Cue reactivity and craving are important aspects of addictions that involve affect, conditional learning, and attentional biases (Elton et al., 2021, Franken, 2003, O'Brien et al., 1998, Robinson and Berridge, 1993, Welberg, 2013). Cue reactivity and craving are of clinical relevance across drug and behavioral addictions, relating prospectively to recovery and relapse (Allenby et al., 2020, Antons et al., 2020, Epstein et al., 2010, Hormes, 2017).

Gambling disorder (GD), classified as a behavioral addiction in the DSM-5 (American_Psychiatric_Association, 2013) and ICD-11 (World_Health_Organization, 2019), is a significant public health concern that shares similarities with and exhibits differences from drug addictions like cocaine use disorder (CUD) (Potenza et al., 2019). Preliminary data from video-tape-based cue-elicitation fMRI tasks comparing neural correlates of gambling urges in GD and cocaine cravings in CUD have suggested shared neural contributions in the posterior cingulate cortex (PCC), precuneus, thalamus, ventral striatum, lingual gyrus, and insula, and group-specific activity in the inferior parietal and anterior cingulate cortices (ACC) (Potenza, 2008). A subsequent larger study suggested shared neural correlates of cue reactivity and craving in the medial prefrontal cortex (mPFC) and ACC for relevant cues in a disorder-specific fashion (Kober et al., 2016). However, diagnostic-group-related neural correlates in the ventromedial and dorsomedial prefrontal cortices were also suggested, as were gender-related differences in insula activation to gambling cues in individuals with GD (Kober et al., 2016). While differences exist in the neurobiologies of GD and CUD, it has been suggested that features such as reduced frontostriatal activation in response to reward anticipation may be shared between GD and substance use disorders (SUDs) (Balodis & Potenza, 2020), as well as related conditions (Mestre-Bach and Potenza, 2023).

To date, most studies of the neural correlates of cue reactivity and urges/cravings for gambling and cocaine have largely focused on activation. However, functional connectivity analyses may provide complementary and unique insights into the neurobiological nature of cue reactivity and craving processes. Activation-related circuit-based findings have suggested that individuals with GD and CUD may differentially engage circuits during processes such as loss-chasing (Qiu and Wang, 2021, Worhunsky et al., 2017), emotion regulation (Picó-Pérez et al., 2022), and processing of near-miss events (Worhunsky et al., 2014). Additionally, resting-state functional connectivity studies have found increases in anterior default mode network (DMN) connectivity, decreases in posterior DMN connectivity, and overall less functional interaction for DMN with limbic and executive networks (Zhang & Volkow, 2019). These findings suggest that disorder-related similarities and differences in functional connectivity during cue responsiveness may provide additional insight into neurocognitive features of addictions beyond what region-specific activation can reveal, and help inform treatment-development-related efforts (Antons et al., 2020, García-Castro et al., 2022, Sutherland et al., 2012).

Connectivity approaches complement regional activity analyses, with the latter providing particular insight into functional networks (or connections within networks) that may more closely align with how the brain operates and links to motivations and behaviors. Traditional seed-based approaches to functional connectivity rely on a priori hypotheses for seed selection, informed by theories arising from the previous literature. This approach is potentially biased by activation-related results, thus limiting the extent to which novel insights may be derived. Data-driven approaches (that may also involve hypotheses but may have fewer restrictions) have the potential to identify novel results not constrained to a priori hypotheses. The intrinsic connectivity distribution (ICD) is a data-driven approach that can assess the degree to which each voxel in the brain is correlated with every other voxel (Scheinost et al., 2012). From a functional standpoint, this metric may be especially sensitive to state-related changes in regional dynamics, reflecting shifts in the broader functional networks in which regions are interacting, and within which they are situated (Salehi et al., 2020). For example, a recent study on smoking and non-smoking compared activation approaches, traditional functional connectivity approaches, and ICD methods. The ICD analyses found unique differences in corticolimbic connectivity that were specific to the current state of nicotine satiety (Yip et al., 2022) Additionally, in a study of youth prenatally exposed to cocaine, a group-by-condition interaction implicated the PCC, with post-hoc analyses showing that the drug-exposed versus non-drug-exposed group showed less PCC connectivity specifically during a stress condition (Zakiniaeiz et al., 2017). These results additionally highlight the link between the ICD approach and inferences about the neural dynamics behind the neurocognitive states important for understanding addictive disorders. Insights into cue reactivity and craving, and how the states that facilitate them are similar and different across diagnoses, could shed valuable insights into addictions.

Although prior ICD studies have been used to investigate functional connectivity in SUDs related to cocaine (Mitchell et al., 2013), tobacco (Garrison et al., 2016, Yip et al., 2022) and alcohol (Zakiniaeiz et al., 2016), as well as with internet gaming disorder (Liu et al., 2021), it has yet to be applied to GD. In this study, we used ICD to examine data from a sample of individuals with GD, CUD and HC subjects as previously reported (Kober et al., 2016). The previous activation-based analyses found increased activity in the ACC and mPFC in response to each group’s disorder-relevant cues. However, these midline nodes have dynamic relationships within broader functional networks, and the regional activity level may not elucidate the entire picture of the neurocognitive processes behind brain-related cue reactivity (BRCR). A recent study using connectome-based predictive modeling on a subset of these data investigated functional canonical networks, such as executive networks and the DMN, implicating regions and network interactions in cocaine cravings and gambling urges (Antons et al., 2023). These whole-brain results suggest that the connectivity of midline DMN nodes may contribute to craving across diagnoses, and our current analyses aim to complement previous work by investigating functional connectivity dynamics using ICD. We hypothesized that the connectivity of midline DMN nodes, such as the PCC and mPFC, would be decreased in BRCR across addictive disorders in a disorder-and-cue-specific fashion. We also hypothesized increased whole-brain functional connectivity in prefrontal regions in healthy versus addicted groups, based on findings relating to top-down control over cue-related reactivity (Albein-Urios et al., 2014, Kober et al., 2010).