Objective. To study the effect of high-dose prednisone on intracranial pressure (ICP), cranial computed tomographic (CT) findings, and clinical outcome in young children with moderate to severe tuberculous meningitis (TBM).
Study Design. Prospective, controlled, randomized study.
Methods. Continuous lumbar cerebrospinal fluid pressure monitoring and contrasted CT scanning were performed in 141 consecutive children with TBM at admission. All children were then randomly allocated to a nonsteroid group (71 children) or a steroid group (70 children) who received prednisone (first 16 children, 2 mg/kg per day; next 54 children, 4 mg/kg per day) for the first month of treatment. ICP monitoring and CT scanning were repeated regularly, and clinical outcome was assessed after 6 months of antituberculosis treatment.
Results. No statistically significant difference in ICP or the degree of hydrocephalus (as demonstrated by CT scan) was found between the steroid and nonsteroid groups after the first month of treatment. Basal ganglia infarcts developed in 16% of children in the steroid group and 24% in the nonsteroid group during the first month of treatment. Neither this incidence nor the eventual size of infarcts present at admission differed significantly between the two treatment groups. Single or multiple tuberculomas were seen on the first CT scans of 7 children (5%), whereas tuberculomas developed in 11 children (8%) at treatment. Both the response of the tuberculomas to treatment and the incidence of new tuberculomas were significantly improved by steroid therapy. Basal enhancement was also significantly less in the steroid group after 1 month of treatment. Steroids lowered mortality in stage III TBM significantly. Similarly, more surviving children in the steroid group had IQs of greater than 75 than did the those in the nonsteroid group. No significant difference was found in the incidence of motor deficit, blindness, or deafness.
Conclusions. Corticosteroids significantly improved the survival rate and intellectual outcome of children with TBM. Enhanced resolution of the basal exudate and tuberculomas by steroids was shown by serial CT scanning. Corticosteroids did not affect ICP or the incidence of basal ganglia infarction significantly.
- TBM =
- tuberculous meningitis •
- ICP =
- intracranial pressure •
- CT =
- computed tomographic •
- CSF =
- cerebrospinal fluid •
- VP =
- ventriculoperitoneal •
- PVL =
- periventricular lucency
More than 30 years after corticosteroids were first introduced as adjuvant therapy in the treatment of tuberculous meningitis (TBM), their value in this condition remains controversial. Some authors concluded that steroids had no place in the treatment of TBM, because the drugs did not benefit clinical outcome at all.1,2Others, however, described improved survival but an increased number of disabled survivors,3 whereas more recent studies found that steroids improved both survival rate and neurologic outcome in patients with TBM.4,5
The outcome of TBM is known to be affected by age,6 stage of the disease at admission,6,7 and whether raised intracranial pressure (ICP) caused by obstructive hydrocephalus is actively treated.8 Improper randomization of patients on account of these variables may thus contribute to the conflicting results of previous studies. Moreover, all the existing studies on the role of steroid therapy in patients with TBM were mainly clinical, with cranial computed tomographic (CT) scanning and ICP monitoring not being routine investigations in any of these studies. Some authors have attributed the beneficial effect of steroids in TBM to the role it plays in normalizing ICP.9 Others proposed that steroids may improve clinical outcome in TBM by enhancing the resolution of the associated vasculitis.10 A recent editorial emphasized the need for randomized clinical trials in determining the value of steroids in patients with TBM, because few such studies have been conducted.11 The exact mechanism by which steroids might improve survival in patients with TBM, however, remains a matter of speculation.
The purpose of this study was to evaluate the effect of high-dose prednisone in young children with moderate to severe TBM in a controlled, randomized trial. The efficacy of steroids was assessed with regard to: (1) ICP and ventricular size changes by means of repeated ICP recording and CT scanning at fixed intervals during the course of therapy; (2) brain parenchymal changes (eg, infarcts and tuberculomata), as demonstrated by CT scanning; and (3) clinical outcome at the time of completion of antituberculosis therapy.
