Pharmacological magnetic resonance imaging (phMRI) is one method by which a drug’s pharmacodynamic effects in the brain can be assessed. Although phMRI has been frequently used in preclinical and clinical settings, the extent to which a phMRI signature for a compound translates between rodents and humans has not been systematically examined. In the current investigation, we aimed to build on recent clinical work in which the functional response to 0.1 and 0.2 mg/70 kg i.v. buprenorphine (partial µ-opioid receptor agonist) was measured in healthy humans. Here, we measured the phMRI response to 0.04 and 0.1 mg/kg i.v. buprenorphine in conscious, naive rats to establish the parallelism of the phMRI signature of buprenorphine across species. PhMRI of 0.04 and 0.1 mg/kg i.v. buprenorphine yielded dose-dependent activation in a brain network composed of the somatosensory cortex, cingulate, insula, striatum, thalamus, periaqueductal gray, and cerebellum. Similar dose-dependent phMRI activation was observed in the human phMRI studies. These observations indicate an overall preservation of pharmacodynamic responses to buprenorphine between conscious, naive rodents and healthy human subjects, particularly in brain regions implicated in pain and analgesia. This investigation further demonstrates the usefulness of phMRI as a translational tool in neuroscience research that can provide mechanistic insight and guide dose selection in drug development.
The implementation of the translational medicine approach has recently gained substantial interest and justification in neuroscience research (Day et al., 2009; Feuerstein and Chavez, 2009; Wan et al., 2009). In particular, neuroimaging methodology is being increasingly used for evaluating pharmacological modulation of brain systems (Fox et al., 2009; Wong et al., 2009; Cole et al., 2012). Translational imaging techniques, such as positron emission tomography (PET) or single photon emission computed tomography (SPECT), are often used to qualitatively and quantitatively assess target engagement of novel and marketed pharmacological compounds. However, the lack of available radioactive ligands for many compounds of interest (or the difficulty in producing them) renders PET and SPECT methodologies sometimes of limited use. In contrast, pharmacological magnetic resonance imaging (phMRI) relies on the endogenous blood-oxygenated level-dependent contrast mechanisms to evaluate drug action on the central nervous system (Breiter et al., 1997; Leslie and James, 2000; Jenkins, 2012). The combination of simultaneous PET-functional MRI (Jenkins 2012) studies would provide the ideal situation as receptor binding (detected by PET), and pharmacodynamics effects (captured by functional MRI) could be determined. If radioligands are not available, phMRI will still provide important information, because dose-response curves can be assessed and potentially the optimal dose determined. Furthermore, phMRI would detect activity induced by the compound across the whole brain, and on the basis of known brain circuits, it is possible to assess whether the compound might have significant adverse effects, such as excessive sedative or potential addictive characteristics (Becerra et al., 2006). In addition, phMRI could detect early brain responses (and potential adverse effects) that do not necessarily have an initial behavioral outcome, because in some cases (e.g., antidepressants), it is required to have patients exposed to the drug for a couple of weeks.
In the simplest form of phMRI, the central hemodynamic response to the compound administration is measured in the absence of any overt physical stimulus or specific cognitive task (Chen et al., 2011). In other words, the signal changes measured are those occurring in direct response to the compound itself (chemical stimulus), in contrast to a modulation design in which the effect of a compound in altering the response evoked by an applied task or stimulus.
Recently, we reported on the central, dose-dependent pharmacodynamic effects of buprenorphine in humans with use of functional imaging methodology, in part, to further understand the neurobiological mechanisms underlying its analgesic properties (Borras et al., 2004; Becerra et al., 2006; Leppa et al., 2006; Upadhyay et al., 2011). Although buprenorphine is an opioid that has potency for multiple opioid receptor types (μ-, δ-, and κ-opioid-like receptor 1/nociceptin) (Sadee et al., 1982; Rothman et al., 1995; Hawkinson et al., 2000; Huang et al., 2001), the partial agonist action of buprenorphine at supraspinal µ-opioid receptor site is thought to be the major driver of its analgesic effect (Lutfy et al., 2003; Ide et al., 2004). Because of the known neuropharmacology of buprenorphine and previous clinical functional imaging work performed with this compound, a unique opportunity exists for evaluating and validating the back-translation of a phMRI signature observed in the brain.
