The authors disclose the following financial relationships within the past 3 years: This study was designed and funded by Bristol-Myers Squibb Company.
Multicenter, randomized, double-blind, active comparator and placebo-controlled trial of a corticotropin-releasing factor receptor-1 antagonist in generalized anxiety disorder†
Article first published online: 8 APR 2010
© 2010 Wiley-Liss, Inc.
Depression and Anxiety
Volume 27, Issue 5, pages 417–425, May 2010
How to Cite
Coric, V., Feldman, H. H., Oren, D. A., Shekhar, A., Pultz, J., Dockens, R. C., Wu, X., Gentile, K. A., Huang, S.-P., Emison, E., Delmonte, T., D'Souza, B. B., Zimbroff, D. L., Grebb, J. A., Goddard, A. W. and Stock, E. G. (2010), Multicenter, randomized, double-blind, active comparator and placebo-controlled trial of a corticotropin-releasing factor receptor-1 antagonist in generalized anxiety disorder. Depress. Anxiety, 27: 417–425. doi: 10.1002/da.20695
- Issue published online: 28 APR 2010
- Article first published online: 8 APR 2010
- Manuscript Revised: 25 FEB 2010
- Manuscript Accepted: 25 FEB 2010
- Manuscript Received: 27 JAN 2010
- orticotropin-releasing factor;
- generalized anxiety disorder;
- HPA axis
Background: Antagonism of corticotropin-releasing factor (CRF) receptors has been hypothesized as a potential target for the development of novel anxiolytics. This study was designed to determine the safety and efficacy of pexacerfont, a selective CRF-1 receptor antagonist, in the treatment of generalized anxiety disorder (GAD). Method: This was a multicenter, randomized, double-blind, placebo-controlled and active comparator trial. Two hundred and sixty patients were randomly assigned to pexacerfont 100 mg/day (after a 1 week loading dose of 300 mg/day), placebo or escitalopram 20 mg/day in a 2:2:1 ratio. The primary outcome was the mean change from baseline to end point (week 8) in the Hamilton Anxiety Scale total score. Results: Pexacerfont 100 mg/day did not separate from placebo on the primary outcome measure. The half-powered active comparator arm, escitalopram 20 mg/day, demonstrated efficacy with significant separation from placebo at weeks 1, 2, 3, 6, and 8 (P<.02). Response rates for pexacerfont, placebo, and escitalopram were 42, 42, and 53%, respectively. Genetic and psychometric rating scale data was obtained in 175 randomized subjects. There was a significant association between a single nucleotide polymorphism (SNP) of the gene encoding plexin A2 (PLXNA2-2016) with the HAM-A psychic subscale score for the entire cohort at baseline (FDR-adjusted P=.015). Conclusions: Pexacerfont did not demonstrate efficacy compared to placebo for the treatment of GAD. Whether these findings are generalizable to this class of agents remains to be determined. Our preliminary genetic finding of an association between a SNP for the gene encoding plexin A2 and an anxiety phenotype in this study merits further exploration. The trial was registered at clinicaltrials.gov (NCT00481325) before enrollment. Depression and Anxiety, 2010. © 2010 Wiley-Liss, Inc.
Generalized anxiety disorder (GAD) is a disabling and often chronic psychiatric condition with a lifetime prevalence of approximately 5%.1, 2 The disorder affects women twice as commonly as men, and patients experience excessive anxiety, irritability, and a variety of associated physical symptoms. The disorder is typically complicated by the presence of comorbid psychiatric conditions, such as depression, and is associated with significantly increased medical utilization and health care costs.3 Standard treatment for GAD includes psychotherapy and pharmacological management with tricyclic antidepressants, selective serotonin-reuptake inhibitors, serotonin–norepinephrine-reuptake inhibitors, or benzodiazepines. Despite these current treatment options, most patients do not experience long-term remission. Novel treatment interventions targeting neurotransmitters beyond the monoaminergic system represent a potentially important research direction.
