From Contrave Saga, Renewed Faith in Trials Built on Trust

Evidence-Based Diabetes Management, September 2015, Volume 21, Issue SP13

The decision by Cleveland Clinic's Steven Nissen, MD, to issue his own press release last spring after the cardiovascular outcomes trial was halted for Contrave turned plenty of heads in the pharmaceutical community. But Nissen and others involved in the saga see what happened with Orexigen as a one-time incident, not a flaw with the process for evaluating the safety of diabetes and obesity therapies.

Several months have passed since the Contrave safety trial imploded—enough time to learn any lessons the experience may have to offer. There is little evidence, however, that the singular, but well-publicized failure, has triggered any appetite to reform an otherwise successful system.

Trials before and since have managed to follow the established 2-part design without a hitch: a handful of corporate officials receive preliminary information they need to submit information for New Drug Applications (NDAs), while everyone else remains ignorant of the numbers, unable to taint the eventual outcome. Indeed, such trials have been required of all new diabetes drugs for several years now, and regulators seem disposed toward making it the norm for any medications thought to have potential implications for long-term cardiovascular health.1

“We’ve probably gone through a dozen of them now, and, to my knowledge, the design has worked as intended with every medication except Contrave. The whole system is a compromise between getting potentially beneficial medicines to market quickly and protecting people from unexpected harm. Using preliminary data in the first step is relatively fast, and it eliminates the risk of approving something truly disastrous, which happened every so often under the old trial regime, when the only things we considered were reductions in A1C [glycated hemoglobin] or body weight. Unfortunately, preliminary data aren’t enough. We need long-term outcome data, and the continuation of the same trial with the same participants is the quickest way to get it,” said Steven Nissen, MD, chair of the cardiology department at the Cleveland Clinic, in an interview with Evidence-Based Diabetes Management.

Nissen should understand the intention of the trial design because he was the one who proposed it to the FDA back in 2008.2 He also has reason to understand how the Contrave safety trial came undone and whether such circumstances are likely to recur, as he chaired the investigative committee that led the study, and he doubts that he will ever see anything like it again.

Contrave is a combination of 2 other drugs, bupropion (an antidepressant) and naltrexone (an opioid antagonist most commonly used to treat alcoholism). Studies found that although the drug had little effect on a majority of individuals in the study, a significant minority did respond. In 1 large trial, 55.6% of Contrave users and 17.5% of placebo users shed at least 5% of baseline body weight in 28 weeks.3

Historically, the FDA has not required separate safety trials for weight-loss medications—the safety information gathered from phase 3 trials was deemed sufficient—but the perceived tendency of bupropion to raise blood pressure,4 particularly when used in combination with a few other compounds,5 reportedly led regulators to demand a long-term cardiovascular outcomes trial, which was dubbed the LIGHT study.

FDA officials and clinical researchers from institutions across the country agreed on the basics of the trial structure with Orexigen, the company that developed Contrave, and Takeda, the company that had agreed to market it in the United States. Some 9000 patients were to be randomized between Contrave and placebo and followed until a set number of cardiovascular events had been recorded. After the first 25% of those events occurred, an independent data monitoring committee would look to see whether Contrave use was associated with at least twice the risk of strokes and heart attacks. If not, the committee would pass the data on to an equally small number of people at the 2 companies, who would use it to complete a portion of the NDA that wouldonly be kept private. Those few people who knew the preliminary results would be sworn to secrecy and cut off from any connection with the trial, while everyone else continued on in ignorance for several more years.

Such a cumbersome setup was estimated to cost around $200 million,6 but simply analyzing user records after a drug’s approval provides far less information. Some would argue that cheaper methods, such as retrospective analyses, provide hardly any information at all. Consider, for example, that millions of men have been using supplemental testosterone for decades now and there is still serious controversy about whether the practice significantly increases cardiovascular risk, significantly decreases it, or has little effect in either direction.7

The problem, according to Nissen and others, is that randomizing patients up front is the only way to get trial and control groups that are truly comparable. Assumptions about the ostensible similarity of people who do and do not use a medication in real-life situations often prove untrue. Users turn out to be sicker or less sick than nonusers in ways that escape detection, and retroactive analysis ends up being confounded by indication. Rofecoxib (Vioxx) was prescribed to 25 million Americans and all of those records failed to alert researchers that the drug significantly increased the risk of heart attacks. Data from a randomized trial (initially misanalyzed and later correctly interpreted by Nissen and others) uncovered a problem that probably killed thousands of patients.8

