Mid-life reversibility of early-established biobehavioral risk factors: a research agenda

Abstract

Objectives:

Epidemiological evidence links exposure to early life adversities --such as childhood maltreatment-- with impaired health and wellbeing in adulthood. Since these effects are usually unrecognized or untreated in childhood, preventive and remediating interventions in adults are needed. Our focus on adulthood prompted three major questions. First, does our increased understanding of mechanisms accounting for the long-term effects of early life adversities (ELAs) help delineate underlying dimensions that underscore key similarities and differences among these adversities? Second, can adults accurately report on adversities they experienced in childhood? Third, can we identify malleable risk processes in adulthood that might be targets for preventive intervention?

Methods:

Supported by the National Institute on Aging, the UK Economic and Social and the Biotechnology and Biological Sciences Research Councils, a network of researchers in human and animal development addressed these questions through meetings and literature review.

Results:

A small number of dimensions may adequately distinguish among a range of cooccurring childhood adversities. Widely-used adult ascertainments of childhood adversity are poorly related to prospective ascertainment. Strategies for preventive interventions should be aimed both at adults who were actually exposed to adversity as well as those who recall adversity, but the targeted risk processes may be different.

Discussion:

Now is an opportune time to support research on adult interventions based on unfolding research on critical periods of sensitivity to adversity in fetal and child development, on improved understanding of risk mechanisms that may persist across the life span and new insights on enhancing neuroplasticity in adults.

Early life adversities (ELAs) come in many forms, including poverty, life in depriving orphanages, nutritional deficiency in utero and beyond, deficiencies in parental nurturing, various forms of maltreatment, and loss of a parent or other attachment figure. A more recent line of research explores the distinctive and long-term effects of racism early in development (e.g., Chae et al., 2018; Pachter & Coll, 2009). Human and animal research has established that ELAs engender a liability for ill health (i.e., they are related to later outcomes, though there is not yet clear evidence of a causal link) that persists through mid-life. While there are interventions for children that offset some of this risk, there are fewer interventions to counter this liability in adults. A coordinated effort to develop interventions for adults is urgent. Future research will be aided by clearer conceptualization and measurement of ELA as well as delineation of useful strategies for targeting and ameliorating risk mechanisms linking ELA to adult health. We briefly summarize evidence underscoring the importance of ELA in adult health. Then we review several approaches to conceptualizing and measuring ELA and discuss the relative advantages and disadvantages of retrospective and prospective ascertainment of ELA. Following, we outline two broad strategies for further research on useful interventions. In one of these, recalled ELA helps to anticipate malignant course and resistance to treatment in established illness, suggesting the existence of clinical subtypes that may benefit from targeted intervention. In a second, information on ELA is useful for identifying disease liability and planning prevention efforts prior to disease onset. Finally, we address the practical challenge of designing prevention trials for health outcomes that only become apparent years or decades in the future and consider recent research that may provide valid intermediary outcomes for rigorous efficacy testing.

The systematic study of the impact of ELA on adult mental and physical health gained momentum a half century ago (e.g., Berkman, 1971; Kiernan, Colley, Douglas, & Reid, 1976). Throughout, the definition of ELA has been broad and has ranged from general indices of adversity such as social class to more specific exposures of fetuses or children to malnutrition and maltreatment from caretakers. As we will note below, there is recent progress in more coherent and empirical based ascertainment of ELA.

Two landmark studies have galvanized this field. The first was the discovery of links between ELA and health at mid-life that came from fastidious records of birth weights compiled by midwives in Hertfordshire, England, to assure proper childhood nutrition after a large number of men failed eligibility to fight in the Boer War (Barker, 2003). Over 60 years later, David Barker and colleagues used these records to report an association between low birth weight and deaths from coronary heart disease decades later. Numerous follow-up studies confirmed this startling observation (e.g., Godfrey, Inskip, & Hanson, 2011), and experimental studies with animals using nutritional challenges (Fleming et al., 2018) and fetal growth restriction (e.g., Schreuder, Fodor, van Wijk, & Delemarre-van de Waal, 2006) suggested that these assocations between diminished fetal growth and cardiac disease were causal and documented mechanisms accounting for them. In a second notable study, Felitti and colleagues investigated the relationship of postnatal adverse childhood experiences (ACEs) recalled by adults, including the physical and psychological abuse they experienced as children. They observed substantial associations between the number of ACEs recalled and the prevalence of risky health behaviors, such as drug abuse, and of cardiometabolic diseases and cancer in mid-life (Felitti et al., 1998). The Barker studies were a model of clear conceptualization of an ELA and the deft integration of human and animal prospective studies to establish causal pathways via risk mechanisms linking ELA and adult disease. The Felitti study yielded an easily administered questionnaire, suitable for large samples, that provide replicable results linking recalled adversity with current and future health.

More recently, several prospective studies of the long-term effects of ELA have assessed childhood exposures using multiple contemporaneous sources including agency records, in order to identify children exposed to ELA and compare their health trajectories to those of children without known adversity. Some investigators have followed such children to mid-life. These studies have, for example, reported associations between several types of ELA, including severe maltreatment, with adult maladaptive social, behavioral, and health outcomes, especially: 1) emerging risk factors for a range of medical illnesses, including chronic inflammation, elevated hemoglobin A1c, and respiratory compromise (e.g., Baldwin et al., 2018; Danese et al., 2009; Lacey, Kumari, & McMunn, 2013; Takizawa, Danese, Maughan, & Arseneault, 2015; Widom, Czaja, Bentley, & Johnson, 2012)¹; 2) risky behaviors, such as prostitution and illicit drug use (Widom & Kuhns, 1996; Wilson & Widom, 2009); 3) impaired economic well-being (Currie & Widom, 2010); and 4) abusive parenting that may perpetuate adversity across generations (Widom, Czaja, & DuMont, 2015). Animal studies that model adverse social exposures in controlled experiments, using strategies akin to those on fetal growth restriction, provide evidence that supports many of these human findings (e.g., Conti et al., 2012; Ganguly & Brenhouse, 2015; Maestripieri, Lindell, & Higley, 2007).

