FLAIR: MiniMed 670G vs. advanced hybrid closed-loop (the future 780G algorithm) in adolescents and young adults
The first head-to-head crossover trial of two automated insulin delivery systems in the hardest-to-control age group (14-29). The advanced hybrid closed-loop system (the algorithm that became the MiniMed 780G) cut daytime high blood sugar without raising lows, compared with the commercially available MiniMed 670G.
Primary endpoints
- Proportion of time glucose was above 180 mg/dL (>10.0 mmol/L) during the daytime (06:00-23:59), tested for superiority
- Proportion of time glucose was below 54 mg/dL (<3.0 mmol/L) over the full 24-hour period, tested for non-inferiority (margin 2%)
Results so far
Yes. Among 113 participants (mean age 19, 62% female), the advanced system reduced daytime time-above-180 mg/dL from 37% (670G) to 34% (advanced), a difference of -3.0 percentage points (95% CI -3.97 to -2.04; p<0.0001). It did this without increasing dangerous lows: 24-hour time below 54 mg/dL was 0.50% on 670G versus 0.46% on the advanced system (difference -0.06%; p<0.0001 for non-inferiority). One severe low-glucose event occurred (in the advanced-system arm) and was judged unrelated to treatment; none occurred on 670G.
The full picture
What was tested and why it matters
Closed-loop insulin systems (sometimes called an "artificial pancreas") link a continuous glucose monitor to an insulin pump and let an algorithm adjust insulin automatically. FLAIR ran the first head-to-head, randomized comparison of two such systems to see whether a newer algorithm could do better than one already on the market.[^1] It compared the commercially available Medtronic MiniMed 670G against an investigational "advanced" hybrid closed-loop system — the algorithm that was later sold as the MiniMed 780G — which adds automatic correction doses on top of background insulin.[^1]
This matters because adolescents and young adults are the hardest group to keep in range: hormones, irregular schedules, and missed mealtime doses push glucose high.[^1] FLAIR deliberately focused on this group.
Who it was for
The trial enrolled people aged 14 to 29 who had lived with type 1 diabetes for at least a year and had an HbA1c between 7.0% and 11.0%.[^2] They were recruited from seven academic centers: four in the United States and one each in Germany, Israel, and Slovenia.[^1]
How it was designed
FLAIR was a multicenter, randomized, crossover trial.[^1] After a run-in period to learn the equipment, each participant used one system for 12 weeks, then crossed over (with no washout) to the other system for another 12 weeks, so everyone tried both.[^1] Because the two systems look and behave differently, neither participants nor staff could be blinded.[^1] There were two co-primary outcomes, both measured by continuous glucose monitor: daytime time spent above 180 mg/dL (tested to see if the advanced system was better), and 24-hour time spent below 54 mg/dL (tested to confirm it was no worse).[^1] As a device study, it carried no drug-trial phase number.[^2]
Key results
Among the 113 participants (mean age 19; 62% female), the advanced system reduced daytime time-above-180 mg/dL from 37% to 34% — a difference of 3 percentage points (95% CI -3.97 to -2.04; p<0.0001).[^1] Crucially, it achieved this without causing more lows: time below 54 mg/dL was 0.50% on the 670G versus 0.46% on the advanced system (p<0.0001 for non-inferiority).[^1] Only one severe low-glucose event occurred during the whole trial, in the advanced-system arm, and reviewers judged it unrelated to the device.[^1]
What it means and what's next
FLAIR showed that smarter algorithms — chiefly automatic correction doses — can squeeze out more high glucose in a tough age group without trading away safety.[^1] The investigational system went on to become the widely used MiniMed 780G. The authors noted that the next priorities are testing these systems in underserved populations, during pregnancy, and in people with impaired awareness of low glucose.[^1]