8th International Conference on Advanced Technologies & Treatments for Diabetes (ATTD 2015)

February 18-21, 2015; Paris, France; Day #1 Highlights - Draft

Executive Highlights

Bonjour from Paris, France, where Day #1 of ATTD 2015 has just wrapped up. While we are certainly enjoying the European scenery, we also don’t feel too far away from home with a Chipotle and Starbucks right next door to the conference center. The meeting officially began with a gorgeous opera performance at the opening ceremony, where we also learned that ATTD 2015 brought in a whopping 2,500 attendees, up from 2,080 in 2014 – one of the largest jumps in attendance the conference has seen in some time. We still remember back in the day, when ATTD was held in tiny venues in Prague and Basel (2008 and 2009), when the venues were not fancy conference centers as big as Paris’ CNIT but charming hotels! This year, ATTD has attracted attendees from 87 countries, with the US as the largest delegation, and the number of European attendees growing quickly (still, US industry probably contributed to a sizable part of the growth). Today’s first half-day featured corporate symposia from Medtronic, Abbott, and Roche, along with an evening keynote on insulin pumps in type 2 diabetes. Please see below for our top highlights from Day #1 and see our preview for what’s in store the rest of the week.  

1. Abbott’s jam-packed standing-room-only corporate symposium was headlined by new hypoglycemia and YSI accuracy data on FreeStyle Libre – the session was so packed with many hundreds of people that organizers had to leave the auditorium doors open to handle the overflow. Boy, was there a lot of enthusiasm in Q&A about how this is being used in the real world, and how much demand there is. Said Dr. Iain Cranston, “Half my patients have not done a fingerstick in two months. They come to rely on Libre.” There is still capacity constraint on the manufacturing front, which is probably only adding to the enthusiasm – all current customers are being served, but are only being sent two sensors (one month’s worth) at one time.

2. Dr. Pratik Choudhary (King’s College London, UK) led a standing-room only workshop (n=50 or so) on Medtronic’s new MiniMed 640G/Enlite Enhanced system with SmartGuard (predictive low glucose suspend), including a hands-on demo and new data from a 40-patient user evaluation study. All were highly positive and signaled lots of market potential for this new foray into automated insulin delivery. Everyone there got to try the pump – great to see.

3. Headlining the opening session, Dr. Yves Reznik (University Hospital of Caen, Caen, France) summarized the predictors and non-predictors of success in the OpT2mise study of pumps in type 2 diabetes; 12-month data will come at ADA. We were a bit surprised that the opening keynote to ATTD was so associated with work from the top sponsor, though Dr. Reznik did mention other approaches to type 2 and pumps.

4. Dr. Jessica Castle (Oregon Health and Science University, Portland, OR) presented a clinicians’ wish list for key improvements to CGM technology, including better real world testing of connectivity/telemetry and fewer “egregious” sensor errors. She’s a star and we found this one most interesting.

5. At a Medtronic workshop focused on the OpT2mise trial, Dr. Ohad Cohen (Sheba Medical Center, Tel Hashomer, Israel) stressed that, contrary to many clinicians’ instincts, pump therapy for type 2 diabetes does not require the complex dosing needed in type 1. At the end of the session, a Medtronic rep recruited attendees for market research on a dedicated pump for type 2 diabetes, confirming that Medtronic is working on something specific (see more on Medtronic’s new type 2 business unit in our report on the company’s F3Q15 performance).

Honorable mention: Dr. Thomas Danne discussed reimbursement and regulatory challenges for CGM, including a mention of Nightscout!

Top Five Highlights

1. Abbott’s standing-room-only corporate symposium was headlined by new hypoglycemia and YSI accuracy data from the 72-patient, 14-day CE Mark trial of FreeStyle Libre – the session was so packed that organizers had to leave the auditorium doors open to handle the overflow. Dr. Tim Bailey (UCSD School of Medicine, San Diego, CA) opened the proceeding with the new pivotal study data from the factory calibrated Libre, which demonstrated an impressive overall MARD of 11.8% vs. YSI (the study had 1,238 paired sensor-YSI points). We first saw topline results from this study at EASD 2014, where Libre demonstrated a similar overall MARD of 11.4% vs. FreeStyle Precision fingersticks – but we didn’t see much hypoglycemia data there. New subgroup analyses showed that the sensor remains accurate in hypoglycemia – MAD was 9.5 mg/dl vs. YSI for glucose values 51-80 mg/dl. As a reminder, the product label recommends a confirmatory fingerstick when hypoglycemic, though commentary during Q&A suggested that patients are routinely skipping these steps, a confirmation of the technology’s real-world accuracy. One quote from Dr. Iain Cranston really put the accuracy in perspective: “Half my patients have not done a fingerstick in two months. They come to rely on Libre” and “This should pass as a glucose meter. It meets all the criteria.” From his lips to FDA’s ears (we doubt this will be easy). Similarly, the sensor maintained its accuracy even when glucose values were rapidly changing – see more on this below. Ultimately, the data on Libre continues to be very strong, and combined with the form factor and initial reception, makes it a product we feel could be transformational for clinicians as much as patients; therapy changes can be much easier with Libre – as can being a successful doctor. Our detail write-up below shares more on the hypoglycemia and rate-of-change data, the behavioral analyses, and interesting Q&A with lots of audience and clinician enthusiasm.

