Pancreatic cancer (PC) is the fourth leading cause of cancer-related mortality in Japan, and 41,235 patients died of PC in 2024 (1). Gemcitabine (GEM) demonstrated superior anticancer efficacy and survival benefits compared with 5-fluorouracil in patients with advanced PC (APC) (23.8% vs. 4.8% and 5.65 vs. 4.41 months, respectively) (2). The combination chemotherapy of nab-paclitaxel plus GEM (AG) was subsequently developed and showed greater survival benefits than GEM monotherapy in patients with metastatic PC in North America, Europe, and Australia (8.5 vs. 6.7 months, respectively; MPACT trial) (3). In recent years, a phase 2/3 randomized controlled trial investigating the efficacy of AG compared with modified fluorouracil/leucovorin plus irinotecan plus oxaliplatin (FOLFIRINOX) or S-1 plus irinotecan plus oxaliplatin (S-IROX) was conducted, and neither modified FOLFIRINOX nor S-IROX appeared to be superior to AG as first-line (1L) chemotherapy for metastatic or recurrent PC (4). However, fluorouracil/leucovorin plus liposomal irinotecan plus oxaliplatin (NALIRIFOX) demonstrated greater survival benefits than AG in patients with metastatic PC (11.1 vs. 9.2 months) (5). However, NALIRIFOX was not covered by the national medical insurance system in Japan in 2025, and AG is therefore most commonly administered as the standard 1L chemotherapeutic regimen in real-world clinical practice in Japan (6). Several real-world studies have also demonstrated the feasibility and clinical activity of AG in patients with APC, including modified administration schedules and exploratory combination approaches (7-10).

The presence of liver metastasis is a poor prognostic factor in patients with APC (11-14). In the MPACT trial, the presence or absence of liver metastasis was included as a stratification factor (3). At that time, to the best of our knowledge, the prognostic significance of the number of liver metastases had not been investigated and remained unclear. Therefore, we previously examined the prognostic significance of the number of liver metastases in patients with APC (15). However, that study included only a small sample size (a total of 37 patients), and the statistical power was limited. Accordingly, we performed an updated analysis with a larger sample size to investigate the prognostic significance of the number of liver metastases in patients with APC who received AG as 1L chemotherapy.

Patients. From January 2015 to July 2024, 103 patients with APC received AG as 1L chemotherapy at Higashiosaka City Medical Center (Higashiosaka, Japan). The patients were categorized into three groups based on contrast-enhanced computed tomography (CT) findings: no liver metastases (NLM), several liver metastases (SLM) (1-9), and multiple liver metastases (MLM) (≥10). Overall survival (OS) was estimated using the Kaplan–Meier method and compared using the log-rank test. We retrospectively compared baseline patient characteristics, treatment response during AG, and clinical course between the non-MLM and MLM groups. Prognostic factors were analyzed using Cox regression models. The data cutoff date was 31 July 2025. All procedures performed in this study were in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual patients. The study was approved by the Ethics Review Board of Higashiosaka City Medical Center (02-1027-A).

Treatment. Nab-paclitaxel (125 mg/m²) and GEM (1,000 mg/m²) were administered intravenously on days 1, 8, and 15, followed by a 7-day rest period, and repeated until disease progression or intolerable adverse events (3). The administered dose was adjusted at the discretion of the physicians and patients. Based on the patients’ general condition, their willingness to receive subsequent treatment after AG, and the results of comprehensive genomic profiling tests (when performed), the indication and regimen for second-line (2L) chemotherapy were determined at the physicians’ discretion.

