A diagnostic nomogram for synchronous lung metastasis at initial diagnosis in osteosarcoma: A retrospective SEER database study

Abstract

Objective

This study aimed to establish a diagnostic nomogram for identifying synchronous lung metastasis at initial diagnosis in osteosarcoma patients, and to descriptively analyze overall survival patterns in patient subgroups.

Methods

A total of 1149 eligible osteosarcoma cases diagnosed between 2010 and 2015 were retrieved from the Surveillance, Epidemiology, and End Results (SEER) database. Candidate predictors were screened by univariate logistic regression (p < 0.10) and entered into backward stepwise multivariate logistic regression (retention p < 0.05) to construct a diagnostic nomogram. Discrimination was evaluated by the area under the receiver operating characteristic curve (AUC), calibration by calibration plot, and clinical utility by decision curve analysis (DCA) with bootstrapping validation. Secondary survival analysis used Kaplan-Meier curves and log-rank tests to describe overall survival patterns.

Results

Synchronous lung metastases were found in 213 patients (18.5%) at diagnosis. The nomogram incorporated age, race, sex, lymph node stage, grade, tumour size, tumour site, and histology. It achieved an AUC of 0.754 (95% confidence interval 0.719 - 0.799) and a bootstrap-calibrated C-index of 0.745. Calibration was satisfactory, and DCA demonstrated superior net benefit compared with traditional staging approaches, including AJCC stage, grade alone, size alone, and N stage alone. In exploratory survival analysis, patients with synchronous metastasis had significantly poorer overall survival, and primary tumour surgery was associated with prolonged median survival.

Conclusion

The diagnostic nomogram, incorporating tumour grade, site, size, and lymph node stage, provides a reliable tool for estimating the probability of synchronous lung metastasis at initial diagnosis, which may assist clinicians in guiding initial staging intensity and treatment planning. However, owing to the cross-sectional design of the SEER database, this model is not designed for predicting future metastatic progression.

Keywords

nomogram osteosarcoma synchronous metastasis diagnostic prediction model SEER database

Introduction

Although osteosarcoma is a rare disease, it is the most common primary bone malignancy that occurs during childhood and adolescence.^1–4 The morbidity due to osteosarcoma is estimated to be 0.2 - 3 patients/100, 000 people annually.⁵ Approximately 20% cases of osteosarcoma have synchronous lung metastases at the time of diagnosis.^6,7 Synchronous metastases is associated with a high mortality rate and can often be misinterpreted as trauma or growing pains, leading to neglect at its onset.⁸ Furthermore, approximately 50% of patients initially presenting without lung metastases eventually develop it during chemotherapy. Despite the development of numerous new therapies for the treatment of osteosarcoma, 30%–40% cases suffer from recurrence, with long-time, post-relapse survival of <20%.^9,10 The early diagnosis and treatment of osteosarcoma is beneficial. Therefore, identifying synchronous metastasis of osteosarcoma at the time of diagnosis is crucial.

The American Joint Committee on Cancer and Enneking classification systems have been widely utilised for the classification of osteosarcoma based on primary staging according to the presence of metastasis and histological grade at diagnosis.^11,12 In addition to factors adopted in the clinical classification, numerous clinical factors such as age,¹³ tumour site,¹¹ and histological type¹⁴ have also been identified as the diagnostic indicators for synchronous metastasis at diagnosis. No single factor can precisely identify synchronous metastasis. To achieve accurate diagnostic prediction at presentation, multiple predictors must be integrated using statistical prediction models.^15–17 Nomograms are new approaches in addition to conventional classification systems to assist in diagnostic decision-making at initial presentation for different tumours.^18–20 Several nomograms are being developed for soft tissue sarcoma^21,22 and osteosarcoma.^23–25

The primary objective of this study was to develop a diagnostic nomogram for identifying synchronous lung metastasis at initial diagnosis in osteosarcoma patients using baseline clinical features available at presentation. The secondary objective was to descriptively analyze overall survival patterns in patient subgroups with and without synchronous metastasis, entirely separate from the diagnostic prediction model. Although survival analysis is methodologically independent from the diagnostic model, descriptive characterization of survival patterns provides essential clinical context for interpreting the diagnostic tool’s utility in practice. This study aims to provide a tool to assist clinicians in determining the necessity of comprehensive staging workup at initial presentation (e.g., chest CT evaluation), not to predict future metastatic progression. Therefore, the diagnostic factors for synchronous lung metastasis in osteosarcoma were examined in a large population-based cohort, and survival patterns were characterized descriptively.

