Psychobiological foundations of coping and emotion regulation: Links to maltreatment and depression in a racially diverse, economically disadvantaged sample of adolescent girls

Abstract

Adolescent risk for depression and passive or active suicidal ideation (PASI) involves disturbance across multiple systems (e.g., arousal regulatory, affective valence, neurocognitive). Exposure to maltreatment while growing up as a child or teenager may potentiate this risk by noxiously impacting these systems. However, research exploring how coordinated disturbance across these systems (i.e., profiles) might be uniquely linked to depressogenic function, and how past maltreatment contributes to such disturbance, is lacking. Utilizing a racially diverse, economically disadvantaged sample of adolescent girls, this person-centered study identified psychobiological profiles and linked them to maltreatment histories, as well as current depressive symptoms and PASI. Girls (N=237, M_age=13.98, SD=0.85) who were non-depressed/non-maltreated (15.1%), depressed/non-maltreated (40.5%), or depressed/maltreated (44.4%) provided morning saliva samples, completed questionnaires, a clinical interview, and a neurocognitive battery. Latent profile analysis of girls’ morning cortisol:C-reactive protein ratio, positive and negative affect levels, and attentional set-shifting ability revealed four profiles. Relative to Normative (66.6%), girls exhibiting a Pro-inflammatory Affective Disturbance (13.1%), Severe Affective Disturbance (10.1%), or Hypercortisol Affective Neurocognitive Disturbance (n=24, 10.1%) profile reported exposure to a greater number of maltreatment subtypes while growing up. Girls exhibiting these dysregulated profiles were also more likely (relative to Normative) to report current depressive symptoms (all three profiles) and PASI (only Pro-inflammatory Affective Disturbance and Hypercortisol Affective Neurocognitive Disturbance). Of note, girls’ cognitive reappraisal utilization moderated profile membership–depression linkages (depressive symptoms, but not PASI). A synthesis of the findings is presented alongside implications for person-centered tailoring of intervention efforts.

1. Introduction

Adolescence is characterized by elevated rates of depressive symptoms and related symptomatology (e.g., passive of active suicidal ideation, PASI; Nock et al., 2013). This may especially be the case for girls, given their increased sensitivity to relational stressors (Hankin et al., 2007) and developmental differences in arousal regulatory, affective, and neurocognitive function (Beauchaine et al., 2019; Shields et al., 2017; Slavich & Sacher, 2019). Recent reports have pointed to a precipitous rise in depression among Black girls in particular (Congressional Black Caucus, 2019; NIMH NOT-MH-21-120, 2020), who are three times more likely to report depression-related sequela than Black boys (Fitzpatrick et al., 2008). Utilizing a racially diverse, economically disadvantaged sample of adolescent girls, the current study aimed towards (a) a more nuanced understanding of the multisystem etiology of girls’ risk for depressive symptoms and PASI, (b) those relational stressors that contribute to such etiology, and (c) to identify putative psychosocial buffers to be leveraged in the design of more effective interventions.

1.1. Psychobiological Function and Depressive Symptomatology

Despite conceptualizations of adolescent risk for depression as involving coordinated disturbance across arousal regulatory, affective, and neurocognitive systems (Cha et al., 2019; Glenn et al., 2017), empirical studies have yet to provide an integrative understanding of how risk coalesces across these levels of analysis (Cicchetti & Dawson, 2002). Researchers have cited an over-reliance on single system approaches as a contributor to this empirical ambiguity (Hostinar et al., 2021), examining systems in isolation or analyzing between-group differences in the outcome of interest (e.g., MDD vs. No MDD) across systems. These studies have been useful in pointing to elevated negative affect and emotionality (Enns et al., 2003; Fergusson et al., 2000), low hedonic capacity (Auerbach et al., 2015), executive function deficits (e.g., poor attentional control and set shifting ability; Wilksinson & Goodyer, 2006), and dysregulated immuno-endocrine functioning (e.g., low cortisol:C-reactive protein (CRP) ratio reflective of pro-inflammatory state, Suarez & Sundy, 2017; high cortisol:CRP ratio reflective of hypercortisolemia, Landau et al., 2021) as potential etiological mechanisms. Here, we propose that a person-centered approach (Cicchetti & Rogosch, 1996; Warmingham et al., 2019) that spans these levels may provide additional nuanced detail of how coordinated psychobiological risk manifests within adolescent girls.

1.2. Psychobiological Function and Maltreatment History

The caregiver-youth relationship is one context in which some children experience acute and chronic forms of relational stress. Child maltreatment, specifically (i.e., severely adverse childhood experiences such as emotional, sexual, and physical abuse, as well as physical and emotional neglect; Cicchetti & Toth, 2016), is known to potentiate risk for depressive symptoms (Humphreys et al., 2020) and PASI (Duprey et al., 2021). Rates of child maltreatment are approximately twice as high for Black relative to White youth (Sedlak et al., 2010), with studies pointing to systemic biases in reporting systems as contributors (e.g., suspected maltreatment for Black youth is more likely to be reported, proceed to investigation, and be substantiated; Detlaff & Boyd, 2021). Also, Black youth with a history of maltreatment are more than five times as likely to report suicidality relative to those without (Fitzpatrick et al., 2008). Evidence suggests that one salient manner by which past maltreatment confers the risk is via insults sustained to youth psychobiological function. A history of maltreatment is known to contribute to dysregulated hypothalamic-pituitary-adrenal (HPA) axis and immune system processes (Coelho et al., 2014; Doom et al., 2013), affect dysregulation (Courtney-Seidler et al., 2014; Crowell et al., 2009), and impaired neurocognitive function (Cowell et al., 2015; Valentino et al., 2009). However, similar to the depression etiology literature, much of this work has examined these systems in isolation as well as by comparing youth with and without a history of maltreatment. As such, whether past maltreatment exposure contributes to heterogeneity in coordinated psychobiological function is not yet known.

1.3. Psychobiological Foundations of Coping and Emotion Regulation

From a coping and emotion regulation perspective, the manner by which past maltreatment contributes to emerging risk for psychopathology may be through adolescents’ compromised ability to efficaciously mange stressors or their involuntary stress responses to them (e.g., emotional reactivity, rumination). Indeed, evidence increasingly suggests that maltreatment exposure may contribute to this risk by noxiously affecting the psychobiological substrates that support coping and emotion regulation and, thus, constrain efficacious strategy use (Gruhn & Compas, 2020). As noted by Cicchetti and Bendezú (in press) and others (Compas et al., 2017), a more nuanced understanding of the psychobiological foundations of coping and emotion regulation is needed. Attention to specific aspects of executive function (e.g., attentional control, set shifting) may highlight foundations that avail adolescents the use of more sophisticated regulatory strategies (e.g., cognitive reappraisal, McRae et al., 2010). However, such a depiction is likely to be incomplete without concomitantly considering affective and arousal regulatory processes also requisite for efficacious coping and emotion regulation.

