Toward a definite diagnosis of Alzheimer's disease

There is a significant, and growing, number of patients who suffer from dementia of the Alzheimer's type (DAT). However, due to clinical variability and symptom overlap, and despite millions of dollars spent in diagnostic work-ups, the diagnosis of DAT remains one of exclusion or by neuropathologic study (usually postmortem). This report evaluates two promising methods, the newly refined clinical criteria sets (e.g., NINCDS-ADRDA) and single-photon emission computed tomography (SPECT), on their capacity for definite, accurate, and early diagnosis of DAT patients. We found that clinical diagnosis of DAT can achieve a weighted average for sensitivity of 72%, specificity of 86%, and diagnostic confidence of 72%; however, when NINCDS-ADRDA criteria are used, these results improve to 88%, 91%, and 92%, respectively. SPECT scans could differentiate DAT patients from normals with 86% sensitivity, 96% specificity, and 98% diagnostic confidence. Comparable figures for DAT versus multi-infarct dementia (MID) are 82%, 81%, and 86%, respectively. We conclude that, although a definite clinical diagnosis of DAT is still elusive and more research is needed, the use of either NINCDS-ADRDA criteria or SPECT scans may enhance diagnostic accuracy.