Materials and methods

Patients

The Surveillance, Epidemiology, and End Results (SEER) database, the largest oncology patient database in the USA, was used to collect data for this retrospective study. Data were extracted using the SEER*stat 8.3.5 software from the official website: https://seer.cancer.gov25. Patients diagnosed with osteosarcoma between 2010 and 2015 were identified using the following ICD-O-3 histology codes: 9180/3 (osteosarcoma, NOS), 9181/3 (chondroblastic osteosarcoma), 9182/3 (fibroblastic osteosarcoma), 9183/3 (telangiectatic osteosarcoma), 9184/3 (osteoblastoma-like), 9185/3 (small cell osteosarcoma), 9186/3 (central osteosarcoma), 9192/3 (parosteal osteosarcoma), 9193/3 (periosteal osteosarcoma), 9194/3 (high-grade surface osteosarcoma), and 9195/3 (intramedullary well-differentiated osteosarcoma). Patients with available data were included in the study. The following variables were collected for each patient: sex (male, female); age (≤24, 25–59, ≥ 60 years); race [white, black, Asian or Pacific Islander (API), American Indian/Alaska Native (AI)]; metastasis status; tumor size (≤8 cm, > 8 cm); tumor site (axial skeleton, spine, ribs, pelvis; extremities: long and short bones of the four extremities); histology (osteosarcoma NOS [ICD-O-3 codes 9180-9186], Parosteal osteosarcoma [ICD-O-3 code 9192]); regional lymph node stage (N0, N1); and tumor grade, categorized as highly and moderately differentiated (ICD-O-3 Grades 1/2) for low-grade tumors, or poorly differentiated or undifferentiated (ICD-O-3 Grades 3/4) for high-grade tumors. SEER records metastasis status only at diagnosis, precluding the analysis of metachronous metastases that develop during follow-up. Cases with missing metastasis status, tumor size, grade, or N stage were excluded from the analysis. Age was categorized into ≤24, 25–59, and ≥60 years based on clinical relevance to osteosarcoma epidemiology and consistent with prior SEER-based osteosarcoma studies. Tumor size was dichotomized at 8 cm based on clinical convention, prior literature, and ROC-derived optimal threshold using Youden index.

Construction of the nomogram

The present study conducted univariate nonadjusted logistic regression for evaluating diverse parameters such as sex, age, race, tumour size, tumour site, grade, histology, and N stage to identify factors associated with synchronous distant metastasis (DM) at diagnosis in patients with osteosarcoma. Logistic regression was selected as the appropriate statistical method for this diagnostic aim, given the cross-sectional nature of SEER metastasis data. Prognostic prediction of future metastatic progression was not attempted due to the absence of time-to-metastasis recording in the database. All candidate predictors were first assessed using univariate analysis, and variables with a liberal threshold of p < 0.10 were considered for multivariate analysis. These candidates were then entered into a multivariate logistic regression model using backward stepwise elimination with a retention criterion of p < 0.05. The final independent predictors were used to construct the nomogram, which visualised the prediction model based on the obtained regression coefficients.

