November 30-December 4, 2015; Vancouver, Canada; Full Report – Diabetes Technology – Draft

Executive Highlights

This report includes our coverage of diabetes technology from IDF 2015. While it was a light meeting for new data or notable commentary (we’d love to see more in Abu Dabi in 2017!), we garnered particularly valuable insight on Abbott’s FreeStyle Libre in the company’s packed corporate symposium. European enthusiasm remains sky high for the 14-day factory calibrated sensor, which has also come to India in a blinded version.

The meeting also brought isolated learning on (i) closing the loop (NIH has funded the International Diabetes Closed Loop trial of a commercial-grade version of UVA’s DiAs system), (ii) the value of pumps over MDI (do we really know? We’d love to see more studies with CGM or Libre), and (iii) non-invasive glucose monitoring (what is the right tradeoff between usability and accuracy?).

Our themes are immediately below, followed by detailed discussion and commentary. Titles highlighted in blue are new additions that were not included in our daily updates, and those highlighted in yellow represent what we felt were the most notable talks of the meeting.

Themes

Continued Enthusiasm for FreeStyle Libre

• Enthusiasm for Abbott’s FreeStyle Libre has not abated after passing the one-year mark on the EU market. The corporate symposium stayed packed till the very end as Abbott shared new accuracy data in China, a reimbursement studies update, and real-world experience with Libre Pro in India. On the data front, Dr. Linong Ji (Peking University People’s Hospital, Beijing, China) shared new findings from a 40-patient, 14-day Chinese pivotal trial of Libre that demonstrated a MARD of just 10.7% vs. YSI, very consistent with the EU pivotal trial (MARD: 11.4%). It was a testament to this technology’s impressive design, though once manufacturing scales up as more countries come online, Abbott will of course need to maintain those results at scale – factory calibration is certainly not easy! As of Abbott’s 3Q15 call, capacity constraints persisted in Europe (patient want this technology!), though were expected to resolve soon. On the reimbursement front, we learned that the two six-month studies – REPLACE (n=210 type 2s on MDI, A1c>7.5%) and IMPACT (n=225 type 1s on MDI or pumps, A1c <7.5%) – will be presented at ATTD and ADA 2016, respectively. These studies were a very big investment and will be highly informative for obtaining broad reimbursement, especially in Europe where patients are less accustomed to paying out of pocket for healthcare.
• The legendary Dr. Viswanathan Mohan provided insight into the strong reception FreeStyle Libre Pro has seen in India – he has used the blinded 14-day sensor in ~1,000 patients in just nine months. Dr. Mohan spoke in glowing terms about the value in optimizing patients’ therapy by helping providers feel more successful and – most importantly – making the invisibility of diabetes more tangible to patients. One of Dr. Mohan’s most fascinating comments was that Libre Pro makes patients “trust SMBG more” because now they see single values in the context of broader patterns. The strong reception bodes well for the Libre Pro sensor that is expected to launch in the US next year (submitted in 2Q15).
• Management separately shared with us that the consumer version of FreeStyle Libre has been submitted to Canadian regulatory authorities. This is the first we had heard of this news. An approval timeline was not provided and we are still not sure how recently it was submitted. Still, it is big positive to see Abbott pushing its geographic footprint forward, and we hope to see the consumer version make its way to the FDA in the near future too. The news could also signal management’s confidence that the manufacturing capacity constraints will be rectified soon.

NIH Grant Will Fund International Diabetes Closed-Loop Trial

• IDF was lighter on closed-loop discussion than even EASD 2015, though a talk from Harvard’s Dr. Frank Doyle provided a major highlight: an NIH grant will fund the International Diabetes Closed Loop (IDCL) trial, a six-month, 240-patient study comparing a commercial-grade version of UVA’s DiAs to sensor-augmented pump therapy. It’s not clear, but it’s possible that the award represents the part of the NIH’s $20 million AP initiative that was not awarded this fall – as a reminder from DTM 2015, the winners were Cambridge ($6.4 million), the DREAM consortium ($2 million), and Boston University/Mass General ($1.5 million), accounting for half of the original allotment. It is not clear whether the group will use this study to support a regulatory submission of the DiAs system, though Dr. Doyle’s language – “to generate the safety and efficacy data needed to satisfy regulatory agency requirements and to demonstrate the clinical effectiveness to facilitate reimbursement” – absolutely hinted at this possibility. Dr. Doyle said the trial will begin in 2016 at ten international study sites – UVA, Harvard University, Mount Sinai School of Medicine, Mayo Clinic, Barbara Davis Diabetes Center, Stanford University, William Sansum Diabetes Center, Academic Medical Center Amsterdam, University of Montpelier, and University of Padova. We assume Dexcom CGM will be a key part of the system, though are not sure what pump will be used or where the control algorithm will reside. Presumably startup TypeZero (who aims to commercialize the DiAs system, per our June 2015 coverage) would be responsible for the FDA submission and commercialization, though that’s not clear either. To be clear, the consortium – as opposed to TypeZero itself – received the award. We’re very glad to see a long-term, large closed-loop study funded; this trial is both longer and larger than the US pivotal study of Medtronic’s MiniMed 670G (n=150, three months).
• The tough questions surrounding automated insulin delivery were not answered at IDF, though we look forward to meatier discussion at ATTD. What’s the appropriate control group in a pivotal study, if any (e.g., the 670G pivotal has no control group!)? What is the incremental benefit of glucagon over insulin alone?  How will patients trade off the level of automation (hybrid, full) vs. the level of glycemic control (low average, high time-in-range) vs. the device form factor (size, hassle factor)? New data is needed to advance the conversation on many of these issues, while others will be addressed once products get on the market. The field certainly seems increasingly comfortable with the idea that artificial pancreas systems are going to look and function differently and provide a spectrum of risk vs. benefit tradeoffs for different patients. No question about it – it will be an exciting couple of years for automated insulin delivery!

What Are the True Benefits of Pumps Over MDI?

• The incremental clinical benefit of pumps was questioned in REPOSE, a two-year randomized study comparing pumps to MDI in very poorly controlled type 1 diabetes (baseline A1c: ~9.2%). To address limitations of prior studies, education was held equal between the groups. At 24 months, pumps showed a very modest 0.2% A1c advantage over MDI (p=0.12) in those with a baseline A1c >7.5% and data from all four study visits (n=208 of the total n=267). Numerical A1c changes were not shown for all participants, but categorically, only 29% of the entire pump group and only 25% of the entire MDI group achieved an A1c <7.5% (p=0.5) – a testament to the poor control these patients were in at baseline (A1c: ~9.2%), and why neither education nor a pump is a magic bullet. Pumps offered no benefit on severe hypoglycemia (it was halved in both groups) and actually led to more DKA. It was disappointing to see so little benefit to pump therapy in this study, and unfortunately, SMBG or CGM data were not collected to supplement the A1c data. On the positive side, pumps did show significantly higher diabetes specific quality of life for diet restrictions and daily hassle, a finding Dr. Simon Heller called “real” and “sustained.” Dr. Heller was not negative on pump use, however – he concluded that care for adults with type 1 should start with MDI and structured training (e.g., DAFNE), with pumps offered later to those limited by MDI. In a follow-up conversation, Dr. Irl Hirsch pointed out a few important caveats to the data: the starting A1c’s were too high to expect a benefit, SMBG data was not collected at all, and based on the results, was not used effectively at the interim visits to titrate insulin.
• Still, the data prompt many important questions. Do we know the true clinical benefit of pumps? If not, how should a pump study be conducted to show the true benefits? Pumps are not like taking a drug – they are totally different therapy for patients AND clinicians. What is fair way to assess the incremental benefit of the technology over MDI? As basal and bolus insulins continue to improve, will the advantages of pump technology decline? How will patients and payers tradeoff CGM vs. pump use? What patients are most likely to benefit from pumps and how should we identify them? We strongly support offering pumps and CGM and closed-loop to those likely to benefit from them, but technologies will increasingly need to demonstrate their value to become widely adopted.

What’s the Bar for a Non-Invasive Glucose Monitor?

• Integrity introduced its non-invasive ear lobe clip device for spot monitoring glucose in type 2 and prediabetes. While the device itself was unimpressive, it did beg many thought provoking questions for the field: What level of glucose monitor accuracy is good enough for type 2 diabetes and prediabetes? How should companies trade off accuracy vs. improved form factor vs. cost? When does the output from a less invasive glucose monitor cross from useful to dangerous? (Integrity’s MARD is 23% - is that good enough for type 2?) These questions have important implications not just for Integrity, but type 2-focused glucose monitors from Abbott, Dexcom/Google, Medtronic, Novartis/Google, and wellness CGM companies like Echo and Sano. (Abbott is arguably the leader right now on the composite profile, having achieved a great form factor with FreeStyle Libre, a lower cost, and accuracy safe enough to dose insulin with factory calibration. But there is clearly a spectrum on the cost, accuracy, and form factor dimensions of any product.) Not every glucose-monitoring device needs to be accurate enough for dosing insulin, but there is presumably some threshold that must be hit. What is that threshold, and how should that be weighed against a smaller or less invasive or less costly device in different patient populations? For instance, if the Novartis/Google contact lens has a MARD of 20% (total speculation on our part), is that good enough to drive therapeutic change in type 2 diabetes and prediabetes? At what point does the device’s clinical inaccuracy outweigh any potential benefit of more frequent monitoring? Tough questions indeed, and ones that will likely be solved only with prototypes tested in clinical trials. We left this session with low overall confidence in Integrity’s product, though perhaps the company can improve with future generations (more details below).

