The ability of EEG using statistical pattern recognition to predict conversion from subtypes of mild cognitive impairment to dementia: A five years follow-up study.

Background: Studies have shown that quantitative EEG is useful in predicting conversion from mild cognitive impairment (MCI) to Alzheimer's disease dementia (ADD) and dementia with Lewy bodies (DLB). As subcortical pathology is present and executive impairment is common in DLB, we hypothesized that EEG could predict conversion in patients with impaired executive function and any subcortical pathology.

Methods: We included 113 patients with MCI from five Nordic memory clinics, 80 (71%) with amnestic MCI, 17 (15%) with dysexecutive MCI (deMCI), 3 (3%) with aphasic, 2 (2%) with visuospatial and 11 (10%) with unspecific MCI. Patients were examined with EEG in a resting state applying the statistical pattern recognition (SPR) method and followed up for five years. Eleven drop-outs were assessed after baseline. Receiver operating characteristic (ROC) analyses were used to examine the ability of EEG to predict conversion.

Results: Sixty patients converted to dementia, 47 to ADD, eight to vascular dementia, two to DLB, one to frontotemporal dementia and two to unspecific dementia. Eight (11%) recovered and 45 (40%) remained MCI stable. ROC analyses revealed that EEG predicted conversion from dysexecutive MCI to dementia with area under the curve (AUC) of 0.92 (95% CI 0.76-100), sensitivity of 89% and specificity of 100%. Subcortical pathology was present in 89% of the dysexecutive MCI converters. EEG did not predict conversion from amnestic MCI to dementia.

Conclusion: This study demonstrates that quantitative EEG using the SPR method predicts conversion from deMCI to dementia disorders with subcortical pathology with high sensitivity and specificity.