Skip to main content
NCBI home page
Discover Oncology logoLink to Discover Oncology
. 2026 Apr 13;17:786. doi: 10.1007/s12672-026-04946-w

Construction and validation of a novel prognostic signature that correlates with immune infiltration based on glycolysis related lncRNAs in osteosarcoma

Shiwei Ma 1, Liling Yang 2, Hongyuan Liu 3, Xue Gong 4, Zhihua Xu 4, Dan Zhang 4,✉
PMCID: PMC13201820  PMID: 41973179

Abstract

Background

Osteosarcoma is the most frequent malignancy for adolescents in bone system, which has caused huge disease burdens for both patients and society. Aberrant activation of glycolysis has been proven to promote osteosarcoma progression. But the prognostic roles of glycolysis-related lncRNAs still remain obscure. We aim to establish a glycolysis-related lncRNA-based signature to predict the prognosis of osteosarcoma.

Methods

Glycolysis-related genes were retrieved from GSEA website. Differentially expressed genes between osteosarcoma samples and the normal samples were determined in GSE36001. Glycolysis-related lncRNA were identified by correlation analysis. The prognosis model was constructed by Cox regression analysis and LASSO analysis. Independent prognostic factors of osteosarcoma were determined by multivariate Cox regression analysis. A prognostic nomogram was further developed with estimation of ROC curves, calibration curves and DCA curves. GO and KEGG enrichment analyses and immune infiltration analyses (ssGSEA, xCELL and CIBERSORT) were conducted to characterize intrinsic features of risk group.

Results

A 11-glycolysis-related lncRNA prognosis model was constructed and validated in validation cohort and several clinical subgroups. ROC curve results showed that the AUC values of training set and validation set were 0.935 and 0.750. Risk score and metastasis status were identified as independent prognostic factors whereby the nomogram was built on them. The robust discriminability of the nomogram was verified. Functional and immunological characterization identified that some cancer-related signaling pathways and mutiple immune infiltrating cells are significantly associated with the signature.

Conclusions

This study revealed the prognostic value of glycolytic-associated lncRNAs in osteosarcoma. The novel glycolysis-associated lncRNA-based model may offer new insights into prognostic evaluation and clinical risk decision for patients with osteosarcoma.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-026-04946-w.

Keywords: Osteosarcoma, lncRNA, Glycolysis, Immune, Nomogram

Introduction

Osteosarcoma is the most frequent cancer in bone system with an incidence rate of 3–4 people per million, which is relatively rare compared with other prevalent cancers like lung cancer and female breast cancer [1]. Epidemiological data reveals that 15–19 years and 75–79 years are the main age periods with occurrence peak of osteosarcoma, and adolescents account for over 70% of all patients whereby causing huge disease burden for adolescents and their families [1]. Osteosarcoma is derived from osteoblast mesenchymal stem cells, thus metaphysis of long bones are the most common primary lessions, especially at the distal femur and proximal tibia [2]. Generally, there are three histopathological subtypes of osteosarcoma: osteoblastic osteosarcoma, chondroblastic osteosarcoma and broblastic osteosarcoma, among which osteoblastic osteosarcoma accounts for about half of all cases [3]. Characterized with local advance and early distant metastasis, osteosarcoma is entitled as a highly invasive malignancy [2]. Previously, surgical operation was the major treatment for osteosarcoma, however the prognosis remained unsatisfactory [4]. Most currently, with the development of multiagent chemotherapy, surgical resection plus chemotherapy has been identified as the standard therapy for osteosarcoma, the clinical outcomes of osteosarcoma patients have been significant improvement, the 5-year overall survival (OS) has risen to about 70% over the past decades [5]. In the meanwhile, other treatment strategies like radiotherapy, immunotherapy and molecular targeted therapy also contributes to the prognosis of osteosarcoma [6]. But, local recurrence and distant metastasis are still the bottlenecks remained to be cracked along the way, to increase survival probabilities of the whole population. Therefore, predicting the prognosis of osteosarcoma accurately attaches significant to clinical treatment and risk decision, which may provide possible priority for survival improvement.

Cellular energetics dysregulation has been ascertained as one of the ten hallmarkers of cancer cell [7]. Glycolysis is the dominant energy-producing process in cancer cells regardless of oxygen flux, the flow of ATP via glycolysis accounts for 60%, this phenomenon is known as the “Warburg effect” [8, 9]. It is generally believed that the abnormal activation of P13K/AKT and RAS/MAPK pathways, and the inactivation of tumor suppressor gene TP53, upregulates HIF-1α and c-Myc whereby enhancing the levels of glycolytic genes and inhibits oxidative phosphorylation through pyruvate dehydrogenase, which ultimately propels the transformation of metabolic reprogramming pattern in tumor [10–12]. In order to satisfy cell survival needs, cancer cells incline to absorb abundant glucose to offset the low energy-producing efficacy via glycolysis, in the meanwhile, macromolecular precursors are also synthesized as the material basis for infinite proliferation of cancer cells [13, 14]. Apart from that, NADPH produced by the glycolysis bypass is indispensable for cellular reduction-oxidation equilibrium state [15]. Glycolysis is reported to promote the anti-apoptosis and chemotherapy resistance of tumor, and also enhancing invasion and metastasis capability [16].Thus glycolysis is pivotal in caner onset and progression, but the concrete mechanisms is not fully understood, especially the associations with the survival prognosis of patients.

