Incorporating ALBI Grade with Geriatric Nutritional Risk Index Enhances Hepatocellular Carcinoma Risk Stratification

Introduction

Hepatocellular carcinoma (HCC) poses a significant global health burden. According to the 2022 estimates, liver cancer is the sixth most frequently diagnosed cancer and the third leading cause of cancer-related mortality [1]. The incidence of HCC is projected to rise [2], driven by a notable shift in underlying causes. Metabolic dysfunction-associated steatotic liver disease has emerged as a major contributor to HCC, with a particularly high prevalence in older individuals [3, 4]. As a result, the incidence of HCC in the older population is increasing and is expected to continue rising. In Korea, for instance, projections indicate that by 2028, individuals aged 80 years and older will comprise an astonishing 21.3% of all patients with HCC [5].

Older patients with cancer, especially those with multiple comorbidities, are significantly more vulnerable to malnutrition. This compromised nutritional status has been linked to poorer treatment outcomes in HCC, including impaired quality of life, reduced effectiveness of anticancer therapy, diminished treatment tolerance, and an increased risk of adverse events, ultimately leading to decreased overall survival (OS) [6, 7]. Estimates suggest that up to 10–20% of cancer-related deaths are attributable to malnutrition rather than the tumor itself [6]. Despite the established influence of nutritional status on HCC prognosis, nutritional assessments are often neglected in routine HCC management. The geriatric nutritional risk index (GNRI) was developed to objectively assess the nutritional status of patients and has been validated for prognostic relevance in numerous HCC studies [8‒11]. Its significance extends beyond geriatric patients to include all adult patients with HCC, as demonstrated by a recent meta-analysis [12].

The albumin-bilirubin (ALBI) grade is a well-established tool for assessing liver function in patients with HCC, enabling effective prognosis stratification [13, 14]. However, a recognized limitation of the ALBI grading system is the imbalanced distribution of patients across grades. Studies have shown a disproportionately low number of patients classified as having ALBI grade 3, while a larger and more heterogeneous population falls within ALBI grades 1 and 2 [15‒17]. This imbalance can hinder the ability of the ALBI grade to precisely differentiate the prognosis. To refine risk stratification, additional parameters that capture prognostic differences are required. Given the established association between nutritional status and HCC prognosis, this study investigated the potential role of the GNRI in conjunction with ALBI grade using data from the Korea Central Cancer Registry (KCCR).

Methods

Patients

The KCCR serves as a comprehensive data resource on the incidence and survival rates of major cancers, including HCC. The KCCR employs a systematic approach to collect detailed HCC data through stratified random sampling of 10–15% of all newly diagnosed HCC cases in Korea, ensuring representation across regions and hospitals [18]. The registry process is multistage, encompassing random case selection, meticulous data extraction by trained personnel, expert validation, personal data anonymization, and data dissemination for research purposes. This random sampling strategy ensures unbiased selection, safeguards patient privacy, and complies with regulations [18].

In total, 18,315 patients newly diagnosed with HCC in the KCCR database between 2008 and 2019 were enrolled in this study. Patients younger than 19 years or those with missing or incomplete clinical data were excluded (online suppl. Fig. 1; for all online suppl. material, see https://doi.org/10.1159/000541647). After these exclusions, 16,416 patients with HCC were included in the final analysis.

Data Collection and Definitions

This dataset incorporated demographic information (age and sex), anthropometric data (height and weight), clinical characteristics (date of diagnosis, viral hepatitis markers, alcohol consumption, ascites, encephalopathy, and performance status), laboratory values (including albumin and bilirubin levels), tumor characteristics (maximal size and number), and treatment methods. The HCC diagnosis was established based on either histological confirmation or dynamic imaging findings on computed tomography and/or magnetic resonance imaging [19]. The Barcelona Clinic Liver Cancer (BCLC) staging system was used for disease staging. OS was calculated as the time elapsed from the date of HCC diagnosis to death from any cause. Mortality data were obtained from death certificates within the national statistical database maintained by the Korean Ministry of Government Administration and Ministry of Home Affairs, with information up to December 2022. The etiology of liver disease was determined by hepatitis B surface antigen positivity for chronic hepatitis B and anti-hepatitis C virus antibody positivity for chronic hepatitis C. The body mass index (BMI) was calculated as body weight (kg) divided by height squared (m²). Patients were categorized as underweight (BMI <18.5 kg/m²), normal (BMI 18.5–24.9 kg/m²), and overweight (BMI ≥25 kg/m²) according to World Health Organization criteria [20]. We categorized patients based on their first-line treatment modality: surgical resection, liver transplantation, local ablation therapy, transarterial therapy (transarterial chemoembolization or radioembolization), systemic therapy (sorafenib, lenvatinib, or other cytotoxic chemotherapy agents), or best supportive care only.

Assessments of Hepatic Reserve Function and Nutritional Status

The ALBI score was determined using serum levels of albumin and total bilirubin, calculated as follows: (log10 bilirubin [µmol/L] × 0.66) + (albumin [g/L] × −0.085). The ALBI grades correspond to the following ALBI scores: grade 1 (≤−2.60), grade 2 (>−2.60 to ≤−1.39), and grade 3 (>−1.39) [13].

