Moniek Hutschemaekers

Moniek Hutschemaekers The effects of testosterone on exposure therapy for social anxiety disorder

Speak Up! The effects of testosterone on exposure therapy for social anxiety disorder Moniek Helena Maria Hutschemaekers

Provided by thesis specialist Ridderprint, Printing: Ridderprint Layout and design: Erwin Timmerman, Funding information The studies within this dissertation were supported by a VICI grant (#453-12-001) from the Netherlands Organization for Scientific Research (NWO) and a consolidator grant from the European Research Council (ERC_CoG-2017_772337) awarded to Professor K. Roelofs. Copyright © 2023 Moniek Hutschemaekers All rights reserved. Save expectations stated by law, no part of this thesis may be reproduced, stored in a retrieval system of any nature, or transmitted, in whole or in part, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, included in a complete or partial transcription, without the written permission of the author, or – when appropriate – of the publisher’s of the presented publications.

Speak Up! The effects of testosterone on exposure therapy for social anxiety disorder Proefschrift ter verkrijging van de graad van doctor aan de Radboud Universiteit Nijmegen op gezag van de rector magnificus prof. dr. J.H.J.M. van Krieken, volgens besluit van het college voor promoties in het openbaar te verdedigen op vrijdag 10 november 2023 om 12.30 uur precies door Moniek Helena Maria Hutschemaekers geboren op 22 april 1991 te Tilburg

Promotor Prof. dr. K. Roelofs Copromotoren Dr. M. Kampman (Pro Persona) Dr. R.A. de Kleine (Universiteit Leiden) Manuscriptcommissie Prof. dr. I. Tendolkar Prof. dr. S.M. Bögels (Universiteit van Amsterdam) Prof. dr. S.G. Hofmann (Boston University, Verenigde Staten)

Contents Chapter 1 General introduction 7 Chapter 2 Neuroendocrinological aspects of social anxiety and aggression related disorders 23 Chapter 3 Endogenous testosterone levels are predictive of symptom reduction with exposure therapy in social anxiety disorder 43 Chapter 4 The enhancing effects of testosterone in exposure treatment for social anxiety disorder: a randomized proof-of-concept trial 65 Chapter 5 Social Avoidance and Testosterone Enhanced Exposure Efficacy in Women with Social Anxiety Disorder: A Pilot Investigation 89 Chapter 6 Summary and General Discussion 111 References Research data management Nederlandse Samenvatting Curriculum Vitae List of publications Acknowledgements 129 153 156 165 166 168

Chapter 1 General introduction

9 General introduction Imagine giving a speech in front of a large audience. While standing there looking at all those faces in front of you, your heart rate will likely rise, you may feel a bit warm and think “my cheeks may turn red” or “I hope they will like my talk.” These are normal reactions to a socially challenging event. However, for some people those feelings turn into a dominating fear, far more challenging and burdensome: Meet Rose. Rose is a 21-year-old bachelor student. Despite the fact that she has good grades and enjoys studying, she experiences problems in completing her bachelor. She did not pass her exam as she is terrified of giving a speech in front of a public. She worries she will tremble, blush and stutter. The thought of a complete blackout and people laughing at her may not reflect a realistic scenario, but feels very real and provokes anxiety. To prevent this scenario from happening, she avoids giving speeches entirely for a few years now. Her fear of public speaking started during secondary school where she had to give a plenary presentation. Although she prepared it very well, she saw some of her classmates laughing which she attributed to her behavior. The thought of doing something stupid made her nervous which in turn made it hard to find the right words. After this experience, she avoided future presentations and became increasingly scared of public performances. Recently, she started to avoid other social situations as well, such as parties, even of good friends. In these situations, she also noticed anxiety symptoms, feeling scared not knowing what to say and being disliked. Rose symptoms meet the diagnostic criteria of Social Anxiety Disorder. As her anxiety symptoms are more and more hindering here social and academic functioning, she seeks treatment to deal with her anxiety. Like Rose, many other individuals are suffering from Social Anxiety Disorder (SAD). This impairing mental disorder is characterized by persistent fear and avoidance of social and performance situations. SAD is one of the most common mental health disorders and it persists when untreated with high levels of impairment in social or occupational function (Aderka et al., 2012; Bruce et al., 2005; Kessler et al., 2005). It can be treated effectively with cognitive behavioral therapy. However, around 45-55% of the individuals with SAD do not profit sufficiently from treatment (Carpenter et al., 2018; Loerinc et al., 2015). Therefore, various (pharmacological) enhancement strategies have been examined to boost therapy outcomes for SAD (Guastella, Howard, Dadds, Mitchell, & Carson, 2009; Hofmann, Fang, & Gutner, 2014; Smits, Rosenfield, et al., 2013a). Based on pre-clinical research, the steroid hormone testosterone might specifically yield promise to enhance exposure therapy effects for those suffering from SAD. This hormone is an important regulator of social behavior both in males and females (Hermans & Van Honk, 2006). Relatively low levels of testosterone have been linked to social fear and avoidance 1

