Acute exercise modulates cigarette cravings and brain activation in response to smoking-related images: an fMRI study
- 785 Downloads
Substances of misuse (such as nicotine) are associated with increases in activation within the mesocorticolimbic brain system, a system thought to mediate the rewarding effects of drugs of abuse. Pharmacological treatments have been designed to reduce cigarette cravings during temporary abstinence. Exercise has been found to be an effective tool for controlling cigarette cravings.
The objective of this study is to assess the effect of exercise on regional brain activation in response to smoking-related images during temporary nicotine abstinence.
In a randomized crossover design, regular smokers (n = 10) undertook an exercise (10 min moderate-intensity stationary cycling) and control (passive seating for same duration) session, following 15 h of nicotine abstinence. Following treatments, participants entered a functional Magnetic Resonance Imaging (fMRI) scanner. Subjects viewed a random series of smoking and neutral images for 3 s, with an average inter-stimulus-interval (ISI) of 10 s. Self-reported cravings were assessed at baseline, mid-, and post-treatments.
A significant interaction effect (time by group) was found, with self-reported cravings lower during and following exercise. During control scanning, significant activation was recorded in areas associated with reward (caudate nucleus), motivation (orbitofrontal cortex) and visuo-spatial attention (parietal lobe, parahippocampal, and fusiform gyrus). Post-exercise scanning showed hypo-activation in these areas with a concomitant shift of activation towards areas identified in the ‘brain default mode’ (Broadmanns Area 10).
The study confirms previous evidence that a single session of exercise can reduce cigarette cravings, and for the first time provides evidence of a shift in regional activation in response to smoking cues.
KeywordsExercise Smoking Cravings Cue-induced cravings Mechanism
Exercise appears to be an effective non-pharmacological method of reducing cigarette cravings and desire to smoke during temporary abstinence from smoking (for a review see Taylor et al. 2007; Ussher et al. 2008). Even short moderate-intensity bouts reduce absolute cravings, responses to cigarette cues, and increase time between ad libitum smoking (Taylor and Katomeri 2007). However, little is known about the neurobiological mechanisms involved. Evidence for such mechanisms could lead to further work to establish the optimal exercise dose (intensity and duration) for relapse prevention in smoking cessation treatment, and also open up new avenues for research into the treatment of other addictions.
During nicotine abstinence, the presence of cigarette images has been shown by fMRI to increase activation in both the mesolimbic (nucleus accumbens, amygdala, and hippocampus) and the mesocortical (prefrontal cortex, orbitofrontal cortex and anterior cingulate) dopamine circuits (Due et al. 2002; David et al. 2005; McClernon et al. 2005; Smolka et al. 2006), in anticipation of both the reinforcing effects and the incentive salience of the drug. The mesocorticolimbic dopamine system is thought to modulate activity in a number of higher brain regions/circuits. Evidence from animal models suggests that there are at least four neurological circuits implicated in nicotine addiction: (1) a reward circuit (including the ventral striatum), (2) a motivational/drive circuit (including the orbitofrontal cortex), (3) a learning and memory circuit (including the hippocampus and amygdala), and (4) a control circuit (including the prefrontal cortex and the anterior cingulate cortex) (Volkow et al. 2003).
Human fMRI studies, involving the visual presentation of non-smoking and smoking cues (e.g., images of people smoking cigarettes), suggest that there are differences in regional brain activation across a widespread network. These include the medial, lateral, and ventral prefrontal regions, as well as areas involved in visual attention (such as the parietal cortex and fusiform gyrus; Due et al. 2002; David et al. 2005; Lee et al. 2005; Lim et al. 2005; McClernon et al. 2005; Smolka et al. 2006). Specific to nicotine, the ventral striatum (including regions of the caudate, striatum, and nucleus accumbens) may be of particular interest as activation in this region signals the presence of nicotine-related environmental stimuli as well as the hedonic effects of nicotine administration (Balfour 2001; Brody 2006; Koob and LeMoal 2001).