  • Dr. Jean-Pierre Riveline (French Institute of Health and Medical Research, Paris, France) explained that the FreeStyle Libre is not currently available for purchase in France, because demand from type 1s has so far exceeded manufacturing capabilities. This was consistent with Abbott’s 4Q14 financial update, where we heard that early sales have come “quicker, faster, and sooner than expected,” and the company was working to expand capacity. We wonder if it will take some time to get the factory calibration all set up as capacity expands. So far, current customers are “being taken care of” though it’s impossible to order more than one month of sensors at once.
  • Dr. Iain Cranston approximated that the cost of FreeStyle Libre is equivalent to that of ~10 fingersticks per day. This is the first such specific estimate we can recall hearing – it approximates to ~$0.40 per strip. As it stands, we believe the Libre system is priced at ~120 euros per month (59 euros for each 14-day sensor, $4 per day) – while this is less expensive than traditional CGM, we don’t believe it’s really being aimed at CGM users but more at type 2 patients who are not testing properly using traditional BGM. Pricing still puts the technology out of the reach of many patients, as there is little reimbursement to date. Abbott continues to work to toward reimbursement (the studies are still ongoing in type 1 and type 2), with expected primary completion dates in May 2015.

2. Dr. Pratik Choudhary (King’s College London, UK) led a standing-room only (N=50 or so) workshop on the new MiniMed 640G/Enlite Enhanced system with SmartGuard (predictive low glucose suspend), including a hands-on demo and new data from a 40-patient user evaluation study. Most notable was the glycemic data from the user evaluation study, in which SmartGuard performed very well – over 2,322 suspend-before-low events (in just four weeks!), the mean lowest sensor glucose value was 3.9 mmol/l (70 mg/dl), and only 3% of suspend-before-low events (n=74) went to the low limit (3.2 mmol/l, 58 mg/dl). This was very impressive hypoglycemia prevention in our view. The pump did not appear to increase mean glucose or glycemic variability. Notably, the Enlite Enhanced sensor (Enlite 2) had a MARD of 9.8% in this user evaluation study, suggesting significantly improved accuracy over the original Enlite (vs. fingersticks). Some of the patient quotes were quite illuminating: “Fantastic piece of technology – life changing”; “Felt confident exercising [for the] first time in a long time”; “Remarkably user friendly”; and “...worked well at preventing low. Silent mode was great – it does its thing in the background.” We also highly enjoyed a demo of the pump in the workshop and exhibit hall, and its design, menus, and highly customizable SmartGuard setup were definite highlights. Below, we enclose many more details from both the user evaluation and our observations from handling the pump.

  • Notably, in addition to the launch in Australia, the MiniMed 640G is now available in the UK and Denmark. An exhibit hall rep spoke very positively about the launch in Australia, noting that patients are “delighted” with a pump that finally looks cool, as opposed to just another device.
  • As a sidenote, ATTD’s organization at this meeting needs some logistical help – people had a lot of trouble registering ahead of time for this meeting and about 25 people were turned away. For a meeting with this price tag for doctors and nurses, organization should be easier. Logistics were also challenging at the opening night of the Exhibit Hall, where we heard from doctors it was pretty time consuming to get a glass a glass of sparkling water or something stronger!

3. In the opening keynote, OpT2mise lead investigator Dr. Yves Reznik (University Hospital of Caen, Caen, France) covered key results from the Medtronic-sponsored trial and summarized the predictors of success with pumps. This fits in with a broader theme of more focus on pumps in type 2 diabetes, and we were glad that Dr. Reznik touched upon other randomized and observational studies as well given that this was the day’s one non-sponsored session. For background, in the main analysis of OpT2mise (published last summer in the Lancet and shown as a late-breaking poster at ADA 2014), A1c declined by 1.1% in those on an insulin pump compared to 0.4% in the MDI group (p<0.001) after 27 weeks in the context of a roughly 20% lower daily insulin dose and less hypoglycemia. The most important predictors of success with pumps were higher baseline A1c, more advanced education, and frequency of daily boluses – some of these secondary analyses were presented as a poster at EASD. Factors that did not impact success with pumps included diabetes duration, older age, the degree of overweight at baseline or weight gain during the trial, compliance (as determined by SMBG frequency), and mild cognitive dysfunction – it was encouraging to see that these factors were not deal-breakers for pump usage, which is typically positioned for younger, tech-savvy patients. Notably, Dr. Reznik confirmed that 12-month OpT2mise results should be presented at this year’s ADA. He also called out the recent ADA/EASD 2015 Position Statement Update for only mentioning insulin pumps once, in the context of LADA. He suggested that pumps could be included in the insulin intensification diagram as an option if patients do not reach goal on MDI. A point well taken, though the cost must come down for sure.

