The immune-related gene-based prognostic index predicts the prognosis and effect of immune checkpoint inhibitors in hepatocellular carcinoma

doi:10.21203/rs.3.rs-2092544/v1

Download PDF

Research Article

The immune-related gene-based prognostic index predicts the prognosis and effect of immune checkpoint inhibitors in hepatocellular carcinoma

https://doi.org/10.21203/rs.3.rs-2092544/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background: Immune-related genes have been confirmed to be related to tumor development and anti-tumor immune response. Therefore, immune-related genes are potential candidate for predicting prognosis and immunotherapy response. We aimed to explore the relationship between immune-related genes and the prognosis, tumor microenvironment and response to immunotherapy in HCC.

Methods: The immune-related gene-based prognostic index (IRGPI) is constructed by 4 hub genes that are most closely related to prognosis among 1.811 immune related genes. We then sequentially compared the differences in clinical characteristics, driver gene mutations, immune microenvironment and immunotherapy between the two groups of IRGPI. Finally, multiplex immunofluorescence staining was used to assess the abundance of infiltrating CD8⁺ T cells in tumor microenvironment between different subgroups.

Results: IRGPI-high significantly related to higher disease progression and shorter survival time. IRGPI is accompanied by a higher TMB and frequency of driver gene mutations that predicts worse prognosis. In tumor microenvironment, the number and activity of tumor-infiltrating immune cells in the IRGPI-low group was significantly greater than that of the IRGPI-high. TIDE and three immunotherapy datasets showed that patients in the IRGPI-low group are more in an immune dysfunction state and respond well to ICIs. Immunostaining on the HCC cohort further confirmed that CD8⁺ T cells were more abundant in IRGPI-low group, while the tumor microenvironment of IRGPI-high group presents a scene of immune desert, which corresponds to our results and partly explains why IRGPI-high group cannot respond well to ICIs.

Conclusions: IRGPI was created based on the immune-related genes in HCC. The novel prognostic model can predict prognosis and response to ICIs in HCC patients.

immune-related genes-based prognostic index

prognostic model

hepatocellular carcinoma

immune checkpoint inhibitors

the tumor microenvironment

multiplex immunofluorescence

Hepatocellular carcinoma (HCC) is the most prevalent primary liver cancer and a leading cause of cancer-associated deaths across the globe. The incidence of hepatocellular carcinoma ranked sixth among all malignant tumors worldwide in 2020, although the mortality rate ranked third [1]. Due to the gradual development of HCC, most patients have indeed missed the opportunity for surgery by the time they are diagnosed with HCC, resulting in a five-year survival rate of less than 30% across the globe. China, the country with the highest number of HCC cases, has a five-year survival rate of only 14.1% [2]. Early-stage patients who undergo surgical resection are often accompanied by a high recurrence rate [3]. Hence, developing new diagnostic and therapeutic strategies to predict and improve HCC prognosis is necessary.

Several breakthroughs have been made in the field of HCC treatment in recent years. Immunotherapy represented by immune checkpoint inhibitors (ICIs) has gained considerable attention in the treatment of HCC [4]. Meta-analyses have shown that ICIs can significantly improve overall survival (OS), progression-free survival (PFS), and overall response rate (ORR) compared with standard therapies [5, 6]. The US Food and Drug Administration (FDA) has approved the targeted drug lenvatinib and the PD-1 blocker pembrolizumab for the first-line treatment of patients with advanced unresectable HCC that is not suitable for localized treatment [7]. Based on the results of IMbrave150, the National Medical Products Administration (NMPA) of China approved the combination of atezolizumab and bevacizumab for the treatment of patients with unresectable HCC who had not previously received systemic treatment [8]. Immune checkpoint inhibitors can block the immune escape of tumor cells and restore the normal tumor-killing activity of the immune system. ICIs provide better therapeutic results and have fewer side effects as compared to standard therapeutic approaches like chemotherapy and radiotherapy [9]. However, only a small number of patients respond to ICIs. In HCC, the response rate to ICIs monotherapy is 15%-23%, which increases to about 30% after combination therapy [10]. Because the cost of ICIs is so high, many patients cannot afford them. Hence, there is a need to explore for identifying ideal cancer patients who may benefit from ICIs.

Conventional biomarkers for assessing ICI efficacy, such as PD-L1 expression, tumor mutation burden (TMB), and microsatellite instability (MSI), are less reliable [11]. Moreover, the efficacy of ICIs depends not only on the features of tumor cells, but also on triggering the immune system to develop a long-lasting antitumor response [12]. Reported studies have revealed that the development of immune-related adverse events (irAEs) was associated with clinical benefits for HCC patients who were treated with ICIs [13]. Tumor-infiltrating lymphocytes (TILs) are an important component in the tumor microenvironment and that facilitates the anti-tumor immune response. The density of TIL in the tumor microenvironment has been verified to be closely related to the efficacy of ICIs [14]. In addition, the upregulation of immune-related genes such as T cell and NK cell proliferation genes indicates that ICIs have a more beneficial effect, which suggests that immune-related genes may serve as potential biomarkers for predicting the effectiveness of ICIs [15, 16]. Although previous studies have proved the value of Immune-related genes in the prognostic evaluation of HCC [17–20], few studies have focused on the value of immune-related genes in evaluating the effect of ICIs. In this study, we established an Immune-Related Gene-based Prognostic Index (IRGPI) that can not only assess the prognosis of the patient, but also evaluate clinical characteristics, TILs, immune subtype and predict the response to ICIs in HCC. In addition, we investigated its interaction with conventional biomarkers and predict that IRGPI can be utilized as an adjunct to conventional biomarkers to give a more accurate basis for the use of ICIs in patients with HCC.

