Maltreated children are more likely to suffer psychiatric disorders over the course of their lifetime. In particular, they are more likely to develop major depression (1–5), bipolar disorder (6), anxiety disorders (2, 3, 7), posttraumatic stress disorder (PTSD) (2, 3), substance abuse (2, 8, 9), personality disorders (10, 11), and psychoses (12). Furthermore, it appears that survivors of early maltreatment differ in critical ways from other individuals with the same psychiatric diagnoses. Disorders emerge earlier in maltreated individuals, with greater severity, more comorbidity, and a less favorable response to treatment (13–15). Maltreated individuals may also have discernible brain abnormalities that are not present in their nonmaltreated counterparts (16, 17). Childhood maltreatment is also linked to a wide array of medical disorders, shortened life expectancy, and reduced telomere length (18, 19). Hence, an understanding of maltreatment as an etiological risk factor is crucial to the development of a science of preventive psychiatry, to the design of effective therapeutic regimens, and to the delineation of an accurate nosology.
Our goal in this review is to advance the thesis (17, 20–23) that affected individuals with childhood maltreatment constitute a critically distinct subtype across depressive, anxiety, and substance use disorders. We also propose that the maltreated subtype may be thought of as a phenotypic specialization (phenocopy) resulting from environmental experience—or more precisely, an ecophenotype.
Why focus on maltreatment? It is maltreatment rather than exposure to other stressors, such as natural disasters, that consistently presents as the antecedent to psychopathology (24, 25). This makes sense. Children are dependent on the adults around them for their survival, and they can endure great hardship if they feel protected and cared for. But when the hardship is the product of their caretakers, and when it is the caretaker who must be protected against, it creates a stressor with far-reaching ramifications.
Our conceptualization of ecophenotypes emerged from a systematic review of the English-language literature on the psychiatric and neurobiological consequences of childhood maltreatment. Details on how the review was conducted, as well as tabulated results of sexual abuse as a psychiatric risk factor, are presented in the data supplement that accompanies the online edition of this article. Studies selected for citation are representative. No contradictory studies showing a significant protective effect of maltreatment were encountered. In this review, we excluded disorders for which research suggests that the vast majority of patients were exposed to some type of abuse or neglect, such as borderline personality and dissociative identity disorder (10, 11, 29, 30). We also excluded schizophrenia and bipolar disorder, which are known to be highly heritable. Instead, we focused on moderately inheritable disorders for which major subsets of patients can be distinguished by positive or negative histories of childhood maltreatment. These disorders include major depression, anxiety disorders, posttraumatic stress disorder, and substance abuse. Childhood maltreatment or early adversity accounts for 30%−70% of the population attributable risk fraction for these disorders (1, 3, 9).
Major Depressive Disorder
Some of the strongest evidence for an association between exposure to childhood maltreatment and the development of major depression can be found in the Adverse Childhood Experiences study (31), which showed that risk for depression increased in a graded, dose-dependent fashion with the number of maltreatment-related adverse childhood experiences. Exposure to one or more adverse childhood experiences accounted for 54% of the population attributable risk fraction for current episodes of depression (1) and 67% for suicide attempts (32). Having five or more adverse experiences increased the relative risk of receiving a prescription for an antidepressant 2.9-fold (6). Long-term prospective studies also indicate about a twofold greater risk attributable to maltreatment (2, 4, 5) (see Figure 2A). These findings are consistent with results of twin studies showing that heritability plays only a minor role in risk for moderate or even severe depressions (33).
FIGURE 2.Forest Plots Showing Odds for Psychopathology in Individuals Exposed to Childhood Sexual Abuse or Multiple Forms of Maltreatment Including Sexual Abusea
a References not included in the main text are provided in the online data supplement, along with further details of the analysis. The forest plots show odds ratios and 95% confidence intervals. Panels A–E address, respectively, diagnoses or suprathreshold symptoms of major depression; diagnoses of posttraumatic stress disorder; diagnoses or suprathreshold symptoms of anxiety disorders, including generalized anxiety disorder, panic disorder, and simple or social phobias; alcohol-related problems, including heavy episodic drinking, abuse, or dependence; and drug-related problems, including use of illicit drugs, abuse, or dependence. Studies were ordered within each cluster by year of publication. Multiple analyses within studies were pooled to provide assessment for overall risk across severity level and gender.
Maltreatment increases the risk for depression in both males and females, although some studies suggest a greater risk for depression in physically abused females than in physically abused males (34, 35). Hence, the greater female prevalence may be due, at least in part, to greater sensitivity to physical abuse and more frequent exposure to childhood sexual abuse (36).
Important clinical differences exist between depressive illnesses with and without childhood maltreatment. Depressions emerge earlier and have a more sustained course (13, 37) in maltreated individuals. These individuals also have more severe mood, neurovegetative, and endogenous symptoms and more comorbidities, particularly substance abuse (13, 22, 37, 38). Psychotic features are also more common, as are suicide attempts and deliberate self-harm (39).
Maltreated patients with depression also differ with respect to treatment response. A recent meta-analysis of depression outcome studies (13) confirmed that childhood maltreatment unequivocally predicts poor treatment outcome. However, it is also possible that maltreated patients with depression respond preferentially to therapies that are less effective for patients with depression who have no history of maltreatment. In a large clinical trial (40), chronically depressed participants received either pharmacotherapy with nefazodone, psychotherapy using the cognitive-behavioral analysis system of psychotherapy, or both treatments in combination. Psychotherapy was clearly superior to antidepressant monotherapy in the subset of participants with childhood trauma, and nefazodone provided little added benefit. In contrast, chronically depressed patients with no history of trauma or loss responded more favorably to nefazodone than to psychotherapy, and they benefited from combination treatment. On the other hand, in another study (41), maltreatment was associated with a poorer response to interpersonal therapy than to cognitive therapy or medication, and with rapid relapse. With hindsight, we can see that factors found over the years to predict treatment resistance in depression (i.e., early onset, comorbid anxiety and substance use disorders, axis II diagnoses, and presence of psychotic features) are the same factors now known to be characteristic of the maltreatment-related ecophenotype.
