DTM 2016 (Diabetes Technology Meeting)

November 10-12, 2016; Bethesda, MD – Day #1 Highlights – Draft

Executive Highlights

Greetings from Bethesda! The 2016 Diabetes Technology Meeting got off to a sprinting start today (pre-conference) with two parallel tracks of rapid-fire 15 minute talks running all day. We’ve picked out our top 10 highlights for the day, six of which came from the fascinating session on implantable continuous glucose sensors. Take a look at our conference preview here for a peek at what’s in store for the next two days.

Top 10 Highlights

1. We saw more detailed data from the 90-day US pivotal trial of Senseonics’ implantable Eversense CGM (PRECISE II; n=90). The sensor performed with an impressive overall MARD of 8.8% vs. YSI in the study (two calibrations per day), with a mean average deviation of 9.6 mg/dl (<80 mg/dl) and a MARD of 8.2% for values ≥80 mg/dl. Based on accuracy metrics alone, Eversense is right up there with the best CGM options today (see picture below).

2. Dexcom’s SVP of Data Annika Jimenez demoed some of the fascinating analyses Dexcom’s Big Data platform can run on both a population and individual level. We’re glad to see big-time investment in Big Data, which should help on the product development, regulatory, and payer fronts.

3. Profusa was the most exciting early-stage implantable sensor startup we saw today. The company’s micro hydrogel sensor is extremely tiny (500 microns in diameter x 5 mm long), flexible, tissue integrating (capillaries can grow into it), lasts over two years, and can sense multiple analytes (oxygen, glucose, lactate, and more). A first-in-human glucose sensing trial is expected to start in early 2017, and its oxygen sensor (Lumee) received a CE Mark in 3Q16. They’ve been at this a long time and it’s great to see progress.

4. OHSU’s Dr. Jessica Castle shared 11 things that patients (and thereby clinicians) want to see in an implantable CGM (insurance coverage was #1, unsurprisingly), a reminder that accuracy is only one piece of the puzzle in glucose sensing.

5. Fenix Group’s Matthew Maryniak provided an investor perspective on implantable CGM, sharing hesitation and general uncertainty in the investment community. An educational gap may exist with investors as to the opportunity in all diabetes technology, let alone implantable CGM. Reimbursement uncertainties have been paramount with the negativity.

6. Glysens CEO Mr. Bill Markle said that the company has a second-gen implantable sensor in development that is “half the size” and “24-month wear” before providing an overview of surveys suggesting patients would opt for the device over alternatives. Earlier, Dr. Joseph Lucisano shared positive interim results from the FIGS-2 human trial of Glysens’ current model.

7. Capillary Biomedical, Biorasis, and LifeCare provided concise updates on their respective implantable CGM projects. Each leverages a different technology to quantify glucose, and with varying form factors: Capillary Biomedical is using pure absorption spectroscopy, Biorasis uses electrochemical sensing, and LifeCare utilizes osmotic pressure.

8. A session on CMS’ Competitive Bidding featured harsh criticism of the program, with focus largely on its deleterious effects and the government’s inadequate safety monitoring.

9. In part one of her double-header, Emerson Consultants’ Ms. Patty Telgener encouraged diabetes technology manufacturers to invest early in clinical outcomes data in order to succeed in a changing healthcare climate.

10. For her next act, Ms. Telgener delivered a seamless talk on the fight for Medicare coverage for CGM. [Note: The scheduled speaker couldn’t make the session, so the amazing Ms. Telgener gave the talk off of slides she had never seen before. Wow!]