Nutritional status was assessed using the GNRI, calculated with the following formula: GNRI = 14.89 × serum albumin (g/dL) + 41.7 × present body weight (kg)/ideal body weight (kg). Ideal body weight was derived based on sex and height using the following formulas: for men, height (cm) − 100 − {(height [cm] − 150)/4}; and for women, height (cm) − 100 − {(height [cm] − 150)/2.5}. To achieve a more refined risk stratification, an optimal cutoff value for the GNRI was determined for each ALBI grade. This approach integrates the strengths of both the ALBI grade and GNRI, enhancing the overall assessment of patient risk.

Statistical Analysis

Data are presented as numbers with percentages or medians with interquartile ranges, depending on the nature of the variables. Statistical tests were chosen based on data type and distribution. For continuous variables, a Student’s t test was used to compare the two groups. For categorical variables, the χ² test was used when the expected frequencies were sufficient, and Fisher’s exact test was applied when the expected frequencies were low. Analysis of variance was used to compare differences in means between more than two groups. Restricted cubic spline (RCS) regression was used to identify GNRI cutoffs specific to each ALBI grade. The Pearson correlation coefficient was calculated to assess the linear correlation between the GNRI and other relevant endpoint markers. An absolute correlation coefficient (|R|) ≥ 0.5 and a p value <0.05 were considered statistically significant. Patient survival was analyzed using the Kaplan-Meier method to estimate survival curves for different groups, with the log-rank test assessing the statistical significance of the differences in survival rates between the groups. Additionally, Cox proportional hazards regression analysis was performed to identify independent factors associated with OS. Statistical analyses were performed using the Stata/IC 17 (StataCorp LLC, USA) and R software version 4.0.3 (plotRCS package, R Foundation, Austria). Two-sided p values <0.05 were considered statistically significant.

Results

Patient Characteristics

The baseline characteristics of 16,416 patients with HCC are presented in Table 1. The mean age was 61 ± 11.4 years, and 79.2% were male. The mean BMI was 24.0 ± 3.4 kg/m². Among the patients, 650 (4.0%) were underweight, 9,948 (60.6%) were of normal weight, and 5,818 (35.4%) were overweight. The etiologies of HCC were distributed as follows: chronic hepatitis B (59.5%), chronic hepatitis C (10.2%), alcoholic liver disease (14.5%), and others (15.9%). Regarding liver function, 7,409 patients (45.1%) had ALBI grade 1, 7,445 (45.4%) had ALBI grade 2, and 1,562 (9.5%) had ALBI grade 3. The mean GNRI value was 100.7 ± 12.0. The median tumor size was 3.1 cm, and 10,202 patients (62.3%) presented with solitary tumors at diagnosis. For BCLC staging, 5,132 (31.3%) patients were classified as stage 0/A, 2,608 (15.9%) as stage B, 5,289 (32.2%) as stage C, 968 (5.9%) as stage D, and 2,419 (14.7%) as indeterminate. Patients were categorized according to ALBI grade, revealing significant differences in clinical factors across groups, including age, sex, BMI, GNRI, laboratory data, tumor characteristics, and BCLC stage.

Table 1.

Baseline characteristics of enrolled patients

Variables	Total (n = 16,416)	ALBI grade 1 (n = 7,409)	ALBI grade 2 (n = 7,445)	ALBI grade 3 (n = 1,562)	p value
Age, years	61.3±11.4	60.4±11.1	62.4±11.5	59.9±11.5	<0.001
Male sex	13,007 (79.2)	5,952 (80.3)	5,800 (77.9)	1,255 (80.3)	<0.001
BMI (kg/m2)	24.0±3.4	24.1±3.2	23.8±3.5	24.0±3.7	<0.001
<18.5	650 (4.0)	299 (4.0)	303 (4.1)	48 (3.1)
18.5–24.9	9,948 (60.6)	4,303 (58.1)	4,642 (62.4)	1,003 (64.2)
≥25	5,818 (35.4)	2,807 (37.9)	2,500 (33.6)	511 (32.7)
Etiology					<0.001
HBVa	9,767 (59.5)	4,775 (64.4)	4,091 (54.9)	901 (57.7)
HCV	1,667 (10.2)	574 (7.7)	936 (12.6)	157 (10.1)
Alcohol	2,374 (14.5)	898 (12.1)	1,172 (15.7)	304 (19.5)
Others	2,608 (15.9)	1,162 (15.7)	1,246 (16.7)	200 (12.8)
ALBI grade
Grade 1	7,409 (45.1)
Grade 2	7,445 (45.4)
Grade 3	1,562 (9.5)
Child-Pugh class					<0.001
A	12,015 (73.2)	7,183 (96.9)	4,829 (64.9)	3 (0.2)
B	3,674 (22.4)	226 (3.1)	2,509 (33.7)	939 (60.1)
C	727 (4.4)	0 (0)	107 (1.4)	620 (39.7)
GNRI	100.7±12.0	109.3±7.6	96.1±8.7	82.4±9.6	<0.001
AST, IU/L	45 (30, 81)	34 (26, 50)	58 (38, 107)	98 (58, 189)	<0.001
ALT, IU/L	35 (22, 56)	31 (21, 47)	37 (23, 62)	47 (28, 88)	<0.001
TB, mg/dL	0.9 (0.6, 1.4)	0.7 (0.5, 0.9)	1.1 (0.8, 1.6)	3.0 (1.8, 6.5)	<0.001
Albumin, g/dL	3.9 (3.3, 4.3)	4.3 (4.1, 4.5)	3.5 (3.2, 3.8)	2.6 (2.3, 2.8)	<0.001
PT, INR	1.1 (1.0, 1.2)	1.1 (1.0, 1.1)	1.1 (1.1, 1.3)	1.4 (1.2, 1.6)	<0.001
Platelet count, ×109/L	146 (100, 201)	162 (124, 209)	131 (87, 194)	112 (74, 182)	<0.001
Tumor size, cm	3.1 (2.0, 6.0)	3.0 (1.9, 5.0)	3.5 (2.0, 7.0)	3.7 (2.0, 7.5)	<0.001
Tumor number (single)	10,202 (62.3)	5,307 (71.8)	4,198 (56.5)	697 (45.0)	<0.001
BCLC stage					<0.001
0/A	5,132 (31.3)	3,174 (42.8)	1,869 (25.1)	89 (5.7)
B	2,608 (15.9)	1,323 (17.9)	1,159 (15.6)	126 (8.1)
C	5,289 (32.2)	1,746 (23.6)	3,078 (41.3)	465 (29.8)
D	968 (5.9)	29 (0.4)	267 (3.6)	672 (43.0)
Unknown	2,419 (14.7)	1,137 (15.3)	1,072 (14.4)	210 (13.4)
AFP, ng/mL	27.7 (5.5, 517.5)	15.8 (4.3, 254.1)	41.5 (7.4, 943.6)	74.1 (7.0, 2,000.0)	<0.001
PIVKA-II, mAU/mL	114.0 (28.0, 1,874.0)	69.0 (25.1, 670.7)	192.0 (29.0, 2,473.5)	486.0 (64.0, 6,698.6)	<0.001

