Introduction

Primary liver cancer ranks fourth globally in cancer-related mortality, with hepatocellular carcinoma (HCC) being the predominant histologic type, constituting 75-85% of cases (Kotsari et al. 2023; Sung et al. 2021; Zhou et al. 2023). Unlike other cancers experiencing a decrease in disease burden, the incidence of HCC continues to rise, and its 5-year survival rate stands at a concerning 12% (Tang et al. 2018; Yang et al. 2019; Zheng et al. 2018). The prognosis of HCC mainly depends on the tumor stage at the time of treatment and the extent of liver function impairment (Gan et al. 2018; Llovet et al. 2021). Patients diagnosed in the early stage may achieve a 5-year survival rate of nearly 70%, while those in advanced stages face a median survival of only 1–2 years (Bangaru et al. 2020). Therefore, there is urgent need to identify and develop more effective therapeutic targets.

Tumors are strongly influenced by the tumor microenvironment (TME), a concept first proposed by Ioannides in 1993, specifically referring to the local environment where tumors develop (Ioannides and Whiteside 1993). TME comprises tumor cells, tumor stromal cells, and non-cellular components of the extracellular matrix (Baghban et al. 2020; Feng et al. 2022). The cells most prominently contributing to HCC progression include regulatory T cells (Tregs), tumor-associated neutrophils (TANs), and tumor-associated macrophages (TAMsWang et al. 2020b; Zheng et al. 2023). Generally, high infiltration of Tregs into tumor tissues correlates with poor prognosis in HCC patients, as well as in many other cancers (Langhans et al. 2019).

Recently, there has been increased focus on chemokine ligands such as Chemokine ligand 17 (CCL17), for which the targeted drug mogamulizumab has been developed(Maeda et al. 2019a). Previous research has shown that aberrantly high expression of CCL17 in various cancers, including leukemia, lung cancer, bladder cancer, breast cancer, gastric cancer, and liver cancer (Bouchet et al. 2020; Gao et al. 2020; Higuchi et al. 2019; Lim et al. 2014; Mishalian et al. 2014; Wang et al. 2018, 2019; Ye et al. 2022), serves as a primary driver for recruiting CCR4-expressing Tregs into the TME. This, in turn, extensively suppresses the anti-tumor immune response and promotes tumor progression(Gao et al. 2020). However, CCL17-positive cells have only been reported to be distributed within the tumor stroma in the context of HCC (Zhou et al. 2016). The direct release of CCL17 by tumor cells has not yet been characterized, and the specific role of CCL17 in the interaction between tumor and tumor stroma in the TME remains undefined. Furthermore, there is no definitive evidence to determine whether CCL17 might be an independent target for cancer therapy or whether it might cooperate with other effective immune checkpoint inhibitors to enhance the anti-tumor effect in HCC. Therefore, the relationship between the expression levels of immune checkpoints and CCL17 in HCC merits further exploration.

Ecto-5’nucleotidase, commonly referred to as CD73, functions as a converter that transforms AMP into extracellular adenosine (Harvey et al. 2020). Previous studies have identified an overexpression of CD73 in many solid tumors, including HCC, and this overexpression is frequently associated with poor prognosis (Harvey et al. 2020; Ma et al. 2020). Additionally, CD73 is linked to tumor metastasis, drug resistance, cancer-stem-cell traits, and has been identified as a novel immune checkpoint (Lupia et al. 2018; Ma et al. 2020). Recent research has shown that CD73 can be expressed by various cell types including immune cells, stromal cells, epithelial cells, endothelial cells, and cancer cells (Da et al. 2022; Yu et al. 2020). However, there remains limited knowledge about the relationship between CD73, CCL17, and CCR4 within the tumor microenvironment in HCC.

