Alcohol is the most widely used intoxicant among adolescents and young adults, with rates of binge drinking and alcohol use disorders peaking between ages 18 and 25 (SAMHSA, 2016), a period commonly referred to as emerging adulthood (Arnett, 2000). Recent national surveys revealed that 39% of 18–25-year-olds reported past-month binge drinking, compared to 25% of adults ages 26 and older (SAMHSA, 2016). This age range also encompasses the final stages of neuromaturation (Gogtay et al., 2004), which may be negatively affected by heavy alcohol use.

Age-related refinements in brain structure and function continue well into emerging adulthood. Region-specific patterns of cortical thinning and gray matter volume reduction have been identified, with higher order association cortices, including prefrontal and temporal regions, maturing last, and varying by sex (Gogtay et al., 2004). Measureable changes in gray matter volume have been observed over a 6-month period in first-year college students, between fall and spring semesters (Bennett and Baird, 2006). These neuromaturational changes are thought to underlie cognitive improvements throughout this age range. In particular, executive functions, subserved by late-developing frontal lobe regions, show dramatic enhancements in late adolescence and emerging adulthood (Casey et al., 2000).

Several neuroimaging investigations have characterized structural brain differences among adolescent drinkers, yet fewer investigations have focused on emerging adults. Extant research has most consistently implicated frontal lobe abnormalities among adolescent drinkers (Silveri et al., 2016). Cross sectional studies have revealed smaller volumes and thinner cortices among adolescent heavy drinkers, particularly in frontal, temporal, and cingulate regions (De Bellis et al., 2005; Medina et al., 2008; Fein et al., 2013; Whelan et al., 2014), with greater alcohol consumption associated with larger volume reductions (De Bellis et al., 2005). A recent voxel-based morphometry study of young adults ages 22–28 identified less gray matter volume among heavy drinkers in anterior cingulate, orbitofrontal, temporal, and insula cortices (Heikkinen et al., 2017). Longitudinal investigations have examined adolescents both before and after the onset of heavy drinking, and described steeper declines in cortical thickness in prefrontal (Luciana et al., 2013; Squeglia et al., 2015) and temporal (Squeglia et al., 2014, 2015) regions in alcohol initiators. One diffusion tensor imaging study ascertained white matter integrity in binge drinkers and controls during late adolescence, and again 3 years later during emerging adulthood. Lower fractional anisotropy was observed at baseline among binge drinkers throughout widespread brain regions (Jacobus et al., 2009; McQueeny et al., 2009), many of which showed additional decline among those who continued binge drinking during the 3-year follow-up period (Jacobus et al., 2013). However, to our knowledge, this is the first longitudinal study to examine trajectories of gray matter volume development in an attempt to shed more light on implications of continued heavy drinking among emerging adult drinkers.

To delineate the influence of ongoing heavy alcohol use on brain gray matter development in emerging adulthood, we collected structural magnetic resonance imaging (MRI) scans at two time-points in a college cohort, at approximately age 18 and again ~2 years later at age 20. We conducted longitudinal voxel based morphometry (VBM) analyses in individuals who were either abstainers/light drinking controls or heavy drinkers at both time-points (i.e., sustained light use or sustained heavy use). Given that cortical pruning is part of the normal developmental process at this age, we predicted that both groups would show significantly reduced gray matter over time; however, sustained heavy drinkers would show a greater decline in gray matter volume compared to sustained light drinkers, especially in frontal and temporal brain regions.

Groups did not differ by age, smoking, FHA or any MINI diagnostic status. Groups also had similar scores on the STAI and BDI. However, groups differed by sex (chi-square = 4.47; p = 0.03). Complete description of demographics and clinical characteristics of study participants are provided in Table 1.

Within-group analysis revealed significant GMV loss in both drinking groups over time (see Figure 1). No significant regional expansion of GMV over time was detected in either group. GMV decline in stable light users was mostly localized to inferior/middle temporal and fusiform gyrus. Gray matter decreases in individuals with sustained heavy alcohol use were lager in magnitude, and included the above regions plus several other brain regions including inferior/middle frontal gyrus (I/MFG), anterior cingulate, insula, thalamus, caudate, and parahippocampal gyri. Longitudinal GMV differences assessed across groups (see Figure 2) showed the heavy use group to have a significantly increased rate of gray matter decline, primarily in fronto-striatal regions, including inferior/medial frontal, anterior cingulate, parahippocampus, precentral gyrus, and insula. A complete description of within-group and between-group results is presented in Tables 2A,B. Further, we detected no group-by-sex or group-by-sex-by-time interactions at the whole brain level. Exploratory analysis of between-group baseline differences in GMV revealed significantly decreased GMV in the heavy drinking group in a subset of regions that also showed exaggerated decline across time. Overlap of regions was mainly localized to superior, medial and inferior frontal gyri after correction for multiple comparisons. Regions showing baseline differences are visually represented in Figure 3.

