on the Brain
Broadly and theoretically, during early brain development, neglect leads to deprivation of input needed by the infant brain at times of experience-expectant maturation, while abusive experiences affect brain development at experience-dependent stages. Both early neglect and abuse therefore have the potential to affect subsequent brain functioning. In practice, the distinction between omission and commission within child maltreatment is not as clear cut. Post and Weiss (1997) describe the "worst case scenario", not infrequently encountered, in which young children experience both abuse and neglect. Child abuse is a potent source of stress, and much of the literature on the relationship between child abuse and neglect and neurobiology has examined the effects of stress on the brain (e.g. De Bellis & Putnam, 1994). Deprivation has been found to result in a stress response as well as in deficits in development.
Trauma and abuse are not synonymous. There continues to be a considerable lack of clarity and tight operational definition about what distinguishes traumatic from stressful events (Shalev, 1996).
Most would agree that trauma includes the following ingredients:
Interesting work has shown, using PET scanning, that when patients suffering from PTSD were presented with vivid accounts of their own traumatic experiences and showed autonomic arousal, there was concomitant heightened activity in their right amygdala and associated areas of the temporal and frontal cortex, as well as in the right visual cortex. At the same time, Broca's area in the left hemisphere, the area concerned with language, was "turned off" (Rauch et al., 1996). The authors suggest that this provides a neurobiological illustration of the clinically encountered tendency in PTSD to re-experience emotions as physical states rather than as declarative verbal memories.
Much abuse and neglect, whilst stressful, is not perceived by the child as trauma, if only because of the very predictable and chronic nature of some forms of maltreatment. While PTSD is not infrequently found in children who have been maltreated (De Bellis, Keshavan, et al., 1999), it is by no means an invariable consequence. Van der Kolk and colleagues consider that the diagnosis of PTSD is most appropriate for traumatised adults, and that children develop much more complex reactions, which are not easily subsumed under that diagnosis (van der Kolk, Weisaeth & van der Hart, 1996).
Chronic abuse and neglect are likely to have a pervasive effect on a child's psychological and biological regulatory process rather than to lead to discrete conditioned behavioural and emotional responses such as are found in PTSD (van der Kolk, 1994). However, a set of criteria for diagnosing PTSD in young children under the age of 4 years has been developed and shown to have good inter-rater reliability (Scheeringa, Zeanah, Drell, & Larrieu, 1995). The set of criteria relies on observed behaviours and has been tested prospectively on 12 young children known to have undergone significant trauma, some of which would be clearly regarded as child abuse.
The ensuing section will commence with a discussion of aspects of the stress response to various forms of child maltreatment. Next, other aspects of neurobiological dysfunction in relation to child abuse and neglect, but not reflecting a stress response, will be considered.
Neurobiological Stress Responses to Child Abuse and Neglect
HPA axis responses
Golier and Yehuda (1998) summarise a number of studies mostly suggesting that some years after experiencing trauma, including sexual abuse, lower cortisol levels are found in conjunction with a possibly enhanced negative feedback in the HPA axis. In childhood studies, findings have been found to be discrepant about the response of the HPA axis to maltreatment. Measures include ACTH response to injected CRH, to!al level of cortisol being secreted, and diurnnal variation of cortisol level.
Hart et al. (1995) found that, in comparison to non-maltreated children, a group of 33 maltreated children aged 4 to 6 years, who were also less socially competent, were shown not to have elevated salivary cortisol levels in situations of stress brought on by conflictual social interactions, which suggests a blunting of a blunting of the HPA axis. This lack of HPA responsiveness may have evolved as a protective factor for the brains of these very vulnerable children. Alternatively, it may reflect lack of novelty of, or familiarity with, stressful situations to which these maltreated children have therefore become habituated and do not respond physiologically.
As part of a longitudinal study of sexually abused girls using a self-selected sample of girls of mean age 11 , an average of 5 years after sexual abuse, De Bellis, Chrousos, et al. (1994) found dysregulation of the HPA axis with a blunted ACTH response to ovine corfico-tropin-releasing hormone (oCRH), but without increased cortisol secretion. The 13 girls were no more depressed than a comparison group at the time of the study, although 54% had a history of severely depressed mood and self-harming behaviour. A hypothesised explanation for this finding is that the previous stress of the abuse had caused CRH hypersecretion with consequent adaptive down-regulation of the HPA axis.
