Article

Biological and perceived stress in motor functional neurological disorders

APAZOGLOU DUCLOUS, Kalliopi, et al.

Abstract

Current models explaining motor functional neurological disorders (FND) integrate both the neurobiological mechanisms underlying symptoms production and the role of psychosocial stressors. Imaging studies have suggested abnormal motor control linked to impaired emotional and stress regulation. However, little is known on the biological stress regulation in FND. Our aim was to study the biological and perceived response to stress in patients with motor FND.

Reference

APAZOGLOU DUCLOUS, Kalliopi, et al. Biological and perceived stress in motor functional neurological disorders. Psychoneuroendocrinology, 2017, vol. 85, p. 142-150

DOI : 10.1016/j.psyneuen.2017.08.023 PMID : 28863348

Available at: http://archive-ouverte.unige.ch/unige:112276

Disclaimer: layout of this document may differ from the published version.

1 / 1 Psychoneuroendocrinology 85 (2017) 142–150

Contents lists available at ScienceDirect

Psychoneuroendocrinology

journal homepage: www.elsevier.com/locate/psyneuen

Biological and perceived stress in motor functional neurological disorders T Kalliopi Apazogloua, Viridiana Mazzolab, Jennifer Wegrzykc, Giulia Frasca Polarac, ⁎ Selma Aybeka,c,d, a Fundamental Neurosciences Department, University of Geneva, Switzerland b Liaison Psychiatry Service, Psychiatry Department, Geneva University Hospital, Switzerland c Service, Geneva University Hospital, Geneva, Switzerland d Neurology Service, InselSpital Bern, Switzerland

ARTICLE INFO ABSTRACT

Keywords: Background: Current models explaining motor functional neurological disorders (FND) integrate both the neu- Functional robiological mechanisms underlying symptoms production and the role of psychosocial stressors. Imaging stu- dies have suggested abnormal motor control linked to impaired emotional and stress regulation. However, little Cortisol is known on the biological stress regulation in FND. Our aim was to study the biological and perceived response Amylase to stress in patients with motor FND. Stress Methods: Sixteen patients with motor FND (DSM-5 criteria) and fifteen healthy controls underwent the Trier Trier social stress test Social Stress Test. Hypothalamo-pituitary-adrenal axis (HPA) response was evaluated with salivary cortisol and autonomous sympathetic response with salivary alpha-amylase. Area under the curve was computed to reflect background levels (AUCg) and change over time (AUCi). Life adversities and perceived subjective stress on a visual analog scale (VAS) were correlated with biological responses. Results: FND patients had significantly higher background levels (AUCg) of both stress markers (cortisol and amylase) than controls. The biological response (AUCi) to stress did not differ between groups for both markers but the subjective response showed an interaction effect with patients reporting higher levels of stress than controls. After stress, controls showed a strong correlation between subjective and objective sympathetic values (amylase) but not patients. The number and subjective impact of adverse life events correlated with cortisol AUCg in patients only. Conclusion: This study confirms a baseline HPA-axis and sympathetic hyperarousal state in motor FND related to life adversities. During a social stress, dissociation between perceived stress and biological markers was observed in patients only, reflecting a dysregulation of interoception capacity, which might represent an endophenotype of this disorder.

1. Introduction linked to a causal psychological stressor (Stone et al., 2011), as re- flected in the new DSM-5 classification (DSM-5, 2013). The dualism Motor functional neurological disorder (FND) − or conversion that the cause had to be either psychological or physical (a so-called disorder (DSM-5, 2013) − is a disabling medical condition affecting a “organic” neurological condition) has been resolved and FND is now large number of young patients (Carson and Lehn, 2017), often with considered a neuropsychiatric condition (Carson, 2014): it manifests chronic disability (Gelauff et al., 2014) due to neurological symptoms, with neurological symptoms that are linked to (and not caused by) such as gait difficulties, tremor or weakness. FND represents the second psychological risk and/or maintaining factors (Hinson et al., 2006; commonest cause for a neurological consultation after headache (Stone Kroenke, 2007; Reuber et al., 2007; Hubschmid et al., 2015). Indeed, et al., 2010). Over the last century, FND has traditionally been viewed there is evidence from increased rates of childhood trauma (Roelofs as “psychogenic” with reference to a psychological cause in form of et al., 2002; Sar et al., 2004) and life adversities (Roelofs et al., 2005)in “conversion” of an intra-psychic conflict into physical symptoms motor FND, which may play a role as predisposing factors. The me- (Kanaan, 2017). Although psychiatric stressors have long been con- chanism on how these psychological factors may predispose or main- sidered as causal factors, experts have agreed recently that the onset tain a neurological motor symptom is still unclear. Evidence from and maintenance of the physical neurological deficit cannot always be neuroimaging studies suggests that the limbic system dealing with

