Hepatocellular carcinoma (HCC) represents a significant global health concern, ranking as the third-leading cause of cancer-related death in 2020 [1]. While surgical resection is a fundamental curative strategy for selected patients, the overall survival (OS) post-operation is often suboptimal. Given the highly heterogeneous nature of HCC and a high postoperative recurrence rate (70%–80% at 5 years), accurately predicting patient prognosis remains challenging [2,3].

Given the evident association between HCC and chronic liver disease, any effort to estimate prognosis should consider not only tumor burden but also the extent of liver function impairment. Existing liver function scores/grades such as Child-Turcotte-Pugh (CTP) and the more recent albumin-bilirubin (ALBI) have been widely utilized and validated [4-6]. Nonetheless, they are not always the optimal choice for risk stratification in patients suitable for liver resection, since most such candidates have good liver reserve, often classified as CTP-A with normal bilirubin levels [5,7]. The indocyanine green clearance test, renowned for its accuracy in assessing liver function, is frequently employed in Japan. However, its technical complexity has limited broader adoption in Western nations [7,8].

Several international staging systems have emerged to stratify HCC patients based on expected outcomes. Notable among these are the American Joint Committee on Cancer tumor, node, metastasis (TNM), the Cancer of the Liver Italian Program (CLIP), the Japan Integrated Staging (JIS) score, and the Barcelona Clinic Liver Cancer (BCLC) staging system [9-12]. The TNM system concentrates solely on the anatomical spread of the disease, whereas the others consider only the CTP grade as a measure of liver function. The BCLC staging system, which takes into account the patient’s performance status, CTP grade, and tumor burden, also offers treatment recommendations but restricts liver resection to patients with a solitary tumor at very early (BCLC-0) and early (BCLC-A) stages [12]. This approach may omit patients who could benefit from surgery despite not conforming to the BCLC criteria, as indicated by emerging studies [13,14]. A multitude of studies have developed prognostic models post-HCC resection; however, most focus on pathological factors such as tumor differentiation and microvascular invasion [15-17]. Thus, there exists potential to construct a robust, preoperatively available model that combines liver function and tumor burden, designed specifically to estimate survival risk in surgical patients with HCC.

In the present study, our objective was to develop a prognostic liver function score based on readily available clinical parameters for patients undergoing HCC resection. Our secondary goal was to craft a simple prognostic model by incorporating the liver function score with preoperative tumor-related parameters.

Study population and data collection

Eleven hundred HCC patients who underwent liver resection at our institute from January 2004 to August 2023 were retrospectively identified. Patients meeting any of the following criteria were excluded from the study: (1) those undergoing a second or subsequent liver resection due to recurrent HCC; (2) individuals diagnosed histologically with combined or mixed hepatocellular-cholangiocarcinoma; (3) those possessing other active malignant diseases at the time of surgery; and (4) those lacking crucial information. Ultimately, 827 patients were included for further analysis.

Data regarding patient age, sex, previous treatment for HCC, and laboratory parameters including hepatitis virus status, total bilirubin levels, albumin levels, aspartate aminotransferase (AST), alanine aminotransferase (ALT), platelet counts, prothrombin activities, alpha-fetoprotein (AFP), des-γ-carboxy prothrombin (DCP), and indocyanine green retention rates at 15 minutes (ICG-R15) were extracted from medical records. The largest tumor sizes were assessed using preoperative computed tomography (CT). The number of tumors was ascertained through a combination of imaging scans, intraoperative probing, and postoperative dissection. For patients with a history of HCC treatment, all nodules identified were factored into the tumor burden assessment, regardless of the confirmation of viable tumor cells through histological analysis. Importantly, transplantation criteria that incorporated tumor size and number, specifically the Milan criterion and the up-to-seven criterion, were employed as parameters of tumor burden. Additionally, the era of hepatectomy (2004–2009, 2010–2016, or 2017–2023), extent of resection (major resection involving three or more segments), surgical approach (open or laparoscopic), with or without concurrent ablation, as well as the incidence of macroscopic residual disease and in-hospital mortality, were also documented. Macroscopic residual tumors were evaluated based on intraoperative observations and the initial CT scan conducted approximately four weeks post-surgery. The study’s endpoint was OS, which was calculated from the date of surgery to the date of the last follow-up or death from any cause.

This study was approved by the institutional ethics committee (approval no.: 2024-05) as a retrospective analysis of collected data, conducted in accordance with the ethical standards of the World Medical Association’s Declaration of Helsinki. The requirement for individual written consent in this retrospective analysis was waived.

