Efficacy and safety of immune checkpoint inhibitors plus chemotherapy in esophageal cancer patients with liver metastases

Abstract

BACKGROUND

The liver represents a common site of distant metastasis in patients with esophageal cancer (EC). Conventional chemotherapy (CMT) presents limited efficacy for EC, and EC patients with liver metastases typically experience a poor prognosis, highlighting an urgent need to explore novel treatment approaches. This study evaluated the overall efficacy and safety of CMT vs CMT combined with immune checkpoint inhibitors (ICIs) in the treatment of EC patients with liver metastases. Furthermore, prognostic factors influencing outcomes in this patient population were identified.

AIM

To evaluate the efficacy and safety of first-line chemoimmunotherapy for EC patients with liver metastases and to analyze prognostic factors.

METHODS

This retrospective study included 126 EC patients with liver metastases at Zhejiang Cancer Hospital between 2014 and 2024. Patients receiving CMT were compared with those receiving CMT + ICI. Analyzed variables included clinicopathological features, treatment history, characteristics of metastasis, systemic and local treatments, overall survival (OS), and treatment-related adverse events (TRAEs). Prognostic factors were evaluated using univariate and multivariate Cox proportional-hazards regression models. Finally, efficacy outcomes and TRAE profiles were compared between the two groups.

RESULTS

A significant difference in median OS was identified between the two groups (10.8 months in the CMT group vs 20.8 months in the CMT + ICI group, P = 0.004). The CMT + ICI group also demonstrated a significantly longer median progression-free survival of 11.7 months (P < 0.001). Patients receiving combination therapy exhibited significantly improved systemic objective response rate and disease control rate. Multivariate analysis identified key factors significantly influencing OS in EC patients with liver metastases: Karnofsky Performance Status score ≥ 70, receipt of local therapy for liver metastases, and the number of cycles of CMT and immunotherapy received. Furthermore, the incidence of TRAEs did not significantly differ between the CMT + ICI and CMT groups.

CONCLUSION

For EC patients with liver metastases, the combination of CMT and ICIs demonstrates significantly superior efficacy compared with CMT alone, while maintaining manageable TRAEs.

Key Words: Esophageal cancer; Liver metastasis; Chemotherapy; Immunotherapy; Local liver-directed therapy

Core Tip: Although chemoimmunotherapy is the standard first-line treatment for metastatic esophageal cancer, its efficacy in the subgroup with liver metastases remains poorly characterized. This study demonstrated that chemoimmunotherapy significantly improved overall survival in this population, with a manageable safety profile. Key prognostic factors were identified, and the addition of local therapy to liver metastases may provide a rational approach to further optimize treatment outcomes.

INTRODUCTION

Esophageal cancer (EC) is the seventh leading cause of cancer-related mortality globally[1]. It is histologically classified into two main types, including esophageal squamous cell carcinoma (ESCC) and esophageal adenocarcinoma. ESCC is the predominant form in China[2,3]. Recent statistics indicated that in 2022, 224012 new EC cases and 187467 EC-associated deaths were recorded in China, accounting for 43.8% and 42.1% of global incidence and mortality, respectively, representing the highest burden worldwide[4]. Approximately 50% of patients present with distant metastasis at initial diagnosis, and over one-third develop metastatic disease following surgery or radiotherapy. Survival outcomes are significantly reduced in advanced EC patients, particularly among those with distant metastases[5], where median survival plummets to 5 months post-metastasis[6]. Distant organ metastasis remains the primary cause of treatment failure and death in EC, severely compromising patients’ survival and quality of life. The liver is one of the most common sites of distant metastasis[7]. Post-esophagectomy hepatic recurrence has been previously reported to be correlated with markedly worse survival outcomes (P = 0.032)[8]. Furthermore, liver metastases have been found to confer a poorer prognosis than metastases to other organs. The paucity of research in this area is largely attributable to the subpopulation's clinically insidious nature and dismal outcomes[9].

Patients with EC liver metastases (ECLM) experience a poor prognosis, and there is no consensus on the most effective management strategies for the disease. Prior to the advent of immunotherapy, palliative systemic chemotherapy (CMT) constituted the prevailing standard of care, yielding response rates of 13%-35.9%, disease control rates (DCRs) of 57%-63.9%, a median progression-free survival (PFS) of 3.6 months, and a median overall survival (OS) ranging from 5.5-6.7 months[10]. The advent of immune checkpoint inhibitors (ICIs) has transformed the therapeutic paradigm for advanced EC. Pivotal trials, including KEYNOTE-590[11], CHECKMATE-649[12], and ESCORT-1^st[13], demonstrated substantial survival benefits with immunotherapy, prompting its shift from later lines to first-line treatment and revolutionizing the CMT-dominated framework. Consequently, chemoimmunotherapy has become the standard first-line regimen for metastatic EC. However, few studies have specifically reported survival outcomes for the liver metastasis subgroup, and the efficacy of chemoimmunotherapy for ECLM patients has still remained elusive.

The ASTRUM-007 trial, a multicenter, double-blind, randomized phase 3 study, revealed that serplulimab in combination with CMT significantly improved PFS and OS compared with the combination of placebo and CMT for patients with locally advanced or metastatic ESCC. In the liver metastasis subgroup, the incorporation of immunotherapy resulted in a significant enhancement of PFS. Although no statistical significance was achieved for OS, a trend indicating a prolonged median OS (mOS) was noted. It is noteworthy that in the liver metastasis subgroup (combined positive score ≥ 10), immunotherapy demonstrated more remarkable improvements in terms of both median PFS (mPFS) and mOS compared with placebo[14]. These findings demonstrate that chemoimmunotherapy may be a viable option for ECLM patients. Nevertheless, due to the paucity of studies, the recommendation based on subgroup analyses of clinical studies remains controversial and requires further research. The objective of this study is twofold: (1) To conduct a comparative analysis of the efficacy and adverse event profiles between CMT alone and chemoimmunotherapy in patients with ECLM; and (2) To explore and identify potential prognostic factors in this specific cohort.

