Abstract
Objective
To analyze risk factors for pediatric osteosarcoma and to construct and evaluate a risk prediction model for pediatric osteosarcoma.
Methods
We retrospectively analyzed data from patients diagnosed with osteosarcoma between 2000 and 2013 (n = 129) from the TARGET database. First, independent prognostic factors associated with osteosarcoma-specific death were identified through Cox proportional hazards modeling. Subsequently, using these independent prognostic factors, a nomogram model for osteosarcoma-specific survival was constructed using SPSS 25.0 and R 4.1.1. The discrimination of the model was evaluated using the C-index, predictive ability was validated through receiver operating characteristic curves and area under the curve values, calibration was assessed using calibration curves, and clinical utility was measured by decision curve analysis. Additionally, Kaplan-Meier survival analysis was performed to test the rationality of nomogram grouping.
Results
The final model included six variables: sex, race, tumor-specific side, tumor-specific region, site of first recurrence, and time of first recurrence. The C-indices of the model for predicting 3-year and 5-year survival rates were 0.802 (95% CI: 0.725–0.880) and 0.787 (95% CI: 0.710–0.864), respectively, indicating good discriminatory ability. Calibration curves showed high consistency between predicted and actual survival probabilities. Decision curve analysis indicated that the model has substantial net benefit across a wide range of mortality risk thresholds. Kaplan-Meier survival analysis showed significant differences in prognosis between high-risk and low-risk groups. The nomogram model constructed in this study can accurately predict 1-year, 3-year, and 5-year survival of pediatric osteosarcoma patients and has high clinical utility.
Conclusion
This model not only provides an effective survival prediction tool for patients but also offers important references for optimizing treatment strategies for pediatric osteosarcoma, with the aim of improving survival rates and quality of life for osteosarcoma patients.
Keywords: Osteosarcoma, Pediatric, TARGET database, Nomogram, Risk prediction model, Survival analysis
Introduction
Osteosarcoma is a non-hematopoietic malignant bone tumor that accounts for approximately 40% of all primary malignant bone tumors, with adolescents representing about 70% of cases. The annual incidence rate is 8.2 per 100,000 in children aged 10–19 years and 1.7 per 100,000 in children under 10 years [1–3]. Although surgical treatment has certain clinical efficacy, approximately 80% of pediatric osteosarcoma patients present with metastatic lesions at initial clinical diagnosis, resulting in poor prognosis. The current standard treatment for osteosarcoma includes neoadjuvant chemotherapy followed by surgical resection of the primary tumor, surgical removal of all clinically significant metastases, and postoperative adjuvant chemotherapy, which has gradually improved the 5-year overall survival rate from 60 to 80% [4, 5]. Despite the tremendous success in osteosarcoma treatment, its high recurrence rate, tendency for metastasis, and high mortality necessitate the identification of factors affecting long-term prognosis and the implementation of targeted preventive measures to improve comprehensive treatment outcomes, extend survival, and enhance patient prognosis.
In recent years, nomograms have been widely applied both domestically and internationally, calculating the probability of clinical events through complex computational formulas [6–8]. With the aid of nomograms, clinicians can assess the risk of clinical events, design individualized treatment plans, determine the use of adjuvant therapies, optimize treatment strategies, and consider appropriate patient counseling [6].
Therefore, establishing a model that can accurately predict the survival of pediatric osteosarcoma patients is of significant importance for improving clinical decision-making and patient prognosis. This study retrospectively analyzed data from pediatric patients diagnosed with osteosarcoma between 2000 and 2013 (n = 129) from the TARGET database to construct and validate a nomogram model capable of accurately predicting the survival of pediatric osteosarcoma patients. We hypothesized that by integrating multiple prognostic factors, this model could provide more precise survival predictions, helping physicians better understand patients’ potential risks, make more reasonable clinical decisions, optimize treatment strategies, and improve patients’ survival rates and quality of life.
