Pathways to post‐traumatic stress disorder and alcohol dependence: Trauma, executive functioning, and family history of alcoholism in adolescents and young adults

Abstract Introduction Family history (FH) of alcohol dependence is likely to increase the risk of trauma exposure, post‐traumatic stress disorder (PTSD), and alcohol dependence. FH of alcohol dependence and trauma has been separately shown to adversely affect planning/problem‐solving aspects of executive function. However, few studies have examined these risk factors in an integrated model. Methods Using data from trauma‐exposed individuals from the Collaborative Study on the Genetics of Alcoholism prospective cohort (N = 1,860), comprising offspring from alcohol‐dependent high‐risk and comparison families (mean age [SE] = 21.9 [4.2]), we investigated associations of trauma (nonsexual assaultive, nonassaultive, sexual assaultive) with DSM‐IV PTSD and alcohol dependence symptom counts, and planning/problem‐solving abilities assessed using the Tower of London Test (TOLT). Moderating effects of family history density of alcohol use disorder (FHD) on these associations and sex differences were explored. Results Family history density was positively associated with PTSD in female participants who endorsed a sexual assaultive trauma. Exposure to nonsexual assaultive trauma was associated with more excess moves made on the TOLT. Conclusion Findings from this study demonstrate associations with PTSD and alcohol dependence symptom counts, as well as poor problem‐solving ability in trauma‐exposed individuals from families densely affected with alcohol dependence, depending on trauma type, FHD, and sex. This suggests that having a FH of alcohol dependence and exposure to trauma during adolescence may be associated with more PTSD and alcohol dependence symptoms, and poor problem‐solving abilities in adulthood.

sexual assaultive trauma together as assaultive trauma, it may be important to separate these two trauma types. For example , Breslau et al. (1998) showed that the highest risk of PTSD was associated with physical assaultive violence, while Lang et al. (2003) showed that sexual assault was associated with increased substance use (Breslau et al., 1998;Lang et al., 2003). Importantly, trauma type may also moderate the association between PTSD and alcohol dependence. For example, some studies found that individuals exposed to nonassaultive trauma were more likely to initiate substance use earlier compared to those with an assaultive trauma history , while other studies found that assaultive trauma, but not nonassaultive trauma, exposure was associated with increases in binge drinking (Cisler et al., 2011). Thus, consideration of trauma type is critical in better understanding potential adverse consequences following trauma exposure.
There are known sex differences in the rates of trauma exposure, as well as PTSD and alcohol dependence diagnoses (Beals et al., 2013;Boudoukha, Ouagazzal, & Goutaudier, 2017;Chung & Breslau, 2008;Erol & Karpyak, 2015). For example, men are more likely to be exposed to trauma during their lifetime, but less likely to meet criteria for PTSD compared to women (Kessler, Sonnega, Bromet, Hughes, & Nelson, 1995). This could potentially be due to the increased likelihood of women experiencing recurring, high impact traumatic exposures (e.g., childhood sexual assault, physical abuse) (Hien, Cohen, & Campbell, 2005). Further, women with lifetime alcohol dependence are more likely to report lifetime PTSD than men (Kessler et al., 1997). These findings further underscore the need to consider the role of trauma type as well as sex when examining the relationship between PTSD and alcohol dependence.
Given the common comorbidity of PTSD and alcohol dependence (Allen, Crawford, & Kudler, 2016;Goldstein, Bradley, Ressler, & Powers, 2017;Kessler et al., 2005), researchers have proposed several hypotheses linking the two disorders. One common theory is the self-medication hypothesis, which suggests that individuals suffering from PTSD symptoms self-medicate with alcohol and subsequently develop an alcohol use problem (Allen et al., 2016;Goldstein et al., 2017). Another hypothesis is the shared-liability model, which suggests that PTSD and alcohol dependence have shared risk factors (Allen et al., 2016;Goldstein et al., 2017), such as having a family history (FH) of alcohol dependence, neurocognitive and behavioral impairments, and genetic risk. This study used an integrated model to investigate the associations of different types of trauma exposure and FH of AUD on PTSD and/or alcohol dependence as well as planning and problem-solving aspects of executive function, while also investigating sex differences.