The difficulty with analyzing real-world data gets worse when a drug gets a reputation for possibly causing some problem. Metformin, for example, got a reputation for causing lactic acidosis, so physicians were far more likely to report metformin-related cases than cases related to other drugs, and it took specific trials to dispel the myth.9,10 As a result, the FDA tends to require long-term randomized trials to monitor outcomes in cases where it suspects possible risk.

The data from the first quarter of the Contrave trial were more than good enough to help secure the drug FDA approval in September 2014. Those early efforts to rule out the possibility that Contrave doubled the risk of heart attack or stroke found instead that the drug was associated with a 41% decrease in the risk of cardiovascular events.11 Had the news of these results been confined to a small number of Orexigen employees whose only responsibility was transferring the data to the NDA, the trial would probably have run its expected course. However, Orexigen CEO Mike Narachi was among the group that learned of the results,12 and although his company had signed a lengthy nondisclosure agreement with the steering committee and the data monitoring committee, Orexigen later argued that it had a more pressing fiduciary obligation to investors—an obligation to secure a new patent that covered Contrave’s possible use as a cardioprotective medication. If Contrave turned out to have this benefit, it would be the first of the new diabetes or obesity drugs to do so—a potentially blockbuster result.

“Orexigen has been working closely with and is committed to continuing to work with the FDA and others to support its regulatory obligations,” the company said in a statement. “Orexigen is also committed to simultaneously meeting its obligations to other regulatory authorities in the United States, such as the SEC [Securities and Exchange Commission], and abroad, such as the EMA [European Medicines Agency], which are relevant to, and have authority over, its business. The company is similarly committed to meeting its fiduciary duties to shareholders.”13

By the time the FDA approved Contrave, Orexigen told regulators that it had shared the preliminary trial data with more than 100 people. The FDA made no public comment at the time but later said that it determined then that the LIGHT study would no longer be sufficient to meet the outcomes data requirement. Orexigen and Takeda would therefore need to run a second trial.

The preliminary trial became fully public when Orexigen filed a Form 8-K with the SEC that explained how it had come to receive a new Contrave patent.14 Investors were delighted. Orexigen share prices jumped 50% on the news that its weight-loss drug might also protect patients against heart attacks and strokes.15 Investigators and regulators, on the other hand, denounced Orexigen for undermining a key safety trial.

First, the FDA released a statement that criticized Orexigen’s behavior and announced the need for a new trial: “The FDA strongly urged Orexigen to protect the interim data from public disclosure, and we are very disappointed by Orexigen’s actions. Even before the FDA became aware that the interim results from the LIGHT trial would become publicly available via patent applications, the agency had determined that [because Contrave had initially shared preliminary results with so many people] the LIGHT trial would not satisfy Contrave’s postmarketing requirement (PMR) related to cardiovascular safety. Therefore, the FDA required Orexigen to complete a second cardiovascular outcomes trial and that requirement remains in effect.”16

Then, a couple days later, the director of the FDA’s Office of New Drugs took the highly unusual step of saying in an interview with Forbes that the preliminary results almost certainly would not be borne out by any completed study: “Step back and think for a second,” [John] Jenkins says. “We required this study because we’re concerned that Contrave may cause adverse cardiovascular events because of its effect on blood pressure and heart rate. So, the likelihood that that drug is going to have an early benefit is highly unlikely. So people need to be very cautious about making medical decisions based on these data, and we’re very concerned that investigators and patients may be unwilling to be in a trial based on these data when they are likely false readings of the actual effect of the drug.”17

Even in retrospect, experts disagree about just when the spread of the preliminary trial data compromised any eventual findings from the completed study. The FDA statement indicates the belief that 100 Orexigen employees and affiliates were enough to sink it, even though company officials had nothing to do with trial management. Nissen believes the trial could have survived that initial lapse, but not the SEC filing that made the preliminary data available to everyone. Darren McGuire, MD, MHSc, who was a member of the LIGHT study’s data monitoring committee, told Evidence-Based Diabetes Management that he believes the study, against all odds, was still producing valuable data when it was stopped.

