Wnt Signaling Pathway Related Gene CTNNBIP1 Act as A potential Prognostic Biomarker and A Comprehensive Prognostic Analysis in Osteosarcoma

doi:10.21203/rs.3.rs-2356189/v2

Download PDF

Research Article

Wnt Signaling Pathway Related Gene CTNNBIP1 Act as A potential Prognostic Biomarker and A Comprehensive Prognostic Analysis in Osteosarcoma

https://doi.org/10.21203/rs.3.rs-2356189/v2

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Purpose

Osteosarcoma is the most common primary bone malignancy in children and young adults with poor outcomes in patients with recurrence and metastasis. The role of Wnt signaling pathway in osteosarcoma was complex and not well specified.

Methods

KEGG Wnt signaling pathway related genes were screened by univariate Cox regression analysis, Boruta feature selection and KM analysis in TARGET-OS cohort for the gene with the highest prognostic value in osteosarcoma. Clinical and immune cell infiltration features were also screened by univariate Cox regression, respectively. A prognostic model was established by the indicators screened above and evaluated in training and validation sets. Then, functional analysis of the differentially expressed genes were performed.

Results

CTNNBIP1 was screened as the KEGG Wnt signaling pathway related gene with the highest prognostic value in osteosarcoma. Metastasis, tumor site and M2 macrophage infiltration in tumor tissue were also identified as prognostic indicators. The model established by these indicators was evaluated in training set and successfully validated in validation set. C-index (0.812 in training set and 0.787 in validation set) and Calibration analysis suggested that the model had high prediction accuracy both in training and validation set. AUCs of the ROC analysis for 1-,3-,5-year prediction of the model were 0.96, 0.81 and 0.81 in training set, 0.80, 0.83 and 0.89 in validation set. Decision curve analyses also showed that patients could get high net benefit from the model. Functional analyses of the differential expressed genes between normal and osteosarcoma patients indicated that Wnt signaling pathway and cell cycle were upregulated and focal adhesion was downregulated in osteosarcoma patients. The overexpression of CTNNBIP1 in osteosarcoma tissues was validated by immunohistochemistry.

Conclusions

CTNNBIP1 was a potential prognostic gene and therapeutic target in osteosarcoma. Prognostic model established by CTNNBIP1, clinical and immune cell infiltration indicators had good prediction accuracy and discriminatory capacity.

CTNNBIP1

osteosarcoma

prognostic model

Wnt signaling pathway

immune cell infiltration

Osteosarcoma is the most common primary bone malignancy in children and young adults with an average incidence of 3.3 per million within different age, sex and races(1–3). While in the 10–24 years group, the average incidence was 7.2 per million among different sex and races, more than 2 times of the average incidence in all ages(1–3) and the incidence was even higher among males of Hispanic ethnicity(1). Obviously, osteosarcoma is a serious threat to children and young adults, ranked the eighth highest childhood cancers at around 2.4% of tumors diagnosed(4, 5). Currently, surgery and chemotherapy are still the main treatment methods for osteosarcoma(6). The standard therapy for osteosarcoma has been used for about four decades and it greatly improved the long-term survival rate of osteosarcoma patients since 1980s. The average 5-year relative survival rate of primary osteosarcoma is 59.9% in all age and races and 65.9% in 10–24 years group by SEER database 2000–2017(1). However, there has been no significant breakthrough in the treatment of osteosarcoma since the standard therapy(6). Moreover, how to improve the survival rate of patients with recurrent and metastatic osteosarcoma still remains problematic, whose relative 5-year survival rate is as low as about 20%(1, 7, 8). Therefore, it’s imperative to develop new therapies for osteosarcoma and finding new therapeutic targets is one of the possible solutions to this problem.

Recent studies have suggested that Wnt signaling pathway is involved in a wide variety of disorders including cancer, sarcoma, coronary artery disease and etc. Interestingly, its role in sarcomas is controversial(9, 10). On the one hand, it can act as a oncogenic factor by facilitating cell proliferation and movement. On the other hand, it also can act as a tumor suppressor by promoting cell differentiation and maturation. Then, what the exact role it plays in osteosarcoma still needs to be further studied. Therefore, the prognostic value of genes involved in KEGG Wnt signaling pathway, which are also called KEGG Wnt signaling pathway related genes, are going to be investigated in this study. Genes, which have an impact on the prognosis of osteosarcoma patients also may be potential therapeutic targets in osteosarcoma. Besides, clinical and immune cell infiltration features are also intended to be explored in order to reinforce the stability and accuracy of the prognostic model.

Although several different prognostic models of osteosarcoma have been established by other researchers, but no model has been established by Wnt signaling pathway related genes and their prognostic value hasn’t been explored either. Moreover, the role of Wnt signaling pathway in osteosarcoma still needs to be further studied.

2.1 Data collection

TARGET-OS data were downloaded from UCSC Xena website (https://xenabrowser.net/hub/) on June 2, 2022. RNA-seq data from 88 different samples, together with their corresponding phenotype and survival data were extracted. 4 samples without complete survival data were excluded and 84 remained. Clinical features, available in at least two-thirds of the samples, were considered to be included for further analysis. Then, age, gender, race, ethnicity, primary site, metastasis data of the samples were collected. GTEX normal tissue RNA-seq data from 379 different subjects were also downloaded as a normal control from UCSC Xena website. Counts data of the TARGET-OS cohort were also converted to Transcripts Per Kilobase of exon model per Million mapped reads (TPM) according to the formula: , where q_i referred to reads mapped to transcript, l_i referred to the transcript length, and referred to the sum of mapped reads to transcript normalized by transcript length.

