Uh, thank you very much and good afternoon, everyone. It's a great pleasure to be here. My name is Lala Liaratna. I'm a clinical associate professor of diabetes at the Imperial College in London. And it's my great pleasure and honor to present to you today, uh, talk about the Freedom II study, a large uh trial that we, uh, concluded in the UK. So these are my disclosures, and the Freedom II study was, uh, sponsored by Abbott Diabetes Care. Now, before we go on to the trial, let me just introduce you to a case and, and then we'll come back to this case again at the end of the presentation. So here is a 55-year-old female who's living with type 2 diabetes for about 12 years. Um, her BMI is 29. Uh, unfortunately, her A1C has been raised for some time, quite high, 9.6%. Um, she struggles with, uh, finger prick testing, only checks glucose a few days before coming to the clinic, uh, generally once a day, and she's taking Humulin I, which is an intermediate acting insulin, 19 units twice a day, together with an SGLT2 inhibitor, andagliflozin. And glipizide, uh maximum dose 160 mg twice a day. Now, she's unable to tolerate metformin and GLP-1 drugs due to gastric side effects. Uh, she has declined statin therapy in the past. So next uh we see her glucose data and you will see um she's only testing the last few days on the morning and they seem pretty normal. Um, and, um, you know, in, in, in uh our UK units in millimoes, they range between 4.9 to 6.3 of indicated milligram per deciliter as well. So based on this information, um, what will you do? Um, so we have 5 options here. Would you increase the morning dose of insulin, increase the evening dose, uh, add in prandial insulin, wait for more glucose data first, and, and wait for a trial, give her a, wait and give her a trial of real-time CGM first. So these are the options. So please uh remember this case and we'll come back to this case at the end of my uh presentation of the uh of the trial results. Now, before we go on to the Freedom II study, I want to take a moment to mention this Mob study, which was done in USA. It was the first large scale randomized controlled trial to look at people treated with basal-only insulin without prandial insulin. And this study used the Dexcom G6 device, uh, included people with high A1C levels, and they found a a statistically significant difference in A1C between group difference of 0.4%, a better time in range, 59%, uh, and lower time below range. However, the study was done in uh recruited between 2018, 202019. And I think reflecting the, at the times of the study, the baseline SGLT2 use was around 9% or so. Um Right, with that background, let's now move on to the freedom 2. So we know there has been major improvement in care for people with type 2 diabetes in the recent 10 years or so, and we now have many people treated with newer agents like SGLT-2 inhibitors and GLP-1 agonists and so on. But despite these newer therapies and being on basal insulin, many people don't achieve uh current glycemic targets, and when I show you the baseline data, you will see it's quite a, this group had very high A1C levels, and um there is lack of high quality randomized controlled trial data showing the benefit of CGM in people treated with these modern therapies and basal insulin. Also, our understanding of the behavioral changes associated with CGM are limited. Um, so the purpose of the Freedom II trial was to determine whether the use of Freestyle Libre 3 CGM system improves HbA1c over 32 weeks compared to self-monitoring blood glucose in adults with suboptimal glycemia. So this is the design of the trial. Uh, this study was conducted in the UK. We recruited participants from 24 secondary and primary care sites. We also had the option for self-referral via our website. And um we needed to, they needed to be treated with basal-only insulin together with other non-insulin agents. Particularly, they need to be on SCLT2 inhibitor and or GLP-1 or dual GIP GLP-1 agonists. And have an A1C between 7.5 to 11% or so. Now, we collected their baseline data using masked CGM um at baseline, and then they were randomized in a 2 to 1 fashion, uh, to either intervention group or the control arm, which was the continuation of self-monitoring blood glucose. But the unique thing about this study is we designed the study in two distinct phases. The phase 1, the 1st 16 weeks, was more focused on self-management. We did not introduce newer therapies unless deemed essential for safety. And in the phase 2, the 2nd 16 weeks of the study, uh, the clinical team, research team was free to introduce newer therapies as supported by local and national and international guidance. We also have an optional extension phase for the control