Measurement of early life adversity

How adversity is assessed, whether through objective or self-report measures, prospectively or retrospectively, has long been a consideration in the study of child adversity. A recent study highlights this reporting discrepancy, as it found poor agreement between prospective and retrospective reports of childhood maltreatment within the same individuals (Baldwin, Reuben, Newbury, & Danese, Reference Baldwin, Reuben, Newbury and Danese2019). The literature provides no clarity about how the reporting of ELA affects inflammatory outcomes. Part of this lack of clarity arises from a lack of formal delineation between prospective measures and retrospective measures of ELA. Few studies on inflammation arising from ELA have prospective and retrospective measures to compare directly. One study that makes the comparison is the Dunedin Multidisciplinary Health and Development Study, which showed that inflammation at age 32 y was more strongly associated with prospective compared to retrospective measures of ELA (Reuben et al., Reference Reuben, Moffitt, Caspi, Belsky, Harrington, Schroeder and Danese2016). However, data from the 1958 British Cohort (N = 7,661) that included prospective and retrospective measures of victimization found more mixed results, with positive associations between ELA and inflammation dependent on the type of ELA exposure (Pinto Pereira et al., Reference Pinto Pereira, Stein Merkin, Seeman and Power2019).

Another related area that requires clarification is how the source of information on ELA influences outcomes. Retrospective measures are, by necessity, self-reports, while prospective measures can include objective information (e.g., court reports), parent reports, or self-reports on ELA. The few studies that do use objective measures of ELA found inconsistent evidence for a link between ELA and inflammation. Osborn and Widom (Reference Osborn and Widom2019) have recently shown that individuals with official reports of child maltreatment (specifically physical abuse) had significantly higher levels of CRP than nonmaltreated individuals, yet two studies using Child Protective Services (CPS) referrals found no association between CRP and maltreatment (Bernard et al., Reference Bernard, Hostinar and Dozier2019; Cicchetti et al., Reference Cicchetti, Handley and Rogosch2015). Other studies of adolescents and adults exposed to objectively and prospectively assessed ELA in the form of institutional care have found no association between neglect exposure and inflammatory biomarkers (Elwenspoek et al., Reference Elwenspoek, Hengesch, Leenen, Schritz, Sias, Schaan and Muller2017; Reid et al., Reference Reid, Coe, Doyle, Sheerar, Slukvina, Donzella and Gunnar2019; Slopen et al., Reference Slopen, Tang, Nelson, Zeanah, McDade, McLaughlin and Fox2019).

Perhaps the ultimate challenge across measures is that the effect sizes between ELA and inflammation are small, and without adequately powered studies the outcomes are predictably inconsistent. A meta-analysis of 25 studies on ELA and plasma levels of inflammation in adulthood found small but significant effect sizes, though most studies included ELA measured broadly and retrospectively in adulthood (Baumeister et al., Reference Baumeister, Akhtar, Ciufolini, Pariante and Mondelli2016). Prospective studies of ELA in large cohorts have also found positive associations between ELA and inflammation, though the significant inflammation biomarker outcome differs between them (Baldwin et al., Reference Baldwin, Arseneault, Caspi, Fisher, Moffitt, Odgers and Danese2018; Bucker et al., Reference Bucker, Fries, Kapczinski, Post, Yatham, Vianna and Kauer-Sant'Anna2015; Flouri, Francesconi, Midouhas, & Lewis, Reference Flouri, Francesconi, Midouhas and Lewis2020). Adequately powered studies in large, longitudinal cohorts with both prospective and retrospective measures of ELA and a consistent set of inflammation biomarkers could help clarify the issue of ELA measurement as it relates to inflammation risk.

Dimensional approaches to the measurement of early life stress type

A second question is whether inflammation outcomes differ depending on the dimension or type of adversity exposure. Numerous studies have found different outcomes when examining separate types of childhood adversity. In meta-analysis of ELA exposure and inflammation in pediatric populations, the types of ELA exposure were too different to compare dimensionally (Kuhlman et al., Reference Kuhlman, Horn, Chiang and Bower2020). Future research would benefit from outlining diversity in the type of adversity exposure to examine the development of a proinflammatory phenotype and risk for psychopathology. McLaughlin et al. (Reference McLaughlin, Sheridan and Lambert2014) propose a dimensional approach to disentangle the effects of ELA on neurodevelopment, and we propose that this dimensional approach could be applied to the study of inflammatory outcomes. The approach distinguishes “threat,” or experiences that threaten an individual's physical integrity, and “deprivation,” which encompasses the lack of species-expected environmental inputs and/or environmental complexity (McLaughlin et al., Reference McLaughlin, Sheridan and Lambert2014). While deprivation and threat often co-occur in contexts where children are victimized, measuring both may clarify differential inflammation and psychopathology outcomes. In this section, we review a few select studies to highlight the measurement questions that remain in this area.

Experiences of threat include those that involve physical or sexual violence, threat of death, or other physical harm (McLaughlin et al., Reference McLaughlin, Sheridan and Lambert2014). In adult populations, there are a few examples of examining “threat” types of childhood adversity as they relate to inflammatory outcomes. In subgroup analyses in a meta-analysis, the type of ELA exhibited contrasting results in specific inflammatory outcomes (Baumeister et al., Reference Baumeister, Akhtar, Ciufolini, Pariante and Mondelli2016). Childhood sexual abuse was positively associated with TNF-α but not with IL-6 or CRP. Childhood physical abuse was associated with TNF-α and IL-6 but not CRP (Baumeister et al., Reference Baumeister, Akhtar, Ciufolini, Pariante and Mondelli2016). An analysis of two large adult cohorts mentioned previously examined different types of childhood adversity on adult levels of inflammation (Pinto Pereira et al., Reference Pinto Pereira, Stein Merkin, Seeman and Power2019). In the British cohort, increased inflammation in adulthood was positively associated with physical abuse but not sexual abuse (Pinto Pereira et al., Reference Pinto Pereira, Stein Merkin, Seeman and Power2019). In the US cohort, physical abuse was not associated with either IL-6 or CRP in adulthood (Pinto Pereira et al., Reference Pinto Pereira, Stein Merkin, Seeman and Power2019). However, childhood sexual abuse was significantly associated with IL-6 and CRP for nonwhites only (Pinto Pereira et al., Reference Pinto Pereira, Stein Merkin, Seeman and Power2019). In studies of pediatric populations, threat experiences of maltreatment and victimization are positively associated with inflammation in some studies (Baldwin et al., Reference Baldwin, Arseneault, Caspi, Fisher, Moffitt, Odgers and Danese2018; Bucker et al., Reference Bucker, Fries, Kapczinski, Post, Yatham, Vianna and Kauer-Sant'Anna2015; Danese et al., Reference Danese, Pariante, Caspi, Taylor and Poulton2007; Flouri et al., Reference Flouri, Francesconi, Papachristou, Midouhas and Lewis2019; Kilic et al., Reference Kilic, Ozlem and Murat2017) but not all (Bernard et al., Reference Bernard, Hostinar and Dozier2019; Cicchetti et al., Reference Cicchetti, Handley and Rogosch2015; Miller & Cole, Reference Miller and Cole2012).

