Outcomes in patients with ischaemic stroke undergoing endovascular thrombectomy Impact of atrial fibrillation

.


Introduction
People with acute ischaemic stroke and atrial fibrillation (AF) have greater risk of intracerebral haemorrhage (ICH), higher mortality, and comparable response to thrombolysis compared to non-AF patients. 1These poorer outcomes with AF-related strokes are often caused by the occlusion of large arteries (i.e.middle cerebral artery) resulting in major ischaemic regions, which together with the impaired collateral flow seen in patients with AF, may lead to serious disability and death. 2,3ndovascular thrombectomy (EVT) has been shown to improve clinical outcomes such as functional independence, successful reperfusion, ICH, and mortality in patients with ischaemic stroke. 4Furthermore, the evidence derived from observational studies that investigated the effect of AF on people following an ischaemic stroke who underwent EVT has been inconsistent. 5onflicting results have been reported in previous studies examining associations between AF and outcomes for patients with ischaemic stroke receiving EVT.One large registry study of patients with ischaemic stroke treated by EVT, demonstrated no significant association between AF and functional status, ICH and mortality. 5However, one post-hoc analysis of multi-centre clinical trial of patients with ischaemic stroke, suggested that AF was associated with a higher risk of ICH at 90 days, even in patients with low INR (<1.13). 6herefore, the aim of this study was to investigate the association between AF and ICH and all-cause mortality at 90 days in patients with ischaemic stroke undergoing EVT.

Materials and methods
A retrospective observation study was conducted with data obtained from TriNetX (https://live.trinetx.com),a global health research network that provides access to electronic medical records retrieved from multiple health care organisations, including specialised medical centres, academic and non-academic medical centres, and community hospitals predominantly in the United States.These data include patient demographics, diagnoses (using International Classification of Diseases, 10 th Revision, Clinical Modification [ICD-10-CM] codes), medical procedures, medications, and laboratory measurements.This global federated network provides only anonymous data of electronic medical records with no patient identifiable information and identification of participating health care organisations, thus, research studies using this network do not require institutional review boards for ethical approval.
In this analysis, the database was searched for patients recorded on the TriNetX network meeting the following criteria were included: (1) entered on the register prior to March 10, 2022, (2) aged 18 years or above, (3) ischaemic stroke between September 1, 2018 and September 1, 2021, (4) received EVT treatment (extirpation or dilation procedure code), and (5) had an ICD-10 diagnosis code of AF.Patients in the historical control group met the same criteria as the AF group but did not have an ICD-10 diagnostic code for AF.The diagnostic and treatment codes used for the study is available in the Supplementary Material Table .1. 27 of the 58 health care organisations (HCQs) in the TriNetX network within the study time period had patients which met the study inclusion criteria.Given the cohorts were predominantly derived from the United States, the start date of September 1, 2018 was chosen because the American Heart Association/American Stroke Association guideline for early management of patients with acute ischaemic stroke was originally published in January 2018, 7 and the end date September 1, 2021 was chosen to allow for at least 90 days follow-up for all participants.The 90 days outcome assessment has been recommended as the standard for all clinical and pre-clinical studies and trials intending to demonstrate the long-term benefit of acute treatment in stroke. 8

Statistical analyses
Baseline characteristics were compared using Chi-square x2 tests for categorical variables and independent-sample ttests for continuous variables.Propensity score matching (nearest-neighbour matching with a tolerance level of 0.01 and difference between propensity scores 0.1) was used to control for differences in the comparison cohorts.A standardised mean difference (SMD) <0.1 were considered well balanced and propensity score density graphs were examined.Using logistic regression, patients with ischaemic stroke who underwent EVT and had AF were 1:1 propensity score matched to patients without AF who had an ischaemic stroke who received EVT, for age, sex, ethnicity, hypertensive diseases, ischaemic heart disease, heart failure, previous transient ischaemic attack, hyperlipidaemia, chronic obstructive pulmonary disease, diabetes mellitus, liver disease, chronic kidney disease, obesity, National Institutes of Health Stroke Scale (NIHSS), intravenous thrombolytics (alteplase) and use of oral anticoagulants.These variables were selected as they may effect on the clinical outcome.Following propensity score matching, logistic regression produced odds ratios (OR) with 95% confidence intervals (CIs) for 90-day incidence of ICH (subcortical, cortical, hemisphere, brain stem, cerebellum, intraventricular, multiple localised) and all-cause mortality, comparing stroke patients with AF who underwent EVT with propensity matched controls (without AF).Kaplan-Meier survival curves for all-cause mortality at 90-days from the date of the index stroke by AF status were used.Statistical significance was set at p < 0.05.

