| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
From the *P.A.I.N. Group, Brain Imaging Center; Department of Psychiatry, McLean Hospital; and Departments of Radiology and Neuroradiology, Massachusetts General Hospital – Athinoula Martinos Biomedical Imaging Center, and Harvard Medical School, Boston, Massachusetts.
Address correspondence and reprint requests to David Borsook, MD, PhD, P.A.I.N. Group, Brain Imaging Center, McLean Hospital, 115 Mill Street, Belmont, MA 02478. Address e-mail to dborsook{at}mclean.harvard.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Previous studies have reported on the effects of opioids on CNS activation (7–10); however, with the exception of the first of these (7), these studies evaluated the effects of painful and nonpainful stimuli after the drug infusion. One study used positron emission tomography (PET) and did not report on changes in subcortical regions aside from the caudate nuclei (7). We used fMRI to define CNS regions activated by unblinded acute administration of small-dose morphine to opioid-naïve individuals. We compared patterns of CNS activation with known circuitry for reward, analgesia, and sedation and with previous imaging studies of other drugs with specific effects on individual behaviors.
|
METHODS |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Experimental Design
Figure 1A shows the experimental design.
|
The experimental protocol consisted of two sessions, 1 wk apart. Infusions of morphine or saline were given in a randomized crossover design: 4 subjects first underwent a saline session followed by a morphine infusion session; the other 4 subjects had the order of the sessions reversed, yielding 8 morphine scans and 8 saline scans. To infuse the drug without producing significant respiratory depression, a total dose of 4 mg/70 kg was delivered over 4 injections (each 2 mL, containing 1 mg/70kg morphine) via an IV catheter placed in an antecubital vein in the right arm over 8 min. This dose is approximately 50% of clinically applied analgesic doses in healthy subjects but was specifically chosen to be overtly nonsedating and to minimize CO2 retention. Each injection took 20 s with a 100-s interval between injections. Respiratory rate was monitored for stability during the 4 infusions and for the duration of the session to insure that it did not decrease to less than 6 breaths/min. Subjects were informed which substance was being infused. Anatomical scans were acquired first (25 min), then subjects were instructed to close their eyes and remain still and that the infusion scan of saline or morphine was going to begin. After 5 min of scanning, drug infusion was started.
Further physiological measures were in place for the subject’s safety. These included heart rate (HR), oxygen saturation, and end-tidal CO2 (ETco2) measures (In-Vivo OmniTrak 3100, Orlando FL).
Subjects were informed that they might experience effects including, but not limited to, nausea, emesis, drowsiness, and a euphorigenic "high" or dysphoric "low". After each infusion, subjects rated their highs or lows on a visual analog scale (VAS) of 0 to 10, with 10 being the maximal high. At the end of the scanning session, they were also asked to report any other effects of the infusion such as nausea or drowsiness.
Brush (25 s off, 30 s on x 4 repeats) and thermal pain (46°C, 25 s off, 30 s on x 4 repeats) stimuli were administered after the morphine or saline infusions (11). The effects of morphine on pain activations are the subject of a separate report; however, pain ratings (Likert numeric scale: 0-10, where 0 = no pain and 10 = maximal pain) are provided in the results.
Scanning was performed with a quadrature head coil and a 1.5 T magnetic resonance scanner (General Electric, Fairfield, CT) modified for echo-planar imaging (Advanced NMR). Imaging followed a standard protocol (11). Reduction of movement in MRI images of deep brain structures including the brainstem was achieved through the use of cardiac-gated acquisition of images.
Scans were acquired using an asymmetric spin echo T2*-weighted sequence in a clustered volume acquisition with cardiac gating (TE = 70 ms, 180° refocusing pulse offset by –25 ms; FOV = 40 x 20 cm; in-plane resolution = 3.125 mm; through-plane resolution = 7 mm; 20 slices oriented perpendicular to the AC-PC axis). The infusion scan lasted approximately 25 min including the 5-min baseline scan and the 4 injections (Fig. 1A). For the infusion scan 250 volumes were acquired with a repetition time (TR) of 6 cardiac pulses. For the sensory scans, 100 volumes were acquired with a repetition time of 4 cardiac pulses. Gating time intervals were recorded to allow T1-correction of image intensities.
