Autism and mood disorders

Abstract

Individuals with autism experience substantially higher rates of mood problems compared to the general population, which contribute to reduced quality of life and increased mortality through suicide. Here, we reviewed evidence for the clinical presentation, aetiology and therapeutic approaches for mood problems in autism. We identified a lack of validated tools for accurately identifying mood problems in individuals with autism, who may present with ‘atypical’ features (e.g. severe irritability). Risk factors for mood problems in autism appear to be largely overlapping with those identified in the general population, including shared genetic, environmental, cognitive, physiological/neurobiological mechanisms. However, these mechanisms are exacerbated directly/indirectly by lived experiences of autism, including increased vulnerability for chronic stress - often related to social-communication difficulties(/bullying) and sensory sensitivities. Lastly, current therapeutic approaches are based on recommendations for primary mood disorders, with little reference to the neurobiological/cognitive differences associated with autism. Thus, we recommend: 1) the development and validation of (objective) tools to identify mood problems in autism and measure therapeutic efficacy; 2) an interactive approach to investigating aetiologies in large-scale longitudinal studies, integrating different levels of analysis (e.g. cognitive, neurobiological) and lived experience; 3) testing potential treatments through high-quality (e.g. sufficiently powered, blinded) clinical trials, specifically for individuals with autism.

Keywords:

• Autism
• mood
• depression
• bipolar
• mental health

Background

Autism Spectrum Disorder (ASD) is one of the most commonly diagnosed, lifelong neurodevelopmental conditions, with an estimated prevalence of 1.9% (Maenner et al., 2020). ASD is clinically characterised by core features of social-communication difficulties and restricted and repetitive behaviours and interests (RRB; American Psychiatric Association, 2013). These core features are often observed early in development, with median age at diagnosis being 4-5 years (Brett et al., 2016).

Alongside core features, it is increasingly acknowledged that a substantial proportion of individuals with autism experience a range of co-occurring physical, neurodevelopmental and neuropsychiatric symptoms. Major depression is among the most common of these co-occurring symptoms (Moss et al., 2015) and bipolar disorder is also experienced at an elevated rate in ASD, as compared to the general population (Croen et al., 2015).

Moreover, co-occurring mood problems have a significant impact on the wellbeing and outcomes of autistic people, contributing to reduced quality of life across developmental stage (Mason et al., 2018; Oakley et al., 2020) and increased mortality through suicide in those without intellectual disability (Hirvikoski et al., 2016; Zahid & Upthegrove, 2017). Indeed, as in the general population, depression has been shown to predict suicidal ideation and attempts by autistic people (including in children; Mayes et al., 2013; McDonnell et al., 2020; Richa et al., 2014) – as well as predict other factors associated with suicide, such as thwarted belongingness and perceived burdensomeness (Pelton & Cassidy, 2017).

Co-occurring mood problems also impact on the family members of individuals with autism, with one study of 192 families demonstrating that mood problems and their associated unmet care need were the strongest predictors of caregiver burden, rather than autism itself (Cadman et al., 2012). Improving understanding of co-occurring mood symptoms in autism and developing more effective interventions for managing these symptoms has, therefore, been declared a leading priority for research (Autistica, 2017; Pellicano et al., 2014).

Hence, we reviewed existing evidence on the prevalence and clinical presentation of mood disorders in ASD and their potential aetiological pathways, including genetic, environmental, cognitive/social and physiological/neurobiological mechanisms. In addition, we assessed the current recommended interventions for managing mood problems in autism - which we argue are hampered by a lack of high-quality randomised controlled trials – and propose critical directions for future research and clinical practice (please see Table 2).

Table 2. Key recommendations for research and clinical practice, based on the evidence reviewed.

Recommendations for research	Recommendations for clinical practice
The development and validation of reliable assessment tools to accurately identify mood problems across the autism spectrum (also for use as clinical outcome measures in intervention studies).	The implementation of routine mental health screening when working with individuals with autism – not only on initial assessment, but regularly over time to identify changes in cognitions/ behaviour that may indicate mood problems.
Methodologically rigorous, large-scale longitudinal studies and data sharing initiatives to identify individual variability in the presentation and risk/ protective mechanisms associated with mood problems across all levels of functioning (e.g. genetic, clinical, cognitive, behavioural, social, physiological/ neurobiological, lived experience) and including individuals with autism diverse in sex, age and cognitive ability.	The routine delivery of high quality specialist training and resources for clinical practitioners, carers and other professionals working closely with individuals with autism to promote the early identification and successful management of mood problems, using relevant modifications where necessary (e.g. psychoeducation, incorporating stress-management techniques).
High quality randomised controlled trials to assess the efficacy of both pharmacological, and particularly psychological, treatments for mood problems in ASD that incorporate broader measures of functioning (e.g. quality of life) and long-term follow-up.	The development of clinical infrastructure for delivering remote, app-based or other digital approaches to support individuals with autism, including those with mild depression symptoms.

Prevalence and characteristics of mood disorders in autism

Prevalence estimates

Major depression and bipolar disorder are among the most common co-occurring psychiatric diagnoses in autism (Moss et al., 2015). Prevalence estimates range from 10-50% for depression (Hollocks et al., 2019; Lai et al., 2019; Wigham et al., 2017) and approximately 5% for bipolar disorder (Lai et al., 2019). These figures are notably higher than those reported in the general population, where prevalence rates of depression and bipolar disorder are estimated at up to 7% and <1%, respectively (Wittchen & Jacobi, 2005).

Variation in prevalence estimates for mood disorders in autism may be driven in part by differences in the demographic composition of research samples, particularly in terms of age, sex and cognitive ability. For instance, the onset of mood problems in the general population is most consistently reported during early adolescence, and with higher rates (or greater chronicity in the case of bipolar) identified in females than males (Andrade et al., 2003; Wittchen & Jacobi, 2005). However, the relationship between mood problems, age and sex in ASD is less clear, partly due to the narrow age range of many previous research samples (e.g. children only) and the under-recruitment of autistic females (Howlin & Moss, 2012; Magiati et al., 2016).

