Practical Ways to Achieve Targets in Diabetes Care: Barbara Davis Center Keystone Conference 2012

July 12-15, 2012; Keystone, CO Full Commentary – Draft

Executive Highlights

As always, the 2012 Barbara Davis Center Keystone Conference attracted some of the most highly regarded thought leaders in diabetes and produced rich dialogue to help HCPs improve diabetes care. Dr. Satish Garg (University of Colorado School of Medicine), once again, organized and chaired the conference, and we are grateful to him for gathering together these movers and shakers in diabetes.

A major theme this year was the cost effectiveness of diabetes care (especially in the face of an uncertain future in healthcare policy). Dr. David Kendall (Distinguished Medical Fellow at Eli Lilly) suggested that although prevention can have serious benefits in specific populations, it would make sense to invest more effort in improving early intervention. In a controversial debate, Dr. Irl Hirsch (University of Washington) and Dr. Amanda Adler (Adenbrooke’s Hospital) took sides on the question of whether insulin analogs are worth their cost. On hypoglycemia, which played a large role in cost-effectiveness discussions, Dr. Adler presented on her opinions for how to reduce hypoglycemia in type 2 diabetes, and Dr. H. Peter Chase (University of Colorado) spoke optimistically about CGM as a solution for hypoglycemia in type 1 diabetes. Dr. Richard Bergenstal (International Diabetes Center) spoke on lessons learned about hypoglycemia from ACCORD, where mortality increased with glucose variability.

Another big topic this year was CGM, predictive technologies, and the closed loop. Dr. Fran Kaufman (Chief Medical Officer, Medtronic Diabetes) reviewed the results of the ASPIRE Trial, expressing hope that long-term use of low glucose suspend would help reduce hypoglycemia by disrupting the vicious cycle of “hypoglycemia begetting hypoglycemia.” Dr. John Pickup (King’s College London School of Medicine) expressed his view that algorithms and pumps have been sufficiently optimized at this time and that the bottleneck for developing a closed loop will be developing more reliable glucose sensing and improving the rapidity and stability of insulin absorption. Dr. Richard Bergenstal (International Diabetes Center) presented his take on improving glucose control through frequent insulin dosing titrations (suggesting that Hygieia’s DIGS device could assist in this process) and through standardized reports from blood glucose meters and CGM. Dr. Helena Rodbard (Endocrine and Metabolic Consultants) also presented on the value of automated clinical decision support systems to improve clinician’s ability to recognize and respond to glycemic abnormalities from structured SMBG data.

Other notable talks included Dr. Shashank Joshi’s (Lilavati and Bhatia Hospital) presentation on the unique challenges that Asian countries face regarding diabetes; a session on the JDRF, industry, and FDA perspectives on diabetes regulatory policies; Dr. Jay Skyler’s (University of Miami) compelling presentation on the importance of preventing microvascular complications as a means of preventing macrovascular complications; Dr. Robert Ratner’s (CSO and CMO of the ADA) exciting overview of five drug therapies (SGLT-2, 11B HSD inhibitors, DGAT, G-protein coupled receptor agonists, and glucokinase activators); and the esteemed Dr. George Eisenbarth’s (Executive Director of the Barbara Davis Center) presentation on type 1 diabetes treatment that emphasized that while we can now predict and diagnose type 1 diabetes, we cannot yet safely prevent the disease. He highlighted that a new approach to prevention under investigation in his lab involves targeting the trimolecular complex.

We are excited to bring you our full Barbara Davis Center Keystone Conference report. Talks not previously published in our daily conference coverage have titles highlighted in yellow.

Table of Contents 

Detailed Discussion and Commentary

What is Making Headlines for Diabetes


Robert Ratner, MD (Chief Scientific & Medical Officer, ADA)

After starting his presentation with a history our nation’s long path to the March 22, 2010 passing of the Affordable Care Act, Dr. Ratner highlighted the rising healthcare costs in diabetes, and the need for diabetes prevention.

  • Dr. Rather highlighted the rising healthcare costs in the US. Today, annual healthcare expenditures have reached $2.6 trillion dollars, marking a seven-fold increase in cost per capita over the past 30 years. (And despite spending a comparatively larger amount of our GDP on health care – 18% GDP vs. EU’s average 7% – the US does not enjoy a proportionally greater quality of care, ranking 19th in mortality.) Yet, Dr. Ratner explained that had any previous historical attempt to nationalize healthcare passed, our nation would be in a much different position. For example, had President Carter’s healthcare plan passed, costs would be down ~$1.7 trillion. Dr. Ratner emphasized that it has been the delay in instituting health reform that has lead to the recent healthcare spending explosion. Dr. Ratner then narrowed his focus to diabetes, in the context of rising healthcare costs. He pointed to a number of illuminating statistics – diabetes results in $153 billion in excess medical costs annually (2007 data); one third of the Medicarebudget is spent on patients with diabetes, even though they only constitute 20% of the patient population; and medical expenditures in diabetes care stem primarily from hospitals with 50% from hospital inpatient care. It is the cost of diabetes, Dr. Ratner said, that really makes it a national problem. He also noted the harrowing trend between the obesity epidemic and type 2 diabetes – the epidemic of diabetes lags behind the trends in obesity, so even though there is some leveling off of obesity, we won’t see the down turn in the diabetes epidemic for some time.
  • He then discussed what we need to see in the future – prevention efforts and extended coverage. He pointed to DPP/DPPOS to demonstrate that that the cost involved in preventing diabetes are miniscule compared to the costs involved in taking care of diabetes. While this is true, of course, since it is challenging to show who is most at risk, and since the majority of people with pre-diabetes do not move on to diagnosed diabetes, there is still reluctance to invest here, which is unfortunate. Second, he commented on the need to reduce the number of uninsured; he pointed to current estimates of 53 million people, approximately 20%, in the US without insurance. We also point out that unfortunately, disproportionately more of this population are at lower socioeconomic levels, and are at higher overall risk of diabetes. For diabetes specifically, those lacking insurance over the course of a year had a 25% reduction in diabetes control compared to those with insurance. Third, he noted that the new provisions in the Affordable Care Act could help on both of these fronts in people with diabetes. Particularly, the extension of coverage to dependents up to age 26 could make a big difference for children with type 1 diabetes, and of course the law’s prevention of coverage denial for pre-existing disease would be huge for people with diabetes.


David Kendall, MD (Eli Lilly, Indianapolis, IN)

Dr. David Kendall, former CSO and CMO of the ADA and now Eli Lilly’s Distinguished Medical Fellow, asked the question – is diabetes prevention cost effective in the current setting of increasing diabetes prevalence? Dr. Kendall made the argument that early, intensive diabetes intervention deserves equal, if not more, attention than diabetes prevention. Globally increasing prevalence of both type 1 diabetes and type 2 diabetes (projected to reach 500 million worldwide by 2030 [we think this estimate is conservative]) are putting a strain on healthcare systems around the world to care for those that are already affected, and many developing countries are not yet equipped to do so. So, he argued, while studies like the DPP and Da Qing demonstrated that prevention for high-risk individuals can indeed produce very positive results (30-60% reduced risk for progressing to diabetes), it really only does so for a subset of the population (those who possess the resources for lifestyle modification, are resistant to many obesity promoting environmental factors, and do not have a genetic predisposition to beta cell deterioration). For those for whom weight loss is a significant challenge and quickly regain it, the onset of disease may only be delayed 2-10 years, and protection against long-term outcomes may not even be conferred. So while prevention can be important, it is also necessary, as demonstrated by early intervention studies like UKPDS, DCCT, and Kumamoto (which pinned down early intensive control as critical for reducing risk of microvascular complications and diabetes related deaths), to focus on achieving blood glucose control “as early as possible, as intensively as possible, as long as possible, as safely as possible, and as rationally as possible” immediately after diagnosis. He painted prevention as a public health challenge and early intervention as a clinical challenge, and he stated, “I always balance clinical trials with clinical reality.” Since there are still such significant challenges in managing established diabetes, he implied it might be more cost-effective at this point in time to invest more effort in improving early intervention.


Judith Fradkin, MD (NIDDK, Bethesda, MD)

Dr. Fradkin’s presentation touched on the expansive scope of what’s going on in diabetes, with the take home message being that we still have lots to learn – that may sound like more of a basic message but she really did underscore multiple areas of study that need more focus. As background, Dr. Fradkin opened with a different interpretation of the 10-year data from the DPPOS, casting a slightly more positive light on prevention than offered by Dr. Kendall. She noted that after the first three years, while the rates of diabetes development became the same, the study had delivered a modified version of the lifestyle change to the other groups. She believed what we were seeing was the reduction of the rate of diabetes in the other two groups as well. Dr. Fradkin particularly emphasized the role of prevention in children at risk of developing type 2 diabetes, considering how difficult this population is to treat. She pointed to the disquieting results from SEARCH showing that those who stand the most to benefit from good glycemic control – youth with diabetes – tended towards the worst control.

  • In addition to prevention, another theme of Dr. Fradkin’s presentation was the importance of determining which therapies work for which patients – this dovetails well, of course, with the heightened emphasis by ADA and EASD to personalize treatment, individualize treatment, and optimize treatments by moving patients to new treatments, especially combinations, when monotherapy or various combinations stop working. She pointed to an NIDDK planning grant for a major multicenter clinical trial to address what is the best class of drugs to add to metformin when metformin alone is no longer adequate in type 2 diabetes as one such effort on this front – we assume this is Dr. David Nathan’s GRADE trial, which is taking some time to get approved and started. We are really looking forward to seeing what this looks like and as a sidenote, we hope that in particular, a variety of SFUs will be used since we would like a major trial to test whether these therapies deserve to be on the market. She also delved into bariatric surgery as another area needing additional exploration – we think it’s a big deal that she said that and were happy to hear that this will be an area of focus for the NIH. She commented that most of the information we have about bariatric surgery are short term and, notably, said that NIDDK had begun a new longitudinal assessment across 10 institutions – we look forward to hearing which institutions and how this will work. She said that we have many lingering uncertainties including: 1) what role bariatric surgery should have in less obese patients; 2) the mechanism of bariatric surgery in diabetes remission; 3) the optimal timing of bariatric surgery relative to other interventions; and 4) the predictors of success in the procedure. She also called for more randomized controlled trials comparing available surgical procedures and comparing these procedures to lifestyle intervention.
  • Dr. Fradkin concluded by highlighting several new studies that the NIDDK is involved with that could shed new light in the diabetes arena. She set the for investigation into gestational diabetes by describing the viscous cycle that can occur when the intrauterine environment predisposes children to an increase risk of type 2 diabetes, begetting future generations that will be exposed to a similarly maladaptive intrauterine environment. We were glad to hear her emphasize this since it hasn’t been discussed as much of late compared to several years ago, by our memory. She then pointed to two studies to increase our knowledge on this front: 1) Lifestyle Interventions in Expectant Moms study (LIFE-moms) which seeks to improve mothers weight and prevent gestational diabetes; and 2) Hyperglycemia and Adverse Pregnancy Outcomes Study (HAPO) follow up to determine the long term consequences of maternal hyperglycemia in the mother and child (we assume this is a follow up since it’s already reported). Dr. Fradkin also remarked on The Environmental Determinants of Diabetes in theYoung (TEDDY), another interesting study being done in collaboration with the Barbara Davis Center - people with type 1 diabetes have been waiting for this study to report for some time. Its primary goal is identifying triggers that cause children to get type 1 diabetes and has the secondary goal of establishing a central repository of data and biological samples. Currently, over 8,000 children with high genetic risk have been enrolled (of which 100 now have developed type 1 diabetes) and will be followed until age 15.


Shashank Joshi, MD (Lilavati and Bhatia Hospital, Mumbai, India)

Dr. Joshi highlighted the increasingly global burden of diabetes and emphasized some of the unique challenges Asian countries face. Dr. Joshi asserted that Asia is the epicenter of the diabetes epidemic – in 2011, China and India led the way in diabetes prevalence with 90 million and 61.3 million cases of diabetes, respectively. Forty-six percent of diabetic deaths in this region occur in people under 60 years of age, and almost 50% of people with diabetes are undiagnosed in Southeast Asia. Dr. Joshi declared that diabetes is a much more widespread problem than diseases like HIV/AIDS or breast cancer, yet does not receive nearly as much publicity. Asian countries are starting to experience the same social and cultural changes that Western countries went through in the past decades, wherein “affluenza” is infecting urban populations, prompting them to consume excess calories and lead more sedentary lives. Compounded with this factor, Asian countries are more crowded, so physical activity is not promoted. Biologically, Asians are at a disadvantage compared to Caucasians and people of African descent. Asian Indians and Chinese have an average body fat composition of 33% and 30%, respectively, compared to 27% for Caucasians. Furthermore, Asians tend to store more fat intra-abdominally so that at the same BMI, Asians have an elevated risk for developing diabetes. The intra-abdominal fat storage (as opposed to subcutaneous fat storage) and relative lack of metabolically active brown fat means that Asians have a blunted insulin response. Therefore, the BMI cutoff for diabetes risk in Asians is 23 kg/m2, compared to 25 kg/m2 for Caucasians, and Asians progress from prediabetes to type 2 diabetes three times faster than Caucasians (who have an average prediabetes duration of 10-15 years). Dr. Joshi estimated that India and China will lose up to $900 billion in national income between 2005 and 2015 because of diabetes.


Robert Ratner, MD (ADA); David Kendall, MD (Eli Lilly, Indianapolis, IN); Judith Fradkin, MD (NIDDK, Bethesda, MD); Shashank Joshi, MD (Lilavati and Bhatia Hospital, Mumbai, India).

Q: You talked about a lot of trials coming up, one of which talked about finding the right drug for the right patient. It’s my sense that drug companies aren’t too keen on us figuring that out. How is that going to change clinical care? Are we going to check C-peptide in patients with type 2 diabetes? Or check their genes? How do you think that is going to play out?

Dr. Fradkin: We have already learned a lot about pharmacogenetics from clinical trials we’ve done. For example, the DPP showed that lifestyle worked in people that had high-risk genes for type 2 diabetes. On the other hand, people who had a variant of the metformin transporter did not respond to metformin. I think pharmacogenomics should be a component of every comparative effectiveness drug trial. In terms of the trial I was speaking about, we will enroll a population that is on metformin and then add one of four different classes of drugs and then go back retrospectively, and I think as we learn more about genes, it’s likely that specific characteristics will emerge.

Q: This question is for Bob. You said that you weren’t sure people were getting their money’s worth in healthcare. How do you think Americans would find out if they are getting their money’s worth?

Dr. Ratner: There are two unique ways to look at that question. First is that Americans loathe being told that they can’t have something. People are talking about death panels and about the health plan limiting ability to get care. The word that was used was rationing. Rationing of healthcare is the ultimate negative in terms of dealing with healthcare reform. However, having said that, Americans love nothing more than a sale. They want to get value for their money. It is nicer to go to Nordstrom and buy a $5,000 suit for $150 dollars than to buy the suit for $5,000. We are spending seven times what other developed countries are spending per capita. Again from a value judgment system, if you are getting seven times the value maybe it is worth it, but we are not. We are not doing as well in terms of outcomes with chronic conditions. You have to look for the sale and you have to look for the value.

Dr. Kendall: In the US in particular, we like nothing more than a sale, but second to that we like nothing more than a buffet. When Bob talks about rationing, what it is, is stepping back from the buffet and saying there are just three choices, because those are the menu items that are the greatest value. You can’t keep going back to the buffet for no good reason.

Q: Speaking of the sale analogy, we all know that what’s on sale is not necessarily what we want to wear. My concern is that the managed care system severely limits the quality of care patients are getting due to the system that defines diabetes on its own terms. So what can I expect from system that is defining type 1 diabetes on C-peptide? What can I expect if I have a system that denies children access to pumps and devices based on that? I think it’s only going to get worse.

Dr. Ratner: We’re in the midst of a swing of the pendulum. We’ve historically had a fee-for-service payment system – regardless of the outcome, doctors were paid for doing more. The pendulum began to swing in the opposite direction when payers said we’re not going to cover this. We don’t care what the outcomes are – it’s too expensive. Both extremes are wrong. The movement currently is to ultimately have what’s called accountable care. That means that you’re going to have bundled payments, and the healthier you can keep your patients, the more you’ll get paid. If it costs you $1,000 a year to put someone on a pump, but it keeps them out of the hospital compared to where they were before, then you’re saving the system money. That’s going to be valued and rewarded. We’re getting to a completely different payment system now, and I think that will ultimately be the solution.

Q: In Boston, Massachusetts we have instituted a nearly universal healthcare system, which the Affordable Care Act is based on. Based off the Massachusetts example, how is the nation looking? Will we be able to achieve our goals?

Dr. Ratner: I sure hope so. Romney Care set up exchanges, guaranteed coverage, took away preexisting conditions, and gave help to those who couldn’t afford it. The issue with Romney Care was that it did nothing in terms of work force, access, or cost. Those are the three major limitations. What Massachusetts is just beginning to implement are cost factors. Built into the Affordable Care Act is an enormous amount of money to test what can decrease cost. It is a bottom up way of looking for cost improvement.

Q: Dr. Joshi, you pointed out that Indians, for the same BMI, are getting type 2 diabetes earlier and within their working lives. Presumably they’re getting complications within their working lives then as well. So are the economic implications even greater in India?

Dr. Joshi: Absolutely. Our economy may be doing well now but bankruptcy may be caused by healthcare costs. And the costs for diabetes mostly come from complications. Most of our healthcare is private, not reimbursed, and this is another challenge. One good thing though is that healthcare is very economical and affordable, even to a middle class worker getting diabetes. In comparison, US drugs are very expensive. Delivery models in India are much more economical.

Q: Results of healthcare intervention studies are all represented in terms of relative risks. Can you explain to patients based on this, how long they could delay diabetes? Could you give us sense from Indian studies?

Dr. Joshi: Indian diabetes prevention programs are very poor. I don’t think lifestyle intervention is a good word. I think the way you look at life is what is important. We need more community intervention. New diabetes intervention programs in Indian are community based, rather than pharmaceutical or individual lifestyle based. So our approach is a little different; it is community focused rather than individual focused. It is very difficult for individuals to change habits.

Q: Just as a clarification – all “prevention” studies are prevention OR delay studies, obviously you can’t prevent death.

Dr. Kendall: I think the modeling that [Dr.] Bill Herman did was invaluable – over the course of a lifespan, what fraction of people who may have developed diabetes never developed diabetes? More importantly if you can delay onset of this disease that is going to consume healthcare dollars for years, it’s a bit like ACE inhibition and aggressive hypertension therapy for renal disease: 15 years of delay in terms of QALYs and health care expenditure is invaluable. These analyses are gong to be invaluable.

Q: I wanted to add a few comments to the nihilist perspective of Dr. Kendall. We’ve introduced a program working with the Indian Health Service replicating the DPP across 36 sites. The program is delivered in groups by culturally appropriate groups in that community. We’ve reached close to 3,000 people in that population. I just want to encourage that we don’t get too dismayed by what it takes to prevent disease. It’s making a tremendous impact in people’s lives.

Dr. Kendall: I was representing other perspectives. Particularly in highest risk communities, some of those programs have been unquestionably the most successful. The advantages are you are not over- resourced, so you have to deal with a modest approach, and it is integrated into your community. I think that is a great part of it and explains a lot of the success and sustainability of those programs.

Q: I have a comment in thinking about or getting prepared for the [healthcare reform debate] at the end of the day today. I would like to point out that we are focusing on adult type 2 diabetes and a bit on adult type 1 diabetes. With regard to the Affordable Care Act, pediatric patients’ needs are very different. So please think about that as we talk later in the day.

Dr. Satish Garg (University of Colorado School of Medicine, Denver, CO): Dr. Joshi, because you’ve come the furthest do you have any questions for any panelists?

Dr. Joshi: Dr. Fradkin, why is healthcare so expensive in the US? The US has the largest pharmaceutical market, but it’s all related to Wall Street. How come you can’t have drugs that are a little less expensive so that more people can get them? There is a huge market in the US and it’s so sad. It’s clear that there are many Indians who are underprivileged, but what is surprising is that there are lots of underprivileged folks in the US without access as well. Maybe you could have an access program for them for drugs, that maybe China or India can export. Byetta this way could be $4 a month, it’s very economical.

Dr. Fradkin: I don’t think I could address all the causes of the high cost in the US but one point I’d like to make in closing is the importance, when you consider costs, of the long-term perspective. Costs of diabetes are really the costs of complications. The US Congressional Budget Office, which decides what the cost of everything is, only considers a 10-year time frame. In DPP we’re already past 10 years, and we couldn’t even start to see complications occurring in the first 10 years. We don’t know if we’ll even see any after 15 years. So this 10 year cost perspective is very problematic. Everyone is looking for the low hanging fruit and they want to save money tomorrow, but that’s not possible with a chronic disease like diabetes.