4. As part of Roche’s session on “The Link Between Performance and Clinical Value of CGM,” Dr. Jessica Castle (OHSU, Portland, OR) provided a most valuable clinical perspective on the key areas of improvement for CGM, especially in preparation for the closed loop. Her list, which was refreshingly patient-oriented, included: (i) optimizing first-day sensor performance, which generally suffers with current designs due to the wound response; (ii) assessing telemetry in the real world to ensure that cell phones and other possibly interfering wireless signals do not lead to dropped sensor data; (iii) identifying and eliminating “egregious” sensor errors while reducing batch-to-batch sensor variability; and (iv) eliminating the acetaminophen effect and other interference issues that could lead to falsely high readings and insulin over-dosing. The list served as an excellent reminder that despite the positives we hear about CGM, the massive improvements over the past few years, and the work on automated insulin delivery, there is still important progress to be made. We salute Dexcom and Medtronic for working and improving on all these fronts – we see all of Dr. Castle’s wishes as addressable.

5. At a Medtronic workshop focused specifically on the OpT2mise trial, Dr. Ohad Cohen (Sheba Medical Center, Tel Hashomer, Israel) stressed that pump therapy for type 2 diabetes does not require complex dosing calculations. He acknowledged that this is counterintuitive for many clinicians who are used to treating patients with type 1 diabetes, admitting that even he couldn’t resist the urge to make small dosing adjustments during the trial. However, for the patient population studied in OpT2mise (i.e., poorly controlled, obese, insulin-resistant, long duration of diabetes), he recommended simply using a single basal rate and fixed bolus doses, arguing that small adjustments are unlikely to be meaningful in the context of large insulin doses (“four units per hour!”) and limited glycemic variability (i.e., since there is usually some residual beta cell function). Overall, Dr. Cohen characterized pump therapy as one of the most appealing options for patients not at goal at the “end” of the type 2 diabetes treatment algorithm. We agree that the potential for greater A1c improvements (1.1% with pumps vs. 0.4% with MDI in OpT2mise) with less insulin and no increased risk of hypoglycemia or weight gain compared to MDI is an exciting prospect. On the other hand, we have also heard speakers including Dr. Timothy Garvey (University of Alabama, Birmingham, AL) suggest that addressing obesity should be a greater priority with these patients and that adding more diabetes medications (particularly those that lead to weight gain) is much less likely to be effective. Food for thought... From that perspective, doubling down on insulin therapy – especially Dr. Cohen’s recommendation to increase the insulin dose if patients are still not reaching target – would not be ideal, though he did also encourage providers to reevaluate patients’ oral medication regimens, eliminate those likely to cause weight gain, and try a GLP-1 agonist. Certainly, there is no single right way to treat type 2 diabetes, and we’re glad to see pump therapy (hopefully) becoming more accepted as another option in the toolbox, particularly globally.

  • At the end of the session, a Medtronic rep asked attendees to take part in European market research for a “dedicated pump” for type 2 diabetes. While the Medtronic/Sanofi type 2 diabetes partnership has been dissolved as of this week (see our Medtronic earnings report from yesterday), the company clearly remains committed to the area, as evidenced by the recent hiring of the highly respected Dr. Bob Vigersky as top medical lead for the type 2 program. We will be watching closely for more details on the proposed device and hope to see what emerges that is clinically effective, designed to fit the needs of type 2 patients, cost-effective, and simple enough to compete with once-weekly GLP-1 agonists.

Honorable Mention

In a Roche symposium, Dr. Thomas Danne (Diabetes Center for Children and Adolescents, Hannover, Germany) discussed reimbursement and regulatory challenges for CGM; more upliftingly, he also shared some ways in which patients are taking a stand. The “18-month half life” it takes to get new technology into patients’ hands in Europe (say nothing of the US), in his view, has been frustrating for providers and patients, not to mention industry. We were pleased to hear Dr. Danne specifically highlight Nightscout as an example of a grassroots patient-led reaction to the extensive regulatory red tape. Of course, the FDA’s recent approval of the Dexcom Share receiver and loosening of the review for apps and devices tied to CGM should help the entire field on this front. Turning to his home turf, he noted with concern that the notorious German authorities will soon rule on reimbursement for CGM (2015 or 2016, per what we heard last week at EASD Diabetes Technology). For background the IQWiG and G-BA authorities have already wreaked havoc in the type 2 diabetes drug landscape. Here too, however, patients have asserted themselves – Dr. Danne displayed a picture of a patient protest to preserve access to rapid-acting insulin analogs for children, and we imagine a similar wave of advocacy could improve the outcome of the government’s upcoming decision on CGM. Despite these bright spots, Dr. Danne acknowledged that the evidence base for CGM’s cost effectiveness is not yet as strong as it needs to be, partially because existing meta-analyses that do not examine CGM in sufficiently targeted patient populations. Ultimately, he suggested that patients should take a greater role in making decisions on reimbursement, and wondered aloud whether it should be patients or scientists (and in what balance) who lead decisions on reimbursement policy.   

