Cumulative trauma load and timing of trauma prior to military deployment differentially influences inhibitory control processing across deployment

Military personnel experience high trauma load that can change brain circuitry leading to impaired inhibitory control and posttraumatic stress disorder (PTSD). Inhibitory control processing may be particularly vulnerable to developmental and interpersonal trauma. This study examines the differential role of cumulative pre-deployment trauma and timing of trauma on inhibitory control using the Go/NoGo paradigm in a military population. The Go/NoGo paradigm was administered to 166 predominately male army combat personnel at pre- and post-deployment. Linear mixed models analyze cumulative trauma, trauma onset, and post-deployment PTSD symptoms on NoGo-N2 and NoGo-P3 amplitude and latency across deployment. Here we report, NoGo-N2 amplitude increases and NoGo-P3 amplitude and latency decreases in those with high prior interpersonal trauma across deployment. Increases in NoGo-P3 amplitude following adolescent-onset trauma and NoGo-P3 latency following childhood-onset and adolescent-onset trauma are seen across deployment. Arousal symptoms positively correlated with conflict monitoring. Our findings support the cumulative trauma load and sensitive period of trauma exposure models for inhibitory control processing in a military population. High cumulative interpersonal trauma impacts conflict monitoring and response suppression and increases PTSD symptoms whereas developmental trauma differentially impacts response suppression. This research highlights the need for tailored strategies for strengthening inhibitory control, and that consider timing and type of trauma in military personnel.


Results
Table 1 provides participant demographics, trauma exposure and PTSD symptoms.Refer to Supplementary Materials for full statistical comparison of demographics by trauma onset.Participants were predominately males aged in their late twenties.The ANOVA and post-hoc t-test showed the adult-onset trauma group was significantly older than the no trauma (p < 0.001) and adolescent-onset (p = 0.010) groups, and had significantly more times deployed and higher combat exposure than other trauma-onset groups (Times Deployed: child p = 0.006, adolescent p < 0.001, no trauma p < 0.001; Combat: child p = 0.005, adolescent p = 0.014; no trauma p < 0.001).The adolescent-onset trauma group had significantly more times deployed than the no trauma group (p = 0.046).The ANOVA and post-hoc t-test of cumulative trauma types (CTT), particularly interpersonal CTT, showed this was significantly higher in the childhood-onset than the adolescent-onset and adult-onset trauma groups (Overall: p = 0.042 and p = 0.003 respectively; and Interpersonal: p = 0.014 and p = 0.004 respectively).

Model fit comparison
Table 2 provides the model fit comparison from the F-test based on the Kenward-Roger's approach by ERP component.For N2 amplitude and latency, there was a significant improvement by adding the interaction of Deployment with Interpersonal and Non-interpersonal CTT.There was no improvement by adding the Deployment*Traumaonset interaction nor a 3-way interaction with PTSD.For P3 amplitude and latency, there was a significant improvement by adding the interaction of Deployment*Interpersonal CTT and Deployment*Non-interpersonal CTT, and Deployment*Trauma-onset.There was no improvement with adding a 3-way interaction for PTSD.

