Reduced heart rate variability (HRV) has proved to be an independent risk factor for cardiac emergencies. The present study aimed to evaluate differences in HRV in healthy volunteers high or low in anxiety sensitivity (AS) and potential changes in HRV following an interoceptive exposure (IE) training. We aimed to identify subject features that are associated with potential changes in HRV. Methods: Data were obtained in five subjects high and five subjects low in AS (aged 19 to 23).
ECG recordings were conducted in supine position in a mock scanner environment during three experimental conditions. Recordings were repeated after three to seven days of IE training. The square root of successive R-R- interval differences (RMSSD) was calculated for HRV assessment. Potential correlations between subject features and change in HRV were tested with Spearman’s rank correlation. Results: No significant HRV differences between subjects high and low in AS in any of the experimental conditions were observed, neither before nor after the IE training. On a descriptive level, subjects high in AS showed lowered HRV compared to subjects low in AS before IE. After IE, subjects low in AS demonstrated increased HRV, while subjects high in AS showed decreased HRV measures. Correlation analyses revealed no significant associations.
Conclusion: Descriptive results indicate that there are AS-related differences in HRV, with subjects high in AS showing lowered HRV as hypothesized. Following IE, subjects low in AS showed an increase in HRV, while a decrease occurred in subjects high in AS. The mechanisms of IE require further investigation. If replicated in a larger sample and with adequate study design, IE could prove to lower individual risk for cardiac complications by increasing HRV. Future design implications are discussed.
Table of contents
Abstract
Introduction
Objective
Health stability through heart rate variability
Relevance of heart rate variability assessment to pathological anxiety
A focus on interoceptive exposure
Methods
Procedure
Measures
Clinical assessments
ECG Recording
Functional MRI data acquisition
Experimental procedures
IE training
Cardiac symptom provocation task
Data preprocessing and analysis
ECG data
Functional MRI data
Statistical analyses
Results
Sample characteristics
Group comparison of heart rate variability
Change in heart rate variability
Characteristics associated with change in heart rate variability
Discussion
Group comparison of heart rate variability
Change in heart rate variability
Characteristics associated with change in heart rate variability
Additional observations
Study limitations
Conclusion and perspective
References
APPENDIX A
APPENDIX B
APPENDIX C
List of tables
Table 1 Demographic and psychological characteristics (pre-training) of the study sample
Table 2 Comparison of cardiovascular parameters between AS groups
Table 3 Brain activation related to the anticipation of cardiac interoceptive stimulation
Table 4 Bivariate correlations between pre-training characteristics and logRMSSD change score
Table B1 Comparison of MDMQ mood profiles between AS groups in the course of ECG recording
Table B2 Cardiovascular parameters as a function of AS group and gender
Table B3 Comparison of logRMSSD indices between the two resting conditions
Table B4 Comparison of MDMQ mood profiles before resting and after resting while listening to EPI noise-profile
Table B5 Comparison of logRMSSD change scores (∆logRMSSD) between AS groups
Table B6 Brain activation during the cardiac symptom provocation task, as revealed by region-of-interest-analyses
Table B7 Comparison of logRMSSD indices between resting and metronome-controlled breathing
Table B8 Comparison of MDMQ mood profiles between resting and metronome- controlled breathing
Table B9 Change in ASI sum score from online screening to post-training
List of figures
Figure 1 Medians of logarithmised RMSSD (logRMSSD) indices as a function of experimental condition and AS group at pre-training (pre) and post-training (post)
Figure A1 Schematic of the cardiac symptom provocation task
Figure C1 Mood profiles obtained with the Multidimensional Mood Questionnaire (MDMQ) in the course of pre-training ECG recording
Figure C2 Mood profiles obtained with the Multidimensional Mood Questionnaire (MDMQ) in the course of post-training ECG recording
Figure C3 Correlations between change in logRMSSD (∆logRMSSD) and pre-training characteristics
List of abbreviations
Abbildung in dieser Leseprobe nicht enthalten
Abstract
Background: Reduced heart rate variability (HRV) has proved to be an independent risk factor for cardiac emergencies. The present study aimed to evaluate differences in HRV in healthy volunteers high or low in anxiety sensitivity (AS) and potential changes in HRV following an interoceptive exposure (IE) training. We aimed to identify subject features that are associated with potential changes in HRV. Methods: Data were obtained in five subjects high and five subjects low in AS (aged 19 to 23). ECG recordings were conducted in supine position in a mock scanner environment during three experimental conditions. Recordings were repeated after three to seven days of IE training. The square root of successive R-R- interval differences (RMSSD) was calculated for HRV assessment. Potential correlations between subject features and change in HRV were tested with Spearman’s rank correlation. Results: No significant HRV differences between subjects high and low in AS in any of the experimental conditions were observed, neither before nor after the IE training. On a descriptive level, subjects high in AS showed lowered HRV compared to subjects low in AS before IE. After IE, subjects low in AS demonstrated increased HRV, while subjects high in AS showed decreased HRV measures. Correlation analyses revealed no significant associations. Conclusion: Descriptive results indicate that there are AS-related differences in HRV, with subjects high in AS showing lowered HRV as hypothesized. Following IE, subjects low in AS showed an increase in HRV, while a decrease occurred in subjects high in AS. The mechanisms of IE require further investigation. If replicated in a larger sample and with adequate study design, IE could prove to lower individual risk for cardiac complications by increasing HRV. Future design implications are discussed.
