Q&A: Dexcom CEO talks rapid pace of CGM innovation, what’s coming in 2020 and beyond

One of the biggest digital health success stories of the two decades is in the diabetes space, especially the continuous glucose monitor (or CGM). The first CGM was approved by the FDA in 1999, but in the last decade we’ve seen CGMs become smaller, longer lasting and more comfortable, and we’ve seen huge strides in datasharing as CGM-phone connectivity went from novelty to commodity. 

Dexcom has been one of the undisputed leaders in this trend — it was the first company to get a CGM through the FDA that did not require fingerstick calibration and the first to connect directly to a phone. The company has seen impressive financial returns, but in some ways has struggled with its own success, running up against manufacturing capacity issues.

On the sidelines of the JP Morgan Healthcare conference last week, MobiHealthNews sat down with Dexcom CEO Kevin Sayer to talk about his company’s latest news, future plans, hot-button issues like noninvasive monitoring and CGMs for Type 2 diabetes, and the myth of “good problems” in healthcare.

Let’s start with your news from the conference. You announced a partnership with Livongo, and something with [Eli] Lilly [and Company]?

Lilly had been announced earlier. We formalized a relationship with Lilly where we’ve been working for quite some time to develop interconnected devices with Lilly as they develop insulin delivery systems. They have pens that are going to have Bluetooth connection capability and they’re working on an insulin delivery system as well for themselves, and our CGM will be integrated into the software platforms that they built. So we’ll provide them CGM data and patients will be able to view CGM data on their apps.

And the Livongo announcement, that’s a data relationship. You know Livongo has a coaching platform and they coach patients based on, right now, fingersticks. Our relationship with them is through our APIs. And what we will do is provide them data, and their coaches and analytics can go through the CGM data and see if there’s recommendations that they would offer patients based upon what they see there. We believe it will provide a much richer experience for those patients.

Livongo’s interesting. They got a lot of press this year because of their IPO, and a lot of people are aware of the space they’re in who weren’t before. But on the hardware side, they’re not super sophisticated; it’s just a white-labeled glucometer that has been around for ages. So it’s interesting to see them connecting to parts of the industry where the technology is cutting edge, like Dexcom.

We’re glad we’re connecting to them. We’ve certainly worked with and spoken to others. Onduo, Verily’s diabetes management platform, has been taking data from us to power their stuff forever. I presented some UnitedHealthcare data yesterday in my presentation whereby they’re working on developing diabetes data and metrics based on CGM data. And we’ve even done some of this stuff on our own. So we’re looking at every possible angle to grow our business and provide patients with what they want. If patients truly want a Livongo experience and their software is better than anything we can develop, then — look, they’re buying sensors, they’re using our sensors, it’s all good. And we have to look at what are we best at. What we’re best at is making great sensors and making sensors that are safe and sensors that are accurate. And we’re good at the app we’ve created and the stuff we’ve done.

But if somebody wants to make a sensor for some completely different use in a market they understand better than us, why not serve up our sensor data to them and let them put it in that platform? That’s been our mantra all along.

I’ve been talking to several folks in the diabetes space the last couple days here at JPM about how incredibly fast the space has moved in the past five years.

It’s moved incredibly fast in the past five years. Because that’s when we were the first company to ever go to the phone. We went directly to the mobile platform in 2015. And once that happened, all these avenues are open because you now have data that can go from a body to a server that’s real medical data. That’s actionable medical data.

One of our board members said something really interesting. I’m a doctor, I’ve been at this forever, but CGM really is what personalized medicine was supposed to be when we started way back when. And this is where you have an opportunity to create personalized medicine for people with actionable information out of this data.

Watching you guys, Abbott, Medtronic and Senseonics to some extent — there’s something of a market share race going on, right? Especially in Europe?

Let’s take a step back. We’ve grown remarkably over the last couple years. We’re over 40% growth, compounded every year since 2010. So our growth trajectory has been nothing short of remarkable. Our strategy has been to make our product very much fit within the Type 1 market for insulin-using patients and to seek reimbursement. We’ve not rolled out highly-expensive cash pay platforms or gone the commercial route that many of our competitors have at this point in time.

I think over the next several years you’ll see us move that way. They have grabbed a bunch of patients in Europe, but our Europe business is growing nicely too.

