Construction of a radiogenomic signature based on endoplasmic reticulum stress for predicting prognosis and systemic combination therapy response in hepatocellular carcinoma

Construction of a radiogenomic signature based on endoplasmic reticulum stress for predicting prognosis and systemic combination therapy response in hepatocellular carcinoma | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Construction of a radiogenomic signature based on endoplasmic reticulum stress for predicting prognosis and systemic combination therapy response in hepatocellular carcinoma Huai-Yu Wu, Shu-Ya Cao, Zheng-Gang Xu, Tian Wang, Gu-Wei Ji, Ke Wang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4535127/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background: Hepatocellular carcinomais one of the most common tumors worldwide. Various factors in the tumor microenvironment can lead to the activation of endoplasmic reticulum stress, thereby affecting the occurrence and development of tumors.The objective of our study was to develop and validate a radiogenomic signature based on endoplasmic reticulum stress to predict prognosis and systemic combination therapy response Methods: Using data from TCGA as a training cohort and data from ICGC as a testing cohort. Univariate Cox regression and multivariate Cox regression analysis were used to identify prognostic-related genes and construct a model. Hepatocellular carcinoma single-cell data obtained from GEO were used to map gene signatures and explore inter-cellular signaling communications. Finally, a radiogenomic signature was used to predict the objective response rate and overall survival. Results: A total of four gene signatures related to ERS (including STC2, MAGEA3, BRSK2, DDX11) were identified. Macrophages were significantly different between high-risk and low-risk groups. The high-risk group showed higher PD-1 and mutated TP53 scores. Drug sensitivity analysis found that most sensitive drugs target the PI3K/mTOR signaling pathway. Further research revealed the expression of STC2 in the endothelial cells, particularly PLVAP+ endothelial cells, and may regulate the reprogramming and function of macrophages. Furthermore, we identified nine radiomic features and established a radiogenomic signature based on ERS that can predict prognosis and response to systemic combination therapy. This signature can guide systemic combination therapy for patients with unresectable hepatocellular carcinoma. Conclusions: We established an ERS prognostic model that can predict patient prognosis. We also found that ERS is closely related to the tumor microenvironment and is mainly manifested in the interaction between tumor-associated endothelial cells and tumor-associated macrophages. Moreover, we constructed a radiogenomic signature based on the ERS. This signature can guide subsequent combination therapy for patients with unresectable HCC. radiogenomics endoplasmic reticulum stress (ERS) hepatocellular carcinoma (HCC) The Cancer Genome Atlas（TCGA） radiomics single-cell Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Liver cancer is currently the sixth most common malignant disease and the third leading cause of cancer death globally[1]. Hepatocellular carcinoma (HCC) accounts for 75–85% of liver cancer cases[2]. Traditional chemotherapy is ineffective against HCC[3], with liver resection and transplantation being the primary treatment options[4]. Immune checkpoint inhibitors targeting programmed cell death-1 (PD-1) based on the tumor immune environment have been approved by the FDA as second-line therapies for advanced HCC[5]. While some patients show significant responses to immunotherapy, approximately 20% to 30%, others exhibit resistance[6], likely due to immunosuppression and immune evasion orchestrated by the tumor microenvironment. Previous studies have shown a positive association between endoplasmic reticulum stress (ERS) and the onset and progression of various liver disorders, including cancer[7-10]. Hypoxia, nutrient deficiency, and rapid protein increase in the tumor microenvironment inevitably lead to the occurrence of ERS[11]. Three different branches of the ERS signaling pathway are involved in chemoresistance, immunosuppression and the induction of immune evasion through different mechanisms[9, 11-14]. Therefore, the abnormal activation of ERS sensors and their downstream signaling pathways has emerged as a critical regulator of tumor development and metastasis, suggesting potential targets for cancer therapy. Radiomics and radiogenomics are emerging fields with a wide range of potential applications. Radiogenomics allows for the non-invasive acquisition of genetic information about tumors[15], which is a more advanced approach than traditional methods involving surgical resection for pathological specimen retrieval. These methods can guide subsequent systemic therapy for HCC patients who are not eligible for surgery. Radiomics is increasingly important in personalized cancer treatment, as quantitative features extracted from radiographic medical images provide valuable high-dimensional data that can enhance clinical decision-making and improve diagnostic and prognostic capabilities[16, 17]. Recent studies have highlighted the diagnostic and prognostic potential of radiomics, particularly in the analysis of tumor and peritumoral regions in HCC, offering insights into the tumor microenvironment[18-20]. Additionally, radiomics shows promise in predicting gene expression prior to surgery, potentially altering prognosis assessments and treatment decisions for cancer patients undergoing surgical procedures[21]. The objectives of our study were to develop and validate a model focused on genes associated with ERS and explore the connection between ERS and the tumor microenvironment. Moreover, radiomics has been used to construct a radiogenomic signature to non-invasively predict the prognosis of surgical patients and the therapeutic response of patients treated with systemic combination therapy. Methods and Materials Study sample The training set of bulk RNA sequencing data and associated clinical information for HCC patients (TCGA-LIHC, n=424) were obtained from the TCGA database. Five tumor samples with missing clinical information were excluded, and 419 samples with complete information were included. The validation set of bulk RNA sequencing data and clinical information (ICGC-JP-LIRI, n=240) were downloaded from the ICGC database. To guarantee a more accurate depiction of gene expression, the bulk data were then transformed into TPM (transcripts per million reads) format. The 253 ERS-related genes were obtained from the Molecular Signatures Database. Additionally, the corresponding somatic mutation data were obtained from the TCGA database. The HCC scRNA-seq datasets GSE156625 (n=58), GSE162616 (n=3), GSE189175 (n=3) and GSE210679 (n=2) were downloaded from the GEO database. For the radiomic analysis, we constructed a radiogenomic signature based on the TCIA cohort (43 patients with complete imaging information and clinical data) from the Cancer Imaging Archive. The surgical therapy cohort consisted of 236 patients who were recruited from our center between 2009 and 2017. The inclusion criteria were as follows: (1) Patients who could undergo R0 resection of HCC. (2) Enhanced CT images before surgery. (3) Patients who did not receive neoadjuvant treatment before surgery. (4) Complete follow-up data. In addition, the systemic combination therapy cohort consisted of 40 patients who received anti-PD-1 plus TACE and anti-VEGF therapy between 2019 and 2023. The inclusion criteria were as follows: (1) At least one systemic combination therapy. (2) No systemic combination therapy was administered before baseline imaging. (3) Enhanced CT images before treatment. (4) Systemic combination therapy for primary HCC or recurrence of HCC. ERS model establishment and validation Differential gene expression was analyzed using the R package “edgeR”, and 35 differentially expressed genes were used in the subsequent analysis. Univariate Cox regression analysis was utilized to investigate the relationship between prognosis and ERS-associated gene expression levels in patients with HCC. Variables with P values less than 0.05 in univariate analysis were entered into multivariate Cox analysis using the stepwise backward selection method. The "survival" package in R was used for all the statistical analyses involving Cox regression, and the results were visualized using the “forest plot” R package. The risk score was calculated as β1 × Exp1 + β2 × Exp2 + β3 × Exp3+βx × Expx. According to the cut-off risk score, the same standard was used to classify patients in the training and validation cohorts into high-risk and low-risk groups. Additionally, Kaplan–Meier survival curves revealed differences in prognosis between the high-risk and low-risk groups. Multi-scale analysis based on bioinformatics To gain a comprehensive understanding of the relationship between ERS and the tumor immune microenvironment, we used CIBERSORT, which calculates tumor-infiltrating immune cell compositions using deconvolution analysis. Subsequently, we divided the samples into high-risk and low-risk groups to compare the immune cell infiltration levels between them. Single-sample gene set enrichment analysis (ssGSEA) was used to assess the correlation between gene signatures and immune infiltration via the “gsva” R package. The “Maftools” package in R was utilized to analyze and compare the quantity and quality of gene mutations between the high-risk and low-risk groups. We collected drug screening data from the Genomics of Drug Sensitivity in Cancer database for drug sensitivity analysis. The drug sensitivity of patients in the TCGA-LIHC cohort and ICGC-JP-LIRI cohort was estimated via the “OncoPredict” R package. ScRNA-seq analysis was performed in R using the package “Seurat”. First, the FindIntegrationAnchors and IntegrateData functions in Seurat were used for the integration of these HCC scRNA-seq data. Subsequently, cells that failed to meet the prescribed quality criteria were removed from the integrated dataset. These criteria included the following: (1) 500<nFeature_RNA<2500; (2) 200<nCount_RNA; (3) percent.mt < 5. As a result, a comprehensive dataset composed of 100073 cells was obtained. Following this, the analysis proceeded using the standard Seurat workflow. The CellMarker database and the “SingleR” R package were used for cell type annotation. The interaction patterns between cells in the tumor microenvironment were analyzed using the “cellchat” R package. Their position in the cellular communication network was determined using the ligand‒receptor database. Radiogenomic signature establishment and validation We constructed a radiogenomic signature based on the TCIA cohort using 3DSlicer software to construct an outline of the regions of interest (ROIs). Next, we validated the signature using surgical therapy cohort and systemic combination therapy cohort. This work was performed by two radiologists who operated under a double-blind protocol to ensure unbiased results. Additionally, reader 1 repeated the work with an interval of at least 1 month following the same procedure. 883 features were extracted from the artery and vein phases respectively through PyRadiomics, which was built into 3DSlicer. A total of 1766 features were retained after excluding features without statistical data. Subsequently, intragroup consistency analysis was performed on the features of the 43 patients drawn by reader 1, and intergroup consistency analysis was performed on the features of the same 43 patients drawn by readers 1 and 2. The features whose ICC＞0.9 were included in the following procedure. Then, we applied Z score normalization to ensure that the radiomic features were measured on the same scale, and Levene’s test was used to verify the homogeneity of variances. We employed the maximum relevance minimum redundancy (mRMR) to retain the most pertinent and reproducible radiomic features for subsequent analyses. Internal validation was performed using the 10-fold cross-validation method. We employed the Lasso algorithm to filter out radiomic features to construct the signature and classify samples. Next, we investigated the value of the signature for the prediction of overall survival (OS) and systemic combination therapy response. Systemic combination therapy responses include complete response (CR), partial response (PR), stable disease (SD), progressive disease (PD), objective response rate (ORR), and disease control rate (DCR) (evaluated every 6~8 weeks) according to the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1. Statistical analysis R version 4.1.1 was used for the statistical analysis. Categorized variables between different risk groups were compared by the Wilcoxon t test, and the log rank test was used to test for significant differences in survival probability between samples, with a P value<0.05 indicating statistical significance. Pearson analysis was used to calculate correlations if the data were normally distributed; otherwise, the Spearman rank test was used. Results Establishment and validation of the ERS model By comparing the gene expression profiles of 374 tumor tissues and 50 normal liver tissues from the TCGA database, we identified 35 differentially expressed ERS-related genes, including 24 upregulated genes and 11 downregulated genes (Fig. 2a) in tumor tissues. Using univariate Cox regression, nine ERS-related genes (p<0.05) associated with OS were found among the 35 DEGs (Additional Table 1). Following the application of multivariate Cox regression analysis based on these ERS-related genes, a prognostic model was developed by defining four