LncRNA is a group of small nucleotide fragments, exerting irreplaceable regulatory effects on cellular signal transduction network. Most currently, the prognostic significance of glycolysis-associated lncRNAs in several cancers has been unfolded, which may contribute to guide personalized treatment [17–19]. However, the roles of glycolysis-related lncRNAs in osteosarcoma have not been systematically assessed. Further investigation of glycolysis-associated lncRNAs may offer new insights into prognosis prediction and therapeutic targets in osteosarcoma.

In this present research, we managed to establish a prognostic risk scoring model based on glycolysis-associated lncRNAs to predict the OS of osteosarcoma. Clinical associations, functional characterizations, immune infiltration levels of the signature were analyzed in sequence. Results of which may offer implications for glycolysis-related lncRNAs studies and serve as basis for further experiments in osteosarcoma.

Methods

Data extraction and determination of glycolysis-related lncRNAs

This research was conducted based on the workflow diagram shown in Supplementary material Fig. 1. The transcriptome data of GSE36001 extraction from Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi? acc=GSE36001), which contained19 osteosarcoma cells and 6 normal cells. We also accessed the transcriptome and clinical data of osteocarcoma from The Cancer Genome Atlas (TCGA) (https://portal.gdc.cancer.gov/projects/TARGET-OS), which contained 84 osteosarcoma samples. Samples without clinical information and survival information were excluded. We typed “glycolysis” as key word to obtain the glycolysis-related genes from GSEA website (https://www.gsea-msigdb.org/gsea/msigdb/human/geneset/BIOCARTA_GLYCOLYSIS_PATHWAY.html? keywords=glycolysis). With “DESeq2” R package, we acquired the differentially expressed glycolysis-related genes (DEGGs) between osteosarcoma cells and normal cells in GSE36001. Genes with |Log2FoldChange| > 1 and P < 0.05 were DEGGs. Correlation analysis was then processed between DEGGs and lncRNAs. lncRNAs with Coefficient > 0.4 and P < 0.05 were identified as glycolysis-associated lncRNAs. All P-values in the correlation analysis were subjected to FDR correction to control for multiple testing.

Establish the signature with glycolysis-associated lncRNAs

TCGA osteosarcoma samples were randomly divided into training samples (50%) and validation samples (50%). To narrow the extent, univariate Cox regression analysis was processed to further determine the glycolysis-associated lncRNAs relative to outcomes. A P-value threshold of 0.05 was used to select variables, ensuring the inclusion of statistically significant factors. The least absolute shrinkage and selection operator (LASSO) regression analysis was employed to establish a prognostic model based on these glycolysis-associated lncRNAs in training samples to calculate the risk score of osteosarcoma patients. Patients were divided into high- and low-risk groups on average according to their scores. With “survminer” and “survival” R packages, we compared whether there was a difference in OS between high- and low-risk groups. Predictive capability of the model was checked out by ROC curve via “timeROC” and “ggplot2” R packages. Furthermore, the risk score distribution was displayed via “ggplot2” R package. Predictive capability of the model was checked out subsequently verified in the validation samples. Furthermore, we evaluated the predictive efficacy of the signature in different clinical subgroups, including ≤ 18 years, > 18 years, female, male, no metastasis and metastasis.

Establishment of nomogram with risk score and clinical information

We next conducted Cox regression analyses to extraction independent prognostic predictors. Parameters with P < 0.05 from the multivariate Cox were further utilized to develop a nomogram to predict OS for patients with osteosarcoma at 1-, 3- and 5-year. ROC curve and correction curve were used to determine the prediction effect of nomogram. Decision curve analysis (DCA) was employed to assess the advantages in clinical benefit of the nomogram at 1-, 3- and 5-year.

Functional enrichment analysis

First, the differentially expressed genes (DEGs) between the training set and the validation set were integrated to form the low-risk and high-risk groups. Second submitted the DEGs to GO and KEGG functional enrichment analyses to identify bioactivities that are significantly activated in high risk samples based on “clusterProfiler”, “org.Hs.eg.db” and “GOplot” R packages. As for gene set enrichment analysis (GSEA) analysis, we downloaded gene set “h.all.v7.4.entrez.gmt” from MsigDB and then compared the enriched hallmarks between the two groups.

Relationship between the risk score and immune infiltration

Single-sample gene set enrichment analysis (ssGSEA) was utilized to compare the differences of 16 cells infiltration levels and 13 function between the training set and the validation set were integrated to form the low-risk and high-risk groups. xCELL algorithm was employed to analyze the infiltration differences of 62 immune cells within high- and low-risk groups. CIBERSORT algorithm was employed to identify the correlation between glycolysis-related lncRNAs and 22 immune cells. Spearman correlation analysis was performed to evaluate the relationships between these glycolysis-related lncRNAs and the immune cell types, as this method is particularly suited for analyzing monotonic relationships that may not follow a linear pattern.

qPCR

This study was approved by the Ethical Review Committee (Biomedical ethics committee of Mianyang Central Hospital). Total RNA was extracted using Trizol reagent (Vazyme, Nanjing, China) following the manufacturer’s protocols. Then, RNA samples were reverse transcribed by Hiscript III Reverse Transcriptase kit (Vazyme, Nanjing, China) and the corresponding RNA expression was evaluated by qRT-PCR with ChamQTM Universal SYBR qPCR Master Mix kit (Vazyme, Nanjing, China). GAPDH acted as an internal reference for normalization.