Data collection and integration

We downloaded an RNA-seq transcriptome profiling dataset including 374 HCC and 50 normal samples, somatic structural variation and matching clinical information including age, sex, stage, tumor-node-metastasis classification over survival time and survival status of HCC from TCGA (https://portal.gdc.cancer.gov/). After removing invalid or partial data from the TCGA database, a total of 365 HCC patient transcriptome and clinical data were included in the training set for subsequent analysis.

Moreover, 1,811 unique immune-related genes (IRGs) were obtained from the Immunology Database and Analysis Portal (ImmPort) database (https://www.immport.org/home) [21].

IMvigor210 dataset contains the transcriptome data and clinical information of 298 urothelial cancer patients who received PD-L1 antibodies. The dataset can be downloaded from http://research-pub.gene.com/IMvigor210CoreBiologies. In addition, we also collected information from 39 patients who used PD-1 antibodies from GSM1648114 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSM1648114) and GSE18220 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE78220) datasets. We analyzed these patients to determine whether the model can predict the effectiveness of ICIs.

Biological function analysis of differentially expressed immune-related genes (DEIRGs)We retrieved immune-related gene expression data of HCC patients from the TCGA databases after gene name conversion and rectification of transcriptome data. DEIRGs between tumor tissues and normal tissues were identified by R package “limma” [22]. Heatmaps were drawn to visualize the differential expression of immune-related genes between tumor and normal tissues.

Next, we performed functional enrichment analysis on these genes to clarify the biological functions of DEIRGs, including Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG). According to the criteria of false discovery rate (FDR) < 0.05, the top 10 most significant GO terms and KEGG signaling pathways were visualized by the R package “ggplot2” [23].

Weighted gene co-expression network analysis (WGCNA)

We evaluated the 555 DEIRGs and constructed a gene co-expression network by R package “WGCNA” [24]. Genes with a high level of topological overlap similarity would be integrated into a module in this network. Genes in the same module usually have a high degree of co-expression. In this study, we used two methods to identify the modular genes that have the most significant impact on clinical characteristics. The module eigengene (ME) represents the first principal component of the module, which is usually used to describe the expression pattern of the module. Module membership (MM) refers to the correlation coefficient between each gene in the same module, which is usually used to describe the reliability of a gene belonging to a module. Finally, we calculated the correlation between each module and the clinical characteristics to determine the module genes closely relevant to the clinical characteristics for subsequent analysis.

Immune-related gene signature development and reliability evaluation

To explore IRGs highly related to OS and assess the prognostic evaluation, univariate Cox proportional hazard regression analysis was performed. With the cutoff value of P < 0.05, the prognosis-related IRGs were identified. of the optional IRGPI model based on prognosis-related IRGs was constructed using the Least Absolute Shrinkage and Selection Operator (LASSO) penalized Cox proportional hazards regression via R package ”glmnet” [25]. The IRGPI score of each HCC patient was calculated by the following formula:

(Herein, Coefi is the coefficient of each selected gene, while Xi is the expression value of IRGs.)

Each patient’s IRGPI score can be calculated using this formula. Patients with IRGPI score ≥ median value were classified as the IRGPI-high group, while those patients whose IRGPI score < median value were classified as the IRGPI-low group. Then, using the R package “survival”, a Kaplan-Meier analysis was used to compare the survival of IRGPI-high and IRGPI-low groups [26]. The IRGPI model’s independent prognostic relevance was further investigated using multivariate Cox regression.

Analysis of gene mutation between IRGPI-high and low group

Based on the somatic mutation data downloaded from the TCGA database, we calculated the total number of non-synonymous mutations of each patient to obtain the TMB. The HCC driver genes were identified by the R package “maftool” [27]. We evaluated the frequency of driver gene mutations in the IRGPI-high and IRGPI-low groups, and the top 20 genes with the highest mutation frequency were designated as potential driver genes for HCC.

Evaluation of tumor-infiltrating immune cells and immune function in the tumor microenvironment

The R package “CIBERSORT”, which is based on the principle of linear support vector regression, was used to calculate infiltration levels for different immune cells and immune function in HCC [28]. Immune and stromal contents for each HCC patient were evaluated by ESTIMATE [29]. The hierarchical agglomerative clustering of HCC was executed as per the tumor-infiltrating immune cells of each patient. Finally, a box plot was used to show the difference in immune infiltrating cells and immunological function between the IRGPI-high and IRGPI-low groups.

Tumor Immune Dysfunction and Exclusion (TIDE)

TIDE is a computational framework to identify factors that underlie immune-dysfunction and immunosuppression of tumor immune escape. The TIDE score calculated by the computational framework consists of two parts: dysfunction score and exclusion score. The dysfunction and exclusion score can be calculated by multiplying the expression of immune-dysfunction and immunosuppression genes by their respective weight coefficients. Compared with the currently widely used biomarkers for evaluating the efficacy of immune checkpoint inhibitors (TMB, PD-L1 expression and IFN-γ), the TIDE score can better evaluate the efficacy of anti-PD-1 and anti-CTLA-4 treatments. To test the efficacy of ICIs, we calculated TIDE scores for patients in the IRGPI-high and low groups to examine whether there were any differences in TIDE scores between subgroups. The underlined study was conducted using the TIDE online application of ICIs response prediction, which is freely accessible with any modern web browser http://tide.dfci.harvard.edu/ [30].

