It was quite an insulin-heavy year at ADA, with a great deal of new phase 3 data for next-generation basal insulins that offer incremental benefits over first-generation basal analogs (e.g., less hypoglycemia, less weight gain, more convenient administration, or lower price) as well as significant progress on the ultra-rapid acting insulin front. You can find coverage of GLP-1/basal insulin combination products in our ADA 2014 Incretin-Based Therapies Report.
We saw new phase 3 data for Sanofi’s Toujeo (U300 insulin glargine), BI/Lilly’s new insulin glargine formulation (LY2963016), and Novo Nordisk’s Tresiba (insulin degludec) as well as a look into the mechanism of action for Lilly’s novel basal insulin peglispro (LY265541). Specifically on BI/Lilly’s investigational insulin glargine LY2963016 (LY), ADA 2014 was the first time data had ever been released (type 1 diabetes, 69-OR; type 2 diabetes, 64-OR), and it appears to act very similarly to Lantus. Speakers concluded that there were no clinically significant differences between the two products in either type 1 or type 2 diabetes. There was very little discussion of what ramifications biosimilar insulins could have, although Dr. Marcus Hompesch (Profil Institute, Chula Vista, CA) did speak during one session on opportunities and challenges with biosimilars, emphasizing that beginning a biosimilar insulin program requires a high “activation energy” and entails many manufacturing challenges. Turning toward Sanofi’s next-generation basal insulin Toujeo, all data from the global phase 3 program that had not already been presented in 2013 were released this year. In EDITION III (68-OR) and EDITION IV (80-LB) Toujeo conferred non-inferior A1c reductions as Lantus (as was the case in EDITION I and EDITION II), but Toujeo did not significantly reduce rates of hypoglycemia based on the pre-specified endpoints (in contrast to EDITION I and II). Also notably on Toujeo, one poster (919-P) presented results of a sub-analysis of patients taking Toujeo at fixed dosing intervals (every 24 hours) or flexible dosing intervals (24 ± 3 hours) and found the change in A1cs to be comparable. On Tresiba, Novo Nordisk presented a poster (402-P) in which it re-analyzed Tresiba’s nocturnal hypoglycemia data using several different definitions of nocturnal and hypoglycemia. It demonstrated that the company’s original claim that Tresiba has a nocturnal hypo benefit over Lantus, (which has been criticized by the FDA) was fairly consistent across different definitions. Another poster of interest (886-P) described the hepato-preferential mode of action for BI/Lilly’s peglispro from a small, open-label euglycemic clamp study. The clinical significance will be made clearer once detailed phase 3 results are released in 2015 (topline phase 3 results were released in May).
An entire session devoted to prandial insulin therapy shared data on promising ultra-rapid-acting candidates from Novo Nordisk (FIAsp), MannKind (Afrezza), and Biodel (BIOD-123). Dr. Tim Heise (Profil Institute for Clinical Research, Chula Vista, CA) presented pretty compelling PK/PD data showing that Novo Nordisk’s faster-acting insulin aspart (FIAsp, aka NN1218) has a clinically significant faster onset of action than its predecessor, Novolog (insulin aspart). The time to 50% maximum concentration for FIAsp was 27 minutes vs. 35 minutes for insulin aspart, which translated to a 22 mg/dl lower postprandial glucose excursions after one hour (26 mg/dl lower after two hours). Other orals in this session from Biodel and MannKind shared full results that augmented data we had previously seen – faster onsets of action, improved postprandial glucose excursions, less risk of hypoglycemia, and advantage (or decrement) in A1c. Based on the data, we have little doubt that these new insulins are indeed faster and offer some improvement over current prandial offerings – the bigger question is whether the incremental gains (in the absence of improved A1c) will be enough to support approval, reimbursement from payers, and sustainable pricing for companies and patients. We’ll find out soon with the recently approved Afrezza (see our report here.)
A valuable talk from Dr. David Klonoff (UCSF, San Francisco, CA) reminded us that ultra-rapid-acting insulins may not necessarily demonstrate lower A1c levels, but they do have meaningful clinical benefits: less postprandial hypoglycemia, less postprandial hyperglycemia, and less time spent out of glycemic range. However, he reminded attendees that the FDA’s 2008 Guidance for Industry on Diabetes Drug Development is quite black-and-white: “For the purposes of drug approval and labeling, the final demonstration of efficacy should be based on reduction in A1c, which will support an indication of glycemic control.” The aforementioned data on FIAsp, BIOD-123, and Afrezza supports Dr. Klonoff’s view, and we wonder how companies can think strategically about designing trials to show enough efficacy that approval and reimbursement are more certain (using CGM is a no-brainer in our view).
This document contains our coverage of presentations related to insulin therapy at ADA 2014. . Presentation titles highlighted in blue were not previously published in our daily highlight reports during ADA. We’ve highlighted in yellow presentations we found particularly notable.
- Executive Highlights
- Insulin Therapy
- Oral Presentations: Basal Insulin Therapy
- New Insulin Glargine 300 U/ml: Glycemic Control and Hypoglycemia in Insulin Naïve People with T2DM (EDITION 3) (68-OR)
- Similar Efficacy and Safety with LY2963016 Insulin Glargine Compared with Lantus Insulin Glargine in Patients with T1DM: The ELEMENT 1 study (69-OR)
- Similar Efficacy and Safety with LY2963016 Insulin Glargine Compared with Lantus Insulin Glargine in Patients with T2DM: The ELEMENT 2 Study (64-OR)
- Evaluation of Immunogenicity of LY2963016 Insulin Glargine Compared with Lantus Insulin Glargine in Patients with T1DM or T2DM (70-OR)
- Superior Glycaemic Control Effects with Insulin Degludec (IDeg) to Insulin Glargine (IGlar) in Diabetic Hemodialysis (HD) Patients Assessed by Continuous Glucose Monitoring (63-OR)
- Oral Presentations: Prandial Insulin Therapy
- Faster-Acting Insulin Aspart Improves Postprandial Glycemia vs. Insulin Aspart in Patients with Type 1 Diabetes Mellitus (T1DM) (129-OR)
- Safety and Efficacy of Ultra-Rapid-Acting Human Insulin Formulation BIOD-123 in Patients with Type 1 Diabetes
- Reduced Hypoglycemia Risk with an Inhaled Insulin Compared to Injected Prandial Insulin in Type 1 Diabetes
- Efficacy and Safety Evaluation of Technosphere Insulin vs. Inhaled Placebo in Insulin-Naïve Type 2 Diabetes (128-OR)
- Injection Depth Does Not Affect the Pharmacokinetics or Pharmacodynamics of Insulin Lispro in Healthy Obese or Normal Weight Subjects (131-OR)
- Rate Ratios for Nocturnal Confirmed Hypoglycemia with Insulin Degludec vs. Insulin Glargine Using Different Definitions (402-P)
- Glycemic Control and Hypoglycemia with New Insulin Glargine 300 U/mL in People with T1DM (EDITION IV) (80-LB)
- New Insulin Glargine 300 U/mL: Efficacy and Safety of Adaptable vs. Fixed Dosing Intervals in People with T2DM (919-P)
- Basal Insulin Peglispro Demonstrates Preferential Hepatic vs. Peripheral Action Relative to Insulin Glargine in Healthy Subjects (886-P)
- Insulin Therapy: Exploring Provider Perspectives on Needle Phobia and Nonadherence (685-P)
- Symposium: Closed-Loop Insulin Delivery – One Step at a Time (Supported by a grant from The Leona M. and Harry B. Helmsley Charitable Trust)
- Symposium: The Need for a Better Basal – What’s on the Horizon
- New Basal Insulins in Development
- What Should Be Required for Regulatory Approval of an Insulin?
- Biosimilars – Opportunities and Challenges
- Building a Better Basal – The Case for Concentrated Insulin
- Symposium: Hypoglycemia and Cardiovascular Disease – Lessons from Outcome Studies (Supported by an unrestricted educational grant from Merck)
- Symposium: Can We Limit the Long-term Decline of Beta Cells in Type 2 Diabetes?
- Symposium: Prevention of Hypoglycemia
- Symposium: Initial Treatment of Type 2 Diabetes – New and Not-So-New Ideas
- Symposium: New Frontiers in Inpatient Diabetes Management
- Corporate Symposium: Shared Decision-Making in Insulin-Initiation – Live Clinician-Patient Conversation on Overcoming T2DM Treatment Barriers (Sponsored by Sanofi)
- The Rationale for Early Insulin Initiation
- Panel Discussion
- Barriers to Insulin Therapy: The Clinician, the Patient, and the Insulin Itself
- Panel Discussion
- New Approaches to Insulin Therapy
- Panel Discussion
- Final Panel Discussion
- Corporate Symposium: Diabetes: Ensayos Clinicos, Tratamientos y Retos (Supported by an unrestricted educational grant from Novo Nordisk and presented only in Spanish)
- “Diabesidad” en el Adolescente (Diabesity in Adolescents)
- Repaso Crítico de la Seguridad en el Uso de las Insulinas Análogas en el embarazo, Niños, y Adolescentes (Safety Review of Insulin Analog Use during Pregnancy, in Children, and in Adolescents)
- Estudios con Nuevas Insulinas Análogas Basales: Últimos Avances (Latest Advances in New Basal Insulin Analogs)
- Administración de Insulina: Lo Que Usted Desconoce que Puede Dañar a su Paciente (Insulin Administration: What You Don’t Know Could Harm Your Patient)
- Panel Discussion
- Oral Presentations: Basal Insulin Therapy
Oral Presentations: Basal Insulin Therapy
Geremia Bolli, MD (University of Perugia, Perugia, Italy)
Dr. Bolli provided the full results of Sanofi’s EDITION III trial of U300 insulin glargine, elaborating on topline results from late last year. As a reminder, EDITION III was a six-month, randomized, open-label trial investigating the effects of the initiation of U300 (n=435) versus standard insulin glargine (n=438) in insulin-naïve type 2 diabetes patients with inadequate control on oral agents alone. Similar to trials EDITION I and EDITION II, results indicated non-inferiority in A1c reduction, with declines of -1.42% with U300 (baseline A1c 8.5%) and -1.46% with standard insulin glargine (baseline A1c 8.6%; p=ns). As noted in topline results, compared to EDITION I and II there was no significant reduction in the percentage of patients with ≥1 severe or confirmed nocturnal hypoglycemic events (≤70 mg/dl) from nine weeks to six months of treatment, at 15.5% with U300 versus 17.4% with insulin glargine (HR 0.89; 95% CI 0.66-1.20). However, this difference became significant in full results when the analysis was extended to include the full six-month study period, with 17.9% of patients experiencing ≤1 event with U300 versus 23.5% with insulin glargine (HR 0.76; 95% CI 0.59-0.99). Though suggestive of benefit, with the improvement in nocturnal hypoglycemia weighted to the initial weeks of treatment, it remains unclear if this will translate to a meaningful difference in the clinical setting – as patients in EDITION III likely had less severe diabetes compared to EDITION I and II patients, the hypoglycemia benefit of U300 may be limited to patients already more prone to hypoglycemia.
- EDITION III aimed to investigate the efficacy and safety of the initiation of U300 versus standard insulin glargine in insulin-naïve type 2 diabetes with inadequate glycemic control on oral antidiabetic medications. Patients were randomized to initiation of U300 (n=435) or standard insulin glargine (n=438) with follow-up of six months. Baseline oral agents were continued with the exception of sulfonylureas due to the potential for confounding hypoglycemia. Patients in both groups demonstrated similar baseline age (58.2 with U300 vs. 57.2 years with standard insulin glargine), BMI (32.8 vs. 33.2 kg/m2), and diabetes duration (10.1 vs. 9.6 years). Use of oral agents was similar between both groups at baseline as well (90.6% vs. 92.0% on metformin, 59.1% vs. 58.6% on sulfonylureas, and 20.7 vs. 22.4% on DPP-4 inhibitors).
- Similar to EDITION I and II, results indicated non-inferiority in A1c reduction, with declines of -1.42% with U300 (baseline A1c 8.5%) and -1.46% with standard insulin glargine (baseline A1c 8.6%) after six months of treatment (p=ns). Likewise, there was no difference in percentage of patients at target A1c <7.0%, at 43% with U300 versus 42% with standard insulin glargine (p=ns). Mean fasting plasma glucose levels were non-significantly different in both groups as well, declining from roughly 180 mg/dl to 120 mg/dl.
- As noted in topline results, compared to EDITION I and II there was no significant reduction in the percentage of patients with ≥1 severe or confirmed nocturnal hypoglycemic events (≤70 mg/dl) from nine weeks to six months of treatment, at 15.5% with U300 versus 17.4% with insulin glargine (HR 0.89; 95% CI 0.66-1.20). However, this difference became significant when the analysis was extended to include the full six-month study period, with 17.9% of patients experiencing ≤1 event with U300 versus 23.5% with insulin glargine (HR 0.76; 95% CI 0.59-0.99). Percentage of patients with ≤1 severe or confirmed hypoglycemic events was additionally significant at any time of day for the full six-month study period (46.2% vs. 52.5%; HR 0.75; 95% CI 0.57-0.99). By time of day, the difference in hypoglycemic events appeared most concentrated in the early morning with continuation to the early afternoon.
- Interestingly, there was a 17% increase in total insulin dose in patients treated with U300 versus standard insulin glargine at the end of the study, at 0.62 versus 0.52 units/kg/day – this has been consistent across the EDITION studies, though the clinical impact remains unclear. Change in body weight remained similar between both groups, at +0.4 kg (0.9 lbs) with U300 versus +0.7 kg (1.5 lbs) with insulin glargine (p=0.278).
Questions and Answers
Q: Can you clarify the time of day of dosing in both groups?
A: There was no question about the timing of dosing. Everyone in the program received basal insulin from anytime between dinnertime and bedtime – so only an evening injection.
Dr. John Buse (University of North Carolina, Durham, NC): At the end of the study, the A1c was identical but the dose with U300 insulin glargine was 17% higher. From your figure, it appeared most of that 17% was titrated after the eight-week mark. Thus, is it possible that the decline in fasting glucose was faster in the U300 arm?
A: A smart question from a smart man. We have data from self-monitoring of blood glucose that does not support your hypothesis, but it needs to be further analyzed.
Q: Why would the dose requirement be higher with U300?
A: The increased insulin requirement is consistent across studies – it was also shown in EDITION IV. The reason isn’t immediately clear. One hypothesis is that when you increase the concentration you slow down the absorption rate – this results in a more constant and pharmacokinetic release but with some loss of the initial effect. I think we need more studies to understand this effect.
Q: Would it be possible that U300 may be better called U250?
A: The official definition is that one unit of insulin is defined by the molecules of insulin in that volume. This is different than the clinical relevance of comparing units across therapies. So I don’t think it practically means anything to give 10% more or 10% less of insulin in units as long as the clinical outcomes are compared.