One component of the dimensional approach to measuring ELA is the grounding of the dimensions in current evidence on neurodevelopment (McLaughlin et al., Reference McLaughlin, Sheridan and Lambert2014). Threat experiences may result in changes to emotional learning networks (Johansen, Cain, Ostroff, & LeDoux, Reference Johansen, Cain, Ostroff and LeDoux2011; LeDoux, Reference LeDoux2003; McLaughlin et al., Reference McLaughlin, Sheridan and Lambert2014). Kraynak et al. looked at pathways from childhood abuse to systemic inflammation and corticolimbic functionality in healthy adults and found that childhood abuse was associated with increased IL-6 and indirectly associated with reduced amygdala– ventromedial prefrontal cortex (vmPFC) connectivity (2019). Consistent with recent neurobiological models of early life threat experiences, associations between childhood physical abuse and adulthood corticolimbic circuit functionality may be partially explained by inflammatory processes (Kraynak et al., Reference Kraynak, Marsland, Hanson and Gianaros2019). Interestingly, while CRP associated negatively with amygdala– ventromedial prefrontal cortex connectivity, it did not statistically associate with childhood physical abuse (Kraynak et al., Reference Kraynak, Marsland, Hanson and Gianaros2019). This study provides preliminary evidence of how a dimensional and neurobiological approach may elucidate ELA–inflammation–psychopathology connections.

In contrast to the construct of “threat,” deprivation experiences are defined as the absence of expected environmental inputs (i.e., cognitive inputs, social inputs, and environmental stimuli that are species- and age-typical in their complexity (McLaughlin et al., Reference McLaughlin, Sheridan and Lambert2014). Experiences of deprivation are central for children who have been institutionalized (i.e., orphanage care), children experiencing neglect, and children in contexts of poverty. Few studies of adult cohorts operationalize neglect as a separate dimensional construct, and the definition of neglect is heterogeneous across and within these studies. For example, emotional neglect was not associated with either IL-6 or CRP in a large US cohort and physical neglect but not emotional neglect was positively associated with inflammation in a large British cohort (Pinto Pereira et al., Reference Pinto Pereira, Stein Merkin, Seeman and Power2019). The Baumeister meta-analysis demonstrated that parental absence was significantly and positively associated with CRP, though the measures of deprivation across the studies were highly inconsistent (Baumeister et al., Reference Baumeister, Akhtar, Ciufolini, Pariante and Mondelli2016).

Studies of adults and children who experienced institutional care early in life provide an opportunity to examine the dimension of deprivation as it relates to inflammatory outcomes. Institutional care typically involves environmental deprivation, caregiver separation, and neglect. Two studies that focused on neglect exposure and subsequent inflammation compared previously institutionalized (PI) adolescents to community controls. One study examined youth adopted out of institutional care before the age of 2 y into well-resourced families in the United States (Reid et al., Reference Reid, Coe, Doyle, Sheerar, Slukvina, Donzella and Gunnar2019) and another examined youth who either remained in institutional care or were randomly assigned to a foster care intervention (Slopen et al., Reference Slopen, Tang, Nelson, Zeanah, McDade, McLaughlin and Fox2019). These studies of adolescents removed from institutional care as infants provide opportunities to look at sensitive periods of adversity exposure as it relates to inflammation and compare differences in behavioral and environmental inputs. Neither study found evidence of increased peripheral inflammation in PI adolescents (Reid et al., Reference Reid, Coe, Doyle, Sheerar, Slukvina, Donzella and Gunnar2019; Slopen et al., Reference Slopen, Tang, Nelson, Zeanah, McDade, McLaughlin and Fox2019). One study found no differences in plasma levels of inflammation in PI and comparison adolescents (Reid et al., Reference Reid, Coe, Doyle, Sheerar, Slukvina, Donzella and Gunnar2019), which replicated findings in PI adults (Elwenspoek et al., Reference Elwenspoek, Hengesch, Leenen, Schritz, Sias, Schaan and Muller2017a). These studies are consistent with a study with the Bucharest Early Intervention Project, in which inflammation (CRP, IL-6, IL-8, TNF-α) was measured in dried blood spots in adolescent follow-up of institutionalized infants randomized to care as usual (n = 68) or foster care intervention (n = 68), and compared to never institutionalized controls (n = 127). The study found no differences in levels of peripheral inflammation markers between the PI and comparison adolescents (Slopen et al., Reference Slopen, Tang, Nelson, Zeanah, McDade, McLaughlin and Fox2019). However, further post hoc analyses found differences in CRP only in PI adolescents who exhibited an increasing adiposity trajectory and glucose dysregulation, though this group of adolescents was a small subset of the full PI sample (Tang et al., Reference Tang, Slopen, Nelson, Zeanah, Georgieff and Fox2018).

Unfortunately, none of the studies reviewed have a dimensional measure of threat and deprivation, and many of the conditions of threat and deprivation co-occur. Current studies suggest that inflammatory outcomes can be diverse even between different threat and deprivation exposures. The field would be well served by research that includes dimensional measures of threat and deprivation in better understand how dimensional constructs of ELA do – or do not – result in a proinflammatory phenotype.

Construct invariance from infancy to childhood and adolescence

Longitudinal construct invariance is the third measurement research question for the study of ELA and inflammation. Many studies measure the family environment in infancy and early childhood, as this proximal family environment is thought to best capture experiences of adversity at very young ages. David, Measelle, Ostlund, and Ablow (Reference David, Measelle, Ostlund and Ablow2017) found that salivary CRP was positively and independently associated with socioeconomic disadvantage and maternal stress in 18-month-olds. In contrast, a study of 125 5-year-olds found no association between family level adversity and salivary IL-6 or TNF-alpha (Riis et al., Reference Riis, Granger, Minkovitz, Bandeen-Roche, DiPietro and Johnson2016). In a prospective, longitudinal cohort of 600 adolescents, exposure to family-level adversity in infancy predicted higher levels of CRP, mediated through higher BMI in childhood (Reid et al., Reference Reid, Doom, Argote, Correa-Burrows, Lozoff, Blanco and Gahagan2020). Finally, a study of 337 11-year-olds found no association between family-level adversity in infancy and IL-6 or TNF-alpha. The study found significant associations only between caregiver depressive symptoms in infancy and higher levels of CRP in childhood (O'Connor et al., Reference O'Connor, Willoughby, Moynihan, Messing, Vallejo Sefair, Carnahan and Caserta2020). Few studies examine ELA in infancy and early childhood, and none yet provide clarity around construct invariance from family adversity to later forms of ELA and inflammation.

