Human epidermal growth factor receptor 2 in combined hepatocellular-cholangiocarcinoma: Prognostic and predictive guide for adjuvant chemotherapy

Abstract

BACKGROUND

Combined hepatocellular-cholangiocarcinoma (cHCC-CCA) is a rare primary liver cancer with poor prognosis and a lack of reliable markers for prognostic assessment or tailored postoperative therapy. The expression of human epidermal growth factor receptor 2 (HER2) in cHCC-CCA and its potential role in predicting prognosis remain unclear. Therefore, investigating the clinical significance of HER2 and its underlying mechanisms may provide valuable insights to optimize diagnosis and treatment strategies in this disease.

AIM

To investigate the prognostic significance of HER2 and related pathological features in surgically treated patients with cHCC-CCA.

METHODS

This retrospective study included 101 patients with cHCC-CCA who underwent surgical resection at Tianjin Medical University Cancer Institute & Hospital between January 2009 and December 2020. Univariate and multivariate Cox regression analyses were performed to identify independent prognostic risk factors. Kaplan-Meier analysis was used to compare survival across different subgroups, and propensity score matching (PSM) was applied to reduce bias. HER2 expression was assessed using immunohistochemistry, and tertiary lymphoid structures (TLS) were characterized using multiplex immunofluorescence.

RESULTS

A total of 87 patients with pathologically confirmed cHCC-CCA were enrolled. Multivariate Cox regression analysis showed that HER2 positivity was an independent protective factor for postoperative overall survival (hazard ratio = 0.551; P = 0.033). HER2 positivity was also significantly correlated with high TLS abundance. Further survival analysis revealed a key interaction between HER2 status and the efficacy of adjuvant chemotherapy: HER2-negative patients experienced a significant survival benefit (P = 0.0082), whereas HER2-positive patients did not (P = 0.61). This differential treatment effect remained consistent after PSM validation.

CONCLUSION

HER2 is an independent prognostic marker for cHCC-CCA post-surgery, correlates with TLS enrichment, and may predict response to adjuvant chemotherapy, supporting precise postoperative treatment stratification.

Key Words: Combined hepatocellular-cholangiocarcinoma; Human epidermal growth factor receptor 2; Tertiary lymphoid structures; Adjuvant chemotherapy; Prognosis; Surgical resection

Core Tip: Combined hepatocellular-cholangiocarcinoma (cHCC-CCA) is a rare liver malignancy with poor prognosis and limited prognostic markers. This study shows that human epidermal growth factor receptor 2 (HER2) expression is an independent prognostic factor and is positively correlated with tertiary lymphoid structures. HER2 status also predicts differential response to adjuvant chemotherapy, with HER2-negative patients benefiting significantly while HER2-positive patients do not. These findings provide a rationale for using HER2 as a biomarker to guide postoperative treatment decisions and support precision stratification of adjuvant therapy in cHCC-CCA, offering practical implications for individualized patient management.

INTRODUCTION

Combined hepatocellular-cholangiocarcinoma (cHCC-CCA) is a rare primary liver malignancy, accounting for approximately 1%-14.2% of all primary liver cancers[1-5]. Pathologically, cHCC-CCA exhibits features of hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinoma (ICC) and has the worst prognosis among the three major liver cancer types.

Despite sharing the American Joint Committee on Cancer 8^th edition TNM staging system with ICC, cHCC-CCA has distinct pathological and prognostic characteristics, making ICC-based management principles unsuitable[6-9]. Current treatment strategies for cHCC-CCA, including surgery, chemotherapy, and targeted therapy, are largely extrapolated from clinical experience with HCC or ICC. Compared with HCC and ICC, cHCC-CCA carries a poorer prognosis, with a median overall survival (OS) of only 32 months after surgical resection[1]. Even after curative resection, the recurrence rate remains high, and the overall prognosis remains poor[10]. Reported postoperative 5-year OS rate ranges from 23.6% to 36.4%[11,12]. Therefore, there is an urgent need to improve patient prognosis following curative surgery.

Postoperative chemotherapy has shown survival benefits for various tumors, including gastric cancer, breast cancer, and ICC[13-16]. However, there is no clear consensus on whether patients with cHCC-CCA should receive adjuvant chemotherapy after surgery. Given its poor prognosis and high-risk nature, accurately assessing cHCC-CCA-specific survival outcomes is essential. Traditional tumor prognostic factors, such as TNM stage, vascular invasion, tumor size, and lymph node metastasis, while having some predictive value, are insufficient to accurately assess the prognosis of patients with cHCC-CCA and effectively guide individualized treatment.

Human epidermal growth factor receptor 2 (HER2), a member of the epidermal growth factor receptor (EGFR) family, is an established prognostic biomarker for various solid tumors. Recently, significant progress has been made in HER2-targeted therapy research, encompassing multiple strategies, such as small-molecule inhibitors, antibody-drug conjugates (ADCs), and cell-based therapies. Among inhibitors, novel agents like zongertinib demonstrate clinical potential in HER2-mutated non-small cell lung cancer (NSCLC) and other solid tumors by selectively inhibiting HER2 while avoiding EGFR-related toxicity. These inhibitors are also synergistic when combined with ADCs or KRAS G12C inhibitors[17]. Research on ADCs such as SHR-A1811 has expanded to biomarker exploration, revealing that hormone receptor status, the distribution of tumor-infiltrating lymphocytes, and the spatial architecture of HER2-positive cells can influence therapeutic efficacy, thereby advancing the development of precision treatment models[18].

