Rimonabant (anti-obesity drug) for prevention of cardiovascular events (CRESCENDO): a randomised, multicentre, placebo-controlled trial - no CVD benefit, psych side effects found
"The findings did not provide evidence of efficacy for rimonabant's prevention of adverse cardiovascular outcomes, but further substantiated its effect of inducing serious neuropsychiatric side-effects.....Marketing authorisation was withdrawn in October, 2008, because of psychiatric side-effects in clinical trials,4 which led to the demand for cessation of all rimonabant trials in November, 2008, by multiple countries in Europe.....The most important concern about rimonabant was related to its neuropsychiatric side-effects, and particularly whether the drug was associated with suicides. In this trial, the rate of suicide was 0·04% for the rimonabant group compared with 0·01% for the control group....The frequency of neuropsychiatric side-effects differed significantly between treatment groups (p<0·0001), and consisted mainly of heightened anxiety, depression, insomnia, dizziness, and mood alteration (table 4). The frequency of serious psychiatric side-effects also differed significantly, occurring in 232 (2·5%) of patients in the rimonabant group compared with 120 (1·3%) in the placebo group (p<0·0001).. There were four suicide events in the rimonabant group versus one in the placebo group, and nine versus five patients, respectively, attempted suicide ......Although this is a profound side-effect and patients were screened in this trial for clinical depression or mental illness, and excluded on the basis of investigator's judgment, the concern is that in clinical practice such screening might not be as rigorous and the incidence of suicides or attempts might be higher than that recorded in this trial. However, the data do not document much variance between the occurrence in randomised controlled clinical trials and observational clinical exposure."|
The Lancet Aug 14 2010
Prof Eric J Topol MD a Corresponding AuthorEmail Address, Prof Marie-Germaine Bousser MD b, Prof Keith AA Fox MBCh c, Mark A Creager MD d, Prof Jean-Pierre Despres MD e, Prof J Donald Easton MD f, Prof Christian W Hamm MD g, Prof Gilles Montalescot MD h, Prof P Gabriel Steg MD i, Prof Thomas A Pearson MD j, Eric Cohen MD k, Christophe Gaudin MD l, Bernard Job MD l, Judith H Murphy MD l, Deepak L Bhatt MD m, for The CRESCENDO Investigators
Blockade of the endocannabinoid receptor reduces obesity and improves metabolic abnormalities such as triglycerides, HDL cholesterol, and fasting blood glucose. We assessed whether rimonabant would improve major vascular event-free survival.
This double-blind, placebo-controlled trial was undertaken in 974 hospitals in 42 countries. 18 695 patients with previously manifest or increased risk of vascular disease were randomly assigned to receive either rimonabant 20 mg (n=9381) or matching placebo (n=9314). Randomisation was stratified by centre, implemented with an independent interactive voice response system, and all study personnel and participants were masked to group assignment. The primary endpoint was the composite of cardiovascular death, myocardial infarction, or stroke, as determined via central adjudication. Analysis was by intention to treat. This study is registered with ClinicalTrials.gov, number NCT00263042.
At a mean follow-up of 13·8 months (95% CI 13·6-14·0), the trial was prematurely discontinued because of concerns by health regulatory authorities in three countries about suicide in individuals receiving rimonabant. All randomised participants were analysed. At the close of the trial (Nov 6, 2008), the composite primary endpoint of cardiovascular death, myocardial infarction, or stroke occurred in 364 (3·9%) patients assigned to rimonabant and 375 (4·0%) assigned to placebo (hazard ratio 0·97, 95% CI 0·84-1·12, p=0·68). With rimonabant, gastrointestinal (3038 [33%] vs 2084 [22%]), neuropsychiatric (3028 [32%] vs 1989 [21%]), and serious psychiatric side-effects (232 [2·5%] vs 120 [1·3%]) were significantly increased compared with placebo. Four patients in the rimonabant group and one in the placebo group committed suicide.
