Autonomic nervous system function before and after trauma-focused psychotherapy in youth with (partial) posttraumatic stress disorder

While trauma-focused psychotherapies have been shown effective in youth with PTSD, the relationship between treatment response and alterations in autonomic nervous system (ANS) associated with


Introduction
Posttraumatic stress disorder (PTSD) is a common mental health disorder that develops in approximately 16% of youth exposed to traumatic events (Alisic et al., 2014).Girls have an almost two-fold increased risk of developing PTSD after trauma exposure relative to boys (Alisic et al., 2014).Youth with PTSD are troubled by frequent re-experiencing of the traumatic event, persistent avoidance, hyperarousal and negative alterations in cognition and mood (American Psychiatric Association, 2013).These symptoms can interfere with social functioning and school performance, have a negative effect on the quality of life of the affected youth and their families (Carrion et al., 2002) and are a crucial factor in shaping the vulnerability to depression, suicidality and cardiovascular disease later in life (Molnar et al., 2001).Furthermore, youth exposed to traumatic events with clinically important symptoms of PTSD but with subthreshold criteria for PTSD (partial PTSD) also demonstrate substantial functional impairment and distress which do not differ significantly from youth meeting full PTSD criteria (Carrion et al., 2002).Current practice parameters therefore recommend that in youth, patients with partial PTSD should also be provided with evidence-supported treatment options (National Institute for Health and Care Excellence, 2018).
Randomized controlled trials have demonstrated efficacy of traumafocused cognitive behavioral therapy (TF-CBT) and eye movement desensitization and reprocessing (EMDR) in youth with (partial) PTSD (Morina et al., 2016).However, less is known about the association between these trauma-focused psychotherapies and changes in autonomic nervous system (ANS) function in youth with (partial) PTSD (Zantvoord et al., 2013).Examining ANS changes associated with treatment response can provide a better understanding of why some youth recover while in some PTSD persists (Zantvoord et al., 2013).This is important because, response varies considerably among individuals, with 30-50% of youth not benefiting sufficiently (Diehle et al., 2015;Morina et al., 2016), leading to persistent symptoms and longer treatment trajectories.
A mounting body of literature emphasizes disturbances in both the sympathetic nervous system (SNS) and parasympathetic nervous system (PNS) in adults with PTSD (Pole, 2007).Increased heart rate (HR) during trauma-related stimuli has been one of the most consistent findings in studies of ANS function in adults with PTSD (Pole, 2007).HR is under joint control of the SNS and PNS and increased HR during stress could be considered as the result of SNS and PNS disbalance in individuals with PTSD (Hopper et al., 2006).The parasympathetic control over HR is exerted through the action of the vagal nerve, in which a higher vagal tone produces a lowered HR compared with the basal firing rate (Porges, 2007).Vagal tone can be measured noninvasively via respiratory sinus arrhythmia (RSA).RSA reflects the rhythmic fluctuation of HR and is quantified as the variability of HR increases with inspiration and decreases with expiration (Grossman and Taylor, 2007).Studies using RSA to examine vagal control in PTSD have rendered more mixed findings, but together indicate an association between PTSD and vagal withdrawal (i.e.reduced RSA) during both rest and stress in adults with PTSD (Campbell et al., 2019), suggesting an overall lower parasympathetic activity in adults with PTSD (Schneider and Schwerdtfeger, 2020).Studies specifically measuring SNS parameters in PTSD have mainly employed skin conductance levels and the pre-ejection period (PEP, the period between the electrical initiation of the heartbeat and the time that blood is ejected into the aorta) as sympathetic indices and suggest heightened SNS activity in adults with PTSD relative to controls (Michopoulos et al., 2015).Furthermore, studies have indicated gender differences in posttraumatic autonomic functioning with women demonstrating decreased parasympathetic control and higher HR relative to men with PTSD (Keary, Hughes and Palmieri, 2009;Seligowski et al., 2021).
Importantly, the ANS undergoes substantial development throughout childhood and adolescence.PNS activity follows a cubic trend, with an exponential increase from infancy, a plateau phase during middle childhood, followed by a decrease in adolescence (Harteveld et al., 2021).SNS activity shows a more linear trend, with a gradual decrease from infancy to adolescence (Harteveld et al., 2021).A recent meta-analysis of youth with PTSD replicated the finding of overall heightened ANS activity during stress tasks found in adults (Siciliano et al., 2022).However, contrary to results in adults, decreased SNS activity at rest corresponded to increased posttraumatic stress symptoms and the correlation between PNS measures and PTSD was non-significant (Siciliano et al., 2022).Divergent ANS findings between youth and adults with PTSD might indicate developmental effects that are not yet apparent until adulthood as well as plastic response to illness over time (Mikolajewski and Scheeringa, 2018;Weems et al., 2019).Together, this highlights that ANS results obtained in the adult PTSD population cannot be readily translated to youth and emphasizes the need for specific studies examining ANS function that take potential age-related effects into account.
There remains a debate as to whether successful trauma-focused psychotherapy may reverse ANS abnormalities associated with PTSD.Results from longitudinal ANS studies in adults with PTSD have been mixed (Zantvoord et al., 2013).The majority of these studies have focused on HR, with most suggesting that HR (reactivity) decreases during trauma-focused psychotherapy (Blanchard et al., 2002;Lindauer et al., 2006), while some did not find a change in HR during trauma-focused psychotherapy (Wangelin and Tuerk, 2015).To date, fewer studies have examined the association between trauma-focused psychotherapy and specific measures of PNS or SNS activity.Most, but not all of these studies, did not find an association between trauma-focused psychotherapy outcome and PNS or SNS activity (Blanchard et al., 2002;D'Andrea and Pole, 2012;Sack et al., 2008).However, studies had limited sample sizes or used crude measures of PNS/SNS activity, precluding a strong synthesis of the current evidence (Bourassa et al., 2021).To the best of our knowledge, only two studies have been published that examine ANS changes associated with trauma-focused psychotherapy response in youth with PTSD.Lipschutz et al. (2017) did not find an association between RSA change and trauma-focused cognitive behavioral therapy (TF-CBT) outcome (Lipschutz et al., 2017), while Shenk et al. (2022) did demonstrate a significant association between increased RSA withdrawal during the course of treatment and symptom improvement (Shenk et al., 2022).The limited amount of studies with mixed findings so far indicate that the question of whether successful trauma-focused psychotherapy in youth may reverse ANS abnormalities associated with PTSD remains to be answered.
In the present study, we first compared ANS activity and stress reactivity in youth (aged 8-18 years) with PTSD or partial PTSD relative to trauma-exposed healthy youth (TEC) and performed exploratory analyses on the moderating effects of age and gender.We then investigated longitudinal changes in ANS activity and stress reactivity in youth with (partial) PTSD treated with trauma-focused psychotherapy (either TF-CBT or EMDR).We used HR, RSA and PEP during a script-driven imagery (SDI) paradigm to compare youth with (partial) PTSD and TEC as well as compare baseline to posttreatment changes in treatment responders relative to non-responders.We hypothesized that youth with (partial) PTSD relative to TEC would show an overall increased ANS stress reactivity as indicated by greater increase of SNS activity (PEP reduction) together with a greater vagal withdrawal (RSA reduction) resulting in a greater HR increase during SDI at baseline.Given ANS developmental change and previous results on gender differences in posttraumatic autonomic functioning, we further hypothesized moderating effects of age and gender on the relationship between ANS stress reactivity and PTSD diagnosis.Last, we hypothesized that traumafocused psychotherapy response would be associated with an increase in vagal control during stress and a reduction of HR and SNS stress reactivity.