Values are presented as mean ± standard deviation, median (interquartile range), or number (%).

BMI, body mass index; HBV, hepatitis B virus; HCV, hepatitis C virus; AST, aspartate aminotransferase; ALT, alanine aminotransferase; TB, total bilirubin; PT, prothrombin time; INR, international normalized ratio; GNRI, geriatric nutritional risk index; ALBI, albumin-bilirubin; BCLC, Barcelona Clinic Liver Cancer; AFP, alpha-fetoprotein; PIVKA-II, protein induced by vitamin K absence or antagonist-II.

^aPatients coinfected with HBV and HCV (n = 195) were also included.

Correlational and Distribution Analysis of GNRI

Correlation analysis showed significant positive correlations between GNRI and albumin levels (R = 0.848, p < 0.001) and BMI (R = 0.566, p < 0.001) (Table 2). There was no significant correlation between age and GNRI (R = −0.089) (online suppl. Fig. 2a). The GNRI distribution was analyzed across various subgroups, including age, sex, alpha-fetoprotein (AFP) level, and etiology. Although significant differences in GNRI distribution were observed among these subgroups, the magnitude of these differences was not substantial (Fig. 1a–d; online suppl. Table 1). Patterns of GNRI distribution were similar across age subgroups, with only a slight decline observed in the elderly subgroup (≥70 years), but the difference was modest (Fig. 1a). However, the GNRI distribution was significantly influenced by both the BCLC stage and ALBI grade (Fig. 1e, f). The GNRI scores were significantly lower in patients with more advanced tumor stages (assessed by BCLC) and poorer liver function (evaluated by ALBI grades). Given these substantial differences in GNRI distribution across ALBI grades, we further explored the prognostic impact of the GNRI within each ALBI grade.

Table 2.

Pearson correlation coefficient of GNRI and clinical factors

Variables	Pearson’s correlation coefficient	p value
Age	−0.089	<0.001
Albumin (g/dL)	0.848	<0.001
TB (mg/dL)	−0.242	<0.001
BMI (kg/m2)	0.566	<0.001
ALT (IU/L)	−0.073	<0.001
Platelet count, ×109/L	0.027	<0.001
Tumor size (largest, cm)	−0.173	<0.001

GNRI, geriatric nutritional risk index; TB, total bilirubin; BMI, body mass index; ALT, alanine aminotransferase.

Fig. 1.

Distribution of GNRI in subgroups of clinical indicators. a Age. b Sex. c AFP level. d Etiology. e BCLC stage. f ALBI grade. GNRI, geriatric nutritional risk index; AFP, alpha-fetoprotein; BCLC, Barcelona Clinic Liver Cancer; ALBI, albumin-bilirubin.

RCS Analysis

To determine the optimal cutoff values of the GNRI for predicting survival outcomes within each ALBI grade, we employed RCS analysis. The analysis revealed significant differences in the optimal cutoff values of the GNRI across ALBI grades (Fig. 2a–c). In ALBI grades 1 and 2, RCS curve analysis of GNRI scores showed a rapid decline in the hazard ratio (HR) until the optimal cutoff value, followed by a gradual stabilization of HR as GNRI increased. This inverse relationship between the GNRI and HR was highly significant for ALBI grades 1 and 2 (overall p < 0.001). In contrast, the association between the GNRI and survival in patients with ALBI grade 3 was unclear. RCS analysis indicated a U-shaped HR pattern, with a moderate decrease, followed by an unexpected increase beyond the cutoff GNRI value. Although the overall p value was significant (p = 0.026), the RCS curve exhibited inconsistencies and differed markedly from those of ALBI grades 1 and 2. This suggests that the GNRI may offer limited prognostic utility in patients with more severe liver dysfunction, as indicated by ALBI grade 3. Additionally, the smaller number of patients with grade 3 compared to grades 1 or 2 may not yield significant clinical insights. Consequently, future analyses should exclude subclassifying ALBI grade 3 based on GNRI levels.

