June 24-28, 2011; San Diego, CA Day #5 Full Notes DRAFT

Executive Highlights

Good night from San Diego and the final day of ADA 2011! We’re sorry to say goodbye to the Gaslamp Quarter and our 17,600 fellow attendees, but this week has us excited for the coming months and years – not least because of today’s strong session of late-breaking oral presentations. A 12-week phase 2b study of TAK-875 showed strong efficacy (A1c reduction of 1.0% at high doses), a fast action profile, and good tolerability with minimal hypoglycemia risk; we look forward to learning more about Takeda’s plans to bring its novel GPR40 agonist into phase 3. Other pharmaceutical highlights were Intarcia’s implantable exenatide delivery system ITCA 650, which has 48-week efficacy similar to Byetta but with no needles and guaranteed adherence (we’ll wait to hear about IP), and Roche (Genentech)’s Lucentis for diabetic macular edema, which was shown to improve a host of secondary efficacy measures in its two 24-month phase 3 studies. Past ADA President Dr. Richard Bergenstal (International Diabetes Center at Park Nicollet, Minneapolis, MN) presented the first-ever analysis of SMBG data from ACCORD, showing that high mortality in the intensive treatment group was associated with “divergence from glycemic target.” As we understand it, the big takeaway from ACCORD hasn’t changed: when glucose isn’t coming down in a stable way, it might be time to relax the regimen. It’s also true that blood glucose mattes, from these striking results. Also of major note, a cost-effectiveness study of the Diabetes Prevention Program suggested that lifestyle intervention was cost-effective and metformin was cost-saving from both a healthcare system and societal perspective; we see this evidence as a win-win patients with prediabetes as well as payors. Unfortunately, prospects for drug therapy in type 1 diabetes remain cloudy. Yesterday’s session on type 1 cure-based therapies included several interesting subgroup analyses and theories about why recent late-stage trials (Tolerx, Macrogenics, Diamyd) have been disappointing relative to earlier studies. Much remains to be learned about the underlying biology of type 1 diabetes and these pharmaceutical targets, and there is no doubt that a true cure is far away indeed. We are grateful to artificial pancreas researchers as it’s clear as day to us this year that the work has multiplied and that they are more ambitious than ever, working under JDRF and the Helmsley Foundation funding (for which we as patients are very grateful) in particular. As always at ADA, we come away refreshed for the long journey toward better therapies for both type 1 and type 2 diabetes – and grateful for the many, many dedicated physicians, researchers, and industry members who are making strides every day.

Oral Presentations

Late Breaking Clinical Studies

A RANDOMIZED, DOUBLE-BLIND, PLACEBO- AND ACTIVE-CONTROLLED, DOSE- RANGING STUDY TO DETERMINE THE EFFICACY AND SAFETY OF THE NOVEL GPR40 AGONIST TAK-875 IN SUBJECTS WITH T2DM

Prabhakar Viswanathan, MBBS, PhD (Takeda, Deerfield, IL)

Dr. Viswanathan presented results from a double-blind dose-ranging study assessing the safety and efficacy of TAK-875, a novel GPR40 agonist. In the 12-week study, subjects were randomized to placebo (n=61), 6.25 mg (n=60), 25 mg (n=61), 50 mg (n=60), 100 mg (n=62), or 200 mg TAK-875 (n=60) oncedaily, or glimepiride (n=62), with or without metformin. At baseline, patients in the various treatment arms had average baseline A1cs between 8.2% and 8.6%. After 12 weeks of treatment, all doses of TAK- 875 conferred significant reductions in A1c beyond placebo; for doses of 50 mg TAK-875 and above, efficacy plateaued at an approximate 1.0% A1c reduction beyond placebo. In addition, TAK-875 treatment provided significant reductions in both fasting and postprandial glucose, with a rapid onset of action. TAK-875 treatment was also associated with a significantly lower rate of hypoglycemia compared to glimepiride; all doses of TAK-875 were well tolerated and treatment-emergent adverse events were balanced across study arms.

• TAK-875 is a novel GPR40 agonist being developed as an adjunct therapy for type 2 diabetes. It has been demonstrated to be a highly selective and potent G protein-coupled receptor 40 agonist, and brings about glucose-dependent insulinotropic action, while conferring low hypoglycemia risk.
• In the 12-week study, subjects were randomized to placebo (n=61), 6.25 mg (n=60), 25 mg (n=61), 50 mg (n=60), 100 mg (n=62), or 200 mg TAK-875 (n=60) once daily, or glimepiride (n=62), with or without metformin. At baseline, nearly three-fourths of participants were on background metformin therapy. Patients were evenly balanced in age (~55 years of age), diabetes duration (~5.5 years), and A1c (~8.4%) at baseline.

• All treatment arms (TAK-875 and glimepiride) brought about significant reductions in A1c beyond placebo over the course of 12 weeks. Dr. Viswanathan noted that the A1c changes with TAK-875 at doses of 50 mg or higher were comparable to that of glimepiride. From the slides shown during the presentation, we estimated the following:

Even at the earliest measured (four-week) time point, a significant reduction in A1c was observed with TAK-875 beyond placebo. With regards to fasting plasma glucose, a close to dose-dependent change was observed; with postprandial glucose (as assessed by two-hour glucose during OGTT), a dose-dependent relationship was observed with increasing doses of TAK-875. For TAK-875 doses of 50 mg and higher, there were significant improvements in beta cell function as assessed by HOMA-B. Meanwhile, there were no significant effects of TAK-875 on body weight.

• With the exception of hypoglycemia, treatment-emergent adverse events were mild in nature and balanced across the study arms. Compared to placebo, those on TAK-875 experienced significantly less of hypoglycemia. Only 2.0% of patients receiving TAK-875 experienced hypoglycemia, compared to 3.3% on placebo, and 16.1% on glimepiride.

Questions and Answers

Q: Were there any changes in insulin an hour post dose, or during the glucose challenge?

A: There were significant increases in the glucose challenge insulin AUCs, but no changes in fasting insulin.

Q: Given that TAK-875 is a fatty-acid agonist, would you expect to see differences with a mixed meal versus a glucose challenge?

A: We have not performed meal challenges yet for TAK-875. We did do a preliminary food effect study, and we found no impact. In the future, we will do standard meal challenges.

Q: In the oral presentation yesterday, reductions in triglycerides were observed. Did you see any change in the lipid profiles?

A: Both in the oral session yesterday as well as in the proof-of-concept US study presented in the President’s Poster session, TAK-875 was shown to decrease triglycerides and LDL levels; however, we didn’t see any consistent change in lipids in our three-month study.

Q: Considering the long effect of the molecule, were there any differences in terms of the duration or intensity of hypoglycemia events?

A: All hypoglycemia was based on self-monitoring of blood glucose, so the degree of detail was not apparent. There didn’t seem to be any difference in the nature of hypoglycemia events as far as we could tell. However, the number of subjects with multiple episodes of hypoglycemia was much higher with glimepiride treatment.

LONG-TERM, INJECTION-FREE TREATMENT WITH ITCA 650, CONTINUOUS SUBCUTANEOUS DELIVERY OF EXENATIDE VIA DUROS DEVICE, LEADS TO SUSTAINED IMPROVED GLYCEMIC CONTROL AND WEIGHT LOSS FOR 48 WEEKS IN METFORMIN-TREATED TYPE 2 DIABETES

Julio Rosenstock, MD (University of Texas Southwestern Medical School, Dallas, TX)

Dr. Rosenstock announced the results of the 48-week extension study of ITCA 650, Intarcia’s implantable DUROS device that delivers continuous subcutaneous administration of exenatide. This extension study builds on positive results from 24-week data presented at EASD 2010. A1c reductions were sustained in all treatment groups and patients continued a trend of losing weight through 48 weeks. Notably, Dr. Rosenstock implied that Intarcia would focus further evaluation of ITCA 650 on the 20 mcg/day and 60 mcg/day doses – this decision was based on the further reduction in A1c and body weight obtained when moving from the 20 mcg/day dose to 60 mcg/day dose in this study. He also noted that these data support further development of ITCA 650 for an extended period of implantation (up to six and 12 months, rather than three months). During Q&A, we also heard Dr. Rosenstock reference an auto-insertion device for this product – we look forward to more details on the insertion/ removal process from a patient and provider perspective. Lastly, Dr. Rosenstock speculated that the DUROS implantable device may improve long-term outcomes, given the guaranteed adherence – we assume this will be an advantage from a payer perspective.