Individuals with a FH of alcohol dependence have been shown
to have disproportionately high rates of trauma exposure and subsequent PTSD compared to individuals without a FH of dependence. For example, one study demonstrated that mother's alcohol dependence status was positively associated with offspring PTSD (Bender, Meyers, McCutcheon, & Bucholz, 2016). This may be due to adverse aspects of the home environment, like having an abusive or neglectful parent or relative (Breslau et al., 2013), as previous studies have demonstrated that having parents with alcohol use problems increases risk for exposure to a traumatic event (Breslau et al., 2013;Dube et al., 2001;Meyers et al., 2014). Further, individuals with a FH of alcohol dependence are at risk, not just because of their environment, but also because of their genetic risk for alcohol dependence, which shares a substantial overlap in genetic influences with PTSD. Studies have shown that genetic influences on PTSD may be around 30%-40% (Banerjee, Morrison, & Ressler, 2017;Nugent, Amstadter, & Koenen, 2008;Stein, Jang, Taylor, Vernon, &  (r = .35) and a moderate, but significant correlation between alcohol dependence and PTSD (r = .34) in female participants (Sheerin et al., 2020). Although there are environmental and genetic contributions to both PTSD and alcohol dependence, most studies only examine dichotomous parental alcohol dependence status. A family history density (FHD) measure is useful for disorders influenced by genetic and environmental components, such as PTSD and alcohol dependence, since it can include other affected family members and account for degrees of relatedness (Cservenka, Gillespie, Michael, & Nagel, 2015;Pandey et al., 2020;Powers, Berger, Fuhrmann, & Fendrich, 2017).
Post-traumatic stress disorder and alcohol dependence also share behavioral impairments, including hyperarousal, irritability, and reckless behavior, which are thought to be due to deficits in executive functions (Koob & Volkow, 2016;Logrip, Zorrilla, & Koob, 2012). Executive functions are central in carrying out efficient adaptive and goal-directed behaviors such as inhibition and attentional control, set-shifting and mental flexibility, working memory, and planning and problem-solving (Packwood, Hodgetts, & Tremblay, 2011;Pennington & Ozonoff, 1996). While there are many studies investigating associations between PTSD and aspects of executive functioning (Flaks et al., 2014;Olff et al., 2014), one aspect of executive functioning, planning, and problem-solving abilities have been understudied in individuals with PTSD. In the few studies that have been conducted (Aupperle et al., 2012) results were inconsistent, limited by small sample sizes of individuals with or without PTSD and only a single study investigated a specific type of trauma (Kanagaratnam & Asbjornsen, 2007;Lagarde, Doyon, & Brunet, 2010;Twamley et al., 2009;Vasterling et al., 2002). This suggests that the relationship between trauma, PTSD, and planning and problem-solving needs further investigation in larger samples of male and female participants.
One task that has been used to assess planning and problem-solving aspects of executive functioning is the Tower of London test (TOLT) (Ruocco et al., 2014). Although activation of different brain areas has been observed depending on the different versions of the TOLT, one area consistently activated is the prefrontal cortex, which plays a significant role in executive function by monitoring activities in other cortical and subcortical regions (Baker et al., 1996;Ruocco et al., 2014). Therefore, executive dysfunction in PTSD and alcohol dependence may be partly due to the shared neurocircuitry of PTSD and addiction, involving hypoactivation of prefrontal regions (Aupperle et al., 2012;Koob & Volkow, 2016). A particularly vulnerable period for prefrontal development is adolescence, when the brain undergoes important developmental changes, including synaptic pruning in this region (Juraska & Willing, 2017;Laviola & Marco, 2011). This period also coincides with peak alcohol experimentation shown to impact brain development (Koob & Volkow, 2016).
