点击显示 收起
Oxford University Clinical Research Unit, Hospital for Tropical Diseases
Pham Ngoc Thach Hospital for Tuberculosis and Lung Disease, Ho Chi Minh City, Vietnam
Centre for Tropical Medicine, Nuffield Department of Clinical Medicine, John Radcliffe Hospital, Oxford, United Kingdom
Background. Tuberculous meningitis occurs more commonly in human immunodeficiency virus (HIV)infected individuals than in HIV-uninfected individuals, but whether HIV infection alters the presentation and outcome of tuberculous meningitis is unknown.
Methods.
We performed a prospective comparison of the presenting clinical features and response to treatment in 528 adults treated consecutively for tuberculous meningitis (96 were infected with HIV and 432 were uninfected with HIV) in 2 tertiary-care referral hospitals in Ho Chi Minh City, Vietnam. Logistic regression was used to model variables associated independently with HIV infection, 9-month survival, and the likelihood of having a relapse or an adverse drug event. Kaplan-Meier estimates were used to compare survival rates and times to fever clearance, coma clearance, relapse, and adverse events.
Results.
HIV infection did not alter the neurological presentation of tuberculous meningitis, although additional extrapulmonary tuberculosis was more likely to occur in HIV-infected patients. The 9-month survival rate was significantly decreased in HIV-infected patients (relative risk of death from any cause, 2.91 [95% confidence interval, 2.143.96]; P < .001), although the times to fever clearance and coma clearance and the number or timing of relapses or adverse drug events were not significantly different between the groups.
Conclusions.
HIV infection does not alter the neurological features of tuberculous meningitis but significantly reduces the survival rate.
Tuberculous meningitis is the most lethal form of infection with Mycobacterium tuberculosis. It is fatal in 30% of cases, despite treatment with antituberculosis chemotherapy, and it is responsible for severe disability in a similar proportion of survivors [1, 2]. Infection with HIV increases an individual's risk of developing all forms of tuberculosis, but the risk of developing extrapulmonary tuberculosis and tuberculous meningitis is increased even more [3]. Consequently, tuberculous meningitis has become a common and serious clinical problem, especially in populations with a high prevalence of tuberculosis and HIV infection.
The diagnosis and treatment of tuberculous meningitis are difficult regardless of the setting, but coinfection with HIV may present special challenges. There are limited data describing the influence that HIV infection has on the clinical presentation of, response to treatment in, and outcome of tuberculous meningitis. To date, the largest published study to have examined this issue included 53 Indian adults (22 were infected with HIV and 31 were uninfected with HIV) [4], and its findings suggested that HIV infection did not alter the presenting clinical features of tuberculous meningitis, although presenting cognitive dysfunction was more severe in HIV-infected patients. The findings suggested that HIV infection may alter the pathological features of tuberculous meningitis: basal meningeal enhancement and hydrocephalus on computed tomography of the brain were less common in HIV-infected adults than in HIV-uninfected adults, and post mortem histopathological assessment revealed fewer basal exudates and larger numbers of acid-fast bacilli in the cerebral parenchyma and meninges of HIV-infected patients. Other smaller case series have suggested that patients with HIV-associated tuberculous meningitis may present with lower leukocyte counts in peripheral blood and cerebrospinal fluid (CSF) [5] and may be more likely than HIV-uninfected patients to have concomitant active extrameningeal tuberculosis [6].
The impact that HIV infection has on responses to treatment and outcome also is unknown, because most studies have been too small to demonstrate significant differences in case-fatality rates and have reported assessments of hospitalized patients only. Katrak et al. compared the outcomes of HIV-infected and HIV-uninfected adults with tuberculous meningitis after 6 months of treatment and reported a significantly higher case-fatality rate in HIV-infected adults (8/22 [36%] HIV infected vs. 3/31 [10%] HIV uninfected; P = .036) [4], whereas other case series [3, 79] failed to demonstrate that HIV infection had a significant impact on in-hospital mortality caused by tuberculous meningitis. Karstaedt et al. reported significantly more treatment failures in HIV-infected patients than in HIV-uninfected patients [6], which suggests that HIV infection may influence responses to treatment and the likelihood of having a relapse, but these are the only available data on treatment responses. To address this important information gap, in the present study, we compared the presenting clinical features, the responses to treatment, and the 9-month outcomes in 96 HIV-infected adults and 432 HIV-uninfected adults consecutively treated for tuberculous meningitis in 2 hospitals in southern Vietnam.
