Substantial research has shown that child maltreatment exerts dire effects on the developmental course. Child abuse and neglect progressively contribute to compromised adaptation on a number of developmental domains and issues central to successful adjustment. These developmental failures pose significant risk for psychopathology (Cicchetti & Toth, in press; Cicchetti & Valentino, Reference Cicchetti, Valentino, Cicchetti and Cohen2006).

The deleterious sequelae accompanying child maltreatment not only result in adverse consequences during childhood but also often initiate a negative developmental cascade that continues throughout the life course (Cicchetti & Tucker, Reference Cicchetti and Tucker1994; Masten & Cicchetti, Reference Masten and Cicchetti2010). The proximal environment involving the nuclear family, as well as more distal factors associated with the community and culture more broadly, transact to undermine normal biological and psychological developmental processes in children who have experienced maltreatment (Cicchetti & Lynch, Reference Cicchetti and Lynch1993, Reference Cicchetti, Lynch, Cicchetti and Cohen1995; Rogosch, Cicchetti, Shields, & Toth, Reference Rogosch, Cicchetti, Shields, Toth and Bornstein1995). Child maltreatment may constitute the greatest failure of the caregiving environment to provide opportunities for normal development (Cicchetti & Lynch, Reference Cicchetti, Lynch, Cicchetti and Cohen1995). Moreover, maltreatment experiences may provide serious challenges to the species-typical organism–environment “coactions” that play important roles in the emergences and timing of normal developmental change (Cicchetti & Lynch, Reference Cicchetti, Lynch, Cicchetti and Cohen1995; Gottlieb, Reference Gottlieb1992).

Findings from the Adverse Childhood Experiences (ACE) study (Anda, Dube, & Giles, Reference Anda, Dube and Giles2006), one of the largest and arguably most important public health studies conducted to date in the United States, demonstrate that childhood adversity is associated with an extraordinary number of adult mental and physical health risk behaviors. The ACE factors that were examined included exposure to childhood emotional, physical, and sexual abuse; substance abuse in the family; parental mental illness; domestic violence; and adult criminal behavior. Adults with elevated ACEs were at increased risk not only for psychosocial and substance abuse problems but also for diseases, including cardiovascular, diabetic, and immunological problems, and early morbidity.

It is striking that over two-thirds of the adults surveyed in the ACE study had experienced one or more types of adverse childhood experiences; 87% of these individuals had experienced two or more types of trauma. According to the Centers for Disease Control (Fang, Brown, Florence, & Mercy, Reference Fang, Brown, Florence and Mercy2012), 1 year of confirmed cases of child maltreatment (including cases of sexual, physical, and emotional abuse, and neglect) costs $124 billion over the life span of the abused and neglected children. The public health implications of the adverse effects accompanying child maltreatment are enormous. Thus, the magnitude of the effects of child maltreatment and the toll they exert on individuals, families, and society should not be underestimated.

Previous research on the sequelae of child maltreatment has focused predominately on psychosocial outcomes. Developmental psychopathologists have advocated research on normal and atypical populations incorporating multiple levels of analyses (Cicchetti & Dawson, Reference Cicchetti and Dawson2002; Cicchetti & Toth, Reference Cicchetti, Toth, Nolen-Hoeksema and Hilt2009). In recent decades, research on the effects of maltreatment increasingly has incorporated multilevel investigations of biological and psychosocial sequelae (Curtis & Cicchetti, Reference Curtis and Cicchetti2013; DeBellis, Reference DeBellis2001, Reference DeBellis2005; Hart & Rubia, Reference Hart and Rubia2012; McCrory, De Brito, & Viding, Reference McCrory, De Brito and Viding2010; Pollak, Cicchetti, Klorman, & Brumaghin, Reference Pollak, Cicchetti, Klorman and Brumaghim1997).

As toxic stressors (Shonkoff, Boyce, & McEwen, Reference Shonkoff, Boyce and McEwen2009), child abuse and neglect are implicated in the disruption of biological systems, including neuroendocrine and immune functioning, physical and mental health outcomes, and neurobiology. Physiological and behavioral responses to maltreatment are interrelated and contribute to an organization of developmental systems that may produce pathological development. Early maltreatment may alter young neural networks, resulting in cascading effects through the course of later development, possibly constraining the child's capacity to adapt flexibly to new challenging situations and environmental demands (Cicchetti & Tucker, Reference Cicchetti and Tucker1994).

The concepts of allostasis and allostatic load (AL) provide an integrative framework for understanding how exposure to chronic stress, such as child maltreatment, potentiates long-term liabilities for physical and mental health (Danese & McEwen, Reference Danese and McEwen2012; Juster et al., Reference Juster, Bizik, Picard, Arsenault-Lapierre, Sindi and Trepanier2011). Allostasis is a process that involves the activation of multiple interactive physiological systems (e.g., cardiovascular, neuroendocrine, immune, and metabolic systems). Epinephrine, norepinephrine, cortisol, dehydroepiandosterone (DHEA), and pro- and anti-inflammatory cytokines are considered the primary mediators of AL because they operate on the cellular level (McEwen, Reference McEwen2003; Rogosch, Dackis, & Cicchetti, Reference Rogosch, Dackis and Cicchetti2011).

