The initial database search revealed a total of 1291 case records that were analyzed for the final inclusion. After removing duplicate records, 870 records were assessed for further inclusion. After the removal of ineligible articles, a total of 118 patients with CNS actinomycosis (95 articles) were analyzed in this systematic review (Fig. 1). Out of 118 patients, 29 were from a case series, and the remaining 89 were individual data. The reference list of all reported cases in this review is given in supplementary material (Additional file 1: s4).

Flow chart of articles selection according to PRISMA guideline

This systematic review identified a total of 118 cases of CNS actinomycosis. The median age of the patients was 45 years. Cases have been reported from 2 months to 90 years of age. The proportion of male patients was 57%. Most cases were reported from Asia (49%), followed by the USA and Europe (Table 1). We found only three cases reported from African countries.

Focal neurological deficits were the most common clinical presentation in CNS actinomycosis (65.3%). The other common presentations were headache, fever, and altered sensorium (Table 2). Meningeal signs were present in 21% of the patients. The mean duration of symptoms (time from the onset of disease to the first presentation in hospital) was 82 days in CNS disease, which indicates the indolent nature of the illness. The cervicofacial region was the most common other site involved in CNS actinomycosis due to contiguous spread of infection (Table 1). Of note, disseminated actinomycosis infection was reported in 19.5% (n = 23) of the cases (Table 1). In these disseminated cases, the other manifestations were pulmonary disease, abdominal actinomycosis and osteomyelitis (10.2%, 5.1% and 4.2%, respectively). Notwithstanding, actinomycosis bacteremia was found in 4 (3.4%) of the cases in this review [19,20,21,22]. One of these patients also had features of endocarditis as a clinical spectrum of disseminated actinomycosis [22]. All of these patients presented acutely (within 14 days of onset of symptoms). Isolated CNS disease was found in 80% (n = 95) of the patients.

In this review, we also searched for various immunodeficient states, risk factors, and comorbidities associated with CNS actinomycosis. Only 12.7% of patients were immunocompromised (Table 1). Among predisposing factors, a history of dental procedures was the most frequent (in 15% of the patients, Table 3). Dental infections (active or in the recent past), head trauma/surgery, and ear infections were the other predisposing factors for CNS actinomycosis (Table 3). Of note, we could not find any risk factors or immunodeficiency state in 49 patients (41.5%). We also compared the clinical presentation of disseminated disease with isolated CNS actinomycosis. The mean age of disseminated cases was significantly increased compared to isolated CNS cases (57 vs 41 years, p =  < 0.003). Furthermore, the duration of illness prior to presentation was shorter in disseminated cases (44 vs 92 days). The proportion of immunodeficient patients was significantly higher in disseminated cases than in isolated CNS cases (23% vs 9.8%, p = 0.02).

In most patients, the diagnosis of CNS disease was made by characteristic histopathological findings and culture. In 55% of patients, actinomycosis was identified in culture. The most common specimen for culture was pus aspirated from a brain abscess (Table 4). CSF (cerebrospinal fluid) culture positivity was found in nearly 17% of the patients. Actinomycosis was isolated from blood in four patients. CSF cytology report was available in only 19 patients in this review. More than two-thirds (68%) of patients showed a neutrophilic pleocytosis with a median protein of 112 mg/dl (Table 4). The molecular diagnosis was established in 12 cases by MALDI-TOF (Matrix-Assisted Laser Desorption/Ionization-Time of Flight), PCR and 16S RNA sequencing methods. The species identification was possible in only 53 cases (Table 1). Actinomyces israelii was the most common species causing CNS disease, followed by A. meyeri and A. viscosus (41.5%, 22.6% and 11.3%, respectively). A. neuii, A. turicensis, A. odontolyticus, A. georgiae, A. naeslundii, and A. oris were the other species identified in the remaining 25% of the cases (Table 1). Actinomycosis is usually considered a polymicrobial infection involving anaerobic and aerobic bacteria. Co-isolates depend on the site or source of infection. In 35 cases (30%), Co-isolates were identified along with actinomycosis. Most isolates were related to fusobacterium species (n = 14), followed by Streptococcus species (Additional file 1 s5, Table 1).

