Letter Urging that Rimonabant (Zimulti) and Etoricoxib (Arcoxia) be Removed from the European Market

July 16, 2007

Dr. Thomas Lönngren  
Executive Director, European Medicines Agency (EMEA)
7 Westferry Circus
Canary Wharf
London E14 4HB
United Kingdom

Dear Dr. Lonngren,

This urgent letter is written in light of an EMEA meeting this week that will discuss, among other topics, a possible change in the regulatory status of the anti-obesity drug rimonabant (Accomplia).


In the past three months, separate FDA advisory committees evaluating the arthritis drug etoricoxib (Arcoxia) and rimonabant voted 20-1 (April 12th) and 14-0 (June 13th, respectively), against approval of these drugs for use in the United States.  These same two compounds, available in the EU since 2002 (etoricoxib) and 2006 (rimonabant), have been documented to have a myriad of problems including safety and efficacy profiles that are dubious at best. The decision of each of these advisory committees was that the risks of the drugs outweighed their benefits and that they were not fit for use by people in the U.S.  We hope that the EMEA will reconsider their stance on these two drugs and we strongly urge that both etoricoxib and rimonabant be promptly removed from the market. If you fail to do so, you are ensuring the continuing damage to the health of Europeans from drugs with no unique benefits and well-documented risks.


The FDA advisory committee that reviewed etoricoxib could not justify recommending its approval due to a totally unacceptable risk/benefit profile.  The EMEA has shown some concern over the Cox-2 inhibitors in general, commissioning two CHMP reports on NSAIDs and recommending that doctors “use the lowest effective dose [of COX-2 inhibitors] for the shortest possible duration of treatment.”[1]  Even with this warning in effect, however, Arcoxia saw $265 million in sales for 2006 in the 63 countries in which it is approved. 

The consideration of etoricoxib involves three outcome variables: relative efficacy for osteoarthritis (OA), relative cardiovascular risk and relative gastrointestinal toxicity (serious complications such as perforation, bleeding and obstruction).   A comprehensive evaluation of each of these three areas shows that the drug is clearly unfit for even the limited use currently recommended in the EU.

The first consideration, relative efficacy, is easiest to discuss because Merck admits, in its conclusions, that “once daily treatment with etoricoxib 60 mg shows comparable efficacy to naproxen 1000 mg (500 mg 2 times daily) and to diclofenac 150 mg (50 mg 3 times daily) in patients with OA.”[2]  Thus, the same therapeutic effect can be achieved through the use of other, and as will be discussed, safer drugs. During the advisory committee hearing, Merck was unable to provide evidence from any study documenting that there were people who had not previously responded to other NSAIDs who, when randomized to get etoricoxib or another NSAID, responded better to etoricoxib.[3]

For the second variable, relative cardiovascular risk, it is clear that etoricoxib causes increased risk of vascular events compared to placebo as well as other available treatments.  As this has been shown to be an issue for the COX-2 inhibitors as a class,[4],[5] it is essential to use a non-selective NSAID comparator such as naproxen, as opposed to diclofenac, which was used by the drug company for comparison.  The American Heart Association found that the relative cardiovascular risk of naproxen compared to placebo was 0.92, not significantly different from 1.  For diclofenac, however, the increased cardiovascular risk was a significant 1.63 times placebo.[6]

Contrasting etoricoxib with naproxen shows that the former drug is, if fact, unsafe.  Dr. David Graham of the FDA presented evidence that though the relative risk of an adverse event was .96 comparing etoricoxib to diclofenac, when naproxen was the comparator the ratio jumped to 2.72 (95% CI: 1.18-6.27).[7]  Adverse events here most commonly include MI, fatal; MI, sudden death; and unstable angina as well as strokes with the greatest numbers of events being MIs and strokes.[8]

Even in comparison to diclofenac, however, there are suggestions of increased cardiovascular risk for etoricoxib.  The EMEA’s CHMP report on NSAID’s describes data from the MEDAL program finding that “cardiorenal events such as oedema, hypertension and cardiac failure were more frequent and severe with etoricoxib [compared to diclofenac], and this may have indirect relevance to long-term thrombotic event rates.” These effects help to explain the statistically significant increase in patients discontinuing etoricoxib because of hypertension-related adverse events.[9]  This is in comparison to diclofenac which, as previously mentioned, has its own risks and should perhaps be reevaluated from a safety perspective as well.  Overall the FDA showed great concern over the cardiovascular data with Dr. Graham concluding that the increased CV risk of etoricoxib presented an “enormous public health and population consequence.”[10]

