Sex/gender differences in brain function and structure in alcohol use: a narrative review of neuroimaging findings over the last 10 years

Abstract

Over the last 10 years, rates of alcohol use disorder (AUD) have increased in women by 84% relative to a 35% increase in men. Rates of alcohol use and high-risk drinking have also increased in women by 16% and 58% relative to a 7% and 16% increase in men, respectively, over the last decade. This robust increase in drinking among women highlights the critical need to identify underlying neural mechanisms that may contribute to problematic alcohol consumption across sex/gender (SG), especially given that many neuroimaging studies are underpowered to detect main or interactive effects of SG on imaging outcomes. This narrative review aims to explore the recent neuroimaging literature on SG differences in brain function and structure as it pertains to alcohol across positron emission tomography (PET), magnetic resonance imaging (MRI), and functional magnetic resonance imaging (fMRI) modalities in humans. Additional work using magnetic resonance spectroscopy (MRS), diffusion tensor imaging (DTI), and event-related potentials (ERPs) to examine SG differences in AUD will be covered. Overall, current research neuroimaging AUD, alcohol consumption, or risk of AUD is limited, and findings are mixed regarding the effect of SG on neurochemical, structural, and functional mechanisms associated with AUD. We address SG disparities in the neuroimaging of AUD and propose a call to action to include women in brain imaging research. Future studies are crucial to our understanding of the neurobiological underpinnings of AUD across neural systems and the vulnerability for AUD among women and men.

Introduction

Over the last ten years, rates of alcohol use disorder (AUD) have increased in women by 84%, relative to a 35% increase in men (Grant et al., 2017). Rates of alcohol use and high-risk drinking have also increased in women by 16% and 58% relative to a 7% and 16% increase in men, respectively, over the last decade (Grant et al., 2017). A recent meta-analysis confirms that the increase in past year drinking and past year binge-drinking over the last fifteen years is nearly driven solely by women (Grucza et al., 2018). The rise in problematic drinking in women is particularly alarming given that women experience worse alcohol-related health consequences, including sex-specific consequences (Andersen, Andersen, Olsen, Grønbæk, & Strandberg-Larsen, 2012; National Institute of Alcohol Abuse and Alcoholism, 2017; Peltier et al., 2019), compared to men. The robust increase in drinking among women underscores the critical need to identify underlying neural mechanisms that may contribute to problematic drinking in women to ultimately develop sex-appropriate treatments for AUD.

It is known that the brain is vulnerable to the neurotoxic effects of chronic alcohol use. Neuroadaptations associated with AUD span multiple brain regions and neurotransmitter systems. For example, individuals with AUD have long-lasting alterations in brain chemistry and lower brain volume and cortical thickness (e.g., dopamine [DA], prefrontal cortex [PFC], white and gray matter) compared to healthy controls (Heinz, Beck, Grüsser, Grace, & Wrase, 2009; Hillmer, Mason, Fucito, O'Malley, & Cosgrove, 2015; Mackey et al., 2018; Nixon, Prather, & Lewis, 2014). Regarding sex, literature dating back nearly two decades suggests that women may be more vulnerable to the neurotoxic effect of alcohol on the brain compared to men; although results are mixed (Hommer, Momenan, Kaiser, & Rawlings, 2001; Mann et al., 2005). Nonetheless, the effect of sex on alcohol-related abnormalities in brain neurochemistry and morphometry remains critically understudied in the last decade.

Differences between women and men can be identified as sociocultural (gender) (Darnall & Suarez, 2009) and biological (sex) (Janine Austin Clayton, 2018; Cornelison & Clayton, 2017; National Institutes of Health, 2019). For the purpose of the present narrative review we will use sex/gender (SG) to acknowledge that findings may be driven by both sociocultural and biological factors (Pace et al., 2018; Streed & Makadon, 2017). It is known that SG differences exist on multiple levels, including neurobiology, neurochemistry and connectivity (Choleris, Galea, Sohrabji, & Frick, 2018; Cosgrove, Mazure, & Staley, 2007; Lind et al., 2017), with relevance for understanding SG differences in the neural mechanisms underlying addiction (Peltier et al., 2019; Sharrett-Field, Butler, Reynolds, Berry, & Prendergast, 2013); yet, even modern neuroimaging studies on addiction often do not account for SG in their analyses. To highlight this problem, a meta-analysis of structural imaging papers published through 2016 in the substance use field found that while female enrollment increased over time, enrollment was significant lower than males, particularly for the alcohol field (Lind et al., 2017). Further, while 79% of structural imaging studies in the substance use field included both sexes, 74% did not analyze by SG or were underpowered to do so (Lind et al., 2017).

To set the stage for this narrative review, we first evaluated imaging studies on AUD, alcohol consumption, or risk of AUD published in the past 10 years (n=252) to determine their SG representation and whether results were analyzed by SG. Consistent with the meta-analysis by Lind and colleagues (2017), we found that women were included in 59%, 81%, and 80% of positron emission tomography (PET), magnetic resonance imaging (MRI), and functional magnetic resonance imaging (fMRI) studies, respectively, of AUD, alcohol consumption, or risk of AUD (see Table 1). However, even in studies including women, female enrollment was approximately two to four times lower than male enrollment. These neuroimaging studies lack sufficient power to examine SG differences, with only 13% and 18% of PET and MRI studies pertaining to alcohol analyzing data by SG (see Table 1). But, this is not just a power issue; even fMRI studies with adequate SG balance fail to study SG differences (see Table 1). This emphasizes the urgent need to include sex as a biological variable (SABV) in the neuroimaging of AUD. Indeed, the National Institutes of Health (NIH) required the inclusion of women in clinical research in 1993 (Clayton & Collins, 2014), with recent requirements to account for SABV in research strategy and analyses in all grant applications (National Institutes of Health, 2019). This is a fundamental step in the advancement of our knowledge on SG-based differences in the underpinnings of addiction, including improving the rigor and reproducibility of our findings to ultimately guide clinical practice and effective treatment for both sexes (Peltier et al., 2019).

Table 1.

Sex/gender representation and analysis in human neuroimaging studies of alcohol risk, alcohol use, or AUD in the past 10 years^a

	No. of Studies	% of Studies Excluding Women	Total N Size (M/W) of Alcohol Sample	% Women of Total Sample	% of Studies Analyzed by Sex
PET	46	41	841/208	20	13
MRI	57	19	3771/1836	33	18
fMRI	149	20	5520/3040	36	11
Totals	252	24	10132/5084	33	13

^aNeuroimaging studies were limited to research conducted in humans in the past 10 years. PubMed search terms were [("PET" OR "positron emission tomography") AND ("AUD" OR "alcohol use disorder" OR "alcohol-dependent" OR "alcohol dependent")]; [("MRI" OR "magnetic resonance imaging") AND ("AUD" OR "alcohol use disorder" OR "alcohol-dependent" OR "alcohol dependent")]; [("fMRI" OR "functional magnetic resonance imaging") AND ("AUD" OR "alcohol use disorder" OR "alcohol-dependent" OR "alcohol dependent")].

