[MUSIC PLAYING]

KATHLEEN DUNGAN: All right, so because I participated in a lot of clinical trials and sat on a lot of consulting and advisory panels, I do have a lot of the disclosures to report. So you can feel free to review those. We all probably understand and recognize the impact that diabetes has on our patients, with the rising numbers in Ohio, across the country, and across the world.

And many of you are familiar with the decades-old research now establishing the value of intensive glycemic control, as opposed to good glycemic control. A number of studies were done-- again, mostly about a decade ago-- that really didn't demonstrate that an intense, intense control.

So A1c 6, 6 and 1/2 provided cardiovascular benefit compared to standard glycemic control. And in particular, the ACCORD trial actually demonstrated increase in risk of death. And the reasons for that were really unclear. A lot of people really think that maybe hypoglycemia might have been one of the factors, as in order to achieve intensive control, you have to use a lot of insulin-- at least at the time that these trials were conducted.

And so the real question is, are there particular strategies or glucose-lowering therapies that can confer benefit without this added risk? And so that's what I want to discuss with you all today. And before we do that, I do want to remind you that, in some populations, we might see different outcomes. And it also may depend upon the duration of time that we follow patients.

So for example, in the UKPDS, patients who had recently diagnosed type 2 diabetes were randomized to good control versus, basically, only treat if symptomatic. And in those initial results in 1997, the patients randomized to-- at the time, they designated a glucose control had improvements in microvascular complications, but not the macrovascular complications.

However, if you then follow these patients another decade or so, then you start to see some modest reductions in mortality, myocardial infarction. And so if we are going to expect that patients achieve some sort of cardiovascular benefit with glycemic control alone, then we have to actually technically follow them for a really long time.

And that's really cost-prohibitive in a number of different ways-- otherwise challenging. So as we think about what types of glucose-lowering strategies or therapies might have some cardiovascular benefit, I would just quickly remind you of the major pathophysiologic abnormalities for which we are targeting with these therapies.

We're looking at, perhaps, agents that might reduce the influx of glucose from the gastrointestinal system-- in particular, the stomach and the-- to a lesser extent, the small intestine-- so those groups of agents. The alpha cells are disregulated, in terms of glucagon-- inappropriate glucagon release. And so there are a number of agents that act in that respect.

Beta cell secretory function is impaired in type 2 diabetes. A number of agents can be used to address that. This is really quite not really well-established yet, but there is probably some autonomic nervous system or sympathetic nervous system disregulations within the central nervous system through which short-acting bromocriptine may exert its glucose-lowering effects.

Peripheral glucose uptake primarily through adipose tissue, and also through skeletal muscle, it is a target for the glitazones. And actually, while patients have glucosuria, the amount of glucosuria that they have is actually down-regulated, compared to normal individuals, in response to the degree of hypoglycemia. And so that's where SGLT2 inhibitors come into play.

And then finally, there is an increase in hepatic glucose output, and of course, metformin is a primary agent that counteracts that. All right, so a number of years ago, the ADA, also in concert with the European Association for Study of Diabetes, came up with some guidelines for management of type 2 diabetes, hyperglycemia.

And it was based on this sort of hierarchical approach, starting with metformin, and then keeping in mind some of the major potential characteristics of these drugs that might influence your treatment for second line therapies. And those classes of medications are there. And at the time, really, all of these classes of therapies were pretty much considered an option for second line therapies.

But they differ by hypoglycemia risk, how effective they are, whether there's weight gain, weight loss, or weight neutral effects-- other unique side effects-- and finally, cost. I'll come back to those guidelines a little bit later in the discussion here, but that has changed in the last year.

So in 2008, the FDA actually came up with some requirements that all new diabetes therapeutics demonstrate cardiovascular safety in either pre-marketing or post-marketing trials. And the way it works is this. So first of all, the phase 2 and 3 studies need to have adjudicated cardiovascular events. And prior to submission, there needs to be a minimum of two years of safety data.

So depending upon the hazard ratio of the cardiovascular endpoints in the studies, the drug could be approved outright or require a post-marketing cardiovascular outcomes trial. So in the initial registry studies, any of the-- any drugs that had an upper bound of that 95% confidence interval above 1.8 were not approvable.

