Clinical Profile, Yield of Cartridge-based Nucleic Acid Amplification Test (GeneXpert), and Outcome in Children with Tubercular Meningitis.

BACKGROUND: GeneXpert MTB/RIF is a test for early, rapid diagnosis of tubercular meningitis (TBM). AIM: The aim of this article was to study the clinical profile, radiological features, yield of GeneXpert, neurosurgical interventions, and outcome of TBM in children. SETTINGS AND
DESIGN: This was a retrospective and prospective observational study.
MATERIALS AND METHODS: Diagnosis was based on the uniform research definition criteria and was staged according to the British Medical Research Council. Mantoux test, analysis of cerebrospinal fluid (CSF), CSF GeneXpert, and radiological investigations were performed.
RESULTS: Of 36 patients, 50% were aged 1-5 years. Fever (100%), headache (82%), altered sensorium (80%), and vomiting (66%) were common features. Twelve (33%) had contact with active case of tuberculosis; 32 received Bacille Calmette Guarin vaccination. Neurological features included severe deterioration in sensorium (Glasgow Coma Scale < 8) (38%), mild and moderate deficit in sensorium (31%), hemiparesis (41%), and involvement of sixth (25%) and seventh (22%) cranial nerves. Cerebral vision impairment (25%), papilledema (25%), and dystonia (22%) were other findings. CSF GeneXpert was positive in 37% (12/33) patients. Hydrocephalus and basal exudates (75%) were noted on neuro-imaging. Surgical intervention was performed in children with hydrocephalus (13/27). Omayya reservoir was placed in seven children, of which five needed conversion to ventriculoperitoneal (VP) shunt; direct VP shunt was carried out in six (6/13). Good outcome was noted in 78% at discharge. Stage III TBM (P = 0.0001), cerebral infarcts (P = 0.0006), and motor deficits (P = 0.03) were associated with poor outcome. Sequelae included learning difficulties with poor scholastic performance (31.5%).
CONCLUSION: GeneXpert has high diagnostic specificity, but negative results do not rule out TBM. CSF GeneXpert provided quick results. Placement of Ommaya reservoir in TBM stage II and III with hydrocephalus was not successful. Hydrocephalus was managed conservatively with success (53%). Copyright:

INTRODUCTION

Extrapulmonary tuberculosis (TB) accounts for 25% of TB cases in children,[1] of which tubercular meningitis (TBM) is the most severe form, resulting in high degree of mortality and morbidity. Poor outcome is directly associated with delayed diagnosis. Definitive diagnosis of TBM by demonstration of mycobacteria in cerebrospinal fluid (CSF), by direct staining, or culture is time-consuming and rarely positive. Early diagnosis and treatment plays a crucial role in the management of TB.

Gene Xpert MTB/RIF test is a cartridge-based fully automated nucleic acid amplification test for the detection of TB and testing for rifampicin resistance. It has a quick turnaround time of 2 hours. In December 2010, the World Health Organization (WHO) endorsed the GeneXpert technology, released a recommendation and guidance for countries to incorporate the new test into their programs. Bhatia R et al.[2] reported the sensitivity of GeneXpert in CSF as 38.24% compared to Bactec culture (14.71%). Data on the yield of GeneXpert in CSF in Indian children are limited. Hence, we planned this study to look for any change in clinical profile, laboratory findings, yield of GeneXpert, role of neurosurgical interventions, and outcome at follow-up in pediatric TBM.

MATERIALS AND METHODS

This retro-prospective observational study was conducted from January 2015 to June 2018, at the pediatric neurology department of a tertiary care pediatric center, India, after obtaining approval from the institutional ethics committee. Prospective participants were included after obtaining a written informed consent from their respective parents and assent from eligible patients.

Children aged 1 month to 16 years with definitive, probable, possible TBM[3] were included in the study; known cases of TBM treated elsewhere and those in whom the diagnosis of TBM could not be established were excluded.