Study design
A graphical overview of the study cohort is depicted in Fig. 1.First, we compared indicators of ANS activity (HR, RSA and PEP) in response to SDI at baseline between 76 (partial) PTSD patients and 27 TEC and investigated potential moderating effects of age and gender.Second, we compared the same parameters between treatment responders (n = 32) and non-responders (n = 16) from baseline to post-treatment.

Participants
All participants were Dutch-speaking, and 8-18 years old.Gender categories were based on the personal identification of participants' gender.Patients were recruited between June 2011 and September 2018 at the outpatient child psycho-trauma center of the Department of Child and Adolescent psychiatry, Levvel in Amsterdam, The Netherlands as part of an RCT comparing TF-CBT and eye movement desensitization and reprocessing (EMDR) (Diehle et al., 2015).Patients were referred by child welfare services, physicians or general practitioners.Diagnoses for PTSD or partial PTSD were established clinically by an experienced child and adolescent psychiatrist or psychologists according to the DSM-IV-TR criteria using joint child reports on individual symptoms on the Clinician-Administered PTSD Scale for Children and Adolescents (CAPS-CA) semi-structured interview (Diehle et al., 2013;Nader et al., 1996) and the caregiver reports from the PTSD scale of the Anxiety Disorders Interview Schedule -Parent Version (ADIS-P) (Verlinden et al., 2014).A symptom was established as present if either child or caregiver reported its presence.Partial PTSD was defined as either fulfilling two of the three PTSD symptom clusters or having one symptom present in each of the three symptom clusters (Stein et al., 1997).Furthermore, participants were required to have a CAPS-CA total score indicating at least mild PTSD symptom severity (>20 points).Exclusion criteria were: acute suicidality, IQ< 70, pregnancy, neurological disorders or serious medical illnesses or meeting the criteria of the following diagnosis: psychotic disorders, substance use disorder or pervasive developmental disorder.Psychotropic medication was required to be stable for at least three weeks before and during trauma-focused psychotherapy.In our sample, one participant was taking sertraline at a stable dose.TEC were recruited through local elementary -and high schools.Exposure to traumatic events was validated according to A1 and A2 criteria of DSM-IV-TR using the life-events checklist of the CAPS-CA.TEC were excluded if they met current or lifetime PTSD or partial PTSD diagnosis using both the CAPS-CA and caregiver reports from the PTSD scale of the ADIS-P or had a current CAPS-CA total score of > 20 points.In accordance with procedures approved by the Institutional Review Board of the Amsterdam University Medical Centers and the Declaration of Helsinki, written informed consent was obtained from all parents or legal guardians.Written informed consent from youth aged 12 years and older and assent from youth aged 11 and younger, was also obtained from the youth themselves.All participants received a monetary incentive for participation (€5 for each assessment).

