September 14-18, 2015; Stockholm, Sweden; Day #5 Highlights - Draft

Executive Highlights

Goodbye Stockholm … what a conference it’s been! Kudos to the EASD organizers for wrapping up Day #5 in style. Those who stayed till the end – a good number of conference-goers – were treated to quite a day of learning.

One major highlight of the day came near the very end, when Dr. Robert Henry presented phase 2 results for J&J’s Invokana (canagliflozin) in type 1 diabetes (demonstrated reductions in A1c, body weight, and insulin doses but a dose-dependent increased risk of ketoacidosis) while another was Dr. Anne Peters who discussed her interpretation of the DKA results. We also heard from Dr. John Buse, who presented sub-analyses (new to us) from Novo Nordisk’s phase 3 DUAL-1 extension trial for Xultophy (insulin degludec/liraglutide), and Dr. Naveed Sattar, who questioned the incremental benefits of the combination over GLP-1 alone. Meanwhile, Dr. Robert Josse (St. Michael’s Hospital, University of Toronto, Canada) presented an update on TECOS safety data, suggesting a possible class effect with increased risk for acute pancreatitis with DPP-4 inhibitors. For much more – including a debate on the utility of GLP-1 agonist/basal insulin combinations – please see our highlights below, followed by full write-ups of selected talks.

Top 5 Highlights

1. Phase 2 results for J&J’s Invokana (canagliflozin) in type 1 diabetes demonstrated reductions in A1c, body weight, and insulin doses but a dose-dependent increased risk of ketoacidosis.

2. Dr. Robert Josse (St. Michael’s Hospital, University of Toronto, Canada) presented an update on TECOS safety data, suggesting a possible class effect with increased risk for acute pancreatitis with DPP-4 inhibitors.

3. The highly regarded Dr. John Buse (University of North Carolina, Durham, NC) presented sub-analyses (new to us) from Novo Nordisk’s phase 3 DUAL I extension trial for Xultophy (insulin degludec/liraglutide), showing broad efficacy irrespective of patients’ baseline A1c or BMI.

4. Dr. Naveed Sattar shared frank commentary on IDegLira’s phase 3 DUAL program, questioning the incremental benefits of the combination over GLP-1 alone.

5. A debate on the utility of GLP-1 agonist/basal insulin combinations featured familiar arguments in favor of the class and some less often cited points against it.

Top 5 Highlights

1. Phase 2 results for J&J’s Invokana (canagliflozin) in type 1 diabetes demonstrated reductions in A1c, body weight, and insulin doses but a dose-dependent increased risk of ketoacidosis. The double-blind trial (n=351) randomized patients to receive one of two doses of Invokana (100 mg or 300 mg) or placebo as an adjunct to insulin for 18 weeks. A slightly higher percentage of patients achieved the composite primary endpoint of A1c reductions ≥0.4% and no weight gain with both doses of canagliflozin vs. placebo (41% and 37% with the higher and lower doses, respectively, vs. 15% with placebo; p<0.001) – this seemed like a pretty easy composite endpoint to reach and we felt no weight gain was surprising given that SGLT-2s usually cause weight loss. Placebo-adjusted weight loss was 5.4% and 3.4% with the two respective doses of canagliflozin (baseline weight = 83-84 kg; baseline BMI = 28 kg/m²), and both doses produced placebo-adjusted A1c reductions of 0.3% (baseline = 7.9%-8% - pretty low). We as usual (“this is a recording”) would’ve liked time in zone data – CGM was used in the last week of the trial but results were not shared – we do think they would improve the results theoretically. Notably, canagliflozin also significantly reduced insulin doses (by 8U/day for the 300 mg dose and 4 U/day for the 100 mg dose – we’re not sure what the baseline was) and led to numerical but not significant reductions in fasting plasma glucose (placebo-adjusted reductions of 11 mg/dl and 9 mg/dl with the two respective doses; baseline = ~180 mg/dl). There were no significant differences between groups for any hypoglycemia parameters, though the rate of severe hypoglycemia was numerically increased in the 300 mg canagliflozin group (7% vs. 3% with 100 mg canagliflozin and 2% with placebo). Notably, there was a significant increase in ketone-related adverse events (11 events and 6 events with the two respective doses vs. none with placebo), all associated with precipitating factors such as pump failure, missed insulin doses, or concurrent illness. While it is debatable in our view whether the benefit/risk profile for Invokana in type 1 diabetes would meet the FDA’s standards based on these results alone, Dr. Henry suggested that the risk of DKA could likely be reduced in phase 3 with more frequent monitoring, lower doses, less of a reduction in basal insulin dose, and/or temporary cessation of treatment during illness or stress. Overall, we remain optimistic about the drug’s prospects in this population, particularly once CGM data from this trial is available – see below for more. We also know it’s going to be used off-label regardless so we urge the company to move forward on it and for the FDA to continue to discuss thoughtfully with sponsors what they want to see for approval.

• The renowned Dr. Anne Peters (USC, Los Angeles, CA) preceded Dr. Henry’s talk by discussing her interpretation of the DKA results and her personal “recipe” for managing the risk. She shared convincing personal accounts from her patients about the less quantifiable benefits of SGLT-2 inhibitors in type 1 diabetes, particularly the reductions in glycemic variability. She framed the ketoacidosis risk as significant but manageable in this population and noted that she has not seen any cases since instituting her protocol – see below for more.