Multiplex immunofluorescence staining and quantitative analysis

40 surgical specimens of HCC from the Shandong Cancer Hospital and Institute and performed multiplex immunofluorescence staining (Table 1). Tissue sections 4 μm thick were deparaffinized in xylene and then rehydrated in 100, 90, and 70% alcohol successively. Antigen retrieval was performed with boiling in antigen retrieval solution EDTA, endogenous peroxidase was inactivated by incubation in 3% H₂O₂ for 15min. Next, the sections were pre-incubated with 10% normal goat serum and then incubated two hours or overnight with primary antibodies: MAPT (1:100 dilution, ab92676, Abcam), GHR (1:200 dilution, ab209790, Abcam), CD5L (1:200 dilution, ab45408, Abcam), CD8 (1:300 dilution, ab199016, Abcam), CCL14 (1:150 dilution, 14216-1-AP, proteintech). Subsequently, the sections were incubated with anti-mouse or anti-rabbit HRP-conjugated Polymer (Vector Lab, CA) for 10min at room temperature. The antigenic binding sites were visualized using the OPAL dye. OPAL-520 (PerkinElmer Inc.), OPAL-690(PerkinElmer Inc.), OPAL-570(PerkinElmer Inc.), OPAL-780(PerkinElmer Inc.), OPAL-620(PerkinElmer Inc.) were applied to each antibody, respectively. After staining, all slides were counter-stained with DAPI for five minutes and mounted in Pro-Long Diamond Antifade Mountant (Thermo Fisher).

To obtain multispectral images, the stained slides were scanned using the Akoya CODEX system (Akoya Biosciences). Images were analyzed and quantified by Phenochart software (v.1.0.12, PerkinElmer Inc.). AI-assisted analyses using Phenochart software were performed to determine the expression levels of MAPT, GHR, CD5L, CCL14 and the numbers of CD8⁺ T cells.

Statistical analysis

All statistical analyses were carried out using R v 4.1.1 (www.r-project.org/), GraphPad Prism version 7.0 and SPSS version 21.0 software (IBM Corporation, Armonk, NY, USA). The Kruskal-Wallis test was used to compare differences between more than two groups, while the Wilcoxon test was used to compare two groups. The overall survival time of the two subgroups was evaluated by the Kaplan-Meier method and visualized by the survival curve. The statistically significant differences were evaluated by the log rank test. The correlation between IRGPI subtypes and clinical features was evaluated using the chi-square test, and the correlation coefficient was determined using the Spearman analysis. A p-value < 0.05 indicated statistical significance.

Establishment of IRGPI in HCC

The TCGA database was used to obtain RNA-seq transcriptomic data, and the expression levels of immune-related genes in tumor and normal tissues were analyzed. The results showed that out of 1811 immune-related genes, 555 genes were found to be differentially expressed in tumor and normal tissues (P< 0.05) (Figure 1A). These results suggested that immune-related genes play an indispensable role in the occurrence and development of HCC.

To further clarify the function of these genes, the underlined immune-related genes were selected as candidates for GO and KEGG enrichment analysis. GO analysis explains the roles of these genes using three aspects: cellular component (CC), molecular function (MF), and biological process (BP). The results of GO analysis suggest that the functions of these genes are mainly focused on the positive regulation of immunity such as cell proliferation, cytokine secretion and cell chemotaxis (Figure 1B). KEGG analysis was performed to investigate the pathways involved in the regulation of downstream genes by immune-related genes. The anti-tumor immune response is closely associated with mitogen-activated protein kinase (MAPK) and antigen processing and presentation, which are the most critical pathways (Figure 1C).

WGCNA was performed to identify the co-expressed gene modules among the 555 DEIRGs, and to investigate the association between genes and HCC. The results obtained from WGCNA show that the genes in the blue module are closely related to the occurrence of HCC (R=0.66, p-value=2e-43) (Figure 2A). Therefore, 58 genes from the blue module were selected for further study.

Univariate Cox proportional hazard regression analysis identified 13 genes that are closely associated with the prognosis of HCC (p-value < 0.05, Figure 2B). LASSO regression prevents overfitting through variable selection and regularization, which helps to improve the accuracy of the model. After minimizing overfitting by LASSO regression, 4 genes were selected as hub IRGs of the model: MAPT, CCL14, GHR and CD5L. Therefore, IRGPI was derived by multiplying hub IRGs with the univariate COX regression coefficient as follows: IRGPI score = [expression of MAPT × 0.401232] + [expression of CCL14 × -0.22304] + [expression of GHR × -0.19147] + [expression of CD5L × -0.09349].

IRGPI predicts survival for HCC patients

According to the median IRGPI score, 374 HCC patients from the TCGA database were classified into IRGPI-low and IRGPI-high subgroups. Finally, a survival-prognosis model was established, using Kaplan-Meier (K-M) method. The K-M survival analysis of the model showed that the survival period of patients in the IRGPI-low group was significantly longer than that of the IRGPI-high group, which implied a remarkable ability in differentiating good or poor clinical outcomes among the two subgroups (Figure 2C). Moreover, based on the 4 hub genes in the model, we divided patients into high and low-expression groups, followed by performing survival analysis. The results showed that patients with high expression of MAPT had a poor prognosis, while patients with high expression of the other three genes had a longer survival time. Consistent with the results of the above Univariate Cox regression analysis, the obtained results reveal that MAPT is a high-risk gene while the remaining genes are low-risk genes (Figure 2D-G).

We further used univariate and multivariate independent prognostic analyses to evaluate the predictive value of the prognostic model. All independent prognostic analysis results were consistent, which shows that the IRGPI can be independent of other clinical characteristics as an independent prognostic factor (Figure 2H, I).

