Adverse childhood experiences and low socioeconomic status with respect to allostatic load in adulthood: A systematic review

Early-life psychosocial stress primes a number of health risk behaviors, and contributes to the development of various mental and somatic disorders in adulthood. It has been reported that adverse childhood experiences (ACEs) and low socioeconomic status (SES) might be associated with allostatic load (AL) in adulthood. In turn, elevated AL index has been found to predict a number of unfavorable health outcomes. Therefore, we aimed to perform a systematic review of studies investigating the association of ACEs and childhood SES with AL in adult populations. Independent online searches covered the publication period up to 20th Jun 2021. A total of 27 studies were included in qualitative synthesis. The majority of eligible studies showed that ACEs (14 out of 19 studies recording ACEs, 73.7%) and low childhood SES (11 out of 12 studies recording childhood SES, 91.7%) are associated with elevated AL in adults. However, several processes were found to mediate or moderate this association. These include educational attainments, social support, health behaviors, adult stress, post-traumatic stress disorder, coping strategies and aging. Moreover, a substantial methodological heterogeneity of approaches to calculating the AL index was observed. Apart from reports from overlapping samples, none of eligible studies used the same set of biomarkers. Findings from this systematic review imply that early-life psychosocial stress might have a lasting impact on biological dysregulations captured by the AL index. Future studies need to explore whether the association between early-life stress and the AL index accounts for the development of specific health outcomes.


Introduction
Adverse childhood experiences (ACEs) can be defined as "any act of commission or omission by a parent or another caregiver that results in harm, potential of harm or threat of harm to a child" (Sideli et al., 2012). This definition captures a number of stressors that are commonly recorded by studies investigating the consequences of ACEs, and include parental loss, emotional neglect and abuse, bullying, physical abuse and sexual maltreatment. Importantly, ACEs have been found to prime a number of health risk behaviors, such as substance use, poor dietary habits, sedentary lifestyle, and mental disorders (Maschi et al., 2013). Moreover, there is evidence that a history of ACEs might predict a number of unfavorable physical health outcomes, such as cardiovascular diseases, musculoskeletal diseases, respiratory diseases and diabetes (Noteboom et al., 2021;Su et al., 2015). The exact mechanisms underlying these observations remain unclear.
One possibility is that ACEs should be analyzed in a broader context of childhood environments operationalized as socioeconomic status (SES) that refers to "the position of an individual at his/her childhood period in an economically and socially stratified society" (Sankar et al., 2019). There are various indices of childhood SES, including, i.e., parental education and income, family structure, household quality and usage of healthcare facilities (Sankar et al., 2019). As similar to ACEs, there is evidence that childhood SES is associated with adult health outcomes, especially with respect to cardiovascular morbidity and mortality as well as all-cause mortality (Cohen et al., 2010). It has been shown that SES may moderate the effects of ACEs on the development of chronic illness in adulthood (Mock and Arai, 2011). However, there are also studies showing that the effects of ACEs on various health outcomes in adulthood remain significant after adjustment for SES (Basu et al., 2017).
Several studies have focused on analyzing specific biological mechanisms to understand how ACEs exert lasting effects on psychological wellbeing and physical health. It has been shown that ACEs are associated with nervous, endocrine and immune system alterations in children and adults. Specifically, a history of ACEs has been shown to correlate with reduced hippocampal volumes as well as abnormalities within the prefrontal cortex and amygdala (Danese and McEwen, 2012). In turn, endocrine and immune alterations emerge as consequences of ACEs. This includes overactivation of the hypothalamic-pituitary-adrenal (HPA) axis (Bernard et al., 2017) with blunted response to psychosocial stress (MacMillan et al., 2009), altered glucose homeostasis (Tosato et al., 2021) and up-regulated immune-inflammatory responses (Baumeister et al., 2016). Moreover, there is evidence that individuals reporting a history of ACEs are at risk of developing obesity (Wiss and Brewerton, 2020). Long-term effects of ACEs can be explained by the fact that these experiences affect maturational processes by acting within sensitive developmental windows (Danese and McEwen, 2012). It is also likely that the effects of early-life stress are the function of specific characteristics of ACEs (timing, duration, severity and multiplicity) and the activity of biological systems that play a buffering or protective role (De Bellis and Zisk, 2014). Given that ACEs exert lasting effects on multiple biological systems, it seems reasonable to approach more comprehensive models that capture a broad range of stress-related biological alterations.