Here, we aimed to determine the extent to which the buprenorphine phMRI signature translates between species by measuring brain responses after 2 i.v. doses (0.04 and 0.1 mg/kg) of buprenorphine in naive rodents. PhMRI measurements were obtained in conscious rats to eliminate possible confounds stemming from drug-anesthetic interactions in the brain. The buprenorphine doses for rodent (Kouya et al., 2002; Christoph et al., 2005) and healthy human (Yassen et al., 2006) phMRI studies were chosen from earlier reported dose-response curves for analgesic effects. Because of the known distribution of opioid receptors, particularly μ-opioid receptors, in the rodent and human brain (Goodman and Pasternak, 1985; Mansour et al., 1987), we hypothesized that robust phMRI responses to buprenorphine would be observed in neuronal structures, such as the ventral and dorsal striatum, thalamus, and cingulate cortex in both species. The data presented in this work provide further support for the use of phMRI as a translational neuroscience tool that can provide a mechanistic evaluation of both novel and marketed pharmacological compounds.
Materials and Methods
Preclinical PhMRI Studies
Male Sprague-Dawley rats weighing 300–350 g were obtained from Charles River Laboratories (Charles River, MA). Animals were housed in pairs and maintained in ambient temperature on a 12:12 light-dark cycle (lights on at 9 AM). Food and water were provided ad libitum. All animals were acquired and cared for in accordance with guidelines published in the Guide for the Care and Use of Laboratory Animals (National Institutes of Health Publication No. 85-23, Revised 1985). The procedure for immobilization and functional imaging in awake animals was approved and monitored by the Massachusetts General Hospital, Institutional Animal Care and Use Committee.
Buprenorphine was purchased from Reckitt Benckiser Pharmaceuticals (Richmond, VA) through the Hospital’s pharmacy. Rats were assigned to 1 of 3 groups: 0.04 mg/kg, 0.1 mg/kg, and saline. Buprenorphine was diluted with saline to produce solutions of 0.04 and 0.1 mg/kg per ml. The 0.04 mg/kg, 0.1 mg/kg, and saline cohorts consisted of 11, 13, and 14 subjects, respectively. The high and low doses were chosen from reported dose-response curves for analgesic effects, with the lower dose at ED50 and the higher dose above ED100 (Kouya et al., 2002; Christoph et al., 2005).
The conscious rat training and imaging paradigm was adapted from previously described methods (Ferris et al., 2006; Becerra et al., 2011a,b). Initially, animals were acclimated over a period of 3 consecutive days before the experiment. On each of these 3 days, rats were lightly anesthetized with 2% isoflurane and placed in a holder similar to the one used for imaging for 60 minutes; the holder was placed in a box to simulate the darkness of the bore of the magnet. During the 60 minutes, the animals were exposed to MRI recorded sounds. The sound level during habituation was 97 dB; it has been added to the text.
PhMRI Data Acquisition.
Immediately before MR scans, rats were anesthetized under 2% isoflurane and then placed into a small animal head and body holder designed for adult male rats weighing 300–350 g (Insight NeuroImaging System, Worcester, MA). The head of the animal was fitted into a cylindrical head holder surrounded by a 5.2-cm diameter birdcage radiofrequency coil. The body of the animal was placed into the body holder for unrestricted respiration but minimal movement. To minimize movement, the lower portion of the body, including both legs (in a prone position), was placed in a specially designed mold. A 24-gauge catheter was placed in the rat tail vein under isoflurane. A PE 10 line was inserted through the catheter. Acquisition of anatomic scans was followed by the drug infusion or phMRI scan. Infusion was performed during the phMRI scan via a syringe connected at the end of the PE 10 line outside the magnet bore while the animal was inside the magnet. The phMRI procedure had the following sequence of steps: 5 minutes baseline, infusion of the drug for over 2 minutes, followed by 20 minutes of scanning. PhMRI data were acquired for drug or saline administration.