• ITCA 650 based on the DUROS device, which relies on an osmotic mini-pump. The DUROS device consists of five components: semi-permeable membrane, osmotic engine, piston, drug formulation, and a diffusion moderator. The semi-permeable membrane allows the entrance of a minute amount of water from the subcutaneous tissue driven by the osmotic gradient. In this trial, the device was implanted every three months (delivering exenatide over a three-month period). Data on 12 week and 24 week usage of ITCA 650 were presented at EASD 2010 (for more information, see the EASD Full Report on December 31, 2010 Closer Look). Dr. Rosenstock cited data showing that the release rate from the DUROS device remains stable over 36 months – he noted that this has been tested with interferon and exenatide.
• ITCA 650 is roughly the size and shape of a matchstick and is inserted in a physician’s office. He noted that the insertion procedure takes 10-15 minutes after applying local anesthesia. The physicians must pierce the skin to make a ~5 mm insertion through which they can slip the DUROS device under the skin. We have also heard estimates that the physicians will make, on average, $250 per procedure per patient, which is higher than we had expected, and we look forward to hearing more about expectations on the reimbursement front.
• Dr. Rosenstock described the design of the initial proof of concept dose-ranging study. Type 2 diabetes patients with an A1c between 7-10% were eligible to participate in the study. At randomization, there were 50 participating sites and 155 patients. During these 12 weeks, there were significant reductions in A1c in all three groups: exenatide BID (0.8%), ITCA 650 20 mcg/day (0.9%), and ITCA 650 40 mcg/day (1.0%).
• At 12 weeks, patients were subsequently randomized to receive various doses of ITCA 650. After 12 weeks, patients on ITCA 650 20 mg/day were randomized to 20 and 60 mcg/day; those on exenatide BID were re-randomized to 40 mcg/day and 60 mcg/day; and those on ITCA 650 40 mcg/day were re-randomized to 40 mcg/day and 80 mcg/day. After 24 weeks, patients had the option to continue treatment for an additional 24 weeks. Approximately 85% of patients volunteered to continue taking ITCA 650 with the same dose.

• In weeks 1-12, the frequency of nausea was initially ~25% in the ITCA 650 20 mcg/day arm compared to ~23% in the exenatide BID arm and ~35% in the ITCA 650 40 mcg/day arm. After 12 weeks, exenatide BID and ITCA 650 40 mcg/day had the same frequency of nausea (20%), while patients on ITCA 650 20 mcg/day reported a very low frequency of nausea (~2%) (all figures were extrapolated from a graph Dr. Rosenstock presented; exact figures were not provided). For comparison, the incidence of nausea with Novo Nordisk’s liraglutide peaks within the first few weeks of treatment, but plateaus at roughly 4% after 12-16 weeks of treatment (for more information on the time course of nausea with liraglutide, see ECO coverage in June 2, 2011 Closer Look). Given that there is not a way to titrate up the dose, we are cautious about this aspect of therapy and how patients will respond “in real life” – we also note that the best doctors “in real life” know how to titrate exenatide so that nausea is lower.
• At week 24, patients were stratified in the following treatment regimens: ITCA 650 20 mcg/day (n=13), ITCA 650 40 mcg/day (n=22), ITCA 650 60 mcg/day (n=22), andITCA 650 80 mcg/day (n=13). From 24-48 weeks, decreases in A1c were sustained. While the time course of A1c was provided, specific figures were not – patients in the two higher doses (60 and 80 mcg/day) ended 48 weeks with ~1.5% A1c reduction from baseline, while patients in the two lower doses (20 and 40 mcg/day) ended the study with ~1.0% A1c reduction. Interestingly, each treatment group experienced further weight loss from week 24 to week 48: 20 mcg/day (2.12 to 2.74 kg weight loss), 40 mcg/day (3.93 to 4.93 kg), 60 mcg/day (3.43 to 3.49 kg), and 80 mcg/day (3.34 to 3.57 kg).
• Dr. Rosenstock presented adverse events of special interest: GI events and insertion site-related adverse events. At the end of 48 weeks, roughly 10.5% of patients reported nausea and 3.5% reported diarrhea. Patients also reported experiencing various skin-related adverse events: irritation (7%), pain (7%), erythema (4.7%), pruritus (3.5%), hematoma (3.5%).

Questions and Answers

Dr. David Kendall (Chief Scientific and Medical Officer, ADA): Were there any significant limitations to this device, such as time to learn the implantation procedure?

Dr. Rosenstock: The device is very easy. As you can imagine, I have my research physician assistant do it. It takes 10-15 minutes to clean the site, and there is plenty of space in the abdomen. You make a small incision of 5 mm. There is a device that puts it into the subcutaneous layer and a lever you pull back.

Q: What if patients don’t come back to you? Does it continue to provide exenatide beyond three months?

Dr. Rosenstock: You can maneuver with the semi-permeable membrane how much you put in formulation into the reservoir. If they don’t come back, they run out of solution. The good thing, especially when we talk about using long acting GLP-1 therapies, is that for whatever reason, if something happens (e.g. some acute event), you can pull it out easily and the levels of exenatide will immediately drop because of the half life.

Q: This is very interesting research. Could you clarify the PK profile of exenatide with this new device?

Dr. Rosenstock: Looking at this evidence and what you can get in terms of exenatide, you get around 250 picograms/ml. When you have studies with LAR, there are huge variations in pharmacokinetics – it’s 300 picograms/ml in general. Here, you have a little less, around 250 picograms/ml, which is much better than exenatide twice daily.

Q: Was there any experience of tissue reaction?

Dr. Rosenstock: There is a little bit of inflammation – we don’t see much of a problem though. It is easily taken out. There is not much fibrosis. And at least in Texas, we have plenty of space.

Q: Do you insert it at the same place?

Dr. Rosenstock: We change it.

Dr. Kendall: Is there any device that can auto-remove the device?

Dr. Rosenstock: No, I don’t think so.

UTILIZING SELF-MONITORED BLOOD GLUCOSE DATA TO FURTHER CHARACTERIZE GLYCEMIC CONTROL IN THE ACCORD TRIAL

Richard M. Bergenstal, MD (International Diabetes Center at Park Nicollet, Minneapolis, MN)

The ACCORD trial has been quite controversial ever since its publication in June 2008, but patient level SMBG data has never been examined. In an important late-breaking oral presentation, Dr. Bergenstal presented an initial analysis of 9.4 million SMBG data points from a 5,347 patient subset (52% of all patients in the trial). The analysis found that the intensive group tested their blood sugar more frequently and had a lower mean blood glucose than the standard group. Frequency of testing was significantly related to A1c, with more SMBG tests per day associated with better glycemic control. Interestingly, a plotted analysis revealed that those who died or had severe hypoglycemia had more unstable modal day glucose profiles. A bucketed analysis revealed that the intensive arm had triple the rate of hypoglycemia and half as many readings >200 mg/dl relative to the control arm. Dr. Bergenstal emphasized that mortality in the study was linked with divergence from glycemic target. In his opinion, after setting A1c and SMBG goals, an inability to reach target might be a sign to relax treatment intensity (of course the opposite of this was done in the intensive group of the ACCORD trial).

• Questions still remain about the cause of the increased mortality in the intensive arm of the ACCORD trial. Data has shown that a drop in glucose, a low A1c level, weight gain, use of TZDs or other medications, and severe hypoglycemia were not the cause of the increased mortality in the intensive arm. Dr. Bergenstal believes problems occur when healthcare providers continually increase therapeutic intensity and without seeing a response.
• This study examined the SMBG data from 2,691 patients from the intensive arm and 2,656 patients from the standard arm of the ACCORD trial. The SMBG sample of patients used for analysis did not differ clinically from the group without adequate SMBG data that was excluded from the study. Data was submitted from 51 out of the 77 ACCORD study centers, creating a dataset with 9.4 million data points.
• The intensive arm of the trial tested more frequently (2.7 times per day vs. 2.0 times per day; p <.0001) and had a lower mean blood glucose (126 mg/dl vs. 157 mg/dl; p<0.0001). As a reminder, the original ACCORD study protocol advised patients only on diet and oral medications to test <7 tests per week in the standard group and >2 tests per day in the intensive group (or four times per day in the intensive group if not achieving target). For those on insulin, <3 tests per day were targeted for the standard group and 4-8 tests per day were targeted in the intensive group.
• Mean A1c was associated with frequency of SMBG, with the intensive arm achieving lower A1c’s at all testing frequencies:

• The 24-hour modal day SMBG profile reveals that both the intensive and standard groups had similar glucose profiles: lowest at 4-6AM and highest from 10PM-12AM. The intensive group had lower glucoses (p<0.0001) at every time point compared to the standard group.
• Glucose profiles were more unstable in those who died and those with severe hypoglycemia. According to Dr. Bergenstal, “It’s not such a good thing to be going up and down with such velocity.”
• The intensive arm had triple the rate of hypoglycemia and half as many readings>200 mg/dl relative to the standard arm:

• Mortality was linked with what Dr. Bergenstal called “divergence from glycemic target.” Looking at those who died in the intensive group, there was a greater incidence of hyperglycemia than in those who stayed alive (hypoglycemia rates were similar between those that died and lived in the intensive group). In the standard arm, those who died had more hypo- and hyperglycemia relative to those who stayed alive.