Neurobiological studies are beginning to provide compelling reasons for considering depression with a maltreatment history as a distinct subtype. Reduced hippocampal size is one of the more prominent neuroimaging findings in major depression. However, Vythilingam et al. (16) reported that reduced hippocampal size was present only in the subset of depressed individuals who had a history of maltreatment. On balance, there is now more consistent evidence for reduced hippocampal size in adults with a history of maltreatment than in adults with major depression. Furthermore, reduced hippocampal volume in maltreated individuals in the absence of depression or any psychiatric history has been observed in recent large-sample studies (42, 43). In short, what has been regarded as a key finding in major depression may instead be a consequence of early stress that serves in turn as a risk factor. Indeed, reduced hippocampal volume can precede and partially mediate risk for depression with early stress (44).
Amygdala activation during exposure to sad or negative faces is another neuroimaging finding linked to major depression (45) that may be limited to depressed individuals with a history of maltreatment (24). Indeed, bilateral amygdala reactivity to emotional expression is enhanced by a history of emotional maltreatment whether or not the individual has depression (46).
Genetic and epigenetic risk factors may also be distinctly different in patients with major depression with and without a history of maltreatment. A comprehensive meta-analysis by Karg et al. (47) found strong support for a gene-by-environment interaction involving the serotonin transporter promoter polymorphism and risk for depression when the environmental experience was childhood maltreatment but only marginal support when the environmental experience involved postchildhood stressful events.
Epigenetic hypermethylation of the Nr3C1 gene results in decreased expression of glucocorticoid receptors and potential hypersecretion of cortisol during stress. Interestingly, Nr3C1 has been found to be hypermethylated in postmortem tissue from suicide victims with a history of maltreatment, but not in suicide victims without maltreatment or in nonsuicide comparison subjects (48).
While some have speculated that individuals without maltreatment who develop depression do so because of a dense family history and a high heritable risk, this supposition is not supported by our unpublished data or by the observation that less severe forms of depression show little evidence of heritability (33). However, nonmaltreated depressed individuals may show an array of noninherited rare copy number variants—short stretches of DNA that are deleted or duplicated between individuals that contribute disproportionately to risk (49).
Finally, depressed patients differ in their risk for autoimmune, metabolic, and cardiovascular disorders based on maltreatment history. This may be related to chronic low-grade inflammation. Longitudinal data show that depression and inflammation are strongly coupled in depressed individuals with maltreatment but not in those without maltreatment (4).
Sexual abuse, physical abuse, and witnessing domestic violence are types of maltreatment that may fulfill the DSM-IV A1 criterion for a traumatic event (50), and they are major risk factors for the development of PTSD (Figure 2B). Scott et al. (2) reported an adjusted odds ratio of 4.86 for lifetime diagnosis of posttraumatic stress disorder in a prospective study of adults with a history of maltreatment. Furthermore, individuals who experienced both childhood adversity and adult traumatic events have been found to be more likely to develop PTSD than those who experienced either type of adverse event alone (51).
However, in recent years there has been growing concern about how well the DSM-IV conceptualization of posttraumatic stress, which is based on exposure to acute life-threatening events in soldiers, applies to maltreated children. Youngsters often experience traumatic or highly stressful events during a substantial portion of their life, which may be perpetrated by one or more family members rather than a faceless enemy. This has led to two important observations. First, DSM-IV criteria are not sufficiently developmentally sensitive. Severely maltreated children often do not meet full diagnostic criteria, as they frequently show symptoms in only two of three category clusters, but may be as impaired as children who meet full criteria (52). Furthermore, risk for posttraumatic stress in children appears to be influenced by frequency of exposure and multiplicity of exposure types rather than the degree to which they witnessed actual or threatened death or serious injury or experienced a threat to their physical integrity. Hence, children may be “traumatized” by repeated exposure to types of maltreatment that do not meet the A1 criterion for a traumatic event, such as emotional abuse (50).
Second, as van der Kolk (50) and others have articulated, traumatized children also show a complex array of problems, such as affective dysregulation, disturbed attachment patterns, behavioral regression, somatic symptoms, and altered attributions and expectancies that are not included in the current DSM conceptualizations and often lead to a host of comorbid diagnoses. Developmental trauma disorder has been proposed as a diagnostic category that more faithfully captures the critical events and clinical presentation of posttraumatic sequelae in chronically maltreated children (50).
However, developmental trauma disorder is best restricted to maltreated individuals with features of posttraumatic stress (see the online data supplement for further discussion). As noted above, many maltreated individuals are more accurately characterized as depressed, and timing of exposure may be a critical determinant. Schoedl et al. (53) found that individuals reporting sexual abuse after age 12 had a 10-fold greater risk of severe PTSD in adulthood than individuals reporting sexual abuse before age 12. Conversely, depressive symptoms were more severe in individuals reporting sexual abuse before age 12 than in those reporting it after age 12 (53).
Multiple lines of evidence suggest that maltreated individuals with PTSD continue to differ from their nonmaltreated counterparts in adulthood. They show greater symptom complexity (54), more comorbid mood disorders (55), and more severe dissociation (56, 57) or alexithymia (58), leading to the designation “complex PTSD” (54, 59, 60). There may also be important neurobiological and genetic differences.
