ENDO – 96th Annual Meeting of the Endocrine Society

June 21-24, 2014; Chicago, IL; Full Report - Diabetes Technology - Draft

Executive Highlights

At this year's ENDO conference, diabetes technology was not a major focus, but we heard an update on the bionic pancreas, commentary on CGM, and a few highlights in mobile health. Dr. Ed Damiano (Boston University, Boston, MA) and Steven Russell (Harvard Medical School/MGH, Boston, MA) discussed the findings of the Beacon Hill and Summer Camp studies, published in the NEJM two weeks prior (by far and away the most popular NEJM article last month with 220,000 views to date, far ahead of the second place rank of 61,000). Their talks highlighted new features of the device, namely a one-time “microburst” dose of glucagon that patients can deploy manually, and an option to temporarily alter the target blood glucose level that we think might be helpful for exercise in particular, and confirmed their ambitious timeline for commercialization in the fall of 2017. The fundraising for trials has gone very well to date but the device development work needs a clear shot in the arm in order to make the timeline. We also heard from the highly regarded Dr. Irl Hirsch (University of Washington, Seattle, WA), who commented that “CGM is going to bypass pump therapy and MDI in future years.” Dr. Anne Peters (USC, Los Angeles, CA) gave a similarly enthusiastic talk on CGM, showcasing her experience with underserved patients in LA and highlighting efforts to obtain Medicare coverage for the technology. Mobile health came up as well – Dr. Ronald Tamler (Icahn School of Medicine at Mount Sinai, New York, NY) presented on the potential of smartphone apps in improving diabetes self-management. In particular, he highlighted the need for easy-to-use, effective apps, since 86% of diabetes patients indicated they were interested in downloading an app, though only 4% used one.

Below are our talks on diabetes technology. Titles of the most notable presentations are highlighted in yellow, while coverage that was not included in our daily coverage is highlighted in blue.

Diabetes Technology

Endocrine Fellows Series: Type 1 Diabetes Care and Management

Making Diabetes Management Disappear: A Bionic Pancreas for One and All

Edward Damiano, PhD (Boston University, Boston, MA)

Dr. Edward Damiano wrapped up the Endocrine Fellows Series with an engaging talk on the bionic pancreas, complete with a live demonstration. The team’s work has received an impressive level of media attention since the results of the Beacon Hill and Summer Camp studies were published in the NEJM in June (by far and away the most popular NEJM article in the past month) and the excitement in the room of fellows was palpable – indeed, the Q&A probably could have continued for another hour. Much of Dr. Damiano’s talk echoed Dr. Russell’s presentation from Day #3 of ADA 2014, but there were a few new tidbits – most notably, the team would like to include Kaiser as one of the study sites for the bionic pancreas pivotal trial testing the final commercial device (dual-chamber pump with algorithms embedded, Xeris glucagon). This could go a long way toward convincing payers early on that the bionic pancreas deserves reimbursement, and Dr. Damiano is hoping to have Kaiser or a similar managed care consortium play a role in a larger post-market approval study as well. Dr. Damiano also mentioned several new features that have been added to the bionic pancreas since last year’s studies, including the ability to give a one-time “microburst” dose of glucagon and the option to temporarily alter the target blood glucose level. He predicted that the final device will be ready for testing in ~18-24 months (end of 2015 to mid-2016) and confirmed that he and Dr. Russell are still aiming for commercialization by the fall of 2017. We learned at FFL 2014 that fundraising to build the device is a current roadblock – approximately $5 million is needed in the next 60 days. As a reminder, the team’s ambitious four-site multi-center outpatient study started on June 16 – we hope to hear interim results as data accumulates over the next year.

  • Dr. Damiano and Dr. Russell would like to use Kaiser or a similar managed care consortium to be one of the sites for the pivotal study of the bionic pancreas in 2015-2016, which could be crucial for future reimbursement efforts. We wholeheartedly agree with Dr. Damiano that engaging payers early in the development process will be very helpful. He even hopes that Kaiser or a similar managed care consortium could play a role in a larger study after the bionic pancreas is approved as well – this could help accumulate more real-world data on the glycemic benefits and safety of the device. Dr. Russell mentioned at ADA that the Beacon Hill/Summer Camp studies may have underestimated the bionic pancreas’ effect size, given that the control groups received much better care than real world management of type 1 diabetes – we expect the multi-center and pivotal studies would not be much different. A real-world post-market approval study could address this unanswered question more robustly.
    • Dr. Damiano hopes that once Kaiser or a similar managed care consortium is convinced of the bionic pancreas’ cost-effectiveness (e.g., significant reduction in ER visits for DKA and hypoglycemia), other payers will line up. This would of course be the holy grail for a device, though demonstrating significant reductions in these severe events might not be possible until the large post-market approval study.  
  • Dr. Damiano mentioned two new features – microburst glucagon and temporary glucose targets – that have been added since last year’s studies. The microburst feature would allow users to administer a one-time dose of glucagon before disconnecting from the device, such as before showering or swimming.  This would reduce the concern about hypoglycemia during short periods of time when users are not wearing the device. The temporary target feature would allow individuals to select a higher blood glucose target range than the default value for a few hours; for example, a truck driver might choose to set the glucose target 20 mg/dl higher during a four-hour drive to even further reduce the risk of hypoglycemia beyond what the device provides using its default target.

