American Diabetes Association 74th Scientific Sessions

June 13-17, 2014; San Francisco, CA Day #5 Highlights - Draft

Executive Highlights

Greetings from the final half-day of ADA 2014 in San Francisco, where the sun was brightly shining (sometimes a rarity during the summer here) to bid conference goers a safe return home. As is typically the case on the last day of ADA, the program was fairly sparse, but we did attend a few notable sessions that we’re sharing with you here. We heard striking data, for example, on suggesting that SGLT-1/2 may be expressed pancreatic alpha cells (where conventionally they had not been thought to exist) and may play a role in glucose regulation from there. We also enjoyed a lively debate over the appropriateness of the new ACC/AHA lipid guidelines as well as a session dedicated to the use of big data for public health surveillance of diabetes. Below, we bring you our top five highlights from the day, followed by full details on the most important material at the end of the report.

Top Five Highlights

1. Dr. Caroline Bonner (European Genomic Institute for Diabetes, Lille, France) presented some striking findings on the expression and activity of SGLT-1/2 in pancreatic alpha cells. Her group found that SGLT-1/2 may be expressed appreciably in pancreatic alpha cells, where it might play a role in glucose regulation – this runs counter to the conventional wisdom that the main role for SGLT-1/2 is glucose reuptake in the intestines and kidney. The expression of SGLT-2 mRNA in particular is increased in obese individuals, but paradoxically falls in individuals with type 2 diabetes – in vitro study of human islets indicates that glucotoxicity is a primary cause. Perhaps as a counterregulatory response, the reduction in SGLT-2 expression is accompanied by an increase in glucagon expression. Administering the SGLT-2 inhibitor dapagliflozin (AZ’s Farxiga) in insulin-resistant rodents caused a significant three-fold increase in plasma glucagon. Until now, the prevailing opinion has been that SGLT-2 is only expressed in the kidney, which means that these researchers may have found a new role for SGLT-2. These findings (if confirmed in further study) have significant implications on the best way use the drug class. During Q&A, Dr. Bonner suggested (as other KOLs such as Dr. Ralph DeFronzo’s group have recently) that SGLT-2 inhibitors could be complemented with GLP-1 agonists due to their glucagon-suppressing activity. An attendee cast the findings in a positive light during Q&A, suggesting that SGLT-2 inhibitors’ combined reduction in hyperglycemia and increase in glucagon secretion might be a great match for brittle type 1 diabetes. While we don’t usually use the word “brittle,” we do agree about this and are very eager to watch type 1 SGLT-2 trials move forward.

2. Dr. Robert Eckel (University of Colorado, Aurora, CO) and Dr. Henry Ginsberg (Columbia University College of Physicians and Surgeons, New York, NY) debated the appropriateness of the new ACC/AHA lipid guidelines. Dr. Eckel provided what we perceived as a somewhat defensive argument about how the ACC/AHA’s new lipid guideline is the most evidence-based assessment of how to use cholesterol-lowering drug therapy to reduce atherosclerotic cardiovascular disease (ASCVD) risk (for background on what the new guideline entails, please see bullets below). Our view by the end of his presentation was that the ACC/AHA had good intentions of very rigorously assessing the entire evidence base out there and coming up with a statement that reflects that, but delivered the message poorly in a way that is misleading many providers into thinking that ACC/AHA are actually recommending against using LDL targets or non-statin therapies. Dr. Ginsberg took issue with the ACC/AHA guidelines in a number of areas, although he started by saying that he welcomed the concept of initiating treatment based on patient risk. From our view, he sounded fairly derisive of ACC/AHA’s claim that these guidelines were strictly evidence based – or the implication that prior guidelines were not – and showed many examples of how the new guidelines used expert opinion, rather than the strict application of trial data to argue that the new guidelines were not as evidence-based as the authors tried to bill it as. He was also disappointed by the removal of specific targets for LDL cholesterol lowering and no mention of other therapies. He showed evidence that the lower the LDL (by whatever means), the lower the risk of a cardiovascular event. Following this maxim, he noted that if statins are not appropriate or fail to reach goal, then there is no clear guidance on how to go further and help the patient – particularly in the case of people with diabetes.

3. Dr. Christel Hero (Institute of Medicine, Gothenburg, Sweden) presented a large study whose results call into question the predictive value of LDL-cholesterol for cardiovascular risk in people with type 1 diabetes. This finding challenges the use of LDL as a marker for cardiovascular disease in this population. As a result, Dr. Hero concluded that there is “no support” for the treatment target of 100 mg/dl. In its place, Dr. Hero’s team proposes substituting the ratio of total cholesterol to HDL-cholesterol. During Q&A, noted cardiologist Dr. Robert Eckel (University of Colorado, Denver, CO) indicated that the low predictive value of LDL is not particularly surprising, though he challenged the notion that the ratio of total cholesterol to HDL cholesterol would serve as a useful predictor of cardiovascular disease. Still, Dr. Eckel congratulated the researchers on a study that furthers our understanding of atherosclerosis in type 1 diabetes.