Meanwhile, RNA expression data and corresponding clinical and survival data of 53 osteosarcoma patients from GSE21257 cohort were downloaded from GEO website (https://www.ncbi.nlm.nih.gov/geo/) on June 2, 2022. The RNA expression data of this cohort were identified by Illumina human-6 v2.0 expression beadchip (using nuIDs as identifier) using tumor tissues from biopsy of osteosarcoma patients, submitted by Centre for Molecular Medicine Norway Apr 08, 2010. The expression data downloaded had already been transformed using the variance stabilizing transformation algorithm and subsequently normalized using robust spline normalization, so here we didn’t need to perform normalization again.

151 KEGG Wnt signaling pathway genes were downloaded from the Gene Set Enrichment Analysis website (http://www.gsea-msigdb.org/gsea/index.jsp) on June 2, 2022.

2.2 Identify gene with the highest prognostic value in Wnt signaling pathway

The prognostic value of the 151 genes in Wnt signaling pathway were preliminarily explored by univariate Cox regression analysis. Genes with P value < 0.01 were selected as candidates for feature selecting by “Boruta” package(11) and KM analysis with samples divided into low and high expression group by median expression of the corresponding gene. Gene considered to be with the highest prognostic value by the analyses above was selected as an indicator for a prognostic model.

2.3 Identify clinical and immune cell infiltration features with prognostic value

The selected clinical features were screened by univariate Cox regression analysis in order to find indicators with prognostic value. Then, indicators with P value < 0.05 in univariate Cox regression analysis were screened by multivariate Cox regression analysis and indicators with P value < 0.05 were selected for the prognostic model.

“xCell” package was adopted to calculate the immune score of 64 immune cells for each sample and univariate Cox regression analysis was adopted to screen the immune cell infiltration features with prognostic value(12). Indicators with P value < 0.05 in training set and could also be validated in validation set were selected for the prognostic model.

2.4 Evaluate the prognostic model in training set.

A prognostic model was established by the indicators screened above and visualized by a nomogram. The prediction accuracy and discriminatory capacity of the model was first evaluated by C-index, calibration analysis, time-dependent ROC analysis and decision curve analysis (DCA) in training set.

2.5 Validate the model in an independent external set.

GSE21257 was set as an independent external validation set. Firstly, the prognostic value of the hubgene was validated by KM analysis in this cohort and the survival characteristics of samples in high and low expression group were also explored. Then, the same methods were adopted to evaluate the prediction accuracy and discriminatory capacity of the model in validation set.

2.6 Identification and functional analysis of the differentially expressed genes between normal and tumor groups.

Differentially expressed genes (DEGs) between normal (GETX cohort) and tumor groups (TARGET-OS cohort) were identified by “DESeq2” package. Then, GO and KEGG clustering, gene set enrichment analysis (GSEA) were performed to explore their functional enrichment.

2.7 Identify hubgene co-expressed genes and perform functional analysis.

Correlation analysis of the expressions between hubgene and other genes was performed in TARGET-OS cohort. Genes with P value < 0.05 were considered to be hubgene correlated genes, which were positively or negatively co-expressed with the hubgene. Then, GO and KEGG clustering were adopted to explore the functional enrichment of these genes. Meanwhile, protein protein interact (PPI) network were performed on STRING website with interaction score ≥ 0.4 (https://cn.string-db.org/) to investigate the interaction among the proteins encoded by hubgene and hubgene co-expressed genes.

2.8 Validate the expression of CTNNBIP1 in tumor tissues by immunohistochemistry.

Paraffin-embedded tissues were obtained from 2 patients with osteosarcoma who underwent surgical resection at the Department of Orthopedics, Jiangxi Provincial People's Hospital. Immunohistochemistry was adopted to explore the expression of CTNNBIP1 in osteosarcoma tumor and paracancer tissues. The CTNNBIP1 antibody used for immunohistochemical staining was obtained from Bioss Inc (http://www.bioss.com.cn/, catalog number: bs-4095R). Both the osteosarcoma tissue section and the normal tissue section adjacent to the cancer were treated by performing a series of steps such as dewaxing, antigen retrieval by microwave, blocking non-specific antigens, primary antibody incubation, secondary antibody incubation (http://www.zsbio.com/, catalog number: ZB-2301), DAB staining (CWBIO, China, catalog number: CW0125M) and Hematoxylin restaining. Finally, the stained osteosarcoma tissue section and the paracancer tissue section were observed under a microscope.

2.9 Statistical analysis

In this study, R software v3.63 was used to process data and generate charts, Cytoscape software v3.7.1 was used to visualize the PPI network on STRING website with interaction score ≥ 0.4 (https://cn.string-db.org/). Flexible statistical methods were adopted for the statistical analysis.

3.1 Clinical characteristics of different cohorts

Clinical characteristics of TARGET and GSE21257 cohorts were shown in Table 1. Because age and overall survival time (OS.time) were not always normally distributed in the two sets, so Kruskal test was adopted in the comparison between the two sets. Results showed that there was no significant difference between the two sets in gender, site, metastasis, OS.time and OS status. In GSE21257, the median of age was elder than that in TARGET cohort, with P value = 0.048. The two cohorts were considered to be similar and comparable. Thus, we considered GSE21257 to be a suitable validation set.