Nevertheless - similar to findings from population-based studies - the onset of depression in autism is most consistently reported during adolescence, with a subsequent increase in severity to early adulthood and a slight decrease in severity into older adulthood (Ghaziuddin et al., 2002; Susan Dickerson Mayes et al., 2011; Uljarević et al., 2020). Likewise, for bipolar disorder, some reports suggest that depressive symptoms are more prevalent earlier in development, with mania only becoming apparent in adults with ASD (Frazier et al., 2002; Vannucchi et al., 2014).

There is also evidence for sex differences in the developmental trajectories of mood problems in ASD. For example, a longitudinal study of individuals with autism followed up from ages 9-24 years showed that males reported higher depression symptoms than females during early adolescence (i.e. age 13), but sex differences were absent by early adulthood (i.e. age 21; Gotham et al., 2015). This trend resulted from developmental stability of mood symptoms in males, but a continuing increase in severity for autistic females. Taken together, these studies emphasise that we need to account for the developmental nature of autism to better understand the emergence and maintenance of co-occurring mood problems.

Diagnostic challenges

Although risk for mood disorders is clearly increased in autism, as compared to the general population, there remain several diagnostic challenges. First, mood problems may be diagnostically ‘overshadowed’ by core autism features (Hollocks et al., 2019; Wood & Gadow, 2010). For instance, social withdrawal is an indicator for depression that may instead be attributed to the core social-communication difficulties characteristic of autism. Similarly, sleep problems are highly associated with both mood disorders and autism (Richdale & Schreck, 2009).

A second diagnostic challenge is that symptoms of mood disorders vary widely between individuals and can present ‘atypically’ in some cases. Atypical symptom presentation may partly result from interactions between the core autism phenotype and/or multiple comorbidities altering, masking or exacerbating the expression of mood disorders (Kerns & Kendall, 2012). For example, while many traditional features of depression – like chronic low mood - are observed in ASD, clinical reports also indicate more autism-specific symptom profiles. These symptom profiles include reduced or increased RRB (e.g. reduced engagement with special interests vs. increase in behaviours like skin picking), psychomotor agitation, regression (e.g. loss of speech), reduced self-care, and severe irritability (Stewart et al., 2006).

Likewise, individuals with autism with co-occurring bipolar disorder exhibit similar mood symptoms to those with bipolar but not autism, though with more mixed features (depression and mania) and fewer episodes of euphoric mood (Borue et al., 2016; Sapmaz et al., 2018; Vannucchi et al., 2014). As an example, mood lability, restlessness, irritability/aggression and psychotic symptoms appear to be frequently present in those individuals with autism diagnosed with bipolar disorder. In addition, up to one-third of autistic people with comorbid psychosis present with ‘affective psychosis’, which includes major/manic depressive episodes and/or bipolar disorder with psychotic features (Larson et al., 2017); and some individuals initially presenting with bipolar disorder have been reported to develop symptoms of catatonia over time (Ghaziuddin & Ghaziuddin, 2020). Moreover, these diagnostic complications can lead to the incorrect diagnosis of psychotic/personality disorders (Skeppar & Fitzgerald, 2013; Vannucchi et al., 2014), elevated antipsychotic prescription and more frequent inpatient hospital (re)admissions (Charlot et al., 2008; Etyemez et al., 2020; Righi et al., 2018).

Last, diagnosis of mood disorders in autism is also hampered by the use of assessment tools that are largely based on criteria developed and validated in the general population. It is therefore unclear how appropriate such tools are for indexing mood disorders in autism (Cassidy et al., 2018). Additionally, many existing measures rely on self-report, which can be challenging for some individuals with autism, who may have particular difficulties identifying and describing internal emotions and sensations (i.e. ‘alexithymia’/reduced interoception; Kinnaird et al., 2019; Murphy et al., 2017) or who are minimally verbal, have other speech and language difficulties and/or co-occurring intellectual disability.

Indeed, though at least 30% of individuals with autism have a co-occurring intellectual disability (MacKay et al., 2018), their representation in the literature remains strikingly low. Existing reports suggest a trend towards increasing mood problems with higher IQ, though the evidence is mixed (Greenlee et al., 2016; Rai et al., 2018). Reasons posited for the association between elevated mood problems and increasing IQ in autism include higher self-awareness/consciousness of one’s own difficulties; and gaps in services and support for autistic people without intellectual disability - who are reportedly three times more likely to have ‘no daytime activities’, compared to autistic people with intellectual disability (Magnuson & Constantino, 2011; Taylor & Seltzer, 2011). Nevertheless, it is likely that reported associations between mood problems and IQ are confounded by the fact that it is easier to capture mental health problems in those who are more able to report on their experiences.

Furthermore, given the internalised nature of many traditionally defined symptoms of mood disorder, caregiver-reported symptoms are unlikely to accurately reflect the experiences of the autistic person themselves. For example, concordance between self- and parent-reported mood problems appears to decline across development, as some individuals with autism become more independent (Andersen et al., 2017; Davidsson et al., 2017). Furthermore, caregiver ratings are susceptible to bias based on their own mood (Bitsika & Sharpley, 2016b).

To begin to address these diagnostic challenges, it is of paramount importance to ensure consistency in mental health assessment and support, maintain a clear record of changes in behaviour and promote strong communication between clinicians, autistic people and their parents/carers (see also Nicolaidis et al., 2015). A primary goal of each of these approaches is to enhance the early and accurate identification of potentially subtle changes in cognitions and behaviour over time, beyond those that may otherwise be considered ‘normal’ for a particular individual and that could suggest underlying mood problems (Ghaziuddin et al., 2002). Tools like the Disability Distress Assessment – already being utilised in the delivery of physical healthcare for individuals with learning disabilities – provide a long-term record of the mannerisms and behaviours (e.g. vocalisations, habits, postures) of individuals at their own baseline, as compared to when distressed (Regnard et al., 2007). Nonetheless, there are no reliable and valid tools to assess mood disorders in individuals with autism with and without intellectual disabilities, across developmental stages, which represents a notable direction for future research (Cassidy et al., 2018; National Insitute of Health and Care Excellence, 2016).

Aetiological pathways to mood disorders in autism

As mentioned above, the development of interventions to improve mental health outcomes for autistic people has been identified as a primary research priority by the autism community (Autistica, 2017). To accelerate the development of more effective, tailored interventions, it is necessary to identify the aetiological mechanisms that give rise to mood problems in different individuals with autism.

