private Bag X385, 0001 Pretoria, South Africa (e-mail: [email protected])
116
lymphocytes since Th1 lymphocytes should provide protective immunity. The rationale for TB immunotherapy is to replace immunopathology with protective antibacterial immunity. This could be achieved by switching off Th2 responses and enhancing Th1 mechanisms against antigens shared by all mycobacteria. Mycobacterium vaccae, a nonpathogenic environmental mycobacterium, has been shown to have such immunogenic properties.5 Preliminary immunotherapy studies have been done with killed suspensions of M vaccae for pulmonary TB in London,6 Kuwait,7 and The Gambia,8 and leprosy in Spain.9 These early studies established the formulation, dose, and safety profile, and some modes of action. Some of the reported benefits were in survival, symptom relief, weight gain,10 radiological resolution, and bacteriological parameters.11 The greatest benefits were seen in studies from areas where good treatment was not available or where the patients had a chronic disease or had drugresistant bacilli. However, these early studies were limited by methodological difficulties such as lack of randomisation, lack of appropriate controls, and the type and dose of chemotherapy given to the participants. These shortcomings prompted us to carry out a phase III randomised placebo-controlled trial of M vaccae in addition to a 6-month course of antituberculosis chemotherapy in patients with newly diagnosed pulmonary TB.
Methods Patients In King George V Hospital, Durban, South Africa, between September, 1994. and March, 1996, 374 patients had newly diagnosed, sputum smear-positive pulmonary TB. The inclusion criteria were: age 18–65 years; no previous treatment for TB in the past 2 years; TB bacilli susceptible to chemotherapy; weight more than 34 kg (women) and 38 kg (men); willing to be in hospital for at least 2 months; and willing to have HIV test after counselling. Exclusion criteria were: pregnancy; tuberculous meningitis; white blood-cell count <3·0⫻109/L; clinical jaundice or serum aspartate aminotransferase concentration >100 IU/L; total serum bilirubin concentration >35 mol/L; serum creatinine concentration >177 mol/L; gout; any serious life-threatening medical disorder such as diabetes, malignant disease, or AIDS; and disorders that needed treatment with corticosteroids. Patients were also excluded if they had an adverse response to the antituberculosis drugs within the first 7 days or did not complete the first 7 days of treatment. Patients were classified by the result of their admission HIV test as positive or negative. A positive HIV test was confirmed by either a second ELISA or a Western blot. Smoking status was defined by never or ever smoked. By protocol, all patients were kept in hospital for the first 8 weeks and for a further 4 weeks if they were smear positive. Patients were then reviewed monthly in clinic until the end of the study at 6 months. The study was conducted in accordance with the Declaration of Helsinki and its amendments, and the Guidelines for Good Clinical Practice as issued by the Committee of Propriety Medicinal Products of the European Union (July 1990). The study was approved by the Institutional Review Board of the University of Natal and each patient gave written, informed consent.
THE LANCET • Vol 354 • July 10, 1999
Variable
Hazard ratio (95% CI)
p value
Treatment HIV positive Age Body-mass index Smoker Smear positive BCG scar
0·96 (0·68–1·36) 1·15 (0·81–1·7) 0·98 (0·96–0·99) 1·06 (1·01–1·11) 0·58 (0·40–0·84) 0·73 (0·60–0·89) 1·48 (0·89–2·45)
0·83 0·43 0·041 0·033 0·004 0·002 0·13
2 for model p<0·0001.
Table 2: Multivariate Cox’s proportional-hazards regression analysis of variables evaluated as factors in sputum conversion occurred after a negative sample, then two further negative samples were required to confirm conversion. Each patient’s bacteriology results from the first 8 weeks were reviewed by the data advisory committee to determine whether and when culture conversion had taken place. If culture conversion did not take place, then the last week at which it was certain that conversion had not taken place was noted.
Figure 1: Trial profile
Randomisation and masking Microbiology On admission, early morning sputum samples were collected on 2 consecutive days. For the first 8 weeks, sputum samples were checked by microscopy and culture at weekly intervals. Thereafter, if patients were able to produce a sputum sample, two samples from consecutive days were collected every month for microscopy and culture. Each sputum sample was processed individually and the positive result was accepted if there was a discrepancy within pairs. If only one sample was available, the result from this sample was used. The growth of a single colony was deemed positive. Drug-susceptibility testing was done on admission samples by indirect absolute concentration.12,13 Resistance was defined as visible growth of 1% or more colonies in the drug-free control tube; results were reported as either susceptible or resistant. Testing was repeated on strains with intermediate (partial) resistance and if intermediate resistance was still seen the strain was reported as resistant. Sputum conversion was defined as two negative cultures on separate occasions—provided that no sample was positive after the first negative sample, If a positive sample M vaccae (n=172)
Placebo (n=175)
Demographic Age (years)* M/F
32 (9) 133/39
32 (10) 131/44
Clinical Body-mass index (kg/m2) Smoking never/ever BCG scar present HIV positive
19·0 (3·1) 57/115 147 60†
19·1 (3·0) 62/113 140 58
Chemotherapy All patients had 6 months treatment with oral antituberculosis chemotherapy. For the first 8 weeks, patients received daily: rifampicin 450 mg (600 mg if >50 kg); isoniazid 300 mg (400 mg if >50 kg); pyrazinamide 1·5 g (2·0 g if >50 kg); and ethambutol 1200 mg. For the 4-month continuation phase, patients continued the rifampicin and isoniazid at the same dose but taken three times a week. Pyridoxine 25 mg daily was taken for the full 6 months.