Among the neuroimaging findings, focal space-occupying lesions (brain abscess) were the most frequent (55%). The majority of these lesions were single abscesses (Table 4). The frontal and parietal lobes were the most common site of brain abscesses (Table 4). Leptomeningeal enhancement, subdural empyema, and ventriculitis were the other neuroimaging findings in CNS actinomycosis (Table 4). Spinal cord involvement in the form of intradural abscess is also found in 19 cases, and the cervical and lumbar region were the site affected most frequently (Table 4). On neuroimaging, brain abscess incidence was significantly higher in disseminated cases compared to isolated CNS disease (73% vs 52%, p = 0.034).

Actinomycosis is a chronic suppurative indolent disease that usually requires prolonged antimicrobial therapy to effectively cure the infection and prevent relapse. In this review, surgical intervention (abscess drainage or tissue debridement) was performed in 79 (67%) patients. The mean antimicrobial duration was 160 days for treating CNS actinomycosis. On a few occasions, therapy was continued beyond one year (up to 720 days).

The mean duration of antimicrobials was 150 days in isolated CNS cases compared to 200 days in disseminated actinomycosis. Single antibiotic use was preferred in 58.7% of cases; however, combination antimicrobials were also used in 41.3% of patients (Table 5). Of note, combination antimicrobial' use was significantly higher in disseminated cases than in isolated CNS disease (65% vs 37%, p = 0.001). There was a wide heterogenicity in antibiotic preference in all cases. Penicillin was the most frequently used antimicrobial, followed by ceftriaxone and ampicillin (47%, 29% and 11%, respectively). Among oral antimicrobials (patients who were discharged after initial iv. Antibiotics), amoxicillin, cotrimoxazole, and doxycycline were preferred drugs. Some recent reports also describe the use of meropenem in CNS actinomycoses [23, 24].

The exact antimicrobial regimen and duration of therapy are not well defined in actinomycosis. We also analyzed the role of combination therapy, type of antibiotics and duration of treatment on clinical outcomes. Survival rate was similar in patients who received combination antibiotic compared to those treated with single drug (Table 6). However, neurological sequelae were significantly less in patients treated with combination therapy (Table 6). Compared to the penicillin-based regimen, ceftriaxone-based treatment was associated with lesser residual deficits though the mortality rate did not differ (Table 6). Those who received more extended treatment (> three months) had significantly fewer neurological sequelae (Table 6). We also analyzed the impact of polymicrobial infections (a common occurrence in actinomycosis) on the clinical outcome of CNS actinomycosis. We did not find any difference in outcomes (mortality and relapse) in polymicrobial infection compared to monomicrobial (actinomycosis alone) (Additional file 1: s6, Table 2). These patients (with polymicrobial disease) had received more extended antibiotic therapy (202 days vs 136 days, p value 0.02) and a higher proportion of combination antibiotics. The clinical significance of polymicrobial infection and the need for antibiotics for other co-isolates is still debatable, it should be decided on case-to-case basis.

The overall mortality rate was 11% (n = 13) in all cases reported with CNS actinomycosis. We have also tried to look for residual neurological deficits in patients whose long-term follow-up (at least one year after treatment initiation) was available. In this review, 22% of patients with CNS actinomycosis had residual neurological sequelae at the end of 12 months (Table 7). We further categorized all patients into survived and non-survived groups to find out various predictors of mortality. The comparison of both groups for different predictors of mortality is shown in Table 7. Advanced age (≥ 65 years) and surgical intervention were found to be statistically significant determinant factors for the outcome (Table 7). After adjusting for other factors (age ≥ 65, immunodeficiency state), only surgical intervention was found to be an independent predictor of survival. The patients who underwent surgery and antimicrobial therapy were less likely to die compared to those treated with antimicrobial therapy alone (adjusted OR 0.14, 95% CI 0.04–0.28, p value 0.039).