Tissue specificity

Though the connection between peripheral inflammation and neuroinflammation has been made (Miller & Raison, Reference Miller and Raison2016), there are still open questions that researchers should continue to acknowledge when studying adversity, inflammation, and psychopathology risk. A meta-analysis of studies examining cerebrospinal fluid, positron emission tomography and postmortem brain tissue in patients with depression found that patients with depression have higher cerebral spinal fluid levels of IL-6 and TNF-α than controls. However, there was little correlation between central and peripheral markers of inflammation in the studies reviewed, which may reflect a lack of connection between central and peripheral inflammation or other possible moderating factors (Enache, Pariante, & Mondelli, Reference Enache, Pariante and Mondelli2019). Nonetheless, measures in blood, and therefore blood draws, are considered the gold standard for inflammatory biomarker measurement in population-based studies. Most studies investigating the links between ELA and inflammation focus on blood as the target tissue of interest in populations of adults, adolescents, and many studies of children.

Large-scale studies and studies of infants and children often use less invasive biomarkers for acceptability and cost. Blood draws can be cost-prohibitive, not feasible, and are not minimally invasive. One less invasive methods of collection for inflammatory biomarkers involve finger prick and blood spot (McDade, Williams, & Snodgrass, Reference McDade, Williams and Snodgrass2007). The blood spot cards are let to dry and can be stored until laboratory analysis with good stability. Levels of inflammation biomarkers, such as CRP, in dried blood spot correlate highly with levels in serum (r = 0.96) (McDade, Burhop, & Dohnal, Reference McDade, Burhop and Dohnal2004). Analysis of ELA-related inflammation abnormalities in dried blood spot showed results comparable to those in serum (Copeland et al., Reference Copeland, Wolke, Lereya, Shanahan, Worthman and Costello2014; Danese et al., Reference Danese, Caspi, Williams, Ambler, Sugden, Mika and Arseneault2011). As such, there is interest in developing further inflammation biomarkers including cytokines from dried blood spots.

Blood draws and dried blood spots may be considered too invasive in studies with healthy infants or young children. Reflecting this, the majority of the studies in infancy and early childhood (0–5 years) in this review use salivary markers of inflammation, which limits our ability to compare these with studies in older children and adults. Several studies have reported poor correlations between saliva and blood for inflammatory markers (Dillon et al., Reference Dillon, Holmes, Birk, Brooks, Lyons-Ruth and Pizzagalli2009; Riis et al., Reference Riis, Granger, Minkovitz, Bandeen-Roche, DiPietro and Johnson2016), likely due to the relative difficulty that most proteins have in passing the multiple barriers between blood and saliva compartments. Cytokines are too large to enter the mouth from the periphery through passive diffusion and instead enter through sites of inflammation or tissue damage, producing large variability between individuals (Bosch, Reference Bosch2014). Thus, while salivary cytokines are mostly indicative of local inflammation (Bosch, Reference Bosch2014), they may still serve as important indicators of health status while taking care to control for dental health confounds (Slavish, Graham-Engeland, Smyth, & Engeland, Reference Slavish, Graham-Engeland, Smyth and Engeland2015). As in any tissue, the type of inflammatory biomarker assessed is important. CRP from saliva has been associated with CRP in circulating blood more consistently than other inflammatory biomarkers, perhaps because CRP is produced in the liver, not in the mouth, and its most likely route into saliva is via blood (Slavish et al., Reference Slavish, Graham-Engeland, Smyth and Engeland2015). A validation study of a high-sensitivity CRP assay in saliva allowed prediction of serum CRP in healthy adults (Ouellet-Morin, Danese, Williams, & Arseneault, Reference Ouellet-Morin, Danese, Williams and Arseneault2011). Markers of salivary inflammation have been stable across time in adolescent samples in the short and long term (when averaging across multiple assessments) (Shields, Slavich, Perlman, Klein, & Kotov, Reference Shields, Slavich, Perlman, Klein and Kotov2019). More research is needed to establish both the health relevance and the mechanisms driving salivary inflammation generally and across development. Further research is needed to determine how ELA relates to salivary inflammation, what increased salivary inflammation means for infants and young children in the short term, whether this measure is associated with blood levels of inflammation, or what the long-term implications of increased salivary inflammation are.

Biomarker (cytokine) selection

Diverse inflammatory markers have been used to test the association between ELA and inflammation. To counteract the literature's heterogeneity, longitudinal and system-wide studies of inflammation regulation with multiple biomarkers that can differentiate trait- and state-specific changes in the context of childhood adversity would improve our understanding of etiology and establish valuable biomarkers. Del Guidice and Gangestad make the argument that IL-6 and CRP assays alone are not enough to establish the existence of inflammatory states due to their multiple roles and states in the body, a concept accepted in immunology research but rarely addressed in behavioral research (Del Giudice & Gangestad, Reference Del Giudice and Gangestad2018). For example, IL-6 and CRP might be elevated for maintenance functions and not necessarily indicative of an aberrant and chronic inflammatory state. Future studies should consider using IL-6 and CRP alongside additional inflammatory biomarkers (e.g., IL-1B, TNF-alpha, and circulating levels of soluble IL-6 receptors such as sIL-6R) (Del Giudice & Gangestad, Reference Del Giudice and Gangestad2018). A meta-analysis of 704 publications on psychopathology and 44 inflammatory biomarkers found that related disorders share similar patterns of inflammatory changes (Yuan et al., Reference Yuan, Chen, Xia, Dai and Liu2019). Longitudinal studies of inflammation regulation with multiple biomarkers could help differentiate trait- and state-specific changes in psychopathology to establish valuable biomarkers in the ELA literature (Yuan et al., Reference Yuan, Chen, Xia, Dai and Liu2019).

Gene-environment correlation

The genetic vulnerability of inflammation is estimated to account for between 13%–30% of within-subjects variability of CRP (Pankow et al., Reference Pankow, Folsom, Cushman, Borecki, Hopkins, Eckfeldt and Tracy2001), and there may be shared, underlying genetic vulnerability to ELA, psychopathology, and inflammation (2017). Immune-related genetic pathways have been found to predict risk for several types of psychopathology (Network & Pathway Analysis Subgroup of Psychiatric Genomics, 2015). Danese & Baldwin (Reference Danese and Baldwin2017) note that an individual exhibiting early behavioral expressions of genetic vulnerability to psychopathology (e.g., emotional dysregulation, oppositional behavior) might increase their risk of maltreatment in childhood. However, the study on victimization by Baldwin et al. (Reference Baldwin, Arseneault, Caspi, Fisher, Moffitt, Odgers and Danese2018) with a genetically informative sample of twins found that exposure to ELA was associated with increased inflammation above and beyond the genetic risk for increased inflammation, further suggesting that while genetics plays a role, additional mechanisms are at play.