Transcatheter arterial chemoembolization (TACE), a treatment that combines chemotherapeutic drug injection with occlusion of the tumor’s blood supply, has emerged as a primary option for patients with BCLC (Barcelona Clinic Liver Cancer) stage B HCC (Dev et al. 2020). The application of TACE, however, is based on traditional tumor characteristics, making it difficult to predict which patients will benefit most from postoperative TACE therapy (Wang et al. 2020a). This complexity highlights the need for reliable indicators for TACE therapy.

This study seeks to investigate the clinical significance of CCR4, CCL17, CD73, and HHLA2 in HCC, exploring their interconnectedness within the tumor immune microenvironment. Additionally, we aim to identify a novel index to predict the effectiveness of postoperative TACE and construct predictive nomograms for overall survival (OS) and recurrence-free survival (RFS), thereby enhancing clinical decision-making in the management of HCC.

Materials and methods

Patients and follow-up

A total of 873 consecutive HCC patients who underwent radical operations at Zhongshan Hospital, Fudan University were enrolled. Patients were stratified into two distinct cohorts based on admission time: an early cohort, comprising 354 patients admitted between 2008 and 2009, served as the training group, while a late cohort, including 519 patients admitted between 2010 and 2012, constituted the validation group. The inclusion criteria were as follows: (1) patients with pathologically diagnosed HCC, (2) patients accepting no preoperative treatment, (3) patients with complete tumor excision and negative incisal margins confirmed by pathology, (4) patients without evidence of other malignant tumors, (5) patients with detailed clinicopathological and follow-up data. Appropriate tumor tissues were subjected to formalin-fixing and paraffin-embedding, and two expert pathologists were assigned to confirm the clinical stage. RFS was defined as the period from surgery to the date of confirmed recurrence o, while OS was recorded from the date of surgery to death.

Tissue microarray (TMA), immunohistochemistry (IHC), and immunofluorescence (IF)

TMAs were prepared from formalin-fixed, paraffin-embedded surgical specimens that were HE-stained to select representative tumor areas. Triplicate cores of 1 mm diameter were taken from each individual tumor.

Experienced pathologists performed the IHC staining for CCL17 (abcam, 2.5 µg/mL), CCR4 (abcam, 5 µg/mL), CD73 (Bio-Techne, 5 µg/mL), CD47 (abcam, 0.3 µg/mL), HHLA2 (Atlas, 0.5 µg/mL), PD-L1 (abcam, 2 µg/mL), CD34 (abcam, 0.22 µg/mL), CD68 (CST, 1:500), CD66b (abcam, 0.026 µg/mL), Foxp3 (abcam, 3.67 µg/mL), CD8 (abcam, 0.4 µg/mL), CD4 (abcam, 1.3 µg/mL). The slides underwent incubation, deparaffinization, rehydration, antigen retrieval (10 min at 120 °C in a citrate buffer, pH 6.0), and endogenous peroxidase inactivation. Then, they were sequentially incubated with primary and secondary antibodies. The preparation for IF was identical to that for IHC, but followed by fluor-conjugated secondary antibody. The experimental procedure of IHC and IF staining of TMA was performed as described in our previous study (Gao et al. 2007; Jing et al. 2019).

Quantification of CCL17, CCR4, CD73, CD47, HHLA2, PD-L1, and immune cell infiltration

CCL17, CCR4, CD73, CD47, HHLA2, and PD-L1 protein expressions in tumor (marked as CCL17-T, CCR4-T, CD73-T, CD47, HHLA2, and PD-L1) were evaluated by immunohistochemistry and quantified using a combined score (CS: % positive cells × intensity). Specifically, the percentage of positive cells was categorized into four levels: 1 (1-10%), 2 (11-50%), 3 (51-80%), and 4 (> 80%), while staining intensity was graded on a scale of 0 (no staining), 1 (light yellow), 2 (tan), and 3 (dark brown). In the survival analysis, low expression was defined as a CS score of 0–4, while high expression was defined as a score of 5–12. Additionally, we assessed the expression levels of these proteins, including CCL17-I, CCR4-I, and CCR4 + CD73+, within the tumor stroma by quantifying the number of positive cells. Moreover, the numbers of CD8+, CD66+, or FOXP3 + cells were quantified in five high-power fields (400X) for each group. The optimal cutoff values for stromal expression were determined using X-tile analysis, as described in our previous study (Zhang et al. 2018).