MATERIALS AND METHODS

Patients’ selection

This study retrospectively analyzed the clinical data of EC patients with liver metastasis who were admitted to the Zhejiang Cancer Hospital (Hangzhou, China) between March 2014 and March 2024. The inclusion criteria were as follows: (1) Clinically diagnosed with liver metastasis originating from EC, including EC patients with liver metastasis at the time of initial diagnosis, as well as those who developed liver metastasis after undergoing radical treatment for EC; (2) Confirmation of measurable liver lesions by liver ultrasound, computed tomography (CT), or magnetic resonance imaging (MRI); (3) The Karnofsky Performance Status (KPS) score was ≥ 60; and (4) The availability of complete medical records. The exclusion criteria were as follows: (1) Co-existence of other pathological types of malignant tumors at the initial diagnosis; (2) Co-existence of active or uncontrolled severe infections; (3) Patients who discontinued treatment following the diagnosis of liver metastasis from EC; and (4) Missing clinical data.

Data collection

This study collected a large amount of data, such as the demographic characteristics, clinical data, and treatment responses of patients. More specifically, the data included basic information, such as age, gender, history of underlying diseases, histological type of EC, primary tumor location, and tumor stage in the TNM staging system. The preliminary assessment included several factors, such as treatment history prior to the development of liver metastasis, the KPS score at the time of liver metastasis diagnosis, the Child-Pugh score at diagnosis, the number of liver metastases, the involved liver lobes, and the status of distant extra-hepatic metastases (e.g., lung, brain, and bone metastases). Additionally, data were collected on the CMT and chemoimmunotherapy regimens administered as first-line treatment after the diagnosis of liver metastasis, including the number of cycles for each regimen, the use of local liver treatments, and the specific methods employed for these interventions. Documented treatment-related adverse events (TRAEs) included hematological toxicity, hepatotoxicity, nephrotoxicity, thyroid dysfunction, and gastrointestinal adverse reactions. It is noteworthy that missing data and loss to follow-up are inevitable challenges in clinical research. In the present study, missing data were minimal (< 2% for all variables) and assumed to be missing completely at random. Thus, they were addressed using the listwise deletion method. To minimize loss to follow-up, a comprehensive hospital electronic medical record system was utilized, linking inpatient and outpatient data, and telephone interviews were conducted at pre-specified intervals to determine survival outcomes. Ultimately, only 3 patients were lost to follow-up (< 3%). Given this negligible rate, the findings remain reliable and valid, and no sensitivity analysis was therefore carried out.

The study was conducted per the Declaration of Helsinki (2013 edition), it was approved by the Medical Ethics Committee of Zhejiang Cancer Hospital [Approval No. IRB-2025-980(IIT)], and the necessity of consent for this retrospective analysis was waived.

Treatment modalities

The study included two groups: The CMT group (n = 39) and the CMT + ICI group (n = 39). Patients in the CMT group received conventional CMT regimens, while those in the CMT + ICI group were treated with CMT plus immunotherapy that adhered to the current NCCN guidelines for the first-line treatment of metastatic EC. The CMT regimens, administered in 21-day cycles, primarily consisted of fluoropyrimidine-platinum doublets (most commonly 5-FU + cisplatin) and taxane-platinum doublets (most commonly nab-paclitaxel + cisplatin). Patients in the CMT + ICI group received ICIs alongside these CMT backbones, with sintilimab, pembrolizumab, camrelizumab, tislelizumab, and other ICIs as the primary agents administered. Throughout the treatment course, adjustments were made to the types and dosages of drugs for some patients based on their clinical conditions and tolerance.

Study endpoints

The primary endpoint was OS, which was defined as the interval from the diagnosis of liver metastasis to death from any cause or the last confirmed follow-up. The secondary endpoint was PFS, which was calculated from the initiation of treatment to documented disease progression, death from any cause, or the most recent confirmed follow-up. The study data were finalized on May 15, 2025, establishing the cut-off date for analysis.

Efficacy and toxicity evaluation criteria

The tumor response evaluation was conducted in accordance with the Response Evaluation Criteria in Solid Tumors (RECIST; version 1.1). Routine assessments were performed at regular intervals throughout the active treatment phase, and the interval between assessments ranged from 6 weeks to 8 weeks. Following the end of treatment, the assessments were conducted at 3- to 6-month intervals. According to the RECIST (version 1.1), tumor response evaluation is a standardized process used to assess the objective efficacy of cancer treatments in patients with solid tumors. The primary objective is to consistently and objectively quantify changes in tumor burden. A baseline assessment should be conducted prior to treatment, involving identifying target lesions. Up to five representative lesions (no more than two per organ) should be selected as “target lesions”, ensuring that they are measurable. Typically, CT or MRI scans are used to measure the longest diameter for non-lymph node lesions or the short axis for lymph node lesions. For non-lymph node target lesions, the minimum measurable size is ≥ 10 mm, while for lymph node target lesions, the minimum measurable size is ≥ 15 mm (short axis). The baseline total is calculated by summing the longest diameter (for non-lymph node lesions) or short axis (for lymph node lesions) of all target lesions. Non-target lesions are recorded based on their presence or absence, including small lesions, unmeasurable lesions, malignant effusions, etc. Patients undergo imaging follow-ups after each predefined treatment cycle, typically using the same imaging modality as the baseline. Follow-up evaluations include re-measuring all target lesions identified at baseline, assessing the overall status of all non-target lesions (e.g., presence, disappearance, definite progression), and recording the appearance of any new lesions. Tumor response is categorized into four classifications based on changes in the sum of target lesions, the status of non-target lesions, and the emergence of new lesions: (1) Complete response (CR): All target lesions have disappeared, and all lymph node target lesions have a short axis of < 10 mm; (2) Partial response (PR): The total diameter of target lesions has decreased by ≥ 30% compared with baseline; (3) Stable disease (SD): Neither PR nor PD criteria are met (i.e., reduction < 30%, or increase < 20% with no absolute increase of ≥ 5 mm); and (4) Progressive disease (PD): The total diameter of target lesions has increased by ≥ 20% from the smallest total diameter observed during the study, with an absolute increase of ≥ 5 mm; new lesions have appeared; or non-target lesions have shown clear progression. In this retrospective study, following treatment, patients' attending physicians routinely performed the necessary clinical assessments and recorded the findings. In cases of disease progression, treatment regimens were modified, or additional therapeutic measures were introduced. The objective response rate (ORR) was defined as the proportion of patients who achieved CR or PR at least once during follow-up. The DCR was defined as the proportion of patients exhibiting CR, PR, or SD at least once during follow-up.