10 Chapter 1 (Enter, Spinhoven, & Roelofs, 2014; Giltay et al., 2012), presumably because the hormone is particularly relevant for preparing a person for socially challenging situations (Wingfield, Hegner, Dufty, & Ball, 1990). Therefore, the overarching aim of this dissertation is to test the potential of testosterone as an enhancer for exposure treatment efficacy for SAD. With this dissertation I aimed to build on fundamental research within the field of neuroendocrinology and to link the acquired knowledge to clinical experimental psychology. Concretely, I aimed to translate well-established experimental findings on the social approach-promoting properties of testosterone in healthy individuals and SAD (Enter et al., 2014; Enter, Spinhoven, & Roelofs, 2016; Enter, Terburg, Harrewijn, Spinhoven, & Roelofs, 2016; Radke et al., 2015; Terburg et al., 2016) to a clinical application in order to improve exposure therapy efficacy in SAD. In this general introduction I will first present a description of SAD, followed by an explanation of exposure therapy and one of its proposed mechanisms of action. Next, I will focus on social avoidance. Firstly by discussing avoidance in the maintenance and treatment of SAD, secondly by providing a neuroendocrinal model of avoidance. Following that model, I will zoom in on the potential of testosterone as an enhancer for exposure therapy for SAD. Finally, I will present the rationale and specific aims of this dissertation and I conclude with an outline of the chapters. Social anxiety disorder Social anxiety disorder (SAD) is the most common and burdensome of all anxiety disorders with a lifetime prevalence of 13% and long-term disability (Aderka et al., 2012; Bandelow & Michaelis, 2015; Bruce et al., 2005; Hendriks et al., 2016). SAD is characterized by an intense fear of social situations in which the individual may be scrutinized by others such as interpersonal interactions (e.g., a conversation), being observed (e.g., eating in public) and performing in front of others (e.g., giving a speech). SAD can be specified into a performance only variant when the fear is restricted to public speaking. These social situations are usually avoided or endured with intense fear or anxiety (American Psychiatric Association, 2013). Individuals with SAD fear that their behavior, physical sensations or appearance will be negatively evaluated by others. They realize that their fear is excessive or unrealistic. SAD usually develops during early childhood or adolescence (Kessler et al., 2005; Schneier, Luterek, Heimberg, & Leonardo, 2004) and typically follows an enduring course without treatment. Moreover, compared to healthy individuals, individuals with SAD have a greater risk of developing comorbid disorders

11 General introduction such as depression and other anxiety disorders (Fehm, Beesdo, Jacobi, & Fiedler, 2008; Lépine & Pélissolo, 2000). Several theoretical models try to explain the maintenance of SAD (Clark & Wells, 1995; Heimberg & Rapee, 1997; Hofmann, 2007; Spence & Rapee, 2016; Wong & Rapee, 2016). All these models share similar aspects, for example increased self-focused attention, negative self-perception and attention to (perceived) social threat. Moreover, these models emphasize the maintaining role of behavioral processes. Specifically, avoidance and safety behaviors. Avoidance refers to the behavioral strategy to prevent exposure to the feared social situation (e.g., avoiding giving a speech or going to a party) whereas safety behaviors refer to all actions aimed to reduce or eliminate social threat while being in the social situation (e.g. holding notes or avoiding eye contact). Avoidance and safety behaviors are thought to play a crucial role in these models since they create a negative feedback loop by which the (social) anxiety remains or increases. That is, when individuals with SAD engage in these behaviors they try to prevent being rejected resulting in a momentary reduction of fear. However, as a result it becomes impossible to critically evaluate their feared outcomes and therefore the anxiety is maintained (Hofmann, 2007; Wong & Rapee, 2016). Cognitive behavioral therapy and exposure for SAD Cognitive behavioral therapy (CBT), is one of the psychological treatments of choice for SAD (Canton, Scott, & Glue, 2012; Pilling et al., 2013, Dutch Treatment Guidelines). CBT can be delivered individually or in a group and is considered the most efficacious and empirically supported treatment for SAD and other anxiety disorders (Hofmann & Smits, 2008a; Norton & Price, 2007; Tolin, 2010). Long term studies show that CBT can have long lasting effects (Leichsenring et al., 2014; Willutzki, Teismann, & Schulte, 2012). However, still many adults with SAD do not benefit from CBT. Response rates (the percentage of the treatment group that is classified as a “responder”) vary between 45–55% (Loerinc et al., 2015). In line with the proposed theoretical models of SAD, CBT aims to challenge dysfunctional beliefs about the likelihood of anticipated social danger (e.g., social rejection), by using cognitive techniques (e.g., keeping thought records and challenging automatic thoughts) as well as behavioral techniques, such as exposure (Hofmann, 2008; Smits, Julian, Rosenfield, & Powers, 2012). Exposure is thought to be one of the most crucial interventions in CBT protocols (Hofmann & Smits, 2008a; Norton & Price, 2007). Exposure treatment involves repeated confrontation with feared stimuli in the absence of the feared outcome. The process of fear extinction, which can be seen as a laboratory 1

12 Chapter 1 analogue for exposure therapy, helps us understand the underlying mechanism contributing to the effects of exposure therapy (Bouton, Mineka, & Barlow, 2001; Craske et al., 2008; Vervliet, Craske, & Hermans, 2013). In this process a conditional stimulus (CS) that was previously paired with an aversive outcome (unconditional stimulus, US) is repeatedly presented without being followed by the US, leading to the loss of the fear response. According to inhibitory learning theory (ILT), a recent exposure model, inhibitory learning plays an important role in extinction learning (Craske et al., 2008; Craske, Treanor, Conway, Zbozinek, & Vervliet, 2014a; Craske, Treanor, Zbozinek, & Vervliet, 2022; Lang, Craske, & Bjork, 1999). It is stated that the original CS – US association does not disappear during exposure, but a new association: US does not predict the US (no US), is learned in addition to the original association (Bouton & King, 1983). After extinction, the CS is associated with two meanings, the original fear association (lower pathway figure 1.1) and the inhibitory meaning (upper pathway figure 1.1). Figure 1.1 An illustration of an exposure model for Social Anxiety Disorder. Exposure treatment is characterized by repeated confrontation with the feared situation (e.g., the public speech, the CS) in absence of the feared outcome (No-US), reflected by the multiple arrows in the upper route. Within exposure, the individual learns a new association between the social situation and the absence of danger. Retrieval of this new association in a social situation results in feeling calmer. According to Inhibitory Learning Theory (ILT), both the original association (lower pathway) as well as the new association (upper pathway) remain. The new association competes for retrieval with the original association. Moreover, both the new association (CS – no US) and the original association (CS – US) compete for retrieval. Inhibitory learning can be maximized by expectancy violation in which the harm expectancies of the individual are altered. For example, Rose would be repeatedly exposed to her feared situation such as giving a public speech (CS) in order to test her explicitly stated harm expectancy: “people will laugh at me and reject me”

