Unresolved-disorganized attachment adjusted for a general psychopathology factor associated with atypical amygdala resting-state functional connectivity.

Background: Recent research has identified a general psychopathology factor (GPF), which explains overlap in presentation of psychopathological symptoms. Unresolved-disorganized attachment (Ud) is another transdiagnostic risk factor that may be relevant to explain differences in patient characteristics within diagnostic classifications. Objective: In the current study, we examined unique relations of resting-state functional connectivity (RSFC) with Ud and GPF. Method: RSFC data were collected from a mixed group of adolescents (N = 74) with and without psychiatric disorder, as part of the Emotional Pathways' Imaging Study in Clinical Adolescents (EPISCA) study. Ud was measured using the Adult Attachment Interview (AAI). Associations between Ud, GPF, and RSFC of the amygdala and dorsal anterior cingulate cortex (dACC) and with amygdala-medial frontal connectivity were examined.
Results: Ud was positively associated with greater functional connectivity between the left amygdala and the left lateral occipital cortex, precuneus, and superior parietal lobule. Furthermore, Ud was negatively associated with left amygdala-medial frontal cortex connectivity. GPF was not significantly associated with dACC or amygdala connectivity. Conclusions: Atypical amygdala connectivity may reflect a vulnerability factor rather than a biomarker of psychopathology. The unique association of Ud and amygdala RSFC, adjusted for a GPF, across participants with and without various classifications of psychopathology illustrates that dimensional approaches based on the AAI may complement psychiatric classifications in clinical research and practice.

Introduction

Psychopathology and unresolved–disorganized attachment (Ud) (Bowlby, 1969, 1980; Hesse & Main, 2000) are different yet interrelated clinical constructs (Lyons-Ruth, Pechtel, Yoon, Anderson, & Teicher, 2016; Patalay et al., 2015; Riem et al., 2019) that can impair adolescent functioning. Ud is characterized by signs of disoriented and/or dissociated, disorganized narratives in cases of loss or abuse, which indicate simultaneous or sequential contradictory strategies to deal with the loss or other trauma, and often display a lack of reflective functioning (Fonagy, Steele, Moran, Steele, & Higgitt, 1991). Ud is thought to be a transdiagnostic risk factor that is relevant across psychopathologies (Lyons-Ruth & Jacobvitz, 2016; Riem et al., 2019) and accounts for patient characteristics not included in diagnostic categories. The general psychopathology factor (GPF), which represents lesser-to-greater severity of psychopathology across disorders, has also been identified as a transdiagnostic factor that represents clinical presentations on a continuum, contrary to diagnostic categorical classifications such as in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), which are not optimal for research, partly because of the overlap between diagnostic categories and difficulty classifying individual patients (Caspi et al., 2014; Patalay et al., 2015; Tackett et al., 2013). The Ud and GPF transdiagnostic factors, however, may provide a useful measure for clinical research, particularly for neuroimaging studies of brain function; however, this prospect has not yet been fully explored.

Early life stress and adversity have negative consequences for physical and mental health, attachment, and psychosocial adjustment across the lifespan (Anda et al., 2006; Felitti et al., 1998). As to mental health, experiences of loss and abuse increase the risk for psychopathology, including post-traumatic stress disorder (PTSD), and anxiety and depressive disorders (Cloitre et al., 2009; Fonagy et al., 1991; MacMillan et al., 2003; McLaughlin, Sheridan, & Lambert, 2014; Schmaal et al., 2016). Neuroimaging studies of child and adult psychopathology are increasingly investigating the effects of early life stress on psychopathology; however, they have neglected the role of attachment. Child maltreatment and neglect, particularly from parental perpetrators, may lead to Ud, which can be measured in patients as a disorganized/disoriented, incoherent, state of mind when narrating childhood attachment experiences (Bowlby, 1969, 1980; Hesse & Main, 2000). Current and future attachment relationships and the transition to adult functioning are negatively impacted by Ud (Hesse, 2016). A meta-analysis of over 200 adult attachment studies showed that the prevalence of Ud within a mixed clinical sample was 43%, which was significantly higher than in normative groups (Bakermans-Kranenburg & van IJzendoorn, 2009), evidence that Ud is a transdiagnostic risk factor that may increase vulnerability to various psychiatric disorders. Furthermore, Ud has a higher representation among individuals with a history of abuse, serious other trauma, or loss (Hesse, 2016; Hesse & Main, 2000).