Where is Digital Health?

• Hearing about global disparities in care, we continue to wonder if digital health – particularly text messaging – can scale to meet a wide variety of challenges in diabetes: growing patients, a shortage of providers, and limited access to care. IDF did not share major new data on this front, though we appreciated Dr. Ambady Ramachandran’s overview of using text messaging to potentially prevent diabetes. His lab’s ongoing two-year study includes 1,050 patients in India and the UK, with one group receiving three personalized lifestyle modification texts per week. Results are currently pending (estimated completion this month)! His talk also reviewed a previous study published in Lancet in 2013, demonstrating a 36% risk reduction in the incidence of type 2 diabetes in the intervention group (text messages; n=271) compared to control (usual care; n=266). The number needed to treat to prevent one case of type 2 diabetes was only 11 patients! We hope to see more focus on scaling behavior change through interventions like text messaging – the low cost, low side effects, and high scalability seem like winners all around. A downside, of course, is patient annoyance factor, and hopefully smart phones can take text messaging to the next level with more personalized and interactive content.

• Who is doing the most exciting and scalable things in type 2 diabetes prevention and care? Will Omada Health eventually scale to help millions? Why hasn’t WellDoc’s BlueStar – which started as a text messaging intervention – scaled faster in its rollout in the US? How can established diabetes drug and device companies help bring low-cost, proven diabetes prevention and lifestyle change technology to market? We continue to hear about isolated attempts to change behavior in diabetes, but the ecosystem is arguably fragmented and siloed – where can collaborations form to change things at a population level?

Detailed Discussion and Commentary

Corporate Symposium: One Year of Flash Glucose Monitoring: The Global Clinical Experience (Sponsored by Abbott)

FreeStyle Libre Accuracy China Study

Linong Ji, MD (Peking University People’s Hospital, Beijing, China)

Dr. Linong Ji shared never-before-seen data from Abbott’s 40-patient, 14-day Chinese pivotal trial of its factory calibrated FreeStyle Libre system, which demonstrated excellent accuracy vs.  both YSI and capillary fingersticks. MARD was just 10.0% vs. YSI (the study had 6,969 paired sensor-YSI points) and 10.7% vs. capillary fingersticks (the study had 6,696 paired sensor-YSI points). The data were impressively consistent across the board as 87% of points were in Zone A of the Consensus Error Grid and 13% in Zone B in both environments, a testament to the device’s accuracy as patients would experience it (i.e., relative to fingersticks). Similar to the EU pivotal trial, accuracy did not significantly deteriorate between days 1-7 vs. days 8-14. The data actually represent a slight improvement on what has been achieved in the EU pivotal trial, where Libre demonstrated an overall MARD of 11.4% vs. FreeStyle Precision BGM (n=13,195 paired points) and 11.8% vs. YSI (n=1,238 paired sensor-YSI points). The overall accuracy in these clinical trials is highly encouraging, and once manufacturing scales up, Abbott will need to maintain those results at scale – we have little doubt on that front, but factory calibration is of course not easy. As a final note, Dr. Ji did not share a timeline on China approval of Libre (or whether the device has yet been submitted to regulatory authorities).

• In a follow-up conversation with management, we learned that FreeStyle Libre (the consumer version) has been submitted to Canadian regulatory authorities. This is the first we had heard of this news publically. A timeline was not provided and we are still not sure when this was submitted. It is positive to see Abbott pushing its pipeline forward, and we hope to see the consumer version make its way to the FDA in the near future. As a reminder, the blinded Pro version is currently under FDA review, with a potential launch in 2016.
• The Chinese pivotal study was conducted similarly to the US studies with three centers across China in type 1 and type 2 patients on insulin therapy. Patients wore two sensors on the back of their arm for 14 days and were asked to: (i) perform eight capillary blood glucose tests daily; (ii) scan the sensor following each test; and (iii) attend three in-clinic eight-hour YSI sessions. Notably, Dr. Ji shared that patients had a mean baseline A1c = 8.6%, ranging from a low of 5.6% to a high of 13.4% for a nice mix of well-controlled and out-of-control patients.
• There was one rather significant difference between the China and US data – the mean lag time was 3.1 minutes in China vs. 4.5 minutes stateside. Dr. Ji could not explain the discrepancy though he hypothesized that it may reflect ethnic differences. Indeed, it’s tough to say what’s causing the difference. Sample size? A different batch of devices?
• MARD was not broken down by glucose range, so accuracy in hypoglycemia is an unanswered question for the Chinese study. Just like the EU, we assume the product label will recommend a confirmatory fingerstick when patients are hypoglycemic.  Still, patients in the real world clearly trust the device enough to never test, so we have little concern about the hypoglycemia data in this study.
• Dr. Ji shared positive data from user experience studies of FreeStyle Libre in the pivotal study. There was no background on how these questions were asked – we assume Yes/No. Said Dr. Ji jokingly, “This is probably due to the fact that I treat them too well.” Jesting aside, the data do point to why European patient uptake has been so strong in these early days, especially in those that have avoided current CGM due to comfort/wearability:
  
  • 100% agreed that the sensor was easy to apply.
  • 100% agreed that applying the sensor was less painful than a routine fingerstick.
  • 100% agreed that the system was packaged with sufficient information to use the system.
  • 100% agreed that the amount of pain felt when applying the sensor to the arm was acceptable.
  • 100% agreed that the amount of bleeding experienced when applying the sensor to the arm was acceptable.
  • 100% agreed that the sensor was comfortable to wear.
  • 97.5% agreed that the sensor was easy to wear due to its small size.
  • 100% agreed that the sensor did not get in the way of daily activities.
  • 100% agreed that scanning the sensor is less painful than pricking a finger.
  • 100% agreed that scanning the sensor is easier than pricking a finger.
  • 100% agreed that the amount of erythema on the arm when the sensor was removed was acceptable.
  • 100% agreed that the amount of edema on the arm when the sensor was removed was acceptable.
• Very few adverse events were reported among patients in the study. Only 9 subjects reported any sort of discomfort related to the sensor insertion and all reports were consistent with what would be expected following insertion of a sensor into the skin. There were no severe adverse events reported and all patients were able to see the trial to conclusion:
  
  • Moderate erythema – 6 patients
  • Bruising – 2 patients
  • Mild pain – 1 patient
• Though not indicated for long-term or deep underwater wear, commentary during Q&A suggested that Libre has been used accurately by divers for up to an hour and to a depth of 30 meters. Dr. Genovese summed up his personal experience with such off-label wear quite succinctly: “It works perfectly!”
• The Chinese Clinical Trial Registry (ChiCTR-OPC-15006147) for the study was last updated in June: A Multicenter Clinical Accuracy Evaluation of the Abbott Sensor Based Interstitial Glucose Monitoring System in Chinese Diabetes Subjects.

Professional Use of FreeStyle Libre Pro: The Clinical Experience in India

Viswanathan Mohan, MD, PhD (Dr. Mohan’s Diabetes Specialties Center, Chennai, India)

A series of case studies presented by Dr. Mohan provided insight into the incredible reception FreeStyle Libre Pro has seen in India. We were floored by his estimate that he has already used the blinded 14-day sensor in ~1,000 patients (whoa!) in just nine months. He spoke in glowing terms about the pattern recognition in Abbott’s Ambulatory Glucose Profile, the value in optimizing patients’ therapy by helping providers feel more successful, and – most importantly – making the invisibility of diabetes more tangible to patients. It sounds like the device has been used far more often in type 2 patients, though Dr. Mohan suggested that the device holds value for anyone on insulin. Especially for patients in India who cannot afford the technology themselves, he impressed upon the audience that two weeks of wear provides a valuable retrospective look to identify trends and can even inform smarter allocation of limited fingersticks. Indeed, we thought that one of Dr. Mohan’s most fascinating comments during Q&A was that Libre Pro makes patients “trust SMBG more” (because now they see single values in the context of broader patterns!). Ultimately, the spectacular reception bodes very well for the Libre Pro sensor that is expected to launch in the US next year (submitted in 2Q15). Abbott did not comment on this timeline (though they were asked) and we assume this is still on track.

• Although Libre Pro has only approved in adults, commentary during Q&A suggested that providers in India are even taking liberties with the sensor in children. The off-label use does not appear to be affecting accuracy in the slightest. Said Dr. Mohan, “We’ve learned a lot of things. I know it’s not approved, but talk about clinical value … We do it!”
• As a reminder, Medtronic announced the launch of CareLink iPro Pattern Snapshot yesterday, a new one-page download report for the blinded iPro2 professional CGM. The report prominently shows a patient’s top three glucose patterns and lists up to six possible causes for each one. We do think FreeStyle Libre Pro’s 14-day wear, factory calibration, and slim form factor is a big step up over iPro2’s three-day wear, the need to retrofit the raw sensor data to fingerstick readings, and the larger transmitter.