Any drugs that had an upper bound above 1.3 were approvable, but required a post-marketing cardiovascular outcomes trial. And then any drugs, whether or not they were superior, that had the upper bound less than 1.3 were approvable without the need for post-marketing.

So that then set the stage for a number of cardiovascular outcomes trials. And in particular, I just wanted to review the MACE cloture. So prior to the FDA guidance in 2008, there were-- it was kind of just the Wild West. So people just sort of pick and choose whatever they wanted.

But that's become a little bit more standardized since then. So we have the classic three-point MACE, which is stroke, non-fatal MI, or cardiovascular death. And the four-point MACE would be adding in other acute coronary events. And then that five-point MACE is hospitalization-- adding in hospitalization for heart failure.

And all of these trials are marked by the need for a good, optimal, and equivalent management of other cardiovascular risk factors, and hopefully similar glycemic control. That's not the case in some of these trials, where it's just easier to get there with some of the drugs. So we'll talk about that.

They are typically higher risk cardiovascular-- higher cardiac risk patients to begin with in order to minimize the duration and the number of patients and expense. And they are endpoint-driven pretty much across the board. So that means that there's no-- there might be a minimum duration, but they are powered according to the number of events.

All right, so this is the landscape up till 2020, and there are a few trials that are still ongoing. I'm not going to cover all of these trials today. But they're color coded. So you can see that light blue there. These are all DPP4 inhibitors. The pink is the SGLT2 inhibitors. Some of these are just different looking-- at different populations, like heart failure or CKD.

The orange is the GLP-1's. And there is one insulin trial, which we'll talk about. There is one TZD trial, but technically, there are several, which I'll just touch on briefly-- and then a alpha-glucosidase inhibitor. So if we just start with the DPP-4 inhibitors, I'll spoil the ending-- none of them showed any cardiovascular benefit, but all of them showed cardiovascular neutrality, in terms of their primary outcomes.

So there are four drugs currently available. They have modest glucose lowering-- probably less glucose lowering potential than any of the other drugs on the market. But they may be, perhaps, better than some of the third line agents. There's no added hypoglycemia-- very well-tolerated, and they're just so-- they're very easy to use.

You do have to renally dose some of them, and they work primarily by increasing endogenous GLP-1 production. The major cautions are still a big question mark, as to whether they might cause pancreatitis. If they do, it's very, very rare. And then we'll talk about heart failure a little bit.

So again, DPP-4 inhibitors basically work through GLP-1. So GLP-1 is a hormone that is produced in response to basically food in the small intestine produced by the L cells in the small intestine. And it is degraded by this enzyme DPP-4. And so a DPP-4 inhibitor inhibits this degradation, increasing endogenous levels.

As a result, the primary effects are glucose-dependent beta cell stimulation of insulin secretion and inhibition of glucose-dependent glucagon. So this is just a summary of these trials. And so the confidence intervals, whether it's a three-point or four-point MACE, were all pretty much neutral.

Oh, I forgot to update this one. So Carmelina was also neutral. So the only one that was not placebo-controlled was Carolina, which was in which an active comparator, glimepiride, was used. On the flip, side, we would say they're all cardiovascular neutral.

However, there was this interesting observation in the trial that took a look at saxagliptin. So that was a the SAVOR-TIMI trial. And there was this peculiar increase in the odds for heart failure hospitalization. And to this day, this remains largely unexplained, although there are lots of fun hypotheses.

And then similarly, but perhaps not as convincing, there was this maybe signal in the alogliptin studies, but absolutely no signal observed with sitagliptin or-- this is a GLP-1 here-- or, for that matter, linagliptin, which is not on this particular slide. And again, this has undergone some additional questions.

We really don't know why it is, but it is in the label and you do want to probably exert some caution, and probably not use at least saxagliptin and alogliptin in patients who have heart failure, or perhaps, are really high risk for heart failure. All right, so I wanted to spend more time on the GLP-1 receptor agonists.