N2 component
The linear mixed model showed significant predictors of N2 amplitude were Site, Deployment, and 2-way interactions of Deployment*Interpersonal CTT and Deployment*Non-interpersonal CTT (Table 3).The fixed effects explained 7% of the variance in N2 Amplitude.Supplementary Materials provides a full breakdown of post-hoc analyses.
After accounting for other variables in the linear mixed model, the main effect of Site had a significant and small effect on N2 amplitude, where amplitude at FCz was 0.78mV greater than Cz (b = − 0.78, SE = 0.22, p = 0.001, Cohen's d = 0.17).The Deployment*Interpersonal CTT interaction (Fig. 1) shows that after accounting for other variables in the linear mixed model, change in N2 amplitude with increasing interpersonal CTT was larger at pre-deployment compared to post-deployment and the effect was small (b = − 0.30mV, p = 0.018, Cohen's d = 0.06).N2 amplitude at pre-deployment decreased by 0.34mV, 95%CI (− 0.11, 0.79) with every 1 additional interpersonal CTT with little change at post-deployment (b = − 0.04mv, 95%CI (− 0.42, 0.50)).The Deployment*non-interpersonal CTT interaction (Fig. 1) shows that after accounting for other variables in the linear mixed model, change in N2 amplitude with non-interpersonal CTT was significantly larger at pre-deployment than post-deployment and the effect was small (b = − 0.96mV, p = 0.009, Cohen's d = − 0.20).N2 amplitude at pre-deployment increased by 0.99mV, 95%CI (− 2.24, 0.27) with every 1 trauma increase in non-interpersonal CTT with little change at post-deployment (b = -0.03mV,95%CI (− 1.30, 1.24)).
The linear mixed model for N2 latency showed significant predictors were Age, Deployment, and Deployment*Non-interpersonal CTT interaction (Table 3).Supplementary Materials provides a full breakdown of post-hoc analyses.After accounting for other variables in the linear mixed model, the main effect of Age had a significant yet small effect on N2 latency, with N2 latency increasing by 0.82ms with every one-year increase in Age at pre-deployment (b = 0.82, SE = 0.26, p = 0.002, Cohen's d = 0.03).The Deployment*Non-interpersonal CTT interaction (Fig. 2) shows that after accounting for other variables, change in N2 latency with increasing non-interpersonal CTT was significantly larger at pre-deployment compared to post-deployment and the effect was small (b = 7.42, p < 0.001, Cohen's d = 0.31).N2 latency at pre-deployment increased by 7.55ms, 95%CI (1.

P3 component
The linear mixed model for P3 amplitude showed a significant main effect of Site, and Non-interpersonal CTT, and 2-way interactions of Deployment*Interpersonal CTT and Deployment*Non-interpersonal CTT, and Deployment*Trauma-onset (Table 4).The fixed effects explained 8% of the variance in P3 Amplitude.Supplementary Materials provides a full breakdown of post-hoc analyses.
After accounting for other variables in the linear mixed model, the main effect of Site had a significant and small effect on P3 amplitude, where amplitude at Cz was 1.51mV greater than at Fz (b = 1.51,SE = 0.24, p < 0.001,    www.nature.com/scientificreports/pre-deployment decreased by 0.05mV, 95%CI (− 0.50, 0.39) with every 1 additional Interpersonal CTT compared to 0.39mV, 95%CI (− 0.84, 0.06) at post-deployment.The Deployment*Non-interpersonal CTT interaction (Fig. 3) shows that after accounting for other variables in the linear mixed model, change in P3 amplitude with increasing non-interpersonal CTT was significantly larger at pre-deployment compared to post-deployment and the effect was small (b = − 0.69, p = 0.036, Cohen's d = − 0.17).P3 amplitude at pre-deployment decreased by 1.39mV, 95%CI (− 0.2.55, − 0.23) with every 1 trauma increase in non-interpersonal CTT compared to 0.70mV, 95% CI (− 1.88, 0.47) at post-deployment.The Deployment*Trauma-Onset interaction (Fig. 4) shows that after accounting for other variables in the linear mixed model, there was a significant decrease in P3 amplitude for the no trauma group (b = 1.37,SE = 0.49, p = 0.006) and significant increase in P3 amplitude for the adolescent-onset group (b = -0.97,SE = 0.39, p = 0.013).Change in P3 amplitude across deployment was greater in the no trauma group compared to the adolescentonset (b = 2.34, SE = 0.70, p = 0.001) and adult-onset (b = 1.28,SE = 0.60, p = 0.034) groups, and adolescent-onset was greater than the adult-onset group (b = 1.06,SE = 0.51, p = 0.037).There was no difference in P3 amplitude between trauma groups at pre-or post-deployment (refer to Supplementary Materials).
The linear mixed model for P3 Latency showed a significant main effect of Site, Deployment, and 2-way interactions of Deployment*Interpersonal CTT, Deployment*Non-interpersonal CTT and Deployment*Trauma-Onset.The fixed effects explained 6% of the variance in P3 latency (Table 4).Supplementary Materials provides a full breakdown of post-hoc analyses.
After accounting for other variables in the linear mixed model, the main effect of Site had a significant and large effect on P3 latency, where latency at Pz was 6.The Deployment*Interpersonal CTT interaction (Fig. 5) shows that after accounting for other variables in the linear mixed model, P3 latency increased with increasing Interpersonal CTT at pre-deployment compared to decreasing at post-deployment and the effect was moderate (b = 1.78, p = 0.015, Cohen's d = 0.43).P3 latency at pre-deployment increased by 0.47ms, 95%CI (− 1.56, 0.2.50) with every 1 additional Interpersonal CTT whereas it decreased by 1.31ms, 95%CI (− 0.3.38,0.77) at post-deployment.Similarly, the Deployment*Non-interpersonal CTT interaction (Fig. 5) shows P3 latency increased with increasing Non-interpersonal CTT at pre-deployment compared to decreasing at post-deployment and the effect was large (b = 4.95, p = 0.013, Cohen's d = 1.19).P3 latency at pre-deployment increased by 2.23ms, 95%CI (− 3.1, 7.55) with every 1 trauma increase in Noninterpersonal CTT compared to a decrease at post-deployment of 2.72, 95%CI (− 8.15, 2.71).
The Deployment*Trauma-onset interaction (Fig. 6) shows that after accounting for other variables in the linear mixed model, there was a significantly longer P3 latency for the childhood-onset (b = 12.17, SE = 3.54, p < 0.001) and adolescent-onset (b = 7.90, SE = 2.36, p < 0.001) groups across deployment and change in P3 latency was significant greater than the adult-onset trauma group (Childhood: b = 12.91, SE = 4.04, p = 0.002; Adolescent: b = 8.64, SE = 3.08, p = 0.005).There was no difference in P3 latency between trauma groups at pre-or postdeployment (refer to Supplementary Materials).

