Herndon, Virginia, May 9-10, 2014; Full Commentary – Draft

Executive Highlights

Over one hundred doctors, scientists, and engineers from around the world recently met outside of the Dulles Airport in Washington, DC at the first annual Tissue Response to Implanted Active Medical Devices conference. The Diabetes Technology Society sponsored the conference and once again brought together a very diverse group of speakers and attendees. The meeting included some promising work in diabetes technology, but also included extensive discussion of tissue response to implants for other medical applications (e.g., cardiac pacemakers, neural stimulators, and orthopedic implants). While most of the public attention to CGM is focused on existing technology (percutaneous sensors from Dexcom, Medtronic, and Abbott), this conference made it abundantly clear that there are other approaches being pursued in the background that may turn out to be clinically and commercially viable in the future.

Notably, there have been significant advances in the field of biomaterials that are likely to impact the design, size, form factor, and usability of future diabetes technology. Long-term implanted glucose sensors were explored almost twenty years ago, most notably by MiniMed and Dexcom, and abandoned because of the size of the devices, poor reliability, and long development cycle times. There are now numerous companies developing highly miniaturized long-term implanted glucose sensors: GlySens, Senseonics (notably absent from the podium at this meeting, though presenting data at ADA 2014 in poster #837-P), Biorasis, and Profusa, to name just a few. In addition, there are academic groups like Caltech (partially funded by Sanofi; see below) and the University of Connecticut that are also active in the field. While these devices are unlikely to threaten the market share of current percutaneous (through the skin) continuous glucose sensors any time soon, it is possible that long-term implanted sensors could be the next “wave” of CGM devices.

Below, we enclose our top 10 highlights from the meeting, followed by detailed coverage of each of these presentations.

Top 10 Highlights

1. Mr. Peter Simpson (Dexcom, San Diego, CA) gave an excellent overview of more than 10 years of Dexcom research and development aimed at addressing the foreign body response and commercializing successive generations of CGM devices with improved accuracy and reliability. The distribution of individual sensor MARD values for the G4 Platinum ranges from 2-5% in vitro, but expands to 5-20% in vivo – the difference is due to biological factors. Dexcom is currently working on future CGM systems that are factory calibrated and plans to have MARD values of 8-10% comparable to fingersticks. R&D efforts are now focused on the effect of biological factors on sensor performance.

2. Dr. Kristin Helton gave the first public presentation on research at her company, Profusa (a startup in South San Francisco). We’ve been waiting a long time on this one. She reported on Profusa’s work to develop a tissue-integrating scaffold that can be used as a platform for fluorescent sensing of a number of different analytes, including glucose. The Profusa team is working with Dr. Michael McShane at Texas A&M to develop a long-term, fluorescent glucose sensor. Both Dr. McShane and others on the Profusa team have worked for many years on fluorescent glucose sensors and are perceived by many as experts in the field. Dr. Helton hopes to present more glucose data by this time next year. Clock is on!

3. Dr. David Gough (University of California San Diego and GlySens, San Diego, CA) challenged the widely held belief that long-term implanted sensors will always be encapsulated by an impermeable membrane. He showed data from GlySens studies of a long-term implanted glucose sensor that demonstrated good sensor performance out to 360 days. The team has completed a small six-month study in humans (n=6) – sensor readings show excellent agreement with YSI readings. In Q&A, Dr. Gough said the team has not yet done accuracy analysis of the data set and is not able to provide data on the mean absolute relative difference (MARD) or Clark error grid. Dr. Gough is a pioneer in this field and it was terrific to see and hear his input.

4. Dr. Axel Scherer (California Institute of Technology, Pasadena, CA) reported on the work of his team at Caltech to develop a fully implanted miniature glucose sensor based on a CMOS design. Notably, Dr. Scherer’s project is funded in part by Sanofi. The glucose sensor is a thin 1.4 mm x 1.4 mm square (the size of Franklin Roosevelt’s eye on a US dime!) and is based on the glucose oxidase enzyme. The key to the miniaturization of the Caltech device being developed by Dr. Scherer’s group is, as we understand it, an inductive power system that wirelessly provides power to the sensor and communicates the result of the glucose sensor reading to an external display device. Dr. Scherer’s team is beginning to explore the operation of their miniaturized devices in animal models.

5. Dr. Dianne Burgess’ (University of Connecticut, Storrs, CT) lab has focused on the use of dexamethasone, a broad and potent anti-inflammatory agent. Her team at the University of Connecticut is developing a fully implanted miniaturized glucose sensor (0.5 mm x 5 mm). The sensor coating consists of microspheres that release dexamethasone, which inhibit the foreign body response. The sensor is inserted into the volar region of the wrist, and a custom wrist-worn watch receives the telemetry from the implanted sensor and can broadcast the sensor signal to a phone or elsewhere. The sensors can also be removed using a needle introduced under the skin. Dr. Burgess and colleagues have licensed their intellectual property to a start-up company, Biorasis Inc, and are attempting to commercialize the technology.