Evidence from animal research suggests that exercise may influence neurological processes that may be implicated in the moderation of drug use and responses to contextual cues. More specifically, exercise has been shown to increase dopamine release in the striatum of rats following treadmill running (e.g., Meeusen et al. 2001; Hattori et al. 1994) and normalize dopamine levels in spontaneously hypersensitive mice (a condition characterized by abnormally low levels of dopamine in the neostriatum and nucleus accumbens; Sutoo and Akiyama 2003). In other studies, exercise reduced self-administered addictive substances, such as amphetamine (Kanarek et al. 1995), cocaine (Cosgrove et al. 2002), and ethanol (McMillam et al. 1995), and there is emerging evidence to suggest that changing the housing status of a non-human primate to allow more locomotive activity results in a reduction of self-administered cocaine and makes the animal less vulnerable to self-administer (e.g., Morgan et al. 2002). Exercise has also been shown to be both reinforcing (rats that run can be trained to lever press for access to running wheels; Iversen 1993) and is linked with conditioned place preference (Lett et al. 2000). Exercise-induced alterations in regional brain activation have been identified in animal models but have yet to be evaluated in humans (Wang et al. 2000) and may be the mechanism by which exercise exerts its effects.
The primary aim of this study was to assess the effects of a single session of exercise (versus a passive control condition) on regional brain activation while viewing smoking-related stimuli in an event-related fMRI study. A secondary aim was to confirm previous findings that an acute bout of exercise will reduce self-reported cigarette cravings.
Materials and methods
Ten smokers were recruited through public poster advertisements and individually screened for suitability. Participants were eligible to take part if they were between the ages of 18–50, smoked at least ten cigarettes a day, had been a regular smoker for 2 or more years, and were not currently making an attempt at smoking cessation. Participants were required to be free from injury or illness that would inhibit their ability to safely exercise at a moderate intensity.
Task and procedure
The control condition involved sitting passively in the laboratory without access to reading materials, mobile phone, or internet for 10 min. Such a brief period of passive sitting has previously been shown to generate stable measures of self-report cigarette cravings (Taylor et al. 2007).
Following a brief familiarization period, participants exercised on a Monarch cycle ergometer, involving a 2-min warm up, then 10 min at a subjective Rating of Perceived Exertion (RPE; Borg 1998) of 11–13 (i.e., fairly light–moderately hard subjective level of exercise intensity/effort). The use of RPE scores as a means of assessing exercise effort are widely accepted as they have been found to have a significant correlation with physiological measures, such as heart rate (Chen et al. 2002). A POLAR heart rate monitor (only visible to the investigator), worn throughout the session, allowed relative exercise intensity to be calculated.
Upon completion of the passive and exercise treatments, participants were required to enter the MRI scanner. Scanning lasted approximately 15 min on both occasions.
fMRI data acquisition
Scanning was performed on a 1.5-T Philips Gyroscan magnet at the Peninsula MRI research center, University of Exeter, UK. A T2*-weighted echoplanar imaging (EPI) sequence was used (TR = 3000 ms, TE = 50 ms, flip angle = 90°, 32 oblique transverse slices in ascending order and matrix size = 3.6 × 3.6 × 4 mm). Two hundred fifty-one volumes were acquired in each of the two runs per subject.
The pictorial stimuli were derived from the International Smoking Images Series (ISIS; Gilbert and Rabinovitch 1998) and consisted of 30 smoking-related (images of hands holding cigarettes, pictures of cigarettes, and people smoking) and 30 neutral images (hands holding pens, items of stationary, and people without cigarettes; see Fig. 1 for examples). Both sets of stimuli were matched for body parts, color contrasts, and size. The smoking-related images have been used in previous studies involving fMRI analysis in smokers compared with non-smokers (David et al. 2005) and abstinent versus satiated smokers (McClernon et al. 2005), using a similar paradigm with no exercise treatment.
