Inflammation and Glucose Sensors: Use of Dexamethasone to Extend Glucose Sensor Function and Life Span in Vivo

Abstract

Background

It has been generally accepted that the acute loss of sensor function is the consequence of sensor biofouling as a result of inflammation induced at sites of sensor implantation, as well as tissue trauma induced by the sensor and its implantation. Because anti-inflammatory therapies are used routinely to control inflammation in a wide variety of diseases, we hypothesized that anti-inflammatory therapy would likely extend glucose sensor function in vivo. To test this hypothesis, we utilized our recently developed mouse model of implantable glucose sensors and the potent anti-inflammatory steroid dexamethasone (DEX).

Method

For this study, glucose sensors were implanted subcutaneously into the head and neck area of mice and sensor function was determined up to 14 days postimplantation. These mice received a daily intraperitoneal injection of DEX at a dose of 1, 6, or 10 mg/kg body weight.

Results

Mice not treated with DEX lost sensor functionality very rapidly, usually within the first 24 hours postimplantation. Mice treated with DEX at the various doses had an increased sensor life span of up to 2 weeks postimplantation. Additionally, sensitivity was maintained in DEX-treated mice as compared to control mice (non-DEX treated). Histologic evaluation of tissue surrounding the site of sensor implantation had almost no inflammatory cells in DEX-treated mice, whereas control mice had an intense band of inflammation surrounding the site of sensor implantation.

Conclusion

To our knowledge this is the first study directly demonstrating that anti-inflammatory therapy can extend glucose sensor function in vivo and supports the key role of inflammation in loss of sensor function in vivo, as well as the uses of anti-inflammatory therapy as a potential key adjuvant in enhancing glucose sensor function and life span in vivo.

Introduction

Diabetes is a chronic disease that has reached epidemic proportions. The key to successful management of diabetes and its associated complications is continuous monitoring of blood glucose. Although there are short duration transdermal glucose sensors on the market, which allow continuous blood glucose monitoring for short periods of time, their functionality is only guaranteed for 3 days postimplantation (DPI).^1–3 Currently, both health insurers and governmental agencies have not been willing to pay the cost of these short duration implantable glucose sensors for diabetes management. Clearly, developing sensors with extended life spans would be more appealing to both the patient and the health care community. Currently, the in vivo loss of sensor function seen in implantable glucose sensors is thought to be in large part the result of the tissue response triad (TRT) of inflammation, fibrosis, and vessel regression, which occurs in the tissue surrounding implanted glucose sensors. This TRT is the result of tissue trauma arising from the insertion of the sensor into the skin, as well as the tissue response to the sensor as a “foreign object.”

Although TRT seen at sites of sensor implantation is histopathologically similar to other forms of tissue inflammation (the “itises”) and could be termed “sensor-itis,” there has been little effort to utilize the vast array of existing effective anti-inflammatory agents (e.g., glucocorticoids and nonsteroidal anti-inflammatory drugs) to suppress sensor-induced TRT directly, thereby extending sensor function and life span in vivo. While there have been efforts to evaluate various drug delivery systems, little is known about the effects of these anti-inflammatory drugs on sensor function in vivo.^4–7 These observations have led us to hypothesize that anti-inflammatory therapy would decrease inflammation at sites of glucose sensor implantation, thereby extending glucose sensor function and life span. Because glucocorticoids, such as cortisone, prednisolone, and dexamethasone (DEX), are powerful anti-inflammatory agents, we focused on this class of anti-inflammatory agents for our initial studies to test this hypothesis. As an in vivo model, we utilized our recently developed murine model of implantable glucose sensors.^8–11 Studies demonstrated that (1) DEX had no effect on glucose sensor function in vitro; (2) in DEX-treated mice the tissue surrounding the site of sensor implantation had markedly reduced, if any, inflammation as compared to control (nontreated) mice; (3) sensor function inversely correlated with inflammation at the site of sensor implantation (i.e., increased inflammation associated with decreased sensor function); and (4) both sensor function and life span were increased dramatically in the DEX model as compared to control.

