Community based chronic tobacco smokers and non-smokers were recruited through a convenient-sampling approach across the South Eastern regions of Scotland between October 2019 and March 2020. Different methods of recruitment were employed, these included: Internet advertisements posted on Gumtree and Craiglist websites, flyers distributed at local businesses (e.g., supermarkets, leisure centers), advertisements published on “The Courier” regional newspaper, and word of mouth. All participants provided written informed consent prior to the beginning of the study. They were rewarded a total of £100 for their full participation. The recruitment flow chart is depicted in Supplementary Figure 1.

Enrolled participants needed to attend two experimental sessions. These sessions were conducted on separate days within the same week (no more than 3 days apart): The first experimental session involved screening procedures and computerized measures of cognitive impulsivity (impulsive choices and risky decision making). This experimental session was conducted at the University of St Andrews School of Medicine in St Andrews. The second session involved a Magnetic Resonance Imaging (MRI) procedure at the Clinical Research Center (CRC), Ninewells Hospital, Dundee.

Participants were screened for eligibility (see Supplementary Table 2) through objective and subjective measurements. Specifically, an exhaled Carbon Monoxide (CO) test was utilized to measure CO levels in participants' breath, while a saliva drug testing kit was employed to determine presence (or not) for cotinine. The presence of cannabis, cocaine, morphine, methadone, amphetamine, methamphetamine, and benzodiazepines was assessed through a urine drug screening test.

Chronic tobacco smokers who reported a recreational use of cannabis (no more than twice per week) mixed with tobacco in the same “spliff” or “joint,” and did not present symptoms of acute intoxication (e.g., conjunctival injection, slurred speech, agitation) at the screening session, were not excluded from the study. In fact, the effect of recreational cannabis use on cortical brain structures is limited (24), whereas the effect of cannabis use on subcortical structures may be better explained by concomitant tobacco smoking (25). Meta-analytic findings also revealed a small effect of cannabis use on cognitive impulsivity with an effect size of 0.30 (26). The Mini International Neuropsychiatric Instrument (MINI) version 7.0.2 (27) was utilized to exclude the presence of DSM V psychiatric disorders (Axis I) for participants. A screening interview was utilized to exclude the presence of chronic conditions (e.g., HIV, diabetes), pregnancy, neurological disorders, and severe head injuries. Participants' patterns of tobacco use, smoking variables (pack years, cigarettes smoked per day, age at regular smoking onset), and weekly use of cannabis and alcohol were assessed through a paper and pencil questionnaire, while severity of nicotine dependence was assessed by using the Fagerström Test for Nicotine Dependence (FTND) (28). Participants' pre- morbid IQ was estimated through the Barona equations (29).

Data obtained from computerized measures of cognitive impulsivity (ADT-5, CGT) were analyzed through analyses of covariance (ANCOVAs) to test the null hypothesis of no differences between chronic tobacco smokers and non-smokers in relation to impulsive choices and risky decision-making outcome measures while controlling for relevant sociodemographic variables such as sex, age, premorbid IQ, educational level, social deprivation (SIMD), and other substances used (cannabis, alcohol). To be analyzed through ANCOVAs, data needed to meet the assumptions of normality, homoscedasticity, homogeneity of variances, and of homogeneity of regression slopes. Data failing assumption of normality and/or homogeneity of variances were Log10 transformed. The non-parametric Kruskall-Wallis H test was employed if data still violated assumptions of normality and/or of homogeneity of variances after transformation. Bonferroni-corrected pairwise comparisons with a significance level set at p < 0.05 were utilized to control for Type-1 error. SPSS v. 26 (SPSS Inc., USA) was utilized for this part of the analysis. Effect sizes (Choen's f) were computed through the software G^*Power.

Neuroimaging data were analyzed by applying a whole-brain voxel-based morphometry (VBM) technique through SPM12 (https://www.fil.ion.ucl.ac.uk/spm/). T1 weighted images were segmented into gray and white matter probability maps, these were subsequently normalized to the Montreal Neurological Institute (MNI) template. Modulation was performed with Jacobian determinants to preserve the total amount of GM and WM in each probability map. Following segmentation and spatial normalization, images were smoothed with an 8 mm Gaussian kernel (32). Total intracranial volume was estimated by the Computational Anatomy Toolbox (http://www.neuro.uni-jena.de/cat/).