Although the etiology of GAD has not yet been clearly elucidated, several lines of evidence suggest that dysfunction in the body's stress response system may contribute to excessive anxiety.4–6 Corticotropin-releasing factor (CRF) is a 41 amino acid polypeptide that coordinates the body's behavioral, endocrine, immune, and autonomic response to stress.7 In addition to its well characterized action on the hypothalamic–pituitary–adrenal axis, CRF also has important extra-hypothalamic effects mediated through two receptor subtypes, CRF receptor-1 (CRF1) and receptor-2 (CRF2). CRF1 has been suggested as a therapeutic target for anxiety and mood disorders given its extensive distribution in brain regions, such as the amygdala, locus coeruleus, hippocampus, and cortex.8–11 Preclinical and open-label clinical studies suggest that antagonism of CRF1 ameliorates anxiety.12–15 Preliminary genetic findings also support a correlation between the CRF system and anxiety.16, 17 Although current animal models of anxiety have not been shown to be specific for any particular anxiety disorder, potential anxiolytic properties of pexacerfont were demonstrated in both the defensive withdrawal and elevated plus maze models of anxiety in rats during preclinical testing (S. Lelas, unpublished data). Antagonism of CRF1 could, therefore, provide a novel therapeutic target for the treatment of anxiety disorders.
Pexacerfont is a highly potent and selective CRF1 receptor antagonist that displays no agonist properties. It is specific for CRF1 receptors and has more than 1,000-fold less affinity for CRF2 receptors, and more than 100-fold less affinity for the CRF-binding protein. In extensive preclinical studies, pexacerfont has been shown to inhibit specific binding of CRF to rat, dog, monkey, and human CRF1 receptors. The functional anxiolytic effects of CRF1 receptor occupancy were demonstrated in two rodent models of anxiety, situational anxiety and elevated plus maze paradigms (S. Lelas, unpublished data). To our knowledge, this is the first published study describing the use of a selective CRF1 antagonist in a large, randomized, and placebo-controlled clinical trial for the treatment of generalized anxiety.
STUDY DESIGN AND POPULATION
This multicenter, randomized, double-blind, active comparator (escitalopram), and placebo-controlled, proof-of-concept study of pexacerfont in GAD was conducted at 50 centers in the United States between July 2007 and January 2008. Patients were treated with study medication for 8 weeks followed by a 1 week medication taper and a 7 week safety follow-up phase. The study was approved by the institutional review board at each study site or a central review board (Western Institutional Review Board). All study subjects provided written informed consent. The trial was registered at clinicaltrials.gov (NCT00481325) before enrollment.
Inclusion criteria for the study were female outpatients aged 18–65, who met Diagnostic and Statistical Manual of Mental Disorders, fourth edition-text revision, (DSM-IV-TR)18 criteria for the diagnosis of GAD, were eligible for study participation. Males were excluded from participation, because results from an earlier initiated study examining the effects of pexacerfont on male spermatogenesis was not yet completed. Diagnosis regarding Axis I disorders was confirmed using the Mini International Neuropsychiatric Interview,19 in addition to a psychiatric evaluation performed by a board-eligible or certified psychiatrist. The presence or absence of any comorbid Axis II diagnoses was determined by the clinical judgment of the evaluating psychiatrist. Additionally, subjects were to meet the following criteria on site-administered rating scales: (1) Hamilton Anxiety scale (HAM-A)20 total score of 18 or greater at screening and baseline; (2) baseline HAM-A total score not more than 30% below the score at screening; (3) HAM-A anxiety and tension item scores of two or greater at both screening and baseline; and (4) Clinical Global Impression of Severity score of four or greater (moderately or more severely ill) at both screening and baseline. Women of child-bearing potential were to have a negative serum or urine pregnancy test within 72 hr before randomization, and agreed to use appropriate birth control throughout the course of the study. Patients with any of the following were excluded from the study: other current Axis I or Axis II psychiatric diagnoses within the past 6 months; serious medical problems that would interfere with safety or efficacy assessments; a 17-item Hamilton Depression (HAMD-17)21 scale score greater than 19 or a HAMD-17 depression item one score greater than 1; any significant alcohol or illicit drug abuse or dependence within the past 12 months; significant suicide risk; and nonresponse to escitalopram or nonresponse to three or more adequate trials of any selective serotonin reuptake inhibitors. Patients could not receive psychotherapy (e.g., individual, group, marriage, or family therapy) during the trial unless participation had been regular (e.g., weekly) for at least 3 months before screening. Prohibited concomitant therapy after randomization included antidepressants, benzodiazepines, antiepileptics, antipsychotics, stimulants, and herbal over-the-counter preparations. Sleeping agents, such as zolpidem, zaleplon, or eszopiclone, were permitted only during the first 4 weeks of treatment with the study medication.
