AETIOLOGICAL ROLE OF OBSTETRIC COMPLICATIONS IN SCHIZOPHRENIA
Schizophrenia, particularly in its severe form, has been reformulated as a neuro-developmental disorder (Reference MurrayMurray, 1994). Several lines of evidence from epidemiological, neuroimaging and post-mortem studies have suggested that at least a portion of adult schizophrenia results from neurodevelopmental impairment early in life. A number of studies have shown possible risk-increasing effects of obstetric complications (OCs) on the development of schizophrenia.
The view that OCs could have a pathogenic effect dates back to the 1950s. Pasamanick et al (Reference Pasamanick, Rogers and Lilienfeld1956) examined OCs in children with behaviour disorders. They concluded, “there exists a relationship between certain abnormal conditions associated with childbearing and the subsequent development of behaviour disorder in the offspring”. They proposed the hypothesis of a continuum of reproductive causality consisting of brain damage incurred during prenatal and paranatal periods, leading to a gradient of injury extending from foetal and neonatal death through cerebral palsy, epilepsy, mental deficiency and behaviour disorder.
Subsequent studies have examined OCs in childhood schizophrenia, adult schizophrenia and the offspring of parents with schizophrenia. McNeil & Kaij (Reference McNeil, Kaij, Wynne, Cromwell and Matthysse1978) reviewed the literature and concluded that OCs are a risk-increasing factor to be taken seriously in the aetiology of adult schizophrenia. Thereafter, more than 20 casecontrol studies and some cohort studies have examined the association between OCs and adult schizophrenia. Although the diagnostic criteria and assessment of OCs varied from study to study, the majority showed an increased rate of OCs in schizophrenia compared with controls. Geddes & Lawrie (Reference Geddes and Lawrie1995), who performed a meta-analysis combining 23 data-sets, found that subjects exposed to OCs were twice as likely to develop schizophrenia. Although most of the subjects included in this meta-analysis were from European populations, similar findings were obtained in a comparison of Japanese patients with schizophrenia and controls (Reference Kunugi, Nanko and TakeiKunugi et al, 1996a ).
However, the mechanisms underlying the relationship between OCs and schizophrenia are unclear. OCs include a variety of obstetric events, such as disease of the mother during pregnancy, abnormal foetal position, prolonged labour, foetal distress at delivery, low birthweight and diseases in early infancy. Although several OCs have been reported to be associated with schizophrenia (Table 1), no specific complication has been consistently found to be associated with schizophrenia. This may be attributable in part to the small sample sizes of the individual studies, and to the difficulty of obtaining obstetric data from a large number of subjects, especially when researchers attempted to obtain recorded obstetric data rather than retrospective information from interviews with mothers of patients with schizophrenia. In addition, it is possible that each single OC has only a weak effect on the development of schizophrenia.
Table 1 Obstetric complications found to be more common in patients with schizophrenia
| Study | Obstetric complications |
|---|---|
| McNeil & Kaij (Reference McNeil, Kaij, Wynne, Cromwell and Matthysse1978) | Pre-eclampsia, inertia of labour |
| Jacobsen & Kinney (Reference Jacobsen and Kinney1980) | Long labour |
| Parnas et al (Reference Parnas, Schulsinger and Teasdale1982) | Bad foetal position |
| Eagles et al (Reference Eagles, Gibson and Bremner1990) | Premature rupture of membranes |
| O'Callaghan et al (Reference O'Callaghan, Gibson and Colohan1992) | Foetal distress |
| Stöber et al (Reference Stöber, Franzek and Beckmann1993) | Maternal infectious diseases during mid-gestation |
| Günther-Genta et al (Reference Günther-Genta, Bovet and Hohlfeld1994) | Umbilical cord complications, atypical presentations |
| Rifkin et al (Reference Rifkin, Lewis and Jones1994) | Low birthweight |
| Kendell et al (Reference Kendell, Juszczak and Cole1996) | Pre-eclampsia |
| Hultman et al (Reference Hultman, Ohman and Cnattingius1997) | Disproportionate birthweight for body length, small head circumference |
| Jones et al (Reference Jones, Rantakallio and Hartikainen1998) | Low birthweight |
| Dalman et al (Reference Dalman, Allebeck and Cullberg1999) | Pre-eclampsia |
| Hultman et al (Reference Hultman, Sparen and Takei1999) | Multiparity, maternal bleeding during pregnancy |
In this context, obstetric records from a large number of subjects are required to clarify the link between OCs and schizophrenia. Furthermore, it is important to examine each individual complication meticulously in relation to any possible causal relationship with schizophrenia. Among the many items included under the term ‘obstetric complications’, we focus here on the role of intra-uterine growth.