The allostatic load (AL) concept was developed to provide better understanding of various consequences related to accumulated stress exposures (Sterling and Eyer, 1988;McEwen, 1998). One of the important observations central to the AL concept is the notion that psychosocial stress is an inevitable part of human life. The physiological effects of stress processes are aimed at promoting adaption of the human body to changing environments while also maintaining homeostasis. The term "allostasis", introduced by Sterling and Eyer (1988), refers to these physiological effects that are activated to achieve "stability through change". In turn, AL reflects the price paid by the body and brain for perpetual adaptation to environmental challenges (McEwen, 1998;McEwen and Stellar, 1993). According to this concept, higher AL is the consequence of excessive stress or ineffectiveness of adaptive allostatic processes (McEwen, 2000).
Chronic activation of allostatic mechanisms (e.g., HPA axis response) can exert systemic and deleterious effects. McEwen and Seeman (1999) further divided the mechanisms of allostasis into: (1) primary mediators (glucocorticoids and catecholamines); (2) primary effects (cellular processes targeted by primary mediators, e.g., immune response and oxidative stress); (3) secondary outcomes (complex processes that are the consequence of primary effects in specific tissues, measured by waist-to-hip ratio, blood pressure or glycated hemoglobin) and (4) tertiary outcomes (disease outcomes being the consequence of AL).
Based on this concept, it has been proposed to operationalize physiological effects of chronic stress using the AL index that captures the number of dysregulations across various biological systems. Therefore, the AL index is calculated based on alterations of several markers grouped into the following categories: (1) cardiovascular and respiratory parameters, e.g., blood pressure, heart rate and peak air flow; (2) metabolic parameters, e.g., body-mass index (BMI), waist-to-hip ratio (WHR), blood lipids as well as fasting glucose and insulin; (3) neuroendocrine markers, e.g., cortisol, epinephrine, norepinephrine and dehydroepinandrosterone (DHEA), and (4) immune parameters, e.g., Creactive protein (CRP) and cytokines (Juster et al., 2010). Importantly, a recent review revealed that almost half of studies do not include the measures of the HPA axis activity (Johnson et al., 2017). Given that the neuroendocrine system plays a key role in the stress response, it has been highlighted that the AL index should include at least some primary mediators to conform with the AL concept (McEwen and Stellar, 1993;Whelan et al., 2021).
To date, several approaches to calculate the AL index have been developed. Two approaches in particular are frequently selected: (1) the first referring to the distribution of specific parameters (e.g., according to quartile distribution) and (2) the second based on conventional clinical cut-offs. Then, the AL index is calculated as the sum of markers with the levels scored beyond quartile-based or clinical cut-offs. The AL index has specific ranges depending on the number of markers being assessed, where higher values indicate greater physiological dysregulations and lower values indicate better adaptation to stress. Importantly, the AL index shares some similarities with other composite constructs that have been developed to predict various physical health outcomes, such as the metabolic syndrome or the Framingham score (Gallo et al., 2014). Indeed, a diagnosis of metabolic syndrome can be established based on the measurements of waist circumference, blood pressure as well as the levels of glucose, triglycerides and high-density lipoproteins (HDL). Similarly, the levels of HDL and systolic blood pressure are recorded to assess the Framingham risk score. All of these parameters are often included to calculate the AL index. However, the percentage of variance in health outcomes predicted by the AL index is higher than that of the metabolic syndrome (Karlamangla et al., 2002;Mattei et al., 2010). Moreover, the AL index has been found to predict functional and cognitive decline (Booth et al., 2015;Karlamangla et al., 2002), frailty (Ding et al., 2017), cardiovascular diseases (Gillespie et al., 2019;Seeman et al., 2001) and all-cause mortality (Castagné et al., 2018). Emerging evidence also indicates that the AL index might hold promise in cancer research and practice. Indeed, significant associations of the AL index with cancer-related stress, post-traumatic growth and resilience in patients with cancer, positive cancer history, tumor pathology as well as cancer-related mortality have been reported (Mathew et al., 2020).