TREATMENT INTERVENTIONS AND RANDOMIZATION
More than 475 human subjects received pexacerfont treatment in Phase 1 and 2 trials before the start of this study. Pexacerfont was well tolerated compared to placebo in these studies. In humans, the half-life of pexacerfont is approximately 4 weeks. Pharmacokinetic modeling of collected data indicated that a loading dose of pexacerfont 300 mg QD for 1 week followed by a 100 mg daily dose is expected to confer greater than 80% receptor occupancy in a majority of patients at steady state trough. This dosing strategy was predicted to provide sufficient drug exposure to exhibit efficacy. Pharmacokinetic samples were collected from patients at weeks 1, 4, 8, 10, and 16, to confirm projected efficacious serum levels (>500 nM) attained during the study.
Subjects were randomly assigned to double-blind treatment with either pexacerfont, matching placebo or escitalopram in a 2:2:1 fashion. Randomization was centralized with no stratification. All treatment arms were delivered in a blinded fashion so that study medications were indistinguishable. Escitalopram treatment was selected as the active comparator because it is indicated for the treatment of GAD and representative of the typical treatment approach using selective serotonin reuptake inhibitors. Eligible patients initially received either pexacerfont 300 mg/day and placebo or escitalopram 10 mg/day for 1 week. After 1 week, dosing was adjusted to pexacerfont 100 mg/day, escitalopram or placebo 20 mg/day through to week 8. After completing week 8, patients entered a 1-week tapering phase followed by a 7-week follow-up phase. To diminish potential SRI-withdrawal effects, patients assigned to escitalopram had their dose blindly decreased to 10 mg for week 9. Patients assigned to pexacerfont or placebo received placebo during week 9. Patients did not receive any study medication during weeks 10–16.
The change in HAM-A total score was prespecified as the primary efficacy assessment of the patient's symptomatology and was obtained utilizing the Structured Interview Guide for the HAM-A (SIGH-A; J. B. Williams, 1996, unpublished). The HAM-A consists of 14 items (each rated 0–4), with higher scores indicating more severe symptoms. All site raters were trained on the standardized administration of the SIGH-A and demonstrated inter-rater reliability by passing a rater training evaluation before rating patients in this study. A secondary assessment of the patient's self-report of level of function was obtained with the Sheehan Disability Scale (SDS).22 Subjects were also evaluated on several additional secondary outcome measures, including change in the HAMD-17 total scores, Clinical Global Impression Severity of Illness (CGI-S), and Global Improvement Scales.23 Patients were classified as treatment responders based on a 50% or greater reduction from baseline HAM-A total scores to endpoint while remission was defined as a final HAM-A total score equal to or less than eight at the last two visits. Additionally, correlations between the HAM-A total score and baseline salivary cortisol were analyzed. Saliva cortisol samples were collected within 1 hr of awakening on three consecutive mornings before randomization.