INTRA-UTERINE PHYSICAL GROWTH AND BRAIN DEVELOPMENT IN SCHIZOPHRENIA
Birthweight
Mean birthweight has been reported to be lower among patients with schizophrenia than among controls (Reference McNeil, Cantor-Graae and NorströmMcNeil et al, 1993; Reference Sacker, Done and CrowSacker et al, 1995; Reference Hultman, Ohman and CnattingiusHultman et al, 1997), unaffected siblings (Reference Lane and AlbeeLane & Albee, 1966; Reference Woerner, Pollack and KleinWoerner et al, 1971) and individuals with an affective psychosis (Reference Rifkin, Lewis and JonesRifkin et al, 1994). However, there are also many studies reporting no significant difference in birthweight between patients with schizophrenia and their healthy siblings, normal controls or live births (Reference Pollack, Woerner and GoodmanPollack et al, 1966; Reference Jacobsen and KinneyJacobsen & Kinney, 1980; Reference McCreadie, Hall and BerryMcCreadie et al, 1992; Reference O'Callaghan, Gibson and ColohanO'Callaghan et al, 1992; Reference Günther-Genta, Bovet and HohlfeldGünther-Genta et al, 1994; Kendell et al, Reference Kendell, Juszczak and Cole1996, Reference Kendell, McInneny and Juszczak2000; Reference Jones, Rantakallio and HartikainenJones et al, 1998). Thus it is not clear-cut as to whether mean birthweight is lower in patients with schizophrenia than in control subjects.
If schizophrenia could be placed within a continuum of reproductive causality from neonatal death, through cerebral palsy to behaviour disorders, as Pasamanick et al (Reference Pasamanick, Rogers and Lilienfeld1956) proposed, then the frequency of low birthweight rather than mean birthweight might be more important, since low birthweight is a more clearly defined risk factor for cerebral palsy. Using Medline, we searched for recent studies comparing the frequency of low birthweight (<2500 g) between patients with schizophrenia and controls; six studies were found that were published in the 1990s. These studies and our own more recent data (Reference Ichiki, Kunugi and TakeiIchiki et al, 2000) are shown in Table 2. A significantly increased frequency of low birthweight among patients with schizophrenia compared with controls was reported in five out of the seven studies. The odds ratios in six out of the seven studies ranged between 1.7 and 3.9. The overall pooled data yielded a highly significant result: 9.5% and 3.9% of approximately 750 patients with schizophrenia and 29 000 control subjects respectively were born with low birthweight (P < 0.00001, by Fisher's exact test, two-tailed, odds ratio 2.6, 95% CI 2.0-3.3). This indicates that low birthweight is a modest - but definite - risk factor for adult schizophrenia.
