Randomised trial of thiacetazone and rifampicin-containing regimens for pulmonary tuberculosis in HIV-infected Ugandans

Randomised trial of thiacetazone and rifampicin-containing regimens for pulmonary tuberculosis in HIV-infected Ugandans

Summary who received treatment for active tuberculosis, thiacetazone has been associated with cutaneous hypersensitivity and recurrent tuberculosis. No controlled trials have investigated the safety and efficacy of thiacetazone-containing regimens compared with alternative regimens among patients with HIV. In a randomised clinical trial of 191 HIV-positive patients with active pulmonary tuberculosis, we examined the safety and short-term efficacy of isoniazid, rifampicin, and pyrazinamide for two months followed by isoniazid and rifampicin for seven months (RHZ) compared with streptomycin, thiacetazone, and isoniazid for two months followed by thiacetazone and isoniazid for ten months (STH). Between May, 1990, and September, 1991, 191 HIVpositive adult Ugandan patients with acid-fast bacilli sputum smear-positive pulmonary tuberculosis (93% confirmed by culture) received either STH or RHZ. Subjects had a standard evaluation that included Mantoux skin test, complete blood count with differential white blood cell count, and chest radiography. After starting therapy, subjects were followed-up over one year for three outcomes: complications of anti-tuberculosis therapy, early sterilisation of cultures, and survival. Of 191 eligible subjects, 90 received STH and 101 received RHZ. The overall one-year survival was similar for STH and RHZ (65% vs 72%), but when controlled for baseline differences in Mantoux reaction size and absolute lymphocyte count, the relative risk of death for STH compared with RHZ was 1·57 (95% Cl 1·0-2·48). Overall, 12 adverse drug reactions occurred in the STH arm (18·2 reactions per 100 person years [PYO]) compared with one in the RHZ arm (1·6 reactions per 100 PYO) for a relative risk of 11·7 (95% Cl 1·52-90·0). 10 cutaneous reactions occurred in the STH arm (15·2 events per 100 PYO) compared with one event in the RHZ arm (1·6 events per 100 PYO) for a relative risk of 9·7 (95% Cl: 1·24, 75·8). A greater proportion of RHZ patients compared with STH patients had sterilised their sputum within two months (74% vs 37%, p<0·001).

Among HIV-positive patients

*Participants are listed at the end of the article. Ugandan Ministry of Health and Department of Medicine, Makerere University, Kampala, Uganda (A Okwera MB, F Byekwaso MB, R Mugerwa MB); Ugandan National Tuberculosis and Leprosy Control Programme, Kampala (A Okwera, F Byekwaso); Department of Medicine, Division of Infectious Disease (J Ellner MD, M Vjecha MD, J Johnson MD), General Internal Medicine and Department of Epidemiology and Biostatistics (C Whalen MD), Case Western Reserve University School of

Medicine, Cleveland, Ohio USA; Division of Tuberculosis Elimination, Centers for Disease Control and Prevention, Atlanta, Georgia (R Huebner PhD)

Correspondence

to: Dr

Christopher Whalen

In developing countries, rifampicin-containing regimens should be given, when possible, to HIV-positive patients to reduce drug toxicity and to prolong survival.

Introduction The HIV epidemic has posed new challenges for tuberculosis control especially in sub-Saharan Africa where both diseases are prevalent.’-3 HIV infection confers the greatest risk for development of active tuberculosis, for reactivation of dormant infection, and for progressive primary infection.4-7 Although HIV-infected patients are at increased risk of tuberculosis, they initially respond well to appropriate antituberculosis therapy,8 if the organism In industrialised is countries, drug-sensitive. antituberculous therapy usually consists of regimens containing isoniazid and rifampicin, whereas in east African countries, standard regimens for treating active pulmonary tuberculosis often contain isoniazid and of thiacetazone instead rifampicin because of thiacetazone’s low cost and ease of administration.99 Before the HIV epidemic, thiacetazone was widely used in developing countries for the treatment of tuberculosis despite the known risks of adverse drug reactions’"’" and recurrent disease.12 Soon after the African HIV epidemic began limitations of thiacetazone were identified. In particular, among HIV-positive patients who received treatment for active tuberculosis, thiacetazone was associated with cutaneous hypersensitivity’3-’6 and recurrent disease." Studies from Kenya showed that cutaneous reactions to antituberculosis therapy were frequent in HIV infection (20%), and that the risk for cutaneous hypersensitivity was 18 times greater in HIVinfected patients than in tuberculosis patients without HIV infection.15 Moreover, some reactions attributed to thiacetazone were characterised by severe cutaneous toxicity, such as Stevens-Johnson syndrome, and could cause death.15,16 In a randomised clinical trial of 191 HIVinfected patients with active pulmonary tuberculosis, we examined the safety and short-term efficacy of isoniazid, rifampicin, and pyrazinamide for two months followed by isoniazid and rifampin for seven months compared with streptomycin, thiacetazone, and isoniazid for two months followed by thiacetazone and isoniazid for ten months.

