Switching back to ENDO! In a late-breaking session that we didn’t get fully written up prior to ADA, we were keenly interested in hearing a presentation and panel discussion on the recent incretin controversies – also the subject of an NIH conference last month. Dr. David D’Alessio (University of Cincinnati, Cincinnati, OH) gave a brief yet valuable overview of the data on DPP-4 inhibitors, GLP-1 agonists, and pancreatitis. He underscored that while all the studies to date have serious limitations, they support further investigation into this issue. Dr. Steven Kahn (University of Washington, Seattle, WA) discussed the methodological limitations of observational studies, an important topic given that many such studies are being used to evaluate the possible risks of new diabetes drugs. Reviewing the role of the FDA, Dr. Robert Smith (Brown University, Providence, RI) specifically highlighted the agency’s restriction policies, such as risk evaluation and mitigation strategies (REMS) and elements to assure safe use (ETASU). He commented that an ETASU has been applied to rosiglitazone, the use of which has since declined by roughly 95%; given the burdens that these restrictions impose on patients, providers, and industry, Dr. Smith said that severe restrictions should be thoroughly justified. Dr. Robert Vigersky (Walter Reed National Military Medical Center, Washington, DC) examined the response to the recent incretin controversies, and shared the recommendations of both the Endocrine Society as well as the FDA. He noted that the FDA, by recommending incretin-using patients to report adverse events in the MedWatch database, might encourage reporting bias.
Detailed Discussion and Commentary
Late Breaking Session: Controversies in Diabetes
Robert Smith, MD (Brown University, Providence, RI)
After reiterating several times that he was not a spokesperson for the FDA, Dr. Robert Smith gave a short (seven-minute) review of the role of the FDA in drug approval, noting that severe restrictions need to be thoroughly justified, given their influence on drug use patterns and often burdensome nature. Dr. Smith outlined the spectrum of actions the FDA can choose to take on a drug post-approval including requiring a REMS (risk evaluation and mitigation strategy), which may include ETASU (elements to assure safe use). Dr. Smith detailed the possible elements of REMS, which he noted have been applied to incretin drugs because of their association with adverse events such as pancreatitis. He then provided the possible components of ETASU, which vary according to the circumstance and can include required training and certification for the healthcare provider in control of dispensation, limiting dispensation to certain healthcare settings, or patient monitoring. Dr. Smith explained that many of these elements were applied to rosiglitazone, leading to a decrease in use of the drug by more 95 percent. He concluded that regulators should have “serious enough concern” about a drug when implementing these restrictions to justify the far-reaching effects and burden to industry, patients, physicians, and healthcare personnel.
LIMITATIONS AND THE PATHWAY FORWARD
Steven Kahn, MB, ChB (University of Washington, Seattle, WA)
Dr. Steven Kahn presented on some of the limitations of observational studies in the evaluation of treatment risks such as pancreatitis. The limitations include the lack of outcome adjudication, incomplete information regarding covariates, and potential confounding factors. He moved on to somewhat unsettling evidence that many anti-GLP-1R polyclonal antibodies on the market are not sufficiently specific for the GLP-1 receptor in both histological staining and Western blot experiments. He ended on a more optimistic note, mentioning that a number of cardiovascular trials of incretins and other glucose-lowering drugs are nearing completion, and will provide a wealth of patient-years of data regarding the risks associated with these drugs categories. He highlighted the European Commission’s SAFEGUARD study, which includes data drawn from patient health records, systematic reviews of existing studies, and mechanistic intervention studies.
- Observational studies on the risks of diabetes drugs face a number of methodological limitations. In clinical trials, a central panel generally adjudicates outcomes, but this isn’t possible in retrospective observational studies. Unreported covariates and confounding factors such as obesity and alcohol use can also lead to incomplete or false interpretations of the data. Dr. Kahn also discussed the impact of biases, especially stimulation bias. He noted that a doctor who is aware of the possible association between incretins and pancreatic cancer is more likely to report a case of pancreatitis in a patient on incretins than in a patient on another agent.
- Many polyclonal anti-GLP-1R antibodies show low specificity. Dr. Kahn cited a histological staining study in which the polyclonal antibodies LS-A1205 and AP23801 both showed positive staining in cells lacking the GLP-1 receptor, whereas the polyclonal antibody AB39072 failed to stain cells that expressed the receptor (Pyke et al., Endocrinology 2013). He also mentioned a study demonstrating that the polyclonal antibodies SC66911 and LS-A120S showed nonspecific binding in Western blot analyses (Panjwani et al., Endocrinology 2013). Monoclonal anti-GLP-1R antibodies performed better in both studies, indicating that their use should be preferred. We note that Novo Nordisk has developed such a monoclonal antibody, as discussed by the company’s chief medical officer, Dr. Alan Moses, at last month’s NIH meeting onincretins and pancreatic risks. For our coverage of Dr. Moses’ talk, see here.
