Journal Logo


An efficient randomised, placebo-controlled clinical trial with the irreversible fatty acid amide hydrolase-1 inhibitor PF-04457845, which modulates endocannabinoids but fails to induce effective analgesia in patients with pain due to osteoarthritis of the knee

Huggins, John P.*; Smart, Trevor S.; Langman, Stephen; Taylor, Louise; Young, Tim

Author Information
doi: 10.1016/j.pain.2012.04.020
  • Free
  • SDC


1. Introduction

Cannabinoid receptors influence nociceptive pathways [22]. Exogenous cannabinoid ligands may presynaptically inhibit neurotransmitter release via CB1 receptors [23], whilst microglial activation and peripheral inflammation may be down-regulated via CB2-receptor activation [46].

Several randomised, controlled trials in multiple sclerosis and HIV-associated neuropathic pain demonstrated the analgesic effect of cannabinoids in humans [1,18,43–45,51,54,56,57], although other studies in peripheral neuropathic pain failed to demonstrate an improvement compared to placebo [10,26,48]. A large body of evidence in behavioural rodent models suggests activation of these receptors may ameliorate chronic pain [13,41].

Use of cannabinoids is known to be associated with psychoactive effects including alterations in perception (blurred vision, visual hallucinations, tinnitus, disorientation, confusion, dissociation, acute psychosis) [36], and there is concern over long-term treatment of individuals with cannabinoids in relation to subsequent risk of mental illness (psychosis, schizophrenia) [43]. The psychoactive effects of cannabinoids also have the added risk of potentially compromising the blinding of controlled clinical trials, leading to possible bias in pain outcomes. It has been postulated that a way to provide the analgesic effects of cannabinoids without the undesirable side effects may be to increase endogenous cannabinoid concentrations rather than administering exogenous compounds [6].

Inhibition of fatty acid amide hydrolase-1 (FAAH1) may be a promising therapeutic target [6], because this enzyme hydrolyzes the endocannabinoid N-arachidonyl ethanolamine (AEA, anandamide) and related fatty acid amides (FAAs), including N-palmitoylethanolamide (PEA), N-oleoylethanolamide (OEA) and N-linoleoyl ethanolamine (LEA), which have diverse biology and mechanism of action [5,15]. One such mechanism is activation of cannabinoid receptors. This does not cause the overt untoward effects associated with exogenous compounds [3,6,11,16,27,32,33].

PF-04457845 is a potent and selective FAAH1 inhibitor that elicits properties associated with analgesia in rodent behavioural models [6]. It inhibits FAAH1 after oral administration to rodents and human volunteers with a concomitant elevation of FAAs [6,32] whilst having an excellent tolerability profile in humans including an absence of effects on cognitive parameters [32]. PF-04457845 therefore provides an opportunity to test whether FAAH1 inhibition would elevate FAAs, including AEA, and exhibit analgesic properties in patients with chronic pain.

Genetic or pharmacological inactivation of FAAH1 produces analgesic, anti-inflammatory, antidepressant and sleep-enhancing phenotypes in rodents, particularly in models of inflammatory pain [2,4,6,12,16,17,19,21,24,25,27,30,33,37,39,40,47]. Also, using electrophysiological single-unit recordings from guinea pig joint afferents, it has recently been reported that local application of another FAAH inhibitor (URB597) reduced noiciceptive pathway firing in a model of osteoarthritis (OA) [50].

We therefore report here the design and results of a randomised, placebo and active-controlled clinical trial of PF-04457845 in patients with OA of the knee using a crossover design intended to provide an efficient approach to decision making. The primary objectives were to evaluate the safety and efficacy (vs placebo) of PF-04457845 in relieving pain. Secondary objectives included investigation of the pharmacokinetics and pharmacodynamics (changes in FAAH activity and FAA concentrations) of PF-04457845 in OA patients. Naproxen, which is considered the standard of care for the relief of OA symptoms, including pain, tenderness, swelling and stiffness, was included as a positive control to assess the validity of the crossover design.

2. Methods

2.1. Patient selection

Patients were recruited from 4 sites in the United States, Canada and Sweden, and the trial was approved by the appropriate institutional review board for each site. Each patient provided written informed consent. Patients were eligible for the trial if they had a diagnosis of OA of the knee based on American College of Rheumatology criteria [7] with X-ray confirmation within the previous 12months (a Kellgren-Lawrence X-ray grade of at least 2 [28]). Patients also had to be aged 18–75years, not be of childbearing potential, willing and able to discontinue all current analgesic therapy (including over-the-counter pain medications and topical analgesics for OA pain) throughout the trial and have a QTc interval up to 470ms and a PR interval up to 210ms based on an electrocardiogram at screening. Key exclusion criteria included a body mass index of >40kg/m2; evidence or history of clinically significant endocrine, pulmonary, gastrointestinal, cardiovascular, renal, psychiatric or neurological disease that could not be confirmed to have been stable (under control) for at least 4weeks; evidence of existing hepatic disease (or history of hepatic disease within the previous year); active malignancy of any type; or a history of a malignancy within 10years, excluding treated basal cell carcinoma. Further key exclusion criteria included symptomatic OA of the hip ipsilateral to the index knee (defined as the more painful knee at screening) which the patient considered to be more painful than the knee; history of diseases other than OA that may have involved the index knee; symptomatic anserine bursitis or acute joint trauma or arthroscopy on the index knee within last year; other severe pain that may impair assessment of OA pain; previous use of any cannabinoid therapy; positive urine drug screening results; personal or family history of psychosis; and evidence of suicidality (as determined by the Columbia Suicide-Severity Rating Scale, C-SSRS [42]). In addition, patients had to have a Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC [8,9]) pain subscore (0–20) of at least 6 at randomisation. Treatment-naive patients had to have a numeric rating scale (NRS) daily pain score (0–10) of at least 4 at screening and randomisation (based on 3 previous daily diary entries), whilst patients who were required to washout from their medication (those on chronic nonsteroidal anti-inflammatory drugs and COX-2 inhibitors) had to have an NRS scale of at least 4 at screening that increased by at least 1 point before randomisation and within 4days of commencing washout.

