Association between hippocampal dose and memory in survivors of childhood or adolescent low-grade glioma: a 10-year neurocognitive longitudinal study

Abstract

Background: Hippocampal avoidance has been suggested as a strategy to reduce short-term memory decline in adults receiving whole-brain radiation therapy (RT). The purpose of this study was to determine whether the hippocampal dose in children and adolescents undergoing RT for low-grade glioma was associated with memory, as measured by verbal recall.

Methods: Eighty patients aged at least 6 years but less than 21 years with low-grade glioma were treated with RT to 54 Gy on a phase II protocol. Patients underwent age-appropriate cognitive testing at baseline, 6 months posttreatment, yearly through 5 years posttreatment, year 7 or 8, and year 10 posttreatment. Random coefficient models were used to estimate the longitudinal trends in cognitive assessment scores.

Results: Median neurocognitive follow-up was 9.8 years. There was a significant decline in short-delay recall (slope = -0.01 standard deviation [SD]/year, P < 0.001), total recall (slope = -0.09 SD/y, P = 0.005), and long-delay recall (slope = -0.01 SD/y, P = 0.002). On multivariate regression, after accounting for hydrocephalus, decline in short-delay recall was associated with the volume of right (slope = -0.001 SD/y, P = 0.019) or left hippocampus (slope = -0.001 SD/y, P = 0.025) receiving 40 Gy (V40 Gy). On univariate regression, decline in total recall was only associated with right hippocampal dosimetry (V40 Gy slope = -0.002, P = 0.025). In children <12 years, on univariate regression, decline in long-delay recall was only associated with right (V40 Gy slope = -0.002, P = 0.013) and left (V40 Gy slope = -0.002, P = 0.014) hippocampal dosimetry.

Conclusion: In this 10-year longitudinal study, greater hippocampal dose was associated with a greater decline in delayed recall. Such findings might be informative for radiation therapy planning, warranting prospective evaluation.

Keywords: hippocampus; memory; pediatric low-grade glioma.

Fig. 1

Regression lines of change in CVLT-SD, CVLT-LD, CVLT-T, semantic cluster and discriminability scores over time without inclusion of any covariates. Unadjusted change in neurocognitive measures over time for individual patients can be found in the supplement.

Fig. 2

Contour plot of the difference in CVLT-SD from baseline as a function of time from RT and volume of: (A) right hippocampus receiving 40 Gy (%) in patients with hydrocephalus, (B) right hippocampus receiving 40 Gy (%) in patients without hydrocephalus, (C) left hippocampus receiving 40 Gy (%) in patients with hydrocephalus, (D) left hippocampus receiving 40 Gy (%) in patients without hydrocephalus. Treating the entire right (Figure 2A and B) or left hippocampus (Figure 2C and D) to 40 Gy was associated with a 5-year and 10-year short-delay recall reduction of 1 SD and 2 SD, respectively, for patients with hydrocephalus, and 0.5 SD and 1 SD, respectively, for patients without hydrocephalus. Contour plot of difference in CVLT-LD from baseline for children <12 years as a function of time from RT and volume of (E) right hippocampus receiving 40 Gy (%), and (F) left hippocampus receiving 40 Gy (%). Treating the entire right (Figure 2E) or left hippocampus (Figure 2F) to 40 Gy was associated with a 5-year and 10-year long-delay recall reduction of 0.9 SD and 1.8 SD, respectively.