JDRF-Helmsley Charitable Trust Closed Loop Intra-Peritoneal Infusion Workshop

April 24, 2017; New York, NY; Full Report – Draft

Executive Highlights

Last month, our team flew coast-to-coast to attend the JDRF-Helmsley Charitable Trust Closed Loop Intra-Peritoneal Infusion Workshop in downtown New York City. The meeting brought together some of the world’s leading thinkers in implantable (and external) pump and sensor technology, including Drs. Jeffrey Joseph, Eric Renard, Roy Beck, Howard Zisser, Eyal Dassau, Henry Anhalt, Katharine Barnard, Mr. Peter Lord, and so many more. At the end of the day, the nearly 60 participants came to the conclusion that IP insulin therapy confers clinical benefit – patients who have used it refuse to give it up, and one IP user in attendance went so far to say that “if you try to take my pump, you’re going to have a problem.” Speakers and others in attendance discussed the hurdles and opportunities in IP delivery; the current landscape of implantable sensors, pumps, algorithms, and insulin; clinical trial design; psychological perspectives; end user needs; funding priorities; and much more. We also heard from several industry players in attendance, including PhysioLogic Devices (implantable pump), Theranova (implantable CGM), and Sanofi (on U400 insulin).

Moderator Dr. Jeffrey Joseph summed up the meeting with encouraging words: “There is a clinical need. Patients want this. Look back at where minimally invasive pumps and sensors were 30 years ago. We need to advocate to move this forward. We need better insulins that are stable in a pump and do not aggregate and clog. We need smaller pumps, long term batteries, and pumps that are easy to put in and take out. We need better materials and designs for catheters that do not occlude. But these tasks are not that difficult. Algorithms developed for a minimally invasive artificial pancreas can be adapted nicely for IP delivery. All the pieces can fall into place. We think this is complex and invasive, but in reality, we’re just catching up with the rest of the medical field.”

One of the main cautionary statements, emphasized time and again, is that those developing implantable systems need to be wary of the competition that they will face in the next five to ten years. Subcutaneous closed-loop systems are already here and will improve significantly in the coming years. Meanwhile, one or more of beta cell replacement/encapsulation therapies and glucose-responsive insulin (previously called “smart insulin” – see the report from the tremendously valuable JDRF meeting in April 2016 on this topic) may be coming in the next decade. How will the value proposition of an implantable system stack up to these, particularly on clinical outcomes, cost (Dr. Joseph estimates ~$8,000 for the system + $1,500-$2,000 for the cost of implantation + $600/year for concentrated insulin + $600/year in physician visits = ~$15,000 over four years), user experience, target market/population segmentation, and prescribing hassle? Is it possible to create a commercial market? We believe these are the central questions for the IP field, and although they’re tough ones to answer in a fast-changing market, they’ve been the questions of interest for decades.

The ~500 type 1 patients using implantable pumps worldwide (seven in the US) love them, and all seem to pass the “I-could-never-live-without-this” test – important early reality checks for the field although questions about the generalizability of these sentiments have arisen over time. As with any long-term research area, there are important technical gaps to overcome, in this case on the hardware side (long-term, reliable, small implantable pumps and CGM sensors). Patient acceptance, clinical burden, reimbursement, and the competitive landscape in the next decade will also be critical to consider in product design. We’re glad to see JDRF, NIH, and HCT funding work in this area and look forward to following it. Read on for talk-by-talk summary.


Table of Contents 

Detailed Discussion and Commentary


Overview of JDRF Efforts in artificial pancreas Systems

Jaime Giraldo, PhD (JDRF, New York, NY)

JDRF’s Dr. Jaime Giraldo opened the workshop by reviewing the Foundation’s artificial pancreas system technology roadmap. See below for a full table, but at a high level, gen 1 entails current technology plus complementary therapies, gen 2 entails more automation, smaller size, and enhanced longevity and fault detection, and gen 3 (an expected 7+ years down the line) is expected to be fully automated and miniaturized. Before gen 3 can be achieved, JDRF believes a number of things must happen. First, CGM and pump access and adoption need to expand – according to Dr. Giraldo, only 8% of patients with type 1 diabetes in the US currently use CGM (we note that no one really knows this stat exactly – we hear double this figure frequently), and 40-50% use pumps. (As noted, the CGM estimate seems quite low to us. Dexcom’s US base alone is 150,000+ patients. Using JDRF’s estimate of 1.25 million US type 1s means Dexcom is more than 10% penetrated in US type 1s alone. Conversely, JDRF’s pump estimate of 40-50% penetration seems like a very high estimate. We ballpark it in the 30-35% range given various reports from Medtronic and others over time.) At this point, coverage is not broad enough and burden is too great – JDRF hopes to address the latter by funding efforts to miniaturize, integrate, extend the use cycle, further automate, and improve insulin pharmacokinetics. And importantly, clinical (beyond A1c!) and economic benefit have to be demonstrated to push regulatory and reimbursement ahead, particularly in populations such as children, seniors, and pregnant women. At the end of this introduction, Helmsley’s Dr. Sean Sullivan emphasized the high degree to which the two non-profits are on the same page when it comes to mission, priorities, and how to get there.    

Gen 1

Gen 2

Gen 3

Current technology

Reduces highs and lows

Complimentary/adjunct therapies (Afrezza, SGLT1/2, Pramlintide)

0-5 year access

Increased infusion set longevity

Fault detection

MEMs pumps

Advanced AP algorithms

More automated

Additional signals (?)

Multihormone (?)

4-7 year access

Fully automated


IP insulin infusion (?)

Implantable (?)