Hart, Gunnar, and Cicchetti (1996) have found that 131 9-year-old children who had been, and possibly were still being, maltreated showed slightly elevated afternoon salivary cortisol levels, which may have been related to the anticipatory stress of returning to an abusive home. Those maltreated children who were classified as depressed using the Child Depression Inventory (Kovacs, 1992) showed attenuated morning cortisol levels. The authors point out that maltreatment is not a unitary concept, and that their results replicate those found by Kaufman (1991) for depressed maltreated children. This is in contrast to depressed children who have not been abused, in whom an alteration in HPA axis functioning has not been found.
Further findings have been reported from a more recent study by Kaufman et al. (1997), in which they compared ACTH response to CRH stimulus in three groups of children, mean age 9.5 years. Thirteen were depressed with additional histories of physical and/or sexual and/or emotional abuse; 13 were depressed without a history of abuse; and 13 were normal matched controls. The depressed children were recruited from inpatient and outpatient psychiatric clinics.
Results showed no differences in the cortisol response to the CRH
stimulus between any of the groups. However, while there was no difference between the ACTH responses of the control, depressed non-abused, and depressed abused children who were now living in a stable environment, seven depressed abused children who were continuing to experience active emotional abuse showed a significantly elevated ACTH response.
The authors suggest the possibility, supported by animal experiments, that stress in early life, in combination with acute or ongoing adversity, leads to a raised ACTH response to CRH. In the absence of a comparison group of non-depressed, abused children who continued to experience abuse, the contribution of depression to these findings is not clear. In the De Bellis, Chrousos, et al. (1994) study described above, there was no indication of ongoing adversity and this supports the explanation postulated here.
The picture is further complicated by findings recently reported by De Bellis, Baum, et al. (1999). They found that 18 prepubertal children (mean age 10.4 years) who were suffering from PTSD following previous significant abuse, and who were now living in stable situations, showed increased levels of 24 hour urinary cortisol in comparison to healthy controls (but not in comparison to non-abused anxious controls). The difference in this sample of maltreated children is that they suffered from PTSD, although adults who suffer from PTSD show lower urinary cortisol levels (Yehuda, Kahana, Binder-Brynes, et al., 1995). The authors suggest that their findings may be explained by a lack or immaturity of adaptation of the HPA axis.
The study of the effects on human infants of extreme deprivation, in particular the emotional and tactile care-giving aspects, has been carried out on infants reared in Romanian orphanages. Postulating a stress response, M. Carlson and Earls (1997) report that at the age of 2 years, these children did not show the usual diurnal variation of cortisol levels that was found in home-reared Romanian comparison children. The children in the orphanages had significantly lower morning cortisol levels, which remained elevated relative to the lower afternoon levels, of the comparison children. The deprived, orphanage-reared children also showed significantly poorer performance on the Bayley Scales of Infant Development and the Denver Development Screening Test.
there is evidence of impairment in the HPA axis following, and during continuing experiences of various forms of abuse and neglect, some showing raised levels of cortisol with others showing protective adaptation expressed by the presence of normal levels of cortisol. What is not clear is the implication for later functioning of continued dysregulation of the HPA axis.
Bremner et al. (1997) have reported 12% left hippocampal volume reduction in 17 adult male and female survivors of severe childhood physical and sexual abuse who now suffered PTSD, in comparison with a non-abused group. There was, however, no associated memory impairment.
Stein, Koverola, Hanna, Torchia, and Mclarty (1997) similarly found statistically significant left hippocampal volume reduction (5%) in 21 women who reported experiences of severe childhood sexual abuse, in comparison to 21 subjects who had experienced no childhood abuse. Seventy-one per cent of the subjects suffered from PTSD and 71% from a dissociative disorder. Severity of the latter correlated positively with degree of hippocampal volume reduction.
Interestingly, there was no associated impairment of memory functioning. The respective researchers, as well as Sapolsky (1996), point out that, for understandable reasons, there is currently no conclusive evidence for a causal explanation in the direction of abuse leading to hippocampal damage, presumably from excess cortisol secretion brought on by the stress of the abuse, although results from many animal experiments, including primates as well as rodents, would support this possibility.
The relationship between hippocampal atrophy and adult short-term memory recall remains unclear. As Stein et al. ( 1997) suggest, it is possible that the relatively small magnitude of the hippocampal damage found in adult survivors of childhood abuse does not impair explicit memory, or may be compensated for during childhood development by the process of neural plasticity. This would explain why trauma (e.g. military combat) first encountered in adult life, when neural plasticity is no longer functioning sufficiently, will affect short-term memory (Bremner et al., 1997).
Differential hippocampal response to stress at different stages of development might also explain why the left and right hippocampi, respectively, were found to be affected by early and late trauma.
