The natural tissue plasminogen activator inhibitor neuroserpin and acute ischaemic stroke outcome

Summary Neuroserpin is a brain-derived natural inhibitor of tissue plasminogen activator (tPA) that has shown neuroprotective effects in animal models of brain ischaemia. Our aim was to investigate the association of neuroserpin levels in blood with functional outcome in patients with acute ischaemic stroke. Due to the potential effect of tPA treatment interfering on neuroserpin levels, we studied two different cohorts: 129 patients not treated with tPA and 80 patients treated with intravenous tPA within 3 hours (h) from symptoms onset. Neuroserpin levels were measured by ELISA. Good functional outcome at three months was defined as Rankin scale score ≤2. In the two cohorts, serum neuroserpin levels on admission were significantly higher than at 24 h, 72 h and in healthy subjects. In non tPA-treated patients, neuroserpin levels decrease at 24 h, but not levels at baseline, were associated with good outcome (for each quartile decrease, adjusted odds ratio [OR] 15.0; 95% confidence interval [CI], 3.5 to 66). In the tPA-treated cohort, high neuroserpin levels before tPA bolus had the stronger effect on favourable outcome (for each quartile, OR 13.5; 95%CI, 3.9 to 47). Furthermore, for each quartile in neuroserpin levels before tPA bolus there was a 80% (95%CI, 48 to 92) reduction in the probability of subsequent parenchymal haematoma. In summary, high serum neuroserpin levels before intravenous tPA and neuroserpin levels decrease at 24 h after ischaemic stroke, independently of tPA treatment, are associated with good functional outcome. These findings support the concept that neuroserpin might play an important role during cerebral ischaemia.


Summary
Neuroserpin is a brain-derived natural inhibitor of tissue plasminogen activator (tPA) that has shown neuroprotective effects in animal models of brain ischaemia. Our aim was to investigate the association of neuroserpin levels in blood with functional outcome in patients with acute ischaemic stroke. Due to the potential effect of tPA treatment interfering on neuroserpin levels, we studied two different cohorts: 129 patients not treated with tPA and 80 patients treated with intravenous tPA within 3 hours (h) from symptoms onset. Neuroserpin levels were measured by ELISA. Good functional outcome at three months was defined as Rankin scale score ≤2. In the two cohorts, serum neuroserpin levels on admission were significantly higher than at 24 h, 72 h and in healthy subjects. In non tPA-treated patients, neuroserpin levels decrease at 24 h, but not levels at baseline, were associated with good Introduction Neuroserpin belongs to the serin proteases inhibitors (serpins) superfamily. It is a natural inhibitor of tissue plasminogen activator (tPA), located almost exclusively in the central nervous system (CNS) (1). Intravenous tPA has demonstrated to be a safe and effective treatment for acute ischaemic stroke within 4.5 hours (h) of symptoms onset (2)(3)(4). This beneficial effect is linked to its primary function as a thrombolytic enzyme in blood. However, tPA is also expressed in CNS and has neurotoxic effects in the neurovascular unit in response to cerebral ischaemia (5,6). Hence, it has been proposed that the local increase of neuroserpin after ischaemia may represent a natural protective response to high tPA levels (7). In line with this hypothesis, it has been demonstrated that neuroserpin attenuates the effect of tPA on NMDA receptor-mediated neuronal death after experimental ischaemia (8).
It has been found that both neuroserpin and tPA expression increase after brain ischaemia in rats. Immunohistochemical analysis has shown elevated neuroserpin staining in neurons surrounding the ischaemic core as early as six hours after stroke, which reaches a peak value at 48 h and remains elevated for one week (7). Additionally, transgenic mice overexpressing neuroserpin in the nervous system show a reduction in brain tPA activity, smaller lesion volume and attenuated microglial response after occlusion of the middle cerebral artery (9), whereas neuroserpin knockout mice have unchanged tPA activity (10). Importantly, Zhang et al. found that neuroserpin may increase the therapeutic window for thrombolytic treatment with tPA in a rat model of embolic stroke (11).
Despite the benefits of neuroserpin in experimental brain ischaemia models, its clinical application for acute ischaemic stroke had not previously been tested in humans. Since neuroserpin is involved in all the aforementioned neuroprotective processes, our aim was to study the association of serum neuroserpin levels and clinical outcome in patients with acute ischaemic stroke.

