Borderline personality disorder and eating disorders are characterized by specific patterns of autistic traits, post-traumatic and catatonic symptoms

Abstract

Purpose

The present study aims to clarify the relationships between autistic traits, post-traumatic symptoms, and catatonic features in individuals with BPD and FED. Specifically, we sought to identify whether distinct patterns of autistic and trauma-related features characterize these conditions and to examine their predictive role in the development of catatonic manifestations.

Methods

49 subjects with BPD, 55 subjects with FED, specifically anorexia nervosa (AN) or bulimia nervosa (BN)/binge-eating disorder (BED) and 55 non-clinical subjects (NCs) were recruited and assessed with the Adult Autism Subthreshold Spectrum (AdAS Spectrum), the Catatonia Spectrum (CS), and the Trauma and Loss Spectrum Self-Report (TALS-SR).

Results

Both FED and BPD groups scored higher on AdAS, CS and TALS-SR than NC, while BPD group scored higher than FED on TALS-SR “Personal Characteristics and Risk Factors” domain. AN significantly differed from BN/BED in specific AdAS and CS domains. A positive correlation among AdAS spectrum, CS and TALS-SR domains was reported. Both AdAS and CS total score were predictors for the inclusion in BPD or FED group, with some specific differences in the predictive role of single domains. For the FED group, only AdAS symptoms predicted greater catatonic symptoms, while for the BPD one, both AdAS and TALS-SR were predictors of greater catatonic features.

Conclusions

Both BPD and FED subjects exhibit more frequent autistic traits, stress-related symptoms, and catatonic manifestations compared to controls, with specific patterns of features associated with each condition.

Level of evidence: Level V: Descriptive studies.

Introduction

Borderline Personality Disorder (BPD) is a severe mental health condition marked by instability in affect regulation, self-image, interpersonal relationships, and impulse control. Individuals often experience emotional dysregulation, chronic emptiness, self-destructive behaviors, and recurrent suicidal ideation or self-harm [1]. While BPD affects about .7–2.7% of the general population, prevalence increases to approximately 12% in outpatient and 22% in inpatient psychiatric settings [2]. The disorder significantly impairs quality of life, contributing to low educational and occupational attainment, interpersonal conflict, risk-taking behaviors, and high comorbidity with mood, anxiety, and substance use disorders, along with increased suicide risk [3, 4]. Women are disproportionately diagnosed, at a 3:1 ratio in clinical settings, though sociocultural and referral biases may contribute to this disparity [1, 4]. BPD is frequently comorbid with feeding and eating disorders (FEDs), including anorexia nervosa (AN), bulimia nervosa (BN), and binge-eating disorder (BED), which are more prevalent in females [5]. This group of disorders represents psychiatric conditions marked by severe disruptions in eating behavior, body image disturbances, and maladaptive coping mechanisms that can lead to life-threatening complications [5, 6]. Multiple researches have consistently shown that FEDs are significantly more prevalent in females compared to males, with epidemiological studies consistently highlighting a pronounced gender disparity across various diagnostic categories [7–9].

Recent research has highlighted elevated autistic traits (ATs) in individuals with BPD or FEDs, suggesting that ATs may act as vulnerability factors in the development and persistence of these disorders [10, 11]. BPD and Autism Spectrum Disorder (ASD) share overlapping features, including emotional dysregulation and interpersonal challenges. BPD individuals often show impulsivity and fear of abandonment, while ASD is associated with social communication deficits and rigid behaviors [12]. Moreover, many studies highlighted how subthreshold ATs are highly prevalent among individuals with BPD, contributing to the severity and complexity of the disorder [10, 13–15]. In particular, subthreshold ATs may exacerbate emotional and social difficulties [11, 14–17].

Similarly, in recent years, there has been growing interest in exploring the potential link between FEDs, especially AN, and ASD [18, 19]. Individuals with FEDs frequently exhibit ATs, such as rigid thinking patterns, difficulties in set-shifting tasks and global processing, social anhedonia, reduced emotional awareness, impaired emotional intelligence and difficulties in emotional processing, and a compromised Theory of Mind (ToM) [20–24]. This has led some to propose that FEDs may represent a female-specific manifestation of autism, with restricted interests focused on food and dieting [25].

As noted in the existing literature, not only ATs have been shown to be associated with an increased risk of developing trauma- and stress-related symptoms after adverse life events, but also BPD and FED (in particular AN and BN) have been reported to often being linked by a shared history of trauma. Trauma is a shared risk factor in BPD and FEDs, with both conditions often linked to adverse childhood experiences, such as abuse or neglect [26–28]. In particular, BPD shows notable overlap with complex PTSD (cPTSD), including symptoms of emotional dysregulation and interpersonal instability and recent studies suggest that a history of trauma can exacerbate, in vulnerable populations, such as those with elevated ATs, either BPD symptoms, such as impulsivity and self-destructive behaviors, or eating disorders [29–31]. In FEDs, trauma may contribute to disordered eating as a maladaptive coping mechanism [32]. In addition, ATs are linked to increased vulnerability to trauma due to heightened risk of bullying, exclusion, or abuse, which may lead to PTSD-like symptoms [33]. This suggests that the presence of ATs may mediate trauma susceptibility in both BPD and FED populations.

Another emerging area of interest is catatonia, not only as a full-threshold motor syndrome but as a spectrum of subthreshold symptoms [34]. Traditionally associated with neurodevelopmental and trauma-related disorders, catatonia has rarely been explored in BPD or FEDs, though symptoms such as mutism or immobility have been observed during periods of intense distress [35–37]. Recent research has highlighted overlapping behavioral, motor, and cognitive features between autism spectrum conditions and catatonia, leading to the proposal of an autistic–catatonic continuum. This conceptual model suggests that catatonic manifestations may, in some individuals, reflect an extreme or stress-induced expression of underlying autistic traits, particularly in the context of trauma or affective dysregulation. While this hypothesis remains debated, emerging evidence supports the existence of shared mechanisms, including alterations in sensorimotor integration, executive function, and social cognition.