Surgical indications, procedures, and follow-up

The possibility of surgical resection was evaluated using a variety of imaging modalities, such as contrast-enhanced CT scans, ultrasonography (US), and magnetic resonance imaging (MRI). Preoperative assessment of liver function entailed a combination of liver biochemistry tests, ICG-R15 testing, and, when necessary, technetium-99m-galactosyl human serum albumin scintigraphy. Liver resection as monotherapy was undertaken when all tumors were considered resectable, while still preserving adequate remnant liver function or volume, based on pre- and intraoperative evaluations. In cases with multiple nodules, where achieving a cancer-free outcome through sole liver resection proved challenging, concomitant ablation was performed. Here, liver resection targeted the largest or the most problematic tumor toablate, whereas ablation therapy was directed at distant, particularly deep-seated tumors—typically those smaller than 2 cm in size—that were difficult to remove via wedge resection. Intraoperative US-guided microwave ablation (Microtaze AFM-712; Alfresa Pharma Corp.) was specifically employed for ablation therapy. When minor extrahepatic spread, such as lymph node and adrenal gland metastases, was confirmed based on pre- or intra-operative findings, simultaneous resection was carried out if it was deemed macroscopically resectable. All procedures were conducted after securing written informed consent from the patient. Follow-up examinations included measuring tumor markers and performing imaging tests such as CT, US, or MRI as needed, initially scheduled for four weeks post-surgery, and subsequently every three months. Beyond the second year, follow-up intervals were customized to the needs of individual cases, occurring every four to six months.

Statistical analysis

Sample size was not formally calculated prior to the study. Patients were randomly divided into a training cohort (n = 498) and a validation cohort (n = 329) in a 6:4 ratio for internal validation. Categorical variables were presented as counts and percentages, while continuous variables were presented as medians with interquartile ranges. Patient characteristics were compared between the training and validation cohorts using the chi-square or Fisher’s exact tests for categorical variables and the Mann–Whitney U-test for continuous variables. A multivariate regression analysis using the Cox proportional hazards model was conducted to identify factors associated with OS in the training cohort. Stepwise backward selection, based on the Akaike information criterion, was used to determine the final parameters included in the model. Prognostic liver function scores were calculated using the coefficients of independent parameters from the subsequent multivariate analysis. The predictive ability of the created liver function scores was then compared with existing scores, including the CTP score, ALBI score [4], Platelet-albumin score [7], and AST-platelet ratio index [18], using time-dependent area under the receiver operating characteristic curve (AUC) analysis. Maximally selected rank statistics identified the cutoff value of the liver function score. A subgroup analysis, stratified by patient characteristics such as the presence or absence of previous treatment for HCC, was performed to assess the prognostic significance of the liver function score, adjusting for potential confounding factors. The discriminatory performance of the final prognostic model was compared with those of TNM and BCLC stages, using time-dependent AUC, Harrell’s concordance index (C-index), and Somers’ D score [19]. The Kaplan–Meier method with the log-rank test was used to compare survival curves among subgroups with different risks. All statistical analyses were performed using R software version 4.2.2 (Foundation for Statistical Computing), with significance set at p < 0.05.

Patient characteristics

Table 1 summarizes the clinical and surgical characteristics of the training and validation cohorts. The prevalence of hepatitis B virus antigen S was higher in the validation cohort than in the training cohort (22.2% vs. 15.1%, p = 0.01), but the other variables were similar.

