Nicotine Patch Alters Patterns of Cigarette Smoking-Induced Dopamine Release: Patterns Relate to Biomarkers Associated With Treatment Response

Abstract Introduction Tobacco smoking is a major public health burden. The first-line pharmacological treatment for tobacco smoking is nicotine replacement therapy (eg, the nicotine patch (NIC)). Nicotine acts on nicotinic-acetylcholine receptors on dopamine terminals to release dopamine in the ventral and dorsal striatum encoding reward and habit formation, respectively. Aims and Methods To better understand treatment efficacy, a naturalistic experimental design combined with a kinetic model designed to characterize smoking-induced dopamine release in vivo was used. Thirty-five tobacco smokers (16 female) wore a NIC (21 mg, daily) for 1-week and a placebo patch (PBO) for 1-week in a randomized, counter-balanced order. Following 1-week under NIC and then overnight abstinence, smokers participated in a 90-minute [11C]raclopride positron emission tomography scan and smoked a cigarette while in the scanner. Identical procedures were followed for the PBO scan. A time-varying kinetic model was used at the voxel level to model transient dopamine release peaking instantaneously at the start of the stimulus and decaying exponentially. Magnitude and spatial extent of dopamine release were estimated. Smokers were subcategorized by nicotine dependence level and nicotine metabolism rate. Results Dopamine release magnitude was enhanced by NIC in ventral striatum and diminished by NIC in dorsal striatum. More-dependent smokers activated more voxels than the less-dependent smokers under both conditions. Under PBO, fast metabolizers activated more voxels in ventral striatum and fewer voxels in dorsal striatum compared to slow metabolizers. Conclusions These findings demonstrate that the model captured a pattern of transient dopamine responses to cigarette smoking which may be different across smoker subgroup categorizations. Implications This is the first study to show that NIC alters highly localized patterns of cigarette smoking-induced dopamine release and that levels of nicotine dependence and nicotine clearance rate contribute to these alterations. This current work included a homogeneous subject sample with regards to demographic and smoking variables, as well as a highly sensitive model capable of detecting significant acute dopamine transients. The findings of this study add support to the recent identification of biomarkers for predicting the effect of nicotine replacement therapies on dopamine function which could help refine clinical practice for smoking cessation.


