Immunological correlation and the expression pattern of 16 TDGs in Pan-cancer.
Pan-cancer analyses aim to determine the types of cancers that may benefit from anti-genes immunotherapy and explore the potential role of those genes in TME. We evaluated the correlation of 16 TDGs with CD4 + and CD8 + T cells in pan-cancer by TIMER2 database. As shown in Fig. 1A, among 32 cancers, the top three ranked genes with the highest comprehensive correlation with CD4 + T cell were ITM2A, IL12B, and BATF respectively. The top three ranked genes with the highest correlation with CD8 + T cell were ITM2A, GPN3, and AHANK. The relationship between these 5 genes and tumor purity, estimate score, immune score and stromal score in TME was further evaluated. The results showed that ITM2A, IL12B, and BATF were positively correlated with estimate score (Figure S1A), immune score (Fig. 1B), and stromal score (Figure S1B) in most cancer species while negatively correlated with tumor purity in the TCGA database (Figure S1C). The above results confirmed the application value of these TDGs in immunotherapy for solid tumors.
ITM2A was selected as a hub gene and may be a potential therapeutic target
Furthermore, we evaluated the expression of those five genes between cancer and adjacent normal tissues in the TCGA database (Figure S2A). ITM2A was lowly expressed in multiple cancer tissues (Figure S2B) relatively while IL12B, GPN3, and BATF were highly expressed in most cancers compared with adjacent normal tissues (Figure S2C-2E). The AHNAK expression was inconsistent in all cancers (Figure S2F), suggesting its tumorigenic heterogeneity. Furthermore, we detected the expression of these genes in various immune cells in the TISCH database and found that ITM2A was mainly and prominently expressed in T cells (Fig. 1C). As a tumor suppressor gene, ITM2A was lowly expressed in most cancers and may be a potential therapeutic target.
ITM2A was a prognostic biomarker for selected cancers
The prognostic value of ITM2A expression for OS and DFS in human cancers was analyzed by the TCGA databases and the GEPIA database (Figure S3). In the TCGA database, lower ITM2A expression was associated with poorer OS in LAML (P = 0.005, Figures S3A), SKCM (P < 0.001, Figures S3B), LIHC (P = 0.003, Figures S3C), KIRC (P = 0.007, Figures S3D) and PAAD (P = 0.043, Figures S3E). Inversely, patients with lower ITM2A expression had better OS in KIRP (P = 0.015; Figure S3F). Correspondingly, patients with low ITM2A expression tend to have a worse DFS in KIRC (P = 0.003, Figures S3G), SKCM (P = 0.021, Figures S3H), LIHC (P = 0.001, Figures S3I). However, ITM2A expression levels did not significantly differ in patients with KIRP (P = 0.076, Figures S3J), PAAD (P = 0.24, Figures S3K), and LAML (P = 1, Figures S3L). These results suggest that ITM2A expression was closely related to the prognosis of various tumor types, especially SKCM, LIHC, and KIRC. Given the clinical benefits of immunotherapy for SKCM and the fact that ITM2A has been reported as a potential prognostic biomarker in LIHC (Zhang et al. 2019), we investigated the immunoregulatory role of ITM2A in KIRC.
Abnormal expression of ITM2A may be an independent prognostic risk factor for KIRC
Pan-cancer analysis suggested that the abnormal expression of ITM2A was associated with the prognosis of multiple tumors (Fig. 2A). Since the equivocal abnormal expression of ITM2A between the tumor and adjacent normal in the TCGA database, we further verified it in GSE10927 and GSE33371. Compared with normal tissue (Fig. 2B) and adenoma tissue (Fig. 2C), ITM2A expression was the lowest in adenocarcinoma tissue. The expression level of ITM2A in TNM stage III/IV was significantly lower than that in early-stage (I/II) patients (Fig. 2D). Further clinical correlation results showed that ITM2A expression was significantly associated with the T stage (Fig. 2E) and M stage (Fig. 2F) while not with the N stage (Fig. 2G). In addition, patients with low differentiation (Grade 3/4) have less ITM2A expression than those with high differentiation (Grade 1/2) (Fig. 2H). In univariate COX regression analysis, TNM stage (HR = 1.922, P < 0.001), Grade (HR = 2.320, P < 0.001), and ITM2A expression (HR = 0.741, P < 0.001) were associated with the OS. In multivariate analysis, the ITM2A expression (HR = 0.827, P = 0.011) could be an independent prognostic factor for KIRC patients (Fig. 2I).