Endpoints
Laboratory ESR (mm/h) Tubercle characteristics Susceptibility Sensitive Low-level resistance Quantity on smear Scanty + ++ +++ Quantity on culture 1–20 colonies + ++ +++ ++++
Patients were randomly assigned intradermal injections of 0·1 mL 109 heat-killed M vaccae or saline on day 8. Randomisation was by means of random, permuted blocks of size 20. A person independent of the study labelled and packaged the supplies with the assistance of the study’s data monitoring agency. Sealed disclosure envelopes were supplied for each vial and were stored in the hospital pharmacy. A sealed copy of the randomisation code was also kept by the project leader (PBF) in case of serious adverse events. To mask investigators to possible reactions to M vaccae, people not involved in patient recruitment, management, or collection of samples carried out the intradermal injections, assessed the response at the injection site, and investigated adverse reactions. Furthermore, the injection site was covered with a bandage to mask patients and staff to the reaction.
98·8 (25·0)
97·8 (26·3)
169 3
167 3
6 14 46 106
5 12 57 101
4 9 18 98 43
3 7 23 91 51
4 64 99
4 77 90
Chest-radiograph grade Minimal Moderate Severe *Mean (SD). †One patient indeterminate status.
Table 1: Characteristics of patients and tubercle bacilli at randomisation
THE LANCET • Vol 354 • July 10, 1999
All endpoints were confirmed by the data monitoring committee. The primary endpoint was the time at which sputum culture conversion (as defined above) occurred in the first 8 weeks. Secondary endpoints were: culture status at 6 months (favourable, unfavourable, or not assessable); radiograpic score at day 56 (minimal, moderate, or severe); change in erythrocyte sedimentation rate (ESR) from day 1 to day 56; and change in weight from day 8 to day 56. Favourable culture status was a negative sputum culture between 3 months and 6 months, with no subsequent culture positive between 4 and 6 months or, a negative culture after 6 months without further treatment and either no result or no sputum sample from months 3, 4, 5, and 6. Unfavourable status was a positive sputum culture between 3 months and 6 months and no culture negative by 6 months. Not assessable status was defined as a patient who was positive or negative at week 8 but did not have a sample or result between 3 months and 6 months and no culture was negative after 6 months. The radiographic score was assessed without comparison with baseline radiographs. Compliance with treatment was checked in the first 8 weeks.
Statistical analysis Analysis of the primary endpoint was by intention to treat. As it was not known how HIV-positive patients might respond to immunotherapy, the data were analysed separately by treatment group and stratified by HIV status. If the effect of treatment did
117
Figure 2: Kaplan-Meier survival estimates of culture conversions by treatment group not differ by HIV status, the conclusions were to be based on the stratified analysis. If the effect of treatment varied by HIV status (ie, if there was evidence at the 5% level of significance of a treatment effect by HIV interaction), then this interaction would be described and the conclusions would be based on HIV-negative patients. Since the primary endpoint was right censored—ie, patients who had not converted by week 8 could convert at a later time—survival analysis was used. The culture conversion of the two treatment groups was compared by Kaplan-Meier method and treatment differences were tested by the log-rank test. Cox’s proportional-hazards regression model were fitted to investigate significant covariates among: age, sex, body-mass index, baseline sputum-culture status, baseline sputum-smear status, HIV status, BCG status, radiographic grade at baseline, and smoking status. For all covariates, marginal tests were done for interaction and, if significant at the 1% level, the interaction was included in the model. Changes in ESR and weight were analysed by linear regression. A 1% significance level was used to test for significant differences between the treatment groups to avoid type-1 error. The overall type-1 error rate was 6·7%. Differences in proportions between the groups were compared by 2 tests. Continuous variables were compared by two-sample t tests. Biochemical and haematological safety will be reported elsewhere. Safety was analysed for all 374 patients.
Results The flow of patients in the study is shown in figure 1. The demographic and clinical characteristics of the patients in the two groups at randomisation are in table 1. There was no interaction between HIV status and response to treatment. The Kaplan-Meier curves for the patients who converted in the first 8 weeks are shown in figure 2. At 8 weeks, 70 patients had converted in the M vaccae group compared with 65 patients in the placebo group (p=0·57). The covariates used in the Cox regression model of culture conversion by 8 weeks are in table 2: after adjustment for these factors there was no significant difference in the time to a negative culture between the two groups (p=0·83). Table 3 shows the culture status at 6 months by treatment group and HIV status. Status at 6 months did not differ between treatment groups for HIV positive (p=0·48) or negative patients (p=0·63). Culture status at 6 months*
HIV negative
M vaccae (n=189)
Placebo (n=185)
All (n=374)
Serious adverse events Recurrence of TB (%) Progression to AIDS Cor pulmonale Pneumothorax Hepatitis
19 (10·1%) 5 3 2 1
16 (8·6%) 5 1 1 2
35 (9·4%) 10 4 3 3
Total serious adverse events (%)
37 (19·6%)
34 (18·4%)
71 (19·0%)
Causes of death TB related Natural HIV related Other/unknown
4 1 3 5
5 4 2 3
9 5 5 8
Total deaths (%)
13 (6·9%)
14 (7·6%)
27 (7·2%)
Table 4: Serious adverse events and causes of death
24 patients in M vaccae group and 27 in the placebo group who were smear negative at week 8 remained in hospital—eg, because of employment difficulties or they were deemed by the ward physician to still be ill. 23 patients in the M vaccae group and 27 in the placebo group who were smear positive at week 8 left hospital—either because they were deemed well enough to leave or wished to leave. The proportion of patients remaining in hospital after day 56 did not differ between treatment groups, stratified by week 8 sputum smear status. 165 (96%) of patients in the M vaccae group and 166 (95%) in the placebo group took their drugs on 51 or more days of the first 8 weeks. Grades on the chest radiographs at day 56 in the placebo groups were: severe (39 [24%] vs 34 [22%]); moderate (84 [54%] vs 94 [60%]); and minimal (33 [22%] vs 28 [18%]), respectively. The mean change in ESR at 8 weeks was –38 mm/h in the M vaccae group and –38 mm/h in the placebo group. The mean weight gain at 8 weeks, was 4·0 kg in M vaccae group, and 4·4 kg in the placebo group (p=0·55). All M vaccae patients had some local induration at the injection site, 186 had erythema, 95 had vesiculation, and 44 patients had ulceration. Among the placebo group, 21 had induration, 17 had erythema, 2 had vesiculation, and 2 had ulceration. Table 4 shows the serious adverse events and causes of death.