Several successful interventions have been developed for children and their families to mitigate the impact of ELAs on relatively short-term, mental health outcomes (e.g., Toth, Gravener-Davis, Guild, & Cicchetti, 2013). Some, with long-term follow-ups, demonstrate impacts on adult cardiometabolic health (Campbell et al., 2014). However, even in those very few countries that carefully monitor childhood adversities and, also, have resources to intervene to prevent its long-term consequences, there will be many individuals who are neither recognized as victims of adversity nor treated early in life to reduce the risk of long-term adverse outcomes. They will arrive in adulthood with significant disease liabilities: they may have had no conspicuous mental or physical disorder but, as a consequence of early adversity, have a high liability for serious illness; or they may have had mental or physical disability in the past and are at higher risk for relapses or new disorders; or, currently ill, may be treatment resistant as a consequence of early adversity. We have, at present, no evidence-based interventions availableto reverse or compensate for these risks nor for enhancing the potential for these indivduals to achieve good health in older age.

The US National Institute on Aging (NIA) and the UK Economic and Social Research Council (ESRC) and Biotechnology and Biological Sciences Research Council (BBSRC) recognized the need to explore the potential for such preventive interventions because: 1) ELAs are highly prevalent (Wildeman et al., 2014)—as high as 20% prevalence of maltreatment in some US subpopulations (using very conservative ascertainment based only on records of child protective services); 2) the magnitude of associations between ELAs and adult health is substantial; 3) adversities, as noted, are associated with a broad range of impairments in adult well-being and health, and 4) emerging evidence suggests that many of the observed associations may be causal. Accordingly, the NIA and ESRC/BBSRC supported a group of senior scientists in the Interdisciplinary Network on Early Adversity and Later Life Reversibility. Below, we describe a series of next steps in research—recommended by the Network--to advance this agenda, beginning with evidence-based consensus on clearest conceptualization and best ascertainment of ELAs.

Conceptualizing and measuring ELA

A major question facing researchers is what adversities they should ascertain and what methods of ascertainment are most reliable and valid. To address this and related questions the Network identified seven major research questions that are as crucial to the next generation of research.

1). Do all early adversities have approximately the same impact on adult health and can they be cumulated into an aggregate adversity score?

In their influential work, Felitti and colleagues asked adults about a number of adverse exposures and simply added these up, weighting each item equally. Their underlying assumption was that different adversities have similar effects, and that effects on a broad range of health outcomes are cumulative: the more you recall, the more likely you are to fall ill (Felitti et al., 1998). Felitti’s ACEs questionnaire has had reliable contemporaneous associations with several medical disorders and has also predicted major health outcomes later in life including all-cause mortality over a ten-year follow-up period. There was a clear threshold effect: only those with six or more recalled adversities showed a notable effect (Brown et al., 2009). Recent work has confirmed the utility of this cumulative approach in studies of the association of ELA with mental health outcomes; these studies include a mixture of informant and self-reports of ELA as well as prospectively measured maltreatment collected contemporaneously in childhood (Cecil, Viding, Fearon, Glaser, & McCrory, 2017; Vachon, Krueger, Rogosch, & Cicchetti, 2015).

This line of investigation often suggests that there is little specificity of effects of many ELAs: very different forms of adversity may anticipate similar distal outcomes. However, comparisons across types of adversity are difficult because of wide variation among children in the severity or chronicity or timing of the adversities they experience (e.g., Jaffee & Maikovich-Fong, 2011). Thus, a particular form of adversity may be more firmly linked to child health in any given sample because of its severity, chronicity, or timing rather than because it poses a unique challenge to the child’s development. Moreover, conventional categorizations of ELA—such as the distinctions between parental sexual abuse and parental neglect-- may lead to similar health outcomes but by very different pathways.

2). Can ELAs be grouped according to their distinctive effects on risk mechanisms?

Research now explores whether distinct dimensions of adversity, dimensions that cut across the more conventional categorizations, have differential impacts on the brain, biology, and behavior, and thus may lead to health outcomes through unique developmental pathways. For example, McLaughlin, Sheridan, and Lambert (2014) have proposed two. First, deprivation – the lack of social and cognitive stimulation important for normal brain development -- is conspicuous in children in extended care in poorly run institutions. McLaughlin and colleagues suggest that deprivation has consequences for development of the cortex in frontal and temporal lobes and is associated with deficits in language and cognitive function. Second, threat is a central feature of physical and sexual abuse. Mediated through activation of the HPA axis, studies of animals exposed to early life threat reveal anatomical and functional changes in hippocampus, amygdala, and ventromedial prefrontal context with impaired fear extinction learning (McLaughlin et al., 2014). In humans, heightened amygdala activity to threatening stimuli has been demonstrated in children exposed to family violence, suggesting a distinct neural pathway leading to their heightened threat sensitivity (McCrory, De Brito, &Viding, 2011).

Distinguishing between deprivation and threat and their related psychological and physiological effects holds potential for elucidating differential pathways through which exposures become biologically embedded, leading to variations in later health outcomes. For example, Geoffroy and colleagues (Geoffroy, Pinto Pereira, Li, & Power, 2016) found that childhood neglect related to both cognitive and mental health across the life-course; while childhood abuse was also related to mental health, it was mostly unrelated to cognition. This work is consistent with recent research by Sheridan, McLaughlin, and colleagues (2018) that extends our knowledge of differential pathways. For example, in a longitudinal study, both deprivation and threat measured at age 5 and 6 anticipated externalizing behavior 10 years later. However, verbal deficits in these children mediated this effect only in the deprived group (Miller et al., 2018). The effect of deprivation on verbal abilities may reflect in part, its impact on processes of associative learning which, in turn, reflects. impairment in fronto-striatal circuits. In fact, these learning deficits were normalized by randomized placement of institutionalized children into foster care in the Bucharest Early Intervention Project. This normalization mediated, in part, the impact of the intervention on these children’s mental health and peer relationships (Sheridan et al., 2018) providing experimental support for the causal role of this pathway in child adjustment. In contrast, children’s exposure to threat, but not deprivation, showed evidence of accelerated aging using DNA methylation patterns (“epigenetic age”) and advanced pubertal stage, both risk factors for a broad range of health outcomes (Sumner, Colich, Uddin, Armstrong, & McLaughlin, 2018).

Although at an early stage, this work offers a novel framework for understanding the long-term effects of ELA on adult health. First, it characterizes adversity through a growing understanding of how children process adverse environments Second it shifts attention from long term outcomes to pathways by which those outcomes are achieved; this offers very specific targets for preventive interventions. Third, through its focus—in part-on neural mechanisms conserved across primate species—it enhances the scientific rapport between research on ELA in human and non-human subjects.