Trauma-focused psychotherapy
Patients were randomly assigned to weekly protocolized sessions for a total of 8 weeks of either TF-CBT or EMDR.The data reported here were obtained as part of a larger study on the long-term efficacy of TF-CBT and EMDR.Treatment was delivered by experienced trauma therapists who were trained in TF-CBT and EMDR before study initiation.Supervision by TF-CBT and EMDR experts was provided throughout the study.Treatment protocols, training and supervision of therapists, as well as treatment fidelity have been described in detail previously (Zantvoord et al., 2019).
Trained psychologists administered the CAPS-CA and the PTSD scale of the ADIS-P to measure PTSD symptoms at baseline and after treatment.The Dutch Revised Child Anxiety and Depression Scale (RCADS (-P)) questionnaire was administered to assess depressive and anxiety symptoms (Chorpita et al., 2000).Symptom change was calculated by subtracting the baseline from the post-treatment CAPS-CA total score.There is no established definition of response criterion or a consensus definition of response terms in the child trauma field.Based on the psychometric properties of the CAPS (-CA) and previous treatment outcome studies using the CAPS-CA we used ≥ 30% reduction of CAPS-CA total score as our primary response criterion for clinically meaningful improvement.A ≥ 50% reduction of CAPS-CA total score was selected as a secondary response criterion entailing substantial clinical improvement (Zantvoord et al., 2021;Zhutovsky et al., 2021).

Script-driven imagery (SDI)
All study participants performed a standardized protocol for scriptdriven imagery (SDI) (Shalev et al., 1992).The SDI protocol and procedure have been described in detail previously (Zantvoord et al., 2019).In summary, patients performed SDI twice, first at baseline within two weeks before the start of trauma-focused psychotherapy and second two weeks after the 8th trauma-focused psychotherapy session.TEC were only assessed once at baseline.A 30 s (range 29-31 s) personalized trauma audio script describing the participant's most disturbing traumatic event and a general 30 sec neutral audio script describing tooth brushing were recorded.During the SDI procedure, the neutral script always preceded the trauma script, both scripts were preceded by a 3 min movie clip depicting neutral landscapes.Participants were instructed to vividly imagine the events described in the scripts with their eyes closed during and for 30 sonds after the scripts.

Psychophysiological measures
Physiological data were measured by the "Vrije Universiteit Ambulatory Monitoring System" (VU-AMS), a lightweight portable device that records ECGs and changes in thorax impedance (impedance Youth with (par�al) PTSD n=76 Lost to follow up n=16 1.
did not finish treatment n=13 2.
no post-treatment CAPS-CA n=3 Randomized n= 64

Responders n=32
Trauma exposed heathy controls n=27 Did not receive treatment or received treatment elsewhere n= 12 cardiography [ICG]) from a six-electrode configuration (VU University, Amsterdam, The Netherlands, www.vu-ams.nl).Participants wore the VU-AMS device during the SDI procedure.
From the VU-AMS recordings, the following outcome variables were computed: HR, RSA, and PEP during both the 30-second neutral and trauma imagery period.HR was directly derived from the inter-beatinterval time series (De Geus and Van Doornen, 1996).RSA was obtained by peak-valley estimation that combined the ECG with the respiration signal assessed with the thorax impedance (De Geus et al., 1995).PEP was extracted from the interval between the Q-onset in the ECG, indicating the onset of left ventricular electrical activity, and the upstroke (B-point) of the ICG signal, indicating the beginning of left ventricular ejection (van Lien et al., 2015).
During automated and visual data cleaning, suspicious IBIs and breathing cycles were corrected or discarded when displaying irregularities.The number of discarded IBIs or breathing cycles did not differ between groups (i.e.(partial) PTSD vs. TEC and responders vs. nonresponders).An automated scoring algorithm was also used to detect crucial landmarks in the ICG, which were then visually inspected and manually corrected.
RSA data was positively skewed and therefore log-transformed to approximate a normal distribution.