Top 10 Highlights

1. Diablo Clinical Research’s Dr. Mark Christiansen presented detailed data from the 90-day US pivotal trial (PRECISE II; n=90) of Senseonics’ implantable Eversense CGM. During the 2Q16 financial update, Senseonics said that the sensor performed with an overall MARD of 8.8% vs. YSI in the study – a marked improvement over the 11.6% in the 180-day EU PRECISE trial, thanks to some algorithmic tweaking. Dr. Christiansen presented more granular accuracy data today: Mean average deviation below 80 mg/dl was 9.6 mg/dl – we’d like to see this broken down even further, given that MARD was 22% in the <70 mg/dl range in the original PRECISE study. In the ≥80 mg/dl range, MARD was an impressive 8.2%. The sensor reported just 7% missed alarms (compare to 17% in PRECISE) and 14% false alarms (compare to 25% in PRECISE) at the hypoglycemic threshold (<70 mg/dl). Performance in the hyperglycemic range (>180 mg/dl) was very strong, with only 4% missed alarms and 6% false alarms. We have previously expressed concerns about the size of the external transmitter, but patients displayed a median wear time of 23.4 hours per day, seemingly taking it off only to charge once every day. This could simply be a manifestation of the “clinical trial effect,” or an indication that the external hardware isn’t too cumbersome (or both). 94% of sensors successfully reported continuous glucose data over the 90-day period (a 6% improvement over the PRECISE data presented at last year’s DTM), and a post-study evaluation suggested that 73% would’ve lasted through 180 days of wear. We wonder if this analysis could supplement the case for a 180-day claim in Europe, which was filed in 2Q16.  From a safety perspective, there were 14 adverse events in seven subjects, pretty low considering ~200,000 hours of total sensor wear time. One of the adverse events was serious, due to a sensor that was implanted too deep. Dr. Christiansen, however, quelled worries about implantation: “I’m an endocrinologist. I don’t like to do stuff with my hands, but I found this device very easy to insert and remove.”

  • Eversense is the first CGM with a MARD below 9%, performing on par with Dexcom’s G5 and with higher accuracy than Libre Pro and Medtronic’s Enlite 3. See the labeled accuracy in the picture below for comparisons, though of course, it’s hard to compare separate studies head-to-head. For what it’s worth, G5 and Eversense are identical in every single accuracy bin, though Eversense bests G5 in overall MARD by a smidgeon 0.2%, and with nearly eight times the number of data points (90 day study). In a retrospective analysis, Senseonics showed overall MARD to be 9.5% when just one calibration was considered per day instead of two, which keeps Eversense well within competitive plans from Medtronic and Dexcom (though behind Abbott’s factory calibrated Libre). On the 3Q16 call, Mr. Goodnow commented that initial in vivo testing of a second-gen sensor, which aims to eliminate fingerstick calibration, is “very encouraging,” and further studies will commence in 1Q17.

  • Senseonics announced FDA PMA submission for Eversense in October, and, assuming a 12-month review (previous guidance), could launch in the US late 2017. Notably, the company plans to talk with FDA about an insulin dosing claim during review, given the low MARD. Eversense is already available through controlled launches in about a dozen clinics in Sweden, Norway, and Germany, with launches in the Netherlands and Italy expected by the end of 2016.

2. Dexcom’s SVP of Data Annika Jimenez demoed some of the fascinating analyses Dexcom’s Big Data platform can run on both a population and individual level. This followed her talk at the DiabetesMine Summit two weeks ago and left many in the room impressed. She showed population-level graphs pulling data from tens of thousands of patients (billions of data points) and comparing a variety of variables: mean glucose by hour of the day over a two-month period in ~30,000 patients; time-in-range vs. CGM sensor utilization (great to show payers: 81-100% CGM utilization brings about two more hours in-range per day vs. 0%-50% and 51%-80%); estimated A1c by age (higher in pediatrics); time-out-of-range by state; average population glucose on holidays; etc. These variables could be selected in a fairly intuitive interface, and data was included right up to August 31, 2016. On an individual level, Dexcom can also use the data to develop individual-level models of glucose, helping clinicians understand risk profiles and giving patients more precise feedback to navigate diabetes each day. Ms. Jimenez affirmed Dexcom’s commitment to building an ecosystem around its data, enabling third parties to build apps that leverage its upcoming APIs. The plan shared two weeks ago remains the same: Dexcom is currently taking developer signups at developer.dexcom.com and will launch its APIs (retrospective data initially) in early 2017.

  • Dexcom has clearly made a big investment here and we’re glad to see it taking best-in-class Big Data technologies from Silicon Valley, with Ms. Jimenez’s leadership. Medtronic has long used its CareLink database for benefit (we believe for regulatory, payer, and algorithm development), and now Dexcom will have such capabilities (or more) too.