Note: PET=positron emission tomography, MRI=magnetic resonance imaging, fMRI=functional magnetic resonance imaging, M/W=men/women

The purpose of this narrative review is to highlight the recent human neuroimaging literature on SG differences in brain function and structure as it pertains to alcohol use across multiple imaging modalities from the last 10 years. We include SG effects from PET, MRI, and fMRI studies examining the neural effects of acute alcohol ingestion, AUD, and in those at risk for AUD. Additional work using magnetic resonance spectroscopy (MRS), diffusion tensor imaging (DTI), and event-related potentials (ERPs) to examine SG differences in AUD will also be briefly covered. We aim to highlight the state of the alcohol imaging field as it pertains to the inclusion of SABV, identify gaps in our knowledge, and provide future considerations for investigating SG differences in the neural mechanisms underlying AUD.

PET Neuroimaging

Background

PET uses radioactively labeled molecules called radiotracers to measure receptors and other chemicals, as well as fluctuations in neurotransmitter levels, in the living brain (E. D. Morris, Lucas, & Cosgrove, 2013). Many radiotracers have been designed to bind to dopamine (DA) and serotonin (5HT) receptors, among numerous others, and are used to measure receptor availability. These techniques have implications for identifying SG differences in neurochemical mechanisms underlying substance use disorders and relating brain mechanisms to important clinical correlates of addiction (Verplaetse, Morris, McKee, & Cosgrove, 2018), including AUD. We first present studies that focused on the dopamine system, and then turn to studies that focused on the opioid system.

Sex/gender differences in PET neuroimaging studies of AUD, alcohol consumption, or risk of AUD

A recent meta-analysis on the dopamine system and sedative drug use, including alcohol, suggests lower striatal D2/D3 receptor availability in alcohol users compared to healthy controls (Kamp et al., 2019). While acute alcohol may be related to modest increases in striatal DA release in social drinkers, in family history positive (FHP) and family history negative (FHN) individuals without AUD, and in AUD (Kegeles et al., 2018; Setiawan et al., 2014; Urban et al., 2010; Yoder et al., 2016), more chronic alcohol use results in dysfunctional DA transmission as evidenced by findings of lower amphetamine-induced increases in DA release in striatum in recently detoxified individuals with AUD vs. controls (Kamp et al., 2019). To date, few PET investigations have examined SG differences in dopaminergic alterations related to alcohol use or AUD. For example, using the D2/3 radiotracer [¹¹C]raclopride, oral alcohol evoked significant ventral striatal DA release in young adult social drinkers (Urban et al., 2010). In this group, men had greater DA release than women in the ventral striatum in response to alcohol (men: ΔBP_ND = −12 ± 8%; women: ΔBP_ND = −6 ± 8%, where ΔBP_ND is the change in [¹¹C]raclopride binding potential) (Urban et al., 2010). Further, in men, but not in women, there was a significant positive correlation of ventrostriatal DA release with positive subjective effects of alcohol (e.g., stimulation, elation, vigor) within the 30 minutes following alcohol ingestion, suggesting that, in men, greater DA release may contribute to the initial reinforcing properties of alcohol (Urban et al., 2010). Though, others using [¹¹C]raclopride PET report no SG differences in striatal DA release in response to oral alcohol among adults with AUD or FHP and FHN without AUD (Kegeles et al., 2018). Regarding AUD risk factors, young adult social drinkers who were FHP for AUD and imaged with [¹¹C]raclopride had higher baseline ventral striatal DA D2/D3 receptor availability (BP_ND=2.35 [0.06]) compared to family history negative (FHN) controls (BP_ND=2.17 [0.02]); an effect driven by social drinking men, not women, in this sample (Alvanzo et al., 2017). Findings suggest alterations in the DA neurotransmitter system in response to alcohol ingestion or in those at-risk for or with AUD, and that these alterations are particularly notable for men. This is consistent with diffusion tensor imaging (DTI) studies (see Other Modalities section) finding differences in the medial forebrain bundle, a key track in the reward circuit in AUD compared to controls (Rivas-Grajales et al., 2018). Future work should further elucidate variations in alcohol-related alterations on the DA system and in SG-based differences in the DA system that may underlie the vulnerability to problematic drinking.

Two other PET investigations examined SG differences in the opioid system in AUD, as opioid receptors have been implicated in AUD (Crowley & Kash, 2015; Mason, 2017; Mitchell et al., 2012). Using the selective kappa opioid receptor (KOR) antagonist radiotracer, [¹¹C]LY2795050, adults with AUD had lower KOR availability than healthy controls in the amygdala and pallidum (A. Vijay et al., 2018). Exploratory analyses suggested that adult men with AUD had lower KOR availability than healthy adult men in 9 regions, whereas adult women with AUD had lower KOR availability in the amygdala compared to healthy adult women (A. Vijay et al., 2018). In this study, comparison of AUD women to healthy women was underpowered to detect differences in multiple regions, underscoring the need to recruit more women. For perspective, in a sample of healthy adult individuals, men had higher KOR availability than women in multiple brain regions, including the anterior cingulate cortex (ACC), frontal cortex, and insula (Aishwarya Vijay et al., 2016), suggesting that differences in KOR are related to sex differences and not simply a question of sample size. In contrast to the KOR system, studies on the mu opioid receptor (MOR) system demonstrated higher MOR availability in the ventral striatum, amygdala, caudate, and putamen in adults with AUD compared to healthy controls using the radiotracer [¹¹C]carfentanil (Weerts et al., 2011). While a group by sex effect was not significant, women overall had lower MOR availability in ventral striatum (BP_ND=1.76 ± 0.06 vs. 1.47 ± 0.08) and cingulate (BP_ND=0.74 ± 0.02 vs. 0.65 ± 0.3) compared to men (Weerts et al., 2011). Findings in the opioid system suggest that there may be underlying SG differences in KOR and MOR availability related to chronic alcohol consumption, albeit the pattern of findings suggests that the KOR and MOR systems may work in opposite directions.

