Celebrex PRECISION Trial: 'Herculean' Effort, But Missing Information Bugs US FDA Panel

Pfizer study finds Celebrex's CV risk comparable to naproxen and ibuprofen, but advisory committee wanted data on confounding factors and reasons for high dropout.

Pfizer Inc.'s 10-year Celebrex (celecoxib) cardiovascular study culminated in a US FDA advisory committee vote supporting a label change that could boost sales of the COX-2 inhibitor. But the panel discussion showed some of the trial's limitations and the difficulties in doing a large population-based study.

FDA asked the panel to vote on whether the PRECISION trial "demonstrated comparable cardiovascular safety for celecoxib as compared to naproxen and ibuprofen." At their joint meeting April 24-25, the Arthritis Advisory Committee and the Drug Safety and Risk Management Advisory Committee voted 15 to 5 with one abstention that it did. (Also see "Celebrex Likely To Get Label Change Saying CV Risk Is Comparable To Ibuprofen, Naproxen" - Pink Sheet, 25 Apr, 2018.)

However, many who voted 'yes' did so with caveats and cited information they wished the study had provided, including explanations for confounding factors and the reasons that 68% of subjects stopped treatment.

Pfizer began the PRECISION trial in 2006 at the request of FDA following the withdrawal of the only other COX-2 inhibitors on the market, Merck & Co. Inc.'s Vioxx (rofecoxib) and Pfizer's Bextra (valdecoxib). The prospective, randomized, double-blind trial included 24,081 patients with osteoarthritis (90% of the study population) or rheumatoid arthritis who had cardiovascular disease or risk factors for cardiovascular disease. They were randomized to receive celecoxib 100-200 mg twice a day, naproxen 375-500 mg twice a day, or ibuprofen 600-800 mg three times a day. Pfizer noted that the dose of celecoxib was limited because labeling restricts use to 200 mg per day for most patients.

"I really appreciate the Herculean effort of the PRECISION trial. It took a long time, there was a lot of work that went into it," said Terri Warholak, University of Arizona College of Pharmacy. "But that said, there's certain things I'm a little worried about."

When reading the briefing materials for the meeting, "I just was at full stop when I saw the dose of the celecoxib because in my mind it's not comparable to the other doses that were studied," she said. In addition, she said there appeared to be differential experimental mortality. And she said she would have liked to have seen more information on the reasons for dropout and confounders that were introduced post-randomization, such as the use of rescue medicines.

Despite these concerns, Warholak voted yes while Eric Tchetgen Tchetgen, University of Pennsylvania, who had similar criticisms, voted no.

Tchetgen Tchetgen said noninferiority trials are susceptible to certain sorts of bias, including noncompliance or discontinuation during the course of the study, switching back and forth between treatment arms, or using other medications. He said all these issues were present in the PRECISION trial.

He noted that there was some evidence presented that the discontinuation was balanced between treatment arms with respect to baseline characteristics, but he said that was not particularly compelling because there was a lot going on post-randomization.

"Pfizer was not always as rigorous as we would have liked them to be in the conduct of the trial," FDA's Judith Racoosin said.

"There were a lot of risk factors for discontinuation that are post randomization factors that would also be related to the endpoint.  None of those were accounted for. The data was not particularly well collected on adherence. I would have liked to have seen a lot more about those data," Tchetgen Tchetgen stated. "A perfect randomized trial is just as good as an observational study and there are a lot of methods for observational studies that should have been considered to further assess the robustness of this trial."

No Evidence Of Excess CV Risk Within Study Parameters

FDA also noted that it felt the study was missing information.

"Pfizer was not always as rigorous as we would have liked them to be in the conduct of the trial," Judith Racoosin, deputy director for safety in the Division of Anesthesia, Analgesia, and Addiction Products, said. "In particular, they did not capture some information that would have been helpful for interpreting the results. For example, information on adherence, closer tracking and analysis of use of rescue therapy for pain, and the specific reasons for discontinuing study treatment."

However, FDA did not find these issues to undermine the study conclusions. In the agency's presentation to the committee, statistical reviewer Bo Li said that "at the doses studied, pre-specified primary analyses results showed no evidence of excess CV risk associated with celecoxib compared to naproxen and ibuprofen." In addition, she said "sensitivity analyses and other CV-related endpoints showed consistent results with the primary analyses."

Other panelists recognized shortcomings with the study, but concluded that it showed celecoxib did not have higher cardiovascular risk than naproxen or ibuprofen.

