Epidemiological study of bacterial meningitis in Tunisian children, beyond neonatal age, using molecular methods: 2014–2017

Background Since the 1990s, the epidemiology of bacterial meningitis worldwide has changed thanks to vaccination. In Tunisia, the main causative pathogens were Streptococcus pneumoniae, Neisseria meningitidis and Haemophilus influenzae serotype b (Hib). Only Hib vaccination was available during our study period. Objectives We performed a laboratory case report based-study of suspected bacterial meningitis in Northern Tunisia from January 2014 to June 2017. Methods CSF samples obtained from children beyond neonatal age with suspicion of meningitis were tested by two real time PCRs, targeting pneumococcus, meningococcus and Hib, and conventional methods. Results Using real-time PCR, 63 were positive including ten supplementary cases compared to conventional methods. A general decrease of bacterial meningitis cases was demonstrated comparing to previous data. Pneumococcus was predominant (69.84%) followed by meningococcus (28.57%) and Hib (1.59%). The main serotypes were 14, 19F, 6B and 23F for pneumococcus and serogroup B for meningococcus. Most cases occurred during cold season and children under one year were the most affected by bacterial meningitis. Conclusion Our study suggests the predominance of pneumococcal cases. It may provide valuable data on meningitis epidemiology before the introduction of pneumococcal vaccine, which may be useful for future evaluation.


Introduction
Bacterial meningitis is a devastating disease associated with a high mortality rate and significant disability in survivors over the world 1-3 . Clinical diagnosis is difficult to assess, among infants. Moreover, bacteriological conventional methods present a limited sensitivity for low-quality cerebrospinal fluid (CSF) samples and antibiotic pre-treated cases 3,4 . Thus, bacterial meningitis cases remain underestimated, highlighting the need to use methods with high-performance detection such as real time PCR 4 .
Children beyond neonatal age are among the most exposed to bacterial meningitis. Historically, the major causative pathogens for this population worldwide were Neisseria meningitidis (N. meningitidis), Streptococcus pneumoniae (S. pneumoniae) and Haemophilus influenzae serotype b (Hib) 1 . However, in the last two decades, bacterial meningitis epidemiology changed thanks to the introduction of conjugate vaccines against the three bacteria. Indeed, the vaccination against Hib reduced the number of Hib meningitis cases by 90% in developed countries 5 . Additionally, the successive introduction of pneumococcal conjugate vaccines (PCV7, PCV10 and PCV13) reduced the number of meningitis cases and limited the possible serotypes replacement 5,6 . Furthermore, the conjugate vaccine against N. meningitidis serogroup C lowered the number of invasive diseases caused by this bacterium in different region of the world. For instance, in the United Kingdom, where meningococcus serogroup C conjugate vaccine was introduced in 1999, the number meningococcal invasive diseases decreased by 81%, from 1998 to 2001, with no evidence of serogroup replacement during the two years following the vaccine introduction 5 .
In Tunisia, a middle-income country, these three pathogens were the major agents of bacterial meningitis in children 7,8 . Most frequent serotypes were 14 and 23F for S. pneumoniae, serotype b for H. influenzae and serogroup B for N. meningitidis 8,9 . Hib vaccination was introduced during the 2002-2005 period, in the national vaccination schedule, leading to the decrease of Hib meningitis incidence 8,10 . However, the use of Hib vaccine was interrupted from 2005 to 2011 causing a rise of Hib meningitis cases 8,11 . Since 2011, it has been reintroduced a second time as a routine vaccination. Furthermore, since April 2019, pneumococcal vaccine has been introduced in the national schedule, using PCV10 12 . However, little data about bacterial meningitis epidemiology have been available since 2011 11 . Most studies focused on characterisation of one pathogen in a very restricted area 13-15 and did not consider the impact of the Hib vaccination reintroduction 11 nor the added value of molecular diagnosis in assessing bacterial meningitis epidemiology. With the aim of estimating the pathogen distribution of meningitis cases in infants and children in Northern Tunisia, we assessed a laboratory case report basedstudy over a three and half-year period using molecular and conventional methods.

Ethics statement and informed consent
The Ethical Committee of Bechir Hamza Children's Hospital in Tunis, Tunisia, approved this study (number 002/2017). The analysis was performed on the leftover of CSF samples obtained for diagnostic purposes. All information related to patient's identity from the sampling of this study (excluding age and sex) was treated confidentially in accordance with ethical guidelines and the Tunisian legal requirements.

