Background and purpose As previously demonstrated, the recanalization rate and clinical outcome in patients with anterior circulation stroke treated with IV thrombolysis (IVT) depend on clot characteristics such as thrombus length and thrombus density. In patients with basilar artery occlusion (BAO) treated with IVT, the recanalization rate has also been shown to depend on thrombus length, although no cut-off value beyond which recanalization would seem impossible has been determined. We aim to evaluate the correlation of clot characteristics with recanalization rate and outcome in patients with BAO treated with endovascular therapy (EVT).
Methods We retrospectively assessed 51 consecutive patients with BAO treated with EVT. Thrombus length and thrombus density (in Hounsfield units, HU) were measured on thin slice non-enhanced cranial CT scan before treatment. Thrombolysis In Myocardial Infarction grade 2–3 was considered successful recanalization and 3-month modified Rankin Scale score 0–2 was considered a favorable outcome. To evaluate the correlation of clot characteristics with recanalization rate and outcome, a binary logistic regression test was computed.
Results Neither thrombus length nor thrombus density correlated with recanalization rate (OR 1.02, 95% CI 0.94 to 1.11, p=0.58 and OR 1.09, 95% CI 0.97 to 1.23, p=0.13, respectively). Thrombus density and thrombus length were not significantly different in patients with (n=41, 80.4%) or without (n=10, 19.6%) successful recanalization (52.3 HU vs 48.4 HU, p=0.07 and 8.2 mm vs 7.5 mm, p=0.91). However, higher clot density was correlated with a favorable outcome (OR 1.31, 95% CI 1.08 to 1.59, p=0.006) whereas thrombus length was not correlated with clinical outcome (OR 0.94, 95% CI 0.86 to 1.03, p=0.20).
Conclusions Successful recanalization does not depend on thrombus length in patients with BAO treated with EVT. Recanalization can therefore be achieved despite high clot burden. Additionally, a high density of thrombi was a strong predictor of a favorable outcome.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Basilar artery occlusion (BAO) is a devastating disease resulting in a mortality of 90% and high morbidity if untreated. With the advent of intravascular thrombolysis (IVT) the prognosis dramatically improved, even before IVT became standard therapy for ischemic stroke.1 However, an advantage of endovascular therapy (EVT) compared with IVT in patients with BAO has never been proven. The Basilar Artery International Cooperation Study (BASICS), a large registry, claimed equal efficacy of IVT and IA thrombolysis.2 Recently, the ENDOSTROKE register study demonstrated a high rate of revascularization with EVT in patients with BAO.3
Thrombus length has been identified as a major predictive factor for successful recanalization in patients treated with IVT to recanalize occlusions in the anterior circulation.4 However, the recanalization rate was not dependent on thrombus length in patients treated with EVT.5 More recently, hyperdense thrombi were shown to correlate with higher recanalization rates,6 ,7 although this is in dispute.8 We therefore aimed to evaluate the predictive value of thrombus properties regarding recanalization rate and outcome in patients with BAO treated with EVT.
Methods and materials
We retrospectively studied consecutive patients with BAO treated with EVT between April 2007 and October 2015. Patients who underwent non-enhanced cranial CT (NECCT) scanning with thin slice reconstructions before treatment were included (n=55). Patients with significant artifacts in the initial NECCT images preventing thrombus evaluation were excluded (n=4).
The severity of symptoms at admission was scored using the National Institutes of Health Stroke Scale scores (NIHSS). Time from onset to treatment was defined as the interval between treatment onset to initiation of treatment. If the time of onset was unknown, the time that the patient was last seen asymptomatic was recorded. Stroke etiology was determined using the Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification9 and a standardized battery of tests including follow-up brain imaging (CT and/or MRI), cervical and intracranial vessel imaging (CT angiography (CTA), MR angiography (MRA), or ultrasound), 24 hour ECG Holter monitoring, transthoracic or transesophageal echocardiography, and further laboratory tests as clinically feasible and appropriate. Clinical outcome was evaluated using the modified Rankin scale (mRS) at 3 months, either by face-to-face consultation or telephone interview. An mRS score of 0–2 was defined as a favorable outcome whereas an mRS score of 3–6 was defined as a poor outcome. Further relevant demographic data were extracted from the patients' charts. A board-certified neurologist collected the clinical data of all included patients.