The 60 images were randomly presented in each scanning session and for each participant using E-prime software (Psychological Software Tools, version 1.1). Images were viewed on a screen placed at the foot of the scanner via a mirror mounted on the head coil. Each image was presented for 3 s. A button, placed in each of the participant’s hands, was pressed upon presentation of the image to ensure attentional focus. The button-press for smoking or neutral images was randomized for hand dominance between participants. After each image, a white screen with a black fixation cross was presented for a randomly determined period of 8, 10, or 12 s, and participants were asked to view this between smoking images to remain focused. The duration of image presentation and the ISI chosen fall between those reported in other event-related studies in this area. David et al. (2005) presented 75 stimuli for 5 s at a frequency of one image each 6 s with a one-second rest period/fixation cross. While McClernon et al. (2005) presented cues for 4 s with a variable inter-stimulus interval fixation of 18 to 22 s, having conducted a thorough review of the literature, we felt our timings suitable to allow the anticipated BOLD response to peak (approximately 4–6 s) and begin to return to baseline levels before the next image presentation.
Subjective measures of cravings
Desire to smoke was assessed using a seven-point scale (1—strongly disagree, 4—neither agree nor disagree, 7—strongly agree) for the statement ‘I have a desire to smoke’ (Tiffany and Drobes 1991). This single item has been shown to be sensitive to the effects of exercise in numerous studies (Taylor et al. 2007). Desire to smoke was verbally assessed outside the scanner at baseline (T1), mid- (T2) and post-treatment (T3) for both conditions.
Subjective measures of cravings were analyzed using SPSS version 14. An initial paired t test was conducted to identify any order effects. To assess the effects of exercise compared with a passive control condition a two-way fully repeated ANOVA was conducted. The main effects of condition and time, and the interaction (condition × time) for the dependent variable were examined, using a Greenhouse–Geisser correction where appropriate. All subjective responses are presented as means (SD).
were analyzed using SPM2 Software (www.fil.ion.ucl.ac.uk/spm). The fMRI images were preprocessed–realigned, normalized, sliced timed (ascending sequence, 32 slices, TR = 3 s) and smoothed (to 6 mm). The movement threshold was 4 mm for transition and 4° for rotation. Event types were modeled for each subject (fixation, non-smoking, and smoking) using a canonical hemodynamic (hrf) with no time derivative. Following estimation, using a general linear model, high pass filter with a cut-off frequency of 128 and applying a Global correction (to avoid artifacts due to whole-brain signal changes between the two sessions), a series of one-sample t tests were carried out. Having created a series of t-contrast images for each subject and each event, a random effects (second level) analysis was carried out to assess which voxels showed consistent activation across all subjects between control and exercise conditions. An uncorrected statistical threshold (p < 0.0001) and a voxel cluster threshold of 24 were used. In order to determine the site of activation, an atlas of Talairach and Tounoux (1998) was used.
The participants (n = 6 males, n = 4 females) had an FTND score of 3.4 (1.6), smoked 13.7 (3.4) cigarettes per day, and had smoked for 8.1 (5.5) years. Participants exercised at a mean (SD) heart rate of 136 (14.6) bpm. Baseline ‘desire to smoke’ was 4.8 (0.47) and 4.4 (0.58) prior to exercise and control, respectively. A paired t test revealed no significant difference between the scores.
Analysis of the button-press responses made by participants during the scanning sessions revealed the following mean (SD) percentage response accuracy scores: following the control treatment, 98.07% (2.04); following the exercise treatment, 99.27% (1.33); therefore, an overall response accuracy of 98.67% (1.77).
Comparisons between control and exercise conditions from the random effects ANOVA revealed that in the control condition, significant activation clusters (uncorrected p < 0.0001, with a cluster threshold of 24) were observed in a number of brain structures thought to be involved in reward processing, including the precentral gyrus, parahippocampal gyrus (learning and memory, Graybiel 2005), and caudate (high proportion of dopamine neurones, e.g., Lynd-Balta and Haber 1994).
The coordinates of peak activation for control and exercise conditions (random effects, uncorrected (p<0.0001) with a cluster threshold of 24
Inferior frontal gyrus
Inferior frontal gyrus
Medial frontal gyrus
Medial frontal gyrus
Medial frontal gyrus
Superior frontal gyrus
Medial frontal gyrus
Medial frontal gyrus
Inferior parietal lobule
Inferior parietal lobule
Inferior parietal lobule
Inferior parietal lobule
Superior parietal lobe
Superior parietal lobe
Superior temporal gyrus
Middle temporal gyrus
Transverse temporal gyrus
Middle temporal gyrus
Comparison from the random effects ANOVA revealed that there was no significant difference in activated voxels between smoking and neutral images across the two conditions (control and exercise), but rather, the effects of exercise were found to be equal for both smoking and neutral images.