In healthy individuals, affective experience such as moderate negative affect helps orient a person to a source of stress (e.g., stressor, internal response to a stressor) and prompts the need to take action (Selye, 1936). Arousal regulatory processes such as well-orchestrated immuno-endocrine function (e.g., healthy cortisol suppression of inflammation) mobilize physiological resources needed to take action without the onset of sickness behaviors that might impede such action (Slavich & Irwin, 2014). Executive functions such as attentional control and cognitive flexibility support coping efforts insofar as they help a person shift attention away from a negative affect-inducing stressor or think differently about a stressor in a way that induces positive affect (Eisenberg & Zhou, 2016). For maltreated individuals, dysregulation across these multiple systems may increase risk for depression by making it difficult to effectively cope with stress or regulate emotions. For example, high negative affect and low positive affect may contribute to attentional biases that interfere with accurate stressor detection (Peckham et al., 2010). Excessive cortisol production that suppresses immune system function (e.g., cellular immunity) may result in an immunocompromised state that hampers ability to manage stressors (Shields et al., 2017). Insufficient cortisol production relative to inflammatory activity may promote fatigue and withdrawal behavior in place of active coping (Slavich & Irwin, 2014). Inability to shift one’s attention or perspective may contribute to perseveration on a stressor and limit cognitive access to alternate coping options (Uddin, 2021). Whether profiles of psychobiological function may be more (e.g., healthy) or less (e.g., unhealthy, linked to maltreatment) amenable to the use of specific strategies and, thereby, decrease or increase depressogenic function (Bendezú, Howland et al., 2021) is not yet known. If so, identification of such profiles may inform the tailoring of treatment efforts to girls’ psychobiological strengths and weaknesses.

1.4. The Current Study: Aims and Hypotheses

The current study had the following aims and hypotheses. Aim 1. Explore within-person patterns (i.e., profiles) of immuno-endocrine (i.e., cortisol:CRP ratio¹), affective (i.e., self-reported positive (PA) and negative (NA) affect), and neurocognitive (i.e., attentional set shifting) function in a racially diverse, economically disadvantaged sample of adolescent girls. Based on existing maltreatment, depression, and emotion regulation research in the developmental psychobiology literature², we expected to identify two profiles: Normative (e.g., moderate cortisol:CRP ratio, low NA, high PA, high attentional set shifting), Multisystem Disturbance (e.g., high or low cortisol:CRP ratio, high NA, low PA, low attentional set shifting). However, we also explored the possibility of additional profiles that previous research may have overlooked due to reliance on single-system approaches and predetermined patterns. Aim 2. Examine maltreatment history as a correlate of profile membership. We expected the likelihood of membership in Multisystem Disturbance (relative to Normative) to increase with a history of exposure to a greater number of maltreatment subtypes. Aim 3. Examine profile membership to depression linkages and cognitive reappraisal as a moderator of those linkages. We expected girls with Multisystem Disturbance (relative to Normative) profiles to report greater current depressive symptoms and PASI. We expected reappraisal to moderate profile membership to depression linkages, such that reappraisal would be associated with lower depressive symptoms and decreased likelihood of PASI for girls with Normative profiles, and not associated (i.e., stably elevated) or positively associated with depressive symptoms and PASI (i.e., stably high likelihood) for girls with Multisystem Disturbance profiles.

2. Method

2.1. Participants

Adolescent girls (N=237, M_age=13.98, SD=0.85) were recruited from an urban setting in upstate New York to participate in a randomized control trial of a depression intervention. We utilized data from the larger study’s baseline assessment. Participants consisted of three different groups: girls with subsyndromal symptoms of depression or who met criteria for clinical depression with (MDD/MAL; 44.4%) and without (MDD/NO MAL; 40.5%) a history of maltreatment, and controls who were non-depressed and non-maltreated (NO MDD/NO MAL; 15.1%). Most girls were Black (62.4%), followed by White (23.2%), multiracial (3.8%), Asian (0.4%), American Indian or Alaskan Native (0.4%), and Other (6.8%). Also, 12.2% of girls were Latina. Average annual household income was $28,353 (SD=$12,734).

2.2. Procedures

All study procedures were approved by the University’s Institutional Review Board (IRB). As described in detail elsewhere (MASKED et al., 2018), informed consent and assent were obtained from parents and girls, respectively. By design, all families participating in the study were required to be eligible for Temporary Assistance to Needy Families (TANF). Efforts to recruit girls with a history of maltreatment included collaborations with University Medical Center’s pediatric social workers, local schools and mental health organizations, and a liaison in the Department of Human Services (DHS) who examined Child Protective Services (CPS) reports for histories of maltreatment. Collaborators at these various institutions contacted potentially eligible families. Interested parents signed a release allowing their contact information to be shared with study staff. Nonmaltreating families were also recruited from a pool of participants similarly eligible for TANF. Restricting income enabled examination of variability in maltreatment-related phenomena without the potentially overriding contributions of socioeconomic status (Sedlak et al., 2010). To ensure the integrity of the data (e.g., variation due to reading ability and literacy), interviews and questionnaires were read to participants, who read along and marked their own answers. Interviewers were graduate students of clinical psychology and post-bac research assistants. In addition to coursework in cultural competency as part of their clinical training, graduate students received training on cultural sensitivity via our standard research protocol given our population, as did post-bac research assistants. Questionnaires were checked by interviewers to ensure that items were not skipped and that answers marked by participants were matched to the appropriate item. Neuropsychological tests were nonverbal, administered with touchscreen technology, with instructions read to participants. Girls and their mothers were interviewed concurrently in different rooms.

2.3. Primary Measures

2.3.1. Immuno-endocrine regulation.

Girls collected saliva samples on two consecutive weekdays via Trident stimulated passive drool immediately upon waking (M_{wake_time}=7:43am, SD=2:12), circumventing morning cortisol waking response (Susman et al., 2007) and morning routine (e.g., food or drink consumption, tooth brushing, smoking) data contamination. The decision to utilize morning cortisol and CRP was motivated by a desire to keep methodologically consistent with the only study to date that has examined the cortisol:CRP ratio with saliva in relation to depression in an adolescent sample (Landau et al., 2021). Girls stored samples in their home refrigerators until they could be collected by study staff and stored in a medical grade ultra-low temperature freezer (−40°C). Samples were shipped on dry ice to Salimetrics Laboratories (State College, PA) for assay. Salivary cortisol (μg/dl) and C-reactive protein (CRP) (pg/mL) were assayed using an enzyme immunoassay kit (Salimetrics, State College, PA). The average intra- and inter-assay coefficient of variation for cortisol and CRP were < 5.0%, 10.0% and 10.0%, 15.0%, respectively. Biomarker levels on each consecutive weekday were averaged to create single cortisol (Cort_ave) and CRP (CRP_ave) indices.

2.3.2. Affect.

Girls reported on the frequency of certain feelings and emotions in the past two weeks using the Positive and Negative Affect Scale for Children (PANAS-C; Laurent et al., 1999). Positive affect (PA) and negative affect (NA) scores were computed by averaging across 12 PA items (Interested, Excited, Happy, Strong, Energetic, Calm, Cheerful, Active, Proud, Joyful, Delighted, Lively; α=.92) and 15 NA items (Sad, Frightened, Ashamed, Upset, Nervous, Guilty, Scared, Miserable, Jittery, Afraid, Lonely, Mad, Disgusted, Blue, Gloomy; α=.89). Higher scores reflect greater frequency of experienced PA and NA.

