CXCL1 and CXCL6 Are Potential Predictors for HCC Response to TACE

Distinct immune patterns of hepatocellular carcinoma (HCC) may have prognostic implications in the response to transarterial chemoembolization (TACE). Thus, we aimed to exploratively analyze tumor tissue of HCC patients who do or do not respond to TACE, and to identify novel prognostic biomarkers predictive of response to TACE. We retrospectively included 15 HCC patients who had three consecutive TACE between January 2019 and November 2019. Eight patients had a response while seven patients had no response to TACE. All patients had measurable disease according to mRECIST. Corresponding tumor tissue samples were processed for differential expression profiling using NanoString nCounter® PanCancer immune profiling panel. Immune-related pathways were broadly upregulated in TACE responders. The top differentially regulated genes were the upregulated CXCL1 (log2fc 4.98, Benjamini–Hochberg (BH)-p < 0.001), CXCL6 (log2fc 4.43, BH-p = 0.016) and the downregulated MME (log2fc −4.33, BH-p 0.001). CD8/T-regs was highly increased in responders, whereas the relative number of T-regs to tumor-infiltrating lymphocytes (TIL) was highly decreased. We preliminary identified CXCL1 and CXCL6 as candidate genes that might have the potential to serve as therapeutically relevant biomarkers in HCC patients. This might pave the way to improve patient selection for TACE in HCC patients beyond expert consensus.


Introduction
Primary liver cancer is one of the leading causes of cancer death worldwide and hepatocellular carcinoma (HCC) comprises more than two thirds of cases [1,2]. The therapeutic strategy is complex and usually evaluated in interdisciplinary tumor board meetings with consensus-based decision-making [3]. Patients are frequently multimorbid, suffering from chronic liver disease with late diagnosis of HCC, limiting the potential treatment modalities. Localized disease may be treated by locoregional therapies such as embolization, microwave-, or radiofrequency ablation if the patient does not qualify for resection or transplantation [4]. Transarterial chemoembolization (TACE) is a leading option in unresectable localized HCC [4]. In conventional TACE, chemotherapeutic agents such as mitomycin C as well as an embolization agent such as ethiodized oil are injected into the tumor feeding artery +/− additional embolization agents such as microsphers [5,6]. TACE aims to induce Curr. Oncol. 2023, 30 3517 tumor ischemia via blocking of the blood supply and enhanced antitumoral chemotherapeutic effects via retention of chemotherapeutics within the tumor environment [6]. TACE is based on the arterial hypervascularity of HCCs, which can be visualized in radiology as the imaging hallmark: arterial hyperenhancement and venous/delayed wash-out [2]. Comparative high-quality multicenter TACE trials remain scarce as the clinical practice and TACE techniques vary globally [2,6]. Of note, performing TACE in ineligible patients may cause harm and reduced overall survival [2]. Therefore, it is of utmost importance to improve the selection of patients who will most likely benefit from TACE. Efforts were made to develop clinical and laboratory scores for TACE patient selection [2,[7][8][9][10][11]. However, the scores are not advised to be used outside clinical trials [2]. Consequently, the patient selection is still based on a low level of evidence, namely expert consensus.
The immunology of HCC could reveal promising targets for improved patient selection. Inflammation such as hepatitis increases the incidence of HCC following an inflammatory-based pathogenesis. The HCC immune landscape comprises a multitude of different cell types, immune cells, immune receptors, ligands, chemokines, or cytokines [12]. Tumor-associated neutrophils, myeloid-derived suppressor cells, or regulatory T-cells can modulate the tumor and immune environment with pro-or antitumorigenic effects [12]. Recent developments in immunotherapy showed promising results, prolonging survival in patients with sensitive tumors [12,13]. However, so far, studies investigating distinct immune features in primary tumor samples that predict response to TACE in HCC patients are lacking.
We hypothesized that response to TACE may depend on different HCC immune patterns. Hence, the aim of this study was to exploratively analyze HCC tissue of TACE responders and non-responders to identify potential biomarker candidates. Those candidates could pave the way to improved evidenced-based clinical decision-making in the future beyond clinical expert consensus.

Database
Tissue samples and patient data used in this study were provided by the University Cancer Center Frankfurt (UCT). Histopathological confirmation of HCC was assessed by expert pathologists of the Dr. Senckenberg Institute of Pathology, University Hospital Frankfurt, Frankfurt am Main, Germany. Written informed consent was obtained from all patients and the study was approved by the Institutional Review Boards of the UCT and the Ethical Committee at the University Hospital Frankfurt (project-number: SGI-10-2020).