In part, this may contribute to why trauma experienced during childhood and adolescence has particularly adverse and enduring consequences on the individual's mental health and cognitive functioning (Dube et al., 2003;Hussey, Chang, & Kotch, 2006;Springer, Sheridan, Kuo, & Carnes, 2007). These consequences have been shown to persist throughout the individual's lifespan, increasing risk for mental health problems, including PTSD and alcohol misuse (Follette, Polusny, Bechtle, & Naugle, 1996;Green et al., 2000;Khoury, Tang, Bradley, Cubells, & Ressler, 2010) Further, models of AUD risk differ for adolescents compared to adults (Fowler et al., 2007). For example, heritability estimates, and the relative importance of genetic and environment risk factors change across adolescence and emerging adulthood (Edwards et al., 2017;Edwards & Kendler, 2013), with research demonstrating that the specific social-environmental factors impacting alcohol misuse in adolescence (e.g., access, parental monitoring, peer substance use) differ adulthood.
It is possible that trauma type and gender may also play a role in the relationship between trauma, FH of AUD, PTSD, and executive function. One study by Meyers et al. (2019) showed neurocognitive differences depending on trauma type, such that sexual assaultive trauma, and not nonsexual assaultive or nonassaultive trauma, was associated with decreased rate of change in frontal theta oscillations during response inhibition, particularly among female participants (Meyers et al., 2019). This study suggests that trauma type, specifically sexual assaultive trauma, may affect executive functioning differently among female individuals compared to other trauma types.
Further, another study examined the impact of fathers' alcohol problems and showed that early adolescent boys, but not girls, had poorer performance on a TOLT than boys from control families (Adkison et al., 2013), suggesting sex differences in the association of paternal alcohol use disorder (AUD) with executive function deficits during early adolescence. However, no prior study has examined neurocognitive risk factors for PTSD among male and female adolescents with a FH of alcohol dependence leaving important questions regarding any potential neurocognitive or behavioral deficits associated with trauma experienced among the offspring of alcoholics unanswered.
In the current study, we investigated the influence of FH of AUD, operationalized as FHD, on the associations between trauma type (nonsexual assaultive, nonassaultive, and sexual assaultive), and PTSD/alcohol dependence symptom counts, as well as planning and problem-solving aspects of executive functioning using an integrated model in trauma-exposed individuals from the Collaborative Study on the Genetics of Alcoholism (COGA) prospective cohort (N = 1,860). Secondary analyses included sex-stratified models. We hypothesized that FHD would moderate these associations depending on the trauma type, such that as FHD increases individuals with high impact traumatic exposures (i.e., sexual assaultive trauma) will have more PTSD symptoms, and that this will also vary by sex since women are more likely to experience sexual assaultive trauma than men.