The present longitudinal study investigated gray matter volume changes over a 2 year period in emerging adult college students who were either sustained light-moderate drinkers or who had continued heavy alcohol use. Consistent with our initial hypotheses, we noted significant brain GMV loss in both groups. However, a direct two group comparison showed a significantly greater rate of loss in students with a sustained pattern of heavy alcohol use over 2 year follow-up. Moreover, this pattern of greater gray matter decline in heavy users occurred largely in regions responsible for emotion, memory, mental flexibility, and decision making, all of which might have a direct impact on student college success (Best et al., 2011).

It is important to recall that cortical pruning and restructuring reflects a normal, beneficial process during adolescence and young adulthood (Gogtay et al., 2004). Although the results noted in abstainers/light drinkers might closely reflect normal developmental processes, the exaggerated GMV-c due to continued heavy drinking would not be fully attributable to such cortical maturation. One possible interpretation could be that alcohol use leads to excess gray matter volume loss similar to that observed in adult alcoholics (Suzuki et al., 2010; Ersche et al., 2013; Sjoerds et al., 2013; Segobin et al., 2014; Droutman et al., 2015). The above hypothesis is supported by animal research that suggests alcohol interferes with neural stem cell proliferation leading to neurodegeneration and changes in brain structure and function (Morris et al., 2010). Animal reports have also shown marked differences in survival of cortical-cells in adolescent animals that were exposed to alcohol compared to controls (Crews et al., 2006; Hansson et al., 2010; Koss et al., 2012). Consistent with the above findings, changes observed in human macro-structural MRI studies such as ours may thus reflect tissue loss or remodeling related to inhibition of cell generation and survival. Preclinical studies also suggest that heavy alcohol intoxication might trigger microglia activation, oxidative stress, and pro-inflammatory changes which in turn leads to neurotoxic degeneration and/or prevents genesis of neurons and glia (Crews and Nixon, 2009; Alfonso-Loeches et al., 2012). Our results are also very consistent with a recent study that used region of interest based analysis to investigate similar longitudinal gray and white matter changes in adolescents before and after they initiated heavy alcohol use (Squeglia et al., 2015). That study observed a similar pattern of accelerated GMV-c in heavy alcohol initiators along with an attenuated growth of several white matter structures. The current study expands on these findings, demonstrating that continued heavy drinking in emerging adulthood leads to excessive gray matter loss in brain structures relevant to everyday functioning. Also, compared to the above study, we employed a voxel-wise whole brain analysis that should enable us to visualize GMV-c with greater spatial specificity.

Previous studies have shown that disruptions to adolescent brain maturation might adversely impact developmental cognitive and motor performance (Burgaleta et al., 2014). In the current study we noted both baseline decreases and accelerated GMV decline of the parahippocampal gyrus (a region primarily responsible for memory function) in the heavy drinking group, which is consistent with and expands results from cross-sectional reports showing reduced GMV in this medial temporal region for both individuals with alcohol dependence and those with family history of alcohol dependence (Suzuki et al., 2010; Sjoerds et al., 2013), The significant interaction of alcohol use on gray matter volume development as shown here might therefore help explain previous longitudinal studies showing worse attention, working memory, and visuospatial performance in adolescents after initiating alcohol use (Tapert et al., 2002; Hanson et al., 2011).