Thomas Blevins, MD (Texas Diabetes & Endocrinology, Austin, TX)
Dr. Thomas Blevins presented phase 3 data from the ELEMENT 1 study indicating that Lilly’s biosimilar LY2963016 insulin glargine had efficacy and safety profiles comparable to Lantus in patients with type 1 diabetes when combined with insulin lispro. In the study, LY2963016 insulin glargine demonstrated non-inferiority to Lantus in terms of A1c reduction; patients achieved A1c reductions of 0.4% with LY2963016 insulin glargine and 0.5% with Lantus at 24 weeks from a baseline of 7.8%, with 0.3% reduction for both treatments observed at 52 weeks. Thirty-five percent of LY2963016 insulin glargine users and 32% of Lantus users achieved A1c levels of 7.0% after 24 weeks; both user groups experienced a 62% incidence of adverse events at 52 weeks, and the total hypoglycemia rate stood at 77 events per patient per year with LY2963016 insulin glargine and 80 with Lantus. Dr. Blevins concluded that the two products in combination with lispro had similar efficacy and safety profiles with no clinically significant differences.
- The ELEMENT 1 study was a phase 3, open-label clinical trial involving 535 patients with type 1 diabetes who were randomly assigned to receive basal-bolus therapy with either the investigational glargine (N = 268) or Lantus (N = 267) in addition to insulin lispro. The study lasted 24 weeks, with a follow-up through 52 weeks. The patients had an average BMI of 25 kg/m2 and an average A1c of 7.8% at baseline, and all of them had been receiving treatment with insulin glargine, insulin detemir, or NPH prior to the study.
- The results of this study demonstrated that LY2963016 insulin glargine is non-inferior to Lantus in terms of the primary endpoint of A1c reduction. After 24 weeks, the group receiving the investigational glargine had achieved an A1c reduction of 0.35% compared to a 0.46% reduction for the group receiving Lantus. Additionally, 35% of the participants in the investigational glargine group had reached the target A1c of 7% compared to 32% of the Lantus group.
- The investigational glargine also demonstrated non-inferiority on a number of secondary endpoints related to safety and efficacy. There was no significant difference between the groups in fasting plasma glucose reduction, body weight change, or daily insulin dose. The frequency of adverse events was 62% in both groups after 52 weeks, and both groups had similar rates of hypoglycemia (77 events/patient/year with LY2963016 and 80 with Lantus after 24 weeks), allergic events (8% of people receiving LY2963016 and 4% of people receiving Lantus), and antibody responses (9% of the LY2963016 group and 6% of the Lantus group).
Questions and Answers
Q: When looking at safety data like insulin antibodies, I’m not interested in mean levels, I want to know if there are a few people who are severely affected. Your presentation addresses the wrong question. Are there 1-2% who have severe problems?
A: Let me defer to the next presentation where you’ll hear more about how antibody levels and response were measured.
Julio Rosenstock, MD (Dallas Diabetes and Endocrine Center, Dallas, TX)
Dr. Julio Rosenstock presented the phase 3 results of the ELEMENT 2 study, which compared the safety and efficacy profiles of Lilly’s investigational LY2963016 insulin glargine to the currently marketed Lantus insulin glargine. The study demonstrated non-inferiority of LY2963016 compared to Lantus in terms of A1c reduction; patients in the LY2963016 group (N = 376) achieved an average A1c reduction of 1.3% after 24 weeks compared to a 1.4% reduction for the Lantus group (N = 380) from a baseline of 8.3%. 49% of LY2963016 insulin glargine users reached the target A1c of 7%, compared to 53% of Lantus users. The frequency of adverse events was also similar – 52% for LY2963016 insulin glargine and 48% for Lantus – as was the number of hypoglycemic events, with 21 events per patient per year for LY2963016 insulin glargine and 22 for Lantus. Dr. Rosenstock concluded that the two insulin glargine formulations have equivalent efficacy and safety profiles and that there is no clinically significant difference between them.
- The ELEMENT 2 study was a phase 3, double-blind clinical trial involving 759 patients with type 2 diabetes who were randomly assigned to receive either the investigational glargine (N = 376) or Lantus (N = 380) for 24 weeks. The average age of the participants was 59, average weight was 90 kg, and baseline A1c was 8.3% for both groups. All patients had received treatment for diabetes for at least 2 years, 60% with insulin and 40% with oral medications.
- The results of this study demonstrated that LY2963016 insulin glargine is non-inferior to Lantus in terms of the primary endpoint of A1c reduction. Both groups had an average A1c of 8.3% at baseline; the average at the end of the study was 7.0% for the group receiving the investigational glargine compared to 6.9% for the Lantus group. Additionally, 49% of the participants in the investigational glargine group reached the target A1c of £ 7.0% compared to 53% in the Lantus group.
- The investigational glargine also demonstrated non-inferiority on a number of secondary endpoints related to safety and efficacy. There was no significant difference between the groups in fasting plasma glucose reduction, body weight change, or daily insulin dose. The frequency of adverse events was 52% with the investigational glargine compared to 48% with Lantus, and both groups had similar rates of hypoglycemia (21 events/patient/year with LY2963016 and 22 with Lantus) and allergic events (6% of people receiving LY2963016 and 7% of people receiving Lantus).
- Dr. Rosenstock was clear that he considers this product to be a biosimilar insulin glargine, despite the regulatory complexity surrounding this designation. He explained that although it will not be classified as a biosimilar by the FDA, it meets the scientific criteria of (i) high similarity to the reference product; (ii) minor differences in clinically inactive components; and (iii) no clinically meaningful differences in terms of safety, purity, and potency.
Questions and Answers
Q: Very nice study. I have one question about antibody response: it’s hard to measure antibodies with a high degree of sensitivity, so can you tell us more about the assay you used?
A: I can’t give you specifics on that, but there will be a presentation later that I encourage you to attend.
Q: This was very nice, congratulations. I have one comment about your second slide where you referred to the regulatory designation in the US and “other geographies.” “Other geographies” includes many, many other countries with different rules!
A: Yes, so in my judgment, it is a biosimilar because it has all the specific criteria. I was speaking from a regulatory perspective; here it has to go through a different pathway, so it’s just regulatory terminology. In essence, scientifically, LY is a biosimilar.
Q: Biosimilar guidelines suggest looking for biosimilarity in the most sensitive population – in this case, patients with type 1 diabetes. In this study, you’ve selected the population that would be least likely to show differences, not people who’ve been using insulin for a long time.
A: We had people who’ve been using insulin for 10 years. 60% of participants were insulin naïve, 40% were previously on insulin – we took the whole type 2 spectrum. For a change, you’re wrong!
Mark Deeg, MD, PhD (Eli Lilly, Indianapolis, IN)
Dr. Mark Deeg presented specific immunogenicity data from the phase 3 ELEMENT 1 and ELEMENT 2 studies of Lilly/BI’s insulin glargine LY2963016 – immunogenicity is one of the key concerns with so-called insulin biosimilars. Overall, there were no worrying differences in overall immunogenicity. Most importantly (and reassuringly), in both studies, the presence or absence of such an antibody response did not affect the degree of A1c-lowering efficacy or any clinical outcomes. For type 1 diabetes patients in ELEMENT 1, a similar percentage of LY2963016 users had detectable anti-insulin antibodies at both baseline (17%) and at 52 weeks (40%) to patients on Lantus (21% and 39% at baseline and 52 weeks, respectively). ELEMENT 2 found similar parallels: 6% of LY2963016 users at baseline and 15% at week 24 showed detectable insulin antibodies, compared to 4% and 11% for Lantus. Out of all the data presented, only at one time point in one study (week 4 in ELEMENT 2) did LY2963016 demonstrate a significant difference in detectable antibodies relative to Lantus, and there the difference was slight and only barely significant (p = 0.047).
Questions and Answers
Q: Did you look at the antibody patterns in ELEMENT 2 between insulin-naïve patients and previously insulin treated subjects?
A: There was a similar treatment-emergent antibody response between both of those groups.
Q: Did you look at neutralizing antibodies?
A: We did not run specific neutralization antibody assays. Most importantly, there were no clinically meaningful differences in outcomes.
Satoshi Funakoshi, MD, PhD (Nagasaki Renal Center, Nagasaki, Japan)
Dr. Satoshi Funakoshi presented the results of a small trial comparing the effects of Novo Nordisk’s Tresiba (insulin degludec) and Sanofi’s Lantus (insulin glargine) using CGM in seven patients with poorly controlled type 2 diabetes (A1c > 8.0%) also on hemodialysis (HD). Hemodialysis has been shown to have an adverse effect on glycemic control in diabetes patients, as glucose levels drop during hemodialysis and have a tendency to rebound afterwards due to counterregulatory hormone action (including glucagon). The study’s results were displayed as overlaid CGM traces. On Lantus, most patients saw the characteristic drop in glucose levels during HD, followed by a sharp rebound immediately afterwards. By comparison, there was no readily apparent drop in glucose during HD with Tresiba, and no visible rebound in most patients. Dr. Funakoshi characterized this flattening of variability as quite surprising for an insulin – we would agree that the difference in the curves was quite striking. Potential causes of this effect include Tresiba’s stability and high binding to albumin (which would minimize loss during HD). From this data alone, it appears that Tresiba may be uniquely suited to use in patients on hemodialysis, at least relative to Lantus. Key study limitations include small size, inter-subject variability, and the sequential exposure to Lantus before Tresiba.
- Insulin is one of the relatively few treatment options available to type 2 diabetes patients with kidney failure. Drug classes such as SGLT-2 inhibitors and (most) DPP-4 inhibitors are contraindicated in this patient population. Additionally, end-stage renal disease is responsible for substantial cost, as well as reduced quality-of-life for patients. According to the CDC’s 2014 Diabetes Statistic Report (which we covered recently), in 2011 a total of 228,924 diabetes patients in the US were on dialysis or received a kidney transplant due to kidney failure.
- Hemodialysis generally leads to temporary reduction in blood glucose, followed by a sharp rise. Glucose, a relatively small molecule, can pass through the dialysis membrane to a greater degree than insulin (a much larger polypeptide). Following the end of HD, blood glucose levels can spike due to a glucagon-driven counterregulatory response. Adding additional variability, insulin can adsorb to the dialysis membrane, and the different binding properties of insulin analogs can affect this binding.
- Methods: Dr. Funakoshi’s study enrolled seven poorly controlled (A1c > 8.0%) type 2 diabetes patients on hemodialysis who were already on treatment with Lantus (baseline doses of 8 – 24 units). After 72 hours of therapy with Lantus (on both on and off-HD days), patients were switched to Tresiba at the same dose. Glycemic control was evaluated using CGM. Concomitant antidiabetic drugs and HD schedules stayed the same throughout the study.
- Results were presented as CGM trace overlays, and the results were strikingly positive in favor of Tresiba. Dr. Funakoshi first displayed data for days in which patients did not receive HD, and even here, it appeared that Tresiba led to less glycemic variability later in the day than Lantus. On HD days, patients on Lantus saw the characteristic dip in blood glucose of 50 – 150 mg/dl during hemodialysis, followed by a substantial rebound (three of the patients peaked at well above 300 mg/dl). By comparison, those trends were not at all visible in patients on Tresiba – no patients crossed 300 mg/dl following hemodialysis.
- Dr. Funakoshi suggested that the glycemic stability seen with Tresiba could be due to the drug’s PK/PD stability, as well as the fact that it has a high degree of binding to albumin (~99%). Lantus, by comparison, does not have the same binding affinity for albumin. This factor is important, as albumin is a very large protein that is not lost to a great extent during dialysis or adsorbed to the dialysis membrane.
- We would have liked to learn whether CGM was patient-blinded in the study, or whether any of the patients had prior experience with CGM. Those factors could have had a significant impact on the outcome of the study. All patients were exposed to Tresiba later in the trial, following their exposure to Lantus. If patients were able to see the CGM traces, patients learning how to interpret CGM, rather than just the comparative effects of Tresiba could have affected the results of the study. The fact that patients’ doses of insulin and other medications were held constant throughout the study helps alleviate this potential bias to some extent.
- Other potential study limitations include the very small number of subjects, the lack of evaluation of plasma insulin or other hormones, and inter-subject variability.
Questions and Answers
Q: At what time of the day did patients take the insulin?
A: Insulin was taken at breakfast time. I think it might be easier to take ultra-long-acting insulin right after hemodialysis.
Q: We know that insulin degludec has a slower absorption rate, which accounts for much of the longer action. There is also evidence that there may be differences in clearance. Do you have information on whether the action was further prolonged in patients with impaired renal function?
A: I do not think so. The steady state is almost the same compared to patients with normal renal function.
Oral Presentations: Prandial Insulin Therapy
Tim Heise, MD (Profil Institute for Clinical Research, Chula Vista, CA)
Dr. Tim Heise presented pretty compelling PK/PD data showing that Novo Nordisk’s faster-acting insulin aspart (FIAsp, aka NN1218) has a clinically significant faster onset of action than its predecessor, Novolog (insulin aspart). This early-stage study in people with type 1 diabetes (n=36) found that the time to first appearance for FIAsp after co-administration with a standardized meal was just 3.6 minutes compared to 9.8 minutes for insulin aspart (it is quite notable that the insulin was dosed at the same time as the meal, since this is likely when many patients actually take their prandial insulin, rather than the 15-30 minutes prior to eating that is optimal for current rapid-acting analogs). The time to 50% maximum concentration for FIAsp was 27.3 minutes compared to 34.5 minutes for insulin aspart. The earlier onset also translated into greater blood-glucose lowering in the first two hours after insulin administration and meal challenge. After one hour, the postprandial blood glucose excursion was, on average, 22 mg/dl lower on FIAsp than on insulin aspart, and after two hours, it was 26 mg/dl lower. Glucose area under the curve after two hours was 26% lower with FIAsp than with insulin aspart, and area under the curve after six hours was 33% lower. Notably, FIAsp achieved this earlier onset of action with no increase in the rate of hypoglycemia (or hyperglycemia) or any other safety or tolerability concerns.
- FIAsp’s structure and formulation: FIAsp’s molecular structure is identical to the insulin analog aspart (Novo Nordisk’s Novolog), which means that the proline at position B28 in the native insulin peptide has been substituted with aspartic acid. Its excipient formulation is what makes it different from Novolog. FIAsp has two excipients: nicotinamide, a vitamin B3 that is an absorption modifier, and arginine, a naturally occurring amino acid that is a stability enhancer. Dr. Heise highlighted that both of the excipients are on the FDA ingredients list for approved drug products for injection. Thus, with the established safety profile of Novolog, FIAsp should have very few safety concerns.
- The study presented was a randomized, double-blind, cross-over, meal challenge study comparing the PK/PD of FIAsp with insulin aspart (n=36 people with type 1 diabetes). Patients received a single dose of FIAsp or insulin aspart (0.2 U/kg) together with a meal (600 kcal standardized liquid meal consisting of 67% carbohydrate, 17% protein, and 16% fat). Prior to dosing, blood glucose was normalized to 100 mg/dl using IV insulin infusion (stopped no later than five minutes before study drug dosing). After a 3-12 day washout period, patients received another meal test, this time with the opposite drug than they received at the first meal test.
- FIAsp had a faster onset of appearance and greater exposure during the first two hours after dosing. FIAsp’s mean onset of appearance was just 3.6 minutes after dosing compared to 9.8 minutes for insulin aspart (a highly statistically significant difference). The time to 50% maximum concentration for FIAsp was 27.3 minutes compared to 34.5 minutes for insulin aspart.
- Total drug exposure (area under the curve over 10 hours), maximum concentration, and time to maximal concentration were essentially equivalent between FIAsp and insulin aspart. This means that FIAsp’s PK curve was simply shifted to the left of insulin aspart. In other words, it has a faster onset but few other PK differences.
