Recombinant Variant of Ciliary Neurotrophic Factor for Weight Loss in Obese Adults: A Randomized, Dose-Ranging Study

Context Obese individuals tend to resist the weight-regulating effects of exogenously administered leptin. A genetically engineered recombinant human variant ciliary neurotrophic factor (rhvCNTF) that signals through leptinlike pathways in the hypothalamus has been shown to bypass leptin resistance in animal models of obesity.

Objective To identify a safe and well-tolerated dose of rhvCNTF that causes weight loss in obese adults.

Design, Setting, and Patients Twelve-week, double-blind, randomized, parallel-group, dose-ranging, multicenter clinical trial conducted at 2 university obesity clinics and at 5 independent clinical research clinics from March 2000 to August 2001, and including 173 nondiabetic obese adults, 82.6% of whom were women, with a mean (SD) body mass index of 41.1 (4.1).

Interventions Patients were randomly assigned to receive daily for 12 weeks subcutaneous injections of placebo (n = 32) or 0.3 µg/kg (n = 32), 1.0 µg/kg (n = 38), or 2.0 µg/kg (n = 33) of rhvCNTF. Another group received 1.0 µg/kg for 8 weeks and placebo for 4 weeks (n = 38), but they were not included in the primary analysis. All participants received instructions for a reduced-calorie diet (World Health Organization formula minus 500 kcal/d).

Main Outcome Measures Change in weight during the 12-week double-blind treatment period and proportion of patients who achieved a weight loss of at least 5%.

Results Of the 173 randomized patients, 123 (71%) completed the double-blind dosing period. Mean (SEM) changes in kilograms from baseline body weights were 0.1 (0.6) for placebo and −1.5 (0.6) for the 0.3, −4.1 (0.6) for the 1.0, and –3.4 (0.7) for the 2.0 µg/kg of rhvCNTF dosage groups (P<.001, test for trend). Two patients (8.7%) in the placebo and 2 (8.3%) in the 0.3-µg/kg, 8 (29.6%) in the 1.0-µg/kg, and 5 (26%) in the 2.0-µg/kg treatment groups achieved a weight loss of at least 5%. Recombinant human variant CNTF was generally well tolerated although adverse events occurred in 75% of patients receiving placebo and 78% to 93% of patients receiving rhvCNTF, in a dose-related fashion, with mild injection site reactions as the most frequently reported adverse event.

Conclusions In this initial, dose-ranging, 12-week study, treatment with rhvCNTF resulted in more weight loss than placebo. These preliminary findings require confirmation in large prospective clinical trials.

Ciliary neurotrophic factor (CNTF), a protein with a molecular weight of 22 kD, is an endogenous neuroprotective factor that is present in Schwann cells and astrocytes but not in the peripheral circulation and is up-regulated during injury to these cells.^1,2 In a study of patients with amyotrophic lateral sclerosis (ALS) evaluating possible neuroprotective properties, CNTF did not alter disease progression. However, CNTF was found to induce marked weight loss in these patients, who in general were not obese at the outset of the trial.³ This result was not fully understood until it was discovered that CNTF and the weight-regulating hormone leptin have a related intracellular signaling mechanism.⁴

Recombinant human variant (rhv) CNTF is a genetically engineered variant of CNTF with increased potency and improved pharmacological properties.⁵ It binds to the CNTF receptor and activates leptinlike intracellular signaling pathways (Janus kinases and signal transducers and activators of transcription 3) in hypothalamic nuclei, which regulate food intake and body weight.^4,6,7

Both leptin and rhvCNTF are capable of causing pronounced weight loss in leptin-deficient obese mice.^7-10 In the diet-induced obese mouse model, leptin levels are already elevated and additional exogenous leptin does not cause weight loss.⁷ Similarly, obese humans have elevated serum leptin levels, and the efficacy of even high doses of exogenous leptin for weight loss has been disappointing.¹¹ This and other published research strongly suggest that obese humans are leptin-resistant.¹² By contrast, rhvCNTF causes weight loss in diet-induced obese mice, suggesting that this agent may bypass the known leptin-resistance present in this animal model.^7,13 This initial dose-ranging study tested the hypothesis that rhvCNTF leads to weight loss in obese humans.

This randomized, placebo-controlled, double-blind, dose-ranging trial of rhvCNTF was conducted at 7 sites in the United States. Each site received institutional review board approval to conduct the study. All patients gave written informed consent before any study procedures were performed. Eligible patients included men and women aged 18 to 70 years with a body mass index (BMI) of 35 to 50, inclusive. (BMI is calculated as weight in kilograms divided by the square of height in meters.) Patients had to have a stable baseline weight range of at least 4 kg and a stable physical activity level in the 12 weeks before study entry. Patients were excluded if they had diabetes mellitus, a history of stroke, myocardial infarction, coronary artery disease, uncontrolled hypertension, or uncontrolled hyperlipidemia. Pregnant or lactating women were also excluded.

