The prognostic significance of large vessel occlusion in stroke patients treated by intravenous thrombolysis

Introduction

Intravenous thrombolysis (IVT) by rt-PA or tenecteplase is the only effective pharmacological treatment for acute ischaemic stroke (AIS) [1,2]. Unfortunately, this treatment is efficient only in a small proportion of patients [3]. Time from stroke onset to treatment by IVT is thought to be the most important factor affecting IVT efficacy [3]. According to guidelines, to save time, the treatment of AIS by IVT can be done based on the results of head computed tomography (CT) without contrast [1,2]. A few studies, additionally implementing computed tomography angio­graphy (CTA), have shown that large vessel occlusion (LVO) is an important poor prognostic factor in patients with AIS, irrespective of whether they are being treated [4-6] or not treated by IVT [7].
In recent years, mechanical thrombectomy (MT) has become a recommended treatment option for patients with LVO [8-12]. Key studies showing the effectiveness of MT in AIS indicated the advantage of brain vessel neuroimaging before the treatment procedure to show detailed LVO location [8-12]. In our Stroke Unit, we introduced a complex neuroimaging protocol on admission, including computed tomography (CT) with and without contrast and CTA, in all consecutive AIS patients, irrespectively of their treatment perspective and whether or not they declared any history of chronic renal disease. The protocol allowed us to gather immediately data on the presence of LVO, its location, or the presence of vascular malformations. All this information allows easier and safer decision-making about AIS treatment and smoother patient flow during the therapeutic window. In this study, we analysed the frequency and the influence of LVO on admission CTA in consecutive patients treated by IVT or IVT and MT, where indicated, on short-term outcome measures (i.e. haemorrhagic transformation [HT] on CT 24 h after treatment, poor outcome as measured by the mRS 3-6 on discharge, and in-hospital mortality).

Material and methods

We retrospectively analysed the data prospectively collected in the Krakow Stroke Data Bank (KSDB). The KSDB is a single-centre, hospital-based registry in which clinical, radiological, and genetic data from AIS cases are included. The systematic collection of data was started in 2007. The study design was approved by the Jagiellonian University Ethical Committee (KBET 54/B/2007). The diagnosis of stroke was made according to the definition by Sacco et al. [13]. The collected patient information has been described in detail elsewhere [14]. For the purpose of the present study, we analysed data from participants of KSDB treated with IVT from June 2014 to December 2018. In each included patient we analysed the following data available within 24 h after stroke: demographics (age and sex); stroke risk factors (hypertension, diabetes mellitus, ischaemic heart disease, atrial fibrillation, hypercholesterolaemia, smoking status, body mass index, and history of stroke); clinical parameters (prestroke independence as measured by the modified Rankin Score [mRS]) [15] (we distinguished patients with mRS = 0 [no disability] from those with mRS > 0 [at least slight disability]); body temperature on admission (we distinguished patients with body temperature ≥ 37°C and with body temperature < 37°C); time from stroke onset to alteplase infusion; and maximum systolic blood pressure and diastolic blood pressures within 24 hours after alteplase infusion. We distinguished patients with maximal systolic blood pressure > 160 mm Hg and ≤ 160 mm Hg (160 mm Hg was the median value); and patients with maximal diastolic blood pressure > 85 mm Hg and ≤ 85 mm Hg (85 mm Hg was the median value). Stroke severity was measured on admission and 24 hours later according to the National Institutes of Health Stroke Scale (NIHSS) [16]. Delta NIHSS was defined as the difference between NIHSS score measured upon hospital admission minus the NIHSS score 24 hours after alteplase infusion (lower scores reflect deterioration; higher scores indicate improvement). We also examined post-treatment haemorrhagic complications on CT taken 24 hours post-stroke (categorized as patients with or without bleeding) and stroke aetiology according to the Trial of Org10172 in acute Stroke Treatment (TOAST) criteria [17]. We also analysed the presence and location of LVO shown on CTA performed on admission. We analysed LVO location. We also distinguished patients with and without LVO on CTA in large vessels of the brain.
We collected data on the use of antibiotics during hospitalization due to stroke-associated infection (SAI) (we distinguished those with and without SAI). The details about the diagnosis of infections have been reported elsewhere [18]. Biochemical data taken while fasting on the 1st day morning were as follows: CRP levels (increased CRP levels was diagnosed as CRP ≥ 10 mg/l); white blood cells (WBC) count; fibrinogen levels; haemoglobin levels; platelet count; glucose levels; creatinine levels, TSH, total cholesterol levels, LDL-cholesterol levels, HDL-cholesterol levels, and triglycerides levels. We analysed the following early outcome measures: post-treatment haemorrhagic transformation (HT) on computed tomography (CT) taken 24 h after stroke (we distinguished patients with or without bleeding). The intensity of bleeding was evaluated according to the ECASS II criteria [19]. Unfavourable outcome on discharge was defined as mRS score [15] from 3 to 6. We also analysed the in-hospital death rate.

