Sex Differences in the Association Between Liver Fibrosis and Clinical Outcomes in Acute Cardioembolic Stroke Population With Nonvalvular Atrial Fibrillation

Background: Liver cirrhosis is a conrmed risk factor for worse clinical outcomes of stroke, however the contribution of liver brosis to cardioembolic stroke (CES) and its short-term outcomes are poorly understood. This study aimed to investigate whether liver brosis is associated with more severe stroke, worse short-term clinical outcomes of acute CES, due to nonvalvular atrial brillation (NVAF), as well as the impact of sex on the association. Methods: Using data of 522 patients with NVAF admitted within 48 hours after acute symptom of CES onset. We calculated Fibrosis-4 score (FIB-4) and dened liver brosis as: likely advanced brosis (FIB-4>3.25), indeterminate (FIB-4, 1.45-3.25), unlikely advanced brosis (FIB-4<1.45). We invested the impact of liver brosis degree on stroke severity on admission, major disability at discharge and all cause death at 90 days stratied by sex. Results: Among 522 acute CES patients with NVAF, the mean FIB-4 on admission reected intermediate brosis, whereas liver enzymes were largely normal. After adjusting for possible confounders, multivariate analyses revealed that likely advanced liver brosis was associated with severe stroke (OR=2.21, 95% CI: 1.04-3.54), major disability at discharge (OR=4.59, 95% CI: 1.88-11.18), and 90-days mortality (HR=1.25, 95% CI: 1.10-1.56). Further grouped by sex, these associations were stronger in males but not signicant in females. Conclusions: In patients with largely normal liver enzyme, likely advanced liver brosis is associated with severe stroke, major disability and all cause death after acute CES due to NVAF; the association unfolded more obvious in males, but not for females. for adjusted for age, BMI, lg NT pro-BNP, eGFR, diabetes mellitus, hypertension, hyperlipidemia, ischemic heart disease, prior history of ischemic stroke/TIA, treatment in hospital, NIHSS on admission. ‡ CES, cardioembolic stroke; FIB-4, brosis-4 score; BMI, Body Mass Index; TIA, transient ischemic attacks; eGFR, estimated glomerular ltration rate; NT-pro BNP, N-terminal Pro-B-type Natriuretic Peptide; NIHSS, National Institutes of Health Stroke Scale; HR, hazard ratio; 95% CI, 95% condence interval.


Introduction
Stroke is a common disease, especially the ischemic stroke, which is the second leading cause of death and third leading cause of disability in adults worldwide.(1) Stroke induced by atrial brillation (AF) is more severe, with more threefold mortality and disability rates than patients without AF. (2) The severity, treatments effectiveness, outcomes of stroke may differ from sex. (3,4) Except for sex, liver disease was associated with in-hospital death after ischemic stroke. (5) Several researches suggested that liver cirrhosis was not only associated with an increased risk of stroke, but an independent risk factor of poor prognosis after stoke. (5,6) Although these evidences emphasized advanced liver disease possibly associated with poor stroke outcomes, the implications of subclinical liver disease-liver brosis for CES severity and short-term outcomes is poorly understood.
Liver brosis, the commonly clinical manifestation of chronic liver disease, is often undetected because of no obvious clinical symptoms. (7) Recent studies suggested that there was a signi cant relationship between liver brosis and all-cause mortality of cardiovascular disease in patients with chronic liver disease, (8,9) as well as ischemic stroke risk.(10) As described above, data are lacking regarding to the contribution of liver brosis on the severity and short-term outcomes of ischemic stroke.
Currently, the China, a super aging country, have an increasing population of CES due to AF, especially the NVAF.
Therefore, we focused on the CES patients with NVAF, and investigated the association between liver brosis, quantized by a valid liver brosis indictor-FIB-4, and CES outcomes using patients without overt liver disease.
Furthermore, Because the severity, outcomes of stroke differ from sex, we also researched effect modi cation of sex on those associations. We hypothesized that liver brosis are associated with more severe stroke, worse functional outcomes, higher risk of death at 90 days in CES patients with NVAF.

Study Population
We performed a retrospective study using data from the First A liated Hospital of Xi'an JiaoTong University, a National Advanced Stroke Center which has both an acute stroke treatment center and a stroke rehabilitation center. Thus, all patients suffering acute CES admitted to the stroke center received consistent therapy in the acute and chronic phase during hospitalization. Patients were included in this study if they met the following requirements: (1) 18 years of age or older; (2) has a history of AF or presented symptom onset accompanied by AF; (3) admitted to the stroke center within 48 hours after onset; (4) without any limitation of physical activities before onset; (5) baseline assessment within 48 hours of CES, including symptom, function and imaging assessment; (6) assessment of CES recovery by cerebral imaging during hospitalization; (7) assessment of functional outcome at discharge and all cause death at 90 days after discharge; Exclusive criteria of this study: (1) missing data on liver chemistries, baseline stroke imaging, or important clinical covariates such as cardiac function, renal function, hospital treatment and CES severity; (2)  Statistical Analysis Data are presented as means ± standard deviation or median and interquartile range for continuous variables and percentages for categorical variables. NT pro-BNP were lg-transformed to minimize skewness and treated as continuous variables. The population characteristics were described by FIB-4 score classi cation and strati ed by sex to explore the distribution of each interval. Kruskal Wallis test and/or chi-square test were used to compare differences among the three groups, as appropriate. Multivariate logistic regression analyses for stroke severity on admission and major disability at discharge and COX regression analysis for all-cause death at 90 days from discharge were performed after adjusting for general risk factors. Statistical signi cance was de ned as two-tailed P < 0.05. All analyses were conducted with SPSS 26.0.