Samples and Study design
From January 2014 to June 2017, we performed a prospective laboratory case report based-study. All CSF samples received in the laboratory of Microbiology of Bechir Hamza Children's Hospital, for bacterial meningitis diagnosis, were investigated. This hospital is the main paediatric hospital (a bed capacity of 347 and at least 30.625 admissions in 2017) gathering patients from all Northern regions of Tunisia (11/24 governorates), the most populated area, and also referred patients from other regional hospitals. The investigated samples (n=253) were obtained from children aged from 1 month to 15 years old, presenting various meningitis signs such as high fever, neck stiffness, photophobia, hypotonia and seizures 3,11 . CSF samples were evaluated by bacteriological conventional methods, i.e. bacterial culture, direct examination for microscopic detection of bacteria and Gram staining, and latex agglutination test 16 . Samples, characterised by at least 5 leukocytes per mm 3 , were selected and considered as suspected bacterial meningitis cases. These CSF samples were considered for PCR when they had a sufficient volume, i.e. at least 200μL. They were stored at -20°C until molecular investigations.. The reference strain of Streptococcus pneumoniae (ATCC 49619) and field strains of Neisseria meningitidis and Hib were used as positive controls for real time PCR tests. Field strains were obtained from CSF for H. influenzae serotype b meningitis case and blood culture for meningococcaemia case and isolated in the laboratory of Microbiology of Bechir Hamza Children's Hospital in Tunis (Tunisia).

Molecular diagnosis of bacterial meningitis
Bacterial DNA was extracted by heat-shock protocol for positive controls and by using the QIAamp DNA Mini Kit (Qiagen, Hilden, Germany) for CSF samples according to the manufacturer's instructions. For detection of bacterial meningitis, we performed a multiplex real time PCR for S. pneumoniae and N. meningitidis detection and a singleplex real time PCR for Hib detection by using TaqMan chemistry 17 .

Typing methods
Meningococcal serogroups were identified by latex agglutination test (Pastorex meningitis, BIO-RAD, Marne-la-Coquette, France) and confirmed by PCR 18 . Pneumococcal serotyping was performed by agglutination test (Pneumotest Latex, Statens Serum Institut Diagnostica, Hillerod, Denmark) for strains collected from 2014 to November 2015. Due to the lack of latex agglutination test, several conventional PCRs were performed for strains collected from December 2015 to October 2016 according to the Centers of Disease Control and Prevention (CDC) and World Health Or-ganization (WHO) recommendations 16 as well as Pai et al. 19 , using the Applied Biosystems kit in GeneAmp PCR System 9700 instrument (Applied Biosystems, Foster City, USA) followed by agarose gel electrophoresis.

Statistical analysis
Statistical significance was determined using EpiInfo software 7.2. The Fisher exact test was performed to compare meningococcal detection proportions by molecular and conventional methods (95% confidence intervals).

Detection rates of bacterial meningitis cases
Overall, 4892 CSF samples were obtained from children admitted in the Bechir Hamza Children's Hospital. Based on inclusion criteria, only 253 cases were considered as suspected bacterial meningitis cases, while the others corresponded to viral meningitis cases (4639 cases). All CSF samples (n=253) were investigated by conventional methods and 183 fulfilled the inclusion criteria for PCR testing (sufficient volume of at least 200μL) and were investigated by real time PCR (Table  1). Among the 183 cases, 31 CSF samples were collected after antibiotic treatment. Using real time PCR, 63 cases were detected positive among the 183 investigated cases while using conventional methods, only 53 cases out of 253 had positive results. All positive cases were confirmed by real-time PCR. Furthermore, ten additional positive cases were detected. The 63 positive cases included 44 cases (69.84%) of pneumococcal meningitis followed by 18 cases (28.57%) of meningococcal meningitis and one case (1.59%) of Hib meningitis. The supplementary detected cases corresponded to six meningococcal meningitis and four pneumococcal meningitis cases. Real-time PCR had a significantly higher rate of detection of meningococcal meningitis than conventional methods given that it revealed one-third more positive cases (P=0.005). Among the 31 CSF samples collected after antibiotic treatment, eleven (35.5%) were positive: eight were positive by real time PCR and three were detected by both methods.