These results stimulate a broad research agenda. First, they prompt independent replication and extension in other laboratories and research groups. Second, other dimensions of adversity may equal threat and deprivation in their impact on child and adult health. For example, high levels of unpredictability in the child’s environment may lead to distinctive maladaptive pathways across development: these include erratic parental behavior, high turnover in residential location, frequent parental co-habitation, or unpredictability of parents’ employment status. Unpredictability has been modeled in animals and is associated with impaired development in human studies (e.g., Baram et al., 2012; Simpson, Griskevicius, Kuo, Sung, & Collins, 2012). Loss through death or separation from an attachment figure--particularly from parents--is an ordinal dimension of ELA that may have distinct effects and, importantly, has been associated with a range of psychiatric disorders independent of genetic effects (Kendler, Neale, Kessler, Heath, & Eaves, 1992). Third, the same strategy can be used to map psychosocial pathways from maltreatment to health outcomes. Among the most promising are the impact of maltreatment on children’s attachment security (Cicchetti & Toth, 2016). Indeed, insecure attachment in these children is a specific target of child-parent psychotherapy that enhances secure attachment, with consequent improvement in children’s conduct problems (Stronach, Toth, Rogosch, & Cicchetti, 2013). This treatment also normalizes cortisol regulation in children (Cicchetti, Rogosch, Toth, & Sturge-Apple, 2011), suggesting promising preventive effects on broad areas of child health.

Research is urgently needed to improve measurement of each of these proposed dimensions to test whether they clarify mechanisms linking childhood exposures to adult health, and to ascertain whether they have distinctive effects on long-term outcomes or lead to similar outcomes via different causal pathways.

Research on poverty provides a unique challenge for both conceptualization and intervention. On the one hand, poverty is a context in which children face many severe adversities and some show alarming deficits in cognitive and social development. Thus, understanding its effects can usefully inform regional, national, and international remedial policies. On the other hand, economic privation is highly associated with many specific ELAs, many of them already reviewed here. Current research has explored whether poverty has a distinct effect, immediate or delayed, on children’s health, independent of other dimensions with which it is often associated such as maltreatment. Support for the causal effects of poverty comes from a natural experiment (the opening of a casino on a Native American reservation) and a planned experiment (voucher support to poor families allowing a move into a better neighborhood) that both improved the mental health and economic productivity of children, an effect sustained into adulthood (Chetty, Hendren, & Katz, 2016; Costello, Erkanli, Copeland, & Angold, 2010). The importance of these two experiments is that they suggest that, however many assaults on children are occasioned by economic privation, the amelioration of overall poverty can have major benefits for healthy child and adult development.

A growing number of studies suggest that the conditions that may accompany early life poverty, including lack of cognitive stimulation, undernutrition, maternal stress, or insufficient parental nurturing, may have long-term effects on neural and psychological development (Eisenberg et al., 2007; Farah et al., 2006; Hackman, Farah, & Meaney, 2010; Noble et al., 2015; Teicher, Samson, Anderson, & Ohashi, 2016). Well-documented associations between early life adversity, educational attainment, and cognitive health, are driving current efforts to elucidate the causal pathways accounting for these associations (National Institute on Aging, 2016; Power & Hertzman, 1997) including efforts to ascertain the extent of genetic influences on these associations (Krapohl & Plomin, 2016). Finally, as noted above, fetal undernutrition, often associated with poverty, has significant long-term consequences for adult health but postnatal undernutrition does as well (Tennant et al., 2014).

3). Does moderate adversity promote resilience?

Dimensions of adversity exposure may not be monotonically related to unfavorable health outcomes. Moderate levels of adversity in childhood may “steel” individuals against adversities later in life, a phenomenon known in some research fields as “stress inoculation,” demonstrated in animal models and with preliminary evidence in humans (Liu, 2015). For example, experimentally controlled brief separation of squirrel monkey offspring from their mothers led, in follow-up, to enhanced cognitive control and exploratory behavior in the offspring (see review: Lyons, Parker, & Schatzberg, 2010). Several studies in children suggest that prior exposure to moderate stress enhances development of executive functioning (Finch & Obradović, 2017). In young human adults exposure to family conflict in early life has a curvilinear relationship with HPA axis reactivity, measured in response to a laboratory interpersonal conflict paradigm, suggesting that moderate levels of early life stress have a buffering effect on subsequent stress reactivity (Hagan, Roubinov, Purdom Marreiro, & Luecken, 2014). In adults, moderate levels of recalled adversity in childhood were associated with the most favorable quality of life and mental health scores in 60-year-olds (Holtge, McGee, & Thoma, 2018). Likewise, in a sample of Northern Ireland adults, moderate levels of recalled adversity in childhood lessened the impact of the civil conflict on the frequency of suicidal thoughts and behavior but not on substance abuse, mood, or anxiety (McLafferty et al., 2018). The quantitative studies do not make clear what children learn from experiencing moderate adversity. More nuanced studies suggest some mechanisms underlying the descriptive terms of “steeling” and “stress inoculation.” For example, in Elder’s well-known study, adolescent boys during the great depression took on adult-like responsibilities to partially offset family hardships (Elder, 1974). This accelerated crystallization of their own career aims, and self-confidence born of early entry to productive roles in their family and beyond in adolescence enhanced their educational and work success. In sum, linkage of mastering coping strategies and deepening self-confidence and self-regulation may underlie many of the advantages of moderate stress in the lives of young people.

4). Are there periods in development where children show heightened sensitivity to particular adversities?

ELAs occurring in critical or sensitive periods – when developing systems are plastic and particularly sensitive to environmental inputs - may have enduring effects on liability for noncommunicable diseases later in development. These early exposures may prime or “program” systems within the fetus or infant to anticipate characteristics of the subsequent environment to which they will be exposed (Hanson, Godfrey, Lillycrop, Burdge, & Gluckman, 2011). According to this view, adverse exposures during critical periods result in irreversible changes; if the environment in later childhood and adult life differs from that “predicted” during fetal life and infancy, the developmental responses may increase the risk of adult disease. Broad categories of adverse exposure that induce developmental responses of this kind include maternal obesity (Godfrey et al., 2017), micronutrient deficiency (Lin et al., 2017), and stress as early as preconception (El‐Heis et al., 2017), as well as during pregnancy and early postnatal periods (Wen et al., 2017).

