The first question related to how the study was conducted, and addressed the exclusion of some data from the final endpoint calculation. Specifically, the removal of certain data relating to silent myocardial infarction, and how this might impact the reliability of the data that pointed towards an endpoint hit.
The response to this one was mixed, but looks weighted towards the opinion that these changes don’t really impact the reliability of the data. The fact that the changes occurred before an interim unblinding fell in favor of the endpoint hit, and the robustness of the data on which the extension is resting (i.e. the CV death data), as well as this data’s chances of being affected by the exclusion of silent MI (i.e. very slim) supported this opinion.
Interpretation – generally positive and favors approval.
The second question again referred to the changes discussed in the first question, and sought to identify the committee’s confidence in the data supporting the primary analysis – i.e. the 3-point MACE reduction.
Again, the panel served up a mixed response to this one, but this time it looks as though the response was slightly weighted against approval. Specifically, a large number of CV deaths were excluded from the trial and classed as not evaluable as part of the final analysis. The uncertainty surrounding the justifiability of these non-inclusions looks to have translated to some loss of confidence in the bottom line numbers, and in turn, the drug’s justifiability as an extended indication. Additional concerns surrounding the exact mechanism of action (MOA) were also raised. To quote:
“Some committee members stated that the trial results do not clearly point to a readily identifiable mechanism to explain the CV mortality results.”
Basically, we don’t really know why this drug reduces risk of CV death, even if the data suggests it does. For some panel members, and as many of these sorts of meetings have shown historically, this isn’t a big issue. For some, however, it can be. It seems there may have been more of the latter group on this panel than the former.
Interpretation – again mixed, but somewhat weighted against approval.
The third question was a sort of combination of the first two, seeking to address the persuasiveness of the findings when looked at in favor of approval.
The answer to this one was clear, but again, mixed in its inference. The panel pretty much agreed across the board that the data suggested Jardiance has a positive impact on CV deaths in diabetes patients, and based this agreement on the low P Value for the data. In other words, statistically, the drug looks to work well. However, based on the lack of understanding of a clear MOA, the panel (again, pretty much unanimously) agreed that the robustness of the results is questionable.
Interpretation – mixed, weighted towards no approval based on a lack of scientific understanding.
The fourth question addressed a specific point of the study relating to heart failure. Essentially they were asked to discuss whether the drug is effective in reducing heart failure, based on the available data.
The answer to this one is clear and simple – no, the data is not sufficient to say with any confidence that the drug is effective in reducing the rate of heart failure. Further, the data is not only insufficient to make this assumption, it suggests that the drug is NOT effective to this end.
Interpretation – negative, but not overly relevant. The labelling expansion is targeting reduced risk of cardiovascular risk, yes, but heart failure is going to make up only a small part of this labelling.
The fifth and final question looked at the so called renal findings. By way of a brief introduction here, the trial threw up some data that suggested an improved profile over placebo in renal risk, and in some of the numbers, even suggested that the drug might have a renal benefit on patients in this indication. The question posed here relates to this benefit, with regards to whether the data is reliable enough to justify any sort of labeling, or further investigation into the renal side of the equation.
Again here the interpretation was profoundly negative. The panel noted that the data was intriguing, but far from usable for either of the above-mentioned goals, and that a lack of proper adjudication pretty much renders it irrelevant to the trial and – in turn – the outcome of this application.
Interpretation – again, negative to some extent, but judged to be totally irrelevant, so it shouldn’t impact the panel’s, or the FDA’s final decision.
After the fifth question, the panel was asked to vote on two points.