sERS-related genes (Fig. 2b). The format for calculating the risk score was as follows: ERSrisk=0.24550*STC2+0.27670* DDX11+0.11410* MAGEA3+0.38278* BRSK2. The ERS risk was selected as the cutoff value for the ROC curve, and the patients were separated into high-risk and low-risk groups. Calibration curves were used to evaluate the discrimination and calibration of our model (Additional Fig. 1). The ERS model C-indexes were 0.646 in the TCGA-LIHC cohort and 0.653 in the ICGC-JP-LIRI cohort. K‒M survival analysis revealed that patients in the high-risk group had a poor prognosis, while those in the low-risk group had a relatively good prognosis (Fig. 2c). To further assess the predictive ability of the model for HCC, a stratified analysis was conducted based on pathological and clinical characteristics such as TNM stage, age, sex, and tumor stage (Additional Fig. 2a). Patients with a high ERSrisk exhibited a greater likelihood of early mortality than did those with a low ERSrisk (Additional Fig. 2b). The area under the curve (AUC) values for 1-, 3-, and 5-year OS predictions were superior to those of other clinical characteristics (Fig. 2d). Variations in ERSrisk were observed between T1 and T2~T4, as well as between stage I and stage II~III, but not in advanced HCC (Fig. 2e). Next, we performed a comparative analysis of mutations in the high-risk and low-risk groups (Additional Fig. 3a and b). We found that the most significant difference in ERSrisk was TP53. The ERSrisk of the mutant type was significantly greater than that of the wild type (Additional Fig. 3c). Moreover, the expression levels of MAGEA3 and DDX11 were increased (Additional Fig. 3d), which suggested that ERS may be closely related to TP53 mutation. In addition, we performed an intergroup comparative analysis of 198 drugs. After cross-validation with the TCGA and ICGC databases, we found that patients in the high-risk group were more sensitive to five drugs, namely, ubrosertib, taselisib, pictilisib, OSI.027, and doramapimod (Additional Fig. 4a), but more resistant to 12 drugs, namely, BI.2536, ERK_2440, ERK_6604, mitoxantrone, PD0325901, RO.3306, SB216763, SCH772984, selumetinib, topotecan, trametinib, and VE.822 (Additional Fig. 4b). Interestingly, four of the five sensitive drugs targeted the PI3K/mTOR signaling pathway; however, most of the resistant drugs targeted the ERK/MAPK signaling pathway (Table 1). This meant that even if they were all proliferation-related drugs, patients with high ERSriskERS risk were more likely to receive drugs targeting the PI3K/mTOR signaling pathway rather than the ERK/MAPK signaling pathway. ERS model correlated with macrophages and endothelial cells Furthermore, we analyzed the differences in infiltrating immune cells between the high-risk and low-risk groups. The results revealed significant differences in a variety of immune cells, including T cells, M0 macrophages, M1 macrophages, dendritic cells, neutrophils, and mast cells. The number of M0 macrophages in the high-risk group was significantly greater than that in the low-risk group, while the number of M1 macrophages was less than that in the low-risk group. Similar results were also observed in the ICGC-JP-LIRI cohort (Fig. 3a). Moreover, ssGSEA was used to assess the correlation between immune cells and ERSrisk, and it was found that ERSrisk was significantly related to macrophages (Additional Fig. 5); therefore, we believe that ERS is closely related to macrophages in the tumor immune microenvironment of HCC. Additionally, we evaluated the relationship between immune checkpoints and ERS. The response to PD-1/CTLA4 immunotherapy in patients with different immunophenotypes was compared between the high-risk and low-risk groups in the TCGA-LIHC cohort. The low-risk group had a greater non-response rate to anti-CTLA4/PD-1 immunotherapy, suggesting that the high-risk group may have a better response to anti-CTLA4/PD-1 therapy (Fig. 3b). Moreover, the TIDE score of the high-risk group was lower than that of the low-risk group, while the PD-L1 score was higher than that of the low-risk group (Fig. 3c), which indicated that the high-risk group had a lower risk of evading immunotherapy and may be more effective in immune checkpoint treatment. We integrated the single-cell data from GEO and removed batch effects. After a standardized procedure, all cells were divided into 23 clusters. Cell type annotations were verified through the “singleR” R package and cellmarker2 database (Additional Fig. 6). The 23 clusters were labeled with eight cell types that were assigned based on the expression of cell type-specific markers: T cells, NK cells, endothelial cells, malignant cells, myeloid-derived cells, HSCs, B cells, and epithelial cells (Fig. 4a). Subsequently, the gene tag was strategically positioned, leading to the identification of significant expression of STC2 within the endothelial cell population (Fig. 4b). Consequently, the endothelial cell population was subsequently extracted, followed by in-depth subpopulation analysis. The endothelial cells were partitioned into ten distinct clusters and subsequently classified into five subtypes: PLVAP+ECs, CD9+ECs, IGFBP3+ECs, PLPP3+ECs, and TFF3+ECs (Fig. 4c; Additional Fig. 7), with predominant expression of STC2 observed in PLVAP+ ECs (Fig. 4d). Since we detected significant differences in macrophages between the high-risk and low-risk groups, we reclassified myeloid-derived cells and divided them into six cell types: pDCs, DC1s, DC2s, TAM1s, TAM2s, and monocytes (Fig. 4e; Additional Fig. 8). However, the gene was not significantly expressed in the macrophages. Considering the differences in the expression of gene tags, we visualized the signaling pathways involving PLVAP+ ECs and found that they were closely related to each cell population (Fig. 4f). Interestingly, the THY1 signaling pathway only existed in PLVAP+ ECs and was not found in other cell populations, especially those closely related to TAM1 (Fig. 4g), which may indicate that there was a unique interaction between PLVAP+ ECs and TAM1. This may indicate that STC2 is involved in the reprogramming of endothelial cells and may regulate the reprogramming and function of macrophages through PLVAP+ECs. The above results suggested that ERS plays a role in endothelial cells, thereby affecting the function of macrophages. Establishment and validation of the radiogenomic signature Although we already understand the important role of ERS in the occurrence and development of HCC and the TME, we hope to make early judgments on patients through non-invasive methods to better guide subsequent treatment. Utilizing resources from the TCIA databases, we embarked on a comprehensive analysis to discern specific radiological features that could predict ERSrisk. 1,766 original image features were extracted from the ROIs. After intragroup consistency analysis and intergroup consistency analysis, a total of 1,162 features whose ICC> 0.9 were retained. Thirty features were identified from the training cohort after Spearman’s correlation analysis and mRMR. The LASSO Cox regression algorithm further narrowed down a fusion signature that retained nine archetypal features (Fig. 5a and b). The resulting signature was formulated as follows (Fig. 5c): radscore= AP-wavelet-LHL-firstorder-Skewness*0.034045 -VP-wavelet-HHH-firstorder-TotalEnergy*0.06068 +VP-wavelet-HLL-glcm-Correlation*0.045428 -AP-wavelet-LLH-glcm-Idmn*0.038591 +AP-wavelet-HLL-gldm-DependenceVariance*0.032164 -AP-wavelet-LLH-glrlm-RunVariance*0.061446 -AP-wavelet-LHL-glcm-Imc2*0.012521 +VP-wavelet-LLL-firstorder-RootMeanSquared*0.052823 -VP-wavelet-LLH-glcm-Correlation*0.058891 The C-index of the radiogenomic signature was 0.647 in the TCG-LIHC cohort. The radiogenomic signature was significantly associated with OS in the TCGA-LIHC cohort (Fig. 5d). Multi-scale validation of the radiogenomic signature We used a surgical therapy cohort to validate the radiogenomic signature and obtained good results, with a C-index of 0.607. The prognosis of the RDS-low group was significantly better than that of the RDS-high group (Fig. 6a). Our study subsequently assessed the relationships between the radiogenomic signature and systemic combination therapy response(Fig. 6b). Interestingly, as shown in Fig. 6c and d, we found that patients in the RDS-low group (45%) had a greater objective response rate than did those in the RDS-high group (5%). In addition, the PD rate was greater in the RDS-high group than in the RDS-low group, which was also confirmed in the entire cohort (Table 2). Next, we explored the ability of the radiogenomic signature to predict systemic combination therapy response. We found that the signature had excellent performance in predicting the ORR, with an AUC of 0.785 and the DCR, with an AUC of 0.768 (Fig. 6e). Finally, Kaplan–Meier plots of progression-free survival (PFS) confirmed that patients in the RDS-low group had a longer PFS than those in the RDS-high group (Fig. 6f). Discussion In our study, four independent risk factors for HCC prognosis, namely, STC2, BRSK2, MAGEA3, and DDX11, were identified based on ERS, and a new prognostic model for HCC was established based on these four genes. Based on this model, the ERSrisk of each sample was calculated separately by dividing the TCGA-LIHC cohort into high-risk and low-risk groups. Furthermore, we discovered that there were notable differences in tumor immune infiltrating cells, primarily macrophages, between the high-risk and low-risk groups. In fact, ERS plays a complex role in the TME[22]. TAMs participate in IRE1-dependent ERS, which promotes cathepsin secretion and promotes a prometastatic phenotype through the synergistic effect of IL-4, IL-6, and IL-10[23]. These results indicate that ERS is involved in the regulation of TAMs in the TME. We combined GEO single-cell data and separated the data into eight categories to investigate the connection between ERS and the TME in HCC in more detail. We discovered that endothelial cells, but not macrophages, express STC2 at a relatively high level, which seems to contradict our previous results. This finding implies that endothelial cells serve as a bridge between ERS and TAMs. Endothelial cells with activated ERS participate in the distribution and functional regulation of TAMs. An increasing number of studies have focused on tumor-associated endothelial cells and revealed numerous important proangiogenic molecules, such as vascular endothelial growth factor (VEGF), which may be targets for antiangiogenic therapy[24]. Previous studies have suggested a key function for tumor-associated endothelial cells in the growth of malignancies[25, 26]. The expression of VEGF and the development of new blood vessels can be promoted in the tumor microenvironment by conditions such as hypoxia and nutritional deprivation, which can trigger ERS and activate the three traditional signaling pathways of the unfolded protein response[27]. In addition to being controlled by the PERK/ATF4 signaling pathway[28], HIF can also promote the expression of STC2 in the hypoxic environment of solid tumors[29]. STC2 can promote angiogenesis by both affecting the expression and phosphorylation of VEGFR2 and participating in the VEGFR2/AKT/eNOS axis[30]. Through further analysis of endothelial cell subpopulations, we found that STC2 was present mainly in PLVAP+ECs. Interestingly, PLVAP+ECs are considered fetal liver-associated ECs, while PLVAP is known to play an important role in liver development and disease[31-33]. The expression of PLVAP is regulated by a variety of substances and mechanisms, among which VEGF plays a key role. VEGF regulates PLVAP expression in a VEGF receptor 2 (VEGFR2)-dependent manner. The PI3K/p38MAPK and MEK1/ERK1/2 pathways are two downstream cascades that can be triggered by VEGF signaling via VEGFR2[34, 35]. STC2 is involved in the phosphorylation of ERK1/2 and VEGFR2, as well as the activation of PI3K. PI3K can initiate multiple cascades and has multiple downstream effects, including p38MAPK and Akt. These substances are believed to be involved in the regulation of PLVAP[34-36]. Surprisingly, this finding is similar to our drug sensitivity results in that the high-risk group was more sensitive to PI3K- and p38-targeted drugs. Based on these observations, it is highly likely that STC2 regulates PLVAP expression via a variety of pathways, including the VEGFR2/PI3K/p38 MAPK pathway and the VEGFR2/MEK/ERK1/2 pathway. Analysis of cell communication between endothelial cells and myeloid-derived cells revealed that this is a large network. Interestingly, among these different cell populations, the THY1 signaling pathway was only expressed in tandem between PLVAP+ ECs and other cell populations, especially TAM1. TAM1 has fetal liver macrophage characteristics[37], and fetal-like PLVAP+ ECs affect fetal-like liver macrophage distribution[31], which indicates that PLVAP+ ECs might affect the layout of TAM1 in some way. Studies have shown that PLVAP+ ECs and TAM1 colocalize in the tumor microenvironment and have immunosuppressive effects, which strongly suggests that ERS mediates the recruitment of PLVAP+ ECs in the HCC tumor microenvironment to TAM1 around new blood vessels through the THY1/MAC1 signaling pathway[37], and provides an immunosuppressive microenvironment for tumor metastasis. Radiogenomics, as an emerging field, holds great potential for various applications[38, 39]. Notably, our research revealed the significant role of ERS in the occurrence and development of HCC. This discovery may pave the way for a new targeted treatment approach for HCC. To facilitate its clinical application, we applied radiogenomics and developed a radiogenomic signature using non-invasive radiomics