2.3.3. Neurocognitive function.

Girls were administered the Cambridge Neuropsychological Test Automated Battery (CANTAB Eclipse, Cambridge Cognition, Cambridge, UK). CANTAB tests several neurocognitive processes and executive functions. We utilized the Intradimensional–Extradimensional (IDED) Set Shifting module, which tests capacity for attentional set formation, and the maintenance, shifting, and flexibility of attention. The module consists of nine stages. Participants complete a stage after providing six consecutive correct responses (i.e., establishing an attentional set). Stage 1 consists of simple discrimination (SD), where participants must correctly choose one of two stimuli of same dimension (e.g. “color-filled shapes”). Stage 2 consists of simple reversal (SR), where contingencies change and the previously correct stimuli is now incorrect. Stage 3 consists of compound discrimination (C_D), where the second dimension is introduced (e.g., “white line”) as a distractor and the participant must continue selecting the previously correct stimuli. At Stage 4, stimulus complexity increases as stimuli from the second dimension become superimposed on the stimuli from the first dimension. Stage 5 consists of compound discrimination reversal (CDR), where the stimulus from the first dimension with a superimposed stimulus from the second dimension that was previously incorrect now becomes the correct choice. Stage 6 consists of an intra-dimensional shift (IDS), where participants are presented with new versions of the stimuli from the two dimensions. Participants must continue to correctly select the stimuli from the first dimension. Stage 7 consists of an intra-dimensional shift reversal (IDR), where participants must correctly select the previously incorrect stimuli from the first dimension. Stage 8 consists of an extra-dimensional shift (EDS), where participants are again presented with new versions of the stimuli from the two dimensions. Here, participants must shift attention from the first dimension to the second dimension, choosing the correct stimuli from the second dimension rather than attempting to choose one of the two stimuli from the now incorrect first dimension. Stage 9 consists of an extra-dimensional shift reversal (EDR), where participants must shift the focus of their selection back to stimuli from the previously incorrect first dimension. The module terminates when a participant completes all nine stages or when the participant fails to establish an attentional set on any given stage after 50 trials. We utilized number of stages completed as an index of attentional set-shifting ability (M=8.21, SD=1.21, Min=2.00, Max=9.00).

2.3.4. Maltreatment history.

Girls completed the Childhood Trauma Questionnaire (CTQ; Bernstein et al., 2003), a 28-item measure assessing girls’ experiences of maltreatment, abuse, and neglect as a child and teenager while growing up. Girls rated a series of statements about specific maltreatment experiences on a 5-point scale (1=never true, 5=very often true). Maltreatment subtypes include emotional abuse (e.g., “People in my family called me things like stupid, lazy, or ugly.”), physical abuse (e.g., “People in my family hit me so hard that it left me with bruises or marks.”), sexual abuse (e.g., “Someone tried to touch me in a sexual way, or tried to make me touch them.”), emotional neglect (e.g., “I thought that my parents wished I had never been born.”), and physical neglect (e.g., “I didn’t have enough to eat.”). Subscales demonstrated good internal consistency (α=.71–.96). Following Walker et al. (1999), severity ratings for each subtype were dichotomized to indicate whether adolescents reported clinically significant maltreatment: emotional abuse (n_yes=63), physical abuse (n_yes=55), sexual abuse (n_yes=30), emotional neglect (n_yes=28), physical neglect (n_yes=34). A count variable was created by summing the number of maltreatment subtypes experienced.

2.3.5. Cognitive reappraisal.

Girls reported on their regulatory strategy use using the Emotion Regulation Questionnaire (ERQ; Gross & John, 2003). The scale includes 10 Likert-type items (1=Strongly Disagree, 4=Strongly Agree). Six items assess cognitive reappraisal (e.g., “I control my feelings about things by changing the way I think about them.”, α=.74) and four items assess expressive suppression (e.g., “I control my feelings by not showing them.”, α=.67). Researchers have observed that certain children (e.g., stress-affected, emotional and behavioral difficulties, girls) tend toward greater overall endorsement of strategy utilization items and have, thus, recommended the use of ratio scores to account for such potential differences in item endorsement base rates (Connor-Smith et al., 2000; Hilt et al., 2010). As such, we divided the cognitive reappraisal score (i.e., average of cognitive reappraisal items) by the sum of all scale item ratings. The resulting ratio score reflects the predominance of cognitive reappraisal utilized as a strategy for managing stressors and difficult emotions.

2.3.6. Depressive symptoms.

Girls reported on current depressive symptoms using the Beck Depression Inventory for Youth (BDI-Y; Beck et al., 2005). The scale includes 20 items (0=Never, 3=Always). Raw total scores (α=.93) were converted to T-scores to ensure scale rating integrity.

2.3.7. Passive or active suicidal ideation (PASI).

Girls and a parent reported on the frequency of girls’ current thoughts of death (e.g., “Sometime children who get upset or feel bad, wish they were dead or feel they’d be better off dead.”) and suicide (e.g., “Sometimes children who get upset or feel bad think about dying or even killing themselves.”) via the Kiddie Schedule for Affective Disorders and Schizophrenia-Present and Lifetime Version (K-SADS–PL; Kaufman et al., 1997). Consensus was established across interviews and the summary score was used. PASI (n_yes=55) was indicated by a threshold (i.e., “recurrent”) summary score of the frequency of thoughts of death or suicide.

2.4. Covariates

Covariates include chronological age, pubertal status, income to needs ratio (INR), body mass index (BMI), medication use, and morning saliva sample time. Though chronological age is confounded with pubertal status, it also affects HPA and immune system functioning in ways not directly mediated by reproductive hormones (e.g., size and structure of key regulatory glands; Linton & Dorshkind, 2004). Chronological age was, thus, controlled for in Aims 2 and 3 analyses. Girls completed the Pubertal Development Scale (Petersen et al., 1988), which consists of five Likert type items (1=no development, 4=development seems complete) about physical development, including body hair, skin changes, growth spurt, breast development and menarche (dichotomous). A mean score (M=3.35, SD=0.48) was computed across items (α=.62). Given that poverty has been linked to immuno-endocrine dysregulation (for review, see Jensen et al., 2017) and executive function deficits (for review, see Blair & Raver, 2016), we controlled for INR in Aims 2 and 3 analyses. INR is an index of household income relative to national poverty line norms (adjusted for number of family members residing in the household). Middle-income family INR ranges between 2.0 and 4.0 while an INR of 1.0 is indicative of poverty (Duncan et al., 1993; Evans & Marcynyszyn, 2004). Though the sample was low income, an examination of INR descriptive statistics suggested that there was variability with respect economic disadvantage to be considered (M=1.13, SD=0.49). Though 45.4% of the sample lived at or below the poverty line, 44.0% of families lived above the poverty line, with 10.6% of families approaching or meeting middle income status cutoffs. As obesity is also associated with a number of our study variables (e.g., inflammation, executive function; Spyridaki et al., 2016; Yang et al., 2018), we controlled for BMI in Aims 2 and 3 analyses. Girls’ height and weight were obtained during their study visit. Age-specific BMI percentiles calculated following Center for Disease Control (CDC) guidelines (M=81.93, SD=20.27). Medications known to influence immune-endocrine function or saliva collection were noted (e.g., corticosteroids, anxiolytics, mood stabilizers, selective serotonin reuptake inhibitors, birth control; Granger et al., 2009). A dichotomous variable was created for girls who were (yes=1) and were not (no=0) taking such medications (n_yes=47). Because wake time can impact morning cortisol levels (Kudielka & Kirchbaum, 2003), we controlled for girls’ waking saliva sample time in Aims 2 and 3 analyses.