A key neuroimaging finding in PTSD, particularly in combat veterans (61), has been reduced hippocampal volume. However, a study of monozygotic twins discordant for combat exposure found reduced hippocampal volume in combat-exposed individuals with posttraumatic stress as well as in their unexposed twins without posttraumatic stress (62). While these results may be confounded by individual drinking history or personality factors common to both twins, it is also possible that reduced hippocampal volume resulted from shared early stress and functioned as a risk factor for posttraumatic stress. As noted above, reduced hippocampal volume has been observed with considerable consistency in adults with a history of maltreatment. While some early studies with small sample sizes found reduced hippocampal size in maltreated adults with PTSD but not those without (63), recent studies with larger samples report reductions that are unrelated to posttraumatic stress (42, 43). Additional neuroimaging findings in PTSD, including amygdala hyperreactivity and reduced medial prefrontal and anterior cingulate response (61), have also been observed in individuals with a history of childhood abuse, including those without PTSD or any psychopathology (43). Studies are clearly needed to ascertain the degree to which these neuroimaging findings are specific to PTSD, are specific to PTSD in the context of a history of maltreatment, or are a more general consequence of exposure to childhood maltreatment.
Similar to findings for depressive illness, a number of genetic polymorphisms appear to modulate risk for PTSD in individuals with a history of maltreatment. The most compelling involves polymorphisms of FKBP5, which regulates cortisol-binding affinity and the nuclear translocation of the glucocorticoid receptor (64, 65). Interestingly, Xie et al. (65) reported that among individuals with the TT genotype of rs9470080, those with no maltreatment history had the lowest risk for PTSD as adults, and those with a maltreatment history had the highest. This suggests that the search for genetic risk factors may be elusive if study subjects are not subtyped by maltreatment history.
The National Comorbidity Replication Study showed that childhood sexual or physical abuse was associated with a 2.03- to 3.83-fold increase in risk for specific phobias, social anxiety disorder, generalized anxiety disorder, and panic disorder with or without agoraphobia (7) (Figure 2C). Childhood adversity accounted for 32.4% of the population attributable risk fraction for anxiety disorders (3). Moreover, exposure to multiple types of childhood adversity increased the likelihood of receiving a prescription for an anxiolytic by twofold (6).
The impact of exposure to childhood maltreatment on the clinical presentation and treatment of anxiety disorders has been understudied. Patients with an anxiety disorder and a history of maltreatment have significantly higher rates of concurrent major depression (37, 66), more significant impairment in social functioning, higher state and trait anxiety scores (66), greater chronicity (37), greater symptom severity, and poorer quality of life (67). Severity increases with the number of types of maltreatment experienced, and emotional abuse and neglect are especially salient risk factors for social anxiety disorder (67, 68). Lastly, in a clinical trial with paroxetine, social anxiety patients with a history of emotional abuse were the most likely to drop out of treatment (68).
Neuroimaging studies in individuals with anxiety disorders, particularly disorders involving intense fear and panic, such as panic disorder, specific phobias, and social anxiety, report evidence for amygdala hyperreactivity, which may stem from underactivity of the prefrontal cortex and insufficient inhibition of the amygdala (69, 70). Overactivation of the insula, a paralimbic region associated with perception of somatic sensations, has also been observed (70, 71). However, as indicated above, heightened amygdala activation has been observed in fMRI studies of adults without psychopathology if they were exposed to childhood maltreatment (43, 46). Moreover, a recent report (72) found that threatening faces produced overactivity in both the amygdala and the anterior insula in maltreated children with normal levels of anxiety. Hence, amygdala and insula findings are not specific to individuals with anxiety disorders. An alternative hypothesis is that enhanced amygdala and insula response to threat emerges as a consequence of exposure to childhood maltreatment and serves as a risk factor for the later development of anxiety disorders.
A substantial body of research shows the important role of maltreatment on risk for drug abuse and dependence (8, 9) (Figure 2D–E), although the nature of the association may be complicated by high rates of substance abuse in maltreating parents and by the possibility of prenatal exposure, prenatal malnutrition, and prematurity. A well-controlled epidemiological and co-twin study of women (8) found that nongenital childhood sexual abuse was associated with a 2.9-fold increase in risk for drug dependence and that sexual abuse involving intercourse was associated with a 5.7-fold increase. Risk was related to the number of different types of maltreatment an individual experienced. Compared with individuals with no adverse childhood events, adults with five or more adverse childhood events are seven to 10 times more likely to report illicit drug use problems, addiction to illicit drugs, and injection drug use (9). The population attributable risk fractions for these outcomes were 56%, 64%, and 67%, respectively (9). Results from the National Longitudinal Study of Adolescent Health and the National Youth Survey provide prospective evidence for a causal relationship between physical abuse and early adulthood substance abuse (73, 74).
A moderate number of studies have reported differences between substance-abusing individuals with and without a history of maltreatment. The maltreated ecophenotype is associated with an earlier age at initiation, a greater likelihood of engagement in risky sexual behaviors (75), a greater risk for recent incarceration (76), greater ratings of psychological distress (77), and a greater risk for comorbid personality disorders (78). Physical maltreatment appears to be a particularly salient risk factor for the development of substance abuse (35) and progression to injection drug use (79).
Substance abusers with a history of maltreatment respond more poorly to treatment, with greater use of substances during treatment and more persistence of substance-related problems after discharge (80–82). Integrative therapies have been developed to address the combined impact of substance abuse and trauma-related psychopathology (83).
Key neuroimaging findings in substance abusers suggest the possibility of a “dopamine deficiency” that may manifest as reduced activation of the ventral striatum (nucleus accumbens) during rewarding or pleasurable tasks (84, 85). Furthermore, deficits in brain regions implicated in salience attribution (the orbitofrontal cortex) and inhibitory control (the anterior cingulate gyrus) may underlie the patterns of compulsive and impulsive behaviors that characterize addiction (86). Although these factors have not been well studied in maltreated individuals, the few relevant studies report reduced sensitivity to reward and decreased basal ganglia response (87), as well as structural and resting blood flow deficits in the ventral striatum, the anterior cingulate, and the orbitofrontal cortex (43, 88, 89). Further research is needed to ascertain whether these deficits are common to substance abusers in general or more specific to the subset with a history of childhood maltreatment.