Introduction
Objective
Heart rate variability (HRV) reflects fluctuations in the time series of consecutive heartbeats and is of vital importance to human health (Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology [Task Force], 1996). Reduced HRV has proved to be an independent risk factor for developing severe disturbances of physical health such as coronary artery disease, cardiopathies or neuropathies (Task Force, 1996; Tsuji et al., 1996), and is further associated with disturbances of mental health such as depression or anxiety disorders (Friedman, 2007; Siepmann et al., 2005). In clinical psychological research, HRV has been discussed as an aetiologically relevant factor in the development of anxiety disorders, largely regarding panic disorder (PD; Friedman & Thayer, 1998). However, to understand the nature and extent of cardiac-autonomic involvement in the onset of clinical anxiety, it is essential to investigate HRV in healthy subjects with no history of anxiety disorders, but who may be especially susceptible to develop clinical forms of anxiety. For this, subjects high in anxiety sensitivity (AS) might represent a suitable sample since they exhibit a similarly high level of fear of anxiety-related bodily symptoms and interoceptive sensitivity compared to patients with PD (Domschke, Stevens, Pfleiderer, & Gerlach, 2010; McNally, 2002). Since HRV is dynamically adapted from beat to beat (Task Force, 1996), and since cardiac control and interoceptive processing are mediated through shared anatomical interfaces (Craig, 2003; Thayer & Lane, 2009; Verberne & Owens, 1998), this raises the question whether reduced HRV can be adjusted through interoceptive exposure (IE). Accordingly, the present study concerns two research objectives: Firstly, we assess AS-related differences in HRV, and secondly, we address the alterability of HRV through a brief interoceptive cognitive-behavioural intervention module in healthy volunteers.
Health stability through heart rate variability
Whether we are exposed to physical stress, intense mental or emotional engagement, our heart is challenged to meet these changing environmental demands (Appelhans & Luecken, 2006). Feeling one’s heart palpitating is frightening to anxiety-prone individuals (Eifert et al., 2000), but fluctuations of heart rate (HR) in response to changing stressors are indeed of cardinal importance for healthiness (Task Force, 1996; Thayer & Lane, 2007). HRV is a well-established index of cardiac vagal autonomic balance demonstrating short- and long- term variability in HR (Loellgen, 1999). It enables a judgement of the heart’s power to adjust the temporal interval between two consecutive heartbeats as a function of physical or mental stress (Ziemssen, Suess, & Reichmann, 2002). This beat-to-beat fine-tuning adjustment is controlled by the integrative action of sympathetic and vagal parasympathetic neural discharges on the sinoatrial node which is the primary pacemaker of the heart (Ziemssen et al., 2002). The activity of the parasympathetic nervous system regulates short-term adjustment of HR in response to changes in blood pressure or respiratory status; the sympathetic division controls changes of HR in response to physical or mental stress (Ziemssen et al., 2002). During routine daily life, proper cardiovascular functioning is achieved via vagal inhibition over sympathoexcitatory inputs allowing flexible adaptation to changing environmental conditions (Thayer & Brosschot, 2005).
Lowered resting or challenge-associated HRV indicates a disruption in autonomic balance towards an increased sympathetic and a decreased parasympathetic tone, and is associated with a reduced adaptation to internal and external stress (Appelhans & Luecken, 2006; Thayer & Brosschot, 2005). Constant sympathetic dominance and decreased vagal function claim enormous energy demands on the organism (Thayer & Brosschot, 2005) and may result in serious cardiac complications or in sudden cardiac death (Task Force, 1996; Thayer & Lane, 2007; Tsuji et al., 1996). Patients diagnosed with PD have a twofold increased risk for coronary artery disease (Gomez-Caminero, Blumentals, Russo, Brown, & Castilla-Puentes, 2005), and a malfunctioning sympathovagal balance, as indexed for instance in diminished HRV, could be one essential pathogenetic linkage mediating the association between PD and an increased risk for cardiac morbidity and mortality (Fleet, Lavoie, & Beitman, 2000; Thayer & Brosschot, 2005; Thayer & Lane, 2007).
The neurovisceral integration of emotional processing with sympathovagal balance is anatomically and functionally controlled through the central autonomic network (CAN; Thayer & Brosschot, 2005; Thayer & Lane, 2009). Certain regions of the cerebral cortex participate in this network, notably the insular cortex, medial prefrontal cortex and orbitofrontal cortex (Thayer & Brosschot, 2005; Thayer & Lane, 2009). Cardiovascular activity as one major autonomic function is controlled inside this neural network as well (Thayer & Lane, 2009; Verberne & Owens, 1998). The mentioned cortical areas are mostly bidirectionally connected with sympathetic and parasympathetic subcortical and medullary structures which are associated with autonomic functioning, e.g. the central nucleus of the amygdale, the periaqueductal grey matter, the parabrachial nucleus and the nucleus of the solitary tract (Thayer & Brosschot, 2005; Thayer & Lane, 2009). Both cardiovascular control and negative affect are mainly associated with right hemispheric activation within the CAN (Ahern et al., 2001; Davidson, 2000). Particularly the afferent branches of the CAN bear certain resemblance with areas mediating interoceptive processing, including vagal afferents to the nucleus of the solitary tract which further ascend to the right anterior insula (Craig, 2003). Thus the CAN maintains proper emotional processing by integrating interoceptive signals regarding physical conditions of the body and by adjusting autonomic functioning in accordance with changing environmental demands (Appelhans & Luecken, 2006; Thayer & Brosschot, 2005). Efferent sympathetic and parasympathetic output fibres constitute the CAN’s principal output (Thayer & Brosschot, 2005). Since these sympathetic and parasympathetic efferent branches dually innervate the sinoatrial node of the heart, output and efficiency of the CAN is directly mapped top-down onto HRV indices which thus serve as a suitable proxy for the CAN’s adaptive functioning (Appelhans & Luecken, 2006; Thayer & Brosschot, 2005). Hence measures of reduced HRV indicate the organism’s failure to dynamically adapt to changing environmental and emotional stressors (Appelhans & Luecken, 2006; Thayer & Brosschot, 2005) which manifests itself for instance in panic attacks, worry (Friedman, 2007) and in perseverative thinking (Thayer & Lane, 2002). Thus measures of reduced HRV may serve as a psychophysiological marker of the organism’s maladaptive emotional engagement, as behaviourally expressed in the failure to inhibit anxiety in a safe environment (Appelhans & Luecken, 2006; Friedman & Thayer, 1998).