Detailed Discussion and Commentary

Industry Workshop (Sponsored by Abbott)

FreeStyle Libre System Accuracy Study

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

Dr. Tim Bailey shared never-before-seen data from Abbott’s 72-patient, 14-day pivotal CE Mark trial of its factory calibrated FreeStyle Libre system, which demonstrated a solid overall MARD of 11.8% vs. YSI (the study had 1,238 paired sensor-YSI points). We first saw top line results from this study at EASD 2014, where Libre demonstrated an overall MARD of 11.4% vs. FreeStyle Precision BGM (13,195 paired points). One criticism of the data shared at EASD was the use of BGM as the reference device, as most CGM studies have in-clinic days with YSI – it was good to see that the 14-day, factory calibration holds up against that standard as well. New subgroup analyses also showed that the sensor remains accurate near the hypoglycemic range, where MAD was 9.5 mg/dl vs. YSI for glucose values in the 51-80 mg/dl range – this is the first time Abbott has shared this data (see below). As a reminder, the product label recommends a confirmatory fingerstick when hypoglycemic, though commentary during Q&A suggested that patients are routinely skipping these steps, testament to its real-world accuracy. Similarly, the sensor maintained its accuracy even when glucose values were rapidly changing as Dr. Bailey shared a comparison of data excluding vs. including all points < 80 mg/dl or changing at > 2 mg/dl. Reported MARD were not statistically different for the two subgroups: 10.4% (exclusive) vs. 11.4% (inclusive).

  • As a reminder, the pivotal study was conducted in four centers across the US 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. Bailey shared that 24% of patients had a baseline A1c > 8.5% and that 22% of patients had a baseline A1c < 7.0% - a nice mix of well-controlled and out-of-control patients.
    • We saw topline results from this study at EASD 2014 where Abbott’s factory calibrated FreeStyle Libre system 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). MARD was lowest on day one (15.7%), improved to 11.9% on day two, and hovered between 10.3% and 11.8% on days 3-14. The study had 13,195 paired FreeStyle Libre-BGM data points (range: 23-498 mg/dl).
  • FreeStyle Libre demonstrated excellent accuracy in the hypoglycemic range vs. both YSI and fingersticks. MAD was just 9.5 mg/dl vs. YSI (n=53) and just 10.0 mg/dl vs. fingersticks (n=901). A bigger testament to Libre’s hypoglycemic accuracy came from Dr. Bailey’s colleague, Dr. Iain Cranston who noted, “Half my patients have not done a fingerstick in two months. They come to rely on Libre.” The data below is cut at 50 mg/dl and 80 mg/dl, so it is tough to know what the accuracy is for <70 mg/dl vs. > 70 mg/dl (i.e., did most of the 50-80 mg/dl points fall in the 70-80 mg/dl bucket?). Even still, based on our experience wearing the device, it is very accurate in hypoglycemia, so the point is more semantics than anything.

Table 1: FreeStyle Libre vs. YSI in Hypoglycemia

Glucose Level (mg/dl)

MAD (mg/dl)


< 50






Table 2: FreeStyle Libre vs. BG Capillary for Low Blood Glucose

Glucose Level (mg/dl)

MAD (mg/dl)


< 50






  • Libre demonstrated impressive results even when glucose values were rapidly changing. Dr. Bailey shared subgroup analysis for results excluding vs. including all points < 80 mg/dl or changing at > 2 mg/dl/min. Impressively, the reported MARD were not statistically different: 10.4% (exclusive) vs. 11.4% (inclusive). We thought this was a clever and smart way to summarize the data.

Table 3: Sensor Results During Times of Rapidly Changing/Low Blood Glucose

Glucose level

Measurement vs. capillary BG

All Data

Excluding results < 80 mg/dl or changing at >2 mg/dl/min





< 100 mg/dl


11.3 mg/dl


10.3 mg/dl


> 100 mg/dl












  • Mr. Bailey 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. Still, the data pointed to why 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.
    • 88% agreed that applying the sensor was less painful than a routine fingerstick (more to the point, we’d add that it’s only required every 14 days).
    • 86% agreed that the sensor did not get in the way of daily activities.
    • 96% agreed that Libre is comfortable to wear.
    • 96% agreed that scanning the sensor is less painful than pricking my finger (We’re not quite sure how these 4% of patients were scanning or what magical lancing device they were using – Adam and Kelly have both used it and say the scanning is completely painless.).
  • Very few adverse events were reported among patients in the study. Only 26 subjects reported any sort of discomfort around the sensor insertion site and all reports were consistent with what would be expected following insertion of a sensor into the skin:
    • Moderate to severe erythema – 4% of the time
    • Other moderate symptoms – 1% of the time
    • Mild symptoms – less than 9% of the time (Moderate vs. mild symptoms were not defined.)
  • Although Libre is only approved for upper arm wear, commentary during Q&A suggested that most patients are taking liberties with sensor placement – the re-location does not appear to be affecting accuracy. Said Dr. Iain Cranston (Portsmouth Hospitals NHS Trust, Portsmouth, England) during Q&A: “Most people using Libre have purchased the devices themselves, so will put it where they damn well like. The upper quadrant of the buttocks is a popular place, and it stays on for two weeks. On the abdomen is another popular place, but more likely to fall off. It’s incredibly anecdotal reporting thus far though.” The upper arm can be a tough sell for patients (less body fat for some; more visible), so we are not surprised to see patients taking initiative, as they have with traditional CGM as well. The company has said it is exploring approval for some of these other sites for wear, though that would simply allow for different marketing and training more than greater uptake.

Questions and Answers

Q: You mentioned Libre’s great precision – why would you want to cross check with fingersticks?

Dr. Bailey: It turns out that the precision of this can replace fingersticks. There are still some caveats with rapid changes or with hypo but a replacement is here and that’s pretty exciting for patients.