⁎ Corresponding author at: Leitende Arztin Neuropsychosomatik, Universitätsklinik für Neurologie, Lory Haus, InselSpital, 3010 Bern, Switzerland. E-mail address: [email protected] (S. Aybek). http://dx.doi.org/10.1016/j.psyneuen.2017.08.023 Received 24 April 2017; Received in revised form 21 August 2017; Accepted 21 August 2017 0306-4530/ © 2017 Elsevier Ltd. All rights reserved. K. Apazoglou et al. Psychoneuroendocrinology 85 (2017) 142–150 psychological stimuli (Voon et al., 2010; Aybek et al., 2015) or trauma Table 1 (Aybek et al., 2014) (amygdala and hippocampus) is aberrantly func- Demographical and Clinical Data. tionally connected to regions responsible for the neurological symptom FND patients (N = 16) Healthy P value (T (supplementary motor area and right temporo-parietal region). The Controls test*, Fisher hypothesis is that there is an emotional hyperarousal state in FND with (N = 15) test °) increased amygdalar response to stimuli, even with positive valence (Voon et al., 2010), as well as a lack of habituation to negative stimuli Age (years) 46 ± 15 39 ± 13 *ns Gender 14F:2M 12F:3M °ns (Aybek et al., 2015). Little is known on this hyperarousal state and in (Fenale:Male) particular few studies looked at biological markers of hyperarousal such Cycle:Menopause 5:9 6:6 °ns as stress biological parameters in motor FND. We thus hypothesized Medication 7 none:5BZD, 3AD, 1AE 14none:1AD °p < 0.01 that this hyperarousal state will be reflected in abnormal biological BDI 8 ± 5.5 3.5 ± 4.6 *p < 0.05 STAI Trait 41.1 ± 12.1 37.9 ± 9.0 *ns measures of the stress systems. State 36.9 ± 10.1 33.7 ± 8.8 *ns The stress response is mediated by two main pathways: the rapid CGI 3 none NA autonomous sympathetic response with epinephrine/norepinephrine 2 minimal secretion and the slower hypothalamo-pituitary-adrenal axis (HPA) 3 moderate response with cortisol secretion. Both can be non-invasively in- 8 severe Mobility NeuroQoL 33.1 ± 4.0 39.7 ± 0.6 *p < 0.01 vestigated by measuring salivary amylase (a protein secreted by the Symptoms duration 73 ± 85 (range 4–300) NA parotid glands during sympathetic stimulation) (Rohleder et al., 2004; (months) Nater et al., 2005) and salivary cortisol (reliably reflecting HPA acti- Symptoms type 6 weakness NA vation) (Hellhammer et al., 2009). A recent study looked at morning 5 tremor awakening cortisol levels in 33 patients with motor FND, which were 2 gait 3 jerks found similar to levels in healthy controls(Maurer et al., 2015) but no studies looked at amylase levels in motor FND and no studies specifi- BZD: benzodiazepines, AD: antidepressant, AE: antiepileptic, NA: not applicable, ns: non- cally tested the biological response to stress. A robust way to probe for significant, BDI: Beck Depression Inventory, STAI: Anxiety Score, CGI: Clinical Global stress response is to expose participants to a social stressor, like a job Impression Score interview setting, and monitor the increase of salivary stress parameters (cortisol and amylase) (Birkett, 2011). We chose the Trier Social Stress 2023 to 0 while subtracting 17 each time”). Both examiner 1 leading Test (TSST) as the most reliable and validated protocol to induce such the interview (VM) and examiner 2 (GFP or JW) refrained from pro- stress, paralleled by a robust cortisol and amylase release (Kirschbaum viding emotional or supporting feedback (such as smile or nodding). et al., 1993b). This is important to induce an uncertainty component and reliably in- The aim of our study was to explore 1) the sympathetic and HPA crease cortisol (Abelson et al., 2014). axis biological functions and 2) perceived stress levels in response to a social stress (TSST) and in relation to life stressors (background hy- 2.3. Saliva samples collection perarousal) in patients with motor FND compared to healthy controls. Nine saliva samples were collected (as shown in Figs. 1–3)by 2. Methods & materials chewing a cotton-swab during 1 min (Sarstedt-Salivette®). Samples were immediately centrifuged (10 min at 3000 rpm) and saliva was 2.1. Participants frozen (−20 °C). All experimental sessions took place from 1:30pm to 4:30pm. Participants were instructed to refrain from heavy meals, Sixteen patients with motor FND were recruited from the Neurology coffee, coke or other fizzy soft drinks, chewing gum and any intense Department of Geneva University Hospital. The diagnosis was estab- physical activity in the hour preceding the session. lished according to DSM-5 criteria of Conversion Disorder (Functional Neurological Symptom) code F44.4 and a board-certified neurologist 2.4. Perceived stress response confirmed the presence of positive functional features (DSM criteria B). Details of the clinical presentation are presented in Table 1. In parallel of saliva sampling, participants filled in a self-report Fifteen age and gender-matched healthy controls (HC) were re- evaluation of stress using a visual analogue scale (VAS: 0 = no stress to cruited through announcements. Exclusion criteria for both groups 10 = very high stress). were: self-report of 1) a neurological condition (past or present), 2) a current psychiatric condition such as psychotic disorder, substance 2.5. Mood and Trauma abuse or depression with acute suicidality, 3) insufficient knowledge of French. Patients suffering from comorbidities such as anxiety or de- Mood was assessed with the BDI depression scale (Beck et al., 1961) pression (without suicidality) were included. All participants provided and STAI anxiety state and trait scales (Laux, 1981). Life events that written informed consent (Swiss Ethics approved protocol CER14-008). occurred within the previous five years were recorded with the Amiel- Upon arrival to the laboratory, participants were asked to fill in all Lebigre questionnaire (Amiel-Lebigre, 1985). The sum of negative (e.g. relevant questionnaires. Following this preparation phase (∼1 h), “Suicide in close family”) and change of life situation events (e.g. subjects were left to relax for 20 min (relaxation phase). They were then “Arrival of a new member of the family in your house”) were computed taken to the test room where the investigator instructed them on the when conducting correlations with biological markers. The Amiel-Le- TSST procedure in the presence of the examiners. They were then given bigre questionnaire provides a subjective rating of each life events on a 10 min to prepare their speech (see below) in the relaxation room. 0–100 scale (0 = no impact on my life, 100 = major impact on my life). Childhood trauma was assessed across 5 domains (sexual, physical 2.2. Stress induction and emotional abuses, physical and emotional neglect) with the Childhood trauma questionnaire (CTQ) (Bernstein and Fink, 1998). As Participants were instructed to prepare and present a video and cortisol stress response following a social stressor is known to be audio-recorded job interview speech in front of two unknown ex- dampened in sexually abused subjects (Schalinski et al., 2015), we re- aminers for 5 min according the TSST protocol (Kirschbaum et al., peated the analysis for subgroups of sexually abused (n = 8) or non- 1993a). Then they performed a 5-min calculation task (“count from abused (n = 8) defined by a cut-off > 5 in the sexual abuse subscore.

143 K. Apazoglou et al. Psychoneuroendocrinology 85 (2017) 142–150

Fig. 1. Objective and Perceived Stress. Stress induction (TSST) occurs at time 0 and lasts 20 min. AUCg: area under the curve with respect to ground. AUCi: area under the curve with respect to increase. Panel A: HPA axis Stress as measured with salivary cortisol showing a main effect of time and a main effect of group (higher AUCg in patients) but no interaction. Panel B: Autonomous Stress as measured with salivary alpha-amylase showing a main effect of time and a main effect of group (higher AUCg in patients) but no interaction. Panel C: Perceived Stress as measured with the 0–10 Visual Analog Scale showing a significant main effect of time across group and a time*group interaction with increased perceived stress in patients after the TSST as compared to before (higher AUCi in patients).