Table 1

Clinical characteristics of TARGET-OS and GSE21257 cohorts
Characteristics	Overall	TARGET	GSE21257	p	test
Sample size	137	84	53
Age (median)	15.10 [12.35, 18.08]	14.35 [12.20, 17.23]	16.67 [13.67, 18.58]	0.048	kruskal.test
Gender
Female	56 (40.9)	37 (44.0)	19 (35.8)	0.440	chisq.test
Male	81 (59.1)	47 (56.0)	34 (64.2)	0.440	chisq.test
Site
Axial skeleton	4 (2.9)	4 (4.8)	0 (0.0)	0.372	chisq.test
Femur	65 (47.4)	38 (45.2)	27 (50.9)
Tibia	36 (26.3)	21 (25.0)	15 (28.3)
others	32 (23.4)	21 (25.0)	11 (20.8)
Metastasis
Absent	102 (74.5)	63 (75.0)	39 (73.6)	1.000	chisq.test
Present	35 (25.5)	21 (25.0)	14 (26.4)	1.000	chisq.test
OS.time (median)	46.00 [22.77, 77.00]	48.65 [20.63, 69.15]	45.00 [27.00, 94.00]	0.122	kruskal.test
OS status
Alive	87 (63.5)	57 (67.9)	30 (56.6)	0.250	chisq.test
Dead	50 (36.5)	27 (32.1)	23 (43.4)	0.250	chisq.test

3.2 Results of the prognostic gene identification

14 genes were screened by univariate Cox regression analysis with prognostic value (P value < 0.05). 11 of them with hazard ratio (HR) > 1 and 3 with HR ≤ 1. The prognostic value of genes with HR close to 1 were limited, such as MYC. While with HR value far from 1 were more suitable as prognostic genes. The minimum and maximum HR values were 0.94 for CTNNBIP1 and 1.1 for FZD3, respectively. Here, CTNNBIP1 with the smallest HR value 0.94 was considered to be one of the best candidates for a prognostic model. Finally, 9 genes with P value < 0.01 (DKK1, MYC, WNT16, PLCB4, PPP3CC, PRKX, CTNNBIP1, WNT10B, WNT11) were select as candidates for the next screening (Fig. 1).

Result of Boruta feature selection showed that CTNNBIP1, PRKX, DKK1, PPP3CC and MYC were genes confirmed according the importance of the genes (Fig. 2A-B) and CTNNBIP1 got the highest importance score. KM analysis of the 9 genes showed that CTNNBIP1, PLCB4, MYC, PRKX were genes with P value < 0.05 between their high and low expression groups respectively and CTNNBIP1 got the smallest P value 0.0024 (Fig. 2C-F). Therefore, taking the above results together, we considered CTNNBIP1 to be a hubgene with the highest prognostic value in Wnt signaling pathway of osteosarcoma and it was selected as an indicator for the prognostic model.

3.3 Identify clinical and immune cell infiltration indicators for the prognostic model

Clinical features were screened by univariate and multivariate Cox regression analysis in TARGET-OS cohort and the results were shown in Table 2. Metastasis and tumor site were considered to be indicators with prognostic value (P < 0.05).

Table 2

Univariate and multivariate Cox regression analysis of the clinical features
Characteristics	Univariate Cox regression			Multivariate Cox regression
Characteristics	Hazard.Ratio (95%CI)	P.Value		Hazard.Ratio (95%CI)	P.Value
Age	1 (0.92–1.09)	0.976		#N/A	#N/A
Ethnicity	0.38 (0.13–1.07)	0.068		#N/A	#N/A
Gender	0.71 (0.34–1.52)	0.382		#N/A	#N/A
Metastasis	4.76 (2.22–10.22)	0		4.43 (2.01–9.75)	0
Race	1.31 (0.38–4.56)	0.672		#N/A	#N/A
Region	0.34 (0.11–1.04)	0.059		#N/A	#N/A
Site (control: axial skeleton)
Tibia	0.1 (0.02–0.51)		0.006	0.1 (0.02–0.52)	0.006
Femur	0.34 (0.1–1.19)		0.092	0.25 (0.07–0.92)	0.038
Others	0.19 (0.04–0.78)		0.022	0.17 (0.04–0.76)	0.02
Surgical strategy	2.34 (0.3-18.02)		0.413	#N/A	#N/A

Immune scores of 64 immune cells for patients in TARGET-OS and GSE21257 cohorts were calculated by “xCell” package. Univariate Cox regression analysis in TARGET-OS cohorts indicated that Plasma cells, MPP, Macrophages, Macrophages M2, Sebocytes, CD4 T cells, Macrophages M1, Keratinocytes were candidate indicators with P value < 0.05 (additional file 1). Then, GSE21257 cohort was used to validate the prognostic value of these cells, result showed that only Macrophages M2 was successfully validated in GSE21257 with P value = 0.018 in univariate Cox regression analysis (additional file 2).

3.4 Establishment and evaluation of the prognostic model in training set.

The survival characteristics of patients in high and low CTNNBIP1 expression groups in TARGET-OS cohort were shown in Fig. 3A. Patients were divided into high and low expression group by median of CTNNBIP1 and more alive patients with longer survival time were found in the high expression group (more green dots in the upper left of the plot). Then, the prognostic model, established by CTNNBIP1, metastasis, site and M2 macrophages infiltration was visualized by a nomogram (Fig. 3B). By summing up the points of each indicator in the nomogram to predict the survival probability of an osteosarcoma patient, we could see that the expression of CTNNBIP1 and M2 macrophages infiltration acted as protective factors, while present metastasis and tumor site at axial skeleton were risk factors for survival. C-index of the model in training set was 0.812 (95%CI: 0.768–0.856). Meanwhile, calibration analysis showed that the predict overall survival rates of 1-, 3- ,5-year were in high agreement with the observed overall survival rates (the solid color line were close to gray dashed ideal line) (Fig. 3C). Results of C-index and calibration analysis suggested high prediction accuracy of the model in training set. AUCs of Time dependent ROC analysis for 1-, 3- and 5-year prediction of the model were 0.96, 0.81 and 0.81, respectively (Fig. 3D). Decision curve analysis showed that patients could got higher net benefit from the model (nomogram) than other strategies (based on 5-year prediction) (Fig. 3E). Besides, although lower than the nomogram strategy, CTNNBIP1 alone was also a good strategy with higher net benefit compared with treat all and treat none strategies.