4. In a morning symposium on the use of big data in diabetes surveillance, Dr. Eric Larson (Group Health Cooperative, Seattle, WA) addressed the challenges for big data utilization. While earlier speakers in the morning symposium addressed specific functionalities of big data (e.g., real-time monitoring capabilities using Electronic Health Records, social media as a diabetes monitoring tool, and big data’s role in diabetes pharmacovigilance), Dr. Larson focused on the hurdles that big data still faces before it can be successfully integrated into diabetes surveillance on a large scale. He highlighted that most of the controversy on diabetes surveillance concerns societal and ethical challenges, including patient privacy, consent requirements, and data ownership/hoarding. Dr. Larson emphasized that “together, all stakeholders – including the general public – must better understand that allowing researchers reasonable access to routinely collected health data will improve health and healthcare for all.” According to Dr. Larson, “the benefits of big data in improving public health surveillance are already evident” – he highlighted current US big data projects for conducting better and more cost-effective clinical research and sharing public health knowledge. Dr. Larson did note that PCORnet has several groups focused on diabetes research and data. In addition to the ethical concerns, technical challenges including data interoperability, incomplete data sets, and regulations must be resolved. Oversight policies and practices must be developed in order to eliminate these barriers to sharing information and maximizing big data’s full potential. He also urged all stakeholders (patients, advocacy groups, physician organizations, academic centers, and industry) to “overcome cultural impediments based on outdated ideas” about the collection and use of everyday health care information because put simply, “there is tremendous ability to move a lot faster than we’ve ever moved before.”

5. Darius Lakdawalla (University of Southern California, Los Angeles, CA) indicated that one in every two Medicare dollars is spent on a person with diabetes. This figure represents a pretty substantial increase from one in three dollars in 2004. Given that overall Medicare spending has increased in this time, this means that people with diabetes are now taking up a larger slice of a bigger pie. Regarding Medicaid, the program’s expansion will likely lead to greater spending on diabetes as more people who are uninsured gained coverage. Had Medicaid not been expanded, Dr. Lakdawalla estimated that spending on people with diabetes would have roughly doubled. Medicaid’s expansion under the ACA adds an additional $15 billion to that. We understand that the hope is that in providing uninsured people with access to care for diabetes, that they will use the emergency room less for already advanced complications and cost Medicare less once they reach age 65. We’ll be back with our full coverage of this presentation in our full report.

 

Detailed Discussion and Commentary

Oral Presentations: ADA Presidents Oral Session

The Glucose Transporter SGLT 2 Is Expressed in Human Pancreatic Alpha Cells and Is Required for Proper Control of Glucagon Secretion in Type 2 Diabetes (386-OR)

Caroline Bonner, PhD (European Genomic Institute for Diabetes, Lille, France)

Dr. Caroline Bonner’s presentation revealed some striking new findings about SGLT-1 and -2 expression and activity in the pancreas. Her group found that SGLT-1/2 mRNA is found to an appreciable extent in pancreatic alpha cells, where it could be serving as a glucose sensor or playing some other role in endocrine signaling. The expression of SGLT-2 mRNA in particular is increased in obese individuals, but paradoxically falls in individuals with type 2 diabetes, apparently due to glucotoxicity. When Dr. Bonner’s team inhibited SGLT2 in human islets using siRNA or dapagliflozin, it led to a substantial stimulation of glucagon expression. Additionally, administering dapagliflozin to an insulin-resistant mouse model caused a significant three-fold increase in plasma glucagon. Until now, the prevailing opinion has been that SGLT-2 is only expressed in the kidney, and these findings (if confirmed in further study) have significant implications on the best way use the drug class. During Q&A, Dr. Bonner suggested (as other KOLs have recently) that GLP-1 agonists might be a complementary partner for SGLT-2 inhibitors due to their glucagon-suppressing activity. An attendee cast the findings in a positive light, suggesting that SGLT-2 inhibitors’ combined reduction in hyperglycemia and increase in glucagon secretion might be a great match for brittle type 1 diabetes.