Statistical analysis

All statistical analyses were conducted using SPSS version 21 (IBM Corporation, Armonk, NY, USA). Associations between CCL17, CCR4, CD73, HHLA2, PD-L1, and other variables, such as tumor size and AFP, were assessed using the Pearson chi-squared test. The Kaplan-Meier method was employed to evaluate the survival curves of OS and PFS for patients, while differences between groups were calculated using the log-rank test. Additionally, univariate and multivariate analyses were executed with the Cox proportional hazards model. A new nomogram was constructed using R software version 3.0.2.

Results

Patient characteristics

We analyzed TMAs comprising 873 HCC tissues sourced from Zhongshan Hospital, categorizing them into two distinct groups: a training cohort consisting of 354 HCC specimens and a validation cohort composed of 519 HCC cases. It is noteworthy that in our cohorts, the etiology of HCC in China predominantly involves hepatitis B virus (HBV) infection, with an observed infection rate of 87.3%. In the training cohort, the OS rates at 1, 3, and 5 years were observed to be 96%, 79%, and 66%, respectively. Concurrently, the RFS rates for the same time intervals were 87%, 68%, and 56%, respectively. Meanwhile, in the validation cohort, the OS rates at 1, 3, and 5 years were recorded at 91%, 56%, and 40%, respectively, with corresponding RFS rates of 76%, 42%, and 29%. Comparisons of training cohort and validation cohort clinical characteristics are listed in Table 1.

Table 1 The comparation between training cohort and validation cohort
Full size table

Tumor expression of CCL17 and CCR4 proteins

To evaluate the expression pattern of CCL17, CCR4 in HCC, we performed IHC on TMAs. Representative microphotographs of CCL17-T and CCR4-T immunostaining are displayed (expressions in tumor). Figure 1A-B shows representative IHC staining of CCL17-T and CCR4-T expressions, categorized as “negative,” “weak,” “moderate,” and “strong,” reflecting ascending levels of expression.

Intriguingly, CCL17 and CCR4 were also found to be widely expressed in tumor stroma, hinting at a subtype of stromal cells, mainly immune cells, expressing these two molecules. This observation aligns with previous reports that identified CCL17 and CCR4 as critical participants in the tumor’s immunosuppressive environment. To further understand CCL17 and CCR4’s role in the tumor microenvironment, expressions in the tumor stroma (CCL17-I and CCR4-I) were also analyzed (Fig. 1A-B).

Fig. 1
figure 1

Expression of CCL17 and CCR4 in HCC tissue samples and their impact on prognosis. Representative micrographs of CCL17 (A) and CCR4 (B) expression within tumor and stroma; Kaplan-Meier survival curves for OS and RFS in HCC patients, high CCL17-T expression and CCL17-I (≥ 30) demonstrated significant associations with poor OS and RFS in the training cohort (C) and validation cohort (D), respectively. (Scale bar = 50 μm)

Full size image

High CCL17 and CCR4 expression impaired HCC clinical outcomes

The training cohort was divided into high and low CCL17-T expression groups according to IHC staining results. Out of 354 patients, 169 (47.8%) were classified into the high-expression group, while 185 (52.2%) were assigned to the low-expression group. Notably, increased CCL17-T levels were primarily found in patients with multiple tumor lesions, poor tumor differentiation, and microvascular invasion (MVI). Additionally, a significant correlation was identified between high CCL17-T expression and several prognostic indicators in HCC, including CD34, the Treg/CD8 ratio, HHLA2, and CD73-T (all P < 0.050, Table 2). Kaplan-Meier survival analysis indicated that elevated levels of CCL17-T and CCL17-I were significantly associated with reduced OS and RFS in HCC patients (both P < 0.050, Table 3; Fig. 1C). In a similar vein, patients with high CCR4-T expression (179, 50.6%) demonstrated markedly shorter OS (P < 0.001) and RFS (P = 0.039), whereas CCR4-I expression did not exhibit a statistically significant impact on HCC patient outcomes (Table 3). These observations were further corroborated by the validation cohort, emphasizing the importance of these findings for the clinical management of HCC (Table 4; Fig. 1D).

