7th International Conference on Advanced Technologies & Treatments for Diabetes (ATTD 2014)

February 5-8, 2014; Vienna, Austria; Day #2; Highlights – Draft

Executive Highlights

The official scientific program of ATTD 2014 kicked off today, and boy was there plenty to listen and react to! Once again, we heard engaging material all over diabetes technology, especially on CGM and the artificial pancreas. A few major highlights came in Dexcom’s jam packed industry workshop, where Dr. Ed Damiano shared topline glucose results from the 2013 Beacon Hill and Summer Camp outpatient studies of the bionic pancreas. In both ambitious and compelling studies, the bionic pancreas improved mean glucose vs. usual care (133 vs. 159 mg/dl [Beacon Hill] and 142 vs. 158-189 mg/dl [Summer Camp]) with a simultaneous reduction in mean time spent <60 mg/dl (1.5% vs. 3.7% and 1.3% vs. 2.2%). Very impressive data and we will hear much more on the studies tomorrow in the morning plenary and Tandem-sponsored afternoon workshop. Dexcom’s symposium was also headlined by a new head-to-head comparison between the G4 Platinum and Medtronic Enlite in 24 patients (each wore both sensors) – an overall MARD of 13.6% with the G4 Platinum “significantly outperforming” the 16.6% MARD seen with the Enlite (p=0.0002). At another session, Dr. Frank Doyle (Sansum Diabetes Research Institute, Santa Barbara, CA) presented impressive pilot results from a model predictive control (MPC) algorithm study. Patients (n=10) experienced 75% time in range, and a superb 97% time in range overnight (although with a low number of visits).

On the CGM front, presentations from Drs. John Pickup (patient narratives on CGM) and Steve Edelman (patient responses to real-time trend arrows) definitely piqued our interest. Though this technology has been around for over a decade, there is still much to learn about how to best use it in practice and how to improve it in the coming years. It further reinforced, particularly Dr. Edelman’s data, how patients who are not lucky enough to have access to CGM “fly blind” in so many aspects of their lives.

In a morning session sponsored by Tandem, Dr. Nate Heintzman (UCSD, San Diego, CA) quickly presented topline results from a survey of Tandem users on their self-reported A1c trend over their first six months of using the t:slim pump (n=289). After a mean of six months, the average A1c drop was 0.4% (baseline: 7.4%; p<0.001) among all respondents. The brand breakdown of those in the experienced pump group suggested that Tandem may be taking a disproportionately higher share of current pumpers not on Medtronic. 

Other talks from the day that made our highlights reel include: 1) Dr. Peter Chase (Barbara Davis Center, Aurora, CO) presenting data on an impressively large 1,912-night study testing predictive low glucose suspend in the home setting; 2) Remarks from Dr. Guido Freckmann (Institute for Diabetes Technology at the University of Ulm, Ulm, Germany) on the strictness of the new FDA accuracy guidelines for point-of-care devices; 3) Study results from Dr. Lori Laffel (Joslin Diabetes Center, Boston, MA) on type 1 diabetes management and psychosocial parameters; and 4) Dr. Noel Schaeffer (Tandem, San Diego, CA) comparing the training duration and usability of Medtronic’s Paradigm Revel and Tandem’s t:slim. Our report also includes a few honorable mentions, followed by an appendix with detailed reports on most of our top ten talks, as well as some others we attended today.

Table of Contents 

Top 10 Highlights

1. In a talk mostly devoted to CGM accuracy in Dexcom’s industry workshop, Dr. Ed Damiano (Boston University, MA) notably shared topline glucose results from the 2013 Beacon Hill and Summer Camp outpatient studies of the bionic pancreas! In both ambitious, practical outpatient studies, the bionic pancreas improved mean glucose vs. usual care (133 vs. 159 mg/dl [Beacon Hill] and 142 vs. 158-189 mg/dl [Summer Camp]) with a simultaneous reduction in mean time spent <60 mg/dl (1.5% vs. 3.7% and 1.3% vs. 2.2%). We’d note that the comparison group in both studies was extremely challenging, as patients in Beacon Hill were highly motivated and in Summer Camp were receiving far better than usual care. Notably, the results project to an A1c reduction of 0.9% in Beacon Hill (a projected 6.2% A1c vs. the usual care projection of 7.1%), rising to an impressive 1.6% A1c reduction in the summer camp study (baseline: 8.2%). In short, this data is unprecedented in diabetes care – dramatic reductions in both mean glucose AND hypoglycemia. The system is also designed to take the thinking out of type 1 diabetes (Dr. Damiano’s ultimate goal) with features like fully reactive control and no need to carb count. His talk also highlighted the head-to-head-to-head CGM data presented at ADA 2013 comparing the Dexcom G4 Platinum, Abbott Navigator, and Medtronic Enlite/Veo/530G. Below, we detail more from the Beacon Hill and Summer Camp studies, including a stimulating Q&A. 

2. Also in Dexcom’s workshop, Dr. Jort Kropff (Academic Medical Center, Amsterdam, Netherlands) presented on the SPACE 2 trial, a very detailed and thorough head-to-head comparison of the Dexcom G4 Platinum and Medtronic Enlite sensors. The four-center, 24-patient study gathered CGM accuracy data over six days in the home setting (vs. fingersticks on the Accu-Chek Aviva meter) and over six hours in an in-clinic assessment (vs. YSI; included breakfast + a 180% insulin bolus to induce hypoglycemia). The 24 patients wore two sensors each. Devices were calibrated per manufacturer’s instructions (it was not specified how many calibration this entailed for the Enlite). Consistent with independent data previously presented from Dr. Ed Damiano, the Dexcom G4 Platinum was more accurate than the Enlite – overall MARD of 13.6% with the G4 Platinum “significantly outperformed” the 16.6% MARD seen with the Enlite (p=0.0002). The G4 Platinum was also significantly more accurate for CGM values <70 mg/dl: 17.6% vs. 24.6% for the Enlite (p=0.005). Dr. Kropff emphasized that both devices were much less accurate in the hypoglycemia region, leaving room for future generations to improve. The G4 Platinum was significantly more accurate than the Enlite on all days of the study. The Clarke Error Grid also supported the MARD findings: 83% of G4 Platinum points fell in Zone A compared to 65% for the Enlite. Looking at individual sensor performance, 15% of Enlite sensors had a MARD <10% vs. a notable 40% of G4 Platinum sensors. In concluding, Dr. Kropff emphasized that collecting both in-clinic and home CGM accuracy data is critical, as the latter adds “essential information” (e.g., a wide range of glycemia and duration of usage).

3. Dr. John Pickup (Guy’s Hospital, London, UK) presented results from a qualitative UK online survey (n=100) to understand patient experiences with CGM (treatment satisfaction, perceived clinical effectiveness, stories). The main purpose of the survey was to answer an open-ended question, “Please describe in your own words your personal experience with CGM, including any benefits, problems, and drawbacks to using the sensor.” The analysis broke down responses into six themes (metabolic control, life on CGM, CGM procedures, technical issues, financial issues, attitudes to CGM), with several sub-themes and example quotes. Overall, patient responses to CGM were “overwhelmingly positive.” A1c and/or hypoglycemia were usually reduced on CGM, and quality of life usually improved, particularly in those with problematic diabetes. The survey did find that there are still significant problems and hassles with CGM, though most patients understand and accept the limitations. For many patients, CGM was described as “life-changing” and sometimes even “life-saving.” Most patients were on Medtronic systems. More complete details are below. We most appreciated Dr. Irl Hirsch’s comment in Q&A, which pointed out that such glowing real-world feedback could be useful in reimbursement discussions – particularly because clinical trials have not shown hugely striking benefits to using CGM. Indeed, this was one of the main reasons Dr. Pickup undertook the study. Read a detailed account of the survey responses below, along with the subsequent Q&A discussion.

4. Dr. Francis J. Doyle III (Sansum Diabetes Research Institute, Santa Barbara, CA), a self-confessed control engineer, showed impressive results of closed loop outpatient clinical studies at UVA and Sansum. The focus was on model predictive control (MPC) algorithms, which Dr. Doyle has been working on for twenty years. MPC was established for the chemical industry. There, a process control system is surrounded by a number of safety layers, each successive layer less likely to be needed. In the same way, safety layers in the MPC algorithm for diabetes include constraints on insulin delivery, glucose zones (which can vary throughout the day), and prediction of low glucose and detection of glucose rate of change. Results for five day outpatient experiments (in Charlottesville and Santa Barbara) were shown for n=10 patients. The data was excellent – time in zone under closed loop control was 75%, and a superb 97% overnight (but only data for a few subjects). Also, Dr. Doyle showed a fascinating and instructive movie of how the algorithm predicts glucose 45 minutes ahead and takes the appropriate action to enable the right outcomes. This work shows the enormous potential of closed-loop technology and the fact that it now works safely and effectively in real-life conditions.

