The current global situation for tuberculous meningitis: epidemiology, diagnostics, treatment and outcomes

Tuberculous meningitis (TBM) results from dissemination of M. tuberculosis to the cerebrospinal fluid (CSF) and meninges. Ischaemia, hydrocephalus and raised intracranial pressure frequently result, leading to extensive brain injury and neurodisability. The global burden of TBM is unclear and it is likely that many cases are undiagnosed, with many treated cases unreported. Untreated, TBM is uniformly fatal, and even if treated, mortality and morbidity are high. Young age and human immunodeficiency virus (HIV) infection are potent risk factors for TBM, while Bacillus Calmette–Guérin (BCG) vaccination is protective, particularly in young children. Diagnosis of TBM usually relies on characteristic clinical symptoms and signs, together with consistent neuroimaging and CSF parameters. The ability to confirm the TBM diagnosis via CSF isolation of M. tuberculosis depends on the type of diagnostic tests available. In most cases, the diagnosis remains unconfirmed. GeneXpert MTB/RIF and the next generation Xpert Ultra offer improved sensitivity and rapid turnaround times, and while roll-out has scaled up, availability remains limited. Many locations rely only on acid fast bacilli smear, which is insensitive. Treatment regimens for TBM are based on evidence for pulmonary tuberculosis treatment, with little consideration to CSF penetration or mode of drug action required. The World Health Organization recommends a 12-month treatment course, although data on which to base this duration is lacking. New treatment regimens and drug dosages are under evaluation, with much higher dosages of rifampicin and the inclusion of fluoroquinolones and linezolid identified as promising innovations. The inclusion of corticosteroids at the start of treatment has been demonstrated to reduce mortality in HIV-negative individuals but whether they are universally beneficial is unclear. Other host-directed therapies show promise but evidence for widespread use is lacking. Finally, the management of TBM within health systems is sub-optimal, with drop-offs at every stage in the care cascade.


Introduction
Tuberculous meningitis (TBM) is an extra-pulmonary form of tuberculosis (TB) characterised by sub-acute or chronic inflammation of the meninges as a result of invasion of the sub-arachnoid space by the bacilli M. tuberculosis. Other forms of central nervous system (CNS) TB such as tuberculoma, cerebral abscess, and spinal TB are not categorised as TBM, even though treatment is similar. TBM most commonly affects young children and individuals with human immunodeficiency virus (HIV). In the absence of TB treatment, TBM is uniformly fatal 1 and even with treatment, outcomes are poor and chronic neurodisability is common. Much of the damage caused by TBM is due to host-derived inflammatory responses, yet our understanding of these processes is limited. In this review we describe the current global situation for TBM in terms of our understanding of the pathogenesis and natural history, epidemiology, diagnosis, and treatment. We also explore how TBM is managed within health systems. Challenges in the field of TBM are shown in Table 1.

Pathogenesis
Primary infection occurs via inhalation of M. tuberculosiscontaining aerosolised droplets, followed by activation of neutrophils, dendritic cells and alveolar macrophages, which engulf the mycobacteria in the terminal alveoli. Infected cells then migrate to lymphoid tissue, resulting in activation of Th1 cells and production of pro-inflammatory cytokines, with resultant inflammatory changes in the lung parenchyma and vasculature 2 . If the mycobacteria reach the vasculature, haematogenous dissemination can occur, with the potential to invade the CNS 3-5 .
The pathogenesis of TBM continues to be debated. A key virulence feature of mycobacteria is the ability to invade the blood-brain and blood-cerebrospinal fluid (CSF) barriers. The invasion mechanisms are not clear, although in vitro and animal data suggest that M. tuberculosis may rearrange the actin of the layers 4,6 . It is also possible that a "Trojan horse" mechanism, by which M. tuberculosis is brought across the blood-brain barrier by infected macrophages and neutrophils, may occur 7 . Rich and McCordock described the development of a caseating "Rich" focus in the context of TBM pathogenesis 8,9 . It is suggested that the Rich focus is formed via activation of microglial cells and astrocytes once the bacilli have gained access to the brain. Once formed, the Rich foci may become activated rapidly or months to years later, resulting in release of M. tuberculosis into the subarachnoid space, triggering an inflammatory cascade 9 . The resulting inflammatory changes may explain some of the clinical features associated with TBM. First, peri-vascular inflammation, particularly of the middle cerebral artery, results in decreased perfusion and cerebral infarct. Second, extension of exudative material to the basal cisterns and the mid-brain leads to disruption of CSF flow, hydrocephalus and raised intracranial pressure. Third, exudates encase cranial nerves, resulting in cranial nerve palsies. Finally, expanding parenchymal tubercles may form tuberculomas and, less frequently, brain abscesses. In contrast (or in addition) to the Rich hypothesis, other researchers have suggested that bacilli reach the CSF during miliary dissemination. Donald and colleagues have proposed a pathogenic mechanism, based on more recent clinical, post mortem and epidemiological data, that articulates the central role of the Rich focus but with an additional component that includes miliary TB 5 .

