2ND INTERNATIONAL CONFERENCE ON ADVANCED TECHNOLOGIES AND TREATMENTS FOR DIABETES (ATTD)

FEBRUARY 25-28, 2009; ATHENS, GREECE — DAY #3 DRAFT

The second full day of the ATTD meeting, although not quite as packed as yesterday, was likewise full of many interesting speakers and discussions. Medtronic stole the show this time round with European experiences of its newest sensor integrated pump systems, and snippets about a six-day sensor, use of CGM data for therapeutic decisions, and combined glucose sensing and insulin delivery sites. To boot, Abbott released news that calibration time for the new Navigator sensor being launched in Europe would be reduced from 10 hours to one hour – a very welcome update. It also has a pediatric claim, from what we understand.

We particularly enjoyed Dr. Stephanie Amiel’s discussion of the new Medtronic X54 insulin pump, which incorporates a low-glucose suspend feature, a step we believe is very important in moving towards the artificial pancreas. We also learned a lot from Dr. Lois Jovanovic who confirmed the benefit of CGM use in patients with gestational diabetes. Finally, we enjoyed the patient perspective provided by Dr. Stuart Bootle and Michal Jelinski who discussed their patients’ experience on insulin pumps as part of the Roche-sponsored symposia.

General Session Highlights

  • Stephanie Amiel, MD (King’s College London, London, UK) discussed the new Medtronic X54 insulin pump that has low-glucose suspend feature – this was an excellent talk as it gave details of this new feature that we hadn’t heard much about yet. According to her work, initial reports from patients using the device were very positive. Patients liked the set-up, equipment, and literature that comes with the pump, and they felt more secure knowing that insulin delivery will shut off if their blood glucose goes into the hypoglycemic range. However, patients sometimes found the alarms irritating (we see this with all CGM devices, it seems!). We see this feature as a big step towards a safe closed loop system — unfortunately, it is not currently available in the US, and we believe it will be some time before it is since this is a PMA submission and the trial is being designed at present.
  • Lois Jovanovic, MD (Sansum Diabetes Research Institute, Santa Barbara, CA) summarized her experience with the use of CGM in pregnancy. As demonstrated by the Hyperglycemia and Adverse Pregnancy Outcomes (HAPO) study, hyperglycemia during pregnancy increases the risk of complications, including birth defects and macrosomia. While patients can often be adequately controlled without CGM, it requires a large number of fingersticks each day (she estimated 8-10) to ensure that glucose levels are within the target range. However, many patients who appear to be in good control may experience significant unreported hyperglycemia, which can be uncovered by the use of a CGM device. It is Dr. Jovanovic’s opinion that the devices can be put to good use at all stages of pregnancy. Her A1c goal for women during pregnancy is 5.0% or less.
  • John Mastrototaro, PhD (Medtronic Diabetes, Los Angeles, CA) discussed improvements to the current Medtronic’s CGM products such as moving to a six-day sensor from a three-day and reducing start-up time from two hours to 30 minutes. He noted other important areas the company was working, such as reducing calibration requirements (and even pursuing a fingerstick replacement claim), reducing sensor size, combining the sensor and insulin delivery sites, and improving sensor accuracy, especially in the hypoglycemic range. He mentioned that Medtronic was investigating the role of glucagon and Symlin in the artificial pancreas, alongside insulin. We heard a great deal overall about the artificial pancreas and he described a Yale study where “time in zone” for someone with an A1c of 7.1% increased from 58% to 82% with the calculated new A1c dropping to 5.8%. He also discussed wireless connectivity from sensors to mobile phones, cars, and home remote monitoring systems, although he didn’t give any sense of when we would see these products. Finally, he discussed Medtronic’s commitment to the artificial pancreas and recent trial data. The success that has been achieved with current closed loop algorithms suggests that a robust ambulatory product, once approved could lead to very good A1c control. In a conversation afterward, Dr. Mastrototaro indicated that Medtronic had mapped out a plan to obtain regulatory approval for an artificial pancreas system step by step, and the company was working with the FDA to make it happen. Despite a genuine commitment to move as fast as possible from the FDA, we imagine that it will still take a considerable time; Europe is already ahead with the low-glucose suspend feature.
  • Thomas Danne, MD (KinderKrankenhaus Auf Der Bult, Hannover, Germany) expressed his wish to create an international center of reference as part of his ISPAD legacy. He has established this in the form of SWEET (Better control in pediatric and adolescent diabetes: working to create Centers of Reference) – which is a European pediatric center focused on improving control for children with diabetes. SWEET has received government and corporate funding and has issued a status report highlighting the broad variation in policy and care across the EU. Its next project is a central repository of information from all countries that can be used for benchmarking. He also hopes to use this facility for telemedicine and patient data.
  • Bruce Buckingham, MD (Stanford Medical Center, Palo Alto, CA) and Eyal Dassau, PhD (Sansum Diabetes Center, Santa Barbara, CA) discussed progress toward a closed loop. The first topic, presented by Dr. Buckingham, was the automatic detection of insulin set failures. This is an extremely important aspect of taking the closed loop into the real world, and we saw Medtronic make mention of it earlier in their development pipeline. It’s hard to distinguish between a set failure and an unannounced meal with a missed bolus, but if we add in meal announcement, it becomes quite possible. Next, Dr. Buckingham outlined results of a closed loop algorithm that was tested using the FDA approved in silico simulator. The algorithm performed well overnight, but ran a little high, casting some doubt over the realism of some of the simulated individuals. Dr. Dassau presented some important infrastructure for closed loop projects – a software platform called APS, a hardware platform that interfaces to devices and virtual patients, and an ‘algorithm on a chip’ which worked well in initial testing.
  • Moshe Phillip, MD (Schneider Children’s Medical Center, Israel) created a Diabetes Technology Center within his Institute to work on the closed loop. He currently has two closed loop approaches. One is the JDRF collaboration previously presented. The second is the “MD-Logic” approach. There was not a lot of visibility into the exact methods behind the MD- Logic algorithm, but it does seem to be guided by clinician heuristics and is tailored to an individual patient over a number of visits to the clinic. The results showed good closed loop control without meal announcement.
  • Claudio Cobelli, MD (University of Padova, Padova, Italy) discussed the use of in silico testing to eliminate the need for animal tests. His is part of a research consortium consisting of members from the University of Virginia, University of Padova (modeling), the University of Pavia (control engineering), and the University of Montpellier, with support from the JDRF Artificial Pancreas project. Their efforts in driving forward closed loop algorithms are significant, and the in silico simulation that was cleared by the FDA last year is a very efficient and cost effective means for development and testing of new closed loop algorithms. A model predictive control (MPC) algorithm tested in this way has received FDA approval for human trials, which are going well. These clinical and in silico trials will also contribute to the refinement of the simulation environment.
  • Francine Kaufman, MD (Children’s Hospital, Los Angeles, CA) cited a study showing that 40% of eighth graders have impaired fasting glucose and 25% have impaired glucose tolerance, often reaching the criteria for pre-diabetes. There are about 3,700 children a year diagnosed with type 2 diabetes of which 26% have hypertension and 59% have dyslipidemia. It’s to be expected that these children might experience cardiovascular events as young as their 30s and 40s. The HEALTHY trial was established to determine if environmental changes can reduce risk factors for type 2 diabetes in schools. The trial runs across 42 schools and follows 6,400 sixth graders for two years – attempting to focus on physical activity, nutrition, and behavior modification. We look forward to hearing results of this important trial that could have profound implications for future generations. On a broader front, California has been making progress on health related legislation such as banning junk food and soda in schools, requiring nutrition information in restaurants, and banning trans fats. There are many other things that Dr. Kaufman believes we still need to do, such as regulating marketing to children, improving the school and work environment, introducing zoning to promote healthy food outlets, and limiting access to junk food.
  • Asimina Mitrakou, MD (National University, Athens, Greece) gave a brief overview of the rationale behind the JDRF Hellas project. The project was designed to investigate the utility of CGM devices in controlling hyperglycemia and hypoglycemia in young adults during periods of moderate-to-intense exercise. Exercise presents a particular challenge for patients who are trying to stay in good glycemic control since glucose levels tend to have a delayed response to exercise. CGM appears to be a valuable tool for managing this problem.
  • Andriani Vazaiou, MD (Athens, Greece) presented the results of the JDRF Hellas pilot study, which examined the effectiveness of CGM in decreasing hypo- and hyperglycemia seen in young adults and associated with exercise. Forty-three patients with type 1 diabetes participated in sessions of four to six hours of intensive mountain climbing. Twenty-eight of the participants elected to use CGM and 13 used SMBG. The results showed that CGM use was effective in decreasing the overall amount of time spent outside the glycemic target zone, and it was also effective in detecting otherwise silent hypoglycemia during the night following the exercise. This indicates that CGM may have an important use in this area, which is encouraging because glucose control during exercise is one of the biggest challenges in diabetes management.
  • Bruce Buckingham, MD (Stanford, Palo Alto, CA) gave a very brief take on the results of the Hellas pilot study and how it might be translated into a clinical guideline. He pointed out that there is no ‘safe’ glucose value at night after exercise, but a possible target is about 130-140 mg/dL. Patients show significant variability in their nighttime glucose variation after afternoon exercise, so it’s very difficult to come up with a safe guidelinethat holds true in all patients. The drop in glucose after activity is about 60-70 mg/dL according to the DirecNet study, but there are no data about glucose drops for specific exercise activities. We would speculate that a low-glucose insulin suspend feature like that on the new Medtronic X54 insulin pump might help to solve this problem.
  • Satish Garg, MD (University of Colorado at Denver, CO, USA) discussed the evidence that CGM use improves outcomes. We have reached the point with CGM technology, he said, that we need to reach out to the majority of patients and encourage them to use it widely, in order for them to realize its medical benefits. As the JDRF CGM study results presented at EASD 2008 showed, improvements were only seen in individuals who used the device and modified their behavior according to the results. CGM reduces glycemic variability, hypoglycemia, and A1c and works for both pumpers and those on MDI. Notably, it doesn’t matter whether we start at higher or lower A1cs – CGM still reduces glycemic variability. If patients start with A1c >9.0% they can double the time spent in normoglycemia. CGM also allows patients to improve their A1c without increasing hypoglycemia. The future hope for CGM is a 12-month implantable or non-invasive sensor.
  • Christos Zoupas, MD (Hygeia General Hospital, Athens, Greece), the mayor of a section of Athens and a physician at Hygeia General Hospital, discussed his experience and opinions regarding CGM. He believes that CGM can be used to gain important improvements in patients’ glycemic control, both in terms of A1c and glycemic variability. In the Hygeia hospital, they use the Medtronic Guardian CGM system and train patients over a period of several months on combined CGM/pump use. Notably, a study conducted in Greek hospitals has shown an A1c decrease of about 1.5% from a baseline of 8.3% after a year of using the Guardian system. Dr. Zoupas thinks that the Medtronic CareLink software is particularly helpful in managing patient data and improving outcomes.
  • John Pickup, DPhil (King’s College London, London, UK) gave a lecture providing another window into the developing world of non-invasive or minimally-invasive glucose sensing technology. His group is developing fluorescent proteins that change their light-transmitting properties in a glucose-dependent manner to detect surrounding glucose levels. He mentioned a variety of different molecules that could be used in this manner, but says that his team has seen the most promise with a bacterial glucose-binding protein encapsulated in micro- capsules. He envisions a system in which these capsules are embedded in the interstitial space and connected to the surface via a fiber-optic cable through which glucose readings can be taken. We were disappointed to hear, in response to a question, that such a system is still five to ten years away from clinical use.
  • Tadej Battelino, MD (University Medical Center, Ljubljana, Slovenia) is a strong proponent of CGM for all patients. He believes that the technology is now mature enough for widespread use. He has collected solid data that CGM is safe in all age groups. He cited evidence from at least three randomized controlled trials that CGM improves metabolic control and reduces hypoglycemia. He asserted that CGM also facilitates education and improves quality of life. The hope is that it will also reduce complications in the long term, and we await trials to validate this hypothesis. In the meantime, there are many CGM trials in process – over ten with Medtronic products alone.