2.2. Trial procedures and clinical end points

This trial was a multicentre, randomised, double-blind, double-dummy, placebo- and active-controlled crossover design. The protocol is registered at (NCT00981357). The trial was conducted between November 2009 and June 2010 and in accordance with the principles of the Declaration of Helsinki and good clinical practice guidelines.

The trial consisted of a screening visit up to 21days before baseline (day 1 of period 1), followed by an initial pain assessment period (PAP) for 1week before baseline to ensure eligibility. Patients then received 2weeks of double-blind treatment (period 1), followed by a 2-week washout period. A repeat PAP was conducted for the following 1week, after which patients received a further 2weeks of double-blind treatment (period 2). All patients had a follow-up visit up to 3weeks after the final dose (Fig. 1A). All patients received single-blind placebo during the wash-in PAP weeks and the 2-week washout period. Patients were randomised to receive either PF-04457845 followed by placebo (or vice versa), or naproxen followed by placebo (or vice versa) on day 1 of period 1. A centralized telerandomisation system was used to manage trial treatment allocations, and patients were allocated sequentially to each of the 4 treatment sequences in a 1:1:1:1 ratio. Unblinded data were provided to an internal review committee (composed of sponsor staff who were not directly involved in any PF-04457845 studies) in order to perform the interim analysis.

Fig. 1.:
Design characteristics of this clinical trial. (A) Overview of trial design. V = visit; D = day within study. Visits 4 and 6 were only conducted for subjects before the interim analysis. (B) Graphical representation of the decision rules applied at the end of the study. Illustrated are means (dashed line) and posterior distributions (solid line). Lines (dotted line) at 0 and −1.8 indicate the cutoff points used in the decision criteria. If the mean WOMAC pain score (difference from placebo) was less than −2.3, then there would be greater than 80% probability that the true difference is less than −1.8 (meeting decision criterion 1) (top). If the mean difference was less than −1.3 then there would be greater than 20% probability that the true difference is less than −1.8 (meeting decision criterion 2a) (center). If the mean difference was less than −0.7, then there would be greater than 90% probability that the true difference is <0, that is, less than placebo (meeting decision criterion 2b) (bottom). At the interim analysis, the criteria are the same, but the posterior distributions would be wider.

Efficacy end points included WOMAC pain subscore (0–20), WOMAC stiffness domain score, WOMAC Physical Function domain score, WOMAC Total score, 11-point NRS from daily pain diaries over each week of each treatment period and use of rescue medication. These end points (except for NRS, which was collected daily) were collected at the start of each PAP (1week before each treatment period) and on day 1 (predose) and day 14 of each treatment period. Blood samples for pharmacodynamic end points (FAAH activity in leukocytes and plasma AEA, OEA, PEA and LEA concentrations) were collected at intervals during the trial for patients recruited before the interim analysis.

Safety end points included adverse events, laboratory safety data, electrocardiograms and vital signs (sitting/standing blood pressure, pulse rate). The C-SSRS was assessed during screening to determine eligibility for the trial. Baseline Hospital and Anxiety Depression Scale (HADS [58]) scores were assessed on day 1 of each treatment period.

2.3. Sample size and trial design

The sample size was based on the primary end point: the WOMAC Pain score after 2weeks of treatment. The primary comparison compared PF-04457845 and placebo, and the secondary comparison compared naproxen and placebo. Traditionally OA trials are large as they are parallel group studies and rely on between-subject comparisons. On the basis of an analysis of unpublished data from Pfizer clinical trials, the intrasubject variance was 0.56 of the intersubject variance. In a parallel group trial, 264 subjects were predicted to give equivalent precision as the planned 72 subjects in the PF-04457845 and placebo sequences of this crossover trial. The sample size was estimated such that the trial would have at least 80% power to detect a difference of 0.9 above the estimate of standard of care (naproxen). This is the equivalent of having 80% power to detect a difference of 0.9.

Naproxen was used as a positive control to assess the validity of a crossover design to assess OA pain. Naproxen has demonstrated clear efficacy in a 2-week period [31,35,49,52]. The crossover design was kept to a 2-period design because of the length of time patients would have been required to be in the trial if additional periods had been included. In addition, there was an anticipated reduction in pain over time, and a 3-period design may not have left a large enough window to observe a pain effect in the third period. Therefore, a design consisting of 2 pairs of sequences was used; the first pair of sequences was for PF-04457845 and placebo, and the second was for naproxen and placebo.

One of the major problems with OA studies is the large placebo effect. The crossover trial was adapted to try to reduce this. A blinded placebo run-in and washout were used so that the patients would know that they would receive both an active compound and placebo, but they would not know when the change in treatment occurred. This blinding of when a treatment period starts was also expected to reduce any psychological carryover. Without this blinding, patients may have guessed whether they were on an active treatment or placebo in the second period given their experiences in the first period on the other treatment.

2.4. Decision rules

Clear decision rules were defined up front to quantify what was required in the primary objective of the trial, and this then affected the design of the trial. For the PF-04457845–placebo comparison, 2-part decision criteria for efficacy and futility were used at both the interim and end of trial analyses. The criteria were based on a Bayesian interpretation of the results assuming a noninformative prior. The criteria used were: (1) clear evidence that PF-04457845 has a greater reduction in pain than standard of care—that is, more than 80% sure that PF-04457845 had a greater than 1.8 reduction in WOMAC pain score compared to placebo (naproxen gave an average 1.8 reduction compared to placebo on the WOMAC pain score from a meta analysis carried out from prior parallel group OA studies [unpublished data]); and (2a) no evidence that PF-04457845 reduces pain less than standard of care and (2b) some evidence that it reduces pain relative to placebo, that is, at least 20% probability that PF-04457845 had a greater than 1.8 reduction in WOMAC pain score compared to placebo and greater than 90% sure that it is better than placebo.