Material and Methods

Glucose Sensor Fabrication and Function in Vitro

Fabrication and in vitro testing of our glucose oxidase-based, Nafion-coated amperometric glucose sensors (needle type) were performed utilizing established protocols.^10,12

Glucose Sensor Implantation in Mouse

For the murine model of implantable glucose sensors we utilized 35- to 40-gram ICR mice from Harlan (Indianapolis, IN). Sensors were implanted in the shaved head and neck area of the mouse utilizing our saline bleb technique as described previously.^9,10 All mice were maintained under specific pathogen-free conditions at the animal facility of the University of Connecticut, Farmington, Connecticut, according to animal care procedures.

Glucose Sensor Function in Vivo

At predetermined time points, postsensor implantation, sensor function, and blood glucose levels were assessed as described previously.¹⁰ Glucose sensor function was tested in each mouse immediately following implantation, designated as 1-hour postimplantation (HPI), 5 HPI, and 1, 2, 3, 7, and 14 DPI. Occasionally, sensors were also tested at 10 DPI. Between sensor function tests, mice were left unrestrained in their cages, without polarization of the sensor. The two-point calibration method was used to calculate sensor sensitivity, expressed in nA/mM, as the ratio between the change in sensor response (Δ nA) and the change in blood glucose concentration (Δ mM) following an IP glucose injection.¹³

Effect of Dexamethasone on Glucose Sensor Function in Vitro

In order to determine if DEX had any negative effect on sensor functionality in vitro, sensors were tested with and without the addition of DEX. Briefly, sensors were tested in phosphate-buffered solution (PBS) with the stepwise increase of glucose (0–30 mM glucose) as described previously.¹² This step was necessary in order to obtain an initial sensor sensitivity value. Next, sensors were placed in PBS with the addition of stepwise DEX (0 to 15 mg/ml) (water-soluble dexamethasone; Sigma, St. Louis, MO). In order to determine if DEX had any negative effect on sensor sensitivity and functionality after DEX treatment, sensors were retested in PBS with stepwise addition of glucose. Because in vivo experiments were designed in a way so that there would be a constant level of DEX present at all times in the animal, we also determined in vitrothe effect of sensor function with a constant level of DEX present. For that, we chose a level of 15 mg/ml of DEX present in the PBS solution. After washing the sensor in PBS and placing it back into PBS solution, glucose was then again added stepwise as described previously.¹⁰ In order to avoid dilution of DEX, all glucose solution used in this step also contained 15 mg/ml DEX. Finally, all sensors were retested in PBS with stepwise addition of glucose in order to confirm initial sensor sensitivity and functionality. Data obtained for this study are expressed as mean ± standard error of the mean (SEM).

Effect of Dexamethasone Treatment on Glucose Sensor Function in Vivo

Mice designated for DEX treatment received 200 µl of DEX [1, 6, or 10 mg/kg body weight (BW)] injected IP the day before sensor implantation (see Figure 1 for protocol). On the day of sensor insertion, 100 µl of DEX (1, 6, or 10 mg/kg BW concentrations) was injected subcutaneously into the head/neck area and the sensor was inserted into the area where DEX was injected. Postsensor implantation, all DEX-treated mice received a daily injection of 200 µl DEX IP over the duration of the study at the various DEX concentrations outlined earlier. A total of 4 mice treated with 1 mg/kg BW, 5 mice treated with 6 mg/kg BW, and 9 mice treated with 10 mg/kg BW were utilized for sensor functionality tests. Occasionally, some time points had to be excluded because of dislodging of the sensor by the mouse. This resulted in the various numbers of mice done for each study. For control studies, 20 mice were utilized for sensor functional testing and histology. As the experiment progressed, this number decreased as a consequence of sacrificing mice at various time points for histological evaluation and dislodgement of sensor by mice. Five mice were utilized for the last sensor functional test point (e.g., 14 DPI). The two-point calibration method was used to calculate sensor sensitivity, expressed in nA/mM, as the ratio between the change in sensor response (in nA) and the change in blood glucose concentration (in mM) following an IP glucose injection.¹³ Data obtained for these studies are expressed as mean ± SEM values.

Figure 1.

Protocol for evaluation of the effect of dexamethasone (DEX) on glucose sensor function in vivo. For these studies, DEX or saline only (non-DEX control) was injected IP 24 hours prior to sensor implantation and daily after sensor implantation.