Questions and Answers

Dr. Phillip Raskin: What was particularly interesting in the patients who died was the increase in the standard deviation. Did you try to corroborate that in a more direct way? Can you give us a take home, practical message about how we should be quantifying variability in our patients’ SMBG levels if it’s as much as a risk factor?

Dr. Bergenstal: We are in the process of carrying this analysis out. It’s of great interest to us. It does look like the variability was particularly an issue. We will have those numbers in due time to see how much of a role stability and variability play.

Q (Australia): I think we should be careful about means and standard deviations. You found that the mean was dropping overnight. But people may have measured overnight only if they were hypoglycemic. And for percentages, people might measure only if they are hypoglycemic.

Dr. Bergenstal: That’s why we compared within-group rather than across groups. Of course, you’ll monitor more though. Your point is well taken.

Q (San Antonio, TX): I’m wondering about the percentages of blood sugars that were pre- prandial and post-prandial?

Dr. Bergenstal: From my clinical experience, the standard group was mostly pre-prandial and the intensive group was a combination. About 70% were pre-meal and 30% were post-meal for the intensive group. For the standard group it was 90% and 10%.

Q: Did you look at the relationship between fingersticks and mortality? I’m thinking that fingersticks would be a marker for patient effort.

Dr. Bergenstal: We don’t at this time. Not that we couldn’t, but we did it to A1c. We’re collecting this data. These are good ideas.

Q: Could it be that the standard deviation was smaller because those who died had a smaller n?

Dr. Bergenstal: I don’t know, it definitely was a smaller number. It does look a little more variable.

Dr. Andy Drexler (Los Angeles): What percentage of patients’ tests were overnight? Can you correlate blood glucose measurements with the insulin algorithms that were used? How many were on long-acting and short-acting insulin?

Dr. Bergenstal: We did take a quick look and didn’t see anything striking enough. Time will reveal these things.

RANIBIZUMAB (ANTI-VEGF) FOR VISION LOSS DUE TO DIABETIC MACULAR EDEMA – RESULTS OF TWO PHASE III RANDOMIZED TRIALS

David Boyer, MD (Retina Vitreous Associates, Los Angeles, CA)

Dr. Boyer presented analyses of phase 3 trials of Lucentis (Roche [Genentech]’s ranibizumab) for diabetic macular edema (DME), bolstering previously presented efficacy results with strong data across a range of secondary efficacy endpoints. Compared to those given sham injections, statistically significantly more patients receiving Lucentis achieved 20/40 vision or better at 24 months – a finding Dr. Boyer highlighted, since 20/40 represents the legal threshold for driving a car as well as an approximate cutoff for ability to read a newspaper. Other notable benefits of Lucentis therapy included lower rates of progression to proliferative diabetic retinopathy, as well as improvements in best corrected visual acuity (BCVA) that became apparent seven days after the start of treatment and were maintained for 24 months. Ocular and systemic safety events were similar to those in past clinical studies of Lucentis, which has been FDA-approved for neovascular age-related macular degeneration (wet AMD) since 2006. As a reminder, the company plans to file a supplementary biologics license application (BLA) by the end of 2011.

• In the double-blinded, multi-center RIDE and RISE trials (n=382; 377), adult patients with diabetes (type 1 or type 2) and DME were randomized to receive monthly intravitreal injections of 0.3 mg Lucentis (n=125; 125) or 0.5 mg Lucentis(n=127; 125), or monthly sham injections (n=130; 127). Exclusion criteria included a history of vitreoretinal surgery in the study eye, panretinal laser photocoagulation or macular laser photocoagulation in the study eye within three months of screening; previous use of intraocular or periocular corticosteroids in the study eye within three months of screening, previous treatment with Lucentis or other anti-angiogenic drugs in either eye within three months prior to day zero of the study; and a history of myocardial infarction or cerebrovascular accident within three months prior to day zero. Macular laser rescue treatment was made available to all patients starting at month three, based on pre-specified criteria, and panretinal photocoagulation was available for all patients when clinically indicated.
• Topline two-year results were released earlier this year for RISE and RIDE (see February 15, 2011 Closer Look and the April 8, 2011 Closer Looks, respectively). Management has said that the trials will continue blinded until month 24, during which patients in the control group can be switched to 0.5 mg Lucentis (data expected 1Q12 for both trials). Open-label extensions will follow.
• Patients had a mean age of 61.7-63.5 years, with mean diabetes duration of 14.5-16.6 years and mean A1c of 7.6-7.7%. They were 65-80% male and 77.6-82.7% Caucasian. 54.7-letters, corresponding to vision of roughly 20/80.
• Dr. Boyer highlighted that statistically significantly more patients in each Lucentis group achieved 20/40 vision or better at 24 months. He explained that 20/40 vision is the legal threshold for being able to drive, and it also roughly represents the cutoff for being able to read a book or newspaper.
• Lucentis treatment was similarly effective whether or not patients’ baseline A1c was above 8.0%. Patients with better baseline glucose control appeared to achieve better BCVA gains in RISE, but this relationship was not borne out in RIDE or in the pooled data.
• As previously noted, significantly more Lucentis patients improved their best- corrected visual acuity (BCVA) by 15 letters or more after 24 months (the studies’ primary efficacy endpoint). Lucentis also led to statistically better improvements in mean BCVA, which were apparent one week after starting treatment and sustained for two years, as well as statistically significant improvements in central foveal thickness (less than a 135 decrease in each sham group vs. decrease of 250 or more in each Lucentis group). Also, over 35% of patients in each Lucentis group improved by two or more steps in the Retinopathy Severity Scale, as opposed to 7% or less for the sham groups.

• Fewer patients in the Lucentis groups required macular laser or panretinal photocoagulation, and fewer Lucentis patients experienced a vision loss of 15 letters or more at month 24.
• Functionality gains were seen in other measures besides BCVA. Patients’ scores on the Visual Function Questionnaire 25 (VFQ-25), a 25-question assessment of perceived visual function, increased by a mean of roughly 3.0 with Lucentis than with sham therapy (by way of reference, a 5-point change is considered equivalent to a 15-letter change). Contrast sensitivity, a measure of people’s ability to distinguish shades of gray from a white background (Pelli-Robson chart), improved by more than 2.0 points in each Lucentis group while declining by 0.2 in both sham groups.

• Lucentis’ safety profile was consistent with Lucentis phase 3 trials for age-related macular degeneration, and the therapy was associated with fewer adverse events associated with diabetic retinopathy, including retinal neovascularization and vitreous hemorrhage, iris neovascularization. Minor ocular adverse events more common in the Lucentis arms included increased intraocular pressure; this included measurements of intraocular pressure taken 60 minutes post-injection.

• Systemic adverse events were generally balanced between groups. Rates of stroke, death, vascular death, and hypertension were slightly higher among Lucentis users than those receiving sham injections. Arteriothrombotic events as a whole were more common among the sham injection groups.

