Video Dual Ketone and Glucose CGM Monitoring for Integration with Insulin Delivery/Pump SystemsInteractive Case-Based Clinic with CGM Experts Board Play Pause Volume Quality 1080P 720P 576P Fullscreen Captions Transcript Chapters Slides Dual Ketone and Glucose CGM Monitoring for Integration with Insulin Delivery/Pump SystemsInteractive Case-Based Clinic with CGM Experts Board Overview CONTINUE TO TEST Back to Symposium Thank you everybody for joining us today. I'm excited to be here in Florence and thank you, Doctor Bergendahl for that excellent presentation. So today I'm gonna be discussing continuous glucose monitoring and integration into automated insulin delivery systems. But as we think about continuous monitoring, can it be for more than one metabolite? These are my disclosures. So as we begin, let's first consider what type one diabetes looks like across the lifespan. Here we see data from the type one diabetes Clinical Registry on 15,881 participants with average A one C plotted on the Y axis and age from Children to the elderly on the Y axis. One could consider it a portrait of type one diabetes, but in truth, this data is a character, meaning it provides a comically exaggerated representation of someone or something. And why is this the case? The truth is the depiction is vastly impacted by technology use in the top graph. We see those on multiple daily injections in black and pump users in green. The pump users have lower mean A one C across the lifespan. The same holds true in the bottom graph where we see sensor users in blue and non CGM users in the deep red. Now, let's put the whole picture together as you see here. Those using MD I and sensor in blue and pump and sensor in black have very different mean glycemic patterns observed. This would suggest that sensor therapy is critical in the management of type one diabetes. And it's important to note that this change in the mean A one C curve is noted with CGM use across all income brackets and regardless of insurance status, the same holds true when we examine the curve across different racial and ethnic backgrounds. So here we see black and non-hispanic, Hispanic or Latino and white non-hispanic. And then these three graphs were going from no technology use, pump use and then layering on CGM use. And I think it's evident the curve is flattening in each subgroup as we move to the right. So understanding the benefits of technology. Now let's take a deeper dive exploring CGM use across the lifespan. First off, what are the advantages of CGM for adults in 2008, the landmark JDFCGM trial demonstrated for participants who were 25 years and older, those who were randomized to sensor therapy, their A one C was 0.53% lower at the six month follow up as compared to those on their usual blood glucose monitoring. Since then, numerous trials have echoed these findings. This has been true for individuals with type one diabetes, regardless of their insulin delivery modality. And for individuals with type two diabetes CGM use has been found to be beneficial for those on intensive insulin therapy, those on basal insulin only and even those only taking noninsulin pharmacologic agents. What about for kids? For youth aged 8 to 14 years in the JDFCGM trial? The group as a whole showed no benefit but only 50% of the cohort wore the sensor six or more days a week. So in a secondary analysis, it was demonstrated that those who wore that sensor six or more days a week represented here in the dark gray bar, there was a drastic change in hemoglobin A one C which was statistically significant. Other studies have struggled to show benefits with CGM use in kids. First, there were first generation CGM studies conducted by direct ned in Children aged 4 to 9 and also in toddlers. And in both of these trials, the studies lasted about six months long. And by the end of six months, only 40% of each cohort was wearing the sensor six or more days a week. But the thing they found was although people weren't wearing the sensor and they didn't see any benefits. In terms of glycemia, parents were very satisfied with CGM. Use. More recently, the sense trial looked at the use of sensor therapy in kids under the age of seven and there was excellent engagement with sensor wear. People were wearing it all of the time and they saw there was a reduction in time in hypoglycemia, but no benefit was noted with time in target range. So the question becomes, is fear of hypoglycemia limiting the ability of younger kids to achieve glycemic targets. Now, let's transition to the elderly in these early studies, elderly was defined as greater than 60 years old. I, I think that may be a little bit off. But um, in this study, looking at individuals from the type one diabetes exchange who wore blinded sensors, they found in those over the age of 60 with a group of 100 and 99 individuals. The median time below range of the group was 6.3% which is equivalent to 91 minutes per day below range. In this graph, we see minutes per day on our Y axis and on our X axis, we're seeing groupings of time, less than 70 time, less than 60 time under 50. And you'll notice in each of these graphs, it's grouped based on hemoglobin. A one C of the individuals. A one C is less than seven, a one C seven to less