Dr. Kendall: I’m not speaking here representing the pharmaceutical industry, but having worked in clinical setting, I’ll use complications as the example. In the US, you can have an MI or stroke without prior authorization, and it’s $100,000 dollars that you have to pay (or the system), and it’s exceedingly expensive. This doesn’t justify the cost of other procedures or the like, but I think Dr. Ratner showed something very instructive. While the acquisition cost of any procedure or product does need to be looked at very carefully, it cannot be looked at in isolation from the cost of care and burden on patients. For example, absence from work is an enormous cost for patients, and probably well outstrips drug acquisition costs.

Where are Insulins and Devices Going?


Irl Hirsch, MD (University of Washington, Seattle, WA)

Dr. Hirsch gave an entertaining talk on whether insulin analogs are worth their cost. He answered with a qualified yes, declaring, “it depends!” He concluded that while analogs clearly reduce the risk of hypoglycemia and improve quality of life for those who use them, their skyrocketing price increases may soon make them inaccessible for many to use. This is, of course, not the fault of patients or providers, and we believe the growing focus on costs of useful therapies, versus focus on cost of complications and benefits of avoiding short- and long-term complications, is unfortunate. For many patients, stability, flatter profiles, and lower hypoglycemia represent immeasurable or less quantifiable benefits conferred by analogs. Dr. Hirsch also expressed distress over the situation, asking the audience, “How can we consider even more expensive therapies, including the artificial pancreas, when many can’t even afford treatments from 10-15 years ago?”

  • Dr. Hirsch reviewed the history of the cost of insulin since its discovery in 1921. For most of insulin’s history, price increases have been relatively modest compared to the cost increase over the last several years, according to Dr. Hirsch. Throughout the 1960s, insulin sold for less than $1/vial, rising to $1.50-$3/vial in 1975. By 2001, when Eli Lilly held 84% of the insulin market, Lilly’s analog insulin lispro (Humalog) cost $35/vial, and Lilly’s synthetic human insulin (Humulin) cost $20/vial. However, as shown in the table below (source: Dr. Hirsch), costs in the last five to seven years have skyrocketed by more than 100% or 200%, compared to 10.6% and 17.5% inflation over the last five and seven years, respectively. We believe this is likely due to the price ceiling increasing on other drugs for people with diabetes and point out that the human insulin increase is on par with other increases, which is surprising since it has been “generic” for so long.
  • Dr. Hirsch compared the straight financial cost of using insulin analogs to the cost of hypoglycemia, concluding that on average a patient might pay an extra $1,100 in costs for analogs over human insulin to avoid a $120 cost/episode of severe hypoglycemia. There was no analysis on the benefits of less mild to moderate hypoglycemia. The severe hypoglycemia cost was derived from a 2003 study in Managed Care Interface where the mean cost was $1,186/severe hypo event, which included a range of $181-$4,924 per episode (the $120/cost today has been adjusted for inflation). Given that 100% of type 1 diabetes patients enrolled in the Helmsley Trust T1D Exchange database (a giant observational study looking at diabetes care in the US) are using analogs, Dr. Hirsch asked if a simple cost/patient/year justified the use of analogs. Assuming average adult patients now take 50 units/day of a vial analog, that would cost about $3,600/year (a conservative estimate), or $1,100/year more than on human insulin. Assuming, also, that analogs reduce the risk of severe hypoglycemic events by 25%, Dr. Hirsch calculated that every 1% reduction in risk costs $44/year. Therefore patients spend an extra $1,100/year to avoid a $120 cost/episode. He acknowledged that this kind of analysis has serious limitations, prefacing his talk by noting how important the topic is and stressing his limited background on cost-benefit analysis.

Insulin Type

2012 Cost/Vial

Present Cost/Unit

% Change in Cost

Humulin R U-500


4 cents

+246% in 5 years

U-100 Humulin/Novolin


7 cents

+114% in 7 years

Insulin lispro (Humalog)


14 cents

+134% in 7 years

Insulin glargine (Lantus)


14 cents

+116% in 7 years

  • Acknowledging that a rigorous cost-effectiveness evaluation takes more into account than simply a direct cost comparison, Dr. Hirsch cited a 2007 ecoCanada study from the Center for Outcomes Research examining incremental cost per quality adjusted life year (QALY) of using insulin analogs compared to human insulin and NPH. For patients with type 1 diabetes, the short-acting analog insulin lispro (Lilly’s Humalog) costs well less than CAN $50,000/QALY, the short-acting insulin aspart (Novo Nordisk’s Novolog) comes close to CAN $0/QALY (the cost of insulin aspart in Canada is very low), the long-acting basal insulin glargine (Sanofi’s Lantus) costs about CAN $90,000/QALY, and the long-acting insulin detemir (Novo Nordisk’s Levemir) costs close to CAN$400,000/QALY. For patients with type 2 diabetes, cost per QALY was at least six times higher inall cases, and the calculation for insulin detemir could not even be performed because patients on detemir in this database had higher A1cs on average than did patients on NPH (this was surprising and we wonder how other factors affected this). If one were to draw, in Dr. Hirsch’s words, conservative cost-effectiveness threshold at CAN $50,000/QALY, then the only therapies that would be “worth it” would be insulin lispro and insulin aspart for type 1 diabetes, and insulin aspart for type 2 diabetes.
  • Dr. Hirsch noted that as expensive diabetes is in general, medications and insulin therapy have skyrocketed in cost. CDC data shows that for 2007, the latest year data is available, therapy is about 20% of the costs of total diabetes care; we will try to find historic comparisons on this front to see if therapy has risen significantly related to total cost of care. Dr. Hirsch also noted a caveat – the costs of blood glucose strips were not included in his analysis. Presumably, strips would be optimal for patients taking either human or analog insulin.
  • In conclusion, Dr. Hirsch declared that whether or not insulin analogs are worth the cost “depends!” on a variety of factors; he also predicted that, although there is a clear consensus that analogs improve risk of hypoglycemia and quality of life, many US patients will be forced to “go back in time” to human insulin because payers and patients will not be able to afford analogs if prices keep increasing. Dr. Hirsch acknowledged that his analyses only took severe hypoglycemia into account and that it is harder to quantify the value of mild and moderate hypoglycemia. We were surprised, especially he (self-reportedly) still spends 40% of his time in the clinic seeing patients and hearing their concerns, that he did not explore the human cost of hypoglycemia in more depth.


Amanda Adler, MD, PhD (Adenbrooke’s Hospital, Cambridge, United Kingdom)

Dr. Adler, Chair of Technology at National Institute for Health and Clinical Excellence (NICE), argued that insulin analogs were not worth their cost in type 2 diabetes. She offered NICE’s perspective on the benefits – or lack there of – of prescribing long-acting insulin analogs versus NPH after metformin and sulfonylureas. While long-acting basal insulins modestly reduce the occurrence of hypoglycemia and insulin detemir (Novo Nordisk’s Levemir) was demonstrated to bring about statistically (but according to Dr. Adler, not clinically) significant favorable changes in weight versus NPH their effects on glycemic control are comparable. A cost per quality-adjusted life year (QALY) analysis (which included not just the cost of the drug itself, but also other costs, e.g., the costs associated with fewer complications) led NICE to conclude that providers should start most, but not all, people with type 2 diabetes on NPH following metformin and sulfonylureas. We were disappointed that A1c was the main focus of her analysis, since we believe multiple other factors come into play; more hypoglycemia actually lowers A1c, of course, so we wouldn’t look only to A1c to determine success of a therapy.

  • Dr. Adler reminded listeners that every healthcare system in the world has to set priorities and make choices; the reality is that the UK’s National Health Service (NHS) cannot afford everything that is effective, and choosing to support a new diabetes drug often means letting go of another drug or program (and not necessarily one in diabetes either). Thus, the challenge in evaluating long-acting insulin analogs (or any new therapy) is coming up with a method of comparison that can apply to all disease areas. To do this, NICE looks at quality-adjusted life years (QALYs), which factor in both length of life and quality of life.
  • Based on the results of a quality-adjusted life year (QALY) analysis by NICE, Dr. Adler supported the recommendation to start most, but not all, people with type 2 diabetes on NPH following metformin and sulfonylureas. (She noted that most trials in diabetes don’t measure QALYs or even length of life; length of life and quality of life were based off of relationships to diabetic complications and side effects.) Results from NICE’s study showed that insulin glargine, insulin detemir, and NPH were comparable in terms of glycemic control, with the long-acting basal insulin analogs having “modest advantages” in terms of hypoglycemia versus NPH. Additionally, insulin detemir showed a statistically significant 1.2 kg (2.6 lb) weight reduction versus NPH. Dr. Adler argued this was not clinically significant – “we’re paying twice the price for a kilogram of weight.” We were surprised by this remark, particularly given that weight gain associated with human insulin isn’t limited to weight gain for the first year only. Weight gained on multiple years on NPH may increase macrovascular complications, of course. Dr. Adler also asked how much the “small difference” in hypoglycemia cost. Cost per QALY gained for insulin glargine (versus NPH) came to 320,029 ($497,613) versus 417,625 ($649,365) for insulin detemir (versus NPH). Dr. Adler explained that a cost per QALY greater than 30,000 ($46,698) was often not approved by the NHS because of the likelihood that the NHS could find greater QALY gains for a lower cost with different therapy.
  • Dr. Adler lamented that there is still high spending on insulin analogs in the UK, despite NICE’s recommendations. Dr. Adler presented data from a BBC News health report that showed from 2000 to 2009, annual NHS spending on what she characterized as “designer insulin” increased from 18 million ($28 million) to 305 million ($475 million) with no corresponding improvement in A1c over this time period. She did not mention any benefits related to patients having more stable insulin overall, less hypoglycemia, or greater time in zone, the latter presumably which has not been measured. Moreover, had just half of those taking analog insulin been put on human insulin instead, the NHS would have saved 250 million ($389 million) – we weren’t sure how she arrived at that figure given that according to Dr. Hirsch, since NPH is about 60% the price of analogs, presumably still close to $300 million would have been spent on insulin, regardless of human versus analog. Dr. Adler also criticized industry for aggressive, and sometimes questionable marketing (referencing Novo Nordisk’s advertisement showing overweight people in bathing suits with the tagline “Once-daily Levemir is changing figures”) as one reason for this trend. That said, she did not shift the responsibility away from diabetes healthcare providers. She called for healthcare providers to realize what these insulin analogs are offering for their costs, and suggested that using human insulin can free up resources to make greater health gains elsewhere in diabetes or in other health areas. She didn’t mention patient input in her process, so presumably patient preferences wouldn’t play a role.
  • Overall, we were disappointed that Dr. Adler characterized insulin analogs as items of ultra-luxury. In one slide, for example, she suggested that use of insulin analogs was as lavish as having Porsches serve as ambulances – she showed a picture of a Porsche ambulance juxtaposed to a Land Rover ambulance (which we presumed would be NPH in this analogy). While cost is always a necessary factor to consider in medical decisions, we don’t think this slide was as persuasive as it could have been! (Interestingly, the reality is that the cost of a Land Rover is comparable to that of a Porsche.)
  • We do see that cost is becoming more and more a factor in decisions whether to approve use of various therapies and think it’s notable that someone as stridently negative on analogs has such a powerful position at NICE. The Germany agency IQWID looked at the insulin analog choice as well, and concluded that insulin analogs have not shown superiority, hence, no higher price could be justified. Manufacturers will have to work harder or at least differently to demonstrate cost effectiveness. Said Dr. Adler: “NHS has a limited amount of money and cannot spend money twice … just like every other healthcare system in the word, public or private, NHS has to set priorities and make choices.” Dr. Adler stressed that the NHS budget is fixed, asking that if NHS paid for various new drugs, what else would be cut? “Will we not be able to pay for education? For diabetes nurses?” She noted that new diabetes drugs had to be compared to cancer drugs and to Alzheimer drug, etc. “NICE has to use measure of effectiveness applicable to all disease areas.”
  • Dr. Adler did stress that doctors and nurses do not always follow NICE guidelines, which she seemed to find frustrating given that there had been no corresponding improvement in A1c from 2000-2009. We would also question whether absent using the insulin analogs, a worsening in average A1cs may have been seen, particularly given the rise of obesity in the UK. Dr. Adler mentioned that devices for the analogs were more attractive, and that primary careHCPs often used with these easier devices. She urged HCPs to try human insulin first. As we understand it, NICE recommends, but does not require (at this point) people with type 2 diabetes to start on NPH.


John Pickup, MD (King’s College London School of Medicine, Guy’s Hospital, London, UK)

Dr. Pickup presented his take on the value of the use of insulin pumps, CGM, and sensor-augmented pump therapy in diabetes, concluding that devices can help most patients (especially those with the worst control on MDI and SMBG) achieve desired clinical and psychosocial goals, but that there is still room for more sophisticated technology (e.g., closed loop systems) to address the needs of those that still do not achieve adequate glycemic control. Based on findings, largely from meta-analyses, he concluded that pump and CGM use is almost always an improvement over MDI/SMBG and that low-glucose suspend pump systems do a promising job of reducing severe hypoglycemia. He notes, though, that the incremental improvement of outcomes achieved with more sophisticated technology is getting smaller and that more advanced technology will bring more modest incremental improvements than we have seen in the past. In all cases, barriers to effective use exist that will inevitably decrease effectiveness of device therapies and he looked forward to greater product improvements emerging.

  • Dr. Pickup began his talk with a short discussion on what he sees as the goals of diabetes management. There are the clinical goals of achieving target A1c, minimizing hypoglycemia and glycemic variability, and preventing complications. Additionally, he stressed the importance of psychosocial goals of improving quality of life, improving treatment satisfaction, and reducing treatment anxiety. These were the endpoints he referenced in his subsequent analysis of the effectiveness of pumps, CGM, and sensor-augmented pump therapy. He mentioned that according to the ADA, most patients should be at an A1c lower than 7%, although mentioned that in terms of survival as a function of A1c, the lowest mortality in diabetes occurs at about 7.6%.
  • Another point he made was that mean values obtained from meta-analyses are often misleading because pooling together data makes it impossible to test for the effects of covariates on outcomes (e.g., how baseline A1c affects change in A1c between treatments). Therefore, in his meta-analyses, he examined individual patient data.
  • In his discussion on insulin pump therapy, he relayed that people with the poorest control on MDI would also continue having the poorest results on pumps, but no matter the starting A1c, switching to a pump almost always reduced A1c. On average, A1c could be reduced by 0.62%, and severe hypoglycemia by 75%, when switching from MDI to pump therapy, but this was a case in which looking at individual patient data provided a more nuanced insight. Those that started with the highest A1c achieved the greatest reduction after switching to pump therapy, and those who suffered from the highest rates of hypoglycemia on MDI could achieve a 10-20 fold reduction in hypoglycemia after switching to pump therapy. We wonder about what the differences in “total time in target zone” would be shown to be – historically, sensor accuracy has not been as strong in hypoglycemia but as accuracy levels improve, it would be instructive to see this analysis.
  • Dr. Pickup then discussed the effectiveness of CGM vs. SMBG based on individual patient data analysis from his meta-analysis of six randomized control trials (RCTs) for patients with type 1 diabetes on CGM or SMBG. As the JDRF CGM trial showed, A1c reductions were greatest in those who wore the CGM five days a week or more (additionalreduction of 0.15% for each 1 day/week increase in sensor usage) and for those with the highest baseline A1c. He also stated that using CGM for seven days had a larger effect, and that for an average person with a 7.0% A1c, seven-day use of CGM would reduce A1c by about 0.9% - but said the effect drops dramatically if patients use CGM only five days a week. He stated that there was good RCT evidence that CGM use reduced mild to moderate hypoglycemia but there was insufficient data for severe hypoglycemia. Overall, we think this bodes very well for manufacturers, since the “hassle factor” associated with various systems is declining, and 24/7 use is increasing in the most recently approved systems, which makes sense since they are easier to use. We also think that in a couple of generations, 24/7 use will be far more common due to better reimbursement and significant product improvements still to come.
  • Dr. Pickup characterized a range of frustrations for CGM users, including lag time, inaccuracy, discomfort in wearing, pain associated with insertion, technology hassles, device malfunctions, skin irritation, body imagine issues, complex training required, costs. We believe virtually all will be improving with future generations of products.
  • He also discussed positive predictors of improved control with CGM, noting that the groups of patients who did best on CGM had good coping skills, had good pattern recognition, and strong support in their personal lives.
  • Dr. Pickup expressed optimism about low glucose suspend (LGS), the Medtronic sensor augmented pump therapy in which the pump is automatically shut off if a low glucose threshold is reached. He characterized this as the first step to a closed loop system and noted that in a recent trial of 31 type 1 diabetes patients in the UK, nocturnal hypoglycemia was reduced by 96% with the use of an LGS insulin pump.
  • To conclude, Dr. Pickup said that what he sees as emerging from all this is that there is an appropriate pathway for the use of pumps and CGM. He made a strong call for new technology that is needed, namely closed loop insulin delivery systems.


Lutz Heinemann, PhD (Science & Co., Düsseldorf, Germany)

Dr. Heinemann addressed two topics during his presentation – the need for better insulins, and the potential impact of biosimilars on the market. He called for better bolus and basal insulin, pointing to a number of developments in ultrafast-acting insulin and ultra long-acting insulin that could improve insulin’s pharmacokinetic profiles. Dr. Heinemann also commented on the arrival of biosimilars (they are coming, he said) with many patents soon to expire. He emphasized the difference between biosimilars and generics, with the former being process-dependent. In Dr. Heinemann’s description of the regulatory process in the EU and US, we found it interesting that insulin in the US will not be considered a biosimilar until April 2020 (currently it is a generic) – we hadn’t realized this. He emphasized that there are quality, efficacy, and safety concerns for biosimilars, with immunogenicity being the most important safety issue. However, he was hopeful that the appropriate clinical trials and pharmacovigilance would help mediate the safety risks. In conclusion, he stated, “The insulin world will look different ten years from now.”

  • Dr. Heinemann noted that although the pharmacokinetic profiles of rapid-acting insulin is better than regular insulin (peaking after two versus three hours), their profiles are still not ideal. While achieving physiological insulin secretion profiles have remain an elusive goal, Dr. Heinemann highlighted developments in ultrafast acting prandialinsulin that could bring us closer to the physiological insulin profile including: 1) Becton Dickinson’s intradermal application; 2) Biodel’s monomeric insulin; 3) Halozyme’s added enzyme to reduce barriers of absorption in subcutaneous tissue; and 4) local heating of the infusion site. He noted that part of the reason we have not seen quicker development of ultrafast-acting insulin in recent history has been that patentability was more difficult in this research area.
  • In addition to better bolus insulin, Dr. Heinemann called for better basal insulin and pointed to Novo Nordisk’s insulin degludec as one development on this front, offering a flatter profile with less intra-individual variability over time compared to insulin glargine (Sanofi’s Lantus).
  • With a number of patent expirations looming for insulin analogs, Dr. Heinemann emphasized that biosimilars are not generics. Whereas generics are chemically synthesized and can result in identical products, small differences in the manufacturing process of biosimilars can result in different products – the question being how large these differences are. In insulin manufacturing, “the process is the product” he said.
  • He commented on the importance for regulatory bodies to recognize the distinction between biosimilars and generics, and noted that clear guidelines for market approvals in the EU and US that recognize the risks of biosimilars exist. In Dr. Heinemann’s description of the regulatory process in the EU and US, we found it interesting that insulin in the US will not be considered a biosimilar until April 2020 (currently it is a generic) – we hadn’t realized this. He emphasized that there are quality, efficacy, and safety concerns for biosimilars, with immunogenicity being the most important safety issue. However, he thought the appropriate clinical trials and pharmacovigilance as part of the approval process could help guard against immunogenicity.


Richard Insel, MD (JDRF, New York, NY); Irl Hirsch, MD (University of Washington, Seattle, WA); Amanda Adler, MD, PhD (Adenbrooke’s Hospital, Cambridge, UK); John Pickup, MD (King’s College London School of Medicine, Guy’s Hospital, London, UK); Lutz Heinemann, PhD (Science & Co., Düsseldorf, Germany)

Q (Dr. Shashank Joshi): I would like to make a comment on biosimilars and changing the way we look at outsourcing and insourcing. The truth is biosimilars from India have not gotten market share in India or China. The biggest challenge is the cost of these copies is only 10 % less than the innovators. When innovators sell insulin in India, China, Germany, the UK, the prices are very different. What will probably happen is insulin manufacturers will have more price pressure on them. You must know currently biosimilars in India have less than 10% of the market share. There have been withdrawals and accidents – there needs to be a lot of pharmacovigilance.

Dr. Heinemann: We are very much on the same track. The term quality is a tricky one, long term quality with insulin manufacturing might look simple but it’s tricky at the end. Price considerations are complex as well. Insulin prices are different overtime and in different countries, but I am not an expert in this. There are many aspects to consider, but the aim of my talk was that there will be numerous changes in this world in the new future.

