June 28, 2016; Rockville, MD; Full Report – Draft

Executive Highlights

Hello from Rockville, MD, where the FDA’s Endocrinologic and Metabolic Drugs Advisory Committee meeting on a potential label update for Lilly/BI’s SGLT-2 inhibitor Jardiance (empagliflozin) wrapped up yesterday. On the committee’s agenda was the question of  whether an additional indication for a cardiovascular mortality benefit should be added to the label for both  Jardiance and SGLT-2 inhibitor/metformin fixed-dose combination Synjardy (empagliflozin/metformin) based on results of the EMPA-REG OUTCOME study. See below for our top five highlights from this nail-biter of a meeting, as well as our coverage of the Open Public Hearing and each panelists’ rationale for their voting decision.

1. The meeting ended with a very close 12-11 vote in favor of an additional indication for Jardiance based on the CV mortality benefit demonstrated in EMPA-REG OUTCOME. As expected, the panel rendered a unanimous “yes” verdict on the first question asking whether the trial had successfully established cardiovascular safety. On the question regarding CV benefit, multiple highly respected voices voted positively, though NIH’s Dr. Judy Fradkin did not.

2. At the crux of much of the debate was whether a single trial designed to demonstrate cardiovascular safety for a 3-point MACE primary endpoint could support a label claim of benefit for the secondary endpoint of cardiovascular death reduction. Although death is a secondary endpoint, the p value is very strong.

3. The FDA and the panel also raised concerns about the lack of clear data on “silent MIs” and non-assessable deaths in EMPA-REG OUTCOME. This reinforced to us the importance of standardization in clinical trials – naively, we had assumed more of this had been established than actually has. Overall, although there is an issue over CV deaths, because many were not demonstrated conclusively to be CV deaths, we put forward our view that this does not matter because all-cause death was similarly reduced. In this circumstance, we believe the absence of a clear pathophysiological mechanism is not important – a death is a death is a death, whatever the mechanism.

4. Several committee members suggested the lack of a clear mechanistic explanation for the cardiovascular benefit seen in EMPA-REG OUTCOME may draw into question the validity of the findings. However, the cardiovascular death curve separation starts so early and is so clear that we doubt that ultimately the results will be questioned. Other CVOTs and perhaps even Big Data could be very helpful over time. We do note that many don’t like not understanding / not having clear explanations – this is probably particularly true for doctors, though we do not have an RCT to prove it.

5. Voting members largely agreed that the reductions in heart failure and adverse renal outcomes shown in EMPA-REG OUTCOME were not sufficient to establish conclusive benefits. Of course the indication for glucose-lowering has been the main indication – depending on what happens with FDA, even someone with healthy glucose may go on empagliflozin, purely for cardiovascular risk reduction! It will be interesting to see where the discussion goes in terms of the dramatic renal results just reported.

Top Five Highlights

1. The meeting ended with a very close 12-11 vote in favor of an additional indication for Jardiance based on the CV mortality benefit demonstrated in EMPA-REG OUTCOME. As expected, the panel rendered a unanimous “yes” verdict on the first question asking whether the trial had successfully established cardiovascular safety. The CV mortality question proved more contentious than we had originally anticipated, though today’s debate was consistent with the relatively negative tone of the FDA’s briefing documents released last week (that’s the last time we’ll listen to anyone who tells us we are over-reacting … ). While panelists agreed that the mortality results were impressive and potentially enormously meaningful for patients, many also echoed the FDA’s concerns about the borderline significance of the results for the primary endpoint, the lack of a clear mechanism of benefit, and flaws in the trial design.  This puts the FDA in an interesting position going forward. While we expect a label update based on the positive vote (albeit nominally positive), based on the discussion we would not be surprised if the FDA took a more neutral approach – that said, we also assume that they will want to show that the billions of dollars invested in the CVOTs have been useful, particularly given these unmatched mortality results. While it would be terrific to see the mortality benefit replicated in additional studies of Jardiance or other SGLT-2 inhibitors, another CVOT anytime soon is a lot to ask. Lilly/BI’s planned heart failure studies for Jardiance may provide the best opportunity to replicate the results, as we assume they will include cardiovascular death as a key endpoint – but within the group that is being discussed (those at high risk of heart disease), who wants to wait that long.