Statistical analysis

Clinical characteristics and outcomes between the groups were compared by the unpaired Student t-test or 2 test, where appropriate. To identify potential independent predictors of outcome, variables with p-value < 0.05 in the univariate regression analysis for the presence of brain haemorrhage, unfavourable outcome as measured by mRS score from 3 to 6, or death on discharge were subsequently included into a multivariate regression analysis. All statistical calculations were performed by Statistica software version 13.3 (TIBCO software INC). P-values below 0.05 were considered statistically significant.

Results

During the study period, 1209 patients registered with the KSDB and 362 (29.9%) received IVT. CTA revealed LVO in 197 patients (54.4%), and 108 patients had MT performed, where indicated. In detail, 46 patients out of 362 (12.7%) had LVO in the symptomatic carotid artery. Among these, 18 (39.1%) had IVT, and 28 (60.9%) had IVT and subsequent MT. In 91 patients (25.1%), LVO was in the M1 region of the symptomatic middle cerebral artery (MCA). Among these, 36 (39.7%) had IVT, and 55 (60.3%) had IVT and MT. In 36 patients (9.9%), LVO was in the M2 region of symptomatic MCA. Among these, 21 (58.4%) had IVT, and 15 (41.6%) had IVT and subsequent MT. In 8 patients (2.2%), LVO was in the M3 region of symptomatic MCA. Among these, 6 (75%) had IVT, and 2 (25%) had IVT and MT. In 16 patients (4.4%), LVO was in the symptomatic basilar artery. Among these, 8 cases had IVT and 8 had IVT plus MT. In 128 patients (35.6%), CTA did not show LVO, and in another 37 patients (10.2%) CTA showed arterial atherosclerotic changes in brain arteries without AC. Haemorrhagic transformation on CT taken 24 h after stroke was found in 75 patients. According to the ECASS II classification, 25 patients (6.9%) developed small petechiae within the infarcted area, 24 patients (6.6%) had more confluent petechiae within the infarcted area, 14 (3.9%) had haemorrhage not exceeding 30% of the infarcted area with some mild space-occupying effect, and 12 (3.3%) had brain haemorrhage exceeding 30% of the infarct volume with significant space-occupying effect. Groups of patients with and without HT did not differ with respect to age and sex distribution. Univariate analysis showed that LVO on admission CTA, AF, hypercholesterolaemia, lower delta NIHSS, higher platelet count, and higher glucose levels correlated with HT (Table 1). Multivariate analysis showed that LVO on admission CTA and lower delta NIHSS were independent factors that affected HT risk (Table 2).
In total 147 patients (40.5%) had an unfavourable outcome, as measured by mRS: 3-6 on discharge. Univariate analysis showed that these patients were significantly older and more often female, more often had LVO on admission CTA, had higher incidence rates of hypertension and AF, had higher body mass index (kg/m2), were more likely to have a history of stroke, more often had body temperature ≥ 37°C on admission, and more often maximal systolic blood pressure > 160 mm Hg and maximal diastolic blood pressure > 85 mm Hg within 24 h after stroke. Patients with unfavourable outcomes were more likely to have mRS scores above 0 pre-stroke or HT on CT, had lower NIHSS than those with a favourable outcome, were more likely to require treatment with antibiotics due to SAI, and were more likely to have CRP levels ≥ 10 mg/l, higher WBC counts, and higher glucose and lower LDL-cholesterol levels (Table 1). Multivariate analysis showed that the presence LVO on admission CTA, older age, female sex, lower delta NIHSS, HT on CT taken 24 h post-stroke, antibiotic treatment due to SAI, more likely CRP levels ≥ 10 mg/l, and higher WBC count affected unfavourable outcome on discharge (Table 2). Multivariate analysis, excluding CRP levels and WBC count (inflammatory markers), showed a similar profile of factors affecting unfavourable outcome on discharge (i.e. LVO, older age, female sex, lower delta NIHSS, HT on CT taken 24 h post-stroke, antibiotic treatment due to SAI) (Table 2).
Thirty-four (9.4%) of the patients died before discharge. Patients who died were older and more likely to be female. Univariate analysis showed that they more often had LVO on admission CTA. Compared to their surviving counterparts, patients who died were more likely to score above 0 on the mRS before the stroke, had lower delta NIHSS, and presented more frequently with HT on CT taken with 24 h post-treatment. They were also significantly more likely to use antibiotic treatment due to SAI. Concerning the biochemical parameters, they presented with CRP levels ≥ 10 mg/l more often and had lower triglyceride levels (Table 1). Multivariate analysis showed that an mRS above 0 before admission, lower delta NIHSS, HT on CT taken 24 h after IVT, and CRP levels ≥ 10 mg/l were independent factors that affected the risk of death up to discharge. Multivariate analysis after the exclusion of CRP did not change the profile of independent factors affecting in-hospital mortality (Table 2).

