Introduction

Tuberculosis has continued to spread unabated for too long. In 2011 there were an estimated 8.7 million new cases of tuberculosis (TB) worldwide, and 1.4 million people died from TB [1]. Central nervous system (CNS) tuberculosis accounts for approximately 1% of all cases of active tuberculosis and associates a high mortality and residual neurologic sequelae, even with adequate treatment [2, 3].

In Romania, the overall incidence of tuberculosis was approximately 100 cases per 100,000 population in 2011, with a decreasing trend from an incidence of 142 per 100,000 population in 2002, but the number of severe cases of TB in children (meningoencephalitis, miliary, cavitary) still maintains a high annual rate (65/100,000 in 2011) [4].

Natural history and clinical manifestations of tuberculosis are different in children and in adults [5], with a higher incidence of disseminated forms and a higher risk to progression to severe forms being reported to occur in children [5]. Note, however, that these differences are reported in different studies and that, in general, there are no unified studies carried out to directly compare the outcome of tuberculosis in the same setting in the same time and in the same diagnostic and treatment conditions.

In our region, different from other medical systems, where TB patients are hospitalized in separate wards (in pediatric hospitals and adult wards of infectious diseases or pneumophtisiology), children and adults are hospitalized in the same ward. This allows us to gain a global vision and compare and contrast the clinical and pathological features of TB meningitis in the two age categories to design appropriate diagnostic followed by an early introduction of treatment.

Here, we present our 10 year experience of managing adult and pediatric patients diagnosed with TBM in Northeast Romania. It is important to stress that, while in the recent years the diagnostic and interventional abilities dramatically improved in Romania, such modern and sophisticated diagnostic tools for TBM were not available in the past. Therefore, our study offers a real-world infectious disease clinic perspective, being thus representative for the clinical environment of developing countries. Such an approach might be useful not only to develop diagnostic and therapeutic guidelines applicable to poor resource settings, but also to identify those critical needs that have to be addressed to dramatically improve the diagnostic and treatment of patients with TBM in the developing world.

Patients and Methods

Results

Clinical stage

The vast majority of patients (>62%) were in the stage 2 of TBM at admission. A smaller fraction of patients was in the first stage (15%). There was a significant difference in the stage of infection at the time of admission between children and adults. Thus, >25% of children and only 8% of adults (p = 0.00012) were diagnosed very early in the infection.

The observation that children with TBM are admitted in earlier stages of infection compared to adults may have direct clinical implications, as a late diagnosis or a late presentation to the hospital may dramatically affect future prognosis.

There were no significant differences with regard to the number of children and adults admitted in stage 3 (23.08% vs. 27.46%) (p>0.05) (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Epidemiological aspects, case definition and stage of TBM on admission.

https://doi.org/10.1371/journal.pone.0133477.t001

Definite diagnosis of TBM was more frequently established in children than in adults (31.16% vs. 17.32%) (p = 0.043), while criteria for a possible TBM diagnosis were predominant among adults (p = 0.044).

A comparison of the presenting clinical features and outcome in the 77 children and 127 adults with TBM is shown in Table 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Clinical features and infection outcome in patients with TBM.

https://doi.org/10.1371/journal.pone.0133477.t002

The vast majority of included children received BCG prophylaxis prior to current infection. Only 19% of patients had systemic signs of bacillary disease (weight loss, night sweats, lethargy, etc.). There were relevant differences between children and adults regarding the systemic signs of bacillary disease (8% vs. 25%, for children and adults, respectively, p = 0.0053).

Neurological state at the time of admission was: coma in 36 patients (18% children vs. 17% adults, p>0.05); other consciousness abnormalities (dizziness, confusion, changes in personality) were observed more frequently in adults (n = 90, 71%) than in children (n = 21, 27%) (p = 0.000001). Cranial nerve palsies were noted prior to therapy initiation in 17 cases (12% vs. 6% for children and adults, respectively, (p>0.05), and developed after 2–7 days of treatment in additional 21 patients (10%) (12% vs. 9% children and adults, respectively).

The average duration of symptoms before hospitalization was 9 days (range: 3–32 days) and was not significantly different between children and adults (Table 2).

Extrameningeal TB presented most frequently as pulmonary tuberculosis, in 80 cases (39%) (35% vs. 42% of children and adults respectively, p = 0.461), as well as adenitis in 3 cases and spondilodiskitis in 5 adults cases.

Major findings on chest radiological exam (Table 3) consisted of miliary pattern, observed in 32 patients (16%) (20% vs. 13%, of children and adults, respectively, p = 0.144), fibrocavitary lesions, diagnosed in 19 (9%) patients (3% children vs. 13% adults, p = 0.0174), upper lobe infiltration, in 16 patients (8%) (4% children vs. 10% adults) (p>0.05) and pleural effusions in 8 cases (4%) (3% children vs. 5% adults, p = 0.528) (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. CSF data and chest X-ray abnormalities.

https://doi.org/10.1371/journal.pone.0133477.t003

Tuberculosis skin test elicited a positive reaction in 21 of 58 tested patients (36%)(35% vs. 37% of children and adults respectively, p = 0.889). CSF smear for acid-fast bacilli was negative in all the cases. CSF culture was positive for M. tuberculosis in 49 cases (31% of children vs. 20% of adults, p = 0.024), and a positive interferon-gamma release test [QuantiFERON-TB Gold test (Cellestis Limited, Australia)] was observed in 40 of serum samples out of 59 tested (60% vs. 68.2% for children vs. adult patients) (p> 0.05).

