Sarculator is a Good Model to Predict Survival in Resected Extremity and Trunk Sarcomas in US Patients.

ABSTRACT

BACKGROUND: Sarculator is an online validated nomogram that predicts overall survival of patients with resected, primary extremity sarcomas. However, its ability to accurately predict outcomes in US patients with sarcoma is unknown. PATIENTS AND METHODS: Patients from the National Cancer Data Base (NCDB) (2006-2016) with resected stage I-III primary extremity or trunk sarcoma were included. Predicted overall survival (pOS) was calculated using the Sarculator algorithm, which includes patient age, tumor size (cm), grade (1-3), and histology, and compared with actual overall survival (aOS). Harrell's C-index was calculated to determine the discriminatory ability of Sarculator (0.7 = good, 0.8 = strong, 1.0 = perfect model), and calibration plots were created. RESULTS: In total, 9738 patients were included. Five-year pOS was 73.7% compared with aOS of 68.9%. The C-index for the entire cohort was 0.726. By stage, the C-index was 0.730 for stage I, 0.708 for stage II, and 0.679 for stage III. By histology, C-indices were highest for leiomyosarcoma (0.745), myxofibrosarcoma (0.722), and other histologies (0.721). By sociodemographic variables, Sarculator performed better for patients < 50 years (C-index 0.722), of other/unknown race (C-index 0.781), with private insurance (C-index 0.715), treated at a center other than a community cancer programs (C-index > 0.7), and with no comorbidities (C-index 0.716). Outcomes by zip code educational attainment and income were not markedly different (all C-indices > 0.7). CONCLUSIONS: Sarculator is overall a good predictor of aOS and useful tool for clinicians to aid in survival prognostication. However, clinicians should be aware of populations for whom Sarculator's predictions may be less accurate. Future work could focus on enhancing the Sarculator algorithm specifically for US patients by including demographic variables.