Isoflurane increases extracellular dopamine concentration and causes trafficking of the dopamine transporter (DAT) in transfected cells. Also, the binding potentials of highly specific positron-emitting DAT ligands are altered by isoflurane in rhesus monkeys. The purpose of this study was to determine the dose-response curve for isoflurane altering the binding potential of one of these ligands ([F-18]FECNT) in humans.
Twenty human volunteers underwent positron emission tomography using [F-18]FECNT. All subjects were scanned while awake and then again after assignment to one of four groups (n = 5 each): awake-control, propofol-control, or light or deep isoflurane anesthesia as defined by Bispectral Index monitoring. Bispectral Index values in the light anesthesia group were 40 +/- 7 (end-tidal isoflurane, 1.02 +/- 0.08) versus 27 +/- 10 (end-tidal isoflurane, 1.6 +/- 0.3) in the deep anesthesia group. The within-subject percent change in putamen binding potential between the awake and second scans was determined for each subject, averaged within groups, and compared across groups.
The [F-18]FECNT binding potential exhibited a biphasic shape as a function of anesthetic dose. The binding potential for the second scan in the awake-control and propofol-control groups was significantly less than the initial scan; for the light anesthesia group, the binding potential was significantly increased during anesthesia, and no change was detected between the two scans in the deeper anesthesia group.
Isoflurane causes a dose-dependent change in the [F-18]FECNT binding potential for DAT consistent with isoflurane causing trafficking of the DAT between the plasma membrane and the cell interior. Concentrations of isoflurane below minimum alveolar concentration causes DAT to be trafficked to the plasma membrane from the cell interior, but no net trafficking occurs at higher concentrations. The data are most easily explained if isoflurane alters the amount of functionally expressed DAT through an indirect pathway. This phenomena should be more fully explored to help make the next generation of anesthetics more mechanistically specific and to reduce undesired side effects.
THE dopamine system is implicated in the control of locomotion, cognition, and endocrine function.1–3Dopaminergic transmissions facilitate prefrontal cortex processes related to cognitive functions.4One model involving the prefrontal cortex suggests that dopamine promotes the likelihood of switching between alternative sources of information both within and between neural circuits, allowing for modifications in the temporal characteristics of an ongoing information sequence.5Suppression of dopaminergic inputs impairs performance on tasks that are mediated through this brain region.4Other investigations have shown that dopamine transmission plays a significant role in wakefulness6and that the wake-promoting effects of psychostimulates are mediated through dopaminergic neuronal transmission.7,8
Early investigations into the molecular effects of anesthetic drugs determined that volatile agents profoundly inhibit excitatory synaptic transmission9–11but do not alter the function of the voltage-gated channels involved in action potential generation.12Hence, considerable interest and effort has been invested in understanding the postsynaptic effects of anesthetic drugs. Evidence also suggests that presynaptic interactions are important targets for anesthetics. Clinically relevant concentrations of halothane and isoflurane have been shown to differentially alter the presynaptic cholinergic regulation of the release of inhibitory neurotransmitters in the striatum.13It has been demonstrated in vitro that volatile anesthetics inhibit dopamine uptake into rat brain synaptosomes14,15and that anesthetics enhance the effect of dopamine uptake inhibition on interstitial levels of striatal dopamine.16
The extracellular dopamine concentration increases during volatile anesthetic administration,16–20and the concentration of dopamine metabolites during isoflurane anesthesia increases as demonstrated by microdialysis in rats.21Because the presynaptic dopamine transporter (DAT) is a key regulator of synaptic dopamine concentration,22anesthetic effects on the DAT may mediate the increases in extracellular dopamine during anesthesia. Eckenhoff and Fagan18found in rat synaptosomes that isoflurane increases DAT density (Bmax) as measured with [H-3]2β-carbomethoxy-3β-(4-fluorophenyl)tropane ([H-3]CFT) and inhibits [H-3]CFT binding by increasing its affinity (kD). Data in figure 4of Tsukada et al. 23seem to show that isoflurane increases the [C-11]CFT binding potential (although this was not the purpose of their manuscript). Previously, we used in vivo and in vitro approaches in rhesus monkeys, rats, and stably transfected cells to investigate the interaction between isoflurane and DAT.24Isoflurane did not change the total amount of DAT protein, but increasing the concentration from 1% to 2% decreased the [F-18]2β-carbomethoxy-3β-(4-chlorophenyl)-8-(2- fluoroethyl)nortropane ([F-18]FECNT) binding potential.25Elfving et al. 26performed ex vivo studies in rats and showed that 1.9% isoflurane decreases the uptake of [I-125]PE2I (a DAT ligand).