IRGPI significantly related to the disease progression

The correlation analysis was performed via a chi-square test to explore the possible correlation between IRGPI and clinicopathologic factors. The results showed that patients in the IRGPI-high group have higher tumor pathological grades, which implied a higher degree of tumor malignancy (Figure 3A). Furthermore, patients in the IRGPI-high group have higher clinical stages and a higher tumor infiltration area (Figure 3B, C). This suggests that IRGPI-high patients had more aggressive tumors, faster tumorigenesis, and a poorer prognosis.

IRGPI is accompanied by a higher TMB and frequency of driver gene mutations

Herein, this study identified the 20 most often mutated genes as driver genes for HCC by studying the somatic structural variation of HCC. Moreover, the frequency of driver gene mutations in the IRGPI-high group was found to be significantly higher than that in the IRGPI-low group, with a higher TMB (Figure 3D, E). A previous study indicated that TP53 or LRP1B mutations are associated with higher TMB and worse survival in HCC [31]. The tumor microenvironment of patients with CTNNB1-mutant had significantly reduced activated immune cells and expression of immune-stimulating molecules. In contrast, the number of M2-type macrophages and active immunological depletion pathways increased significantly [32]. This shows that ICIs are less significant for the IRGPI-high group with greater CTNNB1 mutation frequencies. CTNNB1 mutation-associated aldolase A (ALDOA) phosphorylation promotes HCC cell proliferation, which can be mutually verified with the results of our survival analysis [33].

IRGPI predicts tumor infiltrating immune cells and immune function in the microenvironment

Immune infiltrating cells is one of the important factors affecting immunotherapy. It is necessary to study the abundance and function of immune infiltrating cells between the two groups [14]. CIBERSORT was used to evaluate the relative proportion of 29 types of immune cells and immune function influencing the procedure of anti-tumor immune response in HCC. It is a useful tool for analyzing immune infiltrating cells in the tumor microenvironment. The results showed that the number of tumor-infiltrating cells in the IRGPI-low group was significantly greater than that of the IRGPI-high. Among the 14 immune infiltrating cells, patients in the IRGPI-low group have a higher abundance of B cells, CD⁸⁺ T cells, mast cells, neutrophils, NK cells, and T helper cells, but the abundance of macrophages is lower. In terms of immune function, patients in the IRGPI-low group are more active than IRGPI-high group, such as cytolytic activity, inflammation-promoting, and response to interferon (IFN) (Figure 4A).

Herein, the survival of patients was analyzed based on immune cells. According to the obtained results, the effector cells in anti-tumor immune response such as B cells, CD8⁺ T cells, NK cells, T helper cells and TILs were associated with a good prognosis. Furthermore, macrophages that can secrete immune negative regulatory factors to promote tumor cell immune escape predicted poor survival. (Figure 4B-G) The underlined results are also supported by preliminary research [34-37]. Furthermore, the impact of immune function on the prognosis was analyzed which revealed that the pro-inflammatory factors that promote the activity of immune cells in the tumor microenvironment are related to a good prognosis. (Figure 4H-K)

The immunogenicity of more than 10,000 tumor samples of 33 cancers from TCGA has already been reported, calculating the correlation coefficients among 160 immune characteristics. Furthermore, cluster analysis was performed to obtain the immune expression characteristics of 5 core modules. According to these 5 immune expression characteristics, all non-hematological tumors in the TCGA database are clustered into 6 immune subtypes: wound healing, IFN-γ dominant, inflammatory, lymphocyte depleted, immunologically quiet, and TGF-β dominant (C1-C6). The survival analysis of 6 immune subtypes revealed that C3 had the best prognosis, followed by C1 and C2, and C4 and C6 had the worst prognosis [38]. Matching the IRGPI group with the TCGA immune subtypes, we found that half of the patients in the low-IRGPI group belonged to the C3 with the best prognosis. Three-quarters of patients in the high-IRGPI group are C1, C2, or C4 with relatively poor prognoses. (Figure 5A)

IRGPI predicts responses of ICIs

To verify whether the score can accurately predict the patient's response to ICIs, herein, the difference between traditional biomarkers and the TIDE score between the IRGPI-low group and the high were analyzed. The results showed that the TMB of the IRGPI-high group is greater than the IRGPI-low group. (Figure 5B) Regarding the expression of immune checkpoint molecules, the expression of programmed cell death-ligand 1 (PD-L1, CD274) was considerably higher in the IRGPI-low group than in the IRGPI-high group. However, the expression of cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) was the opposite. (Figure 5C, D)

Recent studies have revealed that tumors have two different immune escape mechanisms. Even when a substantial number of cytotoxic T lymphocytes infiltrate the microenvironment of certain cancers, their functions are vague under the influence of immunosuppressive molecules [39]. In some tumors, immune-negative regulatory cells and factors can eliminate T cells infiltrating the tumor tissues and form a state of immune exclusion [40]. Therefore, the researchers have designed a new computing architecture: the TIDE score. It is also believed that TIDE score values can replace a single biomarker predicting the efficacy of ICI [30]. Herein, TIDE scores were performed on all patients which revealed that patients in the IRGPI-high group had higher immune exclusion levels. (Figure 5E) The finding indicated that fewer immune cells were infiltrating the tumor microenvironment, which was consistent with the outcomes of our analysis of immune cells. However, patients in the IRGPI-low group are more in an immune dysfunction state. (Figure 5F) Previous studies have shown that patients with immune exclusion are resistant to ICIs, and the treatment effect is not as good as that of immune dysfunction [41]. Furthermore, the TIDE score of the IRGPI-high group was significantly higher than the IRGPI-low group, which means that the immunotherapy effect of the high-risk group is poor. (Figure 5G)