In addition to negative effects on attachment relationships, early life adversity is associated with neurobiological alterations that may interfere with brain development (Rinne-Albers, van der Wee, Lamers-Winkelman, & Vermeiren, 2013). Specifically, child abuse and neglect has been associated with altered resting-state functional connectivity (RSFC) of the amygdala and dorsal anterior cingulate cortex (dACC), as well as attenuated cognitive control through the medial prefrontal cortex (Wang et al., 2014). The amygdala and dACC are considered crucial brain structures in detecting and responding to threats (Graham & Milad, 2011; Phelps & LeDoux, 2005), and show heightened reactivity to emotionally negative stimuli in individuals with a history of maltreatment (Teicher, Samson, Anderson, & Ohashi, 2016). The dACC amplifies this fear response to threat through excitation of the amygdala, whereas the medial prefrontal cortex inhibits amygdala activation through a negative-feedback cycle (Feng, Feng, Chen, & Lei, 2014; Schuwerk et al., 2014; Teicher et al., 2016), thereby reducing anxiety and the intensity of emotional reactions. However, there is evidence for impaired inhibition of the amygdala by the medial prefrontal cortex in individuals with a history of adversity (Wang et al., 2014). Impaired emotional regulation related to amygdala and dACC hyperactivation may, in turn, increase the risk for the development of psychopathology.

Indeed, heightened activation of the amygdala and dACC has been implied in PTSD (Lyons-Ruth et al., 2016; Shalev, Liberzon, & Marmar, 2017; Vermetten & Lanius, 2012) and several other psychiatric disorders known to be related to childhood adversity, such as depression and anxiety (Grant et al., 2014; Strawn et al., 2012). For example, in anxiety and depressive disorders the dACC was shown to be involved in location-specific fear network function and fear recovery (Lang et al., 2009; Mechias, Etkin, & Kalisch, 2010; Suarez-Jimenez et al., 2018) A meta-analysis observed abnormalities in the fronto-amygdala circuitry in individuals across the internalizing spectrum, possibly reflecting a general emotional disturbance that is shared across diagnostic classifications (Marusak et al., 2016). Thus, disruptions in the fronto-amygdala circuitry seem to play a general role in psychopathology and may underlie high levels of comorbidity. This may therefore be a potential neural substrate underlying GPF. Another possibility is, however, that these resting-state functional abnormalities are not a true biomarker of general psychopathology, but instead are the result of abuse experiences. Multiple studies provide evidence that brain abnormalities in maltreated individuals are not directly linked to psychopathology because they are found in maltreated individuals with psychopathology but also in resilient individuals with a history of maltreatment but without psychopathology (for a review, see Teicher et al., 2016). This raises the question of whether abuse experiences, Ud, and psychopathology have unique patterns of altered RSFC.

In the current study, we will therefore examine whether Ud and GPF show unique amygdala and dACC RSFC. We were specifically interested in amygdala and dACC connectivity as previous research points to altered connectivity in these regions in maltreated individuals as well as in individuals with psychopathology (Teicher et al., 2016). As previous studies indicate that fronto-amygdala circuitry disruptions are shared by psychiatric disorders (Marusak et al., 2016), we investigated unique associations between Ud, GPF, and connectivity with the amygdala and the medial frontal cortex as an a priori connection of interest. Ud was assessed with the ‘gold standard’ Adult Attachment Interview (AAI) (Cassidy, 2016; Hesse, 2016). It should be noted that Ud represents a current state of mind with respect to childhood attachment experiences, and is derived exclusively from the language used in the narrative about potentially traumatic experiences. Despite low levels of reflective functioning that may obscure self-reports on adverse experiences, the incoherence of the narrative may uncover mental distress not immediately accessible to the adolescent. Our study, therefore, adds to previous neuroimaging studies that examined neurobiological effects of abuse using retrospective self-report questionnaires, without taking into account whether the narrative about the trauma was (dis)organized. We hypothesized that Ud and GPF would show unique RSFC of the amygdala and dACC.