The present study supports previous evidence that an acute session of exercise reduces subjective measures of cravings (e.g., Janse Van Rensburg and Taylor 2008; Taylor et al. 2007). Previous research suggests that exercise may reduce the increase in self-reported cravings in responses to cigarette cues (Taylor et al. 2007). This is the first study to suggest the possible involvement of a neurobiological mechanism.
fMRI scanning post-treatments (control and exercise) identified several brain regions exhibiting differential activation patterns in response to the presented images. More specifically, post-control scanning was associated with increased regional activation in inferior and medial frontal lobes extending to the inferior and superior parietal lobes and middle and superior temporal lobes (including the parahippocampal gyrus). Activation was also seen in the caudate. Compared to the control condition, post-exercise scanning revealed reduced activation in areas of the dorso-lateral prefrontal cortex with a concomitant up regulation of activation in the medial rostal prefrontal cortex (incorporating BA10). The activation patterns noted in the control condition are in agreement with previous studies that have assessed the effects on regional brain activation when deprived smokers are presented with smoking-related stimuli and also reflect activation in some of the neurological circuits identified in nicotine addiction (e.g., reward, motivation, learning/memory, and control, Volkow et al. 2003).
Post-control scanning was associated with increased regional activation in subcortical structures of the limbic system (which is thought to contain a high proportion of dopamine neurones, e.g., caudate and parahippocampal gyrus), suggesting an enhanced perceived pleasure from the stimuli presented. Increased activation in the caudate itself is thought to reflect an enhanced hedonic effect of nicotine (Balfour 2001; Brody 2006; Koob and LeMoal 2001).
Post-control scanning also identified activation of the motivational circuit (as seen via activation of the prefrontal and orbitofrontal cortices (OFC)). According to Goldstein and Volkow (2002), activation in these areas is indicative of an individual experiencing both a heightened conscious experience of the drug cue and an enhanced incentive salience for the drug itself. Specifically, the OFC is thought to be a direct target for the action of drugs of abuse. This is because it receives direct projections from both dopamine cells in the ventral tegmental and from projections from other limbic areas believed to be involved in the rewarding effects of drugs (for review see Volkow and Fowler 2000).
A third area of interest seen in the post-control scanning session is the network responsible with visuo-spatial attention (reflected in activation of the parietal lobe precuneus, inferior and parietal lobes, parahippocampal and fusiform gyrus) possibly suggesting an enhanced attention to the presented stimuli. Indeed, regional activation in these associated areas has been previously shown in studies involving nicotine (Smolka et al. 2006; Due et al. 2002). Together with the activation of the limbic circuit (previously mentioned), the premise that these circuits are activated in unison when deprived smokers are presented with smoking-related cues is believed to reflect an increased attention to stimuli of potential importance to the individual (Lim et al. 2005; Due et al. 2002).
Post-exercise scanning revealed an interesting phenomenon. Although the task requirements and pictorial stimuli remained the same, we noted a substantial reduction of activation in areas of the dorso-lateral prefrontal cortex (compared with control scanning), while other areas of the frontal cortex (specifically BA10) showed increased activation. This finding can be explained by emerging evidence that exercise may selectively impair prefrontal-dependent cognition (Dietrich and Sparling 2004). This ‘Transient Hypofrontality’ hypothesis suggests that during (and for a period following) aerobic exercise there is a temporary inhibition of certain brain regions that are not directly essential to performing and maintaining the exercise, or physiological homeostasis. These are areas of the frontal lobe that are involved in higher cognitive functions (Dietrich 2003). Evidence for this occurrence stems from research that suggests that a single session of exercise selectively impairs performance on tasks that require prefrontal cognition compared to those that do not (Dietrich and Sparling 2004).
Given that there is evidence to suggest that the brain receives no additional resources during exercise (Dietrich 2003) and the premise that global blood flow (and global cerebral metabolism and oxygen uptake) to the brain during exercise remains constant (e.g., Ide and Secher 2002), it is of no surprise that exercise can be seen as placing a large strain on the brain’s limited information-processing capacity (Dietrich 2003). The brain becomes overloaded with having to control for the physiological demand placed on the body by the exercise and is, according to Dietrich (2006), unable to sustain activation in all neural structures at once. Thus, the activation of structures involved in maintenance of physiological homeostasis comes at the expense of the others (i.e., those involved in higher cognitive functioning and addiction processes).