So there are currently six FDA-approved agents. And there is a seventh one, albiglutide, that was approved, but was withdrawn from the market basically just due to a business decision. So you can see their profiles here. The GLP-1's in-- really, I think the totality of evidence is they are the most effective glucose-lowering agent available today, including basal insulin. So it's more effective.

They don't cause hypoglycemia. They have some modest and probably the best weight loss potential of all these agents, and some mild modest blood pressure reduction. And the major downside that prohibits use, other than cost, is the gastrointestinal side effects. And in terms of cautions, again, these are very, very rare.

C cell tumors actually really haven't been reported in association with GLP-1's in humans. All the data is in rodents. But that is something that is being continually reviewed, but these drugs have been out now for 10 years, and have been under study for much, much longer than that. At least some of them are cautioned with advanced CKD, although I can-- you can use many of them down to CKD three or four.

Patients who have gastroparesis probably shouldn't take these drugs. And then there's a question mark again, with respect to pancreatitis. Again, if there is a risk-- it's very rare, but it's still in the label, and I don't typically use them, unless it's a really obviously-- obvious cause of pancreatitis that's remedial, like a gallstone.

And they can be divided into two different groups. So we have the short-acting GLP-1's, which cause intermittent GLP-1 receptor activation-- exenatide and lixisenatide-- and then we have the longer-acting GLP-1's. And don't be fooled by the dosing intervals. Liraglutide is a long-acting because it causes continuous GLP-1 receptor activation.

And therefore, the primary difference then between these two groups of drugs is that the short-acting agents have more pronounced post-prandial glucose-lowering effects-- at least at the meal immediately following administration-- and the longer-acting agents reduce both fasting and post-prandial.

And again, how do they work? We reviewed this previously with the DPP-4 inhibitors. They work primarily through beta cell stimulation of insulin production, and to a lesser extent-- or I'm sorry-- also reduction of glucagon secretion, which its primary role is stimulating hepatic glucose output. There are GLP-1 receptors in the brain, and this is responsible in large part for the appetite suppression, as even people who don't have nausea exhibit weight loss.

And on the heart, there are a lot of postulated mechanisms by which there might be cardiovascular benefit. Some of them might be directly related to GLP-1 receptors and some might be indirectly related to weight loss, blood pressure reduction. Interestingly, these agents consistently result in about a two to four-point increase in heart rate.

And so that was one of the questions early on is whether that might mitigate some of the potential cardiovascular benefits that would be observed. So again, this is-- even more clinical trials have been conducted. All of these are placebo-controlled in a different-- varying population sizes-- three or four-point MACE.

Most of them are three-point MACE. ELIXA was a four-point MACE, and used a very high cardiovascular risk group. But otherwise, they were kind of a mixture. And then we'll go through some of the outcomes. Many of them did have, in their primary outcome, a reduction in the odds of that primary outcome for-- this is liraglutide, semaglutide, REWIND-- which is dulaglutide-- albiglutide.

And a couple of the studies didn't quite make that superiority cut point, but all of them were considered cardiovascular risk neutral. And so we'll talk about a couple of-- a few of these studies here briefly. The first one that came out was the LEADER trial. This was liraglutide.

The primary endpoint was this three-point MACE. And so there was a 13% reduction in odds of that primary outcome. Cardiovascular death was also significantly reduced-- about 22%.

This is SUSTAIN 6, which is semaglutide. It's a subcutaneous semaglutide once weekly. And similarly, their primary outcome was reduced, but this time, it was driven primarily by nonfatal stroke. So the totality of these GLP-1's really suggests that it might be more of an atherosclerotic-- reduction in atherosclerotic burden.

And then we can see here non-fatal MI, which wasn't quite significant, and death was not. This is EXSCEL. This is once weekly exenatide, and this one didn't quite meet that cut here with a hazard ratio of 0.91 and the upper bound of 1.0. So it is non-inferior, but didn't quite make that superiority jump.

And part of the reason for that is this was a very large study, but it was done in a very pragmatic environment. There were fewer visits. There weren't a whole lot of really encouragement to ensure that patients were actually taking the drugs, and 40% of the patients actually had stopped treatment during the trial.