3.5 Validation of the model in GSE21257

KM analysis showed that there was a significant difference in the prognosis of osteosarcoma patients between high and low CTNNBIP1 expression groups in GSE21257 with P value = 0.0063 (Fig. 4A). The survival characteristics of patients in high and low CTNNBIP1 expression groups in GSE21257 cohort showed that there were more alive patients with longer survival time in high expression group (more green dots in the upper left of the plot) and this result was consistent with that in training set (Fig. 4B). Nomogram in GSE21257 showed that the expression of CTNNBIP1 and M2 macrophages infiltration acted as protective factors and present metastasis acted as a risk factor for survival, the same with that in training set (Fig. 4C). While there were also some minor differences in the contribution of different tumor sites to the survival rate between the two sets, which probably due to the small sample size and the absence of patients with tumor site at axial skeleton in validation set. C-index of the model in training set is 0.787 (95%CI: 0.739–0.835), a little lower than that in training set. Calibration analysis indicated that the predict overall survival rate of 1-, 3- ,5-year were in high agreement with the observed overall survival rate (the solid color line were close to gray dashed ideal line) (Fig. 4D). Though C-index in validation set was a litter lower than it in training set, but the results of C-index and calibration analysis still suggested high prediction accuracy of the model in validation set. AUCs of Time dependent ROC analysis for 1-, 3- and 5-year prediction of the model were 0.8, 0.83 and 0.89, respectively (Fig. 4E). Decision curve analysis showed that compared with treat all and treat none strategies, patients could get higher net benefit from the model (nomogram) and CTNNBIP1 strategy (based on 5-year prediction) (Fig. 4F). Meanwhile, when the threshold > 20%, the nomogram strategy was better.

3.6 Identification and functional analysis of the differentially expressed genes between normal and tumor groups.

There were 10759 differentially expressed genes between GTEX normal and TARGET-OS tumor groups with adjusted P value < 0.05 and |log₂fold change| > 1 by “DESeq2” package. The names of genes with adjusted P value < 0.0001 and |log₂fold change| > 8.5 were shown in the volcano plot (Fig. 5A). Biological processes (BP) of GO clustering of the DEGs showed that they were top clustered in mitochondrial gene expression, mitochondrial translation, translational elongation and termination. The top clustered cellular components (CC) by GO clustering were mitochondrial matrix, mitochondrial inner membrane, mitochondrial protein complex, contractile fiber and the top clustered molecular functions (MF) were actin binding, coenzyme binding, electron transfer activity, oxidoreductase activity acting on NAD(P)H. KEGG clustering indicated that besides specific diseases, the DEGs were top clustered in pathways like p53 signaling pathway, thermogenesis, oxidative phosphorylation, focal adhesion, cell cycle and etc. (Fig. 5C, additional file 3). GSEA showed that calcium signaling pathway, JAK-STAT signaling pathway, NF-kappa B signaling pathway, PPAR signaling pathway were top enriched pathways (additional file 4). Besides, we found that Wnt signaling pathway, cell cycle, oxidative phosphorylation and citrate cycle (TCA cycle) were up regulated and calcium signaling pathway, JAK-STAT signaling pathway, NF-kappa B signaling pathway, PPAR signaling pathway and focal adhesion were down regulated in osteosarcoma patients (Fig. 5D).

3.7 Identification and functional analysis of the CTNNBIP1 co-expressed genes.

2740 CTNNBIP1 co-expressed genes in osteosarcoma were identified with P value < 0.05 by Pearson correlation analysis. GO biological processes clustering of these genes showed that they were top clustered in regulation of cell morphogenesis, regulation of actin filament-based process, cell-substrate adhesion, cell junction organization. While focal adhesion, cell-substrate adherens junction, cell-substrate junction were top 3 clustered CC and actin binding, cell adhesion molecule binding, small GTPase binding, Ras GTPase binding were top clustered MF (Fig. 6A). As to KEGG clustering, CTNNBIP1 co-expressed genes were mainly clustered in focal adhesion, adherens junction, regulation of actin cytoskeleton, endocytosis, proteoglycans in cancer and etc. (Fig. 6B). PPI network of the proteins encoded by CTNNBIP1 and the top 100 CTNNBIP1 co-expressed genes ranked by P value indicated that CLSTN1 might interact with CTNNBIP1 (Fig. 6C) and FLNA and MMP11 were hubgenes that might be directly or indirectly associated with CTNNBIP1. Proteins which might interact with CTNNBIP1 were shown on the right side of Fig. 6C and they were mainly involved in the Wnt signaling pathway.

3.8 CTNNBIP1 expression in osteosarcoma tissues by immunohistochemistry.

CTNNBIP1 was stained brown in immunohistochemistry. In patient 1, the cytoplasm of tumor cells was stained brown, whereas nontumor cells, such as vascular endothelial cells, were not stained (Fig. 7A-B). Meanwhile, in the paracancer tissues, cells were not stained brown as well (Fig. 7C-D). The same result was observed in patient 2, the difference was that the paracancer tissue in patient 2 were normal bone tissue (Fig. 7E-H). These results strongly support the overexpression of CTNNBIP1 in osteosarcoma cells.

Currently, more and more biological heterogeneity and molecular aberrations in osteosarcoma have been identified and advances in this area are reported to be of great promise for new treatments of osteosarcoma(13–15). The Wnt signaling pathway is one of the signaling pathways that can affect the proliferation, invasion and metastasis of many tumor cells(16–19). More studies report that Wnt signaling pathway plays a pro-oncogenic role and is hyperactivated in osteosarcoma(20, 21).

