American Diabetes Association 76th Scientific Sessions

June 10-14, 2016; New Orleans, LA; Full Report – SGLT-2 Inhibitors – Draft

Executive Highlights

This document contains our coverage of SGLT-2 inhibitors at ADA 2016. Immediately below, we enclose our themes on the category, followed by detailed discussion and commentary. Talk titles highlighted in yellow were among our favorites from ADA 2016; those highlighted in blue are new full report additions from our daily coverage.

For comprehensiveness, we have included some talks in this report that also overlap with our ADA 2016 DPP-4 Inhibitor report.


  • An intriguing new “fuel energetics” hypothesis for the mechanism of benefit in EMPA-REG OUTCOME generated plenty of discussion at this year’s ADA. The thrust of this hypothesis, outlined in a paper published in Diabetes and a presentation by Dr. Ele Ferrannini (University of Pisa, Italy), is that the shift toward ketone body production with SGLT-2 inhibitors (which has been discussed primarily in the context of increased risk of ketoacidosis) might have cardioprotective effects. In his model, the increased availability of lipid substrates with SGLT-2 inhibitors (due to increased glucagon production and decreased insulin and glucose levels) leads the liver to produce more ketone bodies, which are then taken up by the heart and act as very efficient fuel sources. The increased hematocrit (higher ratio of oxygen-carrying red blood cells to total blood volume) seen with SGLT-2 inhibitors could also lead to more oxygen being delivered to the heart. That combination of more oxygen and more efficient energy utilization could improve the heart’s contractile ability and lessen the strain on a failing heart. A separate paper and presentation by Dr. Sunder Mudaliar (UCSD, San Diego, CA) suggested that a similar hypothesis could apply to the renal protective effects seen in the trial. The basic idea is that a shift by the kidneys toward metabolism of ketone bodies rather than oxygen could help ameliorate renal hypoxia and oxidative stress, which are increasingly being recognized as key contributors to the progression of chronic kidney disease. While these hypotheses do not preclude other previously proposed mechanisms of benefit, the level of attention they received throughout the conference suggests that they will be an important part of the discussion moving forward.
    • If this hypothesis is correct, we look for the field to weigh the beneficial long-term effects of increased ketogenesis on cardiovascular/renal outcomes against the short-term increased risk of DKA. The risk/benefit assessment will likely vary depending on the patient population. Based on current evidence, it seems likely that the benefits will clearly outweigh the risks for patients with longstanding type 2 diabetes and high CV risk, as the EMPA-REG OUTCOME results are most directly applicable to this population and demonstrated no increased risk of DKA. The calculus is somewhat less clear for the broader type 2 diabetes population, but it still seems likely that the benefits will outweigh the risks given that the rates of SGLT-2 inhibitor-associated DKA in type 2 diabetes appear to be low and associated with specific precipitating factors – see our coverage of last fall’s AACE/ACE meeting on the subject for more. The situation may be trickiest in type 1 diabetes, where the risk of DKA appears to be more marked and there is unlikely to be a long-term outcomes trial to evaluate potential cardiovascular and renal benefits. Of course, much more study is needed before making any definitive statements about these tradeoffs, but there is certainly potential for clinicians to face a bit of a paradox.
  • A hypothesis-generating mediation analysis of the EMPA-REG OUTCOME results suggested that changes in hematocrit levels (presumably due to a reduction in plasma volume) could partially account for the 38% risk reduction for cardiovascular death. The covariate mediation analysis examined a host of potential factors to see if they could account for the impressive effect on cardiovascular mortality. Potential mediators were related to glycemia (A1c, fasting plasma glucose [FPG]), vascular tone (systolic blood pressure, diastolic blood pressure, heart rate), lipids (HDL cholesterol, LDL cholesterol, triglycerides), renal factors (log urine albumin-to-creatinine ratio (UACR), estimated glomerular filtration rate [eGFR]), adiposity (weight, BMI, waist circumference), volume status (hematocrit), or other (uric acid). A “change from baseline” analysis meant to determine a potential acute effect suggested that hematocrit levels and a decrease in plasma volume could explain nearly 52% of the overall effect of empagliflozin on cardiovascular death. An “updated mean” analysis meant to tease out chronic effects similarly found that hematocrit levels could account for 52% of the effect, while other volume-related factors such as hemoglobin and albumin levels could explain nearly 46% and 32% of the effect, respectively. While this interesting post-hoc analysis suggests a role for plasma volume in the cardioprotective benefit of empagliflozin, presenter Dr. Silvio Inzucchi (Yale University, New Haven, CT) emphasized that the analysis suggests that the plasma volume is only a partial mediator of the effect and other potential drivers (like ketone bodies) may not have been included in the analysis because they were not adequately measured at baseline and throughout the study
  • The fact that there are now two diabetes drugs with demonstrated renal protective effects is one of our most exciting and surprising conclusions from ADA 2016. Full renal outcomes results from EMPA-REG OUTCOME presented by Dr. Christoph Wanner and published in the NEJM demonstrated a significant 39% risk reduction for incident or worsening nephropathy with Lilly/BI’s Jardiance (empagliflozin) (HR = 0.61; 95% CI: 0.53-0.70; p<0.001). In addition, one of the most positive surprises from the LEADER results was the significant 22% improvement in renal outcomes (HR = 0.78; 95% CI: 0.67-0.92; p=0.003) with Novo Nordisk’s Victoza (liraglutide). These findings are truly profound given the current enormous unmet need for new therapies to treat chronic disease. We hope they might prompt both sponsors to consider conducting a dedicated chronic kidney disease trial for their products, comparable to Lilly/BI’s planned heart failure trials for Jardiance. We also wonder what the implications will be for ongoing and future trials of drugs being developed specifically for diabetic nephropathy – will those agents be required to demonstrate a benefit in addition to that provided by Victoza and/or Jardiance?
    • The FDA’s recent Drug Safety Communication about the risk of acute kidney injury with J&J’s Invokana (canagliflozin) and AZ’s Farxiga (dapagliflozin) could potentially complicate the discussion around the renal effects of SGLT-2 inhibitors. The warning was based on a search of the FDA Adverse Event Reporting System that found 101 confirmable cases of acute kidney injury with the two drugs between March 2013 and October 2015. Jardiance was not mentioned in the warning, and there was no signal of increased risk of acute kidney injury with Jardiance in EMPA-REG OUTCOME. This discrepancy suggests that there could be within-class differences in the effect of SGLT-2 inhibitors on the kidneys, which would be surprising given how homogenous the class is typically considered to be. Results from the ongoing outcomes trials for Invokana and Farxiga should help clarify this question; there could potentially be difficult tradeoffs for clinicians if those trials find both an increased risk of acute kidney injury and a reduced risk of long-term cardiovascular and/or renal risk with these drugs. It is too early to reach any definitive conclusions but we will be keeping a close eye on this area.
  • A session on novel therapeutics for type 1 diabetes underscored the continued interest in the use of SGLT-2 inhibitors in type 1 diabetes. In particular, we saw glycemic variability data from J&J’s phase 2 trial of Invokana (canagliflozin) in patients with type 1 diabetes. In the CGM substudy, MAGE (a measure of glucose variability) was 17 mg/dl improved over placebo with canagliflozin 100 mg and 38 mg/dl improved over placebo with canagliflozin 300 mg. As assessed by CGM, time in range (70-180 mg/dl) improved by roughly 14% placebo-adjusted, driven entirely by a reduction in hyperglycemia, with no major increase in hypoglycemia. This was some of the most compelling data we’ve seen on SGLT-2 inhibitors’ benefits on time-in-range and glycemic variability to date. Similarly, another presentation demonstrated markedly less chaotic CGM tracings with the addition of dapagliflozin (AZ’s Farxiga) to a regimen of GLP-1 agonist liraglutide and insulin in patients with type 1 diabetes.
  • A major theme at the annual TCOYD/The diaTribe Foundation forum was the need for better therapies for type 1 diabetes. In particular, Dr. Jeremy Pettus (UCSD, CA) spoke movingly about his own challenges managing his blood sugars while at ADA, despite the fact that he is well-educated, has health insurance, and has access to all of the best diabetes management tools. Regarding SGLT-2 inhibitors specifically, Dr. Pettus stated, “When I have a patient with type 1 diabetes who takes the drug, they say, ‘My life is better. Maybe I should’ve bolused four units and only bolused for three, and the SGLT-2 inhibitor picks up the slack. It takes away some of the chaos.’ That’s the real value of some of these medications.”
Table of Contents 

Detailed Discussion and Commentary

Symposium: Update from the EMPA-REG OUTCOME Trial

Introduction, Context, and Cardiovascular Outcomes

Bernard Zinman, MD (Mount Sinai Hospital, University of Toronto, Ontario, Canada)

In this presentation, Dr. Bernard Zinman introduced the symposium by reviewing the design and findings of the EMPA-REG trial. Most notably, he presented new sub-analysis data, showing that the heart failure and cardiovascular death outcomes of the EMPA-REG trial for Lilly/BI’s Jardiance (empagliflozin) were consistent across age groups. Also presented in two posters, the analyses stratified the results by three baseline age groups (<65, 65-75, ≥75 years). For cardiovascular death, the hazard ratios were 0.72, 0.54, and 0.55 for the age groups, respectively by increasing age group (p=0.484 for the treatment by age group interaction). Regarding hospitalization for heart failure, hazard ratios were 0.73, 0.66, and 0.45 for the respective increasing age groups (p=0.488 for the treatment by age group interaction). In addition, the analyses calculated the collective hazard ratios for heart failure hospitalization or cardiovascular death: 0.79, 0.59, and 0.52 for the respective increasing age groups (p=0.240 for the treatment by age group interaction). The time to cardiovascular death and heart failure hospitalization appeared to remain mostly consistent across baseline age groups. The analyses also found that reported adverse events were consistent with the known safety profile of Jardiance across age subgroups. The percentage of participants reporting one or more adverse events in the empagliflozin arm was 89.1%, 91.8%, and 90.8% in the <65, 65-75, and ≥75 years age groups, respectively. Specifically, the proportion of participants who reported events consistent with urinary tract infection (15% vs. 20% vs. 26%), volume depletion (4% vs. 7% vs. 7%), and bone fracture (3% vs. 5% vs. 5%) tended to increase with increasing age. These findings support the broad use of Jardiance by age group and hopefully provide further color on how to incorporate the EMPA-REG findings into treatment algorithms, as we look forward to further sub-analyses to provide stronger evidence on the guidance for better personalizing therapy.