Flash Glucose Monitoring: Use of AGP in Clinical Practice in the EU

Stefano Genovese, MD (IRCCS MultiMedica, Sesto San Giovanni, Italy)

Dr. Stefano Genovese saved the best for last as the major highlight of his presentation came in his closing sentence with an update on Abbott’s two six-month FreeStyle Libre reimbursement studies – REPLACE (n=210 type 2s on MDI, A1c>7.5%) will be presented at ATTD 2016 and IMPACT (n=225 type 1s on MDI or pumps, A1c <7.5%) will be presented at ADA 2016. Both studies are listed as “ongoing and not currently recruiting” participants on ClinicalTrials.gov with completion slated for December 2015 and September 2015, respectively. As a reminder, the goal of the type 2 study is to show a change in A1c at six months, while the type 1 study seeks to reduce time spent in hypoglycemia at six months. We can hardly express how much we are looking forward this data as reimbursement is critical around the globe, especially in Europe where patients are less accustomed to paying out of pocket for healthcare. [Libre currently has partial reimbursement only in Sweden.] If these studies report positive results and Abbott obtains widespread reimbursement, the company will have its hands full meeting demand.

• The bulk of Dr. Genovese talk presented the FreeStyle Libre and its accompanying Ambulatory Glucose Profile as clinical tools that can facilitate patient education, improve pattern management, and reduce glycemic variability. Drawing from multiple studies, Dr. Genovese cited glycemic excursions as a driver of microvascular complications and mortality due to the associated oxidative stress. In order to reverse this trend, Dr. Genovese espoused more frequent glucose monitoring – he shared that less than 40% of patients on SMBG in Italy check their glucose more than four times per day whereas ~90% on Libre scan their glucose>four times per day.

FreeStyle Libre Accuracy Study

Timothy Bailey, MD (UCSD, San Diego, CA)

Dr. Timothy Bailey reprised the topline results from Abbott’s 72-patient, 14-day CE Mark pivotal trial – first shared at EASD 2014 – which demonstrated an overall MARD of 11.4% vs. FreeStyle Precision capillary fingersticks (87% of points were in Zone A of the Consensus Error Grid, 13% in Zone B). He opened with a short and sweet introduction to FreeStyle Libre, leading awe-struck attendees through the nuances of flash glucose monitoring: 14-day wear, factory calibrated, approved for wear on the upper arm, etc. Indeed, audiences continue to flock to Libre symposia and this was no exception – the room was packed till the very end! In terms of material itself, the vast majority of Dr. Bailey’s presentation echoed his previous talks on the same topic (ATTD 2015).

Panel Discussion

Q: This is a major paradigm shift to have Libre available. I think the presenters were highly complementary. Can you talk quickly about the various error grids that are used to assess the results?

Dr. Timothy Bailey (UCSD, San Diego, CA): I hate error grids. But they’ve proved to be very useful from a regulatory point of view. They were invented because most glucose monitors were not accurate and it was decided that meters that made less egregious errors were better. However, now in the US, almost 99% of readings are in Zone A and B, so it’s not very useful for comparing technologies.

Q: Bananas and other fruits can cause errors in BGM readings? Do you think that was why there was a higher MARD for Libre vs. capillary?

Dr. Linong Ji (Peking University People’s Hospital, Beijing, China): Yes. Ultimately, we should not interpret just one point though. I like the term “context.” We should use the collective data set to make a clinical judgment.

Dr. Bailey: It’s a good point about bananas and other foods that can interfere with readings. We instructed participants in our trials to do accurate readings. That’s why in reality SMBG may underperform and that’s why the SMBG is different.

Q: Can you talk about the availability and timeline for Libre in Canada and the US?

Mr. Steve Scott (Abbott Diabetes Care, Alameda, CA): The answer to that we’re in the process of trying to get approval for the device in the US, and we’ve submitted the device to Health Canada and can’t really talk about more than that.

Q: Is Libre really CGM in disguise?

Dr. Stefano Genovese (IRCCS MultiMedica, Sesto San Giovanni, Italy): I think it was a good idea of Abbott to design Libre this way. It is a CGM because the sensor continues to measure glucose. However, it is also called “Flash” because the measurement and the possibility to see the measurement is on request. But the actual fact is that the sensor continues to measure everything. It’s the way you use it. The device doesn’t have alarms and this is a problem with CGM, because patients complain about CGM alarms during the night. In the end, Libre gives you all the information a CGM gives you.

Dr. Viswanathan Mohan (Dr. Mohan’s Diabetes Specialties Center, Chennai, India): As far as India is concerned, it is a CGM because we don’t give patients the real-time monitoring capability with the Pro version. We do have Medtronic’s consumer CGM available to patients, but a big difference is the cost. Libre is also less painful and more convenient. No calibrations are VERY convenient. So ultimately, getting Libre for 14 days at lower cost with much less pain is wonderful for us. That’ s why this system has taken off in this country.

Dr. Bailey: I can’t prescribe Libre, but my savvy patients ask about it. SMBG is universally strongly disliked. The big difference is the factory calibration and no fingersticks. I would summarize by saying that CGM and Libre are different technologies, and I have patients that like CGM because it saves them from hypoglycemia at night. But I would say that that’s the minority of patients.

Dr. Mohan: We’ve been saying that Libre replaces SMBG. I have a slightly different take. For us, it doesn’t replace SMBG because we don’t give patients the Pro week after week. That would be too expensive. Instead, we tell them what the problems are and then we tell them when to do SMBG. Most of our patients are type 2s and we tell them that these are the points when you should check – 3 AM, after dinner, their trouble spots, etc. Now patients trust the SMBG a little more and I think Libre and SMBG can be kind of complementary.

Q: Can you talk about the patient experience with and feedback that you’ve received in practice? Are patients seeing a change in lifestyle and outcomes as measured by A1c?

Dr. Genovese: To be honest, I wanted to present some data but I cannot talk about that in a scientific symposium. I can tell you my opinion: the quality of life has DEFINITELY increased. Libre has been used and quality of life has improved a lot. In particular, the number of patients measuring and evaluating their blood glucose more than four times a day has increased to 90% in Italy vs. less than 40% who measure four times a day on SMBG. Libre is increasing the quality of life and changing how patients think about diabetes. I used to say that with SMBG we have a measure of glucose at one point in time. With this approach, you have the trend and the movement of many variables. Patients have much better knowledge about their condition.

Q: Can you talk about the cost in the EU?

Dr. Genovese: The starter kit is something like 150 euros, and it contains the reader plus two sensors. Each sensor is 50 euros, individually. Overall, this comes out to about 4 euros per day. So the cost is higher than SMBG but not so high.

Q: Can a person with an American credit card buy it?

Dr. Genoese: If you go to the website from a particular country, it goes to the country’s website. Another way is if you have a friend in EU.

Mr. Scott: The product is only for sale in those countries and only for use by residents in those countries.

Q: Can you clarify how many times the sensor actually measures glucose?

Mr. Scott: The system measures glucose every minute. Because of limited storage on board, what is put into memory on the body is one data point every 15 minutes. However, the ability to scan and get a reading is available every minute.

Q: Dr. Genovese, have your type 1 patients seen a decrease in the incidence of hypoglycemia?

Dr. Genovese: We are collecting the data. The primary endpoint of the IMPACT study is the time in hypoglycemia. Our expectation is that patients using Libre have fewer hypoglycemic events than those using SMBG because they check their blood glucose far more often. I cannot give you the data in a formal way because we are collecting the data still.

Q: What patients are best for Libre Pro vs. the consumer version?

Dr. Mohan: We have only Libre Pro in India. Certainly, I think it is better in type 2s. In type 1s, I think you need the consumer version because retrospective data itself isn’t as helpful. With a type 1 patient, getting a two-week pattern can be useful but it can also be useless since sometimes there is so much fluctuation that it is quite maddening. For type 2s though, it is fantastic. The majority of our patients have type 2, so it’s fantastic in India.

Q: It’s always really nice to have another tool in the toolbox. Has anyone had any experience with this device in children or women in pregnancy?

Dr. Mohan: Abbott asked me not to talk about this because it’s not approved. But yes, we have used Libre in kids. We did it in a newly diagnosed child recently who travelled to see me from 200 miles away. I was able to identify a lot of hypoglycemia because of Libre and we reduced the night dose. We did a second AGP a few months later. We’ve learned a lot of things. I know it’s not approved, but in terms of clinical value … We do it!

Dr. Genovese: Abbott is doing studies, rest assured. On the use of these instruments, I just want to give an example. Libre is indicated in water for no more than one hour at a depth of no more than one meter. I have a good friend who is type 1 in Italy who loves to diving and he uses Libre diving for one hour at depth of 30 meters. It works perfectly.

Q: Is Libre covered by insurance companies where it is available?

Mr. Scott: Libre has limited reimbursement in Sweden. It is our intention to make the device available to a larger population.

Q: What happens at 14 days?

Mr. Scott: The device ceases to function at 14 days.

