Sex and sobriety: Human brain structure and function in AUD abstinence

Abstract

Women are drinking alcohol as much as men for the first time in history. Women experience more health-related consequences from alcohol use disorder (AUD), like increased prevalence of alcohol-related cancers, faster progression of alcohol-related liver disease, and greater risk for relapse compared to men. Thus, sex differences in chronic alcohol use pose a substantial public health problem. Despite these evident sex differences, our understanding of how these differences present during alcohol abstinence is limited. Investigations of brain structure and function are therefore critical for disen-tangling factors that lead to sex differences in AUD abstinence. This review will discuss current human neuroimaging data on sex differences in alcohol abstinence, focusing on structural and functional brain measures. Current structural imaging literature reveals that abstinent men have smaller gray and white matter volume and weaker structural connectivity compared to control men. Interestingly, abstinent women do not show differences in brain structure when compared to controls; instead, abstinent women show a relation between alcohol use and decreased measures of brain structure. Current functional brain studies reveal that abstinent men exhibit greater brain activation and stronger task-based functional connectivity to aversive stimuli than control men, while abstinent women exhibit lesser brain activation and weaker task-based functional connectivity than control women. Together, the current literature suggests that sex differences persist well into alcohol abstinence and impact brain structure and function differently. Understanding how men and women differ during alcohol abstinence can improve our understanding of sex-specific effects of alcohol, which will be critical to augment treatment methods to better serve women.

Introduction

Alcohol use disorder (AUD) is the most common substance use disorder in the US, contributing to substantial mental and physical health issues (Bouchery, Harwood, Sacks, Simon, & Brewer, 2011). Rates of AUD continually rise; since 2019, rates of AUD have doubled to 30 million people (SAMHSA, 2022). Notably, there are sex differences in these trends; rates have fallen in men but continue to rise in women (Fig. 1). These trends extend beyond drinking behaviors—recent meta-analyses show that chronic alcohol drinking in women is associated with increased risk of death at lower alcohol concentrations, as well as a 35% increase in alcohol-related deaths (e.g., heart attack, stroke), compared to men (Esser, 2024; J. Zhao, Stockwell, et al., 2023). The discrepancy between men and women is a longstanding issue in both mental health and substance use disorders: women are more susceptible to anxiety disorders and also more susceptible to relapse in AUD (Cohen, Feinn, Arias, & Kranzler, 2007; Crum et al., 2018; Kessler et al., 1994). Women also experience worse health consequences from chronic alcohol use compared to men. For example, women have higher exposure to toxic alcohol metabolites, higher prevalence of alcohol-related cancers, and faster progression towards alcohol-related liver disease (Agabio, Campesi, Pisanu, Gessa, & Franconi, 2016; Loft, Olesen, & Døssing, 1987; Rehm et al., 2010). Even though women have increasing rates of AUD and are more likely to relapse, the underlying mechanisms are largely unknown.

Fig. 1. Percent of US adults who engage in binge drinking or heavy drinking in the past month.

Percent of past month binge drinking in adults (18+) in the US (left); percent of past month heavy drinking in adults (18+) in the US (right). Linear regressions show a significant time × sex interaction in past month binge drinking (p = 0.0006) and past month heavy drinking (p = 0.001) (assessed from 2015 to 2022, as binge and heavy drinking criteria changed in 2015). Specifically, past month binge drinking is significantly decreasing over time for men (p < 0.001), but not for women (p = 0.08); past month heavy drinking is significantly decreasing over time for men (p < 0.001), with a non-significant increase in women (p = 0.16). Data from the NSDUH data set (2013–2022).

Limited studies have investigated sex differences in AUD. To date, most research has primarily studied men, largely due to barriers to conducting sex-differences research in AUD. These barriers include: (1) Many previous studies have recruited participants from inpatient treatment units, residential treatment programs, or specific hospitals (state hospitals, Veterans Affairs), which have historically served mainly men and this is exacerbated by the fact that women are less likely to seek treatment, and less likely to be in residential treatment programs (Gilbert, Pro, Zemore, Mulia, & Brown, 2019); (2) In studies of treatment-seeking or -receiving populations, women are more likely to be medicated, which is often an exclusion criteria for investigations of biological or neural mechanisms of AUD (Orlando et al., 2020); (3) Men have had higher drinking rates than women, resulting in a natural selection of men for research studies (See Fig. 1); (4) A longstanding dogma in science has considered female animals/human women as ‘complex’ and generating ‘messy data’—but this dogma is no longer supported and major funding sources now encourage or require the inclusion of both sexes (Becker, Prendergast, & Liang, 2016; McCarthy, Woolley, & Arnold, 2017; Shansky & Woolley, 2016). These limitations, unfortunately, have created a gap in our understanding of sex differences. As women begin to drink as much as men, we are now recognizing the disproportionate effects alcohol has on women.

Despite these barriers, several recent reviews outline the extensive sex differences that exist in AUD. Previous reviews have focused on sex differences in current drinking patterns (e.g., Verplaetse, Cosgrove, Tanabe, & McKee, 2021), stress as a pathway to development of AUD in women (e.g., Guinle & Sinha, 2020; Peltier et al., 2019), and sex differences in neurobiology and cognition (e.g., Fama, Le Berre, & Sullivan, 2020; Flores-Bonilla & Richardson, 2020; Nixon, Prather, & Lewis, 2014). These reviews are often structured through the lens of the Koob and Volkow (2010, 2016) three-step addiction cycle: binge/intoxication, withdrawal/negative affect, preoccupation/anticipation. The conceptual focus of this addiction model paved the way for investigating acute and chronic effects of alcohol. However, there has been little focus on the withdrawal/negative affect phase of the addiction cycle.

