Comprehensive characterization of Fidgetin on tumor immune microenvironment evaluation and immunotherapy in human hepatocellular carcinoma

Tissue samples

From Affiliated Nantong Hospital 3 of Nantong University, 30 pairs of HCC samples and their corresponding normal liver tissues were collected.

Data mining in the TCGA and GEO database

By logging in to the official websites https://portal.gdc.cancer.gov/ and https://xenabrowser.net/, respectively, datasets on the transcriptional expression of FIGN and the corresponding clinical data were obtained from the TCGA (369 primary liver carcinoma and 50 adjacent normal tissue) and UCSC database (n=374) [26]. The total number of cancer types included in this study was 21. For the sake of the subsequent study, the FPKM workflow type data from the RNA-seq method were transformed into TPM format and log2 transformation. The difference in FIGN expression between normal tissues and HCC was further examined using data from GEO (www.ncbi.nlm.nih.gov/geo) (GSE121248, n=107; GSE25097, n=517). GSE121248 included differential gene expression analysis between chronic hepatitis B induced HCC and adjacent Normal tissues using Affymetrix gene arrays. GSE25097 included transcriptome profiling of frozen tissues (tumor and non-tumor) in early-to-advanced HCC patients. Since the data were obtained from TCGA and GEO, there is no need for an ethics check.

Functional enrichment analysis of FIGN gene co-expression network

The LinkedOmics database (http://www.linkedomics.org) examined the co-expression interaction network of FIGN in the TCGA-HCC dataset. In order to show the differences, rank gene maps and heatmaps were created using the Spearman correlation test. The rank coefficient was set as FDR<0.05. Additionally, the “ClusterProfiler” package of R was adopted to summary the Gene Ontology (GO) function and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway that contains the co-expressed genes. Pictures were visualized by “ggplot2” package.

Tumor -immune system interaction database (TISIDB), tumor immune estimation resource (TIMER) database

The expression of FIGN and tumor-infiltrating lymphocytes (TILs) in various human cancers was assessed using the TISIDB database (http://cis.hku.hk/TISIDB/index.php). In order to assess the relationships between FIGN and TILs, Spearman’s test was performed. Additionally, the TIMER, TCGA, and GEPIA2 databases were used to examine the connection between immune infiltrating cells (B cells, CD8⁺ T cells, CD4⁺ T cells, macrophages, neutrophils, and dendritic cells) and FIGN expression in hepatocellular carcinoma. To examine the relationship between FIGN expression and overall survival (OS), disease-free interval (DFI), progression-free interval (PFI), and disease-specific survival (DSS) in HCC and the pan-cancer, data from UCSC-Xena were downloaded and processed using the R software’s “survival” and “forestplot” packages.

Comprehensive analysis of TME in TCGA-HCC subgroups associated the expression of FIGN

To determine the stromal score for each sample, we employed the ESTIMATE method. The immune enrichment score was then calculated based on the 29 immune gene sets using the single sample gene set enrichment analysis (ssGSEA) function of the R package “GSVA” on the gene sets of “c2.cp.kegg.v7.4.symbols.gmt” from the MSigDB dataset. We uploaded the gene expression profiles of the HCC samples to the CIBERSORTx website (https://cibersortx.stanford.edu) using 1000 permutations in order to discover the immune infiltration characteristics of the samples.

Tumor tissues clustering based on the hot tumor related gene markers

The tumors were divided into hot and cold tumors based on the gene markers identified in the previous publication, which included CCL5, CD8A, PDCD1, CD8B, CXCR3, CXCR4, CXCL9, CXCL10, CD4, CD3E, CD274, and CXCL11. To perform unsupervised clustering and offer quantifiable data to support the number of probable clusters within the RNA-seq dataset, the “ConsensusClusterPlus” R program was selected. Here, RNA-seq data from 374 HCC patients in the TCGA database were included in the study. The crucial detection settings included the previously mentioned 1000 repetitions, maximum assessed k of 20, and 80% item resampling.

Immunotherapy analysis and chemotherapeutic response prediction

Immunophenoscore (IPS) was used to examine the data obtained from the TCIA website (https://tcia.at/home) in order to monitor the immunotherapy effects in the low- and high-score patients and assess the benefits and drawbacks of anti-CTLA-4 and anti-PD-L1 therapies. Based on the tumor’s immunogenicity and the effectiveness of the immunotherapy, the IPS scores ranged from 0 to 10. To compare the statistical differences and create the images, the “ggpubr” R tool was utilized.