Q (Dr. Ratner): I’m a strong advocate of cost effectiveness research, but keep in mind that economic analysis always starts with an assumption, and it’s the assumption that needs to be examined very carefully. And as Bernard Shaw said, you can lay 100 economists end to end and not come to a conclusion. The issue of valuing hypoglycemia in cost effectiveness research is highly problematic. People who actually have hypoglycemia can tell you precisely how much it impacts their lives. People who have never had it can’t do so. When you look at the impact on productivity, school attendance, and the ability to even get a good night’s sleep, I think the cost effectiveness analysis of the analogs suddenly begins to take a very different meaning. Additionally, you cannot say you are unable to show a difference in A1c [between human insulin and analogs] because the studies are designed specifically to show equivalence.

Dr. Adler: I agree with you, and that’s what we need to do – we need to quantify that effect on quality of life. It’s reasonable to ask that question about how to value it. Because this was such a problem in what I showed you, we’re doing a study with the London School of Economics to see what effect that does have. I’ll explain that tomorrow.

Dr. Hirsch: I mentioned this in my talk – I was showing the severe hypoglycemia data, and part of the reason is, up until recently with the advent of CGM, we have not been able to do a good job of capturing mild and moderate hypoglycemia, which is applicable to Dr. Ratner’s comment. I’m hoping that the health care economists of the future will try to capture that and I hope that with CGM we’ll be able to do that.

Q: We are interested in how you treat type 2 diabetes in Europe. What is the standard in Europe in using MDI vs. pumps?

Dr. Pickup: It varies tremendously from one country to another. I would say MDI is pretty much the standard throughout Europe, but the uptake of pump therapy is enormously variable, from 20-30% in some countries to less than 5% in the UK. There is debate going on as to why – there are factors like cost and physician acceptance of diabetes technology to be considered. It is an evolving situation.

Q: A question for Dr. Adler – we’ve worked particularly hard in our preschool and children’s practice to get them off NPH at night since they don’t need a peak insulin at night when they’re not eating. I wonder if you have information on basal insulins on preschool children in your country?

Dr. Adler: The work I showed you was for adults and type 2 diabetes, and I’m not a pediatrician, but I do know that with respect to NICE guidelines, both short and long acting analogs are recommended for children with type 1 diabetes.

Q: In terms of the cost of analogs, the absence of severe hypoglycemia does not then indicate reduction in glucose variability. So I was thinking CGM might be a better indicator for cost-benefit ratios.

Dr. Hirsch: Come to my session this afternoon. The big issue about variability is that we can’t even agree how to measure it. I think we have to first have consensus about definitions and how to measure it. I agree with you fundamentally, but we also need to have data. As an example, the FDA does not accept glucose variability. And I hope by the time I’m done with my career they do.

Q: I want to echo Bob Ratner’s comment that the iQWIG analysis has been published multiple times and most of the trials in these types of analysis were designed to show non- inferiority. They were not designed to show superiority. So surprise, surprise when you don’t show improvements with analog insulins, and then you conclude they’re not cost effective. I hope you would look at trials intended to show superiority and make analysis based on those.

Dr. Adler: My understanding of non-inferiority trials is that they require more patients than superiority trials. If there were a difference in the insulins, I would have thought they would have come out. And then we turn to the clinic and the fact is that 80% are using analog insulins with no difference in A1c.

Dr. Hirsch: These meta-analyses are done and based on registration studies, I understand that, so there are flaws there. There are a lot of trials on that are not published if the company doesn’t like the result. We are seeing the best of the best data for these meta-analyses. So if you put all the data in that are not published, it might be worse that what we are seeing now.

Dr. Heinemann: We tried to convince pharma 10 years ago to perform adequately designed studies. At this point in time they laughed at us because they were not forced to do such trials…it would have been a good approach to this question but no company was willing to perform such a study.

Q: There was such big discrepancy between the presentation of the promise of sensors, pumps, and insulins and the fact that we cannot afford the technology we already have. What do you think will happen if we have to go back to regular insulin? Do we put that in pumps? Or switch everyone back to MDI?

Dr. Hirsch: We’re already there. I have patients using regular insulin in their pumps. The question is, is it going to get much worse? In the short term I think it is. I hope that in long term it won’t. It’s a huge concern I have, especially with the artificial pancreas project. It was a concern I had before the JDRF sensor trial – even if we showed a positive result, which we did, would we have problems with reimbursement? And we are. A lot of this could not have been predicted.

I think we’re okay with pump therapy because even many of the Medicaids are now paying for pump. It’s considered part of type 1. The advantage we have with pumps is that we’ve had them for 30 years. I’m now concerned about analogs for type1. I’m worried about CGMs because they’re new. This all gets back to the point that the people who are looking at this data and making these decisions look at the data like data for blood pressure medications, which are pills and not as complex as all of the behavioral problems involved in diabetes management.

Q: I’ve seen a number of patients getting insulin from Walmart.

Dr. Hirsch: When insulin lispro was introduced in the US you needed a prescription. But for NPH and regular insulin you can just go to the drug store and buy it with no prescription. This has huge potential for long term impact with what can happen with pricing. We’ll see what happens with the Affordable Care Act and Health Care reform. My hope among hopes is that getting insulin analogs in the US will be easy. But there is the prescription issue and the cost issue. The goal needs to put the least number of barriers as possible and the biggest barrier right now is strips.

Q (Dr. Skyler): This is a question for Dr. Pickup. In looking at various analyses, the one study that I haven’t seen is CSII + CGM vs. MDI + CGM, because it would seem like CGM is making a difference in outcomes when one is comparing CSII + CGM vs. MDI + SMBG.

Dr. Pickup: That study is being done, and I’m very happy about that. We’re about a year from knowing the definitive answer. From individual patient data, it looks like there is no difference between CSII and CGM concerning effectiveness of CGM. It seems to work just as well for MDI as it does for CSII. I think the trial has to be done, and it is just that the information is not out there at the moment.

Q: When we compare different treatments, it’s not sufficient to ask if there has been a change in A1c and a change in hypoglycemia. We must do these tests simultaneously. We have many situations where there is no difference in A1c in a study by itself and no difference in hypoglycemia in a study by itself.

Dr. Adler: People are looking at that kind of thing, including Dr. Pickup. This kind of thing is even more sophisticated modeling.

Q (Dr. Steve Edelman): With type 2 diabetes, this discussion is a completely different story. Certainly NPH at night works fine. But in type 1, I can tell you, I consult with the executives of the big organizations that implement these restrictions, and they’re on pumps, they’re on CGM, they’re on analog insulins.

Q: I think all this conversation about analogs that is pertinent to type 1 patients, where analyses are different in type 1 and 2 patients, is a perfect example of why type 1 physicians must be at the table when these issues are discussed. Cost effectiveness must be examined, but type 1 physicians need to be at the discussion.

New Therapeutic Options for Diabetes Treatment


Robert Ratner, MD (Georgetown University School of Medicine, Washington, DC)

Dr. Ratner provided an exciting overview of five diabetes drug classes in development. He began his presentation with a nod to those who think that we have all the therapies we need, but just need to use them better saying he “respectfully disagrees.” He explained that the goals of therapy should be to go beyond glycemic control, emphasizing that we need to find therapies that patients accept in order to maximize adherence. We certainly have seen more of this in the last five to seven years and believe future classes as well as generations will bring more. We wonder if today, new therapies that cause weight gain or hypoglycemia or are associated with fractures could even be approved. Then he delved into five new classes of drugs: SGLT-2 inhibitors, 11-beta HSD-1 inhibitors, DGAT inhibitors, G-protein coupled receptor (GPR) agonists, and glucokinase activators (GKAs).

  • SGLT-2 inhibitors: Dr. Ratner decided to begin with SGLT-2 inhibitors, as this class is furthest along in development with dapagliflozin’s prior submission to the FDA (currently put on hold pending additional information) and canagliflozin’s recent submission to the FDA. He noted many advantages of this class including: 1) rapid drops in A1c (whereas many other drugs require dose titration or have a delayed effect); 2) weight loss through obligatory caloric loss with glycosuria (typically weight loss occurs within six weeks); 3) low hypoglycemia; 4) lowered systolic blood pressure (though he noted that whether this is due to volume depletion, an indirect effect of weight loss, or an independent effect of the drug is still unknown); and 5) an insulin- independent mechanism. Given the latter, Dr. Ratner said theoretically, this class of agent could be used in anyone with diabetes, even type 1 diabetes. Interestingly, he noted that although he was taught that polyuria was negative, i.e., a symptom of diabetes, it is now being viewed as a form of therapy. He also addressed some potential negative effects due to polyuria with SGLT-2 therapy, including electrolyte disturbance, urinary tract infections, and fungal genital infections. He also spoke to the increase in bladder cancer risk (though not a statistically significant increase) observed with dapagliflozin, but noted that there was no relationship between overall malignancy and use of dapagliflozin (including a potential protective effect in terms of renal tract malignancies). Dr. Ratner explained that while in most cases all malignancies are examined together, unfortunately, that was not what the FDA did. Given the biological plausibility for bladder cancer, the FDA called for a large-scale cancer study of 16,000 patients over three to five years to show no effect (editor’s note – we believe we understood this correctly though we had not heard this announced before). He concluded his discussion on SGLT-2 inhibitors with his owntake, emphasizing that he was speaking for himself and not ADA: if the cancer profile looks the same for canagliflozin [as dapagliflozin], he doesn’t think this class of agents has a future. This did not appear to be the case in the phase 3 data seen at ADA. He than said that if canagliflozin looks clean (editor’s note – it appeared to at ADA), he’s not sure if dapagliflozin has a future.
  • 11-β HSD-1 inhibitors: Dr. Ratner reminded us that overactivity of the 11-β HSD-1 enzyme causes visceral adiposity, which can result in many of the manifestations of diabetes. He pointed to Incyte’s INCB13739 as a small molecule drug designed to block this particular enzyme. However, Dr. Ratner said in phase 2 trials looking at the drug in people with type 2 diabetes, the drug appeared to work centrally (i.e. on the hypothalamus) with compensatory increase in ACTH, something you wouldn’t want in women. He noted that Incyte’s 11β -HSD has since been pulled and he can find no mention of it in 11-β HSD-1 drugs in development: “As far as I can tell, it’s dead.”
  • DGAT-2 inhibitors: Dr. Ratner provided a brief reminder of the mechanism of action behind DGATs –when diacylglycerol (DAG) is broken down, free fatty acids (FFAs) are released. FFA’s in turn have been implicated in insulin resistance. DGAT inhibitors intervene in this process by blocking the synthesis of DAGs. While studies so far on DGAT-2 inhibitors have been in murine models, with no clinical trials currently listed on, Dr. Ratner said he thought there was great potential here. However, he noted that there could be an issue with DGAT-2 inhibition and gastrointestinal side effects – blocking DGAT-1 has shown massive gastrointestinal side effects – but called for more data to determine what will happen with this therapy.
  • GPR agonists: Dr. Ratner noted that many GPR agonists are being investigated (e.g., GPR-41, GPR-43 and GPR-120) but that by far the greatest amount of data was available on Takeda’s TAK- 875, a GPR-40 agonist. He presented some highlights from phase 2 data on TAK-875 showing a quick fall in A1c that appeared to plateau at 12 weeks. Dr. Ratner also noted that GPR-40 agonists appear to be weight neutral, maybe with a little weight gain, but that we are seeing data from only 12 weeks and need more information. The most reassuring aspect of GPR-40 agonists, he said, is that we were not seeing hypoglycemia.
  • Glucokinase activators: Dr. Ratner drew particular attention to the risk of sustained hypoglycemia with this therapy (“we’re talking hours, not minutes …”). Because glucokinase activators don’t just work on the muscle and liver – importantly they affect the beta cell too – beta cells may sustain insulin secretion even in the case of peripheral hypoglycemia. Dr. Ratner noted that all glucokinase activators have now been taken off development plans, and while he does know of a couple companies trying to develop liver specific glucokinase activators, he thinks this will be very difficult to achieve.


Jay Skyler, MD (University of Miami Miller School of Medicine, Miami, FL)

Dr. Jay Skyler detailed the various efforts under investigation to prevent the onset and slow the progression of type 1 diabetes. He identified his three ideal therapeutic goals for type 1 diabetes: 1) stop immune destruction; 2) preserve beta cell mass; and 3) regenerate or replace beta cells. Unfortunately, a large number of the studies Dr. Skyler reviewed in both the realm of prevention and intervention had disappointing results. However, oral insulin for prevention, and anti-CD3, anti-CD20 rituximab, and abatacept for new onset intervention were bright spots. Encouragingly, many new studies are planned, and new therapeutic targets are being investigated. He concluded his presentation by emphasizing that perhaps we need to approach type 1 diabetes therapy from a different angle – as opposed to evaluating therapeutic agents in isolation, the solution, he suggested, will require a combination-therapy approach.

  • Prevention studies: Dr. Skyler reviewed results from the ENDIT nicotinamide, DIPP nasal insulin, NIP docosahexaenoic acid, DPT-1 parenteral insulin, and DPT-1 oral insulin trials, all of which he said showed no preventative effect. However, a post-hoc analysis of the DPT-1 oral insulin study found that in patients with insulin antibody (IAA) > 80 nU/ml, oral insulin treatment conferred a 4.5-5 year delay in diabetes onset. This delay was further magnified in individuals with IAA > 300 nU/ml. The next step, said Dr. Skyler, is to repeat this study in a prospective way in order to show that the effect of oral insulin was real and robust in these subgroups. Dr. Skyler also expressed hope that taking cow’s milk out of formula would prevent type 1 diabetes onset based on delayed first antibody development in those who drank formula with casein hydrosylate removed vs. those who drank formula with the protein (Knip et al., NEJM 2010).
  • Intervention studies: Dr. Skyler saw GAD-65, anakindra, and canakinumab has having generally disappointing results in slowing diabetes progression. As for DiaPep-277 (heat shock protein peptide 60), Dr. Skyler commented that while a phase 3 study presented at ADA allegedly showed positive effects on C-peptide levels in response to glucagon, the effect disappeared in response to a mixed meal test – this disparity brought the effectiveness of DiaPep-277 for slowing type 1 diabetes into question. Dr. Skyler had a much more positive take on the anti-CD3 otelixizumab and the anti-CD3 teplizumab (which we were surprised to hear given the disappointing results of the Protégé study. For more details see our October 22, 2010 Closer Look at Dr. Skyler highlighted a study in which otelixizumab was given to newly diagnosed patients that found four years later, individuals in the treatment group had a significantly lower insulin dose compared to placebo (Keymeulen et al., NEJM 2005; Keymuelen et al, Diabetologia 2010). He noted that while DEFEND-1 (Gottlieb et al., ADA 2011) showed otelixizumab to have no effect on C-peptide levels, the dose in this study was 1/16 of that used by Keymeulen et al. Dr. Skyler also suggested that abatacept and etanercept seemed to have had positive study results; the latter of which hasn’t been pursued in a full scale trial (but maybe should be, he said).
  • Ongoing and new studies: Encouragingly, Dr. Skyler presented a hearty list of studies assessing new therapeutic compounds (or looking at previously studied therapies in new ways) including proinsulin peptides, the immunosuppressant alefacept, the anti-IL-6 tocilizumab, the combination of thymoglobulin (low dose) + G-CSF, and the combination sitagliptin + lansoprazole, to name a few. Dr. Skyler also highlighted one interesting publication (Couri et al., JAMA 2009) that presented an extreme, but seemingly remarkable therapeutic approach we would have to “scratch our heads about.” A high dose of cyclophosphamide and ATG was used to wipe out a patient’s immune system. Then, the patient was “rescued from death” with AHBMT (autologous human bone marrow transplant). The probability of remaining insulin free following this procedure was 80-100%.
  • A combination approach to type 1 diabetes: Dr. Skyler outlined a potential combination therapy approach beginning with anti-IL-1β (even though anti-IL-1β did not work in mono- therapy studies, he posited it would have a better effect in combination) or anti-TNF in combination with anti-CD3, anti-CD20, or co-stimulation blockade. He suggested GAD, oral insulin, or both should come next, followed by G-CSF (or T-regs themselves). Finally, exenatide, GLP-1, or HIP2b could be added. As to whether this approach would work (or whether regulators would even approve the kind of tactic he outlined), are still questions, he said; nonetheless, hebelieves this type of combination approach is how we need to be thinking about treating the disease.


Irl Hirsch, MD (University of Washington, Seattle, WA)

Dr. Irl Hirsch opened his presentation suggesting that severe hypoglycemia can lead to microvascular complications. He explained that, like hyperglycemic spikes, severe hypoglycemia causes increased inflammatory action and (likely) increased oxidative stress, both of which have been linked to microvascular complications. Furthermore, he questioned the conclusion of the DCCT investigators that A1c explained “virtually all” the difference in risk of complications between the intensive and conventional treatment groups. Dr. Hirsch hypothesized instead that other factors, likely glycemic variability, must have played a role. The impact of glycemic variability on “hard endpoints” needs to be assessed, he said, and his upcoming FLAT SUGAR study (FLuctuAtion reducTion with insulin and GLP-1 Added togetheR) will begin to do just that.

  • Dr. Hirsch suggested that severe hypoglycemia could contribute to microvascular complications. Hyperglycemic spikes result in inflammatory action and increased oxidative stress (both have been linked to microvascular complications). Dr. Hirsch posited that severe hypoglycemia acts through a similar pathway (i.e., inducing an inflammatory response and [potentially] oxidative stress, though he noted the latter was a very new area of investigation). In Dr. Hirsch’s own practice he has seen many cases of retinal bleeding following severe hypoglycemia – anecdotal support for a link between severe hypoglycemia and microvascular complication.
  • Dr. Hirsch painted a complex picture as to what A1c, our fundamental biomarker, actually tells us. He suggested, to the surprise of the audience, that glycemic variability could influence A1c. Dr. Hirsch presented post-hoc analysis from the A1c Derived Average Glucose (ADAG) study, indicating that standard deviation, a measurement of glycemic variability, did impact A1c in people with type 1 diabetes (though it had no impact in people with type 2 diabetes). Of course Dr. Hirsch did not suggest we use A1c to measure glycemic variability, however, the question remains as to what measurement we should use. Standard deviation, coefficient-of-variation, and percent in-range are three of many measurements that have been proposed.
  • In interpreting the DCCT, Dr. Hirsch suggested that investigators gave too much weight to A1c. To set the stage for his discussion, he reminded the audience of the DCCT investigators’ conclusion: “A1c explains virtually all of the difference in risk of complications between the intensive and conventional groups… While other components of hyperglycemia, such as glucose variation, may contribute to the risk of complications, such factors can only explain a small part of the differences in risk between intensive and conventional therapy over time.” In Dr. Hirsch’s statistical analysis, he showed that glycemic exposure (A1c + the duration of diabetes) explained only 11% of the variation of diabetic retinopathy risk in the DCCT, suggesting that other factors (presumably, glycemic variability would be a leading contributor) explain 89% of the variation of diabetic retinopathy risk among individuals independent of A1c. “How could we get this so wrong?” he asked.
  • The next step, said Dr. Hirsch, is to study the impact of glycemic variability on “hard endpoints.” He highlighted the FLAT SUGAR study (FLuctuAtion reducTion with insulin and GLP-1 Added togetheR), set to begin in the next few weeks, as the start to this investigativeprocess. As a reminder, patients with type two diabetes, already receiving insulin, will be transitioned to basal-bolus insulin therapy + metformin during a run-in period, followed by randomization to 26 weeks of either basal-bolus insulin therapy or exenatide + basal insulin. The primary endpoint is differences in coefficient-of-variation at 26 weeks (determined by masked CGM) and the major secondary endpoints are inflammatory activation, oxidative stress, weight, and hypoglycemia.


Robert Ratner, MD (Chief Scientific & Medical Officer, ADA); Jay Skyler, MD (University of Miami Miller School of Medicine, Miami, FL); Irl Hirsch, MD (University of Washington, Seattle, WA).

Q: This is for Dr. Hirsch. I was wondering if you see any data to support that large changes in glycemia is something we should try to eliminate?

Dr. Hirsch: The real answer is, I don’t think we know, but we do know that if we take someone with an A1c of 9.0% or 10.0% and bring them down quickly, we get a counter regulatory response. This is an entire area ripe for study. I suspect it’s not good, but we really don’t know.

Dr. Ratner: When you take someone from very high blood glucose to normal blood sugar levels you get an explosion of retinopathy. That’s been known now for 35 years.

Q: For Dr. Hirsch, your new study (FLAT-SUGAR) looks at people with type 2 diabetes and minimizing glycemic variability. Are you going to do a study in people with type 1 diabetes and how would you do that? With GLP-1s?

Dr. Hirsch: Knowing what I know about GLP-1, I think you could do the study, but it would be off label. What we are seeing anecdotally is very exciting with GLP-1 as far as variability is concerned.

Comment: Just about every other measure of variability is highly correlated with standard deviation. But both for standard deviation and all the other measures, you have to correct for the average A1c or mean plasma glucose because when looking at change in A1c, we have to correct for baseline A1c. The easiest way is to correct with percent coefficient-of- variation. Percent CV is nearly independent of the glucose. Using the rules you’ve developed empirically, we can say that anyone with percent CV with less than 33% is in best quartile and greater than 42% is in the worst quartile.