A two sample T-tests was utilized to assess for GM volume differences between chronic tobacco smokers and non-smokers across the whole brain, including a priori regions of interest (Supplementary Tables 1a, 1b). Whole-brain voxel-wise linear regression models were also computed to investigate the relationship between GM volume reductions and cognitive impulsivity outcome measures (ADT-5, CGT) in chronic tobacco smokers. Proof of concept analyses were conducted by computing whole-brain voxel-wise linear regression models investigating the relationship between GM volume reductions and tobacco exposure variables (cigarettes smoked per day, pack years, FTND, age at regular smoking onset).

Brain regions (including a priori regions of interest) were identified by converting MNI coordinates into Talairach coordinates (https://bioimagesuiteweb.github.io/webapp/mni2tal.html), and by inserting them into the Talairach Daemon Atlas (http://www.talairach.org/daemon.html) (16). A cluster forming significance threshold of p < 0.05 with a minimum of 100 contiguous voxels (k) per cluster was employed for two sample T-tests and for voxel-wise regression models testing the associations between GM volume and cognitive impulsivity measures. This threshold was obtained by applying Monte Carlo simulations (16, 33). For proof of concept analyses a more stringent threshold of p < 0.01 with a minimum of 100 contiguous voxels per cluster was employed. Covariates of no interest consisted in total intracranial volume (TIV), age, and biological sex for two sample T-tests and linear regression models. Results of whole-brain voxel-wise analyses are only reported in the text for a priori regions of interest.

Chronic tobacco smokers and non-smokers were well-matched for age, Scottish index of multiple deprivation (SIMD) scores, and units of alcohol consumed per day at the time of the first experimental session as assessed by independent samples t-tests (p > 0.05). Chi-square test results did also reveal the absence of a statistically significant association between sex and smoking status, χ(1)2 = 0.55, p = 0.46. However, chronic tobacco smokers displayed lower pre- morbid IQ and a lower educational level compared to non-smokers (p < 0.05). Sociodemographic and tobacco smoking characteristics of participants are displayed in Table 1. At the time of the second experimental session (MRI), chronic smoker participants were still matched for sex, age, SIMD, and units of alcohol consumed per day to non-smoker controls (Table 1). No significant changes in smoking characteristics were detected between the two sessions (paired samples t-test p > 0.05).

Voxel based morphometry (VBM) results revealed GM volume reductions in chronic tobacco smokers compared to non-smokers in several cortical and striatal brain regions across the whole brain, predominantly in the left hemisphere. Structures presenting GM volume reductions, cluster sizes, Brodmann's areas (BA), and corresponding Montreal Neurological Institute (MNI) coordinates are reported in Supplementary Table 3. Regions of Interest displayed GM volume reductions in frontal cortices such as the bilateral ACC (14, 21, 29;−11, 26, 18; BA32), left DLPFC (−21, 41, 23; BA9), left VLPFC (−26, 35,−12; BA47), and left OFC (−5, 44, −27; BA11). Striatal gray matter reductions in chronic smokers compared to non-smokers were present in the left caudate (−11, 2, 12), and in the left putamen (−21, −9, 6).

A positive relationship was detected between choice impulsivity scores (ED50 values) and GM volume in regions of interest. These included the left VLPFC (−50, 30, −18; BA47) (p = 0.002; T = 3.20; R² = 0.251), left caudate (−18, −2, 30) (p = 0.002; T = 3.21; R² = 0.336), right caudate (18, 0, 30) (p = 0.002; T = 3.37, R² = 0.281), and left ACC (−17, 42, 5, BA32) (p < 0.0001; T = 4.57, R² = 0.365) of chronic tobacco smokers. These associations are depicted in Figure 1. Furthermore, a negative relationship was detected between “risk taking” (p =0.001; T = 3.62; R² = 0.409) and “overall proportion bet” (p = 0.002; T = 3.34; R² = 0.371) scores of the CGT and GM volume in the left ACC (−9, 33, 21; BA 32) of chronic tobacco smokers (see Figure 2). Supplementary Tables 4, 5 depict the associations between risky decision making, impulsive choices and GM volume in brain regions of no interest (not related to the a priori hypotheses listed in the section Introduction).

