Anatomy of a CGM Sensor
EEEP! EEEP! EEEP! The low-glucose alarm blaring from your continuous glucose monitor at 3 a.m. is a startling, but welcome, sleep disruption. Such lows, if left untreated, are a serious concern for those using insulin pumps or taking multiple daily insulin injections. Study after study has shown that wearing a continuous glucose monitor (CGM)—a device that measures glucose levels every few minutes—can help people who use the data regularly avoid hypoglycemia while, at the same time, lowering average blood glucose levels. This is true for both adults and children with type 1 diabetes (there is much less research on CGM use among people with type 2).
CGMs are a relatively new tool in the diabetes arsenal, and arguably the most advanced. They were a long time coming, requiring manufacturers to overcome a variety of technological and regulatory hurdles. Today, the companies that make CGMs invest considerable effort in constructing devices that provide reliable results, with an eye toward innovations that improve the lives of people with diabetes.
The CGM consists of three basic parts: the wireless monitor, often called a receiver; the transmitter; and the sensor. The handheld monitor, about the size of a cell phone, has a screen where you can check your current glucose level, look at historical data, and get trends about whether glucose is likely to go up or down, and how fast. The transmitter is a simple device—about the size of a quarter—that hooks into the sensor and streams glucose information over radio waves to the monitor.
Compared with the sensor, the hardware for the transmitter and monitor is relatively standard. The sensor, the most sophisticated piece of the CGM, is thinner than a needle and about half an inch long. The sensor is isnerted just under the skin, where it remains in place for several days, detecting glucose in the surrounding fluid.
The sensor uses the same enzyme to measure glucose levels as a test strip: glucose oxidase. This enzyme converts glucose to hydrogen peroxide. The peroxide reacts with platinum inside the sensor, generating an electrical signal that travels through a tiny wire to the transmitter. A computer program in the CGM converts the electrical signal into a glucose reading. These basic features are shared by all CGM sensors. The chemical layers on top of the glucose oxidase keep the sensors functional under the very poor working conditions that exist inside the body.
Both companies that currently make CGMs for the U.S. market, Dexcom and Medtronic, started their CGM journeys over a decade ago. The initial goal was to create a glucose-measuring device that could remain in the body for up to a year. The main obstacle, it turned out, was the immune system. “To keep the thing active in the body for a significant period of time is a challenge,” according to Steven Pacelli, executive vice president of strategy and corporate development at Dexcom.
All day, every day, the immune system is on the hunt for foreign agents in the body to destroy. Normally, that’s a good thing, as it wipes out viruses, unhealthy bacteria, and even cancer cells. But it’s bad news for a glucose sensor that the body sees as an invader. The super-secret ingredients in CGM sensors are the coatings that help persuade the immune system to leave the sensor alone. “The key to the intellectual property is tricking the body,” says Pacelli. “Ultimately, the body wins.”
If You Build It
With some restrictions on what could be shared, Medtronic allowed me to visit one of its sensor manufacturing plants, located about an hour north of Los Angeles among scenic hills. (Dexcom would not allow access to its CGM manufacturing process.) Before I could set foot in the plant, I had to wash my hands plus use a hand sanitizer, and then don a shower cap, booties, and a not particularly flattering gown.
The Medtronic facility occupies 12,000 square feet and employs 300 people who work around the clock, in shifts, to produce Medtronic’s two sensor types: the Sof-sensor and the newer Enlite sensor. The sensor-making process has three steps: fabrication, assembly, and sterilization.
Layer by Layer
The sensors are built on top of what Bahar Reghabi, director of the CGM Value Stream at Medtronic and my CGM tour guide, calls a wafer—a translucent piece of beige plastic about the size of two credit cards. Using metal in place of ink, a machine prints two rows of electrodes onto the wafers. Each electrode, resembling a trident-shaped gold wire, will make up the core of one sensor.
The next phase of sensor fabrication involves layering chemicals onto the wafer-bound electrodes. First, the electrodes are coated in glucose oxidase, using chemistry that fixes the enzyme on the sensor, says Reghabi, so it can’t escape into the body. On top of the glucose oxidase, the manufacturers “paint on” a series of chemicals that help camouflage the sensor from the immune system. “This complex layer of membranes is laid on top of glucose oxidase to keep it active for a long length of time,” says Pacelli. Once all the layers are in place, each sensor is intricately carved from the wafer with an automated laser.