than 8% and a one c greater than 8%. Not surprisingly, the frequency of time below range varied by A one C level. With the greatest amount of time in level one hypoglycemia being in those who had baseline A one CS under 7%. Recognizing the need to minimize hypoglycemia in this age group. The wisdom trial led by Doctor Richard Prattley sought to assess whether real time CGM use could moderate the time below range. And as you can see in this graph, we're looking at time spent below 70 on our Y axis and time during the course of the study on our x axis. You'll see that those who are randomized continuous glucose monitoring are in this bluish gray color and those randomized to standard blood glucose monitoring are in the light yellow. And even by eight weeks after sensor initiation, we see a reduction in the time spent below range in a preplanned extension phase of the study. After that 26 week. Follow up, those who were randomized to the BGM group in the darker gray were able to cross over to sensor use. Not surprisingly, we saw the uh findings from the initial trial echoed where use of sensor therapy in this extension phase led to a reduction in time below range and an increase in time in target range of 70 to 180. And for those who continued with sensor use, the benefits of sensor therapy were sustained. Finally, what about big kids? These are adolescents and young adults who you know, may qualify in the ages of 13 to 24 years old. So we know this is the group who has historically struggled the most with achieving glycemic targets as demonstrated by that graph we saw earlier with the greatest deviation in mean A one C. In the initial Jader F trial, it was noted there was a limited benefit with sensor therapy, but only 30% of the cohort was wearing the sensor six or more days a week. And obviously, if you wore the sensor, again, that group in dark gray, there was a benefit as identified by this change in hemoglobin A one C. More recently, the CGM intervention in teens and young adults RCT led by Doctor Laurie LaFell showed sensor use led to lower hemoglobin A one C dropping from 8.9% at baseline to 8.5% at the six month follow up time period. And we also demonstrated increased time and target range. So I think it's important to note that these big kids are adolescents and young adults see improvement, but we're still really falling short of the glycemic targets we have set. So clinical trials have shown the benefits of CGM across the lifespan adults. We have reduction in A one C and increased time and range older adults. We have reduction in level one hypoglycemia and improvement in gly other glycemic metrics. For Children under the age of 13, we have a reduction in level one hypoglycemia but haven't really demonstrated change in A one C or time in range. And finally, for adolescents and young adults, there's improved glycemic metrics but many are still failing to meet glycemic targets. Recent years have seen vast evolution in CGM technology with improved accuracy, factory calibrations, non adjunctive indications and longer intended duration of wear. And not surprisingly, as these changes occurred penetrance of CGM into clinical care has increased. This is demonstrated by a study looking at the type one diabetes exchange and D PV registries where we see percent CGM use on our Y axis and years on our X axis. And you could see the curves are both going up in both registries. Importantly, these registries also demonstrate that the findings that were first noted in trials hold true with real world use of CGM. So here we're looking at those in the T one D exchange registry. CGM users have the dark circle and you'll see that their A one CS are on average lower. What about for the D PV registry? The same picture holds true sensor use represented by the dark triangle leads to lower mean A one C levels. In early January. A study assessing the point accuracy of two widely used sensors, the Dexcom G7 and the Freestyle Libre three was conducted where a head to head comparison was performed. This was a multi center single arm perspective study that enrolled adults with both type one and type two diabetes. Each participant wore one of the sensors on the back of the arm accuracy was assessed by comparing sensor data to laboratory reference values for a yellow springs instrument which is a bedside glucose analyzer and a capillary blood glucose level, in order to ensure the full duration of sensor wear was covered each um individual in the study underwent a few frequent sample studies. Group one was assessed on days 15 and nine, group two had studies on day 26 and 10. So here we're looking at the accuracy in the 1st 24 hours of sensor. We comparing the sensor to that y si bedside glucose analyzer. And what you'll note is that in the 1st 12 hours, the sensors performed very similarly. However, in the 12 to 24 hour time period, we can see that that freestyle lead grade three had a lower marred indicating improved accuracy. What about when we look at the overall period during the frequent sample testing here, we can see that the MD for the Freestyle Libre three is 8.9% compared to 13.6% for the Dexcom G7. What about if we use a different metric? The percent 2020 agreement here? Again, we can see the Freestyle Libre