2. Dr. Robert Josse (St. Michael’s Hospital, University of Toronto, Canada) presented an update on TECOS safety data, suggesting a possible class effect with increased risk for acute pancreatitis with DPP-4 inhibitors. As we learned at ADA, TECOS had an imbalance in acute pancreatitis with 23 vs. 12 cases (p=0.065) not in favor of sitagliptin. Dr. Josse presented a deeper analysis into the data, showing that severe or fatal pancreatitis occurred rarely, with four severe cases and 19 mild/non-severe cases. The median number of days from randomization to pancreatitis was 517 for the sitagliptin group vs. 528 for placebo. Regarding etiology, ten were of unknown etiology while two were alcohol-related, eight were biliary-related, and four were related to a prior pancreatitis history. Notably, Dr. Josse presented a meta-analysis of the three DPP-4 inhibitor CVOTs (SAVOR, EXAMINE, TECOS), which suggested a statistically significant increased risk of acute pancreatitis with an overall hazard ratio of 1.58 (p=0.033). However, he stressed to attendees the need to interpret these results with caution, noting that the trials had differences in follow-up duration and enrolled patient populations as well as variation in the adjudication procedures. As for pancreatic cancer (9 vs. 14 cases in favor of sitagliptin), Dr. Josse showed that a meta-analysis of SAVOR and TECOS (EXAMINE was not included as no cases were reported) suggested no difference in the risk of pancreatic cancer, despite the fact that the rates were numerically lower with the DPP-4 inhibitor group in both trials (HR=0.54; p=0.066). With the acute pancreatitis risk, we would suspect that these numerical imbalances in favor of DPP-4 inhibitor therapy are likely just due to chance, especially with the low number of cases. Yet increasingly, it appears that acute pancreatitis risk may be a class effect, although so far, we have heard most speakers playing the safety signal down. For example, Dr. Josse emphasized that it may not be a major clinical risk, noting that further post-marketing surveillance will help to determine the clinical relevance despite the meta-analysis’ findings;  Dr. John Buse (UNC, Chapel Hill, NC) similarly suggested at ADA that such a risk would be rare and acceptable within the class’ otherwise strong safety profile. While in the aftermath of EMPA-REG, this information on pancreatitis risk is not the most compelling in terms of CVOT findings, we believe that it could still be somewhat useful for providers when individualizing second-line therapies for patients with a history of pancreatitis (although the evidence is likely not sufficient to drastically change current guidelines quite yet).

3. The highly regarded Dr. John Buse (University of North Carolina, Durham, NC) presented sub-analyses (new to us) from Novo Nordisk’s phase 3 DUAL I extension trial for Xultophy (insulin degludec/liraglutide), showing broad efficacy irrespective of patients’ baseline A1c or BMI. Dr. Buse opened by discussing patient preferences, sharing a meta-analysis of “what is important to the patient about drugs.” Unsurprisingly, treatment benefits (e.g., glycemic control, weight loss) topped this list followed by side effects (e.g., nausea, diarrhea) and treatment burden (e.g., route of administration, frequency). In an echo of his presentation from day #1 of EASD, he then introduced Xultophy as a novel alternative that exceeds expectations on all these fronts: “The efficacy is unique. The tolerability is impressive. The safety is very strong. The quality of life enhancement is really impressive.” To back these claims, Dr. Buse shared data on the efficacy of Xultophy stratified by baseline characteristics (A1c, BMI, and pre-trial medication), demonstrating that the therapy provides benefits across five subgroup analyses. The data are consistent with what has been seen in the total trial population: (i) improvements in glycemic control across all A1c categories vs. component parts; (ii) improvements in hypoglycemia rates across all A1c categories vs. insulin degludec [even at the lowest A1c groups where you might expect more parity]; (iii) improvements in glycemic control across all BMI categories vs. component parts; (iv) improvements in hypoglycemia rates across all BMI categories vs. insulin degludec; and (v) equivalent improvements in glycemic control regardless of baseline medications. See the detailed breakdown in our Appendix below. Dr. Buse acknowledged that these across-the-board benefits make it difficult to identify a population of type 2 patients who would be ill-suited to treatment with Xultophy. It’s hard to argue with these data (though Dr. Leszek Czupryniak did just that in a morning debate on the same day – see #5 below), and we believe in contrast to Dr. Czupryniak that this combination class is one of the most exciting advances on the near-term horizon for type 2 diabetes. We expect that cost will be the most significant barrier to widespread adoption of the class, and the inability to titrate components individually could also be frustrating for some patients. That said, we think adherence would be impressive and we would urge those in control to work out what profile of patient might respond particularly well to this compound.

4. Dr. Naveed Sattar shared frank commentary on IDegLira’s phase 3 DUAL program, questioning the incremental benefits of the combination over GLP-1 alone. Dr. Sattar acknowledged the advantages of IDegLira (one injection instead of two, slower GLP-1 titration, less hypoglycemia and weight gain than insulin, ~0.5%-0.9% lower A1c than the individual drugs), but focused mostly on his view of the drawbacks: the inability to titrate the drugs independently (we’re not sure why a patient couldn’t start out on a low dose of GLP-1), less weight loss and more hypoglycemia than liraglutide alone, and cost (which of course is impossible to assess). Dr. Sattar called for a trial comparing the fixed-combination of IDegLira to a free combination of the drugs (start with liraglutide and titrate up basal insulin to target) – in essence, minimizing insulin to preserve the weight and hypoglycemia advantages of liraglutide as much as possible. Of course, this view also ignores the burden of two injections, which Dr. Sattar seemed to underestimate. In the valuable Q&A (see below), Drs. John Buse and David Russell-Jones tended to agree with Dr. Sattar in broad strokes, though Dr. Buse noted what is perhaps the most remarkable stat in the DUAL program: ~80% of patients got to an A1c <7% with IDegLira. A GLP-1 alone is clearly a great option when possible, but for all those patients failing GLP-1 alone, failing basal insulin alone, or stuck with two injections per day, IDegLira seems like an excellent option (assuming cost is not prohibitive). Dr. Sattar was less convinced, concluding that IDegLira “makes sense” and “has a role in a modest number of patients,” but more studies are needed to judge it. See below for detailed coverage of this talk and the subsequent Q&A.