The study comprised 141 consecutive children with TBM. A clinical diagnosis of TBM was made in all patients on account of the history and typical cerebrospinal fluid (CSF) changes, together with two or more of the following: a strongly positive (>15-mm) Mantoux test (65% of patients); and chest radiograph findings suggesting tuberculosis, ie, a miliary picture or hilar lymph adenopathy, often accompanied by a segmental lesion (60% of patients) and acute hydrocephalus with basal enhancement on CT scanning (83% of patients). In 56 children (40%) Mycobacterium tuberculosis was isolated from gastric aspirate (33 patients) and/or CSF (23 patients). The severity of disease was classified according to the British Medical Research Council classification:12 73 children had stage II and 68 had stage III TBM. Antituberculosis treatment consisted of daily isoniazid (20 mg/kg), rifampicin (20 mg/kg), ethionamide (20 mg/kg), and pyrazinamide (40 mg/kg). All drugs were given as single doses before breakfast for 6 months. Drug compliance could be carefully monitored, because the patients were hospitalized for the 6 months of treatment.
Clinical outcome in surviving children was assessed after 6 months of antituberculosis therapy. Intelligence was measured in 119 children by a clinical psychologist who used mainly the Bayley Scales of Infant Development in children younger than 3 years and the Griffiths Mental Development Scales in children 3 years and older at the time of assessment. Hearing was tested in 116 survivors by means of free-field audiometry or brainstem audiometry when indicated. An ophthalmologist tested vision in 119 survivors, and a physical therapist evaluated motor function in 126 children before discharge. All these individuals were blinded to the treatment status of the patients at admission.
Treatment of Raised ICP
At admission, lumbar CSF pressure was monitored continuously for 1 hour in all patients according to a previously described method. Any of the following, when present, were indicative of raised ICP: a mean baseline CSF pressure of greater than 15 mm Hg, pulse pressure of greater than 3 mm Hg, and B or plateau waves.
Communicating tuberculous hydrocephalus caused by basal cistern obstruction and noncommunicating hydrocephalus secondary to obliteration of the fourth ventricle foramina cannot be differentiated by CT scanning alone, because panventricular dilatation and effacement of the subarachnoid space occur in both types of CSF block. Therefore, we injected 10 mL of air into the lumbar CSF space at the end of the first pressure recording. Demonstration of air in the ventricles by skull radiograph in 116 children (82%) following the above procedure was interpreted as indicating the presence of communicating hydrocephalus. Noncommunicating hydrocephalus was diagnosed in 20 children (14%) whose skull radiographs showed air at the level of the basal cisterns but not in the ventricles. In 5 children in whom air encephalography was not obtained (3 children) or failed (2 children), noncommunicating hydrocephalus was diagnosed clinically on account of sudden deterioration of the level of consciousness or the development of signs of brainstem dysfunction.
All children with communicating hydrocephalus were treated with daily acetazolamide (100 mg/kg) and furosemide (1 mg/kg), which was administered orally at 6- to 8-hour intervals for 1 month. This drug combination has previously been shown by us to normalize ICP in 78% of children with communicating tuberculous hydrocephalus.13The effect of this treatment on ICP was assessed by weekly continuous 1-hour lumbar CSF pressure recordings during the first month of treatment. The children with noncommunicating hydrocephalus were referred for immediate ventriculoperitoneal (VP) shunting surgery.
After staging the degree of TBM (stages II and III) and determining the level of CSF block (communicating and noncommunicating), patients whose parents gave informed written consent were randomly allocated to a steroid or nonsteroid treatment group. Seventy patients received steroids, of whom 37 had stage II and 33 had stage III TBM. Thirty-six of the 71 children in the nonsteroid group had stage II and 35 had stage III disease. The mean age of patients in the steroid group was 38.3 (SD, 28.3) months and in the nonsteroid group was 28.8 (SD 21.9) months.
The first 16 patients in the steroid group received prednisone (2 mg/kg per day), and the remaining 54 patients received 4 mg/kg per day as a single morning dose for the first month of treatment. The daily prednisone dose was doubled when we became aware of a study that showed that rifampicin decreased the bioavailability of prednisolone by 66% and increased plasma clearance of the drug by 45%.14 No serious side effects of steroid therapy were documented, and all patients in the steroid group who survived the first month of treatment were able to complete the course of steroid therapy.