Validation of the nomogram

The area under the receiver operating characteristic curve (AUC) was used to evaluate the nomogram discriminative ability for synchronous metastasis status at diagnosis. The AUC was reported as the primary discrimination metric for this diagnostic model. The concordance index (C-index) was reported secondarily for completeness but should be interpreted in the diagnostic rather than prognostic context. Bootstrapping validation was performed using 1, 000 bootstrap resamples to calculate the corrected C-index. Our diagnostic nomogram was then calibrated and validated with a calibration curve based on 1, 000 bootstrap resamples. The clinical utility of the nomogram was assessed and compared with four traditional staging approaches: AJCC stage (I/II vs. III/IV), tumor grade alone (I–II vs. III–IV), tumor size alone (≤8 cm vs. >8 cm), and N stage alone (N0 vs. N1). For each traditional approach, the predicted probability of synchronous metastasis was calculated based on the observed prevalence in each category, and DCA was performed to compare net clinical benefit across threshold probabilities. R software (version 3.5.1) and SPSS23.0 (IBM Corporation, Armonk, NY, USA) were utilised for data analysis. A P value < 0.05 (by two-tailed test) was considered statistically significant.

Secondary survival analysis (exploratory, descriptive only)

Distinct from the primary diagnostic prediction model, the following survival analyses were performed as exploratory secondary analyses to describe baseline survival patterns in the study cohort. These analyses do not establish prognostic prediction models due to the retrospective cross-sectional design without time-to-metastasis data. The present study regarded overall survival (OS) as secondary descriptive endpoint. OS indicated the duration between the initial diagnosis of osteosarcoma and cancer-related death. On the other hand, difference in survival was analysed using log-rank tests and Kaplan–Meier (K–M) curves.

Results

Patient features and univariate logistic regression

The present study enrolled 1149 eligible cases, of which 213 (18.5%) cases had synchronous metastasis at diagnosis and 936 (81.50%) had no detectable metastasis at diagnosis. Patients in the ≥ 60 years group exhibited a markedly increased synchronous metastasis rate at diagnosis compared with those in the two < 60 years groups (χ² = 12.250; p = 0.002). Table 1 presents the clinical and demographic data of all cases. In the present study, the univariate nonadjusted logistic regression was conducted to analyse several variables, including sex, age, race, grade, tumour site, N stage, surgery, and histology. Variables including the primary tumour size [odds ratio (OR) = 4.94; 95% confidence interval (CI) = 1.01 - 2.01; p < 0.001], axial location [OR = 2.7589; CI = 1.8202 - 4.2408; p = 0.000], high-grade tumours (OR = 5.5921; 95% CI = 3.3336 - 10.0807; p < 0.001), and high regional lymph node (N) stage (OR = 6.1521; 95% CI = 2.5759 - 17.0128; p < 0.001) were positively correlated with the presence of synchronous metastases at diagnosis. Thereafter, these significant diagnostic predictors were included in the multivariable analysis.

Table 1.

Baseline characteristics of 1149 osteosarcoma patients stratified by synchronous lung metastasis status at diagnosis.

Characteristics	Non-metastasis	Metastasis	p	Univariate regression analysis	p
Characteristics	Non-metastasis	Metastasis	p	OR (95% CI)	p
Subjects	936 (81.5)	213 (18.5)
Age			0.002
0-24 years	518 (79.6)	133 (20.4)		Reference
25-59 years	294 (87.5)	42 (12.5)		1.5464 (1.1665, 2.0478)	0.0023
≥60 years	124 (76.5)	38 (23.5)		5.8285 (4.0317, 8.5251)	0.0000
Race			0.409
Black	146 (78.1)	41 (21.9)		Reference
White	700 (82.3)	151 (17.7)		0.7952 (0.5761, 1.1020)	0.166
Others	90 (81.1)	21 (18.9)		0.6726 (0.4071, 1.0988)	0.117
Gender			0.075
Female	441 (83.7)	86 (16.3)		Reference
Male	495 (79.6)	127 (20.4)		1.1743 (0.9215, 1.4979)	0.194
Grade			0.000
Grade I-II	139 (96.5)	5 (3.5)		Reference
Grade III-IV	797 (79.3)	208 (20.7)		5.5921 (3.3336, 10.0807)	0.0000
Tumor size (cm)			0.000
≤ 80	382 (89.9)	43 (10.1)		Reference
>80	554 (68.1)	170 (23.5)		4.4255 (2.8940, 6.8355)	0.0000
N stage			0.0000
N0	898 (83.1)	182 (16.9)		Reference
N1	38 (51.0)	31 (49.0)		6.1521 (2.5759, 17.0128)	0.0001
Histology			0.001
Osteosarcoma NOS (9180-9186)	864 (80.5)	210 (19.5)		Reference
Parosteal	72 (96.0)	3 (4.0)		0.1175 (0.0409, 0.2657)	0.0000
Tumor site			0.001
Extremity	651 (79.4)	169 (21.6)		Reference
Axial	275 (83.6)	54 (16.4)		2.7589 (1.8202, 4.2408)	0.0000

Note: OR = odds ratio; CI = confidence interval.