Furthermore, initial clinical data on CAR-macrophage therapy have shown favorable safety profiles, and promoting T-cell infiltration can remodel the tumor microenvironment, thereby offering a new immunotherapeutic direction for HER2-high tumors[19]. Collectively, these advances indicate that HER2-targeted therapy is evolving toward greater precision, combination approaches, and multimodal strategies. HER2, as an important oncogene, is a key prognostic factor and therapeutic target in cancers such as breast and gastric cancer. However, research on HER2 in cHCC-CCA remains limited, and its prognostic significance and clinical application value continue to be controversial and require further clarification.

Given the lack of consensus regarding adjuvant therapy for cHCC-CCAs and the limitations of conventional prognostic factors, this study was designed to identify more effective predictors. We specifically evaluated the prognostic significance of HER2, along with other pathological features such as tumor differentiation and tertiary lymphoid structures (TLS). By conducting a retrospective analysis of surgically treated patients and integrating these markers with traditional factors, we aimed to provide a refined prognostic framework to support personalized decisions regarding adjuvant chemotherapy.

MATERIALS AND METHODS

Patients

We retrospectively collected data from 101 patients with a confirmed postoperative pathological diagnosis of cHCC-CCA who underwent surgical resection at Tianjin Medical University Cancer Hospital between January 2009 and December 2020. The inclusion criteria were as follows: (1) Age ≥ 18 years; (2) Pathological diagnosis confirmed as cHCC-CCA; (3) R0 resection achieved; and (4) Complete clinical and follow-up data available. The exclusion criteria were as follows: (1) History of other malignancies; (2) Follow-up time < 1 month; and (3) Death within 1 month post-surgery. After applying these criteria, 87 patients were included in the study.

All patients provided written informed consent before surgery. Comprehensive preoperative evaluations and laboratory tests were performed. A multidisciplinary team of experienced surgical oncologists developed the treatment plan. The decision to administer adjuvant chemotherapy, as well as the specific regimen selection, was made collaboratively by a multidisciplinary team. The regimens were primarily based on fluoropyrimidines, gemcitabine, or both. The final treatment plan for each patient was determined by at least two senior oncologists, following the contemporary clinical practice guidelines and based on a comprehensive assessment of the patient's postoperative condition, pathological findings, and overall health status. All treatment options were thoroughly discussed with patients, who were allowed to choose their preferred approach.

The study was conducted following the Declaration of Helsinki (2013 revision) and was approved by the Ethics Committee of Tianjin Medical University Cancer Institute and Hospital, which waived the requirement for additional informed consent. The study adhered to good clinical practice guidelines and relevant local regulations. All patient data were anonymized and de-identified prior to the analysis to protect individual privacy.

Data collection

All patients underwent thorough preoperative evaluation. Blood samples were collected within 3 days before surgery. Comprehensive laboratory and clinical data were collected and analyzed, including demographic information (age and sex) and clinical parameters such as hepatitis status, hepatitis B e antigen status, cirrhosis, serum albumin and total bilirubin levels, prothrombin time, histological grade, and vascular, microvascular, and lymph node invasion. Furthermore, preoperative and postoperative serum markers, including alpha-fetoprotein (AFP), carcinoembryonic antigen, and carbohydrate antigen 19-9 (CA19-9), were carefully recorded. All researchers were blinded to the clinical outcomes during data collection. Laboratory data were obtained within 3 days of treatment and subsequently rechecked every 2-3 months.

The pathological examination results, assessed by our pathologists, included key parameters such as tumor size, number of tumors, tumor components, lymphatic metastasis, and histological type. Patients were classified into two groups based on HER2 expression levels determined using immunohistochemistry (IHC): HER2-positive and HER2-negative. After grouping based on HER2 expression status, patients were further stratified into different subgroups based on whether they received postoperative adjuvant therapy. The study endpoint was OS, defined as the time from surgical resection to death from any cause or the last follow-up. For patients who did not visit the hospital, follow-up information was obtained via telephone.

Statistical analysis

Survival analysis was conducted using the Kaplan-Meier method, and survival curves were compared with the log-rank test. OS was the outcome variable. Variables with P < 0.157 in the univariate Cox regression analysis were included in the multivariate analysis. These variables, along with those that were statistically significant in multivariate analysis, were used to construct the predictive model. Variable selection was performed using the Akaike information criterion (AIC), following the recommendations of Heinze et al[20]. Since AIC corresponds to a significance level of α = 0.157 in nested model comparisons, a threshold of P < 0.157 was used to select variables for inclusion in the multivariate model based on univariate analysis[20]. Hazard ratio (HR) and 95% confidence interval (CI) were calculated.

Continuous variables are expressed as mean ± SD or as median (interquartile range). Normality and homogeneity of variance were assessed; if both assumptions were satisfied, t tests were used for comparisons. When these assumptions were not met, the Wilcoxon rank-sum test was applied. Categorical variables are reported as n (%), and baseline comparisons were performed using Pearson’s χ² test or Fisher’s exact test, as appropriate.

IHC staining

Consecutive sections of formalin-fixed paraffin-embedded (FFPE) tissues were prepared and processed using a Ventana BenchMark XT apparatus (Ventana Medical Systems). Sections were dewaxed and subject to antigen retrieval at 95 °C for 30 minutes in ethylenediaminetetraacetic acid (EDTA) repair solution. The sections were subsequently incubated with primary antibodies at 37 °C for 32 minutes, followed by incubation with a horseradish peroxidase-conjugated secondary antibody [multimer horseradish peroxidase (HRP), Ventana] at room temperature for 10 minutes. Positive signals were visualized using diaminobenzidine, and the sections were counterstained with hematoxylin.