THE 10-YEAR COST-EFFECTIVENESS OF LIFESTYLE INTERVENTION OR METFORMIN FOR THE PRIMARY PREVENTION OF TYPE 2 DIABETES MELLITUS: AN INTENT-TO- TREAT ANALYSIS OF DIABETES PREVENTION

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

After providing background on the DPP and the DPP Outcomes Study (DPPOS), Dr. Herman presented the results from analyses of the 10-year cost effectiveness of lifestyle intervention or metformin for the primary prevent of type 2 diabetes. In the analyses, lifestyle intervention was found to be cost effective ($8,925/quality-adjusted-life-year [QALY] from a health system perspective, and $15,343/QALY from a societal perspective), while metformin treatment was shown to be cost saving. Based on the findings of this study, Dr. Herman suggested that health policy and societal policy should support funding of intensive lifestyle intervention and metformin for the prevention of diabetes.

• The study assessed the cost effectiveness of lifestyle and metformin interventions relative to placebo, using an intent-to-treat analysis spanning the combined 10 years of the DPP and DPP Outcomes Study (DPPOS). Data on resource utilization cost and quality of life were collected prospectively during the DPP and DPPOS, economic analyses were performed from a health system perspective that considered direct medical costs, and sensitivity analyses were performed from a societal perspective that took into account both direct medical costs and direct nonmedical costs (e.g., diet and activity costs, participants’ time, and transportation).
• From both a health system perspective and a societal perspective, lifestyle intervention was found to be cost effective, while metformin intervention was demonstrated to be cost saving.
  • ​Health system perspective: Over 10 years, the cumulative, undiscounted, per- participant direct medical costs of the DPP/DPPOS were approximately $4,500, $2,500, and $750 for the lifestyle, metformin, and control groups. Dr. Herman noted that the cost of lifestyle and metformin intervention were much lower during the DPPOS than they were during the DPP; that is, the incremental costs were low. The 10-year cumulative, undiscounted, per-participant direct medical costs of medical care received outside the DPP/DPPOS were much higher than DPP/DPPOS costs themselves, costing approximately $25,000 in each group. These costs increased year over year, likely due to increased comorbidities and complications for those who developed diabetes. In total, these costs amounted to $24,563, $25,429, and $27,150 per participant in the lifestyle, metformin, and control groups. The bulk of the increased costs in the control group were related to outpatient visits, inpatient care, prescription medications, and self-monitoring supplies and lab tests. Summing all costs, the 10-year cost of lifestyle intervention remained the greatest, while metformin treatment actually cost less than control (the total difference between groups was relatively trivial). Meanwhile, undiscounted mean health utility scores were consistently higher with lifestyle than with metformin or control. Taking everything into account, the undiscounted cost per quality-adjusted-life- year (cost/QALY) was $8,925 ($12,336 discounted) with lifestyle; metformin treatment was cost saving. For context, Dr. Herman noted that cost-saving treatments are quite rare (e.g., the influenza vaccine), while $10,000/QALY is on the low end of the scale for many therapies and procedures (e.g., beta blockers after myocardial infarction, mammographicscreening, hypertension medications for those with diastolic blood pressure over 105 mg/dl) (Neumann, NEJM 2005).
  • Societal perspective: From a societal perspective, the total 10-year per-participant costs for lifestyle, metformin, and control treatment were calculated to be $130,259,$127,066, and $129,355. The undiscounted cost/QALY for lifestyle intervention was$15,343, while metformin was cost saving.

Questions and Answers

Q: Since the DPP, metformin has gotten less expensive, and we’re now increasingly thinking of lifestyle interventions in groups rather than in individuals. Do you have any additional analyses looking at the impact of those changes?

A: Yes, we did use generic pricing of metformin in our analyses. However, we looked at lifestyle as it was implemented in the DPP. If it could be done in a group and maintain the same effectiveness, it would be more cost effective.

Q: Was there any lifestyle intervention in the inpatient setting?

A: Patients were not hospitalized for the lifestyle intervention. We did look at hospitalization outside the study as a cost.

Q: Some efforts are underway to simplify the DPP. Do you have plans to look at some of those newer models?

A: Anything that would decrease the cost while maintaining efficacy would be more cost effective. We’re conducting a per-protocol analysis of participants who adhered to metformin in the study.

Q: My clinic has been doing the DPP since 2004; we’re an urban Indian clinic. According to the Indian Health Service, people with diabetes have annual medical costs of about $15,000 per person per year, while a person without diabetes has medical costs of about $2,500 per year. All the people who have been reversing their diabetes with the lifestyle program are promoting cost savings for the system. From what you said though, there didn’t seem to be much of a difference around this – can you help me understand why this is?

A: I believe the costs you’re citing are for the general diabetes population, and the general population without diabetes. What we found is that the costs for people who developed diabetes was not much higher than the costs for those with impaired glucose tolerance. The patients with diabetes in the DPP did not have advanced complications of diabetes. To the extent that patients will go on to develop more complications and comorbidities, lifestyle and metformin interventions will be more cost effective with longer follow-up.

Symposium

Treatment of Type 1 Diabetes — Update on Clinical Trials

PROTÉGÉ

Nicole A. Sherry, MD (Massachusetts General Hospital for Children, Boston, MA)

Dr. Sherry discussed the results from the phase 3 trial Protégé for MacroGenics’s anti-CD3 therapy teplizumab in individuals with recent onset type 1 diabetes. As was previously announced, although no unanticipated safety issues were identified, the trial failed to meet both its primary (composite of A1c < 6.5% and insulin dose requirements <0.5 U/kg/day at one year) and key secondary endpoints (change from baseline in A1c, insulin dose requirements, and C-peptide levels at one year). However, Dr. Sherry reported the results from a post-hoc analysis that found a significant preservation in stimulated C- peptide levels in individuals treated with the 14-day regimen of 17 mg of teplizumab relative to placebo (p<0.05). This treatment regimen was particularly effective at preserving C-peptide in children ages 8 to 11, participants from the USA, and individuals diagnosed with diabetes within six weeks of the study’s initiation. The post-hoc analyses also indicated that teplizumab significantly reduced insulin requirements while maintaining glycemic control in some individuals. More specifically, a significantly greater percentage of individuals treated with teplizumab (13%) versus placebo (3.1%) achieved an A1c < 7.0% and insulin requirements <0.25 U/kg/day at one year (p=0.006). Furthermore, 5% of teplizumab subjects were off insulin after one year versus none on placebo. Overall, while it appears unlikely that the compound will have much clinical usefulness in the overall recent-onset setting, we did find the results from the post-hoc analyses to be somewhat intriguing. We hope future mechanistic studies will help uncover why these compounds may be more effective in children and other subpopulations.

• The phase 3 trial Protégé investigated the safety and efficacy of the anti-CD3 therapy teplizumab in individuals recently diagnosed with type 1 diabetes (< 12 weeks since diagnosis) that were antibody positive and had detectable C-peptide levels. 516 individuals were randomized to four study groups: a 14 day regimen of 17 mg of total teplizumab (full 14 day regimen) (n=209), a 14 day regimen of 5.6 mg of total teplizumab (1/3 regimen) (n=102), a six day regimen of 4.6 mg of total teplizumab (six day regimen) (n=106), and placebo (n=99). Each individual received a repeat treatment of their particular regimen at six months. The primary endpoint of the study was a composite of A1c <6.5% and insulin dose requirements<0.5U/kg/day at one year. 497 individuals continued long-term follow up beyond one year, and data are still being collected on these individuals. Each arm was largely similar at baseline. Average A1c ranged from 8.1% in the six-day regimen to 8.4% in the 1/3 regimen; average insulin requirements ranged from 0.63 U/kg/day in the six-day regimen group to 0.68 U/kg/day in the 1/3 regimen.
• The primary and key secondary endpoints were not met in the study. The percentage of individuals with A1c <6.5% and insulin requirements < 0.5 U/kg/day on teplizumab (19.3% in the full 14 day regimen group, 13.7% in the 1/3 regimen group, and 20.8% in the six day regimen group) did not differ significantly from placebo (20.4%, no p value provided). Statistical differences were also not found in overall mean change from baseline in A1c, insulin dose requirements, and C-peptide levels.
• Post-hoc analyses suggested that the full 14-day regimen led to statistically significant preservations in C-peptide. Using a non-parametric statistical analysis (after discovering the C-peptide data had a non-normal distribution), it was found that the full 14-day regimen led to a significant preservation of stimulated C-peptide at one year relative to placebo (p<0.05). Furthermore, 40% of the individuals in the full 14-day regimen group exhibited no change in stimulated C-peptide versus just 28% in the placebo group. The full 14 day regimen was found to be particularly effective relative to placebo at preserving stimulated C-peptide in children ages 8 to 11 (55% vs. 29% had no change in C-peptide at one year), participants from the USA (33% vs. 13% had no change in C-peptide at one year), and participants that were diagnosed within six weeks of the trials initiation (59% vs. 32% had no change in C-peptide at one year).

