Positive 8-2 vote at FDA Advisory Committee Meeting for non-adjunctive (insulin-dosing) label claim for Dexcom G5 CGM; labeling discussions start next week; Medicare coverage not before 2018 – July 21, 2016

This piece was expanded July 22, with new news on Dexcom’s analyst call and expanded FDA meeting coverage.

We’re breathing a huge sigh of relief here in DC, where an FDA Advisory Committee voted very positively in favor of a non-adjunctive (insulin dosing) label for Dexcom’s G5 CGM: an 8:2 positive vote for overall benefit:risk, an 8:2 positive vote for safety, and a 9:1 positive vote for effectiveness. The day can only be described as a true nail-biter, with panel discussion that went off the rails and into the weeds at multiple points in our view. At the end of the day, the open public hearing was a big driver of the overwhelmingly positive vote.

An FDA approval seems like a near-certainty now, and FDA-Dexcom discussions will start next week (!) and focus on the post-market study, updated product training and education (particularly for HCPs – what guidance do they need, and how specific should it be for modifying insulin doses?), and the other labeling nuances

This report includes highlights from the FDA panel meeting on Thursday, Dexcom’s fascinating follow-up call Friday morning, Close Concerns’ questions, and an appendix with the panelist-by-panelist vote breakdown. The FDA slides are posted here and we’ve taken pictures of Dexcom’s slides and posted them here.

FDA Panel Meeting Highlights

1. The panel voted overwhelmingly in favor of the labeling update: an 8:2 positive vote for overall benefit:risk, an 8:2 positive vote for safety, and a 9:1 positive vote for effectiveness. Panelists offered fairly limited comment following the vote, but the dreaded “we need more data” was generally absent.

2. The most compelling FDA open public hearing we’ve ever heard appeared to bolster the confidence of the panel. Of the 35+ OPH speakers, there was unanimous support for the labeling change, with the exception of one isolated parent. It was a tremendous series of arguments in favor of the label, summarized in the comprehensive table below. Watch diaTribe’s Facebook Livestreamed video here (part one) and here (part two). Major professional associations that spoke all supported the labeling change: AACE ADA, JDRF, and the Endocrine Society. Abbott actually spoke and supported the label, but in a clever way that buoyed FreeStyle Libre.

3. We give major kudos to the FDA, who was thoughtful, highly encouraging, and generally supportive of Dexcom throughout the day. The Agency clearly understands the value of this labeling update, and seems to be wrestling with the necessary education, label language, and human factors concerns. The FDA actually defended Dexcom’s modeling at many points, citing the challenges of running a useful pre-market RCT. We love that open-mindedness.

4. Much of the day debated whether the modeling and human factors data could support an indication change. Many felt panelists missed the boat in these discussions, but by the end of the day, everyone zoomed out to look at the totality of the evidence. A group of panelists urged skeptics to take the modeling data as a starting point – another tool in the arsenal to aid in assessing the suitability of a dosing claim. (“All models are wrong, but some are useful.”)

5. What information should Dexcom provide to HCPs and patients for making insulin dose adjustments based on CGM trends? Some panelists argued that specific dose adjustment recommendations are impossible (“one size doesn’t fit all”), while others said Dexcom’s proposed general recommendations aren’t specific enough. It’s not clear what the final label will include.

• Directional advice would seem appropriate – Dr. Steve Edelman has shown previously how helpful this could be to his patients. We hope the perfect doesn’t wind up the enemy of the good here. Additionally, we’d very much like to see discussion on a training code or some resources that will assist HCPs. We also believe that other support may help certain patients, such as peer-to-peer support.

6. Dexcom deserves enormous credit for blazing a regulatory trail on non-adjunctive CGM use and we can imagine so many positives that will stem from this, assuming approval of this indication. Several questions seemed to catch Dexcom at least a bit off guard (e.g., education and tech savvy-ness of the human factors study participants), though Drs. Bruce Buckingham, Steven Edelman, and Claudio Cobelli gave outstanding remarks on the clinical need, rationale for the label, and value of simulation modeling in Dexcom’s prepared remarks.

