Group B Streptococcal Neonatal Meningitis

SUMMARY

Neonatal bacterial meningitis is a devastating disease, associated with high mortality and neurological disability, in both developed and developing countries. Streptococcus agalactiae, commonly referred to as group B Streptococcus (GBS), remains the most common bacterial cause of meningitis among infants younger than 90 days. Maternal colonization with GBS in the gastrointestinal and/or genitourinary tracts is the primary risk factor for neonatal invasive disease. Despite prophylactic intrapartum antibiotic administration to colonized women and improved neonatal intensive care, the incidence and morbidity associated with GBS meningitis have not declined since the 1970s. Among meningitis survivors, a significant number suffer from complex neurological or neuropsychiatric sequelae, implying that the pathophysiology and pathogenic mechanisms leading to brain injury and devastating outcomes are not yet fully understood. It is imperative to develop new therapeutic and neuroprotective approaches aiming at protecting the developing brain. In this review, we provide updated clinical information regarding the understanding of neonatal GBS meningitis, including epidemiology, diagnosis, management, and human evidence of the disease's underlying mechanisms. Finally, we explore the experimental models used to study GBS meningitis and discuss their clinical and physiologic relevance to the complexities of human disease.

INTRODUCTION

Bacterial meningitis is a severe and life-threatening infection associated with high morbidity and mortality that affects all populations, especially neonates (1). It is estimated that deaths from meningitis and sepsis in children younger than 5 years are higher than deaths due to malaria, with the most significant impact on the most impoverished communities (2). The neonatal period (first 28 days of life) has the highest lifetime risk of serious infections and is the most vulnerable time for survival. In newborns, group B Streptococcus (GBS; Streptococcus agalactiae) remains the leading causative agent (1, 3). Despite the existence of antibiotic therapies, bacterial meningitis is a significant cause of neurological disability worldwide, with considerable emotional, societal, and economic impact on individuals, families, and communities in both developed and developing countries (2, 4–6). Morbidity has not improved since the 1970s, with some countries reporting a rise in cases (7–10). The absence of specific symptoms, together with a high incidence of negative blood cultures, makes the diagnosis of meningitis more difficult in newborns than in any other age groups (1, 11). Delays in diagnosis and initiation of appropriate treatment have severe, long-lasting consequences for individuals, especially important in the neonatal population, as the pathogenic challenge overlaps a critical period for neurodevelopment. However, therapeutic strategies to decrease neurological sequelae in neonatal meningitis remain sparse and demanding.

For the development of optimized therapeutic interventions, aiming at improving brain injury outcomes, and maximizing the quality of life of affected people, more complete knowledge of GBS meningitis pathology is mandatory. Despite being studied for decades, the mechanisms underlying brain injury and devastating outcomes are not yet fully understood. Further research on meningitis pathogenesis is thus needed. Ideally, the appropriate setting would be the human neonatal brain. However, studies of meningitis in the human clinical setting are severely limited, considering that (i) only the brains of deceased babies can be studied, leading to biased results toward severe pathology; (ii) early stages of the disease are often asymptomatic, and consequently, the window for sampling is missed; and (iii) standard invasive sampling procedures are impractical in live, sick newborns. Thus, animal models that could reflect the complex human physiological functions and systemic interactions are essential tools to provide a breakthrough in understanding neonatal meningitis. A myriad of animal models of GBS disease have been developed, although most use nonnatural infection routes and usually bypass several steps of bacterium-host interactions, and therefore, they may not accurately recapitulate the clinical phenotypes.

In this review, we provide an overview of the current knowledge of GBS neonatal meningitis, including epidemiology, risk factors, clinical presentation, diagnosis, and preventive and management strategies. We also comprehensively review its current pathophysiological concept, including human evidence of the underlying mechanisms of disease. Finally, we outline the relevance of animal models and summarize the different animal models used to study GBS meningitis.

MENINGITIS

Bacterial Meningitis

Bacterial meningitis is defined as the acute inflammation of the meninges in response to bacteria and/or bacterial products that affects the pia mater, arachnoid mater, and subarachnoid space (12, 13). It is the most severe type of meningitis and the commonest bacterial infection of the central nervous system (CNS) in children. It develops very quickly, and the mortality rate can approach 100% if the disease is left untreated. The resulting inflammation can also involve the brain parenchymal vessels (vasculitis), the ventricles (ventriculitis), the inner ear, and the parenchyma itself (encephalitis) (14–16). The first description of a likely epidemic of meningitis dates back to the 17th century and was written by Thomas Willis, an English physician recognized as a seminal figure in the history of neurology and the neurosciences (17). It is believed that the first time the term “meningitis” was used with its modern meaning was in 1803, in a thesis written by the French army surgeon François Herpin (17). However, it was only after publication of the first textbook of neuropathology, written by Edinburgh physician John Abercrombie, that the term became widely disseminated (17).

Etiology.

Acute bacterial meningitis occurs worldwide, can develop in people of all ages, and causes significant morbidity and mortality, with an estimated 10 million cases in 2017 (18) and 300,000 deaths in 2015, according to the World Health Organization (WHO) (5). Surprisingly, although several microorganisms are able to disseminate and cause systemic infection, only a restricted number of pathogens invade the meninges (13, 19). Which causative agents are most common depends on the patient’s age and geographic location. Streptococcus agalactiae, commonly known as group B Streptococcus (GBS), remains the leading cause of meningitis during the neonatal period (1, 3). Other agents that are associated with neonatal meningitis include Escherichia coli, Listeria monocytogenes, coagulase-negative staphylococci, Staphylococcus aureus, and Klebsiella spp. (1, 3, 13). In older infants, children and adults, Neisseria meningitidis and Streptococcus pneumoniae become more common, but cases have substantially declined with the introduction of conjugate meningococcal and pneumococcal vaccines (13, 20, 21). In this review, we focus only on meningitis induced by GBS.

GROUP B STREPTOCOCCAL INFECTION

Neonatal GBS Disease

GBS is a beta-hemolytic, Gram-positive, encapsulated bacterium that commonly colonizes the gastrointestinal and genital tracts of more than 50% of adult individuals (22). GBS is a commensal organism of the normal vaginal and intestinal microbiome of healthy adults. However, due to its opportunistic nature, GBS can transition to a highly invasive pathogen under certain conditions. This bacterium can lead to severe disease in neonates and, occasionally, in postpartum women and individuals with an impaired immune system or underlying medical condition (22–24). It is estimated that GBS is present in the vagina, rectal sites, or both of up to 40% of healthy pregnant women, with maternal colonization as the leading risk factor for neonatal exposure and infection (22).

Two main syndromes of invasive neonatal GBS infections have been recognized, referred to as early-onset disease (EOD) and late-onset disease (LOD), classified by age at onset (22, 25). Cases of very late-onset disease have also been described (26). In EOD, the infection manifests in the first week of life (age 0 to 6 days), with symptoms usually within the first 12 h after birth and almost always clinically apparent in the first 24 to 48 h, whereas LOD is defined by GBS isolation after the first week of life, with cases relatively evenly distributed throughout 90 days of age (age 7 days – 3 months) (7, 22, 27, 28). Cases of very late-onset disease occur beyond 3 months of life (26, 29).

EOD and LOD differ in their pathophysiology and clinical manifestations. EOD usually presents as respiratory distress, followed by septicemia, pneumonia, and, more rarely, meningitis. On the other hand, LOD commonly manifests clinically as bacteremia without focus and meningitis (7, 22, 27, 30–34). Less commonly, LOD can present as cellulitis, osteomyelitis, septic arthritis, necrotizing fasciitis, pneumonia, and adenitis (22, 30, 33, 35). Very late-onset disease still lacks more comprehensive studies. Reports suggest that it is less frequent, can be associated with immunodeficiency syndromes, and may be responsible for meningitis, pneumonia, sepsis, endocarditis, urinary tract infection, and septic arthritis (22, 26, 29). In EOD (including early neonatal meningitis), bacterial acquisition by newborns is most likely due to inhalation of contaminated amniotic or vaginal secretions during labor (vertical transmission) (25, 36). In contrast, maternal colonization is not present in all LOD cases, and the mode of infection remains largely unclear. It has been shown that at the time of LOD diagnosis, mothers had high rates of GBS carriage, and the molecular typing was indistinguishable from that of the infants, suggesting vertical transmission (33, 37). Horizontal transmission through nosocomial sources, nonmaternal caregivers, and possibly infected human milk has also been proposed as a mode for transmission of infection (38–40).

Epidemiology and Risk Factors

Worldwide epidemiological data from 2000 to 2017 reported an estimated overall incidence of invasive neonatal GBS disease of 0.49 per 1,000 live births (41). Estimations vary by geographic region, with an incidence of 0.53 per 1,000 births in Europe and 1.12 per 1,000 births in Africa (41). Recent epidemiology data show that in the Netherlands, the United Kingdom, and Ireland, invasive GBS disease is increasing (8, 9), and in the United States, rates of LOD are now higher than those of EOD (EOD, 0.23 per 1,000 live births; LOD, 0.31 per 1,000 live births) (7). Similarly, recent epidemiologic data from France reported an increase in LOD cases since 2013 (10). Importantly, GBS is estimated to cause the death of 90,000 infants, and 57,000 stillbirths annually worldwide (42). The case fatality rate among cases of EOD is 10.0%, ranging from 5.0% in developed countries to 27.0% in Africa (41, 43), whereas LOD has a mortality rate that can range from 4.0% in developed countries to 12.0% in Africa (41, 43). Among disease-causing isolates, serotyping of the capsular polysaccharides uncovered a high prevalence of serotype III in more than half of all cases (61.5%), followed by serotypes Ia (19.1%), V (6.7%), Ib (5.7%), and II (3.9%) (41). A singular capsulated serotype III GBS clone, belonging to the hypervirulent clonal complex 17 (CC17), is almost exclusively associated with meningitis and LOD cases and often referred to as the hypervirulent clone (44–46).

