Guest blog: Cerebral microbleeds following thoracic endovascular aortic repair

Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article. 

Stroke and cerebral damage are frequent findings after thoracic endovascular aortic repair (TEVAR) with a postoperative clinical stroke rate of 3-4% and silent brain infarcts (SBI) in about 80%.(1-4) However, the mechanism of stroke and subclinical cerebral damage in TEVAR is under-investigated. Current clinical research-efforts such as the STEP-registry (strokes from thoracic endovascular procedures) aim to better understand incidence of, and risk factors for stroke and cerebral damage after TEVAR, and to develop strategies for prevention.(5, 6) More than 60% of patients undergoing arch-TEVAR were reported to have SBIs on diffusion weighted magnetic resonance imaging (DW-MRI) despite protective efforts such as carbon dioxide (CO₂) flushing of the endografts.(5, 6) The aim of the current study is to examine the occurrence of CMBs in patients after TEVAR within the STEP-registry and to evaluate their association with patient- and procedural factors.

Ninety-one patients treated with TEVAR in proximal landing zone (PLZ) 0-3 from September 2018 to January 2020 at the German Aortic Center (Hamburg, Germany) and Marie Lannelongue Hospital (Paris, France) were included in the study.(5) The location and number of CMBs were identified and analyzed with regards to procedural aspects, clinical outcome and Fazekas-score as indicator of preexisting vascular leukoencephalopathy.

Indication for TEVAR was type B dissection, degenerative aneurysm or other aortic disease in 44/91 (48.4%), 34/91 (37.3%) and 13/91 (14.3%) patients, respectively. Anatomical details have been described in detail previously.(5)

PLZ were 0, 1, 2 and 3 in 23/91 (25.3%), 10/91 (11.0%), 47/91 (51.6%) and 11/91 (12.1%) patients, respectively. Seventy-one/91 (78.0%) patients were treated in an elective setting. 23 (25%) patients received branched-TEVAR (B-TEVAR), 15 (17%) patients fenestrated-TEVAR (F-TEVAR) of which 4 (4%) patients had in-situ laser fenestrations. Fifty-three/91 (58%) patients received tubular endografts. The median proximal diameter of the aortic endoprosthesis was 38 (34-46) mm and 37/91 (40.7%) patients received a proximal bare stent. Technical success was reported in all cases. Intraoperative complications, such as prolonged hypotension, iatrogenic dissection of the left subclavian artery, aortic rupture during deployment of a stent-graft and proximal common iliac rupture, were reported in 4/91 (4.4%) patients. No periprocedural ischemic stroke or death occurred within 30 postoperative days (POD). 

On MRI performed within 7 POD (Median 4 (2-7)), a total of 1531 CMBs were detected in 58 (63.7%) patients by two neuroradiological experts; bilateral CMBs were identified in 46/58 (79.3%) patients (P=0.078). CMBs were present unilaterally in the right or left hemisphere in 9/58 (15.5%) and 3/58 (5.2%) patients (P=0.0001) respectively. 

More CMBs were found in the middle cerebral territory vs. the posterior territory and the anterior territory ((3.35 (5.56 SD) vs. 2.26 (4.05 SD) vs. 0.966 CMBs (2.87 SD) (P=0.045)),  Procedural factors associated with the presence of CMBs were deployment in zone 0/1 vs. 2/3 (P=0.001), placement of a branched or fenestrated endograft (P=0.025) and longer procedure time (≥120 min) (P=0.019). Proximal diameter of the endoprosthesis ≥40mm (P=0.016), reoperations linked to primary operation (P=0.017) and atheroma grade 4 and 5 (P=0.048) were significantly associated with CMB in a multivariate logistic binary regression model, whereas compliant balloon (P=0.053) use showed only a tendency.  

Multiple linear regression with Firth regression showed more CMBs in TEVAR with proximal diameter ≥40mm OR 6.8 (95% CI 1.65-41.59; P=0.007) and higher DWM Fazekas-score in postoperative MRI OR 2.6 (95% CI 1.06-7.92; P=0.037, indicator of pre-existing vascular leukoencephalopathy). There was no significant correlation between CMBs and SBIs (P=0.376). 

Since there is no pre-operative MRI data to compare to, a causal association between CMBs and TEVAR cannot be proven by the study design. Nevertheless, the observed rate of CMBs after TEVAR (63,7%) was increased compared to rates reported in the literature among a general population of similar mean age (28%).(7) Although the rate among elderly patients with preexisting cardiovascular diseases may generally be higher, a causal linkage between TEVAR and CMBs can therefore also not be ruled by our findings. With the multi-territorial pattern, an embolic and secondary hemorrhagic origin caused by TEVAR may be hypothesized. The middle cerebral artery territory was most frequently affected due to its higher volume and share of blood.(8), (9) CMBs may be of various origin. As SW-MRI is sensitive to para-, dia- and ferromagnetic compounds, the susceptible MRI lesions referred to as “CMBs” may in this postoperative state also result from embolic micromaterial originating from endovascular devices or dispersed microcalcifications from the aortic arch.

We could not identify a significant correlation between SBI and CMBs (P=0.376). In contrast to SBI, which were found predominantly in the left hemisphere, CMBs were detected bilaterally in the vast majority of patients (79.3 %) and unilateral occurrence was more frequent in the right hemisphere (P=0.0001). While cerebral microbleeds are known to occur predominantly in deep or infratentorial regions in the presence of cardiovascular risk factors, and in the temporal lobe in patients with cerebral amyloid angiopathy, a spatial predilection for the right hemisphere in case of unilateral occurrence has so far not been described.(10, 11)

We found an association of CMBs with the PLZ and atheroma grade of the aortic arch matching previously described associations of stroke, SBI and high-intensity transient signals on transcranial doppler with atheroma grade and PLZ. (12, 13)

Although the occurrence of CMBs after TEVAR did not lead to clinically apparent stroke, a potential long-term effect on cognitive function cannot be ruled out. 

The limitations of this study include its retrospective nature and the non-consecutive patient cohort. Pre-operative MRI weren´t available to safely differentiate between procedure-related and preexisting CMBs.  Multivariable and subgroup analysis are limited by low patient numbers and third error in multivariable analysis. Results should be interpreted with caution and only be used to generate hypotheses for future studies. 

CMBs are present bi-hemispherical in the majority of patients after endovascular arch TEVAR and associated with morphological and procedural factors. The clinical importance of this finding needs to be further examined.