HER2 IHC assessment

HER2 IHC was performed to evaluate HER2 protein expression in cHCC-CCA tumor specimens. Given the absence of a standardized scoring system for HER2 in cHCC-CCA, we used the guidelines established by the American Society of Clinical Oncology/College of American Pathologists for gastric/biliary tract cancer to interpret HER2 staining. Scoring was performed using a four-tier system: Score 0 indicated no staining, or incomplete and faint membrane staining in < 10% of tumor cells; Score 1⁺ was defined as incomplete, faint/barely perceptible membrane staining in ≥ 10% of invasive tumor cells; Score 2⁺ represented weak-to-moderate intensity, complete basolateral or lateral membranous staining in ≥ 10% of tumor cells; and Score 3⁺ denoted strong, complete basolateral or lateral membranous staining in > 10% of tumor cells. To comprehensively assess the association between HER2 expression and clinical outcomes, samples with IHC scores of 1⁺, 2⁺, and 3⁺ were classified as the “HER2-positive” group, while those with a score of 0 were categorized as the “HER2-negative” group for subsequent analyses.

Fluorescent multiplex IHC

Fluorescent multiplex IHC was performed on consecutive 3-μm sections from FFPE cHCC-CCA tissue specimens. Following deparaffinization, antigen retrieval was performed in an EDTA-based buffer with heat induction for 10 minutes. Sections were incubated overnight at 4 °C with primary antibodies against CD20, CD21, CD4, and CD8. After washing, secondary antibody incubation was performed, followed by the sequential application of HRP-conjugated fluorescent opal dyes. After another round of heat-mediated retrieval, nuclear staining was performed using DAPI. Finally, the slides were mounted with an anti-fade medium and examined under a fluorescence microscope.

Propensity score matching

To mitigate potential confounding variables that could influence the evaluation of adjuvant chemotherapy, propensity score matching (PSM) was applied to balance the baseline characteristics between the chemotherapy and non-chemotherapy cohorts. Propensity scores were calculated using clinical and pathological variables, including sex, age, hepatitis B virus (HBV) infection status, presence of cirrhosis, maximum tumor diameter, satellite nodules, microvascular and macrovascular invasion, lymph node metastasis, and immune markers.

Patients were matched using a 1:2 nearest-neighbor matching algorithm. The final matched cohort comprised 7 and 14 patients in the HER2-positive chemotherapy and non-chemotherapy groups, respectively, and 13 and 26 patients in the HER2-negative chemotherapy and non-chemotherapy groups, respectively. After matching, all baseline characteristics were adequately balanced (P > 0.05). Subsequent survival analyses were performed on the matched cohort to further evaluate the survival benefit of adjuvant chemotherapy after controlling for confounding variables.

Software

All statistical analyses were performed using R (version 4.4.3) and SPSS (version 29.0). The R packages “survminer”, “survival”, and “forestplot” were used for the statistical analyses.

RESULTS

Patient enrollment with pathological confirmation of cHCC-CCA

Between January 2009 and December 2020, 101 patients diagnosed with cHCC-CCA who underwent surgical resection were screened. Fourteen patients were excluded based on predefined exclusion criteria, leaving 87 patients. The research team systematically collected clinicopathological characteristics and follow-up data and conducted long-term tracking. All included patients underwent a thorough pathological review to confirm whether they met the diagnostic criteria for cHCC-CCA. Histopathological examination with hematoxylin and eosin staining, combined with immunophenotypic analysis, confirmed that all tumors exhibited distinct biphenotypic differentiation. These features were further validated using specific immunohistochemical markers: GPC3 expression in HCC components and CK19 expression in ICC components. As illustrated in Figure 1A, typical HCC components, characterized by polygonal tumor cells arranged in trabeculae, and ICC components, represented by glandular or cribriform structures, were clearly observed, either intermingled or existing independently within the same tumor tissue. Corresponding immunohistochemical staining confirmed GPC3 positivity in HCC regions (Figure 1B) and CK19 positivity in ICC regions (Figure 1C), establishing the biphenotypic nature of the tumors.

Figure 1 Histopathological features of patients with combined hepatocellular cholangiocarcinoma. A: Hematoxylin and eosin staining showing distinct hepatocellular carcinoma-like and cholangiocarcinoma-like regions; B: Immunohistochemical staining for GPC3 confirms the hepatocellular phenotype; C: Immunohistochemical staining for CK19 confirms the cholangiocellular phenotype.

HER2 expression and baseline patient characteristics

For clinical correlation and subsequent analysis, HER2 protein expression in tumor tissues was assessed using IHC, with positive staining localized to the cell membrane. Representative images show that distinct membranous staining in HER2-positive tumor tissues (Figure 2A), whereas a HER2-negative tumor sample is shown in Figure 2B.

Figure 2 Human epidermal growth factor receptor 2 expression and survival outcomes in patients with combined hepatocellular cholangiocarcinoma. A: Postoperative hematoxylin and eosin (H&E) staining and immunohistochemistry detection of human epidermal growth factor receptor 2 (HER2) in HER2-positive patients; B: Postoperative H&E staining and immunohistochemical detection of HER2 in HER2-negative patients; C: Kaplan-Meier curve comparing overall survival between HER2-positive and HER2-negative patients.