Dexcom Investor Call Highlights

7. Dexcom held a call on Friday morning (just seven minutes of prepared remarks) to discuss the approval, highlighting: (i) FDA discussion on the labeling and post-market study will start this coming week (very impressive - wasting no time!); and (ii) a national coverage decision from CMS won’t come before sometime in 2018. The latter is clearly a long and hard-to-predict process, but we do see it moving ahead now. A strong advocacy community will have more ammunition with this indication to work with CMS.  

8. Dexcom will use this replacement indication as a competitive advantage: “I believe there will be very strong criteria for any system that is going to be called a non-adjunctive CGM. And we now have the opportunity and the ability as we meet with the FDA in the future months to very clearly define what those criteria are going to be. And that's what we intend to do.” Added Kevin Sayer later on, “The outcome of this panel could very well set the tone for all future CGM products.” Dexcom regulatory mastermind Andy Balo felt FDA will set an accuracy threshold for non-adjunctive labeling, and the Agency will consider a product’s composite features (e.g., alarms). This could have important implications for FreeStyle Libre.

9. Management was extremely positive on the potential of this indication to expand adoption of Dexcom CGM: “Medicare is only the tip of the iceberg. I think this is huge.” CEO Kevin Sayer highlighted the marketing benefits of the new indication

10. An automatic shutoff for its CGM (preventing a sensor restart) “remains to be discussed with the Agency going forward.” Dexcom is also looking at disposable transmitters or monthly kits that would contain a one-month transmitter and three 10-day wear G6 sensors. This has important implications for seamless pharmacy distribution. Distribution has been challenging for some patients of late, which reinforces the steadily growing demand.

FDA Panel Meeting Highlights

1. Despite plenty of criticism on Dexcom’s modeling, the panel ultimately voted overwhelmingly in favor of the labeling change: an 8:2 positive vote for overall benefit:risk, an 8:2 positive vote for safety, and a 9:1 positive vote for effectiveness. See the table in the appendix below for a summary of how panelists voted and their fairly limited comments. We were surprised not to hear more from panelists in characterizing their voting. Whereas on the drug side at CDER, we always gained tremendous insight from panelists frequently detailing explanations of their votes, we didn’t hear much from them in the explanations (four simply stated, “I voted, yes, yes, and yes.”). While the panel shared many valuable points, we also point out that one downside of the “no conflict of interest” rule by FDA means that panels often have little experience in the discussion at hand – for example, few were prepared to understand the simulation models and we wonder what preparation they had. This rule is understandable, but we wish there were exceptions – Kelly was sitting next to an engineer during the meeting who had significant experience with and understanding of the models and this guru as a guest to the Panel could have provided very valuable teaching so that the panelists better understood the data.

• NIH’s Dr. Lisa McShane, the lone statistician on the panel, voted no across the board, citing a lack of available clinical trial data on the safety and efficacy of insulin dosing off of CGM. We wonder what kind of clinical data she would have wanted to see that would have been ethically acceptable – much of the data from the simulation explained circumstances where hypoglycemia and other scenarios were present. The only other dissenter was Dr. David Cooke (Johns Hopkins University, Baltimore, MD), who voted in favor of effectiveness, but voted no on safety and the overall risk:benefit ratio (as we had mentioned in our preview for this meeting, he also voted against approval of MannKind’s Afrezza, which was telling); Dr. Cooke also wanted to see a pre-market clinical trial comparing non-adjunctive to adjunctive CGM.
• We were glad that the FDA broke up the voting questions into three parts, adding granularity as it considers the update. With highly positive votes across the board, an approved labeling update seems extremely likely.

2. The most compelling FDA open public hearing we’ve ever heard appeared to bolster the confidence of the panel. Of the 35+ OPH speakers in the 90-minute block (the FDA expanded this twice – first, from 60 to 90 minutes and then from 30 speakers to 35+), there was unanimous support for the labeling change, with the exception of one isolated parent. It was a tremendous series of arguments in favor of the label, summarized in the table below. We have never seen such an impressive group of advocates and clinicians speak in one session – see the list below the table and watch diaTribe’s Facebook Livestreamed video here (part one) and here (part two). Major professional associations that spoke all supported the labeling change: AACE, ADA, JDRF, and the Endocrine Society and although the AADE did not speak (which was surprising given that education is so linked to the subject), the organization wrote a highly persuasive letter to the FDA as we understand it, which presumably was signed by many notable educators. We had a triple hitter from our teams at Close Concerns and diaTribe – Adam’s slides are posted here for Close Concerns, Kelly’s slides are posted here and talking points here for the diaTribe Foundation, and Ava Runge’s slides are posted here for the diaTribe Foundation – all have diabetes (for a collective 50+ years).