Symposium: Closed-loop Devices for Treating Diabetes: What We Have and What We Need

Personalizing the Artificial Pancreas for the Outpatient Setting

Frank Doyle, PhD (Harvard, Cambridge, MA)

Dr. Frank Doyle’s comprehensive overview of closed-loop algorithms was highlighted by big news – the International Diabetes Closed Loop (IDCL) consortium has received an NIH grant to fund its IDCL trial, a six-month, 240-patient study comparing a commercial-grade version of UVA’s DiAs to sensor-augmented pump therapy. It’s not clear, but it’s possible that the award represents the part of the NIH’s $20 million AP initiative that was not awarded this fall – as a reminder from DTM 2015, the winners were Cambridge ($6.4 million), the DREAM consortium ($2 million), and Boston University/Mass General ($1.5 million), accounting for half of the original allotment. (The UC4 AP grant has also been reissued, with award winners to be notified in late 2016.) We do stress that this is speculation on our part and that other grant awardees (or funds allocated) were not discussed. Of course, the DiAs consortium was always a strong candidate, given their experience and encouraging results to date. It is not clear whether the group will use this study to support a regulatory submission, though Dr. Doyle’s language – “to generate the safety and efficacy data needed to satisfy regulatory agency requirements and to demonstrate the clinical effectiveness to facilitate reimbursement” – absolutely hinted at this possibility. We assume Dexcom CGM will be a key part of the system, though are not sure what pump will be used or where the control algorithm will reside. Presumably TypeZero would be responsible for the FDA submission and commercialization, though we’re not sure. To be clear, the consortium - as opposed to TypeZero itself - received the award. Dr. Doyle shared that the trial is slated to begin in 2016 at ten international study sites – UVA, Harvard University, Mount Sinai School of Medicine, Mayo Clinic, Barbara Davis Diabetes Center, Stanford University, William Sansum Diabetes Center, Academic Medical Center Amsterdam, University of Montpelier, and University of Padova –putting a PMA submission in winter 2016/2017 at the very earliest. The news adds to what is shaping up to be an exciting couple of years for commercialized artificial pancreas systems, led by Medtronic’s MiniMed 670G (in a pivotal study slated to wrap up by May 2016)!

• Dr. Doyle opened his presentation by reminding the audience of the exponential acceleration in artificial pancreas research over the past ten years, citing data from the AP Clinical Trial Database. We’ve seen this before, though were still impressed by the thoroughness of the database (which acts as a store of artificial pancreas clinical studies). The database contains information on trials spanning 11 years (2004-2015) and was established by searching PubMed and Web of Science with a number of keywords including: (i) Closed-loop diabetes; (ii) Artificial pancreas; (iii) AP; and (iv) Bionic Pancreas.
• The bulk of Dr. Doyle’s presentation described UCSB’s Model Predicted Control algorithm. It was quite math heavy – often over our heads! – though it was clear that the group is moving quickly and rigorously toward safer and more effective systems. This group (like many others) has its eyes clearly on the prize: a real-world system that maximizes safety and efficiency that can be personalized to individuals.

Questions and Answers

Dr. Robert Eckel (University of Colorado, Aurora, CO): I’m interested in UCSB’s model in terms of the insulin on board. Is that based on patient data?

A: We’re using some data from broad standards of care but the algorithm is not based on patient data. However, that’s a thing we could do.

Q: Is there still a place for glucagon in closed-loop systems?

A: If we go to a longer horizon, I think questions about amylin and glucagon will be in the forefront of our minds. Our lab has focused on insulin, but there is work being done by Dr. Ken Ward (OHSU, Portland, OR) and Dr. Ed Damiano (Boston University, MA). I think this a long-term question.

Dr. Irl Hirsch (University of Washington, Seattle, WA): This insulin on board business is a big deal. Personalization is critical because you have to account for variability in a number of factors, such as when the pump catheter was changed. After all, the set is going to deliver insulin differently on day one vs. day three.

A: Without a doubt, these variables are important. In your example, what we would need is a way to quantitatively correlate the day post-insertion with insulin delivery and absorption. If we had a way to quantify that, putting it in the algorithm is very simply.

Q: When will closed-loop systems reach the market? 

A: When we started working on this, people said they were five to ten years out. In the last five years, we’ve continued to say that they are five years out. From talking to companies, I think we’re seeing parts of artificial pancreas systems creep into practice – such as low glucose suspend – and these are rudimentary steps toward an artificial pancreas. If you’re talking about the full system, I’d still say we’re within five years of getting that to market. We’re testing prototypes and getting closer. There is not going to be one solution. There are going to be different solutions from different companies.

New Sensors for Closed-loop Systems

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

Dr. Jeffrey Joseph gave an enthusiastic overview of glucose sensors, sharing notable optimism on the recent progress (particularly Abbott’s FreeStyle Libre) and disclosing a new startup (Capillary Biomedical) is working on a long-term implantable glucose sensor for the artificial pancreas. He was positive on all currently available and pipeline sensors, though his optimism on FreeStyle Libre stood out – Dr. Joseph called the 11.4% MARD, 14-day factory calibration, and stable sensor accuracy “very exciting” and “amazing.” He acknowledged that the device is not real time (i.e., not for closing the loop), but said, “I’m sure this technology will progress.” Indeed, we wonder if Abbott plans to offer a real-time version at some point down the road. Dr. Joseph optimistically pointed out several companies’ emerging technologies, though didn’t comment on their commercial potential or timing. He devoted slides to implantable CGMs from GlySure (in human studies), Senseonics (90-day data complete; EU launch by end of 2015), BioRasis (UConn, long-term implantable), Microchips (long term implantable), and a notable new company called Capillary Biomedical (more below).

• Dr. Joseph disclosed that a new startup, Capillary Biomedical, is working on a multi-year implantable optical glucose sensor for the artificial pancreas. The idea leverages original work done at Animas in 1995 on an implantable glucose sensor that optically measured glucose in blood flowing through an artery. The Capillary Biomedical idea takes this concept a step further, measuring glucose in an ultra-filtrate of interstitial fluid drawn from the surrounding tissue. Micro-pumps pull tissue fluid through a network of soft, flexible catheters with micro-porous membranes to create the ultra-filtrate. The starfish-looking device appeared to have several sensors for redundancy. Dr. Joseph did not share any timing, expected accuracy, or clinical trial plans.
  
  • Capillary Biomedical’s Angel List web page has more details on the idea. Glucose is determined via infrared spectrometry. A physician implants the device under the skin of the abdomen in the subcutaneous tissue. The implant requires an outpatient surgery that uses local anesthesia. The device will send real-time glucose information wirelessly to an external receiver, mobile phone, and insulin pump. The implanted system only requires periodic inductive charging. We’re not sure if it would be factory calibrated or how large the implant would be.
  • The Angel List page notes that Paul Strasma is CEO, formerly VP Marketing & Clinical Trials at GluMetrics. Dr. Joseph is the only other listed employee, Vice-Chairman & Director of Research. The company has not posted any job on its Angel List page. Based on its billboard webpage, it is located in Irvine, CA and has been around since 2014. The LinkedIn page notes that the  “novel approach aims to overcome the twin challenges of sensor degradation and tissue encapsulation.”
• More broadly, Dr. Joseph noted that current sensors from Abbott, Dexcom, and Medtronic are in the 9-14% MARD range, and importantly for closing the loop, produce fewer large errors. He said the “magic number” for safely closing the loop is a MARD <10%. The “major” recent advance, according to Dr. Joseph, has come in the software realm, where algorithms compensate for sensor drift, detect sensor instability or failure, and optimize calibration. We would point out Dexcom’s work with Dr. Claudio Cobelli’s team as the standout example there, which brought the G4 Platinum’s MARD down to 9.0% with Software 505/G4AP.

Questions and Answers

Dr. Irl Hirsch (University of Washington, Seattle, WA): What about scarring and damage from long-term infusion set wear?

Dr. Joseph: It is a real problem. The damage is likely not from insulin infusion, but from tissue damage when the set is inserted, and then damage from movement. We need to address this.

Adam Brown (Close Concerns, San Francisco, CA): Why hasn’t hospital CGM moved faster? Edwards is out of the game and the Medtronic Sentrino hasn’t done much in Europe. Is this a technological problem or a commercial problem or both?

Dr. Joseph: Hospital administrators want to see a cost-benefit to using the technology. I’m convinced there is a clinical benefit, but there has to be a cost-benefit. In addition, no one knows what to do with the data. The alarm goes off, the nurses does something, but there is not a protocol that has been validated. I think the future is in-hospital closed-loop. But companies have to be convinced they will make money if they invest the resources.

New Insulins and Insulin-Delivery Technologies for Closed-Loop Systems

David Klonoff, MD (Mills-Peninsula Health Services, San Mateo, CA)

Dr. David Klonoff shared a comprehensive overview of new insulins and insulin-delivery offerings for closed-loop systems. He stressed that new insulins or insulin-delivery technologies could revolutionize closed-loop control by both improving glycemic control (greater time-in-range, less hypoglycemia, lower mean glucose levels) and increasing patient convenience. What products are available or soon-to-be available? See below. Dr. Klonoff also discussed Afrezza (seems to be less potent than indicated) and shared optimism for hybrid closed loop systems. His talk reminded us that a lot of things are happening in faster insulin. The bigger question is how much incremental benefit these products will show, and whether they will get paid for.

