Albumin-bilirubin grade as an alternative to Child–Pugh class for evaluating liver function within staging systems for hepatocellular carcinoma

Hepatocellular carcinoma (HCC) is a prevalent malignant neoplasm around the world, characterized by a rising global mortality rate [1]. Generally, HCC occurs in the setting of a chronically damaged liver [2]. HCC differs from other malignancies in that the treatment and outcome are influenced not only by the tumor burden but also by the condition of the hepatic function. Therapeutic regimens for HCC vary according to tumor stages, with different prognosis. Up till now, numerous staging systems for HCC have been proposed and implemented in clinical settings in different countries over the past decades, such as the Barcelona Clinic Liver Cancer (BCLC) [3], the China Liver Cancer (CNLC) [4], the Japan Integrated Staging (JIS) [5], the Cancer of the Liver Italian Program (CLIP) [6], the Hong Kong Liver Cancer (HKLC) staging systems [7]. Traditionally, the Child–Pugh (CP) class is integrated into these staging systems to serves as a semi-quantitative method to assess hepatic function in patients with HCC, including the international normalized ratio, level of serum albumin, level of bilirubin, severity of ascites, and degree of hepatic encephalopathy [8]. CP serves as the foundation for stratifying patients across treatment modalities, and its incorporation into national and international guidelines underscores its widespread acceptance and utility in clinical practice. Despite its robustness and widespread use, the CP class has several inherent limitations, including the equal weights among all 5 indexes as well as the subjective nature of ascites and hepatic encephalopathy evaluation. As such, some specialists deem that hepatic function should be assessed beyond the conventional CP class [3].

The model for end-stage liver disease (MELD) score was initially developed in 2000 for assessing risk of mortality in patients awaiting liver transplant following elective transjugular intrahepatic portosystemic shunts surgery [9]. Subsequently, the modified MELD was proposed based on levels of serum bilirubin, prothrombin time, and creatinine to predict mortality in a wide range of patients with advanced liver disease, namely the classic MELD [10]. In patients with decompensated cirrhosis, MELD has been proven to be a reliable predictor of outcomes. In 2015, the Albumin-Bilirubin (ALBI) score, based only on serum albumin and bilirubin, emerged as an alternative and objective tool for liver functional reserve in HCC [11]. The ALBI has been demonstrated to predict survival, tumor relapse, and post-hepatectomy liver failure of patients receiving potentially curative therapies [12].

Previous studies focused on the prediction performance of MELD or ALBI used alone for HCC prognosis. However, it remains largely unknown whether the MELD or ALBI grade can replace CP class to integrate into the existing staging systems for prognostication and management of HCC. As such, we conducted a comparative analysis of the prognostic performance of the original and modified staging systems in patients newly diagnosed with HCC when CP class was substituted by ALBI grade or MELD grade, to explores whether the ALBI or the MELD grade, a simpler and more objective alternative, could complement or enhance current staging systems to refine prognostic accuracy in selected scenarios.

Better staging systems are crucial to therapeutic decisions of oncologists and to the effective management of patients. In this study, we enrolled a cohort of 568 patients with newly diagnosed HCC, examined the possible prognostic factors of survival, and assessed the prediction performance of the modified staging systems for HCC prognosis when the CP class was replaced by either ALBI grade or MELD grade. Our findings demonstrated the significance of malignancy of the tumor, severity of liver dysfunction, and health status of patients as key factors of prognosis in HCC. Among all the original and modified staging systems, the ALBI-based staging systems performed comparable or even superior prediction ability when the CP class was replaced by ALBI grade, but the incorporation of the MELD grade in staging systems did not yield similar results. Notably, the ALBI-based HKLC system was the optimal prognostic model with superior long-term outcome prediction in different cohorts. Our study confirms acceptable prognostic performance of the ALBI grade across specific HCC staging systems. However, these findings should be interpreted cautiously, given the study's retrospective design and the lack of external validation.