In this study, univariate Cox regression analysis was adopted to screen the prognostic genes in KEGG Wnt signaling pathway. A stringent strategy with P value < 0.01 was accepted in order to reduce the false positive rate and 9 genes were selected to be directly correlated with the survival of the osteosarcoma patients. CTNNBIP1 with the smallest HR value 0.94 was considered to be one of the best candidates for the prognostic model. By Boruta feature selection, the importance score of the 9 genes in the prognosis of osteosarcoma were calculated and CTNNBIP1 was considered to be the most important gene with the highest importance score. In KM analysis, the outcomes of patients in the high expression group of CTNNBIP1 was significantly better than that in the low expression group (divided by median expression of CTNNBIP1) with the lowest P value 0.0024. Besides, immunohistochemical staining of tumor tissues indicated that CTNNBIP1 was overexpressed in osteosarcoma cells (Fig. 7A-H). Therefore, CTNNBIP1 was considered to be the Wnt signaling pathway related gene with the highest prognostic value in osteosarcoma and it was first reported to be a potential prognostic gene in osteosarcoma by us.

Clinical and immune cell infiltration features were also screened by univariate Cox regression analysis, respectively. Metastasis, tumor site and M2 macrophage infiltration were indicators screened with P value < 0.05. These indicators had been reported in osteosarcoma by others researchers. First of all, metastasis was a well-known indicator that would affect the prognosis of osteosarcoma(22, 23). The low long-term survival rate of patients with metastasis still remained an issue that needed to be addressed. Secondly, the primary site of osteosarcoma was also reported to be associated with the survival of osteosarcoma(24, 25). Patient with the tumor site located at the axial skeleton was said to be with the worst prognosis. Fortunately, two-thirds of the tumors emerged around the knee joint in the distal femur, followed by the proximal tibia as the second most frequent site(24). Therefore, in this study, tumor sites were divided into four categories: axial skeleton, femur, tibia and other site. Thirdly, analyses of the immune cell infiltration of osteosarcoma indicated an infiltration of tumor-associated macrophages (TAMs)(26–28). Many researchers declared that M2 macrophages could enhance the stemness and invasiveness of tumor cells and promote the metastasis of osteosarcoma in vitro and in animals and they speculated that M2 macrophages might associated with the poor prognosis in humans(27, 29). However, studies in humans showed that patients with up-regulated M2 macrophages had a better prognosis(30). In this study, our result was consistent with the latter one (HR < 1) and we considered there were several reasons for this. For one thing, conclusions reached by the former studies were mostly based on cell and animal experiments, ignoring the human microenvironment. For another thing, the M2 macrophages score in this study was estimated by algorithms based on the RNA expression profile, which probably could not fully reflect the actual quantity of M2 macrophages. Lastly, more reliable markers were needed to identify different subtypes of macrophages and their functions were also needed to be further specified. Thus, more studies were still needed to further elucidate this issue. All the indicators of the prognostic model were well reflected in the nomogram and the survival rate could be predicted by them (Fig. 3B, 4C).

The model was evaluated in the training set with high prediction accuracy by C index and calibration analysis (Fig. 3C). Time-dependent ROC suggested the model had excellent prediction sensitive and specificity with 1-, 3-, 5-year prediction AUC values: 0.96, 0.81, 0.81, respectively (Fig. 3D). DCA indicated that patients could get more net benefit from the model, which suggested a clinical application value of the model (Fig. 3E). Meanwhile, our model was also successfully validated in an independent external cohort. The nomogram in validation set was similar to that in training set. Minor differences were seen in the tumor site between the two sets, which was mainly due to the small sample size and the lack of patients with axial osteosarcoma in the validation set (Fig. 4C). KM analysis in validation set also suggested a significant difference between high and low expression group of CTNNBIP1. Though C index in validation set was slightly lower than that in training set, but it was also close to 0.8, which still suggested a splendid prediction accuracy of the model. The high prediction accuracy of the model also could be reflected by the calibration analysis in Fig. 4D. Time-dependent ROC in validation set also suggested excellent prediction sensitive and specificity with 1-, 3-, 5-year prediction AUC values: 0.80, 0.83, 0.89, respectively (Fig. 4E). The AUC values were no less than 0.8 both in training and validation set, indicating that the model was stable and with outstanding prediction accuracy and discriminatory capacity. DCA in validation set also suggested a clinical application value of the model (Fig. 4F). One slight difference was that CTNNBIP1 alone could be used as an alternation strategy for clinical decision.

Functional analysis of the DEGs between normal and osteosarcoma patients showed that P53 signaling pathway was one of the top clustered pathways, which was down regulated in osteosarcoma patients with NES = -1.19. The P53 signaling pathway was a well-known tumor suppressor signaling pathway which could lead to cell cycle arrest, cellular senescence or apoptosis and it’s down regulated in osteosarcoma(31, 32). Meanwhile, we found that Wnt signaling pathway, cell cycle, oxidative phosphorylation and citrate cycle (TCA cycle) were up regulated and calcium signaling pathway, focal adhesion were down regulated in osteosarcoma patients (Fig. 5D). It’s known that there were 3 main pathways in Wnt signaling: the canonical pathway, planner cell polarity pathway and Wnt/Ca²⁺ pathway. The canonical pathway was reported to be associated with tumor progression and metastasis and it’s the Wnt signaling pathway we normally discussed. The planner cell polarity pathway was mainly associated with focal adhesion and it negatively interacted with the canonical pathway by PAR-1 and NKD (33). Wnt/Ca²⁺ pathway might had something to do with calcium signaling pathway(33). Here, calcium signaling pathway and focal adhesion were down regulated and Wnt signaling pathway was up regulated. Therefore, we speculated that the canonical Wnt signaling pathway was up regulated and the non-canonical Wnt signaling pathway might be down regulated in osteosarcoma. Previous studies in the literature also demonstrated that canonical Wnt signaling pathway was activated in osteosarcoma(20, 34).

