Biliary tract cancer (BTC) is an invasive adenocarcinoma that can originate from the bile ducts, gallbladder, or ampulla of Vater, and has a poor prognosis (1). It is the second most common hepatobiliary cancer worldwide, with high incidence rates in the Asia-Pacific region and South America (2, 3). Complete surgical resection is the only curative treatment for BTC (4). Despite recent advancement in surgical techniques, perioperative management, and postoperative treatment, the long-term prognosis of surgically treated patients is unsatisfactory (5, 6). Long-term cure is achievable only in patients with early-stage tumors that are amenable to resection (7). However, a large-scale clinical trial demonstrated that disease recurrence occurred in 60% of cases after tumor resection resulting in a 5-year survival rate of under 50%, even among patients administered adjuvant chemotherapy (7). Therefore, great efforts are being made to improve early diagnosis and multimodal treatment including chemotherapy, for better outcomes (6, 8). The survival of patients with BTC is affected by several factors including tumor-node-metastasis stage (9), neurovascular invasion, tumor differentiation, and resection margin status, which are typically assessed intraoperatively.

Recently, a novel liver function assessment model, the albumin–bilirubin (ALBI) grade, was introduced (10). This straightforward model is derived from serum albumin and total bilirubin levels obtained from routine blood tests and has gained attention as a noninvasive index for evaluating liver function. The ALBI grade outperforms the Child–Pugh grade in assessment of liver function, predicting complications, and estimating prognosis in liver disease (11, 12). Another study showed that the modified ALBI (mALBI) grade, which subdivides patients by conventional ALBI score, offers improved accuracy for assessment of liver function and enhanced utility in stratifying patient prognosis (13). A growing body of literature has demonstrated the prognostic relevance of the ALBI score across a wide spectrum of malignancies extending beyond BTC (14-17). However, to the best of our knowledge, the mALBI score has not been investigated in a multicenter case series involving patients after resection of BTC. The purpose of this multicenter study was to evaluate the prognostic utility of the mALBI score in such patients.

Study design. Patients with histologically confirmed perihilar cholangiocarcinoma (PHCC), distal cholangiocarcinoma (DCC), gallbladder carcinoma (GBC), ampullary carcinoma (AC), or intrahepatic cholangiocarcinoma (ICC), diagnosed based on surgically resected specimens, were retrospectively enrolled in this study at four centers between 2001 and 2023. These patients were observed until the follow-up cutoff date (December 2023) or death following surgical resection. Clinicopathological data were obtained from the medical records. Collected clinical data included sex, age, tumor location, presence of preoperative cholangitis, preoperative laboratory parameters (albumin and total bilirubin), and tumor markers [including carcinoembryonic antigen and carbohydrate antigen (CA19-9)], surgical details (type of surgery, operative time, intraoperative blood loss, and concomitant major vascular resection and reconstruction), postoperative hospital stay, postoperative complications, details of adjuvant chemotherapy, recurrence, survival, and pathological findings (tumor differentiation, pathological T factor, lymph node metastasis, lymphatic, vascular, and perineural invasion, and residual tumor status). Pathological findings were classified according to the seventh edition of the Union for International Cancer Control TNM staging system (18). Preoperative cholangitis was diagnosed according to the criteria outlined in the 2018 Tokyo guidelines (19). Postoperative complications were classified according to the Clavien–Dindo classification system (20). Major complications were defined as grade ≥3. This study was approved by the Ethics Committee of Kobe University (approval ID: B240209) and other centers and was conducted in accordance with the Declaration of Helsinki (2024). Informed consent was obtained from all participants using an opt-out approach.

mALBI score. The ALBI score was defined as follows: (log₁₀total bilirubin × 0.66) + (albumin × -0.085) with total bilirubin (μmol/l) and albumin (g/l) determined preoperatively. ALBI grades were then categorized as grade 1 for scores ≤−2.60, grade 2 for scores >−2.60 to −1.39, and grade 3 for scores >−1.39. Consistent with previous studies, ALBI grade 2 was further subdivided using a cut-off value of −2.27 into grades 2a (−2.60≤ ALBI score <−2.27) and 2b (−2.27≤ ALBI score <−1.39) (13). For this study, we defined patients with mALBI grades 1 and 2a as the low mALBI group, and those with grades 2b and 3 as the high mALBI group.