Dr. Cranston: Half of my patients have not done a fingerstick in two months. Patients have come to rely on Libre.

Q: D you have to wear the sensor on your arm? Could you wear it anywhere else? What would be the differences?

Dr. Bailey: At my site, you had to wear it on your arm. In reality, you can’t even imagine all the places patients actually put the devices, but I don’t know the data.

Dr. Cranston: Most people using Libre have purchased the devices themselves, so will put where they damn will like. The upper quadrant of the buttocks is a popular place, and it stays on for two weeks. On the abdomen is another popular place, but more likely to fall off. It’s incredibly anecdotal reporting thus far though.

Q: Is there any data on interfering substances?

Dr. Cranston: I have not seen study data relating to it, though I believe it is there. It is not something I recommend people need to avoid. I think the clinical sheet that comes with it tested against 20 substances, and although they can measure differences in the value from that, they are not clinically different.

Q: Could you tell us more about the first hour of insertion after the new sensor?

Dr. Bailey: The first hour is the calibration. In our study, 100% of the sensors came online. The only sensors that did not perform had to do with problems with insertion. Once they were inserted, they all reported accurate data.

Dr. Riveline: In my experience, patients noted that accuracy was not as good in the first day.

Dr. Cranston: I tend to advise people to not act according to what they see on the sensor in the first 12 to 14 hours. In an anecdotal way, it does appear that the sensor will tend to read a little bit lower on day one. It’s totally anecdotal and someone from the company will probably shoot me for saying that but that’s what I’ve heard.

Dr. Bailey: This is the first presentation of the YSI data; there will be more data pretty soon.

Q: Any allergic reactions to adhesives? Are there any case reports on that? Maybe on social media?

Dr. Cranston: Yes, on social media there are absolutely reports of that. There were photos of that – like people can be allergic to any adhesive. What I’ve found absolutely stunning and what I’ve seen is that they leave them on! It shows how much they like Libre. Normally when you have something that itches, you take it off. Some patients have had quite a nasty reaction because they want to leave the sensor on.

Q: What about the sensor in a hairy area? Or the abdomen? Do people need to be careful about knocking it off?

Dr. Cranston: It’s only a five-millimeter sensor. So if it’s really hairy area, then you might have problems with getting the sensor inserted. On the back of the arm, I have never shaved an arm to put it on. As for knocking a sensor off, of course you can. People tend to put it on the front part of the arm and it gets knocked off. But on the back, it is better.

Q: Can you talk about Libre vs. BGM?

Dr. Cranston: This should pass as a glucose meter. It meets all the criteria.

Q: Would you recommend patients to do more or less fingerstick measurements in the first 24 hours after putting the sensor on?

Dr. Riveline: I would recommend the same number of SMBGs as usual due to the lower accuracy.

Dr. Cranston: I would say it depends on why they have it. But what I tell patients is not to react to what’s happening actually. I tell them to try not to react too much in the first week. Once you have an idea of what your profiles look like, then in the second week, you can react.

Q: What happens to the needle used to insert the sensor?

Dr. Clifford Bailey (Moderator): It’s in and out before you can see it.

Q: Do insurance companies pay in the US, UK, and France?

Dr. Bailey: It’s not clear in the US. Reimbursement is critical for patients to have access. It would be wonderful if this were covered. It’s not covered in the US or France.

Dr. Cranston: In the UK, it has not been looked at by NICE. I’m not aware of any immediate plans to do that. There is the question of whether an individual can get it funded, but that’s a conversation between the patient and insurance companies and I’m not sure that anyone’s managed it.

Dr. Riveline: I think we have to fight to obtain reimbursement but it hasn’t happened yet.

Dr. Clifford Bailey: If you look at the costs incurred by fingersticks compared with this, then this is very competitively priced. I think that that’s going to be a very big factor in actually deciding the mass use of this new device.

Dr. Cranston: Yes, about ten tests a day is the same price.

Q: How do you think about Libre plus a pump?

Dr. Riveline: I think it’s a good thing for a lot of patients but there is no combination with a pump. It depends on the patient. There are patients who don’t want to have alarms. For some patients with a pump, that’s a good thing.

Dr. Cranston: It’s important to say what it’s not. The Libre plus a pump cannot be an artificial pancreas since data is not continuously collected but instead is episodically collected. That said, many Libre users don’t have pumps. They are treating with insulin injections as they normally would. Is it accurate? In my eyes, yes it is. Do they do it? Yes, they do. Some patients have used a wizard with this and nobody’s factored in an arrow-based wizard, but they visually adjust.

Q: In a situation of rapidly changing glucose, there is a lag time. How does this work out for Libre?

Dr. Bailey: The overall lag time is 4.5 minutes. We don’t have data on rapid change. This is when you want to confirm with fingersticks. Particularly when it’s trending down, check with fingersticks.

Dr. Cranston: When they see a vertical arrow, it is easy to flash again in five minutes. I say, don’t work on a single value. The first flash tells you that you need another again. It’s not that hard to get a second look.

Q: How is it possible to have this MARD without calibration?

Dr. Bailey: Everyone is working on this technology. This is the first released product. Every product out there has its advantages. That’s the one thing about this product.

