And that dropout rate occurred despite the fact that a year of treatment with 20 mg of Acomplia was associated with an average weight loss of almost 14 pounds and a waist reduction of 2.4 inches, versus just under four pounds and less than an inch in the placebo group (P<0.001 for both).

Moreover, patients treated with 20 mg of Acomplia and calorie reduction for a year increased HDL cholesterol by 12.6% versus a 5.4% increase in the placebo arm (P<0.001) and decreased triglycerides by 5.3% versus a 7.9% increase for patients taking placebo (P<0.001).

In the year two, 40% of patients maintained on 20 mg achieved "a weight loss of 5% or greater" compared to 19% of patients in the placebo arm (P<0.001) and 17% achieved weight loss of 10% or more versus 8% in the placebo arm (P<0.001).

Moreover, after two years patients taking 20 mg of Acomplia were able to maintain a waist measurement that was about two inches smaller than baseline measurements, compared with a reduction of just 0.87 inches for those in the placebo arm (P<0.001).

Nonetheless, the high attrition rate led F. Xavier Pi-Sunyer, M.D., of the obesity center at St. Luke's-Roosevelt Hospital here and colleagues to conclude that more study is needed to finally confirm a long-term benefit for Acomplia, which is a cannabinoid-1 receptor blocker.

The RIO-North America, which is one of a trio of RIO phase III trials (including RIO-Lipids and RIO-Europe), recruited 3,045 volunteers with a BMI of more than 27 as well as hypertension and dyslipidemia. Randomization occurred between August 2001 and April 2004.

After a four-week placebo-plus-reduced calorie diet (participants were instructed to cut their normal daily caloric intake by 600 calories) run in, patients were randomized to placebo, 5 mg Acomplia, or 20 mg of Acomplia for a year.

For year two patients in the two Acomplia arms were re-randomized to the same Acomplia dose or to placebo. Patients originally randomized to placebo continued on placebo.

Six hundred and two of the original 1,216-patient 5 mg Acomplia group were re-randomized for year two, as were 660 of the original 1,222 patients assigned to 20 mg of Acomplia, while 299 of the original 607 placebo patients continued into year two.

The final analysis included 292 patients from the original placebo arm.

In addition, there were data from 590 patients in the 5 mg arm-including 294 placebo patients-and from 651 patients in the 20 mg arm, including 323 who were randomized to placebo for year two.

In an editorial that accompanied the study, Denise G. Simons-Morton M.D., Ph.D. and colleagues of the National Heart, Lung, and Blood Institute in Bethesda, Md., wrote that an "overriding concern is the failure to obtain final weight measurements on about half of the randomized participants."

They noted that the dropout rate in the RIO-North America trial was similar to that reported for other weight loss drugs such as Xenical (orlistat) and Meridia (sibutramine). By contrast, a meta-analysis of life-style intervention trials reported an average drop-out rate of less than 5%.

The RIO-North America investigators reported that the most significant side effect was nausea (about 11% of the Acomplia group, versus 5.8% of the placebo group).

But, the editorial writers pointed out that patients taking 20 mg of Acomplia reported a 2.7-fold increase in the rate of psychiatric disorders than those taking placebo.

Drugs like Acomplia should be considered within a broader context that includes lifestyle modifications and "their current role should be limited pending further evidence," said the editorialists.

Pi-Sunyer FX et al "Effect of Rimonabant, a Cannabinoid-1 Receptor Blocker, on Weight and Cardiometric Risk Factors in Overweight or Obese Patients RIO- North America, A Randomized Controlled Trial JAMA 2006;295:761-775

Simons-Morton DG et al "Obesity Research-Limitations of Methods, Measurements, and Medications." JAMA 2006; 295:826-828