Patients had CT scans at admission, after 1 month of treatment, and again when followed up at 6 months. The majority of these scans were done with contrast enhancement. The following radiologic features were documented by individuals who were unaware of the treatment status of the patients when the scans were reviewed. The degree of hydrocephalus was expressed as a VP ratio, whereV = the ventricular diameter at the midportion of the body of the lateral ventricles and P = biparietal diameter measured from inner table to inner table. The extent of periventricular lucency (PVL) was graded as: grade I, PVL just visible at the angle of the anterior horns of the lateral ventricles; grade II, PVL extending well into the white matter at the anterior horns of the lateral ventricles but not reaching the cortex; and grade III, PVL extending from the anterior horns of the lateral ventricles to the surface of the brain. Basal enhancement was classified as grade I when slight enhancement was just discernible, grade II when moderate enhancement was clearly visible but not filling the subarachnoid spaces, and grade III whenever enhancement was so severe that the basal cisterns appeared obliterated. All hypodense, isodense, and slightly hyperdense rounded lesions with or without ring enhancement were considered tuberculomas. The site, size, and evolution of all infarcts were documented.
Criteria for Successful Treatment of Raised ICP
ICP was considered effectively normalized in the medical treatment group (communicating hydrocephalus, treated with acetazolamide and furosemide) when the following criteria were met within 4 weeks after treatment had begun: (1) normal lumbar CSF pressure, occasional B waves permitted; and (2) ventricular size the same or less and periventricular edema absent or markedly reduced (mild).
Whenever ICP did not normalize, and follow-up CT scanning after 1 month of treatment showed progressive hydrocephalus (ventricles larger and periventricular edema more than at admission), patients were regarded as having medical treatment failures and referred for VP shunting unless neuroimaging showed signs of severe, irreversible brain damage.
Categorical measurements were analyzed by means of the χ2 test and Fisher's exact test in the case of 2 × 2 contingency tables, with an expected cell frequency of less than 5. When cell frequencies were small, categories in certain variables had to be combined to get large enough frequencies for proper categorical analyses. The nonparametric tests for testing differences in means were the Wilcoxon two-sample test when comparing two groups and the Kruskal-Wallis test when three or more groups were compared. The study was approved by the Ethical Committee of the Faculty of Medicine of the University of Stellenbosch.
Normalization of ICP
The effect of steroids on ICP could be evaluated in only the 116 children with communicating hydrocephalus, because children with noncommunicating hydrocephalus underwent VP shunting shortly after admission.
Lumbar CSF Pressure
No significant difference was found between the baseline pressure and pulse pressure of the steroid and nonsteroid groups at admission and during and at the end of the first month of treatment (Table1).
Ventricular Size and PVL
The mean ventricular size (expressed as the VP ratio) of children in the steroid and nonsteroid treatment groups did not differ significantly after the first month of treatment or at 6 months, when antituberculosis treatment was completed (Table 2). Similarly, no significant difference was found in the number of children in each group who had compensated hydrocephalus, as indicated by absence of only mild PVL, after the first month of treatment (P = .60).
Successful Medical Treatment of Raised ICP
The number of patients in the two treatment groups who complied with the ICP and CT criteria for compensated hydrocephalus decided on at the beginning of the study are shown in Table 3. At the cutoff point of 1 month, 43 (80%) of the 54 children in the steroid group had compensated hydrocephalus compared with 38 (69%) of 55 children who did not receive steroids. This difference was statistically insignificant.
Twenty-five of the 28 children in whom progressive hydrocephalus developed were referred for VP shunting. Three of these children died. The 3 children from whom surgical treatment of hydrocephalus was withheld because of clinical and/or CT evidence of irreversible brain damage also died later.
Cranial CT Changes
Basal Ganglia Infarcts
Unilateral infarction of the basal ganglia was demonstrated in 15 (22%) of 68 patients in the steroid and 12 (17%) of 70 patients in the nonsteroid group at admission. The number of patients with bilateral basal ganglia infarcts at admission were 1 (1.5%) and 5 (7%) in the steroid and nonsteroid groups, respectively. No significant difference in infarct size, as judged by the degree of ex vacuo dilation of the anterior horn of the ipsilateral ventricle or extent of the parenchymal lesion, was found between the steroid and nonsteroid groups after the first month of treatment. New basal ganglia infarcts were demonstrated in 11 patients (16%) of the steroid group and 17 patients (24%) of the nonsteroid group by means of the routine follow-up scan 1 month after admission. This difference in the incidence of new infarcts between the two treatment groups, however, was not statistically significant (P = .14).