Moreover, TRAEs were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (version 5.0). The attending physician was responsible for systematically documenting and grading any TRAEs recorded during the patient’s hospitalization. In contrast, TRAEs reported after discharge were primarily collected through patient self-reports and telephone follow-ups.

Statistical analysis

Intergroup comparisons of categorical variables, including baseline characteristics, treatment features, therapeutic responses, and TRAEs, were performed using the χ² test, continuity-corrected χ² test, or the Fisher's exact test, as appropriate. Kaplan-Meier survival curves were generated to visualize and compare OS and PFS across groups, and between-group differences were assessed using the log-rank test. Univariate survival analysis was conducted via the Cox proportional-hazards model to identify prognostic factors for OS. Clinically relevant variables demonstrating statistical significance (P < 0.05) in the univariate analysis were subsequently imported into a multivariate Cox regression model. The statistical analysis was performed using SPSS 29.0 software (IBM, Armonk, NY, United States). P < 0.05 was considered statistically significant.

RESULTS

Patients’ demographic and clinical data

From January 2014 to March 2024, a total of 423 patients were diagnosed with ECLM at Zhejiang Cancer Hospital. Following the application of rigorous selection criteria, 126 patients were regarded eligible for inclusion in the study. Patients’ exclusion process is illustrated in Figure 1. The selected patients were divided into two groups: Patients receiving CMT alone (n = 87) and those receiving CMT + ICI (n = 39). The median follow-up was 74.7 months. The baseline characteristics of the study participants are outlined in Table 1. The median age of the cohort was 60.0 years, and male patients accounted for over 94% of the cohort. It was noted that nearly 40% of the cohort had received radiotherapy for the primary tumor, presented with distant lymph node metastasis, or undergone CMT prior to metastasis. At the time of initial diagnosis, synchronous liver metastases were present in 50% of cases. A substantial proportion of patients had more than three liver metastases, and a notable percentage exhibited bilateral hepatic involvement. At diagnosis, the majority of patients displayed characteristics, such as squamous cell carcinoma histology, T3-T4 primary tumors, regional lymph node involvement, and dysphagia, mainly accompanied by other cancer-related symptoms. A larger proportion of subjects had middle or lower esophageal primaries, moderate-to-poor differentiation (grade II-III), a KPS score of ≥ 70, Child-Pugh class A-B, and no prior targeted therapy before metastasis. Concurrent extrahepatic metastases occurred in 55 (43.7%) patients, predominantly influencing the lung and bone.

Figure 1 Study flowchart. CMT: Chemotherapy; ICI: Immune checkpoint inhibitor; ORR: Objective response rate; DCR: Disease control rate; OS: Overall survival; PFS: Progression-free survival.

Table 1 Demographic and clinical characteristics of the study cohort, n (%).