Upon admission, CSF, which was available in all cases of TBM, revealed turbidity in 20 patients (10%) (9% vs. 10% for pediatric versus adult patients, respectively) (p> 0.05). In the CSF, polymorphonuclear cells represented >50% of the total leucocyte population in 17 cases (4% vs. 11% for pediatric vs. adult patients, respectively) (p = 0.016). Glucose levels were normal (≥ 0.4g/l) in the CSF at presentation in 69 patients (34%) (29% vs. 37% for pediatric vs. adult patients, respectively, p> 0.05). The initial CSF chloride levels were normal in 92 patients (46% vs.45% for pediatric vs. adult patients, respectively, p >0.05), with decreasing values during the first month of treatment in half of them. The protein CSF levels were elevated in all cases, ranging from 0.4 to 17 g/l (normal range 0.2–0.4 g/l), with very high levels (5–15 g/l) being more frequently seen in adults (18%) than in children (3%) (p = 0.0021).

Treatment was empirically instituted within a few hours to 3 days postadmission in the majority of patients and included a combination of four antituberculous drugs (isoniazid, rifampicin, ethambutol and pyrazinamide) during the initial 2–3 months and two drugs (isoniazid, rifampicin, 3 days weekly) for the next 7–9 months [8].

Drug related hepatic toxicity was recorded in 21 cases (10%) and was more frequent in adults than in pediatric patients (4% vs. 14% of children and adults respectively, p = 0.0145).

Isolates of M. tuberculosis were tested for isoniazid susceptibility only in 4 cases (one strain was resistant).

The overall in-hospital mortality for our group of patients was 7% (15 cases) (8% of children vs. 7% of adults, p>0.05) whilst permanent neurological sequelae (hydrocephalus, cranial nerve palsies, hemiplegia) were seen in 46 patients (23%), more frequently in children than in adults (36% vs. 14% for pediatric vs. adult patients, respectively, p = 0.0121). Patients were discharged after an average duration of hospitalization of 28 days.

Discussion

TBM is the most lethal clinical presentation of Mycobacterium tuberculosis infection, with a mortality rate that may be exceptionally high (6–65%, average 33%), despite treatment with antituberculosis chemotherapy [9]. TBM is also responsible for severe disability in a vast proportion of survivors [9, 10].

In this study, we compared and contrasted the elements used for the diagnosis of TBM in children and adults with the same demographic conditions, the same limitations regarding imaging technology and exposed to the same strains of M. tuberculosis. The vast majority of these patients were admitted in critical condition in the Infections Disease Clinics of the University Hospital of Iasi, Romania. By using standardized clinical criteria we were able to early diagnose patients with TBM in the majority of cases, and to initiate specific treatment within the first days after admission. At the time of the diagnosis in these patients, resources in our hospital being limited, our group is illustrative for real-life practice in poor resources settings.

At first consultation, most of the patients in our study group presented in an already advanced stage of disease (stages 2 and 3), while only 14% of patients presented in stage 1, similar to previous reports by other groups [11, 12], but different from the data reported by Maltezou on a small sample size [13].

The explanation of this situation could be the low level of education and limited accessibility to specialized services (most of patients came from rural areas). Other reason for the delayed presentation may rely on misinterpretation by general practitioner and pediatricians of nonspecific signs and symptoms during the first week of disease (the involvement of CNS was suggested later, when meningeal signs occurred). Moreover, especially in children, late presentation might have been due at least partially to a characteristic rapid progression to severe forms [5].

In our ten-year survey, 24% of patients were found to have positive cerebrospinal fluid cultures for M. tuberculosis, with a higher rate in children than in adults (p>0.05), in agreement with previous studies [14, 15, 16]. Thus, Thwaites et al., reported a 29% positivity rate for M. tuberculosis cultures in 432 uninfected HIV patients with TBM [14] and in a recent study, Blount et al., noted a culture confirmation rate of 27% in children with intra and extrathoracic tuberculosis [17]; Rajeev and Desai found similar positivity rates in India [15, 16]. Conversely, Gijs et al., reported only 11.7% positive M. tuberculosis cultures in South African children [12].

Multiple studies reported that miliary tuberculosis is directly involved in the pathogenesis of tuberculous meningitis in children. In our study nearly 20% of pediatric patients had evidence of miliary tuberculosis, strongly suggesting CNS involvement [10]. These rates were significantly higher than those recently reported by Hristea et al., on patients from Southern Romania, in which miliary tuberculosis was associated to only 7% of cases of MTB [18].