The implications of the decision rules applied to this study are presented in Fig. 1B. With these decision rules, the sample size was 72 subjects per sequence, but an interim analysis was included because after 28 subjects per sequence there was a greater than 90% chance of placebo-like compounds not meeting criterion 2. At the interim analysis, sufficient data were expected to have accumulated to determine whether the positive control had reduced the WOMAC pain subscale as expected and was different for placebo, so the naproxen sequences were only planned to be included up to the interim.

At the interim analysis, the following strategies were defined: (1) if PF-04457845 met the first criterion with clear evidence of efficacy better than historical standard of care, then the trial would be stopped; (2) if the second criterion was met but not the first, the trial would continue beyond the interim; and (3) if the second criterion was not met, this trial would be stopped for futility.

A sample size reestimation was to be carried out at the interim analysis after 28 patients completed the trial for each of the PF-04457845 and naproxen arms of the trial (56 patients). It was anticipated that if the trial continued beyond the interim, a further 44 patients were to complete the PF-04457845 arm by the end of the trial, giving a total of 72 patients on this arm (100 subjects in total). With 72 patients, a compound with a true difference from placebo of 0.9 or less would have had less than 20% chance of achieving the second criterion, and if a compound had a true difference greater than 2.7, it would have had a greater than 80% chance of achieving the first criterion. This was based on the intrapatient variability estimated from a previous OA crossover trial (Pfizer internal data); the upper 80% confidence limit of the estimate of the within-subject standard deviation (3.0) was used for sample size estimation as a conservative measure because the estimate was derived from a small study.

2.5. Statistical analyses

Statistical analysis of the efficacy and pharmacodynamic end points used a mixed effect analysis of covariance model, fitted with random patient effect, period and treatment as fixed effects, utilizing the baseline scores (one for each treatment period) as inter- and intrasubject covariates. The intersubject covariate was the average of the 2 baseline values for each subject, and the intrasubject covariate was the difference between the individual baseline value and the subject’s mean baseline [29].

2.6. Dose selection

PF-04457845 has been demonstrated to be safe and well tolerated after single doses of up to 40mg and multiple doses of up to 8mg once daily (q.d.) for 14days [32]. Further, dose regimens of 0.5mg q.d. to 8mg q.d. all demonstrated a system-based maximal effect on FAAH activity (ie, ≥97% inhibition) with a consequent and significant elevation of the measured plasma FAAs AEA, PEA, OEA and LEA. There was no evidence that increasing the dose of PF-04457845 above 0.5mg q.d. led to any further decrease in FAAH activity. Likewise, the elevations in the FAAs were biologically undifferentiable across the dose range, with the possible exception of PEA, where there was a small increase in concentrations between 0.5mg q.d. and 1mg q.d. PF-04457845. In addition, the pharmacodynamic effects of PF-04457845 were sustained, with FAAH activity and FAA concentrations taking up to 2weeks to return to baseline values after cessation of dosing for doses up to 4 mg. To maximize the potential to demonstrate efficacy in this trial in OA patients, a dose regimen of 4mg q.d. PF-04457845 was selected. This dose regimen was predicted to be both maximally efficacious and to have an off-rate of pharmacodynamic effects that would allow for a crossover design with a 2-week washout.

The dose regimen of naproxen was 500mg twice daily, which is the recommended dose for OA.

3. Results

A total of 74 patients were randomly assigned to treatment, 37 of whom received PF-04457845 and 36 of whom received naproxen. Seventy subjects received placebo and 69 subjects completed the trial, the first 56 of whom were included in the interim analysis. A flow chart demonstrating patient enrollment is provided in Supplementary Fig. 1. There were no important differences in demographic and pretreatment characteristics between each of the 4 treatment sequences (Table 1). In addition, baseline HADS scores revealed no differences between treatment sequences and also indicated the absence of significant anxiety and depression in these patients (Table 1).

Table 1:
Patient demographics and baseline characteristics.a
Supplementary Figure 1.

After the interim analysis, it was clear that the positive control, naproxen, had elicited a reduction in pain (assessed by the WOMAC pain subscale) that was differentiated from placebo, but that there was no evidence of any WOMAC pain difference between PF-04457845 and placebo. Hence, the predetermined decision criterion 2 was not met for PF-04457845. The study was therefore stopped after the interim for futility, although a clear effect of naproxen was apparent. This result was confirmed when the analyses were repeated with data from all available subjects at the end of the trial (Fig. 2).

Fig. 2.:
Fitted mean (80% confidence interval, CI) for difference in WOMAC pain (0–20) from placebo (primary end point). With uninformative priors, the decision criterion “90% sure the compound is better than placebo” is equivalent to saying the 1-sided 90% CI does not cross zero or the right-hand end of the 2-sided 80% CI does not cross zero. Hence, an 80% CI is used. The dotted line represents the target pain score used in the decision criteria. Data are from the full analysis set.