However, no studies to date have examined the intersection of ATs, trauma, and catatonia in these populations.

Therefore, the present study was designed to clarify the interplay between ATs, post-traumatic symptoms, and catatonic features in individuals with BPD and FEDs. The joint examination of BPD and FEDs was motivated by increasing evidence that these disorders share transdiagnostic features, including emotion dysregulation, impulsivity, trauma-related symptoms, and ATs, making them suitable models for studying overlapping dimensional psychopathology. Specifically, we aimed to identify distinct patterns of association among these dimensions and to determine whether autistic and post-traumatic features may predict catatonic manifestations differently in BPD and FED populations.

More precisely, we investigated:

• whether individuals with BPD and FED show higher levels of autistic, post-traumatic, and catatonic symptoms compared to a non-clinical sample (NC);

• whether specific subdomains of these spectra differentiate BPD from FED;

• whether autistic and post-traumatic dimensions predict the severity of catatonic features within each group.

Based on the literature linking autism, trauma, and catatonia, we hypothesize that individuals with BPD and FED will show significantly higher levels of ATs, post-traumatic symptoms, and catatonic features than the NC sample. Within clinical groups, we expect that ATs and post-traumatic symptoms will be significantly correlated with catatonic features and possibly predict the latter.

By addressing these questions, the study seeks to delineate shared and disorder-specific psychopathological pathways across BPD and FED. Clinically, identifying these specific profiles may improve early detection of high-risk individuals, guide differential diagnosis, and support the development of targeted interventions that address both neurodevelopmental vulnerabilities and trauma-related mechanisms.

The present study adopts a dimensional spectrum approach, focusing on continuous measures of autistic, post-traumatic, and catatonic features to identify potential inter-relationships across diagnostic categories. Rather than testing categorical differences or prevalence rates, the aim is to explore patterns of association among these dimensions and their relative contributions within BPD and FED populations.

Materials and methods

Study sample and procedures

A total of 159 participants were recruited and divided into three groups: 49 with BPD, 55 with FEDs (AN n = 15; BN/BED n = 40), and 55 NCs. The recruitment procedure was carried between September 2023 and June 2024. Participants were aged between 18 and 65. Exclusion criteria included individuals under 18, those with significant intellectual or language impairments, schizophrenia, substance abuse, major neurological or medical conditions, or inability to complete questionnaires or provide informed consent. Subjects belonging to the BPD and FED group were recruited from both in and outpatients referring to the Psychiatric Clinic at the Azienda Ospedaliera Universitaria Pisana. NCs were recruited from healthcare personnel, doctors and paramedics, and student trainees and were also excluded if any DSM-5-TR mental disorder was present. A clinical assessment carried by trained psychiatrists, was used to confirm the absence of psychiatric disorders in the NC population. Diagnoses were confirmed by trained psychiatrists according to DSM-5-TR criteria. All participants received full study information and provided written informed consent. The study was approved by the Ethical Committee of Azienda Ospedaliera Universitaria Pisana (AOUP) and conducted in accordance with the Declaration of Helsinki.

Measures

The adult autism subthreshold (AdAS) spectrum

The AdAS spectrum is a self-administered questionnaire consisting of 160 items designed to assess a wide array of autism-related characteristics in individuals without cognitive or linguistic impairments. The instrument is structured into seven distinct domains, each addressing specific behavioral, cognitive, and emotional characteristics associated with the autism spectrum. The Childhood/Adolescence domain focuses on symptoms evident during early developmental stages, including patterns, such as marked quietness or mutism, avoidance of engaging in play with peers, and experiences of social exclusion or bullying. The Verbal and Non-Verbal Communication assess, respectively, atypicalities in spoken interaction, and challenges in social expressivity. The first domain address difficulties, such as the use of a very soft or monotonous tone, difficulties with conversational turn-taking, initiating or terminating dialogs appropriately, and instances of interrupting others in socially incongruent ways. The second investigates discomfort with direct eye contact, aversion to physical affection and reliance on adopting social “roles” to manage interpersonal interactions. The Empathy domain explores deficits in recognizing, interpreting, and responding to the emotions, facial expressions, and mental states of others. The Inflexibility and Adherence to Routine captures cognitive and behavioral rigidity, such as difficulty understanding subtle contextual cues, resistance to changes in daily routines or working methods, compulsive retention of seemingly unnecessary items, and a preference for fixed patterns, lists, or repetitive clothing choices. The Restricted Interests and Rumination domain encompasses characteristic features, such as a preoccupation with narrow or highly specific topics, fascination with numerical or systematic information, and a tendency toward overthinking and excessive elaboration. Finally, the Hyper- and Hyporeactivity to Sensory Input investigates variations in sensory responsiveness, ranging from heightened or diminished sensitivity to auditory, tactile, thermal, or painful stimuli. Items responses are coded in a dichotomous way (yes/no) and domain scores are obtained by counting the number of positive answers. Higher scores correspond to greater autistic features. The AdAS spectrum demonstrated excellent internal consistency for the total score (Kuder–Richardson’s coefficient = .964) as well as for five out of seven domains (all coefficients > .80) and sound test–retest reliability (ICC = .976) [38].