• Post-hoc analyses also indicated that teplizumab significantly reduced insulin requirements in some individuals. Teplizumab treatment led to the elimination of insulin therapy in 5% of the full 14-day regimen group, 4% of the 1/3 regimen group, and 4% of the six day regimen group at one year, while no individuals in the placebo group came off insulin therapy. Combining all the teplizumab groups together, 5% of those treated with teplizumab versus 0% of those on placebo were off insulin at one year (p=0.032). In the full 14-day regimen group, insulin requirements were found to be significantly reduced at each level of A1c between<6.5% to <10%. Furthermore, the percentage of individuals achieving an A1c<7.0% and insulinrequirements <0.25 U/kg/day was significantly higher in the full 14 day regimen group (13%) than the placebo group (3.1%) at one year (p=0.006).
• According to Dr. Sherry, teplizumab was associated with an “acceptable” safety profile given the seriousness of type 1 diabetes. The percentages of individuals experiencing an adverse event and a serious adverse were similar in each group at approximately 99% and 10%, respectively. The percentages of individuals in which dosing was discontinued was significantly higher in each of the teplizumab groups at approximately 22% versus placebo at 11%. Discontinuations in the teplizumab groups were largely driven by adverse events. However, no unanticipated safety issues were discovered. The most commonly reported adverse events that typically occurred more often in the teplizumab groups than in the placebo group included cytokine release syndrome, rashes, headaches, vomiting, and chills (see table below). Infections occurred at similarly rates (both overall and for specific viruses) in both individuals treated with teplizumab and placebo.

DEFEND

Peter Gottlieb, MD (University of Colorado at Denver, Denver, CO) and Paolo Pozzilli, MD (The London School of Medicine and Dentistry, London, UK)

The Belgian Diabetes Registry Trial showed the ability of a 48 mg dose of otelixizumab to durably reduce insulin requirements over 48 months as compared to placebo. However EBV reactivation associated with dosing in this trial suggested that another dose should be considered. The TTEDD trial aimed to find a dose that would have an effect without harmful adverse effects; it identified 3.1 mg as an optimal dose. Most recently, DEFEND-1 trial was a phase 3, randomized, double blind study that explored whether a 3.1 mg otelixizumab dose given over eight days could reduce insulin requirements and preserve beta cell function in new onset type 1 diabetes. There was no difference in C-peptide levels or insulin usage at twelve months between the otelixizumab and placebo treated groups. Notably, there was also very little evidence of EBV reactivation. Generally, side effects were significantly reduced at the dose used in this study. However, this good side effect profile may have come at the expense of therapeutic efficacy.

• The Belgian Diabetes Registry Trial showed the ability of a 48 mg dose of otelixizumab to durably reduce insulin requirements over 48 months as compared to placebo. However EBV reactivation associated with dosing suggested that another dose should be considered.
• The TTEDD trial aimed to find a dose that would have an effect without harmful adverse effects. The study’s dose optimization strategy involved a desensitizing phase in which doses were increased daily and then an efficacy phase in which the same daily dose was administered. A dose of 3.1 mg was identified as optimal.
• The DEFEND trial was a phase 3, randomized, double blind study that explored whether a 3.1 mg otelixizumab dose given over eight days could reduce insulin requirements and preserve beta cell function in new onset type 1 diabetes. 240 patients ages 12-44 years were randomized to receive either otelixizumab or placebo in a 2:1 ratio. The study’s primary endpoint was change in C-peptide. Secondary endpoints included insulin usage and A1c.
• There was no difference in C-peptide levels or insulin usage at twelve months between the otelixizumab and placebo treated groups. C-peptide trends over a twelve- month period were also essentially the same between the two groups. While there was some reduction in A1c levels in otelixizumab treated patients during the course of the study, no reduction was apparent by twelve months. Notably, many of the individuals in both treatment and control groups were well controlled.
• Generally, side effects were significantly reduced at the dose used. There was very little evidence of EBV reactivation in the study. Some adverse events were observed, but these were expected based on previous study results. However, in this study, lymphocyte depletion was reduced as compared to previous studies. This suggests that reduced side effects may have come at the cost of efficacy; the administered dose may have been too low. Dr. Gottlieb noted that further dose ranging studies are necessary to determine the optimal dose of otelixizumab.

Questions and Answers

Q: You asserted that there was suboptimal tolerability in the study where some efficacy was seen. Specially, transient reactivation of EBV was seen. To throw away an efficacious dose for suboptimal tolerability is unfortunate. If you are seeing a beneficial effect, I would think you shouldn’t throw out the baby with the bath water.

Dr. Gottlieb: Tolerx made the decisions. EBV reactivation was transient but the concern was that it occurred in one or two individuals over time and an untoward disorder could develop. The choice of drug dose was probably too low. Especially if you look at teplizumab dosing, you realize you may need a dose that is slightly higher but doesn’t cause side effects.

Q: I am perplexed by the DEFEND protocol as well. It is clear from the Belgian study that one cycle of drug gave efficacy for a certain amount of time, and then that efficacy went away. I would think that the drug needs to be used in repeated cycles. That’s where the field was even several years ago. I was perplexed when DEFEND chose to test one cycle of drug when you knew that one cycle of the drug was not what patients would need.

Dr. Gottlieb: From a personal perspective, I think we are at a point in the field where we need to explore multi-drug combination therapy. The challenge of using monoclonal antibodies in humans is that the development of other antibodies against treatment antibodies can limit the time you get to use them. We have to come up with strategies to circumvent that or explore a different pathway. We are transitioning out to second order experiments, but first order experiments still have a good deal to teach us.

GAD65

Johnny Ludvigsson, MD, PhD (Linkoping University, Linkoping, Sweden)

Dr. Ludvigsson presented the results from the European phase 3 trial that examined the ability of the Gad65 vaccine Diamyd to preserve C-peptide in individuals recently diagnosed with type 1 diabetes. While reportedly well tolerated and safe (data not provided), the vaccine failed to meet both the primary (preservation of stimulated C-peptide) and secondary endpoints (changes in A1c, insulin dose, insulin dose adjusted A1c, and hypoglycemia) of the trial. For the majority of the presentation, Dr. Ludvigsson detailed the results from pre-specified subgroup analyses. It was discovered that stimulated C-peptide was significantly preserved in males, males diagnosed with diabetes for over 90 days before the study, males with baseline fasting C-peptide over 0.32 nmol/l, males aged 12-15 years, males with BMI classification less than 25%, males with a baseline insulin dose of 0.398 to 0.605 IUD/day/kg, and males from non-Nordic countries. The analyses also found that individuals experienced significant preservation of stimulated C-peptide if they were dosed with their first or second injection of Diamyd in the months of March or April, suggesting that seasonality may influence the efficacy of the vaccine. Interestingly, in the previously conducted phase 2 trial that showed significant preservation of C- peptide at 30 months, all individuals in the trial were dosed in March and April. Dr. Ludvigsson concluded by arguing that the positive phase 2 trial and these subgroup analyses lend support for the concept of auto-antigen treatment, and that rather than dismiss this treatment because of a failure to reach a certain endpoint, the scientific community needs to sit down and try to learn from the strategies and techniques used to develop successful allergy immunotherapies and cancer combination therapies.