6. Dr. Buddy Ratner (University of Washington, Seattle, WA) gave the conference’s keynote address and presented new data on the use of micro-porous scaffolds to facilitate tissue ingrowth into implanted devices. The intellectual property from the University of Washington has been licensed to a startup called Heliosonics. Two other startups are using a similar approach: South San Francisco-based Profusa for long-term oxygen and glucose sensors (see above), and iStar for an end ophthalmic implant for the treatment of glaucoma.

7. Dr. Jeffrey Joseph (Thomas Jefferson University, Philadelphia, PA) described his experience evaluating intravascular glucose sensors for use in hospitalized patients, which has included work with GluMetrics, Edwards, Animas (in the 1990s), and others. It was great to hear another excellent overview of the latest in diabetes technology – especially in the hospital where he is such an expert (we have seen him present in this field since 2003; he has contributed a great deal of valuable thinking according to what we hear from the diabeterati).

8. Dr. Andres Garcia (Georgia Institute of Technology, Atlanta, GA) reported on the development of new fluorescent probes to image the progression of inflammation in response to biomedical device implants. These methods are shedding new light on the mechanism of inflammation on encapsulation of implants. Dr. Garcia also shared details on his work to modify receptors on cells that control the adhesion of proteins onto the surfaces of implanted materials.

9. Dr. Shaoyi Jiang (University of Washington, Seattle, WA) shared a review of differences between standard biocompatible polymers, such as polyethylene glycol (PEG), and electrostatic or zwitterionic (ZI) materials. Dr. Jiang characterized the zwitterionic materials as superhydrophilic compared with other materials. Superhydrophilic compounds are highly water soluble and are believed to be very well-tolerated by the body. Ultra-low fouling can be obtained with zwitterionic materials.

10. Dr. Monty Reichert (Duke University, Chapel Hill, NC) reviewed the biocompatibility problem facing continuous glucose monitoring sensors. His talk focused on the work of his laboratory modeling the performance of the Medtronic Sof-Sensor. Dr. Reichert’s team replicated the results from earlier experimental work looking at both signal attenuation and time lag associated with a foreign body capsule.

Detailed Discussion and Commentary

Local Inflammatory Response to Implanted Sensors

Peter Simpson, MS (Senior Director of R&D, Dexcom, Inc., San Diego, CA)

Mr. Peter Simpson gave an excellent overview of more than ten years of Dexcom research and development aimed at addressing the foreign body response and commercializing successive generations of CGM devices with improved accuracy and reliability. Dexcom R&D is now working on addressing the biological issues associated with the response of the body to sensors – Mr. Simpson emphasized that this is critical to solving some of the remaining problems with sensor performance and accuracy, including first day performance and factory calibration. The company’s goal is factory calibrated sensors with MARD values of 8-10% comparable to fingersticks.

• Mr. Simpson began with a review of Dexcom’s fully implanted long-term glucose sensor designed to last for one year. The challenge for that device was the foreign body response, which stemmed from the formation of a cellular barrier surrounding the sensor. That fibrous capsule was associated with increased noise on the sensor and a decrease in sensor signal. Dexcom developed a porous or textured surface led by Drs. Jim Brauker and Vicki Carr-Brendel to promote tissue in-growth to the sensor and minimize fibrous encapsulation. The long-term implanted sensor gave excellent results, but it had an unacceptable failure rate of approximately 25%. Dexcom determined in 2004 to take the technology from the long-term sensor and apply it towards the development of  a short-term percutaneous glucose sensor.
• Mr. Simpson briefly reviewed the Dexcom G4 Platinum continuous glucose monitor and discussed some of the product’s key design features. The G4 Platinum is more accurate than previous systems with a MARD of ~12-14%. The sensor is approximately 11 mm in length, inserts with a 26 gauge introducer needle at 45 degrees, and reaches an average depth of 8 mm. In the G4 Platinum, changes were made in the sensor materials that reduced the occurrence of noise spikes on the first day of use and improved the performance of the device.
  
  • Mr. Simpson explained that the Dexcom sensor is a flexible cylindrical wire vs. the more rigid, flat, and sharp edged Medtronic sensor. Interestingly, the sensing material for the Dexcom CGM is located around the entire circumference of the sensor, a contrast from Medtronic’s approach that locates the sensing material on one side of a planar surface.
• Dexcom is working on future CGM systems with the goal of factory calibration and MARD values of 8-10% comparable to fingersticks. Accordingly, Dexcom is focusing now on the effect of biological factors on sensor performance. The MARD distribution of the G4 Platinum MARD ranges from only 2-5% in vitro, but expand to 5-20% in vivo. Dexcom believes the higher average MARD in vivo and the wider distribution of MARDs in vivo are primarily the result of biological factors.
  