PATIENTS, MATERIALS, AND METHODS
Setting and study participants.
All patients included in the present study were recruited to a randomized, placebo-controlled trial of adjunctive dexamethasone treatment in adults with tuberculous meningitis. Details about the methods of recruitment, treatment, and primary outcome assessment have been reported elsewhere [10].
The study participants were recruited consecutively from 2 centers in Ho Chi Minh City, Vietnam: Pham Ngoc Thach Hospital for Tuberculosis and the Hospital for Tropical Diseases. All adults (>14 years old) with definite, probable, or possible tuberculous meningitis were eligible to enter the study. The entry diagnostic criteria were the following:
1. Definite tuberculous meningitis was defined as clinical meningitis (nuchal rigidity and abnormal CSF parameters) and acid-fast bacilli in the CSF.
2. Probable tuberculous meningitis was defined as clinical meningitis and 1 of the following: suspected active pulmonary tuberculosis on the basis of chest radiography, acid-fast bacilli found in any sample other than from the CSF, and clinical evidence of other extrapulmonary tuberculosis.
3. Possible tuberculous meningitis was defined as clinical meningitis and at least 4 of the following: history of tuberculosis, predominance of lymphocytes in the CSF, illness of >5 days in duration, CSF : blood glucose ratio <0.5, altered consciousness, yellow CSF, and focal neurological signs.
On discharge from the hospital, patients were reclassified as having definite tuberculous meningitis if M. tuberculosis was cultured from the CSF or as not having tuberculous meningitis if another diagnosis was confirmed by microbiological or histopathological assessment, and these patients were excluded from the study.
The ethical and scientific committees of the hospitals, the Health Services of Ho Chi Minh City, and the Oxford Clinical Research Ethics Committee approved the study protocol. Written, informed consent to participate in the study was obtained from all patients or their relatives.
Investigations.
CSF cell counts were measured and biochemical analysis was performed by standard methods, and samples were stained and cultured by standard methods for pyogenic bacteria, fungi, and mycobacteria. Isolates of M. tuberculosis were tested for susceptibility to isoniazid, rifampicin, ethambutol, and streptomycin by use of the proportion method [11, 12]. All patients were tested for hepatitis B surface antigen (AxSYM version 2; Abbott Diagnostics) and antibodies to HIV (Determine HIV1/2; Abbott), and positive results were confirmed by Western blot analysis. For all HIV-infected adults, peripheral blood CD4 and CD8 lymphocyte counts were measured as soon as possible after randomization by use of flow cytometry (FACSCalibur; Becton Dickinson).
Treatment.
Patients previously untreated for tuberculosis received oral isoniazid (5 mg/kg), rifampicin (10 mg/kg), pyrazinamide (25 mg/kg to a maximum of 2 g/day), and intramuscular streptomycin (20 mg/kg to a maximum of 1 g/day) for 3 months, followed by oral isoniazid, rifampicin, and pyrazinamide at the same doses for 6 months. Ethambutol (20 mg/kg to a maximum of 1.2 g/day) was substituted for streptomycin in the regimen for HIV-infected patients and was added to the regimen for 3 months for patients previously treated for tuberculosis. Drugs were given by nasogastric tube to patients unable to swallow. None of the patients received antiretroviral drugs. All patients were randomized to receive either adjunctive dexamethasone treatment or placebo, as described elsewhere [10].
Assessment of outcome.
The attending physicians were responsible for enrolling the study participants, and a daily assessment of all inpatients by the principal investigator ensured uniformity of management between sites and accurate recording of clinical data in individual study notes. Particular attention was given to recording the daily Glasgow coma score, the daily high temperature, and the onset of new focal neurological signs. The coma clearance time was defined as the interval (in days) from randomization to the time when a score of 15 was given on the Glasgow coma scale and the score was repeated for >2 consecutive days. The fever clearance time was defined as the interval (in days) from randomization to the time when a patient had a maximum temperature of <37.5°C for >5 consecutive days. A relapse was defined as the onset of new focal neurological signs or a decrease of 2 points in the Glasgow coma score for 2 days, after >7 days of clinical stability or improvement at any time after randomization. All definitions were set a priori. Disability was assessed by 2 different questionnaires administered after 1, 2, 6, and 9 months of treatment, as described elsewhere [10]. All data were recorded prospectively into individual study notes and entered in an electronic database (Microsoft FoxPro; version 6.0) as soon as each patient completed follow-up, and they were double-checked before analysis.
Statistical analysis.