Post traumatic stress disorder symptoms (PCL) and sub-cluster correlations
Table 5 outlines correlations between PCL and sub-clusters with ERP component and cumulative trauma load at pre-and post-deployment.At pre-deployment, smaller N2 amplitude was associated with higher arousal symptoms and higher interpersonal CTT was associated with higher overall PCL and higher re-experiencing symptoms.

Discussion
This paper was the first to our knowledge to examine the impact of pre-deployment timing and cumulative type of trauma and posttraumatic stress disorder (PTSD) symptoms on inhibitory control processing across military deployment.In line with our hypotheses, (1) timing of trauma explained additional variance over and above cumulative trauma but only for response inhibition, (2) high interpersonal trauma load was associated with enhanced conflict monitoring across deployment, and (3) developmental (child/adolescent) trauma impaired response suppression across deployment.Contrary to our hypotheses, timing of trauma (including developmental trauma) did not impact on conflict monitoring and higher interpersonal trauma load showed a faster response inhibition.These findings are explored below.
In line with the cumulative stress model 7,8 , this study found cumulative trauma load affected the N2 and P3 component suggesting an impact on conflict monitoring (detecting and controlling conflict between incoming stimuli) and response suppression (inhibiting an activated response).Furthermore, higher trauma load,  particularly interpersonal, was associated with childhood trauma onset and higher PTSD symptoms, especially re-experiencing symptoms.Although developmental trauma was not a significant predictor of conflict monitoring as hypothesized these results are consistent with previous research showing exposure to multiple types of trauma, particularly interpersonal, is associated with early-onset trauma, PTSD and impaired inhibitory control [9][10][11][13][14][15] .
More specifically, we found those with higher interpersonal trauma displayed reduced resources toward conflict monitoring (smaller N2 amplitude) at pre-deployment, which was associated with arousal symptoms.This suggests high interpersonal trauma led to less resources for monitoring competing task-relevant stimuli, potentially due to difficulty allocating resources away from arousal symptoms towards processing the task at hand 4 .In contrast, when considering changes across deployment, this group increased resources toward monitoring conflict (increased N2 amplitude) and reduced resources toward response suppression (smaller P3 amplitude and faster P3 latency).Increased resources for conflict monitoring has previously been found in veterans with PTSD and is consistently associated with anterior cingulate cortex (ACC) hyperactivation 19,[23][24][25] .The ACC is activated when anticipating traumatic stimuli in PTSD 42,43 , suggesting high interpersonal trauma may create difficulty regulating threat processing and reactivity or increases intrusive thoughts during deployment which impairs inhibitory control and creates a vulnerability to PTSD.Reduced resources toward response suppression following high interpersonal trauma supports literature showing smaller NoGo-P3 amplitude and hypoactivation in the ACC and orbito-medial prefrontal cortex to the NoGo-P3 is linked with early-onset trauma, PTSD and difficulties in cognitive control and decision making 6,29,[38][39][40][41] .Taken together, high interpersonal trauma may lead to an overwhelmed cognitive system that results in increased effort for monitoring and detecting conflict between activated stimuli in their environment (hypervigilance toward threat) during deployment, thus depleting resources for inhibiting responses.This is consistent with evidence showing PTSD in a military population is associated with difficulty disengaging from internal and external distractions and inhibiting automatic responses 5 .
Contrary to interpersonal trauma, higher non-interpersonal trauma load was associated with delayed speed of conflict monitoring at pre-deployment, and this was not associated with PTSD symptoms.This provides further support for interpersonal trauma being more predictive of PTSD than non-interpersonal trauma [9][10][11][12][13] .Furthermore, in our study, significant changes associated with non-interpersonal trauma in relation to allocation of resources toward conflict monitoring and response suppression were seen in those without non-interpersonal trauma rather than high non-interpersonal trauma.This suggests non-interpersonal trauma has little impact on inhibitory control in the early aftermath of deployment.