Statistical analysis

Statistical analyses were all conducted by R software 4.1.3. The comparison of K-M survival curves in this study was achieved by Cox regression analysis. The comparison of differences of expression levels (genes and immune infiltrating cells) between high- and low-risk groups were completed by Wilcoxon rank sum test. P < 0.05 was deemed as statistically significant.

Results

Determination of glycolysis-associated lncRNAs

Obtained glycolysis-associated genes from GSEA website. In GSE36001, 24 glycolysis-related genes were upregulated in cancer samples, and 31 glycolysis-related genes were upregulated in normal samples (Fig. 1A-B, Supplementary material Table 1). Subsequently, we acquired 451 lncRNAs that significantly correlate with DEGs via correlation analysis in TCGA cohort, which were named by glycolysis-related lncRNAs (Supplementary material Table 2).

Fig. 1.

Fig. 1

Open in a new tab

Identification of DEGGs in GSE36001. A Volcano plot of the DEGs. B Heatmap of the DEGs

Establishment of the prognostic signature

22 glycolysis-related lncRNAs that significantly correlate with OS were screened out in TCGA cohort (Fig. 2A). The 22 glycolysis-related lncRNAs were thrown into LASSO analysis, then a 11-lncRNA signature was then generated (Fig. 2B-C). Risk score = (-0.0776)*OGFR.AS1+(-0.0153)*AC114760.2 + (-0.0742)*LINC01976 + (-0.0251)*AC016813.1 + (0.0913)*AP003419.2 + (-0.0405)*COL4A2.AS1 + (-0.0902)*AL133330.1+(-0.0540)*AL391863.1+ (0.2896)*AC090236.2 + (-0.9196)*SMIM25 + (0.2896)*HIF1A.AS2. The K-M survival curve showed that the prognosis of high-risk group was worse than that of low-risk group (P = 1.507E-04) (Fig. 3A). The risk score and survival time between the high- and low-risk groups was displayed in Fig. 3B-D. There were obviously more dead samples (red dots) in high risk group. AUC from the ROC curve were 0.935, which was favourable (Fig. 3E). Corresponding analyses were conducted again in validation cohort, results of which were consistent with that in training cohort (Fig. 3F-I). AUC from the ROC curve were 0.750 in validation group (Fig. 3J).

Fig. 2.

Fig. 2

Open in a new tab

Establishment of the model based on glycolysis-associated lncRNAs. A Univariate Cox of glycolysis-associated lncRNAs. B-C LASSO regression analysis

Fig. 3.

Fig. 3

Open in a new tab

Evaluation of the signature. A Survival difference between low- and high-risk groups in training samples. B-D Risk score and survival time between the two risk groups in training samples. E ROC curve of the signature in training samples. F Survival difference between low- and high-risk groups in validation samples. G-I Risk score and survival time between the two risk groups in validation samples. J ROC curve of the signature in validation samples

Application of the model in clinical subgroups

In terms of training samples, the survival differences in different risk groups were well distinguished in all the subgroups but > 18 years subgroup (Fig. 4A-F). As for validation cohort, the survival differences in different risk groups were well distinguished in ≤ 18 years, female and metastasis subgroups, while it revealed no significance in > 18 years, male and no metastasis subgroups (Fig. 4G and L).

Fig. 4.

Fig. 4

Open in a new tab

Subgroup analysis of the model. In training samples, A ≤ 18 years subgroup, B > 18 years subgroup, C female subgroup, D male subgroup, E no metastasis subgroup and F metastasis subgroup. In validation samples, G ≤ 18 years subgroup, H > 18 years subgroup, I female subgroup, J male subgroup, K no metastasis subgroup, L metastasis subgroup

Development of a prognostic nomogram

Metastasis status and risk score were deciphered to be independent prognostic factors in osteosarcoma via Cox regression analyses (Fig. 5A-B). The two factors were subsequently utilized to develop nomogram for OS prediction (Fig. 6A). AUCs at 1-, 3- and 5-year from the ROC curves were favourable in both training samples and validation samples (Fig. 6B-C). Calibration curves were close to the ideal line, suggesting excellent predictive efficacy of the nomogram (Fig. 6D-E). Additionally, results from DCA elicited that the nomogram achieves obvious clinical decision-making advantages (Fig. 6F-G).

Fig. 5.

Fig. 5

Open in a new tab

Cox regression analyses. A Univariate Cox regression. B Multivariate Cox regression

Fig. 6.

Fig. 6

Open in a new tab

Development of nomogram for patients with osteosarcoma. A The prognostic nomogram. ROC curves of the nomogram in B training cohort and C validation cohort. Calibration curves of the nomogram in D training cohort and E validation cohort. DCA of the nomogram in F training samples and G validation samples

Functional enrichment analysis

The DEGs between the low- and high-risk groups were determined (Supplementary material Fig. 2A-B). The biological activities that are significantly ativated in high risk group were ascertained to be osteoclast differentiation, toll-like receptor signaling pathway, T cell activation, regulation of inflammatory response, cytokine metabolic process, endoplasmic reticulum lumen, phagocytic vesicle, phagocytic vesicle membrane, immunological synapse and NADPH oxidase complex (Fig. 7A-D). For the GSEA analysis, gene set “h.all.v7.4.entrez.gmt” was used to further investigate the possible biological behaviors of the two groups. Hallmarks were selected upon setting the significant criteria at FDR P value < 0.05, and |normalized enrichment score| > 1. Hallmarks enriched in high-risk subgroup were mainly relevant to PATHWAYS IN CANCER and CYTOKINE SIGNALING IN IMMUNE SYSTEM. Hallmarks enriched in high‐risk subgroup were mainly relevant to TYROBP CAUSAL NETWORK and EXTRACELLULAR MATRIX ORGANIZATION (Fig. 7F).