Interestingly, many of the significantly affected regions overlap with brain systems involved in reward, decision making, emotion and stress, that have been directly linked to addictive behavior (Schoenbaum and Shaham, 2008; Sinha, 2011; Droutman et al., 2015; O'Connor and Kenny, 2015; Silveri et al., 2016). Previous functional studies have shown increased activation to alcohol cues in medial frontal/ACC that underlie attention and motivation (Myrick et al., 2004; Dager et al., 2013). Altered interactions between ACC and parts of the striatum (also implicated in our current study) have shown to be related to motivational shift toward drug-induced cues, thus suggesting potential biological markers of addictive pathology (Myrick et al., 2004; Dager et al., 2013). Another key region identified in our study was inferior frontal gyrus (IFG), which functionally signals drug availability and triggers anticipation of use (Schoenbaum and Shaham, 2008). IFG over-reactivity has also been prospectively linked to escalating drinking in a previous study that used a subset of the current dataset (Worhunsky et al., 2016). It is interesting to note that our supplementary analyses show that some of these frontal differences existed at baseline and seemingly became worse over time. Insula, another region that showed exacerbated GMV loss in our heavy alcohol users, has been shown to play an important role in drug addiction (Droutman et al., 2015). Meta analyses of structural abnormalities in addicted individuals have described significantly reduced insular volume (Ersche et al., 2013). Functional studies have repeatedly observed blunted insular activity in substance-dependent individuals when engaged in decision-making tasks, and have also been shown to predict relapse (Sinha, 2011). The substantial overlap of excess gray matter loss with critical brain systems implicated in addiction might suggest that some of these differences may have been pre-existing. Compulsive drug taking due to poorly modulated decision making as a consequence of inability to learn from negative consequences is a hallmark of addiction. Baseline differences noted in our study were primarily limited to the pre-frontal cortex (PFC) which plays a key role in such decision making processes among other critical executive functions (George and Koob, 2010). Indeed, frontal lobe volumetric differences may be both a pre-existing risk factor as well as consequence of heavy alcohol use in adolescence (Silveri et al., 2016). It is also possible that such volumetric changes might contribute to differential neurochemical transmission sensitivity to alcohol that might in-turn enhance vulnerability to addictive-like behaviors into young adulthood. Given the longitudinal design here, our findings support the view that exacerbated gray matter loss is a consequence of continued heavy use. Although sustained heavy use might have an adverse impact on neurodevelopment, many studies in adolescent and adult alcoholics have demonstrated that certain brain structures and pathways show potential for recovery with alcohol abstinence or curtailment (Gazdzinski et al., 2005; Durazzo et al., 2011, 2016; Dresler et al., 2012; Segobin et al., 2014). Future longitudinal studies should determine the degree of volumetric recovery if drinking is reduced during earlier phases of brain development.

A number of additional factors in college including illicit substance use, stress levels, academic performance, peer pressure and psychiatric status could have an interactive effect with alcohol consumption (Meda et al., 2017). Although the concurrent use of other commonly abused substances such as marijuana were not analyzed here (due to excessive missing rate ~40%), prior studies that performed similar analyses to ours noted no significant interactions between alcohol and drug use with gray matter loss patterns (Squeglia et al., 2014). We measured personality and psychiatric metrics such as the STAI, BDI, MDD, and ADHD and found them to be balanced across our diagnostic groups. Also, cigarette smoking, which is often co-morbid with alcohol use and has been associated with gray matter changes (Franklin et al., 2014) was not significantly different between our groups. The proportions of smokers/non-smokers in our sample can be seen in Table 1. Given that previous cross-sectional studies have suggested differential alcohol effects exerted across sexes (Squeglia et al., 2014, 2015), we looked at potential group by sex and group by sex by time interactions in our dataset, but found none.

Despite its current strengths, our study had several limitations to consider. We did not have sufficient numbers of individuals who transitioned from heavy to light or light to heavy alcohol use, and were therefore unable to examine gray matter changes associated with other trajectories of alcohol use that are of great interest. We also did not capture longitudinal behavioral data that might help better understand relationships between GMV-c and cognitive functioning, which might be a good avenue for future studies. There may be other important interacting variables, such as stress levels, personality factors, or other substance use, which were not examined here but would be critical areas for future inquiries. We did not scan participants before the onset of drinking, making it difficult to determine whether gray matter volume differences were pre-existing; future studies such as the ongoing ABCD project will help delineate patterns of gray matter development in relation to varying patterns of substance use initiation and escalation.

In summary, our study provides important longitudinal evidence regarding excessive gray matter loss in regions related to cognitive control, emotional regulation, and memory that might have both short and long-term implications on student life success. Importantly, regions of abnormal gray matter decline overlapped with those implicated previously in substance abuse, thus providing more evidence toward the theory that initiation and continued alcohol use during young adulthood might confer vulnerability for future, ongoing substance dependence.

SM and AD contributed to design, analysis, and drafting of the manuscript. KH, HT, SR, RW, CA, CF, and GP contributed to the study funding, design, and drafting of the manuscript.