The Genomics of Drug Sensitivity in Cancer (GDSC, https://www.cancerrxgene.org) was used to examine the effects of medications between low- and high-FIGN expression HCC patients in order to predict the reaction of commonly used drugs and possible small molecules. The half-maximal inhibitory concentration (IC50) was assessed by ridge regression, and the prediction accuracy was examined by 10-fold cross-validation based on the GDSC dataset. These effects were predicted using the default parameters of the “pRRophetic” R package. In addition, factors such tissue type of “allSolidtumours” and batch effect of “combat” were deleted, and duplicate gene expression was chosen as the mean value.

Statistical analysis

Student’s t-test, Mann-Whitney U-test, and logistic regression analysis were employed to analyze differences in FIGN expression between the various clinical pathological features. The cutoff value was established using R software’s “pROC” tool. To examine the impact of FIGN on survival, Kaplan-Meier and log-rank tests were run using the “survminer” R package and the Kaplan-Meier website. Using TISIDB and TIMER, it was possible to compare the differences in FIGN expression and TILs between several groups. P<0.05 was regarded as statistically significant, and all data were shown as means with standard deviations.

Availability of data and materials

Publicly available datasets were analyzed in this study. This data can be found at TCGA and GEO datasets (accession number: GSE121248, GSE25097) and in the Supplementary Material.

Expression pattern of FIGN based on pan-cancer analysis and in HCC patients

Total of 21 cancer kinds were filtered out for further study in order to evaluate the transcriptional expression variations of FIGN among various cancer types. The results revealed that FIGN expression was decreased in 16 of the 21 cancer types as compared to the normal tissues. Additionally, compared to normal tissues, FIGN expression was greater in GBM, KIRC, KIRP, and UCEC (Figure 1A).

The transcriptional expression of FIGN was considerably downregulated in the primary tumor compared to paracarcinoma tissues, according to the results of the non-paired TCGA data analysis (Figure 1B, P=0.007). FIGN expression was not substantially different between patients of different ages (Supplementary Figure 1A), tumor stages (Supplementary Figure 1B), T stages (Supplementary Figure 1C), M stages (Supplementary Figure 1D), or N stages (Supplementary Figure 1E), but FIGN expression was significantly different between male and female patients (Supplementary Figure 1F, P<0.05). In Supplementary Table 1, the clinicopathological traits were described in depth. In addition, data sets from GSE121248 and GSE25097 revealed that FIGN mRNA expression was lower in tumors compared to normal tissues (Figure 1C, P=8.9E-08 and Figure 1D, P=1.7E-06).

Notably, FIGN’s mRNA expression was considerably lower in clinical HCC samples than it was in paracarcinoma tissues (Figure 1E, P<0.01) When compared to the corresponding normal tissues, FIGN protein expression was reduced in the main tumor (Figure 1F, 1G, P<0.01). FIGN expression in the tumors was shown to be prone by immunohistochemistry staining on HCC liver sections (Figure 1H–1J, P<0.001). Additionally, histological analysis provided in HPA demonstrated that FIGN protein was downregulated in malignancies (Supplementary Figure 1G). The information gathered showed that FIGN expression in hepatocellular carcinoma tissues was lowered on both the mRNA and protein levels.

The prognostic value of FIGN in hepatocellular carcinoma

The TCGA dataset underwent Cox regression analysis in order to further examine the prognostic value of FIGN. In both the univariate Cox regression analysis (Figure 2A) and the multivariate Cox regression analysis, the findings demonstrated that FIGN was strongly linked with the OS (Figure 2B). High FIGN expression generally predicted a poor prognosis compared to low-FIGN patients, especially in OS (P=0.006, Figure 2C), PFI (P=0.002, Figure 2D), DFI (P=0.049, Figure 2E), and DSS (P=0.004, Figure 2F). The clinical reliability of FIGN was investigated using a ROC curve, and the AUC value of 0.633 demonstrated significant sensitivity and specificity in HCC detection (Figure 2G). Additionally, the AUC values for the 1-year, 3-year, and 5-year OS were, respectively, 0.633, 0.586, and 0.545. (Figure 2H). According to the box plot (P=0.035, Figure 2I), individuals with high FIGN have considerably greater fatality rates than those with low FIGN. These findings suggested that FIGN might be a viable biomarker for HCC therapy.