Negative and positive relationships were also detected between GM volume in regions of interest and smoking variables. These are depicted in Table 2. Regression plots showing the directions of these associations are illustrated in Supplementary Figure 2. Supplementary Table 6 illustrates the associations between tobacco smoking variables and GM volume in brain regions of no interest.

These results are in line with recent findings demonstrating a strong association between chronic tobacco smoking, impulsive choices, and risky decision making (6, 7). That is, chronic tobacco smokers have a greater difficulty in delaying immediate gratification and display a riskier decision-making process in comparison to non- smokers. The current study adds up to the body of knowledge by revealing a cross- sectional relationship between impulsive choices, risky decision making, and GM volume reductions in chronic tobacco smokers not affected by psychiatric comorbidity, poly substance dependence (e.g., alcoholism), and/or chronic medical conditions (e.g., HIV). These deficits were detected in a relatively young sample of cigarettes smokers with a mean age of 28 years old. Thus, it is unlikely that participants were affected by neurocognitive impairments that are a consequence of the aging process (34).

Under a neuroanatomical point of view, the above findings are in line with previous studies revealing GM volume reductions in striatal and cortical structures located in the prefrontal cortex of chronic tobacco smokers (17–19). In accordance with these studies, the current research revealed a negative relationship between tobacco exposure variables and reduced GM volume in prefrontal and striatal brain regions. Specifically, longer pack years were associated to reduced GM volume in the left DLPFC, left ACC, left caudate, and left putamen. Severity of nicotine dependence (FTND score) was also associated with GM volume reductions in the left thalamus, an area strictly related to the development and maintenance of substance dependence for its dopaminergic projections (35). The current study provides support to the findings of Durazzo et al. (21) by revealing a correlation between GM volume reductions in the left ACC and risky decision making in chronic smokers. Furthermore, the current study revealed a positive association between GM volume reductions in the left ACC and the inability to delay gratification displayed by chronic smokers (ED50 values). Indeed, the ACC has been proposed to play a crucial role in the maintenance of addictive behaviors due to its connections with the limbic system and prefrontal cortex, therefore mediating maladaptive emotional and reward-based decisions (36). The association between GM volume reductions in the VLPFC and heightened impulsive choices displayed by chronic smokers extends previous work by endorsing disruptions in brain's executive system as a core feature of nicotine dependent individuals (36).

Notably, impairments in cognitive impulsivity for chronic smokers enrolled in the current study correlated to GM volume reductions localized predominantly in the left hemisphere of the brain. Despite GM deficits being reported frequently in left fronto-cortical and striatal brain regions in chronic tobacco smokers (19, 37, 38), no association between cognitive impairments and lateralization of structural brain deficits in chronic smokers was ever investigated. However, an fMRI study conducted by Clewett et al. (39) revealed greater functional coupling between the left fronto- parietal network and the left insular cortex in chronic tobacco smokers compared to non-smokers while performing a computerized Delay Discounting task (with hypothetical monetary rewards) (39). Greater functional connectivity between these areas was also associated to steeper discounting rates (39). Furthermore, a systematic review conducted by Gordon (40) showed greater peak activation in fronto-cortical brain areas located in the left hemisphere of tobacco deprived smokers while responding to cue-reactivity (craving) stimuli during fMRI investigations (40). Therefore, consistently with the CNDS model, the simultaneous hyperactivation of the left reward system to cigarettes cues, and the inability to delay gratification caused by GM volume reductions in the left VLPFC and left ACC, may induce cigarettes smokers to crave for, and to want immediately the drug of abuse (i.e. cigarettes) despite the known health risks associated to its usage. Indeed, impulsivity has been linked to tobacco craving and smoking relapses by the literature (41).

