Manifestations, complications, and treatment of neurobrucellosis: a systematic review and meta-analysis

Abstract Purpose Central nervous system involvement by Brucella species is the most morbid form of brucellosis disease. Studies on neurobrucellosis are scarce and limited to case reports and series. Brucella is unable to infect or harm neurons without the assistance of monocytes. This raises the question of whether ceftriaxone-based regimens are effective. Methods The primary aim of this study was to identify, evaluate, and summarize the findings of all relevant individual studies in the past 30 years to help better understand the disease. To achieve this, a broad systematic search was undertaken to identify all relevant records. Epidemiological and clinical features of the disease were assessed by the pooled analysis of descriptive studies. Through a meta-analysis, the treatment period duration was compared between the ceftriaxone-based and oral regimens using Standardized mean differences to measure effect size. Results 448 patients were included in the Meta-analyses from 5 studies. A moderate positive effect was found for ceftriaxone-based regimens over oral treatments, and there was a significant difference between these two groups (SMD 0.428, 95% CI −0.63 to −0.22, I 2 = 37.64). Neurobrucellosis has a different clinical picture in pediatric patients. The disease is less chronic in children. Fever, nausea and vomiting, fatigue, and abdominal pain were significantly more prevalent symptoms in children, and Convulsions, ascites, sensorineural hearing loss, and papilledema were significantly more prevalent signs in children than adults. Conclusion It is recommended to initiate the treatment of neurobrucellosis with IV ceftriaxone therapy in combination with oral therapy.


Introduction
Brucellosis is a common zoonosis with minimal mortality [1].However brucellosis has a worldwide incidence, it is endemic in many parts of the world, including the Middle East and the Mediterranean basin [1,2].Central nervous system (CNS) involvement by Brucella species is uncommon and occurs in less than 5% of brucellosis patients [3].Neurobrucellosis is a morbid form of brucellosis disease, presents with inflammatory signs and symptoms [4].Although neurobrucellosis could present with various neurological syndromes, the major clinical picture is consistent with meningitis or meningoencephalitis presentation [5,6].Brucella enters the CNS using the Trojan horse mechanism mediated by brucella-infected monocytes to traverse the blood-brain barrier (BBB) and gain access to the brain parenchyma [7].Brucella-infected monocytes also serve as a source of de novo infection for other cell types such as microglia and astrocytes [7].
Recent experimental findings elucidate that Brucella infection causes deregulation of astrocytes and microglia that creates a microenvironment in the CNS in which secretion of pro-inflammatory mediators leads to destabilization of the glial structure, the damage of the BBB, and neuronal loss [4].The use of parenteral ceftriaxone as an initial component of therapy followed by long-term sequential oral antibiotics seems to reduce the course of the treatment and is effective in the management of the disease [6].Due to the rare nature of the disease, it is not feasible to conduct clinical studies with an appropriate sample size, and data in the Literature are generally restricted to descriptive or retrospective studies [6,[8][9][10][11].
Thus this study aimed to identify, evaluate, and summarize the findings of all relevant individual studies to help better understand the course of the disease and raise the knowledge about both epidemiologic and clinical aspects of the disease.

Protocol and registration
This systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [12], and a PRISMA checklist was completed.This systematic review was registered in PRoSPERo (CRD42020167761).

PICO question
PICo (Population, Intervention, Comparison and outcome) question was relevant to assess: Population (such as patients diagnosed of neurobrucellosis considering inclusion criteria), Intervention (Ceftriaxone with supplementation of standard oral regimens), Comparison (standard oral antibiotic regimens) and outcome (Therapeutic effects of ceftriaxone supplemented regimen comparing to oral regimens).
The formulation of the PICo question was as follows: what are the relative therapeutic effects of different classes of antibiotics with regards to treating neurobrucellosis?