We calculated that 98 evaluable patients per treatment arm for pexacerfont and placebo was required to detect a difference on the HAM-A of 2.5 between the two groups, assuming a standard deviation of 6.2 in each group, a 2-sided t-test error of 0.05 and 80% power. Approximately 49 patients were included in the escitalopram arm to determine assay sensitivity, for approximately 245 evaluable patients for all three treatment arms. The sample size calculations were based upon a review of three GAD studies that demonstrated a mean difference in change from baseline HAM-A total scores between escitalopram and placebo ranging from 1.6 to 3.9, observed standard deviations between 5.9 and 7.4, and an overall effect size of 0.41 for the pooled studies.24
The primary efficacy measure was the mean change from baseline to endpoint (week 8 LOCF) in the site-rated HAM-A total score. Treatment comparisons for the primary analysis were conducted using analysis of covariance (ANCOVA) model, with baseline scores as a covariate and treatment as a main effect. ANCOVA with a mixed-models approach as a secondary examination at change in HAM-A was performed on observed data and supported results of the ANCOVA. Key secondary endpoints were analyzed similar to the primary endpoint. Analyses were performed for each scheduled time point of the assessment as well as for the end of the study. The safety sample included all randomized patients who received more than one dose of study medication. The efficacy sample included all patients in the safety sample who had more than one efficacy evaluation. All p values were two-sided.
Categorical measures, such as the response rate, were analyzed within the framework of the Mantel–Haenszel general association test statistic. Pearson's correlation coefficient were used to estimate the linear relationship between the HAM-A total score and selected laboratory tests.
Due to the relatively large number of centers in the study, randomization was not stratified by center. Any effect due to center was investigated by first pooling centers with small numbers of patients. After pooling, there were 24 sites used as a factor in an analysis of the primary endpoint. This analysis was not statistically significant and consequently there was no “center effect” with regard to the data obtained.
PHARMACOGENETICS SAMPLES AND ANALYSIS
Pharmacogenetic samples were obtained in subjects to assess for the potential relationship between the primary and secondary efficacy measures and genetic polymorphisms in prespecified candidate genes. The purpose of the pharmacogenetic analysis was to assess for potential relationships between the primary and secondary efficacy measures and genetic polymorphisms in prespecified candidate gene regions, including CRF1, glucocorticoid receptor, corticotropin-releasing hormone-binding protein, and other related genes. The genetic marker used in this study was single nucleotide polymorphism (SNP). ANOVA was used to estimate the SNP effect on the baseline psychometric rating scale scores. ANCOVA and logistic regression with baseline psychometric rating scale score as a covariate were used to estimate the SNP effect in change score and response rate of the primary efficacy measure, respectively. The false discovery rate (FDR) was used for multiple testing adjustment24 on the number of SNPs tested in the analysis for each endpoint separately. All subjects that signed informed consent for pharmacogenetics were genotyped (tested) for 150 prespecified SNPs. Statistical analysis to test the SNP genotype effect on the baseline scores were conducted for each of the SNPs on each of the endpoints listed in the protocol as the primary or secondary objectives.
DISPOSITION OF SUBJECTS AND EFFICACY OUTCOMES
Disposition of study subjects is shown in Figure 1. Demographic and clinical characteristics were similar across treatment groups (Table 1), with the exception of a higher percentage of African-American patients randomized in the pexacerfont and escitalopram groups versus placebo.
|Placebo (N=104)||Pexacerfont (N=103)||Escitalopram (N=53)|
|Mean age (year)||39.3||38.5||38.8|
|Duration of illness (year)||2.7||2.3||2.4|
|Race, number (%)|
|Mean HAM-A total score (SD)||24.7||24.0||23.5|
|Mean HAMD-17 total score (SD)||13.3||13.6||13.8|
|Mean SDS total score (SD)||15.1||15.5||17.5|
Primary and secondary efficacy outcomes are summarized in Table 2. Based on the efficacy sample of 247 patients treated up to 8 weeks, pexacerfont 100 mg/d did not statistically separate from placebo (P=.82; 95% confidence interval [CI] for pexacerfont–placebo difference, −2.27–1.80) on the primary endpoint, mean change from baseline to endpoint (week 8 LOCF) on the site-rated HAM-A total score. The half-powered active comparator arm, escitalopram 20 mg/d, showed significant improvement over placebo at weeks 1, 2, 3, 6, and 8 (each week was significant at the P<.02 level; 95% CI for escitalopram–placebo difference, −5.64 to −0.58). Figure 2 shows the weekly HAM-A scores across all treatment groups. HAM-A response rates at endpoint were placebo 42%, pexacerfont 42%, and escitalopram 53%, respectively. Remission rates based upon a final HAM-A total score of 8 or lower at the last two visits were placebo 14%, pexacerfont 12% (P=.70), and escitalopram 34% (P<.01), respectively. Baseline salivary cortisol levels did not correlate with mean baseline HAM-A scores or treatment response.
