Combining ADAS-Cog Assessment with Hypometabolic Region of 18F-FDG PET/CT Brain Imaging for Alzheimer's Disease Detection

摘要

The Alzheimer's disease Assessment Scale-Cognitive subscale (ADAS-Cog) is a convenient tool to perform classifications of AD at mild, moderate, and severe levels with a combination of other neuropsychological tests and SPM12. ADAS-Cog showed a fairly acceptable discrimination for moderate AD with (AUC 0.6 [95% CI, 0.171-1.029]; sensitivity=0.818, specificity=0.600, $ext{PPV}=0.655$, NPV=1.000) at a cut off scores ≥37. The best cut off scores for severe AD diagnosis was ≥48 (AUC 1.000, sensitivity=1.000, specificity=0.786, $ext{PPV}=0.533, ext{NPV}=1.001$). However, ADAS-Cog for mild AD was undetermined (AUC 0%) with all having Montreal Cognitive Assessment (MoCA) scores ≥17. This proved that ADAS-Cog was more reliable to differentiate between severe AD and patients with less effect of cognitive decline ($mathrm{P} < 0.0001$) with PPV of 100% and NPV of 93.3% compared with MoCA (PPV of 59.3% and NPV of 82.4%). These results tested the domains of cognitive parts that showed moderate correlation was associated with the Attention domain. Consequently, the strong correlations were associated with Memory, Concentration, and Abstraction. The most important domains that have a very strong correlation with ADAS-Cog scores were Executive functions, Orientation, and Language. SPM analysis with one-sample t-test within-group analysis was done selecting the total intracranial volume (TIV), MoCA scores, and ADAS-Cog scores as the covariates. The results showed that normal control had bilateral hypometabolism area at medial frontal gyrus, left caudate, left lateral orbital gyrus, right superior frontal gyrus, and left medial frontal gyrus. As expected, all hypometabolism areas across the three AD diagnostic groups were typically detected in clusters of more than 100 voxels showing correlations with reduced activity in multiple sites of the brain. In conclusion, the ADAS-Cog with a combination of PET brain analysis had given promising results in the detection of Alzheimer's disease and measuring the cognitive domain declination.