• In discussing the utility of Afrezza in the real world, Dr. Klonoff echoed commentary we’ve heard from other KOLs that the formulation often feels slightly less potent than indicated. As a reminder, Afrezza is available in 4-, 8-, and 12-unit doses though Dr. Klonoff reported that patients seem to respond more like the doses are in 3-, 6-, and 9- units increments. Overall, he was very positive on Afrezza’s efficacy, explaining that its faster onset of action compared to rapid-acting analogs should be a significant advantage as an adjunct to hybrid closed-loop systems.
• Dr. Klonoff expressed a great deal of optimism about hybrid closed-loop systems despite the fact that partial meal announcements are necessary. Indeed, he cited compelling data that hybrid options can provide very effective meal coverage even while insulin action remains slow. He also acknowledged that different patients will view the benefit/hassle tradeoff differently, appreciating that what is tolerable for some may not be tolerable for others. We fully agree that the question of fully automated vs. daytime hybrid vs. nighttime-only systems is one of patient preference – more on that in diaTribe. Ultimately, we believe a range of options is a good thing for people with diabetes, since all systems and products have pros and cons.

Dr. Klonoff’s New Insulin Landscape Overview

Dr. Klonoff’s New Insulin Delivery Technology Landscape Overview

Symposium: Late-Breaking CSII vs. MDI study – The REPOSE Trial

Late-Breaking CSII vs. MDI Study: The REPOSE Trial

Simon Heller, MD (University of Sheffield, UK)

Dr. Simon Heller presented late-breaking results from a two-year randomized, parallel group study comparing pumps to MDI in very poorly controlled type 1 diabetes (baseline A1c: ~9.2%), holding education equal between the groups. Both the pump (n=132) and MDI (n=132) groups completed the DAFNE course (dose adjustment for normal eating) at baseline, and were randomized to use a pump (Medtronic) or MDI (Levemir and aspart) over 24 months (median age: 41 years). After adjusting for center, DAFNE course, and baseline A1c, pumps showed a modest 0.2% A1c advantage over MDI (p=0.12) in those with a baseline A1c >7.5% and data from all four study visits (n=208 of the total n=267). Numerical A1c changes were not shown for all participants, but categorically, only 29% of the pump group and only 25% of the MDI group achieved an A1c <7.5% (p=0.5). Pumps fared slightly better in a per-protocol analysis, showing a significant (but modest) 0.3% A1c advantage over MDI (p=0.02). DKA was much more common with pumps – 17 events vs. 5 events (13% vs. 4%) – though that difference was confined to the first year of the study, and most DKAs were due to infections (only 18% from set failure in pumpers). The combined rate of severe hypoglycemia was halved overall, though there was no significant difference between pumps and MDI (a testament to the efficacy of DAFNE). Pumps showed significantly higher diabetes specific quality of life for diet restrictions and daily hassle, a finding Dr. Heller called “real” and “sustained.” Dr. Heller concluded that care for adults with type 1 should start with MDI and structured training (e.g., DAFNE), with pumps offered later to those limited by MDI. It’s good to see a long-term, comparative, well-controlled study of pumps vs. MDI (holding education constant), though in speaking to Dr. Irl Hirsch, this study had some important caveats: the starting A1c’s were too high to expect a benefit, SMBG data was not collected at all, and based on the results, was not used effectively at the interim visits to titrate insulin (see our Q&A with Dr. Hirsch below, followed by Dr. Heller’s response).

Q&A with Dr. Irl Hirsch

ADAM BROWN: Dr. Hirsch, what do you think of this study? Would you have designed it differently?

DR. IRL HIRSCH: I commend them for a well-controlled study that ensured equal education at baseline. But they did not do the study appropriately: they studied insulin delivery and left out glucose monitoring. You cannot uncouple those too! We don’t know what happened after the study started with the interim visits – there were no SMBG downloads. How do you make insulin adjustments without glucose data for either MDI or CSII?

In the US, we would download the glucose meter, download the pump, look at the frequency of testing, make suggestions on when glucose testing should be increased, and make adjustments in insulin based on glucose levels. Dr. Heller gave no information on this. My guess is glucose testing was the same in both groups, and it was probably low in both groups, based on our data from the T1D Exchange and STAR-1.

With CGM, we’ve learned that glucose data is the most important thing. Even with SMBG, our first paper in the T1D Exchange showed a relationship between A1c and glucose testing – and we showed that in every age group. If you’re only testing twice per day, you’re not going to do very well.

ADAM: This was a pretty tough group with a mean A1c over 9%. Any thoughts on that inclusion decision?

DR. HIRSCH: If you’re using A1c as a primary endpoint in an adult type 1 diabetes study, don’t include people with A1cs over 9%. They’re not doing the minimum – they are testing infrequently (maybe twice a day) and they’re missing insulin more often. That is why I believe they saw more DKA in the pump group. We learned this the hard way in STAR-1. CGM significantly reduced A1c by 1% for everybody except those with an A1c >9%. They recommended five SMBG tests per day but based on our previous studies it is doubtful this was the case here. The ideal situation would have been documentation of equivalent SMBG in both groups, and then make conclusions about insulin delivery methods.

ADAM: What are the implications of this data?

DR. HIRSCH: My concern is payers are going to look at this and say, “We shouldn’t cover pumps.” Hopefully, the paper, when published, will note the limitations.

ADAM: What studies are we missing in the pump vs. MDI debate?

Dr. HIRSCH: My biggest concern with pumps and closed loop is that pumpers are running out of on-body real estate due to scarring, lipohypertrophy, etc. None of this may matter if we don’t figure out how to reduce that.

Response From Dr. Heller

Dr. Heller: I agree with Irl’s interpretation of our data, but I would challenge strongly his view that these are limitations. The point of the research question is whether poorly controlled adults should go straight onto pumps or whether they should have proper training in the use of insulin.  Then for those who subsequently go onto self-manage their diabetes effectively and find the limitations of MDI prevent them getting to their glucose goals and impair their quality of life, pumps should be generally available – indeed they need to be strongly encouraged to take advantage of such technology.  I believe our data strongly support such a pathway ,which actually could ease the reimbursement barriers to pump therapy in many countries – the people who would benefit the most would be provided with a pump. In summary, our study results argue strongly for much better skills training for all people with type 1 diabetes followed by the availability of pumps for those who experience limitations.

Other Study Details and Outcomes

• Patients had a median age of 41 years, were primarily white, had a median diabetes duration of ~15 years, a mean BMI of 27 kg/m², and more than half (~55%) had microvascular complications. In short, this was a tough group! Importantly, patients could only participate in the study if they had no strong preference for pump therapy vs. MDI. Those who had a clinical indication for pump (e.g., hypoglycemia problems) were excluded from the study.
• Visits occurred at baseline, six months, 12 months, and 24 months. There were no specifics on what these entailed. The DAFNE education program was only completed at baseline and included three courses with 5-8 participants in each course.
• In the unadjusted results, pumps showed a 0.4% A1c advantage, presumably because of the higher baseline A1c (-0.8% vs. -0.4%; baseline: 9.3% vs. 9.0%). As noted above, after adjusting for baseline A1c, center, and DAFNE course, pumps showed a more modest 0.2% A1c advantage over MDI (p=0.12).
• The per-protocol analysis excluded 11 participants that switched from pumps to MDI and seven participants that switched from MDI to pumps. We assume these switches occurred during the trial by patient choice, but it was not specified. The 11 patients that switched from pumps to MDI saw a 0.7% improvement in A1c (10.4% to 9.7% at 24 months), while the seven that switched from MDI to pumps saw a striking 2.0% reduction in A1c at 24 months (10.3% to 8.3%). That could suggest that when patients drive the choice to switch to a pump, the clinical benefit can be much bigger.
• Dr. Heller was very positive on structured training with DAFNE, noting that it reduces the risk of severe hypoglycemia and leads to modest but long-lasting benefits in A1c (“it should be delivered widely”). DAFNE teaches patients flexible intensive insulin therapy to match insulin doses to foods. NICE supports DAFNE but very few adults in the UK go through the program. This is not surprising to us – DAFNE requires a five-day training course (9am-5pm; Monday to Friday) and is only delivered in person at 60+ UK centers. It is a downside that it is not offered online and we wonder how it can be scaled and made much more convenient.
  
  • Structured education is listed as a priority for implementation in the new NICE guidelines released yesterday. More details on DAFNE are here, and the new NICE guidelines are posted here. 
• We certainly agree with Dr. Heller’s conclusion that further work is needed to support people in achieving tighter glucose targets. For those in poor control, we need better strategies! The cost-effectiveness of improving glycemia in this group is high – what are the highest ROI interventions that could do so?
• Said Dr. Heller, “In the UK it is estimated that 6% of adults with type 1 diabetes use pumps compared to 40% in the US.” Dr. Heller did not source these estimates or attempt to explain the discrepancy, though presumably it stems from NICE’s stricter reimbursement criteria for pumps.
• Why conduct REPOSE? According to Dr. Heller, existing evidence on pumps vs. MDI can be biased, as those allocated to the pump may receive more training and attention vs. those on MDI. This study included comparable training for both groups, something few trials have done. The results are consistent with previous meta-analyses showing an ~0.3% benefit of pumps over MDI (Misso et al., Cochrane Review 2010).
  