Putting It All Together
Next to the sensor chemistry laboratory, a couple of dozen gowned and goggled people lean over microscopes, ready for the next batch of sensors. They use tweezers to carefully insert the fragile sensors into tiny protective plastic tubes. Then the sensors move on to the next room, where another group of people, also using microscopes, fit the sensors into their bases. The sensor base is a purple piece of plastic that protects the sensor, makes it easier to handle, and connects to the transmitter. As the sensor is attached to the base, it is pushed inside a hollow needle that will help guide it into the body (the needle is removed after insertion).
The sensor base is backed with several layers of material that will help the sensor stick to the body. The finished sensors are packaged and sterilized. One neat trick is that the sensor base turns from purple to gray after sterilization, providing a visual cue that a product is safe for consumers.
With an initial price tag of between $500 and $1,000, plus the ongoing need for sensors that cost $50 to $100 a pop, a CGM is not an inexpensive diabetes tool. The chemicals, precious metals, and enzymes that are built into the sensor help drive up the cost. Plus, building the sensors is a labor-intensive process.
Getting It Right
Outside of hospitals and laboratories, the blood glucose meter is the gold standard for accuracy. CGMs fall short because they measure glucose levels in the fluid that bathes the body’s cells, not in the blood. Glucose levels in this fluid are related to blood glucose, but there’s a difference and a lag in how quickly the CGM senses change, which can lead to problems with accuracy. That’s why CGMs need to be calibrated a couple of times a day with a measurement from a blood glucose meter. Without that, CGM readings would drift further and further away from blood glucose values. The computer program in the CGM attempts to correct for drift, but calibration remains necessary. “Our goal is to reduce or eliminate the need to calibrate,” says Pacelli.
Another accuracy challenge is that a CGM sometimes gives a faulty reading for biological reasons, such as a small blood clot forming around the sensor. Acetaminophen and other medications can also throw off CGM readings. “There is a lot of physiological interference going on in the body. It’s very complex,” says Pacelli. “We need to detect when it’s doing funky things and compensate for it.”
CGM programs are designed to detect glucose readings that don’t seem quite right, protecting users from false readings. “Technologies today look at noise in the sample,” says Greg Meehan, vice president and general manager of CGMs at Medtronic. “The algorithm has the ability to emphasize certain things and deemphasize noise [electrical fluctuations unrelated to glucose levels].”
Now and Later
The first-generation sensors are good for about three days. By improving the chemicals that coat the sensor, the latest versions can now outsmart the body’s biology for six or seven days. Pacelli says some people wear the sensors for up to two weeks, though that may cause skin irritation and other issues.
Engineers are working to make sensors last even longer and give more reliable results. “We need fewer outliers,” says Meehan. “You have that errant value where your blood glucose will say that you are at 100 [mg/dl] and your sensor says you are at 225. We want to minimize that going forward. We must build better chemistry.”
A better sensor would bring Medtronic closer to its ultimate goal: an artificial pancreas, a system that uses CGM readings to deliver insulin automatically through a pump. “Everything that happens with the [artificial pancreas] and the insulin delivery becomes dependent on the information that’s coming from that sensor,” says Meehan. To improve reading reliability, Medtronic is researching alternatives to glucose oxidase–based detection strategies as well as CGMs that use multiple sensors. Basing each glucose value on several measurements may help make the artificial pancreas safer, says Meehan, by ensuring that insulin doses are calculated from accurate readings.
At Dexcom, engineers are working to make the CGM monitor unnecessary by incorporating a Bluetooth-enabled transmitter into the system. This would allow glucose readings to go to a smartphone, or even a smart watch, Pacelli says, simplifying the life of the person with diabetes.
That’s ultimately the goal for CGMs—to make life with diabetes a little easier by taking some of the “What’s my glucose now?” worry away. With a CGM, you’ll know now, or in just a couple of minutes.
Two of Medtronic’s CGMs include an insulin pump. A single device handles both the CGM data and insulin pump controls. Such sensor-enhanced insulin pumps are considered by some as a stepping-stone to an artificial pancreas, says Greg Meehan, a CGM manager at Medtronic. One of the company’s systems, the MiniMed 530G With Enlite, shuts off the flow of insulin for up to two hours if CGM readings go below a certain threshold and the user doesn’t respond to an alarm.
Layers of Sensor Science