Neural mechanisms of perceptive and affective processing of body stimuli in Anorexia nervosa – are there developmental effects?

Objective Different components of body image processing seem to be reflected by different neural mechanisms. A core symptom of Anorexia nervosa (AN) is a disturbance of body image with correlates found on a neural level. The present study focusses at the neural processing of visual body stimuli of different weight categories in adolescent and adult AN patients. Method 33 adolescents aged 12–18 years (15 AN patients, 18 control participants) and 36 adult women (19 AN patients, 17 control participants) underwent functional magnetic resonance imaging (fMRI) while performing a perceptive and an affective body image task involving photographic stimuli of women belonging to different BMI categories. Results Differential effects on activation depending on the BMI of the women shown in the pictures were found in frontal brain regions, the thalamus, caudate and the fusiform gyrus. Group effects differentiating between AN patients and control participants were seen mainly in the caudate and insula. Discussion During a perceptive task, diminished activation of regions involved in perceptive and evaluative functions as well as emotional reasoning was seen in AN. During evaluation of the neural processing in an affective task there was a tendency towards activation differences reflecting reduced ability of size estimation and impaired integration of visual and body perception with emotions.

size estimation and impaired integration of visual and body perception with emotions.

54
In eating disorders (EDs) abnormal processing of body image, specifically within a neural network for 55 body perception, has been discussed (7) and investigated in several functional Magnetic Resonance 56 Imaging (fMRI) studies. Activation patterns while watching one's own body (8-13) as opposed to 57 those while looking at others', mainly female, bodies (9-12, 14-18), have been studied. On top of 58 instructions to simply look at the bodies, specific tasks like weight rating (14,15,19) or assessing 59 subjective emotions (11,(14)(15)(16)18) enabled researchers to look at differential aspects of body 60 picture processing.

61
Differences between AN patients and healthy participants in body image processing, which have 62 been found in the aforementioned studies, can be allocated to the following neural structures:

63
 In areas which have been described as selectively processing body pictures such as the 64 fusifom body area and extrastriate body area (see above) either enhanced (10) or reduced  The anterior insula was strongly activated in AN patients when comparing their own body to 80 slim bodies (16) and when viewing distorted thin self portraits and rating their satisfaction 81 (19). By contrast, the posterior insula has been identified as belonging to a subnetwork of 82 regions with decreased connectivity in AN patients (26).

83
 The ventral striatum, a region which has been linked to reward reactions, shows a different 84 pattern of activation in AN patients and control participants when confronted with women's 85 bodies of different BMI categories: activation was higher during the processing of 86 underweight stimuli compared with normal weight stimuli in women with AN, and the 87 reverse pattern was observed in healthy females (14).

88
The latter study was conducted with adult AN patients and later enhanced by an adolescent AN 89 group of which data were analyzed separately. As AN typically has its onset in adolescence (27)  To our knowledge, only two studies differentiated between two aspects of body image processing using a more perceptive and a more affective task (14,19). Our aim was to examine changes in brain to different body shapes presented with a standardized picture set. from each BMI category were shown: extremely underweight (BMI 13.5-15.0 kg/m 2 ), underweight  consider delays and graded response times of the BOLD signal to the stimuli. For this a 2γhaemodynamic response function was used (time to response to peak: 5 s, time to undershoot peak: participants. analysis. the group level (multi-study t tests) for adult and adolescent participants separately.

192
In order to clarify effects of picture category on brain activation patterns rather than merely group 193 differences of processing of the single categories, individual β values were extracted for those 194 regions in which group differences (patients vs. control participants) were evident (see Table S1 for 195 regions with group differences; a detailed description of those group statistics would be beyond the 196 scope of the present paper). These β values were used as a measure of the strength of the individual 197 BOLD changes within the respective ROI. They were then subjected to a repeated measures ANOVA 198 with "BMI category" of the stimuli (extremely underweight, underweight, normal weight and 199 overweight) as within-subjects variable and "participant group" (2, "AN patients" and "control 200 participants") and "age group" (adolescents vs. adults) as between-subject factors. Greenhouse-

201
Geisser correction was used when appropriate. For brain regions with significant "participant group" 202 or "age group" interactions, group comparisons via t tests were conducted using Bonferroni-Holm 203 correction to control for multiple comparisons. For some regions that we considered to be of 204 importance for the study the effects were only close to significance. These cases were treated and 205 discussed as tendencies. For regions with significant main effects, planned comparisons were 206 performed in order to determine activation differences between groups for the single picture 207 categories, Bonferroni-Holm correction was also applied. The ANOVA and post hoc tests were carried out using SPSS 21.0 (IBM Corp.).

211
3.1. Participant characteristics appearance ideals assessed by the SATAQ were higher in AN patients with larger group differences in 217 adults than in adolescents. For adults, but not for adolescents, group differences were seen in the 218 appearance comparison behaviour assessed by the PACS and for body dissatisfaction (see Table 2).  BMI as being higher than controls is stronger in adults than in adolescents.

247
Group comparisons via t tests showed that the weight of extremely underweight and normal weight 248 women in the pictures were rated higher by adult AN patients than by healthy women. In 249 adolescents, no significant group differences for the single categories emerged after Bonferroni-

250
Holm correction (see Figure 2). In affective ratings a "BMI category" main effect (F(3,180)=161, p<0.001, part. ²=.729) with highest 258 ratings for underweight, followed by normal weight, extremely underweight and overweight body 259 shapes was seen. A significant "BMI category"*"participant group" interaction (F(3,180)=19.9, 260 p<0.001, part. ²=.249) indicated higher satisfaction ratings in the control group for pictures of 261 normal weight women compared to pictures of the extreme underweight category, but not in AN 262 patients.