So it's really hard to demonstrate any kind of benefit, when you've got that kind of dropout. They didn't drop out from the trial, but dropped out from treatment. And so there's a lot of speculation that, perhaps, if you had done this study in a more controlled manner, that you might have seen no benefit.

And certainly, the direction was there. So these are the individual components-- death from any cause, cardiovascular disease, and heart failure. So this was one of the more recent trials, dulaglutide. This one was unique because it had the largest proportion of patients who did not have cardiovascular disease at baseline.

So in this study, only about a third had CBD. And this study also had the longest follow-up of any of the cardiovascular outcomes trials of 5 and 1/2 years. And they did demonstrate a small, but statistically significant reduction in that primary outcome, and that was driven in large part by stroke-- perhaps a little bit by cardiovascular death.

So I find it interesting that in these CV outcomes trials, that you've got these composite outcomes, but they're driven maybe by different types of events. And to me, that suggests a more global kind of effect. And then PIONEER 6-- this is the first oral GLP-1 to undergo GLP-1 receptor agonists.

This was just published I think last week in the New England Journal. And so this study wasn't powered for superiority. It was designed just to meet that pre-marketing approval standard. And it also had a fairly short follow-up. So if you use a high enough risk population, you can get enough events to satisfy the number needed to treat.

So in this case, the hazard ratio was favorable, but there just wasn't enough power to really demonstrate anything, and the confidence intervals were too wide. Nonfatal MI and stroke weren't statistically different, but curiously, death from cardiovascular disease was reduced by about half.

All right, so moving on to the SGLT2 inhibitors, there are currently four of them on the market. And a few of them have undergone CV outcomes trials. In fact, this-- the first one that was ever reported was empagliflozin. These drugs have modest A1c reduction-- kind of depends on what your baseline A1c is.

They work primarily by blocking renal glucose absorption. Notably, they are ineffective glucose-lowering agents in the presence of chronic kidney disease, so don't expect that, if someone has CKD 3 that they're going to get any glucose-lowering benefit from them.

In terms of cautions, they can cause urogenital infections. This might be anywhere from east infections-- which is fairly common-- potentially UTIs. Although, I do question that to some degree, because patients get some degree of polyuria, polydipsia, and there is a lot-- there's a pretty high frequency of asymptomatic bacteriuria in patients with diabetes.

So they can have some mild fluid and electrolyte shifts. Euglycemic DKA is very rare in type 2 diabetes, but we see it. It's primarily in people who are insulin-deficient-- so people with longstanding diabetes. Amputations were increased in the CANVAS trial-- which was that the cardiovascular outcomes trial dedicated to canagliflozin-- but not in media other studies.

The question remaining is, why might that be, and is it a real finding or not? We really don't know. I always take the conservative approach and try to be cautious in patients who are at high risk. So these agents cause modest blood pressure lowering. They cause some weight reduction, which isn't quite as good as the GLP-1's.

But the effective weight loss is additive, so if you use the two drugs together, you'll get additional weight loss-- and no hypoglycemia. So in terms of the mechanisms of action, the SGLT2 receptor is in the proximal convoluted tubule, and is a primary means by which glucose is reabsorbed from the kidneys. And so that's blocked by the SGLT2 inhibitor.

There's also an SGLT1 that, to a lesser extent, facilitates some reabsorption. And as cardiovascular mechanisms, there's a whole lot of research being done, and you could probably do a whole talk on just the mechanisms. But suffice to say it could be related to the blood pressure effects, possibly autonomic nervous system function, certainly weight loss and loss of visceral adiposity-- a number of different potential mechanisms.

It's not really clear, and it's probably multifactorial. This is the summary of the CV outcomes trials. So there's still one, ertugliflozin, that still is under-- ongoing. And all of these are placebo-controlled, pretty large numbers of patients, slightly different populations, using pretty much three-point MACE-- except for CREEDENCE, which we'll talk about, which is actually really a renal outcomes trial-- but I'm including it here because it did include some cardiovascular endpoints-- and then the follow-up periods.

