Early neurological deterioration in cardiogenic cerebral embolism due to nonvalvular atrial fibrillation: Predisposing factors and clinical implications

Abstract Purpose The aim of the study was to investigate factors which may predispose patients to early neurological deterioration (END) and explore peripheral biomarkers for the prediction of END in cardiogenic cerebral embolism (CCE) patients. Methods Patients diagnosed with CCE within 24 hr of onset between January 2017 and January 2019 were included in this study. END was defined as an increase of ≥2 on the National Institutes of Health Stroke Scale (NIHSS) or the emergence of new neurological symptoms within 3 days of admission. Binary logistic regression was used to investigate the factors associated with END. Receiver operating characteristic (ROC) curves were then generated to determine the predictive value of the potential biomarkers and the optimal cutoff values. Results Of the 129 (male, 55.81%; mean age 71.85 ± 11.99 years) CCE patients, 55 patients with END were identified. Hemorrhage transformation (HT), coronary heart disease (CHD), diastolic blood pressure, cystatin C levels, NIHSS score, and platelet‐to‐lymphocyte ratio (PLR) at admission were independently associated with END. A peripheral cystatin C level ≥ 1.41 mg/L and a PLR ≥ 132.97 were predictive factors for END in CCE patients. The lymphocyte‐to‐monocyte ratio (LMR) was negatively independently associated with HT, and LMR < 2.31 may predict the occurrence of HT in patients with CCE. Conclusions Of the potential predisposing factors considered, increased cystatin C and PLR were associated with END within 3 days of CCE, and a decreased LMR may have predictive value for HT in CCE patients.


| INTRODUC TI ON
Cardiogenic cerebral embolism (CCE) accounts for 15%-20% of ischemic strokes (IS). Atrial fibrillation (AF), especially nonvalvular atrial fibrillation (NVAF), is the most common cause of the embolism, accounting for about 50% of cerebral embolisms (Doufekias et al., 2008). Yamanouchi et al. found that 28% of IS in autopsied elderly patients were CCE, 56% of which were caused by NVAF (Yamanouchi et al., 1997). AF is a common arrhythmia, afflicting about 2% of the world's population. In recent years, the incidence of AF has risen. It is expected that by 2050, the prevalence of AF will rise more than 2.5 times, possibly due to an aging population (Miyasaka et al., 2006). The average age of AF patients has also increased steadily, currently ranging from 75 to 85 years (Lip, 2013;Palacio & Hart, 2002). The mechanism underlying cerebral embolism in patients with AF is complex. The main mechanism of pathogenesis involves a change in the structure and function of the left atrial appendage (LAA), which leads to blood stasis and induction of LAA thrombosis (Yamaji et al., 2002). The embolus then falls off and embolizes the intracranial artery. In CCE, the embolus is often large and causes sudden occlusion of the large and medium intracranial vessels (Hong et al., 2013).
In our previous studies, we have found that patients with CCE often have accompanying early neurological deterioration (END).
END in the early phase of acute ischemic stroke (AIS) generally leads to a marked increase in poor prognosis due to diverse mechanisms (Alawneh et al., 2009). However, the predisposing factors associated with END in AIS patients have not yet been fully elucidated and no unified conclusions have been drawn. Previous studies have suggested that many factors may be associated with END, including systolic blood pressure, neurological functional deficits at admission, hyperglycemia, fibrinogen, and hemorrhage transformation (HT; Chung et al., 2015;Dziedzic, 2008;Seners et al., 2014).
Nevertheless, studies on the predisposing factors for END in CCE patients are quite rare. Studies have shown that the incidence of HT in AIS is about 4%-7%, and HT is associated with END (Seet & Rabinstein, 2012;Seners et al., 2015). However, the incidence of and predisposing factors for HT in patients with CCE are also unclear. The aim of this study was (a) to investigate the incidence of and predisposing factors for END; (b) to identify the incidence and predisposing factors for HT; and (c) to examine possible peripheral biomarkers for the prediction of END and HT in CCE patients.

