Role of Aspirin in Tuberculous Meningitis: A Systematic Review and Meta-analysis
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.266232
Source of Support: None, Conflict of Interest: None
Keywords: Aspirin, infarctions, stroke, tuberculous meningitis
The prognosis of tuberculous meningitis is poor as approximately half of the affected patients die or suffer from severe disabilities., Unfortunately, the major principles of tuberculous meningitis management are based on observational studies and clinical practices rather than clinical trials. Two important cornerstones in this management are the prompt initiation of anti-tuberculosis treatment and administration of adjunctive corticosteroids.
Strokes or infarctions are common complications of tuberculous meningitis that contribute to neurological deficit and worsen the prognosis. Infarctions occur in 15–57% of such patients.,,, The vessels most commonly involved are the proximal middle cerebral artery, medial lenticulostriate arteries, and thalamoperforating arteries. The infarctions in tuberculous meningitis are caused by tuberculous arteritis and entrapment of large-sized arteries in thick basal exudates, leading to arterial spasm and/or thrombosis. Cerebral infarctions are predictors of poor outcome of tuberculous meningitis  and are associated with poor neuro-developmental outcome in patients with childhood tuberculous meningitis. Unfortunately, strategies to prevent and manage infarctions in tuberculous meningitis are unclear.
Aspirin is an anti-inflammatory drug that inhibits the activity of cyclooxygenases. It is the cornerstone of antiplatelet therapy for the secondary prevention of ischemic stroke. However, its role in the prevention or management of stroke in tuberculous meningitis is unclear. A recent Vietnamese trial has suggested that the addition of aspirin to the anti-tuberculosis drug regimen improves the outcome of patients with tuberculous meningitis. However, the addition of aspirin showed no significant benefits in terms of reduction in mortality or prevention of new infarctions. A previous trial has demonstrated that aspirin reduces mortality in adult patients with tuberculous meningitis, whereas another has shown that aspirin does not reduce mortality and morbidity in childhood tuberculous meningitis.
Considering these uncertainties, we conducted this systematic review to assess the role of adjuvant aspirin in treating tuberculous meningitis. The research question formulated for this review was as follows: Does the addition of aspirin to the anti-tuberculosis treatment regimen improve the outcome (in terms of mortality/occurrence of new infarctions) of patients with tuberculous meningitis?
Criteria for considering studies for this review
Types of studies
Only randomized controlled trials were included in this review. No language restriction was applied.
Patients of any age diagnosed with tuberculous meningitis were considered. Diagnosis of tuberculous meningitis was based on clinical, cerebrospinal fluid, and neuroimaging features.
Types of intervention
Intervention group: Aspirin in any dose in addition to the standard anti-tuberculosis treatment.
Control group: Standard anti-tuberculosis treatment without aspirin.
Types of outcome
Death and occurrence of new infarctions confirmed by magnetic resonance imaging were considered as the primary outcome measures. In principle, magnetic resonance imaging was used to define an infarct so that focal neurological deficits occurring subsequent to the presence of tuberculoma, abscess, raised intracranial pressure, etc., could be appropriately labeled and silent infarcts could be detected with this modality. Persisting neurological disability, as reported by the authors at the end of follow-up, and adverse drug reactions, as defined by the authors, were considered as the secondary outcome measures.
Search method for identifying relevant studies
The published medical literature was systematically searched to identify all relevant trials. The electronic databases PubMed and Cochrane Central Register of Controlled Trials were searched without any limitations on year or language. Articles published till September 2018 were included in the review. The keywords used for searching the databases were “Tuberculous meningitis” OR “Tubercular meningitis” OR “Meningeal Tuberculosis” AND “Aspirin” OR “Acetylsalicylic acid” [Table 1]. SCOPUS was also searched using “Tuberculous meningitis” AND “Aspirin” as keywords. Clinical Trials Registry (www.ctri.nic.in, accessed on September 22, 2018) and Current Control Trials (www.controlled-trials.com; accessed on September 22, 2018) were searched using “tuberculosis” and “meningitis” as search terms. Subsequently, reference lists of all shortlisted trials were checked. Finally, the proceedings of annual conferences with available online data were searched. The searches were performed independently by two authors. Any disagreement was resolved by mutual discussion.