The final CT scans after 6 months of antituberculous treatment showed no difference in the infarct size of the two treatment groups.
Seven children (5%) had single or multiple tuberculomas on the initial CT scans. These tuberculomas were located in the brain parenchyma (four children) or adjacent to areas of prominent basal meningovascular enhancement (three children). Six of the seven children survived the first month of treatment. In the four patients who received steroid therapy, the follow-up CT scan after 1 month showed no evidence of tuberculomas in two children and a decrease in size of the tuberculomas with disappearance of the surrounding edema in the other two. In contrast, the CT appearance of the tuberculomas in the two children in the nonsteroid group remained virtually unchanged with regard to size and degree of surrounding edema.
Tuberculomas developed in 11 (8%) of the 141 children in the study during the first month of treatment, 9 of whom did not receive steroid therapy. All these tuberculomas had an extraparenchymal origin, 2 from the ependyma and the rest from the meningovascular enhancement at the circle of Willis. Of the 9 children in the nonsteroid group, 2 died before antituberculosis treatment was completed. In the 7 survivors the final CT scans after 6 months of antituberculosis therapy showed disappearance of the tuberculomas in 4 children and a decrease in the size of the tuberculomas in the other 3. New tuberculomas developed in only 2 children in the steroid group during the first month of treatment. One of these lesions was no longer seen on the final CT scan, whereas the other enlarged. This multiloculated tuberculoma, adjacent to the right middle cerebral artery at the base of the brain, was asymptomatic and spontaneously resolved over the next 2 years. The final CT scans of 4 patients (2 each in the steroid and nonsteroid groups), showed small basal tuberculomas, which were not seen before. The only two tuberculomas in this study that were symptomatic, however, arose from the ependymal lining of the lateral ventricle and presented with symptoms of raised ICP caused by unilateral foramen of Munro obstruction. One of the patients with these lesions died suddenly while being prepared for VP shunting, and the other had an uneventful recovery. Both the course of tuberculomas that were present at admission (P = .05) and the incidence of tuberculomas that appeared during the course of treatment (P = .03) were significantly different between the steroid and nonsteroid groups.
Twelve (20%) of the 60 children in the steroid group and 26 (45%) of 58 children who did not receive steroids had moderate to severe degrees of basal enhancement on the follow-up contrasted scans, which were done after the first month of antituberculosis treatment. This difference between the two groups was strongly significant (P = .004).
Although more children in the steroid group (32 of 62 children) than in the nonsteroid group (24 of 57 children) had enlarged subarachnoid spaces on the CT scans after 1 month, the difference was not statistically significant (P = .299).
Seventeen (12%) of the 141 children, 4 in the steroid group and 13 in the nonsteroid group, died before completing 6 months of antituberculosis therapy (Table 4). This difference in mortality rate between stage III TBM children in the two groups is statistically significant (P = .015). Only 1 death in the steroid group occurred during the first 2 weeks of treatment, compared with 7 in the nonsteroid group. All these children had stage III TBM and signs of decerebration on admission. Severe lung disease, however, was present in only 1 child, from the nonsteroid group, at the time of death. The mortality rate in the patients who received high-dose (4 mg/kg) and low-dose (2 mg/kg) prednisone did not differ significantly.
The clinical outcome of surviving patients at the time of completion of 6 months of antituberculosis treatment is shown in Tables4 and 5. Significantly more children in the steroid group than in the nonsteroid group had IQs of greater than (P = .038), but no significant difference was found between the two treatment groups with regard to motor deficit, blindness, or deafness. Similarly, the clinical outcome of children receiving high- and low-dose prednisone did not differ significantly.
Brain damage in TBM results from the effects of the granulomatous basal exudate, which causes raised ICP attributable to obstructive hydrocephalus and basal ganglia and brainstem infarction secondary to periarteritis of the blood vessels supplying these structures.15 Any drug with the potential ability of enhancing the resolution of this exudate could thus be potentially beneficial in normalizing ICP and decreasing the incidence of cerebral infarction in TBM.
