In clinical practice, predicting the occurrence of ACLF within 90 days in patients with LC-AD is extremely challenging [2]. In this study, ECVIC-liver, which was determined using iodine (water) maps obtained in the equilibrium phase of dual-energy CT, and CLIF-C ADs were identified as an independent risk factor for the progression of LC-AD in the univariate and multivariable logistic regression analysis, and the model constructed by the combination of the two could be used for predicting the occurrence of ACLF within 90 days in patients with HBV LC-AD. The performance of model achieved better results in both the training and validation groups, which may help to provide a more appropriate and intuitive assessment tool for the clinical management of LC-AD patients.

The recent study reported that patients with LC-AD who developed ACLF within 90 days had more severe systemic inflammation than patients with stable disease, with a short-term mortality rate of 53.7% [1, 2]. Several previous studies have successfully predicted prognosis in patients with cirrhosis by monitoring liver function, inflammatory status, and portal hypertension [22,23,24]. Therefore, it is theoretically possible to predict the occurrence of ACLF in patients with LC-AD based on pathophysiological status. In this study, significant differences in several clinical indicators (infection, ascites, urea, Na+, ALB, WBC, PLT, etc.) were observed between the progressive and stable groups, which may to some extent reflect changes in systemic inflammation and portal pressure, as suggested in previous studies [25, 26]. However, the use of serum markers and clinical symptoms to assess the prognosis of cirrhosis still has certain limitations (e.g., susceptibility to extrahepatic factors, subjectivity, and instability of evaluation indices), making these data unsuitable for predicting prognosis in patients with LC-AD [9,10,11]. Therefore, we evaluated the development of 90-day ACLF in these patients using the newly proposed CLIF-C ADs, which was also independently associated with patient prognosis. In contrast to conventional CTP, MELD and MELD-Na scores, the CLIF-C ADs was proposed based on LC-AD specifically and takes into account the value of WBC and age for prognostic evaluation [8]. WBC reflects systemic inflammatory status, while age is negatively correlated with total body muscle mass, so a novel score combining these two factors can help to determine prognostic risk in patients with LC-AD [27].

ECV is a marker used to quantify the status of the extracellular matrix, which is an important component of the cellular microenvironment, and the development of disease is usually accompanied by changes in the cellular microenvironment. Therefore, ECV can reflect the extent of tissue inflammation, liver function, and the degree of fibrosis [21, 28, 29]. In the dynamic progression of cirrhosis, the gradual deposition of collagen in the liver leads to gradual expansion of the extracellular gap. After the distribution of the imaging agent in and out of cells and blood vessels during the equilibrium phase reaches stability, ECV can be quantified by evaluating the distribution of the imaging agent during the equilibrium phase [30]. Several studies initially used conventional contrast-enhanced CT to measure ECV, reporting that liver and spleen ECV values allow for non-invasive assessment of the extent of organ fibrosis and the severity of portal hypertension, but with low diagnostic efficacy [13, 31]. Recently, some studies have further used dual-energy CT equilibrium phase iodine maps to measure ECV, which significantly improves the accuracy of extracellular matrix measurements based on dual-energy CT iodine maps when compared with conventional CT values, making them effective tools for the noninvasive assessment of organ fibrosis and prediction of disease progression [17, 32]. In addition, most previous studies [21, 30, 31, 33] have focused on the use of ECV for staging liver fibrosis, grading the severity of portal hypertension, and predicting the occurrence of decompensation and HCC, while none have attempted to predict the occurrence of ACLF within 90 days in patients with LC-AD. Therefore, we used ECVIC-liver values quantified using dual-energy CT to predict short-term disease progression in patients with LC-AD. Our analysis indicated that ECVIC-liver was independently associated with the occurrence of ACLF at 90 days, and that patients with worse prognosis had higher ECVIC-liver values than patients moving toward stable disease. Our model also demonstrated excellent predictive power. In addition, conventional dual-energy CT parameters (IC, Z, K140, etc.) can be obtained simultaneously when measuring ECVIC-liver, enabling the assessment of hepatic blood flow and liver function. However, the above parameters measured in this study were excluded because of the risk of covariance, probably due to the limited sample size in this single-center population and the fact that the study only measured single-phase dual-energy CT parameters of patients, which reflect relatively limited information about the liver, and multiphase dual-energy CT parameters of patients from multiple centers could be included in the future to reflect the liver status more comprehensively.

In this study, ECVIC-liver and CLIF-C ADs increased with disease progression in patients with LC-AD and differed significantly between the progressive and stable groups. Recent studies have reported that the diagnostic efficacy of ECV for predicting prognosis in patients with cirrhosis is comparable or slightly better than that of CTP and MELD scores and that dynamic monitoring of hepatic ECV is beneficial for patient prognosis [33]. Bak et al. [18] demonstrated that hepatic ECV values increased linearly with increasing cirrhosis stage using dual-energy CT, and its sensitivity and accuracy for the non-invasive prediction of liver decompensation and HCC were higher than those for the MELD score. This finding suggests that dual-energy CT is a reliable and sensitive method for assessment and individualized management in patients with compensated cirrhosis. Baldin et al. [27] demonstrated that high CLIF-C ADs are associated with a higher incidence of organ dysfunction, increased risk of complications, and high short-term mortality, allowing for more accurate short-term prognostic prediction. In this study, we further combined the CLIF-C ADs, which reflects the clinical index, and the ECVIC-liver, which reflects the state of the cellular microenvironment, to construct and validate a model for the occurrence of ACLF in LC-AD patients within 90 days. The AUC value of the model in the validation group was 0.838 and a good calibration curve was obtained, indicating that the model has good predictive value, and the satisfactory DCA curve indicates that the model has high clinical value, but future validation with larger sample sizes from multiple centers is still needed.

This study had several limitations. First, the CT scan protocol used in this study was limited to our center, and the results of ECV measured using different delay times may differ. The accuracy of ECV quantified at different delay times with respect to the extracellular matrix should be further explored in prospective studies. Second, this study included patients with HBV cirrhosis only, and the study follow-up period was only 90 days. Further expansion of the study population and extension of the follow-up period are required to identify additional imaging parameters that may reflect liver prognosis, such as liver and spleen volume, liver surface nodule score, paravertebral muscle density, and fat content. Finally, as a retrospective single-center study, the study may be biased due to the limited total sample size, and without external validation, the generalizability of the model requires more in-depth future studies.

In conclusion, the dual-energy CT quantitative ECVIC-liver can achieve the prediction of ACLF occurrence within 90 days in patients with LC-AD; internal validation showed that the model constructed by combining ECVIC-liver and CLIF-C ADs has good predictive performance and clinical utility. Therefore, the results of this study can be used to assist the clinic to better identify the early disease progression in LC-AD patients.