|Home | About | Journals | Submit | Contact Us | Français|
Physiological changes (such as heart rate and respiration rate) associated with strong pharmacological stimuli could change the blood-oxygenation-level-dependent (BOLD) functional magnetic resonance imaging (fMRI) mapping signals, independent of neural activity.
This study investigates whether the physiological changes per se associated with systemic cocaine administration (1 mg/kg) contaminate the BOLD fMRI signals by measuring BOLD and cerebral blood flow (CBF) fMRI and estimating the cerebral metabolic rate of oxygen (CMRO2) changes.
BOLD and CBF fMRI was performed, and changes in CMRO2 were estimated using the BOLD biophysical model.
After systemic cocaine administration, blood pressure, heart rate, and respiration rate increased, fMRI signals remained elevated after physiological parameters had returned to baseline. Cocaine induced changes in the BOLD signal within regions of the reward pathway that were heterogeneous and ranged from −1.2 to 5.4%, and negative changes in BOLD were observed along the cortical surface. Changes in CBF and estimated CMRO2 were heterogeneous and positive throughout the brain, ranging from 14 to 150% and 10 to 55%, respectively.
This study demonstrates a valuable tool to investigate the physiological and biophysical basis of drug action on the central nervous system, offering the means to distinguish the physiological from neural sources of the BOLD fMRI signal.
Cocaine is a psychostimulant that is frequently abused and has reinforcing characteristics that can establish a state of addiction characterized by motivated, drug-seeking behavior and by a loss of the ability to control use (American Psychiatric Association 1994; Nestler et al. 2001). Cocaine mediated activation of the mesolimbic dopamine system has been observed using a variety of techniques (Marota et al. 2000; Porrino et al. 1988; Stein and Fuller 1992), and these and other studies have illustrated the complexity of the brain’s response to cocaine, whereby activation occurs in many brain regions beyond the limits of the dopaminergic reward pathway (Bardo 1998; Kelley 2004; Nestler 2002). Consequently, the successful development of therapies to treat cocaine addiction may require detailed information about the changes that cocaine induces in the activity of other large-scale neural networks within the brain (O’Brien 2003), and the acquisition of this information may be facilitated by experimental techniques that are able to observe changes in neural activity in vivo after systemic cocaine challenge.
Functional magnetic resonance imaging (fMRI) is a powerful and noninvasive technique to map brain functions ranging from perception to cognition at higher spatial and temporal resolution than alternative techniques, such as positron emission tomography (PET). Traditional fMRI experiments infer changes in neural activity from changes in surrogate signals such as blood-oxygen-level-dependent (BOLD) contrast, cerebral blood flow (CBF), and cerebral blood volume (CBV). While studies based on changes in BOLD contrast comprise the majority of published fMRI research, the relationship of BOLD contrast to neural activation entails a complex interaction between changing regional metabolic demands and compensatory responses by the cerebral vasculature. The interpretation of BOLD signals may not be straightforwardly associated with stimuli that cause significant physiological changes because these stimuli may cause fMRI signal changes that are independent of changes in neural activity.
Changes in BOLD contrast arise from changes in the venous concentration of deoxyhemoglobin and from changes in CBV (Boxerman et al. 1995; Ogawa et al. 1993). While fully oxygenated hemoglobin is diamagnetic, deoxygenated hemoglobin is paramagnetic, and its presence creates microscopic distortions in the magnetic field surrounding erythrocytes. The interaction of water molecules with these field distortions causes a change in the transverse relaxation rate of the blood that is dependent on deoxyhemoglobin concentration and that can be detected by MRI (Thulborn et al. 1982). Increased neural activity leads to a cascade of events that include increased in metabolic by-products and blood flow regulating factors (such as CO2 and NO) that lead to vasodilation, causing a regional increase in CBF. Interestingly, this increase in blood flow overcompensates for the increased oxygen metabolic demand, reducing the deoxyhemoglobin concentration which can be observed as increased BOLD signal.