Covariates
Adjustments were made for a priori selected putative demographic characteristics including age, gender, educational level as well as trauma-related and health indicators associated with PTSD and ANS activity, including smoking status (yes/no), alcohol use (>1 unit per day, yes/no), index trauma type and time since trauma exposure (Harteveld et al., 2021;Iffland et al., 2020).Because it has been linked to HR variability, respiration rate was also included as a covariate (Harteveld et al., 2021).In the baseline to posttreatment analysis (responders vs. non-responders), baseline total CAPS-CA score was added as a covariate.

Statistical analyses
The distribution of baseline clinical, trauma and demographic characteristics across TEC and youth with (partial) PTSD as well as treatment responders and non-responders were examined using X 2 -tests, independent sample t-tests or Mann-Whitney tests as appropriate.For descriptive purposes, independent sample t-tests or Mann-Whitney tests were also used to outline the differences at baseline between ANS values during neutral and trauma SDI, and ANS stress reactivity values across youth with (partial) PTSD and TEC as well as treatment responders and non-responders respectively.Absolute ANS stress reactivity values were calculated by subtraction of averaged ANS values during neutral SDI from averaged ANS values during trauma SDI.
To formally examine the differences in HR, RSA, and PEP between youth with (partial) PTSD and TEC at baseline, we conducted linear mixed model analyses.We used linear mixed models because this method accounts for the dependency of repeated measurements obtained from the same participant over time.In addition, mixed models also have the advantage of allowing missing measurements without discarding data (Gueorguieva and Krystal, 2004).Two models were used: one to estimate the main effects of group (youth with (partial) PTSD vs TEC) and condition (trauma versus neutral SDI), and the other model to additionally estimate the interaction effect of group-by-condition (ANS stress reactivity differences across the groups at baseline).The mixed models were then repeated with adjustments for covariates.In case of a significant interaction, post-hoc analyses examining the main effects of condition were performed in both groups separately.
To examine a moderating effect of gender and age at baseline we performed linear mixed models with 3-way-interaction between group (youth with (partial) PTSD vs TEC)-by-condition (trauma vs. neutral SDI)-by-age as well as group-by-condition-by-gender interaction.In case of a significant interaction, stratified post-hoc tests were conducted.In case the higher-order interaction was non-significant, we removed this term and used the remaining more parsimonious model.Furthermore, LMM with condition-by-age, as well as condition-by-gender interactions, were performed in the (partial) PTSD and TEC groups separately.Mixed models were performed with and without covariates.
To assess the (baseline to posttreatment) effects of trauma-focused psychotherapy response in patients on HR, RSA and PEP, we first tested whether responders and non-responders were comparable at baseline using both the primary ≥ 30% and secondary ≥ 50% response criterion.We then performed LMM with a 3-way-interaction between time (baseline versus posttreatment), condition (trauma versus neutral SDI) and response status (responder versus non-responder) separately for both response criteria.In case of a significant 3-way interaction, posthoc test stratified for response status were performed.When a high-er=order interaction did not contribute significantly to the model, we removed this term and used the more parsimonious model.LMM were additionally performed in the (partial) PTSD and TEC groups separately.We repeated the mixed models with incorporation of covariates in the model.
Statistical analyses were performed using SPSS version 26.0 (SPSS Inc., Chicago IL, USA).The criterion for statistical significance was p < .05.

Participant characteristics
Table 1 shows that relative to TEC, youth with (partial) PTSD had lower educational levels and were more likely to be of non-western ethnicity.In addition, youth with (partial) PTSD were more often exposed to interpersonal and multi-event trauma as well as exposed to trauma at an earlier age.The most common index trauma in patients was sexual abuse, followed by community violence and witnessing domestic violence.In TEC the most common index trauma were accidents followed by the category 'other trauma' from the life-events checklist of the CAPS-CA.52.6% of patients and 29.6% of TEC were exposed to multipleevent trauma (Table 1.).Within the patient group 77.6% of patients met the full DSM-IV diagnostic criteria for PTSD at baseline and the remaining 22.4% met the criteria for partial PTSD.The average baseline CAPS-CA score in patients was M = 52.9 points, SD = 23.1, which is indicative of moderately severe PTSD.The number of discarded IBIs or breathing cycles did not differ between (partial) PTSD and TEC.
Applying the ≥ 30% response criterion yielded 32 responders and 16 non-responders, applying the ≥ 50% response criterion yielded 23 responders and 25 non-responders.Table S1a indicates that there were no significant group differences at baseline for the distribution of sociodemographic, trauma or clinical characteristics between responders and non-responders when using the primary ≥ 30% response criterion.Table S1b shows when applying the ≥ 50% response criterion, responders had a lower CAPS-CA re-experiencing score and female-tomale ratio at baseline (p < .05).Number of discarded IBIs or breathing cycles did not differ between responders and non-responders.

