Schizophrenia’s Etiology: Behavioural Geneticists’ Research and the Research They Ignore

This article was originally published in the December 2016 issue of the Journal of Critical Psychology, Counselling and Psychotherapy: Sedarati, J. (2016). Schizophrenia’s etiology: behavioural geneticists’ research and the research they ignore. The Journal of Critical Psychology, Counselling and Psychotherapy, 16(4), 257-266.

SUMMARY: This article evaluates the evidence in favour of a strong genetic component to so called schizophrenia, with reference to alternative perspectives.

KEYWORDS: schizophrenia, mental health, behavioural genetics, etiology.

Schizophrenia involves a range of cognitive, behavioural, and emotional problems including hallucinations, delusions, disorganised speech and behaviour, and negative symptoms: e.g. avolition (APA, 2013). However, no single symptom is pathognomonic. Lifetime prevalence of schizophrenia is approximately 0.3 per cent to 0.7 per cent, although reports vary based on several factors including race and sex. The current acritical consensus is that genes play a large role in the etiology of schizophrenia (Plomin, Defries, Knopik, & Neiderhiser, 2013; Ripke et al., 2014; Kendler, 1983; Torrey, 1992), however, a substantial body of evidence opposes this view (BPS, 2014; Alemany et al., 2013; Read & Dillon, 2013; Joseph, 1999, 2004b, 2015b; Jackson, 1960). This paper evaluates the evidence in support of a genetic role and argues that family, twin, and adoption studies are plagued by environmental confounds and methodological inconsistencies and thus, cannot be considered strong evidence for a genetic conception of schizophrenia. Furthermore, a discussion of molecular genetic research highlights a consistent inability to replicate findings, with studies of childhood adversity providing support for alternative explanations.

Establishing schizophrenia’s genetic etiology

Researchers have developed several methods for understanding genetic contributions to schizophrenia including; family, twin, and adoption studies, and more recently, molecular genetic studies (Rhee & Ronald, 2014). Schizophrenia tends to run in families (Kendler, 1983), however, it’s now widely recognised that family studies are unable to disentangle genetic and environmental factors. This is because family members share both a common environment and common genes (Joseph, 2015b). Accepting this, researchers now cite studies of pairs of monozygotic (MZ) and same-sex dizygotic (DZ) twins, or adoptees. The basic rationale behind the study of MZ and DZ twins is that because MZ twins share 100 per cent of their DNA and DZ twins only share 50 per cent (on average), then if a particular disorder such as schizophrenia is more frequently observed (i.e. concordant) in pairs of MZ twins compared to DZ twins, twin researchers conclude that the MZ twins’ more similar genes are responsible for the disparity. Based on their results, twin researchers then calculate heritability estimates for a particular trait or disorder.

Twin studies

Joseph (2004a) pooled the concordance rates of all twin studies up until mid- 2001 and calculated MZ pairwise concordance at 40.4 per cent and DZ pairwise concordance at 7.4 per cent. Pre-1962 twin studies are often distinguished from post-1962 studies because of their inferior research methods (e.g. non-blinded diagnoses and potentially biased samples). The pre-1962 studies reported, in general, much higher concordance rates compared to the post-1962 studies whose pooled pairwise concordance rates are only 22.4 per cent (MZ) and 4.6 per cent (DZ). Torrey (1992) aggregated the concordance rates for eight twin studies that met minimal criteria for sampling and zygosity ascertaintment and reported a pairwise concordance rate of 28 per cent for MZ twins and 6 per cent for DZ twins. These figures are significantly lower than the 50 per cent concordance estimate for MZ twins provided in modern textbooks (e.g. Kalat, 2016) and the pooled rates, discussed above. Despite the differences in concordance rates, a common trend among each estimate is that MZ concordance rates are significantly higher than DZ rates. Large heritability estimates for schizophrenia (e.g. 81 per cent, Sullivan, Kendler, & Neale, 2003; 60–80 per cent, Rhee & Ronald, 2014) have also been derived from twin research. These data have led many researchers (e.g. Kendler, 1983) to maintain the position that genetic factors play a key role in the etiology of schizophrenia.

Twin studies are praised for playing a ‘pivotal role… in establishing a genetic contribution to the etiology of schizophrenia’ (Cardno & Gottesman, 2000, p. 12), but despite their popularity, there are several methodological problems with this approach (Joseph, 2004a, 2004b). The ‘equal environment assumption’ (EEA) forms the bedrock of the twin method and maintains that pairs of MZ and DZ twins experience roughly equal environments; ergo any differences between them are attributed to genes. However, it is now understood that MZ twins do experience more similar environments, are treated more similarly by parents (e.g., they are dressed more alike), and experience greater identity confusion, compared to DZ twins (Joseph, 2004b). The invalidity of the EEA means that the disparity between MZ and DZ concordance rates for schizophrenia may not be due to the MZ twins’ more similar genetics, but rather, their more similar environments.