Subjects

and methods

Between May, 1990 and September, 1991, 410 patients at the national referral Tuberculosis Treatment Center at Mulago Hospital in Kampala, Uganda, were screened for active pulmonary tuberculosis with a detailed history and physical examination, collection of expectorated sputum specimens for acid-fast bacilli (AFB) smear and culture, tuberculin skin testing, and chest radiography. Of the 410 patients, 330 had a diagnosis of tuberculosis confirmed by sputum culture or clinical presentation. Of 330 patients, 218 (66%) were infected with

1323

HIV-1(EIA and western blot). Patients were eligible if they were HIV-positive, had at least one sputum smear positive for acidfast bacilli and a chest radiograph consistent with pulmonary tuberculosis, were between 15 and 60 years of age, and lived within a 20 km radius of Kampala. Subjects were excluded if they were severely debilitated (Karnofsky performance score" of less than 50%), had previous or current treatment for active tuberculosis, were pregnant, had jaundice, were severely anaemic (haemoglobin <5 mg/dL), or had a total white blood cell count of less than 3000 cells/mm’. To preserve confidentiality about HIV serostatus, 10% of the subjects enrolled in the cohort did not have HIV infection; therefore, participation in the study did not indicate HIV status. Data from these subjects were excluded from the analysis. The study protocol was approved by institutional review boards at Makerere University and Case Western Reserve University.

Treatment

regimens

and randomisation

randomised to one of two treatments after Eligible patients informed consent. The treatment arms were streptomycin, giving and thiacetazone for two months followed by isoniazid isoniazid, and thiacetazone for ten months (2STH/10TH, hereafter STH) or isoniazid, rifampicin, and pyrazinamide for two months followed by isoniazid and rifampicin for seven months (2RHZ/7RH, hereafter RHZ). Isoniazid was given at 300 mg per day and thiacetazone at 150 mg per day. were

daily doses of rifampicin, pyrazinamide, and streptomycin adjusted according to body weight. For subjects less than 50 kg, the daily doses were rifampicin 450 mg, pyrazinamide 1-5 gm, and streptomycin 0-75 gm; for subjects over 50 kg, daily doses were rifampicin 600 mg, pyrazinamide 2 gm, and streptomycin 1 gm. Neither physician nor subject was blinded to the treatment regimen. Eligible patients were assigned to one of the two treatment arms by random allocation and a block size of The

were

120.

Measurements At baseline, all demographic and clinical information was obtained through interviews and physical examination. At screening and after verbal consent, venous blood was tested for HIV-1(EIA testing, Cambridge Bioscience, Cambridge, MA). All sera that was positive by EIA and one of ten negative sera were retested by EIA and confirmed with subsequent western blot (BioRad Novapath, Hercules, CA). Blood was also collected for colorimetric haemoglobin measurement and manual white blood count and differential. At enrolment, all subjects had Mantoux skin tests with five tuberculin units of purified protein derivative (PPD) (Tubersol, Connaught Laboratories, Ontario, Canada). At 48 hours, a single observer blinded to treatment assignment, recorded size of induration (mm) and the presence or absence of a BCG scar. Cutaneous anergy to PPD was defined as absence of reactivity (0 mm of induration) to the Mantoux PPD skin test. Patients had posterior-anterior chest radiographs taken at baseline, two months, six months, and at the end of therapy. Chest radiographs were interpreted with a standard protocol by a single blinded reviewer. Radiographs were examined for the pattern of involvement, extent of disease, specific radiographic findings, and interval change during treatment and follow-up. The radiographic extent of disease was graded as normal, minimal, moderately advanced or far advanced disease using published criteria established by the National Tuberculosis and Respiratory Disease Association, USA. Extent of disease also was classified as a total zone score representing the number of lung regions involved by disease. Each hemithorax was divided into upper, middle, and lower lung zones by perpendicular lines drawn at equidistant intervals along a line extending from the apex of the lung to the dome of the hemidiaphragm. Each zone was assigned a score of 1 if there was visible disease (infiltrate, effusion, cavity, etc) or 2 if no lesions were present. The scores from all six zones were summed yielding a total score of 6-12. The presence of specific radiographic abnormalities such as