Fig. 2.

RCS curves for each ALBI grade illustrating the association between GNRI and the mortality risk. a ALBI grade 1. b ALBI grade 2. c ALBI grade 3. GNRI, geriatric nutritional risk index; HR, hazard ratio; CI, confidence interval; ALBI, albumin-bilirubin.

Based on the observed patterns, we established distinct GNRI cutoff levels for each ALBI grade: 105 for ALBI grade 1 and 90 for ALBI grade 2. Using these cutoffs, we categorized patients into five groups by incorporating ALBI grade with GNRI: ALBI grade 1 with GNRI >105 (n = 5,239), ALBI grade 1 with GNRI ≤105 (n = 2,170), ALBI grade 2 with GNRI >90 (n = 5,639), ALBI grade 2 with GNRI ≤90 (n = 1,806), and ALBI grade 3 (n = 1,562).

Survival Analysis Incorporating ALBI Grade with GNRI

During the study period, we collected 52,084 person-years of follow-up, resulting in 10,615 recorded deaths. The median OS of all patients was 3.13 years (95% confidence interval [CI]: 3.03–3.22 years). Significant differences in median OS were observed across ALBI and BCLC stages (Fig. 3a and Table 3). Survival analysis was conducted to assess whether incorporating the GNRI with the ALBI grade significantly impacted the prognostic stratification. Among patients with ALBI grades 1 and 2, those with higher GNRI scores exhibited significantly better survival outcomes. The median OS for each group was as follows: ALBI 1 + GNRI >105; 7.42 years (95% CI: 6.99–7.84), ALBI 1 + GNRI ≤105; 4.48 years (95% CI: 4.05–4.85), ALBI 2 + GNRI >90; 2.21 years (95% CI: 2.08–2.36), ALBI 2 + GNRI ≤90; 0.85 years (95% CI: 0.76–1.03), and ALBI 3; 0.40 years (95% CI: 0.34–0.47) (Fig. 3b). Analysis utilizing distinct GNRI cutoffs for ALBI grades 1 and 2 proved to be an effective method for stratifying the prognosis in patients with HCC.

Fig. 3.

Kaplan-Meier curves for OS based on ALBI grade (a), the incorporation of ALBI grade with GNRI (b), the incorporation of ALBI grade with GNRI in BCLC stages 0/A, B, C, and D (c–f). GNRI, geriatric nutritional risk index; ALBI, albumin-bilirubin; BCLC, Barcelona Clinic Liver Cancer.

Table 3.

Median OS stratified by ALBI grades, BCLC stages, and GNRI

Variables	N	Median OS (95% CI)	p value
ALBI			<0.001
Grade 1	7,409	6.28 (6.01–6.58)
Grade 2	7,445	1.80 (0.68–1.91)
Grade 3	1,562	0.40 (0.34–0.47)
BCLC			<0.001
Stage 0/A	5,132	7.79 (7.46–8.12)
Stage B	2,608	2.63 (2.43–2.79)
Stage C	5,289	0.81 (0.75–0.85)
Stage D	968	0.20 (0.17–0.22)
ALBI+GNRI			<0.001
ALBI 1 + GNRI >105	5,239	7.42 (6.99–7.84)
ALBI 1 + GNRI ≤105	2,170	4.48 (4.05–4.85)
ALBI 2 + GNRI >90	5,639	2.21 (2.08–2.36)
ALBI 2 + GNRI ≤90	1,806	0.85 (0.76–1.03)
ALBI 3	1,562	0.40 (0.34–0.47)
BCLC 0/A			<0.001
ALBI 1 + GNRI >105	2,397	N/A (N/A)
ALBI 1 + GNRI ≤105	777	8.77 (8.13–9.62)
ALBI 2 + GNRI >90	1,589	5.41 (5.12–5.84)
ALBI 2 + GNRI ≤90	280	3.65 (2.93–4.33)
ALBI 3	89	3.17 (2.31–4.73)
BCLC B			<0.001
ALBI 1 + GNRI >105	878	4.26 (3.84–4.73)
ALBI 1 + GNRI ≤105	445	2.72 (2.32–3.17)
ALBI 2 + GNRI >90	881	2.04 (1.82–2.33)
ALBI 2 + GNRI ≤90	278	1.04 (0.74–1.32)
ALBI 3	126	0.74 (0.55–1.01)
BCLC C			<0.001
ALBI 1 + GNRI >105	1,131	2.39 (2.12–2.67)
ALBI 1 + GNRI ≤105	615	1.26 (1.08–1.52)
ALBI 2 + GNRI >90	2,224	0.60 (0.56–0.66)
ALBI 2 + GNRI ≤90	854	0.40 (0.35–0.45)
ALBI 3	465	0.34 (0.29–0.38)
BCLC D			0.082
ALBI 1 + GNRI >105	13	1.10 (0.24–2.76)
ALBI 1 + GNRI ≤105	16	0.63 (0.33–0.85)
ALBI 2 + GNRI >90	158	0.18 (0.16–0.23)
ALBI 2 + GNRI ≤90	109	0.26 (0.17–0.37)
ALBI 3	672	0.18 (0.16–0.21)

ALBI, albumin-bilirubin; BCLC, Barcelona Clinic Liver Cancer; GNRI, geriatric nutritional risk index; OS, overall survival; CI, confidence interval.