- The earlier onset of appearance translated into greater blood-glucose lowering in the first two hours after administration and meal challenge. After one hour, the postprandial blood glucose excursion was, on average, 22 mg/dl lower on FIAsp than on insulin aspart, and after two hours, it was 26 mg/dl lower (statistically significant). Glucose area under the curve after two hours was 26% lower with FIAsp than with insulin aspart (statistically significant).
- Total glucose area under the curve over six hours was reduced by 33% with FIAsp compared to insulin aspart.
- The incidence of hypo- and hyperglycemia requiring intervention was similar between FIAsp and insulin aspart. Investigators intervened at 50 mg/dl and 306 mg/dl for hypo- and hyperglycemia, respectively. On FIAsp, 20% of patients required hypoglycemic intervention compared to 23% on insulin aspart. On FIAsp, there were zero interventions for hyperglycemia, and there was one for insulin aspart.
Questions and Answers
Q: What was the actual time between the injection and the beginning of the meal?
A: About one minute. Basically, they injected and started to eat.
Q: What are the requirements to get faster-acting aspart approved by FDA?
A: The regulatory requirement is similar to that of other drugs. Because the molecule is the same, there was no necessity for a phase 2 study. So basically after the phase 1 studies, the company directly moved directly into phase 3 and these trials are ongoing. It will be like any other drug.
Q: What is the objective of the phase 3 program?
A: First objective is obviously safety and efficacy under more real-world conditions. So the phase 3 studies will investigate safety data and will look into A1c, hypoglycemia, possibly also quality of life and other things.
Q: What was the time to maximum insulin concentration vs. aspart?
A: If I recall correctly it was about 1 hr for both – the time to maximum was not significantly different.
Alan Krasner, MD (Biodel, Danbury, CT)
Biodel’s Dr. Alan Krasner presented full results on the phase 2 study of its ultra-rapid-acting human insulin, BIOD-123 – topline data was shared in September 2013. Our takeaway after seeing these full results was that data interpretation is challenging for many reasons – there were large baseline group imbalances in gender, differences in basal insulin doses, and an open-label design. Overall, BIOD-123 was non-inferior to Humalog as measured by change in A1c. There was a 0.17% treatment difference in favor of Humalog (baseline A1c: 7.3%), though the 95% confidence interval [-0.01, 0.35] just barely met the non-inferiority margin of 0.4%. [Note: this was highly similar to what was observed in the phase 3 trial of MannKind’s Afrezza, which had a 0.19% treatment difference in favor of insulin aspart and a [0.02, 0.36] confidence interval that also just barely met the non-inferiority endpoint.] Ten-point profiles and CGM did not demonstrate consistent postprandial glucose differences between BIOD-123 and Humalog, despite a clear postprandial benefit observed during a liquid meal challenge test (~15 mg/dl improvement with BIOD-123). There was higher injection site discomfort with BIOD-123, a problem that has plagued Biodel since Linjeta – however, these events disproportionately came from certain trial sites, suggesting the possibility of ascertainment bias (the trial was an open-label design). Less basal insulin was used in the BIOD-123 arm, and there was a non-significant trend towards higher overnight glucose levels with BIOD-123 – this could suggest basal insulin was sub-optimally titrated in the BIOD-123 arm. Overall, lots more questions than answers flowed out of this study, and it is perhaps not a surprise that Biodel has not firmed up a partnership on this asset.
Questions and Answers
Q: There were differences in meal tests in favor of Biodel. But there were no difference in real life. Was the timing of injection before the meal different?
A: They should have been the same. The instructions were to dose the prandial insulin immediately before the meal – in the liquid meal challenge and also at home. We do believe that is practically when most patients take prandial insulin – just before meals.
Q: Together with the previous speaker, ultra-rapid-acting insulins showed lower postprandial glucose excursions. In your study, there not a change in A1c. What level of postprandial reduction is needed to see change in A1c? Perhaps 10-15 mg/dl may not be enough to see a change in A1c. Second, I’m a clinician – how important is this in my clinical day-to-day management?
A: I don’t know the answer to your first question. I don’t know how much postprandial reduction you need for an A1c reduction. Our data is a little complex because of imbalances by chance in dosing. It’s hard to interpret the A1c data. The liquid meal challenge was more controlled circumstances. Glucose was measured by a lab analyzer. It’s possible that some of the home-based readings result from variability and noise in measurements. And differences in dosing that contributed as well.
I don’t know the answer to using it in clinical practice. I would guess these insulins would be used as rapid-acting analogs are used now. We often say take insulin right before meals. We know based on recently published data that may not be the optimal time to take lispro and aspart. Good ultra rapid-acting analogs may be ideal for taking just before meal.
Dr. Melanie Davies (University of Leicester, UK): Thank you very much. There is some promise here, but there is still some work to do.
Bruce Bode, MD (Atlanta Diabetes Associates, Atlanta, GA)
Dr. Bruce Bode shared full results from the phase 3 trial comparing mealtime Afrezza (Technosphere Insulin, or “TI”) to insulin aspart in patients with type 1 diabetes – we previously covered this data in detail in our preview of the April 1 FDA Advisory Committee for Afrezza, as well our full report on the Advisory Committee. In type 1 diabetes, Afrezza was non-inferior in A1c reduction vs. insulin aspart (-0.2% vs. -0.4%; baseline: 7.9%), a difference that was maintained over 24 weeks. Afrezza had an advantage over aspart for both moderate and severe hypoglycemia (-30% and -43%, respectively) – the benefit came in the period two to five hours post-meal, suggesting the fast-in/fast-out profile reduced late postprandial hypoglycemia. Dr. Bode characterized Afrezza as “weight neutral” vs. the weight gain observed in the insulin aspart arm (-0.4 kg with Afrezza vs. a +0.9 kg with aspart). Meanwhile, Afrezza was associated with a “manageable adverse event profile” – the most common event was cough, which was “mild and transient.” Dr. Bode’s talk was factual and not full of opinion, but it certainly refreshed many of the outstanding questions around this trial in type 1 diabetes: high dropout rates in the Afrezza arm (attributed to study demands, and patients with longstanding diabetes resistant to change), statistical A1c inferiority to aspart (though Afrezza just barely met the non-inferiority margin of 0.4%), the dosing flexibility of Afrezza in type 1 diabetes, and perhaps most importantly, how patients will use this in the real world. Q&A was particularly illuminating.
- In the primary efficacy results, Afrezza was non-inferior to insulin aspart at the pre-specified A1c margin of 0.4%. A1c declined by 0.2% in the Afrezza group vs. 0.4% in the insulin aspart group, both from a baseline of 7.9%. The treatment difference of 0.19% translated to a p-value of 0.016 and a 95% confidence interval of [0.02-0.36] – just sneaking under the non-inferiority margin of 0.4%.
- Overall, Afrezza led to a 30% reduction in total hypoglycemia (9.8 vs. 14 events per subject-month) and a 43% reduction in severe hypoglycemia (8 vs. 14 events per 100 subject months) vs. insulin aspart [p<0.05 for both]. Over the 24-week study period, Afrezza had a lower total hypoglycemia event rate than insulin aspart for every A1c category (5.5-6.5%, 6.5-7%, 7-7.5%, 7.5-8%, >8%). The difference was greatest in the population with an A1c of 7-7.5%, where Afrezza led to a 40% reduction in hypoglycemia (10.5 vs. 17.4 events per subject-month). The smallest benefit of Afrezza on hypoglycemia came in those with an A1c >8% – a 13% reduction (8.6 vs. 9.9 events per subject-month).
- Afrezza and insulin aspart had nearly identical rates of total hypoglycemia from 0-2 hours post meal; the clear advantage for Afrezza came in the 2-5 hour post-meal window, where it had a much lower rate of hypoglycemia for all 24 weeks of the study. Importantly, during the study’s follow-up period (weeks 25-28) – when Afrezza use was discontinued and all patients reverted to insulin aspart – the 2-5 hour hypoglycemia advantage disappeared. This supports the conclusion that Afrezza’s shorter tail of insulin action reduces the occurrence of delayed post-meal hypoglycemia.
Questions and Answers
Q: What was the dose of basal insulin?
A: The average basal insulin dose was about 30 units with aspart, which went up to about 35 units with inhaled insulin. There was a significant drop in fasting plasma glucose with inhaled insulin.
Dr. Tim Heise (Profil, Neuss, Germany): I have many questions on the data. Most importantly, would you recommend your type 1 patients go on TI despite a significantly worse A1c vs. aspart?
A: A1c was not significantly worse by the non-inferiority criteria. We looked at it in every which way.
Q: But the confidence interval didn’t include zero…
A: But it was less than the non-inferiority margin of 0.4%. There are lots of ways to give prandial insulin, and whatever the patient does best with is what you choose.
Q: Why was there so much withdrawal of consent in the TI vs. the control group?
A: About 9% withdrew because of adverse events – 5%+ were related to cough. These people all had longstanding type 1 diabetes. They didn’t like the idea of using inhaled insulin once they started. They quickly dropped out early – not late. The aspart arm and both group dropped out because of e-diaries and frequent adjustment of insulin.
Julio Rosenstock, MD (Dallas Diabetes and Endocrine Center, Dallas, TX)
Dr. Julio Rosenstock presented the results of one of the two most recent phase 3 trials on MannKind’s inhalable insulin Afrezza (Technosphere Insulin; TI) – these two trials were a direct outcome of the FDA’s last Complete Response Letter for Afrezza. The trial Dr. Rosenstock presented investigated Afrezza in 177 type 2 diabetes patients on oral diabetes medications, compared to a cohort on the Technosphere powder excipient alone (n = 176). We summarized these results in our preview for Afrezza’s most recent FDA Advisory Committee meeting, as the meeting briefing documents covered the data from this trial (see section on Trial 175). Nothing very new or surprising emerged during the presentation. Highlights of the data included a seemingly modest 0.4% A1c difference between groups (although interestingly the Technosphere powder comparator group saw a 0.42% absolute A1c reduction from baseline that made Afrezza’s relative efficacy seem relatively low) and a significant 130% increase in hypoglycemia. Dr. Rosenstock attributed the hypoglycemia to the high use of SFUs in the trial, and the overall incidence of hypoglycemia was fairly low for a rapid-acting insulin. Cough was the most common side effect (generally dry and transient, and also associated with the Technosphere powder alone), and there was a very slight decrease in FEV1 (a measure of lung function).
- For a complete chronicle of the most recent Afrezza FDA Advisory Committee meeting, read our report. After the meeting, the FDA announced a three-month extension of Afrezza’s PDUFA date to July 15, 2014 (read our report).
Questions and Answers
Q: From what I could see, there was in fact a reduction in FEV1 that was due to the insulin. What is the explanation for that?
A: It is an interesting observation, but you have to remember that we’re looking at a very miniscule change in volume. But yes, it could very well be driven by the insulin.
Q: Can you give us more detail on the background therapy for these patients?
A: The background therapy is important, as it may explain the hypoglycemia findings. Around two-thirds were on metformin and a sulfonylurea, and it is highly possible that if we had fewer patients on sulfonylureas there would have been less hypoglycemia.
Q: Can you comment on patients who may have been smokers?
A: We had so many screen failures because we did not take any patients with a history of smoking or who were active smokers. With studies of previous inhaled products, it looked like smoking in fact increased absorption.
Q: Were there any patients with any type of pulmonary illness?
A: Any people showing a history of obstructive pulmonary disease were excluded.
Q: How is TI impacted by exercise?
A: I am not sure if they have done a study on exercise with TI. I believe that studies were done with Exubera, and that there was increased absorption during exercise, but this is a totally different mechanism of action.
Amparo de la Pena, PhD (Eli Lilly, Indianapolis, IN)
Dr. de la Pena presented data confirming no differences between the 5-mm and 8-mm injection depths for insulin lispro for a variety of pharmacokinetic and pharmacodynamics parameters. Of note, the 8-mm – but not the 5-mm – injection depth was performed with a pinch-up. Two separate studies were performed: one in North America studying healthy obese participants and one in East Asia studying normal weight participants. Neither study enrolled people with diabetes. While the sample sizes were small (n=16 for both studies), the studies allowed the research groups to collect more mechanistic data. Both studies showed nearly identical immunoreactive insulin and glucose infusion rate curves with the 5-mm and 8-mm injection depths, and no statistically significant differences were observed with any of the parameters characterizing the curves.
- The data stem from two studies conducted at different times, at different sites, and with different populations, and neither study looked at patients with diabetes. Both studies used a randomized two-period crossover design where each participant received insulin injections at 5-mm and 8-mm injection depths. The first study (n=16) looked at normal weight subjects with a mean age of 31 years and mean BMI of 23 kg/m2 in a predominantly East Asian population; the second study (n=16) looked at healthy obese participants with a mean age of 41 years and mean BMI of 34 kg/m2 in a predominantly Caucasian population. Of note, the 8-mm – but not the 5-mm – injection depth was performed with a pinch-up.
- In both studies, the pharmacokinetic and pharmacodynamics curves for insulin injection were nearly identical between the 8 mm and 5 mm injection depths. To measure pharmacokinetics, the investigators administered an injection of insulin lispro and collected blood samples from the participant at time points ranging from 10 to 360 minutes to test for immunoreactive insulin lispro-specific activity (IRI). In the normal-weight subject study, the IRI curves for the 8 mm and 5 mm injections were completely superimposed, showing no differences. For the healthy obese subject study, the 8-mm injection depth was delayed by about 10 minutes at early time points; however, this was not statistically significant and the curves overlapped for the later time points. To measure pharmacodynamics, subjects underwent hyperglycemic clamp and glucose infusion rate was measured after insulin lispro injection. The glucose infusion curves for both injection depths were nearly identical, with an approximately five minute delay for the 8-mm depth in healthy obese subjects that did not reach statistical significance. Accordingly, no statistically significant differences were observed between injection depths for any of the parameters derived from the curves.
Questions and Answers
Q: You seem fairly confident that these results will apply to diabetes patients. How sure are you that this will match clinical observations?
A: There is definitely room for more investigation. We hope to compare between normal weight and obese subjects within a single study. I am not aware of any differences in skin thickness between diabetic and non-diabetic individuals.
Q: Identical findings were shown over twenty years ago using radiolabelling and ultrasound just under the skin, but almost nothing has been done since then. It’s nice to see this done with a modern insulin analog. There have been 7 or 8 different studies showing no difference with pens of different lengths. Small question. Where and how did you give the injections? It seems that 8 mm might get close to muscle.
A: We used a pinch up for the 8 mm and no pinch up for the 5 mm, and we alternated injection sites.
Q: There seems to be a much lower Cmax and a longer tmax in obese patients.
A: We did not have a direct comparison between the two studies. They are too different: different sites, different times, different populations. I would love to do a study where we have both populations in the same study.
Q: Can you comment on tRmax being earlier in both studies?
A: They do look like that at the beginning, but the curves come back together, and the difference is not statistically significant.
