Variceal hemorrhage (VH) is a serious complication of cirrhotic portal hypertension and one of the leading causes of death worldwide (La Mura et al., 2020). The annual incidence of first bleeding events in patients with gastroesophageal varices (GEVs) is 5–15%, and more than 15% of the initial bleeding episodes are fatal (Shukla et al., 2016; Haq and Tripathi, 2017). Despite progress in diagnosis and therapy of VH, the mortality from the initial bleeding episode remains high (Reverter et al., 2014). Even if the bleeding is controlled, the patients still have very high-risk of recurrent bleeding, which is associated with mortality as high as that of the first bleed (O’Brien et al., 2013). Hepatitis B virus (HBV) infection is a serious global public health problem (Revill et al., 2020), and a large proportion of GEV cases is associated with HBV infection (Liu et al., 2016; Lv et al., 2019). Reducing the incidence of HBV-related VH is important to decrease the overall mortality rate associated with cirrhosis. For this purpose, routine bleeding risk assessment in patients with HBV-related GEVs is essential such that appropriate prophylactic measures are administered to avoid the first VH episode occurrence.

Hepatic venous pressure gradient (HVPG) is considered an excellent VH predictor (La Mura et al., 2020). Patients with HVPG ≥12 mmHg face high VH risk; in contrast, this bleeding risk significantly decreases when HVPG is <12 mmHg or when the HVPG is reduced by more than 20% from the baseline level (Eisenbrey et al., 2013; La Mura et al., 2020). However, the HVPG measurement is an invasive procedure, which is not suitable for routine clinical testing in most patients. Generally, the stratification strategy for GEVs is mainly based on endoscopic screening, which is currently recommended by the guidelines for cirrhotic patients (Garcia-Tsao et al., 2017). The size of varices >5 mm and appearance of red wale marks (RWM) are considered as high-risk features of VH (The North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices, 1988). However, a study reported that only 30% of patients with actual bleeding presented with these endoscopic risk features (The North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices, 1988). Therefore, endoscopic screening alone may not be enough to accurately identify patients with high-risk of VH occurrence.

The combination of the clinical indicators and endoscopic features is considered as the appropriate tool to provide a better assessment of VH risk (The North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices, 1988). Currently, the most widely used indexes to stratify high-risk patients are the North Italian Endoscopic Club described NIEC index and revised (Rev)-NIEC index that was proposed by Carlo Merkel et al. (The North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices, 1988; Merkel et al., 2000). Both indexes are a combination of the Child-Pugh classification and endoscopic parameters including the size of varices and RWM (The North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices, 1988; Merkel et al., 2000). However, both the NIEC and Rev-NIEC indexes were developed primarily based on the data on populations with alcohol issues and/or hepatitis C virus infections. Whether they can accurately predict the VH risk of HBV-related GEV patients is unknown. In addition, these models were proposed more than two decades ago. Considerable advances have been made in liver disease treatment in the past 20 years, especially toward the antiviral treatment of HBV infection (Wang and Duan, 2021). As the medical environment changes, the predictive performance of these models may also change. Moreover, despite the greatly improved outcomes of antiviral therapy for HBV-associated cirrhosis, the incidence of HBV-related VH and the associated mortality remains high (Kim et al., 2011). Therefore, it is necessary to develop a new prediction model to predict the risk of first VH occurrence in HBV-related cirrhotic patients with GEVs, especially in the context of widespread antiviral therapy.

Nomograms are visualization tools of prognosis evaluation, which can accurately and quantitatively predict the prognosis for individual patients (Iasonos et al., 2008; Balachandran et al., 2015). Currently, nomograms to predict clinical outcomes have been widely used in many medical fields to help in decision-making (Liang et al., 2015; Guo et al., 2022). However, to date, no ideal nomogram has been established to predict the risk of HBV-related VH. In the current study, our aim was to develop a nomogram specifically dedicated to predict the risk of first VH for HBV-related cirrhotic patients with GEVs and to validate its predictive performance in an independent group of patients.