Materials and methods
Design
Data analysis was conducted using Cox regression to establish the model. The model’s discrimination, calibration, decision curve analysis, and time-to-event data analysis were evaluated using receiver operating characteristic (ROC) curves, Calibration plots, Decision Curve Analysis (DCA), and Kaplan-Meier survival curves, respectively. The dataset was randomly divided into training (70%, n = 90) and validation (30%, n = 39) cohorts for model development and internal validation.
Materials
We retrospectively analyzed data from patients diagnosed with osteosarcoma between 2000 and 2013 (n = 129) from the TARGET database. This study complied with the Declaration of Helsinki.
Osteosarcoma patients were classified according to the International Classification of Diseases for Oncology, Third Edition (ICD-O-3) (ICD-O-3 histological types: 9180–9187, 9192–9195), including osteosarcoma Not Otherwise Specified (NOS), chondroblastic osteosarcoma, fibroblastic osteosarcoma, telangiectatic osteosarcoma, small cell osteosarcoma, Paget’s disease-associated osteosarcoma, central osteosarcoma, periosteal osteosarcoma, intraosseous well-differentiated osteosarcoma, high-grade surface osteosarcoma, and parosteal osteosarcoma. Patients diagnosed solely based on clinical presentation or imaging, those confirmed only by death certificates or autopsy, and those with unknown treatment methods or lacking survival information were excluded.
Sample size justification: Based on the widely accepted principle of having at least 10 events per variable (EPV) for developing predictive models [9], with 68 events (deaths) in our cohort and 6 variables in the final model, our EPV ratio is 11.3, which exceeds the minimum recommended threshold. This sample size is therefore adequate for developing a reliable nomogram.
Methods
Data collection
We retrospectively analyzed data from patients diagnosed with osteosarcoma between 2000 and 2013 (n = 129) from the TARGET database. Baseline data were collected, including age, sex, race, diagnosis time, metastasis site, primary site, tumor-specific side, tumor-specific region, site of first recurrence, time of first recurrence, surgical status, and time of death.
Osteosarcoma patients were classified according to the International Classification of Diseases for Oncology, Third Edition (ICD-O-3). Data extraction and processing were performed using SPSS 25.0 and R 4.1.1 software to ensure data completeness and accuracy. Missing values were handled using predictive mean matching (PMM) multiple imputation method. The extent of missing data was as follows: histologic necrosis ratio (58.9%, n = 76), all other variables had complete data. Due to the high proportion of missing data for histologic necrosis ratio, this variable was excluded from the final model. The exclusion of commonly used prognostic factors such as tumor size and specific histologic subtype was due to their unavailability in the TARGET database.
Model construction and validation
Statistical analysis was performed using SPSS statistical software (SPSS 25.0), and model building and prediction were conducted using R (R 4.1.1) software. First, independent prognostic factors associated with osteosarcoma-specific death were identified through Cox proportional hazards modeling. Subsequently, using these independent prognostic factors, a nomogram model for osteosarcoma-specific survival was constructed using SPSS 25.0 and R 4.1.1. The discrimination of the model was evaluated using the C-index, predictive ability was validated through receiver operating characteristic curves and area under the curve values, calibration was assessed using calibration curves, and clinical utility was measured by decision curve analysis. Additionally, Kaplan-Meier survival analysis was performed to test the rationality of nomogram grouping.
Main observation indicators
①Forest plot of risk factors from Cox regression analysis; ②Comparative survival curves from Cox regression analysis; ③Construction and validation of the nomogram prediction model for patient survival rates.