Taken together, these observations indicate that the association between ACEs and various health outcomes needs to be investigated in the context of social environments. Some studies have investigated whether ACEs and childhood SES are related to the AL index. While much of the early AL studies focused on older adults, less is known about the association between ACEs and the AL index in adulthood. A recent systematic review revealed that low SES is related to elevated AL indices (Johnson et al., 2017). However, this systematic review included studies that analyzed the impact of SES on the AL index across all age groups. In turn, a systematic review of findings in this emerging field has not been performed so far. Therefore, in this study, we aimed to synthesize existing evidence on the association between ACEs, childhood SES and AL indices in adults. Specifically, we aimed to address the following questions: (1) are ACEs and low childhood SES associated with the AL index in adults?; (2) what are the potential mediators and moderators of the association between ACEs, low childhood SES and AL in adulthood?; (3) what is the specific role of psychological mechanisms and psychopathology in the association of ACEs and low childhood SES with AL in adults? and (4) do ACEs and low childhood SES operate through the same pathways leading to AL in adulthood?

Search strategy
Two independent reviewers were involved in online searches (A.P. and B.S.) and used the following combination of keywords: "child* OR youth* OR adolescen* OR teen* OR pediatr* OR paediatr* AND abuse OR neglect OR trauma OR maltreat* OR stress OR socioeconomic AND allosta* ". A total of five databases were searched: MEDLINE, ERIC, CINAHL Complete, International Pharmaceutical Abstracts as well as the Academic Search Ultimate and the Health Source: Nursing/Academic Edition from their inception until 20th Jun 2021. Additionally, references of eligible publication records were analyzed. Discrepancies were resolved through discussion with the third reviewer (B.M.). Online searches were performed in agreement with the PRISMA guidelines (Moher et al., 2009).

Eligibility criteria
The inclusion criteria were as follows: (1) studies reporting the AL index measured based on the levels of specific biomarkers in adults (≥ 18 years of age); (2) studies recording ACEs and/or SES under the age of 18 years and (3) case-control, cross-sectional and longitudinal studies. For the purpose of this review, ACEs were defined as any adversities, e. g., parental loss, emotional abuse, emotional neglect, bullying, physical abuse and sexual maltreatment reported below the age of 18 years. In turn, childhood SES was defined as any measure of social position under the age of 18 years, conceptualized using any of the following variables: (1) parental education; (2) parental income; (3) family structure, (4) household quality and (5) usage of healthcare facilities. The following publication records were excluded: (1) studies that did not calculate the AL index based on the levels of specific biomarkers; (2) studies limited to assessment of the AL index under the age of 18 years; (3) psychosocial stressors and/or SES were not recorded under the age of 18 years; (4) animal model studies; (5) non-original studies (e.g., reviews, commentaries and editorials) and (6) non-English language publications. Publication records that were based on overlapping samples were not excluded due to a variety of specific hypotheses being tested.

Data extraction and quality assessment
The following data were extracted from eligible publications: (1) study design; (2) country where the study was performed (3) age of participants (mean and SD); (4) sex; (5) information on the biomarkers used to calculate the AL index; (6) method of calculating the AL index; (7) methods of recording ACEs and SES; (8) information about covariates included in data analysis; (9) tested mediators and/or moderators and (10) main findings. Mediation was defined as the mechanism or process that underlies the association between ACEs or childhood SES (independent variable) and the AL index through the effect of third variable (mediator). Full mediation was considered to occur in two cases: (1) inclusion of the mediator had dropped the association of ACEs or childhood SES with the AL index to zero and (2) the association of ACEs or childhood SES with the AL index appeared to be significant after inclusion of the mediator. In turn, partial mediation was considered to occur if the reduction in variance of the AL index explained by ACEs or childhood SES had been significant after inclusion of the third variable (variable); however; the direct association of ACEs or childhood SES had also remained significant. Finally, moderation was considered significant if the association between ACEs or childhood SES had been dependent on the effect of the third variable (moderator).