The analysis of the impact that adjunctive dexamethasone treatment has on death and disability was reported elsewhere [10]. For the purposes of the present study, adults were excluded from the analysis if they were not tested for HIV or an alternative diagnosis to tuberculous meningitis was confirmed. Continuous variables were compared by use of the Mann-Whitney U test, and categorical variables were compared by use of the 2 test (or Fisher's exact test, when appropriate). Variables associated with HIV infection (P < .1) by univariate analysis were selected for the multivariate analysis. Logistic regression was used to model variables that were independently predictive of HIV at presentation, relapse, adverse events, and death. A forward, stepwise selection procedure (to enter, P = .05; to remove, P = .01) was used to identify the best model. Allocation to treatment with dexamethasone or placebo was included as a covariate in each analysis, except when presenting clinical features were compared. Times to fever clearance, coma clearance, relapse, and discharge from the hospital were summarized in each treatment group by use of Kaplan-Meier estimates and were compared using the log-rank test. The analysis was performed with SPSS (version 10; SPSS).
RESULTS
Between 4 April 2001 and 29 March 2003, a total of 545 adults with tuberculous meningitis were recruited consecutively to a randomized controlled trial of adjunctive dexamethasone treatment [10]. Eleven of these 545 patients were not tested for HIV infection (9 died before testing could be arranged, and 2 declined testing), and an alternative diagnosis was confirmed for 6 patients (cryptococcal meningitis for 2, cerebral metastases for 2, cerebrovascular accident for 1, and Herpes simplex encephalitis for 1), leaving 528 patients available for analysis. HIV infection was confirmed in 96 patients (18.2%), 73% of whom reported injection drug use. Peripheral blood CD4 and CD8 lymphocyte counts were measured in 79 HIV-infected patients at the start of treatment; these measurements were not performed for 17 of these patients either because they died before testing (n = 8) or because of a shortage of reagents (n = 9). The median CD4 and CD8 lymphocyte counts were 67 × 106 lymphocytes/mL (range, 7 × 106694 × 106 lymphocytes/mL) and 487 × 106 lymphocytes/mL (range, 28 × 106998 × 106 lymphocytes/mL), respectively.
A comparison of the presenting clinical features in HIV-infected and HIV-uninfected patients is presented in table 1. Patients with HIV-associated tuberculous meningitis were significantly (P < .05) more likely to be younger, to weigh less, to be male, to have extrapulmonary/meningeal tuberculosis, and to have a lower Glasgow coma score, hematocrit level, peripheral blood leukocyte count (with fewer neutrophils), and plasma sodium level. Concentrations of aspartate transaminase and alanine aminotransferase were significantly higher in HIV-infected patients, and a greater proportion had hepatitis B surface antigenemia. Male sex (odds ratio [OR], 24.4 [95% confidence interval {CI}, 7.776.9]), younger age (OR, 0.90 [95% CI, 0.860.93]), extrapulmonary/meningeal tuberculosis (OR, 3.20 [95% CI, 1.258.22]), and lower hematocrit level (OR, 0.83 [95% CI, 0.770.90]) were associated independently with HIV-associated tuberculous meningitis at presentation.
M. tuberculosis was isolated from the CSF from a higher proportion of HIV-infected patients than HIV-uninfected patients (table 1). Drug-susceptibility testing of these isolates revealed that 47.5% (19/40) of those from HIV-infected patients were resistant to 1 first-line antituberculosis drug, compared with 39.1% (50/128) of those from HIV-uninfected patients (P = .344, 2 test). In particular, resistance to at least isoniazid and rifampicin (multidrug-resistant isolates) was found in a greater proportion of isolates from HIV-infected patients (12.5%) than from HIV-uninfected patients (3.9%).
The median length of follow-up was 273 days (range, 28442 days); 9 patients dropped out of the study 9 months after the start of treatment (1 HIV-infected patient and 8 HIV-uninfected patients). Regardless of dexamethasone use, HIV infection greatly decreased the chance of survival after tuberculous meningitis (figure 1). A total of 62 (64.6%) of 96 HIV-infected patients and 122 (28.2%) of 432 HIV-uninfected patients died by 9 months after the start of treatment (relative risk of death, 2.91 [95% CI, 2.143.96]). The clinical variables independently associated with death are presented in table 2. A lower CD4 lymphocyte count was associated with a worse outcome in HIV-infected patients by univariate analysis but not by multivariate analysis. Antituberculosis drug resistance was not included as a covariate in this analysis, because it would have restricted the data set to a maximum of 168 patients and because we have published elsewhere a detailed analysis of the impact that drug resistance has on the outcome of tuberculous meningitis [12]. However, table 3 presents the outcomes of HIV-infected and HIV-uninfected patients, grouped according to the drug susceptibility of the M. tuberculosis strain isolated from the CSF. The strength of the association between HIV infection and death caused by tuberculous meningitis was confirmed when outcomes were compared between patients with isolates susceptible to all first-line agents: 66.7% (14/21) of HIV-infected patients died, compared with 19.2% (15/78) of HIV-uninfected patients (P < .001, 2 test). The numbers of patients in the other groups defined according to drug susceptibility were too small to support significant associations with outcome (table 3), although it is clear that multidrug resistance is associated with a poor outcome, regardless of HIV infection status.
