A novel nomogram to predict 90-day mortality in patients with hepatitis B virus-related acute-on-chronic liver failure: a single-center retrospective study

Acute-on-chronic liver failure (ACLF) is a multifaceted condition characterized by progressive deterioration of liver function, multiple organ failure, poor treatment effects, and high short-term mortality [1]. The short-(1–3 months) and medium-term (6 months) mortality of patients with ACLF can reach as high as 50–90%, and the 28-day mortality is 15 fold higher than that of patients with chronic liver disease (CLD) alone [2]. Despite the present popularity of vaccines, hepatitis B virus (HBV) infection still inflicts a devastating impact on human health, affecting approximately 240 million people worldwide [3, 4]. HBV-related ACLF (HBV-ACLF) is a serious complication of chronic hepatitis B (CHB) infection, and can develop at any stage of disease progression [5]. In China, HBV is the major cause of CLD. Moreover, it is estimated that HBV-ACLF accounts for more than 70% of ACLF cases in the region, causing over 100,000 deaths every year [6, 7].

Currently, the treatment of ACLF mainly involves the use of an artificial liver support system (ALSS), cell therapies, and liver transplant (LT) [8]. Nevertheless, an accumulating body of research indicates that ALSS use exerts no significant benefit on the survival of patients with end-stage liver failure [9, 10]. Moreover, because of ethical and safety concerns, the clinical application of cell therapies has so far been limited [8, 11]. In many countries, LT has been applied as a medical intervention to save the lives of patients with ACLF. However, because of the shortage of donor liver grafts, identifying suitable recipients and prioritizing LT is crucial [12].

To date, several predictive scoring systems have been developed to evaluate the prognosis of patients with ACLF, including the Model for End-stage Liver Disease (MELD) score, Chronic Liver Failure Consortium Organ Failure score (CLIF-C OFs), CLIF-C ACLFs, and Chinese Group on the Study of Severe Hepatitis B score (COSSH-ACLFs). However, the MELD score, CLIF-C Ofs, and CLIF-C ACLFs were all established in Western countries, in which the populations are over 70% Caucasians. Furthermore, the main causes of disease were hepatitis C virus and alcohol, rather than HBV. Hence, it is still challenging to use these scoring systems to predict HBV-ACLF prognosis [13, 14]. On the other hand, although the COSSH-ACLFs is a predictive scoring model based on HBV infection, its ability to accurately predict the prognosis of patients with HBV-ACLF is uncertain and requires further verification [15]. In addition, the COSSH-ACLFs also involves multiple parameters and requires complex organ failure assessments. Thus, a more simple prognostic model, which would be more feasible in clinical practice, is needed.

The purpose of this study was to establish a valid but simple prediction tool using laboratory and clinical parameters to predict 90-day mortality in patients with HBV-ACLF and provide guidance for clinical treatment and decision-making.

ACLF is characterized by a rapid progression of multiple organ failure and a low survival rate, making it an urgent global health problem [4, 19]. In Asia, HBV infection and activation are the main causes of ACLF [20]. Further, the mortality of patients diagnosed with HBV-ACLF at admission is high, necessitating further research. It is crucial to develop an accurate and feasible mortality risk prediction model to enable the application of relevant interventions. This will aid in guiding the clinical decision making of medical staff, as well as in ameliorating patient outcomes.

Nowadays, nomograms are widely used as prognostic tools in medical studies. Due to their high accuracy, they play important roles in clinical decision making. In this single-center study, we screened five independent risk factors (age, TBIL, INR, HE, and HDL-C) and established a nomogram to predict 90-day mortality in patients with HBV-ACLF.

First, both the CLIF-C ACLFs and COSSH-ACLFs already include age in their prognostic scoring systems, and both consider old age as a risk factor for poor prognosis in patients with HBV-ACLF. This is logical, as an increase in age leads to significant decreases in liver volume, blood flow, and enzyme metabolism, which reduce the liver’s tolerance to diseases [21]. Second, high serum levels of INR and TBIL generally indicate coagulation failure and liver failure, respectively. Many prognostic studies have reported that the abovementioned parameters are indicators of survival in patients with liver diseases [22, 23]. Furthermore, the parameters are widely used in prognostic scores for patients with ACLF [14, 24, 25]. Additionally, patients with HE, especially those with ACLF, reportedly have higher mortality rates [1, 26]. Verma et al. previously revealed that the presence of HE was independently associated with mortality in patients with ACLF, and that patient mortality tends to increase with higher HE grades [27].

