On July 23, the Food and Drug Administration will make its decision on bezlotoxumab, brand name Zinplava, a Merck & Co., Inc. (NYSE:MRK) monoclonal antibody drug designed to prevent recurrence of C. difficile infection. Market Exclusive’s Inside the FDA series will attempt to give you an inside look at the FDA’s decision making process, its leanings and deliberations, and deduce from there the chances of a given drug’s approval. We will be using FDA briefing documents straight from the agency itself to give our readers a more accurate picture of what is going on.
From there we will summarize market potential and make a brief investment analysis based on the chances we estimate from the FDA briefing documents themselves. This week we will begin with Merck.
C. difficile Infection Background
C. difficile infection is essentially a man-made disease that results from an imbalance in gut bacteria, usually as a result of long courses of antibiotics that disturb the internal microbiome. The microbiome is the community of bacteria in the intestine, most of which are either harmless, helpful, or even essential for digestion. C. difficile is a bacteria that is present in most people and usually harmless. It is kept in check by other bacterial colonies in the gut that compete with it for resources. However, when long courses of antibiotics are taken, especially antibiotics that are broad spectrum and kill many types of bacteria, C. difficile can quickly take over and overpopulate the gut, causing anything from diarrhea to life threatening holes in the intestinal wall.
Think of it as a predator-prey situation. If the predator is taken out of an ecosystem, the prey can often overpopulate the area and cause widespread destruction. Both predator and prey keep the ecosystem in balance.
The danger of C. difficile infection is not from the bacteria itself, but from the toxins it produces as byproducts that cause inflammation in the intestinal wall. Much like necrotizing fasciitis, or flesh-eating bacteria, it is not the bacteria itself that eats flesh but a toxin produced by the bacteria that causes the symptoms.
Bezlotoxumab is not an antibiotic, but an antibody that is designed to lock onto C. difficile toxin B, which is theorized to cause intestinal inflammation. The elderly who are under long term hospitalization and antibiotic treatment are the highest risk group, as the most virulent strains originate in hospitals where antibiotics are common. Data published in 2009 reported 336,600 hospitalizations in the US from C. difficile infection (CDI), and as antibiotic use and resistance grows, those numbers have grown as well.
As of 2011 according to the U.S. Centers for Disease Control and Prevention (CDC), CDI infections numbered half a million with 29,000 deaths often within 30 days of initial diagnosis, so this is serious stuff and not just a painful nuisance like your average intestinal bug and diarrhea episode. In the U.S., 80% of deaths related to CDI occur in patients 65 or older.
The infection itself is treated, once again, with antibiotics as there currently is no other choice. Bezlotoxumab is designed to prevent the recurrence of infection after an initial cure by antibiotics. An FDA panel back on June 9 voted 10 to 5 to recommend the drug for approval, which is generally seen as a shoe-in for approval, but the FDA briefing document is not as simple as that. So let’s get into it.
Inside the Briefing Document
Back in 2006, an early Phase II trial was conducted comparing antibodies against C. difficile toxins A and B with placebo in patients on antibiotics against CDI. 101 patients enrolled in the active arm, and 99 in the placebo arm. Recurrence rate for CDI was lower for the active arm than placebo at 7% vs. 25% with a p value of 0.001. So far so good.
But here is where the problems begin. The protocol for one of the Phase III studies had an endpoint that, in the FDA panel’s eyes, was problematic. The endpoint was CDI recurrence. The problem with that is that patients who failed antibiotic treatment would be counted as “no recurrence” because technically, there is no recurrence if the infection was never cured in the first place. Say, to take an extreme example just for conceptual purposes, that the bezlotoxumab arm had zero patients who reached an initial cure. In that case, the recurrence rate would be zero, and would show success for the drug even though it didn’t actually prevent recurrence as there was nothing to recur.
Because of this statistical problem, the FDA recommended changing the endpoint to global cure rather than CDI recurrence. Global cure means initial cure plus no recurrence. Sounds logical, yes?
But in November 2012, Merck responded to the request by saying it would stick with the CDI recurrence endpoint, saying that they didn’t expect antibodies to C. difficile toxins to have an impact on the clinical cure rate. Was Merck asking for trouble here?