Proponents of the twin method (e.g. Kendler, 1983) argue that MZ twins’ more similar environments and experiences are the result of, rather than the cause, of their more similar phenotypes. This argument has been labelled ‘circular’ (Fosse, Joseph, & Richardson, 2015, p. 4), as it presumes a genetic basis before observations have even been made. Bouchard (1997) accepts that MZ and DZ twins’ environments are unequal, but argues that the differences are ‘trivial’ (p. 134). He states that critics haven’t demonstrated that MZ twins’ more similar experiences are relevant to a particular trait (eg. schizophrenia). The ‘trait-relevant’ defense is another attempt to subtly redefine the EEA. One which Joseph (2004b) argues devolves the twin method into a glorified family study; a method already acknowledged as unable to disentangle genetic and environmental influences. Regarding Bouchard’s notion of trait-relevance, research by Alemany et al (2013) is instructive. The authors reasoned that if disparities in the presence of psychotic experiences in MZ twins are associated with childhood adversity, this would provide support for the view that environmental influences are relevant to particular traits. Their assessment of 226 Spanish adult twins found that within-pair MZ differences in exposure to childhood adversity (e.g. physical, sexual, emotional abuse) were significantly associated with subsequent symptoms of psychosis.

Another point of contention regarding twin studies pertains to the concordance rates of same- and opposite-sex DZ twins. Based on the fact that all DZ twins, regardless of sex, share roughly the same proportion of genes, genetic theory would suggest that both types of twins would be equally concordant for schizophrenia (Jackson, 1960). However, pooled schizophrenia concordance rates in studies that assessed both types of DZ pairs are 11.3 per cent for same and 4.7 per cent for opposite-sex DZ twins (Joseph, 1999). Data from Gottesman (1991, cited in Kalat, 2016) also show disparities in the probability of developing schizophrenia between siblings (9%) and DZ twins (17%). These data contradict one of the basic expectations of twin research: that genetically homogenous individuals will be (on average) equally likely to develop schizophrenia.

Adoption studies

Inherent confounds in twin studies have led some researchers to conduct adoption studies. In theory, adoption studies enable researchers to separate biological and environmental factors by studying individuals who were raised by genetically unrelated caregivers and comparing them to a control group. If an adopted child resembles their biological parents for a particular trait (despite not being raised by them), it is inferred that genes influence the development of that trait. The first schizophrenia adoption study (Heston, 1966) compared adults born to mothers diagnosed with schizophrenia, where mother and infant (2 weeks old) were separated permanently, with a control group. Schizophrenia diagnoses were reported in 5/47 of the offspring. No cases of schizophrenia were found among the control group adoptees and the difference was statistically significant. Despite being consistently cited in support of a genetic etiology of schizophrenia, Heston’s study is rife with flaws (see Joseph 2004b), including poorly defined diagnostic criteria, lack of blinding, and the decision not to assess the psychiatric status of the adoptees’ fathers (i.e. half of the experimental group’s gene pool was ignored). Other high profile adoption studies such as the Danish-American studies (Kety et al., 1994; Rosenthal, Wender, Kety, & Schulsinger, 1971; Rosenthal et al, 1968; Wender, Rosenthal, Kety, Schulsinger, & Welner, 1974, cited in Plomin et al, 2013) purport to have established schizophrenia as a highly heritable disorder but these studies contain a plethora of limiting factors (see Joseph, 2004b, chapter 7). Importantly, none of these studies was able to provide an adequate definition of schizophrenia.

Diagnosis and environmental factors

There are no biological markers or laboratory tests for schizophrenia (Frances, 2014). In fact, ‘We still do not have a single laboratory test in psychiatry.’ (p. 10). Despite Frances’ criticisms of psychiatry, he and others (Flaum and Andreasen, 1990, cited in Carlson, 2012) argue that schizophrenia can be reliably diagnosed. Frances’ assertion is put forward without a source and directly contradicts his earlier admission that ‘all of our diagnoses are now based on subjective judgments that are inherently fallible and prey to capricious change.’ (p. 12). Research by Kirk and Kutchins (1994, cited in BPS, 2014), reported that clinicians who had received additional training in the application of diagnostic criteria could only agree on a broad diagnostic category about 50 per cent of the time, contradicting claims of diagnostic reliability. Evidence also suggests that, since the advent of the DSM-III, the inter-rater reliability of schizophrenia has steadily declined (Carney, 2013). Barnes (2004) reported that compared with Whites, African Americans admitted to state psychiatric hospitals were almost five times more likely to receive a diagnosis of schizophrenia rather than mood disorder. Frances (2014) acknowledges these findings and blames the discrepancies on bias and cultural ignorance rather than racial differences, further confusing his position that current diagnostic methods are reliable. Kinderman (2014) acknowledges that in the realm of physical health, diagnostic reliability is often also poor, however, the fundamental difference between physical and psychiatric diagnoses is that, with respect to physical ailments, objective tests and post-mortem examinations can correct mistakes made by diagnosticians; the same cannot be said for the field of psychiatry. It is unscientific to claim knowledge (e.g., via twin studies) about the etiology of schizophrenia if reliable methods of diagnosis do not exist. Thus, research that does attempt to search for ‘causes’ is inherently confounded based on this lack of objectivity.