In the current study, we developed and validated an individualized nomogram model specifically for HBV-related cirrhotic patients with GEVs to assess the risk of first VH occurrence within one-year. The robustness and predictive power of this nomogram was demonstrated to be superior compared with the traditional NIEC and Rev-NIEC indexes by internal and external validation tests. Importantly, this nomogram could be used as a risk stratification tool to identify patients at different risk levels of VH occurrence, which helps to promote the precise prevention of VH occurrence and personalized management of patients with GEVs.

VH prevention is the primary goal in the GEV patient management (Garcia-Tsao et al., 2017; de Franchis et al., 2022). Several decades ago, researchers noticed the importance of bleeding risk assessment in the development of prevention strategies and have established several models for this purpose (The North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices, 1988; Merkel et al., 2000; Tacke et al., 2007). Among them, the NIEC index is the most widely used, and the risk stratification based on this index is still an important basis for primary prevention strategies (The North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices, 1988; Garcia-Tsao et al., 2017; de Franchis et al., 2022). However, the predictive power of such models are expected to vary among different populations owing to differences in the etiologies of cirrhosis. Based on our research, the predictive performance of the traditional NIEC and Rev-NIEC indexes for patients with HBV-related GEVs is lower than that for the patients with hepatitis C virus/alcohol abuse-related GEVs studied in the original report (The North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices, 1988; Merkel et al., 2000). Furthermore, it is reported that the etiology is an independent VH predictor (Tacke et al., 2007), indicating different progression of VH between different etiologies. Therefore, different evaluation model based on different etiologies may be more effective for VH risk prediction. For patients with HBV infection, antiviral therapy is a very important strategy regardless of the disease stage, which can delay disease progression and improve long-term prognosis (Iloeje et al., 2006; Chen et al., 2006; European Association for the Study of the Liver, 2017; Wang and Duan, 2021). Previous studies have also demonstrated that antiviral therapy is effective in reducing the VH rate (Li et al., 2013; He et al., 2019). Nevertheless, the incidence and mortality of HBV-related VH remains high (Kim et al., 2011). Thus, apart from the antiviral therapy, it is essential to develop other targeted prevention strategies in accordance with bleeding risk of each individual, for reducing the HBV-associated first VH occurrence.

In the current study, we carefully analyzed data from 527 HBV-related cirrhotic patients with GEVs and explored the VH-related impact factors based on one-year follow-up outcomes. As expected from previous literature (The North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices, 1988; Kleber et al., 1991; Merkel et al., 2000; Kim et al., 2019), the sizes of varices as well as the presence of RWM were strong predictors for VH. Endoscopy, as the gold standard technique for the diagnosis of varices, plays an important role in the bleeding risk assessment for patients with GEVs (de Franchis et al., 2022). Patients with large varices have approximately three times higher risk of VH than those with small varices, while the presence of RWM increased the risk of bleeding up to four times (Park et al., 2004; Aggeletopoulou et al., 2018). Further, the liver dysfunction severity, e.g., elevated GGT levels and ascites, was found to be an essential risk factor for VH. However, the liver function indicators were different from the previous studies (The North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices, 1988; Merkel et al., 2000), in which the Child-Pugh classification as a liver function indicator was associated with bleeding. This may be because the Child-Pugh scoring system involves some subjective indexes (hepatic encephalopathy and ascites) and interrelated indexes (serum albumin and ascites) (Pugh et al., 1973), which virtually increase the instability of the prediction in different studies. It has also been previously reported that the Child-Pugh classification is not associated with the bleeding incidence (Kleber et al., 1991). Besides such well-recognized factors mentioned above, HCT and splenic thickness were also found as important factors affecting VH occurrence. HCT is one of the most important indicators of whole blood viscosity (Voerman and Groeneveld, 1989). A decreased blood viscosity is associated with higher bleeding risk and increased bleeding severity (Ohki et al., 1988). Previous reports also indicated that low hematocrit is a risk factor for variceal bleeding (Liu et al., 2006; Zhou et al., 2017). Splenomegaly, excluding diseases of the blood system, is an important manifestation of portal hypertension in cirrhosis (Li et al., 2017). The increased spleen thickness indicates the severity of liver morphological changes, indicating the portal hypertension and gastroesophageal variceal bleeding progression (Bolognesi et al., 2002; Berzigotti et al., 2008; Zhang et al., 2019).