1324

t186 patients had

baseline chest

X-rays available for review (96% of patients

receiving STH and 99% receiving RHZ).

tp<0.05. Table 1: Baseline characteristics of patients treatment assignment

according to

pleural effusion, cavitation, and volume loss was also noted. Twomonth chest radiographs were improved if the total zone score was higher and worse if the total zone score was lower. Chest radiographs were classified as typical tuberculosis if upper lung field infiltrates or cavities were present with or without lower lung field infiltrates or pleural effusions. Chest radiographs were classified as atypical if infiltrates, including cavities, were present predominantly in the lower lung fields or if there was prominent hilar or mediastinal adenopathy or miliary disease. Lower lung field involvement was defined as presence of disease below a line drawn between the pulmonary hili. At least one sputum sample was collected at baseline and after one, two, and six months of treatment, at the end of therapy, and every 6 months thereafter. All sputum specimens were digested with 3% NaOH, concentrated, and stained for AFB with both Ziehl-Neelsen and fluorescent auramine stains at the Uganda Tuberculosis Investigations Bacteriological Unit, Wandegeya. AFB smears were reviewed by trained technicians who graded the smears by the number of acid-fast organisms seen on light microscopy on an ordinal scale from 0-4.’"’ The number of organisms in the two-month smear was decreased if the ordinal score was lower than baseline and increased if the ordinal score was higher. Specimens were cultured for Mycobacterium tuberculosis on Lowenstein-Jensen medium slants, incubated at 37°C in air and examined weekly until positive or for eight weeks. Isolates were confirmed as M tuberculosis if the organism was nitrate positive, niacin positive, had typical colonial morphology, and growth on Middlebrook 7-HIO agar and was not inhibited by 5% thiopencarboxylic acid hydrazine (TCH). We diagnosed pulmonary tuberculosis if the sputum smear showed AFB and the patient had a chest radiograph showing active tuberculosis or the sputum culture was positive for M tuberculosis

Outcomes and

follow-up

Adverse drug reactions and survival were the major outcomes of interest. At monthly intervals during therapy, subjects were evaluated with a standard history and physical examination. Adverse drug reactions were defined as signs or symptoms consistent with known toxicity of the drugs prescribed which resolved when the drug was discontinued. Subjects were routinely evaluated for skin rash, jaundice, nausea, or vomiting, tinnitus and hearing loss. Survival was defined as the interval between the start of therapy and death or, in the case of subjects who were censored, the date they were last seen alive. Subjects were censored if they were lost to follow-up or were alive onI August, 1993. For the first two months of therapy, most subjects

admitted

Old Mulago Hospital where treatment assured by directly observed therapy. After compliance were evaluated at monthly intervals during discharge, patients treatment and quarterly thereafter. Adherence to therapy was assessed by the medical officers at each visit and measured as the frequency of defaulted appointments. Patients who did not keep scheduled appointments were visited by a home visitor and asked to return to the clinic for evaluation. were

to

was

Analysis Demographic and clinical characteristics were compared between treatment two groups to assess the adequacy of randomisation. Categorical variables were compared with the chi squared test and continuous variables with Student’s t-test (or appropriate non-parametric test). Survival distributions for the subjects in each treatment group were estimated with the Kaplan-Meier test and compared with log-rank test. Incidence rates of adverse drug reactions and recurrent disease were calculated with the person-years method and expressed as the the

number of

new

events

per 100 person-years of observation

(PYO). We used incidence density ratio to compare the incidence rates of adverse drug reactions between the two groups. Cox proportional hazards regression analysis was used to determine relative risk for death and to control for baseline differences between the two treatment groups in confounding variables and we analysed the study by intention-to-treat method as assigned by randomisation.