Uncertainty remains as to the causes of the recent rise in cases. In the United States, the dominance of LOD might be explained by the decline in EOD rates, whereas LOD remained stable over the study period (7). Nevertheless, in the United States, a significant increase in cases of LOD owing to serotype III was observed, and other countries with an increased number of cases have reported notable deviation over time in clones responsible for both EOD and LOD (7, 8, 47). Although this could be a reflection of the shift in the lifestyle pattern observed among communities and changes in clinical practices, with the concomitant evolution and adaptation of bacteria (selection and clonal expansion because of selection pressure), these trends are concerning, as a significant increase of virulent clones such as CC17 strains was observed. Indeed, in the Netherlands, since the mid-1990s, the CC17 GBS lineage became dominant in neonates with invasive disease, and within this clone, the intralineage CC17-A1 expanded and became the most prevalent by 2007 (47). In France, the prevalence of CC17 strains also increased over the last 13 years, as well as a multidrug-resistant sublineage of CC17 GBS (10). Recently, epidemiologic data from Portugal also reported the emergence of a CC17 sublineage, characterized by the loss of PI-1 (CC17/PI-2b) and by the resistance to multiple drugs (48). Nevertheless, these rates may underrepresent the true burden of invasive neonatal GBS disease (42). There is still a lack of precise global estimates and wider genomic surveillance of GBS, as most studies focus on high-income countries, with fewer data from other populations, where most of the disease burden is likely to be.

There are a myriad of obstetrical risk factors associated with increased risk for GBS disease, including (i) preterm birth (before 37 weeks’ gestation); (ii) low birth weight; (iii) low birth length; (iv) prolonged rupture of membranes (>18 h); (v) intrapartum maternal fever (temperature of ≥38°C); (vi) history of GBS disease in a previous pregnancy, and (vii) use of invasive measures during delivery, such as frequent intrapartum vaginal examinations and invasive fetal monitoring (1, 27, 30, 49). Newborns born to HIV-infected mothers have an increased risk for invasive GBS disease in both developed and developing countries (50, 51). Regarding LOD in particular, a study revealed that babies born to mothers living with HIV are at higher risk, whereas this was not the case for EOD. While HIV infection is not associated with increased maternal GBS carriage, HIV-exposed infants were 6.85-fold more likely to develop LOD than HIV-unexposed infants (50, 52). Additionally, it was found that foul-smelling fluid is a risk factor for EOD, and maternal GBS bacteriuria is a risk factor for both EOD and LOD (50). The risk of invasive GBS disease in the twin of an infected baby can be as high as 40%. This may be due to shared risk factors such as prematurity, low levels of anti-GBS antibodies in circulation, primary immunodeficiency disorders, or other genetic predispositions, as well as concordant exposures (53–56). During their infancy, twins are also likely to have similar exposures to exogenous sources of GBS, namely, maternal, hospital, and household contacts (53–56). Prematurity (<37 weeks) is a risk factor for LOD and mortality (7, 33, 57–59). According to a recent Active Bacterial Core surveillance, performed in collaboration with the Centers for Disease Control and Prevention (CDC), from 2006 to 2015, prematurity was associated with 42% of all GBS LOD cases and a case fatality rate more than double that observed in full-term babies (3.4% versus 7.8%) (7). Similarly, recent data from France showed that the frequency of preterm infants was higher in babies with late-onset than early-onset meningitis (57).

Prevention of Neonatal GBS Disease—Obstetric Guidance

In 1996, the CDC, in collaboration with several professional societies, first published guidelines recommending either screening for GBS in pregnant women or using a risk factor-based approach to the use of intrapartum antibiotic prophylaxis for the prevention of perinatal GBS infection (60). In 2002 and later in 2010, the guidelines were revised and updated (27, 61). In 2018, the responsibility for curating these guidelines was transferred from the CDC to the American College of Obstetricians and Gynecologists and the American Academy of Pediatrics. The current guidelines followed by the United States, and many other countries, recommend performing universal GBS culture-based screening of pregnant women between 36 and 37 weeks of gestation to detect vaginal and/or rectal carriage of GBS (62, 63). Although many countries follow these recommendations, it should be noted that antenatal screening is still controversial and varies between nations. For example, according to the European consensus conference on GBS, antenatal screening is not recommended, as the conference endorsed an intrapartum non-culture-based diagnostic tool, as a rapid and universal molecular test (64). Nevertheless, to our knowledge, at the time of this writing, the European recommendations were not yet revised, dating from 2014 (64), and there is no universal intrapartum screening program being carried out in Europe. Indeed, some European nations, including Portugal, still recommend culture-based antenatal GBS screening, in accordance with the recommendations from the United States, as well as the National Institute for Health and Care Excellence (62, 63, 65).

Laboratory diagnosis.

Screening for GBS carriage in pregnant women requires high-standard practices, from specimen collection to GBS identification. In brief, a single swab should be used, aiming to increase the culture yield and the likelihood of GBS recovery (66). Specimens should be preferably collected with the same swab, first from the lower vagina followed by the rectum, placed into a nonnutritive liquid-based transport medium (e.g., Amies transport medium), and transported to the laboratory within 24 h (66). Swabs must then be cultured into selective enrichment broth media to reduce the outgrowth of enteric organisms, increasing the sensitivity of downstream screening methods. Acceptable enrichment broths include Todd-Hewitt broth, supplemented with selective agents (e.g., gentamicin, nalidixic acid, and colistin), and some selective and differential media, such as carrot broth and liquid biphasic Granada medium, that are based on the unique ability of hemolytic GBS strains to produce a carotenoid pigment, thereby allowing the GBS enrichment and detection in a single step (62, 67–69). After enrichment, culture-based testing remains the standard, as this method has been shown to increase GBS identification in cultures (62). Direct specimen plating on agar medium, without prior enrichment, is unacceptable (62). When chromogenic tube media are used as the first choice, negative cultures (i.e., weakly hemolytic or nonhemolytic GBS strains) should be subcultured in agar plates (62). Nevertheless, biochemical profiling, direct antigen detection testing, latex agglutination testing, matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) mass spectrometry, DNA probes, and nucleic acid amplification tests are also acceptable choices for the identification of GBS from enrichment broth (66).

Molecular tests, such as real-time PCR, have the potential to be used intrapartum, even on the amniotic fluid, in the case of rupture of membranes, while conventional cultures require more time (24 to 72 h), which prevent them from being used as a point-of-care test. Although such tests could have the potential to positively impact the evaluation of GBS colonization in women at the time of delivery and reduce unnecessary antimicrobial treatments, more evidence-based studies are needed, as conflicting data regarding their efficacy exist. Some reports show that the performances of intrapartum molecular tests for the screening of maternal colonization are higher than those of antenatal culture screening, with higher sensitivity and specificity and higher positive predictive value (70–72). Others, however, suggest that performing nucleic acid amplification tests alone directly on rectovaginal swabs, without incubation in enrichment broth, can give rise to false-negative results, with result accuracy also depending on trained laboratory personnel, and thus may not adequately replace routine prenatal screening (62, 73–75). Therefore, current laboratory diagnosis guidelines recommend always incubating samples for GBS testing in selective enrichment broth before agar plating or molecular testing, as it enhances the sensitivity of detection (63, 66).

Nevertheless, the adoption of commercially available real-time PCR nucleic acid amplification tests on the enriched selective broth has been proposed as the new gold standard for GBS screening in pregnant women, as it could enhance sensitivity for GBS detection compared to conventional culture methods and has a shorter time to result (76–78). The greatest limitation is that it does not provide antimicrobial susceptibility testing, which is particularly concerning in pregnant women reporting an allergy to β-lactams. Also, molecular testing may not be a cost-effective strategy compared to conventional culture techniques, though the costs could be counterbalanced by the reduction of treatment costs in case of GBS invasive disease (79). Finally, some caution must be taken with molecular diagnostic methods, as the loss of target nucleic acid sequences, either through deletion or mutation, can lead to a failure in the detection of organisms. A recent study characterizing 31 GBS isolates from several locations in the United States and Ireland found four types of chromosomal deletions in the CAMP factor (cfb) gene (80). Nevertheless, prospective surveillance showed that these diagnostic-escape mutants represented only less than 1% of the total GBS isolates in three of the four geographic locations analyzed and 7% of GBS isolates in a fourth location, giving rise to a low rate of false-negative results, equivalent to those found when culture-based screenings with selective and differential media, without enrichment, are used (80, 81). Moreover, these mutations were not found in a Slovenian survey (82). However, if selective pressures were to favor such mutations, it could be worrying considering that of the more than 10 different nucleic acid amplification tests cleared by the U.S. Food and Drug Administration, eight of them amplify the cfb gene (83). More worldwide surveys are needed to truly understand the impact of chromosomal deletions in molecular-detection evasion.

GBS prevention practices.

All GBS-positive women should receive appropriate antibiotic prophylaxis during labor, to reduce GBS transmission and consequent invasive neonatal disease (62, 84). GBS antibiotic prophylaxis is not indicated in the case of a cesarian birth prior to labor and with intact amniotic membranes, even in the presence of a positive vaginal-rectal GBS culture (62, 84). Screening can be omitted in cases of history of previous pregnancy with a GBS-infected newborn or in women with GBS-positive urine cultures during pregnancy, as intrapartum antibiotic prophylaxis is automatically recommended (62). Whenever screening results are not available at the onset of labor, intrapartum GBS prophylaxis is indicated for women who meet any of the following criteria: (i) prolonged membrane rupture (>18 h) at term, (ii) preterm gestation (<37 weeks), (iii) intrapartum fever (temperature of ≥38°C), (iv) previous pregnancy with GBS-positive screening, and (v) intrapartum nucleic acid amplification testing positive for GBS (62). In preterm gestation, antibiotic prophylaxis is recommended at the onset of labor only when maternal GBS screening test results are unknown (62).

