Management of tuberculous meningitis in children

Introduction

Tuberculous meningitis (TBM) remains the most devastating form of tuberculosis (TB). A systematic review of TBM in children reported an overall mortality risk of almost 20% [1], yet mortality has been found to be <5% in some large paediatric studies [2,3]. In several studies, severe neurological morbidity is reported in more than 50% of survivors, but this depends largely on the stage of TBM at presentation [1,4]. Because many cases go undiagnosed, little is known about the true incidence of TBM in children, and, even if diagnosed, cases may not be reported [5,6]. In Germany, a low TB-burden country, the prevalence of TBM among TB cases from 2002 to 2009 was estimated to be approximately 1% overall, but was 3.9% in children <5 years, 2.2% in children aged 5–9 years and 1.3% in children aged 10–14 years [7]. In a high TB-burden setting in the Western Cape, South Africa, 5.6% of children (<13 years) with bacteriologically confirmed TB diagnosed at hospital level between 2013 and 2017 had TBM [8]. The World Health Organization (WHO) estimated that in 2019 there were 1.19 million new cases of TB in children (0–<15 years); if 2% of these children had TBM, it would account for ~25,000 cases per year globally. This editorial briefly describes the pathogenesis of TBM in children as well as advances in diagnostics, recent developments in antimicrobial therapy and the role of therapeutic drug monitoring.

Pathogenesis

TB infection occurs through inhalation of aerosolised droplets (<5 μm in size) containing a few M. tuberculosis bacilli, some of which reach the terminal alveoli. Activation of neutrophils, dendritic cells and alveolar macrophages then occurs, which engulf the mycobacteria. Infected cells spread via the lymphatics and may reach the vasculature, causing haematogenous dissemination, with the potential to invade the central nervous system (CNS) [6]. Although the exact pathogenesis of TBM continues to be debated, Rich and McCordock showed that the meninges could not be directly infected by haematogenous spread [9]. They demonstrated that caseating subcortical or meningeal (Rich) foci form, which are activated rapidly, or, alternatively, months to years later, from which bacilli gain access to the subarachnoid space, triggering an inflammatory cascade. After the release of bacilli and granulomatous material into the subarachnoid space, a dense gelatinous exudate forms; it is most florid in the interpeduncular fossa and suprasellar region anteriorly, and it may extend throughout the prepontine cistern and surround the spinal cord. The exudative material in the basal cisterns and the mid-brain leads to disruption of cerebrospinal fluid (CSF) flow, hydrocephalus and raised intracranial pressure. It envelops arteries causing peri-vascular inflammation, particularly of the middle cerebral artery, resulting in decreased perfusion and cerebral infarction. Furthermore, exudates encase cranial nerves, resulting in cranial nerve palsies, and, finally, expanding parenchymal tubercles may form tuberculomas and, less frequently, pseudoabscesses in the brain [6,10].

Prevention of TBM

The primary focus of clinical and public health measures should be to prevent TBM in children. Neonatal BCG vaccination is 60–80% effective in preventing disseminated forms of TB (TBM and miliary TB) in young children after infection [11], and recent shortages of BCG have ssconfirmed this [12]. However, especially in high TB-burden settings, BCG is insufficient to protect all TB-infected children against TBM. In low TB-burden countries, as highlighted in the case reported from France by Huynh and colleagues in this issue of Paediatrics and International Child Health, BCG is often not routinely given [13]. Contact tracing with early identification and treatment of prevalent TB cases together with the provision of TB preventive therapy for infected/exposed children will further reduce the number of TBM cases; unfortunately, these opportunities are often missed [14,15].

Diagnostic approach

Clinical features

If disease develops, early diagnosis is critical. This is frequently challenging, however, because the early symptoms and signs are non-specific, and a high index of suspicion, a good history and clinical examination are the mainstays of a presumptive diagnosis. Prior to progression to TBM, some children may have symptoms of pulmonary TB, such as cough, fever or failure to gain weight. A diagnosis at this point may prevent progression to TBM. Following the discharge of M. tuberculosis bacilli into the CSF, however, decreased activity, further weight loss and vomiting are likely to occur. At this point, older children may present with headache and behavioural changes, yet, in all age groups, acute onset of meningitic symptoms and signs occurs- in only a minority of cases. There is often a history of exposure to an infectious case of pulmonary TB. A delay in diagnosis will be followed by neurological deficits (weakness), loss of consciousness or convulsions [10] and, unfortunately, it is often only when neurological symptoms appear that the diagnosis of TBM is considered and when irreversible neurological damage has already occurred. Other intrathoracic and extrathoracic features of TB disease may assist in diagnosing TBM.