Relevance of heart rate variability assessment to pathological anxiety
Besides its primary medical use in cardiovascular autonomic function measurement and risk assessment after cardiovascular emergencies (Task Force, 1996), HRV analysis is an increasingly applied tool in clinical psychology. Changes in HRV are discussed in terms of an endophenotype concept that characterises pathophysiological conditions of common anxiety disorders, e.g. autonomic involvement in clinical anxiety (Friedman & Thayer, 1998; Thayer & Lane, 2009). Altered HRV was observed in patients with Posttraumatic Stress Disorder (Cohen et al., 2000), Generalized Anxiety Disorder (Pittig, Arch, Lam, & Craske, 2013; Thayer, Friedman, & Borkovec, 1996), Social Anxiety Disorder (Pittig et al., 2013) and in a broad range of studies in patients with PD (Friedman, 2007; Friedman & Thayer, 1998; Pittig et al., 2013). It was shown consistently that patients diagnosed with PD demonstrate a depressed HRV at rest (Klein, Cnaani, Harel, Braun, & Ben-Haim, 1995; Martinez, Garakani, Kaufmann, Aaronson, & Gorman, 2010; Yeregani et al., 1993) as well as under panic-related manipulations triggering panic-like symptoms, such as hyperventilation paradigms or orthostatic maneuvers (Friedman & Thayer, 1998; Martinez et al., 2010). Since most of the reviewed findings result from cross-sectional designs, it is uncertain whether vagal withdrawal precedes clinical anxiety, or whether chronic anxiety results in vagal withdrawal. To our knowledge, no prospective data exists on this controversy. However, there are some findings suggesting diminished vagal cardiac function to be a shared aetiological factor across a broader range of anxiety disorders, strongly in PD and to a smaller extent in Generalized Anxiety Disorder and Social Anxiety Disorder (Pittig et al., 2013), as well as in non-clinical ‘panickers’ who experienced panic attacks but who failed to meet diagnostic criteria for PD (Friedman et al., 1993), and in non-clinical manifestations of anxiety such as high trait anxiety (Fuller, 1992) or behavioural inhibition (Snidman, 1989). Last but not least, decreased vagal activity predicted future anxiety levels in adolescent girls (Greaves-Lord et al., 2010). In conclusion, scarce cross-sectional evidence exists that reduced HRV may precede clinical anxiety levels remaining this relation suggestive rather than definite. However, sporadic reports exist that are not in accordance with vagal hypoactivity in patients diagnosed with PD in response to autonomic challenge (Stein & Asmundson, 1994). After all, these reports are singular and confront good convergence of panic-centric HRV research that showed a consistent association between reduced HRV measures and PD (Friedman & Thayer, 1998). To understand autonomic involvement in clinical anxiety, it is necessary to investigate autonomic activity not only in clinical but as well in subclinical samples with no history of mental or cardiovascular disorder.
Some authors have considered AS being a subclinical dispositional yet malleable risk factor for anxiety, mood and alcohol use disorder, but especially for panic attacks and PD (Calkins et al., 2009; Donnell & McNally, 1990; Maller & Reiss, 1992; McNally, 2002; Schmidt, Lerew, & Jackson, 1999; Schmidt, Zvolensky, & Maner, 2006). AS is defined as the global tendency to misattribute benign physical sensations to a medical emergency having immediate adverse health, mental or social consequences (Reiss, Peterson, Gursky, & McNally, 1986). Since there is a paucity of longitudinal studies addressing the question of whether AS predicts clinical anxiety, a causal relation remains hypothetical. Nevertheless, individuals high in AS seem to be an eligible subclinical sample to investigate, since they share the heightened level of fear of interoceptive sensations that is characteristic for patients diagnosed with PD (McNally, 2002). Due to this concern, subjects with high AS are engaged in attending to internal sensations that indicate anxiety and in monitoring these sensations (Reiss et al., 1986). Medium to large effect sizes were reported both for subjects high in AS (d = 0.63) and subjects diagnosed with PD (d = 0.61) with regard to accurate cardiac interoceptive sensitivity (Domschke et al., 2010). Thus, they may serve as a model for patients with PD in order to comprehend the aetiological processes involved in the ‘incubation’ of clinical anxiety levels (Stein, Simmons, Feinstein, & Paulus, 2007). As reviewed earlier, proliferating research demonstrated an association between PD and reduced HRV measures (e. g. Friedman & Thayer, 1998). Further, Lavoie et al. (2003) reported that AS is associated with a significantly reduced HRV and with cardiac complications in patients diagnosed with coronary artery disease. It has not been explored yet to what extent AS is associated with changes in HRV in mentally and physically healthy subjects with no current or recent psychiatric and cardiovascular disease.
A focus on interoceptive exposure
Since HRV is continuously adapted from beat to beat (Task Force, 1996), there should be a relevant change in HRV observable succeeding certain forms of intervention, possibly beta-blocking medication (Cook et al, 1991; Lampert, Ickovics, Viscoli, Horwitz, & Lee, 2003) or psychotherapy (Garakani et al., 2009). Investigations focussing on the malleability of HRV through psychotherapy are scarce. Garakani et al. (2009) reported an increase in HRV at rest and with paced breathing succeeding 12 sessions of cognitive-behavioural therapy (CBT) in a sample of patients diagnosed with PD. CBT was implemented as an intervention bundle including psychoeducation, breathing training, progressive muscle relaxation, cognitive restructuring, interoceptive and situational in vivo exposure practice. In addition to the improvement of quality of life and well-being, cognitive-behavioural interventions appeared to have beneficial psychophysiological effects. They may execute cardio-protective effects and lower panic patients’ risk for suffering cardiovascular complications by adjusting HRV. The active components of such intervention bundles have not been explored separately yet.