Demographic and clinical characteristics	Total number of cases (n = 126)	CMT group (n = 87)	CMT + ICI group (n = 39)	χ² value	P value
Age (years)				0.451	0.502
> 60	59 (46.8)	39 (44.8)	20 (51.3)
≤ 60	67 (53.2)	48 (55.2)	19 (48.7)
Sex				0.079	0.779
Male	119 (94.4)	83 (95.4)	36 (92.3)
Female	7 (5.6)	4 (4.6)	3 (7.7)
Underlying diseases				0.016	0.900
Yes	43 (34.1)	30 (34.5)	13 (33.3)
No	83 (65.9)	57 (65.5)	26 (66.7)
Dysphagia or cancer-related symptoms				0.182	0.669
Yes	106 (84.1)	74 (85.1)	32 (682.1)
No	20 (15.9)	13 (14.9)	7 (17.9)
KPS score				-	0.094
≥ 70	124 (98.4)	87 (100.0)	37 (94.9)
< 70	2 (1.6)	0 (0.0)	2 (5.1)
Histology				2.098	0.321
ESCC	111 (88.1)	74 (85.1)	37 (94.9)
EAC	3 (2.4)	3 (3.4)	2 (5.1)
Other	12 (9.5)	10 (11.5)	0 (0.0)
Primary tumor location				1.649	0.428
Upper	4 (3.2)	3 (3.4)	1 (2.6)
Middle	51 (40.5)	32 (36.8)	19 (48.7)
Lower	71 (56.3)	52 (59.8)	19 (48.7)
Tumor differentiation grade				1.476	0.452
G1	6 (4.8)	5 (5.7)	1 (2.6)
G2	46 (36.5)	34 (39.1)	12 (30.8)
G3-G4	74 (58.7)	48 (55.2)	26 (66.7)
Tumor stage				0.961	0.327
T1-2	17 (13.5)	10 (11.5)	7 (17.9)
T3-4	109 (86.5)	77 (88.5)	32 (82.1)
Nodal stage				0.007	0.932
N0	15 (11.9)	11 (12.6)	4 (10.3)
N+	111 (88.1)	76 (87.4)	35 (89.7)
Synchronous liver metastasis at initial diagnosis				0.487	0.485
Yes	62 (49.2)	41 (47.1)	21 (53.8)
No	64 (50.8)	46 (52.9)	18 (46.2)
Distant lymph node metastasis				0.054	0.816
Yes	53 (42.1)	36 (41.4)	17 (43.6)
No	73 (57.9)	51 (58.6)	22 (56.4)
Primary lesion surgery				0.184	0.668
Yes	52 (41.3)	37 (42.5)	15 (38.5)
No	74 (58.7)	50 (57.5)	24 (61.5)
Primary lesion radiotherapy				0.227	0.634
Yes	51 (40.5)	34 (39.1)	17 (43.6)
No	75 (59.5)	53 (60.9)	22 (56.4)
Pre-hepatic metastasis chemotherapy				0.483	0.487
Yes	41 (32.5)	30 (34.5)	11 (28.2)
No	85 (67.5)	57 (65.5)	28 (71.8)
Pre-hepatic metastasis targeted therapy				-	0.552
Yes	3 (2.4)	84 (96.6)	0 (0.0)
No	123 (97.6)	3 (3.4)	39 (100.0)
Child-Pugh score				-	1.000
> 6	2 (1.6)	2 (97.7)	0 (0.0)
≤ 6	124 (98.4)	85 (2.3)	39 (100.0)
Oligometastatic liver disease				0.491	0.483
Yes	24 (19)	18 (20.7)	6 (15.4)
No	102 (81)	69 (79.3)	33 (84.6)
Number of liver metastases				1.680	0.195
> 3	80 (63.5)	52 (59.8)	28 (71.8)
≤ 3	46 (36.5)	35 (40.2)	11 (28.2)
Hepatic lobe involvement				5.638	0.060
Left	8 (6.3)	4 (4.6)	4 (10.3)
Right	40 (31.7)	33 (37.9)	7 (17.9)
Bilateral	78 (61.9)	50 (57.5)	28 (71.8)
Extrahepatic metastasis				3.739	0.060
Yes	55 (43.7)	33 (37.9)	22 (56.4)
No	71 (56.3)	54 (62.1)	17 (43.6)
Lung metastasis				2.321	0.128
Yes	60 (32.1)	18 (20.7)	13 (33.3)
No	127 (67.9)	69 (79.3)	26 (66.7)
Bone metastasis				0.182	0.669
Yes	20 (15.9)	13 (14.9)	7 (17.9)
No	106 (84.1)	74 (85.1)	32 (82.1)
Brain metastasis				-	1.000
Yes	3 (2.4)	2 (2.3)	1 (2.6)
No	123 (97.6)	85 (97.7)	38 (97.4)

Statistical significance was set at P < 0.05. For certain variables, the Fisher’s exact test was applied in the absence of χ² values. G1: Well-differentiated; G2: Moderately differentiated; G3: Poorly differentiated; G4: Undifferentiated; T: Tumor; N: Node; CMT: Chemotherapy; ICI: Immune checkpoint inhibitor; KPS: Karnofsky Performance Status; ESCC: Esophageal squamous cell carcinoma; EAC: Esophageal adenocarcinoma.

Treatment regimens and details

Treatment details are summarized in Table 2. In the CMT group, taxane-platinum regimens were administered to 49 (56.3%) patients, fluoropyrimidine-platinum to 23 (26.4%), and other regimens to 20 (15.9%). The CMT + ICI group received taxane-platinum (n = 26, 66.7%), fluoropyrimidine-platinum (n = 8, 20.5%), or other regimens (n = 5, 12.8%). A total of 105 (83.3%) patients continued CMT for ≥ 4 cycles, whereas 38 patients in the CMT + ICI group (97.4%) completed ≥ 4 cycles of immunotherapy. Totally, 39 (31.0%) cases in the two groups underwent local treatment of liver metastases, predominantly ablation (n = 24, 19.0%).

Table 2 Treatment-related characteristics of participants, n (%).

Treatment-related characteristics	Total number of cases (n = 126)	CMT group (n = 87)	CMT + ICI group (n = 39)	χ² value	P value
Chemotherapy regimen				0.852	0.653
Taxane-based ± platinum	75 (59.5)	49 (56.3)	26 (66.7)
Fluoropyrimidine ± platinum	31 (24.6)	23 (26.4)	8 (20.5)
Other	20 (15.9)	15 (17.3)	5 (12.8)
Chemotherapy cycle				0.081	0.776
≥ 4	105 (83.3)	72 (82.8)	33 (84.6)
< 4	21 (16.7)	15 (17.2)	6 (15.4)
Immunotherapy cycle				120.395	< 0.001a
≥ 4	38 (30.2)	0 (0.0)	38 (97.4)
< 4	88 (69.8)	87 (100.0)	1 (2.6)
Liver local treatment				0.746	0.388
Yes	39 (31.0)	29 (33.3)	10 (25.6)
No	87 (69.0)	58 (66.7)	29 (74.4)
Local treatment modalities				2.626	0.662
None	87 (69.0)	58 (66.7)	29 (74.4)
Ablation	24 (19.0)	18 (20.7)	6 (15.4)
Surgery	4 (3.2)	3 (3.4)	1 (2.6)
TACE	4 (3.2)	4 (4.6)	0 (0.0)
Radiotherapy	7 (5.6)	4 (4.6)	3 (7.7)

^aP < 0.05, statistical significance between the two groups.

CMT: Chemotherapy; ICI: Immune checkpoint inhibitor; TACE: Transarterial chemoembolization.

Treatment outcomes

The mOS in the CMT and the CMT + ICI groups was 10.8 and 20.8 months, respectively (P = 0.004). The Kaplan-Meier survival curve is illustrated in Figure 2A. The mPFS in the CMT and the CMT + ICI groups was 5.5 months and 11.7 months, respectively (P < 0.001). Compared with the CMT group, the CMT + ICI group also had a significant survival advantage, and the survival curve is shown in Figure 2B.