3. Profusa was the most exciting early-stage implantable sensor startup we saw today. The company’s micro hydrogel sensor is extremely tiny (500 microns in diameter x 5 mm long), flexible, tissue integrating (capillaries can grow into it), lasts over two years, and can sense multiple analytes (oxygen, glucose, lactate, and more) – watch a video here. A first-in-human glucose sensing trial is expected to start in early 2017, and Profusa’s oxygen sensor (Lumee) received a CE Mark in 3Q16. CEO William McMillan shared plans to secure a CE Mark for glucose sensing in type 2 diabetes and prediabetes in 2018, followed by an expected launch in the top three EU countries later that year. A type 1 diabetes trial is also expected to start in 2018. The tiny sensor is injected under the skin (hypodermic needle) and uses a fluorescence sensing chemistry (calibration details not clear). An on-body patch is worn on the skin that illuminates the sensor, collects data from it, and relays the information to the phone. The company ultimately hopes to have a flexible, thin-film, Bluetooth-enabled disposable patch and/or use a smartphone reader for intermittent sensing (i.e., like Abbott’s LibreLink). The sensor itself is remarkably tiny (see picture below), and the ability to sense multiple analytes gives it a variety of other applications: peripheral artery disease, stroke, tumors, oxygen monitoring during exercise, organ status, microbiome, asthma, chemotherapy, etc. Profusa was incorporated in 2009 (incubated at University of Washington and UCSF), has received $22 million in DARPA and NIH grants (including a “transformative research award” from NIH and lots of military interest), and raised a $13 million B-round of financing in 2015. The team is now 29 people. Though its glucose sensor still has a long way to go, we think this is a company to watch in the implantable CGM area. See pictures below or on the company’s website here. We first met with them years ago – they have been at this a long time and have grown significantly.

4. OHSU’s Dr. Jessica Castle shared 11 things that patients (and thereby clinicians) want to see in an implantable CGM, in her view:

  • Covered by insurance (#1!);
  • Minimally invasive and long wear time;
  • Integrated with existing devices (phones, watches);
  • Excellent accuracy (MARD <10%);
  • Little or no calibration requirements;
  • No interfering substances;
  • No compression artifact;
  • Insulin automation (this is a very big one);
  • Inconspicuous;
  • Safe and comfortable (no rashes); and
  • Accessible data (on the web; with insulin).

Dr. Castle’s wish list highlighted one of the biggest challenges for any CGM company: accuracy is just one piece of making a great sensor. In our view, form factor and cost will drive this field in the future, particularly as it expands beyond early adopters. We were not surprised to see her list insurance coverage first, and to note that more expensive devices must bring superior outcomes. Her second item – a device’s procedure and invasiveness – shared a key insight for implantable CGM companies: “I haven’t held a scalpel in 15 years, and I don’t think I want to start now. Frequency is key. I see patients every 3-6 months. Ideally, we want a device that lasts a year.” It was a good point for more invasive procedures like GlySens, or for shorter-lasting devices like Senseonics’ Eversense – will endocrinologists have the bandwidth to support these technologies? Extra reimbursement “procedure codes” will of course help with some. On accuracy, Dr. Castle advocated for a MARD <10% with no egregious errors (>50% away from reference) and preferably almost all values within 15% of reference. She urged CGM companies to integrate with insulin data (“critical”) and to make data easily accessible on the web.

5. Fenix Group’s Matthew Maryniak provided an investor perspective on implantable CGM, sharing hesitation and uncertainty in the investment community. An educational gap may exist with investors as to the opportunity in all diabetes technology, let alone implantable CGM. In interviews with VCs, Mr. Maryniak found a general reluctance to invest in implantable CGM – the long time horizon for a return on investment and high perceived risks (manufacturing, regulatory, reimbursement) outweigh potential upside in his view. He mentioned that it took Medtronic an estimated eight to nine years before it got a return on the MiniMed acquisition (based on sales, as Medtronic does not disclose profitability), while Senseonics only raised $45 million in its IPO (he also alluded to the fact that this has been around for some time – as a reminder, it was rebranded from Sensors for Medicine and Science in 2011). From a feature perspective, Mr. Maryniak emphasized that implantable CGMs may only address “some” of the new enhancements promised by consistently-improving traditional subcutaneous CGMs: miniaturization, fingerstick replacement, calibration, drug interaction, extended wear (the clear winner for implantable CGM), pump integration, accuracy, reimbursement, and insertion. One investor commented that implantable birth control saw uptake far below expectations, a potential historical analog for implantable CGM. Meanwhile, a former Director at Lilly shared that in “market research on an implantable drug delivery device for T2DM ... consumers overwhelmingly didn’t feel comfortable with having this device under their skin.” (We assume this referred to Intarcia but he did not specify. We note that Lilly has a competitive product to the GLP-1 implantable). Mr. Maryniak noted that Dexcom, Insulet, Tandem, and Senseonics are all “in the range for acquisition"; Dexcom is of course making significant investments and it would be our guess (speculation) that they have no interest in being acquired, given the years building up a very sustainable organization (they are far bigger than the other three and Insulet is meaningfully bigger than the other two). He concluded that “investors want to maximize ROI, and given experience with other implantable devices, they remain weary of implantable CGM. Market penetration of CGM continues to rise, but acceptance of implantable CGM remains uncertain.” While it’s hard from an investor perspective alone to tell whether there will be successes in this arena, we think it’s early to assess whether an implantable CGM could see success, with only ~15% of US type 1s on any CGM at all. In addition to Mr. Maryniak’s insightful point for this audience of mostly non-investors that investor sentiment in implantable CGM has been lukewarm, we note that with the exception of Dexcom, investors have been lukewarm to diabetes devices more broadly for some time, largely due to regulatory and reimbursement concerns.