three had a higher agreement rate at 91.4 as compared to the Dexcom G7, which was at 78.6. Importantly, while it's great to see the overall rates. It's also good to go ahead and look at the overall consecutive days of wear. And what you'll notice on this graph is the MD shows that it is consecutively improved with the freestyle Libre three over the entire duration of wear. So the freestyle Libre three sensor was more accurate than the Dexcom G7 sensor and all metrics evaluated throughout the study period. And this is the first head to head comparison of these two sensors. So in truth, there's a sense of magic when using continuous glucose monitoring for a person with diabetes, continuous glucose monitoring is like Dorothy leaving Kansas and entering the world of Oz and seeing things with so much more clarity and drenched in color. But maybe to truly unlock the potential of sensor therapy, we need to use automated insulin delivery. So as we begin, let's first discuss the anatomy of an automated insulin delivery system. Here we see a continuous glucose monitor which is reading interstitial glucose levels and feeding that information to a control algorithm which then commands the pump to either administer or suspend insulin delivery. And this changes the glucose which leads to a new sensor glucose reading and a continued cycle of alterations based on this physiologic data. Putting it in simpler terms, I would consider continuous glucose monitoring to be the heart of the operation and the algorithms are really the brains. And so what's happened in terms of the evidence behind use of automated insulin delivery. There's been a clear picture that has emerged looking at data from pivotal trials that led to device approval of the various devices that are commercially available. We see that time in target range increased by 9 to 16% over the course of these studies, which equates to 2.2 to 3.9 more hours a day. Additionally, hemoglobin A one C levels had been reduced by 0.3 to 0.5% and time below range either remains unchanged or is reduced. And importantly, in a paper that was just published um alongside with one of my younger colleagues, Elizabeth Considine, we demonstrated that real world data mirrors the findings from clinical trials. But as we think about a ID who benefits the most from this technology, understandably, those with the greatest deviation from glycemic targets have the potential for the greatest benefits. In one post hoc analysis of pivotal trial data. Looking at the control IQ system, it was found that those with hemoglobin A one C is greater than 8.5% had the greatest benefit in two separate real world analyses that relied on GM I the glucose management indicator to define baseline glycemia prior to system use. Those with the highest GM I showed the greatest benefit with A ID therapy. Not surprisingly, another group that can benefit the most are adolescents and young adults. As we saw earlier, they have the greatest deviation from glycemic targets. And finally, those transitioning from multiple daily injections show great benefit from transitioning to automated insulin delivery. Recently, a retrospective observational study including 2044 respondents who self reported mode of glucose monitoring and insulin delivery provided information on glycemic metrics, frequency of severe hypoglycemic events and also had a modified goal score recorded to assess for impaired awareness of hypoglycemia. The group was recruited from the type one diabetes exchange registering and online communities. And here are the demographics of the overall group mean age was 43 years. Three quarters of the group was female and the vast majority were white self reported. A one C was on average 6.9% and the average duration of diabetes was 26.3 years. So, what did that study found find? Notably in this self selected, highly engaged cohort, only 57% of respondents reported achieving an A one C less than 7%. Additionally, 20% of respondents reported having at least one severe hypoglycemic event in the past 12 months. This included 16% of those on A ID, 19% of those on conventional pump therapy and 23% of those on multiple daily injections. In fact, 12% of respondents indicated they had two or more severe hypos in the past 12 months. And finally, regardless of glucose monitoring or insulin delivery modality. One third of all study participants was classified as having impaired awareness of hypoglycemia. And so I think what this shows is that despite the benefits of technology issues still remain and it's not just issues with hypoglycemia with insulin deficiency. Those with type one are still at risk for the acute complication of diabetic ketoacidosis. Prevalence in those with established disease is anywhere between three and 10%. And factors that increase the risk of DK A include suboptimal engagement with care. Individuals who may not be bolusing those who may be missing long acting insulin injections or forgetting to change their insulin infusion site. Adolescents and young adults are at higher risk. Those with multiple daily injections as their mode of insulin delivery are also at higher risk. And finally, those on SGLT two inhibitors have an increased risk due to the increased risk of a near euglycemic ketoacidosis. And