Junkfood Science: Crystal balls and “forward looking statements”

November 23, 2007

Crystal balls and “forward looking statements”

This past week, the financial news reported that retirement investors had sadly been taken in by the “forward looking” statements** released by a drug company projecting enormous profit potentials for a revolutionary weight loss drug in development.

The press statements, like the glowing media reports taken from them, were marketing, of course. A critical look at the actual clinical trial evidence on the drug, the medical literature and the history of every diet drug to date, demonstrated a far less optimistic picture. This is another tragic illustration of why it is important to look critically at the health information we hear, to understand science and go to the original sources to examine the actual study data for ourselves. Basing decisions on the interpretations presented in an abstract, medical meeting or press release can cost more than our savings, but possibly even jeopardize our lives and well-being.

More recent developments are leading to this drug being considered as a treatment for diabetes, heart disease and ‘metabolic syndrome,’ with far more enormous potential profits at stake. The promotions have already begun in the media and literature, making an understanding of the clinical evidence of special importance.


Concerning those investors, the lawyers and courts will battle their case out. They have a law firm, Coughlin Stoia Geller Rudman & Robbins LLP — one with 180 lawyers who’ve successfully won more than $45 billion in aggregate recoveries on behalf of investors — and have filed a class action lawsuit against Sanofi-Aventis, the third largest pharmaceutical company in the world. The complaint filed on November 13 with U.S. District Court, southern district of New York, states that:

[Sanofi] made false and misleading statements and engaged in a scheme to deceive the market and a course of conduct that artificially inflated Sanofi’s stock price and operated as a fraud or deceit on class period purchasers of Sanofi stock by misrepresenting the company’s business.

Their case charges that they weren’t told that Acomplia® (rimonabant, aka Zimulti, Slimona) increases depression. After the FDA Endocrinologic and Metabolic Drugs Advisory Committee’s decision to not recommend Acomplia for approval on June 13, 2007, the company's stock price dropped like a rock.

Coincidentally this past week, a review of the clinical trials looking at Acomplia’s efficacy and safety, was published in the journal Lancet. In this article, Danish researchers, led by Robin Christensen of the Parker Institute, Musculoskeletal Statistics Unit, Frederiksberg, looked at the data on the four randomized, controlled trials that had compared the full dose of Acomplia (20-mg) to placebo. They found that the patients on Acomplia were 2.5 times as likely to have depression and 3 times as like to have anxiety (as measured on the Hospital Anxiety and Depression Scale) that was so severe they had to be taken off of the trial by the investigators.

In an editorial in this same issue of Lancet, doctors at the University of New South Wales wrote that these findings “provide compelling evidence that rimonabant is associated with development of severe adverse psychiatric events.”

These findings are especially striking since people who had a history of serious depression or other psychiatric illnesses had been excluded before study entry,” the editorial noted. Part of the clinical trial protocols mandated that patients, who had been carefully selected before they were eligible to participate, be taken off the trial if they developed a need for antidepressant treatment.

Sanofi’s latest press statement contended that in this review, “extrapolation of certain data appears to us to be the sole result of the authors’ opinions.”

These authors’ conclusions, however, were far more than their opinions. The full dose of Acomplia has been shown in every published clinical trial sponsored by the drug company — as well as several meta-analyses and expert reviews — to be consistently associated with increased risks for serious psychiatric side effects, suicide ideation, depression and anxiety compared to those taking a placebo, as well as having a clear biological explanation. Similarly, the potential health benefits seen to date haven’t been much to write home about, either. The findings of this latest analysis, however, don’t begin to tell the full story.

So, let’s start from the top.


Brief background overview

Last December, as you may remember, the media was saturated with stories of this new “miracle weight loss drug.” Dr. Lou Aronne, M.D., one of the primary investigators for RIO-North America, featured prominently on CBS. It was part of the company’s marketing efforts in anticipation of FDA approval this spring and where JFS picked up the story as a new blog. But the saga began years earlier.

Sanofi-Aventis had submitted an Investigational New Drug application to the FDA Division of Metabolism and Endocrinology Products (DMEP) in May 1999 to begin exploring use of Acomplia in humans.