What this does not explain is why in one of t studies (Bremner et al., 1995) there was short-term memory recall deficit in adulthood following childhood abuse. The association between hippocampal atrophy and adult PTSD following either childhood or adult trauma may be explained by the suggestion that meta-memory, the integrative function of the hippocampus, particularly for emotionally charged memories of trauma, may be affected as reflected by the partial hippocampal atrophy, leading to the phenomena associated with PTSD.
Although there is evidence of damage or shrinking of the hippocampus following traumatic events, amnesia associated with hippocampal damage is usually confined to subsequent events, while retrograde memory, that is, memory for events preceding the hippocampal damage, is spared (Squire, 1992). Also, interestingly, Schacter , Koustaal, and Norman (1996) report a series of 22 women who experienced prolonged and severe child sexual abuse and who showed reduction of volume of the left hippocampus on MRI (magnetic resonance imaging). Yet, both their memories for the abuse and their current memory functioning were intact.
Urinary catecholamine excretion was measured by De Bellis, Lefter, Trickett, and Putnam (1994) in the longitudinal study of sexually abused girls mentioned above (De Bellis, Chrousos, et al., 1994). Twelve sexually abused girls were found to secrete significantly greater amounts of homovanillic acid, a metabolite of dopamine, than a comparison group, indicating higher catecholamine activity. Interestingly, only one girl met DSM-III-R criteria for PTSD. In their further study reported above, De Bellis, Baum, et al. (1999) also showed that abused children suffering from PTSD excreted significantly greater concentrations of baseline noradrenaline and dopamine in comparison to non-abused anxious and healthy controls, measures correlating significantly with duration of abuse. These findings are presumed to indicate an enduring stress response. Increased levels of urinary catecholamine excretion have been found in adult PTSD (Southwick et al., 1993).
Perry (1994) has reported decreased platelet adrenergic receptors in a small group of children with PTSD following serious abuse, a finding suggesting downregulation of the peripheral adrenergic receptors in response to higher levels of circulating catecholamines. Perry also found increased resting heart rate, and abnormal return of heart rate to baseline levels after an orthostatic challenge, suggesting an overactive sympathetic nervous system in these children.
These findings of a continuing effect on catecholamine functioning following abuse primarily apply to children with PTSD. Whether these effects would be found without the PTSD and whether they abate following treatment is less certain. In his uncontrolled trial of medication using clonidine, Perry found an improvement in arousal symptomatology as well as in basal heart rate in the children suffering from PTSd. However, recent guidelines for treatment of children and adolescents with PTSD state that "there is no empirical support for the use of any particular medication to specifically treat PTSD in children" (Cohen, 1998).
In psychiatrically hospitalised boys who had experienced significant neglect with or without abuse in the first 3 years of life, an association has been found with reduced levels of plasma dopamine beta hydroxylase (DβH) (Galvin et al., 1991). This enzyme is involved in the conversion of dopamine to noradrenaline. The reduced blood level of the enzyme, which is correlated with DβH level in the cerebrospinal fluid (CSF), is believed to be long-lasting. Galvin et al. postulate that the early neglect and abuse which these boys suffered led to an overstimulation of the noradrenergic system due to the stress response with enzyme induction. Subsequent reactive repression of enzyme activity leads to the findings of lowered DβH level.
Rogeness and McClure (1996) report similar findings in a group of children (mean age 10.5 years) hospitalised for psychiatric treatment. They found reduced plasma DβH levels as well as lowered blood pressure in association with a history of early neglect, with or without abuse.
These findings bear similarities to those found in rhesus monkeys. Infant monkeys deprived of their mother's care were found to have lower levels of cerebrospinal fluid (CSF) noradrenaline and its metabolites than control monkeys (Higley, Suomi & Linnoila, 1992; Kraemer, Ebert, Schmidt, & McKinney, 1989).
Galvin, Stilwell, Shekhar, Kopta, and McKasson Goldfarb (1997) have gone on to show that among psychiatrically hospitalised boys, those maltreated before the age of 3 years, and those who had lower serum DβH, showed more interference with an aspect of conscience development, namely the valuation of authority- and peer-derived rules of conscience.
These difficulties correspond closely with essential features of conduct disorder. In fact, the same group of researchers had previously reported an association between low serum DβH and undersocialised conduct disorder, as well as with a history of early maltreatment, in a group of boys hospitalised in a psychiatric hospital (Galvin et al., 1995). The authors postulate that the findings regarding processing of values of conscience requires an intact behavioural inhibition system that is, in part, mediated by noradrenaline.