Dr. Cranston: Let’s make it clear that it’s not that this isn’t calibrated. It is. It’s just not calibrated against fingersticks. It’s calibrated in the lot.

Dr. Bailey: It also has to do with drift. Other sensors would drift and you would have to recalibrate. This one doesn’t drift. Standard meters aren’t that accurate too. So factory calibration is great because it gets rid of using inaccurate meters in calibration.

Q: Can I walk across the street to the nearest pharmacy and buy one?

Dr. Riveline: In France, it was possible to have the device directly bought from the company. But it’s no longer available because a lot of type 1 patients asked for the device. So we have to wait for more devices to be built. The cost is 120 euros for the device and 60 euros for each sensor.

Q: Could you comment on the clinical relevance of using Libre?

Dr. Cranston: I don’t think that there is anybody who has come to me with a profile or that I have put a sensor on where we haven’t gotten clinical value from the profile. There is always something we can clinically get out of the profile.

Q: Is there enough data yet to make diabetes diagnoses in previously undiagnosed patients?

Dr. Cranston: I think it’s a really good question. We’re trying to collect some data about that. Traditionally, diagnosis is based on complications. So we’d have to do it indirectly. I have to say, when you see a normal glucose profile on one of these devices, there’s nothing else that looks like a normal glucose profile. Low variability and a completely flat profile is what normal looks like. We’re looking at Libre in post-transplant profiles to see how it changes.

Industry Workshop (Sponsored by Medtronic)

Advanced Hypoglycemia Protection in Action

Pratik Choudhary (King’s College London, UK)

Dr. Pratik Choudhary led a standing-room only workshop (n=50ish) on the new MiniMed 640G/Enlite Enhanced system with SmartGuard (predictive low glucose suspend), including a hands-on demo and new data from a 40-patient user evaluation study. Most notable was the glycemic data from the user evaluation study, in which SmartGuard performed beautifully – over 2,322 suspend-before-low events, the mean lowest sensor glucose value was 3.9 mmol/l (70 mg/dl), and only 3% of suspend-before-low events (n=74) went to the low limit (3.2 mmol/l, 58 mg/dl). This was very impressive hypoglycemia prevention in our view. Notably, the Enlite Enhanced sensor had a mean MARD of 9.8% in this user evaluation study, suggesting significantly improved accuracy over the original Enlite (vs. fingersticks). Dr. Choudhary also shared a number of quotes from patients in the study – it was terrific to hear the positive impact this system has had on participants. We also enjoyed a demo of the pump in the workshop and exhibit hall – we were impressed with the new pump design, menus, and highly customizable SmartGuard setup. More details below!

MiniMed 640G User Evaluation Data

  • Dr. Choudhary gave a sneak peek at results from the 40-patient EU user evaluation study of the MiniMed 640G. Children and adult patients in Copenhagen, London, and Barcelona wore the system for four weeks and evaluated its new features and training.
  • Over four weeks, participants (n=38 on the slide) had an average of 2.3 suspend before low events per day – this translated to 2,322 events, a surprisingly high number and a reminder of how many moderate lows patients with type 1 diabetes face every day. Mean duration of suspends was only ~57 minutes.
  • SmartGuard performed very well – the mean lowest sensor glucose value was 3.9 mmol/l (70 mg/dl), and only 3% of suspend-before-low events (n=74) went to the low limit. Suspend thresholds were set at 3.2 mmol/l (58 mg/dl) – this was impressive performance in preventing lows in our view. Time spent <70 mg/dl was just 3%, and patients spent 75% of the time in the target of 70-180 mg/dl. The study was not powered for glucose outcomes, and there was no control group, so it’s hard to draw significant conclusions from the data. However, the results are certainly encouraging.
  • Compared to baseline, both mean glucose and standard deviation were slightly improved on the MiniMed 640G in adults and children (~9 mg/dl or ~0.5 mmol/l). The study was not powered for glucose outcomes, and the slide did not contain p-values. The data were directionally interesting, though did not enable any conclusions about the system, other than it is not dramatically changing mean glucose or glycemic variability. We assume Medtronic will publish more robust results.
  • Notably, the Enlite Enhanced sensor had a mean MARD of 9.8% in this user evaluation study, suggesting significantly improved accuracy over the original Enlite (MARD of 13-14%). Dr. Choudhary said this in passing and did not discuss whether it was vs. YSI or fingersticks (we later learned that it was, indeed, vs. fingersticks), or how many calibrations per day were needed. He did emphasize that the improved accuracy in part stems from a new algorithm that now sits on the transmitter (as opposed to the pump in prior Medtronic devices).
  • Notably, the Veo’s issue of repeated alarms at night has now been rectified – as noted below, patients can choose whether they want any alarms with SmartGuard, and the specific periods of time where they do or don’t want them.
  • Dr. Choudhary emphasized that patients took some time to adjust to the system. Given years of treating lows on their own, it took many patients 2-3 weeks to feel comfortable that SmartGuard could mitigate hypoglycemia.

Patients Views on the MiniMed 640G
[grammatical errors were included on the slides]

“In day took some carbs and manually restarted.

“Stopped me going low so I didn’t have to deal with the after effects of hypo.”