Penicillin G remains the preferred drug for intrapartum prophylaxis, even though ampicillin can be an acceptable alternative. In case of allergy, women should receive cefazolin (first-generation cephalosporin). Intrapartum antibiotic prophylaxis is most effective if administered 4 h before delivery, aiming to reach levels above MIC needed for GBS in the fetal circulation, cord blood, and amniotic fluid (62). Although this strategy, together with the improvement in neonatal intensive care, has remarkably decreased the incidence and case fatality of EOD, it is ineffective in preventing LOD, with recent studies showing that cases are increasing (7–10, 57, 85). Moreover, while intrapartum antibiotic prophylaxis has been implemented in most high-income countries since the late 1990s, this public health intervention remains logistically challenging and not standard of care in most resource-limited countries, where most births are at home (4, 41, 85). Intrapartum antibiotic prophylaxis has some inherent limitations and remains controversial. The UK National Health Service does not endorse routine universal testing for GBS carriage and GBS antibiotic prophylaxis (86), and in the Netherlands, similar risk-based antibiotic chemoprophylaxis did not prove beneficial (8).

Owing to the transient nature of GBS colonization, which can be intermittent or transitory, the results at the time of screening, even under perfect laboratory conditions, cannot identify all infants at risk. A recent report revealed that nearly half of EOD cases occurred in infants born to women who were not eligible for intrapartum antibiotic prophylaxis (7). Also, a prevention strategy that relies on a substantial amount of perinatal antibiotic use raises concerns about multidrug resistance in GBS or other neonatal pathogens. Indeed, the emergence of multidrug-resistant GBS and E. coli isolates has already been reported (10, 27, 48, 87). In addition, the growing appreciation that pioneer microbiota composition within the gastrointestinal tract are implicated in the development and function of multiple organ systems (88) raises concerns about the potential effects of exposing mothers and infants to antibiotics (89–92). Moreover, as mentioned above, antibiotic prophylaxis while preventing EOD had little impact on LOD incidence or LOD-associated deaths, suggesting that the pathophysiology of GBS LOD is not entirely understood. Also, this further highlights the importance of informing and counseling parents about the risks of LOD on discharge. Doctors should also be aware of this risk, and infants must be immediately assessed should there be any suspicion of LOD.

Maternal vaccination represents the most attractive and promising strategy in all settings to protect both the fetus and newborn from GBS disease morbidity and mortality. The first known evidence that maternal GBS vaccination could decrease the burden of GBS invasive disease in the infant dates back to the 1970s, when an inverse correlation between the levels of maternal antibodies against type III capsular polysaccharides and newborns’ susceptibility to GBS disease was observed (93). Ideally, maternal vaccines would lead to the production of specific high-titer immunoglobulin G (IgG) that is actively transferred through the placenta via the neonatal Fc receptor (FcRn) in the placental syncytiotrophoblast, providing passive immunity for the fetus and newborn and thus preventing both EOD and LOD (94). It has been estimated that a maternal GBS vaccine with 80% efficacy and 90% coverage could prevent 108,000 fetal and infant deaths (42). A potential concern of this strategy is that the neonatal immune responses after primary vaccination in early infancy can be inhibited by maternal IgG antibodies, a phenomenon termed “immunological blunting,” already observed in infants born to women who received a maternal vaccine against pertussis (95, 96). To date, there is no vaccine commercially available to prevent GBS infections, but several vaccine candidates against various GBS capsular types are under development (97, 98).

Clinical Presentation and Laboratory Diagnosis

Early diagnosis and rapid initiation of targeted treatment in response to suspected meningitis play a decisive role in preventing potentially poor outcomes. However, diagnosing meningitis in newborns is difficult, time-consuming, and invasive. Clinical presentation of neonates with meningitis is often subtle and nonspecific, overlaps presentation of neonatal sepsis without meningitis (99), and is absent in more than 30% of cases (1, 100, 101). It includes fever, lethargy, listlessness, high-pitched crying, irritability, feeding difficulties and refusal, vomiting, diarrhea, respiratory distress, bulging fontanelle, and seizures (1, 11, 20, 22, 102). Nuchal rigidity occurs in less than 25% of affected neonates and is often a late finding of the illness, being associated with poor prognosis (1).

Diagnostic tests.

There are several tests that can be used for the diagnosis of GBS meningitis, as discussed below, some of which might also be useful for epidemiological studies where characterization of GBS isolates is important (81, 83).

(i) Bacterial culture.

In the context of invasive disease, detecting GBS generally requires the microbiological culture of the organism from a sterile body site, such as the blood, cerebrospinal fluid (CSF), and pleural fluid. Indeed, blood culture is the most used diagnostic test for detecting the presence of bacteria in the blood. However, when meningitis evaluation is performed only in neonates with confirmed bacteremia, it can lead to missed diagnosis, as up to 38% of babies with meningitis have negative blood cultures (1). Although the reasons for this remain elusive, it may be explained by the timing of culture. In the routine clinical setting, therapy with antibiotics is started before the definitive identification of the organism. Thus, it is possible that, due to previous or ongoing treatment with antibiotics (either transplacentally or postpartum), levels of recovered bacteria from the blood are considerably reduced at the time of sampling and thus below the detection level, resulting in a false-negative result.

There are, however, other possible explanations. Evidence shows that the volume of blood cultured may also interfere with results, as it determines the technique’s sensitivity (103, 104). It is possible that the level of bacteremia in neonates with meningitis is low, contradicting the idea that blood infections in pediatric patients are associated with high bacterial levels in the circulation (105–108). Thus, successful isolation of bacteria depends on the volume of blood cultured. Techniques for drawing blood cultures in neonates are difficult and invasive, particularly in the critically ill, and require trained and practiced health care professionals (109). Also, there is no consensus in the literature regarding the adequate blood culture volumes sampled from pediatric patients (110, 111). According to recommendations by pediatric blood culture bottle manufacturers, the correct volume required for a blood culture bottle ranges between 1 and 4 mL, a volume that might not be able to be obtained from newborns (106, 112–114), thereby significantly decreasing the sensitivity of the test. Bacteremia can also be transient and intermittent, leading to false-negative results. Repeated blood culture sets, as performed for adults, might not be an option for babies due to their small circulating blood volumes and increased risk of anemia. Last, another interesting explanation for the lack of detectable bacteria in the blood is that GBS could reach the meninges by using the lymphatic network. In contrast to intracellular bacteria and viruses, extracellular bacterial dissemination through lymphatics has not received much attention. However, lymphatic tropism has been described for Gram-positive bacteria such as group A Streptococcus and S. aureus (115–117).

The gold standard for confirming bacterial meningitis is a positive CSF culture (1, 118). However, several concerns remain: (i) determining when to perform a lumbar puncture to obtain and analyze CSF is challenging; (ii) the procedure is invasive and can cause additional distress in newborns; (iii) CSF parameters (leukocyte count, glucose, and total protein) are frequently within the normal range; (iv) CSF culture requires medically trained personnel, which are largely absent in middle- and low-income settings. Together, these present clinicians with a dilemma, as treating neonates with antibiotics based only on subtle signs results in overtreatment, whereas delaying treatment until symptoms appear entails an obvious risk of preventable mortality (119). Thus, lumbar puncture is frequently delayed or omitted in clinical practice (20, 119). Also, antibiotic treatment is frequently initiated before lumbar puncture, affecting CSF culture results (118). CSF Gram staining is an inexpensive and well-validated method that can aid in the identification of the causative microorganism. Particularly for neonatal patients, it has been shown that it accurately identifies GBS, with a sensitivity ranging between 80 and 90% (120, 121). However, this can be much lower for other agents, and negative staining does not exclude the disease (120, 121). Also, the detection limit requires a concentrated CSF sample (10⁴ CFU/mL), and its yield is also lower in the cases with antibiotic pretreatment (120, 121). Latex agglutination tests are fast, usually taking less than 15 min, and recommended in cases with negative CSF cultures. However, their use is controversial, and alone, they may have little additional value in diagnosing meningitis (115, 121).

(ii) Blood and CSF cell count, glucose, and protein.

Other approaches have been proposed, but their accuracy in diagnosing neonatal meningitis has been disappointing. In the presence of neonatal bacterial meningitis, it is possible, but not always the case, to find abnormal peripheral blood cell counts (high or low white blood cell counts, elevated ratio of immature to total neutrophils, and low platelet count) or abnormal CSF parameters (elevated CSF white blood cell counts, polymorphonuclear pleocytosis, elevated CSF proteins, and decreased CSF glucose) (101, 122–124). Nevertheless, for neonates with bacterial meningitis caused by GBS, CSF white blood cell counts are often inconclusive, and a normal CSF study does not exclude meningitis (121, 122, 125). C-reactive protein values and immature-to-total neutrophil ratio (I/T) as screening tests for meningitis in culture-negative EOD sepsis can have a negative predictive value of 85% to 90% (126).

(iii) Molecular testing.

As for prenatal screening, bacterial molecular detection testing is becoming more available to diagnose neonatal meningitis, replacing traditional culture techniques. Although it does not provide antimicrobial susceptibility testing, it has the advantage of faster and reliable detection of GBS invasive disease, thus allowing rapid treatment initiation. Importantly, and as already mentioned, the CSF and blood samples are often collected after initiation of antibiotic treatment, enormously compromising bacterial growth. Nucleic acid amplification tests can overcome this limitation, as they are not dependent on viable organisms and can detect small amounts of potential pathogens’ DNA. Indeed, a study showed that nearly 50% (14/31) of infants with PCR-positive and culture-negative results were born to mothers who had received antibiotics (127).

The number of studies assessing molecular testing specifically for GBS meningitis is still limited. Nevertheless, reports among the literature have been increasing and strengthening its potential as a diagnostic tool. According to a large multicenter retrospective study comparing GBS PCR testing with conventional bacterial culture, in both blood and CSF samples, from infants aged 7 days to 3 months, the sensitivity of GBS PCR in the CSF using bacterial culture as a reference was 100%, and the specificity was 97%. In contrast, lower sensitivity was found in blood samples (65%) (128). Importantly, this study found that of the nine patients with negative CSF culture and positive PCR result, six had clinical criteria for GBS diagnosis (evaluated by positive blood culture and high CSF white cell counts) (128). Similarly, a study performed in infants with suspected meningitis, with ages ranging between 1 day and 1 year (61.1% less than 28 days old), found a 63.9% positivity rate among those with negative CSF cultures (125). Thus, based only on CSF cultures, clinicians cannot exclude meningitis with confidence.