Discussion

We found that in patients with AIS treated by IVT solely or IVT and MT, where indicated according to current recommendations [1,2], the presence of LVO on admission CTA was an independent factor affecting the risk of HT on CT taken 24 h after stroke and unfavourable outcome on discharge as measured by the mRS: 3-6, but not in-hospital mortality. A few studies performed in patients treated by IVT solely [4,5] or IVT and intraarterial thrombolysis as bridging therapy [6], before the era of MT, indicated that the LVO on CTA affected poor prognosis. The prognostic significance of the presence of LVO on admission CTA in consecutive patients treated by IVT or IVT and MT, where indicated, was not studied systematically.
According to guidelines, to shorten the time window, the treatment of AIS by IVT can be done based on the results of the head CT without contrast. Our results strengthen the need to perform additionally admission CTA of the brain arteries to show the presence and exact location of LVO in the brain artery. This knowledge is not only helpful in preparing individual treatment plans but also, based on the results of the present study, in predicting short-term prognosis. We would like to emphasize that in our Stroke Unit, in an emergency setting, to shorten the therapeutic window maximally and based on current guidelines [1,2], we do not measure creatinine levels routinely on admission. Only patients who report chronic renal failure on admission have their neuroimaging procedures scheduled individually, after evaluation of the blood parameters of renal efficiency and a consultation with a nephrologist. In all other patients, creatinine levels are routinely studied on the next morning after the admission. In the presented group of patients, we recorded a mean creatinine level of 84.4 ± 25.8 µmol/l (median value, 81 µmol/l; IQR, 68-97 µmol/l; min-max values, 27 and 218 µmol/l). In patients with creatinine levels outside the normal range, consultation with a nephrologist was routinely scheduled, and treatment was introduced. Nobody from the presented group required dialysis.
Haemorrhagic transformation on CT in AIS patients treated by IVT as an early outcome measure was analysed previously by several authors [20-33]. Most of these were analysed as symptomatic haemorrhage according to ECASS II criteria (i.e. the presence of any haemorrhage on CT not detected on previous CT, combined with neurological deterioration of 4 points or more in NIHSS from baseline) [19]. The prevalence of symptomatic haemorrhages in the studies varied from 3.9% [27] to 8.4% [23]. In our study, we were able to find only 12 symptomatic haemorrhages according to the above-mentioned criteria (3.3%), and we were not able to find any significant differences between patients with and without symptomatic haemorrhage (data not shown). When we compared patients with and without HT on CT taken 24 h after stroke, despite taking into account any concomitant changes in neurological deficit, in a multivariate regression analysis, we found that LVO and deterioration of the neurological deficit after treatment affected its risk. Interestingly, our results are in line with previous studies published on this topic [20,29-31], showing the correlation of symptomatic brain haemorrhage with neurological deficit.
In our study, we also measured 2 different early outcome measures: unfavourable outcome (defined as the mRS from 3 to 6 on discharge) and in-hospital mortality. Interestingly, many different early outcome measures were studied so far in AIS patients treated by IVT, such as the following: early recanalization [22,34], in-hospital mortality [30,35], in-hospital mortality or discharge to a nursing home [29], neurological deterioration as measured by an increase of NIHSS by at least 4 points 24 h after stroke [24,29,36], at least by 8 points on day 7 after stroke [24,29], or deterioration by at least 2 points between day 0 and 5 [21], and mRS on discharge [37]. In those studies, most of the authors analysed the influence of a single chosen parameter on the studied early outcome measure. For example, Kimura et al. showed that AF is the only factor influencing the lack of recanalization, and the lack of recanalization correlated with neurological worsening [34]. Sung et al. showed that mortality on discharge in severe stroke was similar in those with and without AF [30]. Jovanowic et al. showed that early outcome measures are similar in males and females [24], while Al Husain confirmed the lack of correlation between sex and recanalization rate [22]. Tang confirmed that diabetes mellitus affects unfavourable outcomes 24 h after stroke, and on day 7 after stroke [29]. Only 2 authors were able to build up logistic regression models showing independent factors affecting in-hospital mortality (age ≥ 80 years, aphasia, conscious disturbances, hypertension, the presence of haemorrhage on CT, and pneumonia) [20,29] and in-hospital mortality or discharge to a nursing facility (older age, longer hospital stay, and large cerebral infarct) [38].
In our study, we showed that unfavourable outcome as measured by mRS: 3-6 on discharge, but not in-hospital, is affected by the presence of LVO on admission CTA; lower delta NIHSS and HT on CT taken 24 h after stroke are common disadvantageous factors affecting both outcome measures in the present study. Unfavourable outcome was also related with antibiotic treatment due SAI, irrespectively of including or excluding markers of inflammation into logistic regression models. Interestingly, as was shown previously [39], age and sex affected the unfavourable outcome but not in-hospital mortality. Our study is mainly limited by its retrospective analysis. The risk of selection bias is possible because we included only patients who agreed to participate in the study. We also analysed outcome measures on discharge, and, unfortunately, we did not collect detailed information about the length of hospital stay; in Poland, patients with AIS are hospitalized for at least 9 days or until death. Another limitation is the relatively long recruitment period, lasting 5 years, so this long period might increase the number of confounding factors because the approach to stroke treatment changes rapidly over time.

Conclusions

In AIS patients treated by IVT solely or by IVT and MT, where indicated according to current guidelines, LVO of the brain vessels shown on admission CTA affects the risk of HT or unfavourable outcome on discharge, but not in-hospital mortality.

Conflict of interest

The authors report no conflict of interest.

References