The estimated difference in WOMAC pain score after 2weeks of treatment are presented in Fig. 2 and Table 2. The least squares means and 80% confidence intervals suggest no effect of PF-04457845 compared to placebo, whereas naproxen demonstrated a mean reduction in pain score of 1.13 compared to placebo. There was no evidence of a carryover effect between treatment periods. The observed WOMAC pain scores did reduce over time across all treatment sequences (Fig. 3). This reduction was expected and was accounted for in the analyses [pfizer data on file]. Despite this reduction, there was still a sufficient window to observe treatment effects because the trial was able to detect a greater reduction in pain after naproxen compared to placebo. Further analyses demonstrated a similar lack of effect of PF-04457845 compared to placebo in the other WOMAC subscales and the total WOMAC score, whereas naproxen demonstrated statistically significant differences compared to placebo for all these end points (Table 2).

Table 2:
Mean differences (80% confidence intervals) from placebo in WOMAC scores at end of treatment.a
Fig. 3.:
Mean (SD) WOMAC pain scores (0–20) for patients receiving PF-04457845 then placebo (A), placebo then PF-04457845 (B), naproxen then placebo (C) and placebo then naproxen (D). Data are from the full analysis set.

PF-04457845 had no effect on average daily pain scores (assessed by 11-point NRS) in weeks 1 and 2 compared to placebo, whereas there was a clear effect of naproxen compared to placebo (Table 3).

Table 3:
Mean differences (80% confidence intervals) from placebo in daily pain score at weeks 1 and 2.a

During the 2-week treatment periods, the percentage of patients who self-administered rescue medication was the same (59%) for PF-04457845 and placebo, whilst less patients (39%) self-administered rescue medication whilst receiving naproxen (Table 4).

Table 4:
Rescue medication,a

No pharmacokinetic parameters were calculated for PF-04457845 as a result of the relatively sparse sampling schedule. However, the maximum mean concentration observed on day 1 was 25.3ng/mL, and it occurred 1h after the dose was administered, which is similar to the exposure achieved (33.9ng/mL) in healthy volunteers after the same administration schedule [32].

Mean FAAH activity was relatively constant in the absence of PF-04457845; however, in the presence of PF-04457845, FAAH activity decreased by a mean of 96% (Table 5). This value included one subject who had no apparent decrease in FAAH activity; when this subject was excluded, mean FAAH activity decreased by 97% (Table 5). Concentrations of the FAAs were relatively constant over time after placebo dosing; however, concentrations of all 4 FAAs increased rapidly and substantially after PF-04457845 dosing and remained elevated throughout the 14-day trial period. This increase was also observed in the subject who had no apparent change in FAAH activity. Statistical analysis of the FAA data suggested that PF-04457845 caused mean increases compared to placebo of 3.4- to 13.5-fold (Table 5). Concentrations of all the FAAs were generally still elevated 8days after cessation of PF-04457845 dosing; however, FAAH activity and FAA concentrations had returned to baseline values by the start of the second treatment period. There was no apparent difference in the effect of PF-04457845 on any of the FAAs between the 2 treatment sequences.

Table 5:
Statistical analysis of FAAH inhibition and FAA concentrations.a

PF-04457845 was well tolerated in patients with OA, with a safety profile that was indistinguishable from placebo. The most common adverse events are presented in Table 6. There were no adverse events that caused functional unblinding and no cannabinoid-type events.

Table 6:
Adverse events reported by at least 5% of patients in any treatment group.a

4. Discussion

This clinical trial found that in OA patients with chronic pain, naproxen but not PF-04457845 demonstrated analgesic activity according to predefined decision criteria.

Although pharmacokinetic parameters were not calculated for PF-04457845 in this study, plasma concentrations were comparable to those previously seen in healthy subjects [32]. In the previous study, PF-04457845 was rapidly absorbed with a median Tmax of 0.5–1.2h and exhibited linear pharmacokinetics on multiple dosing over the dose range of 0.5 to 8mg q.d. Steady state was achieved by day 7. FAAH1 activity was almost completely inhibited (>97%) within 2h after doses of at least 0.3mg (single dose) and 0.5mg q.d. (multiple dose) PF-04457845 and was maintained for at least 1week after the last dose of the multiple-dose regimens. Mean plasma FAA concentrations increased rapidly (∼2h) to a plateau, which was then maintained after PF-04457845. Both FAAH1 activity and FAA concentrations returned to baseline within 2weeks after cessation of dosing at doses up to 4mg. Hence, the onset of pharmacodynamic effects was similar to that of plasma pharmacokinetics, whilst there was a marked delay in pharmacodynamic offset time compared to plasma pharmacokinetics owing to the irreversible nature of the inhibition of FAAH by PF-04457845.

PF-04457845 has been reported to have effects in rodents that can be interpreted as antinociceptive, where pain sensitivity is enhanced to external stimuli [6]. Specifically, at doses that inhibit plasma and brain FAAH, PF-04457845 reduced mechanical hyperalgesia/allodynia in a manner similar to naproxen in the complete Freund adjuvant model in a manner that was mediated by both CB1 and CB2 receptors. At similar doses in a noninflammatory model of pain, it reduced joint compression threshold to a similar extent to celecoxib. Such behavioural end points in these animal models have traditionally been interpreted as indicating likelihood of potential therapeutic benefit in patients with pain due to OA. Indeed, the recent electrophysiological data from another FAAH inhibitor (URB597) in a supposed model of OA would add further weight supporting the likely potential of FAAH inhibition in relieving pain in OA patients [50]. The absence of detectable analgesic properties in this clinical trial was therefore not only disappointing for the clinical utility of this class of agents, but also poses translational questions regarding animal models.

It is clear (Table 4) that the dose of PF-04457845 administered to humans inhibited FAAH1 nearly completely in these patients, as predicted from studies in healthy volunteers [32] and rats [6]. There was one subject who had no apparent inhibition of FAAH activity. However, this subject did have elevations of all 4 plasma FAAs, consistent with the other subjects, so it seems likely that the FAAH activity data are erroneous in this case.