Quality assessment was performed using the tools developed by the US National Heart, Lung and Blood Institute (https://www.nhlbi.nih. gov/health-topics/study-quality-assessment-tools). Separate checklists were used for observational cohort and cross-sectional studies (14 items) as well as case-control studies (12 items). The checklists are provided in Supplementary Table 1. Each item includes a dichotomous (yes/no) question about study quality. One point was scored for each "yes" response and zero points were scored for each "no" response. Due to differences in the number of items used to assess quality of cohort or cross-sectional studies and case-controls studies, quality scores were expressed as the percentage of maximum quality score that can be assigned to specific study. Higher percentage indicates better quality.

Associations between ACEs and the AL index
The majority of studies recording ACEs (14 out of 19 studies, 73.7%) found that they are associated with elevated AL in adulthood. Notably, four out of five longitudinal cohort studies reported a significant association. This association was reported for childhood abuse (physical, emotional and sexual) and emotional neglect, harsh parenting and parental antipathy. The potential effect of multiplicity of ACEs was suggested by Wallace et al. (2020), who showed that adults living with HIV and a history of at least 4 different ACEs show elevated AL. There was substantial variability in the use of specific co-variates (Table 1). These variables included: age (14 studies, 73.7%), sex (12 studies, 63.2%), race/ethnicity (7 studies, 36.8%), adult SES characteristics (10 studies, 52.6%), childhood SES characteristics (2 studies, 10.5%), prenatal and perinatal factors (1 study, 5.3%), mother's BMI (1 study, 5.3%), health behaviors (6 studies, 31.6%), health problems (2 studies, 10.5%), adult stress (3 studies, 15.8%), use of medications and/or adherence to treatment (4 studies, 21.1%), oral contraceptive use (1 study, 5.3%) and time of blood sampling (1 study, 5.3%).
The quality of studies that did not report significant associations (Beckie et al., 2016;Berg et al., 2017a;Evans and De France, 2021;O'Shields and Gibbs, 2021;Thayer et al., 2017) varied between 35.7% and 78.6% (50.0-78.6% in case of studies reporting significant associations). Among studies that did not report significant associations, there were four cross-sectional studies and one cohort study. Among them, one study did not use any covariates due to small sample size (Beckie et al., 2016). In turn, the study by Evans and De France (2021) operationalized limited number of ACEs (violence, separation from family and family turmoil) together with some measures of SES (crowding, noise and housing problems) as the cumulative risk exposure. It is also important to note that the study by Berg et al. (2017a) was conceptually different as the authors did not test direct associations between ACEs and the AL index. Similarly, mediation by other variables recorded in this study was not tested with respect to the association between ACEs and AL. Moreover, although the authors measured the levels of CRP and IL-6, they did not include these markers in the AL index. They found that a history of ACEs might increase the level of depressive symptoms that are associated with the level of IL-6. In turn, the levels of IL-6 were associated with the AL index.
* Arrows refer to direction of association: ↑ -significantly higher AL index in subjects exposed to ACEs and/or low childhood SES; ↔ -not significant association of exposure to ACEs and/or childhood SES with the AL index.