  • How should a pump study be conducted to show the true benefits of the technology? Pumps are not like taking a drug – they are totally different therapy for patients and clinicians. What is fair way to assess the incremental benefit of the technology?

Questions and Answers

Dr. Bernie Zinman (Mount Sinai Hospital, Toronto): Really nice study. Can you tell us about the number on CGM, or were they not allowed to wear CGM?

Dr. Heller: We had no specific policy on CGM.

Dr. Zinman: We did a similar study comparing MDI with pumps. Those that started with a high baseline tended to get a better benefit of pumps than MDI. Did you find that too?

Dr. Heller: We are currently undertaking that analysis. To date, we haven’t seen that effect.

Dr. Irl Hirsch: Very impressive study. You noted the caveats very well. My question has to do with the interim visits. Specifically, what happened with the glucose data – was the data downloaded? Especially with higher DKA in the pump group and not using CGM in general, what was the frequency of home blood glucose monitoring? That’s the key for any insulin delivery. Checking blood glucose enough to make treatment decisions.

Dr. Heller: During training we emphasized the importance of blood glucose and ketone testing. They definitely had the training. Blood glucose monitoring was recommended up to five times per day and more. These were not people with a strong preference for a pump. Some people on pumps didn’t perhaps appreciate it in the way people who want pumps do. Is that a fault of study design? We clearly agonized over that.

Dr. Hirsch: But do you have SMBG data?

Dr. Heller: Only self-report. We don’t have those data.

Q: What about weight gain?

Dr. Heller: We have looked at it, and there were no significant differences in weight gain. I want to emphasize, these patients were poorly controlled, and the majority remained poorly controlled. It’s not surprising they didn’t gain much weight.

Q: Were the MDI patients on analog basal insulin? Were they using it once per day?

Dr. Heller: All these patients were on analog insulin. When funded, the reimbursement in the UK wanted to compare the best MDI practice. All the patients were on twice-daily Levemir and insulin aspart as the bolus insulin.

Q: In people that are physically active, often being on a pump allows them to make smaller adjustments more quickly, leading to overall better control. Did you take that into account?

Dr. Heller: We didn’t take into account physical activity levels. In a surprise to us, when we undertook the study, we thought we would have great difficulty because people didn’t want to risk going low with exercise on MDI. The opposite was true for most centers. When we talked to people on MDI about pumps, they would say, “I think I’m fine, and if I exercise, a pump is going to get in the way.” A lot of the people, such as you describe, who would benefit from a pump during exercise, didn’t think it was going to help them. And some of the people you talk about are probably using pumps already.

Debate: The Pump – Is It for Everyone?

Yes

Betsy Rodriguez, BSN, MSN, CDE (Centers for Disease Control and Prevention, Atlanta, GA)

Ms. Betsy Rodriguez opened this debate by defending her “mixed yes” stance on pumps, stressing that education is one of the most important aspects of successful pump therapy. According to Ms. Rodriguez, “the pump is not for everyone – it just needs to be an option!” She acknowledged that people choose insulin pump therapy for different reasons (better glucose control, more flexibility, fewer needles, etc.), including her daughter, who chose pump therapy because it dramatically reduced the frequency of hypoglycemic episodes. Ms. Rodriguez discussed other benefits of pump therapy, including peace of mind, discretion, and micro-doses for better control, though ultimately concluded that none of this would be possible without adequate pump education. We fully agree, especially considering that cost-effective, scalable education programs for patients are relatively few and far between; the burden of education is often on primary care providers or endocrinologists, who are already overworked and overwhelmed – it is no wonder that education is so often overlooked! Ms. Rodriguez concluded by underscoring value of intermittent “pump vacations”, too, emphasizing that “you can get on it, you can get off it, but at the end it’s about the quality of life.”

No

Dicky Poon (Hong Kong, China)

Mr. Dicky Poon discussed the practical issues that often prevent people with diabetes from using pumps, drawing on his own experience as a person with type 1 diabetes in Hong Kong. His presentation centered on three major factors hindering pump use in Hong Kong: (i) expense (>$600 per month for pump expendables vs. $120 per month for MDI); (ii) very limited tech support; and (iii) the humid subtropical climate, which leads to reliability issues and skin irritation. With these caveats in mind, he asserted quite succinctly that, “it’s too early to say that insulin pump is for everyone.”

Yes

Chuck Eichten (Author, The Book of Better: Life with Diabetes Can't Be Perfect. Make It Better, Portland, OR)

Mr. Chuck Eichten argued in favor of pump therapy, underscoring that it is the best option available for treating diabetes. He began his presentation with a humorous disclosure: “I am absolutely, unequivocally, unapologetically, and at times even hysterically biased in favor of the insulin pump.” He acknowledged that pump therapy is not perfect, but stressed that it provides the better control than MDI and offers major advantages, such as flexibility in eating and sleeping patterns, discretion/privacy, and fewer needles (“1,460 holes vs. 156 holes in yourself”). For those with objections to pump therapy, he offered a compelling analogy: “Remember how afraid you were the first time you met the love of your life and you were so scared to talk and now you are so happy you did? That’s how you will feel about your pump.”

No

Wilf Ruland (Ontario, Canada)

Type 1 father Mr. Wilf Ruland led a moving discussion that focused on Ontario’s inappropriate pump subsidies that practically force patients to use pumps regardless of their insulin delivery preferences. According to Mr. Ruland, the province subsidizes pumps and associated supplies but does not provide health funds for pens, syringes, insulin, CGMs, or pump-equivalent training for those on MDI. He positioned the policy as an inefficient outlay of public health dollars that could also be detrimental to health outcomes. We would add – perhaps most fundamentally – that the policy tactlessly places patient preferences as an afterthought, putting pump vs. MDI decision-making power in the hands of regulators instead of patients. Mr. Ruland alluded to his teenage daughter as an example of a patient affected by the policy, noting that she doesn’t want a pump, but despite excellent management on MDI, is not eligible for funding for her method of insulin delivery or for better pump-equivalent training. We are surprised to hear of this model of targeted subsidies and wonder if outcomes have improved as a result of it. Ultimately, a pump is a tool for insulin delivery, though whether or not it is “useful” can depend on the individual. 

Symposium: Health Information Technology to Prevent, Diagnose and Manage Diabetes

Using Health Information Technology to Prevent and Treat Diabetes

Fran Kaufman, MD (Chief Medical Officer, Medtronic Diabetes, Northridge, CA)

Dr. Fran Kaufman highlighted CGM, online programs, mobile phones, and other health information technologies (IT) as effective solutions for managing diabetes prevention and treatment in large populations. She emphasized the value technology brings to a constrained diabetes healthcare system – providers do not have sufficient time and effort to manage patients with diabetes through 10-minute visits every 90 days. Dr. Kaufman stressed the utility of CGM to enhance diabetes care, whether it’s used real-time, retrospective, integrated into the medical record, or driving closed-loop. We were particularly glad to hear her mention use of CGM in research studies to determine the effectiveness of interventions – we certainly hope to see much more of that in drug studies going forward. Dr. Kaufman expressed optimism for future integration of CGM into fully automated systems for insulin delivery (closed-loop), including systems with multiple data inputs (blood glucose, insulin, heart rate, activity, etc.). Of course, Medtronic is leading the field on closing the loop near-term, with the MiniMed 670G in a pivotal study slated to wrap up in May 2016. More broadly, she discussed the translation of evidence-based intervention programs into online formats, such as the chronic disease self-management programs developed by the Stanford Patient Education Research Center and the Diabetes Prevention Program. These programs were proven effective in prior studies, and online transference has improved their scalability and cost effectiveness. Dr. Kaufman concluded with an example of cell phone monitoring in which an Indian mobile diabetes program sent a series of text messages to one million people, demonstrating efficacy in improving physical activity, increasing consumption of fruits and vegetables, and decreasing consumption of high fat and fried foods. This vividly illustrates the considerable ROI potential of mobile phone technology, in which a vast number of people can participate in a simple, cost-effective intervention. Now that is where diabetes care HAS to go.

Effectiveness of Short Messaging Service (SMS) Using Mobile Phones in the Prevention of Diabetes

Ambady Ramachandran, MD, PhD (India Diabetes Research Foundation, Chennai, India)

Dr. Ambady Ramachandran contextualized his lab’s ongoing study (NCT01570946) evaluating the efficacy of text messaging interventions in type 2 diabetes prevention, providing an in-depth review of several diabetes prevention studies conducted in India. The ongoing study began enrolling participants (n=1,050) in India and the UK in 2011, and results are currently pending (slated to wrap up in December 2015). According to Dr. Ramachandran, the study is based on findings from a series of studies demonstrating that (i) lifestyle intervention is effective at preventing the progression to diabetes; and (ii) mobile phones can allow this intervention to be scalable and reach large populations via simple short messaging service (SMS). As an example, he shared results from the first Indian Diabetes Prevention Program (IDDP-1) in the early 2000s, which followed 531 participants with impaired glucose tolerance for three years to evaluate the effectiveness of lifestyle management and metformin on preventing progression to type 2 diabetes. Study results showed that 55% of the control group developed diabetes, while the lifestyle modification intervention reduced diabetes conversion by 28.5%. There was no additional benefit of using metformin. He noted that compared to DPP studies in the US and Finland, the IDDP-1 showed diabetes onset at a younger age and lower BMI. Next, Dr. Ramachandran highlighted key results from multiple recent studies investigating the efficacy of SMS in reaching a large number of people at a low cost and influencing them to change their behavior to improve health. Most notably, a randomized clinical trial demonstrated a 36% risk reduction in the incidence of type 2 diabetes in the intervention group compared to control. The intervention involved frequent mobile phone messaging with content tailored to participant sub-groups using the trans-theoretical model (a framework for changing behavior). The intervention significantly reduced the incidence of diabetes during the 2-year period, and HDL cholesterol and diet improved as well. We are pleased to see a focus on prevention in India, where diabetes rates are climbing past 69 million, resources are limited, and simple mobile phones are ubiquitous.

