Alexithymia and autism diagnostic assessments: Evidence from twins at genetic risk of autism and adults with anorexia nervosa

Highlights

• Alexithymia has been argued to contribute to some deficits associated with autism.

• It is possible alexithymia affects scores on instruments such as the ADOS.

• This was examined in women with anorexia nervosa (AN), and adolescent twins with high autistic traits/whose twin had high autistic traits.

• Parent-reported alexithymia in the adolescent sample was associated with ADOS outcome, scale and item scores.

• In the AN sample, alexithymic women were more likely to meet diagnostic cut-off, despite few associations between scale/item scores and alexithymia.

Abstract

Background

Alexithymia, a difficulty identifying and communicating one’s own emotions, affects socio-emotional processes, such as emotion recognition and empathy. Co-occurring alexithymia is prevalent in Autism Spectrum Disorder (ASD) and underlies some socio-emotional difficulties usually attributed to autism. Socio-emotional abilities are examined during behavioural diagnostic assessments of autism, yet the effect of alexithymia on these assessments is not known. This study aimed to examine the associations between alexithymia and Autism Diagnostic Observation Schedule (ADOS) assessment scores.

Method

Two previously collected samples of ADOS assessments were used to examine the relationship between alexithymia and ADOS scores. Participants included 96 women with anorexia, and 147 adolescents who were either high in autistic symptoms, or whose twin had high autistic symptoms. We examined 1) the impact of alexithymia on meeting the criteria for autism/ASD, 2) correlations between alexithymia and ADOS subscales, and 3) whether alexithymia predicted scores on specific ADOS items, selected a priori based on existing literature.

Results

In the adolescent group, parent-reported (but not self-reported) alexithymia correlated with both ADOS sub-scales, predicted scores on ADOS items, and predicted meeting clinical cut-offs for an ASD/autism diagnosis. Few associations were observed in the anorexic sample between self-reported alexithymia and ADOS subscale and item scores, but the presence of alexithymia predicted the likelihood of meeting diagnostic criteria for autism/ASD in this sample.

Conclusions

Alexithymia does show relationships with ADOS assessment scores. We discuss potential clinical and research implications, particularly in studies of autism where the ADOS is often the only diagnostic measure used.

Introduction

Alexithymia, a sub-clinical trait characterised by a difficulty identifying and describing one’s own emotional states, is elevated in Autism Spectrum Disorder (ASD; hereafter referred to as “autism”), and several other psychiatric and neuropsychiatric conditions such as substance abuse, schizophrenia and eating disorders (Berthoz, Lalanne, Crane, & Hill, 2013; Pinard, Negrete, Annable, & Audet, 1996; van’t Wout, Aleman, Bermond, & Kahn, 2007; Westwood, Kerr-Gaffney, Stahl, & Tchanturia, 2017), as well as in acquired brain damage or neurodegenerative disease (Sturm & Levenson, 2011; Wood & Williams, 2007). Despite evidence that alexithymia may be relevant to understanding socio-emotional difficulties (such as impaired emotion recognition and empathy) across a broad range of clinical conditions, there has been little attempt to investigate the potential impact of alexithymia on diagnostic practices. We focus here on the role of alexithymia in the diagnosis of autism.

Recent studies have argued that alexithymia, rather than autism, is responsible for several socio-emotional difficulties commonly attributed to autism, including emotion recognition, eye contact and empathy (Bird & Cook, 2013; Bird, Press, & Richardson, 2011; Grynberg, Luminet, Corneille, Grèzes, & Berthoz, 2010; Oakley, Brewer, Bird, & Catmur, 2016). As yet, the causal relationship between autism and elevated rates of alexithymia are unclear, but autism and alexithymia are conceptually distinct. Although between 40 and 65 % of the autistic¹ population are thought to be alexithymic (Berthoz & Hill, 2005; Hill, Berthoz, & Frith, 2004), a figure significantly above the rate of 10 % observed in the general population (Franz et al., 2008), alexithymia is neither necessary nor sufficient for a diagnosis of autism: there exist autistic individuals with and without alexithymia, and alexithymic individuals who do not meet the diagnostic criteria for autism. Therefore, while certainly autistic individuals are more likely to be alexithymic than non-autistic individuals, alexithymia is not always a characteristic of autism, and it has similarly increased prevalence across other psychiatric populations.

However, the findings regarding the association between alexithymia and socio-emotional difficulties have important implications for diagnostic practice. A key piece of evidence clinicians may draw upon when making a diagnosis of autism is behavioural assessment, most commonly the Autism Diagnostic Observation Schedule (ADOS; Lord, Rutter, DiLavore, & Risi, 1999; ADOS-2; Lord et al., 2012). The ADOS is considered by many to be a ‘gold standard’ test for the presence of autism, however, elements of this assessment involve socio-emotional processes known to be associated with alexithymia. Given recent findings that some socio-emotional impairments usually thought to be a feature of autism are attributable to alexithymia, it is important to investigate the impact that alexithymia has on autism diagnostic assessments.