As a selective cannabinoid CB1 receptor antagonist, it was initially pursued as a treatment for nicotine dependence and other addictive behaviors in humans, as it appeared to suppress cravings for nicotine and food in rodents. Three randomized controlled clinical trials on humans were conducted, examining its efficacy for nicotine dependence and smoking cessation: the STRATUS studies (Studies with Rimonabant and Tobacco use): STRATUS-US and STRATUS-EU were 10-week studies and STRATUS-WW was a one-year study. The results have never been published, even years later, but were only presented as abstracts, posters or oral presentations at scientific meetings. However, in a review of Acomplia for the American Heart Association, published last year in the journal Circulation, Kishore M. Gadde, M.D., David B. Allison, Ph.D., concluded: “Abstinence rates in these RTCs [randomized controlled trials] were not highly impressive.” On February 17, 2006, Sanofi reported that the FDA had issued a nonapproval letter for use of Acomplia for smoking cessation.

With that dead end, Sanofi submitted a New Drug Application in April 2005 for Acomplia as a weight loss drug, as well as a treatment for type 2 diabetes, dyslipidemia and metabolic syndrome. According to the FDA, this application was accompanied by data from 36 Phase I studies, 5 Phase II studies, and 8 Phase III studies. [The phases of studies were explained here.] Only four completed Phase III trials were submitted. These were the ones that have been conducted on Acomplia for weight loss to date: the RIOs (Rimonabant In Obesity) — RIO-North America, RIO-Europe, RIO-Lipids and RIO-Diabetes.

The FDA Endocrine and Metabolic Drugs Advisory Committee met on June 13, 2007 to review the safety and efficacy data and make recommendations to the FDA before its final ruling on Acomplia for weight loss, scheduled for the following month. The FDA advisory panel did not recommend Acomplia for approval, citing safety concerns that the risks for adverse psychiatric effects outweighed its benefits for weight loss. In fact, the FDA Advisory Committee also said, based on the efficacy data from the RIO trials, they did not believe that the data “should be viewed as supporting a specific indication for rimonabant as a primary treatment of type 2 diabetes, dyslipidemia, or the metabolic syndrome,” either.

Sanofi quickly withdrew its FDA application before the FDA final ruling. By withdrawing their application, they could avoid outright rejection of the drug by the FDA and would be able to modify their application and come back and try again. In fact, they are currently working to develop a case for rimonabant as a treatment for diabetes, cardiovascular disease and metabolic syndrome. There are Phase III trials underway involving cardiovascular disease (CRESCENDO, AUDITOR, STRADIVARIUS, VICTORIA and ADAGIO-Lipids) and type 2 diabetes (SERENADE, ARPEGGIO and RAPSODI).

JFS looked at RIO-North America as an example, but all of the RIO trials shared similar methodological concerns that are important to understand so that we can more accurately understand why the risks reported by several meta-analyses and expert reviews — including the American Heart Association, Cochrane, the FDA, and the Europe Medicines Agency — actually went so far as to note that the studies likely underestimate the adverse effects and overstate the benefits. Let’s take a critical look at the RIOs.


A few of the major design and methodological concerns

1. Short-term trial period. The duration of any weight loss study needs to be long enough to demonstrate effectiveness for weight loss, meaning go beyond the period of inevitable weight regain. A study also needs to be long enough to determine if the associated changes in health indices (seen during any type of weight loss) hold and to enable reviewers to weigh the long-term complications seen after weight rebound. Most studies, however, stop before these adverse outcomes manifest themselves. As all experts and scientific reviews of the clinical evidence of weight loss products and diets — including the National Institutes of Health, U.S. Preventive Service Task Force, National Academies of Science and the Federal Trade Commission — have concluded, weight regain by year five is the rule, not the exception, and any weight loss intervention under five years cannot be taken as credible evidence for effectiveness.

The RIO trials were only one and two years.

As the FDA Advisory Committee said:

[T]he FDA now seeks demonstration of long-term efficacy with the recognition that patients regain weight quickly after stopping these drugs and that no significant long-term health benefits may be achievable with short-term treatment of obesity.