“Very positive – worked well at preventing low. Silent mode was great – it does its thing in the background.”

“Loved all the info on the home screen. If I treated the hypoglycemia ended up high – should have left it alone.”

“Fantastic piece of technology – life changing. I don’t find the low suspend as useful – I’m already low when it suspends; suspend before low is more...” [rest of quote was cutoff]

“Felt confident exercising first time in a long time – I could exercise without drinking two bottles of Lucozade! Amazing to avoid the horrible scary feeling completely – took me a while to have confidence in it but by my second sensor I was there.

“In day usually took carbs and restarted basal; at night didn’t wake and pump sorted low out. Surprised how it alerted me in the day before I got low. At first not sure how much glucose to take – but found needed much less hypoglycemia treatment now.”

“Remarkably user friendly. Avoided over treatment hypos. Gave me confidence – user override important if I doubted it. Training important – changing hardwired behaviors is tough.”

  • Dr. Choudhary’s last slide succinctly summed up the user evaluation experience: “Patients loved the 640G. Clear, effective, unobtrusive. It makes them feel safe. Would you drive a car without ABS?” We look forward to hearing more on-the-ground patient experiences, since corporate symposia can’t offer the full picture of a device’s pros and cons.

Hands-On Workshop – Setting up SmartGuard

  • In a short hands-on workshop, attendees had the opportunity to setup SmartGuard, which is found in the “sensor options” – “low alerts” menu. Medtronic has done an excellent job of making the setup as easy as possible, and we got the hang of it with only a few glances at the two-page quick-start guide. SmartGuard requires setting up four different parameters: (i) time segment; (ii) low limit; (iii) suspend by sensor options; and (iv) alert options. The approach is somewhat reminiscent of the Tandem t:slim’s insulin profiles (i.e., where a slew of options is bucketed under one heading – in the case of the 640G, the low limit, suspend, and alert options are bucketed under the time segment).

I. Time Segment

II. Low Limit

III. Suspend By Sensor Options

IV. Alert Options

Patients can select up to eight time segments to define a different profile of low settings.

This is a major win, as it would allow patients to set different day vs. night profiles (for example), when hypoglycemia risks and alarm preferences are much different. As a patient said above, “It does its thing in the background.”

The low glucose limit for the time segment. This limit is the lowest sensor glucose value that a patient would like to avoid reaching. All suspend and alert options are based on the low limit identified for the time segment. It can be set from 2.8-5 mmol/l (50-90 mg/dl).

We like the universality of a single low limit, as well as the very understandable language around what this limit means (“value that you would like to avoid reaching”).

1. Suspend before low (i.e., predictive low glucose suspend)

2. Suspend on low (i.e., threshold suspend)

3. OFF

If 1 or 2 is selected, insulin delivery will automatically resume if sensor glucose is above the low limit and trending upwards or if the maximum two-hour suspend time has been reached.

For each time segment, patients can choose whether they want to be alerted:
1. Before low

2. On low

3. Resume basal alert

Notably, patients can turn all these alarms off during a time segment (e.g., at night), meaning the pump will suspend and resume insulin in the background.

  • “Everything in this pump is pretty much customizable,” said the rep in our visit to the Medtronic booth. The ability to shut off all alarms and let the pump do the work in the background is one of the major promises of automated insulin delivery – taking the burden off the patient and putting it in the hands of technology. While the MiniMed 530G badgered patients with loads of alarms and messages, it’s great to see that these can be turned off at night with the 640G.

Hands-On Workshop – Observations on Pump Design

The workshop also allowed us to browse the 640G’s menus and hold it for the first time.