In the short term, mobilization of these systems exerts a protective effect on the body and promotes an adaptive response to stress; however, with chronic activation, physiological reactions to stress become less efficient in protecting the individual. Ensuing damage to the body results in allostatic overload, which, in turn, contributes to change in brain structure and function and the development of various disease states (Hostinar, Sullivan, & Gunnar, Reference Hostinar, Sullivan and Gunnar2013). It is the collective impact of alterations across multiple systems that contributes most strongly to morbidity, rather than changes within any one system (Juster et al., Reference Juster, Bizik, Picard, Arsenault-Lapierre, Sindi and Trepanier2011; Juster, McEwen, & Lupien, Reference Juster, McEwen and Lupien2010; Rogosch et al., Reference Rogosch, Dackis and Cicchetti2011).

Cicchetti and Rogosch (Reference Cicchetti and Rogosch2001) examined the extent to which maltreated children vary with respect to cortisol regulation, a biomarker of AL. Although no differences in cortisol regulation were found between the maltreated and nonmaltreated groups, findings revealed significant within-group variation as a function of maltreatment subtype. In particular, children who had experienced both physical and sexual abuse, in combination with neglect and/or emotional maltreatment, exhibited substantial elevations in morning cortisol levels (i.e., hypercortisolism). In addition, a subgroup of physically abused children showed a trend toward lower morning cortisol relative to nonmaltreated children (i.e., hypocortisolism). Furthermore, the neglected and emotionally maltreated groups did not differ from nonmaltreated children in terms of cortisol regulation.

Relatedly, Cicchetti, Rogosch, and Oshri (Reference Cicchetti, Rogosch and Oshri2011), within an AL framework, investigated the effect of Gene–Environent (G × E) interactions on diurnal cortisol regulation and internalizing symptomatology. Variation in the corticotropin-releasing hormone receptor 1 (CRHR1) TAT haplotype and serotonin transporter linked polymorphic region gene (5-HTTLPR) was identified in a sample of maltreated and nonmaltreated children. G × E effects for CRHR1 and maltreatment and early abuse (before age 5) on diurnal cortisol regulation were observed; CRHR1 variation was related to cortisol dysregulation only among maltreated children. Early abuse and high internalizing symptoms also interacted to predict atypical diurnal cortisol regulation. The interaction of CRHR1, 5-HTTLPR, and child maltreatment (G × G × E) identified a subgroup of maltreated children with high internalizing symptoms who shared the same combination of the genetic variants (two copies of the CRHR1 TAT haplotype and the long–long 5-HTTLPR genotype). The findings provide support for an AL perspective on the effects of the toxic stress associated with child maltreatment on cortisol regulation and internalizing symptomatology as moderated by genetic variation.

There are different models of how the effects of excesses, deficits, and dysregulation in the primary mediators of AL, particularly cortisol, may affect brain structure and function. Further, developmental considerations are important. In this regard, the timing of periods of severe stress may be critical. For example, variation in the age of maturation of brain structures, notably the hippocampus, prefrontal cortex, and the amygdala, may result in differential effects of functioning and health (Lupien, McEwen, Gunnar, & Heim, Reference Lupien, McEwen, Gunnar and Heim2009; Shonkoff et al., Reference Shonkoff, Boyce and McEwen2009). Sensitive periods in the development of these brain structures may generate heightened vulnerability to the neurotoxic effects of excess glucocorticoids, thereby creating a long-term liability as development proceeds. Alternatively, the cumulative exposure to stressful experiences and concomitant dysregulation of the hypothalamic–pituitary–adrenal (HPA) axis may contribute to ongoing wear and tear on these brain structures. These effects of chronic stress are heightened, given that these areas of the brain have dense concentrations of glucocorticoid receptors, thereby promulgating progressive inefficiency in brain structure and function (Gunnar & Vasquez, Reference Gunnar, Vasquez, Cicchetti and Cohen2006). For children in low-income environments and those subjected to abuse and neglect, the consequences may be particularly salient for health outcomes across the life span.

Consistent with the concept of AL, Rogosch et al. (Reference Rogosch, Dackis and Cicchetti2011) conducted a multidomain assessment of stress-sensitive systems among low-income maltreated and nonmaltreated comparison children. An AL composite was created from measurements of salivary cortisol and DHEA, body mass index, waist–hip ratio, and blood pressure. The results indicated that maltreatments and AL independently predicted psychopathology and health difficulties (i.e., parent report of child's physical health status and utilization of health care systems). As AL increased, the level of the child health and psychological problems increased for all low-income children. Child maltreatment had an additive effect, contributing to the degree of physical and mental health problems beyond that accounted for by the AL composite. Therefore, children with both high AL and a history of maltreatment had the most health problems.

