High levels of neurological involvement but low mortality in miliary tuberculosis: a 6-year case-series from the UK
- Navin Venkatraman1,
- Thomas King1,
- David Bell1,
- Gerrit Woltmann2,
- Martin Wiselka1,
- Ibrahim Abubakar3,4 and
- Manish Pareek1,4,5,6⇑
- 1Dept of Infection and HIV Medicine, University Hospitals of Leicester, Leicester, UK
- 2Dept of Respiratory Medicine, University Hospitals of Leicester, Leicester, UK
- 3Research Dept of Infection and Population Health, University College London, London, UK
- 4TB Section, Public Health England, London, UK
- 5Dept of Infection, Immunity and Inflammation, University of Leicester, Leicester, UK
- 6Dept of Health Sciences, University of Leicester, Leicester, UK
- Manish Pareek, Dept of Infection, Immunity and Inflammation, University of Leicester, Jarvis Building, Leicester Royal Infirmary, Leicester, LE1 5WW, UK. E-mail: mp426{at}le.ac.uk
Abstract
Miliary TB: high level of CNS involvement best investigated by head MRI which suggests re-evaluation of NICE guidance http://ow.ly/X3CDK
To the Editor:
Tuberculosis (TB) remains one of the biggest global health challenges. Whilst the greatest burden of active disease is seen in Asia and Africa [1], TB remains a significant issue in the UK. Miliary TB is one of the severest manifestations of TB disease [2–4]. Up-to-date clinicopathological data on miliary TB from the developed world are lacking. We undertook a comprehensive 6-year review (2007–2012) of cases presenting to a single UK centre with an ethnically diverse population with high levels of population exchange with the Indian Subcontinent and Africa. Miliary TB was defined as the presence of miliary nodules on thoracic imaging in patients who presented with symptoms compatible with the diagnosis and either culture of Mycobacterium tuberculosis complex or culture-negative patients with clinical and/or histological features compatible with TB who were commenced on a course of antituberculous therapy (ATT).
42 cases were included in the final analysis (median age 48.5 years and 59.5% males). Most miliary TB cases occurred in foreign-born patients from the Indian subcontinent; 60% of cases in migrants occurred within 5 years of arrival to the UK. HIV testing was performed in 38 (90.5%) out of 42 patients and four (10.5%) out of 38 patients were positive (all foreign-born: Sub-Saharan African, n=3; Indian Subcontinent, n=1).
Overall mortality was 7.1% (three out of 42 patients); two patients died during their course of ATT and one patient died before starting treatment. All patients had samples sent for microbiological culture with bronchoalveolar lavage (BAL) (27 (64.3%) out of 42 patients) and sputum (25 (59.5%) out of 42 patients) being the most commonly positive samples. The organism was isolated in 32 (76.2%) out of 42 patients. BAL and sputum specimens were culture positive in 15 (55.6%) out of 27 patients and 18 (72%) out of 25 patients, respectively. Four (9.5%) out of 42 patients were smear-positive for acid fast bacilli. All organisms were fully sensitive to first-line ATT. When we examined the monocyte:lymphocyte ratio we found that it was raised at presentation/pre-treatment (median 0.47 (interquartile range) 0.33–0.73) and significantly decreased at the end of treatment (0.25 (0.16–0.27); p=0.0001); this was largely driven by the resolution of lymphopenia.
Assessment of central nervous system (CNS) involvement with a lumbar puncture and/or neuroimaging was undertaken in a high proportion (39 (92.9%) out of 42) of patients (table 1). Of these 39 patients, 28 (71.8%) underwent neuroimaging and 31 (79.5%) had a lumbar puncture performed (table 1). We found a relatively high level (38.5%) of neurological involvement, as defined by radiological or cerebrospinal fluid (CSF) evidence of CNS disease (table 1). In our analysis, magnetic resonance imaging (MRI) scans were significantly more likely to be abnormal than computed tomography (CT) scans (13 (65%) out of 20 patients versus two (14%) out of 14 patients (p=0.003)), particularly in identifying tuberculomas which were seen in 55% of cases. Just under three-quarters (31 out of 42) of patients in our study underwent a lumbar puncture; this is a much higher proportion in comparison to previous studies. CSF abnormalities were only detected in five patients overall. Only one patient with normal neuroimaging had an abnormal CSF; patients with an abnormal CSF did, however, have a high yield for culture positivity.