  • The 640G’s color, light-adjusting screen lived up to the hype. It’s not quite an iPhone-quality screen, but it’s a massive improvement over the grayscale screens of previous Medtronic pumps. The black background with white writing is very sharp and high contrast – this change should be especially significant at night, where there is now no need to turn on a backlight.
    • The screen is also much larger than on previous pumps, an advantage most readily apparent in the bolus wizard. Four lines of text fit on the screen (blood glucose correction, IOB, carbs, and the bolus amount), in addition to the top bar containing status icons. Instead of scrolling, an ok button takes the pump into the next screen, where a normal, dual, or square wave bolus can be selected. The screen seems roughly on par in surface areas to the OmniPod PDM’s screen.
    • The color screen has also allowed Medtronic to do more with color-coding and indicators. For instance, toggling certain options changes them from gray (OFF) to lit up with a green indicator (ON). In another case, the reservoir icon is lit up green, orange, or red to indicate how much insulin remains. These are small but meaningful changes, and perhaps taken for granted in the smartphone era.
  • The 640G pump body is slightly curved and feels a bit larger in volume than the Paradigm line of pumps, but not overtly so. In some ways, it looks smaller because it is oriented vertically instead of horizontally. The pump can be used with one hand, unlike previous Medtronic pumps that had to be held horizontally and used with two thumbs (i.e., it’s now like holding a smartphone instead of a video game controller).
  • The pump is much easier to navigate through using a four-way arrow key, a back button, and a select button – as a reminder, older Medtronic pumps had up/down arrows, a back button (ESC), a select button, and a bolus wizard button. The 640G also has a menu button to return to the main menu at any time; we rarely needed to use this in navigating through the pump, but it is helpful. The interface is not quite as easy as the touchscreen on Tandem’s t:slim, but it is a marked improvement over previous Medtronic pumps.
    • Home screen: It’s fantastic to see insulin-on-board on the home screen, side-by-side with the CGM data. We loved that feature of the Animas Vibe (see the diaTribe test drive) and are glad to see it now incorporated in the 640G. The home screen also has many status icons like battery, reservoir, alarms, etc.
    • From the home screen, two menus can be immediately accessed: bolus and basal. The bolus menu allows viewing active insulin (we’re not sure what this screen shows differently from the home screen), the bolus wizard, a manual bolus, and a preset bolus. The basal menu allows access to a temp basal, preset temp basal, and basal patterns. It was great to see these very common pump functions quickly accessible.
    • The pump’s main menu (i.e., pressing the menu button) has nine screens: suspend delivery (i.e., manually suspend the pump), audio options, history, reservoir and tubing, insulin settings, sensor settings, event markers, reminders, and utilities.
  • Like the Tandem t:slim and Insulet OmniPod, Medtronic also has more confirmation screens in the 640G. Messages such as “Changes not saved – exit without saving changes?” and bolus confirmation popups are now built into the pump. This is the way all pumps are going for human factors sake, and though it does add some patient hassle, the safety benefits may be worth it for a drug as dangerous as insulin.
  • The Enlite Enhanced (Enlite 2) is reportedly more accurate, though Medtronic has not shared any new clinical data. As a result, the MiniMed 640G manual still cites the same MARD as the Enlite. However, Enlite 2 is 80% smaller, and uses a new algorithm built right into the transmitter. As noted above, Dr. Pratik Choudhary said the MARD was 9.8% in the user evaluations (we assume vs. fingersticks, though he did not specify). The transmitter also uses a new frequency and can store up to 10 hours of data and backfill the information into the pump.
  • Other new features we learned about: patients can stop bolus delivery while a bolus is in progress; louder and more customizable alarms than on previous Medtronic pumps; the ability to make preset boluses (e.g., “Dinner – 2.5 units”); and the ability to copy basal patterns.
  • See our initial report on the Australian launch for more discussion of the pump’s features.

Questions and Answers

Q: How do you judge how to treat hypoglycemia?

A: There are a couple factors that go into it. If the pump suspends on its own, and there is no active insulin, you don’t need to treat it. That’s the thing people picked up. If it suspended and there was active insulin on board, we didn’t quite get enough hypos to get a clear algorithm. What I came to was this: if active insulin is about the basal rate, you don’t need to do anything. If it was over a unit, you probably need that level of carb based on your insulin:carb ratio. The key thing is that active insulin is on the home screen.

Q: How often did patients have problems with calibration of the sensor?

A: On the whole, we haven’t looked at it specifically. MARD was 9% - the accuracy was much better. It was out of range much less often.

Q: It’s important to note that you’re also avoiding the low levels, where the sensor is less accurate.

A: Yes. The sensor is cutting out insulin at a level of 80-120, and it is more accurate in that range.

Q: So during the night, you don’t use the alarm and don’t correct the low by eating. During the day, you look at active insulin. But I’m inclined to say don’t look at anything.

A: At night, you’re less likely to have active insulin. In the day, it comes from the tail end of boluses, so active insulin plays a slightly bigger part. It took people a bit of time to trust the system. Of course, if you just took six units, and you’re already at four units on board, stopping 0.6 units of basal isn’t going to make a difference.

Q: Can you think about two pump settings, one with a working sensor and one with a non-working sensor? With the working sensor, can you crank up insulin? And if the sensor isn’t working, it goes back to safe mode?

A: That’s a tricky one. You need to have a very switched on patient so that they know what’s happening there. You do run the risk if the sensor misses hypoglycemia, of driving low. But the question is, ‘Would this system tempt you to use a higher basal than normal?’ I tend to go for 50/50. In pediatrics some go for 70/30. I would struggle to say that it’s the right thing to do to run a higher basal.

Adam Brown (Close Concerns, San Francisco, CA): What was your biggest surprise as you did the user evaluations? And what is the one thing you would like to see improved as Medtronic looks to the next generation?

A: The biggest surprise – when people took the system off, we asked them two questions. First, we know that reimbursement for CGM is patchy. Most don’t have it. So the first question was, “Would you use this over the Veo if you didn’t have sensors?” In other words, is the pump on its own any better than the Veo? They all said yes. The menus are clearer. The alarms are more adaptable. There is a quick and easy bolus. That was a surprise. There was an extra button press to validate every bolus, and that annoyed some patients.

The second surprise was how much subliminal discomfort and fear of nocturnal hypoglycemia there is. People are scared of hypos. I was surprised by how much people came back with reassurance. The pump would switch off before patients were in hypoglycemia. Once they had a few nights, they would say, “It suspended twice last night! I didn’t even realize.” I was surprised by the impact of that factor on the patients. It’s the same thing we see with closed loop. Even when people are sleeping calmly, there is a subliminal level of anxiety at night.

Adam: And how about the one improvement?

A: We’d want to see the other half of that algorithm – the closed-loop piece. And then some people didn’t like the extra bolus confirmation screen.

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