Importantly, HDL-C level, which is not included in any of the currently used prediction models, was considered a risk factor in our prognostic nomogram. It is well-known that a low HDL-C level is an independent predictor of cardiovascular risk [28, 29], and some recent studies have shown that blood lipid levels are decreased in patients with inflammatory diseases [20]. The possible underlying mechanism involves the binding of HDL-C particles to lipopolysaccharides, resulting in the inhibition of the activation of inflammatory factors [30]. According to a cross-sectional study, low HDL-C level is a poor prognostic factor for chronic liver failure [31]. In our study, HDL-C levels in patients with HBV-ACLF were lower than the established normal values (0.78–1.81 mmol/L) in both the survival (0.17, 95% CI 0.10–0.23) and non-survival groups (0.20, 95% CI 0.14–0.27). However, in our prognostic nomogram, we concluded that a lower HDL-C level was beneficial for the survival of patients with HBV-ACLF. This result is contrary to the conventional view that HDL-C is a “good cholesterol” [32, 33]. Indeed, recent studies have reported that HDL-C may not always be a “good cholesterol.” Silbernagel et al. found that the protective effect of HDL-C disappeared in patients with unstable coronary heart disease [34]. Moreover, some studies have shown that, unlike in healthy individuals, HDL-C does not exert any vascular protective effects in patients with kidney disease or diabetes; in fact, increased levels of HDL-C may even have harmful effects [35‐37]. We speculated that in a diseased state, HDL-C may have a negative impact on the human body, owing to disease-induced dysfunction. Therefore, in patients with HBV-ACLF, higher HDL-C levels may be more detrimental to patient survival. Nevertheless, more intensive basic experiments and mechanistic research are needed to clarify the abovementioned speculation. It is still unknown whether HDL-C plays a role in the course of HBV-ACLF.

As observed, the areas under the ROCs of the nomogram and COSSH-ACLFs were very similar. Hence, it could be considered that the predictive abilities of the nomogram and COSSH-ACLFs are comparable. The COSSH-ACLFs is a complex score based on organ failure, comprising 11 parameters which are complicated to evaluate [15]. As Chen et al. stated in their article, pulse oxygen saturation or arterial blood gas analysis were recently used in their center to routinely evaluate the respiratory function of patients with ACLF [38]. The complexity of the COSSH-ACLFs significantly limits its application. In the present study, five independent risk factors were selected to build a prognostic nomogram that is simple, intuitive, precise, and universal.

However, the present study has several limitations. First, this was a single-center, retrospective study, which lead to inevitable selection bias, and the majority of the participants were men. Moreover, we excluded patients who were lost follow-up or with incomplete data, which may have affected the efficacy of the nomogram. Second, the course of ACLF changes dynamically. The data in this study were collected at the time of diagnosis rather than at consecutive time points. Moreover, subsequent changes in treatment and nursing care may lead to variations in relevant parameters. Third, further studies are required to confirm whether HDL-C can be used as a prognostic predictor in patients with HBV-ACLF, and to clarify the mechanism of HDL-C in the pathogenesis of HBV-ACLF. Lastly, although the robustness of our prognostic nomogram was checked via internal validation (bootstrap testing) and external validation, external evaluation with a larger, multi-center population is still required.

In this study, we established and validated a simple, prognostic nomogram to predict 90-day mortality in patients with HBV-ACLF. Additionally, this nomogram can be used for early detection of patients with HBV-ACLF, making clinical decisions, and guiding the allocation of medical resources. Although our nomogram was internally validated by internal verification (1000 bootstrap resampling) and assessed by using another validation cohort, this nomogram still requires external validation before its use can be popularized.