Research in the last few decades focuses on a unique, and critically understudied, phase of AUD: alcohol abstinence beyond acute withdrawal. Abstinence is a period of continuous change—in the body and in the brain—that occurs after ceasing alcohol use. Risk of relapse is highest within the first year of abstinence (Zywiak, Connors, Maisto, & Westerberg, 1996); relapse is likely driven by the neural changes that occur during alcohol abstinence and subsequent development of negative affect symptoms, like increased stress sensitivity, anxiety, and depression (Koob, 2013a; Zhao, Weiss, & Zorrilla, 2007; Zywiak et al., 1996). During abstinence, women are more likely to relapse and experience stress-related relapse, compared to men (Logrip, Milivojevic, Bertholomey, & Torregrossa, 2018). Additionally, women's trajectory of AUD is often described as ‘telescoping’—where women have an accelerated path from first drink to the development of AUD and entering treatment, compared to men (Ashley et al., 1977; Piazza, Vrbka, & Yeager, 1989); although evidence in telescoping for women, is mixed with several studies reporting either no evidence of telescoping, or that telescoping is driven by specific variables, like drinking milestones (e.g., Keyes, Martins, Blanco, & Hasin, 2010; Lewis, Hoffman, & Nixon, 2014). Women are also less likely to seek treatment over the course of their chronic alcohol use (Greenfield et al., 2007). Increased risk of relapse and ‘telescoping’ among women, along with increased drinking behaviors and harms in the past decade, suggests that sex plays a prominent role in alcohol use and abstinence, and that we are in a unique time to determine sex differences in alcohol abstinence.

The goal of this literature reviewis to summarize sex differences in the human brain during alcohol abstinence. The first section briefly summarizes the impact of acute and chronic alcohol use on the human brain. The second section introduces and summarizes sex differences in brain structure, function, and connectivity in people with AUD who are abstinent, including gray and white matter volume and structural connectivity, intrinsic functional connectivity, and task-based activation and connectivity. We focus on studies where individuals are abstinent for at least 21 days to control for the impact of withdrawal on brain measures. Finally, the last section introduces possible mechanisms that contribute to sex differences during alcohol abstinence. This review is guided by the following pillars: (1) women are a critical group to study, considering that they have rising rates of use and experience more severe alcohol consequences; (2) abstinence is a critical and understudied phase of AUD; (3) sex differences are complex and vary across different measures of brain structure, function, and connectivity; and (4) sex differences are likely driven by mechanisms related to differential effects of alcohol, hormones, and stress responses. This review will focus on sex differences in the abstinent phase of AUD in humans, specifically focusing on neuroimaging data, which provides non-invasive measurements of brain structure, function, and connectivity. Additionally, these brain measures provide translational opportunities to compare human findings of sex differences to animal models of alcohol abstinence. This review will focus on reports of group (abstinent, controls) × sex interactions to account for known sex differences that exist in controls. In this review, we define sex as the sex assigned at birth (male, female) because sex assigned at birth is the most reported definition in the alcohol literature. There is limited work that investigates gender identity as a construct in alcohol use (see review for impact of sex and gender on drinking: Erol & Karpyak, 2015). We expand on this limitation in the future directions section.

Brain regions relevant to human neuroimaging studies of AUD abstinence

Extensive animal research and recent human neuroimaging research reveals that chronic alcohol exposure causes the brain to undergo neuroadaptations to create a new homeostasis. The neuroadaptations largely affect reward and stress brain systems, consistent with prior work emphasizing the role of reward and stress systems in addiction (Koob, 2008; 2013a). Animal models indicate that initially, alcohol intake causes widespread changes in the reward system, such as increased GABA activity and increased dopamine release in the nucleus accumbens (NAc), and increased dopamine and opioid release in the ventral striatum (Carrillo & Gonzales, 2011; Mitchell et al., 2012; Nestler, 2005; Robinson, D.L., Howard, McConnell, Gonzales, & Wightman, 2009). In rodent stress systems, acute alcohol use increases GABA activity in the extended amygdala, inhibits glutamate synaptic activity and increases GABA release in the amygdala, and inhibits NMDA receptor function and increases stress signaling in the bed nucleus of the stria terminalis (BNST) (Kash, Baucum, Conrad, Colbran, & Winder, 2009; Robinson, S.L., Alexander, Bluett, Patel, & McCool, 2016). Chronic alcohol use contributes to further homeostatic changes, causing a decrease in reward system function and an increase in stress system function. Specifically in rodent reward systems, dopamine release decreases in the NAc and endocannabinoid signaling decreases across the striatum (DePoy et al., 2013; Karkhanis, Huggins, Rose, & Jones, 2016). In rodent stress systems, there is increased corticotrophin releasing factor signaling in the amygdala and increased NMDA receptor function in the prefrontal cortex (PFC; Herman, Contet, & Roberto, 2016; Kroener et al., 2012). Together, animal models of drinking reveal that changes in reward and stress system function commonly lead to large-scale signaling changes across brain regions, which is thought to result in the development of negative affect during alcohol withdrawal and abstinence.

The most prominent effects of alcohol abstinence are seen in corticolimbic brain regions implicated in reward and stress processes. For example, decreases in reward system function are seen in the NAc, ventral tegmental area, insula, and areas of the prefrontal cortex, while increases in stress system function are seen in the amygdala, BNST, hypothalamic-pituitary-adrenal (HPA) axis, habenula, and NAc shell (Koob, 2013a; 2013b). Notably, the BNST is at the core of this network and is responsible for initiating and coordinating stress responses. In humans and rodents, the BNST is also highly interconnected to corticolimbic brain regions impacted by alcohol use and abstinence: the amygdala, anterior hippocampus, NAc, anterior insula, hypothalamus, ventromedial prefrontal cortex (vmPFC), dorsolateral PFC (dlPFC), and dorsal anterior cingulate cortex (dACC) (Avery et al., 2014; Dong, Petrovich, & Swanson, 2000; 2001; Dong, Petrovich, Watts, & Swanson, 2001; Dong & Swanson, 2004; Flook et al., 2020; A. X. Gorka, Torrisi, Shackman, Grillon, & Ernst, 2018; Krüger, Shiozawa, Kreifelts, Scheffler, & Ethofer, 2015).

The BNST network implicated in alcohol abstinence is also sexually dimorphic (Avery et al., 2014; Flook et al., 2020). The BNST is essential for sexual function and expresses hormonal receptors (De Vries & Panzica, 2006; Segovia & Guillamón, 1993). Additionally, the rodent BNST is larger in males and contains more neurons in males—a result of sex-specific neurodevelopmental pruning (Allen & Gorski, 1990). There is also emerging evidence that female rodents express more corticotrophin releasing factor neurons in the BNST, likely contributing to stress-related sex differences (Tsukahara & Morishita, 2020; Uchida et al., 2019). Several corticolimbic brain regions in the BNST are also sexually dimorphic—-specifically, the amygdala and HPA axis (Handa & Weiser, 2014; Hines, Allen, & Gorski, 1992). The amygdala is larger in male rodents, while female rodents exhibit higher baseline activity of amygdala neurons (Price & McCool, 2022). The HPA axis contains sex hormone receptors (i.e., androgen and estrogen receptors), therefore impacting sex-specific stress regulation and processing such that human women exhibit greater stress responses than human men.

