Senseonics, Roche, and TypeZero sign R&D agreement to develop automated insulin delivery system with 180-day sensor; to be tested in NIH-funded iDCL arm at three EU centers – July 27, 2017

Executive Highlights

  • On Monday, Senseonics announced that it will develop a “long-term” automated insulin delivery system with its Eversense XL sensor (180-day implantable CGM), Roche’s Accu-Chek Insight pump, and TypeZero’s inControl AP algorithm. This is an R&D agreement between the players and no commercial terms have been decided on.
  • The system will be tested soon at three EU universities – Amsterdam, Padova, and Montpellier – in an arm of the continually-expanding-and-hard-to-follow NIH-funded iDCL trial. Data is expected to support a CE Mark submission, a major win for the companies. Start date, completion date, and commercialization timing are not clear at this point, but if the trial begins soon, we could imagine a launch is possible in the ~2019-2020 time frame (our speculation).
  • This news establishes Roche and Senseonics as players in the automated insulin delivery landscape – adding another strong pipeline asset for Roche on top of the recent mySugr acquisition. This is also key for Senseonics, placing the first longer-wear implantable CGM in a major closed loop trial. It also adds to TypeZero’s impressive list of partners, which also includes Tandem and Cellnovo.

Senseonics announced on Monday that it has entered an R&D agreement with Roche and TypeZero to develop an automated insulin delivery system. Comprised of Senseonics’ Eversense XL 180-day implantable CGM, Roche’s Accu-Chek Insight pump, and TypeZero’s inControl AP algorithm, the system will be tested as another arm of the NIH-funded International Diabetes Closed Loop (iDCL) trial, eventually delivering the data to support a CE Mark submission.

While this is purely a research agreement at the moment, the ultimate goal of the collaboration is to bring a product to market. Neither Senseonics nor Roche could go into detail on timeline for study start, completion, or commercialization, though we imagine that, should the trial begin fairly soon, an integrated system might hit the market in the ~2019-2020 time frame (our speculation). The announcement brings about two big firsts: (i) confirmation that Roche is in on automated insulin delivery and may look to commercialize a system; and (ii) the first system with an implantable sensor to be tested in a major (pivotal) trial.

Notably, the trial will be funded by an undisclosed chunk of the $12.7 million NIH apportioned to the iDCL back in February. We’d note that this trial remains incredibly confusing to follow, seems to keep getting pushed back, and expands in scope and devices by the day. It’s not clear how much of NIH funding will go to the Tandem/Dexcom/TypeZero trial vs. the Roche/Senseonics/TypeZero trial.

This phase of the iDCL will be conducted at three centers in Europe: University of Amsterdam, University of Padova, and University of Montpellier. No start date has been settled on as the companies work to finalize the study protocol and ensure that the individual components of the system are fully integrated, but Senseonics indicated that Amsterdam’s Professor Hans de Vries, specifically, is “keen on getting started as quickly as possible.” Because the study is part of the blanket iDCL trial, it will have similar design to that of the US iDCL arm testing the Tandem/Dexcom/TypeZero system. That is, a six-month randomized controlled trial comparing closed loop vs. sensor-augmented pump (SAP) with primary outcomes of A1c, time in range, and hypoglycemia – notably to see all three in there. The entire iDCL trial (US + EU) will include ~360 participants; according to, the main protocol of the Tandem/Dexcom/TypeZero study will enroll 120 participants, but Senseonics cautioned that the power analysis to determine sample size in the EU cohort is ongoing, and so couldn’t comment on possible trial size. It would make sense to us if it were in the same ballpark as the US trial (same length, similar primary outcomes, etc.). NIH emphasized that the grants awarded for the four “pivotal” multi-center artificial pancreas trials in February, including iDCL, are for more fully automated devices, meaning this one won’t require mealtime boluses – similar to Tandem/TypeZero’s system and an important differentiator from the first-gen MiniMed 670G. Of course, this doesn’t preclude user interaction altogether. 

Though this is from the NIH grant, we’re not sure if Roche/Senseonics/TypeZero will also commercialize this or a similar system in the US at some point. At this stage, Senseonics management indicated that focus is on the European market, where Eversense is CE Marked. Roche reps added that, since the Accu-Chek Insight pump is currently not registered in the US, it is not expected that the study will support a 510(k) submission. A goal is to generate data to support submissions for future pump systems; what that means is unclear.