B. Misiak et al. studies, 91.7%) have clearly indicated that low childhood SES is associated with elevated AL. All longitudinal studies reported significant associations between childhood SES and elevated AL in adults. Studies measuring childhood SES used the following covariates: age (7 studies, 58.3%), sex (8 studies, 66.7%), race/ethnicity (5 studies, 41.7%), adult SES characteristics (6 studies, 50.0%), parental status characteristics (1 study, 8.3%), health behaviors (4 studies, 33.3%), health problems (1 study, 8.3%), adult stress and/or measures of psychological well-being (3 studies, 25.0%), and time or fasting condition during blood sampling (25.0% studies, 5.3%). One cross-sectional study revealed that social class in childhood (father's occupation at the age of 11 years) was not associated with AL after controlling for educational attainment, attained social class and health behaviors in adulthood (Gale et al., 2016). Quality of studies reporting significant associations ranged between 50.0% and 78.6%. The quality of the study by Gale et al. (2016) was 57.1%.

Joint contributions of ACEs and childhood SES to the AL index
Only three studies jointly investigated the effects of ACEs and childhood SES on AL (Dich et al., 2015;Evans and De France, 2021;Friedman et al., 2015). Importantly, these studies did not investigate whether both exposures interact in the pathway to AL. Dich et al. (2015) found that the impact of family adversity and the combination of family adversity with economic hardship in childhood have greater impact on the AL index than economic hardship alone. In turn, Evans and De France (2021) used a prospective cohort study to investigate the impact of poverty during childhood on AL in adults. They found that this association is significant but is not mediated by other experiences conceptualized as the cumulative risk exposure (crowding, noise, housing problems, violence, separation from family and family turmoil). Finally, Friedman et al. (2015) revealed that low childhood SES (measured by financial situation and parental education) and physical abuse, but not parental divorce and death, are associated with AL in adults. However, the effect of physical abuse on AL was not significant after controlling for adult educational attainment, social relationships, and health behaviors.

Insights into mechanisms: focus on mediators and moderators
Some studies also provided insights into potential mechanisms mediating or moderating the association between ACEs and AL. These studies indicate that social environments may play an important role in this association. For instance, Currie et al. (2019) found that the association between childhood discrimination and AL is significant only in participants representing low cultural continuity groups. Cultural continuity can be defined as the transmission of meanings and values characteristic for specific societies through time and generations (Chandler et al., 1998). In this study, it was measured as the level of cultural identity, engagement and connectedness. In agreement with these observations, social support across the lifespan has been found to partially mediate the association between ACEs and AL (Horan and Widom, 2015a). Similar findings were obtained by Piotrowski et al. (2020), who found that engagement in social interactions as the way of coping suppresses the association between a history of sexual abuse and AL in participants with schizophrenia spectrum disorders. Moreover, Friedman et al. (2015) revealed that a history of childhood physical abuse is not associated with AL after controlling for adult educational attainment, social relationships and health behaviors. Health behaviors, education level and wealth were also found to partially mediate the association between ACEs and AL in the study by Solís et al. (2015). However, Wallace et al. (2020) found that the use of alcohol and cigarette smoking do not mediate the association between a history of ACEs and AL among people living with HIV. Finally, one study found that stress and SES in adulthood also partially mediate this association (Turner et al., 2016). Similarly, economic hardship in adulthood has been found to amplify the effect of ACEs on AL (Berg et al., 2017b).
Two studies demonstrated that educational attainments partially mediate the association between childhood SES and AL in adults (Friedman et al., 2015;Graves and Nowakowski, 2017). Importantly, adult SES and stress have been found to mediate the association of childhood SES (full mediation), ACEs (partial mediation) and AL in adults (Turner et al., 2016). Finally, one study revealed that coping strategies may impact the association between childhood SES and AL (Chen et al., 2012). More specifically, the authors focused on the role of 'shift-and-persist' strategies that rely on positive reinterpretation of stressors (shifting), while persisting with a focus on the future. They found that adults who had grown up in low SES, and reported engagement in this coping strategy, had significantly lower AL. However, no significant effects of shift-and-persist strategies on AL were found in those who had grown up in high SES.