Results
Within the TriNetX network at the time of the analysis, there were 1718 patients with ischaemic stroke who underwent EVT and had AF, and 1388 patients with ischaemic stroke who underwent EVT who did not have AF between September 2018 and September 2021 (shown in Fig. 1).Patients with ischaemic stroke who underwent EVT who had AF were significantly older than those with ischaemic stroke and EVT who did not have AF (mean § SD: 73.6 § 12.6 vs. 61.1 § 14.8, p < 0.0001).The patients with AF had a higher prevalence of white patients (77.8 vs. 67.1%,p < 0.0001) and a lower prevalence of black or African American patients (13.9 vs. 22.7%, p < 0.0001).Patients with ischaemic stroke, EVT and AF had a significantly higher prevalence of history of diabetes mellitus, hypertension, transient ischaemic attack, ischaemic heart disease, heart failure, chronic kidney disease, hyperlipidaemia, chronic obstructive pulmonary disease, obesity and intravenous thrombolysis (shown in Table .1) than those without AF who experienced an ischaemic stroke and received EVT.
After 1:1 propensity score matching, there were 832 patients with AF and 832 patients without AF, and the cohorts were well balanced on all included characteristics (all p > 0.05 and SMDs< 0.1, except for the INR level) (shown in Table.1).After propensity score matching, Kaplan-Meier survival analysis showed the odds of 90day ICH were not significantly different for patients with ischaemic stroke who underwent EVT and had AF compared to the propensity score-matched non-AF patients (OR: 0.92 [95% CI 0.68-1.24];survival probability of ICH within 90-day: 87.1 vs. 85.9%,log-rank test p = 0.54, shown in Fig. 2).
After propensity score matching, Kaplan-Meier survival analysis showed odds of 90-day all-cause mortality were lower for AF patients with ischaemic stroke undergoing EVT compared to the propensity score-matched non-AF patients, but this was not statistically different (OR: 0.79 [95% CI 0.63-1.01];survival probability of allcause mortality within 90-day: 79.4 vs. 75.8%,log-rank test p = 0.049, shown in Fig. 3).

Discussion
In the present study of over 1,650 patients with ischaemic stroke treated by EVT in multiple HCOs, AF patients with ischaemic stroke treated by EVT were not significantly associated with greater ICH or allÀcause mortality at 90Àdays than non-AF patients with ischaemic stroke treated by EVT, when propensity score matched for age, sex, race, and NIHSS, prior anticoagulation status and several comorbidities.
AF is one of the leading causes of large vessel occlusion that mandates EVT treatment.A subgroup analysis of the MR CLEAN trial of 135 patients treated by EVT with and without AF demonstrated no significant difference in the effect of EVT treatment in patients with and without AF, despite a non-significant trend towards a decreased treatment effect in AF patients. 9Similarly, a meta-analysis of 1351 patients with and without AF also showed no difference in the treatment effect size of EVT, functional outcome or symptomatic ICH in patients with AF compared to those without AF. 10ndeed, it has been debated whether AF per se increases the risk of ICH or mortality for ischaemic stroke patients undergoing EVT.A post-hoc analysis of multi-centre clinical trials of 140 patients with ischaemic stroke showed that AF was an independent risk factor for any ICH in patients with ischaemic stroke treated by EVT, even in patients with low INR (<1.13). 62][13] These studies were  small in sample size and outcome events, so these findings must be confirmed in larger cohorts.
5][16] Similar to the present study, two previous retrospective observational studies in the United States (n = 562 AF vs 572 non-AF) and China (n = 407) used 1:1 propensity score matching in patients with and without AF demonstrated no significant difference in ICH (at 1-day and 90 days), mortality or good functional status at 90 days. 17,18The present analyses further supports evidence from previous observational studies showing that presence of AF did not impact on the rate of ICH or all-cause mortality within 90 days of follow-up in patients with ischaemic stroke who underwent EVT, when compared to ischaemic stroke patients without AF.

Limitations
This study has several limitations.The ICD codes derived from the electronic medical records are limited and may differ among HCOs, 19 and the impact of attending different HCOs cannot be determined due to data privacy.It is worth noting that there is a possibility of covert AF in the control group that was diagnosed later but not reported or never diagnosed, which could impact the outcome.Similarly, in the AF group, two possible causes of stroke could be due to AF or other atherosclerotic artery occlusions.The detailed cause of death, such as cardiovascular or cerebrovascular mortality, was not available, and Fig. 2. Kaplan-Meier curve depicting the risk of intracerebral haemorrhage events within 90 days after endovascular thrombectomy (EVT) in patients with ischaemic stroke, with and without atrial fibrillation.reported mortality is 'all-cause mortality', and deaths occurring outside the TriNetX network will not be captured.Although this was a multi-HCO, incorporating more than 50 centres, data may not be representative of the broader population of the US.The propensity score was matched for factors such as age, sex, co-morbidities, stroke severity, IVT, and prior anticoagulation status.However, residual confounding may be present as some variables of interest were not available, such as treatment time to therapy (features of treatment time, such as stroke onset to IVT, time to deployment, time to reperfusion, time door in to door out), degree of reperfusion, rate of good reperfusion, and pre-stroke disability (modified Rankin Scale 2-6).Stroke severity score (NIHSS) for assessment of pre-stroke and post EVT functional status was also not available.In addition, details regarding oral anticoagulation, such as timing of oral anticoagulation reinitiation post stroke, were not available.Further, a previous study has reported surrogate measures for functional status (modified Rankin Scale), such as discharge destination (home, relative's or friend's home, rehabilitation facility, and nursing home) at 90 days in the poststroke cohort, 20 but this surrogate marker was not clearly defined in this network.

Conclusion
In patients with ischaemic stroke undergoing EVT, AF was not significantly associated with increased risk of intracerebral haemorrhage or allÀcause mortality at 90Àday followÀup.This study adds support to the existing evidence that AF does not significantly affect outcomes for patients undergoing EVT.Further research may examine different EVT devices on ICH and mortality for patients with ischaemic stroke and AF and whether type of AF plays a role in clinical outcomes following EVT.

Statement of Ethics
Ethical approval for this study was waived by TriNetX.Research studies using the TriNetX federated network do not require ethical approval or patient informed consent as no patient identifiable information is received.

Fig. 3 .
Fig.3.Kaplan-Meier curve depicting the risk of all-cause mortality within 90 days after endovascular thrombectomy (EVT) in patients with ischaemic stroke, with and without atrial fibrillation.

Table . 1
. Baseline characteristics for patients who underwent endovascular thrombectomy and atrial fibrillation and controls who underwent endovascular thrombectomy without atrial fibrillation before and after propensity score matchings.