“The big problem with preliminary data spreading beyond its intended recipients is that it can reach study patients and site investigators and bias the results. If the preliminary data look bad, patients may drop out of the trial for fear of getting dangerous medication. If the preliminary data look good, they may drop out for fear of getting the placebo. Either way, it biases the study and control groups toward each other and thus toward a null finding,” said McGuire, who codirects a heart disease research group at UT Southwestern.

“The very odd thing with the LIGHT trial was that we expected the public release of the preliminary data to produce mass defections, and perhaps defections would have ruined the trial at some point, but even 3 months after the data became public, we had not observed a change in the drop-out rate that would further bias the results—perhaps because the news didn’t spread so fast as we expected, perhaps because people accepted the FDA statement that the benefits were almost certainly illusory, or perhaps because they felt a duty to go through with a trial and help provide answers…In general, patients have proven very willing to stay with these trials and risk getting the placebo, even after medications reach the market. Patients really do seem to understand that these long-term trials are the only way to know if these medications actually do improve outcomes and that their willingness to stick with trials helps everyone. That attitude, in turn, has helped to make this type of trial design effective,” added McGuire.

Despite warnings from the FDA and many independent researchers that the apparent cardiovascular benefits of Contrave were almost certainly illusory, Contrave sales took off. The preliminary study data were the only data available and they appeared sufficient to convince many physicians and patients to choose Contrave over the alternatives.18

In late March, as sales rose, the LIGHT study’s executive committee voted unanimously to end the trial and publicize its ultimate findings. As expected, by the time half of the preset number of cardiovascular events had taken place, the apparent benefits of Contrave usage had disappeared. Use of the drug through the halfway point was not associated with significantly lower—or higher—risk of stroke, heart attack, or death. Takeda reportedly agreed to several different press releases that announced this finding in different languages, but Orexigen spent 6 weeks rejecting every release that investigators proposed. On May 12, 2015, Orexigen and Takeda announced that the trial had been stopped—in a press release that contained no extra trial data.19

Minutes later, the Cleveland Clinic issued a release of its own, a release that contained analysis of the 50% data.18 Nissen, moreover, told several journalists that day that Orexigen had misled both patients and investors by releasing the 25% data and that it had tried to perpetuate misperceptions by blocking the release of the 50% data.

Orexigen countered with a statement: “Contrary to allegations cited today by a journalist, Orexigen has never misled patients. At the time of the patent issuance in March, we stated plainly and clearly that the effect of Contrave on CV [cardiovascular] morbidity and mortality has not been established and that a larger number of [major adverse cardiovascular events] are required to precisely determine the effect of Contrave on CV outcomes …Takeda and Orexigen agreed that the appropriate manner to wind down the study was to collect the final information from the study and then to present and publish the study results. There was pressure from the [Executive Steering Committee] to release the 50% interim data. We maintained we would not be in a position to release data without access to the full data set, which we have not had and still do not have.”

Even before additional trial data came out, it seemed unlikely that another trial would ever break down in a similar fashion; the precedent of an investigator and his employer choosing to violate a confidentiality agreement to provide important information to the public in what they considered due time made it even harder to imagine. That move nullified any advantage that Orexigen had inadvertently gained through the original leak, and the fact that it went unchallenged greatly reduced the expected gain any future company could expect from any attempt to intentionally manipulate information. Orexigen stock fell from $6.86 per share on May 11, 2015, to $5.02 per share, 2 days later on May 13—nearly $1 lower than it was before the preliminary data became public.20 It was also announced that Takeda had asked Orexigen to pay the entire cost of Contrave’s second long-term safety trial, which will be led by Nissen.

“This trial design places a good deal of trust in all parties, trust that they will meet their obligations and produce the sort of quality research that will serve physicians and patients well. Even if things had turned out far worse in this case, I think most people would support the continued use of this basic trial design, though perhaps with a few modifications. The fact that trobably counts as a tolerably good outcome—is just another argument in its favor,” said McGuire.

