INTRODUCTION

The connection between striatal function and addiction is well founded as this brain region has a major role in subjective drug craving (Volkow et al, 2006) and the maintenance of drug-cue associations (See et al, 2007). In relation to tobacco smoking specifically, the striatum mediates the rewarding effect of smoking (Barrett et al, 2004), the motivation to smoke (Le Foll et al, 2014), and damage to this region due to stroke disrupts nicotine dependence (Gaznick et al, 2014). The striatum also shows persistent reactivity to smoking-related cues even during abstinence, which may contribute to relapse risk during a quit attempt (Janes et al, 2009; McClernon et al, 2009). Recently, larger striatal volume has been linked with greater lifetime exposure to nicotine (Das et al, 2011), which calls into question whether there is a link between striatal function and structural phenotype. Although previous studies have evaluated striatal volume in the context of addiction (Churchwell et al, 2012; Das et al, 2011; Ersche et al, 2012), anomalies in subcortical shape are beginning to emerge as a robust endophenotype in neuropsychiatric disorders (Shaw et al, 2014), suggesting that this measure may provide additional insight into the relationship between striatal morphology and substance abuse.

To assess the link between craving and striatal anatomy, we used a new segmentation methodology coupled with local shape indices along the striatal surface to estimate striatal volume (Chakravarty et al, 2013), surface area, and shape (Raznahan et al, 2014). For volumetric analyses, we assessed both total striatal volume as well as striatal subregions to get a sense of both global and localized structural effects. These measures provide a comprehensive evaluation of striatal morphology not easily captured by standard volumetric analyses alone. We then assessed the link between striatal morphology and two measures of cigarette craving: baseline subjective craving as measured by the brief questionnaire of smoking urges (QSU; Cox et al, 2001) and craving induced by exposure to smoking-related cues (Janes et al, 2013). Although both measures evaluate craving, the QSU assesses craving state prior to exposure to smoking cues, whereas cue-induced craving measures the amount of craving induced by visual stimuli associated with nicotine reward. We hypothesize that smokers who report more intense cigarette craving will have larger striatal volume, which may partly explain the connection between larger striatal volume and lifetime cigarette use (Das et al, 2011). In addition, we will build upon the existent literature by providing the first analysis that links cigarette craving and more nuanced measures of striatal morphology including surface area and shape assessments.

MATERIALS AND METHODS

Participants

Twenty-five nicotine-dependent participants (6 left-handed, 15 women/10 men) participated in all study measures at the McLean Imaging Center of McLean Hospital. Participants were 26.3±5.3 (mean±SD) years old and had an average of 6.4±4.1 pack-years of smoking experience. Pack-years of smoking use were calculated by multiplying the average number of packs of cigarettes smoked per day by the number of years the participant smoked. Inclusion criteria included smoking ⩾10 cigarettes/day over the past 6 months and on average, participants smoked 14.2±3.7 cigarettes per day. All participants were moderately to heavily nicotine dependent as indicated by an average Fagerstrom test for nicotine dependence (Heatherton et al, 1991) score of 6.12±1.1. Participants were assessed by the Structured Clinical Interview for DSM and met criteria for current nicotine dependence. Participants had an average expired carbon monoxide level of 24.4±12.4 ppm and 15±2.1 years of education. Participants were excluded if they had a lifetime diagnosis of the following conditions: organic mental disorder, bipolar depression, or schizophrenia spectrum disorder or depressive symptoms in the past 6 months. Participants were excluded for current substance use disorder other than nicotine dependence. Smokers also were excluded for pregnancy, recent drug use, or excessive alcohol use. Participants were recruited using online advertisements and fliers posted in the Boston metropolitan area. All participants provided both verbal and written informed consent prior to participating in the study and the institutional review board at McLean Hospital approved this study.

Magnetic Resonance Imaging

All participants smoked one of their own cigarettes immediately following signing the informed consent to standardize the time as a cigarette was last smoked. MRI scanning began ∼1.5 h after smoking this cigarette. Scans were acquired on a Siemens Trio 3 Tesla scanner (Erlangen, Germany) with a 32-channel head coil. Multiecho multiplanar rapidly acquired gradient-echo (ME-MPRAGE) structural images were acquired with the following parameters (TR=2.1 s, TE 3.3 ms, slices=128, matrix=256 × 256, flip angle=7°, resolution=1.0 mm × 1.0 mm × 1.33 mm).