Image acquisition, analysis and statistics
Thrombi in the basilar artery (BA) were detected in all NECCT datasets (Brilliance 64, Philips, Best, The Netherlands; 120 kV, 320 mAs, reconstructed slice thickness 2.5 mm) and clot length was calculated by one author as described previously. In short, a region of interest (ROI) was defined in the course of a hyperdense vessel. All pixels between 55 and 80 Hounsfield units (HU) served as a seed for a region-growing algorithm into adjacent pixels between 40 and 80 HU. Thrombus length was then calculated along the middle axis of this defined volume.10 The presence of a hyperdense BA sign was rated by two independent raters. If disagreement was present, a consensus was reached. Cohen's κ was used to assess the agreement.
CTA (80 kV, 280 mAs, 80 mL bolus of 350 mg I/mL, followed by 40 mL saline flush at 5 mL/s) was available in 46 (90.2%) patients. Thrombus length, defined as an opacification gap on CTA, was determined by one author and checked for consistency by another, both blinded to thrombus length measurement on NECCT. A Bland-Altman plot was used to depict the agreement between thrombus length measurement by CTA and NECCT.
Absolute HU values were calculated by two authors blinded to clinical information by placing ROIs into the hyperdense parts of the vessel on every slice, adding the obtained HU values, and then dividing by the number of observed slices (figure 1). Vessel wall calcifications (>100 HU) were excluded from the measurement.7 ,11 When available, CTA was used to localize the thrombus. In order to correct for hematocrit, relative HU values were calculated using the HU ratio (HUclot/HUcalibration). HUcalibration was measured in a vessel distal or proximal to the clot (intracranial vertebral artery, proximal BA, or posterior cerebral artery).7 ,11 ,12
Occlusion site and recanalization
The site of occlusion was categorized as proximal (between the confluence of the vertebral arteries and the origin of the anterior inferior cerebellar artery (AICA)), mid (between the origin of the AICA and the origin of the superior cerebellar artery (SCA) but also any occlusion of more than one segment including mid BA) and top of the BA. Recanalization success was determined by digital subtraction angiography (DSA) after intervention by a rater blinded to clinical outcome. A TIMI (Thrombolysis in Myocardial Infarction) score of 2–3 was considered as successful recanalization and a TIMI score of 0–1 was considered no recanalization.2 ,13 ,14
Statistical analysis was conducted using SPSS software V.22 (IBM, New York, USA). All continuous variables were first tested for normality using the Kolmogorov-Smirnov test. Variables with a normal distribution were recorded as mean±SD and were tested using an independent t-test. The Mann-Whitney U test was used to analyze the variables without a normal distribution. To describe them, the median value and IQR were calculated. Categorical variables are presented in absolute numbers or percentage and were assessed using Fisher's exact test. Logistic regression analysis was performed to determine clinical factors which might have a predictive value regarding successful recanalization and favorable outcome. Statistical significance was defined as p<0.05.