2.5. Data Preparation

2.5.1. Cortisol:CRP ratio.

As per Suarez and Sundy (2017), we divided Cort_ave by CRP_ave and the resulting ratio was log₁₀ transformed for latent profile analysis. No outliers (+/− 3 SDs from the grand mean) were noted in the transformed cortisol:CRP ratio variable.

2.5.2. IDED set-shifting.

A ceiling effect emerged for IDED Set Shifting – stages completed, which has been observed in other studies of adolescent mood disorders (e.g., Dickstein et al., 2007; Dickstein et al., 2016). In all, 61.2% of girls completed Stage 9, 7.5% completed only up to Stage 8 (i.e., did not complete Stage 9), 29.3% completed only up to Stage 7 (i.e., did not complete Stage 8 or 9), 0.7% completed only up to Stage 4, and 1.4% completed only up to Stage 2. Following Leeson and colleagues (2009) as well as others (Barnett et al., 2005; Jazbec et al., 2007), a “pass/fail” variable was created (0 = module not completed, 1 = module completed). Given that the majority of participants who did not complete the IDED module had difficulty completing Stages 7 and 8 which consists of the extra-dimensional (ED) shift, our dichotomized IDED variable can be thought to reflect attentional set-shifting proficiency and inefficiency with respect to ED set-shifting ability in particular. This dichotomous variable was modeled as a categorical indicator in our latent profile analyses.

2.6. Overview of Analyses

2.6.1. Aim 1.

To identify subgroups of girls with unique psychobiological profiles, latent profile analysis (LPA) via MPLUS (Version 8; Muthén & Muthén, 1998-2017) was used and classified girls based on the extent to which they exhibited similar within-person patterns of psychobiological function across our four indicators of interest: cortisol:CRP ratio, negative affect, positive affect, IDED set shifting. Model specification began with a one-class solution and additional, more complex solutions (e.g., two-class, three-class) were evaluated across a series of model fit statistics: Akaike information criterion (AIC; Akaike, 1973), Bayesian information criterion (BIC; Schwarz, 1978), adjusted Bayesian information criterion (adjusted BIC; Sclove, 1987). Simulations studies have demonstrated BIC and Adjusted BIC to be conservative fit statistics for determining superior class specification (Magidson & Vermunt, 2004; Yang, 2006), with lower values indicating the most optimal solution. The entropy statistic was also utilized during model specification (Celeux & Soromenho, 1996), with values approaching 1 indicative of a superior solution. Though we expected to identify at least two profiles, we allowed both theory and model fit indices to guide us to the most informative solution (e.g., 3-profile, 4-profile; Masyn, 2013; Tein et al., 2013). At present, there are no firm guidelines on determining the appropriateness of subgroup size in LPA. However, LPA methodologists have previously suggested that subgroup sizes should not be smaller than 5% of the overall sample size (Weller et al., 2020). Once specified, girls were categorized into subgroups based on the highest posterior probability of class membership. After categorization, distinction analyses comparing mean indicator levels across systems and subgroups helped to characterize the profiles and informed our labeling conventions.

2.6.2. Aim 2.

Multinomial logistic regression was used to examine number of maltreatment subtypes as a correlate of profile membership. Simulation studies have shown that examining correlates of categorical profile membership (i.e., assigning participants to subgroups based on the highest posterior probability of class membership) is suitable for latent profile analysis solutions with an entropy greater than .80 (Clark & Muthen, 2009). Study covariates (e.g., age, pubertal status, INR, BMI, medication use, and morning saliva sample time) and number of maltreatment subtypes were examined in a single model.

2.6.3. Aim 3.

Three regression models were used to examine a) profile membership to depressive symptoms (multiple linear regression) and PASI (logistic regression) linkages, and b) cognitive reappraisal as a moderator of those linkages. As a more stringent test of the predictive capacity of our identified profiles, the number of maltreatment subtypes girls experienced was also included as a covariate in addition to our established set of study covariates³. In two initial models, profile membership and cognitive reappraisal were examined independently as predictors. In our final model, profile membership by cognitive reappraisal (grand mean centered) interactive effects were examined.

2.6.4. Post-hoc analyses.

To shed light on ways that a multisystem, person-centered approach provided unique information relative to conventional methods, we compared girls with and without maltreatment histories on our study variables of interest. Specifically, we conducted t-tests comparing these two groups on our psychobiological indicators, cognitive reappraisal, and depressive functioning.

3. Results

Results are organized by study aim. Descriptive statistics and bivariate correlations for key study variables are presented in Table 1. Cortisol levels across days were positively correlated, as were CRP levels. With the exception of positive and negative affect, which were inversely correlated, no significant between-person correlations across our psychobiological indicators of interest emerged, further supporting our aim of examining within-person profiles of psychobiological functioning. Negative affect was positively correlated with number of maltreatment subtypes, depressive symptoms, and PASI. Positive affect was not associated with number of maltreatment subtypes, but negatively correlated with depressive symptoms and PASI. Cortisol, CRP, and IDED set shifting were not correlated with number of maltreatment subtypes, depressive symptoms, or PASI. Cognitive reappraisal was positively correlated with positive affect and negatively correlated with negative affect, number of maltreatment subtypes and depressive symptoms. Depressive symptoms were positively correlated with PASI.

Table 1.

Descriptives and Correlations for Key Study Variables

		1.	2.	3.	4.	5.	6.	7.	8.	9.	10.	11.
1.	Cortisol (μg/dl) – day 1	—
2.	Cortisol (μg/dl) – day 2	.43*	—
3.	CRP (pg/mL) – day 1	−.11	−.07	—
4.	CRP (pg/mL) – day 2	−.11	−.07	.60*	—
5.	Negative affect	−.01	.04	−.04	−.08	—
6.	Positive affect	.01	.04	−.06	−.04	−.28*	—
7.	IDED set-shifting a	−.11	−.06	.08	.05	−.04	−.08	—
8.	Maltreatment subtypes	−.03	.08	−.05	−.07	.36*	−.17	.02	—
9.	Cognitive reappraisal	.04	.03	−.07	−.11	−.19*	.35*	−.04	−.28*	—
10.	Depressive symptoms	.06	.02	−.05	−.02	.67*	−.37*	−.08	.37*	−.35*	—
11.	PASI a	.11	.02	.01	−.03	.39*	−.21*	−.14	.19*	−.17	.45*	—
Means (Proportions)		0.33	0.30	6419.07	6613.28	1.90	3.11	(0.61)	0.91	0.55	48.94	(0.23)
SD		0.24	0.23	15104.7	17629.2	0.68	0.68	0.49	1.28	0.08	12.06	0.42
Min		0.01	0.02	102.60	135.11	1.00	1.08	0.00	0.00	0.25	34.00	0.00
Max		1.49	1.29	115322.7	181512.4	4.00	5.00	1.00	5.00	0.78	100.00	1.00

Note. CRP = C-reactive protein; IDED = Intra Dimensional-Extra Dimensional, PASI = Passive or active suicidal ideation. IDED set-shifting coded 0 for “module not completed” and 1 for “module completed.” PASI coded 0 for “not indicated” and 1 for “indicated.”

