Regulatory challenges for new drugs to treat obesity and comorbid metabolic disorders

Efficacy

Revised guidance notes for the development of antiobesity drugs have been issued in draft form by the FDA [33] and in final form by EMEA [34]. In both territories, the primary clinical end-point for the approval of a new antiobesity drug is weight reduction and, just as importantly, the maintenance of that weight loss after 1 year's treatment. Clinically meaningful weight reduction after 1 year of treatment is defined as ≥5% (placebo-subtracted and statistically significant) in the USA and >10% of baseline weight, which is also ≥5% greater than that observed with placebo, in Europe. Although weight reduction is the primary end-point of efficacy for antiobesity drug candidates, in the absence of commensurate improvements in a wide range cardiovascular risk factors, weight reduction would be considered by either FDA or EMEA as being a cosmetic outcome with no medical benefit. Consequently, any drug candidate exhibiting this clinical profile would be non-approvable in the USA or Europe. The FDA mandates that an effective weight-management product should provide improvements in blood pressure, lipids, glycaemia, or other beneficial outcomes that are commensurate with the degree of weight loss. To this end, common weight-related comorbidities need to be factored into clinical trials to assess the efficacy of such products. In Europe, EMEA lists ‘secondary end-points’ of efficacy for antiobesity drug candidates as improvements in lipid and glucose metabolism, blood pressure, cardiac function, waist–hip ratio, waist circumference, ultrasensitive C-reactive protein, sleep apnoea episodes and quality of life parameters. EMEA also cites maintenance of weight loss or the prevention of weight regain as an important ‘secondary end-point’ for antiobesity drug candidates.

Safety

The safety criteria for antiobesity drug candidates are particularly onerous because of the poor track record of approved drugs in this therapeutic indication. Furthermore, in the eyes of many healthcare professionals obesity is a risk factor for serious cardiovascular and metabolic diseases including hypertension, Type 2 diabetes and dyslipidaemia, rather than a disease in its own right.

Both regulatory agencies state that antiobesity drugs should not adversely affect cardiovascular function. A subtle difference between the USA and Europe is the focus of these concerns. In the USA, cardiac valvulopathy is highlighted because of their experience with dexfenfluramine and the fenfluramine/phentermine (fen/phen) combination. In Europe, dexfenfluramine and fenfluramine were only approved for use for ≤12 weeks, and as a consequence cardiac valvulopathy was not a safety issue. In addition, the use of dexfenfluramine or fenfluramine together with another centrally-acting anorectic agent was ‘contraindicated’ in the European Summary of Product Characteristics (SPC), thereby circumventing the fen/phen issue also. On the other hand, the fenfluramines have been linked to the rare, but potentially fatal, adverse event of PPH in Europe [25, 26] and it is therefore a focus of safety monitoring and evaluation for current and future antiobesity drugs. For centrally-acting drug candidates, psychiatric adverse events, abuse, dependence and withdrawal side-effects are explicitly noted by FDA and EMEA as potential safety issues to be investigated. In its draft guidance, the FDA also states that it wishes to see drug-mediated weight reduction demonstrated to result from a loss of body fat, rather than lean body mass, using dual energy x-ray absorptiometry or other scanning techniques.

Benefit–risk evaluation

The criteria listed above are the factors on which the benefit–risk evaluation for drug approval is performed, but there is no indication in either guidance document of the relative weight ascribed to each of them. In Europe, ‘hard’ cardiovascular end-points, i.e. myocardial infarction, stroke, morbidity and mortality, are considered to be of far greater relevance than reductions in risk factors like blood pressure, LDL-cholesterol or HbA_1c. Although clinically meaningful improvements in these end-points are sufficient to gain European approval for an antiobesity drug candidate, a commitment to a Phase IV outcome trial is a prerequisite for market authorization, cf the Sibutramine Cardiovascular OUTcome (SCOUT) [35] and Comprehensive Rimonabant Evaluation Study of Cardiovascular ENDpoints and Outcomes (CRESCENDO) [36] studies. In the USA, on the other hand, a benign risk profile is the predominant factor in the FDA evaluation of benefit–risk for an antiobesity drug candidate. This emphasis on safety is based not only on the US experience with dexfenfluramine and fen/phen, but also on the recent high-profile safety scares with other drugs, e.g. rofecoxib (Vioxx®) and rosiglitazone (Avandia®).