Sex differences in the human brain during abstinence

Chronic alcohol use impacts brain structure and function, and abstinence disrupts these changes. Seminal research using computerized tomography brain scans were pivotal for revealing sex differences in alcohol abstinence (e.g., Jacobson, 1986; Mann, Batra, Günthner, & Schroth, 1992), therefore paving the way for current neuroimaging research. Specifically, recent work from our group and others highlights how these changes differ between men and women across several imaging measures. We will discuss baseline differences in brain structure, function, and connectivity, specifically focusing on reported sex differences and interactions with other variables (e.g., measures of drinking, anxiety). Brain structure measures reflect the underlying structural properties of the brain. Brain structure is typically measured using gray matter volume (across the brain or in specific regions), white matter volume (across the brain in major and minor tracts and white matter microstructure), and diffusion tensor imaging of major and minor tracts, which measures the integrity of white matter tracts. Measurements of brain function reflect the activity of the brain, or changes in blood level oxygenation level dependent (BOLD) signal, while a participant is passively viewing images or engaging in a behavioral task in the scanner. Finally, brain connectivity measures reflect the inter-connectedness of brain regions and can be used to reveal brain networks; brain connectivity can be measured at rest or during a task. Structural volume, structural connectivity, and intrinsic connectivity all measure baseline differences in the brain; baseline differences are ideal for identifying trait differences between groups, which may guide development of tasks or interventions. Evidence suggests that baseline brain differences lead to functional, or task-based, brain differences (Vincent et al., 2007; Warbrick, Rosenberg, & Shah, 2017). However, there is also evidence that brain differences may only be revealed during tasks (Zhao, Stockwell, et al., 2023). Thus, it is critical to examine both baseline and task-based differences to have a comprehensive understanding of changes that occur in the brain in AUD abstinence. Notably, the studies performed to date are cross-sectional and rarely include measurements of hormones; these limitations are discussed later in the review.

Brain structure

Gray matter.

The earliest and largest studies investigating sex differences in alcohol abstinence investigated gray matter volume in major brain lobes and regions. Hommer, Momenan, Kaiser, and Rawlings (2001) investigated sex differences in total grey matter volumes in abstinent adults (54% men; average length of abstinence = 21 days) and controls (51% men), revealing that abstinent women had lesser gray matter volume than control women; no differences were detected between abstinent and control men. The same year, Pfefferbaum, Rosenbloom, Deshmukh, and Sullivan (2001) investigated sex differences in abstinent adults (51% men; average length of abstinence = 84 days) and controls (56% men), revealing that abstinent men had lesser gray matter volume than control men; no differences were detected between abstinent and control women. One decade later, Sullivan, Rohlfing, and Pfefferbaum (2010) investigated sex differences in cerebellar and pons volume in abstinent adults (67% men; average length of sobriety = 82 days) and controls (48% men); no group (abstinent, controls) × sex interactions were detected. However, an additional analysis in the abstinent group showed that abstinent men had smaller cerebellar volume than abstinent women. A secondary analysis in a subset of abstinent adults was performed to control for lifetime drinking, but no sex differences were detected in cerebellar and pons volume. Demirakca et al. (2011) investigated sex differences in recently abstinent adults from an inpatient unit (54% men; average length of abstinence = 17 days) and controls (52% men) using voxel-based morphometry to measure voxel-level differences in whole brain cerebrospinal fluid, gray matter, and white matter. No group (abstinent, controls) x sex interactions were detected; however, the authors report that gray matter volumewas smaller in abstinent adults than controls in the amygdala and hippocampus, cingulate gyrus, insula, orbitofrontal cortex, and thalamus. The authors also reported that gray matter volumes in those regions were smaller in men than women. Recently, Sawyer et al. (2017) tested for specific sex effects in abstinent adults (50% men; average length of abstinence = 6.2 years) and controls (48% men) in global gray matter measures, as well as within brain regions involved in reward (amygdala, dlPFC, ACC, insula, hippocampus, NAc, orbitofrontal cortex, subcallosal cortex, temporal pole, and ventral diencephalon). The authors found group (abstinent, control) x sex differences in total reward network volume, such that abstinent men had lower total gray matter volume than control men. There was also a significant group × sex interaction in the dlPFC, where abstinent women had larger gray matter volume than control women. Secondary analyses investigated the relation between duration of heavy drinking and gray matter volume, such that higher duration of heavy drinking in abstinent men was correlated with lower temporal pole gray matter volume, but higher temporal pole volume in abstinent women.

In summary, studies of gray matter volume differences in alcohol abstinence revealed that sex differences are primarily driven by men: abstinent men have lesser gray matter volume than controls (men and women). Interestingly, these differences exist across a variety of brain regions involved in reward and stress processing, supporting decades of animal studies revealing the widespread impact of alcohol abstinence on reward and stress processing.

White matter.

Measurements of white matter include white matter volume and connectivity across the brain. White matter volume measures total white matter volume across the brain, which includes major tracts, minor tracts, as well as white matter microstructure. Structural connectivity methods measure the strength of white matter tracts connecting brain regions in the brain. Structural connectivity is usually assessed using diffusion-tensor imaging, an imaging modality that measures water flow and direction in fiber bundles. The most common measures of white matter integrity are fractional anisotropy, radial diffusivity, and mean diffusivity. The amount of diffusion (water flow) is identified in each voxel along three main axis or directions. Fractional anisotropy (FA) measures the strength of diffusion in the main axis relative to the other two axes and is the most common measure. Structural connectivity studies typically measure the white matter integrity (FA) within the major fiber bundles in the brain (e.g., longitudinal fasciculus). Another approach is to use deterministic or probabilistic tractography methods to identify specific white matter tracts and then measure strength of those connections. Therefore, structural connectivity in humans refers to the literal physical connections between brain regions, like tract tracing methods in rodents. Across all measurements, sex differences are observed between abstinent men and women, and between abstinent men and control men.