Methods

Design and sample

Seventy-four participants from the Emotional Pathways’ Imaging Study in Clinical Adolescents (EPISCA) study (van Hoof, van Lang, Speekenbrink, van IJzendoorn, & Vermeiren, 2015) (N = 77) were involved in the current study. They were recruited from mental health centres and local advertisements according to specified inclusion and exclusion criteria (van den Bulk et al., 2013; van Hoof et al., 2015) (see supplemental material) and available coded AAI (Hesse, 2016). Dropout was due to anomalous magnetic resonance imaging (MRI) findings (n = 2), technical scanning problems, or poor imaging data quality (n = 2). Within this group there were 21 adolescents with childhood sexual abuse (CSA), 28 adolescents with anxiety and/or depressive disorders (DEP), and 25 non-clinical adolescents (CNTR). Unresolved loss or trauma (continuous and categorical) was determined with the AAI coding system (Hesse, 2016). Information about pubertal status was missing for 10 participants. For these participants, pubertal status was imputed using gender and age.

Written informed assent and consent was obtained from all adolescents and their parents. Participants received financial compensation including travel expenses. The medical ethics committee of the Leiden University Medical Centre approved the study (nr. P08.175).

Procedure

After adolescents and their parents had given assent and consent to participate in the EPISCA study they filled out questionnaires, usually at home, and were tested for intelligence quotient (IQ) and interviewed for Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM-IV) classification and attachment representation at the clinic in separate appointments. Scanning was usually performed on separate days, depending on the availability of the scanner.

Adult Attachment Interview

The AAI (Hesse, 2016) (see supplemental material) is a one-hour-long semi-structured interview, validated for adolescents. The AAI asks how the interviewee thinks about the relationship with parents or other primary caregivers in his or her youth, how these experiences have influenced him or her, how the actual relationship with parents or other primary caregivers is, and whether there were any experiences of severe illness, separation, fear, trauma, or loss. The interviewee is asked to give specific examples supporting each evaluation. Not the autobiographical content, but rather the coherence of the narrative matters.

After transcription and coding of the AAI according to the manual (Hesse, 2016) by a certified coder, an attachment representation classification can be given. In organized attachment representations there is one coherent mental strategy with regard to attachment figures, either secure–autonomous or insecure. In Ud representations, different mental strategies with regard to attachment figures are used simultaneously or sequentially, which are often contradictory, as becomes apparent when coding the narrative. A scale score for unresolved loss or trauma of 5.5 or above also renders a classification unresolved–disorganized (Hesse, 2016) (see supplemental material and manual). In the current study, unresolved trauma was examined as a continuous variable to enhance statistical power. A log transformation was applied because the distribution was skewed. Since Ud is hypothesized to be a transdiagnostic factor for psychopathology, like no other (in)secure attachment representation (Cassidy, 2016; Hesse, 2016), we chose to use Ud in our connectivity analyses.

General psychopathology factor

To estimate the effects of psychopathology separately from Ud, we decided to use a GPF. The GPF represents lesser-to-greater severity of psychopathology that is associated with negative emotionality (Tackett et al., 2013), compromised brain integrity (Caspi et al., 2014), lower IQ, higher negative affectivity, and lower effortful control in 1954 children from a birth cohort, aged 6–8 years (Neumann et al., 2016). The GPF has also been shown a significant single-nucleotide polymorphism heritability of 38% (SE = 0.16, p = .008) (Neumann et al., 2016). The use of the GPF was also shown to be valid in girls (Lahey et al., 2015) and in young adolescents (Patalay et al., 2015). The GPF was estimated for our sample using parent and self-report measurements for behavioural and emotional problems in children and adolescents: the Youth Self Report (YSR) (Achenbach, 1991a), Child Behavior CheckList (CBCL) (Achenbach, 1991b), Revised Child Anxiety and Depression Scale (RCADS) (Chorpita, Yim, Moffitt, Umemoto, & Francis, 2000), Trauma Symptom Checklist for Children (TSCC) (Briere, 1996), Children’s Depression Inventory (CDI) (Kovačs, 1992), and Adolescent Dissociative Experiences Scale (A-DES) (Armstrong, Putnam, Carlson, Libero, & Smith, 1997). Principal component analysis was performed using these (sub)scales. The Kaiser–Meyer–Olkin statistic showed sampling adequacy (KMO = .92). There were two components with eigenvalues larger than 1 (eigenvalue component 1 = 9.24, eigenvalue component 2 = 1.40). The scree plot showed an inflection justifying the extraction of one component explaining 61.63% (see Table S2 for an overview of the loadings). Individual factor scores were calculated to estimate the GPF (Lahey, Krueger, Rathouz, Waldman, & Zald, 2017). Factor score coefficients were calculated using the regression method. These coefficients were multiplied with the (sub)scale scores to obtain factor scores, which represent individual standardized scores on the GPF, based on their scores on the constituent scales. All calculations were performed in SPSS with principal component analysis. See Table S1 and Figure S1 in the supplemental material for the mean psychopathology scores across the psychopathology groups.

