The current study was the first to investigate the ceRNA network that associated with TME of SARCs based on TCGA database. We first used the ESTIMATE algorithm and found that SARCs patients with high immune scores had longer overall survival time. This was similar to a previous study which reported the high immune scores has favorable outcomes in bone SARCs patients [13]. While for the correlation between stromal scores and survival in SARCs samples, we failed to find any significant difference.
Subsequently, this study filtered 328 DEGs, 18 DEMs and 67 DELs that commonly existed in both two high vs. low scores groups. Then the GO and pathway analysis showed that many of the DEGs mainly participate in immune processes. Next, we identified hsa-miR-9-5p, hsa-miR-490-3p, hsa-miR-133a-3p, hsa-miR-133b and hsa-miR-129-5p as the top 5 nodes in the ceRNA network, and MMP9, TYROBP, CSF1, CXCR4, FBN1, FLNA, PDGFRB, CYBB, FCGR3A and MYH11 were the top 10 genes in the PPI network, these results indicated that they might play more significant functions in the networks. Finally, we performed overall survival analysis of the 89 DEGs, 14 DEMs and 38 DELs in the ceRNA network and found that 9 mRNAs (APOL1, EFEMP1, LYZ, MEDAG, MYH11, RARRES1, TNFAIP2, TNFSF10 and ZNF385A), 2 miRNAs (hsa-miR-9-5p and hsa-miR-183-5p) and 3 lncRNAs (CTD-2228K2.7, HOTAIRM1 and NCF1C) were closely associated with the overall survival of SARCs patients.
Expect for TNFSF10 (TRAIL), which has been reported in several subtypes of SARCs, the rest prognostic biomarkers are novel for SARCs. APOL1 is a novel BH3-only protein, and its overexpression could induce autophagy and autophagy-associated cell death in several kinds of cancer cells [16, 17]. It was overexpressed in pancreatic cancer, lung adenocarcinoma and papillary thyroid carcinomas compared to matched normal tissues, and their prognostic roles were found in pancreatic cancer and lung adenocarcinoma [18–20]. EFEMP1 can be found in different human tissues, is a member of the fibulin family of extracellular glycoproteins [21]. Its high expression helps enhance the tumor growth in pancreatic carcinoma cells via binding the EGF receptor and activate the MAPK and Akt pathways [22]. Also, EFEMP1 could promote the migration and invasion of bone SARCs via MMP-2 with induction by AEG-1 via NF-κB signaling pathway, and EFEMP1 was also reported to be a poor prognostic indicator of bone SARCs [23]. LYZ encodes human lysozyme and acts as a macrophage marker, its expression levels positively correlate with the numbers of CD68 + pSTAT1 + macrophages [24]. In addition, it could interact with CD34 + cells and neutrophils that may predict an increased risk of thrombosis in essential thrombocythemia patients [25]. MEDAG was involved in the processes of lipid accumulation, adipocyte differentiation, and glucose uptake in mature adipocytes [26]. Overexpression of MEDAG was related to lymph node metastasis and poor prognosis in papillary thyroid microcarcinoma, which was also the first to report its roles in cancer [27]. MYH11, RARRES1 and TNFAIP2 have been commonly investigated in many cancers. MYH11 is a protein that participates in muscle contraction through the hydrolysis of adenosine triphosphate [28]. Its expression have been found to be associated with many kinds of cancer, such as lung cancer, bladder cancer [29], head and neck cancer [30] and colorectal cancer [31]. RARRES1 is also gene that dysregulated in several cancers. It could induce autophagy in prostate cancer and cervical cancer cells [32, 33]. A recent study confirmed that RARRES1 contributed to the regulation of dendritic cells, and acted as a novel immune-related biomarker for glioblastoma [34]. TNFAIP2 is abundant in immune cells like myelomonocytic cells, endothelial cells, peripheral blood monocytes, dendritic cells, intestinal M cell, and macrophages [35–37]. And it played essential roles in inflammation, cell proliferation, angiogenesis, migration, membrane nanotube formation [38]. ZNF385B belongs to the family of zinc-finger genes and encodes transcription factors. In serous ovarian cancer, the increased mRNA levels of ZNF385B contributed to shorter survival time of ovarian cancer patients [39].
For the 2 miRNAs (hsa-miR-9-5p and hsa-miR-183-5p) and 3 lncRNAs (CTD-2228K2.7, HOTAIRM1 and NCF1C), the expression of hsa-miR-9-5p could be downregulated by lncRNA TUG1, and TUG1 overturned the effect of miR-9-5p on the proliferation, colony formation, cell cycle arrest, and apoptosis in bone SARCs cells [40]. Hsa-miR-183-5p was the first found in SARCs, highly expressed hsa-miR-183-5p might be associated with a poor prognosis for Clear cell renal cell carcinoma patients [41]. In addition, it was also considered as the optimal diagnostic biomarkers for hepatocellular carcinoma [42]. The lncRNA HOTAIRM1(HOXA transcript antisense RNA myeloid-specific 1) is a natural antisense transcript of HOXA1 gene and expresses in the myeloid lineage and induced during neuronal differentiation [43, 44]. The authors reported that HOTAIRM1/HOXA1 could downregulate the expression of suppressive molecules released by MDSCs and increase the antitumor immune response [45]. It has been reported to be involved in many malignant diseases [45–48], while it was the first to be found in SARCs. However, there is no study on the roles of CTD-2228K2.7 and NCF1C in TME of SARCs patients, which means that more investigations are needed.