Recombinant human variant CNTF (Axokine) and matching placebo were manufactured by Regeneron Pharmaceuticals, Inc, Tarrytown, NY, and provided for this study.

After a 2-week single blind run-in period of daily placebo injections, patients were randomly assigned to receive placebo or 0.3, 1.0, or 2.0 µg/kg per day of rhvCNTF subcutaneously for 12 weeks. An additional fifth group received rhvCNTF at a dose of 1.0 µg/kg per day for 8 weeks before blindly being switched to receive placebo for the remaining 4 weeks. Data from this dosage group were analyzed in a separate comparison with the 1.0-µg/kg-continuous dosage group. The intent of this withdrawal group was to observe the weight of these patients in the last third of the double-blind dosing period in a controlled and blinded way.

All subcutaneous study drug injections were self-administered daily for the entire study period. Instructions for mild to moderate caloric restriction were given to all patients during the 12-week active treatment phase; these provided a 500-kcal/d deficit relative to the daily caloric requirements calculated according to the World Health Organization formula.¹⁴ Nutritional counseling, including 24-hour food recalls, was provided on 4 occasions. No food diaries were collected.

After the 12-week double-blind dosing period, all patients were asked to return for extended follow-up observations every 3 months for 1 year. Patients and investigators continued to be blinded with respect to the study-drug treatment assignment. At each of these visits body weights and adverse events were recorded. At 6 and 12 months after the last study drug injection, patients received dietary counseling.

Routine hematology and chemistry analyses were performed by a central laboratory (Covance, Indianapolis, Ind). Anti-rhvCNTF antibodies were determined by Regeneron Pharmaceuticals using an enzyme-linked immunosorbent assay. Briefly, microtiter plates were coated with rhvCNTF, which bound antibodies that were present in human serum. These rhvCNTF-bound and immobilized antibodies were detected by the binding of a horseradish peroxidase–conjugated protein A and subsequent addition of an horseradish peroxidase substrate. A mouse monoclonal antibody selective for rhvCNTF and purified by protein A–affinity column was used as a calibrator for the assay.

The primary efficacy end points were weight loss from baseline to the end of the double-blind treatment period at 12 weeks, and the proportion of patients losing a prespecified percentage (≥5%) of their body weight. Analyses using the following data sets were carried out: Intent to treat, last observation carried forward, all available data, and completers. The group that received rhvCNTF for 8 weeks before receiving placebo was not included in these analyses. The intent-to-treat data set included all randomized patients regardless of whether they had a postrandomization weight measured including 6 patients who had elevated prolactin levels at randomization and were excluded (Figure 1); patients without postrandomization weight values were assumed to have had no change in weight. Patients in the last-observation-carried-forward data set had to have at least 1 postrandomization weight measurement, which was carried forward.

Statistical differences between means comparing all dose groups including placebo were first analyzed by analysis of variance with a linear trend test and clinical center as covariate. The prospectively defined analysis plan stated that if the trend test reached statistical significance at the P<.05 level, then each dose group was to be compared with placebo using 1 degree of freedom contrasts in the analysis-of-variance model. Weight-loss-responder analyses, patients with ≥5% weight loss from baseline body weight, were performed using the Cochran-Mantel-Haenszel procedure for all dose groups, stratified by study center. Comparisons of individual dose groups against placebo were also done by the Cochran-Mantel-Haenszel procedure. All statistical tests were 2-sided. Safety data were tabulated and evaluated descriptively. Analyses were performed using SAS statistical software version 8.2 (SAS Inc, Cary, NC).

One hundred ninety-six patients entered a 2-week single-blind run-in period and received daily placebo injections. One hundred seventy-nine patients were subsequently randomly assigned to receive 0.3, 1.0, or 2.0 µg/kg of rhvCNTF or placebo for 12 weeks. Six patients were excluded a few days after double-blind treatment was started because of elevated serum prolactin measurements at baseline (a protocol exclusion criterion), which resulted in 173 evaluable patients (Figure 1). The treatment groups were well matched at baseline with respect to weight and race (Table 1). The majority of the participants were women. Reasons for patients' dropping out up until the end of the double-blind dosing period are displayed in Figure 1.