The Metabolic Syndrome

One of the most contentious areas in terms of drug development has been the status of the Metabolic Syndrome as a therapeutic indication suitable for the market authorization of new drugs. The syndrome was first described by Reaven in his 1988 Banting Medal award lecture [49]. In his groundbreaking lecture, Reaven postulated that insulin resistance and the compensatory hyperinsulinanaemia predisposed patients to hypertension, hyperlipidaemia and diabetes, and thus was the underlying cause of much cardiovascular disease. Although obesity was not included in Reaven's primary list of disorders caused by insulin resistance, he acknowledged that obesity was correlated with insulin resistance and hyperinsulinaemia. Thus, reducing bodyweight and increasing physical activity would be predicted to have a beneficial effect on what he called ‘syndrome X’. An illustration of the key factors of what we now call the Metabolic Syndrome is shown in Figure 2.

It should be noted that as originally envisaged by Reaven, the essential driver of the Metabolic Syndrome was insulin resistance. However, several groups of clinical experts (often with a therapeutic bias, e.g. diabetologists, cardiologists and endocrinologists) have since revisited the definition of the Metabolic Syndrome either to refine or fundamentally alter its diagnostic criteria. Such groups have included the WHO and National Cholesterol Education Program (NCEP) Adult Treatment Panel (ATP) III and the International Diabetes Federation (IDF). Each of these highly respected and authoritative bodies has drawn different conclusions about what cardiovascular risk factors should be included in the definition of the Metabolic Syndrome, whether there is an essential core symptom that is a prerequisite for the diagnosis of the syndrome, and finally, what are the cut-off criteria for reaching a decision on whether or not a patient can be diagnosed as having the Metabolic Syndrome. The WHO produced the first revised definition of the Metabolic Syndrome in 1999 [50] and this definition most closely adheres to Reaven's original hypothesis [51]. This was followed by the NCEP ATP III guidelines published in 2001 [52]. Recently, the IDF published its ‘consensus worldwide definition of the Metabolic Syndrome’ in 2005 [9] and this definition of the syndrome was a major deviation from Reaven's original hypothesis because central obesity was not only included as a risk factor for diagnosis, but also classified as being a prerequisite. The individual definitions of the Metabolic Syndrome according to the WHO, NCEP ATP III and IDF guidelines are reported in Table 4.

A comparison of these very different definitions of the Metabolic Syndrome reveals that there are patients who will meet the criteria for its diagnosis according to one definition, but will fail on both of the others. For example, a man with waist circumference of 90 cm, triglycerides 155 mg dl⁻¹ (1.76 mmol l⁻¹), HDL-cholesterol 35 mg dl⁻¹ (0.9 mmol l⁻¹), blood pressure 135/90 mmHg with a fasting plasma glucose of 105 mg dl⁻¹ (5.82 mmol l⁻¹) would be diagnosed as having the Metabolic Syndrome according to NCEP ATP III, but not WHO or IDF criteria. It might be thought that these differences in classification reflect the perspectives of individual specializations in the metabolic field, but this is also not the case. For example, within a year of the IDF publishing its ‘consensus guidelines on the metabolic syndrome’, two other major diabetes groups, the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD), appeared in print with a damning criticism of the IDF definition [53]. The ADA/EASD critique of the Metabolic Syndrome was much more fundamental than merely questioning the criteria employed for its diagnosis, because it questioned the arbitrary cut-off points applied in all definitions of the syndrome and the lack of a rationale for the exclusion of other significant cardiovascular risk factors, e.g. smoking and a family history of cardiovascular disease. The overall conclusion reached by Kahn and colleagues was that with no irrefutable evidence of any cooperative or synergistic interaction between the risk factors included in the definition of the Metabolic Syndrome, none of the definitions satisfied the requirement for designation as a ‘syndrome’. Moreover, the ADA/EASD recommended that the individual risk factors should be treated in clinical practice, not the Metabolic Syndrome as an entity. Even the IDF did not advocate the use of pharmacological interventions to reduce central obesity, merely citing diet, exercise and a change to a healthy diet to reduce weight by 5–10%. Worse was to come, when Reaven himself stepped into the argument and published a similarly critical review of the IDF consensus guidelines on Metabolic Syndrome [51]. Thus, with competing and divergent definitions of what constituted the Metabolic Syndrome, and internecine warfare breaking out between the respective groups of clinical opinion leaders, the chances of any regulatory agency favouring one definition over all others and using it to register drugs for clinical use in the indication was remote. However, even if a single, agreed definition of the Metabolic Syndrome existed (including one where central obesity was a prerequisite), the fact that a diagnosis of the syndrome is made on the basis of any two (or three, or four) of four (or five) risk factors spanning obesity, Type 2 diabetes, dyslipidaemia and hypertension, means that a diagnosis of Metabolic Syndrome would be clinically meaningless because of the disparate cluster of patients that would be captured under this umbrella definition. Furthermore, in this heterogeneous patient population no single drug would be suitable for treating every patient and the risk of inappropriate prescribing would be significant. When this factor is added into the equation, it is predictable that the regulatory agencies are not prepared currently to consider approving an antiobesity drug as a treatment for the Metabolic Syndrome.

