Background

In the past decades, treatment of patients with an acute ischemic stroke (AIS) has been improved resulting with a decreased number of deaths and an increased number of patients with a long-term post-stroke disability. Accordingly, many researchers have attempted to find a predictive model for AIS outcome. Ideally, the model should include reliable variables that will be measured early after AIS onset directing the course of treatment, and consequently, reducing the time of recovery and healthcare costs. Numerous clinical variables have been employed in the prediction of short- and long-term outcomes, mostly using different scoring systems based on physical examination and radiological imaging [1,2,3].

Age, time from stroke onset to treatment, severity of neurological deficit, infarct size, and localization have been repeatedly reported as the most powerful predictors for stroke outcome [4,5,6]. Besides, different simple blood tests, such as glucose and lipid levels, that would help in AIS prognosis and outcome have been investigated for years but findings are inconsistent. In recent meta-analyses, increased risk of unfavorable AIS outcome in prediabetic and diabetic patients has been shown [7, 8]. On the other hand, studies that investigated the prognostic value of lipid levels have shown that short- and long-term outcomes of the AIS could be associated with triglycerides levels measured within 24 h of the AIS onset. However, results are inconsistent in terms of whether lower or higher triglycerides levels contribute to better outcomes [9, 10]. Similarly, contradictory findings have been reported for other lipid parameters, low-density lipoprotein cholesterol (LDL-C) and high-density lipoprotein cholesterol (HDL-C) [11].

Genome-wide association studies have revealed a certain level of inheritability for AIS depending on stroke subtype and cardiovascular phenotype [12]. Since AIS is a complex disease, many studies have investigated combined effects of hemostatic genetic variants such as F5 (rs6025), F2 (rs1799963), SERPINE1 (rs1799768), F13A1 (rs5985), and FGB (rs1800790) as well as other variants involved in atherogenesis (i.e., in lipid metabolism APOA5 (rs662779)) and blood pressure ACE (rs4340) and the AIS [13,14,15,16,17,18]. Mostly due to negative associations between genetic variants and AIS onset, there are a limited number of studies that explored relationships of candidate gene variants and stroke outcome or severity [19,20,21].

Considering all of the above, we hypothesized that gene variants that alter hemostatic, metabolic, and vascular homeostasis could be associated with AIS severity and outcome. To test the hypothesis, we investigated the association of genetic variants involved in hemostatic (F5 (rs6025), F2 (rs1799963), SERPINE1 (rs1799768), F13A1 (rs5985), and FGB (rs1800790)) and atherogenic (APOA5 (rs662779) and ACE (rs4340)) pathways as well as blood lipid levels with acute ischemic stroke severity.

Materials and methods

Subjects

This cross-sectional observational study was conducted in the University Hospital Centre Sestre milosrdnice, Zagreb, Croatia, from 2009 to 2011. The study included 250 consecutive patients who were presented to the emergency unit of the Neurology Department within 24 h of the onset of the acute ischemic stroke. All of the patients were Caucasians of the Croatian origin. A consultant neurologist made the diagnosis of the acute ischemic stroke based on neurological examination, computed tomography, and medical history. Risk factors, specified according to National Stroke Association, were collected from the medical record of each patient. Risk factors data included lifestyle (smoking status, alcohol use, and obesity), medical (atrial fibrillation, hypertension, and diabetes mellitus) and uncontrollable factors (previous cerebrovascular and cardiovascular events, age, and sex).

Stroke type was determined according to Oxfordshire Community Stroke Project clinical classification (OCSP) and Trial of ORG 10172 in Acute Stroke Treatment (TOAST). According to OCSP, subjects were grouped in four categories: total anterior circulation infarct (TACI), partial anterior circulation infarct (PACI), lacunar infarct (LACI), and posterior circulation infarct (POCI). Based on TOAST classification, patients were divided into five categories: large-artery disease (LAD), cardioembolic stroke, small-artery disease (SAD), stroke of other determined etiology, or stroke of undetermined etiology. At the admission to the emergency unit, the neurological deficit in each patient was obtained using National Institutes of Health Stroke Scale/Score (NIHSS).

