Article Text

PDF

Original research
Predictors and impact of hemorrhagic transformations after endovascular thrombectomy in patients with acute large vessel occlusions
  1. Yu Bin Lee1,
  2. Woong Yoon1,
  3. Yun Young Lee1,
  4. Seul Kee Kim1,
  5. Byung Hyun Baek1,
  6. Joon-Tae Kim2,
  7. Man-Seok Park2
  1. 1 Department of Radiology, Chonnam National University Medical School, Gwangju, Republic of Korea
  2. 2 Department of Neurology, Chonnam National University Medical School, Gwangju, Republic of Korea
  1. Correspondence to Professor Woong Yoon, Department of Radiology, Chonnam National University Medical School, Chonnam National University Hospital, Gwangju, 61469, Republic of Korea; radyoon{at}jnu.ac.kr

Abstract

Background Predictors and impact of hemorrhagic transformation (HT) after thrombectomy remain to be elucidated.

Objective To investigate the independent predictors and impact of each hemorrhagic infarction (HI) and parenchymal hematoma (PH) after thrombectomy in patients with acute stroke due to intracranial large vessel occlusion (LVO).

Materials and methods We retrospectively reviewed data from 400 patients with acute LVO who underwent thrombectomy. Logistic regression analyses were performed to determine independent predictors of HI and PH on post-treatment CT scans. Associations between HT and poor outcome (modified Rankin Scalescore ≥3) at 90 days were analyzed.

Results HT was observed in 98 patients (62 HIs (15.5%) and 36 PHs (9%)). Independent predictors of HI were male sex, atrial fibrillation, and time from symptom onset to groin puncture. Hyperlipidemia (OR=0.221, 95% CI 0.064 to 0.767, P=0.017) and successful reperfusion (OR=0.246, 95% CI 0.093 to 0.651, P=0.005) were independently associated with a lower chance of PH, while hypertension (OR=2.260, 95% CI 1.014 to 5.035, P=0.046) and longer procedure duration (OR=1.046, 95% CI 1.016 to 1.077, P=0.003) were independently associated with a higher chance of PH. Only PH (OR=10.154, 95% CI 3.260 to 31.632, P<0.001) was an independent predictor of poor outcome.

Conclusions PH is independently associated with poor outcome, whereas HI does not predict outcome after thrombectomy in patients with acute LVO. Our findings suggest that rapid and successful reperfusion is essential to prevent PH in patients undergoing thrombectomy for acute LVO. In addition, our study suggests that hyperlipidemia is associated with a lower risk of PH in such patients.

  • stroke
  • thrombectomy
  • hemorrhage

Statistics from Altmetric.com

Introduction

Endovascular thrombectomy using a stent retriever with or without preceding intravenous thrombolysis is now accepted as a standard treatment for selected patients with acute ischemic stroke due to intracranial large vessel occlusions.1 Despite its efficacy, endovascular thrombectomy is more likely to increase hemorrhagic complications than conservative management.2–4

Post-treatment hemorrhagic transformations can be classified into hemorrhagic infarction (HI) defined as stippled petechial hemorrhages within the infarction without mass effect, and parenchymal hematoma (PH) defined as a dense hematoma with a space-occupying effect.5 Knowledge of the clinical consequences and predictors of hemorrhagic transformations is important for accurate prognosis in patients undergoing endovascular thrombectomy. However, the clinical relevance of each hemorrhagic transformation after endovascular thrombectomy remains unclear. In addition, data are limited on the independent predictors of each hemorrhagic transformation after modern thrombectomy in patients with acute ischemic stroke. Accordingly, we aimed to investigate the clinical impact and independent predictors of hemorrhagic transformations (HI and PH) in patients who received endovascular thrombectomy owing to intracranial arterial occlusions.

Methods

Patients

Between January 2011 and February 2016, 426 consecutive patients with acute stroke underwent endovascular thrombectomy at a tertiary stroke center. Of these patients, those who presented with a pre-stroke modified Rankin Scale (mRS) score of ≥3 (n=15), who had infarctions in multiple arterial territories (n=8), and who were lost to follow-up (n=3) were excluded from the analysis. For the remaining 400 patients, we retrospectively analyzed clinical and procedural data, which were prospectively recorded in the database at our institution. Clinical data included demographic features, vascular risk factors, use of antiplatelet or anticoagulant medication at the time of stroke, admission National Institutes of Health Stroke Scale (NIHSS) score, intravenous thrombolysis before thrombectomy, admission systolic and diastolic blood pressure, and admission serum glucose level. Patients were diagnosed as having hypertension or diabetes according to the history at presentation or use of antihypertensive or diabetes medication. Hyperlipidemia was defined as hypercholesterolemia (total cholesterol >220 mg/dL, low-density lipoprotein >160 mg/dL, or triglyceride >150 mg/dL) or use of cholesterol-lowering medication. Smoking was considered in current smokers or those who had stopped <5 years ago. Atrial fibrillation was defined as documented history or diagnosis during hospitalization. Coronary artery disease was defined as documented history of myocardial infarction or angina pectoris. Previous stroke was defined as history of prior ischemic stroke or transient ischemic attack. Admission hyperglycemia was defined as serum glucose >140 mg/dL.6