Gross and Histological Evaluation of Sites of Sensor Implantation

Sites of sensor implantation were evaluated grossly for redness, swelling, warmth, or other signs of inflammation. In general, at least three to five animals were used per time point and drug concentration for the histological tissue reaction evaluation. Animals from the functional sensor study were included in the histological evaluation when this study was terminated, resulting in a higher n value for the 14-day time point. To evaluate the tissue responses to implantation of the glucose sensor at various time points, individual mice were euthanized, and tissue containing the implanted sensors was removed and fixed in 10% buffered formalin. Next, the buffered formalin-fixed tissue was then paraffin embedded. Prior to sectioning the embedded tissue, the sensor was removed from the paraffin-embedded tissue. Removing the sensor while it was embedded in the paraffin resulted in minimum tissue damage as a result of sensor removal. After removal of the sensor, the tissue was reembedded and processed for sectioning.¹² The resulting tissue sections were processed using hematoxylin and eosin (H&E), as well as trichrome-staining techniques. Histopathologic evaluation of tissue reactions at sites of sensor implantation was done on mouse specimens obtained at 1, 3, 7, and 14 days postimplantation of the glucose sensor. Tissues from the sites of sensor implantation were examined for evidence of loss of cell and tissue architecture, acute and chronic inflammation, including giant cell formation, and necrosis, as well as fibrosis and vessel regression.

Results

Effect of Dexamethasone on Glucose Sensor Function in Vitro

To begin these studies we first determined whether DEX had any effect on glucose sensor function in vitrousing the protocol as outlined in the Materials and Methods section. Initially we tested sensors using injectable DEX, which is already dissolved in a carrier vehicle/solution, but found that the vehicle (DEX placebo) induced a sensor response in the absence of glucose. It is possible that the vehicle contained glucose, but this information was proprietary and not available to us. Although there is the possibility that injectable DEX would not affect the sensors in vivo, we chose not to use this formulation of DEX because of this issue. For all our studies, water-soluble DEX from Sigma prepared in sterile PBS was utilized. Our in vitrostudies demonstrated that water-soluble DEX had no effect on sensor function in vitro(Figure 2).

Figure 2.

Evaluation of the effect of dexamethasone on glucose sensor function in vitro. These sequential studies demonstrate that DEX does not have any adverse effects on sensor response in vitro. Data obtained for this study are expressed as mean ± SEM.

Effect of Dexamethasone Treatment on Glucose Sensor Function in Vivo

Once we determined that DEX did not interfere with glucose sensor output, we investigated glucose sensor functionality in mice treated with DEX. For that, glucose sensors were implanted subcutaneously into the head and neck area of mice with and without DEX treatment (for protocol, see Figure 1). Mice received a daily intraperitoneal injection of DEX at a dose of 1, 6, or 10 mg/kg BW. There were no deaths as a result of DEX treatment. To evaluate sensor function in our mouse model, we analyzed sensor function in the mouse at 1 HPI and up to 14 DPI as described previously by our laboratory.¹⁰ In these studies, glucose blood levels paralleled sensor function at both 1 and 7 DPI for DEX-treated mice but not for control mice (representative studies presented in Figure 3). Control mice (i.e., no DEX) lost sensor functionality rapidly, usually within the first 24 hours postimplantation (Figure 3). Alternatively, mice treated systemically with DEX had increased sensor sensitivity with a sensor life span of at least 1 week postimplantation when compared to control mice (Figure 3). Additionally, sensitivity was maintained in DEX-treated mice as compared to control (Figure 3). To evaluate tissue reactions at sites of sensor implantation, mouse tissue was obtained from tissue sites adjacent to the sensors implanted for 1, 3, and 7 DPI. The resulting tissue was processed for histopathologic evaluation as described by our laboratory.¹⁰ Standard H&E was used for this study. Histological evaluation of tissue reactions at the site of sensor implantation indicated that DEX-treated mice had almost no inflammatory cells, whereas control (no DEX) mice had an intense band of inflammation surrounding the site of sensor implantation (Figure 3). The original sensor location is designated by the letter S and residual sensor coatings are designated by a star in Figure 3. A summary of the results of all studies is presented in Figure 4, expressed as sensor sensitivity (nA/mM glucose) versus time postsensor implantation for mice ± DEX treatment (i.e., Figure 4A, mice treated with 1 mg/kg BW DEX; Figure 4B, mice treated with 6 mg/kg BW DEX; and Figure 4C, mice treated with 10 mg/kg BW DEX). As expected, control mice (no DEX) displayed a rapid fall in sensor function within the first 24 hours postsensor implantation (Figure 4). Alternatively, mice treated with DEX at all three concentrations displayed a marked increase in sensor life span as well as sustained sensitivity (Figure 4).

