Steep Dose-response Relationship between Maximum Hippocampal Dose and Memory Deficits following Hippocampal Avoidance Whole Brain Radiation Therapy (HA-WBRT) for Brain Metastases: A Secondary Analysis of NRG/RTOG 0933

Abstract

NRG/RTOG 0933 was a phase II study suggesting that HA-WBRT for brain metastases may help to preserve cognitive function. However, hippocampal dose constraints in clinical trials conducted to date have been driven predominantly by feasibility of sparing while maintaining adequate coverage of the at-risk areas, as the radiation tolerance of the hippocampus remains uncertain. The purpose of this study was to perform an analysis of hippocampal radiation dose-volume effects and memory decline following HA-WBRT using a prospectively collected multi-institutional dataset. 113 patients received HA-WBRT on RTOG 0933. All patients received 30 Gy in 10 fractions. Memory was measured by the Hopkins Verbal Learning Test-Revised (HVLT-R) Delayed Recall (DR) and Total Recall (TR) at baseline and 6 months after radiation. Dose–volume histograms (DVH) were generated for the composite bilateral hippocampi. Dose–response data were fitted to a nonlinear model for D100%, D50%, and Dmax to the bilateral hippocampi. The primary endpoint was memory decline, measured by HVLT-R DR and TR at baseline and 6 months after completion of radiation therapy. The Reliable Change Index (RCI) was used to measure a clinically meaningful decline between baseline and 6 months for HVLT-R DR and TR, defined as a raw score change greater than or equal to 3. Models were also repeated using an RCI threshold raw change score of greater than or equal to 5. Evaluations of statistical significance used the Fisher Exact Test. 27 patients had complete DVH data available in addition to both baseline and 6-month cognitive testing. D100% was per protocol (≤9 Gy) in all cases. Dmax was per protocol (≤16 Gy) in 23 cases, had an acceptable variation (≤17 Gy) in 3 cases and an unacceptable deviation (20 Gy) in 1 case. There was a steep dose-response relationship between higher maximum dose to the bilateral hippocampi and deterioration in short-term memory using a raw score change greater than or equal to 3. For example, a Dmax = 14Gy was associated with a 10% risk and Dmax = 16Gy was associated with a 25% risk (underpowered, p = 0.139). When evaluating for a raw score change of greater than or equal to 5, there was a steep dose-response for Dmax and deterioration in short term memory according to both HVLT-DR and TR, and a significant relationship between maximum dose to the bilateral hippocampi and performance on HVLT-TR (p = 0.028). These data demonstrate a steep dose relationship between maximum dose to the bilateral hippocampi and deterioration in short term memory, even amongst patients who met hippocampal constraints according to the protocol, suggesting that tighter constraints may be beneficial. Future investigations with larger sample sizes will be important to confirm these findings.