The catatonia spectrum (CS)

The CS is a structured self-report instrument composed of 74 items, designed to assess the lifetime presence of behaviors, motor phenomena, and related symptoms characteristic of the catatonic spectrum. The measure is divided into eight domains, each representing a distinct dimension of catatonic symptomatology. The Psychomotor Activity domain corresponds to the catatonic stupor and evaluates reductions or complete inhibition in voluntary motor activity, encompassing behaviors, such as immobility, marked rigidity, or sustained fixed postures. The Verbal Response domain represents mutism and investigates the absence or significant reduction of verbal communication despite preserved capacity for speech. This domain assesses phenomena, such as long periods of silence, minimal verbal output, or resistance to speech. The Repetitive Movements domain investigate stereotypies and focuses on purposeless, repetitive, and non-goal-directed motor behaviors, including repetitive gestures, head or body movements, or ritualistic actions. The Artificial Expressions and Actions domain correspond to mannerisms and examines exaggerated, stylized, or peculiar motor and facial expressions that are incongruent with the social situation. Items in this domain address idiosyncratic gestures, rigid or theatrical movements, and overelaborate forms of behavior that may represent maladaptive attempts at communication or self-regulation. The Oppositivity or Poor Response to Stimuli domain represent catatonic negativism and explores the active or passive resistance to external prompts or attempts to engage the individual. Manifestations include refusal to follow instructions, opposition to movement or assistance, or unresponsiveness to environmental cues. The Response to Instructions Given from Outside domain address automatic obedience and captures excessive or mechanical compliance with external commands or suggestions, even when such actions may be inappropriate or contrary to the individual’s will. The Automatisms domain investigates behaviors characterized by apparently automatic or unconscious motor activity, such as echopraxia or echolalia. Finally, the Impulsivity domain assesses the sudden, unpremeditated initiation of motor or verbal acts, which may be disproportionate to external circumstances or internal states. This domain encompasses episodes of abrupt movement, speech, or aggression, reflecting impaired inhibitory control and disorganization of motor behavior. Responses to each item are recorded dichotomously (“Yes” or “No”), reflecting the lifetime occurrence of each specific manifestation. Higher scores correspond to greater catatonic symptomatology. The CS has demonstrated strong validity and reliability, with excellent internal consistency (Cronbach’s alpha values associated with the CS domains were exceeding the value of .7 and the Cronbach’s alpha value associated with the total score of the scale was excellent α = .954), excellent test–retest reliability (all ICCs above the value of .90), and significant positive convergent validity with other dimensional measures of catatonia [34].

The trauma and loss spectrum—self-report (TALS-SR)

The TALS-SR is an assessment tool designed to evaluate lifetime exposure to various loss and traumatic events, alongside stress-related symptoms, behaviors, and personal traits that may serve as risk factors for developing stress-related syndromes. It comprises 116 items, structured into nine domains, each capturing a specific dimension of trauma- and loss-related phenomena. The Loss events domain catalogs a broad array of loss experiences, including bereavement, dissolution of significant relationships, material or physical losses, and reductions in social or economic standing. The Grief reactions domain spans from typical, atypical, and subsyndromal symptomatology associated with complicated grief. Items assess hallmark manifestations such as longing, yearning, and preoccupation with the lost person or place; persistent daydreaming or reminiscing; attachment to objects associated with the loss; and recurrent, intense pangs of sorrow. The Potentially traumatic events domain enumerates traumatic life events, ranging from DSM-5 qualifying traumas to lower-intensity but psychologically impactful experiences. The Reactions to losses or upsetting events domain captures the emotional, physical, and cognitive reactions associated with loss or traumatic exposure. The Re-experiencing, Avoidance and numbing, and Arousal domains assess the respective core symptom clusters of post-traumatic stress. The Maladaptive coping domain focuses on disfunctional coping mechanisms that may perpetuate distress or hinder recovery following trauma or loss. Finally, the Personal characteristics/risk factors domain comprises a set of six personality-trait items that could be associated with greater vulnerability to trauma and stress response. Items responses are coded in a dichotomous way (yes/no) and domain scores are obtained by counting the number of positive answers. Higher scores correspond to greater trauma- and stress-related symptomatology. All Kuder–Richardson coefficients exceeded the minimum standard of .50, test–retest/inter-rater reliability was excellent, with infraclass correlation coefficients values exceeding .90 for each of the domains [39].

Statistical analysis

A Kruskal–Wallis t tests and chi-square analyses were used to compare age and gender. Due to non-normal data distribution, non-parametric methods were applied. A Kruskal–Wallis test assessed differences in AdAS Spectrum, TALS-SR, and CS scores among BPD, FED, and NCs. To control for potential Type I error inflation due to multiple comparisons, Bonferroni correction was applied to all post hoc tests and correlation analyses. Only results surviving the corrected significance threshold were interpreted as statistically significant.

Spearman correlations examined relationships between AdAS Spectrum, CS, and TALS-SR scores. Correlation analyses were conducted on the overall sample to examine dimensional associations among ATs, post-traumatic symptoms, and catatonic features across both clinical and non-clinical participants. This approach aligns with the spectrum-based conceptual model adopted in the study, which posits that these dimensions vary continuously along a neurodevelopmental continuum, rather than existing as discrete diagnostic constructs.

Multinomial logistic regression assessed whether total scores predicted inclusion in BPD or FED groups versus NCs. Two additional logistic regressions evaluated which questionnaire domains predicted diagnostic group membership. Finally, two linear regressions (one per clinical group) examined whether AdAS Spectrum and TALS-SR total scores predicted CS total scores.

Results

The total sample included 159 participants, divided into three diagnostic groups: 49 with BPD, 55 with FED, and 55 non-clinical controls NCs. Within the FED group, 15 participants were diagnosed with AN, while 40 were diagnosed with either BN or BED.

The three groups were homogeneous in age and gender (see Table 1).

Table 1.

Age and gender comparison between groups

	BPD (Mean ± SD)	FED (Mean ± SD)	NCs (Mean ± SD)	Test statistic	p
Age	30.06 ± 10.43	35.62 ± 14.04	33.96 ± 10.71	5.783	.055

	n (%)	n (%)	n (%)	Chi-square	p
Gender
F	44 (89.8%)	50 (90.9%)	47 (85.5%)	.903	.637
M	5 (10.2%)	5 (9.1%)	8 (14.5%)

BPD borderline personality disorder, FED feeding and eating disorder, NCs non-clinical sample

Results from the Kruskal–Wallis test showed that both BPD and FED groups scored significantly higher than NCs across all domains and total scores of the AdAS Spectrum, TALS-SR, and CS questionnaires. No significant differences were found between BPD and FED, except in the TALS-SR Personal Characteristics and Risk Factors domain, where BPD scored higher (Table 2).