 

Corporate Sponsored Session Highlights

Roche: From Information Management to Insulin Management

  • Ewa Pankowska, MD (Medical Academy, Warsaw, Poland) discussed the role of information management in diabetes care. She believes that access to better patient information can change the standards of medical practice. First, she focused on electronic data collection, including data on insulin administration and glucose measurement. Secondly, she discussed the use of data to manage large groups of patients and get easy access to their treatment results. We are glad to see patient management receiving so much attention at this conference. However, even if data is available a secondary problem will be getting physicians to access it. Overall, we would stress the importance of an easy-to-use system that can interface with existing medical records systems — a system that cannot link in with others will not be as significant of a step forward.
  • Thomas Danne, MD (KinderKrankenhaus Auf Der Bult, Hannover, Germany) discussed the role for data management in overcoming four major problems with insulin therapy: finding the right basal dose, controlling postprandial glucose, making the best use of pump therapy, and avoiding adverse events. We agree with his conclusion that better management tools could help patients to meet their goals across the board—however, we would caution that few if any systems available today meet the standard necessary for such wide-ranging success.
  • Stuart Bootle, MD (Derbyshire, UK) and Michal Jelinski (Poland), both have type 1 diabetes, and talked about their experience with diabetes management and technology. Dr. Stuart Bootle is a practicing general practitioner in the UK and has had diabetes for more than 20 years. He talked about his experience with pump therapy and how it has revolutionized his diabetes care. He also briefly mentioned the Accu-Chek Combo, which includes a data-management system that he says has again changed how he manages his therapy. The company has indicated that the Combo will be available in 2009. Jelinski is a three-time world champion rower and Olympic gold medalist, who has had diabetes for the past five years. Like Dr. Bootle, he described how switching to a pump in just the past year has improved his treatment and fine-tuned his performance in competition. This was a very interesting presentation to get a feel for insulin devices ‘on-the-ground’, and the audience was really getting into the spirit. It was followed by a very interesting Q+A session, which included comments on glucose variability, insulin therapy in children, and other topics.

Abbott: Using Freestyle Navigator in Clinical Practice — Who, When, and How?

  • Gerard Reach, MD (Hopital Avicenne, Bobigny, France) showed how CGM fills an important need for patients, and solves the problems of SMBG. It addresses the double- edged sword of glycemic variability and hypoglycemia. Patients will use it if it is convenient and comfortable.
  • Jan Bolinder, MD (Karolinska Institutet, Sweden) gave us a measured presentation about CGM progress. His main argument was that although we are converts to the CGM cause, we still don’t have enough definitive data from randomized controlled trials to draw rigorous conclusions in many areas – and this needs to be done urgently. This message was a little pessimistic for true believers, but it is surely correct from a pure evidence-based medicine perspective.
  • Bruce Buckingham, MD (Stanford Medical Center, Palo Alto, CA) presented results from DirecNet – a network of centers researching CGM use (among other therapies and technologies). Participants’ A1cs went from 7.1% to 6.8%, and fingerstick frequency went from five/day to four/day, following CGM initiation. The patients who got the most benefit from CGM were looking at the sensor 20-40 times a day. Dr. Buckingham thinks that kids can understand CGM devices better than their parents, which makes them more independent. His results were very positive for CGM, and he sprinkled them with lots of valuable clinical learnings. Notably, during Q&A, when asked what it would take to get a replacement claim, Dr. Arleen Pinkos of the FDA said that manufacturers should put patients in a simulated home situation and see if they stay safe – the FDA would be interested in seeing these data, she said. It was also mentioned during Q&A that Abbott was launching a new sensor in Europe that has a one-hour calibration rather than ten hours.
Table of Contents 

 

Detailed Discussion and Commentary

Continuous Glucose Monitoring

THE CONTINUOUS USE OF REAL TIME GLUCOSE MONITORING COMBINED WITH INSULIN PUMP, AS A MEANS OF NORMALIZING GLUCOSE VARIABILITY AND HBA1C IN IDDM

Christos Zoupas, MD (Hygeia General Hospital, Athens, Greece)

Dr. Zoupas, the mayor of a section of Athens and a physician at Hygeia General Hospital, discussed his experience and opinions regarding CGM. He believes that CGM can be used to gain important improvements in patients’ glycemic control, both in terms of A1c and glycemic variability. In the Hygeia hospital, they use the Medtronic Guardian CGM system and train patients over a period of several months on combined CGM/pump use. Notably, a study conducted in Greek hospitals has shown an A1c decrease of about 1.5% from a baseline of 8.3% after a year of using the Guardian system. Dr. Zoupas thinks that the Medtronic CareLink software is particularly helpful in managing patient data and improving outcomes.

  • Despite the clear benefits of good glycemic control, patients continue to fall short of glycemic targets. Hypoglycemia remains a major problem, and control is poor in two-thirds of patients. The average A1c in patients on pumps is 8.6% compared to the average A1c in type 1 patients overall that is around 9.0%. (We assume these figures are for Greece.)
  • Continuous glucose monitoring can be used to improve patients’ control. The GuardControl trial showed that patients using CGM improved control by about 0.6% from a baseline around 9.0%. In addition, CGM can be used to reduce glycemic variability. A person can gain important benefits from pump and CGM use, including reducing the dawn phenomenon, reducing treatment complexity, and improving quality of life. Dr. Zoupas cautioned that CGM systems are not a cure, but a tool requiring appropriate training for optimal use. He concentrated on the importance of glucose trends over absolute numbers. Notably, a study conducted at Greek hospitals showed a decrease of about 1.5% A1c over 12 months, from a baseline of 8.3%, using the Guardian. Patients’ total insulin dose was reduced significantly, as was hypoglycemia unawareness. These are very impressive results, particularly for the Guardian – we know that system has been improved significantly.
  • Dr. Zoupas’s discussed his experience with the first CGM systems. He was one of the first doctors to use the Premedos insulin pump, and the Biostator glucose control system. Over the years he has seen a significant decrease in the size of pumps and an increase in their tolerability.
  • He advocates use of the Guardian for CGM training. In the first month, he trains patients on the Guardian, combined with training in pump use. In the second month, he starts patients on pump basics. Then he gradually introduces new features. He discussed the CareLink software, which he characterized as a ‘helpful system’ in enabling the easier management of patient data.