The use of the same plasma biomarker in humans and rats excludes reduced inhibition of FAAH1 in humans compared with rats as a factor in the negative clinical result reported here in that the degree of FAAH inhibition and its time course are similar between species. In addition, the elevation of plasma AEA and OEA levels is similar between species, indicating that the impact of FAAH1 inhibition on these FAAs translates between species. It was clearly not practicable to assess central penetration of PF-04457845 or inhibition of brain FAAH in OA patients. Whilst it is true that in both OA patients (reported here) and in healthy volunteers [32] there is scant evidence of central nervous system adverse events caused by PF-04457845, this is in accordance with the phenotype of FAAH-/- mice and of the behaviour of rats after administration of a FAAH1 inhibitor. Indeed, PF-04457845 has been demonstrated to affect sleep architecture in a subtle and yet consistent manner in humans and rats, an effect presumed to reflect central pharmacology [pfizer data on file].

The absence of analgesic activity of PF-04457845 in the clinical trial reported here was not a function of the trial design because naproxen used as a positive control revealed robust effects.

It therefore seems that the lack of translation between the animal model results reported previously for PF-04457845 and the clinical results reported here relate more to the comparative mechanisms in rats and humans of nociceptive sensitisation affected downstream from elevated FAA levels. Although the exact translational difference cannot be concluded from this study, 2 overall hypotheses can be suggested.

Firstly, the activity of PF-04457845 in a model in which frank inflammation is apparent (the complete Freund adjuvant model) might lead to the conclusion that the patients recruited for our trial have insufficient inflammatory stimulus for FAAH1 inhibition to lead to analgesic effects. However, the activity of PF-04457845 in the sodium monoiodoacetate model at time points where there is little inflammation [6] does not support this conclusion. One possibility that is raised from the effects of other inhibitors in rodent models of anxiety [27] is that for analgesic responses to be observed, patients need to be more anxious than was the case for the patients in this trial. The HADS score of patients at baseline (Table 1) varied from a mean±SD of 4.0±2.96 to 5.3±3.44, depending on randomisation sequence, indicating absence of depression or anxiety in these patients. It is interesting to note that fibromyalgia patients, who are either heterogenous or homologous for the C385A FAAH1 polymorphism (and who therefore express less FAAH1), have reduced threat-induced amygdala reactivity as detected by fMRI [20]). This may indicate a role of FAAH1 in humans in the modulation of specific aspects of mood. Whether PF-04457845 would prove to be analgesic in other patient populations, such as those with greater intrinsic anxiety, cannot be directly assessed from these results.

Secondly, FAAH1 inhibitors may affect downstream targets in quantitatively distinct manners between species. For example OEA and LEA activate the TRPV1 channel [38], which is thought to be associated with inflammatory pathways. AEA itself acts as a full agonist at TRPV1 in some systems [59]. N-arachidonyl-serotonin, which both inhibits FAAH (weakly) and antagonizes TRPV1, is more efficacious than a more potent and selective FAAH inhibitor (URB597) in animal models of inflammatory pain [14,34], suggesting that in rodents TRPV1 activation may reduce the efficacy of FAAH inhibition. Hence, we speculate that the simultaneous indirect activation of cannabinoid receptors and TRPV1 may have differential effects between species.

A different possibility is that in humans (more than rodents) simultaneous inhibition of FAAH1 and COX-2 are required for pain amelioration. Under some conditions, AEA and 2-AG are substrates for COX-2 [55]. A rodent study demonstrated that firing of dorsal cord neurones receiving inputs from an inflamed paw were inhibited by COX-2 inhibitors by a mechanism involving CB1 receptors [53].

Both of these latter possibilities seem unlikely as a result of the complete absence of analgesic effects in patients, but they cannot be excluded because our clinical trial made no attempt to examine the additive or synergistic effects of PF-04457845 and naproxen.

Despite the disappointing results with PF-04457845, the data indicate the efficiency of the crossover trial design compared to a parallel group study for this indication. The trial provided clear conclusions using fewer subjects and fewer centres, and it was therefore operationally faster than a more traditional parallel group trial. Recruitment into this trial was rapid, and an advantage of small patient numbers is that a trial can be run at fewer centres (most patients in this trial were recruited from 2 centres), making patients and end points more homogeneous, in addition to the low intrasubject variability intrinsic for crossover (as opposed to parallel group) clinical trials. There were few patient discontinuations in this trial and the use of rescue medication was low, perhaps also reflecting the good patient–investigator rapport that can be established with a small number of patients and centres in a clinical trial.

An advantage of using decision criteria before initiation of the trial in the way that they are described here is that they provided a clear understanding of decisions that would be made, given potential outcomes at trial completion and therefore simplified trial interpretation. In addition, the frank psychoactive properties of exogenously administered cannabinoid agents makes functional blinding of clinical trials with these compounds difficult to achieve. The lack of analgesic activity with PF-04457845 in the present study is not affected by such issues because of the clean profile of adverse events seen with this agent, suggesting that it has the potential to be used to explore the direct analgesic, as opposed to psychoactive, properties of cannabinoids in a variety of indications. This assumes that the level of CB-receptor activation induced by FAA elevation after FAAH1 inhibition is equivalent to that induced by exogenous cannabinoids. This would seem to be the case in animals (as evidenced by pharmacological manipulation of FAAH inhibitor effects using CB1 and CB2 antagonists [6]) but has not yet been demonstrated in humans.

Conflict of interest

The trial was sponsored by Pfizer. All authors are or were employees of Pfizer at the time of this research and may own stock in the company. Medical writing support by Valley Writing Solutions Ltd was also supported by Pfizer. There are no other known conflicts of interest to declare.


