Article Text
PDF
Abstract
Background Different imaging characteristics such as clot burden score, collaterals, and pre-interventional thrombus migration are associated with functional outcome in patients with acute ischemic stroke. Moreover, histological thrombus composition is associated with pre-interventional thrombus migration. We hypothesized that smaller clots may more likely migrate and that collateral status in ischemic stroke patients may mediate this tendency of the clot to migrate.
Methods In this prospective cohort of consecutive ischemic stroke patients, clot burden scores and collateral scores were rated and the retrieved thrombi were histologically analyzed. We then investigated the relationship between clot burden score, probability for thrombus migration, and collateral scores using mediation analysis.
Results 163 patients are included of which 36 (22.1%) had a clot migration. Probability of thrombus migration was significantly associated with lower collateral scores (P<0.01), higher clot burden scores (P<0.01), shorter thrombi (P<0.01), and higher RBC count (P<0.01). In the mediator pathway, higher collateral scores were significantly associated with higher clot burden scores (P<0.01) and younger age (P=0.029). The total effect of an increase in clot burden score by one grade on thrombus migration is composed of the direct effect (+18%, P<0.01) and the collateral score-mediated indirect effect (−5%, P<0.01).
Conclusions Smaller, erythrocyte-rich thrombi tend to migrate more often. Good collaterals seem to have a considerable effect on limiting migration. This supports the hypothesis that larger clots have stronger adherence with the vessel wall and that good collaterals increase the counter pressure distal of the clot.
- stroke
- thrombectomy
- angiography
Data availability statement
Data are available upon reasonable request.
Statistics from Altmetric.com
Introduction
Mechanical thrombectomy has become standard of care for large-vessel occlusion stroke after randomized trials have confirmed its efficacy and benefit for neurological outcome and thereby enables histological analysis of the retrieved thrombus.1 We and others have reported three main thrombus components: fibrin/platelets (F/P), red blood cells (RBC), and white blood cells (WBC).2–6 Moreover, we have shown that histological thrombus features contain important information regarding stroke etiology,3 can predict interventional and clinical outcome, and can themselves be predicted using admission imaging.2
Thrombus migration (TM) is associated with the technical and clinical success of thrombectomy as it is a predictor of lower complete recanalization rates after mechanical thrombectomy and may impact neurological improvement rates.7–10 In a previous study we observed thrombus migration more often in erythrocyte-rich thrombi and is less often in fibrin-rich clots. However, the association with other admission imaging and thrombus characteristics such as clot burden score and collateral score in computed tomographic angiography (CTA) is unknown, even though these parameters seem to be associated with functional outcome in patients with acute ischemic stroke.10–12
We therefore conducted one large study based on thrombus histology to determine whether there is an association of these important imaging parameters with histological thrombus characteristics and with pre-interventional thrombus migration. We hypothesized that smaller clots (represented by higher clot burden scores) may more likely migrate and that collateral status in ischemic stroke patients might mediate this probability.
Methods
Patients
We retrospectively evaluated data of a prospectively collected cohort of 198 consecutive ischemic stroke patients with anterior circulation occlusion in whom thrombectomy with stent retriever was performed. Only patients with occlusion of the M1 segment of the middle cerebral artery and complete diagnostic and histologic workup were included for further analysis: patients with insufficient imaging quality for the required analysis or without follow-up imaging were excluded. Patients who underwent baseline imaging at other institutions and were transferred for thrombectomy (drip and ship) were excluded because admission imaging was not available in these patients.
We primarily investigated the relationship between thrombus histology expressed as a percentage of the main components (fibrin, RBCs, and WBCs) and thrombus migration.
We also performed immunohistochemical tests including analysis of clusters of differentiation co-receptors (CD3, CD20, and CD68/KiM1P). Stroke cause was determined according to the TOAST classification using CT and MR imaging, duplex sonography of the cervical arteries, coagulation tests, long-term electrocardiography, and transthoracic or transesophageal echocardiography.
Additionally, baseline information including age, sex, application of intravenous tPA, and interventional parameters was obtained from patients’ clinical records and imaging on admission (table 1).