Dr. Hirsch: What I am hoping is we will have standardizations for research and for clinical practice. We desperately need standardization in many areas of diabetes. I had a patient by total coincidence who was one of our EDIC patients who had to wear a holter monitor and also wears a pump. She said to me, “This is the best form of birth control I’ve had.” It’s a lot of equipment to wear.

Q: One of my children has not allowed his daughter to have any cow milk. Is there any data to suggest how long they should abstain?

Dr. Skyler: The short answer is we don’t know for sure and don’t even know if the removal of cow’s milk will be beneficial. We do know longer breastfeeding seemed to be beneficial and early exposure to cows milk seemed to be detrimental, and these are of course reciprocal. But there is no way to design this type of study ethically. The critical period is the first year of life in this study; the study is only focusing on that. As to whether it makes a difference beyond that, there is no way of knowing.

Q: I had a question about how cardiac arrhythmias are associated with glycemic variability.

Dr. Hirsch: In young, healthy men, with type 1 diabetes where we see prolonged QT interval it was the duration of time that they were hypoglycemic that made the difference. Now with CGM, we are seeing a lot of this. It’s the duration of hypoglycemia, not the actual number that seems to be the issue.

Dr. Hirsch: This is for the Chief Scientific & Medical Officer of ADA – is there a role for retrospective GGM?

Dr. Ratner: No. But I do think there is a role for real-time CGM.

Q: I have a question about glycemic variability. We are seeing AUC in pump reports. Is this something we need to pay attention to in deciphering glycemic variability?

Dr. Hirsch: The short answer is no, because I still struggle to figure out AUC with reports I get right now as it is not intuitive. Maybe this should go back to Dr. Bergenstal’s group as we try to standardize this. We need to, at the very least, correlate these measurements with something important – inflammatory markers for example. AUC is taking it one step too far for me.

Hypoglycemia and Diabetes


Peter Chase, MD (University of Colorado School of Medicine, Denver, CO)

Dr. Chase’s presentation focused on the currently available methods to reduce hypoglycemia in type 1 diabetes and what we can expect to see in the future. He reviewed the definition of hypoglycemia – a blood glucose level below 70 mg/dl – but suggested perhaps a blood glucose level below 60 mg/dl would be a more appropriate definition given that even individuals without diabetes spend 20 minutes per day in the 60-69 mg/dl range. Dr. Chase reminded us that hypoglycemia is a serious problem, calling it the rate-limiting factor in attaining optimal glycemic control and the number one fear in families with children with type 1 diabetes. He emphasized hypoglycemia unawareness as a major culprit of hypoglycemia. As a solution, he spoke optimistically about the potential of CGM, low glucose suspend systems (LGS), and predictive low glucose suspend systems (PGLS) in preventing hypoglycemia. Dr. Chase also gave a different perspective on insulin analogs from what we heard the day before, emphasizing their necessary role, especially in pediatric patients. (Editor’s note – thank God.) Dr. Chase concluded his presentation with three assertions: (1) we have made progress; (2) there is a need for continued progress; and (3) the balance between safety and moving ahead must remain a high priority.

  • Dr. Chase outlined the causes of hypoglycemia including: (1) increased glucose utilization and insulin secretion with exercise; (2) increased insulin levels during exercise; (3) hypoglycemia begetting hypoglycemia (see page 83 of our ADA 2012 full report at; (4) impaired counterregulatory hormone response in people with type 1 diabetes (especially during sleep and particularly epinephrine). On the latter point, he noted the decreased epinephrine response and hypoglycemia unawareness applies to pediatric groups too. Last was (5) insulin inhibition of hepatic glucose output. Dr. Chase also stressed the need for standard definitions for severe hypoglycemia.
  • Dr. Chase stressed that severe hypoglycemia is the clear rate limiting factor in attaining optimal glycemic control. If it weren’t for this, he stressed, we could push up insulin doses and obtain great A1cs with little variability. He highlighted the encouraging results from the STAR-3 sensor-augmented pump trial, which showed improved A1cs with no increase in hypoglycemia (Bergenstal et al. NEJM 2010).
  • Dr. Chase detailed the tools we have to prevent hypoglycemia now:
    • Reduce the intensity of glycemic control (which, presumably, patients and providers would not want to do given landmark DCCT results).
    • Use of insulin analogs (a strong recommendation from Dr. Chase). Dr. Chase was adamant that there was a role for insulin analogs in pediatric patients (perhaps a nod to Friday’s debate on insulin analogs, which questioned whether they are really worth their cost), stating that anyone who argues there is no place for insulin analogs in very young children needed to come chat with him.
    • Improved management of exercise (e.g., stopping insulin right before exercise).
    • Using an insulin pump. Dr. Chase referred the audience to Dr. John Pickup’s (King’s College London School of Medicine, London, UK) presentation on Friday for more details, where he widely praised the increased use of diabetes technology (see page 7 of our report at
    • Using a CGM. Dr. Chase noted that if you are just measuring number of episodes, you may find hypoglycemia increases with initial CGM use because you are now detecting unrecognized hypoglycemia – this is a good thing. He referred to the landmark JDRF CGM trial that showed decreased time spent in hypoglycemia with CGM use. We believe hypoglycemia detection will only improve over time as CGMs get more and more accurate– the accuracy and reliability of CGMs used in that trial were not as good as the current generation, to say nothing of next generation products from Dexcom (G4) and Medtronic (Enlite).
  • Dr. Chase believes low glucose suspend (LGS), predictive low glucose suspend (PLGS), and preserving C-peptide/insulin production may be useful to prevent hypoglycemia in the future. Closest on the US horizon is LGS, which Dr. Chase said is already used in 35 countries and would be in the US soon in the form of the Medtronic MiniMed 530G (see our report covering its submission to FDA at In discussing PLGS, he cited the major advantage of the predictive alarms: patients may actually avoid hypoglycemia in the first place (since the “predictive” alarm alerts them). As a result, they will also not have to “waste” their epinephrine, which ultimately leads to hypoglycemia unawareness. We certainly look forward to seeing more data on predictive systems. Dr. Thomas Danne noted at ATTD that the Medtronic PILGRIM trial (Predictive Low Glucose Management in Realtime Sensing Insulin Pump Therapy) is expected to report results at ATTD 2013 in Paris.


Amanda Adler, MD, PhD (Addenbrooke’s Hospital, Cambridge, UK)

Dr. Adler reviewed risk factors for, consequences of, and potential mitigation strategies for hypoglycemia in patients with type 2 diabetes. She lamented the lack of a consensus definition for hypoglycemia, which makes comparisons between trials difficult and limits our ability to effectively address the issue. Dr. Adler relayed that rates of hypoglycemia in patients with longstanding type 2 diabetes are similar to those of patients with type 1 diabetes, and the associated hypoglycemic unawareness can become just as severe. She highlighted certain simple measures that could help patients avoid easily preventable hypoglycemia, especially in the in-patient setting: some hospitals are careless in mismatching patients’ meal and insulin schedules or even experience drug dispensing errors due to poor handwriting; she suggested that these represented two easily amendable measures. These small actions were the most concrete suggestions offered during the talk, which we found disappointing because these hardly seem to address the root of the problem. In assessing risk factors for hypoglycemia, she said that patients with low A1c values weren’t necessarily more likely to have hypoglycemia than those with higher values. A study in Scotland demonstrated no difference in A1c between patients that experienced hypoglycemic events and those that didn’t (and even for those on insulin who experienced a hypoglycemic event, the average A1c was pretty high – 7.9%). Factors that did raise the risk of hypoglycemia included age, renal impairment (resulting in the slow clearance of diabetes drugs), hepatic impairment (impaired gluconeogenesis), recent hospitalization, multiple drug therapy, alcohol intake, and duration of insulin treatment (>10 years). She advised for less aggressive treatment in those at higher risk for hypoglycemia. Finally, Dr. Adler stated that newer drugs seem to cause less hypoglycemia, but she also added that most new drugs in development do not list hypoglycemia as a primary endpoint. Additionally, trial recruitment bias and differences in the definition of hypoglycemia make it difficult to interpret the effect these drugs have on hypoglycemia. Finally, she urged for better research aimed specifically at improving hypoglycemia.


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

Dr. Bergenstal offered preliminary findings from SMBG data in ACCORD that linked increased mortality risk with greater “divergence from glycemic targets” – i.e., for patients in the intensive control arm, difficulty achieving low blood glucose values; for patients in the standard control arm, difficulty maintaining blood glucose in a moderate range. Additionally, patients with severe hypoglycemia and patients that died both showed greater overall glycemic variability and larger drops in blood glucose levels from bedtime to morning. For additional detail on these analyses, see our coverage of Dr. Bergenstal’s similar presentations at ATTD 2012 ( and ADA 2011 (

  • As a reminder, the ACCORD study randomized 10,251 volunteers with type 2 diabetes at high risk for CVD to either an intensive control group or standard control group. The intensive group (n=5,128) targeted an A1c less than 6.0% and the standard group targeted an A1c between 7.0 and 7.9%. The study was stopped in February 2008 after the data and safety monitoring board found a 22% increase in mortality in the intensive vs. the standard treatment arm.
  • Many hypotheses have been proposed to account for the 22% increased mortality in the intensive treatment arm – e.g., rapid drop in A1c, too-low A1c, weight gain, use of TZDs and other medicines, and severe hypoglycemia – but Dr. Bergenstal said the data have not supported these hypothesis. He noted that while, in general, severe hypoglycemia does increase the risk of dying, severe hypoglycemia did not explain the increased mortality risk in the intensive vs. standard group in ACCORD. Dr. Bergenstal then asked whether it was actually mild and moderate hypoglycemia that could be implicated in the mortality differences. To investigate this hypothesis, Dr. Bergenstal presented preliminary findings from SMBG data collected from 51 of 77 clinical sites in ACCORD, totaling over 9.4 million SMBG data points.
  • Patients with severe hypoglycemia and patients who died had SMBG profiles with greater variability and greater drops in blood glucose level from bedtime to morning. This relationship was seen in both the intensive and standard care arms.
  • In the intensive treatment arm, greater mortality risk was associated with inability to achieve the intensive arm target of 6.0% A1c. Dr. Bergenstal reminded the audience thathealthcare providers in the intensive control arm were instructed to do anything and everything to get their patients’ A1c down. Comparing SMBG trends between intensively managed patients that were alive vs. deceased at the end of the trial, Dr. Bergenstal demonstrated that mortality risk was greater among people with a high percentage of hyperglycemia or a low percentage of hypoglycemia, as measured by SMBG. In other words, the patient population in the intensive control arm who did not achieve the target A1c of 6.0% (despite every effort to get their A1c down) experienced the most death. Mean SMBG in the deceased group (n=74) was 134.8 mg/dl vs. 125.7 mg/dl (n=2617) among survivors.
  • In the standard treatment arm, mortality risk was higher among patients that did not achieve the control arm target of moderate glucose – i.e., those with higher percentages of hypo- and/or hyperglycemia. In contrast to the severe approach taken by providers in the intensive control arm, providers in the standard treatment arm were instructed to avoid highs and lows and to keep patients in a middle range. Correspondingly, patients with the highest mortality risk in the standard treatment group had a high percentage of hyperglycemia and/or a high percentage of hypoglycemia. Patients that generally achieved middle-range blood glucose values had lower risk of dying. (A high percentage of SMBG readings in the middle range corresponded to the intent of the 7.0-7.9% target, even though we note that A1c does not capture “time in target.”) Dr. Bergenstal concluded, based on the trends in both the intensive and standard groups, that mortality was linked with “divergence from glycemic targets.”
  • Dr. Bergenstal concluded with the clinical implications of these preliminary SMBG ACCORD findings. He noted that when setting A1c and SMBG targets, providers need to evaluate glycemic profiles to achieve safe glucose levels. From preliminary SMBG data in the intensive arm, Dr. Bergenstal gleaned that providers should relax intensive therapy for their patients if A1c and SMBG are significantly divergent from targets, despite reasonable effort and intensity on the part of the patient and provider. Furthermore, based on the preliminary SMBG findings from the standard arm, he recommended that mild-to-moderate hypoglycemia should be minimized, particularly when A1c targets are less intense.


David Kendall, MD (Eli Lilly, Indianapolis, IN); H. Peter Chase, MD (University of Colorado School of Medicine, Denver, CO); Amanda Adler, MD, PhD (Adenbrooke’s Hospital, Cambridge, UK); Richard Bergenstal, MD (IDC, Minneapolis, MN); Shashank Joshi, MD (Lilavati and Bhatia Hospital, Mumbai, India)

Q: Were there distinguishing factors associated with patients that experienced this divergence from targets?

Dr. Bergenstal: We know there are factors published, like cognitive problems, that lead to more hypoglycemia so now we’re trying to sort out in this group their cognitive function vs. their profile. So I think we’re still grappling a little bit with this but if you have a frail elderly person with cognitive dysfunction, we should be careful, but not just be careful with their A1c, but monitoring their blood sugar as well. We’ve seen that you can have high A1c and high risk for lows as well.

Dr. Adler: An epidemiologist would say that the kind of person who can’t meet his goals might be fundamentally different in a way that increases his risk of death. On the other hand my boss has decided they may have died of frustration [laughter].

Q: I am a dietician. As far as hypoglycemia goes, we know nutrition plays a big role. I think it was interesting to hear the idea about blood glucose vs. A1c when thinking about the elderly. People say don’t be as aggressive with elderly, but when it comes to nutrition and lifestyle I find the elderly are the most receptive and benefit the most in terms of quality of life. People will say let them live their life, but it’s actually been extremely rewarding to see their lows disappear.

Dr. Bergenstal: In ACCORD, the single largest cause of severe hypoglycemia was someone who did not follow their meal plan. The data we have, though a little soft, is that not following your meal plan – like skipping meals, etc. – was the single highest cause.

Q: I’m in rural southeastern New Mexico with two dialysis centers. This past year the government changed the regulation of labs we could draw; they call it the bundle effect. Now we can only do A1cs once a year. My patients are used to nephrologists or nurse practitioners taking care of them, and when we send them to primary care physicians, it might take 3 months to get an appointment because they are so overloaded. I don’t have a good endocrinologist at all. I have been seeing, since January, more hypoglycemia at the end of dialysis treatment than I’ve ever seen in 14 years I’ve been doing dialysis. And I am at a loss for what to do because it happens with those people who are out of glycemic control and I can’t get them to see a PCP quickly enough to change their drug regimen.

Dr. Bergenstal: Don’t worry too much about A1cs because that’s not very accurate for monitoring hypoglycemia. Put that money instead into more SMBG.

Q: I’d like to suggest some additional forms of analysis. Analyze the nocturnal values in much greater detail. Also, when characterizing glycemic variability you have to distinguish between total variability, between-day variability, within-day variability, and variability between patients. Those are four types of variability to consider.

Dr. Bergenstal: We have some of that data, but we’ll talk about it.

Q: I’m a diabetes educator from Tucson, and I had a question about ACCORD – I noticed I didn’t see any middle-of-the-night-glucose testing in the modal day data you showed. At our office, we always make sure patients take their blood glucose at the middle of their sleep cycle before we change basal insulin dose.

Dr. Bergenstal: We recommend it very often but we don’t get a lot of it. We do have some data we need to look at more carefully, but it’s scattered and the time of testing was not uniform (e.g., 3 AM or 12 AM).

Dr. Kendall: Dr. Joshi, you’re dealing with very disparate populations in terms of means to support diabetes care. Is there any correlation between means and hypoglycemia risk?

Dr. Joshi: Yes and no. It’s all about the education. SMBG, good diabetes education, and good quality time spent with patients can prevent lots of hypoglycemia.

CGM and Closed Loop


Jay Skyler, MD (University of Miami Miller School of Medicine, Miami, FL)

Dr. Skyler gave an extensive overview of the development of CGM technology. He brought the audience all the way back to the 1967 Nature paper when S. J. Updike proposed an enzyme electrode glucose sensor. He then passed through some of the biggest milestones since then, including: the Biostator glucose-controlled insulin infusion system (commercialized in the late 1970s), the commercially unsuccessful Cygnus GlucoWatch Biographer, and the modern-day CGM systems of Medtronic, Dexcom, and Abbott, He also listed many new CGM proposals including non-invasive technologies – infrared spectroscopy, optical coherence tomography, and polarization changes to name just a few – but noted that most glucose novel sensors have ended up in the graveyard. To conclude, Dr. Skyler reminded the audience that Rome wasn’t built in a day, and neither will the next generations of CGM technology. Instead he said, as he showed a picture of Gaudi’s Sagrada Familia (under construction since 1882), CGM is a work in progress.


Fran Kaufman, MD (Chief Medical Officer, Medtronic Diabetes, Los Angeles, CA)

Dr. Kaufman presented results of the ASPIRE Trial, designed to assess the ability of the low glucose suspend (LGS) feature of Medtronic’s Veo sensor-augmented pump to reduce duration and severity of exercise-induced hypoglycemia. Fifty adult type 1 diabetes patients were randomized to an LGS-on or LGS-off group and were driven to hypoglycemia by fasting and exercise and then observed for 3.5-to-4 hours. After a washout period of three-to-10 days, each repeated the experiment under the alternate LGS condition. On average, duration of hypoglycemia was lower with the LGS feature turned on (hypoglycemia lasted 138.5 minutes with LGS on vs. 170 min with LGS turned off, p<0.01). Of note, the researchers observed an order effect: the study participants in the LGS-on group after the crossover had a significantly longer duration of hypoglycemia than those in the LGS-on group prior to the crossover, but in all cases the LGS-on group always spent less time in hypoglycemia than the LGS-off group. Dr. Kaufman interpreted this to signify that “hypoglycemia begets hypoglycemia” and concluded that long- term use of LGS could help further reduce hypoglycemia by disrupting this vicious cycle. For related information on ASPIRE results, see our coverage of Dr. Satish Garg’s presentations at ADA 2012 ( and ATTD 2012 (

  • The LGS feature shuts off the insulin pump for two hours when a low blood glucose threshold is reached. This can be canceled at any time, and after two hours the pump automatically resumes basal insulin delivery. Three studies have been published to date on efficacy of LGS, and so far all have found that use of LGS decreases hypoglycemia events and does not increase hyperglycemia.
  • In the ASPIRE study, 50 adult type 1 diabetes patients (baseline A1c 8%, average age 34 years, average BMI 27, and 56% male) were randomized to an LGS-on or LGS-off group and were driven to hypoglycemia (<70 mg/dl) by fasting and exercise and then observed for 3.5 to 4 hours. In the LGS-on group, the LGS system would shut off insulin infusion when the patient reached 70 mg/dl, whereas in the LGS-off group, the pump would remain on. After a washout period of three-to-10 days, the experiment was repeated with treatment assignments crossed over – i.e., LGS-on patients switched to LGS-off, and vice versa.
  • On average, duration of hypoglycemia was decreased with the LGS feature turned on (138.5 minutes vs. 170.7 min with LGS turned off; p<0.01). Additionally, blood glucose at the end of the observation period was 91 mg/dl in the LGS-on group compared to 66 mg/dl in the LGS-off group (p<0.001). The lowest blood glucose reading (the nadir) during the experiment was statistically lower in the LGS-off group (57.6 mg/dl compared to 59.6 mg/dl in the LGS-on group, p<0.015) but clinically very similar. No patients experienced any serious adverse events (DKA, severe hypoglycemia, or death).
  • Of note, the results revealed an order effect: the study participants in the LGS-on group after the crossover had a significantly longer duration of hypoglycemia than those in the LGS-on group prior to the crossover, but in all cases the LGS-on group always spent less time in hypoglycemia than the LGS-off group. The LGS-off group was unaffected by the order of treatment, and the order effect was not seen for the nadir endpoint. Age, weight, and duration of diabetes did not explain the order effect, but previously induced hypoglycemia (the experiment could be repeated up to three times in each patient before a successful run was achieved) did. Dr. Kaufman interpreted this to signify that “hypoglycemia begets hypoglycemia” and concluded that long-term use of LGS could help further reduce hypoglycemia by disrupting this vicious cycle.


Aaron Michels, MD (University of Colorado School of Medicine, Denver, CO)

Dr. Aaron Michels provided his perspective on the role of CGM in diabetes management. He began his presentation reminding the audience that the more you check your blood sugar with SMBG the better your glucose control, and likewise, the more you use your CGM, the better your control. He noted that CGM provided many benefits, including lower A1c, decreased hypoglycemia, less glucose variability, and weight neutrality or weight loss. He explained that the JDRF CGM and STAR studies were demonstrative of the important role of CGM in people with type 1 diabetes. Then, Dr. Michels transitioned his focus to people with type 2 diabetes. After highlighting case studies of patients with type 2 diabetes who showed improvements in glycemic control and were able to decrease basal insulin units with CGM wear, Dr. Michels concluded that CGM was beneficial to people with type 2 diabetes on insulin in addition to people with type 1 diabetes.