  • Sensor performance during the first day is affected by the tissue trauma induced by insertion. Histology cross section of a sensor in subcutaneous tissue from a porcine model shows the presence of neutrophils and even keratin brought down from the surface of the skin into the adipose tissue in the vicinity of the sensor. In earlier generations of Dexcom CGM devices,  a small number of patients experienced noise in the sensor signal during the first day of use. Algorithms in the early generation Dexcom CGM devices identified these periods and suppressed the display of data believed to be incorrect.  The latest generation, Dexcom’s G4 Platinum, fixed the first day noise by improving the biocompatibility of the tissue interfacing membrane.
  • Another problem that occurs in some Dexcom sensors is a temporary reduction in sensor sensitivity for up to six to eight hours after sensor insertion. This appears to be a biological phenomenon most likely associated with the presence of metabolically active cells in the vicinity of the sensor, which reduces the availability of glucose at the sensor itself.
  • Dexcom CGM sensor performance is most stable in the period two to seven days after sensor insertion. The final stage of sensor performance can be characterized by reduced sensitivity to glucose and an increase in sensor noise associated with the progression of a foreign body response.
  • Dexcom limits the approved commercial use of sensors to seven days in order to maximize the likelihood that the sensor will function at the highest level of accuracy and reliability, but the sensors are capable of much longer use. A recent study by Luijf et al. (Diab Tech Ther 2013) comparing three different commercially-available CGM systems found two Dexcom CGM sensors that lasted 50 days and one Dexcom CGM that lasted 82 days.

Questions and Answers

Dr. Buddy Ratner: The slide you showed of neutrophil deposition is most certainly associated with bacteria. Glad to hear you cited Jim Brauker’s early work on biomaterials, which inspired many of us to explore the effect of surface structure and porosity on the long-term stability of implanted materials.

Tissue-Integrating Sensors

Kristen Lloyd Helton, PhD (Profusa, South San Francisco, CA)

Dr. Helton is a co-founder of Profusa, an early stage medical device company based in South San Francisco. She reported on Profusa’s work to develop a tissue-integrating scaffold that can be used as a platform for fluorescent sensing of a number of different analytes, including glucose. This was the first public presentation of Profusa’s tissue integrating fluorescent sensor approach, and included data from both animal and human studies using their long-term miniaturized oxygen sensor implants. The Profusa team is working with Dr. Michael McShane at Texas A&M to develop a long-term, fluorescent glucose sensor that is also based on the tissue integrating concept. Dr. Soya Gamsey leads the Profusa glucose sensor team. Both Drs. McShane and Gamsey have worked for many years on fluorescent glucose sensors and are widely respected as experts in the field. Dr. Helton concluded her remarks by saying she hoped to present more glucose data by this time next year.

• Dr. Helton acknowledged the significant improvements in accuracy and performance of current CGM sensors. However, some intrinsic problems still remain with the current generation of sensors: need for once-weekly sensor insertion, the need for frequent calibration, poor performance on days one and two, and a bulky form factor (especially for small children).
• Profusa is attempting to develop long-term implanted sensors for glucose and other analytes. Dr. Helton continued the discussion from the morning about the effect of the foreign body response on sensor performance. She noted that the foreign body capsule can limit diffusion of glucose to the sensor, and also highlighted that the inflammatory response in the vicinity of sensors can include metabolically active cells that change the local glucose concentration. Dr. Helton briefly reviewed two papers she published in 2011 on the effect of motion at the biosensor interface on sensor performance.
• Dr. Helton and colleagues have developed a sensor with porous sizes that minimizes the thickness of the foreign body capsule formed around implants (similar to that described by the world famous biomaterials expert Dr. Buddy Ratner from the University of Washington in his keynote address at the beginning of the meeting – see below). She emphasized the importance of minimizing the size and stiffness of the implanted sensor, but also using materials that foster tissue ingrowth or integration into the sensor in order to bring microvasculature in close proximity to the sensing elements. The sensor is typically implanted 2-8 mm below the skin and then interrogated transdermally with near infrared light. Once the sensor is inserted into the tissue, its operation is non-invasive. The current generation of the Profusa sensor is small (500 microns by 2 mm cylinders), but they expect to achieve further miniaturization in the near future (e.g., 200 microns by 1 mm cylinder). The fluorescent emission from the sensor is then correlated with the concentration of the analyte (either oxygen or glucose). The sensor chemistry is embedded into the sensor polymer scaffold.
• Dr. Helton showed histology results from tissue solid sensors and sensors fabricated with the Profusa approach. A thick fibrous capsule surrounded the solid sensor, whereas there was good integration of local cells into Profusa’s tissue integrating sensor. Dr. Helton showed high-resolution ultrasound images comparing the mechanical stresses around a solid sensor and around a tissue-integrating sensor. The tissue-integrating sensor dramatically reduced the mechanical motion and sheer forces at the biosensor interface compared with the solid sensor.
• Measurement of tissue oxygen is an unmet clinical need. Pulse oximetry is a widely accepted technology, but it measures systemic oxygen and not local tissue oxygen. Profusa has tested their fluorescent tissue oxygen sensor in patients undergoing surgical treatment of peripheral arterial disease. The response time of the sensor is critical, since it can provide feedback to surgeons on whether their intervention (typically balloons or stenting to bypass occlusions in lower limbs) has changed the tissue oxygen concentration in the affected areas (e.g., lower limb or foot).
• Dr. Helton showed sensor data from an endovascular procedure on a 68-year-old patient with diabetes and peripheral arterial disease. The oxygen sensors placed in the foot showed an increase in local tissue oxygen concentration immediately after the procedure, thus providing feedback to the surgeons on the success or failure of their interventions. She also showed data taken for up to 28 days in two patients. In the first patient, the benefit of the surgical procedure was sustained, whereas in the second patient, the tissue oxygen concentration measurements showed a marked decline in local tissue oxygen and the need for a possible follow-up procedure.