Results

Study population A diagnosis of smear-positive active pulmonary tuberculosis was made in 330 of 410 patients screened. Of the 330, 218 (66%) were HIV-1positive and 191 met all study eligibility criteria. The remaining 27 HIV-positive patients were excluded because of poor performance status (2), residence outside of Kampala (8), initial negative sputum smears (6), previous treatment for tuberculosis (3), pregnancy (1) and subject refusal (7). The final study sample included 132 (69%) men and 59 (31 %) women. The mean (SD) age was 30 (8) years range: 16-55 years. We confirmed diagnosis of active tuberculosis in the study cohort by sputum culture in 178 subjects (93%); the remaining 13 subjects had positive sputum smears and chest radiographs showed moderate to far advanced disease. Four of these 13 subjects were in These the STH arm and 9 were in the RHZ arm (p>0-2). were similar to the culture 13 subjects positive subjects with regard to age, sex, Mantoux test reaction size, duration of follow-up and extent of disease at baseline. We randomly assigned 90 subjects to receive STH and 101 to receive RHZ. The discrepancy in the size of the two groups resulted from the block size (120) used in the random assignment of therapy. Table 1 shows the two groups were similar in age, gender, HIV-related conditions, presence of BCG scar, chest radiograph findings, baseline hamoglobin, and white blood cell count, but differed in baseline absolute lymphocyte count and size of the Mantoux reaction. The distribution of HIV-related conditions such as thrush, varicella zoster, and chronic diarrhoea was similar at baseline between the two groups. The proportion of subjects with cutaneous anergy to PPD (0 mm induration) was greater in the RHZ arm but did not reach statistical significance. The proportion of subjects with moderate to far advanced disease on chest radiographs was comparable in the two groups. The duration of follow-up from the start of therapy was similar for both treatment groups, 15-5 months for STD cohorts and 16-1months for RHZ

Follow-up time (days) Figure: Overall survival of subjects according to treatment arm. There was no significant difference in the survival distributions between the treatment arms (pa05) cohorts.

days

to

Follow-up 35

for the STH group

months, and for the RHZ

ranged from 2 days to 32

group 3

months. Survival Of the 191 subjects, 12 (6%) died within 30 days of therapy, 7 in the STH arm and 5 in the RHZ arm. Figure 1 shows that overall survival proportions 12 months after diagnosis of tuberculosis were similar for the two groups, 65% in the STH arm and 72% in the RHZ arm (p>0’2, logrank test). Because of the baseline differences in the size of the Mantoux skin test reaction and the absolute lymphocyte count, further analyses were done to control for these potential confounding variables on survival. The unadjusted relative risk of death in the STH group compared with the RHZ group was 1-01 (95% CI: 0-65, 1-57). However, when adjusting for the Mantoux reaction and the absolute lymphocyte count at baseline in a Cox regression analysis, the relative risk of death for STH compared with RHZ was 1-57 (95% CI: 1-00, 2-48). Adverse effects of therapy During the study, there were 13 adverse drug reactions attributed to antituberculous medications for an overall incidence of drug-related toxicity of 10-0 events per 100 PYO (table 2). Twelve occurred in the STH arm and gave an incidence rate of 18-2 events per 100 PYO. Only one adverse event occurred in the RHZ arm for an incidence rate of 1-6 events per 100 PYO. The rate ratio for adverse drug reactions in subjects receiving STH compared with RHZ was 11-7 (95% CI: 1-52, 90-0). Of the 13 adverse reactions, 11 were cutaneous reactions yielding a combined incidence rate of 8-5 reactions per 100 PYO (table 2). Ten cutaneous reactions occurred in the STH group, for an incidence rate of 15-2 reactions per 100 PYO; one cutaneous drug reaction occurred in the RHZ group for an incidence rate of 1-6 reactions per 100 PYO. The relative risk of adverse cutaneous drug reactions in subjects receiving STH compared with RHZ was 9-7 (95% CI: 124, 75’S).

*Person-years of observation. f95% Confidence interval. Table 2: Incidence rates and rate ratios for adverse reactions