Craving Measures

Craving was assessed just prior to scanning (1.5 h after smoking) using the 10-item QSU (Cox et al, 2001), which is a standard tobacco craving measure. Following scanning (3 h after smoking), participants were shown smoking and neutral images and were asked to rate the amount of craving they experienced when viewing each image on a five-point scale. Smoking and neutral images were comprised of images from various sources including the International Smoking Image Series (Gilbert and Rabinovich, 1999) and those used in our prior work (Janes et al, 2013) and by Kober et al (2010). To control for visual characteristics, all images were converted to gray scale. The smoking images represented people smoking, hands holding cigarettes, and items such as cigarettes. The neutral images, which lacked the smoking element, were matched for content and included images of people, hands holding objects such as paintbrushes, and items such as pens. One hundred and forty-four images of each type were presented. The average difference in craving to smoking vs neutral images was then calculated and was associated with brain morphology. Average reported subjective craving between smoking and neutral images was compared using a paired t-test.

Striatal Volume and Surface Area

Neuroanatomical segmentations are typically limited in accuracy by errors in the transformation estimation, irreconcilable neuroanatomical differences between the neuroanatomy of the template and the subject, and resampling errors after the application of the transformation to the labels (Collins and Pruessner, 2010; Heckemann et al, 2006). MAGeT Brain limits these errors by using a modified ‘multi-atlas’-based strategy while still relying on a single well-defined template. As input, the algorithm uses a single atlas containing the striatum, which was previously defined using a three-dimensional reconstruction of serial histological data (Chakravarty et al, 2006) that was warped to an MRI template (Holmes et al, 1998). The atlas is first customized to a subset of the data set (21 subjects, based on an arbitrary subsample of the cohort being studied) using a nonlinear transformation estimated in a region of interest defined around the subcortical structures (Chakravarty et al, 2008, 2009, 2012). This set of subjects now acts as a set of templates to which all other subjects are now warped, providing 21 candidate segmentations for each subject and effectively uses the neuroanatomical variability of the cohort being analyzed to improve overall segmentation accuracy. The final segmentation is decided upon by using a voxel-wise majority vote (ie, the label occurring most frequently at a specific voxel location is retained; Collins et al, 1995). In our original paper, we demonstrated high overlap between our method and gold standards manually derived on MRI data (Dice Kappa of 0.861 when using MAGeT Brain with the ANIMAL algorithm (Collins et al, 1995) for nonlinear transformation estimation; Chakravarty et al, 2013). As in our recent work (Raznahan et al, 2014), we also quantified local surface area. A surface-based model was warped to each subject using all 21 transformation estimated through the MAGeT Brain segmentation procedure. To ensure homology between vertices, all surface vertices were redefined using a Vornoi diagram (see Lyttleton et al, 2009). The median position was taken for each coordinate at each vertex in order to create a ‘median surface’ representation. This can be considered to be the surface-based homolog to majority-vote label fusion measures and has the benefit of filtering noise and outliers from the surface-based representation. Total surface area was then estimated as the sum of the surface area for each polygon in the surface. If a significant relationship between craving and total striatal morphology was identified, a striatal subregion analysis was then conducted to identify a more localized effect. Striatal subregion volumes were estimated using the MAGeT Brain algorithm as well. In this case, the striatal mask was subdivided in the original histological data set in order to identify the ventral striatum and the caudate and putamen.

Vertex-Wise Analyses of Striatal Shape: Surface Displacement

Shape analysis was carried out using a surface-based methodology proposed by Lerch et al (2008). First, surface-based representations of the striatum defined on the input atlas were estimated using the marching cubes algorithm (Lorensen and Cline, 1987) and morphologically smoothed using the AMIRA software package (Visage Imaging, San Diego, CA, USA). The resulting surfaces are defined by ∼6300 vertices per striatum. The nonlinear portions of the 21 transformations that map each subject to the input atlas were concatenated and then averaged across the template library in order to limit the effects of noise and error and to increase precision and accuracy (Frey et al, 2011; Dorr et al, 2008; Borghammer et al, 2010). The dot product between the nonlinear deformation vector (of the inverse of the averaged atlas-to-subject transformation) and the striatum surface normal (a unit vector describing the direction perpendicular to the surface) is estimated and provides a local measure of inward or outward displacement along the normal (as first described in Lerch et al (2008); and as in the method described in Magon et al, (2014)). Prior to estimating the dot product between the normal to the surface and the deformation field, we explicitly modeled and removed global linear effects not originally accounted for in the initial linear transformations. This ensures that the contributions of overall differences in brain volume are minimized in this analysis.