• We heard compelling testimony from 22 patient advocates and a who’s-who of many of the field’s leading thinkers, researchers, and clinicians.
  
  • OPH patient advocates included Kristine Agnello (T1D); Adam Brown (T1D, Close Concerns head of diabetes technology); Kelly Close (T1D, The diaTribe Foundation); Greg Dooley (father); 15-year-old Caroline Dorn (T1D); Joslin Medalist Dan Fleshler (T1D); Gregg Gerety (T1D); Manny Hernandez (T1D; VP, Livongo Health, Diabetes Hands Foundation founder); Dr. Aaron Kowalski (JDRF Chief Mission Officer), Gene Kunde (CEO, Diabetes Hands Foundation); Jeff Hitchcock (father; Children with Diabetes); Dr. Nicole Johnson (T1D; Students with Diabetes); Michael Keane (father); Dr. Sarah Kimball (T1D, mother) and her 10-year old son Sam Mazlish (T1D); Thomas Lye (T1D, National Federation of the Blind); Matthew Merrinack (spouse); Gerald and Owen O’Connell (father and son, both T1D); Christina Roth (T1D; College Diabetes Network); Ava Runge (T1D, the diaTribe Foundation); Lorraine Stiehl (spouse); and 76-year-old Joslin Medalist Lynn Wickwire (T1D).
  • OPH scientists, researchers, and clinicians included Drs. Aaron Kowalski, Bill Tamborlane, Bill Polonsky, Bob Ratner, Boris Kovatchev, Daniel DeSalvo, Janet McGill, Lori Laffel, Nicholas Argento, and Satish Garg. The doctors helped elevate the patient arguments significantly – they were all very consistent.
• Drs. Robert Ratner, Satish Garg, and Bill Tamborlane brought some much-needed wisdom and perspective, emphasizing during the OPH that fear of the unknown, “patients won’t know what to do,” and resistance to change were also barriers when the field was switching from urine testing to SMBG. All three doctors were practicing during the transition from urine testing to fingersticks, and recounted that clinicians at the time were highly skeptical about the meters, saying that patients would never use them because they were too complex (“even if patients do prick their fingers, they won’t know what to do with the data”), expensive, inaccurate, and they required calibration. Those clinicians failed to acknowledge the highly favorable risk-benefit ratio of fingersticks, relative to urine testing, and underestimated the savvy of patients. Drs. Ratner and Garg noted that CGM is now facing the same hurdles – this added extremely helpful context from our view. Dr. Tamborlane added that putting trust in CGM is not a new concept, as the approval of the MiniMed 530G low glucose suspend technology already established a precedent for making clinical decisions based on sensor readings. It was terrific to see such experienced and influential figures elevate the dialogue to the big-picture level.
• Abbott’s Dr. Mahmood Kezemi began his OPH talk in favor of approval, which was unprecedented in our view and favorable for the day. He did then list key areas where non-adjunctive CGM must mitigate risk. He did not mention the company’s factory-calibrated FreeStyle Libre by name, but it was obvious that the listed risks were chosen carefully: (i) CGM systems that rely on manually inputted BGM data should include risk mitigations for errors; (ii) human factors should address calibration issues, such as inaccurate glucose values; (iii) CGMs should not provide data over the approved length of wear; (iv) risk mitigation should account for acetaminophen interference. Overall, we think Dr. Kezemi’s points could be helpful as the FDA and Dexcom discuss training. It happens that some of what Dr. Kezemi discussed related to what could be seen as advantages of Abbott’s FreeStyle Libre, which Dexcom’s G5 does not have, such as factory calibration (no BGM values inputted), auto-shutoff at 14 days, and no acetaminophen interference. We don’t see most as relevant since Dexcom could obviously instigate auto-shutoff if it wanted to very easily, they are moving toward factory calibration, and acetaminophen interference will be a thing of the past eventually. The consumer version of FreeStyle Libre still hasn’t been filed with the FDA, but CEO Miles White projected earlier this week that approval could come in 1Q17, which we view as an optimistic time horizon. The FDA’s decision on the G5’s label will have important implications for Abbott and other CGM products that seek replacement labeling. 