We conducted additional analyses stratified by BCLC stage to further assess the impact of combining ALBI grade with GNRI on survival outcomes. Kaplan-Meier curves and log-rank tests were used to evaluate survival outcomes for each BCLC stage. The incorporation of ALBI grade and GNRI demonstrated a significant association with OS across BCLC stages 0/A, B, and C (Fig. 3c–e, all p < 0.001). However, this prognostic stratification was not significant for BCLC stage D (p = 0.082) (Fig. 3f). Notably, patients with ALBI grade 2 and GNRI ≤90 had a poor prognosis comparable to those with ALBI grade 3. There were no significant differences in median OS between patients with ALBI grade 2 and GNRI ≤90 and those with ALBI grade 3 across BCLC stages (p = 0.6454, 0.9663, 0.6027, and 0.8395 for BCLC stages 0/A, B, C, and D, respectively).

Cox Proportional Hazards Analysis of GNRI

To assess the independent prognostic value of GNRI, we performed Cox proportional hazards regression analyses (Table 4). The GNRI was dichotomized based on previously established cutoff values and included as a central prognostic factor for patients with ALBI grades 1 and 2. The univariate Cox proportional hazards regression analysis (model 0) revealed that a high GNRI was significantly associated with favorable OS (crude HR, 0.700; 95% CI: 0.670–0.732; p < 0.001). This association remained statistically significant after adjusting for age and sex in the multivariate analysis (model 1). The adjusted HR [aHR] for high GNRI in model 1 was 0.719 (95% CI: 0.688–0.752; p < 0.001). Furthermore, high GNRI retained its independent prognostic value for favorable OS even after further adjustment for additional clinical factors in model 2 (aHR, 0.775; 95% CI: 0.733–0.820; p < 0.001). These clinical factors included age, sex, BMI, platelet count, ALBI grade, BCLC stage, liver disease etiology, and AFP levels. To address potential multicollinearity, we calculated the variance inflation factor (VIF) between GNRI and the included variables. The mean VIF was 1.08, indicating minimal collinearity. A detailed breakdown of the VIF for each variable is provided in the online supplementary Table 2.

Table 4.

Univariate and multivariate analyses of association between GNRI and survival

Subjects	Model 0		Model 1		Model 2
	crude HR (95% CI)	crude P	aHR (95% CI)	adjusted P	aHR (95% CI)	adjusted P
ALBI grade 1 and 2 (n = 14,854)
Low GNRI (n = 3,976)	Reference		Reference		Reference
High GNRI (n = 10,878)	0.700 (0.670–0.732)	<0.001	0.719 (0.688–0.752)	<0.001	0.775 (0.733–0.820)	<0.001
Total patients (n = 16,416)
ALBI 3 (n = 1,562)	Reference		Reference		Reference
ALBI 2 + GNRI ≤90 (n = 1,806)	0.783 (0.728–0.842)	<0.001	0.715 (0.665–0.769)	<0.001	0.887 (0.812–0.968)	0.007
ALBI 2 + GNRI >90 (n = 5,639)	0.495 (0.466–0.526)	<0.001	0.471 (0.443–0.501)	<0.001	0.717 (0.665–0.773)	<0.001
ALBI 1 + GNRI ≤105 (n = 2,170)	0.306 (0.283–0.330)	<0.001	0.288 (0.267–0.311)	<0.001	0.420 (0.382–0.460)	<0.001
ALBI 1 + GNRI >105 (n = 5,239)	0.193 (0.181–0.207)	<0.001	0.186 (0.173–0.199)	<0.001	0.311 (0.286–0.338)	<0.001

Model 0, not adjusted; model 1, adjusted for age and sex; model 2, adjusted for age, sex, BMI, platelet count, ALBI grade, BCLC stages, etiology, and AFP.

GNRI, geriatric nutritional risk index; HR, hazard ratio; CI, confidence interval; ALBI, albumin-bilirubin; BMI, body mass index; BCLC, Barcelona Clinic Liver Cancer.

We repeated the Cox proportional hazards regression analysis incorporating the ALBI grade and GNRI for the entire patient cohort, including those with ALBI grade 3. In model 0, patients classified as ALBI grade 1 with a GNRI >105 had the lowest HR, whereas those with worse ALBI grades and low GNRI levels exhibited significantly higher HRs (p < 0.001). This trend in HR across prognostic categories was consistently observed in the adjusted models (models 1 and 2), with all intergroup differences reaching statistical significance (all p < 0.01).