• To us, this discussion was an excellent illustration of both the pros and cons of the FDA’s 2008 CV Guidance. On the one hand, Jardiance’s cardioprotective effects might never have been revealed without the requirement to conduct a CV safety study, as there were few incentives for companies to invest in cardiovascular outcomes trials for type 2 diabetes drugs in the pre-2008 era. On the other hand, many of the concerns raised during today’s discussion stemmed from the fact that EMPA-REG OUTCOME was primarily designed to evaluate safety rather than efficacy. Based on the experience of the past seven years, we would love to see the FDA move toward a more case-by-case approach to the issue of CVOTs for diabetes drugs: companies could be required to conduct a safety trial for products with a signal of increased risk and required or incentivized (though, how?) to conduct a superiority trial for products with potential CV benefits. We do think many of the questions that this this trial raised should have been decided beforehand and we do believe it’s up to FDA to standardize. We also point out that there is still enormous demand for easy to prescribe and easy to use / take compounds and this oral drug addresses that demand.
• From a limited number of discussions with KOLs, we believe the composite could be called into question by the FDA (the components move in different directions, the Kaplan-Meier curves separate then coverage) and as several panelists shared, the p – value is not as high as many would like (the upper CI is very close to 1.00). That said, although death is a secondary endpoint, the p value is very strong.

2. At the crux of much of the debate was whether a single trial designed to demonstrate cardiovascular safety for a 3-point MACE primary endpoint could support a label claim of benefit for the secondary endpoint of cardiovascular death reduction. Dr. William Hiatt (University of Colorado School of Medicine, Aurora, CO) in particular expressed discomfort with the lack of a clear hierarchical analysis leading from non-inferiority and superiority for 3-point MACE to superiority for CV death. More specifically, several speakers pointed out that the trial was barely significant for superiority for the primary MACE endpoint (p=0.038) and suggested that the Advisory Committee meeting would not even be occurring if the p-value had been slightly higher. We believe 0.05 is the level of “significance” for a reason – while we would understand this argument is the p-value were 0.049, we think it’s a bit of a stretch to demand (effectively) a p value of 0.01. A number of panelists said they felt that they had to believe that the superiority finding for the primary MACE endpoint was valid in order to even consider the validity of the CV death risk reduction – we were a bit surprised by this. NIDDK Director Dr. Judith Fradkin, in particular, noted that she wasn’t entirely convinced of the validity of the primary outcome and has always viewed positive secondary outcomes as hypothesis-generating. In addition, Dr. Peter Wilson (Emory University School of Medicine, Atlanta, GA) noted in his voting commentary that it’s difficult to design a single trial that successfully addresses both safety and superiority – that’s for sure, eight years and billions of dollars and hundreds of millions of out of control patients later!  Dr. Daniel Budnitz (CDC, Atlanta, GA) echoed those sentiments, stating that while he felt that EMPA-REG OUTCOME was excellently designed as a safety trial, it was not ideally designed to demonstrate cardiovascular benefit. While we of course respect these views, we think it’s a lot to expect “ideal” design.

• On the other hand, multiple other committee members felt that CV death is arguably the most clinically meaningful and the “hardest” of the three outcomes within the 3-point MACE composite and that a reduction in CV death was worth seriously considering regardless of the primary endpoint outcome. Several members pointed out the impressively low p-value (0.00003) and large number of events (n=309) for CV death, arguing that the benefit was likely valid on those grounds (this felt like a bit of an understatement). Dr. Michael Proschan (NIAID, NIH, Bethesda, MD) was particularly convinced by the highly significant p-value and the fact that the significance was retained even after strong adjustments for multiplicity. Dr. Marvin Konstam (Tufts University School of Medicine, Boston, MA) also found it reassuring that extremely comparable CV death benefits were seen with two different doses of empagliflozin (10 mg and 25 mg), suggesting that this in essence is like having two separate CVOTs rolled into one.
• Highly regarded statistician Dr. Stuart Pocock characterized the cardiovascular death results as providing “overwhelming evidence of benefit” and “proof beyond reasonable doubt” that the benefit is valid. At the behest of BI, Dr. Pocock spoke to the highly significant p-value (0.00003) for absolute cardiovascular death rates in the empagliflozin vs. placebo groups, emphasizing the rarity of such great statistical significance. He demonstrated how this p-value remained highly significant even after Bonferroni corrections for the 10 pre-specified adjudicated cardiovascular outcomes (p=0.0003) and the 42 pre-specified secondary outcomes (p=0.0012). We found this quite notable.