It is also important to remember that typically, PCR will only detect an already suspected infectious agent, since that is the one that will be targeted in the primer mix. Thus, to overcome this limitation, a PCR multiplex molecular diagnostic assay has been developed, enabling simultaneous testing of multiple infectious agents, including viruses, bacteria, and yeasts. A recent study performed in an Italian hospital over 32 months found that multiplex PCR testing of CSF samples could reliably facilitate meningitis diagnosis, with a sensitivity of 90.9% and a specificity of 100% (129). However, among the 44 patients with bacterial meningitis included in this study, only four were neonates, weakening any conclusions that could be made regarding this age group (129). In a large retrospective study also assessing the clinical use of multiplex PCR testing, 238 of 705 CSF samples were from pediatric patients, and the sensitivity for GBS, using bacterial culture as a reference, was 100% (130). Nevertheless, interpreting these data is challenging, as the analysis did not discriminate between age groups, with patients being between 3 days and 95 years old and the pediatric population being defined as patients younger than 18 years (130). Another study focusing particularly on infants younger than 3 months found that from 12 CSF samples positive for GBS or E. coli, only 5 yielded positive bacterial cultures, supporting the use of multiplex testing aimed at reducing missed diagnoses (131).

However, some investigations raise concerns, as they report false-positive and false-negative results, which could compromise the application of multiplex PCR technology as the standard for diagnosing bacterial meningitis (132–134). According to a recent systematic review and meta-analysis, GBS presented one of the highest proportions of false-positive results (15.4%) (135). However, this analysis was based on only four studies, of which only one was exclusively from pediatric samples; in that study, the patients’ ages were not specified, and thus, it could include samples from birth until the age of 18 years (135, 136). Of the remaining three studies mentioned, one included 291 samples but did not specify whether the samples were from adults, pediatric patients, or both, and most of the CSF samples that were positive for bacteria had been frozen for more than 30 years, which could compromise their quality (135, 137). The other two studies included both adult and pediatric samples, but in one, the pediatric period was not defined, and the analysis did not discriminate by patient age. In the other, despite a large cohort of 1,560 samples, including 299 from infants younger than 2 months, GBS was found only in the CSF of a patient older than 65 years, and most pediatric infectious agents were viruses (132, 135, 138). Thus, despite providing a rapid and straightforward assessment of a broad range of pathogens, multiplex PCR testing still lacks comprehensive clinical validation for neonatal bacterial meningitis, and studies in large cohorts are needed.

Metagenomic next-generation sequencing has significantly increased in recent years and is another promising tool to rapidly and accurately diagnose meningitis, even in cases where CSF cultures are negative and CSF findings are normal (139–141). Compared with conventional PCR-based methods, next-generation sequencing has the key advantage of being completely unbiased, as it is not conditioned by a prior selection of the suspected pathogens and can identify potential infectious agents in a single test, which is particularly important in limited samples like the ones obtained from neonates (141). However, this method is still expensive, time-consuming, and challenging to implement in laboratories with low capacity and minimal technical expertise, such as those found in low-income settings.

(iv) Neuroimaging.

Neuroimaging techniques, such as cranial ultrasound, computed tomography, and magnetic resonance imaging (MRI), have emerged as helpful tools alongside laboratory testing and clinical evaluation. Although these techniques are more valuable for evaluating complications than diagnosing bacterial meningitis, since infected babies may present with no structural changes, thus showing no abnormalities in radiological findings (4), they can be helpful in preliminary diagnosis and aid in clinical management decisions. Cranial ultrasound is the most useful and practical neuroimaging technique, as it can be performed at the bedside without sedation and with little disturbance to the neonate, while also having no risk of radiation exposure. Altered ultrasound was observed in approximately 65% of infants with bacterial meningitis and was as high as 100% in patients with clinical evidence of severe neurological complications (142, 143). Several abnormalities can be found via diagnostic imaging techniques (142, 144). A retrospective study with infants with culture-proven bacterial meningitis found altered MRI in 81% of 75 infants studied, with subdural empyema, brain parenchymal ischemia/infarction, and leptomeninges enhancement being the most frequent findings (145). Despite the small sample size, the results from this study found that almost half the infants presenting abnormal MRI had at least one change in management, such as prolongation of antibiotic treatment (145). Also, the development of cerebral infarcts is suggestive of streptococcal meningitis, as 40 to 59% of neonates and infants infected with a streptococcal pathogen present at least one infarct area (146, 147). A computerized tomography scan may also be helpful before performing a lumbar puncture in cases of suspected increased intracranial pressure, such as focal findings in the neurological exam (148). Finally, it has been proposed that imaging following therapy can help monitor the therapeutic process and predict complications, as discussed further below (“Neuroimaging and other predictors of neurological abnormalities”).

Management of GBS Meningitis

When there is suspicion of meningitis, empirical antimicrobial therapy should be initiated as soon as possible, even before the diagnosis and underlying etiologic cause can be confirmed. Delaying the initiation of treatment is strongly associated with high mortality and poor outcomes (20). GBS is likely to be sensitive to β-lactam antibiotics, such as penicillin G, ampicillin, and first- and second-generation cephalosporins, and also to vancomycin (a glycopeptide antibiotic) (30, 35, 149). In the case of suspected invasive GBS disease, empirical antibiotic administration should include the broad-spectrum drug ampicillin (≤7 days, 100 mg/kg of body weight intravenously [i.v.] every 8 h; >7 days, 75 mg/kg i.v. every 6 h) (35, 150). Immediately upon diagnosis of GBS meningitis and evident clinical improvement, penicillin G (≤7 days, 150,000 U/kg i.v. every 8 h; >7 days, 125,000 U/kg i.v. every 6 h) becomes the preferred drug because it is the most narrow-spectrum and active agent in vitro (30, 35, 150). Nevertheless, ampicillin remains an acceptable choice (30). Synergic combined therapy with ampicillin and gentamicin should be used during the initial days of treatment and continued until sterile blood and/or CSF cultures are obtained (151). Uncomplicated GBS meningitis is treated with a 14-day course, whereas complicated cases (subdural empyema, ventriculitis, intracranial abscess, and suppurative venous sinus thrombosis) should be treated for at least 21 days, although longer courses may be required (30, 151).

Supportive care for neonates with bacterial meningitis, who are frequently ill, should be delivered in a neonatal intensive care unit to ensure adequate oxygenation and perfusion, maintenance of normal fluid and electrolyte balance, nutritional support, and seizure control. The response to therapy and the potential development of complications are monitored clinically through serial neurologic examinations (with particular attention to early signs of increased intracranial pressure) and neuroimaging (152).

Steroids, particularly dexamethasone (an anti-inflammatory corticosteroid), were proposed as a potential adjunctive treatment strategy to reduce inflammation-related neuronal death and improve poor outcomes associated with neonatal meningitis. However, few studies exist on the effects of dexamethasone for the treatment of bacterial meningitis in the neonatal population. Moreover, they do not focus specifically on GBS meningitis, as they usually include patients with bacterial meningitis of all etiologies. The available data are conflicting and difficult to interpret, yet they suggest that therapy with dexamethasone is not beneficial, as it does not improve survival or severe complication rates, and thus not recommended or used in clinical practice (148, 153). A randomized clinical trial with 52 full-term neonates (3 with GBS meningitis) studied the effects of dexamethasone in neonatal bacterial meningitis. One group received dexamethasone (0.15 mg/kg body weight every 6 h i.v., for 4 days) and antibiotic therapy, and the other received only antibiotics. The outcomes in both groups were identical, with similar clinical response, mortality, and complications (153). Nevertheless, some studies showed that dexamethasone therapy could be beneficial. A randomized trial with 80 neonates studied the role of dexamethasone in the treatment of bacterial meningitis. The group treated with dexamethasone (0.15 mg/kg body weight every 6 h for 48 h) presented decreased mortality and inflammatory response, with a reduction of the proinflammatory cytokines tumor necrosis factor alpha (TNF-α) and interleukin 1β (IL-1β) levels in the CSF. It is, however, important to note that only one of these patients had GBS meningitis (154). A meta-analysis including 10 studies showed an apparent benefit in reducing hearing loss among survivors but no effect on mortality or severe neurological complications (155). Likewise, other adjunctive therapies, such as glycerol, immunoglobulins, and granulocyte-macrophage colony-stimulating factor, are not recommended in neonatal meningitis, as they have not been shown to conclusively improve outcomes (152).

Importantly, and as already mentioned for GBS antibiotic prophylaxis, long-term antibiotic therapy is associated with the emergence of resistant bacterial strains and with overlapping fungal infections and might alter the composition of the intestinal microbiota of the neonates and/or lead to an incomplete microbiota recovery (156–158). Although further research is needed to uncover the long-term clinical consequences, it has been suggested that disruptions of the gut microbiota in early life may impact neonatal development and contribute to increased risk of behavioral disorders in later life (159–162). Also, antibiotic treatment needs vascular access, which despite being mandatory is an invasive procedure that can inflict harm (22). The development of new and less invasive approaches specifically targeting pathogenic bacteria while also being neuroprotective is imperative.

Complications and Outcomes.

Long-term sequelae.

Neuroinfections may lead to permanent mental alterations, long after the elimination of the infectious agent. Infants who survive GBS meningitis are at high risk of suffering from complex neurological or neuropsychiatric sequelae later in life. It is estimated that nearly half of all survivors will experience neurodevelopmental impairments (4, 50, 102, 163–166). However, the actual burden of GBS-induced neurodevelopmental impairments is likely underestimated due to challenges in detection, different screening methods, and the choice of neurological functions to examine. Contemporary surveys are scarce and encompass medium- or high-income settings, and thus, information and estimates lack data from developing countries, where the burden of bacterial meningitis is likely greatest. Notwithstanding these limitations, neurodevelopmental impairments can be broadly categorized as intellectual impairment and/or motor, hearing, or vision impairment, and they can vary in severity between mild, moderate, and severe (4).

