We’ve seen this show before. Based on material we reviewed, the scientific case against the GLP-1 drugs has not been adequately documented and conclusions about harm set forth by Dr. Peter C. Butler appear to be premature.
This post was drafted by David H. Schwartz, a principle at Innovative Science Solutions, and John Clark, MD,MSPH, who provides expertise to ISS in litigation involving pharmacovigilence and adverse event reporting. Dr. Clark has consulted with corporations and defense counsel on mass tort and complex litigation for the past three decades. He has worked with individual clients and industry groups on management of some of the country’s largest and most prominent pharmaceutical issues. Dr. Clark is the president and chief medical officer at PCSglobal, a pharmaceutical, biotechnology and medical device industries consulting company specializing in risk-reduction services.
The medical community has been embroiled in a heated debate on diabetic therapies for the better part of the last two decades. The most recent controversy involves a class of diabetes drugs called incretin-based therapies or glucagon-like peptide 1 (GLP-1)-based medications, a topic covered in a recent article in The New York Times.
We’ve seen this show before. Based on material we reviewed, the scientific case against the GLP-1 drugs has not been adequately documented and conclusions about harm set forth by Dr. Peter C. Butler appear to be premature. In this post, we provide a brief summary of the controversy, and describe the problems that we see with the evidence purportedly documenting a health risk from the drug class.
In an upcoming post we use will use this topic to illustrate the steps that should be taken to properly evaluate the available scientific and medical evidence in any drug or device case.
Detractors of the incretin drug class assert that GLP-1 agents can induce acute pancreatitis, perhaps, chronic pancreatitis, and ultimately increase the risk of pancreatic cancer. Skeptics argue that the benefits of the GLP-1 drugs outweigh any risks due to these adverse events (AEs), and that the data demonstrating harm are preliminary and potentially biased.
These arguments are nicely laid out in a pair of dueling editorials published ahead of print in the journal Diabetes Care: an article by Peter Butler and colleagues sets forth the case that GLP-1 therapies pose an undue health risk, while an article by Michael A. Nauck takes the position that the benefits outweigh the potential risks.
Problems With Dr. Butler’s Position
To be sure, Dr. Butler lays out data that is relevant to a comprehensive assessment of the safety of GLP-1 drugs. However, we identified a number of problems with his analysis:
1) Cherry Picking the Data
In Table 1, Butler and colleagues compile and summarize animal studies of GLP-1-based therapies. According to the text, Butler and colleagues interpret these data as supporting the view that a clear safety signal is present. However, they do not provide a methodology for how they selected these studies. This omission indicates the possibility that they focused on studies that support their perspective. Even if an acceptable selection methodology was used, the data are too inconsistent to support his thesis. A proper evaluation of this evidence would involve a systematic set of inclusion/exclusion criteria, followed by a systematic analysis of all the available data.
2) Inappropriate Use of Metrics
Table 2 compiles AE reports from GLP-1-based therapies, which were forwarded to the FDA, and compares them to controls using the FDA Adverse Event Reporting System (FAERS). Butler reports the results of this analysis using Odds Ratios (ORs). However, true ORs cannot be calculated based on this type of spontaneously reported data. It would have been more appropriate to report the findings as a signaling analysis based on either Proportionate Reporting Ratios (PRRs) or Reporting Odds Ratios (RORs). While this may be more a technical point, it highlights the liberties that Butler is willing to take to support his position.
3) Confounding by Indication
Butler lists a number of reasons why diabetics are prone to pancreatitis –factors that are referred to in epidemiology as “confounding bias” or confounders. Confounders are positively associated with both a drug exposure and an AE, and can therefore be responsible for the AE independent of any contribution from the drug. In this case, having a diabetic pancreas could be responsible for pancreatitis in a diabetic population. Thought of in another way, since the diabetic pancreas does not function properly, it is not surprising that diabetic products can be associated with diseases of the pancreas in either reporting systems such as FAERS, or in more formal epidemiological studies.
Butler’s discussion illustrates a type of confounding bias which is somewhat peculiar to studies of risk from medical products. Specifically, whenever a drug is prescribed for a specific disease state (diabetes) and that disease state is also associated with an AE (pancreatitis), the AE can just as reasonably be attributed to the disease state as to the drug. This phenomenon is commonly referred to as “confounding by indication.” Therefore, finding indication-related adverse outcomes in treated patients is not unexpected, and does not necessarily indicate that the medical product was responsible. Butler purports to control for these factors in his analysis, but effective control of confounding by indication is rarely possible in FAERS data.
4) Controlled Epidemiological Data Inconsistent with FAERS Data
The most striking defect in Dr. Butler’s assessment of GLP-1 drugs and acute pancreatitis is provided in Figure 1 from Dr. Nauck’s article. Rather than using FAERS data — for which adjustment for confounders is difficult or impossible — Dr. Nauck reviewed published epidemiological studies which are much less prone to biased results and allow adjustment for at least some confounders. It is quite interesting that in six published studies using typical epidemiological designs for exenatide, the true odds ratios for acute pancreatitis clustered around 1.0 (the no effect value), none of the odds ratios for exenatide exceeded 2.0 (a risk ratio cited by some as a threshold for potential causality), and none of the p-values were <0.05. Likewise, in five published studies of sitagliptin and acute pancreatitis, there was a similar array of odds ratios in the 1.0 – 1.5 range, and the p-values for these comparisons were all <0.05. It should be noted that positive epidemiological findings involving odds ratios between 1.0 and 2.0 are notoriously unreliable, and, by themselves, are not considered evidence in favor of a drug-AE link.
In his write-up, Butler does not address why the spontaneously calculated RORs from FAERS show elevated levels of disproportion, while epidemiological results using superior methodology essentially show no effect. As noted above, the most likely explanation is the inability of Butler to adjust his FAERS ROR statistics for confounding by indication.
5) Failure to Perform Benefit-Risk Evaluation
Nowhere in his review does Butler compare the benefits of the GLP-1 agents to their potential adverse health effects, a methodology known as Benefit-Risk Evaluation (BRE). Current AE assessment practices in the pharmaceutical industry are based on BREs, and such assessments must always be carried out when proposing a significant AE for a drug. Health risks cannot, and should not, be evaluated in isolation of a product’s benefits.
6) No Assessment of Company Data
Finally, we have no sense of what the company data showed relative to the safety of these drugs. There were large data packages submitted to the FDA in support of the safety of these compounds. This data was not properly reviewed and evaluated by Dr. Butler to arrive at a satisfactory conclusion about GLP-1 safety.
After reviewing Butler’s analysis and the relevant scientific data, we agree wholeheartedly with his conclusion: “The case presented here does not prove that these agents are unsafe…”
Did you like this post? We are here to help! Schedule a consult with one of our experts.
Just fill out the form bellow and we will contact you with more information.
On Wednesday July 8, 2020, Dr. David Schwartz of Innovative Science Solutions presented at the IADC 2020 Virtual Annual Meeting on a panel titled The Use of Genetic Testing in the Courtroom. A complimentary copy of the panel presentation is now available for download. Read more
Dr. David Schwartz of Innovative Science Solutions will be presenting at the IADC 2020 Virtual Annual Meeting on a panel titled The Use of Genetic Testing in the Courtroom. Read more
Plaintiff experts having been asserting for decades that all mesotheliomas must be linked to some asbestos exposure. Indeed, this has led to the erroneous (but widespread) view that mesothelioma is a signature disease, only caused by asbestos exposure. Read more
The journey to scientific and commercial success is often complex and always critical, if you are looking for an expert partner to help steer you to confident solutions, contact us todayContact us