Fig. 7.

Fig. 7

Open in a new tab

Functional enrichment. A KEGG. B Biological process (BP). C Cellular component (CC). D Molecular function (MF). E Gene sets enriched in high-risk group and Gene sets enriched in low‐risk group

Relationship between the risk score and immune infiltration

Results from ssGSEA determined that the infiltration of DCs, macrophages, neutrophils, pDCs, T helper cells, Th1 cells and TIL in high risk group were lower (Fig. 8A). Immune signatures like CCR, check point, inflammation promoting, parainflammation and T cell co-inhibition in high risk group were also lower (Fig. 8B). xCELL revealed that the levels of macrophages, iDC, macrophages M1, macrophages M2, monocytes, neurons, pDC, CD4+ T cells, pericytes and sebocytes in high risk group were lower (Fig. 9). We displayed the infiltration of 22 immune cells in training cohort by utilizing CIBERSORT (Fig. 10A). The relationship between the 11 glycolysis-associated lncRNAs and 22 immune infiltrating cells were shown in Fig. 10B. It was worth mentioning that AP003419.2 significantly positively correlates with NK cells resting, while AP003419.2 significantly negatively correlates with NK cells activated.

Fig. 8.

Fig. 8

Open in a new tab

Evaluation by ssGSEA. A Infiltration levels of 13 immune cells between two groups. B Differences of 16 immune signatures between two groups

Fig. 9.

Fig. 9

Open in a new tab

The levels of 62 immune cells between two group by xCELL

Fig. 10.

Fig. 10

Open in a new tab

Evaluation by CIBERSORT. A The infiltration levels of 22 immune cells in training cohort. B Correlations between 11 glycolysis-related lncRNAs and 22 immune infiltrating cells

Validation of glycolysis-related lncRNAs expression levels

As seen in Fig. 11, the expression levels of AL391863.1, COL4A2.AS1, AL133330.1, AC016813.1, HIF1A.AS2, LINC01976, AC090236.2, AC114760.2 and AP003419.2 were significantly higher in osteosarcoma, while SMIM2 and OGFR.AS1 were significantly lower. The detailed sequence of primers used was listed in Supplementary material Table 3.

Fig. 11.

Fig. 11

Open in a new tab

Validation of the expression of the selected glycolysis-related lncRNAs in osteosarcoma and control group (n = 6). *p < 0.05, **p < 0.01, ***p < 0.001

Discussion

Osteosarcoma has caused huge disease burdens for both patients and the society [1, 2]. Although the entire prognosis of osteosarcoma has been optimized over the past decades by multiple therapies, the clinical outcomes of patients with local recurrence and metastasis still remained unsatisfactory [20]. Drug resistance also hindered the progress of current osteosarcoma treatment [21]. Thus, it is necessary to exploit reliable signatures for prognosis prediction in osteosarcoma, to guide clinical risk decision and personalized therapy. Aberrant activation of glycolysis in osteosarcoma cells has been proven to propel malignant phenotypes, including proliferation, invasion, metastasis and drug resistance [22]. lncRNAs are important regulators in cellular signal transduction network. Several lncRNAs have been ascertained to be enrolled in glycolysis modulation in osteosarcoma [23–25]. These lncRNAs relevant to glycolysis may provide new insights into osteosarcoma treatment. Most currently, application of prognostic model based on glycolysis-associated lncRNAs to predict the OS has been reported in different cancer types sequentially, like liver cancer, bladder cancer and renal cancer [26, 27]. However, there is still no comprehensive study to investigate the utility of glycolysis-related lncRNAs in prognosis prediction for patients with osteosarcoma. Herein, a robust glycolysis-related lncRNAs-based prognostic signature was established and validated in osteosarcoma, and its associations with clinical parameters and immune infiltration were also assessed.

