Presentation and course of illness in women

Symptom expression at presentation has important implications for a number of reasons, such as determining treatment regimens and understanding course of illness. Gender differences in clinical presentation and course have been reported as broadly consistent in different countries and cultures (Hambrecht et al., 1992; Goldstein, 1997; Harrison et al., 2001). One of the largest and most comprehensive epidemiological, population-based samples is from the National “Low Prevalence” study from Australia (Morgan et al., 2008). This is an incident sample that includes 1,090 new cases of psychotic disorder: at presentation, women were more likely to have depressive symptoms and less likely to show negative symptoms. Women may also have higher levels of depressive symptoms than men throughout disease progression (Goldstein & Link, 1988; Castle et al., 1993). The prominence of affective symptoms in mentally ill women overall may represent the likelihood that women are more likely to express affective symptoms than men overall, and therefore sex differences in symptom expression in schizophrenia may be related to gender differences per se (in illness expression) rather than schizophrenia per se (Flor-Henry, 1983; see also Chapter 19).

During first illness episodes, levels of negative symptoms are generally lower in women (Vazquez-Barquero et al., 1996; Leung & Chue, 2000; Morgan et al., 2008), but this is not always found to be the case (e.g., Salokangas’ first contact sample aged 15–44 [Salokangas, 1997a,b] and the ABC cohort [Hafner et al., 1993]). Specific positive symptoms such as paranoia or persecutory delusions (e.g., Goldstein and Link, 1988; Goldstein, 1997; Hambrecht et al., 1992) and auditory hallucinations (Tien, 1991; Rector & Seeman, 1992) have been found to be higher in women in some studies. However, the Australian Low Prevalence study (Morgan et al., 2008) suggested that if positive symptoms were defined as an overall score, women presented with as many positive symptoms overall as did men.

However, the problem with prevalence data is that they may be less reliable for determining symptom burden at true illness onset. Drake et al. (2015) addressed this using a large first-episode sample. Consistent with previous studies, they reported that early onset women showed worse overall PANSS (Positive and Negative Syndrome Scale) scores (negative and cognitive symptoms) than later onset cases and that women showed significantly worse depression-anxiety scores. Overall, independent of age at onset, women presented with fewer negative symptoms and more mood symptoms. The association of older age at onset with fewer negative and cognitive symptoms at presentation among women was also consistent with other findings (e.g., Sartorius et al., 1986; Morgan et al., 2008; Grossman et al., 2008). Paranoid symptoms may be more severe in women with older onset than men with older onset (e.g., Castle et al., 1993; Häfner et al., 1993).

In most studies, women with schizophrenia have a better prognosis over 2–10 years than men for many measures (e.g., Angermayer et al., 1990; Jablensky et al., 1992; Häfner et al., 1993; Robinson et al., 1999); though some studies, often smaller ones, find no differences (e.g., McCreadie et al., 1989; Rajkumar & Thara, 1989; and see Angermeyer et al., 1990) and studies that control for premorbid adjustment and baseline symptoms often find these mediate outcome, especially social measures (e.g., Salokangas, 1997b). In general, earlier age of illness onset predicts worse negative and disorganization symptoms and a more malign symptomatic course (e.g., Jablensky et al., 1992; Morgan et al., 2008; Grossman et al., 2008), albeit Selten’s (2007) group did not find this to be the case in a sample restricted to presentations under age 55.

The relationships between age at first presentation, duration of illness, symptoms and gender are complex. Thus, some suggest there are minimal differences between women and men with schizophrenia in familial cases (see Leung & Chue, 2000) or after adjustment for premorbid function (Jablensky & Cole, 1997). In keeping with this complexity, Drake and colleagues (2015) found that although age of onset and sex predicted presenting symptoms, neither predicted medium-term deficit, psychotic, cognitive/disorganized, excitement and dysphoric symptom outcomes after adjusting for baseline scores.

