How might formative trauma exposure account for subsequent psychosis? A principle assumption of biomedical approaches is that the neurological and biochemical abnormalities observed in adult patients have a causal etiological status, with psychological and social conflict being either dismissed as irrelevant, or minimised to the role of catalyst for underlying genetic liability. However a major difficulty with this essentialist framework—what a President of the American Psychiatric Association deemed “a bio-bio-bio model” (
Sharfstein, 2005, p. 3)— is that it ignores the fact that the brain’s primary function is to respond to the environment (
Read et al. 2009). A natural line of enquiry when brain anomalies are identified in any population should be “
What happened to that group to cause this pattern of functioning?” To assume that brain differences exist in a social vacuum, and are solely and causally responsible for schizophrenia, has the same logic as suggesting neural changes during bereavement are the causes of sadness rather than the loved one’s loss. This lack of context—that external events in patients’ lives are largely disregarded—is a serious logical flaw in much neuroimaging research. Indeed, given what is now known about the links between life stress and psychosis, the full extent to which ignoring such variables has jeopardised data integrity is only now becoming fully apparent.
The Traumagenic Neurodevelopmental Model of Psychosis
The traumagenic neurodevelopmental (TN) model of psychosis (
Read, Perry, Moskowitz, & Connolly, 2001) synthesises biological and psychological research to emphasise the similarities between structural and functional abnormalities in the brains of abused children and those of adult patients with psychosis (which, correspondingly, reflect the differences between patients with psychosis and healthy adults, and traumatised and non-traumatised children). A major premise of the TN model is that the heightened stress sensitivity consistently found in patients with psychosis is not necessarily inherited, but caused by formative exposure to abuse and neglect. This is consistent with the original conception of the stress-vulnerability model of schizophrenia (
Zubin & Spring, 1977), which contended that susceptibility to stress could be
acquired “due to the influence of trauma, specific diseases, perinatal complications, family experiences, adolescent peer interactions, and other life events” (p. 109). Therefore, environmental stressors should not be relegated to “triggers” for those genetically predisposed to psychosis, but reconceptualised as causal events.
When the TN model was proposed in 2001, evidence emphasised that detectable neurological differences between individuals diagnosed with schizophrenia and healthy adults could be conceptualised as trauma-related. These included dopamine, serotonin and norepinephrine irregularities, overactivity of the hypothalamic–adrenal–pituitary (HPA) axis, and structural differences, such as cerebral atrophy, hippocampal damage, reversed cerebral asymmetry, and ventricular enlargements. All these features are typically cited to support the disease model of schizophrenia. At least 125 publications, using a range of methodologies, have subsequently provided either indirect support or confirmation of the central premises of the TN model (
Read et al., 2014). The following sections present an abridged summary of this evidence.
HPA Dysregulation. A 2008 review of neurobiological mechanisms in psychosis conclude, a “heuristically useful framework . . . is the concept of ‘behavioral sensitization that stipulates that exposure to psychosocial stress—such as life events, childhood trauma, or discriminatory experiences—may progressively increase the behavioral and biological response to subsequent exposures’ ” (
van Winkel, Stefanis, & Myin-Germeys, 2008, p. 1095 It has long been recognised that adversity exposure can stimulate a cascade of chronic disturbances in the responsivity of the HPA axis (e.g.,
De Bellis et al., 1994;
Heim et al., 2008;
Tarullo & Gunnar, 2006). There is a growing corpus of data outlining associations between psychotic symptoms and hyperreactivity to stress, as evidenced by HPA dysregulation (see
Read et al., 2014 for review). For example, people diagnosed with psychotic disorders who had a history of childhood adversity including emotional maltreatment (
Braehler et al., 2005;CSA, (
Mondelli et al., 2010a) and impaired parental bonding (
Pruessner, Vracotas, Joober, Pruessner, & Malla, 2013) demonstrated greater dysregulation in the stress hormone cortisol, with associations additionally observed between irregularities in cortisol secretion and the severity of positive symptoms (
Belvederi Murri et al., 2012;
Walder, Walker, & Lewine, 2000), disorganized symptoms (
Walder et al., 2000), and cognitive deficits (
Aas et al., 2011;
Halari et al., 2004;
Walder et al., 2000) than those in comparison groups who were not abused. A one-year longitudinal examination of 56 adolescents found that at-risk individuals who transitioned to full psychosis exhibited significantly more elevated cortisol levels than those who did not convert (
Walker et al., 2010).