The use of fMRI to investigate cocaine’s effects on neural circuitry in humans and in animal models have been reported (Breiter et al. 1997; Honey and Bullmore 2004; Luo et al. 2003; Mandeville et al. 2001; Stein 2001). Systemic administration of cocaine and other pharmacological agents can cause significant changes in cardiovascular physiology (heart rate, respiration rate, and blood pressure) (Pitts et al. 1987) and changes in vascular tone, potentially modulating the fMRI signals independent of changes in neuronal activity and complicating the interpretation of BOLD fMRI data (Gollub et al. 1998; Kaufman et al. 2001; Marota et al. 2000; Stein 2001).
In addition to conventional BOLD fMRI, changes in cerebral metabolic rate of oxygen (CMRO2) could also be estimated to serve as a metabolic marker for changes in neural activity (Hyder 2004). Stimulus-evoked CMRO2 changes can be estimated from BOLD and CBF measurements using a biophysical model of the BOLD signal and a calibration experiment (Davis et al. 1998; Hoge et al. 1999a; Kim et al. 1999; Ogawa et al. 1993). A brief and mild hypercapnic challenge (~5% CO2 over a few minutes) does not alter neural metabolism but does modulate CBF and BOLD contrast (Hafkenshiel and Friedland 1952; Kety and Schmidt 1948), revealing the regionally specific coupling of changes in CBF to changes in BOLD under isometabolic conditions. The proportionality constant M relating BOLD and CBF changes reflects the maximum BOLD response that can be expected from a given region. Combining the parameter M with measured changes in CBF and BOLD contrast after a functional stimulation, CMRO2 change can be calculated. This noninvasive technique has been used to map CMRO2 changes after sensory stimulations in both humans (Davis et al. 1998; Hoge et al. 1999a; Kim et al. 1999; Ogawa et al. 1993) and in animal models (Liu et al. 2004; Mandeville et al. 1999).
The goal of this study was to apply the CMRO2 fMRI technique to investigate the potential physiological effects of cocaine on the hemodynamic (CBF and BOLD) signals and to study the coupling among BOLD, CBF, and CMRO2 changes for brain regions involved in the reward pathway. CMRO2 fMRI requires multiple measurements that include BOLD, CBF, or CBV associated with the stimulus and associated with a hypercapnic calibration. This may not be feasible for experiments in which stimuli cannot be repeatedly administered. In this study, we overcame this limitation by using the two-coil arterial spin labeling technique to simultaneously measure BOLD and CBF (Duong et al. 2000a; Silva et al. 1999), making it possible to perform CMRO2 fMRI of a single cocaine administration in the same subject and the same setting.
Eleven cocaine naïve male Sprague–Dawley rats (300–375 g) were studied. The left femoral vein and artery were catheterized to administer cocaine and monitor physiology, respectively. Rats were anesthetized with 2.5% gaseous isoflurane (Phoenix Pharmaceutical, St. Joseph, MO) during surgery and immobilized in an MRI-compatible stereotaxic headset. The isoflurane level was reduced to 1.1% during imaging experiments, and rats breathed spontaneously without mechanical ventilation at all times. Mean arterial blood pressure (MABP), heart rate (HR), and respiration rate (RR) were measured continuously using a Biopak Systems data acquisition system (Santa Barbara, CA). HR and RR were derived from high and low frequency oscillations of the MABP signal, respectively. Before cocaine administration, hypercapnic (5% CO2) challenge experiments were performed for derivation of the calibration parameter (M) map (Davis et al. 1998). During cocaine challenge experiments, a 1.0-mg/kg cocaine dose (1.0 ml/kg injection volume) was administered intravenously via the femoral vein over a period of approximately 10 s.