Investigations

If meningitis is clinically considered, a lumbar puncture is usually the first investigation to be performed to obtain CSF for chemistry, white cell count and bacteriological examination. This should only be undertaken in the absence of neurological deficit, signs of raised intracranial pressure or loss of consciousness. Typical CSF results in TBM include a moderately raised white cell count with a lymphocyte predominance, raised protein and low glucose; however, up to 20% of CSF results are not typical [16]. Although bacteriological confirmation of M. tuberculosis as the cause of meningitis is ideal, culture and molecular tests are frequently negative and, even when positive, results are rarely available in a timescale that is useful to meaningfully impact clinical decision-making. In one study, the sensitivity of culture, Xpert MTB/RIF and GenoType MTBDRplus on CSF in children were 22%, 26% and 33%, respectively, which highlight that a negative CSF test does not rule out TBM [17]. If lumbar puncture is clinically contra-indicated, brain computerised tomography (CT) and/or magnetic resonance imaging (MRI) are often undertaken before a lumbar puncture is attempted. Features of intracranial TB are basal enhancement, hydrocephalus with or without periventricular oedema, infarction, tuberculomas and, rarely, pseudo-abscesses [18]. Communicating versus non-communicating obstructive hydrocephalus is difficult to differentiate through imaging, and some centres use an air-encephalogram to make this distinction [19]. Other supporting investigations are chest radiography for pulmonary TB, and biological samples (respiratory secretions and other extra-thoracic specimens) for bacteriology. Although genotypical tests return results rapidly, mycobacterial culture is required to enable full drug susceptibility testing (DST). A tuberculin skin test or interferon-gamma release assay may assist in confirming M. tuberculosis sensitisation, but these tests are often falsely negative in severe TB, such as TBM. Novel biomarkers are also emerging as potential diagnostic aids [20].

Management of TBM

Management of TBM is multifaceted and should include drugs to kill the organism [21,22], approaches to address raised intracranial pressure (e.g. hydrocephalus) [23], critical care (such as management of hyponatraemia) [24,25] and treatment for the vasculitis and immunological response [26]. Here, only antimicrobial therapy is discussed. It should also be remembered that clinical deterioration, such as raised intracranial pressure owing to new tuberculomas or new brain infarcts, may occur even while on effective anti-TB treatment [27], but, if bacteriological confirmation and DST were not obtained, drug-resistant TBM should also be considered [28].

Drug management

As outlined by Huynh and colleagues [13], pharmacokinetic data for anti-TB drugs are limited in children. Although substantial work has been undertaken recently to address this knowledge gap, few pharmacokinetic studies of anti-TB drugs in CSF have been performed. The limited data that exist suggest that exposure of first-line TB drugs is low in the CSF of children treated for TB [29], and modelling studies suggest that higher dosages are required [30]. Every opportunity to increase our knowledge of CSF penetration of first- and second-line anti-TB drugs should be used, and well planned therapeutic drug monitoring (TDM) of CSF specimens in children on anti-TB treatment may accumulate experience. Although much current practice of treating TB in children is still derived from adult studies, the pharmacokinetics of many anti-TB drugs in children differ substantially from that in adults. Current first-line, drug-susceptible TBM treatment dosages in children are summarised in Table 1. In cases of drug-resistant TBM, a minimum of two but preferably three drugs with moderate-to-good CSF penetration is required to manage the TBM in addition to other MDR-TB drugs in a regimen of four or five effective drugs (Table 1).

Table 1. Anti-tuberculosis drugs, recommended doses, cerebrospinal fluid penetration, common adverse effects and comments

