Neuroimaging for drug addiction and related behaviors (2012)

Functional magnetic resonance imaging (fMRI)

The creation of an MR image requires that the object is placed within a strong magnetic field. Magnetic strength for human MRI scanners ranges from 0.5 to 9.4 T; however, the strength of most clinical MRI scanners is 1.5–3 T. Within a magnetic field, the nuclear spins of certain atoms within the object are oriented either parallel or anti-parallel to the main magnetic field and precess (spin) about the main magnetic field with a certain frequency called the Larmor frequency. Magnetic resonance occurs when a radio frequency (RF) pulse, applied at the (tissue specific) Larmor frequency, excites the nuclear spins, raising them from lower to higher energy states. This is represented by a rotation of the net magnetization away from its equilibrium. Once the magnetization is rotated, the RF field is switched off and the magnetization once again freely precesses about the direction of original main magnetization. This time-dependent precession induces a current in a receiver RF coil. The resultant exponentially decaying current, referred to as the free induction decay, constitutes the MR signal. During this period, magnetization returns to its original equilibrium state (also known as relaxation), characterized by two time constants T₁ and T₂ (Lauterbur, 1973). These time constants depend on physical and chemical characteristics unique to tissue type and hence are the primary source of tissue contrast in anatomical images (Mansfield and Maudsley, 1977). These T₁ and T₂ differences between different tissue types (e.g., gray matter, white matter, and cerebrospinal fluid) yield a high-contrast MR image.

It was not until the 1990s that MRI was used to map human brain function non-invasively, rapidly, with full brain coverage, and with relatively high spatial and temporal resolution. Belliveau et al. (1990), using gadolinium as contrast agent, was the first to introduce the functional MRI (fMRI). This was then immediately followed by a series of fMRI studies using the ‘Blood Oxygen Level Dependent’ (BOLD) signal (Ogawa et al., 1990a,b) as endogenous contrast agent for indirect measure of brain activity (Bandettini et al., 1992; Kwong et al., 1992; Ogawa et al., 1992). Recently, work by Logothetis et al. (2001) has explored a causal relationship between the BOLD signal and neuronal local field potentials (see Logothetis, 2003; Logothetis and Wandell, 2004 for reviews).

fMRI has become perhaps the most widely used functional neuroimaging technique because of its non-invasive nature (unlike PET and SPECT, it does not expose participants to radioactivity) and very high spatial resolution (~1 mm). The limitations of this technique include high susceptibility of the BOLD response to several non-neural and imaging artifacts, especially due to its low signal-to-noise ratio and low temporal resolution (~1–2 s) compared to other techniques, such as EEG (although much higher than that of PET). More recently, the use of fMRI at rest has enabled researchers to investigate resting functional connectivity of the human brain (Rosazza and Minati, 2011). Measures of resting functional connectivity have been shown to be reproducible and consistent across laboratories (Tomasi and Volkow, 2010) and to be sensitive to diseases of the brain including drug addiction (Gu et al., 2010).

Craving

The pharmacological effects of a drug are modulated by non-pharmacological contextual factors (e.g., places, people, or paraphernalia associated with drug intake). As these factors are consistently paired with the pharmacological effects of the drug they are integrated into the intense experience associated with drug use, becoming ‘motivational magnets’ or ‘drug cues’ through Pavlovian conditioning (Berridge, 2007; Berridge et al., 2008). This conditioning shapes an individual’s expectations of the effects of a drug and, in turn, modifies the neural and behavioral responses to the drug. For example, in drug-addicted individuals, attention and other cognitive and motivational processes are biased towards the drug and away from non-drug stimuli culminating in an urgent desire to consume the drug in susceptible individuals (e.g., Johanson et al., 2006).