^a= Spearman’s Rho

^*p<.05.

3.1. Aim 1: Psychobiological Foundation Profiles

LPA model specification supported a four-profile solution (Table S1, see supplementary materials). Model fit indices did not uniformly settle on a single solution. As recommended (Masyn, 2013; Tein et al., 2013), we utilized both theory and model fit indices as conceptual and statistical guide posts and determined that the four profile solution was most informative. Profile specific means and standard deviations for each continuous indicator in each profile are presented in Table S2 (see supplementary materials). Probabilities are reported for binary indicators in each profile (e.g., probability of completing the IDED set-shifting module). To aid interpretation of the cortisol:CRP ratio, raw and standardized cortisol and CRP levels for each profile are presented in Table S3 (see supplementary materials). Profile specific deviations from the sample mean or proportion for continuous and binary indicators, respectively, in each profile are graphically depicted in Figure 1. The Normative profile was consistent with expectation, the largest of all groups (n=158, 66.6%), and characterized by moderate cortisol:CRP ratio levels, low negative affect levels, high positive affect levels, and moderate probability of completing the IDED set-shifting module. Three smaller subgroups emerged whose profiles reflected various within-person patterns of psychobiological dysregulation. The Pro-inflammatory Affective Disturbance profile (n=31, 13.1%) was characterized by the lowest cortisol:CRP ratio levels in the sample, high negative affect levels, low positive affect levels, but also the highest probability of completing the IDED set-shifting module in the sample. The Severe Affective Disturbance subgroup (n=24, 10.1%) was characterized by moderate cortisol:CRP ratio levels, the highest negative affect and lowest positive affect levels in the sample, and moderate probability of completing the IDED set-shifting module. The Hypercortisol Affective Neurocognitive Disturbance (n=24, 10.1%) subgroup was characterized by the highest cortisol:CRP ratio levels in the sample, high negative affect levels, moderate positive affect levels, and the lowest probability of completing the IDED set-shifting module in the sample.

Figure 1.

Profile specific deviations from the sample mean for psychobiological indices.

3.2. Aim 2: Maltreatment History Correlates of Psychobiological Foundation Profiles

Parameter estimates for our multinomial logistic regression models are shown in Table 2. Medication use and saliva sample time were not associated with our dependent variables in Aim 2 or 3 analyses and were, thus, removed in the interest of model parsimony.⁴ The Normative profile (i.e., largest, healthy) was used as the reference profile. Our covariate model was not significant. Neither age, pubertal status, INR, or BMI were associated with profile membership. Our maltreatment model was significant. As hypothesized, relative to Normative, girls exhibiting any of the three dysregulated profiles were more likely to have experienced a greater number of maltreatment subtypes while growing up.

Table 2.

Odds Ratios (95% Lower CI, 95% Upper CI) for Multinomial Logistic Regressions Predicting Profile Membership

Comparison Profile	Pro-inflammatory Affective Disturbance	Severe Affective Disturbance	Hypercortisol Affective Neurocognitive Disturbance
Age	1.08 (0.65, 1.78)	0.91 (0.49, 1.69)	1.23 (0.70, 2.16)
Pubertal status	0.77 (0.33, 1.76)	1.10 (0.37, 3.25)	0.86 (0.33, 2.27)
Income-to-needs ratio	0.99 (0.43, 2.31)	1.98 (0.75, 5.22)	1.15 (0.46, 2.89)
Body mass index	1.00 (0.98, 1.02)	1.01 (0.98, 1.04)	1.00 (0.97, 1.02)
Number of maltreatment subtypes	1.44* (1.04, 2.00)	2.06* (1.47, 2.89)	1.66* (1.18, 2.34)
X2 (df)	27.80 (15)
Nagelkerke’s R2	.14

Note. CI = confidence interval. The Normative profile served as the reference group.

^*p<.05.

3.3. Aim 3: Psychobiological Foundation Profile to Depressive Function Linkages

3.3.1. Depressive symptoms.

Parameter estimates for our linear regression models predicting current depressive symptoms are shown in Table 3. Neither age, pubertal status, INR, or BMI were associated with depressive symptoms. Number of maltreatment subtypes was positively associated with depressive symptoms. Relative to Normative, girls exhibiting any of the three dysregulated profiles were more likely to report elevated depressive symptoms. Cognitive reappraisal was negatively associated with depressive symptoms. A significant interaction between subgroup membership and cognitive reappraisal emerged (Figure 2). For Normative and Pro-inflammatory Affective Disturbance girls, cognitive reappraisal was not associated with depressive symptoms (i.e., stably low). However, for girls exhibiting the Severe Affective Disturbance or Hypercortisol Affective Neurocognitive Disturbance profiles, cognitive reappraisal was negatively associated with depressive symptoms.

Table 3.

Parameter Estimates for a Nested Taxonomy of Multiple Linear Regression Models Predicting Depressive Symptoms

	Profile		Cognitive Reappraisal		Profile × Cognitive Reappraisal
Independent Variables	B	SE	B	SE	B	SE
Intercept	32.33*	11.06	43.08*	11.27	30.48*	10.73
Age	0.62	0.78	0.76	0.76	0.62	0.75
Pubertal status	−0.67	1.34	−0.41	1.31	0.30	1.30
Income-to-needs ratio	1.03	1.29	1.62	1.28	1.23	1.26
Body mass index	0.03	0.03	0.03	0.03	0.02	0.03
Number of maltreatment subtypes	2.52*	0.54	2.29*	0.53	2.10*	0.53
Profile a
Pro-inflammatory Affective Disturbance	8.55*	1.89	8.04*	1.85	8.59*	1.84
Severe Affective Disturbance	16.47*	2.27	16.30*	2.22	16.42*	2.17
Hypercortisol Affective Neurocognitive Disturbance	13.74*	2.11	13.51*	2.06	13.39*	2.02
Cognitive reappraisal			−25.17*	7.56
Normative × cognitive reappraisal					−9.90	9.65
Pro-inflammatory Affective Disturbance × cognitive reappraisal					2.54	21.97
Severe Affective Disturbance × cognitive reappraisal					−50.54*	20.94
Hypercortisol Affective Neurocognitive Disturbance × cognitive reappraisal					−63.59*	23.59

^a= Normative served as the reference profile.

^*p<.05.

Figure 2.

Moderation effects of cognitive reappraisal on profile membership to girls’ depressive symptoms. Effects plotted at − 1 SD and + 1 SD for illustrative purposes.