Questions and Answers

Dr. David Klonoff (UCSF, San Francisco, CA): I think this is a very important area. It could be very important to monitor tissue oxygen levels in patients who have been treated with stents and other devices. How long do the sensors last and can they be removed?

Dr. Helton: We have oxygen sensors in healthy volunteers that have lasted for eight months. We think the sensors will continue to operate for at least 12 months. The sensors are very small and we think that they can be permanently left in the body.

Dr. Klonoff: Aren’t you concerned about potential toxicity from the sensor chemistry?

Dr. Helton: The small size of the sensor means that the amount of sensor chemistry is also very small. We have tested the chemical compounds of the sensor and they meet all the standards for safety and biocompatibility. For example, the LD50 (50% lethal dose) toxicity limits are literally millions of times below the danger levels.

Dr. Vigersky: Have you considered using this for remote monitoring? For heart disease or COPD?

Dr. Helton: Yes, we are looking into that.

Dr. Joseph: One other potential use could be to provide feedback for closed loop control of oxygen delivery at home.

Fully Implanted Sensors: Demise of the “Impermeable Tissue” Hypothesis

David Gough, PhD (UCSD/GlySens, San Diego, CA)

Dr. David Gough is a world-renown expert on glucose sensors, and MiniMed licensed his early work and used it for many years in its long-term implanted sensor program. Dr. Gough acknowledged the widely held belief that long-term implanted sensors will always be encapsulated by an impermeable membrane, but he believes the recent data from GlySens shows that this is not necessarily the case. He reported on recent work by his graduate students at UCSD and his colleagues at the company GlySens (Dr. Gough is a co-founder). As a reminder, GlySens is developing a long-term fully implanted glucose sensor consisting of a co-located glucose and oxygen sensor. The team has completed a small six-month study in humans (n=6). They performed a blood glucose clamp study one day every month for six months. Data from the sensor was compared to YSI readings and showed excellent agreement. Between clamp studies, subjects carried out fingerstick measurements. All sensor readings were blinded to the patients. In Q&A, Dr. Gough said they have not yet done accuracy analysis of the data set and are not able to provide data on the mean absolute relative difference (MARD) or Clark error grid. Said Dr. Gough, “Appropriately designed implanted glucose sensors can provide reliable glucose sensors for up to one year, despite the foreign body response.” This product has been in development for quite a long time, and we wonder what the timing looks like on larger and longer studies.

• The GlySens sensor measures the glucose modulated oxygen dependent current and uses an independent oxygen sensor to give a glucose measurement. The sensor is housed in a small titanium case and transmits data every two minutes (Gough et al., Sci & Translational Med 2010). The team has obtained data from subcutaneously implanted glucose sensors in a diabetic pig model out to 360 days. GlySens has shown continued sensor response in 17 animals for over one year, and in one animal, the sensor functioned out to 520 days. Notably, only occasional recalibration was needed. Their oxygen sensor was tested before implantation and after explantation in the gas phase to confirm stability (Anal Chem 2004). Dr. Gough has submitted this work to Biomaterials for publication.
• In their recent work, the team has observed very little effect of encapsulation in animals. Generally speaking, a dense fibrous relative avascular tissue capsule is expected to form around an implant within a number of weeks, thereby rendering the implanted device ineffective, especially in sensor applications.
• Dr. Gough’s group has developed a model of glucose dynamics including transport from blood to interstitial fluid and diffusion into the sensor. Dr. Gough claims that the delay in the sensor is primarily due to diffusion in the tissue.
• According to Dr. Gough, the long-term GlySens sensor does not need frequent recalibration. This is due in part to the ability to wait for the foreign body response to stabilize before clinical use of the sensor.