Acute inflammatory responses

Acute inflammation in response to physical trauma, infection, or exposure to pathogens during sensitive periods early in life may stimulate microglia and neuroendocrine activity and affect brain development. Acute inflammatory responses can be incited through psychological stress or in the context of physical trauma, as children exposed to ELA are more likely to be exposed to physical injuries (Gilbert et al., Reference Gilbert, Widom, Browne, Fergusson, Webb and Janson2009). Children exposed to ELA are more likely to be exposed to infection (Gilbert et al., Reference Gilbert, Widom, Browne, Fergusson, Webb and Janson2009) and psychological stress and traumatic experiences can increase susceptibility to both infections and recurrent viral infections (Cohen et al., Reference Cohen, Janicki-Deverts, Doyle, Miller, Frank, Rabin and Turner2012; Shirtcliff, Coe, & Pollak, Reference Shirtcliff, Coe and Pollak2009). ELA could contribute to systemic inflammation through these recurrent acute immune perturbations, but this pathway may not be applicable in all contexts. For example, in the studies of PI adolescents and young adults mentioned earlier in this paper, ELA-exposed youth have been found to exhibit higher levels of cytomegalovirus (CMV) titers, likely due to increased exposure to infectious disease and pathogens in institutional care (Elwenspoek et al., Reference Elwenspoek, Sias, Hengesch, Schaan, Leenen, Adams and Turner2017b; Reid et al., Reference Reid, Coe, Doyle, Sheerar, Slukvina, Donzella and Gunnar2019). ELA and CMV exposure were associated with acquired immune system perturbations and markers of T-cell senescence, but not higher levels of inflammation (Elwenspoek et al., Reference Elwenspoek, Sias, Hengesch, Schaan, Leenen, Adams and Turner2017b; Reid et al., Reference Reid, Coe, Doyle, Sheerar, Slukvina, Donzella and Gunnar2019). In addition, McDade's study of a prospective birth cohort in the Philippines found that ELA was associated with adult levels of inflammation only in individuals with low levels of microbial exposure in infancy (McDade, Rutherford, Adair, & Kuzawa, Reference McDade, Rutherford, Adair and Kuzawa2010). It is unclear how ELA and early infection interact, the salience of specific pathogen exposures, and how infection affects the developing immune system. For instance, PI youth are not at increased risk of more inflammation (Elwenspoek et al., Reference Elwenspoek, Hengesch, Leenen, Schritz, Sias, Schaan and Muller2017a; Reid et al., Reference Reid, Coe, Doyle, Sheerar, Slukvina, Donzella and Gunnar2019; Slopen et al., Reference Slopen, Tang, Nelson, Zeanah, McDade, McLaughlin and Fox2019), but they are at increased risk of psychopathology (Gunnar & Reid, Reference Gunnar and Reid2019). In the Philippines cohort, there was no evidence of association between depressive symptomology in adulthood and increased inflammation (McDade, Borja, Adair, & Kuzawa, Reference McDade, Borja, Adair and Kuzawa2013). In contexts where infectious disease burden is high, pathogen exposure may act as a protective factor in the pathway from ELA, inflammation, and psychopathology (McDade, Reference McDade2012). Future work should consider operationalizing pathogen and infection exposure across development to better understand inflammation and psychopathology risk.

Neuroendocrine dysregulation

Psychological stress can indirectly contribute to inflammation by inducing neuroendocrine abnormalities and changes in HPA axis physiology. The HPA axis is a key neuroendocrine pathway involved in processes critical to homeostasis, including an organism's responses to challenge, metabolism, and immune activity. Peripheral and central HPA axis mediators affect brain function (McEwen, Weiss, & Schwartz, Reference McEwen, Weiss and Schwartz1968), and early life experiences can significantly shape the development and subsequent functioning of the HPA axis (Levine, Alpert, & Lewis, Reference Levine, Alpert and Lewis1957). Furthermore, the HPA axis and the immune system relate to each other in dynamic and bidirectional ways. Inflammation can activate the HPA axis and cortisol activates anti-inflammatory and suppresses proinflammatory responses (Barnes & Adcock, Reference Barnes and Adcock2009; Besedovsky et al., Reference Besedovsky, del Rey, Sorkin and Dinarello1986). In situations where chronic stress leads to HPA upregulation, immune cells can become insensitive to the GR-mediated inhibitory effects of cortisol after excessive exposure over long periods, thought to be indicative of “glucocorticoid resistance” (Cohen et al., Reference Cohen, Janicki-Deverts, Doyle, Miller, Frank, Rabin and Turner2012; Kuhlman et al., Reference Kuhlman, Horn, Chiang and Bower2020; Miller, Chen, & Parker, Reference Miller, Chen and Parker2011b).

HPA biology in humans is typically measured in the context of cortisol reactivity to stress and the diurnal rhythm of cortisol secretion, which reflect the ability of the HPA axis to regulate responses to stress and to modulate circadian rhythm. In chronic ELA exposures, the evidence suggests that the HPA axis responsivity can either upregulate, where cortisol becomes chronically elevated, or downregulate, where the HPA axis becomes become flat and unresponsive (McEwen, Reference McEwen2007, Reference McEwen2012). In general, current evidence shows that childhood adversity is associated with heightened activity of the axis (Koss & Gunnar, Reference Koss and Gunnar2018; Sanchez et al., Reference Sanchez, Ladd and Plotsky2001). If the axis is chronically elevated, it may produce a blunted cortisol response (Koss & Gunnar, Reference Koss and Gunnar2018; Sanchez et al., Reference Sanchez, Ladd and Plotsky2001).

Blunted cortisol patterns have been found in studies of PI children (Koss, Hostinar, Donzella, & Gunnar, Reference Koss, Hostinar, Donzella and Gunnar2014; Koss, Mliner, Donzella, & Gunnar, Reference Koss, Mliner, Donzella and Gunnar2016) and in children experiencing neglect and maltreatment (Bernard, Hostinar, & Dozier, Reference Bernard, Hostinar and Dozier2015; Bruce, Fisher, Pears, & Levine, Reference Bruce, Fisher, Pears and Levine2009). Blunted cortisol patterns may lead to different Neuroendocrine×Immune System interactions. Inflammatory cytokine production is typically inhibited when cortisol binds to GRs in healthy immune cells (Barnes & Adcock, Reference Barnes and Adcock2009). Thus, in the case of downregulated HPA activity, low circulating levels of cortisol might induce elevated levels of inflammatory cytokines (Del Giudice & Gangestad, Reference Del Giudice and Gangestad2018). Studies have found insufficient glucocorticoid signaling in adolescents and adults with a history of childhood maltreatment, which could bring about systemic inflammation (Heim, Ehlert, & Hellhammer, Reference Heim, Ehlert and Hellhammer2000; Miller, Cohen, & Ritchey, Reference Miller, Cohen and Ritchey2002; Miller, Freedland, Carney, Stetler, & Banks, Reference Miller, Freedland, Carney, Stetler and Banks2003). However, this finding is not always consistent: the study of 11-year-olds mentioned previously found no association between glucocorticoid resistance and either ELA exposure in infancy or inflammation at 11 years, even though early adversity exposure was positively associated with inflammation (O'Connor et al., Reference O'Connor, Willoughby, Moynihan, Messing, Vallejo Sefair, Carnahan and Caserta2020). It is unclear whether this discrepancy is a result of the need for cumulative adversities to result in glucocorticoid resistance, a level of adversity that was not severe enough to impact glucocorticoid resistance, or an indication of multiple adversity-to-inflammation pathways across development.