3. We give major kudos to the FDA, who was thoughtful, highly encouraging, and supportive of Dexcom’s modeling throughout the day. The Agency clearly understands the value of this labeling update, and seems to be wrestling with the necessary education, label language, and human factors concerns. FDA seemed quite supportive of the protocol Dexcom used for modeling and human factors. FDA’s short presentation (see slides here) ran through Dexcom’s clinical accuracy, model simulation, and human factors studies, mostly reiterating what Dexcom showed. FDA gave a balanced review of the strengths and limitations of the modeling, but was generally supportive of their value – even with the long list of assumptions that went in. Some KOLs with whom we spoke (some there, some watching online) felt a number of the panelists missed the boat on this point, spending too much time criticizing the models and missing the bigger picture: the totality of evidence (clinical accuracy, simulation modeling, human factors, real-world experience), and the value of the models as an important (but by no means only) piece of information. A couple panelists suggested a pre-market trial comparing adjunctive vs. non-adjunctive CGM use would be “small” and “easy to do.” Drs. Courtney Lias and Stayce Beck were straightforward that the FDA had felt a pre-market clinical study would be hard to design and might not stress the G5 enough for the scariest scenarios (presumably this would be unethical); hence, they encouraged Dexcom to do the modeling studies, which they seemed to support. It would have been helpful to have a panelist who could have explained more of the background of the simulation to other panelists or a “guest” to the panel.

• The FDA noted that the trend information provided by CGM helps to inform treatment decisions, but noted the (potentially) lower accuracy of CGM vs. BGM and the potential human factors pitfalls with having the extra information. The emphasis was heavily placed on the latter concern, as FDA’s Dr. James Mullally indicated that users might actually use the additional information incorrectly and make dangerous treatment decisions. As an example, if users see a double arrow up, indicating that blood glucose levels are increasing at a rate of 3 mg/dl/min, will they know whether and how to adjust treatment? The FDA also worried that the near-constant stream of data might lead CGM-naïve users to overtreat, stack insulin, and risk hypoglycemia. It was notable to hear early in the day that the biggest of the FDA’s concerns weren’t related to the technology itself, but could be addressed by education and training.
• We would be surprised to see FDA request a pre-market clinical trial at this stage. FDA’s Dr. Stayce Beck explained the reasons not to do a study in Q&A: “We want to hear from panel what kind of outcomes would be appropriate – we struggled to come up with what they should be, barring safety. Some outcomes, like hypoglycemia, are difficult because they are self-reported. The problem is that severe hypoglycemia events are rare – one in 11 person years. For enough events to distinguish between adjunctive use and the replacement plan, you’d need a decent sample size in the thousands, for six months to a year, to see a difference.” With so many people already using the technology non-adjunctively, it was recognized that it would be risky to run an RCT, pushing approval back further. Panelists eventually agreed, that approving now would be advisable so that Dexcom can begin to release educational materials (more on this below). Once the label has been updated, Dexcom can, with the FDA’s guidance, conduct a comprehensive post-market study to ensure safety.
• Dexcom will still be asked to do a post-market study, but the form is still under discussion. Some suggested registries might be possible to leverage on this front – presumably T1D Exchange has thousands of patients in its registry on CGM, as does the dQ&A panel, which includes both type 1 and type 2 patients. Panelists seem to find the data from a recent study dQ&A did with over 650 Dexcom CGM users of interest – some of this was shared in the Open Panel Hearing, such as the fact that 69% of Dexcom users (n=654) currently dose insulin off the CGM even though this is not in the label. (We’d point out that virtually all of the 20 million people in the US on BGM also dose from it, although it is also not in the label.) If you’d like this dQ&A data, please contact Richard Wood at richard.wood@d-qa.com. The focus of the post-market study will be on safety, which will hopefully reduce the trial design burden – we hope significant time is spent designing this study to reduce the potential for “oh the trial design wasn’t optimal!” effect.
• FDA’s Dr. Courtney Lias explained that the FDA considers evidence from a variety of sources, not just RCTs! She read the definition of “valid scientific evidence verbatim: well-controlled/partially controlled/non-controlled studies, well-documented case histories, reports of significant human experience with a marketed device, and reasonable assurance of safety and effectiveness from qualified experts to be valid scientific evidence. This is why the Agency didn’t automatically require an RCT from Dexcom – the wealth of real world experience (numerous OPH presenters claimed that they or loved ones safely dose off the G5 all the time) and Dexcom’s clinical accuracy studies have made it so that the financial cost of a pre-market trial would probably outweigh any marginal assurance of safety and/or benefit. Presumably the dQ&A study of over 650 users could be helpful data to show as well, if higher numbers are wanted – we also assume that many letters that FDA received would have shown qualitatively that there are many patients dosing now. We do caution that some patients will dose insulin using CGM and have negative experiences – but overall we think it will happen far less frequently with this indication in place.
• Dr. Lias pressed panelists on the content of and patient access to educational materials. This made it seem as if the meeting was more about the nuances of this approval, rather than whether it should be approved at all. We remain impressed by the way the Agency continues to reach out to members of the medical community for advice.
• We have long argued the importance of doctors and nurses to receive better reimbursement for CGM training and hope this will happen if this indication emerges. Presumably there could be a code put together for training that would expand reimbursement, which is sorely needed in our view, in order to attract and retain more HCPs focused on CGM. We would like to hear AADE’s view on this as well as other experts on the training front.
• We are not sure if the FDA will wait to see data from the T1D Exchange REPLACE BG study, which will have data in the fall or winter. Presumably this won’t be a requirement since it was not to begin with, though that may be because the FDA didn’t know about the trial at that stage? Some panelists suggested that FDA could make some additional move if REPLACE BG was not confirmatory, but that waiting for the data was not necessary, particularly since it was not a study that FDA had originally recommended. It does seem like a shame that this meeting hadn’t taken place later in the year, though we appreciate that this may not have been possible, given the time constraints to review the PMA (which was submitted last fall).