The UK National Institute of Health and Care Excellence currently suggest that the delivery of interventions for co-occurring mental health problems in autism should be largely informed by guidance developed using evidence from the general population (National Institute of Health and Care Excellence, 2011). However, it is not known whether co-occurring mood problems in autism share the same underpinning mechanisms as those identified in the general population (Wood & Gadow, 2010). Thus, in the following sections, we summarise evidence for several potential aetiological mechanisms contributing to mood disorders in the general population – and discuss the extent to which they apply to ASD. This evidence particularly pertains to the role of genetic/familial risk, chronic stress, difficulties with emotion awareness/regulation processes, loneliness and dysregulation to physiological/neurobiological systems.

Genetic and familial risk factors

Major depression and bipolar disorder are both highly heritable, with data from epidemiological and twin studies indicating that approximately 40-50% of the risk for depression and up to 85% of the risk for bipolar is genetic (McGuffin et al., 2003; Nestler et al., 2002; Sullivan et al., 2000). Though genetic factors are strongly associated with mood disorders in the general population, identifying specific genes or genetic variants that directly confer risk is challenging, due to the large number of genes implicated and the relatively small effect of each one of them on risk for individual psychiatric conditions (Smoller et al., 2013). Nevertheless, genetic variants and gene-gene/gene-environment interactions have been shown to modulate vulnerability for mood disorders (Harrison et al., 2018; Jawahar et al., 2019).

In line with research from the general population, genetic vulnerability is also an important aetiological underpinning for mood disorders in ASD (Carroll & Owen, 2009; Ghaziuddin et al., 2002). Family history studies have consistently reported that the family members of individuals with ASD also experience high rates of mood problems, with one group reporting rates of 58.9% and 5.9% for major depression and bipolar disorder, respectively (Mazefsky et al., 2008). It is not possible to wholly parse out the contributions of genetic, as compared to environmental, risk factors from family history studies. Nevertheless, the vast majority of parents report the onset of their own mood disorders as being prior to the birth of their child with autism (Mazefsky et al., 2008), suggesting that environmental factors (i.e. stressors like atypical parent-child reciprocal interactions) cannot fully explain the relationship between parent-proband symptoms.

Indeed, genomic studies indicate that many neurodevelopmental and neuropsychiatric conditions share common genetic variance. Of most relevance, a meta-analysis of data from genome-wide association studies (including over 33,000 individuals with neurodevelopmental/psychiatric conditions across 19 countries) highlighted overlap of polygenic risk between autism and bipolar disorder – though not with major depression (Smoller et al., 2013).

For symptoms of major depression, there is evidence that genetic factors modulate the development and function of neurotransmitter systems that are associated with both core autism features and later developing mood problems (Johnson et al., 2015). For instance, specific genetic variants influencing serotonin activity, including homozygosity for the G allele of the serotonin 2 A receptor gene, have been found to be associated with depression symptom severity in autism (Gadow et al., 2014). This is notable, since platelet hyperserotonemia (elevated circulating serotonin) is the most consistently replicated serotonin-related finding in ASD and may influence the efficacy of treatment with selective serotonin reuptake inhibitors (SSRIs) in some individuals with autism (Anderson et al., 2002; Daly et al., 2019; Hranilovic et al., 2007). Providing some support for this suggestion, individuals with primary major depression who are homozygous for the G allele of the serotonin 2 A receptor gene are reported to be around 18% less likely to respond to SSRIs than those homozygous for the A allele (McMahon et al., 2006). Together, these studies suggest that the genetic factors related to elevated likelihood of autism may also confer risk for developing mood problems and/or result in variation in treatment response.

In addition to serotonin, genetic variants influencing other key neurotransmitter systems have been implicated in mood disorders in ASD. For example, one study reported that both a specific dopamine transporter gene polymorphism (DAT1 intron8) and dopamine D2 receptor variant (rs2283265) were associated with increased parent-rated depression symptom severity in children with autism (Gadow et al., 2014; Gadow et al., 2014). Furthermore, autism, major depression and bipolar disorder have all been reported to associate with variation in genes encoding the oxytocin receptor and altered plasma levels of the hormone oxytocin (both increased and decreased levels have been reported), as compared to neurotypical groups (Cochran et al., 2013; Iovino et al., 2018).

These studies were important first steps for developing understanding of the aetiology of mood problems in autism, but they were often based on relatively small samples (e.g. <150 individuals). Given the genetic (and phenotypic) heterogeneity of autism, the question of genetic overlap between ASD and mood disorders will require further investigation in larger cohorts of individuals who are characterised both by genomic profiles and ‘deep phenotyping’ of clinical features (e.g. the Autism Sharing Initiative). This is necessary because variability in genetic profiles that map onto diverse symptom profiles could help to identify which individuals may benefit most from a particular intervention approach for managing mood problems (Loth et al., 2016).

Environmental and individual risk factors

Though genetic factors are clearly relevant to the aetiology of mood disorders in individuals with and without autism, results from a large-scale study of approximately 6000 population-representative twin pairs revealed only modest genetic overlap between core autism features and mood problems, with stronger associations identified for shared environmental influences (Hallett et al., 2010). This finding emphasises the importance of also considering environmental factors and gene-environment interactions to better understand the aetiology of mood problems in autism, particularly in the case of major depression.

Chronic stress and life events

Exposure to chronic stress is suggested to increase risk for several health-related conditions, including mood problems, in the general population – particularly for individuals with a predisposition (e.g. genetic, physiological, cognitive) towards difficulties responding to or regulating stress (i.e. the ‘diathesis-stress' model; Colodro-Conde et al., 2018; Sapolsky, 2006). Stressors described in relation to mood problems in the general population include everyday demands like travelling to work or interacting with peers, as well as major life events like the death of a loved one, divorce or moving home (Kanner et al., 1981).

Based on evidence regarding stress-related features that may confer risk for depressed mood in the general population, it is possible that disruption in responding to environmental stressors may also occur in ASD, at two interrelated levels. These include: 1) an increased vulnerability for exposure to chronic stress and; 2) difficulties responding to and regulating responses to environmental stressors, also associated with dysregulation to stress-related physiological/neurobiological systems.