Patient selection criteria for endovascular thrombectomy included angiographically confirmed intracranial large vessel occlusion, groin puncture that could be initiated within 6 hours from stroke onset for anterior circulation stroke and within 12 hours for posterior circulation stroke, and an initial NIHSS score of ≥4. Exclusion criteria included the detection of intracranial hemorrhage on pretreatment imaging study, acute anterior circulation infarction larger than one-third of the middle cerebral artery territory on CT or MRI, acute posterior circulation infarction with diffuse bilateral pontine ischemic change on diffusion-weighted imaging, and pre-stroke mRS score of >2. The institutional review board approved the study and waived the requirement to obtain informed consent for participation because of the retrospective design of the study.

Endovascular therapy

According to international guidelines, eligible patients who could be treated within 4.5 hours of stroke onset received intravenous recombinant tissue plasminogen activator before the start of endovascular therapy. Cerebral angiography and endovascular therapy were completed under conscious sedation. Stent retriever thrombectomy was performed as the first-line thrombectomy technique. When repeated stent retriever thrombectomy failed to achieve successful reperfusion, secondary contact aspiration thrombectomy with an intermediate catheter or Penumbra reperfusion catheter was performed. The details of thrombectomy techniques have been described previously.7 8

Intracranial angioplasty with or without stenting was performed when severe (≥70%) underlying intracranial atherosclerotic stenosis of the target artery was seen during the thrombectomy. Endovascular thrombectomy was subsequently performed after carotid stenting if patients had a steno-occlusive lesion at the ipsilateral cervical internal carotid artery.

All patients underwent non-enhanced CT scans immediately after and at 24 hours after endovascular therapy. Procedure duration was defined as the time from groin puncture to final angiography. Reperfusion status was assessed on the final angiogram and graded according to the modified Thrombolysis in Cerebral Infarction (mTICI) scale.9 Successful reperfusion was defined as mTICI grade 2b or 3.

Outcome measures

Hemorrhagic transformation was classified into HI or PH according to European Cooperative Acute Stroke Study classification.10 HI was defined as a heterogeneous hyperdensity occupying a portion of an ischemic infarct zone without mass effect, whereas PH was a more homogeneous, dense hematoma with mass effect. A hyperdense lesion detected on the immediate CT scan that disappeared on a follow-up scan obtained 24 hours after treatment was considered to be a contrast staining lesion.11 CT images were independently assessed by two neuroradiologists blinded to clinical information and angiographic findings. Final decisions were made by consensus.

Symptomatic hemorrhage was defined as hemorrhagic transformation that was associated with neurological worsening (≥4-point increment in the NIHSS score). Neurological examination was performed by neurologists after endovascular therapy, at 24 hours and 90 days after stroke, and when any changes in patient’s symptoms ensued. Functional outcome assessment was performed using the mRS at the outpatient clinic 90 days after the stroke. If patients could not attend the outpatient clinic, the mRS score was assessed by telephone interview. A poor functional outcome was defined as a mRS score of 3–6 at 90 days.

Statistical analysis

Univariate analysis was performed using the χ2 or Fisher’s exact test for categorical variables and the Mann–Whitney U test for continuous variables to identify predictors of HI and PH. Then, the independent predictors of HI and PH were determined using a binary logistic regression analysis, respectively. The variables tested in the multivariate binary logistic regression models were those with a P value <0.2 on univariate analysis. Finally, a binary logistic regression analysis was conducted to identify the predictors of poor outcome at 90 days. A P value <0.05 was considered significant. Statistical analyses were performed using SPSS V.23.0 (SPSS Inc, Chicago, Illinois, USA).

Results

Of the 400 patients included in this study, 236 had occlusions in the middle cerebral artery, 105 in the internal carotid artery, and 59 in the basilar artery. Overall, successful reperfusion was achieved in 83.8% (n=335/400) and a good outcome in 46.5% (n=186/400) of patients. The mortality rate was 11.3% (n=45/400). Eighty-one patients (20.3%) underwent contact aspiration thrombectomy as a secondary approach after failure of the first-line approach. Underlying intracranial stenosis was found in 55 patients (13.8%). Thirty-seven (9.3%) had a coexisting steno-occlusive lesion at the proximal cervical internal carotid artery. Overall, HI occurred in 15.5% (n=62/400), PH in 9% (n=36/400), and symptomatic hemorrhage in 3.5% (n=14/400) of patients.