To summarize, at onset, older age and affective symptoms predict better course in women in most studies. Young onset, negative and disorganized symptoms predict worse short- and medium-term outcomes and are less associated with women. In clinical practice, symptoms at presentation still have a role as reliable and obvious indicators of likely course in women (and men). Illness course over the first two years also appears to mediate the effect of sex on long-term outcomes (Harrison et al., 2001), so after this stage, illness history may be a better guide than demography.

The brain in women with schizophrenia

As a neurodevelopmental disorder, it is likely that schizophrenia originates during fetal and early postnatal life and therefore it is also likely that sex differences in early differentiation of the brain during fetal and early postnatal life are key in understanding sex differences in schizophrenia (Goldstein and Walder, 2006). This premise has support from animal studies demonstrating brain abnormalities and behavioral consequences, depending on the timing of the insult during fetal and early postnatal brain development, for female compared with male animals (Goldman et al., 1974; Grimm and Frieder, 1985). Human studies are also consonant with the animal literature (Rantakallio and Wendt, 1985; Goldstein et al., 2014).

Early studies of sex differences in structural brain abnormalities in schizophrenia generally found less pervasive brain abnormalities than men with schizophrenia, although not wholly consistent (Nopoulos, Flaum et al., 1997). (Nopoulos, et al., 1997). More recent work reports region-specific structural brain abnormalities in women with schizophrenia. Some studies have reported smaller volumes of heteromodal association areas among women with schizophrenia than men (e.g., dorsolateral prefrontal cortex and superior temporal gyrus [STG] and orbital prefrontal cortex). Others found smaller volumes of STG in men, and similar abnormalities in women and men in dorsolateral prefrontal cortex (Gur, et al. 2000). Studies have demonstrated varied differences between women with schizophrenia compared with their healthy controls, depending on the prefrontal region assessed (e.g., Gur, et al. 2000; Goldstein, et al. 2002) and inconsistencies across studies may be, in part, due to methodological and sample size differences.

Goldstein and colleagues (2002) reported that structural brain abnormalities occur in nonpsychotic offspring of parents with schizophrenia in a number of brain regions found to be normally sexually dimorphic, and differentially abnormal in adult women and men with schizophrenia. Szeszko et al. (2002) also provided evidence for differential sex effects in brain abnormalities during the premorbid period using MRI and neuropsychological tests in first episode patients. That study reported an association between anterior hippocampal volume and executive and motor functioning in male patients, which was not present in women. First episode studies have been important in taking account of the differing age at onset and treatment exposures between the sexes (see Drake et al. 2015).

There has been an increase in work on sex differences in the brain in schizophrenia over the past 5 years. Earlier studies of sex differences in brain abnormalities in schizophrenia focused primarily on analyses of gray matter volumes. However, emerging literature on sex differences in white matter tracts has been possible with the development of diffusion tensor imaging (DTI) (Kunimatsu, et al., 2012; Savadjiev, et al., 2013). Additionally, the field has become more accepting of the importance of sex differences in schizophrenia and thus increasing numbers of functional brain imaging studies are characterizing sex differences in prefrontal cortical functions in more refined ways (i.e., as brain-clinical phenotypes), (Elsabagh, et al., 2009; Jimenez, et al., 2010; Abbs, et al., 2011; Mendrek, et al., 2011). The role of estradiol and genes in understanding these sex differences is also being explored (Elsabagh, et al., 2009; Irle, et al., 2011; Chen, et al., 2014).

Recent structural analyses of white matter found sex-dependent orbitofrontal white matter density and sulcogyral patterns were related to symptomatology (Uehara-Aoyama, et al., 2011; Joshi, et al., 2012) and an increased diffusion coefficient in the anterior cingulum bundle in men with schizophrenia versus male controls related to negative symptoms that was not found in women with schizophrenia (Kunimatsu, et al. 2012). However, an increased diffusion coefficient in right anterior cingulum and left fornix was found in women with schizophrenia compared with female controls and not between schizophrenia and control males (Kunimatsu, et al. 2012), suggesting a more complex pattern of sex-dependent findings when laterality was taken into account.

