| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
;
||#
From the Departments of *Anesthesiology and Critical Care, Neurosurgery, Neurology, ||Radiology, and ¶Pathology and Laboratory Medicine, Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania; Neuroscience ICU, Duke University Medical Center, Durham, North Carolina; #The Center for Functional Neuroimaging, University of Pennsylvania, Philadelphia, Pennsylvania.
Address correspondence to W. Andrew Kofke, MD, MBA, FCCM, Department of Anesthesiology and Critical Care, University of Pennsylvania, 7 Dulles Building, 3400 Spruce St, Philadelphia, PA 19104-4283. Address e mail to kofkea{at}uphs.upenn.edu.
Abstract
BACKGROUND: Opioids have been linked to limbic system activation and, in animals, to neurotoxicity. Limbic system nonpharmacologic activation patterns have been linked to the Apolipoprotein E (ApoE) allelic distribution. We tested the hypothesis that, in the absence of surgery, small doses of remifentanil produce limbic system activation in humans which varies with dose and ApoE genotype.
METHODS: Twenty-seven ASA I–II volunteers received a remifentanil (UltivaTM) infusion at four sequentially increasing doses: 0, 0.05, 0.1, and 0.2 µg · kg–1 · min–1 while receiving 100% oxygen. Cerebral blood flow (CBF) was measured at each dose globally and in the amygdala, cingulate, hippocampus, insula, and thalamus regions by pulsed arterial spin labeling magnetic resonance imaging. ApoE single nucleotide polymorphisms were determined in each subject.
RESULTS: Significant dose-related CBF increases, without correction for Paco2, were detected in all areas. After normalizing for global CBF to correct for Paco2 effects, the remifentanil-mediated increased CBF in the cingulate persisted, with decreased flow occurring in the hippocampus and amygdala. All these Paco2-corrected effects were reversed in the presence of the ApoE4 polymorphism.
CONCLUSION: Remifentanil at sedative doses produces both activating and depressing effects in various limbic system structures. The cingulate cortex seems to have the most susceptibility to remifentanil activation, and ApoE4 seems to produce relative activation of the hippocampus and amygdala.
Postoperative cognitive dysfunction (POCD) is emerging as an important problem in patient safety (1–6). The pathogenesis of POCD remains elusive. Several reports suggest that opioids at moderate to large doses, have a unique propensity to induce limbic system activation in humans (7–9) and rodents (10–12), with congruent limbic system neurotoxicity having been demonstrated with large doses in rodents (7,10,13,14). In addition, any such effects may be affected by genotype, with the Apolipoprotein E4 (ApoE4) genotype, the gene associated with early-onset Alzheimer's disease (15), and intolerance to neural injury (16), notably being implicated in disparate responses of the limbic system to cognitive stimulation (17,18). One might speculate that there is a relationship between the genotype-based response to opioids and surgery and propensity to POCD. Given that perioperative opioid use is widespread, and that POCD is likely related pathogenetically to limbic system processes and genotype, it is possible that, at routine sedative-analgesic doses, opioid-induced limbic system activation occurs in humans in a genotype-dependent manner. Demonstrating these phenomena would be essential presupposing elements in generating a viable hypothesis that routinely-used perioperative opioids have the potential for genotype-dependent neuroexcitation-mediated limbic system neurotoxicity, and thus could contribute to POCD in humans.
The purposes of this investigation were twofold (1): to determine whether sedative-dose remifentanil, as representative of mu opioids, in the absence of surgery, increases metabolic rate (as indirectly reflected in blood flow) in limbic system structures in normal humans, and (2) whether such activation is linked to the presence of the ApoE4 genotype.
METHODS
This investigation was approved by the Biomedical IRB of the University of Pennsylvania. Twenty-seven healthy ASA physical status I-II volunteers between 20 and 30 yr of age and of both genders were recruited and provided written informed consent. Pregnancy was an exclusion criterion.