Results from the intent-to-treat analysis, the last-observation-carried-forward analysis, and the completers analysis are shown in Table 2. The results for weight loss compared with placebo are consistent among the different data sets and are statistically significant for those who were in the 1.0- and 2.0-µg/kg dosage groups. Decreases in BMI were statistically significant in all analyses at the 1.0- and 2.0-µg/kg dosages compared with placebo. On average, patients lost weight in all active treatment groups while patients receiving placebo showed nonsignificant weight gain (Figure 2). Mean weight loss in the 1.0-µg/kg-dosage group was continuous over the 12-week dosing period.

Patients who were switched in a blinded fashion from rhvCNTF at a dosage of 1.0-µg/kg on day 56 to placebo appeared to stop losing weight, and these patients on average maintained their weight compared with patients who continued to receive rhvCNTF at 1.0 µg/kg (data not shown). Subgroup analyses for the effects of weight loss stratified by patient age (divided by tertiles), BMI (<42 or = 42), and race (white or not white) were not significant. A positive test for trend was found for all analyses of responders who lost at least 5% (Table 2). Eighty-eight percent of all patients lost weight while treated with rhvCNTF.

Due to attrition, the number of patients in each dosage group decreased over time. One-year follow-up data were available for 42% of the patients who received placebo and 44% of those treated with rhvCNTF. For this reason, the post hoc analysis of weight in the extended posttreatment follow-up was conducted comparing all patients treated with rhvCNTF as a combined group vs placebo (Figure 3). Compared with the weight at the end of double-blind treatment at month 3, patients who received placebo gained an average of 1.6 kg while patients treated with rhvCNTF gained 0.1 kg at 6 months after the cessation of double-blind injections. Compared with baseline, at month 9, patients in the placebo group had gained 1.7 kg while those in the combined rhvCNTF group had lost 2.9 kg. At month 12, after cessation of the study drug, those in the placebo group gained 0.5 kg, whereas those in the treatment group had gained 0.8 kg. The difference in weight change from baseline between placebo and rhvCNTF treated patients was statistically significantly different at month 3 (P<.001), month 6 (P = .006), month 9 (P = .006), and month 12 (P = .04), but there was no difference at month 15 (P>.10; t test).

There was no difference between any of the dosage groups and placebo in triglyceride, cholesterol, or insulin levels. However, decreases in fasting serum insulin (P = .08 vs placebo) and triglyceride levels were noted in patients who lost 5% or more of their body weight.

Patients and investigators were asked about treatment group assignment at 8 and 12 weeks. At week 12, of all the patients receiving the study drug, 34% thought they were taking rhvCNTF, 40% thought they were taking placebo, and 26% did not know. The investigators thought 34% of patients were taking rhvCNTF, 48% thought patients were receiving placebo, and 18% said they did not know.

At least 1 adverse event was reported by 75% of patients in the placebo group and by 78.1% to 93.9% of patients in the rhvCNTF groups, with increasing rates related to higher doses (Table 3). The most frequently reported adverse events were injection site reactions, nausea, infection, cough, and musculoskeletal pain. Of these, injections site reactions, nausea, and cough were considered to be drug-related by the investigators. Injection-site reactions consisted of nonindurated, mild-to-moderate redness of the skin, which rarely were pruritic. Injection site reactions, if they occurred, generally appeared within a day after dosing and resolved within 3 to 5 days.

Nausea was reported more frequently in patients treated with higher doses of rhvCNTF and nausea-related dropouts were more frequent in the 2.0-µg/kg dosage group (Table 3). Mean weight loss in the rhvCNTF dosage groups did not differ between patients who reported nausea vs those who did not. Increased cough was reported by 42.2% of patients in the 2.0-µg/kg dosage group while the rates of cough in the lower-dosage groups did not differ from placebo.

A total of 20 patients withdrew from the trial prematurely due to adverse events (Figure 1): 4 (12.5%) of 32 patients in the placebo group and 16 (11.3%) of 141 patients receiving rhvCNTF. No pattern of adverse events was seen in the placebo or lower dosage groups. However, in the 2.0-µg/kg dosage group, 4 patients dropped out due to nausea or nausea and vomiting vs 1 in the 1.0-µg/kg dosage group. In our study, occurrence of herpes mouth sores was similar in placebo and drug-treated patients. One serious adverse event occurred during the treatment phase (a motor vehicle crash that was not considered drug related). No death was reported in this study.

Overall, there were no changes in vital signs, or routine laboratory parameters in patients in the active-treatment or placebo control groups. In a dose-dependent fashion, values of C-reactive protein were slightly elevated early into the dosing period but returned to baseline by the end of the active dosing period. At the end of the double-blind dosing period, between 45% and 87% of patients in the rhvCNTF dosage groups had developed anti-rhvCNTF antibodies. There was no apparent relationship between the dosage of rhvCNTF and the proportion of patients developing antibodies. A total of 9 patients from among all rhvCNTF treatment groups developed in vitro neutralizing anti-rhvCNTF antibodies. Eight of these patients showed weight loss in the range of 0.6 kg to 5.5 kg while 1 patient gained 1.0 kg during the active dosing period. The pattern of adverse events in these 9 patients did not appear to differ from that in other patients treated with rhvCNTF.