Among the 87 included patients, 28 were HER2-positive and 59 were HER2-negative. The median follow-up time was 75.2 months (95%CI: 58.6-91.8), calculated using the reverse Kaplan-Meier method. The collected patient data were divided into HER2-positive and HER2-negative groups. Significant differences were observed in tumor number and vascular invasion between the two groups. After PSM, no differences were observed, indicating a balanced distribution of laboratory and clinical data among the matched groups. Most patients were < 60 years (71.4% vs 62.7%), predominantly male (92.9% vs 71.2%), and had HBV infection (67.9% vs 61.0%) with antiviral therapy. The majority did not have cirrhosis (57.1% vs 66.1%) and had AFP concentrations below 400 ng/mL (75.0% vs 79.7%). CA19-9 levels were above 25 U/mL in 53.6% vs 59.3% of patients. Most did not exhibit macrovascular (82.1% vs 72.9%) or microvascular (92.9% vs 83.1%) invasion.

HER2 positivity as an independent prognostic factor in cHCC-CCA

To further investigate the role of HER2 in cHCC-CCA prognosis, we performed survival analysis. The results showed that HER2-positive patients had significantly better OS than HER2-negative patients did, suggesting that HER2 may be a favorable prognostic factor (Figure 2C).

To systematically assess the impact of other clinicopathological factors, we performed univariate and multivariate analyses incorporating variables such as CA19-9 levels, histological grade, macrovascular invasion, and maximum tumor size. Given the established prognostic value of TLS in cHCC-CCAs, as reported in our previous study, we specifically focused on this feature. Using multiplex immunofluorescence, TLS were characterized in this cohort; representative images (Figure 3) clearly depict the composition and spatial co-localization of distinct immune cell populations within these structures.

Figure 3 Postoperative analysis of tertiary lymphoid structures in patient tissue specimens by immunofluorescence. A: High-magnification view showing the spatial distribution and density of CD20⁺ B cells within the tertiary lymphoid structure (TLS); B: High-magnification view showing the spatial distribution and density of CD4⁺ T helper cells; C: High-magnification view showing the spatial distribution and density of CD8⁺ cytotoxic T cells; D: High-magnification view showing the spatial distribution and density of the CD21⁺ follicular dendritic cell network; E: Visualization of tissue architecture and cell nuclei using DAPI nuclear counterstain; F: Merged image showing the integrated architecture and spatial co-localization of the various immune cell subsets within a representative TLS.

Table 1 shows a summary of the results of the univariate and multivariate analyses for OS. The univariate analysis identified several factors independently associated with OS, including CA19-9 concentration (> 25 vs ≤ 25, HR = 1.698; 95%CI: 1.015-2.838; P = 0.044), histological grade (low vs high/intermediate, HR = 2.469; 95%CI: 1.445-4.213; P = 0.001), macrovascular invasion (HR = 0.513; 95%CI: 0.302-0.872; P = 0.014), maximum tumor diameter (> 5 vs ≤ 5, HR = 1.684; 95%CI: 1.035-2.740; P = 0.036), HER2 positivity (HR = 0.482; 95%CI: 0.281-0.826; P = 0.008), and TLS score (2 vs 3, HR = 2.910; 95%CI: 1.036-8.170; P = 0.043) (1 vs 3, HR = 4.400; 95%CI: 1.542-12.550; P = 0.006) (0 vs 3, HR = 4.569; 95%CI: 1.763-11.841; P = 0.002). Subsequently, factors with P < 0.157 in the univariate analysis were included in the multivariate model. In this model, histological grade (low vs high/intermediate, HR = 2.313; 95%CI: 1.354-3.953; P = 0.002), macrovascular invasion (HR = 0.586; 95%CI: 0.343-1.000; P = 0.050), and HER2 positivity (HR = 0.551; 95%CI: 0.318-0.953; P = 0.033) were significant predictors of OS.

Table 1 Univariate and multivariate analyses of prognostic factors based on overall survival.

Variable	Univariate		Multivariate
	HR (95%CI)	P value	HR (95%CI)	P value
Sex
Male	Reference
Female	1.329 (0.757-2.333)	0.322
Age	1.018 (0.986-1.050)	0.273
HBV
Present	Reference
Absent	1.222 (0.741-2.015)	0.433
Liver cirrhosis
Present	Reference
Absent	1.115 (0.678-1.834)	0.668
ALB (g/L)	0.980 (0.922-1.042)	0.520
TBIL (μmol/L)	1.003 (0.983-1.024)	0.780
PT (sec)	1.016 (0.775-1.332)	0.909
AFP (ng/mL)
≤ 40	Reference
> 40	1.192 (0.736-1.931)	0.475
CA19-9 (U/mL)
≤ 25	Reference		Reference
> 25	1.698 (1.015-2.838)	0.044a	1.285 (0.720-2.294)	0.396
Histological grade
High/median	Reference		Reference
Low	2.469 (1.445-4.213)	0.001a	2.313 (1.354-3.953)	0.002a
Satellite lesions
Present	Reference
Absent	0.923 (0.503-1.694)	0.795
Microvascular invasion
Present	Reference		Reference
Absent	0.575 (0.290-1.141)	0.114a	0.734 (0.326-1.655)	0.456
Macrovascular invasion
Present	Reference		Reference
Absent	0.513 (0.302-0.872)	0.014a	0.586 (0.343-1.000)	0.050
Lymphatic node metastasis
Present	Reference		Reference
Absent	0.566 (0.279-1.148)	0.115	0.733 (0.321-1.670)	0.459
Maximum tumor diameter (cm)
≤ 5	Reference		Reference
> 5	1.684 (1.035-2.740)	0.036a	1.159 (0.671-2.003)	0.596
HER2
Negative	Reference		Reference
Positive	0.482 (0.281-0.826)	0.008a	0.551 (0.318-0.953)	0.033a
TLS score
3	Reference		Reference
2	2.910 (1.036-8.170)	0.043a	2.069 (0.695-6.160)	0.192
1	4.400 (1.542-12.550)	0.006a	2.331 (0.675-8.048)	0.181
0	4.569 (1.763-11.841)	0.002a	2.461 (0.820-7.391)	0.108

^aP < 0.05.