S Heller, C Mathieu, R Kapur, ML Wolden, B Zinman
This poster presents the results of a post-hoc analysis by Novo Nordisk to determine the robustness of their previous finding that treatment with ultra-long acting insulin degludec (trade name Tresiba) led to significantly lower rates of nocturnal hypoglycemia than treatment with insulin glargine (Sanofi’s Lantus) in patients with type 2 diabetes and numerically lower rates in patients with type 1 diabetes. This analysis was likely fueled by FDA criticism of the methods used in the original meta-analysis in degludec’s registration packet. This new study conducts several analyses using different definitions of nocturnal hypoglycemia including i) only confirmed episodes with symptoms; ii) the ADA definition; and iii) a different time frame for the nocturnal period to show that the original findings remain robust no matter which definition of “nocturnal” or “hypoglycemia” is used. The results of these analyses confirmed the findings from the original meta-analysis under nearly all of these conditions. The one exception was when the nocturnal period was extended to 0:01-7:59, in which case hypoglycemia was reduced only in the population of patients with type 2 diabetes treated with basal-bolus therapy (and not in type 1 diabetes or in insulin-naïve type 2 diabetes). Under all other conditions, treatment with insulin degludec led to significantly lower rates in all patients with type 2 diabetes and to numerically but not significantly lower rates in patients with type 1 diabetes. The table below summarizes the rate ratio and 95% confidence intervals for all conditions (rate ratio of 1 indicates an equal rate of hypoglycemia, <1 indicates a lower rate with insulin degludec, >1 indicates a lower rate with insulin glargine). Overall, the data seems to indicate that treatment with insulin degludec may lead to significantly lower rates of nocturnal hypoglycemia in patients with type 2 diabetes compared to treatment with insulin glargine, though the less impressive results with the time period 0:01-7:59 do provide some reason for cautious skepticism.
Table: Rate ratio for nocturnal confirmed hypoglycemia, insulin degludec/insulin glargine
Type 2 insulin-naïve
Type 2 basal-bolus
Nocturnal confirmed hypo, original definition (0:01-5:59)
0.64 [0.48, 0.86]
0.75 [0.58, 0.99]
0.83 [0.69, 1.00]
Nocturnal confirmed symptomatic hypo (0:01-5:59)
0.56 [0.39, 0.80]
0.68 [0.51, 0.91]
0.88 [0.72, 1.08]
Nocturnal ADA documented symptomatic hypo (0:01-5:59)
0.73 [0.56, 0.97]
0.72 [0.55, 0.93]
0.91 [0.74, 1.11]
Nocturnal confirmed hypo, original definition (21:59-5:59)
0.60 [0.45, 0.80]
0.73 [0.59, 0.91]
0.88 [0.76, 1.03]
Nocturnal confirmed hypo, original definition (0:01-7:59)
0.93 [0.75, 1.15]
0.77 [0.60, 0.97]
1.00 [0.86, 1.17]
- This was a post-hoc meta-analysis of six 24- or 52-week randomized, controlled, open-label phase 3a trials involving patients with type 1 and type 2 diabetes and using several definitions of nocturnal hypoglycemia. Definitions included i) confirmed symptomatic episodes; ii) symptomatic episodes with plasma glucose £70 mg/dl (the ADA definition); and iii) the original definition with a different time frame for the nocturnal period (21:59-5:59).
- Insulin-naïve patients with type 2 diabetes treated with basal-only insulin had significantly lower rates of nocturnal hypoglycemia when treated with insulin degludec (N=1279) than with insulin glargine (N=631), using those three definitions. The rate ratios and 95% confidence intervals with the three definitions listed above were i) 0.56 [0.39, 0.80]; ii) 0.73 [0.56, 0.97]; and iii) 0.60 [0.45, 0.80], compared to 0.64 [0.48, 0.86] in the original meta-analysis. A rate ratio of 1 indicates equal rates of hypoglycemia, a ratio <1 indicates a lower rate with insulin degludec, and a ratio >1 indicates a lower rate with insulin glargine.
- Patients with type 2 diabetes on basal-bolus therapy had significantly lower rates of nocturnal hypoglycemia when treated with insulin degludec (N=742) than with insulin glargine (N=248), using those three definitions. The rate ratios and confidence intervals with the three definitions were i) 0.68 [0.51, 0.91]; ii) 0.72 [0.55, 0.93]; and iii) 0.73 [0.59, 0.91], compared to 0.75 [0.58, 0.99] in the original meta-analysis.
- Patients with type 1 diabetes had numerically but not significantly lower rates of nocturnal hypoglycemia when treated with insulin degludec (N=637) than with insulin glargine (N=316), using those three definitions. The rate ratios and confidence intervals with the three definitions were i) 0.88 [0.72, 1.08]; ii) 0.91 [0.74, 1.11]; and iii) 0.88 [0.76, 1.03], compared to 0.83 [0.69, 1.00] in the original meta-analysis.
- Additional analysis using 0:01-7:59 as the nocturnal period demonstrated an advantage of insulin degludec over insulin glargine only for patients with type 2 diabetes on basal-bolus therapy. The rate ratios and confidence intervals were 0.93 [0.75, 1.15] for insulin-naïve patients with type 2 diabetes, 0.77 [0.60, 0.97] for patients with type 2 diabetes on basal-bolus therapy, and 1.00 [0.86, 1.17] for patients with type 1 diabetes.
- Additional analysis of the maintenance period only (after the initial 16-week titration period in each trial) showed that all patients with type 2 diabetes had significantly lower rates of hypoglycemia with insulin degludec than with insulin glargine using all definitions and that patients with type 1 diabetes had significantly lower rates with insulin degludec only with the original definition.
PD Home, RM Bergenstal, MC Riddle, M Ziemen, M Rojeski, M Espinasse, GB Bolli
This study presented the primary results from the EDITION IV phase 3a trial of Sanofi’s new U300 insulin glargine. In the trial, patients with type 1 diabetes on background basal-bolus therapy (n=549) were randomized 1:1:1:1 to once-daily U300 or standard insulin glargine in either the morning or evening while continuing mealtime insulin. As noted in the topline results, U300 was non-inferior to standard insulin glargine in reducing A1c levels (mean change -0.40% with U300 vs. -0.44% with standard insulin glargine; baseline 8.1%) at the end of six months of treatment. Rates of any time confirmed or severe hypoglycemia (<70 mg/dl) were not different between the two groups, although U300 users had reduced nocturnal hypoglycemia in the first eight weeks of treatment (HR 0.69; 95% CI 0.53-0.91). These results are similar to EDITION III, in which benefit to nocturnal hypoglycemia was weighted to the initial weeks of treatment – while different from EDITION I and II, it remains unclear if this will represent a meaningful benefit overall. We are curious if the inclusion of morning administration was able to reduce the risk of nocturnal hypoglycemia overall. Notably, we do note that rates of nocturnal hypoglycemia from week eight to six months of treatment was not a pre-specified main secondary endpoint for this study.
- This global, multi-center, open-label study, patients with type 1 diabetes (n=549) were randomized 1:1:1:1 to either U300 insulin glargine or standard insulin glargine in either the morning or evening. Average baseline A1c was 8.1%, average BMI was 27.6 kg/m2, and average diabetes duration was 21 years for both U300 (n=274) and insulin glargine (n=275) groups. Patients were followed over a six-month period. As noted across the EDITION studies, the total insulin dose at the end of the treatment period was slightly higher for U300 users (+0.19 units/kg/day from baseline) compared to standard glargine users (0.10 units/kg/day from baseline).
- U300 demonstrated non-inferior A1c reductions compared to standard insulin glargine (mean change -0.40% with U300 vs. -0.44% with standard insulin glargine; baseline 8.1%). There were no significant differences between the morning and evening groups.
- Rates of any time confirmed or severe hypoglycemia (<70 mg/dl) were not different between the two groups, although U300 users had reduced nocturnal hypoglycemia in the first eight weeks of treatment (HR 0.69; 95% CI 0.53-0.91). Rates of hypoglycemia were equal between the morning and evening groups. Overall, severe hypoglycemia was seen in 6.6% of the U300 users versus 9.5% of the standard glargine users.
- Notably, similar to EDITION II, patients taking U300 gained significantly less weight versus standard glargine users (difference -0.56 kg [1.2 lbs]; p=0.037). U300 users gained an average of 0.5 kg (1.1 lbs), while insulin glargine users gained an average of 1.0 kg (2.2 lbs).
MC Riddle, GB Bolli, PD Home, R Bergenstal, M Ziemen, I Muehlen-Bartmer, M Wardecki, L Vinet, H Yki-Jarvinen
Dr. Matthew Riddle and colleagues conducted two sub-studies of Sanofi’s insulin glargine (Lantus) 300 U/ml comparing the effects of fixed dosing (FD) vs. adaptable dosing (AD) in 198 type 2 patients. The sub-studies were part of two larger, six-month open label studies comparing glargine 300 U/ml to glargine 100 U/ml: EDITION 1 (basal insulin plus mealtime insulin; n=53 for FD and n=56 for AD) and EDITION 2 (basal insulin plus oral anti-diabetic medications; n=44 for FD and n=45 for AD). To generate the sub-studies, participants who completed the glargine 300 U/ml on-treatment during the main trials were re-randomized to either FD (injections at 24 hour intervals) or AD (injections at 24 ± 3 hr intervals; participants were asked to use an injection interval of exactly 21 hours or 27 hours at least twice a week). Endpoint measurements were taken at month nine (three months after the re-randomization) and the intent-to-treat analysis included data from 194 patients. In both sub-studies, the primary endpoint – change in A1c – was comparable between the FD and AD groups. Similarly, the FD and AD groups had comparable rates of adverse events, overall hypoglycemia, and nocturnal hypoglycemia. Based on this data, the authors conclude that type 2 patients who occasionally adapted the timing of their glargine 300 U/ml injections did not compromise the safety or efficacy of the drug.
- Baseline characteristics were comparable between the FD and AD groups within each sub-study, and between the two sub-studies: average age of 57-61 years, A1c of 7.2-7.5%, and percent male of 43-50%.
- Variability in the timing of injections between the FD and AD groups were measured by recording the time between two consecutive injections during the last seven days before the two endpoint assessments, which occurred one-and-a half and three months following the re-randomization. In the FD group, a low percentage of patients administered injections outside of a 24 ± 1 hr interval (13% for the EDITION 1 sub-study and 11% for the EDITION 2 sub-study). As would be expected, larger percentages were recorded for the AD group (37% and 48%, respectively). Notably, a fraction of patients in the AD group injected their insulin more than three hours above or below the standard 24-hour interval (14% for EDITION 1 and 19% for EDITION 2).
- In both sub-studies, the FD and AD groups experienced similar changes in A1c, fasting plasma glucose, and eight-point SMPG profiles after three months (see table below; note: A1c increased slightly in EDITION 1 sub-study). Furthermore, the participants made only small changes to their mean daily basal insulin dose, and these changes were similar across the FD and AD groups in both sub-studies.
Change in A1c
Mean difference between groups
Change in Fasting Plasma Glucose
Mean difference between groups
- In both sub-studies, the FD and AD groups had similar rates of overall and nocturnal (midnight to 6 am) hypoglycemia (see table below). Only one event of severe hypoglycemia was reported in the sub-studies (in the FD group of Edition 1).
Percent experiencing any hypoglycemia*
Percent experiencing nocturnal hypoglycemia*
* numbers estimated from graph
RR Henry, S Mudaliar, SL Choi, TP Ciaraldi, DA Armstrong, J Pettus, P Garhyan, MP Knadler, SJ Jacober, ECQ Lam, H Linnebjerg, N Porksen, MJ Prince, and VP Sinha
Dr. Robert Henry et al. conducted a single-center, randomized, open-label trial comparing the sites of action of Lilly’s basal insulin peglispro (referred to as LY2605541) vs. Sanofi’s insulin glargine (Lantus) in eight healthy male participants (mean baseline age of 26 years and BMI of 24 kg/m2; all had fasting plasma glucose <108 mg/dl ). The study measured the drugs’ abilities to suppress endogenous glucose production (EGP, which reflects their actions on the liver), as well as their abilities to stimulate the glucose disposal rate (GDR, which reflects their actions outside the liver – i.e., peripheral action). The participants underwent four eight-hour euglycemic clamp studies (maintained at 90 mg/dl): the first three with primed, continuous infusions of LY2605541 (five doses ranging from 5.1 to 74.1 mU/min), and the fourth with insulin glargine (either 20 or 30 mU/m2/min). The investigators used D-[3-3H]-glucose infusion to assess EGP and GDR. Suppression of EGP and stimulation of GDR were observed with increasing concentrations of both insulins. Notably, the LY2605541 dose needed for 100% EGP suppression had little effect on GDR. In contrast, the glargine dose required for a comparable suppression of EGP led to an increase in GDR. These results indicate that in healthy males, LY2605541 exhibits greater hepato-preferential action compared to insulin glargine.
Insulin Therapy: Exploring Provider Perspectives on Needle Phobia and Nonadherence (685-P)
J Krall, K Williams, R Gabbay, L Siminerio
In a BD-supported study, 23 primary care providers and three pharmacists were interviewed to assess their familiarity with and use of smaller and shorter needles as a solution for addressing problems with insulin therapy adherence. A full 70% of the physicians (n=16) reported that needle phobia is the primary challenge to initiating insulin therapy, and a striking 87% of physicians (n=20) stated that the availability of smaller needles would be an important factor in persuading patients to start injections. However, few physicians were familiar with the smallest needle available (BD’s 32-gauge, 4 mm Nano needle) or the fact that shorter needles can be used in any patient regardless of weight. Only 39% (n=9) reported prescribing a specific needle – most physicians instead deferred to default options in prescribing systems or assumed that a pharmacist would choose the best needle. To add insult to injury, the pharmacists in the survey reported referring decisions to PCPs. The authors argue, and we agree, that provider education will need to be revisited in order to increase awareness of these options and ultimately improve patient adherence to insulin therapy.
Symposium: Closed-Loop Insulin Delivery – One Step at a Time (Supported by a grant from The Leona M. and Harry B. Helmsley Charitable Trust)
Do We Need More Rapid Acting Insulins?
David Klonoff, MD (UCSF, San Francisco, CA)
Dr. David Klonoff systematically answered the question posed in his talk title with a resounding “yes,” organizing his thoughts by answering six questions (see below). He showed physiologic, pharmacologic, and clinical data, along with “common sense” to support the benefits of more ultra-rapid-acting insulins: less postprandial hyperglycemia, less late postprandial hypoglycemia, possibly less weight gain, and less glycemic variability. Most importantly, he emphasized that ultra-rapid-acting insulins may not necessarily demonstrate lower A1c levels, something we’ve now seen with MannKind’s Afrezza, Biodel’s BIOD-123, and Halozyme’s Hylenex. We were ecstatic to hear Dr. Klonoff call for study endpoints that look beyond A1c, something that is not currently incorporated into the FDA’s 2008 guidance on developing diabetes drugs.
- Dr. Klonoff on the FDA’s 2008 Guidance for Industry on Diabetes Drug Development: “For the purposes of drug approval and labeling, the final demonstration of efficacy should be based on reduction in A1c. If a drug can improve outcomes other than A1c, then this result would be like trying to fit a square peg in a round hole. Ultra-rapid-acting insulin cause less postprandial hypoglycemia, less postprandial hyperglycemia, and less glycemic variability, but the FDA uses A1c as the only valid endpoint for a diabetes drug.” Dr. Klonoff wondered, “If A1c is the only outcome measure for an insulin, should we avoid a faster acting insulin and stick with a slower acting human regular insulin?” Dr. Klonoff cited Dr. Doug Muchmore’s 2011 JDST article that called for “a composite endpoint that integrates A1c and hypoglycemia risk.” Said Dr. Klonoff, “It’s not just about the A1c… For you pharma companies, if your only endpoint is A1c, you’re going to have a problem. Have some other endpoints.”