White matter volume.

Hommer et al. (1996) were one of the first to investigate sex differences in white matter volume in alcohol abstinence adults (48% men; average length of abstinence = 26 days) and controls (53% men). Group (abstinent, control) × sex differences revealed that abstinent women had lesser corpus callosum volume than abstinent men and controls (men and women). Several years later, Hommer et al. (2001) also investigated sex differences in total white matter volume in abstinent adults (54% men; average length of abstinence = 21 days) and controls (51% men); no group (abstinent, control) × sex (men, women) differences were detected. Pfefferbaum et al. (2001) also assessed white matter volume differences in abstinent adults (51% men; average length of abstinence = 84 days) and controls (56% men) in the study described above. Group (abstinent, control) × sex differences revealed that abstinent men had lesser white matter volume than control men in major brain lobes (prefrontal, frontal, anterior superior temporal, posterior superior temporal, anterior parietal, and posterior parietal-occipital); no differences were detected between abstinent and control women. Ruiz et al. (2013) characterized widespread white matter differences in abstinent adults (50% men; average length of sobriety = 8 years) and controls (50% men), where they found that abstinent men have lesser white matter volume than control men in total corpus callosum volume and in the mid-anterior, central, and mid-posterior subregions. Additional analyses included measurements of drinking; increased duration of heavy drinking in abstinent women correlated with lower white matter volumes in the frontal cortex, temporal lobe, and parietal lobe. Greater quantity frequency index, a measure of total alcohol consumption, also correlated with larger ventricle volume in abstinent women.

There are few studies that investigate sex differences in white matter volume during alcohol abstinence. However, the studies summarized suggest that men drive group × sex differences, while women primarily drive associations between drinking measures and white matter volume changes.

Major white matter tracts.

Most structural connectivity studies in abstinent adults with AUD have focused on major white matter tracts. Rivas-Grajales et al. (2018) investigated white matter integrity of the medial forebrain bundle in abstinent adults (30% men; average abstinence = 7 years) and controls (32% men); the medial forebrain bundle connects corticolimbic brain regions relevant to AUD abstinence (e.g., ACC, amygdala, NAc). Abstinent men show weaker FA in the medial forebrain bundle than control men, while abstinent women show greater FA than control women. Additional analyses in the abstinent group included number of daily drinks; abstinent women with greater number of daily drinks had weaker FA in the medial forebrain bundle, while abstinent men had no correlation between number of daily drinks and FA. Sawyer et al. (2018) also investigated white matter integrity across the whole brain in abstinent adults (47% men; average length of abstinence = 7 years) and controls (46% men). Several clusters across the brain showed differences in FA, including in the corpus callosum (genu and anterior), the left arcuate fasciculus and extreme capsule, and superior longitudinal fasciculi. In all these clusters, abstinent men had weaker FA than control men, while abstinent women had stronger FA than control women.

Seed-based connectivity of white matter tracts.

Seed-based connectivity can identify minor white matter tracts by using a primary brain region of interest (‘seed’) to investigate specific networks implicated in alcohol abstinence (e.g., reward and stress processing networks). Our group examined structural connectivity of the BNST using probabilistic tractography in abstinent adults (53% men; average length of abstinence = 127 days) and controls (45% men) (Flook et al., 2021). We focused on the BNST, based on the BNST's role in stress-related changes during alcohol abstinence, and its corticolimbic connections (vmPFC, anterior insula, amygdala, anterior hippocampus, and hypothalamus). Overall, BNST network connectivity was stronger in abstinent women compared to controls, with no differences between abstinent and control men. Post-hoc analysis showed that the difference between abstinent and control women was strongest for the BNST-anterior insula connection—abstinent women had stronger connectivity compared to control women, while abstinent men had weaker connectivity than control men. Secondary analyses in the abstinent group revealed a correlation with AUD severity scores (AUDIT; Babor, Higgins-Biddle, Saunders, & Monteiro, 2001) and BNST-vmPFC connectivity: abstinent men with higher AUDIT scores had stronger BNST-vmPFC connectivity, whereas abstinent women with higher AUDIT scores had weaker BNST-vmPFC connectivity.

Together, these studies of white matter integrity suggest there is a divergent, sex-dependent impact of alcohol abstinence on white matter tracts in the human brain. Widespread sex differences are seen across major bundles connecting corticolimbic and sensory brain regions. A common theme emerges from these studies: differences between abstinent and control men primarily drove sex differences, but many alcohol-specific associations were primarily detected in abstinent relative to control women (e.g., quantity frequency index, number of daily drinks). However, the study of BNST connectivity revealed that women drive sex differences in BNST network white matter tract integrity. This suggests that sex differences in white matter may differ by network, wherein major white matter tract differences are driven by men, but minor white matter tract differences relevant to stress and alcohol abstinence may be driven by women. Additionally, this suggests that the dose-response effects of chronic alcohol exposure on brain structure may be more prominent in women, despite structural integrity impacting both men and women in these studies.

Brain function

Intrinsic functional connectivity.

Intrinsic (‘resting state’) functional connectivity refers to activation patterns of the brain at rest. Specifically, intrinsic functional connectivity identifies how activation in a seed region is associated with activation (or deactivation) in other regions across the brain. The seed region can be determined using a priori methods, like a specific brain region, or it can be determined by using nodes from an established resting-state networks (e.g., dorsal attention network, salience network). Few studies have investigated sex differences intrinsic functional connectivity in alcohol abstinent adults, despite the ease of collecting this type of data. Of the published studies that have investigated intrinsic functional connectivity differences between alcohol abstinent adults and controls, most only include men (e.g., Bordier et al., 2022; Deng et al., 2022; Yang et al., 2023), do not report on sex differences (e.g., Muller & Meyerhoff, 2020; Srivastava et al., 2021; van Oort et al., 2023), or control for sex differences (e.g., Camchong et al., 2021).

Prior work from our group investigated intrinsic connectivity between abstinent adults and controls, in BNST network connectivity (Flook et al., 2023). This sample overlaps with our prior report on structural connectivity with abstinent adults (55% men; average length of abstinence = 127 days) and controls (45% men). We used a seed-based approach, where we pre-defined nodes of the BNST network (vmPFC, anterior insula, amygdala, anterior hippocampus, and hypothalamus). Abstinent men had weaker BNST-hypothalamus, BNST-anterior hippocampus, and BNST-amygdala connectivity than control men. Abstinent men had stronger BNST-vmPFC connectivity than control men. There were no group differences in women. Secondary analyses examined the correlation of alcohol severity (AUDIT scores) in the abstinent group: men with higher AUDIT scores had weaker BNST-hypothalamus connectivity, while women with higher AUDIT scores had stronger BNST-hypothalamus connectivity.

