Every night, Edward Damiano wakes three to four times to monitor his 11-year-old son’s blood sugar levels. Damiano administers insulin remotely through a pump when his son’s blood sugar reading is high or gives him juice through a straw when his blood sugar falls.
His son, David, who was diagnosed with Type 1 diabetes at 11 months old, sleeps peacefully through it all — and that’s exactly what worries Damiano.
“You can check his blood sugar all night long and he won’t wake up,” Damiano says. “Right now it’s a blessing. But when he goes to college, this is scary to me. I’m not going to be there.”
Damiano, a biomedical engineer at Boston University, is trying to build an artificial pancreas that could help people with Type 1 diabetes maintain healthy levels of glucose in the blood by the time his son leaves for college.
The device he’s working on, intended to mimic the exquisite sensitivity of the pancreas, relies on a computer algorithm that acts like a “brain” to calculate the precise dosage of two hormones, insulin and glucagon, that are needed at any one time. The software is sandwiched between existing technologies — a sensor that reads blood sugar every few minutes and a pump that administers the insulin and glucagon — with which the computer “talks” to dictate dosing.
“It’s mathematical; it’s objective,” said Firas El-Khatib, a biomedical research scientist at Boston University who has worked with Damiano since the beginning on the control algorithms. “It takes the human out of the loop.”
Around the world, roughly a dozen groups are testing similar technologies. Damiano’s team, further along than most, began testing its computer algorithm in humans in 2008 and is in its second round of in-patient human trials. Damiano says he hopes to be performing out-patient trials by 2012 and estimates that the device could be on the market by 2015.
Under normal conditions, levels of glucose in the blood are kept under tight control by insulin and glucagon: Insulin orchestrates the uptake of glucose by cells for use as energy, and glucagon counteracts that effect when blood sugar gets too low by helping the body access stored glucose from the liver.
But in Type 1 diabetes, the insulin-producing cells of the pancreas, known as beta cells, are destroyed by the immune system and can no longer produce the insulin, leaving the cells that produce glucagon to operate in the dark.
Damiano is convinced that the best artificial pancreas should incorporate both hormones — insulin for when blood sugar climbs too high and glucagon for when it falls too low.
One of the hitches, however, is that glucagon is not yet approved by the Food and Drug Administration for long-term use because it breaks down in solution. Several companies are tackling that problem. In the meantime, since a system that uses only insulin is likely to be FDA-approved sooner, Damiano’s team is working on an insulin-only system as well.
It is a common misconception that diabetes is fairly well managed with a few routine insulin injections throughout the day. In fact, maintaining healthy blood sugar levels — a life-or-death matter — is surprisingly difficult, and very few people with diabetes are able to do it effectively. The process is time-consuming and requires constant vigilance. And even those who monitor rigorously in some cases struggle with erratic blood sugar levels.
“The current treatments we have, although much better than they were, are really not solving the problem for the majority of people with Type 1 diabetes,” says Dr. Gordon Weir, a diabetes researcher at the Joslin Diabetes Center and professor of medicine at Harvard Medical School, who is not involved with Damiano’s research. “With current insulin treatment, we very rarely bring the sugar levels into the normal range. I think most, the majority, are in a range where damage is occurring.”
One of the challenges is that people with Type 1 diabetes walk a fine line between the immediate risks of hypoglycemia (low blood sugar) and the longer-term risks of hyperglycemia (high blood sugar). Even a slight overdose of insulin can cause a precipitous drop in blood sugar, leading to sweating, trembling and confusion — and, if the situation persists, to loss of consciousness, seizures, even death.
Since low blood sugar is life-threatening, many people with diabetes tend to be conservative and allow their blood sugar to remain on the high end. But elevated blood sugar for prolonged periods causes insidious damage to the nerves, kidneys, eyes and heart. Diabetes is the leading cause of blindness, kidney failure and amputations, and it dramatically increases the risk of heart disease.
“They’ve got the complications down the road on one hand and the immediate danger of hypoglycemia on the other hand, and then this tremendous workload to try and steer the course between these two extremes,” says Dr. Steven Russell, an endocrinologist at Massachusetts General Hospital who works with Damiano on designing clinical trials for the artificial pancreas.
On top of all that, there is no constant guideline for correct insulin dosing.
The proper amount is a constantly moving target, depending on such factors as the amount of sleep, size of a meal, time of day, state of mind, exercise and more. The “right” insulin dosage in the morning after someone’s just eaten a bagel, for example, may be far too much insulin to deal with the same bagel eaten later in the day. Exercising — requiring glucose uptake by the muscles — can cause a dramatic crash in blood sugar.
Plus, with current technologies, there’s a significant delay between the time when insulin is given and the time it takes to get into the blood. And some people absorb insulin more slowly than others.
“Trying to maintain tight blood glucose control is like having a second job,” Russell says. “It’s mind-numbing work to check blood glucose all the time and make little adjustments.
“But,” he adds, “that’s the perfect sort of thing for a computer.”
Today’s standard of care for people with Type 1 diabetes involves monitoring blood sugar about five times a day. In comparison, the artificial pancreas collects glucose data every five minutes. Dosing decisions are based on a continual flow of information: If a person eats a piece of cake, the artificial pancreas quickly sees the blood sugar rise and quickly adjusts the insulin and glucagon it delivers.
So far, Damiano’s team has tested its algorithm in 15 people in one- to two-day experiments. The first trial, in which they tested adults for 27-hour stretches, demonstrated that safe and effective glucose control was feasible with the two-hormone artificial pancreas. In the second trial, currently underway, they are testing the system in children and adults for 51 hours and have included an exercise component. (Since exercise can lead to increased risk of hypoglycemia, this adds an additional level of challenge to the algorithm’s decision-making process.) Because the trial is ongoing, the team is hesitant to draw early conclusions, but Damiano says that they are very encouraged by the results.
Though better than what diabetes patients experience now, this is still a long way from being as sensitive and efficient as the body of someone without diabetes. In a healthy pancreas, the beta cells read blood glucose levels and secrete insulin continuously, as needed, and the insulin is delivered directly to the veins. With an artificial pancreas, insulin is delivered just under the skin into the fat and is slowly absorbed by the blood. This can lead to a one- to three-hour delay in insulin absorption.
In addition, there are some significant technical hurdles that must be overcome. If a person lies down on the side in which the sensor is placed, the pressure can prevent accurate glucose readings. The sensor requires careful calibration. And the human trials so far have revealed that people vary greatly in the rate of insulin absorption — what’s needed, Damiano says, is a faster-absorbing insulin.
“I feel like I’m in this race,” he says. “I’ve got this seven-year ticking clock.”