3. The FDA and the panel raised concerns about the lack of clear data on “silent MIs” and non-assessable deaths in EMPA-REG OUTCOME. Silent MIs (diagnosed based on ECG measurements) were not included in the primary endpoint, and a secondary analysis found that the benefit for the primary endpoint was no longer statistically significant when those events were included. During the morning presentations, BI asserted that excluding silent MIs is common practice in CVOTs; the FDA acknowledged that many trials have not included them but suggested that the EMPA-REG OUTCOME investigators had not been consistent about their approach across all protocols and statistical analysis plans. This issue was a central point of discussion in the morning Q&A session but did not appear to have a huge influence on the final vote.  Yet again, this is another area where we’d like to see standardization and hopefully we’re headed to that point given the 2014 AHA/ACC guidelines “2014 ACC/AHA Key Data Elements and Definitions for Cardiovascular Endpoints Events in Clinical Trials”. More directly relevant to the CV mortality question was the issue of “non-assessable” or “presumed cardiovascular” deaths. In its briefing document and morning presentation, the FDA noted that 40% of CV deaths in EMPA-REG OUTCOME were categorized as non-assessable, meaning they were presumed to be due to cardiovascular causes but a direct cause could not be ascertained. This is standard practice in cardiovascular trials, but the FDA and a number of panelists expressed concern about the unusually high proportion of such deaths in EMPA-REG OUTCOME – some felt that this called the overall conduct of the study into question. That said, the risk reduction for CV death and all-cause mortality remained significant even when non-assessable deaths were excluded, which seemed to assuage at least some of the panelists’ concerns, reasonably, in our view. Overall, although there is an issue over CV deaths because many were not demonstrated conclusively to be CV deaths, this does not matter because all-cause death was similarly reduced. In this circumstance, we believe the absence of a clear pathophysiological mechanism is not important – a death is a death is a death, whatever the mechanism. The mechanism question arises for CV death, because many CV deaths were not undoubtedly CV deaths – there were questions on some.

4. Although multiple committee members concluded that there was clear cardiovascular benefit based on the mortality results (even if they can’t be explained), several committee members suggested the lack of a clear mechanistic explanation for the cardiovascular benefit seen in EMPA-REG OUTCOME may draw into question the validity of the findings. We would say that the findings have been questioned ever since the results arose – that doesn’t mean, in our view, that fewer patients didn’t die! In summing up the committee’s comments, Chairperson Dr. Robert Smith (Brown University, Providence, RI) noted that the fact that the cardiovascular mortality benefit was not presented with an accompanying plausible mechanism made it that much more difficult to confidently back the validity of the results. Drs. Melissa Li-Ng (Cleveland Clinic, OH), Paul Palevsky (University of Pittsburgh, PA), and David Cooke (Johns Hopkins University, Baltimore, MD) all cited the uncertainty surrounding the mechanism of benefit as reasons for their vote against an updated indication – this is not particularly patient-friendly in our view given the data on CV death. In particular, Dr. Li-Ng stated that, as an endocrinologist, she struggled with the idea of discussing a potential cardiovascular mortality benefit with her patients without a mechanistic explanation to back it up – we saw this irony as profound, overall, since from the beginning of time, patients have wanted to do better and to experience fewer complications and multiple other classes are challenging to address. Dr. David Good (Penn State, Hershey, PA) also expressed some concerns about the unknown mechanism of benefit in his voting commentary, though he ultimately voted in favor of approval – that made a lot of sense from our view. Practically since the moment the full results for EMPA-REG OUTCOME were revealed almost a year ago, the diabetes world has been abuzz with speculation over the mechanism of benefit, with the main hypotheses related to the diuretic, hemodynamic, or neurohormonal effects of empagliflozin. We were particularly intrigued by Dr. Ele Ferrannini’s fuel energetics hypothesis discussed at ADA 2016. Critically, however, these potential explanations are only hypotheses at this stage and it’s a common refrain at conferences that EMPA-REG OUTCOME has generated more questions than answers. From our view, that doesn’t mean the results aren’t valid – we think it’s a bit questionable that panelists would call into question results that they merely couldn’t respond to adequately.