- How does rapid-acting insulin affect mean glycemia and glycemic targets? Dr. Klonoff summarized the answer to this nuanced question quite concisely: rapid-acting insulin has little effect on mean glycemia and a small benefit on postprandial hypoglycemia. He cited some negative reviews comparing rapid-acting insulin to regular human insulin – a Cochrane collaboration review and Germany’s IQWiG both didn’t see a meaningful clinical benefit of using rapid-acting insulin. Still, Dr. Klonoff cited an observational study from the University of Colorado, which tracked A1c and severe hypoglycemia prior to and following the DCCT. While A1c improved during the DCCT, the rate of severe hypoglycemia correspondingly increased. Following the introduction of Humalog in 1996, A1c continued to improve, though there was no added increase in severe hypoglycemia.
- How does more rapid-acting insulin affect glucose in closed loop systems? Dr. Klonoff highlighted the 2008 hybrid closed-loop work of Dr. Stu Weinzimer (Diabetes Care), which used pre-meal boluses to cover 25-50% of the meal – pre-meal boluses led to a lower mean glucose, lower postprandial glucose, and no increase in hypoglycemia. Dr. Klonoff also mentioned the closed-loop work with MannKind’s Afrezza at UCSB/Sansum – post-meal time in zone was significantly better with use of Afrezza prior to meals.
- How can insulin be made to act more rapidly? Dr. Klonoff mentioned the following methods and insulins in development: Novo Nordisk’s FIAsp (currently in phase 3; strong PK/PD data was presented on Day #2 of this meeting); Thermalin’s Fluorolog (ultra-rapid U500); Biodel’s portfolio of candidates (phase 2 BIOD-123 data presented on Day #2); MannKind’s Afrezza; a dissolving microneedle patch (no specific company mentioned); Halozyme’s hyaluronidase; Insuline’s InsuPad (“a nice product” that heats the skin and is being used successfully in Germany); and Roche’s intraperitoneal DiaPort (“not making much progress at this point”).
- How rapidly is insulin released in response to a meal in healthy people? Based on data from as early as 1968 (Curry et al., Endocrinology), healthy people release insulin within a few minutes before/after eating. Dr. Klonoff highlighted the difference between first-phase and second-phase insulin release, noting that even a low dose of early insulin can make a big difference in postprandial glycemia. In people without diabetes, insulin is released when food hits the mouth, far before the glucose hits the blood. Though the cephalic phase of insulin release only represents 1-3% of meal insulin, without it, “control is poor.”
Questions and Answers
Dr. Howard Wolpert (Joslin Diabetes Center, Boston, MA): We should also talk about approval by insurers – they are all fixated on A1c. So we need to get them looking at postprandial glucose as well. Should we suggest some kind of standardized testing protocol?
A: We need new tests and new metrics. The doctors who understand insulin therapy need to work with the FDA to develop additional endpoints. If A1c is the only metric that is considered, then fewer highs and fewer lows is not going to be appreciated.
Symposium: The Need for a Better Basal – What’s on the Horizon
New Basal Insulins in Development
Chantal Mathieu, MD, PhD (KU Leuven, Leuven, Belgium)
Dr. Chantal Mathieu, MD, PhD (KU Leuven, Leuven, Belgium) presented an overview of upcoming basal insulins, focusing on insulin degludec (Tresiba, Novo Nordisk) and Lilly’s peglispro (LY2605541). Dr. Mathieu reviewed data from insulin degludec phase 3 clinical trials and peglispro phase 2 clinical trials. Two themes emerged. First, degludec and peglispro have both demonstrated non-inferiority to insulin glargine (Lantus, Sanofi), in both type 1 and type 2 diabetes patients. Second, both degludec and peglispro tended to reduce nocturnal hypoglycemia compared to glargine in both type 1 and type 2 diabetes patients. Degludec and peglispro also have flatter pharmacodynamic and pharmacokinetic profiles than insulin glargine as a result of longer durations of action. Dr. Mathieu also touched on Sanofi’s U300 triply concentrated insulin glargine, noting that simply increasing the concentration gives a flatter profile and longer duration of action with noninferior A1c reductions. Novel administration procedures and smart insulins (glucose-responsive insulins) were also briefly mentioned.
- According to Dr. Mathieu, current basal insulin therapy has limited flexibility and duration of efficacy. Effects are highly variable between patients and within patients. Current insulins must be administered at the same time every day. They also have problems being mixed with other products and do not always last 24 hours, which she said is the optimal time period from a practical point of view.
- For both type 1 and type 2 diabetes patients, insulin degludec is non-inferior to insulin glargine in reducing A1c, and has a longer half-life, flatter pharmacodynamic and pharmacokinetic profiles, and reduced nocturnal hypoglycemia. In a recent meta-analysis, type 2 diabetes patients had a 32% reduction in nocturnal hypoglycemia risk and type 1 diabetes patients had a 17% reduction in nocturnal hypoglycemia risk when they used insulin degludec, compared to insulin glargine.
- To test dosing flexibility, Dr. Mathieu and Novo Nordisk came up with an “arbitrary design” that alternated dosing times between 40 hours and 8 hours for an entire week. The flexible degludec dosing arm was compared with a fixed dosing degludec arm and insulin glargine arm. The erratic dosing schedule had no deleterious effect: flexible insulin degludec still showed noninferiority versus glargine in terms of efficacy, and still was associated with lower rates of nocturnal hypoglycemia. Similar results were seen in type 1 diabetes patients, with one key difference: flexible dosing was actually associated with 27% lower rates of nocturnal hypoglycemia compared to fixed dosing (p=0.000). Dr. Mathieu addressed this strange finding during the question and answer session (see below).
- Insulin peglispro (LY2605541) phase 2 trials demonstrated more mixed results compared to insulin glargine. Peglispro was noninferior to glargine in terms of A1c change in type 2 diabetes patients (n=176), and was associated with significantly less nocturnal hypoglycemia. Additionally, peglispro users lost weight while insulin glargine users gained weight, likely due to peglispro’s preference for targeting the liver (which has its problems, as acknowledged by Dr. Mathieu). Peglispro actually demonstrated a superior A1c reduction (-0.6%) over insulin glargine (-0.4%) in type 1 diabetes patients (p<0.001), as well as less nocturnal hypoglycemia; however, incidence of overall hypoglycemia was slightly higher with peglispro vs. glargine.
- Basal insulins in the pipeline are being designed with reduced dosing, less frequent administration, and glucose responsiveness in mind. For example, Sanofi is developing a triply concentrated insulin, degludec holds promise for thrice weekly dosing, Hanmi insulin could potentially be administered once a week, and smart insulins may be able to offer glucose responsive activity.
Questions and Answers
Q: You showed CGM data on peglispro but failed to show data on insulin degludec.
A: The people from Novo Nordisk say that at the time when the studies were done, the technology didn’t allow them to come to clear conclusions and that data were all over the place. That’s what they told me. That’s all I can tell you.
Q: In your study on type 1 patients, I didn’t see any difference between the fixed degludec arm and the glargine arm in terms of nocturnal hypoglycemia. Can you elaborate on that?
A: That study is the only study where there was no advantage of nocturnal hypoglycemia for degludec. What I think happened is that we had a surplus of nocturnal hypoglycemia in the first four weeks because we switched people on twice-a-day detemir, for instance, to dose for dose for degludec. We probably overdosed the degludec in the beginning. The doses for degludec even came down in the first weeks of titration, whereas the glargine doses went up.
Q: Do we really need very long-acting insulins? Many patients have different requirements for insulin during the night and daytime. Is 24 hours really a good time?
A: Yeah, I think 24 hours is a good time. Longer than 24 hours, I have the same doubts.
Q: In the US, pharmacies are changing the basal insulins very frequently. So I think that when people are changing from one insulin to another, there needs to be an established ratio to determine dosing.
A: Substituting one analog for another is something that needs to be undertaken by a healthcare provider.
What Should Be Required for Regulatory Approval of an Insulin?
Eric Brass, MD, PhD (UCLA David Geffen School of Medicine, Los Angeles, CA)
Dr. Eric Brass offered his thoughts on the regulatory process for new basal insulins, reminding the audience that “we have a step between research and reaching patients.” After a brief review of the current FDA guidance requiring new diabetes biologics to meet specific efficacy criteria in high-quality clinical trials, he used the saga of Novo Nordisk’s ultra-long acting insulin degludec (see our report here) to illustrate many of the challenges facing companies that want to develop a new basal insulin product. He explained that although it is fairly straightforward to demonstrate non-inferior glycemic control, the uncertainty about how the FDA will weigh the risk/benefit of effects beyond A1c (hypoglycemia, weight, administration and dosing, etc.) makes the approval process very unpredictable and challenging. He noted that benefits that would be significant for many patients, such as convenient administration and reduction in non-severe hypoglycemia, are either not well captured by typical clinical trials or are not considered important by the FDA. The FDA’s new emphasis on cardiovascular risk has presented a major obstacle for the development of new diabetes drugs; insulin degludec received a complete response letter primarily because the phase 3 data did not sufficiently exclude the possibility of an unacceptable cardiovascular risk. Dr. Brass explained why a typical phase 3 trial is not designed to meet the FDA’s expectations for excluding unacceptable risk, leading to the necessity of conducting specific cardiovascular outcomes trials. He concluded by saying that a new consensus is needed on what constitutes an important clinical benefit, that pragmatic, real-world trials may be a more effective way to demonstrate those benefits than a typical controlled trial, and that the challenges in defining the benefits and safety of new insulin products “may be stifling innovation. I think it already has.” Dr. Brass’ suggestions would surely provide greater clarity to drug developers on regulatory expectations and, thus, improve the efficiency of the development process.
- Dr. Brass explained that meeting the FDA’s primary endpoint for trials of a new basal insulin – effectiveness in lowering A1c – is conceptually straightforward. He hastened to add that “straightforward doesn’t mean easy,” as the FDA requires large, highly controlled clinical trials that involve a full spectrum of patients with type 1 and type 2 diabetes. The phase 3 trials for Novo Nordisk’s Tresiba (insulin degludec), for example, involved over 6,800 patients with type 2 diabetes and over 2,100 with type 1; they included both insulin-naïve and previously insulin treated patients; and they used almost all of the available drug classes as comparators.
- In Dr. Brass’ opinion, it is far more complicated to demonstrate that a new basal insulin has clinical benefits over existing therapies. First of all, it is very difficult to demonstrate superior glycemic control in an intensely monitored clinical trial in which all participants are titrated to goal. So-called “convenience benefits” of dosing and administration, though very important to patients and providers, are also masked in clinical trials, where adherence is strictly enforced. Measuring reduction in hypoglycemia, another extremely important issue for patients, is complicated by various factors, including a lack of clarity from the FDA on what constitutes clinically important hypoglycemia (for example, they did not consider degludec’s nocturnal hypoglycemia reduction claim to be clinically significant) and the fact that most hypoglycemia in type 1 diabetes is caused by the short-acting bolus insulin rather than the basal.
- Dr. Brass discussed the FDA’s new cardiovascular safety requirements for diabetes drugs at length, explaining that a typical phase 3 trial is not sufficiently powered to exclude cardiovascular risk. In the case of insulin degludec, the point estimate of the hazard ratio for cardiovascular events in a meta-analysis of all phase 3 trials was 1.39 (below the FDA’s pre-approval threshold of 1.8), but the upper bound of the confidence interval was 2.565, so the FDA couldn’t exclude the possibility of a high increase in risk. The confidence interval was so wide because there were so few total cardiovascular events (39 with degludec and 15 with the comparator), so Novo Nordisk is now required to perform a cardiovascular outcomes trial that is specifically designed to evaluate these risks.
- Because of the huge amount of uncertainty about how the FDA views the benefits and risks of new basal insulins, Dr. Brass concluded that drug developers may need to develop new clinical trial frameworks. He emphasized the need for consensus on the important clinical benefits of new basal insulins and suggested that in order to most effectively measure them, “maybe we shouldn’t be doing well controlled clinical trials.” As an alternative, he suggested “pragmatic trials” that more closely approximate daily life for patients and providers. He also said that the FDA’s expectations about excluding unacceptable cardiovascular risk will require programs specifically designed to ensure that enough events take place to make meaningful inferences.
Questions and Answers
Q: What would the definition of a hypoglycemia benefit be for a new basal insulin?
A: The FDA has said nothing about anything other than severe hypoglycemia, and their comments at the Advisory Committee meeting on insulin degludec suggested that the only hypoglycemia they accept as important is severe. That doesn’t mean they don’t have something else in their own thinking, but in terms of rejecting new programs, severe hypoglycemia is the only thing they’ve suggested they’ll accept as clinically meaningful. They didn’t accept confirmed nocturnal hypoglycemia as clinically important with insulin degludec but they would have accepted severe. Where else along that spectrum they would accept, we just don’t know.
Q: What did the FDA ask Novo Nordisk to do to show clinically significant benefit?
A: Nothing publicly. The decision is always benefit vs. risk. In this case, there was no perceived differentiating benefit, and uncertainty about risk, so they said Novo Nordisk needed to remove the uncertainty about risk. If we could establish unique benefits, the uncertainty about risk would become less of a problem, but it doesn’t seem like they accepted any benefits.
Q: I think perhaps it’s time for the FDA to come up to 2014 standards. Using CGM glycemic variability would make much more sense as a measure of glycemic control. What are your thoughts on that?
A: These are complex things. What the FDA and I would insist on is that anything used to establish a benefit must be of unambiguous clinical importance—it can’t just look nice. Other aspects for glycemic control improvement beyond A1c I think have yet to reach unambiguity that translates to clinical meaning. Where I see immediate opportunity is the development of pragmatic trials that translate to how drugs are used to establish superior glycemic control and open-mindedness about what’s important for patients and clinicians.
Biosimilars – Opportunities and Challenges
Marcus Hompesch, MD (Profil Institute for Clinical Research, Chula Vista, CA)
Dr. Marcus Hompesch provided a concise, well-organized overview of the challenges facing companies that hope to enter the biosimilar insulin market in the near future. He anticipates that many companies will want to pursue these products, as the demand for insulin is skyrocketing and several branded insulins are facing a patent cliff. He warned, however, that beginning a biosimilar insulin program requires high “activation energy” – a large initial investment without certainty of approval. He also discussed the challenges of manufacturing such a complex molecule, reminding the audience that making a biosimilar is not a “copy and paste” process and that subtle structural differences can affect the product’s efficacy and safety. Dr. Hompesch also pointed to regulatory uncertainty as one of the major challenges in this area, as guidelines for biosimilar insulins vary from country to country and are nonexistent in some. The “most important rule” in his mind is that it is the applicant’s responsibility to justify the product to regulators at every step of the process. “So,” he warned, “good luck.” Despite these obstacles, Dr. Hompesch ended on a fairly optimistic note, listing the many companies with biosimilar insulin programs and expressing hope that these products could reduce costs for patients and improve efficiency in the health care system.
- Dr. Hompesch believes the biosimilar insulin market offers great opportunities for companies and that this will be a “crowded space” in the coming years. As we are all too aware, the demand for insulin is exploding as the diabetes pandemic grows, and several market leaders are facing the end of their patent protection (Humalog last year, Lantus in 2015). The combination of these forces is likely to lead many pharmaceutical companies to pursue biosimilars in the near future, he said. Dr. Hompesch also expressed hope that this competition will drive innovation and lead to cost savings for consumers, though he said it was too early to draw definitive conclusions.