Study selection and data extraction
Two reviewers independently (IR, RKG) selected the studies based on the above-mentioned criteria, and any disagreement was resolved by mutual discussion. Selection of studies was performed in two steps. In the first step, the two reviewers independently screened titles and abstracts. In the second step, full articles of the selected studies were obtained. Any subsequent exclusion was recorded with a relevant reason.
Data extraction was independently performed by the two reviewers (IR, HSM) using a pre-formed data extraction sheet. The extracted data included bibliographic information (author, year), participants' characteristics, diagnostic criteria, stage of tuberculous meningitis, HIV status, doses and durations of aspirin administered, anti-tuberculosis regimens in the intervention and control arms, and outcome measures.
Risk of bias assessment
The Cochrane Collaboration's tool was used by the two reviewers (IR, HSM) for independently assessing the risk of bias in each included study. Conflicts were discussed with the third reviewer (RKG) and the post-discussion consensus was adopted. Each study was assessed for selection bias (random sequence generation and allocation concealment), performance bias (blinding of participants and personnel), detection bias (blinding of outcome assessment), attrition bias, and reporting bias. These biases were further classified into low risk, unclear risk, and high risk. The findings were plotted on a risk-of-bias graph using RevMan software (Review Manager (RevMan; computer program) Version 5.3; Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014).
Risk ratio (RR) was used as the measure of treatment effect. The RR and 95% confidence interval (CI) were calculated for all primary and secondary outcomes. The overall treatment effect was calculated using a random-effects model. This model was used because both clinical and methodological heterogeneities were expected between the trials. Statistical heterogeneity was assessed using visual inspection of the forest plots as well as χ2-test and I2 statistics. An I2 value of ≥50% and a P value of <0.10 in the χ2-test were considered to indicate statistical heterogeneity. The data were analyzed using the RevMan software version 5.3 (Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014).
For the purpose of this analysis, the aspirin doses were classified into two categories: a dose of ≤150 mg/day was defined as low, and that of >150 mg/day was defined as high. Publication bias was not assessed using the funnel plot because of the small number of trials.
Assessment of the quality of evidence
The quality of evidence for primary outcomes (death and new infarctions) was assessed using the GRADE approach.
A total of 72 studies were identified by searching the above-mentioned sources and 56 studies remained after eliminating the duplicate titles. The screening of titles and abstracts further excluded 45 articles. Seven more articles were excluded as they either were not randomized controlled trials or did not meet the inclusion criteria. Finally, four articles that met the inclusion criteria were included [Figure 1].,,,
All included studies were randomized controlled trials. Two trials were double-blind trials,, one open-label trial, and one with an undefined label. One study was from Vietnam, one from India, one from South Africa, and one from Pakistan [Table 2].
The four included trials comprised of 546 patients in total. The trial by Schoeman et al. also included an arm of nonrandomized patients (N = 29); this arm was excluded from the meta-analysis as the patients were already taking aspirin at the time of randomization. Of the total 546 randomized patients, 315 received aspirin and 231 received placebo. Of the 315 patients who received aspirin, 145 received a low dose (≤150 mg/day) and 170 received a high dose (>150 mg/day). Two trials included adult patients,, while the other two trials included only children < 15 years old., Further, 283 (51.83%) of the 546 patients were males. The classification of the severity of tuberculous meningitis was as follows: 87 patients had stage I infection, 210 had stage II, and 249 had stage III. In total, 177 patients were microbiologically confirmed as having tuberculous meningitis [Table 2].
The doses and durations of aspirin administered in the included trials are summarized in [Table 3]. Mai et al. used two doses of aspirin, namely 81 and 1000 mg/day, in their trial. Misra et al. used a dose of 150 mg/day. Schoeman et al. also used two doses: 75 mg/day and 100 mg/kg/day. Amin et al. used 60 mg/kg/day of aspirin. The doses and durations of anti-tuberculosis drugs and steroids used are also summarized in [Table 3].
Three trials reported the observations of the follow-up period. The primary outcomes were assessed at 60 days in the trial by Mai et al. The follow-up period was 3 and 6 months in the trials by Misra et al. and Schoeman et al., respectively., The follow-up period was not described in the study by Amin et al.
Death was reported in all included trials. New infarctions were reported in two trials;, one trial reported events in the form of hemiplegia, instead of infarctions per se. Disability assessment was reported in three trials, although all of them used different methods for assessing disability.,, Adverse events were reported in three trials.,, A few other outcome measures were also reported in the trials but were not considered in this review.