  • Two high-profile studies published earlier this year (Merovci et al. & Ferrannini et al., JCI 2014) demonstrated that SGLT-2 inhibitors led to significant increases in plasma glucagon and hepatic glucose production. We heard Dr. Ralph DeFronzo (University of Texas Health Science Center, San Antonio, TX) discuss this point at CODHy Latin America earlier this year (read our report). During that talk, Dr. DeFronzo suggested that this effect could mean that SGLT-2 inhibitors should be paired with a GLP-1 agonist, which could blunt the increase in glucagon – Dr. Bonner echoed this point during Q&A.
  • The current view of SGLT-2 inhibitors largely depends on the view that SGLT-2 transporters are only expressed to an appreciable extent in the kidney. As a result, Dr. Bonner noted, the expression and activity of the transporter elsewhere in the body is poorly understood.
  • Through immunofluorescence staining, Dr. Bonner’s group showed that SGLT-1 and SGLT-2 co-localize with glucagon/alpha cells in pancreatic islets. Further studies using cell purification showed that SGLT-1 and SGLT-2 mRNA is enriched in alpha cells in particular – while expression in beta cells is relatively low or nonexistent.
  • The expression of SGLT genes is deregulated over the course of the progression into type 2 diabetes. Using cross-sectioned samples from human islets, Dr. Bonner’s group showed that SGLT-1 and SGLT-2 mRNA expression increases as individuals progress into obesity and glucose intolerance. However, interestingly, SGLT-2 mRNA expression drops off dramatically as patients progress from glucose intolerance to diagnosed type 2 diabetes. Glucagon mRNA expression, in contrast, spikes substantially as patients progress from glucose intolerance to type 2 diabetes. The drop-off in SGLT-2 expression is remarkable, as SGLT-2 expression in the kidney is significantly up-regulated as patients progress to type 2 diabetes. When Dr. Bonner’s group exposed human islets in culture to varying levels of glucose, it appeared that the drop-off in SGLT-2 expression could be linked with glucotoxicity.
  • Both a knockdown in SGLT-2 using siRNA and the SGLT-2 inhibitor dapagliflozin (AZ’s Farxiga/Forxiga) stimulated glucagon gene expression. Dr. Bonner stated that the two means of blocking SGLT-2 activity also caused a increase in SGLT-1 expression, but in our view it was harder to conclude that from the data. The mechanism by which an SGLT-2 inhibitor might reduce SGLT-2 mRNA is not immediately clear; generally, counterregulatory mechanisms would cause increased expression of a gene if the protein it codes for is being inhibited. During Q&A, Dr. Bonner somewhat cryptically suggested that the action of SGLT-2 inhibition in the pancreas could be at the transcriptional level.
  • In a preclinical model of insulin resistance, Dr. Bonner’s team found that dapagliflozin administration led to an increase in plasma glucagon. The mouse model used was a C57blk lineage treated with S961 peptide. The increase in plasma glucagon (ostensibly from pancreatic alpha cells) was roughly three-fold, and was statistically significant.

Questions and Answers

Q: These results raise the questions of whether SGLT-2 inhibitors might be useful in type 1 diabetes, especially brittle type 1 diabetes. Aside from using them to blunt hyperglycemia due to the glycosuric effect, perhaps it could blunt hypoglycemia by increasing glucagon expression.

A: We are doing studies now in STZ models – that’s something we’re looking into.

Q: Regarding therapeutics, if in theory this is a prominent mechanism, you would expect to see more hypoglycemia with SGLT-2 inhibitors, but if anything you see a reduction in glucose. Does that suggest that this intriguing interaction with islets has less therapeutic impact than the class’ effect in the kidneys?

A: We have to wait and see, but I do think that this drug class is very good. Perhaps it might not be used best as a standalone therapy. It might be better used in combination with a GLP-1 analog to combat the oversecretion of glucagon.

Q: Why was it that you saw relatively low expression of SGLT-1 and SGLT-2 in lean individuals, but then the levels were much higher in obese patients, as were levels of glucagon? In healthy individuals there is no connection between SGLT and glucagon expression.

A: In the kidney, SGLTs are very well characterized as glucose transporters. We are not sure what the function is in the pancreas. Perhaps it is a glucose sensor, or is involved in endocrine signaling. We do know that expression is reduced in type 2 diabetes due to an overload of glucose.

Q: Dapagliflozin is meant to be an inhibitor of the SGLT-2 protein, but in vitro you saw reduced SGLT-2 mRNA expression. How can you explain that?

A: There is some evidence in our lab suggesting that SGLT-2 is inhibited at the transcriptional level.

Q: Generally, one of the strongest determinants of glucagon secretion is insulin secretion. The results we saw in the JCI articles were unexpected, as there was an increase in glucagon but no increase in insulin. Did you study the effect of changing insulin levels in the growth media?

A: We haven’t studied the effect of decreasing insulin concentration in the cell media.

Q: Is there anything known about glucagon levels in humans that have loss-of-function mutations for SGLT-2?

A: I do not know that off the top of my head.