|Placebo (N=98)||Pexacerfont (N=97)||Escitalopram (N=47)||P value (pexacerfont/pbo)||P value (escitalopram/pbo)|
|HAM-A total score||−9.4||−9.6||−12.5||.82||<.02*|
|HAM-A psychic factor||−4.5||−4.7||−6.1||.69||.02*|
|HAM-A somatic factor||−4.9||−4.9||−6.5||.89||.02*|
|HAMD-17 total score||−3.6||−3.1||−6.3||.61||<.02*|
|SDS total score||−6.2||−4.7||−7.6||.17||.28|
Three hundred and fifty-one pharmacokinetic samples were collected and analyzed from 88 patients in the pexacerfont group. By the end of week 1 loading phase, 93% of these patients were above the projected human efficacious concentration of pexacerfont (500 nM), which was sustained until dosing ceased. Mean serum concentration (± standard error) of pexacerfont at weeks 1, 4, 8, 10, and 16 were 2,536 nM (±212), 1,629 nM (±100), 1,633 nM (±125), 854 nM (±72), and 370 nM (±47).
There were 150 SNPs genotyped from 175 subjects whose blood samples were collected with appropriate quality for genotyping. After the FDR multiple testing adjustment, none of the SNPs were significant at the 5% level for either the change score or the responder rates on the HAM-A.
To examine for a potential relationship between phenotype and genotype, baseline HAM-A, HAM-D, and CGI-S scale scores were tested in the subjects for which both genetic and baseline rating scale data were available (n=175). After FDR multiple testing adjustment for the 150 SNPs genotyped, a SNP from the PLXNA2 gene (rs-752016) located on chromosome 1q32 demonstrated a significant association with the baseline HAM-A psychic subscale score (FDR-adjusted P=.015). Two other PLXNA2 SNPs tested (rs-2782948 and rs-11119014) did not demonstrate any association (FDR-adjusted P=.78 and .96, respectively).
SALIVARY CORTISOL MEASUREMENTS
Mean baseline morning salivary cortisol concentrations µg/dL (±standard deviation) for placebo (N=84), pexacerfont (N=88), and escitalopram (N=42) were 0.31 (±0.23), 0.45 (±1.13), and 0.32 (±0.18), respectively. Mean baseline evening salivary cortisol measurements µg/dL (±standard deviation) for placebo (N=83), pexacerfont (N=86), and escitalopram (N=42) were 0.08 (±0.09), 0.97 (±7.88), and 0.11 (±0.18), respectively.
The mean change from baseline to end-of-study morning salivary cortisol measurements µg/dl (±standard deviation) for placebo (N=61), pexacerfont (N=70), and escitalopram (N=34) were 0.07 (±0.20), −0.15 (±1.28), and 0.04 (±0.21). The mean change from baseline to end of study evening salivary cortisol measurements µg/dL (±standard deviation) for placebo (N=63), pexacerfont (N=67), and escitalopram (N=34) were 0.04 (±0.17), −1.08 (±8.94), and 0.01 (±0.09). There were no statistically significant correlations between salivary cortisol measurements and baseline HAMA scores, change from baseline HAMA or treatment responders.