Experience of How This Works In Practice

Marlis Schosser (mySugr, Vienna, Austria)

IDF young leader Ms. Marlis Schosser described diabetes technology from the perspective of a patient with type 1 diabetes. Ms. Schosser opened by emphasizing the complexity of diabetes management: “Many say diabetes is easier now because we have lots of technology, but when you get diagnosed it turns your life upside down…You have to think and react how your pancreas thinks and reacts, which is impossible.” She noted that people with diabetes only get 15 minutes with a doctor every few months, a setup that places the burden of diabetes management on the patient ~90% of the time. So then, what does it take to self-manage diabetes successfully? According to Ms. Schosser, it starts with appropriate tools, education, and motivation. With respect to the first solution, Ms. Schosser proposed that data from diabetes devices need to be more seamlessly integrated into one platform to allow patients to access and glean insights from their CGM, pump, and activity data. This has become a common refrain at diabetes meetings for the past five years, though we think the conversation should start to move past this to more substantive questions– Which data platforms are really nailing the integration? How can we better integrate them into clinical practice? How can we make the data more actionable for patients and providers?

• Ms. Schosser also took time to advocated for online coaching tools and mobile apps that can scale diabetes education. She underscored the value of fun challenges, games, and connecting with the diabetes community, noting that, “we still have a while until a cure, so it’s up to people with diabetes to find solutions and help each other make our lives and diabetes management more fun.”

Open Forum: Public Health and Epidemiology

My “Big Data”: What Is It, How Is It Relevant to My Health, and Why Don't I Have Access to It?

Joachim Roski, PhD (Booz Allen Hamilton, Principal, Washington DC)

Dr. Joachim Roski discussed the future of big data in the context of many challenges: (i) How do we ensure data privacy and security?; (ii) How do we enable data sharing?; (iii) Who owns the data?; and (iv) How should we think about consent? Dr. Roski expressed optimism that these questions will be answered in the future though acknowledged that the field is in its early days – patients, providers, and healthcare organizations are only now beginning to ask themselves the “tough questions” and it is clear that medicine as a whole is only beginning to appreciate the promise of digital health. Dr. Roski concluded that the future is bright and that there is hope for the field as these conversations are becoming more and more common; we certainly agree and have seen the discussion around digital health come increasingly into the spotlight in 2015, particularly as diabetes devices have increasingly embraced connectivity (Dexcom G5, MiniMed Connect, Abbott’s FreeStyle LibreLink).

Applications of Social Media for Diabetes Epidemiology and Prevention Research

Joyce Lee, MD, MPH (University of Michigan, Ann Arbor, MI)

Dr. Joyce Lee provided an overview of the role of social media and the internet in promoting diabetes epidemiology and prevention research. She opened with compelling data illustrating that social media forums (YouTube, Facebook, Twitter, etc.) have now become some of the most prominent places for patients to exchange knowledge. She offered social media as the next frontier where providers can interact with patients both in terms of distributing knowledge and learning from their patients (“lurking”). On the other hand, Dr. Lee also stressed that an appreciation for the differences between social media platforms is key for those joining the community – e.g., Snapchat is more likely to be used by wealthier teens, Facebook is most popular among lower-income youth. She pointed out that each platform has a population that communicates in a very particular way – see the humorous “Social Media Doughnut” below. Dr. Lee also spoke at length about the way social media can be leveraged for research and circled around one question in particular: can social media replace traditional methods of disease surveillance? She spoke to the power of big data and predictive analytics (e.g., The Parable of Google Flu – see below), though acknowledged that the field remains far from a place where the technology can fully supplant more traditional methods of disease surveillance. Ultimately, Dr. Lee concluded that the value in social media is in using it as an ethnographic tool, one that helps us understand the patient experience and the issues with living with disease without necessarily overanalyzing/extrapolating from the results.

• The Parable of the Google Flu: Dr. Lee spent time explaining how and why Google’s flu tracking system made headlines in 2013 after incorrectly predicting more than double the proportion of doctor visits for influenza-like illness (ILI) relative to the CDC. The latter bases its estimates on surveillance reports from clinics across the US, while Google’s flu tracker was built to analyze search terms online and to predict ILI accordingly. However, Google’s researchers made a mistake that Dr. Lee termed “big data hubris,” an implicit assumption that big data are a substitute for, rather than a supplement to, traditional data collection and analysis. In this case, the odds of finding search terms that matched the propensity of the flu but which were structurally unrelated (and did not predict the future) were quite high, resulting in an overestimate of the flu progression. As a result, Google’s flu tracker is often held up as an exemplary use of big data and the errors that can result.
  
  • We’d highly encourage anyone interested in this topic to read the 2014 Financial Times article, “Big Data – Are We Making a Big Mistake.”

Social Media Doughnut 

Corporate Symposium: Non-Invasive Glucose Monitoring (NIGM): Have we reached the Holy Grail? (Sponsored by Integrity Applications)

The “Other Idea”: A Different Approach For A Truly Non-Invasive Glucose Monitoring Device - Past, Present And Future

Avner Gal (CEO, Integrity Applications, Israel)

Integrity Applications CEO Avner Gal introduced GlucoTrack, a CE-Marked and just-launched non-invasive ear lobe clip device for spot monitoring glucose in people with type 2 diabetes and prediabetes. The boldly titled corporate symposium (“Non-Invasive Glucose Monitoring (NIGM): Have we reached the Holy Grail?”) shared an unconvincing overview of the device to a small audience: an MARD of 23% vs. Hemocue fingersticks (53% in Zone A of Clarke Error Grid); an upfront cost of $2,000 and $120 for each six-month ear clip; a one-minute measurement time for each spot reading (it’s not continuous); no measurements <70 mg/dl; and the need to recalibrate the ear clip every six months with an ~20-30 minute calibration with a fingerstick. The device is said to have already launched in some places around the world (not specified), with expansion said to be expected soon. We think it has an uphill battle on many fronts (cost, accuracy, form factor, hassle), though Mr. Gal rightly noted that without early generation CGMs, the field would not be where current sensors are today. Still, the device has been in development for 12 years, and after clinical trials in >750 patients, it is hard to imagine a step function change in accuracy or form factor is coming anytime soon, particularly with just $2.1 million in cash in the bank as of 3Q15 and a market cap of $22 million on the OTC/QB exchange. The company has spoken to the FDA, and per its October announcement, plans to begin a US trial in early 2016. Integrity’s current product roadmap includes adding cloud connectivity, a type 1 indication, models for developing countries and the hospital setting, and more. We look forward to following the company and salute the team for trying to crack the extremely challenging non-invasive glucose monitoring (NIGM) – as Dr. Lutz Heinemann summarized the field in his overview talk, “NIGM has a long history of continuous ambition and failure,” but it is still worth pursuing for patients. (That said, we feel most believe FreeStyle Libre is nearly non-invasive.)  

• Integrity’s device begs many thought provoking questions: What level of glucose monitor accuracy is good enough for type 2 diabetes and prediabetes? How should companies trade off accuracy vs. improved form factor vs. cost? When does the output from a less invasive glucose monitor cross from useful to dangerous? These questions have important implications not just for Integrity, but type 2-focused glucose monitors from Abbott, Dexcom/Google, Medtronic, Novartis/Google, and wellness CGM companies like Echo and Sano. (Abbott is arguably the leader right now on the composite profile, having achieved an outstanding form factor with FreeStyle Libre, a lower cost, and accuracy safe enough to dose insulin with factory calibration. But there is clearly a spectrum on the cost, accuracy, and form factor dimensions of any product.)
  
  • Not every glucose-monitoring device needs to be accurate enough for dosing insulin, but there is presumably some threshold that must be hit. What is that threshold, and how should that be weighed against a smaller or less invasive or less costly device? Does For instance, if the Novartis/Google contact lens has a MARD of 20% (our speculation), is that good enough to drive therapeutic change in type 2 diabetes and prediabetes? At what point does the device’s clinical inaccuracy outweigh any potential benefit of more frequent monitoring? Tough questions indeed, and ones that will likely be solved only with prototypes tested in clinical trials.