Combined CBF and BOLD measurements were made on a 4.7 T/40 cm magnet (Oxford, UK), Bruker Biospec console (Bruker BioSpin, Billerica, MA), and a 20-G/cm gradient insert of 12 cm inner diameter (ID) capable of 120-μs rise time. An actively decoupled 2.3-cm ID surface coil was used for brain imaging and a neck coil for perfusion labeling as previously described (Duong et al. 2000b; Liu et al. 2004; Silva et al. 1999). Images were acquired using the continuous arterial spin-labeling technique with a single-shot, gradient echo, echo-planar-imaging sequence, and continuous arterial spin labeling was accomplished with a 1.7-s square radio frequency pulse to the labeling coil in the presence of 1.0-G/cm gradient along the direction of blood flow in the carotid arteries, satisfying the adiabatic inversion condition. The sign of the frequency offset was reversed during acquisition of non-labeled images. MRI parameters were: data matrix=64×64, field of view (FOV)=2.3 cm×2.3 cm, eight 1.5-mm thick slices, TE=15 ms, and TR=2 s. During the calibration experiments, 30 pairs of images (2 min) were acquired during baseline and 60 pairs during the hypercapnic challenge. During the cocaine challenge experiments, after a stabilization period of 15 min after the termination of the hypercapnic calibration experiment, 50 pairs of images (3.3 min) were acquired during baseline, and an additional 160 pairs were acquired after cocaine administration (11.7 min). High-resolution anatomical images (128×128, RARE) were acquired for co-registration of images with an atlas.
Data were analyzed using custom software developed in the Java programming language (Sun Microsystems). A BOLD image time series was derived from the non-labeled data set of CBF measurements, and a CBF image time series was generated by calculating CBF values pixel-by-pixel for each image (Silva et. al. 1999). BOLD and CBF time series data were analyzed pixel by pixel using a generalized linear model (GLM) approach (Worsley and Friston 1995) with a single-step box-car model for both the hypercapnic challenge and cocaine challenge data sets. P values corresponding to zero correlation were obtained for each pixel and used as thresholds. Percent change maps of BOLD and CBF for the hypercapnic challenge and cocaine challenge experiments were created using only pixels with P values below 0.01, with all other pixels set to 0 to reduce artifacts in maps calculated subsequently. The data illustrated in Fig. 2 were produced from one subject by normalizing each BOLD and CBF repetition with an average of 40 baseline repetitions, and ΔCMRO2 were calculated independently for each repetition.
A biophysical model for calculating relative CMRO2 changes and strategy for calibration using mild hypercapnia (Davis et al. 1998; Hoge et al. 1999a) were used to calculate M maps and CMRO2 percent change maps. M maps were calculated from the hypercapnic challenge BOLD and CBF percent change maps according to Eq. 1, where the subscripts H%chg and %chg designate percent change values obtained during the hypercapnic and cocaine challenge experiments, respectively. The exponents a and β have values 0.38 (Ueki et al. 1988) and 1.5 (Madsen et al. 1998), respectively, and pixels containing physiologically meaningless M values (imaginary values, negative values, and infinite values) were identified and excluded from subsequent map and ROI calculations.
CMRO2 percent change maps were calculated according to Eq. 2, where subscripts indicate percent change maps obtained during the cocaine challenge experiment. A single, group average M map (N=9) was created for use in Eq. 2.
Each slice of the maps and anatomical images from each subject were manually co-registered to an atlas (Paxinos and Watson 1998) using custom software and the following degrees of freedom: x and y translation, independent x and y scale, and rotation about the z axis. The original images were then resampled in the transformation to the atlas coordinate system, and a new set of images was created. Regions of interest (ROIs) were defined and used to measure average values from the ΔCBF, ΔBOLD, ΔCMRO2, and M maps of each subject. ROI analysis was performed on the periaqueductal gray (PAG), substantia nigra (SN), ventral tegmental area (VTA), hippocampus (HIP), primary somatosensory cortex (S1), ventral posterolateral thalamic nucleus (VPL), lateral globus pallidus (LGP), cingulate cortex (CG), nucleus accumbens (NAC), and striatum. Co-registered maps of all subjects were averaged together for visualization.