techniques, which demonstrated promising benefits. The radiomic features that constitute the radiogenomic signature are all wavelet features, which reflect its high stability and efficiency in reflecting tumor heterogeneity. Simultaneously, there are features from the arterial phase and the venous phase, which reflects that enhanced scans at different phases are of great value in assessing the heterogeneity of HCC, which is consistent with previous studies[40, 41]. For patients with unresectable HCC, the preferred treatment option is systemic combination therapy, which includes immunotherapy + targeted therapy + interventional therapy[42, 43]. However, due to individual differences, everyone has different responses to systemic combination therapy, so there is an urgent need for a method that can predict therapeutic response. Here, we demonstrated that ERS is closely related to PD-1 and VEGF, so we constructed a radiogenomic signature based on ERS. The results showed that the RDS high-risk group had a worse response to systemic combination therapy and a shorter PFS, which seems to contradict the results of our previous study. However, after receiving systemic combination therapy, ERS is activated, which leads to immune evasion and enhanced expression and activity of signaling pathways such as VEGF. Therefore, the subgroup with high expression of ERS genes is more likely to develop drug resistance[13], which in turn affects the effectiveness of systemic combination therapy and patient prognosis. These results show that our radiogenomic signature can accurately predict patient response to systemic combination therapy. However, our study has several limitations. Our research primarily focused on genomics and radiomics and lacked experimental validation. This is a key aspect of our future research plans. Additionally, the sample size in our study was insufficient, potentially impacting the statistical significance of the clinical data. In summary, we established an ERS prognostic model that can predict patient prognosis. We also found that ERS is closely related to the tumor microenvironment and is mainly manifested in the interaction between tumor-associated endothelial cells and tumor-associated macrophages. Moreover, we constructed a radiogenomic signature based on the ERS. This signature can guide subsequent combination therapy for patients with unresectable HCC. Abbreviations HCC:Hepatocellular carcinoma PD-1:Programmed cell death-1 FDA:Food and Drug Administration ERS:Endoplasmic reticulum stress TPM:Transcripts per million reads CT:Computed tomography TCGA:The Cancer Genome Atlas Program ICGC:International cancer genome consortium GEO:Gene Expression Omnibus TCIA:The Cancer Imaging Archive GDSC:The Genomics of Drug Sensitivity in Cancer RNA:Ribonucleic Acid TACE:Transcatheter arterial chemoembolization VEGF:vascular endothelial growth factor ssGSEA:Single-sample gene set enrichment analysis ROI:Region of interest mRMR:Maximum relevance minimum redundancy OS:Overall survival CR:Complete response PR:partial response SD:stable disease (SD) PD:progressive disease (PD) ORR:objective response rate (ORR) DCR:disease control rate (DCR) RECIST:Response Evaluation Criteria in Solid Tumors (RECIST) DEG:Differential Expression Analysis，DEG STC2:Stanniocalcin-2 DDX11:DEAD/H-Box Helicase 11 MAGEA3:Melanoma-Associated Antigen 3 BRSK2:BR Serine/Threonine-Protein Kinase 2 ROC:Receiver Operating Characteristic AUC:area under the curve (AUC) TIDE:Tumor Immune Dysfunction and Exclusion PLVAP:Plasmalemma Vesicle Associated Protein TAM:Tumor-associated macrophages LASSO:Least absolute shrinkage and selection operator RDS:radiomics score PFS:progression-free survival TME:Tumor Mircroenviroment Declarations Ethics approval and consent to participate：Ethics approval by the First Affiliated Hospital of Nanjing Medical University Ethics Committee (Ethics number: 2024-SR-368). Consent for publication:Not applicable. Availability of data and materials:The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Competing interests:All the authors declare no competing interests. Funding:This study was funded by the Jiangsu Provincial Key Research and Development Program (Social Development) under project number BE2020708 and National Natural Science Foundation of China (82102150) and the Natural Science Foundation of Jiangsu Province (BK20210968). Authors' contributions:W.H.Y. designed, performed, and analyzed experiments, and wrote the manuscript. C.S.Y. and W.T. contributed to the bioinformatic analyses.W.K.，J.G.W. and X.Z.G. helped in the project design, supervised the progress of the study, and edited the manuscript. All authors read and approved the final manuscript. Acknowledgements:Fundings from the Jiangsu Provincial Key Research and Development Program (Social Development) under project number BE2020708 and National Natural Science Foundation of China (82102150) and the Natural Science Foundation of Jiangsu Province (BK20210968). References Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, Bray F: Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 2021, 71(3):209-249. Singal AG, Lampertico P, Nahon P: Epidemiology and surveillance for hepatocellular carcinoma: New trends. J Hepatol 2020, 72(2):250-261. Marin JJG, Macias RIR, Monte MJ, Romero MR, Asensio M, Sanchez-Martin A, Cives-Losada C, Temprano AG, Espinosa-Escudero R, Reviejo M et al: Molecular Bases of Drug Resistance in Hepatocellular Carcinoma. Cancers (Basel) 2020, 12(6). Qin S, Bai Y, Lim HY, Thongprasert S, Chao Y, Fan J, Yang TS, Bhudhisawasdi V, Kang WK, Zhou Y et al: Randomized, multicenter, open-label study of oxaliplatin plus fluorouracil/leucovorin versus doxorubicin as palliative chemotherapy in patients with advanced hepatocellular carcinoma from Asia. J Clin Oncol 2013, 31(28):3501-3508. Zhu AX, Finn RS, Edeline J, Cattan S, Ogasawara S, Palmer D, Verslype C, Zagonel V, Fartoux L, Vogel A et al: Pembrolizumab in patients with advanced hepatocellular carcinoma previously treated with sorafenib (KEYNOTE-224): a non-randomised, open-label phase 2 trial. Lancet Oncol 2018, 19(7):940-952. Zhu AX, Abbas AR, de Galarreta MR, Guan Y, Lu S, Koeppen H, Zhang W, Hsu CH, He AR, Ryoo BY et al: Molecular correlates of clinical response and resistance to atezolizumab in combination with bevacizumab in advanced hepatocellular carcinoma. Nat Med 2022, 28(8):1599-1611. da Silva DC, Valentão P, Andrade PB, Pereira DM: Endoplasmic reticulum stress signaling in cancer and neurodegenerative disorders: Tools and strategies to understand its complexity. Pharmacol Res 2020, 155:104702. Wu SX, Ye SS, Hong YX, Chen Y, Wang B, Lin XJ, Lin X: Hepatitis B Virus Small Envelope Protein Promotes Hepatocellular Carcinoma Angiogenesis via Endoplasmic Reticulum Stress Signaling To Upregulate the Expression of Vascular Endothelial Growth Factor A. Journal of virology 2022, 96(4):e0197521. Vandewynckel YP, Laukens D, Bogaerts E, Paridaens A, Van den Bussche A, Verhelst X, Van Steenkiste C, Descamps B, Vanhove C, Libbrecht L et al: Modulation of the unfolded protein response impedes tumor cell adaptation to proteotoxic stress: a PERK for hepatocellular carcinoma therapy. Hepatol Int 2015, 9(1):93-104. Liu P, Wei J, Mao F, Xin Z, Duan H, Du Y, Wang X, Li Z, Qian J, Yao J: Establishment of a Prognostic Model for Hepatocellular Carcinoma Based on Endoplasmic Reticulum Stress-Related Gene Analysis. Front Oncol 2021, 11:641487. Chen X, Cubillos-Ruiz JR: Endoplasmic reticulum stress signals in the tumour and its microenvironment. Nat Rev Cancer 2021, 21(2):71-88. Bao MH, Wong CC: Hypoxia, Metabolic Reprogramming, and Drug Resistance in Liver Cancer. Cells 2021, 10(7). Khaled J, Kopsida M, Lennernäs H, Heindryckx F: Drug Resistance and Endoplasmic Reticulum Stress in Hepatocellular Carcinoma. Cells 2022, 11(4). Liu J, Fan L, Yu H, Zhang J, He Y, Feng D, Wang F, Li X, Liu Q, Li Y et al: Endoplasmic Reticulum Stress Causes Liver Cancer Cells to Release Exosomal miR-23a-3p and Up-regulate Programmed Death Ligand 1 Expression in Macrophages. Hepatology 2019, 70(1):241-258. Ming W, Zhu Y, Bai Y, Gu W, Li F, Hu Z, Xia T, Dai Z, Yu X, Li H et al: Radiogenomics analysis reveals the associations of dynamic contrast-enhanced-MRI features with gene expression characteristics, PAM50 subtypes, and prognosis of breast cancer. Front Oncol 2022, 12:943326. Lambin P, Leijenaar RTH, Deist TM, Peerlings J, de Jong EEC, van Timmeren J, Sanduleanu S, Larue R, Even AJG, Jochems A et al: Radiomics: the bridge between medical imaging and personalized medicine. Nat Rev Clin Oncol 2017, 14(12):749-762. Pinker K, Chin J, Melsaether AN, Morris EA, Moy L: Precision Medicine and Radiogenomics in Breast Cancer: New Approaches toward Diagnosis and Treatment. Radiology 2018, 287(3):732-747. Hu Y, Xie C, Yang H, Ho JWK, Wen J, Han L, Chiu KWH, Fu J, Vardhanabhuti V: Assessment of Intratumoral and Peritumoral Computed Tomography Radiomics for Predicting Pathological Complete Response to Neoadjuvant Chemoradiation in Patients With Esophageal Squamous Cell Carcinoma. JAMA Netw Open 2020, 3(9):e2015927. Kim S, Shin J, Kim DY, Choi GH, Kim MJ, Choi JY: Radiomics on Gadoxetic Acid-Enhanced Magnetic Resonance Imaging for Prediction of Postoperative Early and Late Recurrence of Single Hepatocellular Carcinoma. Clin Cancer Res 2019, 25(13):3847-3855. Xu T, Ren L, Liao M, Zhao B, Wei R, Zhou Z, He Y, Zhang H, Chen D, Chen H et al: Preoperative Radiomics Analysis of Contrast-Enhanced CT for Microvascular Invasion and Prognosis Stratification in Hepatocellular Carcinoma. J Hepatocell Carcinoma 2022, 9:189-201. Zheng Z, Guo Y, Huang X, Liu J, Wang R, Qiu X, Liu S: CD8A as a Prognostic and Immunotherapy Predictive Biomarker Can Be Evaluated by MRI Radiomics Features in Bladder Cancer. Cancers (Basel) 2022, 14(19). Di Conza G, Ho PC: ER Stress Responses: An Emerging Modulator for Innate Immunity. Cells 2020, 9(3). Yan D, Wang HW, Bowman RL, Joyce JA: STAT3 and STAT6 Signaling Pathways Synergize to Promote Cathepsin Secretion from Macrophages via IRE1α Activation. Cell Rep 2016, 16(11):2914-2927. Chung AS, Ferrara N: Developmental and pathological angiogenesis. Annu Rev Cell Dev Biol 2011, 27:563-584. Xu Z, Guo C, Ye Q, Shi Y, Sun Y, Zhang J, Huang J, Huang Y, Zeng C, Zhang X et al: Endothelial deletion of SHP2 suppresses tumor angiogenesis and promotes vascular normalization. Nat Commun 2021, 12(1):6310. Zhang L, Xu J, Zhou S, Yao F, Zhang R, You W, Dai J, Yu K, Zhang Y, Baheti T et al: Endothelial DGKG promotes tumor angiogenesis and immune evasion in hepatocellular carcinoma. J Hepatol 2023. Binet F, Sapieha P: ER Stress and Angiogenesis. Cell Metab 2015, 22(4):560-575. Qie S, Sang N: Stanniocalcin 2 (STC2): a universal tumour biomarker and a potential therapeutical target. J Exp Clin Cancer Res 2022, 41(1):161. Law AY, Wong CK: Stanniocalcin-2 is a HIF-1 target gene that promotes cell proliferation in hypoxia. Exp Cell Res 2010, 316(3):466-476. Ren S, Yang Y: The proliferation and angiogenesis in hemangioma-derived endothelial cells is affected by STC2 medicated VEGFR2/Akt/eNOS pathway. Pak J Med Sci 2023, 39(4):1119-1123. Rantakari P, Jäppinen N, Lokka E, Mokkala E, Gerke H, Peuhu E, Ivaska J, Elima K, Auvinen K, Salmi M: Fetal liver endothelium regulates the seeding of tissue-resident macrophages. Nature 2016, 538(7625):392-396. Ramachandran P, Dobie R, Wilson-Kanamori JR, Dora EF, Henderson BEP, Luu NT, Portman JR, Matchett KP, Brice M, Marwick JA et al: Resolving the fibrotic niche of human liver cirrhosis at single-cell level. Nature 2019, 575(7783):512-518. Wang YH, Cheng TY, Chen TY, Chang KM, Chuang VP, Kao KJ: Plasmalemmal Vesicle Associated Protein (PLVAP) as a therapeutic target for treatment of hepatocellular carcinoma. BMC cancer 2014, 14:815. Strickland LA, Jubb AM, Hongo JA, Zhong F, Burwick J, Fu L, Frantz GD, Koeppen H: Plasmalemmal vesicle-associated protein (PLVAP) is expressed by tumour endothelium and is upregulated by vascular endothelial growth factor-A (VEGF). J Pathol 2005, 206(4):466-475. Hamilton BJ, Tse D, Stan RV: Phorbol esters induce PLVAP expression via VEGF and additional secreted molecules in MEK1-dependent and p38, JNK and PI3K/Akt-independent manner. J Cell Mol Med 2019, 23(2):920-933. Ma B, Xu X, He S, Zhang J, Wang X, Wu P, Liu J, Jiang H, Zheng M, Li W et al: STC2 modulates ERK1/2 signaling to suppress adipogenic differentiation of human bone marrow mesenchymal stem cells. Biochem Biophys Res Commun 2020, 524(1):163-168. Sharma A, Seow JJW, Dutertre CA, Pai R, Blériot C, Mishra A, Wong RMM, Singh GSN, Sudhagar S, Khalilnezhad S et al: Onco-fetal Reprogramming of Endothelial Cells Drives Immunosuppressive Macrophages in Hepatocellular Carcinoma. Cell 2020, 183(2):377-394.e321. Hartmann K, Sadée CY, Satwah I, Carrillo-Perez F, Gevaert O: Imaging genomics: data fusion in uncovering disease heritability. Trends in molecular medicine 2023, 29(2):141-151. Liu Z, Wu K, Wu B, Tang X, Yuan H, Pang H, Huang Y, Zhu X, Luo H, Qi Y: Imaging genomics for accurate diagnosis and treatment of tumors: A cutting edge overview. Biomedicine & pharmacotherapy = Biomedecine & pharmacotherapie 2021, 135:111173. Tang VH, Duong STM, Nguyen CDT, Huynh TM, Duc VT, Phan C, Le H, Bui T, Truong SQH: Wavelet radiomics features from multiphase CT images for screening hepatocellular carcinoma: analysis and comparison. Scientific reports 2023, 13(1):19559. Soufi M, Arimura H, Nagami N: Identification of optimal mother wavelets in survival prediction of lung cancer patients using wavelet decomposition-based radiomic features. Medical physics 2018, 45(11):5116-5128. Brown ZJ, Tsilimigras DI, Ruff SM, Mohseni A, Kamel IR, Cloyd JM, Pawlik TM: Management of Hepatocellular Carcinoma: A Review. JAMA Surg 2023, 