Consistent with our hypotheses, trauma onset in adolescence resulted in increased resources toward response suppression and delayed response suppression processing across deployment (increased No-Go-P3 amplitude and longer NoGo-P3 latency).This supports previous research in police officers with sub-clinical PTSD and veterans with PTSD respectively 26,27 .Similar to adolescent trauma, we found childhood trauma resulted in a delayed, but not enhanced, response suppression across deployment.Longer latency for response suppression following childhood and adolescent trauma suggests the need for more time during response suppression processing in order to inhibit the correct response.Although we found no association between delayed response suppression and PTSD in the early aftermath of deployment (4 months), delayed response inhibition has been associated with PTSD, particularly arousal and re-experiencing symptoms, in the years following deployment 27 .This suggests those with trauma onset in childhood and adolescence may be more vulnerable to developing PTSD in the long-term aftermath of deployment.www.nature.com/scientificreports/ The NoGo-P3 activates pre-frontal regions, including the orbitofrontal cortex, which develop during adolescence, alongside developing connectivity between frontal inhibitory and the amygdala and threat detection networks 19,23,24,30 .As the amygdala and threat detection networks develop in childhood, this may suggest childhood trauma impairs speed of response suppression due to increased arousal symptoms from an impaired emotion regulation system, and adolescent trauma impairs allocation of resources toward response suppression and speed of response suppression through impaired frontal inhibitory connectivity during adolescence [30][31][32][33][34][35] .Further research is needed to determine the brain networks and connectivity contributing to impaired response suppression following adolescence and childhood trauma.
Taken together, these findings indicate that like brain injury during development 36 , developmental trauma also has a long-term consequence on inhibitory control.Our findings support the sensitive period of trauma exposure model 16,17 , where developmental trauma appears to impact on inhibitory frontal networks or connections, which are important for response suppression 14,33,34 and is consistent with research showing developmental trauma leads to decreased accuracy for inhibiting responses and increased impulsivity and risk-taking behavior 30,44,45 .Furthermore, it suggests developmental trauma plays a differential and supplementary role to cumulative trauma load during inhibitory control processing.
By adulthood, brain development is largely complete, and we found trauma onset in adulthood did not appear to impact on inhibitory control processing in the early aftermath of deployment.However, adults without trauma exposure displayed decreased resource allocation for response suppression across deployment, but unlike those with high interpersonal trauma they did not get faster at response suppression.The percentage of high PTSD symptoms in those without trauma exposure was 13% at post-deployment and this group went from reporting the lowest to highest avoidance across deployment.Increased PTSD symptoms have been associated with increased attentional threat avoidance during acute stress on deployment 46,47 .Therefore, lack of trauma exposure and low deployment experience may increase avoidance to threat leading to reduced inhibitory suppression processes and a temporary increase in PTSD symptoms in the early aftermath of post-deployment.
Although our findings support and extend current research, there are several limitations.Larger sample size would provide greater power to further differentiate interpersonal trauma types and differentiate individuals with childhood and adolescent onset trauma from childhood onset trauma alone 9,14 .Further, our models explain up to 10% of variance in the ERP components suggesting scope for a wider range of predictors to further understand mechanisms contributing to inhibitory control.
Our research highlights the need for the development and implementation of tailored strategies for strengthening emotion regulation, inhibitory control, and prefrontal functioning in military personnel, particularly those with (a) developmental trauma, (b) high interpersonal trauma load and (c) no prior trauma exposure who display impaired inhibitory control across deployment and may be at risk for PTSD.High post-deployment PTSD symptoms and combat exposure were not significant predictors of conflict monitoring or response suppression in our study, suggesting they did not impact on inhibitory control processing in the immediate post-deployment period.As PTSD can change following deployment, or be delayed, our research highlights the importance of further follow-up at least one-year post-deployment to explore relationships between timing of trauma and trauma load on inhibitory control and PTSD trajectories, consistent with previous research [48][49][50][51] .
In conclusion, this study investigated the impact of timing and cumulative type of trauma as well as PTSD symptoms on inhibitory control processing across military deployment.Our findings extend previous research by showing the supplementary and differential role of interpersonal trauma load alongside timing of trauma on inhibitory control processing.Where developmental trauma appears to impact on response suppression, interpersonal trauma leads to an overwhelmed inhibitory system that impairs conflict monitoring and response suppression.Our findings also reveal a differential impact of childhood and adolescent trauma on response suppression, which highlights the need for research to examine specific critical periods rather than defining trauma before age 18 as childhood trauma.Taken together, this paper supports a combined cumulative trauma and sensitive period of trauma exposure model for inhibitory control processing and highlights the enduring impact of timing of trauma and trauma load on inhibitory control.