>7 year access


Current Knowledge and Landscape
Review of IP Insulin Infusion Systems, Clinical Benefits, and Challenges
Jeffrey Joseph, DO (TJU, Philadelphia, PA)

TJU’s Dr. Jeffrey Joseph, who chaired the meeting, gave a superb overview of the physiological rationale behind IP insulin delivery, as well as the practical limitations that researchers face at this point. He explained that the goal of IP delivery is to get insulin efficiently to the hepatocytes. Two-thirds of hepatocyte blood supply is derived from the portal vein – which is where >50% of insulin infused intraperitoneally enters the circulation. The hepatocytes will uptake as much insulin they need, and the remainder will proceed to enter the peripheral circulation. By allowing the liver to have the first pass at the insulin, IP delivery mimics physiologic insulin much more closely than subcutaneous delivery. Not only is this a more effective way of getting insulin into the body – numerous studies have shown faster onset AND offset of action, and neither Dr. Joseph nor Montpellier’s Dr. Eric Renard could recall an instance of IP pump malfunction causing an over-delivery of insulin (granted, it’s a small number of patients wearing them). Other proven physiological benefits include tighter glycemic control, lower incidence of hypoglycemia, and faster recovery from hypoglycemia. Of course, there are also limitations to such systems: Catheter obstruction with fibrous tissue causing under-delivery of insulin, potential for insulin aggregation within the fluidic system, adhesions to the intestines, cost and burden of implantation and refilling the pump (this could be the biggest market challenge in this area), and risk of infection. Despite these clear drawbacks, there are plenty of well-known limitations of subcutaneous delivery too, and Dr. Joseph said patients who have tried IP delivery generally do not want to go back to subcutaneous (this is a small number of patients who currently are on it). Dr. Joseph’s ambitious vision is for an implantable artificial pancreas with IP delivery to be the standard of care – something that is easy-to-use and user-friendly – for everyone with type 1 diabetes.

  • Dr. Joseph pointed out that IP delivery is “a more mature market than you think, just not in diabetes.” He pointed to the variety of long-term implantable IP catheters for peritoneal dialysis and FDA-approved implantable pumps and catheters for delivering chemotherapeutic drugs to the liver and narcotics to the spine. Diabetes is lagging behind other disease states, he said, but with miniaturization and technological upgrades (e.g., the implantable MiniMed pump is quite old – the first was implanted at Johns Hopkins by the late Dr. Chris Saudek in 1986), Dr. Joseph believes these devices can become mainstream.
  • Dr. Joseph is working with Capillary Biomedical, Inc.to develop a long-term implantable IP insulin delivery catheter with porous membranes to promote the ingrowth of vascular tissue, leading to faster and more consistent delivery into the portal vein. Capillary Biomedical and TJU are also developing flexible continuous subcutaneous insulin infusion catheters (CSII- infusion sets) with multiple holes that sprinkle insulin into a large area of subcutaneous tissue, leading to faster and more consistent systemic insulin absorption. “There has been very little innovation in this space, and this is something we need to pay attention to.” Some termed these wise words, as a number of the most common issues with subcutaneous insulin pumps pertain to infusion sets. If the last stage of delivery can avoid hiccups, that could be a huge win, especially in implantable systems, where an occluded catheter would likely require a minor surgery to fix. There are currently many labs looking at chemical and material approaches to minimize fibrosis, Dr. Joseph said – one recent paper identified a potential therapeutic target for eliminating the foreign body response.

Questions and Answers

Dr. Garry Steil (Boston Children’s Hospital, MA): What’s the best evidence in the literature for how much insulin is delivered portally? How much insulin goes through other mechanisms?

Dr. Joseph: The answer is – I’m guessing I reviewed 40-50 manuscripts – and every one says the majority of insulin is delivered portally. There are a few studies with radioactive markers and they showed that insulin rise occurs faster in the portal vein than the systemic system. But I learned that everyone did something different – the catheter was placed in a different location, the method of delivery was different, animal vs. human work. We can’t make any conclusions. But more than 50% is absorbed portally when delivered IP – that’s my simple answer.

Comment: On the issue of catheter placement, it’s not just higher or lower but where you put it in the cavity – as you go inside toward the mesentery or omentum, you get more portal delivery. That’s a great area for more study with modern technology.

Dr. Sanjoy Dutta: Does the average person need to still insulinize to supplement the peripheral tissue?

Dr. Joseph: Insulin is absorbed from the peritoneum, the liver then takes what it needs, it’s taken up by hepatocytes, then the insulin that leaves the liver comes out through the hepatic vein and goes to the lungs and the rest of the body. Everyone thinks the pancreas controls peripheral insulin delivery, but it’s really the liver, or both. So no,  patients do not require supplementation. Patients have improved blood glucose control with higher portal vein insulin concentrations despite much lower systemic insulin levels.

Current Status of Implantable CGM Suitable for Closed Loop Systems

Dan Burnett, MD (CEO, Theranova, San Francisco, CA)

TheraNova CEO Dr. Dan Burnett commented on a number of implantable sensor mechanisms, asserting that there are tradeoffs to consider regarding accuracy, lag time, attenuation, longevity, power draw, and footprint. With an implantable glycoenzymatic sensor such as Glysens’, for example, the tradeoff is longevity for lag time – theoretically, the higher the enzyme load, the longer the lag, but also the longer it would last in the body. Fluorescence-based sensors (i.e., Senseonics) also face issues with lag time, and thus require compensatory algorithms. Optical sensors (i.e., Dr. Mark Arnold’s spectroscopic near-IR sensor) senses in real-time and doesn’t bleach, so it may be well suited for a long-term sensing implant. (We sat at Dr. Arnold’s table and saw his glucose-sensing photonic integrated chip. It was miniscule – probably 1 cm x 1 cm – and he said it could also sense lactate, ketones, and other compounds based soley on NIR spectra!). Dr. Burnett reminded attendees that his company, Theranova, has JDRF, NIH, and Helmsley-funded early-stage projects with each of the three sensing mechanisms: The glycoenzymatic sensor is in the preclinical and clinical stages at the moment, with impressive data (from a study designed by Drs. Howard Zisser, Eyal Dassau, and Frank Doyle) in pigs showing less lag time IP vs. subcutaneous; the fluorescent sensor is in the preclinical stage and has been placed for over three months at a time in sheep – similarly, the rate of signal appearance and disappearance is much slower in the subcutaneous space vs. IP; finally, the optical sensor is in the preclinical and clinical stages. With aid from the (former) UCSB trio, most of the clinical work done by the acclaimed Dr. Basu at Mayo Clinic, and a few sensors in the works, Theranova has a lot going on (and is also developing implantable pumps). The most obvious questions for implantable CGM are cost, accuracy/calibration, form factor, and implantation hassle relative to where subcutaneous CGMs will likely be in 5-10 years from the major players (e.g., bandage-like).