6. Glysens CEO Mr. Bill Markle said that the company has a second-gen implantable sensor in development that is “half the size” and “24-month wear” before providing an overview of dQ&A and T1D Exchange surveys. He suggested patients might opt for the device over alternatives based on the data. We continue to point out that at ~15% penetration in type 1, there is lots of room for multiple concepts to do very well, as the standard of care changes, assuming CGM is affordable. Earlier, Dr. Joseph Lucisano shared positive interim results from the FIGS-2 human trial of Glysens’ current model. According to Mr. Markle, Glysens is still working on enzyme-loading for the next-gen sensor, but the battery issue (key for shrinking the size) has been worked out. We wonder if the minimized battery could make room for additional enzyme in the chamber to extend duration further. In the dQ&A (n=701 respondents) and T1D Exchange (n=533) online surveys of current, former, and naïve CGM users, Mr. Markle noted that ~90% of the surveyed populations would “likely” or “definitely” use a long-term, implantable CGM, regardless of past CGM experience. A bulk of the remaining 10% said they would not want surgery. As for ongoing studies: The small FIGS-2 trial (n=20 type 1s on insulin) is a 12-month study of the Glysens implantable CGM at two clinical sites. Interim data reported by Dr. Lucisano shows that all procedures have been unremarkable thus far. The implant has been well-tolerated, so much so that all eligible subjects at one site opted to continue to 12 months and six out of eight volunteered for sensor reimplant. Preliminary survey data was also positive. The sensor is performing with less than 1% drift per week, requiring an anticipated 1-2 calibrations per month.

  • In our last meeting with the Glysens team, we learned that bringing the MARD down into the low teens via an improved digital signal processing algorithm is the company’s main priority. If that happens by this fall, the company will proceed to initiate a CE approval trial (n=30) in San Diego, likely in 1Q17. Mr. Markle believes that this study would also contribute to the phase I cohort of the US pivotal trial. We’re glad to see Glysens pushing forward with its fully implantable CGM, but even if things go exactly as planned, the device will still not be market-ready for quite some time (we assume over 1.5 years, since a 1 year study will take time to compete). In the pipeline (for later-gen sensors) is a plan to eliminate the fairly bulky receiver and push data to a smartphone with a relay device (e.g., a watch that receives the RF signal from the sensor and relays it to the phone via Bluetooth). The company also hopes to extend the lifespan to two years, and drive a 40% smaller sensor that will likely have memory capabilities to backfill data (e.g., if the receiver is out of range).

7. Capillary Biomedical, Biorasis, and LifeCare provided concise updates on their respective implantable CGM projects. Each leverages a different technology to quantify glucose, and with varying form factors: Capillary Biomedical is using pure absorption spectroscopy, Biorasis uses electrochemical sensing, and LifeCare utilizes osmotic pressure.