The above-cited work is consistent in that isoflurane affects the DAT but seems to be contradictory in the degree and direction. In some studies it seems to increase the uptake of DAT-specific tracers, whereas in others, it inhibits the binding. The differences could be explained by the different tissues examined (transfected cells vs. in vivo studies and rats vs. nonhuman primates) or by different concentrations of isoflurane used. However, it may also be that the effect of isoflurane on the DAT is more complicated than a direct relation. Of course, alterations in dopamine release and DAT regulation would change the normal transmission of dopamine. Therefore, alteration in the regulatory controls of plasma membrane proteins caused by anesthetics is likely to play a role in the desired aspects of anesthesia or its undesired side effects. We hypothesize that anesthesia induces alterations in dopaminergic neurotransmission, which impairs cognitive abilities and suppresses arousal systems necessary for normal cognitive performance. Furthermore, we hypothesize that these impairments will linger after the anesthetic is complete and contribute to increased recovery time. This study was performed to determine the isoflurane dose–response curve for altering [F-18]FECNT binding to the DAT in human subjects. Each subject acted as his or her own control by being scanned while awake and while anesthetized. Two control groups were also studied: subjects who were not anesthetized, to determine the test–retest variability, and subjects who were induced with propofol but not anesthetized. The latter group was required because propofol was used as the induction agent before isoflurane administration.
Ideally, many different anesthetic levels would be tested in each research subject. However, because of radiation dose limits, only two positron emission tomography (PET) studies could be performed per subject, and using many subjects at several different anesthetic levels is prohibitively expensive. For these reasons, we chose to study two anesthetic levels and enroll 20 subjects for PET studies. The two anesthetic levels were chosen by considering the smallest dose with which the subject would remain motionless during the PET scan and the greatest dose that could be used without compromising the safety of the subject. For convenience of description only, these levels are termed light or lighter and deep or deeper (these labels are not intended to describe any clinical effects).
Materials and Methods
The Emory University Human Investigation Committee (Atlanta, Georgia) approved this project. Twenty subjects completed this study; all gave written informed consent for participation. Subjects between the ages of 21 and 45 yr were recruited from both sexes and all races by posted advertisement. All subjects were free of major medical illness on the basis of history, physical examination, and blood testing; were nonsmokers; were not actively abusing substances or alcohol; were free of all medications; and had no history of complications with anesthetics. Subjects were randomly assigned to one of four groups—awake-control, propofol-control, or lighter or deeper anesthesia—until the group contained five subjects. Among qualified candidates, a filled group was the only criteria that could exclude a subject from a particular group.
Subject monitoring included electrocardiogram, noninvasive blood pressure, pulse oximetry, end-tidal carbon dioxide, and electroencephalogram. The electroencephalographic signal was acquired using Zipprep® electrodes (Aspect Medical Systems Inc., Natick, MA; all impedances < 5 kΩ) applied to the forehead and temple using a frontal-temporal montage. The Bispectral Index was calculated and displayed in real time using an A-2000 electroencephalographic monitor (Aspect Medical Systems Inc.). One intravenous catheter was inserted for fluid administration and [F-18]FECNT injection. Participants fasted overnight and lay in the PET scanner supine with arms positioned at their sides. For the anesthesia condition, anesthesia was induced with a bolus injection of propofol (1–2 mg/kg). After placement of a laryngeal mask airway, the subjects were connected to an Ohmeda anesthesia machine (Madison, WI) via a semicircle breathing system and breathed 100% inspired oxygen. When the airway was secured, an air–oxygen mixture was used, and subjects were allowed to breathe spontaneously or with assistance to maintain end-tidal carbon dioxide within normal range. Isoflurane was delivered into the breathing system using a standard vaporizer. Expired concentrations were monitored on a Datex gas analyzer monitoring system (Ohmeda). The concentration of isoflurane was titrated to one of two levels of hypnotic depth as indicated on the Bispectral Index® monitor (35–45 during lighter and 20–35 during deeper anesthesia). At least 20 min was allowed to ensure that brain anesthetic partial pressure was constant before injection of [F-18]FECNT.
