The drug -- still in clinical trials -- is part of a class called gamma-secretase modulators (GSM), but until now, very little was known as to why the molecule appears to work.
GSMs' target is the accumulation of amyloid beta protein, which forms plaques in the brain and eventually leads to the tragic mind-wrecking disease, symptomised by forgetfulness and dementia.
In a study published in the British journal Nature, a team of 29 US and European researchers found that GSMs had a remarkable dual role.
The drugs not only reduced production of long pieces of amyloid beta that cluster together, but also increased production of short pieces of the same protein that prevent the longer pieces from sticking together.
"In a very general sense, the action of these GSMs on amyloid beta might be analogous to some cholesterol-lowering drugs that can lower LDL, the bad cholesterol that sticks to your arteries, and can raise HDL, the good cholesterol that sweeps out LDL," said senior author Todd Golde, a neuroscientist at the Mayo Clinic in Jacksonville, Florida.
The investigators also found that the GSMs stick to amyloid beta that is already in the brain, keeping it from aggregating.
Amyloid beta is created by an enzyme that chops up a larger protein called APP.
But the teams made the surprising discovery that GSM did not, as had been thought, work by targeting the enzyme. Instead, it works by targeted the structure, or substrate, of the protein itself.
This is good news, as amyloid beta was not previously believed to be "druggable," said Golde in a press release.
"This broadens the notion of what drugs can do and therefore has wide-reaching implication for future drug discovery for many different disorders."
Doctors are in the third and final phase of trials on human volunteers to test the first GSM, a molecule called tarenflurbil, branded as Flurizan.
Several more drugs in the class are set to enter human trials in the next year or two.
Alzheimer's is a disease that is related to age but also has a strong inherited component, with at least three variants of genes among the potential causes.
The number of people worldwide with Alzheimer's is set to rise from 24 million people today to 42 million in 2020 and 81 million in 2040, according to some estimates.