Assessment of the risk of bias
The methods of randomization were adequately reported in only three trials.,, Allocation concealment was adequately described in two trials., Adequate blinding of participants and personnel was performed in only two trials., Blinding of the outcome assessment was properly described in one trial, which was classified as having a low risk of detection bias. The rest of the three trials did not clearly describe the blinding of the outcome assessment, but this bias might not affect the assessment of mortality. Therefore, all these trials were classified as having an unclear risk of detection bias. One trial reported >10% loss to follow-up and was therefore classified as having a high risk of attrition bias. The trials all reported the outcomes mentioned in the Methods section and were therefore classified as having a low risk of reporting bias [Figure 2].
Effects of intervention
Death was reported in all four trials. The pooled analysis revealed 40 (12.70%) deaths among the patients receiving adjuvant aspirin compared with 53 (22.94%) deaths in the control arm. However, the difference in the number of deaths between the two arms was not statistically significant (RR = 0.66; 95% CI = 0.42–1.02). In addition, no significant statistical heterogeneity was observed (I2 = 17%) [Table 4] and [Figure 3].
Pediatric and adult trials were analyzed separately. The trials by Mai et al. and Misra et al. included adults and their pooled analysis showed no significant difference in the number of deaths between the aspirin and control arms (RR = 0.53; 95% CI = 0.28–1.01). The trials by Amin et al. and Schoeman et al. included children and their pooled analysis also showed no significant difference in the number of deaths between the aspirin and control arms (RR = 0.90; 95% CI = 0.34–2.39). No significant difference was observed in the number of deaths between the low-/high-dose aspirin arm and the control arm [Table 4] and [Figure 3].
Unequivocal infarctions, confirmed on magnetic resonance imaging, during follow-up were reported in two trials., The pooled analysis revealed that 15 (15%) patients in the aspirin arm developed new infarctions compared with 21 (32.31%) patients in the control arm. Thus, the addition of aspirin significantly reduced the occurrence of new infarctions (RR = 0.52; 95% CI = 0.29–0.92). No significant statistical heterogeneity was observed (I2 = 0%).
The comparison between the low-dose aspirin and control arms showed a significant reduction in the occurrence of new infarctions (RR = 0.49, 95% CI = 0.26–0.95). However, the comparison between high-dose aspirin and control arms showed no such significant reduction (RR = 0.59, 95% CI = 0.21–1.64) [Table 4] and [Figure 4].
All three trials which included a disability assessment used different scales to quantify disability. Therefore, it was not possible to pool the data and perform a meta-analysis. None of the trials reported a better outcome in terms of disability.
Adverse drug reactions were reported in only three trials.,, Only Mai et al. reported the number and type of adverse events for aspirin and control arms separately. Gastrointestinal bleeding events occurred in 20% patients in the aspirin arm and 13.2% patients in the control arm; this difference was not statistically significant. Intracranial bleeding was confirmed by magnetic resonance imaging in only one patient in the aspirin arm. No significant difference was noted in the occurrence of other side-effects, such as allergic reactions and gastrointestinal, hepatic, cardiac, or respiratory events. Forest plot showed comparison of adverse events [Figure 5]. Misra et al. reported 33 adverse events: vomiting in 28 patients, epigastric discomfort in 1 patient, rashes in 4 patients, and jaundice and altered liver function in 28 patients; the authors did not provide a break-up of these events according to the individual study arms but stated that no significant difference was observed in the occurrence of these events between the two arms. Schoeman et al. reported hematemesis in one patient and intracranial bleeding in one patient; the details of others adverse events were not mentioned.
Quality of evidence
The assessment of the quality of evidence as per the GRADE approach is summarized in [Table 5].
Despite adequate treatment using anti-tuberculosis drugs and adjunctive corticosteroids, morbidity and mortality in patients with tuberculous meningitis remain high. Host-directed therapies, including nonsteroidal anti-inflammatory drugs such as aspirin, can be beneficial in the management of tuberculosis.,
This systematic review assessed the benefits and disadvantages of adding aspirin to the first-line anti-tuberculosis drug regimen. Four relevant studies which met the inclusion criteria and comprised of 546 patients in total were included. These trials included both males and females as well as children and adults. The trials also included microbiologically confirmed and unconfirmed cases. The follow-up period ranged from 60 to 180 days.