Figure 2 Kaplan-Meier survival curves of the chemotherapy group and the chemotherapy + immune checkpoint inhibitor group. A: Overall survival; B: Progression-free survival. CMT: Chemotherapy; ICI: Immune checkpoint inhibitor; OS: Overall survival; PFS: Progression-free survival.

The overall ORR was 69.8%, involving 58.6% in the CMT group and 94.9% in the CMT + ICI group (P < 0.001). The overall DCR was 82.5%, including 74.7% in the CMT group and 100.0% in the CMT + ICI group (P < 0.001). Additional details are presented in Table 3.

Table 3 Comparison of objective response rate and disease control rate between the two groups, n (%).

Healing effect	Total number of cases (n = 126)	CMT group (n = 87)	CMT + ICI group (n = 39)	P value
ORR	88 (69.8)	51 (58.6)	37 (94.9)	< 0.001
DCR	104 (82.5)	65 (74.7)	39 (100.0)	< 0.001

P < 0.05 was indicative of statistical significance between the two groups. CMT: Chemotherapy; ICI: Immune checkpoint inhibitor; ORR: Objective response rate; DCR: Disease control rate.

TRAEs

Table 4 presents TRAEs occurring in ≥ 10% of patients after systemic therapy for liver metastases. The overall incidence of TRAEs was 92.1% in the pooled groups. Hematological toxicities occurred in 87.4% of patients in the CMT group vs 92.3% in the CMT + ICI group (P = 0.60). Anemia developed in over 80% of patients with liver metastases during first-line treatment in both groups. The incidence rates of gastrointestinal TRAEs, such as nausea and vomiting, were 40.2% and 20.5% in the CMT and CMT + ICI groups, respectively (P = 0.03). No significant differences were identified between the two groups for other TRAEs, including fatigue and abnormal liver function. Grade 4 TRAEs occurred in 9 patients, and no grade 5 TRAEs were reported. Among patients receiving immunotherapy, immune-related adverse events included thyroid dysfunction in 3 patients, renal dysfunction in 1 patient, and a skin adverse reaction in 1 patient.

Table 4 Comparison of treatment-related adverse events between the two groups, n (%).

TRAEs	Total number of cases (n = 126)	CMT group (n = 87)	CMT + ICI group (n = 39)	χ² value	P value
Any TRAEs	116 (92.1)	80 (92.0)	36 (92.3)	0.000	1.000
TRAEs ≥ 10%
Hematologic toxicity	112 (88.9)	76 (87.4)	36 (92.3)	0.261	0.609
Anemic	104 (82.5)	71 (81.6)	33 (84.6)	0.169	0.681
Leucopenia	71 (56.3)	49 (56.3)	22 (56.4)	0.000	0.993
Neutropenia	57 (45.2)	40 (46.0)	17 (43.6)	0.062	0.803
Thrombocytopenia	52 (41.3)	33 (37.9)	19 (48.7)	1.293	0.256
Hypoalbuminemia	71 (56.3)	50 (57.5)	21 (53.8)	0.144	0.704
Liver dysfunction	40 (31.7)	32 (36.8)	8 (20.5)	3.289	0.070
Increased γ-GT	15 (11.9)	13 (14.9)	2 (5.1)	1.626	0.202
Increased AST	21 (16.7)	16 (18.4)	5 (12.8)	0.602	0.438
Nausea and vomiting	43 (34.1)	35 (40.2)	8 (20.5)	4.657	0.031a
Fatigue	36 (28.6)	25 (28.7)	11 (28.2)	0.004	0.951
≥ 3 TRAEs	41 (32.5)	32 (36.8)	9 (23.1)	2.304	0.129
Hematologic toxicity	37 (29.4)	28 (32.2)	9 (23.1)	1.077	0.299
Neutropenia	17 (13.5)	14 (16.1)	3 (7.7)	1.628	0.202

^aP < 0.05, statistical significance in pairwise multiple comparisons between two groups.

CMT: Chemotherapy; ICI: Immune checkpoint inhibitor; TRAEs: Treatment-related adverse events; γ-GT: Gamma-glutamyl transferase; AST: Aspartate aminotransferase.

Prognostic factors influencing OS

This study evaluated the impact of significant clinical factors on patients’ OS. Firstly, univariate Cox regression analysis was conducted to identify potential prognostic factors for OS, as detailed in Table 5. The univariate analysis demonstrated that the following factors were potentially associated with OS in ECLM patients: KPS score ≥ 70, distant lymph node involvement, prior targeted therapy before liver metastasis, number of liver metastases (≤ 3), oligometastatic liver disease, liver lobe involvement (left-sided lobe or right-sided lobe vs the reference category, bilateral lobes), CMT regimen (fluoropyrimidine-based ± platinum vs reference category: Taxane-based ± platinum, other regimens vs reference category: Taxane-based ± platinum), CMT cycles (≥ 4), immunotherapy, immunotherapy cycles (≥ 4); receipt of local liver therapy, and local treatment modality (ablation vs reference category: No local therapy).

Table 5 Results of univariate Cox regression analysis.