After cocaine administration, MABP increased from a baseline of 113±14 mmHg (mean±standard deviation) to 137±33 mmHg (P<0.05) during injection and remain elevated at 119±15 mmHg for 1 to 2 min after injection (P<0.05). Respiration rate increased from a baseline of 81±7 to 89±8 bpm (P<0.05) during injection and further increased to 103±16 bpm (P<0.05) 1 to 2 min after injection. Heart rate dipped slightly from a baseline of 405±49 to 394±39 bpm (P=0.13) during injection and increased slightly to 419±42 bpm (P<0.05) 1 to 2 min after injection. All parameters returned to baseline levels within 5 min after cocaine administration.
Figure 1 illustrates the changes in BOLD, CBF, and CMRO2 after systemic cocaine challenge. Maps were co-registered, averaged together, overlaid on anatomical images, and areas outside the brain were masked. The parameter M varied between 0.08 and 0.17 across the regions investigated, consistent with results from a previous study obtained under similar experimental conditions (Liu et al. 2004). M was appreciably higher in the vicinity of large draining veins, consistent with previous observations in humans (Davis et al. 1998) and in rats (Liu et al. 2004). The spatial distribution of the parameter M did not vary significantly between subjects, hence, an average M map was calculated for nine subjects (Fig. 1: panel M, Table 1) and the ΔCMRO2 maps for each animal were calculated using this average M map.
The BOLD signal increased in cortical, extrapyramidal, and midbrain regions after cocaine challenge (Fig. 1: panel ΔBOLD). Negative changes in BOLD after cocaine administration appeared in the striatum and CA1 and CA2 fields of the caudal hippocampus, and negative BOLD was observed on the cortical surface. In contrast, CBF consistently increased after cocaine administration across all studied regions of the brain (Fig. 1: panel ΔCBF). BOLD, CBF, and CMRO2 of different brain structures analyzed are summarized in Table 1. S1 and CG exhibited larger increases than midbrain nuclei, and only a moderate increase was observed in the hippocampus. Modest, but significant increases in CBF within the VTA and SN must be interpreted cautiously because of low signal to noise, susceptibility artifacts, and partial volume effects in these regions. Changes in CMRO2 after cocaine administration were positive in all regions of the brain studied, ranging from 17 to 38% (Table 1 and Fig. 1: panel ΔCMRO2). In cortical regions exhibiting decreases in BOLD, increases were observed in CBF and ΔCMRO2 suggesting that the negative BOLD signal change within these cortical areas is not the result of image or motion artifacts (Table 1: Cortical ROI). The time courses of BOLD, CBF, and CMRO2 for a representative subject (Fig. 2) illustrate the rapid and sustained increases in BOLD and CBF within S1, while increases in BOLD and CBF were considerably smaller within the NAc.
The parameter M is a proportionality constant that describes the changes in BOLD with changes in CBF. Iso-contour lines in Fig. 3a illustrate the expected changes in BOLD and CBF for given changes in CMRO2 within a region having an M value of 0.10 and illustrate that negative changes in BOLD may be observed in regions experiencing dramatically increased oxidative metabolism, potentially leading to erroneous inferences made from BOLD measurements alone. Data points in Fig. 3a indicate the observed changes in BOLD and CBF within regions with M values equal to 0.1 ± 0.02. CMRO2 increased within these and all other regions investigated. Changes in CBF and CMRO2 were linearly coupled by a ratio of approximately 2.8:1 (Pearson’s r=0.92) (Fig. 3b), and BOLD and CMRO2 were also roughly linearly correlated (r=0.79) (Fig. 3c). Two regions, the VTA and PAG, exhibited a negative and near-zero BOLD response but showed a significant increase in CBF, corresponding to an increase in CMRO2 (Fig. 3c). One region, the SN, deviated significantly from the correlative relationship observed between changes in BOLD, CBF, and CMRO2 exhibited by the other regions investigated, and the large increase in BOLD could be an artifact.
