Replacement of Blood Glucose Measurements by Measurements With Systems for Real-Time Continuous Glucose Monitoring (rtCGM) or CGM With Intermittent Scanning (iscCGM): A German View

For people with diabetes, the availability of systems for real-time continuous glucose monitoring (rtCGM) and a CGM system which requires intermittent scanning (iscCGM; also called flash glucose monitoring) is a quite helpful option. These systems provide the patients with much more information about their glucose profile over 24 hours than conventional capillary blood glucose (BG) measurements (film vs photo); however, can rtCGM/iscCGM systems be used as a (partial) replacement for blood glucose self-measurements (BGSM) from a safety point of view?

Currently, in some European countries, for example Germany, the measurement results of one rtCGM system (the Dexcom G5 and G5 Mobile, respectively) and an iscCGM system (FreeStyle Libre from Abbott; system is not on the US market yet) are intended for therapeutic decisions. This means that the therapeutic decisions patients with diabetes are making, especially determination of insulin doses, can be based on the glucose values measured in the interstitial fluid (ISF) and shown on the display of the rtCGM or iscCGM systems from a regulatory point of view. Until today patients were advised to base such decisions on BGSM and use the CGM data only as supporting information (“adjunctive” usage). This is what the regulatory agencies require, not necessarily what patients are doing in practice. Therefore, in Europe measurements made with such CGM systems can replace BGSM (“nonadjunctive use”; except for some exceptional situations, see instructions for use). Using these systems, potential interferences with drugs (eg, acetaminophen) or other substances (eg, vitamin C) have to be considered.

With the rtCGM system, still two BGSM per day are required to calibrate the system; due to the factory calibration of the iscCGM system, such measurements are not necessary.

The extent of use as replacement for BGSM and exceptional situations in which the manufacturers of the two systems regard additional BGSM as necessary are described in the instructions for use (at least those used in Germany) as follows:

1. With the Dexcom G5 therapeutic decisions may only be made (see instructions for use, Rev. 001):

2. - If the instructions were followed and the system was calibrated 12 hours after the initial calibration

3. - On the basis of the tissue glucose value, trend arrow, trend diagram, and/or attainable warnings

4. - If at least three consecutive measurement results from the last 15 minutes are available

When using FreeStyle Libre, additional blood glucose measurements have to be performed (see instructions for use, Rev. 04/2014):

- In phases of rapidly changing glucose values

- To confirm a sensor-reported hypoglycemia or impending hypoglycemia

- If symptoms do not comply with the glucose value displayed by the system

In the DIAMOND study, in which the rtCGM system (Dexcom G4 with software version 505) was used adjunctively (ie, in addition to BGSM), a clear reduction of the number of daily BGSM measurements was observed like in other studies before. ¹

This year, results from two clinical studies with the iscCGM system were presented; in February a study with patients with type 2 diabetes (REPLACE study) and in June one with patients with type 1 diabetes (IMPACT study). In the REPLACE study, the primary study aim of a reduction of HbA1c values was not reached; however, time in the hypoglycemic range could be reduced. ² In the IMPACT study, time in the hypoglycemic range (primary study endpoint) could be reduced without an increase in HbA1c values. ³ In these studies patients adjusted their insulin doses based on the measurement results provided by the rtCGM/iscCGM system.

According to the instructions for use of the rtCGM system, 93% of the measurement results in adults with this system showed a deviation of <±20 mg/dl or ±20% from results of blood samples measured with a laboratory method at glucose concentrations ≤80 mg/dl or >80 mg/dl respectively. 99% of the results were within ±40 mg/dl or ±40% respectively. The package leaflet of the iscCGM system states that at glucose concentrations <75 mg/dl and ≥75 mg/dl, 86.2% of the values were within ±15 mg/dl or ±20% respectively.

For comparison, according to the ISO standard for BGSM systems from 2003, at least 95% of the measured values have to be within ±15 mg/dl or ±20% from a reference method at glucose concentrations <75 mg/dl or ≥75 mg/dl respectively, if a blood glucose monitoring system shall be distributed in the EU. With the stricter ISO requirements from 2013 that have become active in 2016 only ±15 mg/dl or ±15% deviation at glucose concentrations <100 mg/dl or ≥100 mg/dl are allowed. Until now, no such mandatory requirements are available for neither rtCGM systems nor the iscCGM system.