drug

1325

The mean duration of therapy before the diagnosis of adverse drug rash was 24 (12) days. The most common cutaneous presentation among the STH subjects was a diffuse morbilliform exanthem (7). 2 subjects had maculopapular rash involving mucous membranes, vesicles, and bullae; one patient progressed to toxic epidermal necrolysis and died despite supportive care and the immediate cessation of antituberculous therapy and another developed pruritis that resolved when the drug was withdrawn. Treatment was withdrawn in each case as soon as the rash was noted by the medical staff, and subjects with cutaneous reactions remained off STH therapy for an average of 24 (12) days. 5 subjects were switched to RHZ therapy because of the toxicity associated with STH. Because reactions were mild, 5 subjects were rechallenged with STH without incident. The only cutaneous reaction that occurred in the RHZ group was a maculopopular rash that was associated with oral lesions. The antituberculous drugs were stopped and the patient had prednisone therapy; the rash resolved within 29 days. The patient completed a course of therapy with ethambutol, rifampin, and pyrazinamide. We saw two episodes of jaundice in subjects who received STH. In one patient, therapy was stopped after 31 days and restarted 20 days later without recurrence of symptoms. In the other patient, STH was stopped 24 days into treatment and again resumed 59 days later without further incident. Development of cutaneous toxicity was associated with the size of the Mantoux skin test reaction at the start of therapy, the absolute lymphocyte count, and an atypical chest radiograph on presentation. Subjects who developed an adverse drug reaction had smaller reactions to the Mantoux skin test than subjects who did not develop drug reactions (3-2 mm vs 11-3 mm, p<0-005). 8 of the 11 subjects (73%) with cutaneous drug reactions were nonreactive to PPD (0 mm induration) compared with 44 of 180 subjects (25%) without adverse drug reaction The relative risk for an adverse cutaneous (p<0-001). in subjects with anergy to PPD compared to reaction drug those without anergy to PPD was 7-0 (95% CI: 1-94, 25-5). The mean absolute lymphocyte count in subjects with cutaneous drug reactions was lower than in subjects without cutaneous drug reactions (1150 vs 1970 cells/mL, 8 of the 11 subjects (73%) with cutaneous p<0-001). reactions had atypical chest X-rays compared with 70 of 174 (40%) without cutaneous reactions (p<0-05). Age, gender, HIV-related complications, haemoglobin, white blood cell count, radiographic extent of disease, and lung zone scores were not associated with cutaneous drug reactions. an

therapy Of the 171 subjects who were alive after two months of therapy, sputum specimens were available for AFB smears from 146 subjects (85%) and for mycobacterial cultures from 131 subjects (77%). Most subjects without sputum samples were unable to produce sputum; there were no differences in clinical parameters between subjects providing sputum for smear or culture and those who did not. The 13 subjects with culture negative clinical tuberculosis were excluded from this analysis. A greater proportion of subjects receiving RHZ showed a reduction in the quantity of AFB in sputum by two months than STH (90% vs 76%, p<0-05, table 3). Similarly, a greater of subjects receiving RHZ had negative proportion Response

1326

to

*p<0.05. tSputum AFB

smears were

available for 66

subjects receiving STH and 80 receiving

RHZ.

tSputum mycobacterial

cultures

were

available for 57

subjects receiving STH

and 74

receiving RHZ.

**p<0.001. Table 3: Sputum microscopy and two months of therapy

mycobacterial cultures after

sputum cultures for M tuberculosis within

months than subjects receiving STH (74% vs 37%, p<0-001, table 3). Of the 44 subjects who died while on therapy after the first month of treatment, 34 subjects (77%) had follow-up sputum specimens available for culture or staining; 21 were from the STH arm and 13 from the RHZ arm. Of these 34 subjects, 28 (82%) showed partial or complete clearing of AFB on sputum stain, 17 in the STH arm and 11in the RHZ arm. 29 had cultures available, 17 from the STH arm and 12 from the RHZ arm. Of these, 19 (66%) were culture-negative within 2 months of starting therapy. A greater proportion of subjcts who received RHZ were culture negative compared with subjects who received STH (92% vs 47%, respectively; p=0-02). Table 1 shows that at the start of therapy, the extent of disease on chest radiograph was similar for both the STH and RHZ arms as most patients had far advanced or moderately far advanced disease (table 1). Of the 179 subjects who survived longer than one month, 168 (94%) had at least one chest radiograph taken during follow-up that could be compared with the baseline film, 76 patients received STH and 92 received RHZ. There were no statistically significant differences between the two treatment groups with regard to serial changes in lung zone scores or in the extent of disease on chest X-ray two

