Modeling Radiation Dosimetry to Predict Cognitive Outcomes in Pediatric Patients with CNS Embryonal Tumors

Abstract

Materials/Methods: Pediatric patients (n 39) with CNS embryonal tumors were prospectively evaluated with serial cognitive testing before and after post-operative risk-adapted craniospinal (CSI) and conformal primary site irradiation followed by chemotherapy. Patients were characterized according to age (median 7.3 years); median follow-up (42.7 months); male:female ratio 25:14; diagnosis (MB, n 34; sPNET, n 1; ATRT, n 4); tumor location (infratent 35; supratent 4); and risk-classification (average 14; high 25). Average-risk included residual tumor 1.5cm and no evidence of metastases regardless of tumor location. Risk-adapted CSI (23.4 Gy-average risk; 36–39.6 Gy-high risk), conformal posterior fossa irradiation (36.0 Gy-average risk only) and conformal primary site irradiation (CTV 2cm) to 55.8 Gy was delivered at 1.8 Gy/day. Differential DVHs were calculated from composite dosimetry for the total brain, supratentorial brain, infratentorial brain, and left and right temporal lobe volumes and integrated in 5 Gy increments for correlation with IQ evaluations conducted after surgery and 1, 2, and 5 years from completion of therapy using standard instruments; all patients had 3 or more evaluations. A linear mixed model with random coefficients was used to estimate the impact of the specific dose intervals and their fractional volumes on the slope of longitudinal IQ and included age and other covariates. Integrated dosimetry was partitioned into two (low and high) or three (low, intermediate and high) dose intervals with corresponding divisions determined by experimenting over the range of dose in increments of 5 Gy. Model were checked by examining the estimating equation for direction of slope, magnitude of the dose-volume coefficients and their p-value. The divisions which resulted in coefficients with the highest level of statistical significance were chosen.