Fifty-one patients with BAO treated with EVT were included in our study. EVT was started within 6 hours of symptom onset in most cases (n=37, 73%). EVT was performed using three different strategies: IA recombinant tissue plasminogen activator (rtPA) alone (5–80 mg, n=5), mechanical thrombectomy (n=30), and mechanical thrombectomy in combination with IA rtPA (n=16) (table 1). IVT (0.9 mg rtPA/kg body weight) was delivered in eight (15.7%) patients prior to EVT. The baseline characteristics of all patients are summarized in tables 2 and 3. 70.6% of the patients were men (n=36), median age was 72 years (IQR 66–78 years), and median NIHSS at admission was 18 (IQR 11–20). Median thrombus length measured by NECCT was 7.5 mm (IQR 5–14.4 mm). Mean absolute thrombus density was 51.6±6.2 HU, whereas mean relative thrombus density was 1.34±0.15. There was no correlation between thrombus length and density (p=0.16). The site of occlusion was the proximal BA (19.6%), mid BA (41.2%) and top of the BA (39.2%). Thrombus location did not correlate with clinical outcome or recanalization rate (tables 2 and 4). In 42 patients (82.4%) a hyperdense BA (HDBA) sign was present (Cohen's κ=0.74). Median time from onset to treatment was 222 min (IQR 178–462 min). Neither the presence of a HDBA sign nor time from onset to treatment was correlated with recanalization and outcome. The Bland-Altman plot demonstrated a high agreement between thrombus length measured on NECCT and on CTA (figure 2). Interestingly, the thrombi of the three patients whose data points were outside the limits of agreement proved to be of heterogeneous composition (figure 3). Stroke etiology was cardiogenic (58.5%, n=30), large artery atherosclerotic (25.5%, n=13), other confirmed (9.8%, n=5), and unknown (5.9%, n=3) according to the TOAST criteria.9
Successful recanalization was achieved in 41 patients (80.4%). A stent-retriever was used in 32 patients (62.7%). In seven patients (21.8%) no or minimal recanalization was achieved and in one patient (3.1%) the exact number of passes was not documented. For the remaining 24 patients we did not find a correlation between thrombus density and number of stent-retriever passes (p=0.197). However, there was a trend towards longer thrombus with a greater number of passes (r=0.37, p=0.08). In univariate analysis, thrombus length measured by NECCT or CTA15 showed no significant difference between patients with (n=41, 80.4%) or without (n=10, 19.6%) successful recanalization (8.2 mm vs 7.5 mm, p=0.91 and 10.9 mm vs 10.8 mm, p=0.89; table 2 and figure 4). Absolute thrombus density showed a trend towards a higher recanalization rate (p=0.07). However, after correction for hematocrit, relative thrombus densities showed no statistically significant correlation with successful recanalization (p=0.29). Other clinical variables also did not have a correlation with successful recanalization. In the logistic regression analysis, neither thrombus length nor thrombus density had a significant correlation with the recanalization rate (OR 1.02, 95% CI 0.94 to 1.11, p=0.58 and OR 1.09, 95% CI 0.97 to 1.23, p=0.13, respectively; table 3).
A favorable outcome was observed in 13 patients (25.5%). In univariate analysis, absolute thrombus density was statistically different between patients with (n=13, 25.5%) and without (n=38, 74.5%) a favorable outcome (55.7 HU vs 50.1 HU, p=0.004; table 4). This statistically significant difference was also present when relative thrombus densities were compared (1.49 vs 1.28, p<0.001). In a secondary analysis, no differences were found in absolute thrombus densities when patients with local atherosclerotic changes were compared to those without (51.0±7.0 vs 51.8±5.9, p=0.70), and there was no difference regarding recanalization rates (p=0.25) or clinical outcomes (p=0.73). There was no difference in thrombus lengths between patients with and without a favorable outcome (7.3 vs 8.3 mm, p=0.23; figure 5). All patients without a HDBAS showed a poor outcome (n=13, 100%) even though no statistical difference was found (p=0.09). In the logistic regression analysis, thrombus length did not show a correlation with favorable outcome (OR 0.94, 95% CI 0.86 to 1.03, p=0.20). However, high density of thrombi was a significant predictor of a favorable outcome (OR 1.31, 95% CI 1.08 to 1.59, p=0.006; table 3 and figure 6). High baseline NIHSS was correlated with a worse outcome. Symptomatic intracranial hemorrhage per European Cooperative Acute Stroke Study II (ECASS II)16 occurred in five patients (9.8%), all of whom showed a poor outcome.
In this study we analyzed the correlation of thrombus properties with the recanalization rate and clinical outcome in 51 patients with acute occlusion of the BA treated with EVT.
The main results are threefold. First, the recanalization rate was independent of thrombus length. The effectiveness of EVT is well documented and has been reported to range from 60% to 80%,17 ,18 even for high clot burden.19 This is different in patients treated with IVT. Our previous study suggested that patients with acute middle cerebral artery occlusion barely benefited from IVT if the length of the thrombi exceeded 8 mm.4 Strbian et al 14 demonstrated that the recanalization rate in patients with BAO after IVT was dependent on thrombus length, although no threshold could be established beyond which IVT was deemed futile. Similar to our results, Gilberti et al 20 reported that recanalization is independent of thrombus length in BAO treated with EVT. However, they determined thrombus length by DSA whereas we measured thrombus length on initial NECCT.