3.3.2. PASI.

Parameter estimates for logistic regression models predicting PASI are shown in Table 4. Age, pubertal status, and INR were not associated with PASI. However, BMI and number of maltreatment subtypes were associated with increased likelihood of PASI. Relative to Normative, girls exhibiting the Pro-inflammatory Affective Disturbance or Hypercortisol Affective Neurocognitive Disturbance profiles (but not Severe Affective Disturbance) were more likely to report PASI.

Table 4.

Parameter Estimates for a Nested Taxonomy of Logistic Regression Models Predicting Passive or Active Suicidal Ideation

	Profile		Cognitive Reappraisal		Profile × Cognitive Reappraisal
Independent Variables	B	SE	B	SE	B	SE
Intercept	−4.87	3.36	−3.81	3.48	−5.69	3.52
Age	0.03	0.23	0.05	0.23	0.05	0.26
Pubertal status	0.08	0.39	0.11	0.39	0.12	0.41
Income-to-needs ratio	−0.20	0.40	−0.14	0.40	−0.22	0.41
Body mass index	0.03*	0.01	0.03*	0.01	0.03*	0.01
Number of maltreatment subtypes	0.50*	0.14	0.48*	0.14	0.45*	0.15
Profile a
Pro-inflammatory Affective Disturbance	1.40*	0.49	1.36*	0.49	1.32*	0.52
Severe Affective Disturbance	0.98	0.57	0.95	0.58	0.88	0.66
Hypercortisol Affective Neurocognitive Disturbance	1.19*	0.56	1.14*	0.56	1.12	0.60
Cognitive reappraisal			−2.78	2.31
Normative × cognitive reappraisal					3.39	3.42
Pro-inflammatory Affective Disturbance × cognitive reappraisal					−10.70	6.49
Severe Affective Disturbance × cognitive reappraisal					−15.16	7.94
Hypercortisol Affective Neurocognitive Disturbance × cognitive reappraisal					−9.38	7.80

^a= Normative served as the reference profile.

^*p<.05.

Reappraisal was not associated with PASI. No significant profile membership by cognitive reappraisal interaction emerged.

3.4. Post-hoc Analyses

A comparison of girls with and without maltreatment histories on our psychobiological indicators of interest, cognitive reappraisal, depressive symptoms, and PASI is shown in Table S4 (see supplementary materials). Differences between girls with and without maltreatment histories emerged on negative affect, positive affect, depressive symptoms, and PASI. No other significant differences emerged.

4. Discussion

Utilizing a racially diverse, economically disadvantaged sample of adolescent girls, the current study identified distinct profiles of psychobiological function that were meaningfully associated with girls’ past maltreatment exposure and current depressogenic functioning. Our adoption of a person-centered (Cicchetti & Rogosch, 1996), multiple levels of analysis (Cicchetti & Dawson, 2002) approach helped illustrate varied ways in which coordinated disturbance across the immuno-endocrine, affective valence, neurocognitive systems manifest within adolescents (Cha et al., 2019; Glenn et al., 2017). Specifically, latent profile analysis (LPA) of girls’ morning cortisol:CRP ratios, positive affect levels, negative affect levels, and attentional set shifting capacities revealed four profiles. These profiles connected with maltreatment and maladjustment in ways that potentially point to person-specific psychobiological mechanisms the may underlie maltreated girls’ risk for depression during adolescence. Interestingly, cognitive reappraisal to depression linkages varied across profiles, highlighting various sources of psychobiological resilience that might mitigate such risk by differentially supporting the use of this potentially efficacious strategy. Our findings provide nuanced detail of the psychobiological foundations of coping and emotion regulation (Compas et al., 2017), with implications for tailoring of prevention and intervention efforts targeting the emergence of depression in maltreated adolescent girls.

Our study identified four profiles, one reflecting healthy psychobiological function and three reflecting varied forms of psychobiological dysregulation. Negative affect disruptions were common across all three dysregulated profiles, whereas disturbance across other systems studied were not. Negative affect may, therefore, fail to differentiate among profiles of psychobiological dysregulation, consistent with the argument that negative affectivity is a non-specific feature of internalizing difficulties more broadly (Kotov et al., 2017). Person-centered, multilevel studies of adolescent psychobiological function have also shown elevated negative affect to be a common feature across identified dysregulated profiles (Bendezú, Calhoun et al., 2021; Koss et al., 2017). Our study suggests that concurrent attention to positive affect, immuno-endocrine, and neurocognitive function may provide a more nuanced and comprehensive understanding of the heterogeneity involved in adolescent girls’ psychobiological risk for depression.

To our knowledge, Landau and colleagues (2021) is the only study to date that has examined the morning cortisol:CRP ratio using saliva in relation depression in adolescents. No studies have examined the ratio in a racially diverse, economically disadvantaged sample of adolescent girls. Our multi-system, person-centered study helped to address this gap in the literature. In so doing, our results also help to clarify inconsistent findings from studies using variable-centered methods (e.g., linear regression) to examine the cortisol:CRP ratio as it pertains to internalizing symptomatology. Specifically, both high and low cortisol:CRP ratios were observed in our dysregulated profiles. Lower ratios have been associated with depression in females and stress-induced negative affect in depressed adults (Suarez et al., 2015; Suarez & Sundy, 2017). Higher ratios have been associated with anxiety in females and first-onset depression in at-risk adolescents (Landau et al., 2021; Suarez et al., 2015). In agreement with Landau and colleagues, our person-centered findings suggest the possibility that both lower and higher ratios within the same study and sample may reflect qualitatively distinct forms of immuno-endocrine dysregulation and, thus, index risk. If so, conclusions drawn from variable-centered methods (e.g., linear regression) that average across potential subgroups at both ends of the cortisol:CRP ratio distribution may be erroneous (e.g., concluding that higher cortisol:CRP ratios weakly confer risk for internalizing problems when in actuality both higher and lower cortisol:CRP ratios moderately or strongly confer risk).

4.1. Psychobiological Foundation Profiles

Relative to Normative, girls exhibiting the Pro-inflammatory Affective Disturbance profile presented with lower cortisol:CRP ratios. For these girls, lower cortisol:CRP ratios may reflect the inadequate release of cortisol relative to CRP, a pattern often attributed to chronic stress-related glucocorticoid resistance whereby immune cells (e.g., monocytes, macrophages) become less sensitive to cortisol’s anti-inflammatory signaling properties (Slavich & Irwin, 2014). Pro-inflammatory states can also contribute to depressive states (Anacker et al., 2011; Dantzer, 2001), which is consistent with the observation that these girls also exhibited high negative affect and low positive affect levels. Notably, our simultaneous attention to neurocognitive systems revealed a potential psychobiological protective factor: better than average attentional set shifting skills. The capacity to effortfully focus attention on task-related demands and intentionally shift this focus in service of goal-directed behavior is thought to be a core component for the modulation of emotion (e.g., felt sadness; Eisenberg & Zhou, 2016), perhaps reflected in these girls relatively mild elevations in negative affect. That Pro-inflammatory Affective Disturbance girls exhibited better than average attentional set-shifting capacity is also consonant with claims about the adaptive function of chronic stress-related attenuated cortisol release patterns, which are thought to protect adolescents’ developing brains (e.g., hippocampus, prefrontal cortex) against the potential neurotoxic effects of cortisol overexposure (Fries et al., 2005; Gunnar & Vazquez, 2001).