And you do have a well-known brand and a relationship with patients.

We do! And if we do something wrong, they’re also very happy to let us know. It’s OK.

What does 2020 look like for you? What are you excited about?

The first thing we’re excited about is we’re going to double our G6, our current product manufacturing capacity, again this year. We have been capacity-constrained ever since we launched that product. So we’ve not been able to go do some of the aggressive commercial things that we would have liked to have done as far as getting word out to more people.

That being said, we’ve just needed product.

Has that bottleneck been on the manufacturing side?

It has — not only with us, but with our component suppliers. So we’re beefing up our supply chain as well as our internal stuff.

We do have a big G7 year planned. We’ve finalized our product configuration and we’re down pretty much to the end on that where we should be. We’ll start our clinical studies and execute those.

But our 2020 launch is going to be a limited launch; it’s not going to be a big launch. If we’ve learned one lesson from the G6 platform, it’s that we can’t possibly launch this product when we’re not ready to produce tens of millions of them. With G6 we got an early approval and rather than wait until we’d built up a bunch of inventory we rolled it out. And we’ve been running out of product every month ever since, pretty much. We’re not going to do that again. When we get G7 approved, what we’re going to do is we’re going to build up inventory and when we launch it we’ll turn on the faucet and have availability across the board. So that’s more of a 2021 event rather than a 2020 event.

What’s the big leap from G6 to G7, especially in terms of patient experience?

The first thing they get is size. The G7, it’s smaller than a [Freestyle] Libre. It is probably in diameter a little smaller than a quarter, a little bigger than a nickel, and it doesn’t sit up very high. It is very easy to use — it’s literally a two-step insertion.

And how long does it last for?

We’re working on a longer-labeled application, 14 or 15 days. That length of wear will be determined by the accuracy of the data. Our current product is a 10-day standard. So we’ll see. That’s currently the plan of record to get it out that far.

It’s just a different experience for our patients. Right now for example we have a transmitter that pops in, so you have to buy a transmitter and a sensor. No more transmitters. It’s one thing, you peel it off, you pop it in. You don’t have to put anything together. It’s very much ready for prime time.

Does it have a separate receiver or go straight to the phone?

We will offer a separate receiver because there are patients who don’t want to go to the phone. But we’ll also have it go straight to the phone. G6 is the same.

Getting smaller, lasting longer, getting more accurate — is that where this space is now? Is it almost like phones, just a bigger camera, or are there still paradigm-shifting changes to come?

That is such a wonderful question. I’m glad you asked me that. I think, based on the performance of our sensor, while we are very good there are things we could do better. But we’re quite a ways down the path. The things we can make better going forward are very well-known to us; they’re very well-known to our patients. You want to eliminate anything that’s an outlier. Our algorithm is designed to pick up outliers and shut them down, but there may be an occasional one where it doesn’t. So if you could eliminate the big negative experience that’s a good thing.

Certainly, improving experience within various ranges and on various days — these [challenges] are small compared to the problems we’re used to solving. The G7 is a very small physical footprint, I don’t know that we need to go much smaller than that. What’s going to be interesting going forward is this becomes a software and a data thing. This becomes about providing the patient with the experience they want.

So for example, I had somebody in earlier who has two daughters with Type 1 diabetes. His vision of what he wants and what his daughters want is a CGM that connects with pumps, connects with phones, that has all this science and algorithms so his daughters never have to do anything, so they have a fully-automated insulin delivery system. Our Tandem Control IQ joint product that came out late last year is a step along that path.

There are patients who want that and we need to offer a product that will meet those needs nicely. There are others who don’t. There are others who are going to want a pen that’s connected to an app and maybe manage that way. … But if we can create an experience whereby we can, in a very thoughtful way, meet the needs of all patients, I don’t think that’s 10 software iterations [away], but I don’t think it’s one either. I can guess somewhere between two and five.

When we talk about paradigm shifts, I have to ask: Is noninvasive a pipe dream?

With any technology I ask another question first: What does invasive mean? Invasive, the way people define it, means they’re afraid of needles. So I want you to think of something. I’m wearing a G7 on my arm right now. Just touch that. Is there anything invasive about that?

I can’t tell you’re wearing it.