The premature termination of this trial has important lessons for drug development. A drug that was being marketed for weight loss, but being tested for improving cardiovascular outcomes, induced a level of serious neuropsychiatric effects that was deemed unacceptable by regulatory authorities, and both the drug and the trial were abruptly terminated.
Obesity has reached epidemic proportions worldwide. Abdominal obesity, the most dangerous form, is associated with a cluster of diabetogenic and atherogenic metabolic abnormalities that are often referred to as the metabolic syndrome. Although improved lifestyle-nutritional habits and regular physical activity-is the cornerstone of management of obesity, poor clinical practice has led us to consider pharmacological support to help patients to reduce their caloric intake and bodyweight and improve their cardiometabolic risk profile.1-3 Unfortunately, physicians have been left with few options for the pharmacotherapy of obesity, especially abdominal obesity.
The endocannabinoid pathway, with its principal cannabinoid receptor expressed in the brain and adipose tissue, among other organs, has been intensively studied in experimental models and man. Rimonabant-a selective cannabinoid-1 receptor antagonist-not only reduces waist circumference and bodyweight, but also improves several metabolic risk factors by increasing concentrations of HDL cholesterol and adiponectin, and reducing triglycerides, blood glucose, fasting insulin, and leptin.1 Despite its major side-effects of nausea and depression, the efficacy of rimonabant on metabolic parameters has been recorded in many placebo-controlled randomised trials; however, duration of treatment was 12 months or less.1-3 In a study of the effects of 18 months of rimonabant treatment on atherosclerotic progression assessed by intravascular ultrasound, the primary endpoint was not achieved but total atheroma volume was reduced compared with placebo.4 No previous trial was adequately powered to assess clinical endpoints.
The Comprehensive Rimonabant Evaluation Study of Cardiovascular Endpoints and Outcomes (CRESCENDO) trial was designed to test whether long-term rimonabant therapy would have the potential to reduce the risk of cardiovascular death, myocardial infarction, or stroke in a large population with either documented cardiovascular disease or significant risk factors for development of atherosclerotic vascular disorders.
This trial was prematurely discontinued after a mean of 14 months of follow-up, as compared with the plan of a minimum of 33 months of treatment with either rimonabant or placebo. The findings did not provide evidence of efficacy for rimonabant's prevention of adverse cardiovascular outcomes, but further substantiated its effect of inducing serious neuropsychiatric side-effects.
The cumulative effect of obesity, metabolic syndrome, and diabetes poses an important health-care burden.6 The biology of the endocannabinoid pathway, with central and peripheral effect on energy balance, led to several trials suggesting that the endocannabinoid-1 blockade could improve lipoprotein profile, fasting glucose, and insulin.1-3 However, whether prolonged blockade of endocannabinoid-1 could achieve the benefit of a reduction in major adverse cardiovascular events needed to be established. Although this trial attempted to resolve this issue, and fulfilled its ambitious worldwide enrolment, it was discontinued owing to serious neuropsychiatric side-effects7, 8 that occurred both within and outside the clinical trial, in countries where the drug was commercially approved and marketed. Marketing authorisation was withdrawn in October, 2008, because of psychiatric side-effects in clinical trials,4 which led to the demand for cessation of all rimonabant trials in November, 2008, by multiple countries in Europe.
A 1-year period was anticipated for the potential benefits of rimonabant to become evident. This presumption was derived from previous trials that modulate lipoproteins, such as statins, which have generally needed 12-18 months of therapy before divergence of event curves were clearly discerned.9 The absence of any reduction of cardiovascular endpoints, despite the anticipated benefit in metabolic parameters, is disappointing. Although this trial was underpowered with the prespecified goal of having more than 1600 adjudicated primary endpoint events, more than 700 cardiovascular events occurred. Accordingly, the large number of events in this trial does provide some indication to discern whether any benefit might be possible, and the report provides new data for side-effects since the duration of rimonabant therapy was long.
