Pharmacotherapy for Obesity

In the summer of 1957, my pediatrician became so concerned about my weight that he convinced my mother that I should begin taking pills to curb my appetite. I was 7 years old and weighed 70 pounds with a goal of reaching 60 pounds. I still remember how those pills made me feel; I had no appetite and felt chronically anxious. Most remarkable to me that summer was that my hands were always cold. I did reach my target weight but the chronic anxiety continued and the pills were stopped. I promptly regained all the weight that I had lost.

Since my childhood in the 1950s obesity has become substantially more prevalent in the United States,¹ despite high levels of reported efforts to lose weight.² Paradoxically, over this same period, total and cardiovascular mortality have decreased, along with the prevalence of hypertension and hypercholesterolemia.³ Such trends may support the contention that obesity is not a health problem but merely represents a cosmetic one.⁴

Other observations, however, raise doubts about the benign nature of the secular trend in obesity. It now appears that the prevalence of type 2 diabetes mellitus in the United States has increased over the past 10 years.⁵ New evidence also suggests that during this same period the incidence of a first myocardial infarction has not declined in whites, and has increased in blacks, especially in black women.⁶ In addition, there is greater awareness that obesity contributes significantly to other conditions that increase cardiovascular risk (eg, sleep apnea) and seriously erode quality of life (eg, osteoarthritis, back pain). These factors, together with the preoccupation for a leaner appearance, indicate that there will continue to be a large market for obesity treatments.

How far has pharmacotherapy for obesity evolved in the past 40 years? Fortunately, the amphetamine-based anorexiants to which I was exposed are now rarely prescribed. In addition, there is now an emerging consensus that, like other medications for chronic health conditions, long-term treatment with weight loss drugs is preferable to short-term treatment. The journey to successful long-term treatment has been difficult, however. The serotonin-based anorexiants, dexfenfluramine hydrochloride, and fenfluramine hydrochloride, were withdrawn from the market last year because of their association with cardiac valvulopathy.⁷ Last year a serotonin-norepinephrine reuptake inhibitor, sibutramine hydrochloride, became available in the United States. Safety studies suggest that sibutramine is not associated with cardiac valvulopathy, but in some patients it increases blood pressure.⁸ Over a 1-year period at the recommended dosage of 10 mg daily, the placebo-adjusted effect (drug effect minus placebo effect) of sibutramine was approximately an additional 2.7-kg weight loss.⁸

The newest candidate drug for the US market is orlistat, a gastrointestinal lipase inhibitor that prevents absorption from the gut of about 30% of dietary fat. In 1998, 2 trials of orlistat were published. The first study included nondiabetic obese northern Europeans,⁹ and the second study included obese Americans with type 2 diabetes mellitus.¹⁰ The study by Davidson and colleagues¹¹ in this issue of THE JOURNAL is the third multicenter randomized placebo-controlled trial of orlistat to be published.

The 2-year study by Davidson et al began with 1187 obese American adults who had no other significant chronic health conditions. After a 4-week placebo run-in period with a moderately hypocaloric diet, 892 participants (74%) who had maintained pill compliance of at least 75% were randomized either to placebo or to orlistat for 52 weeks while continuing on the diet. At the end of 52 weeks, 576 participants (65%) remained in the study. For the subsequent 52 weeks participants were placed on a weight maintenance diet, and those originally randomized to orlistat were rerandomized either to the initial dose (3 times per day) of 120 mg, to a 60-mg dose, or to placebo. At the end of the study 403 participants remained, representing 45% of those randomized and 34% of those who originally began the run-in period.

The placebo-adjusted effects of orlistat were as follows. During the first year orlistat increased weight loss about 3.15 kg. During the second year's weight maintenance phase orlistat (120 mg) decreased weight regain by about 2.25 kg. By the end of the study orlistat (120 mg) increased the proportion of patients losing more than 10% of initial weight by approximately 17 percentage points (ie, compared with a group of 100 patients treated with diet alone, a group of 100 patients treated with orlistat will have 17 more patients lose more than 10% of their initial body weight).

Orlistat treatment also demonstrated modest improvements in physiologic risk factors. Orlistat resulted in a net decrease in 1-year blood pressures, but this effect was largely driven by an increase in blood pressures in the placebo group. In both groups blood glucose levels increased over 2 years, but the increase was lower in the orlistat group. Blood insulin levels remained nearly constant in the placebo group, but decreased in the orlistat group. Orlistat also resulted in modest reductions in total and low-density lipoprotein cholesterol levels that were independent of its weight loss effect. Orlistat also increased the rate of required vitamin supplementation by about 2-fold (14% of the orlistat group vs 6% of the placebo group).

