Percutaneous fiber-optic sensor for chronic glucose monitoring in vivo
Introduction
Clinical studies have concluded that fine-tuning of insulin administration on the basis of frequent glucose measurements provides substantial advantages in the management of diabetes (Diabetes Control and Complication Trials Research Group, 1993; UK Prospective Diabetes Study Group, 1998). Such treatment can dramatically reduce mortality and complications from the chronic metabolic fluctuation in either hyperglycemia or hypoglycemia. Ideally, insulin could be administered by “artificial pancreas” consisting of a chronically implantable sensor, a pump and an algorithm to adjust dosage continuously.
Presently, insulin delivery technology (insulin formulation and pump designs) is well developed but glucose sensors remain problematic (Steil et al., 2004). A suitable sensor should have a low cost to operate, which is related to the cost of the sensor itself, the cost to install it and the frequency at which it must be changed. It should be ergonomically unobstructive and easy to use to facilitate frequent measurements. It must have sufficient accuracy (as defined clinically by error grid analysis; Clarke, 2005), ideally without requiring intrusive calibration.
In the current clinical environment, choices for glucose monitoring are limited. Intermittent self-monitoring of blood glucose (SMBG) is the clinical method of choice. Typically, glucose measurements are done by pricking a finger (or other more insensitive region such as upper arm) and extracting a drop of blood, which is then applied to a test strip composed of chemicals sensitive to the glucose in the blood sample. An optical meter is used to analyze the blood sample and gives a numerical glucose reading. Most conventional SMBG devices have 95–99% accuracy in Clarke error grid analysis (Clarke, 2005). Because of the associated pain and inconvenience, however, few patients take more than a couple of readings per day, and so are at risk from unexpected and undetected peaks and valleys of glucose concentrations. The FDA has conditionally approved 4 short-term (<72 h) continuous monitoring sensors. All measure interstitial fluid glucose concentrations using the long-established enzymatic method involving glucose oxidase. This assay method consumes glucose enzymatically, creating a concentration gradient that may become steeper as the sensor is walled-off by the foreign body reaction. The glucose oxidase deteriorates over time and is sensitive to pH and temperature (Usmani and Akmal, 1994, Tamada et al., 2002, Wentholt et al., 2006). Limited or uncertain stability and accuracy (<80% in Clarke error grid analysis) prevent them from being used exclusively, so patients must still calibrate and cross-check using SMBG methods. The percutaneous probes are inherently complex and expensive to manufacture and include a multipin electrical connector that must be fixed mechanically to the skin surface near the point of entry. This fixation interferes with physical activity and hygiene and itself limits the long-term use of such percutaneous sensors.
We are developing a family of disposable, minimally invasive, in vivo sensors that can measure various analytes in a patient over a period of several weeks. The key element is a chronically implanted optical fiber (Fig. 1) with size and flexibility similar to a human hair, which would enable it to be both mechanically stable and unobtrusive when implanted in any convenient patch of hairy skin. Optical fibers are thin, lightweight, chemically stable, and generally biocompatible, all desirable properties for medical devices. Fiber communication technology is well established and has the advantages of high capacity, low attenuation, immunity to electromagnetic interference, and inherent electrical isolation that are attractive for this application.
The glucose sensor measures glucose concentration by the much-studied fluorescence resonance energy transfer (FRET) assay based on the selective binding of saccharides by the jack bean lectin Concanavalin A (Con A)(Meadows and Shultz, 1993). We have reengineered the chemistry, however, to overcome the challenges of chemical sensing in vivo:
In order to obtain sufficiently strong fluorescence from a very small sensor volume, it is necessary to incorporate fluorophores at high concentration, but that tends to increase the concentration of glucose required for competitive binding to affect FRET. Replacing linear dextran with betacyclodextrin (which has lower affinity for Con A) allowed us to improve the signal-detection sensitivity 60-fold, while preserving glucose-concentration sensitivity in the physiological range of 0–500 mg/dl (Liao et al., 2005).
In order to take full advantage of the simplicity of a single, percutaneous optical fiber, the fluorescence signals must be split and filtered from the much brighter excitation light. Quantum Dots (Qdot) can be excited at any wavelength shorter than their emissions, whereas organofluorophores such as fluorescein isothiocyanate (FITC) must be excited at a specific wavelength fairly close to their emission. Qdots also have a higher quantum yield, making them suitable for the FRET donor (but not the FRET acceptor). Preliminary results were presented by Liao et al. (2006).
Concentration of glucose is inferred from the ratio of the two fluorescent emissions at 525 nm and 570 nm rather than the absolute intensity of either. The preliminary results confirmed that it makes the assay relatively insensitive to variability of the photonic coupling or deterioration of the sensing materials over time. Because the assay is a reversible binding reaction, it does not consume glucose as does glucose oxidase. Encapsulation by a diffusion barrier of connective tissue might slow the dynamic responses of the sensor but would not be expected to produce a false concentration gradient around the detection site. The principles of the assay are illustrated in Fig. 2 and described in the “Supplementary Material” section.
