A deep learning model for accurately predicting cancer-specific survival in patients with primary bone sarcoma of the extremity: a population-based study

Purpose Primary bone and joint sarcomas of the long bone are relatively rare neoplasms with poor prognosis. An efficient clinical tool that can accurately predict patient prognosis is not available. The current study aimed to use deep learning algorithms to develop a prediction model for the prognosis of patients with long bone sarcoma. Methods Data of patients with long bone sarcoma in the extremities was collected from the Surveillance, Epidemiology, and End Results Program database from 2004 to 2014. Univariate and multivariate analyses were performed to select possible prediction features. DeepSurv, a deep learning model, was constructed for predicting cancer-specific survival rates. In addition, the classical cox proportional hazards model was established for comparison. The predictive accuracy of our models was assessed using the C-index, Integrated Brier Score, receiver operating characteristic curve, and calibration curve. Results Age, tumor extension, histological grade, tumor size, surgery, and distant metastasis were associated with cancer-specific survival in patients with long bone sarcoma. According to loss function values, our models converged successfully and effectively learned the survival data of the training cohort. Based on the C-index, area under the curve, calibration curve, and Integrated Brier Score, the deep learning model was more accurate and flexible in predicting survival rates than the cox proportional hazards model. Conclusion A deep learning model for predicting the survival probability of patients with long bone sarcoma was constructed and validated. It is more accurate and flexible in predicting prognosis than the classical CoxPH model (AU)

Unraveling T cell exhaustion in the immune microenvironment of osteosarcoma via single-cell RNA transcriptome.

Osteosarcoma (OS) represents a profoundly invasive malignancy of the skeletal system. T cell exhaustion (Tex) is known to facilitate immunosuppression and tumor progression, but its role in OS remains unclear. In this study, single-cell RNA sequencing data was employed to identify exhausted T cells within the tumor immune microenvironment (TIME) of OS. We found that exhausted T cells exhibited substantial infiltration in OS samples. Pseudotime trajectory analysis revealed a progressive increase in the expression of various Tex marker genes, including PDCD1, CTLA4, LAG3, ENTPD1, and HAVCR2 in OS. GSVA showed that apoptosis, fatty acid metabolism, xenobiotic metabolism, and the interferon pathway were significantly activated in exhausted T cells in OS. Subsequently, a prognostic model was constructed using two Tex-specific genes, MYC and FCGR2B, which exhibited exceptional prognostic accuracy in two independent cohorts. Drug sensitivity analysis revealed that OS patients with a low Tex risk were responsive to Dasatinib and Pazopanib. Finally, immunohistochemistry verified that MYC and FCGR2B were significantly upregulated in OS tissues compared with adjacent tissues. This study investigates the role of Tex within the TIME of OS, and offers novel insights into the mechanisms underlying disease progression as well as the potential treatment strategies for OS.

A deep learning model for accurately predicting cancer-specific survival in patients with primary bone sarcoma of the extremity: a population-based study.

PURPOSE: Primary bone and joint sarcomas of the long bone are relatively rare neoplasms with poor prognosis. An efficient clinical tool that can accurately predict patient prognosis is not available. The current study aimed to use deep learning algorithms to develop a prediction model for the prognosis of patients with long bone sarcoma. METHODS: Data of patients with long bone sarcoma in the extremities was collected from the Surveillance, Epidemiology, and End Results Program database from 2004 to 2014. Univariate and multivariate analyses were performed to select possible prediction features. DeepSurv, a deep learning model, was constructed for predicting cancer-specific survival rates. In addition, the classical cox proportional hazards model was established for comparison. The predictive accuracy of our models was assessed using the C-index, Integrated Brier Score, receiver operating characteristic curve, and calibration curve. RESULTS: Age, tumor extension, histological grade, tumor size, surgery, and distant metastasis were associated with cancer-specific survival in patients with long bone sarcoma. According to loss function values, our models converged successfully and effectively learned the survival data of the training cohort. Based on the C-index, area under the curve, calibration curve, and Integrated Brier Score, the deep learning model was more accurate and flexible in predicting survival rates than the cox proportional hazards model. CONCLUSION: A deep learning model for predicting the survival probability of patients with long bone sarcoma was constructed and validated. It is more accurate and flexible in predicting prognosis than the classical CoxPH model.

Deciphering the heterogeneity and immunosuppressive function of regulatory T cells in osteosarcoma using single-cell RNA transcriptome.