Five studies investigated the association between ACEs and AL in the context of psychopathological symptoms (Piotrowski et al., 2020;O'Shields and Gibbs, 2021;Scheuer et al., 2018;Thayer et al., 2017;Widom et al., 2015). Scheuer et al. (2018) demonstrated that AL partially mediates the association between childhood physical abuse and depressive symptoms. This effect appeared to be moderated by age and was not significant in older participants (55 -81 years). However, these findings were not confirmed by another study (O'Shields and Gibbs, 2021). O'Shields and Gibbs (2021) observed that AL might be associated with higher levels of depressive symptoms only in females. Importantly, this study did not find significant association between ACEs and AL. Thayer et al. (2017) investigated this association in the context of PTSD. The authors found that post-traumatic stress disorder (PTSD) fully mediates the association between ACEs and AL. In turn, Piotrowski et al. (2020), apart from the association with stress coping, found that parental antipathy contributes to elevated AL both in patients with schizophrenia spectrum disorders and healthy controls. However, the impact of sexual abuse on AL was significant in patients with schizophrenia spectrum disorders but not in healthy controls. Finally, Widom et al. (2015) found that the association between ACEs and AL remains significant after controlling for potential mediators of internalizing and externalizing problems in adolescence.

Discussion
Findings from this systematic review indicate that growing up in low SES and having a history of ACEs are associated with elevated AL in adulthood. Significant positive associations were reported by the vast majority of longitudinal cohort studies. Moreover, we did not find considerable differences between studies reporting positive findings and those reporting negative findings with respect to their quality. Only three studies with heterogenous methodologies jointly investigated the effects of both exposures. One of these studies (Evans and De France, 2021) investigated whether childhood SES and ACEs interact in pathways leading to AL. However, this study did not demonstrate that cumulative exposure to social disadvantage and certain ACEs mediate the effect of poverty and AL.
Findings of this systematic review indicate that the association of childhood SES and ACEs with AL is mediated or moderated by a number of critical processes. Due to similarity of mediators and moderators being tested by eligible studies, it is unlikely to assume that both exposures contribute to AL through different pathways (see Fig. 3 for summary of findings). The following mediators were found to impact the association of childhood SES and ACEs with AL: low social support, health-related behaviors, adult stress and PTSD symptoms. In turn, the use of certain coping strategies might counteract the effect of early-life stress on AL. There is also some evidence that the association between early-life stress may not be significant for older adults. This observation might be explained by the fact that older people tend to accumulate greater coping strategies (Sütterlin et al., 2012). Moreover, it has been shown that age-related increases in AL reach a plateau between 60 and 90 years of age (Crimmins et al., 2006).
From a public health perspective, findings might hold some promise for developing specific interventions for socially disadvantaged children and those exposed to ACEs. To date, several interventions have been proposed to reduce health inequalities with some of them (e.g., provision of suitable housing, residential care, emergency support and early crisis interventions) dedicated to children with low SES or those exposed to ACEs (MacGuire et al., 2020). However, little is known about their efficacy in reducing AL. A recent scoping review (Rosemberg et al., 2020), based on systematic database searches, identified only six studies with the AL index as an outcome. These studies tested the efficacy of the following interventions: (1) osteopathic manipulative treatment (Nuño et al., 2019); (2) second-generation antipsychotics (Berger et al., 2018); (3) cognitive behavioral therapy vs. Tai Chi Chih vs. sleep seminar (Carroll et al., 2015); (4) whole food diet (Soltani et al., 2018); (5) mentor-based supportive expressive program (Ye et al., 2017) and (6) federal nutrition assistance program (McClain et al., 2018). Significant decrease in AL were found for osteopathic manipulative treatment in graduate students, antipsychotic treatment in patients with first-episode psychosis, cognitive behavioral therapy and Thai Chi Chih in older patients with sleep disorders and mentor-based supportive expressive program for patients with metastatic cancer.