Discussion The burden of TB on already overstretched health-care facilities in most developing countries is crippling. An intervention that can decrease expenditure on drugs and in-hospital care, with the attendant danger of nosocomial transmission, is urgently required. However, the results of this study do not justify such a role for M vaccae immunotherapy.
HIV positive
M vaccae
Placebo
M vaccae
Placebo
Favourable Negative at 3–6 months Negative after 6 months
79 13
78 10
38 8
43 5
Unfavourable Positive at 3–6 months
7
8
2
0
Not assessable Negative by week 8 Positive at week 8
4 8
6 15
5 7
6 4
*For complete definitions see text.
Table 3: Culture status at month 6 by treatment group and HIV status
118
Figure 3: Proportion of trial patients per treatment group with negative cultures
THE LANCET • Vol 354 • July 10, 1999
We did not find any benefit of the addition of M vaccae immunotherapy to appropriate TB chemotherapy. How then can the beneficial trends of other studies be explained?10,11 One possibility is that antituberculosis drugs may be of variable quality in some developing countries and consequently M vaccae immunotherapy maybe beneficial in the absence of optimum chemotherapy. Another explanation is that other studies may have had shortcomings in their design that resulted in biased observations. The time that immunotherapy was given and stage of disease may also have been factors. Most of our patients had advanced cavitary disease at baseline and, under such circumstances, conversion to bacteriologically negative sputum may be slower—even with optimum chemotherapy. Since the immune system is continuously presented with a high bacillary load, immunomodulating effects of M vaccae may have been masked. A study in Romania found significantly higher culture conversion rates at 2 months among immunotherapy recipients than placebo recipients.11 In that study, the M vaccae injection was given after 1 month of twice-weekly antituberculosis chemotherapy at which time the bacillary load would have been substantially reduced. None of the secondary endpoints differed between our two groups. The method used to evaluate radiological changes would not detect small changes and the time chosen for assessment was perhaps too soon to identify any differences between groups: radiographs from 6 months and 24 months have not yet been analysed. Although we used a strict definition for culture conversion in the first 8 weeks, we found a surprisingly low frequency in both groups (about 62%) of a negative culture. The expected clearance rate by 2 months, which would be week 9 in this study, is about 70% under routine field conditions and outcomes from controlled clinical trials are usually better.14 In Kigali, Rwanda, all patients had negative sputum by microscopy, irrespective of HIV status, by the end of 3 months on a combination of rifampicin, isoniazid, and ethambutol (or streptomycin).15 This may suggest poor compliance in our trial but we found that more than 95% of patients took their medicine regularly during the initial phase of treatment. Despite a less strict definition for sputum conversion at 6 months, the rates between our two groups did not differ. Similarly, the Romanian study11 did not find a significant difference between treatment groups at 6 months or 1 year. This result would be expected, even if immunotherapy showed remarkable differences at week 8, in a setting where appropriate drugs were available and compliance was reasonably assured. Some workers have suggested that immunotherapy may be useful in situations where chemotherapy alone fails because of drug resistance, lack of suitable drugs, and problems of non-compliance.16 Furthermore, a possible effect may be shown among HIV positive individuals who, being prone to relapse and reinfection, may benefit from multiple doses of immunotherapy.17 However, our results do not support the use of M vaccae immunotherapy for tuberculosis. Durban Immunotherapy Trial Group Writing Committee—P C Onyebujoh (chairman), J B Levin, P B Fourie, V Gathiram. Participating institutions and investigators—MRC National Tuberculosis Research Programme, Pretoria, South Africa (P B Fourie, P C Onyebujoh),
THE LANCET • Vol 354 • July 10, 1999
Durban (L C Tembe, N P Phili, T C P Mthiyane, T Moniwa, G Bayer, I M Ramajoe, T A B Mncwabe, L G M Mallisar, T N M Saul); Centre for Epidemiological Reseqarch in South Africa, ?where (J B Levin); Regional Office, Durban (T H F G Jackson, S Suparsad); London School of Hygiene and Tropical Medicine, London, UK (P E M Fine); King George V Hospital, Durban ( D J S Pendlebury, E Fine, I Houghton, J Clyde, H P Vos, N Padayatchi, A Pala, A Ramjee, M Ramjee, J Ramdeen, I H Masters, G Osbourne, K Naidu, S Bamba, B Mazur, R Czarnocki, K Landers, G Ndlovu, N Maphumulo); University of Natal Medical School, Durban (V Gathiram, A W Sturm, J Moodley, C Pillay, L Roux, R Moodley, A Sarawan, T Jali, F Manickam, A Smith, ? Gopaul); King Edward VIII Hospital, Durban (T Durosanmi, R Moonsammy); Inveresk Clinical Research, Scotland (P Wyld, J McCallum, C Fulton, K Bisset, S Henderson, D Stewart, O Nticinka, D Watson); Stanford-Rook Ltd, London UK (N Tuckwell, S Henderson, D Kennard); University College, London, UK (J L Stanford, G A Rook, J M Grange, A A Zumla). Independent Data Advisory Committee—E Bateman (chairman), P Hopewell, J Darbyshire, L Geiter, A Nunn, P E M Fine, K Weyer.