First-line drugs	Once daily dose^a	CSF penetration	Important adverse effects	Comments
Isoniazid [29]	10–15 mg/kg	Good	Peripheral neuropathy, hepatotoxicity	In South Africa, 15–20 mg/kg/day for TBM
Rifampicin [29]	10–20 mg/kg	Poor	Hepatotoxicity	In South Africa, 20 mg/kg/day and in USA 30 mg/kg/day for TBM. Higher doses currently recommended and investigated [30]
Pyrazinamide [29]	30–40 mg/kg	Good	Hepatotoxicity, skin rashes
Ethambutol [17]	15–25 mg/kg	Poor	Optic neuritis (rare)	In South Africa, TBM regimen replaced with ethionamide as fourth drug^b
Second-line drugs by WHO groups for MDR-TB
Group A
Fluoroquinolones [31]:Moxifloxacin Levofloxacin	10–15 mg/kg 15–20 mg/kg	Moderate–good Moderate–good	For both fluoroquinolones, arthralgia/arthritis, hallucinations. QT prolongation more with moxifloxacin	Probably needs higher doses. Rifampicin reduces moxifloxacin concentration Probably needs higher doses (19–33 mg/kg) [30]. Dispersible levofloxacin tablets acquire higher concentrations than adult formulation owing to better bio-availabilty [45] Fluoroquinolones partially protein-bound [33].
Bedaquiline	>30 kg: 400 mg daily × 2 wks, then 200 mg M/W/F × 22 wks 15–30 kg: 200 mg daily × 2 wks, then 100 mg M/W/F × 22 wks	Almost none	QT prolongation, arthralgia, hepatotoxicity	Dosing for children <6 y or <15 kg currently being determined Does penetrate into rat brain tissue [35].
Linezolid	<15 kg: 15 mg/kg OD ≥15 kg: 10–12 mg/kg	Good	Myelosuppression, peripheral neuropathy, optic neuritis, lactic acidosis	Needs close monitoring for adverse effects
Group B
Cycloserine/terizidone	15–20 mg/kg	Good	Neuropsychiatric adverse effects
Clofazimine [46]	>20 kg: 100 mg/day 10–20 kg: 50 mg/day or 100 mg every 2nd day 5–<10 kg: 50 mg every 2nd day or 100 mg M/W/F	Poor	QT prolongation, skin discoloration, ichthyosis	Good penetration in mouse brain tissue
Group C
Amikacin	15–20 mg/kg	Poor	Ototoxicity (irreversible), nephrotoxicity	Slightly better during acute inflammation. Try to avoid because of hearing loss
Delamanid [36]	>34 kg: 100 mg BD 20–34 kg: 50 mg BD 10–<20 kg: 25 mg BD	Poor	QT prolongation, neuropsychiatric adverse effects	Good brain tissue concentrations in rabbits
Meropenem (plus clavulanic acid)	40 mg/kg 8-hourly IV (max 2 g 8-hourly) plus 8-hourly amoxiclav	Poor to moderate	GIT disturbance, headache, allergic response	Rescue treatment in MDR-TB cases with resistance to fluoroquinolones and more
Ethionamide/prothionamide	15–20 mg/kg	Good	GIT disturbance, hypothyroidism, hepatotoxicity	GIT adverse effects precludes higher doses
Para-aminosalicylic acid [47]	200–300 mg/kg OD or BD	Poor	GIT disturbance, hypothyroidism, hepatotoxicity	Higher single daily dose may provide higher maximum CSF concentration if child can tolerate dose

^aUnless otherwise indicated; ^bSouth African TBM regimen in children: 6 to 9 months of isoniazid, rifampicin, pyrazinamide plus ethionamide (all 4 drugs full duration)

CSF: cerebrospinal fluid; TBM: tuberculous meningitis; MDR-TB: multidrug-resistant tuberculosis; WHO: World Health Organisation; M/W/F: Monday/Wednesday/Friday dosing; GIT: gastro-intestinal.

Effective treatment of TB relies on sufficient concentrations of drugs above the minimal inhibitory concentration (MIC) at the site of disease. In TBM and/or tuberculomas, this means that adequate exposure is required in the CSF and brain tissue. This process is dynamic as, in the early phases of disease, disruption of the blood-CSF barrier during acute meningeal inflammation can lead to reasonable penetration. However, as the inflammatory response subsides with treatment, penetration declines. Some anti-TB drugs are known to penetrate the CSF well, irrespective of inflammation even after oral administration (Table 1). These include isoniazid, pyrazinamide, the fluoroquinolones, cycloserine, prothionamide/ethionamide and linezolid [21,31]. However, some drugs penetrate CSF very poorly, such as ethambutol and rifampicin. Rifampicin is a highly (89%) protein-bound drug and only unbound rifampicin is active in treatment [32]. It is probably mostly unbound rifampicin which penetrates the CSF, which could mean that lower concentrations of free (unbound) rifampicin could be effective as long as concentrations are above the MIC of the organism. Unbound drug is also the active component of other protein-bound drugs [33], including the new drugs bedaquiline and delamanid, both of which are also highly protein-bound. The concentration of bedaquiline was unmeasurably low in the CSF of one adult with TBM [34]; in a rat experiment, bedaquiline concentrations in the brain were much lower than plasma concentrations, but still achieved concentrations higher than the MIC for bedaquiline-susceptible strains [35]. Low concentrations of delamanid were detected in the CSF of both rabbits and humans, but, in the rabbits, the brain concentrations were much higher than in the plasma [36]. The role of these drugs in treating TBM and intracranial TB therefore still needs to be established, although there have been favourable outcomes in some patients on delamanid [36].