We found that the frontal cortex did not act as a single unit. Instead, coupled with the down regulation of areas of the dorso-lateral prefrontal cortex, we found significantly more activation in BA10 (also referred to as the ‘default mode’) post-exercise compared to post-control scanning.
The ‘default mode’ network is thought to be activated during baseline or resting states (Raichle et al. 2001) and there is emerging evidence that suggests that it can also be activated during tasks that require limited cognitive demands (Greicius et al. 2003; Gilbert et al. 2005, 2006). However, evidence of the role that the ‘default mode’ plays in human cognition seems unresolved, and in the context of findings from this study could reflect one (or both) of two theories.
The first theory that could be reflected here is the concept that activation in the ‘default mode’ may reflect an individual’s own mental state (Gilbert et al. 2005). Emerging evidence suggests that parts of the medial PFC have connections in the brain’s emotional processing. According to Simpson et al. (2001), a decrease in focused attention (seen in the exercise condition) may reflect a dynamic interplay between the emotional status of the individual and the cognitive process. This concept is interesting given the accumulating evidence that a single session of exercise can have a positive and consistent effect on mood (e.g., Bartholomew et al. 2005). To our knowledge, there is no fMRI study on the affective response of a single session of exercise on mood. However, our assumption is that activation of the ‘default mode’ during post-exercise scanning may reflect some type of Opponent Process (Solomon 1980) response to the session of exercise, suggesting a state of contentment due to completing the exercise; this consequently causes an alteration in the individual’s cognitive processing (perhaps diverting thoughts away from smoking-related cues).
The second theory that could be considered is that the ‘default mode’ is thought to reflect a baseline state of human brain functioning that is ‘suspended during attention-demanding or goal directed behaviours’ (Raichle et al. 2001) or reflects enhanced attention/thoughts that are directed towards the external environment (e.g., Gilbert et al. 2006). In our study, participants were required to respond via button-press to the smoking and neutral stimuli (participants responded with an accuracy of 98.07% and 99.27% control and exercise sessions, respectively). It could be suggested that exercise enabled participants to perceive the ‘task’ as requiring less cognitive demand. This may be further substantiated by the lack of activation in brain areas associated with visuo-spatial attention (as seen in the control condition), suggesting individuals, although were responding correctly to stimuli, were less occupied with the connotations of the stimuli during the exercise condition scanning.
To our surprise, activation clusters were also found in the cingulate gyrus (anterior cingulate) and the precuneus during post-exercise scanning. These areas have been associated with reward anticipation, and form part of the limbic system. However, there is evidence to suggest that the anterior cingulate is also involved in autonomic functions such as the regulation of blood pressure and heart rate. This seems a plausible explanation given that exercise is associated with changes in these functions (Lechin et al. 1995; Forjaz et al. 1998; Fagard et al. 2001). Furthermore, activation in the precuneus may be the result of the perceived pleasurable effects of taking part in the exercise (e.g., an emotional response to exercise) (Morris and Hardman 1997; Sandlund and Norlander 2000; Berlin et al. 2006). Further studies are needed to identify the exact neurological effects of an acute bout of exercise.
We felt it necessary to match smoking and neutral images closely for content (e.g., pen compared to cigarette), color (image brightness) and body parts (e.g., hands holding pens or cigarettes) in order to ensure control over the stimulus effects, this may have been a problem for two reasons. Firstly, given that a large proportion of smokers self-report declines in concentration during abstinence, it may have been that the images were too similar to each other for the smokers to distinguish between their content, coupled with the relatively short duration of stimulus presentation, smokers may have required additional resources to process these images, thus detracting from their emotional content. Increased activation to neutral cues in the same proportion to smoking cues has also been found by others (e.g., McClernon et al. 2005). Secondly, there is evidence to suggest that neurons will show decrements in firing patterns with the repeated presentation of stimuli that are very similar to each other (e.g., Xiang and Brown 1998). Thus, it may have been that activations to images were dampened towards the end of the protocol, causing a reduced overall activation to the images. The fact that we were unable to distinguish significant differences in regional activation between control and smoking-related stimuli is in contrast with some previous studies (e.g., Due et al. 2002; David et al. 2005) and our own pilot work with seven smokers (following abstinence but no exercise). However, we believe that this does not suggest individuals showed a lack of attentional focus or were not attending to the cues, but rather, this may have been due to methodological issues which we are currently exploring in further research.
