Introduction

The central nervous system (CNS) can be infected with numerous infectious agents, which result in significant morbidity and mortality, despite the effectiveness of available antimicrobials [1–5]. In large international cohorts, the case–fatality rates for adults with pneumococcal, tuberculous, and herpetic CNS infections are 14, 17, and 10%, respectively, despite treatment [6–8]. Thus, risk assessment in this subgroup of patients is of key importance in predicting the likely pathogen leading to rational treatment. Although there are considerable data on the epidemiology of these CNS infectious syndromes, data are lacking to facilitate our understanding of the cumulative epidemiology of community-acquired CNS (CA-CNS) infections. We, therefore, performed a multicenter study to provide an insight into CA-CNS infections and to better understand the burden of infection in community settings, and to explore risk assessment data for the management of these infections.

Materials and methods

Study design

We designed a multicenter, retrospective, cross-sectional study performed in the infectious diseases and neurology departments of participating referral hospitals. The study was led by the Infectious Diseases International Research Initiative (ID-IRI). We approached >50 survey sites in >25 countries to participate in this study. Data were collected retrospectively during the period from 1st May 2012 to 1st May 2014 and included all CA-CNS infection patients over the age of 15 years hospitalized at the survey sites. All epidemiological, clinical, laboratory, therapeutic, and outcome data were collected using a standardized questionnaire (Excel 2010). The participant centers submitted their individual datasets via email to the lead hospital for analysis. Patient confidentiality was considered and the study was approved ethically by the institutional review board of Fatih Sultan Mehmet Education and Research Hospital in Istanbul.

Definitions

The diagnoses were made according to the clinical suspicion of the examining clinicians based on compatible clinical presentations of the patients with CA-CNS infections, and in accordance with standard definitions [2–5, 9]. The CA-CNS infections were defined as infections affecting the CNS, and were acquired in daily life. These infectious disorders were not associated to infections in the healthcare settings.

Chronic CA-CNS infection:

Patients with symptoms for more than four weeks at the time of diagnosis were included in this category [10].

Elderly classification:

Patients over the age of 64 years were classified as elderly [11].

Primary immunosuppression:

Congenital immunodeficiency diseases are classified in this category [12].

Secondary immunosuppression:

Immunodeficiency conditions depending on underlying immune diseases or due to immunosuppressive treatments are defined as secondary [12].

Probable immunosuppression:

Patients with probable immunosuppressive conditions like human immunodeficiency virus (HIV; immune status unknown), diabetes mellitus, malignity, pregnancy, and others were included in this group. Patients were included in this group when they did not have detailed data in our database in determining the level of immunosuppression.

Adverse clinical outcome:

Either death or sequelae were defined as adverse clinical outcomes.

Data sources and measurements:

The microbiological diagnoses of patients compatible with CNS infections were made according to the microbiological techniques below and are presented in detail in Table 1:

  1. a)

    Neurosyphilis: CSF VDRL, serum VDRL/RPR, fluorescent treponemal antibody absorption (FTA-ABS), T. pallidum particle agglutination (TPPA)/T. pallidum hemagglutination test (TPHA), T. pallidum microhemagglutination (MHA-TP) [13].

  2. b)

    Neuroborreliosis: Molecular testing and serology [14].

  3. c)

    Neurobrucellosis: Non-specific CSF culture, blood culture, CSF Rose Bengal test, CSF Wright agglutination test, serum Rose Bengal test, serum Wright agglutination test, CSF-PCR [15].

  4. d)

    CNS tuberculosis: CSF culture for tuberculosis, acid fast staining of CSF, molecular testing (MTB PCR), and response to anti-tuberculosis treatment in probable cases [16].

  5. e)

    Other bacterial infections: Non-specific CSF culture, blood culture, and molecular testing [1, 17].

  6. f)

    Cryptococcal CNS disease: Fungal CSF culture, non-specific CSF culture, blood culture, CSF cryptococcal antigen testing, and molecular testing [17].

  7. g)

    Candidal CNS disease: Fungal CSF culture, non-specific CSF culture, blood culture, and serological testing [18].

  8. h)

    Viral CNS infections: Molecular testing and serology [3, 5].

  9. i)

    CNS toxoplasmosis: Molecular testing and serology [5].