Stratification Analysis

A comprehensive stratification analysis was conducted to evaluate the impact of GNRI on OS in patients with HCC. The analysis included 14,854 patients with ALBI grades 1 and 2, excluding 1,562 patients with ALBI grade 3. Patients were categorized based on several demographic and clinical indicators: age, sex, Child-Pugh class, ALBI grade, BCLC stage, AFP level, platelet count, BMI, liver disease etiology, and year of database entry. The analysis revealed a significant impact of the GNRI on survival in most of the stratified groups (Fig. 4a). Patients with a high GNRI consistently had a lower risk of mortality compared to those with a low GNRI. However, a survival benefit associated with a high GNRI score was not observed in patients with Child-Pugh class C, BCLC stage D, or underweight status (BMI <18.5 kg/m²). There were no statistically significant differences in survival outcomes between the subgroups based on GNRI levels. We further evaluated the association between the GNRI and OS by adjusting the HR for age and sex within each stratum (Fig. 4b). These results were consistent, showing that patients with a high GNRI exhibited a significantly lower risk of mortality in most strata. The insignificant survival differences among Child-Pugh class C, BCLC stage D, and underweight patients persisted even after adjustment.

Fig. 4.

Stratification analysis between GNRI and survival by clinical indicators (a, b) and treatment methods (c, d). Forest plots depict HRs (a, c) and aHRs for age and sex (b, d). GNRI, geriatric nutritional risk index; HR, hazard ratio; CI, confidence interval; aHR, adjusted hazard ratio; ALBI, albumin-bilirubin; LT, liver transplantation.

We conducted further stratified analyses based on the first-line treatment method. Patients who underwent surgical resection, liver transplantation, local ablation therapy, and transarterial therapy exhibited a significantly lower risk of mortality with high GNRI compared to those with low GNRI (Fig. 4c, all p < 0.01). This association was not statistically significant in patients treated with systemic chemotherapy, although there was a trend toward lower mortality (HR, 0.879; 95% CI: 0.767–1.009; p = 0.066). No significant association between GNRI and mortality was observed in the best supportive care group. After adjusting HR for age and sex, these findings remained consistent (Fig. 4d). The adjusted analysis revealed a significant association between high GNRI and lower mortality among patients receiving systemic chemotherapy (aHR, 0.868; 95% CI: 0.756–0.995; p = 0.043).

Discussion

This nationwide registry cohort study investigated the association between the GNRI and OS in patients with HCC. Our findings revealed a significant correlation, with patients exhibiting higher GNRI demonstrating improved survival outcomes compared to those with lower GNRI. Furthermore, we observed that combining the ALBI grade with the GNRI provided a more precise prognostic stratification of OS. This finding was confirmed using multivariate Cox proportional hazards regression and stratified analyses. These results suggest that good nutritional status, as indicated by a high GNRI, may be a significant prognostic factor for patients with HCC, particularly when combined with the ALBI grade.

Nutritional status plays a pivotal role in the prognosis and treatment of patients with HCC [7]. Despite its significance, cancer-related malnutrition in patients with HCC is often unrecognized, underestimated, or inadequately treated in clinical practice. Moreover, malnutrition and a dysbiotic gut microbiome are closely linked to the progression of metabolic dysfunction-associated steatotic liver disease [21], a major risk factor for HCC. Given this intricate relationship, nutritional assessment and management should be the cornerstone of HCC care. In this study, we used the GNRI to objectively assess the nutritional status of patients with HCC. While prior studies have investigated the association between the GNRI and OS in smaller cohorts [10, 11, 22‒24], our research distinguishes itself by thoroughly analyzing a comprehensive dataset of 16,416 patients and accumulating 52,084 person-years of follow-up. The GNRI incorporates albumin level, an indicator of both nutritional status and liver function. This dual role complicates the use of the GNRI alone for assessing the nutritional status of patients with HCC, as liver dysfunction can independently affect albumin levels. To address this issue, we stratified patients based on the ALBI grade and explored the association between the GNRI and prognosis within each ALBI grade. By incorporating the ALBI grade alongside the GNRI, we aimed to provide a more robust assessment of nutritional status and its influence on clinical outcomes in patients with HCC.

The Child-Pugh classification has limitations for HCC, including subjectivity in assessing ascites and encephalopathy, and potentially inapplicable cutoff points for albumin, bilirubin, and prothrombin time in non-cirrhotic patients [14]. In contrast, the ALBI grade provides an objective assessment of liver function and has been validated for use in patients with HCC through numerous studies [14, 25‒28]. However, the distribution of the ALBI grades may limit their prognostic power. Kudo et al. [17] reported that only a few patients were classified as having ALBI grade 3, with a large proportion falling into ALBI grade 2, accounting for half of all patients. Our analysis also found that ALBI grade 3 accounted for less than 10% of patients, while grades 1 and 2 each comprised approximately 45% (Table 1). This uneven distribution suggests the need for subclassification strategies, particularly for patients with ALBI grades 1 and 2.

The combination of ALBI grades and GNRI was significantly correlated with OS across the entire cohort (p < 0.001; Fig. 3b). The GNRI demonstrated significant associations with OS within ALBI grades 1 and 2, with specific cutoff values identified to differentiate prognostic outcomes. However, in patients with ALBI grade 3, the prognostic efficacy of the GNRI diminishes. The significance of incorporating the GNRI into ALBI grades 1 and 2 was independently demonstrated in the multivariate analysis (Table 4). Stratified analysis across various subgroups and treatment methods further demonstrates the robust prognostic value of GNRI (Fig. 4). By incorporating the ALBI grade alongside the GNRI, we achieved a more refined risk stratification that better captured the interplay between nutritional status and liver function, both of which influence patient outcomes.