Immune score, immune cell infiltrated correlation of ITM2A in KIRC
We explored the associations between ITM2A expression and KIRC immune score, stromal score, estimate score, RNAss, and DNAss in the TCGA database. A strong positive correlation was found between ITM2A expression and stromal score (Fig. 3A), immune score (Fig. 3B), and estimate score (Fig. 3C) in TME of KIRC. While ITM2A was negatively correlated with RNAss and DNAss (Figs. 3D and 3E). We also estimated the infiltration levels of TIICs (tumor-infiltrating immune cells) in the TME by “ssGSEA” valgorithm and proved that CD4 + and CD8 + T cells with high expression of ITM2A owned higher ssGSEA scores (Fig. 3F) and Immune activity scores (Fig. 3G). The correlation between ITM2A expression and CD4+ / CD8 + T cell, B cell immune infiltration was further confirmed by TIMER (Fig. 3H), XCELL (Fig. 3I), MCPCOUNTER (Fig. 3J), and QUANTISEQ (Fig. 3K) analysis in TIMMER2 database. Next, the role of ITM2A expression on immune subtypes was investigated and proved that ITM2A expression was closely related to immune subtypes in KIRC (Fig. 3L). In addition, the infiltration level of CD4+/CD8 + T cells was altered by changes of ITM2A CNV (Fig. 3M).
Immunomodulators’ correlation of ITM2A
To further clarify the role of ITM2A in immunotherapy, we explored the associations between ITM2A expression and several well-known Immune Checkpoint (ICP) genes. In the TCGA-KIRC dataset, the expression of ITM2A was positively correlated with PD-L1, PD-1, BTLA, CD200, CD200R1, CTLA4, IDO1, and LAG3 expression (Fig. 4A). There was strong consistency in GSE73731 and GSE53757 suggesting that patients with high expression of ITM2A could benefit from immunotherapy. Given that chemokines and chemokine receptors play a role in regulating the movement of immune cells in TME and recruiting T cells into tumors, we demonstrated that ITM2A was coordinately expressed with GZMA, CXCL10, CD8A, CCL5, IFNG in TCGA-KIEC (Fig. 4B) and validated in GSE73731 and GSE53757 dataset. Together, these results revealed that ITM2A could be a potential biomarker and therapeutic target for T cell changes.
Identification of enrichment pathways for KIRC patients with ITM2A alternation
To explore the underlying mechanism of ITM2A in the regulation of immune response in the TME of KIRC, we performed a GSEA enrichment analysis based on high-ITM2A expression and low-ITM2A expression group by TCGA-KIRC data (Fig. 4C). The results showed that high expression of ITM2A was enriched in immune-related pathways including interferon α response (Figure S4A), interferon γ response (Figure S4B), IL6 JAK STAT3 signaling (Figure S4C), IL2 STAT5 signaling (Figure S4D), TNFα signaling via NFKB (Figure S4E), apoptosis (Figure S4F), inflammatory response (Figure S4G).
Clinical prediction value of ITM2A expression for ICB
To further clarify the role of ITM2A in cancer treatment, we examined the relationship between ITM2A and Pharmacologic in the NCI-60 cell line set in the CellMiner (https://discover.nci.nih.gov/cellminer/) web-tool (Reinhold et al. 2012). Thirty-two antineoplastic drugs were significantly correlated with ITM2A (correlation > 0.3 or < -0.3, p < 0.05) (Fig. 4D and Figure S5). The drugs with the highest correlation included Nelarabine, Chelerythrine, Asparaginase, Hydroxyurea, and Fenretinide. Patients with high expression of ITM2A had higher pharmacologic IC50 suggesting that they were more likely to benefit from these drugs (Fig. 4D). The identification of predictors is crucial for immunotherapy strategies. we then investigate whether the ITM2A expression could predict immunotherapeutic benefits from the IMvigor210 dataset. The IMvigor210 dataset contains complete data on patients with urothelial carcinoma who received PD1 anti-PD-L1 therapy. The results showed that patients with high ITM2A subtype had inflamed immune phenotype while low ITM2A subtype owned desert or excluded immune phenotype (Fig. 4E). The IC (immune cells with the PD-L1 values) analysis further illustrated that ITM2A was significantly positively correlated with IC levels (Fig. 4F) revealing the potential mechanism of ITM2A's role in tumor progression and ICB therapy.
The expression of ITM2A in the KIRC tested by immunohistochemistry
We further examined the expression of ITM2A in the KIRC tissue microarray. Figure 5A illustrates the expression of KIRC and paracancerous tissues in three cases of different TNM and pathological stages. The expression of ITM2A in KIRC tissues was significantly lower than that of para-tumor tissues (Fig. 5B). In addition, ITM2A was lowly expressed in TNM stage III/IV (Fig. 5C) and Fuhrman III/ IV (Fig. 5D) tumor tissues. These results revealed that ITM2A may be involved in the occurrence of KIRC and is related to the process of malignant progression.