Table 1 shows a comparison of characteristics between 62 patients with HI and 302 patients without HI or PH. Diabetes was more common in patients with HI compared with those without HI or PH (30.6% vs 18.9%, P=0.038). Patients with HI had longer procedure duration (270 min vs 230 min, P=0.002) and time to reperfusion (304.5 min vs 260.5 min, P=0.004) compared with those without HI or PH. On multivariate binary logistic regression analysis adjusting for potential confounders (diabetes mellitus, systolic blood pressure at admission, underlying severe intracranial stenosis, cardioembolism, and use of secondary contact aspiration thrombectomy), independent predictors for HI were male sex (OR=1.825, 95% (CI) 1.022 to 3.260, P=0.042), atrial fibrillation (OR=2.192, 95% CI 1.201 to 4.001, P=0.011), and time from onset to groin puncture (OR=1.005, 95% CI 1.002 to 1.008, P=0.002).

Table 1

Univariate analysis for predictors of hemorrhagic infarction after endovascular thrombectomy

Table 2 shows a comparison of characteristics between 36 patients with PH and 364 patients without PH (including 62 patients with HI). Hyperlipidemia was less common (8.3% vs 26.6%, P=0.015) but procedure duration was longer (30.5 min vs 25 min, P=0.019) in patients with PH than in those without it. On multivariate logistic regression analysis adjusting for potential confounders (admission hyperglycemia, ongoing antiplatelet medication, use of intravenous thrombolysis, and secondary contact aspiration thrombectomy), independent predictors for PH were presence of hypertension (OR=2.260, 95% CI 1.014 to 5.035, P=0.046), absence of hyperlipidemia (OR=0.221, 95% CI 0.064 to 0.767, P=0.017), longer procedure duration (OR=1.046, 95% CI 1.016 to 1.077, P=0.003), and successful reperfusion (OR=0.246, 95% CI 0.093 to 0.651, P=0.005).

Table 2

Univariate analysis for predictors of parenchymal hematoma after endovascular thrombectomy

Multivariate binary logistic regression analysis (table 3) showed that independent predictors of poor functional outcome (mRS score 3–6) were older age (OR=1.038, 95% CI 1.015 to 1.063, P=0.001), presence of diabetes (OR=2.044, 95% CI 1.093 to 3.824, P=0.025), higher NIHSS score on admission (OR=1.112, 95% CI 1.053 to 1.173, P<0.001), longer procedure duration (OR=1.016, 95% CI 1.003 to 1.030, P=0.018), unsuccessful reperfusion (OR=3.709, 95% CI 1.729 to 7.956, P=0.001), and occurrence of PH (OR=10.154, 95% CI 3.260 to 31.632, P<0.001). Multivariate binary logistic regression analysis was adjusted for potential confounders such as hypertension, atrial fibrillation, previous stroke, admission hyperglycemia, internal carotid artery occlusion, and underlying severe intracranial stenosis.

Table 3

Univariate and multivariate binary logistic regression analysis showing independent predictors of 90-day poor outcome in 400 patients undergoing thrombectomy

Discussion

Our study suggests that PHs are strongly associated with poor functional outcome, whereas HIs do not independently predict outcome in patients undergoing endovascular thrombectomy for acute ischemic stroke. Only 11% of patients with PH had a good outcome at 90 days, and PH was the most powerful independent predictor of poor outcomes (OR=10.154, 95% CI 3.260 to 31.632, P<0.001). An association between the occurrence of PH and poor outcomes after endovascular treatment has been found in previous studies.12 13 Nogueira et al reported that PH was an independent predictor of both 90-day poor outcome (mRS score ≥3) and 90-day mortality in patients with acute anterior circulation stroke who had various types of endovascular treatment.12 Kaesmacher et al recently showed that PH is independently associated with poor neurologic outcomes (defined as discharge NIHSS score ≥5) after modern thrombectomy in patients with isolated middle cerebral artery occlusion.13 Of interest, these two studies found that HI was also independently associated with poor 90-day and neurologic outcomes. In contrast, our results support the notion that HI is a relatively benign imaging characteristic. In our study, patients with HI did tend to have a lower rate of good outcomes (40.3% vs 51.9%, P=0.094) than those without hemorrhagic transformations. However, after adjustment for potential confounders, HI was not an independent predictor of poor outcome. Additional studies are warranted to determine whether HI represents a prognostic marker of endovascular thrombectomy in patients with large vessel occlusions.

Previous groups have identified several independent predictors of PH after endovascular stroke treatment, such as atrial fibrillation, longer time from onset to treatment, lower baseline Alberta Stroke Program Early CT score (ASPECTS), and wake-up stroke.12 13 Our study additionally found that hyperlipidemia and successful reperfusion are independently associated with a lower chance of PH while a longer procedure duration is independently associated with a higher chance of PH.