Adverse events are summarized in Table 3. In general, pexacerfont was generally well tolerated. Pexacerfont treatment was associated with a higher incidence of nausea and vomiting compared to the other treatments. The most common individual adverse events associated with pexacerfont were nausea, headache, infections, and vomiting. Discontinuation rates due to adverse events were: 10% escitalopram, 5% pexacerfont, and 3% placebo. The incidence of suicidal ideation was low. One subject in the placebo group reported the adverse event of suicidal ideation during the study period. No patients were hospitalized for psychiatric reasons. During the 8-week efficacy phase of the trial, three patients experienced serious adverse events. Two patients were in the placebo group (suicidal/homicidal ideation; incorrect dose administered) and one in the pexacerfont group (incorrect dose administered). There were no clinically significant differences in laboratory measurements, vital signs, or ECGs across treatment groups. Potentially clinically significant liver function abnormalities were seen in one subject on placebo and two on pexacerfont. Adjusted mean changes from baseline weight were: .28 kg for placebo, −.39 kg for pexacerfont, and .54 kg for escitalopram.
|Placebo (N=101)||Pexacerfont (N=98)||Escitalopram (N=51)|
|Nausea, no (%)||12(12)||38(38)||15(29)|
|Vomiting, no (%)||7(7)||11(11)||1(2)|
|Diarrhea, no (%)||14(14)||10(10)||8(16)|
|Dry mouth, no (%)||6(6)||8(8)||3(6)|
|Abdominal pain/distension-no (%)||17(17)||17(17)||14(26)|
|Constipation, no (%)||8(8)||5(5)||6(12)|
|Flatulence, no (%)||2(2)||1(1)||3(6)|
|Headache, no (%)||17(17)||22(22)||12(24)|
|Dizziness, no (%)||4(4)||8(8)||4(8)|
|Somnolence, no (%)||3(3)||4(4)||3(6)|
|Paraesthesia, no (%)||0(0)||2(2)||4(8)|
|Infections, no (%)||18(18)||25(25)||9(18)|
|Fatigue, no (%)||5(5)||7(7)||4(8)|
|Hyperhidrosis, no (%)||0(0)||3(3)||3(6)|
|Decreased appetite, no (%)||0(0)||1(1)||4(8)|
|Serious adverse events|
Despite over two decades of preclinical research and recent open-label clinical studies suggesting that CRF1 antagonism may be associated with anxiolytic properties, this study failed to demonstrate significant anxiolytic properties associated with this investigational drug. This large-scale controlled trial determined that pexacerfont 100 mg/d was not superior in efficacy to placebo for the treatment of Generalized Anxiety Disorder. The active comparator arm, escitalopram, showed statistical separation from placebo and demonstrated that robust signal detection was obtained. Whether pexacerfont's lack of efficacy is specific to this particular compound, due to the dose employed, or is indicative of a more general failure of this particular class of drugs in anxiety disorders remains unknown. There have been several other CRF1 antagonists studied by other pharmaceutical companies to date25; however, negative data from only one placebo-controlled study using a CRF1 antagonist in depression have been published.26
The absence of any beneficial effects of pexacerfont over placebo in anxiety symptoms is somewhat surprising, given the central role that the CRF system is hypothesized to play in the neuroendocrine response to stress. The study results provide potentially important clinical information regarding the CRF system and highlights the need for additional research to further understand this complex system. The human stress response may be more dependent on a critical balance between different compensatory neuroendocrine mechanisms than earlier appreciated. Blocking only one system, such as CRF, may not be sufficient to observe a therapeutic response with regard to anxiety. For instance, there is considerable evidence that CRF and vasopressin work in concert to coordinate the neuroendocrine stress response, with both playing important roles in coordinating behavioral emotions of anxiety.27, 28 Indeed, redundancy within such an important homeostatic system would seem likely.
