Surveying and Optimizing the Predictors for Ependymoma Specific Survival using SEER Data


Purpose: This study used receiver operating characteristic curve to analyze Surveillance, Epidemiology andEnd Results (SEER) ependymoma data to identify predictive models and potential disparity in outcome. Materialsand
Methods: This study analyzed socio-economic, staging and treatment factors available in the SEER databasefor ependymoma. For the risk modeling, each factor was fitted by a Generalized Linear Model to predict theoutcome (‘brain and other nervous systems’ specific death in yes/no). The area under the receiver operatingcharacteristic curve (ROC) was computed. Similar strata were combined to construct the most parsimoniousmodels. A random sampling algorithm was used to estimate the modeling errors. Risk of ependymoma deathwas computed for the predictors for comparison.
Results: A total of 3,500 patients diagnosed from 1973 to 2009were included in this study. The mean follow up time (S.D.) was 79.8 (82.3) months. Some 46% of the patientswere female. The mean (S.D.) age was 34.4 (22.8) years. Age was the most predictive factor of outcome. Unknowngrade demonstrated a 15% risk of cause specific death compared to 9% for grades I and II, and 36% for gradesIII and IV. A 5-tiered grade model (with a ROC area 0.48) was optimized to a 3-tiered model (with ROC areaof 0.53). This ROC area tied for the second with that for surgery. African-American patients had 21.5% risk ofdeath compared with 16.6% for the others. Some 72.7% of patient who did not get RT had cerebellar or spinalependymoma. Patients undergoing surgery had 16.3% risk of death, as compared to 23.7% among those whodid not have surgery.
Conclusion: Grading ependymoma may dramatically improve modeling of data. RT isunder used for cerebellum and spinal cord ependymoma and it may be a potential way to improve outcome.