DTM 2018 (Diabetes Technology Meeting)

November 8-10, 2018; Rockville, MD; Highlights – Day #1 – Draft

Executive Highlights

  • In insulin delivery, we have fascinating MiniMed 670G user experience data from Mayo Clinic/ASU, showing highly varied experiences and satisfaction with the hybrid closed loop (including high Manual Mode use in some users). Also in AID, Dr. Roman Hovorka (how great to get him at this meeting!) shared thoughts fully closed loop in the hospital, including new data on parenteral nutrition – the CGM graphs were truly compelling, once again! In smart pens, Dr. Jane Seley shared views on the emerging landscape (she seemed most excited about add-on caps to diposable pens as a simple plan for adoption), and we heard a high-level talk from Lilly on why connected pens will be useful (it has completed a meal bolusing study, which we’ve previously noted). Finally, Drs. Roy Beck and John Pickup debated MDI vs. pump in type 1 diabetes, mostly agreeing but taking different tacks on each side.

  • In CGM, adhesive juggernaut 3M (of Post-it notes and medical tape) discussed the feasibility of >14-day wear, which remains “elusive” at this stage given skin challenges and adhesive technology. We also enjoyed a round-up of the implantable CGM landscape, including Glysens’ plans to run Gen 3 in trials in 2019, a Theranova’s study comparing IP vs. SC sensing, HCT funding for the first Biorasis human implant, PhysioLogic Devices IP-IP System in Canines (implanted in March 2018), and more. In a talk on CGM in the hospital Dr. Curtis Cook shared that telemetry is the “holy grail,” allowing nurses to remotely monitor glucose/alarms from a patient’s bedside. We also heard updated plans from EyeSense, who claims to have a 29-day, 1 cal/day needle CGM (EU pivotal 2H19).

  • In BGM, we were impressed with Ascensia’s poster sharing real-world Contour app use over 180 days in n=5,870 users – notably, doubled blood glucose checking frequency along with much lower odds of hypo/hyper. Pops also provided an update on its smartphone-attachment BGM, with plans to secure FDA clearance by end of year.

  • We also enclose a concerning talk on treatment persistence in type 2 diabetes (injections are higher than orals, oddly, but not by much), a ranking of mobile health apps by evidence, and news from FDA that the YSI 2300 will be discontinued.

Greetings from the Bethesda North Marriott and DTM 2018! Today was a pre-conference day before the main show starts tomorrow – see our highlights below and read our preview here for what to expect tomorrow and Saturday.

Table of Contents 

Day #1 Highlights

1. Medtronic 670G User Experience Study from Mayo Clinic/ASU – “Varied” and “Contradictory” Experiences; Lots of Manual Mode Use Often/Always

An independent Mayo Clinic/ASU study shared fascinating user experience data from a questionnaire given to n=21 MiniMed 670G users recruited from an outpatient endocrinology clinic (mean age: 44 years; A1c: 6.7%; 5 months of experience on 670G). Responses were quite varied across most questions. On the plus side, two-thirds of users expressed “a high level of satisfaction” with the 670G. However, satisfaction was highly mixed on Auto Mode itself: 52% of respondents were satisfied or very satisfied; 24% were neutral (neither satisfied nor dissatisfied); and 24% were dissatisfied or very dissatisfied. Notably, 62% of respondents said they used manual mode “often” or “always” (i.e., not in closed loop); a smaller 24% used manual mode “rarely”; and only 14% “never” used manual mode (i.e., they were always in closed loop). The least liked features of the 670G were physical design or structural issues (24%); perceived need for increased user input (24%); and high alert frequency (19%). Equally interesting, 43% of respondents reported “better blood glucose control” and 29% liked 670G’s “ease of use”, though many verbatims noted the opposite reactions – that 670G was too conservative or was difficult to use. A notable 14% of users had nothing to say about “most liked” features of Auto Mode, and another 14% had not used it. These data from a small study (in low-A1c patients) certainly illustrate the mixed real-world experiences with the hybrid closed loop – some love it, some find it conservative or demanding in terms of time required. We’d love to see these questionnaires administered in a consistent, structured way across devices – comparisons on user experience would be highly valuable. See key tables below.

2. 3M on Adhesives: Going Beyond 14-Day Wear is “Elusive”