UTILITY OF CONTINUOUS GLUCOSE MONITORING THROUGHOUT PREGNANCY

Lois Jovanovic, MD (Sansum Diabetes Research Institute, Santa Barbara, CA)

Dr. Jovanovic summarized her experience with the use of CGM in pregnancy. High glucose during pregnancy increases the risk of complications, including birth defects and macrosomia. While patients can often be adequately controlled without CGM, it requires a large number of fingersticks each day (she estimated 8-10) to ensure that glucose levels are within the target range. However, many patients who appear to be in good control may experience significant unreported hyperglycemia, which can be uncovered by the use of a CGM device. She expressed the opinion that these devices can be put to good use at all stages of pregnancy, and we heartily agree!

  • High glucose during pregnancy increases the risk of macrosomia. Dr. Jovanovic’s team found that macrosomic babies already have increased visceral fat compared to normal babies, which is known to be a risk factor for later cardiovascular disease and diabetes. The Hyperglycemia and Adverse Pregnancy Outcome (HAPO) study found a strong, association between maternal glucose levels and adverse pregnancy outcomes such as increased birth weight and increased cord-blood serum C-peptide levels. Importantly, there was no threshold at which risk began to increase; any elevation of glucose was associated with an increase in adverse events. This suggests that controlling high blood glucose in pregnant women may have substantial therapeutic value, even in cases of relatively mild hyperglycemia.
  • Over the course of a pregnancy, glucose control declined. Postprandial spikes are about 20 mg/dL higher after 38 weeks of gestation compared to 12 weeks. This indicates that glucose monitoring is particularly important in the time surrounding a meal. Postprandial blood sugar was found to predict the risk of macrosomia. The risk increased rapidly when postprandial spikes exceeded 120 mg/dL. To achieve this, a woman’s A1c has to be less than 5.0%, which Dr. Jovanovic says requires 8-10 glucose measurements/day. This is a very ambitious goal and we would guess very few women achieve it. Fetal risk increases 17 times as maternal glucose levels double.
  • Dr. Jovanovic thinks that CGM has utility throughout the entire pregnancy. In the first trimester, it is clear that the loss of a pregnancy is directly related to maternal glucose. The risk of congenital anomalies in women with type 1 diabetes goes as high as 9%, but is even higher in women with type 2 diabetes, going up to 13%. CGM use in pregnant women can uncover significant unseen hyperglycemia even in women that appear to be in good glucose control. In patients who need medication during pregnancy, CGM can help them maintain good control despite the effects of the medication, and it can also help to decrease blood glucose variability.

Questions and Answers

Q: Do you think that a pregnant woman with type 1 should be controlled to normoglycemia?

A: Actually, we don’t see increased risk of hypoglycemia in pregnancy—women can learn how to prevent lows using CGM. Certainly in type 1 diabetes, you can treat a falling blood sugar. We also make sure that our women don’t drive unless their sugar is above 90 mg/dL. If a woman lives alone, we need support systems to check on them in the middle of the night, but we instruct them to get up during the night regardless.

FLUORESCENCE: SHEDDING LIGHT ON GLUCOSE SENSING

John Pickup, DPhil (King’s College London, London, UK)

Dr. Pickup’s lecture provided another window into the developing world of non-invasive or minimally invasive glucose sensing technology. His group is developing fluorescent proteins that change their light-transmitting properties in a glucose-dependent manner to detect surrounding glucose levels. He mentioned a variety of different molecules that could be used in this manner, but says that his team has seen the most promise using a bacterial glucose-binding protein encapsulated in microcapsules. He envisions a system in which these capsules are embedded in the interstitial space and connected to the skin surface via a fiber-optic cable, through which glucose readings can be taken. We were disappointed to hear, in response to a question, that such a system is still five to ten years away from clinical use.

  • Dr. Pickup identified problems with current sensing technology noting in particular impaired responses, the need for frequent recalibration, and their invasive nature. Over the next ten years, he hopes that sensors can become more reliable (particularly for the closed loop) while also becoming non-invasive.
  • Fluorescence may provide an alternative technology. According to Dr. Pickup, it is very sensitive, non-invasive, free from electro-active interference, and is good for in vivo use. This approach would combine fluorescence intensity with lifetime. Fluorescence lifetime is measured using time-correlated single-photon counting, which measures the time from photon emission to detection. Applications may include “smart tattoo” sensors, fiber optic probes, or some intrinsic fluorescence in the body. Glucose would be bound to a receptor molecule that produces the photonic signal. Many different molecules could be used as receptors, including hexokinase and concanavalin A. In addition, glucose changes cellular NADPH concentrations, which could be monitored from outside the skin because NADPH is fluorescent while NADH is not. Bacterial glucose-binding protein may be another promising receptor candidate, and it can be encapsulated in micro-capsules.

Questions and Answers

Q: Where do you measure your glucose?

A: The intention is first to measure it in subcutaneous tissue using an implanted fiber-optic probe.

Q: Do you see a problem with light scattering?

A: Probably not because the photon lifetime is independent of light-scattering.

Q: When do you expect these results to come to clinical practice?

A: The next phase of the project is clinically testing a prototype fiber-optic probe, but it will take some years. We’re in the five to ten year framework, realistically.

Technology – Health Policy and Benefit

DO WE NEED CENTERS OF REFERENCE TO IMPLEMENT NEW TECHNOLOGIES IN PAEDIATRIC DIABETES?

Thomas Danne, MD (KinderKrankenhaus Auf Der Bult, Hannover, Germany)

As part of his ISPAD legacy, Dr. Danne has set out to create an international center of reference. To this end he established the SWEET project, a European pediatric center focused on improving control for children with diabetes. SWEET has received government and corporate funding and has issued a status report highlighting the broad variation in policy and care across the EU. Its next project is expected to be a central repository of information from all countries that can be used for benchmarking. The hope is to also use this facility for telemedicine and patient data.

  • There seems to be a large variation in funding, behavior and outcomes across nations and clinical centers. Some examples include:
    • The Euro Consumer Diabetes index is a measure of ‘diabetes bang for the buck’ – there is an enormous variation across countries, with Lithuania able to make the best use of its healthcare money in tackling diabetes.
    • The use of insulin analogs in major EU countries varies from 41% (Germany) to 91% (Sweden). Where is evidence-based medicine in this variation?​​
    • In studies, we have seen a large variation in A1c across European centers measured by the same lab.
  • Every innovation has a learning curve – initial results may not be successful, but that doesn’t mean we shouldn’t persevere with them. Sharing experience is the best way to move quickly down the learning curve. For example, just because we had some problems with kids in the JDRF CGM trial doesn’t mean that children can’t be successful with CGM – it’s just about learning how to do it well.
  • Dr. Danne has initiated a European Center of Reference, called the SWEET project (www.sweet-project.eu). This is partially funded with European grants and corporate sponsorship. The goal is to obtain better control in pediatric and adolescent diabetes.
  • So far, the SWEET project has issued a Status Report, which conveys some disappointing conclusions. Only five of 27 EU states have data from 15-18 year olds; only 13 out of 25 have a national pediatric diabetes register; only 13 have a national diabetes plan; and only seven countries have special provisions addressing children with diabetes. Very few EU member states have regulation or specific measures for children in school.
  • The SWEET project intends to address three major issues: 1) structure quality (e.g. what do you need to have a pediatric diabetes center), 2) process quality (how do you deliver the best care), and 3) outcome quality.
  • The next step is to establish a SWEET information technology network to enable secure, private sharing of data on a central server. This will include patient data. The longer-term plan is to implement telemedicine, and also interact with the primary care provider.

EVIDENCE OF IMPROVED HEALTH OUTCOMES FROM CONTINUOUS HOME GLUCOSE MONITORING

Satish Garg, MD (University of Colorado at Denver, CO, USA)

Dr. Garg said we have reached the point with CGM technology that we need to reach out to the majority of patients and encourage them to use it widely, in order that they can realize the medical benefits. CGM reduces glycemic variability, hypoglycemia, and A1c and works for both pumpers and those on MDI. Notably, it doesn’t matter whether we start at higher or lower A1cs – CGM still reduces glycemic variability. If patients start with A1c >9.0% they can double the time spent in normoglycemia. CGM also allows patients to improve their A1c without increasing hypoglycemia. The future hope for CGM is a 12- month implantable or non-invasive sensor. When SMBG meters first came out, it was not clear that there were going to be cost effective, but today there is no doubt. The same argument can be made for CGM. However, the devices do need to be cheaper. There are good arguments about reducing the cost of hypoglycemia (emergency room visits, car accidents), but it’s probably early to do the cost analysis.

PROMOTION OF HEALTHY LIFESTYLE FOR CHILDREN

Francine Kaufman, MD (Children’s Hospital, Los Angeles, CA)

Over the last six years, there have been some positive changes in the struggle against pediatric obesity (and its complications) in the US. Obesity rates appear to be flattening around 16%, although it’s too early to say for sure. In a study cited by Dr. Kaufman, 40% of eighth graders have impaired fasting glucose and 25% have impaired glucose tolerance, and often reach the criteria for pre-diabetes. There are about 3,700 children a year diagnosed with type 2 diabetes of which 26% have hypertension and 59% have dyslipidemia. It’s to be expected that these children might experience cardiovascular events as young as their 30s and 40s.