[F-18]FECNT (110 min half-life) was prepared by a two-step reaction sequence as previously described.27Briefly, alkylation of 1-[F-18]fluoro-2-tosyloxyethane with 2β-carbomethoxy-3β-(4-chlorophenyl)nortropane in dimethyl formamide at 135°C for 45 min affords [F-18]FECNT, which is purified by semi prep, reverse-phase high-performance liquid chromatography to produce a product free from the precursor, 2β-carbo-methoxy-3β-(4-chlorophenyl)nortropane with specific activity 56 MBq/nmol (1.5 Ci/μmol). [F-18]FECNT is at or near quasi-equilibrium 30 min after injection and is rapidly displaced by β-CIT.25The biodistribution, radiation dosimetry, and initial human studies have been reported.28,29
Human [F-18]FECNT studies were collected on a Siemens 921 (Hoffman Estates, IL) scanner. Two [F-18]FECNT PET brain scans (185 MBq ± 5% each) were acquired on each subject as depicted in figure 1. The first was performed while subjects were awake, and the second was performed according to which experimental group they belonged. Five subjects were scanned for each of the four groups depicted in table 1. The awake-control group was scanned while awake both times to determine the test–retest variability. Anesthesia was induced with propofol, which necessitated a second control group to differentiate any effects that might be caused by the propofol induction. The propofol-control group was induced but then allowed to wake up and were scanned again while awake. Subjects in the lighter and deeper anesthesia groups were scanned a second time while anesthetized. During all awake scans, the subject remained still, with the eyes open, and one of the investigators carried on a two-way conversation to ensure that the subject remained awake and alert. The conversation was kept nonemotional, with topics such as TV shows, sports, and current events.
After positioning the subject in the scanner, the head was immobilized using either a thermoplastic (Tru Scan, Annapolis, MD) facemask during awake scanning or tape if the subject was anesthetized. For the second scan, the subject was positioned in approximately the same position, but no additional efforts were made to match the positioning of the first scan exactly. Efforts were taken to ensure that the subject was stable and at equilibrium with the anesthetic, but otherwise, the time between scans was minimized. Two transmission scans were acquired with orbiting Ge-68 rod sources, one immediately before the awake scan and one immediately before the second scan. [F-18]FECNT, 185 MBq ± 5%, was injected as a slow bolus over 5 min with an infusion pump. Scanning started coincident with the start of injection. The acquisition sequence was 28 frames long, starting with 30-s scans and ending with a 15-min scan, for a total duration of 90 min. The exact sequence of number of frames, time per frame in seconds was 10, 30; 5, 60; 5, 180; 5, 600; 1, 900.
At the end of the first scanning sequence, the subject was removed from the scanner, allowed to walk to the bathroom, and then anesthetized (or not) according to their experimental group. Approximately 1 h after the end of the first scan, the subject was repositioned in the scanner, and a second transmission scan was collected with rod windowing to recalculate the attenuation correction because the head could not be repositioned exactly. PET images were then acquired with the same sequence as above except they included a 10-min scan immediately before the [F-18]FECNT injection to measure background activity remaining from the first injection.