Search strategy
A systematic search of literature published between January 1, 1990, and November 25, 2021, was conducted in the following electronic databases: MEDLINE, Web of Science, Scopus, and Cochrane Library.Terms referring to Brucellosis were combined with terms referring to CNS involvement (using both MeSH terms and text words) and we also checked the references of four narrative reviews on neurobrucellosis to supplement our searching [13][14][15][16].Details of the searches and exact search strings are provided in the supplementary data.All the identified records were pooled and duplicate records removed before screening using Endnote X9 citation manager software.Then titles and abstracts were appraised by two independent screeners and irrelevant records were excluded.The full text of potentially relevant papers was retrieved if possible.The full text was independently reviewed by two authors (A.T. and M.S.) and any disagreements were discussed to reach a consensus.

Eligibility criteria
The patients were included in the study was merely among those who had consistent clinical findings of brucellosis and a serum agglutinin titer of ≥1:160 in serum tube agglutination (STA) or a positive blood culture who had symptoms and signs of neurological disease not otherwise explainable concomitant with the presence of any one of the following criteria: (1) isolation of Brucella species from cerebrospinal fluid (CSF); (2) positive enzyme-linked immunosorbent assay (ELYSA) or tube agglutination with coombs test using a cutoff ≥ 1:8 for Brucella antibodies in CSF; (3) presence of inflammatory alteration of CSF including lymphocytosis, increased protein and decreased glucose levels; (4) findings in cranial magnetic resonance imaging (MRI) or computed tomography (CT) scan in favor of neurobrucellosis.Hence there is no consensus on diagnostic antibody titer in CSF for neurobrucellosis, we set the cutoff ≥1:8 as it is highly sensitive and specific [17].
only patients were included that were followed for a minimum of 3 months after they stopped antibiotic treatment and their final status was clearly mentioned.There was no restriction on the severity of symptoms or complications, no restriction on sample size, and only published studies were included.only studies published in English were eligible for inclusion.
Epidemiological features of the disease were assessed by the pooled analysis of descriptive studies.Pooled analyses can only be conducted if the included studies used the same study design [18] therefore we only included studies for the pooled analysis in which the full course of the disease and outcome was well-defined for each patient.Relative therapeutic effects of different classes of antibiotics were assessed by the Meta-analysis on retrospective studies.

Data extraction
For data extraction, two authors (A.T. and M.S.) independently reviewed all eligible articles and extracted the following data using electronic forms designed exclusively for extraction of data from descriptive studies by oracle Database 10 g and oracle Forms 6i.Basic demographic data and details of signs and symptoms, laboratory findings, complications, radiologic findings, and the treatment interventions effect, were collected through data extraction.Conflicts were resolved by group consensus, with M.S. making the final decision for inclusion.

Assessment of risk of bias
The Joanna Briggs Institute (JBI) Critical Appraisal Checklist for Case Reports [19] was used for critical appraisal of each case included in the pooled analysis.JBI Critical appraisal tools have been developed by the JBI and collaborators and approved by the JBI Scientific Committee following extensive peer review [19].The purpose of this tool is to assess the methodological quality of the study and address the possibility of bias in its design.The GRACE checklist [20] was used for the critical appraisal of retrospective studies included in the Meta-analysis.The risk of bias was assessed by two independent reviewers (A.T. and M.S.) and any disagreements were resolved by group consensus.

Data synthesis and statistical analysis
The primary outcome in the Meta-analysis was to compare the treatment period duration between ceftriaxone-based regimens and oral treatment.The effect size was measured by both standardized mean difference (SMD) and Hedges' g which is a corrected effect size that tells you how much one intervention group differs from another [21].Heterogeneity across the included studies for the Meta-analysis was evaluated using the I 2 statistic values, which describes the percentage of variation across studies that is due to heterogeneity rather than chance [21].Heterogeneity is considered substantial when I 2 >50%.Because the heterogeneity statistic I 2 can be biased in small Meta-analyses, we analyzed data using both fixed effect and random effect models [22].
Publication bias was subjectively assessed using funnel plots of precision for asymmetry and the classic fail-safe N test was conducted to assess the risk of publication bias quantitatively.All the analyses for this part were done using the Comprehensive Meta-Analysis (CMA) Software [23].
To synthesize data for pooled analyses of 274 neurobrucellosis cases extracted from descriptive articles, descriptive statistics were used to summarize the characteristics of the data set and were presented as frequencies, percentages for categorical variables, and mean ± standard deviation and median (min-max) for continuous variables.The chi-square test was used to compare the categorical variables.The Shapiro-Wilk test was used to evaluate the distribution of variables, and then the independent-samples T-test and the Mann-Whitney u test were used for the comparisons of continuous variables based on the results of the tests for normality.In the assessment of the relations between the variables, Pearson or Spearman correlation coefficients were calculated based on variable types.A p-value of less than 0.05 was considered to be statistically significant.Statistical analysis was conducted using IBM SPSS statistics 22 (IBM Corp., Armonk, NY, uSA) to synthesize data for the pooled analyses.Plotly open source library for python was used to plot data.