The HEALTHY trial was established to determine if environmental changes could reduce risk factors for type 2 diabetes in schools by attempting to focus on physical activity, nutrition, and behavior modification. On a broader note, California has been making progress on health related legislation such as banning junk food and soda in schools, requiring nutrition information in restaurants, and banning trans fats. There are many other things that Dr. Kaufman believes we still need to do, such as regulating marketing to children, improving the school and work environment, introducing zoning to promote healthy food outlets, and limiting access to junk food.

  • In 1999-2000, obesity prevalence in two- to nine-year-old children reached 13.9%. By 2006 it was 16.3%, but it appears to have been roughly constant since 2003. Has it really reached a maximum? It’s certainly our hope that this is the case and with more work we can bring it back down.
  • Over 90% of all kids with type 2 diabetes are obese. Dr. Kaufman estimated that 30% of type 1 kids are obese, which is higher than the general population. Obesity also leads to other medical conditions that some pediatricians never expected they might see like gall bladder and gynecological conditions.
  • By the eighth grade, 40% of children have impaired fasting glucose and 25% have impaired glucose tolerance. 3,700 children per year are diagnosed with type 2 diabetes in the United States. 26% of these have hypertension and 59% have dyslipidemia. These children can be expected to have cardiovascular events when they are as young as 30 or 40 years old.
  • The HEALTHY trial was established to determine if it is possible to reduce the risk factors for type 2 diabetes in a school setting. It followed 6,400 sixth graders for two years in 42 schools. The interventions include increased physical activity (targeting 150 minutes of exercise every two weeks), nutrition (education, shutting down the soda and junk food vending machines), and a behavioral component (education, discussions with peer groups, goal setting and monitoring). The trial is almost complete and se look forward to hearing results of this important study that could have profound implications for future generations.
  • The environment in the United States is obesogenic. Unfortunately, low income individuals have a difficult time living healthily in such an environment. They do not have access to healthy food limiting their diets to junk food, and they do not have safe parks for exercise. 25% of all products use high fructose corn syrup that adds to calories and may even have specific metabolic issues. Most states are not taking action.
  • In 2003, sodas were banned in Los Angeles public schools. The effort had taken six years. In 2005, sales of soda and junk food were banned in schools across the whole state. Dr. Kaufman is working on requiring nutrition information in restaurants with more than seven sites and banning trans fats.
  • There are many other things that Dr. Kaufman believes we still need to do, such as regulating marketing to children, improving the school and work environment, introducing zoning to promote healthy food outlets, and limiting access to junk food.

The Pathway of Diabetes Care: Next Steps into the Future

MAKING THE MOST OF CGM TECHNOLOGY: UTILIZING CLINICAL EVIDENCE

Tadej Battelino, MD (University Medical Center, Ljubljana, Slovenia)

Dr. Battelino is a strong proponent of CGM for all patients. He believes that the technology is now mature enough for widespread use. He has collected solid data that CGM is safe in all age groups. He cited evidence from at least three randomized controlled trials that CGM improves metabolic control and reduces hypoglycemia. He asserted that CGM also facilitates education and improves quality of life. The hope is that it will also reduce complications in the long term, and we await trials to validate this hypothesis. In the meantime, there are many CGM trials in process – over ten with Medtronic products alone.

  • Years ago, with the introduction of home glucose monitoring, there were many papers concluding that SMBG was not valuable or cost-effective for patients taking insulin, and we are already seeing similar papers regarding CGM.
  • There is currently a strong debate about glycemic variability and whether or not it leads to complications. Prof. Battelino is repeating Dr. Hans de Vries’ oxidative stress experiment with type 1 patients (discussed on Day 1), expecting to see a different (non-negative) result, which would be more in line with the famous Monnier result in type 2 patients (which established a relationship between oxidative stress and glycemic variability).
  • Modern CGM accuracy is perfectly acceptable, despite what some people say. We see>95% of sample points in the Clarke error grid A region when CGM data are compared to a blood glucose reference or even comparing between CGM sensors.
  • There is a physiological delay between interstitial fluid and blood glucose – but this is not an error, it’s a feature! The delay is estimated as 15-20 minutes, which seems to vary by patient. We should look at this as a feature of CGM rather than an error. If the fingerstick glucose is a lot higher than CGM, then that’s evidence that glucose is rising rapidly. Conversely, if the fingerstick result is lower then we can anticipate and treat possible hypoglycemia. In our view, this is an interesting and novel argument that deserves validation in a clinical trial to ensure that intra- and inter-patient variability are not insurmountable confounders.
  • The JDRF CGM trial was very successful in demonstrating that better glycemic control is possible from consistent use of CGM. The NEJM paper didn’t make mention of the fact that all age groups improved A1c if they used CGM at least six days a week, which perhaps gave the impression that CGM was not effective in children and adolescents, an unfortunate editorial decision. However, Dr. Battelino doesn’t like the concept of patient selection for CGM, preferring instead wants to figure out a way to make CGM work for everybody – an ambitious goal.
  • CGM allows pediatricians to tighten control in kids without worrying about risks of hypoglycemia. It’s even possible to use CGM sensors with pre-term kids of low birth weight.
  • Prof. Battelino gave some case studies of combined CGM and pump data (using Medtronic CareLink, pumps and sensors). In all cases, we see the value of the bolus wizard, and the clarity that’s provided by CGM in understanding the causes of hyperglycemia or hypoglycemia.
  • There are many more CGM trials underway, including STAR3, and at least ten trials using Medtronic products alone.

CLOSING THE GAP TOWARDS CLOSED LOOP—THE UK EXPERIENCE

Stephanie Amiel, MD (King’s College London, London, UK)

Dr. Amiel discussed the new Medtronic X54 insulin pump with a low-glucose suspend feature. According to her work, initial reports from patients using the device were generally positive. Patients liked the set-up, equipment, and literature that comes with the pump, and felt more secure knowing that insulin delivery will shut off if their blood glucose goes into the hypoglycemic range. However, patients sometimes found the alarms irritating (we see this with all CGM devices, it seems!). We see this feature as a big step towards a safe closed loop system — unfortunately, it is not currently available in the US, and we believe it will be some time before it is since this is a PMA submission and the trial is being designed at present.

  • Medtronic’s contribution to the closed loop comes in the form of the X54, the first low-glucose suspend pump available. The pump will suspend insulin for two hours after detecting hypoglycemia (that isn’t responded to by a patient), and then will re-suspend if glucose remains low four hours after resuming. The system isn’t a hypoglycemia prevention device, but a protection from severe hypoglycemia. This is particularly important in hypoglycemia unawareness, which affects between 25% and 40% of type 1 diabetes and increases the risk of severe hypoglycemia by a factor of three. Nocturnal hypoglycemia is likewise a common problem that could be alleviated by the X54, and it is particularly dangerous because it is not associated with a counter-regulatory stress response. Alarms may not wake sleeping patients — in fact, he reported that 71% of patients will sleep through their alarm. In addition, nocturnal glycemia may impair the response to subsequent hypoglycemia the next day. The suspension of insulin delivery during nocturnal hypoglycemia could help to prevent severe nocturnal hypoglycemia and subsequent hypoglycemia unawareness.
  • The potential difficulties of interrupting insulin infusion include (rebound) hyperglycemia or DKA. In designing the pump, Medtronic considered studies of the rate of glucose increase with insulin shut-off. These studies suggest that a five-hour or three-hour shut- off may be too long to maintain control, which is why Medtronic chose the two-hour shut-off period for the X54.
  • Patients given self-evaluations of the X54 by Dr. Amiel ‘really liked’ the pump overall. There is some data looking at live operation of the low-glucose suspend, showing that the shut-off really does help to reduce severe hypoglycemia without hyperglycemia, compared to alarm-only CGM systems. Patients liked the literature, the equipment, and the set-up, and felt more secure with the low-glucose suspend feature, but many found the alarms irritating. There doesn’t seem to be much of a hyperglycemic overshoot after suspension, and it usually occurs after eating. Patients may increase the duration of prediction for hypoglycemia alarms to better treat impending hypoglycemia by reducing insulin delivery and not by eating (ed. note – this helps reduce hyperglycemic rebounds). Patients can adjust the trigger for insulin suspension to make it suspend insulin earlier. Dr. Amiel thinks that this approach could be used to prevent hypoglycemia altogether. Indeed, patients sometimes suspend insulin delivery on their own and have success in preventing hypoglycemia.

Questions and Answers

Q: In patients who have residual beta cell function, how does this affect the shut-off of the pump?

A: You’re really asking the wrong person. We do not routinely use pumps early in the course of type 1 diabetes. Some residual insulin secretion is protective against hypoglycemia, so I don’t see it being a serious problem. Those people don’t need that feature as much as others.

Q: Is lag time taken into account?

A: No, not in this device. As I said, this feature is designed to minimize exposure to hypoglycemia, not to prevent it. That is the next step, but this is not designed for that use.

Q: Do patients want the shut-off without the alarm?

A: Spouses objected to being woken up at night. It depends on how you set them, and some patients turned them off. However, almost as much positivity was expressed about the alarms as negativity.

Q: What’s your opinion about separate glucagon infusion?

A: I guess if it could be done easily it would be fine. I have problems envisaging it as a commercial development because it requires a sophisticated delivery device.

Q: How often did the system shut off when the patient was not hypoglycemia?

A: In our small study it did not do so, as far as we are aware.