Enrichment analysis of FIGN gene co-expression network in HCC

To examine the co-expressed protein-coding genes connected to the expression of FIGN in the TCGA database’s HCC data sets. In total, 6635 genes with negative correlations and 13286 genes with positive correlations were presented in Supplementary Figure 2A. Heat maps were used to display the top 50 genes that favorably (Supplementary Figure 2B) and negatively (Supplementary Figure 2C) associated with FIGN expression. The GO function and KEGG pathway enrichment analyses included the top 200 genes that were positively co-expressed with FIGN. Top 10 bubble plots for BP, CC, and MF were displayed (Supplementary Figure 2D–2F). According to KEGG pathway analysis, FIGN co-expression genes were primarily enlisted in immune-related pathways (Supplementary Figure 2G). The findings in this study suggested that FIGN may contribute to an immune-related response in HCC patients.

The KEGG pathway variations between the high- and low-FIGN patients were then evaluated by GSVA and depicted in a heat map (Figure 3A). Drug metabolism pathways tended to be enriched in the low-FIGN group, but immune-related pathways, such as the JAK-STAT signaling pathway, Chemokine signaling pathway, and B cell receptor signaling pathway, etc., were significantly enriched in the high-FIGN group. GSEA was used to enrich the gene sets in order to examine the potential role of the FIGN gene. A total of 2718 gene sets were identified, including the TNF signaling pathway, NF-kappa B signaling pathway, Toll-like receptor signaling pathway, JAK-STAT signaling pathway, and Nature killer cell mediated cytotoxicity (FDR=0.047, P=8.9e-03), as well as drug metabolism-cytochrome P450 (FDR=9.4e-05, P=3.9e-06) (Figure 3B). According to the data presented above, FIGNinteracting genes were primarily involved in the control of immunity and drug metabolism.

Correlation between FIGN and tumor immune infiltration cells

The association between FIGN expression and immune penetrating cells in HCC was discovered using the TIMER database. Results showed that FIGN expression was positively correlated with B cell (R=0.223, P=3.10e-05), CD8⁺ T cell (R=0.203, P=1.52e-04), CD4⁺ T cell (R=0.324, P=7.93e-10), macrophage (R=0.381, P=3.28e-13), neutrophil (R=0.427, P=1.02-16), and dendritic cell (R=0.311, P=4.89e-09), while negatively correlated with tumor purity (R=-0.062, P=2.48e-01) (Figure 4A). Additionally, B cell, CD4⁺ T cell, and dendritic cell infiltration levels were impacted by FIGN CNV (Copy Number Variation) (Figure 4B).

To comprehensively analyze the landscape of the immune microenvironment, CIBERSORTx was taken to examine the infiltration situation of 22 immune cells. Figure 4C shows the immunological characterization based on TCGA-LIHC samples. Additionally, Spearman’s analysis revealed a link between FIGN expression profiles and the infiltration levels of 22 immune cell types (Figure 4D). Neutrophils and activated CD4 memory T cells were the top two immune cells that positively linked with FIGN, while activated dendritic cell and NK cells were the top two immune cells that negatively correlated with FIGN. In conclusion, FIGN showed some relationship with immune invading cells.

The stromal score and immune activity of all samples have been shown by heatmap in order to compare the profiling of stromal cells and immune cells in the TME, which had a significant impact on tumor development, therapeutic effects, and clinical outcomes (Figure 5A). FIGN expression demonstrated a connection with tumor immune cell infiltration (Figure 5B). Additionally, FIGN expression illustrated positive correlations with ESTIMATEScore, ImmuneScore, and StromalScore while presenting negative correlations with TumorPurity, indicating its function in antitumor immunity (Figure 5C). Data sets from the GEPIA2, TIMER, and TCGA-LIHC databases were taken to assess the link between FIGN and distinct immune infiltrating cells in HCC by examining the association between FIGN and immunological markers of the immune cells (Supplementary Table 2). Results from the three databases showed a correlation between FIGN expression and the genes for B cells, natural killer cells, Th1 cells, and Th2 cells (Figure 5D). Together, FIGN may play a significant role in the control of tumor TME and immunity.

The enrichment scores of various immune cell subpopulations, associated functions, and linked pathways were measured using ssGSEA in order to further investigate the relationship between FIGN and immunological state. We discovered that the high-FIGN group had considerably higher levels of the components of the antigen presentation pathway in the TCGA cohort, including aDCs, iDCs, pDCs, APC co-inhibition, HLA, and MHC class I. (Supplementary Figure 3A). Additionally, the high-FIGN group had greater proportions of Tfh cells, Treg cells, Th1 cells, Th2 cells, T cell co-stimulation, and T cell co-inhibition. Furthermore, the high-FIGN patients had higher scores for CCR, check-point, macrophages, neutrophils, and inflammation-promoting activity, while type II IFN response activity was the exact reverse (Supplementary Figure 3B). In the TCGA, the low-FIGN group had a greater enrichment of immune cells, particularly anti-tumor immune cells (using the TIMER, CIBERSORT, CIBERSORT-ABS, QUANTISEQ, MCPCOUNTER, XCELL, and EPIC algorithms) (Supplementary Figure 3C). According to several platforms, the immune bubble diagram demonstrated the affiliation of various immune cells with FIGN expression (Supplementary Figure 3D). B cells and cancer-associated fibroblast (CAFs) were associated with the high-FIGN expression patients. Totally, FIGN were correlated with immune infiltrating cells and thus affected the TME.