2.6. Duration and severity of FND and medication of the cortisol assay was 0.5 nmol/l and inter-assay imprecision at 8.0 nmol/l was 11.5% (coefficient of variation, CV), while intra-assay CV Duration was calculated from onset of main symptom to date of was < 6.0% at the same concentration. inclusion (months). Severity was assessed with the Clinical Global Amylase activity was determined in diluted saliva samples (1:1000) Impression (CGI) score (0 = no neurological impairment, 1 = minimal, on a cobas c-501 analyzer, using a synthetic oligosaccharide as sub- 2 = moderate, 3 = severe, 4 = very severe) and the continuous score of strate and measuring the production of p-nitrophenol with a colori- the “Mobility subscale” of the Neuro-Qol (Quality of Life in metric method (increase of absorbance). The analytical lower limit of Neurological Disorders)(Cella et al., 2012), which evaluates the impact the assay was 3 U/L and inter-assay imprecision 1.1% (CV) at 168 U/L of physical impairment of individuals living with neurological condi- and 1.7% at 52 U/L, while intra-assay CV were < 1.0% for the same tions, rating mobility difficulties in daily life activities with a maximum activities. The analytical procedures were accredited according to in- score of 40 (full mobility). Use of CNS-acting medication such as ben- ternational ISO-15189 norms. zodiazepine, antidepressant, antiepileptic and neuroleptic was re- corded. 2.8. Statistics

2.7. Saliva samples analysis Analyses were performed with R software. For repeated measures analysis, a linear mixed model (lmer) was used, with fixed effects of After thawing, samples were centrifuged again and 100 μl saliva factors group and time and random effects of the factor subject in order aliquots were collected. Cortisol concentration was measured directly to account for individual differences. To correct for potential con- (without extraction) on a cobas e-601 analyzer (Roche) by an electro- founding effects of mood and menstrual cycle, these variables (BDI, chemiluminescent immunoassay (ECLIA). The analytical detection limit STAI-T, cycle) were added as covariates of no interest in the model. For

144 K. Apazoglou et al. Psychoneuroendocrinology 85 (2017) 142–150

Fig. 2. Life events and background Cortisol. Right Panel:Nodifferences were found between groups in terms of number of negative or change events as well as impact (self-rated from 0 no impact to 100 major impact) Left Panel: Significant correlation, in patients only, between number and impact of negative + change events. correlation analyses, a Pearson coefficient was calculated (r) and p biomarkers (main effect of time F(8,232) = 8.01; amylase: F values were estimated using a regression with a p value threshold sig- (8,240) = 7.55, p < 0.001 for both), albeit groups did not react dif- nificance at p < 0.05. ferentially (interaction group*time p > 0.05). The main effect of group was reflected in a significant difference between groups in the AUCg 2.9. AUC calculations values with strong effect sizes for both biomarkers (Cortisol effect size 0.83: FND 1776 ± 696 nmol/l, Controls 1297 ± 438, nmol/l, In order to account for the two different effects in our repeated p < 0.05, t = 2.27, df = 29, Amylase effect size 0.92 U/l: FND measures study, namely the background hormonal levels and the 79′822′500 ± 53′803′345 U/l, Controls 42′309′33 ± 20′343′461 U/l, change over time after stress, we used area under the curve (AUC) p < 0.05, t = 2.66, df = 30). calculations that have shown to be robust in psychoendocrinology Reflecting the absent group x time interaction, we found no statis- (Pruessner et al., 2003). These two formulae resulted in two variables tical differences when calculating changes of concentration in respect of indicating a) the total amount of hormone released (AUCg) and b) the time (AUCi) between both groups. time-dependent changes of hormonal release during and after the social stress (AUCi). 3.2. Life events

3. Results Recent (5-years) life events were analyzed by number and sub- jective impact for events related to change or negative events (Fig. 2). A Demographic data of the two populations are presented in Table 1. two —way ANOVA showed no significant differences between groups Patients and controls were comparable in terms of gender, age, time of for either type of events. We thus merged both types of events for menstrual cycle, state and trait anxiety. FND patients had higher de- further analysis. pression scores than controls (p < 0.05). The total number (mean number ± standard deviation: FND 13.6 ± 9.2, Controls 15.1 ± 10.6) or the subjective total impact of the 3.1. Objective biomarkers (cortisol and amylase) events (FND: 616 ± 512, Controls: 713 ± 602) did not differ between groups either. However, the total number of events significantly cor- Both objective stress biomarkers −cortisol and amylase- showed a related with baseline cortisol (AUCg) in patients, but not in controls significant main effect of group (cortisol F(1,29) = 4.9, p < 0.05, (FND: r = 0.67 p < 0.01, Controls: r = −0.09 p > 0.05 Fig. 2) as did amylase F(1,30) = 8.95, p < 0.01) with higher levels in patients the total impact of events (FND: r = 0.6, p < 0.05, Controls: (Fig. 1). The TSST introduced a statistically significant increase in both r=−0.03 p > 0.05). Amylase values (AUCg) did not correlate

145 K. Apazoglou et al. Psychoneuroendocrinology 85 (2017) 142–150

Fig. 3. Cortisol and Amylase response in patients with history of childhood sexual abuse. Left Panel: Blunted response to stress for both biomarkers in patients with history of childhood sexual abuse (N = 8). Between groups T-tests are presented for each time point. HC:healthy controls with no history of childhood sexual abuse, FND NAbus: patients with motor FND with no history of childhood sexual abuse, FND Abus: patients with motor FND with history of childhood sexual abuse. Right Panel: Subscores of the Childhood Trauma Questionnaire.