HR: Hazard ratio; CI: Confidence interval; CA19-9: Carbohydrate antigen 19-9; HER2: Human epidermal growth factor receptor 2; HBV: Hepatitis B virus; ALB: Albumin; TBIL: Total bilirubin; PT: Prothrombin time; AFP: Alpha-fetoprotein; TLS: Tertiary lymphoid structures.

Notably, in the univariate analysis, in addition to traditional prognostic factors, two less-studied factors-HER2 and TLS scores-were identified. Multivariate analysis confirmed that histological grade, macrovascular invasion, and HER2 positivity were independent prognostic factors for cHCC-CCA.

Association between HER2 positivity and TLS abundance

Interestingly, while the TLS score was a significant prognostic factor in univariate analysis, it was not independent in the multivariate model, indicating the hypothesis that TLS interacts with other covariates, notably HER2.

We initially characterized TLS morphologically. Immunohistochemical and immunofluorescence analyses confirmed the presence of mature TLS, structured as organized aggregates with a core of CD20⁺ B cells surrounded by CD4⁺ and CD8⁺ T cells, along with CD21⁺ follicular dendritic cells (Figure 4A-E). Supporting our hypothesis, subsequent correlation analysis revealed that TLS scores were significantly higher in HER2-positive tumors than in HER2-negative tumors (Figure 4F), indicating a clear association between HER2 status and TLS abundance. Collectively, these findings suggest that the enrichment of TLS within HER2-positive tumors may at least partially contribute to the favorable prognosis observed in this patient group.

Figure 4 Assessment of tertiary lymphoid structures and their prognostic value stratified by human epidermal growth factor receptor 2 status. A: Hematoxylin and eosin staining showing the histopathological structure of tertiary lymphoid structures (TLS); B: Immunohistochemistry (IHC) staining for CD20 within the TLS; C: IHC staining for CD21; D: IHC staining for CD4; E: IHC staining for CD8; F: Correlation analysis between TLS abundance (TLS score) and human epidermal growth factor receptor 2 (HER2) expression; G: Forest plot showing subgroup survival analysis comparing overall survival between HER2-positive and HER2-negative patients across clinicopathological factors.

Differential benefit of adjuvant chemotherapy by HER2 Status

To further investigate the clinical significance of HER2 treatment decision-making, we conducted a subgroup analysis (Figure 4G). Potential confounders for the different subgroups are displayed in Figure 4G. An HR value < 1 indicated better OS for HER2 (+) patients than for the HER2 (-) patients, and statistically significant differences were observed in some subgroups. For example, in the TLS analysis, patients with a score of 0 showed no statistically significant difference in HER2 (+) and HER2 (-) patient groups (HR = 0.992; 95%CI: 0.432-2.273; P = 0.985). In contrast, for TLS scores of 1-3, HER2 (+) patients had lower OS risk than did the HER2 (-) patients (HR = 0.397; 95%CI: 0.189-0.831; P = 0.014). Regarding adjuvant chemotherapy, when it was administered, OS risk did not significantly differ between HER2 (+) and HER2 (-) patients (HR = 0.694; 95%CI: 0.230-2.101; P = 0.519). However, in the absence of adjuvant chemotherapy, HER2 (+) patients had a lower risk of OS than HER2 (-) patients did (HR = 0.311; 95%CI: 0.159-0.606; P = 0.001).

Furthermore, in all subgroups receiving adjuvant chemotherapy, the prognosis did not significantly differ between HER2-positive and HER2-negative patients (HR = 0.694; 95%CI: 0.230-2.101; P = 0.519). However, in the subgroup that did not receive adjuvant chemotherapy, the HER2 (+) group had a significantly better prognosis (HR = 0.311; 95%CI: 0.159-0.606; P = 0.01). These results indicate that adjuvant chemotherapy may provide similar prognostic outcomes for HER2-negative and HER2-positive patients.

To further validate this finding, we compared survival outcomes based on chemotherapy receipt within each HER2 subgroup. Survival curve analysis revealed that in the HER2-negative group (Figure 5A), patients receiving adjuvant chemotherapy had significantly better OS than those who did not receive chemotherapy (P = 0.0082). In contrast, in the HER2-positive group (Figure 5B), the OS did not significantly differ between patients receiving adjuvant chemotherapy and those who did not (P = 0.61).

Figure 5 Comparative analysis of adjuvant chemotherapy benefit in human epidermal growth factor receptor 2-positive and human epidermal growth factor receptor 2-negative patients before and after propensity score matching. A: Kaplan-Meier survival curve for overall survival (OS) in human epidermal growth factor receptor 2 (HER2)-negative patients before propensity score matching (PSM), showing a significant benefit from adjuvant chemotherapy; B: Kaplan-Meier survival curve for OS in HER2-positive patients before PSM, showing no significant difference between chemotherapy and non-chemotherapy groups; C: Kaplan-Meier survival curve for OS in HER2-negative patients after PSM, confirming the continued survival benefit of adjuvant chemotherapy; D: Kaplan-Meier survival curve for OS in HER2-positive patients after PSM, validating no significant benefit from conventional chemotherapy.