Sensitivity for identifying CNS involvement for the different modalities (CT, MRI and lumbar puncture) is outlined in table 1. Overall, MRI was significantly more sensitive (92.9%, 95% CI 66.1–99.8%) than both CT (40.0%, 95% CI 5.3–85.3%; p=0.013) and lumbar puncture (41.7%, 95% CI 15.2–72.3%, p=0.005). Amongst 12 patients with CNS involvement who were investigated with both neuroimaging and lumbar puncture, five (41.7%) had symptoms/signs of neurological involvement. All five patients underwent a lumbar puncture and neuroimaging (as per National Institute for Health and Clinical Excellence (NICE) guidelines) [5]. There were three (60%) abnormal lumbar punctures and four (80%) patients had abnormal neuroimaging (p=1.0). In the seven asymptomatic patients, where NICE recommends lumbar puncture only [5], lumbar puncture was only abnormal in two (28.6%) patients whereas neuroimaging was abnormal in all seven (100%) patients (p=0.02).
This study does have a number of limitations. It is a relatively small, retrospective cohort of patients from a single regional centre in the UK. However, our region does have a relatively high fraction of national TB cases and the study does give insight into the investigation and outcome of patients with miliary TB in a developed world setting. We also found no evidence of temporal change in the clinical management of miliary TB patients. The overall mortality (7.1%) was low in comparison to other modern case series and national mortality figures (11.7% in 2011) [6]. However, most of the large case series have been conducted in the developing world with differing resources, TB prevalence and socioeconomic factors predisposing to the development of miliary TB.
A relatively high proportion of patients in this developed world setting had evidence of CNS involvement on imaging and/or lumbar puncture, which is in keeping with the known pathology and complications of miliary TB [7]. Our findings strongly indicate that, even if an initial head CT is performed for simplicity and logistical reasons, brain MRI should be performed in all patients with miliary TB in resource-rich settings, irrespective of the presence of neurological manifestations. This is in stark contrast to current NICE guidelines, which only recommend neuroimaging for patients with CNS symptoms or signs [5]. The utility of lumbar puncture as a diagnostic tool in previous studies varies greatly and was often restricted to those with neurological features; culture positivity rates from CSF ranged from 30% to 60% in previous studies in comparison to 13% (four out of 30 patients) in our study. Patients in this study may have presented earlier, and there was a clear discrepancy between MRI results that showed abnormalities in 65% of patients compared to lumbar punctures which were only abnormal in 16.7%, suggesting that CSF changes occur relatively late in the presentation of CNS disease in miliary TB.
This study showed that the monocyte:lymphocyte ratio was raised at presentation/pre-treatment and significantly decreased at the end of treatment. Sabin et al. [8] showed that the monocyte:lymphocyte ratio was predictive of the severity of experimental Mycobacterium bovis infection in rabbits. This observation has recently been followed up in large cohorts of HIV co-infected patients in South Africa and active TB patients in China, where extremes in monocyte:lymphocyte ratio predicted subsequent development of TB in this group and the initiation of TB therapy significantly reduced the monocyte:lymphocyte ratio [9, 10]. These observations highlight the importance of the monocyte:lymphocyte ratio in TB pathogenesis; whether these changes are a product of TB infection itself, or a reflection of individual predisposition to development of active disease requires further evaluation in diverse populations and TB phenotypes.
In conclusion, we emphasise the importance of awareness and early diagnostic investigations in high-risk groups to confirm the diagnosis and identification of CNS involvement with a brain MRI in all patients with miliary TB. This clinical approach is important as it allows the optimal management, particularly in terms of the duration of drug treatment, of patients with neurological involvement thereby preventing relapse and augmenting efforts to achieve TB elimination [11, 12]. Despite a high prevalence of co-existing CNS involvement, overall mortality was low; this may have been due to a high index of suspicion in an ethnically diverse population and early initiation of therapy [13]. Obtaining specimens for culture is imperative and has a high yield. Finally, the monocyte:lymphocyte ratio appears to be of value in monitoring treatment response and warrants further evaluation in larger studies as a marker of clinical response in patients being treated for active disease.
Footnotes
Conflict of interest: None declared.
- Received October 14, 2015.
- Accepted December 20, 2015.
- Copyright ©ERS 2016