Questions and Answers

Q: If this device is tied to an iPhone app, what happens if the iPhone dies?

A: The device in the pivotal studies will be nothing like this. We’re proposing a dedicated medical device; it won’t be on an iPhone at all.

Q: Interstitial glucose is delayed by five to ten minutes compared to blood glucose. What happens if there’s a sudden drop?

A: Interstitial glucose is our only signal, and there can be a big differential, so our system gives glucagon preemptively. If it sees a fast enough drop, it backs off insulin and gives glucagon. Glucagon is absorbed rapidly (as fast as juice), begins to work within ~10 minutes, and peaks within ~20 minutes. Because we know there’s a lag, we’ve designed the bionic pancreas to give glucagon preemptively, before the blood glucose becomes too low. The CGM reading could be as high as 180 mg/dl, but if it’s dropping fast enough, it’s giving glucagon.

Q: Your patients had a great A1c in comparison to many patients. What was your inclusion criteria? And if you’re using glucagon for hypoglycemia, does that affect the physiologic counterregulatory response?

A: We saw no indication of a counterregulatory response to hypoglycemia in our C-peptide negative subjects with type 1 diabetes during our inpatient studies, where we measured plasma glucagon levels every 15 minutes. We saw no physiologic response at all and suspect this has to do with the disruption in the paracrine effect of the beta cell on the alpha cell.

Q: Was that part of your criteria?

A: In our inpatient studies and Beacon Hill Study, the subjects had to be C-peptide negative, but we’ve removed that requirement for our two Summer Camp Studies and for our Multi-Center Study. Similarly, initially the upper limit for A1c was 9%, but we’ve removed that as well. As far as A1c is concerned, we take all comers.

Q: What I’ve seen is that patients who go to camp want to learn about diabetes – they’re self-selected. Is that a bias that could affect results?

A: The baseline A1c of the kids at camp was 8.2%, which is not that different from the national average, so they weren’t doing much better at management than the population at large. In that sense, we weren’t getting a self-selected group at camp, though we were with the adults in the Beacon Hill Study. One might think this might bias the results in the Beacon Hill Study. However, notice how well the device does (referring to a slide) with someone who has an average over the study period of over 200 mg/dl – it brings them well below the ADA goal for therapy with virtually no hypoglycemia. Nothing would suggest that people doing worse at baseline would have worse control on the bionic pancreas than someone would who was under good control at baseline. They all come together. It doesn’t seem to matter.

Q: Is there any bioinformation the patient can put in? There’s that mealtime option, but what if people want to give more information? And if you disconnect the sensor and you’re not putting blood glucose values into the algorithm, what happens?

A: If the CGM is offline, you will have to enter blood sugar levels into the device regularly (preferably five to seven times every day while the CGM is offline). There’s nothing more the bionic pancreas can do for you if you’re not going to play some role in providing it information when the CGM is offline. It’ll give you basal insulin, which the device determined previously during periods when the CGM was online, so you won’t go into DKA, but for it to be able to do the most it can for you when the CGM is offline, the user will need to enter frequent blood sugar levels into the device and not forget to use the meal-announcement feature. What other biometrics were you thinking of?

Q: Like in the menstrual cycle, sugar shoots up.

A: It’ll see that higher insulin requirement during menstruation and automatically adapt insulin dosing upwards as the cycle begins, and then downward as the cycle ends.

Q: What about an athlete who exercises at a regular time every day? It would be helpful to put that in.

A: That’s what the temporary glucose target is about. If you use the device with the default settings and find that you sit low during certain activities, you can anticipate this by adjusting the temp target. If you exercise at the same time each day, you can set the temporary target higher at that time. That’s how you’d manage that. Things like heart rate aren’t predictive of glucose. If I play basketball with my son and he’s not stressed, his blood sugar tends to drop rapidly, but when doing the same activity in a competitive situation, like during a basketball game with his peers, his blood sugar tends to rise during the game. The verdict is in glucose, that’s what the pancreas looks at. Other biometrics can be misleading. I have found no other biometric you can trust other than glucose. By the way, look how well the device works; its only input is the CGM. I don’t know how much better the control needs to be.

Q: Did you include adult subjects with other comorbidities?

A: No, they would’ve been excluded.

Q: How close is this to coming to market?

A: You saw the last slide. I can’t be more specific than that, no one can.