Details pertinent to demography, clinical profile, clinical examination, staging, and progression of the disease were noted. Headache aged and Meningeal signs were taken into consideration in children aged >2 year. In symptomatic children, contact with any form of active TB within the last 2 years was considered significant.[4] Staging of the disease was carried out according to the British Medical Research Council.[5] Mantoux test was performed with 1 TU contained in 0.1mL of PPD R23 with Tween 80 solution. CSF was analyzed for cell type and count, protein, and glucose. CSF GeneXpert was performed under Revised National Tuberculosis Control Program (RNTCP). Results of the GeneXpert were classified as Mycobacterium tuberculosis detected (rifampicin sensitive/indeterminate/resistance), M. tuberculosis not detected, invalid test, and error/no result. An independent radiologist analyzed radiological findings of computed tomography (CT) and/or magnetic resonance imaging (MRI) of brain.

Definite, probable, and possible TBM were diagnosed according to the uniform research definition criteria.[3] It is based on a composite score consisting of clinical, CSF, cerebral imaging criteria, and evidence of TB in other systems.

Patients were treated with antitubercular drugs according to the RNTCP guidelines (rifampicin [10–15 mg/kg], isoniazid [10 mg/kg], ethambutol [20–25 mg/kg], and pyrazinamide [30–35 mg/kg]) for 9–12 months.[6] Need for surgical procedure, ventriculoperitoneal (VP) shunt or Ommaya reservoir placement, was based on clinical presentation. Response to treatment was judged by modified Glasgow Outcome Scale (GOS). Low and moderate disability was classified as good outcome, whereas severe disability, persistent vegetative state, and death were classified as poor outcome. Presence of motor deficits, seizures, and cognition was assessed on follow-up for a minimum period of 3 months. Similar data for the retrospective study were collected from the medical records and were followed up on outpatient basis.

Statistical analysis: Results were expressed as median and interquartile range for continuous variables and as percentage and frequency distribution for categorical variables. Chi-square test and Fischer’s exact t test were applied. Analysis was carried out using the Statistical Package for the Social Sciences (SPSS, version 20.0). Two-sided p value of <0.05 was considered significant.

RESULTS

We included 36 children (boys = 23, 64.0%; girls = 13, 36.0%) who met the selection criteria. The median age was 39 months with a age range of 3 months to 16 years. Most common age-group affected was 1–5 years (n = 18, 50%) followed by 11–16 years (n = 7, 19.4%) [Figure 1].

Figure 1

Age-wise distribution of study population

All patients had fever; headache (n = 19 of 23, 82.0%), vomiting (n = 24, 66.0%), and altered sensorium (n = 25, 80.0%) [Table 1] were the other common features. Five (14.0%) complained of cough, and 12 (33.0%) had history of contact with TB. None of them had generalized lymphadenopathy. Thirty-two cases had received BCG vaccination as part of immunization schedule; four were not vaccinated.

Table 1

Clinical presentation among study population

Clinical features	Frequency (n = 36)	Percentage (%)
Fever	36	100%
Headache*	19/23	82%
Poor feeding	29	80%
Altered sensorium	25	69%
Seizures	23	63%
Vomiting	24	66%
Meningeal signs**	27/29	93%
Papilledema	9	25%
Cranial nerve deficits	19	52%
Motor deficits	19	52%

Fourteen (38.0%) patients had severe deterioration in sensorium (Glasgow Coma Scale [GCS] <8), 11 (31.0%) patients each had moderate (GCS 8–13), and mild (GCS 13–15) deficit in sensorium, respectively. Hemiparesis (n = 15, 41%) was more common than quadriparesis (n = 03, 8%). Of cranial nerve deficits (n = 19, 52%), involvement of 6th (n = 09, 25%) followed by 7th (n = 08, 22%) and 3rd cranial nerve (n = 02, 5%) was common. Choroid tubercles were noted in three (8.3%) patients. Ptosis (n = 02), cerebral vision impairment (n = 09, 25%), and papilledema (n = 09, 25%) were the other clinical features. One each had Bickerstaff encephalitis and paroxysmal autonomic instability with dystonia (PAID) syndrome. Two children had spinal arachnoiditis.

Dystonia (n = 08, 22%) and hepatomegaly (n = 03, 8.3%) were other findings. Two children with stage III had diabetes insipidus.

Seven (20%) children had stage I, 18 (50%) had stage II, and 11 (30%) had stage III disease. Fourteen (39%) had definite TBM, 19 (53%) had probable, and three (08.0%) were diagnosed to have a possible TBM.