| Sample
This study utilized data from the COGA prospective study, which has been ongoing since 2004. Details on data collection and procedures have previously been published .
The COGA prospective study comprises offspring, both with and without alcohol use problems, from alcohol-dependent high-risk and community comparison families who were 12-22 years of age at enrollment and have had multimodal assessments (clinical, behavioral, neuropsychological, neurophysiological) approximately every 2 years. For the purpose of this cross-sectional study, data from the last interview were used for each individual. Only traumaexposed individuals (N = 1,860) were included in this study since exposure to a traumatic event had to be present in order for an individual to be asked questions concerning PTSD symptoms. The sample was 47.1% female with a mean age of 21.9 (SD = 4.2) at the most recent interview.

| Ethical statement
Informed consent was obtained from each participant, and experimental protocols were in accordance with US Federal Policy of protection of human subjects ethical guidelines and approved by each site's Institutional Review Board. Parental permission was obtained for all participants under the age of 18.

| Trauma exposure and DSM-IV PTSD and alcohol dependence symptom counts
Trauma exposure was defined using the Semi-Structured Assessment for the Genetics of Alcoholism (SSAGA-IV) interview, which consists of questions regarding potentially traumatic life events involving physical and sexual violence, witnessing violence, and witnessing natural disasters and accidents. Consistent with previous studies (Bender et al., 2016;McCutcheon et al., 2017;Meyers et al., 2019), three types of traumatic exposures were considered: nonsexual assaultive, nonassaultive, and sexual assaultive trauma.
These were lifetime binary variables (0 = no; 1 = yes). Nonsexual assaultive trauma was defined as having been shot, stabbed, mugged or threatened with a weapon, kidnaped, or held captive and/or tortured. Nonassaultive trauma was defined as unexpectedly discovering a dead body, experiencing a natural disaster, or experiencing a life-threatening accident. Sexual assaultive trauma was defined as being raped or otherwise sexually assaulted by a relative or nonrelative. DSM-IV PTSD symptom counts were recorded if an individual endorsed one or more traumatic experiences. A maximum endorsement of 17 symptoms were possible in for the three PTSD symptom criteria defined by the DSM-IV: five re-experiencing symptoms, seven avoidance symptoms, and five arousal symptoms (American Psychiatric Association, 2000). Individuals could endorse a maximum of seven symptoms for DSM-IV alcohol dependence. As only data on DSM-IV PTSD were available, DSM-IV alcohol dependence was used to remain consistent for both PTSD and alcohol dependence.

| Family history density of AUDs (FHD)
Family history of DSM-5 AUD was ascertained for adult subjects and adult family members through the SSAGA, and adolescents through the child and adolescent tailored CSSAGA (Bucholz et al., 1994). A FHD ratio was calculated for each individual from both alcohol and community control families, with reported information for four or more relatives (N = 1,704). We note that while those from community comparison families had lower FHD scores on average, com- analyses. Details regarding the calculation of FHD can be found in Pandey et al. (2020). Briefly, affectedness was defined as the individual's first-and second-degree relatives' (parents, grandparents, parental siblings, full siblings, half-siblings) AUD symptom count values divided by the maximum possible value of 1 (range from 0 to 1). Weighting was applied, such that the weight decreases with the increase in degree of relatedness. This measure, referred to as FHD in this study, has been found to be a sensitive indicator of individuals with DSM-IV alcohol dependence (Pandey et al., 2020).

| Tower of London test
A computerized version of the TOLT was administered using the Colorado assessment tests (CATs) for cognitive and neuropsychological assessment (Davis, 2002). The TOLT was used to assess planning and problem-solving skills (Shallice, 1982) and has been described in a previous publication (Pandey et al., 2018). Briefly, participants were required to move colored beads on a computer screen from a starting position to a goal position in as few one-byone moves as possible. The test consisted of 21 trials with 3, 4, and 5 colored beads placed on respective number of pegs, with seven problems per puzzle-type. Derived performance measures were available for 1,860 trauma-exposed individuals and included average trial time (ATRTI) and excess moves made (EM) for each puzzletype separately and once across all puzzle-types. For the purpose of this study, TOLT performance measures across all puzzle types were used. Consistent with all other measures in this study, data from the last TOLT evaluation were used.

| Statistical methods
A primary path analysis was performed using Mplus (Muthén & Muthén, 1998 to investigate main and interaction effects of FHD and three types of trauma (nonsexual assaultive, nonassaultive and sexual assaultive trauma) for DSM-IV PTSD symptom counts, alcohol dependence symptom counts, and TOLT performance measures (average trial time and excess moves made. Since it is possible that an individual may have multiple types of trauma exposure, all trauma variables were modeled simultaneously. All dependent variables were also modeled simultaneously to account for possible shared liability among PTSD and AD symptom counts and TOLT performance measures (Figure 1). All statistical models were adjusted for familial clustering, sex, self-reported race/ethnicity, age, income, and educational attainment (i.e., highest grade completed), as each of these variables has been previously shown to influence risk for PTSD and alcohol dependence (Powers, Etkin, Gyurak, Bradley, & Jovanovic, 2015;Tripp, McDevitt-Murphy, Avery, & Bracken, 2015).
Since subjects improve in behavioral tasks over time, and not all individuals have participated in the same number of sessions, number of sessions for each individual was used as a covariate on the TOLT performance measure pathways. Secondary analyses examined the same path model described above stratified by sex, to investigate potential sex differences in all associations.