158(4):410-420. Llovet JM, Kelley RK, Villanueva A, Singal AG, Pikarsky E, Roayaie S, Lencioni R, Koike K, Zucman-Rossi J, Finn RS: Hepatocellular carcinoma. Nature reviews Disease primers 2021, 7(1):6. The Cancer Genome Atlas Program database (https://portal.gdc.cancer.gov/) Accessed 13 November 2022 International cancer genome consortium database (https://dcc.icgc.org/) Accessed 13 November 2022 Molecular Signatures Database (https://www.gsea-msigdb.org/gsea/msigdb/index.jsp) Accessed 13 November 2022 GEO database (Home - GEO - NCBI (nih.gov)) Accessed 13 November 2022 the Cancer Imaging Archive (TCIA: https://www.cancerimagingarchive.net/) Accessed 6 October 2023 the Genomics of Drug Sensitivity in Cancer (GDSC: https://www.cancerrxgene.org/) database Accessed 6 October 2023 The CellMarker (CellMarker2.0 (hrbmu.edu.cn) database Accessed 8 April 2023 Tables Table 1 Drug sensitivity analysis Drug Targets Signaling pathway Sensitivity. Uprosertib AKT1, AKT2, AKT3 PI3K/MTOR signaling sensitive Taselisib PI3K (beta sparing) PI3K/MTOR signaling sensitive Pictilisib PI3K (class 1) PI3K/MTOR signaling sensitive OSI. 027 MTORC1, MTORC2 PI3K/MTOR signaling sensitive Doramapimod p38, JNK2 JNK and p38 signaling sensitive BI.2536 PLK1, PLK2, PLK3 Cell cycle resistant ERK_2440 ERK1,ERK2 ERK MAPK signaling resistant ERK_6604 ERK1,ERK2 ERK MAPK signaling resistant Mitoxantrone TOP2 DNA replication resistant PD0325901 MEK1, MEK2 ERK MAPK signaling resistant RO.3306 CDK1 Cell cycle resistant SB216763 GSK3A, GSK3B WNT signaling resistant SCH772984 ERK1, ERK2 ERK MAPK signaling resistant Selumetinib MEK1, MEK2 ERK MAPK signaling resistant Topotecan TOP1 DNA replication resistant Trametinib MEK1, MEK2 ERK MAPK signaling resistant VE.822 ATR Genome integrity resistant Table 2 Therapy response of RDSgroup in the systemic combination therapy cohort high low P N=20 N=20 Response 0.009 CR 0 (0.00%) 2 (10.0%) PR 1 (5.00%) 7 (35.0%) SD 10 (50.0%) 9 (45.0%) PD 9 (45.0%) 2 (10.0%) Additional Declarations No competing interests reported. Supplementary Files Additionalmaterials.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-4535127","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":316295042,"identity":"c29a532a-09af-405d-9e65-bc30e616ef9f","order_by":0,"name":"Huai-Yu Wu","email":"","orcid":"","institution":"The First Affiliated Hospital of Nanjing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Huai-Yu","middleName":"","lastName":"Wu","suffix":""},{"id":316295043,"identity":"ccc9119c-0383-4647-8729-8b0d13c80316","order_by":1,"name":"Shu-Ya Cao","email":"","orcid":"","institution":"The First Affiliated Hospital of Nanjing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Shu-Ya","middleName":"","lastName":"Cao","suffix":""},{"id":316295044,"identity":"0deb4560-74ec-4f5d-bc73-e73d3a8770f7","order_by":2,"name":"Zheng-Gang Xu","email":"","orcid":"","institution":"The First Affiliated Hospital of Nanjing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Zheng-Gang","middleName":"","lastName":"Xu","suffix":""},{"id":316295045,"identity":"bdca9656-bf11-4d78-9171-783dd5ea7b47","order_by":3,"name":"Tian Wang","email":"","orcid":"","institution":"The First Affiliated Hospital of Nanjing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Tian","middleName":"","lastName":"Wang","suffix":""},{"id":316295046,"identity":"52e1bb20-e896-4059-bc7c-6903a2ee1ee3","order_by":4,"name":"Gu-Wei Ji","email":"","orcid":"","institution":"The First Affiliated Hospital of Nanjing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Gu-Wei","middleName":"","lastName":"Ji","suffix":""},{"id":316295047,"identity":"ed3f2667-c81e-45fa-a620-a1cb3cbc10cc","order_by":5,"name":"Ke Wang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAt0lEQVRIiWNgGAWjYBACPihtwM/MfPgBUVrYwGQCg4FkO1uaAWlaDM7zKEgQp4X97MEPH38cNjY+zMNgwFBjE01YC09esuSMhMNmZod5DzxgOJaW20DYYTlmzDwJt23MDvMlGDA2HCZCC/8bM+Y/QC3GzTwGEsRpkQDawpBw28yAmXgtb4wle9L+G0scBgZyAjF+4efPMfzwwybNsL//8OEHH2psCGtBBQmkKR8Fo2AUjIJRgAsAAKgqOCBHTqUmAAAAAElFTkSuQmCC","orcid":"","institution":"The First Affiliated Hospital of Nanjing Medical University","correspondingAuthor":true,"prefix":"","firstName":"Ke","middleName":"","lastName":"Wang","suffix":""}],"badges":[],"createdAt":"2024-06-05 15:29:14","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4535127/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4535127/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":59136902,"identity":"4f686ff9-fd7c-4d4f-a8ed-db3323c171e2","added_by":"auto","created_at":"2024-06-26 18:56:30","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":102823,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eWork flaw of the study design.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-4535127/v1/c154113f347d69e4ad18db27.png"},{"id":59138807,"identity":"2db9a47c-5d36-472d-a94c-fdde40001807","added_by":"auto","created_at":"2024-06-26 19:12:30","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":359810,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eConstruction and validation of the HCC prognostic model based on 4 ERS-associated genes.\u003c/strong\u003e (a) Volcano plot showing differentially expressed ERS-related genes between tumor tissues and non-tumor tissues. Red and green represent upregulated and downregulated genes, respectively. (b) Forrest chart of multivariate Cox regression analysis of the 4 genes associated with ERS in the training set of the TCGA-LIHC cohort (n=429). (c) Kaplan‒Meier curves of OS for the high-risk and low-risk groups in the TCGA-LIHC cohort (n=429) and ICGC-JP-LIRI cohort (n=240). (d) ROC curves of 1-, 3-, and 5-year OS according to ERSrisk , age, sex, TNM stage, and tumor stage. (e) Comparison of clinical characteristics, including tumor TNM stage and tumor stage, between the high-risk and low-risk groups. *P \u0026lt; 0.05, **P \u0026lt; 0.01, ***P \u0026lt; 0.001, ****P \u0026lt; 0.0001.\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-4535127/v1/76a8ce55fa480e49f82a9d77.png"},{"id":59137981,"identity":"0c11b1c9-056e-4b55-89ac-a97e6103ea75","added_by":"auto","created_at":"2024-06-26 19:04:30","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":331398,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThe prognostic risk score correlated with the immune microenvironment.\u003c/strong\u003e(a) Comparison of the degree of immune cell infiltration between the high-risk and low-risk groups in the TCGA-LIHC cohort (n=429) and ICGC-JP-LIRI cohort (n=240). (b) Comparison of the immunophenotypes (CTLA4 and PD-1) between the high-risk and low-risk groups in the TCGA-LIHC cohort (n=429). (c) Relationships between the risk score and the expression levels of immune checkpoint genes and TIDE in the TCGA-LIHC cohort (n=429). *P \u0026lt; 0.05, **P \u0026lt; 0.01, ***P \u0026lt; 0.001, ****P \u0026lt; 0.0001.\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-4535127/v1/cdbac49b95b739adfa893954.png"},{"id":59136906,"identity":"830051db-3c5b-46fe-9fbd-1c3fbe1687eb","added_by":"auto","created_at":"2024-06-26 18:56:30","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":596886,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSingle-cell type expression of the identified ERS-related genes in HCC. \u003c/strong\u003e(a) Twenty-three cell clusters were obtained after the first-level classification, and eight types were identified by marker gene annotation. (b) STC2 was significantly expressed in endothelial cells. (c) Ten cell clusters were obtained after the second-level classification of endothelial cells, and five cell types were identified by marker gene annotation. (d) STC2 was significantly expressed in PLVAP+ECs. (e) Thirteen cell clusters were obtained after the second-level classification of myeloid-derived immune cells, and eight types were identified by marker gene annotation. (f) The cellular communication network in which PLVAP+ECs interact with other cells. (g) The THY1 signaling pathway interacts in which PLVAP+ECs interact with other cells.\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-4535127/v1/1012acf415d4a2cda2c1dc53.png"},{"id":59138808,"identity":"b48ae8fe-ebe7-41e7-8a2e-da43dba51a02","added_by":"auto","created_at":"2024-06-26 19:12:31","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":201002,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eEstablishment and validation of the radiogenomic signature\u003c/strong\u003e (a-b) LASSO regression analysis for radiogenomic signature construction. (c) Nine features and their weights to build the signature. (d) Kaplan–Meier curves of OS for the RDS-high and RDS-low groups in the TCGA-LIHC cohort (n=43).\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-4535127/v1/1b2af77d8f1165586431c33d.png"},{"id":59137984,"identity":"51bc20e9-2345-4312-8fb1-b43959a2a05e","added_by":"auto","created_at":"2024-06-26 19:04:30","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":275846,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePredictive value of radiomics imaging biomarkers for therapeutic response and clinical outcomes in patients treated with combination therapy.\u003c/strong\u003e(a) Kaplan–Meier curves of OS for the RDS-high and RDS-low groups in the surgical therapy cohort (n=236). (b) ROIs at baseline and post-treatment in the systemic combination therapy cohort (n=40). (c-d) The ratio of different therapy responses among the RDS-low group and RDS-high group in combination therapy cohort 1 (n=40). (e) ROC analysis of ORR and DCR in the combination therapy cohort (n=40). (f) Prognostic value of the RDS signature for PFS in patients treated with systemic combination therapy (n=40).\u003c/p\u003e","description":"","filename":"Figure6.png","url":"https://assets-eu.researchsquare.com/files/rs-4535127/v1/ad1bd62c4fc9c0df018c6b4f.png"},{"id":59139745,"identity":"fb35acc4-8b7a-4109-8c47-0171e41be2da","added_by":"auto","created_at":"2024-06-26 19:28:34","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2388607,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4535127/v1/fb1b7a24-c0c9-4660-9c68-282d871d8135.pdf"},{"id":59136908,"identity":"a08e5a3d-cac7-4d65-afd7-ed9130423b95","added_by":"auto","created_at":"2024-06-26 18:56:31","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":13173079,"visible":true,"origin":"","legend":"","description":"","filename":"Additionalmaterials.docx","url":"https://assets-eu.researchsquare.com/files/rs-4535127/v1/eebb0e817e4f9c0b6287b0c4.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Construction of a radiogenomic signature based on endoplasmic reticulum stress for predicting prognosis and systemic combination therapy response in hepatocellular carcinoma","fulltext":[{"header":"Introduction","content":"\u003cp\u003eLiver cancer is currently the sixth most common malignant disease and the third leading cause of cancer death globally[1]. Hepatocellular carcinoma (HCC) accounts for 75\u0026ndash;85% of liver cancer cases[2]. Traditional chemotherapy is ineffective against HCC[3], with liver resection and transplantation being the primary treatment options[4]. Immune checkpoint inhibitors targeting programmed cell death-1 (PD-1) based on the tumor immune environment have been approved by the FDA as second-line therapies for advanced HCC[5]. While some patients show significant responses to immunotherapy, approximately 20% to 30%, others exhibit resistance[6], likely due to immunosuppression and immune evasion orchestrated by the tumor microenvironment.\u003c/p\u003e\n\u003cp\u003ePrevious studies have shown a positive association between endoplasmic reticulum stress (ERS) and the onset and progression of various liver disorders, including cancer[7-10]. Hypoxia, nutrient deficiency, and rapid protein increase in the tumor microenvironment inevitably lead to the occurrence of ERS[11]. Three different branches of the ERS signaling pathway are involved in chemoresistance, immunosuppression and the induction of immune evasion through different mechanisms[9, 11-14]. Therefore, the abnormal activation of ERS sensors and their downstream signaling pathways has emerged as a critical regulator of tumor development and metastasis, suggesting potential targets for cancer therapy.\u003c/p\u003e\n\u003cp\u003eRadiomics and radiogenomics are emerging fields with a wide range of potential applications. Radiogenomics allows for the non-invasive acquisition of genetic information about tumors[15], which is a more advanced approach than traditional methods involving surgical resection for pathological specimen retrieval. These methods can guide subsequent systemic therapy for HCC patients who are not eligible for surgery. Radiomics is increasingly important in personalized cancer treatment, as quantitative features extracted from radiographic medical images provide valuable high-dimensional data that can enhance clinical decision-making and improve diagnostic and prognostic capabilities[16, 17]. Recent studies have highlighted the diagnostic and prognostic potential of radiomics, particularly in the analysis of tumor and peritumoral regions in HCC, offering insights into the tumor microenvironment[18-20]. Additionally, radiomics shows promise in predicting gene expression prior to surgery, potentially altering prognosis assessments and treatment decisions for cancer patients undergoing surgical procedures[21].\u003c/p\u003e\n\u003cp\u003eThe objectives of our study were to develop and validate a model focused on genes associated with ERS and explore the connection between ERS and the tumor microenvironment. Moreover, radiomics has been used to construct a radiogenomic signature to non-invasively predict the prognosis of surgical patients and the therapeutic response of patients treated with systemic combination therapy.