In addition to cortisol dysregulation, maltreatment has been found to be associated with other adverse health outcomes. For example, maltreatment predicts increased risk for hospital-based treatment of asthma, cardiorespiratory, and infectious diseases in childhood. (Lanier, Jonson-Reid, Stahlschmidt, Drake, & Constantino, Reference Lanier, Jonson-Reid, Stahlschmidt, Drake and Constantino2010). Early child abuse has been linked to more health-related symptoms such as sleep and eating difficulties, higher body mass index, high blood pressure, low high-density lipoprotein, leptin deficiency, obesity, fibromyalgia, and compromised immune system functioning in childhood and adolescence (Danese et al., Reference Danese, Caspi, Williams, Ambler, Sugden and Mika2011; Danese, Pariante, Caspi, Taylor, & Poulton, Reference Danese, Pariante, Caspi, Taylor and Poulton2007; Danese & Tan, Reference Danese and Tan2014; Lee, Reference Lee2010; Rogosch et al., Reference Rogosch, Dackis and Cicchetti2011; Shirtcliff, Coe, & Pollak, Reference Shirtcliff, Coe and Pollak2009). Furthermore, health liabilities extend into adulthood, because child maltreatment has been found to predict adult cardiovascular disease, elevated inflammation levels, type II diabetes, HIV risk, self-reported physical symptoms across a range of organ systems, and decreased longevity (Danese et al., Reference Danese, Pariante, Caspi, Taylor and Poulton2007; Felitti et al, Reference Felitti, Anda, Nordenberg, Williamson, Spitz and Edwards1998; Wilson & Widom, Reference Wilson and Widom2011). Moreover, child maltreatment has been linked to shortened leucocyte telomere length, whereby it is thought that maltreatment may influence cellular aging (Tyrka et al., Reference Tyrka, Price, Kao, Porton, Marsella and Carpenter2010).

Miller, Chen, and Parker (Reference Miller, Chen and Parker2011) proposed a conceptual model to explain why those who experience child maltreatment have an enhanced pro-inflammatory phenotype. Miller et al. (Reference Miller, Chen and Parker2011) state that stress that occurs during a sensitive developmental period, that is, when immune function is highly malleable/plastic, becomes embedded in the functioning of the cells that regulate inflammation. Coelho, Viola, Walss-Bass, Brietzke, and Grassi-Oliveira (Reference Coelho, Viola, Walss-Bass, Brietzke and Grassi-Oliveira2014) reviewed 20 studies that examined the association between adults' recollections of child maltreatment and inflammatory markers. Coelho et al. (Reference Coelho, Viola, Walss-Bass, Brietzke and Grassi-Oliveira2014) concluded that a history of child maltreatment was associated with increased levels of C-reactive protein (CRP), an inflammatory biomarker that is indicative of systemic inflammation. Reichlin (Reference Reichlin1993) proposed that the neuroendocrine and immune systems interact. Immune responses are thought to alter neural and endocrine function. In turn, neural and endocrine activity are seen as modifying the function of the immune system. The dysregulation of the HPA axis in maltreated persons, who manifest pituitary hyporesponsiveness to corticotropin-releasing hormone and/or increased glucocorticoid sensitivity to cortisol, has the potential to bring about a systemic proinflammatory state.

Danese et al. (Reference Danese, Moffitt, Pariante, Ambler, Poulton and Caspi2008) studied adults who were currently depressed and had a history of child maltreatment. They found that individuals with both depression and a maltreatment history were more likely to have high levels of high-sensitivity CRP compared to control subjects. In contrast, individuals who were currently depressed, but who had no history of child maltreatment, had a nonsignificant elevation of CRP. In another study, Danese et al. (Reference Danese, Moffitt, Harrington, Milne, Polanczyk and Pariante2009) investigated adults who were exposed to adverse psychosocial experiences during their childhood. They found that these adults were at an elevated risk of depression and had high inflammation levels and metabolic risk markers. These metabolic risks included high blood pressure, high total cholesterol, and low high-density lipoprotein cholesterol. Adults who had not been exposed to child maltreatment did not exhibit elevations in depression, inflammation, and metabolic risks.

Slopen, Kubzansky, McLaughlin, and Koenen (Reference Slopen, Kubzansky, McLaughlin and Koenen2013) conducted a prospective longitudinal study in which they examined acute adverse events at seven time points prior to age 8 and inflammation outcomes at ages 10 (CRP; interleukin-6) and 15 (CRP). Investigations have rarely examined the links between child adversity and inflammation prospectively. Inflammation may be an early mechanism by which adverse experiences in childhood become biologically embedded and adversely impact health throughout the lifespan (cf. Miller et al., Reference Miller, Chen and Parker2011). Exposure to adverse events before age 8 is associated with elevated levels of inflammation at age 10 and in midadolescence. Slopen et al. (Reference Slopen, Kubzansky, McLaughlin and Koenen2013) discovered that the timing of the exposure to adverse events appears to matter with respect to later development of inflammation. They found that adverse events experienced in early childhood and in relatively close proximity to the measure of inflammation were most strongly associated in inflammatory levels.

Bufalino, Hepgul, Aguglia, and Pariante (Reference Bufalino, Hepgul, Aguglia and Pariante2012) reviewed the clinical studies literature on the role of immune genes in the association between depression and inflammation. Dysregulation of the immune system in the pathogenesis of depressive disorders commonly occurred and research has begun to implicate the role of an underlying genetic vulnerability. Functional allelic variants of genes for CRP were among the genes that were found to augment the risk for depression. Genetic variants influence the biological mechanisms by which the immune system contributes to the development of depression (Hadler et al., Reference Halder, Marsland, Cheong, Muldoon, Ferrell and Manuck2010).

Given the paucity of studies conducted on the effects of child maltreatment on inflammation in children, and evidence accumulating that variation in the CRP gene is implicated in vulnerability to depression, in the present multilevel investigation we examine the association among timing of child maltreatment, variation in CRP genotype, neuroendocrine regulation, salivary CRP, and levels of internalizing symptoms.

