DOI: https://doi.org/10.21203/rs.3.rs-2318090/v1
Background: Osteosarcoma(OS) is the most common primary bone malignancy in ado-lescents. The function of lipid metabolism-related lncRNAs in disease progression and prognosis of osteosarcoma remains unclear. This study aimed to explore the role of lipid metabolism-related lncRNAs in osteosarcoma development and prognosis.
Methods: Pearson correlation was used for identification of lipid metabolism-related lncRNAs, and univariate and multivariate Cox regression analyses were used to construct and validate a risk signature to predict the prognosis of OS patients. Functional analysis using Gene set enrichment analysis (GSEA) to elucidate underlying mechanisms. Analysis of potential regulatory mechanisms of lipid metabolism-related lncRNAs using ceRNA networks, and they were preliminarily verified in our tissues using immunohistochemistry (IHC).
Results: We screened two lipid metabolism-related lncRNAs (SNHG17 and LINC00837) to con-struct a risk signature and validated them in the GEO database. The results showed that this risk model was an independent prognostic factor for OS patients. GSEA analysis showed that this signature may be associated with cell proliferation and metabolism-related pathways in OS patients. Cox regression, ROC curve analysis, and a nomogram indicated that the risk model was an independent prognostic factor and it showed potent potential for survival prediction in osteosarcoma. Nomogram integrating risk model and clinical characteristics could predict the prognosis of osteosarcoma patients accurately. Immunohistochemical results showed that CSNK2A2, MIF and VDAC2 were up-regulated in tumor tissues.
Conclusions: In summary, our study demonstrates that lipid-metabolism related-lncRNA could be applied to predict the prognosis of in osteosarcoma accurately.
Osteosarcoma, is a primary malignant tumor originating from mesenchymal tis-sue, mainly in children and adolescents [1]. In the past 50 years, the treatment of osteosarcoma has hardly changed significantly, and the treatment has entered a plateau. About 35–45% of patients with osteosarcoma have chemotherapy resistance and low response to treatment, and these patients are very prone to distant metastasis. In patients with distant metastases, the 5-year survival rate is less than 20% [2–5]. Although finding effective treatment methods is the focus of current osteosarcoma re-search, reliable disease predictors or assessment systems also have important clinical value in evaluating treatment response, supporting clinical decision-making, monitoring disease recurrence and metastasis, and estimating long-term survival [6].In recent years, a large number of studies have found that the metabolic reprogramming of tumors plays an important role in the occurrence and development of tumors [7]. Abnormal fatty acid metabolism has attracted increasing attention as a feature of metabolic reprogramming in tumors [8, 9]. During tumor progression, the availability of nutrients in the tumor microenvironment is constantly changing, and tumor cells utilize lipid metabolism to sustain their rapid proliferation, survival, migration, invasion, and metastasis [10–13]. Previous studies have shown that lipid metabolism-related genes have strong prognostic potential in renal cell carcinoma, hepatocellular carcinoma, diffuse glioma, and ovarian cancer [14–17]. However, the relevant mechanisms of lipid metabolism-related genes in the occurrence, development and prognosis of osteosarcoma are not fully understood.
LncRNA refers to RNA molecules longer than 300 nucleotides without a protein-coding reading frame [18]. It can participate in multiple regulatory processes of biological pathways such as X chromosome silencing, genomic imprinting, chromatin modification, transcriptional activation, transcriptional interference, and intranuclear transport [19–21]. In osteosarcoma, the abnormal expression level of lncRNA is closely related to the survival time of patients, and most of the abnormal expression of lncRNA is negatively correlated with the survival time of patients with osteosarcoma [22, 23]. Recent reports suggested that lncRNAs could be good candidates for tumor biomarkers and possessed high specificity, high sensitivity, and noninvasive characteristics [24]. However, the value of lipid metabolism related-lncRNA in Osteosarcoma prognostic prediction remains unclear.
In the present study, we used multiple bioinformatics methods to comprehensively analyze lipid metabolism-related lncRNA. Constructed a new lipid metabolism-related lncRNA-based risk score model to evaluate the prognostic value of lipid metabolism-related lncRNA in osteosarcoma. Finally, the regulatory roles of two lncRNAs (SNHG17 and LINC00837) in the risk model on downstream target gene ex-pression were validated using our own osteosarcoma patients’ tissue. Our findings may provide theoretical insights into the clinical prognosis and treatment of subsequent osteosarcoma patients.