Furthermore, the clinical contribution of CRF2 receptors in modulating anxiety and stress remains unknown and potentially important in considering the lack of apparent efficacy with a CRF1-only treatment intervention. CRF2 receptors are also located in brain regions involved in anxiety and fear, and recent animal models have suggested that CRF2 antagonism attenuates anxiety.29 Understanding the specific roles and interplay between CRF1 and CRF2 receptors in anxiety needs to be investigated further. Additional work understanding the differential effect that a CRF1 antagonist may have within and outside of the hypothalamic–pituitary axis is also needed. The observation that treatment with pexacerfont did not significantly change basal cortisol levels was considered to be a favorable attribute of the drug from a safety perspective; however, the absence of any inhibition of basal cortisol levels might also have reflected insufficient CRF blockade to achieve therapeutic effect in humans or the engagement of compensatory mechanisms that might have blocked a therapeutic effect. Compounds that produce more robust blockade of the CRF system, as reflected by some significant attenuation of basal cortisol levels, may be necessary in order to yield anxiolytic effects in extra-hypothalamic brain regions.
CRF1 may be particularly activated during acute stress and early phases of anxiety disorders, and blocking these receptors may have limited efficacy in chronic states where stable anxiety levels have been established. In a preclinical model of anxiety disorders,30 CRF antagonists blocked stress-mediated amygdala changes in the early stages of plasticity, but were ineffective once full plasticity and a chronic anxiety state was established. In another study,31 CRF1 antagonists had little effect on social interaction in the absence of agonist infusion. This finding suggests that little activation of CRF1 occurs under routine conditions of the social interaction test. However, when the effects of the CRF1 antagonist on stress-induced reductions in social interaction were tested with animals that were subjected to three sessions of a 30 min restraint stress, the CRF1 antagonist dose–dependently reversed the effect of stress on social interaction. Therefore, the CRF1 receptor may play an important role in the early stages of illness, and antagonists may be best utilized during this period or in stress-induced psychiatric disorders. Indeed, the lack of efficacy in this study may relate to the timing of our intervention, which occurred well after the GAD pathology was established and potentially outside the optimal window of therapeutic opportunity.
Available information regarding pexacerfont's pharmacokinetic properties and preclinical studies suggested that appropriate blood levels were obtained in this clinical trial. Limitations of this study include the inability to determine actual receptor occupancy with pexacerfont (given the lack of a currently available radioactive ligand for human CRF receptors) and toxicology constraints with the compound that prevented the exploration of whether higher dosages of pexacerfont would have been effective.
The observation in this study of an association between a SNP of the gene encoding for plexin A2 and the baseline HAM-A psychic subscale score is consistent with findings from a recent genome-wide association study in anxiety.32 Plexin A2 acts with neuropilins as a receptor for class of molecules called semaphorins.32 Semaphorins are thought to play an important role in neuronal development, apoptosis, and pruning of hippocampal axons.34–36 The potential role of plexins in neuropsychiatric disorders has yet to be clearly established. Our preliminary finding of an association between rs-752016 and an anxiety phenotype in this study merits further exploration. The plexin A2 gene may be a candidate gene for certain subtypes of anxiety as well as other psychiatric disorders.36, 37
Unfortunately, the pathophysiology of anxiety disorders is poorly understood and the development of novel treatment interventions will continue to be challenging, given our current limitations in knowledge regarding the mechanistic underpinnings of these illnesses. Whether a CRF1 antagonist will ultimately prove to be a viable therapeutic target for anxiety disorders remains to be established despite an abundance of preclinical data suggesting its efficacy. Continued research to understand this complex neuroendocrine regulatory system is needed. Future studies on CRF1 and CRF2 antagonists are warranted to determine if they have therapeutic viability in anxiety disorders. Preclinical studies of adjunctive approaches using CRF1 and CRF2 with vasopressin-active therapies may help clarify the therapeutic opportunity. Better elaboration of the degree of basal cortisol reduction for therapeutic efficacy is warranted while in vivo imaging with ligands of CRF receptors would greatly facilitate translational studies.
There remains an urgent need to develop more tolerable and novel treatment approaches that speed onset of therapeutic action and increase remission rates for those suffering from anxiety disorders. The timely development of new treatments beyond the traditional monoaminergic approaches will likely be enhanced by sharing findings from such novel treatment studies and advancing our collective knowledge regarding the pathophysiology of these disorders.