(p > 0.05) with total life events impact (r = 0.33) or number biological values were found at rest (samples before TSST). After stress, (r = 0.24) in patients nor in controls (impact r = 0.33, number a positive correlation was found in healthy controls, but not in FND r = 0.27). Note that both types of events, even when analyzed sepa- patients, between VAS and cortisol values (at 20 min:FND r = −0.13, rately, showed the same pattern of correlations. p > 0.05; Controls r = 0.60, p < 0.05 and 30 min:FND r = −0.17, p > 0.05; Controls r = 0.56, p < 0.05) and VAS and amylase values 3.3. Childhood trauma (at 20 min: FND r = 0.13, p > 0.05; Controls r = 0.66, p < 0.001) (this correlation is illustrated in Fig. 4). FND patients reported more childhood sexual abuse (9.4 ± 5.6 vs. Controls: 5.8 ± 2.1, p < 0.05, effect size 0.85, t = 2.33, df = 29) but 3.5. Duration and severity of motor symptoms and medication no other trauma (Fig. 3). Half of FND patients (50%) had a sexual abuse subscore of > 5, indicating some degree of abuse, whereas only 3 Duration and severity of the disorder (Mobility NeuroQoL) did not healthy controls had a similar score (20%). When we separated FND correlate (p > 0.05 for all Pearson correlations) with objective bio- patients in sexually abused (N = 8) and non-sexually abused subgroups markers or subjective stress (AUCg: dur ∼ cort, r = 0.45, dur ∼ sAA, (N = 8), the curves revealed a blunted response for both cortisol and r=−0.08, dur ∼ VAS, r = −0.14; sev ∼ cort, r = −0.22, sev ∼ sAA, amylase in abused patients as compared to non-sexually abused healthy r=−0.45, sev ∼ VAS, r = 0.02). No differences were found between controls. Pre-stress values were comparable to non-abused FND patients patients under medication (9 subjects) and those without medication (7 but lower levels were found after stress (T-test statistics are represented subjects) for cortisol (AUCg: Med 1963 ± 690, NoMed 1674 ± 784, on Fig. 3, we did not conduct between groups ANOVAs due to small p = 0.45, AUCi: Med 180 ± 264, NoMed 345 ± 248, p = 0.29), sample of subgroups − only 3 abuses controls). amylase (AUCg: Med 90392.5 ± 61979, NoMed 66232.5 ± 41618, p = 0.39, AUCi: Med −4127.5 ± 58436, NoMed −286 ± 18152, 3.4. Perceived stress p = 0.87) and VAS scores (AUCg: Med 334 ± 203, NoMed 366 ± 76, p = 0.72, AUCi: Med 217 ± 181, NoMed 167.5 ± 192, p = 0.63). Reported stress values (VAS) showed a main effect of time (p < 0.001) across groups, reflecting increase subjective stress after 4. Discussion the social stress task (Fig. 1, Panel A). No main effect of group was observed but a significant interaction group*time (p < 0.05) showing 4.1. Background biological markers of stress a higher subjective stress response in patients. Respectively, AUC values of subjective stress were not different between groups using the AUCg This study showed that patients suffering from motor FND have measurement, but differed significantly when calculated in respect of significantly higher levels of salivary cortisol (marker of the HPA axis time (AUCi: FND 196 ± 180, Controls 83 ± 174, p < 0.05, effect size stress system) and amylase (marker of sympathetic autonomous stress 0.65). In both groups, no correlations between VAS scores and system) than matched healthy controls. These results seem to differ

146 K. Apazoglou et al. Psychoneuroendocrinology 85 (2017) 142–150

Fig. 4. Perceived Stress and autonomous response. At each time point perceived stress (Vas score) was correlated with biological measures. After stress a positive correlation was found between VAS score and amylase peak value (at 20 min post-stress) in controls but not in patients, as illu- strated here.

from recent findings (Maurer et al., 2015) of equivalent diurnal cortisol in NES patients compared to controls (Bakvis et al., 2009a; Ponnusamy levels in 33 patients with motor FND and 33 healthy controls. In the et al., 2011) suggesting decreased emotion regulation (Beauchaine, latter study cortisol levels were examined at bedtime and the following 2015) and increased emotional reactivity, factors that may affect the morning upon spontaneous awakening, 30 min later, at noon and at stress response. Indeed, as demonstrated in children with FND (motor 3pm during a two-day hospitalization stay. In our study, subjects were and NES), low baseline vagal tone was related to impaired autonomous tested in the afternoon during a single visit in the laboratory between regulation (lower heart rate increase than healthy controls) following 1.30pm and 4.30pm. The only comparable time point is the 3pm stressful cognitive tasks, reflecting stress vulnerability (Kozlowska measure from Maurer et al., which shows higher values in FND patients et al., 2015). as represented graphically with log values. As raw data are not pro- Our data, together with findings from studies in NES patients, vided no direct statistical can however be done between our studies. suggest that HPA and sympathetic stress parameters are increased in These two studies seem to suggest that awakening cortisol might be both forms of FND (motor and NES). similar to controls in FND but is heightened in the afternoon. In line with our findings, heightened cortisol only in the afternoon have been 4.2. Biological response to a social stress versus perceived response to a reported in patients suffering from non-epileptic seizure (NES) (Tunca social stress et al., 2000; Bakvis et al., 2010a) another form of conversion disorder. A study in 12 NES patients (Bakvis et al., 2011) and 20 healthy controls, Regarding the biological response to stress, patients showed similar with a design comparable to ours, collected salivary cortisol between responses to healthy controls (for both cortisol and amylase) suggesting 1.15pm and 4pm − before and after physical stress induction (Cold a normal physiological adaptation to social stress. To our knowledge, pressure test) − and found significantly higher background levels in no study has reported biological data in response to a social stress in patients. Taking the circadian cortisol rhythm into account, it is thus motor FND and only one study used the same stressor as ours (TSST) in possible that cortisol levels in FND remain high during afternoon hours a cohort of 19 NES patients to look at the effect of stress induction on and do not follow the same diurnal variation as in healthy subjects. In preconscious threat processing but, like in our study, no differences order to confirm this hypothesis further research examining the diurnal between NES and controls were found in cortisol levels responses cortisol slope specifically in patients with motor FND is needed. (Bakvis et al., 2009a). Another study in 19 NES investigated working To our knowledge, data on amylase have never been collected in memory with another stress induction paradigm (Cold Pressure test) motor FND. Our results of increased autonomous sympathetic nervous and here again, patients did not differ from controls in their cortisol system activity in motor FND is in line with increased mean heart rate levels responses (Bakvis et al., 2010b). (reflecting sympathetic arousal) found in children and adolescent suf- It can be concluded that the cortisol response to a social stressor fering from FND (motor and NES) (Kozlowska et al., 2015). Further- (TSST) is normal in both motor FND and NES and that the cortisol re- more, a recent study found decreased vagal tone in motor FND by sponse to a physical stressor (Cold Pressure test) is normal in NES while measuring indexes of heart rate variability (RMSSD: Root mean squared it has not been tested yet in motor FND. However, It should, be un- successive differences of interbeat intervals) reflecting parasympathetic derscored that cortisol and amylase stress response are known to be tone (Maurer et al., 2016). Altogether, this suggest imbalance between blunted in sexually abused patients (Elzinga et al., 2008; Gordis et al., sympathetic and parasympathetic tone in motor FND. 2008; Schalinski et al., 2015). When we separated our patients in two In the NES literature, only one study looked at amylase and found sub-groups we found such dampened response in abused patients. As comparable levels between patients and healthy controls (Bakvis et al., only few controls reported sexual abuse (N = 3), we were not able to 2010a). On the other hand, heart rate variability indexes (CSI: cardio- fully control for this factor. To obtain a definite answer on this topic, sympathetic index) values reflecting increased sympathetic tone have future study designs should test larger groups of non-abused FND pa- been found in NES patients (Ponnusamy et al., 2011) compared to tients as well as abused healthy controls, as abuse may act as a con- healthy controls (but no difference with epileptic patients). Just like in founding variable when stating that stress response is normal in con- motor FND (Maurer et al., 2016), lower vagal tone (RMSSD) was found version disorder.