Acknowledgments This study was funded by Stanford Rook Holdings Ltd (UK) and monitored by Inveresk Research, Scotland, in association with monitors of the sponsoring company. The South African MRC was responsible for overall management of the trial. We thank King George V Hospital and University of Natal Medical School, Durban, and the Natal Provincial Administration for laboratory and clinical support and the Durban and Umlazi chest clinics for help recruiting the patients.
References 1 2
3
4 5
6
7
8 9
10
11
12
13
14
15
16 17
Grange JM. The mystery of the mycobacterial ‘persistor’. Tuber Lung Dis 1992; 73: 249–51. Mitchison DA. The Garrod lecture. understanding the chemotherapy of tuberculosis—current problems. J Antimicrob Chemother 1992; 29: 477–93. Balasubramanian R, Sivasubbramaniam S, Vijayan VK, et al. Five year results of a 3 month and two 5-month regimens for the treatment of sputum-positive pulmonary tuberculosis in South India. Tubercle 1990; 71: 253–58. Rook GA, Hernandez-Pando R. The pathogenesis of tuberculosis. Annu Rev Microbiol 1996; 50: 59–84. Stanford JL, Rook GA, Bahr GM, et al. Mycobacterium vaccae in immunoprophylaxis and immunotherapy of leprosy and tuberculosis. Vaccine 1990; 8: 252–30. Poziak A, Stanford JL, Johnson NMcl, Rook GA. Preliminary studies of immunotherapy of tuberculosis in man. Proceedings of the International Tuberculosis Congress, Singapore 1986. Bull IUALD 1987; 62: 39–40. Bahr GM, Shaaban MA, Gabriel M, et al. Improved immunotherapy for pulmonary tuberculosis with improved Mycobacterium vaccae. Tubercle 1990; 71: 259–66. Stanford JL, Bahr GM, Bypass O, et al. A modern approach to the immunotherapy of tuberculosis. Bull IUTLD 1990; 65: 27–29. Stanford JL, De las Aguas J, Turres P, Gervasoni B, Ravioli T. Studies on the effects of a potential immunotherapeutic agent in leprosy patients. Health Coop Pap 1987; 7: 201–06. Onyebujoh PC, Abdulmumini A, Robinson S, Rock GA, Stanford JL. Immunotherapy with Myobacterium vaccae as an addition to chemotherapy for the treatment of pulmonary tuberculosis under difficult conditions in Africa. Respir Med 1995; 89: 199–207. Corlan E, Marica C, Macabei C, Stanford JL, Stanford CA. Immunotherapy with Mycobacterium vaccae in the treatment of tuberculosis in Romania. Respir med 1997; 91: 12–19. Kleeberg HH, Koornhof HJ, Palmhert H. Laboratory manual of tuberculosis methods. 2nd ed. Revised by Nel EE, Kleeberg HH, Gatner EM. Pretoria: SAMRC Tuberculosis Research Institute, 1980. Kleeberg HH, Blacklock Z, Boulahbal F. A simle method of testing the drug susceptibility of Mycobacterium tuberculosis. Bull IUTLD 1985; 60: 147–50. American Thoracic Society. Medical Section of the American Lung Association: treatment of tuberculosis infection in adults and children. Am Rev Respir Dis 1986; 134: 355–63. Batungwanayo J, Taelman H, Dhote R, Bogaerts J, Allen S, Van de Perre P. Pulmonary tuberculosis in Kigali, Rwanda. Am Rev Respir Dis 1992; 146: 53–56. Etemadi A, Farid R, Stanford JL. Immunotherapy for drug resistant tuberculosis. Lancet 1992; 340: 1360–61. Narain JP, Raviglione MC, Kochi A. HIV-associated tuberculosis in developing countries: epidemiology and strategies for prevention. Tuber Lung Dis 1992; 73: 311–21.