Questions and Answers

Dr. Joe Lucisano (CTO, Glysens, San Diego, CA): On average, we’ve seen on the order of about a nine-minute time lag. The sensor itself has an intrinsic time response of about two to three minutes.

Dr. Burnett: Have you tried putting it IP?

Dr. Lucisano: No, not yet.

Dr. Burnett: Maybe it could be improved?

Dr. Lucisano: Yeah, we also didn’t see an increase in time lag over the duration of wear.

Dr. Burnett: Did you see capsule formation?

Dr. Lucisano: Certainly there’s a mild fibrotic response that occurs, but the sensor is still sensitive enough to give an adequate signal.

Current Status of Algorithms for Closed Loop Intra-Peritoneal Insulin Infusion
Eyal Dassau, PhD (Harvard University, Cambridge, MA)

According to Harvard’s Dr. Eyal Dassau, PID, MPC, and Fuzzy Logic algorithms designed for subcutaneous closed loop therapy could easily be adapted for IP insulin delivery, and sufficient miniaturization of automated insulin delivery systems can be achieved without degradation in clinical performance. In a study using the Roche Diaport for IP delivery or sub-cutaneous delivery, unannounced meals, and a similar algorithm, the DiaPort group spent 66% of time between 70-180 mg/dl, while the subcutaneous delivery group spent 44% of time between 70-180 mg/dl (Dassau et al., Diabetes Obes. Metab.). To test the feasibility of embedding a closed loop controller on a chip, Dr. Dassau and colleagues moved the PID controller to simulated closed loop on a microprocessor (Arduino platform), and found that the CGM was within 70-180 mg/dl an impressive 98% of the time. They were equally successful with a Zone MPC algorithm – they found that challenging it with unannounced meals resulted in equivalent glycemia to when the controller was stored on the computer. In addition, the chip-based design significantly reduced execution time and energy consumption (Chakrabarty et al., IEEE Trans., In Press). By all accounts, it seems like the algorithm piece of implantable automated insulin delivery is the most “ready to go,” relative to CGM sensing and insulin infusion. This will be put to the test in a canine feasibility trial shortly in collaboration with PhysioLogic Devices (pump-sensor system), Vanderbilt’s Dr. Justin Gregory (who will conduct the trial), and UCSB/Verily’s Dr. Howard Zisser. The first iteration in the trial will be with PID, followed by MPC – we can’t wait to see these results!

  • Dr. Dassau took a moment to thank the late great Mr. Alfred Mann. Mr. Mann provided inspiring leadership and significant investment to develop a fully implantable closed loop system including IP insulin and IV glucose sensing. This work was validated in human feasibility studies sponsored by MiniMed in 2000 – without him, this meeting may not have happened.

Current Status of Insulin Sutable for IP Delivery

Eric Petreto, MD (Vice President Device Solutions, Sanofi, Paris, France)

Phoning in from France, Sanofi VP of Device Solutions Dr. Eric Petreto was the source of a paradoxically inspirational reality check: Implantable pumps are outdated (“we’re using the technology of 30 years ago”) and not designed for current insulin, and existing insulins are not suitable for long-term storage in implanted devices. On the bright side, this vacuum has left plenty of room for innovation in the pump space, and Dr. Petreto believes that, once there is are definitions for target populations and value propositions, the field can be ambitious. We felt this was right on target (and probably a level of minimum acceptability). While Dr. Petreto said that current pumps are constructed with technology from 30 years ago, today we have miniaturized electronics, better materials for primary packaging (non-reactive with insulin, could combat clogging), greatly improved power management technology, and connectivity at our disposal. Today, implantable pumps have a shelf life of seven to nine years, but Dr. Petreto stated that could easily be pushed to 12 years or more. An incentive for investment in this technology does exist – as we heard throughout this meeting, patients who are using implantable pumps invariably believe that they have a better quality of life, even with their 1980s-era pumps, and don’t want to give up their devices. (And we presume that companies like Theranova and PhysioLogic Devices are leveraging the modern capabilities outlined by Dr. Petreto to cater to this population and beyond.) We do point out that it’s a very small number of patients on these pumps and until value can be shown more broadly, while we respect these views, we don’t think they will necessarily result in widespread adoption / reimbursement. The other half of the equation is insulin, said Dr. Petreto. Dr. Petreto focused on Sanofi’s Insuman Implantable U400, which supplanted Insuplant being used in implantable pumps in recent years and is now used in four EU countries (France, Sweden, the Netherlands, and Belgium). Implantable insulins are further along than pump technology; said Dr. Petreto: “Can we do highly concentrated insulin? Yes. Can we do it thermostable? Yes. Can we deliver it? Big question mark.” The other question mark is a matter of comfort – the current formulation is only stable for 45 days in a pump. That means a patient needs to go to the doctor for refills ~eight times every year. Dr. Petreto would love to see that improved considerably. We too see that frequency as major barrier, given the burden that diabetologists and patients already face (we suspect not that many patients even go to the doctor two to four times a year). One idea raised was that perhaps companies could sponsor local centers where refills take place – it does seem like nurses, PAs, and others could take over the refills though the volume of people would need to be far higher to make this worthwhile.