Glycolysis-related genes were retrieved from GSEA website. After differential analysis and correlation analysis in GSE36001 cohort and TCGA cohort respectively, we finally acquired 451 glycolysis-related lncRNAs. 22 glycolysis-associated lncRNAs were determined to significantly correlate with the OS of osteosarcoma by univariate Cox analysis. Subsequently, a prognostic signature was constructed by LASSO analysis, which contained 11 glycolysis-related lncRNAs. Some of the 11 glycolysis-related lncRNAs have been reported to be engaged in the onset and development of malignancies. Yu et al. [28]. observed that COL4A2.AS1 can enhance glycolysis and proliferation via miR-20b-5p/HIF1A axis in colorectal cancer cells. Upregulation of COL4A2.AS1 was valuable to infer poor prognosis for patients with breast cancer [29]. AL133330.1 was exploited to be a positive prognostic predictor in laryngeal squamous cell carcinoma, and its aberrant expression also correlated with prognosis [30]. Tang et al. [31]. found that SMIM25 is significantly downregulated in lung adenocarcinoma both in vivo and in vitro, and it was further evaluated as an independent risky factor. SMIM2 serves as a significant prognostic biomarker in mucinous colonic adenocarcinoma, where its upregulation is closely linked to reduced patient survival, implicating its role in tumor progression and offering a potential avenue for targeted therapy [32]. HIF1A.AS2 was unfolded to be a common tumor promoter gene in diverse cancer types to facilitate malignant potential by means of sponging downstream miRNAs in ceRNA network [33–36]. Particularly, miR-129-5p and miR-33b-5p were verified to be the direct targets of HIF1A.AS2 in osteosarcoma cells in vitro, and malignant phenotypes could be significantly affected by regulating HIF1A.AS2 expression [37, 38]. Wang et al. [37]. also identified that the levels of HIF1A.AS2 in osteosarcoma are higher than that in the paracancerous, and patients with higher HIF1A.AS2 levels suffered from poorer prognosis. The expression of LINC01976 in thyroid cancer is significantly associated with altered patterns of immune cell infiltration, suggesting its potential role as a modulator of the tumor immune microenvironment [39]. To date, no studies have reported on the expression patterns or functional implications of AL391863.1, AC016813.1, AC090236.2, AC114760.2, AP003419.2, and OGFR.AS1 in osteosarcoma, highlighting these lncRNAs as novel candidates for future biomarker or therapeutic research. Characterizing these molecules presents a unique opportunity to uncover potentially critical nodes in OS biology, which could lead to the development of much-needed diagnostic indicators or novel therapeutic targets for this aggressive bone tumor. Subsequently, patients were divided into low- and high-risk groups according to the risk scores. The OS differences of patients within the low- and high-risk groups were well discriminated against in both training samples and validation samples. The predictive efficiency of the prognostic signature was identified to be promising by time-dependent ROC curves. To further analyse the value of the model in prognosis prediction, clinical subgroup analysis was executed. It turned out to be that the signature also exerts good predictive capability in different subgroups.

To explore the prognostic utility of the model, we confirmed that the risk score together with clinicopathological characteristics like metastasis status are independent prognostic predictors in osteosarcoma via Cox regression analyses. Nomogram was established accordingly. The integrated nomogram showed favourable capability in survival prediction and advantage in clinical benefit by assessment of ROC curves, calibration curves and DCA.

Functional characterization may help to further understand the intrinsic features of risk group. Thus we submitted the DEGs of high risk group to GO/KEGG functional enrichment analyses to ascertain the biological activities that are significantly enriched. We mainly labeled the enriched biological activities such as: T cell receptor signaling pathway, toll-like receptor signaling pathway, regulation of inflammatory response and cytokine metabolic process. Researches revealed that γδ T cells-mediated cytoxicity to osteosarcoma cells can be enhanced by cytokines like IFN-γ whereby we could hypothesize that more γδ T cells are activated and recruited by cytokines in the tumor microenvironment (TME) to mediate cytotoxicity to osteosarcoma cells with higher malignant potentials in high risk group [40]. Endogenous ligands from host-cell origin to regulate toll-like receptors (TLRs) were also named as damage-associated molecular patterns (DAMPs), which can be activated and secreted to respond to tissue damage via enhancing inflammatory response [41]. In the meanwhile, cancer-related inflammation activated by toll-like receptor was also utilized by osteosarcoma cells to alleviate apoptosis induced by tissue damage [42], which was consistent with the enrichment of toll-like receptor signaling pathway and regulation of inflammatory response in high risk group rendering poorer prognosis. However, the role of reactive oxygen species appeared to be contradictory in osteosarcoma. On the one hand, mitochongdria-mediated iron accumulation was able to propel the progression and Warburg effect of osteosarcoma cells, and reactive oxygen species induced by iron attached importance to these activities [43]. On the other hand, reactive oxygen species could induce cell apoptosis via mitochondria dysfunction in osteosarcoma cells [43]. Thus, the effect of reactive oxygen species in osteosarcoma needed further investigations.

Subsequently, ssGSEA, xCELL and CIBERSORT algorithms were used to estimate the associations between the signature and tumor infiltrating immune cells. Results from ssGSEA and xCELL were complementary and mainly demonstrated that the infiltration levels of DCs (iDC, pDC), macrophages (M1, M2), T cells (CD4+ T cells, T helper cells, Th1 cells and TIL), neutrophils, monocytes, neurons, pericytes and sebocytes were higher in low risk group. Dendritic cells (DCs) were important to enhance T cell activation by antigen presentation to exert anti-tumor immunity [44, 45]. Thus higher infiltration levels of DCs in low risk groups may infer better prognosis, which was consistent with our results. It is generally accepted that high infiltration levels of T cells represent favourable clinical outcomes [46, 47]. We also determined similar results in this present study that low risk groups with higher infiltration levels of T cells harbor more satisfactory clinical outcomes, which may possibly result from enhanced activation by DCs. Macrophages 2 (M2) were reported to facilitate metastasis of osteosarcoma cells and inferred poor outcome [48, 49]. However, we found higher infiltration levels of M2 in low risk group with better prognosis, possibly due to small sample size. Thus, advanced technologies were required for further investigation like single cell sequencing and flow cytometry. Additionally, we determined that some glycolysis-related lncRNAs in the signature are significantly correlated with several immune infiltrating cells via CIBERSORT algorithm, indicating their functional associations. Experimental evidence in depth would be more beneficial to clarify the concrete mechanisms of the reciprocity between glycolysis-related lncRNAs and immune infiltrating cells.

