Objective: Normalization to an appropriate reference region in F-FDG PET imaging may enhance diagnostic performance in Huntington disease (HD). We aimed to identify stable brain areas that could be used to model neurometabolic degeneration in HD correlating imaging (SUVrvalues at the basal ganglia [BBGG]) and clinical parameters (disease burden score [DBS]).
Materials and methods: We performed brain F-FDG PET/CT in 38 manifest HD patients (meanage ± SD, 54 ± 14.3 years; CAGrepeats ± SD, 44.2 ± 3.1), 20 premanifest HD patients (meanage ± SD, 42.7 ± 11.7 years; CAGrepeats ± SD, 40 ± 3.8), and 18 healthy controls (NC; meanage ± SD, 45 ± 13.2 years). For quantitative analysis, we selected (a) defined reference regions from the Montreal Neurological Institute space atlas (pons, whole cerebellum, cerebral white matter, thalamus, and a pons-cerebellar vermis region of interest), and (b) reference clusters obtained by voxelwise statistical comparison across groups (P < 0.05 FWE; extent voxel threshold k = 200). Each candidate reference region and reference cluster was quantitatively assessed using imaging and clinical parameters.
Results: Comparing HD and NC groups, we obtained a reference cluster in the cerebellum, and in temporal and frontal lobes. Comparing manifest HD and premanifest HD patients, we observed reference clusters in the cerebellum, pons, thalamus, parietal lobe, and cuneus. The set of reference regions showed a significant correlation between SUVrvalues at the BBGG and DBS in all HD patients. In premanifest HD patients, the correlation between SUVrvalues at the BBGG and DBS was significant using the pons-cerebellar vermis region of interest, the thalamus as defined reference regions, and the pons and thalamus as reference clusters. In manifest HD patients, the correlation was significant using the temporal and white matter frontal lobe clusters. Variance between SUVrvalues in the set of reference regions and reference clusters was minimal within NC.
Conclusions: The pons may be a stable and reliable region to calculate SUVrvalues to model the neurometabolic degeneration in quantitative F-FDG PET imaging in HD.