FDA’s new Sentinel system is up and running, and the first examples of real world use of the “real world” post-marketing data network are being showcased. One example of compliance with patient selection criteria for Lilly’s Effient indicates some of the profound ways Sentinel may ultimately reshape regulation—but also suggests that the notoriously slow launch of prasugrel was even worse than it looked.
The fourth annual Sentinel Initiative Public Workshop was something of a coming out party for the new FDA active surveillance system. As The Pink Sheet notes, here, the FDA/Harvard “mini-Sentinel” project now has access to data on 126 million people, more than enough to meet the statutory objective set by Congress of 100 million lives by July 2012. It was also the first time, as Harvard’s Richard Platt noted, that there were actual data to present, showing real world uses of the system. Indeed, according to Platt and the FDA officials who spoke at the workshop, Sentinel is now very much part of the drug safety toolbox, with more than 100 safety queries conducted so far.
The use of Sentinel in FDA’s post-market monitoring is a pretty big deal, and exactly how the new tool starts to affect product lifecycles will be an important story for industry to monitor for years to come. (We’ll have lots more in the next issue of The RPM Report, but for starters, here is how Center for Drug Evaluation & Research Deputy Director Robert Temple, the unofficial dean of the FDA review staff, sees it fitting in.)
One of the early reports from Sentinel struck us as especially interesting for what it may portend about where the new capability will take the agency—and the impact it may have on some sponsors during the learning process.
This was a query that looked at compliance with FDA labeling for a new drug: specifically, the contraindication for use of Lilly’s Effient (prasugrel) in patients with a history of stroke or transient ischemic attacks. The point of the study, as Platt described it, wasn’t to look for adverse events per se, but rather to look at how the contraindication affected real world use of the brand.
To answer that question, the Sentinel team analyzed claims data to determine the proportion of Effient users who had a stroke or TIA within the past 12 months, compared to those who received Bristol-Myers Squibb’s Plavix (clopidogrel), which is not contraindicated in that population.
The first thing that jumps out from the data is the dramatic difference in denominators: the query turned up about 153,000 new Plavix users with data on prior medical histories, compared to just 7,000 new Effient users. In other words, Plavix had 96% share of new scripts in this base of payors.
We knew Effient was off to a slow start, but yikes.
As for the actual point of the analysis, the Sentinel team found that approximately 8% of Effient users had a prior history of stroke, and another 2% had a TIA history. That is “substantially fewer” than for clopidogrel, where the prior stroke rate was about 17% and prior TIA was about 8%.
But it isn’t zero.
Remember, this is a contraindication against use in this population, not just a warning. In fact, it isn’t just a contraindication either—it is also the basis of a Risk Evaluation & Mitigation Strategy imposed on the product that requires communications to providers about appropriate patient selection, “emphasizing patients in whom Effient should not be used.”
So, the analysis suggests, about 10% of users of Effient had a contraindication for the agent—a contraindication which Lilly was specifically required to emphasize in its launch materials.
Now, to be clear: the analysis should not form the basis for any profound conclusions. “There are lots of limitations to this kind of quick look at the question,” Platt stressed, noting for example that the codes for prior stroke and TIA “have not been verified in the mini-Sentinel dataset.” (On the other hand, the contraindication is not limited to stroke/TIA within 12 months, so the reported rate of compliance with the labeling is probably too high.)
For Platt, the important thing is the ability “to be able to ask the question and get an answer in a very short period of time is something that wasn’t possible before.”
Fair enough. But still, there are some implications to ponder.
First, the data suggests that the warnings added to Effient labeling aren’t the only reason for the poor launch. After all, if the key factor driving the failure to prescribe the therapy was concern about bleeding risk, you would expect the rate of prescribing in at-risk patients to be far lower.
It also suggests that Lilly may be facing some further headaches on the regulatory front. Platt disclosed that the Sentinel team is preparing to do a follow-up query to look at the rate of strokes reported in users of drugs in the class—a more typical use of the Sentinel dataset. There was no timeline given for that analysis, but with Effient finally showing signs of traction in the US, the last thing Lilly needs is to worry about some kind of drug safety notice triggered by a Sentinel query.
Then there is the question of what the Sentinel data suggest about Lilly’s compliance with its REMS obligations. The company has already completed its first REMS assessment, having submitted an 18 month report in January 2011. That assessment included revisions to the patient and physician communication materials. But the next update is due at the three-year mark (July 2012), and FDA will have the Sentinel data to compare to Lilly’s survey-based assessment data.
Given the still developing knowledge about how to use Sentinel and how to do REMS assessments, it doesn’t seem likely that anyone will come down too hard on Lilly. But still: we would imagine there will be at least some discussion of how to take that apparent 10% rate of use in contraindicated patients down.
All of which underscores the last point for this early case study: the role of Effient as one of the guinea pigs in the new safety system means the product will likely get more attention in the post-market space from FDA than it would have otherwise.
FDA’s interest in the risks of Effient may have cooled given the lackluster adoption of the product, but the decision to impose the REMS (and, we suspect, the request to study compliance in Sentinel) were made based on the expectations that it could be a very widely used drug. (Indeed, Temple told a different conference last year that he had expected prasugrel to become a blockbuster.)
Certainly it isn’t making Lilly happy that Effient didn’t live up to its commercial expectations at launch. But the high expectations are still embedded in the scrutiny the product is receiving from FDA, which may make the costs of that failure be even higher than it looks.
Tuesday, January 24, 2012
Why the Prasugrel Launch Was Worse Than it Looked: A Sentinel Case Study :: Elsevier Business Intelligence