This study demonstrates BOLD, CBF, and CMRO2 functional MRI after intravenous cocaine administration in the anesthetized rat with concomitant physiological measurements (MABP, HR, and RR). Cocaine evoked significant and heterogeneous changes in BOLD, CBF, and CMRO2 within the mesocortical limbic pathway and throughout other regions of the brain. With the exception of negative BOLD changes in some cortical surface areas and within the VTA, BOLD, CBF, and CMRO2, changes were positive. Brain regions exhibiting increases in CMRO2 are consistent with regions identified as having increased metabolism of glucose in 2-deoxyglucose autoradiography studies reported previously (Porrino et al. 1988), and the changes in CBF that were observed are consistent with previously reported results.
BOLD fMRI of cocaine using similar doses has been previously reported in rats (Chen et al. 2001; Luo et al. 2003; Mandeville et al. 2001), although the magnitude and sign of the BOLD responses varied between labs. Negative BOLD near the surface of the cortex (but not in the subcortical structures) associated with cocaine challenge was consistently observed in this study. Such negative BOLD could be due to a marked cocaine-evoked increase in deoxyhemoglobin concentration in the large draining veins on the cortical surface (Krings et al. 1999) particularly at high field strength. Spin-echo BOLD experiments which yield contrast that is less sensitive to macrovascular contributions than the gradient-echo acquisitions used herein may provide insights into the origin of this negative BOLD signal, and may suppress macro-vascular BOLD changes overestimating CMRO2 change (Table 1).
While others have reported scattered negative BOLD responses throughout the brain (Luo et al. 2003), our previous forepaw stimulation studies under identical experimental conditions (Liu et al. 2004) did not show negative BOLD changes on the cortical surface. This suggests that the negative BOLD effect observed in this study is biological in origin and related to the effects of systemic cocaine challenge. Mandeville et al. 2001 did not observe significant negative BOLD changes at 2 T. These differences may be due to differences in experimental conditions (e.g., anesthetics and ventilation protocols) and possibly due to magnetic field dependent effects. Anesthesia, in particular, has been shown to impact the measured BOLD signal (Sicard et al. 2003). In addition, variations between labs may suggest that BOLD fMRI is more susceptible to nonneural physiological modulations and the precise experimental conditions of pharmacological studies, relative to sensory stimulation. For example, BOLD signal change is strongly dependent on baseline CBF and blood oxygenation (Sicard et al. 2003; Uludag et al. 2004). Vasoconstriction after cocaine administration (Kaufman et al. 2001; Luo et al. 2003) has been shown to cause negative changes in BOLD and CBF (Gollub et al. 1998); although this is difficult to reconcile with our and with others’ measurements of increases in CBF or CBV after a cocaine challenge, vasoconstriction is consistent with cocaine’s action as a sympathomimetic.
In contrast to the BOLD data, CBF increases after cocaine administration were reasonably consistent across different labs, despite differences in experimental conditions, anesthetics, and measurement techniques. Positive CBF increases were observed throughout the brain and were particularly pronounced in the neocortices. Our results are in good agreement with Stein and Fuller (1992) who reported CBF increases as measured by using [14C]iodoantipyrene in awake rats after intravenous cocaine administration. Our estimated CBV changes obtained from CBF and the Grubb et. al.’s relation (Grubb et al. 1974) in the cingulate gyrus (CG), nucleus accumbens (NAc), motor cortex (S1), dorsal thalamus (VPL), and striatum are linearly correlated (r=0.96) with the CBV changes obtained with MION contrast agent (Marota et al. 2000), although our overall CBV changes were slightly larger (data not shown).