119
Summary Background Mycobacterium vaccae, an environmental saprophyte, has immunogenic properties that enhance the host immune response. Immunotherapy with M vaccae has been suggested to shorten short-course antituberculosis chemotherapy. We tested the hypothesis that the addition of M vaccae to standard short-course antituberculosis chemotherapy would decrease the time to achieve a negative sputum culture. Methods Patients with newly diagnosed tuberculosis were randomly assigned an injection of saline (placebo) or M vaccae on day 8. All patients received antituberculosis chemotherapy with rifampicin, isoniazid, pyrazinamide, and ethambutol. Sputum samples were checked by microscopy and culture every week for the first 8 weeks and monthly until the end of chemotherapy at 6 months. The primary outcome was the time to a negative sputum culture in the first 8 weeks. Intention-to-treat analysis was used and time to sputum clearance was assessed by log-rank test and Cox’s proportional-hazards regression. Findings 172 patients received M vaccae and 175 patients received placebo. At 8 weeks, 70 patients in the M vaccae group and 65 patients in the placebo group had a negative culture; there was no difference between groups in the time to a negative culture (p=0·83). There was no interaction between HIV status and treatment. Interpretation M vaccae immunotherapy has no benefit when added to standard antituberculosis chemotherapy. Lancet 1999; 354: 116–19 See Commentary page xxx
Introduction Modern chemotherapy for tuberculosis (TB) kills most of the rapidly replicating bacilli within days. However, a “persister” population of slowly replicating or dormant organisms need long-term treatment of at least 6 months.1 Thus, there are two phases in the treatment of TB, early bactericidal activity and tissue sterilisation.2 If optimum drug treatment is stopped early—ie, before 6 months— there is a high relapse rate (about 20%) with organisms An sensitive to the original combination.3 immunotherapeutic intervention to shorten the duration of therapy has been hypothesised since by enhancement of the host immune response, “persister” organisms would be killed faster. Most patients with TB have an immune response that seems inappropriate since it results in gross tissue destruction and progression of the disease: in the guineapig this response is the tissue-necrotising Koch phenomenon.4 Tissue necrosis may be due to the additional activity of Thelper (Th) type 2 lymphocytes to those of Th1 *Members listed at end of paper Correspondence to: Dr P C Onyebujoh, National Tuberculosis Research Programme, Medical Research Council, Private Bag X385, 0001 Pretoria, South Africa (e-mail: [email protected])
116
lymphocytes since Th1 lymphocytes should provide protective immunity. The rationale for TB immunotherapy is to replace immunopathology with protective antibacterial immunity. This could be achieved by switching off Th2 responses and enhancing Th1 mechanisms against antigens shared by all mycobacteria. Mycobacterium vaccae, a nonpathogenic environmental mycobacterium, has been shown to have such immunogenic properties.5 Preliminary immunotherapy studies have been done with killed suspensions of M vaccae for pulmonary TB in London,6 Kuwait,7 and The Gambia,8 and leprosy in Spain.9 These early studies established the formulation, dose, and safety profile, and some modes of action. Some of the reported benefits were in survival, symptom relief, weight gain,10 radiological resolution, and bacteriological parameters.11 The greatest benefits were seen in studies from areas where good treatment was not available or where the patients had a chronic disease or had drugresistant bacilli. However, these early studies were limited by methodological difficulties such as lack of randomisation, lack of appropriate controls, and the type and dose of chemotherapy given to the participants. These shortcomings prompted us to carry out a phase III randomised placebo-controlled trial of M vaccae in addition to a 6-month course of antituberculosis chemotherapy in patients with newly diagnosed pulmonary TB.
Methods Patients In King George V Hospital, Durban, South Africa, between September, 1994. and March, 1996, 374 patients had newly diagnosed, sputum smear-positive pulmonary TB. The inclusion criteria were: age 18–65 years; no previous treatment for TB in the past 2 years; TB bacilli susceptible to chemotherapy; weight more than 34 kg (women) and 38 kg (men); willing to be in hospital for at least 2 months; and willing to have HIV test after counselling. Exclusion criteria were: pregnancy; tuberculous meningitis; white blood-cell count <3·0⫻109/L; clinical jaundice or serum aspartate aminotransferase concentration >100 IU/L; total serum bilirubin concentration >35 mol/L; serum creatinine concentration >177 mol/L; gout; any serious life-threatening medical disorder such as diabetes, malignant disease, or AIDS; and disorders that needed treatment with corticosteroids. Patients were also excluded if they had an adverse response to the antituberculosis drugs within the first 7 days or did not complete the first 7 days of treatment. Patients were classified by the result of their admission HIV test as positive or negative. A positive HIV test was confirmed by either a second ELISA or a Western blot. Smoking status was defined by never or ever smoked. By protocol, all patients were kept in hospital for the first 8 weeks and for a further 4 weeks if they were smear positive. Patients were then reviewed monthly in clinic until the end of the study at 6 months. The study was conducted in accordance with the Declaration of Helsinki and its amendments, and the Guidelines for Good Clinical Practice as issued by the Committee of Propriety Medicinal Products of the European Union (July 1990). The study was approved by the Institutional Review Board of the University of Natal and each patient gave written, informed consent.
THE LANCET • Vol 354 • July 10, 1999
Variable
Hazard ratio (95% CI)
p value
Treatment HIV positive Age Body-mass index Smoker Smear positive BCG scar
0·96 (0·68–1·36) 1·15 (0·81–1·7) 0·98 (0·96–0·99) 1·06 (1·01–1·11) 0·58 (0·40–0·84) 0·73 (0·60–0·89) 1·48 (0·89–2·45)
0·83 0·43 0·041 0·033 0·004 0·002 0·13
2 for model p<0·0001.
Table 2: Multivariate Cox’s proportional-hazards regression analysis of variables evaluated as factors in sputum conversion occurred after a negative sample, then two further negative samples were required to confirm conversion. Each patient’s bacteriology results from the first 8 weeks were reviewed by the data advisory committee to determine whether and when culture conversion had taken place. If culture conversion did not take place, then the last week at which it was certain that conversion had not taken place was noted.
