Abstract
The aim of this study was to describe atypical cerebrospinal fluid (CSF) alterations in tuberculous meningitis (TBM) and to analyse the differences in outcome between patients with typical and atypical profiles. We did a retrospective study during the period of 2000 to 2005 including the cases of TBM assisted in a referral centre for infectious diseases in Rio de Janeiro State, Brazil. Neutrophilic plecytosis at the first spinal tap was found in 32.4% of TBM patients, who had a worse outcome when compared with those patients with typical CSF profiles. One factor that might have a major impact was the delay in starting empirical treatment (27.5 versus 11.6 days). We conclude that, in cases with clinical and epidemiological data compatible with TBM but with an atypical CSF profile, empirical treatment should be considered if CSF culture and direct examination for bacteria are negative.
Introduction
TB is considered by the World Health Organization to be a public health emergency. It is estimated that 30% of the global population is infected by Koch's bacillus, with 9.2 million new cases detected in 2006. 1 The central nervous system (CNS) is the third most common site of extrapulmonary TB and tuberculous meningitis (TBM) is the most frequent manifestation of CNS TB. 2,3
In middle and low income countries the diagnosis of this condition relies mainly on the clinical/epidemiological history and on the typical CSF profile (mononuclear pleocytosis, low glucose and elevated protein levels), as direct examination and cultures for Mycobacterium tuberculosis have a low yield, and culture takes a long time to become positive. 4 Furthermore, in these settings molecular biology techniques for CNS TB diagnosis are not widely available.
In some cases, atypical alterations in CSF profile can occur such as an acellular CSF in 3% to 6% 5 or neutrophilic predominance at the first spinal tap in 15% to 21.3%. 6,7 This last condition mimics bacterial meningitis resulting in a delay of the TBM diagnosis and specific treatment. The outcome for TBM patients with neutrophilic pleocytosis is unknown to date. The aim of this study is to highlight the importance of the recognition of atypical forms of TBM presentations, especially neutrophilic meningitis.
Material and methods
Study setting
The Instituto Estadual de Infectologia São Sebastião is a public institution, which is the referral centre for infectious diseases in Rio de Janeiro State, Brazil. The majority of patients are from the metropolitan area of the state, which has a population of 16 million people.
Subjects
We did a retrospective study of cases assisted during inpatient care over a period of five years (2000 to 2005). The files of the patients were selected if they fulfilled at least two of the following criteria: Meningeal signs associated with subacute or chronic evolution (two to four weeks); Typical CSF alterations (mononuclear pleocytosis, low level of glucose and elevated protein level); Initial neutrophilic pleocytosis of CSF with treatment failure for acute bacterial meningitis; Empirical treatment for TBM based on clinical, epidemiological grounds evolving with good clinical and laboratorial response; A clinical picture compatible with TBM along with positive acid fast bacilli smear or culture for M. tuberculosis in sputum, blood or CSF. patients with typical alterations of CSF parameters (pleocytosis with mononuclear predominance, high protein and low glucose levels); patients with atypical CSF profile (neutrophilic predominance at first tap or acellular CSF, high protein and low glucose levels).
Medical charts were reviewed and data on age, sex, signs and symptoms, co-morbidities, initial CSF analysis results, drug treatment regimens, complications and clinical outcome at the time of hospital discharge were collected. A lumbar puncture was performed on all patients and according to their CSF pattern we split the sample into two groups:
All patients were treated for TBM with a three-drug regimen including rifampin (600 mg/day), isoniazid (400 mg/day) for nine months and pyrazinamid (2000 mg/day) in the first two months for patients who weighed more than 45 kg.
8
Data analysis
Variables collected from patients' charts were inserted in a statistical program (SPSS 13.0 for windows) and analysed for central tendency and dispersion measures. Chi-square tests (Fischer's exact test for small samples) were performed to compare categorical variables and Mann-Whitney U-tests for continuous variables. All results were regarded as significant if a P value < 0.05 was obtained.