Structural findings have continued to underscore sex differences in hippocampal volume, but extended this by considering duration of illness: reduced hippocampal size in men with schizophrenia compared with women did not progress with illness duration, but reduced hippocampal size was found only in women with schizophrenia with long illness duration (Irle, et al., 2011). In a more refined structural analysis of networks of brain regions associated with particular functional domains, covariances of brain regions implicated in verbal memory circuitry (hippocampus, inferior parietal, dorsolateral prefrontal [DLPFC] and anterior cingulate cortices) were consistent with greater abnormalities in associations of hippocampus with anterior cingulate gyrus and DLPFC, and between inferior parietal cortex and prefrontal cortex, with better verbal memory performance in females with schizophrenia than males (Abbs, et al. 2011).

However, not all studies have been consistent. Two studies of early-onset schizophrenia (Thormodsen, et al., 2013; Weisinger, et al., 2013) reported a lack of structural abnormalities by sex in cortical thickness (Thormodsen, et al., 2013; Weisinger, et al., 2013) and insula volume (Shepherd, et al., 2012). Also, in childhood-onset schizophrenia (COS), the hippocampus, caudate, amygdala, thalamus, putamen and pallidum were not found to be sexually divergent (Weisinger, et al., 2013). Lack of sex differences in early onset or COS is consistent with the lack of sex differences in incidence of COS suggesting other factors (e.g., gonadal hormones); genes may “override” mechanisms usually regulating sex differences in brain development.

Methodological variation may contribute to the inconsistencies in results. Most studies are small; a particular concern in studies with unbalanced sex ratios (e.g., Uehara-Aoyama, et al., 2011; Joshi, et al., 2012); many studies represent a mix of adult and early-onset samples (Savadjiev, et al., 2013; Thormodsen, et al., 2013; Weisinger, et al., 2013) including a wide range of illness duration. These factors have been found to influence the direction of sex-dependent effects, for example hippocampal size (Irle, et al., 2011; Weisinger, et al., 2013) and sex-dependent effects on fronto-parietal working memory related brain activity in schizophrenia (Elsabagh, et al., 2009). In addition, analyses have not necessarily compared women and men directly, reporting sex-stratified results alone (Kunimatsu, et al., 2012). This may explain the failure to replicate sex-dependent findings in the corpus callosum (Kunimatsu, et al., 2012; Savadjiev, et al., 2013). Finally, when additional clinical phenotyping (such as negative and positive symptomatology) has been included, sex differences in brain-clinical phenotypes have emerged (Mendrek, et al., 2011; Uehara-Aoyama, et al., 2011; Kunimatsu, et al., 2012; Manuseva, et al., 2012).

Functional studies demonstrating sex differences in brain abnormalities have included electrophysiological event-related potential (ERP). Overall, studies reported lower amplitudes in all four components (P100, N170, N250, P300) in men with schizophrenia compared with women (Lee, et al., 2010; Jung, et al., 2012); with respect to N170 and N250, findings corresponded to source activities in superior temporal gyrus, middle temporal gyrus, insula and inferior frontal gyrus (Jung, et al., 2012). Further, when depressive symptoms in schizophrenia were examined, there was a positive association between depression and early perceptual processing in response to novel stimuli in men and later processing stage abnormalities associated with parietal activity in women (Sumich, et al., 2014). Significantly greater QEEG beta-type amplitudes over left central, temporal, parietal and occipital regions and theta frequencies over left frontal and temporal regions have been reported in women with schizophrenia compared with men with schizophrenia and correlated with greater symptomatology (Manuseva, et al., 2012). These findings underscore the importance of brain-clinical phenotyping in characterizing sex-dependent functional brain abnormalities in schizophrenia.