The Cochrane review of the randomized clinical trials evaluating full dose Acomplia for weight loss compared to a placebo, published on October 18, 2006, found that Acomplia “caused significant more adverse effects both of general and serious nature, especially of nervous system, psychiatric or gastro-intestinal origin.” It also concluded that the reported effects on weight, blood lipids (cholesterol) and blood pressure were not “clinically relevant.” It added:

The observed results should be interpreted with some caution, since the evaluated studies presented some deficiencies in methodological quality. Studies with longer follow-ups after the end of treatment and of more rigorous quality should be done before definitive recommendations can be made regarding the role of this new medication in the management of overweight or obese patients.

2. Extraordinarily high drop-out rates and numbers lost in followup. When a clinical trial has high attrition rates, it is easier to make the treatments appear more effective and with fewer side effects. Most people participating in clinical trials are highly motivated — this would be especially true for a drug promising miracle weight loss, as most obese people have desperately tried every weight loss treatment imaginable — and most who drop out of trials do so because the treatment isn’t working and/or the side effects are so unpleasant they can’t continue. Many researchers will also employ “last observed carried forward” in their statistical analysis, which is especially flawed in weight loss studies as it enables them to show patients maintaining weight loss and their initially improved health indices, rather than the inevitable rebounds with long-term follow-up.

In RIO-North America, for example, the Sanofi researchers started with 1,222 patients in the arm receiving the full 20-mg dose, yet only were able to report on 323 of them left at year two. That’s a mere 26% who completed the study arm!

Similarly, only 67% completed the full-dose arm of RIO-Diabetes at one year.

Overall, only 44.7% of the 2,2176 participants in the full dose arms in all of the RIO trials completed the studies.

There is another reason that attrition and how the RIO results were reported would make the psychiatric and other adverse effects reported likely to be underestimations. The Lancet meta-analyses explained that those on the full dose were 2.5 times more likely to discontinue because of depression and 3 times more likely to stop because of anxiety than a placebo. BUT, these refer only to the serious adverse reactions for which a reason was recorded by the investigators. These “Adverse Events” were those where the researchers took the participants off of the study because, most often, the severity of their side effects made them ineligible to continue per the trial protocol — such as needing treatment for depression or psychiatric problems. These adverse events with known explanations made up less than half (47.7%) of all those who dropped out of the RIO trials.

Of the others in the treatment group who didn’t complete the RIO trials:

33% requested to leave (the investigators didn’t reveal the participants’ reasons)

7% were “lost in follow-up”

6% stopped taking the trial drug (the investigators didn’t reveal their reasons)

4% left because it wasn’t working

2% other reasons (not described)

The big unanswered question is why over half of the participants in the RIO trials dropped out. Several independent reviewers have been highly critical of these trials because no efforts were made to follow-up and evaluate the long-term weight changes, health indices or determine why so many didn’t complete these studies. “It is a mystery to me why they don’t make a more concerted effort to measure the people who dropped out,” said Dr. Denise Simons-Morton, Director of the Clinical Applications and Prevention Program, Division of Epidemiology and Clinical Applications at the U.S. National Heart, Lung, and Blood Institute.

3. Designing a trial that isn’t a “fair test” of the intervention. As in most pharmacological weight loss studies, the drugs are almost always tested in combination with diet and exercise. So, it’s impossible to reliably estimate the changes due to the drug versus caloric restrictions or exercise. As the American Heart Association Acomplia reviewers noted, “in the absence of these ancillary interventions and the structure of a clinical trial, one would suspect that these drugs might be even less effective in real-life clinical practice settings than in clinical trials.”

The Lancet reviewers expressed similar concerns, adding that the “intensity and the enforcement of the restriction in calories in the RIO programme was fairly weak, and produced only modest weight loss of less than 2 kg in the placebo groups.” The weight loss attributed to Acomplia would be considerably less in a more calorie-restricted cohort, meaning “the additional weight loss produced by rimonabant would probably be less than that we reported.”

Similarly, the health indice changes attributed to Acomplia, when lumped with a diet and exercise program resulting in temporary weight loss, are also questionable. In RIO-Diabetes, for example, the full dose over placebo showed an absolute (actual) 0.7% fall in HbA1c, in the context of weight loss. This resulted “in the intriguing claim by the researchers that 57% of this improvement (together with an improvement of much the same size in the concentration of HDL cholesterol) was independent of weight loss,” wrote doctors Stephen J Cleland and Naveed Sattar in a Lancet editorial review of the study. However, they cautioned:

the effect of rimonabant over-and-above weight loss merits closer scrutiny, because these data were generated by a statistical model that will be difficult to follow for most readers and which might have made invalid assumptions, including a disregard of weight loss in the run-in period.