Update on Microvascular Outcomes

Christoph Wanner, MD (University of Würzburg, Germany)

Dr. Christoph Wanner presented full renal outcomes results from EMPA-REG OUTCOME demonstrating a significant 39% risk reduction for diabetic nephropathy with Lilly/BI’s Jardiance (empagliflozin). The results were also published in the NEJM this morning. The hazard ratio for the main renal endpoint of incident or worsening nephropathy (progression to macroalbuminuria or doubling of serum creatinine accompanied by eGFR ≤45 ml/min/1.73 m2 or renal replacement therapy or death due to renal disease) was 0.61 (95% CI: 0.53-0.70; p<0.001). As with the cardiovascular outcomes, the effect appeared early and was sustained throughout the trial; there was no difference between the 10 mg and 25 mg doses of empagliflozin. The results were essentially the same for the combined endpoint of incident or worsening nephropathy or CV death (HR = 0.61; 95% CI: 0.55-0.69; p<0.001). Subgroup analyses demonstrated a consistent effect in subgroups divided by CKD stage, age, sex, race, diabetes duration, A1c, BMI, blood pressure, and concomitant medications. New onset macroalbuminuria was the most common event and therefore demonstrated the most robust results (HR = 0.62; 95% CI: 0.54-0.72; p<0.0001). The effect for hard renal outcomes (doubling of serum creatinine, renal replacement therapy, or death due to renal disease) took longer to appear but was also highly significant by the end of the trial (HR = 0.54; 95% CI: 0.40=0.75; p<0.001). There was no significant risk reduction for incident albuminuria in patients with normal urinary albumin levels at baseline, suggesting that the benefit is likely exerted later in the progression of CKD. There was no signal for adverse events including hypoglycemia, urinary tract infections, acute kidney injury, bone fractures, hyperkalemia, or DKA in patients with impaired kidney function, which should help support a label update to remove the current contraindication for patients with renal impairment.

  • Dr. Wanner outlined one potential mechanism for the positive renal effects, centered around a reduction in glomerular hypertension. He explained that diabetes leads to increased reabsorption of sodium and glucose in the proximal tubule and decreased delivery of sodium to the macula densa. Because the macula densa regulates the glomerular filtration rate based on the amount of sodium it senses, the abnormally low delivery of sodium leads to hyperfiltration. With SGLT-2 inhibition, more glucose and sodium reaches the macula densa, leading to an initial drop in glomerular pressure and glomerular filtration rate. While this short-term decline in GFR initially led to concerns that SGLT-2 inhibitors might be harmful to the kidneys, over the long term the restoration of tubulo-glomerular feedback and reduced glomerular hypertension appears to be quite beneficial. This model is the closest thing to a consensus hypothesis we have heard for the mechanism of renal protection with SGLT-2 inhibitors, though we also heard an intriguing alternative hypothesis related to fuel energetics earlier this week.

Further Insight into the Findings – Mediation Analysis (Results)

Silvio Inzucchi, MD (Yale School of Medicine, New Haven, CT)

Dr. Silvio Inzucchi (Yale School of Medicine, New Haven, CT) presented a covariate mediation analysis of the EMPA-REG OUTCOME results, suggesting that increased hematocrit (presumably as a reflection of plasma volume) was a significant (but still just a partial) contributor to the impressive 38% risk reduction for cardiac death with Lilly/BI’s Jardiance (empagliflozin). The analysis looked at several potential mediators of the effect related to glycemia (A1c, fasting plasma glucose [FPG]), vascular tone (systolic blood pressure, diastolic blood pressure, heart rate), lipids (HDL cholesterol, LDL cholesterol, triglycerides), renal factors (log urine albumin-to-creatinine ratio (UACR), estimated glomerular filtration rate [eGFR]), adiposity (weight, BMI, waist circumference), volume status (hematocrit), and other (uric acid). The analyses involved a “change from baseline” analysis to determine if a covariate may have an acute effect on the outcome and a “updated mean” analysis to determine if it may have a chronic effect on the outcome. The change from baseline analysis found that adjustment for hematocrit levels produced a hazard ratio of 0.791 for cardiovascular death, suggesting that decrease in plasma volume could potentially explain nearly 52% of the overall effect of empagliflozin on cardiovascular death. Adjustment for uric acid levels yielded the second largest change in effect (24%), suggesting it may also have some role, though Dr. Inzucchi noted, however, that any percent change in effect under 30%-40% is unlikely to indicate a significant major mediator of the cardiovascular death effect.

  • In the updated mean analysis of potential chronic factors, hematocrit levels once again emerged as a significant potential mediator of the effect. Adjustment for hematocrit produced a hazard ratio again of 0.791 for cardiovascular death, also a 52% change in the overall effect. Analysis for other volume-related factors such as hemoglobin and albumin also suggested a potential effect: hemoglobin levels accounted for 45.7% of the overall effect and albumin levels for 31.6% of the overall effect. In this analysis, A1c and FPG appeared to affect the outcome more than in the “change from baseline” analysis, with adjustment resulting in a 23% and 29% change in effect, respectively. Overall, however, Dr. Inzucchi concluded that hematocrit appeared to by the most likely potential mediator for the effect, warranting further study to investigate this hypothesis. He did, however, acknowledge that there could be other factors not measured in the EMPA-REG OUTCOME trial that could have mediated some of the effect on cardiovascular death.  Others have proposed that a slight increase in ketone bodies produced by SGLT-2 inhibitors may provide a more efficient fuel source for the heart and that this may be an important mediator as well.


William Herman, MD (University of Michigan, Ann Arbor, MI)

Discussing both the cardiovascular and renal outcomes results, Dr. William Herman argued that EMPA-REG OUTCOME is a game-changer but one with important caveats. He stressed that the very impressive results need to be interpreted strictly in the context of the eligibility criteria for the trial, meaning they should be applied only to older patients with established CVD. He also noted that the results could reflect the reduced use of potentially harmful medications like insulin and sulfonylureas rather than a specific beneficial effect of empagliflozin. Dr. Anne Peters (USC, Los Angeles, CA) offered a similar theory yesterday with regard to the LEADER results for Novo Nordisk’s Victoza (liraglutide). It is an interesting theory in both cases, though it seems unlikely that it would explain the entire benefit in EMPA-REG OUTCOME given the lack of a hypoglycemia difference between the groups and the dramatic and specific benefits on heart failure and CV death with empagliflozin. For patients with earlier-stage type 2 diabetes, Dr. Herman suggested that selection of SGLT-2 inhibitors as a second-line therapy should still be based on the criteria outlined in the ADA guidelines, including efficacy (moderate), hypoglycemia risk (low), weight effect (beneficial), side effects (genitourinary infections), and cost (expensive). In terms of the mechanism of benefit, Dr. Herman suggested that a sum of small effects (on A1c, weight, blood pressure, etc.), decreased plasma volume, renal effects, or alternative fuel sources are all plausible hypotheses. Several of these issues were discussed in more detail during Q&A – see below for a full transcript.

Panel Discussion

Q: The known factors like blood pressure are very important here. As Dr. Wanner pointed out, the IDNT and RENAAL studies controlled for blood pressure. The FDA requires control of blood pressure for renal endpoints. Even if the FDA didn’t care to control for blood pressure in this setting, why didn’t the sponsor? We’re left with a choice between the Cox analysis and scratching our heads.

Dr. Bernard Zinman (Lunenfeld-Tanenbaum Research Institute, Toronto, Canada): The vast majority of patients with type 2 diabetes do have hypertension and the vast majority are on two or three blood pressure medications. I think the blood pressure we reported here was very similar to other studies like this. I think this is the reality of the kind of blood pressure we can achieve in a global setting. I’m not sure whether you’re implying we should’ve intensified blood pressure therapy before randomization? When you look at the difference in blood pressure, there was no difference in the outcome.

Q: Why wasn’t it treated to target in both arms?

Dr. Zinman: It was.

Q: Not successfully.

Dr. Zinman: The target A1c is <7% and I’d like to know how many people in this audience have all their patients under 7%. We didn’t actively intervene on blood pressure or lipids; we left it to the local physicians and they did the best they could. It didn’t in any way affect the outcome.

Dr. Matthew Riddle (OHSU, Portland, OR): This was a relatively advanced CV risk population and quite a few with existing nephropathy. Was this a surprising systolic blood pressure for this population?

Dr. Christoph Wanner (University of Würtzburg, Germany): I’m quite happy with the blood pressure we achieved. Remember in the IDNT and RENAAL trials they had higher blood pressure and the committee hammered on the investigators to bring the blood pressure down. Here we are in a range where we have quite sophisticated blood pressure-lowering agents. A meta-analysis in CKD did not indicate that further lowering would have a benefit. It’s not proven in type 2 diabetes but at least in CKD that’s true.

Q: I understand the mechanism is unclear, but Dr. McGuire suggested that the mechanisms were different for LEADER and EMPA-REG OUTCOME so the combination might be beneficial. If the alternative fuel hypothesis is true, liraglutide shuts down beta-hydroxybutyrate production, so you might lose at least this part of the effect. You shouldn’t just combine them. If you think we should, we need a combined study.

Dr. Zinman: I can’t agree more. You don’t know until you actually do it. A good example is combining ACE inhibitors and ARBs. I already asked how many people in the room are using GLP-1 agonists and SGLT-2 inhibitors and 60% of hands went up. It’s being done, but you’re correct that we need to study it more. It will take awhile, but we need to look at surrogates, hemodynamic responses, and metabolic responses. Several companies are initiating combination studies. We can’t just assume it will be additive.

Q: One thing you mentioned in your slide was oxidative stress, but it wasn’t further discussed, probably because it’s hard to measure. There are ways to measure the end products of oxidative stress if samples are available, and we had two posters here showing that it was correlated with cardiovascular endpoints in the VADT. If samples are available, some investigation could be telling.

Dr. Silvio Inzucchi (Yale University, New Haven, CT): I agree. These trials teach you that you should have saved more samples. We’re somewhat restricted to what we measured at baseline. We don’t have blood left over. It’s led to a lot of speculation.

Q: It’s interesting that the renal effects of empagliflozin were maintained in patients with a GFR under 45. That’s the level at which we’re not meant to be using the drug according to the label. Was there an analysis of cardiovascular outcomes in that subgroup?

Dr. Inzucchi: There was no heterogeneity of effect based on sub-category of GFR. There was a consistent effect for heart failure hospitalization and CV death in those patients.

Q: You showed that for an eGFR <60. What about under 45?

Dr. Wanner: If you look at stage 3 CKD, the reduction in CV outcomes was similar in the two sub-categories and very consistent. CV outcomes were reduced in even stage 3b, which is eGFR <45. But the n was small.

Q: Do you think the hematocrit increase or plasma volume decrease could also explain the imbalance in stroke?

Dr. Inzucchi: One caution is that when we looked at the stroke outcomes, most of the imbalance occurred in patients who had stopped the study drug. We’re still analyzing it but it is a bit of a curious finding. A reduction in circulating plasma volume and the potential for sludging from increased hematocrit is something we’ve considered. A change in hematocrit from a mean of about 41 to about 44 shouldn’t really change plasma viscosity, but there certainly could be outliers. We are now looking at that specific question – were hematocrit or volume-related adverse events associated with stroke – so far, the answer is no.

Q: As a clinician I’m being persuaded to use this drug. Do you have advice on how to treat mycotic infections? Do you stop the drug if they occur?