Girls exhibiting the Severe Affective Disturbance profile displayed psychobiological function characterized by disruptions primarily of the affective valence system. Both low positive affect and high negative affect have conceptual and empirical links to risk for depression (Chorpita & Daleiden, 2002; Olino et al., 2011). However, our concomitant consideration of both neurocognitive and immuno-endocrine functioning uncovered a set of psychobiological protective factors parallel to these girls’ affective disruptions. In addition to exhibiting attentional set-shifting abilities similar to that observed in Normative, girls exhibiting the Severe Affective Disturbance profile displayed a moderate ratio of morning cortisol to CRP levels. Such close correspondence between morning cortisol and CRP levels is believed to index homeostatic equilibrium between the HPA and immune system (Landau et al., 2021). Such countervailing effects are believed to support healthy stressor management, where by cortisol mobilizes metabolic resources needed to efficaciously respond to a stressor while corresponding inflammatory processes facilitate recovery (i.e., physical healing, limit infection) from stress exposure (Slavich & Irwin, 2014). Indeed, homeostatic immuno-endocrine equilibrium has been implicated as a biological buffer of girls’ risk for depression during the adolescent transition (for review, see Slavich & Sacher, 2019).

Relative to Normative, Hypercortisol Affective Neurocognitive Disturbance girls exhibited disrupted functioning across multiple (but not all) systems. First, these girls exhibited relatively higher morning cortisol:CRP ratios. Elevated cortisol:CRP ratios are thought to indicate the overproduction of cortisol to CRP (Guilliams & Edwards, 2010), a pattern often attributed to chronic stress-related glucocorticoid immunosuppression whereby excessive cortisol via its anti-inflammatory signaling properties unduly suppresses immune system function (e.g., cellular immunity). The resulting immunocompromised state can increase risk for psychopathology (e.g., depression, Slavich & Irwin, 2014). That Hypercortisol Affective Neurocognitive Disturbance girls also displayed the lowest attentional set-shifting abilities is further consistent with this claim, as excessive HPA activation and cortisol overexposure noxiously impact neurobiological circuits that support healthy executive function (Shansky & Lipps, 2013). These girls, however, also displayed normative levels of positive affect, consistent with a prior person-centered, multisystem study of depressed adolescent girls’ stress responsivity and its identified Hyperresponsive profile (Bendezú, Calhoun et al., 2021). Positive emotionality can be a source of resilient function (Curtis & Cicchetti, 2007) and buffer of negative emotions and related sequelae (Frederickson, 2004), which may explain these girls’ relatively mild elevations of negative affect.

4.2. Links to Maltreatment History

The validity of our profiles was supported through their connections with maltreatment. Relative to Normative, girls exhibiting profiles thought to reflect psychobiological dysregulation reported having been exposed to a greater number of maltreatment subtypes while growing up as a child and teenager. Exposure to maltreatment has been independently linked to youth affective valence disturbance (e.g., elevated negative affect, emotional lability; Courtney-Seidler et al., 2014; Crowell et al., 2009), which findings from our post-hoc conventional analyses are consistent with. However, by spanning multiple levels of analysis to identify qualitatively unique within-person profiles of coordinated psychobiological function, our person-centered approach strengthens inference about how immuno-endocrine and neurocognitive disruptions that have been previously independently linked to maltreatment (e.g., elevated and blunted cortisol levels; Cicchetti et al., 2010; Doom et al., 2013; elevated CRP and pro-inflammatory cytokine levels; Coelho et al., 2014; Ehrlich et al., 2021; compromised memory and executive functioning; Cowell et al., 2015; Valentino et al., 2009) may interact with one another within maltreated girls.

In so doing, our person-centered, multisystem approach illustrated profile-specific immuno-endocrine (e.g., Severe Affective Disturbance) and neurocognitive (e.g., Pro-inflammatory Affective Disturbance) strengths that may potentially offset the contributions of past maltreatment exposure to profile-specific weaknesses (e.g., affective difficulties common across dysregulated profiles). This finding is noteworthy, insofar as it would be reasonable to hypothesize from the summarized maltreatment research focusing on systems in isolation that maltreatment would be associated with profiles characterized by disturbance across all systems (e.g., Multisystem Disturbance). Our study, however, suggests that psychobiological dysregulation as it relates to maltreatment exposure is far more nuanced and complex. For example, while neurocognitive impairment and maltreatment history linkages are well-documented (Cowell et al., 2015; Gould et al., 2012), our concomitant consideration of affective valence in the Hypercortisol Affective Neurocognitive Disturbance profile pointed to positive affect as potential undocumented strength. These findings highlight the utility of our multisystem approach. As we discuss later, identification of these profiles has the potential to inform tailored care approaches to intervention that might better capitalize on and circumvent maltreated girls’ profile-specific strengths and weaknesses.

4.3. Links to Depressive Symptoms

Our identified profiles were also linked to depressogenic function in expected ways. Relative to Normative, girls exhibiting profiles reflecting psychobiological dysregulation were more likely to report current depressive symptoms, even after adjusting for past maltreatment exposure. When examined in isolation, depression has been independently linked high negative affect (Enns et al., 2003; Fergusson et al., 2000), low positive affect (Auerbach et al., 2015), executive function difficulties (Wilksinson & Goodyer, 2006), and dysregulated immuno-endocrine function (e.g., Landau et al., 2021; Suarez & Sundy, 2017). Our investigation is one the first to provide a more nuanced, comprehensive depiction of the varied ways these systems might interact in an etiological sense to confer risk for depression in adolescent girls.

Depressive symptom levels also varied across profiles by girls’ use of cognitive reappraisal, a finding that extends the psychobiological foundations of coping and emotion regulation literature (Compas et al., 2017). Both Normative and Pro-inflammatory Affective Disturbance girls displayed mild depressive symptom levels irrespective of their cognitive reappraisal utilization, perhaps indicative of their lesser degree of psychobiological impairment and, thus, less requisite use of this specific strategy. Cognitive reappraisal, however, was negatively related to depressive symptoms for girls with more advanced degrees of psychobiological dysregulation. Regarding the Severe Affective Disturbance profile, these girls’ high negative affect and low positive affect levels may constrain the efficacious use of certain coping skills (Compas, 2009). However, if their moderate cortisol:CRP ratios and normative attentional set shifting skills indicate the healthy suppression of inflammatory processes by the HPA, which together support adaptive functional changes in brain regions and circuitry (e.g., amygdala, prefrontal cortex) implicated in executive function and self-regulation (Shields et al., 2017), then it is possible that these girls’ immuno-endocrine and neurocognitive strengths support the use of cognitive reappraisal for managing depressive symptoms in a way that circumvents their affective predispositions. Similarly, Hypercortisol Affective Neurocognitive Disturbance girls may rely on normative positive affect levels to support their use of cognitive reappraisal. Indeed, positive emotionality promotes flexible thinking and adaptive coping with negative emotions (Folkman & Moscowitz, 2000; Fredrickson & Branigan, 2005). This strength might circumvent a possible neurocognitive proclivity towards perseveration (i.e., being “stuck” on stimuli from previously established dimension the ED stages of the IDED set-shifting module; Henry & Bettenay; 2010; Yerys & Munakata, 2006) that would otherwise interfere with reappraisal utilization. In each case, knowledge about girls’ psychobiological profiles may help clinicians leverage clients’ strengths in service of teaching this multi-faceted strategy (e.g., harnessing positive affect to identify the “silver lining,” harnessing executive function to brainstorm numerous reinterpretations, select the most optimal solution, and shift to a different solution in case the first fails) in ways that may circumvent these girls’ weaknesses.