To further validate the accuracy of the model’s ability to predict the effect of immunotherapy, we analyzed three immunotherapy datasets. The IMvigor210 cohort contains 298 patients with urothelial cancer treated with PD-L1 blockers. Each patient was scored and assigned to IRGPI-high and low groups according to the median value. Patients with complete response (CR) and partial response (PR) were defined as a response, and patients with stable disease (SD) and progressive disease (PD) were defined as a non-response. Notably, the objective response rate (ORR) of the PD-L1 blocker was higher in the IRGPI-low than in the IRGPI-high group in the IMvigor210 cohort (chi-square test, P=0.008), and the IRGPI of non-responders were significantly higher than responders (Wilcoxon, p-value < 0.001) (Figure 6A-C). A similar outcome was observed in the GSE78220 and GSE67501 cohort, which contains 39 patients with melanoma and non-small cell lung cancer (NSCLC) treated with PD-1 therapy (chi-square test, P=0.043) (Figure 6D-E). Taken together, the IRGPI can effectively predict the response to ICIs. Immunostaining on the HCC cohort further confirmed that CD8⁺ T cells were more abundant in IRGPI-low group, while the tumor microenvironment of IRGPI-high group presents a scene of immune desert (Figure 7A, B). Statistical analysis of the number of CD8⁺ T cells showed that the proportion of CD8⁺ T cells in the IRGPI-low group was significantly increased than that in the IRGPI-high group (Figure 7C). Survival analysis showed that the survival time of IRGPI-low group was significantly longer than that of IRGPI-high group (Figure 7D). In addition, we followed up the patients for three years. Survival analysis showed that the three-year survival rate of IRGPI-low group was significantly higher than that of IRGPI-high group. Together, these results indicate that IRGPI-low group has a unique immune ecosystem, with increased CD8⁺ T cells. This is one reason why patients in the IRGPI-low group have a better response to ICIs and a longer survival period.

Researchers have developed a greater understanding of HCC pathogenesis and immune evasion as a result of the evolution of molecular biology techniques over the past few decades. Consequently, the rapid development of HCC therapeutic medications will potentially transform HCC into a chronic condition with a long-life expectancy. Early clinical trials of ICIs showed their significant efficacy in causing tumor shrinkage and improving the overall survival of HCC patients. However, only a small number of patients can benefit from it. Conventional biomarkers cannot accurately screen out all tumor patients suitable for ICIs. In this study, we aimed to quantify the comprehensive component of tumor microenvironment in HCC. Due to the lack of datasets of HCC patients with ICIs, we used IRGPI to evaluate response to ICIs in other tumor types. The mechanism of ICIs is to restore the anti-tumor activity of the immune system suppressed by immune checkpoint. The anti-tumor activity is non-specific, which is why ICIs have a broad-spectrum therapeutic effect. With the exception of urothelial cancer, atezolizumab has been demonstrated to be effective in HCC, NSCLC, small cell lung cancer (SCLC) and triple negative breast cancer (TNBC) [8, 42–44]. We also assume that as research continues, sequencing data on HCC patients treated with ICIs would become available as soon as feasible, allowing us to validate the IRGPI’s accuracy in HCC.

Increasing evidence has implied that the immune cell dysfunction induced by immune checkpoints within the tumor microenvironment promotes immune evasion and thus causes tumor growth and progression. In this study, we analyzed the IRGPI in 374 HCC samples and categorized three HCC samples into distinct IRGPI subtypes. Our analysis showed that the tumor microenvironment in the IRGPI-low group has a higher density of immune cell infiltration and is associated with a better prognosis, which showed consistency with the previous studies [45, 46]. The underlined study indicated that the pre-existing immune cells in the tumor microenvironment have an anti-tumor effect and positively influence the response to ICIs. Conventional biomarkers for predicting the efficacy of immunotherapy such as PD-L1 expression, TMB, and MSI usually focus on the expression of tumor cell immunosuppressive molecules and the formation of neoantigens. However, whether there are sufficient quantity and quality of immune cells in the tumor microenvironment to produce a killing effect on the tumor is more important. The method capable of distinguishing between hot tumors rich in immune cells and cold tumors lacking immune cells can more effectively screen potential candidates who can benefit from immunotherapy [45]. IRGPI can clearly show the difference in immune infiltrating cells between subgroups. The immune microenvironment in the IRGPI-low group has sufficient immune cell infiltration and is inflammation-promoting, which is conducive to the progress of the anti-tumor immune response. According to a synergistic analysis with immunosuppressive molecules, tumor cells of IRGPI-low group typically exhibit a high level of PD-L1. This implies that immune cells in the tumor microenvironment are functionally suppressed, and ICIs can effectively awaken immune effector cells to kill tumors. In the IRGPI-high group, the abundance of immune infiltrating cells in the tumor microenvironment is low and the expression of CTLA-4 is increased. CTLA-4 is usually expressed on the surface of antigen-presenting cells and transmits negative immune regulatory signals to T cells in the lymphatic system to induce T cell apoptosis and prevent T cells from migrating to tumor tissues [47]. This helps to explain why IRGPI-high patients’ tumors are resistant to immunotherapy. Instead of blindly employing ICIs in IRGPI-high patients, increasing the number of immune infiltrating cells in the tumor microenvironment should be emphasized. Radiotherapy and anti-angiogenesis therapy can increase the number of immune cells by destroying the immune barrier and reshaping the tumor microenvironment [48]. The complementary effects of the two immunological up-regulated pathways (CTLA-4 and PD-1/PD-L1) may result in a synergistic impact when different ICIs are combined. Thus, the multi-ICIs or the combination of ICIs and other therapies has become a novel strategy to treat HCC [49]. Therefore, patients in the IRGPI-high group may require pretreatment with other treatments to transform cold tumors into hot tumors before using ICIs. The IRGPI-high group has high expression of CTLA-4 and a small number of CD8 + T cells in the tumor microenvironment, which may be an indication for a combination of anti-PD-1/PD-L1 and anti-CTLA-4. Currently, several clinical trials combining PD-1 and CTLA-4 inhibitors have shown promising results, suggesting a novel strategy for overcoming the ineffectiveness of single ICIs [50, 51]. Overall, the IRGPI-low group had significantly higher TILs, PD-L1 expression, and TMB, which corresponded to better ORR of ICIs. Therefore, the combination of IRGPI and standard biomarkers may help to more accurately predict the efficacy of ICIs in HCC. We believe that IRGPI can be used synergistically with current biomarkers such as PD-L1, TMB etc. for clinical decision making of ICIs in HCC, as the cost of gene test technology has decreased and its use has become more prevalent.