These very limited data suggest, again, that there are widespread effects on the brain that occur during alcohol abstinence that differ between men and women. Like structural volume and connectivity, these differences are largely driven by differences between abstinent and control men.

Task-based activation.

Functional activation measures BOLD signal in the brain while a task is performed. Common approaches are to either examine activation in a priori regions of interest thought to be engaged by the task or an exploratory approach of activation across the whole brain. Few studies investigate sex differences in task-based brain activation, likely due to the difficulty with collecting the data compared to structural and intrinsic functional connectivity methods. That is, structural and intrinsic connectivity methods can be performed more easily in larger samples because of relatively short scan lengths, standardized methods, and the ability to combine data across multiple samples and sites. Task-based imaging requires longer scan lengths and more effort from participants—often resulting in smaller, more preliminary studies. Task-based brain measurements will be a critical next step for expanding sex differences research in alcohol abstinence. Notably, the published studies have used emotion processing tasks, which are of interest because chronic alcohol use and alcohol abstinence affect reward and stress processing in the brain. Thus, a primary interest in abstinence research is understanding the impact of alcohol abstinence on emotion processing.

Several small task-based neuroimaging studies have tested for sex differences. A preliminary study from Heinz et al. (2007) investigated whole brain activation to alcohol, positive, or negative images in recently abstinent adults (50% men; average length of abstinence = 14 days) and controls (50% men). No group (abstinent, controls) × sex differences were detected in follow-up analyses, which may have been due to the small sample size being underpowered to find sex differences. Padula, Anthenelli, Eliassen, Nelson, and Lisdahl (2015) investigated group and sex differences in whole brain activation to emotional faces (fear, happy, sad, neutral) in abstinent adults (43% men; average length of abstinence = 941 days) and controls (64% men). Post-hoc analyses revealed that abstinent women show less activation to emotional faces compared to control women, while abstinent men show stronger activation to emotional faces than control men. When viewing fear faces, abstinent women had less activation in the bilateral inferior frontal gyrus; when viewing happy faces, abstinent women had less activation in the caudate, middle frontal gyrus, paracentral lobule, and lingual gyrus than control women. Conversely, when abstinent men viewed fear faces, they had stronger bilateral inferior frontal gyrus activation than control men; when viewing happy faces, abstinent men had stronger activation in the caudate, middle frontal gyrus, paracentral lobule, and lingual gyrus than control men. Sawyer et al. (2019) investigated emotion processing differences in a large sample of long-term abstinent adults (40% men; average length of abstinence = 7 years) and controls (48% men). Emotional responses were evoked to a variety of stimuli: erotic, gruesome, happy, aversive, and neutral (the control condition). Abstinent men had higher activation in the medial orbitofrontal cortex and rostral ACC, but lesser activation in the dlPFC, thalamus, and inferior parietal cortex, than control men when viewing all emotional stimuli. Abstinent women had higher activation in the supramarginal gyrus and superior frontal gyrus compared to control women when viewing emotional stimuli. Notably, these differences were largely seen during aversive and happy images. These findings are consistent with conceptual models that suggest that the impacts of alcohol abstinence on reward and stress systems are different between men and women.

Recently, Radoman, Fogelman, Lacadie, Seo, and Sinha (2024) investigated stress- and alcohol-induced craving in a treatment-seeking sample of adults with AUD (60% men; average length of abstinence = 16 days). Abstinent men with higher craving had lesser activation of the vmPFC, dorsomedial PFC, dorsal ACC, supplementary motor area, and anteromedial thalamus, while abstinent women only showed lesser vmPFC activation, when viewing alcohol (vs. stress) stimuli. In abstinent women, higher alcohol craving was associated with higher activation of the dlPFC during neutral stimuli, but lesser activation of the dlPFC when viewing stress and alcohol stimuli. Additional analyses investigated whether brain measurements predicted future heavy drinking days during eight weeks of AUD treatment; lesser BNST and subgenual ACC activation and greater anterior insula activation when viewing stress stimuli in abstinent women predicted more heavy drinking days during treatment, but not in abstinent men. Greater activation of the hypothalamus when viewing stress stimuli, and greater hippocampus activation when viewing alcohol stimuli, predicted more heavy drinking days in abstinent men, but not in abstinent women. Additionally, when viewing alcohol (vs. stress) stimuli, lesser hypothalamus activation and greater hippocampus activation predicted more heavy drinking days in abstinent men, whereas greater activation of the vmPFC, ventrolateral PFC, and striatum predicted more heavy drinking days in abstinent women. Our group also recently investigated sex differences in task-based activation in early abstinent adults (55% men; average length of abstinence = 127 days) and controls (45% men) during an unpredictable threat task (Zabik et al., 2024). The unpredictable threat task engages the BNST and its corticolimbic network, and prior work reveals that adults with AUD have greater startle to unpredictable (vs. predictable) threats (Clauss, Avery, Benningfield, & Blackford, 2019; Goode, Ressler, Acca, Miles, & Maren, 2019; Gorka, Kreutzer, Petrey, Radoman, & Phan, 2020; S. M. Gorka, Lieberman, Phan, & Shankman, 2016; Grupe & Nitschke, 2013; Moberg, Bradford, Kaye, & Curtin, 2017; Schmitz & Grillon, 2012). For unpredictable threat cues, control men with higher anxiety exhibited higher posterior cingulate activation than abstinent men. For unpredictable threat images, abstinent men with higher anxiety exhibited greater insula and dACC activation than control men, while control men with higher anxiety scores had lower insula activation. No differences were detected in abstinent or control women.