Autonomic nervous system activity during script-driven imagery in youth with (partial) PTSD vs. TEC at baseline
Table 2 describes the unadjusted means of ANS data during neutral and trauma SDI and the stress reactivity values for TEC and youth with (partial) PTSD at baseline.Results of the unadjusted and adjusted main and interaction effects of group and condition at baseline are shown in Table s2.The fully adjusted mixed-model analyses resulted in a significant main condition effect for HR (F = 4.01, df = 1, p < .05)but not RSA (F =.02, df = 1, p = .89)and PEP (F =.13, df = 1, p = .72)for neutral SDI compared with trauma SDI, indicating that trauma SDI at baseline yielded the expected stress activation as measured with HR but not with PEP and RSA.Significant main group effects at baseline (PTSD vs. TEC) were seen for HR (F = 3.84, df = 1, p = .05),which was higher in the (partial) PTSD group as compared to TEC (table s2).No main group effect was found for PEP (F = 1.30, df = 1, p = .26)and RSA (F 2.54, df = 1, p = .12).Importantly, a significant result was found for the group-bycondition interaction for RSA (F = 6.38, df = 1, p = .01)(i.e.differences in RSA stress reactivity across youth with (partial) PTSD and TEC), related to RSA decrease during trauma SDI in youth with (partial) PTSD and RSA increase during trauma SDI in TEC (Table 2), although the posthoc analysis stratified by group did not show a significant main effect of condition for RSA in youth with (partial) PTSD (F = 1.31, df = 1, p = .26)and showed a trend in TEC (F = 3.05, df = 1, p = .09).There were no significant interactions for HR (F =.04, df = 1, p = .85)and PEP (F =.85, df = 1, p = .80).When excluding the patient using sertraline from the analyses, no changes occurred in significance or magnitude of findings.

Moderation by age and gender of ANS activity during SDI in youth with (partial) PTSD vs. TEC at baseline
All moderation analyses are fully described in the Supplementary material and are briefly summarized here.There was no significant effect of the age-by-group-by-condition or age-by-group interaction at baseline, indicating that ANS differences between (partial) PTSD and TEC were independent of age.The fully adjusted gender-by-group-bycondition interactions indicated a moderating effect of gender on differences between youth with (partial) PTSD and TEC during SDI at baseline in PEP (F = 5.13, df = 1, p = .03)and RSA (F = 6.63, df = 1, p = .01)but not for HR (F =.59, df = 1, p = .45).Post-hoc analysis stratified by gender showed a significant group-by-condition interactions for RSA in boys (F = 13.17,df = 1, p = .001)but not girls (F =.14, df = 1, p = .71)and for PEP a trend in girls (F = 3.11, df = 1, p = .08)but not boys (F = 1.96, df = 1, p = .17)(Fig. 2).

Baseline to posttreatment change in autonomic nervous system activity during script-driven imagery in responders vs. treatment nonresponders
At baseline, the responder status-by-condition interaction and main effects of responder status indicated that HR, PEP and RSA (re)activity Fig. 2. Autonomic nervous system activity during script driven imagery in youth with (partial) PTSD vs. trauma exposed healthy controls (TEC) at baseline stratified by gender.All results are adjusted for age, ethnicity, smoking status (yes/no), alcohol consumption (yes/no), index trauma and time since trauma as well as respiration rate for RSA.Error bars indicate SD.The gender-by-group-by-condition interaction indicated a moderating effect of gender on differences in PEP (F = 5.132, df = 1, p = .026)and RSA (F = 6.628, df = 1, p = .012)but not HR (F =.590, df = 1, p = .445).Post-hoc analysis stratified by gender showed a significant group-by-condition interactions for RSA in boys (F = 13.17,df = 1, p = .001)but not girls (F =.14, df = 1, p = .71)and for PEP a trend in girls (F = 3.11, df = 1, p = .08)but not in boys (F = 1.96, df = 1, p = .17).
during SDI before treatment were comparable between trauma-focused psychotherapy responders and non-responders (see table s3a and s3b in supplementary materials for ≥30% and ≥50% response criterion respectively).
The same 3-way interaction and subsequent analyses with the secondary response criterion (≥50% CAPS-CA reduction) are fully described in the Supplementary material and summarized here.We found a significant effect of the time-by-condition-by-responder status for RSA (F = 4.26, df = 1, p = .05)indicating differential baseline to posttreatment changes in RSA stress reactivity between trauma-focused psychotherapy responders and non-responders (Fig. s1 and table s4b In supplementary materials).Post-hoc analysis stratified by response status showed a significant increase of RSA in response to trauma SDI from baseline to posttreatment in treatment responders (F = 4.12, df = 1, p < .05),while non-responders did not show a significant increase of RSA in response to trauma SDI with treatment (F =.15, df = 1, p = .70).We did not identify any significant 3-way or subsequent 2-way interactions with HR or PEP using our secondary criterion.