- Despite this promise, manufacturing a safe and effective biosimilar insulin is a very challenging, complex process. Using the fairly simple manufacturing process for small-molecule generics as a comparison, Dr. Hompesch warned the audience, “sorry, it ain’t that easy.” Insulin is a large, complex molecule, and any subtle difference in its structure could potentially have an impact on its clinical profile. Manufacturing insulin is a multi-step process involving living organisms (usually E. coli or yeast), so there are many opportunities for something to go awry. Additionally, all biological products are potentially immunogenic, so clinical studies are necessary to ensure that there is no excessive risk of a dangerous immune reaction.
- In addition to the manufacturing challenges, the regulatory guidelines for biosimilar insulins are “complex, inconsistent, and incomplete.” Dr. Hompesch used the EMA’s guidelines for biosimilar insulins as a reference but emphasized that there is no clear global standard and that requirements vary from country to country; for its part, the FDA has four guidance documents about biosimilars, though none specific to insulin. His main advice for companies hoping to bring a biosimilar insulin to market was to proceed in a stepwise fashion, as BI/Lilly has with their biosimilar glargine. Flipping to a picture of Presidents Obama and Putin scowling at each other, he stressed that it would be important to better align regulators from different countries in the future, but “I didn’t say it would be easy.”
Building a Better Basal – The Case for Concentrated Insulin
Wendy Lane, MD (Mountain Diabetes and Endocrine Center, Asheville, NC)
In her fast-paced review of concentrated insulins, Dr. Wendy Lane highlighted that these drugs offer several benefits for patients: a flatter action profile, longer duration of action, lower risk of hypoglycemia, smaller injection volume, less discomfort, and a higher dose per injection. Dr. Lane first considered Eli Lilly’s U500, currently the only FDA-approved concentrated insulin in the USA. In comparing it to Sanofi’s U300 insulin glargine (which is not bioequivalent to Sanofi’s U100 glargine [Lantus]), she noted that U300 glargine has a flatter profile, longer duration of action, and less nocturnal hypoglycemia. These advantages allow for the possibility of intensifying therapy with the drug. Dr. Lane also briefly mentioned two other insulins: 1) Novo Nordisk’s U200 insulin degludec (Tresiba), which is bioequivalent to U100 degludec and is more convenient for some patients; and 2) Znsulin (Thermalin), a U500 insulin with a 48 hour action profile.
- Dr. Lane noted that as people become more obese, they require larger insulin doses. She relayed that 2.1 billion people globally are currently overweight or obese, and that the US is a leader in this category. In recent diabetes studies, 35% of participants required ≥60U per day and 21% required ≥80U per day. Illustrating an extreme example, Dr. Lane showed a picture of a patient who requires 500-600 units of insulin per day. She described how current syringes and pens only deliver 100U and 60-80U in a single injection, respectively. Thus, concentrated insulin has the potential to significantly lower the injection burden for patients with high insulin requirements – i.e., obese type 2 patients with severe insulin resistance, as well as patients that are post-op or on glucocorticoid therapy, or patients with systemic infections, genetic defects in insulin action, or rare forms of immune mediated diabetes.
- Concentrated insulins offer several advantages: greater control and predictability, more comfort, increased patient adherence, and prolonged action. The ideal concentrated insulin would have a duration of ≥24 hours, single low volume dosing ideally using a pen, a flat and stable PK profile, minimal risk of hypoglycemia, low variability, and low intrinsic mitogenicity.
- U500 insulin (Eli Lilly) is the only approved concentrated insulin in the US. Its high concentration alters the pharmacokinetics to be similar to that of intermediate acting insulin, and the drug is only available in a vial and syringe. It has a peak action that tapers off in roughly 12 hours, and it works best if given two to three times a day. Patients report improved quality of life compared with U100, primarily because the smaller injection volume results in less pain, less leakage, and fewer injections. While U500 is often used as monotherapy, it can also be used with rapid acting insulin, or off label in a pump, where it performs well (though Dr. Lane noted that it has a long tail and it stacks). She also remarked that U500 has a tendency to confuse pharmacists.
- Sanofi is developing a new U300 insulin glargine, which has a flatter profile compared to U100 insulin glargine (Lantus) and thus is not equivalent in bioavailability. Importantly, the U300 insulin is associated with less nocturnal hypoglycemia. The longer duration and flatter profile gives the potential for intensifying treatment without additional hypoglycemia. The EDITION I trial (an open label 12 month study of U100 vs. U300 insulin glargine) demonstrated an equivalent A1c reduction in both groups, with a 10% higher dose for U300 patients and less nocturnal hypoglycemia associated with U300.
- Novo Nordisk’s U200 insulin degludec (Tresiba) has a smooth, flat action profile that lasts more than 24 hours. U200 insulin degludec is bioequivalent to U100 (i.e., same pharmacokinetics/pharmacodynamics), and provided the same A1c reduction in clinical trials.
- Znsulin (Thermalin) is a U500 insulin in development. It has an expected 48 hour profile due to the novel placement of Zn ions, which lead to “zinc stapled arrays” of hexamers.
Symposium: Hypoglycemia and Cardiovascular Disease – Lessons from Outcome Studies (Supported by an unrestricted educational grant from Merck)
Evidence from ORIGIN
Lars Ryden, MD, PhD (Karolinska Institute, Stockholm, Sweden)
Dr. Lars Ryden discussed the ORIGIN trial with a focus on the relationship of cardiovascular (CV) disease with hypoglycemia (symptoms confirmed by a glucose reading of ≤54 mg/dl) and severe hypoglycemia (symptoms requiring assistance, plus documented glucose ≤36 mg/dl and/or prompt recovery with oral carbohydrate, intravenous glucose, or glucagon). Hypoglycemia in ORIGIN was associated with increased risk of all-cause mortality and CV mortality (HR ~1.20, p<0.05 for both), but those relationships disappeared after controlling for known hypoglycemic risk factors (e.g., sulfonylurea treatment). By comparison, severe hypoglycemia was linked with the trial’s MACE composite endpoint, all-cause mortality, CV mortality, and arrhythmic death (HR=1.8-2.1, p<0.001 for each); even after adjusting for known risk factors of hypoglycemia, severe hypoglycemia was linked to roughly 50% higher CV risk. The glargine group had higher prevalence of hypoglycemia (42% vs. 14%) and severe hypoglycemia (6% vs. 2%) compared to the standard-care group. However, in the standard-care group, severe hypoglycemia was associated with roughly double the risk of major CV events, all-cause mortality, CV mortality, and arrhythmic death relative to the glargine group. Dr. Ryden therefore thinks that severe hypoglycemia caused by glargine was unlikely to be a direct “accelerator” of cardiovascular events. Rather, severe hypoglycemia may simply be a marker of risk, as has been suggested by studies of hospitalized patients (e.g., Kosiborod et al., JAMA 2009; Malmberg et al., Eur Heart J 2005).
- As a reminder, ORIGIN’s main finding was that intensive glargine therapy vs. standard care led to similar rates of major adverse CV events [MACE] in people with type 2 diabetes or prediabetes who were at high risk of CV events. See our full coverage of ORIGIN’s primary results at http://www.closeconcerns.com/knowledgebase/r/10be2669.
- The risk of both non-severe and severe hypoglycemia was predicted by use of glargine and/or sulfonylureas. Dr. Ryden noted that sulfonylureas may be especially harmful from a cardiovascular perspective, because they seem to impair myocardial adaptation to ischemia. Predictors of non-severe hypoglycemia included younger age, lower BMI, depression, and baseline diabetes. Predictors of severe hypoglycemia included older age, limited education, hypertension, renal disease, cognitive decline, and lower attained A1c.
Questions and Answers
Dr. John Yudkin (University College London, London, UK): We are gradually losing focus on what we should be trying to do. The rate of myocardial infarction (MI) in ORIGIN was 0.9 per 100 patient-years, meaning that 4.5% of patients will have an MI over 5 years. It was 0.93 per 100 patient-years in the intensive group. The hypoglycemia rate went up from 0.3 to 1.0 per 100 patient years. So in the standard-treatment group, the ratio of MI to severe hypos was 3:1; with intensive treatment the ratio was 1:1.1, meaning that more people were in the hospital with hypoglycemia than with MI. How do you provide informed consent to a patient when you suggest a treatment that may reduce their CV event rate, but will triple their risk of hospitalization for hypoglycemia?
A: The study that I presented today does not allow us to answer that question. It depends on the specific patient. The risk for CV event seems higher with some therapies than others, but it depends on the character of the patient. You can’t take data from this study to another population. These patients were relatively early in diabetes and with relatively high CV risk.
Dr. David Kendall (Lilly Diabetes): These data affirm that frequent mild-to-moderate hypoglycemia appears not only not-harmful, but may even mitigate the risk of CV events. In ACCORD, deaths tended to be in those patients with less frequent hypoglycemic events. A similar pattern has been seen in other studies. Do you have any thoughts on what may be happening to heart’s electrical system with frequent mild-to-moderate hypoglycemic events?
A: Repeated minor hypoglycemic episodes may make the heart less vulnerable to attack. More-frequent hypoglycemia may train patients’ bodies to respond to severe hypoglycemia. Still, repeated hypoglycemic episodes are not something that you should try to get, because you may unintentionally cause severe hypoglycemia.
Symposium: Can We Limit the Long-term Decline of Beta Cells in Type 2 Diabetes?
Ananda Basu, MD (Mayo Clinic, Rochester, MN)
In this review of insulin therapy’s effects on beta cell function in type 2 diabetes, Dr. Ananda Basu focused on clinical data, from modeling studies by Dr. Claudio Cobelli to the recent multi-center ORIGIN trial. He concluded with a call for more research on the durability of effect with early, tight intensive insulin therapy; non-invasive tests that will allow researchers to correlate beta cell function and mass; and the role of the beta cell in the legacy effect (the delayed, beneficial effect of glucose control on complications risk, which was observed in UKPDS).
- Dr. Basu explained that the insulin response to a glucose challenge is biphasic whether the glucose is given orally or intravenously, but with oral glucose the response is more reproducible and robust. Both phases of the postprandial insulin response become smaller as people move from prediabetes to type 2 diabetes (i.e., postprandial and basal insulin levels are closer together as diabetes progresses). However, intensive therapy with insulin early in the course of type 2 diabetes may preserve postprandial insulin response and glucose control, as has been suggested in a yearlong, uncontrolled proof-of-concept trial in 16 patients (Ryan et al., Diabetes Care 2004) and a larger, two-year randomized controlled comparison to metformin/sulfonylurea in China (Lancet 2008).
- Dr. Basu also mentioned the ORIGIN trial, in which insulin glargine therapy decreased the risk of progression from prediabetes to diabetes, assessed by oral glucose tolerance test (OGTT) at a median of 100 days after discontinuation of therapy (HR=0.80, p=0.05). Glargine-treated patients were also more likely to have mean A1c below 6.5% for the duration of the trial (up to seven years).
Symposium: Prevention of Hypoglycemia
Insulin Regimens to Reduce the Risk of Hypoglycemia
Niyaz Gosmanov, MD (Oklahoma City Veterans Affairs Medical Center, Oklahoma City, OK)
In this very broad overview, Dr. Niyaz Gosmanov touched on various factors affecting the choice of an insulin regimen. In particular, he stressed the need for insulin therapy to minimize the risk of hypoglycemia, emphasizing that A1c is not the only target. According to Dr. Gosmanov, an insulin regimen is a treatment package that replicates physiological insulin effects as closely as possible. A regimen can range from one shot per day to full replacement of all types of insulin, depending on the endogenous insulin production of the patient (which should not be overlooked as often as it is, in Dr. Gosmanov’s opinion). Insulin can be basal (for elevated fasting glucose levels) or bolus (for elevated post-prandial levels); it can also be human or analog. Critically, issues such as cost and flexibility are just as important for the patient. Unfortunately, the two may have an inverse relationship, as seen with the inflexible but cheaper premixed insulins. Another problem highlighted by Dr. Gosmanov is the complexity posed by insulin regimens, as it can be difficult for patients to keep track of multiple shots per day. He suggested the website www.accurateinsulin.org, an interactive tool that helps determine when and how to adjust insulin doses. Finally, Dr. Gosmanov remarked that the best therapy is the one that the patient is likely to take confidently and his provider prescribes confidently and comfortably.
Questions and Answers
Q: In the 4T study, basically the basal-bolus regimen was found to cause less hypoglycemia than twice daily mixtures. What is your opinion on mixtures versus basal-bolus?
A: Today we put too much on mixed insulins. We’re desperate and using them as a last resort. The evening dose has to be given too early. Patients lose flexibility.
Q: We all agree that standard diabetes care is quite effective and safe. I also feel that it’s quite illogical. For example, using a single number from a finger stick to correct high blood sugar seems illogical and dangerous. I think we should try to prevent large increases in the first place. The pancreas monitors change and prevents an abnormal glucose rise. The reason there’s so much hypoglycemia is that the standard approach in how we teach people to use tools is wrong.
A: I completely agree. A finger stick value comes from just one second out of the day. Type 2 patients have more reliability and predictability compared to type 1, but in an ideal world I would use CGM on everyone. But unfortunately I can’t do that.
Q: I was hoping your talk would be a little more specific on insulin regimens to either avoid or pursue. I want to throw out some errors. First, I see a lot of patients who come from primary care who are over-basalized. They need prandial insulin but they get basal. I think that places them at higher risk for hypoglycemia especially if they sleep in. Another is taking premixed insulin at bedtime, which has to be an error. Third, NPH can be a good basal at bedtime, but at dinner it increases the risk of low blood sugar. Fourth, primary care doctors increase Lantus based solely on A1c levels. Any objections that these are erroneous strategies?
A: Nope. It’s all about education to the patient as well as the provider. Those points were all relevant. I initially had slides related to these, but my colleagues suggested they were too simple. Maybe I’ll include them next time.
Q: I have two questions. First, even though we know that premixed insulin has limits, it is the most prescribed insulin across the world. How can you best use premixed insulin? From the point of diet, how can you best split meals to best suit premixed insulin? Secondly, 10 to 20 years down the lane, type 2 patients behave just like type 1 because all insulin runs out. I think the insulin pump definitely has a role to play. So what type of type 2 patients can benefit from a pump?
A: For the first question, I would refer to the website in the presentation. In the developing world, people try to get away with cheaper versions of insulin. Any time you save on costs, you lose something else, usually flexibility. So the patient needs to eat at least two meals. I would recommend injection post-morning and in the late afternoon. As for how to titrate the insulin, I think the website would give more information. For the second question, we have criteria we use for pump selection in type 1 patients. We can apply some of those criteria to type 2 diabetes before recommending them to proceed with the pump. Certain situations that need good sugar control, like pregnancy, may prompt a pump. There’s evidence that type 2 patients can benefit from it.
Q: Any pearls on how to work with patients with an overly large fear of hypoglycemia, who eat their way to high blood sugars before bedtime?
A: Last year ADA had a nice webcast about this topic. That’s exactly where psychological intervention takes precedence. We’re looking for more psychologists at our diabetes center than nurses or other workers. One episode of severe hypoglycemia may push patients back several steps because they don’t want to repeat that.