REALIZING THE DREAM OF CLOSED LOOP: MEDTRONICS'S NEXT GENERATION DEVICES

John Mastrototaro, PhD (Medtronic Diabetes, Los Angeles, CA)

It’s always exciting to see Medtronic’s “futures” presentation, which shows up regularly at technology conferences. Some of the initiatives that Medtronic discussed in the past have now started to turn up in products, such as the X54 pump with the low-glucose suspend feature.

In the presentation, Dr. Mastrototaro discussed improvements to the current CGM products such as increasing sensor life from three days to six days. He also discussed wireless connectivity from sensors to mobile phones, cars, and home remote monitoring systems, although he didn’t give any sense of when we would see these products. Finally, he discussed Medtronic’s commitment to the artificial pancreas and recent trial data. The success that has been achieved with current closed loop algorithms suggests that a robust ambulatory product, once approved, could lead to very good A1c control.

In a conversation afterward, Dr. Mastrototaro indicated that Medtronic had mapped out a plan to obtain regulatory approval for an artificial pancreas system step by step, and the company was working with the FDA to make it happen. Despite a genuine commitment to move as fast as possible from the FDA, we imagine that it will still take a considerable time; Europe is already ahead with the low-glucose suspend feature.

  • Improvements to Medtronic’s CGM products that are under development include:
    • Increasing sensor life from three days to six days (we saw some data with the new six day sensor, implying it is close to market).
    • Reducing startup time from two hours to 30 minutes
    • Reducing fingerstick requirements and pursuing a fingerstick replacement claim (we assume this means using CGM for insulin dosing)
    • Reducing sensor size on the body and “combining the sensor and insulin delivery site into one place”
    • Improving sensor accuracy, particularly in the hypoglycemia range
  • Medtronic is focusing on bringing wireless technologies to market, such as:
    • A smart phone monitor that could display glucose readings on a cell phone display and could automatically upload data into CareLink for retrospective analysis. The phone could even warn others if the person doesn’t respond to a low alarm. This will be important for safety in ambulatory closed loop studies.
    • Home remote monitoring of glucose – Dr. Mastrototaro described this as a ‘fancy baby monitor’. A display/alarm unit can sit by a parent’s bedside, monitoring a child with diabetes, and warn only if necessary – saving a lot of lost sleep. Judging from the quality and detail of the concept, this product seemed a little closer to market.
    • Connectivity in the car via Bluetooth. The product concept was a glucose reading on the navigation system display.
  • The X54 pump with low-glucose Suspend (LGS) is the first step towards the artificial pancreas. LGS doesn’t prevent hypoglycemia; rather, if tries to avoid a severe hypoglycemia event. Test data showed that the algorithm detected hypoglycemia 80% of the time, which can be increased to 90% using a 15-minute prediction algorithm.
  • Medtronic’s artificial pancreas approach uses a PID algorithm with one-minute (rather than five minute) sensor data. Although most researchers have switched to superior MPC algorithms, Medtronic says that its algorithm now works well enough within the errors and time delays of a real life system, and it’s better to spend their time overcoming the safety and regulatory obstacles.
  • There have been three trials of Medtronic closed loop technology to date. With fully closed loop, we saw post-meal hyperglycemia excursions, but overnight control was excellent. Accordingly, Medtronic allowed meal announcement and pre-meal boluses, which was very successful at reducing post-prandial hyperglycemia. They’ve also added insulin on board calculations.
  • Medtronic is investigating the role of glucagon and Symlin (Amylin’s pramlintide) in the artificial pancreas, alongside insulin. Symlin helps control post meal spikes. With glucagon in the equation, it’s possible for the algorithm to be more aggressive with insulin, since there is a ‘brake’. Dr. Mastrototaro used the metaphor of a car with no brakes: you can’t be too aggressive on the gas pedal because if you want to stop you have to coast. However, glucagon adds to the complexity in lots of ways including the physical design of the pump, and the interaction of insulin and glucagon dosing algorithms.
  • It was really exciting to see that even current closed loop performance offers stunning results – in the Yale study, a well-controlled patient with an A1c of 7.1% increased ‘time in the zone’ from 58% to 82%, with a calculated new A1c of only 5.8%. Imagine how closed loop therapy could transform glucose control.
  • The areas of focus for bringing closed loop to market are:
    • Improving sensor performance and detecting failing sensors
    • Failsafe mechanisms
    • Redundant sensors and error checking
    • Insulin delivery accuracy – tubing and site problemsAlgorithm concerns such as sensor delays (interstitial fluid to blood), insulin pharmacodynamics, and absence of counter regulatory mechanism

Questions and Answers

Q: (Dr. Thomas Danne) Hypoglycemia can cause cardiac arrythmia – where is the safety valve for the heart?

A: We are starting to measure hypoglycemia and cardiac function in halter studies. We are also starting to look at activity monitoring. These are just extra pieces of information to help regulate insulin delivery.

Q: (Dr. Arleen Pinkos, FDA) Will we have a “one size fits all” algorithm, or will it be personalized somehow to the individual?

A: In low-glucose suspend – we may need shorter or longer time of suspend depending on the individual. The data suspend for two hours was an average cost/benefit tradeoff. For fully closed loop studies – we run open loop for a while and measure total insulin needs to set the starting parameters for the algorithm, such as sensitivity of insulin delivery. But we’ve shown that the current algorithm is robust to +/- 50% error in insulin sensitivity and won’t cause hypoglycemia or hyperglycemia with those errors. We would also use CareLink to upload artificial pancreas data and have an algorithm to assess changes that need to be made to the parameters on an ongoing basis.

The Loop Club

The Loop Club presentations were all collaborations across many groups. We recognize the efforts of the JDRF Artificial Pancreas Project in aligning the research goals of these world-class centers. It seems that the pace of progress is rapid, as evidenced by all the human trial data we are seeing this year.

THE STANFORD-UCSB-RPI-DENVER PROGRESS TOWARDS A CLOSED LOOP

Bruce Buckingham, MD and Eyal Dassau, PhD (Stanford Medical Center, Palo Alto, CA; Sansum Diabetes Center, Santa Barbara, CA)

This consortium consisted of Stanford, Sansum, Barbara Davis, and Rensselaer with the sponsorship of the JDRF.

The first topic, presented by Dr. Buckingham, was the automatic detection of insulin set failures. This is an extremely important aspect of taking the closed loop into the real world, and we saw Medtronic make mention of it in their development pipeline as well. It’s hard to distinguish between a set failure and an unannounced meal with a missed bolus, but if we add in meal announcement, it becomes quite possible, Dr. Buckingham said.

Next, Dr. Buckingham outlined results of a closed loop algorithm that was tested using the FDA cleared in silico simulator. The algorithm performed well overnight, but ran a little high, casting some doubt over the realism of some of the simulated individuals.

Dr. Dassau presented some important infrastructure for closed loop projects – a software platform called APS, a hardware platform that interfaces to devices and virtual patients, and an ‘algorithm on a chip’ which worked well in initial testing.

  • From a model standpoint, the characteristics of an insulin set failure (SF) are apparent insulin resistance, a consistently rising glucose slope, and a low failure occurrence rate. The group created an algorithm to spot these failures. The algorithm can spot the failures and warn the user much earlier than users currently react.
  • SFs can be easily confused with a missed meal bolus – so to distinguish between the two, we can add meal prompting. The SF detection algorithm has a 75% false alarm rate with no meal announcement, but only 35% false alarms when meals are recorded.
  • Dr. Buckingham also presented results of a Model Based Control algorithm, incorporating insulin on board. The model was tested using the FDA cleared in silico simulator, using both adult and adolescent simulated subjects, during night time and meal times.
  • The algorithm avoided hypoglycemia consistently, but overall results were a little high, particularly with adolescents. It also cast doubt over a few of the simulated patients, who seemed to be extremely insulin resistant.
  • Dr. Dassau presented APS (artificial pancreas software) – a PC software platform to allow easy interaction with and setup of closed loop control algorithms. The software has an easy to use interface, and allows differing sensors, pumps and algorithms to be used. Currently, APS supports Navigator, DexCom and OmniPod and they are working on Roche, Deltec, and Animas insulin pumps.
  • Furthermore, Dr. Dassau showed a complementary hardware platform for closed loop experiments, consisting of interfaces to real sensors and pumps, but a virtual patient for rapid testing.
  • Another goal is to embed an MPC algorithm on a chip. Optimizing in real time could be hard, so instead the solution space is solved offline and transferred to the chip as a lookup table. Seven trials were run (in collaboration with Moshe Phillips’ group) and the system appeared to work well – handling unannounced meals.

SUBCUTANEOUS CLOSED LOOP STUDIES USING THE S.C. MD-LOGIC SYSTEM

Moshe Phillip, MD (Schneider Children’s Medical Center, Israel)

Dr. Phillip created a Diabetes Technology Center within his Institute to work on the closed loop. He currently has two closed loop approaches. One is the JDRF collaboration previously presented. The second is the “MD-Logic” approach. There was not a lot of visibility into the exact methods behind the MD-Logic algorithm, but it does seem to be guided by clinician heuristics and is tailored to an individual patient over a number of visits to the clinic. The results showed good closed loop control without meal announcement.