5. During a Tandem sponsored session, Dr. Nate Heintzman (UC San Diego, CA) quickly presented topline results from a survey of Tandem users on their self-reported A1c trend over their first six months of using the t:slim pump. The survey was emailed to 5,000 t:slim users, of which 289 completed the survey and included their A1c data. After a mean use period of about six months, the average A1c drop was 0.4% (baseline: 7.4%; p<0.001) among all respondents. In the experienced pump user cohort (n=211), the average A1c reduction was 0.2% (baseline: 7.1%; p<0.001) after ~6.2 months. The most robust A1c decline occurred in the new-to-pump group, who saw a 0.9% reduction in A1c after 5.3 months of use (baseline: 8.0%; p<0.001). Overall, increased use of t:slim features and software by previous pump users was associated with a greater though non-significant A1c improvement (details below), and over 50% of respondents indicated that they uploaded data before their doctor’s appointment. Of course, all of this data is subject to self-reporting bias (289/5,000 is quite a low response rate), though it is directionally interesting. It was quite illuminating to see the brand breakdown of those in the experienced pump group: 32% formerly on a Medtronic pump, 21% formerly on Animas, 14% formerly on Deltec, 4% formerly on OmniPod, 1% formerly on Roche, and 1% formerly on other brands. This was particularly fascinating, as Medtronic has about two-thirds of the overall pump market – this suggests that Tandem may be taking a disproportionately higher share of current pumpers not on Medtronic. We also would have been extremely interested in “time in zone” data since A1c differences never tell the entire picture.

6. Dr. Steven Edelman (UCSD, San Diego, CA) presented a subset of results from a fascinating 300-patient survey examining how CGM users respond to real-time trend arrows. His talk – also in Dexcom’s packed industry workshop – shared data from 222 respondents with type 1 diabetes (mean A1c: 6.9%; 75% CSII). On average, they reported making “major modifications” to insulin therapy based on trend arrows. These modifications, which ranged from -48% to +140%, were much larger bolus insulin adjustments than have been previously suggested in the literature (10-20%). We appreciated the very thorough case-based approach that asked participants how much bolus insulin they would take in a certain scenario (e.g., 220 mg/dl CGM reading, flat trend), and then how that would change with different trends (e.g., double up arrows, one down arrow, etc.). Results are briefly summarized below, and much more will be shared in the future on type 2s vs. type 1s; type 1 MDIs vs. pumpers; and more. In showing the summarized results, Dr. Edelman said, “This slide explains why there is so much frustration out there.” Indeed, for the example three-unit dose of bolus insulin, patients reported they would give 7.2 units for a double upward arrow trend and just 1.5 units for a double downward trend. For Dr. Edelman, this illustrates why “CGM really is the standard of care for people with type 1 diabetes” – static BGM values don’t offer enough information to make an informed dosing decision.

Trend Arrow

Mean Patient Adjustment to Bolus Dose*

Example Bolus Dose*

-->

---

3.0 units

^ (up)

+111%

6.3 units

^^ (double up)

+140%

7.2 units

v (down)

-41%

1.8 units

v v (double down)

-48%

1.5 units

*“It has been four hours since your last dose of insulin and meal and your CGM receiver shows a value of 220 mg/dl (matching your fingerstick blood glucose of 220 mg/dl) with an arrow and trend graph that is flat (straight across). If you are not planning on eating or exercising, what dose of insulin would you give yourself to bring your glucose to 120 mg/dl?”

7. Dr. Peter Chase (Barbara Davis Center, Aurora, CO) presented on an impressively large 1,912-night study testing predictive low glucose suspend (PLGS) in the home setting. The trial randomized 45 patients to either PLGS on (942 nights) or PLGS off (970 nights) – the algorithm (30-minute prediction horizon) conducted the randomization each night, allowing each patient to serve as his or her own control. (Note, this is not Medtronic’s PLGM algorithm.) The system does not use alarms, as the goal is “to have people sleep at night” – excellent! Notably, the percent of nights with a glucose <60 mg/dl was 21% with PLGS on vs. 33% with PLGS off (a 36% reduction). In addition, AUC <60 mg/dl was 81% lower on PLGS nights compared to control nights. (This puts into perspective just how much better predictive suspend is than threshold suspend, as ASPIRE in-home found a 38% AUC reduction for episodes <65 mg/dl). Time in range (70-180 mg/dl) significantly improved when PLGS was turned on – 82% vs. 75% (p<0.001). PLGS halved the number of nights with a glucose <60 mg/dl for more than 30 minutes (12% vs. 24%; p<0.001). On the safety side, A1c did not change in the study, and the likelihood of developing ketones did not increase when PLGS was turned on. With ~2,000 measurements, this represents the largest ketone outpatient study ever done in the world. There was slightly more morning hyperglycemia, though the incremental tradeoff for significantly less nocturnal hypoglycemia would likely be worth it for most patients. More details below.

8. During his talk on meter accuracy, Dr. Guido Freckmann (Institute for Diabetes Technology at the University of Ulm, Ulm, Germany) remarked that the new FDA accuracy guidelines for point-of-care (POC) devices are “very strict and will need further discussion.” We have heard similar comments from other leaders in the field since the standards came out at the beginning of January – many people think that this accuracy level will be very challenging for many US point-of-care devices to meet. These requirements are indeed much tighter than the 2013 ISO Standards for POC – the 2013 CSLI standards require that 95% of test values fall within 12.5% of reference for values >100 mg/dl, and within 12.5 mg/dl for values <100 mg/dl; the new FDA POC guidelines, on the other hand, state that 99% of measured values must be within 10% of reference for values ≥100 mg/dl and within 7 mg/dl for values <100 mg/dl (additionally, no value should exceed 20% of reference for values ≥70 mg/dl or 15 mg/dl for values <70 mg/dl). The FDA hospital standards are more challenging still, with 99% of all measurements within 10% variance; this includes nursing homes and assisted living facilities and we do not think they will be in a position to meet the standards. While some believe that the US standards will be impossible for some manufactures to meet, particularly the hospital standards, others point to the tighter control requirements reflecting the FDA’s desire to position itself as improving optics, i.e., improving accuracy in the hypoglycemic range, which the organization has long been under pressure to do. How these requirements will be enforced for offshore entities is a major question; in general, we think supply chain dynamics are becoming increasingly challenging for the FDA to manage. While we think it is very positive that there are finally updated standards - until this January, the FDA had used the 2003 CLSI Standards for POC devices (which were the same as 2003 ISO Standards) – we would be very interested to better understand what evidence there was that the ISO standards were not sufficient. We are eager to see how the US industry will respond to these new standards and whether or not some companies will back out of healthcare facility blood glucose testing. While some have wondered if the US POC market prevails, if the EU will consider tightening its guidelines in the future, as well – we do not envision that as likely at this stage.

9. Dr. Lori Laffel (Joslin Diabetes Center, Boston, MA) announced initial results from the much-awaited TEENS Study, which is the largest, contemporary, cross-sectional study assessing type 1 diabetes management and psychosocial parameters in youth. The results Dr. Laffel presented were for the US and Russia cohorts (n=500 each). The analysis suggested that the majority of youth with type 1 diabetes in the US (75%) and Russia (84%) are not at their A1c target (or close to it), and experience acute complications at relatively high rates (details below). These disappointing results are despite many youth with type 1 diabetes in the US and Russia using advanced management tools like frequent blood glucose monitoring (on average about four time a day) and pumps (63% in US and 19% in Russia). Dr. Laffel hopes that the further analysis of the TEENs Study will offer recommendations on how to broadly improve type 1 diabetes management among the youth, and during Q&A indicated that the study group is looking to present more detailed results from the broader trial at ADA. We look forward to hearing these important results.

10. Dr. Noel Schaeffer (Tandem, San Diego, CA) presented a study comparing the training duration and usability of Medtronic’s Paradigm Revel and Tandem’s t:slim. Participants (n=72 participants, 45 of whom had type 1 diabetes) were randomized to either the Paradigm Revel or t:slim pump. After training, participants were asked to perform seven tasks: 1) set the date and time; 2) set up a new pump profile; 3) deliver a food bolus; 4) deliver an extended bolus with correction; 5) start a temporary basal rate; 6) stop the pump; and 7) resume therapy. On average, patients using the t:slim required less pump training (18 vs. 26 minutes) and experienced 65% fewer task failures when attempting the seven tasks (2.3 errors vs. 0.8 errors). Additionally, participants were asked to rate the difficulty of each of the tasks on a scale of one to seven (difficult to easy); while five of the tasks did not differ in difficulty between pumps, participants did report that it was easier to “[set] up a new pump profile” (5.4 vs. 4.3, p=0.009) and “[deliver] an extended/dual wave bolus” (5.2 vs. 4.0, p=0.009) on the t:slim vs. Paradigm Revel, respectively. At the end of the study, participants completed a pump survey based on their experience during the study. The t:slim received significantly higher marks than the Paradigm Revel on seven characteristics: 1) terminology clarity; 2) screen contrast; 3) word readability; 4) pump size; 5) ease of programmability; 6) preparation from training; and 7) “enjoyability” (p <0.03 for all).

 

Honorable Mentions

  • In Tandem’s sponsored session, Dr. Noel Schaeffer (Manager – User Experience Research, Tandem, San Diego, CA) described the impact human factors research had in the t:slim’s original design. For background, human factors research is a scientific discipline that seeks to optimize the interaction between humans and their environment, or in the case of this session, their insulin pump. Dr. Schaeffer noted that human factors research differs from market research, as it uses evidence-based data to develop safe and effective products for users rather than focus groups to learn what features people believe they want – an important and misunderstood distinction in our view. An example of a design problem Tandem identified and solved using human factors research was that an early prototype of the t:slim included a “DEL” button in the bolus-entry screen. This button was intended to represent “DELETE”; however, participants in usability tests were sometimes confused if it instead meant “DELIVER” [insulin]. Tandem took the feedback and changed the image to a back arrow to represent the button’s intended delete function. During Q&A, Dr. Schaeffer stated that the medical device industry is increasingly using human factor research, as the FDA has mandated companies conduct these analyses.   
  • Dr. Thomas Danne (Kinderkrankenhaus, Hannover, Germany) pressed that the SWEET project, a currently EU-centered pediatric diabetes registry, wants to integrate participants of the US-based T1D Exchange. Dr. Danne explained that its main funder ISPAD has recommended the SWEET project expanding outside Europe in 2014. In addition to the US, Dr. Danne is looking to include patients in India, Brazil, Venezuela, the United Arab Emirates, and Canada. Dr. Danne also commented during Q&A that the biggest barrier the SWEET project is facing is limited funding. Dr. Danne continued to remark, “we would love $25 million from the Helmsley Charitable Trust.”