Osteosarcoma (OS) is a highly invasive malignant neoplasm with poor prognosis. The tumor microenvironment (TME) plays an essential role in the occurrence and development of OS. Regulatory T cells (Tregs) are known to facilitate immunosuppression, tumor progression, invasion, and metastasis. However, the effect of Tregs in the TME of OS remains unclear. In this study, single-cell RNA sequencing (scRNA-seq) data was used to identify Tregs and various other cell clusters in the TME of OS. Gene set variation analysis (GSVA) was used to investigate the signaling pathways in Tregs from OS and adjacent tissues. The CellChat and iTALK packages were used to analyze cellular communication. In addition, a prognostic model was established based on the Tregs-specific genes using bulk RNA-seq from the TARGET database, and it was verified using a Gene Expression Omnibus dataset. The pRRophetic package was used to predict drug sensitivity. Immunohistochemistry was used to verify the expression of candidate genes in OS. Based on the above methods, we showed that the OS samples were highly infiltrated with Tregs. GSVA revealed that oxidative phosphorylation, angiogenesis and mammalian target of rapamycin complex 1 (mTORC1) were highly activated in Tregs from OS compared with those from adjacent tissues. Using cellular communication analysis, we found that Tregs interacted with osteoblastic, endothelial, and myeloid cells via C-X-C motif chemokine ligand (CXCL) signaling; particularly, they strongly affected the expression of C-X-C motif chemokine receptor 4 (CXCR4) and interacted with other cell clusters through CXCL12/transforming growth factor ß1 (TGFB1) to collectively enable tumor growth and progression. Subsequently, two Tregs-specific genes-CD320 and MAF-were screened through univariate, least absolute shrinkage and selection operator regression (LASSO) and multivariate analysis to construct a prognostic model, which showed excellent prognostic accuracy in two independent cohorts. In addition, drug sensitivity analysis demonstrated that OS patients at high Tregs risk were sensitive to sunitinib, sorafenib, and axitinib. We also used immunohistochemistry to validate that CD320 and MAF were significantly upregulated in OS tissues compared with adjacent tissues. Overall, this study reveals the heterogeneity of Tregs in the OS TME, providing new insights into the invasion and treatment of this cancer.

Deep-Learning-Based Model for the Prediction of Cancer-Specific Survival in Patients with Spinal Chordoma.

OBJECTIVE: Spinal chordomas are locally aggressive and frequently recurrent tumors with a poor prognosis. Previous studies focused on a Cox regression model to predict the survival of patients with spinal chordoma. We aimed to develop a more effective model based on deep learning for prognosis prediction in spinal chordoma. METHODS: Patients with spinal chordoma were gathered from the SEER database. Cox regression analysis was conducted to compare the influence of different clinical characteristics on cancer-specific survival. Two deep learning models, namely, DeepSurv and NMTLR, were developed, alongside 2 classic models, for the purpose of comparison. Performance of these models was evaluated by concordance index, Integrated Brier Score, receiver operating characteristic curves, Kaplan-Meier curves, and calibration curves. RESULTS: A total of 258 spinal chordoma patients were included in the current study. The median follow-up time was 94 ± 52 months. Variables used for prognosis prediction consisted of age, primary site, tumor size, histologic grade, extension of surgery, tumor invasion, and metastasis. Comparing with conventional models, each deep learning model showed superior predictive performance, the C-index on the test cohort is 0.830 for DeepSurv and 0.804 for NMTLR, respectively. The DeepSurv model represented the best performance, with area under the curve of 0.843 in predicting 5-year survival and 0.880 in predicting 10-year survival. CONCLUSIONS: We successfully constructed a deep learning model to predict the CSS of spinal chordoma patients and proved that it was more accurate and practical than conventional prediction model. Our deep learning model has the potential to guide clinicians in better care planning and decision-making.

Deciphering the crosstalk of immune dysregulation between COVID-19 and idiopathic inflammatory myopathy.