Other evidence for HPA axis disturbances in psychosis include enlargement of the paraventricular hypothalamic nucleus (PHN), although these findings are currently less consistent.
Tognin et al. (2012) and
Goldstein et al. (2007) have reported increased hypothalamic volumes in people diagnosed with schizophrenia compared to controls, with hypothalamic abnormalities positively associated with anxiety in both studies. However, although some studies have found pituitary alterations in terms of enlargement prior to psychosis transition in high-risk groups (
Garner et al., 2005;
Pariante, 2008;
Pariante et al., 2005), increased volume in chronic psychosis compared to non-patients (
Takahashi et al., 2009), and associations between illness duration and pituitary size (
Habets et al., 2012;
Pariante et al., 2004), other research has not replicated the same morphological pituitary changes in psychosis (
Klomp et al., 2012;
Nicolo et al., 2010;
Upadhyaya et al., 2007).
Structural Cerebral Changes. Other structural changes that may cause HPA axis dysregulation in psychosis include abnormalities in the hippocampus, a cortical region that is essential for consolidating information from short-term to long-term memory. Reduced hippocampal volume is a well-characterized sequala of childhood maltreatment (
Bremner & Narayan, 1998;
Frodl et al., 2010;
Teicher, Anderson, & Polcari, 2012), with imaging studies revealing that early trauma is “associated with remarkable functional and structural changes even decades later in adulthood” (
Dannlowski et al., 2012, p. 286). Hippocampal pathology is also an established and prevalent observation in people diagnosed with schizophrenia, and it has been reported from both post-mortem (neurochemical, genetic, histological, morphometrical) and
in vivo (functional and structural imaging, neuropsychological) investigations (see
Harrison, 2004). Specifically, hippocampal changes in the left cortex may progress after psychosis onset (
Rosoklija et al., 2000;
Shepherd, Laurens, Matheson, Carr, & Green, 2012), with reduced left hippocampal volume associated with hyper-secretion of cortisol (
Mondelli et al., 2010b) and increased stress sensitivity in patients with psychosis (
Collip et al., 2013). In addition, exposure to childhood trauma has been found to significantly predict left hippocampal volume in individuals with first-episode psychosis (
Hoy et al., 2012). Abnormalities consistent with stress-induced pathology have also been detected in hippocampal formations of patients with psychosis (
Rosoklija et al., 2000); there is evidence that this population exhibits abnormally heightened regional blood flow to the left hippocampus in response to cortisol infusion when compared to matched controls (
Ganguli, Singh, Brar, Carter, & Mintun, 2002).
Other cerebral changes associated with childhood trauma are diminished gray-matter volume in the frontal lobes, particularly the anterior cingulate, and the dorsolateral, medial prefrontal, and orbitofrontal regions (
Cohen et al., 2006;
Hart & Rubia, 2012; Thomaes et al., 2012). In terms of subsequent stress-vulnerability, animal models have demonstrated that such morphological changes in the frontal lobes also have consequences for HPA axis regulation (
Holmes & Wellman, 2009). This includes the potential for childhood stress to affect the expression of glucocorticoid receptors in the frontal cortex (
Chiba et al., 2012;
Mizoguchi, Ishige, Aburada, & Tabira, 2003;
Patel, Katz, Karssen, & Lyons, 2008), receptors which mediate the direct effects of hormones secreted in response to stress, and in turn are linked with abnormal inhibitory HPA feedback regulation (
Mizoguchi et al., 2003).
Similarly, loss of gray-matter volume in frontal and prefrontal cortical regions is widely observed in patients with psychosis (
Ellison-Wright & Bullmore, 2010;
Tian et al., 2011;
Williams et al., 2013). This includes pathology consistent with stress-induced changes in the prefrontal cortex (
Benes & Berretta, 2001;
Glantz & Lewis, 2000;
Harrison, 2004) as well as decreases in glucocorticoid receptor expression in the dorsolateral pre-frontal cortex (
Sinclair, Fullerton, Webster, & Weickert, 2012;
Webster, Knable, O’Grady, Orthmann, & Weickert, 2002). A recent study directly comparing abuse exposure and brain volume in 60 patients and 26 matched controls (
Sheffield, Williams, Woodward, & Heckers, 2013) has found that a significant amount of variance in gray-matter volume in psychotic disorders can be accounted for by a history of sexual trauma. The association was not significant for other types of childhood maltreatment, although rates of CSA, physical abuse, emotional abuse and physical neglect were all higher in the patients with psychosis than the healthy controls.