Natural history
The risk and severity of TBM are altered by the status of the host immune response and pathogen virulence, consistent with the damage response framework 10 . Multiple host factors such as age, HIV co-infection, Bacillus Calmette-Guérin (BCG) immunisation, malnutrition, and helminth co-infection may lead to either a deficient or exaggerated immune response to mycobacterial infection.

Age
Several studies conducted in the pre-chemotherapy era identified children who had been exposed to TB and followed them for tuberculin skin test conversion, chest radiograph changes, clinical progression, and mycobacterial culture positivity. Marais and colleagues reviewed these studies and proposed a timetable for TB, in the absence of drug therapy 11 . Following infection with M. tuberculosis, young children are at high risk of progression to TB disease including disseminated forms such as TBM. Both the risk of disease progression and risk of dissemination fall rapidly through childhood, reaching a nadir in the primary school age. In more recent cohorts of children with TBM, the median age is between two and four years 12 . Marais and colleagues also found that children who develop miliary TB or TBM generally do so from one to four months after exposure.

HIV
HIV co-infection is the most significant risk factor for TBM in adults and is believed to attenuate the host immune response in the CSF. The impact of HIV on the clinical presentation of TBM is debated, with some studies suggesting a higher rate of extra-meningeal disease and of TB drug resistance [13][14][15][16] . Other studies report that the clinical course of TBM is unaltered by HIV infection 17,18 .

BCG, helminth infections and nutrition
Neonatal BCG immunization is effective in preventing childhood TB, particularly TBM and miliary TB 19-21 . Efficacy varies from as high as 80% near the poles, to <20% near the equator, possibly due to masking or blocking from environmental mycobacteria or modulation due to helminth infections 20,22,23 . Helminth infections activate a T helper type 2 and/or regulatory T cell response and may down-regulate T helper type 1 and T helper type 17 responses to mycobacterial antigens, potentially decreasing BCG vaccine immunogenicity. Whether this is clinically important is unclear 24-31 . Several epidemiological, clinical and laboratory studies have demonstrated that malnutrition, a form of acquired immunodeficiency, increases susceptibility to TB. In particular, vitamin D and vitamin D receptor polymorphisms are critical for function of macrophages in the context of TBM 32-34 . Vitamin D deficiency was significantly more common in TBM compared to controls and pulmonary TB in an Indian study 35 .

Strain type
There are seven main M. tuberculosis lineages, each associated with varying degrees of proinflammatory host response. The modern lineage x (Beijing) is associated with disseminated extra-pulmonary disease, drug resistance, and possibly poor treatment outcomes 36-39 . One plausible mechanism of Beijing strain survival is a dampened IFN-γ host inflammatory response, promoting high bacterial load.

Epidemiology
Our understanding of the global burden of TBM is poor. Because of inadequate diagnostic test performance and availability, many cases of TBM remain undiagnosed. Autopsy studies in adults with HIV have found high proportions of TB co-infection, commonly with extrapulmonary disease, including meningitis, which is frequently undiagnosed and untreated prior to death 40 . Even those cases diagnosed may not be appropriately reported as individuals with TBM are usually diagnosed in hospitals, which in some contexts, are less likely to be reporting units for TB 41 . Adjuvant immunomodulatory treatment is used in TBM to ameliorate the potentially damaging host response. A 2016 Cochrane systematic review and meta-analysis found that adjunctive corticosteroids reduce mortality in the short term but had no effect on long-term neurological disability in HIV-uninfected patients with TBM; however, the benefit of corticosteroids is unclear in HIV co-infected individuals 102 . Corticosteroids may have a differential effect dependent on leukotriene A4 hydrolase genotype, being beneficial in those with a hyper-inflammatory genotype but detrimental in others. Use of adjunctive aspirin has been shown to be beneficial in adults with TBM 103,104 , with one recent study of high dose aspirin demonstrating prevention of new cerebral infarction in adults 105 . This effect, however, has not been demonstrated in children and requires larger randomised trials in adults 106 . Currently registered clinical trials for TBM are shown in Table 3.
TB mass lesions, usually in the setting of immune reconstitution inflammatory syndrome (IRIS) and optochisamic arachnoiditis, are difficult to treat and often occur near vital structures in the brainstem and spinal cord. Thalidomide, a potent TNF-α inhibitor, has demonstrated clinical benefit in both TB mass lesions and optochiasmic arachnoiditis 107,108 . Reports have also demonstrated a clinical role for other agents that cause TNF-α inhibition, including infliximab and IFN-γ 109-111 . Prophylactic pyridoxine (5-10 mg per day in children and 10 mg per day in adults) is recommended to prevent isoniazid related peripheral neuropathy 93,94 . Neuropathy is more severe during pregnancy, HIV co-infection, alcohol dependency, malnutrition, diabetes, chronic liver disease and renal failure. Treatment of hydrocephalus depends on the level of CSF obstruction. Communicating hydrocephalus, a common complication of TBM, can be successfully treated with one month of acetazolamide 50 mg/kg per day and frusemide 1 mg/kg per day 112,113 .

