Potential specific immunological indicators for stroke associated infection are partly modulated by sympathetic pathway activation

Background Evidence has led to the consideration of immunodepression after stroke as an important contributor to stroke associated infection (SAI). However, so far no specific immunological indicator has been identified for SAI, and the underlying mechanism remains poorly understood. Results SAI patients had significantly higher IL-6 and IL-10 levels and lower HLA-DR levels than no-infection patients within 48h after stroke onset. NA significantly increased IL-10 levels, reduced HLA-DR expression, and decreased IL-6 expression by increasing β-arrestin2 expression which reduced the activation of the NF-κB pathway. Propranolol reversed this effect of NA by reducing β-arrestin2 expression. Materials and Methods A systematic search for eligible clinical studies was applied to pool the differences in peripheral cytokine levels between infection and no-infection stroke patients. The underlying mechanism behind these differences was investigated in vitro by applying norepinephrine (NA) and lipopolysaccharide (LPS) to simulate sympathetic pathway activation and sepsis respectively in THP-1 cells. Propranolol was applied to determine the effect of reversing the activation of the sympathetic pathway. Immunological indicators were also detected to assess the immune activation of THP-1 cells and measurements of the expression of β-arrestin2, NF-κB, IκBα and phosphor-IκBα were performed to assess the activation of the sympathetic pathway. Conclusion IL-6, IL-10 and HLA-DR are good candidate biomarkers for SAI. The activation of the sympathetic pathway could partly account for the specific immunological alterations found in SAI patients including HLA-DR decrease and IL-10 increase, which both could be reversed by propranolol. However, the mechanism underlying IL-6 increase still needs further exploration.


Identification of literature
Systematic electronic searches of the PubMed and Web of Science citation databases were performed from their inception to September 1th, 2015. Six authors performed the data extraction. We also manually searched the conference proceedings of the World Congress of Neurology, the American Academy of Neurology and the European Federation of Neurological Societies from the year 2000-2015 for relevant clinical trials. The reference lists of all relevant studies were also manually searched for this review. The search subject terms were: infection, pneumonia, immune, immunodepression, brain infarction, cerebral infarction, cerebrovascular accident and stroke. The search was restricted to English-language literature.

Inclusion criteria
1. Clinical studies including cross-sectional, casecontrol, or cohort study were considered to be eligible for the study pool, and the placebo arms presented separately in the original articles were also included. Only the most recent publication results from the same studies were included.
2. The subjects of eligible studies should be ischemic or hemorrhagic stroke patients diagnosed through clinical and radiographic evidence with subgroup analysis of infection.
3. The criteria of infection definition originated in the included studies. Diagnosis of pneumonia required fever, pathological percussion sounds and purulent secretion, and additional infiltrates in radiographic findings. For UTI to be diagnosed, dysuria and a positive urine culture were required. Infection was defined as developing infection after at least 48h of hospitalization [1]. 4. The studies permitted assessment of the immunological blood biomarker measurements (eg, IL-6, TNF-α) after stroke in patients with and without infection and identified the population for both groups.
5. Cytokines levels were measured within 48h after stroke onset.

Exclusion criteria
1. Reviews, case reports, case series, in vitro and animal experiments.
2. Treatment arms in randomized controlled trials were excluded due to confounding from drug related changes in the immunological biomarker levels.
3. The objects of studies were not stroke patients. 4. The objects of studies were immunocompromised by chemotherapy or acquired immunodeficiency syndrome.

Quality assessment and data extraction
The randomized controlled studies were assessed in strict accordance with the Jadad scale [2], and the other studies were assessed in terms of the Newcastle-Ottawa Scale (NOS) [3]. The extracted data from each study included the following: first author's surname, publication time, study location, the total number of cerebral infarction patients, the number of patients with infections and the number without infections, Evaluation indexes: the concentration of each immunological blood biomarkers (eg, IL-6, TNF-α). Two reviewers independently carried out data extraction to avoid evaluation deviation. In case of disagreements, the two reviewers discussed in order to reach an accord.