 

Appendix: Detailed Discussion and Commentary

Industry Workshop: Dexcom G4 Platinum – The Pathway to Superior Accuracy, Clinical Outcomes, and the Artificial Pancreas (Sponsored by Dexcom)

PATHWAY TO THE BIONIC PANCREAS: ASSESSING SENSOR ACCURACY IN A HEAD-TO-HEAD STUDY

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

In a talk mostly devoted to CGM accuracy, Dr. Ed Damiano notably shared topline glucose results from the 2013 Beacon Hill and Summer Camp outpatient studies of the bionic pancreas! In both ambitious, practical outpatient studies, the bionic pancreas improved mean glucose vs. usual care (133 vs. 159 mg/dl [Beacon Hill] and 142 vs. 158-189 mg/dl [Summer Camp]) with a simultaneous reduction in mean time spent <60 mg/dl (1.5% vs. 3.7% and 1.3% vs. 2.2%). We’d note that the comparison group in both studies was extremely challenging, as patients in Beacon Hill were highly motivated and in Summer Camp were receiving far better than usual care. Notably, the results project to an A1c reduction of 0.9% in Beacon Hill (a projected 6.2% A1c vs. the usual care projection of 7.1%), rising to an impressive 1.6% A1c reduction in the summer camp study (a projected 6.6% A1c vs. 8.2% at baseline). In short, this data is unprecedented in diabetes care – dramatic reductions in both mean glucose AND hypoglycemia. The system is also designed to take the thinking out of type 1 diabetes (Dr. Damiano’s ultimate goal) with features like fully reactive control and no need to carb count. His talk also highlighted the head-to-head-to-head CGM data presented at ADA 2013 comparing the Dexcom G4 Platinum, Abbott Navigator, and Medtronic Enlite/Veo/530G – as a reminder, Dexcom had a MARD of 10.8%, better than the 12.3% with the Navigator and 17.9% with the Medtronic Enlite/Veo/530G (Dr. Damiano emphasized that the latter used the algorithm that patients would see with the Medtronic MiniMed 530G, which recently came to the market in the US). Below, we detail more from the Beacon Hill and Summer Camp studies. 

  • Unexpectedly, Dr. Damiano shared topline glucose data results from the Beacon Hill and Summer Camp outpatient studies of the bionic pancreas. Dr. Damiano discussed the background on both studies (described below) and showed aggregated data from days 2-5 of each study. Day one data was excluded due to the nature of the algorithm – the device adapts and takes some time to learn about patients, meaning day one is not representative of days 2-5 (e.g., mean glucose was 151 mg/dl on day one in adults vs. 131 mg/dl on days 2-5). In all publications, Dr. Damiano shows day one data.

Beacon Hill Study Results, Days 2-5
(n=20 adults, 100 bionic pancreas days)

 

Bionic Pancreas

Usual Care

 

CGM

% CGM <60 mg/dl

CGM

% CGM <60 mg/dl

Mean

133 mg/dl

1.5%

159 mg/dl

3.7%

Projected A1c

6.2%

7.1%

 

 

Summer Camp Study Results, Days 2-5
(n=32 adolescents, 160 bionic pancreas days)

 

Bionic Pancreas

Supervised Camp Care

Baseline

 

CGM

% CGM <60 mg/dl

CGM

% CGM <60 mg/dl

CGM

Mean

142 mg/dl

1.3%

158 mg/dl

2.2%

189 mg/dl

Projected A1c

6.6%

7.1%

8.2%

 
  • The randomized, crossover Beacon Hill study compared five days on the bionic pancreas to five days of “usual care” (what a patient would normally do, though with the addition of blinded CGM). The study included 20 adult type 1 patients >21 years. The bionic pancreas mobile platform consisted of two Tandem t:slim pumps (insulin and glucagon), a Dexcom G4 Platinum sensor and transmitter, and an iPhone 4S controller. Patients had free run of a three-square mile area of the Boston peninsula. Point of care capillary blood glucose checks occurred during the day via 1:1 nursing. At night, patients slept in a hotel with venous blood glucose monitoring and 1:2 nursing. A total of 100 days on the bionic pancreas were accumulated. For more information, read our diaTribe test drive on the Beacon Hill study.
  • The randomized, crossover Summer Camp study compared five days on the bionic pancreas to five days of supervised camp care. The study took place at Camp Joslin (n=16 boys) and the Clara Barton Camp (n=16 girls) in 2013. Point of care capillary blood glucose checks occurred during the day and night (no venous glucose monitoring!). The same mobile platform was used as in Beacon Hill. Study staff and camp staff provided 24-hour, round-the-clock telemetry to monitor glycemia. A total of 160 days on the bionic pancreas were accumulated. For more information and interviews with trial participants, please see our detailed Closer Look write-up after we visited the study site this past summer.

PANEL DISCUSSION

Dr. Jay Skyler (University of Miami, FL): Dr. DeVries, you and Dr. Damiano showed the same thing in the sensor comparison. Independent studies seem to be showing that. Why hasn’t Medtronic switched and used the Dexcom sensor for their pumps? [Laughter]

Dr. Hans DeVries (Academic Medical Center, Amsterdam, Netherlands): To whom is the question directed?

Dr. Damiano: I’m sure the offer has been made.

Dr. Steve Edelman (UCSD, San Diego, CA): In the Beacon Hill study, why was the control group blinded to their CGM?

Dr. Damiano: The study CGM worn for study purposes had its display blinded and its alarms muted. But we let them do whatever they wanted to with regard to their own glucose monitoring. If you happened to wear a CGM before the study, you could still wear one during the study. The reason we didn’t want to mandate that everyone use CGM was because it would have required a training period for those who had never worn a CGM.

Dr. Edelman: What percentage of patients were on CGM prior to the study?

Dr. Damiano: We had a pretty engaged population of patients in Beacon Hill. The T1D Exchange figure has 8% on CGM. I think we were about 30% in Beacon Hill? But don’t quote me on that. We definitely had a higher percentage on CGM than the population as a whole.

Dr. DeVries: In your studies, glucagon was reconstituted. How often did you refresh the glucagon? What is your projection for a commercially available stable glucagon?

Dr. Damiano: The glucagon formulation that we’re using is the Lilly product. It is human glucagon. We got FDA approval eons ago, I think in 2007, to use that glucagon in solution and reconstitute it at the point of care at the beginning of each experiment. We can run it for up to 27 hours, but not more than that. For multi-day studies, we change the glucagon reservoir daily with a freshly reconstituted one milliliter solution. Going forward, that’s not a practical solution – even over 27 hours, it’s breaking down. There has been lots of effort in recent years from small pharma companies, and even some of the large ones, interested in making a stable, pumpable glucagon. They are also looking into a real true rescue solution – it’s an absurdity to figure out what to do with a glucagon kit when someone is seizing in front of you. Often times, people inject sterile water. It’s an amazing that it ever got clearance. Even first responders are not allowed to use glucagon. There is lots of interest in making Epi-pen like devices. So these small pharma companies are now working on stable formulations, and there has been lots of success – Xeris, Biodel, Latitude. There all have ways to stabilize it, and there are a number of ways to skin that cat. My recommendation is not to make an analog. We don’t need to make it faster; we just need to stabilize it. As long as it doesn’t reduce absorption times, we don’t need faster glucagon. It absorbs in 15 minutes.

Dr. Udo Hoss (Abbott Diabetes Care, Alameda, CA). In the head-to-head study, there was a huge difference in accuracy in the CRC and home use. What is your explanation?

Dr. DeVries: There are some key differences between those settings that affect the results. The CRC provokes hyperglycemia and hypoglycemia. The circumstances put the system to the test more stringently with more rapid rates of change. At home, things could be more stable and you have less hypoglycemia. But you are also integrating a whole week of performance, and over time things may change. We did the CRC on day three, and by and large, performance on that day tended to be the best. 

Dr. Damiano: We saw a similar thing. In Beacon Hill, we checked capillary plasma blood glucose every two hours during the day and venous plasma glucose every 30 minutes at night. We found the MARD was worse in the real world on the G4 Platinum – 16.7% with the bionic pancreas out in the field vs. about 11% in adults in the inpatient setting. The reasons could be that capillary blood glucose is a lot more variable than venous blood glucose. We used alcohol swabs and we did as good a job as we could, but it may not have been enough.

Dr. Skyler: Probably the in-dwelling CGM is more accurate than the fingerstick, which is subject to so much more variability. The increased variability is not because of CGM, but because the BGM is not as good.

Dr. Damiano: I totally agree. We’re calibrating devices [CGMs] that are potentially more accurate than the devices we’re using for calibration [BGMs].

Ian Jørgensen (Tidepool, Palo Alto, CA): In both outpatient studies, Summer Camp and Beacon Hill, what was the practical thing that was hardest. Was it chasing kids, calibrating CGMs, refilling glucagon, monitoring devices?