One crucial component of panic-focused psychotherapy is IE practice for triggering panic-related bodily sensations and tolerating these sensations (Craske & Barlow, 2007). Cardiac complaints, notably chest pain, palpitations or subjectively perceived extra-systoles and tachycardia, are among the most frequently reported cardiac symptoms by patients diagnosed with PD (Barsky, Cleary, Sarnie, & Ruskin, 1994; Fleet et al., 1996), and are usually among the initial sensations triggering the symptom cascade of a full-symptom panic attack (American Psychiatric Association, 2000). Besides, cardiac sensations were reported as one symptom of a typical panic attack by 93.2% of patients interviewed by Andor, Gloeckner- Rist, Gerlach, and Rist (2008). With regard to Clark’s cognitive theory of PD (Clark, 1986), natural autonomic activity is misinterpreted as signal of severe health damage leading to increased fear which further escalates in more intense somatic sensations, creating a reciprocating circle. Cardiac symptoms are of special concern since they involve our pivotal organ and mimic sensations of life-threatening cardiovascular diseases, thus causing subjectively perceived vital threat in subjects who believe that anxiety can cause imminent heart failure or other health damage (Siepmann & Kirch, 2010). Accordingly, IE practice may provide a proper setting to disconfirm these maladaptive beliefs (Salkovskis, Hackman, Wells, Gelder, & Clark, 2007). By engaging in symptom provocation such as holding breath or spinning, subjects learn to tolerate benign somatic sensations (Craske & Barlow, 2007). Considering that habituation-based exposure involves several hours of in vivo confrontation (Salkovskis et al., 2007) and since the implemented number of exposure sessions is fixed in this study, we suppose that our short-time intervention supports primarily belief disconfirmation. Since interoceptive processing and cardiac-autonomic control share common anatomical interfaces as noted earlier (Craig, 2003; Thayer & Brosschot, 2005; Verberne & Owens, 1998), we suppose that cardiac autonomic functioning may be adjusted as well in the course of IE.
Summing up, subjects high in AS are characterised by elevated interoceptive sensitivity and are susceptible to interpret somatic stimuli as health-threatening. We postulate that IE is capable of adjusting HRV by triggering interoceptive processing through exposing subjects to corrected information about bodily sensations and their consequences, and thereby recruiting brain areas that control both interoceptive and cardiac-autonomic processing. The present paper focusses on the neural correlates of interoceptive processing and their association to cardiac control. To our knowledge, this is the first paper investigating HRV and its malleability by means of psychological intervention in a subclinical sample. Primary, this examination tests the hypothesis that subjects being predispositioned with high AS exhibit lower HRV indices compared to subjects low in AS. Secondarily, we address the issue of whether IE is capable of changing HRV towards standard value. We are further interested in answering the question which alterable psychological and which physiological characteristics including neural correlates of interoceptive processing (perception and anticipation) are associated with change in HRV.
Methods
Procedure
Recruitment and preselection of interested volunteers were accomplished via an established online screening tool designed by our study group. To be eligible, subjects needed to be aged between 18 and 40 years, right-handed, fluent in German and high or low in AS at online screening. ‘Low AS’ was defined by AS level ≤ 11, and ‘high AS’ was defined by AS level ≥ 29 corresponding to the Anxiety Sensitivity Index (ASI; Hoyer & Margraf, 2003). Exclusion criteria included recent intake of psychotropic or cardiac active medication, regular smoking, acute or chronic neurological, pulmonary or cardiac disease, history of mental disorder (except for specific phobia) according to criteria of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR; American Psychiatric Association, 2000), lifetime experience of a panic attack and failure to meet magnetic resonance (MR) compatibility. Eligible subjects were phone contacted and screened regarding the defined inclusion and exclusion criteria (see Appendix A for details). There were n = 10 volunteers from the Technische Universität Dresden, aged between 19 and 23 years, attending this study; n = 5 of them were men, n = 5 were women. Investigations took place on three separate days. Briefing on investigation procedures was completed at a first appointment. After the participants gave their written informed consent, clinical assessments started. At pre-training appointment, we investigated participants in a MR-simulator environment (mock scanner) and with magnetic resonance imaging (MRI) before introducing the IE training. The electrocardiogram (ECG) was recorded in the mock scanner. Participants were asked to train IE three times per day during the pre- to post-training interval of three to seven days. At post-training appointment, the re-investigation in the MRI and mock scanner took place. Subjects received 50€ for participation. The study procedure was approved by the local ethics committee.
Measures
Clinical assessments.
Anxiety Sensitivity Index. The Anxiety Sensitivity Index (ASI; Hoyer & Margraf, 2003) is a self-estimation instrument assessing the respondent’s anticipatory fear of anxiety- related symptoms and his or her sensitivity towards these symptoms. AS refers to the belief that anxiety symptoms have negative somatic, mental or social consequences. The respondent is asked to quote his affirmation with 16 items on a 5-point scale with response anchors ranging from 0 (very little) to 4 (very much). The ASI holds adequate test-retest reliability (r = .75) over a period of two weeks (Reiss et al., 1986, Study 1), and an internal consistency reliability of Cronbach’s α = .88 (Peterson & Heilbronner, 1987). With regard to criterion validity, agoraphobics show higher AS levels than patients with other anxiety disorders which have higher AS levels than healthy college students (Reiss et al., 1986, Study 2). Besides, AS demonstrates sufficient construct validity being distinct from trait anxiety (McNally, 2002; McNally & Lorenz, 1987).