The authors thank the investigators and patients who participated in this trial. The authors also thank Kuntal Sinha (Pfizer), who was responsible for the analysis of PF-04457845; Kathryn Wright (Pfizer), who was responsible for oversight of the analysis of FAAH activity and FAA concentrations; Samantha Abel (Valley Writing Solutions Ltd), who provided medical writing support; and Anne Heatherington and Gillian Burgess (Pfizer) for support and critical review of the PF-04457845 clinical development programme.


[1]. Abrams DI, Jay CA, Shade SB, Vizoso H, Reda H, Press S, Kelly ME, Rowbotham MC, Petersen KL. Cannabis in painful HIV-associated sensory neuropathy: a randomized placebo-controlled trial. Neurology. 2007;68:515-521.
[2]. Adamczyk P, Golda A, McCreary AC, Filip M, Przegalinski E. Activation of endocannabinoid transmission induces antidepressant-like effects in rats. J Physiol Pharmacol. 2008;59:217-228.
[3]. Ahn K, Johnson DS, Cravatt BF. Fatty acid amide hydrolase as a potential therapeutic target for the treatment of pain and CNS disorders. Expert Opin Drug Discov. 2009;4:763-784.
[4]. Ahn K, Johnson DS, Mileni M, Beidler D, Long JZ, McKinney MK, Weerapana E, Sadagopan N, Liimatta M, Smith SE, Lazerwith S, Stiff C, Kamtekar S, Bhattacharya K, Zhang Y, Swaney S, Van Becelaere K, Stevens RC, Cravatt BF. Discovery and characterization of a highly selective FAAH inhibitor that reduces inflammatory pain. Chem Biol. 2009;16:411-420.
[5]. Ahn K, McKinney MK, Cravatt BF. Enzymatic pathways that regulate endocannabinoid signaling in the nervous system. Chem Rev. 2008;108:1687-1707.
[6]. Ahn K, Smith SE, Liimatta MB, Beidler D, Sadagopan N, Dudley DT, Young T, Wren P, Zhang Y, Swaney S, Van Becelaere K, Blankman JL, Nomura DK, Bhattachar SN, Stiff C, Nomanbhoy TK, Weerapana E, Johnson DS, Cravatt BF. Mechanistic and pharmacological characterization of PF-04457845: a highly potent and selective FAAH inhibitor that reduces inflammatory and noninflammatory pain. J Pharmacol Exp Ther. 2011;338:114-124.
[7]. Altman RD, Hochberg MC, Moskowitcz RW, Schnitzer TJ. Recommendations for the medical management of osteoarthritis of the hip and knee: 2000 update. American College of Rheumatology Subcommittee on Osteoarthritis Guidelines. Arthritis Rheum. 2000;43:1905-1915.
[8]. Bellamy N. Pain assessment in osteoarthritis: experience with the WOMAC Osteoarthritis Index. Semin Arthritis Rheum. 1989;18:14-17.
[9]. Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW. Validation study of WOMAC: a health status instrument for measuring clinically important patient relevant outcomes to antirheumatic drug therapy in patients with osteoarthritis of the hip or knee. J Rheumatol. 1988;15:1833-1840.
[10]. Berman JS, Symonds C, Birch R. Efficacy of two cannabis based medicinal extracts for relief of central neuropathic pain from brachial plexus avulsion: results of a randomised controlled trial. PAIN®. 2004;112:299-306.
[11]. Boger DL, Miyauchi H, Du W, Hardouin C, Fecik RA, Cheng H, Hwang I, Hedrick MP, Leung D, Acevedo O, Guimaraes CR, Jorgensen WL, Cravatt BF. Discovery of a potent, selective, and efficacious class of reversible alpha-ketoheterocycle inhibitors of fatty acid amide hydrolase effective as analgesics. J Med Chem. 2005;48:1849-1856.
[12]. Chang L, Luo L, Palmer JA, Sutton S, Wilson SJ, Barbier AJ, Breitenbucher JG, Chaplan SR, Webb M. Inhibition of fatty acid amide hydrolase produces analgesia by multiple mechanisms. Br J Pharmacol. 2006;148:102-113.
[13]. Cheng Y, Hitchcock SA. Targeting cannabinoid agonists for inflammatory and neuropathic pain. Expert Opin Investig Drugs. 2007;16:951-965.
[14]. Costa B, Bettoni I, Petrosino S, Comelli F, Giagnoni G, Di Marzo V. The dual fatty acid amide hydrolase/TRPV1 blocker, N-arachidonoyl-serotonin, relieves carrageenan-induced inflammation and hyperalgesia in mice. Pharmacol Res. 2010;61:537-546.
[15]. Costa B, Comelli F, Bettoni I, Colleoni M, Giagnoni G. The endogenous fatty acid amide, palmitoylethanolamide, has anti-allodynic and anti-hyperalgesic effects in a murine model of neuropathic pain: involvement of CB(1), TRPV1 and PPARgamma receptors and neurotrophic factors. PAIN®. 2008;139:541-550.
[16]. Cravatt BF, Demarest K, Patricelli MP, Bracey MH, Giang DK, Martin BR, Lichtman AH. Supersensitivity to anandamide and enhanced endogenous cannabinoid signaling in mice lacking fatty acid amide hydrolase. Proc Natl Acad Sci USA. 2001;98:9371-9376.
[17]. Cravatt BF, Saghatelian A, Hawkins EG, Clement AB, Bracey MH, Lichtman AH. Functional disassociation of the central and peripheral fatty acid amide signaling systems. Proc Natl Acad Sci USA. 2004;101:10821-10826.