We identified six prognostic factors of survival in HCC. An independent risk factor for OS is poor performance status, as measured by the Eastern Cooperative Oncology Group performance status scale. (ECOG-PS), was an independent risk factor of OS. As to the importance of the performance status, there was some disagreement. Specifically, patients with an ECOG-PS score of 1–2 are classified into the advanced stage based on the definition of BCLC system and are unsuitable for receiving hepatic resection. However, CNLC and HKLC systems deem that some patients with ECOG-PS score of 1 can obtain favorable outcomes by radical treatments. Furthermore, the subjective nature of ECOG-PS assessment makes it difficult to precisely determine whether the patient’s general health status at diagnosis of HCC is due to the tumors or to the other underlying diseases [21]. Expectedly, the tumor burden, including the number and size of tumor as well as the presence of vascular invasion and extrahepatic metastasis, could independently predict adverse prognosis in HCC patients. The ALBI grade, CP class, and MELD grade are usually used for evaluating the degree of liver dysfunction in HCC, and the association of ALBI grade with OS was observed. However, some variables encompassed in the CP class (e.g., ascites and prothrombin time) and in the MELD grade (e.g., creatinine and INR) were not considered as independent predictors in the study, meaning the CP class and MELD grade cannot accurately reflect liver function in part of HCC patients. This further supports the optimization of the model for assessing liver functional status.

Decreasing trends in survival from early to advanced stages were observed in different HCC staging systems. As for the comparison of the 11 original staging systems, the HKLC system presented the optimal prognostic performance in the entire cohort, with the highest discriminative ability. Indeed, there was a wide variation in staging systems and treatment algorithms across different geographical regions [22]. Specifically, the development of BCLC, CLIP, and JIS models were based on the Western or Japanese population where HCV is the predominant cause of HCC. However, the CNLC and HKLC systems were constructed in the Chinese society where HBV is the leading etiology of HCC. As such, the HKLC had the best C-index, Wald χ², and AICc values in the study.

While the CP class remains deeply embedded in HCC management and clinical guidelines, its limitations—such as subjectivity and reliance on parameters with variability in measurement—highlight the need for complementary approaches, so we substituted it by MELD grade or ALBI grade in the BCLC, CNLC, JIS, CLIP, and HKLC systems. Our study demonstrated that the replacement of CP class by ALBI grade achieved better or comparable overall prognostic performance in the five systems, but the incorporation of the MELD grade in staging systems failed to do so. Taken together, the ALBI-based HKLC system performed best with superior survival prediction in different cohorts. The ALBI grade involved only two blood test indexes, i.e., serum albumin and bilirubin. Comparably, the ALBI grade was more objective and accessible than the CP class, and had fewer parameters than the other two, thus probably decreasing the misjudgments of liver function. Moreover, the CP class is initially designed for cirrhotic patients, but now the change in the etiologies of HCC results in more patients with no cirrhosis at the diagnosis of HCC [23, 24]. The MELD grade is primarily developed for the risk prediction of mortality in patients awaiting liver transplant following elective transjugular intrahepatic portosystemic shunts surgery [9]. In contrast, the ALBI grade is designed to meet the need for simply and objectively assessing liver function in HCC, and its performance has been validated [11, 25]. Our study ultimately demonstrated that the ALBI grade is a promising alternative to the CP class for evaluating liver function within tumor staging systems. However, our findings should not be interpreted as advocating for ALBI to entirely replace CP. Rather, ALBI may serve as an adjunct or alternative in specific contexts, particularly when objective and reproducible metrics are preferred. It is also important to consider the real-world clinical implications of this shift. In clinical practice, this transition could refine treatment decisions. For instance, in cases where ALBI indicates better liver function compared to CP, patients who might have been excluded from aggressive treatments, such as locoregional therapies or surgical interventions, could now be considered suitable candidates. This would potentially expand therapeutic options and improve patient outcomes. Conversely, ALBI could help identify patients at higher risk of poor outcomes more accurately, supporting decisions for more conservative management or earlier referral to palliative care. By integrating ALBI into clinical decision-making, treatment strategies could be more individualized, allowing for a tailored approach to patient management that maximizes therapeutic benefits while minimizing risks. Future studies should explore this further to confirm the potential benefits of adopting ALBI in daily clinical practice.

The ALBI grade, based on albumin and bilirubin levels, reflects two key aspects of liver function: synthetic capacity and excretory function. Albumin, a protein synthesized by the liver, serves as an indicator of liver synthetic function and correlates with both the severity of liver damage and systemic inflammation [26, 27]. Hypoalbuminemia is often associated with poorer prognosis, as it reflects not only impaired liver function but also malnutrition and chronic inflammation, which can promote tumor progression and compromise the patient’s overall immune response. Bilirubin, on the other hand, is a marker of the liver’s excretory capacity [28]. Elevated bilirubin levels indicate significant liver dysfunction, which may be exacerbated by tumor burden or cirrhosis. In contrast, the CP classification includes subjective measures such as ascites and encephalopathy, which are influenced by clinical interpretation and do not directly reflect the intrinsic liver function. The objective nature of ALBI, by focusing on quantitative markers like albumin and bilirubin, may therefore offer a more reliable assessment of liver function and its interaction with tumor biology, leading to its superior prognostic performance.