Some traumatic forms of child abuse and neglect are repeated experiences. This includes the witnessing of domestic violence, being the recipient of direct physical threat or abuse, and being the victim of some forms of sexual abuse. Following maltreatment, some aspects of reminders of the experience become cues capable of evoking apprehension and fear, even when not accompanied by or following the trauma.
Fear is perceived in the amygdala (LeDoux, 1994). The recognition of these cues as danger signals is conveyed to the amygdala, which sets off a fear response (Gallagher & Chiba, 1996).
Children might thus become hyper-aroused, experiencing raised heart rate and feelings of anxiety when faced with an apparently neutral occurrence or stimulus, which might include a particular smell or a sound associated with the abuse. The children are often not actually aware of the source of the arousing cue. Moreover, their threshold to reacting to stress is lowered, and their reaction shows sensitisation to early indicators of potentially frightening or threatening cues (Perry, Pollard, Blakley, Baker, & Vigilante, 1995).
Some of these children show behaviour that symptomatically resembles ADHD and which is a manifestation of post-traumatic arousal. This hypervigilance and hyperactivity is seen equally commonly in girls and boys who have been abused, unlike ADHD without trauma, which is far more common in boys. The extent to which this behaviour is the same entity as the commonly encountered disorder is not clear .
When interacting with their depressed mothers, infants have been observed to be less active, vocalize less, and show more gaze aversion than a comparable group of infants of nondepressed mothers. Dawson, Hessl, and Frey (1994) also found increased heart rate, a physiological measure of being stressed, in infants of depressed mothers when the infants were engaged in social interaction, either with their mother or with another adult.
This physiological stress response could be the result of both increased sympathetic adrenergic activity or lowered vagal (parasympathetic) tone, which was also noted by Field, Pickens and Fox ( 1995). In the infants of depressed mothers, the physiological stress response was not, however, associated with behaviourally observable distress, in contrast to infants of nondepressed mothers who do show arise in heart rate when displaying negative affect. Stress hormones, both cortisol and noradrenaline, have been found to be elevated in depressed mothers and infants (Field, 1998). Infants of depressed mothers also show less distress when separated from them.
Parasympathetic nervous system response
A very different reaction to trauma, which is found particularly in very young children and in girls, is described by Perry et al. (1995). Here, the child faced with a frightening experience, and unable to escape, "resorts" to dissociation, disengaging attention from the present reality.
In this process the vagus nerve, the parasympathetic part of the autonomic nervous system, becomes activated, leading to a slowing of the heart rate and a fall in blood pressure. It is postulated that as part of the dissociation process, endogenous opiates, associated with dopamine systems that arise in the brain stem and are activated by stress, alter the perception of painful stimuli (Abercrombie & Jacobs, 1988).
As with the arousal response, children will later react to cues or apparently minor reminders of the past traumatic experiences by dissociating. Clinically they are described as having brief, blank, and inattentive spells, which interfere with their functioning.
Kaufman et al. ( 1998) also measured serotogenic functioning in their study reported above (Kaufman et al., 1997), and found a raised level of
plasma prolactin secretion in 10 depressed-abused children. compared to depressed-nonabused and normal controls. The findings held both for children who were no longer being abused and for those still being emotionally abused. It is therefore difficult to determine whether this finding is considered to be part of a stress response to previous abuse or whether it is associated with familial and experiential factors as well as with current psychopathology.
Other Neurobiological Responses to Child Abuse and Neglect
Using MRI scans, De Bellis, Keshavan. et al. ( 1999) have also recently measured the size of brain lateral ventricles and various brain structures in 44 maltreated children suffering from PTSD and 61 healthy, nonabused controls, individually matched for age. size, Tanner stage, gender, and handedness. Their mean ages were 12 years. Some of the children had also participated in the study reported by De Bellis, Chrousos. et al. (1994).
Compared to the controls, the children with PTSD were found to have 7% smaller cerebral volumes ; the total midsagittal area of the corpus callosum, particularly the middle and posterior regions, were smaller ; and total lateral ventricles and cortical and prefrontal cortical CSF volumes were larger.
The brain volume was correlated positively with age of onset of PTSD and negatively with duration of the abuse. The volumes of the left and total lateral ventricles correlated positively; and the area of the total corpus callosum correlated negatively with duration of abuse. PTSD symptomatology correlated positively with ventricular volume. and negatively with intracranial volume and total corpus callosum midsagittal area.
