Aberrant parasympathetic reactivity to acute psychosocial stress in male patients with schizophrenia spectrum disorders

Abstract

Autonomic dysfunction represents a core domain of the pathophysiology of schizophrenia spectrum disorders (SCZ), with aberrant physiologic arousal underlying maladaptive social and cognitive behaviors. Antagonistic parasympathetic and sympathetic systems support autonomic flexibility to appropriately regulate arousal and respond to environmental challenges, which can be modeled using physiologic measures. SCZ patients consistently show heightened basal stress, however, their parasympathetic reactivity to an acute psychosocial stressor is poorly understood. Heart period (HP-arousal), respiratory sinus arrhythmia (RSA-parasympathetic vagal activity), and their relationship were measured in SCZ patients (n=19) and healthy controls (n=20) at baseline and during psychosocial stress exposure. Parasympathetic vagal control of arousal, reflected in RSA-HP coupling, was assessed for the first time in SCZ. Patients demonstrated blunted physiologic reactivity (less change in heart period and respiratory sinus arrhythmia), a unique increase in respiratory sinus arrhythmia relative to baseline during recovery, and elevated arousal was associated with poor cognitive performance and greater positive symptoms. Arousal regulation was tightly controlled by parasympathetic activity in controls only, indicated by a strong association between changes in heart period and respiratory sinus arrhythmia. Results are the first to demonstrate maladaptive, inefficient parasympathetic arousal regulation (RSA-HP decoupling) in reaction to psychosocial stress in SCZ, representing an autonomic profile incompatible with appropriate social and emotional functioning.

1. Introduction

Schizophrenia spectrum disorders (SCZ) are characterized by significant impairments in social and cognitive domains. Autonomic function plays a pivotal role in promoting adaptive social and cognitive behaviors, and may precipitate the onset and severity of SCZ symptomatology (Appelhans and Luecken, 2006; Bär et al., 2005, 2008; Berntson and Cacioppo, 2004; Boettger et al., 2006; Hamilton et al., 2014; Jáuregui et al., 2011; Montaquila et al., 2015). Autonomic dysfunction contributes to the vulnerability for psychosis and disease exacerbation through deficient adaptation to environmental challenges, arousal dysregulation, and heightened stress sensitivity (Porges, 2007). SCZ patients consistently demonstrate aberrant spontaneous measures of autonomic function representing greater basal stress physiologic levels, and this is found to predict poor health (Bar et al., 2005; Clamor et al., 2016; Montaquila et al., 2015); however, parasympathetic reactivity and arousal regulation during an acute psychosocial stressor in SCZ is poorly understood.

Successful adaptation to environmental challenges, including psychosocial stress exposure, relies on a rapid autonomic nervous system (ANS) response to mobilize energy resources, modify arousal, respiration and heart rate, and elevate catecholamine secretion for immediate assessment and reaction to physiological disruptions in homeostasis (Appelhans and Luecken, 2006). The capacity of the ANS to adjust to changes in circumstance is referred to as autonomic flexibility and is critical for initiating and supporting appropriate biobehavioral responses. Autonomic flexibility relies on competing inhibitory, parasympathetic (i.e., vagal) and excitatory, sympathetic regulatory influences on the sinoatrial ‘pacemaker’ node of the heart, which produces beat-to-beat variation in the heart rate, or heart rate variability. This delicate balance of opposing parasympathetic and sympathetic systems differentially regulates visceral state by controlling the interval between consecutive heartbeats, which is reflected in cardiovascular measures including heart rate (or its reciprocal, heart period, the average interbeat interval (IBI)). Respiration modifies the parasympathetic vagal influence on the heart, generating high-frequency rhythmic heart rate oscillations at the frequency of spontaneous breathing (Denver et al., 2007). The relationship between respiration and heart rate variability is represented by the respiratory sinus arrhythmia, which provides a noninvasive, dynamic index of cardiac parasympathetic vagal tone (Lewis et al., 2012; Porges, 2007).

According to the polyvagal theory, a heightened arousal state stimulates vagal suppression, with reduced respiratory sinus arrhythmia, shortened heart period and elevated heart rate, to support mobilization and defensive “flight or fight” strategies. Alternatively, the vagal influence is disinhibited to foster calm, engaging social behaviors in safe, neutral contexts, which is indexed by greater respiratory sinus arrhythmia, longer heart period, and slower heart rate (Porges, 2007). Efficient parasympathetic-mediated arousal regulation is critical for promoting adaptive behaviors. As such, enhanced respiratory sinus arrhythmia and reliable respiratory sinus arrhythmia suppression are associated with greater executive functioning and cognitive performances (Hansen et al., 2003, 2004), social engagement (Hamilton et al., 2014) and emotion regulation processes in healthy individuals (Appelhans and Luecken, 2006; Lane et al., 2009). Conversely, deficient neural regulation of vagal activity fosters maladaptive physiological reactivity with hypersensitivity to the environment, and may contribute to compromised social engagement and affect expressivity in neuropsychiatric patients (Porges, 2007; Rubio et al., 2015).

Appropriate autonomic arousal (heart period) regulation relies on a parasympathetic vagal contribution, with a tight coupling between the change in heart period and change in respiratory sinus arrhythmia (Lewis et al., 2012; Umhau et al., 2002). Therefore, a decoupling between changes in heart period and respiratory sinus arrhythmia may represent inefficient cardiovascular regulation, diminished parasympathetic vagal control, and maladaptive sympathetically-dominated arousal regulation (Montaquila et al., 2015). Attenuated parasympathetic vagal control of arousal, with a limited association between changes in heart period and respiratory sinus arrhythmia, has been observed in groups with heightened stress sensitivity, including pain disorders (Eisenlohr-Moul et al., 2015), borderline personality disorder (Austin et al., 2007), posttraumatic stress disorder (Sahar et al., 2001), and in perpetrators of domestic violence (Umhau et al., 2002). However, the relationship between changes in heart period and respiratory sinus arrhythmia has not been studied in patients with SCZ and may provide insight on characterizing cardiovascular adaptability and symptom etiology.

