A comparison of Aβ amyloid pathology staging systems and correlation with clinical diagnosis

Abstract

Current neuropathological Alzheimer's disease (AD) criteria from the National Institute on Aging-Alzheimer's Association (NIA-AA) incorporate two staging systems for Aβ pathology, namely the Thal Aβ phase (TAP) and the Consortium to Establish a Registry for Alzheimer's Disease (CERAD) methods. The goal of this study was to compare and contrast results obtained with these two different staging systems for Aβ pathology since this is critical for future correlations of Aβ amyloid imaging data with Aβ neuropathology data based on immunohistochemical detection of Aβ deposits. A total of 123 cases, divided into 82 training and 41 validation cases, with a diagnosis of either unremarkable adult brain (normal) or AD and CERAD scores ranging from none to frequent were included. There was no clear and consistent relationship between CERAD and the TAP Aβ scores with the exception of scores for the highest plaque burdens (i.e., CERAD C3 and TAP A3) in the cases studied here. However, we developed an algorithm that relates CERAD scores to TAP scores with high agreement (94 % in training and 98 % in the validation set). In addition, TAP scores were a better predictor of dementia (sensitivity of 94 % specificity 87.7 %) than CERAD scores (sensitivity of 57 % specificity 100 %). Yet, further research is needed to define strategies to relate CERAD and TAP Aβ plaque scores to compare their utility and for determining the clinical associations of these different amyloid staging systems with aging and AD.

Fig. 1

Comparison of agreement between CERAD C and TAP A scores. In AD and normal patients, the overall agreement between CERAD C and TAP A scores is moderate (kappa = 0.57). When assessing individual groups, CERAD C3 has very good agreement (100 %) with TAP A3, while CERAD C1 and C2 cases corresponded to a diverse range of TAP A values

Fig. 2

Algorithm for the conversion of CERAD to TAP scores of subjects with AD neuropathologic change. CERAD C3 can be extrapolated readily to TAP A3. If CERAD C is <3 and Braak B is 3 or the patient was demented, the score can be extrapolated to TAP A3. For CERAD C2 cases, Aβ IHC of the midbrain will differentiate between TAP A3 and TAP A1 or A2 groups. For CERAD C0, Aβ IHC in amygdala and frontal cortex will differentiate between TAP A0 and TAP A1 or A2. CERAD C1 cases require Aβ IHC to be performed on all six brain regions discussed in the text to relate CERAD C1 to a TAP A score