Prognostic nomograms for predicting overall and cancer-specific survival of high-grade osteosarcoma patients

Abstract

Aim: The present study aimed to develop nomograms estimating survival for patients with high-grade osteosarcoma.

Methods: 1990 patients with high-grade osteosarcoma between 1994 and 2013 were retrospectively retrieved from the Surveillance, Epidemiology, and End Results (SEER) database. Data from 12 cancer registries (n = 1460) were used to conduct multivariate Cox analysis to identify independent prognostic factors. Nomograms which estimate 3- and 5-year overall survival (OS) and cancer-specific survival (CSS) were constructed. The nomograms were internally validated for calibration and were also externally validated with an independent patient cohort from 1 cancer registry (n = 530).

Results: Age, primary site, tumor size, use of surgery, and extent of disease were found to be independently associated with OS and CSS (p < 0.05). The nomograms estimating 3- and 5-year OS and CSS were developed based on these prognostic factors. The concordance indices were high in internal validation (0.726 for OS and 0.731 for CSS) and external validation (0.716 for OS and 0.724 for CSS). Internal and external calibration plots demonstrated a good agreement between nomogram prediction and actual observation.

Conclusions: We constructed nomograms that accurately predict OS and CSS of high-grade osteosarcoma patients. The nomograms can be used for counseling patients and establishing risk stratification.

Keywords: Nomogram; Osteosarcoma; Prognostic factor; Survival; Validation.

Fig. 1

(A)–(C) The graphs show defining the optimal cutoff values of tumor size via X-tile analysis. (A) The black dot indicates that optimal cutoff values of tumor size have been identified. (B) A histogram and (C) Kaplan–Meier were constructed based on the identified cutoff values. Optimal cutoff values of tumor size were identified as 8.0 cm and 13.1 cm based on overall survival.

Fig. 2

The flow diagram indicates the process of collecting patients. Based on the inclusion and exclusion criteria, 1990 patients were collected from the SEER database. 1460 patients from 12 cancer registries and 530 patients from 1 cancer registry were assigned into the training and validation cohorts, respectively.

Fig. 3

(A)–(B) The graphs show the nomograms which predict 3- and 5-year (A) overall survival and (B) cancer-specific survival of high-grade osteosarcoma patients. Points of each variable was acquired by drawing a vertical line between each variable and the Points scale. By totaling the points of each variable, we then draw a vertical line between the Total Points scale and overall survival or cancer-specific survival scale to calculate the predicted 3- and 5-year survival.

Fig. 4

(A)–(H) The graphs show the calibration plots for internal validation of (A) actual 3-year and (B) 5-year overall survival; (C) actual 3-year and (D) actual 5-year cancer-specific survival; and external validation of (E) actual 3-year and (F) 5-year overall survival; and (G) actual 3-year and (H) 5-year cancer-specific survival. The dashed line represents an excellent match between nomogram prediction (X-axis) and actual survival outcome (Y-axis). The cohort was divided into ten groups with equal sample size for internal and external validation. Closer distances from the points to the dashed line indicate higher prediction accuracy.