Discussion
In this study, we investigated ANS stress reactivity in traumaexposed youth and examined trauma-focused psychotherapy-related changes in ANS stress reactivity.Our results indicate that before treatment youth with (partial) PTSD are characterized by an RSA decrease during trauma SDI while trauma-exposed healthy controls show an RSA increase.This suggests the withdrawal of vagal activity in response to stress in (partial) PTSD patients and a contrasting increase of vagal activity in response to stress in TEC.Relative to TEC, youth with (partial) PTSD also showed an overall higher HR before treatment, which may reflect an overall SNS and PNS disbalance in youth with (partial) PTSD.In contrast, we did not find evidence for differences in HR and PEP stress reactivity in response to trauma SDI.Our results provide preliminary evidence for a moderating role of gender on ANS differences between youth with (partial) PTSD and TEC, in which RSA differences were mainly found in boys while PEP differences occurred in girls.We did not find evidence for a moderating effect of age.We further found that relative to non-responders, trauma-focused psychotherapy responders were characterized by an overall increased HR stress reactivity, which was independent of time (baseline and posttreatment).We did not observe change in HR, PEP and RSA after successful trauma-focused psychotherapy with our primary (>30% CAPS-CA reduction) response criterion.We did, however, find change in RSA reactivity after successful trauma-focused psychotherapy employing our secondary (≥50% CAPS-CA reduction) response criterion, the latter may suggest an association between treatment response and an increase in vagal control during stress.
Our finding of RSA decrease during trauma SDI in youth with (partial) PTSD suggests lower vagal control during trauma-related stressors through a withdrawal of vagal input.According to Porges, RSA serves as an index of stress reactivity and stress vulnerability (Porges, 2007).High vagal activity enhances cardiac control of environmental demands while vagal withdrawal is associated with a less adaptive capacity and associated with increased vulnerability to dysregulated and excessive responses to (traumatic) stressors (Porges, 2007;Sack et al., 2004).Dysfunctional vagal stress control could either be a pre-existing risk factor in the development of PTSD or a plastic response to PTSD over time.Although we cannot differentiate between these possibilities in the current study, previous research by Mikolajewski et al. (2018) has provided evidence for the former, with RSA and RSA reactivity as pre-existing markers of stress sensitivity that predicts posttraumatic stress symptomatology after trauma exposure (Mikolajewski and Scheeringa, 2018).
Increased vagal withdrawal during stress has been a finding in some but not all studies in adult PTSD (Campbell et al., 2019).Interestingly, the few prior studies which measured RSA reactivity to investigate the relationship between vagal control during stress and PTSD in youth did not find significant differences in RSA stress reactivity between PTSD and TEC (Katz and Gurtovenko, 2015;Kirsch et al., 2015;Scheeringa et al., 2004), which contrasts findings in the current study.Aside from methodological differences in RSA measurements, sample size and analysis, PTSD sample characteristics could be an explanation for divergent results between these studies.For example, the sample of Scheeringa et al. (2004) and Katz and Gurtovenko (2015) consisted of children aged 2-6 years and 6-12 years respectively, which do not, or only partially overlap with the age range in the current study (8-18 years).Although we did not find a moderating effect of age, this does not rule out that age might be a factor in explaining discrepancies between studies.Particularly, as previous research has reported a cubic developmental trajectory of RSA (Harteveld et al., 2021).Another possibility could be that the divergent findings might be related to differences in contextual factors.For example, parenting style and caregiver's PTSD symptomatology have been reported to influence the relationship between PTSD and RSA stress reactivity in the studies of Scheeringa and Katz (Katz and Gurtovenko, 2015;Scheeringa et al., 2004).This might be relevant as a considerable part of our sample was exposed to trauma in the domestic environment and was subject to out-of-house placement and foster care.Future research in older youth could therefore benefit from the inclusion of measures of parental PTSD and parenting style to examine potential moderating effects on the relationship between RSA and PTSD.
In contrast to previous studies in adult PTSD (Pole, 2007), we did not find evidence for differences in HR stress reactivity at baseline in youth with (partial) PTSD relative to TEC.Importantly, only half of the included participants showed the expected HR stress response during SDI.However, even when we limited our analysis to those individuals who did show the expected stress response, we did not identify differences between youth with (partial) PTSD and TEC in HR stress reactivity.Interestingly, prior studies comparing youth with PTSD and TEC also did not find significant differences in HR stress reactivity during exposure to trauma-related stimuli (Jones-Alexander et al., 2005;Kirsch et al., 2015;Scheeringa et al., 2004).Together, these studies suggest divergent findings regarding HR stress reactivity between youth and adults with PTSD.A possible explanation for this is the difference in the type of trauma exposure of included patients and differences in control groups.The majority of studies in adults included patients who were either exposed to motor vehicle accidents or were (male) veterans, while in studies in youth, domestic violence and sexual abuse were more prevalent.Moreover, studies in adults more often compared PTSD patients with non-traumatized controls, which prior research has shown to increase the chance of finding group differences (Meewisse et al., 2007).Also, average baseline CAPS-CA scores in our patient sample indicated moderately severe PTSD.As previous research in youth has demonstrated a correlation between PTSD severity and ANS activity (Siciliano et al., 2022), including a patient sample with more severe PTSD symptoms could have increased chances of finding ANS differences between TEC and PTSD.