Could maltreatment be a nonspecific amplifying factor that “tips the balance” so that individuals at hereditary risk for one disorder or another become more likely to express it? In essence, then, could maltreatment act to enhance the “penetrance” of inherited genetic susceptibilities? This could provide an explanation for the elevations in both prevalence and associated comorbidities.
A richer and more compelling alternative is that the myriad possible outcomes of exposure to childhood maltreatment depend on the timing, type, and severity of exposure, plus a host of genetic factors that influence susceptibility and resilience, and an array of protective factors that attenuate risk. Epigenetic modifications in stress-response systems and neurotrophic factors regulating trajectories of brain development may be the driving force producing the various ecophenotypes. We believe that this explanation best accounts for the available data and suggest that psychiatric disorders presenting in individuals with a substantial history of childhood maltreatment be thought of as ecophenotypic variants or ecophenocopies (see the data supplement for strategies for capturing this in our nosology).
Neurobiological Correlates of Childhood Maltreatment
As indicated above, there is a growing body of reproducible findings linking childhood maltreatment to structural and functional brain differences. The most consistent finding is that of alterations in the corpus callosum, characterized by reduced midsagittal area (90–94) or decreased fractional anisotropy (diminished integrity) on diffusion tensor scanning (95, 96) (Table 1). Another reasonably consistent finding is reduction in hippocampal volume in adults (16, 93, 97–105) but not younger children (91, 92, 106, 107) with a history of maltreatment (Table 2). The hippocampus is likely the most stress-sensitive structure in the brain, and translational studies show that stress or glucocorticoids act on the hippocampus to suppress neurogenesis in the dentate gyrus and provoke remodeling of pyramidal cells in portions of the cornu ammonis, particularly CA3. A recent study (105) found that childhood maltreatment was associated with volume reductions in the same subfields in a relatively large population of young adults, suggesting that the same mechanisms may be at work.
TABLE 1.Childhood Maltreatment and Area or Integrity of the Corpus Callosuma
| Add to My POL
|Number of Subjects||Age (Years)|
|First Author (Reference)||Types of Maltreatment||Diagnostic Requirement||Exposed||Comparison||Mean||SD||Gender||Medication||Main Corpus Callosum Findingsb|
|Teicher (90)||Sexual, physical, or neglect||Inpatients with versus without abuse||28||23||12.9||2.9||Both||No||Decrease in regions IV, III; males more affected than females|
|De Bellis (91)||Sexual, physical, or witnessing domestic violence||PTSD versus typical controls||44||61||12.1||2.3||Both||No||Decrease in regions IV, V–VII; males more affected than females|
|De Bellis (92)||Sexual, physical, or witnessing domestic violence||PTSD versus SES-matched controls||28||66||11.5||2.9||Both||No||Decrease in regions VII, IV–VI|
|De Bellis (144)||Sexual, physical, or witnessing domestic violence||PTSD versus SES-matched or typical controls||61||122||11.7||2.6||Both||No||Decrease in regions VII, I, VI; males more affected than females; reanalysis|
|Teicher (94)||Sexual, physical, or neglect||Inpatients with versus without abuse and controls||28||23 inpatients, 115 controls||12.2||3.4||Both||No||Decrease in regions IV, V–VII; males affected by neglect, females by sexual abuse; partial reanalysis|
|Zanetti (145)||Physical or sexual||BPD with versus without physical or sexual abuse and controls||10 (4 without physical or sexual abuse)||20 controls||29.1||9.1||Both||No||BPD versus controls NS; increase in regions V, VII in BPD with versus without abuse|
|Rusch (146)||Sexual||BPD with versus without sexual abuse and controls||20 (10 without physical or sexual abuse)||20 controls||27.6||6.8||Female||No||Decrease in region V in BPD versus controls; decrease in regions V, VI in BPD with versus without abuse|
|Kitayama (147)||Sexual, physical, or witnessing domestic violence||PTSD versus typical controls||9||9||37.3||9.4||Female||Yes||Decrease in region V and total area|
|Jackowski (95)||Sexual, physical, or witnessing domestic violence||PTSD versus typical controls||17||15||10.6||2.3||Both||No||Decreased FA in middle and posterior|
|Andersen (93)||Sexual||No diagnosis required; 27% with history of PTSD||26||17||19.8||1.4||Female||No||Decrease in region III; sensitive period, ages 9–10|
|Carrion (148)||Sexual, physical, or witnessing domestic violence||PTSD symptoms versus controls||24||24||11.0||2.2||Both||Yes||NS 8.7% decrease in region VII|
|Mehta (120)||Early deprivation, 24 months||Romanian orphans versus controls||14||11||16.1||0.8||Both||No||NS 6.5% decrease in absolute volume|
|Teicher (96)||Peer verbal abuse||No psychopathology||63||Used ratings, not groups||21.9||1.9||Both||No||Decreased FA in region VII; males and females affected to the same degree|
|Frodl (149)||CTQ score||Unaffected relatives with major depression, controls||6 relatives, 4 controls||15 relatives, 20 controls||36.3||12.9||Both||No||Decreased FA in region VII controls with versus without abuse; increase in FA in region VII, relatives with versus without abuse|
TABLE 2.Childhood Maltreatment and Structure and Function of the Hippocampusa
| Add to My POL
|Number of Subjects||Age (Years)|