Characterizing autonomic stress reactivity in SCZ is critical considering its fundamental role in adaptive behavioral states and emotional processing, which are impaired in SCZ. Patients with SCZ show heightened emotional reactivity to stressful events (Docherty et al., 2009), and appraise positive and negative events as being more difficult to manage and less controllable relative to controls (Horan et al., 2005). In addition to elevated stress sensitivity, SCZ patients exhibit acute hyperarousal with reduced parasympathetically-mediated heart rate variability or respiratory sinus arrhythmia (Boettger et al., 2006; Chang et al., 2009; Clamor et al., 2016), and aberrant autonomic activity in response to stress, including delayed recovery of heart rate variability in reaction to a mental arithmetic stressor (Castro et al., 2008), decreased respiratory sinus arrhythmia associated with subjective “real world” stress experiences (Kimhy et al., 2010), and increased heart rate following a socially evaluated cold-pressor test (Rubio et al., 2015). Furthermore, SCZ patients demonstrate diminished stimulation of the parasympathetic system during deep breathing, further suggestive of a deficient adaptive autonomic response (Liu et al., 2016). Moreover, decreased heart rate variability and vagal activity have been found in unmedicated patients with SCZ (Bär et al., 2005; Bär et al., 2012) and unaffected relatives (Bär et al., 2010, 2012), and have been associated with illness duration (Bär et al., 2005) and psychotic symptom severity (Cella et al., 2017; Kim et al., 2004). However, parasympathetic vagal-mediated respiratory sinus arrhythmia, and the ability of the cardiovagal system to flexibly respond to an acute experimental psychosocial stress manipulation has not been evaluated in SCZ patients, and was the goal of this study.

The current study investigated physiologic function before, during and after an acute psychosocial stressor, the Trier Social Stress Test (TSST), and defined the association between cardiac autonomic function and symptomatology in SCZ. The TSST is a reliable stress manipulation which has been used in diverse populations in various forms to robustly induce a well-described ANS response experimentally (Allen et al., 2014). A previous study implemented a modified TSST procedure in patients with SCZ, and found normal increases in heart rate, yet blunted hypothalamic-pituitary-adrenal (HPA) axis reactivity to the psychosocial stress exposure (Brenner et al., 2009). That said, parasympathetic vagal-mediated respiratory sinus arrhythmia has not been examined in SCZ patients in reaction to the TSST. Here we report a novel approach of characterizing the dynamic relationship between respiratory sinus arrhythmia and heart period to determine efficiency of the parasympathetic vagal system in SCZ.

The TSST is expected to induce heightened arousal to facilitate task engagement in control participants, indexed by shortened heart period and reduced respiratory sinus arrhythmia. However, we predict that SCZ patients will exhibit elevated arousal (shortened heart period) due to reduced parasympathetic vagal activity (reduced respiratory sinus arrhythmia) at baseline, consistent with an autonomic profile supporting defensive rather than engagement behaviors in calm, neutral situations. Furthermore, patients are expected to demonstrate altered autonomic stress reactivity relative to controls, with heart period changes less dependent on cardiac vagal responses (i.e., reductions in respiratory sinus arrhythmia) during stress exposure, and a delayed recovery to baseline levels of cardiac output. In controls, arousal regulation is expected to be dependent on a parasympathetic vagal influence, with a strong coupling between the change in heart period and the change in respiratory sinus arrhythmia. Given the heightened stress sensitivity observed in SCZ patients, it is hypothesized that changes in arousal (heart period) will not be efficiently regulated by parasympathetic withdrawal, reflected in a limited association between heart period and respiratory sinus arrhythmia. Considering previous evidence of autonomic abnormalities related to socioemotional, cognitive and psychotic symptom dimensions, we predict that aberrant measures of stress reactivity will be related to greater symptom severity, particularly positive symptoms consistent with Cella et al. (2017), and neurocognitive deficits.

2. Method

2.1 Participants

Twenty-one patients diagnosed with a schizophrenia-spectrum disorder (SCZ) and 21 unaffected control individuals (all males, ages 18–35 years) were studied. Patients were recruited from the Outreach and Support Intervention Services (OASIS) clinic in Carrboro, North Carolina, and referred to the study by their treating psychiatrist, who judged them stable and fit for the study. Control participants were recruited from the community via flyers. Recruitment was restricted to males in accordance with the male predominance in this age range (Abel et al., 2010), and to reduce variability in the small sample. Final analyses were performed with 19 SCZ patients and 20 controls because of recording issues. Patients met criteria for schizophrenia (n = 13), schizoaffective (n = 3), or schizophreniform disorder (n = 3) as assessed by the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (SCID DSM-IV), and diagnoses were confirmed using diagnostic criteria outlined in DSM-V (American Psychiatric Association, 2013; First et al., 1995). All patients were medicated with first-generation (n = 2) or second-generation (n = 17) antipsychotics and had stable symptoms. Four patients additionally reported using antidepressant medications, including Bupropion, Trazodone (n = 2), and Cymbalta, two patients reported using beta-blockers (Metoprolol, Atenolol), and one patient reported using a calcium channel blocker (Amlodipine Besylate). Control participants had no DSM-IV Axis-I diagnosis (confirmed using the SCID), were not taking antipsychotic medications, and did not have first-degree relatives with schizophrenia spectrum disorders. One control participant reported antidepressant use (Escitalopram). All participants spoke English, had normal or corrected vision, and no history of neurological disorders. Participants received $100.00 for their participation.