In a prospective cohort study in England and Wales of a follow-up investigation at 5 years of age of those who had neonatal GBS meningitis, 49% had some form of neurological impairment (165). The long-term consequences can be drastic, with children having a permanent severe or moderately severe disability (13.3% and 17.3%, respectively), including learning difficulties, neuromotor disabilities, seizure disorders, hearing problems, ocular or visual disorders, speech or language problems, and behavior complications (165). A longitudinal study in children 9 to 10 years of age who had experienced neonatal meningitis reported that 63.3% of survivors had a normal outcome. However, 14.3% had a severe outcome (defined as “severe disability or a significant functional impairment: cerebral palsy, significant learning problems, global delay or special education”), 8.1% had a moderate outcome (defined as “moderate to mild disability or an abnormality that impairs function without significant intellectual or developmental impairment: mild cerebral palsy, mild learning problems, and sensorineural hearing loss”), and 14.3% had a mild outcome (defined as “an impairment without disability: neuromotor signs without overt functional loss, isolated hydrocephalus, isolated epilepsy, and borderline learning problems, isolated occurrence of all mABC component scores being below the 5th centile”) (166). Others found that 25% of children who survived GBS meningitis had moderate to mild impairment (grade retention at school, ventriculoperitoneal shunt, delayed fine motor skills), and 19% had severe impairment (cerebral palsy, cortical vision impairment, seizure disorder, profound sensorineural hearing loss) (164). Importantly, in this study, the authors found that in the cases of moderate to mild sequelae, the answers provided by the parents on developmental screening questionnaires were not sufficient to identify children with developmental delays (164). This further supports the idea that sequelae are underestimated and highlights the need for parental counseling at hospital discharge, as well as follow-up evaluations. A recent systematic review and meta-analysis showed that 32% of GBS meningitis survivors who had a follow-up beyond 18 months presented neurodevelopmental impairments (4), of which almost one-fifth were moderate or severe (4).

Neuroimaging and other predictors of neurological abnormalities.

Neuroimaging is not widely used to predict neurodevelopment outcomes in surviving infants, and reported studies are limited. However, MRI shows cerebrovascular involvement, which has been associated with poor neurodevelopmental outcomes in 2 years (4). Different studies have shown ischemic infarction (neonatal stroke), with two observable patterns of injury (deep perforator arterial infarction and more superficial cortical injury) (4, 146, 167, 168). Severe global cerebral vasculopathy, encephalomalacia, and transverse myelitis were also described (169, 170). Perinatal stroke was reported as a complication of neonatal GBS meningitis and sepsis. MRI showed evidence for acute or subacute ischemic infarction, with some patients having cerebral sinovenous thrombosis (171). The stroke location was predominantly cortical, with bilateral involvement and, in some cases, basal ganglia involvement as well. The follow-ups (6 to 48 months after) revealed normal developmental outcomes in 8 of 14 patients and focal neurological deficits and/or developmental delay in 6 patients (171).

GBS disease can also be associated with neonatal encephalopathy (172). Neonatal encephalopathy describes a clinical constellation of neurological dysfunctions in the term infant that can be associated with child mortality and long-term impairment (173, 174). The percentage of term infants with neonatal encephalopathy and coexisting GBS infection is 0.58%, which is more than 10 times higher than that seen among term infants without neonatal encephalopathy. GBS-associated neonatal encephalopathy is associated with two times more mortality risk than neonatal encephalopathy alone (172, 175). Brain imaging, using both cranial ultrasound and MRI, can be used to predict the neurodevelopmental outcome of GBS meningitis. Abnormal cranial ultrasound, extensive bilateral deep gray lesions, and white matter lesions correlated with abnormal motor outcome, whereas extensive bilateral deep gray matter lesions correlated with abnormal cognitive outcome (176). Other predictors of GBS related mortality or neurological abnormalities after discharge have been identified, namely, premature birth, apnea, coma, shock, seizures, lethargy at presentation, bulging fontanel, leukopenia, acidosis, and CSF protein concentration exceeding 300 mg/dL or a glucose value lower than 20 mg/dL (50, 57, 102, 164). Nevertheless, with our current knowledge of neonatal meningitis, it is impossible to accurately predict which infants will die and which infants will survive with or without long-lasting sequelae.

PATHOGENESIS OF BACTERIAL MENINGITIS

Anatomical Considerations

The brain and spinal cord are enveloped by a protective triple-layered membrane, known as the meninges (Fig. 1) (177, 178). The meninges comprise the pachymeninx or the dura mater (“dura” from the Latin meaning “hard”) and the leptomeninges (“lepto” from the Greek meaning “thin”), which consist of the arachnoid mater and pia mater (179). The meningeal membrane closest to the skull is the dura mater, which is a dense, thick, collagenous membrane that contains lymphatic vessels, immune cells, and fenestrated blood vessels (178, 180, 181). The middle layer of the meninges is the arachnoid mater, named after the Greek word “arachne” meaning “spider,” because of the delicate spider web-like appearance. The arachnoid mater is composed of a multilayer of cells connected by tight junctions, regulates the transport of molecules, and is structurally attached to the pia mater (182, 183). The CSF flows between the arachnoid and the pia mater in the subarachnoid meningeal space. The subarachnoid space is occupied by the trabeculae, which are composed of strands of delicate collagen-enriched tissue coated by leptomeningeal cells that connects the arachnoid with the pia mater, forming intercommunicating channels that accommodate the CSF (184).

FIG 1.

Overview of the brain meninges and cerebrovascular barriers. Schematic representation of a coronal section of the brain. The schematic at top right shows the meninges. The meninges comprise the dura and the leptomeninges, which consist of the arachnoid mater and pia mater. The cerebrospinal fluid (CSF) is localized between the arachnoid and the pia mater in the subarachnoid meningeal space. Vasculature within the subarachnoid space is contained within subarachnoid trabeculae, a delicate tissue that connects the arachnoid mater with the pia mater. The pia mater closely follows the contours of the brain and enters its sulci. The schematic at bottom left shows a cross-section of the blood-brain barrier (BBB) in the central nervous system parenchyma. The BBB is part of the neurovascular unit, composed of the BBB endothelium, neurons, interneurons, pericytes, glial cells (e.g., astrocytes and microglia), and the basal membrane. The Virchow-Robin perivascular space comprises the perivascular space surrounding brain-penetrating arteries between endothelial cells and astrocyte end feet. The schematic at bottom right shows the blood-CSF in the choroid plexuses. The CSF is mainly produced by the epithelial cells of the choroid plexus and localized in the brain ventricles, and it flows through the ventricles into the subarachnoid meningeal space. The capillaries of the choroid plexus are fenestrated and lack tight junctions.

The pia (“gentle” in Latin) mater is a thin membrane, often only one cell thick, that closely follows the contours of the brain and enters its sulci, adhering firmly to the surface of the brain parenchyma and spinal cord (178, 179, 184). The pia mater is highly vascularized and is semipermeable to the CSF (179). The glia limitans is composed of a thin barrier of astrocyte end feet surrounding the brain and spinal cord that acts as a physical barrier and insulates the brain parenchyma (185). The CSF is mainly produced by the epithelial cells of the choroid plexus, localized in the brain ventricles, and flows through the cerebral ventricles toward the subarachnoid meningeal space (186). Some of the CSF accesses the brain parenchyma along with the perivascular spaces surrounding brain-penetrating arteries (Virchow-Robin spaces). According to the traditional view, most of the CSF drains into the bloodstream through arachnoid protrusions directly into the venous sinuses of the brain and through perineural routes (including the cribriform plate) (recently reviewed elsewhere [187]). Recent studies have revealed that the CSF could be additionally cleared via lymphatic vessels present in the dura mater (the meningeal lymphatics) and drained to cervical lymph nodes in mouse models (188, 189). These structures are similar to those found in humans (190). In rodents, the meningeal lymphatic network in the dura mater develops postnatally (191, 192), but whether this is the case in human neonates remains to be confirmed.

Central Nervous System Barriers and Bacterial Invasion

Bacterial access to the CNS involves a multistep process that progresses from mucosal colonization to blood dissemination and, last, circumvention of CNS barriers, involving interactions between bacterial structures with host receptors (Fig. 2). Not surprisingly, the CNS is extraordinarily protected. The complex blood-CNS barriers include the blood-brain barrier (BBB) and the blood-CSF barrier (193, 194). The meninges connect these structures and provide an additional interface for peripheral interaction with the CNS (195, 196).

FIG 2.

Generalized GBS virulence factors involved in central nervous system (CNS) invasion. The schematic bacterial illustration depicts the major virulence factors mediating GBS meningitis regardless of the bacterial strain, and thus, they may not all be simultaneously found in a single strain. Several pathways of blood-brain barrier (BBB) penetration have been described, including transcellular, paracellular, and leukocyte-facilitated routes. Upon dissemination, GBS can persist, induce cell damage, bind, and invade the BBB. During bacteremia, the CNS invasion probably occurs via the choroid plexus, the brain parenchyma microvessels, or the dural venous sinuses. The mechanisms that ultimately lead to neuronal injury are not entirely understood. The hypothetical mechanisms involve the release of bacterial products that can cause direct toxicity to neurons, the release of proinflammatory and/or toxic compounds by host cells, and/or direct bacterial stimulation of glial cells. The resulting inflammation can also cause vasculitis, ventriculitis, and encephalitis and can also lead to edema and increased intracranial pressure. Ultimately, these processes contribute to additional neurotoxicity. β-h/c, β-hemolysin/cytolysin; BspC, GBS surface protein C; Cas9, CRISPR-associated protein 9; CovS/R and CiaR/H, two-component regulatory systems; FbsA, FbsB, and FbsC, fibrinogen-binding proteins A, B, and C; HvgA, hypervirulent GBS adhesion; iagA, invasion-associated gene A; Lmb, laminin-binding protein; LTA, lipoteichoic acid; SfbA, streptococcal fibronectin-binding protein A; Srr1 and Srr2, serine-rich repeat proteins 1 and 2.