Regulatory Considerations and End User Perspectives

Regulatory COnsiderations for Implantable CGM Systems

Campbell Hutton (Senior Director of Regulatory Affairs, JDRF, New York, NY)

JDRF’s Ms. Campbell Hutton raised some interesting regulatory considerations for implantable AP systems, beginning with: Where will jurisdiction for these systems lie within the FDA? There is a lot of expertise within the Agency – Division of Chemistry and Toxicology Devices in closed loop and General Hospital Devices in implantable pumps – but who will take the lead? Will the drug manufacturer need to file with CDER in parallel? (We might assume Dr. Courtney Lias’ branch would spearhead the process, with input from other branches as needed.) Regardless, and despite the novelty of an implantable system, Ms. Hutton pointed out that there is already quite a bit that we do know: (i) Regulators have had a decade of experience with various artificial pancreas systems, and Medtronic’s MiniMed 670G has already been approved; (ii) There is a lot of available information about implantable CGM, though no systems are yet approved in the US (Senseonics’ Eversense has been rolling out in Europe since mid-2016 and is under FDA review now); and (iii) There is even more data available pertaining to implantable pumps, and not only for insulin delivery. The Device Division has really propelled the field forward in recent years, and we do not see this as the biggest barrier for this field – assuming a sponsor engages in FDA discussions early and often. The barrier may be more the other divisions that would pursue this work.

  • Ms. Hutton provided a glance at testing requirements the FDA may request: Preclinical (pump/CGM accuracy; algorithm in silico modeling; catheter – mechanical and functional testing; drug stability and compatibility; electrical safety; electromagnetic, MRI, and CT compatibility; bio compatibility); Animal testing (implantation procedure; refill procedure; flow rate; tissue response); human factors (simulated use); clinical studies; and post-approval study.
  • Presumably this is an area where outcomes beyond A1c, particularly time in ranges, would be of high interest.
Expectations and Acceptability of Implanted Thin Pump- Technical and Surgical Perspective
Henry Anhalt, DO (CMO, T1D Exchange, Boston, MA)

T1D Exchange CMO Dr. Henry Anhalt introduced PhysioLogic Devices’ ThinPump system and presented data from the Helmsley Charitable Trust-funded ThinPump Adoption Study, where he and psychologist Prof. Katharine Barnard probed factors that would influence uptake of the system (animal studies this year; US and EU clinical trials expected in 2021) and its potential impact on people with diabetes. For background, the ThinPump system is an implantable, IP artificial pancreas (see below for pictures). It is 70% smaller than the existing Medtronic MiniMed open loop pump, has a fully automatic closed loop mode, and requires no skin-mounted devices or supplies. The system is expected to have a 10+ year life, need refills every three to six months, and work in tandem with Biorasis’ IP glucose sensor (one calibration/week, one-to-two-year sensor life objective). UVA/Padova simulations show that, for meals with <70 grams of carbohydrates, the closed loop mode would control average blood glucose to less than 154 mg/dl with no user input. The early-stage project is definitely ambitious, but Founder and CEO Peter Lord is the developer of previous generations of MiniMed implantable pumps and has support from NIH, JDRF, and an impressive catalogue of collaborators (Verily/UCSB’s Dr. Howard Zisser, UCSB’s Dr. Sumita Pennathur, Harvard’s Drs. Eyal Dassau and Dr. Frank Doyle, Montpellier’s Prof. Eric Renard, Uconn’s Dr. Diane Burgess, etc.).

  • Results from the ThinPump Adoption Study survey (n=581) almost uniformly indicated that people would be willing to try the system. Surveyed patients (mean A1c 7%; 74% female; mean age 43 years; 78% on a pump) had fairly positive views on the device and procedure:
    • ~90% thought that the initial pump device implant, annual sensor replacement, and transcutaneous refill procedures every 3-6 months were either “fully acceptable” or “somewhat acceptable” (we don’t view these as synonymous)  
    • ~83% of patients thought that the system would be at least a seven out of ten in terms of ease of use. (In NPS parlance, the 9s and 10s are of most interest – see below)
    • ~75% said at least seven out of ten when asked how likely they would be to use the pump device when it hits the market. However, only 46% said nine or ten, translating to a pretty low net promoter score. Also, preferences are less and less linked to actual behavior due to reimbursement.
    • Indeed, the consensus sticking point was reimbursement – only ~31% said at least seven out of ten when asked how likely their insurance would cover the system, and ~41% responded with a four or less. Either way, we can’t imagine patients would necessarily even have the right information to characterize this aptly.
    • Overall, these are encouraging results but very difficult to draw big conclusions from. These were mostly current pumpers, whereas the IP field will need to expand the pump market to build a sustainable business. Moreover, what people say they will do in a survey about getting an implanted device vs. what they will do in the real world are two very different things; to boot, we expect their preferences would change if reimbursment were available.
  • Dr. Anhalt included a number of quotes picked up from focus groups conducted with those who participated in the survey, reflecting enthusiasm and concerns about the ThinPump. We note that despite the survey numbers given, the qualitative responses suggest a number of concerns, before reimbursement is even addressed. That said, we agree that patients on IIP appear to love it!
    • “I would wear this but not a traditional pump”
    • “The idea of feeling something kind of snake-like, or even worse, worm-like, inside you and wriggling around…”
    • “It’s just so massive that I wouldn’t even consider it.”
    • “If the battery life was shorter (e.g., five years) could the device be smaller?”
    • “Would the device set off the security alarm at the airport?”
    • Concerns about sensor accuracy.

Questions and Answers

Dr. Mark Arnold (University of Iowa, Iowa City): The patients were concerned about calibration and accuracy. What information did you give them?

A: I don’t think we specifically asked about calibration, unless my memory serves me poorly.

Dr. Katharine Barnard: We didn’t. In terms of accuracy, people who have previously used CGM and had poor accuracy, poor experience with CGM, were most concerned about the accuracy of implantable system, in addition to connectivity issues with implantable device. On accuracy – this thing will be in my tummy for a long time, so what if it fails?

Q: I’ve used an implantable pump since 1990. These people didn’t use a pump previously – those who are using implantable MiniMed wouldn’t give it up and go back. You said ~seven of ten would be willing to try this, when research studies were performed before, physicians asked patients if they wanted to try it, and we did, and when the study closed, people were very unhappy about that. A few of us were able to go to France and maintain the pump that way. All the people I know were not able to give it up, whether for better control or because they rely on it to survive. (Editor’s note – we believe that to date, people using this pump had high hypoglycemic unawareness and saw a very high value in this pump – we don’t necessarily know that this experience is generalizable, though we also don’t know that it isn’t!)