- A number of studies are currently being conducted on the safety risk of diabetes drugs. Four studies each on DPP-4 inhibitors and GLP-1 agonists will likely be completed in the coming two years. Dr. Kahn also discussed the European-Commission-funded SAFEGUARD trial, which aims to assess the cardiovascular, cerebrovascular, renal, and pancreatic safety of most currently marketed non-insulin glucose-lowering agents. He highlighted the study’s disparate work packages, which draw from databases of adverse events, population-based health care databases, systematic reviews of reported pancreatic side effects, and mechanistic intervention studies in type 2 diabetes patients. Dr. Kahn noted that the findings of the observational work packages face the same methodological limitations he mentioned earlier. Nonetheless, he seemed optimistic about the positive impact of the study, given that it may potentially provide millions of patient-years of data. While these will no doubt be extraordinarily interesting, it won’t take the place of fewer patients being studied over a far longer number of years, which seems to be a bigger and bigger problem associated with CVOTs.
POLICY AND PRACTICE CONSIDERATIONS
Robert Vigersky, MD (Walter Reed National Military Medical Center, Washington, DC)
Dr. Robert Vigersky reviewed the recommendations that the Endocrine Society set forth on March 1, following a Journal of the American Medical Association (JAMA) article about the potential risk of incretin drugs. He emphasized that the Endocrine Society has a job to educate its members and the public about the constant “24/7” flow of news information from the popular and medical press. This rapid response team calls on experts to help the Endocrine Society formulate statements on controversial and breaking news. Dr. Vigersky also described the FDA’s drug safety recommendations on the topic, which were issued two weeks later and included similar themes to the Endocrine Society’s.
- On March 1, 2013, the Endocrine Society made three recommendations in response to a paper linking incretin therapies to pancreatitis (Singh et al., JAMA Intern Med 2013). (In an independent analysis of several large claims databases, Singh and colleagues found that patients prescribed exenatide or sitagliptin had a twofold higher rate of hospitalization for pancreatitis.) The recommendations were as follows: (1) patients should know about potential side effects of incretins and should know the symptoms of pancreatitis; (2) HCPs should consider the potential adverse effects when prescribing these treatments, especially in patients with other risk factors for pancreatitis; (3) HCPs should discourage patients from stopping medication on their own, because doing so could increase the risk of short-term and long-term health complications.
- On March 14, the FDA made two recommendations similar to the earlier recommendations of the Endocrine Society: (1) patients should continue to follow medications as prescribed until they talk to their HCPs, and HCPs should continue to prescribe drugs as recommended on drug labels; (2) patients and HCPs should report negative effects of incretin mimetics to the FDA MedWatch program. Dr. Vigersky noted that this recommendation could bias the MedWatch database, by encouraging a higher rate of reporting by incretin-using patients relative to patients not using incretins.
- Dr. Vigersky thanked the ENDO organizers for putting together a panel so quickly after the NIH conference last Wednesday and Thursday, since so many physicians treating diabetes are using incretin drugs. We are curious to hear the continued dialogue aboutthis issue beyond ADA. Our coverage of the NIH write-up that Dr. Vigersky mentioned can be found at http://www.closeconcerns.com/knowledgebase/r/8326d0dd (Day #1) and https://www.closeconcerns.com/knowledgebase/sites/default/files/document_attachments/2013-06-14%20CC%20-%20NIDDK%20workshop%20on%20pancreatitis%20and%20pancreatic%20cancer%20-%20Workshop%20Day%20%232_1.pdf (Day #2).
William Young, MD, MSc (Mayo Clinic, Rochester, MN); David D’Alessio, MD (University of Cincinnati, Cincinnati, OH); Steven Kahn, MB, ChB (University of Washington, Seattle, WA); Robert Smith, MD (Brown University, Providence, RI); Robert Vigersky, MD (Walter Reed National Military Medical Center, Washington, DC)
Dr. Vigersky: Can you tell us more about what it is involved in pancreatic size measurements? What factors should we be thinking about, and how might these affect our decisions?
Dr. Kahn: Measurements of human pancreases becomes difficult because we cannot put everyone in the same environment. We are able to see differences across mice strains, but it gets very noisy in human autopsies. It becomes even more difficult if you are not standardizing samples. On another note, a paper from 2003 shows that diabetes patients have a reduced beta cell volume of about 15%. In the current paper, the controls had a smaller beta cell volume than those with diabetes in the original paper. This might have to do with the fact that one study puts patients on ventilators while the other uses human autopsy. It is known that there is an increased proliferation of alpha and beta cells in individuals on ventilators. One paper from Brussels additionally showed that individuals on ventilators have increased cell replication compared to those not on ventilators.
A: I think the findings in the literature are provocative, and nothing nags at you like the feeling that you are doing something wrong. To be vigilant about side effects is crucial. The current papers generate more hypotheses instead of generating results that we can actually make decisions based on. It is hard to be patient with the popular press, but we really need to be most concerned about the care of our patients. There is already data being accumulated that will give us a more balanced and reasonable assessments of what is happening in the field.
Dr. Vigersky: I’m concerned that all these studies are placebo-controlled except for one. Are we going to get the information we are looking for? We don’t treat our patients with placebos. That’s the problem with the some of the rosiglitazone studies.