Integration of Sub-millimeter Implanted Wireless Sensors

Axel Scherer, PhD (California Institute of Technology, Pasadena, CA)

Dr. Scherer is a professor of physics, applied physics, and medical engineering at Caltech. His work on miniaturized fully implanted continuous glucose sensors is supported in part by Sanofi. The Caltech sensor is a thin 1.4 mm x 1.4 mm square and is based on glucose oxidase – notably, that is the size of Franklin Roosevelt’s eye on a US dime, and the team believes it can be made smaller by a factor of 10 than the University of Toronto or Google X miniaturized glucose sensors. The entire sensor implant is based on a CMOS design (Complementary Metal Oxide Semiconductor - a technology for constructing integrated circuits used in microprocessors, microcontrollers, and image sensors). The key to the miniaturization of the Caltech device being developed by Dr. Scherer’s group is an inductive power system that wirelessly provides power to the sensor and communicates the result of the glucose sensor reading to an external display device. Dr. Scherer’s team is beginning to explore the operation of their miniaturized devices in animal models. The duration of the device operation is not currently known. Dr. Scherer believes that it will be possible to extend the duration of use out to three to six months. He suggested multiple sensors could be removed surgically once every few years.

• Dr. Scherer began his talk by contrasting the rise in health care costs and the fall in the size of consumer electronics. There is a 100-fold reduction in consumer electronic spatial size every 10 years. Ultimately, the limits to the physical size of electronics are power (battery) and communication. Google X has recently reported its work on a miniaturized glucose sensor small enough to be incorporated into a contact lens. This project is still early stage, but we were encouraged to see Google getting into diabetes.
• There are numerous advantages of miniaturization. These devices can be injectable and will not need surgery. There is also evidence that miniaturizing the sensor will reduce the magnitude of the foreign body response around the implant. One of the advantages of the Caltech device is that it is so small that it will not move relative to the surrounding tissue, hence there is no change in the signal impedance. Another advantage is that the fabrication can be done with current Silicon Valley chip manufacturing processes at an extremely low cost (less than ten cents per sensor).
• The Caltech sensor measures the redox current and converts into a binary digital signal. The sensor communicates with a reader worn outside the body using inductive coupling. This has been shown to work with mock sensors embedded up to 1 cm in chicken breasts.
• The sensor chemistry works well in vitro and has been shown to work up to three weeks in vivo. A typical sensor is built using a nafion surface, a hydrogel containing glucose oxidase and a standard reference electrode. The sensor uses standard CMOS technology. The electrode surface area can be enhanced resulting in higher signal to noise and better performance. The scar tissue thickness surrounding these microdevices is approximately 60 microns – quite small. Power requirements are currently at 3mW but can be reduced to 2mW or less. The sensor technology can be applied to other analytes, potentially including insulin.
• Another approach to achieve even further miniaturization would be used to use optical methods rather than inductive methods to deliver power to the implant. Dr. Scherer’s group has designed a sub-millimeter sensor platform that can be powered by infrared light. Since one of the major determinants of the implant size is the inductive power capture, the use of optical methods for delivering energy to the implant should make possible even further miniaturization. Dr. Scherer believes that the current size of 0.4 mm³ volume can be reduced to 0.18 mm³ in the near future and ultimately to 0.03 mm³.

Questions and Answers

Dr. David Klonoff: Have you looked at the tissue response to the sensor?

Dr. Scherer: We are just looking at this in our rodent models. The sensors seem to be working well. The measured thickness of the scar tissue is approximately 60 microns, but these results were obtained after implanting the sensors with a surgical method rather than with a microneedle insertion method.

Dr. Klonoff: Fluorescent sensors may offer the ability for fine-tuning of the response in hypoglycemia and then separately in hyperglycemia.

Role of Immune Cells

Dianne Burgess, PhD (University of Connecticut, Storrs, CT)

Dr. Burgess’ lab has focused on the use of dexamethasone, a broad and potent anti-inflammatory agent. Dexamethasone not only inhibits inflammation, but also inhibits the formation of capsules around implants. Her team at the University of Connecticut is developing a fully implanted miniaturized glucose sensor (0.5 mm x 5 mm). This is a collaboration with other researchers at the University of Connecticut, including Drs. Faquir Jain and Fotios Papadimitrakopoulos. The sensor coating consists of microspheres that release dexamethasone, which inhibit the foreign body response. The sensor is inserted into the volar region of the wrist. A custom wrist-worn watch receives the telemetry from the implanted sensor and can broadcast the sensor signal to a phone or elsewhere. The sensors can also be removed using a needle introduced under the skin. Dr. Burgess and Dr. Papadimitrakopoulos have licensed their intellectual property to a start-up company, Biorasis Inc, and are attempting to commercialize the technology.