We combined the above mentioned six independent predictors and established a novel and easy-to-use nomogram to evaluate the probability of the first VH occurrence. To our knowledge, this is the first report on developing a nomogram for predicting the risk of VH in HBV-related cirrhotic patients with GEVs. This established nomogram demonstrated better predictive performance for VH occurrence compared with the traditional NIEC and Rev-NIEC indexes, as supported by the AUC and DCA curves in the internal and external validation tests. Thus, this nomogram developed specifically to report that GEVs screening strategies for HBV-related cirrhosis offers unique advantages over the traditional risk predicting models.

For efficient prophylaxis, accurate risk stratification and prognosis assessment are important because patients in the high-risk group may benefit from earlier prevention strategies, while patients in the low-risk group could avoid undergoing unnecessary interventions (Tacke et al., 2007). In our study, we stratified patients into low- (<7 points), medium- (7–19 points), and high- (>19 points) risk groups according to their bleeding risk scores. In this hierarchical model, 26.8% of patients were categorized as low-risk group with an annual bleeding risk of 0.7%. Furthermore, the low cut-off point of 7 offered 98.2% sensitivity and 99.3% NPV for VH prediction in the derivation cohort, and 100% sensitivity and NPV in the validation cohort, which means that endoscopic screening and preventive measure should be reduced in patients with low-risk group features to avoid the risk of overdiagnosis and overtreatment. In contrast, the 26.8% annual VH incidence in the high-risk group signifies that patients in this group require closer surveillance and proactive preventive strategies, and it can be accepted that the higher the risk score, the more attention should be given. Therefore, the clinical significance of this study lies in the idea that clinicians can use this nomogram for assistance in understanding the risk for VH and choosing appropriate and individualized treatment strategies according to patient risk level, instead of adhering to a single treatment regimen, which may contribute to improved overall outcomes in all patients with GEVs and allocate medical resources efficiently.

This study has several limitations. First, this is a retrospective study and potential selection bias is inevitable. However, the nomogram was derived from a large sample size cohort of tertiary hospital and it was validated in an external cohort. Nonetheless, data from multicenter prospective studies with larger sample size are still needed to make the results generalizable. Second, the occurrence of bleeding events may be influenced by precipitating factors, such as severe coughing, and constipation. However, this challenge reflects those faced in daily clinical practice and also exists in all large-scale research studies. Although these precipitating factors may shorten the period of the VH event, this shortened period was considered meaningful for early warning of bleeding. Finally, the follow-up duration was one-year, and the model’s predictive performance for the long-term prognosis remains unclear. Conversely, one-year is a reasonable time span for the development of VH risk prediction model. Despite these limitations, our research provides new guidance for the selection of prevention strategies for HBV-related VH and offers an idea to develop a predictive model for VH in patients with GEVs from other etiologies.

In summary, the nomogram established in our study can be useful for estimating the first VH probability within one-year and stratifying the bleeding risks in HBV-related cirrhotic patients with GEVs, which can assist clinicians in determining the appropriate prophylactic strategies. However, the clinical utility and true predictive value of this nomogram needs to be further verified in larger prospective studies.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethical Review Committee of the Beijing Ditan Hospital (Beijing, China). Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements. Written informed consent was not obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

XW, YM, and QZ conceived and designed the project. QZ, SN, KS, XZ, YZ, and YB collected the data. QZ, BZ, and LY analyzed and interpreted the data. QZ, SN, and LY drafted the manuscript. All authors contributed to the article and approved the submitted version.