Image data acquisition

Images were acquired on a Philips 3T MRI system (Philips Healthcare, Best, The Netherlands), equipped with a SENSE-8 head coil. Scanning took place at the Leiden University Medical Center. Before scanning, all participants were introduced to the scanning situation by lying in a dummy scanner and hearing scanner sounds. For each subject, a sagittal three-dimensional gradient-echo T1-weighted image was acquired (repetition time = 9.8 ms; echo time = 4.6 ms; flip angle = 8°; 140 sagittal slices; no slice gap; field of view = 256 × 256 mm; 1.17 × 1.17 × 1.2 mm voxels; duration = 4:56 min) as part of a larger, fixed imaging protocol. Resting-state functional MRI data were acquired, using T2*-weighted gradient-echo echo-planar imaging of 160 whole-brain volumes (repetition time 2200 ms; echo time 30 ms; flip angle 80°; 38 transverse slices; no slice gap; field of view 220 mm; in-plane voxel size 2.75 × 2.75 mm; slice thickness 2.72 mm; total duration of the resting-state run 6 min). Participants were instructed to lie still with their eyes closed and not to fall asleep.

Statistical analysis

Pre-statistics

The FEAT module of the FSL software (FMRIB’s Software Library, http://www.FMRIb.ox.ac.uk/fsl) (Smith et al., 2004) was used to apply the following pre-statistics processing: motion correction using MCFLIRT (Jenkinson, Bannister, Brady, & Smith, 2002), non-brain removal using BET (Smith, De Stefano, Jenkinson, & Matthews, 2001), spatial smoothing using a Gaussian kernel of full-width-at-half-maximum 6.0 mm, and high-pass temporal filtering (high-pass filter cut-off = 100.0 s). Functional scans were registered to the T1-weighted images, which were registered to standard space in order to calculate the transformation matrix for the higher level group analysis (Jenkinson et al., 2002).

Functional connectivity analysis

A seed-based correlation approach was used for the current study (Murphy & Fox, 2017). We created binary masks of the left and right amygdala and left and right dACC. Coordinates of the seed regions were similar to a previous study examining resting-state connectivity in relation to depressive symptoms in partly the same sample (amygdala: x = ± 22, y = −6, z = −16; dACC: x = ± 6, y = 18, z = 28) (Pannekoek et al., 2014). Masks were created as spheres (4 mm radius, similar to Pannekoek et al., 2014) centred on these coordinates. After transforming the masks to native space, the mean time series for each participant were extracted from the voxels in the seed regions. The times series of the left and right amygdala were then entered as regressors in a general linear model (GLM) to examine amygdala connectivity, and the left and right dACC time series were entered as regressors in a separate GLM to examine dACC connectivity. In addition, cerebrospinal fluid, white matter, and the global signal were added as regressors to the model to reduce the influence of artefacts caused by physiological signal sources on the results (Fox & Raichle, 2007).

The temporal derivative of each regressor was added to the model, similarly to Pannekoek et al. (2014), resulting in 10 regressors in each model. Motion parameters were also added to the model. Contrasts of interest were the parameter estimates corresponding to the regressor of the left and right amygdala and left and right dACC, which represents functional connectivity with that region. First-level analyses were performed in native space. These first-level contrast images and the corresponding variance images of connectivity with each seed region were transformed to standard space and submitted to second-level mixed-effects group whole-brain analyses using FMRIB’s Local Analysis of Mixed Effects (FLAME). In the second-level analysis, the positive and negative correlations between amygdala and dACC connectivity and unresolved loss and trauma score and GPF were assessed. Unresolved status and GPF were included in the same model, as the aim of the current study was to examine unique correlates of Ud and GPF. Composite score age and pubertal status, gender, and IQ were confound regressors in the model. Non-parametric permutation inference was conducting using FSL’s randomize with threshold-free cluster enhancement (TFCE) to obtain family-wise error-corrected clusters (p < .05) (Winkler, Ridgway, Webster, Smith, & Nichols, 2014). Finally, because of the prominence of the medial frontal cortex in previous studies of social–emotional processing (Crone, 2014; Waugh, Lemus, & Gotlib, 2014) and general psychopathology (Marusak et al., 2016), connectivity between the amygdala and a medial frontal cortex region of interest (ROI) was examined. The left and right medial frontal cortex was anatomically defined using the Harvard–Oxford cortical atlas (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/Atlases) and used as an inclusive mask for the analysis of amygdala connectivity.