John Pickup, MD (King’s College London School of Medicine, London, UK)

Dr. Pickup gave an excellent review of patch pumps and then offered his opinion on what is next for closed-loop systems. As has been a theme here at Keystone, he expressed optimism for the low glucose suspend (LGS) system as a step toward a fully closed loop. He expressed his personal view that algorithms and pumps have been sufficiently optimized at this time and that the bottleneck for developing a closed loop will be developing more reliable glucose sensing and improving the rapidity and predictability of insulin absorption – this was our sense at this past ADA as well. More reliable and less invasive glucose sensing, as he explained, is currently in development (he described two efforts: fluorescence-based glucose binding proteins and “smart tattoos”). Finally, he warned against the burden of too much technology since the burden of diabetes is already substantial – a patient must already juggle constantly thinking about blood glucose, food intake, exercise, illness, stress; he or she must carry around numerous devices (pump, BGM, CGM, cellular phone, and instruction manuals for every one); and he must keep track of all the people in his life (doctors, nurses, dietitians, family, school, pharmacy) that need to be kept up to date on his diabetes. Dr. Pickup emphasized the need to consolidate and connect all of these people and devices, stating that mobile healthcare should reduce the burden of diabetes technology.

  • Dr. Pickup reviewed the growing number of patch pumps in development or on the market, highlighting particularly noteworthy features and timelines. Interestingly, he explained that patch pumps try to address the top concerns that came up in a long-term CSIIsurvey conducted 30 years ago: smaller/lighter, improved body attachment, and improved cannula insertion and design.
    • Insulet OmniPod: Despite its launch in 2005, Dr. Pickup noted that it has been slow to reach Europe (2010 in the UK and Germany). He explained that its uptake and distribution have not been helped by Insulet’s pricing strategy.
    • Solo MicroPump: Dr. Pickup reminded the audience that the Roche-acquired pump is FDA approved but has not been released. He described the pump’s design, which is intended for 90-day use. It includes a cradle/patch attachment with an inserter/cannula, a detachable two-part mini-pump (reservoir and pump), and a remote control.
    • Valeritas V-Go: This newcomer (its “limited” launch occurred in June 2012) is non- electric and uses a spring-like mechanism. It is both FDA approved and CE Marked. The V-Go has a number of preset basal rates and two-unit manual bolusing on demand. The device can be worn for one day at a time.
    • CeQur PaQ: This pump is also intended for type 2s, but has the advantage over the V- Go of three-day use. It includes a disposable insulin reservoir that attaches to a reusable electronic messenger. Also similar to the V-Go, it has preset basal rates and on-demand bolusing. A clinical trial is occurring in Austria ( identifier: NCT01535612) and the intention is to launch the device in Europe. For more on the device and the trial design, see our recent report at
    • Debiotech Jewel Pump: Dr. Pickup reviewed the two-part patch pump that uses Micro-Electro-Mechanical System (MEMS) technology. He said that the device is currently in a clinical trial in France. We found two studies of the Debiotech Jewel Pump listed on – a technical evaluation of the pump in 28 individuals has recently been completed (NCT01640210), while a semi-closed-loop study using the pump is expected to complete in October 2012 (NCT01640223). Both are in France. This two- part pump was a highlight of the exhibit hall at ADA 2010 (see page 403 of our ADA 2010 report at or our first Closer Look on the company at At that time, we understood that the pump was under FDA review. Since then, we have not heard much about the Debiotech or the Jewel Pump.
    • Cellnovo: After reviewing the features of the patch pump, Dr. Pickup honed in on the touchscreen remote control (with integrated LifeScan blood glucose monitor) and the “interesting” wireless cloud-based connectivity. We heard at ADA that Cellnovo hopes to submit approval to FDA later this year in hopes of launching their product in the US in 2013. Ongoing usability trials are gathering information concerning efficacy and quality of Cellnovo’s pump before release in the EU.
  • Dr. Pickup summarized recent data on Medtronic’s low glucose suspend pump and reviewed the artificial pancreas work occurring in Europe. Dr. Pickup believes patients likely to benefit most from LGS pumps are those with a continued high rate of hypoglycemia or hypoglycemia unawareness even after optimized CSII therapy. This may be due to variable insulin absorption, gastroparesis, unpredictable responses to exercise, a mismatch of insulin and food at meals, or renal impairment. Patients with excessive fear of hypoglycemia and intentional maintenance of a high A1c (to avoid hypoglycemia) are also likely to benefit. Dr. Pickup concluded this section of the presentation by reviewing the EU-funded AP@home project – for extensive detail on the group’s ongoing efforts, see our ATTD report at
  • To improve CGM, Dr. Pickup highlighted fluorescence-based glucose binding proteins, which are not based on glucose oxidase and thus not subject to electrical interference. He explained the mechanism of action and emphasized the day-to-day stability of this sensor technology. As a reminder, BD is working on an optical-based, glucose binding protein (GBP) CGM – the company is currently recruiting for an 18-patient study ( identifier: NCT01469715), with completion slated for October 2012. An announcement was made prior to ADA 2012 that BD is collaborating with JDRF and the Helmsley Charitable Trust’s Sensor Initiative to accelerate development of a more accurate and reliable CGM sensor. For more details, see our report at
  • Dr. Pickup also discussed smart tattoos as another approach to improving CGM. This technology requires subcutaneous or intradermal injection of nano or micro-sensors. Once in place, the sensor is illuminated with light and glucose is recorded as a function of fluorescence changes. He stated that there are smart tattoos currently in animal studies with “reasonably” successful results.


Jay Skyler, MD (University of Miami Miller School of Medicine, Miami, FL); Fran Kaufman, MD (Chief Medical Officer, Medtronic, Los Angeles, CA); Aaron Michels, MD (University of Colorado School of Medicine, Denver, CO); John Pickup, MD (King’s College London School of Medicine, London, UK).

Q: In regard to the low glucose suspend study (ASPIRE), how did the patients interact with the LGS when they saw hypoglycemia. Did they treat their lows with carbohydrates or just rely on the suspend period?

Dr. Kaufman: They had no treatment at all. If they required treatment, the observation time was over.

Q: Do you think that will change how we advise patients?

Dr. Kaufman: This study was designed to look at keeping insulin on or shutting it off. It is not about informing real clinical practice. We would by no means suggest that awake and alert patients should not respond to their lows. You treat your hypoglycemia as you normally would – the LGS has been built like an airbag for if you are not awake or alert.

Q: A comment for Dr. Skyler on uncontrolled patients, especially teenagers with type 1. They want what they call a “GPS” [CGM] so they can see where they are at all the time, but the problem is that I can’t get insurance to pay for it unless they are using a pump.

Dr. Kaufman: I’ve asked my adolescents, while they’re waiting to see me, “Why don’t you write down your priorities?” And I had only one person write down diabetes as priority. Controlling their diabetes is really not what they’re focused on trying to do – they have other things on their minds.

Regulatory Perspectives on Diabetes Drugs & Devices


William Hiatt, MD (University of Colorado School of Medicine, Denver, CO)

Dr. Hiatt provided an overview of efficacy and safety considerations in the development and regulation of diabetes drugs, discussing endpoint selection, the use of active comparators, and the current cardiovascular risk assessment requirements for diabetes drugs. In addition, he touched on the regulatory hurdles for obesity drugs, stating that we can expect similar cardiovascular risk assessment requirements in the upcoming future.

  • He discussed endpoint selection and the use of active comparators in clinical trials. Dr. Hiatt emphasized the importance of selecting an appropriate primary endpoint – if the primary endpoint is not met, it doesn’t matter what secondary endpoints are met; the trial is considered a failure. This is perhaps an obvious truism but we think his point was about how incredibly much thought needs to go into it; there are a number of failed trials where different design work on dosing etc. may have changed the outcomes, such as in anti-CD3 (20/20 hindsight is easy to have of course). Moreover, Dr. Hiatt stressed that the importance of using composite endpoints only where they can really be justified, i.e., where they make clinical sense - for example, in a CV study, combining nonfatal MI with stroke would make sense, but including hospitalizations would not. He questioned how clinically relevant surrogate endpoints (e.g., A1c, blood pressure, and weight loss), as they do not necessarily translate to improved outcomes. We would have liked more discussion on this point; it may be they do not translate in the initial duration of the trial, but follow up data on UKPDS showed that tighter control, as measured by A1c, did impact macrovascular outcomes (though it took a very long time to show this). Subsequently, Dr. Hiatt continued his discussion around efficacy by talking about design necessities in active comparator non-inferiority trials, emphasizing the ethical imperative (where you cannot deny a patient an active drug when the established therapy convincingly bests the placebo, and therefore must compare the new drug to the established therapy). He also noted the nuance in how active-comparator trial objectives are phrased – if a trial is a non-inferiority trial, it cannot prove superiority (only non-inferiority) even if the data supports the higher marking.
  • Dr. Hiatt reviewed the current CV requirements for diabetes drug approvals. As a reminder, companies developing diabetes drugs must: 1) establish an independent CV endpoint committee; 2) adjudicate all CV events from all phase 2 and 3 trials; 3) conduct a meta-analysis and create a prospective analysis plan; and 4) achieve an upper bound of less than 1.8-fold risk in a two-sided 95% confidence interval pre-approval (the upper bound is between 1.3 and 1.8, then an additional phase 4 safety study must be conducted post-approval). Using liraglutide (Novo Nordisk’s Victoza) as an example, Dr. Hiatt noted that the upper bound of the 95% confidence interval of liraglutide versus all comparators was 1.61 (the upper bound was 2.83 versus placebo, and 1.66 versus active comparators); ultimately, the FDA’s Endocrinologic and Metabolic Drugs Advisory Committee voted 8-5 in favor of approval with this data in hand.
  • Interestingly, Dr. Hiatt also discussed adverse event reporting. He said it was important but lamented how imprecise it was – e.g., reporting on chest pain is challenging to make consistent among patients. He said, of course, adjudication committees were essential to have established beforehand, though we can still see the challenges here.


Richard Insel, MD (Chief Scientific Officer, JDRF, New York, NY)

Dr. Insel emphasized that in order for JDRF to achieve its three goals surrounding type 1 diabetes – to prevent, treat, and cure the disease – it must continue to take a proactive role in working with regulatory agencies, including the FDA, to help make them smarter about type 1 diabetes on behalf of patients. In our view, JDRF has done a very impressive job of this in the last several years in particular. Dr. Insel highlighted the collaborations that JDRF has undertaken with the FDA in order to facilitate better regulatory clarity, transparency, and consistency: JDRF employs full-time regulatory experts to guide dialogue with regulatory agencies in the US; JDRF has cohosted workshops with the NIH and FDA on the artificial pancreas and islet-derived stem cells; and JDRF conducts educational seminars and advocates for classes of agents in one-on-one meetings with the FDA. As an example of a highly successful collaboration that JDRF has had with the FDA, Dr. Insel recounted that JDRF convened a clinical panel in 2010 to draft a guidance for the approval of artificial pancreas systems, which the FDA took into account in its development of a draft guidance policy that it drafted up “in record time” by December 2011. Dr. Insel stated that JDRF was pleased with the FDA’s recent responsiveness to recommendations provided by JDRF. He concluded that there is a vital role for foundations, such as JDRF, in influencing regulatory agencies. We would certainly agree, particularly for foundations where leadership is as strong as it is at JDRF.


Matthias Schweitzer, MD (Roche, Mannheim, Germany)

Dr. Schweitzer expressed a generally unenthusiastic opinion of the approval processes prescribed by regulatory agencies around the world. He argued that the major deterrents to industry reaching its full potential to innovate and provide patients with drugs they desperately need are the inconsistency and unpredictability of regulatory agencies. Various countries have myriad different approval procedures and policies, and given the enormous number of new medical products in development, it can be difficult for globally active companies to deal with the numerous different requirements in each arena. Unpredictability is a problem, Dr. Schweitzer argued, because company managers must believe a product has a reasonable chance of being approved before even allowing development to begin. In a large company that is active in many therapeutic areas, as Roche is, budgetary constraints may force an essential diabetes drug to be nixed in favor of an easier-to-approve alternative in a different department. Better predictability would allow companies to better gauge the probability of approval for drugs before pursuing development and allow greater room for innovation.


Terry Gregg (President and CEO, Dexcom, San Diego, CA)

In his overview of the device regulatory process, Mr. Gregg reviewed the device classification system, the costs of device development, the design control process, and the differences between 510(k) and PMA submissions. In particular, he emphasized the need for close communication and collaboration with the FDA for class 3 devices (devices for which premarket approvals [PMAs] are required).

  • He began by differentiating the three classes of devices: 1) class 1, which includes devices like tongue depressors; 2) class 2, which require a 510(k) premarket clearance based on substantial equivalence to a marketed device (e.g., insulin pumps and blood glucose monitors); and 3) class 3, which require premarket approval (PMA) and apply to novel devices like CGM, LGS, and the artificial pancreas. (Mr. Gregg noted that it costs approximately $170-200 million to take a class 3 device from concept to clinical studies to approval.) He emphasized the distinction between the class 2 and class 3 devices (as these are most relevant to diabetes) and ensuing regulatory process. For class 2 devices, risks are well understood, standards of evidence involve substantial equivalence, and clinical data is not always required (however, Mr. Gregg emphasized that the FDA’s view on the latter is evolving and he thinks more clinical trials will be required, and agreed that they should be). In contrast, for class 3 devices, risk levels are not sufficientlyunderstood, valid scientific evidence of safety and effectiveness is necessary, and clinical studies are almost always required. Mr. Gregg also noted that there is less reliance on FDA guidelines and standards, so collaboration with the FDA is critical in the PMA process.
  • To paint a fuller picture, Mr. Gregg laid out the steps a company might go through to get a device approved. The design control process includes: 1) deciding on specifications and requirements for the artificial pancreas, which includes a sensor, a pump, an infusion set, an algorithm and insulin; 2) the initial product development cycle; 3) verification and validation (showing that every step in the process makes sense); 4) clinical testing (both feasibility and pivotal studies); and 5) submission. He noted that it often takes over four years from concept to approval, and that many devices never completing the process. This intensive process for devices requiring PMAs to demonstrate safety and efficacy as required by the FDA (compared to safety and performance for CE Marking) is often why device availability in the US lags behind Europe.


Courtney Harper Lias, PhD (OIVD Associate Director of Toxicology, FDA)

Dr. Lias focused on regulatory issues surrounding diabetes devices and stressed the FDA’s commitment to providing products to patients in a reasonable time frame while still ensuring their safety and efficacy. To this end, the FDA is currently pursuing a program called the Medical Device Innovation Pathway 2.0, which aims to, according to the FDA’s website, “make the regulatory process more efficient and timely” by increasing contact and collaboration between the FDA and innovators and providing additional tools and resources for informed decision-making. Dr. Lias also provided her thoughts on the directions in which various device classes are headed. In the blood glucose monitoring (BGM) arena, a challenge will be to tighten accuracy, given that home care regimens are now different from what they were in the 1990s when BGM standards were instituted. She foresees that accuracy will also need to improve in CGMs and hopes that they may one day replace self-BGM devices. FDA communications with device companies have been encouraging on this front; industry has planned many incremental changes in the future to make sensors more accurate, comfortable, and easy to use. In terms of software and data management, Dr. Lias highlighted the need to standardize the format of data output to facilitate patient-provider interactions. In the remote monitoring field, she stressed the need for security in data transmission. Importantly, she asserted that the FDA fully supports a movement toward consolidation of devices, potentially using mobile phone applications. She expressed uncertainty as to how this could be regulated, but did not offer any potential solutions. Finally, she reported that the draft guidance for approval of an artificial pancreas system was published in December 2011 and is currently being finalized based on feedback from both patients and clinicians. In order to make the review process more efficient, the previously disparate groups reviewing the artificial pancreas regulatory requirements have been consolidated into one group.


William Hiatt, MD (University of Colorado School of Medicine, Denver, CO); Matthias Schweitzer, MD (Roche, Mannheim, Germany); Terry Gregg (Dexcom, San Diego, CA); Courtney Lias, PhD (FDA, Silver Spring, MD).

Q: With regards to the FDA and AP, one of the speakers talked about nine months being typical time to device approval. When can we expect the AP to be available?

Dr. Lias: The systems being developed for suspending at a threshold are fairly far along. Medtronic announced that it had submitted for review and the others are in clinical studies. I expect the suspend systems will start at a broad range for hypoglycemia and hyperglycemia and over time tighten that target. The data from current studies and algorithms are very encouraging and we are hopeful these devices will be developed soon. For a fully closed loop, it will definitely be over a year.

Mr. Gregg: The definition of AP has many sisters. What is an AP? Courtney [Lias] mentioned the closed loop, which is different from LGS. It is a system, so it comprises a sensor, a pump, an algorithm, an infusion set and insulin. So should any of these systems fail, they all fail, so it brings up the issue of a requirement for redundancy. I tried a long time ago to create a fully implanted AP and we failed. But today I have great hope that while I am still working we will see an AP, even before we see a cure.

Q: Thank you for telling us how much it costs to bring your products to market. I won’t ask you why you haven’t sought a pediatric indication for the Gen 4, but will you in the future? The lack of such indication is seriously limiting its use for third-party providers.

Mr. Gregg: We agree. The pediatric population is a little different. Since we’ve had this in development we wanted to go through adult population first. The Gen 4 today is really a Gen 4.5 after a restart we had. We have received FDA approval for pediatric trials. Actually the first patients for this trial will be here in Colorado later this month. I was just talking with Dr. Chase yesterday; other sites have IRB approval, and we’re going through the process. We expect this to take several months to complete. The goal is to submit to agency before end of this year.

Q: Could the FDA enforce that the companies allow an open platform? For example, one CGM may be better where as a pump from a different manufacturer might be the ideal pump. Do you see the FDA’s role of making a more open platform so we can suit what is best for patients?

Dr. Lias: We have addressed this in our draft guidance to some degree. We believe they should be regulated as a system. This doesn’t mean it is impossible to do what you’re saying, but it is difficult in terms of how software interfaces work together. We can’t design the things ourselves – we can’t say you’re the best insulin pump and you’re the best CGM – but we can encourage aspects that make them safer. What drives everything is the science. We are happy to talk to people about that, and want what makes sense for patients.

Dr. Satish Garg (University of Colorado School of Medicine, Denver, CO): As most of you know, there are only 2,000 of us [endocrinologists] taking care of 24 million people in the US. In other words, it’s totally impossible. Thus, 90% of patients are not seen by endocrinologists. So it makes sense that patients be advised and use mobile technologies. Why are we creating bigger hurdles for pharmaceutical and device companies to allow these to be available in the marketplace?

Dr. Lias: I am curious as to what hurdles you believe to be in place. Our center is communicating broadly with industry about medical mobile applications. We’re proposing even that there is no regulatory bar for some uses there, for example, if it is just to communicate well-known information. For other uses, the bar needs to be set for risk of it going wrong. The degree of regulatory scrutiny really needs to make sense with what you’re actually doing. With regard to mobile apps, we know that’s the direction the world is going in. We’re not standing in way of people developing those products, and we’re encouraging people to work together so that they can be available to patients.

Dr. Garg: Does the mobile phone then become a medical device then?

Dr. Lias: The phone itself, no, but the software application might be depending on what it does. Apple does not want to be a medical device, and we have not said they are.

Mr. Gregg: We [Dexcom] work with both Android, Apple, and the FDA. I congratulate the agency on the rapid adoption on how to achieve this with the understanding that there are safety concerns. For example, if you download Angry Birds on your phone, will that disrupt the signal? Do software upgrades corrupt signal? This protocol must maintain itself in the face of constant upgrades. We go to the agency and sit down with them and say, this is how we look at the landscape, what do you think? We’re doing this together. We’re all racing as fast as we can.

Dr. Richard Bergenstal (IDC, Minneapolis): This is for Courtney and Bill, I’ve been impressed with your guidance saying that you’re looking at CGM as a more valuable marker for the time someone spends in hypoglycemia. Across the way to your partners in drug development, do you talk to each other – for example saying that CGM could provide a good tool to show a certain insulin causes less hypoglycemia? Can you show them the wisdom about thinking about CGM?

Dr. Lias: We do have members from the CDC [Editor’s note – we didn’t catch this and are not sure she said CDC] in our group. Can you use CGM as a marker for efficacy? If a drug sponsor wanted to propose CGM as a surrogate, the Center for Drugs would come talk to us about it. I believe time in range is valuable, but the question remains whether that can be the only end point that applies.

Dr. Bergenstal: So you’d welcome them coming over and speaking to you.

Dr. Lias: Sure sure, we talk to them a lot more often than we used to.

Dr. Hiatt: I’ve noticed the significant evolution in their thinking about drug trial endpoints. They traditionally improve drugs based on changing numerical endpoints. They are moving away from those endpoints to more clinical endpoints, which is more consistent with how CV drugs have been looking at endpoints. So you would have to demonstrate that there is clinical significance to time in range. The challenge going forward in diabetes drugs is whether you can you show that there is less renal injury or syncope or MI/stroke or death as a result of the surrogates – which I think makes it more challenging but more clinically relevant.