Table 1 . Patient characteristics

Characteristic	Training cohort (n = 498)	Validation cohort (n = 329)	p
Age (yr)	70 (63–76)	71 (63–77)	0.41
Sex, male	383 (76.9)	248 (75.4)	0.62
Previous treatment for HCCa)			0.91
TACE	108 (21.7)	67 (20.4)
RFA	16 (3.2)	7 (2.1)
TACE + RFA	45 (9.0)	27 (8.2)
Drug therapy	15 (3.0)	10 (3.0)	0.54
HAIC	14 (93.3)	8 (80.0)
Targeted therapy	1 (6.7)	2 (20.0)
HBs-Ag positive	75 (15.1)	73 (22.2)	0.01
HCV-Ab positive	282 (56.6)	165 (50.2)	0.07
Total bilirubin (mg/dL)	0.7 (0.5–0.9)	0.7 (0.6–0.9)	0.60
Albumin (g/dL)	4.0 (3.6–4.3)	3.9 (3.6–4.3)	0.82
AST (U/L)	42 (28–63)	39 (28–59)	0.34
ALT (U/L)	35 (23–58)	33 (21–55)	0.14
Platelet count (×104/μL)	15.1 (11.0–19.7)	14.7 (11.2–21.0)	0.44
Prothrombin activity (%)	84 (77–93)	85 (78–93)	0.78
ICG-R15 (%)	15.5 (10.0–22.7)	14.3 (9.3–23.2)	0.37
Child-Turcotte-Pugh score			0.78
5	366 (73.5)	240 (72.9)
6	92 (18.5)	68 (20.7)
7	33 (6.6)	18 (5.5)
8	5 (1.0)	3 (0.9)
9	2 (0.4)	0 (0)
AFP (ng/mL)	16 (7–145)	15 (6–222)	0.58
DCP (mAU/mL)	78 (27–673)	86 (26–1,121)	0.46
ASAL score			0.68
Low (< 2.7)	338 (67.9)	218 (66.3)
High (≥ 2.7)	160 (32.1)	111 (33.7)
Largest tumor size (cm)	3.5 (2.5–5.0)	3.5 (2.5–5.5)	0.69
Tumor number	1 (1–2)	1 (1–2)
Beyond Milan criterion	207 (41.6)	141 (42.9)
Beyond up-to-seven criterion	132 (26.5)	102 (31.0)
BCLC stage			> 0.99
0	59 (11.8)	39 (11.9)
A	255 (51.2)	170 (51.7)
B	113 (22.7)	74 (22.5)
C	71 (14.3)	46 (14.0)
TNM stage			0.66
IA	59 (11.8)	39 (11.9)
IB	203 (40.8)	122 (37.1)
II	134 (26.9)	103 (31.3)
IIIA	31 (6.2)	25 (7.6)
IIIB	57 (11.4)	29 (8.8)
IVA	8 (1.6)	6 (1.8)
IVB	6 (1.2)	5 (1.5)
Era of surgery			0.71
2004–2009	173 (34.7)	107 (32.5)
2010–2016	193 (38.8)	127 (38.6)
2017–2023	132 (26.5)	95 (28.9)
Major hepatic resection	87 (17.5)	59 (17.9)	0.93
Laparoscopic resection	138 (27.7)	92 (28.0)	> 0.99
Concomitant ablation	96 (19.3)	78 (23.7)	0.14
Macroscopic residual disease	7 (1.4)	7 (2.1)	0.43
In-hospital mortality	6 (1.2)	3 (0.9)	> 0.99
Median follow-up time (mon)b)	93.5 (80.4–104.6)	85.1 (73.0–98.6)	0.56

Values are presented as number (%) or median (interquartile range).

HCC, hepatocellular carcinoma; TACE, transcatheter arterial chemoembolization; RFA, radiofrequency ablation; HAIC, hepatic arterial infusion chemotherapy; HBs-Ag, hepatitis B virus antigen S; HCV-Ab, anti-hepatitis C virus antibody; AST, aspartate aminotransferase; ALT, alanine aminotransferase; ICG-R15, indocyanine green retention rates at 15 minutes; AFP, alpha-fetoprotein; DCP, des-γ-carboxy prothrombin; ASAL, aspartate aminotransferase-albumin; BCLC, Barcelona Clinic Liver Cancer; TNM, American Joint Committee on Cancer tumor, node, metastasis.

^a)Some patients who underwent TACE and/or RFA as previous treatments overlap with those who received drug therapy.

^b)95% confidence interval for the values in the parentheses.

Multivariate analysis and creation of liver function score

Focusing on parameters related to liver function, albumin, and AST were identified as the most significant predictors in the multivariate analysis (Table 2). Based on the coefficients derived from a subsequent multivariate analysis, the AST-albumin (ASAL) score was calculated using the following formula: exp [AST (IU/L) × 0.005 – albumin (g/dL) × 1.043] × 100. The ASAL score consistently demonstrated superior predictive ability in terms of time-dependent AUC compared to other scores in both the training (Fig. 1A) and validation cohorts (Fig. 1B). However, the ALBI score outperformed the ASAL score exclusively at the 24-month mark in the training cohort.

