Those who keep an eye on diabetes technology conferences will know that Roche are running an industry symposium at ATTD (Advanced Technology and Treatments for Diabetes) in 2024.
Their session has presentations from some of the most senior and highly regarded endocrinologists on offer. The program is shown below:
What this tells us is that Roche have:
- A new CGM
- A novel CGM that has a level of accuracy worth mentioning
- It uses predictive algorithms for something
Whilst that’s not a huge amount of detail, it’s a good start. It appears we have a new CGM, it has accuracy that’s either very good, or good enough and potentially uses one of the new consensus standards to prove that accuracy, and it has predictive features.
If we use our best search tools, it doesn’t take too long to find another piece of research, published by Roche Diabetes in June 2023.
This brief article talks about the perception of existing CGM users towards a hypothetical system that could predict glucose levels 2 hours out (most current systems predict at most, 30 mins).
The conclusion it drew was:
Increasing the glucose prediction to up to 2 hours would be seen as a potential improvement regarding reductions in hypoglycemia distress and fear of hypoglycemia by both people with type 1 and type 2 diabetes.https://doi.org/10.2337/db23-1794-PUB
On its own, this wasn’t all that interesting, but taken in context with the “novel” CGM presentation at ATTD24, it suggests that Roche’s new system may predict glucose levels out much longer than those currently available on the market.
Okay, but come on, what’s it going to be called?
Once again, we go back to web trawling, and this time having a search for Roche Trademarks. Google, once again, is very helpful and provides a database of trademarks. And plenty of those are for Roche.
They were both filed on the same date, and both seem to cover similar things. The date is in 2020, which suggests that this has been planned for some time.
If I was betting on it though, I’d expect that Roche would go with Accu-Chek Smartguide over Ilustra, as it plays to their heritage in the blood testing market, and feels a little more solid and dependable than Ilustra.
Is there anything else to consider?
We’ve not yet discussed the CGM marketplace. Right now, in Europe at least, there’s a lot of competition.
Almost everyone offering CGM talks about a MARD in the 8%-10% range. The prices go from roughly £35-£55 per sensor. Lifespan is usually 7-14 days (except for Eversense, but that’s a slightly different proposition).
Will Roche jump in to the middle of this pack? When you bear in mind that the annualised out of pocket cost of Dexcom’s lowest priced product is £1216, while Abbot’s is £1255, I’d expect that it wouldn’t be too different from these.
Similarly, in terms of sensor life, we know that apart from the Eversense implantable system, almost everyone has aimed for 14 days, with the notable exceptions of Dexcom (10 days) and Medtronic (7 days). Would an accurate two hour blood glucose prediction with a shorter sensor life offer the right kind of trade off?
Then there’s the Medtronic InPen system. It uses Medtronic’s CGM with the InPen app to provide advice around bolusing plus the CGM alerts associated with highs and lows. This type of “Smart MDI” doesn’t look like it’s available anywhere else yet.
What about MySugr?
Roche already owns MySugr, the Diabetes companion app. MySugr integrates Accu-Chek meters via Bluetooth and Novo Smartpens and has one of the more advanced bolus calculators. It also has integration with CGM data, even if not always real time, can use step data from Google Fit and Apple Health and can track location.
It’s not a huge step to suggest that a new CGM with relatively good accuracy and response times would use the same technology that is contained within mySugr to provide easy access to insulin and activity data as part of the prediction model. This would certainly help the two hour prediction gain better accuracy.
Would it be integrated directly? It’s possible, but it might change the sense of what MySugr is.
Given the branding of a new CGM as “SmartGuide”, I imagine some of this technology ending up in a new app, building on the MySugr “connected” framework. I don’t think mySugr will go away, but I don’t expect it to change much. That would also enable the new “Smartguide” to potentially use on phone machine learning to better customise the two hour prediction algorithm to each user.
Where does this leave us?
Pretty much everything here is speculation, guided by a few snippets of information that we already know.
- The technology to pull together a load of data in a phone app to enhance prediction capabilities with continuous glucose data.
- A list of patents for CGM that would enable them to produce a market competitive product at the right cost.
- Research that suggests that longer term glucose predictions reduce diabetes distress in those using insulin.
- Multiple trademark filings that cover the devices and services discussed here.
While this initially looks like a way to break in to the CGM market with something that no-one else is doing yet, perhaps it’s more than that.
Better predictive algorithms for MDI combined with a lower cost sensor could create an alternative to hybrid closed loop for insulin users that don’t have or don’t want access to the latest technologies. That’s a huge opportunity.