Multivariate analysis and nomogram

Table 2 displays the multivariate analysis of the significant variables such as sex, age, race, tumour site, tumor size, histology, grade, N stage. The nomogram in this study was developed based on these identified independent variables (Figure 1). Each variable in the constructed nomogram was assigned a corresponding score using a point scale. The total score, derived from the bottom scale, was then summed to estimate the probability of synchronous DM at diagnosis in osteosarcoma cases. This represents the diagnostic probability of already-present metastasis, not the future risk of metastatic progression.

Table 2.

Predictive logistic multivariate regression model parameters.

Intercept and covariate	β	p	SE	OR	95% CI
Intercept	-2.7251	0.0000	0.700	0.1110	0.04434, 0.2669
Age
0-24 years	Reference				Reference
25-59 years	0.4878	0.0044	0.187	1.6287	1.1657, 2.2874
≥60years	1.6388	0.0000	0.421	5.1491	3.2765, 8.2606
Race
Black	Reference				Reference
White	-0.2242	0.2421	0.192	0.7991	0.5498, 1.1671
Others	-0.4457	0.1239	0.289	0.6403	0.3602, 1.1241
Gender
Female	Reference				Reference
Male	0.1401	0.3327	0.145	1.1708	0.8813, 1.5581
Grade
Grade I-II	Reference				Reference
Grade III-IV	1.6939	0.0000	0.435	5.4453	2.9546, 10.6931
Tumor size (cm)
≤80	Reference				Reference
>80	1.0177	0.0000	0.262	2.7360	1.6312, 4.6087
N stage
N0	Reference				Reference
N1	1.4031	0.0067	0.517	3.9017	1.4650, 11.6709
Histology
Osteosarcoma, NOS	Reference				Reference
Parosteal	-1.3088	0.0137	0.532	0.2661	0.0831-0.6962
Tumor site
Extremity	Reference				Reference
Axial	0.7364	0.0041	0.256	2.0786	1.2629, 3.4623

Note: β, regression coefficient; CI, confidence interval.

Figure 1.

Nomogram for osteosarcoma patients to predict the risk of metastasis.

Validation and performance of the prediction nomogram

For the diagnostic prediction of synchronous metastasis status at diagnosis, our nomogram had a C-index value of 0.759 (95% CI: 0.719 - 0.799) in predicting metastasis, and this value was then calibrated by bootstrapping validation as 0.745. Additionally, the ROC curve was plotted (Figure 2(a)) with the AUC value of 0.754 for discriminating synchronous metastasis status at diagnosis, demonstrating favourable discriminating ability. The calibration curve was highly consistent between the nomogram-predicted and measured results in estimating synchronous metastasis probability at diagnosis among osteosarcoma cases (Figure 2(b)). DCA is a new approach that is utilised for evaluating the prediction models. The DCA demonstrated the satisfying net benefits of our nomogram across many threshold probabilities. Our constructed nomogram exhibited better net clinical benefits in the diagnostic identification of synchronous metastasis among osteosarcoma cases than the four traditional staging approaches (AJCC stage, tumor grade alone, tumor size alone, and N stage alone) (Figure 2(c)). Specifically, the nomogram demonstrated superior net benefit across threshold probabilities of 10%-60%, whereas individual staging variables showed limited or no net benefit in this range.

Figure 2.

The performance evaluation of the nomogram in predicting lung metastasis of osteosarcoma. (a) ROC curve demonstrating the discriminative ability of the nomogram. (b) Calibration curve comparing the predicted and observed metastasis probabilities. (c) Decision curve analysis (DCA) showing the clinical net benefit of the nomogram compared to conventional prediction methods.