Overview of Psychosocial Considerations for Implantable Systems from a Physician’s Perspective
Katharine Barnard, PhD (Bournemouth University, UK)

The always-insightful Dr. Katharine Barnard reviewed psychosocial considerations for implantable systems, exploring topics such as diabetes burden, individualization of technology, adoption, and expectation management. Her mantra when it comes to technology is: “Right patient, right device, right time, right support.” Some of our favorite quotes from her talk are below.  

  • “Closed loop is associated with positive impact and decreased burden. But alarms, connectivity, and intrusion in daily life are downsides. It’s NEVER about glycemic control. Rarely do people complain about that; it’s about how glycemic control is ruining their lives.”
    • Fears and concerns about technology are no different for IP systems vs. closed loop systems. People worry that they may become too complacent or dependent on the system. And what if it gives too much insulin? Pain, also.”
  • “Access to trained healthcare providers is one of the biggest barriers to uptake, adoption, and continued use.” She underscored the continued support and reassurance patients need when starting a new device or therapy – this will be particularly true of IP devices, since patients cannot troubleshoot the actual implanted device on their own.
  • “People expect [PhysioLogic Device’s ThinPump] to do a better job from a biomedical perspective – they just want a better quality of life. We sometimes forget the absolute relentless nature of knowing there’s no cure and you’ll live with this forever – wake up in morning with blood glucose of 17.2 mmol/l (310 mg/dl), that ruins the rest of your day. It’s about feeling better overnight, knowing you’re less likely to get a hypo, getting better sleep, waking up with a blood glucose of 5.5 mmol/l (100 mg/dl) that lasts until lunch – it makes people feel better and relieved that the risk of complications goes down.”
  • For whom do technologies work? That’s an impossible question to answer. People with diabetes aren’t some automated robotic army. They are all different. We need technologies that suit people across the spectrum.”
  • “Patients don’t really care about these systems; they care more about their lived experience. I have an iPhone, and I have no idea how it works inside the box, I just want to be able to turn it on and call my friend.”
  • “Will this help me have a more normal life, give me more freedom in my life to enjoy the things I want to do, reduce the burden, allow me to think less about diabetes? People want a better quality of life, to be less exhausted, and hopefully to be happier.” (Editor’s note: this is important to note, but translating it into value and reimbursement is absolutely essential; some patients did not particularly address this.)
  • “Uptake and continued use of diabetes technology remains poor and low. That’s not unique to any specific country.”
  • “According to Dr. Rich Bergenstal’s research, CSII users spend 4.4 hours weekly on diabetes tasks, while it’s 3.4 hours for MDI. Is it worth the tradeoff? For some yes, for some no.”
    • Dr. Barnard added that the great Dr. Richard Rubin’s son once said: “At least once every 15 minutes I have to think about my diabetes.”
  • “We always say we want to reduce burden, but then we throw hurdles in the way.”
  • “This is not just about me – I also want to reduce burden for my family. How can they worry less and do less to support me? Improved sleep for loved ones, less stress for partners. I want to argue less about blood glucose levels. I want to, for once, talk about something other than diabetes.”
  • “One of the benefits of ThinPump is that it could give you less stuff to carry. It may seem trivial, but it’s kind of not.”

Questions and Answers

Dr. Sanjoy Dutta: Issues with implantable systems aren’t different from other devices. That’s encouraging, right?

A: The perception is that people don’t want a snake or worm implanted, but one thing that came out clearly is that actually people are not concerned about the surgery. One woman said if she doesn’t ever trust the device, she’ll just have it taken out.

Q: A new thing about devices in general is the term “cuddle killer”…

A: Yeah, can you feel it, depending on the size. Same thing with pump therapy. When you have sex, I always just say take it off, but they don’t think to do that.

Q: I wonder if psychosocial aspects will be different for adolescents and adults. I recently had a college student with type 1 on a pump who doesn’t swim anymore because he’s embarrassed about injection site reactions. When we’re 40, we don’t really care. How would that play out here?

A: Actually, for adolescents, the perception externally is that they don’t care about glycemic control, but that’s not my perception at all. They are worried, but that’s balanced by the social aspect of being different. And kids can be not very nice to each other sometimes – they can be mean. They are mean. I think from a “will they want to wear it perspective,” I think they will want to, and depending on where the scar is, that’ll play a big part in whether they want it. A lot of the kids say “it means my mum can sleep.” For the kids, they really want their parents to have better quality of life. But yes, there are different issues.

PwD Perspective on Implantable Systems

Mr. Greg Peterson and Ms. Alecia Wesner

In a special segment, Mr. Greg Peterson (who has used implantable MiniMed pumps since the ‘90s) and Ms. Alecia Wesner (who wears CGM and a pump and has been in three subcutaneous artificial pancreas trials) faced off in a fiery debate over implantable systems.