Disability was assessed 9 months after the start of treatment in 344 survivors. Surprisingly, HIV infection did not worsen the outcome in survivors: the proportion that had a complete recovery by 9 months after the start of treatment was similar in HIV-infected patients (21/32 [65.6%]) and HIV-uninfected patients (177/310 [57.1%]) (P = .601), and there was a trend toward a smaller proportion of severely disabled survivors among HIV-infected patients (2/34 [5.9%] vs. 53/310 [17.1%]; P = .092). Adjunctive dexamethasone treatment did not alter significantly the proportions of severely disabled survivors in either group (data not shown).
Fever cleared in a median of 8 days (range, 170 days) in HIV-infected patients and 9 days in HIV-uninfected patients (P = .973, log-rank test). Coma cleared in a median of 7 days in both groups (range, 170 days in HIV-infected patients and 181 days in HIV-uninfected patients; P = .824, log-rank test). Discharge from the hospital occurred significantly earlier for HIV-infected patients (median, 29 days [range, 7102 days] for HIV-infected patients vs. 51 days [range, 4276 days] for HIV-uninfected patients; P < .001, log-rank test]. Adjunctive dexamethasone treatment did not change significantly the time to any of these end points within the subgroups (data not shown).
The treatment of tuberculous meningitis was complicated by relapse of the disease in 71 (16.4%) of 432 HIV-uninfected patients and 18 (23.1%) of 78 HIV-infected patients (P = .584). The median time to relapse from the start of treatment was 37 days in HIV-uninfected patients and 58 days in HIV-infected patients (P = .106, log-rank test). Death followed relapse in 36 (50.7%) of 71 HIV-uninfected patients and 15 (83.3%) of 18 HIV-infected patients (P = .012). Multivariate analysis of all patients revealed that a miliary pattern on a chest radiograph, a longer duration of symptoms, and a lower Glasgow coma score were associated with relapse (table 4). Adjunctive dexamethasone treatment did not alter significantly the proportion, timing, or outcome of relapse in each group (data not shown).
A total of 59 (61.5%) of 96 HIV-infected patients and 262 (60.6%) of 432 HIV-uninfected patients had an adverse event (P = .930). The median time to an adverse event was 17 days (range, 1214 days) for HIV-infected patients and 16 days (range, 1246 days) for HIV-uninfected patients (P = .835, log-rank test). Hepatitis was the most common adverse event observed (table 5), and it occurred in a similar proportion of patients (16/96 [16.7%] of HIV-infected patients vs. 66/432 [15.3%] of HIV-uninfected patients; P = .734) and at a similar time (median, 13 days [range, 1118 days] in HIV-infected patients vs. 14 days [range, 1246 days] in HIV-uninfected patients; P = .774, log-rank test) in both groups. Multivariate analysis of all patients revealed that concomitant active extrapulmonary/meningeal tuberculosis was independently associated with the development of hepatitis (OR, 2.49 [95% CI, 1.444.31]). Adjunctive dexamethasone treatment did not alter significantly the number, type, timing, or outcome of adverse events within the HIV-infected and HIV-uninfected subgroups, although it was associated with a reduced number of severe eventsin particular, severe hepatitisin all patients [10].
DISCUSSION
Elsewhere, we have reported the results of a randomized, double-blind, placebo-controlled trial in 545 adults with tuberculous meningitis and with or without HIV infection [10]. Adjunctive dexamethasone treatment was associated with a significant reduction in the risk of death but was not associated with a significant reduction in the proportion of severely disabled survivors. A stratified analysis suggested that the effect of adjunctive dexamethasone treatment on survival was homogeneous across the subgroups defined according to HIV status (stratified relative risk of death, 0.78 [95% CI, 0.591.04]) [10].