Table 2.

Comparison AdAS Spectrum, TALS-SR and CS scores between BPD, FED and NCs

	BPD (Mean ± SD, mean rank)	FED (Mean ± SD, mean rank)	NCs (Mean ± SD, mean rank)	Statistic (H)	p	Pairwise comparison
AdAS Spectrum
Child/adolesc	11.47 ± 3.95, 110.44	10.02 ± 4.28, 96.99	3.14 ± 3.50, 35.89	79.723	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .136
Verbal comm	8.45 ± 3.39, 105.24	7.84 ± 3.92, 97.58	2.69 ± 2.94, 39.93	64.765	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = 396
Non-verbal comm	13.59 ± 5.03, 105.18	12.89 ± 4.81, 101.83	3.89 ± 3.81, 35.74	78.127	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .710
Empathy	5.61 ± 2.90, 105.77	4.73 ± 2.88, 94.45	1.54 ± 2.16, 42.59	57.942	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .208
Inflex. and adherence to routine	19.35 ± 6.96, 102.94	19.22 ± 7.31, 102.60	6.67 ± 6.00, 36.96	73.624	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .970
Restricted interests and rumination	11.65 ± 4.95, 105.28	10.73 ± 4.39, 98.83	3.56 ± 3.61, 38.65	68.544	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .475
Hyper/hyporeact. to sensory inputs	8.14 ± 3.72, 108.70	6.89 ± 3.78, 96.67	1.98 ± 2.64, 37.75	73.091	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .182
Total score	78.26 ± 25.13, 108.03	72.31 ± 24.99, 100.82	23.49 ± 21.63, 34.21	83.823	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = 425
Bonferroni correction applied: significant for p < .007
TALS-SR
Loss events	4.53 ± 1.78, 98.37	4.50 ± 2.24, 96.96	1.84 ± 1.56, 43.35	52.329	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .875
Grief reactions	12.79 ± 6.48, 95.39	13.40 ± 7.15, 98.43	4.53 ± 5.17, 44.61	48.092	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .735
Pot. traum. events	6.43 ± 3.39, 100.31	5.98 ± 3.73, 93.92	1.91 ± 2.29, 44.49	49.213	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .475
Reaction to losses or upsetting events	9.84 ± 4.19, 104.26	8.44 ± 4.46, 93.05	2.84 ± 3.22, 41.80	57.927	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .212
Re-experiencing	4.86 ± 2.60, 97.23	4.40 ± 2.94, 89.75	1.87 ± 2.24, 51.17	32.303	<.001*	NC-FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .401
Avoid. and numb	5.43 ± 2.76, 103.19	4.86 ± 3.13, 94.50	1.20 ± 1.76, 41.37	59.263	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .330
Maladaptive coping	3.20 ± 2.25, 102.76	2.52 ± 2.40, 89.44	.49 ± 1.07, 46.55	48.024	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .125
Arousal	2.73 ± 1.72, 101.31	2.33 ± 1.86, 89.67	.56 ± 1.03, 47.62	44.433	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .180
Pers. charact./risk factors	2.90 ± 1.39, 107.26	2.21 ± 1.89, 86.19	.62 ± 1.01, 45.61	53.367	<.001°	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .016*
Total score	52.71 ± 18.17, 104.41	48.65 ± 24.43, 96.00	15.85 ± 15.94, 38.87	66.266	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .350
Bonferroni correction applied: significant for p < .005
CS
Psychomotor activity	9.22 ± 4.08, 102.98	9.33 ± 3.72, 104.05	2.44 ± 2.59, 35.47	79.063	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .905
Verbal responses	4.41 ± 2.49, 99.23	4.71 ± 2.50, 103.35	1.00 ± 1.53, 39.52	66.429	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .646
Repetitive movements	2.51 ± 2.00, 92.72	2.78 ± 1.93, 100.29	.74 ± .97, 48.37	42.132	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .393
Artificial expressions	1.92 ± 1.73, 94.84	1.94 ± 1.70, 95.15	.47 ± .77, 51.63	34.540	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .971
Oppositivity	3.61 ± 2.23, 102.24	3.33 ± 1.88, 99.06	.74 ± 1.14, 41.12	61.737	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .721
Response to instructions	2.90 ± 1.93, 91.50	3.25 ± 1.89, 99.13	1.14 ± 1.47, 50.81	35.672	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .379
Automatisms	4.45 ± 2.84, 101.42	3.85 ± 2.44, 94.55	1.33 ± 1.60, 46.37	46.266	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .443
Impulsivity	5.73 ± 3.54, 105.74	4.87 ± 3.38, 97.18	.84 ± 1.37, 39.88	66.140	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .339
Total score	34.75 ± 15.31, 103.97	34.07 ± 13.08, 104.72	8.71 ± 8.61, 33.93	84.281	<.001*	NC–FED: p < .001*
						NC–BPD: p < .001*
						BPD–FED: p = .934
*significant for p < .05; Bonferroni correction applied: significant for p < .006

BPD borderline personality disorder, FED feeding and eating disorder, NCs non-clinical sample, AdAS Spectrum Adult Autism Subthreshold Spectrum, TALS-SR Trauma and Loss Spectrum—Self-Report, CS Catatonia Spectrum

Spearman correlations showed significant positive associations between all AdAS, CS, and TALS-SR domains and total scores (Tables 3, 4, 5).

Table 3.