(table 4). Discussion Our study showed that treatment with STH, compared with RHZ, gave higher rates of adverse drug reactions, a higher proportion of culture-positive patients after two months of therapy, and a worse prognosis. These findings indicate that RHZ is a safer and more effective regimen than STH in HIV-infected individuals with active tuberculosis. Other studies have shown that cutaneous reactions to STH are more common and often more severe in HIV-positive than in HIV-negative patients with tuberculosis.15.16 However, the safety of regimens with and without thiacetazone has not been established in HIVinfected patients with tuberculosis. Because HIV infection is linked to greater risk of adverse drug reactions to a wide variety of medications,21,22 a formal comparison of alternative regimens was needed. In our study, adverse drug reactions were 12 times more likely with STH than with RHZ. Although this estimate of risk lacked precision, the result was statistically significant. Even at the lower confidence interval, adverse drug reactions were 50% more likely in STH arm than RHZ arm. The spectrum of adverse reactions ranged from mild to severe. Mild cutaneous reactions were most common and accounted for most of the risk associated with STH. The more

Table 4: The

radiographic extent of disease

after 2 months of

therapy reactions included reversible hepatitis and toxic epidermal necrolysis. The frequency and severity of cutaneous drug reactions ascribed to STH were comparable to previously published results in HIVsevere

positive patients. In this study, the size of the Mantoux skin reaction, the absolute lymphocyte count, and an atypical presentation on chest radiograph were associated with the development of adverse drug reactions to antituberculous therapy. Nunn et a115 reported that hypersensitivity to thiacetazone associated with anergy to tuberculin among HIVinfected individuals,15 but they could not estimate the risk because Mantoux skin tests were not uniformly available on subjects at the start of therapy. The findings of our study indicate that patients with anergy to PPD were 4 times more likely to develop a cutaneous drug reaction than patients without anergy to PPD. We also found that cutaneous drug reactions were associated with lower absolute lymphocyte counts and atypical presentation of tuberculosis on chest X-ray. These results suggest that in HIV-positive patients greater immunosuppression, as indicated by Mantoux reaction, absolute lymphocyte count, and atypical chest X-ray appearance, is predictive of cutaneous drug reactions to antituberculous therapy. After the first two months of treatment, the response to therapy was better with RHZ than with STH. RHZ patients compared with STH were more likely to show a reduction in the quantity of organisms seen on AFB smear, although a majority of patients in both treatment arms had reductions in the number of organisms seen on acid-fast smear. As for sputum cultures at two months, a significantly greater proportion of patients on RHZ had sterilised the sputum compared with patients on STH. These results are consistent with earlier studies that have shown enhanced antimicrobial activity of rifampicin or pyrazinamide containing regimens compared with thiacetazone.23 A recent prospective study from Kenya 25 suggests that patients treated with rifampicin-containing regimens have improved survival when compared with thiacetazonecontaining regimens. When controlling for confounding variables that may influence survival, such as age, sex and education, the difference in survival between the regimens was of only borderline statistical significance. Moreover, because this study was observational in design and was

because

therapy was not randomly allocated, unrecognised confounders, such as severity of HIV infection, mitigated any observed survival difference. In our study in which treatment was randomly assigned, one-year survival

STH

greater in the RHZ arm than in the 65%) but the difference did not reach

was

(73% vs significance. However, at baseline, arm

statistical

reaction and the

mean

the mean size of the Mantoux absolute lymphocyte count were

lower in RHZ group than STH group, indicating that the RHZ arm may have had a greater degree of immunosuppression at the start of therapy. When controlling for these baseline differences in a Cox regression analysis, there was a survival benefit for RHZ compared with STH. The risk of death in STH subjects was 1-6 times greater than RHZ subjects. The beneficial effect of rifampicin-containing regimens is consistent in both studies and strongly suggests that treatment with these regimens will improve survival. The observed survival distribution in our study is comparable to the survival of HIV-infected individuals with active tuberculosis reported from both industrialised and developing countries. The overall survival proportion at one year in this study (70%) is consistent with one-year survival seen in other studies in the US8,26,27 and in Africa."25 However, poor survival occurs despite evidence of an adequate response to antituberculous therapy. This result supports the hypothesis that active tuberculosis may modify the course of HIV infection by stimulating the release of cytokines, such as tumour necrosis factor, which may enhance HIV replication in infected tissues. Previous studies have shown that HIV-infected individuals with drug-sensitive tuberculosis often die of other HIVrelated complications and not tuberculosis its elf. 8,26 Indeed, there is now epidemiological evidence that the frequency of opportunistic infections increases in HIVinfected patients with tuberculosis and that overall survival is reduced compared with HIV-infected patients without tuberculosis but with the same degree of immunosuppression as measured by CD4+ lymphocyte count.28 The results of this study raise difficult issues for developing countries with limited financial resources for tuberculosis treatment. In HIV-positive patients, the risk and severity of cutaneous toxicity due to thiacetazone are greatly increased, arguing strongly for the use of alternative regimens. This study provides specific evidence that an alternative regimen, RHZ reduces the risk of drug-related complications and may prolong survival. In deciding upon the optimal treatment regimen for pulmonary tuberculosis in developing countries where HIV infection is prevalent, the potential benefits of alternative therapies must be weighed against their overall costs. In a formal cost-effectiveness analysis from the World Bank,35 the cost per HIV seronegative patient was greater with rifampicin-containing regimens than thiacetazone-containing regimens, but when examining the cost per effective treatment course, RHZ short-course therapy was most cost-effective. This analysis only accounted for financial costs and did not assess the effects of alterations in survival, the risk of fatal complications associated with thiacetazone treatment, or the impact of recurrent infection on perpetuating transmission of tuberculosis in the community. Moreover, the analysis did not account for the high prevalence of HIV infection in developing countries. If these factors were included in the cost-effectiveness analysis, it is likely that rifampicincontaining regimens would be preferable, even when considering the cost per patient. There are several limitations to this study that bear upon the interpretation of the results. First, the large block size for randomisation led to unequal assignment of patients to treatment arms. We do not think that the observed differences in Mantoux reaction size and absolute lymphocyte count are explained by the 1327