Figure 3.

Correlation of glucose sensor response versus sensor-induced tissue reactions in dexamethasone- and nondexamethasone (control)-treated mice. For these studies, mice were injected with DEX (6 mg/kg BW) using the protocol presented in Figure 1. Control mice received no DEX injections. At 1, 3, and 7 days postsensor implantation, glucose sensor function was determined and the tissue surrounding the sensor was removed and processed for histopathology as described previously.¹⁰ These studies clearly demonstrate that DEX treatment of mice increased sensor function/life span dramatically and decreased inflammation at the site of sensor implantation. The original sensor location is designated by a star and residual sensor coatings are designated by an S.

Figure 4.

Effect of various concentrations of dexamethasone on sensor function in vivo. For these studies, mice received various concentrations of DEX [1 (A), 6 (B), or 10 (C) mg/kg BW] using the protocol outlined in Figure 1. Sensor function was evaluated in DEX- and non-DEX-treated mice for up to 14 days postsensor implantation. Using our mouse model, these results clearly demonstrate that DEX extended glucose sensor function in vivo dramatically. Data obtained for this study are expressed as mean ± SEM.

Particularly in the early time points (e.g., 5 HPI in Figures 4A and 4C for DEX treatment) the sensor response was increased slightly when compared to data obtained from one HPI DEX-treated animal. Possible explanations include variations from sensor to sensor as well as the degree of implantation trauma. Additionally, it should be noted that sensors used for these studies utilized a Nafion coating as the outer sensor membrane coating. Previous studies by Mercado and colleagues¹⁴ demonstrated that Nafion coating is mineralized with calcium phosphate in vivo. Mineralization of the outer membrane coating Nafion accelerates the susceptible cracking of the Nafion sensor membrane, ultimately resulting in a decrease of sensor functionality depending on membrane damage or total sensor loss. Because DEX does not have any effect on preventing mineralization of the outer membrane, we believe that this mineralization is the result of sensor functionality loss caused by 14-day postsensor implantation even in the presence of DEX. Occasionally, mice dislodged their sensor prior to termination of the experiment, thus resulting in the various n values for the different DEX dosages utilized for the animals. These data clearly demonstrate the “proof of principle” that anti-inflammatory drugs, such as dexamethasone, can extend glucose sensor function and life span in vivodramatically.

Discussion

Implantable materials and devices, such as drug delivery systems, pacemakers, artificial joints, and organs, play an important role in health care today. In addition to these devices, implantable monitoring devices such as sensors have great potential for improving both the quality of care and the quality of life of patients and animals. Potentially these sensors can measure a wide variety of analytes in the blood and tissue, which would be critical in the early diagnosis and treatment of diseases. Unfortunately, the development of these implantable sensors has been hampered by the inability of currently designed implantable sensors to overcome their rapid loss of function in vivo. This loss of sensor function is thought to be the result of the acute and chronic tissue reactions to the implanted sensors. Specifically, these tissue reactions are a result of various factors, including (1) tissue injury and inflammation as a result of the surgical implantation of the device, (2) inflammation at the implantation site as a result of “foreign body” reactions to the device, (3) tissue toxicity of the inflammatory cells recruited to sites of sensor implantation, i.e., the “bystander effect,” and (4) the release of tissue toxic factors from the functioning of the device and/or the chemical breakdown of the device and its coating. For example, implantable glucose sensors are usually based on the usage of glucose oxidase, an enzyme specific for glucose. The enzyme continuously breaks down glucose into gluconic acid and hydrogen peroxide, both of which are tissue toxic, as well as potentially “sensor toxic.” The hydrogen peroxide is broken down further in reactive oxygen radicals, which are also toxic. In the case of a needle-type hydrogen peroxide-detecting glucose sensor, glucose oxidase needs to be immobilized on the platinum wire by using a carrier protein such as albumin and toxic cross-linking agents such as glutaraldehyde. Ultimately, these chronic inflammatory reactions at sites of sensor implantation result in tissue destruction and fibrosis with complete loss of sensor function in vivo. We refer to this tissue response to implantation of devices as the tissue response triad of inflammation, fibrosis, and vessel regression.