5. A debate on the utility of GLP-1 agonist/basal insulin combinations featured familiar arguments in favor of the class and some less often cited points against it. Dr. Stephen Gough (University of Oxford, UK) argued for the use of these fixed-ratio combinations, noting that they are “more than the sum of the component parts” in terms of efficacy, tolerability, safety, and quality of life. Specifically, he reviewed the positive safety and efficacy data from the DUAL trials of Novo Nordisk’s Xultophy (insulin degludec/liraglutide), pointed to the GI side effect disadvantages of free-mixing GLP-1 agonists and insulin, and noted the simplicity of the fixed dose combination for patients. These points echo those from Dr. John Buse in multiple presentations during this conference, and those we have heard repeatedly at conferences over the past few years as the DUAL trials have reported impressive results. Dr. Leszek Czupryniak (Medical University of Lodz, Poland) was tasked with what he acknowledged was the more difficult task of arguing against these combinations, but he joked that he had almost convinced himself by the time he was done. He suggested that the pathophysiological rationale for the combination is somewhat questionable since both components act at least in part by increasing insulin levels. He believes it would be more logical to start with a GLP-1 agonist and then add insulin if necessary; in our view, the answer likely varies depending on the patient (for example, the preferred approach would be different for someone with a very high baseline A1c vs. someone for whom weight loss is the main goal). Dr. Czupryniak also noted that the A1c reductions achieved in some trials of these combinations were fairly modest (~0.5%) and claimed that “you can do that with lifestyle in 90% of patients” for a fraction of the cost. Ugh – theoretical arguments like this make little sense to us – all that is needed is to look at the data to see how many patients are successful with lifestyle alone (maybe Dr. Czupryniak could let us know how he does it). Cost was one of the main drawbacks Dr. Czupryniak cited, and the one that we think will be most relevant for patients – of course the ROI of the drugs are rarely discussed (because ROIs are not calculated that we know of). Ultimately, while landmark study DPP shows he is correct that a successful lifestyle intervention can be more effective than the best new drug, the problem lies in implementation – we’d love to see routine implementation be programs that integrate therapy, technology, and lifestyle such as Omada or other behavioral programs. See below for more detailed discussion of this timely topic.

Honorable Mention

• Dr. Juris Meier (St. Josef Hospital, Bochum, Germany) suggested that it may be worthwhile to pivot GLP-1 combination therapy research toward newer therapies (e.g., secretin, cholescystokinin, and gastrin) and away from existing avenues of development (e.g., GIP and glucagon agonists). The commentary came during a presentation on GLP-1 combination therapies during which Dr. Meier acknowledged the class’s efficacy and tolerability relative to type 2 alternatives. Given this potential, Dr. Meier highlighted the interest in additive effects when co-administering GLP-1 agonists with PYY, GIP, and glucagon, summarizing the host of literature in this field. In particular, Dr. Meier highlighted the additive impact on energy intake of PYY and GLP-1 co-administration and the “interesting” combination of GLP-1 and glucagon co-agonists. On the latter, Dr. Meier expressed healthy skepticism based on a number of theoretical concerns: (i) the stimulation of hepatic glucose by glucagon; (ii) the fact that glucagon antagonists are also being explored in type 2 diabetes treatments; (iii) the synergistic effects on the induction of nausea; (and (iv) the synergistic effects to increase heart rate. Dr. Meier raised similar theoretical concerns about combination of GIP and GLP-1 co-agonists. He pointed out that GIP has already proven ineffective in patients with type 2 diabetes and that the drug counteracts the suppression of glucagon by GLP-1. Alternatively, Dr. Meier discussed the potential of exploring other hormones in the GI tract that he sees as far more promising and synergistic with incretin-based therapy: secretin, cholescystokinin, and gastrin. We haven’t heard about a lot of those and would like to know more. In closing, despite the progress that has been made in GLP-1 combination therapy (with respect to GIP, glucagon, and PYY), Dr. Meier argued that it may be worthwhile to refocus our efforts on other research areas moving forward.

Detailed Discussion and Commentary

Trending Topics in SGLT-2 Inhibition

Efficacy and Safety of Canagliflozin, a Sodium Glucose Co-transporter 2 Inhibitor, as Add-on to Insulin in Patients with Type 1 Diabetes Mellitus

Robert Henry, MD (University of California San Diego, CA)

Dr. Robert Henry presented phase 2 results for J&J’s Invokana (canagliflozin) in type 1 diabetes demonstrating reductions in A1c, body weight, and insulin doses but a dose-dependent increased risk of ketoacidosis. The double-blind trial (n=351) randomized patients to receive one of two doses of Invokana (100 mg or 300 mg) or placebo as an adjunct to insulin for 18 weeks. A higher percentage of patients achieved the composite primary endpoint of A1c reduction ≥0.4% and no weight gain with both doses of canagliflozin vs. placebo (41% and 37% with the higher and lower doses, respectively, vs. 15% with placebo; p<0.001). Placebo-adjusted weight loss was 5.4% and 3.4% with the two respective doses of canagliflozin (baseline weight = 83-84 kg; baseline BMI = 28 kg/m²), and both doses produced placebo-adjusted A1c reductions of 0.3% (baseline = 7.9%-8%). Canagliflozin also significantly reduced insulin doses (by 8U/day for the 300 mg dose and 4 U/day for the 100 mg dose) and led to numerical but not significant reductions in fasting plasma glucose (placebo-adjusted reductions of 11 mg/dl and 9 mg/dl with the two respective doses; baseline = ~180 mg/dl). There were no significant differences between groups for any hypoglycemia parameters, though the rate of severe hypoglycemia was numerically increased in the 300 mg canagliflozin group (7% vs. 3% with 100 mg canagliflozin and 2% with placebo). That is definitely not a positive though we assume that over a longer period of time as insulin dosing changes are managed better, this would go down. We would guess however that an FDA Advisory Committee would need a lot of advocates (who would show up) to get around this. We’d love to know how many severe hypos caused hospital visits. As expected, female genital mycotic infections were increased with canagliflozin (10 events and 9 events with the two respective doses vs. 6 events with placebo) though not by that much. Notably, there was a significant increase in ketone-related adverse events (11 events and 6 events with the two respective doses vs. none with placebo), including serious DKA (6 events and 4 events with the two respective doses vs. none with placebo). All events were associated with precipitating factors such as pump failure, missed insulin doses, or concurrent illness.