INTRODUCTION
Tobacco smoking is the world's leading cause of preventable death. It is largely driven by the reinforcing effects of nicotine-the primary addictive chemical in cigaretteswhich activates nicotinic acetylcholine receptors located on dopamine neurons to release dopamine in mesolimbic brain regions such as the striatum which encodes reward 1,2 .
Abstinence is hard to sustain. All available treatments for smoking cessation have limited success rates 3 and most smokers relapse within 6 months of a quit attempt 4 . One of the most widely-used smoking cessation treatments is the transdermal nicotine patch (NIC). NIC is a type of nicotine replacement therapy which releases a steady, low dose of nicotine that is absorbed through the skin. Just like cigarette smoking, NIC releases dopamine in reward regions of the brain. There is evidence that nicotine replacement therapies reduce cigarette use 5 . It may be possible that NIC alters reward via its own effect on dopamine to reduce cigarette use, but smoking-induced dopamine release in the striatum during NIC treatment has not been thoroughly investigated. Certain traits, such as nicotine dependence level and nicotine clearance rate, have been identified as predictors of treatment efficacy. Higher nicotine consumption level is strongly associated with higher nicotine dependence and both are associated with poorer treatment outcomes 6,7 . Nicotine metabolism ratio (NMR) has recently been identified as a biomarker for predicting the success of smoking cessation with NIC compared to varenicline, an FDA-approved smoking cessation treatment that modulates dopamine release 8, 9 .
Several positron emission tomography (PET) studies have examined the dopamine response in vivo in tobacco smokers [10][11][12] . These studies used amphetamine as a teststimulus which releases robust amounts of dopamine via several neurobiological mechanisms. Unlike amphetamine, nicotine or cigarettes release much smaller amounts of dopamine lasting only minutes 13 . Some studies have investigated smoking-induced dopamine-release 14-18 and a few other studies have assessed the smoking-induced dopamine response to cessation treatments [19][20][21] . However, the results of these studies have A c c e p t e d M a n u s c r i p t 5 been inconsistent. The inconsistencies may be attributed to: (1) an inability of conventional tracer kinetic models to detect small and short-lived dopamine responses, (2) a weakness of nicotine as a test-stimulus and/or, (3) an excessive delay between the stimulus and the PET scan. First, studies that use conventional time-invariant tracer kinetic models, such as SRTM (simplified reference tissue model) 22 , measure the average smoking-induced changes in dopamine binding over the scan duration. The parametric endpoint used in these studies, BP ND (receptor availability), is a steady-state parameter that fails to accurately capture the small and transient (on the order of minutes) alterations in the dopamine system elicited by cigarette smoking 23 . Second, many routes of nicotine delivery including nicotine gum, nicotine patch, IV nicotine, and cigarette smoking have been tried. Administering nicotine in unnatural ways (i.e., other than cigarette smoking) may not produce a sufficiently robust dopamine response. Third, the dopamine response to cigarette smoking is brief so PET scanning too long after smoking may simply miss the response. In the present study, we used an appropriate model, a naturalistic stimulus, and a rigorous experimental design.
The current study is the first double-blind crossover design, including two PET scans per subject-one following 1-week on NIC and one following 1-week on a placebo patch (PBO)-to assess the effect of NIC on the striatal cigarette-induced dopamine response, at the voxel resolution. We evaluated whether nicotine dependence level (assessed as smoking pack-years), and nicotine clearance rate (assessed via NMR), affected NIC-induced changes to the transient dopamine response. We used motion-tracking technology and listmode reconstruction to correct any intra-frame motion that might have occurred while subjects smoked inside the scanner during the scan 24,25 . We also used the time-varying kinetic model LSRRM (linearized simplified reference region model) 26 , which models transient dopamine release peaking instantaneously at the start of the stimulus and decaying exponentially. Our two outcome measures were magnitude of dopamine release, and spatial extent of dopamine release in the precommissural striatum 27 . We expected that the presence of NIC would alter the magnitude and location of the dopamine response to Downloaded from https://academic.oup.com/ntr/advance-article/doi/10.1093/ntr/ntac026/6518132 by guest on 08 February 2022 A c c e p t e d M a n u s c r i p t 6 smoking, based on previous microdialysis and PET literature showing that repeated nicotine injections increase dopamine release in mesolimbic brain regions 1,28,29 . We also expected that the putative biomarkers for treatment success (dependence level and nicotine metabolism rate) would affect the magnitude or spatial extent of the dopamine response, based on prior literature demonstrating the dopamine release was related to nicotine dependence 20 and nicotine metabolism 30,31 .

Subjects
Thirty-five tobacco smokers (16 female) were studied. Subjects had no history or evidence of significant medical disorders on physical exam and did not meet DSM-5 criteria for current or past psychiatric or substance abuse diagnosis (except nicotine dependence  Table 1). On intake day, smoking status was confirmed by spirometry to measure carbon monoxide (CO) levels >11 parts per million (ppm) and by urine samples to measure cotinine-the primary metabolite of nicotine-levels >150 ng/ml (NicAlert cotinine test strips; Nymox Pharmaceutical). On scan day, overnight abstinence was confirmed by CO levels <10 ppm or ≤50% of their intake level. Pre-scan plasma nicotine and metabolites (cotinine and 3-hydroxy-cotinine) were collected. Pregnancy and lactation were exclusionary.
Menstrual cycle phase was not controlled and use of hormonal contraception was not exclusionary.