Statistical analysis. Descriptive statistics are presented as medians with ranges for continuous variables, and as numerical values with percentages for categorical variables. Continuous variables were compared using the Kruskal–Wallis test, whereas categorical variables were analyzed using Pearson’s chi-square test or Fisher’s exact test. OS was defined as the interval from the date of surgery to the date of death from any cause or last follow-up. Recurrence-free survival (RFS) was defined as the interval between surgery and relapse, death, or the last follow-up. Kaplan–Meier survival curves were used to estimate OS and RFS, and the differences between the curves were assessed using the log-rank test. Univariate analysis was conducted to identify potential risk factors for OS and RFS. Variables with a value p<0.10 in the univariate analysis, along with clinically relevant factors were included in a multivariate Cox proportional hazards model to identify factors independently associated with postoperative survival. The results were reported as hazard ratios (HRs) with 95% confidence intervals (CIs). Statistical significance was set at p<0.05. All statistical analyses were performed using JMP statistical software version 18 (SAS Institute Inc., Cary, NC, USA).

PSM analysis. PSM was used to balance patient characteristics between the high and low mALBI groups. The matching algorithm used was nearest-neighbor matching with a 1:1 ratio, and the caliper was set to 0.02. The variables age group, sex, tumor location, preoperative cholangitis, CA19-9, T category, N category, vascular invasion, lymphatic invasion, perineural invasion, and R category were matched.

Ethics statement.This study was approved by the Ethics Committee of Kobe University (approval ID: B240207) and was conducted in accordance with the principles of the Declaration of Helsinki (2024). Informed consent was obtained from all participants using an opt-out approach in which patients were given the opportunity to decline participation after being informed about the study.

Characteristics of patients and assessment of efficacy. The initial cohort comprised 716 patients. After excluding patients with distant metastases (M1) (n=28), macroscopic residual cancer after resection (R2) (n=28), double cancer (n=22), and those lost to follow-up (n=4), 634 patients were enrolled in the study (Figure 1). The baseline data of these patients are presented in Table I.

A total of 108 (17%) patients were clinically or pathologically diagnosed with PHCC, 155 (24%) with DCC, 140 (22%) with GBC, 140 (22%) with AC, and 90 (14%) with ICC. The patients had a median age of 71 years (range=25-89 years) and 382 (60%) were men. According to mALBI grade, 202 (32%) patients were in the high mALBI group, and 432 (68%) were in the low mALBI group. Pancreaticoduodenectomy was most commonly performed (n=306, 48%), followed by hepatectomy (n=202, 31.9%), and gallbladder bed resection (n=82, 13%). A total of 23 (4%) patients underwent concomitant portal vein resection and reconstruction (PVR), and 6 (1%) others underwent concomitant hepatic artery resection and reconstruction. Microscopic residual cancer was observed in the resection margins in 104 patients (16.4%). A total of 165 (26.0%) patients were administered adjuvant chemotherapy after surgical resection. After 1:1 PSM, the high mALBI group (n=131) was compared with the low mALBI group (n=131) having balanced baseline characteristics (Table I).

Associations between mALBI grade and clinical variables after PSM. The clinicopathological characteristics of patients according to mALBI are shown in Table I. No significant differences in sex, age, body mass index, tumor location, carcinoembryonic antigen, CA19-9, PVR, hepatic artery resection and reconstruction, operative time, intraoperative blood loss, postoperative complications, major complications, adjuvant chemotherapy, or any of the pathological characteristics were observed between the groups. Although the high mALBI group had a significantly longer postoperative hospital stay (p=0.045), there were no significant differences in the incidence of major postoperative complications (p=0.922).

OS. The median duration of follow-up for the whole cohort was 94 months. Patients in the high mALBI group (n=202) had significantly worse prognosis in terms of OS than those in the low mALBI group (n=432) (median=39 vs. 164 months; p<0.001) (Figure 2A, upper panel). After PSM, the high mALBI group (n=131) had a significantly worse prognosis in terms of OS than the low mALBI group (n=131) (median=44 vs. 70 months; p=0.015) (Figure 2B, upper panel).

In analyses stratified by tumor location, patients in the high mALBI group consistently demonstrated significantly worse OS than those in the low mALBI group across all tumor sites. In extrahepatic cholangiocarcinoma, the median OS was 35 months for the high mALBI group (n=133) and 61 months for the low mALBI group (n=130) (p=0.007). In GBC, the median OS was 50 months (n=24) versus 165 months (n=116), respectively (p=0.013). Among patients with AC, the median OS was 69 months for the high mALBI group (n=35), whereas it was not achieved for the low mALBI group (n=105) (p<0.001). Similarly, in ICC, the median OS was 39 months for the high mALBI group (n=9) and was not achieved for the low mALBI group (n=81) (p=0.039) (Figure 3).

Patients were further categorized according to surgical procedure into a hepatectomy and non-hepatectomy (including pancreaticoduodenectomy) groups, and OS was evaluated by mALBI group within each surgical category. In the hepatectomy-treated group, the median OS was 34 months for the high mALBI group (n=65) and 62 months for the low mALBI group (n=138) (p<0.001). In the non-hepatectomy group, the median OS was 48 months (n=137) and 165 months (n=294), respectively (p<0.001) (Figure 4).