Background: The coronavirus disease (COVID-19) pandemic is a serious threat to public health worldwide. Growing evidence reveals that there are certain links between COVID-19 and autoimmune diseases; in particular, COVID-19 and idiopathic inflammatory myopathies (IIM) have been observed to be clinically comorbid. Hence, this study aimed to elucidate the molecular mechanisms of COVID-19 and IIM from a genomic perspective. Methods: We obtained transcriptome data of patients with COVID-19 and IIM separately from the GEO database and identified common differentially expressed genes (DEGs) by intersection. We then performed functional enrichment, PPI, machine learning, gene expression regulatory network, and immune infiltration analyses of co-expressed genes. Results: A total of 91 common genes were identified between COVID-19 and IIM. Functional enrichment analysis revealed that these genes were mainly involved in immune dysregulation, response to external stimuli, and MAPK signaling pathways. The MCODE algorithm recognized two densely linked clusters in the common genes, which were related to inflammatory factors and interferon signaling. Subsequently, three key genes (CDKN1A, IFI27, and STAB1) were screened using machine learning to predict the occurrence of COVID-19 related IIM. These key genes exhibited excellent diagnostic performance in both training and validation cohorts. Moreover, we created TF-gene and miRNA-gene networks to reveal the regulation of key genes. Finally, we estimated the relationship between key genes and immune cell infiltration, of which IFI27 was positively associated with M1 macrophages. Conclusion: Our work revealed common molecular mechanisms, core genes, potential targets, and therapeutic approaches for COVID-19 and IIM from a genomic perspective. This provides new ideas for the diagnosis and treatment of COVID-19 related IIM in the future.

Using network pharmacology to fabricate crosslinked hyaluronan-chondroitin sulphate-resveratrol composite hydrogels for cartilage regeneration.

Cartilage tissue engineering provides a new approach for the treatment of cartilage damage. The combination of drug system with a tissue scaffold could be highly beneficial. Resveratrol (RES) is a potent anti-inflammatory agent, but its target genes and molecular mechanism of cartilage repair remain to be further studied. We used systems biology and network pharmacology methods to explore the mechanism of RES for chondrocyte and macrophages. Meanwhile, crosslinked hyaluronan-chondroitin sulphate-RES hydrogels (cHA-CS-RES) were constructed based on the target prediction results. Byin vitroandin vivoexperiments, we investigated its anti-inflammatory and pro-chondrogenesis. The results showed there were 12 hub genes potentially interacting in the RES-chondrocyte-macrophage network.In vitroexperiments were used to further verify the validity of the predicted hub genes. The composite hydrogels were successfully fabricated, and maintenance of the characteristic was further confirmed.In vitrostudy, cHA-CS-RES showed high cell viability, anti-inflammatory and pro-chondrogenesis abilities.In vivostudy of cartilage defects confirmed that the cHA-CS-RES groups were significantly better than the control group. Network pharmacology was used to predict and screen the target proteins of RES critical to cartilage tissue engineering. Moreover, cHA-CS-RES composite hydrogel showed good cartilage repair effects, anti-inflammatory and pro-chondrogenesis abilities.

Expression of lactate-related signatures correlates with immunosuppressive microenvironment and prognostic prediction in ewing sarcoma.

Objectives: Ewing sarcoma (EWS) is an aggressive tumor of bone and soft tissue. Growing evidence indicated lactate as a pivotal mediator of crosstalk between tumor energy metabolism and microenvironmental regulation. However, the contribution of lactate-related genes (LRGs) in EWS is still unclear. Methods: We obtained the transcriptional data of EWS patients from the GEO database and identified differentially expressed-LRGs (DE-LRGs) between EWS patient samples and normal tissues. Unsupervised cluster analysis was utilized to recognize lactate modulation patterns based on the expression profile of DE-LRGs. Functional enrichment including GSEA and GSVA analysis was conducted to identify molecular signaling enriched in different subtypes. ESTIMATE, MCP and CIBERSORT algorithm was used to explore tumor immune microenvironment (TIME) between subtypes with different lactate modulation patterns. Then, lactate prognostic risk signature was built via univariate, LASSO and multivariate Cox analysis. Finally, we performed qPCR analysis to validate candidate gene expression. Result: A total of 35 DE-LRGs were identified and functional enrichment analysis indicated that these LRGs were involved in mitochondrial function. Unsupervised cluster analysis divided EWS patients into two lactate modulation patterns and we revealed that patients with Cluster 1 pattern were linked to poor prognosis and high lactate secretion status. Moreover, TIME analysis indicated that the abundance of multiple immune infiltrating cells were dramatically elevated in Cluster 1 to Cluster 2, including CAFs, endothelial cells, Macrophages M2, etc., which might contribute to immunosuppressive microenvironment. We also noticed that expression of several immune checkpoint proteins were clearly increased in Cluster 1 to Cluster 2. Subsequently, seven genes were screened to construct LRGs prognostic signature and the performance of the resulting signature was validated in the validation cohort. Furthermore, a nomogram integrating LRGs signature and clinical characteristics was developed to predict effectively the 4, 6, and 8-year prognosis of EWS patients. Conclusion: Our study revealed the role of LRGs in immunosuppressive microenvironment and predicting prognosis in EWS and provided a robust tool to predict the prognosis of EWS patients.