4. Much of the day included debate on whether the modeling and human factors data could support an indication change. Many felt panelists missed the boat in these discussions, but by the end of the day, everyone zoomed out to look at the totality of the evidence. Many of the panelists, frustratingly, got bogged down in the analysis of the simulations (they don’t take into account heterogeneity, patient learning, complications, exercise, stress, tech-savvy, varying rate of change adjustment, etc.), arguing that they weren’t very informative. Drs. Claudio Cobelli, Courtney Lias, George Grunberger, and Andrew Bremer, among others, urged the panel to take the modeling data as a starting point – another tool in the arsenal to aid in assessing the suitability of a dosing claim. (“All models are wrong, but some are useful.”) Though not the total package, Dr. Cobelli explained that the simulations are powerful because they allow for thousands of trials that directly compare the SMBG and CGM in the same exact virtual user and with all else equal, and because they cheaply enable investigation in certain scenarios that would be dangerous and unethical to study in humans.

• Several panelists also criticized Dexcom’s human factors studies, in which subjects were asked how they would treat a variety of CGM readouts. Many wondered if larger, more representative, or more applicable studies needed to be done. One panelist was disappointed that the studies just investigated how often the patient made the right decision – he would’ve liked to see the protocol go one step further, figuring out exactly how the patient executed the treatment. There was some criticism that the human factors studies were small, as only ~15 patients were included in each user group. The FDA and Dexcom defended the study design, noting that the Agency’s guidance was followed and the studies were quite robust (five in total, 136 patients). One panelist proposed that a survey of existing users would’ve been more convincing than the human factors studies: “I was disappointed with the human factors studies. Many people in the field are using this non-adjunctively. I didn’t see a survey; we need better understanding of how it’s being used. It wouldn’t be very expensive and could be done quickly. The information could be extracted from current patient users.” We though this was a good point and wonder how registries could be leveraged to this effect – more dQ&A data on this front could easily be developed and shown (650 users were written in early July and some data was shown in the Open Panel Hearing – contact Richard.Wood@d-qa.com for more information).
• Said Dr. Gregg Gerety (Albany Medical Center Community Division, NY – a well known center for patients looking for technology expertise) in the OPH: “There is no need to spend research dollars to prove that non-adjunctive use is safe and effective. Extensive real world use has already proven it.” It seemed like many of the panelists agreed with Dr. Gerety, they just wanted to see a compiled source of data, as opposed to accepting word of mouth. Panelist Dr. Kathleen Wyne (Ohio State University, Columbus, Ohio) noted that there is a wealth of de facto data on non-adjunctive CGM use from clinical trials: “We have data from studies showing that when people use this device, their A1c improves with less hypoglycemia … all with less fingerstick monitoring. These trials are telling us something similar to the real world, that non-adjunctive use is working … they just weren’t set up to answer that question.” Fascinating commentary – to her, existing trials showing the benefits of adjunctive use are actually showing the benefits of non-adjunctive use because the trial participants report fingersticking so much less frequently. Buy that or not, there has to be a way to leverage the wealth of data floating around from current off-label users.