LDL-cholesterol Is Not a Good Marker of Cardiovascular Risk in Type 1 Diabetes: Observational Study in 30,778 Patients—A Report from the National Diabetes Register in Sweden (381-OR)

Christel Hero, MD (Institute of Medicine, Gothenburg, Sweden)

Dr. Christel Hero presented a large study (n=30,778), which calls into question the predictive value of LDL-cholesterol for cardiovascular risk in people with type 1 diabetes. The study drew records from the Swedish national diabetes registry, the hospital discharge register, and the cause of death register using each patient’s national ID number. Mean follow-up was 7 years and outcomes included both fatal and non-fatal cardiovascular events. The results show that LDL levels are, at best, a very weak predictor of cardiovascular disease in patients with type 1 diabetes. This finding challenges the use of LDL as a marker for cardiovascular disease in this population. As a result, Dr. Hero concluded that there is “no support” for the treatment target of 100 mg/dl. In its place, Dr. Hero’s team proposes substituting the ratio of total cholesterol to HDL-cholesterol. During Q&A, noted cardiologist Dr. Robert Eckel (University of Colorado, Denver, CO) indicated that the low predictive value of LDL is not particularly surprising, though he challenged the notion that the ratio of total cholesterol to HDL cholesterol would serve as a useful predictor of cardiovascular disease. Still, Dr. Eckel congratulated the researchers on a study that furthers our understanding of atherosclerosis in type 1 diabetes.

  • The study included 30,778 patients with type 1 diabetes. For the study, records were drawn from the Swedish national diabetes registry, the hospital discharge register, and the cause of death register using each patient’s national ID number. Mean follow-up in the study was seven years and outcomes included both fatal and non-fatal cardiovascular events.

Question and Answer

Dr. Robert Eckel, MD: Just a comment, atherosclerosis in type 1 may be a whole different process. The fact that you found that LDL is not a great predictor is really not surprising. With regards to using the ratio of total cholesterol to HDL cholesterol, levels of HDL cholesterol have already been reported to be higher in type 1 diabetes. You may be including HDL that may not be as functional; it might be glycosylated or have other modifications. I still think this is an outstanding study just to support the concept that atherosclerosis in type 1 is not the same animal – it may be more fibrotic and it may be more calcified.

Q: When you talk about using lipid-lowering medications, were only statins used?

A: Because of the way medications are entered into the registry we cannot tell.

Q: For the subpopulation on lipid lowering treatment – do we know their LDL values before they were prescribed medication?

A: No, we do not.

Current Issues: Are the New ACC/AHA Guidelines for Lipids Appropriate for Diabetes?

Pro

Robert Eckel, MD (University of Colorado, Aurora, CO)

Dr. Robert Eckel provided a somewhat defensive argument about how the ACC/AHA’s new lipid guideline is the most evidence-based assessment of how to use cholesterol-lowering drug therapy to reduce atherosclerotic cardiovascular disease (ASCVD) risk (for background on what the new guideline entails, please see bullets below). Our view by the end of his presentation was that the ACC/AHA had good intentions of very rigorously assessing the entire evidence base out there and coming up with a statement that reflects that, but delivered the message poorly in a way that is misleading many providers into thinking that ACC/AHA are actually recommending against using LDL targets or non-statin therapies.

  • Dr. Eckel heavily emphasized that guidelines are not meant to dictate and restrict, but to support and guide and that the ACC/AHA guideline does not mean that providers can no longer use LDL goals or other LDL-lowering medications besides statins, but just states that there has yet to be any evidence that these strategies are beneficial (in part because there has not been an incentive yet to design a trial to prove that using LDL targets is beneficial).
  • For background, he reminded the audience that this process began five years ago when the NHLBI set out to update the outdated ATP-III, but when the NHLBI announced it would step away from making clinical guidelines, ACC/AHA stepped up to the plate to finish the work they had started – Dr. Eckel made this statement in defense of the mounting criticism ACC/AHA have received for the new guideline, suggesting that they were simply carrying out a mission they had been charged with and that had already undergone substantial internal and external review when the NHLBI was working on it.
  • He briefly showed some of the evidence used to inform the guideline – notably the Cholesterol Treatment Trialists group has published evidence showing benefit of statin therapy independent of cholesterol level, and benefits across a wide range of patient demographics and disease types (Lancet 2010). He acknowledged that the 7.5% 10-year ASCVD risk cut off was somewhat arbitrary, but based on evidence that even a 5% risk would warrant intervention.
  • For background, the AHA/ACC guideline recommends treating people for high cholesterol using statin therapy if they fall into one of four risk categories of people most likely to benefit from statin therapy: (i) people with existing atherosclerotic cardiovascular disease (ASCVD – e.g., stroke, coronary disease, aortic aneurysm, peripheral vascular disease, etc.); (ii) LDL ≥190 mg/dl and age ≥21 years; (iii) primary prevention with diabetes, aged 40-75 years and LDL 70-189 mg/dl; (iv) primary prevention without diabetes but with 10-year risk score of ≥7.5% (using the new risk calculator), aged 40-75 years, LDL 70-189 mg/dl.
    • The guideline also states that no evidence exists for using certain LDL targets, so it simply recommends that you put patients that meet any of the above criteria on a statin and monitor for adherence. The previous LDL treatment paradigm had been centered around meeting certain LDL goals. Dr. Eckel emphasized that goal setting in clinical practice may be very useful, but there just haven’t been trials conducted to prove that.
    • Another major point of the new guideline is that non-statin therapies have not been proven to provide ASCVD risk reduction benefits or safety profiles comparable to statin therapy. Again, Dr. Eckel stated that if someone still needs residual LDL-lowering after going on a statin, it is of course good clinical practice to try another LDL-lowering drug, but the evidence just does not exist yet to show that this is beneficial.