Clinical characteristics and clot compositions between patients with and without clot migration
The study was approved by the Ethics Committee of the University Münster and the Westfalian Chamber of Physicians, Münster, Germany (2017–233 f-S). All study protocols and procedures were conducted in accordance with the Declaration of Helsinki.
Interventional thrombectomy
Interventional thrombectomy was performed by experienced neuroradiologists under general anesthesia. In all patients we used a biplane neuro-X-ray system (Allura Xper FD20/20, Philips, Best, The Netherlands). Interventional therapy included positioning of a 6F guiding catheter in the cervical segment of the internal carotid artery. In most of the patients a pRESet stent retriever device (Phenox, Bochum, Germany) with a size of 2×4 or 3×6 mm was used. All angiograms were reviewed postinterventionally and graded using the modified TICI score.13
Histology and immunohistochemistry
5 µm-thin sections of paraffin-embedded, formalin-fixed samples were prepared. Sections were stained with hematoxylin and eosin as well as with Elastica van Gieson and Prussian blue following a standard protocol. For immunohistochemical staining, sections were placed on slides, deparaffinized, rehydrated, and then washed in aqua dest. If necessary, for antigen retrieval, slides were microwaved twice for 7 min each and then washed in a buffer of phosphate buffered saline (PBS), bovine serum albumin, and Triton X-100 (PBT) three times for 5 min each. The slides were blocked in 20% goat serum diluted in PBT for 30 min and then incubated with the following antibodies: CD3 (mouse monoclonal, 1:25, pH 6.1, DAKO, Glostrup, Denmark) and CD20 (mouse monoclonal, 1:700, pH 6.1, DAKO) and CD45 (mouse monoclonal, 1:800, pH 6.1, DAKO). For CD68 staining supernatant from KiM1P hybridoma cells, kindly provided by Prof. Klapper, Institute of Pathology, Kiel (1:5000, mouse monoclonal) was used. After washes in PBT, slides were incubated with a biotinylated goat anti-rabbit secondary antibody (E0432; 1:500 dilution; Dako) for 45 min at room temperature after incubation with the ABC kit (SK6100; Vectastain avidin-biotin complex-horseradish peroxidase, Vector Laboratories, Burlingame, CA, USA) for 45 min. The signal was developed using a 3,3–diaminobenzidine substrate kit (SK4100; Vector Laboratories), and the sections were counterstained with hematoxylin.
Quantification of stainings
Using an Olympus BX43 microscope and digital camera, photographs of thrombus material was photographed (magnification 40 x). For quantification of erythrocytes, fibrin, and other cellular components, ImageJ software (ImageJ 1.47 n, National Institute of Health, Bethesda, MD, USA) was used. Quantification of fibrin, RBC, and WBC was performed manually while for quantification of immunohistochemical stainings, the pictures were converted to gray scale (8-bit), threshold was set, and particles were analyzed automatically (areas covered by the respective cells (%) were measured).
Evaluation of CT parameters
Evaluation of thrombus migration between admission CTA and interventional digital subtraction angiography (DSA) was performed as recently described.4 7 14 In brief, evidence for thrombus migration is evaluated using different approaches of direct evidence of thrombus migration between admission CTA and DSA as well as indirect evidence of TM by comparing infarct patterns in the lenticulostriate artery territory to more distal occlusions in the DSA.
Clot burden was determined according to the clot burden score on baseline CTA.11 Briefly, a score of 10 on the clot burden score indicates that no occlusion is present. Two points are deducted for lack of contrast opacification in the supraclinoid internal carotid artery and both the proximal and distal M1 segment. One point is deducted for lack of opacification in the M2 branches, the A1 segment, or the infraclinoid internal carotid artery.
The collateral score was evaluated on baseline CTA as previously described.15 Collateral score grades distal arteries filling with a 4-point scale, with 0 constituting absent collaterals (0% filling of the occluded territory), 1 for poor collaterals (>0% and≤50% filling of the occluded territory), 2 for moderate collaterals (>50% and<100% filling of the occluded territory), and t3 for good collaterals (100% filling of the occluded territory). All ratings were performed by two experienced neuroradiologists, and disagreements were settled by joint discussion.