Figure 1 .
Figure 1.Estimated Marginal Trends from the linear mixed model with 95% Confidence Intervals in N2 Amplitude by Interpersonal and Non-interpersonal Cumulative Trauma Type across Deployment.N2 amplitude is a negative ERP component so smaller values indicate larger magnitude.

Figure 2 .
Figure 2.Estimated Marginal Trends (with 95% Confidence Intervals) from the linear mixed model for N2 Latency by Non-interpersonal Cumulative Trauma Type across Deployment.

Figure 3 .
Figure 3.Estimated Marginal Trends (with 95% confidence intervals) from the linear mixed model for P3 Amplitude by Interpersonal and Non-interpersonal Cumulative Trauma Type across deployment.

Figure 4 .
Figure 4.Estimated Marginal Means (with Standard Error) from the linear mixed model for P3 Amplitude by Trauma Onset across Deployment.

Figure 5 .
Figure 5.Estimated Marginal Trends (with 95% Confidence Intervals) from the linear mixed model for P3 Latency by Interpersonal and Non-interpersonal Cumulative Trauma Type across Deployment.

Figure 6 .
Figure 6.Estimated Marginal Means (with Standard Error) from the linear mixed model for P3 Latency by Trauma-Onset across Deployment.

Table 1 .
30, Participant demographics, trauma exposure and PTSD symptoms by Trauma onset.PCL Posttraumatic stress disorder symptoms, mTBI Mild Traumatic Brain Injury.All values refer to mean (standard deviation), unless indicated otherwise as percentage.P-values are from ANOVA for continuous variables and Fisher's test for categorical variables.

Table 2 .
Results from the F-test assessing model fit comparison for N2 and P3 Component.