Characteristics of Study Population
The dataset of our study consisted of 522 CES patients with NVAF. Standard liver chemistry tests were generally in normal range of the study sample, with 11.7% of aspartate aminotransferase (AST) > 40 IU/L, 11.1% of alanine aminotransferase (ALT) > 40 IU/L. The median value of FIB-4 is 2.28 which re ect intermediate probabilities of brosis. 109, 280, 133 subjects were classi ed into unlikely advanced brosis group, indeterminate group, likely advanced brosis group based on FIB-4 score by analysis, respectively. The population in likely advanced brosis group had higher values for age, CHA2DS2-VASc score, HAS-BLED score and higher proportion for smoker, diabetes melitus, hyperlipidemia, ischemic heart disease, then followed by indeterminate group and unlikely advanced brosis group. No difference was found in sex, BMI, hypertension, prior history of ischemic stroke/TIA, drug use before admission (anticoagulant, antiplatelet and lipid-lowering use) and treatment in hospital (intravenous thrombolysis and endovascular revascularization) among three groups (Table 1). Patients in the likely advanced brosis group had signi cantly higher NIHSS score on admission, mRS score at discharge, 90-days mortality and proportion of severe stroke, major disability compared with those in the indeterminate group and no advanced brosis likely group. The proportion for severe stroke was 32.8%, 34.8%, 46.7%, for major disability is 25.9%, 28.3%, 49.3%, and for 90-days mortality was 5.2%, 3.6%, 8.0% in the no advanced brosis likely group, indeterminate group and the likely advanced brosis group respectively (Table 1).

Characteristics, Stroke Severity and Short-term Outcomes of Study Population Grouped by Sex
When classi ed by sex, the likely advanced brosis group had higher value for age and INR both in males and females, but beyond that, the male patients had higher CHA2DS2-VASc score, HAS-BLED score and higher proportion for smoker, hyperlipidemia, intravenous thrombolysis and endovascular revascularization in likely advanced brosis group, then followed by indeterminate group and unlikely advanced brosis group. No difference was observed in sex, BMI, hypertension, prior history of ischemic stroke/TIA, drug use before admission among three groups both for males and females (Table 2).  Both male and female patients in likely advanced brosis group had the higher NIHSS score, mRS score, more severe stroke, and more major disability than their counterpart (Table 2). For 90-days mortality, 8.0% male patients suffered death in likely advanced brosis group at 90 days, which is higher than other two groups, but there was no signi cant difference among three groups in females (     Table 5).
Whereas the relationship between FIB-4 and severe stroke, major disability and 90-days mortality was not signi cant in the female group, no matter FIB-4 as a continuous or a categorical variable (Model 4, Table 3, 4; Model 2, Table 5).