Results

Clinical sample characteristics

The mean age of the participants was 15.42 years (SD 1.67, range 12–20 years), and they had a total mean IQ of 103.28 (SD 8.89, range 81–119); 85.1% (n = 63) of the participants were female. As to cultural background, 1.4% were Asian (CSA n = 1), 93.2% were Caucasian (CSA n = 20, DEP n = 25, CNTR n = 24), 1.4% were Surinamese (DEP n = 1), and 2.7% were Latin American (DEP n = 2). Four adolescents (5.4%; CSA n = 2, DEP n = 2) were on stable selective serotonin reuptake inhibitor (SSRI) use (three fluoxetine, one sertraline). Pubertal stage was assessed according to the following categories using the Pubertal Development Scale (PDS) (Petersen, Crockett, Richards, & Boxer, 1988): prepubertal (CSA n = 1), early pubertal (DEP n = 1), midpubertal (CNTR n = 6, DEP n = 1), late pubertal (CSA n = 10, DEP n = 14, CNTR n = 13), and postpubertal (CSA n = 10, DEP n = 12, CNTR n = 6).

Clinical assessment of this sample (as detailed in the supplemental material) revealed that the mean score for post-traumatic symptoms (TSCC) was 34.13 (SD 22.72, range 0–98), for depression (CDI) 12.84 (SD 9.17, range 0–40), for anxiety (RCADS) 25.88 (SD 14.96, range 0–70), for dissociation (A-DES) 1.44 (mean total score; SD 1.42, range 0–6.37), for youth self-report problems (YSR) 18.78 (SD 11.13, range 0–44), and for reported internalizing problems by parents (CBCL) 13.60 (SD 9.68, range 0–42).

Unresolved loss or trauma score (AAI)

The AAI (Hesse, 2016) mean score for unresolved loss or trauma in this sample was 2.42 (SD 1.81, range 1–8). Based on the AAI (Cassidy, 2016; Hesse, 2016), 36.5% of the adolescents were classified as secure (CNTR n = 13, DEP n = 11, CSA n = 3), 41.9% as dismissive (CNTR n = 11, DEP n = 11, CSA n = 9), and 21.6% as Ud (CNTR n = 1, DEP n = 6, CSA n = 9). No adolescents in this sample were classified as preoccupied (Cassidy, 2016; van Hoof et al., 2015). See Table 1 for the mean GPF, age, IQ, and PDS scores for the Ud and non-Ud groups and see Table S1 for general psychopathology scores for the separate groups (CSA-PTSD, internalizing, control, and Ud versus non-Ud). There was no significant correlation between Ud and GPF (Pearson r = .203, p = .083, covariance 0.84).

Table 1.

General psychopathology scores, age, total IQ, and PDS scores for the Ud and non-Ud groups.

	Ud (N = 16)		Non-Ud (N = 58)
	M	SD	M	SD
GPF	0.38	0.97	−0.11	0.99
Age	15.56	1.63	15.38	1.69
Total IQ	99.38	8.40	104.36	8.89
PDS scores	4.19	0.98	4.22	0.73

M, mean; SD, standard deviation; GPF, general psychopathology factor; IQ, intelligence quotient; PDS, Pubertal Development Scale; Ud, unresolved–disorganized attachment.

Amygdala connectivity

Whole-brain analysis showed a significant positive association between Ud and connectivity of the left amygdala, with left lateral occipital cortex (LOC), precuneus, and superior parietal lobule (Brodmann area 7), after adjusting for GPF, age, pubertal status, IQ, and gender (family-wise error corrected p < .05) (Table 2 and Figure 1(a)). The whole-brain analysis did not reveal significant associations between amygdala connectivity and GPF.

Table 2.

Cluster size, lowest p-value, and coordinates of the significant clusters resulting from the analyses with the amygdala as seed region.