2.2 Clinical and neurocognitive measures

2.2.1 Clinical assessments

The Clinical Assessment Interview for Negative Symptoms (CAINS), a well-validated, comprehensive measure of negative symptomatology, was used to assess motivation, pleasure and emotional expression (Kring et al., 2013) in patients and control participants. Patients were additionally administered the Structured Clinical Interview - Positive and Negative Syndrome Scale (SCI-PANSS) to estimate symptom severity along positive, negative and general psychopathology subscales (Kay et al., 1987; Opler et al., 1992). All participants self-reported drug and alcohol use on the Alcohol Use Scale/Drug Use Scale (AUS/DUS), a reliable scale developed to examine substance use in SCZ (Drake et al., 1990).

2.2.2 Neurocognitive tests

Participants performed a computerized battery of neurocognitive assessments to estimate verbal intelligence using the North American Adult Reading Test (NAART; Uttl, 2002), evaluate executive function using the Continuous Performance Test-Identical Pairs (CPT-IP; Cornblatt et al., 1989) and Stroop Test (van Erp et al., 2015), and assess memory using the Auditory Verbal Learning Test (AVLT; Geffen et al., 1994) and Visuospatial Sequencing Test (VST). The Stroop interference score was calculated using the ‘Golden’ method to account for color naming and word reading (refer to Tables 1 and 3) (van Mourik et al., 2005; Golden 1978). Neurocognitive assessments were administered through the Computerized Multiphasic Interactive Neurocognitive DualDisplay System (CMINDS), and have demonstrated strong agreement with standard tests and high levels of test-retest reliability (O’Halloran et al., 2008; van Erp et al., 2015).

Table 1.

Demographics of Study Groups

Demographic Information	SCZ Patients (N=19)	CON group (N=20)	Statistic
Age (years)	26.26 ± 4.19	23.65 ± 4.67	F(1,37) = 3.37
Handedness (r/l)	15/4	17/3	χ2(1) = 0.24
Education (years)	13.37 ± 1.80	15.20 ± 1.67	F(1,37) = 10.84**
Avg parental education (years)	15.60 ± 2.29	15.89 ± 2.71	F(1,31) = 0.11
Race			χ2(3) = 5.62
White	9	10
Black	9	4
Asian	1	5
Hispanic	0	1
Clinical
CAINS-MAP	10.44 ± 6.12	2.39 ± 3.20	F(1,32) = 23.85***
CAINS-EXP	2.94 ± 3.17	0.44 ± 0.78	F(1,32) = 10.44**
Illness Duration (# of years)	6.29 ± 5.58
PANSS
Positive subscale	14.79 ± 5.07
Negative subscale	14.05 ± 4.75
General subscale	28.42 ± 7.74
Typical (1st gen.) antipsychotics
Haloperidol	1
Fluphenazine	1
Atypical (2nd gen.) antipsychotics
Olanzapine	2
Risperidone	2
Quetiapine	1
Aripiprazole	4
Clozapine	6
Paliperidone	2
Frequency of Drug & Alcohol Use (AUS/DUS)a
Tobacco	1.00 ± 1.10	0.58 ± 0.84	F(1,33) = 1.66
Alcohol	1.06 ± 1.24	1.84 ± 1.38	F(1,33) = 3.03
Marijuana, THC	0.94 ± 1.73	0.79 ±1.23	F(1,33) = 0.09
Neurocognitive Assessments
CPT-IP (Avg D-Prime) b	2.93 ± 0.81	3.38 ± 0.64	F(1,34) = 3.42
NAART (# of words)	30.44 ± 12.21	45.21 ± 7.18	F(1,35) = 20.37***
VST (tot. correct)	15.42 ± 3.22	19.10 ± 2.38	F(1,37) = 16.57***
AVLT (tot. recalled)	20.71 ± 6.05	29.33 ± 4.95	F(1,33) = 21.44***
AVLT (tot. recognized)	11.47 ± 3.10	14.10 ± 1.07	F(1,37) = 12.78**
Stroop c	9.05 ± 8.07	3.73 ± 2.59	F(1,37) = 7.88**
Questionnaires
Perceived Stress Scale d	17.50 ± 7.01	15.35 ± 4.89	F(1,36) = 1.22
Daily Stress Inventory-Freq e	24.44 ± 14.13	22.10 ± 6.61	F(1,36) = 0.44
Daily Stress Inventory-AIR f	2.83 ± 1.30	2.70 ± 0.84	F(1,36) = 0.13

Values presented as mean ± standard deviation. Abbreviations: CAINS-MAP: Clinical Assessment Interview for Negative Symptoms- Motivation and Pleasure Scale; CAINS-EXP: CAINS-Expression Scale; PANSS: Positive and Negative Syndrome Scale; THC: tetrahydrocannabinol; CPT-IP: Continuous Performance Test-Identical Pairs; NAART: North American Adult Reading Test; VST: Visuospatial Sequencing Test; AVLT: Auditory Verbal Learning Test.