The chronic biocompatibility and mechanical integrity of Sencil for in vivo application have been evaluated by animal studies. Histology of chronically implanted sites has verified the biocompatibility of the Sencil. Very small quantities of PEG (approximately 10 μg) and Con A (<10−12 g/Sencil) may be left behind if the matrix detaches when the spent Sencil is plucked out (Liao et al., 2005), but this should have negligible effects even cumulatively (the lethal dose of Con A in mice is 50 mg/kg by intravenous injection).
The polyimide coat of the optical fiber improves its mechanical toughness greatly, but it discourages cell and protein adhesion (Tan et al., 2001, Drumheller and Hubbell, 1994), which are needed for percutaneous fixation. The optical fibers separated easily from the PEG matrix and tended to slip out of the skin. To address both problems, we decided to bind a layer of collagen covalently to the polyimide coating because collagen is a well-known cell adhesion promoter (Darwin and Kivirikko, 1995, Jokinen et al., 2004) and it can be cross-linked to the PEG matrix (Liao et al., 2006).
In this study we demonstrate the in vitro performance of the prototype sensor, and summarize the improvement of chronic stability with collagen coated adhesion enhancement in vivo.
Section snippets
Reagents and materials
All chemicals involved were purchased from Sigma (St. Louis, MO). Tetramethylrhodamine isothiocyanate (TRITC) and 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide, hydrochloride (EDAC) was purchased from Invitrogen (Carlsbad, CA). α-acryloyl, ω-N-hydroxysuccinimidyl ester of poly (ethylene glycol)-propionic acid, MW 3400 (PEG-NHS 3400) was purchased from Shearwater Inc. (Huntsville, AL). The silica optical fiber was purchased from OceanOptics with 100 μm core, 110 μm cladding and 10–25 μm thick
Chronic biointerface integrity with adhesion enhancement
Before we conducted in vivo studies in rats, we measured the adhesion enhancement between the sensing matrix and the optical fiber in vitro. A peak-force meter was clamped to the free end of the Sencil. The sensing matrix bulb was pulled until it separated from the fiber (in vitro pull test). We measured a peak tension force of only 0–2 g for untreated 100 μm fibers. Separation force increased to 40 g after collagen treatment. For comparison, 70 g is the mean peak force for removing hair from the
Conclusions
The prototype glucose Sencil described here constitutes the first step in the development of a novel class of low-cost chemical sensors suitable for long-term management of ambulatory patients. Key elements of feasibility for glucose sensing can be considered in the context of underlying mechanisms and prior assay development.
The next step toward clinical implementation will be validation of the measurements in chronically implanted animals by comparison with a “gold standard” such as
Acknowledgements
This research was funded by the Alfred E. Mann Institute for Biomedical Engineering at the University of Southern California. We thank Zhen Wang for assistance with the rat experiments.
References (21)
- et al.
Anal. Biochem.
(1994) - et al.
J. Biol. Chem.
(2004) - et al.
Biomaterials
(1995) - et al.
Neurosci. Lett.
(2001) - et al.
Anal. Chim. Acta
(1993) - et al.
Adv. Drug Deliv. Rev.
(2004) - et al.
Biophys. J.
(2001) Diabetes Technol. Ther.
(2005)- et al.
Annu. Rev. Biochem.
(1995) Carbohydr. Res.
(1973)
Cited by (69)
Advancements in optical fiber sensors for in vivo applications – A review of sensors tested on living organisms
2024, Measurement: Journal of the International Measurement ConfederationFiber-optic semi-distributed Fabry-Perot interferometer for low-limit label-free detection of CCL5 cancer biomarker
2024, Optics and Laser TechnologyMultifunctional microcoaxial fibers for electrochemical, optical and spectroelectrochemical sensing
2023, Sensors and Actuators B: ChemicalRole of nanomedicine for targeted drug delivery in life-threatening diseases
2023, Nanobiotechnology for the Livestock Industry: Animal Health and NutritionMinimalistic design and rapid-fabrication single-mode fiber biosensors: Review and perspectives
2022, Optical Fiber TechnologyCitation Excerpt :From a technological perspective, modern biosensing devices can be divided into electrical and optical technologies; from Clark’s 1962 first electrochemical biosensor [10], the field of electrical biosensors has been successfully established, thanks to several solutions that include electrochemical impedance spectroscopy [11], transistors [12], lab-on-chip [13] among others. More recently, optical biosensors have been proposed, with the main goal of increasing performances and detection limits by exploiting the high sensitivity of optical methods [14]; in parallel to “labelled” biosensors, based on methods such as fluorescence [15] or spectrophotometry [16], label-free methods are the main research direction as they allow precise detection, at very low concentration, without the need for any activation or devices external to the sensor itself [17]. Surface plasmon resonance (SPR) has provided a boost to the performances of optical biosensors thanks to high sensitivity together with ease of fabrication [18–22].
- 1
Present address: 12F-1, No. 86, Sec. 1, Huamei W. Street, West District, Taichung City, Taiwan.
- 2
Present address: Department of Chemistry, University of Southern California, CA, USA.
- 3
Present address: School of Pharmacy, University of Southern California, CA, USA.
- 4
Present address: Department of Biomedical Engineering, University of California at Davis, CA, USA.
- 5
Present address: Baylor College of Medicine, Houston, TX, USA.