Exclusion criteria were hemorrhagic stroke and transient ischemic attack as well as AIS with symptoms lasting longer than 24 h.

Sestre milosrdnice University Hospital Center ethics committee approved the study and all participants signed informed consent.

Samples

Blood samples for serum and DNA analysis were collected by venepunction in tubes without anticoagulant and with K2EDTA (BD, Franklin Lake, NJ, USA), respectively. All samples were taken the next morning after admission to neurology intensive care unit (within 24 h of admission). After spontaneous blood clotting, serum samples were centrifuged at 1500g for 10 min and immediately analyzed. Isolated DNA samples were stored at + 4 °C until genotyping.

Methods

Determination of total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), triglycerides (Tg), and glucose was done on biochemistry analyzer AU2700 (Beckman Coulter, Brea, CA, USA) using original reagents and manufacturer protocols.

DNA was extracted from EDTA blood samples using QIAamp DNA Blood Mini Kit (Qiagen, Hilden, Germany). F5, F2, SERPINE1, F13A1, APOA5, and FGB genotypes were determined by polymerase chain reaction followed by restriction fragment length polymorphism (PCR-RFLP) method, whereas ACE genotypes were done by PCR described by Saracevic et al. [22]. PCR and RFLP conditions are available in supplementary material. Within each run of samples, three controls (wt/wt; wt/mt; mt/mt) and one blank sample were tested. Accuracy of F5, F2, SERPINE1, F13A1, and ACE methods was determined by participation in external quality assurance provided by DGKL (now Reference Institute for Bioanalytics), while for APOA5 and FGB methods, 10% of all samples were retested.

Statistical analysis

Kolmogorov–Smirnov test was used to determine the normality of distribution. Age is presented as median and range (min–max). Other results are presented as mean and standard deviation, median and interquartile range, or as counts and proportions where appropriate. The difference between the observed and expected genotype frequencies for Hardy–Weinberg equilibrium calculation and the distribution of genotypes according to TOAST, OCSP, and NIHSS were tested using chi-squared test or Fisher exact test depending on the number of genotypes. Univariate logistic regression analysis was used to identify those variables (selection criterion of P < 0.05) that were subjected to forward stepwise multivariate modeling. Values < 0.05 were considered significant. Statistical analysis was performed using the MedCalc® statistical software version 17.9.1 (MedCalc, Ostend, Belgium).

Results

Demographic data and biochemistry parameters in stroke patients are presented in Table 1. According to OCSP, majority of patients had LACI (34.8%) and TACI (37.6%) types of stroke. TOAST classification revealed that 34.7% of patients had SVD type of stroke. The most common risk factor in the studied group was hypertension (80.1%) followed by FA (27.9%), DM type 2 (27.0%), and obesity (25.8%).

Table 1 Demographic data, types of stroke, and biochemistry parameters in the stroke patients (N = 250)
Full size table

Distribution of the studied genetic variants has shown no deviation from Hardy–Weinberg equation.

To assess the differences according to stroke severity, subjects were grouped using NIHSS as follows: in the first subgroup were subjects with a minor to moderately severe/severe stroke (NIHSS score 0–20), and in the second subgroup were subjects with a very severe stroke (NIHSS score 21–42) [23, 24]. Results revealed a statistically significant difference in the distribution of the APOA5 -1131T>C genotypes between subgroups (P = 0.026). Distribution of wildtype (T) and mutant (C) alleles of APOA5 variant between studied subgroups was not statistically significant (P = 0.051, data not presented). There were no significant differences in genotype distribution between subgroups for the rest of the studied variants (Table 2).