Furthermore, the prognostic evaluation incorporating the ALBI grade with the GNRI remained significant across each BCLC stage. In stages 0/A, B, and C, the groups with a favorable ALBI grade and high GNRI consistently exhibited significantly better OS (all p < 0.001, Fig. 3c–e). However, in BCLC stage D, nutritional assessment using the GNRI showed no significant association with prognosis (p = 0.082, Fig. 3f). This finding aligns with real-world clinical practice, where BCLC stage D is associated with an extremely poor prognosis, with a median OS of only 0.2 years in our study. The limited lifespan at this stage makes it challenging to analyze the impact of nutritional status. Additionally, as anticancer treatment is not feasible in BCLC stage D, differences in treatment responses based on nutritional status do not exist. This advanced disease state may limit the utility of nutritional assessments in predicting the prognosis. GNRI could be effective for prognosis in HCC patients with preserved liver function (ALBI 1 and 2) and feasible treatment (BCLC 0-C), suggesting nutritional status impacts treatment efficacy and compliance.

The authors acknowledge several limitations of their study that need to be considered when interpreting the findings. First, this study was conducted solely in Korean patients, which may limit the applicability of the results to other ethnic groups. Further studies are required to validate these findings in diverse populations. Second, this study focused solely on GNRI as a nutritional assessment tool. Although the GNRI is a valuable measure, it may not capture the full spectrum of nutritional status. Other nutritional parameters such as dietary intake, nutrition-related symptoms, muscle mass, physical performance, and systemic inflammation were not assessed. This may limit the comprehensiveness of the nutritional evaluations. As this study relied on a national registry database, these aspects could not be evaluated. However, we believe that a detailed analysis conducted with a large patient cohort can validate the effectiveness of the GNRI. Further studies are required to evaluate these additional aspects. Finally, our study may not fully capture the impact of GNRI on patients receiving systemic chemotherapy. At the time of the study, sorafenib and lenvatinib were the only approved systemic therapies for HCC, predating the introduction of immunotherapy. This limitation hinders the generalizability of our findings in patients with advanced HCC, given the current treatment landscape where immunotherapies have become the mainstay. Further research is needed to explore the role of GNRI in patients undergoing immunotherapy.

Despite these limitations, the strengths of our study lie in the large nationwide registry cohort and the novel approach of combining the ALBI grade with GNRI for improved prognostication. Our findings underscore the pivotal role of nutritional status in the survival of patients with HCC. Particularly in patients with preserved liver function, integrating the ALBI grade with the GNRI enhances comprehensive risk stratification for HCC. Overall, this study emphasizes the importance of comprehensively evaluating both liver function and nutritional status when assessing the prognosis of patients with HCC.

Acknowledgments

This study was conducted with data provided by the Korea Central Cancer Registry and the Korean Liver Cancer Association.

Statement of Ethics

This study was approved by the Institutional Review Board of the Catholic University of Korea (Approval No. UC24ZASI0019). The need for informed consent was waived by the Institutional Review Board of Catholic University of Korea (UC24ZASI0019).

Conflict of Interest Statement

The authors declare no conflicts of interest pertaining to this study.

Funding Sources

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (RS-2023-00208767).

Author Contributions

Conception or design of the work: H.N. and S.H.B. Data analysis and interpretation and drafting the article: H.N. Critical revision of the article: P.S.S., S.W.L., D.S.S., J.H.K., J.W.J., C.W.K., and S.H.B. All authors have read and approved the final version of the manuscript.

References

1. Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I,. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA CancerJClin. 2024;74(3): 229–63.

2. Rumgay H, Arnold M, Ferlay J, Lesi O, Cabasag CJ, Vignat J,. Global burden of primary liver cancer in 2020 and predictions to 2040.JHepatol. 2022;77(6): 1598–606.

3. Rivera-Esteban J, Munoz-Martinez S, Higuera M, Sena E, Bermudez-Ramos M, Banares J,. Phenotypes of metabolic dysfunction-associated steatotic liver disease-associated hepatocellular carcinoma. Clin Gastroenterol Hepatol. 2024;22(9): 1774–89.e8.

4. Choi S, Kim BK, Yon DK, Lee SW, Lee HG, Chang HH,. Global burden of primary liver cancer and its association with underlying aetiologies, sociodemographic status, and sex differences from 1990-2019:aDALY-based analysis of the Global Burden of Disease 2019 study. Clin Mol Hepatol. 2023;29(2): 433–52.

5. Chon YE, Park SY, Hong HP, Son D, Lee J, Yoon E,. Hepatocellular carcinoma incidence is decreasing in Korea but increasing in the very elderly. Clin Mol Hepatol. 2023;29(1): 120–34.

6. Muscaritoli M, Arends J, Bachmann P, Baracos V, Barthelemy N, Bertz H,. ESPEN practical guideline: clinical Nutrition in cancer. Clin Nutr. 2021;40(5): 2898–913.

7. Bischoff SC, Bernal W, Dasarathy S, Merli M, Plank LD, Schutz T,. ESPEN practical guideline: clinical nutrition in liver disease. Clin Nutr. 2020;39(12): 3533–62.

8. Bouillanne O, Morineau G, Dupont C, Coulombel I, Vincent JP, Nicolis I,. Geriatric nutritional risk Index:anew index for evaluating at-risk elderly medical patients. AmJClin Nutr. 2005;82(4): 777–83.