Dr. Damiano: During the camp study, the hardest thing was wireless connectivity to the cloud service. That was a big challenge because we were out in the middle of the woods. We did it all right in the girls camp. Whenever a kid’s device stopped sending data to the cloud, we would bring them to our command center and use Wi-Fi. At the boys’ camp, there was one day where we had no cloud service for the whole day. We brought all of the kids into the mess hall and had a slumber party – it was like something out of a Harry Potter scene at Hogwarts. All 16 kids slept in the mess hall and our study staff looked at the devices manually every few minutes all night long. There are lots of challenges in all of these studies.

Q: When do you most commonly inject glucagon? How does glucagon work around physical activity?

Dr. Damiano: It varies tremendously from subject to subject. Where glucagon is given the most is in the late postprandial period 3-4 hours after a meal and during and after exercise. And with activity, even moderate activity is a challenge – even walking around downtown Boston can really drop blood glucose.

Dr. Claudia Graham (Dexcom, San Diego, CA): Steve, you demonstrated that there is a 100% difference in somebody’s dosing response depending on the arrows and direction of change. To dose off a static measurement of 200 mg/dl, people went to 7.5 units or down to 1.5. Did you ask what they would have done if they just did a static SMBG measurement?

Dr. Edelman: Part of the question was that they confirmed every sensor glucose with a blood glucose.

Howard Look (Tidepool, Palo Alto, CA): Did you do any correlative studies to get data off the devices – this was what people said, but what do people do in real life?

Dr. Edelman: These folks are all over the country. Even self-reported A1c was anonymous. They did get a $25 gift card for completing the questionnaire, but it was totally anonymous. There is no way to confirm those answers.

Mr. Look: It might be interesting to pull the data.

Dr. Edelman: Let’s call the National Security Agency and have them get the data for us. [Laughter]

Pumps and Sensors

PATIENT NARRATIVES ON THEIR EXPERIENCES OF CGM

John Pickup, MD (Guy’s Hospital, London, UK)

Dr. John Pickup presented results from a qualitative UK online survey (n=100) to understand patient experiences with CGM (treatment satisfaction, perceived clinical effectiveness, stories). The main purpose of the survey was to answer an open-ended question, “Please describe in your own words your personal experience with CGM, including any benefits, problems, and drawbacks to using the sensor.” The analysis broke down responses into six themes (metabolic control, life on CGM, CGM procedures, technical issues, financial issues, attitudes to CGM), with several sub-themes and example quotes. Overall, patient responses to CGM were “overwhelmingly positive.” A1c and/or hypoglycemia were usually reduced on CGM, and quality of life usually improved, particularly in those with problematic diabetes. The survey did find that there are still significant problems and hassles with CGM, though most patients understand and accept the limitations. For many patients, CGM was described as “life-changing” and sometimes even “life-saving.” Most patients were on Medtronic systems. More complete details are below. We most appreciated Dr. Irl Hirsch’s comment in Q&A, which pointed out that such glowing real-world feedback could be useful in reimbursement discussions – particularly because clinical trials have not shown hugely striking benefits to using CGM. Indeed, this was one of the main reasons Dr. Pickup undertook the study. We thought this was a brilliant point and hope this type of real-world, anecdotal experience data could be used all over the world to support better reimbursement of diabetes technology, particularly as the field advances towards automated insulin delivery.

  • One hundred UK patients on CGM participated in this online survey – 50 were children (mean age: 10 years) and 50 were adults (mean age: 44 years). Nearly all participants (87%) were on CSII, with 2% on MDI and 11% on “both MDI and CSII.” Most patients (71%) reported wearing CGM 75-100% of the time. We did not catch all the details on the CGM brand slide (it flashed by very quickly), though the vast majority of participants were on some model of Medtronic CGM (38% were specifically on Veo/Enlite). Two-thirds of participants received NHS funding for CGM and one-third self-funded CGM.
  • The analysis broke down responses into six themes: metabolic control, life on CGM, CGM procedures, technical issues, financial issues, and attitudes to CGM. Dr. Pickup presented several sub-themes and example quotes within each theme – they are listed below in the order of the aforementioned major themes.

Sub-themes

Quotes

Lower A1c

Less blood glucose variability

Detection and prediction of hypoglycemia, reduced frequency and severity of hypoglycemia.

“I have not had a severe hypoglycemia for four months, and only one when I needed medical help in 18 months of usage. Previously, I was in the hospital 2-3 times per month.”

“Her A1c has dropped a little (probably 0.5-1% on average), but we are happier that it is achieved with much less variation in blood glucose levels.”

Hypoglycemia alerts give confidence whilst driving and at work. Independence at school, but for some it might disrupt school life.

Reduced stress for patient and caregiver. Reassurance and security, more confidence, independence, improved energy, mood, and quality of life.

Stress of viewing poor control and perceived failures to manage diabetes; obsession with data.

Easier to sleep and feel safe at night; can be a lifesaver and reduces fear of nocturnal hypoglycemia. But alarms can disturb sleep; lying on sensor can produce aberrant readings.

Better control during exercise and sport; need to avoid damage to sensor during exercise.

Encourages less snacking, different bolus profiles for meals, and different bolus timing. Reduces postprandial glucose levels.

Often reduces number of SMBG tests.

 

“She said it is like having a mummy in her pocket.”

“The psychological impact was huge for us and our son. We feel that we can listen to our son’s needs much better and he feels more in control of his own body. He loves it and says it is his best friend. It gives him a voice.”

“It’s easy to become obsessed with the inevitable spikes.”

“The low suspend is a life-saving piece of equipment and I would never be without that. It is a no brainer really that this is superior to anything else at the moment on the market - it saves lives.”

Use of trends and patterns provides a fuller picture vs. SMBG; helps with adjustment of basal and bolus insulin.

Many hospital staff are supportive and helpful. Others have poor knowledge about CGM and training.

“We see the full picture, which gives us a much better understanding of his diabetes.”

Not accurate and reliable enough.

Lag time is usually known and knowledgeable.

Lifetime can be shorter than the expected 6-7 days.

Time of calibration is important.

Predictive alarms and trends arrows are valuable; most would like louder alarms.

Insertion is uncomfortable/painful; sensitivity to tape and adhesion.

Difficulty of interpreting data/graphs; time needed to review and think about data.

Wearing an extra piece of equipment.

“I have had mixed experiences using my CGM. Sometimes it is close to my blood glucose reading and others it is massively out.”

“The alarms sometimes drive me mad.”

Expensive for those self-funding; lack of routine NHS funding is annoying.

“I love CGM, but I am frustrated that the NHS won’t fund it for me. I worry that I may have to stop using my CGM when finances get tight.”

Some HCPs are supportive and understanding; other have negative reactions and consider it to be a waste of money or untested or unusable in <18 years.

Most patients understand and accept the limitations; it’s hard work, but worth it.

Most have overwhelmingly positive view of CGM.

“CGM, if understood for its limitations, is the best tool to allow for quality of life and proactively managing your diabetes.”

“CGM has changed our lives and how I manage.”

 

Questions and Answers

Dr. Irl Hirsch (University of Washington, Seattle, WA): I found the comments very compelling and similar to what we see in our CGM patients. I’m wondering how much impact this will have on NHS and reimbursement. In our RCTs, we try to show what we can, but these comments are more impactful…

A: I was on the NICE Committee that reviewed and eventually approved insulin pumps. The NICE committee took a lot of notice of patient anecdotes and views. Some of the HCPs suspected they took more notice of that than the science and RCTs. Also, there was a tendency – surprisingly – to take quite a large amount of notice of observational studies. I do sense there is a movement towards looking at all data in this area. Sometimes randomized controlled trials have not been done particularly well. They don’t tell the whole story. Sometimes committees are looking for other ways to make decisions. That was one of the reasons why we did this study – we thought that might be the case to persuade NICE approval type bodies in going that direction [with CGM].

Q: Have you validated some of the response – A1c and glycemic variability – with meter downloads?

A: We specifically made the study anonymous so we couldn’t check against the medical records. The other point was that this was a UK-wide survey, so we didn’t have regular access to records. I take the point – there is lots one could do. With follow-up interviews we may try to sort out things in detail.

Dr. Yogish Kudva (Mayo Clinic, Rochester, MN): Have you had a chance to look at whether the geographic location of subjects had any impact? Was there a difference between more specialized clinics vs. less specialized clinics?

A: That’s very interesting. We cannot tell location by and large from the responses. I wouldn’t be surprised. From the pump data, there is a geographical difference in outcomes. You also have to take into account the level of deprivation – not only the level of care. It’s a very complicated mixture when we’re thinking about geography. That would be very worthwhile to look at.

Q: Of those on Veo, was there anything that emerged for those on LGS vs. not?

A: We haven’t looked in that sort of detail. Sometimes you are not able to tell if they are on an LGS pump unless they specifically said that. It would be worth going back and looking at it.

Dr. Roman Hovorka (Cambridge University, UK): Half of the population was children with diabetes. But given the mean age of 10 years, it was the parents filling out the survey. Were there differences between children and parents’ comments?

A: We haven’t looked at it in that kind of detail. One would suspect qualitative differences in parent vs. patient responses. That would be interesting to look at that. Our idea about this is to go back and do more detailed individual interviews with some of the patients.

Dr. Hovorka: We have experience as well - there is a difference with children and parents.