In laboratory settings, a craving state is usually achieved by exposing participants to images depicting drug-related stimuli. Using this technique with cocaine users, PET [¹¹C]raclopride studies have revealed that cocaine cue videos can elicit a significant release of DA in the dorsal striatum and this increase is positively associated with self-reported drug craving especially in severely addicted individuals (Volkow et al., 2006, 2008). Another PET study showed that chronic cocaine abusers retain some level of cognitive control when instructed to inhibit cue-induced craving as quantified by lower metabolism with cognitive inhibition in the right OFC and the NAcc (Volkow et al., 2010). These results are consequential as there is a significant association between DA D₂ receptor binding in the ventral striatum and the motivation for drug self-administration, as measured by [¹¹C]raclopride (Martinez et al., 2005) and [¹⁸F] desmethoxyfallypride (Heinz et al., 2004).

Studies measuring CBF, glucose metabolism, or BOLD have also shown that drug cue-induced craving in drug-addicted individuals is associated with activations in the perigenual and ventral ACC (Maas et al., 1998; Childress et al., 1999; Kilts et al., 2001; Wexler et al., 2001; Brody et al., 2002, 2004; Daglish et al., 2003; Tapert et al., 2003, 2004; Grusser et al., 2004; Myrick et al., 2004; McClernon et al., 2005; Wilson et al., 2005; Goldstein et al., 2007b), medial PFC (Grusser et al., 2004; Heinz et al., 2004; Tapert et al., 2004; Wilson et al., 2005; Goldstein et al., 2007b), OFC (Grant et al., 1996; Maas et al., 1998; Sell et al., 2000; Bonson et al., 2002; Brody et al., 2002; Wrase et al., 2002; Daglish et al., 2003; Tapert et al., 2003, 2004; Myrick et al., 2004) insula (Wang et al., 1999; Sell et al., 2000; Kilts et al., 2001; Brody et al., 2002; Daglish et al., 2003; Tapert et al., 2004), ventral tegmental area and other mesencephalic nuclei (Sell et al., 1999; Due et al., 2002; Smolka et al., 2006; Goldstein et al., 2009c). Brain regions that are involved with memory processing and retrieval are also activated during craving, including the amygdala (Grant et al., 1996; Childress et al., 1999; Kilts et al., 2001; Schneider et al., 2001; Bonson et al., 2002; Due et al., 2002), hippocampus, and brainstem (Daglish et al., 2003). Of note is evidence showing that these effects are observed even when controlling for the effects of pharmacological withdrawal (Franklin et al., 2007).

In general, findings from craving studies in drug abusers suggest increased mesocortical (including the OFC and ACC) activation when processing drug cues and that drug expectation plays a significant role in this process. Such evidence in part explains the difficulty for drug abusers to focus on other non-drug related cues. Interestingly, in females but not in male cocaine abusers a PET study showed decreases in metabolism in prefrontal regions involved with self-control following exposure to cocaine cues, which could render them more vulnerable (than males) to relapse if exposed to the drug (Volkow et al., 2011). This finding is consistent with preclinical studies suggesting that estrogen may increase the risk for drug abuse in females (Anker and Carroll, 2011).

EEG has also been used to investigate the reactivity to drug-associated stimuli across different drugs of abuse. For example, increased cortical activation has been reported in response to drug cue exposure in alcohol-dependent patients (quantified by EEG dimensional complexity) (Kim et al., 2003), and in cocaine-addicted individuals (quantified by high beta and low alpha spectral power) (Liu et al., 1998). Another study of cocaine-addicted individuals showed an increase in beta spectral power along with a decrease in delta power while handling cocaine paraphernalia and viewing a crack cocaine video (Reid et al., 2003). This pattern was also observed when comparing these individuals to healthy controls during rest (Noldy et al., 1994; Herning et al., 1997) and this increase in beta was associated with amount of prior cocaine use (Herning et al., 1997). In nicotine addiction, an increase in theta and beta spectral power was observed in response to cigarette-related cues (Knott et al., 2008). Higher cortical activation in response to drug cues has also been reported in ERP studies. For example, increased amplitude of P300 and other P300-like potentials have been reported in response to drug cues in alcohol- (Herrmann et al., 2000) and nicotine- (Warren and McDonough, 1999) addicted individuals. Increased LPP amplitudes have also been reported in response to drug-related pictures compared to neutral pictures in alcohol- (Herrmann et al., 2001; Namkoong et al., 2004; Heinze et al., 2007), cocaine- (Franken et al., 2004; van de Laar et al., 2004; Dunning et al., 2011), and heroin- (Franken et al., 2003) addicted individuals.