The gene expression dataset (GSE12865) was downloaded from Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/Geo/) of the National Center for Bio-technology Information (NCBI). The gene expression dataset GSE12865 included 12 samples from pediatric osteosarcoma and 2 samples of healthy osteoblast cells on the basis of the platform of GPL6244 (Affymetrix Human Gene 1.0 ST Array). We down-loaded the level 3 RNA-seq expression data of 88 patients with OS from TCGA data-base (https://portal.gdc.cancer.gov). Subsequently, the latest clinical data of patients with OS were downloaded from the TARGET database (https://ocg.cancer.gov/programs/target). Finally, both RNA-seq expression data and valid clinical information of 85 patients were used as a training set for analysis in this study. For the validation set, the RNA-sequencing data and corresponding clinicopathological features of 34 patients with OS were downloaded from the GSE16091 database.
After using lipid-specific keywords (fatty acyls, glycerolipids, glycerophospholipids, sphingolipids, sterol lipids, prenol lipids, saccharolipids, and polyketides), 21 lipid me-tabolism-related pathways and five lipid metabolism related gene sets were collected from the Kyoto Encyclopedia of Genes and Genomes (KEGG) website (http://www.kegg.jp/blastkoala/) and the Molecular Signatures Database (MisDB) web-site (https://www.gseamsigdb.org/gsea/msigdb/index.jsp), respectively (reference: Lipid metabolism gene-wide profile and survival signature of lung adenocarcinoma). After removing overlapping genes, a total of 1045 lipid metabolism-related genes were obtained. Lipid metabolism-related differentially expressed genes (DEGs) between 12 samples from pediatric osteosarcoma and 2 samples of healthy osteoblast cells of GSE12865 were screened through R package limma. The parameters set for differential expression analysis were false discovery rate (FDR) < 0.05 and |log2 fold change| (logFC) > 1. Regarding the screening of lncRNAs, we extracted the common lncRNAs from TCGA and GSE16091 datasets, and the Pearson correlation analysis was carried out to analyze lipid metabolism-related lncRNAs, the lncRNAs with |Pearson R| > 0.3 and P value < 0.05 were referred to as lipid metabolism-related lncRNAs and considered for subsequent analysis.
Univariate Cox regression analysis and multivariate Cox regression analysis were further performed to screen lncRNAs significantly associated with survival and determine the coefficients to construct the risk model. The risk-score formula was calculated as follows: risk score = ExpGene1 × Coef 1 + ExpGene2 × Coef 2 + ExpGene3 × Coef 3. Coef is the coefficient and Exp is the normalized expression value of each sig-nature gene. The risk model of 2 lipid metabolism-related lncRNAs was built in the TCGA-TARGET-OS training set and validated in the GSE16091 validation set, and patients were divided into high-risk and low-risk groups according to the medium value. Subsequently, univariate and multivariate Cox regression analyses were used to evaluate the availability of the prognostic model.
Single-sample gene set enrichment analysis (ssGSEA) was performed to assess the the enrichment of the 24 immune infiltrating cells(Activated B cell, Activated CD4 T cell, Activated CD8 T cell, Central memory CD4 T cell, Central memory CD8 T cell, Effector memory CD4 T cell, Effector memory CD8 T cell, Gamma delta T cell, Immature B cell, Memory B cell, Regulatory T cell, T follicular helper cell, Type 1 T helper cell, Type 17 T helper cell, Type 2 T helper cell, Activated dendritic cell, CD56 bright natural killer cell, CD56dim natural killer cell, Eosinophil, Monocyte, Natural killer cell, Natural killer T cell, Neutrophil, Plasmacytoid dendritic cell, Immature dendritic cell, Macrophage, Mast cell, and MDSC) in the tumor samples. And the relationship be-tween risk score and immune cell infiltration was calculated by Spearman correlation.
The nomogram was constructed by multivariate Cox regression analysis, and the calibration curves were established to show the consistency between the predicted probability of nomogram and the observed overall survival rate of OS patients. Draw decision curves and evaluate the clinical effectiveness of the nomogram by estimating the net benefits within the threshold probability range.
Cytoscape v3.9.0 was used to visualize the correlations between lncRNA (SNHG17 and LINC00837) and differentially expressed lipid metabolism-related genes. The target miRNAs of SNHG17 and LINC00837 were predicted by StarBase database. MiRDIP v4.1 online tools (http://ophid.utoronto.ca/mirDIP/index_confirm.jsp)) were used to screen the corresponding target mRNAs of these miRNAs, and Cytoscape v3.9.0 was used to integrate the regulation relationship between lncRNA-miRNA and miRNA-mRNA to construct ceRNA network.