Statistical analysis
①SPSS 25.0 and R 4.1.1 software. ②The Cox proportional hazards model is a statistical method widely used in survival analysis to assess the impact of one or more independent variables on survival time. In this study, the Cox model was used to analyze which factors were associated with the survival of pediatric osteosarcoma patients. ③A nomogram is a graphical prediction tool based on statistical model results (such as Cox model) that integrates multiple prognostic factors to predict the probability of specific clinical events. In this paper, the nomogram was used to predict 1-year, 3-year, and 5-year survival probabilities for pediatric osteosarcoma patients. ④The ROC curve is a graphical tool used to evaluate the performance of binary classification models, showing the relationship between true positive rate (sensitivity) and false positive rate. In this study, ROC curves were used to assess the predictive ability of the nomogram model, especially for 1-year, 3-year, and 5-year survival predictions. ⑤Calibration curves are used to evaluate the accuracy of prediction models, i.e., the consistency between model-predicted probabilities and actual event occurrence proportions. In this study, calibration curves were used to assess the accuracy of the nomogram model in predicting 1-year, 3-year, and 5-year survival. ⑥DCA is a method used to evaluate the clinical utility of clinical prediction models by directly considering the impact of model prediction results on clinical decision-making. In this study, DCA was used to assess the clinical value of the nomogram model in predicting 1-year, 3-year, and 5-year survival. ⑦Kaplan-Meier survival analysis is a non-parametric method used to estimate survival probabilities, capable of handling censored data (i.e., data from patients without complete follow-up information). In this study, Kaplan-Meier survival analysis was used to evaluate survival differences between high-risk and low-risk groups, thereby verifying the rationality of the nomogram model grouping.
Results
Trial flow chart
Based on inclusion and exclusion criteria, data from 129 patients were ultimately included for analysis. The entire process ensured data integrity and scientific rigor of the study. See Fig. 1.
Fig. 1.
Forest gram of each risk factors in Cox survival analysis
Clinical data statistics
The 129 pediatric osteosarcoma patients had a mean age of (5,476.37 ± 1,822.06) days. Other clinical characteristics are detailed in Table 1.
Table 1.
Patient clinical characteristics
| Item | Quantity |
|---|---|
| Age (days) | 1,828.66 ± 1,249.54 |
| Sex | |
| Male | 74 |
| Female | 55 |
| Race | |
| Non-Hispanic or Latino | 83 |
| Hispanic or Latino | 14 |
| Unknown | 32 |
| Diagnosis time (days) | 5,476.37 ± 1,822.06 |
| Metastasis site | |
| Bone | 4 |
| Lung | 18 |
| Bone and lung | 3 |
| Other sites | 104 |
| No metastasis | 1 |
| Primary site | |
| Upper limb bones | 13 |
| Lower limb bones | 113 |
| Trunk bones | 3 |
| Tumor-specific side | |
| Right | 32 |
| Left | 37 |
| Other | 60 |
| Tumor-specific region | |
| Proximal | 39 |
| Distal | 56 |
| Other | 34 |
| Site of first recurrence | |
| Lung | 31 |
| Bone | 7 |
| Other | 91 |
| Time of first recurrence | |
| Within 1 year | 36 |
| 1–2 years | 25 |
| 2–3 years | 3 |
| More than 3 years | 65 |
| Surgery | |
| No | 1 |
| Limb-sparing surgery | 69 |
| Amputation | 15 |
| Other | 44 |
| Histologic necrosis ratio | |
| Less than 25% | 17 |
| 25%−50% | 16 |
| 51%−75% | 10 |
| Greater than 75% | 13 |
| Unknown | 76 |
| Survival time | |
| Within 1 year | 11 |
| 1–3 years | 39 |
| 3–5 years | 11 |
| More than 5 years | 68 |
Statistical analysis
Figure 1 depicts the results of Cox survival analysis, which identified independent risk factors for patients’ overall survival time (p ≤ 0.05) including:
Sex (female) (HR = 0.56, 95%CI: 0.33–0.95).
Race (HR = 1.19, 95%CI: 1.00-1.42).
Tumor-specific side (HR = 0.72, 95%CI: 0.52–0.99).