 

CURRENT STRATEGIES FOR THE TREATMENT OF THE SEVERELY INSULIN RESISTANT PATIENT

Irl B. Hirsch MD (University of Washington, Seattle, USA)

This very practical talk by the legendary Dr. Hirsch focused on the obese individual and the use of concentrated insulin. The interest in U-500 insulin is being driven today by the increase in obesity. Typically, it’s used for obese patients taking more than 200 units/day. But the first issue for the clinician is to understand the etiology of insulin resistance. If it’s syndromic, then it might be possible to treat the root cause and/or use concentrated insulin. For non-syndromic obese patients requiring >1 unit/kg/day, Dr. Hirsch recommended switching from analog insulins to NPH twice daily with different sites. That’s because gains in insulin action get attenuated at higher doses and that a smaller sub-cutaneous depot gives more consistent results. It is also much cheaper. For those patients requiring >2 units/kg/day then U-500 is recommended. There isn’t a lot of hard evidence for the use of U-500 at present, but A1c and weight gain do not seem to be affected by concentrating the insulin. Patients really appreciate the smaller volumes. Furthermore, observational evidence suggests that pumping U-500 can lower A1c, even though it is a regular insulin. Trials would be welcome.

  • Today, interest in U-500 is driven by the increase in obesity, and it’s used in obese patients taking >200 units/day, where U-100 becomes impractical and inconvenient. Large volumes of insulin can be painful to inject or infuse. Patients require multiple shots, and pumpers require daily set changes. In the early 1950’s U-500 insulin was developed as a response to insulin antibodies from animal insulin. Today it can be used in non-syndromic insulin resistance, obesity with >200 units/day, post-operative or post-transplant, with high doses of steroids, systemic infection or pregnancy with underlying type 2 diabetes.
  • When treating patients who take large volumes of insulin, it’s first important to understand the etiology of the insulin resistance and whether it’s syndromic or non-syndromic. Syndromic insulin resistance includes lipodystrophy and a range of other rarer causes. These include genetic causes, endocrinopathies, insulin receptor antibodies (Type B insulin resistance) and mitochondrial diabetes.
  • Dr. Hirsch recommends that non-syndromic patients taking >1 unit/kg/day of insulin glargine (Lantus, Sanofi) should switch to NPH insulin twice per day at two different sites. Once patients take over 1 unit/kg/day there is not an increase in insulin action over a 24-hour period with U-100 insulin glargine, probably linked to the larger sub-cutaneous depot. The larger the depot, the more variable the absorption, so taking more injections per day is preferable. NPH is also a lot cheaper than an analog insulin and there is little risk of hypoglycemia in these patients. Adding GLP-1 can also be a real benefit (by reducing weight, insulin dose and A1c). Patients should also continue taking rapid acting insulin analogs at mealtime.
  • U-500 insulin is recommended for non-syndromic patients taking >2 units/kg/day, and it seems to perform well. There are no randomized controlled trials but in a review, A1c dropped from 10% to 8.4%, pumping made no change in A1c, and there was no difference in weight from baseline. Data also suggests that it can be used with good glycemic control and without excessive weight gain for at least nine years. There is no data regarding U-50o and carb counting versus set doses at meals, although most type 2 patients who take set mealtime doses do well. Correction doses remain more controversial, since U-500 is a regular insulin.
  • U-500 can also be used successfully in pumps. In a small (n=37) observational study, A1c was reduced from 8.6% to 7.4% after one year. Rapid acting analogs can also be added at mealtimes. 

Questions and Answers

Q: You showed Show NPH vs glargine. But I was wondering if any studies have looked at detemir?

A: I have not seen data on detemir, but I expect it exists. We may see more studies on analogs, to see where we get decreasing returns versus NPH. I am also very interested to see what happens with the new U-300 [glargine].

Q: What is the best argument to the insurance company to approve U-500?

A: The real issue comes down to the individual. But hypoglycemia prevention works most of the time, also future planning of pregnancy. But insurance is a huge struggle.

 

PATIENTS BARRIERS TO THE USE OF PUMP AND SENSORS

Ragnar Hanas, MD, PhD (Uddevalla Hospital, Uddevalla, Sweden)

Dr. Hanas gave a packed presentation on the many factors that cause patient dissatisfaction with pumps and CGM. Rates of pump discontinuation are relatively low, particularly when patients initiate pumps at diagnosis. The main reasons that patients discontinue their pump include ‘burnout’ (operator fatigue from operating the pump, changing sets etc), site problems and body image issues. Interestingly, many patients said they would go back if pumps were improved (by making them smaller, less visible and having better needles). For CGM, patients tend to use fewer sensors over time, but discontinuation is also relatively low. The key reasons for discontinuing are accuracy, bad sensors, alarms, cost and pain/irritation at the site. There are also body image issues and a ‘confusion factor’ with the therapy. Although we didn’t include it in the notes, Dr. Hanas also included an excellent practical section on skin care – a frequent concern with infusion sets and sensors.

  • There are many studies on rates and reason for pump discontinuation. In one US study, 18% of young people discontinued pumps over a 0.1-4.5 year period. In another study, the majority of people who discontinued said that they would go back if the pumps were improved. Most requested improvements were a smaller, less visible pump and better needles. Typical reasons for stopping the pump included ketoacidosis, burnout, site problems, body image issues and weight gain. People who stopped were older, with less support at home and had more hypoglycemia. In another study, reasons given included painful insertion, pump visible to others and scarring.
  • If patients start on the pump at diagnosis of diabetes, then the dropout rates seem to be much lower (around 1%). In a study of 160 children who started CGM and pumps at onset of diabetes, A1c was not significantly improved (although the data was suggestive that it was) compared to pumps alone, and there was a significant decline in CGM use. However, pump and CGM dropout rates were lower than the other studies (4%).
  • The key reasons patients cite for discontinuing CGM are accuracy, bad sensors, alarms, cost and pain/irritation at the site. Patients are also frustrated with continuing blood glucose measurements, that the CGM and the SMBG don’t match, by the alarms caused by inaccurate readings, and by continuing alarms after action was taken (due to lag time). There are also body image issues and a ‘confusion factor’ with the therapy. Dr. Hanas recommended that the CGM is set to limit alarms to 2-3 times/day at most and the snooze should last 2-3 hours for high blood glucose and 30 minutes for low glucose.
  • Reimbursement is very different across health care systems and can be approved based on many varied factors. These include severe hypoglycemia, an unsatisfactory A1c, number of tests per day, pregnancy and pediatric use.

 

TIME LAG OF GLUCOSE BETWEEN INTRAVASCULAR AND INTERSTITIAL COMPARTMENT; IMPLICATIONS FOR GLUCOSE SENSING

Ananda Basu, MD (Mayo Clinic, Rochester, MN)

Dr. Ananda Basu presented results from an important and very detailed study examining the physiological time lag of glucose transferring between the intravascular and interstitial fluid spaces. The study was published in December in Diabetes. The study concluded that in the overnight fasted state in adults without diabetes, the physiological delay of glucose transport from the vascular to the interstitial space is 5-6 minutes, shorter than was previously thought. Following an insightful question from Dr. Roman Hovorka, Dr. Basu conceded in Q&A that “equilibration time” may be what matters more for closed-loop control. Dr. Claudio Cobelli is currently modeling this aspect of the data, and the plan is to present it at ADA.

Various Algorithms in Order to Cope with Glucose Control

SAFE CONTROL ALGORITHMS TO PERSONALIZE THE OUTPATIENT ARTIFICIAL PANCREAS

Francis J. Doyle III, PhD (Sansum Diabetes Research Institute, UCSB, Santa Barbara, CA, USA)

Dr. Doyle, a self-confessed control engineer, showed impressive results of closed loop outpatient clinical studies at UVA and Sansum. The focus was on model predictive control (MPC) algorithms, which Dr. Doyle has been working on for twenty years. MPC was established for the chemical industry. There, a process control system is surrounded by a number of safety layers, each successive layer less likely to be needed. In the same way, safety layers in the MPC algorithm for diabetes include constraints on insulin delivery, glucose zones (which can vary throughout the day), prediction of low glucose and detection of glucose rate of change. Results for five day outpatient experiments (in Charlottesville and Santa Barbara) were shown for n=10 patients. The data was excellent – time in zone under closed loop control was 75%, and a superb 97% overnight (but only data for a few subjects). Also, Dr. Doyle showed a fascinating and instructive movie of how the algorithm predicts glucose 45 minutes ahead and takes the appropriate action to enable the right outcomes. This work shows the enormous potential of closed-loop technology and the fact that it now works safely and effectively in real-life conditions.

  • The four main components of a model predictive control (MPC) algorithm are the predictive model, the constraints, the cost function and the dynamic optimizer. On top of this, an independent safety overlay ensures that the system is behaving correctly and minimizes risk to the patient. The key concept is that the control system is surrounded by concentric safety layers, each of which is intended to be used less frequently than prior layers.  The MPC system can include safety constraints, such as ‘hard’ constraints on insulin delivery (based on insulin on board and time of day) or ‘soft’ constraints on glucose (target glucose zones that can change by time of day, loosening up at night). The safety layer also contains a low glucose predictor (LGP) and glucose rate of increase detector (GRID) that can predict hypoglycemia and hyperglycemia and take appropriate action.
  • The MPC model can be tuned to the individual patient. The key aspect of this is fine tuning insulin sensitivity, which can be done over a short period of time by monitoring pump and CGM data and taking into account personal medical data and diaries. This is known as the adaptive advisory system.
  • Dr. Doyle described a control to zone clinical trial, which tested safety control algorithms outside the clinic. This is a five-day outpatient trial of a portable artificial pancreas system (pAPS), consisting of 4 hours of inpatient system setup and testing and 26 hours of outpatient closed loop control. The system was based on a portable computer and the Animas OneTouch Ping pump and Dexcom G4 Platinum CGM. The results presented were pilot data for ten participants, testing a safer design with variable zone MPC and asymmetric weights (faster actions for low glucose than high glucose) and a patient adapted baseline using the adaptive system
  • A superb movie demonstrated how the algorithm looks 45 minutes ahead and then optimizes ongoing insulin delivery to ensure that the patient stays in the zone. The algorithm alternately gives and suspends insulin based on its predictions, and the patient gives meal boluses as usual.
  • The pilot results from the study were very impressive – patients experienced 75% time in range (70-180 mg/dl) and 97% time in range overnight, admittedly with a low number of visits. Time spent below 70 mg/dl was only 4%.
  • Two future studies (funded by NIH) of 72 and 84 visits will be conducted by Summer 2014. There will be three sites, and each patient will make two visits.