Table 2 . Multivariable analysis of prognostic factors using stepwise backward selection in the training cohort

	HR (95% CI)	SE	z	p	Coefficient
HBs-Ag positive	0.57 (0.35–0.93)	0.247	–2.265	0.02	–0.560
HCV-Ab positive	0.69 (0.49–0.96)	0.170	–2.223	0.03	–0.378
Albumin	0.41 (0.30–0.57)	0.164	–5.389	< 0.001	–0.882
AST	1.01 (1.00–1.01)	0.002	3.274	0.001	0.006
Prothrombin activity	0.99 (0.98–1.00)	0.006	–1.616	0.11	–0.009
AFP ≥ 20 (ng/mL)	1.41 (1.05–1.87)	0.147	2.322	0.02	0.341
Beyond up-to-seven criterion	1.60 (1.17–2.19)	0.159	2.961	0.003	0.472
ASAL score parameters
Albumin	0.35 (0.27–0.47)	0.142	–7.358	< 0.001	–1.043
AST	1.01 (1.00–1.01)	0.002	3.334	< 0.001	0.005

The sixteen parameters tested were as follows: age, sex, hepatitis B virus antigen S (HBs-Ag), anti-hepatitis C virus antibody (HCV-Ab), total bilirubin, albumin, aspartate aminotransferase (AST), alanine aminotransferase, platelet counts, prothrombin activities, indocyanine green retention rates at 15 minutes, alpha-fetoprotein (AFP) ≥ 20 ng/mL, des-γ-carboxy prothrombin ≥ 40 mAu/mL, Milan and up-to-seven criteria, and era of hepatectomy (2004–2009, 2010–2016, or 2017–2023).

HR, hazard ratio; CI, confidence interval; SE, standard error; ASAL, aspartate aminotransferase-albumin.

Fig 1. Time-dependent area under the receiver operating characteristic curve analysis for overall survival estimation in the training (A, C) and validation (B, D) cohorts. AUC, area under the receiver operating characteristic curve; ASAL, aspartate aminotransferase-albumin; ALBI, albumin-bilirubin; PAL, platelet-albumin; CTP, Child-Turcotte-Pugh; APRI, aspartate aminotransferase-platelet ratio index; TNM, American Joint Committee on Cancer tumor, node, metastasis; BCLC, Barcelona Clinic Liver Cancer.

Cutoff determination of the ASAL score and subgroup analysis

Using maximally selected rank statistics, the optimal cutoff for the ASAL score was determined to be 2.7. When classifying the training cohort into groups with high ASAL scores (≥ 2.7, n = 160) and low scores (< 2.7, n = 338), the 5-year OS rate was significantly lower in the high ASAL score group compared to the low group (39% vs. 70%, p < 0.001) (Fig. 2A). This pattern was replicated in the validation cohort (ASAL score high: 38% vs. low: 68%, p < 0.001) (Fig. 2B). When patients were categorized by treatment history for HCC, BCLC stage, TNM stage, and era of surgery, the high ASAL score consistently correlated with lower OS in both the training and validation cohorts. Adjusted hazard ratios ranged from 1.88 to 4.24 in the training cohort and from 1.55 to 15.03 in the validation cohort, with all subgroups displaying hazard ratios greater than 1. No significant interactions were observed across subgroups (all p for interaction > 0.05) (Fig. 3).

Fig 2. Kaplan-Meier estimates of overall survival according to the ASAL score and prognostic model: (A) ASAL score in the training cohort, (B) ASAL score in the validation cohort, (C) prognostic model in the training cohort, (D) prognostic model in the validation cohort, (E) prognostic model limited to treatment-naive patients in the training cohort, and (F) prognostic model limited to treatment-naive patients in the validation cohort. ASAL, aspartate aminotransferase-albumin.

Fig 3. Forest plot of subgroup analysis revealing the association between high ASAL score and overall survival across training and validation cohorts. Hazard ratios were adjusted for age, sex, hepatitis B virus antigen S status, anti-hepatitis C virus antibody status, largest tumor size, tumor count, and alpha-fetoprotein levels. ASAL, aspartate aminotransferase-albumin; CI, confidence interval; BCLC, Barcelona Clinic Liver Cancer; TNM, American Joint Committee on Cancer tumor, node, metastasis.