  • The MD-Logic approach is to tailor a patient specific algorithm based on a medical way of thinking and learning from experience. Clinicians use “intelligent trial and error” using a weighted consideration of known factors. Engineers were requested to create an algorithm that has these characteristics. We know that the MD-Logic algorithm also contains a prediction component to allow for delays (e.g. insulin action, CGM sensors), and it is tuned to prevent hypoglycemia. However, based on what was presented, it was a little difficult for us to understand exactly what type of algorithm was used and its strengths and weaknesses.
  • Patients visit the clinic with three days of prior data, which is used to set initial parameters for the algorithm. They then start with a fasting session, in which their insulin dosing and glucose responses are recorded. The system runs in closed loop mode for a while and results are collected. The algorithm is then re-adjusted to their individual parameters, and apparently takes two to three sessions to produce good results. Note that the algorithm appears precisely tailored to each patient, but it’s not clear whether it can robustly handle intra-patient(i.e. day to day) variations or disturbances like exercise.
  • The results shown were good. Overnight control was 110 mg/dl, and peak excursions after meals were around 200 mg/dl, without any meal announcement. Next steps are to improve post- prandial response by delivering more insulin closer to the meal, although this might not be possible with just insulin alone.

SUBCUTANEOUS MODEL PREDICTIVE CONTROL CLINICAL TRIAL POWERED BY IN SILICO STUDIES

Claudio Cobelli, MD (University of Padova, Padova, Italy)

This consortium consists of members from the University of Virginia, University of Padova (modeling), the University of Pavia (control engineering) and the University of Montpellier, with support from the JDRF Artificial Pancreas project. Their efforts in driving forward closed loop algorithms are significant.

The in silico simulation that was cleared by the FDA last year is a very efficient and cost effective means for development and testing of new closed loop algorithms. An MPC algorithm tested in this way has received FDA approval for human trials, which are going well. These clinical and in silico trials will also contribute to the refinement of the simulation environment.

  • In January 2008, the FDA cleared an in silico simulator of type 1 diabetes patients, which could be used to test closed loop algorithms instead of performing animal studies. To create this simulation, triple tracer data from 204 patients was used to generate models of the many ‘fluxes’ (insulin secretion, insulin absorption, glucose production etc). The data was adapted to model type 1 patients and the inherent patient-to-patient variations. 300 simulated patients were created (adults, adolescents and children) and sensor and pump models (Navigator, Guardian, DexCom, OmniPod and Cozmo) were added. This is a ‘maximal model’ –i.e. very detailed and complex.
  • Using the in silico simulation as a basis, the team created an MPC (model predictive control) algorithm for an artificial pancreas. Conveniently, the model needs only one parameter – “Q” – to account for individual variations. Q is characterized as the “aggressiveness of control” and is a function of weight, basal insulin needs and insulin:carbohydrate ratio. An additional layer of hypoglycemia prevention supervision was added to the model – known as ‘Brakes’ – which predicts incipient hypoglycemia, as presented on Day 1 of this meeting.
  • In April 2008, after only three months, the FDA gave regulatory approval for human clinical trials of the MPC closed loop based entirely on the in silico testing. It’s great to see such rapid movement by both the researchers and regulators.
  • The clinical trials are to cover 24 individuals, and 18 have been completed to date. There are two overnight hospital visits. The first is open loop and four weeks later there is a closed loop visit. The closed loop portion starts after dinner (around 9:30pm) until the following lunch time. The trial uses the FreeStyle Navigator CGM and the OmniPod insulin pump.
  • Preliminary results show excellent control overnight. We see some post-prandial hyperglycemia after breakfast, where control could be improved. However patients spend 77% of the time in the normoglycemia zone (70-140 mg/dl) in closed loop, compared to 62% in open loop and hypoglycemia (<70 mg/dl) decreased from 20 to 4 episodes (although in a small sample) when moving to closed loop. It would be interesting to see the reported differences for average glucose and the ‘estimated A1c’.

JDRF Hellas Project: Hypoglycemia and Exercise

INTRODUCTION: HYPOGLYCEMIA DURING EXERCISE

Asimina Mitrakou, MD (National University, Athens, Greece)

Dr. Mitrakou gave a brief overview of the rationale behind the JDRF Hellas project. The project was designed to investigate the utility of CGM devices in controlling hypoglycemia and hyperglycemia in young adults during periods of moderate-to-intense exercise. Exercise presents a particular challenge for patients who are trying to stay in good glycemic control since glucose levels tend to have a variable (hyperglycemia) and delayed (hypoglycemia) response to exercise. We believe that CGM can be a valuable tool for managing this problem.

  • Research has shown that physical activity can optimize health and fitness benefits, including a reduced risk of cardiovascular disease and weight control, as well as improved insulin sensitivity and a reduction in daily insulin needs.
  • Glycemic control during and after exercise needs to be better than in periods without exercise. Intense exercise can cause either hyperglycemia or hypoglycemia. Typically, hyperglycemia occurs during exercise and then a period of hypoglycemia follows later. It is likely that the treatment of hypoglycemia due to exercise may counterbalance the benefit of such exercise in some children.
  • Children with diabetes should be encouraged to participate in any sport, but should use glucose monitoring prior to, during, and after the activity. While SMBG has traditionally used for this purpose, the convenience of CGM was being tested in the Hellas project. The project was to help people learn about their responses to exercise and to help them control their blood sugar during and after exercise using CGM or SMBG technology.

THE ADVANTAGES OF CONTINUOUS GLUCOSE MONITORING SYSTEM ON MOUNTAIN CLIMBING IN YOUNG ADULTS WITH TYPE 1 DIABETES

Andriani Vazaiou, MD (Athens, Greece)

Dr. Vazaiou presented the results of the JDRF Hellas pilot study, which examined the effectiveness of CGM in decreasing hypo- and hyperglycemia seen in young adults and associated with exercise. Forty- three patients with type 1 diabetes participated in exercise sessions of four to six hours of intensive mountain climbing. Twenty-eight of the participants elected to use CGM and 13 used SMBG. The results showed that CGM use was effective in decreasing the overall amount of time spent outside the target glycemic zone, and it was also effective in detecting otherwise silent hypoglycemia during the night following the exercise. This indicates that CGM may have an important use in this area, which is encouraging because glucose control during exercise is one of the biggest challenges in diabetes management.

  • Dr. Vazaiou presented the results of the JDRF Hellas pilot study investigating the use of CGM systems in young adults during exercise. The trial enrolled 43 patients with type 1 diabetes with an average age of 23 years, A1c of 7.5%, and a BMI of 24 mg/kg2. 25 patients in the study were on MDI, two were taking two injections per day, one was on three injections/day, and 13 were using a pump. The average disease duration was 14 years. For the study, 28 patients elected to use CGM and 13 used SMBG. Participants went on extended rock climbing trips of four to six hours, and their glycemic control was monitored before, during, and after exercise.
  • Prior to exercise, subjects’ basal rate was reduced by 50% if they were on a pump or 30% if they were on MDI. Blood glucose readings were taken every time they were needed for a correction prior to exercise. If children were under 150 mg/dL before exercise, 15 g of carbs were provided. Patients over 150 mg/dL were treated with insulin.
  • The results showed an overall benefit of CGM in reducing the time spent out of euglycemia during exercise. Before exercise, two out of 41 subjects had hypoglycemia, and 18 had hyperglycemia. Twenty out of 41 had hypoglycemia during the exercise, and there was no difference in the number of patients experiencing hyperglycemia between the control group and the CGM group. In those patients who experienced hypoglycemia, those in the CGM group had fewer episodes than those in the control group, and the episodes they did have were of shorter duration. The overall amount of time spent in hypoglycemia was reduced by about 20% in the CGM group. There was no difference in the number of subjects experiencing hyperglycemia between the control group and the CGM group, and there was no difference in the number or duration of hyperglycemic events between the two groups. Overall, fewer patients with CGM had hypoglycemic or hyperglycemic events, and those in the CGM group spent on average about 60 fewer minutes above or below the normal limit.
  • There was no difference in the rate of hypoglycemia or hyperglycemia after exercise between the two groups, but CGM may help to diagnose otherwise silent hypoglycemia. 33% of subjects developed diagnosed hypoglycemia after exercise. Importantly, an additional ~25% had undiagnosed hypoglycemia that was only detected on the CGM reading, pointing to the value of CGM in detecting nocturnal hypoglycemia. Nocturnal hypoglycemia was most common in people who had experienced an episode of hypoglycemia in the previous 12 hours. 87% of sensor values were within the A+B regions of the Clarke error grid, and Dr. Vazaiou mentioned that this result includes two patients with failure of the sensor, so their calibration was not as good and the reported value likely understates the true clinical accuracy of CGM.

Questions and Answers:

Q: Did the CGM alarms go off?

A: Yes, the alarms going off prevented episodes, but some children did not respond to their alarms. The majority responded well.

Q: Did you have trouble with the sensors falling off?

A: No, we have had a lot of experience putting these on well.

CONCLUSIONS: THE USE OF CONTINOUS GLUCOSE MONITORING

Bruce Buckingham, MD (Stanford University, Palo Alto, CA)

Dr. Buckingham gave a very brief take on the results of the Hellas pilot study and how it might be translated into a clinical guideline. He pointed out that there is no ‘safe’ glucose value at night after exercise, but a possible target is about 130-140 mg/dL. Patients show significant variability in their nighttime glucose variation after afternoon exercise, so it’s very difficult to come up with a safe guideline that holds true in all patients. The drop in glucose after activity is about 60-70 mg/dL according to the DirecNet study, but there are no data about glucose drops for specific exercise activities. We would speculate that a low-glucose insulin suspend feature like that on the new Medtronic X54 insulin pump might help to solve this problem.

