Risk factors for cranial irradiation-related late neurocognitive toxicity: A prospective cohort study.

Abstract

2066 Background: Neurocognitive dysfunction is a common complication of cranial irradiation, occurring in up to 50% of irradiated brain tumor patients. Symptoms often severely compromise quality of life long before patients succumb to their tumor; however, risk factors are poorly understood, and consequently, prognostication and prevention have not been possible. The objective of this study was to evaluate the prognostic value of vascular risk factors for the development of irradiation-related brain injury. Methods: This single-institution prospective cohort study included patients with malignant primary brain tumors who received cranial irradiation as part of their initial tumor-directed therapy. Three putative vascular risk factors – homocysteine, total cholesterol, apoprotein E genotype (ApOE) – were measured and dichotomized (above vs. below the laboratory normal). Univariate analyses compared each risk factor with four measures of neurocognitive dysfunction: mini-mental status exam (MMSE), MRI white matter changes at 6 months (MRI), physician (Phys) assessment, and patient (Pat) assessment. Decision analysis was used to construct a prediction algorithm. Results: 80 patients were included in this analysis. Elevated homocysteine was the most powerful and consistent predictor of neurocognitive toxicity, followed by elevated triglycerides and the ApOE genotype (Table). Logistic regression revealed a highly significant (p<0.01) association between homocysteine level and each of the four outcome variables. Decision tree analysis using homocysteine level (high vs. low) and ApoE genotype (yes vs. no) provided the most efficient predictive algorithm. Conclusions: Two putative vascular risk factors (homocysteine level > 14 and ApOE genotype) provide moderate ability to predict post-radiation neurocognitive dysfunction using a variety of simple but clinically meaningful definitions of dysfunction. This predictive algorithm should be validated in prospective trials. If these findings are corroborated, studies examining additional risk factors as well as studies looking at risk factor mitigation will be appropriate. [Table: see text]