Study population and patients characteristics
We studied two different cohorts of patients with acute ischaemic stroke. The first group included patients who did not receive tPA treatment and the second group included patients treated with intravenous tPA. These cohorts were separately analysed due to the potential interaction between tPA treatment and neuroserpin levels on clinical outcome. Sample size was calculated using EPIDAT software (http://dxsp.sergas.es/ApliEdatos/Epidat/cas/default. asp) assuming an alpha and beta errors of 0.05 and 0.2, respectively.
The first cohort was selected from 190 consecutive patients with a first-ever ischaemic stroke admitted in a single center within 12 h of symptoms onset who were previously independent for their daily living activities. Patients with cancer (n=4), chronic inflammatory (n=5), severe hepatic (n=4), renal (n=2), haematological (n=2), or infectious diseases in the 15 days prior to inclusion (n=5) were excluded. Twenty-six patients who received thrombolytic treatment, seven patients who did not agree to participate in the study and six patients lost during the follow-up were also excluded, thus, a total of 129 patients were finally included in the study.
The second cohort consisted of 80 patients with acute ischaemic stroke treated with intravenous tPA (0.9 mg/kg for 1 h) within 3 h of symptoms in three university hospitals into a common research project on biomarkers of reperfusion injury. All patients had middle cerebral artery (MCA) occlusion on prebolus transcranial duplex examination (TCDx) and were evaluated by cranial CT, neurological and functional scales during a follow-up of 90 days. Patients with prior disability and known infectious, inflammatory or cancer diseases at the time of treatment were excluded.
This research was carried out in accordance with the Declaration of Helsinki of the World Medical Association (2000) and approved by the Ethics Committee of the participating hospitals. Informed consent was obtained from each patient or their relatives after full explanation of the procedures.

Clinical variables
All patients were admitted to an acute stroke unit and treated following the European Stroke Organisation guidelines (12). Medical history recording potential vascular risk factors, blood and coagulation tests, 12-lead electrocardiogram, chest x-ray, and carotid ultrasonography were performed on admission. Stroke subtype was classified according to the TOAST criteria (13) and stroke severity was assessed by a certified neurologist using the National Institute of Health Stroke Scale (NIHSS) on admission (or before tPA treatment when applicable), at 24 ± 6 h and 48 ± 6 h. NIHSS scores range from 0 to 42, with higher scores indicating increasing severity (14). Early neurological deterioration (END) was diagnosed in those patients who worsened four points or more on NIHSS score within the first 48 h. Functional outcome was evaluated at three months using the modified Rankin Scale (mRS) by an internationally certified neurologist with broad experience in the field. This scale classifies patients in seven categories: 0, no symptoms at all; 1, No significant disability despite some symptoms; able to carry out all usual activities; 2, Slight disability; unable to carry out all previous activities, but able to look after own affairs without assistance; 3, Moderate disability; requiring some help, but able to walk without assistance; 4, Moderately severe disability; unable to walk and attend to own bodily needs without assistance; 5, Severe disability; bedridden, incontinent and requiring constant nursing care and attention; 6, Dead (15).Good outcome was defined as a mRS score ≤2, thus including independent patients; and "poor outcome" was defined as a mRS score from 3 to 6, including dependent or dead patients.

Neuroimaging variables
CT scans were carried out on admission and between days 4 and 7. Patients who received tPA were studied before infusion and at 24-36 h after treatment. Early CT signs of infarction were evaluated on admission, and hypodensity volume and hemorrhagic transformation (HT) were assessed on the follow-up computed tomography (CT). HT was classified in haemorrhagic infarction type 1 (HI1) and 2 (HI2), and parenchymal haematoma type 1 (PH1), type 2 (PH2), and remote (rPH) according to ECASS II definition (16). HT was defined as symptomatic when it was associated with END. Hypodensity volume was calculated on the follow-up CT by using the formula 0.5xaxbxc, where a and b are the largest perpendicular diameters, and c is the number of 1-cm thick sections that contain the lesion. Brain edema was classified as grade I if effacement of the cortical sulci, grade II if ventricular asymmetry, and grade III if shifting of the structures of the midline were observed. Malignant edema was diagnosed if midline shift and compression of the basal cisterns were associated with a decrease in the level of consciousness compared to the baseline clinical status on admission. Severe cerebral edema grouped patients with grade III or malignant brain edema. Neuroradiologists who had no knowledge of the patients' clinical and laboratory results made all neuroimaging evaluations.