If ADOS scores in part reflect several processes now known to be associated with alexithymia, it could be expected that alexithymic individuals would appear to demonstrate greater autistic symptomatology in ADOS assessments than non-alexithymic individuals. Hypothetically, in the extreme, an individual may receive a diagnosis of autism when in fact they simply have a high degree of alexithymic traits (a false positive). Another possibility is that if scores on ADOS assessments are highly confounded by alexithymia, and ADOS cut-off scores were determined in populations of autistic individuals with a high degree of co-occurring alexithymia, then individuals who have autism but low levels of alexithymia may not meet ADOS cut-offs and therefore not receive a diagnosis (a false negative). While diagnosis should not be based on any single instrument, in practice the diagnosis of autism in adults, or where a caregiver cannot provide developmental history, may rely heavily on behavioural assessments including the ADOS. As adult diagnosis of autism becomes increasingly common (see Bastiaansen et al., 2011), examining factors that potentially confound behavioural assessments is important. Furthermore, in research, the ADOS is often the only measure administered to confirm diagnosis (Falkmer, Anderson, Falkmer, & Horlin, 2013).

The ADOS comprises four modules, administered dependent upon the verbal ability and age of the individual. Scoring algorithms for use with the latest edition of the ADOS (ADOS-2) yield two subscale scores for Modules 1 to 3: social affect (SA), and restricted and repetitive behaviours (RRB) (Gotham, Risi, Pickles, & Lord, 2007). Published algorithms also allow these same subscale scores to be derived for Module 4 of the ADOS (Hus & Lord, 2014). The symptoms assessed in the ADOS have not been investigated for their association with alexithymia before now. However, based on existing literature, it is possible to predict that alexithymia likely impacts at least a subset of the items included in the ADOS assessment.

For adolescents and adults with fluent verbal ability, the ADOS includes an item which assesses the ability to communicate one’s own emotions. Specifically, individuals are asked to identify times that they felt a given emotion, and to describe what that emotion feels like. This clearly requires an ability that is by definition impaired in alexithymia. Indeed, previous research shows that, when asked similar interview questions to those contained in the ADOS, alexithymic individuals (without autism) produce fewer emotion words, and more null responses (“I don’t know”), than individuals with fewer alexithymic traits (Wotschack & Klann-Delius, 2013). Thus, alexithymic individuals would be expected to show greater impairment on this ADOS item than those without alexithymia.

The ADOS also assesses whether an individual comments on others’ emotions or displays empathy. Although empathy deficits are often assumed in autism, research suggests it is alexithymia rather than autism that predicts reduced empathy (Bird & Cook, 2013; Bird et al., 2010). In the general population, alexithymia is inversely related to empathy, and alexithymic individuals report lower scores on empathy questionnaires, and demonstrate atypical neural responses to others’ pain compared to non-alexithymic people (Grynberg et al., 2010; Guttman & Laporte, 2002; Moriguchi et al., 2007). Alexithymic individuals also perform worse than typical individuals on tasks requiring the recognition of emotion from faces (Grynberg et al., 2012; Jongen et al., 2014) and voices (Heaton et al., 2012), and alexithymia, rather than autistic traits, accounts for the difficulties in facial and vocal emotion recognition in the autistic population (Cook, Brewer, Shah, & Bird, 2013; Heaton et al., 2012). Such deficits would likely lead to a reduced ability to interpret accurately, and presumably a reduced propensity to comment upon, others’ emotions (Coll et al., 2017).

An ADOS assessor will consider unusual eye-contact, and whether appropriate facial expressions are directed to the assessor. Eye-tracking investigations have shown that alexithymia, rather than autism symptom severity, predicts fixation of the eye region in autism (Bird et al., 2011). Production of emotional facial expressions has also been shown to be affected in alexithymic individuals, and reduced emotional expression has been linked to alexithymic traits rather than autistic traits in groups with and without autism (Brewer, Biotti et al., 2016; Brewer, Cook, & Bird, 2016; McDonald & Prkachin, 1990; Trevisan, Bowering, & Birmingham, 2016). Limited or abnormal facial expressions and eye contact may thus be a feature of alexithymia and contribute to higher scores on these ADOS items.

Individuals’ insight into typical social situations (e.g. what it means to be a friend versus an acquaintance, why people might seek romantic relationships) is examined during an ADOS assessment. At present there is limited evidence about alexithymic individuals’ social insight, per se. Nonetheless, alexithymic individuals experience difficulties in their interpersonal lives, including cold social functioning (Spitzer, Siebel-Jürges, Barnow, Grabe, & Freyberger, 2005; Vanheule, Desmet, Meganck, & Bogaerts, 2007), and reduced satisfaction in intimate relationships (Humphreys, Wood, & Parker, 2009). Alexithymia also impacts social reward (Foulkes, Bird, Gökçen, McCrory, & Viding, 2015) and the ability to draw typical inferences about others’ character (Brewer, Collins, Cook, & Bird, 2015), which could underpin some of the interpersonal problems reported by alexithymic individuals. Interpersonal difficulties and abnormal social responses could feasibly reflect and/or contribute to atypical insights into social situations and roles as measured by the ADOS.

