BiomaRkers for AnorexIa NErvosa and autism spectrum Disorders- longitudinal study

There are unique challenges to identify the individual contributions of different factors on the development and ongoing nature of anorexia nervosa. It has long been theorised that autism comorbidity may further complicate the picture and exacerbate the devastating impact of the illness. Therefore, taking a science- and theory-driven approach, we will follow up young women with anorexia, who were treated for anorexia, and young women who never had an eating disorder, who have previously participated in our funded study from MRC/MRF.

The first time we saw these young women, we carried out a wide variety of tests, questionnaires, interviews, and MRI scans. These different forms of research techniques allowed us to examine how they think and react to social images and situations (such as looking at faces or storytelling), and to look at whether their illness and levels of autism symptoms had an impact on their responses.

In this proposed further research, we will see these young women again two years later and carry out similar testing and interviews. This will allow us to look at clinical features, how their answers and reactions have changed over time. We will also be able to examine changes over time in the areas of their brains and neural pathways which are used to carry out different types of problem solving in these tasks. We are then going to use machine learning, a set of advanced computing and statistical techniques, to develop a model which predicts likelihood and length of recovery from anorexia. As we are aiming to see all participants from the first study again, rather than only the patient and recovered groups, our findings will be inherently robust because we will be able to compare between the developmental trajectories of the healthy controls and the patient and recovered young women. This means that we will be able to examine the independent, but interlinked, effects of anorexia nervosa, autism symptoms, and aging, with the ability to control for normal developmental changes alongside the impact of illness.

Using this longer-term approach to seeing participants is groundbreaking, as it will help us to move the field towards an understanding of the lifelong perspective on illness pathways and co-morbidities, tested and analysed in novel ways which harness insights from the rapidly-advancing world of big data and machine learning. This will enable us to gain new insights, potentially identifying underlying mechanisms which underwrite illness development and maintenance. That knowledge, in partnership with leading experts and policymakers, has the potential to develop targeted interventions which can be implemented nationally and internationally paving the way to reducing stigma and shame around anorexia nervosa and creating recognition of its interaction with autism spectrum conditions.

Aims: The proposed study aims to make fundamental discoveries in the field of eating disorders by accelerating understanding of the mechanisms underpinning illness maintenance in anorexia nervosa (AN).

Objectives: Utilising a high-quality database of multimodal baseline data, we will examine longitudinal changes in social-emotional and cognitive processing in young people with and without AN and those who have recovered from AN. We will additionally investigate longitudinal changes in autistic symptoms and their role in behavioural and brain responses to social-emotional and cognitive stimuli. Finally, we will explore patterns in the longitudinal, multimodal data that predict illness outcome.

Design: Longitudinal, mixed methods multi-approach comparative study.

Methodology: We will follow-up 160 young people who took part in the initial baseline assessment two years later. The sample includes 70 young people who had a current diagnosis of AN, 70 aged-matched healthy participants, and 20 young people who had recovered from AN when they first completed the baseline assessment. All participants will be female and aged 12 - 25 years. We will conduct the same battery of clinical and behavioural tests and examine longitudinal changes in autistic symptoms, central coherence, set-shifting, and theory of mind. We will also examine changes in the neural systems that underpin implicit responses to positive and negative facial affect, field dependence, and theory of mind. Furthermore, we will apply science driven, Big Data methods to identify patterns in the robust, multimodal data that predict good or poor illness outcome at the 6- and 12-month follow-up points using recurrent neural networks. Using two follow-up time points will enable us to examine the effectiveness and clinical utility of the final model. Our contingency plan of over-recruiting by 10% at the baseline assessment will ensure we will meet the sample size goal at the second assessment point.

Anorexia nervosa (AN) has devastating consequences including not only on the individual due to lasting effects of starvation on the brain leading to treatment resistance, and ultimately death, but also on families' and carers' wellbeing. AN also has substantial financial impact on societal level costing UK economy approximately £1.25 billion per annum. Steady accumulation of evidence suggests that treatments should target the cognitive and emotion processing difficulties. New, emerging evidence also highlights the importance of developing interventions for implementation of treatments that take presence of comorbid autistic symptoms into considerations. However, as the aetiology of AN and the complex interplay with autistic symptoms and AN still remains uncertain, development of effective interventions is extremely difficult. Our pilot audit data shows that people with both conditions AN and ASD have longer admission and require more clinical resources (data and publications from our research group available upon request from PI).

We propose that improved understanding of the underpinning mechanisms in AN during early stages of illness through longitudinal examination of autistic symptoms, and brain and behavioural responses to key cognitive and social-emotional processes. Better understanding of the underlying neural systems of AN during the early stages of illness, will accelerate the understanding of AN and allow for the development of new neurobiological models of AN. Emphasising the links between behavioural and brain systems help shift focus from environmental drivers, such as social and family, to include biological influences across the lifespan. This approach can help lift some of the shame and stigma associated with the diagnosis. Stigma and shame have negative impact on agency in help-seeking, which further contribute to long duration of illness and treatment resistance due to denial.

Identification of longitudinal neural inefficiencies present in young people with AN, allows for better understanding of the illness-related underlying mechanisms that contribute to illness progression as well as the neuroprogressive changes that occur as a consequence of the illness. These findings could improve understanding of what mechanisms behavioural or non-invasive neurobiological treatments (e.g. TMS) should be targeting during the early stages of illness. The tasks used also easily lend themselves to a test battery that could be used to examine the effectiveness of new or currently available treatments. The findings can also shed light on the mechanisms through which the treatments work to allow further understanding of their effectiveness and their further improvement.

Finally, the findings from the study will also shed light on longitudinal, multimodal markers that predict good or poor illness outcome. Thus, the proposed project will allow us to harness knowledge about predictors of potential chronicity early and identify differences between treatment responders and non-responders. Treatments could then be developed to specifically target the non-responders who do not benefit from currently available treatments. Therefore, the findings from the proposed study have substantial practical clinical implications and will help further understanding of the aetiology of AN as well as make contributions to evidence based medicine to target chronicity and treatment resistance.