[18]. Ellis RJ, Toperoff W, Vaida F, van den Brande G, Gonzales J, Gouaux B, Bentley H, Atkinson JH. Smoked medicinal cannabis for neuropathic pain in HIV: a randomized, crossover clinical trial. Neuropsychopharmacology. 2009;34:672-680.
[19]. Gobbi G, Bambico FR, Mangieri R, Bortolato M, Campolongo P, Solinas M, Cassano T, Morgese MG, Debonnel G, Duranti A, Tontini A, Tarzia G, Mor M, Trezza V, Goldberg SR, Cuomo V, Piomelli D. Antidepressant-like activity and modulation of brain monoaminergic transmission by blockade of anandamide hydrolysis. Proc Natl Acad Sci USA. 2005;102:18620-18625.
[20]. Hariri AR, Gorka A, Hyde LW, Kimak M, Halder I, Ducci F, Ferrell RE, Goldman D, Manuck SB. Divergent effects of genetic variation in endocannabinoid signaling on human threat- and reward-related brain function. Biol Psychiatry. 2009;66:9-16.
[21]. Holt S, Comelli F, Costa B, Fowler CJ. Inhibitors of fatty acid amide hydrolase reduce carrageenan-induced hind paw inflammation in pentobarbital-treated mice. comparison with indomethacin and possible involvement of cannabinoid receptors. Br J Pharmacol. 2005;146:467-476.
[22]. Hosking RD, Zajicek JP. Therapeutic potential of cannabis in pain medicine. Br J Anaesth. 2008;101:59-68.
[23]. Howlett AC, Barth F, Bonner TI, Cabral G, Casellas P, Devane WA, Felder CC, Herkenham M, Mackie K, Martin BR, Mechoulam R, Pertwee RG. International Union of Pharmacology. XXVII. Classification of cannabinoid receptors. Pharmacol Rev. 2002;54:161-202.
[24]. Huitron-Resendiz S, Sanchez-Alavez M, Wills DN, Cravatt BF, Henriksen SJ. Characterization of the sleep–wake patterns in mice lacking fatty acid amide hydrolase. Sleep. 2004;27:857-865.
[25]. Jayamanne A, Greenwood R, Mitchell VA, Aslan S, Piomelli D, Vaughan CW. Actions of the FAAH inhibitor URB597 in neuropathic and inflammatory chronic pain models. Br J Pharmacol. 2006;147:281-288.
[26]. Karst M, Salim K, Burstein S, Conrad I, Hoy L, Schneider U. Analgesic effect of the synthetic cannabinoid CT-3 on chronic neuropathic pain: a randomized controlled trial. JAMA. 2003;290:1757-1762.
[27]. Kathuria S, Gaetani S, Fegley D, Valino F, Duranti A, Tontini A, Mor M, Tarzia G, La Rana G, Calignano A, Giustino A, Tattoli M, Palmery M, Cuomo V, Piomelli D. Modulation of anxiety through blockade of anandamide hydrolysis. Nat Med. 2003;9:76-81.
[28]. Kellgren JH, Lawrence JS. Radiological assessment of osteo-arthrosis. Ann Rheum Dis. 1957;16:494-502.
[29]. Kenward MG, Roger JH. The use of baseline covariates in crossover studies. Biostatistics. 2010;11:1-17.
[30]. Kinsey SG, Long JZ, O’Neal ST, Abdullah RA, Poklis JL, Boger DL, Cravatt BF, Lichtman AH. Blockade of endocannabinoid-degrading enzymes attenuates neuropathic pain. J Pharmacol Exp Ther. 2009;330:902-910.
[31]. Kivitz AJ, Moskowitz RW, Woods E, Hubbard RC, Verburg KM, Lefkowith JB, Geis GS. Comparative efficacy and safety of celecoxib and naproxen in the treatment of osteoarthritis of the hip. J Int Med Res. 2001;29:467-479.
[32]. Li GL, Winter H, Arends R, Jay GW, Le V, Young T, Huggins JP. Assessment of the pharmacology and tolerability of PF-04457845, an irreversible inhibitor of fatty acid amide hydrolase-1, in healthy subjects. Br J Clin Pharmacol. 2012;73:706-716.
[33]. Lichtman AH, Leung D, Shelton CC, Saghatelian A, Hardouin C, Boger DL, Cravatt BF. Reversible inhibitors of fatty acid amide hydrolase that promote analgesia: evidence for an unprecedented combination of potency and selectivity. J Pharmacol Exp Ther. 2004;311:441-448.
[34]. Maione S, De Petrocellis L, de Novellis V, Moriello AS, Petrosino S, Palazzo E, Rossi FS, Woodward DF, Di Marzo V. Analgesic actions of N-arachidonoyl-serotonin, a fatty acid amide hydrolase inhibitor with antagonistic activity at vanilloid TRPV1 receptors. Br J Pharmacol. 2007;150:766-781.
[35]. Makarowski W, Zhao WW, Bevirt T, Recker DP. Efficacy and safety of the COX-2 specific inhibitor valdecoxib in the management of osteoarthritis of the hip: a randomized, double-blind, placebo-controlled comparison with naproxen. Osteoarthritis Cartilage. 2002;10:290-296.
[36]. Martín-Sánchez E, Furukawa TA, Taylor J, Martin JLR. Systematic review and meta-analysis of cannabis treatment for chronic pain. Pain Med. 2009;10:1353-1368.
[37]. Massa F, Marsicano G, Hermann H, Cannich A, Monory K, Cravatt BF, Ferri GL, Sibaev A, Storr M, Lutz B. The endogenous cannabinoid system protects against colonic inflammation. J Clin Invest. 2004;113:1202-1209.
[38]. Movahed P, Jönsson BA, Birnir B, Wingstrand JA, Jørgensen TD, Ermund A, Sterner O, Zygmunt PM, Högestätt ED. Endogenous unsaturated C18 N-acylethanolamines are vanilloid receptor (TRPV1) agonists. J Biol Chem. 2005;280:38496-38504.