Validation of chemotherapy benefit using PSM

To validate the robustness of the aforementioned findings and control for potential confounding biases, we performed 1:2 PSM between HER2-positive and HER2-negative patients (baseline characteristics are presented in Tables 2 and 3). In the matched cohort, the HER2-negative group included 13 patients who received adjuvant chemotherapy and 26 who did not, whereas the HER2-positive group included 7 patients who received adjuvant chemotherapy and 14 who did not. Survival analysis of the matched cohort (Figure 5C and D) further supported our conclusions: Among HER2-negative patients, those receiving adjuvant chemotherapy had significantly better OS than did the non-chemotherapy group (HR = 0.62; 95%CI: 0.43-0.89; P = 0.010), indicating that adjuvant chemotherapy provides an independent survival benefit for HER2-negative patients.

Table 2 Clinicopathological characteristics of patients in the human epidermal growth factor receptor 2 positive group after propensity score matching.

Characteristic	Median (IQR) or n (%)		P value
	Chemotherapy group (n = 7)	Non-chemotherapy group (n = 14)
Sex			0.533
Male	7 (100.0)	12 (85.7)
Female	0	2 (14.3)
Age	57 (54-65)	56.5 (53.8-61.5)	0.793
HBV			0.870
Present	4 (57.1)	10 (71.4)
Absent	3 (42.9)	4 (28.6)
Liver cirrhosis			0.640
Present	4 (57.1)	5 (35.7)
Absent	3 (42.9)	9 (64.3)
ALB (g/L)	41.1 (38.1-43.1)	42.4 (39.5-45.8)	0.232
TBIL (μmol/L)	22.9 (13.5-24.8)	17.7 (13.6-21.1)	0.117
PT (sec)	11.2 (10.6-11.9)	11.2 (10.7-12.3)	0.911
AFP (ng/mL)			0.877
> 40	4 (57.1)	6 (42.9)
≤ 40	3 (42.9)	8 (57.1)
Satellite lesions			1.000
Present	1 (14.3)	2 (14.3)
Absent	6 (85.7)	12 (85.7)
Microvascular invasion
Present	0	0
Absent	7 (100.0)	14 (100.0)
Macrovascular invasion			0.844
Present	2 (28.6)	2 (14.3)
Absent	5 (71.4)	12 (85.7)
Lymphatic node metastasis			0.100
Present	2 (28.6)	0
Absent	5 (71.4)	14 (100.0)
Maximum tumor diameter (cm)			1.000
≤ 5	3 (42.9)	7 (50.0)
> 5	4 (57.1)	7 (50.0)
CA19-9 (U/mL)			1.000
≤ 25	3 (42.9)	7 (50.0)
> 25	4 (57.1)	7 (50.0)
Histological grade			1.000
High/median	4 (57.1)	8 (57.1)
Low	3 (42.9)	6 (42.9)
TLS score			0.782
3	2 (28.6)	3 (21.4)
2	2 (28.6)	5 (35.7)
1	2 (28.6)	2 (14.3)
0	1 (14.3)	4 (28.6)

IQR: Interquartile ranges; CA19-9: Carbohydrate antigen 19-9; HBV: Hepatitis B virus; ALB: Albumin; TBIL: Total bilirubin; PT: Prothrombin time; AFP: Alpha-fetoprotein; TLS: Tertiary lymphoid structures.

Table 3 Clinicopathological characteristics of patients in the human epidermal growth factor receptor 2 negative group after propensity score matching.

Characteristic	Median (IQR) or n (%)		P value
	Chemotherapy group (n = 13)	Non-chemotherapy group (n = 26)
Sex			0.713
Male	8 (61.5)	19 (73.1)
Female	5 (38.5)	7 (26.9)
Age	52.0 (47.0-63.5)	56.5 (51.8-62.3)	0.290
HBV			0.837
Present	9 (69.2)	15 (57.7)
Absent	4 (30.8)	11 (42.3)
Liver cirrhosis			0.378
Present	2 (15.4)	9 (34.6)
Absent	11 (84.6)	17 (65.4)
ALB (g/L)	45.6 (39.6-48.6)	42.3 (39.2-44.6)	0.104
TBIL (μmol/L)	17.7 (12.9-20.6)	14.9 (12.1-21.3)	0.870
PT (seconds)	10.7 (10.5-11.5)	11.1 (10.8-11.5)	0.241
AFP (ng/mL)			0.183
> 40	2 (15.4)	11 (42.3)
≤ 40	11 (84.6)	15 (57.7)
Satellite lesions			0.852
Present	2 (15.4)	2 (7.7)
Absent	11 (84.6)	24 (92.3)
Microvascular invasion			1.000
Present	1 (7.7)	2 (7.7)
Absent	12 (9.23)	24 (92.3)
Macrovascular invasion			0.883
Present	3 (23.1)	4 (15.4)
Absent	10 (76.9)	22 (84.6)
Lymphatic node metastasis			0.852
Present	2 (15.4)	2 (7.7)
Absent	11 (84.6)	24 (92.3)
Maximum tumor diameter (cm)			1.000
≤ 5	7 (53.8)	14 (53.8)
> 5	6 (46.2)	12 (46.2)
CA19-9 (U/mL)			0.254
≤ 25	9 (69.2)	13 (50.0)
> 25	4 (30.8)	13 (50.0)
Histological grade			0.904
High/median	5 (38.5)	8 (30.8)
Low	8 (61.5)	18 (69.2)
TLS score			0.234
3	2 (15.4)	3 (11.5)
2	1 (7.7)	4 (15.4)
1	0	4 (15.4)
0	10 (76.9)	15 (57.7)

IQR: Interquartile ranges; CA19-9: Carbohydrate antigen 19-9; HBV: Hepatitis B virus; ALB: Albumin; TBIL: Total bilirubin; PT: Prothrombin time; AFP: Alpha-fetoprotein; TLS: Tertiary lymphoid structures.