Discussion
In this study, we found that the higher FIB-4 score was independently associated with more severe strokes on admission, a less favorable functional outcome at discharge and a higher 90-days death risk among patients with CES due to NAVF. More importantly, our study also suggest that sex could modify this association, when classi ed by sex, the association become stronger in male patients, but not signi cant in females. Notably, we also observed these associations in patients with generally normal range of standard liver chemistries.
FIB-4 score is a validated non-invasive tool to assess live brosis in HIV and chronic hepatitis C virus co-infection, hepatitis C mono-infection and non-alcoholic fatty liver disease (NAFLD) populations. (12,13) NAFLD is the main cause of most clinically covert liver brosis, especially after excluding the in uence of alcohol abuse.(23) Liver imaging and pathological diagnosis of those CES patients are not available in this study. However, FIB-4 have been veri ed to have good accuracy in the identi cation of liver brosis in patients with NAFLD.(15) To our knowledge, there was no evidence to con rm the association between subclinical liver disease and CES characteristics due to NVAF and clinical outcomes as well as the sex hybrid effect in present studies. Liver brosis may be associated with the risk of cardiovascular disease or mortality in the general population. (24) Remarkably, in NAFLD, cardiovascular disease is more commonly responsible for death than is liver disease. (8) Although several previous studies have concluded that the severe abnormal liver enzyme may be signi cate risk factor for stroke and short-term outcome, (5,25,26) however they focused on the nonspeci c liver enzyme index and heavy alcohol use were not excluded Or they studied validated live brosis but not concern for liver enzyme index and all type of stroke were included.(5) In contrast, a novel association of severe stroke on admission, major disability at discharge and 90-days death after discharge for the recognized liver brosis indictor (FIB-4) was observed in the CES patients with NVAF of our study which of different pathogenesis from others. Moreover this novel nding revealed that live brosis may represent more severe stroke on admission, worse functional outcomes at discharge and a higher 90-days mortality risk, without obvious clinical manifestations of liver disease, abnormal liver enzyme and con rmed liver disease, which share the same conclusion as those observation-studies that liver brosis could occur without special attention of many subjects, and almost 75% of subjects with liver brosis had normal liver chemistries levels. (23) To sum up, our study presented the evidence that FIB-4 index is independently associated to severe stroke, short-term outcome of CES due to NVAF.
Stroke affected on males and females differently. Stroke risk, severity, reaction to endovascular therapy and outcomes could be differ from sex, and studies generally suggested that older females suffered from more severe stroke, worse prognosis, and higher mortality than males. (3,4,27,28) We also observed the higher NIHSS score, mRS score and more death in women patients (Table S1). Therefore we grouped sex to eliminate the hybrid effect, and found that liver brosis became a stronger risk factor for server stroke, major disability and 90-days death in males, but not in females. The results of the present study was not consistent with the observation among liver brosis population showed that sex would not affect the association between liver disease and stroke mortality. (29) Maybe because of the different research population which they researched almost did not suffer from AF.
The possible mechanisms in the association between live brosis and poor prognosis of CES patients with NVAF may include metabolic pathways, immune-in ammatory and coagulopathy.(29-31) Liver participate to a variable degree in the acute ischemic stroke, and is responsible for the synthesis and metabolism of blood coagulation factors and brinolytic enzymes associated with pathophysiology of stroke.(30, 32) What's more, liver also can produced enzymes proportional to the injury size, to response signals from cerebral infarction.(32) But whether these mechanisms can also apply to explain our ndings referring to subclinical liver brosis remain uncertain and requires further studies to verify.
Our nding about sex can modify the relationship of liver brosis and the short-time outcomes of CES patients with NVAF, which share a different view of another study. (29) The disagreement might be related to discrepancies in several aspects between these two studies, including different study populations, durations of stroke after onset and timings to assess the outcomes. In Baik's study (29), all types of acute stroke were investigated, and the patients they included with younger age, little atrial brillation, and milder function outcome at discharge. Moreover, their follow-up time was over 3 years. Present studies concluded that traditional cardiovascular risk factors varies by sex. (33)(34)(35) Women have speci c risk factors, including gestational hypertension and pre-eclampsia, gestational diabetes, and placental disorders such as intrauterine growth restriction and stillbirth. (36)(37)(38) It is worth noting that females remained an independent poor prognostic factor for CES due to AF, a prospective cohort study revealed the etiological stroke subtypes independently related to unfavorable outcomes at 90 days were CES in women, but large artery atherosclerosis in men. Those seemly suggested that AF affect females more than males in short-term outcome of stroke.(39) So in our study, AF maybe display an important role for the outcomes of females, and liver brosis also presented sex difference related to the outcome of CES due to AF, which may be explained for the comorbidity affect patients differently classi ed by sex (39) and the sex differences in liver function, disease pathogenesis and metabolic genes for maintaining homeostasis in the liver. (40) But this hypothesis still need further studies to verify.
Strengths of our study include the aimed population with strict inclusion criteria, the availability of standardized outcome assessments, and the exclusion of patients with overt liver disease.This is the rst study that shows liver brosis is associated with stroke severity, outcomes due to NVAF, as well as sex could modify the relationship. It con rms and illustrates the high burden of liver brosis in CES patients with NVAF. In our study, almost 90% of subjects with liver brosis had normal liver chemistries levels, indicating even through the stroke patients with normal liver enzyme levels also need accept live brosis evaluation such as the FIB-4 sore, especially for male patients.
There were several limitations in the present study. First, it shares all the limitation of single-center observational study, so the generalizability of our results may be limited. To minimize the biases caused by an observational study, our study included consecutive patients admitted during the study period thereby. Second, although our use of liver brosis indice-FIB-4 is consistent with other studies (41,42) to observe how liver brosis affect other disease processes, but we did not apply advanced liver imaging or liver biopsy to con rm the diagnosis of liver brosis in our study population. Third, self-reported alcohol use may have been imprecise.

Conclusion
We concluded that among CES patients with NVAF, liver brosis indictor-FIB-4 were associated with admission stroke severity, major disability at discharge, and 90-days mortality despite their liver enzymes ranged normally. When classi ed by sex, the association was more signi cant in males but not in females. Our work suggests that new risk assessment and therapeutic targets aimed at liver brosis, indicated as FIB-4 in CES may bene t those who with poor outcomes. Moreover, considering the sex discrepancy, different treatment options will be it will be necessary to make different treatment options in order to improve the prognosis in different sex.