Study Design
The study was approved by the Yale Human Investigation and Radiation Safety Committees.
Subjects wore a nicotine patch (21mg, daily) for 1-week and a placebo patch for 1-week in a double-blind, randomized, counterbalanced, placebo-controlled, cross-over design.
A c c e p t e d M a n u s c r i p t 7 Following 1-week on NIC and then overnight abstinence, smokers participated in a 90minute [ 11 C]raclopride PET scan while continuing to wear the NIC/PBO from the previous day. While lying in the scanner during continuous scanning they smoked a cigarette, as previously described 24 . Briefly, subjects smoked one cigarette of their own brand, with their dominant hand, at their own pace (typically 3-min to complete a whole cigarette), starting at mid-session (35- into low and high pack-years groups using a median split to assess the effects of nicotine dependence on smoking-induced dopamine release. Low and high pack-years smokers were matched for age, sex, MNWQ, and QSU scores (p>0.05). High pack-years smokers were significantly older, smoked more cigarettes per day, smoked for more years and had higher FTND scores (p<0.02) than the low pack-years smokers. We chose to use packyears as a measure of dependence rather than FTCD for the following reasons: (1) FTCD has poor reliability and validity possibly due to dichotomous scoring 32 , (2) FTCD does not account for duration of smoking which is an important factor in level of dependence, and (3) FTCD would not be appropriate for the current median split analysis because it is a categorical variable and thus, the resulting groups would not be as clinically meaningful.
2.3.2 NMR: NMR was calculated as the ratio between 3-hydoxy-cotinine and cotinine. Participants were divided into slow and fast NMR groups using the clinically-A c c e p t e d M a n u s c r i p t 8 established cutoff NMR ratio of 0.31, based on 8 to assess the effects of nicotine metabolism on smoking-induced dopamine release. Fast metabolizers had significantly higher NMR (p<<0.01) than slow metabolizers. Slow and fast metabolizers were matched for age, sex, and all other smoking characteristics (p>0.2).

Demographic data statistical analysis
A chi-squared test was used to evaluate group differences in the categorical variable (sex).
Student's t-tests were used to evaluate group differences in continuous variables such as basic demographics (age), smoking questionnaires (e.g., FTND), and smoking measures (e.g., cigarettes smoked per day) between conditions and between smoker subgroups.

Imaging data acquisition
A 3T structural MRI scan for anatomical localization was collected from each subject (Trio and Prisma, Siemens Medical Systems, Erlangen, Germany). [ 11 C]raclopride, a D 2/3 antagonist, was synthesized as previously described in 13 . Before each PET scan, a 6-min transmission scan was acquired for attenuation correction. Injected activity was compared between conditions and between smoker subgroups using student's t-tests. To determine the effect of mass on the D 2 receptor, we used the highest A c c e p t e d M a n u s c r i p t 9 2.6 Image pre-processing and post-processing PET data were reconstructed using MOLAR 37 , motion-corrected using Vicra (Polaris Vicra Tracking System; Northern Digital), and de-noised with a 3 x 3 x 3 voxel HYPR filter 38,39 , according to the methods previously described 24,39 . A 3D Gaussian filter ( =2 voxels) and was applied. PET data were aligned to the subject's MRI and then spatially normalized to a standard MNI-152 template. A standardized 1004-voxel mask delineating the precommissural striatum and its sub-regions (left and right, ventral striatum (VS), dorsal caudate (DC), dorsal putamen (DP)), based on Martinez et al. 27 , was applied to all subject images in template space prior to analysis.
The LSRRM model 26   (multilinear reference tissue model) 42 , using the Akaike information criterion corrected for small data sets 43 . LSRRM estimates 4 parameters; MRTM (nested within LSRRM) estimates 3. A cluster-size threshold (CST=15 voxels) was applied to all significant voxels to correct for multiple comparisons 24,41 . For a lengthy discussion of CST to correct for multiple comparisons, see paper by our group 41 as well as others 44-46 . A c c e p t e d M a n u s c r i p t  (2) ‗Spatial Extent of Dopamine Release' Images. Images were produced by summing the binary masks for all subjects and dividing by the number of subjects, for each condition.
‗Spatial Extent of Dopamine Release' induced by smoking was defined as the number of voxels activated by smoking and compared between groups and conditions, by sub-region.
All 35 subjects were used to create ‗Spatial Extent of Dopamine Release' images. ‗Spatial Extent of Dopamine Release' images were also created for low and high pack-years groups, and slow and fast metabolizer groups, for each condition. Statistical significance in the mean number of voxels activated by smoking in each subregion of the mask was assessed with a permutation test (two-tailed, p<0.05, Bonferroni corrected by 1004 voxels in the mask), as previously described 24 . For each permutation test, one hundred thousand random resamplings scans were performed to achieve a zero mean difference between conditions in each subregion. Each set of re-samplings was used to create a null distribution for 2 arbitrary groups reflecting the actual sizes of the two experimental cohorts. In order to assess the likelihood of any given difference occurring by chance, we tested whether the mean value of the difference between PBO and NIC conditions was different from the zero (the null distribution) for each subregion. The null hypothesis of all permutations tests was that the mean difference in number of smoking-induced activated voxels between conditions was not different from zero, in a given subregion.