First, on average, individuals with autism are more likely to be exposed to chronic stressors and trauma, as compared to their neurotypical counterparts. In terms of chronic stressors, daily environmental demands (e.g. shopping, travelling on public transport) have been shown to be perceived as more stressful (or intense) by those with, compared to without, ASD (Gillott & Standen, 2007; Magnuson & Constantino, 2011). This may partly result from a reduced threshold for arousal and differences in the cognitive processing of environmental demands (e.g. feeling a lack of control over one’s own life and stressors; van Heijst et al., 2020), as indicated by the high rates of irritability and sensory sensitivities identified in ASD.

Further to chronic stressors, many individuals with autism report having been the victim of bullying, with potentially profound consequences, including reduced positive self-evaluation and increased depression symptom severity (Burrows et al., 2017; Hoover & Kaufman, 2018; Rai et al., 2018). Across development, stigmatisation and discrimination represents a substantial barrier for some autistic people in settings like education and the workplace, as well access to healthcare (Crane et al., 2019; Evans-Lacko et al., 2014).

The risk of exposure to adverse life events, such as unemployment, income insufficiency and violence/abuse, is also estimated to be up to two-fold higher for autistic, as compared to non-autistic, individuals (Berg et al., 2016; National Autistic Society, 2016). As in the general population, individuals with ASD who experience cumulative and traumatic adverse life events are at elevated likelihood of also experiencing mood problems (Rigles, 2017; Taylor & Gotham, 2016). Of note, exposure to early adverse events have been identified even prior to birth, with pregnancy complications like maternal medical illness estimated to be more prevalent in mothers of children with ASD, compared to mothers of children with typical development (Beversdorf et al., 2005). Thus, taken together these findings suggest that stressful life events – from chronic daily stressors to discrete traumatic experiences – may be implicated in mood problems in ASD.

Arousal/emotion regulation difficulties

In addition to increased stress exposure, some individuals with autism may also have difficulties with responding to and regulating arousal and stress. The ability to identify and appraise one’s own emotion has been argued to be the crucial first stage of successfully regulating responses to environmental demands or stressors (Gross, 2015). This is because emotional appraisal involves the consideration of, not only one’s current emotion, but also the situation or thoughts that preceded this emotion and which strategies could be implemented to alter these situational factors or thoughts (e.g. through distraction, re-evaluation, behavioural change), if relevant (Gross & Thompson, 2007).

As alluded to earlier in this review, current estimates suggest that difficulties identifying and describing emotion (i.e. ‘alexithymia’) affect at least 40-65% of individuals with autism, as compared to approximately 10% of individuals in the general population (Bird & Cook, 2013; Kinnaird et al., 2019). Alexithymia may hinder the implementation of a range of adaptive emotion regulation strategies, particularly situation selection or modification, whereby individuals find it more challenging to identify the situations and contexts that trigger specific negative emotions (Gross & Thompson, 2007). In support of this, alexithymia has been found to be indirectly associated with concurrent depression severity in ASD, via elevated emotion regulation difficulties (Morie et al., 2019).

Attentional and cognitive processes implicated in arousal and emotion regulation may also increase vulnerability for mood problems in autism, similarly to mechanisms identified the general population (Beck, 1976). For example, in preferential viewing paradigms, both autistic and non-autistic depressed individuals have been found to orient faster to negative emotional stimuli (e.g. sad vs. happy faces) and spend less time attending to positive stimuli, on average, as compared to never-depressed individuals (Unruh et al., 2018). However, as with executive functioning differences, the presence and effect sizes of relationships between attentional biases and mood problems in autism are inconsistent (Bergman et al., 2020; Hollocks et al., 2014).

More robust effects have been identified for associations between rumination – the tendency to dwell on negative past events or cognitions – and mood problems in ASD. For instance, higher self-reported general and anger rumination have been shown to correlate with increased depression symptom severity in both adolescents and adults with autism (Gotham et al., 2014; Patel et al., 2017). It should be acknowledged that, to date, studies assessing associations between attentional and cognitive biases with symptoms of mood disorder in ASD have predominantly been cross-sectional in design. Therefore, prospective longitudinal modelling is necessary in order to test whether these biases may precede mood problems, or are a feature of current depressed state. For example, it is possible that during episodes of acute depression, some individuals have a more negative outlook on their experiences and internal thoughts than in periods between depressive episodes.

Having said this, broader cognitive differences associated with autism may increase the likelihood of rumination and thus represent a precursor, rather than consequence of mood problems. For instance, rigid thinking patterns and perseverative tendencies represent components of RRB that are a core feature of autism, which may extend to perseverative attention to/thoughts about negative events (Keenan et al., 2018). Furthermore, reduced inhibitory control, experienced by some individuals with autism, may exacerbate repetitive negative cognitions (Geurts et al., 2014) and impede the ability to engage in adaptive emotion regulation strategies like attentional redeployment or cognitive restructuring (i.e. the ability to ‘switch’ to, or become distracted by, other thoughts or activities).

Overall, the evidence above suggests that heightened exposure to (and difficulties responding to and regulating) chronic environmental stressors likely constitutes another aetiological pathway to mood problems for individuals with autism. As noted above, this pathway is likely to be complex and include multiple, interrelated physiological, neurobiological, environmental and cognitive mechanisms. Nevertheless, interventions that bolster coping and regulation strategies for managing everyday stressors, across different contexts, need to be tested.

Social processes

Finally the ability to describe or share emotions with others can facilitate extrinsic emotion regulation – for example, by eliciting support and comfort from caregivers or friends. Indeed, social support has been posited as an important buffer against stress in the wider population (Cohen & Wills, 1985) and for individuals with autism, specifically (Bishop-Fitzpatrick et al., 2018; Hollocks et al., 2014; Mason et al., 2018; Renty & Roeyers, 2006).

Conversely, loneliness has been identified as a significant risk factor for depression symptoms in population-based studies (Cacioppo et al., 2006) and research is now beginning to suggest that the same may be true in the case of autism (Smith & White, 2020). For example, loneliness has been shown to be associated with higher depression symptom severity in adults with autism, as well as non-autistic depressed individuals (Hedley et al., 2018; Hedley et al., 2018; Mazurek & Kanne, 2010).