Magnetic Resonance images were acquired in a Bruker Biospec 4.7 T/40-cm horizontal magnet (Oxford Instrument, Oxford, UK) equipped with a Biospec Bruker console (Bruker, Billerica, MA) and a 20-G/cm magnetic field gradient insert (ID=12cm, 120-μs rise time). Functional images were continuously acquired for 25 minutes and were obtained with a gradient echo EPI (Echo-Planar Imaging) pulse sequence [TE/TR (Echo Time/Repetition Time) = 12ms/2.5s, FOV (Field-of-View) = 3cm, 1.5mm slice thickness, 64×64 matrix] 600 volumes.
PhMRI Data Analysis.
PhMRI data were preprocessed using FSL (http://www.fmrib.ox.ac.uk/fsl/). Preprocessing steps consisted of motion correction, brain extraction, registration to a reference template, and spatial smoothing using a Gaussian filter of 0.7 mm FWHM (Full width half maximum). Because the brain response to an infusion could be in the order of minutes, no temporal filtering (high pass) was applied to eliminate slow fluctuations from the data, because it might eliminate the signal of interest. Spikes in the data were determined using the outlier time point detection and exclusion tool in FSL. Infusion responses were estimated using a general linear model; it was modeled on the basis of plasma pharmacokinetics of buprenorphine in rats (Ohtani et al., 1994; Gopal et al., 2002) and using a ramp model for humans (Pendse et al. 2010, Upadhyay et al. 2012). A linear drift EV (explanatory variable) and EVs to explain intensity changes (spikes) and the motion parameters were also added to the design matrix to capture nuisance signals. Group-level statistical analysis was conducted using a mixed-effects model comparing 0.04 mg/kg buprenorphine with saline and 0.1 mg/kg buprenorphine with saline conditions. Multiple comparison correction was performed on each group-level statistical map using a using a Gaussian mixture modeling approach (Pendse et al., 2009), in which a posterior probability threshold of 0.5 was implemented. Activation and deactivation statistical maps obtained from group-level comparisons were registered to a reference anatomic template according to the Paxinos Atlas 4th Edition that has been aligned with anatomic MRI brain images (Fig. 1) (Becerra et al., 2011a,b). The latter enabled a determination of which regions possessed significant activation and deactivation after 0.04 and 0.1 mg/kg buprenorphine administration.
Clinical PhMRI Studies
The methodology and results for the clinical phMRI investigation of buprenorphine have been described and reported previously (Upadhyay et al., 2011; Upadhyay et al., 2012). in brief, 24 healthy, right-handed males were included in this study. Two i.v. doses of buprenorphine [0.1 (N = 12) and 0.2 mg/70kg (N = 12)] were investigated. Buprenorphine was purchased from Reckitt Benckiser Pharmaceuticals, Inc. through the McLean Hospital pharmacy. Each subject underwent two scanning sessions that were ∼2 weeks apart, where a placebo (saline solution) or buprenorphine (drug plus saline solution) was administered. A randomized, double-blinded, crossover study design was implemented. All volunteers gave informed consent before study participation. The Institutional Review Board at McLean Hospital approved this investigation.
Subsequent to anatomic scanning, subjects underwent a 25-minute phMRI scan, during which either drug or placebo (saline solution) was administered. All infusions were controlled by an automatic micro-injector (Medrad Spectris, Columbus, OH). In the phMRI scan, a 5-minute baseline was collected before the first infusion of buprenorphine or saline alone for the placebo scan. Four infusions, totaling 8 ml, were performed at 5, 7, 9, and 11 minutes. The full drug or placebo delivery was performed over four injections to minimize respiratory depression, which is known to occur after opioid administration. This procedure has been implemented in previous phMRI studies in which pharmacodynamic effects of morphine have been investigated (Borras et al., 2004; Becerra et al., 2006; Leppa et al., 2006; Upadhyay et al., 2011).
Infusion responses to 0.1 and 0.2 mg/70kg buprenorphine were also estimated using a general linear model. Here, the main EV, modeling the effects of i.v. administration of buprenorphine, consisted of a ramp function. The profile of the ramp function was based on the pharmacokinetics elicited by the four closely spaced infusions of buprenorphine, in which drug concentrations were expected to increase quickly during buprenorphine administration (Yassen et al., 2006) but remain stable after the infusions were completed and for the duration of the phMRI scan. Similar to rodent phMRI analysis, group-level statistical analysis was conducted using a mixed-effects model comparing drug and placebo conditions, whereas the multiple comparison correction was performed on each group-level statistical map with use of the Gaussian mixture modeling approach and a posterior probability threshold of 0.5.