Results
As depicted in Figure 1, 2784 records were identified by applying search strategy.After screening for duplicates and irrelevant studies, 448 records were sought for retrieval, and finally, after applying inclusion criteria, 142 studies were included for pooled analysis including 274 cases, and 4 studies were included for the Meta-analysis through electronic searches plus the pooled analysis of the therapeutic effects of ceftriaxone-based regimens conducted in this study, finally, 5 studies entered the Meta-analysis including 448 cases.The risk of bias for these articles is reported in Supplementary data, Table 1.

Epidemiologic characteristics
Extracting the data from 142 studies, resulted in finding 274 neurobrucellosis cases, meeting the inclusion criteria.161 males and 113 females enrolled (M/F, 1.4:1).The mean age of neurobrucellosis patients was 31 ± 17.82, median = 29(0 to 82).Two maternal transmissions were reported in infants [24][25][26].Patients were categorized into two age groups: 41 pediatric patients were defined by age ≤12 years and 233 adolescent and adult patients were defined by age >12 years, and the rationale behind this classification was the different therapeutic protocols in this age group because of the doxycycline side effects in children.To compare the onset of disease between age groups, the duration of symptoms was tested for normality, and the null hypothesis for normal distribution was rejected (p ≤ .001).Log-normal distribution of the onset of disease was suspected by examining quantile-quantile plots, therefore LN (onset of disease) was tested for normality, and failed to reject the null hypothesis for normal distribution (p = .095).It seems neurobrucellosis has log-normal distribution for the duration of symptoms (see supplementary data, Figure 1).The mean duration of symptoms for pediatric patients was 73.34 ± 111.96, median = 30(2 to 430) days from onset of symptoms, and the mean duration of symptoms for adults was 121.87 ± 165.20, median = 60(1 to 850), and the distributions in the two groups differed significantly (Mann-Whitney U = 1574, N P =32, N A =144, p = 0.005).The Mood's median test was also rejected for the assumption of the same medians between two age groups for the onset of disease (p = 0.013), (see Figure 2).The frequency table for the nationality of all included cases is provided in Supplementary data, Table 2, and plotted in Figure 3 note that this figure doesn't represent the worldwide distribution of the disease and only represent the included articles distribution in this study around the world.