The authors discuss possible explanations for their findings, suggesting that these brain size changes could be due to the traumatic early childhood experiences, and therefore possibly related to increased catecholamine concentrations (Simantov et al.. 1996; Todd. 1992) and possibly raised cortisol levels. Impoverished environments and lack of mental stimulation in the early lives of these children could also be contributing factors. As the authors show and discuss. there is a positive correlation between IQ measures and brain size. An important implication for these abused children with PTSD is, therefore, an impairment in their cognitive functioning.
It is important to note that this childhood study did not find a predicted decrease in hippocampal volume, in contrast to adult survivors of childhood abuse. De Bellis, Lefter, et al. ( 1999) discuss explanations for the differences in hippocampal volumes between children and adults, both of whom had suffered childhood abuse and PTSD. Differences may be due to differences in methods of measurement.
In preclinical rat studies, age-dependent changes in sensitivity to the neurotoxicity of glutamate receptor blockers have been found such that cell death in corticolimbic regions did not occur in immature animals, but increased towards adulthood (Farber et al.. 1995). Moreover, in human adults the extent of hippocampal atrophy following major depression correlated positively with the total life-time duration of depression (Sheline, Wang, Gado, Csernansky, & Vannnier, 1996). These findings indicate that brain changes depend both on maturational factors and course of illness.
Functional, structural, and chemical changes in the brain
There have been diverse examples of an association between functional, structural, and chemical changes in the brain, and a history of child abuse and neglect, several of which are summarised here.
Particular relationships have been found between the ERP responses of maltreated and non-maltreated children to affect-containing stimuli. Specifically, pictures of happy and angry faces interspersed between neutral pictures were shown under 2 conditions to 23 maltreated and 21 non-maltreated children aged 9 years. The children were instructed to press a button when seeing either angry or happy faces.
While being as accurate in their responses, the maltreated children were slower. Unlike the nonmaltreated children, they were also found to have higher amplitude of ERPs when instructed to respond to the angry, rather than the happy, faces (Pollak, Cicchetti, Klorman, & Brumaghim, 1997).
The implication is that being presented with negative affect carries a different meaning to maltreated children and elicits a physiologically measurable different response. This may be adaptive in abusive situations but may become maladaptive when used indiscriminantly in different settings.
In a group of 115 psychiatric inpatient children, a significant association has been shown between a history of child physical, sexual, or psychological abuse, and EEG abnormalities in the left side of the frontal and temporal region of the brain (Ito, Teicher, Glod, & Ackerman, 1998; Ito et al., 1993).
This follows an earlier study by the same group which showed that, using a questionnaire designed to test limbic system dysfunction in adults, adult psychiatric outpatients with a history of childhood physical or sexual abuse had a significantly higher score, indicating limbic system dysfunction (Teicher, Glod, Surrey, & Swett, 1993). The limbic system includes the amygdala, hypothalamus, hippocampus, and prefrontal cerebral cortex.
There are early and continuing EEG changes in both infants and their depressed mothers. The pattern is one of relatively increased right frontal lobe activity, which has been recorded from 1 week of age, through to 3 years of age (Jones, Field, & Fox, 1997).
Similar EEG changes have also been shown by Dawson, Frey, Panagiotides, Osterling, and Hessl (1997). They describe the EEG changes in 13-15-month-old infants of depressed mothers as relatively decreased left frontal lobe EEG activity. The authors point out that the left frontal lobe is specialised for the expression of positive emotions such as joy, whereas the right frontal lobe is concerned with sadness and other negative emotions. The right frontal region is specifically activated during withdrawal-related negative affective states (Davidson, Ekman, Saron, Senulis, & Friesen, 1990).
A preliminary study with 10 non-patient adult subjects who had experienced early trauma and 10 comparison subjects found that the traumatised subjects showed greater right-sided brain activity when recalling a traumatic memory than did the comparison group (Schiffer, Teicher, & Papanicolaou, 1995). Although the numbers were small, the authors speculate that their findings indicate that traumatic memories may be preferentially stored in the right cerebral hemisphere.
In comparison to depressed and normal children, sleep disturbances have been found in prepubertal children who had been physically and sexually abused and who had been referred for inpatient or outpatient psychiatric treatment. The disturbances included difficulty in falling asleep and increased activity during sleep.
The disturbance was worse in the physically abused children and was found to be related to the abuse rather than to PTSD (Glod, Teicher, Hartman, & Harakal, 1997).
Much of what has been discussed above about the deleterious effects of trauma, abuse, and neglect on the developing brain and subsequent child development has been amplified descriptively in intelligible lay language in a new book by Karr-Morse and Wiley (1997). The book is replete with reputable references as well as quotes from extensive conversations with researchers in the related fields.