Extent of white matter lesion is associated with early hemorrhagic transformation in acute ischemic stroke related to atrial fibrillation

Abstract Background Hemorrhagic transformation (HT) after stroke, related to atrial fibrillation (AF), is a frequent complication, and it can be associated with a delay in the (re‐)initiation of oral anticoagulation therapy. We investigated the effect of the presence and severity of white matter disease (WMD) on early HT after stroke related to AF. Methods A consecutive series of patients with recent (<4 weeks) ischemic stroke and AF, treated at the Hyper Acute Stroke Unit of the Imperial College London between 2010 and 2017, were enrolled. Patients with brain MRI performed 24–72 h from stroke onset and not yet started on anticoagulant treatment were included. WMD was graded using the Fazekas score. Results Among the 441 patients eligible for the analysis, 91 (20.6%) had any HT. Patients with and without HT showed similar clinical characteristics. Patients with HT had a larger diffusion‐weighted imaging (DWI) infarct volume compared to patients without HT (p < .001) and significant difference in the distribution of the Fazekas score (p = .001). On multivariable analysis, HT was independently associated with increasing DWI infarct volume (odd ratio (OR), 1.03; 95% confidence interval (CI), 1.01–1.05; p < .001), higher Fazekas scores (OR, 1.94; 95% CI, 1.47–2.57; p < .001) and history of previous intracranial hemorrhage (OR, 4.80; 95% CI, 1.11–20.80; p = .036). Conclusions Presence and severity of WMD is associated with increased risk of development of early HT in patients with stroke and AF. Further evidence is needed to provide reliable radiological predictors of the risk of HT in cardioembolic stroke.

While the majority of the HTs are asymptomatic and believed to be innocuous in some studies (Annan et al., 2015;Berger et al., 2001;Libman et al., 2005), HT was associated with adverse clinical outcomes such as mortality and disability in other studies (Dzialowski et al., 2007;Paciaroni et al., 2018Paciaroni et al., , 2008Park et al., 2012). Notably, the presence of HT is associated with a delay in the initiation of oral anticoagulation therapy after stroke, in patients with atrial fibrillation (AF), which may increase the risk of early stroke recurrence (Paciaroni et al., 2018). A better understanding of the factors underlying HT may aid the early identification of patients at high risk of HT leading to an earlier and safer initiation of anticoagulation therapy in selected patients (Paciaroni et al., 2016).
So far, most of the studies on early HT after acute ischemic stroke did not control for the effect of white matter disease (WMD) (He et al., 2019;Kerenyi et al., 2006;Kunte et al., 2012;Marsh et al., 2013Marsh et al., , 2016Muscari et al., 2020;Paciaroni et al., 2008;Tan et al., 2014;Valentino et al., 2017). Moreover, the effect has been investigated in cohorts with mixed stroke etiologies and with a focus on thrombolysis (Bivard et al., 2016;El Nawar et al., 2019;Fierini et al., 2017;Kongbunkiat et al., 2017). Specifically, the association between WMD severity and HT, after stroke related to AF, has not been established, yet the decision to initiate the anticoagulant is particularly relevant in this context.
The aims of the present analysis of the early initiation of direct anticoagulation after stroke in patients with atrial fibrillation (EIDASAF) study were (1) to evaluate the incidence and the characteristics of early HT in patients with acute ischemic stroke related to AF and (2) to investigate the effect of the presence and severity of WMD on the development of early HT.

METHOD
Data included in the present study were derived from the database of the EIDASAF study (D'Anna et al., 2020). EIDASAF was an observa- For the purpose of the present analysis, we considered only patients who had a brain MRI performed 24-72 h from their stroke onset and who had not been started on anticoagulant treatment ( Figure 1) in line with previous studies (Paciaroni et al., 2018(Paciaroni et al., , 2008. The brain MRIs were performed with a 1.5T or 3T MRI machine (Siemens AG, Munich, Germany), following an institutional protocol for acute stroke. This protocol included diffusion-weighted imaging (DWI) at B1000 and apparent diffusion coefficient (ADC) sequences, a fluid attenuated inversion recovery (FLAIR) sequence, a T1, T2 weighted and gradient echo (GRE) sequences (T2-MRI). The following MRI brain variables were analyzed: (1) stroke volume on DWI, (2) subcortical white matter hyperintensity (i.e., leukoaraiosis) rated according to the Fazekas scale (Del Zoppo et al., 1992) and (3) presence of HT. Stroke lesion volumes were delineated manually on DWI sequences on axial slices using MRIcroN (version 12/2009) (Wahlund et al., 2001). The program automatically calculated the infarct volume in milliliter. The Fazekas classification system is a scale ranging from 0 (no WMD) to 3 (high WMD) (Fazekas et al., 1993). HT was defined as hypointensity on GRE sequence, within the infarcted area or outside the infarct zone, corresponding with isoor hypointensity on T1-weighted and iso-or hyperintense signals on T2-weighted/fluid attenuation recovery images detected only on brain MRI performed during hospitalization. HT was categorized into hemorrhagic infarction (HI) or parenchymal hemorrhage (PH) according to the European cooperative acute stroke study (ECASS) classification (Del Zoppo et al., 1992). HI was defined as small petechiae along the margins of the infarct (HI-1) or as more confluent petechiae within the infarcted area but without space-occupying effect (HI-2). PH was defined as hematoma in <30% of the infarcted area with some slight space-occupying effect (PH-1) or as dense hematoma ≥30% of the infarcted area with substantial space-occupying effect or as any hemorrhagic lesion outside the infarcted area (PH-2). In case of more than one hemorrhagic lesion on brain scan, the worst possible HT category was assumed. HT was considered symptomatic if it was not seen on the admission brain scan and there was, subsequently, either a suspicion of hemorrhage or a decline in neurological status (an increase of ≥4 points in NIHSS). These MRI brain scans were inspected and rated by an experienced stroke researcher (MM) who was blinded to clinical information. The study analysis began after the stroke researcher reached high reliability (i.e., intraclass correlation analysis >0.90) on a single researcher's repeat assessments of 50 randomly selected brain scans.