Table 1 The efficacy of the microbiological methods used in the diagnosis of community-acquired central nervous system (CA-CNS) infections
Full size table

The detection of pathogens was directly related to the technical capacity of the participating centers and the diagnostic approaches of the examining clinicians. HIV parameters (e.g., CD4+ T-cell counts and HIV-RNA levels) indicating severity of illness were not in our database. Hence, HIV-positives are included in a separate block.

Statistical methods

Chi-square, Pearson’s chi-square, continuity correction of Pearson, and Fisher’s exact test were used to analyze the data. We used a cut-off p-value of p < 0.05 to indicate significance for all tests. Continuous variables were analyzed by using histograms and the Kolmogorov–Smirnov test for normality. After this separation, a Mann–Whitney U-test was used for variables that were not distributed normally.

Results

A total of 2583 patients with CA-CNS infections were included from 37 referral centers in Albania, Bosnia and Herzegovina, Croatia, Czech Republic, Denmark, Egypt, France, Hungary, Hong Kong, Iran, Italy, Kazakhstan, Kosovo, Pakistan, Portugal, Romania, Serbia, Syria, Texas/USA, and Turkey. Table 1 highlights the microbiological techniques and their positivity rates used in the diagnosis.

Infectious syndromes

There were 2603 infectious CNS syndromes identified in 2583 cases. Patients were classified as meningitis (n = 1292, 50%), encephalitis (n = 208, 8%), meningoencephalitis (n = 888, 34.4%), and meningoencephalomyelitis (n = 1, 0.04%). There were 144 (5.6%) patients with suppurative intracranial infections. These were brain abscesses (n = 97), subdural empyemas (n = 14), epidural abscesses (n = 31), other abscesses (n = 2), other syndromes (n = 69), and arachnoiditis (n = 2). The distribution of infectious CNS syndromes is presented in Table 2.

Table 2 The presentation of the patients according to etiological agents
Full size table

Microbiological data

In this study, a CNS pathogen was disclosed in 1079 (41.8%) patients. However, no infecting pathogen was detected in 1504 (58.2%) patients. The diagnoses were established with either direct or indirect methods, or with the combination of different methods, as indicated in Table 1. We identified Streptococcus pneumoniae (n = 206, 8%) and Mycobacterium tuberculosis (n = 152, 5.9%) as the frequent causative agents. The distribution of infecting pathogens according to the ultimate diagnosis and immunosuppressive states are presented in Tables 3 and 4, respectively. Common pathogens in CA-CNS infections are presented in Fig. 1.

Table 3 The distribution of infecting pathogens according to the ultimate diagnosis
Full size table
Table 4 The distribution of ultimate microbiological diagnoses according to the immunosuppression status of the patients
Full size table
Fig. 1
figure 1

Common pathogens in community-acquired central nervous system (CA-CNS) infections

Full size image

In 77 out of 2583 (3%) patients, multiple etiological agents were detected. In 75 cases, two causative agents were identified and in two patients, three agents were recorded. When these patients were categorized according to clinical presentations, there were 43 cases of meningitis, in which 87 different organisms were identified [S. pneumoniae (n = 20), T. pallidum (n = 10), enteroviruses (n = 9), cryptococci (n = 7), EBV (n = 5), streptococci other than pneumococci (n = 4), N. meningitidis (n = 3), VZV (n = 3), WNV (n = 3), coagulase-negative staphylococci (n = 2), Borrelia spp. (n = 2), HHV-6 (n = 1), CMV (n = 1), Brucella spp. (n = 1), Listeria spp. (n = 1)]. In 26 patients with meningoencephalitis, 53 agents were recorded as multiple pathogens [TBEV (n = 16), Borrelia spp. (n = 13), VZV (n = 10), M. tuberculosis (n = 3), HSV (n = 3), enteroviruses (n = 3), EBV(n = 2), S. pneumoniae (n = 1), coagulase-negative staphylococci (n = 1), Cryptococcus (n = 1)]. In five patients with encephalitis, ten agents [HSV-1 (n = 4), S. aureus (n = 1), S. pneumoniae (n = 1), VZV (n = 1), WNV (n = 1), EBV (n = 1), enterovirus (n = 1)] were identified. Finally, in three patients with brain abscesses, six pathogens were disclosed [streptococci other than pneumococci (n = 2), Neisseria spp. (n = 2), CMV (n = 1), EBV (n = 1)].