Steady-state parameters
Parametric images of steady state parameters, R 1 and BP ss , are available from LSRRM.   Figure S4 illustrates the process of removing the order effect from the images comparing NIC and PBO conditions (see Supplemental Material).

Subjects
Twenty-eight (excluding outlier and BLD) smokers (13 female) were included in the final analysis. The exclusion of the outlier and BLD subjects did not change demographic or smoking measure data (p>0.5), nor did it change any findings of significance between subgroups (Table S1).

Smoking Characteristics
On average, subjects smoked less cigarettes per day during the week they were on NIC (10.9±1, p=0.002) and PBO (11.2±1, p<0.001) patches relative to baseline (at intake; 14.4±1), demonstrating compliance with, and possibly effectiveness of, the patch protocol.

Blood measures
Plasma nicotine and cotinine levels were higher under NIC than PBO condition (p<0.02), demonstrating compliance with the patch usage protocol (  Craving scores decreased whereas enjoyment and energy scores increased following cigarette smoking, regardless of condition ( Figure S1) or scan order (Table S3). Excluding outlier and BLD subjects did not impact average subjective ratings.

Injection parameters
On average, there were no significant differences in injected radiotracer activity between low and high pack-years, fast and slow metabolizers, first and second scans, or NIC and PBO scans (

Compliance with protocol and summary of findings
This is the first double-blind, randomized, placebo-controlled, cross-over study to examine cigarette-induced dopamine release following short-term nicotine replacement therapy.
Smokers complied with the medication regimen, as evidenced by higher blood nicotine and cotinine levels under NIC condition compared to PBO and reduced smoking during the patch protocols relative to baseline, consistent with prior literature 5 . In-scanner cigarette smoking decreased craving and increased enjoyment and energy ratings, as expected. Because subjective ratings of craving, enjoyment and energy were not different between conditions, we did not predict that they would be related to brain changes. Subgroups and conditions were well-matched on injection parameters. This study has three main findings: (1)

Removal of the scan order effect
We found a significant effect of scan order on the magnitude of dopamine release, such that greater cigarette-induced dopamine release was observed in the first scan session relative to the second, regardless of condition. This effect may be related to the novelty of the situation (i.e., smoking a cigarette in the scanner) on the first scan 49,50 . While this is an unintended effect of study design, the effect of order was successfully removed as evidenced by the non-significant difference in the magnitude of dopamine release between individuals with a NIC scan first relative to a PBO scan first ( Figure S4).