• Dr. Henry suggested that rates of DKA could be lower in future type 1 diabetes trials based on lessons learned from this study. While participants were educated about the risk and advised to test ketones in this trial, the full extent of the concern was not widely known when the study began, and future trials will likely include more extensive education for both patients and investigators. Investigators in this study were also advised to down-titrate patients’ basal insulin (by 10% for baseline A1c <8% or 20% for baseline A1c >8%) prior to initiating treatment, which will likely not be part of the protocol in future trials now that the association between reduced insulin doses and DKA risk is better understood. Dr. Henry also pointed to more frequent ketone monitoring, temporary cessation of treatment during illness or stress, and lower doses (such as 25 mg or 50) of canagliflozin as additional potential mitigating factors in future studies.
• We remain optimistic (perhaps it’s just hopeful) that J&J will choose to advance Invokana into phase 3 for type 1 diabetes. While it is debatable whether the benefit/risk profile would meet the FDA’s standards based on these results alone, Dr. Henry suggested during Q&A that one of the main goals of this study was to determine the most effective design for phase 3, and we suspect that with some alterations (such as lower doses of canagliflozin or a more obese study population), the benefits might more clearly outweigh the risks. We are also very eager to see the CGM data from this trial (patients wore CGM for a week before randomization and during the final week of the study), as reduced hypoglycemia and hyperglycemia may be among the most meaningful benefits of SGLT-2 inhibitors in type 1 diabetes from a quality of life perspective even if A1c isn’t different. 

Questions and Answers

Q: The conclusion might be that SGLT-2 inhibitors in type 1 diabetes work. They will improve A1c and quality of life, but you have to learn something. I think the approach of reducing the basal dose and not reducing the bolus dose so much was not right. We’ve seen in our studies, if you reduce the bolus dose you reduce the rate of hypoglycemia as well. The bolus dose was too high and the basal was too low in this case. It’s good to have experience for handling exercise. With acute exercise, you can cover it with carbohydrates or reduction of the bolus dose, but with the refilling phase of muscles we have no experience. It works but we have to learn.

A: The intention of the phase 2 study was to find out factors and features to modify before a definitive registration trial. This started before the whole concept of DKA in type 1 diabetes was recognized as an issue. We thought it might be but not at the level it became, in the 5-6% range. We learned a great deal, which will allow the development of additional studies to answer questions more clearly. I agree with your comment about not primarily reducing basal insulin, but that’s what we learned from this study. Most of the reduction if not all should be postprandial. From previous experience we thought basal reduction was appropriate. That’s the reason for doing studies, so you can design the next set to be more accurate. I’m confident we’ve learned a great deal. The paper with further details on DKA is under review now.

Q: Were adverse events any different between MDI and CSII treatment?

A: There was no difference. I think 38% of patients were on basal-bolus and about 62% on CSII and people on both types got DKA.

Q: Lexicon has presented data on its dual inhibitor with no DKA or no euglycemic DKA reported. Here we do see it. Do you think canagliflozin could be more prone to produce DKA than the dual inhibitor? Could it be mechanism based?

A: Canagliflozin’s SGLT-1 effect is transient in the first one to two hours when the concentration gets high in the proximal duodenum, but the effect is transient, not systemic. The concentration is very low to inhibit SGLT-1 in other parts of the body. I don’t believe there is any relationship between the contribution of SGLT-1 and ketoacidosis. There could be, but the SGLT-1 inhibition is so rapid and short-lived, it seems unlikely. Empagliflozin in short-term studies did develop ketoacidosis, though there were excessive reductions in insulin doses. The point is that empagliflozin is a pure SGLT-2 inhibitor and they got DKA. I don’t believe SGLT-1 is a component. It’s primarily an effect on glucagon and insulin in people with a precipitating factor.

Update on Ketoacidosis with SGLT-2 inhibitors

Anne Peters, MD (USC, Los Angeles, CA)

Dr. Anne Peters’ presentation focused specifically on the DKA results from the phase 2 trial of Invokana presented by Dr. Robert Henry, and on anecdotes from her patients about the less quantifiable benefits of SGLT-2 inhibitors in type 1 diabetes. She opened her talk with convincing personal accounts, including a video of one of her patients, about the dramatic improvements in quality of life Invokana can produce for people with type 1 diabetes, primarily due to reductions in glycemic variability. She noted that even some of her patients who have experienced ketoacidosis have begged to resume treatment afterwards and are able to manage the risk by testing ketones regularly and being aware of precipitating events. Based on the results from this study and her clinical experience, Dr. Peters framed the risk of DKA in type 1 diabetes as significant but manageable. She emphasized that every event in this study was associated with a precipitating event but also noted that the participants were all educated about the risk and how to manage it, suggesting that patient education is not a cure-all. In type 2 diabetes, she noted that rates of DKA have been very low in clinical trials, including EMPA-REG OUTCOME, and that all of the cases she has seen clinically have occurred following surgery. She closed by sharing her personal clinical protocol for off-label use of SGLT-2 inhibitors in type 1 diabetes (see below). Our take-home message was that successful off-label use of SGLT-2 inhibitors in type 1 diabetes is clearly possible when the patient and the provider are both very engaged and well educated about the risk. However, we suspect that the FDA will require a more robust body of evidence and a clearer understanding of the contributing factors and mitigation strategies before it would be willing to approve a broad type 1 diabetes indication.