Influential factors (independent variables)	P value	HR	95%CI for HR
			Lower bound	Upper bound
Age (> 60/≤ 60 years)	0.537	1.124	0.775	1.630
Sex (male/female)	0.781	1.124	0.493	2.560
Underlying diseases (yes/no)	0.324	1.220	0.822	1.810
Dysphagia or cancer-related symptoms (yes/no)	0.217	0.737	0.454	1.196
KPS score (≥ 70/< 70)	0.013a	6.117	1.463	25.573
Histology
ESCC, dummy variable	0.164
EAC	0.163	1.563	0.835	2.928
Other	0.168	2.269	0.709	7.267
Tumor location
Upper, dummy variable	0.819
Middle	0.595	1.317	0.478	3.628
Lower	0.798	0.952	0.652	1.390
Tumor differentiation grade
Grade 1, dummy variable	0.298
Grade 2	0.140	2.044	0.791	5.281
Grade 3-4	0.121	2.110	0.821	5.427
Tumor stage (T1-2/T3-4)	0.121	1.539	0.892	2.656
Nodal stage (N0/N+)	0.758	1.092	0.623	1.915
Synchronous liver metastasis at initial diagnosis (yes/no)	0.795	1.050	0.726	1.519
Distant lymph node metastasis (yes/no)	0.036a	1.492	1.027	2.168
Primary lesion surgery (yes/no)	0.080	0.712	0.488	1.041
Primary lesion radiotherapy (yes/no)	0.765	0.944	0.649	1.373
Pre-hepatic metastasis chemotherapy (yes/no)	0.441	0.854	0.571	1.277
Pre-hepatic metastasis targeted therapy (yes/no)	0.009a	4.809	1.477	15.656
Child-Pugh score (> 6/≤ 6)	0.107	3.224	0.778	13.361
Number of liver metastases (> 3/≤ 3)	0.012a	1.650	1.115	2.441
oligometastatic liver disease (yes/no)	0.001a	0.434	0.261	0.725
Hepatic lobe involvement
Bilateral lobes, dummy variable	0.024a
Left lobe	0.070	0.462	0.200	1.066
Right lobe	0.021a	0.616	0.409	0.929
Extrahepatic metastasis (yes/no)	0.334	1.201	0.828	1.743
Lung metastasis (yes/no)	0.614	1.118	0.724	1.728
Bone metastasis (yes/no)	0.299	1.300	0.792	2.135
Brain metastasis (yes/no)	0.161	0.366	0.090	1.491
Chemotherapy regimen
Taxane-based ± platinum, dummy variable	0.012a
Fluoropyrimidine ± platinum	0.627	1.115	0.720	1.727
Other	0.003a	2.231	1.313	3.792
Chemotherapy cycle (≥ 4/< 4)	< 0.001a	0.261	0.160	0.425
Immunotherapy (yes/no)	0.005a	0.551	0.364	0.836
Immunotherapy cycle (≥ 4/< 4)	0.004a	0.538	0.353	0.820
Liver local treatment (yes/no)	0.005a	0.546	0.358	0.833
Local treatment modalities
None, dummy variable	0.047a
Ablation	0.027a	0.565	0.340	0.937
Surgery	0.217	0.528	0.192	1.455
TACE	0.409	1.530	0.557	4.203
Radiotherapy	0.064	0.424	0.171	1.050

^aP < 0.05, statistical significance included in multifactor Cox regression analysis.

HR: Hazard ratio; 95%CI: 95% confidence interval; TACE: Transarterial chemoembolization; KPS: Karnofsky Performance Status; ESCC: Esophageal squamous cell carcinoma; EAC: Esophageal adenocarcinoma.

To minimize multicollinearity, variables with P < 0.05 in univariate analysis were entered into the multivariate Cox regression model after excluding factors demonstrating high correlation. The results of the multivariate Cox regression are detailed in Table 6. The analysis revealed that a KPS score ≥ 70, receipt of ≥ 4 CMT cycles, completion of ≥ 4 immunotherapy cycles, and administration of local liver therapy were significantly associated with improved OS.

Table 6 Results of multivariate Cox regression analysis.

Influential factors (independent variables)	P value	HR	95%CI for HR
			Lower bound	Upper bound
KPS score (≥ 70/< 70)	0.049a	5.452	1.008	29.479
Oligometastatic liver disease (yes/no)	0.156	0.592	0.286	1.222
Distant lymph node metastasis (yes/no)	0.843	1.046	0.670	1.633
Pre-hepatic metastasis targeted therapy (yes/no)	0.378	1.853	0.470	7.310
Hepatic lobe involvement
Bilateral lobes, dummy variable	0.449
Left-sided lobe	0.736	0.858	0.353	2.088
Right-sided lobe	0.206	0.686	0.382	1.231
Chemotherapy regimen
Taxane-based ± platinum, dummy variable	0.187
Fluoropyrimidine ± platinum	0.498	0.849	0.528	1.363
Other	0.123	1.570	0.885	2.785
Chemotherapy cycle (≥ 4/< 4)	< 0.001a	0.139	0.072	0.266
Immunotherapy cycle (≥ 4/< 4)	< 0.001a	0.225	0.132	0.383
Liver local treatment (yes/no)	0.021a	0.549	0.330	0.913

^aP < 0.05, statistical significance.

HR: Hazard ratio; 95%CI: 95% confidence interval; KPS: Karnofsky Performance Status.

DISCUSSION

The CMT + ICI group demonstrated a significant advantage

The present study demonstrated that adding immunotherapy to systemic CMT could improve survival outcomes in EC patients with liver metastases, and systemic adverse events remained within acceptable limits. The combination group (CMT plus immunotherapy) exhibited significantly superior ORR and DCR compared with CMT alone. Notably, pre-specified subgroup analyses were performed to evaluate the influences of different immunotherapy and CMT regimens (including fluoropyrimidine-base ± platinum and taxane-based ± platinum) on clinical outcomes in ECLM patients. The results indicated no statistically significant differences among these regimens. Since the choice of specific regimen did not demonstrate a discernible impact on the primary outcomes, a detailed discussion on this aspect was regarded unnecessary in this study. Multivariate Cox regression analysis identified a KPS score of ≥ 70, local therapy for liver metastases, over 4 CMT cycles, and over 4 immunotherapy cycles as significant independent prognostic factors for survival.

Compared with CMT alone, the combination group exhibited a significantly longer mOS (20.8 months vs 10.8 months, P = 0.004) and PFS (11.7 months vs 5.5 months, P < 0.001). These findings align with those of a separate study involving over 500 EC patients, including 103 with baseline liver metastases. In that study, the subgroup receiving CMT plus serpelimab (n = 71) demonstrated improved PFS compared with the CMT plus placebo subgroup (n = 32; 5.7 months vs 4.3 months; P = 0.03). Although OS did not reach statistical significance in that cohort, a numerical advantage was noted (13.7 months vs 10.3 months)[14]. The results of a previous study are consistent with those of the current study[14].