13 General introduction (US) when she gives a speech in different contexts (see Figure 1.1). When she experiences that she is not being rejected repeatedly, she no longer expects social rejection to occur when giving a public speech. Avoidance behavior Avoidance and safety behaviors in SAD As described in the different psychological models (Clark & Wells, 1995; Heimberg & Rapee, 1997; Hofmann, 2007; Spence & Rapee, 2016; Wong & Rapee, 2016), avoidance behavior plays a crucial role in the maintenance of SAD. As mentioned above, avoidance is a behavioral strategy to prevent exposure to the feared situation, which can take different forms. For example, an individual with SAD could completely avoid social situations such as a party, a speech or a job interview. Also, avoidance behavior can take more subtle forms in which individuals aim to reduce their distress or hide it, such as avoiding eye contact, speaking quietly, or taking somebody with them to the feared situation (those types of more subtle avoidance behaviors can also be classified as safety behaviors, Cuming et al., 2009; Wells et al., 1995). Although avoidance can be adaptive in threatening situations (e.g., running away when being attacked by a dangerous animal), and can result in temporary relief in the short run, avoidance is unnecessary in absence of real danger and can become maladaptive in the long run. Critically, systematic avoidance behavior hinders the individual to learn that fears are unsubstantiated. Rose can never experience that she can chat at a party and might not be rejected if she avoids going there at all. The same is true when she uses safety behaviors such as looking away when people make eye contact. Moreover, successful avoidance usually results in a temporary reduction in anxiety and is therefore reinforcing and persistent (Wong & Moulds, 2011). As such, social avoidance behavior is a major factor that maintains fear of social situations in individuals with SAD (Arnaudova, Kindt, Fanselow, & Beckers, 2017; Clark & Wells, 1995). Therefore, reducing avoidance behavior is the core target of treatment for SAD. Safety and subtle avoidance behaviors are typically discouraged during exposure sessions through modeling by the therapist and by providing explicit instructions (Craske, Treanor, Conway, Zbozinek, & Vervliet, 2014b; Hofmann & Otto, 2017), since they may also hamper the efficacy of exposure therapy (Wells et al., 1995). A therapist might for instance stimulate Rose not to use notes during the speech neither holding something in her hands, stimulating effective approach behavior. Otherwise, Rose may feel that she could only give the speech because of her notes in her hand (Piccirillo, Taylor Dryman, & Heimberg, 2016). In fact, the use of safety behaviors could increase the likelihood of 1

14 Chapter 1 feared outcomes (e.g., people may actually reject her because she avoids looking at them; Cuming et al., 2009; Piccirillo, Taylor Dryman, & Heimberg, 2016). A number of studies indeed showed that the use of safety behaviors impedes exposure efficacy in SAD (McManus et al., 2009; Morgan & Raffle, 1999; Piccirillo et al., 2016). However, studies in other anxiety disorders show that the controlled use of safety behaviors may be beneficial in the context of exposure treatment (Milosevic & Radomsky, 2008), as it might help individuals in taking the first steps in exposure and it promotes acceptability. Avoidance tendencies Individuals with SAD do not only show overt social avoidance such as avoiding a social event, they also show more automatic and implicit avoidance tendencies that can be picked up in speeded experimental tasks. An example is the biased information processing toward threatening stimuli. This is usually characterized by initial increased attention to negative emotional information, such as angry faces followed by attentional avoidance of these stimuli, specifically avoidance of the eyes, in order to regulate anxiety provoked by the initial registration of threat (the CS), (Chen & Clarke, 2017). In addition, biased action tendencies seem to play a particular role in individuals with SAD. These action tendencies can for instance be assessed by means of social Approach Avoidance Tasks (AAT: Rinck & Becker, 2007), in which participants respond to social stimuli (for example faces) by pushing (avoidance) or pulling (approach) a joystick. Socially anxious individuals typically show avoidance of social stimuli; i.e. stronger avoidance tendencies compared to approach tendencies toward angry, but also happy faces (Heuer, Rinck, & Becker, 2007; Loijen, Vrijsen, Egger, Becker, & Rinck, 2020; Roelofs, Putman, et al., 2010; Roelofs, van Peer, et al., 2009a) and even neutral faces, compared to non-social stimuli and healthy controls (Kuckertz, Strege, & Amir, 2017). Angry faces communicate potential threat and neutral and happy faces are ambiguous to individuals with SAD and therefore labeled as threatening, and thereby activating avoidance mechanisms (Heuer et al., 2007; Roelofs et al., 2010). Previous work showed that automatic avoidance tendencies assessed by an AAT could predict real life avoidance in specific phobia (Rinck & Becker, 2007). These findings indicate that automatic avoidance tendencies may underly overt avoidance behavior, an important maintaining factor in SAD. Moreover, it could be hypothesized that these avoidance tendencies hinder successful exposure, presumably since they prevent individuals with SAD to really engage in the exposure.