"I do acknowledge all the limitations, statistical and otherwise that have been discussed," said Yves Rosenberg, of the National Institutes of Health's National Heart, Lung and Blood Institute.  "They definitely need to be taken into account in the interpretation of results of the trial. However, we never look at one clinical trial in a vacuum. We look at in the context of other research, meta-analyses, and look at the consistency of the results as a whole."

Within the context of the trial and what was studied, "this ultimately looks fairly reassuring that there is not an excess cardiovascular risk for the patients evaluated in this trial at the dose tested," Rosenberg stated.

Findings Challenge View Naproxen Has Superior CV Safety, Nissen Says

The results of the trial were published in the New England Journal of Medicine in November 2016. The primary endpoint was the first occurrence of an adverse event that met the Antiplatelet Trialists' Collaboration (APTC) criteria, a composite of cardiovascular deaths, non-fatal myocardial infarction, or nonfatal stroke.

In the intention-to-treat analyses, a primary outcome event occurred in 188 patients in the celecoxib group (2.3%), 201 patients in the naproxen group (2.5%), and 218 patients in the ibuprofen group (2.7%) (hazard ratio for celecoxib vs. naproxen, 0.93; 95% confidence interval (CI), 0.76 to 1.13; hazard ratio for celecoxib vs. ibuprofen, 0.85; 95% CI, 0.70 to 1.04; P˂0.001 for noninferiority in both comparisons).

In the on-treatment analysis, a primary outcome event occurred in 134 patients in the celecoxib group (1.7%), 144 patients in the naproxen group (1.8%), and 155 patients in the ibuprofen group (1.9%) (hazard ratio for celecoxib vs. naproxen, 0.90; 95% CI, 0.71 to 1.5; hazard ratio for celecoxib vs. ibuprofen, 0.81; 95% CI, 0.65 to 1.02; P˂0.001 for noninferiority in both comparisons).

Following slow accrual of APTC events, Pfizer proposed and FDA agreed to lowering the study power from 90% to 80%. When APTC event accrual remained slow, FDA agreed in 2011 to increasing the non-inferiority margin for the modified intent-to-treat analysis to 1.4 from 1.33 while keeping the non-inferiority margin for the intent-to-treat analysis at 1.33.

Trial investigator Steven Nissen, of the Cleveland Clinic, presented the results of the trial to the committee. He said the findings challenge the widely-held view that naproxen provides superior cardiovascular safety. He also noted that adherence and retention were lower than typical CV outcome trials, but similar to other NSAID pain studies with no strong evidence for an effect on the primary noninferiority findings.

Pfizer submitted the study data to FDA on June 23, 2017 and requested that the results be included in the celecoxib US package insert.

Drug's Mechanism Unclear

FDA's Sharon Hertz, director of the Division of Anesthesia, Analgesia, and Addiction Products, noted some remaining questions from the data in response to a question from Steven Solga, University of Pennsylvania Perelman School of Medicine, who said the meeting was much different from the panel's 2014 meeting addressing CV risks with NSAIDs. He questioned why the agency had not invited experts who gave presentations then to the current meeting.

"We spoke about it as perhaps the most investigated question in the history of medicine," Solga said of the discussion of CV risk at the 2014 meeting.

Hertz replied that the purpose of the earlier meeting was to "further our understanding of CV risk in the context of all of the work that had been going on, all of the epidemiological studies, all of those things that had sometimes conflicting data, often had different methods [from] different countries with different standards."

The agency always worries "whether there are underlying factors that we can't identify that may be contributing to the outcome," she said. So, when possible, we try to get a prospective clinical trial.

"Now that we're here, really the question is what did we learn from this clinical trial and how should we think about these data and how should that influence our thinking about appropriate labeling" that ultimately will hopefully help clinicians manage their patients, Hertz said.

"To be perfectly honest, I'm not sure we know the mechanism," she added. "Is it blood pressure, is it platelet, what is it? There's many possibilities and perhaps there's many concurrent processes contributing" to it, she added.

FDA noted in its briefing document that an estimated 70 million prescriptions were dispensed from retail pharmacies for celecoxib, ibuprofen, or naproxen single ingredient products in 2017. Of these prescriptions, ibuprofen single ingredient products accounted for the largest portion of use at 64% (44.8 million prescriptions), followed by naproxen single ingredient products at 26% (18 million prescriptions) and celecoxib at 10% (7.3 million prescriptions). Ibuprofen and naproxen formulations are available over-the-counter, but there is no OTC formulation of celecoxib.