Spatially varying effect of NIC on magnitude of cigarette-induced dopamine release
The ability to capture subtle fluctuations in endogenous dopamine concentration within a single-scan is novel, and possible thanks to sophisticated modeling techniques.  52 or no group differences in smoking-induced dopamine release, e.g., 21 . The parameter presented herein, is analogous to BP ND in the literature because it evaluates the change in the level of dopamine binding. However, is sensitive to short-lived changes in dopamine binding and is not dependent on scan duration.
Thus, this endpoint has allowed us to capture and characterize nuances in the change in magnitude of the dopamine response caused by NIC, at voxel resolution.
We found that the magnitude of cigarette-induced dopamine release was greater under NIC compared to PBO condition in ventral striatum, suggesting an additive reinforcing effect of nicotine on the brain's dopamine system. This is consistent with previous microdialysis and PET literature showing that a nicotine challenge following repeated daily injections increases dopamine releases in a dose-dependent manner 1,28,29 , regardless of the effects of treatment 15 . Rollema and colleagues theorized that despite chronically elevated nicotine levels due to NIC, cigarette smoking still causes steep increases in nicotine levels and in DA release, thus, maintaining its reinforcing effect 53 . This additive rewarding effect may be critical for driving smoking behavior and may decrease the likelihood of successful cessation. It is also important to note that subjects were overnight abstinent. Thus, the cigarette smoked in the scanner was the first cigarette of the day, which is the most pleasurable 54 . Further, we found sub-regional variation in the effect of NIC on smokinginduced dopamine release consistent with previous reports. One study found that continuous microinjections of nicotine into the rat brain both increased and decreased [ 11 C]raclopride dopamine release in the striatum, suggesting that local extracellular dopamine levels may  55 . An fMRI study found that NIC increased smoking cue-induced activation in the caudate while decreasing putamen activation relative to PBO 56 , supporting our findings.
Heterogeneity in the spatial pattern of pharmacologically-induced dopamine release has also been reported for other drugs. For example, Yoder et al. 57 found that individual dopamine responses to alcohol in the striatum were highly localized and varied widely by subject.
Taken together, we can deduce that the pattern of cigarette-induced dopamine release is a complicated, highly localized phenomenon such that NIC treatment may not necessarily affect the striatum homogeneously, or even in a single direction.
4.5 Differences in spatial pattern of brain response to smoking by dependence Nicotine dependence level was related to cigarette-induced dopamine release. Moredependent smokers activated more voxels than less-dependent smokers regardless of the condition, suggesting that the longer people engage in their addictive behavior, the more of the striatum they must recruit to process rewarding and enjoyable stimuli. This is consistent with prior literature, e.g., 20 . Animal studies have demonstrated that repeated nicotine administration enhances psychomotor responses, the rewarding effects of nicotine, and striatal dopamine release in response to nicotine 28 . In humans, nicotine gum-induced dopamine release in the ventral striatum was shown to be positively correlated with the degree of nicotine dependence 20 .
Our study also showed that more-dependent smokers activated more voxels in dorsolateral as opposed to ventral regions of the striatum than less-dependent smokers, suggesting a migration of dopamine activation from the goal-directed to the habit-formation striatum over time 24,58 . We believe our finding links previous work in cocaine-dependent animals to nicotine-dependent humans who are dependent on nicotine for years as opposed to

Limitations
This study included a number of limitations that were primarily related to PET scanning and analyses.

Examination of Sex Differences
One of the goals of this study was to examine sex differences. We found that the overall pattern of was similar for male and female subjects when both NIC and PBO conditions are combined. However, due to small cell sizes, we were underpowered to examine sex differences in dopamine magnitude and spatial extent by individual NIC and PBO conditions.

Model limitations
LSRRM models the dopamine response as starting (and peaking) at the time of stimulus and decaying exponentially thereafter. The assumption of an instantaneously-peaking response may be too restrictive for our data. Microdialysis experiments suggest that the peak magnitude of the dopamine response following a pharmacological stimulus could have a latency period, e.g., 53 . The true range of start-times and peak-times for the dopamine A c c e p t e d M a n u s c r i p t 21 response is unknown and may vary based on treatment and subject. Our group has developed a suite of time-varying ‗ntPET' models that are specifically tailored to characterize the transient changes in neurotransmitter binding caused by a drug stimulus, e.g., 61 . We have used the lp-ntPET model e.g., 41 , to not only detect, but also characterize brief smokinginduced dopamine transients at the voxel level-revealing novel sex differences in the spatiotemporal signature of the dopamine response 24

CONCLUSIONS AND IMPLICATIONS
This is the first study to show that nicotine patch alters highly localized patterns of cigarette smoking-induced dopamine release and that levels of nicotine dependence and nicotine clearance rate contribute to these alterations. This current work included a homogeneous subject sample with regards to demographic and smoking variables, an absence of comorbidities, and subject compliance with treatment, as well as a highly sensitive model capable of detecting significant acute dopamine transients. The findings of this study add support to recent identification of biomarkers for predicting the effect of nicotine replacement therapies on dopamine function.
A c c e p t e d M a n u s c r i p t 23

Funding
The study was supported by the National Institutes of Health R01DA038709 (EDM). YZ was supported by the SCORE Career Enhancement Core Program (U54AA027989).

Declaration of Interests
The authors declare no conflict of interest.

Data and code availability statements
The data used in this study are not openly available at this time because the study is ongoing. Software and code are available upon request.   and NIC conditions across all subjects (n=25). Dopamine release was increased by NIC compared to PBO in the VS. Dopamine release was decreased by NIC compared to PBO in the DC and DP. Regions are labeled in a representative slice. Right side of the brain is on the right.