Summary

The examination of SG differences in PET neuroimaging of AUD is drastically understudied. To our knowledge, only five PET investigations in the last decade examined SG differences related to alcohol use or risk of AUD across the DA and opioid receptor systems. These studies implicate striatal regions and the amygdala as key regions vulnerable to alcohol use or in genetic risk of AUD. Results from these studies should be replicated in AUD populations and in larger samples, particularly AUD women, as they may have important implications for SG-based neural differences in the involvement of DA, KOR, and MOR in AUD. Further, SG differences in the neurochemical mechanisms underlying AUD may have implications for the efficacy of pharmacotherapeutics for AUD such as naltrexone, a non-selective opioid antagonist, which may be more effective in males compared to females with AUD (Garbutt et al., 2005).

MRI

Background

MRI is an imaging technique used to examine how disease affects brain structure, such as brain volume and cortical thickness. As such, MRI has been employed to look at the neurotoxic effects of chronic alcohol consumption on the brain. Findings suggest that chronic alcohol use or AUD is associated with lower gray and white matter volumes compared to controls. For example, individuals with AUD demonstrate significant reductions in gray matter volume in corticostriatal-limbic circuits compared to healthy controls (for meta-analysis see (Yang et al., 2016)). However, robust SG disparities exist in the structural imaging literature. As previously highlighted, 74% of structural imaging studies of substance use disorders conducted through 2016 did not evaluate SG effects or did not use appropriate analytic approaches to do so (Lind et al., 2017). Here, we review the recent MRI literature on SG differences in alcohol-related brain morphometry.

Sex/gender differences in MRI studies of AUD, alcohol consumption, or risk of AUD Adolescence

Emerging evidence suggests that SG differences in brain morphometry emerge in childhood and continue to differentiate in adolescence and adulthood (Giedd, Raznahan, Mills, & Lenroot, 2012; Lenroot et al., 2007; Luders, Gaser, Narr, & Toga, 2009). At baseline, boys have larger total brain size (by ~8-10%) than girls (Giedd et al., 2012; Lenroot et al., 2007). Structural differences in boys and girls have been found in caudate, amygdala, and hippocampus (Giedd et al., 2012) – all regions implicated in addiction. Cortical and subcortical gray matter volumes generally follow an inverted U-shaped trajectory, with adolescent females peaking earlier than males (Giedd et al., 2012; Lenroot et al., 2007). Through adolescence, males tend to have a steeper rate of increase in white matter volume compared to females (Lenroot et al., 2007). In adulthood, in studies where brain size is controlled, women have more regional gray matter concentration than men (Luders et al., 2009).

In adolescence, the initiation of harmful drinking patterns may pose a risk for the development of AUD in adulthood. Multiple brain regions have been implicated in early engagement with alcohol use, including the hippocampus, NAcc, and PFC (for a comprehensive review see (Ewing, Sakhardande, & Blakemore, 2014)), and may be particularly at risk for the neurotoxic effects of alcohol (i.e., brain volume loss). Overall, sexual dimorphism in brain morphometry is complicated, especially when considering the effect of substances of abuse on the brain. This section will highlight what is known about SG differences in brain morphometry associated with alcohol use in adolescence and adolescent-onset AUD.

Brain Volume.

Recent structural MRI studies reveal SG differences in brain volume related to heavy drinking in adolescence and adolescent-onset AUD spanning the PFC, thalamus, and putamen. In PFC, adolescent females with AUD demonstrated smaller PFC volumes and smaller PFC white matter volumes than their healthy female counterparts (cc³; anterior ventral PFC: AUD mean=0.0473, SD=0.0040; HC mean=0.0554, SD=0.0064; p<0.03), whereas adolescent males with AUD had larger PFC volumes and PFC white matter volumes than their healthy male counterparts (cc³; anterior ventral PFC: AUD mean=0.0586, SD=0.0063; HC mean=0.0531, SD=0.0071; p<0.009) (Medina et al., 2008). This is consistent with gray matter atrophy in PFC being more strongly associated with heavy drinking in females compared to males (S. Seo et al., 2019). Adolescent female binge drinkers also exhibited ~8% thicker frontal cortices compared to same-sex controls, which was associated with worse cognitive performances, while male binge drinkers had ~7% thinner cortices compared to same-sex controls (Squeglia et al., 2012). This suggests delayed cortical thinning, a process promoting efficient neural processing in adolescence, in adolescent female drinkers. These results are consistent with functional findings in frontal regions for heavy drinking in adolescence (see fMRI section). For example, males with adolescent-onset AUD or binge-drinking males had greater frontal activation in response to a spatial working memory task compared to same-sex controls (Cohen’s d=0.6), whereas females with adolescent-onset AUD or binge-drinking females showed less frontal activation to a spatial working memory task compared to same-sex controls (Cohen’s d=1.2) (Squeglia, Schweinsburg, Pulido, & Tapert, 2011). For female binge drinkers, less frontal activation was associated with poorer working memory and attention on the spatial working memory task (Squeglia et al., 2011).

Others found SG differences in volume loss in the thalamus and putamen in adolescents with AUD. The pattern of brain volume loss was reversed from that demonstrated in the PFC; males with AUD had smaller thalamus (mm³; mean=16277, SD=939) and putamen (mm³; mean=9395, SD=975) volumes compared to non-AUD males (mm³; mean=16502, SD=819; mean=10088, SD=710, respectively), whereas females with AUD had larger thalamus (mm³; mean=16355, SD=934) and putamen (mm³; mean=9623, SD=1092) volumes than non-AUD females (mm³; mean=15920, SD=867; mean=9237, SD=878, respectively) (Fein et al., 2013). Although, Seo and colleagues (2019) found that atrophy of gray matter in thalamus and other regions was more strongly associated with heavy drinking in females compared to males (S. Seo et al., 2019). Taken together, these studies suggest AUD is associated with small to medium effect sizes on volume when stratified by sex.

SG differences in structural findings in adolescent AUD, however, are not consistent. While frontal and temporal volumes were reduced overall in adolescents with AUD or young adults with a history of excessive drinking compared to healthy controls, an effect of SG on reduced grey matter volumes in frontal and temporal brain areas was not significant (Brooks et al., 2014; Heikkinen et al., 2017). This is somewhat consistent with findings from the National Consortium on Alcohol and NeuroDevelopment in Adolescence (NCANDA) that gray and white matter volume and surface area did not differ between adolescent girls and boys with moderate to high alcohol use; however, smaller gray and white matter volumes were present in adolescent girls compared to boys with no to low alcohol use (Pfefferbaum et al., 2016).

