Metabolic Profile of Cerebellum in Posterior Fossa Tumor Survivors: Correlation With Memory Impairment

Abstract

AimsThe cerebellum is a key structure in working and procedural memory. The aim of the present prospective exploratory study was to investigate, the metabolic characteristics of the cerebellum in posterior fossa tumor (PFT) survivors using 3D proton magnetic resonance spectroscopy imaging (3D MRSI), to determine whether metabolites could be useful biomarkers of memory impairment. Materials and methodsSixty participants were included in the IMPALA study, divided into three groups: 22 irradiated PFT, 17 nonirradiated PFT, and 21 healthy controls matched with irradiated PFT for age, sex, and handedness. PFT survivors were treated at least 5 years ago, either by surgery or a combination of surgery, chemotherapy, and radiotherapy. All participants underwent working and procedural memory tests and multimodal MRI including a 3D MRSI sequence. N-acetylaspartate (NAA), choline (Cho), creatine (Cr), and lactate (Lac) metabolite values were extracted from the cerebellum for comparisons between groups, correlations with neurocognitive test scores, and radiotherapy doses. ResultsMedian (range) age at neurocognitive tests was 18 (7–26) years. Median Cho, Cr, NAA, and Lac values, and the ratio of NAA to the sum of metabolites were significantly lower for PFT survivors than for healthy controls (p < 0.05). Scores on working and procedural memory tests were significantly lower for PFT survivors (p < 0.004) and correlated with median and maximum Cho and NAA values (0.28 <r < 0.49, p < 0.04). Except for creatine, the other metabolites were not significantly different between irradiated and nonirradiated survivors. MRSI values in the cerebellum were not correlated with either total dose or doses received by this structure. ConclusionResults revealed changes in cerebellar metabolic values in PFT survivors that were closely correlated with memory deficits, suggesting that some metabolites could be used as markers of cognitive decline, but this will require validation on a larger sample size.