5. Voting members largely agreed that the reductions in heart failure and adverse renal outcomes shown in EMPA-REG OUTCOME were not sufficient to establish conclusive benefits. Despite apparent reductions in both complications, most panelists agreed that these exploratory endpoints were not collected and adjudicated rigorously enough to warrant any definitive conclusions. BI was not, of course, seeking a label revision to reflect either of these outcomes at this stage, though the kidney results were only recently announced and it is impossible to forecast what BI/Lilly will be doing on this front. While panelists expressed interest in future investigations designed specifically to probe for benefits on heart failure and renal function, especially given the preliminary suggestion from EMPA-REG OUTCOME that Jardiance could have protective effects on these endpoints, we felt it would be quite a lot to ask for a second outcomes trial at this stage. Lilly/BI have already announced plans to conduct two dedicated heart failure trials for Jardiance and while it is certainly possible that we will see a dedicated renal outcomes trial in the future as well, we are not positive on this. Of course the indication for glucose-lowering has been the main indication – depending on what happens with FDA, even someone with healthy glucose may go on empagliflozin, purely for cardiovascular risk reduction! It will be interesting to see where the discussion goes in terms of the dramatic renal results just reported.

• Dr. Peter Wilson (Emory University, Atlanta, GA) championed the need for updated heart failure metrics in clinical trials. He argued that the field’s approach to quantifying heart failure in clinical trials must evolve and suggested hospital days and changes in outpatient medication dose as potential measures to kick-start the conversation. Heart failure has emerged as an important endpoint in CVOTs for diabetes drugs (with the signal of benefit in EMPA-REG OUTCOME and signal of increased risk in SAVOR and EXAMINE), and we expect that future trials will make an effort to evaluate it more rigorously and consistently.

Detailed Discussion and Commentary

Open Public Hearing

• Dr. George Grunberger, representing AACE, shared his thoughts on the three main questions he has been inundated with since the full EMPA-REG OUTCOME results were released: (i) is the benefit a class effect?, (ii) should all patients with type 2 diabetes be on an SGLT-2 inhibitor or at least on empagliflozin?, and (iii) when will AACE change its diabetes algorithm to reflect the EMPA-REG OUTCOME results? Dr. Grunberger provided rather conservative answers to each of these questions overall – AACE typically does not advocate strongly for or against a position, but his presence was notable. He said that we simply don’t know at this point if the effect seen in EMPA-REG OUTCOME is a class effect, as the CANVAS and DECLARE CVOTs for J&J’s Invokana (canagliflozin) and AZ’s Farxiga (dapagliflozin), respectively, won’t report for several years – while this is true, we didn’t feel this was part of today’s debate. Dr. Grunberger also argued that it’s premature to suggest all patients with type 2 diabetes should be on empagliflozin or an SGLT-2 inhibitor based on a single trial in which only one component of the primary endpoint was significant for superiority – we also don’t think anyone was suggesting this be the case. He pointed out that all participants in EMPA-REG OUTCOME notably already had a history of cardiovascular disease and he would at most recommend widespread empagliflozin usage in patients that matched the trial’s inclusion criteria – given that this was the question at hand, this is the part of the commentary that is most pertinent. Notably, Dr. Grunberger suggested that post-marketing data and real-world evidence will help determine if a broader recommendation of empagliflozin usage in all patients with type 2 diabetes is justified – it would have to be quite a lot of “Big Data” in our view to come to this conclusion. Finally, Dr. Grunberger noted that AACE doesn’t feel the need to update its guidelines as it already highly recommends SGLT-2 inhibitors as a second-line therapy after metformin and highlighted the note in the latest update that SGLT-2 inhibitors may offer a “possible benefit” on cardiovascular outcomes – once again, AACE is ahead of its time! Although someone without a lot of knowledge on the field may look at Dr. Grunberger’s take on the EMPA-REG OUTCOME results as a bit of a reserved one, Dr. Grunberger provided a valuable broad contextual overview of the link between diabetes and increased cardiovascular risk in his speech and emphasized that AACE recognizes the need for new drugs and indications to better support patients with diabetes. Overall, Dr. Grunberger’s position appeared to reflect that he was impressed by the EMPA-REG OUTCOME results but not quite ready yet to recommend broad changes in current prescribing practices for the entire class – this seemed very reasonable to us.
• Speaking on behalf of The diaTribe Foundation, our very own Ms. Helen Gao (Close Concerns, San Francisco, CA) remarked on patients’ stake in the addition of Jardiance’s cardioprotective effects to the label. Recalling the historic moment of seeing the EMPA-REG OUTCOME results unveiled to the sound of thunderous applause at EASD, Ms. Gao remarked that the average patient has likely not heard of these exciting results, and may not be fully aware of the worrisome connection between diabetes and cardiovascular complications. An indication of Jardiance’s cardioprotective benefits on its label would therefore be instrumental in encouraging conversations between patients and their physicians regarding cardiovascular health. Quoting the musical Hamilton, Ms. Gao explained that diabetes patients and others in the field often feel “outgunned, outmanned, outnumbered, and outplanned” in the face of the diabetes epidemic. Publicizing Jardiance’s cardioprotective effects on its label would assure health care professionals and patients that they have another tool against diabetes and its cardiovascular complications. See Ms. Gao’s full remarks here.
• Our own Ms. Emily Regier (Close Concerns, San Francisco, CA) provided commentary on how the addition of a cardiovascular indication to Jardiance’s label could add value for diabetes patients and their providers. She noted that the EMPA-REG OUTCOME results came as a staggering and hopeful surprise to the diabetes community and that this type of cardioprotective outcome would have a significant impact on the healthcare system even if it only applies to a select patient population. Ms. Regier acknowledged that the methodological and statistical subtleties of the trial may be imperfect but stressed the importance of not letting these details overshadow the real lesson from EMPA-REG OUTCOME: the importance of discussing and being attentive to the cardiovascular risks associated with diabetes. Indicating Jardiance’s cardioprotective benefits on the label would be an important impetus for such discussions between patients and providers, and would ensure that the massive investments of time and money into this trial produced as much value as possible. See Ms. Regier’s full remarks here.