Buprenorphine-Induced Activation in Rats and Humans.
In the conscious rat, phMRI of 0.04 and 0.1 mg/kg i.v. buprenorphine yielded dose-dependent phMRI activation in cortical structures, such as the somatosensory, motor, and cingulate cortices and in subcortical structures, such as the thalamus, hippocampal formation, striatum, and periaqueductal gray (Fig. 1A; Supplemental Tables 1–3; Table 1). A highly similar, dose-dependent phMRI activation pattern was observed in the human buprenorphine phMRI dataset, in which functional effects of 0.1 and 0.2 mg/70kg i.v. buprenorphine were evaluated (Fig. 1B; Supplemental Tables 1–3; Table 2). For the human 0.1 mg/70kg buprenorphine cohort, significant phMRI activation did not survive the multiple comparison correction. However, at subthreshold level, activation patterns similar to those obtained at the higher dose were observed (Supplemental Fig. 1). In Fig. 1, structures labeled on the rat and human brain highlight those regions where significant activation was measured for the higher doses of buprenorphine tested in each species (rodent: 0.1 mg/kg i.v. buprenorphine; human: 0.2 mg/70kg i.v. buprenorphine). Table 3 indicates changes in brain structures observed to follow a dose response in the phMRI signals. Structures have been generically matched, because they do not necessarily preserve the same functions and characteristics across species. No similarity across brains was observed for areas displaying more activity at the low dose than at the high dose.
Buprenorphine-Induced Deactivation in Rats and Humans.
The 0.04 mg/kg rodent cohort possessed significant phMRI deactivation (drug less than saline) in cortical and subcortical areas: somatosensory cortex, cingulate cortex, thalamus, hypothalamus, and hippocampal formation (Fig. 2; Supplemental Tables 1–2). The 0.1 mg/kg rodent cohort did not show a greater level of deactivation in regions such as the cingulate, but robust deactivation was present in mammillary nuclei, amygdala, dentate gyrus, and brainstem (Fig. 2). Thus, with respect to phMRI deactivation, a clear dose-dependent relationship was not observed across all brain structures. PhMRI deactivation that surpassed the multiple comparison correction was not observed for either buprenorphine dose evaluated in the human studies (unpublished data).
Low-Dose Versus High-Dose Buprenorphine in Rats and Humans.
To further elucidate dose-response relationships in rodent and human datasets, a group-level comparison was performed between doses evaluated in rats and humans. In the rat, 0.1 mg/kg buprenorphine yielded significantly greater phMRI responses throughout the brain, in comparison with the 0.04 mg/kg dose (Table 1). The greater phMRI responses to the higher dose were observed in cortex (i.e., cingulate, somatosensory cortex, and insula), subcortical structures (i.e., caudate-putamen, hippocampus, and thalamus), and cerebellum; 0.04 > 0.1 mg/kg buprenorphine was only observed in the fimbria-fornix and pons. A very similar trend in humans was observed when comparing 0.1 and 0.2 mg/70 kg buprenorphine doses. For example, the cortex, subcortical structures and cerebellum showed greater activation for the 0.2 mg/70 kg buprenorphine dose. However, unlike the rodent dataset, the lower buprenorphine dose (0.1 mg/70 kg) demonstrated greater phMRI activation in some cortical areas.