Con

Henry N. Ginsberg, MD (Columbia University College of Physicians and Surgeons, New York, NY)

Dr. Henry Ginsberg took issue with the ACC/AHA guidelines in a number of areas, although he started by saying that he welcomed the concept of initiating treatment based on patient risk. However he was strongly derisive of ACC/AHA’s claim that these guidelines were strictly evidence based - or the implication that prior guidelines were not – and showed many examples of the use of expert opinion, rather than strict application of trial data. However, his definition of applicable trial data was the randomization of the exact risk groups, rather than meta-analyses. Meanwhile, he was disappointed by the removal of specific targets for LDL cholesterol lowering and no mention of other therapies. He showed evidence that the lower the LDL-C (by whatever means), the lower the risk of a cardiovascular event. Following this maxim, he noted that if statins are not appropriate or fail to reach goal, then there is no clear guidance on how to go further and help the patient – particularly in the case of people with diabetes.

  • Dr. Ginsberg welcomed the stratification of patients by risk when determining whether to prescribe a statin. The new ACC/AHA guidelines divide patients into four groups based on risk (clinical atherosclerotic cardiovascular disease (ASCVD), patients with LDL cholesterol (LDL-C) >190 mg/dl, people with diabetes and LDL-C between 70-189 mg/dl, but no existing ASCVD), and give guidance on the appropriate intensity of treatment with statins, but no targets for LDL-C.
  • The new ACC/AHA goals are supposed to minimize expert opinion and be fully evidence-based, but Dr. Ginsberg believes that this is not true. He presented extensive evidence (for each of the four risk groups) that the randomized controlled trials considered by the committee did not randomize patients at the specified level of risk at baseline – so the recommendation is necessarily based on meta-analyses.
  • Dr. Ginsberg gave his own recommendations for each of the four groups:
    1. For patients with existing ASCVD, we should treat them all with high intensity statins (not just the under 75).
    2. For patients with LDL-C >190 mg/dl, “common sense and integration of a large body of data from numerous sources say it is OK to treat” with statins.
    3. Diabetes is associated with higher event rates, so Dr. Ginsberg would treat with high intensity statins, since there is no evidence that high intensity statins are not just as good and their safety has been established. He stated, “I wouldn’t differentiate based on primary or secondary prevention”.
    4. Dr. Ginsberg noted that that the evidence is just as good for 5% as for 7.5%, but he was “OK with the recommendations”.
  • Since in Dr. Ginsberg’s view, there were plenty of exceptions and inconsistencies in the use of the data, he was disappointed that the Expert Panel couldn’t find evidence for the use of targets (for LDL or non-HDL). He stated, “…All the evidence we know shows that any way to lower LDL reduces cardiovascular disease, irrespective of the exact mechanism of LDL lowering”. He also went on to posit that “lowering LDL more is better than lowering LDL less” – implying that it makes sense to set goals.
  • Dr. Ginsberg also wanted guidance on what to do if high dose statins didn’t succeed in lowering LDL below 110mg/dl – presumably other drugs should be used to reach a target, but non-statins are rejected. He favors the ADA guidelines, which do set targets. He believes that there should be non-statin choices for LDL >70 mg/dl. High triglycerides and low HDL-C should be approached with diet and exercise, but the possibility of combinations to treat severe dyslipidemia should be supported (fibrates and possibly niacin).

Rebuttal

Robert Eckel, MD (University of Denver, Aurora, CO)

Dr. Robert Eckel rebutted Dr. Ginsberg’s arguments that the ACC/AHA guideline was not quite as evidence-based as it claims to be. His points boiled down to the idea that where the evidence does not yet exist, one must then exercise expert clinical judgment. For example, in response to the criticism that the ACC/AHA does not have clinical trial evidence to support the recommendation that everyone with LDL ≥190 mg/dl should be on a statin, Dr. Eckel said, “When you reach the need for a heart transplant or liver transplant, you don’t need a trial to tell you that you need the transplant,” suggesting that an LDL of ≥190 is so farcically high that it is a no brainer to treat it.