The most common pathways in high-risk group and low-risk group were studied by gene set enrichment analysis (GSEA), and the genes related to lipid metabolism differentially expressed in SNHG17 and LINC00837 were analyzed by gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway by implementing the cluster Profiler R package.
Most analyses were performed using R software and results with a P value. The Kaplan-Meier method and log-rank test were used to evaluate the difference of overall survival rate between high-risk group and low-risk group. Receiver operating characteristic (ROC) curves and area under curve (AUC) values were used to determine the prediction accuracy of the risk model. According to the following clinicopathological features: age (16 years old), gender (male or female), metastatic status (metastatic or non-metastatic), Student’s t-test was performed to compare the risk scores between subgroups.
Primary OS adjacent tumor-free tissues and tumor tissues were collected from the department of Orthopedics, the 920th Hospital of Joint Logistics Support Force of Chi-nese People’s Liberation Army.
Tumor tissues were fixed with 10% formaldehyde and embedded in paraffin. Paraffin-embedded tissue sections were deparaffinized, and the sections were then incu-bated by the primary antibody against CSNK2A2, ODC1, MIF and VDAC2 (1:500, Abcam, USA) overnight at 4°C, using a secondary antibody (30–40 µL, Abcam, USA) at room temperature incubated for 1 h. The nuclei were stained by hematoxylin solution. The images were determined under light microscopy.
A total of 219 differentially expressed lipid metabolism-related genes were identified from GSE12865. A heatmap was created to show the hierarchical clustering analysis of these 219 lipid metabolism-related genes (Fig. 1A), and a volcano plot was constructed to reveal the 219 significant differentially ex-pressed lipid metabolism-related genes (Fig. 1B).
And then, the lncRNA data of OS patients were screened from the gene expression of TCGA-TARGET-OS and GSE16091, and a total of 116 common lncRNAs were identified. By Pearson correlation analysis between these lncRNAs and differentially ex-pressed lipid metabolism-related genes, 113 lipid metabolism-related lncRNAs were screened for further study.
We conducted univariate Cox regression analysis in TARGET-OS and GSE16091 to further investigate the role of these lipid metabolism-related lncRNAs in predicting prognosis of OS patients. The results revealed that 4 lipid metabolism-related lncRNAs were intimately correlated with the overall survival of OS patients (P < 0.05) (Fig. 2A). In addition, we performed multivariate Cox regression analysis to increment the predictive robustness of the five lipid metabolism-related lncRNAs in the TARGET-OS training set. Two lipid metabolism-related lncRNAs were filtered to build a risk model, and the risk score was computed as 0.60286 ×expression value of SNHG17 + 0.64943 ×expression value of LINC00837 (Fig. 2B). We classified patients in the TARGET-OS training set into high-risk and low-risk groups according to the median risk score. The distributions of risk score, survival status of patients with OS in TARGET-OS training set, and the expression values of SNHG17 and LINC00837 are shown in the heatmap (Fig. 2C). The results of Kaplan-Meier analysis showed that the clinical outcome was poorer in the high-risk group (Fig. 2D), and ROC curves demonstrated that TAR-GET-OS training pooled lipid metabolism-related lncRNA signatures were highly ac-curate in predicting overall survival with an AUC value of > 0.65 (Fig. 2E). In the meantime, we also researched the relationship between risk score values and clinicopathological characteristics of OS patients in the TARGET training set. There were no statistically significant differences in the risk scores of OS patients by age ( < = 16 and > 16) and gender (female and male), and patients with metastatic disease had higher risk score values than those with non-metastatic disease (Fig. 2F).
We implemented further validation of the risk model in the GSE16091 dataset. Patients were grouped into low and high risk groups in accordance with the median risk score, and the risk score was calculated using the same formula as in the previous section. Consistent with the results from the TARGET-OS training set, we found that death status was predominantly distributed among patients with higher risk score values (Fig. 3A), with shorter survival times for patients in the high-risk group (Fig. 3B). A heatmap was produced showing the expression values of SNHG17 and LINC00837 in the low and high risk groups (Fig. 3A). The ROC curve also illustrates the high efficiency of the risk model in the prediction of prognosis of OS patients in the GSE39055 dataset (AUC > 0.7, Fig. 3C).