There are certain limitations of this study. Firstly, because there is no dataset on immunotherapy for HCC, the model to predict response to immunotherapy can only be verified by other types of tumors. Secondly, our model is mainly based on the data of HCC in Western countries. HCC in Western countries is mostly caused by hepatitis C virus or fatty liver disease, while 85% of HCC patients in China have hepatitis B virus infection. The gene mutation spectrum of HCC in China is different from that of Western countries, which may affect the response to ICIs. Hence, clinical research based on Chinese data is required to produce better therapeutic outcomes [33]. Finally, more research is needed to determine the molecular biological mechanisms through which immune-related genes influence prognosis and immunotherapy response in HCC. Research on lung cancer has found that not only does the overall TILs of the tumor has clinical significance, but TIL also has clinical significance at the different tumor site (central or peripheral), which cannot be detected by conventional genomics, so another bioinformatic technology is needed for analysis. We hope to provide some useful clues for later research.

In this study, we established a prognostic model based on immune-related genes. This model can predict the prognosis of patients with HCC, and more importantly, can examine the response to ICIs. In addition, we also analyzed the influence of IRGPI on immune infiltrating cells in the tumor microenvironment and the association with other biomarkers. Hence, it has been suggested that the underlined biomarker could be used in combination with standard biomarkers to help clinicians better identify HCC patients who could benefit from immunotherapy.

ICIs: immune checkpoint inhibitors; HCC: hepatocellular carcinoma; IRGPI: immune-related gene-based prognostic index; TCGA: The Cancer Genome Atlas; GEO: Gene Expression Omnibus; WGCNA: Weight Gene Co-expression Network Analysis; TIDE: Immune Dysfunction and Exclusion; TMB: tumor mutation burden; MSI: microsatellite instability; TIL: tumor infiltrating lymphocytes; IRGs: immune-related genes; ImmPort: Immunology Database and Analysis Portal; DEIRGs: differentially expressed immune-related genes; GO: Gene Ontology; KEGG: Kyoto Encyclopedia of Genes and Genomes; FDR: false discovery rate; ME: module eigengene; MM: module membership; OS: overall survival; LASSO: Least Absolute Shrinkage and Selection Operator; CC: cellular component; MF: molecular function; BP: biological process; MAPK: mitogen-activated protein kinase; IFN: interferon; ALDOA: aldolase A; PD-L1: programmed cell death-ligand 1; CTLA-4: cytotoxic T-lymphocyte-associated protein 4; CR: complete response; PR: partial response; SD: stable disease; PD: progressive disease; ORR: objective response rate.

Ethical approval and consent to participate

All the patients were informed about the purposes of the study. All investigations conformed to the principles outlined in the Declaration of Helsinki and were performed with permission by the responsible Medical Ethics Committee of Shandong First Medical University, China.

Consent for publication

Not applicable.

Data availability statement

The datasets analyzed during the current study are available in The Cancer Genome Atlas, [https://portal.gdc.cancer.gov/] and Gene Expression Omnibus, [http://research-pub.gene.com/IMvigor210CoreBiologies]; [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSM1648114]; [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE78220].

Competing interests

The authors have no conflicts of interest related to this publication.

Funding

This work was financially supported by the grants from the Department of Science & Technology of Shandong Province (2021CXGC011102).

Authors’ Contributions

Study concept and design: (Jialin Qu, Fenghao Sun), acquisition of data: (Jialin Qu, Fenghao Sun, Hui Guan), drafting of the manuscript (Jialin Qu, Fenghao Sun), critical revision of the manuscript for important intellectual content (Ligang Xing, Xiaorong Sun), administrative, technical, or material support (Yichen Hou, Haoran Qi), and study supervision (Ligang Xing, Xiaorong Sun). All authors have made a significant contribution to this study and have approved the final manuscript.

Acknowledgements

Not applicable.