The BBB is a complex, specialized system formed by a thick continuous glycocalyx, a tight monolayer of nonfenestrated specialized brain microvascular endothelial cells (BMEC) of the capillary wall, two basement membranes (vascular basement membrane and glia limitans), and astrocyte end feet that ensheath the vascular tube (Fig. 1) (197, 198). The BBB is responsible for ensuring homeostasis within the CNS, tightly regulating the flow of nutrients and molecules for normal brain function. To enable the BBB barrier properties, maintain cerebral homeostatic hemodynamic response, and prevent bacteria from gaining access to the brain parenchyma, continuous cross talk between different cell types occurs at a complex functional and anatomical structure called the neurovascular unit. This unit is composed of the BBB endothelium, neurons, interneurons, pericytes, glial cells (e.g., astrocytes and microglia), and the basal membrane (19, 197).

In contrast to the BBB, the blood-CSF barrier acts as a controlled gate, allowing immunosurveillance under normal conditions (reviewed elsewhere [193]). The main compartments of the blood-CSF barrier are the choroid plexus and the leptomeningeal vasculature. The blood vessels within the choroid plexus are fenestrated and lack tight junctions, facilitating the passive transport of blood cells and different molecules into the stroma (Fig. 1) (199–201). The choroid plexus is composed of an additional layer of ependymal cells sealed by tight junctions that cover the choroidal stroma, providing a barrier between the fenestrated blood vessels and the CSF. The endothelial cells of pial and arachnoid microvessels also have transendothelial electrical resistance (201). The arachnoid mater forms a barrier at the interface between the CSF and the fenestrated vessels of the dura mater, created by the tight junctions between the cells of the continuous outer layer (193).

Blood-borne bacteria can invade the meninges through several routes, such as the choroid plexus, brain parenchyma microvessels (the capillaries of the CNS barriers), or the arachnoid villi of the venous sinuses (19, 202). Current evidence supports the idea that bacteria invade the subarachnoid space through the choroid plexus and/or the capillaries of the CNS barriers (19). For bacteria to cross the choroid plexus, they first have to interact with the capillary endothelium and then cross the monolayer of ependymal cells. The crossing could also occur following bacterial interaction with the capillaries of the leptomeninges and/or the capillaries of the brain parenchyma. When the cerebral vasculature is the leading site of entry into the CNS, the main obstacle is the endothelium layer. In vitro BBB models suggest that upon adhesion to brain endothelium, bacteria cross by paracellular, transcellular, or leukocyte-facilitated routes (19). Models of GBS invasion suggest the transcellular (203) and paracellular routes (204). Although models with cell lines can be a rather reductionist approach and not replicate the in vivo scenario, postmortem examination of a fatal human case of neonatal GBS meningitis supports both choroid plexus and endothelial invasion, as bacteria were observed associated with both structures and in clusters around brain vessels (46, 205). Changes in BBB integrity in bacterial meningitis can also be mediated by direct cellular injury induced by pathogen-derived cytotoxins, such as GBS β-hemolysin/cytolysin (β-h/c), and/or indirectly as a consequence of systemic inflammation (206, 207). Indeed, BBB change can be disruptive or nondisruptive, whether or not there is a physical disruption of the BBB (reviewed elsewhere [207]).

Extracellular pathogens that can cause meningitis all share some key aspects that endow them with heightened ability to survive in the extracellular environment, such as the ability to resist opsonophagocytic killing mechanisms and the ability to obtain iron, a critical nutrient for their growth and survival, from the bloodstream (19). A minimum bacteremia threshold is considered necessary to cause meningitis (19, 208, 209). However, neonatal meningitis can occur in the absence of detectable bacteremia and in the presence of normal CSF parameters (122). Nevertheless, high bacterial levels in the circulation might increase the possibility of bacterial interaction with the CNS barriers, considered a prerequisite for meningeal invasion. Interestingly, this ability decreases with age, increasing again in the elderly, with different pathogens being responsible for meningitis at different ages (13, 19, 20). Also, meningitis in the newborn is often associated with sepsis (16). In fact, the clinical presentation of neonatal meningitis is similar to that observed in neonatal sepsis without meningitis (210). The concomitant systemic inflammation can lead to BBB changes, including those that are nondisruptive, consequently promoting neuroinvasion of pathogens. CNS dysfunction during systemic infection is common, causing a syndrome known as septic encephalopathy (207). Once the CNS is reached, the pathogen benefits from an environment with limited host defense mechanisms, which is more favorable for bacterial growth and spread over the entire surface of the brain and spinal cord (16).

GBS Virulence Factors and Brain Invasion

GBS encodes a wide range of virulence factors that collectively contribute to its dissemination and pathogenesis (Fig. 2). The ability of GBS to invade the CNS reflects a complex interplay between bacterial surface proteins and their associated BMEC receptors. Many of these factors are involved in adherence, penetration, and invasion of the BBB (summarized in Table 1). A few examples are discussed below. Their potential blockade could be used as preventive and treatment approaches.

TABLE 1.

GBS virulence factors associated with brain invasion^a

Virulence factor	Properties	Reference(s)
β-h/c	Pore-forming toxin that promotes the injury of BMEC. Key mediator in causing an acute inflammatory response in brain endothelium.	206
Cas9	Regulates a significant portion of the GBS genome. Promotes interaction with host cells, penetration into the brain, and vaginal persistence.	245
Fibrinogen-binding proteins Fbs (FbsA, FbsB and FbsC)	Promote bacterial adhesion or invasion of endothelial cells. Promote brain colonization.	221, 223, 224, 227
GBS surface protein C (BspC)	Interacts with host cytoskeleton component vimentin, promoting GBS adherence to BMEC.	246
Hypervirulent GBS adhesion (HvgA)	Specific to the hypervirulent CC17 clones. Responsible for GBS tropism to the CNS.	46
Invasion-associated gene A (iagA)	Facilitates BMEC invasion but not BMEC adherence or intracellular survival. Mediates interactions between GBS and BMEC independently of its capsule.	211
Laminin-binding protein (Lmb)	Involved in GBS invasion of BMEC. Contributes to GBS neurotropism.	214, 215
Lipoteichoic acid (LTA)	Found in higher quantities in infants with EOD and LOD; critical for BMEC invasion. Shed LTA may act as competitive inhibitor for BMEC uptake.	212
PilA	Contributes to GBS’s initial attachment to BMEC. It is necessary and sufficient for BMEC adherence.	213
PilB	Contributes to the process of bacterial internalization. It is necessary and sufficient for BMEC invasion.	213
Serine-rich repeat (Srr) glycoproteins	Srr2 promotes bacterial adhesion. Srr1 plays a part in the binding of GBS to BMEC through a “dock, lock, and latch” mechanism.	219, 220, 225, 226
Streptococcal fibronectin-binding protein (SfbA)	Contributes to fibronectin binding and BMEC invasion. Promotes BBB penetration and the development of GBS meningitis.	217
Two-component regulatory system CovS/R	Control of virulence sensor/regulator. Regulates the expression of several genes important for GBS pathogenesis, such as HvgA and β-h/c	46, 239, 240
Two-component regulatory system CiaR/H	Promotes GBS survival in BMEC. Confers GBS resistance to antimicrobial peptides. Prevents GBS trafficking to the lysosome and increases its survival in phagocytic cells.	243, 244

^aBBB, blood-brain barrier; BMEC, brain microvascular endothelial cells; CNS, central nervous system; EOD, early-onset disease; GBS, group B Streptococcus; LOD, late-onset disease.

Invasion-associated gene A (iagA) promotes BMEC uptake of bacteria, and its deletion leads to reduced risk for the development of meningitis in a mouse model (211). The iagA gene encodes an enzyme for biosynthesis of diglucosyldiacylglycerol, a cell membrane glycolipid that anchors lipoteichoic acid (LTA). Cell membrane anchoring of LTA is critical for BMEC invasion (211). Loss of the cell membrane anchor for LTA, in addition to removing a potential invasive factor from the GBS surface, results in LTA shedding, which may also act as a competitive inhibitor for BMEC uptake mechanisms (211). In fact, epidemiologic investigations have shown higher quantities of cell-associated LTA in clinical GBS isolates from infants with invasive disease than strains isolated from asymptomatically colonized infants (212). Genes encoding surface-associated pili contribute to GBS adherence to and invasion of BMEC. It was shown that, while the pilus protein PilA promotes GBS initial attachment and adherence to BMEC, PilB contributes to bacterial internalization and is sufficient for BMEC invasion (213). The surface-expressed protein hypervirulent GBS adhesion (HvgA) is CC17 specific and is important for GBS’s barrier-breaching potential. HvgA acts at different steps of the infectious process, being critical for intestinal colonization by GBS and translocation across both the intestinal barrier and the BBB (46). GBS binding to human laminins, a family of abundant glycoproteins of the basement membrane, through laminin-binding protein (Lmb) contributes to GBS invasion of the BMEC (214) and may be involved in GBS neurotropism (215).

GBS can exploit the host plasmin(ogen) system to enhance its ability to migrate across the BBB and enter the CNS (216). The streptococcal fibronectin-binding protein (SfbA) is present in multiple GBS serotypes and contributes to fibronectin binding and BMEC invasion, promoting BBB penetration and the development of meningitis (217). GBS also binds to human fibrinogen, which is increased during the acute inflammatory response or infection (218), through surface bacterial adhesins, including the serine-rich repeat (Srr) glycoproteins (Srr1 and Srr2) and the fibrinogen-binding proteins Fbs (FbsA, FbsB, and FbsC) (219–224). Srr1 expression is associated with increased attachment to and penetration of BMEC and contributes to disease progression (219). Srr2 contributes to meningitis by promoting bacterial dissemination and invasiveness (220). Moreover, Srr1 expression is absent in CC17 clones, whereas Srr2 is highly expressed and CC17-specific (225). Additionally, it was recently demonstrated that Srr2 binding to BMEC was mediated through the interaction with two specific integrins (α5β1 and αvβ3), contributing to the development of meningitis in a mouse model of infection (226). Importantly, the authors showed that the expression of α5β1 and αvβ3 in the barriers of the brain is higher in neonatal animals than adults, which helps explain the higher susceptibility to brain invasion by GBS during early life (226).