Study results about the measurement accuracy of the iscCGM system were published, which however do not provide any direct statement regarding replacement of BG measurements by iscCGM. ⁴ But the study offers interesting suggestions that shall be discussed in the following. In a simulation study, Kovatchev and colleagues calculated a minimal accuracy of a mean absolute relative difference (MARD) ≤10% for CGM to reach sufficient safety when using the results for insulin dosing decisions. ⁵ However, the shortcomings of MARD have to be considered. ⁶ In the study by Bailey and colleagues using the iscCGM system, only approximately half of the sensors showed a measurement accuracy that is sufficient for insulin dosing according to these model calculations. ⁴ The user did not have any evidence if the currently used sensor is among those that reveal sufficient measurement accuracy in this regard or not. An objectification of accuracy would require a BG measurement. Furthermore, with mean MARD values ~15% the observed accuracy of the iscCGM system on the first day, that is, the day of sensor insertion, was lower than on the other 13 days of the sensor life time with a mean of ~11%. ⁴ However, measurement accuracy after the run-in-phase was markedly good when considering that this system does not require daily calibration. The results under consideration of the replacement claim already stimulated a scientific discussion.^7,8

In the IMPACT study, a randomized controlled trial, thus showing a higher degree of evidence than the simulation study, patients red their glucose values up to 15times per day with the iscCGM system but only performed BGSM 0.5 times a day. ³ There were no product-related hypoglycemic events or serious adverse events (SAEs) reported. The 13 observed product-related (sensor) adverse events (AEs) were local skin reactions at the insertion site, which in some cases were regarded as allergic reactions. The published data do not imply that the mentioned days/situations with higher MARD values have any effects relevant to safety. The question is whether and how these study conditions can be transferred to the use under everyday conditions. Therefore, data from adequately created registers would be desirable.

Also rtCGM systems show a so-called run-in phase with higher MARD values in the phase directly after sensor insertion. In the following days the MARD usually improves. For one rtCGM system overall MARD was 9% and after the first day of use it was 8% compared to a laboratory measurement. ⁹ A certain drift that leads to a deterioration of measurement quality over the wearing time is compensated by (re)calibrations in rtCGM systems. However, the calibration procedure has to be performed adequately; data about how good this works in praxis are scarce. An advantage of conventional BGSM systems is that “poor” batches of test strips can be identified through measurements with control solution; however, this is not realized by most patients. ¹⁰

For an adequate (ie, efficient) and safe replacement of BGSM by CGM measurements, an adequate training of the users is as important as a high analytical measurement quality. People with diabetes that are used to make therapeutic decisions only based on their preprandial BGSM may be overloaded by the amount of information that is provided by a rtCGM or iscCGM system. Through adequate training in the use of constantly updated glucose values and presentation of glucose curves and trend arrows, the user should be taught how to handle this information, that is, how to interpret and safely use it for therapeutic decisions. In doing so the user should also be pointed to the limitations of such systems (eg, so-called pressure-induced sensor attenuation ¹¹ or a temporary elevation of measurement results caused by intake of acetaminophen that was described for CGM systems). ¹² Training goes far beyond the description of technical properties and handling of the systems and is not to be confused with an instruction session.

It has to be communicated that CGM systems determine the glucose concentration in ISF and not directly in blood. Especially in phases of rapid glucose changes in one of these two compartments, considerable deviations compared to the other compartment may occur. Such deviations do usually not represent measurement errors but physiological differences.

Due to these reasons, attending training is obligatory to ensure that the clinical use of these systems is accessible to all patients regardless of their state of knowledge.

Based on manufacturer’s indications—as well as the other aspects mentioned—users should act with caution regarding therapeutic decisions. They should not rely exclusively on measurement results from CGM systems and at least during the transition period, conventional BGSM should be performed in parallel. This approach should help to safely use the new positive possibilities these systems offer but also to detect their limits. If these prerequisites are fulfilled, safe therapy decisions can be made.

Especially at the day of insertion of a new sensor, CGM results are to be used restrainedly. If any doubts as to the measurement quality of the sensor occur during the use of a CGM system, BGSM for clarification are strongly recommended. The user should therefore always carry a BGSM system (meter, test strips and lancing device) as backup; the iscCGM system has an integrated meter in the reader. This is also important in case of unpredictable temporary failures that cannot be influenced by the user like after sauna visits.

A further problem that arises with lasting failure of the sensor is the replacement of the medical product. In Europe, new sensors for the iscCGM system can currently only be ordered via an online portal, but this will change in near future. rtCGM sensors are ordered from the manufacturer per hotline, fax, or online, or the user has a prescription to obtain the sensors from a pharmacy/supplier. Blood glucose test strips are obtainable in every pharmacy also on weekends.

In a hearing of the FDA (Federal Drug Administration) in July 2016, where data about the use of the rtCGM system mentioned above for therapeutic decisions were excessively presented and discussed, the panel gave a positive vote about an effective and safe nonadjunctive use. ¹³ Recently this rtCGM system got approval by the FDA for nonadjunctive use, given that an adequate training and information program is established. As stated above, the iscCGM is not approved by the FDA, but recently a version that allows for retrospective display of glucose curves became available in the United States (Libre Pro).

The clinical studies performed with rtCGM/iscCGM systems and the positive vote of the FDA panel concerning the replacement of BGSM by rtCGM suggest a paradigm shift in the future. Then conventional BGSM will become obsolete in most cases, but not in general.

It is important to clarify that in contrast to BGSM (there is an ISO standard that describes requirements for the technical performance) there are no mandatory clinical or technical standards for rtCGM/iscCGM systems. A technical as well as a clinical standard for rtCGM and iscCGM systems has to be established to ensure a clear framework for the nonadjunctive use of these systems in diabetes therapy.