CMT remains the primary treatment for advanced EC, with mOS ranging from 5.5 months to 6.7 months for platinum/fluoropyrimidine or platinum/taxane doublets[10,15]. Recent clinical trials, such as KEYNOTE-590[11] and ESCORT-1^st[13], provided noticeable evidence supporting the efficacy of immunotherapy in metastatic disease. However, due to the heterogeneity in PD-L1 expression among primary tumors, lymph nodes, metastatic sites and the potential immunosuppressive mechanisms, such as the liver-specific microenvironment regulation[16], immunotherapy has shown limited efficacy in patients with liver metastases. Existing studies on non-small cell lung cancer, endometrial cancer, gastric cancer, and breast cancer have indicated that PD-L1 expression is higher in metastatic tumors than that in primary tumors. In colorectal cancer patients with liver metastases, it has been confirmed that PD-L1 expression is elevated in liver metastases compared with primary tumors. Additionally, the accumulation of myeloid-derived suppressor cells, the proliferation of inhibitory T cells, and increased secretion of inhibitory cytokines (e.g., TGF-β, IL-10, etc.) in the liver metastatic microenvironment can impact the effectiveness of anti-PD-L1 immunotherapy[17]. However, research on ECLM remains relatively scarce, and the mechanisms underlying resistance to immunotherapy have not been thoroughly explored. Consequently, the therapeutic impact on liver metastases warrants further investigation.

Compelling evidence suggests that liver metastases sequester systemic T cells, particularly CD8+ T cells, through interactions with FasL+CD11b+F4/80+ hepatic macrophages, leading to CD8+ T-cell apoptosis[18,19]. This sequestration has the potential to deplete systemic CD8+ T cells, which are critical for immunotherapy response, thereby limiting the efficacy of monotherapy. Mechanistically, fluoropyrimidines and platinum agents may enhance immunotherapy by: (1) Upregulating PD-L1 expression via JAK/STAT signaling pathway[16]; and (2) Promoting platinum-induced dendritic cell recruitment, sensitizing tumors to immune checkpoint blockade[20]. These synergistic mechanisms, supported by our clinical results, highlight that chemoimmunotherapy improves survival in EC patients with liver metastases.

TRAEs and independent prognostic factors

The adverse events associated with chemoimmunotherapy remained within acceptable limits. Hematologic toxicities constituted over 80% of all TRAEs, and no significant difference was identified between the chemoimmunotherapy and CMT-alone groups. Similarly, the majority of adverse events, including abnormal liver function and hypoalbuminemia, exhibited comparable incidence rates. The incidence of grade ≥ 3 adverse events (predominantly hematologic toxicities) also demonstrated no significant intergroup difference. These findings are consistent with those of a previous study[14], which reported hematological toxicity rates of 78.7% and 78.6% in the chemoimmunotherapy group and the CMT group, respectively, and most non-hematological toxicities exhibited comparable frequencies. Collectively, these data indicated that the addition of immunotherapy did not exacerbate treatment-related toxicity. It is noteworthy that the chemoimmunotherapy group demonstrated significantly reduced incidence rates of nausea and vomiting vs the CMT group. This phenomenon may be ascribed to the enhanced reduction in tumor load and the improved patient performance status that resulted from the therapeutic efficacy.

Univariate Cox regression was employed as the primary screening method. All clinically meaningful covariates that achieved statistical significance (defined as P < 0.05) in univariate analysis were involved in the final multivariate Cox proportional hazards model. It is noteworthy that due to collinearity between the variables "immunotherapy (yes/no)" and "immunotherapy cycles (≥ 4/< 4)", which were both independent prognostic factors, only the latter was retained in the final multivariate regression model, as it was considered clinically more relevant. Therapeutic adequacy, defined as the completion of ≥ 4 CMT cycles and ≥ 4 immunotherapy cycles, emerged as a significant prognostic factor. These findings highlight the critical importance of adequate therapeutic exposure. Premature discontinuation due to adverse events or socioeconomic factors may compromise survival outcomes, providing valuable insights for clinical management.

Additional local therapy may improve treatment efficacy

Local liver-directed therapy was identified as an independent prognostic factor. Modalities included radiofrequency ablation (RFA), metastasectomy, transarterial chemoembolization (TACE), and radiotherapy. Due to concerns of collinearity, detailed stratification was excluded from the multivariate analysis. Univariate analysis revealed a significant survival benefit from RFA. A small-scale study (n = 40) of oligometastatic EC patients (≤ 3 liver lesions) treated with fluoropyrimidine + platinum CMT demonstrated superior mOS (13 months vs 7 months, P = 0.011) and PFS (P = 0.03) with microwave ablation compared with CMT alone[21]. Similarly, a single-arm study (n = 16) reported a mOS/PFS of 14.5/7.5 months with ablation plus CMT, in which outcomes exceeded historical CMT benchmarks. These data support our conclusion that ablation, particularly RFA, improves survival in this patient population.