As a complement to models of fetal and infant programming, Power and Hertzman (1997) articulate a pathways model whereby ELA is linked to health outcomes by a series of sequentially contingent life experiences. For example, in their study of the 1958 British cohort, they found that social class at birth was linked to a range of health outcomes, but the links were stronger for educational attainment, which was partially contingent on - but not equivalent to - social class at birth. Unlike critical period mechanisms, which suggest the need for intervention efforts at precise times in early development, pathway models suggest the potential of interventions at later stages of the life span to offset or moderate one or more of the contingent sequential adversities associated with ELA, a topic we explore further below. Current evidence suggests that both processes are at play, and consideration of both fetal programming and pathways models is essential for understanding constraints on later life plasticity and for selection of targets for later life intervention.

5). Do characteristics of the child play a role?

Differences among individuals across the life span can exacerbate or ameliorate the effects of ELA. In sum, it is clear that the impact of ELAs on long term development can be moderated by child characteristics in the years both during and immediately following ELAs and that individual differences may continue to play a moderating role in adulthood.

For example, resilient maltreated children effectively utilize a personal strategy of emotional reserve along with a controlled and rational interaction with others; presumably this reduced the likelihood of being targets of further maltreatment (see: Cicchetti, 2013) for a review). Differences among children and adolescents in locus of control and self-regulation may also play a decisive role in their capacity to offset severe adversity. First, locus of control and self-regulation are dependent on robust function of prefrontal cortex, amygdale and hippocampus. These structures—some studies suggest—are adversely affected in most children by severe ELA (Teicher, Samson, Anderson, & Ohashi, 2016). Second, in initial studies interventions that promote self-regulation in adolescents exposed to chronic and severe adverse experiences—interventions that model natural variation in these qualities in exposed adolescents-- lead not only to protection of hippocampal and amygdale injury by adversity but, additionally, to diminished insulin resistance (Brody, Gray, et al., 2017; Brody, Yu, Chen, & Miller, 2017); as noted above, insulin resistance is an important pathway that may lead from ELA to a range of chronic illness in adulthood.

Resilience in adults, to the effects of severe adversities in their childhood, is less well studied. However, successful strategies for overcoming the long term effects of severe economic adversity include positive cognitive reappraisal of current stresses as well as optimistic persistence; adults with these profiles show a lessened allostatic load despite harsh economic conditions of childhood (Chen, Miller, Lachman, Gruenewald, & Seeman, 2012)

While the intervention studies with adolescents need replication, their initial findings—when added to studies of resilient adults--support a more general concept that redirection of health trajectories of those exposed to ELA may be possible throughout the lifespan. A key may be resilience built around the realted constructs of self-regulation, mastery, persistence, and locus of control. In some cases, children and adults may already have these capacities. In other cases, these qualities may be enhanced by planned interventions. In both cases, a restorative mechanism may induce compensatory brain changes (Halfon, Larson, Lu, Tullis, & Russ, 2014).

While differences among children or adults in these sources of resilience may reflect genetic influences, we now have direct evidence of the role of genetic variation in moderating the impact of ELAs. For example, highly heritable traits such as verbal intelligence (Bagley & Mallick, 2000) can moderate the effects of ELA. Moreover, preliminary evidence has identified specific polymorphisms of the child (e.g., Cicchetti, Rogosch, Sturge-Apple, & Toth, 2010) that moderate the impact of maltreatment on psychological development and psychopathology.

An entirely different role for individual differences, harder for many to countenance, is the role of the child in evoking ELAs. There is substantial evidence, for example, that children with physical, neurological, or behavioral abnormalities are at increased risk of abuse (Jones et al., 2012). Somewhat weaker evidence suggests a role of difficult child temperament (e.g., (Mackenzie, Nicklas, Brooks-Gunn, & Waldfogel, 2011). Most arresting are studies suggesting heritable characteristics of children may evoke harsh parenting (Jaffee et al., 2004) and even more severe maltreatment (Schulz-Heik et al., 2010). These evocative characteristics of children may be amplified by parents who are challenged, frightened, or overwhelmed by them (Bugental et al., 2002; Bugental & Shennum, 1984; Martorell & Bugental, 2006). Thus, harsh parenting or frank maltreatment may reflect, in some cases, an escalating transaction between parents made vulnerable by their own risk factors and evocative characteristics of their children.

6). Can we detect, early in development, those children exposed to severe adversity those who are most likely to suffer consequent long-term health problems?

Early indicators of biological embedding of ELA may permit identification of resilient children and the mechanisms of that resilience, of factors that may determine the collapse of this resilience later in development, and may represent the earliest indicators of the impact of ELA on subsequent psychobiological processes that accelerate disease processes. These indicators help improve precision ascertainment of health risks attributable to ELA, before disease processes take hold.

The allostatic load model proposed by McEwen and Seeman (McEwen, 1998; McEwen & Seeman, 1999), articulates how multiple or excessive “hits” to systems that serve to maintain homeostasis and promote health in the face of normal day-to-day challenge can become dysregulated under severe or cumulative adversity. To maintain homeostasis, stress activates the autonomic and central nervous systems that, in turn, influence cortisol, adrenalin, the immune system, and metabolism. Severe and unrelenting challenges to these systems can result in gradual and long-term dysregulation in these homeostatic processes, causing them to become overused and dysregulated, resulting in allostatic load and overload (McEwen, 2004). Several approaches have been used to cumulate indices of systems affected, including those based on assays of urinary cortisol, norepinephrine and epinephrine, dyslipidemia, blood pressure, and waist-hip ratio, validated through prediction of future health and mortality (Seeman, McEwen, Rowe, & Singer, 2001).

Adults with a documented history of maltreatment show evidence of high allostatic load (Widom, Horan, & Brzustowicz, 2015). Allostatic load is also manifested in children exposed to severe adversity (Evans, 2003). One component of this load, serum cholesterol, may be elevated in the short term in children with ELA but decline to levels below children without ELA (e.g., Trickett, Noll, Susman, Shenk, & Putnam, 2010). Indeed, low serum cholesterol has been reported in some children exposed to ELA who go on to develop psychiatric difficulties (Laurent et al., 2013; White et al., 2017).

More recently, gene expression profiles have been shown to be sensitive to adverse rearing conditions in animals (Meaney, 2001) and humans (Miller et al., 2009). Animal studies suggest that effects of maternal deprivation on offspring gene methylation profiles can be observed in the first month postpartum (Massart, Nemoda, et al., 2016). Effects of prenatal stress in mothers on gene expression profiles in the developing fetus are apparent in both human (Monk et al., 2016) and monkey placentas (Massart, Suderman, et al., 2016).