Data analysis and statistical methods
All of the included studies after systematic review were pooled according to their immunological blood biomarkers (eg, IL-6, TNF-α). For studies with median (IQR) or mean (confidence interval) data, we estimated the standard deviation according to the RevMan user guideline [4]. Review Manager 5.2 (RevMan5.2) (The Cochrane Collaboration, UK) was applied to perform all statistical analyses.
Statistical heterogeneity among all included studies was considered to be statistically significant if P<0.05 or I 2 >50% by the assessment of Chi 2 test and inconsistency (I 2 ) statistics [5]. Subsequently, a random effects model would be applied on account of significant heterogeneity. Otherwise, a fixed-effects model was selected. As the data of blood cytokines levels were continuous values with a wide variation among eligible studies, the Standard mean differences (SMD) were calculated to estimate whether blood cytokines were significantly different between the groups of patients with and without infection. Additionally, publication bias was assessed through Egger's test together with Begg's test (STATA 12.0, Stata Corp. College Station, Texas). P<0.05 was regarded to be statistically significant.

Eligible studies
As shown in S1, the primary literature search identified 1706 potentially relevant citations. A review of the titles and abstracts eliminated 1550 articles (479 reviews, 928 basic experiments, and 143 citations with no-relevance). The remaining promising 156 articles were investigated further in greater detail through full-text assessment. From 156 eligible articles, ten had all necessary information for inclusion in the meta-analysis, with biomarker(s) levels expressed as mean ± SD or median (IQR).

IL-6
Data for IL-6 was available in seven trials of 473 patients. Because the I 2 statistic in the fixed effects model demonstrated statistically significant heterogeneity (I 2 , 96%; P<0.00001) and a wide variation in the IL-6 levels among studies was found, a random effects model was used to pool the Std. mean difference for the included studies. The pooled result demonstrated that SAI patients had a significantly higher IL-6 level than patients without infection (Std.MD 2.35; 95%CI 0.82-3.89; P=0.003).

TNF-α
Four studies (236 patients) were collected to estimate the difference in the TNF-α level between patients with and without infection. Because the I 2 statistic in the fixed effects model demonstrated statistically significant heterogeneity (I 2 , 73%; P =0.001) and a wide variation in TNF-α levels among studies was found, a random effects model was used to estimate the difference in the TNF-α levels between the two groups by the pooling the Std. mean difference. Compared to patients without infection, the TNF-α level of patients with infection was a little lower, but the difference was not statistically significant (Std.MD −0.02; 95%CI −0.50-0.47; P = 0.94).

IFN-γ
Two studies with 99 patients detected IFN-γ level in patients and compared the difference between the groups with and without infection. There was no heterogeneity between studies (I 2 = 0%, P=0.40) when IFN-γ levels were compared between patients with and without infection. Since this wide variation in the data was not corrected for potential confounders, we also estimated the difference between the IFN-γ level of the groups with and without infection by pooling the Std. mean difference in a random effects model. Patients with infection had a lower IFN-γ level than those without, but the difference was not statistically significant (Std.MD −0.21; 95%CI −0.67-0.24).

IL-10
Three studies including 190 patients were used to compare the difference in the IL-10 level between the groups with and without infection. Due to the wide variation in the data and the high heterogeneity score for the included studies (I 2 = 78%, P=0.0003), we also estimated the difference in the IL-10 level between the groups by pooling the Std. mean difference in a random effect model. Patients with infection had a significantly higher IL-10 level than patients without infection (Std.MD 1.08; 95%CI 0.09-2.06; P = 0.03).

HLA-DR
The difference in the HLA-DR level between patients with and without infection was evaluated in three studies containing with 101 patients. A random effects model was used to calculate the pool values due to a wide variation in the HLA-DR levels between studies, although significant heterogeneity was not found (P=0.89, I 2 = 0%). Compared to patients without infection, the HLA-DR level in patients with infection was significantly lower (Std.MD −0.93; 95%CI −1.35-−0.51; P <0.001).