Method

Participants

Participants included 489 children aged 8 to 12 (Mage = 9.72, SD = 0.99; 52.4% male) who attended a summer day camp research program designed for school-aged low-income children. The sample included both maltreated children (n = 267) and nonmaltreated children (n = 222) and was racially and ethnically diverse (66.0% African American, 10.0% Caucasian, 19.6% Hispanic, and 4.5% another race/ethnicity). Informed consent was obtained from parents of maltreated and nonmaltreated children for their child's participation in the summer camp program and for examination of any Department of Human Services (DHS) records pertaining to the family.

Children in the maltreated group were recruited through a DHS liaison, who examined Child Protective Services reports to identify children who had been maltreated. Children living in foster care were not recruited for the current investigation. The DHS liaison contacted eligible families and explained the study. Parents who were interested in having their child participate provided signed permission for the contact information to be shared with project staff. These families were representative of those receiving services through DHS. Comprehensive reviews of all DHS records for each family were conducted. Coding is based on all available information and does not rely on DHS determinations. Maltreatment information was coded by trained research staff, doctoral students, and clinical psychologists, using the Maltreatment Classification System (MCS; Barnett, Manly, and Cicchetti, Reference Barnett, Manly, Cicchetti, Cicchetti and Toth1993), a nosological system for classifying child maltreatment. Adequate reliability has been obtained (weighted κ = 0.86–0.98; Manly, Reference Manly2005; Manly, Kim, Rogosch, & Cicchetti, Reference Manly, Kim, Rogosch and Cicchetti2001).

Because maltreating families primarily have low socioeconomic status (National Incidence Study; Sedlak et al., Reference Sedlak, Mettenburg, Basena, Petta, McPherson and Greene2010), nonmaltreating families were recruited from those receiving Temporary Assistance to Needy Families in order to ensure socioeconomic comparability between maltreated and nonmaltreated families. A DHS liaison contacted eligible nonmaltreating families and described the project. Parents who were interested in participating signed a release allowing their contact information to be given to project staff for recruitment. Families who received preventative DHS services due to concerns over risk for maltreatment were not included with the nonmaltreated comparison group. In order to further verify a lack of DHS involvement, trained research assistants interviewed the mothers of children recruited for the nonmaltreated group using the Maternal Child Maltreatment Interview (Cicchetti, Toth, & Manly, Reference Cicchetti, Toth and Manly2003) and reviewed DHS records in the year following camp participation to assure that all information had been assessed.

As shown in Table 1, children in the maltreated and nonmaltreated groups were comparable on a number of demographic variables, including child age, maternal education, maternal marital status, and history of receiving public assistance. Groups differed on gender such that more males comprised the maltreated group, χ² (1) = 5.78, p = .02. Groups also differed on ancestry informed race/ethnicity such that there were more African American children in the nonmaltreated group, χ² (3) = 16.77, p = .001.

Table 1. Demographic and descriptive characteristics of sample

*p < .05. **p < .001.

Procedure

Families were approached and asked if they would agree to have their child attend a weeklong day camp and participate in the research (for detailed descriptions of the camp procedures, see Cicchetti & Manly, Reference Cicchetti, Manly, Brody and Sigel1990). Children were transported by bus to the camp each day, with travel time averaging 45 min. At the camp, children were assigned to groups of 10 (5 maltreated, 5 nonmaltreated) same-age and same-sex peers. Each group was led by three trained camp counselors, who were unaware of the maltreatment status of children and the hypotheses of the study. Camp lasted 7 hr per day for 5 days, providing 35 hr of interaction between children and counselors. In addition to the recreational activities, after providing assent, children participated in various research assessments and provided morning and afternoon saliva samples. Trained research assistants, who also were unaware of research hypotheses and maltreatment status, conducted individual research sessions with children, in which questionnaires and other research measures were administered. If any concerns over danger to self or others emerged during research sessions, then clinical psychologists were available for consultation and referrals were made as needed. The counselors, who had been trained extensively for 2 weeks prior to the camp, also completed assessment measures on individual children, based on their 35 hr of observations and interactions with the children in their respective groups.

Measures

MCS

The MCS (Barnett et al., Reference Barnett, Manly, Cicchetti, Cicchetti and Toth1993) was used to delineate diverse features of child maltreatment that individual children experienced. The MCS utilized DHS records detailing investigations and findings regarding maltreatment occurrences in identified families. Rather than relying on official designations and case dispositions, the MCS codes all information available on a designated family from DHS records, making independent determinations of maltreatment experiences. In particular, the MCS codes all incidents that have been documented in DHS records and, based on operational criteria, designates the subtypes of maltreatment individual children have experienced (i.e., emotional maltreatment, neglect, physical abuse, and sexual abuse), the severity of each type of maltreatment, the frequency of occurrence, the developmental periods in which maltreatment occurred, and the perpetrators of maltreatment.

Children were classified into four groups based on the timing of their maltreatment experiences. Children who experienced maltreatment before age 5, but not since then were classified as “early, not recently maltreated.” Children who experienced maltreatment both before age 5 and since then were classified as “early and recently maltreated.” Children whose first maltreatment experience occurred between age 5 and the time of study, were classified as “recent onset maltreated.” Children without maltreatment histories were classified as “nonmaltreated.”

DNA collection, extraction, and genotyping

Using an established protocol, trained research assistants obtained DNA samples from participants by collecting saliva using the Oragene DNA Self-Collection kits. DNA was purified from 0.5 ml of Oragene-DNA solution using the DNAgenotek protocol for manual sample purification using prepIT-L2P. Sample concentrations were determined using the Quant-iT PicoGreen dsDNA Assay Kit (P7589, Invitrogen).