So in terms of their primary outcomes, the hazard ratio was reduced for empagliflozin, canagliflozin. There are some peculiarities there that we won't really discuss a lot, but they did kind of pool these two studies together to get that endpoint.

Canagliflozin-- we'll discuss that one in more detail. And then DECLARE had two primary endpoints. One was this three-point MACE which was not statistically significant, and then they had another endpoint, which is cardiovascular death and heart failure, which was.

So if we start off with EMPA-REG-- again, this was the first one of the CV outcomes trials to actually demonstrate benefit. And so this is the breakdown. The three-point MACE reduced about 14%, and then that was largely driven by cardiovascular death. And there was this-- maybe suggestion that there may be some effect on strokes there, but that wasn't statistically significant.

In terms of some of the components-- so here's just death from any cause is reduced-- so has a ratio of 0.68-- and then hospitalization for heart failure reduced by about 1/3. So here's similar results from the canagliflozin trial. Again, this one had some peculiarities in the study design, which some people criticized, because originally, it was designed as a couple of separate trials that they put together.

But they did receive consistent results, in terms of their primary endpoint. Now, again, I want to call out the renal outcomes, because these were somewhat unexpected. So this is from EMPA-REG, and if you look at the eGFR over time, from baseline to 190 weeks, you see this concerning drop in eGFR early on within those first four weeks that then just sort of levels out throughout the rest of the study-- whereas, with the placebo group, you see this continued drops, such that there is a significant difference.

And so you don't want to be too alarmed, if you see that increase in creatinine after starting these drugs. It's kind of similar phenomenon with the ACEs and ARBs. So this was their renal outcome. This was the secondary outcome. It was reduced by almost half in the EMPA-REG study. And it was reduced significantly in the canagliflozin CANVAS trial.

So DECLARE was one of the oddballs. The MACE outcomes wasn't different, so it was similar, whereas their other endpoints-- secondary endpoints were significant. So this was death or heart failure reduced 13%. This is a renal composite outcome, again. Death from any cause wasn't reduced.

So CREEDENCE was just published a couple of months ago. We actually just reviewed this in Journal Club yesterday. And this was done in patients who had type 2 diabetes and CKD with albuminuria. So they had to have an eGFR between 30 and 90, and they had to have macroalbuminuria greater than 300.

And then their primary outcome was this end-stage renal doubling of creatinine, or renal cardiovascular death. Wait, what happened to my-- why won't it go? OK, well, some reason, my chart isn't showing up. Well, suffice it to say that, in this study, the primary outcome was reduced significantly, as well as their primary renal outcome.

And I can go to that other slide here to show you. These are the actual hazard ratios. I don't know why the chart's not showing up here. I'm sorry. Here it is. So their hazard ratio of that primary outcome was reduced by about 30%. And also, their-- they did have another renal outcome-- several renal outcomes, which were also reduced.

And if you look at the individual subgroups in CREDENCE, that's shown here. There wasn't a significant difference. The p-values for interaction were similar across the baseline categories of eGFR, and the renal specific outcome wasn't different across these categories either.

And so that really raises the question-- we know that, in these patients-- especially with the more advanced chronic kidney disease-- that they're really not getting glycemic benefit. And so the cardiovascular benefit is mediated by other things.

And the last group of drugs that I wanted-- well, actually, the second-to-last group of drugs that I wanted to review are just the TZDs. Now, these really haven't been studied systematically in post-marketing FDA study trials, but they have been studied in terms of CV outcomes, so I wanted to present them here.

The A1c reduction with these drugs, again, depends on their baseline, but it's pretty good-- about 1%, maybe 1 and 1/2%. The two drugs that are approved, pio and rosi-- pioglitazone zone is really the only one that's used anymore. There's a lot of drama in the past, related to rosiglitazone, that many of you know about.

They're primarily working as insulin sensitizers. They can increase the risk of heart failure or heart-- at least hospitalization for heart failure about two-fold. They can cause edema. They can cause weight gain, just general subcutaneous fat, and not so much visceral fat.

They can cause fractures in postmenopausal women. There's this question as to whether it might increase the risk of bladder cancer. So these drugs, I'll call out, although they do have some concerning cautions, they're not generally first or even second line for most people. But they have the greatest durability probably of all of these agents.