The opioid buprenorphine is a well-known analgesic that has multiple opioid receptor targets (μ-, δ-, κ-, and ORL1/nociceptin receptors) in the central nervous system. Because of buprenorphine’s clinical efficacy as an analgesic for acute (Johnson et al., 2005) and neuropathic (Hans, 2007; Pergolizzi et al., 2009; Pergolizzi et al., 2010; Vadivelu and Anwar, 2010; Guetti et al., 2011) pain and the known analgesic doses used in rodents (Kouya et al., 2002; Christoph et al., 2005) and humans (Yassen et al., 2006), phMRI of this particular opioid was an ideal test case in which the translatability of a phMRI readout could be elucidated. For human studies specifically, the doses evaluated did not yield any reported subjective feelings by healthy volunteers to the drug versus placebo during the scanning session, thus enabling a double-blinded, cross-over (drug versus placebo) study paradigm to be performed. The exposures in terms of Cmax reached in the human studies for 0.1 and 0.2 mg/70 kg buprenorphine were 0.96 and 2.08 ng/ml, respectively (Upadhyay et al., 2012). Although plasma samples were not obtained in the present rodent phMRI studies (male Sprague-Dawley rats weighing 300–350 g), previous pharmacokinetic assessment of buprenorphine in rat (male Wistar rats weighing 225–250 g) by Yassen et al. (2006) suggests plasma exposures of 1–10 ng/ml for the 0.04 and 0.1 mg/kg doses evaluated in our experiments. Thus, the exposures evaluated between human and rodent phMRI work fall within a similar range.
In this report, we have showed that phMRI activation after i.v. administration of buprenorphine back-translates to conscious, naive rodents from healthy human subjects and that phMRI responses to buprenorphine were observed in a number of brain regions that facilitate pain processing in both species. As summarized in Table 3, phMRI indicates a similarity across species in brain structures responding to different buprenorphine doses (with the caveat of the correspondence across species of brain substrates). The areas are also well known to have large density of μ-opioids, and the increased phMRI response could be attributable in part to a larger receptor occupancy. This functional imaging work further demonstrates the usefulness of phMRI with respect to defining the extent to which dose-dependent changes in brain activity occur after systemic administration of an opioid.
phMRI Induced by Buprenorphine in Rats and Humans.
Common phMRI activation was observed in a number of brain regions for the higher doses of buprenorphine (i.e., 0.1 mg/kg and 0.2 mg/70 kg for rat and human, respectively). These regions included the somatosensory cortex, cingulate, insula, thalamus, striatum, hippocampal formation, periaqueductal gray, and cerebellum, which are all implicated in pathophysiological conditions that include pain (Elman et al., 2011; Borsook, 2012; Tracey and Dickenson, 2012) and have high densities of opioid receptors (Zubieta et al., 2000; Peng et al., 2012). Thus, in large part, the phMRI signature for buprenorphine was observed to translate between conscious, naive rodents and healthy human subjects in a dose-dependent manner. However, in the rodent dataset specifically, we observed that the 0.04 and 0.1 mg/kg doses of buprenorphine induced deactivation in structures, such as the dentate gyrus, entorhinal cortex, and mammillary nucleus. This is in contrast to the clinical cohort, in which no significant deactivation was observed at either buprenorphine dose tested. A similar inconsistency has been observed for other pharmacological compounds, such as ketamine. Here, phMRI activation in the brain is the dominant effect observed in awake, naive rodents and healthy human subjects (Chin et al., 2011; Baker et al., 2012; De Simoni et al., 2012). However, in the preclinical setting, phMRI deactivation is observed in regions, such as the cerebellum, substantia nigra, and periaqueductal gray; deactivation is minimally observed in the human subgenual cingulate cortex.
One limitation of the current study was an inability to exactly determine the physiologic cause of this discrepancy between preclinical and clinical findings. It is speculated, however, that the difference in distribution pattern or densities of various opioid receptor types between rats and humans (Mansour et al., 1995; Peng et al., 2012) in conjunction with distinct binding affinities at each opioid receptor (μ: 2.7 ± 0.4 pM, δ: 33 ± 1.6 nM, κ: 2.1 ± 0.2 pM and ORL1/nociceptin: 25 ± 0.3 μM) (Brown et al., 2011) may contribute to the variability observed between rodent and human buprenorphine-induced phMRI signatures. Considering that the 0.04 mg/kg dose in rats corresponds to the ED50 dose of buprenorphine [0.1 mg/kg: ED100; 0.15 mg/70 kg buprenorphine or higher: clinically analgesic (Escher et al., 2007)], the measured deactivation may arise as the result of inhibitory processes uniquely occurring in rats. Buprenorphine is known not to induce changes in blood gases (Hreiche et al., 2006) that could significant alter the blood oxygenation level dependent signal. Therefore, it is unlikely that the observed differences arise from buprenorphine-induced physiologic adverse effects. The difference in phMRI activity potentially denotes a species difference in biologic systems in the brain but requires further investigation.