Variations in the gene encoding for CRP were determined using assays for single nucleotide polymorphism (SNP) rs1417938 purchased from Applied Biosystems, Inc. (ABI; Bedford, MA) as C7479322 10. Individual allele discriminations were made using Taq Man Genotyping Master Mix (ABI Catalog No. 4371357) with amplification in an ABI 9700 thermal cycler and analyzing the endpoint fluorescence using a Tecan M200. If a genotype could not be determined after the first run, then it was repeated up to four times. The call rate for the CRP SNP was 99.2%. CRP SNP distribution did not deviate from Hardy–Weinberg equilibrium, χ² (1) = 0.89, ns. The frequency distribution of the CRP SNP was as follows: 66.9% TT, 25.4% AT, and 3.1% AA. Genotypes AA and AT were combined in these analyses because of the low frequency of AA.

Ancestry

For ancestral proportion testing, DNA from study participants was subjected to SNP genotyping of the Burchard et al. panel of 106 SNPs (Lai et al., Reference Lai, Tucker, Choudhry, Parnell, Mattei and Garcia-Bailo2009; Yaeger et al., Reference Yaeger, Alvial-Bront, Abdul, Nolan, Grann and Birchette2008), known to be informative for ancestry from Africa, Europe, and Native America. The SNPs were genotyped using the iPLEX platform from Sequenom Bioscience, Inc., which uses the Sequenom MassArray. Samples are subjected to single base primer extension (SBE) with fluorophore-labeled nucleotides from primers designed for SNPs of interest. The samples including the SBE products were placed on the iPLEX platform and matrix assisted laser desorption ionization time of flight was used to identify the allele based on the fluorophore passing the detector at the expected time associated with the mass of the SBE primer. The SNP genotyping results were then recoded and uploaded into STRUCTURE v2.3.4, which uses algorithms developed by Pritchard and colleagues (Falush Stephens, & Pritchard, Reference Falush, Stephens and Pritchard2003, Reference Falush, Stephens and Pritchard2007; Hubisz, Falush, Stephens, & Pritchard, Reference Hubisz, Falush, Stephens and Pritchard2009). Three SNP tests were excluded based on high allele call rates of the non-DNA containing wells. The data from remaining 103 loci were uploaded into the software and set to analyze with an Admixture model of ancestry and initialization of the simulation on the GALA cohort (initialize of POPINFO). The simulation was set to run with a burn-in of 10,000, 1,000 MCMC repetitions, and assuming three populations within the group. The results of the simulations were subsequently identified as percent association to each ancestry group based on the known ancestry of the GALA cohort.

To facilitate Gene × Race/Ethnicity and Maltreatment × Race/Ethnicity interaction tests, a grouping variable using ancestral proportion continuous scores was created using multinomial logistic regression to classify cases. This was done because an ancestry-informed race/ethnicity grouping variable is more readily interpretable in the case of any potentially statistically significant interactions. Parent-reported race/ethnicity (coded 1 = African American, 2 = Caucasian, 3 = Hispanic, 4 = other race/ethnicity) was predicted from proportion African ancestry and proportion Native American ancestry. We did not include proportion European ancestry because of collinearity concerns with the other ancestral proportion scores. The parent-identified Caucasian group was selected as the reference group. The overall multinomial logistic regression was significant, LR χ² (6) = 606.60, p < .001, and together the two ancestry proportion scores explained a very large amount of variance in parent-identified group membership (Nagelkerke R ² = .81). The results indicated that higher African ancestry scores predicted a significantly greater likelihood of being in the parent-identified African American versus Caucasian group (B = 0.48, odds ratio [OR] = 1.62, 95% confidence interval [CI] = 1.21, 2.17), Hispanic versus Caucasian group (B = 0.38, OR = 1.46, 95% CI = 1.09; 1.95), and other race/ethnicity group versus Caucasian group (B = 0.39, OR = 1.48, 95% CI = 1.10, 1.97). Higher Native American ancestry scores significantly predicted a greater likelihood of being in the parent-identified Hispanic versus Caucasian group (B = 0.06, OR = 1.06, 95% CI = 1.09) and other race/ethnicity group versus Caucasian group (B = 0.04, OR = 1.04, 95% CI = 1.06). Native American ancestry did not significantly differentiate between parent-identified African American and Caucasian groups. The mean probability of predicted group membership from the two ancestry proportion scores was 0.84 (SD = 0.18) with probabilities ranging from 0.28 (SD = 0.28 among those parent-identified as other) to 0.90 (SD = 0.16 among those parent-identified as African American). Ancestral proportion scores were more likely to reclassify individuals who parent-identified as Hispanic or other race/ethnicity compared to those who parent-identified as African American or Caucasian.

