In the study, PET data were analyzed with an automated computer program that performed an observer-independent statistical comparison with an age-matched reference database. This approach has proved more accurate than observer-dependent approaches, the investigators said.

The longitudinal, prospective study included 30 patients who met the established diagnostic criteria for mild cognitive impairment but who had no diagnosable cerebral or systemic cause for the condition. Thus, this group likely contained a high proportion of pre-Alzheimer's patients, the researchers said. The group consisted of 16 women and 14 men, with an average age of 70.
All patients underwent neuropsychological evaluation, a PET imaging scan, and genetic screening for the APOE-e4 allele, which has been identified as a major susceptibility gene. About 16 months later, patients underwent a second neuropsychological evaluation to determine who had progressed to Alzheimer's dementia.

PET scans suggested a likelihood of Alzheimer's dementia in 13 patients, and 11 of these progressed to dementia by the end of the study. Of the 17 patients with no suggestive PET findings, only one progressed to dementia. This test had a sensitivity of 92%, a specificity of 89%, and a test accuracy of 90%. The positive predictive value was 85%.

Genetic tests identified susceptibility in 17 of the patients, and nine progressed to clinical dementia by the end of the study. Of the 13 patients without genetic susceptibility, three progressed to dementia. Thus, the researchers concluded this test had a sensitivity of 75%, a specificity of 56%, and a test accuracy of 63%. The positive predictive value was 53%.

Eight patients had both suggestive PET scans and genetic susceptibility, and eight patients had neither. All patients with both progressed to dementia, while all patients with neither remained clinically stable. Of the 14 patients with either a suggestive PET scan or genetic susceptibility, only four progressed to dementia. The researchers calculated that combining both tests yielded a sensitivity of 67% (95% CI=35-90), a specificity of 100% (95% CI=81-100), and a test accuracy of 87%. The positive predictive value was 100%.

Combining both tests may be especially useful for identifying subgroups of patients with especially high or especially low risk, the investigators said.

"The current study underlines that a genetic disposition does not necessarily represent a determined prognosis, thus, the need for measures that allow the definition of the actual onset of a disease process is apparent," Dr. Drzezga said. "Molecular imaging could play an important role in this context."