Family history of AUD may also be a risk factor for alterations in brain volume in adolescence. In the hippocampus, male FHP adolescents had larger left hippocampal volume than FHN males, an effect not significant for females or in the right hippocampus (Hanson et al., 2010). A trend showed that males overall tended to have smaller hippocampal volume than females (Hanson et al., 2010). Hippocampal volume was not predictive of alcohol use at follow-up (Hanson et al., 2010). Another study found a positive association between left nucleus accumbens volume and FHP in adolescent females only (R²=0.11, p=0.006 in females vs. R²=0.005, p=0.53 in males) (Cservenka, Gillespie, Michael, & Nagel, 2015). Findings suggest that risk factors for AUD, such as FHP, may be related to alterations in hippocampal and nucleus accumbens volume during a peak time in neurodevelopment for male and female adolescents. However, a recent study found no evidence of a SG by family history interaction for regional volumes in hippocampus, amygdala, nucleus accumbens, orbitofrontal gyrus, inferior frontal gyrus, and middle frontal gyrus in young adults even after controlling for recent alcohol use (McPhee et al., 2018).

Adulthood

SG differences in brain morphometry that emerge in adolescent alcohol use may continue into adulthood, with increased vulnerability to disruptions in brain volume and cortical thickness as individuals progress toward AUD (Peltier et al., 2019). This section will highlight what is known about perturbations in alcohol-related brain morphometry in adulthood; although there are relatively few studies that examine SG differences in alcohol-related alterations in brain structure in adults and those studies that do examine SG differences demonstrate mixed results.

Brain Volume.

Structural findings in adults suggest that moderate to chronic alcohol use may be associated with deficits in frontal areas, and that alterations in frontal regions may be greater in heavy drinking men. In young adults, associations between reduced cortical thickness in frontal regions (e.g., dlPFC, anterior cingulate cortex [ACC], orbitofrontal cortex [OFC]) and drinking were driven by men; an effect not observed in women (V. Morris et al., 2019). In a longitudinal cohort study of over 30 years, moderate-to-high alcohol consumption was also associated with greater hippocampal shrinkage in men compared to women (Topiwala et al., 2017). This is consistent with findings of reduced cortical thickness, reduced gray matter volumes, and smaller reward system volumes, including dlPFC, in men with AUD compared to same-sex controls (Momenan et al., 2012; Sawyer et al., 2017), and findings of reduced white matter volume in corpus callosum in heavy drinking older men or abstinent men with AUD compared to same-sex controls but not in women (Kapogiannis et al., 2012; Ruiz et al., 2013). Lastly, three studies found that men with AUD had smaller reward system and cerebellar volume compared to women with AUD (Sawyer et al., 2016; Sullivan, Rohlfing, & Pfefferbaum, 2010); although, SG findings are mixed and likely due to the inclusion of a disproportionate sample of men compared to women (Cardenas, Hough, Durazzo, & Meyerhoff, 2020).

Conversely, smaller brain volumes in women drinkers or in women with AUD compared to same-sex controls or men with AUD have also been demonstrated. For example, women with AUD exhibited a larger decrease in cortical thickness when compared to same-sex controls (mean difference in right precentral and right postcentral gyrus=0.402 mm and 0.358 mm, respectively) than did men with AUD compared to same-sex controls (mean difference in right insula and right medial frontal gyrus=0.291 mm and 0.328 mm, respectively) (Momenan et al., 2012). Further, a recent study showed that heavy drinking women had more gray matter loss than men, especially in the OFC; although the interaction effect was weak overall (Thayer et al.,2016). Structural data also suggests that years of heavy drinking in abstinent women with AUD may be associated with reduced frontal and temporal white matter (1-1.5% volume loss for each additional year of heavy drinking) (Ruiz et al., 2013), which is consistent with findings that women with AUD have greater alterations in white matter compared to men with AUD (Srivastava et al., 2010). Alcohol expectancy may be related to alterations in gray matter in problem drinking women; positive alcohol expectancy was associated with less gray matter volume in the right posterior insula, a region involved in drug craving, in non-dependent social-drinking women compared to men (Ide et al., 2017). Thus, psychological factors associated with gray matter volume loss may represent a neural phenotype of risk for AUD in the transition from social to more habitual drinking habits (Ide et al., 2017). It should be noted that one study found that women with AUD had 4.4% larger reward system volumes compared to same-sex controls, whereas men with AUD had 4.1% smaller reward region volumes compared to same-sex controls (Sawyer et al., 2017), indicating further discrepancies in neuroanatomical mechanisms associated with AUD.

Yet, there are still other voxel-based morphometry studies demonstrating no SG differences in global or regional brain volume loss, including insula volume, in women and men with AUD compared to same-sex healthy controls (Demirakca et al., 2011; Senatorov et al., 2015); although, there was evidence of widespread brain volume loss overall in individuals with AUD or who were undergoing alcohol withdrawal compared to healthy controls, consistent with the larger literature of brain volume loss with chronic, excessive alcohol consumption.

It should be noted that abstinence from alcohol may be associated with recovery of brain volume. For instance, longer recent sobriety was associated with larger white matter volumes in abstinent women with AUD compared to men (Pfefferbaum, Rosenbloom, Deshmukh, & Sullivan, 2001; Ruiz et al., 2013), suggesting that abstinence from alcohol may reverse some of the neurotoxic effects of heavy, chronic drinking, particularly in women. Others found that alcohol abstinence reversed gray matter volume and cerebral ventricle volume changes in alcohol-dependent men and women (Sawyer et al., 2017; van Eijk et al., 2013). For example, total ventricular volume was 1.8% lower per year of sobriety (Sawyer et al., 2017). Thus, abstinence from alcohol may be related to the recovery of gray and white matter volumes in individuals with AUD, although whether SG modulates recovery is unclear.

Summary

The MRI literature is larger than that of PET regarding the imaging of SG differences in AUD but is still lacking in terms of recent investigations and our overall understanding of SG differences in brain morphometry associated with adolescent alcohol use and the transition to AUD in adulthood. Overall, adolescent girls demonstrated smaller PFC volumes and less frontal activation compared to same-sex controls, whereas boys generally demonstrated the opposite pattern, and this may have implications for cognitive deficits associated with alcohol use. Family history should be considered a risk factor in the development of AUD, as alterations in hippocampus and nucleus accumbens have been associated with FHP in this peak time of neurodevelopment. Findings are largely mixed for adults with AUD. Findings suggest reduced gray and white matter volume in AUD men only, in AUD women only, or no SG differences in reduced brain volume regarding problem drinking or AUD compared to healthy controls. Overall, structural neuroimaging results indicate that smaller brain volumes are associated with alcohol use and that SG difference findings are generally mixed.

fMRI

Background

fMRI is an imaging technique that detects local fluctuations in blood flow, blood volume, and oxygenation to examine brain activity, often in response to cognitive tasks or stimuli. Previous fMRI studies have shown that brain areas in the salience network, such as the insula and anterior cingulate cortex (ACC), or regions associated with reward processing, including ventral striatum, medial and dorsolateral PFC (mPFC and dlPFC, respectively), amygdala, and OFC, are activated by alcohol-related cues in heavy drinkers and in individuals with AUD (Heinz et al., 2009; D. Seo et al., 2011; Wrase et al., 2007). Further, the taste of alcohol alone can elicit increases in blood-oxygen-level-dependent (BOLD) PFC activity in drinkers, a response that positively correlates with drinking behavior (r=0.4, p=0.02) and alcohol craving (r=.33, p=0.05) (Filbey et al., 2008). However, because SG differences are understudied, findings of sex-dependent effects on neural responses in AUD samples are not consistent between studies. We summarize below the scant literature examining SG differences in functional activity in response to alcohol-related stimuli in subclinical (i.e., moderate to heavy drinkers not yet diagnosed with AUD) and AUD populations.