Each participant was NPO for 6 h. Intraarterial and IV catheters, 22G or 20G, were placed. The subject was positioned in the head coil of the magnetic resonance imaging (MRI) scanner and continuous physiologic monitoring was initiated. Each subject received 100% oxygen by Mapleson D circuit and tight-fitting mask and oxygen flow more than 10 L/min. Ear plugs were inserted and the eyes were covered. A normal saline infusion was begun at 30 mL/h, and ondansetron (2 mg) and glycopyrrolate (0.1 mg) were given IV.
Each study required 60 min to complete (Fig. 1). A baseline anatomical MRI was conducted for the first 8 min. At 8 min, the baseline perfusion MRI was performed, requiring 7 min to complete. At 15 min, the remifentanil infusion was started at the first dose of 0.05 µg · kg–1 · min–1. A second perfusion MRI scan was obtained at 23 min. At 30 and 45 min, the rate of the remifentanil infusion was increased to 0.10 and 0.20 µg · kg–1 · min–1, respectively, with associated perfusion MRIs performed at 38 and 53 min, respectively. An arterial blood gas was determined at the onset of each perfusion MRI, 7–8 min into each remifentanil dose level. The rate of the continuous remifentanil infusion as well as the normal saline was controlled by a volumetric pump on the outside of the radiofrequency (RF) shield of the MRI suite, beyond the magnetic field, with IV tubing being fed through a designated conduit in the wall. Remifentanil infusion tubing was connected to the maintenance IV tubing via a T-piece connecter placed on the IV catheter luer lock connection. The accuracy of the volumetric pump was verified before its use in the study.
|
MRI Methods
Using pulsed arterial spin labeling (PASL) perfusion functional MRI (fMRI), quantitative cerebral blood flow (CBF) data at each dose were obtained for 11 subjects (Group 1), and relative cerebral blood flow (relCBF) data at each dose were obtained for all 27 subjects (Group 2).
MRI Imaging Parameters and Analysis
Imaging Acquisition
MR scanning was carried out on a 1.5-T Siemens Sonata whole-body scanner, using the standard birdcage head coil. Before the functional scans, high-resolution T1-weighted anatomic images were obtained. The PASL sequence was a modified version of the flow-sensitive alternating inversion recovery "FAIR" technique (19), in which a saturation pulse was applied at TI1 = 800 ms after the global or slice-selective inversion (20). For optimal labeling, a hyperbolic secant (HS) inversion pulse was generated using the MATPULSE software (21), with 15.36 ms duration, 22 µT RF amplitude, and 95% tagging efficiency. A gradient of 0.7 mT/m was applied along with the HS pulse during tag, while the HS pulse was applied in the absence of gradient during control. The slab of the slice-selective inversion was 10 cm thick. The saturation pulse was applied to a 10-cm slab adjacent and inferior to the selective inversion slab in both label and control acquisitions. A delay time (w = 1.2 s) was inserted between the saturation and excitation pulses to minimize transit related effects in arterial spin-labeling images (22).
Twenty-seven subjects underwent the remifentanil dosing paradigm. For 11 of 27 subjects (Group 1), a "dummy" gradient and RF pulses preceded the scan to allow tissue to reach steady-state magnetization, and M0 image was acquired for the reconstruction of quantitative CBF. The other 16 subjects did not receive the dummy scan and M0 image was not acquired because of technical problems. Thus, Group 2 includes all 27 subjects using the relative perfusion signal (the direct difference from the subtraction of interleave control and labeled images) for further analysis (denoted as relCBF). The relative perfusion signal would not have affected the comparison of dose effect within subject, but may have compromised the power for intersubject analysis because of scanner variability unaccounted for by the M0 images. Each subject received four PASL scans, each with 120 acquisitions that took 6 min.
Imaging Data Analysis
fMRI processing and analysis were performed primarily with the Statistical Parametric Mapping software package (SPM99, Wellcome Department of Cognitive Neurology, UK, implemented in Matlab 5, Math Works, Natick, MA), with some modifications providing for perfusion analysis (http://cfn.upenn.edu/perfusion/software.htm).