Mantoux test was positive in 50% (15/30) children with a mean induration of 16.7mm (±SD 9.4mm). Significantly elevated Erythrocyte Sedimentation rate (ESR) (ESR >50 mm/h) (n = 26, 72.2%) and abnormal findings on Chest X ray (n = 04, 11%) were noted. One patient each had Miliary TB, right lower lobe consolidation, and two had hilar lymphadenopathy. None of our study cohort had human immunodeficiency virus infection.

CSF was analyzed for cell count, protein, glucose, and GeneXpert. CSF pleocytosis (>5 cells/mm3), lymphocytic predominance was observed in all. Protein concentration (upper limit of normal range 40 mg/dL) was elevated in 34 (94.4%) children; concentration of >300 mg/dL was noted in 7 (19.4%) patients. CSF glucose (<40 mg/dL) was reduced in 24 (66.6%) patients.

CSF GeneXpert was carried out in 33 children; results were positive in 37% (12/33) cases. M. tuberculosis was sensitive to rifampicin in 11 children and resistant in one child. There was no association between GeneXpert positivity and CSF pleocytosis (P = 0.27), high CSF protein (P = 1.0), or low glucose (P = 0.46) levels in the CSF.

Features of neuroimaging are provided in Figures 2 and 3. Hydrocephalus (27, 75%) and basal exudates (26, 72.0%) were the most common findings.

Figure 2

Characteristic features on neuroimaging

Figure 3

Basal exudates with temporal horn and fourth ventricular dilatation seen in a child with TBM

There was no association between positive GeneXpert in CSF and the presence of hydrocephalus (P = 0.20), basal exudates (p = 0.39), and cerebral infarcts (P = 0.43) [Figure 4]. However, GeneXpert was positive in 40% cases with hydrocephalus compared to 12% without hydrocephalus.

Figure 4

Left thalamic, right caudate and lentiform infacts in a child with TBM

Management and outcome: All were initiated on category 1 antitubercular therapy (ATT). One child had rifampicin resistance and was started on levofloxacin, cycloserine, amikacin along with isoniazid, pyrazinamide, and ethambutol. Dexamethasone (0.6 mg/kg/day) was started in all, followed by oral prednisolone at discharge and continued for 6–8 weeks. One child developed hepatitis and required change of drug. Vision impairment was noted in a child after 5 days of starting ATT. Choroid tubercles with optic neuritis were noted on fundal examination. Ethambutol was stopped and pulse dose of methylprednisolone was prescribed, following which there was a gradual improvement of vision.

Of those with hydrocephalus, 14 (53%) were managed conservatively, others were managed surgically [Figure 5]. Ommaya reservoir was placed in seven (26%) children. But in view of progression of symptoms, it was converted to VP shunt in 5/7 cases. Direct VP shunt was placed in six (22%), of whom five had stage II TBM, whereas one had stage III TBM. Revision surgery was needed in one child due to shunt malfunction secondary to blockade.

Figure 5

Surgical management of TBM (original)

During the course of hospital stay, three (8.3%) patients died. Treatment outcome was good in 28, (78%) and poor in eight (22%). Stage III TBM (P = 0.0001), cerebral infarcts (P = 0.0006), presence of motor deficits (P = 0.03), and low GCS at presentation (P = 0.03) were associated with poor outcome. CSF pleocytosis (P = 0.669), protein (P = 0.549), glucose (P = 1.00), hydrocephalus (P = 0.396), and basal exudates (P = 0.392) did not correlate with any adverse outcome.

The mean duration of follow-up was 8.28 months (±SD 3.7). At follow-up, there were three deaths (after a mean duration of 3 months) amounting to six deaths. Learning difficulties with poor scholastic performance in the form of reading and writing difficulties were evident in 6/19 (31.5%) of school-going children at follow-up. Facial nerve palsy was noted in one child. Epilepsy (3/30, 10.0%), motor deficits (14/30, 46.6%) were noted at follow-up. Among 14 children with motor deficits, 10/14 (71%) had independent ambulation, 3/14 (22%) were ambulant with support, and 1/14 (7%) was bed bound.