Conventional TACE and Tumor Response Assessment
Patients were treated with cTACE in clinical routine as described in previous studies [14]. Modified response evaluation criteria in solid tumors (mRECIST) was used to assess response to TACE [15]. Complete (CR) or partial response (PR) was defined as response. Stable or progressive disease was defined as no response.

Ribonucleic Acid (RNA) Isolation and Immune Profiling Analysis
Pre-TACE formalin-fixed paraffin-embedded (FFPE) tissue samples were retrieved from the archive of the Dr. Senckenberg Institute of Pathology, University Hospital Frankfurt. Representative tumor material of the primary tumor was retrieved by punching out a 1 mm core. RNA was isolated using the truXTRAC FFPE total NA Kit (Covaris, Woburn, MA, USA) based on focused ultrasonication and column purification according to the manufacturer's instructions. NanoString nCounter ® Platform and PanCancer immune profiling panel were used to enrich a commercially available function-specific panel of 770 genes by hybrid capture technique (NanoString, Seattle, WA, USA). The PanCancer immune profiling panel includes genes from different immune cell types (among others, B cells, T cells, NK cells); common checkpoint inhibitors; CT antigens; and genes covering both the adaptive and innate immune response. NanoString nSolver ® software v4 and implemented nCounter ® advanced analysis module v2.0.134 were used for subsequent raw data processing and normalization by internal controls following differential supervised analysis between TACE responders (n = 8) and non-responders (n = 7) according to mRE-CIST. Quality control was performed with default settings as previously described [16]. One sample (TACE responder) was flagged in quality control as it exceeded the threshold of the binding capacity, and we excluded this sample for further analysis. Gene expression of TACE non-responders was set as the baseline for the comparative analysis. For pathway analysis of differentially expressed genes, Enrichr [17,18] was used for functional enrichment analysis for Gene Ontology to identify gene sets for biological processes. For further differential expression analysis, we used the following cut-offs: log2 fold change ≤−2 or ≥2 and p < 0.05 after Benjamini-Hochberg (BH) correction. T-distributed stochastic neighbor embedding (t-SNE) analysis and plots were performed in Python 3.7.6. [19].

Study Population
In total, 15 patients met the inclusion and exclusion criteria ( Figure 1). The STARDflowchart of patient inclusion is shown in Figure 1. Our data set comprised seven TACE non-responder (median age, 66 (54-83); men, five) and eight TACE responders (median age, 65.5 (55-75); men, seven). The groups did not differ significantly between clinical characteristics including the size of the target lesions, HCC etiology, Barcelona clinic liver cancer (BCLC) stage, Child-Pugh score, model for end-stage liver disease (MELD) score, or selected laboratory values such as alpha-fetoprotein (AFP), albumin, or bilirubin (Table 1). Further patient characteristics are shown in Table 1.

Unsupervised Clustering
Unsupervised t-SNE plots of all normalized mRNA data reveal two distinct groups of responders and non-responders ( Figure 2a). NanoString pathway score analysis uses functionally annotated genes with subsequent hierarchical cluster analysis for dimensionality reduction. However, this approach did not clearly separate responders from non-responders ( Figure S1). Importantly, responders reveal a general trend of upregulation in the majority of pathways; downregulation is only seen in the pathway scores describing complement and cancer/testis (CT) antigen ( Figure 2b).

TACE Responders Revealed Genetic Subgroup Signatures
All patients are included for preliminary expression analysis in dependence of TACE response status and are depicted in volcano plots for further analyses and subsequent selection for further functional classification. TACE responders differ in immune responserelated genes (Figure 3a). A total of 25 genes are strictly deregulated (log2 fold-change ≤ −2 or ≥2, BH-p < 0.05) in TACE responders (Figure 3b). A total of 92% (23/25) of significantly deregulated genes were up-regulated in TACE responders and only two genes (8%, 2/25) were down-regulated ( Table 2). The top three deregulated genes are the upregulated CXCL1 (log2fc 4.98, BH-p < 0.001) and CXCL6 (log2fc 4.43, BH-p = 0.016), as well as the downregulated MME (log2fc −4.33, BH-p = 0.001). The majority of genes (28% (7/25)) are associated with chemokines and all of them are upregulated in responders.

