FDA posts decision letter for FreeStyle Libre 2 clearance, including accuracy data and vitamin C interference – June 23, 2020

FreeStyle Libre 2 cleared last week as iCGM with “warning/limitation” against use with AID systems – warning based on potential interference with high doses with ascorbic acid (vitamin C)

This week, we got a glance at the 30-page decision summary for Abbott’s FreeStyle Libre 2, which was cleared last week as an iCGM. The summary confirms that Abbott used Dexcom’s G6 as the predicated device for iCGM clearance and gives a detailed breakdown of FreeStyle Libre 2’s accuracy in both adults and pediatrics (FreeStyle Libre 2 is cleared for users down to 4 years old). Importantly, the summary also includes some background and context for the “Warning/Limitation” announced last week that “The [FreeStyle Libre 2] System must not be used with automated insulin dosing (AID) systems, including closed loop and insulin suspend systems.” The decision to include this limitation comes from potential interferences for users who take high doses of ascorbic acid (vitamin C). Based on the FDA’s language, it seems the Agency is being extra cautious CGMs driving AID systems; the FDA is assuming users will be more conscious of potential interferences when injecting insulin manually, but when FreeStyle Libre 2 glucose values are used to automatically drive an AID system, the Agency fears users will be less likely to “recognize the interference prior to erroneous treatment.”

Ascorbic Acid (Vitamin C) Interference

• Last week’s FDA clearance announcement came with a surprise “Warning/Limitation” listed at the bottom of Abbott’s press announcement that “The [FreeStyle Libre 2] System must not be used with automated insulin dosing (AID) systems, including closed loop and insulin suspend systems.” Outlined in the FDA’s decision summary, a multi-center clinical study (n=60 adults) measuring interference between ascorbic acid and FreeStyle Libre 2 showed high doses of ascorbic acid positively bias FreeStyle Libre 2 readings compared to YSI. Each adult in the study wore two FreeStyle Libre 2s (one on each arm) for at least 10 days. After a 1000 mg dose of ascorbic acid, FreeStyle Libre 2 readings showed a maximum average bias of +9.3 mg/dl – the maximum bias was seen “approximately 2 to 3 hours” after each dose. When second and third 1000 mg doses of ascorbic acid were administered (each dose was four hours apart), the maximum average bias jumped to +18.4 mg/dl and +19.7 mg/dl, respectively. Abbott’s bench studies of FreeStyle Libre 2 showed that interference was proportional to ascorbic acid concentration, but not glucose.

  • Naturally, the FDA’s biggest fear with the positive bias from ascorbic acid comes when true glucose values are low. A +10 mg/dl positive bias when true glucose values are 60 mg/dl could certainly lead users to miss or undertreat lows. As outlined below, FreeStyle Libre 2 actually shows a (much smaller) positive bias in the hypoglycemic range even when no interference is present.

• For context, 1000+ mg/dl doses of ascorbic acid are well above mean ascorbic acid intakes, but not uncommon. According to Mayo Clinic, the recommended daily amount of ascorbic acid for adults is 65-90 mg per day, but the upper limit is 2,000 mg per day. The USDA estimates the amount of vitamin C in one cup of orange juice at 124 mg, though vitamin C supplements, including Emergen-C or Airborne, usually come in 1000 mg doses.

• Ultimately, the FDA’s decision to include the AID limitation on FreeStyle Libre 2 boils down to a fear that users will be less likely to be conscious of potential interferences when using a closed loop system vs. manually dosing insulin. The FDA’s decision letter summary notes that many other BGMs have potential interference with ascorbic acid, but have labeling that indicates these risks. The FreeStyle Libre 2 reader screen comes with a warning: “Do not take high doses of vitamin C (more than 500 mg per day). This may falsely raise your Sensor readings. Supplements like Airborne or Emergen-C have high doses of vitamin C. Read labeling for all supplements to determine vitamin C content.” With this warning, the FDA is comfortable allowing users to manually dose insulin based on FreeStyle Libre 2 readings alone, but not to drive an AID system.

  • From the FDA decision letter: “… when values from the Freestyle Libre 2 system are used automatically by an AID system to start or stop insulin or other medications, the user has less opportunity to recognize the interference prior to erroneous treatment. The risks introduced by this interference when used with an AID system are greater than for similar iCGM devices, yet the benefits are largely equal. Therefore, the benefits do not outweigh the risks for that use. Thus, the Freestyle Libre 2 system must not be used with AID systems.”

• We discussed the competitive and partnership implications for the AID contraindication in more detail last week. Abbott remains confident that the limitation will eventually be removed – Abbott’s work with AID has been long-awaited, though it may not have a huge impact on revenue.

Accuracy Comparison Between FreeStyle Libre 2, Dexcom G6, Medtronic Guardian Sensor 3, and Senseonics Eversense

Accuracy of Abbott FreeStyle Libre 2 in Adults

146 adult subjects (91% type 1s, 9% insulin-requiring type 2s)

Figures in parenthesis below represent the lower-bound of the 95% confidence interval

Accuracy of Abbott FreeStyle Libre 2 in Pediatrics

139 adult subjects (99% type 1s, 1% insulin-requiring type 2s)

Figures in parenthesis below represent the lower-bound of the 95% confidence interval

Accuracy of FreeStyle Libre 2 in Adults Compared to iCGM Special Controls and other CGM systems

--by Albert Cai and Kelly Close