Q: There’s been lot of discussion about devices vs. medical management and maybe even mobile devices. A pump can easily use an algorithm to tell the patient how much insulin to take based on carbohydrate intake. I’ve been told there is a regulatory or industry on other devices, whether it’s an app or a meter, that could do the same calculation such that a patient on MDI could input carbohydrate intake and calculate insulin dosage.

Dr. Lias: Insulin dosing calculators are regulated as medical devices by the FDA. They haven’t yet been developed for BGMs, though there are standalone software apps approved for that. We are seeing a lot more interest in doing that, both for type 1 diabetes and type 2 diabetes. They haven’t been submitted yet, and we can only review what comes to us.

Q: Help me understand the logic in providing downloads for professionals that look different for patients at home. When I want to teach my patient about the download, the information they are getting is different. It really is a hindrance in helping patients interpret their own data. [Applause from audience]

Dr. Lias: There is a big push to making that not the case. There should be a standard data format that patients and provider can use to increase their communication and understanding between them. If I were to design it, every device would go into a single type format so the HCP wouldn’t also have to keep reinterpreting every format they see.

Q: Could the FDA enforce that?

Dr. Lias: We have limited ability to force that to happen. We have some tools to encourage that to happen, but it is challenging for us to force that to happen.

Mr. Gregg: I would suggest the market dynamics would enforce it – the market itself will dictate what tools will be used.

Dr. Irl Hirsch (University of Washington, Seattle, WA): This is with regards to standardization. For Courtney, some have been over critical with how we’ve been over zealous in terms of safety in some of these topics, but I’m surprised how little attention has been put towards bolus calculators for insulin pumps. Depending on which pump you have, you can have a different dose recommendation – pumps are giving disparate results. I’d like to see standardization or at least the FDA looking into it.

Dr. Lias: Unfortunately, I’m not aware of a standardization effort, but I will definitely take your comments back.

Q: I have a question related to drugs in the adult world. A single-dose pill, as opposed to a per-kilo approach, is always used, and it may be very overwhelming for a 90-lb, 65-year-old woman compared to a 300-lb man. Does this affect your thinking process on how drugs are approved, dosing, how they come to market, and how patients can get access? Primary care practice won’t change unless there is a literature saying that is the case because providers are so afraid of lawsuits.

Dr. Hiatt: The pharmacology of every drug that comes to market is very well known. Dosing recommendations depend on a variety of factors: metabolism (whether it is renal or liver), body weight, or renal/hepatic insufficiency. Dosing is more of an issue in children. If there is a particular drug that has a per kilo requirement, that would be a labeled indication.

Insulin Resistance, Incretins, & Diabetes


Aaron Cypess, MD, PhD (Joslin Diabetes Center/Harvard University, Boston, MA)

Dr. Cypess expressed optimism for the use of brown fat in the treatment of obesity and diabetes. Brown adipose tissue (BAT) is different from white adipose tissue (WAT) in that it burns energy instead of storing it (as a cold-protective mechanism). He argued for BAT’s relevance as a treatment target, relaying that 1) brown fat could be found in up to 100% of the adult human population upon activation (though he presented evidence that brown fat was detected at basal conditions only in ~4% and ~7% of adult men and women, respectively); 2) there are multiple imaging techniques (e.g. PET/CT and MRI) now available that can measure BAT activity; 3) BAT has the potential to consume 100-200 kcal/day and perhaps even more with sustained activation; and 4) once stimulated, BAT is activated within seconds. Dr. Cypess gave a compelling talk, though we are a bit less optimistic than he was about the utilization of BAT for treatment of obesity and diabetes. For people who do not have a high baseline level of BAT, it seems that currently the only reliable way to activate the genesis or activity of BAT is by cold exposure (participants in studies wore a cold-vest at 57 F), and don’t see this as a viable treatment option for a chronic disease. Perhaps if a more convenient method of activating BAT (i.e., a pharmaceutical option) were developed, it might have more potential as a treatment option.


Gerald Shulman, MD, PhD (Yale University School of Medicine, New Haven, CT)

Dr. Gerald Shulman focused on insulin resistance in the muscle and importantly, how insulin resistance in the muscle can lead to resistance in the liver; thus, he posited that improving muscle glucose sensitivity helps improve resistance in the liver. As for a mechanism, Dr. Schulman hypothesized that free fatty acids in the muscle were affecting the translocation of GLUT-4 to cellular membranes, which in turn was causing insulin resistance (as a reminder, the GLUT-4 transporter mediates the transport of glucose from the bloodstream into the muscle). He presented findings from murine models that suggested elevated levels of diacylglycerol (DAG; a lipid intermediate) inhibited PI-3 kinase activity, which is required to translocate GLUT-4 to the membrane. Murine experiments, he explained, have indicated that a similar pathway (whereby increased DAG increases insulin resistance) occurs in the liver. Dr. Schulman concluded his presentation by identifying exercise as a potential solution to muscle (and liver) insulin resistance. Abnormal patterns of energy storage in the muscle, Dr. Schulman said, can be reversed with exercise (which in turn improves conditions in the liver); with exercise you get increased muscle glycogenesis, decreased hepatic de novo lipogenesis, and decreased liver triglyceride synthesis.


Nestor de la Cruz-Muñoz, MD (University of Miami Miller School of Medicine, Miami, Florida)

Dr. Nestor de la Cruz- Muñoz, a bariatric surgeon, reported on the effects of various types of bariatric surgery (including Roux-en-Y gastric bypass (RYGB), gastric banding, and sleeve gastrectomy) on gut hormones (including leptin, GLP-1, GIP, CCK, and Peptide-YY [PYY] among others). Very interestingly and encouragingly, he presented the finding that post-prandial GLP-1 and PYY (a hormone associated with appetite reduction that is produced by L cells, the same cells in the small intestine that produce GLP-1) secretion increased after RYGB and sleeve procedures. These increases could not be explained by weight-loss alone, suggesting that they were produced as a result of bypassing much of the digestive tract. Accordingly, another group found that ileal transposition (connecting the ileum [the most distal part of the small intestine where L cells reside] directly to the stomach) results in 3.5-4 fold increases in post-prandial GLP-1 and PYY secretion. These findings may explain why, as Dr. de la Cruz-Muñoz relayed, bariatric surgery helps patients with diabetes decrease their blood sugars even before significant weight loss occurs.


Steve Edelman, MD (University of California San Diego School of Medicine, San Diego, California)

Dr. Edelman presented a very positive outlook on the potential combination of GLP-1 agonists and basal insulin therapy (he noted his presentation would not address DPP-4 inhibitors). Notably, he said the GLP-1 analogs (available and in development) represent one of the greatest classes in the treatment of diabetes that we’ve seen in a long time. He presented data from six studies investigating GLP-1 agonists with basal insulin therapy demonstrative of why he thought treating patients with this combination therapy was better than just adding more insulin. These studies showed the following: 1) greater A1c reductions with GLP-1 + basal insulin vs. placebo + basal insulin; 2) better weight change profiles with added GLP-1 therapy vs. placebo; 3) improved fasting and post prandial glycemia; 4) minimal hypoglycemia with GLP-1 therapy (while Dr. Edelman noted that rates of hypoglycemia were generally higher with GLP-1 therapy over placebo, this was in the context of greater A1c reduction); and 5) worse side effect profile for gastrointestinal events with GLP-1 therapy including nausea, vomiting, and diarrhea. Dr. Edelman concluded that GLP-1 agonist and basal therapy insulin work well together and make very good physiological sense.


John Pickup, MD (King’s College London School of Medicine, Guy’s Hospital, London, UK); Satish Garg, MD (University of Colorado, Denver, CO); Aaron Cypess, MD, PhD (Joslin Diabetes Center/Harvard University, Boston, MA); Gerald Schulman, MD, PhD (Yale University, New Haven, CT); Nestor De La Cruz-Munoz, MD (University of Miami, Miami, FL); Steve Edelman, MD (University of California San Diego School of Medicine, San Diego, California).

Q (Dr. Amanda Adler): A question for Dr. Edelman. Why do trials for incretins show an impressive placebo effect?

Dr. Edelman: I noticed that too. In some placebo studies, it was quite flat and no improvement was observed, but in some cases there was. I think it really depends on the duration of diabetes, baseline A1c, etc.

Dr. Adler: Placebo is quite cost effective, I understand. [Laughter]

Q: This is for Aaron. The induction of brown fat ultimately has the effect of raising body temperature. What happens if we pharmacologically increase brown fat in Washington, DC where it’s 105 degrees?

Dr. Cypess: My parents live in Potomac, MD so I’m familiar with this issue. One of the things to appreciate when we cool people down with a cooling vest is there body temperature core is only 0.5 degrees lower. However if you move to pharmacological interventions, what are you going to worry about? I think we can expect pharmacological intervention to lead to about a 0.5-1 degree Celsius increase, so we shouldn’t have to worry about hyperthermia.

Q (Dr. Edelman): I’ve had two patients with bypass that lost a lot of weight and went off all their diabetes medicines, but are now developing severe hypoglycemia. I’m not sure what to do with them. I have one on a glucagon pump and another on diazoxide.

Dr. Munoz: We’re seeing some hypoglycemia after bypass, usually a couple years later. Most patients developing hypoglycemia resolve with dietary changes. A lot of it has to do with how the brain thinks the body is eating more carbs since the food reaches the intestine so much more quickly. If we separate liquids and solids, decrease unaccompanied carb intake, and work on the diet side, about 95% resolve on their own. Some go to pharmacologic therapy, but I haven’t had anyone where we’ve thought about another surgical procedure or reversal.

Q: For Dr. Edelman, looking at pre and post control blood sugars, have you noticed that by normalizing people’s appetites with GLP-1 analogs and basal insulin, you are affecting weight that way?

Dr. Edelman: I have seen it. It’s really the reduction in their feverish appetite that helps out quite a bit. Have I studied it? No. But it really helps.

Q: A question for Dr. Shulman - regarding the 45 minutes of exercise and its role in hepatic fat. Resistance or aerobic exercise? And intense or moderate? This is from a practicality perspective.

Dr. Shulman: The studies we did were in sedentary participants and we worked them at 65% V02 max, so it was more aerobic. Mechanistically I think we’re activating AMPK, which causes GLUT4 translocation and phosphorylates ACC to inhibit it, lowering malonyl-coA and promoting fat oxidation. Also, with chronic activation of AMPK, you get increased mitochondrial biogenesis. Activity, I think is working mostly through those three mechanisms. The exercise we did was more aerobic than anaerobic.

New Directions in Diabetes of Young


Georgeanna Klingensmith, MD (University of Colorado, Aurora, CO)

Dr. Klingensmith acknowledged that the data she presented were just observational, but still underscored the potential power of this tool. She reviewed the mission and demographics of the registry and some of the initial reference data compiled thus far of about 24,000 enrollees. Certain statistics demonstrating that a large proportion of patients are not meeting treatment targets struck us as particularly disheartening: in every age group, average A1c was at least 0.6% above target; additionally, in every adolescent and adult age group, at least 60% are above their A1c target; 44% of adult patients do not meet blood pressure goals even on ACE/ARBs; 40% of adults are overweight, 30% of adults are obese, and 30-40% of children in the database are above BMI targets.


Philip Zeitler, MD (The Children’s Hospital, University of Colorado Denver, Denver, CO)

Dr. Zeitler gave a compelling talk, comprehensively reviewing the results of the TODAY trial, which compared time to failure (loss of glycemic control) of three treatment regimens (metformin, metformin + rosiglitazone, and metformin + intensive lifestyle) for pediatric patients with type 2 diabetes. The underlying hypothesis was that insulin resistance in puberty plays a role in the appearance of type 2 diabetes. Accordingly, after 72 months the trial demonstrated that the addition of rosiglitazone reduced loss of glycemic control compared to metformin monotherapy (failure rates of 38.6% and 51.7%, respectively), but the addition of lifestyle intervention did not (failure rate of 46.6%). The median time to failure for all three regimens was the same in those that did fail (10.3, 12.0, and 11.8 months to failure for metformin + rosiglitazone, metformin + lifestyle, and metformin alone, respectively). Dr. Zeitler also underscored how incredibly rare pediatric type 2 diabetes is, especially before puberty, warning attendees to be skeptical of any such diagnosis. Nearly 10% of patients initially diagnosed with type 2 diabetes by pediatric endocrinologists and enrolled in the TODAY trial tested antibody positive, and therefore should have been diagnosed with type 1 diabetes on a background of obesity. For our full commentary on the TODAY trial, see our May 3, 2012 Closer Look at


Georgeanna Klingensmith, MD (Barbara Davis Center for Childhood Diabetes, Aurora, CO)

Dr. Klingensmith, as a last minute substitute speaker, briefly highlighted the lack of manpower in both pediatric and adult endocrinology. She expressed a sense of urgency to recruit younger people to work in diabetes and supported the inclusion of HCPs in the decision making process with policy makers and insurance companies to ensure that we will be able to keep improving diabetes outcomes.


Georgeanna Klingensmith, MD (Barbara Davis Center for Childhood Diabetes, Aurora, CO); Philip Zeitler, MD (The Children’s Hospital, University of Colorado Denver, Denver, CO)

Q: I’m a primary care pediatrician in a 500,000 person town. When I did my residency in the mid 70s, we did not need CDEs or pediatric endocrinologists for kids with diabetes, we did it all. For the most part we wandered around in the dark, but we did a pretty decent job. I’ve always considered diabetes to be a primary care type illness; today we’re not teaching residents how to manage it and they’re not getting experience in the hospital. I think if we’re going to improve our manpower problem, I think we need to start teaching residents how to take care of type 1 diabetes.

Dr. Klingensmith: The majority of pediatricians, particularly in larger communities don’t have time to do that. They have a 10-12 mi visit per patient and find it too expensive to take care of someone with type 1 diabetes.

Q: Again I think reimbursement for primary care needs to change through the whole system, and you need to be reimbursed for the time it takes to take care of these patients.

Q: For Dr. Zeitler: you mentioned that about 9% of the relatively small population of kids diagnosed with type 2 are autoantibody positive.

Should we be screening for autoantibodies for kids we diagnose with type 2 diabetes? Does that change our management, do they go to insulin quicker?

Dr. Zeitler: There are a couple of issues there: if you’re contemplating type 2 diabetes as a diagnosis, I would absolutely get autoantibodies, particularly if it is a Caucasian kid, but even among minorities you’ll find antibody positive kids. They will indeed progress to insulin sooner. Any kid who is antibody positivity needs to be treated as type 1 diabetic. What is more complicated is where to have that done? Many community labs don’t have well standardized antibody assays. That’s where many hospitals send their assays. And even Mayo doesn’t even have all of the antibodies. It’s a problem to address- antibody standardization. But definitely test, because you can run into trouble.

Q: Dr. Zeitler, are there any studies with incretins or DPP-4 inhibitors in kids with type 2 diabetes? What do we do with our antibody positive obese kids on insulins when the problem is getting worse?

Dr. Zeitler: A number of pharmaceutical companies are undertaking trials with exenatide, liraglutide, and saxagliptin. They are recruiting slowly. As far as what to do with an obese kid with type 1 diabetes, you have to deal with type 1 as you would with any other patient, but they also need intensive addressing of their lifestyle. Certainly you need to deal with 1 and the background that it’s presenting against, and you do need to be quite aggressive comorbidities: lipids, hypertension, sleep apnea.

Dr. Klingensmith: Obese type 1 patients may have insulin resistance, but not to the degree of a type 2 patient. Additionally, with the insulin resistance associated with puberty you get less bang for your buck with medicines.

Comment (Dr. Fran Kaufman, Medtronic): Phil, I don’t know if you emphasized enough about how educational materials are now available from the TODAY trial for free on the NIH website.

Dr. Zeitler: Yes, those were just recently made available on the TODAY Study public website There are extensive materials available for teaching children about type 2 diabetes.

Comment (Dr. Fran Kaufman, Medtronic): I would still say that maybe the issues we face with managing children at school are very important, especially as therapy gets more complex. Right now in America we are at the Supreme Court discussing diabetes management at school as to who can do what, where and when. Maybe eventually there will be time for national policy where we get something that makes sense so we don’t have to battle this from state to state, school to school

Dr. Zeitler: You’ve always been an optimist. The ADA will be publishing TODAY materials as well.

Q: I have a lot of kids who have acanthosis, but I work them up and find nothing. What do we do with them?

Dr. Zeitler: Make them lose weight. It’s a marker of insulin resistance, but it tells you nothing about metabolic state. It’s a very nonspecific symptom. A general workup for a child over the age of puberty would include lipids, A1c, liver tests, think about vitamin D deficiency, ask about sleep apnea, and if all that’s normal they just need to lose weight.

Q: Dr. Klingensmith, would you mind elaborating about insulin resistance you see in patients with type 1 diabetes? Occasionally I resort to a study about “out of control type 1s” who use metformin and it says they might respond differently?

Dr. Klingensmith: I’m only saying that patients with type 1 diabetes who have been studied appear to have insulin resistance of obesity and adolescence, not of the severity of patients with type 2 diabetes, so they might not respond as well to metformin. The T1D Exchange is recruiting overweight adolescents with type 1 diabetes to repeat that in a larger trial.

Dr. Zeitler: A lot of trials up to now have been flawed because there were no insulin reading protocols.

Dr. Klingensmith: I am in agreement. When you do put a child on metformin who has a family history of type 2 diabetes but clearly has type 1, you do have to be careful about insulin dose and remind families that if this is successful then the insulin dose may need to be decreased.

Q: Are any of you using GLP-1s in overweight patients with type 1 diabetes. We’ve all tried Symlin, and it hasn’t been successful.

Dr. Klingensmith: I was talking to someone at lunch who said they’ve never had a Symlin prescription refilled.

Dr. Zeitler: I recently had a type 1 patient and her doctor put her on liraglutide and she lost 15 lbs. I don’t know what that means.

Dr. Klingensmith: I think we’re waiting on results of studies.

Q: Sometimes some of our type 2 kids are autoimmune negative, so we start them on metformin at an A1c of 6.5%-7%. They come back and their A1c is 10% and higher. Sometimes we wonder if we should check to see if they are antibody positive again. Do you do that?

Dr. Zeitler: In terms of kids becoming antibody positive after developing diabetes, some kids do but it’s a small number. Out of 700 there were 10 in our study that developed antibodies. They did have a greater chance of failing as you would expect. One thing is that there is no evidence that treating kids with type 2 diabetes with lifestyle alone is effective. All these kids should be started on metformin immediately. If you wait and let them fall apart it will be hard to get them back on track. The ADA 2012 recommendation for adults doesn’t support treatment with lifestyle and diet for patients at diagnosis anymore because it is ineffective. Another thing that is really important is that you have to reconsider diagnosis based on the course of disease. One common cause of antibody positivity in the TODAY trial was that the result from a common lab came back negative and then when it was done by our lab came back positive.

Dr. Klingensmith: I would add, and I’m not adult endocrinologist, but if you read the literature, mostly labs are just measuring the GAD antibody, and it’s not appropriate approach to just measure 1 antibody; you can’t exclude type 1 diabetes based on that.

Q: Dr. Zeitler, on lifestyle intervention, I was struck by the psychosocial concerns in these populations. It seems like there is lots of room for improvement in lifestyle, but it’s a difficult population to implement change in because of the lack of psychosocial support. How much of the lifestyle intervention was psychosocial based?

Dr. Zeitler: If we redesigned the study we’d have a social work arm. But these kids really worked hard – they were very adherent to the intervention, and it still didn’t work. They came to their visit, took their pills, and really worked hard at the lifestyle arm. It wasn’t like they didn’t do what they were supposed to so it’s important to recognize that the intervention was successfully delivered. It may very well be that a kid does well for a few weeks and then grandpa is diagnosed with cancer and everything falls apart. Case management intervention may be important, but that may be too hard because they live in a whole milieu that is so resistant to change.

Q: About lifestyle intervention in the TODAY study - were they required to exercise? For kids in lower economic groups, it is more difficult to exercise so how did you address those issues?

Dr. Zeitler: In the past, exercise was part of daily life if you were poor, in particular. Now you have to pay to exercise so this is a problem. We did not provide them with a health club membership like DPP did. The behavior interventionist would go exercise with them at a local rec center, and we would help them get discounts at rec centers. An interventionist would teach them how to exercise; we gave them balls, things to exercise with, provided information on things they could do. We kept a record of how much they exercised, and that data is still being processed.

Q: I don’t know of any guidelines for comorbidities and treatment in children and adolescents.

Dr. Zeitler: There are guidelines from the AHA and ADA on the treatment of comorbidities in at-risk children. They are a little less aggressive than adult guidelines. The doses are the same because these are adult sized children.

Q: Are children of BMI 35-40 candidates for bariatric surgery?

Dr. Zeitler: Candidates, sure, but good candidates? That’s less clear to me. It’s unusual for children with a BMI of 40 to come out of supportive environments. Unclear whether they’d be successful for bariatric surgery. In general it’s less successful in children than adults. For some children it’s their only hope, but it needs to be done in a very carefully constructed pediatric experience. Teen-LABS is a national multi center trial looking at outcomes of bariatric surgery in children. It hasn’t been published but results are due soon.