We have conducted a GSEA in the high and low risk groups to investigate potential pathways and hallmarks with the aim of identifying pathways relevant to the risk model. Among the results of KEGG pathway enrichment, we found enrichment for CELL_CYCLE, HUNTINGTONS_DISEASE, OOCYTE_MEIOSIS in the high-risk group of patients (Fig. 4A), while enrichment for ASTHMA, COMPLE-MENT_AND_COAGULATION_CASCADES, and ECM_RECEPTOR_ INTERACTION in the low-risk group of patients (Fig. 4B). Among the hallmarks enrichment results, we discovered that ADIPOGENESIS, DNA_REPAIR, and E2F_TARGETS were enriched in the high-risk group of patients (Fig. 4C), whereas ANGIOGENESIS and APICAL_SURFACE were enriched in the low-risk group of patients (Fig. 4D). With these results suggesting that in osteosarcoma, lipid metabolism-related lncRNA risk signatures may be associated with cell proliferation and metabolism-related pathways. Next, we were to calculate the level of infiltration of 28 immune cells in each sample. Spearman correlation analysis yielded risk scores associated with central memory CD8 T cells (P value < 0.001,r = -0.32), macrophages (P value < 0.001,r = -0.25), natural killer cells (P-value < 0.001,r=-0.30), and plasmacytoid dendritic cells (P-value = 0.03,r=-0.20) were significantly negatively correlated (Fig. 5). As these findings suggest, the lower the risk score the higher the immune cell infiltration in OS patients.
We subsequently conducted univariate and multivariate Cox regression analyses to examine whether risk, gender, age and metastasis were independently correlated with prognosis in OS. Univariate Cox regression analysis showed that risk and metastasis were strongly associated with prognosis (Fig. 6A), and multivariate Cox regression analysis further proved that risk score and metastasis were independent prognostic factors in patients with OS (p < 0.05, Fig. 6B). On the basis of the independent prognostic factors (risk score and metastasis), a nomogram predicting 1/3/5-year survival in OS patients was established (Fig. 6C). An excellent agreement was found between observed and predicted 1-year and 3-year overall survival rates, as shown in the calibration plots (Fig. 6D). The magnitude of the correction is further illustrated by the decision curves (Fig. 6E). In the nomograms, moreover, the AUC values at 1, 3 and 5 years reached 0.935, 0.772 and 0.828, respectively, indicating that the nomograms exhibited superior survival prediction (Fig. 6F).
Analysis of the correlation between lipid metabolism-related lncrna (SNHG17 and LINC00837) and lipid metabolism-related differentially expressed genes. A lipid metabolism-related lncRNA-mRNA network was then established and visualized using CytoScape 3.9.0 software (Fig. 7A). A KEGG enrichment analysis showed that these genes were enriched in a variety of metabolism-related pathways, including carbon metabolism, the citric acid cycle (TCA cycle), metabolism of glyoxylic esters and dicarboxylic esters (Fig. 7B). The results of GO term annotation indicated that these genes have significant association with lipid metabolism (Fig. 7C). For further elucidation of how lipid metabolism-related lncRNAs (SNHG17 and LINC00837) regulate lipid metabolism-related genes by sponging miRNA in OS, we separately searched for miRNA interactions with SNHG17 and LINC00837. The SNHG17-miRNA pairs were sourced from the ENCORI database. There was no predicted miRNA that could bind to LINC00837. From this, a SNHG17/miRNA/lipid metabolism-related gene network in OS was constructed (Fig. 7D).
To further verify the effect of SNHG17 upregulation in OS on the expression levels of downstream target genes including CSNK2A2, ODC1, MIF and VDAC2. Detection and analysis of 7 tumor tissues and 6 adjacent tissues of patients with osteosarcoma by immunohistochemistry (IHC). IHC staining demonstrated that the expression level of CSNK2A2, MIF and VDAC2 was dramatically upregulated in the tumor tissues as compared with adjacent tissues, but there was no significant difference in the expression level of OCD1 between tumor tissues and adjacent tumor tissues (Fig. 8).
Osteosarcoma is the most common primary bone malignancy in children and adolescents. Although improvements in therapeutic strategies were achieved, the outcome remains poor for most patients with metastatic or recurrent osteosarcoma[25]. Therefore, it is imperative to identify novel and effective prognostic biomarker and therapeutic targets for the disease. Studies have emphasized that both lipid metabolism and lncRNA can be involved in the proliferation, metastasis and drug resistance of tumor cells[7, 26, 27]. Thus, lipid metabolism-related lncRNA abnormalities may be a potential predictor of osteosarcoma metastasis and patient survivorship. In the present study, we identified a potential prognostic two-lipid metabolism-related lncRNAs signature from GEO database, which included SNHG17 and LINC00837.