Similarly, we did not find evidence for differences in PEP stress reactivity at baseline in youth with (partial) PTSD relative to TEC.This corroborates results from a recent meta-analysis on SNS stress reactivity in youth with PTSD, which revealed that the relation between SNS activity during stress tasks and PTSD was non-significant (Siciliano et al., 2022).The only prior publication on PEP of youth with PTSD also found that PEP stress reactivity in isolation did not significantly differ between youth with and without PTSD (Cohen et al., 2020).Together, these findings provide preliminary evidence that isolated SNS stress reactivity does not seem to be able to discriminate between youth with PTSD and TEC.A possible explanation for the absence of the expected HR and PEP stress response in half of the included participants could be that the effectiveness of a stressor paradigm to elicit a stress response is subjective to developmental change (Gunnar et al., 2009).Throughout childhood and adolescence, marked changes are found in the ability of the same stressor paradigm to provoke a stress response (Gunnar et al., 2009).These changes in stressor paradigm effectiveness may be related to ANS development and psychological changes during childhood and adolescence (Gunnar et al., 2009;Harteveld et al., 2021).Interestingly, however, we did not find a moderating effect of age on stress reactivity in our sample.Longitudinal studies measuring stress reactivity over the course of development from childhood and adolescence into adulthood will eventually be necessary to examine the role of developmental change as an explanation for the different stress reactivity findings in adults and youth with PTSD.
Our study did show generally higher HR during both neutral and trauma SDI at baseline in youth with (partial) PTSD relative to TEC.This finding suggests a heightened heart rate throughout the SDI procedure independent of script content.One possibility that could account for the increased HR during neutral SDI, is a generally increased HR in youth with (partial) PTSD.However, a previous meta-analysis did not find a generally increased HR in youth, which contrasts with findings in the current study (Siciliano et al., 2022).Another possibility is that the increased HR during neutral script in youth with PTSD might reflect increased anticipatory anxiety relative to TEC or the lack of HR increase during trauma script might be related to avoidance of trauma script imagery.Importantly, the neutral script always preceded the trauma script and participants knew that they would be exposed to a trauma script.Unfortunately, we did not collect (ambulatory) ANS data outside of the SDI procedure.Further studies, with a collection of ambulatory ANS data together with stress reactivity data, would be needed to conclude if higher HR in PTSD reflects general heightened ANS activity or is limited to the period of (anticipation of) the SDI paradigm.
Our exploratory moderation analysis provides preliminary evidence for a moderating role of gender on HR, RSA and PEP differences at baseline in youth with (partial) PTSD relative to TEC.Prior studies have shown a markedly increased risk of PTSD in girls relative to boys, identified clear gender differences in ANS development and have found notable gender differences in autonomic functioning among traumaexposed individuals (Alisic et al., 2014;Harteveld et al., 2021;Seligowski et al., 2021).Together, these findings could propose a role of dimorphic ANS development in explaining the increased PTSD prevalence in girls.Surprisingly, however, a recent meta-regression analysis, did not find a moderating role of gender on the association between posttraumatic stress symptoms and different ANS measures (Siciliano et al., 2022) in youth.To the best of our knowledge, prior cross-sectional studies comparing ANS measures between youth with PTSD and TEC have not reported on potential moderating effects of gender (Cohen et al., 2020;Jones-Alexander et al., 2005;Katz and Gurtovenko, 2015;Kirsch et al., 2015;Scheeringa et al., 2004).The exploratory nature of our moderation analysis and paucity of comparable studies precludes a strong synthesis of evidence at this time and emphasizes the need for additional ANS studies that examine gender differences throughout development.
In the baseline to posttreatment comparison between traumafocused psychotherapy responders and non-responders, we did not observe a change in HR or PEP after successful treatment.The former contrasts with the majority of studies in adults with PTSD (Wangelin and Tuerk, 2015;Zantvoord et al., 2013).To the best of our knowledge, this is the first reported study to longitudinally examine PEP in relation to trauma-focused psychotherapy response.Our results suggest that PEP, as a measure of SNS activity, may not directly change with successful treatment.However, to allow a reasonable interpretation of these findings, it should be taken into account that we only found overall lower PEP and higher HR in youth with PTSD relative to TEC at baseline but failed to observe differences in HR and PEP stress reactivity.Also, although the average total CAPS-CA reduction in the current sample could be considered relatively low (for in-depth discussion of response rates, see (Zantvoord et al., 2019) and the number of patients in the non-responder group was small (n = 16), which reduces power and raises the possibility of a type II error.