^aFrequency of tobacco (0=no use,1=occasionally, 2=<10/day, 3=11–25/day, 4>25/day; frequency of drugs/alcohol (0=no use, 1=1–2X/month, 2=3–4X/month, 3=1–2X/week, 4=3–4X/week, 5=almost daily)

^bCPT-IP: average sensitivity index (D-prime) for 2,3, and 4-digit trials

^cStroop interference score (seconds) = Incongruent color words - [(congruent color words*color naming)/(congruent color words+color naming)]

^dPerceived Stress score reflects sum of all scale items (reversing responses for items 4,5,7,8)

^eDaily Stress Inventory-Freq: frequency of stressful events

^fDaily Stress Inventory- average impact rating (AIR) of endorsed events

^*p < 0.05,

^**p < 0.01,

^***p < 0.01

2.2.3 Self-report stress and affect questionnaires

The appraisal of stressful events, and frequency and impact of daily stressors were evaluated using the Perceived Stress Scale (PSS), which has good internal consistency in a clinical sample (Cohen et al., 1983; Hewitt et al., 1992) and the Daily Stress Inventory, which is a reliable assessment of the source and impact of minor stressful events (DSI; Brantley et al., 1987). Subjective Stress and Affect (SSR) ratings were collected at six time-points (baseline, before TSST, after TSST, recovery 1(rec1), recovery 2(rec2), recovery 3(rec3)) to assess five affective composites (stressed, happy, irritated, depressed, and overwhelmed), and validate the stress manipulation. Clinical and demographic information are presented in Table 1.

2.3 Psychophysiological recordings

2.3.1 Data acquisition

Cardiovascular activity was recorded at 1kHz using the ECG amplifier module of Biopac and the AcqKnowledge acquisition software package (Biopac Systems, Inc., CA). Three self-adhering, disposable electrodes (Ultratrace) were positioned below the right collarbone, over the rib cage on the left side, and above the pelvic bone on the right side.

2.3.2 Signal processing

Interbeat intervals (IBIs) were extracted from the raw ECG data using in-house software implemented through LabView2014 version 14.0f2 (developed by Drs. Maria Davila and Greg Lewis, UNC). IBIs were visually inspected and missed R-wave detections and errors were corrected using CardioEdit (Brain-Body Center, Chicago, IL). Respiratory sinus arrhythmia was quantified using standard adult parameters that define respiratory sinus arrhythmia from frequencies between 0.12 to 0.40 Hz, and analyzed using the Porges-Bohrer method of respiratory sinus arrhythmia analysis in CardioBatch (Brain-Body Center, Chicago, IL). As previously described in Lewis et al. (2012), respiratory sinus arrhythmia estimates were calculated from sequential filtered time series over several short (30 second) epochs, which were averaged to generate the final estimate of respiratory sinus arrhythmia. Each epoch estimate was transformed by its natural logarithm (ln(ms²)) to ensure that the distributions of respiratory sinus arrhythmia estimates were normal. Heart period was calculated as the average time interval in milliseconds between successive R-waves.

2.4 Procedure

Informed consent was provided in accordance with the University of North Carolina at Chapel Hill Institutional Review Board. Participants completed an interview session and a psychophysiological recording session on separate days. Participants were permitted to smoke and consume caffeine until two hours prior to the experiment, consistent with their habitual smoking and drinking patterns.

2.4.1 Psychophysiological recording session

Participants completed a urinary toxicology screen upon arrival at the psychophysiological recording session, followed by ECG electrode placement. Continuous, short-term electrocardiograms were recorded at 5 time-points to generate reliable estimates of heart rate parameters (Lewis et al., 2012; Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology, 1996): baseline resting state before stress (bas, 3 minutes), during the stress manipulation (TSST, 13 minutes), resting state after stress (recovery 1 (rec1), 3 minutes) and a final resting state approximately 80 minutes post-stress onset (recovery 2 (rec2), 3 minutes) to capture stress reactivity and recovery.

2.4.2 Stress protocol

The procedure for the acute psychosocial stressor (TSST) has been described in detail elsewhere (Kirschbaum et al., 1993). Briefly, the current protocol consisted of a 3 minute preparation period (rather than 10 minutes in the original protocol; Kirschbaum et al., 1993), followed by a 5-minute mock job interview and 5 minutes of challenging serial subtraction (i.e., subtract 7 from 2000). The committee consisted of two individuals in white lab coats who were introduced as “academic experts” (to distinguish them from medical professionals) trained in evaluating nonverbal behavior. The committee was instructed to refrain from providing any facial feedback. Participants were informed that they were being video-recorded for subsequent performance evaluation.

2.5 Statistical analyses

2.5.1 Clinical and neurocognitive assessments

Statistical analyses were performed using IBM SPSS statistical software, version 24.0 (IBM, 2016). Demographic, neurocognitive, clinical and self-report questionnaire data were analyzed using analyses of variance (ANOVAs) for continuous variables, and chi-square tests for categorical variables. Composite scores were calculated for executive function (reversed, z-transformed Stroop interference score (van Mourik et al., 2005) minus z-transformed CPT d-prime discrimination score) and memory (sum of z-transformed AVLT recall, delayed recall, recognition, and VST scores). NAART (number of correctly identified words) was used to estimate verbal intellectual ability.

2.5.2 Psychophysiological measures and SSR ratings

A Group (Controls, SCZ patients) X Time (bas, TSST_prep, TSST_speech, TSST_math, rec1, rec2) X Measure (heart period, respiratory sinus arrhythmia) repeated measures (rm)-ANOVA was performed to evaluate psychophysiological measures of stress reactivity. Polynomial and repeated within-subject contrasts were implemented to assess linear and quadratic trends and the comparison of adjacent levels, respectively. Age was entered as a covariate for the ECG analyses to account for age-mediated physiological changes (O’Brien et al., 1986; Voss et al., 2015). Stress and affect ratings were assessed using a Group X Affect (Happy, Stressed, Depressed, Overwhelmed, Irritated) X Time (6 time-points) rm-ANOVA with repeated within-subject contrasts. Post-hoc pairwise comparisons were Bonferroni-corrected for multiple comparisons and Greenhouse-Geisser epsilon corrections were used when the sphericity assumption was violated for all rm-ANOVA analyses.