\u003c/p\u003e"},{"header":"Methods and Materials","content":"\u003ch3\u003eStudy sample\u003c/h3\u003e\n\u003cp\u003eThe training set of bulk RNA sequencing data and associated clinical information for HCC patients (TCGA-LIHC, n=424) were obtained from the TCGA database. Five tumor samples with missing clinical information were excluded, and 419 samples with complete information were included. The validation set of bulk RNA sequencing data and clinical information (ICGC-JP-LIRI, n=240) were downloaded from the ICGC database. To guarantee a more accurate depiction of gene expression, the bulk data were then transformed into TPM (transcripts per million reads) format. The 253 ERS-related genes were obtained from the Molecular Signatures Database. Additionally, the corresponding somatic mutation data were obtained from the TCGA database. The HCC scRNA-seq datasets GSE156625 (n=58), GSE162616 (n=3), GSE189175 (n=3) and GSE210679 (n=2) were downloaded from the GEO database.\u003c/p\u003e\n\u003cp\u003eFor the radiomic analysis, we constructed a radiogenomic signature based on the TCIA cohort (43 patients with complete imaging information and clinical data) from the Cancer Imaging Archive. The surgical therapy cohort consisted of 236 patients who were recruited from our center between 2009 and 2017. The inclusion criteria were as follows: (1) Patients who could undergo R0 resection of HCC. (2) Enhanced CT images before surgery. (3) Patients who did not receive neoadjuvant treatment before surgery. (4) Complete follow-up data. In addition, the systemic combination\u0026nbsp;therapy cohort consisted of 40 patients who received anti-PD-1 plus TACE and anti-VEGF therapy between 2019 and 2023. The inclusion criteria were as follows: (1) At least one systemic combination therapy. (2) No systemic combination therapy was administered before baseline imaging. (3) Enhanced CT images before treatment. (4) Systemic combination therapy for primary HCC or recurrence of HCC.\u003c/p\u003e\n\u003ch3\u003eERS model establishment and validation\u003c/h3\u003e\n\u003cp\u003eDifferential gene expression was analyzed using the R package \u0026ldquo;edgeR\u0026rdquo;, and 35 differentially\u0026nbsp;expressed\u0026nbsp;genes were used in the subsequent\u0026nbsp;analysis. Univariate Cox regression analysis was utilized to investigate the relationship between prognosis and ERS-associated gene expression levels in patients with HCC. Variables with P values less than 0.05 in univariate analysis were entered into\u0026nbsp;multivariate Cox analysis using the stepwise backward selection method. The \u0026quot;survival\u0026quot; package in R was used for all the statistical analyses involving Cox regression, and the results were visualized using the \u0026ldquo;forest plot\u0026rdquo; R package. The risk score was calculated as \u0026beta;1 \u0026times; Exp1 + \u0026beta;2 \u0026times; Exp2 + \u0026beta;3 \u0026times; Exp3+\u0026beta;x \u0026times; Expx. According to the cut-off risk score, the same standard was used to classify patients in the training and validation cohorts into high-risk and low-risk groups. Additionally, Kaplan\u0026ndash;Meier survival curves revealed differences in prognosis between the high-risk and low-risk groups.\u003c/p\u003e\n\u003cp\u003eMulti-scale analysis based on bioinformatics\u003c/p\u003e\n\u003cp\u003eTo gain a comprehensive understanding of the relationship between ERS and the tumor immune microenvironment, we used CIBERSORT, which calculates tumor-infiltrating immune cell compositions using deconvolution analysis. Subsequently, we divided the samples into high-risk and low-risk groups to compare the immune cell infiltration levels between them. Single-sample gene set enrichment analysis (ssGSEA) was used to\u0026nbsp;assess\u0026nbsp;the\u0026nbsp;correlation\u0026nbsp;between\u0026nbsp;gene\u0026nbsp;signatures\u0026nbsp;and\u0026nbsp;immune\u0026nbsp;infiltration via the \u0026ldquo;gsva\u0026rdquo; R package.\u003c/p\u003e\n\u003cp\u003eThe \u0026ldquo;Maftools\u0026rdquo; package in R was utilized to analyze and compare the quantity and quality of gene mutations between the high-risk and low-risk groups. We collected drug screening data from the Genomics of \u0026nbsp;Drug Sensitivity in Cancer database for drug sensitivity analysis. The drug sensitivity of patients in the TCGA-LIHC cohort and ICGC-JP-LIRI cohort was estimated via the \u0026ldquo;OncoPredict\u0026rdquo; R package.\u003c/p\u003e\n\u003cp\u003eScRNA-seq analysis was performed in R using the package \u0026ldquo;Seurat\u0026rdquo;. First, the FindIntegrationAnchors and IntegrateData functions in Seurat were used for the integration of these HCC scRNA-seq data. Subsequently, cells that failed to meet the prescribed quality criteria were removed from the integrated dataset. These criteria included the following: (1) 500\u0026lt;nFeature_RNA\u0026lt;2500; (2) 200\u0026lt;nCount_RNA; (3) percent.mt \u0026lt; 5. As a result, a comprehensive dataset composed of 100073 cells was obtained. Following this, the analysis proceeded using the standard Seurat workflow. The CellMarker database and the \u0026ldquo;SingleR\u0026rdquo; R package were used for cell type annotation. The interaction patterns between cells in the tumor microenvironment were analyzed using the \u0026ldquo;cellchat\u0026rdquo; R package. Their position in the cellular communication network was determined using the ligand‒receptor database.\u003c/p\u003e\n\u003ch3\u003eRadiogenomic signature establishment and validation\u003c/h3\u003e\n\u003cp\u003eWe constructed a radiogenomic signature based on the TCIA cohort using 3DSlicer software to construct an outline of the regions of interest (ROIs). Next, we validated the signature using surgical therapy cohort and systemic combination therapy cohort. This work was performed by two radiologists who operated under a double-blind protocol to ensure unbiased results. Additionally, reader 1 repeated the work\u0026nbsp;with an interval\u0026nbsp;of\u0026nbsp;at least\u0026nbsp;1\u0026nbsp;month following the same procedure. 883 features were extracted from the artery and vein phases respectively through PyRadiomics, which was built into 3DSlicer. A total of 1766 features were retained after excluding features without statistical data. Subsequently, intragroup consistency analysis was performed on the features of the 43 patients drawn by reader 1, and intergroup consistency analysis was performed on the features of the same 43 patients drawn by readers 1 and 2. The features whose ICC＞0.9 were included in the following procedure. Then, we applied Z score normalization to ensure that the radiomic features were measured on the same scale, and Levene\u0026rsquo;s test was used to verify the\u0026nbsp;homogeneity\u0026nbsp;of\u0026nbsp;variances. We employed the maximum relevance minimum redundancy (mRMR) to retain the most pertinent and reproducible radiomic features for subsequent analyses. Internal validation was performed using the 10-fold cross-validation method. We employed the Lasso algorithm to filter out radiomic features to construct the signature and classify samples. Next, we investigated the value of the signature for the prediction of overall survival (OS) and systemic combination therapy response. Systemic combination therapy responses include complete response (CR), partial response (PR), stable disease (SD), progressive disease (PD), objective response rate (ORR), and disease control rate (DCR) (evaluated every 6~8 weeks) according to the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistical analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eR version 4.1.1 was used for the statistical analysis. Categorized variables between different risk groups were compared by the Wilcoxon t test, and the log rank test was used to test for significant differences in survival probability between samples, with a P value\u0026lt;0.05 indicating statistical significance. Pearson analysis was used to calculate correlations if the data were normally distributed; otherwise, the Spearman rank test was used.\u003c/p\u003e"},{"header":"Results","content":"\u003ch3\u003eEstablishment and validation of the ERS model\u003c/h3\u003e\n\u003cp\u003eBy comparing the gene expression profiles of 374 tumor tissues and 50 normal liver tissues from the TCGA database, we identified 35 differentially expressed ERS-related genes, including 24 upregulated genes and 11 downregulated genes (Fig. 2a) in tumor tissues. Using univariate Cox regression, nine ERS-related genes (p\u0026lt;0.05) associated with OS were found among the 35 DEGs (Additional Table 1). Following the application of multivariate Cox regression analysis based on these ERS-related genes, a prognostic model was developed by defining four sERS-related genes (Fig. 2b). The format for calculating the risk score was as follows: ERSrisk=0.24550*STC2+0.27670* DDX11+0.11410* MAGEA3+0.38278* BRSK2. The ERS risk was selected as the cutoff value for the ROC curve, and the patients were separated into high-risk and low-risk groups. Calibration curves\u0026nbsp;were\u0026nbsp;used to evaluate the discrimination and calibration of our\u0026nbsp;model (Additional Fig. 1). The ERS model C-indexes were 0.646 in the TCGA-LIHC cohort and 0.653 in the ICGC-JP-LIRI cohort. K‒M survival analysis revealed that patients in the high-risk group had a poor prognosis, while those in the low-risk group had a relatively good prognosis (Fig. 2c).\u003c/p\u003e\n\u003cp\u003eTo further assess the predictive ability of the model for HCC, a stratified analysis was conducted based on pathological and clinical characteristics such as TNM stage, age, sex, and tumor stage (Additional Fig. 2a). Patients with a high ERSrisk exhibited a greater likelihood of early mortality than did those with a low ERSrisk (Additional Fig. 2b). The area under the curve (AUC)\u0026nbsp;values for 1-, 3-, and 5-year OS predictions were superior to those of other clinical characteristics (Fig. 2d). Variations in ERSrisk were observed between T1 and T2~T4, as well as between stage I and stage II~III, but not in advanced HCC (Fig. 2e).\u003c/p\u003e\n\u003cp\u003eNext, we performed a comparative analysis of mutations in the high-risk and low-risk groups (Additional Fig. 3a and b). We found that the most significant difference in ERSrisk was TP53. The ERSrisk of the mutant type was significantly greater than that of the wild type (Additional Fig. 3c).\u0026nbsp;Moreover, the expression levels of MAGEA3 and DDX11 were\u0026nbsp;increased\u0026nbsp;(Additional Fig. 3d), which suggested that ERS\u0026nbsp;may be closely related to TP53 mutation.\u003c/p\u003e\n\u003cp\u003eIn addition, we performed an intergroup comparative analysis of 198 drugs. After cross-validation\u003c/p\u003e\n\u003cp\u003ewith the TCGA and ICGC databases, we found that patients in the high-risk group were more\u003c/p\u003e\n\u003cp\u003esensitive\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eto five drugs, namely, ubrosertib, taselisib, pictilisib, OSI.027, and doramapimod (Additional\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eFig. 4a), but more resistant to 12 drugs, namely, BI.2536, ERK_2440, ERK_6604, mitoxantrone,\u003c/p\u003e\n\u003cp\u003ePD0325901, RO.3306, SB216763, SCH772984, selumetinib, topotecan, trametinib, and VE.822 (Additional Fig. 4b). Interestingly, four of the five sensitive drugs targeted the PI3K/mTOR signaling pathway; however, most of the resistant drugs targeted the ERK/MAPK signaling pathway (Table 1). This meant that even if they were all proliferation-related drugs, patients with high ERSriskERS risk were more likely to receive drugs targeting the PI3K/mTOR signaling pathway rather than the ERK/MAPK signaling pathway.\u003c/p\u003e\n\u003ch3\u003eERS model correlated with macrophages and endothelial cells\u003c/h3\u003e\n\u003cp\u003eFurthermore, we analyzed the differences in infiltrating immune cells between the high-risk and low-risk groups. The results revealed significant differences in a variety of immune cells, including T cells, M0 macrophages, M1 macrophages, dendritic cells, neutrophils, and mast cells. The number of M0 macrophages in the high-risk group was significantly greater than that in the low-risk group, while the number of M1 macrophages was less than that in the low-risk group. Similar results were also observed in the ICGC-JP-LIRI cohort (Fig. 3a). Moreover, ssGSEA was used to assess the correlation between immune cells and ERSrisk, and it was found that ERSrisk was significantly related to macrophages (Additional Fig. 5); therefore, we believe that ERS is closely related to macrophages in the tumor immune microenvironment of HCC.\u003c/p\u003e\n\u003cp\u003eAdditionally, we evaluated the relationship between immune checkpoints and ERS. The response to PD-1/CTLA4 immunotherapy in patients with different immunophenotypes was compared between the high-risk and low-risk groups in the TCGA-LIHC cohort. The low-risk group had a greater non-response rate to anti-CTLA4/PD-1 immunotherapy, suggesting that the high-risk group may have a better response to anti-CTLA4/PD-1 therapy (Fig. 3b). Moreover, the TIDE score of the high-risk group was lower than that of the low-risk group, while the PD-L1 score was higher than that of the low-risk group (Fig. 3c), which indicated that the high-risk group had a lower risk of evading immunotherapy and may be more effective in immune checkpoint treatment.