There are some ongoing studies that actually are going to assess that in a little bit more detail. But at least compared to sulfonylureas and metformin, they have greater durability. They take longer to fail. They have a very, very low risk of hypoglycemia.

And one of the reasons that pioglitazone is, perhaps-- will have more cardiovascular benefit is because there's a differential effect on lipids. I won't be able to spend a lot of time on them, but those effects seem to favor pioglitazone.

Rosi may increase LDL cholesterol, although it's unclear what the ramifications of that-- because there are favorable changes in LDL particle size. All right, so the proactive trial was done in 2005-- again, before there was any sort of standardization of any kind of endpoints, really.

And so they had this really huge in point of mortality, MI, stroke, other cardiovascular events any kind of vascular interventions, amputations, and some other secondary endpoints. And so their primary endpoints did not meet a superiority but then if you look at a more traditional three-point MACE endpoint that was statistically significantly reduced over the duration of the study.

The RECORD trial was the only one that's really been done with rosiglitazone, and it was done because of this increased risk that was observed in shorter-term studies with rosiglitazone. And a really famous meta-analysis was published in the New England Journal of Medicine that prompted the FDA to actually mandate that all these CV outcomes trials be done in all of the drugs.

And that included the rosiglitazone, which had a follow-up of over five years, and didn't find a significant difference. Again, there's a whole lot of drama related to whether this-- the results of this study were really reliable. The only clear thing that we know is that the hazard ratio for heart failure was increased significantly, and again, that's across both pio and rosi.

And I will point out that these drugs don't actually cause myocardial remodeling, per se. They don't actually cause overt heart failure. They're in people who are otherwise predisposed to heart failure. So they might be people who have LVH, or otherwise are just asymptomatic.

And then the last-- actually, two more studies-- so BARI-2D was a study that compared two different-- it was a two-by-two factorial intervention. So in the first intervention, patients were randomized to prompt revascularization causation-- whether that was CABG or PCI-- or they were randomized to an insulin sensitization strategy versus an insulin provision strategy.

So in that sensitization strategy, that included TZD. And so the point is that there wasn't any difference in outcomes among those two medication strategies. There was benefit in those patients who went-- who were needing CABG-- to have CABG done sooner, rather than delayed.

And then the last TZD study was published more recently, a few years ago. This was not done in patients with diabetes, but was done in patients who had stroke or TIA, and had evidence of insulin resistance. So there's a particular test that was done called the HOMA-IR that patients had to meet that cutoff. The median follow-up was about five years, and they did demonstrate a reduction in their primary outcome, which was a combination of stroke or MI.

All right, so now getting back to the guidelines, in late 2013, ADA-- again, in conjunction with EASD-- came out with some guidelines that really tried to call out more explicitly what should be your second line drug after metformin. First of all, they consider-- they're recommending early combination, if they're A1c is more than one or two points above the target.

And if patients had atherosclerotic cardiovascular disease, then they recommended adding a GLP-1 or SGLT2 inhibitor, depending on what-- which is your preference. If they had a predominance of heart failure or early renal disease, then it was recommended to do an SGLT2 inhibitor.

Again, they're still stating if they had adequate eGFR. So that's not quite consistent with some of these trials, which actually enrolled patients with lower eGFRs, and actually so showed some cardiovascular benefit. But again, the struggles there is, are you looking for cardiovascular benefit or are you looking for glucose-lowering benefit? So this was couched within glucose lowering.

And then if a SGLT2 couldn't be used or the additional therapy was needed, then a GLP-1 could be added. Other agents demonstrating cardiovascular safety could then be added, like some of the DPP-4 inhibitors-- again, keeping in mind that, perhaps, maybe saxa and maybe alogliptin might increase the risk of heart failure.

And then you're considering other major characteristics or qualities-- hypoglycemia, weight gain, and [INAUDIBLE], which are most important for the patient. And going from there, for hypoglycemia, we have several drugs that don't cause hypoglycemia, and several third line agents, and then some advice on how to use insulin and sulfonylureas as safely as possible.