| Study population
From January 2017 to January 2019, 129 consecutive patients with CCE were enrolled in the study at the Department of Neurology and Emergency Department of Shengjing Hospital affiliated with China Medical University. The inclusion criteria were as follows: (a) acute onset, confirmed by cranial CT or MRI, with AIS, and in line with the diagnostic criteria for CCE: Patients may be caused by emboli in the heart. (1) At least one related cardiac source of emboli must be determined; (2) clinical and neuroimaging findings are similar to LAA ischemic stroke; and (3) previous transient ischemic attack (TIA) or stroke in more than one vascular area or evidence of systemic embolism supports the clinical diagnosis of cardiogenic stroke (Quan et al., 2020;Sacco et al., 2013;"Stroke--1989

| Data collection
Demographic variables, traditional risk factors, and routine laboratory test results were collected using the hospital's electronic  (Hacke et al., 1998). The abnormal area of DWI was manually delineated on the MR sequence. In order to calculate the total DWI lesion volume of each patient, the DWI abnormal area was added and multiplied by the section thickness (mm) and the intersection gap (mm). Two neurologists, who were blinded to the clinical data, interpreted all the neuroimaging studies and reached consensus.
All patients in our study had a history of AF or were recently diagnosed with AF after admission. The existence of a history of AF indicated that the patient was clearly diagnosed by a cardiologist and had been received medication or ablation of AF before hospitalization prior to hospitalization in this study. The recent diagnosis of AF indicated that patients with no history of AF were diagnosed by a neurologist at admission based on electrocardiogram or Holter electrocardiogram (Kirchhof et al., 2016).

| Definition of END
Stroke severity was evaluated according to the NIHSS upon admission and at 24 hr postadmission, and both assessments were performed by the same neurologist (Lee et al., 2017). END was defined as follows: (a) an increase of more than 2 points in the total NIHSS TA B L E 1 Demographic, clinical, and laboratory data in cardiogenic embolism patients with and without END score compared to the score at admission; (b) an increase of more than 1 point in the NIHSS specificity subitems, namely level of consciousness (1a-1c) or motor capacity (5a-6b); or (c) new neurological deficits despite no change in the NIHSS score (Seners et al., 2014).

| Treatment
According to the guidelines of the European Heart Association and the European Stroke Society in 2016, anticoagulant therapy was initiated 3 days after stroke onset, and anticoagulant therapy was initiated 2 weeks after onset in patients with severe stroke and a high risk of HT (Kirchhof et al., 2017). Warfarin was given at a dose of 2.5 mg/day, and prothrombin time (PT) and international normalized ratio (INR) were reexamined after 3 days of anticoagulation therapy.

| Statistical analysis
The Mann-Whitney U test or Student's t test was used to compare nonnormally or normally distributed variables, respectively. Continuous variables are presented as mean ± standard deviation if normally distributed; otherwise, the median interquartile range (IQR) is presented.
The categorical variables are expressed in terms of frequency and percentage, and the data were compared using a chi-square test. The in-

| Sample characteristics
In total, 129 patients met the inclusion criteria and were included in this study. The patients had a mean age of 71.85 ± 11.99 years, and 55.81% were male. None of the enrolled patients received intravenous tPA (IVT) and endovascular treatment (EVT). The baseline clinical data and laboratory results are shown in Table 1.

| Demographic, clinical, and laboratory data in cardiogenic embolism patients with and without END
Among the patients, 55 (42.64%) showed END within 72 hr of onset.
We found that the patients with END were older (74.96 ± 12.30 vs. 69.53 ± 11.28; p = .01) and had a higher incidence of hyperten- Abbreviations: CHD, coronary heart disease; Cys C, cystatin C; END, early neurological deterioration; NIHSS, National institute of Health Stroke Scale.

| Demographic, clinical, and laboratory data in cardiogenic embolism patients with and without HT
Since HT is related to END in CCE patients, we determined which factors were associated with HT. We found that 37 ( (Figure 2).