• Dr. Ludvigsson reviewed the results from the 15-month European phase 3 trial that examined the efficacy of Diamyd’s GAD65 vaccine Diamyd versus placebo in preserving beta cell function in recently diagnosed individuals with type 1 diabetes (within three months of enrollment in the study). Participants were required to be between the ages of 10 to 20 years old and GAD autoantibody positive. The double-blind study randomized 334 individuals to one of three arms: 20 mcg of Diamyd given twice a day on day 0 and 20 (n=108), 20 mcg of Diamyd given four times a day on day 0, 30, 90, and 270 (n=111), and placebo (n=115). A mixed meal tolerance test (MMTT) was administered at baseline, 3, 9 and 15 months to assess changes in C-peptide levels. At baseline, average A1c was 7.0%, fasting C-peptide was 0.28 pmol/ml, stimulated C-peptide was 0.66 pmol/ml, and insulin dose/kg was 0.56. Differences inage were present between the arms. 10 to 11 year olds comprised 33.9%, 31.8%, and 23.4% of the four dose Diamyd arm, two dose Diamyd arm, and placebo arm. Meanwhile, 16 to 20 year olds comprised 18.3%, 11.2%, and 21.6% of the same respective arms.
• Both Diamyd treatment regimens failed to meet the primary (change in stimulated C-peptide) and secondary endpoints (change in insulin dose, A1c, and hypoglycemia) of the trial. While change in meal stimulated C-peptide was numerically lower in both Diamyd arms relative to placebo at 15 months, the treatment effect was not statistically significant in the two dose arm (p=0.19) or the four dose arm (p=0.12). Similarly, while change in fasting C-peptide was numerically lower in both Diamyd arms, the treatment effect was not statistically significant in the two-dose arm (p=0.08) and the four-dose arm (p=0.16). Although exact data was not provided, Dr. Ludvigsson revealed that changes in A1c, insulin dose, insulin dose adjusted A1c, and hypoglycemia were also not significantly different between each Diamyd arm and the placebo arm.
• Treatment with Diamyd did, however, significantly preserve stimulated C-peptide levels at 15 months in several subgroups in the combined four dose and two dose Diamyd arms relative to placebo. These subgroups included males (p=o.0093), males diagnosed with diabetes for over 90 days before the study (p=0.0440), males with baseline fasting C-peptide over 0.32 nmol/l (p=0.468), males aged 12-15 years (p=0.0132), males with BMI classification less than 25% (p=0.0389), males with a baseline insulin dose of 0.398 to 0.605 IUD/day/kg (p=0.0334), and males from non-Nordic countries (p=0.0048). Also interesting, subgroup analyses revealed that individuals in the combined Diamyd arms experienced significant preservation of stimulated C-peptide if they were dosed with their first of second injection in March or April (p=0.0244), suggesting that seasonality may influence the efficacy of the vaccine. Dr. Ludvigsson noted that in the previously successful phase 2 trial for Diamyd (which showed significant preservation of both fasting and stimulated C-peptide at 30 months) dosing in March and April. Another difference between the phase 2 and phase 3 trial, according to Dr. Ludvigsson, was the percentage of participants that received influenza vaccinations. Because the phase 3 trial took place during the swine flu outbreak, many more participants in this trial received an influenza vaccination. While not statistically significant, individuals in the phase 3 trial that were vaccinated for influenza more than 150 days after the 1st Diamyd injection trended towards experiencing a significant preservation of stimulated C-peptide (p=0.0713). Meanwhile individuals that were vaccinated for influenza within 150 days of their first Diamyd injection showed no trend toward stimulated C-cell preservation (p=0.876).

Questions and Answers

Q: What do you think the prospect is for GAD65 vaccines?

A: Well, we have to sit down and look at all the data. We are certainly disappointed, but we cannot completely throw away these findings or even this drug. Maybe we have to look at different dosing strategies, etc.

CTLA4IG

Tihamer Orban, MD (Joslin Diabetes Center, Boston, MA)

Abatacept (CTLA4-Ig, BMS’ Orencia) is a selective T cell costimulation modulator that impairs the full activation of T cells. This randomized, multicenter, controlled clinical trial explored whether abatacept was capable of preserving beta cell function in patients newly diagnosed with type 1 diabetes. Adjusted mean C-peptide was 59% higher in the abatacept treated group as compared to the control group at two years. C-peptide declines were slowed by 9.6 months in the abatacept treated group as compared to placebo. While C-peptide had declined by 67% from baseline in the placebo treated group by two years, it fell by only 46% in the abatacept treated group. Fewer abatacept treated subjects had C-peptide levels below 0.2 nmol/L at two years. Significantly more abatacept treated patients had an A1c of less than 7% at two years, but there was no difference in insulin usage between the groups. The protocol did result in adverse events, though not severe. These results suggest that abatacept may be useful in slowing the decline of beta cell function in those with recent onset type 1 diabetes.

• Abatacept (CTLA4-Ig, BMS’ Orencia) is a selective T cell costimulation modulator which impairs the full activation of T cells. It has been approved in the United States for use in rheumatoid arthritis and juvenile idiopathic arthritis. It functions by binding to the CD80/86 complex, thus impairing the signaling between antigen presenting cells and T cells.
• This randomized, multicenter, controlled clinical trial explored whether abatacept was capable of preserving beta cell in function in patients newly diagnosed with type 1 diabetes. It enrolled 112 people ages 6-45 with stimulated C-peptide levels greater than or equal to 0.2 nmol/L and detectable autoantibodies. Patients were randomized 2:1 to receive either abatacept at a dose of 10 mg/kg with a maximum of 100 mg/dose or to receive placebo. Patients received 27 intravenous infusions over two years. The study’s primary outcome measure C- peptide levels after two hours of a four-hour mixed meal tolerance test. Secondary outcomes included rate of change of C-peptide, time to first stimulated C-peptide less than 0.2 nmol/L, A1c levels and insulin doses at regular time points throughout the study, and several safety and mechanistic parameters.
• Adjusted mean C-peptide was 59% higher in the abatacept treated group as compared to the control group at two years. The mean C-peptide for the abatacept group was 0.375 nmol/L and that for the placebo group 0.266 nmol/L. C-peptide declines were slowed by 9.6 months in the abatacept treated group as compared to placebo. However, C-peptide levels declined in parallel for abatacept and placebo treated patients from six months of treatment onwards. Dr. Orban suggested this might indicate that T-cell activation decreases over time.
• While C-peptide had declined by 67% from baseline in the placebo treated group by two years, it fell by only 46% in the abatacept treated group. Fewer abatacept treated subjects had C-peptide levels below 0.2 nmol/L at two years. At one year, C- peptide had fallen by 28% in the abatacept treated group versus 46% in the placebo treated group. It is notable that those receiving abatacept lost the same amount of C-peptide by two years as the placebo group had lost by one year. At two years, 32.5% of abatacept treated patients had C- peptide levels below 0.2 nmol/L as compared to 42.9% of placebo patients.
• Significantly more abatacept treated patients had an A1c of less than 7% at two years, but there was no difference in insulin usage at two years. 47.2% of those in the abatacept group and 25.8% of those in the placebo group achieved this A1c metric. Differences in insulin usage were only observed between the groups at six and twelve months.
• The protocol did result in adverse events, though not severe. 22% of abatacept treated patients experienced infusion reactions, while 17% of placebo patients did. There was no difference in infection or neutropenia rates between the two groups and no severe hypoglycemic events were reported.

• These results suggest that abatacept may be useful in slowing declines in beta cell function in those with recent onset type 1 diabetes. It may also be useful in prevention studies in those at high risk of diabetes or as one of the components in a combination therapy.

ABATE (ANTI-CD3 MAB)

Stephen E. Gitelman, MD (University of California San Francisco, San Francisco, CA)

Dr. Gitelman presented the results from a phase 2 study that examined the ability of two courses of the anti-CD3 therapy teplizumab (Macrogenics) administered 12 months apart to preserve beta cell function in individuals recently diagnosed with type 1 diabetes. Encouragingly, treatment with teplizumab led to a significant reduction in C-peptide loss at 24 months, the primary endpoint of the trial. However, the beta cell preserving effect of teplizumab largely occurred within the first six months after the first course was dosed. For the remaining 18 months of the trial, C-peptide levels in both the teplizumab and control arms decreased at approximately the same rate, and this rate was not meaningfully affected by the second course of teplizumab treatment. A similar pattern was observed for both insulin use and A1c, which were only found to be significantly improved in the teplizumab arm between months 12 and 18 and months 6 and 15, respectively. Attempting to explain why the second course of teplizumab treatment had limited effectiveness, Dr. Gitelman presented data to suggest the expression of anti-idiotype antibodies may have attenuated responses to the therapy. As expected, adverse events typically associated with anti-CD3 therapy were observed in the trial, but very few serious adverse events were reported. Dr. Gitelman closed with a discussion on the next steps for the teplizumab program, which include: 1) conducting mechanistic studies to evaluate the parameters that change in responders to the therapy as well as how the efficacy of the second course of treatment could be enhanced; 2) further analyzing results from the phase 3 trials for teplizumab; 3) pairing teplizumab with other drugs that work through other mechanisms to achieve synergy; and 4) evaluating the efficacy of teplizumab in a prevention setting (TrialNet has recently launched a type 1 diabetes prevention trial for the drug).