Dr. Irl Hirsch (University of Washington, Seattle, WA): The science is great – we’re all wowed. But where I live, I struggle to get CGM, I struggle to get strips and insulin. Working at the beginning with all the carriers and payers and the ACO infrastructure, my concern is that we’ll just be separating the haves and the have-nots.

A: Yes, well, unfortunately, this device will even further the gap between the developing world and the West. We’ve only focused on the West, and the only way it will work there is with reimbursement.

Q: Are you working with payers?

A: We are engaging private payers and we will engage CMS. If we do not engage both now, the outcomes we build in to satisfy the FDA may not satisfy CMS. The multi-center study we’re doing now is like a mini-pivotal. We’ll add 8-10 more sites in the pivotal study and one is potentially Kaiser, so they’ll be able to gain early access to the data and see first hand how the bionic pancreas works.

Q: The CMS population may be more important than young people because of all the hypoglycemia. We need to cut and slice the data. The data is very clear that you can reduce hypoglycemia. When 20% of people with a duration of longer than 40 years have had a severe hypoglycemic event in the past year, this is a no-brainer.

A: Even having Kaiser won’t be enough, but it’s a good start. For the rollout after clearance, we want to do a much bigger study to get people on it quickly and hopefully have Kaiser or a similar managed care consortium play a role (we haven’t asked them yet). Hospital visits for DKA and severe hypoglycemia are expensive, so if we can get Kaiser to run a very large study for a few years right after FDA clearance is obtained, and they see tremendous value – they will see for themselves that they’ll benefit from large savings on acute visits right away – that’s what we hope to achieve.

Q: That would get you another New England Journal paper.

A: That would be fine; but, more importantly, I just want all the other payers to line up.


Smartphone Use By Patients With Diabetes (MON-1005)

A Ross, G Boyd-Woschinko, D Kaiser, A Alifarag, D King, M Diefenbach, R Tamler

This study sought to document the prevalence and use of smartphones among patients with diabetes. The researchers surveyed 120 English-speaking patients at the Medicare/Medicaid and diabetes faculty practice at the Mount Sinai Medical Center. Results indicated that smartphone apps are not presently used in diabetes self-management and that there are strong socioeconomic differences in smartphone ownership. The researchers hope that these findings will enable the design of a more user-friendly and targeted app – the team is planning clinical trials in the coming year.

  • Of the 120 patients surveyed who owned a cell phone, 81 patients (68%) owned a smart phone. Android OS (41%) and Apple’s iOS (47%) were the most prevalent operating systems.
  • Smartphone ownership varied significantly based on demographic factors – ownership was higher among faculty practice patients vs. those at the Medicare/Medicaid clinic (88% vs. 46%, p<0.01), among graduate degree holders vs. subjects who never attended college (100% vs. 46%, p <0.01), and among Caucasians compared to minorities (95% vs. 56%, p <0.01).
  • Seventy participants (86%) said that they would be interested in using a diabetes self-management app, although only one patient was found to be using an app. We see this willingness to adopt mHealth technology as a great sign for the industry, although it’s clear that current product offerings are just not clinically useful for patients. 
  • The following properties were ranked as most important in a diabetes self-management app (in order of decreasing importance): ease of use, legibility, functionality for tracking blood glucose, weight, blood pressure, and food intake, recording medication administration, and reminders about medications and doctors appointments. The emphasis on usability does not surprise us, given the day-to-day hassles and time commitment of managing diabetes although given how many patients miss appointments according to many clinicians with whom we speak, we’d keep the reminders a bit higher about meetings! (We would also emphasize Skype and other meetings more often.) “Ease of use” also goes beyond simply app design in our view – for broad adoption and widespread, it should make living with diabetes much easier.

Provider Attitudes Toward Smartphone Use By Patients with Diabetes (MON-1006)

A Ross, G Boyd-Woschinko, D Kaiser, A Alifarag, D King, M Diefenbach, R Tamler

In conjunction with the poster above, Dr. Ronald Tamler and colleagues presented a second study that examined 70 healthcare providers’ perceptions of their patients’ smartphones preferences. Of the 70 surveyed clinicians, 32 said that at least half of their patients suffered from diabetes. The goal of the survey (administered at a medical education event) was to determine how designers might best tailor an app to improve diabetes self-management. A sizeable majority of providers (76%) reported that they would like to incorporate a smartphone app into the management of their diabetic patients – of course, interest is one thing, but the devil will be in the details in terms of HCPs’ adoption of mHealth technology. Reimbursement is clearly the big elephant in the room – anything that does not save providers time and that they cannot get paid for strikes us as a long shot in the marketplace.