Although previous findings have suggested that low total cholesterol and low-density lipoprotein values are associated with increased risk for hemorrhagic transformation after acute ischemic stroke,14–16 the association between hyperlipidemia and lower hemorrhagic risk following endovascular thrombectomy has not been reported previously. Our study is the first to demonstrate that hyperlipidemia has a paradoxically protective effect on the occurrence of PH after modern endovascular thrombectomy in patients with acute large vessel occlusion. The exact mechanism of association between hyperlipidemia and decreased risk of hemorrhagic complications in patients with acute ischemic stroke remains unclear. It has been suggested that hyperlipidemia may have a protective role on the integrity of cerebral small vessels.17 18 Hyperlipidemia also has antioxidant activity by neutralizing free radicals and anti-inflammatory activity by downregulation of vascular endothelial growth factor.19 In addition, concomitant statin use has similar neuroprotective effects, such as endothelial protection and anti-inflammatory activity.20–22 Prospective studies in a larger cohort would be needed to confirm the relationship between hyperlipidemia and occurrence of PH in patients undergoing endovascular thrombectomy.

Previous studies have suggested that successful reperfusion has a protective effect on the risk for hemorrhagic transformation after endovascular treatment. Kaesmacher et al showed in a cohort of 299 patients that complete reperfusion (TICI 3) was independently associated with a lower risk for HI (OR=0.408, 95% CI 0.210 to 0.789, P=0.008) following endovascular thrombectomy.13 Wang et al reported that successful recanalization (TICI 2b or 3) was independently associated with any hemorrhagic transformation (OR=0.42, 95% CI 0.19 to 0.95, P=0.038) following endovascular stroke therapy, including intra-arterial thrombolysis and mechanical thrombectomy using a MERCI device, Penumbra, or Solitaire stent.23

Our results add to the existing evidence of benefits of successful reperfusion as measured by hemorrhagic complications in patients undergoing endovascular stroke treatment. Similar to previous studies, successful reperfusion was independently associated with a lower risk for PH (OR=0.246, CI 0.093 to 0.651, P=0.005) after modern thrombectomy. This protective effect of successful reperfusion might be caused by a substantial reduction in the severely ischemic regions that are prone to developing disruptions of the blood–brain barrier.

We also found that longer procedure duration was independently associated with an increased PH risk. This observation is consistent with the results of a recent study by Alawieh et al that included 197 patients with acute stroke treated with a direct aspiration first-pass technique (called ADAPT).24 They showed that patients in the late recanalization group (procedure duration >35 min) had a higher incidence of PH (PH2 type) than those in the early recanalization group (procedure duration ≤35 min) (13.3% vs 2.5%, P=0.006). In addition, they found a significant positive correlation between procedure duration and the incidence of PH2 (Pearson’s correlation, R2=0.447, P=0.03).

A history of diabetes mellitus, atrial fibrillation, longer time from symptom onset to groin puncture, intravenous thrombolysis before endovascular therapy, higher baseline NIHSS score, and higher admission glucose level all are identified as predictors of HI following endovascular stroke treatment.12 13 Similarly, our study found that a history of atrial fibrillation, longer time from symptom onset to groin puncture, and male sex were independently associated with HI occurrence. Although the relationship between male sex and risk for hemorrhagic complications has not been reported in patients undergoing endovascular treatment, this association has been found in patients managed with medical treatment. Kalinin et al, in their study of 535 patients with acute middle cerebral artery stroke, found that male sex was independently associated with any hemorrhagic transformation (OR=2.41, 95% CI 1.12 to 5.15, P=0.027).25

Our study has some limitations. It was a retrospective single-center study, which may compromise the external validity of the results. We did not differentiate types of HI and PH because the numbers of cases in each category were small. In addition, several potential predictors, such as international normalized ratio, baseline ASPECTS, and collateral status, were not assessed.

Conclusions

PHs on post-treatment CT scans are independently associated with poor functional outcome, whereas HIs do not independently predict outcome after endovascular thrombectomy in patients with acute ischemic stroke. Our study suggests that achievement of rapid and successful reperfusion is essential to prevent occurrence of PH in patients undergoing endovascular thrombectomy owing to acute large vessel occlusions. In addition, our study suggests a new finding that hyperlipidemia is associated with a lower risk for PH in such patients.

References

View Abstract

Footnotes

  • Contributors YBL: acquisition, analysis, and interpretation of data; writing of the manuscript. WY: conception and design of the study; analysis and interpretation of data; writing and revision of the manuscript. YYL, SKK, BHB, J-TK, MSP: acquisition, analysis, and interpretation of data. All authors approved the final version to be published.

  • 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 WY reports being a consultant to Stryker Inc.

  • Patient consent Not required.

  • Ethics approval This study was approved by Chonnam National University Hospital (CNUH) institutional review board.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

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.