Optical and Fluorescence-Based Glucose Transduction

Optical and Fluorescence-Based Glucose Transduction

Optical glucose transduction turns a glucose level into a change in light instead of an electrical current. Most commercial CGM sensors work electrochemically [1]. The fluorescent exception is Senseonics' implanted Eversense, whose current model, Eversense 365, the FDA cleared in September 2024 for up to a year of wear [2]. Chemists who helped develop its sensing molecule wrote in 2024 that Senseonics was the only company to bring this type of sensor to market [3]. Needle-free optical monitoring is still research; as of October 2026, the FDA says it has not authorized any smartwatch or smart ring that measures glucose on its own [4].

Why measure glucose with light

"Transduction" is the step where a sensor turns what it detects into a signal its electronics can read. In most CGMs, an enzyme on an electrode (usually glucose oxidase) reacts with glucose and produces an electrical signal [1]. An optical sensor works differently. Glucose changes how a sensing material handles light, in Eversense's case how brightly it fluoresces, and the sensor's optical system measures that change [2].

The appeal for a long-term implant is chemistry that doesn't depend on an enzyme. Senseonics researchers call their glucose-indicating polymer "abiotic," meaning synthetic rather than biological [5].

How the Eversense sensor turns glucose into light

The Eversense 365 sensor is a small capsule, about 3.5 mm by 18.3 mm, that a healthcare provider inserts under the skin [2]. In a hydrogel on its surface sits a fluorescent indicator built on boronic acid, a chemical group that binds sugars reversibly [5][3]. When glucose binds, the indicator fluoresces more strongly. The sensor measures that change in brightness, and the transmitter worn on the skin over it converts the signal into a reading in mg/dL [2][6].

The implant draws its power by radio frequency from that transmitter. It takes a measurement every 5 minutes, and the transmitter relays glucose data to a phone app over Bluetooth Low Energy [2].

The oxidation problem

Here's what surprised us: the first sensors implanted in people lost their signal during the first day, because the boronic acid had been oxidized into phenols that no longer respond to glucose [3]. The chemists attribute that to the body's inflammatory response to the implant [3].

Developers tried two defenses. A platinum layer about 3 nm thick, sputtered over the gel, shielded the boronic acids from reactive oxygen species and extended sensor stability in the body from hours to at least 6 months [3]. The second was a silicone collar holding 1.75 mg of dexamethasone acetate, an anti-inflammatory steroid; the FDA's review estimated that about 0.18 to 0.35 mg is released over 90 days [6].

A third fix came later. In the PROMISE study, 43 participants wore a second sensor whose gel added a "sacrificial" boronic acid meant to absorb oxidative attack and spare the glucose-binding sites [7]. Ninety percent of those modified sensors lasted 180 days, against 65% of the standard sensors [7]. Their MARD (mean absolute relative difference, the average gap between sensor readings and lab reference values) was 8.5%, compared with 9.1% [7].

From 90 days to a year

The FDA approved the first Eversense on June 21, 2018, for adults, for up to 90 days, as an adjunct to fingerstick meters [6]. In its pivotal PRECISE II study of 90 adults, MARD was 8.5%, and 94.3% of readings fell within 20 mg/dL or 20% of lab values [6]. Those figures came from reprocessing the raw sensor data with an updated algorithm after the study ended [6].

The Eversense E3, approved on February 10, 2022, changed the sensor design to allow up to 180 days of use [8]. Eversense 365 has since superseded it. Its label covers adults 18 and older for up to a year, lets its readings replace fingersticks for treatment decisions, and allows connection to automated insulin delivery systems [2].

The FDA's review of Eversense 365 rests on the ENHANCE study of 110 adults, in which 92.0% of readings were within 20% of the lab reference [2]. The published results report an overall MARD of 8.8% with mostly one calibration per week, and 90% of sensors lasted the full 365 days [9]. Accuracy held steady through the year, with MARD in each 30-day window between 7.8% and 9.9% [2]. It was weaker at the low end: readings below 54 mg/dL had a MARD of 17.9% [2]. Our comparative accuracy and MARD analysis puts these numbers next to other sensors.

What fluorescence changes for the wearer

Eversense 365 still asks for fingerstick calibrations. After a 24-hour warm-up, it needs four within 36 hours, then one a day through day 13, then one a week from day 14 [2].

Optical sensors also have their own interference profile. A substance can raise the signal by binding to the hydrogel or by emitting light in the sensor's spectral range, or lower it by absorbing the excitation light or quenching the fluorescence [7]. For Eversense 365, the labeling says tetracycline-class antibiotics can falsely lower readings. Mannitol or sorbitol given intravenously, or in irrigation or peritoneal dialysis solutions, can falsely raise readings [2]. Whether any of these medicines matters for you is a question for your care team.

The system is also contraindicated for people who can't take dexamethasone, and the transmitter must be removed before an MRI [2]. Our article on implantable fluorescence sensors covers insertion and removal.

Optical sensing without a needle

On February 21, 2024, the FDA warned people not to use smartwatches or smart rings that claim to measure blood glucose without piercing the skin. These devices don't directly test blood glucose, the agency said, and it has not authorized any that measure or estimate glucose on their own [4].

The most extensive needle-free optical data we found use Raman spectroscopy. A laser shines into the skin, and a spectrometer reads the faint light that skin molecules scatter back [10]. A 2018 study aimed a table-top Raman instrument at interstitial fluid about 250 µm below the skin at the base of the thumb [11]. Thirty-five people with diabetes used it at home for 60 days, and its calibration stayed stable for more than 10 days [11]. A 2023 follow-up tested a portable device in 160 people at home for at least 15 days after calibration, and in a subset with type 2 diabetes, MARD was 14.3% [10].

Two caveats apply. The developer, RSP Systems, funded the 2018 study, and its employees wrote that paper and led the 2023 one [11][10]. The 2023 paper also lists the field's long-standing problems as accuracy, calibration stability and robustness, noting that most earlier attempts lasted only a few hours under controlled conditions [10]. Our overview of non-invasive optical glucose sensing technologies covers other light-based approaches.

What is still unknown

The 2018 Raman paper put the gap plainly: many non-invasive technologies had been tested, and none had entered general clinical use [11]. The studies we read don't show a needle-free optical device matching the 8.8% MARD reported for the implanted sensor [9][10]. Until one does, the FDA-reviewed accuracy data we found for optical glucose sensing all come from fluorescence under the skin [6][2].