Sex/gender differences in fMRI studies of AUD, alcohol consumption, or risk of AUD Adolescence

Reactivity to alcohol-related cues may be an important factor contributing to adolescent alcohol use and the trajectory to more habitual drinking in adulthood. For example, alcohol-naïve adolescent girls showed greater activation to alcohol picture cues than boys in frontal areas (Nguyen-Louie et al., 2018), which is consistent with a prior study in girls with adolescent-onset AUD (Tapert et al., 2003). However, once moderate-to-heavy drinking started, the pattern reversed such that boys showed greater activation than girls in response to alcohol picture cues (Nguyen-Louie et al., 2018). Thus, differences in activation to alcohol-related stimuli emerge prior to alcohol initiation in early adolescence and continue to differentiate once moderate to heavy drinking started in late adolescence.

Brain areas related to cognition and affective processing may also be affected by early alcohol use in adolescence. In adolescent binge drinkers or those with adolescent-onset AUD, boys had greater frontal activation in response to a spatial working memory task compared to same-sex controls (Cohen’s d=0.6), whereas girls who were binge drinkers or had adolescent-onset AUD showed less frontal activation to a spatial working memory task compared to same-sex controls (Cohen’s d=1.2) (Squeglia, Schweinsburg, Pulido, & Tapert, 2011), suggesting that the magnitude of differences was greater for girls than boys. For binge-drinking girls, poorer cognitive performance was associated with less frontal activation (Squeglia et al., 2011). A longitudinal fMRI study of adolescent males and females with a family history of AUD (age range: 8.5 to 17.6 years; 3-4 scans per participant) demonstrated that males showed decreased activation in amygdala and precentral gyrus with age in response to negative vs. neutral words on an affective word task (Hardee et al., 2017). Adolescent females showed persistent activation in the amygdala and precentral gyrus with age in response to negative vs. neutral words (Hardee et al., 2017), a finding consistent with the notion that girls/women may be more reactive to negative emotional stimuli. Overall, SG-dependent alterations in frontal activation and limbic processing of negative stimuli may lead to differences in risk for AUD before adulthood (Hardee et al., 2017).

Adulthood

Alcohol Cue-Induced Activation.

As previously mentioned, brain areas associated with reward become activated in response to alcohol-related cues in heavy drinkers and in those who transitioned to AUD. Regarding SG differences in alcohol cue-induced activation, findings suggest that women with AUD activated reward circuits, cognitive control circuits, part of the salience network, including anterior insula, and the default-mode network (DMN) in response to cue-induced ‘high-risk’ decisions to drink compared to ‘low-risk’ decisions to drink relative to healthy control women (Arcurio, Finn, & James, 2015). In response to alcohol taste cues, heavy drinkers activated the ACC, bilateral amygdala, right lateral OFC, anterior insula, thalamus, and putamen, but men showed greater response in the left amygdala compared to women (Claus, Ewing, Filbey, Sabbineni, & Hutchison, 2011). Neuroadaptations related to AUD in these regions may be related to maladaptive regulation of neural circuits related to reward and cognition and may have implications for the likelihood of high-risk drinking in response to alcohol cues. For example, a recent resting state fMRI study found that drinking was associated with SG differences in thalamic connectivity in relation to AUDIT scores (Zhornitsky et al., 2018). In men, connectivity between thalamus and salience and executive control regions correlated with AUDIT scores (Zhornitsky et al., 2018). In women, thalamic connectivity to posterior cingulate cortex correlated with AUDIT scores (Zhornitsky et al., 2018).

Other functional studies have not shown significant SG differences in cortical or subcortical activation in response to alcohol-related cues or cognitive tasks in non-dependent drinkers and in those with AUD (Garbusow et al., 2016; Zehra et al., 2019; Zhu et al., 2015). For example, a recent study found no SG differences in BOLD signal in the NAcc during a Pavlovian-to-instrumental transfer (PIT) task in detoxified individuals with alcohol dependence (Garbusow et al., 2016). The sample of women was significantly smaller than men, so this may have limited their ability to detect an effect of SG. Future studies need to replicate results and include sufficient numbers of women for adequate analysis of SG within the NAcc, as it is a key region associated with the reinforcing properties of alcohol and other substances of abuse.

Emotional Cue-Induced Activation.

In addition to reward processing, the ACC, mPFC, and amygdala may play a role in affective processing. Long-term alcohol abstinent men with AUD had altered temporal limbic activation in response to emotional faces compared to healthy men, suggesting that deficient amygdala and hippocampus activation may underlie impairments in the processing of emotionally salient stimuli in AUD (Marinkovic et al., 2009). However, that study included only men. When examining both sexes, results suggest frontal brain activation in response to affective or arousing stimuli, albeit in opposing directions depending on SG (Ide et al., 2018; Padula, Anthenelli, Eliassen, Nelson, & Lisdahl, 2015; Sawyer et al., 2019). For example, men with AUD showed greater activation than healthy men in response to fearful faces in the inferior frontal gyrus, whereas women with AUD showed lower activation than healthy women in response to fearful faces (Padula et al., 2015). The same pattern of findings was found in response to happy faces in the left caudate, right middle frontal gyrus, left paracentral lobule, and right lingual gyrus (Padula et al., 2015). Most recently, Sawyer and colleagues (2019) found that alcoholic men had lower activation compared to non-alcoholic men in frontal brain areas in response to all emotional stimuli, whereas alcoholic women had greater activation than non-alcoholic women in superior frontal cortex and supramarginal gyrus in response to happy and aversive stimuli, respectively (Sawyer et al., 2019). Further, in response to uncertain and predictable threat, individuals with AUD had greater right insula and dorsal ACC activation compared to healthy controls but a significant effect of SG on functional activation was absent (Gorka, Kreutzer, Petrey, Radoman, & Phan, 2019). Nonetheless, findings suggest alterations in the functional processing of emotional stimuli in both women and men with AUD.

Acute Alcohol-Induced Activation.