Commentary from Voting Members

Following the final panel vote, each panelist was given a few minutes to explain his or her decision. Included below are summaries of each statement for the voting question of whether EMPA-REG OUTCOME provides substantial evidence of cardiovascular mortality reduction. The comments are split by “yes” and “no” votes and arranged in alphabetical order by speakers’ last names. The panel vote was incredibly close – 12-11 in favor of approval of an additional indication for cardiovascular death benefit. The other voting question of whether the EMPA-REG OUTCOME results met the bar to demonstrate cardiovascular safety for empagliflozin was met with a unanimous “yes” vote – or, as Dr. Proschan put it, “yes, because DUH!”

Yes

• Dr. Leslie Cho (Cleveland Clinic, Cleveland, OH) offered one of the strongest yes votes on the panel, describing the CV mortality results as “very convincing.” She emphasized how unusual it is to see such a clearly positive result from a trial conducted under the 2008 Guidance; she also noted that all the cardiologists on the panel had voted yes. She also suggested that it may be time for the field to move away from its traditional focus on only atherosclerotic endpoints in cardiovascular outcomes trials. She acknowledged the uncertainty around the mechanism but argued that what really matters is the number of deaths, “the actual bodies on the ground.”
• Cardiologist Dr. James de Lemos (University of Texas Southwestern, Dallas, TX) voted yes, swayed by the robustness of the cardiovascular mortality finding and the strongly significant reduction in all-cause mortality. Despite this, he suggested that the reported size of the effects was likely an overestimate due to misclassification of some events.
• Dr. David Good (Penn State Medical Center, Hershey, PA) cited the “reasonably robust” results for cardiovascular death in his vote in favor of approval. That said, he noted that he would have voted against an indication on the primary endpoint and expressed concerns about the unknown mechanism driving the benefit.
• Dr. Brendan Everett (Brigham and Women’s Hospital, Boston, MA) was a tentative yes vote, pointing to the importance of mortality as an endpoint but strongly encouraging Lilly/BI to confirm the results in a second trial. He concurred with the concerns about the robustness of the findings for the primary endpoint, but he framed those results primarily as a means to “get through the door” to look at the individual components. He described the effect on mortality as “clear and substantial” and pointed to the high number of events and low p-value as drivers of his decision to vote yes. That said, he expressed some hesitation about the fact that the results have not been replicated and indicated that he “would not object” to waiting for more studies before updating the label. We felt this view reflected some lack of experience in the field – the 2008 Guidance did not cover what would be needed for “benefit” although updated Guidance covering at least standardization would be helpful in our view.   
• Dr. William Hiatt (University of Colorado School of Medicine, Aurora, CO) said that he found himself compelled by the sensitivity analyses demonstrating a sustained cardiovascular death benefit for empagliflozin to vote in favor of approval. He noted that he had initially had some concerns about the EMPA-REG OUTCOME data, especially with regards to the results for the primary MACE endpoint, but he found the evidence presented by the multiple sensitivity analyses for cardiovascular death hard to argue with. He emphasized that, despite issues with the trial design, he ultimately believes that the results for cardiovascular death are valid and that he’d “hate to see it denied a claim uniquely on how the trial was run.” As he is particularly highly respected and has significant experience on the EMDAC front, we take Dr. Hiatt’s views especially seriously.
• Dr. Marvin Konstam (Tufts University, Boston, MA) shared conflicted views, ultimately voting yes although agreeing at least in part with the rationale of his colleagues who voted no. “The vote is very representative of the back-and-forth that’s been going on in my mind all day long.” He justified his affirmative vote with three critical pieces of information – an extremely small p-value for CV mortality, the high number of cardiac events, and his opinion that EMPA-REG OUTCOME was really two trials in one due to the inclusion of two dose groups (while we would not say there was broad agreement on this point, we also point out that there isn’t clear guidance on what is required to shown benefit – this is clearly grounds for question among voting members).
• Patient representative Dr. Richard Lumley voted yes, stating that he has been able to keep his type 2 diabetes under control due to medications like empagliflozin and that he believed the drug is safe and effective. While this may appear a bit of a simple explanation, his speech did reflect the fact that there is not at this stage significant downside advocating that patients at particularly high risk of CV disease consider taking the drug, excluding cost.
• Dr. James Neaton (University of Minnesota, Minneapolis, MN) was a strong positive vote, explaining simply that EMPA-REG OUTCOME demonstrated an “extreme difference on a very clinically relevant outcome,” and noting that he found the scientific analysis to be rigorous and compelling.
• Biostatistician Dr. Michael Proschan (NIAID, Rockville, MD) voted yes, expressing his confidence in Jardiance’s benefit for cardiovascular mortality. He acknowledged that the 38% risk reduction reported in EMPA-REG OUTCOME is likely an overestimation due to sampling bias, but noted that he would have been compelled by even a 20% risk reduction. We found this particularly persuasive.
• Dr. Morris Schambelan (UCSF, San Francisco, CA) voted yes though like some other panelists, expressed uncertainty. He noted that he had swung back and forth throughout the day but was ultimately impressed enough with the impact on this “most important outcome” to vote yes. He emphasized that the results should only be applied to the specific population studied in the trial and jokingly wished the FDA good luck in deciding how to interpret the panel’s very conflicted input.
• Endocrinologist Dr. Abraham Thomas (New York University Langone, New York, NY) voted yes, expressing hope that a cardiovascular indication may increase the likelihood that insurers will cover Jardiance. He found the all-cause mortality data especially convincing, remarking that this benefit is something patients deserve to know about. We found his vote particularly important as he has led several EMDAC meetings historically and is especially highly respected.
• Dr. Susan Yanovski (NIH, Bethesda, MD) emphasized the impressive magnitude of the results for CV death and all-cause mortality, as well as the large size of the multi-center study. While she acknowledged the limitations imposed by the fairly large proportion of non-determinable deaths in EMPA-REG, like a number of others, she was reassured by the sensitivity analysis showing statistically significant reductions in CV death and all-cause mortality even after excluding these deaths.

