With the most common Type 2 form, the body gradually loses its ability to use insulin, so the confused pancreas churns out extra until eventually its cells wear out. Most at risk are the overweight.
Genetics research is showing diabetes is far more complicated than those simple demarcations:
—First, there is MODY, shorthand for six different subtypes thought to account for 2 percent of all diabetes. Each is caused by a single, different gene. Suspicions arise when patients are extra hard to treat, especially skinny people diagnosed with Type 2 diabetes or young adults with diabetic relatives who abruptly seem to develop Type 1.
Consider Dan Humphries of Shawbury, England, who at age 16 was diagnosed with Type 1. His mother, a nurse with diet-controlled diabetes, questioned the diagnosis. But doctors insisted he was too skinny for other diabetes. They prescribed insulin that had Humphries passing out from low blood sugar even with small doses.
His mother sought out Britain's Peninsula Medical Center in Exeter, where Hattersley performed a gene test that showed Humphries' pancreas could make its own insulin. But a gene called HNF1-alpha was essentially putting that production to sleep.
Over a decade of research, Hattersley had found that old diabetes drugs called sulfonylureas neutralize that gene so insulim production resumes. Sure enough, Humphries, now 19, is fine with a quarter-tablet morning and night.
—After that are the 16 genes that are discovered so far to play a role in Type 2 diabetes, and at least 14 in Type 1.
Surprisingly, the Type 2 genes do not affect how the body uses insulin, thought to be the trigger. Instead, they alter how the pancreas makes insulin to start with, explains Dr. David Altshuler of Harvard University and the Massachusetts Institute of Technology.
So how healthy the pancreas begins could determine how vulnerable the patient is to other diabetes triggers, like obesity.
Collins points to one potential drug target: A gene with the sole job of getting zinc to insulin-creating cells. Zinc is a key part of the recipe; too little or too much, and insulin is not secreted.
But randomly choosing a gene to target is ''a shot in the dark,'' cautions Eric Schadt of Merck & Co., who will urge another approach at this week's meeting, hosted by the National Disease Research Interchange.
Monday in the journal Nature, Schadt reports finding how multiple genes work together in computer-like networks that suggest which will be master control switches, and thus good drug targets. Already, Merck has begun checking whether one network of obesity genes really might predict which overweight people will contract diabetes.