![Sample our Behavioral Sciences journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/110801/TOC-Subject-Sample-2021-Behavioral-Sciences.jpg"})
ABSTRACT
Introduction
We studied the ability of latent factor scores to predict conversion from mild cognitive impairment (MCI) to Alzheimer’s disease (AD) and investigated whether multimodal factor scores improve predictive power, relative to single-modal factor scores.
Method
We conducted exploratory factor analyses (EFAs) and confirmatory factor analyses (CFAs) of the baseline data of MCI subjects in the Alzheimer’s Disease Neuroimaging Initiative (ADNI) to generate factor scores for three data modalities: neuropsychological (NP), magnetic resonance imaging (MRI), and cerebrospinal fluid (CSF). Factor scores from single or multiple modalities were entered in logistic regression models to predict MCI to AD conversion for 160 ADNI subjects over a 2-year interval.
Results
NP factors attained an area under the curve (AUC) of .80, with a sensitivity of .66 and a specificity of .77. MRI factors reached a comparable level of performance (AUC = .80, sensitivity = .66, specificity = .78), whereas CSF factors produced weaker prediction (AUC = .70, sensitivity = .56, specificity = .79). Combining NP factors with MRI or CSF factors produced better prediction than either MRI or CSF factors alone. Similarly, adding MRI factors to NP or CSF factors produced improvements in prediction relative to NP or CSF factors alone. However, adding CSF factors to either NP or MRI factors produced no improvement in prediction.
Conclusions
Latent factor scores provided good accuracy for predicting MCI to AD conversion. Adding NP or MRI factors to factors from other modalities enhanced predictive power but adding CSF factors did not.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Data availability statement
Data used in preparation of this article were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (http://adni.loni.usc.edu/). As such, the investigators with the ADNI contributed to the design and implementation of ADNI and/or provided data but did not participate in the analysis or writing of this report. A complete listing of ADNI investigators can be found at: https://adni.loni.usc.edu/wp-content/uploads/how_to_apply/ADNI_Acknowledgement_List.pdf
Supplementary material
Supplemental data for this article can be accessed here
Notes
1. PCA and EFA are often confused with each other. The produced index variables are called components in PCA and factors in EFA. The two methods differ in that PCA is aimed at accounting for most variance of the manifest variables without considering the latent structure of these variables, but EFA is meant to identify the number of latent variables and the latent structure that can explain the correlations between the manifest variables. Thus, PCA does not distinguish between the shared and unique variance of a manifest variable, but EFA does (Costello & Osborne, Citation2005).
2. This means that the MCI sample in ADNI is all amnestic-MCI (a-MCI). Note that a-MCI is the subtype of MCI that is at increased risk of converting to AD (Petersen, Citation2011).
3. One subject, who was diagnosed as AD at 12 months but converted back to MCI at 24 months, was removed from the final analyses.
4. Among the 92 MCINC subjects, only 6 reverted back to a not-impaired diagnosis within 24 months. Thus, the classification of MCI is very reliable in the ADNI 1 dataset.
5. The MRI data were acquired during a screening session, which occurred within 28 days of the baseline session. Although MRI data were also gathered during the baseline session, the screening session data were richer, and hence, those data figured in our analyses.
6. We thank an anonymous reviewer for suggesting CFA analyses in addition to EFA analyses.
7. Modification indices are the changes in chi-squared values if a certain path was added or a certain constraint was removed.
8. Notably, all variables loaded on this factor (entorhinal cortex, amygdala, hippocampus, and temporal pole) are involved in memory consolidation (Izquierdo & Medina, Citation1993; Landi et al., Citation2021).
9. Neuroinflammation refers to inflammatory responses within the central neural system (CNS), which are triggered by CNS insults, such as protein misfolding and aggregation. Recent evidence has suggested that excessive neuroinflammation can cause neuron damage and contribute to deterioration in brain diseases (Calsolaro & Edison, Citation2016; Heneka et al., Citation2015).
10. We followed the convention of setting the threshold at .5 in ROC analyses throughout the paper. Please note that the threshold can be adjusted to improve sensitivity or specificity at a relative cost to each other.
11. All p values of the DeLong’ tests were corrected for multiple comparisons using the Benjamini-Hochberg procedure (Benjamini & Hochberg, Citation1995).
12. We thank an anonymous reviewer for suggesting additional analyses based on the Jak/Bondi criteria (Bondi et al., Citation2014).
13. Again, one subject, who was diagnosed as AD at 12 months but converted back to MCI at 24 months, was removed from the final analyses.