Cortisol and DHEA

The camp context allowed for a setting where a consistent collection of saliva samples could occur at uniform times across the camp week. Each child arrived at camp via camp bus transportation. Each day upon arrival to the camp at 9:00 a.m., trained research assistants obtained saliva samples from each child. Given the approximate 45-min camp bus transportation time and time spent being greeted by camp staff, children had been awake at least 1 hr prior to providing the morning saliva samples, thus avoiding the period of the dynamic cortisol awakening response (Susman et al., Reference Susman, Dockray, Schiefelbein, Herwehe, Heaton and Dorn2007). Children had not consumed food or drink for at least 30 min before each saliva sample was obtained. Samples were collected following the method recommended by Schwartz, Granger, Susman, Gunnar, and Laird (Reference Schwartz, Granger, Susman, Gunnar and Laird1998). The children were asked to chew Trident^® sugarless original flavor gum to stimulate saliva flow and then passively drool through a short drinking straw into a 20-ml plastic vial. The samples were immediately frozen and stored at –40 °C. Each week, the samples were shipped overnight on dry ice for next-day delivery to Salimetrics Laboratories (State College, PA) for assay. After thawing, each sample was processed by placing four to five 1-ml aliquots into 1.8-ml cryogenic storage vials and freezing at –80 °C. Upon assay, samples were thawed to room temperature and centrifuged at 3000 rpm for 15 min. The clear top plastic of the sample was pipetted into appropriate test tubes/wells.

Salivary cortisol (micrograms/deciliter) was assayed using an enzyme immunoassay kit (Salimetrics, State College, PA). This kit is commercially available and uses 25 µl of saliva. Its lower limit of sensitivity is 0.007 µg/dl (range up to 1.8 µg/dl) with average intra- and interassay coefficient of variation of <5.0% and 10.0%, respectively. Cortisol was assayed from saliva for each day across the week that it was collected.

Salivary DHEA (pg/ml) was also processed using an enzyme immunoassay kit (Salimetrics, State College, PA). This kit uses 550 µl of saliva. Its lower limit of sensitivity is 10.0 pg/ml (range up to 1000 pg/ml) with average intra- and interassay coefficient of variation of <5.0% and 15.0%, respectively. DHEA was assayed from saliva for 2 days (Tuesday and Thursday) because of less variability in DHEA levels.

To calculate the cortisol/DHEA ratio, cortisol values in micrograms/deciliter and DHEA values in pictograms/milliliter were transformed into nanomoles/liter values. A cortisol/DHEA ratio was then calculated for Tuesday morning and Thursday morning. These ratio values evidenced high skew and kurtosis and were log-transformed. The two log-transformed values were averaged to yield the cortisol/DHEA ratio value for analyses. See Table 2 for descriptive statistics across maltreatment groups.

Table 2. Descriptive statistics of study variables by maltreatment timing groups

*p < .05.

Salivary CRP

Saliva samples were collected via passive drool once per day for 3 days in the afternoon. All samples were assayed for CRP using a commercially available immunoassay without modification to the manufacturer's recommended protocol. The test volume was 15 µl, with a range of standards from 93.75 to 3000 pg/ml. Samples were thawed to room temperature, centrifuged at 3000 rpm for 15 min to remove mucins, and diluted 1:10 prior to assay. Intra- and interassay coefficients of variation were less than 10% and 15%, respectively. Log transformations were conducted on each value to address skew and kurtosis and were then averaged across days resulting in a mean CRP value. Previous research has demonstrated a moderate to strong association between salivary CRP correlates with plasma CRP (Ouellet-Morin, Danese, Williams, & Arseneault, Reference Ouellet-Morin, Danese, Williams and Arseneault2011). See Table 2 for descriptive statistics across maltreatment groups.

Internalizing symptoms

The Child Behavior Checklist Teacher Report Form (TRF; Achenbach, Reference Achenbach1991) was used to assess children's internalizing symptomatology. After observing and interacting with children in their respective groups over the course of the camp week, camp counselors completed the TRF on individual children in their group. The camp counselors were unaware of the maltreatment status of the children in their group, as well as the research hypotheses. The TRF is a widely used and validated instrument to assess symptomatology by teachers. In the present study, because counselors are able to observe similar behavior to that of teachers, the camp counselors' ratings were used to provide an analogous assessment of child functioning by an adult external to the family. The TRF contains 118 total items rated for frequency that assess two broadband dimensions of child psychopathology: internalizing behavior problems and externalizing behavior problems. In this study, we employed the internalizing behavior problems broadband scale (composed of withdrawal, somatic complaints, and anxiety/depression subscales). Average interrater reliability (κ) among pairs of counselors' was 0.67. The counselors' scores for each child were averaged to obtain individual children's raw and T scores for the internalizing dimension. The mean internalizing T score was 48.29 (SD = 7.63). See Table 2 for descriptive statistics for each maltreatment group.

Results

The data analytic strategy for this investigation involved three sets of analyses. The first set utilized analyses of covariance (ANCOVAs) to investigate the G × E interactive effects of maltreatment and CRP genetic variation on salivary CRP levels. The second set of analyses employed ordinary least squares regression to examine whether the relation between salivary CRP levels and children's internalizing symptoms varied by maltreatment experiences. The final set of analyses was multiple group path models used to explore associations between neuroendocrine functioning (as measured by cortisol/DHEA ratio), inflammatory processes (as measured by salivary CRP), and internalizing symptoms and to investigate whether these relations were moderated by maltreatment timing.

Preliminary analyses are presented in Table 2. No differences in CRP SNP rs1417938 allele frequencies (TT vs. AT/AA) were found between the four maltreatment timing groups (nonmaltreated, early and not recent maltreatment, early and recent maltreatment, and recent onset maltreatment). Thus, we did not find evidence for a gene–environment correlation for CRP and maltreatment timing. Moreover, the four maltreatment timing groups did not differ on cortisol/DHEA ratio and internalizing symptoms. The results indicated a statistically significant difference between maltreatment timing groups on salivary CRP levels, F (3, 476) = 3.34, p = .03; however, none of the Bonferroni contrasts were statistically significant.

