18F-AV-1451 Autopsy Study

Sensitivity and specificity of 5 independent readers' interpretations of ante-mortem flortaucipir PET imaging for detection of a pattern of flortaucipir neocortical uptake that corresponds to neurofibrillary tangles (NFT) Score of B3 (Hyman et al., 2012; Montine et al., 2012). NFT B scores range from B0 (Braak Stage 0; no NFTs in the brain) to B3 (Braak Stage V/VI; widespread NFTs in the brain). Sensitivity and specificity are percentages that can range from 0 to 100%. The hypothesis tested was that, of the 5 independent imaging physicians, at least 3 will have the lower bounds of 2-sided 95% CIs ≥50%, for both sensitivity and specificity.

Sensitivity and specificity of 5 independent readers' interpretations of ante-mortem flortaucipir imaging for detection of a pattern of flortaucipir neocortical uptake that corresponds to high levels of Alzheimer's disease neuropathologic change (High ADNC) as defined by National Institute on Aging-Alzheimer's Association (NIA-AA) criteria. ADNC categories are None, Low, Intermediate and High, with High indicating the most severe level of AD-related pathology changes in the brain (Hyman et al., Alzheimers Dement. 2012 Jan;8(1):1-13). The hypothesis tested was that, of the 5 independent imaging physicians, at least 3 will have the lower bounds of 2-sided 95% CIs ≥50%, for both sensitivity and specificity.

Sensitivity and specificity of majority interpretation of AD pattern tau PET scan corresponding to NFT Score of B3. The 95% confidence intervals (CI) provided for specificity and sensitivity were based on the Wilson score method. The hypothesis tested was that majority read results had the lower bound of the 2-sided 95% CI greater than or equal to 55% for both sensitivity and specificity.

Sensitivity and specificity of majority interpretation of of AD pattern tau PET scan corresponding to NIA-AA autopsy diagnosis. The 95% CIs provided for specificity and sensitivity were based on the Wilson score method. The hypothesis tested was that majority read results had the lower bound of the 2-sided 95% CI greater than or equal to 55% for both sensitivity and specificity.

Fleiss' Kappa statistics were used to assess inter-reader agreement for the diagnostic decisions associated with primary outcome 1. Fleiss' kappa is a statistical measure for assessing the reliability of agreement between a fixed number of raters when assigning categorical ratings to a number of items or classifying items. Fleiss' kappa can range from 0 to 1 with 1 indicating perfect agreement between the readers. Results are reported as overall agreement, calculated as proportion of scans where reader pairs agreed, divided by the total number of scans read by each reader pair.

Sensitivity and specificity of 5 independent readers' interpretations of ante-mortem flortaucipir PET imaging for detection of a pattern of flortaucipir neocortical uptake that corresponds to neurofibrillary tangles (NFT) Score of B3 (Hyman et al., 2012; Montine et al., 2012). NFT B scores range from B0 (no NFTs in the brain) to B3 (widespread NFTs in the brain). Sensitivity and specificity are percentages that can range from 0 to 100%. For this analysis, B scores of B2-B3 were considered truth positive, and B scores of B0-B1 were considered truth negative.