  • The vast majority of providers (83%) owned a smart phone, although many (70%) thought that “fewer than half” of their patients owned a smartphone. Given the above finding that 68% of patients reported owning a smart phone, it seems that HCPs underestimate smartphone ownership.
  • Notably, that only 11 of the providers (16%) believed that patients with a smart phone would utilize an app for diabetes self-management. Whether this lack of confidence is a reflection of their experience with the relatively poor quality of apps to date, doubt about adherence to a mobile application, or unfamiliarity with what an app entails was not explored. Clearly, this is an area begging for further investigation.
  • Most providers (84%) noted that their patients usually do not bring their blood glucose meter to appointments. Combined with the notoriously illegible/unreliable nature of paper logbooks, we believe this is a critical gap that mHealth can fill. Devices like Glooko’s MeterSync cable, Telcare’s BGM, and LifeScan’s VerioSync are important steps in this direction.
  • A majority of providers (76%) reported that they would like to incorporate a smartphone app into the management of their diabetic patients. Although surveys are notoriously weak in ascertaining how a clinician would act in practice, their reported willingness to use an app is an encouraging sign and implies that there is potential for mHealth technology to become a more central part of diabetes care.

Symposium: Meeting Patient Needs in the Evolving Landscape of Type 1 Diabetes

Meeting Patient Needs in the Evolving Landscape of Type 1 Diabetes

Anne Peters, MD (University of Southern California, Los Angeles, CA)

After presenting data from the T1D Exchange to illustrate the many challenges patients and providers still face in managing type 1 diabetes, Dr. Anne Peters made a persuasive case for the potential of CGM to vastly improve the lives of patients, regardless of their age. Dr. Peters said it is “almost impossible for me to manage patients these days without CGM.” Dr. Peters then discussed efforts to introduce her underserved patients in East LA to the technology after her center was “kicked out” of the JDRF CGM study because her patients were unable to adhere to its intricate and complex protocol. Instead, Dr. Peters conducted her own small trial of CGM in this patient population, most of whom were very low income, had only a high school education, and had blood glucose levels typically in the range of 350-400 mg/dl. Additionally, about 50% of them had already suffered complications of diabetes. Though some patients found the technology difficult to use, 80% of them indicated that CGM made it easier to adjust insulin and avoid lows and that they would continue to use it if they had access. Unfortunately, few do, since Medical (California’s insurance for low income people) does not cover CGM. Dr. Peters wrapped up her presentation by showing a video she created of her older patients advocating for Medicare coverage of CGM. The patients powerfully talk about the immense impact it has had on their health and peace of mind. A bill is currently before Congress that, if approved, would allow CGM to be covered by Medicare. Dr. Peters expressed our shared hope that such strong statements from patients would help move this effort forward. 

  • Data from T1D Exchange shows that CGM use has increased over the past few years but is still only 14% penetrated in older patients. Patients in the Exchange who used CGM had a lower A1c than those who did not (7.2% compared to 7.7%), and given the high risks of severe hypoglycemia in the elderly and the low percentage with an A1c <7%, it is clear from our view that access to this technology could offer immense benefits for these patients.
  • Patients in Dr. Peters’ video sent a powerful message about the positive impact of CGM on their daily lives. They described the reassurance they felt knowing they would be alerted if they were in danger of hypoglycemia – “I no longer think about whether I’m going to wake up in the morning” – and said “it makes no sense” that patients over 65 years would not be “entitled to the best medical care.” Dr. Peters said that this video has been shown to members of Congress and urged the audience to continue pressuring their representatives to move the Medicare CGM Coverage Act out of committee.

Symposium: New Treatment Options

Artificial and Bionic Pancreas Technology

Steven Russell, MD, PhD (Harvard Medical School, Boston, MA)

In front of a packed, standing-room-only audience, Dr. Russell shared topline results from the Beacon Hill and bionic pancreas Summer Camp studies previously presented at ATTD 2014 and ADA 2014, which were simultaneously published online in the New England Journal of Medicine in late June (“Outpatient Glycemic Control with a Bionic Pancreas in Type 1 Diabetes”). Echoing his previous talks, Dr. Russell emphasized the impressive glycemic control exhibited during closed loop in both studies and noted the significant reduction/lack of an increase in hypoglycemia in adults and children, respectively. Notably, he also spent a few minutes comparing these results to those of alternative studies in artificial pancreas technology. He highlighted data from a four-week study of overnight closed-loop control presented by Dr. Hood Thabit (University of Cambridge, Cambridge, MA), at ADA 2014 – while the results were promising, he noted that the nighttime-only approach does have limitations. Looking ahead, Dr. Russell expressed excitement about the team’s ambitious four-site multi-center outpatient study that started on June 16, which will compare 11 days of glycemic control with the Bionic Pancreas to 11 days of usual care in adults with type 1 diabetes. He emphasized the real-world nature of the experiment, which will allow subjects to live their daily routines unencumbered and “should really test the system in a robust way.”

Panel Discussion

Q: Given the link between LDL levels and CVD risk, is there any concern about SGLT-2 inhibitors?