B ec k-Depression-Inventory II. Severity of depressive symptoms was evaluated using the Beck-Depression-Inventory II (BDI II; Hautzinger, Keller, & Kuehner, 2006). This self- report rating inventory is composed of 21 groups of items matching DSM-IV major depression criteria. Each item group is a list of four statements arranged in increasing intensity. The respondent is asked to choose the most appropriate alternative that best describes the way he or she felt during the past two weeks including the present day. The inventory is internally consistent (Cronbach’s α = .89) and concurrent to other self-report tools for depression (r = .72 to r = .89; Kuehner, Buerger, Keller, & Hautzinger, 2007). C ardiac Anxiety Questionnaire. The Cardiac Anxiety Questionnaire (CAQ; Eifert et al., 2000), translated into German by Hoyer & Eifert (2004), is a self-report survey measuring one’s attention to and fear of aversive cardiac sensations as well as cardio-phobic and cardio- protective behaviour. The respondent indicates for 17 items on a 5-point Likert scale anchored from 0 (never) to 4 (always) his or her agreement with the described behaviour. The German version of the CAQ is internally consistent as measured with Cronbach’s α = .83 for the total score (Hoyer et al., 2008). Correlation coefficients demonstrate appropriate convergent validity between CAQ total score and concurrent anxiety measures like the Anxiety Sensitivity Index (r = .69) or the Body Sensations Questionnaire (r = .66; Eifert et al., 2000).
Cl austrophobia Questionnaire. The Claustrophobia Questionnaire (CLQ; Radomsky, Rachman, Thordarson, McIsaac, & Teachman, 2001; unpublished German translation; Alpers, 2002) is a self-assessment tool developed to assess a person’s fear of suffocation and restriction. The respondent is asked to estimate accordance with 26 items on a 5-point scale with response anchors ranging from 0 (not at all anxious) to 4 (ex tremely anxious). The CLQ total score demonstrates adequate predictive power of subjective fear (r = .52), somatic sensations (r = .64) and anxiety cognitions (r = .64) during exposure to a small enclosed space, and high internal consistency (Cronbach’s α = .95; Radomsky et al., 2001).
C o m posite International Diagnostic Interview. The Composite International Diagnostic Interview (DIA-X/CIDI; Wittchen & Pfister, 1997) is an established standardised interview technique providing a basis for DSM-IV diagnosis of mental disorders on axis I. It comprises screening questions and a computer-based interview guideline for the enquiry of ICD-10- and DSM-IV -compatible diagnostic classifications. Interrater-reliability measures range from κ = .49 to κ = .83 (Wittchen & Hoyer, 2011). Regarding convergent validity, correlation coefficients between CIDI diagnoses and psychiatric diagnoses by a clinician vary from κ = .39 respecting psychotic disorders to κ = .82 respecting panic disorder (Wittchen & Hoyer, 2011).
Initial physiological conditions questionnaire. This self-created questionnaire is a self-report survey asking for actual well-being, attention, sleep quality, nicotine-, alcohol-, drug- and medication-intake the previous day and current date of appointment. It is handed out on arrival to the ensure participants’ comparability regarding their physical conditions (see Appendix A for details).
Multidimensional Mood Questionnaire. The Multidimensional Mood Questionnaire (MDMQ; Mehrdimensionaler Befindlichkeitsfragebogen: MDBF, Steyer, Schwenkmezger, Notz, & Eid, 2002) is a self-report questionnaire which assesses actual well-being in three bipolar dimensions (valence: bad to good mood, alertness: tired to awake, calmness: restless to calm) with 12 items to be answered on a 5-point scale ranging from 1 (not at all) to 5 (ve ry much). All scales can be divided into two parallel test halves that can be applied to get a mood-profile in time and that we alternated during our measurements. The MDMQ meets high internal consistency with Cronbach’s α = .86 up to α = .96 (Steyer et al., 2002). With regard to criterion validity, MDMQ dimensions demonstrate sufficiently high correlation coefficients with measures of daily hassles and uplifts (Steyer et al., 2002).
Physical Fitness Questionnaire. Basic motor function status including cardiorespiratory fitness, strength, flexibility and coordination was assessed using the Physical Fitness Questionnaire (Fragebogen zur Erfassung des motorischen Funktionsstatus: FFB-Mot, Boes et al., 2002). The standard version of this self-report inventory is composed of 20 items and was utilised in this study. An extended version is available measuring low physical fitness (activities of daily living) or high fitness (sport scale). The agreement with all items is self-evaluated on a 5-point scale anchored from (no difficulties in performing this motor task) to (not able to perform this motor task). The FFB-Mot total score features a high level of internal consistency (Cronbach’s α = .87 to α = .92; Boes et al., 2002). The convergent validity measures between the total score and physical fitness tests are moderately high with r = .58 for men, and r = .56 for women (Boes et al., 2002).
E CG Recording. Two channels of ECG were acquired with 5000 Hz sampling rate in supine position from two electrodes attached to the back in a mock scanner environment by using Brain Vision HR recording device (Brain Vision ExG Amplifier, Brain Vision Recorder, Munich). The mock scanner environment is a re-creation of a MRI scanner and is almost identical to a real MRI environment except for the magnetic field. It includes a ‘head coil’ and headphones allowing the presentation of echo planar imaging (EPI) sequence noise profile. Subjects were placed headfirst in the mock scanner tube like they were placed in the real MR scanner tube.