Age distribution

The mean age of the cases was 47.63 (SD ± 19.8) years. 446 (17.3%) patients were elderly, with a mean age of 74.58 (SD ± 7.02) years. Among the patients with any identified pathogen, the numbers of patients over 65 years and less than 65 years of age were 250 (23.2%) and 829 (80.9%), respectively. Although S. pneumoniae [59/250–147/829 (p = 0.039)], Listeria spp. [13/250–14/829 (p = 0.004)], and VZV [41/250–50/829 (p < 0.001)] caused significantly more infections in the elderly, M. tuberculosis [23/250–129/829 (p = 0.011)], enteroviruses [2/250–89/829 (p < 0.001)], and TBEV [13/250–79/829 (p = 0.032)] affected adults significantly more than the elderly (Table 2).

Chronicity

96 (8.9%) out of 1079 patients with any identified pathogen presented with clinical features of chronic meningitis. Treponema pallidum (24/96–0/983; p < 0.001), Brucella spp. (6/96–8/983; p = 0.001), M. tuberculosis (37/96–115/983; p < 0.001), and Borrelia spp. (10/96–28/983; p = 0.001) were more likely to be causing chronic infections, while streptococci other than pneumococci (0/96–55/983; p = 0.012), Neisseria meningitidis (0/96; 45/983; p = 0.046), VZV (0/96–91/983: p = 0.003), enteroviruses (1/96–90/983; p = 0.011), and TBEV (0/96–92/983; p < 0.001) frequently presented non-chronic patterns. Streptococcus pneumoniae (0/96–206/983; p < 0.001) always presented with acute patterns (Table 2).

Immunosuppressive states

There were 741 immunosuppressive conditions in 603 (23.3%) out of 2583 patients. Of these, there were 170 (6.6%) confirmed (primary and secondary) immunosuppressive patients and 571 (22.1%) probable immunosuppressive conditions in 508 cases. The distribution was as follows:

  1. 1.

    Primary immunosuppression: One hyper IgM syndrome (type 1), two hypogammaglobulinemia, one mucocutaneous candidiasis, one IgA deficiency.

  2. 2.

    Secondary immunosuppression: Acute lymphocytic leukemia, bone marrow transplantation, amyloidosis, acute myelocytic leukemia, autoimmune hemolytic anemia, leukemia, myelofibrosis, pyoderma gangrenosum, multiple myelomas, and myelodysplastic syndrome were seen in one patient. Two monoclonal gammopathies, two chronic lymphocytic leukemias, two large cell lymphomas, three cirrhosis, six solid organ transplantations, seven splenectomies, 31 with chemotherapy, and 121 systemic corticosteroid use were other conditions.

  3. 3.

    Probable immunosuppression: 263 diabetes mellitus, 116 malignities, 84 HIV infections, 27 pregnancies, 15 alcoholics, ten chronic liver diseases, six Addison’s disease, six end-stage renal disease, seven systemic lupus erythematosus, three rheumatoid arthritis, three intravenous drug users, two ulcerative colitis, and two psoriasis. Hypersplenism, acute renal failure, Behçet’s disease, positive anti-GM/GSF antibody, brain tumor, congenital leukodystrophy, disseminated sclerosis, Down syndrome, generalized herpetic infection, genital herpes, Crohn’s disease, non-Hodgkin lymphoma, panhypopituitarism, scleroderma, and ankylosing spondylitis were seen in one patient.

When the patients with a etiologically proven agent was considered (n = 1179), S. aureus was more frequent in diabetics (11/121, 9.1%) compared to non-diabetics (37/958, 3.9%) (p = 0.017). Coagulase-negative staphylococci were more common in patients with secondary immunosuppression (6/95, 6.3%) than the others (18/984, 1.8%) (p = 0.015) and in patients with concurrent malignancies (n = 4/61, 6.6%) compared to those without concurrent malignancies (n = 20/1018, 2%) (p = 0.042). Listeria monocytogenes was more common in secondary immunosuppression (9/95, 9.5%) than those without (18/984, 1.8%) (p = 0.015) and in malignancy patients (n = 5/61, 8.2%) compared to those without malignity (22/1018, 2.2%) (p = 0.015). Finally, cryptococci was the most frequent in etiologically proven CNS infection patients with HIV infection (n = 13/50, 26%) compared to HIV-negative patients (n = 14/1029, 1.4%) (p < 0.001).