For each subject, functional images were realigned, coregistered, and smoothed. Perfusion weighted image series were then generated by pairwise subtraction of the label and control images for all subjects. For the 11 subjects with M0 image, perfusion weighted image series was converted to absolute CBF image series based on a single compartment continuous arterial spin-labeling perfusion model (20).
The perfusion or CBF image series for each scan were averaged to produce single mean image associated with each condition (the baseline and three dose levels). Global quantitative CBF or perfusion signal intensities were extracted for each subject from the mean perfusion or CBF images. The mean images were normalized to a 2 x 2 x 2 mm3 Montreal Neurological Institute template in the standard Talaraich space and used for further region of interest (ROI) analysis. The ROIs were determined a priori to be the amygdala, cingulate, hippocampus, insula, and thalamus selected from an automated anatomical labeling ROI library (23) in the SPM Marsbar toolbox and illustrated in Figure 2. For all subjects (Group 2), the relative perfusion signals were extracted and dose-induced perfusion signal changes were calculated in each ROI. For each subject of Group 1, the quantitative CBFs in each ROI were calculated.
|
Evaluation of Genetic Polymorphism
The ApoE, 
2, 3, 4 (24), single nucleotide polymorphism was evaluated by the University of Pennsylvania Molecular Diagnosis and Genotyping Facility using polymerase chain reaction and restriction fragment length polymorphism, digesting with HhaI restriction enzyme. Forward primer: GCA CGG CTG TCC AAG GAG CTG CAG GC. Reverse primer: GGC GCT CGC GGA TGG CGC TGA G.
Statistical Analysis
Physiologic data for both groups underwent repeated measures ANOVA. CBF data underwent repeated measures ANOVA across the four doses for each ROI. Then, to internally account for CO2 effects, changes in regional CBF (rCBF) were normalized to global CBF by dividing the ROI by the global CBF at that dose, and were statistically evaluated in a similar manner. relCBF data were divided by the baseline signal to derive the fractional change in CBF for each remifentanil dose. Then the fractional change in each ROI was divided by the fractional change in global CBF to derive a CO2-independent normalized measure of CBF change in each ROI relative to global CBF. For Group 1, repeated measures ANOVA was performed on the CBF data, and where significance was indicated by P < 0.05, a paired t-test was done at each dose to evaluate for changes from the baseline dose. Many of the datasets in the Group 2 data were not normally distributed. In this group, a Friedman's test was done to detect whether there was any dose effect and then if one was detected at P < 0.05, a post hoc Wilcoxon's test was done comparing baseline data with each of the larger remifentanil doses.
To evaluate for an ApoE4 effect all of the data collected during remifentanil administration were pooled followed by a directed evaluation for this effect by ANOVA, comparing CBF in each ROI for subjects with and without the E4 genotype. SPSS (SPSS, Chicago, Il) and Analyze-it (Analyze-It Software, LTD, Leeds, UK) statistical packages were used.
RESULTS
Demographic characteristics of the enrolled subjects are shown in Table 1. Physiologic data for the two groups of subjects are shown in Tables 2 and 3. In both groups a statistically significant dose effect was observed for Paco2 and pHa.
|
|
|
|
|
relCBF Results: Group 2
These data express the relative change in CBF within one individual. Thus the data are expressed as fractional change in regional CBF compared with baseline with subsequent normalization for global CBF, similar to the procedures used for Group 1. The results for Group 2 are summarized in Figure 5. Significant dose effects, without correction for Paco2, were detected in all areas (Fig. 5A). Post hoc comparisons with baseline indicate increased CBF with larger doses of remifentanil in all areas. Normalizing the change in rCBF for global changes and thus eliminating Paco2 effects (Fig. 5B) indicates a decrease in ROI CBF relative to global in the amygdala and hippocampus, whereas an increase in this measure occurred in the cingulate.
