The team identified 18 proteins in particular whose abundance could distinguish those with Alzheimer's disease from those without it, for an overall accuracy of about 90 percent -- that is, it correctly classified individuals who had been clinically diagnosed with Alzheimer's disease 95 percent of the time and classified as negative those without the disease 83 percent of the time.
This panel of proteins was equally effective when applied to another, completely separate set of patient samples, the researchers noted.
But "the home run of the paper," said Levey, was the finding that, when applied to blood samples collected from patients who were diagnosed with mild cognitive impairment -- a condition that often, but not always, precedes Alzheimer's -- the panel could predict who would ultimately develop Alzheimer's with 81 percent accuracy, 30 months before clinical diagnosis, on average.
Calling the study "very interesting," Levey nevertheless noted two caveats. The first was its relatively small sample size. The other was its use of proteins that have no obvious relationship to Alzheimer's.
"The blood has thousands of proteins, and they started with 120 proteins that they could measure," he said. "I don't think if one were to try to make a biomarker for Alzheimer's that you would necessarily choose these 120 proteins."
Wyss-Coray agreed that the team's decision to focus specifically on signaling proteins might seem like "a bit of a crazy idea." But given the disease's target organ, it makes sense, he said.
"Cells receive input through hundreds of different receptors, and it responds with some output, usually a signaling protein," Wyss-Coray explained. "So, by thinking in terms of this output, we decided to look specifically at signaling proteins and see if there are changes between patients who are healthy versus those with Alzheimer's disease or other dementias."
Others have also m
All rights reserved