This is also highlighted by functional fMRI studies of specific cognitive and emotional tasks in schizophrenia. In general, fMRI tasks of visual-spatial ability (mental rotation; Jimenez, et al., 2010), object location memory (Shipman, et al., 2009) and phonemic dichotic listening (Hahn, et al., 2011) demonstrate better performance and fewer brain activity deficits in women than men with schizophrenia. However, in a standard n-back fMRI task of working memory, duration of illness further influenced sex-dependent effects on fronto-parietal working memory activity with reduced DLPFC and inferior parietal cortices in men and reduced inferior frontal and superior temporal gyri activity in women (Elsabagh, et al., 2009). Sex differences in visuospatial performance and pattern of brain activation comparing schizophrenia patients with controls suggested a diagnosis-by-sex interaction in accuracy and reaction time. Thus, men with schizophrenia performed significantly worse on a 3-D mental rotation task with brain activity deficits in precuneus, prefrontal and parietal cortex (Jimenez, et al., 2010). Women with schizophrenia perform better on object-location tasks, but there was a significant interaction of sex by diagnosis for object-exchange and object-shifting tasks, with no sex differences on delayed-recognition tasks (Shipman, et al., 2009).

Brain activity associated with emotional stimuli also varied by sex, dependent on hormonal responses (Goldstein, et al., 2015) and symptomatology (Mendrek, et al., 2011). Using a visual stress challenge (negative affective stimuli versus neutral stimuli) while uniquely collecting blood throughout functional magnetic resonance imaging (fMRI), males with psychosis (schizophrenia and bipolar psychosis) showed hyperactivity across all hypothesized stress response regions, including hypothalamus and anterior cingulate cortex, whereas females showed hyperactivity only in hippocampus and amygdala and hypoactivity in orbital and medial prefrontal cortices. Hypercortisolemia was associated with hyperactivity in prefrontal cortices in men and hypoactivity in prefrontal cortices in women with psychoses. Findings suggested disruptions in neural-hormone associations in response to stress are sex-dependent in psychosis, particularly in prefrontal cortex, a finding that was shared with schizophrenia and bipolar psychoses (Goldstein, et al., 2015). In an alternative strategy for eliciting emotion response circuitry, during exposure to sad versus neutral film excerpts, there was an inverse correlation between positive symptoms and activity in hippocampus and parietal and occipital cortices among women, while men with schizophrenia exhibited prefrontal, temporal and anterior cingulate cortices and caudate and cerebellum deficits that were positively correlated with negative symptoms (Mendrek, et al., 2011).

Sex differences in brain-clinical phenotypes also vary by genotype (Zhang, et al., 2011; Radulescu, et al., 2013). Some of these genes have been implicated in glutamatergic and dopamine pathways (Martins-de-Souza, et al., 2010; O’Tuathaigh, et al., 2010; von Wilmsdorff, et al., 2010; Bychkov, et al., 2011; Holley, et al., 2013; Bertholet, et al., 2014; Chen, et al., 2014). Using a genetic knock-out (KO) mouse model of a dopamine-pathway signaling gene, greater deficits were reported in male KOs compared with female KOs in locomotion and the striatum, but this could be reversed with 17β-estradiol (Chen, et al., 2014). These studies suggest that genes, hormones and clinical presentation should all be considered as we refine our understanding of sex differences in brain abnormalities in schizophrenia.

Gonadal hormones and symptoms in women with schizophrenia

The majority of women develop psychotic illnesses like schizophrenia in early adulthood. This coincides with the development of a mature (adult) hypothalamic-gonadal axis. Here, we consider a range of ways in which the reproductive and stress axes may influence an illness like schizophrenia in women. First, there is some evidence that psychotic symptomatology varies with menstrual phase, such that higher symptom levels have been associated with lower estrogen levels (Endo et al., 1978; Huber et al., 2004; Bergemann et al., 2002). In addition, low estrogen phases of the menstrual cycle have also been associated with higher rates of hospital admission (Huber et al., 2004) and with poorer cognitive performance, in particular verbal and spatial memory, and perceptual-motor speed (Hoff et al., 2001). Second, disruptions in a number of endocrine axes, whatever the cause, are unlikely to be specific to schizophrenia, but they may have important consequences for the nature and course of cognitive and other functional deficits in the disease process. Third, during episodes of relapse and worsening psychosis, women with schizophrenia may be more likely to lead chaotic lifestyles, with poor nutrition and self-care. This may also coincide with oligo- or amenorrhea. It is clear that many women with schizophrenia suffer menstrual disruption over varied lengths of time.