Finally, in the case of serious gastrointestinal toxicity, there was no benefit to etoricoxib compared with diclofenac. The rate of serious confirmed GI events with etoricoxib was 0.30 per 100 patient years vs. 0.32 with diclofenac, not significantly different.  The EMEA’s CHMP report states that “data from e.g. CLASS trial and MEDAL programme show no significant benefit of the Cox-2 inhibitors for ‘complicated’ gastrointestinal events” and that “although most evidence suggests a gastrointestinal benefit for Cox-2 inhibitors, further data are needed.”[11]   FDA advisory committee members agreed with this finding and stated that any GI advantage was related to incidence of dyspepsia and uncomplicated ulcers,[12] though even these effects are reduced or eliminated in patients taking aspirin/anti-platelet agents.[13]  The benefit of preventing these minor disorders, while not insignificant, is certainly outweighed by the increased cardiovascular risk posed by the drug.  This point was illuminated by Dr. Felson of the AAC who calculated, using conservative sponsor estimates, that compared to naproxen the number needed to kill due to a cardiovascular event would be 600 and the number needed to prevent a GI death would be 1200, thus using etoricoxib “you kill twice as many people as you save.”[14]

Importantly, it has also come to light recently that the GI complications of NSAID use are not as much of a problem as had previously been assumed.  During the Arthritis Advisory Committee (AAC) meeting Dr. Fries clarified that the often quoted number of 16,500 deaths a year due to upper GI bleeding from the use of all NSAIDs combined, which had emanated from his research, was highly inflated and that the actual number would be lower than 2,500.[15]

The case of etoricoxib seems to be a clear example of a drug with an unfavorable risk benefit profile.  The drug poses significant cardiovascular risks, in the absence of benefit in efficacy or reduced serious GI complications compared to older NSAIDs. As mentioned above, members of the FDA advisory committee were of the opinion that there was no specific patient population that would clearly benefit from the drug.[16]

Thousands, probably tens of thousands of patients have already had needless heart attacks because they took one of the marketed or previously marketed COX-2 drugs instead of clearly safer alternatives.  When the question of cardiovascular risk in already approved NSAIDs was raised Dr. Jenkins of the FDA told the AAC in their deliberation of etoricoxib “as you obtain more information, you have to apply that new information…  Science has changed,” he asked “what should our regulatory position be today in 2007, given what we know about cardiovascular risk, about GI benefit, about tolerability, about benefit as far as efficacy compared to other therapies?”[17]  The AAC deliberated this point and decided that etoricoxib’s risks far outweighed its benefits to the point that it was unfit for use.   We urge the EMEA to ask this same question. Taking into account the new information that has come to light since the drug was approved in 2002, we believe the EMEA will arrive at the same conclusion and quickly remove etoricoxib from the market.


The FDA advisory committee assembled to review rimonabant for obesity treatment in the United States expressed an utter lack of confidence in the available data characterizing the safety and risk-benefit profiles of the drug,[18] which has already been used by over 130,000 people in the EU.[19]

One major concern of committee members was the low quality and ambiguity of the current data.[20]  The four published RIO trials, upon which all of Sanofi’s efficacy and much of their safety conclusions rely, were plagued by discontinuation rates of 32%-49%.[21],[22],[23],[24]  Combined with other questions that have arisen regarding the methodological quality of the four studies including method of randomization, allocation concealment, and blinding,[25] high attrition rates serve to throw conclusions on safety and efficacy into doubt.

Flaws aside, however, the information from these trials still shows a worrisome incidence of adverse events caused by rimonabant.  Most notably 26% of rimonabant 20-mg treated subjects vs. 14% of placebo treated subjects experienced a psychiatric symptom as an adverse event. Rimonabant was seen to cause a statistically significant increase in the incidence of depression, anxiety, insomnia, and panic attack[26] including a “doubling of risk of depression in the subgroup of patients who did not even have a previous history of psychiatric illness” according the FDA advisory committee.[27]  The evidence for increased depression is of particular concern for a drug targeted towards the obese, a population that has been shown to have a significantly higher incidence of depression and eating disorders compared to non-obese individuals.[28],[29]

The biggest safety concern, however, regards strong evidence that rimonabant causes an increase in suicidality including suicidal tendencies.  FDA meta-analysis of the rimonabant trials for the advisory committee showed an 80-100% increase in relative risk for suicidal ideation for the rimonabant groups, representing a .3% increase in absolute risk.  This finding was seen as biologically plausible, based on current knowledge of the CB1 receptor, and consistent with the increased incidence of depression in the clinical trials following use of the drug.[30]  FDA safety and efficacy reviewer Dr. Amy Egan stated of the link between rimonabant and psychiatric disorders associated with suicidality “we strongly believe that it is causal.”[31]

Dr. Egan also stated that one patient would be harmed (by causing an episode of suicidal ideation) for every three hundred treated.   She noted that this number may even be low “given that: higher percentages of rimonabant patients dropped out of studies due to psychiatric adverse events.”[30]

Another issue with the information taken from the studies done thus far is the lack of long term safety data.  Of the studies performed to date, two had durations of two years, while the other two were one year.  Because rimonabant is the first drug of its class, there is no data evaluating the long term effect of antagonizing the widespread cannabinoid system.  Weight lost while using rimonabant is regained after discontinuation of use,[32],[33] which means that if the drug is to be effective at all it will have to be prescribed on a long term basis.  Given this fact, the complete lack of data on rimonabant use in humans over an extended period of time is cause for significant concern.