| RE SULTS
Rates of traumatic exposures in COGA's prospective study have been previously described McCutcheon et al., 2017;Meyers et al., 2019). Briefly, exposure to nonassaultive trauma was the most prevalent (N = 1,479; 79.5%) followed by nonsexual assaultive trauma (N = 916; 49.2%) and sexual assaultive trauma (N = 364; 19.6%). Rates of sexual trauma were greater among female participants (N = 306 vs. N = 58), and rates of nonsexual assaultive trauma were greater among male participants (N = 616 vs. N = 300). 36.7% of individuals who reported experiencing at least one of the three trauma types endorsed at least one PTSD symptom, and 56.1% endorsed at least one DSM-IV alcohol dependence. See Table 1 for descriptive statistics.
F I G U R E 1 Pathway model used to investigate main and interaction effects of trauma exposure and family history density of alcohol use disorder (FHD) with post-traumatic stress disorder and alcohol dependence symptom counts, and planning and problem-solving measures from the Tower of London Test (average trial times, excess moves made). Note: Not pictured but are also included in this model are covariates: sex, self-reported race/ethnicity, age, income, educational attainment (i.e., highest grade completed), number of TOLT sessions.
Consistent with prior studies, significant main effects of nonsexual assaultive, nonassaultive, and sexual assaultive trauma were observed for DSM-IV PTSD symptom counts such that having any of the three types of trauma was positively associated with PTSD symptom counts (Table 2; Figure 2). Significant main effects of nonsexual assaultive and nonassaultive trauma were observed for DSM-IV alcohol dependence symptom counts such that having either trauma type was positively associated with alcohol dependence symptom counts (Table 2; TA B L E 1 Demographic information, trauma history, mean DSM-IV post-traumatic stress disorder (PTSD) and alcohol dependence (ALC) symptom counts, and mean family history density ratio of alcohol use disorder (AUD) among trauma-exposed individuals in COGA's prospective study  In the sex-stratified analysis, significant associations were observed such that all three types of trauma were positively associated with PTSD symptom counts in both male and female participants (Table 3; Figure 3). The significant associations of trauma type and FHD observed in the full sample were only observed in male participants, such that male participants, but not female participants, who reported a nonsexual assaultive and nonassaultive trauma, as well as increasing FHD, had more alcohol dependence symptom counts (Table 3; Figure 3a). A significant moderating effect of FHD on the association between sexual assaultive trauma and PTSD symptom counts was observed in female participants, but not male participants, such that sexual assaultive trauma was positively associated with PTSD symptom counts as FHD increases in female participants only (Table 3; Figure 3b).

TA B L E 2 Main and interaction effects
A main significant effect was observed for nonsexual assaultive trauma and the TOLT excess moves made, such that having a nonsexual assaultive trauma was positively associated with excess moves made (i.e., more excess moves made; Table 2, Figure 2). In the sex-stratified analysis, this association was only observed in female participants (Table 3, Figure 3b). No significant associations were observed for average trial time in the full or sex-stratified samples.

No moderating effects of FHD on the associations between trauma
and TOLT performance measures were observed in either the full or sex-stratified analyses.

| D ISCUSS I ON
Consistent with prior research, trauma exposure was positively associated with PTSD and alcohol dependence symptom counts, and FHD was positively associated with alcohol dependence symptom counts. While FHD did not moderate the associations between trauma and PTSD in the full sample, a sex-stratified analysis showed that the association between sexual assaultive trauma and PTSD symptom counts was moderated by FHD in female participants only.
One trauma type, nonsexual assaultive trauma, was associated with more TOLT excess moves made, which was shown in the sex-stratified analysis to be driven by the female participants.

