Hypometabolism in PiB-positive, population-based, cognitively normal subjects with and without APOE-ɛ4

Abstract

Pittsburgh compound B (PiB) accumulation occurs in about a third of cognitively normal (CN) subjects. The relationship of neurodegenerative biomarkers, namely FDG PET, and PiB positivity in CN subjects is not well understood. We hypothesized that subtle hypometabolism may occur in close temporal association with the early findings of PiB positivity. Therefore, we assessed hypometabolism in different stages of PiB positivity in CN subjects and related it to APOE status. 617 CN subjects in the Mayo Clinic Study of Aging were analyzed. We used two methods to identify any hypometabolic changes in PiB positive subjects. First, we used penalized logistic regression and cross-validation to develop a parsimonious multivariable region of interest model of “optimal” cortical brain regions of hypometabolism associated with PiB positivity. This method was first validated in our CN and AD populations and then employed in CN PiB negative and positive subjects. Second, we used a voxel-wise approach (SPM) to look at hypometabolism in various PiB-value subsets (dose-finding approach) namely those CN subjects with PiB values < 1.4 (n=351), >1.4 to 1.48(n=66), >1.48 to 1.73(n=66), >1.73 to 205(n=66), and >2.05(n=66). In subjects with PiB < 1.4, APOE -vs+ were compared for metabolic differences. Our penalized logistic regression model demonstrated hypometabolism in signature AD regions, namely the angular gyrus and posterior cingulate, in a cross-validated multivariable ROI model (AUC=0.65) in the CN PiB positive subjects. SPM analysis demonstrated a trend (non-FDR, P=0.005) for hypometabolism in these same regions beginning at the 1.48–1.73 PiB-value subset that became most significant in these regions but with additional, more diffuse cortical regional hypometabolism in the >2.05 subset (non-FDR, P=0.005). (Figure 1a) No hypometabolism was seen in APOE positive subjects (PiB <1.4) as compared to negatives at this same significance level (p=0.005).(Figure 1b) Hypometabolism in PiB positive, CN subjects in AD-signature cortical regions was seen at low levels of PiB positivity suggesting that neurodegeneration and related hypometabolism occurs early in the amyloid deposition process. No independent effect of APOE e4 carriage alone on hypometabolism was seen, although APOE e4 carriage was associated with greater PiB positivity. Pattern of hypometabolism seen in CN PiB positive subjects vs. CN PiB negative subjects (a) and APOE+ vs – PiB negative subjects (b) by SPM analysis.