RESULTS
Note that 441 patients were included in the analysis (study flow chart in Figure 1). Ninety-one patients (20.6%) had any HT of which 68 (15.4%) were HI and 23 (5.2%) were PH. Eight patients had a symptomatic HT (two in the patients with HI and six in the patients with PH). The demographics and clinical characteristics of patients with and without HT are shown in Table 1. Patients were similar in both groups but patients with HT more frequently had suffered an intracranial hemorrhage previously (p = .009) and less frequently had peripheral vascular disease (p = .027).
In terms of neuroimaging findings, patients with HT had a larger infarct volume compared to patients without HT (p < .001) ( Table 2).
There was also a statistically significant difference in the distribution of the Fazekas score according to the four different classes in the two cohorts (p = .001) ( Table 2).
The distribution of the Fazekas score across the four different classes differed significantly when we compared patients without HT to patients with HI and PH (p < .001) (Table 3). Notably, all these patients with PH showed a WMD severity score of 3 according to the

Independent predictors of early HT
On multivariable analysis ( Abbreviations: HI, hemorrhagic infarction, for analysis purpose we considered HI-1 and HI-2 together (HI); ICH, intracranial hemorrhage; MT, mechanical thrombectomy; NIHSS, National Institutes of Health Stroke Scale; tPA, tissue plasminogen activator.
On multivariable analysis ( ICH (OR, 7.53; 95% CI, 1.37-41.21; p = .020). We did not find any statistically significant independent predictors of PH (Table 6). Of note, Fazekas score and MT did not show any variation within PH categories; hence, they were automatically excluded from the respective regression model.

DISCUSSION
The main finding of our analysis is that the presence and severity of WMD is associated with an increased risk of development of early HT.
In previous studies, the relationship between WMD and risk of HT was controversial. Paciaroni et al. (2018) did not find a statistically significant association between WMD and HT in patients with acute stroke and AF assessed with a second CT scan or MRI performed 24-72 h from stroke onset. In their study, leukoaraiosis was only dichotomized into absent versus present (Wahlund et al., 2001). A retrospective study of 122 consecutive stroke patients who underwent MRI of the brain 6-60 h after stroke onset did not find that the extent of chronic white matter lesions, classified according to the Fazekas criteria, differed between patients with or without HT (Kablau et al., 2011).
Similarly, Pande et al. (2020) did not show a statistically significant difference in terms of presence of microvascular changes in patients with and without HT. In both studies, the majority of strokes were classified as noncardioembolic. Conversely, another retrospective study of 207 patients presenting with acute stroke and AF and/or rheumatic heart disease evaluated with brain MRI or CT scan up to 7 days after admission supported a potential role for leukoaraiosis in predicting HT after acute ischemic stroke (Wei et al., 2019).
Moreover, Bivard et al. (2016) found an association between small vessel disease and increased risk of HT in 229 patients with ischemic stroke. However, 127 patients (55.5%) had been treated with intravenous thrombolysis before the follow-up scan and only 59 patients (25.8%) had cardioembolic stroke. Despite the limitation of our retrospective study design, our study strongly suggests that WMD has an effect on the frequency and severity of HT in cardioembolic stroke. Compared to previous studies, our patient cohort was rather large and homogenous, and it was limited to AF-related stroke.
Moreover, all included patients underwent a brain MRI 24-72 h from the stroke onset and before the initiation of the anticoagulant treatment.
In previous studies, the proportion of stroke patients with HT varied largely between 50.81% (Wei et al., 2019) and 5.9% (Valentino et al., 2017) in the literature. Difference in the time interval between stroke onset, in the modality of follow-up imaging and in the population of stroke patients included in the studies partially explains the discrepancies between the reported different rates. MRI, as used in our study, is more sensitive than CT for detection of blood components, especially the deoxyhemoglobin and hemosiderin, after acute ischemic stroke (Arnould et al., 2004;Renou et al., 2010).
Similar to previous studies, the volume of the brain infarct (Butcher et al., 2010;Tong et al., 2001)  Our study has several limitations. The reported associations in our nonrandomized study could undoubtedly be influenced by numerous potential confounders even though adjusted statistical models were used to reduce them. The retrospective nature of data collection is another limitation of the study. The participants in this study were all ischemic stroke patients with AF who represent a specific population and may limit the generalizability of our findings. Finally, a long-term follow up was not available for analysis. The strengths of our study include the sample size and the homogeneity of the subjects In conclusion, the presence and severity of WMD was associated with an increased risk of early development of early HT in patients with acute stroke and AF. Our findings provide additional information to predict the occurrence of HT and as consequence they enable safer initiation of anticoagulation therapy, even during the acute phase of cardioembolic stroke. Further evidence is needed to provide reliable radiological predictors of the risk of HT, in cardioembolic stroke, to better characterize the cohort of patients who may benefit from early initiation of anticoagulation.

ACKNOWLEDGMENTS
The EIDASAF study was supported by an unrestricted grant by Bristol-Myers Squibb and Pfizer to Imperial College London (PI: Roland Veltkamp). RV is an investigator of the Imperial BRC and is partially funded by the European Union's Horizon 2020 research and innova-tion program under grant agreement No. 754517. PB is supported by