Annals of Internal Medicine publishes Dexcom DIaMonD type 2 cohort data – August 22, 2017

This afternoon, Dexcom announced that Annals of Internal Medicine published data from the type 2 cohort of Dexcom’s DIaMonD study testing CGM in MDIs. The very positive data was initially presented by one of the principal investigators, IDC’sDr. Rich Bergenstal, at ATTD in February: Compared to participants with type 2 diabetes using BGM alone for glucose monitoring (n=79), those using CGM (n=79) experienced a statistically significant 0.3% A1c advantage at 24 weeks: -0.8% with CGM vs. -0.5 with SMBG (baseline for both groups: 8.5%; p=0.02). Unsurprisingly, the benefit of CGM rose with a higher baseline A1c – those starting at ≥9.0% saw a mean 1.4% reduction (inclusion criteria mandated that baseline A1c be between 7.5%-9.9% - improvements would almost certainly be even greater with a higher starting point). At 24 weeks, median time per day between 70-180 mg/dl increased 80 minutes in the CGM group (a 10% improvement from baseline) and 42 minutes in the control group (a 5% increase, measured by blinded CGM) (p=0.01). Low baseline biochemical hypoglycemia (11 minutes per day and 12 minutes per day in CGM and control group, respectively), precluded detection of meaningful changes and there was no severe hypoglycemia in either group, meaning improvements came from decreased time >180 mg/dl. That said, median time <70 mg/dl was reduced by 64%, and median time <60 mg/dl was zero in the CGM group. Despite improved time-in-range and mean one fewer fingerstick per day for the CGM group, there were no significant group differences in perceived quality of life – but we’ll take better health outcomes.

Since the results were initially reported at ATTD, there has been a lot of excitement around shifting perceptions of CGM use in type 2 patients – subanalyses suggested that A1c differences favoring the CGM group were seen regardless of age, baseline A1c, education level, and diabetes numeracy. Similar to the type 1 data (ADA 2016, JAMA 2017), CGM wear time was very strong: 93% of the type 2 cohort was using CGM ≥six days per week at six months (mean 6.7 days per week in participants >60 years old (n=42), and mean 6.8 days/week in participants with baseline A1c >9.0%)). Of course, this was in the context of a clinical trial and while it is not completely clear exactly how it would compare to real-world, this study was intentionally designed to mirror standard care, and numerous KOLs have been impressed by the wear time (notably Drs. Bruce Buckingham and Roy Beck at Keystone, and Dr. Bergenstal at AADE). Dexcom VP of Medical Affairs Dr. David Price noted on the phone that he and his team were “really surprised” that so many type 2s were so willing to wear CGM so diligently. This, of course, will only improve as the form factor improves over time – the improvement will be substantial.

The -0.3% relative reduction in A1c from a high baseline may be fodder for skeptics, but type 2 is a notoriously challenging population to treat and make fast changes, and the study was “pretty standard-care feeling” (according to Dr. Bergenstal). That is to say that there was minimal hand-holding – medications, including insulin, didn’t meaningfully change throughout the study (although they may have after), and the protocol included no structured review of glucose patterns. Given the low degree of hypoglycemia seen in both groups, Dr. Price is confident that titration and coaching could make for further improvements. Dr. Bergenstal also shared at ATTD that he is eager to explore these frontiers.

This is the third publication of data from the DIaMonD study in a medical journal – type 1 in JAMA and type 1 pump extension phase in Lancet D&E – a big win for the field. We are glad to see a real-world-reflective study of CGM in type 2s, something Dr. Bergenstal has pointed out was lacking, and we hope that this data will drive the further uptake and reimbursement for all type 2s on MDI. We believe every patient with type 2 diabetes should at least be on professional CGM twice a year.

• According to Dr. Price, the next logical step, a Dexcom-sponsored and Jaeb Center-run study of CGM in type 2s on just basal insulin (MOBILE), will soon be up on clinicaltrials.gov. We first heard about this study from Dr. Beck at Keystone and are excited to see if these patients derive similar benefits as those on basal+bolus. Dr. Price also shared that Dexcom is in discussions about trials “earlier down the type 2 road” – presumably type 2s on orals and possibly even prediabetes. In our view, people can learn so much about the impact of behavior and lifestyle on glycemia by using CGM, so even these populations would see benefits. We could imagine Dexcom waiting until smaller, more cost-effective devices (like the sensors with Verily) are available before delving into these populations.
  
  • Also in the study pipeline are trials of CGM in children under eight years old (SENCE), adolescents (CITY), and the elderly (WISDM). Dr. Price stated these trials will either use the Dexcom G5 system or the Dexcom G6 system – this is unlike DIaMonD, which used G4, so it’s likely to have even better results.
• These results tell a different story from Abbott’s six-month REPLACE study comparing FreeStyle Libre to SMBG in type 2s with a baseline A1c of 8.8% (ATTD 2016): That study disappointingly missed its primary endpoint – similar 0.3% A1c reductions with both SMBG and Libre – but the most compelling was actually the hypoglycemia data, which improved markedly with FreeStyle Libre overall, overnight, and particularly for dangerous hypoglycemia (<55 mg/dl). Taken together, we think REPLACE and DIaMonD show continuous glucose monitoring (either traditional or Flash) can drive meaningful improvements in hypoglycemia, time-in-range, and hyperglycemia in type 2s on insulin – exactly where the improvements are depend on many things like population, baseline A1c, baseline hypoglycemia, perception of diabetes technology, etc.
  
  • What would happen if a pump were added into the mix? In the OpT2mise study, a treatment effect of o.7% was found in favor of pump therapy vs. MDI. In the type 1 pump extension, there was no significant difference in A1c, though adding a pump prompted a solid time-in-range benefit on top of that from CGM alone (+83 minutes/day, adjusted mean difference). However, the pump group saw slightly more hypoglycemia <70 mg/dl) (+15 minutes per day), while the MDI group had slightly less hypoglycemia (-9 minutes per day; p<0.001). We wonder if similar trends would be observed in a type 2 cohort – clearly, type 2 patients can benefit from CGM regardless of their therapy of insulin delivery if they use insulin.
• Dr. Price told us that the objective is to get the G5 Mobile non-adjunctive use post-market study up and running at 10 sites (n=>1,000 patients) sometime this year. The comprehensive post-market study to ensure the safety of non-adjunctive use was condition of FDA approval – debated heavily at the July Advisory Committee meeting. We’re not sure what form this study will take, but have little doubt it will show safety and effectiveness (especially given the T1D Exchange REPLACE BG study).

Close Concerns Questions

Q: Would those with A1c >10% experience similar/greater reductions in A1c? What about in those with higher baseline levels of biochemical hypoglycemia?

Q: How much bigger would A1c/time-in-range improvements be with more instruction/medication adjustment?

Q: How will this data further reimbursement and adoption efforts?

Q: Is real-world wear time in type 2s similar to that seen in this study?

Q: How could CGM be in people with prediabetes at very high risk of converting to type 2?

-- by Brian Levine and Kelly Close