Corticosteroids remain among the cornerstones in the treatment of the systemic vasculitis syndromes, the suggested mechanism of action being the inhibitory effects of this drug on monocytes, macrophages, and T and B lymphocytes.16 A decrease in CSF cytokine levels (interleukin 1β and tumor necrosis factor) was recently demonstrated in children with bacterial meningitis treated with dexamethasone.17 This finding offers a possible explanation for the previously described beneficial effect of dexamethasone on the incidence of severe and bilateral sensorineural hearing loss in children with Haemophilus influenzae meningitis. No significant difference could, however, be demonstrated between the CSF cytokine levels of a smaller group of steroid- and nonsteroid-treated patients with TBM in a recent study reported from our hospital.18
Raised ICP and Obstructive Hydrocephalus
Feldman et al19 experimentally showed that intracisternal injected, heat-killed tubercle bacilli caused no leptomeningeal inflammatory reaction in rabbits sensitized to tuberculoprotein when these animals were pretreated with intrathecal hydrocortisone acetate. Apart from a study by O'Toole et al,9 who reported that dexamethasone decreased the opening lumbar CSF pressure of children with TBM significantly during the first week of therapy, no other studies on the role of steroids on ICP in childhood TBM have been reported. We have previously shown that antituberculosis treatment combined with acetazolamide and furosemide successfully normalized ICP in 78% of children with communicating hydrocephalus caused by TBM.13 The results of the present study show that high-dose prednisone therapy did not further improve the number of children whose hydrocephalus became compensated with the above-mentioned regimen. A possible anti-inflammatory effect of steroids on the basal exudate was shown by the significant reduction in the degree of meningovascular enhancement seen on CT scans after the first month of treatment. However, these changes do not seem to be functionally significant, because no difference could be demonstrated between the ICP and ventricular size of the steroid- and nonsteroid-treated patients in our study.
Parenchymal CT Lesions
The predominantly basal exudate associated with TBM results in periarteritis of the vessels of the circle of Willis and their branches in the subarachnoid space.15 The ensuing panarteritis often results in occlusion of especially the medial branches of the middle cerebral arteries with infarction of the basal ganglia.20In a recent prospective study, we found that 21% of 198 children with stage II and III TBM had CT evidence of unilateral basal ganglia infarction, and 10% had evidence of bilateral basal ganglia infarction. New infarcts developed in an additional 22% of patients during the first month of treatment.21 Steroids are known to decrease morbidity and mortality in conditions associated with systemic necrotizing vasculitis, such as giant cell arteritis and poliarteritis nodosa.16 Mackay10 reported a dramatic clinical improvement of localizing signs caused by presumed periarteritis in a few adult patients with TBM shortly after the introduction of steroid therapy. However, the effect of steroids on TBM-associated vasculitis has never been objectively evaluated in a randomized study by means of serial CT scanning and repeated clinical assessment.
In the present study, steroids did not seem to have affected the vasculitis associated with TBM significantly, as evaluated by CT scanning, because no difference in either the eventual infarct size or the incidence of new infarcts could be shown between the steroid- and nonsteroid-treated patients. The almost similar incidence of residual hemiparesis in the two treatment groups at follow-up supports these CT findings.
Intracranial tuberculomas are known to appear or to paradoxically increase in size in patients while being treated for TBM.22These lesions are usually discovered accidentally when follow-up CT scanning is performed routinely or when new neurologic signs develop during the course of antituberculosis therapy.23 Because of a lack of prospectively conducted clinical studies, however, both the incidence of these lesions in TBM as well as the possible preventive role of steroids on their development are unknown.
The present study showed that although the incidence of these lesions during the first month of treatment was unexpectedly high (8%), they were almost always asymptomatic. The sudden deterioration and death in one of our patients with an ependymal tuberculoma, however, emphasizes the possible disastrous effects of these lesions when causing a proximal CSF block.
Late-onset tuberculomas have been noted to become symptomatic shortly after cessation of steroid therapy and to improve when steroids were subsequently recommenced.23 Our finding of a significantly reduced incidence of late-onset tuberculomas in children who received high-dose prednisone during the first month of treatment has not been reported before. The significantly enhanced resolution of tuberculomas already present at admission in our steroid-treated patients would further support the beneficial role of steroids on intracranial tuberculomas. This finding lends some scientific support to the generally accepted idea that intracranial tuberculomas resolve faster when steroids are added to antituberculosis treatment, a concept that has not before been confirmed in a controlled, randomized study.