unbalanced assignment because consecutive subjects were randomised. Second, the conclusions regarding the clearance of bacilli from the sputum must be interpreted with caution because specimens were not available for all surviving subjects. The two most likely reasons for missing data were inability to produce sputum, suggesting response to therapy, or death. Third, because the study was not blinded, it is possible that detection bias occurred through greater surveillance for outcomes, particularly adverse outcomes, in the STH arm. We feel this potential bias was minimised because we evaluated subjects at predetermined intervals. We conclude that RHZ is a more effective and safer regimen than STH in HIV-infected individuals with active pulmonary tuberculosis. Rifampicin-containing shortcourse chemotherapy should be given to HIV-infected individuals to reduce drug toxicity and to prolong

into thiacetazone (thioacetazone*) side-effects. Tubercle 1966; 47: 33-73. 11 Ferguson GC, Nunn AJ, Fox W, Miller AB, Robinson DK, Tall R. A second international co-operative investigation into thiacetazone sideeffects. Tubercle 1971; 52: 166-81. 12 A Cooperative Study in East African Hospitals, Clinics and Laboratories with the Collaboration of the East African and British Medical Research Councils. Isoniazid with thiacetazone (thioacetazone) in the treatment of pulmonary tuberculosis in East Africa. Third report of fifth investigation. Tubercle 1973; 54: 169-79. 13 Hira SK, Wadhawan D, Kamanga J, et al. Cutaneous manifestations of human immunodeficiency virus in Lusaka, Zambia. Am Acad 19: 451-57. 14

among HIV-seropositive patients with tuberculosis treated with thiacetazone. Weekly Epidemiologzcal Record No 1/2, 10 January 1992. 15 Nunn P, Kibuga D, Gathua S, et al. Cutaneous hypersensitivity reactions due to thiacetazone in HIV- seropositive patients treated for tuberculosis. Lancet 1991; 337: 627-30. 16 Dukes

CS, Sugarman J, Cegielski JP, Lallinger GJ, Mwakyusa DH. Severe cutaneous hypersensitivity reactions during treatment of tuberculosis in patients with HIV infection in Tanzania. Trop GeogrMed

survival. Makere

University-Case Western University Research Collaboration. Mycobacteriology Laboratory: T Aisu, A Morrisey. Data Management and Analysis: D Hom, C Daylally. Advisory Board: P Eriki, T Daniel. Home Visitation: J Nakibali, S Nyole. Study Coordination and Monitoring: R Wallis, K Edmonds. The authors thank staff of Makerere University—Case Western Reserve University Research Collaboration and the Ugandan National Tuberculosis and Leprosy Programme. This work was funded by a cooperative agreement from the Centers for Disease Control and Prevention (CCU 501881).