The fibrinogen-binding proteins FsbA and FsbC mediate bacterial adhesion to or invasion of BMEC (221, 224), whereas FsbB mediates GBS invasion into epithelial cells (223). FsbC was also shown to contribute to GBS hematogenous dissemination and brain colonization in a mouse model of infection (224). Interestingly, the ability of the hypervirulent CC17 clones to bind fibrinogen is more dependent on FbsB than FbsA, with the latter being less highly expressed in the analyzed strains (227). The expression of FbsC is absent in clinical GBS isolates belonging to the hypervirulent clones, similarly to Srr1 (224).

Two additional factors that contribute to the pathogenesis of meningitis are the surface polysaccharide capsule and β-h/c. The sialic acid-enriched capsule present on the GBS surface aids in bacterial survival in the bloodstream by restricting complement deposition and thus protecting from neutrophil-mediated opsonophagocytosis (228, 229), and also by inhibiting the generation of chemotactic complement component C5a, thereby reducing phagocyte recruitment (230). GBS sialic acid can also interact with inhibitory Siglecs (231, 232), and our understanding of its role in GBS immunity continues to expand. For example, it was recently demonstrated that GBS sialic acid interacts with Siglec on platelets (Siglec-E, murine; Siglec-9, human), contributing to bacterial pathogenesis by damping platelet activation and increasing bacterial survival in the bloodstream (233). Although not directly involved in GBS neurotropism, GBS sialic acid-mediated immune evasion contributes to increased bacterial survival, which could promote meningitis development. The GBS pore-forming toxin β-h/c, encoded by the cylE gene, promotes the direct damage of several cell types, including BMEC. Notably, β-h/c appears to be an important mediator of the inflammatory response in the brain endothelium through the release of cytokines (206). Nevertheless, it has been suggested that this toxin is not essential to GBS virulence, as in animal models of adult (206) or neonatal (234) GBS meningitis, β-h/c-deficient mutant bacteria were found in the brain.

The expression of many virulence factors is controlled by two-component regulatory systems, which enable bacteria to sense and adjust to microenvironmental changes during their colonization and infectious cycle (235, 236). Briefly, typical two-component systems are formed by a membrane-associated histidine kinase receptor and a cognate response regulator in the cytoplasm, the latter of which is able to activate or repress several genes. Among the more than 20 two-component systems described in GBS (237, 238), CovS/R (control of virulence sensor/regulator) is the most studied, being involved in regulating the expression of several important genes for GBS pathogenesis, such as HvgA and β-h/c, whose role in GBS meningitis is described above (46, 239, 240). Decreasing the CovR function, using GBS strains lacking the covR gene, led to increased expression of cylE (which is involved in β-h/c expression) and enhanced the ability of GBS to colonize the brain in a murine model of GBS hematogenous meningitis (241, 242). Another important two-component regulatory system for serotype III GBS strain is CiaR/H, as it promotes GBS survival in BMEC by limiting endocytic trafficking to the lysosome (243, 244). The mechanism of survival was not restricted to BMEC and was also observed in professional phagocytic cells. Increased survival via CiaR is likely due to increased resistance to host killing mechanisms, mainly antimicrobial peptides, lysozyme, and oxidative stress (243, 244).

A recent study showed that CRISPR-associated protein 9 (Cas9) modulates GBS virulence, playing a major role in colonization and disease pathogenesis. Cas9 regulates over 17% of the GBS genome, including genes that encode virulence and metabolic factors and two-component systems, namely, CiaR, which were downregulated in a Cas9-deficient mutant strain. Also, Cas9 contributes to GBS pathogenesis by promoting interaction with host cells, penetration into the brain, and vaginal persistence (245). A GBS surface antigen I/II protein, BspC, interacts with the host cytoskeleton component vimentin, promoting GBS adherence to BMEC. Additionally, it was found that BspC contributes to GBS meningitis in a hematogenous murine model (246).

An alternative mechanism for BBB invasion, by means of modulation of host endothelial cells by GBS, was proposed (204). GBS can downregulate the expression of tight-junction proteins by modulating the expression of the transcriptional repressor Snail homolog 1 (SNAI1) in host endothelial cells. This promotes BBB disruption, facilitating the paracellular passage into the CNS and inducing progression to meningitis (204). Recently, the potential contribution of hormonal regulation to GBS virulence was also proposed (247). The authors suggest that the high blood concentration of estradiol and progesterone in neonates, due to fetal impregnation throughout pregnancy, promotes bacterial dissemination by modulation of the intestinal barrier permeability, allowing crossing through Peyer’s patches (247).

Central Nervous System Inflammation—Clinical Evidence

The therapeutic use of neuroprotective treatment has the potential to reduce the burden of long-term sequelae. However, developing such optimized therapeutics has been challenging, as understanding the mechanisms leading to neuronal injury in GBS meningitis remain elusive. Early recognition of bacterial components through activation of innate immune receptors such as Toll-like receptors triggers the immune response and signals the production of key inflammatory mediators, including cytokines (248). During classical neuroinflammatory diseases, the CNS is invaded by blood-borne leukocytes that produce cytokines such as IL-1β, IL-6, and TNF-α (249). These cells also produce reactive oxygen species, which further fuel the inflammation (249). Although human data are scarce, it is believed that these are contributing factors to the sequelae observed in GBS meningitis.

Several studies of pediatric meningitis have evaluated the cytokine levels in the CSF and blood of patients (summarized in Table 2). A prospective study of infants younger than 6 months found that the proinflammatory cytokines TNF-α, IL-1β, IL-6, IL-8, and IL-12, as well as the anti-inflammatory cytokine IL-10, were significantly elevated in the CSF of those who had bacterial meningitis (250). However, the study should be interpreted with caution, as 77% of the infants had received antibiotics prior to the lumbar puncture, which could have affected results (250). A different study of children 0.1 to 10.1 years old showed higher CSF concentrations of IL-6 and IL-10 in patients with bacterial meningitis than in children with viral encephalitis, febrile convulsion, or epilepsy and healthy controls (251). In children up to 14 years with bacterial meningitis, the CSF concentrations of TNF-α, IL-6, and IL-8 were significantly elevated compared with those in children with viral meningitis and uninfected controls (252). TNF-α levels were also found to be significantly increased in children (<12 years) with bacterial meningitis, compared with children with aseptic meningitis (253). A study of infants (<6 months) with bacterial meningitis found that the CSF levels of IL-6 were 10 times greater than in those with aseptic meningitis, and TNF-α levels were detected in 60% of children with bacterial meningitis but not in children with viral or aseptic meningitis. Plasma levels did not show significant differences (254). In neonates with GBS disease (sepsis and/or meningitis), half of the blood samples showed detectable levels of TNF-α, as did all urine samples and the only CSF sample tested (255).

TABLE 2.

Altered cytokines during pediatric bacterial meningitis^a

Age	Sample studies	Infectious agent	Findings	Reference(s)
Neonates	CSF Blood Urine	GBS	TNF-α detected in the blood samples of 50% of the patients with GBS disease and in all urine and CSF samples.	255
<6 mo	CSF	GBS, Staphylococcus aureus, Enterococcus faecalis, Streptococcus pneumoniae, Listeria monocytogenes, Haemophilus influenzae, Escherichia coli, Neisseria meningitidis, Klebsiella pneumoniae	Elevated TNF-α, IL-1β, Il-6, IL-8, IL-10, IL-12. TNF-α detected in 60% of patients with bacterial meningitis.	250, 254
0.1-10.1 yrs	CSF	Enterococcus, S. pneumoniae, E. coli, K. pneumoniae, GBS	Elevated IL-6 and IL-10	251
<12 yrs	CSF	Mostly N. meningitidis	Elevated TNF-α	253
<14 yrs	CSF	Not specified	Elevated TNF-α, IL-6, and IL-8	252

^aCSF, cerebrospinal fluid; GBS, group B Streptococcus.

Altogether, these studies suggest a dysregulation in the cytokine milieu within the CSF in patients with bacterial meningitis; however, they should be carefully interpreted, as they are not exclusive of GBS meningitis and frequently include samples from patients with a wide age range.

Lessons from Animal Models

Importance of animal models.

Animal models are widely used to elucidate the pathophysiology of diseases and explore the efficacy of candidate vaccines and drugs before clinical testing in humans (256, 257). Nevertheless, there has been limited success in applying such findings in human clinical settings, and their validity has been debated (258–260). It is estimated that only one-third of animal studies are translated to the level of human randomized trials, and of these, only 10% result in approval of the drug for use in patients (258). Factors contributing to such limitations include design flaws in the animal experiments, insufficient statistical power, lack of external validity of some animal models that might not faithfully mirror a given human disease, and neutral or negative animal studies remaining unpublished (258, 259).

Preclinical studies that better predict the effect in humans are imperative, as the financial and human costs of clinical trials are tremendous. Initiating a clinical trial without the endorsement of robust animal data withholds therapies that might be more successful (259). The most crucial factor to consider when choosing a model system for preclinical experiments should be whether it closely mirrors human disease (260). No single animal model will perfectly reflect a particular human disease, which presents differently between patients. Nevertheless, differences between species and strains could be exploited to pave the way for new therapies and preventive strategies, as they can contribute to the understanding of disease development and progression and host response (256). Thus, efforts should be made to switch to animal models that are appropriately designed, genetically, experimentally, and/or physiologically, as such models are more likely to predict and recapitulate the complexity of the human disease and can advance the field substantially (260, 261).

Experimental models of GBS meningitis.