If weight gain is the major factor, then considering all these other ancillary things as well as GLP-1's and SGLT2s-- there are some third line agents that could be considered, and then trying to minimize insulin and sulfonylureas, as well as TZDs. And then finally, cost is a big issue. And to be honest, I think all of these are important for almost every patient that we see, so it's somewhat artificial to try to break it down and say one is the most important.

But perhaps, with a shared decision making approach, you can identify which is the most important for you and the patient. So in terms of Intensifying to injectable therapies, the guideline really quickly took a little bit of a turn, and the first injectable now is no longer basal insulin. It's a GLP-1.

So if they're still not at goal, then you can consider adding basal insulin. And then there are a couple of caveats. So if you're A1c is crazy high, patient's really symptomatic with hyperglycemia symptoms, or you think that type 1 diabetes might be possible, then you're going to use insulin. And also, think about using a combination straight away, if the A1c is above 10 or 2% above your target.

And then we'll get to additional glucose-lowering therapies late-- in a couple of minutes. So as we approach towards insulin, then, again, we come back to this question of, what role is hypoglycemia playing, and how can we potentially mitigate that? In the major cardiovascular outcomes trials done prior to the FDA-mandated era, severe hypoglycemia was associated with an increased odds of mortality.

But the really peculiar thing is that those odds of mortality were generally higher in patients who received standard treatment compared to those who received intensive therapy. So really, there's-- it's a very complex relationship, and some of the severe hypoglycemias that confers that mortality risk is actually probably mediated by underlying treatment adherence and underlying chronic illness, and not necessarily that treatment-induced hypoglycemia.

That's not to say that hypoglycemia is something that should be ignored at all. And one thing that's commonly misunderstood is that you can just get rid of hypoglycemia by raising the A1c target-- or the glucose targets, for that matter. In fact, that's not the case, in most patients.

So if you look at a cross-cut of a population-- this is over 27,000 patients-- the risk of hypoglycemia is fairly similar across categories of A1c for both type 2 and type 1 diabetes. Really, what was important predictor is how long the patient had been requiring insulin therapy. And by extension, that-- a surrogate of that-- at least for type 1 diabetes-- is how long they've had diabetes.

So that's really important. And as you look across therapy groups, no doubt, the more complex insulin injection regimen, the greater the risk of hypoglycemia-- and weight gain, for that matter, as well. So again, why do we see this risk? So there's a whole bunch of different mechanisms by which you might postulate that hypoglycemia could increase cardiovascular risk-- inflammation, abnormal coagulation.

The sympathoadrenal response is likely to cause an increase in catecholamines, and that could have heart rhythm abnormalities, endothelial dysfunction. All of these probably play some role. This is a case that was published a few years ago that was really interesting, because this patient had continuous glucose monitoring-- demonstrated hypoglycemia overnight.

And then you see these ECG changes-- patient became bradycardic and developed some ectopic beats. And so that's, in general, the progression of prolonged hypoglycemia is increase in risk of arrhythmias. But insulin is definitely needed, particularly over time. We see beta cell secretion-- secretory function declines, while insulin resistance tends to remain relatively stable. And so insulin is generally still needed, the longer you have diabetes.

This is just-- gives you a snapshot of the types of basal insulins currently available, ranging from shorter-acting to ultra long-acting basal insulins here. And just to demonstrate for you the differences in some of these basals, we used to call insulin glargine a flat profile, but now we know it's not really flat at all. It's just flatter than NPH.

And it doesn't last a full 24 hours in everybody. Detemir, again, flat, compared to NPH-- perhaps a little bit shorter duration than glargine. But then the newer kids on the block-- this is glargine U300-- you probably know it better as Toujeo-- does really have a relatively flatter profile compared to glargine U100, and lasts longer.

And degludec-- and it's kind of similar-- and there's some ongoing head-to-head studies to try to tease out, oh, is there some tiny difference between the two of them? But I sort of lump them together. So again, they're flatter profile, longer duration. Because of this, they tend to cause less hypoglycemia, particularly overnight. And they're more flexible, so you can give them a-- if you give them a couple hours early or a couple hours late, it really isn't going to affect the glucose control as much as you would see with glargine NPH.