2.5.3 Heart period and respiratory sinus arrhythmia moderation analysis

The contribution of parasympathetic vagal influence on arousal regulation was assessed using a multiple linear regression model to determine whether group status moderated the association between changes in respiratory sinus arrhythmia (ΔRSA) and changes in heart period (ΔHP) from the first resting state to the stress introduction (bas and TSST_prep) and from the end of the stress protocol to the initial recovery period (TSST_math and rec1). To further evaluate respiratory sinus arrhythmia moderation by group, simple regression slopes of ΔRSA on ΔHP were tested for patients and controls using Hayes’ PROCESS macro (Hayes, 2013), as utilized in Lewis et al. (2012).

2.5.4 Clinical, neurocognitive and physiological correlations

Partial correlations corrected for age were performed to assess the relationship between physiological measures (respiratory sinus arrhythmia and heart period before, during and after stress, change in heart period and respiratory sinus arrhythmia during stress onset and recovery), and clinical measures (illness duration, positive, negative, general PANSS subscales, motivation and pleasure, expression CAINS scores) in patients, and neurocognitive measures (NAART, CPT, VST, AVLT scores, Stroop, executive function and memory composite scores) in patients and controls separately, and with combined groups.

3. Results

3.1 Physiologic measures²

3.1.1 Heart period

A main effect of Time for heart period indicated that heart period shortened significantly during stress exposure (F(2.32, 64.88) = 4.65, p = 0.01,η_p² = 0.14) across groups, thus validating the stress manipulation. Patients demonstrated shorter heart period at baseline (p < 0.05), and a more gentle decline or blunted heart period change in response to stress, represented in the significant quadratic effect of Group by Time on heart period (F(1, 28) = 10.06, p < 0.01, η_p² = 0.26). Planned within-subject contrasts revealed that controls exhibited a steeper decline in heart period from “bas” to “TSST_prep” compared with patients (F(1, 28) = 8.02, p < 0.01, η_p² = 0.22) (Table 2; Figure 1a).

Table 2.

Physiologic measures before, during and after psychosocial stressor

		Controls	SCZ Patients	Statistic
Baseline	HR	67.35 (11.33)	79.90 (14.15)	F(1,37) = 9.39**
	HP	923.45 (155.96)	777.41 (132.60)	F(1,37) = 9.87**
	RSA	6.53 (1.05)	5.38 (1.71)	F(1,37) = 6.52*
TSST-prep	HR	84.52 (12.53)	87.21 (15.33)	F(1,35) = 0.34
	HP	731.83 (116.93)	712.55 (122.85)	F(1,35) = 0.24
	RSA	5.66 (0.74)	4.85 (2.21)	F(1,35) = 2.18
TSST-speech	HR	81.08 (11.87)	87.20 (13.41)	F(1,35) = 2.17
	HP	761.49 (107.66)	707.67 (104.80)	F(1,35) = 2.37
	RSA	6.32 (0.69)	5.47 (1.84)	F(1,35) = 3.52
TSST-math	HR	79.32 (12.93)	84.84 (13.28)	F(1,36) = 1.69
	HP	782.73 (124.21)	729.17 (113.30)	F(1,36) = 1.93
	RSA	6.12 (0.94)	5.59 (1.70)	F(1,36) = 1.38
Recovery 1	HR	65.98 (11.55)	75.25 (13.51)	F(1,36) = 5.16*
	HP	943.56 (173.32)	825.97 (140.33)	F(1,36) = 5.28*
	RSA	6.24 (1.22)	5.84 (1.75)	F(1,36) = 0.68
Recovery 2	HR	65.63 (10.73)	75.58 (13.16)	F(1,33) = 5.97*
	HP	946.98 (153.63)	819.78 (124.71)	F(1,33) = 7.27*
	RSA	6.33 (1.06)	6.11 (1.62)	F(1,33) = 0.21

Values are presented as mean (standard deviation)

Abbreviations: HR: heart rate (beats per minute, bpm); HP: heart period (milliseconds); RSA: respiratory sinus arrhythmia (ln(ms²)); TSST: Trier Social Stress Test

^*p < 0.05,

^**p < 0.01

Figure 1. HP and RSA before, during and after stress manipulation.

Visualization of HP (A) and RSA (B) during 6 task recordings for SCZ patients (dashed line) and Controls (solid line) (age as covariate).

3.1.2 Respiratory sinus arrhythmia

Planned contrasts demonstrated a linear effect of Group by Time on respiratory sinus arrhythmia (F(1, 28) = 7.35, p = 0.01, η_p² = 0.21). Specifically, patients exhibited a modest increase in respiratory sinus arrhythmia between baseline and rec2 (F(1, 28) = 9.73, p < 0.01, η_p² = 0.26), while control subjects failed to recover respiratory sinus arrhythmia levels to their, admittedly much greater, initial levels by rec2 (Table 2; Figure 1b).

