Introduction
Accurate prognostication is central to multidisciplinary tumour board (MDT) decision-making in hepatocellular carcinoma (HCC). Although the albumin–bilirubin (ALBI) grade is an established prognostic marker, its real-world performance across tumour burden, performance status (PS), and MDT-directed treatment pathways remains incompletely defined. The incremental value of tumour biology, reflected by alpha-fetoprotein (AFP), relative to conventional scores is also unclear.
Aims & Methods
We evaluated whether ALBI differentiates survival across subgroups defined by tumour burden (up-to-7; UT7), performance status (PS), and MDT-recommended treatment pathway, and if AFP adds incremental value to ALBI. We retrospectively analysed consecutive patients with new or recurrent HCC discussed at the Singapore General Hospital / National Cancer Centre Singapore weekly liver MDT (2021–2024). Of 306 cases, 289 with complete data and ≥6 months follow-up were included. Tumour burden (up-to-7; UT7), performance status (PS 0–2), and MDT-recommended treatment pathways (surgery, locoregional, systemic) were recorded. Overall survival (OS) was analysed using Kaplan–Meier and Cox regression models. Model performance was compared using likelihood ratio testing.
Results
Median age was 70 years (IQR 63–77), and 75.2% were male. Hepatitis B (38.6%) and MASLD (26.1%) were the predominant aetiologies. In the overall cohort, ALBI demonstrated clear stepwise survival separation (p<0.001), with median OS not reached for ALBI 1 (37.4%), 21.4 months (95% CI 17.8–25.9) for ALBI 2 (49.1%), and 8.6 months (95% CI 5.9–11.4) for ALBI 3 (13.5%).
ALBI remained prognostic across tumour burden strata. Within UT7, median OS was not reached, 23.1 months (95% CI 18.4–28.9), and 10.2 months (95% CI 6.7–14.6) for ALBI 1–3 (p<0.001); beyond UT7, 32.6 months (95% CI 24.1–NR), 14.7 months (95% CI 10.2–19.9), and 6.1 months (95% CI 3.8–9.4), respectively (p<0.001). Consistent stratification was observed across PS subgroups (0–2; n=276). ALBI also differentiated outcomes across MDT-recommended treatment pathways, with the greatest separation observed among locoregional therapy candidates (p<0.001).
On multivariable analysis, ALBI was a strong independent predictor of mortality (HR ≈2.8 per grade increase, p<0.001). AFP provided additional prognostic value beyond ALBI (HR ≈1.4 per log₁₀ increase, p<0.001), significantly improving model fit (p<0.001). In contrast, Child–Pugh class and composite tumour–liver scores, including up7-ALBI, did not improve prognostic performance beyond ALBI alone.
Among patients undergoing locoregional therapy, both ALBI and AFP remained independent predictors of survival. A simple 3×2 ALBI–AFP risk matrix further stratified outcomes (p<0.001), identifying a low-risk group (ALBI 1, AFP <20 ng/mL) with favourable survival and a very high-risk group (ALBI 3, AFP ≥20 ng/mL) with poor outcomes.
Conclusion
In a real-world MDT cohort, ALBI robustly stratifies survival across tumour burden, performance status, and treatment pathways, and is the dominant determinant of prognosis. AFP provides complementary information reflecting tumour biology and enhances risk discrimination. A simple ALBI–AFP framework enables clinically meaningful stratification, particularly among locoregional therapy candidates, supporting its integration into MDT-based decision-making.
Disclosure
The authors have no conflict of interest to declare.