We found that the session of exercise had the ability to down regulate and shift regional brain activation in response to both smoking and neutral cues (rather than an isolated effect on smoking cues). Thus, it may be suggested that exercise has an overall tonic effect on the way in which information was processed (e.g., a shift from ‘reward seeking’ mode (seen post-control session) to the ‘default mode’ (seen post-exercise session) of brain function). In this study, we used participants who were deprived from nicotine. Future studies may wish to compare a range of cognitive activation tasks in smokers and non-smokers to validate the role of exercise in affecting the manner in which the brain processes information.
Limitations exist in the present study, which should be addressed in future research. Firstly, while differences in regional activation were identified between post-exercise and control conditions, it is not clear what molecular changes occurred. Secondly, the absence of a difference between activation in response to smoking and neutral images may be due to methodological factors such as the length of presentation of the stimuli and the ISI. Thirdly, the effects reported are from moderately heavy smokers: The observed effects may have been different among heavier smokers, with greater reactivity to the smoking cues. Fourthly, more vigorous exercise may have produced different effects, given the premise that a minimum level of exercise intensity is required to force the redistribution of resources in the brain (Dietrich 2003). All measures were conducted outside the scanner, future studies may wish to measure subjective cravings at given time points throughout the scanning session. Lastly, like many other fMRI studies, this study may have been limited by sample size, thereby, limiting the power to detect differences in post-treatment activation, although clearly, we were able to report significant reductions in self-reported cravings.
This study adds further evidence that a single session of exercise has the ability to reduce the desire to smoke, experienced during cigarette abstinence. In terms of regional brain activation, in the control condition we found activation clusters in areas associated with reward processing and visuo-spatial attention, suggesting individuals experience an increase in cravings and emotional attachment toward the smoking-related stimuli during temporary nicotine abstinence. In contrast, following the exercise session, we found no such activations, but rather activation in regions associated with the brain ‘default mode’. Thus, it may be that the benefits of exercise are that it acts as a tool to increase the strain on the brain’s information-processing capacity, causing hypo-activation in areas of the frontal cortex that are involved in reward processing and linked with cigarette cravings. More specifically, following exercise, increased activation of the medial rostral prefrontal cortex (including BA 10) was observed with no increased activation in other regions involved in nicotine cravings, such as the dorso-lateral prefrontal areas.
We have shown that changing the emotional processing for smoking-related stimuli is linked to the pathophysiology of cigarette cravings (Lim et al. 2005). Of further importance in shifting the emotive response to cigarette cues is the premise that increased attentional bias to smoking-related images has been found to be a successful predictor of relapse to smoking (Waters et al. 2003).
This is the first study to show shifts in regional brain activation in response to smoking-related images, following exercise, and during abstinence.
Thanks to Dr Jon Fulford for his technical support and Dr Tim Rees for his thoughtful suggestions on the manuscript.
Disclosure/Conflict of Interest
None of the authors have received financial reward/benefit for this research.
- Borg GAV (1998) Borg’s Perceived Exertion and Pain Scales. Human Kinetics, Champaign, IllGoogle Scholar
- Gilbert DG, Rabinovitch NE (1998) International Smoking Image Series with neutral counterparts, 1.2 edn. Southern Illinois University at Carbondale, Carbondale, ILGoogle Scholar
- Talairach J, Tounoux P (1988) Co-planar stereotactic atlas of the human brain: 3-dimensional proportional system: An approach to cerebral imaging. Thieme, StuttgartGoogle Scholar
- Ussher M, Taylor AH, Faulkner G (2008) Exercise interventions for smoking cessation. Cochrane Rev (4):CD002295Google Scholar