Tumor-specific region (HR = 0.64, 95%CI: 0.45–0.91).
Time of first recurrence (HR = 1.00, 95%CI: 1.00–1.00).
Figure 2 is the survival analysis plot of patient Cox regression.
Fig. 2.
Cox regression survival analysis gram
Figures 3, 4, 5, 6, 7, 8 shows the Kaplan-Meier survival analysis plots for sex, race, tumor-specific side, tumor-specific region, and time to first recurrence. Figure 3 showed longer OS in females than males (log-rank test, p = 0.005). Figure 4 (log-rank test, p = 0.1) showed no statistical difference between racial groups, although a trend toward different survival patterns was observed. Figure 5 (log-rank test, p = 0.2) showed a trend towards longer OS in patients with right-sided tumors. Figure 6 (log-rank test, p = 0.06) with pairwise comparison showed no statistical difference between proximal and distal locations. Figure 7 (log-rank test, p = 0.02) with pairwise comparison showed that patients without recurrence had significantly longer OS than those with recurrence at other sites. Figure 8 (log-rank test, p < 0.001) with pairwise comparison showed significant differences between patients with recurrence within 1 year and other groups.
Fig. 3.
Kaplan-meimer curve of patient's survival rate based on sex
Fig. 4.
Kaplan-meimer curve of patient's survival rate based on race
Fig. 5.
Kaplan-meimer curve of patients survival rate based on tumor-specific side
Fig. 6.
Kaplan-meimer curve of patients survival rate based on tumor-specific region
Fig. 7.
Kaplan-meimer curve of patients survival rate based on initial recurrence location
Fig. 8.
Kaplan-meimer curve of patients survival rate based on initial recurrence time
Figure 9 shows a nomogram for predicting 1-year, 3-year, and 5-year survival rates of patients. This visualization tool integrates the identified independent prognostic factors to generate individualized survival probability estimates. The nomogram allows clinicians to calculate a total score for each patient based on their specific clinical characteristics, which can then be translated into predicted survival probabilities at these key time points.
Fig. 9.
Prediction nomogram of 1-year, 3-year, and 5-year survival rates
Figures 10, 11, 12 shows the ROC curves of the nomogram predicting 1-year, 3-year and 5-year survival rates, with respective AUC values of 0.645 (95% CI: 0.475–0.816), 0.802 (95% CI: 0.725–0.880) and 0.787 (95% CI: 0.710–0.864).
Fig. 10.
ROC of predicting 1-year survival rate
Fig. 11.
ROC of predicting 3-year survival rate
Fig. 12.
ROC of predicting 5-year survival rate
Figures 13, 14, 15 display calibration plots for the nomogram’s predictions of 1-year, 3-year, and 5-year survival rates. These calibration curves demonstrate the agreement between predicted survival probabilities and actual observed outcomes. Notably, the model shows higher calibration accuracy for 3-year and 5-year survival rate predictions compared to the 1-year predictions, indicating that the nomogram performs particularly well for medium to long-term prognostic assessments.
Fig. 13.
Calibration plot for the nomogram’s prediction of 1-year survival rate
Fig. 14.
Calibration plot for the nomogram’s prediction of 3-year survival rate
Fig. 15.
Calibration plot for the nomogram’s prediction of 5-year survival rate
Figures 16, 17, 18 present the clinical decision curve analysis (DCA) for the nomogram’s predictions of 1-year, 3-year, and 5-year survival rates. The DCA curves demonstrate the net benefit of using the nomogram across different threshold probabilities, helping to evaluate its clinical utility. The analysis reveals that the nomogram offers particularly good clinical benefit for predicting 3-year and 5-year survival rates, suggesting that the model provides valuable prognostic information that could influence clinical decision-making for medium to long-term treatment planning.
Fig. 16.
Clinical decision curve analysis for the nomogram’s prediction of 1-year survival rates
Fig. 17.