15 General introduction Neuroendocrinal regulation of social avoidance Hypothalamus Pituitary Male Testes Female Ovaries GnRH FSH, LH Testosterone Amygdala Striatum PFC Social threat A. B. Figure 1.2. Illustration of the HPG-axis and a simplified model of the approach enhancing properties of testosterone. A: Illustration of the hypothalamic–pituitary–gonadal axis (HPG-axis) activity of which leads to testosterone production. B: Simplified neural model of the proposed threat approach enhancing properties of testosterone based on neuroimaging studies verifying effects of testosterone administration during social threat challenges (such as visual presentation of an angry facial expression): testosterone enhances amygdala activity, dopaminergic projections from the amygdala to the striatum, and has been associated with reduced connectivity between the prefrontal cortex (PFC) and the amygdala. Note. GnRH = Gonadotropin-releasing hormone, FSH = Follicle-stimulating hormone, LH = Luteinizing hormone. + indicates excitatory connection; - indicates inhibitory connections. 1

16 Chapter 1 Testosterone Produced by the Hypothalamus-Pituitary-Gonadal (HPG)-axis (see figure 1.2A), testosterone constitutes an important regulator of social motivational behavior in both males and females, including social approach and avoidance behavior (Hermans & Van Honk, 2006; Mazur & Booth, 1998). Gonadotropin-releasing hormone (GnRH) is secreted from the hypothalamus, which stimulates the production of luteinizing hormone (LH) and follicle stimulating hormone (FSH) in the pituitary gland. This triggers the production of testosterone and estradiol in the gonads (i.e., testes and ovaries). Testosterone follows a diurnal cycle with the highest levels upon wakening (Diver, Imtiaz, Ahmad, Vora, & Fraser, 2003). Next to the pre- and early postnatal organizational effects on the brain structure, testosterone also affects emotion, motivation and behavior later in life (Lombardo et al., 2012; McHenry, Carrier, Hull, & Kabbaj, 2014). Endogenous testosterone levels can be assessed by blood samples (Serum Testosterone), but also less invasively via (passive drool) saliva samples (Salivary Testosterone). There is intra- variability in testosterone levels, depending on for example, time of the day and menstrual cycle as well as inter-individual variability (e.g., age and sex). Typically, the lower end of the serum concentration range is 4-5 times higher in healthy males compared to the upper end of the healthy female range (Clark et al., 2019; Kanakis, Tsametis, & Goulis, 2019). In addition to testosterone assessment, testosterone can be administrated to individuals (exogenous testosterone) through different methods (i.e., sublingually or using gels or via injections), enabling determination of a causal relationship between testosterone reactivity and its effects on social motivational behavior. Within this dissertation, we applied a single dose (0.5 mg) sublingual testosterone administration. This well-established method has been used in healthy and anxious individuals, showing consistent psychophysiological and behavioral effects approximately 4-6 hours after administration (Bos, Panksepp, Bluthé, & Honk, 2012; Tuiten et al., 2000). Testosterone interacts with other neurotransmitters and -peptides such as oxytocin, vasopressin and dopamine (de Souza Silva, Mattern, Topic, Buddenberg, & Huston, 2009). Moreover, the HPG-axis works in antagonism with the Hypothalamus-Pituitary-Adrenal (HPA)-axis. Specifically, cortisol (end-product of the HPA-axis) has an inhibitory effect on the production and actions of testosterone and vice versa (Toufexis, Rivarola, Lara, & Viau, 2014). Both the HPG and HPA-axis are important in the regulation of social motivational behavior (Mehta & Josephs, 2010; Roelofs, van Peer, et al., 2009b), but within this dissertation I will focus on the HPG-axis and its end product testosterone.

17 General introduction Social challenge hypothesis Social events are usually associated with a temporary surge in testosterone levels. The social challenge hypothesis (Wingfield et al., 1990), originally based on testosterone and aggression associations in monogamous birds (Wingfield, Lynn, & Soma, 2001) and later also established in primates (Muller & Wrangham, 2004) and humans (Bateup, Booth, Shirtcliff, & Granger, 2002; Neave & Wolfson, 2003), is the most predominant theory of testosterone reactivity. It states that testosterone levels rise in preparation to a challenging encounter in which social status might be threatened, such as giving a speech in front of a public, and thereby initiating approach motivation and reducing fear (Archer, 2006; Bos, Panksepp, et al., 2012). Following this hypothesis, a rise in testosterone levels in preparation to public speaking, may stimulate Rose to fully approach this challenging situation, rather than using safety behaviors or avoiding it completely. Consistent with this hypothesis, in both animal and human studies, low levels of endogenous testosterone have been linked to socially submissive, anxious, and avoidant behaviors (Archer, 2006; Josephs, Sellers, Newman, & Mehta, 2006; Sapolsky, 1991), whereas high basal testosterone levels are related to social dominance and approach behavior (Maner, Miller, Schmidt, & Eckel, 2008; Mazur & Booth, 1998). Importantly, reduced levels of endogenous testosterone have been found in those suffering from SAD (Giltay et al., 2012) and other social avoidance-related disorders such as depression (Almeida, Yeap, Hankey, Jamrozik, & Flicker, 2008; Giltay et al., 2012). The threat-approach facilitating properties of testosterone have been linked to its effects on the amygdala (and its connectivity with the prefrontal cortex, PFC) and striatum: biasing the amygdala toward reward anticipation and threat approach (see figure 1.2B, Hermans et al., 2010; Volman, Toni, Verhagen, & Roelofs, 2011; Radke et al., 2015). Testosterone as a possible enhancer of exposure therapy in SAD Experimental studies using testosterone administration The social motivational enhancing effects of testosterone have been established not only by correlational but also by more causal testosterone administration studies. For example, administration of a single dose testosterone (0.5 mg sublingual vs placebo) to healthy (female) participants prior to exposure to a threat cue, has been shown to reduce fear, to enhance reward sensitivity and to promote social approach motivation (Bos, van Honk, Ramsey, Stein, & Hermans, 2012; Enter et al., 2014; Terburg et al., 2016). Critically, administration of a single dose of testosterone (0.5 mg sublingually versus placebo), specifically in females with SAD, prior to an eye-tracking experiment, result1