| D ISCUSS I ON
In this study, we first determined the incidence of END in CCE patients and found that HT, CHD, diastolic blood pressure, NIHSS scores, cystatin C levels, and PLR at admission were independently associated with END. Furthermore, we found that cystatin C levels and PLR had value in predicting END. We also investigated factors related to HT in CCE patients and confirmed that LMR was negatively associated with HT and could also predict HT in CCE patients.
A large number of studies have confirmed that END is an independent risk factor for poor prognosis in stroke patients (Hui et al., 2018;Simonsen et al., 2016;You et al., 2019). Consequently, predictive factors for END are also of great interest. It has been reported that large artery atherosclerosis (Kim et al., 2016), initial glycemic variability (Hui et al., 2018), uric acid levels (Huang et al., 2018), mean platelet volumes (Oji et al., 2018), and fibrinogen levels (Lee et al., 2017) (Gill et al., 2016). These conclusions are consistent with our results.
HT is independently associated with END, and the mechanism behind this correlation is complex. Usually, HT occurs in large artery atherosclerosis patients. Reperfusion injury may play a major role, and severe brain edema, damage to the blood-brain barrier, and excessive release of neurotoxic inflammatory mediators after HT may also be involved (Jeon et al., 2014;Juttler et al., 2014;Su et al., 2016). These patients also have higher NIHSS scores, which are consistent with our results. The incidence of AF in patients with CHD is significantly higher than that in healthy people, and CHD usually indicates more severe arteriosclerosis and is associated with poor prognosis in patients with AIS. Our results suggest that CHD is associated with END in patients with CCE (Drakopoulou et al., 2019;Ferrari & Fox, 2016;. Blood pressure (BP) may contribute to the development of END and to poor outcomes by affecting cerebral perfusion (Toni et al., 1998). Other studies have also shown the importance of BP for prognosis after IS (Geeganage et al., 2011;Ko et al., 2010). Our study showed that increased diastolic blood pressure was associated with END in CCE patients, which is consistent with previous F I G U R E 1 Receiver operating characteristic (ROC) curves for serum Cys C (a) and peripheral PLR (b) Cys C, cystatin C; AUC, area under the curve; END, early neurological deterioration; PLR, platelet-to-lymphocyte ratio; CCE, cardiogenic cerebral embolism studies. Several studies have shown that cystatin C is associated with IS and could be considered a risk factor for AIS (Wang, Li, et al., 2019). Additionally, cystatin C is also independently associated with shorter survival in IS patients (Winovich et al., 2017). Kim et al. showed that cystatin C was a useful predictor of END in elderly patients with AIS (Kim et al., 2017). Our study confirmed that cystatin C is not only independently associated with END, but that cystatin C levels ≥ 1.41 mg/L can predict END in CCE patients. The PLR has been used as a convenient novel biomarker indicating inflammation, thrombosis, and plaque instability (Turkmen et al., 2013). Sung et al. reported that the PLR correlated with the severity of neurological impairment in patients after AIS (Sung et al., 2019 HbAlc (%) 6.58 ± 2.24 6.25 ± 1.33 .297 White blood cell count (10 9 /L) 9.93 ± 5.50 8.05 ± 3.36 .020 Neutrophil count (10 9 /L) 7.60 ± 5.28 5.61 ± 3.21 .010 Lymphocyte count (10 9 /L)  (Park et al., 2018). Therefore, it is possible that decreased LMR may also suggest a poststroke inflammatory reaction in HT. Furthermore, this reaction may be involved in infarct evolution and may worsen the clinical outcome of patients with AIS.
The present study has a few limitations. First, it is a single-center, retrospective study, and the sample size was small. Second, the PLR and LMR are dynamic indicators. We only recorded PLR and LMR at admission and did not remeasure them during the period of hospitalization. Third, we focused only on relatively early deterioration, and no long-term follow-up was conducted. Moreover, the sensitivity and specificity of cystatin C, PLR, or LMR are relatively low with all AUC less than 0.7, so their predictive values are of limited.

| CON CLUS IONS
To the best of our knowledge, this is the first study to show that HT, CHD, diastolic blood pressure, cystatin C levels, NIHSS, and PLR at admission are independently associated with END. Furthermore, peripheral biomarkers, such as cystatin C and PLR, were predictive of END in CCE patients. LMR had a negative independent association with HT, and reduced LMR may have predictive value for HT.
However, further studies with larger cohorts are needed to verify our findings.

CO N FLI C T O F I NTE R E S T
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

AUTH O R CO NTR I B UTI O N S
Lin Cong collected clinical data, performed follow-up of patients, completed statistical analysis, and wrote the manuscript. Weining Ma designed the study and contributed to editing the manuscript.
All authors read and approved the final manuscript.

PEER R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1002/brb3.1985.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data of our study will be available via connecting with Dr.