Table 2 Studies comparing the frequency of low birthweight in patients with schizophrenia and control subjects
| Study | Schizophrenia group | Control group | Odds ratio (95% CI) and P value | Comment | ||
|---|---|---|---|---|---|---|
| n | LBW1 | n | LBW1 | |||
| Ichiki et al (Reference Ichiki, Kunugi and Takei2000) | 312 | 30 (9.6%) | 517 | 24(4.6%) | 2.2(1.3-3.8) P=0.005 | Japanese case-control study |
| Dalman et al (Reference Dalman, Allebeck and Cullberg1999) | 238 | 16 (6.7%) | Data not shown | 1.7 (1.0-3.0) | Swedish cohort study | |
| Hultman et al (Reference Hultman, Sparen and Takei1999) | 167 | 11 (6.6%) | 835 | 31 (3.7%) | 1.8 (0.9-3.7)2 P=0.092 | Swedish cohort study |
| Jones et al (Reference Jones, Rantakallio and Hartikainen1998) | 76 | 6 (7.9%) | 10 498 | 360 (3.4%) | 2.4(1.0-5.6) P=0.03 | Finnish cohort study |
| Sacker et al (Reference Sacker, Done and Crow1995) | 49 | 7 (14.7%) | 16 812 | 7062 (4.2%) | 3.9 (1.9-8.1) P<0.01 | British cohort study; schizophrenia was narrowly defined by PSE category |
| Rifkin et al (Reference Rifkin, Lewis and Jones1994) | 100 | 152 (15%) | 67 | 12 (2%) | 11.6 (1.5-90.4)2 P=0.0022 | British case-control study. Controls were patients with affective psychosis |
| Heun & Maier (Reference Heun and Maier1993) | 43 | 2 (4.7%) | 74 | 2 (2.7%) | 1.8 (0.2-12.9)2 P=0.622 | German case-control study. Controls were siblings of subjects with schizophrenia |
| Total3 | 747 | 71 (9.5%) | 28 803 | 1124 (3.9%) | 2.6 (2.0-3.3) P<0.00001 | |
Twin studies
Since monozygotic (MZ) twins share the same genetic material, it is advantageous to investigate intra-pair differences in birthweight for MZ twins discordant for schizophrenia in order to determine the effect of birthweight on the development of the illness. Pollin & Stabenau (Reference Pollin, Stabenau, Rosenthal and Kety1968) analysed 100 pairs of MZ twins discordant for schizophrenia or severity of schizophrenia and provided evidence of lower birthweight in the twins with schizophrenia than in their co-twins. In contrast, Gottesman & Shields (Reference Gottesman and Shields1976), who examined 82 pairs from five studies, reported lower birthweight to be equally distributed among twins with schizophrenia and their well co-twins. However, Torrey (Reference Torrey1977), who re-examined the same data, excluding partially concordant pairs, concluded that lower birthweight was more frequent in the affected twins than in the unaffected co-twins.
More recently Onstad et al (Reference Onstad, Skre and Torgersen1992) examined 16 MZ twin pairs discordant for schizophrenia, and found that there was no significant difference in mean birthweight between those with schizophrenia and the unaffected twins. McNeil et al (Reference McNeil, Cantor-Graae and Torrey1994), who investigated 23 MZ twin pairs discordant for schizophrenia, reported that birthweight was not significantly lower in the twins with schizophrenia than in the well co-twins.
To summarise the data from studies on MZ twin pairs discordant for schizophrenia, there is no clear-cut evidence for lower birthweight in twins with schizophrenia when compared with their unaffected co-twins. However, the sample sizes were too small to detect the possible weak relationship between mean birthweight and the development of schizophrenia. Furthermore, given the sample sizes, it is difficult to draw any conclusion with respect to the relationship between low birthweight and subsequent schizophrenia. Another limitation was that most twin studies obtained their obstetric data from maternal or other relatives' recall rather than from hospital records.
Head circumference at birth
McNeil et al (Reference McNeil, Cantor-Graae and Norström1993) found a reduced head circumference at birth in patients with schizophrenia compared with controls. Subsequently, Kunugi et al (Reference Kunugi, Takei and Murray1996b ) found a smaller head circumference at birth for gestational age in patients with schizophrenia compared with controls. Hultman et al (Reference Hultman, Ohman and Cnattingius1997) also provided evidence for smaller head circumference at birth in patients with schizophrenia.