The pooled analysis of the results of these trials revealed that the addition of aspirin to the anti-tuberculosis regimen does not significantly reduce mortality. Two trials included only children,, whereas the two trials included only adults., To make our comparison more homogenous, we analyzed the pediatric and adult trials separately as well; however, this stratified analysis did not demonstrate the superiority of aspirin over placebo in reducing mortality [Table 4].
In the primary prevention of cardiovascular diseases, aspirin reduces deaths, ischemic strokes, and myocardial infarction. However, we found no benefit of aspirin in reducing mortality among patients with tuberculous meningitis. Aspirin is commonly prescribed in two doses. The low doses of aspirin (75–150 mg) inhibit platelet aggregation and can prevent ischemic cerebrovascular diseases. In contrast, the high doses of aspirin (>150 mg) have anti-inflammatory effects through the inhibition of pro-inflammatory prostaglandins and thromboxane A2. The trials included in our study also used aspirin at different doses. Mai et al. and Schoeman et al. used both low and high doses., Misra et al. used only the low dose, whereas Amin et al. used only the high dose. A subgroup analysis based on the aspirin dose revealed no significant reduction in mortality using the low or high doses of aspirin compared with that using the placebo [Table 4]. It may be reiterated that the dose schedules used in the trials listed are arbitrary and there is no scientific explanation to the low-dose versus high-dose regimen. Additionally, there have been no transitional dosage schedules to judge the outcomes related with intermediate doses. Even the drug-and-dosage regimen of anti-tuberculosis medication and steroids, across the trials, is not uniform.
Notably, the addition of aspirin leads to a significant reduction in the risk of new infarctions among patients with tuberculous meningitis., The pooled analysis revealed that approximately 15% patients developed new infarctions in the aspirin arm compared with approximately 32% patients in the control arm. Thus, the addition of aspirin led to approximately 17% absolute risk reduction in the occurrence of new infarctions. This is equal to a number needed to treat approximately six patients, indicating that approximately six patients with tuberculous meningitis need to be treated with adjunctive aspirin to prevent the occurrence of one new infarction. Further, the subgroup analysis performed based on the aspirin dose revealed that the addition of a low dose of aspirin significantly reduces the risk of occurrence of new infarctions [Table 4]; this finding has important implications for reducing the risk of infarctions in tuberculous meningitis.
We did not pool disability as an outcome because different trials used different methods for assessing disability. Thus, the effect of aspirin on disability could not be determined. We speculate that the prevention of new infarctions affects disability. A recently conducted retrospective study found that adjunctive aspirin can improve the outcome of patients with tuberculous meningitis; however, this study was not included in the meta-analysis as it was not done in a randomized controlled fashion.
Aspirin use was not associated with a significant increase in adverse effects, such as allergic reactions and gastrointestinal, hepatic, cardiac, or respiratory events. No significant increase was noted in gastrointestinal or intracranial bleeding in the aspirin arm compared with that in the control arm. These estimates were available from a single trial only as the other trials did not provide a detailed breakup of adverse events. However, the safety of high-dose aspirin appears to be established to a considerable extent.
We used a random-effects model for pooling of data as we expected the effect size to differ between the trials due to differences in age distribution and severity of tuberculous meningitis as well as slight differences in the doses and durations of aspirin. Our main comparisons revealed no significant statistical heterogeneity. Furthermore, we could not perform a subgroup analysis based on the classification of tuberculous meningitis (definite versus the rest), severity of disease, HIV status, and drug resistance categories as sufficient data was not available from the included trials.
The GRADE approach was used to assess the quality of evidence for death and occurrence of new infarctions. We found low-quality evidence for the effect of adjuvant aspirin in reducing mortality, indicating that it is not superior to the anti-tuberculosis drug regimen alone in reducing mortality. Thus, further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. We found moderate-quality evidence for the effect of adjuvant aspirin in preventing the occurrence of new infarctions, indicating that its addition to the standard regimen helps in preventing their occurrence. Thus, further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
In conclusion, aspirin reduces the risk of new infarctions in patients with tuberculous meningitis but does not alter the risk of death. The evidence found in our review is of moderate-to-low quality. Thus, further research is warranted to formulate a better evidence-based recommendation.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5]
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5]