Small nucleolar RNA host gene 17 (SNHG17) is a novel cancer-related lncRNA of the SNHG family which is highly expressed in various tumors as well as may be in-volved in proliferation, apoptosis, invasion, metastasis, drug resistance and other bio-logical functions of cancer cells[28]. SNHG17 also can regulate some gene expression via binding to relative miRNA as a competitive endogenous RNA (ceRNA)[29]. A large number of studies have also shown that SNHG17 plays an important pathogenic role in a variety of tumor-associated diseases. Such as SNHG17 can upregulated PAX6 to exert its carcinogenic role acted as a ceRNA of miR-375[30], promoting epithelial-mesenchymal transition (EMT) progress, proliferation and invasion of ESCC cells by sponging miR-338-3p, thereby activating SOX4[31], and may regulate H2AX signaling via miR-328-3p in renal cell carcinoma[32]. In summary, numerous evidences indicate that SNHG17 plays an important role in tumor development and has important clinical application value in tumor diagnosis and prognosis. At present, there are few studies on LINC00837, only one study shows that LINC00837 may be positively correlated with resting dendritic cells, but its impact on tumorigenesis, diagnosis and prognosis is still unclear[33].
Therefore, we speculate that the abnormal expression of SNHG17 and LINC00837 may be related to the prognosis and immunosuppressive microenvironment of OS patients. To further evaluate the prognostic value of SNHG17 and LINC00837, we con-structed a prognostic risk model for lipid metabolism-related lncRNAs using univariate and multivariate COX regression analysis in GEO training cohort. And subsequently, we constructed a prognostic nomogram that integrated the risk score based on this model and some significant clinical features including sex, age and metastasis. The results showed that the risk score effectively predicted prognosis in the GEO training cohort and was validated in the GEO internal validation cohort. The signature and nomogram were further validated by Kaplan–Meier survival analysis, calibration plots, receiver operating characteristic (ROC) curves and decision curve analysis (DCA). And the AUC values of 1, 3, 5 years of the nomogram reached 0.935, 0.772, 0.828, respectively, suggesting that the nomogram exhibited superior survival predictive ability. In summarising, several validation methods have proven the robustness of the risk model, and we have confidence that the risk model will be extensively applied to individualised risk management. In an addition, we discovered that the label-based risk score was significantly associated with metastasis, which indicated that labeling was also a better predictor of osteosarcoma metastasis.
To identify pathways associated with the risk model, we performed a GSEA to examine potential pathways and features in both high and low risk groups. KEGG results suggest that lipid metabolism-related lncRNA risk features may be associated with cell proliferation and metabolism-related pathways. When lipid metabolism is dysregulated, it can lead to impaired tumour microenvironment and bone remodelling, resulting in poor prognosis of osteosarcoma[34]. Studies have shown that metabolic re-programming is an important feature of immune cell activation, which can affect their immune function because of different metabolic characteristics[35, 36]. The infiltration level of immune cells in tumors is an important indicator for prognostic judgment and treatment effect evaluation[37]. Given the closeness of metabolic reprogramming to the tumour immune microenvironment, the immune environment between high- and low-risk populations was explored. Subsequently, we then calculated the level of infiltration of 24 immune cells in patients with OS. Spearman's correlation analysis yielded a significant negative correlation between risk score and central memory CD8 T cells, macrophages, natural killer cells and plasmacytoid dendritic cells. It is suggested that the lower the risk score of OS patients, the higher the degree of immune cell infiltration. On these grounds, it is plausible to conclude that risk score and immune status are inter-related with poor prognosis.