147 K. Apazoglou et al. Psychoneuroendocrinology 85 (2017) 142–150

Regarding perceived stress, our FND cohort reported a significantly hypothesized that elevated cortisol in patients with motor FND renders higher subjective response (VAS AUCi values) than controls, when their them more susceptible to external stimuli. biological responses (cortisol and amylase AUCi) were comparable to controls. Such dissociation between subjective perception and objective 4.4. Biological stress and FND symptom measurement in FND has also been observed in motor perception; pa- tients with functional tremor subjectively reported 65% more tremor Imaging studies in FND have suggested that an abnormal hyperar- time than objectively measured by actigraphy (Parees et al., 2012b). ousal state could influence sensorimotor processing underlying the Perception arises from integrating sensory information from the en- physical . One highlighted increased amygdala ac- vironment with internal predictions about the expected data (priors) tivity −reflecting the hyperarousal − not only to threat stimuli (fear (Edwards et al., 2012) and it has been hypothesized that in FND more faces) but also positive stimuli (happy faces) (Voon et al., 2010) and weight is put on prior expectation (Parees et al., 2012a), thus biasing another showed lack of physiological habituation over time in FND perception towards what is expected (forward prediction) with less patients to negative stimuli (sad and fear faces) (Aybek et al., 2015). In weight on the actual feedback information. One could hypothesize that both these studies, aberrant functional connectivity was found between in our stress experiment, FND patients expected more stress, which amygdala and supplementary motor areas, suggesting a link between resulted in higher subjective VAS scores. amygdalar hyperarousal and motor control in patients with motor FND. When looking at correlation between VAS scores and objective No studies to date have directly tested a link between baseline increased biological values at each time point, a positive correlation was observed cortisol levels and amygdalar hyperactivity in FND but an experimental in controls just after the stress test for amylase and cortisol, reflecting study was done in healthy controls (van Marle et al., 2009). Stress one’s capacity to accurately detect shifts in arousal states. In patients, (confirmed by increased cortisol, increased heart rate and decreased such correlation was absent despite the fact that stress response (AUCi heart rate variability) induced a shift in amygdala activity towards of cortisol and amylase) was similar to that of controls. This might enhance stimuli detection of fear and anger (better sensitivity) but at a occur due to an already aroused system at baseline, making it difficult cost, as even happy stimuli triggered such response (lower specificity). to detect any further change in level (ceiling effect). Or this might re- This shift occurs only in a stress condition and has a role in promoting flect an impaired detection of biological cues, which accords with survival, hence a preference for higher sensitivity, at a cost of lower findings of poorer interoceptive awareness found in subjects with FND specificity. Evidence of such loss of specificity −even at baseline (Ricciardi et al., 2016). Under stress, healthy individuals are known to without stress induction − has been observed in patients with motor display better interoceptive awareness (Schulz et al., 2013), which may FND (Voon et al., 2010) who displayed similar levels of amygdalar account for the better correlation between biological stress cues and activity when viewing fearful and happy faces whereas healthy in- perceived stress we found in the control group. To confirm this hy- dividuals controls showed a clear differential amygdalar activity pothesis, further research is needed and in particular a test addressing (higher for fearful and lower for happy faces). Future studies should specifically interoceptive awareness before and after stress induction in verify whether this shift is correlated with background cortisol values in FND. FND patients.