Subject’s mental and emotional actions during resting MR scanning might feature a notable inter-individual variability and seemed barely controllable through simple resting instructions (Brennan et al., 1988; Muehlhan, Lueken, Wittchen, & Kirschbaum, 2011). Nearly 16% of student participants experienced anxiety and 8.5% of them showed cardiac reactions during supine rest in a mock scanner apparatus (McGlynn, Karg, & Lawyer, 2003). Hence, to avoid confounding sympathetic anxiety-related changes in HR, the experimental setting was standardised in the following way: ECG recording was performed twice at rest and once under metronome-controlled breathing, while each recording period lasted six minutes. Resting ECG was registered first without any stimulation and next while presenting an EPI noise-profile to create a more authentic MRI experience. This was to include an additional stressor and to test if state influence would alter HRV. At both resting periods, we instructed participants to lay feasibly fixed but relaxed, to move and speak as little as possible and to breathe spontaneously. Participants were not attended to any specific task, but were asked to stay awake. To avoid interference with spontaneous breathing, we instructed metronome-controlled breathing at last, with six deep breathes per minute (5 s of inspiration and 5 s of expiration) as recommended by Ziemssen et al. (2002). A rising and falling bar, apparent via a mirror projection system on a monitor behind the mock scanner tube, indicated the onset of inspiration and expiration (Atempace V1.2, Zentrum für Neuropsychologische Forschung, Universität Trier). Respiration recording was not possible during mock scanner assessment, but since metronome-controlled breathing accentuates vagal activity as indexed in the high frequency (HF) respiratory component (Berntson et al., 1997; Siepmann et al., 2005), this was used as manipulation check for matching the metronome. Further, to control for potential stress or anxiety in reaction to mock scanner investigation, we asked participants to indicate their well-being with the help of the MDMQ immediately before ECG recording started, after both resting conditions and after metronome-controlled breathing. The effect of the EPI noise-profile as an additional stressor was inferred from comparing HRV indices between both resting conditions within AS group. The following rule is defined: If both resting HRV indices do not differ significantly from each other, further analyses will be conducted with the first ‘true resting’ HRV index in addition to HRV index under metronome-controlled breathing. If both resting HRV indices differ significantly, we will conduct further analyses with these two indices additionally to HRV index under metronome- controlled breathing. The latter index is conducted anyway since we expect HRV assessment to be most accurate under standardised metronome-controlled breathing within HF power band thus allowing for correct vagal attribution (Pagani et al., 1986; Pinna, Maestri, La Rovere, Gobbi, & Fanfulla, 2006).
Functional MRI data acquisition. We acquired whole brain imaging data on a 3- Tesla Trio-Tim MRI whole-body scanner (Siemens, Erlangen, Germany). Functional images were obtained using a T2* weighted gradient EPI sequence allowing excellent time resolution (repetition time TR 2500 ms, echo time TE 25 ms, interleaved acquired 41 slices 3 x 3 x 3 mm, 3.0 mm slice thickness, field of view FOV 192 mm, flip angle 80°). Anatomical images were completed with a high-resolution Magnetization Prepared Rapid Gradient Echo Imaging (MPRAGE) sequence (repetition time TR 1900 ms, echo time TE 2.26 ms, sagitally sequentially ascending slices 1 x 1 x 1 mm, field of view FOV 256 mm, flip angle 9°). Geometric distortions in EPI originating from magnetic field inhomogeneity were corrected including field mapping in the realign and unwarp options (repetition time TR 488 ms, echo time TE 1 4.92 ms and TE 2 7.38, transversally acquired 41 slices 3 x 3 x 3 mm, 3.0 mm slice thickness, field of view FOV 192 mm, flip angle 60°). Standard presentation devices and MR- compatible peripheral equipment (head coils, MR-Confon headphones and Lumitouch keypad) were used. The paradigm was presented on a monitor arranged at the head end of the MR tube and was visible to participants via a mirror projection system.
Experimental procedures
IE training. The IE practices corresponded to those described by Lang, Lang-Helbig, Westphal, Gloster, and Wittchen (2011) in their manual-based psychotherapy for patients diagnosed with PD, but was implemented as a study-related intervention. All nine exercises were introduced at pre-training by a study assistant who was paying attention to accurate performance. Training involved e. g. shaking one’s head rapidly from side to side for 30 s, running on the spot for one minute, spinning around on the spot for one minute or breathing through a straw for one minute. This was to provoke somatic symptoms that potentially cause anxiety and distress. We asked participants to indicate symptom intensity, anxiety and discomfort produced by each exercise on a 10-point scale with response anchors from 0 (not at all) to 10 (extremely). The three exercises triggering the most discomfort and anxiety were highlighted and repeated self-governed by all participants thrice per day between pre- and post-training. We enhanced compliance via the following motivational strategies: We highlighted importance of regular and accurate practice, and appealed to subject’s honesty in performing the exercises. Participants were instructed to minute all exercises in a prepared online-based schedule. Adherence to the schedule was checked daily for missing practices; be it that someone missed one practice, subjects were reminded of daily exercises by email. Besides, subjects received one cinema coupon after complete and regular performance. The exercises were repeated together with a study assistant at post-training appointment and were followed by a resting phase of 30 min before MR scanning started.
Cardiac symptom provocation task. The newly designed paradigm corresponded to interoceptive paradigms published by Paulus and associates (e.g. Lovero, Simmons, Aron, & Paulus, 2009) and was used to analyse neural activation patterns that are associated with the anticipation and perception of interoceptive cues. The paradigm combined a continuous performance task and cued confrontation with interoceptive stimuli (interoceptive cue condition). The continuous performance task served as baseline while participants were assigned to indicate the horizontal direction of black arrows. The interoceptive cue condition involved two phases. During the anticipation phase, the colour of the arrows changed signalising the imminent stimulation phase. During the stimulation phase, participants listened either to computer-generated heartbeats (interoceptive cue condition) or sinus tones (exteroceptive control condition) of two different frequencies (50 vs. 100 beats per minute [bpm]). Participants were instructed to imagine hearing their own heartbeats. The subjects’ task was to count all perceived heartbeats or tones and, in the quarter of all trials, to enter the counted signals via the keypad. Thrice during MRI scanning, we asked participants to estimate their well-being via MDMQ. Stimulus ratings were performed prior to and after MRI scanning in the mock scanner. The paradigm comprised two runs which involved the three task conditions (see Figure A1). The baseline phase started with the presentation of a fixation cross displayed for 30s and a familiarisation phase comprising four runs of the continuous performance task participants were engaged in. The duration of the interstimulus interval varied from 3,000 to 9,000 ms. Baseline procedure was repeated 16 times per run. Presentation of the coloured arrow during the anticipation phase lasted from 5,000 to 10,000 ms and was repeated 16 times. The continuous performance task was performed analogue to baseline condition. In 37.5% of all trials, stimulation followed; in 62.5% of all trials, anticipation was followed by a fixation cross displayed for 3,000 ms. In half of the trials heartbeats were presented, and in the other half sinus tones. Stimulation appeared six times per run.