\u003c/p\u003e\n\u003cp\u003eWe integrated the single-cell data from GEO and removed batch effects. After a standardized procedure, all cells were divided into 23 clusters. Cell type annotations were verified through the \u0026ldquo;singleR\u0026rdquo; R package and cellmarker2 database (Additional Fig. 6). The 23 clusters were labeled with eight cell types that were assigned based on the expression of cell type-specific markers: T cells, NK cells, endothelial cells, malignant cells, myeloid-derived cells, HSCs, B cells, and epithelial cells (Fig. 4a). Subsequently, the gene tag was strategically positioned, leading to the identification of significant expression of STC2 within the endothelial cell population (Fig. 4b). Consequently, the endothelial cell population was subsequently extracted, followed by in-depth subpopulation analysis. The endothelial cells were partitioned into ten distinct clusters and subsequently classified into five subtypes: PLVAP+ECs, CD9+ECs, IGFBP3+ECs, PLPP3+ECs, and TFF3+ECs (Fig. 4c; Additional Fig. 7), with predominant expression of STC2 observed in PLVAP+ ECs (Fig. 4d).\u003c/p\u003e\n\u003cp\u003eSince we detected significant differences in macrophages between the high-risk and low-risk groups, we reclassified myeloid-derived cells and divided them into six cell types: pDCs, DC1s, DC2s, TAM1s, TAM2s, and monocytes (Fig. 4e; Additional Fig. 8). However, the gene was not significantly expressed in the macrophages.\u003c/p\u003e\n\u003cp\u003eConsidering the differences in the expression of gene tags, we visualized the signaling pathways involving PLVAP+ ECs and found that they were closely related to each cell population (Fig. 4f). Interestingly, the THY1 signaling pathway only existed in PLVAP+ ECs and was not found in other cell populations, especially those closely related to TAM1 (Fig. 4g), which may indicate that there was a unique interaction between PLVAP+ ECs and TAM1. This may indicate that STC2 is involved in the reprogramming of endothelial cells and may regulate the reprogramming and function of macrophages through PLVAP+ECs.\u003c/p\u003e\n\u003cp\u003eThe above results suggested that ERS plays a role in endothelial cells, thereby affecting the function of macrophages.\u003c/p\u003e\n\u003ch3\u003eEstablishment and validation of the radiogenomic signature\u003c/h3\u003e\n\u003cp\u003eAlthough we already understand the important role of ERS in the occurrence and development of HCC and the TME, we hope to make early judgments on patients through non-invasive methods to better guide subsequent treatment. Utilizing resources from the TCIA databases, we embarked on a comprehensive analysis to discern specific radiological features that could predict ERSrisk. 1,766 original image features were extracted from the ROIs. After intragroup consistency analysis and intergroup consistency analysis, a total of 1,162 features whose ICC\u0026gt; 0.9 were retained. Thirty features were identified from the training cohort after Spearman\u0026rsquo;s correlation analysis and mRMR. The LASSO Cox regression algorithm further narrowed down a fusion signature that retained nine archetypal features (Fig. 5a and b). The resulting signature was formulated as follows (Fig. 5c):\u003c/p\u003e\n\u003cp\u003eradscore=\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;AP-wavelet-LHL-firstorder-Skewness*0.034045\u003c/p\u003e\n\u003cp\u003e-VP-wavelet-HHH-firstorder-TotalEnergy*0.06068\u003c/p\u003e\n\u003cp\u003e+VP-wavelet-HLL-glcm-Correlation*0.045428\u003c/p\u003e\n\u003cp\u003e-AP-wavelet-LLH-glcm-Idmn*0.038591\u003c/p\u003e\n\u003cp\u003e+AP-wavelet-HLL-gldm-DependenceVariance*0.032164\u003c/p\u003e\n\u003cp\u003e-AP-wavelet-LLH-glrlm-RunVariance*0.061446\u003c/p\u003e\n\u003cp\u003e-AP-wavelet-LHL-glcm-Imc2*0.012521\u003c/p\u003e\n\u003cp\u003e+VP-wavelet-LLL-firstorder-RootMeanSquared*0.052823\u003c/p\u003e\n\u003cp\u003e-VP-wavelet-LLH-glcm-Correlation*0.058891\u003c/p\u003e\n\u003cp\u003eThe C-index of the radiogenomic signature was 0.647 in the TCG-LIHC cohort. The radiogenomic signature was significantly associated with OS in the TCGA-LIHC cohort (Fig. 5d).\u003c/p\u003e\n\u003cp\u003eMulti-scale validation of the radiogenomic signature\u003c/p\u003e\n\u003cp\u003eWe used a surgical therapy cohort to validate the radiogenomic signature and obtained good results, with a C-index of 0.607. The prognosis of the RDS-low group was significantly better than that of the RDS-high group (Fig. 6a). Our study subsequently assessed the relationships between the radiogenomic signature and systemic combination therapy response(Fig. 6b). Interestingly, as shown in Fig. 6c and d, we found that patients in the RDS-low group (45%) had a greater objective response rate than did those in the RDS-high group (5%). In addition, the PD rate was greater in the RDS-high group than in the RDS-low group, which was also confirmed in the entire cohort (Table 2). Next, we explored the ability of the radiogenomic signature to predict systemic combination therapy response. We found that the signature had excellent performance in predicting the ORR, with an AUC of 0.785 and the DCR, with an AUC of 0.768 (Fig. 6e). Finally, Kaplan\u0026ndash;Meier plots of progression-free survival (PFS) confirmed that patients in the RDS-low group had a longer PFS than those in the RDS-high group (Fig. 6f).\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn our study, four independent risk factors for HCC prognosis, namely, STC2, BRSK2, MAGEA3, and DDX11, were identified based on ERS, and a new prognostic model for HCC was established based on these four genes. Based on this model, the ERSrisk of each sample was calculated separately by dividing the TCGA-LIHC cohort into high-risk and low-risk groups.\u003c/p\u003e\n\u003cp\u003eFurthermore, we discovered that there were notable differences in tumor immune infiltrating cells, primarily macrophages, between the high-risk and low-risk groups. In fact, ERS plays a complex role in the TME[22]. TAMs participate in IRE1-dependent ERS, which promotes cathepsin secretion and promotes a prometastatic phenotype through the synergistic effect of IL-4, IL-6, and IL-10[23]. These results indicate that ERS is involved in the regulation of TAMs in the TME. We combined GEO single-cell data and separated the data into eight categories to investigate the connection between ERS and the TME in HCC in more detail. We discovered that endothelial cells, but not macrophages, express STC2 at a relatively high level, which seems to contradict our previous results. This finding implies that endothelial cells serve as a bridge between ERS and TAMs. Endothelial cells with activated ERS participate in the distribution and functional regulation of TAMs.\u003c/p\u003e\n\u003cp\u003eAn increasing number of studies have focused on tumor-associated endothelial cells and revealed numerous important proangiogenic molecules, such as vascular endothelial growth factor (VEGF), which may be targets for antiangiogenic therapy[24]. Previous studies have suggested a key function for tumor-associated endothelial cells in the growth of malignancies[25, 26]. The expression of VEGF and the development of new blood vessels can be promoted in the tumor microenvironment by conditions such as hypoxia and nutritional deprivation, which can trigger ERS and activate the three traditional signaling pathways of the unfolded protein response[27]. In addition to being controlled by the PERK/ATF4 signaling pathway[28], HIF can also promote the expression of STC2 in the hypoxic environment of solid tumors[29]. STC2 can promote angiogenesis by both affecting the expression and phosphorylation of VEGFR2 and participating in the VEGFR2/AKT/eNOS axis[30].\u003c/p\u003e\n\u003cp\u003eThrough further analysis of endothelial cell subpopulations, we found that STC2 was present mainly in PLVAP+ECs. Interestingly, PLVAP+ECs are considered fetal liver-associated ECs, while PLVAP is known to play an important role in liver development and disease[31-33]. The expression of PLVAP is regulated by a variety of substances and mechanisms, among which VEGF plays a key role. VEGF regulates PLVAP expression in a VEGF receptor 2 (VEGFR2)-dependent manner. The PI3K/p38MAPK and MEK1/ERK1/2 pathways are two downstream cascades that can be triggered by VEGF signaling via VEGFR2[34, 35]. STC2 is involved in the phosphorylation of ERK1/2 and VEGFR2, as well as the activation of PI3K. PI3K can initiate multiple cascades and has multiple downstream effects, including p38MAPK and Akt. These substances are believed to be involved in the regulation of PLVAP[34-36]. Surprisingly, this finding is similar to our drug sensitivity results in that the high-risk group was more sensitive to PI3K- and p38-targeted drugs. Based on these observations, it is highly likely that STC2 regulates PLVAP expression via a variety of pathways, including the VEGFR2/PI3K/p38 MAPK pathway and the VEGFR2/MEK/ERK1/2 pathway.\u003c/p\u003e\n\u003cp\u003eAnalysis of cell communication between endothelial cells and myeloid-derived cells revealed that this is a large network. Interestingly, among these different cell populations, the THY1 signaling pathway was only expressed in tandem between PLVAP+ ECs and other cell populations, especially TAM1. TAM1 has fetal liver macrophage characteristics[37], and fetal-like PLVAP+ ECs affect fetal-like liver macrophage distribution[31], which indicates that PLVAP+ ECs might affect the layout of TAM1 in some way. Studies have shown that PLVAP+ ECs and TAM1 colocalize in the tumor microenvironment and have immunosuppressive effects, which strongly suggests that ERS mediates the recruitment of PLVAP+ ECs in the HCC tumor microenvironment to TAM1 around new blood vessels through the THY1/MAC1 signaling pathway[37], and provides an immunosuppressive microenvironment for tumor metastasis.\u003c/p\u003e\n\u003cp\u003eRadiogenomics, as an emerging field, holds great potential for various applications[38, 39]. Notably, our research revealed the significant role of ERS in the occurrence and development of HCC. This discovery may pave the way for a new targeted treatment approach for HCC. To facilitate its clinical application, we applied radiogenomics and developed a radiogenomic signature using non-invasive radiomics techniques, which demonstrated promising benefits. The radiomic features that constitute the radiogenomic signature are all wavelet features, which reflect its high stability and efficiency in reflecting tumor heterogeneity. Simultaneously, there are features from the arterial phase and the venous phase, which reflects that enhanced scans at different phases are of great value in assessing the heterogeneity of HCC, which is consistent with previous studies[40, 41].\u003c/p\u003e\n\u003cp\u003eFor patients with unresectable HCC, the preferred treatment option is systemic combination therapy, which includes immunotherapy + targeted therapy + interventional therapy[42, 43]. However, due to individual differences, everyone has different responses to systemic combination therapy, so there is an urgent need for a method that can predict therapeutic response. Here, we demonstrated that ERS is closely related to PD-1 and VEGF, so we constructed a radiogenomic signature based on ERS. The results showed that the RDS high-risk group had a worse response to systemic combination therapy and a shorter PFS, which seems to contradict the results of our previous study. However, after receiving systemic combination therapy, ERS is activated, which leads to immune evasion and enhanced expression and activity of signaling pathways such as VEGF. Therefore, the subgroup with high expression of ERS genes is more likely to develop drug resistance[13], which in turn affects the effectiveness of systemic combination therapy and patient prognosis. These results show that our radiogenomic signature can accurately predict patient response to systemic combination therapy.\u003c/p\u003e\n\u003cp\u003eHowever, our study has several limitations. Our research primarily focused on genomics and radiomics and lacked experimental validation. This is a key aspect of our future research plans. Additionally, the sample size in our study was insufficient, potentially impacting the statistical significance of the clinical data.\u003c/p\u003e\n\u003cp\u003eIn summary, we established an ERS prognostic model that can predict patient prognosis. We also found that ERS is closely related to the tumor microenvironment and is mainly manifested in the interaction between tumor-associated endothelial cells and tumor-associated macrophages. Moreover, we constructed a radiogenomic signature based on the ERS. This signature can guide subsequent combination therapy for patients with unresectable HCC.