Survival analysis for metastases

As an exploratory secondary analysis distinct from the primary diagnostic model, K-M analysis exhibited decreased OS in patients with synchronous metastasis at diagnosis (Figure 3(a); p < 0.001), elder patients (Figure 3(b); p < 0.001), those with axial location (Figure 3(c); p < 0.001), those with high-grade osteosarcoma (Figure 3(d); p < 0.001). Furthermore, OS in patients who received surgery at the primary site was significantly higher than those who did not receive surgery (Figure 3(e); p < 0.001). However, this observation must be interpreted with extreme caution due to inherent treatment-selection bias: patients without detectable metastasis at diagnosis are more likely to undergo definitive surgery, while those with extensive synchronous metastasis may be deemed unresectable. Causal inference about surgery’s survival benefit cannot be drawn from this retrospective analysis.

Figure 3.

Kaplan-Meier survival curves of Osteosarcoma for clinical risk factors. (a) Metastasis, (b) Age, (c) Tumor_site, (d) Grade, (e) surgery.

Discussion

This study developed and internally validated a diagnostic nomogram for synchronous lung metastasis at initial diagnosis in osteosarcoma using a population-based SEER cohort of 1149 patients. The overall prevalence of synchronous metastasis was 18.5%, a rate lower than that reported in a single-centre study (29.5%),²⁶ likely reflecting differences in referral patterns and disease severity between population-based and hospital-based cohorts. The model incorporated age, race, sex, tumour size, tumour site, histology, grade, and N stage, and demonstrated moderate discriminative ability with an AUC of 0.754 (95% CI 0.719–0.799) and a bootstrap-calibrated C-index of 0.745. Calibration was satisfactory, and decision curve analysis indicated consistent net clinical benefit across threshold probabilities of 10%–60%, outperforming individual traditional staging variables including AJCC stage, tumour grade alone, tumour size alone, and N stage alone. However, the moderate AUC suggests that critical predictive information is not captured by SEER variables; notably, serum alkaline phosphatase, MRI-defined soft tissue extension, response to neoadjuvant chemotherapy, and detailed lung nodule characteristics (number, size, laterality) are all absent from the database yet likely contribute substantially to metastasis risk. Future models incorporating such variables may achieve substantially higher discriminative performance.

Our findings align with and extend the existing literature on clinical predictors of synchronous metastasis in osteosarcoma. Advanced age (≥60 years) was associated with a markedly increased synchronous metastatic rate, consistent with the distinct biological behaviour of osteosarcoma in older adults, in whom the disease often presents as a secondary malignancy with more aggressive features. Neither race nor sex was significantly associated with synchronous metastasis prevalence in our cohort, a finding that concords with prior SEER-based studies.^27,28 Some single-centre series have reported higher prevalence in males and in tumours of the tibia and femur²⁹; these discrepancies likely reflect the fact that these are the most common primary sites rather than a true sex- or site-specific biological predisposition. Large tumour size and high tumour grade each independently predicted synchronous metastasis in our model, corroborating evidence from Munajat et al.³⁰ and Zheng et al.,³¹ and consistent with the biological paradigm that larger, high-grade tumours have undergone more cell divisions and acquired a greater propensity for haematogenous dissemination.^32,33 Tumour site and N stage further refined risk stratification: axial tumours carried a relatively higher probability of synchronous metastasis despite their lower absolute frequency, likely attributable to delayed detection due to deep anatomical location and proximity to vital vascular structures facilitating early haematogenous spread. The survival implications of these diagnostic factors were reflected in descriptive analyses: patients with synchronous metastasis experienced markedly poorer overall survival, and those with axial tumours and higher stage fared worst, in agreement with Lin et al.³⁴ and other reports.³⁵ It is important to distinguish this diagnostic tool from models predicting metachronous metastasis during chemotherapy. Approximately 50% of initially non-metastatic patients develop lung metastases during neoadjuvant treatment, yet SEER’s cross-sectional recording precludes analysis of this dynamic process. Consequently, clinicians must not apply this nomogram to guide surveillance intensity during follow-up or adjuvant therapy decisions. Compared with the existing osteosarcoma nomogram by Wang et al., which predicted cancer-specific survival,³⁶ our model differs fundamentally in aim (diagnostic vs. prognostic) and outcome (synchronous metastasis at diagnosis vs. survival), precluding direct performance comparison. To our knowledge, this is the first diagnostic nomogram specifically designed for identifying synchronous lung metastasis at initial diagnosis in osteosarcoma.