Data preprocessing and analysis
ECG data. ECG data were preprocessed using Brain Vision Analyzer Software 2.0.2 (Brain Products, Munich) for filtering and segmentation, and using Kubios HRV 2.1 (Biosignal Analysis and Medical Imaging Group, Kuopio, Finland) for HR determination as well as for time and frequency analysis. To ensure adjustment of the subject’s HR to the task, all data sets were devoid of the first minute of recording. High-pass (70 Hz) and low-pass filtering (0.53 Hz) allowed removing low frequency components due to motion and nonsinoatrial node originating cardiac beats. As recommended, prior to spectral analysis, the R-R interval time series were transformed into an equidistantly time series using cubic spline interpolation with a sampling rate of 4 Hz, and R-R intervals deviating 0.25 s from local average were corrected (Berntson et al., 1997). For power spectrum estimation, autoregressive modelling with model order set at 16 was used for extracting ‘respiratory’ HF power band (0.15 - 0.4 Hz). RMSSD time domain measure, defined as the square root of successive R-R- interval differences (Baron & Wasner, 2005), was chosen for HRV index computation since it has proven as a robust index reflecting high-frequency variations in HR following parasympathetic control of HRV (Siepmann et al., 2005; Task Force, 1996). The mentioned cardiovascular parameters were calculated thrice per subject: at rest without any stimulation, at rest while presenting EPI noise-profile, and under metronome-controlled breathing. A logarithmic transformation was applied to compare RMSSD indices to standard values reported by Agelink et al. (2001). Change in RMSSD (∆logRMSSD) was defined as the difference between pre-training logarithmised RMSSD (logRMSSD) and post-training logRMSSD.
Functional MRI data. Functional and structural MRI data were analysed using Statistical Parametric Mapping Software, version 8 (SPM 8; Welcome Trust Centre for Neuroimaging, UCL, London, UK), implemented in MATLAB 7.1 (Mathworks Inc., Sherborn, MA, USA). Prior to statistical analyses, functional MRI (fMRI) data quality was assessed by inspection for visible head motions. Images were realigned and unwarped for head motion correction; divergence from the defined quality parameters of less than 1° rotation and less than 1 voxel translation in time were accepted. Functional and anatomical images were coregistered and normalised into the stereotactic space of Talairach and Tournoux (1988) to the 305 MNI (Montreal Neurological Institute, Quebec, Canada) standard template. Finally, a weighted average algorithm for smoothing with 8 mm full-with half- maximum kernel was applied. First-level model specification and estimation were conducted on the basis of the general linear model. The expected blood oxygen level-dependent (BOLD) signal change was modelled by a canonical hemodynamic response function. A total of 13 regressors per run were entered into the fMRI design model, including coloured arrow, baseline black arrow, fixation cross following anticipation without stimulation, heartbeats at 50 bpm, sinus tones at 50 bpm, heartbeats at 100 bpm, sinus tones at 100 bpm, as well as six motion parameters estimating translation, rotation and dispersion; the number of counted signals was entered only for the first run. T-contrasts were defined for each subject as prerequisite for second-level one-sample t-test and two-sample t-test comparing AS-related activation patterns. For a hypotheses-driven analysis, a region-of-interest(ROI)-analysis was performed using WFU Pickatlas (Maldjian, Laurienti, Kraft, & Burdette, 2003) and the implemented automated anatomical labelling atlas (Tzourio-Mazoyer et al., 2002). ROIs were anatomically defined according to those brain structures of the aforesaid neurovisceral integration network that correspond to brain areas being associated with the perception of interoceptive signals: right anterior insular cortex (AIC; BA 13), right anterior cingulate cortex (ACC; BA 24), right orbitofrontal cortex (OFC; BA 11, 12, 47) and medial prefrontal cortex (mPFC; BA 9, 10; Craig, 2009; Paulus & Stein, 2006; Schuenke, Schulte, & Schumacher, 2006; Thayer & Lane, 2009). BOLD signal changes were considered statistically significant if they equalled or exceeded p < .001 (uncorrected) with a minimum cluster size of 10 voxels. Beta values were extracted for each subject from significant activated clusters as revealed by ROI-analysis during the anticipation of interoceptive stimuli and from contrasting interoceptive and exteroceptive perception (HEART>TONE). Beta values represent the mean activity over voxels within each cluster and were extracted using REX toolbox for SPM 8 (Whitfield-Gabrieli, 2009).