Spearman correlation coefficients between AdAS Spectrum and CS domains scores in the overall sample

	Psychom. act	Verbal resp.	Rep. mov	Artificial exp	Oppos	Resp. to instr	Autom	Impulse	Total score
Child/adole	.619**	.544**	.509**	.474**	.508**	.310**	.455**	.502**	.628**
Verbal comm	.612**	.557**	.512**	.471**	.603**	.399**	.534**	.567**	.676**
Non-verbal comm	.757**	.617**	.557**	.488**	.668**	.523**	.620**	.607**	.775**
Empathy	.590**	.518**	.432**	.337**	.426**	.250**	.354**	.365**	.534**
Inflex. and routine	.705**	.557**	.644**	.514**	.601**	.466**	.637**	.577*	.747**
Restrict. int	.749**	.638**	.571**	.517**	.624**	.449**	.602**	.583**	.759**
Hyper/hyporeact	.682**	.547**	.610**	.482**	.571**	.395**	.566**	.553*	.702**
Total score	.762**	.637**	.632**	.535**	.650**	.464**	.623**	.613**	.785**

* p < .05; ** p < .001

Table 4.

Spearman correlation coefficients between AdAS Spectrum and TALS-SR domains scores in the overall sample

	Child./adol	Verbal comm	Non-verbal comm	Empathy	Inflex./routine	Restrict. int	Hyper/hyporeact	Total score
Loss events	.455**	.443**	.465**	.395**	.561**	.497**	.540**	.534**
Grief reac	.462**	.451**	.542**	.370**	.568**	.542**	.529**	.567**
Pot. traum. events	.486**	.469**	.503**	.394**	.569**	.533**	.505**	.564**
Reac. to losses	.552**	.466**	.597**	.460**	.602**	.646**	.593**	.635**
Re-exp	.442**	.436**	.486**	.346**	.517**	.527**	.508**	.531**
Avoid./numb	.5326**	.443**	.616**	.403**	.595**	.569**	.573**	.610**
Maladap. coping	.519**	.502**	.511*	.399**	.543**	.530**	.535**	.572**
Arousal	.543**	.440**	.534**	.423**	.587**	.542**	.523**	.586**
Pers. charact	.521**	.527**	.584**	.395**	.579**	.572**	.534**	.605**
Total score	.586**	.543**	.640**	.469**	.674**	.657**	.630**	.685**

* p < .05; ** p < .001

Table 5.

Spearman correlation coefficients between CS and TALS-SR domains scores in the overall sample

	Psychom. act	Verbal resp.	Rep. mov	Artificial exp	Oppos	Resp. to instr	Autom	Impulse	Total score
Loss events	.526**	.510**	.413**	.372**	.447*	.507**	.383**	.459**	.534**
Grief reac	.575**	.490**	.469**	.412**	.477**	.503**	.428**	.499**	.567**
Pot. traum. events	.495**	.442**	.526**	.478**	.479**	.475**	.448**	.535**	.564**
Reac. to losses	.592**	.488**	.419**	.365**	.436**	.443**	.463**	.466**	.635**
Re-exp	.484**	.456**	.403**	.369**	.360**	.430**	.452**	.497**	.531**
Avoid./numb	.617**	.589**	.458**	.432**	.450**	.582**	.555**	.568**	.610**
Maladap. coping	.504**	.492**	.530**	.498**	.403**	.394**	.573**	.599**	.572**
Arousal	.582**	.497**	.464**	.360**	.494**	.439**	.439**	.573**	.586**
Pers. charact	.501**	.518**	.467**	.535**	.539**	.468**	.533**	.636**	.605**
Total score	.650**	.584**	.543**	.493**	.535**	.563**	.550**	.616**	.685**

*p < .05; ** p < .001

Multinomial logistic regression indicated that higher AdAS and CS total scores significantly predicted inclusion in the BPD or FED groups versus NCs (Table 6).

Table 6.

Multinomial logistic regression analysis with group category as the dependent variable and AdAS Spectrum, CS and TALS-SR total scores as independent variables

	B (SE)	EXP (B)	CI (95%)		p
			Lower limit	Upper limit
BPD
Intercept	−5.288 (.863)	–	–	–	<.001*
AdAS Spectrum total score	.039 (.014)	1.039	1.011	1.069	.007*
TALS-SR total score	.032 (.012)	1.032	.990	1.075	.133
CS total score	.113 (.039)	1.119	1.036	1.209	.004*
FED
Intercept	−4.517 (.783)	–	–	–	<.001*
AdAS Spectrum total score	.030 (.014)	1.030	1.002	1.059	036*
TALS-SR total score	.023 (.021)	1.024	.983	1.006	.257
CS total score	.123 (.040)	1.131	1.047	1.223	.002*

Classification table
	NC	BPD	FED	Percent correct
NC	50	2	3	90.9%
BPD	3	18	28	36.7%
FED	5	18	29	55.8%
Overall percentage	37.2%	24.4%	38.5%	62.2%

BPD borderline personality disorder, FED feeding and eating disorder

Cox and Snell R²: .561; Nagelkerke R²: 631; *significant for p < .05

Results of the logistic regression performed showed a predictive value of AdAS Spectrum Childhood Adolescence and Non-Verbal communication domains toward the inclusion in both BPD or FED groups, although the first appear to be more predictive, as reported by a greater EXP(B), of the inclusion in BPD and the second of the inclusion in FED group (see Table 7).

Table 7.