Transcranial duplex ultrasound assessment
For the tPA group, TCDx assessments were performed using a General Electric Vivid 7 Pro or an Aplio 50 Toshiba SSA-700 devices equipped with multi-frequency transducers on admission and at 2 h after tPA bolus by investigators with extensive experience in TCDx monitoring. MCA occlusion was diagnosed in presence of no flow or minimal flow (TIBI 0 or 1) or blunted or dampened signals (TIBI 2 or 3) in the symptomatic artery (17). MCA recanalisation was diagnosed as complete in presence of stenotic o normal flow (TIBI 4 or 5), or as partial if baseline arterial pattern improved one grade or more on TIBI grade in the follow up TCDx examination (17). Partial or complete MCA recanalisation at 2 h was considered to be early MCA recanalisation. Continuous monitoring of the symptomatic MCA was maintained during the 1-h tPA infusion. Echo-contrast was used when there was an insufficient temporal bone window.

Laboratory determinations
Biochemistry, haematology and coagulation tests were assessed in the central laboratory of each participating center. Blood samples, drawn from all patients on admission (before tPA treatment when applicable), at 24 ± 6 and 72 ± 24 h, were collected in glass chemistry test tubes, centrifuged at 3,000xg for 10 minutes, and the serum was immediately frozen and stored at -80ºC until analysis. Serum neuroserpin levels were determined in a laboratory blind to the clinical outcome and neuroimaging findings. For neuroserpin quantification, a sandwich ELISA was performed as described previously (18,19). Each sample was assayed in duplicate and intraassay coefficients of variation were always <15%. Clinical investigators were unaware of the laboratory results until the end of the study, once the database was closed. The normal range of serum neuroserpin was defined as the percentiles 5-95 of values in 22 healthy subjects, and corresponded to 12-37 ng/ml. Results from Western blot studies of serum samples suggested that most neuroserpin was non tPA-binded (see Supplementary information available online at www.thrombosis-online.com).
Active matrix metalloproteinase -9 (MMP-9) was determined in patients treated with tPA using an activity assay (GE Healthcare) in serum samples at baseline and at 24 ± 6 h after treatment.

Outcome variables
In the two cohorts, the primary endpoint was good functional outcome at three months. In the non-tPA group, secondary variables   were hypodensity volume on the follow-up CT and END, whereas in the tPA group they were parenchymal haematoma, severe cerebral edema and early MCA recanalisation.

Statistical analysis
The results are expressed as percentages for categorical variables and as mean (SD) or median [quartiles] for the continuous variables as appropriate. Proportions were compared using the chisquare or Fisher test, while the continuous variables between groups were compared with the Student's t-or the Mann-Whitney tests. Spearman's or Pearson's analyses were used for bivariate correlations. ANOVA was used to analyse the relationship between stroke subtypes and neuroserpin levels. Mixed analysis of variance (MANOVA) was used to study the effect of time, and groups (good and poor outcome) by time (baseline to 72 h) interactions on neuroserpin levels. The MANOVA analysis did not include the healthy controls as this group was only analysed at baseline. Neuroserpin was categorised in quartiles to calculate the adjusted odds ratios of outcome variables for each quartile increase. When the frequency of the outcome event did not show a progressive increase for each neuroserpin quartile, the odds ratio (OR) for each quartile was calculated taken the first one as the reference value. The influence of neuroserpin on lesion volume was assessed by multiple lineal regression models. Logistic regression and lineal models were ad-justed for the baseline variables related to each outcome variable and neuroserpin in the bivariate analyses. All significant variables were forced to enter in the models (enter approach). Results were expressed as adjusted OR with the corresponding 95% confidence intervals (95%CI). The statistical analysis was conducted using SPSS 15 (SPSS Inc., Chicago, IL, USA) software.

Results
In  [18,30] ng/ml). ǠTables 1 and 2 show baseline clinical characteristics and neuroserpin levels in the two cohorts. Serum neuroserpin levels were not different by sex or stroke subtype and did not correlate with age. No significant associations were found between neuroserpin and baseline variables in the non tPA-treated cohort. In the tPA treated cohort, baseline neuroserpin levels showed a mild negative correlation with NIHSS score (coeff.= -0.386), serum glucose (coeff.= -0.333), systolic (coeff.= -0.301) and diastolic (coeff.= -0.303) blood pressure (all p-values <0.01), and were higher in patients with distal MCA occlusion (p=0.009) and without early CT signs of infarction (p=0.043).