Dr. Samuel Dagogo-Jack (President-Elect, Medicine & Science, ADA): LDL is a significant predictor of CVD in diabetes. Any increase in LDL should be avoided. There is a tendency for a modest increase in LDL with SGLT-2 use that hasn’t been fully explained. There has been some data, although at the present time, I’m not well-informed or positioned to talk about the long-term impacts of LDL on CV event rates.

Q: What do you think about combinations of SGLT-2 and DPP-4 inhibitors?

Dr. Robert Vigersky (Walter Reed Army Medical Center, Washington, DC): I don’t know about any data on these combinations. But these drugs are approved for combinations.

Q: Any comment on TZDs and bladder cancer?

Dr. Vigersky: the new info is that pioglitazone has been banned in India. It’s still available in France and the US. The data to support that has been from retrospective studies. In my view, the French study was not convincing. There were a lot of confounding factors. At the present time, I use these agents. I’ll use them in lower doses, but in terms of their teratogenicity, I’m a little bit skeptical. I expect a randomized control trial.   

Q: Can you talk about glucagon inhibitors in the pipeline and any potential for those?

Dr. Steven Russell (Harvard Medical School, Boston, MA): In type 2, inappropriate glucagon secretion can work against therapeutic efforts to maintain blood glucose control. There are glucagon inhibitors in the pipeline, but I’m not very well informed about the latest in that line of research.

Q: Some patients with type 1 diabetes may have very high H-LDL cholesterol levels and thus higher risk of CVD. What are your thoughts on that?

Dr. Dagogo-Jack: The issue is that some families may have certain mutations and are thus more at risk of having high H-LDL levels. Some are protected against it and others are not. HDL levels are not the whole story. The functionality needs to be further determined in the clinic.

Q: Can you comment on glucagon being inappropriately high in diabetes? Does glucagon make the liver more resistant to insulin?

Dr. Russell: Especially in type 2 – and sometimes type 1 – there can be paradoxical glucagon released in response to meals. Really, that shouldn’t happen. It might be helpful if we could suppress that endogenous release that is not in response to hypoglycemia. A separate question is whether the fact that the bionic pancreas gives glucagon intermittently to prevent hypoglycemia increases the insulin requirement. The amount of insulin that the bionic pancreas gives, for people who are in good control during usual care, is the same as usual care for both kids and adults. In adults, we did have people who had high average blood glucose during the usual care who used more insulin on the bionic pancreas than during usual care. We also found that that they ate 50% more carbohydrates during the bionic pancreas arm than the usual care arm, perhaps because they felt they could safely do that. Those are two reasons that might account for the increase in insulin use in those patients, so it doesn’t appear that they needed excessive insulin because of glucagon administration. Also, we haven’t seen any evidence for tachyphylaxis with glucagon.

Q: Did you have any technical problems with glucagon?

Dr. Russell: Currently, glucagon is not stable. We had to reconstitute it every day. However, several companies are interested in making glucagon more stable. The current leader is Xeris, who is making a human glucagon in a non-aqueous solvent. And it seems to be stable for at least 2 years at room temperature. We’re in the middle of a euglycemic clamp study with it right now. Other pharmaceutical companies, including small and large companies, are also developing stable glucagon formulations and analogs. We’re confident we will have a stable, pump-able glucagon to use in the device.

Q: Did you have issues with battery life?

Dr. Russell: With the preliminary device, we did have to leave it plugged in during the night and charge it a couple times during the day. However, the complete device will only need a couple AAA batteries that can be switched out every couple weeks or so.

Q: Can you talk about health care access for the military and veterans?

Dr. Vigersky: I can only comment on the military. If someone served for 20 years, then they are entitled to complete healthcare benefits for life for themselves and their spouse. Military facilities do have GLP-1 agonists on their formularies. You can get SGLT-2 inhibitors, just like you can with private insurers, but you have to fill out forms saying why the patient can’t take other agents or has failed on other agents before they’ll dispense SGLT-2 inhibitors. So, you can certainly get it, if it’s appropriate. Bromocriptine is not on formularies. Insulin analogs are on formularies, although insulin detemir is not available in tent form.

Q: The other question is how much do dopamine agonists have an effect on growth hormone levels?

Dr. Vigersky: In those people with growth hormone deficiency, they very rarely have hypoglycemia because they have other countering mechanisms intact. I don’t know if anyone has tested this yet. But my suspicion based on general knowledge is that it may have a minor effect of suppressing GH.  