Multinomial logistic regression analysis with group category as the dependent variable and AdAS Spectrum domains scores as independent variables

	B (SE)	EXP (B)	CI (95%)		p
			Lower limit	Upper limit
BPD
Intercept	−4.662 (.750)	–	–	–	<.001*
Child/adole	.361 (.111)	1.434	1.153	1.784	.001*
Verbal comm	−.112 (.141)	.894	.679	1.177	.425
Non-verbal comm	.291 (.118)	1.338	1.062	1.686	.013*
Empathy	.087 (.167)	1.090	.786	1.521	.604
Inflex. and routine	−.006 (.077)	.994	.854	1.157	.973
Restrict. int	.015 (.126)	1.016	.793	1.301	.902
Hyper/hyporeact	.014 (.163)	1.015	.738	1.396	.929
FED
Intercept	−4.053 (.674)	–	–	–	<.001*
Child/adole	.282 (.107)	1.326	1.075	1.636	.008*
Verbal comm	−.077 (.135)	.926	.710	1.207	.568
Non-verbal comm	.297 (.115)	1.345	1.074	1.685	.010*
Empathy	.006 (.161)	1.006	.733	1.380	.972
Inflex. and routine	.077 (.073)	1.080	.936	1.246	.292
Restrict. int	−.016 (.121)	.984	.776	1.248	.892
Hyper/hyporeact	−.096 (.160)	.908	.664	1.242	.546

Classification table
	NC	BPD	FED	Percent correct
NC	52	1	2	94.5%
BPD	2	27	20	55.1%
FED	6	15	34	61.8%
Overall percentage	37.7%	27.0%	35.2%	71.1%

BPD borderline personality disorder, FED feeding and eating disorder

Cox and Snell R²: .550; Nagelkerke R²: .619; *significant for p < .05

For the CS, Psychomotor Activity predicted both BPD and FED, while Impulsivity was specific to BPD (Table 8).

Table 8.

Multinomial logistic regression analysis with group category as the dependent variable and CS domains scores as independent variables

	B (SE)	EXP (B)	CI (95%)		p
			Lower limit	Upper limit
BPD
Intercept	−3.835 (.668)				<.001*
Psychomotor activity	.255 (.122)	1.291	1.016	1.639	.037*
Verbal responses	.319 (.214)	1.376	.904	2.096	.136
Repetitive movements	.043 (.286)	1.044	.596	1.828	.881
Artificial expressions	.389 (.350)	1.476	.743	2.930	.266
Oppositivity	.436 (.257)	1.546	.934	2.560	.090
Response to instructions	−.118 (.250)	.889	.544	1.451	.637
Automatisms	−.152 (.239)	.859	.538	1.372	.525
Impulsivity	.377 (.179)	1.458	1.026	2.072	.036*
FED
Intercept	−3.923 (.666)				<.001*
Psychomotor activity	.244 (.120)	1.276	1.008	1.615	.042*
Verbal responses	.410 (.212)	1.507	.994	2.284	.053
Repetitive movements	.270 (.281)	1.310	.755	2.271	.337
Artificial expressions	.423 (.350)	1.527	.769	3.033	.227
Oppositivity	.457 (.257)	1.579	.954	.2.615	.076
Response to instructions	.125 (.244)	1.133	.703	1.828	.608
Automatisms	−.351 (.239)	.704	.440	1.125	.143
Impulsivity	.232 (.180)	1.261	.885	1.796	.199

Classification table
	NC	BPD	FED	Percent correct
NC	51	2	2	92.7%
BPD	4	28	17	57.1%
FED	6	12	37	67.3%
Overall percentage	38.4%	26.4%	35.2%	73.0%

BPD borderline personality disorder, FED feeding and eating disorder

Cox and Snell R²: .569; Nagelkerke R²: .641; significant for p < .05

Linear regression in the FED group showed AdAS total score significantly predicted higher catatonic features (Table 9), while in the BPD group, both AdAS and TALS-SR total scores were significant predictors (Table 10).

Table 9.

Linear regression analyses with CS total score as a dependent variable and AdAS Spectrum and TALS-SR total scores as independent variable in the FED group

	B (SE)	Beta	t	p
Intercept	7.633 (4.547)	–	1.679	.100
AdAS Spectrum total score	.316 (.061)	.601	5.150	<.001*
TALS-SR total score	.069 (.064)	.125	1.073	.289

AdAS Spectrum Adult Autism Subthreshold Spectrum, TALS-SR Trauma and Loss Spectrum—Self-Report

R²: .436; adjusted R²: .413; *significant for p < .05

Table 10.

Linear regression analyses with CS total score as a dependent variable and AdAS Spectrum and TALS-SR total scores as independent variable in the BPD group

	B (SE)	Beta	t	p
Intercept	−1.281 (7.241)	–	−.177	.680
AdAS Spectrum total score	.218 (.073)	.358	2.996	.004*
TALS-SR total score	.360 (.101)	.427	3.582	.001*

AdAS Spectrum Adult Autism Subthreshold Spectrum, TALS-SR Trauma and Loss Spectrum—Self-Report

R²: .367; adjusted R²: 340; *significant for p < .05

Discussion

In the present study, we aimed to analyze the presence and correlates of ATs, stress-related symptoms, and catatonic features among individuals with BPD, FED, and NCs. Results showed that BPD and FED subjects scored significantly higher than NCs in all three instruments, aligning with literature reporting high rates of ATs [11, 12, 16, 17, 19, 25, 40–44], stress-related symptoms [29, 30, 45–47], and catatonic manifestations [29, 30, 36, 46–49] in these populations. Specifically, literature has linked ATs with AN and increasingly with other FEDs [25].

To support our results, Dudas et al. [16] found that BPD individuals exhibited a strong drive to systemize, similar to individuals with high ATs, in line with a growing body of research identifying higher ATs prevalence in BPD [50]. Likewise, trauma has been shown to be a key factor in FED development, especially in BN, as reported in Brewerton’s review [51]. The higher scores of TALS in BPD support the close relationships and overlap between BPD and PTSD variously stressed in the literature [46, 52].

