February 15-18, 2017; Paris, France; Day #1 Highlights – Draft

Executive Highlights

Salutations de Paris! Day #1 of the 10^th annual ATTD is a wrap – Wednesday was a valuable day of symposia, mostly manufacturer-sponsored, headlined by our first look at data from real-world use of Abbott’s FreeStyle Libre (plus a string quartet following the opening remarks!). See below for our top takeaways from the day.

Top Five Highlights

1. Dr. Ramzi Ajjan presented real-world data from >55,000 FreeStyle Libre users (also in a press release), showing that as sensor scanning increased, (i) estimated A1c dropped from 8.0% to 6.7%; (ii) time spent with glucose <70, 55 and 45 mg/dl decreased by 15%, 40%, and 49% respectively – this was particularly moving; (iii) time >180 mg/dl fell a whopping 44% from 10.4 to 5.8 hours per day; and (iv) time in range increased 40% from 12.0 to 16.8 hours per day. In addition, users scanned 16.3 times per day on average (consistent with IMPACT), with some scanning as frequently as 50 times per day! Data from a cohort this large may help to convince payers that people in the real world use and see compelling benefits from Libre. We were particularly moved by the data as it reinforced lots of what we hear in the “real world.”

2. A packed Lilly-sponsored panel riffed on connected care in diabetes, highlighting the promise and some key challenges: (i) R&D often relies too much on feedback from early adopters, failing to design for the broader patient population (e.g., not seeing an endo, perhaps with A1c in the double digits); (ii) patients want to spend less time thinking about their diabetes, while developers usually seek to maximize time customers spend in an app; and (iii) technologies will need to address different priorities – more focus on patient-driven tools in type 1 and more focus on providers in type 2.

3. mySugr and Prosciento (formerly Profil) presented a retrospective analysis suggesting a 1.3% estimated A1c reduction over six months in 440 randomly selected high-risk Logbook app users (baseline estimated A1c: 9.0%). A mySugr newsletter also confirmed today that two German payers will reimburse its population health management (first mentioned at JPM), and Roche and mySugr will finally launch in the US this spring (including the new Accu-Chek Guide BGM).

4. Tandem presented a retrospective analysis comparing CGM outcomes with the t:slim G4 vs. previously published Medtronic CareLink data for sensor-augmented pump users. t:slim G4 users had statistically significant lower rates of hypoglycemia vs. Medtronic users – roughly half as much time spent <70 mg/dl, translating to ~39-75 minutes less time in hypo per day. The exploratory analysis is encouraging, though comparing across studies is always difficult.

5. Dr. Raimund Weitgasser (Clinic Diakonissen, Salzburg, Austria) commented that Roche will explore the Accu-Chek Insight CGM’s performance over 14-15 days – it is currently intended for seven-day use. He also confirmed that it launched in the Netherlands, Sweden, Norway, and Denmark via specialized diabetes centers (in line with Roche 4Q16).

Top Five Highlights

1. Abbott Shares Real-World FreeStyle Libre Data: More Scanning=Lower A1c (8% vs. 6.7%), Up to 49% Less hypoglycemia

Dr. Ramzi Ajjan presented compelling real-world data from >55,000 FreeStyle Libre users (also in a press release) suggesting that, when they scanned at higher frequencies, their A1cs were lower and they spent less time in hypoglycemia. De-identified data was collected over a period of 18 months (October 2014 – May 2016), and compiled to nearly 400 million glucose data points, 64 million scans, and 1.2 million fingerstick blood glucose readings. Through this mass of incredible data points came some strong trends: As scanning increased, (i) estimated A1c dropped 8.0% to 6.7% (see figure below); (ii) time spent with glucose <70, 55 and 45 mg/dl decreased by 15%, 40%, and 49% respectively; (iii) time >180 mg/dl fell from 10.4 to 5.8 hours per day; and (iv) time in range increased from 12.0 to 16.8 hours per day. In addition, users scanned 16.3 times per day on average, with some scanning as frequently as 50 times per day! We were not surprised by these findings in one sense– Libre is a great technology that makes obtaining glucose data less painful, and more frequent glucose data and trends gives actionable information to make therapeutic change. What is compelling here is real-world data from a 55,000-patient-strong cohort, especially because it confirms data from Libre’s RCTs. We hope this stands as further evidence (especially for payers) that patients derive significant real-world benefits from Libre, are willing to scan at high frequencies outside of a clinical trial setting, and can make appropriate therapeutic and behavioral adjustments.