|First Author (Reference)||Types of Maltreatment||Diagnostic Requirement||Exposed||Comparison||Mean||SD||Gender||Medication||Main Hippocampus Findingsb|
|Bremner (97)||Physical or sexual||PTSD versus healthy controls||17||17||41.3||6.6||Female||Yes||L decreased 12%|
|Stein (99)||Sexual||PTSD or dissociative identity disorder versus SES-matched controls||21||21||31.1||6.4||Female||Yes||L decreased 5%|
|Driessen (103)||CTQ score||BPD versus healthy controls||21||21||29.6||6.5||Female||Yes||L, R decreased 16%|
|Vythilingam (16)||Physical or sexual||Major depression with versus without abuse and controls||21||11 major depression, 14 controls||31.4||6.9||Female||No||L decreased 15% in major depression with physical or sexual abuse versus control; NS for major depression without physical or sexual abuse versus control|
|Schmahl (104)||Physical or sexual||BPD with abuse versus comparison without BPD||10||23||30.3||8.0||Female||Yes||L decreased 11%, R 16%|
|Bremner (63)||Sexual||Abuse with PTSD, abuse without PTSD, and controls||10 with PTSD, 12 without PTSD||11||34.9||7.5||Female||No||L, R, decreased 19% for sexual abuse with PTSD versus control; NS for sexual abuse without PTSD versus control|
|Brambilla (102)||Physical or sexual||BPD versus healthy controls||10||20||33.0||8.9||Both||No||L, R decreased 6.8%, most marked in BPD with abuse|
|Pederson (150)||CTQ severe to extreme, pubertal||Abuse with PTSD, abuse without PTSD, controls||17 with PTSD, 17 without PTSD||17||25||6||Female||?||NS 2.8% decrease on L for abuse with PTSD versus control; NS 6.3% decrease on L for abuse without PTSD versus control|
|Vermetten (100)||Physical or sexual||Dissociative identity disorder with PTSD versus comparison||15||23||37.8||9.0||Female||Yes||L, R decreased 19.2%|
|Cohen (114)||ELSQ high versus low, 0–12 years||No psychopathology||122||84||39.9||17.2||Both||No||L, R decreased (p=0.07 and p=0.06)|
|Zetzsche (151)||Physical or sexual||BPD with versus without physical or sexual abuse and controls||14 BPD with 11 BPD without physical or sexual abuse||25||26.7||6.7||Female||Yes||L decreased 5% (p=0.07), R decreased 6% (p=0.03) for BPD versus control; NS for BPD with versus without physical or sexual abuse|
|Andersen (93)||Sexual||No diagnosis required; 27% with history of PTSD||26||17||19.8||1.4||Female||No||Decreased 6.8% bilaterally; sensitive periods, ages 3–5, 11–13|
|Bonne (152)||Sexual, physical, emotional||PTSD with versus without abuse, controls||11||11 PTSD, 22 controls||35.9||10.4||Both||No||Decreased 9% bilaterally for PTSD versus control; NS for PTSD with versus without abuse|
|Weniger (153)||Physical or sexual||PTSD, dissociative disorders, and controls||10 PTSD 13 dissociative disorders||25||32.||7.1||Female||Yes||Decreased 18% bilaterally for PTSD versus control; NS for dissociative disorders versus control|
|Lenze (154)||CECA score||Remitted major depression with versus without abuse, controls||19||12 Remitted major depression, 24 controls||48.5||14.9||Female||Yes||Decreased L for remitted major depression versus control; abuse NS contribution|
|Soloff (155)||Physical or sexual||BPD with versus without physical or sexual abuse, controls||20 with 14 with-out physical or sexual abuse||30||26.6||7.9||Both||No||Decreased R, L for BPD versus control; NS for BPD with versus without physical or sexual abuse|
|Weniger (101)||Physical or sexual||BPD, controls||24||25||32.5||6.5||Female||Yes||Decreased 12% bilaterally (with or without comorbid PTSD)|
|Gatt (156)||ELSQ score||No psychopathology||89||Used ratings, not groups||36.2||12.7||Both||No||Decreased gray matter volume R, L with ELSQ ratings and MET polymorphism of BDNF|
|Frodl (98)||CTQ score||Major depression, healthy controls||43||42||44.1||12.4||Both||Yes||NS gray matter volume; emotional neglect: decreased white matter volume on L in females, L and R in males|
|Thomaes (157)||Physical or sexual||Complex PTSD, controls||33||30||35.5||11.0||Female||Yes||R decreased (p<0.04); R inverse correlation with abuse severity (p<0.02)|
|Landré (158)||Sexual||PTSD, unexposed controls||17||17||24.8||4.7||Female||No||NS|
|Sala (159)||Physical or sexual||BPD, matched controls||15 BPD (6 physical or sexual abuse)||15||33.5||7.9||Both||Yes||R decreased 12.7% for BPD versus control; R, L decreased for BPD with versus without physical or sexual abuse|
|Everaerd (160)||List of Threatening Life Events||No psychopathology; 5HTTLPR genotyping||357||Used ratings, not groups||23.7||5.6||Both||No||Gene-by-abuse-by-gender; decreased R, L for males with S′-allele and severe adversity (p<0.002)|
|Teicher (42)||CTQ and ACE scores||No diagnosis required; 46% exposed history major depression||104||89||21.9||2.1||Both||No||Decreased 6% in L subfields dentate gyrus and CA3; not related to major depression or PTSD|
|Dannlowski (43)||CTQ score||No psychopathology||148||Used ratings, not groups||33.8||10.4||Both||No||R decreased (p<0.05)|
|Carballedo (161)||CTQ score||No psychopathology, with versus without family of history major depression||20 positive, 20 negative family history||Used median split ratings||36.5||13.1||Both||No||Decreased L, R hippocampal heads in subjects with emotional abuse and positive family history|
|Children and adolescents|
|De Bellis (91)||Sexual, physical, or witnessing domestic violence||PTSD versus typical controls||44||61||12.1||2.3||Both||No||NS 2.2% increase|
|Carrion (148)||Sexual, physical, or witnessing domestic violence||PTSD symptoms versus controls||24||24||11.0||2.2||Both||Yes||NS 7.6% decrease|