However, there are certain limitations in study. Firstly, samples from real clinical patients are needed to get preciser verification of the level of the 11 glycolysis-associated lncRNAs. Secondly, prospective study with large osteosarcoma cohort from multiple centers will make the signature and corresponding results more convincing. Thirdly, more experimental studies are required to further unfold the obscure regulatory axes and functional characterizations of the 11 glycolysis-related lncRNAs.

Conclusion

This study constructs and validates a robust prognostic model based on glycolysis-related lncRNAs in osteosarcoma. The model could be applied as an independent prognostic predictor for osteosarcoma. Functional and immunological characterization identified its associations with some cancer-related signaling pathways and mutiple immune infiltrating cells. The glycolysis-related lncRNA-based signature may provide insight into prognosis evaluation and risk decision-making for osteosarcoma patients in clinics.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1. (13.9KB, xlsx)
Supplementary Material 2. (41.3KB, xlsx)
Supplementary Material 3. (10.2KB, xlsx)
Supplementary Material 4. (144.6KB, docx)
Supplementary Material 5. (223.6KB, docx)

Acknowledgements

We are grateful for the research data provided by the GEO database.

Author contributions

SM and DZ conceived and designed the study, prepared Fig.s and approved the final draft. SM, LY ang HL authored or reviewed drafts of the paper and approved the final draft. SM, XG, ZX, ang DZ analyzed the data.

Funding

The work was supported by Natural Science Foundation of Mianyang Central Hospital (2020FH08).

Data availability

The transcriptome data of GSE36001 extraction from Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi? acc=GSE36001). The transcriptome and clinical data of osteocarcoma extraction from The Cancer Genome Atlas (TCGA) (https://portal.gdc.cancer.gov/projects/TARGET-OS). The glycolysis-related genes extraction from GSEA website (https://www.gsea-msigdb.org/gsea/msigdb/human/geneset/BIOCARTA_GLYCOLYSIS_PATHWAY.html? keywords=glycolysis), click on “Download gene set” on the page to download it. If you cannot access it directly, you may need to register an account for GSEA.

Declarations

Ethics approval and consent to participate

This research has been approved by the Ethics Review Committee of Mianyang Central Hospital.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