3M’s Dr. Del Lawson provided the rare perspective of an adhesive manufacturer, sharing insights from recent clinical trials and R&D. Most importantly for CGM companies, he noted that >14-day wear with reliable function remains an “elusive” goal, given what is known about adhesives right now. This comment in Q&A certain drew a reaction from the audience: “At 21 days, when you smell what comes off, it is not a healthy situation. If you’re saying, ‘I want to go for 30 days adhesive,’ think again. Please come talk to us, as there are some things we have to fix.” Dr. Lawson spent much of his talk on challenges, with the skin, though there were some interesting technical tidbits too: (i) chest and abdomen are the best wear locations for extended wear; “We have a lot of problems when we do arm studies – door knocks, clothing, etc.”; (ii) having an adhesive “skirt” is very important for extended wear, and a larger skirt is optimal for (e.g., the Dexcom G6 has a large adhesive skirt, meaning the adhesive size is larger than the on-body wearable vs. FreeStyle Libre has a smaller adhesive skirt); and (iii) silicone-based adhesives are less irritating, but don’t stick as well to the skin for extended wear (e.g., Senseonics Eversense). This talk reminded us that improving adhesives is not as simple as making them “less irritating,” since changing the materials can result in meaningful product tradeoffs (e.g., less sticky). Further, hearing from 3M – maker of Post-its and Scotch tape – that 21-day wear is difficult makes us think 14-day wear CGMs may be here for a while – unless companies get an indication that includes over-tape. Adam has used Simpatch over-tape for a long time and he characterizes it as “outstanding”; on Amazon, it is now available for Dexcom, Abbott, and Medtronic sensors, with great reviews.

3. Implantable CGM and Pumps Galore: GlySens Gen 3 in Trials in 2019, Theranova to Launch IP vs. SC Sensing Study with JDRF/Montpellier, HCT Funds First Biorasis Human Implant, PhysioLogic Devices IP-IP System in Canines in March 2018

Glysens, Theranova, Biorasis, PhysioLogic Devices, LifeCare, and Dr. Mark Arnold (University of Iowa) all presented updates on their implantable CGM and/0r IP insulin technology. Many of these companies participated in a similar panel at DTM 2016, and it was great to hear that many have made progress on a number of fronts, including technological (miniaturization!), funding, and clinical work. Still, they are all rather early stage and none shared regulatory submission/launch timing – by the time they are more market-ready, what will the needle sensor landscape look like? Wear duration will certainly fall in favor of implantable devices, but will that balance the always-improving user-experience of subcutaneous CGMs (bandage-like wearables?) and apps? And by how much can extended wear implantables drive down cost? Notably absent from the updates session were Capillary Biomedical and Profusa, both of which presented at DTM 2016.

  • GlySens is now on its gen 3 implantable Eclipse 3 ICGM system, with plans to enter clinical trials in 2019. [Note: The “I” in “ICGM” likely means “implantable” rather than the new FDA “iCGM” pathway.] CEO Mr. Bill Markle overviewed the gen 3 system’s features: Fully-implanted (no external components, unlike Senseonics); designed for two-year life; Bluetooth-connectivity and no dedicated receiver (straight to iOS app, with high-fidelity connection up to 6-8 feet); aiming for 1-2 fingersticks/month, though the company may move to 1x/week in the interest of driving MARD down; 40% smaller than the gen 2 device; onboard glucose calculation and data storage. The outpatient/in-office procedure is 15-20 minutes has been performed using just local anesthesia. Mr. Markle showed pictures depicting a barely-visible 1.25-inch incision, along with clean sensors at explant (suggesting minimal or no foreign body response at that time). GlySens’ path has been rife with delays, and the first two generation sensors were never deployed on the market or hit their pivotal/submission milestones. Direct-to-smartphone – so long as it’s reliable – and the smaller form factor are major improvements in gen 3, but it seems like accuracy and possibly lag time and signal stability remain the biggest areas for improvement. Mr. Markle notably didn’t display any data on these fronts, which we found to be a notable omission, given that human trials with the gen 3 system are slated to begin next year.

    • GlySens will shortly announce the completion of a ~$10 million internally-led round, which should carry the company deep into 2020. We suspect that most of the funds will be allotted toward clinical investigation, though it sounds like more work may be needed on the algorithm side as well. We look forward to the readout of the first trial in 2019 – a two-year fully-implanted sensor certainly has its appeal, and due to the potential wear time (4x longer than Eversense XL), we suspect that per-day pricing could come in at a fraction of current CGM.

  • Theranova is set to embark on a “first-of-its-kind” study in collaboration with JDRF and Prof. Eric Renard (Montpellier) investigating intraperitoneal (IP) vs. subcutaneous glucose sensing while delivering insulin intraperitoneally. This is a step on the way to development of a fully-implanted IP automated insulin delivery system with Bluetooth connectivity to allow (at least) monitoring from an app. After reviewing the literature supporting IP insulin delivery (e.g., 7-8x faster kinetics, lower risk of hypo, no need for meal announcements in closed loop), Theranova’s Mr. Chris Hanson matter-of-factly stated: “We figured that if we’re in the peritoneal cavity anyway for insulin delivery, why not sense there as well?” He pointed to literature from both swine and humans demonstrating that glucose-sensing lag time is significantly less-pronounced in the IP vs. subcutaneous space and that the company’s IP sensor can function lag-free for 90 days. As Theranova remains in “stealth mode,” Mr. Hanson didn’t show pictures or specifically describe the company’s implantable CGMs or pumps. As of April 2017, the company had three different sensor mechanisms in development: (i) A glycoenzymatic sensor in the preclinical and clinical stages; (ii) a fluorescent sensor in the preclinical stage; and (iii) an optical sensor in the preclinical and clinical stages. We’ve never heard any details on the company’s implantable pump.