Most studies included in our systematic review reported partial mediation. This observation suggests that other processes, not tested by eligible studies, may also explain how early-life stress impacts AL in adults. Certain antecedents and direct consequences of ACEs related to, i.e., parental care, relationships with other relatives and friends as well as delivery of professional support following ACEs, can further impact, at least theoretically, the association between ACEs and the AL index. Another possibility is that this association may simply reflect the effect of subsequent stressors. Indeed, one of the studies included in this systematic review demonstrated that adult stress mediates the association between ACEs and AL (Turner et al., 2016). Moreover, ACEs may also give rise to various psychological processes that are related to unfavorable mental health outcomes and biological responses to stress. These include emotional dysregulation, negative schemata (i.e., mental constructs of meaning) and cognitive biases (Bloomfield et al., 2021;Gawęda et al., 2018). Although most of these processes have not been investigated with respect to AL, there is evidence that they might be related to single parameters captured by the AL index. Regarding cognitive biases, it has been found that increases in cortisol in the context of stress is related to decreased level of attention for threat bias (McHugh et al., 2010). In turn, experimental induction of emotions might increase cortisol levels (Mikkelsen et al., 2021) and could be the focus of future AL research.
The present systematic review provides some evidence on the role of psychopathology in the association between early-life stress and AL. In particular, some studies suggest that AL is associated with depressive symptoms. However, one study showed that this association is significant only in females (O'Shields and Gibbs, 2021). In one study, AL was found to mediate the association between ACEs and AL indices (Scheuer et al., 2018). Notably, sex differences have been widely reported by studies investigating AL. A recent systematic review demonstrated that men tend to have higher AL index than women in studies that detect sex differences (Kerr et al., 2020). However, women show gender-specific variation for numerous socio-cultural factors such as age, race/ethnicity, adversities, social support, and health behaviors that influence associations between AL and mental health.
In one study, AL was found to mediate the association between ACEs and depressive symptoms (Scheuer et al., 2018). Interestingly, the association among AL and depressive symptoms has also been reported in patients with psychosis, suggesting that this relationship might be independent of current diagnostic classifications . One study included in our systematic review showed that a history of childhood sexual abuse is associated with elevated AL in patients with schizophrenia spectrum disorders but not in healthy controls (Piotrowski et al., 2020). In turn, parental antipathy was associated with AL in both groups of participants. One study also demonstrated that PTSD fully mediates the association between ACEs and AL (Thayer et al., 2017). These observations suggest that the impact of ACEs on AL might be greater for those who develop psychopathology. However, cross-sectional or case-control design of studies investigating Fig. 2. Frequency of specific biomarkers used to calculate AL indices across eligible studies. Abbreviations: BMI, body mass index; CRP, C-reactive protein; DBP, diastolic blood pressure; DHEA, dehydroepiandrosterone; DHEA-S, dehydroepiandrosterone sulfate; HbA1c, glycated hemoglobin; HDL, high-density lipoproteins; HOMA-IR, the homeostatic model assessment for insulin resistance; HRV, heart rate variability; ICAM-1, intercellular adhesion molecule-1; IL-6, interleukin-6; LHFSP, low and high frequency spectral power; LDL, low-density lipoproteins; PEF, peak expiratory flow; RMSSD, the root mean square of successive differences; SBP, systolic blood pressure; SDRR, standard deviation of R-R intervals; TC, total cholesterol; TNF-α, tumor necrosis factor-α. psychopathology do not allow us to conclude about causality.
Although eligible studies have almost consistently reported that low childhood SES and ACEs are related to elevated AL indices, there was a substantial methodological heterogeneity of approaches to calculating the AL index. Apart from reports from overlapping samples, none of eligible studies used the same set of biomarkers. It is also important to note that cardiovascular and metabolic markers as well as CRP were more often included to measure AL than were primary mediators, such as cortisol or catecholamines. Additionally, certain studies did not include any primary mediators. Similarly, some inconsistencies were observed with respect to the methods of calculating AL indices, with sexspecific percentiles being rarely determined. This methodological heterogeneity does not allow us to perform a qualitative synthesis of evidence in the field. In this regard, a consensus statement on calculating AL indices is needed soon.