As mentioned earlier, bullying and peer rejection may be one factor contributing to feelings of loneliness. For example, a qualitative investigation of responses to bullying among adolescents with autism reported that some individuals developed a lasting mistrust in other people, especially if they did not identify at least one friend to turn to for support (Humphrey & Symes, 2010). Relatedly, though many individuals with autism report friendships, some research suggests that self-reported friendship quality – as defined by factors like feeling cared for and low conflict/betrayal – may be lower for autistic than neurotypical individuals, on average (Mazurek & Kanne, 2010; Whitehouse et al., 2009).

Aside from peer problems, social motivational differences often associated with autism have also been suggested to increase risk for loneliness. For example, motivational differences have been observed in some individuals with autism when responding to both social and non-social ‘reward’ stimuli (e.g. happy faces vs. financial incentives; Chevallier et al., 2012; Dawson et al., 2005; Kohls et al., 2011). Reduced social motivational processes have been suggested to result in reduced pleasure during social interactions and, consequently, increased social withdrawal and loneliness for some individuals with autism (Bitsika & Sharpley 2016a; Han et al., 2019).

Nevertheless, in some cases, anxiety and sensory processing differences also commonly associated with autism may result in social withdrawal and loneliness, rather than motivational factors. For instance, high levels of anxiety in social situations and/or sensory sensitivities in social environments, may lead to social withdrawal, which in turn exacerbates social anxieties (Bellini, 2006; Spain et al., 2018).

Physiological and neurobiological risk factors

In the previous two sections, we provided an overview of existing literature pertaining to genetic and familial, as well as environmental and individual risk factors for mood disorders in autism. Evidence from this literature suggests that genetic factors influencing early neurodevelopmental trajectories may increase the likelihood of later mood problems in some individuals with autism. Furthermore, an increased vulnerability for exposure to chronic stressors and difficulties responding to and regulating responses to stress may also contribute to the high rates of mood problems observed in ASD. Notably, interactions between physiological/neurobiological and environmental risk factors for mood problems must also be taken into account. Of particular importance, chronic stress, or ‘allostatic load’, is associated with disruptions to physiological and neurobiological systems that are trans-diagnostically implicated in mental health outcomes in the general population (Holzman & Bridgett, 2017; McEwen & Stellar, 1993; National Institute of Mental Health, 2019).

Similarly, stress-related disruptions to physiological and neurobiological pathways have been identified in ASD (Patriquin et al., 2019). Dysregulation of the hypothalamic pituitary adrenal (HPA) axis in some individuals with autism is a primary example of a stress-related physiological profile that may be associated with mood problems. Briefly, the HPA-axis responds to stressors by promoting the release of corticotrophin releasing hormone from the hypothalamus, which in turn triggers adrenocorticotropin releasing hormone from the anterior pituitary gland and lastly cortisol from the adrenal glands (Smith & Vale, 2006). Both hypo- and hyper-activation of the HPA-axis have been reported in ASD, as compared to the neurotypical population, generally suggestive of HPA-dysregulation in response to stress (Taylor & Corbett, 2014). Hypo-activation of HPA-axis function, as assessed by salivary cortisol levels, has been shown to correlate with increased self-reported depression symptom severity in autistic girls, and parent-reported symptoms in autistic boys (Bitsika et al., 2016; Sharpley et al., 2016). Of importance, while hyper-activation of the HPA-axis can reflect acute stress exposure, hypo-activation may be indicative of chronic stress, as prolonged stress hormone release (e.g. cortisol) induces a compensatory down-regulation, or ‘blunting’, of HPA function (Hollocks et al., 2014).

Additionally, stress-related autonomic nervous system (ANS) dysregulation has also been reported in autism and associated with mood problems (Klusek et al., 2015). The ANS is fundamental for regulating all organs and tissues of the human body (except skeletal muscles; Wehrwein et al., 2016) and is comprised of two major divisions: the sympathetic pathway, implicated in heightened arousal states in response to environmental stressors (e.g. increased heart and breathing rate, pupil dilation); and the parasympathetic pathway, implicated in resting state processes (e.g. reduced heart and breathing rate, digestion; Karemaker, 2017). As with HPA-axis activity, there is no uniform ‘autonomic profile of autism’. However, where autism-comparison group differences have been observed, evidence to date suggests that autism is most consistently characterised by ANS hyper-arousal or sympathetic dominance (Klusek et al., 2015). Indices of sympathetic dominance – particularly reduced heart rate variability – have been found to associate with increased depressive symptoms in samples of autistic children (Muscatello et al., 2020; Neuhaus et al., 2014). Nevertheless, associations between ANS dysregulation and mood problems have not yet been confirmed in adult samples (Cai et al., 2019). Moreover, while other physiological and neurobiological profiles associated with autism overlap with those identified in mood disorders, including reward network differences and elevated proinflammatory markers, research to comprehensively characterise these features in cohorts of individuals with autism is undoubtedly lacking (Chevallier et al., 2012; Mitchell & Goldstein, 2014; Schwarz et al., 2020).

Summary of aetiological factors

The aetiological pathways for mood disorders (predominantly depression) in autism clearly involve several interacting genetic, cognitive, social/behavioural systems and physiological/neurobiological mechanisms (see also White et al., 2018). Many of these mechanisms are largely overlapping with those identified in the general population, including genetic factors modulating neurodevelopment and neurotransmitter activity, chronic stress, difficulties with emotion awareness/regulation processes, loneliness and stress-related disruption to physiological and neurobiological networks.

However, it is also true that individuals with autism experience unique risk factors. For instance, autism itself may be associated with greater exposure (e.g. different neurodevelopmental trajectories, reduced threshold for arousal and stress, perseverative thinking) and/or increased vulnerability for stressors and life events (e.g. bullying; Au-Yeung et al., 2019). This is important, given longstanding questions around whether comorbid psychiatric conditions share the same underpinning mechanisms as in the general population, and therefore whether ‘true’ comorbidity is possible in the case of autism (Wood & Gadow, 2010). The evidence to date indicates that existing diagnostic and intervention pathways are likely to be applicable (albeit in a modified form) for some autistic people.

Nevertheless, to best inform diagnostic and intervention approaches, it is also critical to investigate individual variability in the presentation and risk factors for mood problems in autism, which has rarely been reported in previous studies. Acknowledging individual variability is of primary importance for developing more individualised, needs-based (as opposed to diagnosis-based) detection and intervention approaches for mood problems – identifying who may benefit most from specific interventions and minimising unwanted side effects and disruption to compensatory mechanisms (Ure et al., 2018).