Roche Sponsored Session: From Information Management to Insulin Management

INFORMATION MANAGEMENT—THE VALUE IN DAILY LIFE OF DIABETES PROFESSIONALS AND THEIR PATIENTS

Ewa Pankowska, MD (Medical Academy, Warsaw, Poland)

Dr. Pankowska discussed the role of information management in diabetes care. She believes that access to better patient information can change the standards of medical practice. First, she focused on electronic data collection, including insulin administration data and glucose measurement data. Secondly, she discussed the use of data to manage large groups of patients and get easy access to their treatment results. We are big fans of patient management, and hugely excited that it is getting so much attention at this conference. However, even if the data are made available, a secondary problem will be getting physicians to use it. Overall we would stress the importance of an easy-to-use system that can interface with existing medical records systems—a system that cannot link in with others will not be as significant of a step forward.

  • Dr. Pankowska believes that access to better patient information can change the standards of medical practice. First, she focused on electronic data collection, including insulin administration data and glucose measurement data. She thinks that collation of diverse patient data enables the rapid identification of management problems. We certainly agree with that, and have heard similar comments from many speakers at this meeting—most notably Dr. Irl Hirsch. She stressed particularly how data enables fine-tuning of insulin therapy and the split between basal and bolus therapy. Some patients are managed best with a 50/50 basal/bolus split, but others do better with a greater percentage of basal or bolus. Over time after diagnosis, Dr. Pankowska has observed that the proportion of basal insulin increases. She also mentioned the importance of advanced bolus modes for maintaining glucose control.
  • Secondly, she discussed the use of data to manage large groups of patients and get easy access to their treatment results. Her group, like others, has found that it is more difficult to reach treatment goals in adolescents compared with children. The use of data management allows her to monitor other diseases that may impact patients’ diabetes or glucose control.
  • The next step, in her opinion, is taking patient data online in a way that is helpful for patients and health care professionals alike.

DAILY LIVES OF DIABETES PATIENTS—A NEED FOR SPECIAL REQUIREMENTS IN SPECIAL SITUATIONS

Thomas Danne, MD (KinderKrankenhaus Auf Der Bult, Hannover, Germany)

One of our favorite speakers, Dr. Thomas Danne, discussed the role for data management in overcoming four major problems with insulin therapy: finding the right basal dose, controlling postprandial glucose, making best use of pump therapy, and avoiding adverse events. We agree with his conclusion that better management tools could help patients to meet their goals across the board— however, we would caution that few if any systems available today meet the standard necessary for such wide-ranging success.

  • Dr. Danne discussed the results from the JDRF study, pointing out that CGM is ‘not for everyone’, referring to the lack of a clear effect in children and adolescents. However, he still thinks that new technologies can help to address problems with insulin therapy.
  • Problems in insulin therapy include: finding the right basal dose, controlling postprandial glucose rise, making best use of pump therapy and avoiding adverse effects. Dr. Danne believes that data management can play a role in overcoming each of these.
  • He pointed out the importance of the circadian variation of basal insulin requirements, noting that this variation was age-dependent. In a study in 2004, certain German diabetes centers had higher A1cs, and this was correlated with a higher use of basal insulin compared to prandial insulin in those centers. He pointed out the importance of separating the basal and prandial component, and said that basal insulin should be 30-40% of total daily dose in children. With data management software, this ratio is very easy to see for any given patient.
  • Estimating carbohydrate count is important for controlling postprandial glucose. Carbohydrate to insulin ratios vary throughout the day, becoming lower in the evening. In addition, fat and protein affect glucose levels differently, and this needs to be taken into account. The variations from different foods may be compensated for using the more advanced bolus features on modern pumps. Data management allows doctors to assess the success of interventions, and data management can help to find problem areas for attention.
  • The crucial point in reducing pediatric A1c is the number of boluses given per day. According to a 2008 study by Dr. Danne, patients with more than 12 boluses per day had an average of 7.3% while those with less than two had an average of 10.4%. Statistical analyses can help focus attention on this factor.
  • A recent paper in 2009 Diabetologia suggests that ‘dead in bed’ syndrome may be due to cardiac arrhythmia. Dr. Danne thinks that data management could help doctors monitor other medical parameters besides glucose, potentially catching this and other dangerous disease interactions.

NEW SOLUTIONS IN PUMP THERAPY—WAYS TO IMPROVE INSULIN MANAGEMENT IN DIABETES PATIENTS

This session was a very interesting discussion of the role of technology in patient’s lives, given by two type 1 diabetes patients. Dr. Stuart Bootle is a practicing general practitioner in the UK and has had diabetes for more than 20 years. He talked about his experience with pump therapy and how it has revolutionized his diabetes care. He also briefly mentioned the Accu-Chek Combo, which includes a data- management system that he says has again changed how he manages his therapy. During their 4Q08 earnings call, Roche indicated that the Combo would be launched in Europe in 1Q09 and will be available in the US later in 2009.

The second speaker was Michal Jelinski, a three-time world champion rower and Olympic gold medalist, who has had diabetes for the past five years. Like Dr. Bootle, he described how switching to a pump in just the past year has improved his treatment and fine-tuned his performance in competition.

This was a very interesting presentation to get a feel for insulin devices ‘on-the-ground’, and the audience was really getting into the spirit. It was followed by a very interesting Q+A session, which talked about glucose variability, insulin therapy in children, and other topics.

Stuart Bootle, MD (Derbyshire, UK)

  • Dr. Bootle pointed out the importance of self-management and cooperation between patients and doctors in achieving management goals. He thinks that technology can be a catalyst for improving self-management. He focused on the difficulties people with diabetes face in coping with daily life while also maintaining good glucose control. Even with pump therapy, Dr. Bootle has had problems with glycemic variability. In front of the audience, he reached down his pants to show his insulin pump, as a demonstration of how awkward it is to use even the best modern technology.
  • The Accu-Chek Combo has revolutionized how he uses insulin pump therapy. He says that he has moved from pump therapy to an ‘insulin management system’. He thinks that the Combo helps him catalog and access important information. He can label test results as pre- or post-prandial, easily change basal and bolus delivery, calculate his on-board insulin, and take into account the food he had been eating and the activity he had been involved with.
  • He emphasized that healthcare providers should give patients treatment options and let them choose the treatment that is best for them.

Michal Jelinski (Poland)

  • Jelinski. a three time world champion rower, discussed his experience with diabetes. He has had diabetes since 2003, and has been on pump therapy since 2008. He was in remission for several years, but then had difficulty during training for the 2008 Olympic games, so decided to switch to a pump, with which he now trains successfully. He has reduced his insulin dosage by about two-thirds after switching to the pump and takes six to eight blood glucose readings/day.
  • As a sports professional, Jelinski needs to have his pump therapy to be easy, quick, discreet, safe, individual, and flexible. Going on a pump gave him more control over his basal rate, which is usually quite low during training (as little as 0.1 U/hr).

Questions and Answers

Q: During teenage years, A1c is slightly higher. What is a normal A1c in a non-diabetic healthy teenager?

Dr. Panowska: There is a lack of data showing the correlation between the A1c and glycemic profile in non-diabetics, so I can’t really answer your question.

Dr. Danne: Puberty is a very difficult time, and insulin resistance during puberty does account for the onset of type 2 diabetes in that age group. There is reason to believe that glucose control is more difficult.

Dr. Hanas (from the audience): Teenagers had exactly the same A1cs as adults when we examined this many years ago.

Q: Dr. Bootle, why did you go on an insulin pump?

A: I had a medical reason to go on a pump because of nighttime hypos. I think that as doctors we have to sell technology a bit. Patients have to form a relationship with the technology, and I don’t think most people will want to go on one until they see a need. Our role as doctors is to help people form a relationship with their technology.

Q: Isn’t it a bit disappointing to see above-target results with a lot of bolusing in children?

Dr. Danne: It’s a good point, and we need to think about that. Usually even people who use the pump irregularly don’t want to part with it, because it gives them flexibility. We’re all looking forward to new technologies to help these people get their glucoses to target.

Q: How can we make people motivated to change?

A: I think that’s the role of general practice. Often times, general practitioners are scared of changing things with the insulin pump because diabetes care is handled by specialists. I think there’s a lot of unresolved grief and difficulty that we never have a chance to address. I actually believe that specialists and primary care need to become closer.

A: I think if you set the stage as early as possible to lessen the hatred and negative feelings about diabetes, you’ll help people to achieve better control.

Q: Eva, would you start a child at three years at a basal of 25%?

A: It depends on the age and also the duration of diabetes. In very small children, I would probably use 10-15% of the total dose, but it depends a lot on the remission period.

Q: Do you think, Dr. Danne, that using variability information is useful in daily life?

A: I believe in glucose variability, and I would try to reduce it using technology. I’m not trying to put too much emphasis on this, however, because it’s to some extent unavoidable, and since we’re not sure about this in the scientific community, I don’t think it’s a top priority.

Abbott Corporate Symposium: Using FreeStyle Navigator in Clinical Practice – Who, When and How?