|De Bellis (162)||Sexual, physical, or witnessing domestic violence||PTSD versus typical controls||9||9||10.6||1.6||Both||Yes||NS at baseline or while followed longitudinally for >2 years|
|Chugani (163)||Early deprivation, mean 38 months||Romanian orphans versus epilepsy control||10||7||10.3||3.9||Both||No||Decreased PET glucose metabolism in L temporal region, including hippocampus|
|De Bellis (92)||Sexual, physical, or witnessing domestic violence||PTSD versus SES-matched controls||28||66||11.5||2.9||Both||No||NS 1.8% decrease|
|Tupler (107)||Sexual, physical, or witnessing domestic violence||PTSD versus SES-matched or typical controls||61||122||11.7||2.6||Both||No||NS gray matter volume; increased white matter volume; reanalysis|
|Carrion (106)||Sexual, physical, or witnessing domestic violence||PTSD symptoms||15||0||10.4||8–14||Both||Yes||Inverse correlation (r=–0.48) between volume and cortisol level over 12–18 months|
|Mehta (120)||Early deprivation, 24 months||Romanian orphans versus controls||14||11||16.1||0.8||Both||No||L, R decreased 16% absolute, NS after adjusted for brain volume|
|Rao (44)||Early life adversity||Major depression, high risk, and controls||30 major depression 22 high risk, 35 controls||Ratings of exposure within each group||14.9||1.8||Both||No||Decreased R, L with early life adversity in high risk and controls; hippocampal volume partially mediated risk for major depression with early life adversity|
|Carrion (164)||Sexual, physical, or witnessing domestic violence||PTSD symptoms versus controls||16||11||13.9||2.0||Both||Yes||Abnormal (decreased) R BOLD response on verbal memory task|
|Maheu (165)||Caregiver deprivation—emotional neglect||Orphans or foster care versus controls||11||19||13.5||2.6||Both||No||Abnormal (increased) L BOLD response to fearful and angry versus neutral faces|
|Tottenham (121)||Early deprivation, 63 months||Orphans versus healthy controls||34||28||8.9||2.1||Both||?||NS 2.5% decrease on L in late adoptees (after 15 months)|
|Edmiston (166)||CTQ score||No psychopathology||42||Used ratings, not groups||15.33||1.37||Both||No||Decreased with total scores on R, L in females; decreased with emotional neglect on R, L in males and females|
|Lupien (122)||Mothers with chronic major depression||Exposed versus controls||17||21||10||Both||No||NS|
There are also associations between exposure to early maltreatment and the attenuated structural or functional development of the neocortex (93, 108–113), including the anterior cingulate (109, 114–116), the orbitofrontal (89, 116, 117) and dorsolateral prefrontal cortex (88, 115), and the visual and auditory cortex (Table 3).
TABLE 3.Childhood Maltreatment and Structure and Function of the Cerebral Cortexa
| Add to My POL
|Number of Subjects||Age (Years)|
|First Author (Reference)||Types of Maltreatment||Diagnostic Requirement||Exposed||Comparison||Mean||SD||Gender||Medication||Main Cortical Findingsb|
|De Bellis (91)||Sexual, physical, or witnessing domestic violence||PTSD versus typical controls||44||61||12.1||2.3||Both||No||Increased prefrontal cerebrospinal fluid (volume loss)|
|De Bellis (109)||Sexual, physical, or witnessing domestic violence||PTSD versus typical controls||11||11||10.2||29||Both||No||Decreased N-acetyl aspartate/creatine ratio in anterior cingulate|
|Carrion (108)||Sexual, physical, or witnessing domestic violence||PTSD symptoms versus controls||24||24||11.0||2.2||Both||Yes||Decreased frontal asymmetry|
|Chugani (163)||Early deprivation, mean=38 months||Romanian orphans versus epilepsy control||10||7||10.3||3.9||Both||No||Decreased PET glucose metabolism in R, L orbital frontal gyrus, infralimbic prefrontal cortex|
|De Bellis (92)||Sexual, physical, or witnessing domestic violence||PTSD versus SES-matched controls||28||66||11.5||2.9||Both||No||Increased prefrontal cerebrospinal fluid (volume loss)|
|De Bellis (167)||Sexual, physical, or witnessing domestic violence||PTSD versus typical controls||43||61||12.1||2.3||Both||No||Increased R, L superior temporal gyrus gray matter volume; reanalysis|
|De Bellis (144)||Sexual, physical, or witnessing domestic violence||PTSD versus SES-matched or typical controls||61||122||11.7||2.6||Both||No||Increased prefrontal cerebrospinal fluid (volume loss); reanalysis|
|Brambilla (102)||Physical or sexual||BPD versus healthy controls||10||20||33.0||8.9||Both||No||NS in temporal lobes and dorsolateral prefrontal cortex|
|Richert (168)||Sexual, physical, or witnessing domestic violence||PTSD symptoms versus controls||23||24||11.0||2.2||Both||Yes||Increased middle inferior ventral prefrontal gray matter volume; reanalysis (108)|
|Cohen (114)||ELSQ high versus low, 0–12 years||No psychopathology||122||84||39.9||17.2||Both||No||Decreased anterior cingulate total volume|
|Kitayama (169)||Physical, sexual||PTSD versus healthy controls||8||13||39.3||8.2||Both||?||Decreased R anterior cingulate volume|
|Andersen (93)||Sexual versus healthy controls||No diagnosis required; 27% with history of PTSD||26||17||19.8||1.4||Female||No||Decreased total frontal gray matter volume; sensitive period, ages 14–16|
|Tomoda (111)||Sexual versus healthy controls||No diagnosis required; most without axis I, II disorders||23||14||19.7||1.4||Female||No||Decreased occipital gray matter volume in BA 17–18; sensitive period, before age 12; partial reanalysis (93)|