The same caveats are likely in order for the other Acomplia diabetes studies. Sanofi press releases have already been reporting, before the SERENADE study has been published, that Acomplia “significantly” lowered HbA1c levels by 0.5% over placebo.

4. Use of false surrogate endpoints versus report actual clinical outcomes or mortality data. There are a number of health indices (serum glucose, cholesterol levels, and blood pressure) that change during weight loss, as they do with serious stresses and illnesses. But, these health indices are also well-known to rebound, often reaching higher levels, after the weight loss intervention has been stopped.

The average weight lost at one year among all four of the RIO trials was 10.3 pounds among those taking the full dose over those on a placebo. (The FDA found the 5-mg dose not effective at all for weight loss.) That’s 0.19 pound a week attributed to Acomplia. But, like all weight loss studies, the RIO trials show striking regains nearly to baseline among the two-year data reported. “No follow-ups were reported after discontinuation of active treatment, thus any [additional] weight regain could not be assessed,” said the Lancet reviewers.

Why the FDA Advisory Committee found no support among the RIO trial data for Acomplia as a treatment for type 2 diabetes, dyslipidemia, or the metabolic syndrome is also clearer when one looks at the actual trial data. As the American Heart Association reviewers noted, not only was the weight loss efficacy no better than the modest effects observed with other anti-obesity drugs; in all four RIO trials, Acomplia “led to a negligible reduction in systolic blood pressure...no reductions in total cholesterol or LDL-cholesterol...The National Cholesterol Education Program guidelines emphasize that lowering LDL-C is of primary importance in reducing cardiovascular morbidity and mortality.”

RIO-Diabetes, for example, enrolled 1,047 diabetics with stable weight (BMIs 27- 40) who had required medications (metformin or sulphonylurea) for at least 6 months. Among the two-thirds who completed the one-year intervention period on the full dose, compared to the placebo group, actual total cholesterols changed 0.06 mmol/L, LDL-cholesterol 0.04 mmol/L, fasting insulin lowered 0.3 µIU/mL, and HbA1c fell 0.7%. There was no change in the numbers diagnosed with metabolic syndrome, with 26% showing improvement and 27% developing metabolic syndrome.

The drug also did not show any benefits on diabetes medication needs, concluded the FDA Advisory Committee. No changes were seen among 75.9% among the intervention group, 11.3% needed increased medications and 11.9% saw dosages lowered.


Adverse effects — how significant are the psychiatric and neurological side effects?

Earlier this year, Acomplia was turned down for reimbursement by health insurance in Germany, the European Union’s largest market. It also failed to get approval for reimbursement as a weight loss drug in France. The European Medicines Agency has allowed Acomplia to be on the market, but their review found it doubled risks for depression “in all types of patients,” leading to the addition of strong warning labels and mandating that it “must no longer be used in patients [taking antidepressants].” In the U.K, since Acomplia’s release in June 2006, 364 psychiatric reactions have been reported, the Department of Health's Medicines and Healthcare Products Regulatory Agency (MHRA) said on July 19th.

The most comprehensive review of the adverse reactions seen in the preclinical and clinical evidence on Acomplia was done by the FDA Endocrinologic and Metabolic Drugs Advisory Committee. Its 88-page Briefing Document concluded:

Among the most significant adverse events throughout the Phase 3 program were those in the primary System Organ Class Psychiatric Disorders, specifically depressive events, anxiety, psychomotor agitation, and sleep disorders. In the pooled RIO studies, for subjects receiving the same treatment during the whole study, 26% of rimonabant 20-mg treated subjects vs. 14% of placebo treated subjects experienced a psychiatric symptom reported as an adverse event. Specifically, 9% of rimonabant 20-mg treated subjects vs. 5% of placebo treated subjects reported symptoms of depression (depressed mood; depression; depressive symptom; or major depression).