Meet The Professor Sessions

Continuous Glucose Monitoring

Howard Wolpert, MD (Joslin Diabetes Center, Boston, MA)

The highly regarded Dr. Howard Wolpert gave a clinically focused overview of the role of CGM in minimizing hypoglycemia. His excitement regarding the technology was evident (“With the use of CGM, patients can minimize glycemic variability AND reduce hypoglycemia!”), although his presentation also delineated some of the shortcomings that have limited adoption of the technology – sensor burnout, inaccuracy, and alarm fatigue. Dr. Wolpert emphasized that a patient’s understanding of the device is critical to its optimization, from understanding postprandial patterns to making dietary adjustments to grasping physiological lag to accurately calibrating with a BGM. We were very interested to hear his view that carbohydrate-counting based methods are an “obsolete and inaccurate” method of mealtime insulin dosing. In Q&A, he expressed some reservations about the Beacon Hill study of the bionic pancreas, as not all participants in the control group (“usual care”) wore CGM. This was a valid criticism that we also heard from Dr. Roman Hovorka at ADA 2014, though we’d note it was a deliberate choice by Drs. Russell and Damiano – the goal of Beacon Hill was to compare real-world efficacy of the bionic pancreas to what patients are currently using in practice. At ADA, Dr. Russell higFKAPOOR

hlighted that CGM is only used by ~9% of US patients, though it was actually used by 45% of study participants during the usual care arm of Beacon Hill. This is of course a critical question as closed-loop systems move towards commercialization – What is the appropriate comparator? (Pump + CGM? SMBG + MDI? Patients’ current therapy, whatever it might be?)

  • Dr. Wolpert did not seem particularly surprised that CGM adoption has been limited, given issues with sensor burnout, inaccuracy, and alarm fatigue. Dr. Wolpert noted that the hassles of the technology often outweigh the benefits for many patients, particularly given issues with device accuracy and software malfunctions. Indeed, he suggested that the quality-of-life benefits with CGM often outweigh the hassles only for patients suffering from significant hypoglycemia unawareness and those with the greatest risk for episodes of severe hypoglycemia. However, Dr. Wolpert is confident that the next generation of CGMs will address these issues – we believe that the already-available Dexcom G4 Platinum has made significant strides on the accuracy and reliability fronts, and we are also confident that future sensors from both Medtronic and Dexcom will continue the trend towards a better hassle-benefit ratio.
  • Patients often don’t understand the implications of physiologic lag time: interstitial fluid vs. blood glucose. Data has demonstrated that CGMs are relatively slow in reporting postprandial glucose spikes (in particular, spikes occurring at a rate greater than 2 mg/dl/min), exercise-induced rates of decline, and hypoglycemic recovery. In order to avoid overtreatment, Dr. Wolpert encouraged patients to refrain from treating based on CGM data alone during these periods and to calibrate devices based on steady-state glucose levels.
  • Dr. Wolpert implied that carbohydrate-counting based methods are an obsolete and inaccurate method of mealtime insulin dosing. He cited the different glycemic indices of various meals, emphasizing the need to match insulin action profiles with carbohydrate absorption rate. Additionally, Dr. Wolpert highlighted that non-carbohydrate calories (e.g., fatty acids) have an impact on insulin resistance and gastric emptying – this was demonstrated in his influential 2013 paper (Wolpert et al., Diabetes Care 2013), which compared closed-loop control after meals with identical carbohydrate and protein content, but different levels of fat. The high fat dinner required 40% more insulin than the low fat dinner, and despite the additional insulin, the high fat meal caused more hyperglycemia. Dr. Wolpert called for alternative dosing algorithms depending on meal’s fat content.
  • Dr. Wolpert spoke positively about the value of CGM alarms, though asserted that the optimization of alarm settings requires careful consideration – the goal is to avoid “nuisance” alarms that tend to undermine adherence. In most cases, we have heard experts call for alarms to be set quite widely at the beginning, with tightening over time as patients gain comfort with the system. Dr. Wolpert reminded the audience that the alarm threshold needs to be set ahead of the actual target number (higher than the low threshold and lower than the high threshold) in order to account for the lag of interstitial monitoring.

Questions and Answers

Q: What do you think about the efficacy of the bionic pancreas vs. therapy via CGM?

A: I would have liked to see CGM vs. the bionic pancreas data in the Beacon Hill Study, as opposed to the bionic pancreas vs. “usual care.” These issues with study design can really confuse results.

Q: How do you teach patients carb-counting when the meal composition influences insulin boluses so much?

A: What we do at mealtime is that we give 50% of the insulin bolus up front and 50% postprandially. There is a lot of individual variability in responses to insulin and carbohydrates. There are standard formulas, but I’m hesitant to apply them to everyone because of that individual variability. I wish I could give you better guidance.