Besides, many studies had revealed that Wnt signaling pathway could promote tumor progression and metastasis by regulating cell cycle, epithelial-mesenchymal transition (EMT), acquisition of stem cells properties of cancer stem cells (CSCs), extracellular matrix modulation, cell adhesion, angiogenesis and immune tolerance(9, 10, 35–37). The up-regulation of oxidative phosphorylation and citrate cycle (TCA cycle) suggested an increase in energy metabolism in osteosarcoma cells(38) and this was also consistent with the results of GO clustering. CTNNBIP1 co-expressed genes were mainly clustered in focal adhesion, adherens junction, regulation of actin cytoskeleton and this suggested that CTNNBIP1 might also affect the prognosis of osteosarcoma patients by crosstalk with these pathways. In the PPI network, CLSTN1 was a member of the calsyntenin family, FLNA was a protein involved in focal adhesion and MMP11 was a matrix metalloproteinase. All the related pathways and genes above could affect the migration ability of osteosarcoma cells and thus affect the metastasis of osteosarcoma.

In general, we found CTNNBIP1 was up-regulated in osteosarcoma patients and high expression of CTNNBIP1 was associated with a good prognosis. Wnt signaling pathway was also found to be activated in osteosarcoma patients, as well as cell cycle, energy metabolism and focal adhesion. CTNNBIP1 might affect the prognosis of osteosarcoma patients by Wnt signaling, cell cycle, cell adhesion and extracellular matrix modulation.

As to CTNNBIP1, it (Catenin Beta Interacting Protein 1) encoded a protein, which could bind β-catenin and prevent the interaction between β-catenin and TCF family members. It’s a negative regulator of the canonical (β-catenin dependent) Wnt signaling pathway(39, 40). Moreover, CTNNBIP1 might have consequences for cadherin-based adhesive function in certain circumstances(41). To date, many studies had demonstrated that the canonical Wnt signaling pathway acted as a pro-oncogenic role in osteosarcoma(20, 21), while CTNNBIP1 was a negative regulator of it. Therefore, high expression of CTNNBIP1 could suppress the activation of the canonical Wnt signaling pathway and thus result in activation of cell cycle and inactivation of cell adhesion, which would contribute to the progression and metastasis of osteosarcoma. However, the fact was that Wnt signaling pathway was activated in osteosarcoma patients with high expression of CTNNBIP1. A possible explanation for this was that CTNNBIP1 was just one of the many factors that could regulate the Wnt/β-catenin signaling and the expression alteration of CTNNBIP1 might be secondary to the changes of Wnt signaling pathway(40, 42). Besides, CTNNBIP1 was also found to be associated with other tumors through Wnt signaling, such as hepatocellular carcinoma(43), papillary thyroid cancer(44), lung cancer(45), gastric adenocarcinoma(46), colorectal cancer(47), cervical cancer(48), and etc. While the prognostic value of CTNNBIP1 in osteosarcoma was first reported by us and we found that CTNNBIP1 probably could affect cell adhesion and extracellular matrix modulation in Wnt signaling dependent and / or independent manners in osteosarcoma, which still needed further research.

To sum up, CTNNBIP1 was the Wnt signaling pathway related gene with the highest prognostic value in osteosarcoma. Metastasis, tumor site and M2 macrophage infiltration were also found to be with prognostic value in osteosarcoma. The prognostic model established by them was successfully validated in an independent external cohort, the prediction accuracy and discriminatory capacity of the model were excellent both in training and validation set. Functional analyses suggested that CTNNBIP1 might affect the prognosis of osteosarcoma by Wnt signaling and other pathways, like cell cycle, cell adhesion and etc.

In the end, some limitations of this study should be noticed. For one thing, our results were mainly based on bioinformatic analysis. The detailed mechanism still needed further validation by experiments. For another thing, more data were needed to further validate our model, large sample size studies were welcome.

CTNNBIP1 was a potential prognostic gene in osteosarcoma. Prognostic model established by CTNNBIP1, clinical and immune cell infiltration indicators had good prediction accuracy and discriminatory capacity. CTNNBIP1 could be used as a potential therapeutic target for osteosarcoma.

Biological process

Cellular component

CSCs

Cancer stem cells

Confidence interval

DCA

Decision curve analysis

DEGs

Differentially expressed genes

EMT

Epithelial-mesenchymal transition

GEO

Gene expression omnibus database

Gene ontology

Hazard ratio

KEGG

Kyoto Encyclopedia of Genes and Genomes

Kaplan-Meier

Molecular function

Overall survival

PPI

Protein protein interact

TAMs

Tumor-associated macrophages

TPM

Transcript per million

Ethics approval and consent to participate

This work was approved by the Ethics Committee of Jiangxi Provincial People's Hospital (NO. 2022-059) and all patients / legal guardians of patients signed consent for the use of their specimens and related information.

Consent for publication

Not applicable.

Availability of data and materials

All the original data were downloaded from the public databases on June 2, 2022. TARGET-OS and GTEX data were downloaded from UCSC Xena website (https://xenabrowser.net/hub/). GSE21275 dataset was downloaded from GEO website (https://www.ncbi.nlm.nih.gov/geo/). 151 KEGG Wnt signaling pathway genes were downloaded from the Gene Set Enrichment Analysis website (http://www.gsea-msigdb.org/gsea/index.jsp).

Competing interests

The authors declare that they have no competing interests.