Evaluation. At diagnosis, all patients underwent contrast-enhanced CT, and the presence or absence of liver metastasis and the number of liver metastases were evaluated by physicians and radiologists. A 64-multidetector row CT scanner (Aquilion TSX-101A; Toshiba Medical Systems, Tokyo, Japan) or a 128-slice helical CT scanner (Brilliance iCT SP; Philips, Amsterdam, the Netherlands) was used. Baseline CT images were analyzed using SYNAPSE VINCENT software version 5.3 (Fujifilm, Tokyo, Japan) to determine the presence of sarcopenia. The total skeletal muscle area was measured on a single axial CT image at the level of the third lumbar vertebra using Hounsfield unit thresholds of −29 to +150 for skeletal muscle and was normalized by the square of the height to calculate the skeletal muscle index (SMI) (16-20). The SMI cutoff values for identifying sarcopenia were 43.75 cm²/m² for men and 38.5 cm²/m² for women (21). We compared the SMI at diagnosis with that at the end of AG to evaluate changes in skeletal muscle mass during AG (16, 18). Performance status (PS) (0 vs. 1-2) was assessed according to the Eastern Cooperative Oncology Group scale by physicians or nurses in the ward or outpatient clinic (22). Treatment response was assessed by CT using the Response Evaluation Criteria in Solid Tumors version 1.1. Objective response was defined as complete or partial response, and disease control was defined as complete response, partial response, or stable disease maintained for ≥4 weeks. Dose intensity was defined as the delivered dose of each individual agent divided by the planned dose of AG and was calculated based on body surface area measured using the DuBois formula. The relative dose intensity (RDI) was defined as the ratio of the actual dose intensity to the planned dose intensity for a given period, similar to that described in the Hryniuk model (23).

Statistical analyses. Baseline patient characteristics, treatment responses during AG, and clinical courses were compared between the non-MLM and MLM groups using the chi-squared test and the Mann–Whitney U-test. OS was calculated from the start of AG to death from any cause or censoring. Progression-free survival (PFS) was calculated from the start of AG to the first documentation of progressive disease according to the Response Evaluation Criteria in Solid Tumors version 1.1, or discontinuation of AG because of intolerance of adverse events. OS and PFS were analyzed using the Kaplan–Meier method and compared using the log-rank test. Prognostic factors with p-values of <0.05 in the univariate analysis, as well as clinically important factors, were entered into a multivariate Cox proportional hazards model to identify independent prognostic factors. A p-value of <0.05 was considered statistically significant. Statistical analyses were performed using JMP version 13.0 (SAS Institute, Cary, NC, USA).

Evaluation of number of liver metastases. Among the 103 patients, 52 (50%) were categorized into the NLM group, 23 (22%) into the SLM group, and 28 (27%) into the MLM group based on contrast-enhanced CT findings.

OS according to number of liver metastases. Figure 1 shows the OS curves according to the number of liver metastases. The median OS was 12.0 months [95% confidence interval (CI)=8.7-16.0] in the NLM group. The median OS was 9.9 months (95%CI=5.7-11.5) in the SLM group, which was significantly shorter than that in the NLM group (p=0.012). The median OS was 4.6 months (95%CI=3.5-6.7) in the MLM group, which was also significantly shorter than that in the NLM group (p<0.001).

Patient characteristics between non-MLM and MLM groups. Patient characteristics in the non-MLM and MLM groups are shown in Table I. Patients with PS of 1-2, bone metastasis, a high neutrophil-to-lymphocyte ratio (NLR) (>5), hypoalbuminemia (<3.3 g/dl), or elevated C-reactive protein (>1.0 mg/dl) were more frequently observed in the MLM group than in the non-MLM group. The median carbohydrate antigen 19-9 (CA 19-9) level was significantly higher in the MLM group than in the non-MLM group. No other statistically significant differences were observed between the two groups.

Treatment response and clinical course between non-MLM and MLM groups. Treatment response and clinical course in the non-MLM and MLM groups are presented in Table II. A ≥50% decrease in CA 19-9 within the first eight weeks was observed significantly less frequently in the MLM group than in the non-MLM group. The median PFS in the MLM group was significantly shorter than that in the non-MLM group (2.8 vs. 4.5 months, p=0.003). No other statistically significant differences were observed between the two groups.

