Low ficolin-3 levels in early follow-up serum samples are associated with the severity and unfavorable outcome of acute ischemic stroke

Patients with ischemic stroke included in the present work were admitted to two centers: the Department of Neurology, University of Pecs, Hungary (39 patients:20 men and 19 women, aged 49-84 years) and the Department of Neurology, Kútvölgyi Clinical Centre, Semmelweis University, Budapest, Hungary (26 patients:10 men and 16 women, aged 58-87 years) (Table 1). The management of ischemic stroke was in accordance with the guidelines of the Stroke Council of the American Heart Association/American Stroke Association [32] None of the patients were treated by intravenous thrombolysis. Patients with stroke were enrolled upon the first occurrence of acute ischemic stroke only; all patients had neuroimaging (most of them brain MRI, but at least cranial CT). No patients had hemorrhagic infarction. All patients with definite acute clinical symptoms were enrolled regardless of etiology i.e. lacunar or territorial infarct caused by thrombosis or emboli. Exclusion criteria were infectious diseases, fever < 4 weeks before stroke, an elevated WBC, erythrocyte sedimentation rate (ESR), high-sensitivity CRP (hsCRP, cut-off value < 10 mg/L), procalcitonin on admission (cut-off value < 0.05 ng/mL), positive chest X-ray, hemorrhagic stroke defined by an acute cranial CT scan, and those who declined to participate in the study. Almost all patients had hypertension and elevated cholesterol/triglyceride levels. All patients were therefore treated for such risk factors; nevertheless the effect of such treatments on the ficolin pathway is unlikely. An evidence-based guideline [33] was followed to detect post-stroke infectious complications (in short, physical and laboratory measures including WBC, ESR, hsCRP, PCT, fever, abnormal urine, chest X-ray or positive cultures). Such complications occurred on the 4^th day as an average, and were located to the respiratory system and urinary tract even in the absence of catheterization; in addition, thrombophlebitis occurred in a single case.

At the time of admission, severity of stroke was assessed using the National Institutes of Health Stroke Scale (NIHSS) [34]. Blood samples were obtained at the time of admission (admission samples: the median time from the onset of symptoms was 7 hours in the Budapest cohort and 8.5 hours in the Pecs cohort), and 72 to 96 hours later (follow-up samples). Five and four patients in the Pecs and Budapest cohorts, respectively, developed infections. The outcome of disease was assessed with the modified Rankin scale [35].

Serum samples were also taken from 100 healthy volunteers as controls (Table 1). Additionally, 134 patients with significant carotid atherosclerosis served as controls (Table 1). In agreement with international guidelines, significant carotid atherosclerosis was defined as 70-100% stenosis of the carotid artery determined by Duplex scan sonography. The examination was indicated in the case of other vascular disorders or risk factors of vascular disorders. None of the patients had definite residual signs and no symptoms suggesting acute ischemia. Some of these patients had either peripheral arterial disease or coronary disease, and in these patients carotid Duplex scans were performed to detect asymptomatic severe carotid stenosis (as a common comorbidity). Some of the patients had non-specific symptoms (i.e. dizziness) or transient ischemic attack previously; in these patients diagnostic carotid Duplex scans were performed. Lacunar strokes defined by neuroimaging were no exclusion criteria.

Serum samples of the patients and of the controls were stored at -80°C in the Hungarian laboratories until transported on dry ice to Copenhagen.

The study was approved by the local ethics committees, and all patients and control subjects gave informed consent.

The serum concentrations of the proteins ficolin-2 [36] and ficolin-3 [37] were determined by ELISA-based methods at the Laboratory of Molecular Medicine, Department of Clinical Immunology, Rigshospitalet, Copenhagen, Denmark. Briefly, microtiter plates were coated with either monoclonal anti-ficolin-2 antibody (FCN216) or monoclonal anti-ficolin-3 antibody (FCN334) in phosphate buffered saline (PBS) overnight at 4°C. Samples diluted 1:50 or 1:640 in sample buffer (PBS-T with 1% mouse serum and bovine serum) were added in triplets to washed wells and incubated for 3 hours at 37°C. Ficolin-2 was detected with biotinylated monoclonal anti-ficolin-2 antibody (FCN219) and ficolin-3 was detected with biotinylated monoclonal anti-ficolin-3 antibody (FCN334) by incubation overnight at 4°C. Washed wells were incubated for 1 hour at 37°C with HRP-conjugated streptavidin. Plates were developed for 15 min with OPD (o-phenylenediamine) substrate solution and stopped by adding 1M H₂SO₄. The optical density was measured at 490 nm. A standard dilution series of pooled human serum were added to each assay as were a sample control. The lower limit of detection in these assays is 5 ng/ml of ficolin-2 and 1 ng/ml of ficolin-3. The inter-assay coefficient of variation (CV) is 7.1% and 4.7% and the intra-assay CV 4.3% and 3.9% for the ficolin-2 and ficolin-3 assay, respectively.

Human S100β concentrations were measured by an ELISA method (BioVendor, Modrice, Czech Republic). In our previous study, we found that the concentration of S100β was the highest 72 hours after the onset of stroke, therefore concentration was determined at this timepoint [27].

