STANFORD, Calif. -- A way to predict the development of Alzheimer's disease years before its clinical onset has been described by investigators here.
STANFORD, Calif., Oct. 15 -- A way to predict the development of Alzheimer's disease years before its clinical onset has been described by investigators here.
A group of 18 cell-signaling proteins in blood plasma, involved in inflammatory and immune processes, appears to distinguish Alzheimer's from controls with close to 90% accuracy, and could one day offer a predictive diagnostic test, according to its developers.
The experimental protein panel, if verified in additional tests, could also predict which patients with mild cognitive impairment may progress to Alzheimer's-type dementia within six years, reported Tony Wyss-Coray, Ph.D., of Stanford, and colleagues, online in Nature Medicine.
"Our technology enables us to 'listen' to the chatter of cells communicating with each other and determine if there's anything abnormal," Dr. Wyss-Coray said. "Our data indicate blood contains a highly specific, biological signature that can characterize Alzheimer's disease years before a clinical diagnosis can be made.
The authors, speculating that it might be possible to screen for the early signs of Alzheimer's disease by looking for changes in the concentrations of cell signaling proteins, collected 259 stored plasma samples from patients who later went on to develop Alzheimer's disease, as well as those with mild cognitive impairment, late-stage Alzheimer's, and non-demented controls.
They measured levels of 120 known signaling proteins, and then used the data to develop training sets of microarrays for predictive analysis, which eventually yielded a panel of 18 proteins.
They then set about to see whether the predictive protein microarrays could distinguish between molecular Alzheimer's or non-Alzheimer's phenotypes in samples blinded to analysts. The samples included plasma from patients with Alzheimer's disease and other types of dementia, and from non-demented controls.
They found that predictive analysis of microarrays correctly classified samples as Alzheimer's disease with 90% accuracy, and ruled out Alzheimer's correctly in 88% of samples.
Samples from eight of nine patients with Alzheimer's disease confirmed on autopsy were classified correctly, and 10 of the 11 other dementia samples received a "non-Alzheimer's" classification.
In unsupervised test runs, the 18 markers were able to sort Alzheimer's samples from those of non-demented controls when the two were mixed or "clustered" on test panels.
"Similarly, unsupervised clustering based on the 18 predictive signaling proteins led to a good separation of all Alzheimer's samples from the plasma samples of individuals with other neurological diseases or with rheumatoid arthritis," the authors wrote.
To see whether the protein panel could predict whether patients with mild cognitive impairment could progress to Alzheimer's disease, the authors applied it to an analysis of samples from patients with mild cognitive impairment at the time of diagnosis in two published cohort studies.
They found that the predictive analysis of microarrays correctly classified 20 of 22 patients with mild cognitive impairment who progressed to Alzheimer's two to five years after diagnosis, and correctly ruled out Alzheimer's in eight patients with mild cognitive impairment who went on to develop other forms of dementia.
"I really think it has enormous potential," said Lennart Mucke, M.D., director and senior investigator of the Gladstone Institute of Neurological Disease at the University of California San Francisco, who was not involved in the study. "Most researchers in this field agree that there is an urgent need for better lab tests for Alzheimer's disease, and this study has addressed this need admirably."