|
Genotype Results
ApoE genotype distribution for all of the subjects (Group 2) is shown in Table 4. There was only one ApoE4 subject in the 11 subjects in Group 1, and thus a genotype analysis was done only for the entire set of subjects in Group 2. The analysis of pooled dosing data (baseline versus all three doses averaged together) for an effect of ApoE4 is shown in Figure 6. Subjects with the E4 genotype showed essentially no change in global CBF because of remifentanil compared with an approximately 20% increase observed in those without the E4 gene. These data include an effect of Paco2. Evaluation of the normalized data indicate that, compared with global CBF, the E4 gene was associated with higher flows in the amygdala and hippocampus but with a lower flow in the cingulate.
|
|
DISCUSSION
Using perfusion MRI, we observed a pattern of relatively reduced CBF in the amygdala and hippocampus and increased CBF in the cingulate with sedative doses of remifentanil. Notably, this pattern was reversed in subjects with the ApoE4 genotype. Moreover, our data confirm previous studies in humans (8,9,25) and rodents (10–12), demonstrating the potential for partial limbic system activation with small-dose mu opioids.
Methodological Issues
All of the subjects were given 100% oxygen as a safety measure. Hyperoxia produces cerebral vasoconstriction (26). This was probably present in our subjects and may have decreased their CBFs or altered their cerebrovascular responses to remifentanil.
One of the commonly used methods of accounting for Paco2 effects is to use literature data on the impact of changes in Paco2 on CBF. rCBF activation studies indicate that CO2 effects are neither predictable nor uniform throughout the brain (27). Thus, we chose to evaluate each ROI normalized to the global CBF [as suggested by Ramsay et al. (27)] at each remifentanil dose with its consequent change in Paco2. In doing so, any information regarding global CO2-independent effects of remifentanil was lost, and important ROI effects that were the same as global effects may have been missed.
Each subject received ondansetron and glycopyrrolate. Glycopyrrolate has been reported to not cross the blood–brain barrier (28), and ondansetron has been reported to have no effect on intracranial pressure (29). Both of these reports suggest that the cerebrovascular effects of these drugs are minimal, and not likely to have significantly affected the results of this study, although neither drug has undergone scrutiny for specific cerebrovascular effects.
We chose to use progressively increased infusions of remifentanil, but were constrained by a need to minimize time in the scanner for purposes of comfort and safety. These considerations led to the experimental design which entailed starting CBF data acquisition 8 min after a change in infusion dose. Computer modeling of a paradigm such as this by Minto et al. (30) indicates that by 8 min, when CBF measurements began, the pharmacodynamic effects as measured by electroencephalogram are about 75% of maximal with a near maximal effect present by 15 min after the step change in the infusion rate. Similarly, the UltivaTM (remifentanil) package insert (31) indicates development of a new steady-state concentration 5–10 min after a change in infusion rate. Thus, although it is likely that our subjects had not achieved completely maximal dose-related effects, it is reasonable to suggest that the pharmacodynamic effects were consistently near maximal during CBF determinations.
Relevance to POCD
Cognitive evaluations that would allow anything other than reasoned speculation on the role of opioids in POCD were not conducted. Nonetheless, the data do provide important pharmacological support for subsequent clinical studies specifically assessing the hypothesis that perioperative opioid use could contribute to POCD. The rationale for such a hypothesis follows.
The data indicate that sedative dose remifentanil in humans increases CBF, and probably metabolic rate in the cingulate, while decreasing it in the amygdala and hippocampus. Cingulate activation in humans with small doses of fentanyl and remifentanil was also reported by Firestone et al. (32) and Wagner et al. (9) respectively. Moreover, Kofke et al. (7) reported increased metabolic rate measured on positron tomography in the cingulate and temporal lobes in volunteers given large doses of remifentanil with neuromuscular blockade and mechanical ventilation. This suggests that our currently reported similar observations of increased flow in similar brain areas do, in fact, reflect metabolism. Data from animal studies seem to indicate a relationship between such metabolic activation and subsequent injury (7,10). The insula and cingulate have a role in memory and attention (33–35). Taken altogether, these observations support the notion that neuroactivation of such limbic areas with perioperative use of opioids might have a role in the genesis of POCD.