Thus, an important priority for future research is to investigate the mechanisms associated with mood problems in multi-diagnostic cohorts, including individuals with ASD and those with primary mood disorders. This approach is valuable because focussing on single diagnostic groups likely oversimplifies our understanding of symptom profiles and their underpinning mechanisms, when comorbidity between neurodevelopmental and neuropsychiatric conditions is the rule, rather than the exception. This is especially true of bipolar disorder, for which psychiatric multimorbidity has been posited as a marker in youth (McElroy, 2004). Identifying symptom profiles and contributing mechanisms that are shared across clinical groups (i.e. transdiagnostic features), as compared to those unique to a particular group (e.g. ASD), could indicate which existing intervention approaches are likely to be effective in ASD and where modifications or novel intervention approaches are most required.

Interventions for mood disorders in autism

As noted earlier, current recommended intervention approaches for managing mood problems are largely based on evidence from the general population. This is particularly concerning in the case of pharmacological treatment, since there is limited understanding of the underlying neurophysiology(ies) of autism, limited evidence of the effectiveness of existing compounds in ASD populations and limited translation of findings for novel compounds in animal/cellular models to human studies (Ghosh et al., 2013; King et al., 2009). Moreover, individuals with autism may be at elevated risk of suffering from the side effects of medication (Howes et al., 2018; Williams et al., 2013).

In terms of psychological and psychosocial interventions, treatment targets identified from general population research (e.g. negative cognitions/behaviours, self-awareness, emotion regulation) do seem to be relevant for autism and the use of existing approaches like Cognitive Behavioural Therapy (CBT) and mindfulness-based therapies are supported in everyday clinical practice (Benevides et al., 2020; Chandrasekhar & Sikich, 2015). However, due to a considerable lack of high-quality (e.g. sufficiently powered, blinded, follow-up) randomised controlled trials for the treatment of mood problems in autism, it remains unclear which approaches will be most effective for whom and which autism-specific modifications to therapies are most beneficial.

Pharmacological interventions

To date, there has been very little evidence to indicate which pharmacological interventions are effective for managing mood problems in ASD, an issue that reflects broader limitations in the clinical development of therapeutics for autistic people (please see Ghosh et al., 2013). The lack of evidence for the efficacy of pharmacological treatment for mood problems in autism is somewhat surprising, since prescription rates for psychotropic medications in ASD are high (Hsia et al., 2014; Murray et al., 2014) and believed to have increased up to threefold in the US between 1992 and 2001 (Aman et al., 2005; Howes et al., 2018; Maddox et al., 2018).

Current clinical recommendations for the use of selective serotonin reuptake inhibitors (SSRIs) for mood problems, such as sertraline, fluoxetine, citalopram and escitalopram, are largely based on evidence from typically developing groups (Vasa et al., 2016). This is of concern, however, as some individuals with autism have been shown to exhibit different neural responses to pharmacological challenge, as compared to neurotypical individuals (Pretzsch et al., 2019). Moreover, the use of SSRIs in ASD may result in heightened adverse side effects, including agitation, impulsiveness, hyperactivity, stereotypy and insomnia, and it has been suggested that they should therefore only be considered on a ‘case-by-case basis’ (Williams et al., 2013).

For mood instability, evidence for the effectiveness of pharmacological treatment is limited to case reports or case series, despite individuals with autism with comorbid bipolar disorder being at elevated likelihood of concurrently using multiple psychotropic medications (Spencer et al., 2013). One systematic review reported mood stabilisers (lithium, valproate) to be preferable over atypical antipsychotics/stimulants that are associated with greater risk of severe long-term adverse events (Vannucchi et al., 2014). However, the major caveat to this suggestion is that the evidence drawn from case studies is generally based upon individuals who exhibit severe behavioural problems and for whom treatment may not be specifically targeted for managing mood disorder.

Due to a lack of robust evidence for the efficacy of pharmacological interventions and issues regarding safety and side effects, risperidone and aripiprazole are currently the only medications approved for individuals with autism by the US Food and Drug Administration – both for targeting irritability (which may or may not indicate mood problems, as mentioned in the opening section of this review; Howes et al., 2018). Furthermore, approximately 60% of autistic adults and their parents report apprehension over the use of medications that are not scientifically proven to be effective in ASD (Wallace et al., 2013). The complete absence of randomised controlled trials for medications targeted at managing mood problems in ASD is therefore of great concern, given that combination therapy with pharmacological and psychological intervention is reported to confer the greatest improvement in depression symptoms in the wider population (March et al., 2004).

Psychological and psychosocial interventions

In the same way as pharmacological interventions, the effectiveness of psychological and psychosocial interventions for mood problems in autism remains inconclusive, though there has been more research published in this area (see White et al., 2018; Table 1). Cognitive Behavioural Therapy (CBT) is perhaps the most widely documented intervention approach, particularly with reference to the modifications that may be required when working with individuals with autism. Some of the most frequently identified CBT modifications recommended for clinical practice include: specialised therapist training and focus on inter-disciplinary care; incorporating concrete tools and support (e.g. a longer course of highly structured sessions, visual aids, focus on specific behaviours vs. abstract cognitions); addressing core autism features (e.g. focus on interpersonal skills); psychoeducation regarding autism and mood problems; incorporation of special interests; and increased parental involvement (Cooper et al., 2018; Connor Morrow Kerns et al., 2016; Moree & Davis, 2010). As outlined in the previous section, maladaptive thought patterns like rumination have been posited as a key underpinning mechanism for depression symptoms in autism and therefore it seems intuitive that CBT may be a beneficial intervention approach.

Table 1. Design and results of selected psychological/psychosocial intervention studies for mood problems in autism, referenced in this review.