Statistical analysis
For categorical data, absolute and relative frequencies are given. Univariable distribution of metric variables is described by median and IQR. Patients with good (no clot migration) vs poor (clot migration) outcome were compared by Mann-Whitney U test for metric outcome variables and by chi-square test for categorical outcome variables. Mediation analyses16 17 were used to evaluate whether collateral score mediates the probability of thrombus migration at different levels of clot burden in patients undergoing mechanical thrombectomy. Mediation models consider the impact of an intervening variable M (mediator) that is hypothesized to transmit the influence of independent variables X onto an outcome Y. The basic mediation model employed in this analysis is depicted in figure 1, and the underlying regression models were defined according to Mackinnon and Dwyer.18
Mediation analysis path diagram for assessing the mediation effect of collateral score on thrombus migration in patients undergoing mechanical thrombectomy. RBC: red blood cells in thrombus histology; β: regression coefficients.
Regression models
Accounting for the ordinal nature of the mediator collateral score, an ordered probit model was selected for regression of the mediator on the independent variables. The relationship between the dichotomous outcome variable thrombus migration yes/no and the independent variables including mediator was modeled with binary probit regression analysis.
Independent variables were selected using backward selection algorithms. All predictors that were significant for either the mediator collateral score or the outcome variable thrombus migration were included in the analysis. In addition, common predictive parameters (age, sex) were added to the analysis.
Mediation analysis
Mediation was assessed according to requirements defined by Baron and Kenny.16 First, the mediator collateral score was regressed on the independent variables (clot burden score, age, sex, internal carotid artery location of thrombus, thrombus lengths, and RBC count) to confirm that the independent variables are significant predictors of the mediator. Second, the dependent variable (probability of thrombus migration) was regressed on the independent variables and the mediator to confirm that both the independent variables and the mediator are significant predictors of the dependent variable.
Mediation analysis was conducted employing algorithms proposed by Imai et al19 and Imai et al20 that allow estimation of causal mediation effects for linear and nonlinear relationships with continuous and discrete mediators, and various types of outcome variables. Effects were assessed through the average causal mediation effect (ACME) and the average direct effect (ADE) with clot burden score as an independent condition. In this context, ACME represents the expected change in the probability for thrombus migration when the mediator collateral score increases by one grade, while the clot burden score itself is held constant. ADE represents the expected change in the probability for thrombus migration when the clot burden score increases but the mediator collateral score is held constant. Both metrics, ACME and ADE, on average add up to the total impact of clot burden score on the probability for thrombus migration. The strength of mediation effects was evaluated at two different base levels of the clot burden score with low thrombus burden level (clot burden score=8) and high thrombus burden level (clot burden score=3). Furthermore, percentage proportions of ACME and ADE in relation to the total effect were calculated.
Confidence intervals of the mediation analysis were derived using quasi-Bayesian approximation. All analyses were carried out using R 3.6.2 with the mediation package 4.5.0 and in SPSS version 26 (IBM Corporation, Armonk NY).
Results
Patients and rates of clot migration
Of 198 patients in whom clot material was collected, 11 patients had to be excluded due to insufficient histological analysis and 24 patients were excluded because of missing follow-up imaging or insufficient imaging quality. Of the remaining 163 patients, 36 (22.1%) patients were identified with either direct (15) or indirect (21) signs of clot migration prior to endovascular therapy. Interrater reliability concerning the classification of clot migration was excellent (Cohen’s kappa 0.89).
Association of clot migration and histology
Patients with proven clot migration had significantly higher levels of RBC than patients without clot migration (median 50% vs 26%, P<0.001). We also observed a significant association of higher rHU and clot migration (P=0.025; table 1). Lower amounts of fibrin were significantly more often observed in patients in the clot migration group compared with the no clot migration group (43.5% vs 62.0%, P<0.001; table 1). In patients with proven clot migration the pre-interventional median embolus length was statistically noticeably shorter compared with the patients with no clot migration (9 mm vs 12 mm; P<0.030). A higher clot burden score was significantly more often observed in patients in the clot migration group. Intravenous thrombolytic treatment was not associated with the risk of thrombus migration (P=0.686). Other baseline criteria likewise did not showa significant difference between the groups with and without clot migration (see table 1).