The intended clinical application of this nomogram is to guide the intensity of initial staging at the time of osteosarcoma diagnosis, not to predict future events. In resource-limited settings or emergency departments, a patient with a high predicted probability—based on age ≥60 years, axial tumour site, tumour size >8 cm, and high grade—may be prioritised for urgent chest CT staging rather than routine scheduling. However, the nomogram is not a replacement for imaging confirmation, nor is it designed for screening asymptomatic populations or for counselling about future metastatic risk. The decision to pursue aggressive surgical resection in patients with synchronous metastasis should be made within a multidisciplinary tumour board, with full acknowledgement that the observed survival benefit of surgery in this retrospective cohort may be largely attributable to selection of patients with favourable disease profiles.

Several limitations warrant careful consideration. First, the retrospective design of the SEER database introduces inherent confounding and selection biases. Second, and critically, SEER records metastasis status as a cross-sectional variable at diagnosis only; it does not capture the timing of metastasis detection or subsequent metachronous progression. Consequently, the nomogram estimates the probability of already-present synchronous metastasis and must not be applied to predict future metastatic risk. Third, only internal bootstrap validation was performed; no independent external cohort was available. External validation in a multi-institutional, prospective setting is essential before clinical implementation. Fourth, SEER lacks several clinically relevant variables known to influence metastasis risk and survival, including response to neoadjuvant chemotherapy, specific surgical margins, serum ALP and LDH levels, detailed lung nodule characteristics on imaging, and performance status and comorbidity data—all of which may have contributed to the moderate discriminative performance of our model. Fifth, the study population was derived from the United States, and generalizability to other populations, particularly in Asia, remains to be established. Despite these constraints, the nomogram provides a useful, transparent framework that can be refined as more granular data become available.

Conclusion

In conclusion, the diagnostic nomogram established in this study, incorporating tumour grade, site, size, and lymph node stage, offers a reliable method for estimating the probability of synchronous lung metastasis at initial diagnosis in osteosarcoma patients. Within the constraints of cross-sectional SEER data, this tool may assist clinicians in guiding initial staging intensity and treatment planning at presentation. It is not a prognostic tool and should not be used to predict future metastatic progression. Prospective studies with longitudinal metastasis monitoring are warranted to develop prognostic models for metachronous metastasis.

Footnotes

ORCID iD

Yanli Niu

Ethical considerations

The current analysis did not involve individual participant data. Consequently, separate ethical approval is not required.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Program Project of Science and Technology Innovation Guidance of Zhengzhou City; 2024YLZDJH175, Key Scientific Research Project of Colleges and Universities in Henan Province; 26A310002, the Plan of Science and Technology of Kaifeng City; 2503167.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.*

Generative AI statement

The authors declare that no Generative AI was used in the creation of this manuscript.

References

Gross

Hameed

. Osteosarcoma. Surg Pathol Clin 2025; 18(3): 449–470. https://doi.org/10.1016/j.path.2025.01.009

Beird

Bielack

Flanagan

, et al. Osteosarcoma. Nat Rev Dis Primers 2022; 8(1): 77. https://doi.org/10.1038/s41572-022-00409-y

Yan

Dong

Chen

. Advances in osteosarcoma research: Pathogenesis and emerging therapeutic strategies. Crit Rev Oncol Hematol 2026; 222: 105252. https://doi.org/10.1016/j.critrevonc.2026.105252