Clinical decision curve analysis for the nomogram’s prediction of 3-year survival rates
Fig. 18.
Clinical decision curve analysis for the nomogram’s prediction of 5-year survival rates
Discussion
Pediatric osteosarcoma is a rare malignant tumor prone to invasive destruction and systemic metastasis, accounting for approximately 2% of all pediatric cancers [10, 11]. Although wide excision and neoadjuvant chemotherapy have significantly improved the prognosis of patients with extremity osteosarcoma [12]– [13], the rarity of pediatric osteosarcoma has resulted in a lack of large-scale population-based studies to evaluate its survival rates and prognostic factors. Therefore, identifying risk factors and prognostic determinants for pediatric osteosarcoma has important clinical value for early diagnosis, treatment, and prognostic prediction.
Multiple prognostic factors may influence survival in osteosarcoma, but these factors have not yet been integrated, and single prognostic indicators have limited accuracy in predicting patient outcomes. Nomograms are commonly used statistical methods that can achieve high robustness and precision in predicting patients’ overall survival probabilities. Song et al. [14] established a nomogram for predicting overall survival in pediatric osteosarcoma patients using the SEER database, identifying age, tumor site, and metastasis as important prognostic indicators. Zhang et al. [15] developed a nomogram to predict post-operative survival in osteosarcoma patients, which proved superior to the traditional AJCC staging system for 1-, 3-, and 5-year cancer-specific survival and overall survival. However, these findings have not been fully validated in other populations, potentially limiting their generalizability due to potential biases. While previous studies have explored survival predictions for various osteosarcoma populations, the current study specifically constructed and validated a prognostic model designed for pediatric osteosarcoma patients based on the TARGET database. This model not only considered surgical status as a factor but also integrated multiple independent prognostic variables including sex, race, tumor location, and recurrence patterns. Through this design, our model provides a more comprehensive assessment of survival probabilities for pediatric osteosarcoma patients, thereby offering more precise support for clinical decision-making.
The innovations of this study include: ①It considered not only common prognostic factors such as age, sex, and surgical status but also included factors such as tumor location and race, thus providing a more comprehensive prognostic assessment tool; ②It utilized pediatric-specific data from the TARGET database with internal validation through training and validation cohorts, further enhancing the model’s reliability. ③The model demonstrates good discrimination with C-indices of 0.802 and 0.787 for 3-year and 5-year survival predictions, respectively, which are comparable to or better than previously published nomograms for pediatric osteosarcoma.
Limitations of this study include: ①As a retrospective study based on the TARGET database, selection bias may exist; ②The TARGET database does not provide detailed information on chemotherapy regimens, radiation therapy, and molecular markers, limiting in-depth analysis of treatment effects; ③The lack of external validation is a significant limitation that should be addressed in future studies before clinical implementation; ④The events-per-variable ratio of 11.3, while exceeding the minimum threshold, is still relatively modest and may increase the risk of overfitting; ⑤Other variables not included in the model may influence prognosis, such as specific genetic mutations or molecular markers. Increasing research suggests that specific gene mutations or expression level changes are closely related to the biological behavior and prognosis of osteosarcoma. For example, abnormal expression of genes such as p53, RB1, and MDM2 has been confirmed to be associated with poor prognosis [16–18].
To further identify prognostic factors affecting pediatric osteosarcoma, future work is planned: ①Consider including more prognostic factors, such as molecular markers and gene expression profiles, to further improve the accuracy and personalization of the prediction model; ②Explore the impact of new treatment methods such as targeted therapy and immunotherapy on the prognosis of pediatric osteosarcoma as they develop, and incorporate them into the prognostic model; ③Conduct multicenter, prospective cohort studies to collect more detailed clinical and pathological information to validate and refine existing prediction models with external validation cohorts.