2. Lilly Panel on the State of Connected Care and Challenges Ahead

A Lilly-sponsored panel titled “The Value of Connected Care in Diabetes” with Glooko’s Rick Altinger, BCG’s Chris Bergstrom, our own Adam Brown, Dexcom’s Steve Pacelli, Helmsley Charitable Trust’s David Panzirer, and Lilly’s Dr. Howard Wolpert identified a range of important challenges in connected care for diabetes:

• Developers often rely too much on feedback from early adopters, failing to reach the broader patient population. Dr. Howard Wolpert explained how uptake of early CGM was exclusive to technophiles, or those well-versed with technology and highly-engaged in diabetes self-management – the burden of using the technology was simply too high. Mr. David Panzirer argued that the problem persists: “we in diabetes are used to hearing from the vocal minority, but the great majority of people with diabetes in the US are adults not seen in an endocrinology clinic.” He emphasized that developers need to look beyond the “uber user,” beyond the “educated consumer” whose ears are perked waiting for the next product. Imagine the difference the field could make on a population level if we designed diabetes devices instead for patients who are truly struggling? Mr. Panzirer pointed to patients with double-digit A1c as an opportunity for targeted connected care technology – these are the individuals who add the greatest cost to the healthcare system, and it’s an easier task to get them down to 8% than to get someone with a 7.2% down to 6.5%. Notably, on this theme as well, Glooko CEO Rick Altinger announced that the company has just signed a contract to enter its diabetes management platform into 106 sites, mainly in primary care, to extend reach beyond those in an endocrinology clinic.
• Patients want to spend LESS time thinking about their diabetes, while developers often seek to maximize time customers spend in an app. Our own Adam Brown pointed out this irony, which explains some of the discord between “great apps” and “great diabetes apps” –  the metric for success for a diabetes app is “less time spent on diabetes, more time spent on life.” Added David Panzirer, “The second you ask patients with diabetes to do more, you lose a majority of them. Make devices easier to use. Something that asks you to take a picture of every meal is not going to gain mainstream traction.” Insurance and administrative hassle also adds diabetes hassle, something many panelists commented on too. Dexcom’s Steve Pacelli hoped that the “home-run” second-gen Dexcom/Verily CGM (flexible, bandage-sized, less expensive, 2020-2021 launch) would ideally be available for pick-up at a drug store, or for online purchase on something like Amazon Prime  We see huge potential in direct-to-consumer models that offer less insurance headache and less hassle overall.
•  The ideal technology design for type 1 and type 2 diabetes may be distinct. Adam broadly described type 1 diabetes as more “patient-driven,” while type 2 diabetes is often more “provider-driven” – that distinction (while far from perfect, and not to generalize!) does drive different design priorities for different groups. We certainly keep hearing that HCPs are drowning, and as we have written many times, there would be tremendous upside in providing better tools to titrate insulin, select the right therapy, spend less time on EMRs, etc. as many organizations are trying to do. Dr. Wolpert agreed that while there’s a lot of overlap between the two, there’s a big need for population management and surveillance in type 2 diabetes. The goal, of course, is for fewer to end up in the emergency room with severe hypoglycemia (to say nothing of kidney disease, cardiovascular disease, strokes, etc.) and expensive hospital bills that could have been avoided.
• For quotable quotes from this session, on these and other themes, see the detailed discussion and commentary below.

3. mySugr Retrospective Data Shows 1.3% estimated A1c reduction in 440 High-Risk Users over 6 mos; Two German Payers Will Reimburse mySugr Population Management

mySugr and Prosciento (formerly Profil) presented a retrospective analysis suggesting a 1.3% estimated A1c reduction over six months in 440 randomly selected high-risk Logbook app users (baseline estimated A1c: 9.0%). To be in the retrospective analysis, patients had to have a mean baseline blood glucose of ≥183 mg/dl (estimated A1c >8%) and high engagement on the mySugr Logbook app (logging ≥5 days/week for ≥6 months). Mean blood glucose fell 18%, from 211 mg/dl at baseline to 173 mg/dl, an impressive drop. Both high blood glucose index (HBGI) and low blood glucose index (LBGI) improved (see the poster here). It’s great to see the company digging into data from its nearly one-million strong (!) user base, especially through this important research partnership with Prosciento to clinically validate digital health. We look forward to seeing more outcomes, especially when the Logbook app is couple with coaching. Separately, mySugr announced a few notable industry updates today:

• Following the first mention at JPM, two German payers will reimburse for mySugr’s Coaching and population health management: “We’re proud to be the first digital diabetes service that will be reimbursed by a German health insurer. BBKK and UKV policy holders will soon receive proactive coaching from diabetes educators based on their real-time therapy data thanks to mySugr Coaching.” The insurers cover over 2.6 million lives and are among the top 10 German private health insurance companies – whoa. At JPM, CEO Frank Westermann also mentioned this includes usage-based unlimited strips and a connected BGM for $850 per year (~$71 per month). The meter was not specified, but we assume it is Roche’s Accu-Chek Connect, a brilliant business move for both companies and a trend we see continuing to expand in BGM.
• Roche and mySugr will finally launch in the US this spring, and include the new Accu-Chek Guide BGM. As a reminder, this partnership was signed last April, though the US launch has taken longer than expected. mySugr users will be able to download a voucher in the app and redeem it for the new Accu-Chek Guide meter at local pharmacies. Once paired with the mySugr app, the Roche meter syncs data directly to it. Remarkably, mySugr’s user base of 900,000+ is up 50% from April 2016, when it stood at 600,000+.  
• mySugr updated its bolus calculator to improve post-meal calculations and offer easier setup – see the blog post here. As a reminder, this is approved in Europe as a Class IIb device and included in mySugr Pro; according to the blog post, the FDA is a “work in progress.” We see big potential for this in the US, given that most patients are not on pumps and most bolus calculator apps are not approved (and some are flat-out dangerous).

4. Tandem Retrospective Data Suggests Hypo, Time-in-Range Advantages for t:slim G4 over Medtronic MiniMed SAP

Tandem presented a retrospective analysis (n=3,046) comparing CGM-based outcomes with the t:slim G4 (downloaded from t:connect) vs. previously published CareLink data for Medtronic sensor-augmented pump users (n=7,916; Battelino et al., Diabet Med 2015). As shown in the table below, Tandem found statistically significant lower rates of hypoglycemia in t:slim G4 users vs. the previous Medtronic publication – roughly half as much time spent <70 mg/dl (3%-3.6% vs. 6.3%-8.5%; p<0.001), translating to ~39-75 minutes less time in hypoglycemia per day with t:slim G4 – wow! We are thrilled to see the company using outcomes beyond A1c to inform HCPs and patients so comprehensively. This effect was consistent regardless of sensor usage (<25% to >75%). t:slim G4 users wearing a sensor >25% of the time also spent significantly more time-in-range: ~23-81 more minutes per day. Results were mixed for hyperglycemia, as shown below. Overall, these data are an interesting exploratory analysis – we do emphasize that comparing across studies is challenging, even though Tandem ensured similar inclusion criteria (T1/T2 using an SAP, at least six months of sensor use, and more than 15 days of CGM use during the first six months). The press release suggests Tandem’s intuitive touchscreen interface may be responsible for the results, though we’d note the improvement could also be from the more accurate Dexcom sensor vs. Medtronic’s older Enlite. Interestingly, approximately half of the Medtronic users in the comparator publication were using low glucose suspend, so Tandem’s hypoglycemia results are encouraging from that perspective. We’d emphasize that these studies are not really comparable, as they occurred at different times (2015-2016 vs. 2011-2013), in different regions of the world (US vs. Europe), and Tandem’s user population is probably more likely to be early adopters in better control.

• A separate poster shared self-reported, retrospective hypoglycemia outcomes in a survey of Tandem users. Tandem pump users reported a 52% reduction in severe hypoglycemia compared to their prior therapy. Data were collected from patients with either type 1 (90%) or type 2 (10%) diabetes who had been using a Tandem pump for at least 1 year (n= 1,370). Participants voluntarily filled out an online survey with questions about the occurrence of mild hypoglycemic events (able to treat the low themselves), medium hypoglycemic events (needed assistance from someone but did not need to go to the hospital), and severe hypoglycemic events (needed assistance from a medical professional and/or needed to go the hospital). Questions focused on both the year before starting a Tandem pump (using previous method of diabetes therapy) and the past year using their Tandem pump. Even when controlling for CGM usage, Tandem pump users reported a 52% reduction in severe hypoglycemia, a 31% reduction in medium hypoglycemia, and a 15% reduction in mild hypoglycemia, compared to previous methods of diabetes therapy. Furthermore, there was a statistically significant reduction in ambulance rides due to severe hypoglycemia of 58% and in days spent at the hospital due to severe hypoglycemia of 50%. Again, the results must be interpreted through the lens of retrospective data (and in this case, self-reported), but we’d note that a severe hypoglycemia is more likely to be remembered.