Construction of the prognostic model

Among the tumor-related parameters (i.e., AFP, DCP, Milan criterion, and up-to-seven criterion), AFP ≥ 20 ng/mL and beyond the up-to-seven criterion remained significant factors (Table 2). We subsequently developed a straightforward scoring model, assigning scores of 0 or 1 for low or high ASAL score, low or high AFP, and within or beyond the up-to-seven criterion (Table 3). We categorized patients into four groups based on their total scores, ranging from 0 to 3. The survival curves for these groups were distinctly separated in both the training and validation cohorts (both p < 0.001, Fig. 2C, 2D). This scoring model demonstrated improved discriminatory ability compared to the TNM and BCLC stages in terms of time-dependent AUC (Fig. 1C, 1D), the C-index, and Somers’ D score in both cohorts (Table 4). Moreover, even within the treatment-naive cohort, this scoring model effectively separated the survival curves (Fig. 2E, 2F) and showed superior C-index and Somers’ D scores compared to the BCLC and TNM staging systems (Table 4).

Table 3 . Definition of the scoring system

	Score
	0	1
ASAL score	Low	High
AFP (ng/mL)	< 20	≥ 20
Up-to-seven criterion	Within	Beyond

ASAL, aspartate aminotransferase-albumin; AFP, alpha-fetoprotein.

Table 4 . Comparison of prognostic performance (C-index and Somers’ D) between our model, TNM, and BCLC stages in all patients and treatment-naive subgroups across training and validation cohorts

	Training cohort		Validation cohort
	C-index	Somers’ D	C-index	Somers’ D
Our model
All patients	0.674	0.639	0.722	0.764
Treatment-naive	0.663	0.589	0.748	0.809
BCLC stage
All patients	0.663	0.627	0.689	0.727
Treatment-naive	0.623	0.534	0.693	0.675
TNM stage
All patients	0.672	0.614	0.675	0.650
Treatment-naive	0.636	0.530	0.672	0.606

C-index, concordance index; TNM, American Joint Committee on Cancer tumor, node, metastasis; BCLC, Barcelona Clinic Liver Cancer.

In this study, we developed a preoperative prognostic model for HCC patients undergoing liver resection by incorporating the novel ASAL score with conventional tumor-related parameters (AFP and up-to-seven criterion). The ASAL score demonstrated superior performance compared to existing scores at most postoperative time points, and emerged as an independent prognostic factor, independent of TNM and BCLC stages. Additionally, our straightforward scoring model effectively stratified patients into four distinct prognostic groups, and showed enhanced discriminatory ability compared to the BCLC and TNM stages. These results were confirmed in an independent cohort, highlighting the potential utility of the ASAL score and our prognostic model in guiding clinical practice.

Liver function parameters such as transaminases and albumin are routinely used to assess hepatic inflammation and preserved liver function. Notably, abnormal levels of AST and albumin were reported as the most prominent characteristics of HCC patients with cirrhotic hepatitis [20]. AST is predominantly localized within the mitochondria of liver cells, whereas ALT is primarily located outside these organelles in liver cells [21]. Liver injury, including mitochondrial damage, results in a greater release of AST relative to ALT [22]. Cancer cells increase energy production through augmented aerobic glycolysis, unlike normal cells [23]. AST supports this process by facilitating the transport of nicotinamide adenine dinucleotide hydrogen into mitochondria through malate-aspartate shuffling [24]. Consequently, patients with HCC are more likely to exhibit elevated AST levels compared to ALT. Additionally, liver fibrosis may reduce AST clearance, contributing to its accumulation in the blood [25]. A higher AST level is also associated with an increased hepatitis B viral load, which correlates with decreased OS in HCC patients [26]. Takeishi et al. [27] noted that a high AST level was significantly associated with tumor size and the rate of tumor capsule formation with cancer cell infiltration in non-B, non-C HCC. Albumin, conversely, is a globular protein synthesized exclusively by liver cells that serves as an indicator of liver synthetic function, nutritional status, and systemic inflammation [28]. As the predominant protein in plasma, human serum albumin performs several biological functions, including regulating plasma oncotic pressure, maintaining vascular permeability, and influencing immune responses. Furthermore, albumin acts as a vital and independent biomarker for a wide range of human diseases [29]. Various studies have underscored the importance of serum albumin levels, used alone or as part of combination biomarkers (e.g., C-reactive protein to albumin ratio), in predicting HCC prognosis following curative resection [29]. In a comprehensive study involving non-transplant HCC patients (n = 1,889), Carr et al. [30] demonstrated that AST and albumin exhibited the highest hazard ratios for survival among patients with both elevated and low serum AFP levels, and across different tumor sizes. They also revealed that abnormal serum AST and albumin levels were linked with more aggressive tumor traits, such as multifocality and portal vein tumor thrombosis. Peng et al. [20] found that a high AST to albumin ratio significantly correlated with more advanced tumors and poorer overall and recurrence-free survival in HCC patients post-liver resection. Therefore, the combination of AST and albumin may serve as an indicator of liver function and potential tumor aggressiveness, thereby offering a robust predictor of OS in HCC patients.