The most important concern about rimonabant was related to its neuropsychiatric side-effects, and particularly whether the drug was associated with suicides. In this trial, the rate of suicide was 0·04% for the rimonabant group compared with 0·01% for the control group. Although this is a profound side-effect and patients were screened in this trial for clinical depression or mental illness, and excluded on the basis of investigator's judgment, the concern is that in clinical practice such screening might not be as rigorous and the incidence of suicides or attempts might be higher than that recorded in this trial. However, the data do not document much variance between the occurrence in randomised controlled clinical trials and observational clinical exposure.
The low incidence of serious neuropsychiatric side-effects might have been counterbalanced by increased improvement in protection from cardiovascular death, heart attack, or stroke. For thrombolytic therapy in acute myocardial infarction, a benefit in survival is noted but at the risk of inducing an intracranial haemorrhage in about one in 200 patients, which is apt to be associated with major neurological deficit. This trial documents a risk of death induced by rimonabant many orders of magnitude less than this rate, albeit in a disorder that is not immediately life-threatening with other treatment alternatives. Nevertheless, the low absolute risk of neuropsychiatric effects led to termination of the trial by the regulatory authorities in multiple participating countries, suggesting that the tolerance for risk of new molecular entities is considerably lower now than a decade or two ago. Although patients receiving rimonabant had many other side-effects, almost all these would be expected to be resolved from discontinuation of the drug. The crucial issue about a favourable trade-off of improved cardiovascular outcomes overriding serious side-effects might never be resolved, since clinical development for endocannabinoid-1 blockers from other manufacturers has also been discontinued.
With such an important cluster of diseases of obesity, metabolic syndrome, and diabetes, innovative approaches are urgently needed. Endocannabinoid blockade could have proven viable, if a genome-wide association study had been done to establish what sequence variants are linked with suicides, suicide attempts, or significant neuropsychiatric side-effects. In many such pharmacogenomic studies such as those for statin myopathy,10 platelet response to clopidogrel,11 lack of efficacy of interferon for hepatitis C virus infection,12 and hepatoxicity for lumiracoxib,13 only a few events were needed to establish the principal common gene variants associated with the outcome of interest. Many drugs with very large clinical development programmes had their development stopped because of rare side-effects, such as angioneurotic oedema with omapatrilat, or hepatitis with ximelagatran. As with this trial, knowledge about the appropriate clinical criteria for inclusion, with use of genomic data indicating high risk of side-effects at the time of study entry, could have pre-empted or at least reduced the serious psychiatric side-effects. For example, a study14 showed that single nucleotide variants in a tyrosine kinase receptor gene (NTRK2) involved in neurotrophic signalling were associated with more than four-fold increased risk of suicide. Furthermore, the clinical benefit of endocannabinoid-1 blockade could have been manifest in patients with a particular gene variant in the endocannabinoid-1 or related pathways.
In recent years, many clinical trials have been prematurely discontinued, usually because of the business interest or status of the sponsor. However, in this trial, regulatory authorities rather than the investigators, data and safety monitoring board, or sponsor, dictated its early discontinuation. In fact, in one country considering whether to discontinue all clinical research with rimonabant, the chair of this trial's data monitoring committee unsuccessfully argued the case for continuation of the trial. In this setting, the performance of a long-term trial for an expanded cardiovascular indication was affected by the track record of the drug in clinical practice that was marketed for weight loss. The lesson of putting a global research project in jeopardy because of unanticipated serious side-effects occurring with open-label commercial use is invaluable. The present regulatory environment has changed, and serious adverse events are not tolerated. Roflumilast, a new drug being considered for approval by the US Food and Drug Administration for treatment of chronic obstructive pulmonary disease, was associated with three completed suicides and two suicide attempts compared with no such events in the placebo group.15 Accordingly, the predicament with rimonabant is not an isolated scenario. The new precedent here is the capacity for such regulatory agencies to stop an ongoing clinical research project. Hopefully, the experience and data from this clinical trial will prove useful in planning future drug development.
Figure 1 shows the trial profile. 18 695 patients were enrolled: 9314 were assigned to placebo and 9381 to rimonabant. Table 1 shows baseline characteristics of participants, and table 2 the drugs that patients were taking at baseline.
