Video The Foundational Role of iCGM for Personalized A1c Optimization Play Pause Volume Quality 1080P Fullscreen Captions Transcript Chapters Slides The Foundational Role of iCGM for Personalized A1c Optimization Overview CLICK HERE FOR CME CREDIT Back to Symposium My name is Concha Terasa. I am a diabetologist of adults and I work as a full professor in advanced medical and surgical technology and methodology at the Department of Science University in Magna Grecia, Catanzaro, Italy. The title of the presentation is The Foundational Role of ICGM for Personalized A1C Optimization. Dear colleagues, as you know, continuous glucose monitoring system is a cornerstone in the modern management of diabetes. Uh, and the CGM um outperforms the traditional blood glucose monitoring both in, uh, clinical practice and research. However, continuous glucose monitoring systems are not just systems measuring interstitial glucose, but it's something more. A very complex uh uh devices with different features and opportunities to personalize the treatment and optimize glucose control. Here you have a list of different features and characteristics of continuous glucose monitoring system and we are going to focus on some characteristics. With some colleagues, uh, very recently, we tried to group all these characteristics and we, uh, identify three different domains. So one including essential features of CGM, major drivers of choice, and additional features. Among the essential features, we have accuracy and certification and I'm going to focus on these two characteristics. And in the next slide, I will try to convince you that it's time to rethink the way to define accuracy. As you know, we have two different accuracy, the analytical accuracy and the clinical accuracy. And analytical accuracy is assessed by a formula. Here you have the formula of the MARD, the mean absolute relative difference. And the MARD is evaluated by calculating the difference between the sensor glucose, the reference glucose. Dividing the value for the reference glucose and multiplying by 100. As you know, a CGM with a MARD equal or lower than 10% is considered a reliable, accurate, um, continuous glucose monitoring system. Then we have the clinical accuracy. In a few words, the clinical accuracy expresses the safety of our individuals with diabetes. So when they use the Uh, continuous glucose monitoring of glucose, uh, data to take an action. For example, injecting insulin or taking carbohydrates for a low value. The MARD is not really the absolute number to evaluate the accuracy of continuous glucose monitoring system and I'll show you that in the next slide. Here we have an example um related just to one individual and the 3 different cases for the same individual. Well, in the first case, you have a number, 6 number of pairs, sensor glucose and the reference glucose pair, and we have a mud equal to 9.19. Now, let's see what's happened if we add another pair in the low range. For example, between 50 and 60 mg. We have a different mark that is a 10.26. But look at the third case. In this case, we have added a new person but in the high glucose range, we have a mark that is lower than 9%. So what is the message? The mud is a strongly, I would say influenced by the reference glucose value we are considering and this makes sense if you remember the formula we use to calculate the mud because we have the reference value as denominator in the ratio. There is also another limit. Um, yes, and it, it is related to the way and what the mud define. We have just a number. The mud is in general represented as a number, as a percentage, but we have no idea about the variability around the mud. And uh if we had a variability, maybe it would be strongly influenced by the numbers of pairs. Indeed, as you well know from the statistical rules, That when you have a lot of numbers, a lot of pairs in this case, you have a more stable MARD and also the variability is very, is very low. But if you have a low numbers of pairs, of course, the stability of MARD is high, is low, and also the um variability is high. So the key message of this um slide is another one. MARD depends also on the number of the measurement. Finally, the number in general is not evaluated for the rate of change, but as we know, our individuals with diabetes use the rate of change to take an action. For example, before injecting meal bolus uh and so on. So We have to be sure that the accuracy evaluated in general for a CGM device is applicable not only for static values but also for dynamic values. Here, we have an example with two different systems, the System A and the System B. Well, when you are in the low rate of change, um, see in particular the middle of the slide. Yes, we have a quite acceptable mud and it is also similar between the system A and the system B. But what What happened when the rate of change is high in both the direction. We have a worsen of the demand, the incredible high value. Maybe we may have some devices performing better than other, but here the message is that analytical accuracy should be evaluated also for the rate of change because the rate of change is used in real life from our individuals with diabetes. Very recently, the Food and Drug Administration, and we are very happy for that, um, defining this new criteria for the ICGM. What is the ICGM? With the term ICGM we are referring to integrated continuous glucose monitoring system. That is the system connected with the insulin pump or other medical apps, for example, helping individuals to calculate the meal volus or also to take action in the routine life. In this case, the CGM must have specific characteristics and identified by the FDA. As you can see in this slide, we do not have a MARD. We have a glucose range on the left of the slide. You have a performance against the reference reading and you have the lower boundary of a 95% confidence interval. What does it mean? Just, I, I show you an example. In the low range and specifically for glucose value lower than 70 mg per deciliter, you must have more than 85% of the readings within a 5 15 mg difference between the sensor glucose value and the reference