Table 2 Demographic and clinical characteristics of patients in training cohort
Full size table

CD73 and HHLA2 inversely correlated with HCC prognosis

We subsequently explored the associations between immune checkpoints and clinical outcomes in HCC. CD73, CD47, HHLA2, and PDL1 were selected for analysis due to their recognized immunosuppressive role in HCC. Expression levels were assessed through IHC as depicted in Fig. 2A-B. In the training cohort, the rates of high CD73, CD47, HHLA2, and PDL1 expression were 58.8%, 36.2%, 43.8%, and 40.4%, respectively. In the validation cohort, the percentages of high CD73, CD47, HHLA2, and PDL1 expression were 63.0%, 34.3%, 48.2%, and 40.3%, respectively (Fig. 2C-D). Survival analysis demonstrated that high CD73 or HHLA2 expression independently correlated with poor OS and RFS, whereas CD47 and PDL1 did not serve as predictors (Tables 3 and 4).

Fig. 2
figure 2

Expression of CD73, CD47, HHLA2, and PDL1 in HCC tissue samples. Representative micrographs illustrating the expression of PDL1, HHLA2, CD47, and CD73 in tumor tissue (B) and the positive or negative expression of CD73 in the stroma (A). Statistical expression profiles of CD73, CD47, HHLA2, and PDL1 in training (C) and validation cohorts (D). (Scale bar = 50 μm)

Full size image
Table 3 Univariate and multivariate analyses for OS and RFS in patients with HCC in training cohort
Full size table
Table 4 Univariate and multivariate analyses for OS and RFS in patients with HCC in validation cohort
Full size table

CCL17 and CCR4 interact closely within the tumor microenvironment

We stained 873 HCC specimens with CD4, CD68, CD8, CD66b, and Foxp3 antibodies, revealing distinct immune cell types in the TME, where CD8 signifies cytotoxic T cells, CD66b is a surface marker of TAN, and Foxp3 represents Treg (Fig. 3A). We observed that high CCL17-T was correlated with increased Treg infiltration, reduced CD8 + T cell infiltration, and a higher Treg/CD8 ratio. This suggests that HCCs with high CCL17-T expression exhibited more Treg infiltration and a lack of CD8 + T-cell infiltration, implying an intimate relationship between CCL17-T and an anergic T-cell response against tumor cells (Fig. 3B). High CCL17-T expression also highly correlated with CD73 (Fig. 3B).

We proposed that CCL17, with its chemotactic function, might recruit TAN into the tumor stroma to form a suppressive barrier. Indeed, our analysis revealed that high CCL17 expression was associated with a significant increase in TAN infiltration (p < 0.01). Tumor tissues with high CCL17-T levels generally had high CCR4-I expression, suggesting that HCC cells might secrete CCL17 to recruit CCR4 + stroma cells (p < 0.001, Fig. 3B). Intriguingly, CCR4-I expression was found to be highly correlated with Treg infiltration (r = 0.4036, p < 0.001, Fig. 3B), leading us to speculate that CCR4 + CD73 + cells might be a special subset of Treg exerting an immunoinhibitory function.

Next, we further explored the relationship between CCL17, CCR4, and tumor angiogenesis in human HCC tissues. A CD34 antibody was used to stain vascular endothelial cells. Tumor tissues with high CCL17-T levels had significantly increased vascular endothelial cells genesis compared to those with low CCL17-T expression (Fig. 3B).

