INTERLABORATORY COMPARISON OF NEUROPATHOLOGY ASSESSMENTS IN ALZHEIMERS-DISEASE - A STUDY OF THE CONSORTIUM TO ESTABLISH A REGISTRY FOR ALZHEIMERS-DISEASE (CERAD)

Concerns about intercenter variation in methods and interpretation pro mpted CERAD investigators to examine standardization of the neuropatho logical assessment of Alzheimer's disease (AD). Contiguous frontal lob e sections derived from autopsy brains of eight patients clinically di agnosed as having probable AD and two cognitively normal individuals w ere distributed to 24 neuropathologists from 18 medical centers in the United States and Canada. Using their routine staining method(s), neu ropathologists determined the rank order of severity of AD neuropathol ogy in these cases, as well as semiquantitative and quantitative senil e plaque and neurofibrillary tangle frequencies. Ranking of the ten ca ses revealed 75% inter-rater reliability among the 24 raters. Semiquan titative analyses showed reasonable inter-rater agreement, whereas qua ntitative measures yielded significant differences between raters for plaque and tangle counts (p < 0.0001). These differences reflected var iation in stain sensitivity, staining technique (even when the same st ain was used), and interpretation of the histological findings. Rating s on the cases with the highest proportions of diffuse plaques showed the greatest dependence upon stain sensitivity and variability in inte rpretation. This study indicates that greater attention to quality imp rovement is needed for the neuropathological evaluation of AD, particu larly when pooling data in multicenter studies such as CERAD.