  • From a cost perspective, a surgery-requiring implantable IP seems like it would be more expensive than subcutaneous sensor and insulin infusion combo upfront, but Prof. Eric Renard pointed out in Q&A that regular human insulin does the trick in the IP space. “Regular insulin in the peritoneal cavity is at least as quick as fast-acting insulin in the subcutaneous space. You don’t need fast-acting. Analogs were designed for quick subcutaneous absorption, but you don’t need this for IP. Also, the distribution of insulin is different – you can increase the insulin infusion without inducing hypoglycemia.” Due to this lower hypoglycemia risk, Mr. Hanson asserted that IP closed-loop algorithms might not have to be as complicated in design as those for subcutaneous. This was interesting to hear.

  • At this point, there are no insulins approved in the US for IP delivery, while Sanofi’s Insuman is approved in Europe. Even if it should be approved in the US, Sanofi’s Dr. Eric Petreto pointed to a couple of barriers in IP insulin at a workshop last year: (i) 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.” (ii) The current formulation of Insuman is only stable for 45 days in a pump. Both Thermalin (preclinical) and Arecor (preclinical) have U1000 insulins in development. 

  • Biorasis announced that the Helmsley Charitable Trust will fund the first human transplantation of its long-term, injectable/implantable CGM, “Glucowizzard.” Notably, Glucowizzard is expected to be the sensing element for PhysioLogic Device’s IP-IP closed loop system (see updates below). Like GlySens, Biorasis has shrunk its sensor significantly in the past two years – as it stands, the CGM is 0.75 x 6 mm, small enough to be injected through and harvested back using a needle. The vision is to inject the rice-grain-sized sensor on the top side of the wrist such that a watch can power it with light and receive glucose measurements sent from an LED light source through the skin (Dr. Fotios Papadimitrakopoulos said the team has developed its own watch, though a consumer smartwatch could theoretically be used with a customized underside attachment). Glucowizzard has thus far been tested in rats, mini-pigs, rabbits, and dogs, performing with: (i) high stability out to 55 days – as of 2016, the company wanted to extend wear time to six months, so it has a ways to go; (ii) 9.8% error (MARD, we assume) over 14 days (100% in Consensus Error Grid zones A+B); and (iii) a lag time of 5-10 minutes. Dr. Papadimitrakopoulos devoted a solid chunk of the talk to efforts to combat the foreign body response to extend wear. Localized, low-dose dexamethasone release from “microspheres” – similar to the approach taken by Senseonics – has suppressed inflammation and fibrosis for two months in rabbits and six months in rats in studies. At this stage, the single-day does of dexamethasone is 75x smaller than the minimum therapeutic dose – safe by any measure. At this stage, Biorasis is not approaching Senseonics’ accuracy and wear time, though the wrist form-factor to power/receive data is attractive, theoretically.

  • PhysioLogic Devices was not on the panel, but did have a poster showing that the pre-clinical ThinPump IP-IP closed loop system was implanted in canines in March 2018. The feasibility trial used a Harvard algorithm and was conducted in collaboration with Vanderbilt’s Dr. Justin Gregory and UCSB/Verily’s Dr. Howard Zisser. In the canine model, IP glucose sensing lag time vs. YSI was just 3 minutes at baseline; at 50 days, lag was six minutes with dexamethasone and nine minutes without dexamethasone, suggesting the corticosteroid coating attenuates the foreign body response. As a reminder, ThinPump has a fully automatic closed loop mode, requires no skin-mounted and is expected to have a 10+ year implant life (requiring refills every three to six months). US and EU clinical trials are slated for 2021, as of the last update in April 2017.

    • PhysioLogic Devices is currently seeking $3 million in a convertible note round. We also overheard in hallway chatter that the company is working with Thermalin on U1000 insulin – we first heard at the NIH AP Workshop in 2016 that the company had a U1000 monomeric analog in preclinical development for miniaturized artificial pancreas systems.

  • Dr. Andreas Pfützner provided an update on LifeCare’s osmotic sensor, SenCell, which is set to enter human pilots in 2H19. Within the miniature sensor (2 x 1 x 0.5 cm), there are two compartments separated by a semi-permeable membrane: In one, there is dextran bound to ConA (a protein, which can be toxic and will therefore be replaced by a different agent). When glucose enters the compartment, it displaces the dextran, increasing the amount of free sugar molecules in the compartment and thereby increasing osmotic pressure. Because the physics are not chemistry-consuming, the sensor could theoretically last indefinitely. SenCell has undergone preclinical proof-of-concept studies with a small working sensor model. In a portion of the talk that elicited plentiful “oohs” and “ahs,” Dr. Pfützner said that LifeCare has been working with CantiMED, which 3-D prints nano-pressure sensors. CantiMED uses an “electron print microscope,” which lets gas flow into a chamber, and where an electron beam makes contact with the gas molecules, the matter solidifies – using this method, one can print with great precision on any substrate down to a resolution of 10 nm! We found the nano-sensing method to be pretty neat: At baseline osmotic pressure, electrons hop from the anode to the cathode of the sensor, resulting in a certain resistance. At higher osmotic pressure (due to higher glucose concentration), the sensing element is now strained, so there’s a change in resistivity, which can then be quantified and correlated to the glucose signal. LifeCare has shown that the force on the membrane (which is directly related to glucose concentration) has a linear relationship to electric signal. The goal is to have each miniaturized implant carry ~16 nanosensors to lend redundancy. Dr. Pfützner didn’t give any timing on the product, but shared that 3-D printing of batches en masse could “decrease the cost of production to very, very, very low amounts. We’re glad to see a very different sensing approach with low-cost potential!