Statistics

RMINC (https://github.com/mcvaneede/RMINC), a statistical image analysis software package built to work in the R environment, was used for all surface-based vertex-wise analyses. A linear model was used and all imaging results were corrected using a 10% false discovery rate (FDR) correction (Genovese et al, 2002). All analyses with respect to PSYRATS subscales were conducted using the values of the assessment. SPSS was used to conduct correlation and regression analyses. Correlations were conducted to evaluate the relationship between QSU and cue-induced craving measures, whereas regression analyses were used to evaluate associations between brain morphology and craving measures. For the volumetric analyses, total brain volume, age, and sex were included as control variables.

RESULTS

Subjective Craving and Smoking History

Participants reported experiencing significantly greater craving when viewing smoking vs neutral images (t=7.6, p<0.0001). Smoking vs neutral cue-induced craving was positively correlated with craving as measured by the QSU (r=0.42, p=0.035).

Striatal Morphology

Striatal volume

A significant positive association was found between left striatal volume and craving as measured by the QSU (t=2.6, p=0.016; Figure 1) and craving induced by exposure to smoking vs neutral cues (t=2.3, p=0.032; Figure 1). Given this significant association, a follow-up subregion analysis was conducted. Dorsal, but not ventral, striatal volume was related with both the QSU (t=2.3, p=0.026) and smoking vs neutral cue-induced craving (t=2.3, p=0.03). Craving measured by the QSU was significantly associated with the caudate (t=3.3, p=0.004), but not the putamen. Smoking vs neutral cue-induced craving was not related to either dorsal striatal subregion. No significant associations were found between right total striatal volume and either craving measure.

Figure 1
figure 1

Left striatal volume and surface area relationships with craving. A positive relationship was found between craving (1. QSU and 2. cue-induced) and left striatal volume and surface area.

PowerPoint slide

Full size image

Surface area

A positive association was found between the left striatal surface area QSU (t=2.5, p=0.02; Figure 1) and smoking vs neutral cue-induced craving (t=0.02, p=0.007, Figure 1).

Striatal shape

A significant relationship was found between smoking vs neutral cue-induced craving and right striatal shape deformations for both local contractions (t-statistic range: −2.7<t<−3.95) and expansions (t-statistic range: 2<t<3.56) (FDR corrected 10%; Figure 2). Similar trends were found for the left striatum, but were not significant (Figure 2). No relationship was found between striatal shape and QSU.

Figure 2
figure 2

Cue-induced craving and striatal shape deformations. A significant association was found between smoking cue-induced craving and both local (a) contractions and (b) expansions in the right striatum. Colored regions indicate the range of t-statistics and regions of significance after FDR correction (q=0.1), only applicable for the right striatum. An arbitrary threshold (t=2) is chosen for trend-level results for the left striatum, which are not significant after correction.

PowerPoint slide

Full size image

DISCUSSION

These current findings show a relationship between cigarette craving and striatal morphology. Both subjective craving, as measured by the QSU, and craving induced by exposure to smoking cues were associated with larger left striatal volume and surface area. Cue-induced craving was significantly related with more profound right striatal shape deformations. Given the striatum’s functional role in craving (Volkow et al, 2006), and cue-reactivity (See et al, 2007) the current findings suggest a link between striatal morphology and its underlying function. In addition, our novel surface area and shape findings provide more nuanced insight into the relationship between striatal morphology and craving.

The relationship between craving and left lateralized striatal volume and surface area is consistent with previous findings linking stimulant abuse with greater left striatal volume (Das et al, 2011;Churchwell et al, 2012, Ersche et al, 2012). The connection between the left striatum and substance abuse may be tied to the more dominant role of the left hemisphere in reward and approach behavior (Davidson and Irwin, 1999). In fact, the left striatum shows greater activation following unexpected reward delivery (McClure et al, 2003) and sustained activation following reward feedback (Delgado et al, 2000). Greater positive incentive motivation also is associated with higher dopamine D2 receptor availability in the left relative to the right striatum (Tomer et al, 2008). It is tempting to speculate that the relationship between left striatal volume and craving is due to greater left striatal D2 receptor availability. Such a hypothesis is plausible given that D2 binding potential is positively associated with striatal volume (Woodward et al, 2009), and craving (Volkow et al, 2006).