Tissue Response Triad and Implantable Sensors

The tissue response triad of inflammation, fibrosis, and vessel regression occurs in the tissue surrounding implanted glucose sensors. In addition, insertion of a biosensor into subcutaneous tissue is a triggering event for the inflammatory cascade. Key to this inflammatory cascade is the movement of fluids (edema) and cells, i.e., recruitment of inflammatory cells, including polymorphonuclear leukocytes, monocytes/macrophages, and lymphocytes, into the sites of sensor implantation. The increased vasopermeability induced by vasoactive mediators triggered by the implantation of the sensor results in edema formation in the skin, with the potential of plasma proteins biofouling the sensor, as well as triggering the release of leukocyte chemotactic factors, which recruit inflammatory cells from the vasculature into the tissue site. These recruited inflammatory cells not only contribute to biofouling of the implanted sensor directly, but also induce massive tissue injury because of toxic oxygen radicals and hydrolases released by these inflammatory cells. Furthermore, these inflammatory cells are also directly toxic to the tissue cells located near the implanted sensor. Ultimately the inability of these inflammatory cells to “clear” or remove the sensor triggers the recruitment of fibroblasts to the site of sensor implantation. Fibroblasts once activated at the site of sensor implantation produce a dense scar (collagen), which induces vessel regression away from the sensor (decreasing sensor access to blood glucose and oxygen), as well as slowing glucose diffusion to the sensor. Thus, minimizing TRT at sites of sensor implantation is key to extending glucose sensor function in vivo.

Efforts to Overcome Tissue Response Triad

Previously, efforts to overcome various aspects of TRT, thereby extending the in vivolife span of implantable glucose sensors, have focused on the usage of various sensor polymer coatings in an effort to “hide or stealth” the sensor from detection and the related tissue reactions. Unfortunately, these approaches have not been successful, and the use of various coatings has seen limited success as a result of the body's innate and acquired host defense systems (immunity) that can detect minute differences between normal tissue elements and foreign materials such as sensor coatings. Alternative approaches to extending sensor life span in vivohave been proposed, including incorporation of bioactive drugs and peptides and proteins into the sensor coatings or usage of sensor-associated drug delivery systems. In the case of these types of coatings it has been found that (1) generally only “analyte-permeable coatings” can be used as sensor coatings, thus limiting the type of coating available for implantable sensors; (2) binding of sufficient quantities of bioactive agents, such as peptides and proteins, can be difficult and often do not remain active after being bound to the sensor coating; (3) the intense tissue reactions (proteins and cells) frequently “mask” or degrade the bioactive agents on the coatings and limit their effectiveness; and (4) because of the limited quantities of bioactive agents that can be incorporated into sensor coatings, the coating and therefore the sensor itself have a limited life span in vivo and must be replaced frequently. In our experience, traditional drug delivery systems such as microbeads frequently do not incorporate (load) and/or release bioactive agents in quantities and for durations that are useful for implantable sensors. Because of these limitations, current drug delivery systems frequently require injections of large quantities of the drug delivery system in order to have any sustained anti-inflammatory effects. Additionally, drug delivery systems releasing bioactive agents often have negative “bystander” effects on the sensor and its function because of the release of toxic breakdown products from the drug delivery system itself. For example, in the case of biodegradable systems, breakdown of the drug delivery system itself results in the release of tissue toxic and sensor toxic by-products that hinder sensor function in vivo. Ultimately, the combination of inflammation, fibrosis, and loss of blood vessels also decreases tissue levels of glucose and oxygen as a consequence of glucose and oxygen consumption by inflammatory cells in close proximity of the sensor implantation site. Because glucose and oxygen are both essential to glucose sensor function in vivo, the consumption of these molecules by inflammatory cells results in a loss of sensor function and life span in vivo. Clearly, new approaches will be needed in the future to extend the function and life span of implantable glucose sensors to aid in the successful management of diabetes.