Health systems
The care cascade for TBM has not been specifically investigated. For TB more generally, large gaps exist at every stage from incidence to diagnosis, from diagnosis to treatment and from treatment initiation to favourable outcome 114-116 . However, there are specific challenges for health systems in their management of TBM, including rehabilitation following treatment conclusion, that suggest gaps could be even larger than for TB as a whole.
Symptoms of TBM are non-specific and community sensitisation is poor. Even where sensitisation to TB has occurred, this is usually to the symptoms and signs of pulmonary TB, such as cough, weight loss, night sweats and haemoptysis 117 . If individuals with TBM do present for evaluation, it is commonly to primary care facilities, at which staff training and experience can be limited and appropriate investigations, such as lumbar puncture or cerebral imaging, are often unavailable. Multiple presentations to healthcare workers are common and are associated with delayed diagnosis and treatment initiation 118,119 . Referral from primary care to hospital can lead to further delays and frequently incur costs to patients and families. These delays can lead to clinical deterioration 120 .
Even when an individual presents to a hospital for evaluation, the availability of Xpert is variable globally 61 . Although AFB smear is widely available, it has limited sensitivity and culture, if available, returns results in a timescale that is not clinically useful. Although drug therapy is usually provided freely by national TB programmes and is available at peripheral care settings, the specialist services required to appropriately manage TBM, such as paediatrics, neurology, neurosurgery and rehabilitation, are usually only available in specialist centres in larger urban areas and frequently incur costs for patients.
Surveillance data for TBM is limited, with cases rarely reported if patients die before starting treatment. Many patients with TBM are cared for in hospitals for most of their illness and given that many hospitals are not TB-reporting units, these cases are not reflected in regional, national or global reports. Disaggregation of data occurs to the level of pulmonary TB and extrapulmonary TB, without specifying the site of disease. Some electronic registers allow the recording of International Classification of Diseases 10 th revision (ICD-10) codes, but these are often poorly completed.

Conclusions
TBM is the most severe form of TB and is associated with high mortality and morbidity. Our understanding of the global burden is limited but it is likely that most cases are not diagnosed and appropriately treated. At least 100,000 cases are likely to occur each year. The current diagnostic tools are largely inadequate, and treatment is usually started only once substantial neurological damage has occurred. Treatment, both anti-mycobacterial and anti-inflammatory, require further investigation and optimisation, and improvements at every stage along the care cascade are urgently needed.

Data availability
No data are associated with this article.
report of its use in a one-year-old infant. Yale J Biol Med. 1946;18: 221-226. actually available and the barriers to implementation of such testing at the healthcare system level in those where the testing does not exist.
The treatment discussion is nicely delineated, with a historical and current perspective discussion. Missing from the discussion is commentary on treatment in the pediatric population and treatment initiation of antiretroviral medications in patients with TBM, both challenges to those at the bedside treating such patients. It would be worthwhile to expand the treatment section to include supportive care given the importance of neurointensive care efforts including ICP monitoring, treatment of hyponatremia, when to consider shunting etc.
The healthcare systems section of the paper is important and novel and would benefit perhaps from a visual of the current system at the primary, secondary and tertiary care health systems levels and major knowledge gaps/policy gaps in clinical practice.
Expansion of Table one is also important in terms of future directions beyond implications of challenges in TBM--how do we study and fill this knowledge gap is essential in TBM.
Overall, while the article does not provide new data on TBM, the review provides an important summary on our current state of knowledge, with an emphasis on knowledge gaps. I applaud the authors for their work on this important disease, and look forward to further fruitful work from the consortium.

Is the review written in accessible language? Yes
Are the conclusions drawn appropriate in the context of the current research literature? Yes No competing interests were disclosed.

Competing Interests:
Reviewer Expertise: Clinical research in neuroinfectious diseases, Studies on access to neurological care in resource-limited settings I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.