The present study compared the presenting clinical features, response to treatment, and outcome in 96 HIV-infected and 432 HIV-uninfected patients with tuberculous meningitis. The results suggested that HIV does not alter the presenting neurological features of tuberculous meningitis but does alter the extracerebral findings. Our data suggested that these findings relate partly to the epidemiological pattern of HIV infection within the study population (young male injection drug users with a high prevalence of viral hepatitis) and partly to the effects of systemic immunosuppression (low weight, low hematocrit level, and high prevalence of extrapulmonary/meningeal tuberculosis).
But these data will give clinicians scant reassurance, given the poor 9-month survival rate in HIV-infected patients with tuberculous meningitis. Comparison of the survival curves (figure 1) reveals significant differences; most striking is the nearly constant case-fatality rate in HIV-infected patients throughout follow-up. We were unable to record accurate causes of death, and our ability to diagnose other opportunistic infections was limited. Therefore, because most of the patients presented with extremely low CD4 lymphocyte counts and antiretroviral drugs were not available, the differences between the survival curves may be explained by other undiagnosed, fatal opportunistic infections. However, other factors may have contributed to the high case-fatality rate. A higher proportion of M. tuberculosis isolates from HIV-infected patients was resistant to 1 first-line antituberculosis drug. We recently examined the effect that drug resistance has on outcome of tuberculous meningitis and concluded that, although the impact of multidrug resistance was devastating (the 9-month case-fatality rate was 100%), the consequences of isoniazid and/or streptomycin resistance on outcome was probably minimal [12]. The influence that HIV infection has on outcome of drug-resistant tuberculous meningitis is unknown, and the small sizes of the patient groups make it difficult to draw firm conclusions from the present study and a similar one [14]. However, the strong association between death and HIV infection in patients with tuberculous meningitis caused by M. tuberculosis strains that are susceptible to all first-line drugs suggests that factors other than drug resistance are responsible for the poor outcome of our HIV-infected patients.
Despite the detrimental effect that HIV infection has on survival, we observed no significant differences between the incidence of severe disability in HIV-infected and HIV-uninfected survivors after 9 months of treatment. This is a surprising and important finding that supports the likely influence of other opportunistic infections on survival after HIV-associated tuberculous meningitis and offers hope to those in whom these infections may be prevented or diagnosed more accurately and treated. Furthermore, there were no significant differences between other measurements of treatment response (fever clearance and coma clearance), which suggests that HIV infection does not compromise antituberculosis drug activity, despite its suppression of the immune system. Only time to discharge from the hospital was significantly shorter for HIV-infected patients, which is probably explained by the limited number of beds in the HIV ward and the need to discharge patients early, rather than by a faster recovery time in this group. More resources to treat tuberculous meningitis in this setting are urgently required.
Studies assessing the use of antiretroviral drugs will need to consider the frequency and timing of relapses and adverse events, because our data suggest that both parameters influence outcome [10]. Clinical deterioration after the administration of antituberculosis drugs is a common phenomenon in tuberculous meningitis, and it is often labeled "paradoxical" because it occurs during treatment and may have an inflammatory basis [15]. Corticosteroid treatment is suggested, although no controlled trials have addressed the issue and there are no widely accepted definitions of relapse. We prospectively applied a definition for relapse on the basis of our clinical experience. Although there was no significant difference between the frequency and timing of relapses between HIV-infected patients and HIV-uninfected patients, significantly more deaths occurred in HIV-infected patients once relapse occurred. As was discussed above, these deaths may have been caused by other undiagnosed opportunistic infections. Disseminated tuberculosis, diagnosed by a miliary pattern on a chest radiograph, was an independent risk factor for having a relapse in HIV-uninfected patients. Paradoxical inflammatory reactions are a more plausible explanation for the relapses in this group, because results of magnetic resonance imaging suggest that many of these patients have multiple, widespread intracerebral granulomas [16]; bactericidal drugs, such as isoniazid, may cause the diffuse release of bacillary contents and provoke inflammation and edema that could cause clinical deterioration.
Our data suggest that the type, frequency, and timing of adverse events during treatment are not affected by HIV infection. This is surprising, particularly given the significantly higher prevalence of hepatitis B infection in our HIV-infected patients, compared with that in our HIV-uninfected patients. We did not test for hepatitis C infection, but a high prevalence of this infection was likely, given that 73% of the HIV-infected patients were injection drug users [17]. A larger study is required to investigate this issue, particularly given the prospect of an increased risk of hepatitis drug reactions when antiretroviral drugs are used. Collectively, extrapulmonary/meningeal tuberculosis was independently associated with the development of hepatitis during treatment, and adjunctive dexamethasone treatment prevented severe hepatitis reactions [10]. These data suggest that some of these events may have an immune basis and may have resulted from the effect of disseminated and hepatic release of mycobacterial products after the initiation of antituberculosis chemotherapy.