4.3. Influence of life events 4.5. Limitations

We found a strong positive correlation in patients between back- One major limitation of our study is the homogenous sample of ground cortisol levels (AUCg) and number of events (r = 0.67, motor FND, which limits the generalizability of our findings to the p < 0.01) as well as subjective impact of events (r = 0.6, p < 0.05), understanding of FND in a broader way. Moreover, the small sample confirming a link between life adversities and biological regulation of size did not allow correcting for a potential major confounder, sexual the HPA axis system. There are two ways to interpret this correlation: abuse, or differentiating patients according to the duration of the dis- life events may cause the stress hyperarousal by over stimulating the ease. Another potential cofound is the use of CNS acting drug in a large stress system − “the more adverse events − the more cortisol secre- proportion of our patients but no differences were detected between the tion”. If this were a physiological response, however, the correlation subgroups of patients with and without drugs. Also, the effect of such should also be seen in healthy controls who had similar number of life drugs on HPA axis is rather inhibitory according to the literature: events. An alternate interpretation, is that high background cortisol benzodiazepines (Korbonits et al., 1995; Locatelli et al., 2010), anti- levels are not a consequence of adverse events but constitute a pre- depressants (Mason and Pariante, 2006) and antiepileptic drugs disposition in certain individuals to develop FND. A longitudinal study (Rabinovitz et al., 2004) have been shown to lower cortisol secretion. looking at links between HPA axis arousal and life events followed a So even if such effect was possible in our sample, the significant dif- cohort of children from age 9–14yo until they reach 18yo and collected ference we found with higher values in patients should not be explained prospectively the occurrence of negative life events (LeMoult et al., by a medication effect. The choice of a self-report questionnaire to as- 2015) as well as onset of a psychiatric disorder (depression). Life events sess life events was mainly driven by feasibility reasons but has the predicted the onset of psychiatric illness only in subjects with baseline major drawback of being subject to recall bias and lacks objectivity in elevated AUCg cortisol. The authors argued that elevated total cortisol the judgment of impact of the life events. Also, it does not take into production potentiate susceptibility to environmental adversity. In account the nature of the events, like a potential secondary gain, which FND, such mechanism can be postulated arguing that the hyperarousal may be important in FND (Nicholson et al., 2016). of the HPA axis might predispose individuals to develop FND rather than being a mere consequence of adverse life events. Adverse events 5. Conclusions may then act as non-specific triggers. Other factors reflecting very early autonomic hyper-reactivity, such as altered feeding, sleeping or tactile First, our study confirms a hyperarousal stress state in motor FND by reactivity in infants, have been shown to predict the subsequent de- showing increased biomarkers of HPA axis and autonomous system velopment of functional symptoms (Rask et al., 2013). In line with this when the circadian cycle is at his trough (in the afternoon). Future concept, triggers of a physical nature (Stone et al., 2009; Parees et al., studies should aim at confirming the hypothesis that this hyperarousal 2014) and not only psychological stressors like adverse life events have is linked to the physical symptom of FND, possibly through the fol- been linked to FND onset. In one NES study, baseline cortisol levels lowing mechanisms: by favoring automatic habitual motor response were found to positively correlate with attentional bias (longer reaction mediated by basal ganglia at the expense of controlled response times) to angry masked stimuli (Bakvis et al., 2009b). One can mediated by prefrontal control.

