The finding by an international team of researchers is set to be the most influential gene discovery for Alzheimer's in the last 20 years.
It could have major implications for understanding the causes of the disease and may lead to new treatments.
Researchers led by John Hardy, from the Institute of Neurology at University College London used new sequencing techniques to home in on the TREM2 gene.
Additional TREM2 sequencing was then performed, in part, by scientist Aleksandra Wojtas.
These studies led to identification of a set of rare variants in TREM2 that occurred more often in 1,092 Alzheimer's patients than in a control group of 1,107 healthy people.
The most common variant, R47H, was then evaluated in follow-up studies of a large number of Alzheimer's disease patients and controls.
More From This Section
Researchers spearheaded the direct genotyping and analysis of R47H in DNA samples from 1,994 Alzheimer's disease patients and 4,062 "control" participants - individuals verified not to have Alzheimer's.
Studies showed unequivocally that the R47H variant of TREM2 substantially increases the risk of Alzheimer's disease.
"The TREM2 variant may be rare, but it is potent. In our series, it was present in 1.9 per cent of the Alzheimer's patients and in only 0.37 per cent of the controls," Minerva Carrasquillo, a neuroscientist with the Mayo Clinic.
"R47H isn't fully penetrant - meaning that not all people who have the variant will develop Alzheimer's and in those who do, other genes and environmental factors will also play a role - but like APOE 4 it does substantially increase risk," said Carrasquillo in a statement.
"R47H is the first goldilocks variant to show strong association with Alzheimer's disease," researcher, Steven Younkin said.