Subsequently, we divided the patients into two groups based on the ratio of Treg to CD8 and performed survival analysis. As illustrated in Fig. 3C, HCC patients with Treg/CD8 > 1 were associated with a higher probability of recurrence and significantly shortened survival time, which was corroborated in the validation cohort. We continued to explore the prognostic value of the specific group of cells, CCR4 + CD73+, defined earlier. Using 13 as the cutoff value, patients with higher CCR4 + CD73 + exhibited poor OS and RFS in both the training and validation cohorts.

Fig. 3
figure 3

Tumor-infiltrating immune cells and their correlation with CCL17, CCR4, CD73, as well as their impact on prognosis. (A) IHC images show positive staining for CD8 (cytotoxic T cells), CD66b (tumor-associated neutrophils), and FOXP3 (regulatory T cells), along with relevant intra-tumor negative controls. (B) High CCL17-T expression significantly correlates with increased Tregs, lower CD8 + counts, and the ratio of Tregs to CD8, as well as CCR4-I. The scatter plot indicates a close relationship between CCL17-T and CD73 protein expression. Both high CCL17-T and CCR4-T expression are associated with elevated CD34 + counts. High CCR4-I exhibits a strong correlation with high Treg infiltration. (C) Kaplan-Meier survival curves for OS and RFS of patients with HCC based on the Treg/CD8 ratio and CCR4 + CD73 + cell count. (Scale bar = 50 μm; *p < 0.05, **p < 0.01, ***p < 0.001)

Full size image

Univariate and multivariate analyses for OS and RFS in patients with HCC

Univariate Cox proportional regression analysis identified various risk factors for OS in the training cohort, including tumor number, size, differentiation, MVI, AFP, GGT, CD34, CD68, CD66b, Treg/CD8, CCR4-T, CCL17-T, CCL17-I, HHLA-2, CD73-T, and CCR4 + CD73 + cells. Factors associated with RFS were similar and included liver cirrhosis (all P < 0.050, Table 3). Multivariate analysis revealed that high CCL17-T maintained its prognostic significance for predicting both RFS [HR 1.774(1.274–2.470), P = 0.001] and OS [HR 1.818 (1.242–2.661), P = 0.002], along with tumor size, MVI, Treg/CD8, CCR4-T, HHLA-2-T, CD73-T, and CCR4 + CD73 + cells (Table 3).

In the validation cohort, tumor number, size, age, differentiation, MVI, GGT, CD34, CD66b, Treg/CD8, CCR4-T, CCL17-T, CCL17-I, HHLA-2, CD73-T, and CCR4 + CD73 + cells have been proved to be prognostic variables for both OS and RFS in univariate analysis (all P < 0.050, Table 4). Multivariate analysis revealed that high CCL17-T maintained its prognostic significance for predicting both RFS [HR 1.407 (1.117–1.771), P = 0.004] and OS [HR 1.633 (1.274–2.095), P < 0.001], along with tumor size, MVI, differentiation, CD34, Treg/CD8, CCR4-T, CCL17-I, HHLA-2-T, CD73-T, and CCR4 + CD73 + cells for OS, and age, tumor size, MVI, Treg/CD8, CCR4-T, CCL17-I, HHLA-2-T, CD73-T, and CCR4 + CD73 + cells for RFS (Table 4).

Investigating the impact of CCL17, CCR4, and CCR4 + CD73 + cell profiles on clinical outcomes in HCC

In our study, 155 patients in the training cohort and 329 in the validation cohort experienced recurrence following radical surgical therapy. Figure 4 illustrates that there was no significant difference in OS between patients who underwent post-operative TACE and those who did not, across both cohorts. Further analysis revealed that HCC patients’ OS, regardless of their CCL17-T expression levels, was not markedly improved by post-operative TACE. A similar observation was made among patients with elevated CCR4-T levels. However, an intriguing pattern emerged within a specific subset of HCC patients; those with low CCR4-T expression significantly benefited from post-operative TACE in terms of OS. This beneficial effect was also observed in patients with low CCR4 + CD73 + expression, suggesting that TACE could offer considerable clinical advantages for HCC patients characterized by either low CCR4-T or CCR4 + CD73 + profiles. These findings imply that integrating TACE with targeted anti-CCR4 and anti-CD73 therapies might provide a novel and effective strategy for enhancing the survival of post-operative HCC patients. Considering the ongoing exploration of CCR4 and CD73 inhibitors in various clinical trials, a synergistic approach that includes TACE represents a promising avenue for advancing HCC post-operative treatment.