Another way to examine differences in alcohol-related activation is to utilize an acute alcohol challenge. Two studies conducted by Marinkovic and colleagues found conflicting results regarding SG effects on activation of the ACC, a region thought to be involved in executive cognitive function, in response to acute alcohol administration in healthy social drinkers (Marinkovic, Rickenbacher, Azma, & Artsy, 2012; Marinkovic, Rickenbacher, Azma, Artsy, & Lee, 2013). While both studies found that alcohol vs. placebo administration attenuated ACC activity during a cognitive control task (0.11 vs. 0.08% signal change) (Marinkovic et al., 2012 [0.11 vs. 0.08% signal change]; Marinkovic et al., 2013), one found that social-drinking women reported feeling more intoxicated than men and had lower activity in the left ACC compared to men following acute alcohol administration (Marinkovic et al., 2012) and the other did not find an effect of sex on ACC activity following alcohol administration (Marinkovic et al., 2013). However, the second study also did not find an effect of SG on cognitive control, suggesting that lower activity in the ACC following alcohol consumption impairs cognitive control in social drinkers of both sexes. This decrement in ACC activity may lead to greater increases in drinking behavior (Marinkovic et al., 2013).

Summary

Functional neuroimaging findings demonstrated activation to alcohol-related cues, emotional cues, and acute alcohol administration spanning adolescence and adulthood. Drinking in adolescence may be related to SG-dependent alterations in frontal activation. In adults, findings are generally mixed, with some studies demonstrating greater activation in women, some demonstrating greater activation in men, and some showing no SG differences in functional activity in response to cues. Results are also inconsistent regarding the effect of acute alcohol ingestion on brain activation. As with other imaging modalities, findings regarding SG differences in the functional neuroimaging of AUD are limited by inadequate sample size or lack of analysis (see Table 1); thus, further underscoring the need to examine functional neural mechanisms underlying AUD in women and men.

Other Modalities

MR spectroscopy (MRS) has been used to detect metabolic changes related to AUD, including changes in GABA and N-acetylaspartate, a putative marker of neuronal function, though few MRS papers have studied SG differences in alterations in brain metabolism in AUD. To our knowledge, only one such study has been conducted in the last ten years. In that study, Yeo et al. (2013) found no SG differences in brain metabolite concentrations in individuals with chronic, heavy alcohol use vs. healthy controls (Yeo et al., 2013). This is in contrast to previous findings of lower N-acetylaspartate levels, indicative of neuronal loss, in frontal gray matter in recently detoxified alcohol-dependent women with AUD compared to healthy women; an effect not significant between recently detoxified men with AUD and healthy men (Schweinsburg et al., 2003). Discrepant findings may be related to sample size differences between studies or higher N-acetylaspartate levels found in the healthy women compared to healthy men in the latter study (Schweinsburg et al., 2003). Nonetheless, it is possible that women with AUD may be more vulnerable to gray matter injury compared to their male counterparts (Schweinsburg et al., 2003). These results should be replicated in future studies.

Diffusion tensor imaging (DTI) is an MRI technique that can characterize tissue microstructure such as the principle direction of axonal fibers based on restricted diffusion of water molecules in tissue (Nixon et al., 2014). DTI has been used, albeit limitedly, to examine SG differences in AUD. For example, abstinent men with AUD had lower fractional anisotropy in corpus callosum, superior longitudinal fasciculus, and the arcuate fasciculus plus extreme capsule (FA; mean=0.68, 0.44, 0.43, respectively) compared to same-sex controls (mean=0.73, 0.47. 0.47, respectively), whereas abstinent women with AUD had higher FA (mean=0.72, 0.50, 0.49, respectively) in the same regions compared to same-sex controls (mean=0.69, 0.45, 0.45, respectively) (Sawyer et al., 2018). Findings were similar in the medial forebrain bundle, a white matter pathway connecting the VTA to the NAcc; abstinent men with AUD had lower FA (mean=0.54 vs. 0.62) and higher radial diffusivity (RD; mean=0.53 vs. 0.44) compared to same-sex controls, whereas abstinent women with AUD had higher FA (mean=0.58 vs. 0.55) and lower RD (mean=0.47 vs. 0.50) compared to same-sex controls (Rivas-Grajales et al., 2018). These findings suggest sexual dimorphism in the reward system in relation to AUD. These studies are consistent with findings in adolescent binge-drinkers (Squeglia et al., 2012), suggesting that structural alterations in cerebral white matter may arise with acute alcohol use and, thus, precede chronic patterns of alcohol consumption (Sawyer et al., 2018). In a recent large study of 218 participants from the Human Connectome Project, an advanced diffusion model was used to examine white matter integrity in individuals FHP for AUD (Waters, Sawyer, & Gansler, 2019). Although FHP was related to reduced white matter density compared to FHN controls, there was no interaction between SG and family history (Waters et al., 2019). Nonetheless, there is some indication that structural abnormalities in white matter related to AUD may be SG-dependent; specifically, that white matter abnormalities may be greater in men with AUD.

Likewise, we must include recent findings using event-related potentials (ERPs), a measure of brain function during cognitive engagement, to examine SG differences in heavy drinkers (Nixon et al., 2014; J. L. Smith, Iredale, & Mattick, 2016). In a study of female heavy drinkers, women exhibited impaired inhibitory processing and performance monitoring compared to less heavy drinkers. This was demonstrated on a stop-signal task with longer stop-signal reaction time, a larger P3 increase for successful vs. failed inhibition trials, and smaller error-related negativity (ERN) amplitude (J. L. Smith & Mattick, 2013; Janette L Smith & Mattick, 2014). When examining both women and men in the same task, longer stop-signal reaction times were related to heavy drinking in women only; SG differences were not significant for P3 or ERN amplitude (J. L. Smith et al., 2016). Thus, findings suggest that heavy drinking women may have both behavioral and physiologic deficits related to alcohol consumption and inhibitory processing, whereas men only demonstrate physiologic deficits. Such findings may have implications for alcohol-related disinhibition in the development of AUD. Further, using ERP in an alcohol cue reactivity task, male binge drinkers exhibited greater P3 amplitude reactivity to alcohol cues vs. no-alcohol cues compared to male light drinkers; an effect not significant in women (Petit, Kornreich, Verbanck, & Campanella, 2013). This result suggests that binge drinking men may have greater alcohol cue reactivity compared to light drinkers and women, and this may be a possible mechanism by which men develop AUD. SG differences in behavioral inhibition and alcohol cue reactivity should be further investigated as it relates to ERP and other imaging modalities.