Broadly, these data suggest that drug-associated stimuli are related to significantly higher neural activations, suggesting an increase in incentive salience and arousal when drug-associated stimuli are encountered or anticipated by drug-addicted individuals. These results corroborate theories that posit addiction as an alteration to the brain’s motivation and reward systems (Volkow and Fowler, 2000; Robinson and Berridge, 2001; Goldstein and Volkow, 2002), where processing is biased towards drugs and conditioned cues and away from other reinforcers as associated with craving (Franken, 2003; Mogg et al., 2003; Waters et al., 2003).

Loss of inhibitory control and bingeing

Inhibitory control is a neuropsychological construct that refers to the capacity to control the inhibition of harmful and/or inappropriate emotion, cognition, or behavior. Critically, the disruption of self-controlled behavior is likely to be exacerbated during drug use and intoxication as modulated by a compromise in an essential function of the PFC: its inhibitory effect on subcortical striatal regions (including NAcc) (Goldstein and Volkow, 2002). This impairment in top-down control (a core PFC function) would release behaviors that are normally kept under close monitoring, simulating stress-like reactions in which control is suspended and stimulus-driven behavior is facilitated. This suspension of cognitive control contributes to bingeing; a discrete period of time during which an individual engages in the repeated and unabated consumption of the substance often at the expense of behaviors needed for survival including eating, sleeping, and maintaining physical safety. These periods usually discontinue when the individual is severely exhausted and/or unable to procure more of the drug.

Neuroimaging studies suggest the involvement of thalamo-OFC circuit and the ACC as neural substrates underlying bingeing behavior. Specifically, it has been reported that addicted individuals have significant reductions in D₂ receptor availability in the striatum (see Volkow et al., 2009 for a review), which in turn is associated with decreased metabolism in the PFC (especially OFC, ACC, and dorsolateral PFC), and that these impairments cannot be fully attributed to impaired behavioral responses and motivation (Goldstein et al., 2009a). As these PFC regions are involved in salience attribution, inhibitory control, emotion regulation, and decision-making, it is postulated that DA dysregulation in these regions may enhance the motivational value of the drug of abuse and may lead to loss of control over drug intake (Volkow et al., 1996a; Volkow and Fowler, 2000; Goldstein and Volkow, 2002).

Indeed, there is evidence showing that these regions, particularly the OFC, are critical in other disorders of self-control involving compulsive behaviors such as obsessive-compulsive disorder (Zald and Kim, 1996; Menzies et al., 2007; Chamberlain et al., 2008; Yoo et al., 2008; Rotge et al., 2009).

Although it is difficult to test compulsive drug self-administration in humans, clever laboratory designs have overcome some of the practical constraints encountered when studying bingeing in humans. For example, in a recent fMRI study, non-treatment-seeking cocaine-dependent individuals were permitted to choose when and how often they would self-administer intravenous cocaine within a supervised 1-h session. Repeated self-induced high was negatively correlated with activity in limbic, paralimbic, and mesocortical regions including the OFC and ACC. Craving, by contrast, positively correlated with activity in these regions (Risinger et al., 2005) (also see Foltin et al., 2003). Simulating compulsive drug self-administration vis-à-vis other compulsive behavior (such as gambling when it is clearly no longer beneficial) may offer invaluable insight into the circuits underlying loss of control in addictive disorders. Interestingly, oral MPH significantly decreased impulsivity and improved the underlying ACC responses in cocaine-addicted individuals (Goldstein et al., 2010).