The important distinction was observed between the MELD grade and the ALBI grade in predicting outcomes for HCC patients. MELD demonstrated lower prognostic performance compared to ALBI and CP. The MELD was originally developed to assess mortality risk in patients with end-stage liver disease, particularly those being considered for liver transplantation. But in China, a low proportion of HCC patients received liver transplants. Furthermore, the MELD primarily focuses on factors related to liver failure, such as bilirubin, INR, and creatinine. While these parameters are critical in end-stage liver disease, they may not adequately capture the complexities of liver function or tumor biology specific to HCC. As shown in Table 2, our univariate and multivariate survival analyses of the entire HCC cohort demonstrated that elevated levels of albumin and total bilirubin—both components of the ALBI grade—were associated with poorer prognostic outcomes. Conversely, the creatinine and INR, which are integral to the MELD grade, did not show similar predictive value in our study. MELD places a strong emphasis on renal function (creatinine) and coagulopathy (INR), which, while critical in cirrhosis, are less directly relevant to the oncologic burden of HCC. For example, HCC patients with well-compensated liver function but significant tumor burden may have low MELD scores, leading to an underestimation of their prognosis. In contrast, ALBI’s reliance on albumin and bilirubin provides a direct assessment of liver synthetic function and the severity of liver dysfunction without being confounded by renal parameters or regional treatment disparities. This discrepancy may explain the MELD score's comparatively lower performance, as it may not be as tailored to the unique characteristics of Chinese HCC patients.

The findings from our subgroup analysis indicate that MELD and ALBI scores are valuable prognostic tools in patients with non-resectable HCC, as these patients often exhibit compromised liver function. However, in patients with stable liver function, the utility of these scores diminishes. This highlights the importance of considering liver function status when applying MELD and ALBI scores in clinical practice. Although the ALBI grade enhances objectivity by focusing on albumin and bilirubin levels, it does not account for the subjective parameters present in the CP class. The CP class remains a cornerstone of liver function assessment due to its incorporation of variables such as ascites and hepatic encephalopathy, which are critical for both prognosis and therapeutic decision-making. In clinical scenarios where ascites and INR are crucial for treatment planning, ALBI could serve as a complementary measure rather than a replacement. This approach would allow clinicians to benefit from ALBI’s objectivity and reproducibility while retaining the holistic evaluation provided by the CP class. In future research, we will implement various strategies to standardize the assessment of ascites and hepatic encephalopathy. This will facilitate a more comprehensive comparison of the performance of different HCC staging systems and ensure a thorough evaluation of liver function in HCC patients. By addressing these considerations, we hope to strengthen the clinical utility of liver function assessments in managing HCC.

In this study, we evaluated the performance of the CP, MELD, and ALBI-based staging systems in both cirrhotic and non-cirrhotic patients, as 36.4% of our cohort were non-cirrhotic. In non-cirrhotic patients, our results showed that CP, MELD, and ALBI-based staging systems performed comparably, with the HKLC staging system demonstrating superior predictive ability. Future studies should focus on further validating these models in specific subgroups, such as patients without ascites, to refine their use in clinical practice.

Moreover, TNM-AJCC 8th performed well in several aspects, notably having the low AICc score (TNM-AJCC 8th 2486.196 vs. ALBI-HKLC 2469.079). First, the TNM-AJCC staging system is extensively validated and used worldwide for various types of cancers, including HCC. Given its widespread clinical application, there is ample data supporting its predictive accuracy and reliability. Although TNM-AJCC may not emphasize liver function, its straightforward assessment of tumor progression provides strong prognostic power in many clinical scenarios. Second, the TNM-AJCC system primarily focuses on the anatomical aspects of the tumor, including tumor size (T), lymph node involvement (N), and distant metastasis (M). These anatomical factors are highly relevant for predicting outcomes; as such, the TNM-AJCC system is widely applied in the field of surgical resection. As nearly half of the study population received curative treatments, tumor anatomy plays a crucial role in their prognosis, which explains the strong performance of the TNM-AJCC system in the current study. Third, the factors evaluated by the TNM-AJCC system—tumor size, lymph node involvement, and metastasis—are directly linked to the biological behavior of the tumor. Advanced HCC is often associated with larger tumors and metastasis, both of which significantly affect survival. Thus, TNM-AJCC can effectively capture the aggressiveness and progression of the tumor. In addition, we observed that the CP class outperformed MELD and ALBI in terms of homogeneity in both the JIS and CLIP staging systems, as reflected by the higher Wald χ² values. This may suggest that in specific staging systems like JIS and CLIP, which integrate factors such as tumor morphology and liver function differently, CP remains a reliable indicator of liver function and can provide better stratification of patients within these models. However, as shown by other metrics (C-index and AICc), ALBI-based systems demonstrated overall superior prognostic performance across most of the staging systems.