Conversely, among HER2-positive patients, no significant difference in OS was observed between the chemotherapy and non-chemotherapy groups (HR = 0.91; 95%CI: 0.72-1.15; P = 0.42). This consistent finding after PSM confirms that HER2-positive status may help identify a patient subgroup that does not derive significant benefits from conventional adjuvant chemotherapy.

DISCUSSION

cHCC-CCA is a rare primary liver tumor characterized by histological features of HCC and ICC. It accounts for a relatively small proportion of primary liver cancers[21-24]; however, it ranks as the second leading cause of liver-cancer mortality worldwide[25]. The complex nature of cHCC-CCAs is often underexplored, and conventional diagnostic methods, such as histopathology and radiological imaging, frequently provide limited insight. Currently, no standardized therapy exists for cHCC-CCA[1,26], and surgical resection remains the only potentially curative treatment. However, strict eligibility criteria limit surgical intervention to only a few patients[27]. According to the National Comprehensive Cancer Network guidelines, patients with Child-Turcotte-Pugh (CTP) class A-or carefully selected CTP class B-who have suitably located tumors, adequate liver reserve, suitable future liver remnants, and no portal hypertension may be eligible for surgical resection[28]. However, even among those who meet these criteria, the recurrence rate of cHCC-CCA remains high. Consequently, effective perioperative management strategies are urgently needed to reduce recurrence. Postoperative adjuvant chemotherapy has shown survival benefits in patients with various tumors, highlighting the importance of identifying the patient population most likely to benefit from such therapy.

Recently, HER2-targeted therapy has advanced significantly, with a shift in focus from traditional monoclonal antibodies to more precise and diversified strategies. ADCs represent a particularly promising development, with agents such as T-DXd transforming the treatment landscape by successfully extending indications to HER2-low breast cancer[29]. Concurrently, bispecific antibodies (e.g., zanidatamab) have shown notable efficacy through dual-targeting mechanisms, offering a new therapeutic option for patients with advanced diseases[30]. Moreover, the scope of HER2-targeted therapy continues to expand across additional cancer types, as exemplified by orally active agents such as zongaitinib for HER2-mutant NSCLC[17]. Despite these advances, the expression pattern and clinical significance of HER2 in cHCC-CCA remain unclear.

In our cohort, 32.2% (28/87) of patients were HER2-positive. In the multivariate analysis, HER2 positivity was confirmed as an independent favorable prognostic factor, along with histological grade and macrovascular invasion. Notably, although the TLS score was a significant prognostic factor in univariate analysis, it did not remain independent in the multivariate model. Given the established prognostic value of HER2 and TLS, we hypothesized that their effects might be interrelated. Indeed, subsequent correlation analysis confirmed that HER2-positive tumors had significantly higher TLS scores than HER2-negative tumors, suggesting that HER2 status may influence the immune microenvironment, particularly TLS abundance. Recent studies indicate that TLS provides a vital local microenvironment for immune responses against tumor cells and is considered a marker of favorable clinical outcomes in various malignancies. TLS can enhance tumor-specific immune responses and facilitate adaptive immune activation. One of the primary effector functions of B cells within TLS is the production of disease-specific antibodies[31]. TLS has been associated with improved prognosis across multiple cancer types[32], and its prognostic value often remains independent of TNM staging, as documented in lung[33], colorectal[34], and pancreatic cancers[35]. However, its prevalence varies significantly across cancer types and among patients[31]. In cHCC-CCA, TLS scores have been linked to patient survival.

Notably, while intratumoral TLS is associated with prolonged survival, abundant TLS in adjacent tissues paradoxically predicts poor outcomes. The spatial distribution and density of TLS reflect key features of the cHCC-CCA immune microenvironment, correlate significantly with prognosis, and serve as potential biomarkers of immunotherapy response[36]. Our findings further demonstrate that TLS spatial distribution and density are significantly correlated with HER2 status, suggesting that the prognostic value of HER2 may be mediated, at least in part, by its influence on TLS formation and subsequent immune modulation. Available evidence suggests a close relationship between HER2 signaling and the regulation of the tumor immune microenvironment. One plausible hypothesis is that HER2 pathway activation may remodel the local chemokine milieu through downstream signaling cascades, thereby promoting immune cell recruitment and TLS formation. Specifically, HER2 signaling can activate the PI3K/AKT and MAPK pathways, which converge on NF-κB-dependent transcriptional programs and modulate the expression of inflammatory mediators and chemokines[37]. Meanwhile, studies have shown that PI3K/NF-κB activation enhances chemokine secretion, with tumor-intrinsic PI3K signaling upregulating CXCL13 in an NF-κB-dependent manner[38]. Furthermore, TLS formation is largely orchestrated by lymphoid chemokines such as CXCL13, CCL19, and CCL21, which recruit B cells, T cells, and dendritic cells and guide their spatial organization into TLS-like structures[39,40]. Notably, spatial transcriptomic and immunological analyses in classical HER2-driven tumors, such as HER2-positive breast cancer, have also documented the presence of TLS or TLS-like immune aggregates, providing biological support for the notion that HER2-activated contexts may favor immune organization, including TLS formation[41]. Collectively, these lines of evidence support a parsimonious model in which HER2 pathway activity may facilitate TLS development by enhancing lymphoid chemokine expression and promoting immune cell aggregation. Based on the mechanistic hypothesis outlined above, future studies could investigate this process in cHCC-CCA through integrated multi-omics analyses. Validating the correlation between HER2 expression and the spatial distribution of TLS in larger cohorts, along with exploring the underlying mechanisms, would help deepen our understanding of this biological process.

