In a head-to-head trial comparing three currently available continuous glucose monitors (CGMs) for real-world use in diabetes, the newly approved implantable glucose monitor Eversense (Senseonics) performed the best.
Findings from the comparison of the 90-day Eversense with the 7-day Dexcom G5 (not the newly licensed Dexcom G6) and the 14-day Abbott Freestyle Libre Pro (not the patient-use version) were presented here on June 22 at the American Diabetes Association (ADA) 2018 Scientific Sessions by Rabab Z. Jafri, MD, of Massachusetts General Hospital, Boston.
"This was conducted in the outpatient setting and was a direct comparison of the performance of the [three CGMs]. In this real-world use setting all of the CGM devices had lower accuracy than reported from in-clinic studies," Jafri told Medscape Medical News.
Overall, the Eversense was found to be more accurate than the Dexcom G5, and both were more accurate than the Libre Pro. However, Jafri said it's important to note that, as per labeling, the Eversense and Dexcom G5 were calibrated twice daily, whereas the Libre Pro doesn't allow calibrations.
"We've previously shown that the Dexcom G5 is sufficiently accurate to provide the input to the bionic pancreas. These data suggest that the Senseonics Eversense is likewise sufficiently accurate to provide the input to the bionic pancreas devices in future studies," she said.
The data also showed that all three CGMs performed worse when patients' blood glucose readings were in the hypoglycemic range.
Asked to comment, session moderator Anders L. Carlson, MD, medical director of the International Diabetes Center, Minneapolis, Minnesota, told Medscape Medical News, "I think this study reassures us that the Eversense and Dexcom are adequate and will work well in a closed-loop setting. We'll need to see where the Libre falls into that."
Which One to Use?
Carlson added, "We have seen in clinical practice that all CGMs don't always do great in the extremes [of blood glucose]. I think you have to know your patients. If it's someone you're worried about having hypoglycemic unawareness or frequent hypoglycemia, then the Libre may not be your first choice because you want something with alarms, and perhaps based on this, higher accuracy at those extremes."
On the other hand, he said, "if your patient wants not so much to have the CGM alert them to lows but to kind of modify behavior and lifestyle, I don't know that it matters all that much. And of course, there's the cost difference [with Libre being less expensive]."
Regarding the Eversense 90 days, just approved in the United States the day before the ADA conference, Carlson said: "I think we all are anxious to see how patients respond to something that's implantable. The European experience looks positive, but I think we'll have to create new workflows and procedure rooms, and train nursing staff and so on, so there's going to be a little bit of build-up to that."
MARDs Matter, Plenty of Good Options Now, Match CGM to Patient
In the study, 23 adults with type 1 diabetes wore all three CGM devices simultaneously at home for 6 weeks. Twice-daily calibrations for the Eversense and Dexcom G5 were performed as per labeling specifications using the Nova StatStrip Xpress meter. All patients had type 1 diabetes for at least a year, used insulin pumps for at least 6 months, and used a CGM for at least 1 month.
The primary outcome was the mean absolute relative difference (MARD) versus plasma glucose values measured with the Nova StatStrip Xpress meter; that is, the difference between the sensor versus reference value, whereby the smaller the value, the more accurate the reading.
Generally, a MARD less than 10% is considered highly accurate.
In this study, MARD was 14.8% with the Eversense, 16.3% for the Dexcom G5, and 18.0% with the Libre Pro. The difference between the Eversense and Libre Pro was significant (P = .004).
In contrast, Jafri noted, MARDs reported by the respective manufacturers based on in-clinic trials were 8.8% for Eversense, 9% for Dexcom G5 in adult and 10% in pediatric studies, and 12.3% for Libre Pro.
Performance for all three in the current study was worse in the hypoglycemic range (plasma glucose < 70 mg/dL), with MARDs of 24.9%, 23.6%, and 36.1% for the Eversense, Dexcom G5, and Libre Pro, respectively. Here, differences were significant between the Libre Pro versus Dexcom G5 (P = .004) and Eversense versus Libre Pro (P = .046).
At the hyperglycemic plasma glucose range (> 180 mg/dL), MARDS were 12.8%, 13.3%, and 17.3% for Eversense, Dexcom G5, and Libre Pro, and was significant for Dexcom (P = .001) and Eversense versus Libre Pro (P = .003).
In a separate two-way comparison, MARDs were 15.1% for Eversense and 16.9% for Dexcom G5 (P = .008). Here, accuracy between the two didn't differ in the hypoglycemic range but did in the hyperglycemic range, with MARDs of 12.1% with Eversense versus 13.8% for Dexcom G5 (P = .001).
Carlson concluded: "I think it's great that we have good options now. It all goes back to knowing your patient and trying to match the right technology to the right patient."
American Diabetes Association 2018 Scientific Sessions. June 22, 2018; Orlando, Florida. Abstract 14-OR.
Comments (SA): The G6 CGM is already launched, as well as 180 days Eversense CE marked in Europe. 
www red DiabetologNytt