Gene Ontology Term Enrichment Analysis and Cell Type Profiling
Using Enrichr gene ontology term enrichment on the whole differentially expressed data, the biological processes were mostly related to cytokine-cytokine receptor interaction of chemokines (CC subfamily, CXC subfamily), the class I helical cytokines (mainly y-chain utilizing), and TNF family ( Figure S2), as well as chemokine signaling pathways involving JAK 2/3, Src, PI3K, Itk, and PKC ( Figure S3). Next, the nCounter ® advanced analysis module was used to measure the abundance of cell types whose expression is largely specific to certain immune cell populations [20,21]. The vast majority of immune infiltrates are increased in TACE responders (Figure 4a). The total number of tumor-infiltrating lymphocytes is increased in TACE responders. Of note, the relative number of CD8-cells to T-regs is highly increased in responders whereas the relative number of T-regs to TILs is highly decreased (Figure 4b).
significantly deregulated genes were up-regulated in TACE responders and only two genes (8%, 2/25) were down-regulated ( Table 2). The top three deregulated genes are the upregulated CXCL1 (log2fc 4.98, BH-p < 0.001) and CXCL6 (log2fc 4.43, BH-p = 0.016), as well as the downregulated MME (log2fc −4.33, BH-p = 0.001). The majority of genes (28% (7/25)) are associated with chemokines and all of them are upregulated in responders.   number of CD8-cells to T-regs is highly increased in responders whereas the relative number of T-regs to TILs is highly decreased (Figure 4b).

Figure 4.
Cell type profiling reveals increased immune infiltrates in responders. The nCounter  advanced analysis module of the PanCancer immune profiling panel uses genes whose expression is largely specific to certain immune cell populations to measure the abundance of these cell types. The module assumes that each cell type's characteristic genes are expressed exclusively and consistently within the cell type. Under this model, a cell type's abundance can be measured as the average log-scale expression of its characteristic genes. The cell type abundance measurements are plotted against response. True = responder, false = non-responder.

Discussion
In this study, we used NanoString technology to exploratively analyze different immune patterns of HCC patients who were classified as responder and non-responder to repetitive TACE. Patients' response to TACE was evaluated using mRECIST response assessment. We used pre-TACE tissue samples of the primary tumor and performed a multitude of NanoString technology analyses such as pathway scoring, gene set enrichment analysis, differential expression analysis, and cell type profiling to build a holistic model of different HCC immune signatures that might be associated with response to TACE. We revealed that the immune pathway scores in TACE responders are type scores are shown in responders (true) and non-responders (false). The nCounter ® advanced analysis module of the PanCancer immune profiling panel uses genes whose expression is largely specific to certain immune cell populations to measure the abundance of these cell types. The module assumes that each cell type's characteristic genes are expressed exclusively and consistently within the cell type. Under this model, a cell type's abundance can be measured as the average log-scale expression of its characteristic genes. The cell type abundance measurements are plotted against response. True = responder, false = non-responder.