Dr. Coric, Dr. Stock, Dr. Feldman, Dr. Pultz, Dr. Wu, Dr. Dockens, Dr. Huang, Dr. Emison, Ms. Terrye, and Ms. Gentile are employees of Bristol-Myers Squibb and own stock in the company. Dr. Grebb and Dr. Oren were formerly employed at Bristol-Myers Squibb during the course of this study. Dr. D'Souza, Dr. Goddard, and Dr. Zimbroff served as clinical study sites during this study. Dr. Zimbroff and Dr. Goddard received consulting and/or lecture fees from Bristol-Myers Squibb. Dr. Goddard reported receiving grants or consulting fees from Astra-Zeneca, Bristol-Myers Squibb, Janssen, and Orexigen. Dr. Shekhar reported receiving research support from Johnson and Johnson and Eli Lilly. No other potential conflict of interest relevant to this article was reported.
Our deepest appreciation to the patients who participated in this clinical trial. We are grateful for the hard work and efforts of the following principal investigators and their clinical staff: Richard Weisler M.D., Ronald Brenner M.D., Valerie K. Arnold M.D., Donald Garcia M.D., Mahmoud S. Okasha M.D., Andrew Goddard M.D., Jason D. Baron M.D., Michael Downing M.D., Tanya Vapnik Ph.D., Nick Vatakis M.D., Louise Thurman M.D., Mary Stedman M.D., John Stoukides M.D., Leslie Taylor M.D., John Stoukides M.D., Elizabeth Reeve M.D., Mildred Farmer M.D., J. Gary Booker M.D., John S. Carman M.D., James Barbee M.D., Norman E. Rosenthal M.D., Olga Brawman-Mintzer M.D., Mark DiBuono M.D, Boadie Dunlop M.D., Kurian Abraham M.D., John E. Barkenbus M.D, Al-Li Wu Arias M.D., Naresh Emmanuel M.D., Nizar El-Khalili M.D., Anne C. Fedyszen M.D, William Fuller M.D., Bradley N. Gaynes M.D., M.P.H., Lawrence D. Ginsberg M.D., John K. Heussy M.D., Alexander E. Horwitz M.D., Alan M. Jonas M.D., George Joseph M.D., Vernon L. Kliewer M.D., Susan G. Kornstein M.D., David Sert Krakow M.D., Michael Lesem M.D., Michael T. Levy M.D., Craig McCarthy M.D., Matthew Menza M.D., Charles H. Merideth M.D., Irina Mezhebovsky M.D., Janice Miller M.D., Paul R. Miller M.D., Leslie Moldauer M.D., Edward R. Norris M.D., Margarita Nunez M.D., Teresa A. Pigott M.D., Alfredo N. Rivera M.D., Jennifer Rosenberg M.D., Jon Chaffee M.D., Daniel Lieberman M.D., Andrew Winokur M.D., Phebe M. Tucker M.D., and Frederick W. Reimherr M.D.
We also thank the Pexacerfont Development Team for their outstanding implementation of this study protocol, including Olive Watson-Coleman R.N., Cynthia McHugh, Sarah Bevivino, Tamara Bratt, Jennifer Mack, Kathleen Monaco, Lorenzo Biscotti, and Caroline Clairmont; and to Anne Paccaly, Yan Feng, and Anisha Mendonza for their pharmacokinetic modeling, simulation, and analyses in supporting our dosing strategy. We thank the biomarkers group for their support including Randy Slemmon, Ashok Dongre, Jiwen Chen, and Mark Curran. The study benefited immensely from the wisdom and energy of Jack A. Grebb M.D. and Daniel L. Zimbroff M.D., who both passed away before this publication.
- 18Diagnostic and Statistical Manual of Mental Disorders, fourth edition-text revision: DSM-IV-TR. American Psychiatric Association, 2003.
- 23DHEW Publication No. (ADM) 76-338. ECDEU Assessment Manual for Psychopharmacology. Rockville, MD. National Institute of Mental Health; 1976..
- 24Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Roy Stat Soc B Met 1995;57:289–300., .