Fig. 4
figure 4

The relationship between CCL17-T, CCR4-T, and CCR4 + CD73 + Cells and the efficacy of TACE therapy in HCC with postoperative recurrence. Significantly improved prognosis associated with low CCR4 expression and CCR4 + CD73 + < 13 in HCC patients undergoing postoperative TACE after tumor recurrence, shown in both training cohort (A) and validation cohort (B)

Full size image

Establishing new nomogram models for OS and RFS prediction

We developed a nomogram to visually represent the predicted clinical outcome, drawing from multivariate regression analyses in both the training and validation cohorts. The multivariate Cox proportional regression model identified significant predictors as independent prognostic factors for OS: CCL17-T high, CD34 high, CCR4 + CD73 + ≥ 13, CCR4-T high, CD73-T high, HHLA2, MVI, Treg/CD8 ≥ 1, and tumor size ≥ 5 cm. These factors were then utilized to create a time-based nomogram model to forecast the survival probability of HCC patients at 1-, 3-, and 5- years post-surgery (Fig. 5A). The multivariate analysis also highlighted eight significant prognostic factors for RFS. A corresponding nomogram model was developed to predict 1-, 3-, and 5-year RFS (Fig. 5B).

The model’s calibration curve showcased a strong match between the predicted and actual survival rates at 1-, 3-, and 5- year for both cohorts. Additionally, a decision curve analysis was conducted to assess the clinical value and potential benefits of our model. The findings from this curve suggested that therapeutic strategies informed by our nomogram might enhance clinical outcomes (Fig. 5C and E). Predicted outcomes from the nomogram closely matched the real RFS in both cohorts. Notably, our new nomogram models displayed superior net benefits across a broader threshold probability spectrum (Fig. 5D and F). Furthermore, our OS and RFS nomograms demonstrated a higher area under the receiver operating characteristic curve when compared to the BCLC and CLIP (Cancer of the Liver Italian Program) stage systems in both cohorts (Fig. 5G-H).

Fig. 5
figure 5

Nomogram Predictive Model for OS and RFS. (A and B) OS and RFS nomograms, respectively. (C and E) Calibration curves designed for predicting OS at 1 year, 3 years, 5 years. (D and F) Calibration curves designed for predicting RFS at 1 year, 3 years, 5 years. Nomogram-predicted possibility of survival marked on the x-axis; actual survival marked on the y-axis. (G and H) AUC curves for nomogram, BCLC, and CLIP. Dashed lines demonstrate the net benefit of the nomogram in each of the curves across a range of threshold probabilities. The horizontal solid black line represents the assumption that no patients would experience the event, and the solid gray line represents the assumption that all patients will die or relapse

Full size image

Discussion

In this study, we established a correlation between high expression levels of CCR4, CCL17, CD73, and HHLA2 in tumor tissues and the poor prognosis in HCC patients. HCC-derived CCL17 exhibited a robust relationship with CD73, present on both tumor and stroma cells, indicating its significant association with heightened CD73 expression in the TME. Interestingly, we recognized a stromal cell subtype, CCR4 + CD73 + co-expressing cells, which were closely associated with Tregs and predicted unfavorable outcomes. Patients with low CCR4 + CD73 + infiltration showed considerable improvement in OS after undergoing post-operative TACE. Moreover, patients with suppressed CCR4 expression also benefited from post-operative TACE. This data emphasizes the critical role of combining TACE with anti-CCR4 and anti-CD73 therapies in managing solid tumors. Additionally, we constructed novel nomogram models for predicting OS and RFS, enhancing comprehensive patient management and informed clinical decision-making.