Overview of Findings

Few neuroimaging studies include sufficient numbers of women to power an analysis by SG. Those that have examined SG differences in brain function and structure related to alcohol risk, alcohol use, or AUD generally demonstrate mixed findings. PET neuroimaging of drinking implicates SG-dependent differences in striatal DA release and in opioid receptor availability; social-drinking men have greater alcohol-evoked DA release compared to women and in AUD, men demonstrate regional differences in KOR availability compared to women. MRI and fMRI studies are largely mixed, suggesting either reduced gray matter volume or region-specific activation in AUD men only, in AUD women only, or no SG differences in cortical or subcortical brain volumes or activity. One MRS study found no SG differences in brain metabolites of heavy drinking men and women. Nonetheless, common brain regions implicated in these SG-dependent findings are in cortico-limbic areas. Cortico-limbic pathways should be considered in future work examining SG differences in the neuroimaging of AUD. The degree of mixed results on SG-based differences in brain function and morphometry in AUD may be a result of the low rates of female enrollment in AUD neuroimaging studies and the low rates of AUD neuroimaging studies that analyze data by SG (see Table 1). It would also benefit the state of field by including quantitative values within original research publications to identify the magnitude of SG differences in the neuroimaging of alcohol use. The findings of this review critically underscore the need to include more female participants in neuroimaging studies, possibly by over-recruiting women and/or conducting stratified analyses, to further elucidate SG differences in neural mechanisms underlying AUD and to ultimately improve SG-appropriate pharmacotherapeutic treatment of AUD.

Future Considerations to Address Sex/Gender Disparities

Inclusion of Sex as a Biological Variable (SABV)

Historically, women have been excluded from scientific research because they were of childbearing age; because there was a general consensus that women were less affected by a number of medical conditions and thus, could be treated similarly to men; and because it was thought that the inclusion of women in scientific research added a layer of complexity and thus, constituted additional costs to research (Howard, Ehrlich, Gamlen, & Oram, 2017; Mazure & Jones, 2015). While these misperceptions are changing, as highlighted by Table 1 and throughout this review, there are often insufficient numbers of women included in research on the neuroimaging of AUD to study SG differences. Even when sufficient numbers of women are included, as in the case of fMRI research, results are still not analyzed by sex. With the NIH requirement to include women in clinical research and recent added requirements for including SABV in all grant applications (Janine A Clayton & Collins, 2014; National Institutes of Health, 2019), we are hopeful that our understanding of SG differences in the neural mechanisms underlying AUD will continue to improve in order to guide future research, clinical practice, and pharmacotherapeutic treatments for AUD.

Certainly, with the call to action for the inclusion of women in research study populations, we should consider SG-specific barriers that may preclude women from participation in research studies and affect access and entry to treatment. For example, the subject burden may be higher in women as women may be more likely to experience stigma and financial constraints and may be less likely to take time off work, find transportation, and obtain affordable childcare while participating in research or seeking treatment for substance use disorders (Greenfield, Back, Lawson, & Brady, 2010; Tuchman, 2010). Further, there are other logistical considerations, such as scan costs and other recruitment challenges, that may deter recruitment of adequate sample sizes to power the detection of sex effects. Nonetheless, it is important that the NIH prioritizes women’s health research, that the field upholds the NIH guidelines to include and analyze SABV in not only neuroimaging research but all research domains, and that the we consider the socioeconomic barriers women may face in scientific research.

Consideration of Stress-Related Brain Regions

Findings presented in the current review suggest that alcohol-related disturbances in brain function and structure in women and men may overlap with stress pathophysiology. Brain regions such as the PFC, amygdala, insula, ACC, and hippocampus demonstrate alterations in volume and activation by SG as a function of alcohol use (e.g., AUD vs. healthy controls), and these structures are also heavily implicated in stress-related drinking behavior (Logrip, Milivojevic, Bertholomey, & Torregrossa, 2018; Peltier et al., 2019; Zakiniaeiz, Scheinost, Seo, Sinha, & Constable, 2017). For example, alcohol and stress both induce disturbances in PFC activity in both women and men (Peltier et al., 2019; D. Seo et al., 2011; D. Seo et al., 2013). Alcohol-related cues activated the mPFC and dlPFC in individuals with AUD (George et al., 2001; Grüsser et al., 2004), and lesions of the ventromedial PFC enhance hypothalamic-pituitary-adrenal (HPA) axis reactivity in response to stress (Sinha, 2008). AUD individuals had lower activation in the ventromedial PFC in response to stress, and lower ventromedial PFC activity was predictive of greater days of alcohol use after relapse (D. Seo et al., 2013). Further, abstinent individuals with AUD had blunted amygdala functional connectivity in response to stress, suggesting disruptions in stress processing that may be related to chronic alcohol use (Wade et al., 2017). Thus, there is robust evidence for maladaptations in the functional response of cortico-limbic brain regions related to addiction and stress in men and women with AUD (Nikolova, Knodt, Radtke, & Hariri, 2016; Peters, Peper, Van Duijvenvoorde, Braams, & Crone, 2017; D. Seo et al., 2013; van Wingen, Mattern, Verkes, Buitelaar, & Fernández, 2010; Zakiniaeiz et al., 2017).

This overlap in neural networks may be important as there is evidence suggesting that women are more likely to drink to avoid negative reinforcers (e.g., negative mood, anxiety), whereas men are more likely to drink for positive reinforcement (e.g., euphoria, social aspects) (for comprehensive reviews see (Logrip et al., 2018; Peltier et al., 2019)). AUD women activate less and AUD men active more in frontal regions in response to fearful stimuli compared to same-sex controls (Padula et al., 2015) and healthy men activate the insula (Lee et al., 2014), an area related to inhibitory control, in response to stress. Thus, it is possible that women may be less able to process negative emotional stimuli compared to men, thereby drinking to regulate negative emotion. Future work in the neuroimaging field should consider examining SG differences in stress-related pathophysiology as it relates to AUD, develop radiotracers to answer fundamental questions about differences in stress-related neurochemistry in women and men with AUD, and how SG-dependent alterations in these neural mechanisms may influence clinical correlates of problem drinking and treatment.

Overlap between AUD and Mental Health/Substance Use Disorders

Many of the studies cited within this review analyzed AUD samples diagnosed with other psychiatric comorbidities including various substance use disorders. For example, AUD and tobacco use are highly comorbid. Smoking is associated with region-specific brain volume reductions compared to never smokers, including gray matter volume loss in the PFC, ACC, and other regions overlapping with the neurotoxic effects of alcohol on regional brain volumes (Fritz et al., 2014). Smoking is also associated with reduced cortical thickness of the left medial OFC (Kühn, Schubert, & Gallinat, 2010). Studies have demonstrated that chronic cigarette smoking was associated with larger cortical gray matter volume loss and alterations in cortical thickness in individuals with AUD (Durazzo, Cardenas, Studholme, Weiner, & Meyerhoff, 2007; Durazzo, Mon, Gazdzinski, & Meyerhoff, 2013; Kühn et al., 2010; van Holst, de Ruiter, van den Brink, Veltman, & Goudriaan, 2012). To further complicate the picture, men and women smokers differ with regard to availability of β2-nicotinic acetylcholine receptors (nAChR), to which nicotine binds and activates, and DA D2/3 receptor availability (Cosgrove et al., 2012; Cosgrove et al., 2014; Verplaetse et al., 2018), and it has been shown that alcohol interacts with nAChRs (Larsson & Engel, 2004).