The Abbott symposium was very well attended, reflecting strong interest in the FreeStyle Navigator, particularly in Europe. The overall tone of the presenters was more “manufacturer neutral” than we saw with DexCom yesterday, but once again there was enthusiastic support for CGM. The only slight cloud in the sky was the call for yet more clinical trials to establish a fact base for policy making and reimbursement.

Today’s news from Abbott was that they have now reduced sensor calibration time from ten hours to one hour – a very welcome step forward – and that they now have European approval for more than one patient to use the receiver (the on-board data can be expunged), which bodes well for more part- time type 2 use.

 

CGM: WHERE ARE WE?

Gerard Reach, MD (Hopital Avicenne, Bobigny, France)

Dr. Reach showed how CGM fills an important need for patients, and solves the problems of SMBG. It addresses the double-edged sword of glycemic variability and hypoglycemia. Patients will use it if it is convenient and comfortable.

  • Glycemic variability is really the root cause of most diabetes problems, leading to hypoglycemia, complications and poor quality of life. As well as intra-day variability, inter-day variability is very common, and leads to real difficulties in insulin dosing.
  • In interviews, patients exposed to CGM say that they appreciate ‘filling in the gaps’ when they didn’t have glucose data, they are afraid of hypoglycemia, and they want to prevent glucose excursions, not treat them. Analysis of logbook data showed that patients state that they would increase insulin dosage, but they don’t in practice because of fear of hypoglycemia.
  • SMBG has many problems – it gives only an infrequent snapshot, it’s often used passively not proactively to adapt insulin dose, and adherence is very poor (40% adults and 26% children) – we aren’t sure how adherence was defined.
  • CGM works very well providing it is used. In CGM studies, a 10% increase in adherence gave a 41% increase in the probability of a 0.5% reduction in A1c. The accuracy of CGM is more than adequate - the Clarke error grid accuracy data represent excellent work by the manufacturers.
  • CGM will be used by patients if: it is comfortable to use, if they believe it is useful to them (less fear of hypoglycemia, educational value, improves A1c and glycemic variability), and if it doesn’t create new problems.

REAL TIME CGM: WHO AND WHEN?

Jan Bolinder, MD (Karolinska Institutet, Sweden)

A patient quote makes the value of CGM clear: “This is really the first time in 23 years of diabetes that I have taken the control of my diabetes, instead of the opposite.”

Dr. Bolinder gave us a measured presentation with little in the way of fireworks. His main argument was that although we are converts to the CGM cause, we still don’t have enough definitive data from randomized controlled trials to draw rigorous conclusions in many areas – and this needs to be done urgently.

  • CGM gives instant glucose measurements, automatic hypoglycemia and hyperglycemia alarms, a time trend and predictive ability. It can be thought of as sailing a boat using a modern GPS device for navigation, seeing where you are on the map at all times, rather than using old-fashioned occasional point observations of the sun and stars.
  • Physicians should use CGM for patients with poor control, experiencing severe hypoglycemia or hypoglycemia unawareness, who are pregnant, starting pump therapy, or are ill.
  • Dr. Bolinder reviewed the major CGM randomized controlled trials :
    • The GuardControl study showed that CGM lowers A1c (by 1% after three months), for all segments - adults, teenagers and children.
    • In the STAR1 trial (six months, 140 patients) – there wasn’t an improvement in A1c, but hypoglycemia was reduced compared to control.
    • The JDRF CGM trial showed that adults reduced A1c 0.5%, but children and adolescents didn’t have a statistically significant reduction. The outcome was associated with how much the device was used.
  • Applying the highest standards of proof, and combining and summarizing the results of seven significant CGM randomized controlled trials to date, Dr. Bolinder concluded that:
    • For patients with poor glycemic control, CGM really reduces A1c; with well controlled patients, it is not clear if it does reduce A1c (“but wait for the JDRF results tomorrow!”)
    • CGM probably (but not definitely) improves glycemic variability.
    • Compliance is the key to success, regardless of the age of the patientThere is no proof that CGM reduces severe hypoglycemia, but it possibly reduces time in the hypoglycemia range.
    • There is no evidence that CGM reduces patient fear of hypoglycemia or improves quality of life.
  • Although this sounds like a pessimistic interpretation, we agree with Dr. Bolinder’s assertion that more evidence from well-designed and adequately powered randomized controlled trials is needed to persuade reimbursement authorities and policy makers. Dr. Bolinder stressed that there is no need to persuade physicians and patients – they are already on the train. Fortunately, there are already at least 16 planned or ongoing randomized controlled CGM trials.

VIDEOS

Abbott showed two video testimonials from Navigator patients. The following are some of the points raised by the patients:

Patient 1

  • The Navigator has definitely reduced the number and severity of hypoglycemia episodes.
  • Using the Navigator I can definitely tighten my control without having to worry so much about hypoglycemia.
  • In three months my A1c went from 8.1% to 6.9%.
  • My wife likes the instant feedback.
  • I like being able to wake up and see how I ran overnight and being able to look at that over a few days and maybe change my basals.
  • I want it to be more discreet.
  • I wouldn’t give it up.

 

Patient 2

  • I was amazed what I was missing, even testing 12 times a day or more.
  • Without the device when I went for a run I would have to stop and check several times during a workout.
  • It gives me a view of the trends, for meals and exercise and driving. I can see where it’s at and where it’s going.
  • I don’t lose the data like I did with a meter.
  • After 90 minutes of initial training, I was really comfortable with every aspect of it.
  • I wish they would have a smaller transmitter.
  • My wife loves it.

DIRECNET: NAVIGATOR PILOT AND FEASIBILITY STUDY IN CHILDREN WITH TYPE 1 DIABETES

Bruce Buckingham, MD (Stanford Medical Center, Palo Alto, CA)

Dr. Buckingham presented results from DirecNet – a network of centers researching CGM use. His results were very positive for CGM, and he sprinkled them with lots of valuable clinical learnings. Participants’ A1cs went from 7.1% to 6.8%, and fingerstick frequency went from five/day to four/day. The patients who got the most benefit from CGM were looking at the sensor 20-40 times a day. Dr. Buckingham thinks that children can understand CGM devices better than their parents, which makes them more independent.

  • A study of 28 children for three months showed that on average they spent 6.6 hours each day above 200 mg/dl. New CGM users can be initially shocked by this and can be tempted to overcorrect and stack insulin. Bolus wizards are very important for new CGM users.
  • DirecNet has not seen any differences in the results of CGM depending on patient age, or sensor site on the body.
  • The patients who got the most benefit from CGM were looking at the sensor 20-40 times a day. Note that taking away children’s watches made them look at the sensor more.
  • Participants’ A1cs went from 7.1% to 6.8%, and fingersticks went from five/day to four/day.
  • Some interesting learnings from DirecNet:
    • Kids can understand CGM devices better than their parents, which makes them more independent.
    • MDI patients don’t do quite as well as pumpers, and the drop out rate was higher.
    • Patients who had more lows used the device less.
    • People who are well controlled seem to respond best to CGM. The issue is to get the less motivated people to use it and benefit from it.
  • The JDRF CGM trial showed that patients who used the device at least six days a week all improved A1c, regardless of age. The team apparently tried four times to get this figure into the NEJM article, but were only successful in getting it into the editorial. We also got a two-second peek at the data for the patients with baseline A1c <7%, which is to be formally presented later in the conference. From what we saw subliminally, it looks to be very good…
  • As part of the JDRF CGM trial an online teaching tool was developed that contains excellent coaching, followed by a multiple choice quiz on how to interpret and use CGM. It can be found at https://studies.jaeb.org/ndocs/extapps/CGMTeaching

Question & Answer Session

Q: (Dr. Danne) – Does the choice of CGM manufacturer affect the outcomes in kids in the JDRF trial?

(Dr. Buckingham) It is true that there are a lot of differences between the devices. But we are never going to analyze the JDRF data like that [by manufacturer]. In any case the technology moves quickly and there will be improvements. There are also bigger factors at work – for example, diabetes control is not so important to adolescents.

Q: (Satish Garg). What do you have to do to get a replacement label (i.e eliminate need for confirmatory fingersticks)?

(Unknown respondent) I think Abbott has a lot of experience in applying for that label, but they were unsuccessful in the USA. However, they sort of got it in Europe. (They still need a confirmatory fingerstick for low or rapidly changing glucose.)

(Arleen Pinkos, FDA): There’s a big difference between the CE Mark and FDA Approval. I have always said that manufacturers should put patients in a simulated home situation and see if they stay safe. The FDA would be interested in seeing these data.

(Dr. Bode) Abbott introduced the 10-hour warm up specifically to reduce ‘play’ so they could go for replacement claim. It was really a safety mechanism to get to optimal data.

(Joe Bugler, Abbott): Abbott is launching a new sensor right now that has only a one-hour calibration window rather than ten hours.

Q: What about pregnancy – it’s not approved?

(Jan Bolinder) I agree that pregnancy is an ideal indication. But we lack the data to convince the authorities. That’s the message with my lecture – we need more trials.

Comment: (Bruce Bode) I agree with Dr. Bolinder, we don’t have hard data on severe hypoglycemia. In fact in STAR1 there was a negative trend. Alarms didn’t wake up the person. I believe that an auto shutoff would have eliminated severe hypoglycemia in that trial. But to do a good CGM hypoglycemia study, the numbers would have to be astronomical since it’s a relatively rare event.