Models of ELA, such as those of McLaughlin, Sheridan, and Lambert (2014), may widen the scope of search for initial indicators. For example, early indicators of biological embedding in children suffering from ELAs that deprives them of cognitive and social stimulation might include reduced cerebral cortical thickness and performance deficits on complex cognitive tasks.

An equally strong case for “psychosocial embedding” can also be made. There is good evidence that child maltreatment, for example, is prospectively associated with children’s problems in emotion regulation, impaired attachment security, withdrawal from or aggression with peers, school problems as well as personality problems, cognitive deficits, and both internalizing and externalizing behavioral problems (for a comprehensive review, see: Cicchetti & Toth, 2016). These childhood difficulties may help identify children who were both exposed to adversity and are increased risk for later problems. In addition, some of these behavioral or relational problems might reflect a causal path connecting early adversity to health outcomes in mid-life. For example, attachment insecurity in infancy and toddlerhood anticipates cardiorespiratory disease and diabetes at age 32 (Puig, Englund, Simpson, & Collins, 2013).

For indicators of both biological and psychosocial embedding, it remains an important research task to determine their persistence into adulthood. Is there a marker that accurately identifies the subgroup of adults with a history of ELA who are vulnerable to later health impairments that are not yet apparent? Beyond identifying this subgroup, there is a broader problem of whether we—at the current state of our knowledge—can validly identify adults who have been exposed to ELAs, whether or not such exposure renders them vulnerable.

7). Can we ascertain in adults their exposure to adversity when they were children?

Most research reports on the links between ELAs and adult health derives from studies using retrospective recall through questionnaires or structured interviews. In most cases, with notable exceptions (e.g., Barnes, Noll, Putnam, & Trickett, 2009), these procedures have only been validated against other adult ascertainment such as detailed clinical interviews (e.g., (Bernstein et al., 1994; Bifulco, Bernazzani, Moran, & Jacobs, 2005; Wingenfeld et al., 2011) or with each other (e.g., Wingenfeld et al., 2011). Recent efforts have focused on the relationships between retrospective and prospective measures of ELA (including agency reports, on children in long-term longitudinal studies) and suggest that agreement is modest, with average kappa in across 16 studies, with 25,741 unique subjects, equaling 0.19 (Baldwin, Reuben, Newbury, & Danese, 2019). Fifty-six percent of adults who reported severe adversity in childhood did not have evidence of actual exposure in concurrent childhood records. Fifty two percent of adults with prospective childhood records did not report adversity as adults (Baldwin et al., 2019). Importantly, prospective and retrospective measures may be measuring different constructs. For example, those who recall adversity, in comparison to those prospectively ascertained, score higher on measures of neuroticism, lower on agreeableness (Reuben et al., 2016), and higher on perceived pain (Raphael, Widom, & Lange, 2001)², providing preliminary evidence that retrospective recall of childhood adversity may be associated with an individual’s high reactivity to physical and emotional stimuli.

Meanwhile, prospective measures cannot be considered foolproof. Agency reports, for example, have biases particularly in that many severely maltreated or traumatized children are never detected (Shenk, Noll, Peugh, Griffin, & Bensman, 2016). In addition, comparisons between prospective and retrospective measures in the same sample must reflect scientific consensus that both measures are aimed at the same construct. Retrospective measures are often the sole source of data in many large, population samples of adults Although some of these have revealed associations similar to those seen with prospective measures (e.g., for economic outcomes (Currie & Widom, 2010; Pinto Periera, Li, & Power, 2017), for HbA1c (Thomas, Hypponen, & Power, 2008; Widom, Czaja, Bentley, & Johnson, 2012), and for markers of chronic inflammation (Baumeister, Akhtar, Ciufolini, Pariante, & Mondelli, 2016; Pinto Pereira, Merkin, Seeman, & Power, 2019) it is unclear if the pathways from recalled adversity to these health outcomes are the same as those from prospective, contemporaneous measurement.

A major challenge now is to improve the accuracy of adult recall of ELA; this is warranted because population samples are crucial for understanding the long-term impact of ELA on adult health, moderating factors, and identification of potential beneficiaries of adult interventions. Further, very large. population-based cohorts—even those followed from pregnancy and birth—may be unable to effectively obtain crucial, sensitive information about child maltreatment until the children reach an age when they can report it themselves. Equally important, prospective studies with comprehensive ascertainment of ELAs have limited value when the scientific focus is on a relatively rare outcome such as schizophrenia.

One key approach to improvements seems likely to come from the development of structured interviews to replace questionnaires (Barnes et al., 2009). The Comprehensive Trauma Interview (CTI) represents one such effort (Noll, Horowitz, Bonanno, Trickett, & Putnam, 2003). The CTI is a semi-structured interview assessing many forms of child maltreatment. To date, such efforts remain limited and development of more accurate and reliable retrospective reports of ELA remains a critical target for future research efforts. Drawing on experience with the CTI, key elements likely to be important to the development of better measurement tools include providing subjects a more detailed definitions of ELA and using follow-up queries to aid in the reduction of false positives.

Additional strategies to improve retrospective recall might include building on basic memory research, using strategies to recreate contexts of adversity, e.g., pictures of sites where adversity may have occurred (Goodman & Melinder, 2007; Memon, Meissner, & Fraser, 2010). Finally, one might search for behavioral patterns (e.g., Kumsta et al., 2015; Shaffer & Sroufe, 2005), distinctive gene expression profiles (e.g., Provençal et al., 2012), markers of chronic inflammation (e.g., Miller et al., 2009), or neurobiological markers (e.g., Gee et al., 2013) that – in combination – are sensitive to ELA.

What is needed most is a comprehensive research program that addresses three major questions. First, what long term effects of actual exposure occur independently of whether or not they are remembered later in development? Second what processes account for why some people remember severe ELAs when there is little evidence of actual exposure, and what processes lead others to forget or conceal those experience years later? Finally, what role do these memory processes play in enhancing or reducing risk of long-term effects of ELA? This program of research can be partially anchored by evidence that the memory of ELAs reliably anticipates major health outcomes, and some of the pathways between the two are being explored. For example, the prospective association of memories of ELA – as revealed in the ACEs questionnaire – with lung cancer appears partly attributable to increased smoking among those recalling adverse childhoods (Brown et al., 2010). Future research may establish more detailed biological mechanisms linking recalled ELA and illness incidence. For example, recalled ELA has been associated with general indices of cellular senescence such as telomere shortening (Puterman et al., 2016), known risk factors for subsequent illness such as chronic inflammation (e.g., Miller & Raison, 2016), and brain changes that may be the basis of a broad range of maladaptive behaviors (McCrory, De Brito, & Viding, 2011).