  • 1.Sadykova LR, Ntekim AI, Muyangwa-Semenova M, Rutland CS, Jeyapalan JN, Blatt N, et al. Epidemiology and Risk Factors of Osteosarcoma. Cancer Invest. 2020;38(5):259–69. [DOI] [PubMed] [Google Scholar]
  • 2.Eaton BR, Schwarz R, Vatner R, Yeh B, Claude L, Indelicato DJ et al. Osteosarcoma Pediatr Blood Cancer. 2021;68 Suppl 2(e28352. [DOI] [PubMed]
  • 3.Brion R, Regnier L, Mullard M, Amiaud J, Rédini F, Verrecchia F. LIM Kinases in Osteosarcoma Development. Cells. 2021;10(12). [DOI] [PMC free article] [PubMed]
  • 4.Ohno T, Abe M, Tateishi A, Kako K, Miki H. Osteogenic sarcoma. A study of one hundred and thirty cases. J Bone Joint Surg Am. 1975;57(3):397–404. [PubMed] [Google Scholar]
  • 5.Belayneh R, Fourman MS, Bhogal S, Weiss KR. Update on Osteosarcoma. Curr Oncol Rep. 2021;23(6):71. [DOI] [PubMed] [Google Scholar]
  • 6.Harrison DJ, Geller DS, Gill JD, Lewis VO, Gorlick R. Current and future therapeutic approaches for osteosarcoma. Expert Rev Anticancer Ther. 2018;18(1):39–50. [DOI] [PubMed] [Google Scholar]
  • 7.Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. 2011;144(5):646–74. [DOI] [PubMed] [Google Scholar]
  • 8.Vander Heiden MG, Cantley LC, Thompson CB. Understanding the Warburg effect: the metabolic requirements of cell proliferation. Science. 2009;324(5930):1029–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Koppenol WH, Bounds PL, Dang CV. Otto Warburg’s contributions to current concepts of cancer metabolism. Nat Rev Cancer. 2011;11(5):325–37. [DOI] [PubMed] [Google Scholar]
  • 10.DeBerardinis RJ, Lum JJ, Hatzivassiliou G, Thompson CB. The biology of cancer: metabolic reprogramming fuels cell growth and proliferation. Cell Metab. 2008;7(1):11–20. [DOI] [PubMed] [Google Scholar]
  • 11.Chen X, Zhao X, Cai H, Sun H, Hu Y, Huang X, et al. The role of sodium hydrosulfide in attenuating the aging process via PI3K/AKT and CaMKKβ/AMPK pathways. Redox Biol. 2017;12:987–1003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Kruiswijk F, Labuschagne CF, Vousden KH. p53 in survival, death and metabolic health: a lifeguard with a licence to kill. Nat Rev Mol Cell Biol. 2015;16(7):393–405. [DOI] [PubMed] [Google Scholar]
  • 13.Lu J, Tan M, Cai Q. The Warburg effect in tumor progression: mitochondrial oxidative metabolism as an anti-metastasis mechanism. Cancer Lett. 2015;356(2 Pt A):156–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Gatenby RA, Gillies RJ. Why do cancers have high aerobic glycolysis? Nat Rev Cancer. 2004;4(11):891–9. [DOI] [PubMed] [Google Scholar]
  • 15.Hamanaka RB, Chandel NS. Cell biology. Warburg effect and redox balance. Science. 2011;334(6060):1219–20. [DOI] [PubMed] [Google Scholar]
  • 16.Ngo H, Tortorella SM, Ververis K, Karagiannis TC. The Warburg effect: molecular aspects and therapeutic possibilities. Mol Biol Rep. 2015;42(4):825–34. [DOI] [PubMed] [Google Scholar]
  • 17.Zhong X, He X, Wang Y, Hu Z, Huang H, Zhao S, et al. Construction of a prognostic glycolysis-related lncRNA signature for patients with colorectal cancer. Cancer Med. 2023;12(1):930–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Zou J, Gu Y, Zhu Q, Li X, Qin L. Identifying glycolysis-related LncRNAs for predicting prognosis in breast cancer patients. Cancer Biomark. 2022;34(3):393–401. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Zheng Z, Lai C, Li W, Zhang C, Ma K, Yao Y. Identification of a Novel Glycolysis-Related LncRNA Signature for Predicting Overall Survival in Patients With Bladder Cancer. Front Genet. 2021;12:720421. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Durfee RA, Mohammed M, Luu HH. Review of Osteosarcoma and Current Management. Rheumatol Ther. 2016;3(2):221–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Hameed M, Dorfman H. Primary malignant bone tumors–recent developments. Semin Diagn Pathol. 2011;28(1):86–101. [DOI] [PubMed] [Google Scholar]
  • 22.Feng Z, Ou Y, Hao L. The roles of glycolysis in osteosarcoma. Front Pharmacol. 2022;13:950886. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Shen Y, Xu J, Pan X, Zhang Y, Weng Y, Zhou D, et al. LncRNA KCNQ1OT1 sponges miR-34c-5p to promote osteosarcoma growth via ALDOA enhanced aerobic glycolysis. Cell Death Dis. 2020;11(4):278. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Pu F, Liu J, Jing D, Chen F, Huang X, Shi D et al. LncCCAT1 interaction protein PKM2 upregulates SREBP2 phosphorylation to promote osteosarcoma tumorigenesis by enhancing the Warburg effect and lipogenesis. Int J Oncol. 2022;60(4). [DOI] [PMC free article] [PubMed]
  • 25.Pan X, Guo J, Liu C, Pan Z, Yang Z, Yao X, et al. LncRNA HCG18 promotes osteosarcoma growth by enhanced aerobic glycolysis via the miR-365a-3p/PGK1 axis. Cell Mol Biol Lett. 2022;27(1):5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Bai Y, Lin H, Chen J, Wu Y, Yu S. Identification of Prognostic Glycolysis-Related lncRNA Signature in Tumor Immune Microenvironment of Hepatocellular Carcinoma. Front Mol Biosci. 2021;8:645084. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Li T, Tong H, Zhu J, Qin Z, Yin S, Sun Y, et al. Identification of a Three-Glycolysis-Related lncRNA Signature Correlated With Prognosis and Metastasis in Clear Cell Renal Cell Carcinoma. Front Med (Lausanne). 2021;8:777507. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Yu Z, Wang Y, Deng J, Liu D, Zhang L, Shao H, et al. Long non-coding RNA COL4A2-AS1 facilitates cell proliferation and glycolysis of colorectal cancer cells via miR-20b-5p/hypoxia inducible factor 1 alpha subunit axis. Bioengineered. 2021;12(1):6251–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Yao Y, Zhang T, Qi L, Zhou C, Wei J, Feng F, et al. Integrated analysis of co-expression and ceRNA network identifies five lncRNAs as prognostic markers for breast cancer. J Cell Mol Med. 2019;23(12):8410–19. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Qian L, Ni T, Fei B, Sun H, Ni H. An immune-related lncRNA pairs signature to identify the prognosis and predict the immune landscape of laryngeal squamous cell carcinoma. BMC Cancer. 2022;22(1):545. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Tang X, Hua X, Peng X, Pei Y, Chen Z. Integrated Dissection of lncRNA-miRNA-mRNA Pairs and Potential Regulatory Role of lncRNA PCAT19 in Lung Adenocarcinoma. Front Genet. 2021;12:765275. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Gao Y, Ren J, Chen K, Guan G. Construction and validation of a prognostic signature for mucinous colonic adenocarcinoma based on N7-methylguanosine-related long non-coding RNAs. J Gastrointest Oncol. 2024;15(1):203–19. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Guan HM, Li WQ, Liu J, Zhou JY. LncRNA HIF1A-AS2 modulated by HPV16 E6 regulates apoptosis of cervical cancer cells via P53/caspase9/caspase3 axis. Cell Signal. 2022;97:110390. [DOI] [PubMed] [Google Scholar]
  • 34.Si J, Ma Y, Lv C, Hong Y, Tan H, Yang Y. HIF1A-AS2 induces osimertinib resistance in lung adenocarcinoma patients by regulating the miR-146b-5p/IL-6/STAT3 axis. Mol Ther Nucleic Acids. 2021;26(613 – 24. [DOI] [PMC free article] [PubMed]
  • 35.Chen M, Wei X, Shi X, Lu L, Zhang G, Huang Y, et al. LncRNA HIF1A-AS2 accelerates malignant phenotypes of renal carcinoma by modulating miR-30a-5p/SOX4 axis as a ceRNA. Cancer Biol Med. 2021;18(2):587–603. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Wang Y, Zhang G, Han J. HIF1A-AS2 predicts poor prognosis and regulates cell migration and invasion in triple-negative breast cancer. J Cell Biochem. 2019;120(6):10513–18. [DOI] [PubMed] [Google Scholar]
  • 37.Wang X, Peng L, Gong X, Zhang X, Sun R. LncRNA HIF1A-AS2 promotes osteosarcoma progression by acting as a sponge of miR-129-5p. Aging. 2019;11(24):11803–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Lin H, Zhao Z, Hao Y, He J, He J. Long noncoding RNA HIF1A-AS2 facilitates cell survival and migration by sponging miR-33b-5p to modulate SIRT6 expression in osteosarcoma. Biochem Cell Biol. 2020;98(2):284–92. [DOI] [PubMed] [Google Scholar]
  • 39.Zhou Y, Dai X, Lyu J, Li Y, Bao X, Deng F, et al. Construction and validation of a novel prognostic model for thyroid cancer based on N7-methylguanosine modification-related lncRNAs. Med (Baltim). 2022;101(42):e31075. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Li Z, Xu Q, Peng H, Cheng R, Sun Z, Ye Z. IFN-γ enhances HOS and U2OS cell lines susceptibility to γδ T cell-mediated killing through the Fas/Fas ligand pathway. Int Immunopharmacol. 2011;11(4):496–503. [DOI] [PubMed] [Google Scholar]
  • 41.Erridge C. Endogenous ligands of TLR2 and TLR4: agonists or assistants? J Leukoc Biol. 2010;87(6):989–99. [DOI] [PubMed] [Google Scholar]
  • 42.Li GZ, Deng JF, Qi YZ, Liu R, Liu ZX. COLEC12 regulates apoptosis of osteosarcoma through Toll-like receptor 4-activated inflammation. J Clin Lab Anal. 2020;34(11):e23469. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Ni S, Kuang Y, Yuan Y, Yu B. Mitochondrion-mediated iron accumulation promotes carcinogenesis and Warburg effect through reactive oxygen species in osteosarcoma. Cancer Cell Int. 2020;20:399. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Kawano M, Tanaka K, Itonaga I, Iwasaki T, Miyazaki M, Ikeda S, et al. Dendritic cells combined with doxorubicin induces immunogenic cell death and exhibits antitumor effects for osteosarcoma. Oncol Lett. 2016;11(3):2169–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Yu Z, Ma B, Zhou Y, Zhang M, Long H, Wang Y, et al. Allogeneic tumor vaccine produced by electrofusion between osteosarcoma cell line and dendritic cells in the induction of antitumor immunity. Cancer Invest. 2007;25(7):535–41. [DOI] [PubMed] [Google Scholar]
  • 46.Pagès F, Berger A, Camus M, Sanchez-Cabo F, Costes A, Molidor R, et al. Effector memory T cells, early metastasis, and survival in colorectal cancer. N Engl J Med. 2005;353(25):2654–66. [DOI] [PubMed] [Google Scholar]
  • 47.Sato E, Olson SH, Ahn J, Bundy B, Nishikawa H, Qian F, et al. Intraepithelial CD8 + tumor-infiltrating lymphocytes and a high CD8+/regulatory T cell ratio are associated with favorable prognosis in ovarian cancer. Proc Natl Acad Sci U S A. 2005;102(51):18538–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Cersosimo F, Lonardi S, Bernardini G, Telfer B, Mandelli GE, Santucci A et al. Tumor-Associated Macrophages in Osteosarcoma: From Mechanisms to Therapy. Int J Mol Sci. 2020;21(15). [DOI] [PMC free article] [PubMed]
  • 49.Cheng Z, Wang L, Wu C, Huang L, Ruan Y, Xue W. Tumor-derived Exosomes Induced M2 Macrophage Polarization and Promoted the Metastasis of Osteosarcoma Cells Through Tim-3. Arch Med Res. 2021;52(2):200–10. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1. (13.9KB, xlsx)
Supplementary Material 2. (41.3KB, xlsx)
Supplementary Material 3. (10.2KB, xlsx)
Supplementary Material 4. (144.6KB, docx)
Supplementary Material 5. (223.6KB, docx)

Data Availability Statement

The transcriptome data of GSE36001 extraction from Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi? acc=GSE36001). The transcriptome and clinical data of osteocarcoma extraction from The Cancer Genome Atlas (TCGA) (https://portal.gdc.cancer.gov/projects/TARGET-OS). The glycolysis-related genes extraction from GSEA website (https://www.gsea-msigdb.org/gsea/msigdb/human/geneset/BIOCARTA_GLYCOLYSIS_PATHWAY.html? keywords=glycolysis), click on “Download gene set” on the page to download it. If you cannot access it directly, you may need to register an account for GSEA.

Articles from Discover Oncology are provided here courtesy of Springer

RESOURCES