Panel Discussion – Selected Questions and Answers

Q: Are we setting ourselves up for never having data, because we can’t design the studies to answer the questions rigorously? In other words, are we going to stop using parachutes because there is no randomized controlled trial?

Dr. Henry Ginsberg, MD (Columbia University College of Physicians and Surgeons, New York, NY): There are two current trials of non-statin drugs that lower LDL. Hopefully people will be getting these drugs on top of statins, [in the trials], but if you take the strict use of the term ‘evidence-based’ there may not be extrapolation to the trials that say a lower LDL is better than a high HDL. Now the Panel has put themselves into a corner.

Dr. Robert Eckel, MD (University of Colorado, Aurora, CO): On the Panel, we were discouraged from expert opinion, since this is the current rule of the land. But the guidelines will be updated on a regular interval based on trial evidence, (although it’s not clear how they support this process). Dr. Ginsberg discards data based on the stringency of the trial design and that’s a concern. In the clinic what I recommend is that you document that the guideline has been applied to the patient and beyond that you then have a lot of room to do other things. If you are in primary care, you can get consultation about subsequent steps. In fact, 80% of the decisions we make in the clinic go beyond the current guidelines.

Q:  We have still avoided the whole problem of why people with diabetes have more atherosclerosis than others. It has been demonstrated that LDL and HDL are modified by glycation. That may be a link with atherosclerosis in the diabetic patient. Why hasn’t there been more attention to that process as opposed to just pushing the LDL-lowering lower and lower?

Dr. Ginsberg: I think we had three large trials that looked at lowering A1c, all of them I think showed difficulty of doing that in large clinical trials. Maybe they were all in patients who were irreversible but none of those trials in which A1c went from 8.3% to 7% or 7.4% to 6.4% showed benefit, and of course the ACCORD trials showed some detriment. I think glycated LDL and HDL exist and relate to control. Certainly people with very high A1cs may suffer some consequence. The in vivo applicability is still less certain than it was in mice but I think we have failed at proving in a late stage cohort of diabetics that lowering their A1c reverses their disease over the next several years. It’s been very disappointing, as you know.

Q: What happens in patients with high triglycerides. Should we use fibrates?

Dr. Ginsberg: If you look at fibrate trials, the monotherapy trials are even more positive for high triglycerides, even those trials that were negative overall.

Dr. Eckel: Imagine we have a patient with 6.9% A1c and triglycerides of 350 mg/dl, and LDL around 75-80 mg/dl on an appropriate dose of statin. For that patient I would add a fibrate and would tell her about the trial evidence. This again goes beyond the guidelines.

Q: Before I agree with Henry’s point about targets I do want to say these guidelines are tour de force of examining the evidence. I really compliment the committee. Having been on a guideline committee, my heart goes out to members of the guideline committee – it is thankless, and no matter what you come up with, it is often criticized often both ways for opposing reasons. Having said that, in Canada in 2009 when we came up with a target LDL of 2.0 mmol/L, we were totally aware it was not evidence based. There was no question about that we understood it, but it was a public health recommendation that was very easy for people to follow. Sometimes I think there can be a tyranny of evidence or evidence-based policy that actually ends up worse off. I think you could take your evidence and make it very user friendly. I think that’s one problem – the translation of the evidence to make it user friendly for the practicing clinician.

Dr. Eckel: The process wasn’t a simple one, and we kept getting reminded that trials were not designed to reach 100 mg/dl or 70 mg/dl and ultimately the committee in consensus voted for the guidelines as they stand. Practicality is problematic. I see nothing wrong with setting goals – if you want to interpret evidence to set a number. I still set goal with my patients, but keep in mind to realize that the goal is not in line with the guideline.

Q: The guidelines mention lifetime risk, but the calculators are strongly weighted to age – so should we wait for young people to get older for the risk to increase before we treat them?

Dr. Eckel: Lifetime risk is so appropriate, especially for women, and the Framingham score doesn’t look at lifetime risk.  There is never going to be a study that examines it, so we never are going to have the data. We should treat now, rather than wait in certain circumstance.

Dr. Rob Ratner, MD (American Diabetes Association, Alexandria, VA): We need to use common sense and use the guidelines as recommendations. Unfortunately we deal with organizations that don’t always apply common sense and yet pay for coverage. Have we seen denial of service for the use of lipid measurement as quality indices yet? [A few people raised their hands]. But it is recommended that we measure lipids three months after starting statins and annually. Goal setting behavior is not wrong, if you feel it’s appropriate then just include it in your note. And if decisions are going beyond the guidelines you should be fine (for reimbursement).