Q: A comment on over-basalization: one reason is because tools being put out by drug companies who make the analogs never mention over-basalization. They never talk about it. So physicians just tell the patient, “We’re moving to 7%” and so the patients keep increasing the dosage. Some people never know there’s a max, and next thing they move it up to 90 units before seeing you again.
A: I see some of your concerns, but there are also some counterpoints.
Symposium: Initial Treatment of Type 2 Diabetes – New and Not-So-New Ideas
Short-Term Intensive Insulin Therapy Early in the Course of Type 2 Diabetes
Ravi Retnakaran, MD (Leadership Sinai Centre for Diabetes, Toronto, Ontario)
Dr. Ravi Retnakaran delivered an overview of previous data demonstrating the effectiveness of short-term intensive insulin therapy (IIT) and made a case for treating patients with type 2 diabetes early on with IIT followed by maintenance therapy. In prior studies, IIT was shown to have positive effects on β-cell function even after treatment ended; however, this effect was dependent on administering the treatment soon after the diagnosis of type 2 diabetes. Furthermore, this effect slowly declined over time. Thus, Dr. Retnakaran suggested following the ITT treatment with additional long-term therapy to maintain the gains in β-cell function. Dr. Retnakaran described three clinical trials testing this maintenance drug concept: BEST, LIBRA, and RESET IT. Of these, BEST had largely negative results, RESET IT is ongoing, and LIBRA results were announced the Sunday of ADA.
- Short-term intensive insulin therapy (IIT) has been shown to improve β-cell function even after treatment cessation. Beta-cell function declines in response to multiple factors in type 2 diabetes, and insulin therapy has been shown to counteract several of these. For example, insulin can reduce glucotoxicity (due to chronic hyperglycemia), lipotoxicity (due to chronically elevated free fatty acids), inflammation, and resistance to incretins. Dr. Retnakaran cited a 2008 Lancet study by Weng et al. which showed that A1c decreased from a baseline of ~9.5% to ~8% following two to five weeks of treatment but further decreased to ~6.5% one year after patients switched from medical treatment to diet and exercise. Remission was more durable in groups initially treated with insulin compared to those initially on oral medications such as metformin.
- The earlier patients start on IIT after diagnosis, the more their beta-cells recover. Dr. Retnakaran described a model by which β-cell dysfunction has both reversible and irreversible components, with the latter increasing over time. Thus, amelioration of dysfunction should be more effective early in the course of the disease. He uses this principle as well as supporting evidence to argue for the use of IIT in early type 2 patients.
- The beta cell improvements seen after IIT decline over time; however, strategies to maintain the improvement are under investigation in several clinical trials. One strategy would be to add a long-term oral medication after IIT cessation. The BEST trial evaluated the DPP-4 inhibitor sitagliptin (Merck’s Januvia) for this purpose but found no difference vs. placebo. The LIBRA trial, whose results were released during ADA 2014, studied the GLP-1 agonist liraglutide (Novo Nordisk’s Victoza). Finally, the ongoing RESET IT trial investigates the effect of administering IIT every 3 months –i.e., using insulin as both the initial and maintenance therapy.
Questions and Answers
Q: Do you think combinations of drugs as maintenance therapy may yield even more long-term benefits?
A: Well, the truth is there’s still a ways to go in testing combinations. There may be combinations that provide an even greater benefit.
Q: Can you define how early in diabetes is “early,” and the duration of treatment in IIT?
A: The duration before treatment is initiated is very important; with a longer duration, you see a weaker response. Typically, patients within the first five years of diagnosis are in good shape for seeing a response. The duration of insulin treatment is an interesting question: in LIBRA we used a 4-week duration, and it varies for other studies. However, we see a considerable effect even at 2 weeks, so a shorter duration is possible.
Symposium: New Frontiers in Inpatient Diabetes Management
Are There Alternatives To Insulin?
Roma Gianchandani, MD (University of Michigan, Ann Arbor, MI)
Dr. Roma Gianchandani discussed potential alternatives to insulin therapy as the primary method of treating hyperglycemia in hospitals. Dr. Gianchandani cited conflicting data on insulin’s efficacy for achieving proper glycemic control and reducing hospital complications, noting that the drug has been repeatedly linked with an increased risk of hypoglycemia and is the most error-prone medication in hospitals. Dr. Gianchandani proposed that incretins (GLP-1 agonists and DPP-4 inhibitors) have several desirable characteristics as replacements for insulin: they operate via a mechanism that increases insulin secretion, decreases glucagon secretion, and reduces the risk of hypoglycemia, while remaining relatively easy to use and decrease the need for glucose monitoring. Although gastrointestinal (GI) side effects are a serious concern, and limited data has assessed safety outcomes, she maintained that these drug classes are the most viable alternatives to insulin therapy.
- In a study by Umpierrez et al. (2002), 38% of adults admitted to an Atlanta hospital had hyperglycemia (26% with known history of diabetes; 12% without). Based on these data, Dr. Gianchandani emphasized that achieving better glycemic control remains a challenge in hospital settings.
- The NICE-SUGAR trial demonstrated that tight glycemic control (81-108 mg/dl), achieved via insulin infusion, in critically ill patients is actually associated with higher mortality than patients maintained under conventional glycemic control (< 180 mg/dl). Dr. Gianchandani emphasized that while tight glycemic control is typically thought to improve patient outcomes, this study is one important example that introduces uncertainty into the use of insulin therapy in hospital settings.
- “Insulin is the most error-prone medication used in hospitals.” In addition to being labor intensive and requiring hourly checks and complex calculations, Dr. Gianchandani emphasized the severe risk of hypoglycemia as the primary reason for identifying an alternative to insulin therapy.
- Dr. Gianchandani argued that incretins (GLP-1 agonists and DPP-4 inhibitors) are the most viable options to replace insulin therapy, due to a host of desirable characteristics, including: an action mechanism that increases insulin and decreases glucagon, low risk of hypoglycemia, ease of use, and decreased glycemic variability (and, therefore, decreased need for glucose monitoring).
- In a review of multiple studies, incretins were successful in controlling blood glucose levels, regardless of the method of delivery (continuous infusion, with controlled nutrition therapy, subcutaneous injection, or in conjunction with steroids, which raise blood sugar). Notably, hypoglycemia was not negligible in the majority of studies, but use of incretins did tend to reduce the prevalence of hypoglycemia relative to insulin treatment. Dr. Gianchandani emphasized the high prevalence of GI problems and expressed some concern that concomitant insulin use was still needed in particular cases in which incretins alone failed to reduce blood glucose levels.
- The reduced number of injections, nursing time, and risk of hypoglycemia suggest that incretins could be an attractive and cost-effective alternative to insulin therapy. However, as opposed to insulin therapy for which clinicians have extensive experience and protocols are readily available, Dr. Gianchandani recognized that more data on safety outcomes in hospital use is necessary before use of incretins becomes widely adopted.
Questions and Answers
Q: What kind of patient is a good candidate for incretin therapy?
A: I would suggest a patient who is a surgical candidate.
Q: Isn’t there evidence that DPP-4 inhibitors and GLP-1 agonists may be more effective in patients who have had diabetes less than 10 years or are prediabetic?
A: Many of the studies I reviewed involved patients who were prediabetic. But yes, these drugs do need insulin for effect, so in patients who have progressed in the disease, you may need insulin plus an incretin agent for results.
Q: Has it been shown that better glycemic control improves patient outcomes?
A: There’s not a lot of data. Anecdotally, people with high blood sugar during surgeries tend to return more often with infections.
Q: What is the cost of using a DPP-4 inhibitor versus insulin in a hospital?
A: We’re looking into that. When you give insulin in a hospital, the actual cost reflects more than just insulin. It includes the nurses, the time required to check blood sugar multiple times. That said, DPP-4 inhibitors are very expensive right now, but I think that if you can increase their use, it will be more cost-effective.
Q: What do you think about subcutaneous GLP-1 agonists?
A: There was a suggestion for using these drugs before coming to the hospital. It’s a very tantalizing therapy. I do know people who are thinking of that in future trials.
Q: Why are you so supportive of incretin therapy given the concerning GI side effects?
A: The limitations of insulin drives us to GLP-1 agonists. That driving force includes nursing time, hypoglycemia, etc. Though if you use a computerized protocol and different targets and are stringent with your application of insulin, then hypoglycemia is not that big a deal.
Computerized Insulin Algorithms
Rattan Juneja, MD (Indiana University School of Medicine, Indianapolis, IN)
Dr. Rattan Juneja presented the results of a pilot study of SUGAR, a computerized insulin-dosing algorithm for managing in-hospital insulin therapy. The algorithm recommends an insulin infusion rate based on the patient’s blood glucose level and a target range set by the healthcare provider with periodic reminders to enter an updated blood glucose value. Results from this pilot study demonstrated that patients achieved blood glucose levels in the target range (80-110 mg/dl or 100-150 mg/dl in different hospitals) in an average of four to six hours, with reduced variability over the course of treatment. The overall rate of hypoglycemia was 3.6%, with only 0.1% of patients reaching a blood glucose level of <40 mg/dl. Overall, this study suggests that computerized insulin algorithms could be useful tools for managing patients in the hospital setting.
- Current procedures for managing hospitalized patients on insulin are limited in their efficacy. Dr. Juneja presented statistics from 2009 showing that of the 700-800 patients treated at Indiana University Health every day, 300-400 of them received insulin, a percentage that he believes is fairly typical for many hospitals. Not all of these patients are diagnosed with diabetes, and many of the professionals caring for them are not well trained in insulin therapy, leading to ineffective treatment and high rates of hypoglycemia. In Dr. Juneja’s opinion, some of the factors contributing to this problem are lack of coordination between nurses giving insulin and staff in charge of food and transportation, inadequate glucose monitoring, indecipherable insulin orders, and a lack of understanding of basal/bolus therapy.
- The SUGAR system uses a computerized insulin dosing algorithm and reminder alarms to help hospital personnel provide more effective insulin therapy. This system, short for “Systematic Utilization of Glucose Assessment and Response,” was tested in eight hospitals in 2006-2007. The central component is an intravenous insulin dosing tool that recommends an infusion rate based on the patient’s current blood glucose and the high and low targets entered by a doctor or nurse. An alarm goes off every hour reminding the nurse to enter a new blood glucose value, and the algorithm adjusts the recommended infusion rate based on the new value and recent trends.
- An additional tool not yet approved by the FDA applies the same principles to basal/bolus therapy. This system provides a standardized form for insulin orders, sets default blood glucose targets and values for carbohydrate ratios and insulin sensitivity based on the patient’s weight, and alerts nurses to enter a new blood glucose value at various times (e.g., before each meal, at bedtime, and at 3 am). Providers can alter the default settings for individual patients, but the goal is to provide some sort of structure for personnel who are unfamiliar with basal/bolus therapy.
- Results from the Indiana pilot study suggest that computerized insulin algorithms can help patients reach their blood glucose targets with a significantly lower risk of hypoglycemia. The average blood glucose level out of the 334,000 measurements taken in the study was 106 mg/dl, with an average time to target of six hours if the goal was 80-110 mg/dl and four hours if the goal was 100-150 mg/dl. The variability of a typical patient’s blood glucose was also drastically lower with the computerized system. In all, 10.1% of patients had a blood glucose measurement of <80 mg/dl at some point, but only 0.1% of them ever dropped below 40 mg/dl. The overall rate of hypoglycemia was 3.6%, much lower than the rates of up to 12.1% seen in other hospital-based studies.
Questions and Answers
Q: Can you comment on the cost of the software and the time on the hospital system side required for education, IT, etc.? And are there safety parameters in place, like a limit on the maximum amount of insulin given?
A: We did not do a cost analysis. Our hospital system decided this was important, so the cost was absorbed into the operating budget. There are safety parameters – we have stops if more than 50 units of insulin are delivered.
Corporate Symposium: Shared Decision-Making in Insulin-Initiation – Live Clinician-Patient Conversation on Overcoming T2DM Treatment Barriers (Sponsored by Sanofi)
The Rationale for Early Insulin Initiation
Om Ganda, MD (Joslin Diabetes Center, Boston, MA)
Since beta cell function is known to begin to decline years before diabetes diagnosis, Dr. Ganda urged symposium attendees to consider insulin sooner rather than later when treating patients with diabetes. Citing results from the ADOPT trial, Dr. Ganda emphasized that monotherapy (sulfonylurea, metformin, or thiazolidinedione) frequently fails to bring patients to blood glucose goals. However, in a study of 133 newly diagnosed patients with A1c ≥9.7%, 51% were able to reduce fasting blood glucose levels to <110 mg/dl on CSII after one year. Forty-five percent of 118 patients on MDI reached remission, whereas only 27% of patients on oral hypoglycemic agents reached remission. Further, patients receiving CSII and MDI had significantly higher acute insulin response through one year following therapy initiation than patients receiving only oral agents. Though the ADA/EASD guidelines emphasize the need for targeted and individualized care, Dr. Ganda finds that guidance beyond initiating metformin is limited and somewhat vague. He hopes that the upcoming NIDDK-funded GRADE study will help providers make better decisions when treating patients early on in the duration of the disease. Not only will the study compare the effects of insulin glargine against liraglutide, sitagliptin, and glimepiride head-to-head in patients who have been diagnosed within the last five years, it will also assess the impact of adding basal insulin and intensifying with rapid-acting insulin on patients who are unable to meet A1c goals.
Dr. George E. Dailey, III (Scripps Whittier Diabetes Institute, La Jolla, CA): I certainly use insulin earlier than I used to. I think the ORIGIN trial was particularly useful in demonstrating the safety of basal insulin earlier in the course of the disease.
Dr. Anne Peters (Keck School of Medicine of USC, Beverly Hills, CA): It turns out that, obviously, we have more choices now than ever. And one of the good points is that people are doing better and A1cs are going down. I think it’s because we’re working with our patients better. But we also have all different ways to treat our patients. I think the use of basal insulins and insulin pens has made it easier to initiate insulin. Back then, it was more of a barrier.
Dr. Om P. Ganda (Joslin Diabetes Center, Boston, MA): Well, with all the new drugs that we have, can we afford to postpone the use of insulin? No, that’s not the case. From the numbers we’ve seen, we can still be making some progress. I have a strange feeling that we still have a lot of people who could be using basal insulin who are not doing that yet. This could reflect the barriers we’re talking about.
Dr. Peters: Maybe we should cheer over the numbers from JAMA and the CDC. Over the past 20 years, we have done an amazing job with our patients. We’ve increased he number of patients on statins. We don’t have them all, but we’re doing better. There has been a corresponding 56% reduction over the past 20 years in cardiovascular events. We’re reducing hospital rates from hyperglycemia, but hypoglycemia has been going up. We’re doing better as a group in treating our patients.
Dr. Ganda: We’re doing better with lipid control and blood pressure management. And that’s a nice reduction in cardiovascular outcomes when it comes to microvascular disease. But, I’m afraid we might be seeing more renal disease. But that could be one factor that could be further addressed by better glycemic control.
Dr. Peters: We’re also seeing such an epidemic of diabetes that doesn’t seem to be abating. I like having a combination of options – it makes diabetes care more successful. I feel more and more that I can get patients to target. When I trained a long time ago, all I had was NPH, regular, and sulfonylureas. We’ve come a long way since then.