The Dopamine System. The notion of chemical imbalance as an explanation for a range of mental health difficulties has proven to be a popular theory for justifying expenditure on pharmaceuticals rather than expend resources towards psychosocial research and interventions. Correspondingly, the dopamine hypothesis (which posits that the positive symptoms of schizophrenia are partly attributable to disturbed, hyperactive signal transduction in the dopaminergic system), is probably one of the most widely cited theories of psychosis. In clinical terms the hypothesis underlies the widespread administration of neuroleptics (which are claimed to correct dopamine dysregulation), and has been deemed “one of the most enduring ideas in psychiatry” due to its ubiquity as an explanatory illness model (
Howes & Kapur, 2009, p. 562). However, the dopamine hypothesis was crafted from indirect evidence—the seemingly beneficial sedative effects of neuroleptics. On discovering their main mode of action was blocking dopamine receptors, the argument developed that schizophrenia itself must therefore result from dopamine over-activity. Thus rather than designing a therapeutic agent to treat a disorder, a disorder was hypothesised to fit the drug; and, as pointed out by
Jackson (1986), is as logically tenable as claiming headaches are induced by a lack of aspirin.
The role of antipsychotic medication in creating neurological changes will be discussed more fully below, although the drugs themselves have been linked with hyperactivity in the dopamine system (
Snyder, 1974). However, as with the neuroanatomical, there is also evidence that abnormalities in the dopamine system can be explained in terms of psychosocial stress. For example, it is known that childhood adversity may increase sensitivity in the mesocorticolimbic dopamine system (Oswald et al., 2007;
Trainor, 2011;
Wand et al., 2007), and that chronic early-life stress is associated with dopamine overactivity in response to adulthood stress in both animals (
Cabib, Puglisi-Allegra, & Ventura, 2002) and humans (
Pruessner et al., 2010). In this respect, one of the most well-evidenced abnormalities relating to dopamine transmission in psychosis—hyperactivity of striatal dopamine systems, including elevated transmitter release, and heightened presynaptic and extracellular levels (
Howes & Kapur, 2009)—is consistent with research findings in the stress literature. Similarly, the finding that dopamine function in the prefrontal cortex appears lower in individuals diagnosed with schizophrenia relative to controls (
Abi-Dargham et al., 2002;
Meyer-Lindenberg et al., 2002) would support a model based on the effects of chronic stress. An additional finding consistent with a stress-based account of psychotic symptoms is the contention that abnormal activity in the subiculum induces dopamine hyperactivity in the striatum and midbrain (
Lodge & Grace, 2006,
2011). Subiculum function is believed to influence the kind of enhanced threat perception and “aberrant salience” underlying paranoid and delusion presentations (
Roiser, Howes, Chaddock, Joyce, & McGuire, 2013) and is known to be affected by childhood adversity (
Teicher et al., 2012).
Bentall (2009) summarises this more sophisticated approach to understanding the role of dopamine in psychosis by emphasising that dysregulation can be clearly understood when considering one of dopamine’s main functions: anticipation of threat. As such, “the dopamine system becomes sensitized as a consequence of adverse experiences that predate the onset of the illness” (
Bentall, 2009, p. 175).
Neuroanatomy and Antipsychotic Medication
A final relevant factor for interpreting changes in brain morphology in people diagnosed with schizophrenia is the impact of neuroleptic medication. Although largely ignored as a clinical issue during the development of first-generation antipsychotics (
Reveley, 1985), it has been known for several decades that the density of brain tissue decreases relative to medication dosage (
Lyon et al., 1981). In recent years more controlled and technologically sophisticated work has reinforced the claim that the drugs are causally implicated in neurodegeneration; in effect, that long-term medication use may lead to some of the neurological anomalies traditionally ascribed to psychosis itself (
Moncrieff & Leo, 2010;
Smieskova et al., 2009).