• Dr. Peters shared her personal protocol for off-label use of SGLT-2 inhibitors in type 1 diabetes. She stated that she has not seen any cases of DKA since developing this “recipe,” but stressed that this is her personal approach and that it is key to continue doing trials to fully characterize the class’ risks and benefits.
  
  • Demonstrated adherence to treatment is a prerequisite. Dr. Peters stressed that she does not prescribe SGLT-2 inhibitors to patients who have not been adherent to insulin therapy and SMBG in the past. She also asks patients to do fingerstick ketone testing every day for a week before initiating the SGLT-2 inhibitor to ensure that they will continue to do so once they are on the drug.
  • Start the SGLT-2 inhibitor at the lowest dose possible. Dr. Peters will often prescribe doses of 25 mg or 50 mg and tell patients to break their pills in half or in quarters. Depending on future clinical trial results, we could imagine such low doses becoming the standard for type 1 diabetes.
  • Reduce basal insulin by 10% and adjust the dose based on glucose and ketone levels. Dr. Peters said that in most cases, patients’ glycemic control will improve with no change in ketones but that it is important to test regularly in this initial phase to be sure.
  • Stop the SGLT-2 inhibitor and test ketones in the case of illness, intensive exercise, or other potential precipitating events. Dr. Peters does not require that her patients test ketones regularly once they are stable on the drug, but she advises them to test if they get sick, reduce their insulin dose, start training for a marathon, etc., and to hold the SGLT-2 inhibitor if ketones are high.
  • If ketones are high in the context of normal blood glucose (<150 mg/dl), treat with carbohydrates, insulin, and liquids. She stressed that patients need to consume enough carbohydrates to be able to take enough insulin dose to correct the ketosis without becoming hypoglycemic. She also advises patients to have a sports drink or other sugary beverage on hand to replenish fluids and glucose in case they become nauseous or start vomiting.

Questions and Answers

Q: DKA in type 2 diabetes is a misdiagnosis. DKA in type 1 diabetes is related to illness. The problem you reported was that the first patient reduced insulin during inflammation, and the person in the video reduced insulin by 25%. I think it’s a mistake to reduce basal insulin because ketoacidosis is a sign of insulin deficiency and the treatment is insulin. If you’re taking an SGLT-2 inhibitor, it’s a clear prandial approach, and the problems with ups and downs come from bolus insulin. Remember the hypoglycemia rate from bolus insulin is 8-10 times higher than at night. It comes from miscalculation of carbohydrates and corrections. If you treat type 1 diabetes with an SGLT-2 inhibitor, you should reduce the bolus only. Don’t reduce the basal, especially in cases of illness.

A: I used to reduce prandial insulin as well. In the trial they reduced basal insulin, which I find slightly easier since you don’t get hypoglycemia at night. All of my patients were self-adjusting their prandial insulin, so they are reducing both. You have to reduce both because otherwise you get hypoglycemia. That’s why I give more insulin and carbohydrates when patients get sick. You have to balance carbohydrates and insulin. I monitor ketones as patients reduce the insulin dose, to avoid ketosis. It’s all a balance. These are well-educated patients, and even doing the best they can when they take the drug variability improves. To me it’s about patients finding their lives easier to manage because there’s a more predictable response to each prandial dose. I’ve found an approach that works for me, but there’s far more than one way.

Q: There were some cases with empagliflozin and sotagliflozin with fairly draconian reductions in insulin doses. Was that also true for the canagliflozin cases?

A: That’s a good question. We’ve submitted a paper with the details. It wasn’t quite as draconian but there was a trend toward higher risk with greater reduction. That’s why I stressed that it’s a balance. You have to eat enough carbohydrates to give enough insulin. It’s education.

Q: In patients with type 2 diabetes with DKA, have you measured lactate?

A: I’ve measured lactate in patients I’ve seen with type 2 diabetes and it was elevated but not highly elevated. I’m not as familiar with what lactate levels are supposed to be in DKA. It was also higher in patients with type 1 diabetes. I’m not sure what to make of the data. That’s why we need to systematically collect data.

Q: What advice would you give to a patient with type 1 diabetes who exercises heavily? Do they have to reduce basal insulin?

A: I had a patient who went to Disneyland where you walk like crazy and it’s hot, and she went into DKA because she needed so much less insulin. It is the same with any unexpected exercise. I tell people don’t take the drug if you are going to suddenly increase your exercise. If you’re running a marathon, hold the drug. I’ve trained people not to take it on a hike because you don’t want to be at risk. DKA resolves quickly once you don’t take the drug. Not taking it for a day or two is far wiser.

Q: So it’s not a drug for very active patients?

A: That’s why I get baseline ketones. These patients are active, so I thought I would see a lot more ketones at baseline. I take care of lots of athletes. I’m not sure I would give it to patients if they were doing triathlons, but in the off-season they could take it. It can vary depending on the circumstances.

Dr. David Matthews: I want to make it clear that we’re talking about off-label use. That’s important because otherwise manufacturers get into trouble.

A: My slides said off-label every time. I take all responsibility for off-label use and fully document what I do. That’s why I keep stressing that we need clinical trials. If you follow what I say, you can’t say I recommended it per se, but I am sharing what I do to help others if they choose to use SGLT-2 inhibitors in an off-label manner. 