Strategies for research on preventive intervention

Based on the best evidence available, the Network anticipated two productive avenues for research on prevention. The first capitalizes on a large volume of data derived from studies using measures of recalled adversity. While, as noted, uncertainties remain about determinants of these memories, evidence suggests that they constitute important liabilities for ill health and are an adequate base for new research on prevention. A second avenue would focus on indivduals who have sustained actual exposure to ELAs as determined by contemporaneous, prospective measurement.

Strategy one: Ameliorative prevention in established illness based on retrospective ascertainment of ELAs.

One area where the measures of recalled adversity have immediate importance is for the prevention of a malignant course or resistance to treatment in individuals already diagnosed with disease. For example, in the mental health field, recalled adversity predicts both a malignant course and resistance to treatment for major depression and a malignant course for bipolar illness (Agnew-Blais & Danese, 2016; Nanni, 2012). Similar investigations are just beginning in more general medicine. For example, recalled adversity is associated with poor response to bariatric surgery and may be associated with greater pain and treatment refractoriness in migraine (Lodhia et al., 2015; Tietjen, 2016). Taken together, these data suggest a program of research that should bring useful results rapidly. Two answerable questions are pressing: is recalled adversity a liability for malignant course and treatment resistance for a broader range of prevalent psychiatric and medical disorders? If so, are novel treatments required for those who recall ELA versus those with the same condition who do not? These findings suggest that recalled ELA may represent a critical individual difference factor important for tailoring treatments and monitoring adherence and should be assessed in clinical trials to better understand differential treatment response. For example, some patients with depression have elevated indices of chronic inflammation and respond selectively to antidepressants that target downstream effects of inflammation on brain metabolism (see: Haroon & Miller, 2017, for a review). Prospective data, from a longitudinal study assessing ELAs in childhood, suggest that ELA is distinctively associated with the subset of depressed indivduals with elevated indices of inflammation (Danese et al., 2008).

Strategy two: Targeting risk mechanisms in adults to pre-empt disease onset

A second strategy seeks to develop preventive interventions for adults who have been exposed to severe adversity and are therefore liable to disease onset but are still relatively healthy. To accomplish this aim, research must 1) reliably and validly identify samples of adults at risk, 2) identify causal and malleable risk mechanisms, and 3) develop scalable interventions that successfully target those risk mechanisms thereby pre-empting adult disease.

1). Tractable samples of adults at risk.

The field has already drawn heavily on a small number of prospective samples where subjects have been followed into adult life (e.g., Barnes et al., 2009; Danese et al., 2009; Widom et al., 2012), but our Network’s review (Network on Reversibility, 2016) has identified at least 12 other underutilized cohorts, some of which may be amenable to additional data collection. In addition, a sample of high-risk twins, on whom ELA exposure was carefully and contemporaneously documented, is entering adulthood and will soon be a valuable resource (Jaffee et al., 2013). Likewise, children - such as those at the Mt. Hope Center studies - in whom maltreatment and its short-term impact were meticulously ascertained have reached or will soon reach adulthood. The same is true of children in samples with varying length of time in institutional care from birth on (e.g., Rutter & O’Connor, 2004), although caution must be taken to select samples where institutional care entailed severe deprivation (Woodhouse, Miah, & Rutter, 2017). In addition to these samples, long-term follow-up of children with ELAs who were randomly assigned to ameliorative or preventive interventions will be of continuing value, although the strongest mechanistic evidence for malleable targets will derive from interventions that focus narrowly on specific risk mechanisms, i.e. putative, treatment-sensitive intervention targets associated with the embedding or persistence of risk from ELA such as adverse parent-child relationships (e.g., Brody, Yu, Chen, Beach, & Miller, 2016; Stronach et al., 2013), rather than from broad-based, multimodal interventions (e.g., Campbell et al., 2014; Nielsen et al., 2018). Animal models, as they do in clinical pharmacology, may provide initial proof of concept opportunities for testing interventions long before they are ready for human trials. For example, rodent models have clarified gene expression mechanisms that link impaired maternal care (low licking and grooming) to stress reactivity and behavioral constraint in adult animals. We are a long way from attempting direct intervention on specific gene expression mechanisms in human preventive trials to offset ELAs. However, it is illuminating that rodent experiments have demonstrated that agents targeted at gene expression mechanisms can reverse the effects maternal deprivation on stress reactivity (Champagne & Meaney, 2007).

We anticipate at least three immediate uses for human cohorts described above. First, human cohorts identified by prospective assessment will offer solid data for validating improved methods of retrospective ascertainment. Second, they will help to clarify risk mechanisms in the causal chain. And finally, a few may be amenable to subsampling for experiments or micro-trials to test the malleability of those mechanisms. For example, considering a possible pathway from prospectively ascertained ELA to insulin resistance in adulthood, a micro-trial might test whether standard treatment for insulin resistance is equally effective in individuals with and without a documented history of ELA. Parallel work in existing adult samples with retrospective assessment will offer a triangulation of evidence, and/or help clarify the differential pathways by which prospectively and retrospectively assessed ELA predict adult health and identify potential intervention targets along these pathways.

2). A choice for defining targets for intervention: aiming at distinctive processes in childhood or at chains of circumstance across the lifespan.

Currently, we don’t know enough to select risk mechanisms as targets for intervention to prevent the adult health consequences of childhood adversity. However, research will advance if it distinguishes two broad classes of targets. First, are risk mechanisms that are influenced by severe adversity occurring during critical periods in fetal and child development and have enduring effects without requiring additional adversity later in development. Fetal under-nutrition-- previously noted--provides one of the clearest examples of an adversity in early life that has a unique and lasting impact because it occurs at a specifically sensitive period in development. Accordingly, current prevention trials are aimed at improving maternal nutrition before and during pregnancy. With ELAs of this type, an obvious first choice is to focus intervention on processes unfolding in the critical childhood period. However, new research—noted below--raises hopes that critical periods may be reopened in adulthood thereby providing a second window of opportunity for prevention.