Siegel, R.L., K.D. Miller, and A. Jemal, Cancer statistics, 2020. CA Cancer J Clin, 2020. 70(1): p. 7-30.
Allemani, C., T. Matsuda, V. Di Carlo, et al., Global surveillance of trends in cancer survival 2000-14 (CONCORD-3): analysis of individual records for 37 513 025 patients diagnosed with one of 18 cancers from 322 population-based registries in 71 countries. Lancet, 2018. 391(10125): p. 1023-1075.
Fujiwara, N., S.L. Friedman, N. Goossens, et al., Risk factors and prevention of hepatocellular carcinoma in the era of precision medicine. J Hepatol, 2018. 68(3): p. 526-549.
Prieto, J., I. Melero, and B. Sangro, Immunological landscape and immunotherapy of hepatocellular carcinoma. Nat Rev Gastroenterol Hepatol, 2015. 12(12): p. 681-700.
Jacome, A.A., A.C.G. Castro, J.P.S. Vasconcelos, et al., Efficacy and Safety Associated With Immune Checkpoint Inhibitors in Unresectable Hepatocellular Carcinoma: A Meta-analysis. JAMA Netw Open, 2021. 4(12): p. e2136128.
Chu, P.Y. and S.H. Chan, Cure the Incurable? Recent Breakthroughs in Immune Checkpoint Blockade for Hepatocellular Carcinoma. Cancers (Basel), 2021. 13(21): p. 5295.
Finn, R.S., M. Ikeda, A.X. Zhu, et al., Phase Ib Study of Lenvatinib Plus Pembrolizumab in Patients With Unresectable Hepatocellular Carcinoma. J Clin Oncol, 2020. 38(26): p. 2960-2970.
Finn, R.S., S. Qin, M. Ikeda, et al., Atezolizumab plus Bevacizumab in Unresectable Hepatocellular Carcinoma. N Engl J Med, 2020. 382(20): p. 1894-1905.
Finn, R.S., B.Y. Ryoo, P. Merle, et al., Pembrolizumab As Second-Line Therapy in Patients With Advanced Hepatocellular Carcinoma in KEYNOTE-240: A Randomized, Double-Blind, Phase III Trial. J Clin Oncol, 2020. 38(3): p. 193-202.
Pinter, M., R.K. Jain, and D.G. Duda, The Current Landscape of Immune Checkpoint Blockade in Hepatocellular Carcinoma: A Review. JAMA Oncol, 2021. 7(1): p. 113-123.
Chan, T.A., M. Yarchoan, E. Jaffee, et al., Development of tumor mutation burden as an immunotherapy biomarker: utility for the oncology clinic. Ann Oncol, 2019. 30(1): p. 44-56.
Fridman, W.H., L. Zitvogel, C. Sautès-Fridman, et al., The immune contexture in cancer prognosis and treatment. Nat Rev Clin Oncol, 2017. 14(12): p. 717-734.
Xu, S., R. Lai, Q. Zhao, et al., Correlation Between Immune-Related Adverse Events and Prognosis in Hepatocellular Carcinoma Patients Treated With Immune Checkpoint Inhibitors. Front Immunol, 2021. 12: p. 794099.
Yi, M., D. Jiao, H. Xu, et al., Biomarkers for predicting efficacy of PD-1/PD-L1 inhibitors. Mol Cancer, 2018. 17(1): p. 129.
Im, S.J., M. Hashimoto, M.Y. Gerner, et al., Defining CD8+ T cells that provide the proliferative burst after PD-1 therapy. Nature, 2016. 537(7620): p. 417-421.
Zhang, P.F., C. Gao, X.Y. Huang, et al., Cancer cell-derived exosomal circUHRF1 induces natural killer cell exhaustion and may cause resistance to anti-PD1 therapy in hepatocellular carcinoma. Mol Cancer, 2020. 19(1): p. 110.
Wang, K., X. Chen, C. Jin, et al., A novel immune-related genes prognosis biomarker for hepatocellular carcinoma. Aging (Albany NY), 2020. 13(1): p. 675-693.
Liu, T., H. Wu, J. Qi, et al., Seven immune-related genes prognostic power and correlation with tumor-infiltrating immune cells in hepatocellular carcinoma. Cancer Med, 2020. 9(20): p. 7440-7452.
Wang, Z., J. Zhu, Y. Liu, et al., Development and validation of a novel immune-related prognostic model in hepatocellular carcinoma. J Transl Med, 2020. 18(1): p. 67.
Zhang, G., L. Su, X. Lv, et al., A novel tumor doubling time-related immune gene signature for prognosis prediction in hepatocellular carcinoma. Cancer Cell Int, 2021. 21(1): p. 522.
Bhattacharya, S., S. Andorf, L. Gomes, et al., ImmPort: disseminating data to the public for the future of immunology. Immunol Res, 2014. 58(2-3): p. 234-9.
Ritchie, M.E., B. Phipson, D. Wu, et al., limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res, 2015. 43(7): p. e47.
Ginestet, C., ggplot2: Elegant Graphics for Data Analysis. Journal of the Royal Statistical Society Series a-Statistics in Society, 2011. 174(1): p. 245-245.
Langfelder, P. and S. Horvath, WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics, 2008. 9: p. 559.
Friedman, J., T. Hastie, and R. Tibshirani, Regularization Paths for Generalized Linear Models via Coordinate Descent. J Stat Softw, 2010. 33(1): p. 1-22.
Therneau, T., A package for survival analysis. R package 2.37-2. 2012.
Mayakonda, A., D.C. Lin, Y. Assenov, et al., Maftools: efficient and comprehensive analysis of somatic variants in cancer. Genome Res, 2018. 28(11): p. 1747-1756.
Newman, A.M., C.L. Liu, M.R. Green, et al., Robust enumeration of cell subsets from tissue expression profiles. Nat Methods, 2015. 12(5): p. 453-7.
