Become An Expert
Log In

Top 10 of the ATTD - part 1

May 18, 2022

At the end of April, I attended the ATTD, the annual convention on diabetes technology. It was in Barcelona this year, and it was super fun as usual. A bit frustrating that I couldn't see everything, but luckily most sessions were recorded and you can also catch up afterwards.

In this article I discuss over 5 important updates in the field of diabetes technology from this congress:

  1. Mylife CamAPS FX partners with Libre 3
  2. Libre 3 is smaller and has better (overall) accuracy than Dexcom G7
  3. Medtronic wants to make carbohydrate counting obsolete thanks to a smartwatch
  4. Results of the iLet insulin-only closed-loop system
  5. Omnipod 5 works just as well or even better for type 2 diabetes

Next week I'll tell you 5 more updates from the ATTD, stay tuned!


1. Mylife CamAPS FX partners with Libre 3

On the first day of the ATTD, it was announced that Ypsomed, Cam APS FX and Abbott will develop a closed-loop system with the Libre 3 by the end of this year. Cam APS FX is the 4th closed-loop system that will partner with Libre. In the past, agreements with Abbott have also been signed by Tandem (Control IQ), Insulet (Omnipod 5) and Bigfoot Biomedical (Bigfoot Autonomy). It is, however, the first time that a time has been specified when the integration of a libre sensor in a closed-loop system will be completed. At Ypsomed's ATTD booth, they immediately said that its commercialization would not be before this year.

The partnership with Libre will likely make closed-loop systems more affordable and accessible. Libre sensors are cheaper than those from Dexcom, and the Libre 3 will -normally- be sold at the same price as Libre 2. So there is a good chance that this closed-loop system will be reimbursed for more people with diabetes.

Apart from that, the mylife - CamAPS FX - Libre 3 system has a number of unique features:

  • the Libre 3 is the smallest and most accurate (?cfr infra) sensor, and lasts longer than Dexcom (14 days)
  • the YpsoPump is a nice, small pump (smaller than Tandem and Dana i), which works with prefilled insulin "Pumpcarts", and can be controlled from your mobile phone
  • and the CamAPS algorithm has a target value down to 80 mg/dl, and is the only one approved during pregnancy, and from the age of 1 year.

The disadvantages are that the YpsoPump is still a catheter pump (not a patch pump), and that you always have to have your mobile phone nearby for the algorithm to work. There is no direct communication between the pump and the sensor.

2. Libre 3 is smaller and has better (overall) accuracy than Dexcom G7

There were demos of both Libre 3 and Dexcom G7 on the ATTD, so I put them side by side and took a picture. The Libre 3 sensor seems to be slightly smaller than Dexcom G7 🔎. Just like with Dexcom G7, the Libre 3 has a small piece of adhesive underneat though, which was not included with the demo version of the Libre 3.

  • Libre 3 is the successor of the Libre 2 sensor, and is a real-time CGM. So you no longer have to scan to see your blood sugar level, the blood glucose readings just come to your phone. There is not even a reader available anymore. Libre 3 has a CE label and is already available in Germany and the Netherlands. I hope to find out this year when Libre 3 comes to Belgium. For America, FDA approval was requested as iCGM last year.
  • Dexcom G7 is the successor of the Dexcom D6 and is a sensor and transmitter in 1. It lasts for 10 days (+12 hours) and has a warm-up time of only 30 minutes. In my previous blog post I discussed it extensively and compared it with the other new CGMs. It has been CE marked since March 2022, and the response from a first group of users in England was presented at ATTD and has been very positive. However, there was no news about when the G7 would be launched in which country. The US is also eagerly awaiting the approval of Dexcom G7 as iCGM, which is expected to be before this summer. Once approved, Tandem Control IQ and Omnipod 5 have already agreed to integrate the Dexcom G7 into their closed-loop system within months of approval. This is less clear for other closed-loop systems (eg Diabeloop and Cam APS FX). (PS: the Dexcom G8 was also mentioned here! 😲)

Back to the ATTD: Abbott showed a recent study with the Libre 3 in America, where the MARD (Mean Absolute Relevant Difference: a standard parameter for accuracy) was 7.8%! (Publication to follow.) This is the lowest MARD that could be demonstrated in a CGM system so far. Before that, Dexcom G7 had the honor of having the lowest MARD: 8.2% on the arm and 9.1% on the stomach. With some Dexcom warriors, the 7.8% MARD of Libre 3 was immediately nuanced. Although you really shouldn't compare studies in this way (because the study population is completely different), it seems that the Dexcom G7 does outperform the Libre 3 when you look at the first 7 days.