At least theoretically, obtained results may differ due to differences in the methods of recording childhood SES and ACEs. Most importantly, retrospective measures may be a source of recall bias. In turn, documenting the reports of ACEs by interviewing third parties is not always possible and may also be biased. More specifically, family members might be directly or indirectly involved in ACEs. Additionally, victims do not always disclose a history of ACEs to third parties due to feelings of guilt, shame and humiliation (Negrao et al., 2005;Stuewig and McCloskey, 2005). Results from a meta-analysis assessing the association between parameters of HPA axis activity (wake-up cortisol, cortisol awakening response and diurnal cortisol slope) and a history of maltreatment indicate that the way of obtaining reports of maltreatment might be important (Bernard et al., 2017). Indeed, the authors found that maltreatment is associated with significantly higher awakening cortisol levels only in studies of agency-referred samples. Moreover, effect size estimates for awakening cortisol levels were significantly higher for studies of agency-referred samples than for studies that had recorded maltreatment based on self-reports.
Nevertheless, some studies indicate that self-reports of ACEs are reliable and stable over time (Bernstein et al., 1994;Fisher et al., 2011;Pinto et al., 2014). Another important factor, not thoroughly addressed by studies included in this systematic review, is the lapse of time between exposure and outcome. On the basis of a meta-analysis, Miller et al. (2007) found that the HPA axis activity is elevated at stressor onset but decreases as time passes. In agreement with results of this meta-analysis, Raymond et al. (2021) demonstrated that adults first exposed to ACEs between the ages of 3 and 7 years (the critical period of amygdala development) showed greater cortisol awakening response and lower cortisol reactivity compared to those first exposed before 3 or after 7 years of age. One of eligible studies addressed this point by showing that age moderates the mediating effect of the AL index on the association between childhood abuse and depressive symptoms (Scheuer et al., 2018). This effect was not significant for older participants (at the age of 55 -81 years). Finally, other characteristics of ACEs, including duration, severity and multiplicity might also be important from a neurodevelopmental perspective.
The present systematic review has some limitations that need to be highlighted. First, methodological heterogeneity of eligible studies did not allow us to perform quantitative synthesis of the evidence. Similarly, a variety of approaches limit the possibility to conclude anything regarding the strength of the associations between ACEs, childhood SES and AL. Another limitation is related to heterogeneity of potential mechanisms and covariates that were tested by eligible studies. This does not allow us to establish consistency of findings with respect to potential mediators or moderators at this time. It should also be noted that our systematic review is characterized by a predominance of crosssectional and case-control studies and so causal associations should be interpreted with caution.
In conclusion, the present findings suggest that early psychosocial stress might have a lasting impact on biological dysregulations captured by the AL concept and its indexing. Most likely this association is mediated and moderated by several processes, such as educational attainment, social support, health behaviors, adult stress and PTSD symptoms, coping strategies and aging. However, it remains unknown whether the relationship between early-life stress and AL is relevant for the development of specific disease outcomes. Addressing this point would provide grounds for developing specific interventions that aim to prevent the development of unfavorable, stress-related outcomes by targeting modifiable risk factors. However, studies in this field need to take into consideration various shortcomings recognized in our review. Most importantly, a unified approach to indexing AL is needed to provide better generalization of findings and their functional significance. More comprehensive approaches to characterization of early-life stress Fig. 3. Overview of main findings showing mediators and moderators for the association of childhood socioeconomic status (SES) and adverse childhood experiences (ACEs) with allostatic load (AL). Abbreviations: BMI, body-mass index; CRP, C-reactive protein; DBP, diastolic blood pressure; DHEA, dehydroepiandrosterone; ECG, electrocardiography; HR, heart rate; IL-6, interleukin-6; SBP, systolic blood pressure; TNF-α, tumor necrosis factor-α, WHR, waist-to-hip ratio.
while also scrutinizing a thorough definition of ACEs with SES is also warranted. Finally, addressing a broader trajectory of cascading consequences from early-life stress through AL and disease outcomes over time is required.