Mediation analysis
Indirect mediator path a
Results of the probit regression model were as follows: higher collateral scores were significantly associated with higher clot burden scores (β=0.30; P-value<0.01), younger age (β=−0.015; P-value=0.029). Thrombus length and RBC count were not significantly associated with collateral scores.
Direct path b
Results of the binomial probit model confirm that the probability of thrombus migration is significantly associated with lower collateral scores (mediator with β=−0.55; P-value<0.01), higher clot burden scores (β=0.29; P-value<0.01), smaller thrombus length (β=−0.07; P-value<0.01) and higher RBC count (β=0.02; P-value<0.01). Age and sex were not significant direct predictors of thrombus migration.
Mediation analysis
All requirements for the mediation relationship according to Baron and Kenny16 were fulfilled: regression analysis of path a and path b confirmed significant association of the independent predictors and the mediator collateral score with the probability for thrombus migration. Furthermore, significant coefficients were observed for the regression of collateral score on the independent predictors. All effect metrics of the mediation analysis were significant at P-value<0.05.
The total average proportion of effects on the probability for thrombus migration that is mediated by collateral score was calculated at 41% (P-value=0.022). In patients with high clot burden score (low thrombus burden), the share of effects mediated by collateral score was higher than in patients with low clot burden score (high thrombus burden) with 69% vs 12% (P-value=0.022). In total, an increase in clot burden score by one grade (lower thrombus burden) was associated with an increase in the probability for thrombus migration by 13% (P-value=0.014). The total effect is composed of the direct effect ADE (+18%, P-value<0.01) and the mediated indirect effect ACME (−5%, P-value<0.01). Thus, patients with lower thrombus burden (higher clot burden score) had an increased risk for thrombus migration (direct path b, tables 2 and 3, figure 2). However, for higher collateral scores, that were shown to be associated with lower thrombus burden, an opposed indirect mediator effect with reduced risk for thrombus migration was observed (indirect path a, tables 2 and 3, figure 2). If evaluated at different levels of clot burden score, the mediation effect via collateral score was higher for patients with low thrombus burden (risk reduction by 8.5%, P-value<0.01) than for patients with high thrombus burden (risk reduction by 1.8%, P-value<0.01).
Direct and mediation effects on probability of thrombus migration at low thrombus burden (clot burden score ≥8) and high thrombus burden (clot burden score ≤3). ACME: average causal mediation effect; ADE: average direct effect
Regression analysis for mediation path a (probit regression model to ordered factor response with dependent variable collateral score) and direct path b (binomial probit regression model with dependent variable probability for thrombus migration)
Results of the mediation analysis. Low clot burden score is ≤3 (high thrombus burden); high clot burden score is ≥8 (low thrombus burden)
Discussion
Several imaging characteristics such as clot burden score and leptomeningeal collaterals in admission CTA have been shown to be associated with functional outcome in patients with acute ischemic stroke.10–12 However, to date a possible association of these parameters with underlying thrombus histology and the risk of pre-interventional thrombus migration is unknown even though the occurrence and therefore detection of TM is of great importance because it is a predictor of lower complete recanalization rates after mechanical thrombectomy and may impact neurological improvement rates.7–10
The main finding of our study is that smaller thrombi (consistent with higher clot burden scores) tend to migrate more often but that this is less likely in patients with good collaterals. Moreover, we found that thrombus histology impacts the risk of thrombus migration, meaning that erythrocyte-rich thrombi migrate significantly more often.