Funding

There was no funding support for this study.

Authors' contributions

Jian WEN and Lijia WAN conceived the study. Jian WEN performed bioinformatics analyses by R and immunohistochemistry. Xieping Dong reviewed the paper. All authors contributed to the article and approved the submitted version.

Acknowledgements

Not applicable.

Cole S, Gianferante DM, Zhu B, Mirabello L. Osteosarcoma: A Surveillance, Epidemiology, and End Results program-based analysis from 1975 to 2017. Cancer. 2022;128(11):2107-18.
Mirabello L, Troisi RJ, Savage SA. Osteosarcoma incidence and survival rates from 1973 to 2004: data from the Surveillance, Epidemiology, and End Results Program. Cancer. 2009;115(7):1531-43.
Damron TA, Ward WG, Stewart A. Osteosarcoma, chondrosarcoma, and Ewing's sarcoma: National Cancer Data Base Report. Clin Orthop Relat Res. 2007;459:40-7.
Siegel RL, Miller KD, Fuchs HE, Jemal A. Cancer statistics, 2022. CA Cancer J Clin. 2022;72(1):7-33.
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.
Gill J, Gorlick R. Advancing therapy for osteosarcoma. Nat Rev Clin Oncol. 2021;18(10):609-24.
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.
Mialou V, Philip T, Kalifa C, Perol D, Gentet JC, Marec-Berard P, et al. Metastatic osteosarcoma at diagnosis: prognostic factors and long-term outcome--the French pediatric experience. Cancer. 2005;104(5):1100-9.
Singla A, Wang J, Yang R, Geller DS, Loeb DM, Hoang BH. Wnt Signaling in Osteosarcoma. Adv Exp Med Biol. 2020;1258:125-39.
Danieau G, Morice S, Rédini F, Verrecchia F, Royer BB. New Insights about the Wnt/β-Catenin Signaling Pathway in Primary Bone Tumors and Their Microenvironment: A Promising Target to Develop Therapeutic Strategies? Int J Mol Sci. 2019;20(15).
Kursa MB, Rudnicki WR. Feature Selection with the Boruta Package. Journal of Statistical Software. 2010;36(11):1 - 13.
Aran D, Hu Z, Butte AJ. xCell: digitally portraying the tissue cellular heterogeneity landscape. Genome Biol. 2017;18(1):220.
Yu F, Yu C, Li F, Zuo Y, Wang Y, Yao L, et al. Wnt/β-catenin signaling in cancers and targeted therapies. Signal Transduct Target Ther. 2021;6(1):307.
Liu J, Xiao Q, Xiao J, Niu C, Li Y, Zhang X, et al. Wnt/β-catenin signalling: function, biological mechanisms, and therapeutic opportunities. Signal Transduct Target Ther. 2022;7(1):3.
Shoaib Z, Fan TM, Irudayaraj JMK. Osteosarcoma mechanobiology and therapeutic targets. Br J Pharmacol. 2022;179(2):201-17.
Parsons MJ, Tammela T, Dow LE. WNT as a Driver and Dependency in Cancer. Cancer Discov. 2021;11(10):2413-29.
Galluzzi L, Spranger S, Fuchs E, López-Soto A. WNT Signaling in Cancer Immunosurveillance. Trends Cell Biol. 2019;29(1):44-65.
Bugter JM, Fenderico N, Maurice MM. Mutations and mechanisms of WNT pathway tumour suppressors in cancer. Nat Rev Cancer. 2021;21(1):5-21.
Albrecht LV, Tejeda-Muñoz N, De Robertis EM. Cell Biology of Canonical Wnt Signaling. Annu Rev Cell Dev Biol. 2021;37:369-89.
Chen C, Zhao M, Tian A, Zhang X, Yao Z, Ma X. Aberrant activation of Wnt/β-catenin signaling drives proliferation of bone sarcoma cells. Oncotarget. 2015;6(19):17570-83.
McQueen P, Ghaffar S, Guo Y, Rubin EM, Zi X, Hoang BH. The Wnt signaling pathway: implications for therapy in osteosarcoma. Expert Rev Anticancer Ther. 2011;11(8):1223-32.
Sheng G, Gao Y, Yang Y, Wu H. Osteosarcoma and Metastasis. Front Oncol. 2021;11:780264.
Yang C, Tian Y, Zhao F, Chen Z, Su P, Li Y, et al. Bone Microenvironment and Osteosarcoma Metastasis. Int J Mol Sci. 2020;21(19).
Grünewald TGP, Cidre-Aranaz F, Surdez D, Tomazou EM, de Álava E, Kovar H, et al. Ewing sarcoma. Nat Rev Dis Primers. 2018;4(1):5.
Smeland S, Bielack SS, Whelan J, Bernstein M, Hogendoorn P, Krailo MD, et al. Survival and prognosis with osteosarcoma: outcomes in more than 2000 patients in the EURAMOS-1 (European and American Osteosarcoma Study) cohort. Eur J Cancer. 2019;109:36-50.
Koirala P, Roth ME, Gill J, Piperdi S, Chinai JM, Geller DS, et al. Immune infiltration and PD-L1 expression in the tumor microenvironment are prognostic in osteosarcoma. Sci Rep. 2016;6:30093.
Wolf-Dennen K, Gordon N, Kleinerman ES. Exosomal communication by metastatic osteosarcoma cells modulates alveolar macrophages to an M2 tumor-promoting phenotype and inhibits tumoricidal functions. Oncoimmunology. 2020;9(1):1747677.
Corre I, Verrecchia F, Crenn V, Redini F, Trichet V. The Osteosarcoma Microenvironment: A Complex But Targetable Ecosystem. Cells. 2020;9(4).