Our findings that steroids reduced mortality in TBM are in agreement with the results of three recently reported studies on the role of steroids in this disease. A study from China24reported that steroids significantly reduced mortality in adults with stage II and III TBM, whereas Girgis et al5 found the mortality rate to be especially reduced in stage II disease. In a recent controlled, randomized trial of dexamethasone in TBM, Kumarvelu and Ahuja4 also found a trend toward better survival in patients who received steroids, but the small number of patients in their study prevented them from reaching conclusions of statistical significance. Our study in a larger patient cohort confirms that steroids do indeed improve survival in TBM significantly.
The very low mortality rates for both stage II and III TBM in our study is especially significant, because the mean age of the patients was younger than 36 months, and the outcome of the disease is known to be worse in young children.6 We think that the active monitoring and treatment of raised ICP probably was the most important factor in this regard. Steroids, however, further decreased mortality in our patients with stage III TBM from 17% to 4%.
Some of the older studies on TBM suggested that steroids improved the mortality rate and increased the number of disabled survivors.3 In their controlled randomized study, Kamarvelu and Ahuja4 found that dexamethasone not only decreased mortality in adults with TBM but was also associated with fewer permanent neurologic sequelae in the survivors. Our findings of a significant improvement in cognitive function of corticosteroid-treated TBM survivors correlates with the trend toward improved intellectual outcome in steroid-treated survivors in the above-mentioned study. However, steroids did not significantly decrease the incidence of permanent motor deficit (hemiparesis and quadriparesis) in our patients. This apparent inability of steroids to modify the periarteritis of TBM and the associated infarcts corresponds to our finding that steroids did not significantly reduce the number or size of basal ganglia infarcts as demonstrated by CT scanning.
Pathologically the improved clinical outcome in our steroid-treated patients with TBM cannot be explained by an effect on supratentorial structures, because no significant difference in ICP, ventricular size, or the incidence of basal ganglia infarction could be demonstrated between the two treatment groups. We have previously shown by magnetic resonance scanning that the often copious tuberculous exudate filling the basal cisterns can affect the underlying brainstem extensively.25 Patients in the present study who received steroids had less basal enhancement on CT scans after 1 month of treatment than the control patients. Steroids may thus improve the clinical outcome in TBM by way of a possible beneficial effect on the associated border zone brainstem encephalopathy.
We thank the medical superintendent of Tygerberg Hospital for permission to publish the results of our study. We are indebted to the South Africa Medical Research Council for financial support. We also thank Salomé Engelbrecht for assistance in preparing the manuscript.
- Received November 10, 1995.
- Accepted March 14, 1996.
Reprint requests to (J.F.S.) Department of Paediatrics and Child Health, Faculty of Medicine, University of Stellenbosch, PO Box 19063, Tygerberg 7505, Republic of South Africa.
- Ghosh S,
- Seshadri R,
- Jain KC
- ↵Chitale VR, Kasaliwal. Our experience of ventriculo-arterial shunt using Upadhyaya valve in cases of hydrocephalus associated with tuberculous meningitis. In: Rickham PP, Hecker WC, Prévot J, eds. Progress in Pediatric Surgery. Baltimore, MD: Urban and Schwarzenburg; 1982:224–231
- Humphries M
- McAllister WAC,
- Thompson PJ,
- Al-Habet SM
- ↵Dastur DK, Lalitha VS. The many facets of neuro-tuberculosis: an epitome of neuropathology. In: Zimmerman HM, ed. Progress in Neuropathology. New York, NY: Grune and Stratton; 1973:351–408
- Donald PR,
- Schoeman JF,
- Beyers N,
- Carlini SM,
- Olson KD,
- McCracken GH
- ↵Feldman S, Behar AJ, Weber D. Experimental tuberculous meningitis in rabbits: II. Effect of hydrocortisone on the hypersensitivity reaction of the meninges. Arch Neurol. 3:420–424
- Schoeman JF,
- Van Zyl LE,
- Laubscher JA,
- Donald PR
- Shaw PP,
- Wang SM,
- Tung SG,
- et al.
- Copyright © 1997 American Academy of Pediatrics