S t atistical analyses. Nonparametric testing was applied due to small sample size and non-normally distribution of most variables. Testing for normality was performed using Kolmogorov-Smirnov test. We performed Mann Whitney U -test to evaluate group differences in demographic, psychological and cardiovascular characteristics. Statistically significant differences were defined as p < .05. Changes in HRV over time as well as differences in HRV between the two resting conditions were tested with Wilcoxon signed-rank test and Bonferroni-corrected for multiple testing (adjusted α = .025). Associations between changes in HRV and psychological and physiological characteristics including neural correlates of interoceptive processing in the entire sample were tested using Spearman’s rank correlation. In cross-sectional research, decreased HRV indices have been associated with anxiety-related traits (Fuller, 1992), depressive symptoms (Siepmann et al., 2005), poor physical fitness (Buchheit et al., 2005; Goldsmith, Bigger, Bloomfield, & Stein, 1997), high BMI and accelerated HR (Kuch et al., 2001). We wanted to see if reported associations apply as well to HRV change score and entered BMI, FFB-Mot, HR and BDI II sum scores in correlation analysis. ASI, CAQ and CLQ total scores as further anxiety-related traits and logRMSSD initial values were added. Change in HRV (∆logRMSSD) was defined as the difference between pre-training logRMSSD and post-training logRMSSD. Further, the beta values of activated clusters in defined ROIs during the anticipation and perception of interoceptive stimuli were entered in our analyses. Last but not least, since we expect IE to activate brain areas associated with interoceptive processing and cardiac control, we wanted to see if the extent of performance quality correlates with change in HRV. Performance quality in IE was defined as the difference of mean discomfort ratings felt during the three homework exercises from pre- to post-training. All calculations were conducted using SPSS 21 (Chicago, IL, USA).
Results
As outlined above, we were interested in answering the questions of (a) whether subjects high in AS feature lower HRV indices in terms of logRMSSD as compared to subjects low in AS, (b) whether HRV can be changed through IE in both AS groups and (c) which psychological and physiological characteristics including neural correlates of interoceptive processing (perception and anticipation) are associated with a change in HRV.
Sample characteristics
As shown in Table 1, both groups were comparable with respect to demographic and psychological characteristics. Subjects high in AS (n = 5) equalled subjects low in AS (n = 5) with regard to gender ratio and age. Concerning clinical measures, subjects high in AS had significantly higher CAQ scores than subjects low in AS, indicating a higher level of fear of cardiac sensations, cardio-phobic behaviour and cardio-protective care respectively. As intended, participants high in AS scored significantly higher than participants low in AS in the ASI which was used for study inclusion and group assignment. No significant differences emerged between the subgroups concerning depressive symptoms as expressed in BDI II sum scores, with all participants scoring below clinical criteria for mild depressive disorder. FFB- Mot sum scores were below standard values reported by Boes et al. (2002) suggesting that physical fitness was comparatively low in the entire sample.1
Table 1 Demographic and psychological characteristics (pre-training) of the study sample. Medians (I QR) are listed
Abbildung in dieser Leseprobe nicht enthalten
Note. Figures placed in brackets indicate possible range of total scores; IQR = interquartile range; HAS = subjects high in anxiety sensitivity; LAS = subjects low in anxiety sensitivity; BMI = body mass index; ASI = Anxiety Sensitivity Index; BDI II = Beck-Depression-Inventory II; CAQ = Cardiac Anxiety Questionnaire; CLQ = Claustrophobia Questionnaire; FFB-Mot = German version of the Physical Fitness Questionnaire. a χ² test. bASI sum score obtained at online screening. cBDI II, CLQ and FFB-Mot sum scores obtained at date of informed consent. dCAQ total score obtained at pre-training. * p < .05.
Group comparison of heart rate variability
Table 2 summarises cardiovascular parameters as a function of AS group at both sessions. Medians of logRMSSD indices are illustrated in Figure 1. Such as with regard to psychological assessments, the two groups did not differ in mood state in the course of pre- training ECG recording, as measured with the MDMQ (see Figure C1 and Table B1 for details). At post-training, low-AS participants scored significantly lower in the MDMQ dimensions alertness and calmness while resting without any stimulation (see Figure C2 and
Table B1 for details).
Table 2 C omparison of cardiovascular parameters between AS groups. Medians (IQR) are listed
Abbildung in dieser Leseprobe nicht enthalten
Note. AS = anxiety sensitivity; IQR = interquartile range; HAS = subjects high in anxiety sensitivity; LAS = subjects low in anxiety sensitivity; Resting = resting without any stimulation; HR = heart rate measured in beats per minute (bpm); pre = pre-training; post = post-training; logRMSSD = logarithmic transformation of the square root of successive R-R-interval differences measured in milliseconds; HF = high frequency measured in hertz; Resting with EPI = resting while presenting EPI noise-profile; EPI = echo planar imaging; Metronome = metronome-controlled breathing.
For all cardiovascular parameters, no significant differences between AS groups emerged. At a descriptive level, subjects high in AS demonstrated lower pre-training logRMSSD indices as compared to subjects low in AS, both at rest (Mdn for HAS: 1.61; Mdn for LAS: 1.75) and at rest while listening to EPI noise-profile (Mdn for HAS: 1.65; Mdn for LAS: 1.87) as well as during metronome-controlled breathing (Mdn for HAS: 1.69; Mdn for LAS: 1.80), as predicted. Besides, no difference was observed with regard to pre-training HF total power under metronome-controlled breathing between both AS groups, indicating that participants matched the metronome. Based on the fact that pre-training resting HR and logRMSSD at rest and under metronome-controlled breathing were largely in line with established standard values reported by Agelink et al. (2001), we expect participants to be owners of an intact heart. 2
[...]
1 Standard values obtained on adults aged between 33 and 40 years. Means (standard deviation) of the FFB-Mot standard version are provided. Men: 88.5 (8.1), women: 79.9 (8.3). Medians of the present sample: Men (HAS: 40.00, LAS: 34.00), women (HAS: 37.00, LAS: 35.00).
2 Standard values obtained on adults aged between 17 and 25 years. Means (standard deviation) are provided. HR (bpm) at rest: men 68.9 (10.5), women 76.7 (13.5); logRMSSD (ms) at rest: men 1.65 (0.32), women 1.58 (0.32); logRMSSD (ms) under metronome-controlled breathing: men 1.76 (0.24); women: 1.65 (0.25). Note, that Agelink et al. (2001) performed metronome-controlled breathing with 6 s inspiration and 4 s expiration, while we instructed 5 s inspiration and 5 s expiration. See Table B2 for details.
-
-
-
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X.