Shao XJ, Xiang SF, Chen YQ, Zhang N, Cao J, Zhu H, et al. Inhibition of M2-like macrophages by all-trans retinoic acid prevents cancer initiation and stemness in osteosarcoma cells. Acta Pharmacol Sin. 2019;40(10):1343-50.
Zhang C, Zheng JH, Lin ZH, Lv HY, Ye ZM, Chen YP, et al. Profiles of immune cell infiltration and immune-related genes in the tumor microenvironment of osteosarcoma. Aging (Albany NY). 2020;12(4):3486-501.
Synoradzki KJ, Bartnik E, Czarnecka AM, Fiedorowicz M, Firlej W, Brodziak A, et al. TP53 in Biology and Treatment of Osteosarcoma. Cancers (Basel). 2021;13(17).
Thoenen E, Curl A, Iwakuma T. TP53 in bone and soft tissue sarcomas. Pharmacol Ther. 2019;202:149-64.
Katoh M, Katoh M. WNT signaling pathway and stem cell signaling network. Clin Cancer Res. 2007;13(14):4042-5.
Vijayakumar S, Liu G, Rus IA, Yao S, Chen Y, Akiri G, et al. High-frequency canonical Wnt activation in multiple sarcoma subtypes drives proliferation through a TCF/β-catenin target gene, CDC25A. Cancer Cell. 2011;19(5):601-12.
Matsuoka K, Bakiri L, Wolff LI, Linder M, Mikels-Vigdal A, Patiño-García A, et al. Wnt signaling and Loxl2 promote aggressive osteosarcoma. Cell Res. 2020;30(10):885-901.
Gustafson CT, Mamo T, Shogren KL, Maran A, Yaszemski MJ. FH535 Suppresses Osteosarcoma Growth In Vitro and Inhibits Wnt Signaling through Tankyrases. Front Pharmacol. 2017;8:285.
Pridgeon MG, Grohar PJ, Steensma MR, Williams BO. Wnt Signaling in Ewing Sarcoma, Osteosarcoma, and Malignant Peripheral Nerve Sheath Tumors. Curr Osteoporos Rep. 2017;15(4):239-46.
Shen S, Yao T, Xu Y, Zhang D, Fan S, Ma J. CircECE1 activates energy metabolism in osteosarcoma by stabilizing c-Myc. Mol Cancer. 2020;19(1):151.
Tago K, Nakamura T, Nishita M, Hyodo J, Nagai S, Murata Y, et al. Inhibition of Wnt signaling by ICAT, a novel beta-catenin-interacting protein. Genes Dev. 2000;14(14):1741-9.
Stow JL. ICAT is a multipotent inhibitor of beta-catenin. Focus on "role for ICAT in beta-catenin-dependent nuclear signaling and cadherin functions". Am J Physiol Cell Physiol. 2004;286(4):C745-6.
Gottardi CJ, Gumbiner BM. Role for ICAT in beta-catenin-dependent nuclear signaling and cadherin functions. Am J Physiol Cell Physiol. 2004;286(4):C747-56.
Valenta T, Hausmann G, Basler K. The many faces and functions of β-catenin. Embo j. 2012;31(12):2714-36.
Huang R, Zhang L, Jin J, Zhou Y, Zhang H, Lv C, et al. Bruceine D inhibits HIF-1α-mediated glucose metabolism in hepatocellular carcinoma by blocking ICAT/β-catenin interaction. Acta Pharm Sin B. 2021;11(11):3481-92.
Bi W, Huang J, Nie C, Liu B, He G, Han J, et al. CircRNA circRNA_102171 promotes papillary thyroid cancer progression through modulating CTNNBIP1-dependent activation of β-catenin pathway. J Exp Clin Cancer Res. 2018;37(1):275.
Chang JM, Tsai AC, Huang WR, Tseng RC. The Alteration of CTNNBIP1 in Lung Cancer. Int J Mol Sci. 2019;20(22).
Kosari-Monfared M, Nikbakhsh N, Fattahi S, Ghadami E, Ranaei M, Taheri H, et al. CTNNBIP1 downregulation is associated with tumor grade and viral infections in gastric adenocarcinoma. J Cell Physiol. 2019;234(3):2895-904.
Kang DW, Lee BH, Suh YA, Choi YS, Jang SJ, Kim YM, et al. Phospholipase D1 Inhibition Linked to Upregulation of ICAT Blocks Colorectal Cancer Growth Hyperactivated by Wnt/β-Catenin and PI3K/Akt Signaling. Clin Cancer Res. 2017;23(23):7340-50.
Jiang Y, Ren W, Wang W, Xia J, Gou L, Liu M, et al. Inhibitor of β-catenin and TCF (ICAT) promotes cervical cancer growth and metastasis by disrupting E-cadherin/β-catenin complex. Oncol Rep. 2017;38(5):2597-606.

No competing interests reported.

Download PDF

Editorial decision: Reject after review
16 Dec, 2022
Editor assigned by journal
16 Dec, 2022
Submission checks completed at journal
16 Dec, 2022
First submitted to journal
15 Dec, 2022

You are reading this latest preprint version

Wnt Signaling Pathway Related Gene CTNNBIP1 Act as A potential Prognostic Biomarker and A Comprehensive Prognostic Analysis in Osteosarcoma

Status:

Version 2

Abstract

Purpose

Methods

Results

Conclusions

Figures

1 Introduction

2 Materials And Methods

3 Results

4 Discussion

5 Conclusion

List Of Abbreviations

Declarations

Ethics approval and consent to participate

Consent for publication

Availability of data and materials

Competing interests

Funding

Authors' contributions

Acknowledgements

References

Additional Declarations

Supplementary Files

Status:

Version 2