5. Roche Confirms Dec 2016 Launch of Accu-Chek Insight CGM in Select Markets; Will explore 14-15-Day CGM Wear

Dr. Raimund Weitgasser (Clinic Diakonissen, Salzburg, Austria) shared that the Roche Accu-Chek Insight CGM launched in the Netherlands, Sweden, Norway, and Denmark via specialized diabetes centers. We heard that the launch was successful in Roche’s 4Q16 earnings supplementary materials, but this was a definitive confirmation of the specific geographies. No further details were shared on the launch, though Dr. Weitgasser reviewed the accuracy (MARD: ~10.6%) and lag time (~5 min) shared at DTM and gave a light pipeline teaser: The system will be evaluated “quite extensively” in the next few years, and Roche will perform experiments to examine the sensor’s stability over 14-15 days (it is currently intended for seven-day wear with two calibrations per day). See our EASD coverage for a deeper dive into the CGM – how does it stack up to other options? – along with pictures. We are glad to see Roche investing in this field, since more competition with Abbott, Dexcom, and Medtronic should drive even faster innovation.

Detailed Discussion and Commentary

Flash Glucose Monitoring: Real World Benefits Across the Clinical Spectrum (supported by Abbott)

Real world benefits of Flash Glucose Monitioring in a multinational database

Ramzi Ajjan, MD (University of Leeds, UK)

Dr. Ramzi Ajjan presented compelling real-world data from >55,000 FreeStyle Libre users (also in a press release) suggesting that, when they scanned at higher frequencies, their A1cs were lower and they spent less time in hypoglycemia. De-identified data was collected over a period of 18 months (October 2014 – May 2016), and compiled to nearly 400 million glucose data points, 64 million scans, and 1.2 million fingerstick blood glucose readings. Through this mass of incredible data points came some strong trends: As scanning increased, (i) estimated A1c dropped 8.0% to 6.7% (see figure below); (ii) time spent with glucose <70, 55 and 45 mg/dl decreased by 15%, 40%, and 49% respectively; (iii) time >180 mg/dl fell from 10.4 to 5.8 hours per day; and (iv) time in range increased from 12.0 to 16.8 hours per day. In addition, users scanned 16.3 times per day on average, with some scanning as frequently as 50 times per day! We were not surprised by these findings in one sense– Libre is a great technology that makes obtaining glucose data less painful, and more frequent glucose data and trends gives actionable information to make therapeutic change. What is compelling here is real-world data from a 55,000-patient-strong cohort, especially because it confirms data from Libre’s RCTs. We hope this stands as further evidence (especially for payers) that patients derive significant real-world benefits from Libre, are willing to scan at high frequencies outside of a clinical trial setting, and can make appropriate therapeutic and behavioral adjustments.

• FreeStyle Libre users in this study had almost identical usage and fingerstick habits to when they were enrolled in the IMPACT trial (type 1). This demonstration was a major victory for Abbott and shows the RCT results can be generalized to real-world use. In the six-month IMPACT study of FreeStyle Libre in type 1 patients, Libre users scanned 15.1 times per day, on average – similar to the 16.3 times per day shown in the real-world data. Furthermore, during IMPACT, patients decreased SMBG frequency to once every two days (i.e., for every 30 scans, patients performed a fingerstick only once). These ratios were almost identical to the real-world scanning to SMBG ratio (32 to 1 in this trial). This data validates the behavioral findings (relating to testing) in IMPACT, and demonstrates that the more patients use Libre, the better they do (dose response).
• The real-world A1c vs. scan frequency curve is a near-perfect continuation of the A1c vs. SMBG frequency curve established in Miller et al. 2013 (see image below). It makes logical sense that higher frequency glucose checking will lead to lower A1c, but the advantage of flash glucose monitoring is its convenience, ease, and lack of pain allow for levels of daily testing that would be unthinkable with a fingerstick. 

• We were surprised that hypoglycemia decreases so meaningfully (see red  across the spectrum of testing frequency, implying that a majority of the users are on insulin. At the most, some users were spending more than 40 minutes per day <55 mg/dl! Abbott did not collect data on type 1s vs. type 2s in this cohort, but we know the technology has been very popular in type 1 (and might assume this was mostly type 1s). Abbott has never publicly shared the type 1-type 2 breakdown of Libre users, but we would be curious to find out.