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eHCC:Hepatocellular carcinoma\u003c/p\u003e\n\u003cp\u003ePD-1:Programmed cell death-1\u003c/p\u003e\n\u003cp\u003eFDA:Food and Drug Administration\u003c/p\u003e\n\u003cp\u003eERS:Endoplasmic reticulum stress\u003c/p\u003e\n\u003cp\u003eTPM:Transcripts per million reads\u003c/p\u003e\n\u003cp\u003eCT:Computed tomography\u003c/p\u003e\n\u003cp\u003eTCGA:The Cancer Genome Atlas Program\u003c/p\u003e\n\u003cp\u003eICGC:International cancer genome consortium\u003c/p\u003e\n\u003cp\u003eGEO:Gene Expression Omnibus\u003c/p\u003e\n\u003cp\u003eTCIA:The Cancer Imaging Archive\u003c/p\u003e\n\u003cp\u003eGDSC:The Genomics of \u0026nbsp;Drug Sensitivity in Cancer\u003c/p\u003e\n\u003cp\u003eRNA:Ribonucleic Acid\u003c/p\u003e\n\u003cp\u003eTACE:Transcatheter arterial chemoembolization\u003c/p\u003e\n\u003cp\u003eVEGF:vascular endothelial growth factor\u003c/p\u003e\n\u003cp\u003essGSEA:Single-sample gene set enrichment analysis\u003c/p\u003e\n\u003cp\u003eROI:Region of interest\u003c/p\u003e\n\u003cp\u003emRMR:Maximum relevance minimum redundancy\u003c/p\u003e\n\u003cp\u003eOS:Overall survival\u003c/p\u003e\n\u003cp\u003eCR:Complete response\u003c/p\u003e\n\u003cp\u003ePR:partial response\u003c/p\u003e\n\u003cp\u003eSD:stable disease (SD)\u003c/p\u003e\n\u003cp\u003ePD:progressive disease (PD)\u003c/p\u003e\n\u003cp\u003eORR:objective response rate (ORR)\u003c/p\u003e\n\u003cp\u003eDCR:disease control rate (DCR)\u003c/p\u003e\n\u003cp\u003eRECIST:Response Evaluation Criteria in Solid Tumors (RECIST)\u003c/p\u003e\n\u003cp\u003eDEG:Differential Expression Analysis，DEG\u003c/p\u003e\n\u003cp\u003eSTC2:Stanniocalcin-2\u003c/p\u003e\n\u003cp\u003eDDX11:DEAD/H-Box Helicase 11\u003c/p\u003e\n\u003cp\u003eMAGEA3:Melanoma-Associated Antigen 3\u003c/p\u003e\n\u003cp\u003eBRSK2:BR Serine/Threonine-Protein Kinase 2\u003c/p\u003e\n\u003cp\u003eROC:Receiver Operating Characteristic\u003c/p\u003e\n\u003cp\u003eAUC:area under the curve (AUC)\u003c/p\u003e\n\u003cp\u003eTIDE:Tumor Immune Dysfunction and Exclusion\u003c/p\u003e\n\u003cp\u003ePLVAP:Plasmalemma Vesicle Associated Protein\u003c/p\u003e\n\u003cp\u003eTAM:Tumor-associated macrophages\u003c/p\u003e\n\u003cp\u003eLASSO:Least absolute shrinkage and selection operator\u003c/p\u003e\n\u003cp\u003eRDS:radiomics score\u003c/p\u003e\n\u003cp\u003ePFS:progression-free survival\u003c/p\u003e\n\u003cp\u003eTME:Tumor Mircroenviroment\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eEthics approval and consent to participate：Ethics approval by the First Affiliated Hospital of Nanjing Medical University Ethics Committee (Ethics number: 2024-SR-368).\u003c/p\u003e\n\u003cp\u003eConsent for publication:Not applicable.\u003c/p\u003e\n\u003cp\u003eAvailability of data and materials:The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003eCompeting interests:All the authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003eFunding:This study was funded by the Jiangsu Provincial Key Research and Development Program (Social Development) under project number BE2020708 and National Natural Science Foundation of China (82102150) and the Natural Science Foundation of Jiangsu Province (BK20210968).\u003c/p\u003e\n\u003cp\u003eAuthors\u0026apos; contributions:W.H.Y. designed, performed, and analyzed experiments, and wrote the manuscript. C.S.Y. and W.T. contributed to the bioinformatic analyses.W.K.，J.G.W. and X.Z.G. helped in the project design, supervised the progress of the study, and edited the manuscript. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003eAcknowledgements:Fundings from the Jiangsu Provincial Key Research and Development Program (Social Development) under project number BE2020708 and National Natural Science Foundation of China (82102150) and the Natural Science Foundation of Jiangsu Province (BK20210968).\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eSung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, Bray F: Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 2021, 71(3):209-249.\u003c/li\u003e\n\u003cli\u003eSingal AG, Lampertico P, Nahon P: Epidemiology and surveillance for hepatocellular carcinoma: New trends. J Hepatol 2020, 72(2):250-261.\u003c/li\u003e\n\u003cli\u003eMarin JJG, Macias RIR, Monte MJ, Romero MR, Asensio M, Sanchez-Martin A, Cives-Losada C, Temprano AG, Espinosa-Escudero R, Reviejo M et al: Molecular Bases of Drug Resistance in Hepatocellular Carcinoma. Cancers (Basel) 2020, 12(6).\u003c/li\u003e\n\u003cli\u003eQin S, Bai Y, Lim HY, Thongprasert S, Chao Y, Fan J, Yang TS, Bhudhisawasdi V, Kang WK, Zhou Y et al: Randomized, multicenter, open-label study of oxaliplatin plus fluorouracil/leucovorin versus doxorubicin as palliative chemotherapy in patients with advanced hepatocellular carcinoma from Asia. J Clin Oncol 2013, 31(28):3501-3508.\u003c/li\u003e\n\u003cli\u003eZhu AX, Finn RS, Edeline J, Cattan S, Ogasawara S, Palmer D, Verslype C, Zagonel V, Fartoux L, Vogel A et al: Pembrolizumab in patients with advanced hepatocellular carcinoma previously treated with sorafenib (KEYNOTE-224): a non-randomised, open-label phase 2 trial. Lancet Oncol 2018, 19(7):940-952.\u003c/li\u003e\n\u003cli\u003eZhu AX, Abbas AR, de Galarreta MR, Guan Y, Lu S, Koeppen H, Zhang W, Hsu CH, He AR, Ryoo BY et al: Molecular correlates of clinical response and resistance to atezolizumab in combination with bevacizumab in advanced hepatocellular carcinoma. Nat Med 2022, 28(8):1599-1611.\u003c/li\u003e\n\u003cli\u003eda Silva DC, Valent\u0026atilde;o P, Andrade PB, Pereira DM: Endoplasmic reticulum stress signaling in cancer and neurodegenerative disorders: Tools and strategies to understand its complexity. Pharmacol Res 2020, 155:104702.\u003c/li\u003e\n\u003cli\u003eWu SX, Ye SS, Hong YX, Chen Y, Wang B, Lin XJ, Lin X: Hepatitis B Virus Small Envelope Protein Promotes Hepatocellular Carcinoma Angiogenesis via Endoplasmic Reticulum Stress Signaling To Upregulate the Expression of Vascular Endothelial Growth Factor A. Journal of virology 2022, 96(4):e0197521.\u003c/li\u003e\n\u003cli\u003eVandewynckel YP, Laukens D, Bogaerts E, Paridaens A, Van den Bussche A, Verhelst X, Van Steenkiste C, Descamps B, Vanhove C, Libbrecht L et al: Modulation of the unfolded protein response impedes tumor cell adaptation to proteotoxic stress: a PERK for hepatocellular carcinoma therapy. Hepatol Int 2015, 9(1):93-104.\u003c/li\u003e\n\u003cli\u003eLiu P, Wei J, Mao F, Xin Z, Duan H, Du Y, Wang X, Li Z, Qian J, Yao J: Establishment of a Prognostic Model for Hepatocellular Carcinoma Based on Endoplasmic Reticulum Stress-Related Gene Analysis. Front Oncol 2021, 11:641487.\u003c/li\u003e\n\u003cli\u003eChen X, Cubillos-Ruiz JR: Endoplasmic reticulum stress signals in the tumour and its microenvironment. Nat Rev Cancer 2021, 21(2):71-88.\u003c/li\u003e\n\u003cli\u003eBao MH, Wong CC: Hypoxia, Metabolic Reprogramming, and Drug Resistance in Liver Cancer. Cells 2021, 10(7).\u003c/li\u003e\n\u003cli\u003eKhaled J, Kopsida M, Lennern\u0026auml;s H, Heindryckx F: Drug Resistance and Endoplasmic Reticulum Stress in Hepatocellular Carcinoma. Cells 2022, 11(4).\u003c/li\u003e\n\u003cli\u003eLiu J, Fan L, Yu H, Zhang J, He Y, Feng D, Wang F, Li X, Liu Q, Li Y et al: Endoplasmic Reticulum Stress Causes Liver Cancer Cells to Release Exosomal miR-23a-3p and Up-regulate Programmed Death Ligand 1 Expression in Macrophages. Hepatology 2019, 70(1):241-258.\u003c/li\u003e\n\u003cli\u003eMing W, Zhu Y, Bai Y, Gu W, Li F, Hu Z, Xia T, Dai Z, Yu X, Li H et al: Radiogenomics analysis reveals the associations of dynamic contrast-enhanced-MRI features with gene expression characteristics, PAM50 subtypes, and prognosis of breast cancer. Front Oncol 2022, 12:943326.\u003c/li\u003e\n\u003cli\u003eLambin P, Leijenaar RTH, Deist TM, Peerlings J, de Jong EEC, van Timmeren J, Sanduleanu S, Larue R, Even AJG, Jochems A et al: Radiomics: the bridge between medical imaging and personalized medicine. Nat Rev Clin Oncol 2017, 14(12):749-762.\u003c/li\u003e\n\u003cli\u003ePinker K, Chin J, Melsaether AN, Morris EA, Moy L: Precision Medicine and Radiogenomics in Breast Cancer: New Approaches toward Diagnosis and Treatment. Radiology 2018, 287(3):732-747.\u003c/li\u003e\n\u003cli\u003eHu Y, Xie C, Yang H, Ho JWK, Wen J, Han L, Chiu KWH, Fu J, Vardhanabhuti V: Assessment of Intratumoral and Peritumoral Computed Tomography Radiomics for Predicting Pathological Complete Response to Neoadjuvant Chemoradiation in Patients With Esophageal Squamous Cell Carcinoma. JAMA Netw Open 2020, 3(9):e2015927.\u003c/li\u003e\n\u003cli\u003eKim S, Shin J, Kim DY, Choi GH, Kim MJ, Choi JY: Radiomics on Gadoxetic Acid-Enhanced Magnetic Resonance Imaging for Prediction of Postoperative Early and Late Recurrence of Single Hepatocellular Carcinoma. Clin Cancer Res 2019, 25(13):3847-3855.\u003c/li\u003e\n\u003cli\u003eXu T, Ren L, Liao M, Zhao B, Wei R, Zhou Z, He Y, Zhang H, Chen D, Chen H et al: Preoperative Radiomics Analysis of Contrast-Enhanced CT for Microvascular Invasion and Prognosis Stratification in Hepatocellular Carcinoma. J Hepatocell Carcinoma 2022, 9:189-201.\u003c/li\u003e\n\u003cli\u003eZheng Z, Guo Y, Huang X, Liu J, Wang R, Qiu X, Liu S: CD8A as a Prognostic and Immunotherapy Predictive Biomarker Can Be Evaluated by MRI Radiomics Features in Bladder Cancer. Cancers (Basel) 2022, 14(19).\u003c/li\u003e\n\u003cli\u003eDi Conza G, Ho PC: ER Stress Responses: An Emerging Modulator for Innate Immunity. Cells 2020, 9(3).\u003c/li\u003e\n\u003cli\u003eYan D, Wang HW, Bowman RL, Joyce JA: STAT3 and STAT6 Signaling Pathways Synergize to Promote Cathepsin Secretion from Macrophages via IRE1\u0026alpha; Activation. Cell Rep 2016, 16(11):2914-2927.\u003c/li\u003e\n\u003cli\u003eChung AS, Ferrara N: Developmental and pathological angiogenesis. Annu Rev Cell Dev Biol 2011, 27:563-584.\u003c/li\u003e\n\u003cli\u003eXu Z, Guo C, Ye Q, Shi Y, Sun Y, Zhang J, Huang J, Huang Y, Zeng C, Zhang X et al: Endothelial deletion of SHP2 suppresses tumor angiogenesis and promotes vascular normalization. Nat Commun 2021, 12(1):6310.\u003c/li\u003e\n\u003cli\u003eZhang L, Xu J, Zhou S, Yao F, Zhang R, You W, Dai J, Yu K, Zhang Y, Baheti T et al: Endothelial DGKG promotes tumor angiogenesis and immune evasion in hepatocellular carcinoma. J Hepatol 2023.\u003c/li\u003e\n\u003cli\u003eBinet F, Sapieha P: ER Stress and Angiogenesis. Cell Metab 2015, 22(4):560-575.\u003c/li\u003e\n\u003cli\u003eQie S, Sang N: Stanniocalcin 2 (STC2): a universal tumour biomarker and a potential therapeutical target. J Exp Clin Cancer Res 2022, 41(1):161.\u003c/li\u003e\n\u003cli\u003eLaw AY, Wong CK: Stanniocalcin-2 is a HIF-1 target gene that promotes cell proliferation in hypoxia. Exp Cell Res 2010, 316(3):466-476.\u003c/li\u003e\n\u003cli\u003eRen S, Yang Y: The proliferation and angiogenesis in hemangioma-derived endothelial cells is affected by STC2 medicated VEGFR2/Akt/eNOS pathway. Pak J Med Sci 2023, 39(4):1119-1123.\u003c/li\u003e\n\u003cli\u003eRantakari P, J\u0026auml;ppinen N, Lokka E, Mokkala E, Gerke H, Peuhu E, Ivaska J, Elima K, Auvinen K, Salmi M: Fetal liver endothelium regulates the seeding of tissue-resident macrophages. Nature 2016, 538(7625):392-396.\u003c/li\u003e\n\u003cli\u003eRamachandran P, Dobie R, Wilson-Kanamori JR, Dora EF, Henderson BEP, Luu NT, Portman JR, Matchett KP, Brice M, Marwick JA et al: Resolving the fibrotic niche of human liver cirrhosis at single-cell level. Nature 2019, 575(7783):512-518.\u003c/li\u003e\n\u003cli\u003eWang YH, Cheng TY, Chen TY, Chang KM, Chuang VP, Kao KJ: Plasmalemmal Vesicle Associated Protein (PLVAP) as a therapeutic target for treatment of hepatocellular carcinoma. BMC cancer 2014, 14:815.\u003c/li\u003e\n\u003cli\u003eStrickland LA, Jubb AM, Hongo JA, Zhong F, Burwick J, Fu L, Frantz GD, Koeppen H: Plasmalemmal vesicle-associated protein (PLVAP) is expressed by tumour endothelium and is upregulated by vascular endothelial growth factor-A (VEGF). J Pathol 2005, 206(4):466-475.\u003c/li\u003e\n\u003cli\u003eHamilton BJ, Tse D, Stan RV: Phorbol esters induce PLVAP expression via VEGF and additional secreted molecules in MEK1-dependent and p38, JNK and PI3K/Akt-independent manner. J Cell Mol Med 2019, 23(2):920-933.\u003c/li\u003e\n\u003cli\u003eMa B, Xu X, He S, Zhang J, Wang X, Wu P, Liu J, Jiang H, Zheng M, Li W et al: STC2 modulates ERK1/2 signaling to suppress adipogenic differentiation of human bone marrow mesenchymal stem cells. Biochem Biophys Res Commun 2020, 524(1):163-168.\u003c/li\u003e\n\u003cli\u003eSharma A, Seow JJW, Dutertre CA, Pai R, Bl\u0026eacute;riot C, Mishra A, Wong RMM, Singh GSN, Sudhagar S, Khalilnezhad S et al: Onco-fetal Reprogramming of Endothelial Cells Drives Immunosuppressive Macrophages in Hepatocellular Carcinoma. Cell 2020, 183(2):377-394.e321.\u003c/li\u003e\n\u003cli\u003eHartmann K, Sad\u0026eacute;e CY, Satwah I, Carrillo-Perez F, Gevaert O: Imaging genomics: data fusion in uncovering disease heritability. Trends in molecular medicine 2023, 29(2):141-151.\u003c/li\u003e\n\u003cli\u003eLiu Z, Wu K, Wu B, Tang X, Yuan H, Pang H, Huang Y, Zhu X, Luo H, Qi Y: Imaging genomics for accurate diagnosis and treatment of tumors: A cutting edge overview. Biomedicine \u0026amp; pharmacotherapy = Biomedecine \u0026amp; pharmacotherapie 2021, 135:111173.\u003c/li\u003e\n\u003cli\u003eTang VH, Duong STM, Nguyen CDT, Huynh TM, Duc VT, Phan C, Le H, Bui T, Truong SQH: Wavelet radiomics features from multiphase CT images for screening hepatocellular carcinoma: analysis and comparison. Scientific reports 2023, 13(1):19559.\u003c/li\u003e\n\u003cli\u003eSoufi M, Arimura H, Nagami N: Identification of optimal mother wavelets in survival prediction of lung cancer patients using wavelet decomposition-based radiomic features. Medical physics 2018, 45(11):5116-5128.\u003c/li\u003e\n\u003cli\u003eBrown ZJ, Tsilimigras DI, Ruff SM, Mohseni A, Kamel IR, Cloyd JM, Pawlik TM: Management of Hepatocellular Carcinoma: A Review. JAMA Surg 2023, 158(4):410-420.\u003c/li\u003e\n\u003cli\u003eLlovet JM, Kelley RK, Villanueva A, Singal AG, Pikarsky E, Roayaie S, Lencioni R, Koike K, Zucman-Rossi J, Finn RS: Hepatocellular carcinoma. Nature reviews Disease primers 2021, 7(1):6.\u003c/li\u003e\n\u003cli\u003eThe Cancer Genome Atlas Program database (https://portal.gdc.cancer.gov/) Accessed 13 November 2022\u003c/li\u003e\n\u003cli\u003eInternational cancer genome consortium database (https://dcc.icgc.org/) Accessed 13 November 2022\u003c/li\u003e\n\u003cli\u003eMolecular Signatures Database (https://www.gsea-msigdb.org/gsea/msigdb/index.jsp) Accessed 13 November 2022\u003c/li\u003e\n\u003cli\u003eGEO database (Home - GEO - NCBI (nih.gov)) Accessed 13 November 2022\u003c/li\u003e\n\u003cli\u003ethe Cancer Imaging Archive (TCIA: https://www.cancerimagingarchive.net/) Accessed 6 October 2023\u003c/li\u003e\n\u003cli\u003ethe Genomics of Drug Sensitivity in Cancer (GDSC: https://www.cancerrxgene.org/) database Accessed 6 October 2023\u003c/li\u003e\n\u003cli\u003eThe CellMarker (CellMarker2.0 (hrbmu.edu.cn) database Accessed 8 April 2023\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003e\u003cstrong\u003eTable 1\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cstrong\u003eDrug sensitivity analysis\u003c/strong\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" align=\"\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eDrug\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eTargets\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.571428571428573%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eSignaling pathway\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.518987341772153%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eSensitivity.