group. We don't have those results yet, so we'll be focusing on the 16 and 32 week results today. Now our primary endpoint was the between group difference in HbA1c at 16 weeks. We had a key secondary endpoint, the 32 week difference in A1C, and we obviously collected a lot of other data including CGM metrics, the changes to medications, um, we tried to measure uh uh uh the um accelerometer data, safety, patient reported outcome measures, and also qualitative interviews. So this was a large trial. This is the participant flow. We screened 469 participants, and we randomized 303, 2 to 1 randomization, 198 into the intervention arm, and 105 allocated to the control group. These are our baseline data. The mean age of the population was around 61 years, um, 67% were male, uh, mean BMI around 31, and uh mean diabetes duration around 16 years or so. Um, the total daily dose of insulin was around 34 units per day. Remember, this is all basal insulin, and in keeping with the study, um, protocol, 87% taking SGLT2 inhibitors, 86% metformin, about 1/3 sulfonylurea. Um, and if you look at the top right-hand, uh, uh, part of the, uh, graph, you will see the combinations of therapy, around 45% or so taking two other non insulin agents, around 35% or so taking 3 other non-insulin agents. And we had, uh, 26% taking uh GLP-1 or dual GIP GLP-1 agonist, and 19% taking both SGLT2 and GLP-1. Uh, moving on to baseline glycemia, the baseline A1C was quite high. Uh, it was 73 millimor per mole, and the mean A1C 0 8.8%, and you can see the two arms, two groups are very nicely balanced, uh, with identical A1C at baseline, and they were undertaking about two finger prick tests per day. Now, in terms of CGM data, you can see that timing range was only around 40% at baseline. Uh hypoglycemia was very low, less than 1% time below range, but quite concerningly, around a quarter of the glucose levels, 24.9%, were actually in the very high glucose range. Level two hyperglycemia, greater than 13.9, no more than 250 mg per deciliter. Let's now look at the key results. Here we're looking at the primary outcome and the secondary outcome. The the HbA1c in the intervention group improved from 8.8 to 8.1, and in the control group, 8.8 to 8.6, and leaving with a between group difference of 0.6%, which was obviously highly clinically and statistically significant. And the second phase of the trial, the intervention arm improved further from 8.1 to 7.8. The control group also improved up to 8.3. Still, we did have a significant between group difference of 0.5% or so. Again, both clinically and statistically highly significant. Now looking at HbA1c data in a uh uh uh a slightly different way here, we're looking at the percentage of people who had a reduction of at least 1%, 0.5%, or 0.35%. For example, you can see if you take the middle area, the 0.5% or more reduction, 65% of the intervention group had at least 0.5% HbA1c reduction compared to 27% in the control group. And you can see for all three different A1C magnitudes, significantly more people achieve this target at 4 months, and at 8 months, similar trend was noted. And all of these are statistically significant in favor of the intervention. Let's now look at some of the sensor-based metrics, and here we're looking at the all-important timing range, um, and you see at baseline, around 40% intervention I'm 39.7. At the end of the 1st 4 months, it improved 54.0, and then at the end of the second phase, it further improved to 60.2%. So basically starting around 40%, going up to 60% in the intervention arm. And if you look at the between group difference between the two, there's around 10.4 and 10.6% points for the two phases of the trial. Which is roughly about 2.5 hours more spent in the target glucose range with the use of the sensors. This is a very interesting slide. This is the speed at which the timing range improved, and here you are looking at the intervention group. You will see that very significant increase in timing range at the very beginning, within the first two weeks of the trial, the timing range growing from around 40% to over 50% in the intervention group, remaining stable after that until the phase two when there's further increase in the timing range. Uh, during the second phase of the trial. So it really shows the power of information. People were able to use the glucose data to optimize their therapy very quickly before any intervention from the clinical or research team. Here we're looking at the time above range. On the left, you see time above 180, on the right, you'll see time above 250, um, and uh you'll see at both 4 months, uh, as well as the 8 months, there is significant reduction in hyperglycemia, um, with the intervention. If you look at