We are very excited to hear that Senseonics and Roche are in the closed loop game; along with Abbott’s entrance earlier this month (partnering with Bigfoot), the automated insulin delivery field just got a lot more interesting with big and small players partnering to bring systems to market. With Abbott and Senseonics becoming sensor components of systems, there are now more options (no-calibration; 180-day implantable) for patients to choose from on the sensor front, along with diversified pump options. We’ll keep up to date on the iDCL trial’s progress, and hope to see this system hit the European market as soon as possible.

  • We have confirmed that main protocol of the iDCL in the US (testing the Tandem/Dexcom/TypeZero system) has yet to commence. At ADA, UVA’s Dr. Boris Kovatchev (the study’s PI) presented promising data from the completed training phase (n=20) – he confirmed the main protocol will begin “soon,” but offered no additional details. In February, Dr. Kovatchev estimated that the trial would complete in 2Q18.
  • Since Roche began distributing for Senseonics in May 2016, expanded the agreement in December, and later invested in the company in May, this collaboration wasn’t a surprise. It does, however, suggest that Roche is in on closed loop, which could, along with the recent mySugr acquisition, invigorate a challenging global diabetes business (down 7% in 2Q17) – we’d assume that Roche would be the company, out of the three, to eventually commercialize the system, but these details are not ironed out at this point. We wonder what else Roche has in mind for automated insulin delivery – it could, for example, devise a system with its own soft-launched Accu-Chek Insight CGM. The pump market has been challenging for many players, so much so that Roche ceased US pump sales early this year – but perhaps that positive data in this and other future studies could lead to a market re-entry stateside.
  • Likewise, Senseonics and TypeZero established an R&D agreement in May to use Eversense data in closed loop and decision support systems. This is Senseonics’ first partnership, and though reports of the 8.8% MARD implantable Eversense are very positive, we wonder how it will be received by patients in a long-term automated insulin delivery study. Obviously, not having to change sensors for a whole six-month trial is a huge plus, and on-body transmitter wear-time was outstanding in clinical studies (median 23.5 hours in PRECISE and 23.4 hours/day in PRECISE II – the EU and US pivotal trials). TypeZero’s inControl AP algorithm will now be used in Tandem/Dexcom, Cellnovo, and Roche/Senseonics systems, making it the most ubiquitous algorithm, by our math – whether all these actually come to market remains to be seen, especially because this study hasn’t even started.
    • Senseonics and TypeZero have also been working on open loop decision support for pen users in parallel. Senseonics’ first priority is to get its automated insulin delivery program up and running, so there haven’t been any announcements to date on decision support. Senseonics management hopes to make an announcement in the near future, and called open loop support “highly attractive” because of the enhanced patient reach. 
  • According to Senseonics management, there are now “plenty” of patients using Eversense in Europe. Revenue is not meaningful at this point, and we’ll wait until the quarterly report to have a better sense of the real scale here – we assume the number of patients is in the hundreds. Per the 1Q17 update. it sounded like launches continue to ramp, with Germany (with distributor Roche, who has upped sales force from four reps to its entire intensive diabetes sales team) and Sweden (with distributor Rubin Medical) going “full bore”, and market presence in eight European countries expected by mid-2017 (Sweden, Germany, Italy and Netherlands in limited fashion, plus four more).
    • Senseonics remains in active discussion with FDA regarding approval, slated for 4Q17 (per its 1Q17 update). As of the same update, the management team expected there will be an advisory committee meeting that could take place in early fall.

Close Concerns Questions

Q: Will commercial terms be signed? If so, what is the timeline for EU launch?

Q: When will the iDCL main phase actually start, enabling data collection on commercial systems?

Q: Which company will submit and commercialize this Roche/TypeZero/Senseonics system, if it reaches that stage?

Q: How much of the NIH funding will go to this trial vs. the US arm?

Q: Will the companies aim to commercialize this particular system in the US? If not, will a system with different components from one or more of the companies?

Q: How will patients view the implantable 180-day Eversense CGM in the closed loop context – it would require at least two in-office procedures per year.

Q: Will Roche develop an automated insulin delivery system around its Accu-Chek Insight CGM?

Q: Does this mean that Roche aims to come back to the US pump market? On what timeline?


-- by Brian Levine, Adam Brown, and Kelly Close