Real-time prognosis prediction with conditional survival analysis for skull base chordoma based on SEER

Skull base chordoma (SBC) is a rare, locally aggressive malignant bone tumor with a poor prognosis due to its location and recurrence. Despite advances in surgery and radiotherapy (RT), long-term survival remains uncertain. Traditional survival analyses are limited by their static nature, failing to capture the dynamic changes in survival probabilities over time. To address this, we applied conditional survival (CS) analysis for more precise evaluation of evolving survival rates in SBC patients.

Data of 717 SBC patients [2000-2019] from SEER (Surveillance, Epidemiology, and End Results) database were obtained. Using CS analysis, we evaluated survival probabilities over time and developed the first SBC-specific CS-nomogram. Key clinicopathological factors were incorporated into the model via least absolute shrinkage and selection operator (LASSO) and multivariate Cox analysis. The nomogram was validated with training and validation cohorts. Calibration curves, C-index, time-dependent receiver operating characteristic (ROC) curves and decision curve analysis (DCA) were used to evaluate model performance.

CS analysis showed a steady overall survival (OS) improvement in SBC patients over time. The 10-year survival probability rose from 61% at diagnosis to 98% after 9 years. Eight clinicopathological factors, significant predictors of OS, were incorporated into the CS-nomogram. The model had robust predictive accuracy, with C-index values of 0.703 (training) and 0.731 (validation). Calibration curves and DCA indicated good agreement between predicted and actual outcomes with significant net clinical benefit.

Combining CS analysis and a nomogram, we developed a new tool for dynamic, individualized survival predictions in SBC patients. The CS-nomogram can improve clinical decision-making and patient counseling, bringing hope and more precise prognostic evaluations for long-term survivors.