RFS. The difference in RFS between the high and low mALBI groups was significant according to Kaplan–Meier analysis (median=25 vs. 131 months; p<0.001) (Figure 2A, lower panel). After PSM, the high mALBI group (n=131) tended to have a worse prognosis in terms of RFS than the low mALBI group (n=131) (median=29 vs. 47 months; p=0.063) (Figure 2B, lower panel).

Risk factors for OS in univariate and multivariate analyses. In univariate Cox regression analysis, preoperative CA19-9 level, PVR, differentiation, T-stage, N-stage, lymphatic invasion, vascular invasion, perineural invasion, residual cancer, adjuvant chemotherapy, and mALBI were significantly associated with OS (Table II). Variables with a value p<0.10 in the univariate analysis along with clinically relevant factors, were included in the multivariate model. Multivariate analysis revealed that high mALBI (HR=1.584, 95% CI=1.077-2.331; p=0.020), lymph node metastasis (HR=2.235, 95% CI=1.404-3.557; p<0.001), vascular invasion (HR=1.915, 95% CI=1.209-3.032; p=0.006), and residual cancer (HR=2.421, 95% CI=1.449-4.044; p<0.001) were independent prognostic factors for poor OS (Table II).

Risk factor for RFS in the univariate and multivariate analyses. In univariate Cox regression analysis, preoperative CA19-9 level, differentiation, T-stage, N-stage, lymphatic invasion, vascular invasion, perineural invasion, residual cancer, and adjuvant chemotherapy were significantly associated with RFS (Table III). Although in the univariate Cox regression analysis, high mALBI was not significantly associated with poor RFS (p=0.067), it met the threshold for being included in the multivariate analysis; this showed that a high mALBI was an independent prognostic factor for poor RFS (HR=1.447, 95% CI=1.023-2.046, p=0.037) along with the classic factors, lymph node metastasis, vascular invasion and residual cancer.

In this multicenter study using PSM, the mALBI grade was demonstrated to be a robust independent prognostic factor in patients who underwent resection for BTC remaining significantly associated with OS after adjustment for established tumor-related variables. These findings highlighted the prognostic value of preoperative hepatic functional reserve in this patient population and indicated that objective functional status has a meaningful impact on long-term outcomes. Accordingly, the mALBI grade might serve as a simple, objective, and clinically applicable tool for preoperative risk stratification in patients with BTC.

The prognostic significance of ALBI score has been consistently demonstrated in various malignancies other than BTC (14-17). In patients with advanced gastric cancer, the ALBI score was demonstrated a valuable independent prognostic indicator of postoperative recurrence and disease-specific survival (14). Notably, Ju et al. (21) further demonstrated the ALBI score to be an independent prognostic factor even in patients with gastric cancer undergoing curative treatment, supporting its utility beyond advanced disease settings. Similarly, in advanced colon cancer, a high ALBI score was shown to be an independent risk factor for OS and disease-free survival (15). In pancreatic cancer, patients with high ALBI scores and elevated CA19-9 levels exhibited the poorest prognosis, characterized by a significantly reduced 5-year survival rate (16). In the context of BTC, the ALBI score had superior prognostic value compared with the Child–Pugh score in patients with unresectable extrahepatic cholangiocarcinoma, with higher ALBI scores correlating with poorer 1- and 2-year survival rates and an overall worse prognosis (17). More recently, Aydemir et al. (22) reported that both ALBI and platelet–ALBI grades are significant prognostic factors in cholangiocellular carcinoma, further reinforcing the importance of liver function-based indices in biliary malignancies. The tumor site subgroup analysis revealed that patients with extrahepatic cholangiocarcinoma, GBC, AC, and ICC in the high mALBI group had a significantly worse prognosis. The mALBI grade was a valuable prognostic factor in all tumor sites of BTC (Figure 3). Moreover, the mALBI grade was a useful prognostic factor not only in the hepatectomy-treated group, but also in the non-hepatectomy group (Figure 4). These results emphasize that the mALBI grade is a valuable prognostic factor in BTC.

The application of PSM in this study allowed for a more rigorous evaluation of mALBI by balancing key baseline characteristics including age, tumor stage, and invasion status. Notably, before matching, the median OS in the low mALBI group was 164 months which is an exceptionally favorable survival for patients with BTC, whereas after matching, the median OS in the low mALBI group was 70 months. This discrepancy possibly suggested baseline imbalances particularly in tumor location, in the pre-matching low mALBI cohort (Table I). After adjustment for tumor location and other prognostic variables by using PSM, the mALBI grade remained an independent prognostic factor. Although PSM was performed to balance baseline characteristics, the covariates included tumor-related factors known to strongly affect prognosis. Therefore, we additionally conducted a multivariate Cox proportional hazards analysis incorporating these tumor-related factors to confirm whether the mALBI grade remained an independent prognostic factor.