Interestingly, in analysis with the secondary response criterion (≥50% CAPS-CA reduction), we observed a significant baseline to posttreatment increase in RSA in response to trauma SDI in responders, while in non-responders RSA stress reactivity did not change from preto posttreatment (Fig. S1).Based on this finding one could speculate that trauma-focused psychotherapy response might be associated with an increase of vagal control in response to stress.However, interpretation of this finding should be done with caution, as we only found a trend in the condition-by-response-by-treatment interaction with our primary ≥ 30% response criterion.This could be due to more pronounced group differences when employing a stricter response criterion or could be related to larger power with more balanced sample sizes in the ≥ 50% response criterion.Future longitudinal replication studies with larger samples are therefore warranted to determine the relationship between trauma-focused psychotherapy response and ANS activity over time.
Several limitations of this study should be acknowledged.First, youth with (partial) PTSD and TEC differed on several key demographic and trauma-related variables, in particular type of trauma exposure.Relative to TEC, youth with (partial) PTSD were more often exposed to interpersonal and repeated traumatic events.Although inclusion of trauma-exposure type did not change the overall direction and significance of results, it cannot be ruled out that group differences are also related to differences in trauma-exposure rather than PTSD diagnosis alone.Also, the sample size of the (partial) PTSD group (n = 76) exceeded that of the TEC (n = 27), creating the risk for both false positive and false negative findings.To mitigate these limitations future studies should therefore aim at more balanced PTSD and TEC groups, both regarding sample size and trauma exposure.Also, although the majority (77.6%) of included patients had a full PTSD diagnosis, the remaining 22.4% had a partial PTSD diagnosis.Including youth with partial PTSD increased clinical heterogeneity and might have lowered overall group differences between patients and TEC, however, by including youth with partial PTSD, our sample better reflects the reallife clinical setting, which adds to the ecological validity of our findings.Second, ANS (re)activity is related to many state and trait factors, we have tried to account for most of these confounding factors in the current study.However, we omitted inquiry on menstrual status, (oral) contraceptive use and BMI, which are factors that influence the ANS and may influence cognitive and emotional processing involved in PTSD and trauma-focused psychotherapy response (Spear, 2009;Sundström Poromaa and Gingnell, 2014).In theory, this could thus have created a spurious relationship between ANS function and PTSD status as well as treatment outcome.Future studies should therefore aim to inquire about these variables as well.Third, youth with (partial) PTSD were randomized to receive either TF-CBT or EMDR, for the current analysis both treatment conditions were collapsed.Due to limited power, it was not feasible to examine differences between treatment responders and non-responders separately for both treatment conditions.Importantly, the efficacy of both treatment conditions has been shown comparable in an RCT with considerable sample overlap with the current study (Diehle et al., 2015).Furthermore, because we did not have a long-term follow-up assessment, we could not assess if there are ANS changes that are expressed later in development.This emphasizes the need for future treatment studies with long-term naturalistic follow-up of ANS function after trauma-focused psychotherapy.Finally, our study had a considerable drop-out rate, as 25% of randomized patients were lost to follow-up.Drop-out further reduced the power of our longitudinal responder vs. non-responder comparison, limiting the probability of observing significant results.Although dropout in the current study reflects routine clinical practice and ANS measures did not differ between treatment completers and non-completers, there is a possibility that drop-out could have influenced our main findings through attrition bias.

Conclusions
Our findings provide new insights into the ANS function underlying PTSD in youth and trauma-focused psychotherapy response.We demonstrated overall higher HR as well as withdrawal of vagal control in response to stress in youth with (partial) PTSD, with some preliminary evidence that the latter might be restored with successful traumafocused psychotherapy.Future research is merited to test the robustness and generalizability of these findings in larger independent cohorts.If replicated, the ANS differences identified here could serve as new targets for (pharmacological augmentation) interventions to improve treatment outcome in youth with (partial) PTSD (Brunet et al., 2018).

Fig. 1 .
Fig. 1.Flow diagram of included patients.Response is defined as ≥30% reduction in CAPS-CA total score from pre-to post-treatment.

Table 1
Subject characteristics.
ANS = autonomic nervous system; HR = heart rate; RSA = respiratory sinus arrhythmia; PEP = pre-ejection period.Trauma reactivity values represent averaged ANS values during trauma imagery minus their respective averaged values during neutral imagery.a Comparison using independent sample t-tests or Mann-Whitney U tests as appropriate b Values were ln-transformed for analyses and back-transformed for representation of the means.