Q: I just want to focus on the young adult type 1 diabetic with 10-year diabetes duration. They are often ignored. We discussed this yesterday in the new ADA position statement on type 1 diabetes, and I don’t feel this is reversing now. We presented some data from the EDC and Scottish registries that show that risk for a type 1 younger than 40 years with 20 years diabetes duration is very close to that 7.5% cut-off – about 6.9% 10-year risk and after that the 7.5% cut point – statin therapy. Could you comment on whether a type 1 diabetes young adult with 20-year disease duration in their 30s really should be on moderate intensity therapy? My other question is why blood pressure trials have goals but none of the lipid trials do. I’ve never understood why blood pressure is treated so differently from lipids in terms of trial design.

Dr. Ginsberg: In type 1s, it comes down to taking a gestalt view of everything we know about atherosclerosis and multifactorial disease, but LDL cholesterol and other atherogenic lipoproteins are critical. I always talk to students about Japan 30-40 years ago when everyone’s LDL was 60 mg/dl, and everyone smoked two packs a day, etc. but there was no coronary disease. If you have other risk factors for atherosclerosis, having lower LDL is beneficial. So you have to have a leap of faith or a leap of knowledge to make some of those decisions. But should a 10-year-old with five-year disease duration be on a statin? If you did lifetime risk where would they be? I think the biggest conundrum is the 15-year-old with type 2 diabetes, and should they be on an ACE or statin? In type 1 with prolonged duration, I think they should be on a statin.

Dr. Eckel: I think the disease may be somewhat different from type 2 diabetes, and that needs careful scrutiny with ways to study it in terms of basic science and clinical trials. In type 1 I just haven’t heard. In America there are 1.5 million type 1s so it’s not a trivial disease, and we clearly need more data. In terms of numbers of hypertension and not lipids I agree; I have the same problem. I maintain numbers in my clinic because they’re useful but I defend the guideline as it stands because that’s what the committee decided was most evidence-based.

Q: In the PCSK9 trials they are going to analyze quartiles and tertiles of LDL, which might prove that targets make a difference.

Dr. Ginsberg: I imagine that there are pre-specified analyses in terms of LDL achieved, but even if that’s done, the evidence based purists wont abide because it’s not a randomized sub-group.

Q: As a foreigner, can we apply these guidelines to other ethnic groups? I know it would be great if we made or own guideline based on our own RCTs but we don’t have enough numbers of RCTs, and we can’t afford to do that.

Dr. Eckel: Part of the approach of these guidelines was to use trials that have a global population. There was no restriction of the US or North America. The entire world participated in these clinical trials. Keep in mind the recommendations for African Americans and Caucasians were based on populations in this country that were black or white. I think you can with comfort use the risk estimator for our population and just realize unfortunately there’s not enough evidence for people from certain populations – they haven’t been represented specifically enough in clinical trials. But I think you can be comfortable using the calculator for black or white and just know that it’s not that far off.

Dr. Ginsberg: I think there’s one important thing about this question – in Asians there are clearly differences in pharmacokinetics in the metabolism of statins. Much lower doses appear to be as beneficial in terms of LDL-lowering and also higher doses appear to be associated with more adverse effects, particularly myopathy and liver enzyme elevations.

Q: What is the official position of the guideline committee on follow up, success, failure, what to do?

Dr. Eckel: Follow up should be from 3-12 months to assess adherence. It is not to set goals but if you’re setting goals, you can do that. The guidelines said no assessment of lipids going forward. I was routinely measuring lipids in people every six months and getting back to them. That’s probably not necessary. You don’t see people’s lipids really change if they remain on the same dose of a statin. The problem that needs to be assessed is adherence.

Dr. Robert Ratner (Chief Scientific & Medical Officer, American Diabetes Association): Our practice committee is discussing whether to change ours to meet the standards of the ACC/AHA or continue using our target-based standards that go beyond LDL and incorporate non-HDL and triglycerides. By a show of hands, how many of you feel as though the ADA guidelines need to become consistent with the ACC/AHA [very few hands are raised]. How many of you believe that we should continue with the targets? We may modify them [majority of hands are raised].