The phMRI activation induced by the higher buprenorphine doses is in accordance with other opioids, particularly those with agonist properties at the μ-opioid receptor site. For example, preclinical and clinical studies observe a robust phMRI signal in the striatum, hippocampus, and cingulate after infusion of distinct analgesic opioids: buprenorphine, morphine (Becerra et al., 2006; Liu et al., 2007), and remifentanil (Leppa et al., 2006; Liu et al., 2007). Moreover, across three independent clinical studies involving healthy volunteers (Becerra et al., 2006; Leppa et al., 2006), phMRI activation is consistently observed for the three μ-opioid agonists or partial agonists. However, in rodent studies, both phMRI activation and deactivation has been observed after administration of buprenorphine, morphine, and remifentanil (Liu et al., 2007). Such a discrepancy in the phMRI signal across the preclinical work is likely to be a function of whether functional imaging was performed in an awake or anesthetized state (Liang et al., 2012) and the type of anesthetic used (Hodkinson et al., 2012). Because binding potentials and behavioral outcomes for different opioids at different opioid receptors also differ [e.g., hydromorphone: µ-opioid agonist; butrophanol: mixed µ-opioid agonist/antagonist with some kappa activity (Schlaepfer et al., 1998)], a degree of inconsistency in phMRI responses between species and opioids tested can be expected. Despite the differences observed between preclinical phMRI studies, the common set of brain structures or systems modulated by buprenorphine, morphine, and remifentanil indicate a rodent-human phMRI signature for opioid-induced effects in the brain. To further determine the consistency of brain modulation across opioids in multiple species, a systematic preclinical investigation using the same anesthetized or awake imaging paradigm for all drugs tested would be of benefit.
Use of phMRI Approaches.
Since the advent of phMRI, a number of mechanisms of action and compounds have been evaluated in the preclinical domain. With use of phMRI, drug action on rodent brain function has been evaluated for a range of compounds that include amphetamine (Chen et al., 2005), nicotine (Gozzi et al., 2006), remifentanil (Liu et al., 2007), and cocaine (Schwarz et al., 2004). Previous studies have also used phMRI as a means to characterize the effects of a pharmacological pretreatment (eg, 5HT-2A receptor antagonist) on a particular challenge (e.g., phencylidine) (Gozzi et al., 2008; Gozzi et al., 2010; Chin et al., 2011; Large et al., 2011; McKie et al., 2011; Baker et al., 2012). Such studies can be highly informative with regard to understanding the underlying neurobiology (receptor and brain circuit level) that drives a certain cognitive phenomenon. To a much lesser extent, phMRI has been performed in nonhuman primates, in which dopamine function has been more often evaluated (Zhang et al., 2001; Jenkins et al., 2004; Luan et al., 2008).
Although phMRI has been frequently used in preclinical and clinical settings, the number of examples in which phMRI responses are translated between rodents and humans is limited to a few cases. Of interest, ketamine phMRI infusion responses have been measured in naive rats (Littlewood et al., 2006; Chin et al., 2011) and healthy human subjects (De Simoni et al., 2012); robust responses were detected in the cingulate, frontal cortex, and hippocampal formation in both species (Bifone and Gozzi, 2012). Similarly, a highly convergent phMRI signature has been observed between rodents (Liu et al., 2007) and humans (Leppa et al., 2006) after acute remifentanil administration, in which activation was present in brain structures, such as the striatum, thalamus, hippocampus, and cingulate cortex. The current study builds on these earlier preclinical and clinical investigations and provides an additional example of how phMRI can offer a translational pharmacodynamic measure. Furthermore, a plausible next step is the forward and backward translation of other functional imaging end points, such as functional connectivity. On the basis of the work of Schwarz et al. and Vollm et al., functional changes in the brain driven by an amphetamine challenge elicited correlated responses in reward circuitry in rodents and humans, respectively (Vollm et al., 2004; Schwarz et al., 2007). Furthermore, brain circuitry, such as the default-mode network, that were initially observed in nonhuman primates and humans (Raichle et al., 2001; Vincent et al., 2007), have recently been measured to a great extent in the rodent brain (Becerra et al., 2011b; Upadhyay et al., 2011; Lu et al., 2012). This is also true for other networks, such as the striatal system (Becerra et al., 2011b; Jonckers et al., 2011; Liang et al., 2011; Majeed et al., 2011). The preservation of such fundamental brain circuits across species further supports the use of functional imaging, particularly resting-state functional MRI, as a translational tool informing on the pharmacodynamic effects on brain function. An investigation in which resting-state, functional connectivity changes elicited by a pharmacological challenge in a patient population (e.g., patients with neuropathic pain) and corresponding preclinical model (e.g., spinal nerve ligation model) would further inform on the use of functional imaging in drug discovery and development.