Evidence of catatonic symptoms in FED includes case reports of AN subjects developing catatonia [49, 53], while Dell’Osso et al. found increased catatonic traits in BPD, especially in psychomotor activity and impulsivity domains [36]. Interestingly, TALS-SR Personal Characteristics and Risk Factors domain was the only one to differ significantly between BPD and FED, with BPD scoring higher. This supports existing research indicating BPD individuals’ heightened stress sensitivity due to underlying personality traits and altered stress responses. Indeed, emerging evidence suggests that women with BPD may exhibit altered physiological stress responses compared to those with other personality disorders [54]. In this framework, Bandelow et al. highlighted greater rates of abuse and trauma in BPD than in controls [55], and Sinai et al. found that BPD women with sexual trauma were significantly exposed to PTSD symptomatology if compared to those without an history of sexual abuse [56]. This vulnerability may be mediated by maladaptive emotional regulation, which, in turn, could be underlain by the presence of ATs [57], contributing to self-harm and suicidality [58]. While to a lesser extent, trauma-related maladaptive behaviors also appear in FEDs. Johnsen et al. noted that individuals with the binge/purge subtype of AN experienced more non-interpersonal trauma than restrictive types [59]. Intriguingly, ToM and social cognition impairments have been identified in AN and BN, with differing brain activation (respectively, hyper- and hypo-activation in the fronto-temporal brain region) patterns linked to illness severity [60]. In line with this evidence, Monteleone et al. observed that BN and BED individuals are particularly reactive to emotional abuse, which relates to impulsivity and feelings of ineffectiveness [61] with trauma and PTSD also shown to negatively affect FED treatment outcomes [29, 30].

Interestingly, our results revealed significant positive correlations across all domains of the three questionnaires assessing autistic traits, catatonic symptoms, and trauma-related features. Literature supports these findings, highlighting a link between high autistic traits and greater exposure to traumatic events throughout life [62–65]. Ruminative thinking, narrow interests, and sensory reactivity may increase trauma exposure and vulnerability to post-traumatic symptoms in ASD [62], while cognitive inflexibility hinders adaptation to trauma, and communication issues can limit help-seeking [63]. Haruvi-Lamdan et al. emphasized maladaptive coping, sensory overstimulation, and social degradation as amplifiers of stress in ASD [64]. In addition, PTSD symptoms in ASD often co-occurs with anxiety and depression, where thought suppression and perseverative thinking contribute to symptom severity [66]. Similarly, the relationship between autism and catatonia has been widely documented, to the point that catatonia is included as a specifier of ASD diagnosis [15, 67]. To support this evidence, Yurumez et al. found catatonia in ASD children, linked to motor stereotypies and impaired functioning [68]. Dell’Osso et al. noted significant ASD–catatonia symptom overlap, with possible genetic underpinnings [69]. In this framework, a review showed catatonic symptoms in ASD often mirror ASD features, such as mutism, negativism, stereotypies, and motor abnormalities [36]. Shiloh et al. described a PTSD case presenting with catatonia, thus leading to speculate that catatonia may complicate PTSD presentations [70].

Our multinomial logistic regression confirmed that high AdAS Spectrum and CS scores significantly predicted inclusion in FED and BPD groups. Autistic traits are recognized as vulnerability factors for psychiatric disorders and are often elevated in FED, especially early onset cases [43, 71, 72]. AN, particularly among females, frequently overlaps with ATs [73], and catatonic features have been observed in FED, including cases of AN-related catatonia [49, 53, 74]. BPD and ASD share features, such as emotional dysregulation, social difficulties, self-harm, trauma susceptibility, and sensory reactivity [36, 46, 75–77]. Meanwhile, it is known that AT are common among BPD subjects and can play a significant role for suicidality symptoms as well as for lifetime exposure to physical and sexual abuse [11]. In addition, it seems that AT may have an important impact for the development of post-traumatic stress disorders among BPD [13]. The predictive role of catatonic and autistic traits for FED and BPD supports a dimensional overlap between autism and catatonia [15, 67]. Though still understudied, catatonic symptoms have been documented in BPD, especially in the Psychomotor Activity and Impulsivity domains of the CS, suggesting frequent motor and automatic behaviors in this population [36].

Our results identified the AdAS Spectrum Childhood and Adolescence and Non-Verbal Communication domains as significant predictors of inclusion in both BPD and FED groups. The Childhood and Adolescence domain was more predictive of BPD, while Non-Verbal Communication better predicted FED inclusion. This aligns with literature indicating early behavioral abnormalities and interpersonal trauma in BPD patients [55, 78–80]. Notably, the Childhood and Adolescence domain captures social difficulties, such as bullying or isolation, potentially marking early pathways toward BPD [57]. These findings support early intervention strategies during childhood for BPD [81].

Conversely, impaired non-verbal communication emerged as the most relevant autistic trait for FED. FED patients often exhibit social phobia, difficulty with facial/emotional recognition, and altered expression [61, 82, 83]. Harrison reported FED individuals avoid eye contact and engage in self-directed gestures more frequently than controls [82]. Monteleone linked such behaviors to stress responses [61], while other studies noted high alexithymia and low empathy in AN women [84–88]. Though emotion regulation may improve with AN recovery, interpersonal deficits often persist [89]. BN patients also show similar impairments, though typically less severe [84, 90, 91]. Still, some studies found no significant difference between FED and controls in emotion recognition [91–93].

Regarding catatonic traits, the CS domain Psychomotor Activity predicted inclusion in both clinical groups, while Impulsivity was specific to BPD. These results align with the prevalence of altered motor activity among both eating disorders and BPD [94, 95]. Meanwhile, some studies highlighted the presence of altered psychomotor activities underlined by neurological soft signs in BPD [95]. A systematic review by Penfold et al. hypothesized a possible association between chronic fatigue and BPD [95]. In addition, rumination, which is evaluated in the CS domain Psychomotor activity, seems to be significantly represented among BPD subjects and may play a role in the maintenance and exacerbation of the active phase of the disorder [96]. Chronic fatigue and rumination have also been reported in FED [97, 98]. On the other hand, impulsivity is notably a diagnostic feature of BPD [1], which has also been examined from a neurobiological point of view [99]. Shorub et al. analyzed BPD risk of suicidality based on impulsivity levels and pointed out a significant correlation between impulsivity levels and suicidality [100].