• The University of Connecticut sensor is based on glucose oxidase and uses limiting membranes to eliminate interferences from common pharmacologic agents such as acetaminophen. Earlier tests were done on percutaneous sensors in a rat model and showed good quantitative agreement with reference blood glucose readings. Recent studies have been done with fully implanted version in mini-pigs and have continued to show encouraging results, although it has been challenging to keep the proximity reader in place on the pigs. Importantly, foreign body response may also be different in different animal models. In the mini-pig, there may also be differences depending on the skin thickness from one location on the pig to another.
• Dr. Burgess is now exploring the effect of implant size and insertion needle diameter on the foreign body response. Her team is hoping to reduce the sensor size to 0.3 mm x 3 mm. The acute inflammatory response depends on the size of the insertion needles, whereas the final chronic response appears to depend on the implant size. Dr. Burgess believes that minimizing the size of the implant will enable the team to better reduce the foreign body response and extend the lifetime of the sensor beyond three months.
• Dr. Burgess also provided a brief history of the foreign body response. She noted that the time course of the foreign body response has been well characterized by researchers (Grainger et al., Nature Nanotechnology 2013). Giant cells are mobilized to break down the implant, but they are not able to do so because of the size of the implants. Instead fibroblasts are activated and recruited by immune cells – this leads directly to the formation of a foreign body capsule. After seven days, the acute body response to sensor insertion has passed.

Questions and Answers

Dr. Klonoff: what are the systemic effects of dexamethasone? What are your plans for removing the sensor?

Dr. Burgess: The dose is extremely small – well below the level expected to have a systemic effect on man. We were not able to detect dexamethasone using radiolabels at this level.

Dr. Ratner: The dexamethasone strategy may inhibit the foreign body capsule, but it may also inhibit vascularization that could be beneficial to long-term sensor viability.

Dr. Burgess: We have done some studies with dexamethasone to suppress the foreign body response and VEGF to increase the long-term vascularization.

Dr. Reichert: We are also playing in this space. We are doing studies now using dexamethasone release from textured coatings.

Dr. Burgess: We also have a new method for inhibiting migration of the sensor that we believe will be successful. That is the subject of a pending patent application.

Using Nothing To Improve The Performance of Implanted Glucose Sensors And Insulin Delivery Systems

Buddy Ratner, PhD (University of Washington, Seattle, Washington)

Dr. Buddy Ratner gave the conference’s keynote address and presented new data on the use of micro-porous scaffolds to facilitate tissue ingrowth into implanted devices. The intellectual property from the University of Washington has been licensed to a startup called Heliosonics. Two other startups are using a similar approach as well: Profusa for long-term oxygen and glucose sensors and iStar for an end ophthalmic implant for the treatment of glaucoma.

• Dr. Ratner discussed the current methods of insulin delivery and continuous glucose sensing (“percutaneous”) and the serious biocompatibility problems associated with them: protein accumulation on the sensor occurs immediately, macrophage attacks occurs by 48 hours, frustrated phagocytosis occurs by five days, and encapsulation/walling off the sensors typically occurs by three weeks. He noted that encapsulation is seen in many implanted devices and may be a source of countless adverse events reported every year to the FDA.
• Dr. Ratner highlighted two approaches to address the foreign body reaction: novel engineered porous scaffolds and non-fouling materials. The problem with most scaffolds is that they have a broad distribution of pore sizes. One approach is to use micro-spheres and process them to leave a porous structure with a single or dominant characteristic size (i.e., a sphere template with porous hydrogels). A student in Dr. Ratner’s lab, Andrew Marshall, found that pore size had a significant effect on vascular density around an implant. Further, different pore sizes were found in repeated studies to have different impacts on the formation of fibrous capsules around the implant. Above 60 microns, the collagen capsules were dense; below 30 microns, there was greater tissue in-growth into the sensor and a thinner collagen capsule surrounding the implant.
• In addition to the porous scaffold approach, Dr. Ratner is also working on a no-fouling surface that inhibits adsorbed proteins on the surface of the implant. After this idea was proposed in 1995, many groups attempted to develop non-fouling surfaces. Recent work on zwitterionic structures shows promise in minimizing the adhesion of proteins to surfaces.
• Typical scar formation and encapsulation may be replaced by vascular integration promoted by new types of biomaterials and new biomaterial structures. Insulin delivery and glucose sensing might be revolutionized through the use of these new approaches for tissue integrating capsules and materials. Advances in 3D printing technology may also make it possible to explore different pore sizes and geometries.
• Dr. Ratner has gone on to commercialize this approach in a start-up called Heliosonics. One application is for an ophthalmic implant in a company called iStar, which received CE mark for a scleric drain for the treatment of glaucoma.
• Profusa, a start-up company in South San Francisco (see above), is developing an oxygen sensor and a glucose sensor using a similar approach. Sensing nanospheres were embedded in the scaffold material for the measurement of tissue oxygen. After 170 days, sensors with the porous scaffold structure retained a rapid response to oxygen, whereas the solid sensors had a slow response indicative of fibrous encapsulation around the sensor.