  • University of Iowa’s Dr. Mark Arnold discussed his group’s recent work using near-infrared spectroscopy to detect glucose in a diabetes setting and urea in a hemodialysis setting. Dr. Arnold’s work with glucose-sensing is compelling, but we were perhaps most intrigued by the dialysis work. His lab is using the optical sensor to optimize dialysis “dosage” (time) by continuously tracking urea concentration in the dialysis output – once the concentration drops below a threshold, clinicians can be assured that the dose is sufficient. Theoretically, this tool could help cut down on time, cost, and maximize turnover in dialysis centers – a huge quality of life burden and high-cost for the system. Dr. Arnold said that he ultimately hopes to get the sensor down to the size of a photonic chip (which comfortably fits on a fingertip) and a multidisciplinary team is now needed “to take it farther to get to the point where we can put it in people.”

4. Dr. Jane Jeffrie Seley: Smart Pen Caps “Easier in Some Ways” than Smart Pens (More Freedom of Choice of Insulin); Reusable Smart Pens with Manual-Fill Cartridges Coming Down the Pike?

New York Presbyterian’s Dr. Jane Jeffrie Seley highlighted a selection of players in the smart insulin pen/pen cap landscape, punctuated by her enthusiasm for the growing market: “I really welcome this into my practice – some people think too much data is a curse, but … it’s a blessing for us.” She noted that caps are “easier in some ways” than durable pens since they might give patients more freedom of choice with respect to insulins and continued use of disposable pens. She foreshadowed that, in the future, some reusable pens will allow users to manually fill cartridges with any insulin of their choice (with a syringe, like a pump reservoir); she didn’t elaborate, but we imagine patients would be given a connected pen and an empty cartridge and then prescribed vials of insulin, which may be lower hassle than prescribing pre-filled cartridges. For smart pens, Dr. Seley prefers disposable pen add-ons, which are easier to learn to use since they don’t require the patient to re-load the insulin cartridge, though we’d note that they still require the user to transfer the cap from disposable pen to pen and to recharge the cap. The beauty, as Dr. Seley pointed out, is there is a bounty of devices in development, which will allow patients (and payers) to choose from a suite of options. Many providers – particularly educators – are understandably eager to have granular data on insulin injection history, but Dr. Seley emphasized that a number of pieces have to fall into place in order for a fast ramp among consumers. First and foremost, cost and reimbursement – one audience member put a finer point here during Q&A, claiming that large numbers of her type 2 patients decline to use apps as soon as they find out doing so would boost the cost of their data plans (this is important and not often talked about!). Dr. Seley believes cost and reimbursement will improve over time, as outcomes and cost-savings are demonstrated and pharma players transition their bases over to connected devices. In addition to access, Dr. Seley wondered how much training devices will require – how intuitive will it be to use them, and how much provider time will it take to get patients up and running? Lastly, the patient has to perceive that the device has added value to their routines (or at least not adding burden) – considering most dose capture devices are integrated into the injection device and passively upload dose, there is not too much friction associated with their use; however, this needs to be shown in a wide variety of patients, rather than current early adopters. Naturally, adding in features that patients regard as value-adds – e.g., bolus calculators and titration – will make them stickier. 

  • Dr. Seley spoke briefly about six companies manufacturing smart pens or caps:

    • Companion Medical InPen. According to Dr. Seley, the Bluetooth-enabled InPen offers valuable features such as an insulin bolus calculator that incorporates insulin on board (IOB), dose reminders, and temperature alerts. In her view, the pen’s one-year battery life is a limitation. New to us, though the pen’s different color options currently serve to provide options, in the future, different colors may correlate to different insulins (i.e., basal or bolus). InPen currently only supports prandial insulin cartridges (Novolog and Humalog), to our knowledge.

    • Common Sensing Gocap. The Bluetooth-enabled dose capture pen cap (Gocap) remains in limited supply, direct from the company currently, though Dr. Seley imagines it will scale soon. She noted that the Lantus, Apidra, and Novolog compatible caps are now available, and positioned the 510(k)-exempt status – Gocap doesn’t have a bolus calculator – as both an advantage (no FDA hurdles) and a detriment (less value to the user).

    • Emperra Esysta pen. Currently only available in the EU.

    • DiabNext Clipsulin. The pen attachment is said be launching globally this month, as of EASD, though timing with this company is always a moving target. Dr. Seley highlighted that the device works with almost every insulin pen available, logs in one-unit increments only (different attachment needed for two-unit or half-unit dosing in the future?), and stores 200 doses on the device itself and all data in the app.