Other neuroimaging work demonstrated that alterations in ventral striatal and amygdala activity in response to stress in problem drinkers was mediated by anxious/depressive symptomatology (Nikolova et al., 2016). Since AUD is also highly comorbid with mood and anxiety disorders, it remains especially important to consider clinical diagnoses in neuroimaging studies of AUD as these comorbidities are often not accounted for in study design or analysis. Overall, neuroimaging work should control for smoking status and other psychiatric comorbidities in studies of AUD, as this may improve our understanding of the differential effects of alcohol vs. tobacco smoking [or depression/anxiety, etc.] on the brain, and parse out whether these differences are SG-dependent. Importantly, future work, including the Adolescent Brain Cognitive Development (ABCD) and National Comorbidity Survey – Adolescent Supplement studies, will also be able to address SG differences in variability of brain development during adolescence as it relates to substance use (e.g., marijuana, e-cigarettes) and beyond (e.g., traumatic brain injury, sleep, psychopathology, psychosocial factors [parental income, family structure]) and how that may be associated with the development and maintenance of AUD in adulthood (Conway, Swendsen, Husky, He, & Merikangas, 2016; Volkow et al., 2018).

Hormones and Neurosteroids

Alterations in sex hormone levels may play a role in differences in neural mechanisms underlying alcohol use in males and females. It is known that progesterone and its metabolites regulate neuronal function and interact with other neurotransmitter systems, including GABA, DA, and serotonin systems, to modulate the effect of alcohol on the brain (for review see (Peltier et al., 2019). Increasing evidence suggests that ovarian hormones account for SG differences in alcohol-motivated behavior, possibly through these mechanisms (Anker & Carroll, 2010; Becker, Perry, & Westenbroek, 2012). Women with consistently greater estradiol levels (mean=375.5 pmol/liter) demonstrate greater alcohol consumption than women with lower estradiol levels (mean=353.3 pmol/liter) (Muti et al., 1998), and this may be related to the mediating role of estradiol on DA efflux in women but not men (Yoest, Cummings, & Becker, 2014).

This may be particularly true in adolescence as greater sex hormone production may be associated with greater alcohol use in adolescence (Braams, Peper, Van Der Heide, Peters, & Crone, 2016; Westling, Andrews, Hampson, & Peterson, 2008). Higher testosterone and estradiol production has also been linked to earlier initiation of drinking in adolescence, particularly in boys (de Water, Braams, Crone, & Peper, 2013), and this relationship may have an impact on brain connectivity. For example, functional connectivity between the amygdala and OFC may be disrupted by testosterone in adolescent alcohol use. Lower functional connectivity between the amygdala and OFC was associated with higher recent and lifetime alcohol consumption, an effect modulated by testosterone levels in boys only (Peters, Jolles, Van Duijvenvoorde, Crone, & Peper, 2015). In other words, higher testosterone levels were associated with lower amygdala-OFC connectivity and increased alcohol use in boys. Findings were not significant for girls even when controlling for contraceptive use (Peters et al., 2015). A subsequent study by the same group found no SG differences in amygdala-OFC connectivity or alcohol use in a larger sample of over 200 adolescents and young adults who also drank alcohol (Peters et al., 2017); although, this study did not examine the modulating effect of sex hormones. In healthy, adult premenopausal women, however, testosterone administration reduced functional coupling of the amygdala and OFC, suggesting that testosterone may produce alterations in regulatory control over the amygdala (van Wingen et al., 2010).

Thus, it appears that sex steroid hormones may modulate neuronal activity and, in turn, may impact alcohol-motivated behavior. Sex hormone-induced disruption of connectivity between the amygdala and OFC may also be associated with greater alcohol use. Nonetheless, the contribution of sex steroid levels in association with alcohol-related perturbations in neural functioning is worth further investigation. Relatedly, we should continue to elucidate the complex associations between alcohol use and aging, as recent findings suggest that moderate drinking was associated with reductions in neuropsychological decline and AUD-related pathology in octogenarian men compared to rare/never drinkers but not in women (Wardzala et al., 2018).

Conclusion

Neuroimaging has informed our understanding of functional and structural alterations related to acute and chronic alcohol use and alcohol use disorder. There is evidence of SG differences across imaging modalities; although this work is limited. Functional and structural imaging data are mixed but common areas susceptible to the neurotoxic effects of alcohol are apparent, including cortico-limbic regions like the prefrontal cortex, anterior cingulate cortex, striatum, and amygdala, as well as alterations in dopamine. Because results on SG differences in the neuroimaging of AUD in women and men are sparse, there is an urgent need for the study of SG differences in alcohol-related alterations in brain chemistry, structure, and function and a need for replication in larger samples of women.

Future experiments should strive to power studies adequately for the analysis of SG in neuroimaging data, including over-recruiting women for adequate female enrollment. Highlighting this point, we found that while 59%, 81%, and 80% of PET, MRI, and fMRI studies of AUD over the past 10 years included women, respectively, only 13%, 18%, and 11% analyzed neuroimaging data by sex. Future consideration should also be made regarding the role of stress pathophysiology when examining SG differences in AUD as AUD and stress-related neurocircuitry overlap. Sex hormones, psychiatric comorbidities, and psychosocial factors may also be associated with perturbations in neural networks underlying AUD in women compared to men; thus, should be considered in future imaging investigations. With the recent mandate on inclusion of sex as a biological variable in all clinical research, SG differences in the neural substrates underlying alcohol use disorder will greatly advance our understanding of SG-dependent alterations in brain function and structure in alcohol use disorder with the goal of tailoring pharmacotherapeutic treatment strategies for alcohol use disorder, particularly in women.

Significance Statement.

Rates of AUD have increased in women by 84% in the last decade, relative to a 35% increase in men. It is critical to understand sex/gender (SG) differences in neural mechanisms underlying AUD. This review aims to explore recent neuroimaging literature on SG differences in brain function and structure as it pertains to alcohol in humans. Current research is limited and mixed regarding SG effects in the neuroimaging of AUD. This may be related to low female enrollment and underpowered analysis to account for SG. Findings underscore the importance of including women in neuroimaging research to guide treatments for AUD.