Primary outcome
In the non tPA-treated cohort, 74/129 (57.4%) patients showed good functional outcome at three months (ǠTable 1). Younger age, lower levels of systolic and diastolic blood pressure and serum glucose, and less stroke severity were associated with good functional outcome. Median neuroserpin levels on admission were slightly higher in patients with good outcome, but showed a sharp fall at 24 h that persisted at 72 h. In contrast, patients with poor outcome showed a less marked decrease in neuroserpin levels at 24 h that also remained at 72 h (ǠFig. 1A). Notably, the higher the neuroserpin decrease quartile at 24 h, the better the Rankin score distribution (ǠFig. 2A). In logistic regression analyses, for each quartile in neuroserpin levels decrease, but not in neuroserpin levels at baseline, there was a 15 (95%CI, 3.5 to 66) -fold increase in the probability of good functional outcome (ǠTable 3).
In the tPA-treatment group, good outcome at 90 days was recorded in 38/80 (47.5%) patients. Lower systolic and diastolic blood pressure, platelet count, serum glucose and less stroke severity were associated with good outcome, whereas early CT signs of infarction and proximal MCA occlusion were more frequent in the poor outcome group (ǠTable 2). Median serum neuroserpin levels at baseline were significantly higher in the good outcome group than in the poor outcome group. Furthermore, the greater the quartile of neuroserpin levels at baseline, the better the distribution of mRS score (ǠFig. 2B). In this cohort, high neuroserpin levels before tPA bolus had the stronger effect on favourable outcome (for each quartile increase, OR 13.5; 95%CI, 4.9 to 47), since after treatment neuroserpin levels fall to comparable values at 24 h and thereafter in the two outcome groups (ǠFig. 1B and Table 3).

Non-tPA treated cohort: END and hypodensity volume
Seventeen patients (13.1%) showed END. NIHSS score at baseline but not other variables was associated with END. The higher the quartile of neuroserpin levels decrease at 24 h the lower the rate of END (36.4%, 17.6%, 11.8%, 0% for 1 st Q to 4 th Q, respectively). For each quartile increase there was a 70% (95%CI, 32 to 97) reduction in the probability of END, after adjusting for baseline stroke severity.

tPA-treated cohort: Haemorrhagic transformation, severe brain edema and early MCA recanalisation
In the tPA-treatment group, CT at 24-36 h showed no signs of infarction in 13 (16.2%) patients, while in the remainder median hy-podensity volume was 27.5 [3, 107.3] ml. There was a negative correlation between serum neuroserpin levels on admission and the hypodensity volume (coeff.= -0.550, p<0.001).
HI was observed in eight patients (HI1 in 7, HI2 in 1), PH in 19 (PH1 in 7, PH2 in 8 and rPH in 4; symptomatic in 7) and severe cerebral edema in 10 (grade III in 2 and malignant in 8). Systolic blood pressure and neuroserpin concentrations on admission were significantly higher in patients who did not develop PH. The lower the quartile of neuroserpin levels at baseline the higher the rate of PH (55.0%, 19.0%, 15.8%, 5% for neuroserpin 1 st Q to 4 th Q). In the multivariate analysis, for each quartile increase in neuroserpin levels before tPA bolus there was a 80% (95%CI, 48 to 92) reduction in the probability of subsequent parenchymal haematoma. All patients with severe cerebral edema showed neuroserpin levels on admission within the first quartile (p<0.001).

Discussion
This is, to the best of our knowledge, the first study that evaluates the relationship between serum levels of neuroserpin and neurological deficit in patients with ischaemic stroke. Neuroserpin showed similar results regarding functional outcome in the two studied cohorts, although it disclosed a distinct serum profile within the first hours after stroke. In non tPA-treated patients, a sharp fall of neuroserpin levels at 24 h, but not levels at baseline, was independently associated with good outcome, whereas in the tPA-treated cohort high neuroserpin levels before tPA bolus had the stronger effect on favourable outcome since, after treatment, neuroserpin levels fall to values at 24 h and thereafter which were comparable in the two outcome groups. This favourable effect on outcome was supported by less frequency of END, and lower rate of PH and severe cerebral edema after tPA treatment in patients with high levels of neuroserpin. Therefore, the present findings   show an association between neuroserpin and human ischaemic stroke.
Neuroserpin is a natural tPA inhibitor in the central nervous system (20) that has shown neuroprotection in animal models of cerebral ischaemia and also seems to play a role in inflammation (21,22). The infarct volume, the proteolysis of the basement membrane and the number of apoptotic cells decrease when exogenous neuroserpin is intracortically injected after MCA occlusion (7). These effects have been related to the blockade of the deleterious extravascular effects of tPA through the formation of complexes between the two molecules (neuroserpin-tPA) in the ischaemic brain as well as by reducing tPA and uPA activity (7,9,(23)(24)(25)(26)(27). tPA may exert a harmful role in ischaemic brain injury by promoting excitotoxic damage, microglial activation and extracellular matrix degradation (6,(28)(29)(30), so we may hypothesise that high neuroserpin levels after cerebral ischaemia may represent an innate neuroprotective response to counteract the endogenous activity of tPA, which is also up-regulated, thereby limiting brain injury (31).
In our cohorts, neuroserpin serum levels peaked on admission and subsequently went down close to those levels found in healthy population. These findings might suggest an early release of neuroserpin from the ischaemic brain tissue to the intravascular space through the damaged blood-brain barrier (BBB), and a greater demand of neuroserpin in the injured brain in order to form tPA-inactivating complexes in response to ischaemia. While neuroserpin expression remains high in the area surrounding ischaemic brain tissue for several days after the ictus (7,9), the temporal profile of serum neuroserpin levels has not been investigated in animal models of focal cerebral ischaemia. The distinct time window from onset to sampling (within 12 h in non tPA-treated and within 3 h in What is known about this topic? • Neuroserpin expression increases after ischaemic stroke in animal models. • Neuroserpin displays a neuroprotective role in animal models of brain ischaemia, most probably by inhibiting tissue plasminogen activator (tPA)-mediated neurotoxicity.