Questions and Answers

Dr. Richard Bergenstal (International Diabetes Center, Minneapolis, MN): With being able to predict and refine carb and insulin ratios, in your studies were there no manual boluses with meals?

A: No, there were – patients gave their usual boluses. What we adapt and refine is the carb ratio for the basal insulin.

Q: How do you deal with sensor errors?

A: The advantages of having a zone are that noise from the sensor gets attenuated and filtered. Noises within the zone are effectively erased, although the noise that pushes you out of the zone, that’s a different matter.

Q: How do you manage patient expectations? This might be really exciting but a little bit scary.

A: This is an interesting challenge, but in our experience in Santa Barbara we have a smallish community, a dozen or two subjects who get acclimated to working with engineers. So we don’t have to worry about that. But we do have a behavioral component in the grant to study not only apprehension with the technology but also how to design the best user interfaces.

 

EVOLUTION OF A TREAT-TO-RANGE ALGORITHM

Bruce Buckingham MD (Stanford University School of Medicine, CA, USA)

The superb Dr. Buckingham told the tale (complete with Dr. Seuss analogies) of the development of a treat-to-range algorithm. In treat-to-range controllers, the algorithm doesn’t become active unless the patient’s glucose falls out of range (100-160 mg/dl). This handicaps the algorithm from providing tighter control – instead it’s acting more like a ‘safety net’. Of course, Dr. Buckingham gave the clear impression that he’d like to see the technology graduate to full time closed loop control. In this presentation, Dr. Buckingham described the evolution of treat-to-range algorithms via in silico studies of glucose excursions, canine studies and inpatient pilots. In the most recent cohort study, the algorithm had a significant effect in mitigating the effects of missed meal boluses or over-bolusing by 20%, typically avoiding hyperglycemia and hypoglycemia. Although the results were not perfect, they appeared far superior compared to doing nothing, and safety was greatly improved. But it’s difficult to do any better with mismanaged meals, since insulin delivery is deliberately delayed until glucose reaches a hyperglycemic threshold. In his summary, Dr. Buckingham noted that breakfast is a difficult meal to handle, that with better sensors the buffer zone may not be necessary (so the algorithm can operate all the time), and that (in work not presented), very good overnight closed loop control is possible.

  • In treat- to-range controllers, the algorithm doesn’t become active unless the patient’s glucose falls out of range  (typically 100-160 mg/dl). This ‘handicaps’ the algorithm from obtaining tight control, but instead the controller acts as a safety net. Unlike full-time closed loop, this approach provides a ‘buffer zone’ in case of incorrect sensor readings, and doesn’t interfere with patients who can obtain good control themselves.
  • The development of a treat-to-range algorithm started with in silico work, moved to canine studies, and then to inpatient pilot studies. It was funded in large part by the JDRF and NIH, and involved groups at UVA, Padova, Stanford and Sansum. Radio-labeled meals allowed the development of in silico and canine studies of missed meals. In turn, these gave confidence that inpatient studies could be performed safely. In the first such studies, patients bolused at breakfast and were not given insulin at lunch, yet the PID algorithm did a reasonable job - despite being constrained on insulin delivery rates and glycemic thresholds. Parameters were tuned as a result.
  • JDRF then sponsored a multi-center cohort trial with support from Medtronic to further refine the approach. Each of the three cohorts consisted of nine subjects and each cohort was intended to be repeated if safety criteria based on glycemic range and ketones were not met. The trial tested a missed meal bolus at lunch, and 120% overbolus at lunch. Both experiments controlled the occurrence of hyperglycemia and hypoglycemia with reasonable success. For missed meal, peak post-prandial glucose was around 240 mg/dl and for overbolus the nadir was 60 mg/dl (50 mg/dl is the safety threshold).
  • Dr. Buckingham concluded that breakfast is a difficult meal even in hybrid mode, that with better sensors the ‘buffer zone’ may no longer be necessary (ie the algorithm can operate continuously) and that overnight closed loop control can be obtained with a PID algorithm.

Questions and Answers

Dr. Richard Bergenstal (International Diabetes Center, Minneapolis, MN): What will it take for us to feel comfortable in predicting hypoglycemia and so have the algorithm kick in earlier? Is it about better sensors, or other safety considerations?

A: Just using insulin to cover a meal with today’s typical insulin analogs is not enough. You won’t get better coverage until you have something to protect you on the back end like glucagon, or you use a faster insulin. If you had more reliable sensors, then you could have the insulin kick in earlier. But you need to be able to tell when the sensor fails. You can’t have a sensor that’s gone erratic and then deliver insulin on that. So you need a way to detect a sensor failure without a patient pricking their finger. The system would then revert to open loop control. You particularly need to deal with errors for overnight control - patients don’t need to be woken all the time for different errors.

 

PREVENTION OF NOCTURNAL HYPOGLYCEMIA USING PREDICTIVE LOW GLUCOSE SUSPEND (PLGS)

Peter Chase, MD (Barbara Davis Center, Aurora, CO)

Dr. Peter Chase presented on an impressively large 1,912-night study testing predictive low glucose suspend (PLGS) in the home setting. The trial randomized 45 patients to either PLGS on (942 nights) or PLGS off (970 nights) – the algorithm (30-minute prediction horizon) conducted the randomization each night, allowing each patient to serve as his or her own control. (Note, this is not Medtronic’s PLGM algorithm.) The system does not use alarms, as the goal is “to have people sleep at night” – excellent! Notably, the percent of nights with a glucose <60 mg/dl was 21% with PLGS on vs. 33% with PLGS off (a 36% reduction). In addition, AUC <60 mg/dl was 81% lower on PLGS nights compared to control nights. (This puts into perspective just how much better predictive suspend is than threshold suspend, as ASPIRE in-home found a 38% AUC reduction for episodes <65 mg/dl). Time in range (70-180 mg/dl) significantly improved when PLGS was turned on – 82% vs. 75% (p<0.001). PLGS halved the number of nights with a glucose <60 mg/dl for more than 30 minutes (12% vs. 24%; p<0.001). On the safety side, A1c did not change in the study, and the likelihood of developing ketones did not increase when PLGS was turned on. With ~2,000 measurements, this represents the largest ketone outpatient study ever done in the world. There was slightly more morning hyperglycemia, though the incremental tradeoff for significantly less nocturnal hypoglycemia is worth it in our view. Overall, we found the results highly encouraging and wonder if the team will drive to a pivotal study to take this algorithm to market.

  • Overnight median glucose was slightly higher on PLGS nights (132 mg/dl) vs. control nights (125 mg/dl), though this was attributed to less time spent in hypoglycemia and not an increase in hyperglycemia. Median morning blood glucose was 144 mg/dl with PLGS on vs. 129 mg/dl when PLGS was off (p<0.001) – said Dr. Chase, “I would be pleased if my teenagers woke up at 144 mg/dl.” There were slightly more mornings with a glucose >180 mg/dl when PLGS was used – 27% vs. 21% (p=0.001). However, there was no significant difference in the number of mornings with a glucose >250 mg/dl. Given the reductions in hypoglycemia shown in this study, we think the small tradeoff for slightly higher morning glucose levels is worth the safety benefit.
  • Median pump shutoff duration was 71 minutes, and notably, at least some pump suspension occurred on 76% of nights when PLGS was turned on (a clear reminder of just how often hypoglycemia occurs). The length of pump suspensions was fairly evenly distributed, with 20% lasting 1-30 minutes, 12% lasting 31-60 minutes, 28% lasting 61-120 minutes, and 17% lasting >120 minutes.
  • To be clear, this study did not use the Medtronic PLGM algorithm that will be incorporated into the MiniMed 640G pump (see Dr. Danne’s talk on the PILGRIM study). The trial used a Medtronic pump, CGM, and a bedside laptop running the control algorithm. We assume the Enlite CGM was used, though Dr. Chase did not specify.

 

PREDICTIVE LOW GLUCOSE MANAGEMENT WITH SENSOR AUGMENTED CSII IN RESPONSE TO EXERCISE

Thomas Danne, MD (Auf der Bult, Hannover, Germany)

Dr. Thomas Danne reviewed the results from the PILGRIM study testing Medtronic’s predictive low glucose management (PLGM) system, first presented at ADA 2013 (357-OR). Twenty-two adolescents (mean age: 15 years; mean A1c: 8.0%) exercised with the PLGM system to induce a drop in blood glucose. Of the 16 patients who reached the hypoglycemic threshold for PLGM activation, predictive suspension was successfully activated in 15 of the experiments and prevented hypoglycemia (reference blood glucose ≤63 mg/dl) in 12 of the 15 experiments (80% of the time – “a big success”). Mean sensor glucose at predictive suspension was 92 mg/dl, mean suspension time was 90 minutes, and mean sensor glucose upon insulin resumption was 97 mg/dl. As a reminder, international launch of Medtronic’s MiniMed 640G (new pump platform with PLGM) is expected by July 30, 2014 (see our JPM 2014 coverage). Dr. Danne concluded his presentation with his vision of the next key step in the field: combining overnight closed-loop control with daytime PLGM. He shared new preliminary data on hypoglycemia alarms from the DREAM 4 at-home pilot #2 study (overnight closed loop with MD-Logic vs. sensor-augmented pump (SAP) therapy) – under nocturnal closed loop (162 nights), patients experienced 50 hypoglycemia alarms, less than half of the 113 alarms experienced on SAP (160 nights). Full results will be shared later in the meeting, which we assume will include glucose data.