The pathophysiological explanation of this may be that larger clots have stronger adherence with the vessel wall and that good collaterals increase the counter pressure distal of the clot. Our results are supported by an in vitro study that tested the frictions of thrombi with different erythrocyte/fibrin compositions.21 In this study Gunning et al found that the friction of erythrocyte-rich thrombi was lower compared with those with low erythrocyte percentages, suggesting that erythrocyte-rich thrombi are more likely to migrate even though conditions may be influenced by other factors such as haematocrit, blood pressure, and vessel anatomy.21 In addition, patients with a smaller clot burden are more likely to have patent anterior cerebral arteries and posterior communicating arteries, which may result in increased collateral flow. Conversely, higher collateral scores also result in higher clot burden scores through retrograde filling of distal branches in proximal occlusions. These findings are in line with a study of the Mr. Clean trial investigators who demonstrated that patients with lower thrombus burden had higher collateral scores.10 Of note, the results concerning the impact of thrombus migration on clinical outcome are controversial. The study by Alves et al found that thrombus migration was associated with improved functional outcome, although there was an association with worse reperfusion status after thrombectomy.10 In contrast, the study by Kaesmacher et al found that thrombus migration was not associated with clinical outcomes but when defining a difference between baseline NIHSS and NIHSS at discharge of ≥8 or an NIHSS score at discharge ≤1 as “substantial neurologic improvement” there was a significant association with thrombus migration (p=0.048).7 This may be explained by the different patient collectives as the Kaesmacher study similar to ours only included patients with M1 occlusion and used the same approach of assessing direct and indirect signs of thrombus migration.7 In this context, our study establishes an important link between the solely imaging-based results of the Mr. Clean investigators and the Kaesmacher study, and the pathophysiological underlying thrombus composition. Moreover, recent studies found that vascular wall components may be present in retrieved thrombus material, especially in erythrocyte-rich thrombi, if devices reached the M2 segment of the MCA and the P2 segment of the PCA and after a higher number of device passages, whereas successful recanalization tended to be less frequent in patients with thrombi containing vessel wall components.22 Another study found that clot perviousness in CT was associated with higher amounts of erythrocytes and lower fibrin percentages, offering a possible explanation for the higher rates of successful thrombectomy and favorable clinical outcome seen in such patients.23 Of note, in our study the administration of intravenous thrombolysis was not associated with the risk of thrombus migration, which is in line with the results by Kaesmacher et al but contradicts the findings of Alves et al.7 10 A possible explanation may be that in our cohort, like in the Kaesmacher study, only patients with M1 occlusion were included and that the evaluation of indirect thrombus migration was the same as in the study by Kaesmacher et al, whereas Alves at al did not report indirect thrombus migration.7 10 The final goal of further research on the topic of thrombus histology and thrombus imaging would be to combine different admission imaging characteristics for an individual patient selection for thrombectomy (including different devices) and iv-thrombolysis based on thrombus histology.
Our study has some limitations partly attributed to its single center, observational design. Moreover, collateral quality influences the visibility of the distal clot end and therefore thrombus burden assessment. The other limitations include the method of assessing thrombus migration. Contrast may not reach the proximal clot face in all patients on CTA and therefore thrombus migration may be suspected (false positive). Moreover, the results are likely to underestimate the true incidence of TM for several reasons. First, the discrepancies observable between CTA and DSA represent a short time period, and TM may occur prior to imaging (false negative direct evidence for TM). Second, TM may occur in the distal M1 segment without reliable indirect imaging signs because this occlusion pattern lacks end‐artery involvement (false negative indirect evidence for TM). Third, TM may occlude perforators for such a short time that irreversible ischemia does not occur (false negative indirect evidence for TM). Strength of our study is the comparably large consecutive cohort with homogenous histopathological and imaging work-up. To translate our findings into clinical practice the results should be confirmed by prospective multicenter studies which account for different imaging protocols and CT scanner types.
Conclusions
Smaller, erythrocyte rich thrombi tend to migrate more often but this effect is less likely in patients with good collaterals. This supports the hypotheses that larger clots have stronger adherence with the vessel wall and that good collaterals increase the counter pressure distal of the clot but should be confirmed by prospective multicenter studies.
Data availability statement
Data are available upon reasonable request.
Ethics statements
Ethics approval
The study was approved by the Ethics Committee of the University Münster and the Westphalian Chamber of Physicians, Münster, Germany (2017-233-F-S).
References
Footnotes
PBS, HK, HK and UH contributed equally.
Contributors All authors contributed to study design, data collection, interpretation, and writing and revising the manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial, or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.