Figure 1: Trial profile
Randomisation and masking Microbiology On admission, early morning sputum samples were collected on 2 consecutive days. For the first 8 weeks, sputum samples were checked by microscopy and culture at weekly intervals. Thereafter, if patients were able to produce a sputum sample, two samples from consecutive days were collected every month for microscopy and culture. Each sputum sample was processed individually and the positive result was accepted if there was a discrepancy within pairs. If only one sample was available, the result from this sample was used. The growth of a single colony was deemed positive. Drug-susceptibility testing was done on admission samples by indirect absolute concentration.12,13 Resistance was defined as visible growth of 1% or more colonies in the drug-free control tube; results were reported as either susceptible or resistant. Testing was repeated on strains with intermediate (partial) resistance and if intermediate resistance was still seen the strain was reported as resistant. Sputum conversion was defined as two negative cultures on separate occasions—provided that no sample was positive after the first negative sample, If a positive sample M vaccae (n=172)
Placebo (n=175)
Demographic Age (years)* M/F
32 (9) 133/39
32 (10) 131/44
Clinical Body-mass index (kg/m2) Smoking never/ever BCG scar present HIV positive
19·0 (3·1) 57/115 147 60†
19·1 (3·0) 62/113 140 58
Chemotherapy All patients had 6 months treatment with oral antituberculosis chemotherapy. For the first 8 weeks, patients received daily: rifampicin 450 mg (600 mg if >50 kg); isoniazid 300 mg (400 mg if >50 kg); pyrazinamide 1·5 g (2·0 g if >50 kg); and ethambutol 1200 mg. For the 4-month continuation phase, patients continued the rifampicin and isoniazid at the same dose but taken three times a week. Pyridoxine 25 mg daily was taken for the full 6 months.
Endpoints
Laboratory ESR (mm/h) Tubercle characteristics Susceptibility Sensitive Low-level resistance Quantity on smear Scanty + ++ +++ Quantity on culture 1–20 colonies + ++ +++ ++++
Patients were randomly assigned intradermal injections of 0·1 mL 109 heat-killed M vaccae or saline on day 8. Randomisation was by means of random, permuted blocks of size 20. A person independent of the study labelled and packaged the supplies with the assistance of the study’s data monitoring agency. Sealed disclosure envelopes were supplied for each vial and were stored in the hospital pharmacy. A sealed copy of the randomisation code was also kept by the project leader (PBF) in case of serious adverse events. To mask investigators to possible reactions to M vaccae, people not involved in patient recruitment, management, or collection of samples carried out the intradermal injections, assessed the response at the injection site, and investigated adverse reactions. Furthermore, the injection site was covered with a bandage to mask patients and staff to the reaction.
98·8 (25·0)
97·8 (26·3)
169 3
167 3
6 14 46 106
5 12 57 101
4 9 18 98 43
3 7 23 91 51
4 64 99
4 77 90
Chest-radiograph grade Minimal Moderate Severe *Mean (SD). †One patient indeterminate status.
Table 1: Characteristics of patients and tubercle bacilli at randomisation
THE LANCET • Vol 354 • July 10, 1999
All endpoints were confirmed by the data monitoring committee. The primary endpoint was the time at which sputum culture conversion (as defined above) occurred in the first 8 weeks. Secondary endpoints were: culture status at 6 months (favourable, unfavourable, or not assessable); radiograpic score at day 56 (minimal, moderate, or severe); change in erythrocyte sedimentation rate (ESR) from day 1 to day 56; and change in weight from day 8 to day 56. Favourable culture status was a negative sputum culture between 3 months and 6 months, with no subsequent culture positive between 4 and 6 months or, a negative culture after 6 months without further treatment and either no result or no sputum sample from months 3, 4, 5, and 6. Unfavourable status was a positive sputum culture between 3 months and 6 months and no culture negative by 6 months. Not assessable status was defined as a patient who was positive or negative at week 8 but did not have a sample or result between 3 months and 6 months and no culture was negative after 6 months. The radiographic score was assessed without comparison with baseline radiographs. Compliance with treatment was checked in the first 8 weeks.
Statistical analysis Analysis of the primary endpoint was by intention to treat. As it was not known how HIV-positive patients might respond to immunotherapy, the data were analysed separately by treatment group and stratified by HIV status. If the effect of treatment did
117
Figure 2: Kaplan-Meier survival estimates of culture conversions by treatment group not differ by HIV status, the conclusions were to be based on the stratified analysis. If the effect of treatment varied by HIV status (ie, if there was evidence at the 5% level of significance of a treatment effect by HIV interaction), then this interaction would be described and the conclusions would be based on HIV-negative patients. Since the primary endpoint was right censored—ie, patients who had not converted by week 8 could convert at a later time—survival analysis was used. The culture conversion of the two treatment groups was compared by Kaplan-Meier method and treatment differences were tested by the log-rank test. Cox’s proportional-hazards regression model were fitted to investigate significant covariates among: age, sex, body-mass index, baseline sputum-culture status, baseline sputum-smear status, HIV status, BCG status, radiographic grade at baseline, and smoking status. For all covariates, marginal tests were done for interaction and, if significant at the 1% level, the interaction was included in the model. Changes in ESR and weight were analysed by linear regression. A 1% significance level was used to test for significant differences between the treatment groups to avoid type-1 error. The overall type-1 error rate was 6·7%. Differences in proportions between the groups were compared by 2 tests. Continuous variables were compared by two-sample t tests. Biochemical and haematological safety will be reported elsewhere. Safety was analysed for all 374 patients.