for those who use insulin pumps, it's well known that insulin infusion sets are the achilles heel of the system. A recent survey conducted by the type one diabetes exchange online community found that of the 700 respondents, 41% indicated they had one or more infusion set failure per month and data from the type one diabetes exchange clinical registry has shown rates of DK A are actually higher than those on MD I therapy. So when we think about things, regardless of insulin delivery method, education is essential to help avoid episodes of diabetic ketoacidosis. And what's the crux of this education? It's talking about the need to monitor for ketones. So what methods do we currently have for monitoring? Ketones? People can do blood tests where they're going to be measuring beta hydroxybutyrate levels. It provides a discrete value but we do know the cost is somewhat prohibitive being more expensive per strip. The alternative is to use urine ketone measurements. Here, we're assessing aceto aetate. It's a Colom meric test. So it's a very, um subjective in terms of how we quantify it. And with these urine ketone strips, it's relatively affordable, but they do run the risk that you don't throw them out after their expiration date and you may get unreliable results. But how often are people checking for ketones? And so this is data that came from the type one Diabetes Clinical Registry. And specifically the question here is, did people monitor for ketones when they had nausea or vomiting? So in the X axis, we can see age cohorts and on the Y axis percentage of the total group. So you'll see that in kids, nearly half of those under age 13 reported that if they had vomiting, they were always checking their ketone levels. What about for adults? Here? We're looking at those over the age of 26. The exact opposite is true. More than 50% say that they never check for ketones. This is the same data from the type one diabetes exchange. But here, the question asked was how often do you check for ketones if your blood sugar is over 300. Again, for those over age 26 we see more than half of the group says they never check when we talk about the kids. There's still who some who check, but it's much more variable. So we know there needs to be a lot more work done to encourage people to check for ketones and the timing that they need to do it. So the question becomes, is there a potential for other methods to go ahead and monitor for ketones? And we know that there's continuous ketone monitors that measure beta hydroxybutyrate levels in the interstitial fluid that are in development to date. A study on the feasibility of these systems has been published in that trial. There were 12 healthy participants who are on low carbohydrate diets. Each of these participants wore three ketone sensors on the back of their arms for up to 14 days. Data was compared to capillary ketone meter and there was a single retrospective calibration on this graph. We see capillary ketone monitor as compared to the sensor blood sensor ketone monitor. And what you'll note is there's a linear response over the range studied. So really suggesting this could be helpful. So what's the future of diabetes tech? The addition of a continuous ketone monitor will help keep people with diabetes safe and alert them to metabolic decompensation. Earlier in my view, it can provide providers and people with diabetes alike with courage. So in summary, based on the American Diabetes Association standards of care, we know that use of continuous glucose monitoring is beneficial and should be offered to individuals with diabetes. Additionally, we know A ID should be considered in all people with type one diabetes and that continuous ketone monitors could help alert individuals to metabolic decompensation early and potentially avoid episodes of diabetic ketoacidosis. So in our trip to Oz, our person with diabetes can get the benefit from the sensor as the heart of the operation, the brains of the algorithm and in the future, the courage from continuous keto monitors and we as health care providers are like toto we are just along for the ride, but together we can travel down the yellow brick road to the city of Oz and increase our time and range and improve outcomes for people living with type one diabetes. So now what I'd like to do is share a case study that might help us put into perspective how some of this comes into clinical practice. So I wanna talk about a young woman who I recently saw in the hospital back in January. She's a 10 year old with type one diabetes on an automated insulin delivery system. She was diagnosed with diabetes in April of 2022 and her parents are very engaged with her care and she presented on January 19th with hyperglycemia, lethargy, nausea and multiple episodes of emesis. And by report, the parents said, well, we thought she just had a viral illness. Things have been going around the house. So that must have been what's happening to her. And her initial labs showed her glucose was 743. Her ph was 7.15 bicarb 11 and her beta hydroxybutyrate was 5.94. So clearly, this young girl presented in diabetic ketoacidosis and she's admitted to the pediatric intensive care unit and started on a two bag system for correction of her