Results
Demographics
We included 37 patients in the study period. Men represented 73% (27/37) of the cases; the mean age at onset of disease was 28.2 years (range 1 to 80 years). All patients were undergraduate workers, most living in the poorest areas of the state (52.4%).
CSF analysis
Patients with typical alterations of CSF parameters represented 64.8% (24/37) of the sample and in the atypical group 32.4% (12/37) of patients presented with a neuthrophilic predominance at the initial tap. One patient (2.8%) had an acellular CSF (protein level 85 mg/dL and glucose of 50 mg/dL) and a positive CSF culture for M. tuberculosis. Nine patients (24.3%) had HIV, six in the group of typical TBM profile and three in the atypical group. The CSF alterations of each form are summarized in Table 1.
Cerebrospinal fluid (CSF) alterations observed in 36 patients during a five-year period (2000–2005), Instituto Estadual de Infectologia São Sebastião. (One patient with acellular CSF was excluded from this analysis)
Outcome
The period of time between the onset of symptoms and the beginning of treatment was 17.5 days for the typical profile group and 28.5 days for the atypical group (P = 0.46). Outcome variables also presented different means comparing the two groups; in the typical CSF profile group the length of hospital stay was 58.3 days, the need of intensive care was seen in 45.8% of the cases and case fatality rate was 4.2%. In the atypical group hospital stay was shorter, 47 days (P = 0.49), there was more admissions to the intensive care unit, 66.7% (P = 0.49) and case fatality rate was six times higher, 25% (P = 0.11). Incidence of neurological sequels at discharge was the same in both groups (40%).
Discussion
Our study showed that the diagnosis of TBM could be misleading. As typical alterations of CSF could be absent we showed that more than a quarter of the cases seen at our institution had an initial neutrophilic predominance. In this clinical scenario, clinicians usually start treatment for acute bacterial meningitis and only consider other diagnosis after negative bacterial cultures or treatment failure. We observed that patients with neutrophilic predominance at first CSF analysis wait longer for the correct treatment. This fact may explain the higher case fatality rate in this group of patients as seen in the study of Hosoglu and Verdon et al., who described this delay as an important predictor of death. 9,10
Confirmation of a clinical suspicion of TBM has always been difficult; low yields and slow growth of M. tuberculosis in culture make bacteriological diagnosis an unreliable tool to guide the beginning of treatment. 11 Diagnosis of TBM is based on clinical and epidemiological data and CSF typical profile. When this pattern is absent, it becomes even more difficult for the physician to consider TBM and the delay of treatment could worsen the prognosis. In one large case series study from Brazil, the authors highlighted a considerable proportion (about one-fifth of the sample) of cases with a initial neutrophilic predominance in CSF analysis. 6 In another study by Sutlas et al., 15% of their case series had a neutrophilic CSF at first tap. 7 The incidence of neutrophilic meningitis in both studies is very similar to our findings, but they did not analyse the possible differences between the two groups in other CSF laboratorial parameters and in case fatality rate. We also found a higher CSF culture positivity for M. tuberculosis in the atypical group, which could represent a criterion for severity.
Our study was performed in a reference centre for meningitis, which may have influenced the sample selection and the severity of the presentations. Another major limitation was the sample size, which certainly interfered with inferential statistic analysis.
Diagnosing TBM demands a high index of suspicious even when CSF alterations show a typical profile, and treatment delay is an important predictor of poor outcome. Our results emphasize the importance of prompt empirical treatment for TBM in patients with a compatible clinical and epidemiological data, even without the expected CSF alterations. Faster and more sensitive diagnostic procedures in poor income settings, as well as more research, are urgently needed in order to find better criteria for the diagnosis of TBM, especially those with atypical CSF profile at first spinal tap.
Footnotes
Acknowledgments
We are grateful to the medical documentation staff of Instituto Estadual de Infectologia São Sebastião for their assistance.