The limited studies investigating task-based activation suggest that sex differences are also present during emotion, threat, and reward processing. Sex differences were found between abstinent and control groups. First, overall, abstinent men show higher activation compared to control men. Second, abstinent women show varying differences when compared to control women; abstinent women show overall lesser activation to emotional faces, but greater activation to broad emotional images, compared to control women. Finally, abstinent men and women show varying patterns of brain activation when craving and future drinking risk are assessed: abstinent men with higher alcohol craving show lower activation, while abstinent women show both higher and lower activation depending on the stimuli. Variation in brain activation in abstinent men and women also predicted more heavy drinking days, which contrasts with structural and intrinsic connectivity studies that report only abstinent women showing a relation between drinking and brain measures. These data high-light the importance of assessing emotional processing broadly, as well as utilizing tasks that probe threat and reward.

Task-based connectivity.

Functional connectivity is determined by using a ‘seed’ brain region to identify correlated regions that show a similar pattern activation (or deactivation) during specific task events or conditions. In the study by our lab, described above, we also investigated BNST task-based connectivity (Zabik et al., 2024). Abstinent women had weaker BNST-vmPFC, BNST-thalamus, and BNST-dlPFC connectivity than control women when viewing unpredictable threat cues. Abstinent men had stronger BNST-vmPFC connectivity, but weaker BNST-thalamus connectivity, than control men. Our study suggests that task-based functional connectivity can reveal additional network-level sex differences in alcohol abstinence. Notably, these sex differences are primarily driven by men; however, we also report differences between abstinent and control women. Considering the paucity of studies investigating task-based connectivity, this reflects a critical gap in the literature: studies of brain connectivity are critical because brain regions operate in networks to coordinate responses to stimuli in tasks.

Summary of sex differences in brain networks during alcohol abstinence

The summarized studies suggest widespread sex differences in brain structure and function during early abstinence. In structural brain studies, men primary drove differences across structural volume and connectivity. Abstinent men had smaller gray and white matter volume and weaker structural connectivity compared to control men. Interestingly, while there were rarely differences between abstinent and control women, women were also more likely to have correlations between structural brain measures and measures of alcohol drinking. Specifically, many of the studies included secondary analyses with measures of alcohol use severity (lifetime drinking, total lifetime drinking, use severity, and heavy drinking days) where abstinent women showed a relation between alcohol use and brain measures. In functional brain studies, sex differences in activation were driven by both abstinent men and women. A preliminary study on intrinsic connectivity showed that abstinent men also had weaker intrinsic connectivity than control men. Task-based functional imaging studies revealed that overall, abstinent men exhibited greater brain activation to aversive stimuli than control men, while abstinent women exhibited lesser brain activation than control women. See Table 1 for comprehensive review.

Table 1.

Brain regions implicated in sex differences during AUD abstinence.

Brain Region	Structure	Function
Reward processes
Nucleus accumbens (NAc)	GM volume: abstinent men < control mena	Activation to alcohol: higher activation in abstinent women predicted more heavy drinking days during treatmente
Ventral tegmental area (VTA)	—	—
Insula	GM volume: abstinent men < control mena WM integrity with BNST: abstinent men < control men, abstinent women > control womenb	Activation to stress: higher activation in abstinent women predicted more heavy drinking days during treatmente Activation to unpredictability: abstinent men > control menh
Ventromedial prefrontal cortex (vmPFC)	GM volume: abstinent men < control menc WM volume: abstinent men < control menc	Intrinsic connectivity with BNST: abstinent men > control menf Activation to alcohol: higher craving associated with lesser activation in abstinent men and women; higher activation in abstinent women predicts more heavy drinking days during treatmente Connectivity with BNST to unpredictability: abstinent men > control men, abstinent women < control womenh
Dorsolateral prefrontal cortex (dlPFC)	GM volume: abstinent men < control men; abstinent women > control womena WM volume: abstinent men < control menc	Activation to emotional faces: abstinent men < control meng Activation to neutral: higher craving in abstinent women predicted higher activatione Activation to alcohol and stress: higher craving in abstinent women predicted lower activation during stress and alcohol stimulie Connectivity with BNST to unpredictability: abstinent women < control womenh
Anterior cingulate cortex (ACC)	GM volume: abstinent men < control mena WM volume: abstinent men < control mend	Activation to emotional faces: abstinent men > control meng Activation to alcohol: higher craving associated with lower activation in abstinent mene Activation to stress: lower activation in abstinent women predicted more heavy drinking days during treatmente Activation to unpredictability: abstinent men > control menh
Stress processes
Amygdala Bed nucleus of the stria terminalis (BNST)	GM volume: abstinent men < control mena See Insula	Intrinsic connectivity with BNST: abstinent men < control menf Activation to stress: lower activation in abstinent women predicted more heavy drinking days during treatmente
Hypothalamus-pituitary-adrenal (HPA) axis	—	Intrinsic connectivity with BNST: abstinent men < control menf Activation to alcohol: lower activation in abstinent men predicted more heavy drinking days during treatmente Connectivity with BNST to unpredictability: abstinent men < control men, abstinent women < control womenh
Habenula	—	—

Results described in table reflect significant sex interactions reviewed in text from the following studies.

^aSawyer et al., 2017.

^bFlook et al., 2021.

^cPfefferbaum et al., 2001.

^dRivas-Grajales et al., 2018.

^eRadoman et al., 2024.

^fFlook et al., 2023.

^gSawyer et al., 2019.

^hZabik et al., 2024. GM = gray matter; WM = white matter.

Together, the data presented in this review suggest that sex differences exist during alcohol abstinence across different imaging measures, and that these sex differences are complex. The complexity of the findings does not suggest that neuroimaging sex differences in AUD abstinence are not reliable. Rather, it supports our longstanding recognition of the brain—the brain is complex, which means multiple methodologies must be used to fully characterize differences. In this review, we summarized findings from two primary lenses: (1) structural differences, using studies of structural volume and connectivity, and (2) functional differences, using studies of intrinsic connectivity, task-based activation, and task-based connectivity. Structural and intrinsic functional connectivity differences give us insight on trait differences between groups and across individuals, while task-based differences allow us to probe highly specific behavioral and emotional processes. Thus, structural and functional differences uniquely contribute to our understanding of sex differences in alcohol abstinence, which may help us understand the effect of sex on symptoms, behaviors, and relapse.