Proinflammatory effects on immune cells in development may only be reliably observed when immune cells are stimulated in vitro, not when measured in circulation (Kuhlman et al., Reference Kuhlman, Horn, Chiang and Bower2020). If youth exposed to ELA exhibit elevated concentrations of cortisol, this upregulation may be masking the proinflammatory phenotype in children if inflammation is measured with circulating markers (Kuhlman et al., Reference Kuhlman, Horn, Chiang and Bower2020; Tarullo & Gunnar, Reference Tarullo and Gunnar2006). Chronic HPA activation during sensitive developmental periods can induce epigenetic changes in the GR, reducing signaling in a compensatory manner (Weaver et al., Reference Weaver, Cervoni, Champagne, D'Alessio, Sharma, Seckl and Meaney2004). This signal reduction could lead to resistance to cortisol's anti-inflammatory properties. ELA is associated with DNA demethylation in the glucocorticoid response elements of the FK506 binding protein 5 (FKBP5) gene and reduced sensitivity to glucocorticoids’ inhibitory effect on IL-6 production in peripheral blood immune cells stimulated in vitro (Klengel et al., Reference Klengel, Mehta, Anacker, Rex-Haffner, Pruessner, Pariante and Binder2013). A study in 18-year-old females retrospectively reporting ELA did not find higher levels of peripheral IL-6, but in vitro analyses found pronounced cytokine responses to bacterial challenge and evidence for decreased efficacy of cortisol to regulate the inflammatory response (Miller & Chen, Reference Miller and Chen2010). HPA axis and immune system interactions highlight the need to consider how neuroendocrine responses to adversity may increase risk for psychopathology through alterations in the co-regulation of multiple systems. Few studies on childhood adversity investigate HPA axis functioning with immune functioning (Kuhlman et al., Reference Kuhlman, Horn, Chiang and Bower2020). More complex models might clarify how childhood adversity impacts the brain and behavior through both HPA- and immune-mediating systems.

Gut microbiome dysbiosis

Growing evidence suggests that the microbiome plays a role in the pathway between ELA, stress-mediating and immune-system development, and mental health. The gut microbiome is involved in the immune system development and function (Hooper, Littman, & Macpherson, Reference Hooper, Littman and Macpherson2012). Its effects include changes in inflammation (Cani, Osto, Geurts, & Everard, Reference Cani, Osto, Geurts and Everard2012; Wells et al., Reference Wells, Brummer, Derrien, MacDonald, Troost, Cani and Garcia-Rodenas2017) and immune competence in response to pathogen exposure and infection (Mackos, Maltz, & Bailey, Reference Mackos, Maltz and Bailey2017; Zanella Terrier, Simonet, Bichard, & Frossard, Reference Zanella Terrier, Simonet, Bichard and Frossard2014). Gut microbiome dysbiosis has been associated with psychological impairments (Cryan & Dinan, Reference Cryan and Dinan2012; McVey Neufeld, Mao, Bienenstock, Foster, & Kunze, Reference McVey Neufeld, Mao, Bienenstock, Foster and Kunze2013). ELA may alter communication between the gut and the brain through altered vagus nerve and enteric nervous system signaling, HPA axis activation, immunomodulation, and inflammation (Cryan & Dinan, Reference Cryan and Dinan2012; Gareau, Silva, & Perdue, Reference Gareau, Silva and Perdue2008; Konturek, Brzozowski, & Konturek, Reference Konturek, Brzozowski and Konturek2011; Segerstrom & Miller, Reference Segerstrom and Miller2004). Psychosocial stress has been associated with changes in gut microbiota structure and activity (Mackos et al., Reference Mackos, Maltz and Bailey2017). In rodent and nonhuman primate models, early maternal separation has transient and long-term effects on the gut microbiota (Bailey & Coe, Reference Bailey and Coe1999; O'Mahony et al., Reference O'Mahony, Marchesi, Scully, Codling, Ceolho, Quigley and Dinan2009). In murine models, ELS-induced alterations of the microbiome persist into adulthood (Garcia-Rodenas et al., Reference Garcia-Rodenas, Bergonzelli, Nutten, Schumann, Cherbut, Turini and Corthesy-Theulaz2006; Jasarevic, Howard, Misic, Beiting, & Bale, Reference Jasarevic, Howard, Misic, Beiting and Bale2017; Jasarevic, Rodgers, & Bale, Reference Jasarevic, Rodgers and Bale2015). Changes to the gut microbiome in these models have been associated with anxiety-like behaviors and activation of stress-mediating systems (Gareau et al., Reference Gareau, Silva and Perdue2008). Rodents exposed to stress exhibit inflammation, altered gastrointestinal (GI) function and permeability, and differences in immune activity (Gareau et al., Reference Gareau, Silva and Perdue2008). In humans, differences in the gut microbiome have been found in adults who have developed PTSD after trauma exposure and adults with a history of childhood trauma (Hemmings et al., Reference Hemmings, Malan-Muller, van den Heuvel, Demmitt, Stanislawski, Smith and Lowry2017). ELA-exposed children have also been found to exhibit gut microbiome differences (Callaghan et al., Reference Callaghan, Fields, Gee, Gabard-Durnam, Caldera, Humphreys and Tottenham2020). Although conducted in a small sample of children, this study observed that gut bacteria levels were correlated with prefrontal cortex activation to emotional faces (Callaghan et al., Reference Callaghan, Fields, Gee, Gabard-Durnam, Caldera, Humphreys and Tottenham2020). The relationship between ELA-induced gut dysbiosis and inflammation are likely bidirectional (Cryan & Dinan, Reference Cryan and Dinan2012; Hooper et al., Reference Hooper, Littman and Macpherson2012).