Dr. Zinman: In the trial there were not a lot of recurrent infections. Patients tend to treat themselves with over the counter drugs. My own practice is that if a patient had recurrent infections as part of their history of diabetes, I wouldn’t use this class. Similarly, if they have a history of pancreatitis I tend not to use DPP-4 inhibitors. You need to individualize, but in my experience it’s not a big problem. The one patient I had with recurrent infections was a male who got balanitis and although it was treated he got a recurrence and stopped the drug. It’s generally not a problem.

Dr. Inzucchi: I agree that if there are recurrent adverse events from any medication, you need to avoid them. I treat the infections with topical antifungals.

Q: What is your explanation for the increase in creatinine at the onset of treatment?

Dr. Wanner: The increase in creatinine and drop in GFR is because of the reduction in intraglomerular pressure and hyperfiltration. It varies; it can be neutral but it can be quite substantial initially. We know from the ACE inhibitors, where we have a longer history of looking into this, that the more intraglomerular pressure or GFR drops at the beginning, the more the patient benefits long term in terms of RAS blockade. We have to look into it in terms of the SGLT-2 inhibitors, but for the moment I would assume it’s the same.

Q: Hepatic injury was statistically reduced. What was the definition and absolute numbers?

Dr. Zinman: We didn’t specifically measure hepatic injury. I really can’t give greater detail on that. I think it’s worth looking into. Weight-reducing agents can improve hepatic fat and liver enzymes, and it’s definitely going to the left, but I don’t have more details to provide.

Q: Taking your point that this may be a cardiac drug, and you can argue whether it’s really a renal drug, you may or may not need diabetes to have a benefit. Do you have any data on individual patients who had virtually no A1c response vs. a greater response? Was there any difference in outcomes? As clinicians, if we’re using this for diabetes and there’s no A1c improvement, should we withdraw or keep going because glucose is not what we’re actually treating?

Dr. Darren McGuire (UT Southwestern, Dallas, TX): That comment was a bit tongue in cheek, but I believe gauging efficacy by A1c is cutting it well short. There are many plausible beneficial effects and I don’t personally think A1c has anything to do with it.

Dr. Inzucchi: I agree. What you’re proposing is looking at data split by change in A1c. We know baseline A1c doesn’t predict the effect on CV mortality. It would be interesting to look at the impact of the delta in A1c.

Dr. Riddle: This was an interesting and thought-provoking mediation analysis. Will you do a similar analysis for the renal outcomes? Do you have a feeling whether it would show the same or different patterns?

Dr. John Lachin (George Washington University, Washington, DC): The first phase of the analysis is not complete and it will include looking at heart failure and renal outcomes.  

Oral Presentations: Novel Therapeutics in Type 1 Diabetes

Canagliflozin (CANA) Improves Glycemic Control and Reduces Glycemic Variability in Patients with Type 1 Diabetes Mellitus (T1DM) Inadequately Controlled with Insulin

Maria Alba, MD (Janssen R&D, Raritan, NJ)

Dr. Maria Alba presented a full data set from a phase 2 study investigating J&J’s SGLT-2 inhibitor Invokana (canagliflozin) in type 1 diabetes – the primary presentation of the study’s results was at last year’s EASD, but Dr. Alba’s talk included some new data demonstrating markedly reduced glycemic variability with canagliflozin. The original data presentation demonstrated a 0.3% placebo-adjusted reduction in A1c with both the 300 mg and 100 mg canagliflozin doses from a baseline of ~8%, along with weight loss and reduced insulin dose. New data in this presentation demonstrated striking across-the-board improvements in multiple measures of glycemic variability, headlined by a ~15% improvement of time-in-range driven by reduced hyperglycemia. These were some of the best data we have seen on SGLT-2 inhibitors’ benefits on time-in-range and glycemic variability in type 1 diabetes. Taken as a whole, these data make a compelling case for keeping most type 1 patients on the lower 100 mg canagliflozin dose, as the 300 mg dose appeared to have diminishing returns on efficacy but many more adverse events like mycotic infections, DKA, and severe hypo. 

  • The improvements in glycemic variability were striking, even at the lower canagliflozin dose. CGM assessments were performed in a subgroup of 89 patients. Based on these data, it may make sense for the vast majority of type 1 diabetes patients to stick with the lower dose, as the higher dose may result in more of an increase in euglycemic DKA than efficacy. 

Table: Changes in Measures of Glycemic Variability and Glucose Control


Cana 100 mg vs. placebo

Cana 300 mg vs. placebo

Daily mean glucose (mg/dl)

25.4 ± 28.7

22.4 ± 32.2

Daily glucose standard deviation by SMBG (mg/dl)

14.5 ± 15.9

16.2 ± 20.4

Glucose standard deviation in CGM substudy (mg/dl)

6.8 ± 16.8

13.8 ± 13.9

MAGE in CGM substudy

16.5 ± 36.5

37.8 ± 37.0

Change in % time spent in range (70-180 mg/dl)



  • Adverse event data reminded us that while canagliflozin has a lot of upside in type 1, there are some important potential risks to consider and manage.  
    • Hypoglycemia: There appeared to be a slight reduction in the event rate of any documented symptomatic hypoglycemia with both canagliflozin doses, although the magnitude of the effect was small. The event rate was 56.1% per patient-year in the placebo group, 50.6% in the canagliflozin 100 mg group, and 47.3% in the canagliflozin 300 mg group, with the treatment differences achieving statistical significance. However, the number of patients with severe hypoglycemia, though small, was unbalanced not in favor of canagliflozin: two patients on placebo, three patients on canagliflozin 100 mg, and eight patients on canagliflozin 300 mg. In our view, this represents yet another reason why there may be little reason to push type 1 patients onto the high dose.
    • DKA: There was a clear dose-dependent increase in the incidence of both serious and non-serious ketone-related adverse events with canagliflozin. Compared to no cases in the placebo group, five patients on canagliflozin 100 mg and seven on canagliflozin 300 mg required hospitalization for DKA, the latter representing 6% of that study group. We still believe that this potentially very serious side effect can be managed, especially now that we are learning more about it.
    • Others: Unsurprisingly, there was an increase in overall adverse events with canagliflozin driven largely by genital mycotic infections, which affected a full 21% of women in the canagliflozin 300 mg group.
  • Study design: This 18-week study aimed to assess canagliflozin’s effect (at both 100 mg and 300 mg doses) in 351 type 1 diabetes patients inadequately controlled on insulin. In a two-week pre-randomization phase, it was recommended that patients’ insulin dosages be reduced by 10%-20%; once canagliflozin or placebo therapy was initiated, insulin doses were titrated to a target. At baseline, average patient age was around 42 years, average BMI was around 28 kg/m2, mean A1c was 7.9%, and mean diabetes duration was 22 years. A little more than half of the study population used insulin pumps, and 11%-12% had prior DKA.

Questions and Answers

Q: The rates of severe hypo were relatively small. It’s dangerous to over-analyze those results. At face value, there was a quadrupling of severe hypos with high dose canagliflozin. What do you think is underpinning this?

A: There was no pattern to what caused the severe hypoglycemic events, though there is a clear numerical increase in cases with the 300 mg group. There was no specific pattern to the cause, and none of the patients lost consciousness.

Dapagliflozin as Additional Treatment to Liraglutide and Insulin in Patients with Type 1 Diabetes: A Randomized Clinical Trial of 12 Weeks

Paresh Dandona, MD, PhD (University of Buffalo, NY)

Dr. Paresh Dandona (University of Buffalo, NY) presented updated results from a single center, prospective, randomized, placebo-controlled trial of SGLT-2 inhibitor dapagliflozin (AZ’s Farxiga) as an add-on to GLP-1 agonist liraglutide (Novo Nordisk’s Victoza) and insulin in patients with type 1 diabetes. Dr. Dandona had previously shared results for the first 10 participants in the trial in a poster at ADA 2015 and results for 16 participants at the AACE/ACE meeting on DKA. The latest presentation at ADA 2016 included updated results for the 30 participants who have now completed the trial. The participants all had type 1 diabetes for at least a year and had been on 1.8 mg liraglutide plus insulin therapy for at least six months prior to study initiation. The participants were randomized to treatment with either 10 mg dapagliflozin or placebo for 12 weeks. At 12 weeks, dapagliflozin treatment produced a mean A1c reduction of 0.6% (baseline A1c=7.8%, p<0.01) and 15 mg/dl reduction in average weekly glucose concentration (p<0.05 vs. baseline, p=0.07 vs. placebo). Furthermore, the dapagliflozin-treated group experienced a mean weight loss of 1.9 kg (p<0.05). Most notably, Dr. Dandona showed CGM tracings in which dapagliflozin treated markedly improved glycemic excursions and time in range – a “taming of the shrew” as he put it. He also demonstrated that the withdrawal of dapagliflozin resulted in greater glycemic variability compared to the triple therapy and the further withdrawal of liraglutide introduced even greater glycemic chaos. The use of CGM tracings in this, admittedly small, trial further add to the bulk of evidence for the beneficial effects of SGLT-2 inhibitors on time in range for patients with type 1 diabetes. Anecdotally, we’ve heard from patients and physicians that SGLT-2 inhibitors are able to take away some of the unpredictability and smooth out glycemic excursions in type 1 diabetes.

  • In terms of safety and adverse events, Dr. Dandona noted that two participants withdrew from the trial due to DKA. In looking at the underlying factors associated with DKA, his team found that dapagliflozin therapy was associated with significant changes in urine volume and glycosuria, increases in free fatty acid, increases in hormone-sensitive lipase, and increases in beta-hydroxybutyrate. These findings are consistent with the DKA results for this trial that Dr. Dandona previously presented at the AACE/ACE meeting on DKA in October. Hypoglycemia event rates were comparable between the dapagliflozin and placebo groups.

Oral Presentations: Beyond Basal Insulin in Type 2 Diabetes—Treatment Intensification Options

Comparison Between SGLT-2 Inhibitors and DPP-4 Inhibitors Added to Insulin Therapy in Type 2 Diabetes: A Systematic Review with Indirect Comparison Meta-Analysis

Se Hee Min, MD (Seoul National University Hospital, South Korea)

Dr. Se Hee Min presented the results of an indirect meta-analysis that compared the effect of SGLT-2 inhibitors plus insulin (SGLT-2i/INS) vs. DDP-4 inhibitors plus insulin (DPP-4i/INS) in type 2 patients. The group performed a systematic review that yielded 14 studies which investigated either SGLT-2i/INSU vs. placebo/INS or DPP-4i/INS vs. placebo/INS (five and nine studies, respectively). Results of the covariate-adjusted indirect comparison using meta-regression analyses showed that SGLT-2i/INS provided greater reductions in A1c (weighted mean difference [WMD] of -0.24%; 95% CI: -0.43 to -0.05%), as well as greater reductions in fasting plasma glucose (WMD -18.0; 95% CI: -28.5 to -7.6 mg/dl) and body weight (WMD -2.38 kg [lbs]; 95% CI: -3.18 kg [7.0 lbs] to -1.58 kg [3.5 lbs]). No difference in hypoglycemia was observed with SGLT-2i/INS compared to DPP-4i/INS (RR 1.19; 95% CI: 0.78-1.82). Dr. Min remarked that in the absence of head-to-head comparisons, these results could serve as the best available evidence for selecting oral agents in patients uncontrolled on insulin. These findings are not surprising to us, as we feel that it is widely understood that the DPP-4 class is slightly inferior in efficacy, but its solid tolerability and safety profiles keep it as a mainstream treatment option.