    • Lilly and Novo Nordisk. “Both companies are working on connected health, sharing as much data as possible. This is what we want: everything in one place, we can look at one chart, and have all the data that we need there. One way to download, one way to look – that’s how we’ll integrate all this into our practice.”

5. Ascensia Contour App data: n=5,870 users, doubled blood glucose checking frequency at 180 days, lower odds of highs/lows

An Ascensia poster presented real-world data from n=5,870 Contour Diabetes app/connected Contour Next One BGM users, comparing app data from use in the first 30 days to >180 days of use. Impressively, blood glucose checking frequency more than doubled from 2.0 times daily in the first 30 days to 4.5 times daily after 180 days (p<0.001) – this sort of increase in engagement is rarely seen in digital health studies this long and is an excellent sign for Ascensia. Hypoglycemia (<70 mg/dl) and hyperglycemia (>180 mg/dl) were also twice as likely in the first 30 days relative to after 180 days of use. Data looked notably similar in other analyses of cutoffs (<50, >250 mg/dl) and multiple events of hypo/hyperglycemia – there was 40%+ more risk of events in the first 30 days vs. >180 days. It’s excellent to see some real-world data from Ascensia, indicating that the app’s pattern recognition and suggestions are engaging BGM users and driving nice glycemic impact.

6. EyeSense FiberSense 29-Day Wear, 1 Cal/Day Needle CGM Demonstrates 14.7% MARD Vs. YSI; EU Pivotal Trial Slated for 2H19, Potential Launch in Late 2020

A single-center, prospective study (n=18) investigated the performance and safety of FiberSense, a real-time, 29-day wear CGM with a percutaneous fiber optic glucose sensor equipped with ConA-based chemistry. Participants wore the FiberSense either on their abdomen (n=9) or upper arm (n=9) for up to 29 days and also wore the Dexcom G4 as a comparator for one week of the study. Accuracy was comparable between the two systems: FiberSense MARD vs. YSI with one fingerstick/day was 14.7% as compared to 15.7% for the G4. %20/20 analyses were also comparable (73.5% vs. 73.2%). 99.5% of FiberSense glucose readings were within Zones A and B of the Consensus Error grid, with 73.9% in Zone A. There were no serious device-related adverse events, and no severe sensor site reactions were observed. Based on our own experiences with CGM and the 3M adhesive talk today (see above), we are surprised that 29-day wear was seen in this study and wonder how replicable that would be in the real world without significant over-tape.  

  • An EU pivotal trial is expected to initiate in 2H19 (the poster also referenced plans for pivotal trials in the US and China) with a CE Mark expected in 2020 and a potential launch in late 2020. The company expects to launch the 29-day sensor with Bluetooth direct to the phone and at a price point potentially below that of Abbott’s FreeStyle Libre – presumably that all hinges on the indicated wear time. The device has a warmup of a “few hours” and can pair with either a smartphone app or a dedicated receiver. It is clearly early stages for FiberSense, though we’re always interested to see new players pushing to develop innovative, longer-lasting sensors.

7. CGM In The Hospital: Telemetry to Central Nursing Station is the “Holy Grail”; Large, Outcomes-Focused Studies Needed

Mayo Clinic’s Dr. Curtis Cook covered CGM in the hospital, concluding (as expected) that large, multi-center, randomized controlled trials are needed to look at outcomeshow does using CGM impact mortality, length of stay, hyperglycemic/hypoglycemic episodes, and cost? Yes! – as we have said for years, we’d love to see something like NICE SUGAR repeated with modern-day CGM, tested in thousands of patients, using connectivity, and funded by large foundations or NIH. (Another possibility is to add automated insulin delivery, as Dr. Roman Hovorka has done in smaller studies; see below.) To date, most CGM studies in the hospital have looked at accuracy and been limited by small sample size and single-center location. Dr. Cook noted some of the challenges to implementing CGM for hospital blood sugar management, including staff training, patient selection, no FDA approval, IT challenges, cost, and safety. Still, he believes that “Glucometry” (glucose telemetry) is the “holy grail” – continuous measurement and transmission of CGM data from bedside to a centralized station that could be monitored by nurses/trained professionals (just like cardiac monitoring). “As an endocrinologist, I would love to see that, especially because glucometry can provide healthcare providers with more glucose data points, trend, and rate of change. He noted the pilot “Glucose Telemetry System” study published in JDST (mentioned at ENDO Fellows), which tested a Dexcom G4 and central nursing station monitoring system in n=5 type 2s in the general hospital ward. CGM alarms were set up for glucose readings <85 mg/dl. Over the four days of CGM observation, patients spent 65% of the time in-range (“70-179 mg/dl”) and only 0.3% of the time <70 mg/dl. No patients had a CGM glucose value <54 mg/dl. A larger Glucose Telemetry System study at the VA (see our May coverage) is now recruiting. For more on this topic, see Adam’s article about his recent hospital stay and scary lessons learned – including the desperate need for CGM.