It has been posited that antipsychotics can create dopamine over activity (which, in its simplest terms, can be understood as a compensatory effort to overcome the inhibitory effect of the medication:
Snyder, 1974). Postmortem examinations, for example, do not support the dopamine theory of schizophrenia, in that while medicated patients demonstrated elevated levels, drug-free patients exhibited normal concentrations of dopamine (
Haracz, 1982;
Jackson, 1986). A similar lack of results has also been reported in examinations of dopamine metabolites amongst living patients (
Haracz, 1982). A heightened sensitivity in dopamine receptors has been detected, although this in turn is acknowledged to be partly a result of antipsychotic use. For example, postmortem examinations found that increases in K
D dopamine receptors (but not B
max binding sites) were attributable to neuroleptics (
Mackay et al., 1982). Animal models likewise demonstrated that spontaneous hyperactivity of striatal dopaminergic mechanisms was the result of chronic neuroleptic administration (
Murugaiah et al., 1982).
Numerous studies have demonstrated that antipsychotics lead to reductions in gray-matter volume and enlargements in lateral ventricle volumes (see
Weinmann & Aderhold, 2010) and, more equivocally, in the thalamus and the cortex (see
Navari & Dazzan, 2009). Critically (and contrary to the claims of some industry-sponsored studies; e.g.,
Lieberman et al., 2005) both animal models and research with patients with psychosis has also confirmed that this is not an issue limited to older, first-generation drugs. For example, placebo-controlled research with primates confirmed that 18 months of treatment with olanzapine and haloperidol, at equivalent dosages to human therapeutic use, resulted in reduced volumes of between 8% and 11% across all major brain regions (
Dorph-Petersen et al., 2005). Later research, also administering olanzapine and haloperidol to macaque monkeys, likewise found that exposure to both drugs resulted in a 14.6% reduction of gray-matter volume in the left parietal lobe, a lower glial cell number, higher neuron density (
Konopaske et al., 2007), and a significant 20.5% lower astrocyte number (
Konopaske et al., 2008). Longitudinal research with human participants has confirmed that atypical antipsychotic medication contributes to the shrinkage of brain matter (Ho, Andreasen, Ziebell, Pierson, & Magnotta, 2011). A recent meta-analysis of 30 longitudinal MRI studies (representing 1046 people diagnosed with schizophrenia, 780 controls, and a median duration of 72.4 weeks follow-up) not only support this association, but report no effect for either symptom severity or psychosis duration on neuroanatomical abnormalities (
Fusar-Poli et al., 2013). The latter distinction is an important one, as it refutes a model of psychosis as progressive and neurodegenerative; i.e., that underlying disease processes are causing brain shrinkage. A more recent longitudinal analysis also found strong evidence for medication-induced brain changes, but conversely claimed that the latter is linked to psychotic relapse (
Andreasen, Liu, Ziebell, Vora, & Ho, 2013). However, this interpretation of psychosis as a malignant, progressive disease has been disputed (
Zipursky, Reilly, & Murray, 2013), and it should additionally be noted that Andreasen et al. did not clearly distinguish clinical deterioration from drug-induced effects (i.e., as severity increases, dosage is likely to be raised, meaning the two variables are not necessarily independent, despite being treated as such in the paper’s statistical analysis). In this regard, the associations between medication dosage and clinical severity is an important consideration for future research, as while the severity of psychotic symptoms appear to be weakly associated with cerebral shrinkage, they are significantly correlated with medication exposure (
Ho et al., 2011). The difficulty in finding comparison groups who have not been medicated compounds this problem.
Taken together there is strong evidence for inferring that chronic antipsychotic consumption causes serious neurological abnormalities, and that these alterations remain significant when controlling for substance use, and the duration and severity of psychotic symptoms (
Fusar-Poli et al., 2013;
Ho et al., 2011). While pharmaceutical companies claim that such changes are indicative of a disease process necessitating “medical correction” (see
Read, 2013c), other researchers counter that “the effect [brain atrophy] is causal, and not some artefact of an underlying schizophrenia disease process” (
Bentall & Morrison, 2011, p. 172). When revealing the results of the first large-scale longitudinal study (
Ho et al., 2011) to the media, the eminent neuroscientist Nancy Andreasen (
New York Times, 2008) stated that:
The big finding is that people with schizophrenia are losing brain tissue at a more rapid rate than healthy people of comparable age. Some are losing as much as 1 percent per year. That’s an awful lot over an 18-year period . . . Another thing we’ve discovered is that the more drugs you’ve been given, the more brain tissue you lose . . . The prefrontal cortex . . . is being shut down by the drugs. That reduces the psychotic symptoms. It also causes the prefrontal cortex to slowly atrophy.