Combination of Insulin Degludec/Liraglutide (IDegLira): Key Results from the DUAL Programme

Optimising Individualised Therapy and Improving Outcomes

John Buse, MD, PhD (UNC, Chapel Hill, NC)

Dr. John Buse opened his lecture on therapy optimization by discussing patient preferences. He shared a meta-analysis of “what is important to the patient about drugs,” showing that treatment benefits (e.g., glycemic control, weight loss) top this list followed by side effects (e.g., nausea, diarrhea) and burden (e.g., route of administration, frequency). In an echo of his presentation from day #1 of EASD, he then introduced Novo Nordisk’s Xultophy (insulin degludec/liraglutide) as a novel alternative that exceeds expectations on all these fronts. He presented sub-analyses (new to us) from the phase 3 DUAL-1 extension trial showing broad efficacy irrespective of patients’ baseline A1c, BMI, or pre-trial medication: (i) improvements in glycemic control across all A1c categories vs. component parts; (ii) improvements in hypoglycemia rates across all A1c categories vs. insulin degludec; (iii) improvements in glycemic control across all BMI categories vs. component parts; (iv) improvements in hypoglycemia rates across all BMI categories vs. insulin degludec; and (v) equivalent improvements in glycemic control regardless of baseline medications. See below for the detailed breakdown. Dr. Buse acknowledged that these across-the-board benefits make it difficult to identify a population of type 2 patients who would be ill-suited to treatment with Xultophy. It’s hard to argue with these data (though Dr. Leszek Czupryniak did just that in a morning debate on the same day), and we agree that this combination class is one of the most exciting advances on the near-term horizon for type 2 diabetes. We expect that cost will be the most significant barrier to widespread adoption of the class, and the inability to titrate components individually could also be frustrating for some patients.

• Xultophy led to improvements in glycemic control across all A1c categories vs. component parts. As Dr. Buse pointed out, this was just as true for patients in good control.

• Xultophy led to improvements in hypoglycemia rates across all A1c categories vs. insulin degludec. This was true even in patients with low A1cs where we would expect relative parity.

• Xultophy led to improvements in glycemic control across all BMI categories vs. component parts.

• Xultophy led to improvements in hypoglycemia rates across all BMI categories vs. insulin degludec.

• Xultophy led to equivalent improvements in glycemic control regardless of baseline medications.

Commentator

Naveed Sattar, MD (University of Glasgow, UK)

Dr. Naveed Sattar shared frank commentary on IDegLira’s phase 3 DUAL program, questioning the incremental benefits of the combination over GLP-1 alone. Dr. Sattar acknowledged the advantages of IDegLira (one injection instead of two, slower GLP-1 titration, less hypoglycemia and weight gain than insulin, ~0.5%-0.9% lower A1c than the individual drugs), but focused mostly on his view of the drawbacks: the inability to titrate the drugs independently, less weight loss and more hypoglycemia than liraglutide alone, and cost. Dr. Sattar called for a trial comparing the fixed-combination of IDegLira to a free combination of the drugs (start with liraglutide and titrate up basal insulin to target) – in essence, minimizing insulin to preserve the weight and hypoglycemia advantages of liraglutide as much as possible. Of course, this view also ignores the burden of two injections, which Dr. Sattar seemed to underestimate. In the excellent Q&A (see below), Drs. John Buse and David Russell-Jones tended to agree with Dr. Sattar in broad strokes, though Dr. Buse noted what is perhaps the most remarkable stat in the DUAL program: ~80% of patients got to an A1c <7% with IDegLira. A GLP-1 alone is clearly a great option when possible, but for all those patients failing GLP-1 alone, failing basal insulin alone, or stuck with two injections per day, IDegLira seems like an excellent option (assuming cost is not prohibitive). Dr. Sattar was less convinced, concluding that IDegLira “makes sense” and “has a role in a modest number of patients,” but more studies are needed to judge it.

• Dr. Sattar complimented IDegLira’s nicely conducted registration trials, an excellent faculty, and robust quality of life/patient-reported outcomes. He said the combination of degludec and liraglutide “makes sense,” and recommended reading an editorial from Dr. Stefano Del Prato (Lancet Diabetes & Endocrinology 2014).
• Is IDegLira a means to put degludec “back in play?” Dr. Sattar called insulin degludec “too bloody expensive,” an opinion many of his colleagues also share. We would like the focus to be on the benefits of the dual therapy rather than pricing of one – we believe the pricing is high merely because Novo Nordisk knew it wouldn’t be covered in most places anyway but could start a set of conversations.
• Dr. Sattar summarized IDegLira’s “potential benefits” and what he finds “less convincing.” The below summarizes both his slides and talking points.

• Dr. Sattar proposed an “alternative clinical strategy,” starting with GLP-1 first and adding basal insulin independently. He reminded attendees that a few GLP-1s are now available, including once-weekly options. For those who do need basal insulin, Dr. Sattar recommended starting on top of GLP-1 and up-titrating the insulin – thereby minimizing hypoglycemia and maximizing weight loss (vs. the fixed-combination, which presets the insulin/GLP-1 doses). Dr. Sattar said IDegLira could be considered for those with persistent nausea on a GLP-1.
  
  • Dr. Sattar called for a trial comparing the fixed-combination of IDegLira to a free combination (start with liraglutide and titrate up basal insulin to target). Such a trial would compare all the standard outcome measures, including glucose, weight, nausea, hypoglycemia, and patient-reported outcomes. Dr. Sattar cautioned that this trial “requires some thought on how you make it fair” between the two arms.
• “Many require injectable agents at some point. Things have changed in my mind with EMPA-REG outcome. Where do we place SGLT-2s?” Dr. Sattar said that many are using SGLT-2s third and fourth line right now, but with empagliflozin’s positive CVOT results, the class may be used even earlier. Consequently, once patients get to injectables, they may be older and have more co-morbidities.

Questions and Answers

Dr. Buse: I generally agree with you. The advantages in an algorithmic way are much clearer vs. just insulin alone. That’s in opposition to GLP-1 alone as a choice. In the trials, 80% on IDegLira ended with an A1c <7%. The number that didn’t do well on IDegLira was quite small. But for those who go up 50 dose steps, what do you do then? That’s another remaining question. I think your comments are fair. There is likely a continued role for all three approaches. Starting with GLP-1, starting with insulin, and starting with the combination. Insulin alone is the weakest choice.

