Memorandum

Tandem identifies CGM software bug in Control-IQ pivotal trial; study paused until remote update is available this month – March 6, 2019

Very “rare” anomaly related to handling CGM data in edge cases; no reportable adverse events; US launch now expected in “2H19” (one quarter extension)

In unexpected news, Tandem announced that it has identified a minor CGM-related software bug in the ongoing Control-IQ hybrid closed loop study. The six-month pivotal trial (n=168) has now been paused and will resume once a software update is deployed through Tandem’s Device Updater. An update is expected before the end of March, implying a study delay of ~3-4 weeks.

The “rare” anomaly relates to how the t:slim X2 pump handles erroneous CGM data “under specific conditions.” We confirmed with the team that these are “extreme corner cases,” and, unsurprisingly, there have been no reportable adverse events. The issue brings an “increased risk of hypoglycemia,” though the actual clinical impact is likely to be tiny or non-existent (in our view). Tandem may publicly release the specific corner cases at upcoming investor conferences.

Currently, ~120 out of 168 participants  (71%) have already completed the pivotal trial (14+ years), which was previously expected to end in April and report data at ADA in June. We’d guess the results could still come at ADA, though perhaps the trial will now end sometime in May.

A Control-IQ launch is now expected in the “second half of this year” – a one-quarter broadening of last week’s updated plan to launch between “summer and end of Q3” (September). The brand-new pediatric study (down to six years) will now commence following availability of the updated software – an ~one-month delay. Last week’s call highlighted potential for a pediatric and adult launch (6+years) out of the gate, though this minor delay could imply a launch for 14+ years first.

This is a small setback for Tandem, assuming it can deploy the update by the end of this month, finish the trial, and get the Control-IQ algorithm submitted to FDA soon thereafter. Control-IQ still comfortably stands to be second-to-market with hybrid closed loop in the US, and it will leapfrog Medtronic with auto boluses and no fingersticks.

It was laudably proactive of Tandem to announce this news, as it concerns extreme edge cases and is an issue with the Dexcom-owned TypeZero algorithm and use of Dexcom CGM data.

Tandem did a clever job of sharing the negative news as an advertisement for its remote software update tool – a fair point and an advantage it has over Medtronic’s 670G for now. It is impossible to test every edge case for an AID system, and great products will never be perfect at launch – but they will need an infrastructure for continuous improvement. Tandem and Dexcom have done an excellent job on pioneering remote software updates in diabetes technology, going all the way back to the G4 Software 505 algorithm and most recently with Basal-IQ.

  • How long will it take the FDA to review Control-IQ, once it is submitted? The first-in-class MiniMed 670G had a priority three-month FDA review; should we expect a similar or even faster timeline for Control-IQ? On one hand, FDA has already cleared two of the Control-IQ system components (G6 iCGM and t:slim X2 ACE pump), the Control-IQ algorithm has more pre-market, robust clinical trial data than the 670G had, and G6 is more accurate than Guardian Sensor 3. Still, Control-IQ does add automated correction boluses for the first time, a no-fingersticks CGM, and the first components cleared with interoperability in mind. Assuming a summer submission, a launch in 2H19 for 14+ years seems very reasonable; getting an expansion to 6+ years would be upside.

 

--by Adam Brown and Kelly Close