Table 2 Distribution of genotypes according to NIHSS subgroups
Full size table

Univariate logistic regression analysis is done to identify predictors for very severe stroke (NIHSS ≥ 21). Results have shown that APOA5 genotype (TC + CC) [OR 2.40 (95% CI 1.10–5.25); P = 0.034], age [OR 1.08 (95% CI 1.04–1.13); P < 0.001], and CRP concentrations [OR 1.01 (95% CI 1.00–1.02); P = 0.006] are associated with a higher risk for a very severe stroke. Further, results revealed that F5 genotype [OR 2.6 × 10–9 (95% CI 0.00–0.00); P = 0.047], triglycerides level [OR 0.55 (95% CI 0.34–0.91); P = 0.019], and obesity [OR 0.28 (95% CI 0.10–0.73); P = 0.010] are associated with a lower risk for a very severe stroke. However, in multivariate logistic regression analysis adjusted for age and sex, only APOA5 genotype (TC + CC) [OR 3.26 (95% CI 1.29–8.23); P = 0.012], age [OR 1.06 (95% CI 1.01–1.11); P = 0.018], and obesity [OR 0.25 (95% CI 0.08–0.77); P = 0.016] remained associated with a stroke severity (Table 3).

Table 3 Multivariate model adjusted for age and sex
Full size table

Further, we have tested the distribution of the studied genetic variants according to OCSP and TOAST classifications. Results have shown a statistically significant difference in the distribution of F5 Leiden genotypes between OCSP classification groups (P = 0.025), while the difference was not found for other studied genotypes (Table 4). However, we assume that this difference is a result of no GA genotypes in 2 groups. Thus, we have not performed further analysis. Similarly, there was no statistically significant difference in genotype distribution between groups according to TOAST classification (Table 5).

Table 4 Distribution of genotypes according to OCSP classification (N = 242)
Full size table
Table 5 Distribution of studied genotypes according to TOAST classification (N = 245)
Full size table

Discussion

In this study, we investigated the association of the AIS severity and conventional risk factors, such as age, sex, FA, DM, hypertension and use of tobacco, lipid parameters, CRP and F5, F2, F13A1, FGB, SERPINE1, ACE, and APOA5 variants.

The major finding in our study is that C allele carriers (TC+CC genotype) of rs662779 (APOA5 -1131T>C) variant have 3.26 times higher odds for a very severe stroke than non-carriers (TT genotype) [OR 3.26 (95 CI% 1.29–8.23); P = 0.012]. APOA5 -1131T>C variant has been associated with higher lipid levels, particularly triglycerides, that are shown to be a risk factor for atherosclerosis and stroke [25,26,27]. In fact, it has been reported that APOA5 -1131T>C decreases apolipoprotein A5 levels resulting in decreased triglycerides catabolism and consequent hypertriglyceridemia. Patients in our study had lipid parameters within levels recommended by National Cholesterol Education Program Adult Treatment Panel III and that is in accordance with the study of Can Demirdöğen et al. [28] but in contrary to the study of Maasz et al. [29]. More conflicting results revealed the studies that investigated the association between lipid levels and stroke severity. Simundic et al. have reported that there are higher serum lipid levels at admission associated with the higher AIS severity (P = 0.030) [30] while Jain et al. have found an association between lower triglycerides levels and worse NIHSS score (P = 0.004) [10]. Similar evidence has been published by Weir et al. [31]. Furthermore, the association between higher triglycerides levels and less severe clinical symptoms at admission (P = 0.014) has been found by Pikija et al. [32]. All of these studies measured lipid levels within the first 24 h of admission, as it was the case in our study. In a case–control study, Simundic et al. also reported that stroke patients overall had lower lipid levels than the healthy subjects [28]. Several other authors have described similar results. It could be explained with disturbances in lipid metabolism during the early hours of the stroke onset due to acute inflammation response and cerebral ischemia. Thus, the investigators questioned the prognostic value of lipid parameters determined within 48 h of the stroke onset. Regarding that, Perovic et al. have shown that triglycerides levels at admission are significantly lower than at discharge (P < 0.001). The authors also reported the poor functional outcome in patients that had decreased HDL-C levels measured at 48 h after the AIS onset and at the discharge [33]. Present study results relate lower triglycerides with a very severe AIS in univariate regression analysis [OR 0.55 (0.34–0.91); P = 0.019], but that association disappeared when multiple variables were entered in the regression model (P = 0.306). Still, it remains unclear why APOA5 -1131T>C variant and triglycerides have a reciprocal role in the risk of stroke severity in our study. A partial explanation could be found in previously mentioned studies and the inability to measure baseline (prior to stroke) triglycerides levels. Secondly, altered concentrations of apolipoprotein A5 may exert its role more extensively on some metabolic pathway in lipid metabolism other than that of triglycerides. Further studies are needed on a larger number of subjects (including lager groups of stroke subtypes) and different study design to resolve this question.