What does this paper add?
• It analyses the relationship between neuroserpin serum levels after ischaemic stroke and several clinical variables, providing first clinical evidence that neuroserpin, as previously demonstrated in experimental models, could display a critical role during cerebral ischaemia. • It shows an association between serum levels of neuroserpin and functional outcome after ischaemic stroke in both patients treated with tPA and patients that did not receive tPA. • It shows a significant correlation between neuroserpin serum levels and the incidence of haemorrhagic transformation (HT) in patients treated with tPA. tPA-treated patients) and the higher stroke severity in tPA-treated patients may explain the differences in neuroserpin values at baseline between the two cohorts. Furthermore, our laboratory assays suggest that neuroserpin is not substantially detected when it forms complexes with tPA (see Supplementary Information available online at www.thrombosis-online.com), so we may hypothesise that this complex generation resulted in the sharp fall of neuroserpin levels found in the two outcome groups after exogenous tPA infusion. In contrast, in non tPA-treated patients a distinct binding of neuroserpin to the endogenous tPA could result in a different consumption between the two outcome groups. It has been suggested that the neuroserpin/tPA axis regulates the permeability between the vascular and nervous system compartments (9,32). Pretreatment with neuroserpin before tPA administration in animal models of embolic stroke leads to a significant decrease in tPA-induced BBB permeability (11). In line with these experimental findings, we have found an association between high neuroserpin levels at baseline and lower probability of PH and severe cerebral edema in tPA-treated patients. Clinical studies have demonstrated the relationship between PH, brain edema and increased levels of MMP-9, a proteolytic enzyme overexpressed by the action of tPA (34,35). tPA-mediated MMP-9 overexpression occurs in the 8 h following tPA infusion in patients with acute stroke (36) and, in animal models of stroke, brain and plasma MMP-9 levels correlate at 24 h but not earlier (37). Accordingly, we found a negative correlation between the magnitude of neuroserpin levels decrease at 24 h, potentially consumed by forming tPA-neuroserpin complexes and thus inhibiting tPA activity, and MMP-9 concentrations at 24 h (coeff. = -0.554, p < 0.0001). MMP-9 before tPA infusion had, however, an unexplained slight positive correlation with neuroserpin levels. In addition, it has been recently described a novel enhancing effect of tPA on the BBB permeability that is mediated by the activation of platelet-derived growth factor-CC during brain ischaemia (38).
An intriguing finding in this study was the significant association of high neuroserpin levels at baseline with distal MCA occlusions and early MCA recanalisation. Neuroserpin is mainly expressed in the brain and its tPA inhibitory effect has not been described in the vascular compartment (1,39) where plasminogen activator inhibitor (PAI-1) is the primary serpin that regulates tPA thrombolytic activity, so we anticipated a neutral effect on the tPA thrombolytic activity. Because we have no data on arterial occlusion and recanalisation in non tPA-treated patients, the causal mechanisms of this unexpected result should be further investigated.
The main limitation of this study is that we cannot rule out that ELISA detects a small proportion of neuroserpin-tPA complex, so the free neuroserpin profile in blood in the hours following stroke, and consequently the potential mechanisms underlying its effects are so far difficult to explain.
In conclusion, high neuroserpin serum levels on admission and neuroserpin demand are associated with a better clinical outcome in acute ischaemic stroke patients. The role of neuroserpin in human ischaemic stroke should be the focus of further research.