Barriers to Insulin Therapy: The Clinician, the Patient, and the Insulin Itself
Anne Peters, MD (Keck School of Medicine of USC, Beverly Hills, CA)
Dr. Peters spoke on behalf of Dr. Alan Garber, who was sick and unable to make the symposium. She described various sources of resistance to insulin therapy in type 2 diabetes patients. In a survey of 1,267 patients, it was found that misconceptions about insulin were all too common, especially in those who were unwilling to take insulin. Results from the Translating Research Into Action for Diabetes (TRIAD) study found that insulin-naïve type 2 diabetes patients who did not fulfill their insulin prescriptions often “planned to improve health behaviors instead,” or feared the effects of insulin. Dr. Peters stressed that healthcare providers could be to blame. If a doctor is nervous or hesitant to use insulin, then patients can sense that and also start to feel nervous or hesitant. She encouraged physicians to have confidence in the patient and focus on the patient’s goals over everything else. In shared decision-making, patients need to be informed and have the freedom to choose between options. Dr. Peters presented a three-step model for clinical practice in which the physician and patient talk about choices, options, and decisions. Following an audience response session, Dr. Peters concluded with a discussion with two of her patients. Mike, tall and soft-spoken, expressed an adamant fear of needles. Even though Dr. Peters and Mike had talked several times about insulin injections, he refused and so Dr. Peters respected his request. The second patient, Elaine, described her traumatizing diagnosis by an insensitive, almost brutish doctor, and contrasted that experience with Dr. Peters’s collaborative methods. “When you’re with her [Dr. Peters], no one else seems to matter,” Elaine said. “It’s of utmost importance for a patient to feel that way.”
Q: When initiating insulin therapy for type 2 diabetes, I am most likely to:
1. Emphasize the treatment benefits to motivate patients (59%)
2. Ask patients about their fears of insulin (28%)
3. Wait until patients voice negative associations (2%)
4. Reassure them that they’ll soon learn to manage insulin therapy well (12%)
Dr. Ganda: I agree with answer one that emphasizing the benefits is most important. Unless patients are given some background, they’re not going to understand why insulin might be needed.
Dr. Dailey: I think answer two is where we need to focus additional effort. It’s difficult for the patient to express concerns. It is very important to uncover and address them earlier.
Dr. Peters: I sometimes forget how scary it can be to be a patient. For patients who are getting diagnosed, it’s overwhelming to hear the words: “You have diabetes.” Understanding fear and concerns about that is important. Patients will ask about side effects of drugs, but I always talk about it in context with the benefits. One of my patients is Don Rickles, the comedian. He said to me, “Tell me, what’s the worst thing that can happen with me taking statin?” So I said, “It could kill you.” He now calls me Dr. Death, so perhaps I could have chosen a different side effect.
Dr. Ganda: One quick comment. I think answer four is a perfectly reasonable thing, but that’s easier said than done. We have the luxury of nurse educators and nutritionists next to me, but not everyone has that.
Dr. Dailey: A lot of people relate to the worst case scenario. For example, when patients read about the product.
Dr. Peters: The internet is terrible; often I tell people not to look on the internet. I think it’s up to us to help people interpret the package insert and tell them what side effects are realistic.
New Approaches to Insulin Therapy
George Dailey, III, MD (Scripps Whittier Diabetes Institute, La Jolla, CA)
Dr. Dailey provided a high-level overview of strategies for maintaining glycemic control in patients when basal insulin therapy is not sufficient. Dr. Dailey opened his presentation by encouraging a “stepwise” approach to therapy that gradually adds medications in an effort not to overwhelm a patient. He emphasized that ideal glycemic control requires mimicking physiological insulin secretion and that the incorporation of rapid-acting analogs with quick action profiles (two to three hours) into treatment regimens can be a particularly effective option upon failure of basal insulin therapy alone. That said, Dr. Dailey advocated more strongly for GLP-1 agonists, DPP-4 inhibitors, and SGLT-2 inhibitors that he believes offer great potential as insulin alternatives or adjuncts. Citing one study, Dr. Dailey described that treatment with insulin glargine plus twice-daily exenatide resulted in greater A1c reduction and less weight gain relative to glargine treatment alone in type 2 diabetes patients. However, in looking to the future, Dr. Dailey drew attention to the high-concentration and longer-acting analogs under investigation, such as Lilly’s U500 Humulin and Novo Nordisk’s insulin degludec, respectively, arguing that there will be a place for each of these in the market based on pricing and patient preferences. Next, Christine, a patient of Dr. Dailey’s, briefly addressed the issue of patient-provider dialogue and trust. She recalled the trauma she experienced when first diagnosed with diabetes due to providers who did not take the time to fully explain the intricacies of the disease. However, in Dr. Dailey, Christine emphasized that she had found a “partner” who she trusted wholeheartedly and who convinced her to take insulin – which she described as a “nuisance” – by explaining that the benefits outweighed her concerns about weight gain. In this sense, she explained that she was incredibly grateful for a clinician who involved her in the decision-making process and who “really works with me even when I don’t want to do something.”
Q: In a patient on basal insulin with normal fasting glucose whose A1c is beginning to rise, I would:
1. Use a “basal bolus” strategy (44%)
2. Add a GLP-1 receptor agonist (56%)
Dr. Ganda: I think you know that basal bolus was certainly a very popular choice until GLP-1 receptor agonists became available. Many patients are concerned with weight gain. Even with basal insulin alone, there can be a case made for adding a GLP-1 agonist, especially for overweight and obese patients. It’s a good choice, provided you can get coverage. We have many agents available, like once-weekly exenatide, so there are more options for the patient.
Dr. Dailey: What about Elaine?
Dr. Peters: First of all, Elaine clearly was at a point where she started to need something else other than basal insulin. I didn’t want her to get hypoglycemic, so adding a GLP-1 agonist was a wonderful option. I was worried about her because she tends to get GI effects from meds, so we increased the dose slowly to get to a point where she could tolerate it. Her A1c is now amazing and I’m able to further reduce insulin. It turned out to be a win-win situation for me. No mealtime insulin, and she can lose weight.
Dr. Dailey: What’s also nice about GLP-1 agonists is that you can see results within a few weeks. But we know 20% of people simply don’t respond to the drug.
Dr. Ganda: Not all patients are as lucky as Elaine. Some patients have problems with lifestyle, like staying on top of diet and exercise, and so not all patients will be candidates for this combo. We can still get good control in other ways though.
Dr. Peters: I find adding one shot of insulin after the biggest meal is good. It’s not much extra work; it’s not too daunting. It’s also a good time to help patients deal with snacking and eating in the evening, so adding the shot of is another chance to work on lifestyle since we want to make the insulin work better.
Dr. Dailey: Do you have any other tricks to cover “grazing” after dinner?
Dr. Peters: I write prescriptions for dogs frequently. People can walk the dog before sleeping. I also do home makeovers. We found if you take the TV out of the kitchen and move it to the living room, you eat less. Obviously, having healthier foods in the house is good, like having vegetables to snack on instead of unhealthy things. I like to find one healthy snack people like and gradually move toward more healthy snacks.
Final Panel Discussion
Q: Can you talk about new directions in insulin therapy?
Dr. Peters: We want to give insulin in a physiological way. That’s why these longer-acting insulins are useful to me, especially with my type 1 patients. And I think rapid-acting insulins will be of more use to the artificial pancreas. With my type 2 patients, maybe I need insulins that are more stable, and maybe I don’t want to give them as often. I think, in terms of development, it really depends on the patient population, although I think simplicity and avoiding hypoglycemia are universal key points.
Dr. Ganda: I think insulin analogs have very been helpful since we’ve had them, as they’ve significantly reduced hypoglycemia – especially nocturnal hypoglycemia. The other point I want to make is that the obesity epidemic is here to stay. And when we consider giving higher doses of insulin, we have to weigh that risk against the potential increase in weight. This is a problem. So with higher concentration insulins that are available, we can use less insulin, and we might still get benefits. So in terms of future development, working on higher concentration insulins is a great idea since the U500 is the only more concentrated insulin we have right now, and it does not have a predictable course of action.
Dr. Dailey: The 50% reduction in hypoglycemia we saw between NPH and glargine in the Treat to Target trial will not be seen again. From now onward, we’re going to be talking about lesser reductions. Anything that lowers this barrier will make life easier on us.
Dr. Peters: I think people are beginning to understand that the key is having the time to work with patients. I think it’s about partnership with patients and our teams. We all need to work together. In fact, physicians don’t need to be the most closely allied individual to a patient if other individuals can help and the patient feels more comfortable with him or her. The key is that there is a lot of good we can do collectively.
Q: How do I go about reducing the amount of insulin I’m taking?
Dr. Peters: It takes patients who are capable of self-titration, because most of this has to be done by patients. Patients titrate down gradually, and then we communicate if there’s an issue.
Q: Since ADA guidelines no longer specifically recommend A1c less than 7.0%, and hypoglycemia is a problem with older patients in particular, and since the population is growing older, isn’t having 75% of patients under 8% good?
Dr. Ganda: We need to individualize treatment goals. When I said 7%, that’s an overall acceptable level. It’s not just age, but comorbidities and things like renal failure, kidney disease. About the older population, the ADA actually said last year that because of comorbidities, we really need to personalize the goal for older patients. In those situations, tight glucose control is not as important.
Dr. Dailey: You’re right. Benefits in reducing microvascular complications require several years to see. If the patient has less than 10 years left, it’d be pushing it to get them under 7%.
Corporate Symposium: Diabetes: Ensayos Clinicos, Tratamientos y Retos (Supported by an unrestricted educational grant from Novo Nordisk and presented only in Spanish)
While we do not have truly fluent Spanish speakers on our core team, we tried to glean as much as we could from this session and bring you some high level takeaways.
“Diabesidad” en el Adolescente (Diabesity in Adolescents)
Ximena Lopez, MD (University of Texas Southwestern Medical Center and Children’s Medical Center, Dallas, TX)
Dr. Ximena Lopez started off her presentation by emphasizing the huge lack of knowledge about the treatment of type 2 diabetes in children. She then presented a study on glycemic control in adolescent type 2 diabetes patients comparing the efficacy of metformin monotherapy, metformin plus lifestyle intervention, and metformin plus thiazolidinediones (TZDs). In addition, she reviewed guidelines for when to start treatment with insulin in children with type 2 diabetes as well as FDA approval for type 2 diabetes drugs for adolescents.
- Dr. Lopez described the results of the TODAY study, which revealed the prevalence of type 2 diabetes in adolescents from Hispanic and African American backgrounds and with lower socioeconomic status. The study examined glycemic control in adolescents ranging from ten to seventeen years old (n=699), who had been diagnosed with diabetes for less than two years. They were treated with metformin, metformin plus rosiglitazone, or metformin plus intense lifestyle intervention. All were overweight, and 65% were female. ~40% were Hispanics, and ~30% were African American. 42% had a family income <$25,000, and only 17% had parents with a university degree. All patients started with A1c <8%.
- Different genders and minorities responded differently to treatment. Overall, there was no difference between metformin monotherapy and lifestyle intervention, and while the TZD group performed better compared to metformin monotherapy (p=0.006). However, boys responded best to metformin and lifestyle intervention (p=0.06). Interestingly, African Americans responded the worst to metformin monotherapy (p=0.003 compared to metformin and TZDs, p=0.008 compared to metformin and lifestyle intervention). On the other hand, Hispanics responded the worst to metformin and lifestyle intervention, although there were no statistically significant differences between the three treatment groups for this subpopulation. A baseline A1c of <6% or >6% was a predictor of treatment failure; subjects with A1c >6% were more likely to fail treatment.
- Dr. Lopez recommended starting children with type 2 diabetes on insulin therapy if they had DKA, an unclear distinction between type 1 diabetes and type 2 diabetes, or A1c >9%.
- Metformin is the only oral drug approved by the FDA for type 2 diabetes patients <18 years. Dr. Lopez suggested intensifying treatment at A1c >7%, and stressed the importance of monitoring glucose levels.
Questions and Answers
Q: It’s interesting that metformin works better in Caucasian children than in African American children, but this is flipped for adults. Why is that?
A: We’re unsure of that – it could be something related to adolescence.
Q: Why not use DPP-4 inhibitors for adolescents?
A: Currently, there are ongoing studies for children, but there are no results yet. It is possible that it could affect the mineral density of children’s bones, or spermatogenesis.
Q: What happens if metformin fails, but the patient hasn’t yet reached the thresholds that you recommended for insulin use?
A: Start the patient on insulin therapy.
Israel Hartman, MD (University of Texas Southwestern Medical Center, Dallas, TX)
Dr. Israel Hartman reviewed the safety of insulin analog therapy during pregnancy and for children and adolescents. He highlighted necessity for more studies showing how insulin analogs work in pregnant women and children, since drugs from these classes are the only diabetes treatments approved for pregnancy and for young children.
- During pregnancy, glycemic control is even more important. Hypoglycemia is incredibly dangerous for both the mother and child, and hyperglycemia can cause abnormalities when A1c rises to over 1% above goal. It’s also important to take insulins that aren’t immunogenic (i.e., don’t cause the body to antibodies against it).
- Dr. Hartman also emphasized the extreme variability in the action profile of insulins and insulin analogs in children. Insulin glargine (Sanofi’s Lantus) is approved for children over six years old, and insulin detemir (Novo Nordisk’s Levemir) is approved for children over two years old. In insulins, there is a lot of variability for children, especially in kids from ages six to eleven years.
Eduardo Montoya, MD, PhD (University of Barcelona, Barcelona, Spain)
Dr. Eduardo Montoya described the characteristics of an ideal basal insulin: efficacy over a long action period, a flat profile, low variability, and flexibility throughout the day. He reviewed some of the newer insulin analogs that have improved action profiles, such as peglispro. He showed the results of one crossover study in which patients on peglispro had lower incidences of hypoglycemia (p=0.037) and better weight effects (p=0.001) than those on glargine. Moving on to insulin degludec, he highlighted graphs displaying degludec’s flatter action profile.
Jaime Davidson, MD (University of Texas Southwestern Medical Center, Dallas, TX)
After covering some statistics on the current state of insulin administration, Dr. Jaime Davidson reviewed some best practices for avoiding lipohypertrophy (an accumulation of fat under the skin that can occur from injecting insulin at the same site too frequently). In most other countries, the majority of patients use pens, but in the US less than 50% of patients use them. In addition, 48% of patients have lipohypertrophy. Dr. Davidson asserted that there is no difference in skin thickness between people of different ages or races, and that it was imperative to educate patients on how to inject themselves correctly, due to the huge variation in effects when injecting into muscle versus the subcutaneous layer. He recommended needles of 4 mm in length for all patients except for those in the third trimester of pregnancy, when skin thickness changes.
In the panel discussion, Dr. Ximena Lopez, Dr. Israel Hartman, and Dr. Eduardo Montoya answered questions from the audience regarding cardiovascular risk in insulin analogs, recommendations for diabetes treatments during pregnancy, and bariatric surgery in adolescents. From what we gleaned, Dr. Lopez answered the question about bariatric surgery, responding that it has not been proven to reduce diabetes in adolescents, although some bypasses are relatively safe for adolescents.
--By Adam Brown, Eric Chang, Hannah Deming, Jessica Dong, Varun Iyengar, Nina Ran, Emily Regier, Katherine Sanders, Joseph Shivers, Wilbur Song, Tony Thaweethai, Manu Venkat, Michelle Xie, Rebecca Xu, John and Kelly Close