  • Mr. Peterson was as enthusiastic about IP insulin infusion as we have seen just about anyone about any therapy:The pump is ancient, but the physiological response to IP insulin is absolutely, undeniably, wonderfully perfect…You get irritable on sub-q. On IP, you’re happy and you feel better. It’s glorious. If you try to take my pump, you’re going to have a problem. I don’t care about the surgeries or the lump in my body. This technology is the answer to our prayers.” He went on to explain that there are six Americans who have implantable pumps – three of which were in the room – and none of them would go back to subcutaneous infusion if given the choice.
    • Ms. Lindsey Inserra, who has an implantable pump out of necessity due to difficulty absorbing insulin through traditional routes, said she wouldn’t trade her pump for the world. A small woman, she commented that the pumps hit her bones, making her uncomfortable, and she has had pump failures, but those don’t change her mind. If not for this pump, she’d either be tethered to an external pump with a catheter traveling to her IP space or have to practically live in an ICU. She expressed optimism for future iterations of implantable systems, which will presumably address safety issues and work with improved insulins. (Hearing this patient was very interesting – we do not necessarily believe her experiences could be generalized for an entire population of type 1 patients, interesting though they are.)
    • Mr. Chris Witkowski, the third person in the room with an implantable pump, echoed the fervor of Mr. Peterson: “The most important thing to have is IP delivery. I travel internationally, I can travel 12 hours, and I don’t have to worry about anything. People don’t want to have to think about diabetes constantly, and I’m one of those people now, but not on subcutaneous delivery. Relying on subcutaneous delivery lowers my efficiency to do my job. Alecia made a comment about concerns in case something went wrong –well we were all on subcutaneous insulin and we had the same concerns, but once we tried IP, we loved it. I think Alecia is one of the outliers, in the minority there … Life is so much better with this pump. I built my own company, I’ve had arguments with the CIA and DoD on this pump, and you just can’t function as well on subcutaneous insulin as you do on implantable. Helmsley and JDRF, I’ve been talking to you for a decade, when are you going to do it? There’s no question here, this is by far the most superior form of treating type 1 diabetes, and it’s not even comparable to something else. When are you going to do it?? Please! This is life-changing, this is the answer we are looking for.”
      • Ms. Wesner took issue with Mr. Witkowski’s claim that he can do his job better with IP insulin, because it implies that everyone else with type 1 diabetes doesn’t do their job as well as they can. This is where the debate became a bit fiery, and while the passion of the participants was heartfelt, we did not feel it was altogether relevant for the discussion. Mr. Witkowski, an n of 1, believes that he functions better using an implantable pump – of course, that doesn’t translate into a need for everyone to have this technology (including those who may be managing diabetes well on MDI or traditional subcutaneous pump therapy). The value of enthusiasm is hard to quantify – the value of “time in zone” showing different technologies would be helpful to have.  
  • Ms. Wesner offered a more balanced account, beginning with the fact that she’s a little scared by the thought of having an implantable pump, and she knows she’s not the only one. She acknowledged the others’ rave reviews, but fears, for example, getting in a car accident and not being able to rip off her pump. She is not totally implant-naïve; she once participated in a three-month trial with the Senseonics Eversense implantable CGM, and in her view, that “it was not a fun experience.” (We note this is an n=1 experience and challenging to generalize.) Toward the end of her portion, it seemed that she’s not against IP delivery, but is more concerned about safety. She pointed out that both of the men at her table with implantable pumps had broken systems at the time and had to go to France to get them repaired. “I don’t want to do that – I run a company, have a full-time job. I don’t want to think about diabetes. As much as I don’t want to carry devices, I will if I know I’m safe.” There’s no one universal therapy for patients – everyone has different risk-benefit calculations that lead them to settle on different options. The big question is how this market will segment and whether an IP-based system can be assigned a value that would generate reimbursement and would drive a sustainable business.
  • According to a T1D Exchange slide, people on IP insulin experience much less severe hypoglycemia than their subcutaneous counterparts. In one study, subcutaneous insulin led to 19 severe hypoglycemia episodes per 100 patient-years, while the intensively controlled group experienced 62 per 1oo patient-years. In a separate study, the subcutaneous control group experienced 15.2 severe hypoglycemia episodes per 100 patient-years, while the IP intensively controlled group experienced just 2.5. These data are rather remarkable, and as Mr. Witkowski noted, severe hypoglycemia is just the tip of the iceberg. His comment reminds of us the last JDRF Workshop on hypoglycemia, where Dr. Betsy Seaquist of the Regents of the University of Minnesota discussed the significance of the 55-70 mg/dl blood glucose range. Our view is this varies for patients …  

Current Gaps and Challenges

Breakout Session A: Technology Gaps & Needs - Landscape and Commercial Opportunities

Howard Zisser, MD (Verily, Mountain View, CA) and Sumita Pennathur, PhD (UCSB, Santa Barbara, CA)

Verily’s Dr. Howard Zisser and UCSB’s Dr. Sumita Pennathur led one breakout session where attendees identified pain points and areas for innovation in current implantable systems (pumping mechanisms, insulin concentration, biocompatibility, sterilization, and the current pump cycle).

