O3‐12‐03: Why we need two cutoffs for amyloid imaging: Early versus Alzheimer's‐like amyloid‐positivity

Our group and others have spent considerable effort to define a quantitative cutoff between PiB(+) and PiB(-) subjects. This cutoff has typically been set to divide cognitively normal subjects with no amyloid deposition from the 25-50% of controls in whom the process of amyloid deposition had already begun. These PiB(+) controls typically have lower cortical PiB retention than observed in AD. Given the frequent presence of clinically silent amyloid deposition in these controls, it seems likely that a similar percentage of non-AD dementia patients could have clinically silent amyloid deposition. That is, this amyloid may have led to AD years later, but their current clinical dementia is due to another pathophysiological process. Therefore, a second, higher cutoff is probably necessary to ascribe the underlying pathology of a clinical dementia syndrome to AD. The same may apply to MCI. PiB-PET data was collected in 62 cognitively normal controls (69±12 yrs), 45 AD subjects (71±11 yrs) and 54 MCI subjects (69±9 yrs). Tissue ratios were calculated for cortical regions-of-interest and normalized to cerebellum (SUVR; 50-70 min). A global cortical average was calculated. An “early-amyloid” cutoff was calculated by a cluster analysis (k=2) approach using only the 62 cognitively normal controls. An “AD-like” cutoff was determined using the 62 controls plus the 45 AD subjects (k=3). Using only the 62 cognitively normal controls, a global cortical “early-amyloid cutoff of 1.53 SUVR units was calculated. After addition of the 45 AD subjects, cluster analysis yielded an AD-like cutoff of 1.80 SUVR units. When the 1.53 “early-amyloid” cutoff was applied, 10/62 (16%) controls, 26/54 (48%) MCI subjects and 45/45 (100%) AD were early-PiB(+). When the 1.80 AD-like cutoff was applied, 6/10 (60%) controls, 21/26 (81%) MCI and 42/45 (93%) AD subjects who were early-PiB(+) remained AD-like-PiB(+). Cutoffs designed to detect the very first signs of amyloid deposition in cognitively normal controls may be too liberal to accurately ascribe the underlying pathophysiology of a clinical dementia syndrome to AD. A more conservative cutoff may be necessary to increase the specificity of amyloid imaging for the identification of clinical dementia (or even MCI) syndromes caused by AD.