Another related construct is the compromised self-awareness in drug-addicted individuals. Dysfunctional self-awareness and insight characterize various neuropsychiatric disorders, spanning classic neurological insults (e.g., causing visual neglect or anosognosia for hemiplegia) to classic psychiatric disorders (e.g., schizophrenia, mania, and other mood disorders), as recently reviewed (Orfei et al., 2008). As a cognitive disorder (Goldstein and Volkow, 2002), drug addiction also shares similar abnormalities in self-awareness and behavioral control that can be attributed to an underlying neural dysfunction. For instance, studies in alcohol abuse have reported that alcohol reduces the individual’s level of self-awareness by inhibiting higher order cognitive processes related to (attending, encoding or sensitivity to) self-relevant information, a sufficient condition to induce and sustain further alcohol consumption (see Hull and Young, 1983; Hull et al., 1986 for reviews). Moreover, a recent study has shown that cocaine-addicted individuals manifest a disconnect between task-related behavioral responses (accuracy and reaction time) and the self-reported task engagement, highlighting the disruption in their ability to perceive inner motivational drives (Goldstein et al., 2007a).

Specifically, abnormalities in the insula and medial PFC regions (including ACC and medial OFC), and in subcortical regions (including the striatum), have been associated with insight and behavioral control, and with interrelated functions (habit formation and valuation) (Bechara, 2005). These considerations expand the conceptualization of addiction beyond its association with the reward circuit, neurocognitive impairments in response inhibition, and salience attribution (Goldstein and Volkow, 2002; Bechara, 2005) and neuroadaptations in memory circuits (Volkow et al., 2003), to include compromised self-awareness and insight into illness (see Goldstein et al., 2009b for a review).

Studies employing EEG have reliably reported low-voltage beta frequencies (Kiloh et al., 1981; Niedermeyer and Lopes da Silva, 1982) in alcoholics. This beta activity, which may reflect hyperarousal (Saletu-Zyhlarz et al., 2004), has been shown to correspond to the quantity and frequency of alcohol intake, reliably differentiating between ‘low’ and ‘moderate’ alcohol drinkers (determined by pattern of alcohol consumption), as well as familial history of alcoholism (Ehlers et al., 1989; Ehlers and Schuckit, 1990). Simultaneous increases in delta were reported in high-binge drinkers compared to non- and low-binge young adult alcohol drinkers (Polich and Courtney, 2010), and with concomitant increase in theta and alpha frequencies 25 min post-binge-like cocaine dosing (Reid et al., 2006).

Inhibitory control has widely been studied by quantifying N200 and P300 ERP components in go/no-go tasks; these components, thought to measure successful behavioral suppression and cognitive control (Dong et al., 2009) and generate from ACC and associated regions, are increased when a response is withheld (no-go trial) within a series of positive responses (go trials) (Falkenstein et al., 1999; Bokura et al., 2001; Van Veen and Carter, 2002; Bekker et al., 2005). Blunted N200 amplitudes have been reported in individuals with alcohol (Easdon et al., 2005), cocaine (Sokhadze et al., 2008), heroin (Yang et al., 2009), nicotine (Luijten et al., 2011), and even internet (Cheng et al., 2010; Dong et al., 2010) addiction. However, binge drinkers showed larger N200 and smaller P300 as compared to controls, in a sustained attention pattern matching task (Crego et al., 2009) and face recognition task (Ehlers et al., 2007), which may actually be consistent with emotional processing impairment (motivation, salience) more than with loss of control.

Animal models of addiction have provided important clues about the neurobiology underlying bingeing behavior (Deroche-Gamonet et al., 2004; Vanderschuren and Everitt, 2004) showing that these behaviors involve DA, serotonergic, and glutamatergic circuits (Loh and Roberts, 1990; Cornish et al., 1999). However, the utility of animal studies rests on the degree to which these behaviors overlap with inhibitory self-control in humans. In particular, it is difficult to ascertain the degree to which such behaviors may be relevant to the putative cognitive deficits that may underlie impaired inhibitory control in humans. Neuroimaging studies circumvent this limitation by investigating the neural substrates underlying these cognitive deficits and by providing a link to the corresponding behavioral manifestations.

This work was supported by grants from the National Institute on Drug Abuse [1R01DA023579 to R.Z.G.] and General Clinical Research Center [5-MO1-RR-10710].