Laboratory findings
195 of 274 (71%) included cases had examined for CSF agglutination test which 179 (92%) of these cases had CSF titer ≥1:8.Among 95 cases with negative titer or no CSF agglutination assessment, 39 cases had positive CSF culture and 62 cases had inflammatory alteration of CSF including lymphocytosis and increased protein and decreased glucose levels, and 58 cases had positive MRI findings in favor of neurobrucellosis.CSF parameters descriptive statistics are summarized in Table 3.
91.8% of cases had CSF protein level >45 mg/dl, and 62.1% cases had CSF glucose level< 40 mg/dl, and 88.0% of cases had CSF leukocyte count >10 cells, and 92.8% of cases had CSF lymphocytic predominance >50%.The correlations between different CSF parameters are summarized in Table 4 (Before the analysis of titration data, logarithmic transformation was performed based on log10).The frequency table of CSF titration is provided in Supplementary data, Table 3. CSF culture was done in 162 of 274 (59%) included cases and only 62 (38%) of these cultures were positive for Brucella species.
In pediatric patients, 11 patients receiving ceftriaxone-based regimens with the mean duration of 6.45 ± 5.05, median = 6(1.5 to 18), had no significant difference in treatment period with 23 patients receiving oral treatment with the mean duration of 4.02 ± 2.04, median = 3(1.5 to 10), and by the mean difference of 2.43(95% CI −1.02 to 5.89, p = 0.151) failed to reject the null hypothesis.There was no meaningful difference in the treatment period between patients receiving less than one month of IV ceftriaxone and those receiving at least one month of IV ceftriaxone in adults, by the mean difference of −0.52 (95% CI −1.64 to 0.61, p = 0.361).
The relations between treatment period and patient characteristics were assessed by Pearson correlation coefficients for continuous variables and Spearman correlation coefficients for discrete variables, in order to find factors that prolong the treatment period, summary statistics are presented in Table 9. Duration of symptoms (r = 0.296, p = 0.028), CSF Protein levels (r = 0.227, p = 0.042), VIII CN involvement (r = 0.215, p = 0.039) upper motor neuron syndrome (r = 0.215, p = 0.038), and positive imaging (r = 0.322, p = 0.002) were correlated to prolonged treatment period in adults receiving oral therapy.These factors had no meaningful correlation with prolonged treatment periods in adults receiving ceftriaxone-based therapy.

Meta-analysis
The results of our meta-analyses are shown in Tables 10 and 11.The primary outcome in the Meta-analysis was to compare the treatment duration between ceftriaxone-based regimens and oral treatment.4 retrospective studies were identified through database searching plus the pooled analysis from this study met the inclusion criteria to enter the Meta-analysis.448 patients were included in this analysis.A moderate positive effect (SMD = 0.428, 95% CI −0.63 to −0.22, p < 0.001) was found for ceftriaxone-based regimens over oral treatment, and there was a significant difference between these 2 groups.Although heterogeneity was not substantial (I 2 =37.64, p = 0.170), a random effect model was also considered because of bias in  the heterogeneity of small Meta-analyses.The forest plots for both SMD and Hedges' g measure of effect, are provided in supplementary data, Figure 2 and Figure 3. Publication bias was subjectively assessed for asymmetry in the funnel plot of precision (see Figure 4).The bottom of the plot (which smaller studies appear towards there) shows a higher concentration of studies on the right side of the effect size, this would reflect the fact that there is a publication bias in this study.The classic fail-safe N test was conducted to determine how many missing non-significant studies would nullify the observed effect if we locate and include them in the analysis.The fail-safe N is 4, this means we would need to locate and include 4 null studies so that the p-value exceeds 0.05 and therefore the observed effect becomes insignificant.This shows that the possibility of publication bias is considerable in this study.