No

• The CDC’s Dr. Daniel Budnitz felt that the cardiovascular mortality benefit is likely benefit but shared a number of concerns with the trial that led him to vote against an indication. He argued that the EMPA-REG OUTCOME trial did not meet the FDA’s criteria for single-trial indication updates. He suggested that the trial was excellently designed to demonstrate cardiovascular safety but not superiority and the results were not highly reliable as the “non-assessable” CV deaths were not clearly described (this is true although it does not affect all-count mortality). Furthermore, while he acknowledged the impressively low p-value for cardiovascular death, he noted that methodological issues drew the statistical strength of the analysis into question (BI/Lilly’s biostatistician expert himself did not appear to agree). Finally, he emphasized that while it’s clear that mortality is an important clinical benefit, it’s unclear if the EMPA-REG OUTCOME results are generalizable to a larger US type 2 diabetes population given the trial’s rather narrow inclusion criteria. We found this a bit frustrating since in our view, a positive vote has nothing to do with how generalizable the analysis are – we would assume at this stage that most panelists would accept that the analysis is not generalizable beyond the population that was studied. This is, of course, still a sizable number of people who could benefit from the knowledge – which also has nothing to do with the vote. 
• Pediatric endocrinologist Dr. David Cooke (Johns Hopkins University, Baltimore, MD) voted no, stating that it is risky to let EMPA-REG OUTCOME stand alone to demonstrate a cardioprotective effect for Jardiance. He admitted that the data are very suggestive of a benefit, but he argued that it is unwise to make such sweeping conclusions based on a single trial, especially given our complete lack of knowledge of the mechanism underlying the cardiovascular benefits. We believe that perhaps the 2008 Guidance must be re-assessed – we think in this age of Big Data acquisition, we would be very disheartened to see two CVOTs required. Also, “unwise” to make “sweeping” conclusions sounds odd for a multi-year, multi-thousand person trial.
• Highly respected NIDDK Director Dr. Judith Fradkin noted that while she felt there was substantial evidence of a cardiovascular mortality benefit, ultimately trial design and methodology issues led her to vote no. She was impressed by the number of events, the magnitude of reduction, the sensitivity analyses, and the fact that the cardiovascular mortality effect was sustained over time. That said, she strongly felt that a positive secondary outcome could only be hypothesis-generating at best (this was a bit surprising to hear given that the secondary outcome was death) and wasn’t convinced of the validity of the primary outcome given the methodological issues in the study previously discussed. While she felt that an additional dedicated study is needed for a label update, she did emphasize that doctors will see the EMPA-REG OUTCOME data and draw their own conclusions. Given the highly contentious legal environment, we don’t believe this is likely to be the case with all US doctors. As a side note, she stated that she believes a neutral cardiovascular effect in other SGLT-2 inhibitor CVOTs would not necessarily preclude a true cardiovascular benefit for empagliflozin, given the differences in selectivity among the products in the class.
• Consumer representative Ms. Diana Hallare attributed her reluctance to the high proportion of undetermined deaths and the controversy over silent MI. Echoing other panelists who voted no, she highlighted the need for a second trial to provide more clarity. Ms. Hallare also asked for a follow-up investigation that addresses ethnicity, as the EMPA-REG OUTCOME participant pool did not show much diversity. We very much agree with her on the ethnicity point, although we see this as a separate very important discussion – it should be no means be the sole driver of the decision in our view, but there should be a motivating factor related to the decision.
