Alcohol dependence results in two different clinical forms: "uncomplicated" alcoholism (UA) and Korsakoff's syndrome (KS). Certain brain networks are especially affected in UA and KS: the frontocerebellar circuit (FCC) and the Papez circuit (PC). Our aims were (1) to describe the profile of white matter (WM) microstructure in FCC and PC in the two clinical forms, (2) to identify those UA patients at risk of developing KS using their WM microstructural integrity as a biomarker. Tract-based spatial statistics and nonparametric voxel-based permutation tests were used to compare diffusion tensor imaging (DTI) data in 7 KS, 20 UA, and 14 healthy controls. The two patient groups were also pooled together and compared to controls. k-means classifications were then performed on mean fractional anisotropy values of significant clusters across all subjects for two fiber tracts from the FCC (the middle cerebellar peduncle and superior cerebellar peduncle) and two tracts from the PC (fornix and cingulum). We found graded effects of WM microstructural abnormalities in the PC of UA and KS. UA patients classified at risk of developing KS using fiber tracts of the PC from DTI data also had the lowest scores of episodic memory. That finding suggests that WM microstructure could be used as a biomarker for early detection of UA patients at risk of developing KS.
Keywords: Korsakoff's syndrome; Papez's circuit; alcoholism; classification; frontocerebellar circuit; tract-based spatial statistics.
© 2015 Wiley Periodicals, Inc.