Even though the cross-sectional nature of the current research does not allow to directly imply causation, the above findings may be explained by neuroscience- based paradigms. According to the “tobacco Induced neurotoxicity theory of adolescent cognitive development” (TINACD) (42), exposure to tobacco during adolescence (a developmental period characterized by intense neurostructural and neurochemical maturation) may cause long-lasting deficits in frontal brain areas modulating cognitive functions mostly related to top- down inhibitory control and decision making (42). For this reason, adult smokers who initiated tobacco use during adolescence may have more difficulties quitting smoking, and relapse more frequently, in comparison to individuals who started smoking at a later developmental stage (42). This theory has been availed by animal models revealing unique, and long- lasting, cellular alterations and structural changes in fronto-cortical and striatal brain regions of rats exposed to nicotine during adolescence (43, 44). In support to the TINACD, the current study revealed a positive correlation between younger age at regular smoking onset (mean = 16 years) and GM volume reductions in the left VLPFC (BA 47), (Table 2). As described previously, GM volume reductions in the left VLPFC (BA 47) were also associated to heightened impulsive choices in the current sample (Figure 1).

An alternative interpretation to the above findings consists in the presence neurocognitive endophenotypes in substance dependent populations (45). In fact, studies have shown abnormal brain structures in fronto-cortical and striatal brain regions, and related cognitive impulsivity impairments, in substance dependent individuals and their drug- naïve biological siblings (46, 47). However, because of the lack of longitudinal studies, the causal relationship between chronic tobacco smoking and neurocognitive impairments remains unclear (48).

Considering that impaired cognitive impulsivity may be determinant in fostering tobacco smoking during different stages of the drug addiction cycle (initiation, maintenance, relapse) (8, 11), results from this study may inform smoking cessation treatments, and neuroscience-based interventions for tobacco use and dependence. Specifically, the identification of GM volume reductions in brain areas related to different aspects of cognitive impulsivity (ACC to risky decision making, VLPFC, ACC, and caudate to impulsive choices), and the proposed left lateralization of these neurocognitive impairments, may aid the improvement of technology-based neuromodulation interventions such as transcranial direct current stimulation (tDCS) and transcranial magnetic stimulation (rTMS). In fact, such interventions have provided limited and contrasting results for the treatment of nicotine dependence (49).

Moreover, the inability to delay gratification, commonly featured by chronic smokers (7), should be considered a primary target for cognitive rehabilitation therapies (CRTs) (50). Studies have also proposed Episodic Future Thinking (ETF) interventions as effective means to decrease delay discounting rates, cigarette self- administration, and cigarette demand (51).

The current study should be considered in light of several limitations. First, the size of the sample enrolled in the current study was relatively small, therefore limiting the power of risky decision making (CGT) results. However, post-hoc power calculations for choice impulsivity (ADT-5) results revealed an achieved power of 0.94 with a very large effect size of 0.5. Additionally, participants were all recruited from the same deprived metropolitan area, thus generalizability may be limited to Scottish chronic smokers from a low socio-economic background (SES). Studies have proposed a relationship between low SES, brain development deficits, and cognitive impairments (52). Although, chronic tobacco smoking has been recently found to mediate these associations (53). Another limitation consists in the utilization of self- reported questionnaires to assess alcohol consumption patterns and smoking characteristics of participants (pack years, cigarettes smoked per day, age at regular smoking onset) as participants may have provided inaccurate information. Nonetheless, a rigorous screening procedure involving different objective measurements to assess smoking status (exhaled CO, salivary cotinine) and to exclude the presence of other substances in participants' system (urine analysis) was employed. The recruitment of a younger chronic smoker sample not affected by polysubstance dependence and characterized by minimal alcohol and cannabis use, with the utilization of stringent exclusion and inclusion criteria, may be considered strengths of the current study.

The current study revealed an association between left fronto-cortical and striatal GM volume reductions and impaired cognitive impulsivity in chronic tobacco smokers. GM volume reductions in the left VLPFC correlated to heightened impulsive choices and to younger age at regular smoking onset. Considering that the cross-sectional nature of the current research limits inferences of causal effects, longitudinal studies would be required to elucidate if these neurocognitive impairments are representations of pre-morbid endophenotypes or are caused by tobacco exposure on the developing adolescent brain.