MARD Libre 3 (%)   MARD Dexcom G7 (%)
  arm abdomen
Day 1-3 8,6 Day 1 11,9 12,9
    Day 2 8,4 8,6
    Day 4 7,2 7,7
Day 7-8 8,7 Day 7 7,2 8,1
Day 9-12 6,4 Day 10 7,6 9,3
    Day 10,5 6,9 8,8
Day 13-14 7,0      
Overall 7,8 Overall 8,2 9,1


3. Medtronic wants to make carbohydrate counting obsolete with a smartwatch

Ali Daniaty, the VP of R&D and Operations at Medtronic, echoed what he said in his interview with Nerdabetic. Namely, that Medtronic is working on a fully closed-loop system with an optional meal announcement. You could do this by pressing a button on your mobile ("one touch bolus") or by wearing a smartwatch with Klue technology ("Klue bite bolus"). Klue recognizes the typical meal movements of the wrist and can thus recognize when you eat, and even how much you eat (for example, if you make a meal movement several times). The full closed-loop system is expected to work better when you announce your meals. There is also talk that Medtronic might develop its own smartwatch.

A lot of interesting study results about the MiniMed 780G were also discussed at the ATTD:

  • Real-world data from meanwhile 25 396 users of the MiniMed 780G: the mean TIR (time-in-range = glycemia 70-180 mg/dl) was 74.3%, GMI (= estimated HbA1c) 6.8%, TBR (time-below-range = glycemia <70 mg/dl) 1.8%, <54 mg/dl 0.5%, and time in auto mode 91%. The results were roughly the same in each country, and remained the same the longer the system was used. On average, the TIR rose 11%, but as with other closed-loop systems, the lower your TIR at the start, the more TIR you will rise. Even if you have a TIR of 80%, you can still expect a small improvement when you start with the MiniMed 780G.
  • If you switch directly from insulin pens to a MiniMed 780G, we see the same good glycemic results. So there is no reason at all to give people with type 1 diabetes who would like a closed-loop system a (stop-before-low) insulin pump first and then a closed-loop system. These results were discussed:
    • ADAPT study: 75 adults with type 1 diabetes on MDI (multiple daily injections) and a Libresensor were randomized to MiniMed 780G or further MDI. Baseline HbA1c was 9%. Results after 6 months: TIR in the MiniMed 780G group increased from 36.4 to 70.6%, in the MDI group the TIR remained the same: from 42.6 to 43.6%. TBR was low everywhere (around 2%).
    • Polish study: 37 adults with type 1 diabetes on MDI and fingersticks (so no CGM!) were randomized to MiniMed 780G or further MDI (and fingersticks). Baseline HbA1c was lower (7.1% in MiniMed 780G group and 7.4% in MDI group). After 3 months, the TIR in the MiniMed 780G group rose from 69.3 to 85%. In the MDI + fingerstick group the TIR remained the same (from 62.8% to 61.5%). Importantly, there was much less hypoglycemia in the MiniMed 780G group (TBR decreased from 5.8% to 1.8%, in the MDI group this remained the same at around 5%).
  • When switching from the Guardian 3 to Guardian 4 sensors, the glycemia results of the MiniMed 780G remained the same (mean TIR 73.4%). Although I honestly just assumed that, it's nice to see this confirmed.

Finally, at the Medtronic symposia, there was a very strong emphasis on using the right settings. The best results with the MiniMed 780G are seen at a target value of 100 mg/dl and an active insulin time of 2 hours, but not enough people use these settings. So if you have to increase your target value or active insulin time because of hypoglycemia, it is important to check over time whether you can lower it again.

4. iLet insulin-only readout

 (The iLet, a “bionic pancreas,” is being developed by Beta Bionics, led by Ed Damiano, a Boston University professor. Photo by Craig F. Walker)

Beta Bionics has been working on a bihormonal bionic pancreas for almost 10 years now, which they call the iLet. They have already developed a nice pump with 2 reservoirs: 1 for insulin and 1 for (dasi)glucagon. The integration of glucagon is not easy because it is a less stable and above all an expensive product. Perhaps that's why they tested everything first with only insulin.