The total number of participants enrolled was greater than was expected because of the very rapid accrual of patients once all participating sites had been activated. Although enrolment was completed, study drug administration and follow-up in the trial was prematurely ended on Nov 6, 2008, because of a request by the regulatory agency for cessation of clinical research of rimonabant in Ireland, France, and Germany. At that time, the mean exposure to the drug had been 13·8 months (95% CI 13·6-14·0), whereas the anticipated exposure if the trial had been completed was planned to be a minimum of 33 months.
The HR for the primary efficacy endpoint was 0·97 (95% CI 0·84-1·12, p=0·68; figure 2). 375 events occurred in the placebo group and 364 in the rimonabant group, which was less than half of the anticipated 1600 events. Table 3 shows the specific types of events for the overall intention-to-treat population, and populations with overt cardiovascular disease or at increased risk of cardiovascular disease.
Primary and secondary efficacy endpoints for the overall intention-to-treat population, and populations with overt cardiovascular disease or at increased risk of cardiovascular disease
Figure 3 shows the time-to-event curves for the primary endpoint for the two subgroups by treatment assignment. For patients with overt vascular disease the HR for rimonabant was 0·95 (95% CI 0·80-1·13, p=0·60). As postulated, the curves began to diverge after 1 year, but this effect was not sustained during longer follow-up, with a rapidly diminishing number of patients followed up after 2 years (figure 3). However, for patients without overt vascular disease, the event rates were much lower (HR 1·89, 95% CI 1·40-2·55, p<0·0001). There was no indication of any treatment effect (HR 1·03, 0·79-1·35, p=0·81). Figure 4 shows the differences between the treatment effects by subgroup for the endpoint of all-cause mortality. For the rimonabant group, all-cause mortality decreased by 8·9% (HR 0·91, 0·71-1·17, p=0·46), whereas for patients without vascular disease, all-cause mortality increased by 15·7% (HR 1·16, 0·83-1·61, p=0·39). Despite the opposing findings, there was no significant interaction of treatment by subgroup for all-cause mortality (HR 0·79, 0·52-1·19, p=0·26).
Table 4 shows the adverse effects of the study drug, with the frequency and the number of events per 100 patient-years. Gastrointestinal side-effects, particularly nausea and diarrhoea, were significantly increased with rimonabant (p<0·0001; table 4). The frequency of neuropsychiatric side-effects differed significantly between treatment groups (p<0·0001), and consisted mainly of heightened anxiety, depression, insomnia, dizziness, and mood alteration (table 4). The frequency of serious psychiatric side-effects also differed significantly, occurring in 232 (2·5%) of patients in the rimonabant group compared with 120 (1·3%) in the placebo group (p<0·0001). There were four suicide events in the rimonabant group versus one in the placebo group, and nine versus five patients, respectively, attempted suicide.
a Scripps Translational Science Institute, La Jolla, CA, USA
b Hopital Lariboisire, Paris, France
c University of Edinburgh, Edinburgh, UK
d Brigham and Women's Hospital and Harvard University, Boston, MA, USA
e Centre de Recherche, Institut Universitaire de Cardiologie et de Pneumologie de Quebec, Quebec, Canada
f Rhode Island Hospital and Brown University, Providence, RI, USA
g Kerckhoff Klinik, Bad Nauheim, Germany
h Centre Hospitalier Universitaire Pitie-Salptrire, Paris, France
i INSERM U-698, Universite Paris 7 and Hopital Bichat, AP-HP, Paris, France
j University of Rochester, Rochester, NY, USA
k Sunnybrook Health Sciences Centre, Toronto, ON, Canada
l Sanofi-Aventis, Paris, France
m VA Boston Healthcare System, Boston, MA, USA
Corresponding Author Information Correspondence to: Prof Eric J Topol, Scripps Translational Science Institute, 3344 North Torrey Pines Court, La Jolla, CA 92037, USA
The webappendix lists the investigators and participating sites in the CRESCENDO trial