|Tomoda (115)||Harsh corporal punishment versus healthy controls||No diagnosis required; most without axis I, II disorders||23||22||21.7||2.0||Both||No||Decreased gray matter volume in dorsolateral, anterior cingulate, and medial prefrontal|
|Carrion (148)||Sexual, physical, or witnessing domestic violence||PTSD symptoms versus controls||24||24||11.0||2.2||Both||Yes||Increased R, L inferior and superior prefrontal gray matter volume; reanalysis (108)|
|Carrion (170)||Sexual, physical, or witnessing domestic violence||PTSD symptoms versus controls||30||15||13.2||2.1||Both||No||Decreased L ventral and inferior prefrontal gray matter volume; inverse correlation between prebedtime cortisol level and L ventral gray matter volume|
|van Harmelen (171)||Emotional abuse or neglect||Major depression or anxiety disorders versus controls||84||97||37.5||10.4||Both||Yes||Decreased L dorsomedial prefrontal gray matter volume, independent of psychopathology|
|Sheu (88)||Harsh corporal punishment versus controls||No diagnosis required; 63% no lifetime history||19||23||21.9||2.1||Both||No||Increased T2 relaxation time (decreased regional cerebral blood volume) in R, L dorsolateral prefrontal|
|Hanson (89)||Physical versus health controls||No diagnosis required||31||41||11.8||1.1||Both||Yes||Decreased R orbital frontal, dorsolateral, temporal, and left and right parietal lobes|
|Frodl (98)||CTQ score||Major depression and healthy controls||43||42||44.1||12.4||Both||Yes||Physical neglect: decreased prefrontal gray matter volume|
|Thomaes (157)||Physical or sexual||Complex PTSD and controls||33||30||35.5||11.0||Female||Yes||Decreased R dorsal anterior cingulate, R orbitofrontal|
|Landré (158)||Sexual||PTSD and unexposed controls||17||17||24.8||4.7||Female||No||NS regional measures of cortical thickness|
|Tomoda (112)||Parental verbal abuse versus healthy controls||No diagnosis required, 48% with history of mood disorder||21||19||21.2||2.2||Both||No||Increased L superior temporal gyrus gray matter volume|
|Edmiston (166)||CTQ score||No psychopathology||42||Used ratings not groups||15.33||1.37||Both||No||Decreased dorsolateral, orbitofrontal, subgenual prefrontal gray matter volume|
|Gerritsen (172)||List of Threatening Life Events||No psychopathology; BDNF polymorphism||568||Used ratings not groups||23.4||5.4||Both||No||Decreased anterior cingulate and medial orbitofrontal at 1.5-T but not 3-T; G×E BDNF versus events for subgenual anterior cingulate|
|Carballedo (161)||CTQ score||No psychopathology, with versus without family history of major depression||20 positive 20 negative family history||Used median split ratings||36.5||13.1||Both||No||Decreased L dorsolateral and medial prefrontal, R anterior cingulate with emotional abuse and positive family history|
|Tomoda (128)||Witnessed domestic violence versus healthy controls||No diagnosis required, 59% past psychiatric history||22||30||21.7||2.2||Both||No||Decreased gray matter volume and thickness in R lingual gyrus (BA 18), decreased thickness R, L V2 and L occipital pole; sensitive period, ages 11–13|
While maltreatment may be associated with alterations in the striatum/basal ganglia (87, 88, 114) and cerebellum (118, 119), most studies have not reported structural differences in the amygdala (91–93, 97, 102, 114). However, increased amygdala volume has been reported in children with institutional deprivation or rearing by chronically depressed mothers (120–122), while smaller amygdala volumes have been observed in adults with childhood trauma and borderline personality disorder or dissociative identity disorder (100, 101, 103, 104). Nevertheless, there is good evidence of enhanced amygdala reactivity in maltreated individuals (17, 43, 46, 72).
Second, there appear to be sensitive periods when these regions are maximally susceptible to the effects of stress. Following this path of inquiry, our group examined the relationship between age at exposure to sexual abuse and observed alterations in brain morphology in a preliminary sample of young adult women (93). We found the hippocampus to be maximally susceptible to maltreatment in women exposed between the ages of 3 and 5 years. However, when maltreatment occurred at ages 9–10, the midportion of the corpus callosum was maximally susceptible, and at ages 14–16, the prefrontal cortex was affected. Thus, there appear to be specific windows of vulnerability in development that determine the negative effects of exposure. These observations are supported by translational research showing that synaptic density in the hippocampus but not the prefrontal cortex of rats is sensitive to the effects of early (preweaning) stress, while the opposite is true with regard to peripubertal stress (123, 124). Rao et al. (125) provided additional support for an early hippocampal sensitive period in humans, reporting that degree of parental nurturance at age 4, but not at age 8, predicted hippocampal volume at age 14.
Third, the effects of maltreatment on brain functioning may not appear immediately after exposure (124). Several studies have reported reductions in the gray matter volume of the hippocampus in adults with a history of maltreatment but not in maltreated children (Table 2). This pattern of results is consistent with translational studies showing that effects of early stress on the hippocampus first emerge during the transition between puberty and adulthood (124). The delay between exposure and neurobiological change may be particularly relevant, as a comparable time lag often occurs between exposure and emergence of depression or posttraumatic stress disorder (126).