  • Dr. Min noted that her group focused on SGLT-2 inhibitors and DPP-4 inhibitors because the drugs do not require injections and do not contribute to significant weight gain. Thus, the team views them as preferable add-on agents compared to GLP-1 agonists and TZDs. Furthermore, Dr. Min highlighted that SGLT-2 inhibitors and DPP-4 inhibitors have complementary effects to insulin with regards to hypoglycemia and weight gain.
  • To perform their systematic review, Dr. Min and colleagues searched Medline, Embase, LILACS, the Cochrane Central Register of Controlled Trials, and for randomized controlled trials before June 2015 that compared either SGLT-2i/INSU vs. placebo/INS or DPP-4i/INS vs. placebo/INS. They included only trials that were ≥12 weeks long and that measured A1c as an endpoint. The search yielded 14 trials – five SGLT-2 inhibitor studies and nine DPP-4 inhibitor studies.
  • The indirect analysis involved comparing each drug’s efficacy over placebo. After the initial evaluation yielded no differences between SGLT-2i/INSU and DPP-4i/INS, the authors performed a meta-regression that guided a subsequent covariate adjustment, which yielded the results presented above.

Questions and Answers

Q: The DPP4i/INS studies generally looked at a single dose while the SGLT-2i/INS studies usually had a lower dose and a higher dose. Did you include both doses or the maximum dose?

A: We included the data from the maximum approved dose.                                                           

Q: I’m not sure if you could do this because you would need patient-level data, but would you consider comparing the two drugs by baseline A1c category?

A: Thank you for your question. We did not stratify the patients by A1c.

Efficacy and Safety of Ipragliflozin, an SGLT-2 Inhibitor, Add-on to Insulin in Japanese Patients: Results of a Double-Blind, Placebo-Controlled Study

Hisamitsu Ishihara, MD, PhD (Nihon University School of Medicine, Tokyo, Japan)

Dr. Hisamitsu Ishihara (Nihon University School of Medicine, Tokyo, Japan) presented results from a trial of SGLT-2 inhibitor ipragliflozin (Astellas/Kotobuki’s Suglat) in Japanese patients with type 2 diabetes on basal insulin therapy. The 16-week, randomized, double-blind, placebo-controlled, parallel-group study (n=262) investigated ipragliflozin as an adjunct to basal insulin monotherapy or basal insulin/DPP-4 inhibitor dual therapy. The trial found that iprgraliflozin treatment produced an mean A1c difference of -1.07% compared to placebo (baseline A1c=8.67% in ipragliflozin group, 8.62% in placebo group, p<0.001). Participants treated with a DPP-4 inhibitor at baseline experienced greater A1c reductions than those not on a DPP-4 inhibitor (-1.2% vs. -0.84%, p=0.042 for interaction). Participants in the ipragliflozin-treated group overall also experienced a mean 40.3 mg/dl greater reduction in fasting plasma glucose (p<0.001) and a 1.07 kg (~2.36 lbs) body weight reduction (p<0.001) compared to placebo. Ipragliflozin treatment was also associated with an increase in adiponectin (0.33 µg/ml, p=0.022) and a greater decrease in C-peptide (−0.22 ng/ml, p < 0.001) compared to placebo. In terms of adverse events, more participants in the ipragliflozin-treated group experienced hypoglycemia (29.7% vs. 14.9% in the placebo group), urinary tract infection (2.3% vs. 1.1%), genital infection (4% v. 0%), and fluid/volume depletion (2.3% vs. 1.1%). Dr. Ishihara characterized all hypoglycemia episodes as mild and noted that none of the participants experienced symptoms of ketoacidosis.

Oral Presentations: Treatment and Management of Complications—Can a Dog Really Smell Hypoglycemia?

Canagliflozin (CANA) Slows Progression of Renal Function Decline Independent of Glycemic Effects

Hiddo Heerspink, PhD, PharmD (University Medical Center Groningen, Groningen, Netherlands)

Dr. Hiddo Heerspink presented results from a post-hoc analysis of a randomized, double-blind, two-year study on the effect of canagliflozin (J&J’s Invokana) on renal function in 1,450 patients with type 2 diabetes. The participants were randomized to canagliflozin 100mg daily (n=483) or 300mg daily (n=485), or to glimepiride titrated to 6-8mg daily (n=482). Reduction in A1c was comparable between the glimepiride, canagliflozin 100 mg, and canagliflozin 300 mg groups after one year (0.81%, 0.82%, 0.93%, respectively) and after two years (0.55%, 0.65%, 0.74%, respectively). Decline in annual eGFR was greater with glimepiride (3.3; 95% CI: 2.8-3.8) than with canagliflozin 100 mg (0.5; 95% CI: 0-1.0) or canagliflozin 300 mg (0.9; 95% CI: 0.4-1.4; p<0.01 for each canagliflozin group vs. glimepiride). Both doses of canagliflozin also provided greater reductions in albuminuria (measured via UACR) compared to glimepiride. The positive effects of canagliflozin on eGFR and UACR persisted when controlling for A1c, systolic blood pressure, and body weight. Thus, the authors concluded that canagliflozin may slow the progression of kidney disease independent of its glycemic effects. In closing, Dr. Heerspink noted that the CREDENCE trial is underway to assess the effect of canagliflozin 100mg/day on renal endpoints in patients with diabetic kidney disease. The presentation of positive renal outcomes results from the EMPA-REG OUTCOME trial for Lilly/BI’s Jardiance (empagliflozin) was one of the high points of ADA 2016; those results along with this and other promising analyses of phase 3 trials make an SGLT-2 inhibitor class effect on kidney disease seem more and more plausible.

  • In opening, Dr. Heerspink explained that SGLT-2 inhibition has been shown to lower A1c, body weight, blood pressure, and albuminuria. As noted above, the EMPA-REG OUTCOME trial for Jardiance not only demonstrated reductions in cardiovascular risk, but also suggested a renal protective benefit of the drug. These benefits were observed within weeks of starting therapy, indicating that they may be independent of glycemic control. Dr. Heerspink stated that because the study lacked an active comparator, it was difficult to draw conclusions.
  • This study recruited 1,450 patients with type 2 diabetes with an A1c of 7.0-9.5% who were on metformin (dose of at least 2,000 mg/day, or 1,500 mg/day if intolerant of higher dose). Baseline characteristics were similar between the three groups: average age 55-56 years, percent female 45-50%, average A1c 7.8%, duration of diabetes 6.5-6.7 years, systolic blood pressure 120-30 mmHg, diastolic blood pressure 78-79 mmHg, eGFR 89-91 ml/min/1.73m2, UACR 8.2-9.7 mg/g, and ACE/ARB use 59-62%.
  • Additional outcomes included albuminuria (UACR). A greater decrease in albuminuria was observed with canagliflozin 100 mg vs. glimepiride (difference of -31.7; 95% CI: -48.9 to -8.6) and with canagliflozin 300 mg vs. glimepiride (difference of -49.3; 95% CI: -62.2 to -31.9). This effect persisted when controlling for A1c, systolic blood pressure, and body weight; it was also observed in a subgroup of patients with diagnosed microalbuminuria. Regarding trends, albuminuria increased in the glimepiride group across the two-year study. In the canagliflozin 100mg group, it remained stable in the first year and increased in the second year. In the canagliflozin 300mg group, it decreased in the first year and returned to baseline in year two.
  • The study also measured the percent of patients who experienced a 30% or 40% decline in eGFR. Dr. Heerspink noted that the number of events was small, leading to wide confidence intervals. However, within the microalbuminuria subgroup, the percentage of patients that experienced a 30% declined was statistically significantly lower with canagliflozin 100mg compared to glimepiride (HR 0.37; 95% CI: 0.15-0.90).

Questions and Answers

Q: You have shown an effect on your renal endpoints. Your analysis here suggests that this requires long-term treatment with an SGLT-2 inhibitor to exploit the full potential benefit of the treatment. Obviously, when we talk about long-term treatment, we come to the issue of urogenital infection. Was this more common in patients with renal impairment?

A: To fully understand the effect of canagliflozin and the whole SGLT-2 inhibitor class, you need long-term trials because renal function declines steadily over time. And of course, you have to take into account the safety of the drug. We have seen that people with low eGFR have more adverse effects, but this is also seen in the placebo group. So the relative risk of the drug on adverse effects is similar between people with high eGFR and low eGFR.

Q: How frequently did infection occur in this group?

A: The data have been published already. We know that urinary tract infections occurred in 5% of glimepiride-treated patients and in 6% of canagliflozin-treated patients.

Q: These findings are of intense interest and are somewhat similar to reports of decreased cardiovascular disease risk with these agents, which appears to be independent of their dosing and their A1c-lowering effect. Since there was apparently a change in albuminuria between the groups, can you tell us anything about the mechanism?

A: Of course, this is a very important question. We have done a lot of studies on SGLT-2 inhibitors and their glucose lowering effect without knowing the mechanisms. I wouldn’t say the effect on albuminuria was a small effect; I would say it’s a big effect, especially in patients with already-established microalbuminuria. The mechanism, in my view, relates to the blockage of sodium reabsorption. If you start treatment with an SGLT-2 inhibitor, you block the SGLT-2 transporter and block sodium uptake. This leads to more sodium delivery to the distal tubules, which activates the macula densa and restores tubuloglomerular feedback, which leads to constriction of the afferent arteriole, thereby reducing glomerular pressure. Previous studies have shown that other mechanisms that reduce glomerular pressure lead to renal protection, and now we may have a new mechanism.

Q: Given the data you presented here, and what has already been published for empagliflozin, it seems that SGLT-2 inhibitors may be one of the most effective ways to combat renal impairment in people with type 2 diabetes. What about retinopathy in patients?

A: That’s a very good question. I have to go back to the data to answer that question. I can answer the question for cardiovascular endpoints. This patient population was a low-risk population, so the cardiovascular endpoints were similar between the treatment cohorts. The number of events was very low, so it was hard to draw conclusions.

Q: How does eGFR compare to direct measurement of GFR in this patient population?

A: That’s a very good question. There have been very few studies comparing measured GFR and estimated GFR in people on an SGLT-2 inhibitor. We did a study published in 2013 on dapagliflozin and GFR where we saw an acute fall in measured GFR and in eGFR. So measured GFR tracks with the eGFR endpoint.