Authors (Date)	Country	Intervention	Frequency, duration, follow-up	Sample size	Sample composition (Intervention/ Control)	Study design	Outcome measure(s)	Treatment effect (pre-post)	Group effect (intervention- control)
Spek et al. (2013)	Netherlands	Group-based mindfulness	9 weekly 150-min sessions (+ 40-60-min daily meditation, 6 days per week)	N = 42	M:F = 13:7/ 14:7; M (SD) age = 44.4 (11.1)/40.11 (11); IQ > 85	Randomised-controlled (treatment as usual)	Depression (SCL-90-R); Anxiety (SCL-90-R); Positive affect (Dutch GMS); Rumination (RRQ)	p < 0.05, d = 078; p < 0.05, d = 0.76; p < 0.05, d = 0.79; p < 0.001, d = 1.25	p = 0.02; p = 0.02; p = 0.02; p < 0.001
Sizoo and Kuiper (2017)	Netherlands	Group-based CBT vs. mindfulness	13 weekly 90-min sessions (+ homework) – follow-up to 26-weeks	N = 59	M:F = 38:21^aM (SD) age = 35.10 (9.22)/ 39.40 (10.81) IQ > 75	Sequential randomisation without control	Depression (HADS); Anxiety (HADS); Positive affect (Dutch GMS); Negative affect (Dutch GMS); Rumination (RRQ); Social (SRS-A)	p < 0.001, d = 0.51(0.73) CBT(mindfulness); p < 0.001, d = 0.82(0.76); p = 0.001, d = 0.40(0.59); p < 0.001, d = 0.38(52); p = 0.003, d = 0.32(0.56); p = 0.02, d = 0.22(0.28)	p = 0.88; p = 0.22; p = 0.47; p = 0.53, p = 0.69; p = 0.12
Gaigg et al. (2020)	UK	CBT vs. mindfulness	Online programme	N = 54	M:F = 8:1 (CBT); 12:2 (mindfulness)/ 11:4 M (SD) age = 40.30 (12.70; CBT); 42.50 (10.30; mindfulness)/ 43.20 (12.70) IQ ≥ 81	Randomised-controlled (waitlist)	Depression (HADS); Generalised anxiety (GAD-7); Social anxiety (LSAS); Trait anxiety (STAI-T); Anxiety (BAI)	p = 0.02^b (CBT); p = 0.003 (mindfulness)	NS (NR); p = 0.15 for anxiety scales
McGillivray and Evert (2014)	Australia	Group-based CBT	9 weekly 120-min sessions	N = 32	M:F = 19:7 / 13:3 M (SD) age = 20.27 (4.39)/ 20.50 (3.40) IQ ≥ 75	Sequential allocation with control (waitlist)	Depression (DASS); Anxiety (DASS); Stress (DASS); Automatic Thoughts (ATQ); Anxious Thoughts (ASSQ)	p < 0.05, d = 0.40; p < 0.01 d = 0.29; p < 0.01 d = 0.48; p < 0.01 d = 0.44; p < 0.05 d = 0.37	p > 0.05 for all scales
Santomauro et al. (2015)	Australia	Group-based CBT	10 weekly 60-min sessions (+ 1 60-min booster session 4 weeks later) – follow-up to 34-weeks	N = 23	M:F = 12:11^a M (SD) age = 16.00 (1.33)/ 15.50 (1.43) IQ ≥ 85	Randomised-controlled (waitlist)	Depression (BDI-II); Depression (DASS); Emotion regulation (ERQ)	NR; p = 0.04, d = 0.80; NR	p = 0.89; p = 0.07; p = 0.71
Russell et al. (2019)	UK	Low intensity CBT (guided self-help)	9 weekly 30-45-min sessions – follow-up to 24-weeks	N = 70	M:F = 24:10/ 27:8 M (SD) age = 35.30 (13.60)/ 40.20 (12.60) IQ NR	Randomised-controlled (treatment as usual)	Depression (PHQ-9); Depression (BDI-II); Depression (GRID-HAM-D-17)	NA – feasibility pilot	NA – feasibility pilot
Conner et al. (2019)	US	Emotion awareness and regulation skills	16 weekly 45-50-min sessions – follow-up to 13-weeks	N = 17	M:F = 15:2^a M (SD) age = 14.94 (1.54)^a IQ ≥ 82	Experimental without control	Depression (PROMIS); Anxiety (PROMIS); Irritability (ABC-I); Emotion regulation (EDI); Stress (RSQ)	p = 0.001, d = 0.96; p = 0.04, d = 0.59; p = 0.01, d = 0.67; p < 0.001, d > 1.77	NA – no control group
Schiltz et al. (2018)	US	Group social skills intervention	16 weekly 90-min sessions	N = 70	M:F = ∼22:2/ 23:2 M (SD) age = 13.25 (1.07)/ 13.52 (1.92) IQ ≥ 68	Randomised-controlled (waitlist)	Depression (CDI); Negative Mood (CDI); Interpersonal Problems (CDI); Ineffectiveness (CDI); Anhedonia (CDI); Negative self-esteem (CDI)	p = 0.01, d = 0.82; p = 0.32, d = 0.79; p = 0.85, d = 0.27; p = 0.34, d = 0.42; p = 0.03, d = 0.55; p = 0.04, d = 0.48	p = 0.01; p = 0.32; p = 0.85; p = 0.52; p = 0.03; p = 0.04

^aData not provided split by group.

^bNot possible to calculate effect size from descriptives. CBT: Cognitive Behavioural Therapy; M:F: Male:Female; SCL-90-R: Symptom Checklist-90-Revised; GMS: Global Mood Scale; RRQ: Rumination Reflection Questionnaire; HADS: Hospital Anxiety and Depression Scale; SRS-A: Social Responsiveness Scale – Adult; GAD-7 : General Anxiety Disorder − 7 items; LSAS: Liebowitz Social Anxiety Scale; STAI-T: State-Trait Anxiety Inventory - Trait; BAI: Beck’s Anxiety Inventory; DASS: Depression Anxiety Stress Scale; ATQ: Automatic Thoughts Questionnaire; ASSQ: Anxious Self-Statements Questionnaire; BDI-II: Beck’s Depression Inventory – Second Edition; ERQ: Emotion Regulation Questionnaire; PHQ-9 : Patient Health Questionnaire − 9 items; GRID-HAM-D-17 : GRID-Hamilton Rating Scale for Depression − 17 items; PROMIS: Patient-Reported Outcomes Measurement Information Systems; ABC-I: Aberrant Behaviour Checklist – Irritability; EDI: Emotion Dysregulation Inventory; RSQ: Response to Stress Questionnaire; CDI: Children’s Depression Inventory. M(SD): Mean (Standard deviation); p: significance value; d: Cohen’s d effect size.