Symposium: Using Big Data for Public Health Surveillance of Diabetes

Controversies and Future Directions in Using Big Data for Diabetes Surveillance

Eric Larson, MD, MPH (Group Health Cooperative, Seattle, WA)

Dr. Eric Larson (Group Health Cooperative, Seattle, WA) closed the big data symposium with a presentation on the remaining challenges in diabetes surveillance. He highlighted that most of the controversy on diabetes surveillance concerns societal and ethical challenges, including patient privacy, consent requirements, and data ownership/hoarding. Dr. Larson emphasized that “together, all stakeholders – including the general public – must better understand that allowing researchers reasonable access to routinely collected health data will improve health and healthcare for all.” According to Dr. Larson, “the benefits of big data in improving public health surveillance are already evident” – he highlighted current US big data projects for conducting better and more cost-effective clinical research and sharing public health knowledge (NIH’s Health Care Systems Research Collaborative, the FDA’s Mini-Sentinel surveillance system for post-approval safety of medical products and drugs, eMERGE Network for combining EMRs and genomic network data, and the National Patient-Centered Clinical Research Network [PCORnet]). Wider acceptance of big data in diabetes surveillance remains to be seen, though Dr. Larson did note that PCORnet has several groups focused on diabetes research and data. In addition to the ethical concerns, technical challenges including data interoperability, incomplete data sets, and regulations must be resolved. Dr. Larson noted that oversight policies and practices must be developed in order to eliminate these barriers to sharing information and maximizing big data’s full potential. He also urged all stakeholders (patients, advocacy groups, physician organizations, academic centers, and industry) to “overcome cultural impediments based on outdated ideas” about the collection and use of everyday health care information because put simply, “there is tremendous ability to move a lot faster than we’ve ever moved before.”

  • Dr. Larson highlighted current projects in the US using big data to conduct better and more cost-effective clinical research and share public health knowledge.
    • The NIH Health Care Systems Research Collaborative seeks to engage health care organizations as research partners in large-scale studies and translate this research into clinical systems.
    • The “phenomenal” Mini-Sentinel project, a pilot project sponsored by the FDA to create an active surveillance system to monitor the safety of FDA-regulated medical products post-approval. It was established in 2009. So far, Mini-Sentinel has codified more than 126 million data sources from health insurers, integrated delivery systems, and research goals. This codification process and its firewall protective capability will be critical for future networks to engage with big data and get patient approval.
    • The National Patient-Centered Clinical Research Network (PCORnet) is “the biggest new kid on the block,” as it essentially creates a “super highway” for patient-centered discovery and clinical research. Currently, 29 Clinical Data Research Networks (CDRNs) and 50 Patient-Powered Research Networks (PPRNs) are included in the PCORnet. This massive “network of networks” exemplifies the opportunity to leverage shared data sets for implementing clinical trials and engaging patients in research. Notably, patient cohorts in Tennessee, Louisiana, and New York are focused on diabetes research and data.
  • Citing a 2013 JAMA paper, “The Inevitable Application of Big Data to Healthcare,” Dr. Larson emphasized that leveraging the collection of patient and practitioner data could improve the quality and efficiency of health care delivery. In his opinion, the first revolution in modern medicine was the digitalization of medical records, and we are on the cusp of the next revolution: big data. Big data is essentially “continuously generated information,” a concept that Dr. Larson believes could expand the capacity of research, help disseminate knowledge, and importantly, transform healthcare by delivering the benefits of research to patients’ bedsides. Additionally, big data can help with coordinating care, targeting services to the most at-risk populations, and identifying eligible groups for research. Notably, Dr. Larson pointed out that these big data applications have the potential to reduce healthcare costs, particularly in research as studies are easily targeted and populations can be identified within their regular clinical care.

Using Big Data for Diabetes Pharmacovigilence

Nicholas Tatonetti, PhD (Columbia University, New York City, NY)

Dr. Nicholas Tatonetti provided a high level explanation of its use in pharmacovigilance. Currently, 50% of primary care practices are using an EMR, and 75% of all patient encounters are captured within an EMR. As a result, there is more than one petabyte (1015 bytes) of healthcare data in the US alone. This wealth of information is beginning to be tapped, using big-data-analysis techniques. Dr. Tatonetti indicated that two of the key challenges of big data in this area are uncharacterized biases (i.e., unknown confounding variables) and missing data. These obstacles can be overcome with advanced statistical tools, including statistical correction of uncharacterized bias (also known as SCRUB) and latent signal detection, respectively.

  • As an example of the power of big data in pharmacovigilance, Dr. Tatonetti told the story of how he unearthed a diabetes-related drug-drug interaction between paroxetine (a depression drug) and pravastatin. Over one million people in the US are on both of these drugs, and when Dr. Tatonetti surveyed a healthcare database for a blood glucose interaction, this combination was the top hit. While neither paroxetine nor pravastatin affect blood glucose levels as a monotherapy, the observational data indicated that together, they raised blood glucose levels by 18 mg/dl (p<0.001) in healthy people and by 60 mg/dl (p<0.05) in people with diabetes (Tatonetti et al., Pharmacology & Therapy 2011). This finding has been confirmed in rodent models; Dr. Tatonetti did not indicate if it is being tested in humans. 

 

-- by Adam Brown, Hannah Deming, Jessica Dong, Katherine Sanders, Jenny Tan, Manu Venkat, and Kelly Close