those, uh, orange bars starting from around 60% down to around 40%. Starting from 25% down to 11.4%, so quite notable reductions in hyperglycemia and all these are highly statistically significant. This is time below range, as you will recall, our time below range was quite low at baseline, and we did not increase those time below ranges. In other words, the improvements were seen without increasing the hypoglycemia. There are a few other uh CGM metrics. Uh, this is the glucose management indicator, and again, it's very nice to see a very comparable difference in GMI compared to lab A1C around 0.5%. I mean glucose was also lower, uh, standard deviation was lower, but the CV of the glucose was not different between the two groups. Um, here we're looking at the sensor-based metrics for the day and nighttime. If you look at the timing range, um, you see both daytime and nighttime. Uh, there is significantly better timing range with the intervention, but the difference between the control group and the intervention group was much bigger at nighttime. Uh, similarly, time above range, a bigger reduction in hyperglycemia at nighttime compared to daytime. They both were, both were significant, but bigger difference at night for the above 10. But when it comes to about 13.9 or 250, a comparable reduction in both day and nighttime. Uh, time below range is very low and no different either day or night, and mean glucose was also, uh, comparably reduced in both day and nighttime. What you see here is the uh glucose profile on the left control group, you can see it's much more wider and the median glucose is higher than the intervention arm with the narrow bands and lower median glucose, and this is further improved at the 8 months time point. What about the sense usage? Here, uh, we see on the top panel, um, the median sense usage. For 16 weeks, you can see very high sense usage of 97.6%, and the 2nd 16 weeks also very high, 98.2%. In other words, patients really loved this device during the trial, and they were pretty much using it all the time. Uh, in the control group, the use of finger prick testing was roughly around 2 tests per day, and that was stable, uh, throughout the period. Let's now uh look at the medication changes. Um, here, uh, we're looking at the total daily dose, uh, on the left in units on the right, uh, using kilograms per body weight, but, uh, you will see immediately that the two groups are identical, and we did not see a difference in the total daily dose of insulin either at 4 months or 8 months. Which is obviously quite interesting given that we noted significant improvements in A1C, but the story is slightly more detailed. If you look at here, we look at the people who do not change their basal insulin versus those who increase by at least 10% or so, and those who decrease by 10% or so. You'll see interestingly, around 17, you know, close to 18% of people in the intervention group actually reduced the insulin. Uh, and 42% no change. And if you look at the eight month time point, even more people reduced their insulin 26.8. So it's quite interesting. There is a, a, you know, there is, uh, although if you take the total doses, they were no different, but there are differences in those who increase and decrease the insulin during the study. Uh, what about those starting bolus insulin? So at base, at 4 months, remember, we only introduced, it was safe to do, it was a safety reason. Unfortunately, we had 3 protocol deviations in the intervention group, as 2 people were passed out at one for safety, uh, but, you know, this sort of thing happened in large studies, but, uh, really only very small percentage, 97% did not start any new insulin. At the 8 month time point, yes, more people in the intervention group did start Py insulin, but you will remember the total daily dose was no different, meaning the CGM did help people to achieve more balance of basal and bolus insulin. Uh, what about non-insulin glucose lowering medications? You can see again, 97%, uh, in the 1st 4 months did not start any new non-insulin medications, and it's uh identical between the two groups. And the 2nd, 8 months also, uh, the two groups were identical with very similar percentage of people starting non-insulin glucose lowering medications. Now, this slide shows you the different groups of medications. You see SGLT2. Metformin, sulfonylua, GLP-1, and so on. So if it's newly started, it's in green color, uh, when everything stopped in red. So what you see, uh, first of all, two messages. Firstly, the two groups are almost identical, nothing statistically different. Uh, in terms of starting new therapies, there were a little bit more people starting on GLP-1, GIP agents, actually numerically slightly