When assessing more localized left striatal subregions, both the QSU and smoking cue-induced craving measures were associated with the dorsal, but not ventral striatum. On further inspection, only the QSU was associated with more localized volume in the caudate, indicating that the specificity of the volumetric association depends on the type of craving evaluated. This specific relationship between caudate volume and QSU is consistent with the caudate’s role in craving (Volkow et al, 2006) and reward processing (Delgado et al, 2000). In contrast, smoking cue-induced craving is associated with enhanced dorsal striatal volume, which includes both the caudate and putamen. These findings suggest that craving, induced by exposure to smoking-related cues may involve more distributed striatal structures, whereas subjective craving, independent of cue-exposure, is associated with more localized caudate morphology.

A relationship between smoking cue-induced craving and right dorsal striatal shape was found independent of overall striatal volume, specifically in the ventral anterior putamen and the right caudate tail. The ventral anterior putamen not only is larger in occasional stimulant users (Mackey et al, 2014), but this striatal subregion receives projections from the insula (Chikama et al, 1997), a brain region implicated in maintaining nicotine dependence (Naqvi et al, 2007) and smoking cue-reactivity (Janes et al, 2010). The insula and striatum may interact to mediate nicotine dependence, as damage to both of these brain regions disrupts smoking (Gaznick et al, 2014), and suggests a possible link between ventral anterior putamen deformations and cue-reactivity due to striatal-insula interactions. The fact that cue-reactivity was associated with shape differences in the caudate tail is particularly interesting given the role of this striatal subregion in visual habits (Fernandez-Ruiz et al, 2001). This is relevant to the current findings owing to the habitual nature of smoking cue-reactivity (Everitt et al, 2001). The caudate tail facilitates visual habits as this striatal subregion interacts with inferior temporal brain areas as part of the ‘visual’ corticostriatal loop, thus providing a ‘link in a visual pathway underlying oculomotor control and visuo-motor associations’ (Saint-Cyr et al, 1990, pg.1). Owing to this link, the caudate tail facilitates automatic responding toward environmental stimuli, which have been consistently rewarded (Yamamoto et al, 2013). The caudate tail most likely maintains the automatic and habitual responding to the smoking-related images used in the current study, as these images are ‘consistently rewarded visual stimuli’ owing to their repeated pairing with the reinforcing experience of nicotine. As subcortical shape measures are beginning to emerge as a robust endophenotype in neuropsychiatric disorders (Shaw et al, 2014), it is possible that individuals with the variability in striatal shape, we report, might be predisposed to form stronger visual habits. However, this possibility is speculative and requires additional testing.

Although the current work shows a clear link between striatal morphology and craving, additional research will aid in our understanding of how striatal morphology contributes to addiction and other habitual behaviors. For instance, future studies should consider different populations of smokers, as we restricted our sample to a relatively young, healthy population. Whether variation in striatal morphology continues to be associated with craving following brain changes associated with aging remains to be determined. Whereas our sample size is comparable to prior anatomical studies of nicotine dependence (Brody et al, 2004), research involving larger sample sizes may address some of these additional questions. In addition, the link between craving and striatal morphology prompts the question of whether variation in striatal morphology could be linked with differences in other ‘visual habits’ relevant to normal learning or other forms of craving and compulsive disorders. For example, enhanced striatal surface area has been associated with obsessive–compulsive disorder (OCD; Shaw et al, 2014), suggesting that larger striatal surface area may be a feature of compulsive disorders including OCD and addiction. Finally, our research focused only on the relationship between striatal morphology and cigarette craving as previous research has shown a clear link between stimulant abuse and enhanced striatal volume when substance abusers and healthy controls were compared (Das et al, 2011; Churchwell et al, 2012, Ersche et al, 2012). It is unclear whether variability in striatal morphology, in non-abusing populations also may explain individual differences in craving severity for other, more natural, reinforcers. Collectively, our findings indicate that striatal morphology is related to this brain regions’ role in craving and that morphological assessment of the striatum, beyond standard volumetric analysis, provides more nuanced information regarding the relationship between striatal morphology and function.

FUNDING AND DISCLOSURE

This work was supported by the National Institute on Drug Abuse Grant K01DA029645 (AJ). The other authors declare no conflict of interest.