Questions and Answers

Dr. Jeffrey Joseph (Thomas Jefferson University, Philadelphia, PA): We may see vascularity at 90 days, but we usually lose it subsequently. Do you see vascularity that persists?

Dr. Ratner: We see excellent vascularity out to 270 days. The tissue in the implant typically looks like the local tissue. We have no reason to believe the vascularity won’t persist indefinitely.

Tissue Response to Vascular Sensor Implantation

Jeffrey Joseph, DO (Thomas Jefferson University, Philadelphia, PA)

Dr. Jeffrey Joseph described his experience evaluating intravascular glucose sensors for use in hospitalized patients, which has included work with GluMetrics, Edwards, Animas (in the 1990s), and others.

• All objects placed in the bloodstream are subject to cellular adhesion and thrombus formation. However, vessel size and flow affect the extent of adhesion and clotting – placement of catheters in large vessels with high flow rates makes them less susceptible to adhesion and clotting (e.g., time-resolved high-resolution ultrasounds show that thrombus forms on catheters and then dislodges). Thrombus formation and break-up on catheters in the bloodstream is a very dynamic process, which begins following protein and platelet adhesion to a surface.
• Dr. Joseph showed data from catheters in canine studies – all of the catheters, even catheters with anti-thrombogenic coatings, were found to induce thrombus in vessels with low flow rates. These studies were done as part of the development of an intravascular fluorescent glucose sensor by GluMetrics – a now defunct company based in Irvine, California. The sensor worked well, but it was not reliable due to thrombus formation in the vessel.
• Dr. Joseph has also worked recently with Edwards Lifesciences (Irvine, CA), who is also developing an intravenous blood glucose monitoring system for use in hospitalized patients. The Edwards product currently uses a heparin solution that flushes the catheter to reduce the risk of thrombus formation on the sensor.
• There are two companies developing microdialysis glucose sensors for the hospital: Maquet CMA Mircodialysis and A. Menarini Diagnostics.
• Dr. Joseph discussed his work with Animas Corporation from the 1990s on a long-term implantable optical blood glucose sensor. He showed data from canine studies in which a miniature optical spectrometer was placed around an artery in an animal and used to measure blood glucose. Dr. Joseph characterized this as an “implanted extravascular sensor.” In another approach, they placed a miniature optical fiber directly into the vessel. He showed data from two animals with excellent correlation to reference glucose measurements.
• Finally, Dr. Joseph discussed home and ambulatory monitoring of vital signs. Millar Instruments has developed a long-term implantable blood pressure sensor that has become the gold standard in veterinary research. In addition, Konigsberg Instruments has developed a long-term implantable sensor to monitor patients in heart failure.

Cell Biomaterial Interactions: Engineering Tissue Response to Implants

Andres Garcia, PhD (Georgia Institute of Technology, Atlanta, GA)

Dr. Garcia reported on the development of new fluorescent probes to image the progression of inflammation in response to biomedical device implants. These methods are shedding new light on the mechanism of inflammation on encapsulation of implants. Dr. Garcia also shared details on his work to modify receptors on cells that control the adhesion of proteins onto the surfaces of implanted materials.

• Dr. Garcia’s research used hydrocyanine fluorescent probes to image reactive oxygen species and study the progression of the inflammatory response to implants (Selvam et al., Biomaterials 2011). Reactive oxygen species (ROS) are secreted by macrophages as part of the inflammatory response. The fluorescent probes emitted in the infrared and could be used with full-body imaging using IVIS devices. The research revealed that inflammatory cells produce ROS, and ROS readings were found to correlate with fibrosis. These methods were then used to assess the effect of therapeutic drug delivery to reduce implant-associated inflammation. Current research by Dr. Garcia’s group has also focused on differentiating between the effect of infection and inflammation. In these studies, he used two separate fluorescent probes, one looking at nitric oxide (NO) and the other at ROS (Suriet al. JBMRA 2014 and Dijanski et al., Acta Biomater 2014).
• The passive strategy to modulating inflammation relies on coatings to resist adsorption of proteins onto the implant surface; the active strategy relies on the release of anti-inflammatory and bioactive agents. Despite extensive research, the passive strategy has shown reduction in short- and intermediate term fibrous capsule formation, but does not appear effective over the long-term. PEG hydrogel coating was found to reduce protein absorption and cell adhesion and was used in conjunction with an anti-inflammatory agent (IL-1Ra released via inflammation-related proteases). Anti-inflammatory coatings appear to work well with subcutaneous tissue, but work poorly in the brain – in neural electrodes, there was no difference between uncoated electrodes, the PEG coating alone, and the PEG coating with the anti-inflammatory agents. The most important factors appeared to be the trauma of electrode insertion and the continued presence of the electrode in the tissue. There was some benefit with the PEG and anti-inflammatory agents on neuronal survival in the vicinity of the electrode.
• An alternative strategy is to try to promote integration of tissue into the implant (i.e., engineered biomaterials for tissue integration). Many groups are using adhesion receptors to foster local adaptation of the tissue to an implant. Dr. Garcia’s laboratory works primarily on bone and has shown that manipulation of these receptors improves integration of titanium implants in bone (Garcia et al., 2005). Although the NIH turned down their initial grant applications, Dr. Garcia’s group was successful in showing improved titanium implant osseo-integration using these methods (Petrie et al., Science & Trans Med 2010).