During the first year, the dropout rate because of adverse events was significantly higher for orlistat than for placebo (9.1% vs 4.0%, P = .01). These adverse events were largely related to steatorrhea. Although the study's overall dropout rates were high, they were similar for subjects receiving orlistat (54%) and for those receiving placebo (57%). This finding suggests that the addition of orlistat to a lifestyle intervention is unlikely to improve patient retention.

Although the primary outcomes of the study were the 1-year and 2-year changes from initial body weight, the investigators did not attempt to obtain the 1-year or 2-year body weights from the large proportion of participants who dropped out. Instead the data were analyzed using the last observation carried forward method in which the last body weight recorded in the study is used as a proxy for the true final body weight. This method of analysis has been criticized because it makes the often faulty assumption that the outcome remains constant at the last observed value.¹² This assumption is probably invalid for obese persons who have dropped out of a weight loss study. The best alternative would have been to measure the final 1-year and 2-year body weights of all dropouts. If that were too costly then the weights of a smaller random sample of dropouts could have been measured and the final 1-year and 2-year weights for the remaining group of dropouts could be imputed.¹²

Ignoring the putative weaknesses of the study and assuming the results are unbiased, it appears that orlistat causes modest improvements in body weight and other important risk factors. However, 4 key questions remain unanswered.

First, will weight loss drugs such as orlistat result in clinically meaningful changes in physiologic risk factors in the absence of a lifestyle intervention? Being on a diet and participating in a physical activity program may enhance the effectiveness of drugs. To date, no studies have been carried out that estimate the effect of drug in the absence of a weight loss diet. It is likely that in the real world clinical setting some patients simply may be prescribed a weight loss medication without being placed on a concomitant lifestyle program.

Second, will the modest changes in physiologic risk factors translate into meaningful improvements in health events that are important to patients? Physiologic risk factors are important to clinicians who monitor their patients. But patients list outcomes such as having a heart attack, stroke, sudden death, or simply improving their quality of life as important health events. For instance, experimental studies demonstrate that in addition to lowering blood pressure, pharmacotherapy for hypertension also decreases the risks of important health events.¹³ However, no published experimental study has estimated the impact of weight loss on important health events. The National Institute of Diabetes and Digestive and Kidney Diseases recently announced its intention to carry out such a trial.¹⁴ Perhaps just as important, longer-term studies with more prolonged exposure to the drug and with larger numbers of subjects may be necessary to more fully define serious adverse effects that are not apparent in shorter-term studies.

Third, given the selection criteria and the high dropout rates that characterize the studies of weight loss drugs, will the yet-to-be-proven health benefits of these drugs be generalizable to any but a small minority of obese patients? Even if appropriate studies demonstrate that pharmacotherapy for obesity has efficacy in reducing morbidity and mortality, the public health effectiveness of pharmacotherapy for obesity in reducing morbidity and mortality in the general population of obese persons may be quite modest.

Fourth, is an inordinate emphasis being placed on treatment of obesity rather than on its primary prevention? Despite some limited research or prevention,^{15 - 16} however, neither clinicians nor public health workers have any tools with proven effectiveness in preventing obesity. It is hard to believe that any approach to primary prevention or to treatment of obesity will be successful at the population level without fundamental changes in cultural perceptions and expectations regarding physical activity and dietary intake. Certainly the public health community has a vital role in helping to bring about and sustain these changes. Without the active help of the clinical community, however, public health efforts will likely fail.

It is important to acknowledge that on an individual basis the clinician's decision to treat an obese patient with weight loss medication may be a reasonable one, despite the uncertainties about the long-term benefits of pharmacotherapy in the population. For some obese patients who respond well to weight loss drugs and can tolerate the adverse effects, pharmacotherapy undoubtedly will be beneficial. Although I believe that my own pediatrician was acting out of sincere concern for my health when he prescribed those diet pills, for me pharmacotherapy was not useful and, in retrospect, probably not appropriate. Those preoccupied with finding solid evidence for meaningful improvements in health at the population level should continue to search for broadly applicable and effective approaches for treatment and prevention of obesity.