Clinical Implementation: To facilitate practical application of this nomogram, clinicians can use the following approach: (1) Collect the six required variables for each patient (sex, race, tumor-specific side, tumor-specific region, site of first recurrence, and time of first recurrence); (2) Assign points for each variable according to the nomogram scale; (3) Sum the points to obtain a total score; (4) Use the total score to read the corresponding survival probabilities at 1, 3, and 5 years from the nomogram. This tool can be particularly useful for risk stratification, treatment planning, and patient counseling. For example, patients identified as high-risk (low predicted survival probability) may benefit from more aggressive treatment protocols or enrollment in clinical trials. The nomogram can also facilitate shared decision-making by providing visual and quantitative survival estimates that can be discussed with patients and families.
Conclusion
This study successfully constructed and validated a predictive nomogram model for pediatric osteosarcoma based on the TARGET database. The model incorporates six key prognostic factors: sex, race, tumor-specific side, tumor-specific region, site of first recurrence, and time of first recurrence. This nomogram demonstrated good discrimination, calibration, and clinical utility, particularly for 3-year and 5-year survival predictions. The model not only provides an effective tool for clinicians to predict individual patient survival but also offers valuable references for treatment strategy optimization in pediatric osteosarcoma. By identifying patients at high risk for poor outcomes, this model may facilitate personalized treatment approaches, potentially improving survival rates and quality of life for pediatric osteosarcoma patients. However, external validation in independent cohorts is necessary before widespread clinical implementation.
Acknowledgements
We would like to thank the TARGET database for providing the data used in this study and all the patients who contributed to this database. We also acknowledge the statistical support provided by the Department of Biostatistics at Jinan University.
Author contributions
Jianfeng Li: Conceptualization of the study, research design, manuscript drafting, and critical revisionJiayi Li: Statistical analysis, data interpretation, and manuscript preparationJianjun Wang: Data collection, database management, and critical review of the manuscriptBingquan Li: Project supervision, funding acquisition, final approval of the manuscript, and overall scientific guidanceAll authors participated in drafting, reviewing, and approving the final manuscript. They agree to be accountable for all aspects of the research, ensuring its accuracy, integrity, and scientific rigor.
Funding
This study did not receive any funding in any form.
Data availability
The data that support the findings of this study are available from the Therapeutically Applicable Research to Generate Effective Treatments (TARGET) database (https://ocg.cancer.gov/programs/target). Restrictions apply to the availability of these data, which were used under license for the current study. Data are available from the authors upon reasonable request and with permission of TARGET.
Declarations
Ethics approval and consent to participate
This study used publicly available data from the TARGET database. All patient data were de-identified. The study was conducted in accordance with the Declaration of Helsinki. This study used de-identified data from the TARGET database.
Competing interests
The authors declare no competing interests.
Consent to publish
Not applicable. No individual patient data or images are presented in this manuscript.
.