Neurological symptoms “appeared commonly” in the clinical studies and were more frequent among those receiving the full dose compared to placebo, noted the FDA. These included sensory changes, motor impairments, and cognitive difficulties, but the trial reports didn’t characterize them well or evaluate them in detail. “Motor impairment occurred with greater frequency in the rimonabant 20-mg group than in the placebo group – 1.7% and 0.12%, respectively – and was driven predominantly by 'tremor' and 'balance disorder.'" Neurological disorders were seen in 14.3% of the Rio treatment participants compared to 9.4% in placebo; and mental impairment and memory problems were reported in 2.1%, compared to 1.3% in placebo. “In RIO-Diabetes and in SERENADE,” the FDA said, “approximately 5% of rimonabant 20-mg treated subjects vs. 1.2% of placebo treated patients experienced paresthesia, dysaesthesia, or hypoaesthesia.”

Sanofi had been asked by the DMEP to submit additional data on specific psychiatric events. Adverse event reports forms don’t record multiple symptoms when they occur, and the additional data proved by Sanofi, said the FDA, “included 155 additional symptoms associated with depressed mood disorders and 208 additional symptoms of anxiety disorders in subjects receiving rimonabant 20-mg versus 49 and 51, respectively in the placebo group. Overall, the number of associated symptoms reported by subjects reporting a psychiatric adverse event was 427 in the rimonabant 20-mg treated group and 118 in the placebo-treated group.”

Overall, in the four RIO trials, the relative risks for both psychiatric adverse events and incidence of suicidality in the full dose versus placebo group were 1.9, concluded the FDA Advisory Committee. Relative risks for neurological adverse events were 3.1 among diabetics.

Because of the lack of follow-up of the roughly half of patients who withdrew from the trials needing anti-depressant medications, “the results of the above analyses should be viewed as incomplete at best and at worse as an underestimate of rimonabant’s risk for suicidality,” said the FDA.

The FDA committee also had access to additional safety and adverse event reports for the ongoing trials not yet published. Data from ongoing trials as of December 18, 2006 found 11 cases of suicidality among the treatment group, including one completed suicide, and 6 cases of suicidability among the placebo groups. Since then, the FDA said it has received two additional reports among the treatment group, one of a suicide attempt and another of psychiatric homicidal ideation. “Subsequent to the sponsor’s submission of the safety update in March 2007,” said the FDA, they’ve received three more cases of suicide attempts or ideations among the CRESCENDO trial and another in the RAPSODI trial.

Thus far, there have been two completed suicide deaths among those taking Acomplia – one in RIO-North America and one in the ongoing STRADIVARIUS study.

Both the Lancet reviewers and the FDA caution that these adverse reactions are significant, especially given these cases represent only a fraction of those with known outcomes. These are additionally notable because they were seen among a carefully screened study population that excluded those with depression or a history of depression or psychiatric illness, and will likely underestimate the risks in a real-life setting.

As the Lancet authors concluded their analysis: “Taken together with the recent US Food and Drug Administration finding of increased risk of suicide during treatment with rimonabant, we recommend increased alertness by physicians to these potentially severe psychiatric adverse reactions.”


© 2007 Sandy Szwarc. All rights reserved.


* [from Sanofi press releases]:

Forward Looking Statements

This press release contains forward-looking statements as defined in the Private Securities Litigation Reform Act of 1995. Forward-looking statements are statements that are not historical facts. These statements include financial projections and estimates and their underlying assumptions, statements regarding plans, objectives and expectations with respect to future operations, products and services, and statements regarding future performance. Forward-looking statements are generally identified by the words “expect,” “anticipates,” “believes,” “intends,” “estimates,” “plans” and similar expressions. Although sanofi-aventis’ management believes that the expectations reflected in such forward-looking statements are reasonable, investors are cautioned that forward-looking information and statements are subject to various risks and uncertainties, many of which are difficult to predict and generally beyond the control of sanofi-aventis, that could cause actual results and developments to differ materially from those expressed in, or implied or projected by, the forward-looking information and statements. These risks and uncertainties include those discussed or identified in the public filings with the SEC and the AMF made by sanofi-aventis, including those listed under “Risk Factors” and “Cautionary Statement Regarding Forward-Looking Statements” in sanofi-aventis’ annual report on Form 20-F for the year ended December 31, 2004 [2005, 2006 or 2007, varying by press release date]. Other than as required by applicable law, sanofi-aventis does not undertake any obligation to update or revise any forward-looking information or statements.

Bookmark and Share