Efficacy of Dapagliflozin-Saroglitazar Combination for Treatment of NAFLD in Young Diabetics

Kiran Pal Singh, MD (Fortis Hospital, Mohali, India)

Dr. Kiran Pal Singh’s single center, randomized, open-label study investigated the effects of combining dapagliflozin and saroglitazar (a PPAR alpha/gamma agonist) vs. dapagliflozin alone on A1c and the liver in 56 young patients with diabetes and non-alcoholic fatty liver disease (NAFLD). The patients (ages 20-35 years) had been on dietary and metformin therapy for ≥3 months and received either dapagliflozin 10 mg alone (n=28) or in combination with saroglitazar (4 mg; n=28). Transient elastography (Fibroscan) was used to measure hepatic steatosis (via the controlled attenuation parameter), as well as hepatic fibrosis (via liver stiffness). The 24-week data showed that compared to dapagliflozin alone, the dapagliflozin-saroglitazar combination led to significantly greater reductions in triglycerides and liver fat content, as well as higher rates of transaminase normalization. Dr. Pal Singh thus concluded that this drug combination may have potential benefit in NAFLD.

  • In opening his talk, Dr. Pal Singh highlighted that NAFLD commonly coexists with impaired glucose tolerance and type 2 diabetes, and is believed to be the hepatic component of the metabolic syndrome. He relayed that the major risk factors of NAFLD (dyslipidemia, type 2 diabetes, obesity, metabolic syndrome) overlap with those for diabetes. The pathogenesis of hyperlipidemia and hyperglycemia in NAFLD is multifactorial, and includes insulin resistance and increased production and retention of lipids in hepatocytes due to impaired apolipoprotein secretion or beta oxidation. NAFLD carries several risks, including increased liver-related and cardiovascular mortality, as well as the risk of cirrhosis.
  • Dr. Pal Singh highlighted that despite the risks associated with NAFLD, an optimal treatment has not yet been established. In his view, an ideal drug should aim to improve insulin sensitivity, glucose metabolism, and dyslipidemia. It should also reduce steatosis and transaminitis, and should halt the progression of liver fibrosis. He cited several drugs currently under investigation, including GLP-1 agonists, SGLT-2 inhibitors, PPAR alpha/gamma agonists, Farnesoid X receptor agonists, and leucine-metformin-sildenafil combinations. See our competitive landscape for what this market’s latest looks like.
  • PPAR alpha/gamma agonists are a promising therapy for NAFLD. PPAR alpha activation leads to fatty acid oxidation, which decreases triglycerides and increases HDL, ultimately leading to lipid lowering. PPAR gamma activation acts on adipogenesis and lipogenesis to increase glucose uptake and lower blood glucose levels. Dr. Pal Singh mentioned that while several members of the glitazar class have failed clinical trials for NAFLD due to their poor side effect profile (muraglitazar, tesaglitazar, farglitazar, aleglitazar), saroglitazar’s unique structure makes it less likely to cause weight gain and edema. Previous trials have already shown that it reduces triglycerides and lowers A1c.
  • Patients in both groups had similar demographic, clinical, and biochemical parameters: mean age of 29-30 years, gender breakdown (39-42% female), mean duration of diabetes of 2 years, BMI of 30, baseline A1c of 7.9%-8.2%, triglycerides of 374-399 mg/dl, abnormal transaminases in 61%-64% of each group, and liver fat content of 324-336 dB/m. 

Table 1: 24-week results from the study


dapagliflozin-saroglitazar combination

dapagliflozin only

p value



24 wks



24 wks



















Liver fat content








Abnormal Transaminases












Questions and Answers

Q: For your longer term study, have you considered having at least a sample of the patients in both groups undergo evaluation by MRI or liver biopsy?

A: No, we did not do any MRI scanning or biopsies. We just used the fibroscan.

Q: Did the fibroscan give you any measure of fibrosis in addition to liver fat?

A: Yes, we have done that, but we have not analyzed the data. The fibroscan gives both parameters.

Oral Presentations: Treatment Choices after Orals in Type 2 Diabetes

Empagliflozin Compared with Glimepiride as Add-On to Metformin for 4 years in Patients with Type 2 Diabetes

Afshin Salsali, MD (Boehringer Ingelheim, Fremont, CA)

Dr. Afshin Salsali provided an overview of results from a 104-week extension of the EMPA-REG H2H-SU trial in which empagliflozin (25 mg) as an add-on to metformin demonstrated mean A1c, weight, and blood pressure reductions versus the sulfonylurea glimepiride in patients with type 2 diabetes. Specifically, empagliflozin showed a small but significant reduction in A1c from baseline compared to glimepiride at week 208 (-0.18% vs. glimepiride, baseline A1c=7.92%, p=0.02). Notably, though not surprisingly, empagliflozin achieved this A1c with a much lower rate of hypoglycemia adverse events (3% of participants on empagliflozin reported confirmed hypoglycemia, compared to 28% of participants on glimepiride), and the time of onset of first rescue therapy for severe hypoglycemia was much later. Further, empagliflozin provided a 4.6 kg (~10 lbs) weight advantage versus glimepiride, which was sustained for the duration of the study extension period (baseline=83 kg, p<0.001). Empagliflozin also reduced systolic and diastolic blood pressure (-3% and -2% respectively) while glimepiride did not. As expected, on the negative side, empagliflozin was associated with a higher rate of urinary tract infection and genital infection adverse events versus glimepiride (20% and 14% vs. 16% and 4%, respectively). These results further strengthen the argument for the use of empagliflozin as a second-line therapy for type 2 diabetes over sulfonylureas. That said, we recognize that sulfonylureas are often chosen for cost reasons rather than due to their efficacy or safety profile and we hope studies such as this can put pressure on payers to better reimburse newer, safer, more effective medications.

Linagliptin (LINA) as Add-on to Empagliflozin (EMPA) and Metformin in Patients with Type 2 Diabetes (T2DM): Two 24-Week Randomized, Double-Blind, Parallel-Group Trials

Baptist Gallwitz, MD (Eberhard Karls University, Tubingen, Germany)

Dr. Baptist Gallwitz presented positive phase 3 data on the glycemic efficacy of Lilly/BI’s Tradjenta (linagliptin) as add-on to Lilly/BI’s Jardiance (empagliflozin) and metformin compared with placebo. These findings come from two 24-week randomized, double-blind, parallel group studies of Tradjenta vs. placebo, as add-on to Jardiance at either the 10 mg or 25 mg doses and metformin in participants with type 2 diabetes. Participants received metformin and either Jardiance at the 10 mg (study 1; n=352) or 25 mg (study 2; n=354) doses for 16 weeks in an open-label period, which was followed by a randomization to 24 weeks of double-blind, double-dummy treatment with additions of Tradjenta (n=126) or placebo (n=130). The findings demonstrated that at 24 weeks, the Tradjenta group resulted in statistically significant reductions in A1c compared with placebo: from a baseline A1c of ~8%, participants achieved placebo-adjusted A1c reductions of 0.32% and 0.47% in the Jardiance 10 mg and 25 mg dose groups, respectively. In addition, the proportion of participants who reached A1c levels below 7% at week 24 in the Tradjenta add-on groups was more than double that of the placebo groups (26% vs. 11% in study 1; 36% vs. 15% in study 2). There were no significant changes in body weight with either treatment group. Regarding safety and tolerability, no new signals emerged: the placebo group reported more adverse events than the Tradjenta add-on group, with three hypoglycemic events occurring in the placebo arm of study 2 (Jardiance 25 mg + metformin). Ultimately, these findings support an SGLT-2 inhibitor/DPP-4 inhibitor combination as a promising treatment option for patients who are inadequately controlled with metformin and an SGLT-2 inhibitor. Such a combination has certainly received increasing attention, as Merck has recently expressed greater excitement on the potential of a Januvia (sitagliptin)/ertugliflozin combination. We look forward to seeing longer-term data on this approach, as these data could potentially help craft more individualized guidance into treatment algorithms.

Questions and Answers

Q: What was the baseline A1c?

A: Around 8%.

Q: Did you add lina or did you switch the empagliflozin to an empa/lina combination?

A: We did not use the fixed-dose combination. We added the lina on.

Q: Do you have glucagon levels?

A: We do not have these yet.

Q: Did you perform a meal test?

A: No, we did not.

Dapagliflozin + Exenatide QW Reduced Body Weight and Improved Glucose Tolerance in Nondiabetic Obese Adults: A Randomized, Placebo-Controlled, Phase 2 Study

Jan Eriksson, MD, PhD (Uppsala University, Uppsala, Sweden)

Results from a phase 2 proof of concept study (n=50) of combination therapy with AZ’s Farxiga (dapagliflozin) and Bydureon (exenatide once weekly) demonstrated significant ~4 kg weight loss and glycemic improvements vs. placebo in patients with obesity but not diabetes. Participants in the double-blind, single-center study were randomized to receive either active treatment or double placebo for 24 weeks, followed by a 28-week open-label extension study; data from the extension study will be presented at EASD in September. After 24 weeks, the combination led to significant placebo-adjusted weight loss of 4.1 kg (baseline weight = 103-106 kg [227-234 lbs]; baseline BMI = 35-36; p=0.0007) or 4.2% (p=0.0005). As in most obesity drug trials, there was a wide range of responses, but far more patients achieved ≥5% weight loss with the active treatment than with placebo (36% vs. 4%). MRI analysis of body composition showed that almost all of the weight loss was due to loss of adipose tissue, with no significant change in lean tissue. The combination also produced a modest but significant 0.2% placebo-adjusted A1c reduction (baseline = 5.6%, p=0.0004), a significant drop in the proportion of patients with impaired fasting glucose and impaired glucose tolerance, and a significant placebo-adjusted blood pressure reduction of 6.4 mm Hg (p=0.026). Adverse events were fairly balanced, with slightly more GI side effects in the active treatment group.

  • These results are encouraging, though as noted during Q&A, the real test will be how the combination stacks up against each of its components alone. AZ is currently conducting a phase 3 study (n=660) of that comparison in patients with type 2 diabetes that is expected to complete in December 2017 (primary completion May 2016). As presenter Dr. Jan Eriksson noted, GLP-1 agonist/SGLT-2 inhibitor combinations are very appealing for obesity due to their complementary mechanisms of weight loss (calorie loss with the SGLT-2 inhibitor and reduced appetite/caloric intake with the GLP-1 agonist). The same could also be said for glycemic control, as the reduction in glucagon production with GLP-1 agonists could help mitigate the increased glucagon production that blunts some of the efficacy of SGLT-2 inhibitors. We expect AZ to focus primarily on type 2 diabetes with this combination but find the potential in obesity very interesting as well, particularly for a GLP-1 agonist with more potent weight effects like Novo Nordisk’s Saxenda (liraglutide 3.0 mg) or semaglutide.

Questions and Answers

Q: Could you clarify the timing of the last dose vs. the glucose measures?

A: A glucose tolerance test was performed at baseline and at 24 weeks. The dose was taken half an hour before the glucose tolerance test was started.

Q: Although I acknowledge this was a proof of concept trial, why didn’t it include monotherapy + placebo arms? You’re really asking whether there’s a synergistic or additive effect and this study didn’t answer that.

A: It’s a proof of concept for the combination and we showed robust weight loss vs. placebo, but I agree, we want a study against monotherapies as well. That would have increased the study size so we couldn’t have done it at a single center.