Diabetes Management Tools and Applications

Ronald Tamler, MD, PhD, MBA (Icahn School of Medicine at Mount Sinai, New York, NY)

In one of the most entertaining presentations of the afternoon, Dr. Ronald Tamler made a case for the expansion and development of mHealth. Citing the ubiquitous nature of smart phones in our society (in an informal poll of the lecture hall, more than 80% of the room acknowledged having a smart phone), Dr. Tamler argued that this method of patient outreach can improve the management of diabetes, while also allowing providers a novel opportunity to interact with patients. He noted that in a survey of patients at Mount Sinai Hospital, 96% carried their smartphones on them, a far cry from the ~30% of patients with diabetes who bring their glucose meters to doctor’s appointments. Dr. Tamler further highlighted that 86% of patients with diabetes were interested in downloading an app for diabetes, though only 4% use one – in our view, this speaks to the absence of highly useful apps for diabetes, which will no undoubtedly require FDA approval. In particular, Dr. Tamler emphasized that apps can help adherence by improving health literacy, optimizing the affordability of a regimen, streamlining the monitoring process, and empowering patients with reminders and motivational messages. These benefits have been observed in a few studies that have hinted that mHealth intervention may reduce A1c levels and slow the progression of diabetes. However, Dr. Tamler emphasized that most studies of mobile health were performed before the advent of smartphones, meaning the future may hold even more promising products for the field.

  • In September 2013, the FDA released final guidance on mobile medical applications, which outlines the Agency’s tailored approach to mobile apps. Most importantly, it defines what products will be regulated by the FDA: (1) apps that are used as an accessory to an already regulated medical device (e.g., a secondary display for a CGM) and (2) apps that transform a mobile platform into a medical device (e.g., a glucose meter that plus into a smartphone). We believe that the most useful apps for diabetes are likely to require FDA approval/clearance, since they will ideally help patients make therapeutic decisions and/or interface with FDA-regulated products. 
  • Dr. Tamler noted that the challenge of clinician reimbursement for mHealth remains one of the biggest obstacles to the industry’s growth – “You cannot bill for the vast majority of non-face-to-face contact.” How do you evaluate and compensate a clinician’s effort? Is the time they spend online as valuable as the time spent in person? We would argued that “it depends” on the device being used, the frequency of contact, the patient and provider training, and many other factors. The critical question of reimbursement for mHealth is one that we have not seen a lot of movement on – WellDoc has perhaps done the most impressive job of garnering reimbursement for its BlueStar product, and that required two randomized controlled trials. Of course, that product is also designed to save clinicians’ time, something that many mHealth product do not do at this stage (i.e., they simply provide more data).
  • Dr. Tamler noted that 86% of patients with diabetes were interested in downloading an app for diabetes, though only 4% use one. He noted that the majority of current apps are “garbage” and that current product ratings systems are relatively useless to patients – there is a negligible correlation between ratings and user satisfaction (we would also add “meaningful clinical impact” to that correlation). Dr. Tamler also highlighted the speed with which technology changes and the ability to develop apps inexpensively. However, he noted in Q&A that the high cost of maintenance tends to undermine app success.
  • Among apps currently available, Dr. Tamler recommended Track 3 Diabetes ($6), Glucose Buddy (free), and WellDoc. As of our last coverage of WellDoc in January, the company had announced closing of a $20 million Series A round of financing, led by Merck’s Global Health Innovation Fund. The investment was expected to fund a dedicated sales force to regionally rollout BlueStar, WellDoc’s FDA-approved mobile prescription therapy for type 2 diabetes.
  • Most mobile phone intervention studies came before the advent of smartphones. For example, Dr. Tamler detailed a 2013 study of men with impaired glucose tolerance in India: only 18% of patients that received intermittent texts with motivational and practical advice progressed to type 2 diabetes over the course of two years vs. 27% of men who received generic lifestyle advice alone. Given the lag between new consumer technology and clinical research, we expect to more robust mHealth data in the coming years.
  • Dr. Tamler highlighted the MySinai Diabetes App, a novel platform under development that seeks to build on recent improvements to motivational messaging and user interfacing. In particular, he emphasized the integration of social media within the app and informed attendees that a clinical trial with insulin-requiring type 2 patients is in the works. 

Questions and Answers

Q: Can you explain the problems with billing a little more? 

A: The only non-face-to-face interaction you can bill for is the evaluation of CGM data. So the answer is that you cannot bill for the vast majority of non-face-to-face medicine. It’s a big problem.

Q: It seems as though if a patient is engaged in anything – it doesn’t even have to be the best app – they’re going to accomplish something and improve their performance?

A: It can go both ways. And this is why it’s good that the FDA got interested in this. There are some scary apps. Imagine an insulin app programed by a high school senior that does bad math. So, the answer is that it has to be done the right way. It has to be based on solid clinical evidence and done in a way that engages the patient, because I bet that you use only five apps out of the 80 or so you have on your smart phone. If you’re diabetes self-management app is not one of those five, then it is useless. It has to be scientifically and medically appropriate and get people to go back and use it. Otherwise, it doesn’t do anyone any good. Currently, there’s no solution to that. We’re trying to make our app so annoying that it forces you to use it. You see it with children’s apps that say, “You haven’t interacted with me for three days. What’s wrong with you?!” The app needs to be able to do that. But that can be annoying too. So there’s a medical challenge for accuracy and appropriateness, and there’s a behavioral science aspect to it.