  • Capillary Biomedical’s Dr. Jeffrey Joseph revealed the company is working on an implantable system that derives a blood glucose reading from pure absorption spectroscopy of ultra-filtered blood (plasma). A slight negative pressure forces blood into the mini system, where the blood is then filtered and then analyzed by MEMS optic technology. Because the sensor has no enzyme or fluorescent chemistry, it can theoretically function for “many years,” though it is still early and the company needs to show it can move tissue fluid, extend beyond six months to more than a year of use, and show that readings correlate closely with reference values – preliminary data in animals has been encouraging. Capillary is also developing in the insulin delivery space and has set its sights on fully-implantable automated insulin delivery. The company has a JDRF partnership to test a seven-day wear, novel subcutaneous insulin infusion catheter.
  • Biorasis is developing a rice grain-sized, long-term, injectable/implantable CGM called “Glucowizzard.” At the moment, the company is focused on improving biocompatibility to improve accuracy and extend sensor lifetime to six months. Biorasis wants to eventually use Glucowizzard in an artificial pancreas setting, so lag time needs to be low, meaning tissue integration is critical. The current approach is to release a steady dose of dexamethasone and growth factors to reduce inflammation and induce angiogenesis surrounding the sensor. Biorasis aims to initiate feasibility testing – in vitro, in vivo, and lagtime characterization – of its injectable prototype in the next year. The company also plans to test on the International Space Station – cool!
  • Last but not least, LifeCare’s Dr. Andreas Pfützner showed off the company’s osmotic sensor, SenCell. Within the miniature sensor, there are two compartments separated by a semi-permeable membrane. In one, there is dextran bound to ConA (a protein). When glucose enters the compartment, it displaces the dextran, increasing the amount of free sugar molecules in the compartment and thereby increasing osmotic pressure. The company has demonstrated a linear correlation between blood glucose and osmotic pressure. Since this sensor is not chemistry-consuming, it could last indefinitely. In a proof-of-concept porcine study, the glucose profile tracked a reference Dexcom trace well, especially when additional osmotic sensors were fastened to the animal to normalize for movement artifacts. Dr. Pfützner shared that LifeCare is working with a company in Germany to print 3D sensors in nanoscale.

8. A session on CMS’ Competitive Bidding featured harsh criticism of the well-intentioned program, with focus largely on its deleterious effects and the government’s inadequate safety monitoring. Diabetes Patient Advocate Coalition’s Ms. Christel Marchand Aprigliano, University of Arizona’s Dr. David Marrero, and National Minority Quality Forum’s Dr. Gary Puckrein all pointed to Dr. Marrero and Dr. Puckrein’s recent study showing that competitive bidding: (i) Increased the percentage of beneficiaries who acquired diabetes testing supplies from retail outlets; (ii) Increased the percentage of beneficiaries who stopped being compliant in SMBG by 23% - a metric correlated with mortality; and (iii) Increased mortality by 27.7% in the competitive bidding group vs. 14% in the control arm. The results of the program, according to the speakers, are decreased access to strips (causing patients to test less) and lower quality strips (causing patients to trust readings, and ultimately test, less). More concerning, however, is the way CMS has monitored the program. After an evaluation, CMS claimed that there has been no disruption and no harm to Medicare beneficiaries. According to Drs. Marrero and Puckrein, this conclusion came from bad science – no baseline values or matched control groups were included, the study cohorts were unstable and unrepresentative of Medicare beneficiaries, and there is a lack of transparency and incomplete disclosure of the methodology. Both men had strong words, as Dr. Puckrein exclaimed “it’s just appalling” and Dr. Marrero added that he is “little freaked out” and that “they didn’t sample like [he] would sample as a scientist.” Dr. Marrero concluded by recommending that CMS be held to the same safety monitoring standards as other clinical trials and that competitive bidding be suspended until CMS can demonstrate the ability to effectively monitor the effects of the program. The one thing this session lacked that we would have liked to see was a voice representing CMS on the panel – the discussion could’ve put more emphasis on ‘so what can we do next?’

  • Dr. Marrero produced a number of quotable quotes and stories as he shared his thoughts on CMS’s competitive bidding:
    • “My Mom called saying she had to move to a new meter because of competitive bidding. Nobody educated her on how to use it. It was more complicated than the current meter, which I helped her pick out. She wanted to stop testing. This is the nature of the beast. And the second problem, she got strips that didn’t come from the company that produced the meter. I found a strip that gave me a 130-point spread across three separate tests. This is where dangerous therapeutic decisions are made. I performed 600 repetitions with these strips, and discovered they weren’t accurate.”
    •  “I do clinical trials for a living, and if I did something that increased mortality of participants, the trial would be stopped.”
    • “This is standard practice in the rest of the world in trying to understand the safety and efficacy of a therapy. The observation that there is not an impact on beneficiaries is frankly wrong. The fact that they’re about to start another round is of great concern to me and others.”

