Premium
Prediction of cognitive performance from IQ biomarkers and MR volumetrics in the ATN framework depends on diagnosis
Author(s) -
Whittington Alex,
Gunn Roger N.
Publication year - 2021
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1002/alz.056191
Subject(s) - neuroimaging , dementia , biomarker , cognition , hippocampal formation , alzheimer's disease neuroimaging initiative , psychology , neuropsychology , amyloid (mycology) , effects of sleep deprivation on cognitive performance , cognitive impairment , medicine , neuroscience , pathology , disease , chemistry , biochemistry
Abstract Background The A/T/N classification system has emphasised the importance of considering multi‐channel biomarker readouts in AD research. Further, it is important to establish the relationship of these signals to cognitive performance and their relevance to enable optimum inclusion criteria for clinical trials. The IQ analytics platform provides powerful outcome measures for both Amyloid and Tau PET. In this work, we also consider hippocampal volume and investigate the relationship of these A/T/N biomarkers with cognitive performance at different stages of the disease process. Method Amyloid IQ , Tau IQ and MR Volumetric analytics were performed on cross‐sectional data from 605 subjects (CN=376, MCI=173, Dementia=56) downloaded from ADNI where each subject had [ 18 F]Florbetapir, [ 18 F]Flortaucipir and a T1 MRI scan. Amyloid IQ determines the overall amyloid burden as an Amyloid Load percentage (Figure 1 and 2) whilst the Tau IQ algorithm deconstructs the more complex Tau signal into global signal and local signal (Figure 3 and Figure 4). The hippocampal volumes for each subject were derived in SPM. We performed a multiple regression analysis with ADAS‐COG as the outcome variable and Global Tau Load, Local Tau Load, Amyloid Load and Hippocampal Volume as predictors. All possible linear models (including interaction terms) were evaluated on each diagnosis group separately and the optimum model was selected at in each case by choosing the model with the lowest BIC. Result The hippocampal volume was a statistically significant variable in the optimum model in all diagnosis groups but the statistically significant IQ analytics outcome measures were dependent on the diagnosis (Table 1); in the CN group the key parameter was Amyloid Load, in the MCI group, interestingly, both Local Tau Load and Amyloid Load were statistically significant, finally in the Dementia subjects only the Global Tau Load was found to be statistically significant. When evaluated on all subjects simultaneously, the optimum model explained 49% of the variance in ADAS‐Cog scores. Conclusion The IQ analytics biomarkers provide orthogonal information that are related to cognitive function in different ways across the disease spectrum. Our results imply that the novel Local Tau Load outcome measure is an important biomarker for stratifying and following MCI subjects.