Subsequent anticancer treatment after AG. The initiation of subsequent anticancer treatment after AG was determined based on the physician’s evaluation of the patient’s general condition, the patient’s willingness to receive further treatment, and the results of comprehensive genomic profiling tests, when performed. At the data cutoff date, one patient in the non-MLM group continued AG. Among the remaining 102 patients, 58 (57%) received subsequent anticancer treatment. These included S-1 (an oral fluoropyrimidine derivative) in 31 patients (29%) (22 in the non-MLM group and nine in the MLM group), modified FOLFIRINOX in 14 patients (14%) (13 in the non-MLM group and one in the MLM group), surgery in three patients, GEM plus S-1 in two patients, liposomal irinotecan plus fluorouracil/leucovorin in two patients, pembrolizumab in one patient, GEM monotherapy in one patient, GEM combined with carbon ion beam therapy in one patient, carbon ion beam therapy alone in one patient, radiotherapy with concomitant AG in one patient, and radiotherapy alone in one patient. Forty-four patients (43%) did not receive any anticancer treatment after AG.

Univariate and multivariate analyses of OS. Univariate and multivariate analyses were performed to identify prognostic factors for OS (Table III). In patients without MLM, univariate analysis identified PS (0 vs. 1-2), the presence of SLM, and CA 19-9 [≥41 × upper limit of normal (ULN) vs. <41 × ULN U/ml] as significant prognostic factors. Multivariate analysis identified PS (0 vs. 1-2), primary tumor site (head vs. body or tail), and the presence of SLM (yes vs. no) as independent prognostic factors for OS. In all patients, univariate analysis identified PS (0 vs. 1-2), the presence of MLM, and CA 19-9 (≥41 × ULN vs. <41 × ULN U/ml) as significant prognostic factors. Multivariate analysis identified PS (0 vs. 1-2), primary tumor site (head vs. body or tail), the presence of MLM (yes vs. no), and CA 19-9 (≥41 × ULN vs. <41 × ULN U/ml) as independent prognostic factors for OS.

To clarify the prognostic significance of the number of liver metastases in patients with APC who received 1L AG, we retrospectively evaluated OS curves, patient characteristics, treatment responses during AG, and clinical courses. OS was stratified according to the number of liver metastases (0 vs. 1-9 vs. ≥10). Because of a poorer baseline general condition and greater tumor burden, patients with MLM had significantly shorter OS and PFS with AG, as well as a lower frequency of CA 19-9 reduction (>50%) within the first eight weeks, compared with patients without MLM. There was no statistically significant difference in the RDI of AG during the first eight weeks between the two groups. In addition, patients with MLM were less likely to achieve disease control with AG than those without MLM, although this difference did not reach statistical significance (p=0.056).

In our previous study with a small sample size, there was no statistically significant difference in median OS between the NLM and SLM groups (11.0 vs. 11.2 months, p=0.41), whereas a significant difference was observed between the NLM and MLM groups (11.0 vs. 6.4 months, p=0.010) (11). In the present updated analysis with a larger sample size, a statistically significant difference in median OS was observed between the NLM and SLM groups (12.0 vs. 9.9 months, p=0.012), as well as between the NLM and MLM groups (12.0 vs. 4.6 months, p<0.001). In the multivariate analysis, the presence of SLM or MLM was identified as an independent prognostic factor for OS. Patients with MLM more frequently exhibited elevated NLR and CRP levels, hypoalbuminemia, and poorer PS, suggesting the presence of a hyperinflammatory and catabolic systemic state. These findings raise the possibility that MLM is associated with enhanced inflammatory signaling pathways, potentially accelerating cancer-related cachexia and limiting treatment continuity. Compared with peritoneal dissemination–dominant or locally advanced disease, liver-dominant multiple metastases may represent a hematogenous-dominant phenotype that reflects intrinsically aggressive tumor biology, rather than simply increased tumor burden, particularly in light of the extremely poor prognosis observed in patients with MLM. Therefore, the number of liver metastases may serve as a surrogate marker of aggressive tumor biology rather than merely reflecting disease volume. These observations warrant further translational investigations to elucidate the tumor-intrinsic mechanisms underlying a hematogenous dissemination phenotype that drives MLM-dominant progression.