Stress-induced nutrient metabolism disruptions

Disruptions to nutrient metabolism is an unexplored pathway by which ELS can lead to inflammation. To function properly, the developing immune system requires macronutrients (i.e., protein, fat, carbohydrates) and micronutrients (e.g., iron, zinc, B vitamins etc.). Psychosocial stress and dysregulated neuroendocrine pathways can disrupt nutrient metabolism even in the context of adequate nutrient intake (Monk, Georgieff, & Osterholm, Reference Monk, Georgieff and Osterholm2013; Suchdev et al., Reference Suchdev, Boivin, Forsyth, Georgieff, Guerrant and Nelson2017) and lead to through disruptions in immune functioning. There are many required nutrients for immune system development, but we will use iron as an illustrative example. Iron deficiency is one of the most common forms of malnutrition, with an estimated 4 in 10 children under 5 being affected (Stevens et al., Reference Stevens, Finucane, De-Regil, Paciorek, Flaxman and Branca2013), and children exposed to ELA are often at higher risk of iron deficiency. Adequate iron intake and utilization is necessary for normal white blood cell and cytokine function (Coe, Lubach, & Shirtcliff, Reference Coe, Lubach and Shirtcliff2007). In murine models, early iron deficiency increased expression of proinflammatory cytokine gene pathways in the adult brain even after iron repletion. Neonatal iron deficiency also reduced anti-inflammatory gene expression and altered cell-mediated immune response genes in infant mice (Tran, Singh, Wallin, & Georgieff, Reference Tran, Singh, Wallin and Georgieff2018). Research from a nonhuman primates showed that exposure to early life stress resulted in lower levels of iron and iron status subsequently predicted greater deficits in innate immune system functioning (Coe et al., Reference Coe, Lubach and Shirtcliff2007). Few studies exist in human models, but one study found that a randomized control trial of iron supplementation was associated with decreased levels of inflammation in a cohort of adolescents, even after exposure to ELA in infancy (Reid et al., Reference Reid, Doom, Argote, Correa-Burrows, Lozoff, Blanco and Gahagan2020). ELA-induced nutrient dysregulation could be a mediator that is ripe for research and intervention.

Adipose tissue

Adipose tissue generates and regulates about one-third of circulating inflammatory markers (Black, Reference Black2003). Obesity in childhood is associated with low-grade inflammation (Visser, Bouter, McQuillan, Wener, & Harris, Reference Visser, Bouter, McQuillan, Wener and Harris2001), and ELA is a risk factor for obesity in experimental animal models and human studies (Bzostek & Beck, Reference Bzostek and Beck2011; Danese & Tan, Reference Danese and Tan2014; Gundersen, Mahatmya, Garasky, & Lohman, Reference Gundersen, Mahatmya, Garasky and Lohman2011; Jelleyman & Spencer, Reference Jelleyman and Spencer2008; Weinberg et al., Reference Weinberg, Dietz, Stoyak, Melhem, Porta, Payne and Brent2013). The links between ELA and increased adiposity could arise from a biologically and behaviorally driven “thrifty” phenotype characterized by increased energy intake, increased storage, and/or decreased energy expenditure (Danese & Tan, Reference Danese and Tan2014). Childhood stress has been linked with emotional eating and unhealthy dietary patterns (Michels et al., Reference Michels, Sioen, Braet, Eiben, Hebestreit, Huybrechts and De Henauw2012), perhaps due to “self-medicating” with energy dense foods (Baldwin & Danese, Reference Baldwin and Danese2019). Impaired reward processing and reduced self-control from ELA may lead to increased intake of high-fat, high-sugar foods (Evans, Fuller-Rowell, & Doan, Reference Evans, Fuller-Rowell and Doan2012; Hostinar, Ross, Chen, & Miller, Reference Hostinar, Ross, Chen and Miller2015b). The neuroendocrine pathway between ELA, obesity, and inflammation could be a potent one, as glucocorticoids contribute to adipose tissue development (Dallman et al., Reference Dallman, Strack, Akana, Bradbury, Hanson, Scribner and Smith1993) and cortisol plays a central role in regulating food intake and metabolism (Dallman, Reference Dallman2010; Dallman, Pecoraro, & la Fleur, Reference Dallman, Pecoraro and la Fleur2005). Thus, early adversity exposure could increase the risk for children and adolescents to engage in eating behaviors that contribute to increased adiposity and inflammation.

ELA could result in obesity through decreased energy expenditure from metabolic hormone alterations and mental health problems (Baldwin & Danese, Reference Baldwin and Danese2019). ELA is associated with changes in leptin levels, which are responsible for lipolysis (Danese et al., Reference Danese, Dove, Belsky, Henchy, Williams, Ambler and Arseneault2014; Panagiotaropoulos et al., Reference Panagiotaropoulos, Papaioannou, Pondiki, Prokopiou, Stylianopoulou and Gerozissis2004). Mental health problems associated with child adversity are also associated with physical inactivity (Baldwin & Danese, Reference Baldwin and Danese2019). Impaired functioning in neuroendocrine and leptin pathways could dysregulate thermogenesis and lipolysis, leading to decreased energy expenditure and increased adipose tissue growth (Danese & Tan, Reference Danese and Tan2014). Inflammation itself can induce fatigue (Dantzer et al., Reference Dantzer, O'Connor, Freund, Johnson and Kelley2008), and thus inflammation from ELA exposure could cyclically contribute to reduced energy expenditure, increased adipose tissue, and increased inflammation.

Studies have found that ELA was associated with adolescent levels of inflammation only in the context of accelerated BMI growth trajectories (Reid et al., Reference Reid, Doom, Argote, Correa-Burrows, Lozoff, Blanco and Gahagan2020; Tang et al., Reference Tang, Slopen, Nelson, Zeanah, Georgieff and Fox2018). In a number of studies, adiposity attenuated or abolished several of the associations between ELA and inflammation (Matthews et al., Reference Matthews, Chang, Thurston and Bromberger2014; Pinto Pereira et al., Reference Pinto Pereira, Stein Merkin, Seeman and Power2019), though others have found that the association between adversity and inflammation was independent of obesity in both adult and pediatric samples (Baldwin et al., Reference Baldwin, Arseneault, Caspi, Fisher, Moffitt, Odgers and Danese2018; Danese et al., Reference Danese, Caspi, Williams, Ambler, Sugden, Mika and Arseneault2011; Danese et al., Reference Danese, Pariante, Caspi, Taylor and Poulton2007; O'Connor et al., Reference O'Connor, Willoughby, Moynihan, Messing, Vallejo Sefair, Carnahan and Caserta2020). This adversity–obesity–inflammation pathway seems to be of increased importance in studies of childhood adversity and inflammation in adolescence.

