Derivation and validation of a risk classification tree for patients with synovial sarcoma

BACKGROUND: Synovial sarcoma (SS) accounts for 8%–10% of all soft‐tissue sarcomas. Clinical presentation and outcomes vary, yet discrete risk groups based on validated prognostic indices are not defined for the full spectrum of patients with SS. METHODS: We performed a retrospective cohort study using data from the SEER (surveillance, epidemiology, and end results program) database of SS patients who were <70 years of age at diagnosis. We constructed a recursive partitioning model of overall survival using a training cohort of 1063 patients with variables: Age at diagnosis, sex, race, ethnicity, primary site, tumor size, tumor grade, and stage. Based on this model, we grouped patients into three risk groups and estimated 5‐year overall survival for each group. We then applied these groups to a test cohort (n = 1063). RESULTS: Our model identified three prognostic groups with significantly different overall survival: low risk (local/regional stage with either <21 years of age OR tumor <7.5 cm and female sex), intermediate‐risk (local/regional stage, age ≥ 21 years with either male sex and tumor <7.5 cm OR any sex with appendicular anatomic location) and high risk (local/regional stage, age ≥ 21 years, tumor size ≥7.5 cm and non‐appendicular location OR distant stage). Prognostic groups were applied to the test cohort, showing significantly different survival between groups (p < 0.0001). CONCLUSIONS: Our analysis yields an intuitive risk‐classification tree with discrete groups, which may provide useful information for researchers, patients, and clinicians. Prospective validation of this model may inform efforts at risk‐stratifying treatment.