The present updated study newly demonstrates the prognostic stratification according to the number of liver metastases (0 vs. 1-9 vs. ≥10). Such stratification may provide a simple and practical measure to inform treatment strategies, facilitate shared decision making with patients and their families, and support the design of clinical research. First, given the extremely poor prognosis of patients with MLM (median OS, 4.6 months), rapid initiation of chemotherapy without delay, together with proactive supportive care, appears essential. Interestingly, during AG treatment, 11 of 28 patients (39%) with MLM showed an increase in skeletal muscle mass. This finding may reflect attenuation of systemic inflammation and cancer-related wasting through disease control of MLM. In addition to proactive supportive care, the catabolic state may be partially reversible. We previously reported that an increase in skeletal muscle mass during first-line AG was associated with a higher prevalence of 2L chemotherapy (18). Induction of 2L chemotherapy after AG is important for improving OS (11, 12, 24-26). Therefore, timely transition to 2L therapy in patients with MLM may further contribute to improved survival. Second, considering the dismal prognosis of patients with MLM, earlier initiation of advance care planning may be warranted. Prognostic information based on the number of liver metastases may assist in shared decision making with patients and their families, helping them to use their limited remaining time meaningfully. Third, because patients with MLM account for approximately one-third of the population and have an extremely poor prognosis, this factor should be considered as a stratification variable in randomized controlled trials. In general, patients enrolled in clinical trials tend to have better PS and are expected to have a more favorable prognosis. Thus, patients with MLM may be underrepresented in clinical trial settings. However, imbalance in the proportion of patients with MLM, given their dismal prognosis and shorter PFS, may distort the apparent efficacy of investigational agents. To more accurately evaluate the true effect of a test drug, stratification or pre-specified subgroup analyses focusing on MLM should be considered.

Study limitations. It is a retrospective study conducted at a single institution; therefore, selection bias cannot be excluded, and the statistical power is limited because of the relatively small sample size. In addition, there is no universally accepted definition of MLM, although we categorized patients into NLM, SLM, and MLM groups (0 vs. 1-9 vs. ≥10) based on clinical practicality, usability, and patient distribution. The cutoff value of ≥10 was selected on the basis of clinical feasibility and previous exploratory findings; however, further studies with larger datasets are needed to examine alternative cutoff values. Moreover, future investigations incorporating genomic and transcriptomic profiling will be necessary to determine whether MLM-dominant disease harbors distinct molecular characteristics, such as enrichment of proliferation-related pathways, inflammatory signaling cascades, or altered stromal composition, compared with peritoneal dissemination–dominant or locally advanced disease. A better understanding of these biological differences may contribute to the development of phenotype-adapted therapeutic strategies and more precise clinical trial design.

In conclusion, this updated analysis demonstrated that the number of liver metastases provides prognostic stratification in patients with APC receiving AG and may serve as a useful parameter for treatment decision making and for stratification in future clinical trials.

The Authors declare no conflicts of interest in relation to this study.

Conceptualization: Kiyotsugu Iede. Data curation: Kiyotsugu Iede. Formal analysis: Kiyotsugu Iede, Terumasa Yamada. Funding acquisition: Kiyotsugu Iede, Terumasa Yamada. Investigation: Kiyotsugu Iede. Methodology: Kiyotsugu Iede. Project administration: Kiyotsugu Iede. Resources: Kiyotsugu Iede, Terumasa Yamada, Hirotoshi Takayama, Masaru Sasaki, Tomo Ishida, Masafumi Yamashita, Yukako Mokutani, Tsukasa Tanida, Jin Matsuyama, Ken Nakata. Software: Kiyotsugu Iede. Supervision: Shusei Tominaga, Terumasa Yamada. Validation: Terumasa Yamada.

The Authors would like to thank Angela Morben, DVM, ELS, from Edanz (https://jp.edanz.com/ac), for editing a draft of this manuscript.

None.

No artificial intelligence (AI) tools, including large language models or machine learning software, were used in the preparation, analysis, or presentation of this manuscript.