Meltzer

Helman

. New horizons in the treatment of osteosarcoma. N Engl J Med 2021; 385(22): 2066–2076. https://doi.org/10.1056/NEJMra2103423

Strauss

Frezza

Abecassis

, et al. Bone sarcomas: ESMO-EURACAN-GENTURIS-ERN PaedCan Clinical Practice Guideline for diagnosis, treatment and follow-up. Ann Oncol 2021; 32(12): 1520–1536. https://doi.org/10.1016/j.annonc.2021.08.1995

Silva

JAM

Marchiori

Macedo

, et al. Pulmonary metastasis of osteosarcoma: multiple presentations in a single patient. J Bras Pneumol 2022; 48(2): e20210478. https://doi.org/10.36416/1806-3756/e20210478

Bacci

Rocca

Salone

, et al. High grade osteosarcoma of the extremities with lung metastases at presentation: treatment with neoadjuvant chemotherapy and simultaneous resection of primary and metastatic lesions. J Surg Oncol 2008; 98(6): 415–420. https://doi.org/10.1002/jso.21140

Zhang

. Survival of patients with primary osteosarcoma and lung metastases. J BUON 2018; 23(5): 1500–1504.

Zhang

Yang

Zhao

, et al. Progress in the chemotherapeutic treatment of osteosarcoma. Oncol Lett 2018; 16(5): 6228–6237. https://doi.org/10.3892/ol.2018.9434

10.

Mettmann

Baumhoer

Bielack

, et al. Solitary pulmonary metastases at first recurrence of osteosarcoma: Presentation, treatment, and survival of 219 patients of the Cooperative Osteosarcoma Study Group. Cancer Med 2023; 12(17): 18219–18234. https://doi.org/10.1002/cam4.6409

11.

Kundu

. Classification, imaging, biopsy and staging of osteosarcoma. Indian J Orthop 2014; 48(3): 238–246. https://doi.org/10.4103/0019-5413.132491

12.

Alexander

Binitie

Letson

, et al. Osteosarcoma: An evolving understanding of a complex disease. J Am Acad Orthop Surg 2021; 29(20): e993–e1004. https://doi.org/10.5435/JAAOS-D-20-00838

13.

Harting

Lally

Andrassy

, et al. Age as a prognostic factor for patients with osteosarcoma: an analysis of 438 patients. J Cancer Res Clin Oncol 2010; 136(4): 561–570. https://doi.org/10.1007/s00432-009-0690-5

14.

Klein

Siegal

. Osteosarcoma: anatomic and histologic variants. Am J Clin Pathol 2006; 125(4): 555–581. https://doi.org/10.1309/UC6K-QHLD-9LV2-KENN

15.

Zhao

Xia

, et al. Feasibility of machine learning-based modeling and prediction to assess osteosarcoma outcomes. Sci Rep 2025; 15(1): 17386. https://doi.org/10.1038/s41598-025-00179-z

16.

Kirshtein

, et al. Data-driven mathematical model of osteosarcoma. Cancers (Basel) 2021; 13(10): 2367. https://doi.org/10.3390/cancers13102367

17.

Zhao

Meng

Dai

, et al. Prediction of patients with high-risk osteosarcoma on the basis of XGBoost algorithm using transcriptome and methylation data from SGH-OS cohort. JCO Precis Oncol 2025; 9: e2400732. https://doi.org/10.1200/PO-24-00732

18.

Wang

, et al. Nomograms predicting long-term overall survival and cancer-specific survival in head and neck squamous cell carcinoma patients. Oncotarget 2016; 7(32): 51059–51068. https://doi.org/10.18632/oncotarget.10595

19.

Winarno

GNA

Harsono

Suardi

, et al. Nomogram development for predicting ovarian tumor malignancy using inflammatory biomarker and CA-125. Sci Rep 2024; 14(1): 15790. https://doi.org/10.1038/s41598-024-66509-9

20.

Shang

Zhang

, et al. A nomogram incorporating treatment data for predicting overall survival in gastroenteropancreatic neuroendocrine tumors: a population-based cohort study. Int J Surg 2024; 110(4): 2178–2186. https://doi.org/10.1097/JS9.0000000000001080

21.