Statistical Analysis and Image Processing
Statistical analysis of physiological measures (Super ANOVA software, SAS Institute Inc., Cary, NC) involved a two-way analysis of variance with drug treatment (saline, morphine) and time of measurement as factors, followed by post hoc Student’s t-tests. The corrected threshold for significance was P < 0.01. The integer output of subjective ratings was segregated by drug treatment (saline, morphine) and statistically analyzed by Student’s t-test.
Functional data were corrected for T1-effects resulting from cardiac gating acquisition. Data preprocessing and statistical analysis were performed using FSL 3.1 software (http://www.fmrib.ox.ac.uk/fsl). Functional data were smoothed with a Gaussian filter of 6 mm. Statistical maps were generated using a generalized linear model approach with 3 possible responses (Fig. 1B): a constant change after infusion of the drug (Step); a gradual increase in activation (Ramp); and a rapid onset of activation followed by a gradual decay (Decay). Morphine penetrates into brain tissue rapidly, with a half-life of 30 min (12); the response models used in the analysis reflect the most likely temporal responses in the CNS. Individual statistical maps were registered to the corresponding high-resolution anatomical scan and transformed into the standard Montreal’s Neurologic Institute brain template using FLIRT (fMRI Linear Image Registration Tool, part of FSL). Group analysis within and between groups was conducted using FLAME (high-level analysis tool, part of FSL), a mixed-effects based analysis. Regions of activations were identified on anatomical images using cluster analysis with a threshold of P = 0.01. Time courses were extracted using Matlab (Mathworks, Natick, MA). Coordinates of the most significant activated voxel in the cluster were transformed into Talairach atlas. Volumes of activation were determined from the overlap of the activated cluster map with the anatomical structure.
Morphine alters cerebral blood flow (CBF) by its actions on brainstem respiratory control modulating CO2 levels (12). To determine the potential global perfusion changes resulting from the drug, ETco2 and HR readings were analyzed.
|
RESULTS |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
HR and ETco2 measures before and after morphine administration for each group were not statistically different (analysis of variance; P > 0.05) (Fig. 2C), although ETco2 showed a trend toward an increase.
|
Small-dose opioids, including morphine, do not affect mechanical sensation after non-noxious brush stimuli in healthy subjects, despite the potential analgesic effects of the opioid infusion. To confirm that regionally specific brain activation could be discerned against a potential increase in global CBF or analgesic effects, BOLD signal during a control brush stimulus was compared after both morphine and saline infusions. A change in brain activity after brush sensation in the presence of morphine would suggest the presence of significant effects of morphine on global blood flow or glucose metabolism. No differences between VAS scores for brush for the two infusions were observed (data not shown). As expected, during brush stimulation, positive signal change was observed in the same location within S1 after both morphine and saline infusions. There was no significant difference in activation in S1 cortex by brush after morphine or saline infusions (Fig. 1D; P > 0.001), suggesting that changes in global CBF do not contribute significantly to the differences between CNS responses to morphine and saline infusions.
The average rating of "high" after morphine was 3.4 ± 1.14 (mean ± sem), significantly different from an average rating of 0.3 ± 0.45 after saline (t-test, P < 0.002).