GBS colonization is not exclusively found in humans, which must be considered when developing an animal model of GBS disease. Indeed, GBS was first isolated in 1887 from contagious bovine mastitis and later named Streptococcus agalactiae (“a-” meaning “no” and “galactiae” meaning “milk”) because of the observation that infected dairy cattle were unable to produce milk or had decreased production (262, 263). GBS is now recognized as a common commensal and sometimes a pathogen among several animals, including other mammals, reptiles, and fish (225, 264, 265). A myriad of animals have been used as models of GBS meningitis, including different rodents (206, 266–274), rabbits (275–278), rhesus monkeys (279), piglets (280), zebrafish (204, 281, 282), and Drosophila (283). Different routes of infection have also been used to deliver GBS in animal models. The choice of infection route is also relevant, as it has to induce disease akin to that in humans, lead to bacterial dissemination, and include key steps of the pathogenesis while being reproducible within the limits of biological variations. Until very recently, only two main models existed: (i) hematogenous models by the intraperitoneal or intravenous route (206, 217, 219, 267, 268, 281, 284–286) and (ii) direct brain infection by the intracisternal route (269–272, 274, 279, 287). Models of the intranasal, intra-amniotic, and intragastric routes of infection were also reported (46, 266, 273, 279).

All these models have made important contributions, as they made it possible to identify the mechanisms by which GBS disseminates, enters the CNS, and leads to meningitis. However, their applicability as preclinical models that can reflect neonatal human disease is limited (examples of main findings/hallmarks of meningitis obtained from animal models are summarized in Table 3). The neonatal immune system is a highly regulated system, rather than an immature one, but phenotypically and functionally different from that in adults (288, 289). Moreover, it has been suggested that the neonatal BBB permeability is more restricted to opening than that in adults (290). Remarkably, adult animals are often exploited for modeling GBS disease, and many mechanisms underlying GBS meningitis were extrapolated from these models (206, 217, 219, 267, 268, 284, 286). A significant number of adult models contributed to the generalized concept that GBS meningitis is a classical acute neuroinflammatory disease, characterized by an intensive influx of inflammatory cells, especially neutrophils (206, 217, 219, 284, 286). This was further sustained by studies showing that cells of the BBB endothelium infected with a virulent strain have an increase in the transcription pattern of the chemokines IL-8, Groα, and Groβ, which could contribute to amplified neutrophil recruitment (206). This initial wave would further act on CNS-resident cells and trigger a local response, inducing the inflammatory cascade. Few studies have also shown a role for IL-1β in protecting the brain from GBS colonization (267, 268).

TABLE 3.

Animal models of GBS meningitis^a

Age	Animal model	Infection route	Findings/hallmarks of meningitis	Reference(s)
Neonatal/juvenile	Mouse	Oral	Bacterial translocation across intestinal barrier precedes brain invasion.	46
		i.vag.	Bacteremia does not predict brain colonization. Survivors present long-term sequelae. Meningeal congestion, vascular hyperemia, edema and hemorrhage. Reactive microglia and astrocytes.	234
	Mouse, outbred	i.nasal	Bacteria present in the lumen of blood vessels and in the perivascular connective tissue. Perivascular phagocytic cells surrounding meningeal, cerebral, and periosteal vessels.	273
	Pig	i.ventricular	Cortical damage and generalized hypoxia/ischemia. Cerebral hypertension and edema	280
	Rat	i.c.	Cell death in the cortex and hippocampus. Intense granulocytic inflammation in the subarachnoid and ventricular space. Anti-TNF treatment did not affect CSF bacteria and parameters or death	269, 274
	Rat	i.c.	Reactive oxygen species and lipid peroxidation contribute to cortical necrosis and hippocampal cell death. Infiltration of granulocytes in the subarachnoid and ventricular space. Treatment with iNOS inhibitor leads to higher CSF bacterial titers and increased severity	287, 292
	Rat	i.c.	Decreased long-term memory during adulthood, associated with decreased levels of BDNF in the hippocampus and cortex.	270
	Rat	i.c.	BDNF as an adjunctive therapy to protect from neuronal injury	272
	Rat	i.c.	Subarachnoid and ventricular inflammation, vasculopathy and neuronal injury. Intense astrogliosis.	271
	Rhesus monkey	i.v. i.c. i.amniotic	Bacteremia does not predict outcome, but a positive CSF culture does.	279
	Zebrafish (larvae)	Surrounding the heart	Neutrophils present at the site of bacterial injection, but there was no migration to the brain. Increase in IL-1β and CXCL8 transcripts.	282
Adult	Mouse	i.p i.v.	Deficiency in IL-1β or IL-1R leads to higher bacterial levels in the brain	267, 268
	Mouse, outbred	i.v.	Exploration of GBS virulence factors contributing to brain invasion. Bacterial load in the brain associated with influx of inflammatory cells, especially neutrophils and substantial hemorrhage. Levels of brain colonization correlate with histopathologic abnormalities. Meningeal thickening, thrombosis of meningeal vessels, leukocytic infiltration, and areas of bacterial microabscess formation.	206, 217, 219, 284, 286
	Zebrafish	i.p.	Increased relative expression of IL-1β and IL-6 in the brain. Cerebral swelling and edema in infected animals.	281

^aBDNF, brain-derived neurotrophic factor; CSF, cerebrospinal fluid; i., intra; i.c., intracranial/intracisternal; i.p., intraperitoneal; i.v., intravenous; i.vag., intravaginal; iNOS, inducible nitric oxide synthase.

Another drawback in these studies is the use of artificial routes of infection that do not accurately reflect the mother-to-newborn transmission of GBS and the normal course of the disease in humans (bacteremia-meningitis sequence) and therefore can generate a significant amount of misleading data. Neuronal damage in bacterial meningitis is not a monocausal event and is frequently studied as such, with animal models targeting only a particular step of the pathophysiological process (organ colonization, septicemia, or meningitis). Indeed, intense granulocytic inflammation in the subarachnoid and ventricular space was observed in neonatal models, similar to that observed with hematogenous models in adults, but bacteria were directly administered into the brain (274). Histopathological analyses have also shown the presence of hemorrhage, even in the brains of mice without bacterial load in the brain (284, 286). Adult rats that experienced neonatal GBS meningitis showed decreased levels of brain-derived neurotrophic factor (BDNF) in the hippocampus and cortex and decreased levels of nerve growth factor in the hippocampus. BDNF is crucial in sustaining physiological processes of the healthy adult brain, with a role in synaptic plasticity and long-term potentiation, and thus influences learning and memory (270). Likewise, surviving animals had decreased long-term memory compared to the uninfected control group (270). Despite progress, and the recognition that brain infection by GBS leads to neuronal injury with areas of cortical necrosis and hippocampal cell death (269, 271, 272, 274, 287), the mechanistic events leading to long-term sequelae remain largely unknown.

More recently, a mouse model of intravaginal colonization and vertical transmission was developed, reflecting the likely natural route of infection (234). This model reproduces the crucial and clinically relevant steps of GBS infection in human newborns: (i) mother-to-infant vertical transmission; (ii) bacterial colonization rate in blood, lungs, liver, gut, and brain of newborns; (iii) GBS survivors experiencing long-term neurological sequelae. In this model, no parenchymal invasion by leukocytes or inflammatory cytokines was observed in the neonatal brain, despite local and systemic infection (234), contradicting the previously accepted idea. In a zebrafish larva model, neutrophils were observed at the site of bacterial injection, but there was no migration to the brain (282). A sparse rather than purulent inflammatory response was also documented in postmortem analysis of human babies with meningitis (291). Also, in the vertical model, bacteremia did not predict outcome (234). The same was observed in a rhesus monkey model upon intra-amniotic infection (279) and also correlates with human data (1, 11). Nevertheless, several knowledge gaps remain. This model implies that other mechanisms are responsible for the observed sequelae and is a promising tool for studying neuronal damage in GBS meningitis and developing new neuroprotective approaches to reduce the burden of neurological sequelae in survivors.

CONCLUSION

Neonatal bacterial meningitis remains a common cause of infection worldwide, with GBS as its leading agent. Despite significant advances in neonatal intensive care and recent efforts at prevention and early diagnosis of disease, mortality and morbidity remain unacceptably high. Survivors of GBS meningitis are at risk of developing neurodevelopmental impairments, with numerous consequences for their personal and familial life and their community and society.

Recent advances have been made in understanding the pathogenesis of GBS meningitis, such as a more comprehensive knowledge of GBS virulence factors contributing to brain invasion. However, the mechanisms that lead to neuronal injury remain largely unknown, and continued efforts are needed to develop new therapeutic and neuroprotective approaches. Animal models that better mimic the clinical pathophysiology are promising tools to study therapies and prevention strategies. They will make it possible to model GBS-induced diseases more accurately, with features that are more similar to those described in humans and thus have a higher likelihood of translating to the clinic.

Biographies

Teresa Tavares (M.D.) is a training doctor at the Centro Hospitalar Universitário do Porto (CHUP), in Porto, Portugal. She studied medicine at the Instituto de Ciências Biomédicas Abel Salazar of the University of Porto (ICBAS-UP) and graduated with a master’s degree in 2020. Currently, she is practicing medicine and beginning her residency training at CHUP. She started studying neonatal meningitis caused by GBS in 2016 at ICBAS-UP. Her research interests focus on neuroimmunology, the neonatal period, pediatric infections and neurodevelopment, and animal models of disease.

Liliana Pinho (M.D.) studied medicine at the University of Beira Interior (2001 to 2007) and then specialized in pediatrics in 2014 and neonatology in 2018 at Centro Hospitalar Universitário do Porto (CHUP), Centro Materno Infantil do Norte. She has worked as an assistant in the Department of Neonatology of that institution since 2015, with particular interest and focus on the field of neonatal infectiology.

Elva Bonifácio Andrade (PhD) graduated in biochemistry (University of Porto [UP]) in 2005 and obtained her Ph.D. in biomedical sciences at the Instituto de Ciências Biomédicas Abel Salazar (ICBAS-UP) in 2013. Her Ph.D. work focused on the characterization of the innate immune mechanisms leading to neonatal susceptibility to GBS infections. During her postdoctoral studies, she developed a mouse model that recapitulates GBS neonatal infection pathogenesis, with features similar to those described in humans. Since 2020, she has been a research scientist at i3S (Instituto de Investigação e Inovação em Saúde, University of Porto), where she established a new line of research focused on understanding the complex neuroimmune interactions during GBS disease, aiming at developing new therapeutic and neuroprotective strategies. She is also an invited assistant professor at ICBAS-UP and at the School of Health, Polytechnic Institute of Porto, where she has been engaged in teaching immunology and microbiology.