• AbATE was a phase 2 open-label study that randomized 81 individuals with recently diagnosed type 1 diabetes (enrollment within eight weeks of diagnosis) to treatment with teplizumab (n=56) or intensive diabetes management (n=27). Individuals in the teplizumab arm received two separate 14-day courses of the drug, once at the beginning of the study and once 12 months later. The primary endpoint of the study was mean change in four hour AUC C-peptide following a mixed meal tolerance test (MMTT) at 24 months. Secondary endpoints included insulin use, time to undetectable C-peptide, and A1c. At baseline, the average age was 12 years, BMI was 20 kg/m², and insulin use was 0.39 U/kg/day in both arms. A1c was slightly lower in the teplizumab arm (7.02%) than in the control arm (7.29%). Of the 52 individuals that received at least a single dose teplizumab, 29% of individuals did not complete the entirety of both courses due to pre-specified criteria for drug discontinuation (i.e. low CD4 count, increased INR, development of IgE antibodies, etc).
• While C-peptide declined between baseline and two years in both groups, C-peptide loss was significantly reduced with teplizumab treatment. Using intent to treat (ITT) analysis, mean four hour AUC C-peptide levels following a MMTT were found to decrease by 45% in the teplizumab arm and 77% in the control arm (p=0.002). A similar result was observed using a per protocol analysis (which necessitated that individuals had received 80% of the expected drug does in each of the 2 dosing cycles). When four hour AUC C-peptide levels were plotted at six-month intervals, it was discovered in both the ITT and per protocol populations that C-peptide levels were stabilized by teplizumab through six months, but decreased steadily for the remainder of the trial (without any noticeable change upon the second course of dosing) at a rate that was similar to what was observed in the control arm. In addition to preservation of C- peptide, it was found that a significantly greater percentage of individuals in the teplizumab arm had a clinically meaningful C-peptide response during a MMTT (>0.2 pmol/ml) at 24 weeks (80% vs. 40%, p=0.001) as well as detectable C-peptide during a MMTT (90% vs. 65%, p=0.001).
• In the ITT population, insulin use was significantly lower at 12 and 18 months (p<0.001), but not at 24 months. Similar to the C-peptide results, teplizumab treatment appeared to stabilize insulin requirements through six months, but thereafter, insulin requirements rose at a rate that matched that observed in the control group.
• A1c was significantly lower at 6, 12, and 15 months only (p<0.05) in the teplizumab arm. Again teplizumab was found to stabilize A1c through the first six months of the study, which was followed by rise in A1c for the remainder of the trial at a rate that matched that observed in the control arm. Dr. Gitelman noted, however, that the lower baseline A1c in the teplizumab arm could have influenced these results.
• To Dr. Gitelman, these results suggested that the second course of teplizumab treatment had little overall therapeutic effect. To further substantiate this claim, he noted that declines in circulating T cells were smaller and the occurrence of rashes and viral reactivation were less frequent with the second course of treatment. When looking at responders and non- responders to the first course of treatment at 12 months, it was found that anti-idiotype antibodies were more common in non-responders than responders (p<0.01), suggesting that the expression of these antibodies may contribute to an attenuated response to the second course of the treatment.
• Teplizumab treatment was associated with a number of expected adverse events; however, the occurrence of serious adverse events was rare. The serious adverse events associated with teplizumab treatment include one case of diarrhea, two cases of cytokine release syndrome, one case of infection/cellulitis, and one case of infection with normal ANC. Other adverse events experienced in the trial are listed in the table below (ITT population). Although rates of infections were high, Dr. Gitelman stressed that rates of infection were also high in the placebo group and that no unusual opportunistic infections were observed with teplizumab treatment.

Questions and Answers

Q: Can you comment on whether there were any changes in autoreactive T cell reactivity toward beta cell antigens in the teplizumab arm?

Dr. Gitelman: We are currently analyzing that data.

Q: How you looked at whether any particular genes are associated with response to the drug?

Dr. Gitelman: We did not look at that in this trial.

Q: Your data showed that you were getting the best effect at six months. Have you thought about switching to dosing every six months?

A: When we looked at the initial data from the previous phase 2 study, the decline seemed like it began to occur after 12 months. So, I still think that this dosing scheme seems reasonable. You also have to remember that there is also individual variance. Not all individuals will begin exhibition C-peptide decline following treatment at the same time.

DIAPEP277

Itamar Raz, MD (Hadassah University Medical Center, Jerusalem, Israel).

DiaPep277 is a synthetic peptide derived from the heat shock protein 60 that induces anti-inflammatory T cells and could block the destruction of beta cells (manufactured by Diamyd). Preliminary studies in NOD mice demonstrated that even treatment of animals with clear-cut diabetes with DiaPep277 could stop deterioration of glucose metabolism. DiaPep277 has been explored in both preclinical and clinical trials. Eight phase 2 studies were conducted in adults and children with either newly diagnosed or established type 1 diabetes. These studies showed the compound to be safe and tolerable and a trend towards efficacy was seen in all phase 2 studies though only one study reached statistical significance. Two phase 3 studies are in progress. DIA-AID1 is enrolling patients ages 16-45 years old who were diagnosed with diabetes within the past three months in Europe, the United States, and Israel. DIA-AID 2 is similarly designed but is enrolling newly diagnosed patients ages 20-45 years old at 115 medical centers worldwide.

• DiaPep277 is a synthetic peptide derived from the heat shock protein 60 that induces anti-inflammatory T cells and could block the destruction of beta cells. It is an antigen to both T and B cell receptors and is involved in adaptive immunity. By activating Toll- Like Receptor 2, it activates an anti-inflammatory response that might protect beta cells.
• Preliminary studies in NOD and streptozocin-induced diabetic mice demonstrated that treatment with DiaPep277 could stop deterioration of glucose metabolism. Histology confirmed preservation of insulin production and reduced infiltration of immune cells into islets.

• DiaPep277 has been explored in both preclinical and clinical trials. It showed safety and efficacy in animal models and in Phase 1 trials was shown to be safe and tolerable in type 1 patients with established disease. Eight phase 2 studies were then conducted in adults and children with either newly diagnosed or established type 1 diabetes. These studies showed the compound to be safe and tolerable and a trend towards efficacy was seen in all phase 2 studies; a statistical difference in C-peptide levels between treatment and control groups was seen when all four studies were combined, while in study 420, the drug did significantly reduce declines in C- peptide levels. Treatment additionally induced production of IL-10 but did not induce shifts in cytokine responses to bacterial antigens.
• Two phase 3 studies are in progress. DIA-AID1 is enrolling patients ages 16-45 years old who were diagnosed with diabetes within the past three months and have a fasting C-peptide of more than 0.2 nmol/L. It is recruiting in Europe, the United States, and Israel. Patients receive either 1.0 mg DiaPep277 or placebo four times a year for two years. So far, DiaPep277 has had an excellent safety profile in this study. The most common adverse event has been injection site discomfort in the first hours after injection. DIA-AID 2 is similarly designed but is enrolling 450 newly diagnosed patients ages 20-45 years old at 115 medical centers worldwide.

Questions and Answers

Q: Studies with DiaPep277 have routinely involved older subjects. Why not go to younger subjects?

Dr. Raz: There have been two studies in young patients. They also showed signs of a positive effect, but the effect was not as strong as that seen with other patients. So we are concentrating on a group that gives us a better chance to show activity.

Current Issue

Basal Insulin — Is NPH and NPH-Based Premixed Insulin a Competitive Candidate in the Era of Analogs?

INTRODUCTION AND SESSION OVERVIEW

Bernard H. Charbonnel, MD (University of Nantes, France)

Dr. Charbonnel led off the morning session with a comprehensive historical review of insulin, progressing from beef and pork insulin in 1922, to NPH in 1950, and finally concluding at the end of the century with short- and long-acting analogs. He called pump therapy the “gold standard” in type 1 diabetes because it most closely mimics physiologic control. For type 2 diabetes, the literature suggests that short-acting prandial insulins are associated with a greater A1c reduction, but more weight gain. Thus, Dr. Charbonnel believes that initiation of insulin treatment should occur with basal insulin, with further intensification as necessary.