Q: All pediatricians in our community in Las Vegas do screening fasting insulins, so 50% of our referrals are kids with A1cs of around 5%, but with insulin elevated. I don’t treat those children, and I don’t treat with metformin, but I know people who do. This other group is what I would consider prediabetic in adults: A1cs of 5.8% or 5.9%. In these kids I do an OGTT and they are always normal. How do you treat these kids?

Dr. Zeitler: There’s no reason to measure insulin levels – it doesn’t tell them anything new about the child. Elevated triglycerides are a good marker! But what do you care about that? We don’t treat that. In terms of metformin, there is no evidence that this is a useful thing to do in adolescents. There is no evidence that treating insulin resistance itself is beneficial. If you look at progression rates, the progression from insulin resistance to type 2 diabetes in children is very low. Maybe there is a benefit to introducing metformin during puberty to prevent beta cell loss, but those are research questions. For the kids with prediabetes – just make them lose weight. For OGTT, I’d stop. If you have a child with A1c less than 6.1%, the chance that they will revert back to normal insulin tolerance is very high. Of greater concern at an A1c of 6.1-2%, is if you can demonstrate they have loss of beta cell function. That group may need to be of more concern. Prediabetes during puberty needs to be defined at somewhat higher A1c.

Dr. Klingensmith: What about the fatty liver disease study with metformin?

Dr. Zeitler: The TONIC study was problematic. What it showed was that metformin provided no improvement in fatty liver disease measured by liver function test. The problem with that study was that the placebo group lost a lot of weight so they actually improved as much as the metformin group did. So it’s hard to know what would happen with just metformin. It is hard to know if metformin is useful in a child who is not losing weight. A small subset of children got biopsied before and after, but pathology only improved with metformin in a very small group of kids.

Diabetes Complications


David Kendall, MD (Eli Lilly, Indianapolis, IN)

Dr. Kendall provided an overview of how we should address cardiovascular disease (CVD) in the context of diabetes. He began by stating that no more studies are needed to show a reduction in microvascular complications with improved glucose control – there have already been enough studies firmly establishing that relationship. Next, he reviewed the ACCORD study, suggesting that the individuals with greater mortality risk in the intensive control group of the study included: 1) those with long-standing poor glucose control; 2) those with a history severe hypoglycemia; and 3) those who were less responsive to intensive therapy. With regard to the latter, he echoed Dr. Bergenstal’s message earlier that morning, that when patients are unable to achieve their target A1c goal despite a good-faith effort to do so, providers should reassess targets for these patients. Next, Dr. Kendall referenced ORIGIN and the ACCORD blood pressure trial, to say that neither insulin glargine (Sanofi’s Lantus) versus standard care (in ORIGIN) nor intensive systolic blood pressure (SBP) control versus standard control (in ACCORD) led to lower risk for cardiovascular events. Dr. Kendall concluded his presentation saying that the collection of results from the ACCORD, ADVANCE, and VADT studies affirmed the current standards of care in diabetes.

  • Dr. Kendall began with a historical perspective, detailing commonalities observed across landmark trials including the UKPDS and DCCT/EDIC and the more recent ACCORD, ADVANCE, and VADT trials. He noted important distinctions between the UKPDS and DCCT/EDIC trials and the recent clinical trials: 1) the intensive A1c targets in UKPDSand DCCT/EDIC would correspond to the standard targets in ACCORD, ADVANCE, and VADT; and 2) the recent randomized clinical trials were conducted in patient populations with longer- standing duration of diabetes. He explained that every study showed a reduction in microvascular complications with improved control, commenting that no more studies were needed to prove that better control decreased microvascular complication risk. In contrast, he said that neither CV risk nor mortality risk was reduced in any one of these glucose-lowering trials (in the initial trial); on the contrary, ACCORD showed an increased mortality risk with intensive glycemic control.
  • Delving into the unexpected results in ACCORD, Dr. Kendall suggested that those with long-standing poor glycemic control, those with a history of severe hypoglycemia, and (maybe) those who were less responsive to intensive therapy were likely the ones at increased risk of mortality. He pointed to the linear relationship between A1c and mortality risk in the intensive group in ACCORD showing increased risk with intensive therapy versus standard therapy when average A1c was greater than 7.0%. Dr. Kendall echoed what Dr. Bergenstal alluded to earlier that day – when patients are unable to achieve their targeted A1c goal despite giving a good-faith effort, providers should step back and reassess what the appropriate target should be.
  • He highlighted ORIGIN and the ACCORD blood pressure trial results as evidence that neither insulin glargine treatment nor intensive systolic blood pressure control has been shown to decrease cardiovascular risk. Dr. Kendall commented that in ORIGIN, the cardiovascular risk curves for the standard care group versus the insulin glargine group over seven years of follow-up were perhaps the most superimposable graphs he had ever seen. Similarly, he presented data from the ACCORD blood pressure trial, which showed that more intensive control of systolic blood pressure (<120 mmHg target) did not further reduce the risk of cardiovascular disease beyond standard care (<140 mmHg target). However, intensive control was associated with reduced rates of total and non-fatal stroke. (In Dr. Kendall’s experience, it often required three to six antihypertensive drugs to get people to target in the intensive control group in the ACCORD blood pressure trial.)
  • Dr. Kendall concluded that the standards of care in diabetes were affirmed by ACCORD, ADVANCE, and VADT. He noted that none of these trials suggested that we should throw out glucose control or blood pressure control, but rather that our current standards are reasonable. He proposed “a rational approach in 2012” for comprehensive diabetes management founded on three pillars: 1) early more intensive glycemic control to limit the risk of microvascular disease; 2) glycemic control and control of systolic blood pressure to limit the risk and progression of renal and retinal disease; and 3) targeting LDL-c and systolic blood pressure to limit the risk of CVD with the caveat that more aggressive blood pressure targets should be considered for those at the highest risk of stroke.


Jay Skyler, MD (University of Miami Miller School of Medicine, Miami, FL)

In his presentation on microvascular disease in type 1 diabetes, Dr. Skyler suggested that modern approaches to minimize the risks of complications (meticulous glycemic control and early and appropriate use of anti-hypertensive agents) can considerably reduce the risks of death and coronary artery disease via their impressive effect on preventing renal disease. Well known studies have repeatedly shown the benefits of tight glycemic control for reducing microvascular complications, but there is less encouraging data on reducing macrovascular complications. Dr. Skyler posited, though, that one way to reduce coronary artery disease may be through reducing nephropathy. He stated that while avoiding renal disease in itself is desirable, the benefits do not end there. In a study published in Diabetologia (Borch-Johnson et al., 1985), relative mortality of patients with type 1 diabetes was found to be up to 50 times higher for those with proteinuria than for those without proteinuria (relative mortality for those with proteinuria peaked in both men and women around age 35). Furthermore, the relative mortality from cardiovascular disease was found to be up to 65 times higher in patients with type 1 diabetes with proteinuria (~30 years after onset) compared to the general population (relative mortality from cardiovascular disease was found to be about five times higher in patients with type 1 diabetes without proteinuria compared to the general population). Therefore, Dr. Skyler concluded that “renal disease heralds both cardiovascular disease and death,” so our ability to prevent renal disease has major implications for reducing cardiovascular disease and death.


Marian Rewers, MD, PhD (University of Colorado School of Medicine, Denver, CO)

Dr. Rewers provided a refreshingly optimistic outlook on efforts to reduce the burden of both type 1 and type 2 diabetes. On type 1 diabetes, he predicted that we will know what triggers the onset of disease in less than 10 years through TEDDY, a study documenting environmental exposures in children who are at a high risk for developing type 1 diabetes (Dr. Rewers previously stressed that the incidence of type 1 diabetes was not compatible with a genetic model, suggesting that the developmental trigger had to be environmental). He believed after identifying the trigger – whether a virus, gut microbe, or diet – a vaccine and public health screening program would soon follow. On a more somber note, while he thought important strides have been made in treating many of the complications of diabetes, he saw an unfortunate exception in terms of type 1 diabetes and cardiovascular disease, emphasizing that current preventive efforts are not enough. On type 2 diabetes, he focused on the increasing prevalence of the disease in pediatric patients. While the age distribution for type 2 diabetes onset is shifting slowly downward, he was confident that “we can change,” pointing to the drastic reduction in smoking in the US (“an overnight miracle”) as an example of what could be achieved in type 2 diabetes. When considering treatment, Dr. Rewers echoed the message of individualization that we heard at ADA, saying that we need to look at each individual patient and set the lowest A1c target possible without increasing the risk of hypoglycemia. In addition, he emphasized the importance of quality education (enabling patients to become their own doctors), telemedicine, and effective programs to transition patients from pediatric to adult diabetes care. He concluded his presentation with a look towards the future, hoping that by the time he retired, children would be attending summer camps rather than “diabetes” summer camps.


Jay Skyler, MD (University of Miami Miller School of Medicine, Miami, FL); David Kendall, MD (Eli Lilly, Indianapolis, IN); Marian Rewers, MD, PhD (University of Colorado School of Medicine, Denver, Colorado).

Q: You mentioned doing serum creatinines and GFRs. I was wondering, if I have pediatric patients with poor diabetes control and normal urine microalbumin, who should I do those tests on?

Dr. Skyler: I’d wait until age 20, but I don’t have the evidence to say that.

Q: In the metabolic control study, is there any comparison on how they’re coping with the burden of diabetes compared to those started on a traditional method?

Dr. Rewers: Data are being collected; it is tough. They selected patients from families with someone who has diabetes or someone from a highly educated or motivated family. Even with that, the burden was very high.

Dr. Kendall: Speaking from the ACCORD experience, one of toughest things to do after ACCORD ended was to get the very intensive patients to back off on their intensive control because they felt better on it. I think quality of life cannot be discounted, and good glycemic control, I think, is a very important consideration.

Q: I have a question about the relationship between cow’s milk protein and diabetes development. Is there enough evidence to make recommendations?

Dr. Rewers: No, absolutely not enough. I have to be honest that in the observational studies we’ve done - DAISY and some others in Europe – we have not confirmed this association. I get this question often, and the right answer I think is one should follow the American Pediatric Association recommendations – that moms are encouraged to breast feed as much as possible until age seven, eight, or nine months, then slowly introduce small amounts of cows milk, cereal, and eggs while still breast feeding.

Q: What if moms consume cow’s milk?

Dr. Rewers: That could mean casein and bovine insulin could be transmitted through breastfeeding and I think there is no right answer here, there is no strong evidence. We have to wait until September of next year when we will know more.

Dr. Skyler: The final result [for TRIGR] will be in 2017.

Q: I’m a pediatric nurse wondering about treatment of microvascular disease or screening

– when do you start that? What drugs can we use to treat children?

Dr. Rewers: I am quite aggressive and provide adult criteria to recommend treatment in terms of lipid and blood pressure lowering, which is not a universal view. When you go to pediatric meetings you can be killed for suggesting use of statins for a 10-year-old child with an LDL of 180. There is a trial in Europe that is finalizing enrollment that is testing aggressive blood pressure and lipid lowering in teenagers.

Dr. Skyler: There is a caveat there – stop ACE inhibitors/ARBs, and probably statins, in women attempting pregnancy.

Q: I think my patients with type 2 diabetes die of cardiovascular disease. What are people with type 1 diabetes dying of? And Marian, I was disturbed by the last slide you presented that says the modern era has earlier development of cardiovascular disease. Can you elaborate?

Dr. Kendall: There are components of type 1 diabetes (like glycemic variability, inflammatory stress, etc.), so I think yes they develop CVD but they do so in the environment of a very comprehensive CV risk profile. However, patients with type 1 diabetes are beginning to look more like patients with type 2 diabetes so we may begin to treat them in the same way, but I think it is the additional years of oxidative stress and glycemic variability that account for more of the risk than in people with type 2 diabetes where hypertension and other risks predominate. We are significantly better at capturing and identifying cardiovascular disease now too, so I think capture bias may be part of it.

Dr. Rewers: I agree. Our patients are living longer and they have to die of something, so unfortunately the default is cardiovascular disease. But it is not the full explanation of the data from Pittsburgh. Honestly, I

picked the most dramatic example. I think it is fair to say we are not waiting to battle against cardiovascular disease.

Dr. Skyler: I disagree. The same data from Pittsburgh demonstrates reduced overall mortality in the newer era. If there is less renal disease and less mortality, it is certainly not fatal coronary disease.

Dr. Kendall: I do think what Jay pointed out, that there is a push in this community saying “oh a little microalbuminuria, a little change in renal function” is okay. I think the data Jay showed, whether it applies to pediatrics or not, demonstrates that the absence of renal disease is a very positive predictor for health outcomes in diabetes. So a little change in renal function should not be discounted.

Dr. Georgeanna Klingensmith (University of Colorado School of Medicine, Denver, CO): I wanted to follow up and comment on quality of life in people with diabetes. Childhood studies on quality of life have shown that those kids with better control report better quality of life. Just as people with type 2 diabetes are reluctant to relax on their therapy because they feel better (see the first comment by Dr. Kendall earlier in this session), I think we can agree with that in type 1 diabetes too.

Q: There is a study ongoing in the UK, and I believe they’ve reached 200-300 participants with type 1 diabetes using ACE inhibitors and statins as well. Hopefully, we’ll have some data on that in a couple of years. Also, data from the TODAY study (pediatric type 2 diabetes) where they used ACE inhibitors and statins will be available soon.

Dr. Rewers: Unfortunately, Dr. Hirsch is not here, but we had an interesting conversation earlier. We have the first evidence that glycemic variability – which you can obtain from CGM – is directly related to the amount of coronary atherosclerosis. Cross sectional data showed that people who were more variable had more plaque in their arteries after adjusting for A1c and other factors.

Q: For patients that are intolerant to statin therapy, what is the next best recommendation?

Dr. Kendall: Dietary interventions to lower LDL are tough. Even if you never go to a steakhouse again, you get maybe 4-5% improvement. There still is a place for resins, which are relatively easier to take. For reasons entirely unknown to me, they also lower blood glucose a bit, so I tend to use them in statin intolerant patients.

Dr. Skyler: I do that too.

Self-Blood Glucose Monitoring


Richard Bergenstal, MD (International Diabetes Center, Minneapolis, MN)

Dr. Bergenstal provided valuable insight on how to improve the use of SMBG and CGM in clinical practice. He emphasized the importance of effective communication between patients and providers, which includes collaborating on goals and teaching patients what numbers (like A1c and SMBG values) and trends in glucose control mean. Next, he suggested that frequent insulin dosage titration could improve glycemic control – to that effect, he presented study results (Bergenstal et al., Diabetes Technology and Therapeutics 2012) on the safety and efficacy of the Hygieia Diabetes Insulin Guidance System (DIGS) that gives weekly, automated insulin dose adjustment recommendations. After 12 weeks, the technology improved average glucose by 11 mg/dl (baseline 174 mg/dl) and A1c by 0.5% (baseline: 8.4%) across all patients, while concomitantly reducing the frequency of hypoglycemia by 25%. This is impressive early data and we look forward to the larger phase 2 study, where insulin adjustments will not be double-checked by a live physician. Dr. Bergenstal concluded his presentation echoing what we’ve heard from many presenters at Keystone – that we need greater standardization in diabetes care. He emphasized that a uniform report from every brand of meter and CGM could help patients and clinicians better understand what the reports signify. For more on this hot topic, see our comprehensive report at on the Helmsley Charitable Trust and International Diabetes Center’s recent expert panel on standardizing the presentation of glucose monitoring data.

  • Dr. Bergenstal urged providers to explain to patients what SMBG and A1c values actually mean and how these values are connected. In particular, he asserted that good glycemic control is more than just A1c – it also involves minimizing hypoglycemia and glycemic variability. Dr. Bergenstal explained that this is something we need to explain to patients, FDA, and NIH. We’re hearing more and more of these sentiments from leading KOLs, and it’s coming at the right time, too – we believe that many early artificial pancreas systems (e.g., low glucose suspend, control to range) may have a marginal benefit on A1c, but big benefits on glycemic variability and hypoglycemia. This is still clinically meaningful in our view, and something FDA still seems to be grappling with.
  • Dr. Bergenstal also encouraged providers to collaborate with their patients to come up with goals that patients agree upon and understand. This statement reminded us of the new ADA/EASD position statement on individualizing therapy. For more information, see our report at
  • Dr. Bergenstal asked providers to teach patients the difference between individual readings and patterns in blood glucose levels and how to appropriately respond to those numbers. Encouragingly, we’re already starting to see new technologies that are helping patients and clinicians do this (e.g., the OneTouch Verio IQ, the InsulLinx, the iBGStar, the CareLink Pro 3.0) – we have no doubt that more are in the pipeline, and it will be interesting to watch FDA and its role in this process evolution. Dr. Bergenstal also highlighted the importance of giving patients a consistent method about when to test blood glucose levels, suggesting that testing before bedtime and breakfast were most important.
  • Dr. Bergenstal presented a study evaluating the Diabetes Insulin Guidance System (DIGS) software being developed by Hygieia, which automatically advises patients on insulin dose adjustments based on blood glucose readings (Bergenstal et al., Diabetes Technology & Therapeutics 2012). We note that he did not reference the technology or company by name. The study lasted 16 weeks (a four week baseline run-in period, followed by 12 weeks of intervention) in people with type 1 and type 2 diabetes (n=46). During the intervention period, DIGS processed patients' glucose readings and provided insulin dosage adjustments on a weekly basis. If approved by the study team, the adjusted insulin dosage was communicated to the patients. Baseline A1c was 8.9% across all participants, which declined to 8.4% after the run-in period. The data showed that:
    • The technology was safe – the clinical team agreed with the dose recommendation an impressive 1,731 out of 1,734 times.
    • After 12 weeks, the technology improved average glucose by 11 mg/dl and A1c by 0.5% across all patients. During the run-in period, the weekly average glucose for all participants was stable at 174  mg/dl, which declined to 163 mg/dl by the end of 12 weeks of DIGS dosage adjustments (a 6% improvement; p <0.03). Mean A1c decreasedfrom 8.4% to 7.9% (p <0.05) across all participants during the study. For patients with type 2 diabetes in the study, weekly mean glucose improved by 7.4%. However, among patients with type 1 diabetes, weekly mean glucose did not significantly improve. Dr. Bergenstal attributed this to a decline in hypoglycemia among patients with type 1 diabetes in the study in particular.
    • In patients with frequent hypoglycemia, DIGS insulin dosage adjustments increased mean weekly glucose by 18.7% (152  mg/dl to 180 mg/dl), but the frequency of hypoglycemia significantly declined by 36.3% (167.3 to 106.6 events per patient-year; p=0.047). For these patients, A1c did not significantly change (8.1% to 8.2%; p=0.5) – although clearly “time in zone” would likely have improved substantially.
    • In patients without frequent hypoglycemia, weekly mean glucose declined by 12.2% (178.8  mg/dl to 157 mg/dl), while hypoglycemia remained unchanged. During the 12-week active study phase, daily total insulin dosage was increased, and A1c decreased from 8.5% to 7.8% (p=0.0005).
  • Dr. Bergenstal concluded by calling for standardization in blood glucose meter and CGM reports. He thought the key components of this uniform report would include statistics, a visual component, and a day-by-day breakdown. He emphasized that patterns should be assessed over 14 days (repeating to the audience, “Three days does not describe a patient, three days does not describe a patient…”). However, to develop a uniform report, he said that first we need to standardize definitions of the components of a glucose profile. This would include deciding on an additional number to describe “significant” hypoglycemia or hyperglycemia (he suggested <55 mg/dl for significant hypoglycemia and >250 mg/dl for significant hyperglycemia) as well as a measure of glycemic variability (whether that be standard deviation, coefficient of variation, or interquartile range). We noticed that standardization in diabetes care was a prominent theme at this year’s Keystone conference, something we’ve heard at other recent diabetes technology conferences as well. We hope progress continues on this front, especially on the heels of the recent Helmsley Charitable Trust and International Diabetes Center expert panel on standardizing the presentation of glucose monitoring data (see our report at


Lutz Heinemann, PhD (Science & Co, Düsseldorf, Germany)

Dr. Heinemann gave an evaluation of the current state of SMBG devices and offered solutions for improving “total system accuracy” when it comes to self-testing. He reminded the audience that the existence of a quick and relatively accurate method of SMBG is a real strength of current SMBG technology, especially given that about 40 years ago this was not the case. Such advancements have allowed patients to control and adjust their own therapies. In his view, a real weakness of SMBG today is that some patients may not be adequately trained in handling devices and analyzing the data. Dr. Heinemann emphasized that high “total system” accuracy should be the goal when improving SMBG performance today, because in the end, it is the accuracy of the device in the hands of the patient that matters. Requirements for total system accuracy include ease of use of the device, patient education, device accuracy, device handling, and coordination between a patient’s needs and the quality of the device (e.g., a pregnant woman with type 1 diabetes needs much more accurate readings compared to an elderly type 2 patient on oral agents). He called for improvements in evaluation of device accuracy during the approval process, citing a study published in 2010 where 11 out of 27 (40%) CE Marked blood glucose meters examined did not meet ISO 15197 accuracy standards (+15 mg/dl at <75 mg/dl and +20% at >75 mg/dl) – we would be very interested to see such a study conducted of FDA-approved meters, since safety AND effectiveness are taken into account here in the US (vs. just safety in Europe). Dr. Heinemann also suggested that a manufacturer-independent process of quality assessment might be necessary – he would especially like to see a standardized form of evaluation. In order to improve patient competency with BGM devices, Dr. Heinemann proposed a program in which patients complete a training program to obtain a certificate of competence (much like a driver’s license) that confers certain rights (e.g., access to test strips) and certain obligations (e.g., proof that they are testing adequately).