4.4. Links to Passive or Active Suicidal Ideation (PASI)

Findings obtained when PASI was the outcome of interest complemented but also diverged from our depressive symptoms findings. Relative to Normative, girls exhibiting the Pro-inflammatory Affective Disturbance or Hypercortisol Affective Neurocognitive Disturbance profiles (but not Severe Affective Disturbance) were more likely to report PASI. This finding is consistent with recent conceptualizations of adolescent girls’ risk for suicidality as failure of the arousal regulatory, affective valence, and neurocognitive systems (Beauchaine et al., 2019; Crowell et al., 2009). However, our study extends the literature by empirically illustrating how coordinated disturbance across such systems may manifest within adolescent girls to confer such risk. Given that only Pro-inflammatory Affective Disturbance or Hypercortisol Affective Neurocognitive Disturbance girls were more likely to report PASI, it is possible that these profiles reflect more pathological states of psychobiological dysregulation, perhaps due to disruptions across multiple as opposed to a single (e.g., affect) system (e.g., Severe Affective Disturbance).

Contrary to expectation, cognitive reappraisal was not significantly associated with PASI and this association did not vary by girls’ psychobiological functioning. This finding runs counter to evidence suggesting that the tendency to use cognitive reappraisal in response to difficult emotions helps reduce risk for suicidal ideation (Forkman et al., 2014; Ong & Thompson, 2019), and that girls’ ability to do so may depend on their psychobiological foundations (Compas et al., 2017). It is possible that unexamined individual differences in exposure to recent stressful life events may have contributed to our nonsignificant findings (e.g., Boyes et al., 2016; Duprey et al., 2021; Franz et al., 2021). From a stress sensitization perspective (Hammen et al., 2000), girls with maltreatment histories and resulting dysregulated psychobiological function may be most prone to suicidal ideation when they are also unduly exposed to recent life stressors (Heim & Nemeroff, 2001). For Black youth, experiences with racial/ethnic discrimination may be one specific form of contextualized life stress that has been understudied in this respect. Black preadolescent youth are unduly exposed to racial discrimination (Argabright et al., 2021), with experience sampling studies suggesting that these youth encounter on average five experiences per day (English et al., 2020). If so, their maltreatment-related heightened sensitivity to ongoing racial discrimination may also have implications for how efficacious their use of coping and emotion regulation strategies might be in managing their depressogenic mood and thought patterns (Gruhn & Compas, 2020). Some studies have suggested that cognitive reappraisal and related strategies (e.g., positive thinking) may “backfire” for minoritized youth when attempting to manage depressotypic thoughts and feelings in the face of racial/ethnic discrimination, due to (a) limited positive reinterpretations about experienced racial transgressions and threats in offensive contexts, as well as (b) inadvertent internalization of the message that depressotypic thoughts about discrimination are the problem to be resolved rather than discrimination itself (Perez & Soto, 2011; Perzow et al., 2021; Wadsworth et al., 2020). Research is needed to understand whether girls with maltreatment-linked dysregulated psychobiological profiles ability to utilize such strategies for effectively managing PASI depends on minoritized stress exposure.

4.5. Limitations and Future Directions

The lack of evidence with respect to one of our focal anticipated outcomes is also noteworthy. Though our findings provided support for our anticipated Normative profile, we did not find any evidence of a Multisystem Disturbance profile. Future studies utilizing larger sample sizes and recruitment strategies that target more severe clinical presentations (e.g., inpatient) may be needed in order to identify this profile. Nevertheless, our identified profiles, those which provided close but incomplete approximations of Multisystem Disturbance, illustrated unique patterns of psychobiological strengths and weaknesses that pointed to person-specific sources of clinically informative resilience.

Our study also had several limitations that point to directions for future research. Our findings are limited to a racially diverse, economically disadvantaged sample of adolescent girls, some of whom were depressed and/or maltreated, who were not treatment-seeking. Future research is needed to determine whether our findings generalize to other genders, non-maltreated, and non-clinical populations of adolescents. Indeed, given that our sample was economically disadvantaged and poverty has been linked pro-inflammatory profiles (for review, see Jensen et al., 2017), our Pro-inflammatory Affective Disturbance profile may be unique to this demographic aspect. Though our sample size was similar to other notable person-centered studies of psychobiological function (Del Giudice et al., 2012; Koss et al., 2017; Turpyn et al., 2015), it was small for our approach. Though common in LPA, unequal subgroup sizes emerged and may have reduced power to detect effects in our maltreatment and depression analyses and precluded examination of moderation effects by race. Thus, cautious interpretation and replication in studies with larger sample sizes is warranted, moderation findings from which may strengthen inference about whether effects are most pronounced for Black, economically disadvantaged adolescent girls in particular. Though we controlled for girls’ waking saliva sample time and use of pain medications (e.g., Aspirin), morning immuno-endocrine levels can be impacted by variables which were unmeasured in our study (e.g., sleep quality, sunlight, chronic stress). Thus, replication in studies that account for these factors is warranted. Cortisol levels are increased when chewing gum compared to unstimulated passive drool, whereas salivary CRP levels are not. Although confounds related to collection method were constant across participants, this pattern of effects from gum chewing may have led to slightly higher cortisol levels. Studies using passive drool methods may find different cortisol:CRP ratio distributions among identified profiles. Our study relied on retrospective self-reports of child maltreatment, which can be potentially impacted by several biases (e.g., forgetting, subsequent life-events; Hardt & Rutter, 2004; Tajima et al., 2004). We also focused on the specific number of subtypes girls experienced over the course of their childhood and early adolescence. Child maltreatment, however, is multidimensional (Manly et al., 2001), and maltreatment to psychobiological dysregulation linkages in our study may have been better explained were differences in timing (i.e., developmental period during which maltreatment initially occurred) and chronicity (i.e., number of developmental periods during which maltreatment occurred) examined. Future studies focusing on these temporal aspects may find that profiles reflecting greater degrees of psychobiological dysregulation may not only be linked to greater number of subtypes experienced, but also earlier timing and more chronic exposure over the course of youths’ development. Our cross-sectional design precludes causal inference about linkages among child maltreatment, psychobiological functioning, and depressogenic outcomes (Kraemer et al., 1997). Future research incorporating longitudinal designs may be better situated to test whether identified profiles function as psychobiological mechanisms of risk involved in past maltreatment exposure (i.e., antecedent) to future depressogenic function (i.e., outcome) linkages.