Discussion
In this study, we used NanoString technology to exploratively analyze different immune patterns of HCC patients who were classified as responder and non-responder to repetitive TACE. Patients' response to TACE was evaluated using mRECIST response assessment. We used pre-TACE tissue samples of the primary tumor and performed a multitude of NanoString technology analyses such as pathway scoring, gene set enrichment analysis, differential expression analysis, and cell type profiling to build a holistic model of different HCC immune signatures that might be associated with response to TACE. We revealed that the immune pathway scores in TACE responders are upregulated in all but two pathways. A total of 92% of the genes with the highest deregulation are upregulated in responders and the top deregulated genes are the upregulated CXCL1 and CXCL6, as well as the downregulated MME. The genes are strongly associated with the biological processes of cytokine-cytokine receptor interactions and chemokine signaling. The pathway alterations are corroborated by increased scores of tumor-infiltrating lymphocytes in responders with high CD8/T-reg and low Treg/TIL ratios.
HCC commonly arises in the background of cirrhotic liver, often as a consequence of chronic inflammation caused by viral hepatitis, alcohol abuse, metabolic disease, or exposure to a variety of substances [12]. Inflammation, as a crucial step in HCC development, is not only corroborated by reduced HCC incidence in people who are vaccinated against viral hepatitis, but also in the increased application of systemic immuno-oncological therapy [12,22]. Of note, several trials and therapeutic approaches combine immune therapy with locoregional therapies [22]. The high immunological activity in the liver is reflected in CD8 T-cells, which are major targets in immune therapy. It is known that locoregional therapies such as TACE lead to the release of tumor-associated antigens (TAAs). TAAs play a crucial role in the activation pathway of CD8 cells and their anti-tumoral activity [22]. This is in line with our results, as TACE responders reveal increased levels of CD8+ T-cells relative to T-regs. Furthermore, the number of TILs is known to reduce the rate of tumor recurrence and to prolong survival [22,23]. We also reveal an increased total number of TILs in TACE responders. Not only antitumoral cells can be found infiltrating the liver or cancerous tissue, but also immunosuppressive and, therefore, tumor-promoting cell types. Tregs lead to an immunosuppressed milieu, promoting tumor progression [22]. In our study, we reveal a strong decrease in Tregs/TILs ratio in responders, which is in line with the immuno-oncological concept. Our results suggest that pretherapeutic immune patterns may influence the response to TACE. This is of additional importance, taking into account the fact that ablative locoregional therapy can increase the spread of tumor antigens for oncolytic treatment and improve response in combination with immunotherapy [22]. Furthermore, our results show that not only response to immunotherapy but also response to the locoregional therapy itself seems to be influenced by different immune signatures in HCC. In this study, TACE responders have elevated levels of numerous members of the CCand CXC-subfamily of chemokines, with the most prominent levels of CXCL1 and CXCL6. CXCL1 is known to promote neoplastic transformation, tumorigenesis, and angiogenesis in multiple cancer types by binding CXCR2 (CXCR2 was also increased in our analysis, data not shown). In addition, CXCL1 was described to be secreted to promote immune cell recruitment and to alter the immune milieu in liver cancer [24]. CXCL1 was further described as a factor of HCC aggressiveness, promoting proliferation and invasion, while our results indicate that CXCL1 overexpression may increase TACE response. High CXCL6 expression is described as being associated with angiogenic effects and poor prognosis in HCC [25]. CXCL6 is upregulated in TACE responders, potentially as an effect of its angiogenic ability, increasing the effect of chemoembolization. MME is a known factor in cancer cell-cell signaling and it was described to be altered by comparing pre-neoplastic tissue vs. HCC [26]. In our study, the chemokine signaling leads to the activation of JAK2/3 and PI3K. JAK and PI3K are part of signaling pathways contributing to HCC development and progression, while JAK2 inhibition has been shown to induce growth arrest in HCC cell lines [27]. However, our study has several limitations that warrant discussion. We performed a retrospective analysis and selection bias cannot be ruled out. With 15 patients (8 responders, 7 non-responders), our study population is very small, and generalizability may not be presumed. However, we followed a rigorous study design of patient inclusion and exclusion. Next, in the vast majority of cases in this setting, tissue is only available from biopsies, which strongly hinders sufficient RNA isolation due to the sample size. Thus, unstandardized interventional procedures and the limited availability of high-quality tissue specimens did not allow for a higher number of samples in our explorative study. Furthermore, this strongly limits the opportunity to perform additional immunohistochemistry as verification on protein level. Given these considerations, and the fact that data on this topic are lacking so far, the results of the present study are of high clinical interest.
Our aim was to exploratively analyze different HCC immune patterns using NanoString technology. However, due to the retrospective nature of this study and limited tissue availability, confirmatory analysis such as qPCR, in vitro, and in vivo experiments should be addressed in further prospective studies.

Conclusions
In summary, HCC immune signatures may reveal biomarkers for TACE response prognostication in HCC patients. Chemokines, and especially the CC and XCX subfamily with CXCL1 and CXCL6 as two leading representatives, seem to be promising response biomarkers. Our preliminary study sheds light on the treatment-influencing relevance of the HCC immunology for TACE response beyond systemic therapy. However, our results should not be overstated, as they were derived following an explorative approach and they need to be corroborated in further multicenter studies.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/curroncol30030267/s1, Figure S1: Unsupervised clustering using NanoString pathway score analysis; Figure  The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; prepartion, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Boards of the UCT and the Ethical Committee at the University Hospital Frankfurt (project-number: SGI-10-2020).
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data are available upon reasonable request from the corresponding author.