Oral Presentations: Modulators of Adipose Tissue Inflammation

SGLT2 Inhibition by Empagliflozin Attenuates Obesity-induced Insulin Resistance and Inflammation by Enhancing Fat Utilization and Macrophage Alternative Activation

Tsuguhito Ota, MD, PhD (Kanazawa University, Japan)

Dr. Tsuguhito Ota presented the results of his animal study that investigated the effects of empagliflozin on insulin resistance in obese mice given either a high-fat diet (HFD) alone or containing either 0.01% or 0.03% empagliflozin. Empagliflozin suppressed HFD-induced weight gain (16% with the 0.03% dose). The drug also increased energy expenditure – more specifically, Dr. Ota noted that the induction of UCP1 expression observed with the drug suggested that it promotes browning of white adipose tissue. In skeletal muscle, empagliflozin increased fat utilization, accompanied by elevated AMPKα and ACC phosphorylation, and increased expression of genes involved in fatty acid oxidation. Empagliflozin was also shown to improve HFD-induced glucose intolerance, hyperinsulinemia, and hepatic steatosis; furthermore, it increased insulin signaling in the mouse liver and white adipose tissue. These tissues also had reduced accumulation of both macrophages and T cells with empagliflozin, along with a change in the macrophage composition from mainly M1 (CD11c+CD206-) cells to predominantly M2 (CD11c-CD206+) cells. As we hear more obesity data on type 2 diabetes drugs, these findings point to the potential of the SGLT-2 inhibitor class in these broader indications.

Questions and Answers

Q: Some of the effects that you saw in terms of fatty acid oxidation and browning are potentially mediated by FGF21. Could that be playing a role?

A: We measured FGF21 in plasma, but it was not changed significantly, so I don’t think so.


Ertugliflozin Effectively Improves Glycemic Control as Monotherapy in Patients with T2DM (130-LB)

S Dagogo-Jack, M Davies, J Frias, G Derosa, A Darekar, K Focht, G Golm, J Johnson, D Saur, and S Terra

Dr. Dagogo-Jack and colleagues presented interim, 26-week results from the ongoing 52-week, randomized, double-blinded, phase 3 VERTIS MONO trial investigating the safety and efficacy of Merck/Pfizer’s ertugliflozin in patients with type 2 diabetes. The study recruited 461 patients with inadequate glycemic control on diet and exercise (A1c 7-10.5% with no anti-diabetic agents taken within 8 weeks of starting the study), who were randomized to placebo (n=153), ertugliflozin 5mg daily (n=156), or ertugliflozin 15mg daily (n=152). Twenty-six-week data showed that both ertugliflozin doses provided statistically significantly greater A1c reductions vs. placebo (0.99% for the 5mg dose and 1.16% for the 15mg dose); similar results were observed for fasting plasma glucose, body weight, and 2-hour post-prandial glucose. In addition, a greater proportion of patients achieved an A1c <7% in the ertugliflozin groups vs. placebo.

  • Participants who met glycemic rescue criteria received open-label metformin. The FPG thresholds for rescue were >270 mg/dl from day 1 to week 6; >240 mg/dl from week 6 to week 12; and >200 mg/dl from week 12 to 16.  The placebo group had a higher rate of patients receiving glycemic rescue (26%) compared to the ertugliflozin 5mg (2%) and 15mg group (3%).
  • The three groups had comparable baseline characteristics, with an average age of 56-57 years, % male of 54-59%, duration of type 2 diabetes of 5 years, body weight of 91-94 kg [201-207 lbs], BMI of 33 kg/m2, A1c of 8.1-8.4%, FPG of 179-181 mg/dl, baseline 2-hr PPG of 256-263 mg/dl, SBP of 130 mmHg, DBP of 78-79 mmHg, and eGFR of 86-89 mL/min/1.73m2
  • The retention rate through week 26 was 90% across the three groups and 83% for the study drug. 

Table 1: Change in A1c from baseline to 26 weeks 



Week 26

Change from baseline to wk 26















Ertugliflozin 5mg







Ertugliflozin 15mg







Table 2: Results of primary and secondary endpoints at 26 weeks. 


Ertugliflozin 5mg vs. placebo

Ertugliflozin 15mg vs. placebo



p value


p value

A1c (%)





FPG (mg/dl)





Body Weight

-1.76 kg [3.9 lbs]


-2.16 kg [4.8 lbs]


2-h PPG (mg/dl)





SBP (mmHg)





DBP (mmHg)






Odds Ratio

p value

Odds Ratio

p value

Proportion of pt with A1c <7%





* Nominal p value

# Difference in least squares means based on a constrained longitudinal data analysis model

  • Regarding adverse events, genital mycotic infections were more common in female participants receiving ertugliflozin vs. placebo. In male participants, genital mycotic infections were numerically higher with ertugliflozin vs. placebo. The incidence of UTI was numerically highest in the placebo group. Furthermore, the rates of symptomatic hypoglycemia and hypovolemia were similar between the three groups.

Effect of Ertugliflozin plus Sitagliptin on Glycemic Control vs. Either Treatment Alone in Subjects with T2DM Inadequately Controlled with Metformin (125-LB)

R Eldor, R Pratley, G Golm, S Huyck, Y Qiu, S Sunga, J Johnson, S Terra, J Mancuso, S Engel, and B Lauring

Dr. Eldor and colleagues presented interim, 26-week results from an ongoing 52-week randomized, double-blind, phase 3 trial comparing the safety and efficacy of Merck/Pfizer’s ertugliflozin plus sitagliptin (Merck’s Januvia) versus either drug alone in patients with type 2 diabetes. The study recruited 1,233 patients inadequately controlled (A1c 7.5-11%) on stable metformin (≥8 wks at ≥1,500 mg/day), who were randomized to one of five groups: ertugliflozin 5mg or 15mg daily plus sitagliptin 100mg daily, ertugliflozin 5mg or 15mg daily alone, or sitagliptin 100mg daily alone. Data at 26 weeks showed that co-administration of ertugliflozin with sitagliptin led to significantly greater reductions in A1c (1.5% for both groups) compared to either drug alone (1%-1.1% across the three groups; p<0.002). A similar effect was observed with fasting plasma glucose and percentage of patients achieving A1c <7%. Co-administration also led to significantly greater reductions in body weight and systolic blood pressure compared to sitagliptin alone. Static beta-cell responsivity increased across all treatment arms and no difference was observed with the co-administration groups. 

  • The retention rate across all groups at 26 weeks was 92-96%. The authors provided the ranges of baseline characteristics across all five groups: average age of 55 years, % male of 51-62%, mean A1c of 8.50-8.57%, duration of T2DM of 6-7 years, eGFR of 92 mL/min/1.73m2, weight of 88-90 kg [194-198 lbs], and BMI of 32-33 kg/m2

Table 1: Data from 26 weeks showing greater reductions in A1c, FPG, weight, and SBP with combined ertugliflozin + sitagliptin versus either drug alone.

Reduction from baseline:

ERTU 5mg

ERTU 15mg

SITA 100mg

ERTU 5mg + SITA 100mg

ERTU 15mg + SITA 100mg







A1c (%)






FPG (mg/dl)






Body weight


2.7 kg  [6.0 lbs]

3.7 kg    [8.2 lbs]

0.7 kg       [1.5 lbs]

2.5kg#                         [5.5 lbs]

2.9kg#                        [6.4 lbs]







Pt with A1c <7%

66 (26%)

79 (32%)

83 (34%)

127 (52%)

120 (49%)

* p<0.002 vs. individual treatments

# p<0.005 vs. sitagliptin treatments (comparisons to ertugliflozin alone were not performed)

p<0.001 based on model-estimated odds ratio comparing ERTU+SITA vs. individual treatments

  • The safety profiles were similar between the five groups, with the exception of a higher observed rate of genital mycotic infections in the groups that included ertugliflozin. The authors note that there was no meaningful difference in urinary tract infection incidence between the groups, and that the rate of hypovolemia and symptomatic hypoglycemia were low across treatment groups.

Effect of Empagliflozin (EMPA) on Bone Fractures in Patients with Type 2 Diabetes (T2DM)(126-LB)

S Kohler, S Kaspers, A Salsali, C Zeller, and H Woerle

BI presented a retrospective analysis of pooled clinical trial data demonstrating no increased risk of bone fractures with empagliflozin (Jardiance, partnered with Lilly) in patients with type 2 diabetes. Pooled safety data from 15 phase 1-3 clinical trials and four extension studies was analyzed from patients who were randomly assigned to receive empagliflozin 10mg (n=4,221), empagliflozin 25mg (n=4,196) or placebo (n=4,203). Data from a head-to-head phase 3 study of empagliflozin vs. glimepiride was analyzed separately. The researchers evaluated the rate of bone fracture adverse events (AEs) in the total population and in subgroups divided by gender, baseline age, and baseline eGFR. The pooled analysis found no statistical difference in the number of bone fracture adverse events with empagliflozin vs. placebo but an increased risk in both groups in females, older patients, and those with moderate renal impairment. In the head-to-head study, empagliflozin 25 mg did not increase the risk of bone fracture adverse events compared to glimepiride 1-4mg. In the same study, neither treatment group saw bone mineral density T-scores outside the normal range. This retrospective analysis, along with results from the EMPA-REG OUTCOME trial, provides reassuring evidence that there is no clear link between empagliflozin and bone fractures. Bone fractures have been cited as a potential risk with SGLT-2 inhibitors because the drugs alter renal sodium and glucose reabsorption, resulting in possible changes in renal reabsorption of calcium and phosphate and possible modulation of bone metabolism. The FDA released a Drug Safety Communication last September strengthening the label warning on fracture risk for J&J’s SGLT-2 inhibitor Invokana (canagliflozin); it stated at the time that it would evaluate the risk with Jardiance and AZ’s Farxiga (dapagliflozin) to determine whether updates were warranted. We assume that this data, combined with the lack of any harmful signal in the EMPA-REG OUTCOME trial, makes any additional warnings unlikely.

Effect of Empagliflozin on Diabetic Ketoacidosis in Patients with Type 2 Diabetes: Pooled Clinical Trial Data (127-LB)

S Lund, F Solimando, S Kohler, C Zeller, and S Kaspers

In this poster, BI shared results from a meta-analysis that examined the effects of SGLT-2 inhibitor empagliflozin (Lilly/BI’s Jardiance) on diabetic ketoacidosis (DKA) in patients with type 2 diabetes.  Data from over 15,000 patients with type 2 diabetes from 18 randomized clinical trials (varying from phase 1 to 3 and in duration from 8 days to 4 years) were included in the meta-analysis.  4,558 of these patients received empagliflozin 10 mg, 5,520 patients received empagliflozin 25 mg, and 5,599 patients received a comparator drug.  Baseline characteristics were similar between the two groups.  DKA events were only reported in less than 0.1% of patients, and no difference was seen between patients given comparators or empagliflozin.  However, significantly more patients using empagliflozin had high urinary ketone levels (≥ 1+) than patients on comparators. The low event rate for DKA and comparable rates between empagliflozin and comparators across the 18 trials is reassuring; we believe the link between empagliflozin and high urinary ketone levels should be monitored. The general consensus surrounding the link between SGLT-2 inhibitors and DKA appears to be that the event rate is fairly low in patients with type 2 diabetes but urinary ketone monitoring may be useful, especially for patients on insulin.