These findings are intriguing since some studies reported that patients with schizophrenia have reduced whole-brain (especially cortical) volume (Reference Zipursky, Lim and SullivanZipursky et al, 1992; Reference Harvey, Ron and Du BoulayHarvey et al, 1993). These structural changes are thought to be non-progressive, and to originate early in life (Reference RobertsRoberts, 1991; Reference MurrayMurray, 1994). Since head circumference accurately reflects intracranial volume (Reference Bray, Shields and WolcottBray et al, 1969) and correlates with the number of cells in the brain (Reference Winick and RossoWinick & Rosso, 1969), smaller head circumference at birth might reflect primary neurodevelopmental impairment in schizophrenia dating from the prenatal period. Small head size at birth was shown to be a predictor of neurological impairments and lower IQ at the age of 5 years among low-birthweight infants (Reference Gross, Kosmetatos and GrimesGross et al, 1978).
PRENATAL UNDERDEVELOPMENT AND BRAIN DAMAGE
Figure 1 illustrates mechanisms that may underlie the possible association between prenatal underdevelopment and central nervous system (CNS) damage. Genetic predisposition to schizophrenia may be associated with lower birthweight. In accordance with this, Mednick et al (Reference Mednick, Mura and Schulsinger1971), in their Danish high-risk sample, reported that children born to a parent with schizophrenia evidenced a higher incidence of mildly low birthweight. Although McNeil & Kaij (Reference McNeil and Kaij1973) did not find a significantly lower mean birthweight in the offspring of mothers with schizophrenia than those of control mothers, more of the former were small for their gestational age. The Jerusalem Infant Development Study (Reference Marcus, Auerbach and WilkinsonMarcus et al, 1981) showed that infants born to patients with schizophrenia tended to have low to low-normal birthweights. More recently, Bennedsen et al (Reference Bennedsen, Mortensen and Olesen1999) investigated the offspring of 1537 women with schizophrenia, and found that pre-term birth, low birthweight and being small for gestational age were significantly more common in the offspring of women with schizophrenia, compared with those of control mothers.
Fig. 1 Possible mechanisms underlying the link between low birthweight and central nervous system (CNS) damage.
Pregnancy complications such as malnutrition, smoking, pre-eclampsia and diseased placenta are known to be associated with lower birthweight (Reference Lubchenco and AveryLubchenco, 1975). These environmental insults may cause chronic hypoxia in the foetus with resultant abnormalities in the CNS. Intriguingly, Wright et al (Reference Wright, Takei and Rifkin1995) found that maternal influenza infection during the second trimester was associated with lower birthweight in the offspring. This may account, at least in part, for a possible mechanism underlying the relationship between maternal influenza and adult schizophrenia (e.g. Reference Mednick, Machon and HuttunenMednick et al, 1988; Reference O'Callaghan, Sham and TakeiO'Callaghan et al, 1991; Reference Kunugi, Nanko and TakeiKunugi et al, 1995).
Furthermore, several perinatal complications secondary to low birthweight or premature birth are likely: respiratory failure, jaundice leading to kernicterus, dyshaemopoietic anaemia and infectious diseases, for example. Small infants do not tolerate labour and delivery well, and may suffer foetal distress and aspiration of meconium. Such infants have also an inability to conserve heat, and hypoglycaemia is frequently observed (Reference Lubchenco and AveryLubchenco, 1975). All these untoward conditions are likely to have detrimental effects on brain development in the perinatal period. Additional primary perinatal complications such as premature rupture of membranes, umbilical cord complications, inertia of labour and prolonged labour may also cause hypoxia and ischaemic brain damage.
In severe cases, prematurity is often complicated by intraventricular haemorrhagic infarction and periventricular leucomalacia, which are associated with subsequent neurological handicaps such as cerebral palsy and less prominent developmental disabilities of motility, cognition and behaviour (Reference Vohr and MentVohr & Ment, 1996; Reference VolpeVolpe, 1997). It is an open question whether milder forms of these pathological conditions are related to schizophrenia.