Previously, CCL17 expression was only detected in peripheral blood neutrophils and TANs. In contrast, our findings revealed its expression in both tumor and stroma cells, as well as its receptor CCR4. Our results concur with previous report suggesting the stromal expression of CCL17 correlates with TANs (Zhou et al. 2016). Similarly, CCR4 expression in stroma was positively related to FOXP3 + Treg, known for inhibiting self-antigen immune responses (Maeda et al. 2019b). These observations align with those found in bladder and prostate cancer studies (Bergamaschi et al. 2020; Maeda et al. 2019b).

Our in-depth analysis indicated a strong association between CCL17 and the high Treg/CD8 ratio, indicative of an immunosuppressive environment (Bergamaschi et al. 2020; Mao et al. 2019). To substantiate our findings, we evaluated the relationship between CCL17 and various immune checkpoints, notably PD-L1, CD47, CD73, and HHLA2. Remarkably, CCL17 expression correlated robustly with CD73 but not with PD-L1, CD47, or HHLA2. With IF assays, we discovered that CCR4 and CD73 co-expressed on a unique subtype of stromal cells, which correlated highly with Treg and we termed it as CCR4 + CD73 + cells. Tregs co-express CD39, CD73 has been recognized for its pivotal role in immunosuppression and CD39-CD73-Adenosine axis is a crucial pathway in immune-related and malignant diseases, facilitating Treg induction and promoting immunosuppressive actions (Fu et al. 2016; Wang et al. 2021; Xia et al. 2023). Numerous studies have also highlighted the expression of both CCR4 and CD73 in Tregs (Huang et al. 2021; Plitas and Rudensky 2020). Building on these insights, we postulated that CCR4 + CD73 + might represent a unique Treg subtype. However, this proposition warrants further investigation. Our subgroup analysis reaffirmed that patients with elevated CCR4 + CD73 + infiltration exhibited worse OS and RFS than those with reduced infiltration, across both training and validation cohorts. This indicates that CCR4 + CD73 + Tregs may have substantial immunosuppressive capabilities, akin to other Treg types. Notably, when considering mechanisms to counteract cancer’s immune evasion tactics, CD73 emerged as more prevalent and bore a greater prognostic significance than PD-L1 and CD47 in HCC.

Recent therapies have placed considerable emphasis on anti-checkpoint treatments due to their efficacy in limiting Treg infiltration and tumor progression (Langhans et al. 2019). These include targeted therapies against significant checkpoints, such as mogamulizumab for CCR4 and CD73-siRNA delivery for CD73 (Azambuja et al. 2020; Maeda et al. 2019b). Given our findings and the extensive HCC specimen analysis, we propose that a combinatorial approach involving systematic anti-checkpoint therapy (targeting CCR4 and CD73) and TACE might offer a promising avenue for post-operative HCC management.

Although post-operative TACE is a common therapeutic strategy to prevent tumor recurrence, its benefits are limited to a subset of HCC patients (Dev et al. 2020; Wang et al. 2020a). We identified a direct correlation between CCR4 expression, CCR4 + CD73 + stromal cells in stroma, and adverse prognosis post-TACE. Hence, integrating TACE with anti-CCR4 and anti-CD73 therapies might enhance therapeutic efficacy and extend HCC patient survival significantly.

However, our study has limitations. Being a single-institution retrospective study, our nomogram models require external validation. Though our data is clinically-derived, in vitro functional experiments could augment its validity. Additionally, our study’s scope was limited to specific markers and checkpoints; further expansion is warranted.

In summary, our systematic examination of CCL17, CCR4, CD73, and HHLA2 offers a fresh perspective on refining HCC management and clinical decision-making. The identification of CCR4 + CD73 + cells as a novel stromal cell type associated with adverse prognosis provides crucial insights for post-operative TACE strategies, potentially revolutionizing HCC patient care.