Ciliary neurotrophic factor signals through hypothalamic intracellular pathways that are leptinlike. Both rhvCNTF and CNTF were shown to bypass leptin resistance in animal models of obesity.^4,6,7,13 In this initial double-blind, randomized, placebo-controlled study, 12 weeks of daily subcutaneous injection therapy with rhvCNTF resulted in more weight loss than injection with placebo. On average, patients in the 1.0-µg/kg dosage group experienced continuous weight loss during the double-blind dosing period. It is possible that this weight loss results from rhvCNTF circumventing leptin resistance in obese humans.¹³

Treatment of obesity is limited by rebound weight gain after cessation of weight loss programs.^15-17 Early studies with fenfluramine,¹⁸ as well as recent trials using sibutramine,^19,20 show that weight regain begins almost immediately after drug treatment is stopped. It appears that more rapid weight loss is followed by more rapid weight regain.²¹ Hence, a weight loss regimen that does not result in rapid weight regain would be highly desirable.

Patients treated with rhvCNTF did not appear to experience immediate weight rebound when drug treatment was stopped. Toward the end of the 1-year follow-up period patients treated with placebo started to lose some weight while patients treated with rhvCNTF started to gain some weight, indicating that the 2 groups might converge at some later time. However, these post hoc findings should be considered preliminary in nature because of high attrition (42% in the patients receiving placebo and 44% in patients receiving rhvCNTF) and nonstudy related factors in these freely living individuals that start to play a role. Prospective studies are currently under way to confirm these retrospective findings.

Results from animal studies carried out in diet-induced obese mice⁷ showed that mice treated with rhvCNTF did not exhibit compensatory overeating when drug injections were stopped. These animals did not experience a rapid regain of the body weight lost during rhvCNTF administration. In contrast, placebo-treated pair-fed control animals consumed a large amount of food after removal of food restriction and did regain weight rapidly.⁷ This difference may be explained by the observation that rhvCNTF suppresses hypothalamic hunger signals in rodents, including expression of neuropeptide Y.⁴ Such hunger signals are produced in elevated amounts in rodents that are subjected to restricted food intake. It is possible that suppression of hunger signals may be a mechanism operative in the obese patients treated with rhvCNTF. The absence of increased hypothalamic hunger signals while patients are treated with rhvCNTF might lead to a lack of compensatory overeating after treatment cessation, which results in prolonged maintenance of weight loss.

At the end of active dosing, weight change was not different between the 1.0- and 2.0-µg/kg dosage groups; however, the 2.0-µg/kg dosage was associated with more adverse events. Nausea had been observed in patients with ALS who were treated with CNTF in earlier studies³ and also in volunteers treated with high single-doses of rhvCNTF in a first-in-human study.²² It is likely that rhvCNTF in addition to binding to the hypothalamic CNTF receptors also interacts with binding sites in the area postrema, a neighboring region in the brain responsible for nausea and vomiting. At the doses used in this study, reactivation of herpetiform mouth sores was not different between placebo and drug-treated patients. Earlier studies with CNTF in patients with ALS³ and a phase 1 study with rhvCNTF in obese volunteers²² showed reactivation of herpetiform mouth sores. The doses used in the latter study were considerably higher (up to 16 µg/kg) than in our study.

Despite the high rates of adverse events, it appears that a dosage of 1.0 µg/kg or less of rhvCNTF is generally well tolerated. Weight loss in patients with and without nausea was comparable, indicating that nausea was not the factor effecting weight loss. Patients in the 2.0-µg/kg-dosage group had higher incidence of nausea, vomiting, and cough, but weight loss was not greater than in the 1.0-µg/kg dosage group. Taken together, these observations support investigation of rhvCNTF at a dose of 1.0 µg/kg for future studies. There were no deleterious changes of cardiovascular parameters including blood pressure, heart rate, and heart valve function and anatomy (echocardiographic data not shown).

Along with other therapeutic proteins administered subcutaneously, such as insulin, rhvCNTF shares the potential to generate antibodies in humans. Anti-rhvCNTF antibodies developed in many patients treated with rhvCNTF, but the presence of binding antibodies, neutralizing anti-rhvCNTF antibodies, or both did not appear to affect total weight loss or adverse event profile.

In summary, this short-term, dose-ranging study resulted in more weight loss in those taking rhvCNTF than in those taking placebo. These preliminary findings need to be confirmed in large-scale prospective studies.