With the goal of providing a conceptualization for interpreting findings across neuroimaging measures, we provide an analogy for the relations between brain structure and volume, intrinsic connectivity, and task-based activation and connectivity: the journey of grapes becoming wine (Fig. 2). First, the grape vine structure is critical for the development of the grapes, like the structural networks of the human brain. Second, the grapes typically grow in places where the vine structure is most robust, like functional brain networks developing strongly along white matter structural connections. However, grapes can also develop in areas with fewer vines, like intrinsic brain connections that develop outside of major structural connections. Finally, the flavor of the wine is dependent on the terroir, or composition of the soil based on environmental factors. Similarly, psychological processes evoked by tasks are the final output resulting from the brain's structural and functional connections, while influenced by individual factors.

Fig. 2. Contribution of baseline and task-based imaging measures for understanding sex differences in alcohol abstinence.

Images designed by Maxwell Blackford (www.maxwellblackford.com).

Why do women show different patterns of brain differences than men during alcohol abstinence?

There are several theories that attempt to describe the factors that contribute to sex differences in AUD, as well as in alcohol abstinence. These theories primarily center around two over-arching biological systems: hormones and stress responses. Notably, these systems are also highly overlapping. However, there is extensive data to support the separable impact of hormones and stress responses on sex differences in alcohol use, maintenance, and abstinence. Here we will provide a general review of the role that hormones and stress responses play in sex-specific responses to alcohol that may lead to sex differences in brain function during alcohol abstinence. For more extensive reviews on hormones, see Lenz et al. (2012) and Peltier et al. (2021), and for stress, Oyola and Handa (2017).

Hormone response to alcohol beyond acute intoxication

One potential mechanism underlying sex differences in the brain in early abstinence is sex hormones. Even though women drink less than men, women have more alcohol-related harms and illnesses. A potential reason for the greater impact of alcohol on women is likely sex differences in the metabolism and hormone responses to alcohol. For example, women reach a higher blood alcohol concentration at comparable body weight to men; higher blood alcohol concentrations are due to women having a lower proportion of water in their bodies than men, resulting in a higher concentrations of alcohol (Ely, Hardy, Longford, & Wadsworth, 1999). Higher blood alcohol concentration leads to higher tissue exposure to both alcohol and its toxic metabolite, acetaldehyde. This increased toxin exposure likely increases risk for severe alcohol-related illnesses, like liver inflammation, cardiovascular disease, and certain cancers (Agabio et al., 2016). Notably, alcoholic liver disease progresses more rapidly in women, and persists longer in abstinence (Loft et al., 1987; Rehm et al., 2010).

Acute and chronic alcohol use also differentially impact hormone regulation in men and women. In men, acute alcohol decreases testosterone levels (Vatsalya, Issa, Hommer, & Ramchandani, 2012); higher baseline testosterone levels in adolescent boys is predictive of risky drinking in adulthood (Costello, Sung, Worthman, & Angold, 2007; de Water, Braams, Crone, & Peper, 2013). Interestingly, higher testosterone levels in men and women are linked to risky drinking in both sexes (La Grange, Jones, Erb, & Reyes, 1995). Alcohol also decreases progesterone levels in men and women (Sarkola, Mäkisalo, Fukunaga, & Eriksson, 1999), and decreases levels of allopregnalone in both sexes, but this decrease is associated with alcohol liking in men (Pierucci-Lagha et al., 2006). Alcohol is also more rewarding to women due to hormone regulation: women drink more in the luteal menstrual phase when estradiol and progesterone are highest (Becker & Koob, 2016; Martel, Eisenlohr-Moul, & Roberts, 2017). The relation between estradiol and progesterone with drinking outcomes is, however, largely mixed due to administration of alcohol differing between studies (intravenous vs. oral administration) (Peltier et al., 2019). Hormones play a modulating role in drinking behaviors and outcomes, and therefore may contribute to sex differences reported during alcohol abstinence.

While extensive research investigates the role of sex hormones in AUD, very limited studies investigate their role during alcohol abstinence. It is likely that testosterone levels increase and estrogen levels decrease during abstinence (Forquer, Hashimoto, Roberts, & Wiren, 2011; Heberlein et al., 2016). A recent study found differences in progesterone levels and craving between abstinent pre- and post-menopausal women, such that high progesterone in post-menopausal women was associated with lower alcohol craving (Weinland, Muhle, Kornhuber, & Lenz, 2021). Additionally, recent interest in allopregnanolone, a neurosteroid, has surfaced from research demonstrating decreases in allopregnanolone during chronic alcohol use and acute withdrawal, as well as recent applications of allopregnanolone as a treatment for stress-related disorders (e.g., depression, posttraumatic stress disorder) (Boero, Porcu, & Morrow, 2020; Cagetti, Pinna, Guidotti, Baicy, & Olsen, 2004; Gatta, Camussi, Auta, Guidotti, & Pandey, 2022; Maldonado-Devincci et al., 2014; Romeo et al., 1996). Interestingly, allopregnanolone has actions on GABA receptors and corticotrophin signaling—both of which are altered in alcohol abstinence. These very limited data suggest that, like brain neurotransmitters, sex hormones also experience alterations during alcohol abstinence; it will be critical for future research to determine how these changes occur over abstinence.

Stress response in alcohol abstinence

Sex differences in response to emotional stimuli are reported across several emotion domains—most notably responses to negative or threatening stimuli. The differences in stress responses parallel the epidemiological differences, where women are at greater risk for anxiety disorders and stress-related alcohol relapse (Cohen et al., 2007; Crum et al., 2018; Kessler et al., 1994). In mice, females exhibit greater HPA-axis responses to acute stress than males, and these responses are greater when estradiol levels are high (Babb, Masini, Day, & Campeau, 2013; Heinsbroek, Van Haaren, Feenstra, Endert, & Van de Poll, 1991; Iwasaki-Sekino, Mano-Otagiri, Ohata, Yamauchi, & Shibasaki, 2009; Lovick, 2012; Viau & Meaney, 1991). In humans, studies of sex differences in stress responses have mixed findings. During stress inductions, healthy men exhibit greater cortisol elevation than healthy women, but anxious women show a greater cortisol elevation than anxious men (e.g., Earle, Linden, & Weinberg, 1999; Kirschbaum, Pirke, & Hellhammer, 1995, 1999; Peeters, Nicholson, & Berkhof, 2003). Interestingly, sex differences are also evident in stress responses in the brain—in stress induction studies, men show a correlation between increased dorsomedial PFC activation and lower stress, but women show a correlation between increased dorsomedial PFC activation and higher stress (Goldfarb, Seo, & Sinha, 2019; Seo, Ahluwalia, Potenza, & Sinha, 2017).