There are several inherent limitations within the study. First, the retrospective nature may result in case selection bias, so some findings might not be generalizable to other geographical regions. As such, a multi-center study with a large sample is needed to verify the conclusions. Second, our study is the predominance of HBV-related HCC cases in our cohort, which may restrict the generalizability of our findings to populations with other etiologies, such as HCV or non-alcoholic fatty liver disease (NAFLD)-related HCC. This is particularly relevant given the growing incidence of NAFLD-associated HCC in many regions. However, in areas with a high burden of HCC, such as China, Mongolia, and parts of Eastern Africa, chronic HBV infection continues to be the primary risk factor. Despite this limitation, we performed subgroup analyses in patients with cirrhotic HCC, those receiving curative treatments, and those undergoing non-curative treatments to compare the prognostic performance of different staging systems. Future studies with multi-ethnic cohorts and more diverse etiologies will be essential to confirm whether the ALBI-based staging systems is superior to other HCC staging systems across different HCC populations. Incorporating patients with HCV or NAFLD-related HCC will provide a more comprehensive understanding of its utility and help ensure broader applicability of the results. Thirdly, some experts insisted that a new staging system for HCC should be constructed based on tumor-related factors and ALBI grade, instead of simply replacing the CP class with ALBI grade in the existing staging systems. We had no objections to the method of re-constructing a new model, but considered the direct replacement method more acceptable to the clinicians and more readily generalizable to different geographical areas. Fourthly, to evaluate the robustness of the models, decision curve analysis and calibration plots were utilized. However, due to the single-center retrospective design and limited sample size, we recognize that the statistical models may be at risk of overfitting. Future analyses should incorporate cross-validation methods and external validation, to assess model stability and predictive accuracy. Fifthly, unlike the CP class, ALBI and MELD does not incorporate clinical variables such as ascites or hepatic encephalopathy, which are essential for comprehensive liver function assessment and often influence treatment strategies like TACE or liver transplantation. In patients with stable liver function, where bilirubin, INR, and creatinine, and albumin levels remain within normal ranges, the utility of ALBI or MELD may be limited. In these scenarios, tumor-specific factors, including tumor size, vascular invasion, and metastatic burden, become more pertinent for prognostication and treatment decision-making. Therefore, while the ALBI and MELD are valuable tools in assessing liver function, a more nuanced approach that incorporates tumor characteristics is essential for accurate prognostication; moreover, ALBI should be viewed as a complementary tool rather than a universal replacement for CP, particularly in clinical contexts where these parameters are critical. Future studies could explore hybrid models integrating ALBI with additional clinical variables to address these gaps. While our findings demonstrate the potential prognostic advantages of ALBI, its integration into clinical practice requires careful consideration. Transitioning from the CP class to ALBI should be approached through a stepwise framework: (1) Conduct pilot studies within selected institutions to evaluate the feasibility of incorporating ALBI into routine workflows; (2) Use international multi-center, prospective validation cohorts to confirm ALBI's prognostic performance in diverse patient populations with different HCC etiologies (e.g., NAFLD and HCV-related cases). Moreover, future research should focus on real-world trials to evaluate the impact of ALBI-based staging systems on clinical decision-making, patient outcomes, and workflow efficiency. Practical guidelines could then be developed based on these findings to support the gradual adoption of ALBI in routine practice.

To conclude, our study demonstrates that the ALBI grade provides comparable or even superior overall prognostic discrimination to the CP class when integrated into the widely used HCC staging systems. ALBI may serve as a complementary or alternative measure that may enhance prognostic accuracy in certain contexts. Especially, the ALBI-based HKLC system showed the best performance in survival prediction. Further studies are needed to improve the staging strategies for HCC. However, further validation in larger, multi-center cohorts representing diverse patient populations is essential before ALBI can be definitively recommended for routine clinical use.