  • The PLGM algorithm used a 30-minute prediction horizon and suspended insulin delivery when a predictive threshold <80 mg/dl was reached. AS background, the PLGM algorithm suspends basal insulin for at least 30 minutes and for a maximum of two hours – the resumption of insulin delivery between 30 minutes and two hours is variable and is based on current and predicted sensor glucose.
  • The PILGRIM study used a Medtronic Veo pump, an Enlite sensor, a smartphone controller running the algorithm, and communication translator. As a reminder, Medtronic’s MiniMed 640G with PLGM will use a new pump platform that includes a simpler user interface, a color screen, and a waterproof design. For more details, see our Medtronic F2Q14 report.

Questions and Answers

Q: Regarding exercise, are these algorithms applicable to daily life?

Dr. Danne: Exercise is the biggest challenge a diabetologist has. Some of the kids had enormous rises in blood glucose because they were so stressed on the bike with two good-looking nurses watching them. Others had rapid falls in blood glucose that were completely unexpected. Anything that can improve management of exercise is a good thing. Computer modeling will probably be helpful, along with advanced dose calculators with individualized predictions. 

Dr. Yogish Kudva (Mayo Clinic, Rochester, MN): Out of the 22 subjects that exercised in the study, 16 seemed to have glucose lowering and six did not. Do you have insights into the differential responses? Did it have to do with baseline fitness or V02 max?

Dr. Danne: No. All my predictions didn’t work out. The super well trained hockey player and the untrained computer nerd had the same response. The individual characteristics were very difficult to predict. We haven’t made out what characterizes those who didn’t reach the low threshold from those who did.

Dr. Satish Garg (Barbara Davis Center, Aurora, CO): A similar case happened in ASPIRE. We cannot say which individuals will not achieve hypoglycemia despite all these bouts of exercise. 

 

Industry workshop: Optimizing Insulin Pump Use Through Human Factors Research (Supported by Tandem Diabetes Care)

PRELIMINARY ANALYSIS OF SELF-REPORTED OUTCOMES USING THE T:SLIM INSULIN PUMP

Nate Heintzman, PhD (UC San Diego, San Diego, CA)

Dr. Nate Heintzman quickly reviewed results from a survey of Tandem users on the A1c change they experience during approximately their first six months using the pump. The survey was emailed to 5,000 t:slim users, of which 289 completed the survey and included their A1c data. After a mean use period of about six months, the average A1c drop was 0.4% (baseline: 7.4%; p<0.001) among all respondents. In the experienced pump user cohort (n=211), the average A1c reduction was 0.2% (baseline: 7.1%; p<0.001) after ~6.2 months. The most robust A1c decline occurred in the new-to-pump group, who saw a 0.9% reduction in A1c after 5.3 months of use (baseline: 8.0%; p<0.001). Overall, increased use of t:slim features and software by previous pump users was associated with a greater though non-significant A1c improvement (details below), and over 50% of respondents indicated that they uploaded data before their doctor’s appointment. Of course, all of this data is subject to self-reporting bias, though it is directionally interesting. During the subsequent panel discussion, the session’s moderator Dr. Tim Bailey (Advanced Metabolic Care and Research, Escondido, CA) highlighted this last result, emphasizing that few offices, including his own, can claim such a pre-appointment data upload rate. Indeed, when Dr. Bailey asked HCPs in attendance about their practice’s pre-appointment data upload rate, nobody indicated that at least 50% of his/her patients upload data before their appointment.

  • The analyzed survey was emailed to 5,000 t:slim users in September 2013. Responses were received from 749 users. Of these, 411 people completed the survey, including 289 people who reported their A1c data. A1c-inclusive respondents had a baseline A1c of 7.4%. Fifty-two percent of respondents were female, the average participant was ~40 years old, and 70% (n=211) had previously been on a pump. Among experienced pump users, the previous-brand breakdown was 32% formerly on a Medtronic pump, 21% formerly on Animas, 14% formerly on Deltec, 4% formerly on OmniPod, 1% formerly on Roche, and 1% formerly on other brands. This was particularly fascinating, as Medtronic has about two-thirds of the overall pump market – this suggests that Tandem may be taking a disproportionately higher share of current pumpers not on Medtronic.
  • During Q&A, Dr. Heintzman expounded on the A1c data, noting that this trend was not significantly impacted by a person’s gender or age. Dr. Heintzman highlighted that the latter result speaks to the intuitive usability of the t:slim, since people who were older (and thus likely to be less experienced with touchscreen technologies) benefited equally from the t:slim as those who were younger (and thus more likely to be frequent users of touchscreen technologies). 
  • Overall, increased use of t:slim features and software by previous pump users was associated with a greater A1c improvement, though the detected trends were not statistically significant. Patients who used the bolus calculator daily (n=170; 59% of A1c-inclusive respondents) had an average A1c drop of 0.26%, more than double mean reduction of 0.12% among those who used it weekly and more than triple the 0.07% reduction for those who never used the bolus calculator (a statistically significant trend). Looking at the impact of use of the t:slim’s extended bolus on A1c, a similar though not quite as dramatic trend was seen: people who delivered an extended bolus daily (n=37; 13%) had an average A1c drop of 0.27%, weekly (n=129; 45%) had an A1c decline of 0.24%, and never (n=44; 15%) had an A1c reduction of 0.14%. Reflecting on this result, Dr. Heintzman suggested that t:slim educators might need to focus more on the extended bolus to try and increase use of the feature.
  • Over 50% of survey respondents indicated that they uploaded data before their doctor’s appointment; more frequent uploads were associated with a better A1c. Specifically, people who uploaded their pump data weekly using the t:connect software (n=27; 13%) had an average A1c drop of 0.32%, slightly greater than the 0.28% decline among those who did so monthly (n=56; 26%); those who did so just before their appointments (-0.29; n=46; 22%); and those who never uploaded (-0.12%; n=82; 39%). Perhaps unsurprisingly, this data suggests that the greatest A1c benefit comes from uploading data at least once before a doctor’s appointment, and that additional uploads have a reduced marginal benefit.

Questions and Answers

Dr. Bailey: Was there any A1c trend by gender or by age?

A: We did take a look at that and found that regardless of the gender or age of the respondent, there was no significant difference in the reduction in A1c. I think that speaks to the usability of the interface even for those who did not grow up with an iPhone or iPad.

Dr. Bailey: What portion of respondents did not download the data? As we treat more people with diabetes it is important that we have easy access to their pump data.

A:  More than half the patients reported downloading their data at least once before their doctor appointment.

Dr. Bailey: Are their any practitioners in the audience who have more than half of their patients uploading data for any device? [No hands raised.]

 

Oral Presentations

Lori Laffel (Joslin Diabetes Center, Boston, MA)

Dr. Lori Laffel announced initial results from the much-awaited TEENS Study, which assesses type 1 diabetes management and psychosocial parameters in youth aged eight to 25 years. The study is the largest, contemporary, cross-sectional study of its kind (and was sponsored by Sanofi). The results presented today were based upon data from 1,000 participants in the US and Russia (n=500 for each country). The majority of youth with type 1 diabetes in the US (75%) and Russia (84%) are not at their A1c target, and experience acute complications at relatively high rates. These disappointing results are despite the fact that many youth with type 1 diabetes in the US and Russia use good management tools like frequent blood glucose monitoring (on average about four times a day) and pumps (63% in US and 19% in Russia). However, Dr. Laffel maintained some optimism, pointing to the success of some participants as evidence that it is possible to reach glycemic targets, despite facing many challenges. Dr. Laffel hopes that the further analysis of the TEENs Study will offer recommendations on how to improve type 1 diabetes management among the youth, and during Q&A indicated that the study group hopes to present results from the broader trial at ADA.

  • The TEENs Study is the largest, contemporary, cross-sectional study assessing type 1 diabetes management and psychosocial parameters in youth aged 8-25 years old (n=5,960 patients in 20 countries). The study’s inclusion criteria are 1) being 8-25 years old, 2) having been diagnosed with type 1 diabetes before one turned 18 years old, 3) having had type 1 diabetes for at least one year, 4) receiving care at a clinic or hospital that treats at least 100 people with type 1 diabetes, and 5) having not undergone a major change in insulin regimen (i.e., going from pump to MDI or vice versa) during the past three months.
  • The results Dr. Laffel presented were based on data collected at 25 US and 20 Russian centers (n=500 people from each country). Data was collected by surveys, interviews, medical record reviews, and/or questionnaires completed by participants, parents, and centers. Participants were stratified by three age groups - 8-12 year olds, 13-18 year olds, and 19-25 year olds. The burden of diabetes was assessed with the PAID (20-item) questionnaire for youth 13 years and older, and PAID-PR (18-item) for parents of children 18 years or younger.
  • In this sub-study, about half of the participants were female and, as expected, the majority were Caucasian. Duration of diabetes increased with older age groups, and 13-18 year olds needed the highest total daily insulin dose (people are more insulin resistant during puberty). Additionally, American youth tended to be heavier than those living in Russia.
  • The majority of youth in the US (75%) and Russia (84%) with type 1 diabetes were not at their A1c target. As a reminder, the A1c recommendation for a person 18 or younger is <7.5% and for a person over 18 is <7.0%.