Eight months later in July 2013, Merck acknowledged the statistical problem and attempted a compromise. That compromise was to place the global cure endpoint as a secondary endpoint for both Phase III trials, and to pool the global cure endpoint data from both trials to arrive at an efficacy conclusion. This, rather than analyze each global cure rate separately for both trials. The FDA did not like the pooling data idea, because it would mix variables from both trials and lead to a less conclusive result, but Merck stuck to its guns on the issue. Why? Did Merck believe that the technical passage of a primary endpoint would convince human beings with their own free will just because an endpoint was achieved, when they didn’t like the setup? Did they think that reporting the passage of an endpoint would be good for PR? A combination of both? Hard to say.
The problems got worse from here. In one of the Phase III trials, the clinical cure rate (meaning the initial cure before recurrence is measured) was lower for both active arms than it was for the placebo arm. One active arm trialed actoxumab and bezlotoxumab together, which are designed to attack CDI toxins A and B respectively. Keep in mind that patients in the active arms were taking all drugs since the beginning of antibiotic treatment for initial cure. The initial cure rate for the combination arm was 74.7%. For bezlotoxumab alone it was 77.5%. The initial cure rate on placebo, meaning antibiotics plus no antitoxin, was 82.8%. This is precisely what the FDA panel had feared, because a lower initial cure rate means by definition a higher no-recurrence rate as the two are inverse. It skews the CDI recurrence endpoint in favor of bezlotoxumab. For this trial, there was no statistical significance in global cure rate for the two active arms over placebo, and that is taking into account the inherent skewing.
Throughout the panel briefing review, the FDA injects caution to almost every positive conclusion due to this inherent skewing by the primary endpoint. Though statistical significance was reached for the bezlotoxumab arm over placebo in terms of recurrence, the FDA repeatedly questions the actual significance of the technical statistical significance given that initial cure rates were lower for the active arms over placebo. A read of the original document almost conveys a tone of subliminal frustration with Merck over its insistence to maintain the primary endpoint as CDI recurrence rather than global cure.
The following two sentences from the briefing are particularly significant:
“In both [Phase III] trials, the difference [in initial cure rate] was in favor of placebo and significantly better for placebo in Study P001. Thus, a negative effect of bezlotoxumab on clinical cure of the initial CDI episode cannot be ruled out.”
Following these lines, the panel notes that the efficacy of bezlotoxumab is still better assessed by global cure. In other words, this is effectively the FDA panel rejecting recurrence rate as a primary endpoint and picking global cure on its own recognizance. And then this sentence:
Overall, these results suggest that bezlotoxumab may negatively affect cure rate of the initial CDI episode.
In the document itself, this sentence is not included in the safety section of the analysis, which is section 9. It is included as part of the efficacy analysis, section 8, but it does seem that in the minds of the FDA panel at least, it is in fact a safety rather than efficacy issue, especially because it is literally the final sentence before the safety analysis section of the briefing.
From a technical standpoint, there were no significant safety issues besides the safety issue surreptitiously tucked into the efficacy section, but it seems clear that the panel is at least concerned that administration of bezlotoxumab may hinder the initial cure rate.
Nevertheless, the panel voted 10 to 5 to recommend approval, probably based on the endpoint of lower recurrence technically being met, and the technical fact that in terms of strictly defined safety as in adverse events following treatment, there were no differences between arms. Based on that and the rigid bureaucratic structure of these FDA panels, they voted to recommend approval, but who knows what is going on behind closed doors.
With around half a million infections a year and around 30,000 deaths, bezlotoxumab could theoretically generate blockbuster revenues if there is full market penetration. There probably won’t be even if approved because doctors could be wary of the drug lowering initial cure rates. Whether it actually has any effect on that is uncertain but according to this panel’s analysis of the data, it could be. When it comes to the decision on the 23rd, the FDA could even ask for another Phase III trial to ensure that bezlotoxumab does not affect initial cure rates. Then again it could simply approve it based on the official panel vote.
We estimate the chances of approval at 60%. Approval may give Merck a small boost especially if biotech has a good opening final week of July. A complete response letter (CRL) addressing the initial cure rate issue should not bring the stock down dramatically though.