Yoshihara, K., M. Shahmoradgoli, E. Martínez, et al., Inferring tumour purity and stromal and immune cell admixture from expression data. Nat Commun, 2013. 4: p. 2612.
Jiang, P., S. Gu, D. Pan, et al., Signatures of T cell dysfunction and exclusion predict cancer immunotherapy response. Nat Med, 2018. 24(10): p. 1550-1558.
Wang, L., K. Yan, X. He, et al., LRP1B or TP53 mutations are associated with higher tumor mutational burden and worse survival in hepatocellular carcinoma. J Cancer, 2021. 12(1): p. 217-223.
Chen, L., Q. Zhou, J. Liu, et al., CTNNB1 Alternation Is a Potential Biomarker for Immunotherapy Prognosis in Patients With Hepatocellular Carcinoma. Front Immunol, 2021. 12: p. 759565.
Gao, Q., H. Zhu, L. Dong, et al., Integrated Proteogenomic Characterization of HBV-Related Hepatocellular Carcinoma. Cell, 2019. 179(2): p. 561-577 e22.
DeNardo, D.G. and B. Ruffell, Macrophages as regulators of tumour immunity and immunotherapy. Nature Reviews Immunology, 2019. 19(6): p. 369-382.
Paul, M.S. and P.S.J.T.i.C.B. Ohashi, The Roles of CD8+ T Cell Subsets in Antitumor Immunity. 2020. 30(9).
Wang, S.S., W. Liu, D. Ly, et al., Tumor-infiltrating B cells: their role and application in anti-tumor immunity in lung cancer. 2018.
Zou, W. and N.P. Restifo, T(H)17 cells in tumour immunity and immunotherapy. Nat Rev Immunol, 2010. 10(4): p. 248-56.
Thorsson, V., D.L. Gibbs, S.D. Brown, et al., The Immune Landscape of Cancer. Immunity, 2019. 51(2): p. 411-412.
Wherry, E.J. and M. Kurachi, Molecular and cellular insights into T cell exhaustion. Nat Rev Immunol, 2015. 15(8): p. 486-99.
Gajewski, T.F., H. Schreiber, and Y.X. Fu, Innate and adaptive immune cells in the tumor microenvironment. Nat Immunol, 2013. 14(10): p. 1014-22.
Spranger, S., R. Bao, and T.F. Gajewski, Melanoma-intrinsic β-catenin signalling prevents anti-tumour immunity. Nature, 2015. 523(7559): p. 231-5.
Herbst, R.S., G. Giaccone, F. de Marinis, et al., Atezolizumab for First-Line Treatment of PD-L1-Selected Patients with NSCLC. N Engl J Med, 2020. 383(14): p. 1328-1339.
Liu, S.V., M. Reck, A.S. Mansfield, et al., Updated Overall Survival and PD-L1 Subgroup Analysis of Patients With Extensive-Stage Small-Cell Lung Cancer Treated With Atezolizumab, Carboplatin, and Etoposide (IMpower133). J Clin Oncol, 2021. 39(6): p. 619-630.
Schmid, P., S. Adams, H.S. Rugo, et al., Atezolizumab and Nab-Paclitaxel in Advanced Triple-Negative Breast Cancer. N Engl J Med, 2018. 379(22): p. 2108-2121.
He, Y., Z. Jiang, C. Chen, et al., Classification of triple-negative breast cancers based on Immunogenomic profiling. J Exp Clin Cancer Res, 2018. 37(1): p. 327.
Rooney, M.S., S.A. Shukla, C.J. Wu, et al., Molecular and genetic properties of tumors associated with local immune cytolytic activity. Cell, 2015. 160(1-2): p. 48-61.
Snyder, A., V. Makarov, T. Merghoub, et al., Genetic basis for clinical response to CTLA-4 blockade in melanoma. N Engl J Med, 2014. 371(23): p. 2189-2199.
Bernstein, M.B., S. Krishnan, J.W. Hodge, et al., Immunotherapy and stereotactic ablative radiotherapy (ISABR): a curative approach? Nat Rev Clin Oncol, 2016. 13(8): p. 516-24.
Xing, R., J. Gao, Q. Cui, et al., Strategies to Improve the Antitumor Effect of Immunotherapy for Hepatocellular Carcinoma. Front Immunol, 2021. 12: p. 783236.
Schachter, J., A. Ribas, G.V. Long, et al., Pembrolizumab versus ipilimumab for advanced melanoma: final overall survival results of a multicentre, randomised, open-label phase 3 study (KEYNOTE-006). Lancet, 2017. 390(10105): p. 1853-1862.
Hellmann, M.D., L. Paz-Ares, R. Bernabe Caro, et al., Nivolumab plus Ipilimumab in Advanced Non-Small-Cell Lung Cancer. N Engl J Med, 2019. 381(21): p. 2020-2031.

Table 1 Clinical features between IRGPI-high and IRGPI-low group in Shandong Cancer Hospital and Institute cohort

Clinical features	IRGPI-high	IRGPI-low
Total cases	20 (50%)	20 (50%)
Age
< 65	8 (40%)	10 (50%)
≥ 65	12 (60%)	10 (50%)
Gender
Male	15 (75%)	17 (85%)
Female	5 (25%)	3 (15%)
Stage
Stage I	13 (65%)	12 (60%)
Stage II	6 (30%)	8 (40%)
Stage III	1 (5%)	0
Stage IV	0	0
Tumor
T1	13 (65%)	12 (60%)
T2	7 (35%)	8 (40%)
T3	0	0
T4	0	0
Node
N0	19 (95%)	20 (100%)
N1	1 (5%)	0
N3	0	0
N4	0	0
Metastasis
M0	20 (100%)	20 (100%)
M1	0	0

Download PDF

Version 1

posted

You are reading this latest preprint version

The immune-related gene-based prognostic index predicts the prognosis and effect of immune checkpoint inhibitors in hepatocellular carcinoma

Status:

Version 1

Abstract

Figures

Background

Methods

Results

Discussion

Conclusion

Abbreviations

Declarations

References

Tables

Status:

Version 1