5. What information should Dexcom provide to HCPs and patients for making insulin dose adjustments based on CGM trends? There was quite a bit of disagreement on this point, since the simulation modeling used certain assumptions (e.g., +10%, +20%, +30%), but there aren’t agreed upon guidelines. Some panelists argued that specific dose adjustment recommendations are impossible (“one size doesn’t fit all”), while others said Dexcom’s proposed general recommendations aren’t specific enough (“increase insulin with upward trend,” “decrease insulin with a downward trend”). There is probably a happy medium here – we would favor the label incorporating “starting point” guidelines. We also think it would be very helpful for Dexcom to think creatively about what would help patients in addition to the new indication and the HCP guidance. One mention of peer support got shot down fairly quickly, but we can imagine that supplementing the indication with potential for this could be quite appealing to certain patients if presented optimally (the appropriate risks in place, etc.). Particularly in the era of digital health, we urge HCPs to think more creatively about how patients can be involved, since online learning is a default for many patients, particularly given that they receive fairly little time with HCPs. Peer to peer learning would only ever be a supplement, but it may be a very useful one – panelist Anna McCollister-Slipp mentioned that she frequently goes online for education that supplements what she has learned individually or from her doctor, mentioning TuDiabetes (Diabetes Hands Foundation) and diaTribe.org.

• How will Dexcom need to expand its training materials for HCPs or different patient populations? Many expressed concern about HCPs’ ability to handle this new paradigm of managing insulin-dependent diabetes: dosing off CGM values and trend arrows. What guidelines and education should Dexcom be responsible for, and how much should professional associations develop? Patient rep Anna McCollister-Slipp suggested such guidelines have diverse patient input. Separately, many argued that adolescents, CGM naïve individuals, newly diagnosed, etc. may need more robust training than Dexcom has proposed. It’s interesting to us that patients have been dosing using CGM for years, and no one has really worried about this question. Now that it is going to be a formal indication, much more education can ensue! Notably, panelists spoke strongly against a mandatory training requirement, which would have too much burden and be unnecessary for many patients ... most agreed that training should be diverse in every respect to appeal to various populations (in-person vs. online vs. in-app).

6. Dexcom deserves enormous credit for blazing a regulatory trail on non-adjunctive CGM use and we can imagine so many positives that will stem from this, assuming approval of this indication. We can imagine some negatives as well – there are, of course, SO many downsides of current therapy that happen in the dark of the night that do not get much attention. Overall, we believe the indication would acknowledge far more fully what is happening in the “real world” outside randomized controlled trials.

• Several questions seemed to catch Dexcom at least a bit off guard (e.g., education and tech savvy-ness of the human factors study participants), though Drs. Bruce Buckingham, Steven Edelman, and Claudio Cobelli gave outstanding remarks on the clinical need, rationale for the label, and value of simulation modeling in Dexcom’s prepared remarks. The OPH and fairly positive FDA leadership guiding the discussion significantly helped Dexcom today, who probably would have seen a closer vote under other circumstances. We are curious what the impact on other organizations will be – both Medtronic’s MiniMed 670G (which doesn’t need a dosing claim, as we understand it, since it’s only basal modulation) as well as Abbott’s FreeStyle Libre.
• We’ve taken pictures of Dexcom’s slides and posted them here.  

Dexcom Investor Call Highlights

7. Dexcom held a call Friday morning (just seven minutes of prepared remarks) to discuss the approval, highlighting: (i) FDA discussion on the labeling and post-market study will start this coming week (wasting no time!); and (ii) a national coverage decision from CMS won’t come “before sometime in 2018.” Said management, “Andy [Balo] hits the ground Monday and is back on the plane Wednesday. We are going to get moving on getting this labeling finalized and done and see how quickly we can get it out and position this.” CEO Kevin Sayer was careful not to put a timeline on the approval, but we would guess sometime this year is a reasonable assumption. Following the approval, Dexcom will then approach CMS to apply for reimbursement coverage, including a designation into the DME category – “This process is inherently unpredictable,” said CEO Kevin Sayer. He expressed major commitment to pursuing it.