• Epidemiologist Dr. Susan Heckbert (University of Washington, Seattle, WA) argued that the CV mortality data – though intriguing and hopeful – are not sufficient to warrant an indication on the label. She expressed particular concern over the fact that CV mortality was not the primary endpoint and emphasized that the FDA typically requires converging evidence from two separate studies before adding indications to a drug label. Given the p value for death, we would characterize the data as beyond intriguing and hopeful – as for two studies, we would like to see FDA discuss this with manufacturers since we can’t imagine there would be funding for all companies to do two studies to show benefit.
• Dr. Melissa Li-Ng (Cleveland Clinic, Cleveland, OH) voted no, stating that she would find it difficult to tell her patients that the drug reduces all-cause mortality based on one study. She pointed to the lack of a clear mechanism of benefit as her main specific concern and more broadly stated that she would like to see the benefit confirmed in a second study before touting it to her patients. We are curious to know if she was referencing “all” type 2 patients or those at high risk of CV disease – like those in the trial. While we would understand that she would be challenged to suggest this drug for all type 2 patients, this was not the question posed.
• For Dr. Kevin McBryde (NIH, Bethesda, MD), the lack of participant diversity was the red flag that led to his “no” vote. He noted that only 5% of EMPA-REG OUTCOME participants were black, compared to approximately 17% of Americans with diabetes. He pointed out that African-American patients may even have an increased risk of cardiovascular events and mortality, and called for a second trial with a component focused on minority populations. As with Ms. Hallare, we agree this is a grave concern but not one that should drive the decision to this question at hand. We remain frustrated that we see this arise at so many major and large trials – this seems like an addressable problem.
• Cardiologist Dr. Yves Rosenberg (NIH, Bethesda, MD) voted no, arguing that the data must be held to a more rigorous standard since Jardiance is the first drug in its class to offer the suggestion of a cardioprotective effect. He echoed the sentiments of skepticism that the CV mortality data came from only one trial and constituted only a secondary endpoint, additionally taking issue with the lack of racial diversity in the EMPA-REG OUTCOME study population. We continue to be surprised to see lack of ethnic diversity in clinical trials as it has come up time and again as a critical concern. This is an addressable problem as stated earlier – so we are disappointed to see such poor statistics (though, again, we do not think it should be the driver of the results). We had not anticipated that there would be different standards sought for the “first” to show cardioprotection … while we do not favor this, we would like to see more large-scale research week begin.
• Committee chair Dr. Robert Smith (Brown University, Providence, RI) shared that he was very conflicted about his decision, though he ultimately chose to vote against an updated indication. He noted that while he feels the cardiovascular mortality data is very convincing in the context of a single trial, he felt a level of uncertainty over whether a second trial would be able to reproduce the effect seen in EMPA-REG OUTCOME and that uncertainty is reflected in his negative vote. Given that the separation of the curves happened so quickly, we were a bit surprised to hear this  - though this may have also fed into his decision “nearly” to vote positively.
• Dr. Peter Wilson’s (Emory University School of Medicine, Atlanta, GA) negative vote was based on his belief that a higher level of evidence is required to justify an indication for cardiovascular mortality. He felt that since empagliflozin is the first compound in its class to report results, the benefit needs to be substantiated before an indication update can be contemplated. Dr. Wilson also expressed concerns with the use of a single study to assess cardiovascular safety and superiority. Finally, Dr. Wilson also suggested that the indication update may result in empagliflozin being widely overused in an inappropriate patient population (such as younger patients not at high risk for cardiovascular disease).