Another potential use of phMRI would be in gold standard compounds for which their proven analgesic efficacy is well documented in patients and in phMRI studies. For instance, morphine and buprenorphine have significant activity in cingulate cortex, basal ganglia, and other structures associated with analgesia. The lack of activity in such areas in a test compound could lead to further examination of the compound as an analgesic. phMRI use could be further complemented with measurements of pain stimuli modulation after the administration of an analgesic. The comparison of brain areas activated with pain, modulated with a drug, and detected with phMRI will likely help determine the mechanisms of pain modulation by the analgesic and potentially indicate possible targets.
Imaging techniques are being increasingly used as translational neuroscience tools, particularly in drug discovery and development efforts. Methods such as PET and SPECT are useful when characterizing a central nervous system disease state (eg, β-amyloid build up in patients with Alzheimer’s disease (Nordberg, 2004) or dopaminergic terminal function in patients with Parkinson’s disease (Sawle et al., 1993; Marek et al., 1996) and also as a means to determine whether a pharmacological compound has engaged its target [eg, buprenorphine at μ-opioid receptor (Zubieta et al., 2000) or aprepitant at NK-1 receptors (Keller et al., 2006)]. Proving the presence of specific target in a preclinical model or disease state and robust target engagement are extremely important to determine. However, such measures do not inform on the effects that a pharmacological treatment can have on brain function. Moreover, although end points such as target engagement are valuable and can be used to help make go/no-go decisions, proof of robust target engagement as measured by PET or SPECT does not necessarily correspond to clinical efficacy. Thus, a multidimensional dataset that includes a pharmacodynamic biomarker reflecting the functional consequences of target engagement may add value in providing objective data that inform on compound or dose selection for clinical trials. Moreover, a pharmacodynamic biomarker informing on brain function may also indicate how compounds or mechanisms of actions can be differentiated at the brain systems level. PhMRI is one method by which pharmacodynamic effects on brain function can be assessed. The results of this study further support the use of phMRI as a translational tool in neuroscience research, which can inform on pharmacodynamic properties of novel and marketed pharmacological compounds.
The authors thank Paul Serrano for his assistance with the manuscript.
Participated in research design: Becerra, Upadhyay, Bishop, Baumgartner, Schwarz, Coimbra, Maier, Iyengar, Evelhoch, Bleakman, Hargreaves, Borsook.
Conducted experiments: Becerra, Upadhyay, Chang, Bishop, Anderson, Nutile, George, Borsook.
Contributed new reagents or analytic tools: Becerra, Upadhyay, Baumgartner, Schwarz, Coimbra, Sunkaraneni, Maier, Wallin.
Performed data analysis: Becerra, Upadhyay, Chang, Bishop, Anderson, Wallin.
Wrote or contributed to the writing of the manuscript: Becerra, Upadhyay, Chang, Bishop, Anderson, Baumgartner, Schwarz, Coimbra, Wallin, Nutile, George, Maier, Sunkaraneni, Iyengar, Evelhoco, Bleakman, Hargreaves, Borsook.
- Received October 24, 2012.
- Accepted January 30, 2013.
L.B. and J.U. contributed equally to this work.
This work was supported by the Imaging Consortium for Drug Development (to D.B. and L.B.).
- explanatory variable
- magnetic resonance imaging
- positron emission tomography
- pharmacological magnetic resonance imaging
- single photon emission computed tomography
- Copyright © 2013 by The American Society for Pharmacology and Experimental Therapeutics