As discussed earlier, long-term clinical outcome trials are an important factor that could alter the status of the Metabolic Syndrome as a therapeutic indication for the registration of new antiobesity drugs. EMEA is very focused on ‘hard’ end-points in outcome trials, and both Abbott and Sanofi-Aventis agreed to perform Phase IV trials, i.e. SCOUT and CRESCENDO, to determine whether antiobesity drug therapy using sibutramine and rimonabant, respectively, together with lifestyle intervention, would significantly reduce cardiovascular morbidity and mortality in overweight/obese subjects with a high degree of high cardiovascular risk, e.g. in overweight or obese patients, plus the presence of coronary, cerebral and/or peripheral arteriopathy, impaired glucose tolerance or Type 2 diabetes, and at least one other established vascular risk factor. Comparing the patient inclusion criteria for these two outcome trials with the various definitions of the Metabolic Syndrome reveals that many of these subjects would be classified as having the Metabolic Syndrome according to the criteria devised by the WHO, NCEP ATP III or IDF. If, in this heterogeneous population, sibutramine is ultimately shown significantly to reduce cardiovascular morbidity and mortality, it is possible that the regulatory agencies will need to reconsider their position on the Metabolic Syndrome as a therapeutic indication, and on the approval of antiobesity drugs for its management.

Type 2 diabetes

The spreading epidemic of obesity is a major driver for the growing number of diabetes sufferers [3, 5, 54–57]. More significantly, there is an exponential relationship between increasing BMI and the relative risk of developing diabetes, e.g. women with a BMI > 35 have a ∼60-fold greater probability of being diagnosed with Type 2 diabetes than women with a BMI < 22 [58].

Improvements in glycaemic control, e.g. plasma glucose and insulin concentrations, insulin resistance and HbA_1c, are the appropriate secondary end-points for evaluating the efficacy of antiobesity drug therapy in subjects with Type 2 diabetes. Although improvement in these secondary end-points with accompanying reduction in bodyweight and visceral adiposity is highly beneficial in the management of Type 2 diabetes in overweight or obese patients, a decision on whether to approve any antiobesity drug specifically for the treatment of Type 2 diabetes would require its benefits to be compared with those delivered by current antidiabetic drugs. A reduction in plasma HbA_1c is generally accepted as the most appropriate primary end-point for the evaluation of antidiabetic drug candidates, with a reduction ≥0.5% considered to be clinically meaningful. The first generation of antidiabetic drugs, including the sulphonylureas, thiazolidinediones and metformin, generally produce reductions in HbA_1c of the order of 1.0–1.5% when administered as monotherapy [59–64]. More recent introductions, including the postprandial glucose regulators, e.g. nateglinide, GLP-1 agonists, e.g. exenatide and liraglutide, and the DPP-IV inhibitors, e.g. sitagliptin, evoke smaller decreases in HbA_1c in the region of 0.7–0.8% [65–69].