A limitation of the CTP score arises from its inherent subjectivity in evaluating conditions such as ascites and encephalopathy [4]. Patients with HCC who have a CTP score of 5 may exhibit a wide range of clinical manifestations, from the absence of chronic liver disease to chronic inflammation or well-compensated cirrhosis, all resulting in varying long-term prognoses [31]. The ALBI grade offers an objective measure based on continuous albumin and bilirubin levels, yet its applicability to surgical candidates often hinges predominantly on serum albumin levels [32]. Significantly, our findings indicated that albumin played a more critical role than AST in terms of its impact on regression coefficients. Furthermore, liver function scores based on continuous albumin values demonstrated superior prognostic capabilities compared to the CTP score or the AST-platelet ratio index in this study. Kokudo et al. [32] put forward the albumin-Indocyanine Green Evaluation grade for making surgical decisions. Nevertheless, their choice of the ICG-R15 parameter lacked a statistical basis. Additional research is essential to determine the most effective combination of liver function parameters focused on albumin for predicting HCC prognosis after liver resection.

Among the various staging systems for HCC, the JIS score utilizes the TNM classification for tumor staging. The Okuda system and the CLIP score incorporate the extent of tumor involvement in the liver (≤ 50% or > 50% of the total liver) [10,33], yet this parameter may not be feasible for classifying surgical patients, as those with HCC occupying more than 50% of the liver are rarely considered suitable for resection. The Milan criteria and the up-to-seven criteria, which are based on the number of tumors, their size, vascular invasion, and the presence of extrahepatic metastases, are straightforward metrics applicable to liver transplantation. Recent studies have indicated the up-to-seven criteria as an effective measure for directing the choice among liver resection, transcatheter arterial chemoembolization (TACE), and systemic treatments in patients with BCLC stage B HCC [34,35]. Nong et al. [34] observed in their study of 340 patients with BCLC-B HCC undergoing either resection or TACE that those who met the up-to-seven criteria experienced significantly better OS following liver resection compared to TACE (p < 0.001). Furthermore, AFP, a vital tumor marker for HCC, is included in several staging systems, such as CLIP and the Chinese University Prognostic Index, with varying thresholds [10,36]. In their analysis of 2,579 patients with HCC, Hsu et al. [37] identified the optimal cutoff values for AFP in prognosis prediction as either ≥ 20 ng/mL or ≥ 400 ng/mL (adjusted hazard ratios of 1.545 and 1.471, respectively). By adopting a cutoff value of ≥ 20 ng/mL, we validated its significance in prognosis. Ultimately, employing a combination of the ASAL score, up-to-seven criterion, and AFP offers a comprehensive assessment of the prognosis for surgical patients with HCC.

This study has several limitations in addition to its retrospective nature. First, since the ASAL score and our prognostic model are based on Japanese subjects at a single institution, there is bias relating to the prevalence of hepatitis virus and the status of antiviral treatment. Second, including cohorts who have previously undergone treatment for HCC may introduce bias, as their prognosis and preoperative liver function may differ from those of treatment-naive patients. We attempted to address this by conducting subgroup analyses excluding treatment-experienced patients, yet the findings remained consistent. Third, in patients undergoing combined liver resection with ablation, not all ablated lesions were biopsied, resulting in many lesions without histological confirmation of HCC. Finally, we investigated only OS because our data did not permit analysis of outcomes such as recurrence-free survival, disease-specific survival, and postoperative liver failure. Continuously collecting data may yield more detailed and precise results.

In conclusion, integrating the novel ASAL score with the up-to-seven criterion and AFP demonstrated promise for prognostication in HCC patients undergoing liver resection. Our prognostic model could assist in tailoring decisions regarding perioperative adjuvant therapy and follow-up intensity, rather than merely determining eligibility for surgical resection. However, further external validation studies across diverse patient populations are necessary to confirm its clinical utility and applicability.

The authors would like to thank all the patients and medical staff at the institution who contributed to this study.

None.

No potential conflict of interest relevant to this article was reported.

Conceptualization: SI. Data curation: SI, TA, MK, TN. Methodology: SI. Supervision: NY. Writing - original draft: SI. Writing - review & editing: TA, NY.