9. In part one of her double-header, Emerson Consultants’ Ms. Patty Telgener encouraged diabetes technology manufacturers to invest early in clinical outcomes data in order to succeed in a changing healthcare climate. She said the technologies that save money in the long term and improve long term outcomes will benefit under Accountable Care Organizations as they replace fee-for-service with alternate payment models.  If a novel or existing technology can keep a patient out of the hospital and in better overall health, providers and hospitals costs can save costs and be more profitable under the alternative payment models. Because provider organizations will be incentivized to find innovative solutions to empower patients, Ms. Telgener is a big believer in designing clinical trials with outcomes in mind – understand how a given value proposition fits within future payment methodologies and define in which segment of the population the product works. In addition, she endorsed pilot programs with specific payers, a good way to demonstrate cost-savings. Yet there are significant hurdles for industry and patients in technology, starting with outdated Medicare benefit categories. Ms. Telgener illustrated this concept with the durable medical equipment (DME) benefit category: In order to be considered, a product must have a three-year expected life. This may have been an adequate condition when the policy was written, but modern disposable pumps/CGMs and software are left behind – they don’t fit in the box. Ms. Telgener really hopes that the categories “can come into the 21st century,” but until they do, these outcomes-augmenting technologies will be inaccessible for huge segments of the population. Payer actions also present challenges: Consolidation of policies (Anthem/Cigna, Aetna/Humana, etc.) and negotiations with manufacturers (see United Healthcare’s decision to fund only Medtronic pumps) will likely narrow patient choices.

10. For her next act, Ms. Telgener delivered a seamless talk on the fight for Medicare coverage for CGM. [Note: The scheduled speaker, Erika Miller of Cavarocchi, Ruscio, and Dennis Associates, of Washington, DC couldn’t make the session, so the amazing Ms. Telgener gave the talk off of slides she had never seen before. Wow!] Medicare currently covers professional, but not personal CGM, despite clinical guidelines and evidence that it is critical – 16% of older Americans with type 1 diabetes experience seizures or episodes of unconsciousness, for which average inpatient admissions costs >$17,000, and the rate of hospitalization has risen 11.7% since 1999. Dexcom’s Ms. Claudia Graham, the session moderator, chimed in with Medicare’s perspective: Because CGM requires fingerstick calibration, it is seen as adjunctive and therefore ‘precautionary’ and not “medically necessary.” Dexcom, along with other CGM players, have been working to obtain a dosing claim and to eliminate calibrations, believing that doing so would satisfy Medicare’s criteria. At this point, however, Medicare cannot fit CGM into a benefit category, despite legislative advocacy like the CGM Access Act of 2015. Ms. Telgener called coverage for type 1s “of urgent need” and forecasted progress on that front, while type 2 coverage is “a little bit down the road.” She added that the CGM coding was established in 2002 for professional CGM and, like Groundhog Day, it keeps coming back. Luckily, the Endocrine Society and AACE are working to make unique codes for personal and professional use. CGM is such a game-changer for so many, and allowing Medicare patients to leverage the technology would likely save CMS money in the long run.

Detailed Discussion and Commentary

Long-Term Implantable Glucose Sensors

Tissue-Integrating Sensors

Natalie Wisniewski, PhD (Profusa, South San Francisco, CA)

