In 2000, when doctors at the University of Alberta in Edmonton announced that they had successfully transplanted pancreatic islets into diabetic patients, people around the world began clamoring for the therapy. Because islets secrete the right amount of insulin at the right time, they can control the concentration of glucose in the blood far more precisely than a patient can do, even with five or six injections a day. But islets must be harvested from suitable organ donors, and there just are not enough to go around.

Now another group, led by Alex Rabinovitch at the same university, is targeting a far better class of donors: the patients themselves. By administering chemical signals, or growth factors, they hope to provoke the patients' pancreas to grow its own beta cells, which are the insulin-making engines of the islets. It would not be the first such use of a growth factor; for instance, erythropoietin stimulates the production of red blood cells. This time, however, the object is not to speed up a process but to restart one that has stopped.

In type 1 diabetes, the beta cells essentially die off, victims of a ferocious autoimmune reaction. But in the late 1970s autopsies revealed that a very few beta cells continue to bud off from the pancreatic ducts, apparently reproducing the original process by which these cells are born. Diabetologists had long regarded duct cells as mere chaff, to be removed from the islet tissue they were trying to transplant. Now, however, mounting evidence suggests that that chaff may be pure gold.

"Our colleagues in surgery have a paper saying that patients with the best insulin response to islet transplantation were those whose islets were most 'contaminated' with duct cells," Rabinovitch says. Why do patients benefit from added duct cells, seeing that they have plenty of their own intact? He does not know for sure but suspects that the duct cells need to be cheek by jowl with the islets to receive chemical feedback. He has been studying many possible growth factors and has struck pay dirt with two of them.

One is gastrin, which is made in the stomach, and the other is epidermal growth factor, or EGF, an all-purpose growth chemical. Their role in beta-cell generation was first suggested in 1993 by Stephen Brand and his colleagues at Harvard Medical School. They found an unusually large number of beta cells in mice with mutations that caused them to churn out more than the usual amount of the two factors. Rabinovitch reproduced the result by administering the factors to a strain of mice bred to model autoimmune diabetes.

In newly diabetic mice, the results were spectacular. After two weeks of treatment, the beta-cell mass had tripled, the insulin content had increased eightfold, and the blood glucose concentration had fallen to normal levels. It did not go so well in mice with long-standing diabetes: their glucose levels fell substantially, yet they needed some immunosuppressive medication. That drug regimen made possible islet transplantation in the first place, but because it can damage the kidneys, it is suitable only for patients with dangerously uncontrolled diabetes. Even a partial regeneration in patients would at least get around the shortage of donor tissue and would probably require less immunotherapy; native cells would need defense from autoimmunity alone, not from that and organ rejection combined.

Phase II human trials of gastrin and EGF began in August. They should yield results on safety and efficacy by the spring of 2006. The enrolled subjects include those with both type 1 and the more common type 2 diabetes, in which there is no autoimmune reaction and the islets make insulin, but not enough.

A great result, Rabinovitch says, would be if the type 2 patients no longer needed any insulin and if the type 1 patients could get by with less. A little of the body's own insulin goes a long way, because beta cells that produce it react so sensitively to the momentary rise and fall of blood glucose. The better the glucose control, the better the patient's chances will be for avoiding nerve damage and other long-term complications of diabetes.