The "Missing Point": Importance of frequent monitoring of glucose levels by type 2 diabetes patients

Irl Hirsch, MD (University of Washington, Seattle, WA)

Dr. Irl Hirsch did not speak about Integrity’s device specifically; he talked broadly about glucose monitoring in type 2 diabetes. He concluded that (i) based on the totality of current evidence, SMBG results in a small reduction in A1c in non-insulin users; (ii) glucose testing and CGM for those who do take insulin seems to be underutilized – lots of hypoglycemia is missed right now with fingersticks; and (iii) A1c is not a good marker to make treatment decisions for many patients (blood glucose data is better). Dr. Hirsch did comment briefly on Integrity’s device in Q&A, noting less concern with the MARD (“it will get better”) and questioning whether GlucoTrack would have value in the hospital and long-term care settings.

• SMBG studies in non-insulin-using type 2s do a poor job of replicating real life clinical care in the usual primary care setting, especially for the subjects randomized to more frequent SMBG. Dr. Hirsch noted that “the teaching, use, and interaction of SMBG is simply not yet part of the culture of usual primary care.” Dr. Hirsch did not provide any recommendations to counteract this; we wonder if blinded CGM could be helpful in this regard, since intermittent SMBG is extremely challenging to interpret.
  
  • Studies like STeP and several meta-analyses have shown small but significant effects of SMBG in non-insulin users (~0.2-0.3% A1c benefit). But studies have design and bias problems that confound results: (i) funding biases (e.g., STeP was funded by Roche; DiGEM was funded by NHS –the results were as expected based on those two founders); (ii) starting at an A1c too low to expect a benefit (e.g., 7.5% in DiGEM was too close to goal); and (iii) the intensive SMBG arm often tests too infrequently (e.g., less than once per day) to expect a benefit.
• Many patients with type 2 are at risk for hypoglycemia. Dr. Hirsch said that 23% of all prescriptions for type 2 diabetes are for a sulfonylurea, with another ~6 million Americans using insulin in 2012 (June 2015, UBS Large Cap Pharma Monthly Handbook; CDC Report, 2014). He further estimated that ~40 million people are using inulin globally (Gal, Bernstein Research, September 2013). Assuming 5% of these 40 million people have an episode of severe hypoglycemia at some point in their life translates to 2 million people worldwide. “The number is probably much higher,” said Dr. Hirsch.
  
  • “The type 2 diabetes hypoglycemia data is real, and without CGM, we miss the majority of the hypoglycemia in patients who receive insulin. In one example of 52 patients receiving insulin with type 2 diabetes (A1c: 7.7%), CGM detected 64 hypoglycemia events in 42 patients, translating to 79 events per patient-year (mean duration per event: 60 minutes). The frequency caught with SMBG was only 17 events per patient-year. “We are spending so much time in hypoglycemia; too much time,” said Dr. Hirsch, particularly given concerns about cognitive functioning, cardiac arrhythmias, and inflammatory activation.
• “A1c is a great biomarker when assessing glycemic control in a population or clinical trial. But A1c levels are problematic when comparing individuals.” Dr. Hirsch pointed to the ADAG study, which established ADA’s estimated average glucose equation by comparing A1c values to mean glucose measured via CGM. An A1c of 8.0% translated to an average blood glucose of 183 mg/dl in the study, but a tremendously wide confidence interval among patients: 147 mg/dl to 217 mg/dl. The same was true for an A1c of 7.0%: a mean of 154 mg/dl, but a range from 123 mg/dl to 185 mg/dl. “One can’t compare the A1c levels between two people,” said Dr. Hirsch. He expounded on this point by noting particular concerns in patients with aortic valvular diseases and iron deficiency.
  
  • “You need more data to make treatment decisions. It doesn’t make sense to make decisions based on an A1c or six tests per week,” said Dr. Hirsch. 

“Holy Grail”: History Of The Efforts To Develop A Non-Invasive Glucose Monitoring Device

Lutz Heinemann, PhD (Science & Co, Dusseldorf, Germany)

Dr. Lutz Heinemann gave an honest talk on non-invasive glucose monitoring (NIGM), summing up the field succinctly: “a long history of continuous ambition and failure” (most recently C8 MediSensors, who even raised money from the respected GE Ventures!). Though NIGM has an “image issue,” Dr. Heinemann said it warrants further investment (“we owe it to our patients”), and the concept is possible “in principle.” The challenges for scientists and companies, however, are numerous: precision of measurement, specificity, long-term stability, calibration, and real-world usability. After reviewing several failed approaches (e.g., C8, Roche, Pendra, Solianis), Dr. Heinemann concluded that non-invasive glucose monitoring is not a hot topic, and it is “tricky” to know the current status of devices in development. He noted only a small number of published clinical studies (none recently), with most being small and inconclusive. Despite the field’s negative image and years of frustration, Dr. Heinemann optimistically reminded attendees that scientists are “impressively creative” in finding new approaches – we should not lose hope. He believes trust should be re-established in this area of research through sufficient funding, transparent data evaluation, and better cooperation between academia, companies, and investors. Ultimately, any glucose monitor must be robust enough for daily life and offer precision and accuracy sufficient for clinical purposes. We appreciated Dr. Heinemann’s honest appraisal of the field, though it seems unlikely major funding will begin flowing into this area, particularly with minimally invasive products like Abbott’s FreeStyle Libre on the market and Dexcom/Verily [Google Life Sciences] and Novartis/Verily in the works.

Panel Discussion

Q: Why is this not indicated for type 1 diabetes? And does skin complexion affect the results? What about lag time?

Mr. Gal: Our technology is not impacted by skin pigmentation. We did trials with a variety of skin colors.

As for type 1, we definitely understand the need. The fluctuations in type 1 are much higher and faster than in type 2s. We can use device for type 1 diabetes, but we decided we have to go to market step by step. That was a wise decision. With all the sorrow we cannot fulfill the need for type 1 diabetes, we believe in the future we can submit the device for type 1 diabetes.

By definition we have lag time; we cannot run away from that. Lag time does impact the accuracy. In the results you saw we take into consideration lag time.

Dr. Irl Hirsch: A practical question in the clinical trials. Were patients wearing the device 24/7? Were there problems with it falling out of the ear?

Mr. Gal: The device is intended for spot measurement – not continuous. You have to clip it on the ear lobe for about one minute. Each time you want to perform a measurement, it has to be clipped on. During the clinical trials, it depended on the group. In the clinic, the research team clipped the device onto the ear lobe. In the home simulated, measurements were done by the subject.

Dr. Hirsch: I have a real interest in inpatient or long-term care facility patients. Your higher MARDs don’t concern me personally. Both in SMBG and CGM over time, it gets better. I’m not worried about that. The potential huge use for this technology is in the hospital as the MARDs get better. That is where we struggle. Not just in the hospital, but in long-term care and nursing homes. We really struggle with glucose data. Have you thought about that?

Mr. Gal: We are talking about a variety of other devices based on the same technology. One at a time. It’s hard enough to start to get into the market, penetrate, and stay stable. We have lots of ideas for our team, and we could go until 2040 and not make everything.

Q: What happens in hypoglycemia values <70 mg/dl, or values over 500 mg/dl? What does the device show?

Mr. Gal: The device recommends either a repeat measurement or using the conventional invasive device.

Dr. Heinemann: Irl, do you see any light at the end of the tunnel for a clearer view on the role of SMBG in patients with type 2 diabetes not on insulin? In Germany, this is still a hotly debated topic. Some areas get test strips; others not. My view on the literature is its difficult to interpret.

Dr. Hirsch: It is. With conventional SMBG the conclusion – which is biased – is a small but significant improvement for patients not on insulin. My biggest issue – and I am biased because I probably see more of these than PCPs – is that the A1c is not a good metric to make treatment decisions. And because of that, fingerstick testing becomes more important. Maybe what needs to be done is if you have a patient with a known reason why A1c does not give an accurate reflection of glycemia, those are patients we need to focus on glucose to make decisions. After all, all these biomarkers are just reflections of glucose. If a biomarker doesn’t work, there is nothing wrong with going back to what works. It’s glucose that’s the enemy. I showed data on cardiac valvulopathies, which very few people appreciate. That’s the reason why as an endocrinologist, if I’m listening to the heart and I hear a heart murmur, there is a good chance A1c is not reflective of glycemia. In those patients we should pay for more strips.

Dr. Heinemann: Such non-invasive devices could avoid SMBG measurements in the future.

Dr. Heinemann: When are you coming to market in the near future? Do you have a CE Mark?

Mr. Gal: The device is currently available in some territories around the world. We are expanding distribution agreements around world. Right now it is there. We users around world – not too many to be honest – in the few hundreds. We just started recently. We expect to accelerate – not dramatically – and definitely increase the devices around world.

Q: What is the ideal way to use this tool?

Mr. Gal: The major benefit is the ability to monitor more frequently. What happens now is doctors recommend patients with type 2 diabetes (non-insulin treated and insulin treated) measure once a day or twice a day. They don’t try to convince them to measure 7-8-10 times per day. All the doctors know that if they try to convince patients to monitor more frequently, they won’t do it. IDF recommends daily 4-5 times. All the physicians will agree – none of the patients will do that. “Let’s try to convince them to do it once or twice a day.” With such a device as GlucoTrack, you can measure as many times as you like. With an invasive device, the more you measure, the more you pay.

-- by Adam Brown, Varun Iyengar, Ava Runge, and Kelly Close