Close Concerns Questions

Q: In the era post-FDA 2008 CV Guidance and as cardiovascular outcomes garner more and more attention, how can the field develop clearer standards on the conduct and evaluation of CVOTs?

Q: How should “non-assessable” deaths be assessed, and can the field find a consensus of opinion within this community of clinicians and researchers? How should silent MI be treated in CVOTs?

Q: What are the circumstances in which positive results for a secondary outcome would be clinically meaningful and sufficiently substantial to justify additional indications for a therapy? Are there specific criteria that would make development more predictable?

Q: Can the field establish a method to assess the validity of secondary outcomes even if new hypotheses emerged during the course of a long-term clinical trial (i.e. realizing Jardiance would pass standards for both safety and superiority)?

Q: How can CVOTs and other clinical trials better recruit minority patient populations?

Q: If the benefit to CV mortality doesn’t make it onto the Jardiance label quite yet, what level of provider or patient education on the EMPA-REG OUTCOME results is available?

Q: What will be the impact of the FDA’s decision on an indication for cardiovascular mortality on how diabetes guideline-writing committees view the data?

Q: How will the FDA’s decision on the Jardiance label impact Lilly’s reimbursement negotiations for the drug?

Q: Will the EMA raise similar concerns over the trial design and methodology in its review of the EMPA-REG OUTCOME data?

Q: Will there be an Advisory Committee meeting for the LEADER results demonstrating cardioprotection for Novo Nordisk’s Victoza (liraglutide) and, if so, will similar concerns arise in that meeting?

Q: We’ve heard from many pharmaceutical companies on their interest in pursuing diabetes candidates with cardioprotective benefits in addition to glucose-lowering. How would the inclusion (or rejection) of an indication for cardioprotection on the Jardiance label impact industry thinking on this issue?

--by Abigail Dove, Helen Gao, Payal Marathe, Emily Regier, and Kelly Close