Biomaterials which Elicit No Immune Response: Zwitterionic Functional Materials

Shaoyi Jiang, PhD (University of Washington, Seattle, WA)

Dr. Jiang gave a review of differences between standard biocompatible polymers, such as polyethylene glycol (PEG), and electrostatic/zwitterionic (ZI) materials. The premise is that hydration is a fundamental parameter for creating biocompatible materials. PEG materials become less hydrophilic at elevated temperatures, while zwitterionic material helps to increase hydration and remains highly hydrophilic at body temperature (“superhydrophilic”). Notably, hydrogel coated glucose sensors with a zwitterionic coating in whole blood were found to continue to function up to 42 days.

• There are naturally occurring zwitterionic and biomimetic materials. Ultra low fouling can be obtained with zwitterionic materials. Hydrogels have low fouling, but adding zwitterionic materials changed the fouling significantly (Carr et al., Biomaterials 2010; Carr et al., Biomaterials 2011).

Dr. Jiang’s group showed no capsule around implanted hydrogels after three months of subcutaneous implantation in a mouse model (Zhang L. et al., Nature Biotechnology 2013). Dr. Jiang was also able to show no immunological response from zwitterionic materials compared with PEG for implants in circulation in the body. There were high levels of immune markers after one week for the PEG coatings and low levels for the materials with the zwitterionic coatings. Dr. Jiang suggested that zwitterionic coatings could be used to improve preservation of biocompatibility of cells and proteins in the body.

Modeling Lag Time for Glucose Sensors

William “Monty” Reichert, PhD (Duke University, Durham, NC)

Dr. William Reichert discussed his team’s basic science work on sensor lag time. The team has found that maximizing sensor accuracy and minimizing sensor lag requires a thin capsule and blood vessels in the vicinity of the sensor. He and his colleagues are now using fibrin gels as an in vitro test bed for investigating the effects of cells on glucose consumption in the vicinity of the sensor. The team is trying to develop an in vitro test bed with Medtronic to evaluate the effect of biofouling on sensor performance.

• In recent years, our views of the sensor-tissue interface have changed. The sensor response to glucose now appears to more perfusion limited than diffusion limited. We no longer worry as much about the effect of adsorbed protein films on sensor function. However, there is increasing attention on the role of metabolic effects, especially from immune cells, on sensor performance.
• Dr. Reichert and his students are interested in developing a model for how biofouling affects sensor performance from first principles. This follows Heidi Koshwanez’s work in Dr. Reichert’s laboratory modeling the performance of the Medtronic Sof-Sensor. Sensor lag can be characterized by the time lag between the sensor maximum and the reference blood glucose maximum. Dr. Reichert and colleagues have developed a source-sink model of glucose sensor performance, including the effect of foreign body response (Anal Bioanal Chem 2010). Their model has replicated the results from earlier experimental work looking at both signal attenuation and time lag associated with a foreign body capsule (Koschwanez 2008).
• The team has found that maximizing sensor accuracy and minimizing sensor lag requires a thin capsule and blood vessels in the vicinity of the sensor. The Duke team has also attempted to repeat the work of Klueh et al. from the University of Connecticut (2007), which showed that protein adhesions on a sensor in blood causes an attenuation of the sensor signal, but not an increase in the lag time – a protein biofouling layer imposes no transport resistance to glucose detection. Surprisingly, increasing the density of these thin films does not appear to increase the lag time. However, the presence of macrophages changes the temporal response of the sensor. Accumulated white blood cells (but not red blood cells) generate a glucose depletion zone near the sensor surface that affects the sensor reading.
• Klueh et al have recently reported that inserting macrophages at the location of a sensor causes a drop in the sensor signal (Biomaterials, 2014). The Duke team is now using fibrin gels as an in vitro test bed for investigating the effects of cells on glucose consumption in the vicinity of the sensor. Duke is trying to develop an in vitro test bed with Medtronic to evaluate the effect of biofouling on sensor performance.

--by Tom Peyser, Adam Brown, and Kelly Close