[39]. Naidu PS, Kinsey SG, Guo TL, Cravatt BF, Lichtman AH. Regulation of inflammatory pain by inhibition of fatty acid amide hydrolase. J Pharmacol Exp Ther. 2010;334:182-190.
[40]. Naidu PS, Varvel SA, Ahn K, Cravatt BF, Martin BR, Lichtman AH. Evaluation of fatty acid amide hydrolase inhibition in murine models of emotionality. Psychopharmacology. 2007;192:61-70.
[41]. Petrosino S, Di Marzo V. FAAH and MAGL inhibitors: therapeutic opportunities from regulating endocannabinoid levels. Curr Opin Investig Drugs. 2010;11:51-62.
[42]. Posner K, Oquendo MA, Gould M, Stanley B, Davies M. Columbia Classification Algorithm of Suicide Assessment (C-CASA): classification of suicidal events in the FDA’s pediatric suicidal risk analysis of antidepressants. Am J Psychiatry. 2007;164:1035-1043.
[43]. Rice ASC. Should cannabinoids be used as analgesics for neuropathic pain? Nat Rev Neurol. 2008;4:654-655.
[44]. Rice ASC, Lever I, Zarnegar R., 2008. Cannabinoids and analgesia, with special reference to neuropathic pain. In: McQuay HJ, Kalso E, Moore RA, editors., Systematic reviews in pain research: methodology refined. IASP Press, Seattle, pp. 233-246.
[45]. Rog DJ, Nurmikko TJ, Friede T, Young CA. Randomized, controlled trial of cannabis-based medicine in central pain in multiple sclerosis. Neurology. 2005;65:812-819.
[46]. Romero-Sandoval A, Eisenach JC. Spinal cannabinoid receptor type 2 activation reduces hypersensitivity and spinal cord glial activation after paw incision. Anesthesiology. 2007;106:787-794.
[47]. Russo R, Loverme J, La Rana G, Compton TR, Parrott J, Duranti A, Tontini A, Mor M, Tarzia G, Calignano A, Piomelli D. The fatty acid amide hydrolase inhibitor URB597 (cyclohexylcarbamic acid 3′-carbamoylbiphenyl-3-yl ester) reduces neuropathic pain after oral administration in mice. J Pharmacol Exp Ther. 2007;322:236-242.
[48]. Salim K, Schneider U, Burstein S, Hoy L, Karst M. Pain measurements and side effect profile of the novel cannabinoid ajulemic acid. Neuropharmacology. 2005;48:1164-1171.
[49]. Schnitzer TJ, Kivitz AJ, Lipetz RS, Sanders N, Hee A. Comparison of the COX-inhibiting nitric oxide donator AZD3582 and rofecoxib in treating the signs and symptoms of osteoarthritis of the knee. Arthritis Rheum. 2005;53:827-837.
[50]. Schuelert N, Johnson MP, Oskins JL, Jassal K, Chambers MG, McDougall JJ. Local application of the endocannabinoid hydrolysis inhibitor URB597 reduces nociception in spontaneous and chemically induced models of osteoarthritis. PAIN®. 2011;152:975-981.
[51]. Svendsen KB, Jensen TS, Bach FW. Does the cannabinoid dronabinol reduce central pain in multiple sclerosis? Randomised double blind placebo controlled crossover trial. BMJ. 2004;329:253-257.
[52]. Svensson O, Malmenas M, Fajutrao L, Roos EM, Lohmander LS. Greater reduction of knee than hip pain in osteoarthritis treated with naproxen, as evaluated by WOMAC and SF-36. Ann Rheum Dis. 2006;65:781-784.
[53]. Telleria-Diaz A, Schmidt M, Kreusch S, Neubert AK, Schache F, Vazquez E, Vanegas H, Schaible HG, Ebersberger A. Spinal antinociceptive effects of cyclooxygenase inhibition during inflammation: involvement of prostaglandins and endocannabinoids. PAIN®. 2010;148:26-35.
[54]. Wade DT, Makela P, Robson P, House H, Bateman C. Do cannabis-based medicinal extracts have general of specific effects on symptoms in multiple sclerosis? A double-blind, randomized, placebo-controlled study on 160 patients. Mult Scler. 2004;10:434-441.
[55]. Woodward DF, Carling RW, Cornell CL, Fliri HG, Martos JL, Pettit SN, Liang Y, Wang JW. The pharmacology and therapeutic relevance of endocannabinoid derived cyclo-oxygenase (COX)-2 products. Pharmacol Ther. 2008;120:71-80.
[56]. Zajicek J, Fox P, Sanders H. Cannabinoids for treatment of spasticity and other symptoms related to multiple sclerosis (CAMS study): multicentre randomised placebo-controlled trial. Lancet. 2003;362:1517-1526.
[57]. Zajicek JP, Sanders HP, Wright DE, Vickery PJ, Ingram WM, Reilly SM, Nunn AJ, Teare LJ, Fox PJ, Thompson AJ. Cannabinoids in Multiple Sclerosis (CAMS) study: safety and efficacy data for 12 months follow up. J Neurol Neurosurg Psychiatry. 2005;76:1664-1669.
[58]. Zigmond AS, Snaith RP. The Hospital Anxiety and Depression Scale. Acta Psychiatr Scand. 1983;67:361-370.
[59]. Zygmunt PM, Petersson J, Andersson DA, Chuang H, Sørgård M, Di Marzo V, Julius D, Högestätt ED. Vanilloid receptors on sensory nerves mediate the vasodilator action of anandamide. Nature. 1999;400:452-457.

Appendix A Supplementary data

Supplementary data associated with this article can be found, in the online version, at


Crossover design; FAAH1; Osteoarthritis; Pain; PF-04457845; Translation

© 2012 Lippincott Williams & Wilkins, Inc.