18 Chapter 1 ed in alleviation of gaze avoidance toward angry facial expressions in individuals with SAD (Enter, Terburg, et al., 2016). Participants showed less aversion of gaze towards the eye-regions of negative facial expressions after testosterone versus placebo. Moreover, testosterone administration led to increased approach behavior toward social threat (e.g., angry faces) on a social approach avoidance task (Enter, Spinhoven, et al., 2016). Finally in an EEG study, it was found to result in reduced automatic threat processing of angry faces in individuals with SAD versus healthy controls (van Peer, Enter, van Steenbergen, Spinhoven, & Roelofs, 2017). In light of these consistently established prosocial and approach enhancing properties of testosterone in general and specifically in SAD, testosterone may be a potential candidate to boost exposure effects in SAD, by targeting within session avoidance behavior. This is exactly what I aim to investigate using the current dissertation. Boosting treatment effectiveness by means of testosterone would extend a vast line of research conducted over the past two decades on pharmacological enhancement of CBT. Among potential pharmacological enhancement methods, various randomized controlled trials have found most support for D-cycloserine (DCS), a partial N-methyl-D-Aspartate (NMDA) receptor agonist, associated with fear extinction consolidation (Hofmann et al., 2006; Rodebaugh, Levinson, & Lenze, 2013; Smits et al., 2020). Additionally, few studies examined effects of pharmacological enhancers such as Yohimbine and Oxytocin in SAD (Guastella et al., 2009; Smits, Rosenfield, Davis, et al., 2013). These enhancement strategies all aimed to target the process of (extinction) learning in SAD. Although the results are encouraging, none of these pharmacological enhancers directly acts on acute within-session social-approach behavior, essential for effective exposure. This brings me to the next section, on the potential mechanism of action of testosterone in exposure therapy. Testosterone as a potential enhancer of exposure therapy As mentioned, one of the proposed mechanisms of action of exposure is that individuals with SAD learn that exposure to a feared stimulus (e.g., a social situation) does not lead to their feared outcome (social rejection). To learn this, it is needed that individuals with SAD approach the feared situation, rather than avoiding it. That is, they have to approach the social interaction and have to get engaged in it. The social challenge hypothesis (Wingfield et al., 1990, 2001) states that testosterone levels rise in preparation to a socially challenging situation (such as exposure) and thereby initiates approach motivation. Testosterone reactivity may therefore be important for successful exposure treatment by stimulating within-session approach behavior. Crucially, testosterone reactivity can be experimentally manipulated for example by testosterone administration, resulting in social approach behaviors in individuals with SAD (Enter, Spinhoven, et al.,

19 General introduction 2016; Enter, Terburg, et al., 2016). Based on these fundamental and experimental findings on testosterone, we expect that testosterone reactivity (or administration) prior to an exposure session improves engagement and approach behavior within the exposure session, resulting in improved corrective learning, as assessed by retention of learning in the following exposure session(s) (see Figure 1.3 for an illustration). Figure 1.3 An illustration of the proposed mechanism of action for testosterone as an enhancer for exposure therapy. A. The upper part of the figure depicts subtle avoidance and safety behaviors during a speech exposure, such as holding notes, talking quietly, hiding face behind hair etc. According to Inhibitory Learning Theory this negatively affects inhibitory learning, thereby reducing the efficacy of exposure treatment. If anything, individual learns confirmation that absence of social rejection can be attributed to safety behaviors, leading to maintained fear for the next exposure session. B. The lower part of the picture illustrates our hypothesis that Testosterone reactivity or administration stimulates within-session approach behavior and engagement: no notes, talking at higher volume, open posture. We expect that reducing subtle avoidance and safety behaviors in individuals with SAD, results in more effective exposure and improved corrective learning. Note. The dotted lines reflect hypotheses rather than established findings. Aim of this dissertation To summarize, SAD is one of the most common anxiety disorders with detrimental consequences when left untreated. It is characterized by avoidance behavior, which is the core target of exposure therapy. Although efficacious, the therapy leaves ample room for improvement (response rates vary between 45–55%). Considering the anxiolytic, avoid1

20 Chapter 1 ance alleviating, and prosocial properties of testosterone, testosterone might have the potential to enhance exposure treatment efficacy for SAD. Therefore, the overarching aim of this dissertation is to examine whether endogenous or exogenous testosterone increases can enhance exposure efficacy for SAD. Specifically, the first aim is to review the current scientific knowledge on social motivational properties of the HPG-axis and its potential role in social motivational deficiencies underlying affective disorders, such as SAD. Second, we aim to test 1) whether endogenous testosterone is predictive of exposure outcomes, 2) if administrating testosterone to individuals with SAD prior to exposure can improve exposure efficacy and 3), whether automatic avoidance behavior toward social stimuli may moderate the effects of exposure enhancement with testosterone. Outline of this dissertation Chapter 2 presents a theoretical overview of steroid hormones testosterone and cortisol and their relationship with social motivational behavior and psychopathology such as aggression related disorders and SAD in specific. Chapter 3 describes a proof-ofconcept study in which we translated the social challenge hypothesis of testosterone into the clinical practice. In a sample of 73 participants with SAD, this study sought to test whether endogenous pre-treatment testosterone increases, enhances efficacy of a standardized exposure therapy session for SAD, as measured by fear levels during exposure and change in social anxiety symptoms following one standardized exposure session. In chapter 4 the results of a placebo controlled randomized proof-of-concept trial are presented. Concretely, we tested the augmentative potential of administrating one dose of testosterone (0.5 mg sublingual, vs placebo) prior to a speech exposure session for females with SAD (N = 55). Within session fear and social anxiety symptoms were the primary and secondary outcome measures, respectively. In chapter 5 we tested the hypothesis that highly avoidant participants benefit more from the testosterone-enhanced therapy described in chapter 4. We measured pre-treatment automatic avoidance tendencies toward social stimuli with an approach-avoidance joystick task in the same sample as the study described in chapter 4 and tested if these tendencies moderate the effects of testosterone enhanced exposure. Additionally, we tested whether these avoidance tendencies are relatively stable or whether they vary with (testosterone enhanced) exposure efficacy. Finally, I will close with a discussion in chapter 6, presenting an overview and discussion of all findings in light of the existing literature, followed by an evaluation of strengths, limitations and implications for the clinical practice, as well as considerations for future research and concluding remarks.