  • Dr. Heinemann noted that a number of factors influence total blood glucose meter system performance: modifying features like temperature, air pressure, etc.; batch-to-batch variations in test strips; number of handling steps; and touching a slice of apple. This concept of “total accuracy” reminded us of Dr. Barry Ginsberg's talk on BGM accuracy at the recent Clinical Diabetes Technology Meeting. In that presentation, he showed that compared to counting carbs and insulin absorption variability, meter accuracy isn't as big a deal as it might otherwise seem. For more information, see our report at


Helena Rodbard, MD, FACP, MACE (Endocrine and Metabolic Consultants, Rockville, MD)

In the wake of smart phones and rapidly evolving technologies in diabetes management, Dr. Rodbard’s presentation reminded the audience of the importance of using technology to improve diabetes outcomes. She suggested that automated decision support tools could help reduce clinician inertia (one of the major contributors to poor patient glucose control) by improving a clinician’s ability to recognize and respond to glycemic abnormalities seen in structured SMBG data. Dr. Rodbard also reviewed the DECIDE study (Rodbard et al., Diabetes Care 2012), which found that use of structured testing data, combined with a decision support tool and/or an educational DVD, enhanced clinicians' ability to correctly identify significant glycemic patterns and make appropriate therapeutic decisions to address those patterns. She concluded with tempered optimism – “What we’ve got now is not perfect, but we’ve come a long, long way.”

  • Clinician inertia, a major contributor to poor patient glucose control, is driven by a number of factors, including (1) competing clinical demands during time-limited office visits averaging seven to ten minutes (“How much can you do?”); and (2) lack of clinical information (e.g., it’s difficult to make sense out of three months of data).
  • Dr. Rodbard suggested computer assisted decision support (CADS) is a more accessible tool for patients and providers in the management of type 2 diabetes (Rodbard et al., J Diab Sci Tech 2011). CADS is a computerized system to support PCPs, physician assistants, and nurse practitioners in the management of patients with type 2 diabetes. Patients can upload SMBG data by an automated telephone download, and the system will generate graphs and statistics, identify problems (e.g., high before breakfast, low before dinner), and recommend changes in therapy (e.g., add GLP-1, change medication dosages). She explained that the program is currently being tested at four different sites (three military and one civilian) and data will hopefully be available soon. This sounds like a very encouraging use of data analytics and potential way to overcome clinician inertia.
  • While Dr. Rodbard said systems like simple pattern recognition in the OneTouch Verio IQ were easy and provided helpful information, she asked the audience to recall that the retail price of each strip for the OneTouch Verio IQ is $1.55 – “That’s an expensive way of getting advice.”
  • The DECIDE study, which used an automated decision support tool (DST) and/or a 28 minute educational DVD with instructions on identifying glycemic abnormalities, helped clinicians interpret and appropriately respond to structured SMBG data (Rodbard et al., Diabetes Care 2012). Dr. Rodbard noted that endocrinologists and diabetologists were excluded from this study. Two hundred and eighty eight clinicians (39.6% family practice physicians, 37.9% general internal medicine physicians, and 22.6% nurse practitioners) were randomized to review 30 cases involving type 2 diabetes patients using structured SMBG data alone (STG) (n=72); STG and a decision support tool (DST; n=72); STG and the 28 minute DVD training (n=72); or STG, the decision support tool, and the DVD (DST+DVD; n=72). The goal was to identify the principal glycemic feature (e.g., hypoglycemia) and the best course of therapeutic action.
    • Dr. Rodbard concluded that use of the decision support tool, the DVD, and the combination of these two, improved clinicians’ abilities to identify the primary glycemic abnormality and appropriately respond. More clinicians in the decision support tool group (49%; p <0.0001), DVD group (51%; p <0.001), and the combined decision support tool and DVD group (55%; p <0.001) were able to identify the principal glycemic abnormality and prescribe the appropriate therapy compared to the STG group (33%).
    • However, Dr. Rodbard noted even after training and with these tools, many clinicians still failed to identify the correct glycemic abnormality. We would further underscore the fact that these were fairly ideal circumstances: participation in a clinical trial, top technology, and access to structured testing data. In the real world, these three pieces could be absent in many settings, reinforcing how far we have to go towards making data collection and analysis easier for providers. This is especially the case for non-diabetes specialists that treat patients with type 2 diabetes, where time is limited and expertise is often less refined in data interpretation and medication adjustment.


Emanuele Bosi, MD (Vita-Salute San Raffaele University, Milan, Italy)

While it is widely recognized that SMBG is highly valuable in the management of type 1 diabetes and insulin treated type 2 diabetes, its utility in non-insulin-treated type 2 diabetes has been disputed in the scientific literature. The PRISMA study, the largest-ever RCT of SMBG in type 2 diabetes patients, randomized participants to either intensive self-management (ISM; which required performing three four-point SMBG profiles every week and equipped healthcare providers with a colorful “metabolic dashboard” software for visualizing the glucose data) or active control (discretionary, unstructured blood glucose testing). At the end of 12 months, the ISM group had experienced a slightly (but statistically significantly) greater mean A1c decline than the control group: 0.39% vs. 0.27% by intent- to-treat analysis and 0.45% vs. 0.24% by per-protocol analysis. We last covered this study in a Roche sponsored corporate symposium at ATTD 2012 (see our report at Cost- effectiveness has not yet been evaluated to our knowledge; we are unsure whether payors will think the study’s benefits justify high SMBG reimbursement in the general population. At ATTD, Dr. Bosi said that patients in the ISM group used an average of two-to-fourfold more strips than their control group counterparts. However, since people with the highest baseline A1c tend to have both the highest risk for complications and the greatest glycemic benefits from SMBG, we speculate that subsequent analysis could show relatively higher cost-effectiveness in these patients.

  • PRISMA was a multi-center (39 diabetes clinics in Italy), prospective (12 month), randomized, parallel study – 1,024 patients (mean baseline A1c: 7.3%, 6.2 years diabetes duration, mean age: 60 years, ~40% female, mean BMI: 30.5 kg/m2, 40.6% on one oral agent, 41% on two oral agents) were randomized to an intensive structured monitoring (ISM) group or an active control (AC) group. In the ISM group, patients were asked to take three daily profiles per week (each consisting of four scheduled measurements) and were trained on how to respond to SMBG values outside of the target range with changes in diet and exercise. At follow-up visits, changes in medications were made based on SMBG readings, A1c, and hypoglycemic events. The AC group was asked to take three daily profiles in the week prior to month three and month five follow-ups. Changes in medications were based on A1c and hypoglycemic events.
  • There was a significant and sustained improvement in A1c over 12 months in both the intent to treat (ITT) population (n=949) and the per protocol (PP) population (n=553). In the ITT population, A1c at month 12 declined by 0.12% more in the ISM group than the AC group (0.39% in the ISM group vs. 0.27% in the AC group; p<0.05). In the per protocol (PP) population (n=553), this difference was amplified (-0.45% in the ISM group vs. -0.24% in the AC group; p<0.001). The proportion of patients achieving risk targets (using Kovatchev low blood glucose and high blood glucose indices) was not significantly different in the ITT population, but higher in the ISM group in the PP population (90.0% of the ISM group reached risk target vs. 82.5% of the AC group; p=0.038).
  • We note that the study did have a fairly high percentage of patients (42%) that did not follow the protocol. To us, this speaks to the difficulty of encouraging non-insulin treated type 2s to carry out structured testing. On top of that, this was of course a clinical trial setting, meaning real world rates – where patients must pay for their own strips – would likely be much lower. Dr. Bosi’s slides also indicated that 199 out of the 237 participants excluded from the per- protocol analysis in the ISM group were non-compliant with the SMBG regimen. Still, those that did comply saw a larger A1c benefit, and we hope to see more well conducted trials examining structured SMBG in this patient population. Moreover, we wonder how intermittent CGM might play a role in the future given the promising results of the Vigersky study published in Diabetes Care in 2011 (see our report at
  • The frequency of mild hypoglycemia increased in the ISM group, but in most cases was asymptomatic, and the increase in frequency could be a reflection of detection bias (i.e., greater frequency of testing means hypoglycemia is more likely to be detected). Severe hypoglycemia did not increase in the ISM group.
  • The ISM protocol also led to slightly larger reduction in BMI. The difference attained statistical significance in the per-protocol analysis (roughly 0.6 vs. 0.3 kg/m2).
  • There was no difference in quality of life between the groups as measured by a survey adapted from the DCCT DQOL questionnaire. However, the ISM group felt significantly more in control of their diabetes, as compared to the AC group that felt more strongly that factors influencing their control of the disease was more up to chance (as quantified by the locus of control measure utilized in Peyrot et al., Diabetes Care 1994).


Peter Chase, MD (University of Colorado School of Medicine, Aurora, CO); Richard Bergenstal, MD (International Diabetes Center, Minneapolis, MN); Lutz Heinemann, PhD (Science & Co, Düsseldorf, Germany); Helena Rodbard, MD, FACP, MACE (Endocrine and Metabolic Consultants, Rockville, MD); Emanuele Bosi, MD (Vita-Salute San Raffaele University, Milan, Italy).

Dr. Chase: One of my pet peeves is a family picking up a new meter, if they don’t have insurance perhaps it’s the cheapest one at the store, and you can’t download it when they get there. Or when a new meter comes out and the company provides it to patients, but the download program is not yet available, and they haven’t kept a logbook. In the old days everybody kept a logbook. This is a real dilemma but one of my top pet peeves.

Q: I’m a family nurse practitioner and CDE in Columbus, GA where sweet tea is a food group. I do agree with you that we should have a more defined number for hypoglycemia, but at the same time sometimes a patient’s ideal A1c is not 6.5% or 7.0%. I have a patient that feels bad at 75 or 78 mg/dl, so I tell my patients to treat their hypoglycemia at 70 mg/dl or when symptomatic.

Dr. Bergenstal: Life today is all about personalizing, and on each of these reports you just put in what goal you have and move the ideal range around, but I agree that not everyone is even ready to set it at 70 mg/dl. When you’re coming in at 300 mg/dl, 100 mg/dl feels pretty crummy to start with.

Q: I want to make a comment on modal data. I also think it’s the greatest for a 30-second glance. But one error introduced by patients is with outliers, when maybe one day they tested a lot of times. You need some redundancy to ensure the modal day is accurate.

Dr. Bergenstal: There are pros and cons with all of this. Take a quick snap shot but have some other views. I’m fine with that.

Q: What are some of the legal and regulatory hurdles in creating these automated support systems or advisors? We doctors are not living with them and there’s no way that all of us are available all the time.

Dr. Rodbard: I’m not the right person to talk about regulatory or legal aspects. Of course there is always an issue when there is a machine or automated system giving advice. When Dr. David Rodbard [her husband] started working on these kinds of systems 25 years ago, that was my main concern (Pernick et al., Diabetes Care 1986). How are we going to get through the legal hurdles when something that doesn’t know the patient is giving medical advice? Regarding the payers, that is a really big issue we are faced with. Just recently I was told that in state of Oregon, for people with non-insulin dependent type 2 diabetes, insurance allows only one strip per week. That’s ludicrous. I agree with you that regulatory and legal hurdles are major issues. Unless we can show the value for money they are spending for strips and ancillary tools to interpret that data and make it useful – downloading the data, running it through a computer, generating a report, and giving the clinician enough time to digest and interpret the results – it’s going to be an uphill battle. The DECIDE study shows that such tools can improve clinical performance, and indicate that even more intensive training and decision support are both necessary and justifiable.

Dr. Bergenstal: I wanted to ask our friend Courtney Lias (FDA) to make a comment.

Dr. Courtney Lias (Office of In-Vitro Diagnostics, Associate Director of Toxicology, FDA, Silver Spring, MD): Some simple clinical support tools are already out there. We do think pattern recognition can be beneficial to patients. We are hearing more discussion about more advanced clinical support tools including recommendations for dosing and dosing adjustments. We believe there could be value in that. We are seeing a lot more discussion in type 2 than type 1, perhaps because that’s a lower hanging fruit. We do need to see some safety information first. From internal discussions, I can say that people are generally open to the idea.

Dr. Satish Garg (University of Colorado School of Medicine, Aurora, CO): For Dr. Lias while you are here – none of the meters that have been approved have been approved for adjusting the dose of insulin, but all data show that that is what the providers tell the patients to do. So we are really doing this off label.

Dr. Lias: Meters are cleared for the use for self-monitoring. It’s not any more descriptive than that, but obviously they are used for dosing insulin in type 1 diabetes. I wouldn’t call it “off label,” but it hasn’t been studied. Obviously it is used for insulin dose adjustments. The question is around the algorithms. The inquires we are seeing are for prescription, not over the counter types. We’ll talk to companies, but we’re only one step. The studies we need to see are not the same that the payers need to see. I encourage companies to talk to the payers early. Don’t wait until you get on market.

Dr. Bergenstal: It’s such a circular and tough process. Payers won’t pay until they see the studies. Many people are pushing ahead with clinical decision support. I am happy to hear you are open to dialogue for this clinical decision support. It’s giving you a suggestion, but maybe you are more likely to do it. Whereas the artificial pancreas is actually delivering [the dose], [clinical decision support] is one step short of that. So you think it could be possible.

Dr. Garg: Is anyone in the audience on the legal end of this?


Q: I’m a nurse practitioner and CDE. I have a question about downloaded data. We see teenagers that test when they feel like it. What is the usual number of tests that is enough to use your standard deviation and averages to be powerful enough for analysis?

Dr. Bergenstal: The more the better. But if patients are doing fewer tests, we need longer periods of time to accumulate that data. For that automated device I showed you [the Hygieia Diabetes Insulin Guidance System (DIGS)], it won’t give you a dose adjustment each week if you only go one week with very few tests; you’ll have to go two weeks before it gives you an adjustment. If you’re on insulin you need at least four a day.

Dr. Bosi: If you want a number I would say six for type 1 patients on intensive insulin therapy.

Dr. Georgeanna Klingensmith (University of Colorado School of Medicine, Denver, CO): You get what you get. So if it is 2.8 tests per day or 3.2 tests per day, which is the average for a teenager, you have to use that information; you have to do what you have to do. I have gotten to a point where I don’t go beyond 1.0 units/kg or 2.0 units/kg because what if they gave it all one day. They don’t measure their carbs, so you have to do what you have to do.

Dr. Chase: Some people think a teenager goes up in A1c because they are going down in number of blood sugar checks per day. So I sometimes use bribery.

Q: I am a safety and pharmacovigilance physician. From yesterday’s slide on the distribution of diabetes throughout Europe, it was very interesting that Finland was mentioned as number one. I believe maybe one contributor is Chernobyl. I do know that most of us who have been exposed, and I myself was exposed, have autoimmune disease and our offspring have autoimmune disease. If it’s true, as Finland was first to be hit with that radiation, maybe it will be helpful data to Japan in the future.

Dr. Bosi: The etiology of type 1 diabetes in Europe has been studied for many years. The fact that there was a gradient with highest incidence in the north and progressively declining toward the south has been known for many years. Finland had the highest rate (except for Sardinia, which is an exception) already before the Chernobyl disaster. So I don’t know whether the disaster has further increased the incidence of autoimmune diseases, but I don’t see very convincing data around the area. But I don’t think the gradient is a consequence of Chernobyl because it was pre-existing. Why that is, there are several theories, but honestly nobody knows exactly.

Beta-Cell Imaging & Targets for Prevention of Diabetes


George Eisenbarth, MD, PhD (University of Colorado School of Medicine, Aurora, CO); Aaron Michels, MD (University of Colorado Denver, Denver, CO)

Dr. Eisenbarth opened the presentation by noting that while we can quite accurately predict and diagnose type 1 diabetes, we cannot yet safely prevent the disease. He remarked that the insulin peptide is the primary target for the autoimmune response in NOD mice, and that the same was also likely true in humans. He highlighted that a promising approach to prevention may involve specific targeting of the trimolecular complex formed by MHC (a cell surface molecule on antigen presenting cells), a T cell receptor, and the insulin peptide. Following Dr. Eisenbarth, Dr. Michels presented data from preclinical studies that have examined the effects of blocking insulin peptide antigen presentation on type 1 diabetes development. In particular, one study found that by mutating a single amino acid in the sequence of insulin to block the presentation of insulin antigens, diabetes could be prevented in 100% of NOD mice for 60 weeks. Dr. Michels concluded by noting that ongoing work in his lab was attempting to identify compounds that are capable of interacting with the MHC cell pocket that binds the insulin antigen in order to specifically block insulin antigen presentation.

  • Screening for four major autoantibodies (ZnT8, IA-2, GAD65, and IAA) can be very powerful for the prediction of the onset of type 1 diabetes. Only ~5% of children positive for one autoantibody (AA) progress to type 1 diabetes 15 years after initial detection. However,~70% and ~90% of children who are positive for two and three AAs, respectively, develop type 1diabetes within 15 years after detection of the AAs. Additionally, age of onset can be predicted by age of first autoantibody-positive detection and level of insulin autoantibody (IAA).
  • Dr. Eisenbarth noted that no reliable method to date has been identified to prevent type 1 diabetes. Previous studies have examined insulin administration (nasal, oral, and systemic administration); however, none of these trials demonstrated a delay in diabetes onset for more than several months or for a few years. Type 1 diabetes prevention studies currently ongoing include a study examining Macrogenic’s anti-CD3 therapy teplizumab ( identifier: NCT01030861), two studies examining oral insulin administration (NCT00419562 and JDRF’s Pre-Point Study), a study examining nasal insulin administration (NCT00336674).
  • Dr. Eisenbarth highlighted the trimolecular complex formed by MHC, the insulin B:9-23 peptide, and a T cell as a particularly intriguing target for preventing islet autoimmunity. In the NOD mouse model, the MHC cell surface molecule on antigen presenting cells (APCs) presents peptide antigens to T cells, and Dr. Michels’ group believes that thepresentation of the insulin B:9-23 peptide to the T cell is crucial for the initiation of insulin autoimmunity and subsequent islet destruction. Disrupting this process, therefore, may help delay or prevent the onset of type 1 diabetes.
  • Dr. Michels discussed data that showed that the pocket in the MHC molecule, which binds peptides, presents peptides to T cells, and specifically interacts with the insulin B:9-23 peptide had been identified. Mutating one amino acid of the insulin B:9-23 peptide (B16:YA) in NOD mice prevented diabetes development for 60 weeks. This amino acid is important for the T cell recognition of the insulin peptide. Therefore, a screen was performed for small molecules that interact with the pocket of the MHC molecule (e.g. pocket 9). Early studies are currently underway to see whether identified compounds can block insulin antigen presentation and prevent diabetes development in preclinical models.


George Eisenbarth, MD, PhD (University of Colorado School of Medicine, Aurora, CO); Aaron Michels, MD (University of Colorado, Denver, CO)

Dr. Aaron Cypress (Joslin Diabetes Center, Boston, MA): Of the four autoantibodies you discussed, Dr. Eisenbarth, to what extent are they causative and to what extent are they markers of islet cell destruction?

Dr. Eisenbarth: We used to think they were not causative, but in the NOD mouse model, if none of them are present, diabetes doesn’t develop. To pick between the different ones, we don’t have the information to do so, but I think it would be interesting to target insulin autoantibodies specifically to see if their removal changes the rate of progression. Others have no relationship to rate.

Dr. Satish Garg (University of Colorado School of Medicine, Aurora, CO): We’ve always said that high levels of glucose may be the reason cells are dying, but it’s the same reason they make insulin, so is it something else that might be killing the cells?

Dr. Eisenbarth: That’s not the fundamental reason they’re dying. The fundamental reason is that the T cells are directly killing them and that insulin peptides are being recognized in a harmful manner by the immune system. I think the primary reason might well be immunologic, but at edge of this process perhaps we can modulate it a little it.

Dr. Jay Skyler (University of Miami Miller School of Medicine, Miami, FL): Where do you think we go from here and do we have hope to prevent type 1 diabetes?

Dr. Michels: I definitely think we will see type 1 diabetes delayed and prevented in the upcoming years. And that might be with the oral insulin trials already underway. We need safe and specific therapies that can be used by three, four, five, or six year old children.

-- by Kira Maker, Jessica Dong, Adam Brown, Vincent Wu, Ben Kozak, Joseph Shivers, Eric Chang, and Kelly Close