Analysis

Abstract

In this issue of Journal of Diabetes Science and Technology, Nakamura and Balo report on accuracy and efficacy of the Dexcom G4 Platinum Continuous Glucose Monitoring System. The authors demonstrate good overall performance of this real-time continuous glucose monitoring (RT-CGM) system, although accuracy data of the next generation RT-CGM system, the G4AP, is already available. Also, now that MARDs seem to move to single-digit numbers, the question comes up how low we need to go with accuracy. Results of the study also showed a reduction in time spent in hypoglycemia, although the clinical relevance should be questioned. To date, few trials have demonstrated a reduction of severe hypoglycemia. Conventional RT-CGM, without threshold suspension or closing the loop, might be insufficient in preventing severe hypoglycemia.

Real-time continuous glucose monitoring (RT-CGM) assists in optimizing glycemic control.¹ RT-CGM systems assess interstitial glucose levels and provide glucose values every 5-10 minutes with trends and alarms. Randomized controlled trials investigating the effect of RT-CGM on glycemic control show that RT-CGM lowers HbA1c levels, especially in compliant, suboptimally controlled type 1 diabetes mellitus (T1DM) patients.^2-6 There is also increasing evidence that RT-CGM reduces mild to moderate hypoglycemia in well-controlled patients^7,8 and severe hypoglycemia in patients with impaired awareness of hypoglycemia.^9,10 To date, the effect of RT-CGM on quality of life is not convincing,¹¹ although RT-CGM might increase treatment satisfaction,^6,12 decrease diabetes-related physiological distress,⁶ and decrease fear of hypoglycemia.¹³ A recent qualitative study of Pickup and colleagues showed that sensor accuracy and reliability, (false) alarms, and other technical issues remain barriers of RT-CGM.¹⁴ Theoretically, improving these barriers will enhance adherence to RT-CGM, which might result in better glycemic outcomes and quality of life.⁵ Therefore, improving the technical capabilities of RT-CGM systems remains essential.

In this issue of Journal of Diabetes Science and Technology, Nakamura and Balo present a prospective, open-labeled, nonrandomized study evaluating the accuracy and efficacy of the fourth generation RT-CGM system, the Dexcom G4 Platinum (DG4P) system. This system is replaces the Dexcom SEVEN PLUS CGM system and received the CE mark and FDA approval in 2012. The DG4P system uses an advanced version of the glucose oxidase-based sensor which can be used for up to 7 days and must be calibrated twice-daily. Like its predecessor, the DG4P system displays “actual” glucose values every 5 minutes.

Seventy-two T1DM and T2DM patients were enrolled to participate in a DG4P system session for up to 7 days. Thirty-six participants wore 2 sensors simultaneously for the purpose of analyzing sensor precision. All participants were invited to participate in 3 in-clinic sessions, on days 1, 4, and 7, for the purpose of obtaining venous blood samples for YSI blood glucose determination. During home use, participants were asked to perform at least 7 finger pricks per day.

Adequate accuracy assessment methods were used to evaluate the accuracy of the DG4P system.¹⁵ Mean absolute relative difference (MARD) was 13% when calibrated with YSI. Clarke error grid (CEG) analysis showed 81% of values falling within CEG zone A and 98% of values falling in zone A+B. Clinical accuracy was analyzed using continuous glucose error grid analysis, although this method has been criticized for being too forgiving.¹⁶ Of the values, 80% were accurate within the hypoglycemia range, 98% of values were accurate within the euglycemia range and 96% of values were accurate within hyperglycemia range. Accuracy assessments of the DG4P system were quite similar when compared with either YSI or SMBG values. The authors also demonstrated that the DG4P system reached a high level of reliability, with 94% of sensors lasting 7 days and good accuracy on the last day. A longer life span of sensor could also, in theory, improve adherence to CGM. These results are in line with other studies evaluating the accuracy of the DG4P system^17-19 and show that the DG4P system is among the best of commercially available CGM systems. However, by the time accuracy and reliability data get reported, the next generation CGM system is often already launched.²⁰

Although accuracy of CGM systems is increasing, it will likely never become perfect because of the compartment problem (assessing glucose values in the interstitium). However, how low do we need to go with accuracy? To date, CGM systems are not approved to be used as a stand-alone device. Patients still need to perform finger pricks to confirm hypo- and hyperglycemic episodes and make insulin dosing decisions. Being able to use sensor values in diabetes self-management will possibly increase treatment satisfaction with RT-CGM. Recently, Kovatchev and colleagues were the first to evaluate, in silico, what level of accuracy is needed for using RT-CGM values in diabetes self-management and demonstrated that further improvement of MARD below 10% did not contribute to better glycemic outcomes.²¹ More studies are needed to verify these findings. On the other hand, the FDA approval of threshold suspension is in fact a judgment that confirmatory finger pricks are not always needed to make a treatment decision based on CGM glucose values, which thereby are deemed accurate enough.

Furthermore, it was suggested that patient satisfaction with RT-CGM accuracy is 1 of the key determinants of quality-of-life-related benefits.²² Therefore, improving perceived accuracy of RT-CGM systems is an important goal. Perceived accuracy can be improved by enhancing the absolute level of accuracy of RT-CGM systems, but another key element in improving satisfaction is education and expectation management. Therefore, besides improving technical capabilities of RT-CGM systems, educational programs should be developed for the purpose of improving confidence in using RT-CGM data.²³

In the presented study, the authors demonstrated good diagnostic features of the DG4P system in the hyperglycemic range, with hyperglycemia detection rates and true alert rates over 90%, which could reduce another main barrier of CGM, alarm, and alert fatigue.¹⁴ However, the diagnostic feature of the DG4P system in hypoglycemic range, with approximately 1 in 5 false detection and alert rates, remains unsatisfactory and needs further improvement.

Results of the study also show a reduction in time spent in hypoglycemia during 1 sensor session. However, the clinical relevance of this outcome should be questioned. Reduction in hypoglycemia was evaluated in the very short term, without a control group and participants performed at least 7 finger pricks per day, which could also contribute to improved glycemic control.

As stated before, the effect of RT-CGM on HbA1c reduction, without increase in hypoglycemia, has been established.^1-6 RT-CGM trials focusing on hypoglycemia prevention also demonstrated decrease in mild-to-moderate hypoglycemia.^7,8 However, an effect on severe hypoglycemia has never been established in an randomized controlled trial evaluating the effect of conventional RT-CGM (without threshold suspension) on glycemic control.^7,8,24 Only 1 observational study showed a reduction in severe hypoglycemia.⁹ Recently, Ly and colleagues demonstrated a reduction in severe hypoglycemia in patients with impaired awareness of hypoglycemia when using RT-CGM with threshold suspension.¹⁰ Although most trials were not powered to show differences in severe hypoglycemia, it may be suggested that conventional RT-CGM (without threshold suspension) is not enough to prevent severe hypoglycemia and that closing the loop is essential for that purpose.