Contrast	Region	Voxels	p	x	y	z
Unresolved loss and trauma +	Left superior parietal lobule	80	.02	−20	−56	48
	Left superior parietal lobule	25	.04	−24	−52	34
	Left lateral occipital cortex	8	.05	−28	−66	34
Unresolved loss and traumaa –	Right medial frontal cortex	189	.02	12	50	−16

aResults from the region of interest (ROI) analysis with the medial frontal cortex as a priori defined ROI.

Figure 1.

(a) Significant positive association between unresolved loss and trauma and left amygdala–lateral occipital cortex connectivity, resulting from the whole-brain analysis, Threshold Free Cluster Enhancement (TFCE) family-wise corrected, p < .05. (b) Significant negative association between unresolved loss and trauma and left amygdala–medial frontal cortex connectivity, resulting from the region of interest (ROI) analysis, TFCE family-wise corrected, p < .05. The right side of the brain corresponds to the left hemisphere and vice versa.

Connectivity between the amygdala and the medial frontal ROI was significantly and negatively correlated with Ud, controlling for GPF, age, puberty status, IQ, and gender (family-wise error corrected p < .05) (Table 2 and Figure 1(b)), indicating reduced connectivity in individuals showing higher levels of unresolved loss and trauma. No significant association was found between GPF and amygdala–medial frontal cortex connectivity.

dACC connectivity

The whole-brain connectivity analyses with the dACC as seed region did not reveal associations with Ud or GPF (however, see supplemental material).

Discussion

The aim of this study was to investigate whether unresolved loss or trauma (Ud) as assessed with the AAI and a GPF were differentially associated with amygdala and dACC functional connectivity in a mixed group of adolescents with CSA-related PTSD, anxiety and depressive disorders and without psychiatric symptoms. After adjusting for GPF, pubertal status, age, gender, and IQ, we found that individuals with higher levels of Ud showed stronger left amygdala connectivity with the lateral occipital cortex, precuneus, and left superior parietal lobule compared to individuals with lower levels of Ud. In addition, Ud was negatively associated with left amygdala–medial frontal cortex connectivity. Our study suggests that across diagnoses, Ud is associated with specific RSFC of the left amygdala. This finding means that individuals’ functional connections vary according to attachment status, regardless of specific psychopathology.

The amygdala is part of the limbic network and is involved in fear, fight, flight, and freeze reactions to traumatic experiences controlled by the hypothalamic–pituitary–adrenal axis (Rinne-Albers et al., 2013). The urgency of these conditions, as well as their long-term impact, may hamper specific reflective functioning called mentalization, i.e. reflection about thoughts, emotions, and behaviour of the self and others (Luyten & Fonagy, 2015). A critical element of Ud is that individuals coded high for Ud often lack reflective functioning and awareness of their own psychological states. Our findings on the connectivity of the amygdala may reflect a lack of awareness of fear states, rather than regulatory deficits alone. Because Ud is assessed through coding of language rather than self-reported psychopathology, this indirect method of assessing mental distress may reveal the neural basis of psychopathology better than self-reported mental health.

Our finding that amygdala connectivity with the lateral occipital cortex is enhanced in individuals with Ud shows parallels with our previous study, which showed enhanced LOC–hippocampus connectivity in Ud (van Hoof et al., in press). Altered LOC activity is associated with atypical processing of emotional stimuli and has been implicated in higher level visual processing, including emotional scene perception (Sabatinelli et al., 2011). In addition, we observed enhanced functional connectivity with the precuneus and the left superior parietal lobe in adolescents with Ud. The precuneus is part of the default mode network (resting consciousness) and has been associated with self-consciousness, memory, directing attention in space, both when an individual makes movements and when imaging or preparing them, visuospatial mental operations, and mental imagery/modelling other people’s views (Cattaneo & Rizzolatti, 2009; Cavanna, 2007; Cavanna & Trimble, 2006). The left superior parietal lobe is involved in spatial orientation, and visual and sensorimotor input from the hand. Both precuneus and superior parietal lobe involvement have been implicated in PTSD with symptoms of dissociation, depersonalization, and derealization (Nicholson et al., 2015), and in particular the precuneus may be associated with Ud and its relative lack of reflective functioning.