| Sex differences in the moderating effects on the association of trauma and DSM-IV PTSD symptom counts
This study extends previous work demonstrating that trauma is associated with risk for PTSD and alcohol dependence, as well as F I G U R E 2 Associations of trauma and family history density of alcohol use disorder (FHD) with post-traumatic stress disorder symptom counts, alcohol dependence symptom counts, TOLT average trial times, and TOLT excess moves made. Note: Only significant pathways are displayed. Not pictured but are also included in this model are interaction variables (non-sexual assaultive trauma X FHD, non-assaultive trauma X FHD, sexual assaultive trauma X FHD) and covariates (self-reported gender, race/ethnicity, age, income, educational attainment (i.e., highest grade completed), number of TOLT sessions). TOLT, Tower of London Test associations between FHD and the risk for alcohol dependence (Breslau et al., 2013;Giaconia et al., 1995;Knopik et al., 2004;Pandey et al., 2020), by examining these important risk factors together with measures of executive functioning in an integrated model. Findings from this study demonstrate that the association between trauma, specifically sexual assaultive trauma, and PTSD is moderated by FHD, such that increasing FHD increased the association with PTSD in female participants with sexual assaultive trauma, Since research has shown that sexual assault is most commonly associated with PTSD in women (Kessler et al., 1995), it is possible that other risk factors for PTSD development, such as a FH of alcohol dependence, may increase the risk for developing PTSD to a greater extent or exacerbate the effects of exposure to sexual assaultive trauma in women, which could explain our observed result that increasing FHD of alcohol dependence increased the association with PTSD in female participants with sexual assaultive trauma.
Prior studies have shown that children who grow up in a household with at least one parent with alcohol dependence are more likely to experience trauma compared to children with unaffected parents (Dube et al., 2001;Meyers et al., 2014). However, these studies have only examined dichotomous parental alcohol dependence status, whereas FHD measures may be more useful (Cservenka et al., 2015;Pandey et al., 2020;Powers et al., 2017). Future studies should investigate how different levels of FHD may influence PTSD symptoms and risk, as well as examine molecular genetic risk of alcohol dependence on PTSD in female individuals who have been exposed to a sexual assaultive trauma.

| Association of trauma and FHD on planning and problem-solving aspects of executive function
Overall, our findings suggest that nonsexual assaultive trauma was associated with excess moves made, which may suggest poor TA B L E 3 Sex differences in main and interaction effects of trauma exposure, FHD and DSM-IV PTSD and alcohol dependence (ALC) symptom counts, and TOLT performance measures (average trial time [ATRTI]; excess moves made [EM])  was associated with poor performance, using an executive function composite score. Further, Malarbi et al. (2017) demonstrated that trauma-exposed children showed cognitive deficits, especially those with a PTSD diagnosis. However, these studies have been limited by sample sizes (e.g., Bucker et al. (2012) recruited 30 trauma-exposed children and 30 age-and sex-matched controls) or available data on trauma type (Corbo et al., 2016;Malarbi et al., 2017). One study has shown that sexual trauma, and not physical or interpersonal trauma, moderates psychotherapy outcome for PTSD (Markowitz, Neria, Lovell, Van Meter, & Petkova, 2017), which underscores the impor- Given the heterogeneous nature of PTSD and the results of our study, perhaps one specific pathway to developing PTSD symptoms could be investigated by linking trauma type, specifically nonsexual assaultive trauma, to cognitive function. Individuals who have been exposed to nonsexual assaultive trauma are more likely to develop PTSD compared to those exposed to nonassaultive trauma (Cisler et al., 2011;Kessler et al., 2005;Ozer et al., 2003;Resnick et al., 1993), and poor executive functioning has been observed among individuals with PTSD (Flaks et al., 2014;Olff et al., 2014

| CON CLUS IONS
Our study has shown that increasing FHD of alcohol dependence in sexual assaultive trauma-exposed female participants was associated with increased PTSD symptom counts. Nonsexual assaultive trauma was associated with poorer TOLT task performance in our full sample. In the sex-stratified analysis, this association was only observed in female participants. These findings suggest that specific types of traumatic exposure are more likely to be associated with PTSD and planning and problem-solving aspects of executive function. Future studies should explore whether any observed poor planning/problem-solving performance precedes or follows trauma exposure, considering trauma type and sex differences. Further, future studies should investigate whether these aspects of executive function precede or follow PTSD and/or alcohol dependence onset.
Understanding the complex etiology of these commonly comorbid disorders may inform early intervention and treatment strategies aimed at reducing the severity and endurance of PTSD and alcohol dependence. Michael Conneally, PhD, Raymond Crowe, PhD, and Wendy Reich, PhD, for their critical contributions. This national collaborative study is supported by NIH grant U10AA008401 from the NIAAA and the NIDA.

CO N FLI C T O F I NTE R E S T
There are no conflicts of interest to disclose.

PEER R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1002/brb3.1789.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.