Chapter 2 Neuroendocrinological aspects of social anxiety and aggression related disorders Enter, D., Hutschemaekers, M. H., & Roelofs, K. (2018). Neuroendocrinological aspects of social anxiety and aggression-related disorders. In Routledge international handbook of social neuroendocrinology (pp. 635-655). Routledge.

25 Neuroendocrinological aspects of social anxiety and aggression related disorders Introduction Steroid hormones, like cortisol and testosterone, play an important role in the regulation of social motivational behavior. Whereas testosterone facilitates threat approach, presumably by facilitating dopaminergic projection from the amygdala to the striatum (de Souza Silva, Buddenberg, Huston, Topic, & Mattern, 2008; Hermans et al., 2010; Radke et al., 2015), cortisol increases threat avoidance, particularly in high socially anxious individuals (van Peer et al., 2007; van Peer, Spinhoven, Dijk, & Roelofs, 2009). Interestingly, social motivational disorders, such as social anxiety, and aggression-related disorders show an imbalance in these steroid hormones: social anxiety has been associated with increased cortisol stress-responses and decreased testosterone levels (Gerra et al., 2000; Giltay et al., 2012; Roelofs, Minelli, Mars, van Peer, & Toni, 2009), while aggressive psychopathologies have been linked to increased testosterone levels (Glenn, Raine, Schug, Gao, & Granger, 2012; Montoya, Terburg, Bos, & van Honk, 2012; Volman et al., 2016). In this chapter, we discuss the role of these steroid hormones and the neuropeptide oxytocin in social psychopathologies, especially social anxiety and psychopathy. First, we will give a description of the neuroendocrine aspects of social motivational behavior, including social approach and avoidance behaviors. Then we will focus on the neuroendocrine aspects of social anxiety and aggression-related disorders. Finally, motivational and psychiatric findings will be integrated, followed by a research agenda, aiming to provide starting points for clinical applications. Social motivational action The term motivation reflects a broad concept related to anything that may prompt the person to act in a certain way, or to develop an inclination for specific behavior. In this chapter though, we will focus largely on social motivational actions that can be roughly divided into social approach and social avoidance (Davidson, Ekman, Saron, Senulis, & Friesen, 1990; Gray, 1994). These action tendencies involve a basic response to stimulus valence. They are mediated by primary motivational systems of the brain -whereby reward potentiates behavioral activation, while punishment promotes behavioral inhibition or avoidance - and are thought to underlie every complex emotional responding (Carver & White, 1994; Gray & MacNaughton, 2003). Successful social functioning depends on adaptive regulation of these social approach and avoidance responses. Both automatic defensive action tendencies and more instrumental (or goal-directed) mechanisms shape an individual’s behavior. When an individual encounters a social 2

26 Chapter 2 stimulus (e.g., an angry facial expression directed at him/her), he/she will engage in an automatic defensive freeze and flight-or-fight response, a quick and automatic sequence of defensive responses stages (Bradley, Codispoti, Cuthbert, & Lang, 2001). During threat exposure in particular, an initial freezing response is activated during which the individual ceases all ongoing activity and perception is enhanced to quickly assess the situation in order to optimize subsequent fight-or flight responses (Blanchard, Griebel, Pobbe, & Blanchard, 2011; Lojowska, Gladwin, Hermans, & Roelofs, 2015; Roelofs, Hagenaars, & Stins, 2010). This is an automatic process, and the evaluation directly results in a behavioral disposition towards the stimulus: aversive stimuli generally elicit the tendency to move away from the stimulus and appetitive stimuli will elicit a tendency to move towards the stimulus (Lang, Bradley, & Cuthbert, 1997). Such automatic tendencies can also influence more complex, instrumental approach–avoidance decision making (Geurts, Huys, den Ouden, & Cools, 2013; Guitart-Masip, Duzel, Dolan, & Dayan, 2014). For instance, Ly, and colleagues (2014) tested such influence in 45 healthy human individuals using an experimental set-up in which automatic freezing reactions towards negatively (versus positively) valenced stimuli were disentangled from instrumental approach–avoidance decisions (guided by monetary rewards and punishments). Critically, the transfer of valence (and related automatic reactions) to the instrumental approach–avoidance actions were systematically tested. The valence of angry (versus happy) faces was indeed found to transfer to instrumental decision making, in such a way that it induced an instrumental avoidance bias. The extent of freezing elicited by the angry faces was significantly correlated to the instrumental avoidance bias. Both automatic freeze–fight–flight tendencies and more instrumental approach and avoidance biases have been suggested to play a prominent role in the maintenance and perhaps even cause of psychopathology (Blanchard et al., 2011; Rudaz, Ledermann, Margraf, Becker, & Craske, 2017; Turk, Lerner, Heimberg, & Rapee, 2001; Wong & Moulds, 2011). Aggression, for instance, has been conceptualized as a defensive response system in which automatic fight - responses are triggered too easily and in which instrumental threat– approach tendencies become well-learned and rewarded (Blair, 2013; Blanchard et al., 2011; Ly et al., 2016). On the contrary, persistent avoidance in anxiety disorders has been thought of as a defensive response system in which automatic flight –response are easily triggered and in which instrumental threat– avoidance tendencies become rewarded and well learned (Blanchard et al., 2011). Rolls, (2000) emphasized the importance of facial expressions as input for these systems, as they convey social information. When applied in social approach–avoidance tasks (AATs), healthy people show a general tendency to move away from angry expressions and to approach happy faces (Bradley et al., 2001; Chen & Bargh, 1999; Heuer et