In addition, a negative association between Ud and amygdala connectivity with the medial frontal cortex was found, possibly indicating altered amygdala inhibition by the medial frontal cortex. Such an impaired functional connectivity within the amygdala–medial (pre)frontal circuit increases the propensity for excessive fear as it promotes amygdala hyperactivity and diminished medial prefrontal control. It suggests less rational, cortical control by the medial frontal cortex, such as evaluating choices, handling errors, and cognition of social interaction (Crone, 2014; Waugh et al., 2014), and may explain why emotions and behaviour have dominance over cognitions in cases of Ud. Aghajani et al. (2016) previously suggested that disrupted basolateral amygdala–medial prefrontal connectivity might be a reliable neural marker of PTSD and a prominent feature of paediatric PTSD in part of the same sample (Rinne-Albers et al., 2013; Shin & Liberzon, 2010; Sripada et al., 2012; Wolf & Herringa, 2016). Our findings show that the atypical amygdala–medial (pre)frontal connectivity could additionally be a neural marker of Ud, as the current dysconnectivity was associated with Ud and not with GPF. As previous work demonstrated a rather strong association between PTSD and Ud, longitudinal research should investigate the developmental relationships between these phenomena and their underlying neural structures and functions (Harari et al., 2009).

Whereas the amygdala is part of the limbic network, the dACC is part of the salience network and is involved in selection of stimuli that are deserving of our attention, judgement, and discrimination, social sensitivity, and many autonomic functions (motor and digestive functions, regulation of blood pressure and heart rate) (Seeley et al., 2007). The anterior cingulate cortex contains spindle or von Economo neurons (von Economo & Koskinas, 1925), which allow rapid communications across areas, aiding the frontoparietal mirror neuron system in mentalization. The dACC connects primarily to cognitive brain regions such as the medial prefrontal cortex and is active in concert with the basolateral amygdala during the appraisal and expression of fear (Etkin, Egmer, & Kalisch, 2011; Teicher et al., 2016).

Evidence for the lateralization of amygdala function is accumulating (Baas, Aleman, & Kahn, 2004; Sergerie, Chochol, & Armony, 2008). The dominant notion seems to be that the right amygdala mediates relatively global and transient emotional responses, while the left amygdala seems to serve more specific and sustained forms of emotional responding. In more detail, the right basolateral amygdala is thought to encode precise affective features (e.g. punishment), while the left centromedial amygdala is thought to process general affective valence (e.g. good versus bad) (Styliadis, Ioannides, Bamids, & Papedelis, 2014). In this respect, it is interesting that Ud was found to be associated with the left amygdala, as attachment representation comprises a profound, sustained form of relating to others.

Our findings are consistent with previous studies showing that unresolved loss or abuse is a transdiagnostic risk factor for increased vulnerability to psychopathology in general (Lyons-Ruth et al., 2016; Riem et al., 2019). Moreover, these findings indicate that amygdala alterations previously found in patients with PTSD, depression, or anxiety disorders are not a specific biomarker for individual mental disorders, but instead may be common to several disorders with overlapping psychopathological symptoms. Amygdala atypical resting-state connectivity seems to be related to underlying factors associated with childhood attachment experiences which may be predisposing for vulnerability to fear generation and mental disorders (Admon, Milad, & Hendler, 2013).

Limitations

Some limitations should be considered. Owing to the fairly small sample size and the restricted ranges of age, IQ, gender, and ethnicity, the results should be replicated in larger samples with a wider array of clinical diagnoses. The GPF was based on self-report of symptoms and not on clinical interview scores. In addition, because of the cross-sectional design of our study, the results should be interpreted with caution to avoid reverse causality and, therefore, definitive conclusions about cause and effects cannot be drawn.

Conclusion

We found that Ud is uniquely related to amygdala RSFC across psychopathologies, possibly indicating that disrupted amygdala connectivity reflects a vulnerability factor rather than a biomarker of psychopathology. This study shows that the search for underlying dimensions of attachment and psychopathological symptoms across and beyond conventional diagnostic classifications may uncover commonalities and differences at the neural level explaining the aetiology of common disorders. Ud may be especially helpful in uncovering underlying dimensions of psychopathology, since coding of the AAI is not dependent on the (conscious) content of the narrative about past experiences or on a high level of reflective functioning about loss or trauma. Instead, the AAI is coded for coherence of mind, the (in)coherent use of language in the verbatim transcribed interview, without the participant’s awareness.

Click here for additional data file.