In conclusion, the results of the present study suggest that HIV infection does not alter the neurological features of tuberculous meningitis but dramatically decreases the survival rate, although the incidence of severe disability in HIV-infected survivors may be equivalent to or less than that in HIV-uninfected survivors. Studies that assess the impact and optimal timing of antiretroviral treatment on outcome of HIV-associated tuberculous meningitis are urgently required. Nevertheless, early diagnosis and treatment of all patients before the onset of coma are likely to remain critical to a successful outcome of tuberculous meningitis.
Acknowledgments
We thank the doctors and nurses at Pham Ngoc Thach Hospital and the Hospital for Tropical Disease who cared for the patients; the administrative and laboratory staff of Pham Ngoc Thach Hospital, and, in particular, Dai Viet Hoa, Mai Nguyet Thu Huyen, Tran Huu Loc, and Pham Hoang Anh; and Tim Peto (Oxford University, Oxford, United Kingdom), for his advice regarding the design and execution of this study.
References
1. Thwaites GE, Tran TH. Tuberculous meningitis: many questions, too few answers. Lancet Neurol 2005; 4:16070. First citation in article
2. Garg RK. Tuberculosis of the central nervous system. Postgrad Med J 1999; 75:13340. First citation in article
3. Berenguer J, Moreno S, Laguna F, et al. Tuberculous meningitis in patients infected with the human immunodeficiency virus. N Engl J Med 1992; 326:66872. First citation in article
4. Katrak SM, Shembalkar PK, Bijwe SR, et al. The clinical, radiological and pathological profile of tuberculous meningitis in patients with and without human immunodeficiency virus infection. J Neurol Sci 2000; 181:11826. First citation in article
5. Thwaites GE, Chau TT, Caws M, et al. Isoniazid resistance, mycobacterial genotype and outcome in Vietnamese adults with tuberculous meningitis. Int J Tuberc Lung Dis 2002; 6:86571. First citation in article
6. Karstaedt AS, Valtchanova S, Barriere R, et al. Tuberculous meningitis in South African urban adults. QJM 1998; 91:7437. First citation in article
7. Dube MP, Holtom PD, Larsen RA. Tuberculous meningitis in patients with and without human immunodeficiency virus infection. Am J Med 1992; 93:5204. First citation in article
8. Yechoor VK, Shandera WX, Rodriguez P, et al. Tuberculous meningitis among adults with and without HIV infection: experience in an urban public hospital. Arch Intern Med 1996; 156:17106. First citation in article
9. Schutte CM. Clinical, cerebrospinal fluid and pathological findings and outcomes in HIV-positive and HIV-negative patients with tuberculous meningitis. Infection 2001; 29:2137. First citation in article
10. Thwaites GE, Nguyen DB, Nguyen HD, et al. Dexamethasone for the treatment of tuberculous meningitis in adolescents and adults. N Engl J Med 2004; 351:174151. First citation in article
11. Collins CH, Grange JM, Yates MD. Tuberculous bacteriology: organization and practice. 2nd ed. Oxford: Butterworth-Heinemann, 1997. First citation in article
12. Thwaites GE, Lan NTN, Dung NH, et al. Effect of antituberculosis drug resistance on response to treatment and outcome in adults with tuberculous meningitis. J Infect Dis 2005; 192:7988. First citation in article
13. British Medical Research Council. Streptomycin treatment of tuberculous meningitis. BMJ 1948; 1:58297. First citation in article
14. Seaworth BJ. It is too early to discount the contribution of isoniazid to the treatment of tuberculous meningitis. J Infect Dis 2005; 192:102. First citation in article
15. Afghani B, Leiberman JM. Paradoxical enlargement or development of intracranial tuberculomas during therapy: case report and review. Clin Infect Dis 1994; 19:10929. First citation in article
16. Thwaites GE, Farrar JJ. Miliary tuberculosis and meningitis. Practical Neurol 2002; 2:2301. First citation in article
17. Chau TTH, Mai NTH, Phu NH, et al. Malaria in injection drug abusers in Vietnam. Clin Infect Dis 2002; 34:131722. First citation in article