The iLet system is characterized by the fact that very little interaction is required from the patient and the caregiver:

  • You just have to enter the weight of the patient and the system chooses how much insulin it will give. It is a learning algorithm.
  • The only thing you can adjust is the target value, which you can set for different times of the day.
  • You don't have to count carbs either! As a patient you only have to enter which meal you are going to eat (breakfast, lunch or dinner) and the size of the meal (normal, more than usual, less than usual or much less than usual).

The results of their pivotal trial for their insulin-only iLet system were read out on the ATTD: 440 adults and children (>6 years) were randomized to either the insulin-only iLet or standard diabetes care with a Dexcom G6. After 13 weeks, the TIR in the iLet group increased from 51% to 65% (in the control group the TIR increased slightly to 54%). This is a slightly less good TIR result than the other studied closed-loop systems, but this would be due to the fact that their study population had a greater diversity (in terms of ethnicity, in terms of baseline technology use, and in terms of baseline HbA1c). And although you should never compare results from different studies (and different study populations), I secretly did it here:

670G: Garg et al. DTT 2017; 780G: Carlson et al. ADA 2020; Control IQ: Brown et al. NEJM 2019; Diabeloop: Benhamou et al. Lancet Digital Health 2019; Omnipod 5: Brown et al. Diabetes Care 2021; CamAPS FX: Tauschmann et al. Lancet 2018; Tidepool Loop: Lum et al. DTT 2021; Beta Bionics: El-Khatib et al. Lancet 2017; Beta Bionics insulin-only: presented at ATTD; Inreda: Blauw et al. Diabetes Care 2021

5. Omnipod 5 works as well or better for type 2 diabetes

Omnipod 5 is a closed-loop system with an algorithm on the Omnipod, coupled with the Dexcom G6. Since the FDA approval in January 2022, more than 100 people have already started with this in America, and they achieve the same results as in the studies (so they said on ATTD). The CE label was requested at the end of 2021, hopefully this will be sorted out this year!

The Omnipod 5 can be controlled with a separate PDM or via an android app, for now only available on 7 types of Samsung mobile phones. This will be expanded to more types later and an iOS app is also being worked on. A simulator app was also shown on the ATTD (Google Play, Apple Store), but I have not been able to download it yet. I suspect this will only be possible in America?

In the meantime, Insulet is working hard to expand their indication to type 2 diabetes. At the ATTD they discussed the first results of the Omnipod 5 in 24 adults with type 2 diabetes on insulin. Metformin, GLP-1 analogs and SGLT2 inhibitors were allowed to continue if the patients were already taking them. The patients were divided into 2 groups:

  • 12 patients who injected insulin 4x/d: after 8 weeks the TIR increased from 46.5 to 60.5% (+14%!), while the total insulin daily dose fell from 92 to 63 U per day.
  • and 12 patients who injected insulin only 1x/day: after 8 weeks the TIR increased from 32.2 to 56.6% (+24%). Here the total daily insulin dose rose from 30 to 42 U per day, but this had no effect on weight. In both groups it was shown that the weight of the patient did not increase significantly.

These are very good results for glycemic control! Nearly all patients chose to continue wearing the system for 6 months, and these results will be discussed at ADA 2022. It is striking that these people were not required to count carbohydrates! Because people with type 2 diabetes are not used to this and because this may not be necessary for them, it was decided to work with fixed meal portions. The grams of carbohydrates in a typical meal were calculated and patients could then choose whether to eat a normal, small or large meal. It is also striking that during the first 4 weeks they only had to announce their meals if they wanted to, so really "optional". This naturally makes the step to a closed-loop system much more feasible.

Insulet (Omnipod 5) is not the only closed-loop system that is fully committed to people with type 2 diabetes. Dr Hovorka is also currently studying Cam APS FX in people with type 2 diabetes. The results will be read at the EASD 2022.

This was only part of the best news from the ATTD congress 2022. If you want to find out more about what I learned in Barcelona, ​​be sure to check out the 2nd part of my top 10 of the ATTD next week!

What do you think of these developments in the meantime? Is there anything you would like to try for yourself in the future? Let me know in the comments!



Key AID System Developments from ADA 2024

The Evolution of Diabetes Technology: Insights from Professor Chant...

AID Systems Update Spring 2024


Lees meer hier
Schrijf je in op de nieuwsbrief


Read more here
Subscribe to the newsletter