more in the control group, but not significant. And few reds, meaning few people stops off an aura, presumably due to hypoglycemia related things. When we look at some patient reported outcome measures, and here we're looking at the glucose monitoring satisfaction survey, uh, you can see the total score, the openness, emotional burden, behavioral burden, and worthwhileness, all the subskills, they were significantly better with the intervention compared to the control. Uh, hypoglycemia confidence was much better than those using the sensors, but the hypofe was no different. We also did a qualitative interviews with participants, and this was led by Profk Bernard Kelly, and you can see at the baseline, we heard the struggle of testing, uh, finger prick testing and so on. Uh, at the end of the 8 months, uh, we really found, um, you know, the main thing that people commented was that the sensors really helped them to identify the impact of food. And behavior on their glucose levels and some of the foods that they thought were healthy were actually not so healthy, and they were able to modify their behavior in order to achieve better, better control. In terms of safety, uh, the, the generally trial the product was very safe. Uh, we had one, death unfortunately due to unrelated ischemic heart disease in the intervention arm, 2 severe hypoglycemic episodes in the control group. Uh, apart from that, there were no other unanticipated device related adverse events or ketosis or hyperosmer coma and so on. The study was we recently published a paper in The Lancet Diabetes and Endocrine, within the last month or so. This paper is free to for you to access, and I'd like to pay tribute to all the investigators that contributed to this trial to make this a success. Now let's go back to our case, and here is back to the case, 55-year-old female living with type 2 diabetes for 12 years with chronically elevated HbA1c, treated with twice daily basal insulin, MA gliflozin and glipizide. And these were the glucose data that you saw at the beginning, the finger prick tests are quite a few, and, and they are mostly normal in the morning. So, what we did, we did um give this person a Freestyle Libre 3 device, and um what you see on the, on the next slide is the 1st 2 weeks of the sensor data showing yes, in the morning, the glucose levels are good, but during the day they progressively go up, particularly spiking after the evening meal or so, and then the glucose slowly comes down through overnight. So the part of the, the, the, the patient used the Libre 3 for 4 months and they gained the insight into the impact of food on her glucose levels, made changes to her diet, avoiding chocolate, sweets, and, and eating less bread, and so on. And then she was able to actually reduce her insulin dose from 19 units to 14 units, and her timing range improved from 41% to 85%, and A1c improved from 9.6 to 6.9. You can see that the improvement in timing range in the bar here. And if you look at the AGP you can see the very clear difference in uh uh uh uh the AGP with much less high glucose level and significantly better timing range. Um, and, and this was uh sustained at the 8 month time point. Um, and further reduction of insulin from 10 in the morning to 12 and 12 at night. Um, and you can see the HbA1c 9.6 at the start, 6.9, 6.5, and 6.6 at 12 months. Uh, she was also able to, uh, lose some weight. So really impressive improvements in her glycemia with reduced insulin doses by having that information, feedback about her, uh, food and so on. So it's one example. Uh, of how sensors might be able to help people with type 2 diabetes. So, dear friends and colleagues, um, just to conclude by saying that use of real-time CGM in people living with type 2 diabetes on basal insulin led to clinically and statistically significant improvements in A1C at both 4 months and 8 months time point. Uh we observed in participant reported outcomes, improvement of the quality of life and participant reported outcomes. Uh, improvement in time and range, uh, we did not increase time below range, uh, and they were sustained during self-management as well as, uh, clinician intervention. Now, I want to pay tribute to my joint chief investigator, uh, uh, Associate Professor Emma Wilmott from University of Nottingham. Uh, our investigated up and down UK, uh, are listed here. I want to pay tribute to, uh, them all. Our collaborators particularly want to say thank you to Rich, um, for your advice at the early phase of the planning of the study, our trial steering committee, and, and participant identifying centers, and thank you, uh, very much to you for your attention. And very happy to answer any questions. Thank you.
Related Presenters