At the end of the scanning sessions, none of the subjects reported nausea, dysphoria, or drowsiness to the infusions. Subjective ratings for euphoria (after the infusion) and pain ratings (after noxious heat stimulation) were recorded. The VAS ratings were 7.6 ± 0.43 for saline and 6.6 ± 0.6 for morphine; the difference between saline and morphine ratings was significant (P < 0.05), indicative of a mild analgesic response to the drug.
fMRI revealed a number of regions that were significantly activated by morphine compared with saline in cortical and subcortical locations (Table 1 and Figs. 2 and 3). No structure achieved significant activation following the decay model. Most activated structures displayed either a ramp response or a step response as detailed below.
|
|
Morphine induced significant increases in BOLD signal in the nucleus accumbens (NAc), putamen (Pu), sublenticular extension of the amygdala (SLEA), hypothalamus (Hy), and substantia nigra (SN) (Fig. 3A) and decreases in signal in the thalamus (Th) and periaqueductal gray (PAG)/ventral tegmentum (VT) (Fig. 3B). Corresponding time courses show gradual changes in signal beginning with drug infusion and continuing over the course of the scan in the NAc, SLEA, SN, Pu, Th, and PAG/VT. The increase in signal in the Hy occurs with a ramp time course similar to that seen in the GOb. Activation within the striatum was clearly delineated and showed localization to the Pu, with no signal change in the caudate or globus pallidus (Fig. 3A).
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
In the current study, fMRI revealed that small-dose morphine produced changes in BOLD in several regions of the CNS. The regionally specific brain activity resulting from morphine reflected its predominant effect on µ-opioid receptors (12). It is important to note that we saw no significant signal change in at least one area of high µ-receptor density, the caudate nucleus. This suggests that morphine-induced brain activity in this cohort is not merely a reflection of µ-receptor distribution.
After small-dose morphine infusions, drug-naïve healthy volunteers experienced mild, but statistically significant, euphoria. Furthermore, significant positive signal change was observed in the NAc, SLEA, and GOb, regions classically associated with reward function (15). Subjects used in this study were drug-naïve and, therefore, neither in a drug-deficit state nor experiencing any opioid withdrawal. Thus, the activation reported here may represent the initial response of CNS reward systems to a drug of potential addiction. Although morphine induced BOLD changes similar to those seen with other drugs of abuse including cocaine (16) and methamphetamine (17) (Table 2A, Fig. 4), some cortical areas showed opposite responses to morphine and cocaine. For instance, we observed decreased signal in the Th and frontoparietal cortex after morphine infusion, whereas cocaine activated these regions. Recent work suggests that negative BOLD signal changes be interpreted as a general diminishment in neural activity (18). BOLD changes in this region may distinguish opioid depressant effects from the stimulant effects of drugs such as cocaine. However, because the cocaine studies were done in addicts, whereas our current study subjects were drug-naïve, we cannot exclude that the differences we saw were related to prior exposure.
|
|
Morphine produced an increased signal in the NAc, as seen with cocaine (16) and other rewarding stimuli (19). Previous studies have shown that aversive stimuli trigger decreases in activity in the NAc (11). These data are consistent with the hypothesis that the NAc may be involved in discriminating whether a motivationally salient stimulus is rewarding or aversive.
In addition to activation in the NAc, we observed an increase in activation in GOb and a decrease in activation in the DLPFC (Fig. 2). The reason for decreased BOLD signal in the DLPFC (Fig. 2) is unknown, but similar decreases have been reported in other cortical areas after other rewarding stimuli (20). We also observed significant bilateral activation in the SN and Pu but not in the caudate or globus pallidus (Fig. 3). Although not part of the classic reward circuitry, both the Pu and SN have been reported to be involved in reward (19).
Morphine is also a potent sedative hypnotic (12). After morphine infusion, we observed decreased BOLD signal in the brainstem, medial thalamic nuclei and in cortical regions involved in sedation including prefrontal and parietal regions (21). Diminished neural activity in the brainstem and medial thalamic nuclei is consistent with the clinical literature showing progressive diminishment in sensory, motor, and cognitive function to the point of unconsciousness with increasing morphine dosage (12).