Detection rates according to sex and age
For bacterial meningitis cases, approximately two-thirds corresponded to boys (68.25%). Age information was available for 156 suspected meningitis cases: 49 of them were positive cases whereas 107 of them were negative. Among the positive cases, the average age was 2 years and 4 months (SD= 2 years and 2 months, IQR= 3 years and 6.5 months, range= 2 months-11 years) and the median age was 10 months. Most positive cases corresponded to children aged one year and younger (57.14%). For children aged between 1 and 2 years, the rate was lower (8.16%). For children aged between 3 and 4 years, the rate was 18.37% (Table  2). For the 107 negative cases, most children were aged one year and younger (63.55%) ( Table 2).

Distribution of meningitis cases according to years and seasons
Pneumococcal meningitis cases were the most observed during our study period; six cases in 2014 and eighteen cases each in 2015 and 2016. For meningococcal meningitis, the number varied from three cases (2015) to six and seven cases (2016 and 2014 respectively). Only one case of Hib was detected in 2015 ( Fig. 1). The distribution of bacterial meningitis cases revealed that most cases had occurred in winter and spring, with 28 cases (44.44%) and 18 cases (28.57%) out of 63 respectively. (Fig. 2).

Discussion
Bacterial meningitis is a major concern for public health worldwide and especially in developing countries 5,16,20,21 . This study describes the distribution of the three main causative bacteria of meningitis cases in Northern Tunisian children beyond neonates, after the reintroduction of Hib vaccination, and before the introduction of pneumococcal vaccination, from January 2014 to June 2017. It presents valuable results, which can be enlightening on bacterial meningitis in Tunisia given that it concerns the most populated governorates in Tunisia This study suggested an important decrease or absence of Hib cases in Northern Tunisia (only one case detected in our study) and the predominance of pneumococcal cases. Such epidemiological study should be followed by others studies from Central and Southern Tunisia targeting the majority of health centres to confirm this conclusion. Our results are informative at the national stage and may contribute to reshaping the general view on national vaccination strategy by adapting the selection of pneumococcal vaccine to the circulating serotypes. At the regional level, it may be also of great interest to other low-and middle-income countries experiencing the same situation such as Egypt, Malaysia or Iran 23,24 .
On the other hand, the added value of this study was the use of real time PCR, which allowed the detection of 10/62 additional cases to those obtained by conventional methods, especially for meningococcal cases (P=0.005). Indeed, it was previously demonstrated that molecular methods, targeting microbial genome, have the advantage to detect unviable bacterium. Therefore, they could overcome N. meningitidis alteration during transportation at ambient temperature and detect more positive cases 17,25 . Our significant results highlighted the need for an optimum framework to preserve meningococcus viability especially rapid transportation at 37°C 25 . Indeed, in high income countries, Hib meningitis cases decreased by 90% after introduction of Hib vaccination and in the middle-and low-income countries such as Morocco, they decreased by 75% 5,26 . Nowadays, this pathogen is considered to be eradicated in certain regions such as Europe 3 . Nevertheless, the recorded case corresponded to a vaccinated patient. This patient was given all Hib vaccine doses and thus immunological investigation was recommended. Other reports described Hib meningitis cases after vaccination and explained that it might be due to immunodeficiency or non-respect of vaccination schedule such as lack of booster shots 27 . Even though our study was restricted to the Northern regions of Tunisia, it presented valuable and different results from previous studies in the area 8,10,11 . In this study, we did not include neonatal meningitis cases because the epidemiology is different from children beyond neonatal age since other main pathogens, such as Escherichia coli, Listeria monocytogenes and Streptococcus agalactiae, are implicated. Furthermore, results obtained in our laboratory during the study period showed that S. pneumoniae, N. meningitis and H. influenzae serotype b were not detected among neonatal cases (unpublished data). Also, a depth analysis of 6A/6B pneumococcal strains detected by PCR should be carried for precise serotype identification.

Our findings demonstrated that bacterial meningitis
In conclusion, this study suggested a depletion or absence of Hib meningitis cases and the predominance of pneumococcal cases in Northern Tunisia, between 2014 and mid-2017. This finding gave an idea of a possible efficiency of Hib vaccination reintroduction, which should be confirmed by other studies targeting other regions of the country, and might be instructive at the local and regional level. On the other hand, it provided valuable data on meningitis epidemiology before the introduction of pneumococcal vaccine, which will be important to evaluate the effectiveness and the serotype coverage of the recently introduced pneumococcal vaccine PCV10.