YES

Paolo Rossetti, MD, PhD (Polytechnic University of Valencia, Spain)

In this side of the debate, Dr. Rossetti argued that NPH was a competitive candidate in an era of analogs. He began by establishing a proper framework for comparing insulins against each other. In his view, PK/PD studies are the gold standard, but to interpret the data properly, (1) both glucose infusion rate (GIR) and plasma glucose should be examined; (2) subjects with type 1 diabetes should be used because healthy volunteers are “very troublesome;” and (3) dose must be examined. In type 1 diabetes, Dr. Rossetti showed that long-acting analogs have modest, but greater A1c efficacy as well as lower incidence of nocturnal and severe hypoglycemia when compared to NPH. However, he called these clinical benefits “expensive” after examining their cost-effectiveness. For type 2 diabetes, he said there was “definite evidence” that NPH and long-acting analogs are equivalent in terms of glycemic control (A1c), though noted glargine and detemir are superior to NPH in terms of incidence of hypoglycemia (30-50%). A cost-effectiveness analysis also revealed that the number of total QALYs gained in type 2 diabetes is quite small with long-acting analogs. As a result, Dr. Rossetti advocates for NPH as the starting insulin in type 2 diabetes; however, he also supported an individualized treatment approach, especially if the aim is strict metabolic control. In such cases, long-acting analogs may become cost effective due to their significant reduction in hypoglycemia.

NO

Francisco Javier Ampudia-Blasco, MD, PhD (University Hospital of Valencia, Valencia, Spain)

Dr. Ampudia-Blasco presented data to support the use of long-acting insulin analogs over NPH-based insulins. In type 1 diabetes, he cited data showing a 31% reduction in symptomatic hypoglycemia and a 54% reduction in nocturnal hypoglycemia when using long-acting insulin analogs (insulin glargine or detemir) compared to NPH insulin. In type 2 diabetes, he advocated initial use with analogs only in patients with “brittle” type 2 diabetes or patients prone to hypoglycemia or with a long duration of diabetes. We were surprised that he suggested using NPH for type 2 diabetes at the beginning of disease (citing more stable diabetes and the ability of insulin resistance to confer resistance to hypoglycemia). He concluded that long-acting insulin analogs are as effective as NPH and that when hypoglycemia is a limiting factor, both analogs are superior compared to NPH insulin.

• Dr. Ampudia-Blasco presented a patient-level meta-analysis including 3,175 patients with type 2 diabetes from six randomized controlled studies comparing long-acting insulin analogs and NPH insulin. He noted that the use of long acting insulin may enable forced titration algorithms without (or less) hypoglycemia, suggesting that physicians may be able to employ a more aggressive treatment algorithm with long-acting analogs. In combination with oral agents, NPH insulin was similarly effective as insulin glargine or detemir in lowering A1c. In addition, insulin detemir induced less weight gain. During Q&A, many criticized the use of this meta-analysis, given its inclusion of trials using NPH once daily, and advocated for the inclusion of trials using NPH twice daily.

Questions and Answers

Q: I think we have to split the discussion of type 1 and type 2 diabetes. So first, for type 1 diabetes, can you comment about the general conclusion by Dr. Ampudia-Blasco, Dr. Rossetti?

Dr. Rossetti: Basically, when I was invited to speak here to defend NPH, I was a little surprised because I have been working for a long time with type 1 diabetes and Javier has actually shown my own data showing insulin analogs can be superior to NPH. Of course, I would say that NPH is not the best choice in type 1 diabetes because of its variability. But we do have data that demonstrates the importance of the reduction in relative risk for type 1 diabetes. The reduction in relative risk of hypoglycemia is actually associated with a better counter-regulatory response to hypoglycemia in glargine-treated patients as compared to NPH. We have data demonstrating this reduction in hypoglycemia and translating to a real benefit. So in type 1 diabetes, I agree that analogs are perhaps the better choice. However, I would like to say that a good job can be done with NPH in type 1 diabetes. If you split NPH into three or four doses, the nocturnal dose is so small that the peak will be very small as well, so you are reducing the risk of hypoglycemia. Remember that the issue of hypoglycemia risk is derived from the large effect of NPH; it also depends on the insulin dose. So NPH works well and the only difference is in nocturnal hypoglycemia.

Dr. Ampudia-Blasco: Obviously, I would say that with NPH insulin, which we have used for a long time and is working well, we can use it at least three or four times daily. This is what we do in pregnant women with type 1 diabetes. But we should remember that we are treating human beings. If you achieve the same goal and reduce the number of injections, I think this is a major benefit, so I use NPH insulin only in the case of pregnant women.

Q: We can switch this discussion to type 2 patients, which is a different issue. What are you comments?

Dr. Rossetti: When looking at type 2 diabetes, we should always look at the absolute incidence of hypoglycemia and not just the relative risk. If we look at the absolute risk in type 2 diabetes patients, it is about 8-10 episodes/patient/year, so the risk is quite low. Therefore, a 30% reduction amounts to two episodes less per year. Do you think this has clinical implications? I don’t think so. Instead, if we spend this money in education that has a greater impact on A1c reduction than the use of analogs, I think we should achieve a better result. In type 2 patients, I would start with NPH insulin and not insulin analogs, since there is no data showing that the reduction in this hypoglycemia results in better outcomes or counter-regulatory responses to hypoglycemia. There is a subset of type 2 diabetes patients that do not experience hypoglycemia at all. So why spend more money to treat them if we can achieve the same results with NPH?

Dr. Ampudia-Blasco: In type 2 diabetes, it’s different, in my opinion, where the patients have to pay for long acting insulin. It seems reasonable to start with NPH insulin and reserve long acting analogs for those at risk for hypoglycemia. In my country, in Spain, although we have economic troubles, even for GPs and specialists, the use of long acting insulin analogs is well established. If you were to start with the safest strategy to reduce the goals of hypoglycemia, you can also think about the number of injections. I assume that for people with type 2 diabetes, most will need two injections of NPH insulin. If you can achieve this with a once-daily injection of glargine or detemir, it can be a major advantage.

Q: If I understood your slide correctly, compared to type 1 diabetes, the duration of effect seems to be longer in type 2 patients compared to type 1 patients. Is that true?

Dr. Rossetti: Type 2 diabetes patients are not good models for clamp studies because some of them may retain insulin secretion capabilities so it may not accurately reflect the pharmacodynamics. All three insulins have >24-hour duration of action. If you look at clinical studies, you have the same metabolic control as obtained with glargine or detemir, so it’s not surprising to have such PD data.

Q: Does obesity play a role? Does fat in the subcutaneous tissue play a role in delaying insulin absorption?

Dr. Rossetti: I think there are no clinical data. Detemir appears to be less effective in more obese patients compared to insulin glargine and NPH.

Q: One of the difficulties of using basal insulin in type 1 diabetes is the lack of flexibility. When we move to twice daily NPH, we get far better results with flexibility. That’s one of the real world findings we have in the UK. We got better glycemic control with twice-daily NPH compared to once-daily glargine. With NPH, there was also no increased hypoglycemia.

Dr. Ampudia-Blasco: In the meta-analysis I showed, some were once-daily NPH and some were twice- daily. But I completely agree with you. Adjusting insulin doses according to carbohydrates and improving flexibility is the key to success.

Q: I would stress that in this meta-analysis, the best result would be comparing one injection of glargine vs. one injection of NPH. It is quite obvious you would have less nocturnal hypoglycemia with this type of study if you do once glargine vs. twice-daily NPH.

Dr. Stephanie Amiel (King’s College, London, UK): I think the other issue in type 2 diabetes is the magnitude of reduction in weight gain and hypoglycemia risk. For every 130 type 2 patients you put on a long acting analog, you could actually employ a fairly highly qualified diabetes educator. Talking about what we want to spend our money on, if you have a patient with a high risk of hypoglycemia or weight gain, a long-acting analog is a good choice, but for the majority, it might be better to give them insulin and teach them how to use it properly.

Q: There are some parts of the world where the only available basal insulin is NPH, so it is good to see that it is still an alternative. We should bear in mind though that there are places where it is the only option. When you split the dose three or four times, how much do you use and when do you dose?

Dr. Rossetti: You can give one dose at bedtime, around 11 pm, of about 0.15-0.2 U/kg and the rest can be split into two or three doses between 0.05-0.1 U/kg doses at each meal, depending on the composition and size of the meal.

--by Adam Brown, Ben Kozak, Lisa Rotenstein, Joseph Shivers, Sanjay Trehan, Nick Wilkie, Vincent Wu, and Kelly Close