27 Neuroendocrinological aspects of social anxiety and aggression related disorders al., 2007; Roelofs, Minelli, et al., 2009; Inge Volman, Toni, et al., 2011). Social AATs using emotional faces have therefore been used to objectively measure the motor responses that are brought about by the automatic and instrumentally driven tendency to approach or avoid a certain stimulus (Chen & Bargh, 1999; Heuer et al., 2007; Roelofs, Elzinga, & Rotteveel, 2005; Rotteveel & Phaf, 2004). A commonly used type is a manual reaction time task which requires participants to approach and to avoid socially appetitive and aversive visually presented stimuli (happy and angry faces, respectively) by pulling (approach) or pushing away a joystick (avoidance) (see Figure 2.2E). In zooming versions of the AAT, pulling or pushing the joystick increases or decreases the size of the picture respectively, giving the impression of moving towards or moving away from the participant (Heuer et al., 2007). Affect–behavior congruence (i.e., approaching happy or avoiding angry faces) leads to quicker responses than when automatic tendencies need to be overridden, as is the case with affect–behavior incongruence (i.e., approaching angry or avoiding happy faces). Highly socially anxious individuals have been shown to avoid socially threatening (i.e., angry) faces, compared to low anxious controls (Heuer et al., 2007; Roelofs, Putman, et al., 2010), while psychopathic offenders show diminished avoidance tendencies of angry faces, compared to controls (von Borries et al., 2012). Neurobiology underlying social motivational behavior Approach and avoidance-related behaviors are mediated by complex interacting neural networks, which can be categorized in the so-called emotional network, reward network, and cognitive control network (Cremers & Roelofs, 2016), which will be broadly described hereafter. The amygdala plays a central role in the emotional network; its subnuclei process salient information from the environment, such as emotional facial expressions, and trigger behavioral responses in response to these environmental stimuli. The basolateral amygdala (BLA) receives input from the thalamus and sensory cortices (such as fusiform gyrus, involved in face processing), whereas the central amygdala (CeA) orchestrates autonomic responses by projections to the periaqueductal gray (PAG) initiating freeze, to brainstem nuclei for release of neurotransmitters, and the hypothalamus for release of oxytocin, corticotropin releasing hormone (CRH), and gonadotropin releasing hormone (GnRH). This eventually leads to enhanced cortisol and testosterone levels, respectively. The amygdala is also connected to the reward network, which comprises the ventral tegmental area (VTA), striatum (including the nucleus accumbens (NAcc)), and medial prefrontal cortex (mPFC) (Haber & Knutson, 2010). Striatal dopamine transmission is essential for the adaptive regulation of social behavior as it is involved in reward learning 2

28 Chapter 2 (i.e., obtaining social reward but also avoiding punishment; see Delgado, (2009), behavioral activation, and motivational behavior (Cools, 2008; Yacubian & Büchel, 2009). The anterior prefrontal cortex plays a crucial role in the cognitive control network as it is involved in the regulation of emotion (Damásio, 1994; Rolls, 1999). It also has a role in social motivational behavior as it inhibits the amygdala, making it possible to control and override automatic behavioral approach and avoidance tendencies (Roelofs, Minelli, et al., 2009; Volman, Roelofs, Koch, Verhagen, & Toni, 2011). Furthermore, it modulates mesolimbic striatal activity (Grace, Floresco, Goto, & Lodge, 2007; Wager, Davidson, Hughes, Lindquist, & Ochsner, 2008). Naturally, this description is a highly simplified one, and many other brain regions partake in these networks (Cremers & Roelofs, 2016). Hormonal regulation of social motivational behavior Testosterone The hypothalamus–pituitary–gonadal (HPG) axis with its end product testosterone plays a key role in the neuroendocrine regulation of social motivational behavior in both sexes. Testosterone levels follow a pulsatile, seasonal, and diurnal cycle in which levels are highest upon waking and typically decline by 50% during the day (Dabbs, 1990). Gonadotropin-releasing hormone (GnRH) is secreted from the hypothalamus, which stimulates the production of luteinizing hormone (LH) and follicle-stimulating hormone (FSH) in the pituitary gland, which in turn triggers production of testosterone and estradiol in the gonads (i.e., testes and ovaries). The secreted estradiol and testosterone in turn inhibit the hypothalamus and pituitary, thus forming a negative feedback loop. In addition, small amounts of testosterone are produced in the adrenal cortex and synthesized in the brain from cholesterol and other steroid precursors. Testosterone is able to cross the blood–brain barrier, and besides having (epigenetic) organizational effects on brain structures during pre- and early postnatal development, testosterone also influences emotion, motivation, and behavior later in life (i.e., activational effects; Lombardo et al. (2012; McHendry, Carrier, Hull, & Kabbaj. (2014). Actions of testosterone are brought about directly via androgen receptors but also via metabolites such as estradiol, dihydrotestosterone, and 3-diol, which binds to the aminobutyric acid (GABA-A) receptor (Balthazart & Ball, 2006; Wood, 2008). The effects can either be slow and long-lasting (i.e., hours–days) via a genomic pathway featuring intracellular steroid receptors, or rapid (i.e., seconds–minutes) via membrane-bound (steroid) receptors, which exert non-genomic actions in the cell. Importantly, testosterone acts through a steroid-responsive network which includes the amygdala, hypothalamus,