263
Direct group comparisons within the age groups revealed higher satisfaction ratings for extremely 264 underweight bodies in adult AN patients and lower ratings for normal weight bodies. Adolescent AN patients rated their satisfaction for all weight categories similarly to control participants (see Figure   266 2). 267 268 3.3. Activation patterns 269 Beta values of activated clusters were subjected to separate ANOVAs for the perceptive and the 270 affective task respectively. Significant effects in brain regions previously seen as involved in body 271 image processing were found for "BMI category" of the women in the pictures, "participant group", 272 "age group" and "participant group" and "BMI category" interaction (see Table 3 Table   334 3 for ANOVA main effects and interactions).
335 Figure 3: Activation patterns (beta values) in AN patients (green) and control participants (blue)  (7) were found. Frontal regions also showed a pattern of decreasing activity from extremely underweight to 368 overweight body pictures. Regions belonging to the prefrontal cortex (BA 46, 47) exhibited this 369 pattern in both tasks, but inferior frontal regions (BA 9) more specifically in the affective task.

370
Prefrontal regions have been shown to be activated to a greater extent in AN patients while watching 371 others' bodies or reading body words (9, 36), but also in control participants (Sachdev et al., 2008).

372
Inferior frontal regions have even more frequently been linked to body image tasks. Looking at 373 bodies vs. chairs caused increased inferior frontal activation in healthy participants (Suchan et al., 374 2012).Regions of BA 9 have been activated in response to body images, but less to images of interior 375 design (Friederich et al., 2007) and in body-shape self-comparison in AN patients (Friederich et al.,376 2010) as well as in controls (Sachdev et al., 2008, Wagner et al., 2003 and have been seen in the 377 context of self-reflection and self-judgement (21). Medial frontal regions have been linked to 378 cognitive aspects of emotional processing, paying attention to emotions and fear processing (37, 38).

379
These mechanisms have possibly occurred in all participants in our study when looking at strongly 380 underweight other bodies which are not as common as and probably inducing more (negative) 381 emotions than normal weight or overweight bodies. Generally, the pattern of increased activation of 382 different regions for strongly underweight and, to a lesser extent, underweight bodies could be seen 383 as increased alertness and attention in response to these unusual body shapes. Similarly, increased set in an overlapping sample, but occured in AN patients only (39). In contrast to the MRI activation patterns described here, those increased reactions observed at electrodes over central brain areas number of brain regions are assumed to be generators of the Late Positive Potential, activation 390 patterns in specific brain regions cannot be directly linked to enhanced amplitudes. categories in the affective task. In the studies of Fladung et al. ( , 2013 there was also a greater 393 effect for this region when participants fulfilled an affective than a mere perceptive task with 394 enhanced activation for underweight in comparison to normal weight body pictures in AN patients.

395
In our study this reward-related brain structure did not show greatest activation for extremely which does not seem to be consistent with increased activation for higher BMIs in our study, but is 432 also part of the fear and emotional network (16). Therefore activation decrease for extremely 433 underweight bodies in patients might point towards an impaired integration of visual and body 434 perception with emotions, especially fear, which might play a role in control participants, but not in 435 AN patients. The insula is connected with many structures for emotional processing and bodily 436 experience, among those frontal regions and the striatum which show reduced activity in our task as 437 well (see above). Dysfunction within the fronto-insula-limbic-striatal circuit has been discussed in the context of failure to regulate inputs from higher level cognitive representations about the than the anterior section in the previous literature, the posterior section of the insula has also been 445 discussed in the context of eating disorders. It has been identified as part of a subnetwork of 446 connections with decreased connectivity in AN and discussed as a central "hub" within this network 447 (26). Specifically the posterior insula has been associated with sensorimotor processing with 448 functional connections to motor and somatosensory cortical areas and may thus be involved in the 449 processing of own bodily signals such as body size (Ehrlich, Lord et al. 2015 A strong tendency towards a difference in activation patterns between adults and adolescents was 455 found for the affective task in the caudate body. Here generally higher activation, specifically with 456 significance for (extremely) underweight pictures, was seen in adolescents as compared to adults. To 457 date, few studies on body image processing have been conducted with adolescent participants. In 458 our EEG study, we saw differences between adolescents and adults in the form of increased 459 responses towards extremely overweight bodies in adolescents (39). Regarding brain activation 460 patterns, however, no differences regarding overweight bodies were found in this study. As 461 discussed above, the caudate body is involved in perceptual processes, but also connected to anxiety 462 in a self-comparison condition (21). During adolescence, the striatum is known to undergo is stronger in adolescents compared to children and adults (51), and the region is also involved in evaluation and different emotions related to such a process, which is often linked to ideas about 470 one's own appearance and body shape during adolescence, could lead to an increased caudate 471 activation in adolescents compared to adults. This could be especially relevant when looking at 472 underweight shapes, which are more likely to be targets for social comparison and evaluation 473 processes.

475
In a task using representations of bodies belonging to different BMI categories, several brain regions 476 differentiated well between the distinct categories and therefore appear to be involved in body size components, e.g. using elements of reframing or acceptance therapy to manipulate and evaluate 486 participants' attitude towards their body. It would also be interesting to acquire longitudinal data in 487 the course of treatment and, in order to further clarify the processes involved, to assess neural activation patterns in the context of e.g. dopamine depletion, as dopamine has been seen to in bold print).