Principal Components Analysis of Brain Metabolism Predicts Development of Alzheimer Dementia

Article Figures & Data

Download figure
Open in new tab
Download powerpoint
FIGURE 1.
Patterns of regional brain metabolism. (A) ADCRP derived by PCA (P < 0.05) and (B) significant regions derived by SPM t test (familywise error–corrected P < 0.05), overlaid on MRI template image. Voxels with negative region weights and hypometabolism are shown in “cool” colors, and regions with positive region weights and hypermetabolism are depicted in “hot” colors. Data are presented in neurologic orientation.
Download figure
Open in new tab
Download powerpoint
FIGURE 2.
Hazard ratios for different predictors, penalized by ridge regression to suppress effects of multicollinearity among them. Normalized ¹⁸F-FDG uptake stems from linear combination of normalized ¹⁸F-FDG uptake in VOIs with significant hypometabolism and that in VOIs with significant hypermetabolism. APOE reference: APOE positive; sex reference: female. FAQ = FAQ total score; N = number of subjects; n. = normalized. All continuous variables were z transformed.
Download figure
Open in new tab
Download powerpoint
FIGURE 3.
Kaplan–Meier curves for test dataset. (A) Risk strata determined using PES values alone. (B) Risk strata determined using PI derived from combined model (including PES values). Med. = medium.

TABLE 1

Clinical and Demographic Characteristics of Training and Test Datasets

TABLE 2

Characteristics of Cox Regression Models

TABLE 3

Separation of Risk Strata by Different Models

	Hazard ratio			Median time to conversion (mo)			Pairwise log-rank P value
Parameter	Low risk	Medium risk	High risk	Low risk	Medium risk	High risk	Low vs. medium	Medium vs. high	Low vs. high
PES values alone	1	4.62	9.70	120	68	36	1.2 × 10⁻⁵	0.0007	3.0 × 10⁻¹³
PI derived from combined model (including PES values)	1	4.75	15.92	120	96	32	8.3 × 10⁻⁸	1.2 × 10⁻⁵	2.0 × 10⁻¹⁶