Unfolded Protein Response-Related Signature Associates With the Immune Microenvironment and Prognostic Prediction in Osteosarcoma.

Background: Osteosarcoma is a highly malignant bone tumor commonly occurring in adolescents with a poor 5-year survival rate. The unfolded protein response (UPR) can alleviate the accumulation of misfolded proteins to maintain homeostasis under endoplasmic reticulum stress. The UPR is linked to the occurrence, progression, and drug resistance of tumors. However, the function of UPR-related genes (UPRRGs) in disease progression and prognosis of osteosarcoma remains unclear. Methods: The mRNA expression profiling and corresponding clinical features of osteosarcoma were acquired from TARGET and GEO databases. Consensus clustering was conducted to confirm different UPRRG subtypes. Subsequently, we evaluated the prognosis and immune status of the different subtypes. Functional analysis of GO, GSEA, and GSVA was used to reveal the molecular mechanism between the subtypes. Finally, four genes (STC2, PREB, TSPYL2, and ATP6V0D1) were screened to construct and validate a risk signature to predict the prognosis of patients with osteosarcoma. Result: We identified two subtypes according to the UPRRG expression patterns. The subgroup with higher immune scores, lower tumor purity, and active immune status was linked to a better prognosis. Meanwhile, functional enrichment revealed that immune-related signaling pathways varied markedly in the two subtypes, suggesting that the UPR might influence the prognosis of osteosarcoma via influencing the immune microenvironment. Moreover, prognostic signature and nomogram models were developed based on UPRRGs, and the results showed that our model has an excellent performance in predicting the prognosis of osteosarcoma. qPCR analysis was also conducted to verify the expression levels of the four genes. Conclusion: We revealed the crucial contribution of UPRRGs in the immune microenvironment and prognostic prediction of osteosarcoma patients and provided new insights for targeted therapy and prognostic assessment of the disease.

[Biomechanical analysis and effectiveness evaluation of zone â +â ¡+â ¢ reconstruction of hemipelvis with rod-screw prosthesis].

Objective: To analyze the biomechanical properties of the rod-screw prosthesis based on a pelvic three-dimensional finite element model including muscle and ligament, and evaluate the effectiveness of zoneⅠ+Ⅱ+Ⅲ reconstruction of hemipelvis with rod-screw prosthesis in combination with clinical applications. Methods: A total of 21 patients who underwent hemipelvic tumor resection (zoneⅠ+Ⅱ+Ⅲ) and rod-screw prosthesis reconstruction between January 2015 and December 2020 were selected as the research subjects. Among them, there were 11 males and 10 females; the age ranged from 16 to 64 years, with an average age of 39.2 years. There were 9 cases of chondrosarcoma, 7 cases of osteosarcoma, 3 cases of Ewing sarcoma, and 2 cases of undifferentiated pleomorphic sarcoma. According to the Musculoskeletal Tumor Society Score (MSTS) staging, there were 19 cases of stage ⅡB and 2 cases of stage Ⅲ. Preoperative Harris Hip Score (HHS) and MSTS score were 54.4±3.1 and 14.1±2.0, respectively. Intraoperative 15 cases underwent extensive resection, 5 cases underwent marginal resection, and 1 case underwent intralesional resection. The CT image of 1 patient after reconstruction was used to establish a three-dimensional solid model of the pelvis via Mimics23Suite and 3-matic softwares. At the same time, a mirror operation was used to obtain a normal pelvis model, then the two solid models were imported into the finite element analysis software Workbench 2020R1 to establish three-dimensional finite element models, and the biomechanical properties of the standing position were analyzed. The operation time, intraoperative blood loss, and operation-related complications were recorded, and the postoperative evaluation was carried out with HHS and MSTS scores. Finally, the local recurrence and metastasis were reviewed. Results: Finite element analysis showed that the peak stress of the reconstructed pelvis appeared at the fixed S1, 2 rod-screw connections; the peak stress without muscles was higher than that after muscle construction, but much smaller than the yield strength of titanium alloy. The operation time was 250-370 minutes, with an average of 297 minutes; the amount of intraoperative blood loss was 3 200-5 500 mL, with an average of 4 009 mL. All patients were followed up 8-72 months, with an average of 42 months. There were 7 cases of pulmonary metastasis, of which 2 cases were preoperative metastasis; 5 cases died, 16 cases survived, and the 5-year survival rate was 72.1%. There were 3 cases of local recurrence, all of whom did not achieve extensive resection during operation. The function of the affected limbs significantly improved, and the walking function was restored. The HHS and MSTS scores were 75.2±3.0 and 20.4±2.0 at last follow-up, respectively, and the differences were significant when compared with those before operation (t=22.205, P<0.001; t=11.915, P<0.001). During follow-up, 2 cases of delayed incision healing, 2 cases of deep infection, 1 case of screw loosening, and 1 case of prosthesis dislocation occurred, and no other complication such as prosthesis or screw fracture occurred. Conclusion: The stress and deformation distribution of the reconstructed pelvis are basically the same as normal pelvis. The rod-screw prosthesis is an effective reconstruction method for pelvic malignant tumors.