So a good few cardiovascular outcomes trials were conducted. This is the ORIGIN trial. It was really done in patients with pre-diabetes, or very, very early. There's absolutely no difference between glargine and standard of care. This was the DEVOTE study, which was a comparison of glargine to degludec.

There is really no difference in any of their cardiovascular outcomes in this trial, but the really interesting thing is severe hypoglycemia was reduced by about 40%, as well as nocturnal hypoglycemia reduced by about half, with a similar A1c. Now, they did see a slightly lower fasting glucoses, and that's probably just the effect of the long-acting-- the ultra long-acting drugs.

All right, so if you're going to add one of these drugs, are you going to use them with insulin? That's really the biggest question that makes people nervous is, OK, what do I do when I'm mixing all these drugs together? I would say, in general, we're continuing metformin, because it's an insulin sensitizer and minimizes the amount of insulin that you need, and in turn, will hopefully minimize hypoglycemia.

If you're using a DPP-4 inhibitor and you're going to add a GLP-1, you want to stop the DPP-4 inhibitor, because there's absolutely no additive glycemic benefit for using both at the same time. And the H 1 is generally more effective. Sulfonylureas, generally, you're going to stop, or at least reduce the dose, when adding insulin.

TZDs-- again, you'll either stop or reduce the dose. The concern there is an increased risk of edema, and potentially weight gain. And then SGLT2 inhibitors-- generally, I would continue those, but again, I beware wary of those patients with more advanced type 2 diabetes that might be more at risk for developing DKA.

And so they should be aware of some sick day rules. There are whole articles that are published on how to manage this in patients with type 1 diabetes. But in type 2 diabetes, I would think about just those really prolonged duration. So outstanding questions-- first of all, are these class-specific effects or drug-specific effects?

That could probably only be answered by head-to-head studies with some of these that have shown disparate outcomes from one trial to another. I guess my personal bias is that I think that there are probably class effects for both the GLP-1's and SGLT2s, but I could be wrong.

What about if you use a combination of GLP-1 and SGLT2? Will you get additive cardiovascular effects? That we have no idea, although you do see additive weight loss in combinations-- shorter-term drugs-- trials, when used in combination. Another question might be, well, do we use metformin at all?

And so people who have new onset type 2 diabetes-- why should we go with metformin first, when I can-- when I know I can maybe provide additional benefit with an SGLT2 or GLP-1? I don't know the answer to that, and we don't have really data to support that either way. And might there be other populations, like lower cardiovascular risk, pre-diabetes, patients without diabetes?

So there are ongoing trials that are assessing that question, at least for SGLT2 inhibitors, in terms of patients with heart failure and CKD. And then finally, I would say, well, are these results meaningful enough for my patient? And I really think it depends upon the patient. So for example, the number needed to treat for EMPA-REG is around 60.

That number is going to be lower for patients who are at higher cardiovascular risk, or at least at higher risk of heart failure, perhaps. For the LEADER, which is liraglutide, the number's a little bit-- is about the same. For the injectable semaglutide-- a little bit lower number, but still somewhere in that 40 to 60 range.

In REWIND, for example-- so they had a larger number of patients without cardiovascular disease, so you can maybe more confidently say this-- the number you needed to treat was 60, whereas those who had cardiovascular disease only, the number needed to treat was 18. If you're limiting it to the highest risk patients with previous known cardiovascular disease, then yes, that number you needed to treat would be lower. And by reference then, your statins are 40.

All right, so just a couple take-home points, and then we'll stop-- all the drugs studied don't demonstrate any increase in cardiovascular risk, so that was the point of all these trials. Some of the therapies improved glucose control and cardiovascular risk, so that's pretty exciting, having lived through that era of trial after trial after trials showing no benefit back in the ACCORD era.

There were some unexpected findings that I just touched on very briefly, like that increase in risk of heart failure, which we can't explain. Some of it may just be that, if you do enough p-values, you're going to come up with something that you can't explain. And then finally, we should tailor the glucose-lowering therapy to address complications. So I'll stop there. Thank you.