29 Neuroendocrinological aspects of social anxiety and aggression related disorders hippocampus, and PAG, among other limbic areas (Wood, 1996), and hence influences the flight–fight response. Naturally, testosterone interacts with other neurotransmitters and peptides, such as serotonin (probably via estradiol), vasopressin, oxytocin, and dopamine. With regard to the latter, testosterone enhances dopamine transmission in the mesolimbic system, which in turn can lead to increased reward sensitivity and augmented motivational behavior by promoting dopaminergic projections form the amygdala to the striatum (de Souza Silva et al., 2009; Hermans et al., 2010; Welker, Gruber, & Mehta, 2015). Baseline hormone levels are in general predictive of psychological traits and behavior (Welker et al., 2015), whereas social events are typically associated with a temporary surge or decline in hormone levels (Casto & Edwards, 2016; Maner et al., 2008; Sapolsky, 1991). The social challenge hypothesis states that testosterone levels rise in preparation to a challenging encounter in which social status might be threatened, thereby initiating approach motivation and simultaneously reducing fear (Archer, 2006; Mazur & Booth, 1998; Wingfield et al., 1990). Several studies featuring single-dose testosterone administration, which leads to a transient increase in testosterone levels, to healthy female participants confirmed the causal relationship between testosterone and its effects on the social motivational system. The findings show that testosterone administration reduces fear and sensitivity to threat and punishment, enhances reward sensitivity, and promotes social approach motivation aimed at achieving social status (i.e., social reward; see for a review (Bos, Panksepp, et al., 2012; Enter et al., 2014). These actions have been suggested to be brought about by anxiolytic effects (GABA, androgen receptors; (McHenry et al., 2014) and upregulation of the dopaminergic system (de Souza Silva et al., 2009), in addition to biasing the amygdala towards threat approach (Radke et al., 2015) and reducing prefrontal control over the amygdala (Schutter & van Honk, 2004; van Wingen, Mattern, Verkes, Buitelaar, & Fernández, 2010; Volman, Toni, et al., 2011). Although associated with aggression (Montoya et al., 2012), the effects of testosterone on social motivational behavior depend on social context and individual differences and thus do not entail aggressive behavior per se, but could also lead to prosocial behavior when this is more appropriate to ensure an increase in social status (Boksem et al., 2013; Carré et al., 2017; Eisenegger, Haushofer, & Fehr, 2011; Mehta & Josephs, 2010; Stanton & Schultheiss, 2009; van Honk, Terburg, & Bos, 2011; sample sizes in these studies ranged from n = 54 to n = 121). Cortisol For decades cortisol has been a popular biomarker to index acute and chronic social and psychological stress (Hellhammer, Wüst, & Kudielka, 2009). Individual differences in the 2

30 Chapter 2 diurnal pattern are associated with psychopathology (Adam et al., 2017); however, most research has focused on stress-induced cortisol surges. Like testosterone, this hormone follows a pulsatile and diurnal pattern, in which levels are high in the morning, surging within 30–40 minutes after waking, followed by a steep drop for a few hours and a steady decline until the lowest point at bedtime. Cortisol is the end product of the hypothalamus–pituitary–adrenal (HPA) axis. The hypothalamus secretes corticotropin-releasing hormone (CRH), which stimulates the anterior pituitary to release adrenocorticotropic hormone (ACH); this travels via the bloodstream to the adrenal cortex where it stimulates the production of cortisol. Cortisol in turn inhibits the pituitary and the hypothalamus, forming a negative feedback loop, and is able to exert both rapid non-genomic and slow genomic effects in the brain (Joëls, Pu, Wiegert, Oitzl, & Krugers, 2006). Cortisol binds to glucocorticoid and mineralocorticoid receptors in brain areas important in regulating the fight–flight response, such as frontal areas, amygdala, and hippocampus (Lupien, Maheu, Tu, Fiocco, & Schramek, 2007). It has an important role in regulating homeostatic systems, affecting arousal, metabolic processes, and the immune system (Sapolsky, Romero, & Munck, 2000). During the initial phase of the stress response, epinephrine from the adrenal medulla triggers norepinephrine release in the basolateral amygdala, among other regions, which induces an increase in vigilance by prioritizing sensory processing and activation of the amygdala (Osborne, Pearson-Leary, & McNay, 2015). Subsequent cortisol release regulates the stress response by downregulating amygdala responsivity and decreasing anxiety-driven selective attention to threat (Henckens, van Wingen, Joëls, & Fernández, 2010, n = 72; Putman & Roelofs, 2011; van Peer et al., 2009, n = 21, small effect sizes), besides affecting activity in areas involved in the planning and execution of motor responses (Montoya, Bos, Terburg, Rosenberger, & van Honk, 2014, n = 20). Animal research has shown that higher cortisol levels are associated with social avoidance behavior (Sapolsky, 1990). Studies featuring stress-induced cortisol surges and cortisol administration in healthy humans extend these findings by showing that elevated levels of cortisol are associated with increased avoidance of social threat on the AAT (Roelofs et al., 2005, n = 22, small to medium effect sizes; van Peer et al., 2007, n = 40, lare effect sizes). The HPG axis works in antagonism with the hypothalamus–pituitary–adrenal (HPA) axis, in such a way that the end product of the latter (i.e., cortisol, released in response to stress) disrupts production and inhibits actions of testosterone, which in turn inhibits the stress-induced activation of the HPA axis at the hypothalamus (Viau, 2002). Both neuroendocrine axes are important in the regulation of social–motivational behavior and show a complex interaction: basically, higher basal cortisol levels, and low testosterone, are associated with social subordination stress and avoidance behavior, whereas higher basal testosterone and low cortisol facilitate social dominance and approach behavior