By comparison, antiobesity drugs evoke clinically meaningful reductions in HbA_1c in obese diabetic subjects. Orlistat shows the most consistent effects with additional HbA_1c reductions of 0.3–0.5% at 1 year when given in combination with oral antidiabetic agents [70–72]. The effects of sibutramine on plasma HbA_1c in obese diabetics are rather more variable, with reported reductions ranging from 0.1% at 26 weeks [73] to 2.2% at 52 weeks [74]. When the data are viewed overall, the HbA_1c-reducing action of sibutramine is likely to be of the order of 0.5–0.7% when this drug is given in combination with oral antidiabetic agents [73–75]. Rimonabant has been evaluated both as an adjunct to oral antidiabetic therapy in the 1-year RIO Diabetes trial [76] and as monotherapy in the Study Evaluating Rimonabant Efficacy in drug NAive DiabEtic patients (SERENADE) trial [77]. In the former, rimonabant evoked a highly respectable 0.7% fall in plasma HbA_1c when combined with either metformin or a sulphonylurea [76]. When used as monotherapy in newly diagnosed, treatment-naive, obese (average BMI ∼34 kg m⁻²) Type 2 diabetics, rimonabant produced a mean decrease in plasma HbA_1c of 0.5% at week 26 [77]. In a population that is predominantly obese, weight gain is an unwelcome side-effect for a significant proportion of antidiabetic therapies, particularly the sulphonylureas, thiazolidinediones and insulin [78–81]. On the other hand, metformin and some of the newer agents, including pramlintide, exenatide and sitagliptin, are either weight-neutral or induce a small degree of weight loss [67, 68, 80, 82]. As expected, when antiobesity drugs are used to treat Type 2 diabetes in overweight or obese subjects, these drugs produce weight loss [70–77, 83], which, depending on the agent, accounts for all or part of their benefit on glycaemic control. However, subjects with Type 2 diabetes are fairly resistant to the actions of centrally- or peripherally-acting antiobesity drugs. Consequently, the degree of mean weight loss in obese Type 2 diabetics is smaller than that observed in other obese patient populations, e.g. the ‘healthy’ obese (obesity without significant comorbid metabolic disease) or obesity with dyslipidaemia. For example, rimonabant decreased bodyweight by 3.9 kg when used as an adjunct to oral diabetic drugs [76] and 3.8 kg when used as monotherapy [77]. For comparison, in the healthy obese population, the placebo-subtracted reductions in bodyweight were 4.7 kg [84, 85], while in obese subjects with dyslipidaemia it was 5.4 kg [86].

Viewed overall, the antiobesity drugs deliver metabolic benefits in Type 2 diabetes that compare favourably with those of the newer generation of antidiabetic agents. The reason why antiobesity drugs have not until now received serious consideration for treatment of Type 2 diabetes is driven by four factors.

1. The lack of pivotal trial data where HbA_1c, not weight loss, has been used as the primary end-point of efficacy.

2. Until the recent SERENADE trial, no evaluation had been made of an antiobesity drug given as monotherapy in Type 2 diabetes.

3. Type 2 diabetes does not occur exclusively in overweight or obese subjects and approval in this therapeutic indication would imply that a drug would be suitable for all categories of patient, irrespective of BMI. The utility of antiobesity drugs as adjunctive treatments or monotherapy in normal-weight subjects with Type 2 diabetes has not been demonstrated.

4. Until recently, EMEA and FDA had insisted that any beneficial effect of an antiobesity drug on glycaemic control due to weight loss should be discounted. Only improvements beyond those resulting from reduced bodyweight would be taken into consideration for approval of an antiobesity drug in Type 2 diabetes.

FDA has very recently changed its stance on this issue. In its ‘Draft Guidance for Industry. Diabetes Mellitus: developing drugs and therapeutic biologics for treatment and prevention’[87] it states that ‘The FDA's current thinking is that a sponsor can gain approval for the treatment of Type 2 diabetes for a drug or biologic whose principal mechanism of action appears to be weight loss by showing a clinically meaningful and statistically significant improvement in glycaemia’. To date, there has been no indication of what such a clinical development programme would look like. In Europe, there has not been any communication from EMEA to suggest that it is also considering a similar strategy for the development of antiobesity drugs for the indication of Type 2 diabetes.