Non-invasive CMRO2 fMRI offers a valuable surrogate measure of neural activation. Increased neural activity has been shown to be correlated with an increased rate of glucose metabolism (CMRGlu) (Kennedy et al. 1976; Sokoloff et al. 1977) and CBF (Jones et al. 2004), however, the magnitude of stimulus-evoked changes in CMRO2 remains controversial. Following Fox and Raichle’s initial study using PET (Fox et al. 1988), stimulus-evoked CMRO2 changes have been reported to be negligible (Fox et al. 1988; Madsen et al. 1998), smaller than the CBF and CMRGlu increases (Davis et al. 1998; Hoge et al. 1999b; Kim et al. 1999; Mandeville et al. 1999), or increased by 200–400% (Hyder et al. 1996). A recent study suggests that neural metabolism during stimulation is primarily oxidative (Kasischke et al. 2004). Although CMRO2 changes associated with cocaine administration have not been reported previously for comparison, the magnitudes of CMRO2 changes reported in this study were consistent with those reported using various sensory stimuli.
Future experiments using micro-PET imaging may provide a valuable basis for comparison of ΔCMRO2 in small animals after sensory stimulation or pharmacological challenge. Nonetheless, a few studies have demonstrated the consistency of MRI-based ΔCMRO2 over a wide range of experimental conditions (Sicard et al. 2003; Uludag et al. 2004). The mean ΔCMRO2 values are not skewed by noise (although the uncertainties increase) due to the self-correcting nature of the ΔCMRO2 formalism (Davis et al. 1998; Liu et al. 2004). Finally, the use of a two-coil configuration yielded a relatively high signal to noise ratio of the CBF signal, and the use of simultaneous measurements of BOLD and CBF avoided variations that could be present if the two measurements were made in a sequential manner. This approach makes it possible to perform BOLD and CBF measurements and estimate CMRO2 changes after cocaine administration in a single setting, within the same animal.
This study demonstrates for the first time CMRO2 fMRI of cocaine and to study the potential physiological perturbations caused by systemic cocaine challenge on the hemodynamic (BOLD and CBF) fMRI signals. Multimodal fMRI measurements provide complementary surrogate assays of neural activation that facilitate the identification of image artifacts and corroborate inferences of neural activity. The approach herein presents a valuable tool to study the physiological and biophysical basis of drug actions on the central nervous system and offers the means to discriminate the physiological from neural effects of drugs on the fMRI signals. This approach could have many other applications, such as fMRI of acute stroke and its dynamic evolution, where ischemia-induced physiological perturbation is expected to alter the cerebral hemodynamics independently of changes in neural activity.
This work was supported in part by research grants from the National Institute of Neurological Disorders and Stroke (NINDS, R01 NS045879), the American Heart Association (SDG 0430020N), and the National Institute on Drug Abuse (NIDA, R01DA13517). All experiments were performed in accordance with guidelines for the care and use of mammals in neuroscience and behavioral research (National Research Council 2003) and in compliance with the laws of the United States.
Karl F. Schmidt, Center for Comparative Neuroimaging, University of Massachusetts, Worcester, MA, USA.
Marcelo Febo, Center for Comparative Neuroimaging, University of Massachusetts, Worcester, MA, USA.
Qiang Shen, Yerkes Imaging Center, Emory University, 954 Gatewood Road, Atlanta, GA 30329, USA, Tel.: +1-404-727-9991, Fax: +1-404-712-9917.
Feng Luo, Center for Comparative Neuroimaging, University of Massachusetts, Worcester, MA, USA.
Kenneth M. Sicard, Center for Comparative Neuroimaging, University of Massachusetts, Worcester, MA, USA.
Craig F. Ferris, Center for Comparative Neuroimaging, University of Massachusetts, Worcester, MA, USA.
Elliot A. Stein, Neuroimaging Research Branch, National Institute on Drug Abuse, National Institutes of Health, Baltimore, MD, USA.
Timothy Q. Duong, Yerkes Imaging Center, Emory University, 954 Gatewood Road, Atlanta, GA 30329, USA, Tel.: +1-404-727-9991, Fax: +1-404-712-9917.