Brain functioning and behavior

ELA can instantiate a chronic inflammatory state through brain and behavioral changes. The Neuroimmune Network Hypothesis outlines the process by which ELA prompts crosstalk between the neural and immune systems to create a loop of increased threat sensitivity, inflammation, and unhealthy behaviors that can result in increased risk for chronic inflammation (Nusslock & Miller, Reference Nusslock and Miller2016). ELA is associated with impaired executive function, impaired reward processing, and hippocampal and prefrontal cortex abnormalities in human and nonhuman animal studies, which could lead to impaired ability to suppress unwanted behavior (Brunson et al., Reference Brunson, Kramar, Lin, Chen, Colgin, Yanagihara and Baram2005; Danese & McEwen, Reference Danese and McEwen2012; Dillon et al., Reference Dillon, Holmes, Birk, Brooks, Lyons-Ruth and Pizzagalli2009; Matthews & Robbins, Reference Matthews and Robbins2003; Mehta et al., Reference Mehta, Gore-Langton, Golembo, Colvert, Williams and Sonuga-Barke2010; Pechtel & Pizzagalli, Reference Pechtel and Pizzagalli2011; Phillips, Howes, Whitelaw, Robbins, & Everitt, Reference Phillips, Howes, Whitelaw, Robbins and Everitt1994; Pryce, Dettling, Spengler, Spaete, & Feldon, Reference Pryce, Dettling, Spengler, Spaete and Feldon2004). Individuals exposed to ELA are at a higher risk of substance and alcohol abuse disorders (Anda et al., Reference Anda, Croft, Felitti, Nordenberg, Giles, Williamson and Giovino1999; Dube et al., Reference Dube, Felitti, Dong, Chapman, Giles and Anda2003), which contribute to increased inflammation (Shiels et al., Reference Shiels, Katki, Freedman, Purdue, Wentzensen, Trabert and Chaturvedi2014). Though less relevant for children, adolescence might be a salient time for proinflammatory phenotypes to emerge due to increases in risky behaviors, such as smoking and habits that contribute to weight gain (Raposa, Bower, Hammen, Najman, & Brennan, Reference Raposa, Bower, Hammen, Najman and Brennan2014), and impulsive behaviors that increase stress exposure (Lovallo, Reference Lovallo2013; Raposa et al., Reference Raposa, Bower, Hammen, Najman and Brennan2014).

Another behavioral pathway that could follow from ELA to promote inflammation is through persistent or recurring distress and self-harming behaviors. Maltreatment has been associated with heightened threat perception (Leppanen & Nelson, Reference Leppanen and Nelson2009; Pollak & Kistler, Reference Pollak and Kistler2002) and heightened amygdala activation in response to negative emotional stimuli (McCrory et al., Reference McCrory, De Brito, Kelly, Bird, Sebastian, Mechelli and Viding2013; Tottenham et al., Reference Tottenham, Hare, Millner, Gilhooly, Zevin and Casey2011). Depression and inflammation are bidirectionally associated in longitudinal studies of adults and adolescents (Matthews et al., Reference Matthews, Schott, Bromberger, Cyranowski, Everson-Rose and Sowers2010; Miller & Cole, Reference Miller and Cole2012). An individual's perception of stressors, emotional symptoms, and behavioral responses when distressed could increase their inflammation levels (Danese & Lewis, Reference Danese and Lewis2017). An individual's increased exposure to stressors and their emotional symptoms can further lead to chronic levels of stress and individuals may engage in self-harm to relieve feelings of distress (Baldwin, Arseneault, Caspi et al., Reference Baldwin, Arseneault, Caspi, Moffitt, Fisher, Odgers and Danese2019), offering another behavioral pathway by which ELS can contribute to increased inflammation.

Sleep

There are reciprocal links between sleep and inflammation, and we direct the reader to a thorough review by Irwin (Reference Irwin2015). Sleep can regulate the immune response: sleep disturbance has been linked with reduced vaccine efficacy (e.g. Prather et al., Reference Prather, Hall, Fury, Ross, Muldoon, Cohen and Marsland2012; Spiegel, Sheridan, & Van Cauter, Reference Spiegel, Sheridan and Van Cauter2002), impaired response to infectious challenge (Cohen, Doyle, Alper, Janicki-Deverts, & Turner, Reference Cohen, Doyle, Alper, Janicki-Deverts and Turner2009; Toth, Reference Toth1995), and upregulation of inflammation (Irwin, Reference Irwin2015). Sleep influences the HPA axis, which in turn regulates the immune system through the diurnal cortisol rhythm (Irwin & Cole, Reference Irwin and Cole2011; Slavich & Irwin, Reference Slavich and Irwin2014). In experimental and naturalistic epidemiological studies, sleep deprivation is associated with increased expression of proinflammatory cytokines and elevated inflammation (Irwin, Wang, Campomayor, Collado-Hidalgo, & Cole, Reference Irwin, Wang, Campomayor, Collado-Hidalgo and Cole2006; Miller et al., Reference Miller, Kandala, Kivimaki, Kumari, Brunner, Lowe and Cappuccio2009). Maternal separation in rodent models decreased sleep duration and disrupted sleep architecture for stress-affected animals (Mrdalj et al., Reference Mrdalj, Pallesen, Milde, Jellestad, Murison, Ursin and Gronli2013). A history of childhood trauma is associated with increased sleep problems (Gregory & Sadeh, Reference Gregory and Sadeh2016), independent of current PTSD or depression diagnoses (Noll, Trickett, Susman, & Putnam, Reference Noll, Trickett, Susman and Putnam2006). Unfortunately, prospective evidence that demonstrates a mediating role for inflammation in the association between sleep disruptions and psychopathology is not yet available (Irwin, Reference Irwin2015). To our knowledge, no studies have extended this work to investigate the specific pathways between ELA and sleep disruption as they relate to inflammation and psychopathology.

Age of assessment

More research is needed to understand how ELA across development contributes to an inflammatory phenotype and how inflammatory phenotypes change across development. As noted, acute psychosocial stress activates peripheral inflammation and chronic psychosocial stress is associated with systemic, low-grade inflammation. This transition is thought to be a driver of psychopathology, especially after exposure to childhood adversity. However, it is unclear when an acute stressor becomes a chronic stressor (Rohleder, Nater, Wolf, Ehlert, & Kirschbaum, Reference Rohleder, Nater, Wolf, Ehlert and Kirschbaum2004), and this gap is especially prevalent in studies of children. Longitudinal assessments of inflammation over longer periods of time, repeat assessments of inflammation over a shorter window (for an example, see McDade et al., Reference McDade, Tallman, Madimenos, Liebert, Cepon, Sugiyama and Snodgrass2012), and studies on repeated acute stress might help fill this gap. As these studies do not yet exist, we review the ELA and inflammation literature with a developmental lens.

Adulthood

Much of the evidence linking ELA and later inflammation has been found in adults. Consistent with the animal literature, adults exposed to childhood maltreatment are at greater risk of increased levels of circulating inflammatory markers (Baumeister et al., Reference Baumeister, Akhtar, Ciufolini, Pariante and Mondelli2016; Danese & Lewis, Reference Danese and Lewis2017) and greater proinflammatory responses in response to subsequent stressors (Carpenter et al., Reference Carpenter, Gawuga, Tyrka, Lee, Anderson and Price2010; Gouin et al., Reference Gouin, Glaser, Malarkey, Beversdorf and Kiecolt-Glaser2012; Kiecolt-Glaser et al., Reference Kiecolt-Glaser, Gouin, Weng, Malarkey, Beversdorf and Glaser2011). In clinical adult populations, depressed individuals with a history of ELA exhibit increased inflammatory responses to acute psychosocial stress (Pace et al., Reference Pace, Mletzko, Alagbe, Musselman, Nemeroff, Miller and Heim2006) and show higher levels of inflammation in unstimulated plasma (Danese et al., Reference Danese, Moffitt, Pariante, Ambler, Poulton and Caspi2008). Table 2 collates the studies of adult clinical populations from the Baumeister meta-analysis, specifically focusing on populations with psychiatric diagnoses of first episode of psychosis (FEP), major depressive disorder (MDD), and PTSD as compared with health controls (Baumeister et al., Reference Baumeister, Akhtar, Ciufolini, Pariante and Mondelli2016). A history of ELA appears drive increased inflammation in clinical samples (Baumeister et al., Reference Baumeister, Akhtar, Ciufolini, Pariante and Mondelli2016), as patients without ELA had immune profiles similar to healthy controls (Dennison et al., Reference Dennison, McKernan, Cryan and Dinan2012). Further, psychologically healthy controls with a history of ELA exhibited higher IL-1B compared to controls without ELA exposure (Di Nicola et al., Reference Di Nicola, Cattaneo, Hepgul, Di Forti, Aitchison, Janiri and Mondelli2013).