The combination of mechanical tissue injury from sensor insertion and the presence of a foreign irritant (the sensor) causes an acute inflammatory tissue response. The acute response is then followed by a chronic inflammatory response, ultimately resulting in fibrosis and blood vessel regression. Therefore, approaches that suppress or block these tissue responses to the sensor would likely extend sensor function and life span in vivo. Therefore, we hypothesize that inhibition of this inflammatory cascade using powerful anti-inflammatory agents such as glucocorticoids, e.g., DEX, will reduce inflammation and thereby sensor biofouling and extend sensor function and life span in vivo.

Glucocorticoids and Inflammation

The discovery of adrenocorticoids, a type of steroid, as biologically active agents in the adrenal glands in the late 1920s helped scientists realize the importance of these compounds as therapeutic agents. The realization of the anti-inflammatory activity of these agents sparked the production of steroid drugs, which has had a profound impact on disease treatment to the present day. Early on, these drugs were found to ease pain and inflammation related to rheumatoid arthritis, which led to its use as anti-inflammatory, antiallergic, and immunosuppressive agents, eventually expanding its use in many other inflammatory conditions. Today they are used in diseases related to allergy, dermatology, and respiratory. Soon after isolation of these agents, synthetic analogs of these compounds were created and used to treat a variety of ailments. These analogs include the glucocorticoids cortisone, prednisolone, and DEX produced in the 1950s and all are still used today. The efficacy and duration of these compounds vary by the position of certain chemical groups and of these, DEX is the most effective in both these categories.

Dexamethasone and Glucose Sensor Function in Vivo

Although inflammation is generally accepted as a key player in the loss of glucose sensor function in vivo, it is surprising that there have been only a few studies to directly investigate the role of inflammation in sensor function in vivo. In fact, the only prior study that we could find in the literature was a study by Ward and colleagues,¹⁵ which used sensors with drug delivery systems to deliver DEX locally with the sensor. These studies failed to demonstrate that DEX significantly extends sensor function in vivo. This may be in part because of the uses of a very large sensor (8 × 4.5 × 2.5 cm), which required a 7-cm incision to implant. The size and implantation trauma from such a large implantable sensor likely caused massive tissue injury and inflammation, which DEX could not overcome. Additionally, unlike our studies we predosed the animal and tissue site with DEX to minimize any inflammation as a result of implantation trauma.

For our present studies on the in vivo effect of DEX on glucose sensor function in vivo we focused on suppressing the combination of mechanical tissue injury from sensor insertion and the presence of a foreign irritant (the sensor), both responsible for causing an acute inflammation followed by a chronic inflammatory response. To achieve this goal we not only predosed the mice systemically with DEX 24 hours prior to sensor implantation, but also used a DEX injection, i.e., “dexamethasone bleb,” at the site of sensor implantation just prior to sensor implantation. We believe that this approach “primed” the implantation site to minimize inflammation at that site as a result of implantation trauma to the tissue. We believe this preimplantation protocol decreased initial tissue injury as a result of the implantation-associated inflammation. We further believe that the sustained DEX dosing, which suppressed inflammation at the implantation site, was required to extend sensor function in the mice. We did not utilize a large animal pool for this study, as our goal was to demonstrate the “proof of principle” that the anti-inflammatory drug DEX could suppress inflammation at sites of sensor implantation and enhance sensor function and life span. The results of these studies are so dramatic and clear-cut over all the concentrations and time points that usage of a larger animal pool would not change the outcome. For these studies we utilized an in vivo range of 1–10 mg/kg BW DEX, which has been used in nonsensor studies to suppress inflammation.¹⁶ For future studies we plan to develop other protocols to use lower levels of DEX as well as other anti-inflammatory agents to extend sensor life span in vivo.

The ability of DEX treatment to significantly extend glucose sensor function in our mouse model is exciting and supports the concept that suppressing inflammation at sites of sensor implantation will extend sensor function and life span. These data also suggest that short-term systemic or topical DEX will help suppress the initial inflammation resulting from sensor implantation, as well as the tissue reaction to the foreign nature of the sensor itself. We do not believethat long-term uses of systemic DEX are practical in patients with diabetes because of both systemic and local side effects induced by these steroids, but we do believe these studies clearly demonstrate the critical role of inflammation in the loss of sensor function seen in vivo.

Abbreviations

BW

body weight

DEX

dexamethasone

DPI

days postimplantation

H&E

hematoxylin and eosin

HPI

hours postimplantation

IP

intraperitoneal

PBS

phosphate-buffered solution

SEM

standard error of the mean

TRT

tissue response triad