Second, we found that high attenuating thrombi were correlated with a better clinical outcome. Several previous studies have already demonstrated higher attenuating thrombi to be associated with better recanalization rates either by IVT4 ,11 ,12 or EVT.6 ,7 ,12 It is well known that erythrocyte-rich thrombi have a higher density than platelet-rich thrombi on NECCT.21 ,22 It has been claimed that erythrocyte-rich thrombi with little fibrin content are more susceptible to pharmacological23 and mechanical therapy.24 In a recent systematic review and meta-analysis, the authors concluded that patients with a mean thrombus density of 55.1±3.1 HU had a good angiographic outcome whereas those with a mean thrombus density of 48.9±1.9 HU had a poor angiographic outcome.25 However, we did not find a correlation with recanalization but with clinical outcome. We hypothesized that low attenuating thrombi might be caused by atherosclerotic changes (figure 7) and partial volume effects may have blurred high attenuating thrombi, but our secondary analysis did not substantiate this hypothesis. Patients with atherosclerotic changes in the BA did not have lower attenuating thrombi and did not experience lower recanalization rates and clinical outcome.
Third, contrary to anterior circulation stroke, recanalization did not automatically translate into a good clinical outcome. We thus suggest that thrombus length may not be the pivotal factor when it comes to clinical outcome. This is in line with previous results suggesting that other factors such as time to treatment and collateral status might substantially influence clinical outcome.3 Patients should be treated irrespective of thrombus length with the treatment modality according to the advice of national or local guidelines.
There are several reasons why the influence of thrombus density on clinical outcome and recanalization rate should be regarded with caution. First, there is no standard procedure to measure thrombus densities (one or more RoI) and results are thus more or less subjective. Second, in the posterior circulation, especially in the BA, beam hardening artifacts caused by the temporal bones and vessel calcifications are fairly common and may substantially alter density measurements. Third, the claim that high attenuating thrombi are associated with better recanalization rates is in dispute. Yilmaz et al 8 reported on 70 patients treated with EVT in whom thrombus density was not predictive of recanalization success. The abovementioned reasons and others such as slice thickness, different treatment modalities and devices, and the prevalence of atherosclerotic changes in the posterior circulation also limit the comparability between studies. Thus, more sophisticated and standardized methods to measure thrombus density including filtering are needed.26
Our study has some limitations. The retrospective nature of the study and limited sample size (type II error) does not allow for generalization of our results. Further, the study period from 2007 to 2015 limits the comparability of devices used as the effectiveness dramatically changed. In our cohort we achieved a higher rate of recanalization with second-generation devices (n=32) of 78.1% compared with 57.1% with first-generation devices (n=19). In addition, we did not evaluate every possible variable such as collateral status or medical history that may be predictive of recanalization and outcome. Furthermore, considering the limited number of patients, multilogistic models were not used for further analysis of the independent predictive value of thrombus density toward a favorable outcome. Four patients in our cohort were excluded from this study because of substantial artifacts preventing thrombus length and density measurement. With regard to their successful recanalization and poor outcome, the relationship between thrombus properties and recanalization and outcome might be underestimated.
Successful recanalization with EVT can be achieved despite a high clot burden in patients with acute occlusion of the BA. High density of thrombi seems to be a predictor of better clinical outcome. Standardized methods to measure thrombus density are needed.
Contributors LS and UJ-K planned the study. LS, CR, OJ and UJ-K conducted the imaging part of the study. JM conducted the clinical part of the study. LS conducted the statistical analysis and wrote the first version of the manuscript. LS and UJ-K are the guarantors of the study.
Funding This work received funding from the ‘Schlaganfallnetzwerk Schleswig-Holstein’ and the ‘Damp Foundation’.
Competing interests None declared.
Ethics approval Ethics approval for this study was obtained from the ethics committee of the Medical Faculty of the Christian-Albrecht-University Kiel.
Provenance and peer review Not commissioned; internally peer reviewed.