FIGN was correlated with immune checkpoints and enhanced immunotherapy effects

The effects of immunotherapy mostly depended on the activation or inhibition of immune checkpoint gene activity. 24 immune checkpoint genes were identified in heatmap analysis to compare the expression patterns of immune checkpoints in normal and malignant tissues in HCC patients (Figure 6A). FIGN expression was positively connected with the expression of CD40LG (R=0.322, P=2.8e-03) and CD160 (R=0.282, P=9.5e-03), but its expression was negatively correlated with the expression of CD40 (R=-0.499, P=1.3e-06), CD86 (R=-0.354, P=0.001), TNFSF18 (R=-0.422, P=6.3e-05), and TNFRSF9 (R=-0.375, P=4.3e-04) (Figure 6B).

Between high- and low-FIGN groups, the expression of immune-checkpoint genes was compared (Figure 6C). High expression of CD28 (P=6.83e-06), CD27 (P=0.000), CD244 (P=0.002), PDCD1LG2 (P=0.003), CD86 (P=0.005), and CD48 (P=0.007) was shown in a Kaplan-Meier plot to be substantially linked with a better prognosis for HCC, while high expression of TNFSF4 (P=2.93e-07) and NRP1 (P=0.007) illustrated opposite survival outcome (Figure 6D). These findings suggested that FIGN may be linked to the HCC immunotherapy checkpoints.

In order to classify patients with qualitatively different hot tumors and cold tumors based on the expression of 12 hot tumor signature genes, the ConsensusClusterPlus R package was used [11, 12]. The results showed that K=3 appeared to be the best option for dividing the entire cohort into subtypes A (n=137, cold), B (n=90, hot), and C (n=197, non-response) (Supplementary Figure 4A–4D). FIGN expression was significantly higher in hot tumors than in cool tumors (P=3.5e-07, Supplementary Figure 4E), suggesting complex aspects should be taken into account when evaluating the effectiveness of immunotherapy, particularly in clinical trials. Thus, the findings above suggested that FIGN may have a role in the response to immunotherapy. Based on the TCGA RNA-seq data cohort, the association of FIGN with the predictors of response to immunotherapy was established in order to further investigate the therapeutic effects of FIGN. The findings showed that FIGN was significantly linked to the chosen parameters (Supplementary Figure 4F). Additionally, the expression of immunotherapy checkpoints in hot, cold, and non-responsive cancers varied significantly (Supplementary Figure 4G). The information gathered here positively showed that FIGN may have boosted therapeutic effectiveness in the treatment of HCC.

Expression of FIGN predicted sensitivity of HCC patients to clinical immunotherapy and chemotherapy

In the present study, TIDE score was utilized to further investigate the relevance of FIGN in predicting the immunotherapy response of HCC patients. As a consequence, patients with high FIGN had higher CD8 scores (Figure 7A) and CD274 scores (Figure 7B) than patients with low FIGN. Additionally, individuals with high levels of FIGN expression had higher Exclusion scores (Figure 7C), MDSC scores (Figure 7D), and TIDE scores (Figure 7E) than patients with low levels of FIGN expression. Additionally, TIDE and FIGN were found to be negatively correlated (Figure 7F). According to the data above, immunotherapy had a negative effect on patients who expressed high levels of FIGN.

Consistently, we investigated the CTLA-/PD-1 inhibitor response to ICI therapy further. The patients in the low-FIGN group had a substantial therapeutic response in HCC patients (Figure 7G). Next, we investigated the relationship between FIGN expression levels and drug sensitivity in HCC patients. Interestingly, we found that the IC50 values of Docetaxel, Etoposide, Doxorubicin, Gemcitabine, Pazopanib, Midostaurin, CEP.701 and Lenalidomide were higher in patients with high FIGN expression (Figure 7H). Furthermore, the IC50 values of PF.562271, XMD8.85, GSK.650394, Bleomycin, BMS.754807, IPA.3, AUY922 and BIRB.0796 were higher in patients with high FIGN expression (Figure 7I). Taken together, results here suggested that the expression of FIGN was associated with drug susceptibility.