In this study, we assessed the association between three non-invasive indices of liver fibrosis (APRI, FIB − 4 and NFS) and HT in patients with acute ischemic stroke. The present study of 2033 patients demonstrated high APRI was independently associated with increased risk of HT after adjustment for potential confounders. Subgroup analyses further confirmed these findings. Moreover, the addition of liver fibrosis indices to conventional prognostic factors could significantly improve the risk prediction for HT after ischemic stroke.
Although it is well known that liver fibrosis is associated with cardiovascular diseases[19–22], only a few studies have assessed the relationship between liver fibrosis and stroke specifically. The study by Kim et al. showed a significant positive association between acute ischemic stroke and liver fibrosis. [6] In addition, a Korean study involving 395 patients with ischemic stroke or transient ischaemic attack showed that liver fibrosis was an independent predictor of mortality during long-term follow-up.[9] Recently, a cohort study using data from the Virtual International Stroke Trials Archive–Intracerebral Hemorrhage indicated that two liver fibrosis indices-APRI and FIB-4 were associated with increased risk of mortality in patients with acute intracerebral hemorrhage.[15] To our knowledge, the association of liver fibrosis indices with HT after ischemic stroke has not been reported previously. In the present study, we found that higher APRI was associated with increased risk of HT after ischemic stroke.
Because of an excessive concern over the bleeding risk after thrombolysis among stroke patients, families and physicians[3], it is necessary for us to explore novel and effective biomarkers for HT. Liver fibrosis indices could significantly improve the predictive power for HT beyond established traditional risk factors (NRI: 17.3% for APRI, 21.6% for FIB-4, and 12.1% for NFS, respectively), suggesting that liver fibrosis indices at baseline may be a potential biomarkers in the prediction of HT in patients with acute ischemic stroke. Therefore, we speculated that liver fibrosis indices might be useful in risk stratification of HT after acute ischemic stroke and could help the selection of high-risk patients in clinical practice. If patents have high APRI levels at admission, they may be at high risk of HT and should cautiously receive therapies with bleed risks. Further studies are needed to justify our findings.
The possible explanations for the positive relationship between liver fibrosis and the risk of HT could be considered in several ways. There are common risk factors between liver fibrosis and HT. Liver fibrosis is associated with atherosclerosis[23], diabetes[24], dyslipidemia[25] and arterial stiffness[26], which are associated with HT[27, 28]. Moreover, liver fibrosis is associated with inflammation, oxidative stress and endothelial dysfunction[29], which also participate in the occurrence and development of HT.[1, 30] In our study, the association of liver fibrosis with HT remained still significant in multivariate analysis adjusting for common potential confounders. Future investigation is warranted to explore underlying mechanism.
This study has several limitations that should be mentioned. First, because liver imaging or liver biopsy data are not available in our study, we explored the association between three noninvasive liver fibrosis indices and HT. Therefore, these findings, to some extent, might not reflect the real relationship between liver fibrosis and HT. However, these indices have been proved to have better predictive value for identifying liver fibrosis.[31, 32] Second, alcohol use is an important confounder for liver fibrosis[33], but it is difficult to determine the accurate alcohol consumption in clinical practice. Therefore, we perform subgroup analyses to explore potential interactions, but we could not find significant interaction effect by alcohol consumption. Third, we assessed the relationship between NFS and HT based on 1011 stroke patients because body mass index was not available for other 1022 individuals, which may lead to the potential selection bias.