glucose value. And so you have also other indication for other glucose range, but what is interesting is that the FDA is also Identified specific characteristics for the rate of change. With some colleagues, I have been honored to be part of this advisory board. We have identified something more than the FDA. We have totally endorsed, as you can read from this manuscript, the indication from the Food and Drug Administration, but We have extended not only, uh, we have, we have extended to the whole continuous glucose monitoring system, the characteristics for the ICGM. So any CGM should have specific characteristics and measure of accuracy that are not the mud. And we proposed clinical, um, new parameters for clinical investigation and device validation. In particular, we focused For example, on the performance, um, of the sensors and accuracy at each anatomical bursite, uh, in adult or in children or adolescents. For example, we also focused on pregnant women. We must have specific measurement and specific tests evaluating the accuracy in pregnant women. And so on. We also to investigate the accuracy of CGM system when there are possible interfering substances limiting the accuracy of the reading of the interstitial glucose reading. And we have also proposed some number of paired readings, for example, for each site, that should be a minimum of 2500 for younger children or at least 10,000 for adults. So the number of the pairs, it, it looks like a huge number, but we need this information to be sure that our patients with diabetes are using a very accurate medical device in the practice. And this is not, this is not the only manuscript about this topic. Later, a new manuscript has been published by a, a very important working group. Here we have the names of the International Federation of Clinical Chemistry. These are all people working so hard in The evaluation of accuracy of continuous glucose monitor system and in vitro devices. And here you have more details, more details regarding also the clinical study we have to design for evaluating the accuracy of the CGM. Again, the distribution across the different glucose range. Also the minimum data we may have for, we may have for defining good accuracy for our system. And also the characterization of performance. They also focus on the reference range, what is the best reference methods as a comparator for the interstitial glucose value. But the final message of this manuscript is that the math cannot be reduced to a single number evaluated in a very generic situation. But what we really need is a multi-dimensional standardization of performance, as I hope to have explained by showing these last two slides. But we are lucky because we have to say that some companies, very robust companies are performing these, uh, different methods to evaluate the accuracy of medical devices, in particular, of course, the continuous supercosmic system. And here you have the results from um a manuscript recently published and uh this manuscript report to the data, reports to the data about the accuracy of a fifty-day factory calibrated continuous glucose monitoring system will improve the sensor design. Well, here you have the mud because uh uh still You know, still some people is uh Using the demand in their mind, but we have to change the mind of these colleagues. And uh here you have the different situation I have um illustrated before because here you have a number of pairs, the number of participants, and of course, the limits within, you can accept a difference between the sensor glucose value and the reference glucose value. And you have adults, you have pediatric people with diabetes, you have a different glucose range, a low glucose range, a normal glucose range, or high glucose range. And in this manuscript, you also can find the evaluation of accuracy according to the rate of change. So, We are, we have the possibility with some medical devices to have this information that of course, tell us more than the mud and the parameter user here are much more accurate in the evaluation of the mud of the sensors. In the first slide, I mentioned um the accuracy and the certification. Of course, in essential feature certification, we cannot have in Europe a medical device into the market without a CM market. So, how the the certification is uh achieved by the companies for a medical devices. It's a very complex mechanism and in this manuscript, Basil and myself, we try to uh explain all these, all the steps included in the new uh MDR that is the medical device regulation. We are going to have also an update of the MDR and I believe it will be in the next year. Anyway, we have different steps and the two main steps are the involvement of the notified body and the UDA. Uh, well, these are the two steps I, I, I would like to share with you and tell, tell you something about that. The notified body, as you know, are private agency, um, totally involved in the evaluation of the technical documentation. They do not need the evaluation by the um Expert panel. The expert panel um is mandatory for class 3 medical device, but continuous glucose monitoring are class 2B medical device. So they can ask for something to the expert panel, but the evaluation of the expert panel is not limiting the evaluation of the notified body. So they have to determine the conformity of medical, uh, of medical device with the quality management system. They have to perform a technical evaluation about the medical device and at the end, they can recommend the CE marketing for the medical device. However, no notified bodies also involved in the post-marketing clinical for low up or post-marketing, post-marketing surveillance. But unfortunately, the UAM is not just in implement. So we do not have any information in UDEME. Actually, we cannot have access to the UDEME just uh um uh local and the country um. Uh, a country committee involved in the uh regulatory of medical devices can have access to the EUDAMed. However, the Udemed is very important tool because in the Udemed, all the companies are obligated to report the data from the post-marketing surveillance or post-marketing clinical follow-up. So