Clinical trial number
Not applicable.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Ottaviani G, Jaffe N. The epidemiology of osteosarcoma. Cancer Treat Res. 2009;152:3–13. 10.1007/978-1-4419-0284-9_1. [DOI] [PubMed] [Google Scholar]
- 2.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. 10.1002/cncr.24121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Kager L, Zoubek A, Pötschger U, et al. Primary metastatic osteosarcoma: presentation and outcome of patients treated on neoadjuvant cooperative osteosarcoma study group protocols. J Clin Oncol. 2003;21(10):2011–8. 10.1200/JCO.2003.08.132. [DOI] [PubMed] [Google Scholar]
- 4.Bielack SS, Kempf-Bielack B, Delling G, et al. Prognostic factors in high-grade osteosarcoma of the extremities or trunk: an analysis of 1,702 patients treated on neoadjuvant cooperative osteosarcoma study group protocols. J Clin Oncol. 2002;20(3):776–90. 10.1200/JCO.2002.20.3.776. [DOI] [PubMed] [Google Scholar]
- 5.Isakoff MS, Bielack SS, Meltzer P, Gorlick R. Osteosarcoma: current treatment and a collaborative pathway to success. J Clin Oncol. 2015;33(27):3029–35. 10.1200/JCO.2014.59.4895. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Iasonos A, Schrag D, Raj GV, Panageas KS. How to build and interpret a nomogram for cancer prognosis. J Clin Oncol. 2008;26(8):1364–70. 10.1200/JCO.2007.12.9791. [DOI] [PubMed] [Google Scholar]
- 7.Balachandran VP, Gonen M, Smith JJ, DeMatteo RP. Nomograms in oncology: more than Meets the eye. Lancet Oncol. 2015;16(4):e173–80. 10.1016/S1470-2045(14)71116-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Wang Y, Li J, Xia Y, et al. Prognostic nomogram for intrahepatic cholangiocarcinoma after partial hepatectomy. J Clin Oncol. 2013;31(9):1188–95. 10.1200/JCO.2012.41.5984. [DOI] [PubMed] [Google Scholar]
- 9.Peduzzi P, Concato J, Kemper E, Holford TR, Feinstein AR. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol. 1996;49(12):1373–9. 10.1016/s0895-4356(96)00236-3. [DOI] [PubMed] [Google Scholar]
- 10.Gatta G, Botta L, Rossi S, et al. Childhood cancer survival in Europe 1999–2007: results of EUROCARE-5–a population-based study. Lancet Oncol. 2014;15(1):35–47. 10.1016/S1470-2045(13)70548-5. [DOI] [PubMed] [Google Scholar]
- 11.Ward E, DeSantis C, Robbins A, Kohler B, Jemal A. Childhood and adolescent cancer statistics, 2014. CA Cancer J Clin. 2014;64(2):83–103. 10.3322/caac.21219. [DOI] [PubMed] [Google Scholar]
- 12.Meyers PA, Schwartz CL, Krailo MD, et al. Osteosarcoma: the addition of muramyl tripeptide to chemotherapy improves overall survival–a report from the children’s oncology group. J Clin Oncol. 2008;26(4):633–8. 10.1200/JCO.2008.14.0095. [DOI] [PubMed] [Google Scholar]
- 13.Whelan JS, Bielack SS, Marina N, et al. EURAMOS-1, an international randomised study for osteosarcoma: results from pre-randomisation treatment. Ann Oncol. 2015;26(2):407–14. 10.1093/annonc/mdu526. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Song K, Song J, Chen F, Lin K, Ma X, Jiang J. Prognostic nomograms for predicting overall and cancer-specific survival of high-grade osteosarcoma patients. J Bone Oncol. 2018;13:106–13. 10.1016/j.jbo.2018.09.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Zhang C, Guo X, Xu Y, et al. Lung metastases at the initial diagnosis of high-grade osteosarcoma: prevalence, risk factors and prognostic factors. A large population-based cohort study. Sao Paulo Med J. 2019;137(5):423–9. 10.1590/1516-3180.2018.0381120619. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Miller CW, Aslo A, Won A, Tan M, Lampkin B, Koeffler HP. Alterations of the p53, Rb and MDM2 genes in osteosarcoma. J Cancer Res Clin Oncol. 1996;122(9):559–65. 10.1007/BF01213553. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Wadayama B, Toguchida J, Shimizu T, et al. Mutation spectrum of the retinoblastoma gene in osteosarcomas. Cancer Res. 1994;54(11):3042–8. [PubMed] [Google Scholar]
- 18.Onda M, Matsuda S, Higaki S, et al. ErbB-2 expression is correlated with poor prognosis for patients with osteosarcoma. Cancer. 1996;77(1):71–8. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data that support the findings of this study are available from the Therapeutically Applicable Research to Generate Effective Treatments (TARGET) database (https://ocg.cancer.gov/programs/target). Restrictions apply to the availability of these data, which were used under license for the current study. Data are available from the authors upon reasonable request and with permission of TARGET.