acidosis. So how does she do overall with her glycemia? Let's explore the data that we saw from her last office visit, which was back in November 14th, 2023. And as I said, she's doing pretty well with her glycemic management. Her hemoglobin A one C on that visit was 7.6%. Here's data from her A GP report from her omnipod five system that she's on and you can see that overall she's doing pretty well. Her time and target range is 59%. The one thing that was concerning is her time below range was at 4% and we know consensus guidelines recommend we keep our time below range less than 4%. And so in conversation with the family, it was acknowledged that she often was overriding suggested corrections and at times stacking insulin doses, she was very frustrated by post meal hives. And so talking through things with the family, there was a plan to create greater parental oversight to go ahead and strengthen her insulin to carbohydrate ratio to encourage her not to stack her insulin. It was recommended to only give a correction every 2 to 3 hours and to follow what the pump recommends don't just add additional insulin to it. So in many ways, allowing the technology to do more for her and preventing her from feeling undue burden by trying to fix things on her own. But now let's focus back on the issue that brought her in to the hospital on this particular day. So you can see she arrived on January or she on January 18th, the issues began and let's just walk through what happened and what her CGM can tell us and how this might inform the education we use going forward. So when speaking to her family, we pulled up this data and showed that she had a scheduled site change so that you could see that happens on the afternoon, on the day before her presentation. And you could see her sensor glucose starts trending high. And what does the family do as represented here in these purple bars? They make multiple attempts to correct that glucose level. So they're potentially stacking insulin really trying to get things down from where she was. They actually stayed up overnight and continued to try to correct the glucose by giving frequent corrections. And yet you'll see that this sensor was reading high from about 930 the night prior to presentation until when she arrived in the hospital. And the family actually felt like we need to do even more. Let's go ahead and set some temporary basal rates. And despite all of this work, they weren't seeing any benefit. So she finally presents in the, er, she's diagnosed with DK A and her pump was discontinued, we went ahead and started her on the drip. And so, you know, presented on the 19th, feeling better by the 20th. And so in our clinical practice, we provide a sick day management tools to everyone in our clinic and they're personalized, which you'll see here is, it says if you have a sick day, a pump failure, prolonged high blood sugar or vomiting. Here's exactly what we want this individual to do. The date it was generated and the amount of insulin total daily insulin dose, the person has, we provide very specific recommendations. Talk about when to measure ketones, when to present to the er and how to interpret ketone levels. And most importantly, it gives people advice on exactly what to do based on the glucose level and whether or not ketones are present. And so the big thing that we recognize with this young girl's care is that despite the parents being highly engaged with her care, her ketosis went undetected during that entire time, her glucose was above target. They did not go ahead and check for ketones and they never recognized that truly, it was a site failure that led to progressive metabolic decompensation and DK A. And so the reason I think this case is so poignant is we have many individuals who have suboptimal glycemia and are at risk for DK A and so across a wide berth of um, people with diabetes, there's the potential of having continuous ketone monitors serve as a means to get us to be alerted to situations sooner. And I just like to share that, you know, as a person living with diabetes myself, I think it's important that those who live in glass houses should not throw stones. And so to be completely transparent, here's my own data. And what you'll see is that just at the beginning of February, I had a situation where I was doing quite well. And then overnight, you can see my system was giving me repeated automated corrections and I probably was getting alerts and it looks like I may have just acknowledged them and went back to sleep. And it wasn't until the early morning hours where I finally went ahead and said something is up. So I tried to give a correction through my pump. I finally woke up at six AMI finally checked for ketones. I gave a correction with an injection, discontinued my pump and ended up changing my sight two hours later and finally resolving the ketosis I myself had experienced. So the reason I think this is important is to recognize many of us need education on the importance of measuring ketones. And it would be great to have technology further reduce this burden. Thank you. Published Created by Related Presenters Jennifer L. Sherr, MD, PhD Professor of Pediatrics, Pediatric Endocrinology Yale University School of MedicineNew Haven, CT
