Nomogram for Predicting Disease-Specific Survival in Osteosarcoma

Nomogram for Predicting Disease-Specific Survival in Osteosarcoma

Osteosarcoma (OS) is the most common malignant primary bone tumor, predominantly affecting children, adolescents, and young adults. Statistical studies indicate that up to 20% of OS patients present with clinically detectable metastatic tumors, with over 85% of these metastases occurring in the lungs. Identifying risk factors that reflect the biological characteristics and survival of OS patients is crucial for developing better interventions. Nomograms have emerged as a viable predictive model, offering prognostic probabilities for individual patients. Unlike the traditional tumor-node metastasis (TNM) staging system, nomograms integrate multiple prognostic factors, providing a more precise evaluation of prognosis. This study aims to develop and validate a nomogram and risk model for predicting disease-specific survival (DSS) and risk stratification in OS patients.

Patients diagnosed with OS between 2004 and 2015 were obtained from the Surveillance, Epidemiology, and End Results Program (SEER) database. Inclusion criteria included a pathological diagnosis of OS between 2004 and 2015 and OS as the first and only malignant cancer diagnosis. Patients with unknown demographic, clinicopathological, or therapeutic follow-up information or those who died from other causes were excluded. A total of 1639 patients were included in the analysis. Variables extracted included age at diagnosis, race, year of diagnosis, histologic grade, tumor size, node metastasis, distant metastasis, surgery of the primary tumor, radiotherapy, chemotherapy, vital status, and survival time. The primary outcome measure was DSS, defined as the time from diagnosis of OS to death attributed to this tumor. The SEER data-use agreement was assigned, and patients’ informed consent was waived as the data is publicly available and de-identified. All procedures adhered to the Declaration of Helsinki.

Patients were randomly assigned into three cohorts: the training cohort, validation cohort I, and validation cohort II. Baseline characteristics were described using the Chi-square test. Kaplan-Meier survival analyses and log-rank tests were applied to estimate DSS among different subgroups. Univariate and multivariate Cox analyses were performed to identify independent prognostic factors. A nomogram model was plotted based on multivariate analysis results to predict survival outcomes. The concordance index (c-index) was calculated to evaluate the nomogram’s performance, and calibration curves were plotted to assess predictive accuracy. A risk stratification model was developed based on each patient’s total score from the nomogram, dividing patients into low-risk and high-risk groups using X-tile software. All statistical analyses were conducted with R software, with statistical significance determined by a two-sided P < 0.05.

A total of 1639 patients were enrolled: 881 in the training set, 408 in validation set I, and 350 in validation set II. In the training set, significant factors for DSS included age at diagnosis, histologic grade, tumor size, distant metastasis, and surgery of the primary tumor. Age at diagnosis (≥18 vs. <18: HR 2.07, 95% CI 1.63–2.63, P 5 cm vs. <2 cm: HR 2.09, 95% CI 1.29–3.38, P = 0.003), distant metastasis (metastasis vs. without metastasis: HR 3.07, 95% CI 2.35–4.01, P < 0.001), and surgery of primary tumor (local resection vs. no surgery: HR 0.31, 95% CI 0.22–0.44, P < 0.001; amputation vs. no surgery: HR 0.49, 95% CI 0.33–0.71, P < 0.001) were significant.

A predictive nomogram integrating these five independent risk factors was constructed. The total score predicted 1-year, 2-year, and 3-year DSS for individual patients. The C-indexes in the training (0.72, 95% CI 0.68–0.76), validation I (0.78, 95% CI 0.76–0.82), and validation II (0.80, 95% CI 0.77–0.83) cohorts suggested excellent accuracy. In the training cohort, the area under curve (AUC) values for predicting 1-year, 2-year, and 3-year DSS were 0.75, 0.69, and 0.67, respectively. In validation cohort I, the AUC values were 0.71, 0.76, and 0.74, respectively, and in validation cohort II, they were 0.82, 0.76, and 0.76, respectively. Calibration curves indicated good consistency between the nomogram-based predictive outcomes and actual prognosis results.

A risk stratification model divided patients into low-risk (total score <59) and high-risk (total score 59–106) groups. High-risk patients were younger, had higher histopathologic grades, larger tumor burdens, and more distant disease. They were less likely to have received chemotherapy and radiotherapy. The median DSS in the low-risk and high-risk groups was 61.6 months (95% CI 55.6–72.4) and 47.4 months (95% CI 40.5–56.8), respectively. Kaplan-Meier curves indicated that the risk stratification model effectively differentiated DSS in all subgroups.

This study identified five demographic and clinicopathologic characteristics as independent prognostic factors in OS patients: age at diagnosis, histologic grade, tumor stage, distant metastasis, and surgery of the primary tumor. A risk classifier based on these features stratified patients into different prognosis statuses. The large sample size and broad eligibility criteria enhance the clinical applicability of these findings.

Patients aged >18 years had poorer outcomes compared to younger patients. The presence of metastases was significantly associated with DSS. Higher-grade tumors and larger tumor sizes were associated with lower survival rates, consistent with previous publications. Surgery, both local resection and amputation, was associated with better outcomes, while the survival benefit of chemotherapy and radiotherapy remains controversial. Previous studies have shown mixed results regarding the impact of chemotherapy and radiotherapy on survival in OS patients.

The nomogram and risk stratification model developed in this study provide a quantitative method to predict DSS for individual patients and classify them into distinct risk subgroups. These tools offer critical information for indicating prognosis and guiding clinical decisions. However, the retrospective nature of this study and the lack of information on some potential prognostic factors, such as surgical margins and detailed treatments, are limitations. Future prospective and translational studies are needed to validate these findings.

In conclusion, this research identified potential prognostic factors for predicting DSS in OS patients. The nomogram and risk stratification model provide valuable tools for predicting survival outcomes and classifying patients into different risk subgroups. Future large-scale, multi-center, and prospective studies are necessary to validate these findings and further refine prognostic tools for OS patients.

doi.org/10.1097/CM9.0000000000001837

Was this helpful?

YesNo

0 / 0