So when they’re solving the invasive problem, they’re not really solving a problem, because a needle puts a sensor in you but that sensor is the width of a hair, it goes in on a very shallow basis, it doesn’t hurt. So what are you solving by creating a noninvasive platform unless you get wild accuracy?

There are three problems to solve in sensors: If you can make it perform better than the other guys, or as good; if you make it more convenient; and if you significantly reduce the cost. So if one of those three things happen and you can maintain the other two, you have a product offering. I’ve never seen anything on the noninvasive side that meets the performance standards. It may be low cost if you’ve got it reading from your watch, but if it doesn’t achieve performance it doesn’t matter.

We look at all of them. We’re not blind to them, but we haven’t seen them make a difference. And if one of them did? Quite candidly, we’d look real hard at making it part of our product family because we’re not going to give the space up easily.

I had another question about moving into the Type 2 market — is that a clinically motivated thing or is it more of a business move?

People with Type 2 diabetes are diagnosed and they’re given a pill and told “walk more, eat less and take this.” And over time they get worse and worse and worse, and by the time you intervene sometimes it’s even too late. You learn more from one sensor session than you ever learn from any of these fingersticks, and when these patients are provided with information they do incredibly well.

We saw in my presentation yesterday the improvement in the program at UnitedHealth Group, how much better a patient was after going through their program. Average glucose values, glucose spikes, everything goes down.

We saw another pilot from Intermountain Healthcare where they just took a small group of 100 patients — 50 on the CGM full time, 50 on meters. They believe the annual cost savings in the CGM was $5,000 a year per patient, based on hospitalizations, lab tests, drug costs, doctor visits, phone calls. It was a big savings across the board.

Look, we’d be disingenuous to say everything isn’t business-motivated, because it is. We’re a company. But we look at a Type 2 population around the country that’s much larger than the Type 1 population and the interaction with those people is more recommendation, it’s more education, it’s more diagnostic. Versus in our Type 1 and our Type 2 intensive insulin populations, it’s fully integrated into drug delivery and dosing a drug. This is a little bit different.

Look at the Type 2 drugs that are out. Next time you go home and watch television, there are more ads for Type 2 diabetes drugs than any drug out on the market. They spend a fortune on Type 2 diabetes drug advertising — Novo, Lilly, Novartis, everybody. Their Type 2 drugs are everywhere because this condition is so rampant and people are afraid of getting on insulin, but we don’t know when the right time to put a patient on insulin is.

Well, you learn that by putting on a CGM [and tracking their condition]. When the drugs don’t work anymore, it’s time to make a change. Or if they don’t work anymore because you’re eating horribly or you’re not exercising, you need to try this or we’re going to move you to insulin. So there’s a diagnostic element here, but there’s also an element where the patients just do better. It becomes a question of getting reimbursement and getting a product and a payment [model] whereby payers will pay for it.

We have found in all of our experiences all over the world that when our product is reimbursed we do fabulously well. Unless it’s extremely inexpensive, cash pay to have a consistent market is hard. So we’ve got to develop our dataset and we’ll continue to work with all of these companies, with the insurance companies, with Onduo, with Livongo. We’ve got multiple irons in the fire and we’ll figure it out. For us the important thing is that they all wear Dexcom sensors.

Any final thoughts?

We haven’t talked much about all the things we accoplished last year. $1.47 billion off a billion-dollar year. That type of organic growth you just don’t see. And that kind of organic growth given all the things — if you’ve listened to all our earning calls you’ve heard all the statements we’ve made about what a struggle it was for us to have enough manufacturing capacity. Scaling this technology is not simple. Yes, everyone’s trying to get here, but the sheer capital it takes and the money; … it’s easy to build 10,000 sensors. It’s not easy to build 20 million. So we’ve learned and we’ve had these inflection points and this year was another one. I think we’re positioned to do good things in 2020 and beyond.

That’s what you call a “good problem,” right? Everyone wants your product but you can’t make it fast enough?

You know, people say that to me all the time, but I think with our device it’s the most horrible problem you can possibly imagine. Because with our device, our patients’ very lives depend on it and when they have to wait seven days for the next one, I don’t have an answer. So, it is a good problem if you’re in the car business. It’s a good problem if you’re selling iPhones. It’s not a good problem if someone needs your sensors or they’re going to die, so I don’t want that problem any more ever again.