In our study, the most consistent difference regarding the chest X-ray abnormality between the two groups of patients were the fibrocavitary aspect, more frequently seen in adult (13%) than in pediatric patients (3%)(p = 0.0174).

Definite diagnosis of TBM was more frequently achieved in children (31%) than in adults (17%). This difference may rely on the frequent miliary pattern and the highest rate of successful ex vivo isolation of M. tuberculosis in children, which were similar to a previous report of Thwaites et al., in an uninfected HIV population with TBM in Vietnam [13].

Unlike the results of other investigators [10, 12, 19], in our study the proportion of children presenting with coma, seizures and cranial nerve palsies was similar to that observed in adults (18% vs. 17%, 14% vs. 9% and 12% vs. 6%, respectively). MTB also frequently (80%) associated an abnormal mental state, but in children, coma and seizures were noted in a much lower proportion than in the recent study of Blount [17]. The duration of symptoms before admission was relatively short (9 days). Both these clinical aspects were different from those previously reported by Hosoğlu et al. [19].

Biomarker testing also revealed several remarkable features of MTB infection in children and adults: (i) no relevant difference was found between children and adults regarding CSF biomarkers, except for higher CSF protein content in adults (> 5 g/l); (ii) similarly, the levels of chloride and glucose in the normal range were similar between children and adults; this pattern could explain the delay of treatment in TBM. These observations might have a major therapeutic impact as, in the absence of significant changes of the CSF biomarkers, the decision to initiate treatment might be delayed, which may be reflected in a poor survival rate.

MTB complications were significantly more likely to develop in children (36%) and hydrocephalus was the most frequent of them (26%). This is a common complication of TBM which was reported to occur in 15–18% of adults and 57–98% of children [10, 11, 20, 21].

IFN-γ release assay was positive in 68% of cases and had a higher specificity than tuberculin skin test. Yet one of the limitations of the IFN-γ release assay is that it cannot distinguish between M. tuberculosis infection and active disease [6].

Decision to start antituberculosis treatment was empirical in all cases. WHO guidelines recommend a 6 or 9 months course of antituberculosis treatment [1]; however, other guidelines recommend a prolonged therapy extended to 9 or 12 months if low suspicion of drug resistance or up to 24 months if high suspicion of drug resistance. Association of corticosteroids may reduce the number of deaths [8, 22]; therefore, all the patients in our cohort received corticosteroids associated to specific antituberculosis treatment. Similar with the data from a recent systematic review by Donald [23] we found that drug-induced hepatotoxicity was considerable lower in children than in adults. In the same review, when considered papers describing antituberculous drug hepatotoxicity, authors noted variable frequency from less than 1% to 50% in some studies involving children treated for tuberculous meningitis, differences explained by disease severity [23].

Clinical deterioration after administration of antituberculosis drugs is a common feature in TBM, and it is known as “paradoxical”. In our study, cranial nerve palsy occurred in 21 (10%) and coma developed in 17 (8.33%) after the introduction of adequate therapy, without relevant differences between children and adults. Note that these features might be due to bactericidal drugs, such as isoniazid, which may cause the diffuse release of bacillary contents which is responsible of inflammation and edema [10, 24]. Despite a dramatic initial course, 70% of our patients responded promptly to the treatment with antituberculosis drugs and subsequently fully recovered.

The in-hospital mortality rate was 7%, which is among the lowest rates mentioned in the literature (which range from 7 to 57%) [11, 13, 21, 25, 26]. The low mortality rate in our study could be an argument against a high level of resistance in North Eastern Romania. According to WHO recent report, levels of multidrug-resistant tuberculosis (MDR-TB) remain worryingly high in some counties, notably countries in Eastern Europe and Central Asia, where 9–32% of new cases have MDR-TB and more than 50% of previously treated cases have MDR-TB [1]. Another reason for the low mortality in our study may relay to the use of a rigorous, standardized diagnostic strategy in these patients. As such, our results demonstrate that even in a poor resource environment, prognostic of TBM may be acceptable if rigorous diagnostic strategies are in place.

In conclusion, adults were more frequently admitted in the late stages of the diseases and atypical CSF findings such as normal CSF chloride and glucose level were more common than in children. Conversely, children with TBM were more likely to be contacts of tuberculosis cases, to have a normal CSF protein level, a positive CSF culture, and an evolution with neurological complications in a higher proportion comparing to adults.

The rapid evolution towards a negative outcome of the TBM in our cohort strongly suggest that, in areas with a high incidence of TB an improved therapeutic success can be achieved by improving the diagnostic abilities by implementing rapid diagnostic tests such as interferon-gamma release test or PCR for Mycobacterium detection, as well as cerebral imaging. Altogether, these approaches may lead to an earlier initiation of treatment in patients with associated risk factors. This is particularly important for a clinical condition as TBM, which evolve quite a while before admission.