148 K. Apazoglou et al. Psychoneuroendocrinology 85 (2017) 142–150

Second, our study confirms abnormal interoceptive abilities in pa- J. Vis. Exp.: JoVE 56, 3238. http://dx.doi.org/10.3791/3238. Carson, A., Lehn, A., 2017. Epidemiol. Handb. Clin. Neurol. 139, 47–60. tients with motor FND under stress with a dissociation between per- Carson, A.J., 2014. Introducing a ‘neuropsychiatry' special issue: but what does that ceived and objective stress. mean? J. Neurol. Neurosurg. Psychiatry 85, 121–122. Third, our study confirms that adverse life events play a role in FND Cella, D., Lai, J., Nowinski, C., Victorson, D., Peterman, A., Miller, D., Bethoux, F., fi Heinemann, A., Rubin, S., Cavazos, J., Reder, A., Sufit, R., Simuni, T., Holmes, G., as, they are linked to a hyperarousal state, which is speci c to motor Siderowf, A., Wojna, V., Bode, R., McKinney, N., Podrabsky, T., Wortman, K., Choi, FND. It does, however, not imply adverse events have a causal link and S., Gershon, R., Rothrock, N., Moy, C., 2012. Neuro-QOL: brief measures of health- future studies should aim to refine the role of life events as potential related quality of life for clinical research in neurology. Neurology 78, 1860–1867. triggers or mediators of the disorder. DSM-5, 2013. Diagnostic and Statistical Manual of Mental Disorders (DSM-5). American Psychiatric Association, Washington, DC. Finally our study opens the door to future research aiming to un- Edwards, M.J., Adams, R.A., Brown, H., Parees, I., Friston, K.J., 2012. A bayesian account derstand why such stress hyperarousal occurs in FND. There is to date of ‘hysteria'. Brain 135, 3495–3512. little evidence of a genetic background for FND but familial cases have Elzinga, B.M., Roelofs, K., Tollenaar, M.S., Bakvis, P., van Pelt, J., Spinhoven, P., 2008. Diminished cortisol responses to psychosocial stress associated with lifetime adverse been reported (Stamelou et al., 2013). Gene-environment interaction events a study among healthy young subjects. Psychoneuroendocrinology 33, should also be studied and the recent development of epigenetics offers 227–237. this possibility (Frodl, 2017). Frodl, T., 2017. Do (epi)genetics impact the brain in functional neurologic disorders? Handb. Clin. Neurol. 139, 157–165. Gelauff, J., Stone, J., Edwards, M., Carson, A., 2014. The prognosis of functional (psy- Conflict of interest chogenic) motor symptoms: a systematic review. J. Neurol. Neurosurg. Psychiatry 85, 220–226. – fl Gordis, E.B., Granger, D.A., Susman, E.J., Trickett, P.K., 2008. Salivary alpha amylase - Authors, KA, VM, JW, GFP, SA, declare no con ict of interest related cortisol asymmetry in maltreated youth. Horm. Behav. 53, 96–103. to their work presented in the manuscript entitled “Biological and Hellhammer, D.H., Wust, S., Kudielka, B.M., 2009. Salivary cortisol as a biomarker in Perceived Stress in motor Functional Neurological Disorders”. stress research. Psychoneuroendocrinology 34, 163–171. Hinson, V.K., Weinstein, S., Bernard, B., Leurgans, S.E., Goetz, C.G., 2006. Single-blind clinical trial of psychotherapy for treatment of psychogenic movement disorders. Financial disclosure Parkinsonism Relat. Disord. 12, 177–180. Hubschmid, M., Aybek, S., Maccaferri, G.E., Chocron, O., Gholamrezaee, M.M., Rossetti, ffi The authors have no disclosure. A.O., Vingerhoets, F., Berney, A., 2015. E cacy of brief interdisciplinary psy- chotherapeutic intervention for motor conversion disorder and nonepileptic attacks. Gen. Hosp. Psychiatry 37, 448–455. Grants Kanaan, R.A., 2017. Freud's hysteria and its legacy. Handb. Clin. Neurol. 139, 37–44. Kirschbaum, C., Pirke, K.M., Hellhammer, D.H., 1993a. The ‘Trier Social Stress Test'–a tool for investigating psychobiological stress responses in a laboratory setting. Ambizione Grant PZ00P3_147997 for SA. Neuropsychobiology 28, 76–81. Kirschbaum, C., Pirke, K.M., Hellhammer, D.H., 1993b. The trier social stress test − a tool Acknowledgments for investigating psychobiological stress responses in a laboratory setting. Neuropsychobiology 28, 76–81. Korbonits, M., Trainer, P.J., Edwards, R., Besser, G.M., Grossman, A.B., 1995. We thank Dr Michel Rossier (Laboratory of clinical chemistry and Benzodiazepines attenuate the pituitary-adrenal responses to corticotrophin-re- toxicology, Central Institute of the Hospital of Valais, Sion) for the help leasing hormone in healthy volunteers, but not in patients with Cushing's syndrome. Clin. Endocrinol. (Oxf.) 43, 29–35. with cortisol and amylase analysis, Dr Deepa Pothalil and Dr Dimitri Kozlowska, K., Palmer, D.M., Brown, K.J., McLean, L., Scher, S., Gevirtz, R., Chudleigh, Horn for their help in recruiting patients. C., Williams, L.M., 2015. Reduction of autonomic regulation in children and ado- We thank Dr Ulrike Rimmele for advice on methodolody, statistical lescents with conversion disorders. Psychosom. Med. 77, 356–370. Kroenke, K., 2007. Efficacy of treatment for somatoform disorders: a review of rando- analyses and critical review of the manuscript. mized controlled trials. Psychosom. Med. 69, 881–888. Laux, L., 1981. Das State-Trait-Angstinventar (STAI): Theoretische Grundlagen und References Handanweisung. Beltz Test, Göttingen. LeMoult, J., Ordaz, S.J., Kircanski, K., Singh, M.K., Gotlib, I.H., 2015. Predicting first onset of depression in young girls: interaction of diurnal cortisol and negative life Abelson, J.L., Erickson, T.M., Mayer, S.E., Crocker, J., Briggs, H., Lopez-Duran, N.L., events. J. Abnorm. Psychol. 124, 850–859. Liberzon, I., 2014. Brief cognitive intervention can modulate neuroendocrine stress Locatelli, V., Bresciani, E., Tamiazzo, L., Torsello, A., 2010. Central nervous system-acting responses to the trier social stress test: buffering effects of a compassionate goal or- drugs influencing hypothalamic-pituitary-adrenal axis function. Endocr. Dev. 17, ientation. Psychoneuroendocrinology 44, 60–70. 108–120. Amiel-Lebigre, F., 1985. Méthodes et échelles d'évaluation des comportements. EAP, Issy Mason, B.L., Pariante, C.M., 2006. The effects of antidepressants on the hypothalamic- les Moulineaux. pituitary-adrenal axis. Drug News Perspect. 19, 603–608. Aybek, S., Nicholson, T.R., Zelaya, F., O'Daly, O.G., Craig, T.J., David, A.S., Kanaan, R.A., Maurer, C.W., LaFaver, K., Ameli, R., Toledo, R., Hallett, M., 2015. A biological measure 2014. Neural correlates of recall of life events in conversion disorder. JAMA of stress levels in patients with functional movement disorders. Parkinsonism Relat. Psychiatry 71, 52–60. Disord. 21, 1072–1075. Aybek, S., Nicholson, T.R., O'Daly, O., Zelaya, F., Kanaan, R.A., David, A.S., 2015. Maurer, C.W., Liu, V.D., LaFaver, K., Ameli, R., Wu, T., Toledo, R., Epstein, S.A., Hallett, Emotion-motion interactions in conversion disorder: an FMRI study. PLoS One 10, M., 2016. Impaired resting vagal tone in patients with functional movement dis- e0123273. orders. Parkinsonism Relat. Disord. 30, 18–22. Bakvis, P., Roelofs, K., Kuyk, J., Edelbroek, P.M., Swinkels, W.A., Spinhoven, P., 2009a. Nater, U.M., Rohleder, N., Gaab, J., Berger, S., Jud, A., Kirschbaum, C., Ehlert, U., 2005. Trauma, stress, and preconscious threat processing in patients with psychogenic Human salivary alpha-amylase reactivity in a psychosocial stress paradigm. Int. J. nonepileptic seizures. Epilepsia 50, 1001–1011. Psychophysiol. 55, 333–342. Bakvis, P., Spinhoven, P., Roelofs, K., 2009b. Basal cortisol is positively correlated to Nicholson, T.R., Aybek, S., Craig, T., Harris, T., Wojcik, W., David, A.S., Kanaan, R.A., threat vigilance in patients with psychogenic nonepileptic seizures. Behav.: 2016. Life events and escape in conversion disorder. Psychol. Med. 46, 2617–2626. E&B 16, 558–560. Parees, I., Kassavetis, P., Saifee, T.A., Sadnicka, A., Bhatia, K.P., Fotopoulou, A., Edwards, Bakvis, P., Spinhoven, P., Giltay, E.J., Kuyk, J., Edelbroek, P.M., Zitman, F.G., Roelofs, K., M.J., 2012a. ‘Jumping to conclusions' bias in functional movement disorders. J. 2010a. Basal hypercortisolism and trauma in patients with psychogenic nonepileptic Neurol. Neurosurg. Psychiatry 83, 460–463. seizures. Epilepsia 51, 752–759. Parees, I., Saifee, T.A., Kassavetis, P., Kojovic, M., Rubio-Agusti, I., Rothwell, J.C., Bhatia, Bakvis, P., Spinhoven, P., Putman, P., Zitman, F.G., Roelofs, K., 2010b. The effect of stress K.P., Edwards, M.J., 2012b. Believing is perceiving: mismatch between self-report induction on working memory in patients with psychogenic nonepileptic seizures. and actigraphy in psychogenic tremor. Brain 135, 117–123. Epilepsy Behav.: E&B 19, 448–454. Parees, I., Kojovic, M., Pires, C., Rubio-Agusti, I., Saifee, T.A., Sadnicka, A., Kassavetis, P., Bakvis, P., Spinhoven, P., Zitman, F.G., Roelofs, K., 2011. Automatic avoidance tenden- Macerollo, A., Bhatia, K.P., Carson, A., Stone, J., Edwards, M.J., 2014. Physical cies in patients with psychogenic non-epileptic seizures. Seizure 20, 628–634. precipitating factors in functional movement disorders. J. Neurol. Sci. 338, 174–177. Beauchaine, T.P., 2015. Respiratory sinus arrhythmia: a transdiagnostic biomarker of Ponnusamy, A., Marques, J.L., Reuber, M., 2011. Heart rate variability measures as emotion dysregulation and psychopathology. Curre. Opin. Psychol. 3, 43–47. biomarkers in patients with psychogenic nonepileptic seizures: potential and lim- Beck, A.T., Ward, C.H., Mendelson, M., Mock, J., Erbaugh, J., 1961. An inventory for itations. Epilepsy Behav. 22, 685–691. measuring depression. Arch. Gen. Psychiatry 4, 561–571. Pruessner, J.C., Kirschbaum, C., Meinlschmid, G., Hellhammer, D.H., 2003. Two formulas Bernstein, D.P., Fink, Laura, 1998. Childhood Trauma Questionnaire: a Retrospective for computation of the area under the curve represent measures of total hormone Self-report: Manual. Psychological Corporation, Orlando. concentration versus time-dependent change. Psychoneuroendocrinology 28, Birkett, M.A., 2011. The trier social stress test protocol for inducing psychological stress. 916–931.