• Management said T1D Exchange’s REPLACE BG study is important, but is not designed for this labeling. We’re not sure the FDA will wait for it. “I can't speak for what the FDA is going to do, but our involvement in that study is, we provided product because we felt it was extremely important. And the only sensor used in the study is the Dexcom sensor, as far as dosing. We've heard anecdotal stories from investigators around the country who've been involved and the anecdotes have been very positive so far. But we don't know what the data is going to look like and we have no control over this at all. Really, the best investigators in the country are involved in this. We have a great deal of respect for Roy Beck down in Florida, who's leading this. It’s funded by the Helmsley Charitable Trust and the T1D Exchange. And we think it'll be a very important piece of information for the diabetes community, but it's not something that’s designed for this labeling. So, I don't know what FDA is going to – they're aware of it.”

8. In line with previous comments, Dexcom will use this replacement indication as a competitive advantage: “I believe there will be very strong criteria for any system that is going to be called a non-adjunctive CGM. And we now have the opportunity and the ability as we meet with the FDA in the future months to very clearly define what those criteria are going to be. And that's what we intend to do.” Added Kevin Sayer later on, “The outcome of this panel could very well set the tone for all future CGM products.” Management noted that the bar for all new CGM devices is rising: “I can tell you based on our trial discussion with FDA for Gen 6, the barrier for even our own future devices is very high, higher than what we’ve had in the past.” This works in Dexcom’s favor, since it has a very positive relationship with the FDA and has secured the biggest regulatory wins over the past few years (G5 putting a class III mobile medical app on a smartphone; Dexcom Share and the downclassification of secondary display of CGM data; etc.).

• Dexcom regulatory mastermind Andy Balo felt FDA will set an accuracy threshold for non-adjunctive labeling, and the Agency will consider a product’s composite features (e.g., alarms). “...they're definitely going to put an accuracy threshold for the sensors really to meet in order to have a non-adjunctive claim. I also feel like they're definitely going to be looking at all those features, the contextual information, the alerts and alarms, and really if you look at what we presented yesterday, the key for us to have accurate alerts and alarms, which our device were. Hyperglycemia was like 98% accurate and for low it was 92% accurate. And those are the features we have discussed with the FDA. And then Courtney Lias, who is a Director of that group, basically asked the panel yesterday to recommend what features on a continuous glucose monitoring device would be required. And I think alerts, accuracy, reliability will be the key features in the future products.”
• Does this approval create a playbook for other companies to obtain a non-adjunctive label claim? Management was skeptical. “All of the standards for non-adjunctive CGM have yet to be published. But the reason we're here is because of several years of incredible performance of this product. And if you heard those testimonies yesterday, you know from where I speak. I don't know how you document and put that into a standard, but this thing didn't get approved yesterday because of simulations. This got approved yesterday because we have honored our commitment to our patients by delivering a technology that performs the way that we said it would, and the way CGM was envisioned back when I started working on this in 1994. Maybe there is a path for others, but I know how we got here and that's where we are.”

9. Management was extremely positive on the potential of this indication to expand adoption of Dexcom CGM: “Medicare is only the tip of the iceberg. I think this is huge.” CEO Kevin Sayer highlighted the marketing benefits of the new indication: “One of our takeaways from yesterday is, we've somewhat downplayed this [indication] on our investor call saying, ‘we're going to get it and life will go on.’ After listening to the patients yesterday, we think it's going to open a lot of doors and we will be able to position the company in a much different fashion from our consumer campaigns, our physician campaigns, and certainly to a general practitioner. We can say, ‘Hey, we have a treatment here that can replace fingersticks for the patients that you treat and give them continuous data.’ I think it does open some doors, and it is a very big deal.”