 

 

8 – 12 years

13 – 18 years

19 – 25 years

 

US

Russia

US

Russia

US

Russia

Median A1c (%)

7.9

8.8

8.4

9.0

8.2

8.0

% at A1c target

32

27

25

15

16

18

 

  • Generally poor glycemic control was associated with high rates of acute complications (i.e., diabetic ketoacidosis [DKA] and severe hypoglycemia). Unfortunately, DKA is particularly common among people 13-18 years old in the US and Russia. Severe hypoglycemia was defined by a person having a seizure or loss of consciousness.

 

Occurrence of DKA and severe hypoglycemia reported as percentage of persons with at least one event in the previous three months.

 

8 – 12 years

13 – 18 years

19 – 25 years

 

US

Russia

US

Russia

US

Russia

DKA (%)

3.1

7.7

5.7

8.5

1.6

3.2

Severe hypoglycemia (%)

1.5

3.1

0.4

0.8

3.3

N/A

 

  • Poor outcomes occured despite participants utilizing  management tools at a relatively high rate – demonstrating the difficulty of managing type 1 diabetes in the youth. Overall, the youth tend to check their blood glucose four times a day. However, both in the US and in Russia the frequency of BGM appears to decline with increasing age. The majority of participants in the US and Russia carb count, with the practice being more common in the US. Similar to BGM frequency, however, the proportion of youth in the US with type 1 diabetes who carb count declines as they progress from childhood to young adulthood. But only 16-27% of youth in the US and Russia exercise daily, with a greater proportion of children (8-12 years) and teens (13-18 years) being physically active in the US than in Russia, and a larger percentage of young adults (19-25 years) being physically active in Russia.

 

 

8 – 12 years

13 – 18 years

19 – 25 years

 

US

Russia

US

Russia

US

Russia

Daily exercise ≥ 30 min/day (% of participants)

41

15

24

13

16

24

Daily BGM frequency

7

5

4

4

3

3

Use of carb counting (% of participants)

95

66

86

52

80

64

 

  • Pumps had greater penetration in the US cohort (63%) than in the Russia subpopulation (19%). Dr. Laffel did not present the prevalence of CGM use; however, during Q&A she indicated that the technology was not very common in the study population.
  • Dr. Laffel characterized the reported burden due to diabetes as being comparable in the US and Russia, and noted that it was greater for parents than for youth. For background, PAID and PAID-PR scores can range from zero to 100 and higher scores indicate a greater burden. Patients aged 13 to 25 years reported a median PAID score of 15 in the US and 25 in Russia. Parents of children with type 1 diabetes aged eight to 18 years had a median PAID-PR score of 44 in the US and 53 in Russia (parents of people with type 1 diabetes older than 18 years did not complete a PAID-PR questionnaire). Parents’ burden did not appear to decline as their children became teenagers.

 

8 – 12 years

13 – 18 years

19 – 25 years

 

US

Russia

US

Russia

US

Russia

Median PAID score (patient)

N/A

N/A

14

23

16

28

Median PAID-PR score (parent)

45

56

46

52

N/A

N/A

 

Questions and Answers

Dr. Steven Edelman (UC San Diego, CA): Was there a difference in DKA in patients on pump and MDI?

A: We have not looked at the country level yet with respect to the difference in DKA rates. In the entire data set we will look at that question. Will hope to announce that result at ADA.

Dr. Edelman: What was the usage rate of CGM?

A: CGM is not used very often. We do not have that data yet from the Russian and US cohorts, but again, we might have that for the entire cohort results.

 

Technology in Movement: Innovation in Blood Glucose Monitoring Industry Symposium (supported by Sanofi)

SIMPLIFYING LIFE WITH DIABETES: DATA AS SERVANT AND NOT MASTER

David Klonoff MD (Mills Peninsula Medical Center, San Francisco, CA, USA)

Dr. Klonoff made the case for better using data to improve glycemic control. He made ten claims, in which he advocated for SMBG, using structured pre- and post-meal measurements. He also believed in using decision support software such as bolus calculators, not least because it can be challenging to do all the mathematics associated with figuring out the right bolus. He noted that patients need training in carb counting – they tend to significantly underestimate with the higher carb meals. Treat to target regimes work well for improving glucose control and also can be automated. Dr. Klonoff reiterated that A1c informs long-term control and that SMBG readings can be used to calculate A1c. Finally, he warned us that many approved meters don’t conform to the appropriate standards, (although the Sanofi meters did). His take-home message was that patients should perform the tests correctly with accurate tools, and then react appropriately to the data.

  • Dr. Klonoff provided ten examples of how data can help patients improve their outcomes. They were:
    • Knowing blood glucose allows therapy adjustments
    • Structured intervention is needed to use data. Structured testing means obtaining pre- and post-meal SMBG. Brackets before/after the same meal are particularly useful. Pattern analysis can then identify any abnormality, their frequency and their causes. Finally, changes can be made to diet, exercise, stress, insulin, oral agents.
    • Decision support software helps manage blood glucose. Decision support software provides patient-specific advice. It’s important to get the data
    • Dose calculations improve patient outcomes. Patients make errors in bolus calculations 63% of the time. Accordingly, glucose meters with automatic bolus calculators can be very valuable to patients.
    • Poor numeracy limits the use of numbers. The math required to count carbs and calculate boluses is actually quite tricky, and mistakes are made.
    • Poor carb/weight estimates limit the use of numbers. Estimates of carbs by patients did not climb as fast as actual carb content in a test of eight meals of increasing carb content. Patients The highest carb meals
    • Treat to target regimens can improve fasting levels. A1c usually decreases with treat to target regimen of long acting insulin. Although the risk of hypoglycemia can increase, titration can be automated to reduce risks.
    • A1c numbers inform overall control. The DCCT was clear on this point.
    • A1c levels can predict mean blood glucose and vice versa.  The relationship between A1c and estimated average glucose has been ascertained. Dr. Kovatchev has outlined a method of inferring A1c from blood glucose data.
    • SMBG numbers can be misleading. Many approved blood glucose monitors do not perform to the standards.
  • Patients will do well if they (1) do the tests, (2) use accurate tools, (3) operate the tools correctly and (4) react appropriately to the data.

 

Innovation in the Treatment of Diabetes

CAN WE ACHIEVE TIGHTER CONTROL WITH BASAL INSULIN? (LECTURE SUPPORTED BY NOVO NORDISK)

Bruce Bode, MD, FACE (Atlanta Diabetes Associates, Atlanta, GA)

This lecture was supported by Novo Nordisk, and Dr. Bode did a very good job of presenting the high points of Novo Nordisk’s insulin degludec (Tresiba). As a reminder, insulin degludec is a very long-acting basal insulin analog that has an extremely flat profile. In treat-to-target clinical trials against glargine, A1c was non-inferior and hypoglycemia was generally lower, particularly at night. Dr. Bode also discussed the DEVOTE cardiovascular safety trials required by the FDA for approval, and the strong results for iDegLira, a fixed ratio combination of insulin degludec and liraglutide.

  • There is data to suggest that 80% of patients are not at goal. The reasons include hypoglycemia, patient attitudes to insulin and adherence. Hypoglycemia continues to be a main obstacle for HCPs to efficiently treat patients with insulin. We have improved how far we can lower A1c at the same level of hypoglycemia, but hypoglycemia and insulin administration still greatly affects patient attitudes. 28% of patients said they have issues with timing of insulin, and 33% miss a shot at least one day a month. Additionally 73% of HCPs say their patients miss shots on a regular basis. The cost of severe hypoglycemia is a major economic burden, and moderate hypoglycemia affects productivity in the workplace.
  • Insulin degludec (Tresiba, Novo Nordisk) is a new form of insulin which self-associates into long chains of hexamers which then disassemble slowly and evenly in the body. It has a half-life of 25 hours, compared to 12 hours with insulin glargine (Lantus, Sanofi). It has a flat acting profile and is taken once per day. Because it’s so flat, it takes three days to get to a steady state and has a lower risk of hypoglycemia. It is also much less sensitive to timing of the injection.
  • Phase 3 studies of insulin detemir assessed the effects of insulin detemir (versus insulin glargine) for more than 8,000 patients with type 1 and type 2 diabetes. They were treat-to-target trials with careful and consistent measurement of hypoglycemia.
  • The various degludec studies all showed that A1c was non-inferior, and in general, severe hypoglycemia was lower, particularly at night. Severe hypoglycemia was 17% lower for type 2 diabetes, (yet 10% higher for type 1 diabetes), and at night 32% lower for type 2 diabetes and 17% lower for type 1 diabetes. In trials of flexible dosing time, A1c was equivalent to insulin degludec and insulin glargine dosed at fixed times. Patients reported that the key benefit of insulin degludec is that patients know what to expect when they wake up in the morning. Degludec comes in a U100 or U200 pen and Dr. Bode also noted that patients used 12% less insulin than glargine.
  • The FDA sent Novo Nordisk a complete response letter after reviewing their submission for insulin degludec, although the drug was approved in the EU and elsewhere. At issue was cardiovascular safety, although many commentators found the FDA’s decision surprising. Novo Nordisk is now required to perform cardiovascular outcomes trials (DEVOTE) before approval in the USA.
  • IDegLira is a fixed ratio combination of insulin degludec and liraglutide. In studies, IDegLira demonstrated the strengths of both insulin and GLP-1 agonists (glycemic control, no weight gain) and lowered their risks (lower hypoglycemia, nausea).

 

-- by Adam Brown, Hannah Deming, Hannah Martin, Manu Venkat, and John and Kelly Close