Interactive effects of maltreatment and CRP gene on salivary CRP levels

The unique and interactive effects of child maltreatment and CRP genetic variation were examined using a 2 (CRP TT vs. AT/AA) × 2 (maltreated vs. nonmaltreated) ANCOVA with age, gender, and ancestry-informed race/ethnicity included as covariates. All Covariate × CRP Gene and Covariate × Maltreatment interactions were included in this model as recommended by Keller (Reference Keller2014). There were 21 children who were identified as having a race/ethnicity other than African American, Caucasian, or Hispanic. Small cell sizes precluded the investigation of Race/Ethnicity × Gene and Race/Ethnicity × Maltreatment interactions when including these children. Therefore, the 21 children were excluded from G × E analyses. The results indicated that younger children evidenced higher salivary CRP levels, F (1, 423) = 12.24, p = .02. Main effects of race/ethnicity, F (1, 423) = 0.02, gender, F (1, 423) = 0.01, maltreatment, F (1, 423) = 0.35, and CRP genotype, F (1, 423) = 0.10, were all nonsignificant. Moreover, none of the Covariate × CRP Gene or Covariate × Maltreatment interactions was statistically significant. However, the interaction of maltreatment and CRP genotype was significant, F (1, 423) = 5.26, p = .02, indicating that the association between maltreatment and salivary CRP levels depends on CRP genetic variation.

Because main effects of race/ethnicity and gender as well as interactions of Race/Ethnicity × Maltreatment, Race/Ethnicity × CRP Gene, Gender × Maltreatment, Gender × CRP Gene, Age × Maltreatment, and Age × CRP Gene were all nonsignificant, a second model was tested that trimmed these effects from the model. This allowed for the inclusion of the 21 children in the other race/ethnicity group. The results of this ANCOVA with age included as a covariate were similar to those presented above; however, the interaction of maltreatment and CRP genotype was marginally significant, F (1, 454) = 3.01, p = .08.

To clarify the extent to which timing of maltreatment may be influential in these associations, a 2 (CRP TT vs. AT/AA) × 4 (1, nonmaltreated; 2, early, not recent maltreatment; 3, early and recent maltreatment; 4, recent onset maltreatment) ANCOVA was tested again including age, gender, and ancestry-informed race/ethnicity included as covariates. As above, all Covariate × CRP Gene and Covariate × Maltreatment interactions were included in this model. Younger children evidenced higher salivary CRP levels, F (1, 411) = 12.13, p = .001. Consistent with the above model, the main effects of race/ethnicity, F (1, 411) = 0.42, gender, F (1, 411) = 0.34, maltreatment timing, F (1, 411) = 0.34, and CRP genotype, F (1, 411) = 1.27, were all nonsignificant. The interaction of maltreatment timing and CRP genotype was significant, F (3, 411) = 4.90, p = .002, indicating that the relation between maltreatment timing and salivary CRP levels depends on CRP genetic variation. See Figure 1 for a graphical representation of the results. As above, none of the Covariate × Gene and Covariate × Maltreatment Timing interactions was statistically significant.

Figure 1. C-reactive protein single nucleotide polymorphism rs1417938 moderates the association between maltreatment timing and salivary C-reactive protein.

Because main effects of race/ethnicity and gender as well as interactions of Race/Ethnicity × Maltreatment, Race/Ethnicity × CRP Gene, Gender × Maltreatment, Gender × CRP Gene, Age × Maltreatment, and Age × CRP Gene were all nonsignificant, an additional model was tested that trimmed these effects from the model. This allowed for the inclusion of the 21 children in the other race/ethnicity group. The results of this ANCOVA with age included as a covariate were similar to those presented above; however, the main effect of maltreatment timing was significant, F (1, 450) = 3.42, p = .02, which is consistent with the results presented in Table 2. The interaction of maltreatment timing and CRP gene remained significant in this model, F (3, 450) = 3.48, p = .02.

Follow-up analyses revealed that among children with recent onset maltreatment, those with at least one A allele (AT/AA group) evidenced significantly higher salivary CRP levels than did those children in the TT group, t (62) = –2.56, p = .01. There were no significant differences between genotypes for children in any of the other maltreatment timing groups (i.e., nonmaltreated, early, not recent maltreated, and early and recent maltreated).

Salivary CRP and internalizing symptoms: Moderation by maltreatment

We also sought to examine the association between salivary CRP and child internalizing symptoms and to determine the role of maltreatment in this relation. We employed ordinary least squares regression with age, maltreatment (binary variable: maltreated vs. nonmaltreated), salivary CRP (mean centered), and the interaction of salivary CRP and maltreatment included as predictors of children's internalizing T scores. Gender and race/ethnicity were initially included as covariates but were not significantly related to internalizing scores and were therefore trimmed from the final model. The results indicated a marginally significant association between age and internalizing scores such that younger children evidenced marginally significantly higher internalizing T scores (B = –0.64, SE = 0.36, p = .08). None of the other main effects were significant. The interaction of maltreatment status and salivary CRP was marginally significant, B = 1.95, SE = 1.15, p = .09).