149 K. Apazoglou et al. Psychoneuroendocrinology 85 (2017) 142–150

Rabinovitz, S., Mostofsky, D.I., Yehuda, S., 2004. Anticonvulsant efficiency, behavioral cortisol are associated with sexual abuse in childhood. PLoS One 10, e0136921. performance and cortisol levels: a comparison of carbamazepine (CBZ) and a fatty Schulz, A., Lass-Hennemann, J., Sutterlin, S., Schachinger, H., Vogele, C., 2013. Cold acid compound (SR-3). Psychoneuroendocrinology 29, 113–124. pressor stress induces opposite effects on cardioceptive accuracy dependent on as- Rask, C.U., Ørnbøl, E., Olsen, E.M., Fink, P., Skovgaard, A.M., 2013. Infant behaviors are sessment paradigm. Biol. Psychol. 93, 167–174. predictive of functional somatic symptoms at ages 5–7 years: results from the co- Stamelou, M., Cossu, G., Edwards, M.J., Murgia, D., Parees, I., Melis, M., Bhatia, K.P., penhagen child cohort CCC2000. J. Pediatr. 162, 335–342. 2013. Familial psychogenic movement disorders. Mov. Disord. 28, 1295–1298. Reuber, M., Burness, C., Howlett, S., Brazier, J., Grunewald, R., 2007. Tailored psy- Stone, J., Carson, A., Aditya, H., Prescott, R., Zaubi, M., Warlow, C., Sharpe, M., 2009. chotherapy for patients with functional neurological symptoms: a pilot study. J. The role of physical injury in motor and sensory conversion symptoms: a systematic Psychosom. Res. 63, 625–632. and narrative review. J. Psychosom. Res. 66, 383–390. Ricciardi, L., Demartini, B., Crucianelli, L., Krahe, C., Edwards, M.J., Fotopoulou, A., Stone, J., Carson, A., Duncan, R., Roberts, R., Warlow, C., Hibberd, C., Coleman, R., Cull, 2016. Interoceptive awareness in patients with functional neurological symptoms. R., Murray, G., Pelosi, A., Cavanagh, J., Matthews, K., Goldbeck, R., Smyth, R., Biol. Psychol. 113, 68–74. Walker, J., Sharpe, M., 2010. Who is referred to neurology clinics?–The diagnoses Roelofs, K., Keijsers, G.P., Hoogduin, K.A., Naring, G.W., Moene, F.C., 2002. Childhood made in 3781 new patients. Clin. Neurol. Neurosurg. 112, 747–751. abuse in patients with conversion disorder. Am. J. Psychiatry 159, 1908–1913. Stone, J., LaFrance Jr., W.C., Brown, R., Spiegel, D., Levenson, J.L., Sharpe, M., 2011. Roelofs, K., Spinhoven, P., Sandijck, P., Moene, F.C., Hoogduin, K.A., 2005. The impact of Conversion disorder: current problems and potential solutions for DSM-5. J. early trauma and recent life-events on symptom severity in patients with conversion Psychosom. Res. 71, 369–376. disorder. J. Nerv. Ment. Dis. 193, 508–514. Tunca, Z., Ergene, U., Fidaner, H., Cimilli, C., Ozerdem, A., Alkin, T., Aslan, B.U., 2000. Rohleder, N., Nater, U.M., Wolf, J.M., Ehlert, U., Kirschbaum, C., 2004. Psychosocial Reevaluation of serum cortisol in conversion disorder with seizure (pseudoseizure). stress-induced activation of salivary alpha-amylase: an indicator of sympathetic ac- Psychosomatics 41, 152–153. tivity? Ann. N. Y. Acad. Sci. 1032, 258–263. Voon, V., Brezing, C., Gallea, C., Ameli, R., Roelofs, K., LaFrance Jr., W.C., Hallett, M., Sar, V., Akyuz, G., Kundakci, T., Kiziltan, E., Dogan, O., 2004. Childhood trauma, dis- 2010. Emotional stimuli and motor conversion disorder. Brain 133, 1526–1536. sociation, and psychiatric comorbidity in patients with conversion disorder. Am. J. van Marle, H.J., Hermans, E.J., Qin, S., Fernandez, G., 2009. From speci ficity to sensi- Psychiatry 161, 2271–2276. tivity: how acute stress affects amygdala processing of biologically salient stimuli. Schalinski, I., Elbert, T., Steudte-Schmiedgen, S., Kirschbaum, C., 2015. The cortisol Biol. Psychiatry 66, 649–655. paradox of trauma-related disorders: lower phasic responses but higher tonic levels of

150