Liver-directed radiotherapy represents a common local treatment modality for EC patients with hepatic metastases. The current first-line systemic therapy for metastatic EC involves CMT combined with immunotherapy. However, the efficacy of this regimen is limited by the liver's capacity for T-lymphocyte siphoning, where hepatic macrophages sequester CD8+ T cells. Notably, radiotherapy may overcome this limitation by eliminating immunosuppressive macrophages and reducing T-cell sequestration, thereby potentially enhancing immunotherapeutic response[18]. In a pilot study of 7 EC patients with liver metastases receiving systemic CMT plus proton beam radiotherapy to hepatic lesions, 5 patients concurrently received PD-1 checkpoint inhibitors. This approach achieved a mOS of 35.5 months and a mPFS of 8.7 months[22]. Previous studies have proposed several mechanisms to explain the synergistic efficacy of combining radiotherapy with immunotherapy. These include: (1) The release of tumor antigens through local radiotherapy, which stimulates immune responses and may elicit the abscopal effect; (2) The induction of tumor cell ferroptosis by immunotherapy, enhancing radiosensitivity; and (3) The remodeling of the hepatic immune microenvironment by liver-targeted radiotherapy, potentially mitigating the sequestration of T lymphocytes in the liver[18,23,24]. These mechanisms provide strong support for the enhanced therapeutic outcomes achieved through the combination of radiotherapy and immunotherapy. While the integration of local hepatic therapy with chemoimmunotherapy holds promise for improving outcomes in this population, the absence of a significant association of radiotherapy with immunotherapy in the present study may be attributable to the limited sample size.

Metastasectomy is considered potentially curative in certain cases. A retrospective cohort study of 34 patients with gastroesophageal cancer and liver or lung metastases reported a mOS of 52 months in the subgroup with esophageal liver metastasis (n = 11) following resection[25]. In contrast, Ichida et al[26] found only a borderline OS improvement (13 months vs 5 months, P = 0.06) in patients with recurrent disease who underwent resection compared with those who did not. These mixed outcomes, being consistent with our findings, demonstrate that the role of metastasectomy remains controversial and highlights the need for refined patient selection criteria.

TACE is frequently utilized for hepatic metastases. In a study of colorectal cancer (n = 98), drug-eluting bead TACE combined with systemic therapy significantly improved mPFS (12.1 months vs 8.4 months, P = 0.008) and DCR (87.0% vs 67.3%, P = 0.022) compared with systemic therapy alone[27]. However, data of TACE for EC patients are limited. In the present study, no significant benefit from TACE was identified, which might be due to sample size limitations, warranting further investigation.

The present study demonstrated that combining local therapy with systemic treatment may improve outcomes in EC patients with liver metastases. Several mechanisms may underlie this synergistic effect, as supported by previous research. Localized treatment may induce the release of tumor antigens, thereby activating systemic immune responses, a phenomenon known as the abscopal effect. Immunotherapy, in turn, may sensitize tumors to local interventions by promoting ferroptosis. Additionally, local therapy has the potential to remodel the immunosuppressive liver microenvironment, which may mitigate the sequestration of T cells in the liver, thereby enhancing their survival and function in this organ[18,23,24]. These findings highlight the potential of an integrated treatment strategy, combining systemic therapy (e.g., CMT or chemoimmunotherapy) with aggressive local control of liver metastases, to enhance patient prognosis. Although this multimodal approach may enhance therapeutic efficacy, several critical challenges remain to be addressed, including the long-term survival benefits, the identification of the optimal patient population, the selection of appropriate local treatment modalities, the timing of intervention, and whether it can ultimately be regarded as a standard treatment regimen. In conclusion, research on EC with liver metastases is evolving from an emphasis on phenotypic evaluation toward a concentration on mechanistic elucidation and the implementation of precision-based interventions. Future studies should prioritize the elucidation of the molecular mechanisms underlying liver metastasis, the dynamic characterization of the liver immune microenvironment, and the development of innovative interdisciplinary treatment strategies. These approaches may involve the combination of systemic therapies with local interventions, the integration of immunotherapy with anti-angiogenic agents, and the investigation of combined traditional Chinese medicine and Western therapeutic modalities. Addressing these pivotal challenges will necessitate the design and execution of rigorously structured phase III clinical trials.

Limitations

The present study has several limitations. As a single-center, retrospective study with a relatively small proportion of EC patients with liver metastases, the limited sample size might introduce statistical bias, potentially impacting the generalizability and reliability of the findings. Additionally, although the groups were balanced, the proportion of male patients and those with ESCC was relatively high. Given that gender and primary tumor pathology may influence treatment response and prognosis, this could introduce potential bias. Furthermore, some patients who were initially diagnosed without distant organ metastasis underwent radical EC treatment before developing liver metastases. These patients might receive different treatment strategies during this period, potentially introducing confounding factors that could affect the comparison between treatment groups. With the advent of the immuno-oncology era, biomarker evaluation has become a notable element in predicting treatment response and guiding clinical decision-making, a concept widely accepted in the academic community. Studies on liver metastases have also indicated that resistance to immunotherapy may be linked to the expression levels of specific biomarkers. During the design phase of this study, the inclusion of biomarker testing was carefully considered. However, due to the extended timeline of the research, routine biomarker assessments, such as PD-L1 testing, had not yet become standard clinical practice at the outset of immunotherapy. This represents a limitation inherent to the study. It is acknowledged that future research should systematically incorporate key biomarkers into the core analytical framework. The next phase of our research will therefore include an expanded sample size and more rigorous prospective studies to obtain more precise results.

CONCLUSION

In conclusion, the present study examined and compared the efficacy of CMT + ICI as a first-line treatment for EC patients with liver metastases vs CMT alone. The results demonstrated that CMT + ICI was associated with superior efficacy and manageable adverse effects. Multivariate Cox regression analysis identified a KPS score of ≥ 70, local treatment for liver metastases, ≥ 4 cycles of CMT, and ≥ 4 cycles of immunotherapy as independent prognostic factors in these patients. It is noteworthy that the optimal treatment strategy for EC patients with liver metastases remains undefined and warrants further validation through large-scale prospective clinical trials. Given the ongoing advancements in this field, the objective is to refine therapeutic strategies to enhance both the prognosis and quality of life for these patients.

ACKNOWLEDGEMENTS

The authors express their sincere gratitude to Professor Yu-Jin Xu from Zhejiang Cancer Hospital for his invaluable insights into the study design and his critical review of the manuscript. The authors also wish to thank Jin-Yu Wang for his assistance with the statistical analysis.