DISCUSSION

Diagnosis of tuberculosis remains a difficult task despite the availability of various batteries of investigations. A good clinical history and examination with high index of suspicion plays an important role in early diagnosis. As the available investigations have their limitations and take sufficient turnaround time, there is a need for a test that detects TB early. GeneXpert MTB/RIF is filling this gap as it gives results in ~ 2hours.[7] It has comparably greater sensitivity,[28] thus, making it a suitable option for early detection of TBM.[910]

The mean age of presentation was 39 months with age group 1–5 years being the most commonly affected (50%) similar to the study by van Well et al.[11] (< 5 years, 82%),[8] indicating that pediatrician has to be highly suspicious when children of this age-group present with symptoms suggestive of TBM. We report a male preponderance (63.0%) similar to previous studies (70%–74%).[12131415] In our study, 88.8% received BCG vaccination compared to 16% reported by Farinha NJ et al.[16] Lower rate of BCG vaccination reported by the latter could be attributable to the study period (1977–1997) and geographical area (United Kingdom).

The most common presenting symptoms included fever, headache, altered sensorium, seizures, vomiting, and irritability. Fever was the predominant symptom in all children with a mean duration of 21.8 (±SD 17.4) days, which was similar to previous studies.[1718] Farinha NJ et al.[16] (66%) and Nabukeera-Barungi N et al.[19] (83%) have reported lesser prevalence of fever in their patients. We noted seizures and vomiting in 66% of cases similar to previous reports (50%).[1718] Farinha NJ et al.[16] reported that seizure is a common presentation (53%). Generalized tonic–clonic seizure was common (41.6%), followed by focal seizures (22.2%). Altered sensorium (69%) with a mean duration of 5.2 days was observed in our patients similar to the observation of Nabukeera–Barungi N et al.[19] (72%). Irritability (28%, mean duration 11 days) was reported in our patients; Nabukeera-Barungi N et al.[19] reported irritability in 85% of cases in their study. TBM must be considered as one of the possible diagnoses, and it must be investigated in children with fever, headache, and altered sensorium.

In our study, two children with stage III had diabetes insipidus; they responded well to desmopressin and vasopressin infusion. Two children had spinal arachnoiditis, which gradually improved on ATT and hyaluronidase therapy.

Thirty-two (88.9%) patients received BCG vaccination as part of immunization schedule; four were not vaccinated with BCG. This observation prompts us to introspect the role of BCG vaccination in preventing TBM. Khemiri M et al.[20] too noted that TBM is a life-threatening condition even in BCG-vaccinated children. Lower vaccination rate (41.7%) is reported by Wasay M et al.[21] from Pakistan who noted lesser tuberculoma formation in vaccinated patients. Güneş A et al.[15] too reported that the proportion of vaccinated patients remained less (24.3%).

We elicited history of contact with TB in 34% of cases, similar to the report by Gupta BK et al.,[22] Yaramiş A et al.[23] (66.0%), and Farinha NJ et al.[16] (47.0%) reported a higher percentage of history of contact with TB. Thus, it is important to elicit this history, and that family members must be screened for TB.

Uniform consensus definition helps to stratify patients and maintain a uniform approach for the diagnosis of TBM. In our study, definite TBM was present in only 39%, probable TBM in 53%, which is comparable to a study by Solomons RS et al.[3] (11.9%, 73.6%) and Nabukeera-Barungi N et al.[19] (15.0%, 42.5%). Mantoux test was positive in 50% patients, whereas Güneş A et al.[15] have reported lower Mantoux positivity in their study. In addition, significantly elevated ESR (72.2%) and abnormal chest X-ray (11.0%) were noted in our patients.

CSF analysis provides a valuable clue in the diagnosis of TBM.[24] In our study, CSF results were abnormal in all. Low levels of CSF glucose (66.6%) and elevated levels of protein (94.4%) comprise other characteristic abnormalities, which were evident in our series. This was similar to a study conducted by Aketi L et al.[25] in which low levels of CSF glucose and high CSF protein was present in 91.6% of their cases. Normal CSF glucose levels were observed in 33.4% children, indicating that CSF glucose >40 mg/dL is not against the diagnosis of TBM.