Images from the 921 scanner were reconstructed with measured attenuation correction, zoom factor 8, and Shepp-Logan reconstruction filler cutoff at 1 cycle/cm. This produced images with an in-plane pixel size of 1.17 mm, 8 mm isotropic resolution, and axial slice thickness of 3.375 mm. At the early and late time points, the cerebellum and striatum were easily discerned in the PET images. Regions of interest were manually drawn over appropriate images then overlaid on all images to obtain the time–activity curves used in kinetic modeling. Separate regions of interest were drawn for the awake and experimental data sets.
Kinetic Modeling—DAT Availability
The PET studies used a within-subject experimental design that allows each volunteer to act as his or her own control. The parameter of interest was the percent change in [F-18]FECNT binding potential between the two PET scans. The reference tissue method30allows kinetic modeling to be performed without collecting arterial blood samples. Parameters that govern the uptake of radiotracer in the putamen, such as the binding potential, can be determined using data from a reference region devoid of specific uptake (cerebellum). The parameters are R , the putamen:cerebellum ratio of blood flow times extraction; k2, the rate at which activity reenters the blood from the brain; and the binding potential. Usually, there is no radioactivity in the subject before injection of the radiotracer. However, in this application, the analysis of data from the second scan must account for radioactivity remaining from the first injection. A generalization of the reference tissue method,31which yields binding potential estimates that are independent of initial radioactivity, was used without modification as has been done before.24,32
All results were expressed as the percent change in binding potential between the subjects’ first and second scans. Standard tests were performed to verify normalcy and equal variance of the data. Then, statistically significant differences between groups or between the control groups and no change were determined by unpaired t tests. A difference was considered significant at P < 0.05.
The propofol induction provided a smooth, rapid onset of anesthesia and allowed for ease of placement of the laryngeal mask airway device. The expired end-tidal isoflurane concentrations maintained unresponsiveness to verbal stimulation during lighter and deeper anesthesia. Table 2shows the changes in the physiologic variables produced by these levels of anesthesia and control conditions. No pharmacologic intervention was required for any of the subjects even though there was a significant decrease in mean arterial pressure during lighter and deeper anesthesia (mean ± SD: 23 ± 7 and 36 ± 12 cmHg less than the control groups, respectively). There was no statistical difference in the magnitude of the changes between the lighter and deeper anesthetic groups, but the reduction in mean arterial pressure in both groups was significantly different from both control groups (P < 0.05). Bispectral Index values decreased acutely on induction of anesthesia and returned to baseline within 10 min in the propofol-control group. The Bispectral Index values were maintained within the target range for the lighter (35–45) and deeper (20–35) anesthesia groups. The mean values differed significantly between the groups (40 ± 17 vs. 27 ± 10; P < 0.05). Correspondingly, the isoflurane end-tidal concentration differed significantly between the lighter (1 ± 0.08) and deeper (1.6 ± 0.3) anesthetic groups. The weight of the deeper anesthesia group was significantly greater than the other groups.
Typical time activity curves for the putamen and cerebellum for the four groups are shown in figure 2. Between 10% and 20% of the uptake measured in the putamen during the second scan is due to remaining activity from the first injection. Less than 2% of the activity in the cerebellum reference region remains from the first injection. When plotted on a logarithmic scale, the putamen and cerebellum curves are close to parallel. Typical washout rates at the end of the first scan in each set are 25–30% for the putamen and 25–35% for the cerebellum. When these washout rates are equal, the tracer is said to be in a state of quasi-equilibrium. These data indicate that the tracer is close to quasi-equilibrium by the end of scanning for most imaging sessions.
Included on the graphs in figure 1is a copy of the second scan putamen data shifted downward so that it starts at the origin (dashed line). This is a simplistic representation of the data from the second scan corrected for activity remaining from the first (it does not take into account that activity from the first injection continues to clear during the second study). In the awake-control scan, the dashed line is similar to the solid square line, indicating that approximately the same fraction of injected activity bound to the putamen after each injection. However, the cerebellar curve is higher for the second injection, indicating that more activity was available in the brain for binding. Because more activity was available but the same amount bound, the binding potential must have decreased. In the propofol-control case, the same amount of activity was available for binding (similar cerebellar curves), but less of the second injection was bound in the putamen, also indicating a decreased binding potential. This is in sharp contrast to the lighter anesthesia group, in which the fraction that bound to the putamen increased during the second scan, indicating increased binding potential. In the example from the deeper anesthetic group, both the cerebellum and putamen curves are slightly increased, indicating that the binding potential change, if any, was minimal. Results of the more rigorous calculations of the change in binding potential from the kinetic modeling for these examples are given in the figure caption.