• We would love to see additional data detailing change from baseline. What kind of control did patients start with on Libre and how did they change over time by the end of the observation period? It also remains possible that patients who scanned more are simply more engaged and were already in good control at baseline, and then maintained that control in the study. Perhaps a follow-up analysis will show whether Libre use led to decreases in A1c and hypoglycemia from baseline.
• As a reminder, the FreeStyle Libre consumer version is under FDA review, with a US launch now expected in the “second half of 2017,” back 2-3 quarters from the previously ambitious 1Q17 expectation. Abbott has filed BOTH an adjunctive and a replacement claim for the device with the FDA – a smart hedge to get to market faster. See our Abbott 4Q16 report for more details.

Industry Sponsored Workshop

Lilly-Sponsored Panel on Connected Health: Quotable Quotes

David Panzirer (Helmsley Charitable Trust, New York, NY); Howard Wolpert, MD (Joslin Diabetes Center, Boston, MA); Steven Pacelli (Dexcom, San Diego, CA); Rick Altinger (Glooko, Palo Alto, CA); Adam Brown (Close Concerns, San Francisco, CA); Chris Bergstrom (Boston Consulting Group, Boston, MA)

On Easing the Burden of Diabetes

• David Panzirer: “The second you ask patients with diabetes to do more, you lose a majority of them. Make devices easier to use. Something that asks you to take a picture of every meal is not going to gain mainstream traction.”
• Adam Brown: “I want to spend less time on diabetes – less time spent in an app equals more time spent on life. The irony is that an app developer’s goal is usually to get people to spend more time in an app. That’s a funny conundrum.”
• Rick Altinger: “It’s not about the data, as data will become a commodity over time. We need to produce actionable information that supports decision-making, where we’re helping clinicians and patients make better, faster, data-driven decisions to drive better outcomes."
• David Panzirer: “Quality of life is often thrown around as a buzz word, but there’s really something to it. Tie technology to minimizing the amount of time people spend thinking about their diabetes and to minimizing the risk of comorbid depression, because all of this has an impact on outcomes.”
• Adam Brown: “At diaTribe and Close Concerns, we’re excited about models that are going to shortcut the annoying insurance headache and instead go direct to patient. Ordering your device on Amazon Prime – that’s really compelling.”

On Reaching Beyond Early Adopters

• David Panzirer: “We in diabetes are used to hearing from the vocal minority, but the great majority of people with diabetes in the US are adults not seen in an endocrinology clinic. Don’t assume if you build it, they will come – you really need to go out and do the market research. The educated consumer is already on my product; I need to reach out to others.”
• Adam Brown: “Early adopters often drive this field, which can be a mistake. Designing stuff for someone who’s already doing well is not going to do as much for diabetes as designing something for someone who’s really struggling.”
• Dr. Howard Wolpert: “Early CGM technology required a fairly engaged technophile to derive benefit – the tradeoff between benefits and hassles wasn’t great, so it demanded a willingness to deal with the hassles of first-generation technology. Our current model still doesn’t really lend itself to what people with diabetes need most, but we have potential, with almost everyone having a smartphone in their pocket.”
• Rick Altinger: “As we see more and more people with diabetes, technology has to take root in a primary care world.”
• Steven Pacelli: “It’s on us as the manufacturer to develop low-cost, simpler products that expand the opportunity for 420 million people worldwide affected by diabetes – type 1 and type 2.”

On Technology for Type 1 vs. Type 2

• Adam Brown: “This isn’t a perfect analogy, but I often characterize type 1 as a more patient-driven disease, while type 2 is more provider-driven – especially if you are a type 2 not on insulin, you may feel less ownership of your disease. So in type 2, developing tools for providers might be more critical there.”
• Dr. Howard Wolpert: “Talking about type 1 vs. type 2 diabetes, there’s a lot of overlap, but in terms of a customized solution for type 2 it comes down to connectivity. Look at the patients most expensive to the system – what’s missing is some kind of surveillance, so that we don’t see as many patients coming into the ER with severe hypoglycemia.”
• David Panzirer: “What companies really want is well-characterized patients that they can recruit into a clinical trial really quickly – type 1 research has benefited from the T1D Exchange. Registries play a huge role, and we have well-characterized type 1 patients sitting there ready to enter trials.”

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