Dr. Sattar [responding]: I think that’s fine. [Laughter]

Dr. Russell-Jones: I agree with all. The use will vary country by country according to where it fits in. Speaking on the UK, what impressed me the most was people uncontrolled on basal insulin. This is a real potential way to go with intensification. The other group is the GLP-1s failures going to add insulin. This seems a sensible way to do it with lower hypoglycemia rates, etc. Those are the two areas in which this will have utility.

Dr. Buse: The other area I haven’t really thought about is when I see a patient that is very reluctant to go to the next step. This might be a patient where you might say – “This is the next step. You are very likely to get to where you want to be: below or near 7% without a lot of stumbles. For patients that are much more willing and adventurous, you could try one and switch to another. But if I only get one shot on goal, I do think this is the single best shot we have. Pun not intended.

Q: How many patients made it to the maximum dose?

Dr. Russell-Jones: In the first trial, it was around 50%.

Q: What do we do about that?

Dr. Russell-Jones: About 70% reached the target for those on the max dose. The number not reaching the target is small. There is an ongoing trial to look at a higher dose, or possibly a different combination.

Dr. Buse: I’m not sure if they were smart or lucky in the ratio of insulin and GLP-1, but it has worked out for the majority of patients. Some have wondered about a leaner ratio with more GLP-1, or a stronger ratio with more insulin for more powerful glucose lowering. Still, 80% of people getting to target is pretty good.

Q: I totally agree with Dr. Sattar on the trial design and outcome. But you might be forgetting one simple reason for the trials: there are a growing number of patients on insulin and GLP-1 taking two injections. Patient preference is a major player.

Sattar: You might be right, but you need to do that study. We’ve got more orals and more time between diagnosis and thinking about injectables. People are getting picked up earlier. They usually have some other co-morbidities. You’re right that patient preference might be one advantage of the combination. But until you do that trial, you don’t know if that is definitely the case.

Q: In Denmark, the Diabetes Association says that only a small fraction of spending is on drugs, and only 1.3% is on insulin. Sometimes we just fool ourselves in talking about cost. It just lets us say we don’t need to do anything, “It’s so costly...” As doctors, we are costly to the system. We need to be wider in our views.

Dr. Sattar: I agree. Costs, as a proportion, are modest. But the number of people with diabetes is going up. People are living longer. Drugs prices are going up. We can’t let this run away. What are the outcomes, and what do patients prefer? We don’t have enough data on outcomes. I would love to see long-term outcomes. If there is more hypoglycemia and less weight loss compared to liraglutide alone, what does that mean for long-term outcomes? I don’t know.

Michael Berger Debate: Do We Need Fixed Combinations of GLP-1 Agonists and Insulin?

Panel Discussion

Stephen Gough, MD (University of Oxford, UK) and Leszek Czupryniak (Medical University of Lodz, Poland)

Dr. Stephen Gough: I want to thank my colleague because I think it was a really entertaining talk. I hadn’t realized what a comedian you were. To bring it back to clinical science, I think you twisted the facts because the title was fixed-ratio combinations. The best way to treat diabetes might be to give a cup of tea but it doesn’t work. How many do achieve control when they titrate insulin? Maybe in your practice it’s different, but my patients don’t do it.

Dr. Leszek Czupryniak: Thanks for the compliment. That opens the door for my comedy career, though I would rather you say you were killed by my arguments and I have the finest mind in diabetes research. Seriously, I think we in central Europe and the UK differ in one measure and that’s the possibility of patient education. We like nurses, we like educators, people who work with patients on insulin and the mode of titration. The major barrier is initiating insulin. I was lucky enough in the past to spend time with Dr. Rury Holman and when I saw him working there he said “you need insulin,” and the patient probably knew who he was and said ok. In Poland we start negotiations that take weeks or months. Once they do it, we manage quite well to titrate insulin, especially basal analogs. Clearly we could do better with more nurses and education. I don’t see it as a major problem that we don’t get to targets with insulin. The next step is adding more prandial insulin, or we have GLP-1 agonists but the price is high. I did twist the facts a bit and used manipulation techniques, but that’s what debate is all about.

Q: What I don’t like is fixed-dose combinations. Why don’t we use once-weekly GLP-1 and have our hands free to titrate insulin?

Dr. Gough: It’s a good question. One of the things that we’re not saying is that one size fits all. What I’m saying is that we need to have a range of therapies so we can individualize the patient in front of us. While you may be right that fixed-dose combinations can be inflexible, clinical studies show that they work. The DUAL I study showed that 80% could get to target without weight gain and hypo. Of course, that won’t suit everybody but I think it’d be wrong to dismiss FDCs just because we can’t titrate separately.

Q: Do you think doctors should push more on the lifestyle side before entering the pharmaceutical realm?

Dr. Czupryniak: To be a bit more impartial, since we were playing our assigned roles here, I also think fixed-dose combinations have some benefits. One we haven’t mentioned is that patients take one injection with two agents. You can’t overlook that. It has nothing to do with individualization. I think we all feel a bit wary that we want to increase insulin or the GLP-1 agonist but we have to do the same with the other. We can’t compare it directly to a fixed-dose combination of oral drugs.

Q: What remains obscure is the side effects of the combination. The crucial point is whether it is paralleling those of insulin.

Dr. Gough: I’m not quite sure. In terms of side effects, we know there are side effects with insulin and with GLP-1 agonists, and if you take both together you are likely to get the side effects of both. When you take them the way they have been administered in trials, the side effects of each are less. You get the benefit of insulin and a GLP-1 agonist but you see less of the GI effects than with the GLP-1 agonist alone. It’s similar with basal insulin, where you expect to see weight gain and you know hypoglycemia is increased, and both of those are reduced or mitigated when they’re combined.