9. Wang D, Hu X, Xiao L, Long G, Yao L, Wang Z,. Prognostic nutritional index and systemic immune-inflammation index predict the prognosis of patients with HCC.JGastrointest Surg. 2021;25(2): 421–7.

10. Si Y, Xu P, Xu A, Wang P, Zhao K. Geriatric nutritional risk index asaprognostic factor in patients with hepatocellular carcinoma following transarterial chemoembolization:aretrospective study. Medicine. 2022;101(51): e32322.

11. Hiraoka A, Kumada T, Tada T, Hirooka M, Kariyama K, Tani J,. Geriatric nutritional risk index as an easy-to-use assessment tool for nutritional status in hepatocellular carcinoma treated with atezolizumab plus bevacizumab. Hepatol Res. 2023;53(10): 1031–42.

12. Xu X, Kang F, Zhang N, Niu Y, Jia J. Geriatric nutritional risk index and the survival of patients with hepatocellular carcinoma:ameta-analysis. Horm Metab Res. 2023;55(10): 692–700.

13. Johnson PJ, Berhane S, Kagebayashi C, Satomura S, Teng M, Reeves HL,. Assessment of liver function in patients with hepatocellular carcinoma:anew evidence-based approach-the ALBI grade.JClin Oncol. 2015;33(6): 550–8.

14. Demirtas CO, D’Alessio A, Rimassa L, Sharma R, Pinato DJ. ALBI grade: evidence for an improved model for liver functional estimation in patients with hepatocellular carcinoma. JHEP Rep. 2021;3(5): 100347.

15. Hiraoka A, Kumada T, Tsuji K, Takaguchi K, Itobayashi E, Kariyama K,. Validation of modified ALBI grade for more detailed assessment of hepatic function in hepatocellular carcinoma patients:amulticenter analysis. Liver Cancer. 2019;8(2): 121–9.

16. Hiraoka A, Michitaka K, Kumada T, Izumi N, Kadoya M, Kokudo N,. Validation and potential of albumin-bilirubin grade and prognostication inanationwide survey of 46,681 hepatocellular carcinoma patients in Japan: the need foramore detailed evaluation of hepatic function. Liver Cancer. 2017;6(4): 325–36.

17. Kudo M. Newly developed modified ALBI grade shows better prognostic and predictive value for hepatocellular carcinoma. Liver Cancer. 2022;11(1):1– 8.

18. Kim BH, Yun EH, Lee JH, Hong G, Park JY, Shim JH,. Advancing Korean nationwide registry for hepatocellular carcinoma:asystematic sampling approach utilizing the Korea Central Cancer Registry database.JLiver Cancer. 2024;24(1): 57–61.

19. Korean Liver Cancer Association KLCA and National Cancer Center NCC Korea. 2022 KLCA-NCC Korea practice guidelines for the management of hepatocellular carcinoma.JLiver Cancer. 2023;23(1):1– 120.

20. World Health Organization. 2023 Body mass index: adult weight status categories. 2024.

21. Bauer KC, Littlejohn PT, Ayala V, Creus-Cuadros A, Finlay BB. Nonalcoholic fatty liver disease and the gut-liver Axis: exploring an undernutrition perspective. Gastroenterology. 2022;162(7): 1858–75.e2.

22. Imaoka Y, Ohira M, Kobayashi T, Honmyo N, Hamaoka M, Onoe T,. Impact of geriatric nutritional risk index after initial hepatectomy for hepatocellular carcinoma:aretrospective cohort study with the Hiroshima Surgical Study Group of Clinical Oncology(HiSCO).JGastrointest Surg. 2023;27(6): 1152–8.

23. Li L, Wang H, Yang J, Jiang L, Yang J, Wu H,. Geriatric nutritional risk index predicts prognosis after hepatectomy in elderly patients with hepatitisBvirus-related hepatocellular carcinoma. Sci Rep. 2018;8(1): 12561.

24. Kanno H, Goto Y, Sasaki S, Fukutomi S, Hisaka T, Fujita F,. Geriatric nutritional risk index predicts prognosis in hepatocellular carcinoma after hepatectomy:apropensity score matching analysis. Sci Rep. 2021;11(1): 9038.

25. Nam H, Yang H, Chun HS, Lee HA, Nam JY, Jang JW,. Impact of low skeletal muscle mass on long-term outcomes in hepatocellular carcinoma treated with trans-arterial radioembolization:aretrospective multi-center study. Cancers. 2023;15(21): 5195.

26. Pinato DJ, Sharma R, Allara E, Yen C, Arizumi T, Kubota K,. The ALBI grade provides objective hepatic reserve estimation across each BCLC stage of hepatocellular carcinoma.JHepatol. 2017;66(2): 338–46.

27. Hiraoka A, Kumada T, Kudo M, Hirooka M, Tsuji K, Itobayashi E,. Albumin-Bilirubin(ALBI) grade as part of the evidence-based clinical practice guideline for HCC of the Japan society of hepatology:acomparison with the liver damage and child-pugh classifications. Liver Cancer. 2017;6(3): 204–15.

28. Hiraoka A, Kumada T, Michitaka K, Toyoda H, Tada T, Ueki H,. Usefulness of albumin-bilirubin grade for evaluation of prognosis of 2584 Japanese patients with hepatocellular carcinoma.JGastroenterol Hepatol. 2016;31(5): 1031–6.