The mechanism underlying the association between a high mALBI grade and poor prognosis in patients with BTC remains unclear. Several factors may contribute to this association. A high mALBI grade reflects a reduced serum albumin level and an elevated total bilirubin level; hypoalbuminemia is associated with malnutrition and cachexia and may influence macrophage activation and antitumor immune responses, thereby facilitating tumor progression (23, 24). Hyperbilirubinemia reflects tumor-related biliary obstruction and disease extent in patients with BTC; however, it also represents impaired hepatic functional reserve, chronic inflammation, and impaired physiological reserve. In addition, recent studies have highlighted the importance of host-related factors such as nutritional status and body composition in BTC. Utsumi et al. (25) demonstrated that preoperative myosteatosis and prognostic nutritional index are significant predictors of survival in older patients with BTC, while Ikuta et al. (26) reported that reduced psoas muscle thickness is associated with poor postoperative outcomes. These findings suggest that mALBI may partly reflect systemic nutritional and metabolic conditions that contribute to oncological outcomes. Therefore, the prognostic value of the mALBI grade in BTC likely reflects a composite effect of both tumor-related factors and host-related conditions. In this study, the mALBI grade stratified prognosis in both the hepatectomy-treated group, in which liver function directly affects surgical strategy, and the non-hepatectomy group, in which liver function is generally considered to have limited influence on the operative procedure (Figure 4). Furthermore, the importance of hepatic functional reserve in surgical oncology is supported by studies in hepatocellular carcinoma, such as that by Kori et al. (27), which identified liver function-related factors as key determinants of long-term survival after repeated hepatectomy.

The mALBI score may also influence the feasibility of adjuvant chemotherapy. Miwa et al. (28) reported that patients with gastric cancer with a high mALBI score were less likely to continue postoperative adjuvant chemotherapy. Similarly, Kanda et al. (14) showed that an elevated mALBI score was associated with difficulty in maintaining adjuvant therapy. However, in the present study, there were no significant differences in the administration of adjuvant chemotherapy between the high and low mALBI groups (32.3% vs. 35.4%; Table I). This difference from previous studies of other malignancies is attributable to the lack of effective adjuvant chemotherapy regimens for BTC. Further data collection is required to confirm these findings.

The findings of the present study have important clinical implications. Because mALBI can be easily calculated from routine blood tests, it serves as a simple, objective, and cost-effective preoperative prognostic indicator. Patients with high mALBI scores may require perioperative management interventions such as nutritional intervention, optimization of liver function, and close postoperative monitoring; however, further studies are needed to validate these strategies. The incorporation of mALBI into preoperative risk stratification may help optimize treatment strategies and improve postoperative management of BTC.

Study limitations. Firstly, this was a retrospective multicenter study, and it was not possible to completely exclude selection bias. Although PSM improved comparability between the groups, larger-scale prospective validation studies are needed to confirm these findings. Secondly, although we employed the mALBI classification, a direct comparison with other liver function assessment models, such as the Child–Pugh score or liver stiffness measurements, was not performed and warrants further investigation. Although previous studies, such as that by Wang et al. (17), have suggested that the mALBI score may have superior prognostic accuracy compared to the Child–Pugh score in patients with PHCC or DCC, this was not evaluated in the present study. Future studies should explore more comprehensive prognostic models incorporating multiple risk factors. Thirdly, this study did not evaluate dynamic changes in the mALBI score postoperatively or after adjuvant therapy, which may provide additional insights into its prognostic role.

This study demonstrated that mALBI grade is an important prognostic indicator of OS in patients after resection of BTC, even after adjusting for confounding variables through PSM. These findings support incorporating the mALBI score into preoperative risk stratification to optimize treatment strategies and patient management. Further prospective studies are required to validate these results and to explore potential therapeutic interventions for patients with high mALBI scores.

The Authors declare no conflicts of interest.

TT and HY contributed to the study conception and design. TT performed data analysis and drafted the manuscript. YU contributed to the clinical data collection. MA, DT, MY, YU, TO, KU, HM, IO, SK, and TF critically reviewed and revised the manuscript for important intellectual content. TF provided supervision. TT and HY verified the authenticity of all raw data. All the Authors read and approved the final manuscript and agree to be accountable for all aspects of the work.

The Authors would like to thank Editage (www.editage.jp) for English language editing.

No funding was received.

During the preparation of this manuscript, a large language model (ChatGPT, OpenAI) was used solely for language editing and stylistic improvements in select paragraphs. No sections involving the generation, analysis, or interpretation of research data were produced by generative AI. All scientific content was created and verified by the authors. Furthermore, no figures or visual data were generated or modified using generative AI or machine-learning-based image enhancement tools.