The insufficiency of data is particularly alarming in the context of a drug that is centrally acting, modulating the effects of a widespread and relatively little studied neurotransmitter system.  CB1 receptors have been observed in olfactory and cortical regions of the brain (neocortex and pyiform cortex), hippocampus, and amygdala, basal ganglia, thalamic and hypothalamic nuclei, cerebellar cortex, and brain stem nuclei as well as peripheral areas of the autonomic nervous system, liver, muscle, gastrointestinal tract, and adipose tissue,[34] pituitary gland, immune cells and reproductive tissues.[35]  Animal data suggests that abnormal CB1 receptor function can result in reproductive malfunction,[36],[37],[38],[39] increased mortality,[40] and adverse cardiovascular effects.[41]  It is reasonable to think these effects might be relevant at human dosing levels because, as Dr. Karen Davis-Bruno of the FDA explained, “there are limited, if any, differences between exposures generating the desired pharmacologic effect and those associated with significant animal toxicity.”[42]

A recent thorough review of rimonabant by the Cochrane database of systematic reviews concluded that 1) that average weight loss is “modest” and 2) more rigorous studies of efficacy and safety are required to “fully evaluate the benefit risk ratio of this new drug.”[43] This is an alarming characterization for a drug that is currently approved in 42 countries and marketed in 20.[44]  Until such additional studies are presented to allay concerns about the possible dangers of rimonabant use, we firmly believe that the general public should not be exposed to rimonabant.

In summary, we urge that the EMEA review the latest data guiding the FDA decisions not to allow U.S. marketing for either etoricoxib or rimonabant. It is highly likely, if not certain, that when you do you will decide to spare Europeans from these two drugs that have been thought too dangerous, relative to their benefits, to be allowed on the U.S. market.


Sidney M. Wolfe, M.D.
Public Citizen Health Research Group

Benjamin Wolpaw
Staff Researcher

[1] "Press Release European Medicines Agency Concludes Action on COX-2 Inhibitors." 27 June 2005.

[2] FDA Arthritis Advisory Committee Meeting Acroxia 20 and 60 Mg for Symptomatic Treatment of Osteoarthritis Briefing Document. Merck. 2007.

[3] Arcoxia (Etoricoxib). Arthritis Advisory Committee Meeting, 12 Apr. 2007, U.S. Food and Drug Administration (CDER). Capital Reporting Company.

[4] Public CHMP Assessment Report for Medicinal Products Containing Non-Selective Non Steroidal Anti-Inflammatory Drugs. EMEA. 2006.

[5] Psaty, Bruce, and Noel Weiss. "NSAID Trials and the Choice of Comparators - Questions of Public Health Importance." NEJM (2007).

[6] AHA recommendations: Antman, et al.   Circulation.  March 2007.

[7] Graham, David J. "An Epidemiological Perspective on Etoricoxib." FDA. AAC. 12 Apr. 2007.

[8] FDA Presentation at 2/05 meeting

[9] Public CHMP Assessment Report for Medicinal Products Containing Non-Selective Non Steroidal Anti-Inflammatory Drugs. EMEA. 2006.

[10] Graham, David J. "An Epidemiological Perspective on Etoricoxib." FDA. AAC. 12 Apr. 2007.

[11] Public CHMP Assessment Report for Medicinal Products Containing Non-Selective Non Steroidal Anti-Inflammatory Drugs. EMEA. 2006.

[12] Arcoxia (Etoricoxib). Arthritis Advisory Committee Meeting, 12 Apr. 2007, U.S. Food and Drug Administration (CDER). Capital Reporting Company.

[13] Public CHMP Assessment Report for Medicinal Products Containing Non-Selective Non Steroidal Anti-Inflammatory Drugs. EMEA. 2006.

[14] Arcoxia (Etoricoxib). Arthritis Advisory Committee Meeting, 12 Apr. 2007, U.S. Food and Drug Administration (CDER). Capital Reporting Company.

[15] ibid

[16] ibid

[17] Arcoxia (Etoricoxib). Arthritis Advisory Committee Meeting, 12 Apr. 2007, U.S. Food and Drug Administration (CDER). Capital Reporting Company.