The decreases in signal in these regions after small-dose morphine parallel observations of decreased glucose metabolism in some of these brain regions after much larger morphine doses in polysubstance abusers (22). Our data are also consistent with PET data for other sedative-hypnotics with different mechanisms, including propofol (21), and the benzodiazepine midazolam (23) (Fig. 4 and Table 2B). These studies reported decreases in relative CBF in the Th, parietal, prefrontal cortex, and TL (23). The common effect of morphine and other sedating drugs on a specific subset of cortical regions suggests that the BOLD changes we report in these regions reflect morphine’s sedative effects. In contrast to these cortical inhibitory effects, positive signal change in S1 somatosensory cortex to brush after morphine and saline infusions were not statistically different, indicating that regionally specific brain activation could be observed after morphine infusion.
Small-dose morphine produced a small but significant analgesic response. We observed activation in a number of structures that are involved in opioid-related endogenous analgesia, including the PAG, the Hy, the aCG, and reward-related regions such as the NAc, Hi, or amygdala (24) BOLD signal changes in these areas after morphine infusion were largely opposite to those reported by our group for noxious stimuli or infusion of the opioid antagonist naloxone (Table 2C) (1). In addition, the observed decreased activation in the aCG was opposite to that previously reported for fentanyl (7); these differences may relate to differences in experimental design (gender differences, drug kinetics, and dose equivalence). The decreased signal in the PAG after small-dose morphine is similar to the late response observed in the PAG during a noxious thermal stimulus (11), suggesting that morphine and noxious stimuli activate endogenous inhibitory systems within the PAG in the same manner.
Expectancy and placebo-like effects are possible (25), as this study was unblinded. In general, placebo effects refer to expectancy that modulates the experience of the outcome. Expectancy effects on outcomes depend on a learned response and this effect is observable in human brain activity (25). In the current study, subjects were carefully screened to exclude prior morphine exposure, thus they had a minimal basis for learned responses to morphine. Furthermore, they were informed that morphine may produce aversive (e.g., nausea, dysphoria) or nonaversive (e.g., euphoria, high) symptoms. Accordingly, it would be expected to observe a distribution of hedonic ratings among subjects. However, all subjects only reported euphoric experiences. The placebo effect may account for observed changes in reward circuitry (i.e., hedonic high); however, activation in sedative and endogenous analgesic circuitries are less likely (but may occur) to be an expectancy effect (viz. no reports of drowsiness and no expectancy of pain stimulation). Although the contribution of expectancy to the morphine effects cannot be excluded, the regions of activation implicate circuitry that has behavioral effects with morphine (i.e., sedation, addiction, endogenous analgesia).
Morphine is generally viewed as an inhibitory drug. Hence, positive fMRI activation in some regions in response to it may appear counterintuitive. To interpret morphine effects one needs to consider recent work reporting that fMRI BOLD signal reflects local field potentials determined by input to and integrative activity in neurons, as opposed to action potentials out of these cells (18). The release of inhibitory activity on dopaminergic systems, along with direct opiate effects, will alter the input to, and integrative processes of, afferent cells such as those in the NAc, SLEA, and paralimbic regions, resulting in an excitatory process overwhelming an inhibitory process.
fMRI revealed that small-dose morphine produces changes in BOLD in several regions of the CNS. Activation in reward circuitry was similar to that seen after drugs of abuse and other rewarding stimuli. In somatosensory cortex and NAc, morphine induced decreases in BOLD signal, which is opposite to what is seen after a painful stimulus, suggesting the direction of signal change in these areas may be a CNS marker for pain and analgesia. The circuit-based approach to drug evaluation described here indicates that functional imaging may have significant applications in drug development.
|
ACKNOWLEDGMENTS |
|---|
|
Footnotes |
|---|
Supported, in part, by grants to D. B. (012581 & 13650) from the National Institute of Drug Abuse (NIDA), Bethesda, Maryland.
|
REFERENCES |
|---|
|
TopAbstractIntroductionMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  D. D. Langleben, K. Ruparel, I. Elman, S. Busch-Winokur, R. Pratiwadi, J. Loughead, C. P. O'Brien, and A. R. Childress Acute Effect of Methadone Maintenance Dose on Brain fMRI Response to Heroin-Related Cues Am J Psychiatry, March 1, 2008; 165(3): 390 - 394. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