Intravenous vs oral treatment

Long-term intravenous (IV) treatment of children with TBM is rarely achievable, as most TBM cases occur in resource-limited settings. In some cases, short-term IV therapy is indicated because of an inability to receive medication orally. Achieving higher concentrations of rifampicin in CSF is not necessarily dependent on IV administration, as a recent study showed concentrations above the MIC with oral or IV administration in adults [37]. To achieve the same plasma concentrations and therefore probably CSF concentrations of rifampicin, children need much higher milligram per kilogram doses than adults owing to differences in drug absorption, distribution, metabolism and elimination [38]. In South Africa, a rifampicin dosage of 20 mg/kg/day is recommended for children with TBM. However, this may still be too low in some patients [30], and, when TDM can be undertaken, it is especially important for rifampicin as concentrations vary greatly between patients and even in the same patient. Other drugs with concentration- and dose-related adverse effects may also benefit from TDM, such as cycloserine/terizidone, linezolid and amikacin.

Therapeutic drug monitoring

TDM can be very useful for tailoring dosages for TB patients in order to achieve optimal therapeutic exposure while limiting toxicity [39]. This kind of personalised approach has been advocated for the management of drug-resistant TB [40], and a strong case can be made for TDM in the treatment of TBM. However, although Huynh and colleagues recommend TDM be undertaken and that it should be more universally available, especially in TBM cases [13], TDM in this conventional format is probably going to be difficult to achieve. Not only is there a need for high-level pharmacology laboratories with skilled staff, but also the collection of CSF specimens, the timing of CSF sampling after dosing, and the interpretation of the data are further challenges. However, if multi-site data on CSF concentrations of anti-TB drugs in children are collected and information gathered over time, it might become a valuable resource in guiding clinicians in the treatment of TBM. Innovative approaches to the implementation of TDM in low-resource settings require more focus.

Current ongoing studies to shorten and improve TBM treatment in children

TBM-KIDS is a phase I/II open-label clinical trial to evaluate the pharmacokinetics, safety and treatment outcome of high-dose rifampicin with or without levofloxacin versus standard care of children with TBM. Children are randomised for the first 8 weeks of treatment to isoniazid and pyrazinamide with (a) high-dose rifampicin and ethambutol, (b) high-dose rifampicin with levofloxacin substituted for ethambutol, or (c) standard care with a dosage of rifampicin recommended by WHO together with ethambutol. After 8 weeks, all arms are treated for 10 months with standard dosages of isoniazid and rifampicin. Recruitment was completed in 2020 and the trial results are expected in 2021 [41]. The SURE trial evaluates a short intensive antimicrobial regimen for TBM as well as the value of aspirin to prevent ischaemic stroke [42]. Children are randomised to a WHO-based standard of care regimen for 12 months or to a regimen of isoniazid at 20 mg/kg, rifampicin at 30 mg/kg, pyrazinamide at 40 mg/kg and levofloxacin at 20 mg/kg (all given daily for 6 months). Additionally, children are randomised in a factorial manner to receive either aspirin at 20 mg/kg daily or placebo for the first 60 days of treatment. Recruitment began in 2021 and the trial is expected to last for several years. When the results of these two trials are available, it is likely that our insight into TBM treatment in children will change dramatically.

Research priorities

A key research priority is to better understand the relationship between ingested dose, the resulting blood concentration and finally the subsequent CSF exposure for all first- and second-line TB drugs, in children of different ages, ethnicities and nutritional statuses, with and without HIV infection. Ultimately, however, it would be important to move beyond these pharmacokinetic parameters and evaluate pharmacodynamic measures so that treatment outcome is evaluated in children with different drug regimens using different drug dosages. Much can be learned from adult studies and trials to evaluate higher dosages of rifampicin, the inclusion of linezolid or the addition of fluoroquinolones, which have either recently been completed or are under way [21,43,44]. The role of TDM either for the treatment of individual children or to gather sparse-sample pharmacokinetic data needs to be systematically evaluated to determine whether a more personalised medicine approach results in a better outcome. Finally, efforts to develop cheap, easy and feasible TDM approaches which are suitable for implementation in low-resource settings need urgent escalation.

Conclusion

Although far more could be done to prevent TBM in children and to diagnose it early, improved treatment, including improved antimicrobial treatment, would also be of substantial benefit. Understanding of the antimicrobial treatment of TBM in children is slowly changing and new drugs, new dosages and new regimens are being evaluated. TDM could provide an important insight into individual clinical treatment and, more broadly, improve our understanding of drug pharmacokinetics. Although translating this kind of precision medicine approach to the majority of children with TBM will be challenging, low-cost solutions could dramatically change the treatment and outcome in children with TBM.

Acknowledgments

JAS is supported by a Clinician Scientist Fellowship jointly funded by the UK Medical Research Council (MRC) and the UK Department for International Development (DFID) under the MRC/DFID Concordat agreement (MR/R007942/1).

Additional information

Funding

JAS is supported by the MRC/DFID Concordat agreement [MR/R007942/1].