Coadministration of Canagliflozin (CANA) and Phentermine (PHEN) for Weight Management in Overweight and Obese Adults (319-LB)

PA Hollander, HE Bays, J Rosenstock, ME Frustaci, A Fung, and N Erondu

In a late-breaking poster, phase 2 results of the co-administration of J&J’s Invokana (canagliflozin) and phentermine showed significantly greater weight loss vs. placebo in adults with overweight and obesity. This study was a four-arm, 26-week study and evaluated the efficacy and safety of Invokana + phentermine, phentermine alone, Invokana alone, and placebo in 334 adults with BMIs of 30-50 kg/m2 and without type 2 diabetes or with hypertension and/or dyslipidemia and BMIs of 27-50 kg/m2. The findings demonstrated that at 26 weeks, from a mean baseline BMI of ~37 kg/m2, the Invokana + phentermine group achieved significantly greater weight loss (7.5%) compared to the other groups (4.1%, 1.9%, and 0.6% for Invokana, phentermine, and placebo, respectively). In addition, the combination therapy group had a significantly higher proportion of participants achieving ≥5% weight loss compared to placebo (67% vs. 18%). Only 18% and 42% of the Invokana and phentermine groups’ participants achieved weight loss of at least 5%. On other secondary endpoints, the combination therapy group also had a significant reduction in systolic blood pressure, with a placebo-subtracted reduction of 4.2 mmHg. Regarding safety and tolerability, no new signals emerged from these data; however, the combination therapy, phentermine, and Invokana groups all experienced increases in heart rate of 3.5 bpm, 4.1 bpm, and 0.7 bpm, respectively – an observation whose potential consequences on cardiovascular events have not yet been determined and will likely warrant further investigation. Ultimately, these findings are powerful in support of using phentermine with SGLT-2 inhibitors in chronic weight management, and we are interested to see whether Janssen will move forward with phase 3 trials, and ultimately pursue an obesity indication. Combination therapies have certainly gained significant attention within obesity management (in addition to diabetes), and we have specifically heard greater enthusiasm for the potential of GLP-1 agonists and SGLT-2 inhibitors in weight management. With so much evidence already accumulated on these various diabetes drugs, we look to see how industry, healthcare providers, and the FDA envision the movement of these products into obesity, given the troubled commercial environment for obesity compounds to date and given the extreme high need for therapy to help patients.

Dapagliflozin Induces Ketosis in Patients with Type 1 Diabetes (949-P)

H Ghanim, N Kuhadiya, S Khan, M Garg, K Green, S Abuaysheh, and P Dandona

This poster presented data showing that the addition of dapagliflozin (AZ’s Farxiga) to liraglutide (Novo Nordisk’s Victoza) and insulin led to glycemic improvements but an increased risk of DKA in patients with type 1 diabetes. This dataset included 30 patients; we saw results for 16 patients at the AACE/ACE meeting on SGLT-2 inhibitors and DKA last October. Participants received once-daily doses of either 10mg dapagliflozin (5mg the first week) or a placebo for 12 weeks. 26 patients completed the study (17 taking dapagliflozin and 9 taking placebo), and two of the four who dropped out developed DKA within a day after increasing the dapagliflozin dose. For those who did complete the study, both A1c and average glucose were decreased in patients taking dapagliflozin compared to placebo (-0.66% vs. 0% and -15 mg/dl vs. +3.1 mg/dl, respectively). In addition, the incidence of hypoglycemia was comparable between the two groups (28% with placebo and 30% with dapagliflozin). As expected, patients on dapagliflozin had increased glycosuria, urine volume, blood glucagon, and free fatty acid concentrations compared to those on placebo. Notably, patients on dapagliflozin also had increased plasma concentrations of acetoacetate and urinary ketone bodies as well as increased concentrations of hormone sensitive lipase (indicating increased lipolysis), suggesting an increased risk of DKA. The authors concluded that caution must be exercised when decreasing insulin doses and increasing dapagliflozin doses in type 1 diabetes. Many in the field have gone further and advised against any off-label use of SGLT-2 inhibitors in type 1 diabetes outside of clinical trials until the risk of DKA is better understood. We agree that this is prudent advice in most cases, though some clinicians like Dr. Anne Peters (USC, Los Angeles, CA) with extensive knowledge of this issue (to say nothing of her patients) have been able to make it work. We advocate for patients understanding risks well and making personalized benefit/risk tradeoffs.

Empagliflozin and Microvascular Outcomes in EMPA-REG OUTCOME (1086-P)

C Wanner, C Lee, HJ Woerle, M Mattheus, SE Inzucchi, B Zinman

This poster presented the microvascular outcomes data of the EMPA-REG OUTCOME trial, finding that empagliflozin used in addition to standard of care reduced the risk of a composite microvascular outcome in people with type 2 diabetes and high CV risk, driven by a reduction in new or worsening nephropathy. The trial included 7,020 patients with a median treatment duration of 2.6 years and median observation time of 3.1 years, with final vital status available for 99% of patients. The trial defined composite microvascular outcome as time to the first incidence of a number of possible events: initiation of laser therapy for retinopathy or of renal replacement therapy, vitreous hemorrhage, diabetes-related blindness, progression of nephropathy, doubling of serum creatine with eGFR of at most 45 mL/min/1.73m2, or death caused by renal disease. The composite microvascular outcome occurred in a significantly lower percentage of patients on empagliflozin (pooled; 14.0%) than placebo (20.5%; HR of 0.62 [95% CI 0.54, 0.70]; p<0.001). Specifically, the hazard ratios for initiation of laser therapy for retinopathy, vitreous hemorrhage, and new or worsening nephropathy were 0.69 (0.43, 1.12) (p=0.134), 0.93 (0.51, 1.71) (p=0.815), and 0.61 (0.53, 0.70) (p<0.001), respectively. The overall reduction was thus primarily driven by the reduced incidence or decreasing severity of nephropathy in the empagliflozin group relative to placebo.

  • Of the participants, 4,687 received empagliflozin and 2,333 received placebo. All participants had type 2 diabetes (A1c 7%-9% for drug-naïve patients and 7.0%-10% for those on stable glucose lowering therapy), established cardiovascular disease, and eGFR of at least 30 mL/min/1.73m2. Participants received either placebo, 10 mg dose of empagliflozin, or 25 mg dose. The study continued until at least 691 participants experienced an event included in the primary outcome: cardiovascular death, non-fatal myocardial infarction, or non-fatal stroke.

Impact of Exposure to SGLT2 Inhibitors on Incidence of Diabetic Ketoacidosis Among Danish Type 2 Diabetes Patients (1098-P)

M Linnemann Jensen, B Carstensen, G Anderson, F Persson, J Nolan, M Ridderstråle, and M Jørgensen

This observational study by the prestigious Steno Diabetes Center aimed to determine the relationship between exposure to insulin, SGLT-2 inhibitors, and other glucose-lowering drugs and the incidence of DKA in a Danish population with type 2 diabetes. The study included 415,670 Danish residents with diabetes who had at least had one filled prescription for a glucose-lowering drug between 1995-2015; drug exposure length was recorded from the first prescription fill. Results demonstrated 4,045 total DKA events over the 20-year period, translating to an incidence of 1.34 events/1,000 patient-years. We were encouraged to see that the DKA incidence rate decreased by 5.6% per year between 1995 and 2015, perhaps suggesting improvements in glucose control or diagnosis/treatment of DKA. The analysis found that exposure to any glucose-lowering drug led to a significant 30% increased risk of DKA relative to no treatment. In terms of specific therapies, patients on insulin had the highest incidence of DKA at 6.0 events/1,000 patient-years; exposure to insulin led to a six-fold increased risk. Most notably, exposure to an SGLT-2 inhibitor on top of insulin and other glucose-lowering drugs carried a hazard ratio of 2.5 (95% CI: 1.1-5.5) based on six events among 4,524 patients. Overall, the authors concluded that although the addition of an SGLT-2 inhibitor may increase the risk of DKA, this increased risk may not be clinically relevant given the low absolute event rate. Their main conclusion was that DKA was and is rare in type 2 diabetes, both with and without SGLT-2 inhibitor therapy. These results are consistent with the consensus from last fall’s AACE/ACE meeting that while SGLT-2 inhibitors may increase the risk of DKA in type 2 diabetes, the absolute event rates are too low to warrant any major changes in prescribing patterns. 

Improved Treatment Satisfaction in People with Type 1 Diabetes Mellitus (T1DM) Treated with Canagliflozin (CANA) (1184-P)

AL Peters, HW Rodbard, A Slee, and S Traina

This poster investigated patient satisfaction among patients with previously uncontrolled type 1 diabetes who were treated with J&J’s SGLT-2 inhibitor Invokana (canagliflozin). The study’s analysis was based on patient-reported outcomes from an 18-week, randomized, double-blind, placebo-controlled study of patients with type 1 diabetes who were on background basal insulin plus bolus therapy. Patients were randomized to either canagliflozin 100 mg, 300 mg, or placebo, with treatment satisfaction assessed at baseline and at week 18 using the two versions of the Diabetes Treatment Satisfaction Questionnaire (DTSQ) – DTSQ status (DTSQs, which measures baseline satisfaction) and DTSQ change (DTSQc, which measures changes in satisfaction). As background, DTSQs scores range from 0 to 36 with higher scores indicating greater satisfaction. DTSQc scores range from -18 to +18, with positive scores indicating increases in satisfaction compared to baseline. The findings demonstrated that the canagliflozin-treated arms showed greater increases in DTSQc compared to placebo, though all participants improved in all six categories of the DTSQc regardless of treatment group. Notably, canagliflozin-treated patients spent much less time in hyperglycemia and hypoglycemia. Product-method mediation analysis indicated that reduction in total insulin dose and reduction in glycemic variability was responsible for between 4% and 9% of the change in satisfaction with canagliflozin. Concluding, the study suggested that decreased doses of insulin, glycemic variability, and body weight may contribute to the increase in satisfaction among canagliflozin-treated patients. However, the levels and factors of contribution to this rise in satisfaction has not been precisely determined. With these findings, the poster’s authors note that not only can such therapies increase satisfaction, but they may also improve adherence.

  • Baseline demographics and diabetes characteristics were similar among the three patient groups. A1c levels were 7.9%, 7.9%, and 8% for the placebo, canagliflozin 100 mg, and 300 mg groups, respectively. Mean ages were 41.8, 42.2, and 42.6 years for the placebo, canagliflozin 100 mg, and 300 mg groups, respectively. BMI measurements were also similar.
  • Canagliflozin 100 and 300 mg reported higher increases in satisfaction after 18 weeks. Baseline DTSQs scores were very similar (29.0, 28.4, and 28.6 with canagliflozin 100 mg, 300 mg, and placebo, respectively). However, at the end of the trial, those patients treated with canagliflozin showed greater improvements in treatment satisfaction compared to those on placebo. In addition, patients in the canagliflozin-treated groups showed DTSQc scores that were 4.8 (1oo mg) and 5.5 (300 mg) points higher than the placebo group.