Finally, considering that catatonic features could be one of the most severe presentations in psychiatry, we focused on investigating which symptoms may predict catatonic features in each diagnostic group as expected, considering the above-described strong link between ASD and catatonia. Our results highlighted a predictive value of autistic traits on the development of catatonic features in both groups. Interestingly, in the BPD group, also the TALS-SR score exerted a predictive effect on the CS total, suggesting that in the illness trajectory this disorder, in addition to the importance of a neuroatypical basis, the role of trauma is also central. In line with these findings, the relationship between AT and catatonic features has been widely reported [37, 68, 101]. In addition, it is known that catatonia may be the resultant of traumatic experiences [35]. These results accounts for the data previously described and support the perspective of a longitudinal view of the disease trajectory, which starting from a neuroatypical basis, develops into different phenotypes depending on the characteristics of the subject and life experiences [46, 69].

Our work should be considered in light of some limitations. First, the sample size was relatively small; consequently, it is difficult to make comparisons between males and females in the same group, also considering that females are notably more affected by both FED [8] and BPD [102]. Moreover, the small number of male participants in our sample precluded meaningful analysis of gender-specific effects. In addition, the subgroup analyses, particularly those involving smaller samples, should be interpreted with caution due to limited statistical power. Second, this is a cross-sectional work, thus it does not allow us to draw conclusion on the temporal relationship between autistic traits, catatonic manifestation and trauma- and stress-related symptoms in BPD and FED. Third, given the fact that lots of our evaluation was made by self-report questionnaires, we cannot exclude misinterpretation and misunderstanding of symptoms by patients with the risk of over or under estimation of the symptomatology. Another methodological limitation is the absence of gold-standard diagnostic assessments to accompany the spectrum questionnaires. An additional limitation is the possibility of partial content overlap between the instruments used (AdAS Spectrum and CS in particular), which may contribute to the very high intercorrelations observed. Although these instruments were chosen, because they capture complementary dimensional aspects of neurodevelopmental, motor/behavioral and trauma-related psychopathology, some items indeed ask about similar phenomena (e.g., rumination, slowness, and repetitive behavior). Moreover, psychiatric comorbidities were not taken into account for the purpose of this study. Finally, the analytical approach was deliberately limited to correlations and linear regressions, which were appropriate for the exploratory scope and modest sample size of this study. While more complex modeling could be applied in larger data sets, our aim was to ensure statistical validity and interpretability within the present design.

Taken together, our findings suggest that while autistic traits, post-traumatic symptoms, and catatonic features are inter-related across BPD and FED, they combine in distinct ways in each disorder. In BPD, catatonic manifestations appear more strongly linked to trauma-related processes, whereas in FED, they seem primarily associated with autistic dimensions. This differentiation supports a dimensional model of psychopathology that bridges neurodevelopmental and trauma-based frameworks and underscores the need for integrative, personalized clinical approaches.

The present findings should be interpreted as exploratory given the modest sample size and the multidimensional scope of the study. While the results align with theoretical expectations, they warrant replication in larger, independent samples using longitudinal and multi-method designs to substantiate causal and diagnostic implications. It should also be noted that alternative conceptual frameworks exist regarding the relationships among ATs, post-traumatic symptoms, and catatonic features. While the present study interpreted these dimensions through a spectrum-based, transdiagnostic lens, other models emphasize categorical distinctions or disorder-specific mechanisms.

Strength and limits

A key strength of this study is the comprehensive assessment of autistic traits, post-traumatic symptoms, and catatonic features using validated dimensional instruments across distinct clinical populations and robust statistical analyses that support the reliability of the findings. The comparison with NCs and the subdivision of FEDs into AN and BN/BED adds granularity to the findings. However, the study is limited by its relatively small sample size and cross-sectional design, which restricts causal interpretations. Furthermore, reliance on self-report measures may introduce reporting bias or misinterpretation of symptoms.

What is already known on this subject?

Prior studies have shown that both BPD and FED are frequently associated with autistic traits and histories of trauma. Emerging literature has also begun to explore catatonic symptoms in these populations, but the interplay between autistic traits, trauma-related symptoms, and catatonia in BPD and FED remains under-investigated.

What your study adds?

This study is the first to demonstrate specific patterns of autistic, post-traumatic, and catatonic symptoms in BPD and FED patients, highlighting shared and distinct psychopathological trajectories. It shows that while autistic traits predict catatonia in both conditions, trauma-related symptoms predict catatonia only in BPD. These findings support a dimensional approach to psychopathology and suggest targeted avenues for early intervention, clinical assessment, and tailored treatment strategies.

Conclusion

In summary, our results provide preliminary evidence that ATs, post-traumatic symptoms, and catatonic features are significantly inter-related in individuals with BPD and FEDs. The data suggest that although there are shared associations among these domains, the specific pattern and strength of relationships differ between the two clinical groups. These findings are consistent with the notion of overlapping neurodevelopmental and trauma-related dimensions of psychopathology, although they do not establish causality. Given the exploratory, cross-sectional design and modest sample size, future research with larger and longitudinal cohorts, combining diagnostic and spectrally informed tools, is needed to clarify mechanisms, temporal dynamics, and clinical implications.

Author contributions

Conceptualization: All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Barbara Carpita, Gabriele Massimetti, Benedetta Nardi, Leonardo Massoni and Margarita Pustynnikova. The first draft of the manuscript was written by Benedetta Nardi, Leonardo Massoni, Margarita Pustynnikova and Barbara Carpita and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Availability of data and materials

No data sets were generated or analyzed during the current study. The data sets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee of Comitato Etico Regionale per la Sperimentazione Clinica della Toscana—sezione AREA VASTA NORD OVEST (protocol code: 23910_Dell’osso; date of approval: 20th April 2023).

Informed consent

Informed consent was obtained from all individual participants included in the study.

Competing interest

The authors declare no competing interest.