Fourthly, prolonged amenorrhea, oligomenorrhea or hypo-estrogenemia is associated with a wide array of premature aging effects. Thus, severe mental illness such as schizophrenia may be associated with an increased risk of fracture, especially osteoporotic fracture and especially in young women, and in those treated with prolactin-raising agents (Abel et al., 2007). Finally, although evidence of the beneficial effects of adjunct estrogen therapy in schizophrenia for women is relatively weak (see Chapter 23 for a detailed account), clinicians caring for women with schizophrenia should include menstrual and reproductive health checks as part of routine care planning.

Women with schizophrenia as mothers

Chapter 4 of this book considers broader issues relating to women as mothers. Here, we consider some particular difficulties for women with schizophrenia and their children. Women with a mental illness who become pregnant represent a high-risk group of mothers, not only because the mental illness may be more likely to recur or worsen following childbirth (see later in chapter), but also because they are more likely to be exposed (and to expose their fetus) to a range of adverse circumstances. We may say that mothers with mental illness have relatively poor “maternal condition.” This means greater risk of poor antenatal care, poor nutrition, use of prescribed medications, illicit drugs and alcohol, greater likelihood of smoking and of continuing to smoke through pregnancy and fewer positive benefits available from material and emotional supports (Abel et al., 2005).

Much recent attention has been paid to concerns about fetal exposure to psychotropic medication during pregnancy (Abel, 2013). However, the risks of relapse for women with schizophrenia off medication during pregnancy are likely to be considerable and to expose the mother-to–be and her fetus to increased anxiety and stress, which itself is associated with a range of obstetric complications, such as prematurity. The experiences of pregnant women with severe mental illness is covered in more detail in Chapter 10. Undoubtedly, far more research is required in this area and should include examination of differences across countries and cultures.

Parenting infants and preschool children is a stressful occupation, and missed sleep and raised levels of anxiety together with the isolation and emotional demands experienced in the early stages of infants’ lives can combine to precipitate or exacerbate periods of illness. Long-term mental illness can leave mothers without the support of a long-term relationship and relationships generally may be fractured and characterized by violence or abuse. Ill mothers are also far more likely to lose custody of their child after delivery, fear of which may exacerbate psychotic symptomatology during pregnancy and/or pregnancy-related anxiety (Abel et al., 2005). Mothers diagnosed with schizophrenia may be subject to high levels of scrutiny and monitoring from social services and these interventions are often experienced as hostile and threatening. Nearly all the mothers with severe mental illness interviewed by Stanley et al. (2003) reported fears that they might lose their children as a result of their mental health problems and a third of these women noted that such fears had restricted their willingness to seek help; as one interviewee remarked: “that’s why I won’t always ask for help. I’m worried that they will come and take them off me.”

Children whose mothers suffer with schizophrenia can be exposed to a variety of parenting problems including inconsistency and lack of predictability; emotional inaccessibility and unresponsiveness; and role reversal, which may involve inappropriate expectations on children and overinvolvement, where children are incorporated into paranoid or threatening delusions (Falkov, 2013). Such delusions may indicate a high level of risk for the child. However, an Irish study (Somers, 2007) of 37 children, aged 8–16, who were living with a parent with schizophrenia found no differences between the sample children and the matched controls in terms of physical health, positive family feelings, friendships, hobbies and household tasks. However, children with a parent with schizophrenia were less likely to have contact with relatives, were more likely to miss school and were more likely to have behavioral problems and “strange behavior” or mental health problems. The majority of sample children evinced an acute awareness of the stigma associated with mental illness and saw it as something that had to be kept hidden. It is clear that social isolation and social attitudes are heavily implicated in the impact of parental mental illness on children.