Symposium: The Good Heart, the Bad Bone, and the Ugly Alpha Cell – What About them SGLT2 Inhibitors?

Potential Benefits of SGLT-2 Antagonists in Prevention and Safety of Use in Patients with Heart Failure

David Aguilar, MD (Baylor College of Medicine, Houston, TX)

Dr. David Aguilar provided a cardiologist’s perspective on the intersection between heart failure and diabetes and suggested that a diuretic effect is the most plausible explanation for the heart failure benefit seen in EMPA-REG OUTCOME. He reviewed evidence demonstrating that diabetes and heart failure often coexist and are a dangerous combination. He noted that cardiologists treating heart failure have historically viewed diabetes drugs in terms of potential increased risk and celebrated the potential for a more positive view in light of the EMPA-REG OUTCOME results. Dr. Aguilar provided a long list of potential mechanisms for the reduced risk of heart failure hospitalization seen in the trial: glucose control, weight loss, adverse effects of other medications in the placebo group, uric acid reduction, blood pressure lowering, reduced preload and afterload, volume/sodium depletion, and renal effects. Of these, he (like many others) believes that a diuretic effect is the most plausible explanation; he suggested that patients with existing diastolic dysfunction could be especially sensitive to the effects of a mild diuretic. He also granted some credence to the ketone body utilization hypothesis put forward by Dr. Ele Ferrannini and Dr. Sunder Mudaliar earlier in the day but stated that much more work needs to be done to validate it. He also noted that positive hemodynamic effects and reductions in arterial stiffness are potential explanations as well; he described the other possibilities as unlikely due to the small effect size and/or lack of a demonstrated connection to cardiovascular outcomes. Dr. Aguilar suggested that the mechanism underlying the CV death reduction is even less clear than that underlying the heart failure benefit and called for further studies to elucidate it.

Questions and Answers

Q: For all the outcomes we’re looking at, there was no dose effect. I don’t know what the explanation is, but we might want to think about that.

A: Perhaps maximal diuresis occurs with the lower dose. If you look at volume depletion it’s not that different.

Q: If the answer is diuresis, I’m frankly puzzled. Most patients with known heart failure are already on diuretics, and they’re often instructed to adjust them to maintain weight. Why would a different form of diuresis perhaps act as a superior mechanism to reduce heart failure risk and potentially death?

A: That’s a super question. The answer is yes, it is different. As for the follow-up question of why, I don’t know. There are other neurohormonal changes with activation of RAS, but we have a body of literature showing that loop diuretics don’t change survival. What’s different? Why doesn’t heart rate increase? We don’t have good explanations. Maybe there’s something different in kidney signaling related to glycosuria.

Comment: I agree it’s not atherosclerosis. Another syndrome people with diabetes also get is microvascular cardiomyopathy and no one’s addressing that. There might be a signal that the brain is affected by SGLT-2 inhibitors. There’s a paper on arterial compliance, but I don’t put a lot of emphasis on that. It probably was a reduction in blood pressure.

A: I agree there might be effects on arterial compliance. That was a small study, so there’s not a lot of data behind it, but it did change a little. I really think we need more data, and there’s a potential role for microvascular function. Maybe if someone has an infarct they’re less susceptible to arrhythmias.

Understanding the Risk of DKA when Hyperglycemia is Treated with SGLT-2 Inhibitors

Simeon Taylor, MD, PhD (University of Maryland, Baltimore, MD)

Dr. Simeon Taylor reviewed the available evidence on the risk of DKA with SGLT-2 inhibitors and offered a set of clinical recommendations. His main hypothesis is that SGLT-2 inhibitors increase susceptibility to other precipitating factors for DKA (e.g., insulin dose reductions, pump failure, infection, surgery) but do not themselves cause overt ketoacidosis. He reviewed mechanistic evidence demonstrating that SGLT-2 inhibitors promote a shift from carbohydrate to lipid metabolism, somewhat comparable to the body’s response to starvation. This leads to significantly higher levels of ketone bodies in the blood, which makes patients more vulnerable to overt DKA under the right circumstances. With the caveat that this dataset is far from perfect, he also reviewed FDA Adverse Event Reporting System (FAERS) data on the rates of SGLT-2 inhibitor-associated DKA, concluding that there appears to be an approximately 14-fold increased risk of ketoacidosis among SGLT-2 inhibitor-treated patients compared to DPP-4 inhibitor-treated patients. However, this may overestimate the risk because at least 30% of the SGLT-2 inhibitor-associated reports of ketoacidosis were observed in type 1 rather than type 2 diabetes. If the type 1 diabetes-associated reports are removed from the database, then Dr. Taylor estimated a five to 10-fold increased risk in type 2 diabetes. As for the clinical implications, Dr. Taylor advised against prescribing the class off-label in type 1 diabetes, as titrating insulin to avoid both DKA and severe hypoglycemia is very challenging. As he put it, “treating people with type 1 diabetes is like trying to thread a needle, and SGLT-2 inhibitors narrow the width of the eye of the needle.” He did stress that he is open to further research on the class in type 1 diabetes, citing the possibility for non-glycemic benefits like amelioration of renal hyperfiltration. In type 2 diabetes, Dr. Taylor suggested that key risk factors for DKA include surgery, low-carb diets, rapid weight loss, and insulin deficiency. He even suggested that clinicians consider avoiding SGLT-2 inhibitors entirely in truly insulin-dependent patients, regardless of diagnosis. He closed by cautioning against relying on urine ketone tests or severe hyperglycemia to diagnose DKA, as there is evidence that SGLT-2 inhibitors can actually reduce ketonuria by promoting reabsorption of ketone bodies and a number of the reported cases of DKA have occurred at blood glucose levels <250 mg/dl.

  • As a sidenote/postscript to Dr. Taylor’s talk, we note that we have been surprised to learn how many patients with both type 1 and type 2 diabetes do not have experience identifying DKA (2/5 of those with type 2 diabetes in a recent dQ&A survey say they are not confident they could recognize the symptoms of DKA), do not have a ketone meter even if they are on an SGLT-2, and may not have insurance coverage for ketone meters and strips. Although we agree the DKA risk can be managed for patients, we do not think patients in the US (and possible elsewhere) have appropriate education in this sphere. For more information on this data, contact Richard Wood ( 

Questions and Answers

Q: The evidence seems to be that the effects of this class in general are not dose-dependent but the risks including DKA and bone fractures are dose-dependent.

A: I think it depends on which benefits you’re talking about. They may have some degree of dose dependence. A1c lowering is slightly more at the higher dose. I think the peak effect is maximized at the lower dose, but the biggest difference is 24 hours later; when you use the higher dose it’s still giving you something but the lower dose is falling off at the end. There could be complex factors relating to whether you need 24-hour coverage.

Q: Can you speculate about the use of SGLT-2 inhibitors in type 1 diabetes?

A: They are not approved in type 1 diabetes and I absolutely do not recommend them. I think it’s a legitimate research question but I don’t think the drug should be used routinely in type 1 diabetes. Clearly some people can take them safely, but the real question is what is the proven benefit of using it? But I don’t want to use my personal opinions. The drugs are simply not approved there.

Q: Based on the presentations at the AACE/ACE consensus conference, quite a few patients developed DKA without antibodies, so I don’t think that’s a helpful way to rule it out. The pharmaceutical companies excluded people with high A1cs from their trials, and it was those people with high A1cs and low insulin secretory ability who got DKA.

A: I agree, I don’t think it’s a full explanation but it’s possible that patients with antibodies have greater risk.

ADA Diabetes Care Symposium: Lagniappe – A Little Something Different

The EMPA-REG Outcome Study Enigma: Are Fuel Energetics the Answer? Cardioprotection in EMPA-REG: A Thrifty Fuel Hypothesis

Ele Ferrannini, MD (University of Pisa, Italy)

Dr. Ele Ferrannini hypothesized that a shift toward ketone body metabolism could have been responsible for the cardioprotective effect in EMPA-REG OUTCOME. This hypothesis was also outlined in a paper just published in Diabetes. Dr. Ferrannini explained that patients with type 2 diabetes have consistently higher lipid oxidation than people without diabetes, and the glucosuria produced by SGLT-2 inhibitors amplifies this effect by decreasing glucose and insulin levels and increasing glucagon levels. This effect has become a common point of discussion over the past year in the context of the potential increased risk of DKA, but Dr. Ferrannini suggested that it could produce positive effects as well. In his model, the increased availability of lipid substrates leads the liver to produce more ketone bodies, which are then taken up by the heart. Ketone bodies are very efficient fuel sources, meaning that at any given level of oxygen they generate more ATP compared to other substrates like glucose. Dr. Ferrannini noted that SGLT-2 inhibitors also lead to increased hematocrit (the ratio of oxygen-carrying red blood cells to total blood volume) and therefore could increase oxygen delivery to the heart. That combination of more oxygen and more efficient energy utilization could improve the heart’s contractile ability and lessen the strain on a failing heart. Dr. Ferrannini emphasized that this is only a hypothesis and that it should not preclude other more “conventional” explanations such as a diuretic effect, but it certainly adds an interesting new dimension to the discussion. At a BI/Lilly-sponsored corporate symposium later that evening, Dr. Lawrence Blonde (Ochsner Health System, Jefferson, LA) and Dr. Carol Wysham (Rockwood Clinic, Spokane, WA) both described this hypothesis as very compelling, and we are sure we have not heard the last of it.  

The EMPA-REG Outcome Study Enigma: Are Fuel Energetics the Answer? Ketone Bodies in Diabetic Heart Failure and Kidney Disease: Friend or Foe?

Sunder Mudaliar, MD (UCSD, San Diego, CA)

Dr. Sunder Mudaliar outlined a similar fuel energetics hypothesis as that presented by Dr. Ele Ferrannini (also published in a paper in Diabetes) to explain the renal protective effects demonstrated in EMPA-REG OUTCOME. He explained that the kidneys have very high energy needs and rely primarily on oxidative metabolism – their level of oxygen consumption is second only to that of the heart. The kidneys’ workload and oxygen requirements are even higher in diabetes due to abnormally high glucose reabsorption, and Dr. Mudaliar noted that the resultant renal hypoxia and oxidative stress are increasingly being recognized as independent pathways involved in the progression of chronic kidney disease. SGLT-2 inhibitors could relieve some of this stress by shifting renal metabolism toward ketone bodies instead of oxygen. Because ketone bodies are a more efficient fuel source, the kidneys do not need to expend as much energy or consume as much oxygen. Dr. Mudaliar acknowledged that these changes are subtle but argued that they could translate to large differences in clinical outcomes over time. That said, he emphasized that this model is only a hypothesis and advocated for more studies measuring oxygen consumption in the kidneys and heart to test it.

-- by Melissa An, Shivani Chandrashekaran, John Erdman, Nina Ran, Emily Regier, Ava Runge, Manu Venkat, Alasdair Wilkins, Sarah Wilkins, Yrenly Yuan, and Kelly Close