Outcomes

In this study, 477 (18.5%) patients survived with sequelae and 227 (8.8%) died during hospital stay. Motor deficit (n = 128, 5%), cranial nerve involvement (n = 92, 3.6%), cognitive impairment (n = 70, 2.7%), persisting headache (n = 38, 1.5%), and epilepsy (n = 26, 1%) were the most frequent sequelae. Overall, 1879 (72.7%) patients were discharged with complete cure. Among the patients with any etiologically proven pathogen, 711 (65.9%) patients had complete cure, 261 (24.2%) survived with sequelae, and 107 (9.9%) patients died. Adverse clinical outcomes in patients with an etiologically proven pathogen [368/1079(34.1%)] were significantly more common than those without confirmed etiological diagnosis [336/1504 (22.7%)] (p < 0.001). The Glasgow Coma Scale (GCS) scores of the patients with a etiologically proven CNS pathogen on admission were significantly lower in patients with adverse clinical outcomes (mean: 11.94 ± 3.69) compared to survivors with complete cure (mean: 13.37 ± 2.86) (p < 0.001). The outcome distributions in CA-CNS infection patients are presented in Table 5.

  1. a)

    Mortality rate by specific pathogens: The fatality rate was significantly higher for L. monocytogenes [(8/107–19/972); (p = 0.003)], S. aureus [(10/107–38/972); (p = 0.021)], M. tuberculosis [(28/107–124/972); (p < 0.001)], and S. pneumoniae [(33/107–173/972); (p = 0.002)], while it was significantly lower in enterovirus [(1/107–90/972); (p = 0.006)] and TBEV [(1/107–91/972); (p = 0.005)] infections, and where no death was seen due to Borrelia spp. infections [(0/107–38/972); (non-applicable statistics)].

  2. b)

    Rate of neurological sequelae by specific pathogens: Persistence of sequelae was significantly higher in herpes simplex virus (HSV) type 1 cases [(24/261–24/818); (p < 0.001)], while it was less frequent in enterovirus [(13/261–78/818); (p = 0.029)] and TBEV [(8/261–84/818); (p < 0.001)] infections.

Table 5 The outcomes of the CNS infection patients
Full size table

Discussion

CNS infections have significant mortality and permanent neurological sequelae worldwide [17]. Despite the advances in management, 27.3% of patients with CA-CNS infections experienced unfavorable outcomes in this study, including persistent sequelae and death. The most frequent sequelae were motor deficits (5%) and cranial nerve involvements (3.6%), in accordance with the current literature [19]. Cognitive impairment, persisting headache, and epilepsy were seen in descending order (Table 5). We found that CA-CNS diseases due to S. aureus, M. tuberculosis, S. pneumoniae, and Listeria spp. were the most fatal etiologies, while the patients with CA-CNS infections due to enterovirus, TBEV, and Borrelia spp. were less likely to die. Furthermore, HSV-1 was associated with sequelae most commonly, while enteroviruses and TBEV were less likely to be complicated with sequelae.

Meningitis and encephalitis, known as two intersecting syndromes, were reported to be the most frequent CNS infections [17, 20, 21]. The use of neuroimaging before lumbar puncture has generated considerable debate, with some investigators pointing to delays in antibiotic administration, reduced likelihood of identifying a pathogen, and an increase in mortality [22]. In this study, half of the cases were classified as meningitis and 8% as encephalitis, and an unspecified 34% of patients were classified as meningoencephalitis. Considering the high frequency of cases presenting with encephalitic pattern in this study, the detection of brain parenchymal involvement with imaging together with positive EEG findings is directly suggestive of encephalitis [23] and leads to early administration of antivirals that reduce unfavorable outcomes [8]. Added to that, suppurative intracranial infections, where brain abscesses dominated in two-thirds and epidural abscesses in one-third, made up 5.6% of all CA-CNS infections. Thus, these data indicate the necessity of using radioimaging methods and EEG in an unspecified CA-CNS infection patient considering the high frequency encephalitis and suppurative intracranial infections.