However, it must be noted that high quality randomised controlled trials of CBT for mood problems in ASD are still very sparse (Walters et al., 2016). Some quasi-experimental studies report that individuals with autism allocated to individual or group-based CBT show significantly reduced depression symptoms, including when compared to a waitlist-control group (McGillivray & Evert, 2014; Sizoo & Kuiper, 2017). Conversely, other investigations have yielded no significant effects of CBT on depression symptom severity in ASD, particularly in comparison to alternative treatment approaches like mindfulness (Gaigg et al., 2020; Santomauro et al., 2016). Hence, while CBT for mood problems in autism is often identified as a potentially beneficial intervention approach in everyday clinical practice (Chandrasekhar & Sikich, 2015), the research to practice gap needs to be bridged.

Aside from CBT, mindfulness-based therapies are beginning to be explored for their effectiveness in treating mood problems in ASD. Mindfulness exercises like breathing, relaxation and meditative techniques may exert their beneficial effects in several ways, including stress-reduction, improved awareness of one’s own internal physiological and emotional states and emotion regulation (Pagni et al., 2020). As with CBT, some modifications to mindfulness approaches may be required when working with autistic people, like avoiding ambiguous metaphors and imagery, slower pace of sessions and enhanced planning and routine around mindfulness ‘homework’ outside of sessions (Spek et al., 2013).

In some support for the effectiveness of mindfulness-based therapies, one randomised controlled trial demonstrated a significant reduction in depression symptoms from baseline to post-therapy in autistic adults, as compared to a waitlist-control group, with medium effect sizes of Cohen’s d = 0.76-0.78 (Spek et al., 2013). However, while treatment gains from mindfulness-based approaches have been reported to persist at a longitudinal follow-up of approximately three months, it remains unclear whether improvements persist beyond this period (Kiep et al., 2015). Furthermore, there is no robust evidence comparing the added benefits of modifications to mindfulness-based therapies in ASD, such as the optimal number/length of sessions, individual vs. group therapy and the age range within which mindfulness approaches are most effective (Hartley et al., 2019).

Although CBT and mindfulness are currently the only emerging evidence-based interventions for managing mood problems in autism (Benevides et al., 2020), other psychological and psychosocial interventions are now being developed and piloted. One example is the Emotional Awareness and Skills Enhancement (EASE) program, targeted at improving emotion regulation skills, including attentional redeployment and cognitive restructuring. Following the EASE program, a significant decrease in both self- and parent-reported depression symptoms, of medium-to-large effect size (Cohen’s d = 0.52-0.96), was observed in 20 adolescents with autism (Conner et al., 2019).

Additionally, data from psychosocial programs like the Program for the Education and Enrichment of Relational Skills (PEERS®) also demonstrated pre- to post-intervention declines in depression symptom severity in adolescents with autism, suggested to be driven by increased positive peer interactions (Schiltz et al., 2018). Similar to the potentially protective role of positive peer relationships for reducing loneliness and increasing social support, positive family interactions may also be of relevance for the outcomes of individuals with autism with co-occurring mood problems (Shochet et al., 2019). However, a recent systematic review of the application of family therapy in ASD yielded no studies that met the methodological standards required for inclusion (Spain et al., 2017).

Lastly, the UK National Institute of Health and Care Excellence released a research recommendation to investigate the clinical and cost-effectiveness of facilitated self-help for mild depression symptoms in autistic adults (NICE, 2016). Emerging technologies, including evidence-led app-based interventions, may assist this effort - with additional benefits of reducing the financial cost to services, improving access and reducing waiting times for support (NHS England, 2019). To our best knowledge, the Autism Depression Trial (ADEPT) is currently the only registered randomised pilot trial to begin to investigate the use of guided self-help for mild depression in ASD (Russell et al., 2017). The ADEPT feasibility study confirmed that retention rates were higher for guided self-help, as compared to treatment as usual, in 70 autistic adults across nine sessions. However, results from a full-scale randomised controlled trial have not yet been published (Russell et al., 2019). Of importance, future trials should be informed by the development and validation of more robust clinical outcome measures for assessing change in mood symptoms in autism (McConachie et al., 2015).

Future directions and conclusions

To conclude, existing evidence suggests that comorbid mood disorders are significantly more common in ASD than the general population. Nevertheless, mood problems may be challenging to identify, due to the lack of validated measures for use with individuals with autism, who often present with ‘atypical’ features that are not captured by existing diagnostic tools. The potential aetiological underpinnings of mood symptoms in autism include several interrelated genetic, cognitive/behavioural, social and physiological/neurobiological, mechanisms. Many of these mechanisms are largely overlapping with those identified in the general population and other clinical groups, but may be exacerbated either directly or indirectly by autism-related neurodevelopmental trajectories, core autism features and lived experiences of autism. Similarly, emerging evidence-based psychological interventions specified for use in other clinical groups - namely CBT and mindfulness-based therapy - appear to be effective in a modified form for reducing depression symptom severity for some autistic adults, at least in the short-term.

Nevertheless, as compared to other commonly co-occurring psychiatric conditions, like anxiety, research focussing on mood disorders in ASD remains notably lacking (Smith & White, 2020). Adequately powered studies investigating bipolar disorder are particularly sparse (Borue et al., 2016). In future research, large-scale longitudinal studies that include individuals with autism diverse in sex, developmental and cognitive ability are required to characterise individual differences in the presentation and aetiologies for mood symptoms in autism, aiding earlier and more accurate identification. Furthermore, high-quality randomised controlled trials to assess intervention efficacy for improving mood problems and broader quality of life are necessary to ensure the best outcomes for autistic people and their families.

Author contributions

BO conceptualised the article, performed the literature review and wrote the manuscript in full; EL and DM critically revised the article; all authors approved the final version for submission.

Disclosure statement

Prof. Murphy receives consultancy fees from F. Hoffmann-La Roche, is supported by the NIHR Biomedical Research Centre and reports grants from the Innovative Medicines Initiative outside the submitted work. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care. There are no other declarations.

Additional information

Funding

The authors are supported by the Innovative Medicines Initiative 2 Joint Undertaking under grant agreement No 777394. This Joint Undertaking receives support from the European Union’s Horizon 2020 research and innovation programme and EFPIA and SFARI, Autistica, AUTISM SPEAKS. The views expressed are those of the author(s) and not necessarily those of the IMI 2JU.