Clinical application of 3D-printed patient-specific guide plate combined with computer navigation in acetabular reconstruction following resection of periacetabular tumors.

Background: The precise acetabular reconstruction has historically been a challenging procedure. 3D-printed patient-specific guide (PSG) and computer navigation (CN) technologies have been used to assist acetabular component positioning and pelvic reconstruction. This precise reconstruction approach may translate into clinical benefit. Methods: The clinical data of 84 patients who underwent periacetabular malignant tumor resection and screw-rod-acetabular cage system reconstruction in our center from January 2013 to December 2020 were retrospectively analyzed. Patients were divided into four groups: free hand (FH) group, PSG group, CN group, and PSG combined with computer navigation (PSG + CN) group. The operation time, intraoperative blood loss, and number of fluoroscopy views were recorded. The oncological prognosis, radiographic measurements of the acetabulum, limb function data, and postoperative complications were compared among groups. And finally, we evaluated the risk factors for mechanical failure of the prosthesis. Results: The postoperative X-ray and computed tomography (CT) scan revealed that the vertical offset discrepancy (VOD) between affected side and contralateral side was 8.4±1.9, 5.9±2.2, 4.1±1.3, and 2.4±1.2 mm in each groups; the horizontal offset discrepancy (HOD) was 9.0±1.9, 6.1±2.2, 3.2±1.3, and 2.1±1.2 mm, correspondingly; the abduction angle discrepancy (ABAD) was 8.6°±1.8°, 5.6°±2.0°, 2.5°±1.3°, and 1.8°±0.9°, respectively; the anteversion angle discrepancy (ANAD) was 5.9°±1.6°, 3.6°±1.7°, 2.9°±1.6°, and 1.9°±0.9°, correspondingly. Statistical results show that the PSG + CN group was superior to the FH group and the PSG group in terms of acetabular position and limb function (P<0.05). Body mass index (P=0.040) and resection type (P=0.042) were found to be the high-risk factors for mechanical failure of the prosthesis. Conclusions: PSG + CN has potential advantages in improving the accuracy and safety of acetabular positioning and reconstruction.

Comparison between trabectedin and doxorubicin in soft-tissue sarcomas: a systematic review and meta-analysis.

BACKGROUND: This study sought to evaluate the differences between trabectedin and doxorubicin in the treatment of soft-tissue sarcoma (STS). METHODS: Multiple databases, including PubMed, Web of Science, Cochrane Library, and China National Knowledge Infrastructure, were searched to retrieve relevant articles. Ultimately, the full text of 10 studies involving the use of trabectedin and doxorubicin in STS were reviewed. Review Manager 5.2 was used to evaluate the heterogeneity of the results of the selected articles. Forest plot, bias, and sensitivity analyses were carried out on the included articles. RESULTS: Ten papers that met the criteria were included in this analysis. STS patients receiving trabectedin had longer progression-free survival than those receiving doxorubicin [overall mean difference (MD) =1.36, 95% confidence interval (CI): 1.04, 1.68, I2=6%, fixed-effects model]. The experimental group also had a longer overall survival period than the control group (MD =3.92, 95% CI: 0.23, 7.60, P=0.04 and I2=83%, random-effects model), and the experimental group had a better disease control rate than the control group (relative risk =1.2, P=0.03 and I2=45%, fixed-effects model). From the publication bias analysis and sensitivity analysis, we can guarantee the results are robust and unbiased. DISCUSSION: Our research showed that STS patients who received trabectedin had better clinical effects and a longer survival time than those who received doxorubicin.