Profusa was the most exciting early-stage implantable sensor startup we saw today. The well-funded South San Francisco-based company’s micro hydrogel sensor is extremely tiny (500 microns in diameter x 5 mm long), flexible, tissue integrating (capillaries can grow into it), lasts over two years, has low cost of goods, and can sense multiple analytes (oxygen, glucose, lactate, pH, CO2, creatinine, sodium, potassium, etc.) – watch a video here. A first-in-human glucose sensing trial is expected to start in early 2017, and Profusa’s oxygen sensor (Lumee) received a CE Mark in 3Q16. CEO William McMillan shared plans to secure a CE Mark for glucose sensing in type 2 diabetes and prediabetes in 2018, followed by an expected launch in the top three EU countries later that year. A type 1 diabetes trial is also expected to start in 2018. The tiny sensor is injected under the skin (hypodermic needle) and uses a fluorescence sensing chemistry (calibration details not clear). An on-body patch is worn on the skin that illuminates the sensor, collects data from it, and relays the information to the phone. The company ultimately hopes to have a flexible, thin-film, Bluetooth-enabled disposable patch and/or use a smartphone reader for intermittent sensing (i.e., like Abbott’s LibreLink). The sensor itself is remarkably tiny (see picture below), and the ability to sense multiple analytes gives it a variety of other applications: peripheral artery disease, stroke, tumors, oxygen monitoring during exercise, organ status, microbiome, asthma, chemotherapy, etc. Profusa was incorporated in 2009 (incubated at University of Washington and UCSF), has received $22 million in DARPA and NIH grants (including a “transformative research award” from NIH and lots of military interest), and raised a $13 million B-round of financing in 2015. The team is now 29 people. Though its glucose sensor still has a long way to go, we think this is a company to watch in the implantable CGM area.

 

What Do Clinicians Want in an Implantable CGM?

Jessica Castle, MD (Oregon Health & Science University, Portland, OR)

OHSU’s Dr. Jessica Castle shared 11 things that patients (and thereby clinicians) want to see in an implantable CGM (see table below). Her wish list highlighted one of the biggest challenges for any CGM company: accuracy is just one piece of making a great sensor. In our view, form factor and cost will drive this field in the future, particularly as it expands beyond early adopters. We were not surprised to see her list insurance coverage first, and to note that more expensive devices must bring superior outcomes. Her second item – a device’s procedure and invasiveness – shared a key insight for implantable CGM companies: “I haven’t held a scalpel in 15 years, and I don’t think I want to start now. Frequency is key. I see patients every 3-6 months. Ideally, we want a device that lasts a year.” It was a good point for more invasive procedures like GlySens, or for shorter-lasting devices like Senseonics’ Eversense – will endocrinologists have the bandwidth to support these technologies? We do think procedures will be appealing to some due to procedure codes and corresponding payments. On accuracy, Dr. Castle advocated for a MARD <10% with no egregious errors (>50% away from reference) and preferably almost all values within 15% of reference. She urged CGM companies to integrate with insulin data (“critical”) and to make data easily accessible on the web.

What do patients (and clinicians) want in an implantable CGM?

1. Covered by insurance. “Insurance coverage will drive access. How does cost compare to current CGM devices? More expensive devices must be superior.”

2. Minimally invasive and long wear time. Ideally lasting one year. Does the procedure require a scalpel and training? Does the implant have to be removed, and if so, can it be re-implanted in the same location? “You have to consider where infusion sets will be too.”

3. High usability with integration. What external devices are needed? Ideally an implantable CGM will be integrated with existing devices (smartphones, watches) and avoid second on-body components. How good is the telemetry (distance, interference)?

4. Excellent accuracy that is at least comparable to current CGM devices with a MARD <10%. No egregious errors (more than 50% away from reference), and preferably, almost all values within 15% of reference. Accurate across glucose ranges (hypoglycemia, euglycemia, hyperglycemia), appropriate alarms, and no false alarms. “It’s very troubling if you’re missing hypoglycemia, and it’s very annoying to be alarmed in the middle of night when you’re not actually in hypoglycemia.” Minimal delay approaching physiologic delay.

5. Minimal calibration or factory calibration.

6. No interference with acetaminophen, high dose vitamin C, etc. “People on average probably take acetaminophen at least once a month. We are also managing patients with CKD and want to use this across the whole breadth of different patients.”

7. No compression artifact (i.e., false lows from pressure placed on the sensor while sleeping).

8. Insulin automation. “A pump with open loop is going to be a thing of the past very soon.”

9. Inconspicuous and not readily visible to others.

10. Safe and comfortable, without significant adverse effects (adhesive rash, infection, scarring).

11. Data that is easily accessible on the web for both providers and patients. Good pattern recognition and a sharing function that allows real-time data and alarms to be shared with others. “Integration with insulin data is critical. If I’m reviewing someone’s data and they have nocturnal hypoglycemia, I need to know why that is: too much basal overnight, bolusing for snack at bedtime, or something else? Unless I have access to the insulin data, I cannot tell.”

 

 

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