• Management also highlighted the advantages of an insulin-dosing claim with G6, which will require just one calibration per day, further mitigate calibration risks, and improve the patient experience. “CGM is very quickly becoming a standard of care in diabetes management. We've moved away from fingersticks. Some of the discussion yesterday was on potential for calibration errors. Well, guess what? We get to Gen 6 and you move to one calibration point a day. People were talking about the inconvenience of washing your hands. With G6, you wash your hands in the morning, take a fingerstick, calibrate your system, and you're done for the day and you're managing the rest of the day just on your smartphone. This is huge in terms of opening up the market. Medicare, again, being an important piece, but certainly not the end all for us.”

10. An automatic shutoff “remains to be discussed with the Agency going forward.” We wonder if the FDA would impose this requirement, given the potential for G5 users to extend the sensor wear, but potentially observe worse accuracy after seven days (anecdotally, Adam gets good accuracy up to day 10, but then sees a drop-off; Kelly has seen good accuracy beyond this). This issue was brought up by a couple panelists, but was not addressed extensively in the FDA meeting. We imagine many patients would be extremely frustrated by an automatic shutoff, since extending the wear allows them to save money – more days of sensor wear for the same price. Notably, if the label says to replace the sensor and the patient doesn’t, it is obviously not the fault of FDA or the company if something negative ensues.  

• Dexcom is looking at disposable transmitters or monthly kits that would contain a one-month transmitter and three 10-day wear G6 sensors. This has important implications for seamless pharmacy distribution. We love the idea of a one-month pack and believe it would significantly enhance the patient experience and hassles of obtaining CGM. The current G5 transmitter lasts only three months (to the day, no wiggle room), creating reimbursement and reordering hassles – we’ve heard of some patients (Adam included) who are left without CGM for a week or two as the reorder processes and the transmitter has already died. We expect Dexcom to eventually move to a fully disposable model (like FreeStyle Libre), though this one-month model sounds promising too.

Close Concerns Questions

• Will Dexcom get the exact indications for use it has proposed? “The Dexcom G5 is designed to replace fingerstick blood glucose testing for diabetes treatment decisions.” That would be a huge marketing boon for Dexcom and could really drive adoption of CGM. It could also help in payer discussion, as Dr. Bill Tamborlane pointed out: CGM becomes cost saving when strips decline to two per day.   
• Will Dexcom need to conduct any additional human factors, clinical studies, or development of educational materials? Will Dexcom need to provide more specific insulin-dose adjustment advice?
• How will training take place? Is there a way to ensure that more “free” time from HCPs will not be required? We would like eventually to see new codes put in place to ensure more training can take place.
• How long will it take the FDA and Dexcom to agree on the label? Will this be approved this year? Will the FDA wait for the outcomes of REPLACE BG?
• What will the post-market study commitment look like? Will a registry study suffice, or will FDA require a prospective study?
• Would FDA limit this approval to: certain populations; calibration with certain BGMs only; beyond day one; etc.? How will the FDA evaluate sensor restarts?
• Where will the FDA set the bar for non-adjunctive CGM use? Will it be set at a <10% MARD?
• How will Abbott’s FreeStyle Libre (consumer version) fare in regulatory discussions, given the factory calibration (a positive for safety) and lack of alarms (potentially a negative in the context of this meeting)?
• What can be learned from users in Europe dosing off Dexcom’s label? Dexcom presented fairly sparse data from a post-market follow-up survey done in Germany and Sweden. 200 surveys were sent, with 62 completed in Germany and 23 in Sweden. The population was 53% pediatric and 47% adult. Self-reported outcomes:
  
  • 29% reported always dosing off their CGM
  • 48% reported dosing most of the time off their CGM
  • 10% reported dosing some of the time off their CGM
  • 4% rarely dosed off their CGM
  • 8% never dosed off their CGM
• How will this indication impact adoption of CGM?
• How will clinicians feel about this label? Will there be initial skepticism or resistance?
• Will this label finally crack Medicare reimbursement, or will other barriers stand in the way? How long will Medicare reimbursement take? What are the private payer implications of this label?
• Is there a way to involve community health workers and patients in training as a supplement to healthcare provider training? We believe peer to peer support done effectively could well support healthcare provider training – it would never be suggested as a replacement, though to date, that is how it has been used informally. How various pieces of the ecosystem can work together better is an area we would like to see receive more attention, particularly given that to date, patients have not been able to achieve optimal management. This is due to many reasons, and we believe broader interaction with HCPs would be extremely helpful and that interaction with other patients may also lend support to some patients.

Appendix: Panelist Voting Summary with Comments

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