  • Currently all available pumps are positive displacement piston pumps, while Dr. Pennathur believes MEMS (microelectromechanical systems) could open up “all kinds of crazy pumping with small delivery devices.” She alluded to Cam Med’s electrochemical Evopump, a “bandage-like patch pump” which we first saw at ATTD. The pump has a unique design, and is discrete, low-cost, and extremely accurate, but only holds 300 units and is intended for disposal following three days of use. Obviously, in this form, Evopump wouldn’t be suitable for implantation, but attendees pointed out that it can be made reusable with more concentrated insulin and reservoirs that still open following an electrolysis reaction, but have flexible, piston-like diaphragms so they only let out a little bit each time.
    • One major hurdle for miniaturization is the greater probability of clogging when concentrated insulins are enclosed in smaller spaces. Case Western’s Dr. Michael Weiss said the protein would be more likely to unfold and deposit on tubes, on valves, and between the diaphragm and the bottom of the pump chamber in a smaller space. Between his lab’s work to optimize the stability of concentrated insulin and Sanofi’s work to optimize surfactants in insulin, Dr. Weiss believes this is solvable problem. Perhaps solving the problem is more crucial than we think – currently, insulin deposits are rinsed away from pumps with 0.1M NaOH, which many attendees believed may be too concentrated and detrimental to the health of the patient.
    • Another hurdle, brought up by Dr. Pennathur, is that FDA hasn’t approved any MEMS technology and probably won’t for a while. While we’re not completely sure why the Agency would be so hesitant, one possibility is that no such device has been submitted to the FDA. For instance, the long-in-development Debiotech JewelPump debuted at ADA 2010, but has not had a meaningful timing update in years.
  • Biocompatibility, particularly of the clog-prone catheter, was also a major topic of discussion. TJU’s Dr. Jeffrey Joseph asserted that a soft, flexible catheter is least likely to induce a fibrous reaction (“biosymbiosis,” as termed by Dr. Zisser) and having multiple holes/slits would be ideal should one be occluded. T1D Exchange’s Dr. Henry Anhalt, former CMO at Animas, said that his former company was considering coating the tips of teflon catheters with 5-FU (a chemotherapy agent that inhibits macrophage function and collagenous tissue deposition) or another antibiotic. Theranova CEO Dr. Dan Burnett countered that 5-FU and “steroid tricks” are actually not necessary in the peritoneal cavity, and that simple mechanical measures ensuring that the fibrin sheath has nowhere to latch are effective. According to Glysens CTO Joe Lucisano, efforts to consistently prevent adhesion with VEGF and steroids have failed, and engineering the implantable sensor with benign materials has worked for his product thus far – the only compound that he has known to prevent fibrosis is porcine extracellular matrix. We would add that a recent JDRF- and Helmsley-funded study demonstrated that inhibition of a central component of the foreign body response (CSF1R) selectively halts the fibrotic cascade in the presence of implanted biomaterials in rodents and non-human primates.
  • From an algorithm standpoint, Drs. Eyal Dassau and Marc Breton confirmed that faster-on/faster-off insulins would make PID, MPC, and Fuzzy Logic Algorithms all work better. Dr. Breton added that all current systems work without a bolus, just not that well –faster on/faster off insulin would obviously improve delay, and therefore, and algorithm’s performance. An interesting mini-debate ensued: Dr. Dassau pointed out that, in simulations, his algorithm with current insulins can get to 65% time in range for an unannounced meal. “It’s all about performance and expectations.” Mr. Peter Lord responded that he doesn’t view moving to fully closed loop (i.e., no bolus or meal announcement) as a gradual process; “I really do believe there’s a point at which people will just stop bolusing. I think we really have to hit that threshold.” That threshold will of course differ greatly by the patient.
  • Percusense CEO Dr. Rajiv Shah called for innovation on other issues he deemed “non-trivial” such as sterilization and real-time testing. If you’re going to break the mold, you need to develop test strategies to de-risk design in real time. A lot of new things are contemplated, and a there is a lot of complexity associated with determining whether different approaches make sense.” Yes! We love this high-level view of what it takes to make a system function.
  • The traditional four-year pump cycle doesn’t make sense for a lot of reasons, but Dr. Breton pointed out that if a future pump stays in the body for 16 years, the current reimbursement/warranty cycle would be completely senseless. One solution would be to extend the length of this cycle to reflect the use period of the implantable, though we have to imagine pumps introduced onto the market in 5+ years will come standard with over-the-air software update capabilities.

Breakout Session B: Clinical pathway and anticipated challenges, Target Product Profiles with emphasis on projected outcomes, and target populations for these systems. Implications on clinical trial design

Dr. Eric Renard (University of Montpellier, France) and Dr. Korey Hood (Stanford University, Palo Alto, CA)

University of Montpellier’s Dr. Eric Renard and Stanford’s Dr. Korey Hood moderated a second breakout session where attendees discussed clinical trials that should be performed, trial design, and target population. We didn’t hear this session directly, but here’s what we gleaned from the recap:

  • It is known that people on IP insulin therapy have less severe hypoglycemia, but do they have a lower risk of DKA? This has never been studied (and would be a simple study to perform) – when IP delivery stops, how long does it take for ketones to go up? Is it slower than with subcutaneous insulin delivery? Also, from what we understand, this group is more prone to severe hypo, and we aren’t sure how that affects these stats and results.
  • The group agreed that there should be more trials seeking to understand why an implantable system may be better from a psychological perspective. Do implantable systems make patients feel safer? How is this valued?
  • A system should be designed for all type 1 patients, not just “difficult” ones (e.g., hypo unaware, poor insulin absorption). Dr. Renard added that “if it is the best way of treating diabetes, then it should be offered to anyone with type 1 diabetes.” We agree that a broad design approach makes sense, as “difficult” groups are difficult to build a business on; that said, it was not discussed how value will be assessed for future automated insulin solutions, much less traditional closed loop systems.
  • With regards to trial design, the group decided that the best endpoint would be time in range (hear, hear!), and it was said quality of life aspects, such as time spent on tasks, mental and physical burden, and health economics should also be included. We think this is a great starting point list, and each of these components should be toward the top of most trials assessing a diabetes therapy. Dr. Hood suggested having alternative designs beyond RCTs – for example, if an individual does not succeed in one arm, then he/she should be randomized to a different option. “Sequential, multiple randomizations are smart.” In this way, it would be easier to determine which subsets of type 1 patients excel on which therapies. The bigger question for this field is, “What endpoints would payers care about, and would achieving those endpoints with incremental improvements drive sustainable reimbursement for this field?”
    • The comparator group would likely have to be minimally invasive subcutaneous closed loop systems, which will only get better over the next few years. This is a key point, because the rapid improvement in the subcutaneous space (smaller, cheaper and more accurate sensors; better algorithms; faster insulins) will result in stiff competition for a first-generation implantable system. Does IP delivery actually result in cost-effective therapy and better outcomes? If not, will patient preference alone be enough to sway payers?


Funders’ Perspectives
Innovation, New talent, Consortia, Partnerships (other funders, industry)

Guillermo Arreaza-Rubin, MD (NIH, Bethesda, MD)

NIH’s Dr. Guillermo Arreaza-Rubin shared that the Institutes gave a remarkable $177 million to glucose control technologies from 2010-2016 ($21 million on 12 active implantable glucose control projects), and intends to release at least five FOAs for FY2018-2019. The amount of money NIDDK plans to contribute was not disclosed, but the funding priorities are: (i) Support a novel generation of components for integrated systems including implantable and IP-IP; (ii) Support pilot and transitional clinical studies of emerging automated glucose control technologies; (iii) Testing technologies in older adults and relevant subpopulations; and (iv) Support advanced clinical trials on systems with robust feasibility data. Pencils were very active when Dr. Arreaza-Rubin articulated these priorities! Dr. Arreaza-Rubin then presented his view of goals and challenges in the landscape: (i) Improve collaboration and coordination among funders; (ii) Attract and train new talent including bio-engineers working in other fields – we were pleased to see Dr. Sumita Pennathur, a mechanical engineer at the meeting, though she was only drawn into diabetes by her daughter’s diagnosis; (iii) Be willing to fund cutting edge/high risk research; (iv) Promote academic/industry collaboration (we’d love to hear more details here); (v) Support studies for the design of more user-friendly, personalized systems based on input/preference by patients/users; (vi) Studies to improve accuracy, reliability, durability, usability, and safety of technologies; (vii) Long term comparison of system in terms of glycemic and patient-reported outcomes, including quality of life; and (viii) Include cost-effectiveness in studies. This is an awesome and incredibly broad set of goals, and we hope all funders and even industry can be on the same page in achieving them!