Results The flow of patients in the study is shown in figure 1. The demographic and clinical characteristics of the patients in the two groups at randomisation are in table 1. There was no interaction between HIV status and response to treatment. The Kaplan-Meier curves for the patients who converted in the first 8 weeks are shown in figure 2. At 8 weeks, 70 patients had converted in the M vaccae group compared with 65 patients in the placebo group (p=0·57). The covariates used in the Cox regression model of culture conversion by 8 weeks are in table 2: after adjustment for these factors there was no significant difference in the time to a negative culture between the two groups (p=0·83). Table 3 shows the culture status at 6 months by treatment group and HIV status. Status at 6 months did not differ between treatment groups for HIV positive (p=0·48) or negative patients (p=0·63). Culture status at 6 months*
HIV negative
M vaccae (n=189)
Placebo (n=185)
All (n=374)
Serious adverse events Recurrence of TB (%) Progression to AIDS Cor pulmonale Pneumothorax Hepatitis
19 (10·1%) 5 3 2 1
16 (8·6%) 5 1 1 2
35 (9·4%) 10 4 3 3
Total serious adverse events (%)
37 (19·6%)
34 (18·4%)
71 (19·0%)
Causes of death TB related Natural HIV related Other/unknown
4 1 3 5
5 4 2 3
9 5 5 8
Total deaths (%)
13 (6·9%)
14 (7·6%)
27 (7·2%)
Table 4: Serious adverse events and causes of death
24 patients in M vaccae group and 27 in the placebo group who were smear negative at week 8 remained in hospital—eg, because of employment difficulties or they were deemed by the ward physician to still be ill. 23 patients in the M vaccae group and 27 in the placebo group who were smear positive at week 8 left hospital—either because they were deemed well enough to leave or wished to leave. The proportion of patients remaining in hospital after day 56 did not differ between treatment groups, stratified by week 8 sputum smear status. 165 (96%) of patients in the M vaccae group and 166 (95%) in the placebo group took their drugs on 51 or more days of the first 8 weeks. Grades on the chest radiographs at day 56 in the placebo groups were: severe (39 [24%] vs 34 [22%]); moderate (84 [54%] vs 94 [60%]); and minimal (33 [22%] vs 28 [18%]), respectively. The mean change in ESR at 8 weeks was –38 mm/h in the M vaccae group and –38 mm/h in the placebo group. The mean weight gain at 8 weeks, was 4·0 kg in M vaccae group, and 4·4 kg in the placebo group (p=0·55). All M vaccae patients had some local induration at the injection site, 186 had erythema, 95 had vesiculation, and 44 patients had ulceration. Among the placebo group, 21 had induration, 17 had erythema, 2 had vesiculation, and 2 had ulceration. Table 4 shows the serious adverse events and causes of death.
Discussion The burden of TB on already overstretched health-care facilities in most developing countries is crippling. An intervention that can decrease expenditure on drugs and in-hospital care, with the attendant danger of nosocomial transmission, is urgently required. However, the results of this study do not justify such a role for M vaccae immunotherapy.
HIV positive
M vaccae
Placebo
M vaccae
Placebo
Favourable Negative at 3–6 months Negative after 6 months
79 13
78 10
38 8
43 5
Unfavourable Positive at 3–6 months
7
8
2
0
Not assessable Negative by week 8 Positive at week 8
4 8
6 15
5 7
6 4
*For complete definitions see text.
Table 3: Culture status at month 6 by treatment group and HIV status
118
Figure 3: Proportion of trial patients per treatment group with negative cultures
THE LANCET • Vol 354 • July 10, 1999
We did not find any benefit of the addition of M vaccae immunotherapy to appropriate TB chemotherapy. How then can the beneficial trends of other studies be explained?10,11 One possibility is that antituberculosis drugs may be of variable quality in some developing countries and consequently M vaccae immunotherapy maybe beneficial in the absence of optimum chemotherapy. Another explanation is that other studies may have had shortcomings in their design that resulted in biased observations. The time that immunotherapy was given and stage of disease may also have been factors. Most of our patients had advanced cavitary disease at baseline and, under such circumstances, conversion to bacteriologically negative sputum may be slower—even with optimum chemotherapy. Since the immune system is continuously presented with a high bacillary load, immunomodulating effects of M vaccae may have been masked. A study in Romania found significantly higher culture conversion rates at 2 months among immunotherapy recipients than placebo recipients.11 In that study, the M vaccae injection was given after 1 month of twice-weekly antituberculosis chemotherapy at which time the bacillary load would have been substantially reduced. None of the secondary endpoints differed between our two groups. The method used to evaluate radiological changes would not detect small changes and the time chosen for assessment was perhaps too soon to identify any differences between groups: radiographs from 6 months and 24 months have not yet been analysed. Although we used a strict definition for culture conversion in the first 8 weeks, we found a surprisingly low frequency in both groups (about 62%) of a negative culture. The expected clearance rate by 2 months, which would be week 9 in this study, is about 70% under routine field conditions and outcomes from controlled clinical trials are usually better.14 In Kigali, Rwanda, all patients had negative sputum by microscopy, irrespective of HIV status, by the end of 3 months on a combination of rifampicin, isoniazid, and ethambutol (or streptomycin).15 This may suggest poor