\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eUprosertib\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eAKT1, AKT2, AKT3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.571428571428573%\" valign=\"top\"\u003e\n \u003cp\u003ePI3K/MTOR signaling\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.518987341772153%\" valign=\"top\"\u003e\n \u003cp\u003esensitive\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eTaselisib\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003ePI3K (beta sparing)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.571428571428573%\" valign=\"top\"\u003e\n \u003cp\u003ePI3K/MTOR signaling\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.518987341772153%\" valign=\"top\"\u003e\n \u003cp\u003esensitive\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003ePictilisib\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003ePI3K (class 1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.571428571428573%\" valign=\"top\"\u003e\n \u003cp\u003ePI3K/MTOR signaling\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.518987341772153%\" valign=\"top\"\u003e\n \u003cp\u003esensitive\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eOSI. 027\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eMTORC1, MTORC2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.571428571428573%\" valign=\"top\"\u003e\n \u003cp\u003ePI3K/MTOR signaling\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.518987341772153%\" valign=\"top\"\u003e\n \u003cp\u003esensitive\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eDoramapimod\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003ep38, JNK2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.571428571428573%\" valign=\"top\"\u003e\n \u003cp\u003eJNK and p38 signaling\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.518987341772153%\" valign=\"top\"\u003e\n \u003cp\u003esensitive\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eBI.2536\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003ePLK1, PLK2, PLK3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.571428571428573%\" valign=\"top\"\u003e\n \u003cp\u003eCell cycle\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.518987341772153%\" valign=\"top\"\u003e\n \u003cp\u003eresistant\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eERK_2440\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eERK1,ERK2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.571428571428573%\" valign=\"top\"\u003e\n \u003cp\u003eERK MAPK signaling\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.518987341772153%\" valign=\"top\"\u003e\n \u003cp\u003eresistant\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eERK_6604\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eERK1,ERK2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.571428571428573%\" valign=\"top\"\u003e\n \u003cp\u003eERK MAPK signaling\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.518987341772153%\" valign=\"top\"\u003e\n \u003cp\u003eresistant\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eMitoxantrone\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eTOP2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.571428571428573%\" valign=\"top\"\u003e\n \u003cp\u003eDNA replication\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.518987341772153%\" valign=\"top\"\u003e\n \u003cp\u003eresistant\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003ePD0325901\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eMEK1, MEK2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.571428571428573%\" valign=\"top\"\u003e\n \u003cp\u003eERK MAPK signaling\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.518987341772153%\" valign=\"top\"\u003e\n \u003cp\u003eresistant\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eRO.3306\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eCDK1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.571428571428573%\" valign=\"top\"\u003e\n \u003cp\u003eCell cycle\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.518987341772153%\" valign=\"top\"\u003e\n \u003cp\u003eresistant\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eSB216763\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eGSK3A, GSK3B\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.571428571428573%\" valign=\"top\"\u003e\n \u003cp\u003eWNT signaling\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.518987341772153%\" valign=\"top\"\u003e\n \u003cp\u003eresistant\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eSCH772984\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eERK1, ERK2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.571428571428573%\" valign=\"top\"\u003e\n \u003cp\u003eERK MAPK signaling\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.518987341772153%\" valign=\"top\"\u003e\n \u003cp\u003eresistant\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eSelumetinib\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eMEK1, MEK2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.571428571428573%\" valign=\"top\"\u003e\n \u003cp\u003eERK MAPK signaling\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.518987341772153%\" valign=\"top\"\u003e\n \u003cp\u003eresistant\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eTopotecan\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eTOP1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.571428571428573%\" valign=\"top\"\u003e\n \u003cp\u003eDNA replication\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.518987341772153%\" valign=\"top\"\u003e\n \u003cp\u003eresistant\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eTrametinib\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eMEK1, MEK2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.571428571428573%\" valign=\"top\"\u003e\n \u003cp\u003eERK MAPK signaling\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.518987341772153%\" valign=\"top\"\u003e\n \u003cp\u003eresistant\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eVE.822\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.95479204339964%\" valign=\"top\"\u003e\n \u003cp\u003eATR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.571428571428573%\" valign=\"top\"\u003e\n \u003cp\u003eGenome integrity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"21.518987341772153%\" valign=\"top\"\u003e\n \u003cp\u003eresistant\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTherapy response of \u0026nbsp;RDSgroup in the systemic combination therapy cohort\u003c/strong\u003e\u003c/p\u003e\n\u003cdiv align=\"\"\u003e\n \u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd width=\"22.264875239923224%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"25.719769673704416%\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp; \u0026nbsp;high \u0026nbsp;\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.103646833013435%\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp; \u0026nbsp;low \u0026nbsp;\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"25.911708253358924%\"\u003e\n \u003cp\u003e\u003cstrong\u003eP\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"22.264875239923224%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"25.719769673704416%\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp;\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;N=20\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp;\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.103646833013435%\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;N=20\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"25.911708253358924%\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"22.264875239923224%\"\u003e\n \u003cp\u003e\u003cstrong\u003eResponse\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"25.719769673704416%\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.103646833013435%\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"25.911708253358924%\"\u003e\n \u003cp\u003e\u0026nbsp; 0.009 \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"22.264875239923224%\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; CR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"25.719769673704416%\"\u003e\n \u003cp\u003e0 (0.00%)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.103646833013435%\"\u003e\n \u003cp\u003e2 (10.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"25.911708253358924%\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"22.264875239923224%\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; PR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"25.719769673704416%\"\u003e\n \u003cp\u003e1 (5.00%)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.103646833013435%\"\u003e\n \u003cp\u003e7 (35.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"25.911708253358924%\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"22.264875239923224%\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"25.719769673704416%\"\u003e\n \u003cp\u003e10 (50.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.103646833013435%\"\u003e\n \u003cp\u003e9 (45.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"25.911708253358924%\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"22.264875239923224%\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; PD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"25.719769673704416%\"\u003e\n \u003cp\u003e9 (45.0%)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.103646833013435%\"\u003e\n \u003cp\u003e2 (10.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"25.911708253358924%\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"radiogenomics, endoplasmic reticulum stress (ERS), hepatocellular carcinoma (HCC), The Cancer Genome Atlas（TCGA）, radiomics, single-cell","lastPublishedDoi":"10.21203/rs.3.rs-4535127/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4535127/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground:\u003c/strong\u003eHepatocellular carcinomais one of the most common tumors worldwide. Various factors in the tumor microenvironment can lead to the activation of endoplasmic reticulum stress, thereby affecting the occurrence and development of tumors.The objective of our study was to develop and validate a radiogenomic signature based on endoplasmic reticulum stress to predict prognosis and systemic combination therapy response\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods:\u003c/strong\u003eUsing data from TCGA as a training cohort and data from ICGC as a testing cohort. Univariate Cox regression and multivariate Cox regression analysis were used to identify prognostic-related genes and construct a model. Hepatocellular carcinoma single-cell data obtained from GEO were used to map gene signatures and explore inter-cellular signaling communications. Finally, a radiogenomic signature was used to predict the objective response rate and overall survival.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults:\u003c/strong\u003eA total of four gene signatures related to ERS (including STC2, MAGEA3, BRSK2, DDX11) were identified. Macrophages were significantly different between high-risk and low-risk groups. The high-risk group showed higher PD-1 and mutated TP53 scores. Drug sensitivity analysis found that most sensitive drugs target the PI3K/mTOR signaling pathway. Further research revealed the expression of STC2 in the endothelial cells, particularly PLVAP+ endothelial cells, and may regulate the reprogramming and function of macrophages. Furthermore, we identified nine radiomic features and established a radiogenomic signature based on ERS that can predict prognosis and response to systemic combination therapy. This signature can guide systemic combination therapy for patients with unresectable hepatocellular carcinoma.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions:\u003c/strong\u003eWe established an ERS prognostic model that can predict patient prognosis. We also found that ERS is closely related to the tumor microenvironment and is mainly manifested in the interaction between tumor-associated endothelial cells and tumor-associated macrophages. Moreover, we constructed a radiogenomic signature based on the ERS. This signature can guide subsequent combination therapy for patients with unresectable HCC.\u003c/p\u003e","manuscriptTitle":"Construction of a radiogenomic signature based on endoplasmic reticulum stress for predicting prognosis and systemic combination therapy response in hepatocellular carcinoma","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-06-26 18:56:26","doi":"10.21203/rs.3.rs-4535127/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"9921243a-ebaf-48e5-856b-e0d222de4202","owner":[],"postedDate":"June 26th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2024-09-30T08:38:43+00:00","versionOfRecord":[],"versionCreatedAt":"2024-06-26 18:56:26","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4535127","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4535127","identity":"rs-4535127","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}