“The fact that the system proved itself capable of turning a potential disaster into a tolerable outcome—and a clean slate for a second trial probably counts as a tolerably good outcome—is just another argument in its favor,” said McGuire.

Nissen agrees that trust is key to the success of the 2-stage trial and says that he sees no signs that the Contrave experience has led to any general diminution of trust among researchers, regulators, and corporations. “Are there some steps that might further decrease the risk of data leaks, like a flat ban on unblinding executives or a refusal to consider any data that has leaked out in new drug applications? Possibly,” said Nissen. “For the most part, though, secrets have stayed secret, both when preliminary data have been good and when preliminary data have been bad, because nearly everyone accepts the fundamental truth that such things have to stay secret for the system to work.” References

1. Sager PT, Seltzer J, Turner JR, et al. Cardiovascular Safety Outcome Trials: a meeting report from the Cardiac Safety Research Consortium. Am Heart J. 2015;169(4):486-495.

2. Herper M. The great drug debate. Forbes website. Published July 3, 2008. Accessed August 20, 2015.

3. Apovian CM, Aronne L, Rubino D, et al; COR-II Study Group. A randomized, phase 3 trial of naltrexone SR/bupropion SR on weight and obesity-related risk factors (COR-II). Obesity.

2013;21(5):935-943. doi: 10.1002/oby.20309.

4. Mayo Clinic staff. Medications and supplements that can raise your blood pressure. Mayo Clinic website.

art-20045245. Published March 13, 2013. Accessed August 20, 2015.

5. Fava M, Rush AJ, Thase ME, et al. 15 years of clinical experience with bupropion HCI: from buproprion to bupropion SR to buproprion XL. Prim Care Companion J Clin Psychiatry. 2005;7(3):106-113.

6. Silverman E. Orexigen and Takeda feud over cost of controversial diet drug trial. Wall Street Journal website. Published May 13, 2015. Accessed August 20, 2015.

7. Morgentaler A, Miner MM, Caliber M, Guay AT, Khera M, Traish AM. Testoterone therapy and cardiovascular risk: advances and controversies. Mayo Clin Proc. 2015;90(2):224-251.

8. Vioxx (drug). Topics page. New York Times website. Accessed August 20, 2015.

9. Misbin RI. The phantom of lactic acidosis due to metformin in patients with diabetes. Diabetes Care. 2004;27(4):1791-1793.

10. Inzucchi SE, Lipska KJ, Mayo H, Bailey CJ, McGuire DK. Metformin in patients with type 2 diabetes and kidney disease: a systematic review. JAMA. 2014;312(24):2668-2675. doi:10.1001/jama.2014.15298.

11. Feuerstein A. Orexigen weight-loss pill shows surprise heart safety benefit. The website. Published March 3, 2015. Accessed August 20, 2015.


12. Herper M. Orexigen CEO pledged to keep data secret but not from himself. Forbes website. Published May 20, 2015. Accessed August 20, 2015.

13. Silverman E. Fat chance: FDA chastises Orexigen for disclosing interim trial data. Wall Street Journal website.

for-disclosing-interim-trial-data/. Published March 4, 2015. Accessed August 20, 2015.

14. United States Securities and Exchange Commission. Orexigen Therapeutics, Form 8-K.

Published March 3, 2015. Accessed August 20, 2015.

15. Barber J. Shares in Orexigen Therapeutics surge on interim data released from LIGHT study of Contrave. FirstWordPharma website. Published March 3, 2015. Accessed August 20, 2015.

16. Husten L. Orexigen released interim data without approval of trial leaders. Forbes website. Published March 3, 2015. Accessed August 24, 2015.

17. Herper M. Top FDA official says Orexigen study result ‘unreliable,’ ‘misleading.’ Forbes website. Published March 5, 2015. Accessed August 20, 2015.

18. Williams S. Obesity statistics take a wrong turn in this latest study. The Motley Fool website.

aspx. Published July 5, 2015. Accessed August 20, 2015.

19. Orexigen Therapeutics provides statement on termination of the LIGHT study and updates on Contrave collaboration with Takeda Pharmaceuticals [press release]. San Diego, CA: Orexigen Therapeutics Inc; May 12, 2015.

20. Orexigen Therapeutics. Google Finance website.