Wang

Guo

Zhang

, et al. Prediction of soft tissue sarcoma grading using intratumoral habitats and a peritumoral radiomics nomogram: a multi-center preliminary study. Front Oncol 2024; 14: 1433196. https://doi.org/10.3389/fonc.2024.1433196

22.

Han

, et al. A novel nomogram and prognostic factor for metastatic soft tissue sarcoma survival. Front Endocrinol (Lausanne) 2024; 15: 1371910. https://doi.org/10.3389/fendo.2024.1371910

23.

Liu

Xie

Zhang

, et al. Survival nomogram for osteosarcoma patients: SEER data retrospective analysis with external validation. Am J Cancer Res 2023; 13(3): 900–911.

24.

Yang

Weng

, et al. Nomogram for predicting disease-specific survival in osteosarcoma. Chin Med J (Engl) 2022; 135(9): 1126–1128. https://doi.org/10.1097/CM9.0000000000001837

25.

Yan

Yang

, et al. Constructing a nomogram for predicting pelvic osteosarcoma: A retrospective study based on the SEER database and a Chinese cohort. Br J Hosp Med (Lond) 2024; 85(8): 1–17. https://doi.org/10.12968/hmed.2024.0339

26.

Buddingh

Anninga

Versteegh

, et al. Prognostic factors in pulmonary metastasized high-grade osteosarcoma. Pediatr Blood Cancer 2010; 54(2): 216–221. https://doi.org/10.1002/pbc.22293

27.

Liu

Cui

, et al. Epidemiology and nomogram of pediatric and young adulthood osteosarcoma patients with synchronous lung metastasis: A SEER analysis. PLoS One 2023; 18(7): e0288492. https://doi.org/10.1371/journal.pone.0288492

28.

Liu

Tang

Peng

, et al. Lung and bone metastases patterns in limb osteosarcoma: Surgical treatment of primary site improves overall survival. Medicine (Baltimore) 2023; 102(42): e35671. https://doi.org/10.1097/MD.0000000000035671

29.

Estrada-Villaseor

Escamilla-Uribe

De la Garza-Montano

, et al. Association of metastasis with clinicopathological data in Mexican patients with osteosarcoma, giant cell tumor of bone and chondrosarcoma. Asian Pac J Cancer Prev 2015; 16(17): 7689–7694. https://doi.org/10.7314/apjcp.2015.16.17.7689

30.

Munajat

Zulmi

Norazman

, et al. Tumour volume and lung metastasis in patients with osteosarcoma. J Orthop Surg (Hong Kong) 2008; 16(2): 182–185. https://doi.org/10.1177/230949900801600211

31.

Zheng

Chen

Wang

, et al. A clinical prediction model for lung metastasis risk in osteosarcoma: A multicenter retrospective study. Front Oncol 2023; 13: 1001219. https://doi.org/10.3389/fonc.2023.1001219

32.

Tirtei

Michelsen

Haveman

, et al. Prognostic factors in newly diagnosed high-grade osteosarcoma-A systematic review. Cancer Med 2025; 14(14): e71044. https://doi.org/10.1002/cam4.71044

33.

Ripley

Rusch

. Lung metastases. Lung metastases. Abeloff's clinical oncology. Churchill Livingstone, 2014, pp. 764–777.

34.

Lin

Deng

Zhang

, et al. The extremity localized classic osteosarcomas have better survival than the axial non-classics. World J Surg Oncol 2018; 16(1): 39. https://doi.org/10.1186/s12957-018-1344-3

35.

Xin

Wei

. Prognostic factors in osteosarcoma: A study level meta-analysis and systematic review of current practice. J Bone Oncol 2020; 21: 100281. https://doi.org/10.1016/j.jbo.2020.100281

36.

Wang

Zhanghuang

Tan

, et al. A nomogram for predicting cancer-specific survival of osteosarcoma and ewing’s sarcoma in Children: A SEER database analysis. Front Public Health 2022; 10: 837506. https://doi.org/10.3389/fpubh.2022.837506