Dose-Volume Effects Of Partial-Brain Radiation Therapy On Late Neurocognitive Processing Speed And Executive Function

We investigated the dose-volume effects of partial-brain radiation (RT) on late neurocognitive function in adult patients with benign and low-grade intracranial tumors, which has not been well-described. From 2004-2013, 32 patients (10 pituitary adenoma, 9 low grade glioma, 7 meningioma, 2 craniopharyngioma, 4 other) underwent prospective processing speed and executive function testing using serial Trail-Making Test B evaluations pre-RT, 6 and 18 months post-RT. Following resection (n = 27), biopsy (n = 3), or no surgery (n = 2), patients received partial-brain RT to a median dose of 54 Gy in 1.8 Gy fractions. Median age was 47 years (range, 25-71) and 19 were male. Those with disease progression or <6 months follow-up were excluded. Predicted test Z scores after RT were determined using a standardized regression based formula. A worsening ΔZ ≥2 standard deviations (SDs) from the mean was considered clinically significant. Univariate and multivariate regression models were used to evaluate the impact of patient age, gender, pre-RT Z score, tumor location (frontal, temporal, or neither) and generalized equivalent uniform dose (gEUD) to normal brain on 6 and 18 month changes in performance. To test the effect of volume irradiated on neurocognitive outcome, a range of parameters (a = 1, 5, 14, 50 denoting mean to maximum dose to normal tissue) were empirically assessed in separate gEUD models. Prior to RT, only 2/32 patients performed worse than 2 SDs from the age-adjusted expected mean. Six months post-RT, 7/31 patients had clinically significant worsening performance, compared to 7/26 patients at 18 months with available test data (4 with known decline and 3 additional patients). In the univariate model, only increasing patient age (P = .021) and worse pre-RT Z score (P = .004) predicted for a worsening performance. On multiple regression, the most significant factor was RT dose, with an estimated increase of mean ΔZ by 0.15 (std error 0.06) for every unit increase in gEUD (P = .018). Age (P = .036) and pre-RT Z score (P = .041) also remained significant, whereas, tumor location did not (P = .267). Only gEUD models incorporating higher a-values (a = 14, 50) were significant, suggesting a small normal tissue volume effect on dose tolerance. This is the first prospective study to demonstrate that the dose of partial-brain RT is the strongest predictor of decline in processing speed and executive function, and that the volume effect of normal irradiated brain may only be a minor component of this late complication. Whether injury to white matter fibers of complex neural networks incites this compromise warrants further evaluation of radioprotective interventions, especially in older patients and those with baseline impairment.