Postdiagnostic pretreatment risk stratification for chondrosarcoma survival using routinely available registry variables: a SEER-based temporal validation study

Postdiagnostic pretreatment risk stratification for chondrosarcoma survival using routinely available registry variables: a SEER-based temporal validation study. Download PDF Abstract This study developed and temporally validated a postdiagnostic pretreatment survival risk-stratification framework for chondrosarcoma using routinely available variables from the Surveillance, Epidemiology, and End Results registry. Patients diagnosed from 2000 to 2015 formed the development cohort, and those diagnosed from 2016 to 2021 formed the independent temporal validation cohort.

Discrimination, time-specific calibration, risk-group event enrichment, model interpretability, histological heterogeneity, missing-data sensitivity, model-selection stability, and competing risks were evaluated. Among 4066 eligible patients, the temporal validation cohort included 1168 patients and 215 overall-survival deaths. The locked Cox proportional hazards-random survival forest ensemble achieved a C-index of 0.861, with 1- and 3-year AUCs of 0.903 and 0.870.

The high-risk group comprised 35.4% of the validation cohort but accounted for 80.5% of observed overall-survival deaths. In competing-risk analysis, the 1- and 3-year cumulative incidences of chondrosarcoma-attributed death in the high-risk group were 24.2% and 38.7%, respectively. Age, summary stage, tumor grade, and histological subtype were the leading permutation-importance variables; discrimination remained substantial after exclusion of histology and within the ICD-O-3 9220/3 chondrosarcoma NOS subgroup.

Medical Disclaimer: This article is for informational purposes only and does not constitute medical advice. Always consult with qualified healthcare professionals for medical decisions and treatment options.