References 1

2

3 4

5

6

7

8

9 10

Styblo K. Overview and epidemiologic assessment of the current global tuberculosis situation with an emphasis on control in developing countries. Rev Infect Dis 1989; 11: (suppl 2) : S339-46. Eriki PP, Okwera A, Aisu T, Morrissey AB, Ellner JJ, Daniel TM. The influence of human immunodeficiency virus infection on tuberculosis in Kampala, Uganda. Am Rev Respir Dis 1991; 143: 185-87. DeCock KM, Soro B, Coulibaly IM, Lucas SB. Tuberculosis and HIV infection in sub-Saharan Africa. 4AM 1992; 268: 1581-87. Di Perri G, Danzi MC, De Checchi G, et al. Nosocomial epidemic of active tuberculosis among HIV-infected patients. Lancet 1989; 2: 1502-04. Daley CL, Small PM, Schecter GF, et al. An outbreak of tuberculosis with accelerated progression among persons infected with the human immunodeficiency virus: an analysis using restriction-fragment-length polymorphisms. N Engl ’ Med 1992; 326: 231-35. Selwyn PA, Hartel D, Lewis VA, et al. A prospective study of the risk of tuberculosis among intravenous drug users with human immunodeficiency virus infection. N Engl_7 Med 1989; 320: 545-50. Moreno S, Baraia-Etxaburu J, Bouza A, et al. Risk for developing tuberculosis among anergic patients infected with HIV. Ann Intern Med 1993; 119: 194-98. Small PM, Schecter GF, Goodman PC, Sande MA, Chaisson RE, Hopewell PC. Treatment of tuberculosis in patients with advanced human immunodeficiency virus infection. N Engl _7 Med 1991; 324: 289-94. Keers RY. Pulmonary Tuberculosis. A Journey Down the Centuries. Bailliere Tindall, London: Cassell Ltd, 1978. Miller AB, Fox W, Tall R. An international co-operative investigation

1328

D-e-rmatol 1988;Organization, Geneva. Severe hypersensitivity reactions World Health

1992;44:308-11. 17 Perriens JH, Colebunders RL, Karahunga C, et al. Increased and tuberculosis treatment failure rate among human

mortality

immunodeficiency virus (HIV) seropositive compared with HIV seronegative patients with pulmonary tuberculosis treated with "standard" chemotherapy in Kinshasa, Zaire. Am Rev Respir Dis 1991; 144:750-55. 18

Kanofsky DA, Burchenal JH. The clinical evaluation of chemotherapeutic agents in cancer. In: MacLeod CM, ed. Evaluation of chemotherapeutic agents. New York: Columbia University Press, 1949:191-205.

19 Classification of Pulmonary Tuberculosis. Diagnostic standards and classification of tuberculosis, chapter 6. New York: National Tuberculosis and Respiratory Disease Association, 1969. 20 International Union Against Tuberculosis and Lung Disease. Technical Guide for Sputum Examination for Tuberculosis by Direct Microscopy. Bull IUATLD 1986; 61 (suppl 2) : 1-16.

Coopman S, Johnson R, Platt R, Stern R. Cutaneous disease and drug reactions in HIV infection. N Engly Med 1993; 328: 1670-74. 22 Gordin F, Simon G, Wofsy C, Mills J. Adverse reactions to trimethoprim-sulfamethoxazole in patients with the acquired immunodeficiency syndrome. Ann Intern Med 1984; 100: 495-99. 21

23 Mitichson D. Basic mechanisms of chemotherapy. Chest 1979; 76 (suppl): 771-81. 24 Hawken M, Nunn P, Gathua S, et al. Increased recurrence of tuberculosis in HIV-1-infected patients in Kenya. Lancet 1993; 342: 332-37.

P, Brindle R, Carpenter L, et al. Cohort study of human immunodeficiency virus infection in patients with tuberculosis in Nairobi, Kenya: analysis of early (six-month) mortality. Am Rev Respir Dis 1992; 146: 849-54. 26 Chaisson RE, Schecter GF, Theuer CP, Rutherford GW, Echenberg DF, Hopewell PC. Tuberculosis in patients with the acquired immunodeficiency syndrome: clinical features, response to therapy, and survival. Am Rev Respir Dis 1987; 136: 570-74. 27 Stoneburner R, Laroche E, Prevots R, et al. Survival in a cohort of human immunodeficiency virus-infected tuberculosis patients in New York City. Arch Intern Med 1992; 152: 2033-37. 28 Whalen C, Lahart C, Hom D, Simberkoff M, Ellner J, Horsburgh C. Accelerated clinical course of HIV infection after tuberculosis. Am J Resp Crit Care Med (in press). 29 Barnum HN. Cost savings from alternative treatments for tuberculosis. 25 Nunn

Soc Sci Med 1986; 23: 847-50.