Fourth, maltreatment also appears to affect the development of sensory systems and pathways that process and convey the adverse experience. For example, parental verbal abuse is associated with decreased fractional anisotropy in the arcuate fasciculus, which interconnects Wernicke’s and Broca’s areas (127), and with alterations in gray matter volume in the auditory cortex (112). Conversely, witnessing domestic violence is associated with a reduction in gray matter volume in the primary and secondary visual cortex (128) and with decreased fractional anisotropy in the inferior longitudinal fasciculus, which interconnects the visual cortex and the limbic system to shape our emotional and memory response to things that we see (129).
Figure 3 places these findings in context by showing that many of the identified neuroanatomical abnormalities are interconnected and are components of a circuit regulating response to potentially threatening stimuli. Briefly, the thalamus and sensory cortex process threatening sights and sounds and convey this information to the amygdala (130). Prefrontal regions, particularly the ventromedial and orbitofrontal cortex, modulate amygdala response, perhaps turning it down with the realization that something is not actually a threat or, in other cases, irrationally amplifying it (130). The hippocampus also processes this information and plays a key role in retrieving relevant explicit memories (130). The amygdala integrates this information and signals the paraventricular nucleus of the hypothalamus, which in turn regulates autonomic (e.g., heart rate) and pituitary-adrenal hormonal responses and signals the locus ceruleus, which regulates the intracerebral noradrenergic response. The hippocampus, through the subiculum and bed nucleus of the stria terminalis, also modulates paraventricular response, particularly to psychological stressors (131).
FIGURE 3.Neurocircuit Regulating Stress Response to Threatening or Salient Stimulia
a Childhood maltreatment alters development of regions and pathways within this circuit, which serves to reprogram response to subsequent stressors, resulting in either exaggerated or blunted responses. Based primarily on LeDoux (130). ACTH=adrenocorticotropic hormone; BNST=bed nucleus of stria terminalis; PVN=paraventricular nucleus of hypothalamus.
Hence, childhood maltreatment, by affecting the development of key components of this system, reprograms response to subsequent stressors. The influence of maltreatment on autonomic and hypothalamic-pituitary-adrenal response to psychological stressors has been evaluated in a series of studies using the Trier Social Stress Test. Heim et al. (132) first reported that women with a history of physical or sexual abuse had heightened cortisol, ACTH, and heart rate response to stress challenge. Subsequent studies have generally painted a different picture, with evidence emerging for a blunting of cortisol response in adults with a history of maltreatment (133–136). Nevertheless, some individuals show an augmented response, consistent with an enhanced fight-or-flight reaction, and others show a blunted response, consistent with freezing. This divergent pattern of response may be influenced by the type (137) and timing (138) of maltreatment.
Psychosocial Correlates of Exposure
Simultaneous to disruptions in brain development that occur with exposure to mistreatment are alterations in the development of psychological structures. Alterations have been observed in the form of poor self-concept, feelings of worthlessness, and negative views of the world. Furthermore, victims of maltreatment show deficits in what is called deontic reasoning (reasoning about duties and obligations we owe one another), which puts victims at increased risk for future victimization (139). Victims of maltreatment are also more likely to show insecure attachment, associated with diminished expectations of support as well as poor emotion regulation capacities (140).
The first question is whether interventions exist that can reduce a child’s risk of abuse and neglect. The Nurse-Family Partnership has been shown in randomized controlled trials to reduce the incidence of abuse (particularly physical abuse) and neglect of first-born children of high-risk mothers (141). There is also emerging evidence for the efficacy of other interventions against the emergence or reoccurrence of physical abuse. However, no interventions have been shown to be effective in reducing risk for sexual abuse, emotional abuse, witnessing domestic violence, or recurrence of neglect (141).
The second question is whether preemptive interventions exist that can reduce long-term risk for psychiatric illness in maltreated children prior to the emergence of psychopathology. This is an important but largely unexplored area. Third, are there good acute treatments with long-term benefits for maltreated children with psychopathology? Trauma-focused cognitive-behavioral therapy for sexually abused children with symptoms of posttraumatic stress has the most evidence of efficacy (141), but long-term outcome studies are sparse. Assessing and treating parents may also be critical, as maltreatment is often associated with parental psychopathology and parenting problems (26). Recent efforts to develop neurobiologically informed treatments provide preliminary evidence that lower posttreatment cortisol levels may be associated with reduced effects on hippocampal development (106).
Finally, what can be recommended for adults with ecophenotypic variants of major depression, anxiety disorders, substance abuse, or posttraumatic stress? Results of a recent meta-analysis show that depressed individuals with a history of maltreatment respond more poorly to treatment (13), suggesting that standard first-line recommendations for depression may be inadequate for these individuals. The finding that the cognitive-behavioral analysis system of psychotherapy was more effective than nefazodone in maltreated individuals with chronic depression (40) is intriguing, but research is needed to ascertain whether these findings apply to other medications, to other systems of therapy, and to maltreated individuals with less chronic conditions. Integrative trauma-focused treatments have been developed for maltreated individuals with substance abuse that are more helpful than standard treatments, although the results have been far from ideal (83). Childhood maltreatment is often associated with development of insecure attachment patterns (24), and mentalization-based therapy appears to have beneficial effects in patients with insecure attachment patterns across a range of disorders, including major depression, substance abuse, and borderline personality disorder (142). Efforts to reduce allostatic load and inflammation (19) may also be of benefit for maltreated individuals.
Recent recommendations for adults with maltreatment-related posttraumatic stress are to adopt a sequential approach that begins with safety, education, stabilization, skill building, and development of the therapeutic alliance before endeavoring to revisit or rework the trauma, as this may be destabilizing (143). Overall, we suspect that unknowingly mixing maltreated and nonmaltreated subtypes in treatment trials may have left us with an incomplete understanding of risks and benefits. Stratifying study subjects by maltreatment history may provide more definitive insights and delineate a clearer course of action for each subtype.