Current WHO guidelines recommend using GeneXpert as an initial investigation for TBM. However, its low sensitivity and ability to confidently rule out disease are debated.[26] Data is lacking in pediatric population as most of the studies are centered around adult population. GeneXpert was positive in CSF in 37% (12/33) cases in our study. Rifampicin-resistant TB was noted in one child. In a study by Bhatia R et al.,[2] sensitivity of GeneXpert in CSF was 38.24% compared to Bactec culture (14.71%). A higher, sensitivity of GeneXpert was reported by previous studies (59.3%–72.3%);[2728] Heemskerk AD et al.[28] also reported an independent association with higher CSF volume, lactate levels, and lower CSF:blood glucose ratio. In contrast, we found no association between GeneXpert positivity and pleocytosis, high protein and low glucose levels in the CSF. However, in those with CSF cell count >100/mm3, 6/12 had positive GeneXpert compared to 6/21, with CSF cell count <100/mm3. To the best of our knowledge, no similar comparative study from India on TBM in children to compare association between yield of CSF GeneXpert positivity and other elements of CSF.

Neuroimaging (CT/MRI) was performed for all our patients. It provided a diagnostic clue to TBM in 75% of patients, and helped in early diagnosis and initiation of treatment in children with TBM. MRI is superior to CT in detecting basal enhancement and identifying infarcts, thus helping in diagnosis and determining prognosis; CT has a role to play in detecting hydrocephalus.[29]

Hydrocephalus is a common associate of TBM. Güneş A et al.[15] have reported that 90.3% had hydrocephalus. We managed children with hydrocephalus conservatively (53%). In those with hydrocephalus under pressure, without imminent risk of herniation, and close to medical facility, Ommaya reservoir placement (26.0%) (to avoid shunt dependency) was carried out with reservoir tapping as and when required. However, we had to convert it to VP shunt in 71% (5/7) cases indicating this option is not very successful in our cohort. Lin J et al.[30] performed Ommaya reservoir placement in 12 children with hydrocephalus. Four (33%) required Ommaya reservoir followed by VP shunt placement; two children developed serious disabilities and one of them eventually died. They proposed that Ommaya reservoir placement can prevent placement of VP shunt in TBM hydrocephalus. The conversion to VP shunt in our series was high, possibly indicating more severe hydrocephalus.

We noted good and poor outcome in 78% and 22% patients, respectively. Mortality in our study during the hospital stay was 9.4%. In contrast, Aketi L et al.[25] noted lower cure rate (35.5%), sequelae (35.5%), but higher mortality rate (29.0%). Better outcomes may be reflective of better supportive care and early identification in our cohort. Seizures, hydrocephalus, basal exudates, CSF cell count, protein, and glucose exert no influence on outcomes; motor deficits, stage of disease, low GCS at presentation, and cerebral infarcts are associated with poor outcome at discharge. No BCG vaccination, stage III disease, focal weakness, raised intracranial pressure, low GCS score, seizures, basal exudates, and infarcts on CT scan were associated with poor outcome.[113,31] On follow-up, learning difficulties and motor disabilities were noted in 31.5% and 46.6% children, respectively. However, we could not standardize behavioral abnormalities using a standardized questionnaire as part of the study was retrospective.

The strengths of our study include analyzing the yield of GeneXpert in diagnosing TBM, correlating GeneXpert with CSF, and radiological parameters. We also assessed the outcomes in these children and the association with variables such as GCS, laboratory, and radiological findings. To the best of our knowledge, this is the first study conducted on Indian children, wherein GeneXpert with various CSF and radiological parameters were analyzed.

Limitations of the study included a small sample size, single centre experience and partly retrospective data wherein some data were not available. We recommend a prospective study in larger population to evaluate the sensitivity of GeneXpert in TBM.

CONCLUSION

Use of consensus case definition of TBM helped us to classify and diagnose children with TBM, and in early initiation of treatment. GeneXpert is highly specific for the diagnosis of TBM; however, negative results do not rule out TBM. Use of CSF GeneXpert in children with TBM provided quick results but it was positive in only 37%. Clinical findings, biochemical, CSF, and radiological investigations continue to play an important role in the diagnosis of TBM. Placement of Ommaya reservoir in children with hydrocephalus in TBM stage II and III is not very successful.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.