Serum CRP concentrations were measured by particle-enhanced immunturbidimetric assay, using an automated laboratory analyzer (Roche Cobas Integra 400, Basel, Switzerland).

Statistical analysis was performed using the GraphPad Prism 3.0 (GraphPad Software Inc, San Diego, CA, http://​www.​graphpad.​com) and SPSS 13.0 (SPSS Inc., Chicago, IL) software. Between-group differences were evaluated by the Mann-Whitney test. Correlations between the variables were expressed using non-parametric Spearman's correlation coefficients. The categorical variables were compared with the χ² test for trend. The association between the serum concentration of selected proteins and the outcome of stroke was calculated by multiple logistic regression, adjusted for the sex and the age of the patients. All tests were two-tailed. All data are presented as median values with the 25^th to 75^th percentiles in parentheses unless stated otherwise.

The serum levels of ficolin-2, ficolin-3 and CRP were measured in the samples obtained from 65 stroke patients on admission and 3-4 days later (follow-up samples), as well as in the sera of 100 healthy volunteers and 134 patient controls (patients with severe carotid atherosclerosis without acute stroke) (Figure 1). Compared to both healthy controls and patient controls, both ficolin-2 ficolin-3 levels were significantly lower both in the admission and follow-up sera of stroke patients. When all patients were considered, CRP levels were significantly higher in the admission samples than in the sera of healthy controls but were nearly equal to that measured in the sera of patient controls. By contrast in the follow up samples, CRP levels were significantly higher as compared to both control groups. When patients who developed infections were not considered, the difference between stroke patients and controls became non-significant (data not shown). As for the two controls groups, all the three variables had significantly higher concentration in the sera of patient controls than in the healthy controls, although the difference in the ficolin-3 levels was small.

Ficolin-3 concentrations measured in the follow-up samples but not in the admission samples exhibited a significant, negative correlation with indirect measures of stroke severity i.e. the NIH score determined on admission (Figure 2, panel A). Patients were divided into two groups in a similar manner to Foerch 2005 [26]; those with a NIH scale of < 16 with relatively good expected outcome and those with NIH scale of ≥ 16 with poor expected outcome, and the ficolin-3 levels were compared accordingly. There were significantly (p = 0.017) lower ficolin-3 levels in the former than in the latter group. By contrast, no significant differences in the ficolin-2 levels (p = 0.309) were found between the two groups (data not shown).

In addition, we found significant negative correlation between ficolin-3 concentrations and the S100β level measured in the follow-up samples but not in the admission samples (Figure 2, panel B). The levels of ficolin-2 did not correlate with the S10B concentrations (data not shown).

CRP concentrations in follow-up samples were significantly higher in patients with high (≥ 16) NIH score (Figure 2, Panel C), but did not significantly correlate with the S100β levels (Figure 2, panel D).

The levels of the ficolins and CRP were related to the outcome of the disease, as assessed by the modified Rankin scale (Figure 3). When patients were divided according to unfavorable (3 to 6) and favorable (0 to 2) modified Rankin scores, ficolin-3 levels were lower in the former group, supporting the association with an unfavorable outcome. The difference was significant only in the follow-up samples, while almost significant in the admission samples (Figure 3, panels A and B). When the 9 patients, who developed infectious complications were excluded, CRP levels both in admission and follow up samples were significantly higher in the patient group with unfavorable compared to favorable outcome (Figure 3, panels C and D).

We confirmed these data by performing a multiple logistic regression analysis. Unfavorable (modified Rankin scale 3 to 6) vs. favorable (modified Rankin scale 0 to 2) outcome was regarded as a dependent variable, whereas ficolin-3 levels, CRP levels, age and sex were considered as independent variables (Table 2). Since both ficolin-3 and CRP levels were included in the analysis, those 9 patients who developed infectious complications were excluded.

Both ficolin-3 and CRP levels measured in the follow up samples were significantly associated with the outcome of the disease: lower ficolin-3 and higher CRP values were found in the unfavorable compared to the favorable outcome group. Similar but only, marginally significant (ficolin-3) or weakly significant (CRP) associations were found when the admission samples were analyzed.

Next, in order to assess the strength of association between the low ficolin-3 and high CRP levels on the one hand and the unfavorable outcome of the disease on the other hand, we repeated the analysis as above in the follow-up samples by including ficolin-3 and CRP levels as low/high values. Ficolin-3 levels below or equal to the median (16 μg/ml for both the admission and follow up samples) were considered low, while those above the median value were considered high. CRP levels above median (7.7 mg/L) were considered high (Table 3). In the analysis, adjusted for sex and age of the patients, both the low ficolin-3 and the high CRP levels significantly predicted an unfavorable outcome, with odds ratios of 5.6 and 3.9, respectively.

Finally, we assessed the relationship between the baseline NIH scale as an indirect measure of the severity of the stroke, the concentration of the S100β in the follow up samples as an indicator of the infarct size, and the outcome of the disease assessed by the modified Rankin scale (Figure 4). Both measures exhibited highly significant correlation to the outcome: patients with high baseline NIHSS scale had much worse outcome than those with low NIH scale, and patients with unfavorable outcome had higher serum S100β concentrations at 72 hours than those with a favorable outcome.