3.1.3 Moderation analysis: vagal contribution to changes in heart period

The model representing the coupling of change in respiratory sinus arrhythmia (ΔRSA) and change in heart period (ΔHP) was confirmed, as ΔRSA significantly predicted 61.4% of the variance in ΔHP during the onset of stress (R² = 0.61, F(3, 33) = 17.48, p < 0.001), and 47.4% of the variance in ΔHP during recovery (R² = 0.47, F(3, 33) = 9.93, p = 0.001). This effect was significantly moderated by an interaction between ΔRSA and group status, which accounted for 15.3% of the variance in ΔHP (ΔR² = 0.15, F(1, 33) = 13.07, p = 0.001) during stress onset, and 12.3% of the variance in ΔHP during recovery from stress exposure (ΔR² = 0.12, F(1, 33) = 7.72, p < 0.01). A decline in respiratory sinus arrhythmia accompanied a significantly shorter heart period during stress onset (t = 5.53, p < 0.001), and increased respiratory sinus arrhythmia was significantly associated with a longer heart period during recovery (t = 4.91, p < 0.001) for control participants; however, this relationship between ΔRSA and ΔHP was not found for SCZ patients (ps > 0.05) (Figure 2, Figure S1).

Figure 2. Simple slopes of ΔRSA on ΔHP at stress onset and recovery.

Simple slopes of ΔRSA on ΔHP for SCZ patients (dashed line) and Controls (solid line) between baseline and speech preparation (stress onset) (A), and between the math test and first recovery resting state (Recovery) (B). Simple slopes plots (Hayes and Matthes, 2009) illustrate the difference between groups at +/− 1 SD of the sample distribution for each change metric: Stress onset (A): −2.43 ms HP (small) and −244.95 ms HP (large), 0.20 ln(ms²) RSA (small) and −1.54 ln(ms²) RSA (large); Recovery (B): 21.1 ms HP (small) and 232.6 ms HP (large), −0.72 ln(ms²) RSA (small) and 1.06 ln(ms²) RSA (large).

3.2 Subjective stress/affect ratings

Stress exposure modified subjective measures of affect revealed in a main effect of Time (F(3.17, 107.60) = 17.24, p < 0.001, η_p² = 0.34). Participants reported greater levels of “stressed” and “happy” throughout the physiological recording session which was revealed in a main effect of Affect (F(1.75, 59.41) = 56.00, p < 0.001, η_p² = 0.62), and these ratings were differentially impacted by stress exposure, indicated by a Time X Affect (F(7.55, 256.73) = 15.14, p < 0.001, η_p² = 0.31) interaction. For both groups, repeated contrasts indicated that stress exposure elicited changes in subjective affect between “before TSST” and “after TSST” (F(1, 34) = 9.83, p < 0.01, η_p² = 0.22), and “after TSST” and “rec1” time-points (F(1, 34) = 50.68, p < 0.001, η_p² = 0.60). Planned pairwise comparisons revealed that ratings of “happy” deteriorated (p < 0.001) while ratings of “stressed” (p < 0.01) and “irritated” (p < 0.01) increased following stress onset for both groups; however, ratings of “stressed” significantly increased and “happy” decreased between “baseline” and “before TSST” for controls and were not significantly different from baseline until “after TSST” for patients. Patients demonstrated a unique enhancement of “irritated” ratings between “baseline” and “after TSST” which was not found for controls (p < 0.05). Patients additionally reported greater levels of “depressed” during the “rec2” time-point relative to control participants (p < 0.05)(Figure 3).

Figure 3. Subjective affect ratings in response to stress exposure.

Stress and affect ratings collected at 6 time-points throughout the stress manipulation to measure subjective ratings of stressed, happy, irritated, depressed and overwhelmed.

3.3 Clinical, neurocognitive, and stress correlations

Significant correlations are presented in Table 3. No significant correlations were found for illness duration, negative and general PANSS subscales, CAINS symptom scores, NAART or VST.

Table 3.

Partial correlations between clinical, neurocognitive and physiologic measures

PANSS symptoms	Stress Measure	SCZ Patients
		r	P	df
PANSS Positive	HP_BeforeStress	−0.69	0.02	16
	HP_DuringStress	−0.59	0.01	16
	HP_AfterStress	−0.58	0.01	16
Neurocognitive Measures		Combined Groups
CPT D-prime	HP_BeforeStress	0.40	0.02	33
	HP_AfterStress	0.34	0.04	33
	HP_StressOnset	−0.41	0.02	31
	HP_Recovery	0.42	0.01	31
AVLT_recall	HP_Recovery	0.39	0.03	30
AVLT_delayed recall	HP_Recovery	0.34	0.04	34
AVLT_recog	RSA_BeforeStress	0.32	0.05	36
	HP_Recovery	0.35	0.04	34
Stroop	HP_BeforeStress	−0.36	0.03	36
	HP_StressOnset	0.36	0.03	34
Executive Function Score	HP_BeforeStress	0.40	0.02	33
	HP_AfterStress	0.34	0.04	33
	HP_StressOnset	−0.36	0.04	31
	HP_Recovery	0.46	0.01	31
Memory Score	HP_Recovery	0.38	0.03	30
	RSA_Recovery	0.43	0.01	30
		Controls
CPT D-prime	HP_Recovery	0.5	0.04	15
AVLT_delayed recall	HP_BeforeStress	−0.46	0.05	17
AVLT_delayed recall	HP_DuringStress	−0.51	0.03	16
AVLT_recog	RSA_Recovery	0.73	0.001	15
Executive Function Score	RSA_Recovery	0.57	0.02	15
Memory Score	HP_Recovery	0.55	0.03	13
		SCZ Patients
AVLT_recall	RSA_StressOnset	−0.5	0.05	14

Abbreviations: HP: heart period (milliseconds); RSA: respiratory sinus arrhythmia (ln(ms²); HP_StressOnset: change in heart period during the onset of stress; HP_Recovery: change in HP during recovery; RSA_Recovery: change in RSA during recovery; PANSS: Positive and Negative Syndrome Scale

^aContinuous Performance Test (CPT) - average sensitivity index (d′)