To date, few studies have focused on the roles of two lncRNAs, SNHG17 and LINC00837, in the occurrence and progression of osteosarcoma [31]. To further elucidate how the lipid metabolism-related lncRNA (SNHG17 and LINC00837) regulate lipid metabolism-related gene, via sponging miRNAs in OS, we searched for miRNAs interacting with SNHG17 and LINC00837, respectively. Results show that no miRNAs are expected to bind to LINC00837, ceRNA network analysis of SNHG17 showed that SNHG17 could regulate the expression of four genes including CSNK2A2, MIF, ODC1 and VDAC2 via binding to different miRNAs. But the relationship between these genes and the pathogenesis and prognosis of osteosarcoma remains to be further studied. Therefore, in this study, we performed immunohistochemical detection on tumor tis-sues and adjacent tissues of patients with osteosarcoma, and the results showed that CSNK2A2, MIF and VDAC2 genes were up-regulated in tumor tissues of patients with osteosarcoma, but there was no significant difference in the expression of ODC1 gene, which may be due to the large difference in positive signal values between different samples. Previous study showed that CSNK2A2 gene can be used as a prognostic biomarker for hepatocellular carcinoma and may be involved in the pathological process of abdominal aortic aneurysm[38, 39]. MIF is an inflammatory cytokine involved in the carcinogenesis of many cancer types, it has important roles in angiogenesis, im-munity and metastasis in melanoma cell lines[40]. And VDAC2 can induces apoptosis and affects tumor development[41, 42].Taken together, these data and results may demonstrate that SNHG17 may as a competing endogenous RNA (ceRNA) to regulate the expression of lipid metabolism related genes including CSNK2A2, MIF and VDAC2 through binding with hsa-miR-505-3p, hsa-miR-451a, hsa-miR-6839-3p and hsa-miR-370-3p respectively, thereby further regulating the formation and development of osteosarcoma.
Combining the above findings, this study constructed a risk model based on two lipid metabolism-related lncRNA signature, and evaluated the model using multiple validation methods to further validate the robustness and accuracy of the risk model. In particular, we not only investigated the accuracy of the risk model predictions, but also analysed the possible regulatory mechanisms of the two lipid metabolism-related lncRNAs with prognostic value, providing guidance for subsequent studies on the molecular mechanisms of osteosarcoma and targeted therapies.
However, there were certain limitations to our study, this was a retrospective study with a small sample size of osteosarcoma and lacked the support of more experimental evidence. There is therefore a need for further studies to validate the accuracy of the model using independent cohorts combined with experiments such as immunohistochemical analysis or PCR. Notwithstanding these limitations, we are constructed two lipid metabolism-related lncRNA signatures that have good prognostic value in osteosarcoma. They require further research to elucidate their roles in the progression of osteosarcoma.
In conclusiveness, in the current study, our data clearly indicated that the lipid metabolism-related lncRNA signature can predict the prognosis of osteosarcoma in patients. The signature provides accurate prediction and thus facilitate clinical diagnosis and treatment.
Ethics approval and consent to participate
This study was conducted in accordance with the guidelines of the Declaration of Helsinki and was approved by the Ethics Committee of the 920th Hospital of the Joint Logistic Support Department of the People's Liberation Army (No. 2022-039-02). All patients signed written informed consent prior to enrollment.
Consent for publication
Not applicable.
Availability of data and materials
The datasets analysed in this study are all available in public databases. The p-ublic databases are all open access, which are available in the Gene Expressio-n Omnibus GSE12865(https://www.omicsdi.org/dataset/geo/GSE12865) and GSE16091(https://www.omicsdi.org/dataset/geo/GSE16091), the Cancer Genome Atlas ,TCGA-OS(https://portal.gdc.cancer.gov/projects/TARGET-OS), the Therapeutically Applicable Research To Generate Effective Treatments, TARGET(https://target-data.nci.nih.gov/Publ-ic/OS/clinical/).
Competing interests
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential competing interest.
Funding
This study was supported by the Technical Innovation Talents Training Object Project, China (Grant No. 202005AD160146); and Grants from Yunnan Orthopedics and Sports Rehabilitation Clinical Medicine Research Center (Grant No. 202102AA310068).
Author Contributions
CL conceived the original ideas of this manuscript and reviewed the finished manuscript and executed supervision throughout the process. LS, MG, BQ, LP, HS and JR executed the data collection and data analysis. ZT and HF prepared the manuscript, tables, and Figs. All authors have read and approved the manuscript.
Acknowledgments
We like to acknowledge the TARGET and the GEO (GSE12865 and GSE16091) network for providing data.
Author details
1 Clinical Medical College of Dali University, Dali, China. 2 Department of Orthopaedic, The First People’s Hospital of Dali City, Dali, China. 3 Department of Orthopaedic, Kunming Medical University, Kunming, China. 4 Department of Orthopedics, The 920th Hospital of Joint Logistics Support Force of Chinese People’s Liberation Army, Kunming, China. 5 Department of Orthopaedic, Yunnan University of Chinese Medicine, Kunming, China.