PREMATURITY AT BIRTH AND SUBSEQUENT NEURODEVELOPMENTAL ABNORMALITIES: IMPLICATIONS FOR SCHIZOPHRENIA
In the general population, low birthweight is a risk factor for major neurological impairments, such as cerebral palsy, significant developmental delay and sensory deficits, particularly in very low-birthweight (< 1500 g) infants. Even when such major impairments do not occur, low birthweight is also associated with reading difficulties, language delay, poor visual motor integration, behavioural problems and lower IQ at school age (Reference Aylward, Pfeiffer and WrightAylward et al, 1989; Reference Lukeman and MelvinLukeman & Melvin, 1993). These neurodevelopmental abnormalities are attributable to perinatal brain damage. Stewart et al (Reference Stewart, Rifkin and Amess1999), who examined magnetic resonance imaging (MRI) scans of adolescents who had been born very preterm, found that more than half of such individuals had abnormalities of ventricles, corpus callosum and white matter, and that reading, adjustment and neurological impairments were related to the brain abnormalities.
These observations in the general population led to the hypothesis that a subgroup of patients with schizophrenia who were subject to prenatal underdevelopment might show neurodevelopmental abnormalities from early in life. Indeed, Rifkin et al (Reference Rifkin, Lewis and Jones1994) found that lower birthweight correlated significantly with poorer premorbid social and cognitive ability among patients with schizophrenia. Cannon et al (Reference Cannon, Jones and Gilvarry1997) obtained further evidence for lower birthweight correlating with poorer premorbid adjustment in terms of sociability and schooling. Fish et al (Reference Fish, Marcus and Hans1992) found in their Jerusalem Infant Development Study that there was a negative correlation between birthweight and motor development at 10 years of age in the offspring of mothers with schizophrenia. Torrey et al (Reference Torrey, Taylor and Bracha1994), who examined 23 pairs of MZ twins discordant for schizophrenia, found that there were two subgroups of early and late divergence groups. In the early divergent group, the affected and unaffected twins became permanently different from each other in motor skills or unusual behaviour at the age of 5 years or before. Of note is the fact that in all the twin pairs with an intra-pair birthweight difference of 20% or more, the twin who became ill was the lighter one and had an early age of divergence.
All these observations support the possibility that a subgroup of patients with schizophrenia suffer prenatal underdevelopment and this group tends to show poor premorbid functioning.
PERINATAL BRAIN DAMAGE AND SCHIZOPHRENIA
If a portion of schizophrenia originates from prenatal underdevelopment and associated perinatal brain damage, then there should be a correlation between prenatal underdevelopment and morphological changes of the brain in patients with schizophrenia. Although numerous studies investigated structural brain abnormalities in schizophrenia, few have examined the possible relationship between prenatal development and brain morphology in schizophrenia. In the Danish high-risk study, lower birthweight was associated with ventricular enlargement and periventricular damage in a sample genetically predisposed to schizophrenia (Reference Silverton, Finello and MednickSilverton et al, 1985; Reference Cannon, Mednick and ParnasCannon et al, 1989). In contrast, Mukherjee et al (Reference Mukherjee, Schnur and Reddy1993) did not find such a relationship in chronic patients with schizophrenia. These studies were limited by the use of X-ray computerised tomography rather than MRI in measuring brain structures. In addition, the study of Mukherjee et al (Reference Mukherjee, Schnur and Reddy1993) was inadequate in sample size (n=24) to detect a moderate effect size. Further studies using MRI in larger sample sizes are needed to clarify the relationship between prenatal underdevelopment and structural brain abnormalities in schizophrenia.
CONCLUSION
We looked at the possible aetiological role of prenatal underdevelopment in schizophrenia. Pooled data from the recent world literature showed that low birthweight (< 2500 g) is a modest, but definite, risk factor for schizophrenia. Thus, in a small proportion of patients, schizophrenia appears to originate from perinatal hypoxic brain damage associated with prenatal underdevelopment. This subgroup tends to show poorer premorbid psychosocial functioning, and might be placed within the continuum of reproductive casualty from cerebral palsy to behaviour disorders, which Pasamanick et al (Reference Pasamanick, Rogers and Lilienfeld1956) proposed. To elucidate this issue, studies are needed to examine the possible relationship between prenatal underdevelopment and structural and functional abnormalities of the brain in schizophrenia.
eLetters
No eLetters have been published for this article.