Dyslipidaemia

Obesity, a Westernized diet that is high in fat and cholesterol, together with other cardiometabolic risk factors, e.g. smoking and lack of physical exercise, exert a highly deleterious effect on plasma lipid profiles. Hyperlipidaemia is an important modifiable risk factor for the development and progression of cardiovascular disease. The causal link between LDL-cholesterol and coronary heart disease is now firmly established [88]. Compelling evidence also supports an independent link between atherosclerosis and low serum concentrations of HDL-cholesterol [89] with high serum levels of triglycerides [90, 91].

Many people with atherogenic dyslipidaemia would also be diagnosed as having the Metabolic Syndrome according to the WHO, NCEP ATP III or IDF classifications (see Table 4). Treatment aims are to obtain plasma concentrations of LDL-cholesterol <2.58 mmol l⁻¹ (100 mg dl⁻¹), total cholesterol <5.15 mmol l⁻¹ (200 mg dl⁻¹), HDL-cholesterol >1.55 mmol l⁻¹ (60 mg dl⁻¹) and triglycerides <1.7 mmol l⁻¹ (150 mg dl⁻¹) [4]. In common with other related metabolic disorders, e.g. obesity, insulin resistance, Type 2 diabetes and Metabolic Syndrome, it has been shown that dieting, increased physical activity and a switch to a low fat/low cholesterol diet has a major positive impact on all atherogenic, dyslipidaemic and cardiovascular risk factors. Key drugs to treat elevated apolipoprotein-B and/or atherogenic dyslipidaemia are the statins, cholesterol-absorption blockers, bile-acid sequestrants, nicotinic acid and the peroxisome proliferator-activated receptor α agonists (the fibrates). In addition, there are interesting, novel, lipid-regulator, drug candidates in clinical development, the most exciting being the cholesterylester transfer protein (CETP) inhibitors, e.g. R1658 (Roche/Japan Tobacco) and anacetrapib (Merck). The actions of the current crop of drugs and the CETP inhibitors to improve the major components of dyslipidaemia are reported in Table 5. What is immediately apparent is how surprisingly broad their spectrum of beneficial effects is, e.g. the statins are principally known as LDL-cholesterol/VLDL-cholesterol-lowering drugs, but they also substantially increase HDL-cholesterol and reduce serum triglycerides.

An extensive search of both the EMEA and FDA websites for guidelines on the efficacy/safety criteria necessary for the registration of new dyslipidaemic agents revealed no guidance on the subject. However, it is probably safe to assume that two Phase III pivotal studies of the design employed [92] with both placebo and active comparator arms, patient numbers in excess of 1000 and a 1-year treatment duration would be the minimum required for registration of any new drug in this therapeutic indication.

For comparison, the effects of the antiobesity drugs sibutramine, orlistat and rimonabant are also shown in Table 5. Orlistat has been reported to produce moderate improvements in plasma lipid profiles in obese patients with hypercholesterolaemia [93–95] or Metabolic Syndrome [96]. Orlistat is the only antiobesity drug to have been shown to reduce LDL-cholesterol. In addition to its beneficial effects on plasma lipid profiles, orlistat also decreased bodyweight by 1–6 kg in these subjects. In the only double-blind, randomized, placebo-controlled study that has been performed with sibutramine in obese patients with dyslipidaemia, it decreased plasma triglycerides by 17% and increased HDL-cholesterol by 6% at week 24 [97]. It also decreased bodyweight by 4.3 kg. In the RIO-Lipids trial, rimonabant decreased plasma triglycerides by 12%, increased HDL-cholesterol by 8% and decreased bodyweight by 5.4 kg at 1 year [86]. It is evident that although antiobesity drugs have some beneficial effects on plasma lipid profiles and bodyweight in obese patients with dyslipidaemia, their actions are modest compared with those of existing agents. Moreover, rimonabant and sibutramine do not lower plasma LDL-cholesterol (the primary action of the statins), and no outcome trial data are available to demonstrate a reduction in cardiovascular morbidity and mortality with long-term antiobesity drug treatment. Contrasted against this is a wealth of positive outcome data for the statins, e.g. WOSCOPS [98], AFCAPS/TexCaps [99], the simvastatin 4S study [100, 101], the pravastatin PPP [102] and LIPID [103] trials. On this basis, there is no compelling case to support the approval of antiobesity drugs specifically for the treatment of any type of dyslipidaemia.