8. Injection Therapies Show Higher Persistence than Orals (though both are low); >50% of Type 2s Don’t Receive A1c Test Within One Year of Initiating Treatment

McKinsey’s Ms. Nisha Subramanian showed data demonstrating significant variation in adherence across therapy type in diabetes. She noted a “huge, steep drop off” at the 2-3-month mark across multiple disease areas, not just diabetes, hypothesizing that in diabetes the trend is likely due to an initial influx of multiple add-on therapies and difficulties surrounding expectation setting. (We’d also point to “long-term” motivators to “avoid complications” as not very effective or engaging.) Notably, she found that injection therapies have a higher persistence rate than orals (except for SGLT-2s over GLP-1s), possibly because injectables require a more rigorous schedule that patients find easier to follow. It’s also possible that an injectable “feels” like a more important medication, is deployed in people with more advanced disease who approach their care with greater urgency, or that those medications (especially insulin) have a more noticeable effect on how the individual feels day-to-day. The steepest adherence drop-off past six months tends to be within orals and non-insulin medication. For non-insulin therapies, SGLT-2s show slightly higher persistence over GLP-1s and DPP-4s through 10 months. On average, ~40-50% of patients stop taking a given therapy within the first year, which is the real headline stat of this stalk and a very concerning one.

  • Ms. Subramanian shared sobering and surprising statistics: 57% of patients with type 2 diabetes do not receive an A1c test within a year of initiating treatment; and of the patients who do receive an A1c test and are found to be not at goal, a whopping ~70% do not experience therapy escalation. This is shameful and wrong. Moreover, many patients initiate with non-standard-of-care treatments (e.g., ~13% type 2 patients start sulfonylureas as their first line therapy). As we heard more than one attendee proclaim today, we need to go back to basics of education and emotional support! And optimized therapy and not therapy that would not be FDA approved today.

  • Earlier in the day, we received some fast facts from Dr. Deborah Greenwood on adherence:

    • 25%-40% of patients do not fill their primary prescription.

    • 45% of patients self-report not taking their medication and not persisting at six months.

    • 18%-26% discontinue insulin therapy within 12 months.

    • HCPs describe only 20%-30% of patients as “very successful” with insulin therapy.

    • 38% of patients self-report missed, mistimed, or reduced doses in the last 30 days.

9. Lilly’s Connected Pen Study on Missed Meal Boluses Complete; Even “Passive Users” Can Benefit from “Really Rich and Useful” Connected Pen Data

In her discussion surrounding the potential for connected pens to boost patient adherence and improve diabetes outcomes, Lilly’s Ms. Jennal Johnson noted that the company’s 12-week, single-arm smart pen study investigating the frequency and effect of missed meal boluses has been completed (as of July). So far, this study is one of the very few clinical trials that have been conducted in the area – Ms. Johnson also referenced the impressive (our opinion, not hers) Joslin study evaluating use of Common Sensing’s Gocap paired with Dexcom’s G4 CGM data (n=31), as well as the ongoing, 24-week crossover Emory study of Insulclock in type 2 diabetes (n=100), anticipated to complete by the end of 2018. We’d also note a very interesting Stanford/UVA/Mt. Sinai poster presented at ADA, showing that 27% of meals had either a late or missed meal bolus in adults and adolescents (n=24) using smart pens. Ms. Johnson listed a few pilots, including Innovation Health’s collaboration with Sanofi to leverage One Drop’s mobile app and Common Sensing’s Gocap, and Novo Nordisk’s NFC-enabled NovoPen 5 Plus pilot in Sweden. She explained that connected pens not only provide richer insulin dose, timing, and temperature data, but if applied appropriately, can yield actionable insights for providers and people with diabetes. She proposed several add-ons for connected pens, advocating for the passive collection of context (e.g., nutrition, exercise, etc.), alerts/reminders (e.g., Companion Medical’s basal insulin reminders), and the integration of CGM to identify missed mealtime boluses in real time. Still, she cautioned industry members in the room to “keep it simple” especially when it comes to patient-facing data visualization. Highlighting the value of pairing previously invisible insulin data with CGM traces, Ms. Johnson urged the audience to remember that the ultimate goal of avoiding hypoglycemia requires more than patients taking an insulin dose; rather, patients must take the right dose at the right time. We were disappointed not to hear a more substantive update on Lilly’s connected pen efforts, which have been notably quiet since the announcement one year ago – we have yet to see what the device looks like that fits on the disposable KwikPen platform. See Lilly’s Blogger Summit from this past May for more details.

  • During Q&A, Ms. Johnson explained the distinction between active and passive users of technology, asserting that even passive users can benefit from connected pens. While active users might use “every bit and part” of a given device, a passive user “sets it on auto pilot and might use one function.” However, Ms. Johnson said, the “data is still coming in and is going to be really rich and useful” in informing healthcare providers in their discussions with patients. This is the beauty of passive dose capture: Unlike other innovative technologies that fundamentally change diabetes management for the patient, once a smart pen/pen cap is paired with a phone, there is little additional lifestyle change foisted upon the patient (unless he/she wants it).