Q: Is there a logical explanation for that?

Dr. Gough: First of all, the final dose is less. From a GI point of view, there are probably lots of problems related to dose escalation, and it’s a simple titration algorithm with the fixed dose combination, so it’s a slow titration. We know there’s a glucose-sensitive mechanism with the GLP-1 agonist, which clearly offers a protective role when patients would be taking high doses of insulin.

Q: In a separate scenario, side effects would probably be bearable while free mixing?

Dr. Gough: It might be, but that’s not how we use them. We go in with a starting dose and then we go up to the bigger dose. Maybe we should be using GLP-1 agonists in a different way.

Q: Can you do any subtle titrations with GLP-1?

Dr. Gough: That’s where we tend to see more side effects.

Q: What about compliance and price?

Dr. Gough: I think compliance is good. There’s not clear data on that, but if you look at early patient-reported outcome data, it suggests it is well tolerated and patients prefer it to the individual components. On price, unfortunately the discussions always revert to drug acquisition costs. We don’t talk about the treatment costs of diabetes. We spend lots of money on diabetes and about 80% of it is on complications. Of course the acquisition costs are higher, though they’re not quite as high as you’ve explained because we don’t necessarily use those doses. But we should look at the treatment costs.

Dr. Czupryniak: The costs are substantial, and from a practical point of view we care more about the money spent by the patient, not the government. The cost will remain high because the drugs are effective and they’re new and you have to cover the expenses for development. Compliance is probably better, though I’d still argue that these are not necessarily drugs to use together in the majority of patients. I think it should be a sequential thing. It would be better for many patients to do earlier initiation of a GLP-1 agonist and maintain it further. In some of them some insulin will be necessary, and they might be eligible for a fixed combination. But a tight, fixed titration that doesn’t allow any changes is something where you might have serious doubts.

Oral Presentations: Biomarkers of Atherosclerosis

RVX-208 Acts Via An Epigenetic Mechanism to Lower Major Adverse Cardiovascular Events (MACE) in Patients with Atherosclerosis and Especially in Those with Diabetes Mellitus

Norman Wong, MD (Chief Scientific Officer, Calgary, Canada)

Dr. Norman Wong provided an overview of Canadian biotech Resverlogix’s BET bromodomain inhibitor RVX-208. The company arrived on our radar at JPM 2015, after it presented early evidence of a marked reduction of cardiovascular disease in a post-hoc analysis of patients with and without diabetes. Dr. Wong presentation shared this data again – a 64% reduction in the incidence of MACE in all patients (n=499) and an even greater 77% reduction in patients with diabetes (n=192). As we have noted before, the implications are intriguing though tough to wrap our head around as the mechanism of the risk reduction remains a question mark. Dr. Wong did note that RVX-208 therapy has been shown to lower glucose, though neither this nor the lipid effects can explain the full MACE benefit. He suggested that an inflammation-mediated effect is plausible, though the jury is still out given that RVX-208 is known to target multiple pathways with known roles in cardiovascular disease (e.g., coagulation). Ultimately, we will look forward to additional clinical data to shed further light on this field. As we learned at JPM 2015, RVX-208 is moving into a large phase 3 trial called BETonMACE that will enroll a pool of type 2 diabetes patients and with a primary endpoint of a 30% reduction in MACE. Showing cardioprotection in phase 2 is one thing; replicating those results in humans on the big stage (phase 3) is an entirely separate challenge. Considering the dire need for any means of cardiovascular risk reduction in diabetes, we have our fingers crossed.

Questions and Answers

Q: Did you perform in vitro studies?

A: We have done ex-vivo studies, but not in vivo studies.

Q: Which pathways are responsible for diabetes effect?

A: I don’t have a lot of data to back up what I’m saying, though research has shown that inhibitors can play a very important role in inflammation mediation. The cascading effect means that a lot of genes affect cardiovascular disease. In our own animal model, using a beta inhibitor decreases atherosclerosis hugely. We think inflammation is one big pathway. And coagulation and complement are others pathways that plays into this.

Is the Brain the New Pancreas?

Identified Sources and Effects of Insulin in the Cerebral Cortex

Gábor Tamás, PhD (University of Szeged, Hungary)

Dr. Gábor Tamás’s valuable overview of his neuroscience work provided insight – albeit often over our head – into the effect and interaction of insulin with neocortical neurons. The very in-depth presentation concluded with a number of takeaways that will likely be of interest to our basic science readers: (i) that insulin mRNA is expressed by neurogliaform interneurons of the rat cerebral cortex; (ii) that external insulin suppresses excitation in cortical circuits; (iii) that neurogliaform cells release insulin in response to glucose and glibenclamide; (iv) that human interneurons express insulin mRNA and proinsulin; and (v) that insulin expressing subpopulation of inhibitory interneurons are positioned to math neuronal activity and metabolic supply. Whew! You can access the details of Dr. Tamás’s talk here. The clinical implications remain an open question, but we are grateful to Dr. Tamás for opening our eyes to this exciting avenue of work [that has implications for our understanding of the intersection of Alzheimer’s disease and insulin as well].

Questions and Answers

Q: You explained that external insulin suppresses excitatory neurons. How can you explain that insulin deficiency results in brain stimulation? We usually find the opposite.

A: You have to think about the mechanism. What insulin actually does is that it turns up a part of cortical inhibition. This tonic inhibition is not equally well distributed over interneurons and pyramidal cells. You have to think about epileptic discharges work and the suppression of discharges. Epilepsy is not always excitatory overload. It can also be the result of suppressed inhibitory action. In this case, this is one of the scenarios I can think of.

-- by Melissa An, Adam Brown, Helen Gao, Varun Iyengar, Emily Regier, and Kelly Close