Q: What about privacy when patients communicate data via mobile phones?

A: There are multiple matters that need to be addressed. First of all, information itself needs to be private. There are plenty of people out there who don’t use passwords on their phone. If it gets stolen, it’ll be subject to the inquisitive people that took the phone. The second part of this is the security of data transmission. There are ways of encrypting the data and ensuring that patients are aware that when they are transmitting data, it could be interrupted. But we do need to encrypt information to send it safely. That said, this is an issue for regular email communication, too. So the solution is that there needs to be a succinct encryption and privacy process, and there are solutions for that.

Q: Can you speak about the challenges of patients contacting doctors during non-work hours?

A: I’ve actually had experience with this when one of my hospital patients wanted an insulin refill immediately. This issue pertains to eHealth in general. That’s why it’s so important to have patient portals within any electronic medical system. I’m a big advocate of MyChart, which is used in my clinic. I tell my patients categorically that I don’t respond to emails. You have to call the office or go through MyChart. Patients find it empowering. And when I am away, my inbox is sent to another physician. But I agree, if you just hand out your email, then it’s a slippery slope, because patients love access. But there are problems associated with that.

Q: What happens to “garbage” apps?

A: There are apps that are just OK. There are apps that do a lot and are good. And there are apps that are truly garbage. In the app store, two different reviewers assign a score to an app out of 30 possibly points. If an app gets less than 10 points, then the app is horrible to use and has limited functionality. And there are a lot of them. But this is just a snapshot. You can develop an app very quickly. It’s easy to leapfrog competitors. But what is difficult is the maintenance and improvement of an app. Development is not that costly. But updating is particularly costly.

Managing Hyperglycemia on the Ward and in the ICU: Challenges, Controversy, and New Technology

Steven Russell, MD, PhD (Massachusetts General Hospital, Boston, MA)

Dr. Steven Russell advocated for the development and further validation of automated glycemic management technologies for the hospital environment, illustrating their efficiency in reducing hypoglycemic events. He began by presenting the Leuven studies, which demonstrated that intensive insulin therapy reduced mortality rates by 42% in the ICU. Later studies, however, failed to show such benefits, including the very large NICE-SUGAR study. Almost all of these studies had significant occurrences of severe hypoglycemia (<40 mg/dl), which has been shown to be an independent predictor of mortality. It is possible that benefits of tight glycemic control are being offset by harm due to hypoglycemia. In a more recent study of tight glycemic management in children post-cardiac surgery performed at Children’s Hospital in Boston, nurses were asked to frequently check real-time CGM in patients, which allowed for tight glycemic control without significant hypoglycemia. While effective, this approach unfortunately required significant nursing time. Dr. Russell then discussed a preclinical study testing a system for closed-loop automated glycemic regulation. The study was done at Boston University in a pig model of insulin deficiency and insulin resistance, meant to be a challenging case similar to the challenges found in critically ill patients. In this model, the system tightly controlled glycemia without hypoglycemia. This favorable data led to the development of the GlucoSTAT system for humans, which includes a Navigator CGM, a control algorithm, a dual-infusion Symbiq pump, and insulin/dextrose intravenous delivery. Dr. Russell showed results demonstrating that the Navigator had a MARD of ~11% in a mixed population of ICU patients if it was calibrated every six hours. He argued that use of CGM to obtain BG information has several advantages including not requiring more IV access and no risk of artifactually high glucose measurements resulting from measuring glucose infused at a site distal to the measurement site. In a clinical feasibility study of the GlucoSTAT system at Massachusetts General Hospital, the blood glucose of people with type 1 and 2 diabetes was controlled during fasting, after a meal, during simulated tube feeding (sipping Boost continuously over two hours) and after abrupt discontinuation of the simulated tube feeding (a situation that often leads to hypoglycemia). Time within the very tight range of 70-120 mg/dl was strong: 77% (type 1) and 58% (type 2 diabetes). There was no time <60 mg/dl. The mean glucose levels were 107 mg/dl (type 1) and 125 mg/dl (type 2). The total daily insulin doses of the subjects prior to participation in the trial ranged from 0.4 to 2.9 u/kg/day (>7-fold range), so the system was able to handle a large range of insulin sensitivities. The results demonstrated feasibility and were comparable to results of a recently published study from the Czech Republic using a different closed-loop system in cardiac ICU patients. The team is planning future studies of closed-loop control in the hospital ward and ICU environments. The goal is to have the technology to provide tight glycemic control without hypoglycemia so that the Leuven hypothesis can finally be rigorously tested.

-- by Melissa An, Adam Brown, Hannah Deming, Varun Iyengar, Emily Regier, Manu Venkat, and Kelly Close