Consequently, the diagnostic capacity of the hospitals treating CA-CNS infections determines the early diagnosis, which is key to rational therapies [2, 4, 8, 17]. We found that the etiological diagnosis was established in 41.8% of all patients. The data are scarce in the literature in that context. In an African study, 35% of the cases with meningitis [24] and in a Turkish systematic review, 32.6% of the cases [25] were reported to have an etiological diagnosis. In addition, 3% of the CA-CNS infections had multiple infecting agents, which must not be overlooked by the treating clinicians. The most frequent infecting pathogens in this study were S. pneumoniae and M. tuberculosis. TBEV, VZV, enteroviruses, and HSV infections are among the other most frequent CA-CNS infections. Hence, these data highlight the importance of pneumococcal immunization, producing herd immunity and decreasing invasive disease [22], and widespread tuberculosis elimination programs. Accordingly, N. meningitidis, which is the agent of epidemic meningitis, was a still noteworthy pathogen and had a share of 1.6% in CA-CNS patients. Although Streptococcus suis was reported to be one of the major causes of meningitis in some parts of Asia [22], Streptococcus anginosus was recovered mainly from patients in European countries among the non-pneumococcal species in this study. In a recent meta-analysis, the most frequent causes of brain abscesses were streptococci and S. aureus [26], and our data were in accordance with this. Pneumococcal CNS disease and VZV infections were the most frequent infections in the elderly, followed by M. tuberculosis and L. monocytogenes infections. In the literature, pneumococcal and listerial CNS diseases have been known to be quite frequent in elderly patients [4, 27]. However, our data affirm that, in accordance with the high tuberculous CNS disease worldwide [28], and as specific cell-mediated immunity declines in the elderly for VZV infections [29], they should not be underestimated in the elderly. According to our data, S. pneumoniae, S. aureus, L. monocytogenes, and VZV infections were seen relatively more frequently in older patients, while M. tuberculosis and enterovirus infections were observed in adult CA-CNS cases. These differences may be explained owing to the decreasing immunity with advancing age and epidemiological trends [4, 21, 30].

Listeria monocytogenes meningitis was more often found in patients with acquired immunodeficiencies [2, 31], while cryptococcal CNS disease was common in HIV-infected patients [32, 33]. Our data were in accordance with that for both pathogens. In addition, we have disclosed that S. aureus was a significant agent in diabetic patients, while coagulase-negative staphylococci were common in patients with secondary immunosuppression and malignancy patients. These two groups of pathogens were reported to be the leading agents in various subgroups of immunocompromised patients [34, 35]. However, to the best of our knowledge, this is the first report interrelating immunosuppressive states and staphylococcal CA-CNS infections. Our data may, therefore, support the recommendations in using anti-Gram-positive agents as part of the combination regimen in bacterial CA-CNS infections [2, 3, 7].

In the literature of the past two decades [32, 33, 36], the most frequent agents causing chronic CNS infections were cryptococci and M. tuberculosis, depending on the HIV status. In HIV-positive patients, cryptococcal disease predominated [32, 33] and in HIV-negative patients, CNS tuberculosis [36] was the most frequent diagnosis in chronic CNS infections. In this study, 4.3% of the cases presented with a chronic pattern extending over 4 weeks, and neurosyphilis, brucellar meningitis, neuroborreliosis, and CNS tuberculosis were significantly more likely to present chronic courses. In contrast, cryptococcal meningitis was seen only in 2.2% of cases with chronic CNS disease, although it was still the most frequent CA-CNS infection in HIV-positive individuals. This shift was most probably due to the increasing efficiency of HIV treatment worldwide, with the resultant decrease in opportunistic infections [37].

We believe that our study has important implications for the management of CA-CNS infections. Streptococcus pneumoniae and M. tuberculosis were the leading CA-CNS pathogens. Considering the high frequency of patients predisposed as encephalitis, meningoencephalitis, or suppurative intracranial infections, both MRI and EEG should be applied to all patients with probable CA-CNS infections. The need to improve the diagnostic capacity in centers treating CNS infections is another serious concern. Accordingly, high sequelae forming infections like HSV-1 were important in the differential diagnosis. In addition to pneumococcal disease and CNS tuberculosis, both common in all ages, VZV, S. aureus, and L. monocytogenes should be taken into consideration in the elderly population. Accordingly, staphylococcal, cryptococcal, and listerial infections should be considered for immunocompromised patients. Finally, clinicians need to consider that neurosyphilis, brucellar meningitis, neuroborreliosis, and CNS tuberculosis may have a predilection for prolonged courses where alternative strategies are needed.