The risks to children of mothers with schizophrenia continue throughout their lives (Bee, et al., 2013; Bee, et al., 2014) although the level of risk may vary as levels of mental illness fluctuate. Generally, the study of this vulnerable group has been considered as “high-risk research” and has focused on the quantification of genetic risk. However, as noted earlier, many key social and environmental effects require consideration in the poor outcomes of these children. Thus, as discussed in Chapter 9, the pregnancies of women with schizophrenia are more likely to be unplanned, unwanted or a result of coerced sexual encounters than those of women in the general population.

Few epidemiological studies have assessed the parenting outcomes for mothers with serious mental illness. Kumar et al. (1995) reported on a small case-series of women admitted to a single UK mother-and-baby unit and suggested that mothers with schizophrenia had the poorest outcome, with 50% being separated from their babies at discharge. One follow-up study of women with postpartum psychosis found that they were more likely to experience further psychiatric episodes than mothers with non-psychotic affective disorders, endangering the relationship between mother and child further (Videbech & Gouliaev, 1995). In the largest reported sample, Salmon et al. (2003) described clinical and parenting outcomes in 1,081 joint mother-baby admissions (mean age 30 years), including 224 women with schizophrenia, 155 with bipolar disorder and 409 with non-psychotic depression. Poor outcomes were associated with a diagnosis of schizophrenia, behavioral disturbance, low social class, psychiatric illness in the woman’s partner and the absence of a good relationship with a partner. More detailed analysis of an overlapping UK mother-baby sample of 1,153 consecutive admissions (n=239 schizophrenia; n=693 affective disorder) suggested that mothers with schizophrenia were characterized by staff as having more complex clinical and psychosocial problems, and were considerably more likely to have a range of poor parenting outcomes compared to mothers with affective disorder (Abel et al., 2005). Only half of the mothers with schizophrenia were discharged home to care for their infants without any level of formal supervision by the social services, in contrast to the great majority of mothers with affective disorders (91% and 80% for unipolar depression and bipolar disorder respectively). Successful parenting was related, in part, to stability within the family, and access to financial and social resources. Thus, mothers with schizophrenia who have better parenting outcomes may be protected by certain factors such as supportive marital and other relationships, and higher social class.

How the expression of severe mental illness is mediated by life course experiences is critical to understanding the intimate nexus between genetic and environmental predictors of outcome for children of parents with mental illness. This includes not only prenatal exposures but also rearing environment and exposure to stressors and adversity. Animal models demonstrate that the experience of deprived rearing environments can permanently change gene expression (Meaney and Szyf, 2005) and chronically alter vital homeostatic systems (Pryce, et al., 2004).

There is evidence that children of mothers with severe mental illness such as schizophrenia experience a significantly poorer child rearing environment than those of well mothers, with children of depressed mothers somewhere in between (Goodman 1987; Donatelli, et al., 2010). Weintraub (1987) also reported that families of children of a parent with schizophrenia functioned significantly worse than those with well parents. However, in their study (Weintraub, 1987) families with a parent with affective psychosis did not differ from those with a parent with schizophrenia, while in contrast, in the New England Family Studies high-risk design, children of parents with affective psychosis fared better than those with parents with schizophrenia (Donatelli, et al., 2010). In the Rochester high-risk sample of children of women with schizophrenia, low socioeconomic status and maternal illness chronicity were more important predictors of early outcomes of children to age 4 years than maternal diagnosis per se (Sameroff, et al., 1987). When long-term outcomes were examined, high-risk children who go on to develop severe mental illnesses themselves reported poorer relationships with their parents than those who do not (Burman, et al., 1987). Far more translational research is needed to identify modifiable risk factors for high-risk children, so that appropriate interventions can be developed and implemented for both mother and child across the developmental life course, commencing in the antenatal period.