it is the, uh, the, the, the only, the only possibility we uh might have to know what happens after a sensor has been um uh introduced into the market. So there is a, there is still something to clarify. There is still some work to do and especially we should have the opportunity as a physician, but in general, all the stakeholders involved in the management of the medical devices should have access to the Technical documentation and also to the data provided by the companies um evaluating the accuracy of the continuous glucose monitoring. So we have a limit so far because the perceived risk benefit cannot be evaluated independently by any of the stakeholder. But again, we have a lot of people working in this direction. We have the European Diabetes Forum, the IDF, and each of these, uh, each of these different, um, panel of expert is pushing exactly in this direction. That is improve the way and to define the accuracy. So we have, as I told you in advance in the first slide, we have to rethink the way to define accuracy of medical devices, in particular, continuous glucose monitoring. And some of these colleagues also sent a question for a written answer to European Commission and they asked for a new, new way to evaluate the quality and the safety of diabetes management devices in the European market. And there is also very interesting uh podcast by uh John Pemberton. John Pemberton is very much involved in general, in the uh diabetes technology and diabetes management strategy. And I suggest you to follow this episode. the episode number 35 where John Pemberton along with Omar Moser discussed about the limits of the marg and again they commented, they proposed the new way to define the accuracy of CGM. A reliable and accurate CGM is able also to, to give us the opportunity to, to, to, to have a good research. As I, I, as I said before, continuous glucose monitoring system out Perform a blood glucose monitoring in the research as well in the clinical practice. And in this, in this meeting, in this congress, you have the opportunity to, to, to, to follow very interesting oral presentation about the opportunities. Uh, we may have from very accurate continuous glucose monitoring system. And here you have some examples to very interesting abstract demonstrating in a prosthetic study how using a CGM we may have a very robust information about the impact of the CGM itself on the occurrence of acute events such as mortality or diabetes-related uh complication. In detail, an accurate continuous glucose monitoring system. Reduces the occurrence of mortality and morbidity in individual with type 2 diabetes. And these two abset offer the same result in two different population, but the main message is the same. We can reduce the mortality and complication while using a very accurate continuous glucose monitoring system. And it is clear because we have individuals who can take the right decision um when they have a measurement on their system. For example, they can adjust better uh insulin treatment or they can also manage better, for example, hm. The diet or the carbohydrate assumption when, for example, a hypoglycemic event is announced by the continuous glucose monitoring system. And here we have another very interesting abstract by Bergensthal, uh, because this abstract is related to a total different situation that is the use of CGM in pre-diabetes. In other words, can a CGM, um, be helpful to To, uh, suggest what is going on and if there is an increased risk to develop diabetes according to the data, the answer is yes. Especially if we take into consideration that I'm in tight range. So in the time in tight range, we may calculate from a CGM. Can, um, can predict the development of pre-diabetes more than the traditional tests we are normally using in the clinical practice such as the oral glucose tolerance test or plasma glucose alone. And again, in the field of early glucose abnormality here, we have an absence from Ramzir Jan, we may have the same opportunity. Indeed, if you look at the graph for me, you're right, you see three different flat line, the different color of the line. Um, are related to different coefficient of variation, but this is not what I would like to focus on. See how the tier is flat according to different A1C values and see what happens on the left side of the slide where we are considered a time in tight range. The time in tight range is superior to the time in range, you know, to exactly define the 5.7 that is the A1C level from normal, normal, normal A1C to abnormal A1C value. So again, of course, just a very accurate CGM can help us to evaluate the percentage of time in tight range because as we know and as I have also demonstrated in the previous Like the low range are very critical for the evaluation of accuracy. So especially in the low range, we must be sure if we would like to use the CGM to predict early diabetes, early glucose, sorry, abnormalities. My conclusions. We have to rethink the accuracy of continuous glucose monitoring and we have to apply not only to the ICGM but to the all CGM. We need a very accurate system because our individuals with the diabetes deserve the best when they use the system in their routine life. We have to assess the accuracy not only for static values but also in dynamic. Situation and I'm referring to the rate of change, uh, the continuous glucose monitoring can offer as opportunity to our individuals with diabetes. We need more data. We need more data from clinical trials, from real-world analysis. We need also to have the opportunity to understand the process before the CE market in Europe. So we need to Uh, be aware about the data from the companies and the detail, the technical details of a CGM. But as you can see, we have a lot of diabetes community working in this direction. And I'm sure that very soon we will have very accurate and only accurate devices in the market. And I thank you for your attention. Published September 23, 2026 Created by Related Presenters Concetta Irace, MD, PhD Full ProfessorAdvanced Medical and Surgical Technology and MethodologyDepartment of Health ScienceUniversity Magna Graecia Catanzaro, Italy