Discussion
CNS involvement by Brucella species is the most morbid form of brucellosis disease [4].The inflammatory pathogenic mechanisms involved in damage to CNS have been recently elucidated by both in vivo and in vitro evidence [28].The Trojan horse mechanism of Brucella traversing the BBB by Brucella-infected monocytes and then de nonvo infection of microglia and astrocytes has been described by Miraglia et al., 2018 [7].In this study, 92.8% of cases had CSF lymphocytic predominance which is evidence that the integrity of the BBB is altered by neurobrucellosis since the infection of CNS activates the BBB and modifies its integrity, and is mediated by astrocytes and microglia with immune basis [28].Brucella is recognized by glial cells throughout TLR2, NLRP3, and AIM2 which induces astrogliosis and BBB endothelial cells activation, so this leads to an increase in the permeability of BBB to monocytes and neutrophils [4,[29][30][31].Brucella is unable to infect or harm primary cultures of mouse neurons However, when neurons were co-cultured with microglia, Brucella caused a significant neuronal loss [32].This neuronal loss is not due to apoptosis, rather the microglial primary phagocytosis of live neurons causes neuronal death, since blocking microglial phagocytosis prevented neuronal loss without increasing the number of apoptotic neurons, thus phagocytosis was the cause of death and not its consequence [32].
Although it seems immune mediators play an important role in the pathophysiology of the disease, there is not enough evidence in favor or against the effect of steroids with regards to treating neurobrucellosis.In cases misdiagnosed as autoimmune diseases like SLE or MS, receiving methylprednisolone without any proper antibiotic therapy led to the progression of the disease after about one week of initial improvement of the symptoms, therefore we recommend avoiding the initiation of the corticosteroids before proper antibiotic therapy [33,34].81% of included patients were from Turkey, Saudi Arabia, India, and Iran which indicates although neurobrucellosis is rare, it is still an important health issue in these four countries and they need preventive health care programs to decrease the burden of the disease since the eradication of brucellosis in humans is only possible by the control of the disease in animals.
Neurobrucellosis has a different clinical picture in pediatric patients.The disease is less chronic in children, and the median of the course of the symptoms is 30 days shorter in this population.Fever, nausea and vomiting, fatigue, and abdominal pain were more prevalent in children.Also, convulsions, ascites, sensorineural hearing loss, and papilledema are more prevalent signs in children.one of the most important clinical pictures in children was ventriculoperitoneal shunt infection (17.1%), especially in patients with a history of hydrocephalus.This infection may lead to a peritoneal infection presented with abdominal pain.Removing VP-shunt was the key factor to treat these patients.Temporary placement of external ventricular drains could help patients with high intracranial pressure to complete the treatment course.After completion of the treatment and recovery, the VP-shunt could be placed again [35][36][37][38][39][40].VIII CN involvement was significantly less common in children with only 2 cases observed.
In the assessment of the relations between treatment period and patients characteristics, there was a meaningful positive correlation (r) between treatment period and parameters like duration of symptoms, CSF protein levels, positive imaging, VIII CN involvement, and upper motor neuron syndrome in patients treated with oral antibiotics which means these parameters are correlated with a prolonged course of treatment.But the correlations between these parameters and treatment period were no longer meaningful in patients who received ceftriaxone-based regimens.This indicates that ceftriaxone-based regimens neutralize the effects of treatment prolonging factors and this is another evidence for the superiority of ceftriaxone-based regimens over oral treatments.
The results of the pooled analysis were in favor of 1.14 (95% CI 1.99 to 0.3) month shorter treatment period with ceftriaxone-based regimens in adults.There was no meaningful difference in the treatment period between patients receiving less than one month of IV ceftriaxone and those receiving at least one month of IV ceftriaxone, therefore it is recommended to initiate the treatment of neurobrucellosis with at least 14 days of IV ceftriaxone therapy in combination with oral therapy.The overall treatment period is dependent on the patient's characteristics, especially CSF parameters but 6 months of treatment seems to be sufficient as a general recommendation, although the treatment should be continued until the normalization of CSF parameters.There was not enough data to conclude the superiority of any regimens in children and results were not meaningful.
The results of the meta-analyses also showed a moderate positive effect (SMD 0.428, 95% CI −0.63 to −0.22) of ceftriaxone-based regimens in shortening the treatment period over oral therapy in adults.

Conclusion
unfortunately, studies on neurobrucellosis are scarce and limited to case reports or series with limited cases.thus the risk of bias in this study is considerable and at least 4 more studies on the treatment of neurobrucellosis are needed to decrease the risk of publication bias.By the way, the summary of the past 32 years of research on neurobrucellosis favors the superiority of parenteral ceftriaxone treatment in combination with doxycycline and rifampin compared to oral regimens.

Figure 1 .
Figure 1.PrIsMa flow diagram.Wos, web of science; nlM, national library of Medicine.

Figure 2 .
Figure 2. Violin plot mixed with box plot of onset of disease for each age group.Mood's Median test: null Hypothesis: the Median of duration are the same across categories of age group is rejected, total N = 176, Median = 60, test statistics = 7.234, df = 1, asymptotic sig = 0.007, Yate's continuity correction: X2 = 6.214, p = 0.013.

Figure 3 .
Figure 3. Distribution of included neurobrucellosis cases around the world based on the cases included from the past 30 years for the pooled analysis.

Table 8 .
therapeutic failure.age/sex onset(days) imaging findings CSF titer csf culture csf profile treatment regimen