Table 2. Early adversity and inflammation: adult at inflammation assessment, clinical populations

∅ = null association between ELA exposure and inflammatory outcome.

✓ = positive association between ELA exposure and inflammatory outcome.

CTQ = Child Trauma Questionnaire; CRP = C-reactive protein; ELA = early life adversity; FEP = first episode of psychosis; HC = “heathy controls”; IL-6 = interleukin-6; MDD = major depressive disorder; PTSD = posttraumatic stress disorder

The Baumeister meta-analysis found that age did not have a moderating effect on the relationship between ELA and inflammation in adulthood (Baumeister et al., Reference Baumeister, Akhtar, Ciufolini, Pariante and Mondelli2016). It could be possible that the moderating effect of age is most salient in pediatric populations. Alternatively, adulthood could be a life stage in which the cumulative effects of ELA are easier to capture through peripheral inflammatory cytokines. One study of adults in midlife found support for the stress accumulation model: ELA and recent life event stressors were independently associated with higher levels of inflammation, controlling for each other's effects, and their interaction was not significant (Hostinar, Lachman, Mroczek, Seeman, & Miller, Reference Hostinar, Lachman, Mroczek, Seeman and Miller2015a). In that study, mediation analyses found that ELA and recent life stressors had unique and additive contributions to inflammation through norepinephrine pathways, adiposity, and inflammation-inducing behaviors (Hostinar et al., Reference Hostinar, Lachman, Mroczek, Seeman and Miller2015a). In adults, inflammation might depend on behavioral mediators rather than age.

Sex differences

Experimental, nonhuman animal models suggest that early life stress has sexually dimorphic effects on immune dysregulation (Ganguly & Brenhouse, Reference Ganguly and Brenhouse2015). In adult populations, a meta-analysis found no evidence of a moderating effect of sex on the relationship between ELA and inflammation (Baumeister et al., Reference Baumeister, Akhtar, Ciufolini, Pariante and Mondelli2016). The pediatric meta-analysis was underpowered to examine sex differences. Nonetheless, the association between ELA and CRP was specific only to young women in the genetically informed twin study mentioned previously (Baldwin et al., Reference Baldwin, Arseneault, Caspi, Fisher, Moffitt, Odgers and Danese2018). Sex differences could arise from sex differences in neuroendocrine regulation. For example, male, not female, adolescents exposed to violence in the past year exhibited lower cortisol reactivity to a laboratory stressor (Peckins, Dockray, Eckenrode, Heaton, & Susman, Reference Peckins, Dockray, Eckenrode, Heaton and Susman2012). Sleep disruption may be another avenue by which sex moderates the relationship between ELA, inflammation, and psychopathology. Women appear more likely to develop dysregulated inflammatory responses from sleep disturbance while men appear to be more likely to develop cardiovascular disease and cancer, but not depression (Irwin, Reference Irwin2015). The field needs studies that are well-powered to examine sex differences in pediatric populations, as exposure to ELA and the biological sequalae of inflammation could exhibit important sexually dimorphic effects in humans.

Ecological and demographic differences

Examination of heterogeneous populations and contexts could reveal important mechanisms of how ELA contributes to inflammation and psychopathology. Many of the studies reviewed have been conducted in high-income country contexts, though children in low- and middle-income country contexts are no less exposed to ELA. However, the processes by which ELA may instantiate inflammation might be different for children outside of high-income country and often urban contexts. Studies in representative cohorts in the Philippines, as mentioned previously, and Tanzania suggest that the typical high-adversity, high-inflammation associations may be moderated by contextual factors that are unique to high-microbial, rural, or low- or middle-income country contexts (Hadley & Decaro, Reference Hadley and Decaro2014; McDade, Hoke, Borja, Adair, & Kuzawa, Reference McDade, Hoke, Borja, Adair and Kuzawa2013). In a pediatric population, a large study of Tanzanian children under 5 found that CRP was positively associated with adversity quantified as maternal illiteracy, larger household size, and disease load (Hadley & Decaro, Reference Hadley and Decaro2014). Interestingly, membership in the least poor wealth category was associated with higher levels of CRP. As this sample of Tanzanian children were more likely to be stunted, with an average height-for-age of −1.66, and an average CRP level of 4.4 mg/l, it is difficult to compare with the studies of infants conducted in higher-income countries (Hadley & Decaro, Reference Hadley and Decaro2014). However, this study points to important contextual considerations when researching young children, ELA, and inflammatory biomarkers in more diverse country contexts.

Even in the context of Western, educated, industrialized, rich and democratic (WEIRD) societies, under-represented ethnic and race groups in predominantly white countries might experience differences in context that contribute to increases in inflammation after ELA exposure. For example, Pereira's analysis of the large US cohort mentioned previously found that sexual abuse was associated with all inflammatory markers in nonwhite adults but not white adults in the United States. IL-6 was higher by 36.3% in nonwhite adults exposed to childhood sexual abuse versus 9.89% in white adults exposed to childhood sexual abuse (Pinto Pereira et al., Reference Pinto Pereira, Stein Merkin, Seeman and Power2019). Experiences of racial or ethnic discrimination may constitute an aspect of child victimization or threat perception that is currently understudied. In a study of perceived discrimination, there was a positive association between child-reported discrimination and CRP, as measured by dried blood spots in a group of children with a mean age of 12 y (Goosby, Malone, Richardson, Cheadle, & Williams, Reference Goosby, Malone, Richardson, Cheadle and Williams2015). This was a small study, but emphasizes the need to operationalize, measure, and assess how different forms of child adversity, such as racial discrimination, can contribute to inflammatory outcomes and risk for psychopathology. Thus, specific populations may experience moderating factors such as discrimination in addition to the types of adversity that are typically measured in the context of childhood adversity exposure and inflammation (Goosby et al., Reference Goosby, Malone, Richardson, Cheadle and Williams2015; Pinto Pereira et al., Reference Pinto Pereira, Stein Merkin, Seeman and Power2019; Slopen et al., Reference Slopen, Lewis, Gruenewald, Mujahid, Ryff, Albert and Williams2010).