Because of the potential importance of maltreatment timing in these relations, an additional regression was conducted with the four-group maltreatment timing variable: (nonmaltreated; early, not recent maltreated; early and recent maltreated; and recent onset maltreated). Three dummy codes were created for the four groups such that the recent onset maltreated children were the reference group. Age, the three dummy coded maltreatment timing variables, salivary CRP (mean centered variable), and the three interaction terms (dummy code by salivary CRP) were included in this model. None of the dummy codes were significantly related to internalizing and consistent with the above model, and younger children demonstrated marginally significantly higher internalizing T scores (B = –0.65, SE = 0.36, p = .07). The interaction of the dummy code comparing recent onset maltreated to nonmaltreated children by salivary CRP was significant (B = –5.23, SE = 1.62, p = .001) as was the interaction of the dummy code comparing recent onset maltreated to early onset, not recently maltreated children by salivary CRP (B = –5.69, SE = 1.88, p = .003). The interaction of dummy code comparing recent only to early and recently maltreated children by salivary CRP was nonsignificant.

To probe these interactions, three new sets of three dummy codes were created such that each maltreatment timing group was coded as the reference group in one set. Three additional regression models were tested in which a different maltreatment timing group was the reference group in each model and new interaction terms were created with the new dummy codes in each model. The simple slope for recent onset maltreated children was statistically significant (Y = 3.86X + 54.45; p = .006). However, simple slopes for the three other maltreatment timing groups were nonsignificant (early and recent maltreated: Y = 0.40X + 55.08; early, not recent maltreated: Y = –1.82X + 55.00; nonmaltreated: Y = –1.37X + 54.04). Thus, results indicate that higher levels of salivary CRP are associated with higher levels of internalizing symptoms for recently maltreated children only. See Figure 2 for a graphical representation.

Figure 2. Maltreatment timing moderates the relation between salivary C-reactive protein levels and internalizing symptomatology.

We also investigated the effect of child internalizing symptoms on salivary CRP levels (reverse direction of effect) and moderation by maltreatment timing. The results indicated that higher child internalizing symptoms were associated with higher levels of salivary CRP only for recently maltreated children. We chose to present the model in which we explore the effect of salivary CRP on internalizing symptoms given prior literature suggesting that inflammation may temporally precede depressive symptoms (Gimeno et al., Reference Gimeno, Kivimaki, Brunner, Elovainio, De Vogli and Steptoe2009).

Relations among neuroendocrine functioning, immune functioning, and internalizing symptoms

To examine the relations among neuroendocrine functioning and immune functioning and internalizing symptoms among maltreated and nonmaltreated children, multiple group path analysis was conducted in Mplus Version 7.0 (Muthén & Muthén, Reference Muthén and Muthén1998–2012). Missing data were handled using full information maximum likelihood. Model fit was estimated with the chi-square statistic, comparative fit index (CFI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR). Child age, gender, race/ethnicity, and cortisol/DHEA ratio were entered as exogenous variables, salivary CRP was entered as a mediator, and internalizing T score was the endogenous variable. Because child gender and race/ethnicity did not uniquely predict salivary CRP or internalizing symptoms, they were trimmed from the final model.

To test for moderation of these associations by timing of maltreatment, a model that fully constrained all paths to be equal across the four maltreatment groups was tested. This model evidenced inadequate fit to the data, χ² (19) = 27.73, p = .09, CFI = 0.80, RMSEA = 0.06, SRMR = 0.09. Next, a model was tested that relaxed the following constraints across the four groups (i.e., allowed them to vary across the four groups): relation between cortisol/DHEA ratio and salivary CRP; relation between cortisol/DHEA ratio and internalizing symptoms; and relation between salivary CRP and internalizing symptoms. This model evidenced good model fit, χ² (10) = 9.10, p = .53, CFI = 1.00, RMSEA = 0.001, SRMR = 0.06, which was significantly better than the fully constrained model: Δχ² (9) = 18.63, p < .05. This suggests significant moderation of the associations by maltreatment timing. The results of the partially unconstrained model are presented in Figure 3. Consistent with early models described above, younger children evidenced higher salivary CRP levels (b = –0.23, p < .001). Among nonmaltreated children and early, not recent, maltreated children, lower cortisol/DHEA ratios were associated with higher salivary CRP levels (b _non = –0.20, b _early = –0.25; ps < .01); however, among early and recent maltreated children and recent only maltreated children, cortisol/DHEA ratios were unrelated to salivary CRP levels. Consistent with earlier models, higher salivary CRP levels were significantly associated with higher internalizing symptoms among recent onset maltreated children only (b = 0.39, p < .001). Among nonmaltreated children, early, not recent maltreated children, and early and recent maltreated children, salivary CRP was unrelated to internalizing symptoms. Finally, higher cortisol/DHEA ratios were marginally associated with higher internalizing symptoms among recent maltreated children only (b = 0.22, p = .06). The results were not indicative of a mediational process for any of the subgroups of children. Instead, results suggest that associations among neuroendocrine functioning, immune functioning, and internalizing symptoms depend on the occurrence and developmental timing of maltreatment.

Figure 3. Multiple-group model of relations among neuroendocrine functioning, immune functioning, and internalizing symptomatology. Standardized estimates are shown. N, Nonmaltreated; E, early, not recent maltreatment; ER, early and recent maltreatment; R, recent maltreatment. †p = .06, *p < .05, **p < .01, ***p < .001.