The association between alexithymia and RRBs has received comparably less research interest than social abilities. Nonetheless, alexithymia appears to be related to unusual sensory abilities, both in non-autistic (Kano, Hamaguchi, Itoh, Yanai, & Fukudo, 2007; Nyklíček & Vingerhoets, 2000) and autistic populations (Milosavljevic et al., 2016), and is related to hypersensitivity to touch and pain and self-reported sensory processing sensitivity (Liss, Mailloux, & Erchull, 2008; Nyklíček & Vingerhoets, 2000; Sivik, 1993). Thus, there may be some overlap between the sensory issues reported in autism and those experienced by individuals with alexithymia, but at present there is limited evidence concerning the rest of the RRB scale.

Alexithymia does not only co-occur with autism, and it is important to consider the impact of alexithymia on autism assessments in groups for which assessment for co-occurring autism is sometimes indicated. One group that may be of particular interest is individuals with eating disorders, and specifically anorexia nervosa (AN).

Research into socio-emotional functioning in AN has reported deficits in recognising emotion from faces (Kucharska-Pietura, Nikolaou, Masiak, & Treasure, 2004; Zonneville-Bendeck, van Goozen, Cohen-Kettenis, van Elburg, & van Engeland, 2002), reduced empathy (Morris, Bramham, Smith, & Tchanturia, 2014), and poor theory of mind (Russell, Schmidt, Doherty, Young, & Tchanturia, 2009; Tchanturia et al., 2004). Poor social-cognitive functioning may be in part the result of generally diminished cognitive function due to starvation; though some social difficulties appear to begin before the onset of eating disorders (and thus before the effects of malnourishment) (Gillberg & Råstam, 1992; Westwood, Lawrence, Fleming, & Tchanturia, 2016).

These deficits have been commented upon by some as being similar to the difficulties seen in autism (Treasure, 2013). Indeed, autism is overrepresented in populations with eating disorders (Huke, Turk, Saeidi, Kent, & Morgan, 2013; Westwood, Kerr-Gaffney et al., 2017; Westwood, Mandy, & Tchanturia, 2017; Westwood, Eisler et al., 2016), with some studies reporting a rate of ASD in AN of up to 52.5 % (see Westwood & Tchanturia, 2017). Individuals with AN also score higher on self-report measures of autistic traits (Westwood, Eisler et al., 2016). Thus, the evidence suggests that autistic-like behaviours, indexed by experimental tasks, self-report measures and clinical behavioural assessments, occur in a sizeable proportion of individuals with AN.

Alexithymia may play an important role in the heterogeneity of the emotional problems and autistic-like difficulties observed in AN. Alexithymia has been linked to both eating disorder symptomatology and social difficulties in anorexia (even after depression, anxiety and state of starvation are accounted for; Brewer et al., 2019; Courty, Godart, Lalanne, & Berthoz, 2015). Emotion recognition deficits may arise in cases of eating disorder where there is co-occurring alexithymia (Brewer, Cook, Cardi, Treasure, & Bird, 2015), and emerging evidence suggests that alexithymia in AN accounts for the autistic symptoms in some of these individuals (Westwood, Kerr-Gaffney et al., 2017). The presence of alexithymia in AN may also have important implications for how to assess autism in this group; indeed, some reports have indicated that autism assessments based on developmental data versus current self-report symptoms yield different findings regarding whether rates of autism are higher than would be expected in AN samples (see Stewart, McEwen, Konstantellou, Eisler, & Simic, 2017). However, it has also been argued that the association between eating disorders and alexithymia may largely be driven by joint associations with depression (Marchesi, Ossola, Tonna, & De Panfilis, 2014; Montebarocci, Surcinelli, Rossi, & Baldaro, 2011); whether this can also account for the association between AN and autism is unclear.

Understanding the impact of alexithymia on autism diagnostic measures will yield important insights for both research and clinical practice. Considering the relationship between alexithymia and ADOS scores in populations with and without autism, including populations known to have high rates of alexithymia (e.g. AN) would illuminate whether high alexithymic traits increase ADOS scores. This paper thus examined the association between ADOS scores and alexithymia, using previously collected data from two groups: AN patients, and adolescents with autism or with increased likelihood of showing subthreshold autistic traits due to having a co-twin with autism. The extent to which alexithymia predicted ADOS scores was examined, and our predictions were that individuals with alexithymia would: a) be more likely to meet criteria for an autism/ASD diagnosis; b) score higher on the ADOS than those without alexithymia, and c) score higher on certain ADOS items that tap processes previously linked to alexithymia. To test the latter prediction, in order to avoid an increased risk of Type I error due to multiple comparisons, the decision was made not to examine all items included in the ADOS, but to focus selectively on a subset of items for which previous research predicted an association. These included: a) Comments on others’ emotion; b) Unusual eye contact; c) Facial expressions directed to examiner; d) Communication of own affect (recorded for Module 4 participants only); e) Insight into typical social situations.