Although the ApoE4 genotype has not been conclusively related to POCD in noncardiac surgery (36,37), studies by Heyer et al. (38) and Kofke et al. (39) suggest that there is such a link in the context of perioperative cerebral ischemia. Assuming perioperative opioid use in their patients, it thus becomes reasonable to speculate that the differential neuroactivation we report may be tied to these observations.
Relevance to the Pathogenesis of Alzheimer's Disease
There is evidence that the ApoE4 allele predisposes to earlier onset Alzheimer's disease (15,16). Our ApoE4-specific observations may suggest a role for endogenous opiates in the pathogenesis of Alzheimer's disease. Notably, Reiman et al. (17) report an ApoE4-associated hypometabolism in the posterior cingulate and temporal lobes of non-demented young (17) and middle-aged (18) adults with concordance for areas destined to become severely hypometabolic with symptomatic Alzheimer's disease. Thus, the lack of activation in cingulate CBF with concurrently increased CBF in hippocampus and amygdala that we observed with remifentanil in ApoE4 subjects is of interest.
In summary, our data indicate that a sedative dose mu opioid can produce modest limbic system activation in humans. We have provided an essential presupposing element in generating a hypothesis for future studies that opioids have a potential for limbic system neurotoxicity in humans. Moreover, we demonstrate a potentially relevant differential activation pattern related to ApoE genotype status.
ACKNOWLEDGMENTS
The authors gratefully acknowledge Iris R. Karafin for assistance with preparation of the manuscript, Jia Guo, MD (Research Specialist, University of Pennsylvania) with arterial blood gas analysis, Patrick Neligan, MD (Assistant Professor, Department of Anesthesiology and Critical Care, University of Pennsylvania) for assistance with some of the studies, and Thomas F Floyd, MD (Assistant Professor, Department of Anesthesiology and Critical Care, University of Pennsylvania) for critiques of the APSF grant and ongoing advice and support throughout the project.
Footnotes
Accepted for publication March 15, 2007.
Supported by Anesthesia Patient Safety Foundation, NIH Training Grant for Dr Blissitt (5 T32 GM 07612-25), and Support for Imaging from NIH Grants NS045839 and RR002305.
Reprints will not be available from the author.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  F Skorpen, E. Laugsand, P Klepstad, and S Kaasa Variable response to opioid treatment: any genetic predictors within sight? Palliative Medicine, June 1, 2008; 22(4): 310 - 327. [Abstract] [PDF]
|
|
|
|
 |
|
 K. T. S. Pattinson, R. Rogers, S. D. Mayhew, B. J. MacIntosh, M. C. Lee, and R. G. Wise Remifentanil-Induced Cerebral Blood Flow Effects in Normal Humans: Dose and ApoE Genotype Anesth. Analg., January 1, 2008; 106(1): 347 - 347. [Full Text] [PDF]
|
|
|
|
|
|
R. Ramani Remifentanil-Induced Cerebral Blood Flow Effects Anesth. Analg., January 1, 2008; 106(1): 347 - 347. [Full Text] [PDF]
|
|
|
|
|
|
W. A. Kofke, J. A. Detre, J. Wang, and P. A. Blissitt Remifentanil-Induced Cerebral Blood Flow Effects Anesth. Analg., January 1, 2008; 106(1): 347 - 348. [Full Text] [PDF]
|
|
|
|
|
|
R. D. Sanders and M. Maze Translational Research: Addressing Problems Facing the Anesthesiologist Anesth. Analg., October 1, 2007; 105(4): 899 - 901. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