Well-known risk factor for AIS, obesity, has been recorded in one-quarter of all studied patients and multivariate regression analysis revealed that obese patients had a lower risk for severe stroke (P = 0.016). Recent studies showed that obese stroke patients had better long-term outcomes than non-obese patients, so-called obesity paradox in stroke. The Framingham Heart Study in a 10-year follow-up study on 782 stroke patients (87% ischemic stroke) has shown that moderately increased weight was protective for AIS when compared to normal weight subjects [HR = 0.72 (95%CI 0.53–0.99); P = 0.041] and that overweight and mildly obese patients had better 10-year outcome [34]. Further, a 30-month follow-up study from TEMPiS trial has found that overweight and obese patients have better survival and better combined outcomes of survival and non-functional status than patients with normal weight. Indeed, the authors reported that underweight patients have the higher mortality risk than the very obese patients [HR 2.42 (95% CI 1.55–3.76); P < 0.001] vs [HR 0.36 (95% CI 0.20–0.66); P < 0.001] [35]. Similar results have been shown in Yang et al. on Korean patients with AIS during 6-month follow-up [36]. On the other hand, an investigation which was done on the large scale of patients (71,617) has shown that the risk for death by stroke in obese did not significantly differ from the risk in normal weight patients, measuring death by stroke as the most severe outcome within 1 week post-stroke [37]. The results of our study suggest that obesity is favorable for non-severe stroke and probably/possibly better short-term outcome in patients with AIS. The mechanism supporting that observation has yet to be explained. Possible reasons why obesity could be protective for severe stroke could be found in a pro-inflammatory context of stroke. Firstly, TNF-α receptors from adipose tissue may bind TNF-α thus decreasing pro-inflammatory processes. Furthermore, increased lipid levels in obese patients could play a role in binding endotoxin lipoproteins decreasing inflammatory cytokines [38].

As expected, our results have shown that the older age is associated with a higher risk for severe stroke (P = 0.018). Age has been established as an unmodifiable stroke risk factor for decades and our results just underpin the fact that the older patients would more likely suffer from the serious neurological deficits and death [39,40,41].

Our study revealed a significant difference in the distribution of F5 Leiden genotypes between groups according to OCSP classification (P = 0.025). Since OCSP classification is based on clinical syndromes and is not specific for the stroke etiology, and due to the low number of F5 Leiden heterozygous subjects in OCSP classification groups, we may only speculate about F5 Leiden role.

Finally, we have failed to demonstrate the association of other studied gene variants (F5, F2, ACE, SERPINE1, F13A1, and FGB) and AIS severity. Frequency of the variants did not differ from other Caucasian studies nor between studies subgroups. We may speculate that that may be in a line with published meta-analyses suggesting that there are no independent associations between these variants and ischemic stroke despite the synergistic effect which has been found by some researchers [16, 42,43,44].

A major limitation of our study is that the stroke subtypes were unevenly distributed and thus we have not investigated the effect of APOA5 -1131T>C variant on the severity of the stroke subtypes. Besides, due to inconsistently collected data for modified Rankin score, we have not tested the association of APOA5 -1131T>C variant and short-term outcome.

The results of the presented study have brought the conclusion that C allele carriers of APOA5 -1131T>C variant are more frequent in the group of AIS patients with a very severe stroke. Furthermore, older AIS patients have more odds to suffer from a severe stroke while, conversely, obese patients would have less severe AIS.