Consistent with our observations, accumulating evidence across multiple tumor types reveals a strong association between HER2 upregulation and enhanced immune infiltration within the tumor microenvironment. For example, in HR-positive/HER2-positive breast cancer, molecular subtyping has identified a dimmunomodulatory subtype characterized by markedly increased immune cell infiltration and heightened sensitivity to HER2-targeted therapies[42]. Similarly, studies in colorectal cancer show that elevated HER2 expression correlates with increased levels of T cells, B cells, macrophages, and other intrinsic immune cells, indicating a shift from an immune-cold to an immune-hot phenotype[43]. In endometrial carcinomas, HER2-positive tumors exhibit higher immune checkpoint gene expression and increased immune cell infiltration[44]. Collectively, these pan-cancer findings reinforce the role of HER2 in shaping the immune microenvironment, supporting the biological plausibility of our results.

Given the interplay between HER2 and the immune microenvironment, the central finding of our study was the differential response to adjuvant chemotherapy based on HER2 status. Kaplan-Meier analysis showed that HER2-positive patients did not benefit from adjuvant chemotherapy, as indicated by no substantial difference in OS between those who received it and those who did not. Therefore, HER2 may serve as a useful biomarker for identifying patients who are likely to benefit from adjuvant chemotherapy. This finding provides a crucial foundation for individualized treatment decisions for cHCC-CCA, a disease for which standard treatment options are still lacking.

This is the first study where HER2 expression levels were linked with prognosis and response to adjuvant chemotherapy in patients with cHCC-CCA. The findings provide personalized clinical guidance. Compared with those of previous studies, our results identified a distinct role of HER2 as a favorable prognostic factor in cHCC-CCA. However, unlike other cancers, in some solid tumors such as ICC, HER2 overexpression is often associated with strong invasiveness and poor prognosis[45]. This result contrasts sharply with patterns seen in breast and gastric cancers, where HER2-positive tumors are highly sensitive to targeted therapy. In those malignancies, HER2 serves as a therapeutic target for specific agents; however, our study focused solely on conventional chemotherapy. These findings suggest that HER2 positivity in cHCC-CCA may indicate intrinsic resistance to standard chemotherapy regimens.

Based on these findings, we propose two non-mutually exclusive mechanisms that may collectively explain the observed resistance to chemotherapy in HER2-positive cHCC-CCA. First, HER2 positivity may be associated with a distinct immune context, potentially characterized by specific TLS patterns, which could intrinsically reduce chemotherapy efficacy; in addition, chemotherapy may disrupt pre-existing anti-tumor immune responses. Second, the intrinsic heterogeneity of cHCC-CCA, comprising hepatocellular and cholangiocellular lineages, may confer resistance to conventional chemotherapy regimens developed primarily for ICC. Thus, HER2 positivity may serve as a marker of a highly resistant and heterogeneous tumor phenotype. However, the precise mechanisms underlying these observations require further investigation.

The present study provides meaningful insights; nevertheless, some important limitations should be acknowledged. First, as a single-center retrospective analysis, the overall sample size remains limited despite including 87 patients, a constraint largely attributable to the rarity of cHCC-CCA. The limited sample size may introduce selection and information biases and restrict the generalizability of our findings. PSM was used to balance baseline characteristics; nonetheless, residual confounding factors cannot be entirely excluded. Therefore, the observed associations should be interpreted with caution and validated in multicenter prospective studies. Second, the limited sample size also impacted the statistical power of the subgroup analyses. After PSM, only seven HER2-positive patients received adjuvant chemotherapy, resulting in considerable uncertainty in the treatment effect estimates. The finding that HER2-positive patients did not appear to derive significant benefit from adjuvant chemotherapy should be regarded as exploratory, and its clinical relevance requires further confirmation in larger prospective studies. Third, this study was designed to explore the overall clinical relevance of HER2 protein expression rather than to strictly distinguish therapeutic targets. Accordingly, validation using fluorescence in situ hybridization was not performed to confirm gene amplification status. While this methodology reflects the translational nature of the study, it also represents a methodological limitation. Finally, to improve prognostic accuracy and therapeutic stratification in cHCC-CCA, future studies should integrate multicenter resources, address the challenges of recruiting rare cases, expand cohort sizes, and systematically incorporate more comprehensive molecular profiling to validate and extend the findings of this research. Establishing a cHCC-CCAspecific HER2 testing protocol through multicenter collaboration and integrating protein expression with gene amplification analysis would help to more precisely define the clinical significance of HER2 in this rare tumor.

In summary, this study shows that HER2 is an independent prognostic factor in cHCC-CCA after surgical resection and that its expression is positively correlated with TLS abundance. Our findings suggest a potential role for HER2 as a predictive biomarker to guide adjuvant chemotherapy decisions in cHCC-CCA: HER2-negative patients may benefit from adjuvant chemotherapy, whereas HER2-positive patients might require alternative therapeutic strategies. Overall, this study highlights the value of identifying biomarkers to refine precision treatment stratification in patients with cHCC-CCA.

CONCLUSION

HER2 expression is an independent favorable prognostic factor for patients with surgically treated cHCC-CCA and is significantly associated with higher TLS abundance. Furthermore, HER2 status is a predictive biomarker for response to adjuvant chemotherapy: HER2-negative patients derive a clear survival benefit, while HER2-positive patients do not. These findings support using HER2 for precise postoperative risk stratification, guiding the clinical therapeutic application of adjuvant chemotherapy for HER2-negative patients.

ACKNOWLEDGEMENTS

We express our sincere appreciation to the patients and surgical teams at the Tianjin Medical University Cancer Institute and Hospital for their essential role in this research.