^bAuditory Verbal Learning Test (AVLT) - total recalled

^cAuditory Verbal Learning Test (AVLT) - recalled after delay

^dAuditory Verbal Learning Test (AVLT) - total recognized after delay

^eStroop interference score (seconds) = Incongruent color words - [(congruent color words*color naming)/(congruent color words+color naming)]

^fExecutive Function Composite score = reversed Stroop interference z-score + CPT-d′ z-score

^gMemory Composite score = AVLT(recall, delayed recall, recognized) z-scores +VST (total correct) z-score

4. Discussion

The goal of this study was to elucidate the pathophysiological mechanisms underlying key cognitive and affective symptom domains of SCZ, and characterize parasympathetic reactivity and recovery from stress exposure in SCZ patients using a well-validated acute psychosocial stressor. A novel analytical approach was employed to assess dynamic parasympathetic vagal control of arousal for the first time in SCZ patients. The psychosocial stress protocol (TSST) was overall effective in inducing a physiological and subjective stress response in patients and controls. Despite comparable subjective ratings of perceived stress (PSS), SCZ patients demonstrated elevated arousal and diminished parasympathetic control of cardiac output prior to stress onset relative to controls. Patients additionally showed blunted parasympathetic reactivity and abnormal arousal regulation in response to stress exposure. Aberrant physiologic activity was associated with enhanced symptom severity and neurocognitive deficits, indicating a significant relationship between maladaptive arousal regulation and psychosis.

Patients exhibited enhanced physiologic arousal at baseline relative to controls (shorter heart period, smaller respiratory sinus arrhythmia) representing hypersensitivity to a neutral environment and a maladaptive autonomic profile optimizing defensive, avoidance behaviors incompatible with appropriate social engagement and emotional reactivity. This finding is consistent with previous reports of disrupted parasympathetic-mediated (high-frequency) heart rate variability in SCZ patients (Montaquila et al., 2015), unmedicated patients (Alvares et al., 2016), and in unaffected first-degree relatives (Castro et al., 2009; Jáuregui et al., 2011), suggesting that parasympathetic dysfunction may be a valuable vulnerability marker in the early detection and prediction of psychosis. However, parasympathetic dysfunction is not specific to SCZ, and aberrant autonomic function may indicate susceptibility for cognitive and affective symptoms shared across neuropsychiatric disorders (Quintana et al., 2016).

Despite similar contradictory shifts in subjective ratings of happiness and stress during the stress manipulation, patients demonstrated a unique maladaptive autonomic profile. Control participants exhibited an appropriate parasympathetic vagal withdrawal to support stress task engagement, reflected in reduced respiratory sinus arrhythmia during the stress manipulation, and vagal disinhibition (increased respiratory sinus arrhythmia) following stress to facilitate recovery (Porges, 2007). In contrast, patients with SCZ revealed blunted heart period reactivity to stress exposure and a distinct divergence in respiratory sinus arrhythmia from baseline to post-stress recovery. The unexpected stronger vagal disinhibition and increased respiratory sinus arrhythmia found during recovery time-points in patients (as compared to their initial levels) may indicate a slow recovery of parasympathetic vagal suppression over the course of the experiment due to desensitization to the environment. The blunted physiological reactivity could represent deficits in autonomic flexibility, or an inability for the patients to adapt appropriately to environmental challenges and behavioral demands. A previous study (Liu et al., 2016) also showed that SCZ patients had attenuated parasympathetic vagal enhancement (reduced respiratory sinus arrhythmia) in response to deep breathing, a special form of physiological change. Together with the current results, these findings suggest that the restricted autonomic flexibility necessary for appropriately responding to environmental challenges might be global and not only occur in one or two specific situations. The rigidity of the parasympathetic vagal system in patients and their diminished adaptive autonomic response to opposing biobehavioral demands may reflect suppressed interaction of brain stem and higher regulatory neural networks in SCZ (Williams et al., 2004).

The polyvagal perspective proposes that neural regulation of the ANS conforms to a phylogenetic response hierarchy, with the phylogenetically newer myelinated vagal system providing primary regulation. The rapid dynamics of this system dominate regulation of the heart, supporting social communication and calm behavioral states, and suppressing the more primitive sympathetic and unmyelinated dorsal vagal systems (Porges, 2007). When situational demands cannot be met by control over the vagal inhibitory system, the lower systems are then recruited to support adaptive behaviors, as in the fight-or-flight state of sympathetic dominance. Accordingly, the higher-order myelinated vagal pathways in patients may be deficient, leaving the unmyelinated vagal component or sympathetic system to regulate arousal, consistent with the Jacksonian principle of dissolution (Porges, 2007).

Arousal regulation was tightly controlled by parasympathetic activity in control participants, as parasympathetic-mediated respiratory sinus arrhythmia reduction in response to stress onset and increase during recovery was accompanied by a significant shortening and subsequent lengthening of heart period. However, patients revealed an uncoupling of the expected heart period and respiratory sinus arrhythmia relationship, as changes in arousal (heart period) did not appear to rely on a parasympathetic vagal contribution. As such, the altered cardiovascular adaptability observed in SCZ patients may reflect the failure of the sympathetic system to regulate arousal in the absence of efficient parasympathetic vagal control (Montaquila et al., 2015). Accordingly, modulation of arousal in patients may be hampered by inefficient increases in sympathetic and/or reductions in unmyelinated parasympathetic activity not reflected in respiratory sinus arrhythmia magnitude changes. Aberrant parasympathetic activity and potential lack of dynamic sympathetic activity may contribute to impaired arousal regulation in patients, reflected in aberrant heart period. The mechanisms underlying the atypical relationship between change in respiratory sinus arrhythmia and change in heart period in patients are unknown; however, the reduced respiratory sinus arrhythmia levels at initial baseline suggest that a diminished capacity for myelinated vagal control and a chronically altered parasympathetic-sympathetic balance may be responsible. Results from the current study suggest that deficiencies in vagal control and unbalanced arousal regulation may underlie the aberrant autonomic reactivity observed in patients, and support maladaptive social and emotional symptom domains.