[18] Summary Minutes. FDA Advisory Committee Meeting, 13 June 2007, FDA.

[20] Summary Minutes. FDA Advisory Committee Meeting, 13 June 2007, FDA.

[21] Van Gaal, Luc F., Aila M. Rissanen, Andre J. Sheen, et al. "Effects of the Cannabinoid-1 Receptor Blocker on Weight Reduction and Cardiovascular Risk Factors in Overweight Patients: 1-Year Experience From the RIO-Europe Study." Lancet 365 (2005):  1389-1397. 

[22] Pi-Sunyer, F, Louis J. Aronne,   and Hassan M. Heshmati. "Effect of Rimonabant, a Cannabinoid-1 Receptor Blocker, on Weight and Cardiometabolic Risk Factors in Overweight or Obese Patients (RIO-North America:  a Randomized Controlled Trial)." JAMA 295.7 (2006):  761-775. 

[23] Scheen, Andre J., Nick Finer, and Priscilla Hollander. "Efficacy and Tolerability of Rimonabant in Overweight or Obese Patients with Type 2 Diabetes: a Randomised Controlled Study." Lancet 368 (2006):  1660-1672.

[24] Despres, Jean-Pierre, Alain Golay, and Lars Sjostrom. "Effects of Rimonabant on Metabolic Risk Factors in Overweight Patients with Dyslipidemia." NEJM 353 (2005): 2121-34.

[25] Curioni, C., and C. Andre. "Rimonabant for Overweight or Obesity (Review)." Cochrane Database of Systematic Reviews (2006). 

[26] FDA Briefing Document Zimulti (Rimonabant) Tablets 20 Mg. FDA Advisory Committee - June 13, 2007.   22-28.

[27] Summary Minutes. FDA Advisory Committee Meeting, 13 June 2007, FDA.

[28] Faith, Myles S., Patty E. Matz, and Marie A. Jorge. "Obesity - Depression Associations in the Population." Journal of Psychosomatic Research 53 (2002):  935-942.

[29] Golay, Alain, Anne Laurent-Jaccard, Frank Habicht, Jean-Pierre Gachoud, Mireille Chabloz, Anne Kammer,   and Yves Schutz. "Effect of Orlistat in Obese Patients with Binge Eating Disorder." Pharmacology and Therapeutics 13 (2005):  1701-1708.

[30] Egan, Amy G. "Rimonabant Safety and Efficacy Review." FDA. Endocrinologic and Metabolic Drugs Advisory Committee Meeting. 13 June 2007.

[31] "Panel: Diet Drug's Safety Not Proven." AP (From NBC) 13 June 2007. http://today.msnbc.msn.com/id/19212295/print/1/displaymode/1098/

[32] Pi-Sunyer, F, Louis J. Aronne,   and Hassan M. Heshmati. "Effect of Rimonabant, a Cannabinoid-1 Receptor Blocker, on Weight and Cardiometabolic Risk Factors in Overweight or Obese Patients (RIO-North America:  a Randomized Controlled Trial)." JAMA 295.7 (2006):  761-775. 

[33] Prescrire Editorial Staff. "Rimonabant." Prescrire International 15 (2006):  123-126.

[34] Prescrire Editorial Staff. "Rimonabant." Prescrire International 15 (2006):  123-126.

[35] Pertwee, RG. "The Pharmacology of Cannabinoid Receptors and Their Ligands." International Journal of Obesity 30 (2006): s13-s18. www.nature.com/ijo

[37] Wang, Haibin, Huirong Xie, Yong Guo, et al . "Fatty Acid Amide Hydrolase Deficiency Limits Early Pregnancy Events." The Journal of Clinical Investigation 116 (2006):  2122-2131.

[38] Wang, Haibin, Yong Guo, Dingzhi Wang, et al . "Aberrant Cannabinoid Signaling Impairs Oviductal Transport of Embryos." Nature Medicine 10 (2004):  1074-1080.

[39] Liu, W.M., E.K. Duan, and Y.J. Cao. "Effects of Anandamide on Embryo Implantation in the Mouse." Life Sciences 71 (2002):  1623-1632.

[40] Zimmer, Andreas, Anne M. Zimmer, and Andrea G. Hohmann. "Increased Mortality, Hypoactivity, and Hypoalgesia in Cannabinoid CB1 Receptor Knockout Mice." PNAS 96 (1999):  5780-5785. 

[41] Pacher, Pal, Sandor Batkai, and George Kunos. "The Endocannabinoid System as an Emerging Target of Pharmacotherapy." ASPET 58 (2006): 389-462.

[42] Davis-Bruno, Karen. "Non-Clinical Overview: CNS Toxicity with Rimonabant." FDA. Endocrinologic and Metabolic Drugs Advisory Committee Meeting. 13 June 2007.