The model parameters determined from the general reference tissue fits were varied to determine if a change in flow could account for the difference seen between the awake and anesthesia scans. Figure 3shows the putamen data from a subject in the lighter anesthesia group. The awake putamen curve was fit using the awake cerebellar curve as the reference region, and the binding potential was determined to be 6.07. When the flow-sensitive model parameters (R and k2) were allowed to vary in the fitting of the anesthesia data while holding the binding potential fixed at the awake value, the calculated curve for the anesthesia scan predicted that more activity should enter and wash out of the brain than what was seen (fig. 3). The model cannot suitably fit the lighter anesthesia data unless the binding potential is allowed to vary. When the putamen curve from the anesthesia scan was fit allowing all parameters to vary, the binding potential was 7.33, a 21% increase over the awake condition.
The right and left putamen time–activity curves were averaged to obtain a single curve that represented total putamen activity. The percent change in [F-18]FECNT binding potential in the putamen between scans 1 and 2 for each subject was calculated and averaged within the group. Group results are displayed in figure 4. The binding potential increased in the lighter anesthesia group. It was not significantly changed in the deeper anesthesia group but was significantly decreased in the awake- and propofol-control groups. Both isoflurane anesthesia groups were significantly different from the awake- and propofol-control groups. The awake- and propofol-control groups were not significantly different from each other.
This study demonstrates that isoflurane alters the binding potential of [F-18]FECNT in a dose-dependent manner. The binding potential is significantly greater during lighter anesthesia as compared with its paired awake scan. This is in contrast to the test–retest experiments (awake-control group) where the binding potential decreased during the second scan and the deeper anesthesia group, where the binding potential was unchanged between the awake and anesthesia scans. Results from this work are consistent with the literature. We previously showed that [F-18]FECNT binding potential is decreased in rhesus monkeys when the anesthetic is increased from 1% to 2% isoflurane.24Data from Tsukada et al. 23show an increase in [C-11]CFT (a similar DAT ligand for PET imaging) binding potential in rhesus monkeys between awake and 1% isoflurane. These studies taken together imply an inverse parabolic response of the DAT to isoflurane exposure.
The uptake of [F-18]FECNT in the putamen is much greater when lighter isoflurane anesthesia is administered. This type of behavior is always suspicious of a blood flow effect, and it is known that isoflurane increases flow via vasodilatation of the cerebral blood vessels.33An increase in the peak of the cerebellum curve and then faster washout can be seen during the lighter isoflurane condition (fig. 1), which is another indication of increased flow. However, [F-18]FECNT comes close to quasi-equilibrium, which argues against the measured changes in binding potential being due to a flow effect. Also, in our previous work, we manipulated an increase in blood flow but did not see an increase in binding potential.24Here, the expected tissue curve has been calculated assuming that the binding potential did not change and that any differences are due to a change in blood flow (fig. 3). The model cannot explain the anesthetic curves unless the binding potential changes in addition to flow. Finally, blood flow is expected to increase more during the deeper anesthetic condition than during the lighter anesthetic condition, but the binding potential is less during the deeper condition. For these reasons, the results of this study are unlikely to be caused by changes in cerebral blood flow induced by the anesthetics.