  • JDRF’s Dr. Sanjoy Dutta urged attendees to consider what the landscape will look like in five years – “what will your product be compared to when it is released?” Other therapeutic approaches – encapsulated islets, beta cell regeneration, glucose-responsive insulins, SGLT-2 inhibitors in type 1, basal insulin/GLP-1 combos – are advancing and being funded as well, so the value proposition of a competing system needs to be well-planned and thought out.
  • Helmsley Charitable Trust’s Dr. Sean Sullivan emphasized that the Trust weighs regulatory and reimbursement plans heavily when they consider funding opportunities. His added that HCT wants to ensure that people will use these systems if they come to fruition. People who have used implantable pumps don’t want to give them up, but how many people would adopt the systems if they came to the market?
  • Dr. Jeffrey Joseph estimates that an implantable AP system will cost ~$15,000 over four years: ~$8,000 for the system + $1,500-$2,000 for the cost of implantation + $600/year for concentrated insulin + $600/year in physician visits. “The implantable AP System will need to function for at least three years to significantly decrease the total cost when compared with the cost of current insulin pump/infusion set/CGM systems. I believe an initial product will be an external needle-type CGM communicating with an implantable insulin pump with IP delivery.”
    • Prof. Eric Renard indicated that an implantable pump currently costs ~35,000 euros and can be used for ~eight years. Accounting for other closed loop components, a CGM in Europe is ~3,000 euros/year, insulin is ~5,000 euros/year, implantation costs ~6,000 euros, and physician visits for pump refills cost ~4,000 euros/year. That sums to ~140,000 euros (~$158,000) over eight years. Prof. Renard explained that the high costs are related to the low volume of patients using IP pumps. Still, we would note that this is an order of magnitude more expensive than current products.
    • According to PhysioLogic Devices President Mr. Peter Lord, the company’s ThinPump IP-IP system “will achieve a 15% to 25% reduction in overall cost of therapy over products like the Medtronic 670G by extending the longevity, extending the refill interval, and eliminating disposables.” By his analysis, parity would occur at five years. He pointed to pain pumps (Synchromed and Prometra), for which reimbursement is approximately $20,000 for implantation and hardware – he believes ThinPump will be priced “at that level or less” due to the cost-effectiveness of MEMS pumping systems.
    • Mr. Lord believes that a reliable IP automated insulin delivery system would have to rely on an IP sensor, not a needle sensor: “My experience tells me that it will require the protected environment of the IP space and larger, catheter mounted, redundant sensors to provide the required absolute reliability for an automatic insulin delivery system.” While an IP sensor may be ideal, there is certainly an argument to be made, in our opinion, for using an implantable pump with a subcutaneous needle-type sensor.
Questions and Answers

Dr. Dan Burnett (CEO, Theranova, San Francisco, CA): I was in venture capital for a few years, and I invested in one of the first trans-aortic catheter valves. It was used in a small patient group at first, but now it’s used in others. If you swing for a home run and try to hit it in one fell swoop, I wouldn’t invest. But if you give $4M and try to get a system in one group of patients, and then iterate from there, that’s much more appetizing for me. I can’t raise $50 million to get a home run, but the $4 million study is doable.

Dr. Sanjoy Dutta: But then the labeling will only reflect that indication.

Dr. Burnett: I would say just create a post-market registry. It will be used off label. If you don’t want to support it all the way with grants, taking it in bites is probably a more reasonable approach.

Dr. Sumita Pennathur (UCSB, CA): I got into this field because my daughter was diagnosed, but now I feel like it would be so great to tell people in the MEMS field to come and listen to you guys. So many people who spoke today made me want to make a difference.

Dr. Dutta: Yes, we’re constantly trying to bring new blood into the field. (Editor’s note: a lot more should be done toward this goal in our view – we are not certain how success is being measured, but this is one of the biggest problems in the field in our view.)

Dr. Henry Anhalt: I don’t think there’ll be one solution for every patient. I’m reluctant to make assumptions around venture capital – I think there are plenty of examples of funding that didn’t make any sense, some resulted in successful outcomes, and some resulted in flushing money down toilet. I would be careful to say that going for a fully implanted system would result in a failure. Having JDRF and Helmsley behind these systems does make an impact. Development should not be stepwise, but rather occurring at the same time. I don’t know what that looks like, I just don’t want to make assumptions about venture capital.

Prof. Eric Renard: We can discuss improving step by step, but we need a global system. One system with a sensor, a pump, and insulin. Of course it’s a big amount of money but you won’t achieve the goal if you don’t tackle it together. If Al Mann didn’t give a large amount of money for the implantable pump, we wouldn’t have one. We’re now against the wall. We need a system for most type 1s. Of course we can work on [Roche’s] DiaPort, but that’s not for everyone. We can’t skip this problem. We want fully implanted device that targets all patients’ needs.

Dr. Dutta: JDRF has another funding tool, led by Dr. Jonathan Behr, the Type 1 Diabetes Fund. It’s another way we can partner with innovative ideas that lean toward development.

Mr. Paul Strasma (CEO, Capillary Biomedical, Irvine, CA): Development of a subcutaneous system was made possible by collaborating systems. There have been a number of sensors here, but not a pump that can be tested with all of these at this point.

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