compliance in our trial but we found that more than 95% of patients took their medicine regularly during the initial phase of treatment. Despite a less strict definition for sputum conversion at 6 months, the rates between our two groups did not differ. Similarly, the Romanian study11 did not find a significant difference between treatment groups at 6 months or 1 year. This result would be expected, even if immunotherapy showed remarkable differences at week 8, in a setting where appropriate drugs were available and compliance was reasonably assured. Some workers have suggested that immunotherapy may be useful in situations where chemotherapy alone fails because of drug resistance, lack of suitable drugs, and problems of non-compliance.16 Furthermore, a possible effect may be shown among HIV positive individuals who, being prone to relapse and reinfection, may benefit from multiple doses of immunotherapy.17 However, our results do not support the use of M vaccae immunotherapy for tuberculosis. Durban Immunotherapy Trial Group Writing Committee—P C Onyebujoh (chairman), J B Levin, P B Fourie, V Gathiram. Participating institutions and investigators—MRC National Tuberculosis Research Programme, Pretoria, South Africa (P B Fourie, P C Onyebujoh),
THE LANCET • Vol 354 • July 10, 1999
Durban (L C Tembe, N P Phili, T C P Mthiyane, T Moniwa, G Bayer, I M Ramajoe, T A B Mncwabe, L G M Mallisar, T N M Saul); Centre for Epidemiological Reseqarch in South Africa, ?where (J B Levin); Regional Office, Durban (T H F G Jackson, S Suparsad); London School of Hygiene and Tropical Medicine, London, UK (P E M Fine); King George V Hospital, Durban ( D J S Pendlebury, E Fine, I Houghton, J Clyde, H P Vos, N Padayatchi, A Pala, A Ramjee, M Ramjee, J Ramdeen, I H Masters, G Osbourne, K Naidu, S Bamba, B Mazur, R Czarnocki, K Landers, G Ndlovu, N Maphumulo); University of Natal Medical School, Durban (V Gathiram, A W Sturm, J Moodley, C Pillay, L Roux, R Moodley, A Sarawan, T Jali, F Manickam, A Smith, ? Gopaul); King Edward VIII Hospital, Durban (T Durosanmi, R Moonsammy); Inveresk Clinical Research, Scotland (P Wyld, J McCallum, C Fulton, K Bisset, S Henderson, D Stewart, O Nticinka, D Watson); Stanford-Rook Ltd, London UK (N Tuckwell, S Henderson, D Kennard); University College, London, UK (J L Stanford, G A Rook, J M Grange, A A Zumla). Independent Data Advisory Committee—E Bateman (chairman), P Hopewell, J Darbyshire, L Geiter, A Nunn, P E M Fine, K Weyer.
Acknowledgments This study was funded by Stanford Rook Holdings Ltd (UK) and monitored by Inveresk Research, Scotland, in association with monitors of the sponsoring company. The South African MRC was responsible for overall management of the trial. We thank King George V Hospital and University of Natal Medical School, Durban, and the Natal Provincial Administration for laboratory and clinical support and the Durban and Umlazi chest clinics for help recruiting the patients.
References 1 2
3
4 5
6
7
8 9
10
11
12
13
14
15
16 17
Grange JM. The mystery of the mycobacterial ‘persistor’. Tuber Lung Dis 1992; 73: 249–51. Mitchison DA. The Garrod lecture. understanding the chemotherapy of tuberculosis—current problems. J Antimicrob Chemother 1992; 29: 477–93. Balasubramanian R, Sivasubbramaniam S, Vijayan VK, et al. Five year results of a 3 month and two 5-month regimens for the treatment of sputum-positive pulmonary tuberculosis in South India. Tubercle 1990; 71: 253–58. Rook GA, Hernandez-Pando R. The pathogenesis of tuberculosis. Annu Rev Microbiol 1996; 50: 59–84. Stanford JL, Rook GA, Bahr GM, et al. Mycobacterium vaccae in immunoprophylaxis and immunotherapy of leprosy and tuberculosis. Vaccine 1990; 8: 252–30. Poziak A, Stanford JL, Johnson NMcl, Rook GA. Preliminary studies of immunotherapy of tuberculosis in man. Proceedings of the International Tuberculosis Congress, Singapore 1986. Bull IUALD 1987; 62: 39–40. Bahr GM, Shaaban MA, Gabriel M, et al. Improved immunotherapy for pulmonary tuberculosis with improved Mycobacterium vaccae. Tubercle 1990; 71: 259–66. Stanford JL, Bahr GM, Bypass O, et al. A modern approach to the immunotherapy of tuberculosis. Bull IUTLD 1990; 65: 27–29. Stanford JL, De las Aguas J, Turres P, Gervasoni B, Ravioli T. Studies on the effects of a potential immunotherapeutic agent in leprosy patients. Health Coop Pap 1987; 7: 201–06. Onyebujoh PC, Abdulmumini A, Robinson S, Rock GA, Stanford JL. Immunotherapy with Myobacterium vaccae as an addition to chemotherapy for the treatment of pulmonary tuberculosis under difficult conditions in Africa. Respir Med 1995; 89: 199–207. Corlan E, Marica C, Macabei C, Stanford JL, Stanford CA. Immunotherapy with Mycobacterium vaccae in the treatment of tuberculosis in Romania. Respir med 1997; 91: 12–19. Kleeberg HH, Koornhof HJ, Palmhert H. Laboratory manual of tuberculosis methods. 2nd ed. Revised by Nel EE, Kleeberg HH, Gatner EM. Pretoria: SAMRC Tuberculosis Research Institute, 1980. Kleeberg HH, Blacklock Z, Boulahbal F. A simle method of testing the drug susceptibility of Mycobacterium tuberculosis. Bull IUTLD 1985; 60: 147–50. American Thoracic Society. Medical Section of the American Lung Association: treatment of tuberculosis infection in adults and children. Am Rev Respir Dis 1986; 134: 355–63. Batungwanayo J, Taelman H, Dhote R, Bogaerts J, Allen S, Van de Perre P. Pulmonary tuberculosis in Kigali, Rwanda. Am Rev Respir Dis 1992; 146: 53–56. Etemadi A, Farid R, Stanford JL. Immunotherapy for drug resistant tuberculosis. Lancet 1992; 340: 1360–61. Narain JP, Raviglione MC, Kochi A. HIV-associated tuberculosis in developing countries: epidemiology and strategies for prevention. Tuber Lung Dis 1992; 73: 311–21.
119