A potential confounder to the data is that the subjects were engaged in conversation during all awake scanning. From our experience, nearly all volunteer subjects fall asleep during long PET research studies if left alone. It is known that DAT plays an important role in sleep regulation and is necessary for the specific wake-promoting action of amphetamines and modafinil.8Because of this, permitting a subject to fall asleep was considered to be a greater confounder than applying a stimulus to assure that the subjects remained awake and alert. We chose to engage the subjects in nonemotional conversation as a way of maintaining the subjects in a more natural and constant baseline state. We were not able to find any literature indicating that conservation activates the putamen or cerebellum or affects the uptake of a DAT ligand. However, it remains a possibility that conversation could alter the amount of DAT expressed on the plasma membrane and hence the results of this study.
Other potential confounds are that the weight of the deeper isoflurane group was greater than the others and that the mean arterial pressure was reduced in the lighter and deeper groups when compared with control groups. The binding potential is the ratio of DAT-bound to free activity in the brain, which is not directly affected by patient weight. (No subjects were obese or had any weight-related medical problems.) Also, the weights of the control and lighter isoflurane groups were not significantly different, but the binding potential was significantly changed. Therefore, it is unlikely that the results seen are a result of different subject weights. Similarly, the mean arterial pressure was reduced in both of the isoflurane groups, but the binding potential increased in the lighter group and decreased in the deeper group, so it is unlikely that a change in arterial pressure is responsible for the results reported here.
[F-18]FECNT has been used to investigate the DAT under several different conditions. It is a cocaine analog25that competes with cocaine31and other DAT-specific ligands34,35for binding to the DAT. Therefore, similar to cocaine, it is likely that [F-18]FECNT binds to plasma membrane–expressed functional DAT. Hence, manipulations that change the [F-18]FECNT binding potential likely change the amount of functional DAT expressed on the plasma membrane. The change could be due to isoflurane decreasing the amount of DAT or to changing in its affinity for [F-18]FECNT. We have previously shown that isoflurane can induce DAT to be trafficked into HEK cells in vitro 36without changing the total amount of DAT protein.24Therefore, we hypothesize that isoflurane causes DAT to be trafficked between the plasma membrane and the cytoplasm and that the trafficking causes the DAT to change its affinity for [F-18]FECNT from high-affinity binding when on the plasma membrane to low affinity when in the cytoplasm.
It is doubtful that isoflurane has a direct interaction with the DAT because lighter isoflurane administration causes the [F-18]FECNT binding potential to increase. It is known that [F-18]FECNT has a high affinity for the DAT.25For a direct interaction to be responsible for the anesthetic effect, there must be a remarkable coincidence at play. It would imply that isoflurane directly binds to DAT molecules that are in a low-affinity state and changes them into a high-affinity confirmation that binds [F-18]FECNT with similar affinity as when isoflurane is not present. We think it is more likely that isoflurane activates an indirect mechanism that changes the amount of functional plasma membrane–expressed DAT. If this is true, then it is likely that other monoamine transporter systems respond analogous to the dopamine system during isoflurane exposure. This is consistent with 1.9% isoflurane causing an increase in [H-3]-(S )-citalopram (a serotonin transporter ligand) binding in ex vivo rat experiments.26
This work is consistent with the growing body of literature that implicates both presynaptic and postsynaptic sites of anesthetic action. It gives clear evidence that the presynaptic dopaminergic system is affected by isoflurane in a dose-dependent manner. The simplest mechanism that explains the data is that isoflurane affects machinery that is common to all monoamine transporter systems rather than proteins specific to certain systems. This could be tested by performing similar experiments with a serotonin transporter PET ligand. One likely candidate for the common machinery is protein kinase C.37
At this point, it is unclear whether altering the dopaminergic system is required for anesthesia or a side effect of isoflurane administration, nor is it clear to what extent dopaminergic alteration is responsible for the lingering effects of the anesthetic after it has cleared from the body. Further experiments are needed to determine whether dopamine transporter alteration is common across different anesthetics and to determine whether return to normal cognitive functioning correlates with the dopaminergic system returning to normal. If dopamine transporters are affected by a wide range of anesthetics, it is more likely that this is required for effective anesthesia. However, if only a few anesthetics affect dopamine transporters, then this property is unlikely to be necessary, and selecting against it could be used to make the next generation of anesthetics more mechanistically specific and shorten the recovery period.