In contrast, where ELAs have long-term impact via a pathways process, there are opportunities for preventive interventions at several points in development. Already noted, as an example, is the effect of poverty in childhood on subsequent diminished educational attainment that—later in development—anticipates poor health. If the pathways model is correct, then measures to alleviate poverty in early childhood and/or measures to enhance educational attainment later in development should effectively offset some of the consequent developmental deficits.

Childhood adversities whose long-term effects arise primarily from their impact on processes during a critical or sensitive period early in development.

Discovering adversities of this kind will depend, first and foremost, on precise developmental timing of their major effect in human development. There are critical periods for visual, auditory, and language development, and ELAs that result in severe disruption of normal eyesight, hearing, or language development during these periods will have lasting deficits when children are exposed to normal visual or auditory stimuli or language later in development. To cite a simple animal model, in ordinary development the effect of monocular occlusion in a critical early period results in irreversible amblyopia. However, several behavioral and pharmacological interventions can evoke neuroplasticity equal in kind and quantity to the early critical period of sensory development and allow not only the restoration of normal vision but the re-imposition of enduring amblyopia when monocular occlusion is applied to these adult animals (e.g., Vetencourt et al., 2008). Several current clinical trials aim at comparable restoration of sensory impairment in human subjects through reopening critical periods.

However, it is unclear to what extent most ELAs, such as severe maltreatment, have lasting effects by mechanisms analogous to visual and auditory loss in early childhood (for a discussion, see: Callaghan & Tottenham, 2016). To explore this possibility, research must ascertain with precision the developmental onset and the offset of the adversity, a task that is difficult for persistent adversities like poverty or adverse family environments. Further, even if the time of offset and onset can be determined, we require evidence of a persistent deficit despite the restoration of a normally-expected environment.

To return to an animal example, impaired maternal care of a rat dam within a critical 7-day period after birth has lasting effects on her female pups’ capacity to care for their own offspring. Although this effect persists in ordinary environments, it can be offset by rearing these offspring in enriched environments post weaning (Champagne & Meaney, 2007). In human adoption studies by Michael Rutter and his colleagues (Kumsta et al., 2015; Sonuga-Barke et al., 2017) children were reared in unfavorable institutional settings from birth but adopted at various ages into adopting families providing them reasonable care. These children showed persistence of autism like behaviors, attachment deficits and attentional deficits through their most recent follow-up (age 22–25) but only if they were adopted after 6 months of age. These deficits were not influenced by variation in the quality of adoptive rearing but were highly dependent on the length of institutional care. Thus, the period between 6 months and 42 months, the highest age for adoption in this sample, may constitute a critical period for acquiring specific social and attentional abilities. It may be impossible to acquire these skills later in development, in ordinary rearing environments, without special approaches to intervention that are informed by a deep knowledge of the risk mechanisms unfolding from 6 to 42 months.

Adoption designs are also useful for specifying the timing of adversity during prenatal development. For example, in a study of children adopted at birth, prenatal depression in birth mothers affected cortisol dysregulation on the pathway to depression in adopted children (Laurent et al., 2013). In a further refinement of the adoption method, using children conceived by in vitro fertilization, Lewis, Rice, Harold, Collishaw, and Thapar (2011) showed that maternal depression was associated with children’s depression regardless of whether the mother’s own (genetically-related) or a donated (not genetically related) fertilized ovum was implanted. Taken together, these two studies provide strong evidence that exposure to maternal depression, very specifically in the prenatal period, has a causal relationship to subsequent behavioral development. However, postnatal maternal depression also has a major impact on child development and it remains unclear whether prenatal depression leads to specific developmental problems distinct from those associated with maternal depression at any later point during childhood or adolescence. That is, are the developmental processes affected by maternal depression in the fetal period unique?

Where rigorous design has established a “critical period” for a specific adversity, intervention development will benefit from recent research on mechanisms of neural plasticity that explain both the sensitivity as well as the onset and offset of these periods. These insights also hold potential for enhancing the effectiveness of interventions at midlife. Described in detail elsewhere, (Bavelier, Levi, Li, Dan, & Hensch, 2010; Werker & Hensch, 2015), molecular brakes control both the onset and termination of critical periods of openness to the environment by their action on gabaergic parvalbumin cells that—in response to triggers that are now well characterized—facilitate neural growth and organization and, when inhibited, freeze that development. Thus, stimuli during the critical period can have a lasting impact on brain organization and function. A range of interventions in adults may reopen windows of plasticity and, potentially, correct deficits stemming from ELAs. Recent studies in mice demonstrate the role of locomotion in restoring plasticity to the cortex through molecular mechanisms that block the braking action on cortical sensitivity, likely through direct action on the parvalbumin cells (Stryker, 2014). While molecular mechanisms have not been clearly delineated, a broad range of evidence suggests that exercise can restore cognitive function in sedentary older adults through stimulation of increased BDNF (Erickson et al., 2011), a protein that can induce neuroplasticity through action on parvalbumin cells, and regrowth of brain structures—such as the hippocampus—that ordinarily shrink with age. Exercise also improves mood and cardiovascular and metabolic health, suggesting broad-ranging benefits. More recently, research has explored the effectiveness of neurofeedback, a procedure where subjects learn how to self-activate specific brain regions. Under normal circumstances, activation of the ventral tegmental area enhances hippocampal function and memory for more complex relationships and contexts (Murty & Adcock, 2017); preliminary data suggest neurofeedback can help human subjects to selectively activate this neural center, which plays a critical role in new learning (MacInnes, Dickerson, Chen, & Adcock, 2016). In addition to these induced changes, increases in neuroplasticity may occur naturally in adult life. For example, animal studies (Champagne & Curley, 2016) point to brain plasticity in mothers around the time of the birth of their first offspring. Leveraging the plastic potential of the adult brain to reverse or compensate for health risks due to ELA is a promising area of future research, with initial questions focused on demonstrating the malleability of specific risk mechanisms via plasticity-inducing manipulations. However, it remains for future research to discover whether these mechanisms of enhanced plasticity—such as those evoked by exercise or neurofeedback--duplicate those where critical periods have been reopened and whether this enhanced plasticity of adult brains facilitates preventive interventions for either “critical period” or “pathways” mechanisms or both.

Childhood adversities that initiate a long chain of adverse circumstance with cumulative effects across the life span.