10. Cambridge’s Dr. Roman Hovorka on Fully Closed Loop in the Hospital – Very Powerful Data, New n=40 study

Cambridge’s Dr. Roman Hovorka summarized the team’s stunning data on use of fully automated insulin delivery in the hospital general ward, including a new study just completed in patients on parenteral/enteral nutrition (n=40). The glucose results were again incredible in patients on up to 15 days of full closed loop (subcutaneous CGM/pump), summarized in the first modal day plot below. He also reviewed the team’s powerful closed loop study in inpatient type 2s presented at ADA (simultaneously published in NEJM!), as well as the EASD poster showing a profound ~9.3-hour/day improvement in time-in-range in the sub-population of hemodialysis patients (n=19) – that’s over a 30-percentage point improvement in time-in-range, the largest the team has even seen. Dr. Hovorka emphasized that the Cambridge system is fully automated (no meal announcement), highly adaptive (very in the hospital, given insulin sensitivity changes), and has seen impressive patient satisfaction. On the latter, 100% of patients in the NEJM study (62/62) would recommend it to a friend/family member entering the hospital. Beyond the general ward, Dr Hovorka also believes there is a role for closed loop in the critically ill, leveraging IV insulin/dextrose infusion and subcutaneous CGM. The team published a 2013 study to this effect (Leelarathna et al., Critical Care), again noting “enormous” difference in time-in-range with closed loop. We cannot wait to see larger studies of closed loop in the hospital in both non-critical care and critical care – the outcomes and cost savings should be a slam dunk, assuming a study is large enough to power outcomes like length of stay, morbidity/mortality, and readmissions.

  • Cambridge’s hospital algorithm targets 100-180 mg/dl (5.6-10 mmol/l) and often delivers 10-20 units per hour or more. The highest dose was 70 units per hour, indicating just how much insulin resistance is seen in the hospital! Algorithms in the hospital must be highly adaptive, since changes in insulin sensitivity can be significant and frequent.

  • Cambridge is currently using the FreeStyle Navigator 2 CGM (no acetaminophen interference), a subcutaneous insulin pump, a CGM receiver, and a tablet running the control algorithm. We don’t believe FreeStyle Navigator 2 is still available very widely commercially (if at all), and we wonder if Cambridge will move to a different CGM for future studies.

11. Pump vs. MDI in T1D – Drs. Roy Beck and John Pickup Mostly Agree

In the now-common debate on pumps vs. MDI in type 1, Drs. Roy Beck and John Pickup agreed on the big stuff – both technologies are valuable – though discussed different nuances. As we usually see with thoughtful diabetes debate participants, this one didn’t deliver an “either/or” answer, but a “both/and.” See the key points below. To us, the biggest questions in this area are about cadence, connectivity, and automation: (i) when should a type 1 receive a CGM? (i.e., at diagnosis?); (ii) when should a pump be added on top of CGM, and in whom?; and (iii) how will smart pens/CGM compare to pump/CGM? How will automation in both areas change outcomes, quality of life, and treatment burden?

  • CGM – Dr. Beck highlighted the oft-shown T1D Exchange data on A1c by device category (pump vs. MDI, CGM vs. no CGM): while pumpers have a lower A1c than MDI users, in those using CGM, there is no significant A1c difference between pumps and MDI. Looking to clinical trial data, Dr. Beck highlighted the same A1c reductions with MDI and pumps in the JDRF CGM trial, as well as the 2017 JAMA publications from DIAMOND and GOLD (both testing CGM in MDI and showing benefit). Concluding, Dr. Beck noted that “without CGM, a pump wins” – it is associated with lower A1c and a higher percentage meeting the ADA target relative to MDI alone. Once CGM is added, however, “it is a tie” between pump and MDI – mean A1c and percent meeting ADA targets are similar for pump and MDI users, supported by both registry and clinical trial data. In his last slide, he answered what we believe is a better framing of the debate: if you could only have a pump or CGM, which would you choose? Dr. Beck chose CGM, showing pictures of G6, FreeStyle Libre, Eversense, and Guardian Connect.

  • Pump – Dr Pickup started in his usual humorous tone, “I know what some of you are thinking, ‘Pumps are so last year, darling.’” He agreed with Dr. Beck that MDI users can get excellent diabetes outcomes – e.g., DCCT participants on pumps had an A1c of 6.6% vs. 7% in MDI users (“Not much worse” – but that was technology build a long time ago!). However, such outcomes on MDI are not routinely seen in the average patient, the average clinical trial, and at the average clinic. Dr. Pickup argued persuasively that pumps have been around for over 40 years now, meaning the evidence and knowledge base is much greater than with CGM. We aren’t sure it is a fair comparison – one is delivering a drug, after all, while one is a diagnostic. We also know a lot about pumps that is “still uncertain” about MDI/CGM: long-term glycemic outcomes on pumps are well established; long-term treatment satisfaction and quality of life is generally good on pumps; the discontinuation rate of pump therapy over the long term is low (<5%); and pumps are effective in many patients struggling on MDI (e.g., disabling hypoglycemia and/or elevated A1c, even after best structured education). (Dr. Pickup used the term “MDI failures,” which is not language we see as patient-friendly or provider friendly.) Dr. Pickup noted several important benefits of pumps, including “special pharmacology of pump delivery” (constant basal, improved predictability, accurate adjustment); alleviating burdensome multiple daily injections (especially for meals, where bolus injections are often missed and raise A1c); covering high-fat, high-protein meals with extended boluses; and coping with frequent, unplanned exercise. Taking a page out of Dr Beck’s book, he even cited the DIAMOND extension phase, where MDI/CGM users crossed over to the Omnipod – time-in-range was 83 minutes/day higher in the pump/CGM group vs. MDI/CGM. Concluding, he noted, “Insulin pumps are still a jolly good choice for treatment in type 1 diabetes.” It’s funny to hear “still” – we see it getting so much better as closed loop systems near.