The significant correlation between maladaptive stress reactivity and elevated positive symptom severity in patients further confirms an association between autonomic dysregulation and psychiatric symptoms (Cella et al., 2017; Henry et al., 2010). The relationship between impaired cognitive performance and physiologic indices of heightened arousal supports a connection between prefrontal-mediated tasks, particularly executive control, and vagal function (Thayer et al., 2009). Hyperresponsivity to stress onset (greater change in heart period) and deficient recovery (less change in heart period) accompany inferior executive functioning, confirming that dynamic autonomic regulation is critical for successful cognitive functioning.

Limitations should be addressed in future studies. Notably, the inferential power of the study is limited given the modest sample size restricted to male participation, and may have underpowered our results for some tests. However, the sample size was greater than (Albus et al., 1982; Breier et al., 1988; Jansen et al., 2000), or comparable (Akar et al., 2015; Boettger et al., 2006) to other studies investigating the stress response in SCZ. Additionally, results were robust within the sample with moderate to large effects, and augment previous reports of aberrant autonomic activity in SCZ. While recruitment for this study was restricted to males in accordance with the male predominance in this age range (Abel et al., 2010), and to reduce variability in the sample, future studies elucidating sex differences in the parasympathetic stress response in SCZ will be critical for improving external validity and generalizability of the results.

Medication exposure, smoking and caffeine consumption, body-mass-index (BMI) and physical activity are all thought to modify peripheral autonomic output (Blom et al., 2009; Foley and Kirschbaum, 2010). While participants were instructed to refrain from smoking and consuming caffeine prior to the study session, these behaviors could have influenced heart rate parameters of stress reactivity. Furthermore, BMI and physical activity of the participants were not recorded, even though they are both thought to impact physiological parameters (Koenig et al., 2014; Rimmele et al., 2009). Patients with SCZ are expected to have higher BMI and reduced physical activity, due in part to their antipsychotic medication exposure (Haring et al., 2015) and sedentary lifestyle (Stubbs et al., 2016; Vancampfort et al., 2017), which may result in reduced respiratory sinus arrhythmia (Koenig et al., 2014). The explanatory power of our data on other potential confounds (e.g., BMI and physical activity) is limited, and controlling for these variables is warranted in future studies.

We did not exclude participants based on medication use and all patients in this sample were on antipsychotic agents, consistent with American Psychiatric Association guidelines for the treatment of schizophrenia (Patel et al., 2014), and reflective of the male schizophrenic population. The distinct impact of medication exposure on autonomic function remains unclear. While antipsychotic medications may have an effect on parasympathetic activity, the effect has been proposed to be only minimal and predominately insignificant (Bar et al., 2005; Boettger et al., 2006). Several participants in the current study were taking additional agents, including antidepressants and beta blockers, which have been shown to differentially influence physiological measures (Malfatto et al., 2003; O’Regan et al., 2015); although, these effects may be medication specific (Yeh et al., 2016). Despite the aforementioned effects of medication exposure, dysregulated parasympathetic function is found in medication-free patients with psychotic disorders (Alvares et al., 2016), and in unaffected relatives of SCZ patients (Castro et al., 2009; Jáuregui et al., 2011). Consequently, it seems unlikely that the current physiological results are merely attributable to medication use. Future studies can use autonomic reactivity tasks (e.g., TSST), which are proposed to be more sensitive in differentiating illness status (Liu et al., 2016), to elucidate the impact of specific medications on parasympathetic reactivity and to determine vulnerability.

In light of the limitations discussed above, this study offers valuable insight into the pathophysiological mechanisms underlying debilitating symptoms of SCZ, and may facilitate the identification of therapeutic targets for improved treatment efficacy. The current study applied a novel analytical approach for elucidating dynamic autonomic control which can extended to other clinical populations to refine characterization of autonomic reactivity and symptom etiology. This is the first study to demonstrate aberrant parasympathetic mediated HRV during the completion of an acute psychosocial stressor in male patients with SCZ, and report inefficient parasympathetic arousal regulation, indexed by the decoupling of respiratory sinus arrhythmia and heart period in response to and recovery from stress exposure. Patients’ aberrant autonomic profile at baseline and blunted arousal response to stress may indicate restricted autonomic flexibility necessary for appropriately responding to environmental challenges, thus promoting maladaptive cognition and behavior. Results reinforce the potential value of considering autonomic dysfunction as a central deficit in the social and cognitive symptoms of SCZ, and targeting the autonomic system using biofeedback techniques and vagal nerve stimulators to restore psychophysiological resources and foster appropriate social engagement, executive function and cognitive regulation in SCZ patients (Clamor et al., 2016; Howland, 2014).

Highlights.

• Novel analytical approach used to characterize parasympathetic arousal regulation.

• Schizophrenia spectrum patients show aberrant autonomic flexibility during stress.

• Patients exhibit maladaptive basal autonomic activity relative to healthy controls.

• Atypical physiologic activity is associated with symptoms and cognitive deficits.

Abbreviations

RSA

respiratory sinus arrhythmia

HP

heart period

HRV

heart rate variability

ANS

autonomic nervous system

IBI

interbeat interval

TSST

Trier Social Stress Test