Although pathway models are complex, they may provide a broader range of potential intervention targets at various points across the lifespan. However, even when tested in carefully conducted longitudinal studies spanning childhood and adulthood—with repeated assays across time—there are two major threats to their validity. The validity of these models may be confounded and the confounds may become more abundant as the complexity of pathway models increases. This is tractable problem if the potential confounds are known and measurable. In these circumstances, many investigators have turned to analytic strategies--such as directed acyclic graphs-- to clarify not only the mediating and moderating steps linking childhood factors with adult health, but also to incorporate as many potential confounders as possible (e.g., Kamphuis, Turrell, Giskes, Mackenbach, & van Lenthe, 2012).³

A second threat comes from unmeasured confounders; the most important may be genetic. The confounding arises because genetic factors can influence measures of environmental adversity including parenting, measures of stress and peer and teacher relationships (see review: (Kendler & Baker, 2007), as well as measures of health status. Many reports document that the genetic factors that influence environmental variation also influence child, adolescent, and adult development and, thereby, account for a notable portion of the observed covariance between environment and development. For example, South, Schafer, and Ferraro (2015), using a large twin subsample of the Midlife in the United States (MIDUS) study, found that virtually all of the observed association between recalled childhood adversity and self-reports of current health were accounted for by genetic influences common to both. Comparable findings, using twin studies, have been reported for associations of adversity measures with childhood neurodevelopmental disorders (Dinkler et al., 2017) and adult criminal behavior and violence (Forsman & Langstrom, 2012), but not for ADHD symptoms in adults (Capusan et al., 2016). Adoption studies are useful here as well: in a longitudinal adoption study, parental depression was associated with adolescent and young adult psychopathology in a sample of biological parents rearing their own children and in adopted parents and their adopted children, suggesting environmental causality; however, the link between parental alcoholism and child psychopathology was found only for children reared by their biological parents, suggesting that this association was due to genes common to parents and children (King et al., 2009; Marmorstein, Iacono, & McGue, 2012).

Where evidence favors a pathways model for the long-term effects of adversity, the allostatic load model--noted above--has been a rich source for hypotheses and investigation of risk mechanisms that may be targets for preventive intervention. For example, Rasgon and McEwen (2016) have reviewed animal and human research documenting links among severe and sustained stress, elevated glucocorticoids and amino acids, chronic inflammation, and insulin resistance, with ultimate effects on reduced brain plasticity, all of which, to a point, are reversible by a range of interventions from exercise to dietary restriction. Because these pathways can be initiated in the pre-conception phase (Godfrey et al., 2017), as well as at various stages in fetal and child development, a constantly improving map of these interrelationships will offer multiple opportunities for preventive interventions. Pathway models are also important for linking early adversity to health outcomes via cognitive pathways and related outcomes such as education (Power & Hertzman, 1997) or behaviors that impair self-regulation and care, increase risk-taking, or lead to disrupted social relationships (e.g., Colman & Widom, 2004). Indeed, research on neuroplasticity, noted above, may be particularly useful in interventions directed at these behaviors: enhanced neuroplasticity, through such strategies as exercise or neurofeedback, may increase the effectiveness of preventive interventions aimed specifically at self-regulation, risk-taking, or disruptive patterns of social relationships; they may also be relevant for some steps in metabolic pathways as our understanding of their relationship to brain function improves.

3). Selecting outcomes for preventive intervention trials.

A major problem confronting prevention research in cohort studies over extended periods is the long-time gap between ELA and the specific health outcomes to which it has been linked. We have already noted the utility of the allostatic load model to search for early indicators of embedding of ELAs. Extensions of this model provide many potential mediators between ELA and health outcomes, such as chronic inflammatory states and insulin resistance that appear later in life. These markers that anticipate later disease represent useful intermediate outcomes for testing the efficacy of preventive interventions.

Recent research offers new approaches to measuring fundamental processes of premature biological aging, some of which can be detected in early adulthood and are, in turn, related to a range of health outcomes that have been linked to ELA. Like allostatic load and other indices, measures of these aging processes may constitute potential outcomes for intervention research; they may appear earlier in development and hence provide valid outcome measures that reduced the time gap between intervention and outcome. For example, Belsky and colleagues (2015) summed changes in 18 biomarkers—including telomere length, HbA1C, triglycerides, mean arterial pressure and BMI, from ages 26 to 38 to compute a pace of aging score. Many of these measures overlap those drawn from the allostatic load model but, measured over several time periods, they add rate of change that may increase their precision as predictors of a broad range of health outcomes. This rate of change score correlated with other aging indicators such as grip strength, retinal arterioles, and older facial appearance, and was anticipated by adverse child experiences, such as lower SES and diminished child health, but also by a potential genetic indicator: age of longest-lived grandparent. Using a comparable strategy, Horvath (2013) identified 353 CpG sites whose methylation correlated with chronological age. Individual variation in methylation at these sites allows computation of “methylation age” that has been associated with all-cause mortality (Chen et al., 2016). Horvath (2013) reports that this methylation aging score is 100% heritable in infants but declines to 39% in older individuals suggesting increasing environmental influences. Indeed, Brody and colleagues (2016) have presented preliminary data that parental depression—a presumed stress for their children--anticipates methylation age in adolescents that can be offset by intervention to reduce harsh parenting by these parents.

Taken together, data on allostatic load, pace of aging, and methylation clocks suggest that evaluating efficacy of preventive interventions in healthy adults to forestall medical disorders need not endure the many years between intervention and the emergence of the disorders selected for study. Biomarkers of multi-system dysregulation and aging may be a valid, near term outcome for a first series of randomized trails.

Conclusion

Human and animal research is establishing clear links between ELA and poor health in adulthood. Basic research in behavioral and physical development is clarifying targets for preventive intervention in adults to offset the risks of ELA. Retrospective accounts by adults of their ELA has helped distinguish subgroups of adult psychiatric patients that will have a malignant course and be treatment resistant. Research on the specific treatment needs of this subgroup is both promising and urgent. Equally urgent is research on whether recalled ELA has similar significance for medical disorders. Research on the prevention of ill health, originating in ELA, will be facilitated by a greater use of cohorts where ELA has been documented in childhood. Effective trials will identify and target potentially malleable behavioral, psychological, or neurobiological processes that can reverse or compensate for the persistent risk due to ELA, utilizing current research on the effects of ELA on risky behavior, metabolic and inflammatory processes neurodevelopment and neuroplasticity.