Dr. Kahn: For the studies I showed on my slide, all except one had a placebo group. But it was just the one study that found a difference compared to sulfonylureas.
Dr. Smith: I think that’s an issue that you need both sides for, ultimately. If you’re only doing comparative studies, then you can find yourself in a situation where it’s difficult to determine what the background is upon which the drug is acting. Ideally, one would like to see the effect of the drug without a confounding background, but also something from the real world where drug interactions could interfere. I expect that in the mix of data available, we’ll see both types of data. Also, there is a fair amount of wisdom going into study design right now, partially in terms of how studies are funded. Some of the requests that have been put forward are for pooling and examining data available in the US and elsewhere, from corporate sources, which will help data quality. There is also sensitivity to mistakes that have been made in ensuring the completeness and quality of data that comes from industry studies. I’m optimistic that we’re going to have a huge amount of good data to help inform this question. We have to be careful about when the popular press finds literature and reports on it. I would ask everyone to be cautious when they read these papers, and I would ask the authors of these papers to frame the papers appropriately so that we don’t cause mass fear.
Dr. Vigersky: Let me touch on something that came up earlier and had to do with the rosiglitazone story. What are the kinds of lessons have we learned? What do we know to avoid?
A: That is a hard question. There are hard lessons. As people may know, the week before last there was an advisory committee meeting that reexamined the rosiglitazone issue of CV risk and mortality, and that was following two prior meetings. I think the general view of people in the advisory committee was that the increased risk associated with rosiglitazone is not clear. There does not appear to be really good evidence of increased mortality while taking rosiglitazone, and an increased risk of CV events while taking rosiglitazone is definitely not clear. There was a lot of press and discussion about potentially negative events when perhaps the evidence was not that clear. Ultimately, the FDA made a REMS and ETASU and put limitations on distribution of the drug. Basically, the use decreased very, very dramatically. Each individual here can make a judgment on the appropriateness of that decision. However, now it is extremely difficult to study the drug in controlled trials. Multiple members of the committee, including me, believe that controlled trials are the only way we are ever going to answer the questions of what risks are associated with each drug. Eliminating the ability to do these controlled studies is the trouble with dealing with data that is unclear.
Dr. Kahn: I agree, and I want to go back to the reporting bias. Certain things make much better press. The rosiglitazone story started in the New England Journal of Medicine, and who knows what would have happened if it had started in a smaller journal. We have since seen research that insulin causes cancer. The ORIGIN trial showed no evidence cancer risk. The European agency has looked at all the data and concluded that glargine was not associated with increased cancer risk. We should learn not to rush judgments because, for example, we will lose participants for studies. I would call on the popular and medical press to exercise the correct judgment on what they choose to publish.
Questions and Answers
Q: Is there data being collected on pancreatic cancer in autopsies in diabetics and non- diabetics?
A: I don’t know if there are studies doing that. I think the SAFEGUARD study may try to do that.
Q: One difficulty we have as clinicians is turning down the noise in determining what will be fallacy or reality. The three examples I can think of are ACE inhibitors, rosiglitazone, and phentermine. How do we as clinicians separate out all this noise, when some of it is real early on?
A: You really never know about side effects until you get a million or two million patients out there. We don’t have a robust post-marketing system to collate the millions of exposures that happen when drugs get out there. It can be difficult to have clarity about something like ACE inhibitors and cough, or other myriad effects. One of the strongest biases we have is clinician bias. We remember the dramatic cases and forget the more normal patients. Given that we don’t have a well organized system, we have to look at ways that we can get large data sets that are controlled enough to be interpreted and provide real conclusions. I agree, a sharp clinician will pick this up early on, but it’s hard to do that physician-by- physician, so there needs to be systematic analyses before we can have judgments.
Q: What happened last week at NIDDK, what was the resolution?
A: It was a two-day workshop about pancreatitis, cancer, and diabetes. Both endocrine and exocrine experts were present. The first day was about understanding risks. On the second day, there was a
presentation of evidence from a UCLA diabetes paper, the FDA, and industry, and afterwards a panel discussion. There were clearly a lot of limitations brought up, but they came to the same agreement: that we have hardcore data from CV outcome studies that will be forthcoming over the next 6 to 36 months and will inform us about the risk of pancreatic disease since those are pre-specified endpoints, and that we should support looking at the FDA adverse reporting system. Some limitations were brought up about the paper, such as that it wasn’t an age-matched cohort, and whether some type 1 diabetics were in the study population.
Q: I just wanted to make a comment about reporting bias. I recently reported one of my patients who was on incretins and had pancreatitis to AERS, but I never reported the two patients with pancreatitis who were not on incretins. This presentation reminded me that reporting those two cases is something I should do. I think this example illustrates the limitations of the AERS system, and why the FDA says it shouldn’t be used for epidemiologic studies. Reporting bias is a serious issue.
-- by Hannah Martin, Manu Venkat, Stephanie Kahn, Joseph Shivers, and Kelly Close