    • Dr. Pickup cited “more than 1 million people” using pumps worldwide – a sign of the technology’s uptake – though of course this now also applies to CGM with Abbott’s 1+ million FreeStyle Libre users, likely 350,000+ Dexcom CGM users, and likely over 125,000 Medtronic CGM users. Dr. Pickup estimated 5% of type 1s globally are using pumps, while only 1-2% are on CGM; that math doesn’t quite work out, unless he is not counting FreeStyle Libre as a CGM – lots of the CGM numbers are, of course, type 2, which is not the case as much for pumps.

12. Top 10 Mobile Apps by Amount of Evidence: Dexcom G5, Senseonics’ Eversense, and WellDoc’s BlueStar Take the Lead

Children’s Mercy Hospital’s Dr. Mark Clements reviewed the top 10 mobile apps by total number of registered trials, journal articles, and meeting items (i.e., abstracts/posters). We were somewhat surprised to see Senseonics’ Eversense coming in so high on the list (#2!), beating out WellDoc’s BlueStar and Fitbit’s app, though well behind Dexcom’s G5. This ranking is particularly surprising given that Fitbit is the wearable device provider for nearly 95% of NIH-funded research studies using a wearable device, suggesting that a large proportion of these studies may not be leveraging the Fitbit app or at least calling it out in the published data. Dr. Clements highlighted the G5/G6 mobile app, expecting that direct-to-Watch communication will increase adoption of Dexcom and other CGM systems. He also referenced Companion Medical’s InPen app, noting that it provides a variety of reminders (e.g., insulin temperature, battery life, dosing, insulin age), and One Drop Mobile, remarking that patients can receive push notifications even outside of the app. Rimidi, he noted, is an example of a platform in which the healthcare team can set individualized reminders for patients. Last, he noted Klue, a new Apple Watch app for automatically detecting meal start and eating speed with an automatic bolus reminder, which we just saw demoed at DiabetesMine last week.

13. Pops! Diabetes Care Expects 510(k) Clearance for All-in-One, Smartphone-Adhered BGM; Poster Shows 1% A1c Decline at 6 Months (n=5)

A poster from Pops! Diabetes Care indicates that the novel all-in-one BGM and digital platform is currently pending FDA 510(k) clearance, and we confirmed with the company that the milestone is expected before the end of the calendar year, with a launch soon to follow. The poster itself showed that use of the “Pops! one Mobile Platform” reduced A1c by ~1% (baseline: 8.9%) over six months in n=5 adolescents with type 1 diabetes – the study enrolled 50 individuals into the single-armed, prospective trial, but the remaining 45 participants had not yet completed the trial. As shown in the picture below, the whole testing kit can adhere to the back of a smartphone, where three stationary lancets reside next to three strips on a disposable case module (i.e., each module provides three glucose measurement opportunities; see a YouTube demo here). The meter communicates with a companion mobile app via Bluetooth, and the app also allows caregivers and providers to remotely monitor patients’ glucose readings and other information. The system will be sold to payers through a subscription model, with the goal of patients receiving all supplies at no cost. A rep told us there will be some level of automated coaching in the app (e.g., “You got it!”), but the company doesn’t intend to incorporate human coaching at this stage. Pops! CEO Mr. Lonny Stormo has significant experience in the field, having spent the last 11 of his 30 years at Medtronic in VP roles; despite his diabetes expertise and the novel BGM design, making headway in an already tough BGM market will be no easy feat. That said, we really like the form factor focus on a slim BGM that fits onto the phone i.e, (always with someone), especially for those that don’t want/cannot access CGM.

14. YSI 2300 being phased out, no plans for next-gen 2900 to seek FDA clearance; talk to FDA for acceptable comparator methods

In a talk from FDA’s Dr. Lisa Landree, we learned that the popular YSI 2300 analyzer is “being phased” out by the company, and oddly, there are “no plans” to get FDA clearance for the next-gen YSI 2900. She noted that there are “many” FDA-cleared lab-based methods that can be used for accuracy comparison of BGMs and CGMs; anyone planning to use the YSI 2300 as a comparator in the future should talk to the FDA.


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