Abstinence from alcohol represents a unique stressor. Acute alcohol intake increases peripheral cortisol levels (i.e., HPA-axis activity) in adults without an AUD (Richardson, Lee, O'Dell, Koob, & Rivier, 2008). In chronic alcohol users however, acute alcohol intake does not impact peripheral cortisol levels; HPA-axis activity to acute alcohol is likely blunted because chronic alcohol users have sustained HPA-axis activation (Blaine, Nautiyal, Hart, Guarnaccia, & Sinha, 2019; King, Munisamy, de Wit, & Lin, 2006; Wemm et al., 2013).

Thus, alcohol abstinence disrupts HPA-axis allostasis, likely disrupting responses to threatening and aversive stimuli. Low or blunted peripheral cortisol responses to threat and alcohol cues are reported during alcohol abstinence, with two studies linking a blunted cortisol response to a higher risk of relapse (Junghanns et al., 2003; 2005; Sinha et al., 2009); however, other studies show high cortisol response to stress is linked to early relapse (e.g., Sinha et al., 2011). Acute alcohol intake also reduces startle response specifically to unpredictable threat (i.e., not to predictable or certain threat) and decreases threat-related brain activation in the central nucleus of the amygdala and BNST (Gorka et al., 2016; Hur et al., 2018; Moberg et al., 2017). This is consistent with studies from our lab that show extensive sex differences in alcohol abstinence in BNST network structural connectivity and functional activation and connectivity in response to unpredictable threat, and that these differences were also dependent on anxiety levels (Flook et al., 2021, 2023; Zabik et al., 2024). Together, stress-related sex differences, along with altered threat processing in AUD, reflect an inflection point for women to be disproportionally impacted by a unique stressor like alcohol abstinence.

Future directions

The study findings reviewed suggest that men and women experience alcohol abstinence differently, and these differences are reported across numerous brain imaging measures. Most of the studies we reviewed were cross-sectional, comparing abstinent adults to controls, and only investigated one time-point of alcohol abstinence. This represents a major gap in knowledge: the brain continues to change and recover from chronic alcohol during abstinence, and therefore sex differences in brain structure and function may fluctuate in different directions during the recovery period. Together, though, these findings set the stage for alcohol research to integrate women, and specifically investigate sex or gender, into study designs and analyses with the intention of delineating sex effects in neural and behavioral mechanisms of AUD. We have several suggestions for future directions to further investigate sex differences in alcohol abstinence.

Integrating sex as a biological variable

While extensive progress is being made to address sex differences, we also have substantial gaps to fill. This review revealed complex sex differences across several brain imaging methods. Notably, the studies we summarize are preliminary, as most report small sample sizes; many studies in the alcohol abstinence literature are not powered to test for sex differences. Thus, the simplest way to address investigations of sex differences is to increase sample sizes. Larger studies will help us determine whether the differences we see are consistent across large samples of abstinent men and women, and likely reveal additional differences that current studies are not powered to detect. Additionally, moving beyond studying emotion processing will greatly contribute to our understanding of threat, reward, and cognition during abstinence. Therefore, two primary avenues to address this gap in knowledge are: (1) include sex as a variable of interest and include these analyses in reported studies and (2) develop a working group to identify a standard set of tasks to implement, therefore leveraging larger sample sizes.

Sex and gender in alcohol abstinence

A limitation of current alcohol research is the lack of differentiation between sex and gender. Inclusion of participants who are non-binary or transgender will be essential for continuing to serve underserved individuals. Notably, non-binary and transgendered individuals are at higher risk for binge drinking and risky drinking (Messman & Leslie, 2019; Scheim, Bauer, & Shokoohi, 2016; Staples, Neilson, George, Flaherty, & Davis, 2018), which emphasizes the need for socially- and culturally-sensitive treatment pathways (Cochran & Cauce, 2006; Nuttbrock, 2012). This review's focus on recognizing sex differences in alcohol abstinence is the first step towards championing health care pathways for other underserved populations. To this end, researchers should consider partnering with community-based organizations to both recruit and educate non-binary and transgender people on alcohol use. Studies that incorporate sex hormone measures, as well as measures of social transition (e.g., pronouns, adopting a new name), will be important for identifying the separate and overlapping influence of sex and gender on alcohol use and abstinence.

Forward- and reverse-translation studies of alcohol abstinence

This review focused on human neuroimaging studies in alcohol abstinence. However, many of the studies discussed were influenced by decades of animal research. A critical barrier for alcohol research is the development of forward- and reverse-translational studies. To mind this gap, researchers should identify a colleague whose research goals align with their own, but whose species is vastly different from the one they study. Further, identify a common pathway, behavior, or brain region of interest to investigate together. Cross-species translation will also be critical for delineating sex differences in alcohol use, and whether sex differences stem from hormonal response to chronic alcohol or other pathways (e.g., stress).

Conclusions

The field of alcohol is entering a new era of research: an era where we more deeply investigate women. Despite alcohol being a commonly used substance throughout human history, we still have much to learn about its long-term effects on humans. We continue to discover how and why men and women experience AUD abstinence differently, and how these differences are driven by a multitude of factors. In the spirit of this review, we presented compelling neuroimaging data that suggests that sex differences in alcohol abstinence exist across measures of brain structure of function, and that these differences may be driven by chronic effects of alcohol and their interplay with hormones and stress-responding. Future research that expands analyses to include sex as a variable, incorporate sex hormone measurements, and integrate cross-species investigations will be paramount for defining the role of sex in alcohol abstinence. Additionally, longitudinal brain imaging studies will allow the field to delineate between effects of chronic alcohol use and recovery-related brain changes. Sex differences in brain measures during alcohol abstinence will also need to be contextualized—which sex differences in the brain are relevant to stress responding or hormonal impacts of alcohol, and which sex differences reflect slower or faster brain recovery during abstinence? Studies probing these questions, as well as addressing future directions we outlined above, will be essential for developing broader conceptual models of alcohol use and recovery. The complexity of alcohol's impact on the brain and how it affects men and women differently during abstinence will be an ever-evolving avenue of research, which will ultimately lead to impactful change in the alcohol field and, we hope, will also positively impact successful recovery in men and women.