Previous sections suggested that the methodologies and findings of twin and adoption studies are largely untenable. Despite this, Torrey (1992) asserted that genetics remains ‘the only clearly defined etiological factor’ (p. 168) for schizophrenia. Evidence associating psychosis with adverse childhood experiences suggests otherwise (Read & Dillon, 2013). In addition to Alemany et al. (2013, discussed above), Varese and colleagues (2012) conducted a meta-analysis of 41 studies published from 1980 to 2011 to investigate the association between childhood adversity and psychotic symptoms. They reported significant associations between adversity (e.g. physical, sexual, emotional abuse) and psychosis across all research designs. The findings suggest that upon eliminating factors of abuse, psychosis prevalence could be reduced by one third. Despite garnering little attention from behavioural geneticists, these and other data (e.g. Janssen, et al., 2004; Shevlin, Dorahy, & Adamson, 2007) imply that adversity is an important determinant of psychosis.

The multi-decade search for genes

Replication, replication, replication

For decades, researchers have been in search of ‘genes for’ schizophrenia. Originally, psychiatrists anticipated identifying single genes for a range of traits and disorders, but this has not occurred (Frances, 2014). In 1988, Sherrington and colleagues studied seven British and Icelandic families (N = 104) and published findings supporting a genetic link between two DNA polymorphisms on the long arm of chromosome five and schizophrenia. Sherrington et al. claimed the results provided ‘the first concrete evidence for a genetic basis to schizophrenia’ (p. 167). However, in the same year, a similar study using a Swedish kindred (N = 81) was unable to replicate their findings (Kennedy et al., 1988). Other failed attempts to replicate the genetic linkage were documented by Owen (1992). More recently, Faraone and colleagues (2008) admitted that after the publication of thousands of gene association studies, few findings have been replicated (see Joseph, 2013). In 2014, Ripke et al. published the largest ever molecular genetic study of schizophrenia. The genome-wide association study (GWAS) examined the genes of approximately 150,000 people and identified 108 genetic loci where schizophrenia patients’ DNA sequences differed from controls. However, the researchers concluded that these loci only explained a small fraction (3.4%) of the total variance. This is not an unusual finding. Thus far, genes have only been able to explain 1–5 per cent of the variation in psychiatric inscriptions such as depression, anxiety disorders, and schizophrenia (Plomin & Simon, 2013, cited in James, 2014). Also, the majority of the 108 loci had not previously been identified. Therefore, it is possible that the findings will not be replicated, as has been the general trend for 50 years (Joseph, 2013, 2015a).

Genetic dogmatism

‘We’re still waiting’ (Kupfer, 2013, para 1), admitted the Chair of the American Psychiatric Association’s DSM-5 Task Force with regards to the multi-decade search for biomarkers and genes for so called mental illnesses including schizophrenia. Despite the failure to find consistent genetic linkages (Frances, 2014), the DSM-5 (APA, 2013) claims there is ‘a strong contribution for genetic factors in determining schizophrenia’ (p. 103) and that there are a number of risk alleles that each contribute a small proportion to the total population variance. The ‘disappointingly small’ (Maher, 2008, p. 18) proportions in reported variance relative to the large heritability estimates derived from twin studies has been dubbed ‘the missing heritability’ problem. In opposition to their negative results, geneticists remain steadfast in their beliefs about the high heritability of psychological traits. The fact that thousands of failed replications and critical reviews of methodologically unsound research have not led the genetics community to re-evaluate its stance regarding the importance of genes in relation to schizophrenia and other mental disorders (Joseph, 2015a) should be of concern to any ‘scientific’ community.

A ‘firestorm’ (Kirk, Gomory, & Cohen, 2013, p. 29) of criticism has been directed at the field of psychiatry for its attitude towards mental health (Kinderman, 2014; Frances, 2014; Szasz, 1988). Schizophrenia patients and other individuals deemed ‘mentally ill’ have been routinely institutionalised and medicated against their will, despite there being ‘no demonstrable histopathological or pathophysiological evidence to support the claim that schizophrenia is a disease’ (Szasz, 1988, p. 105). The ‘disease-model’ approach to mental health (Kinderman, 2014) is fueled by a multi-billion dollar pharmaceutical industry financially incentivised to convince patients that ‘abnormal’ or distressing experiences are the result of innate traits which only pills can alleviate (Whitaker, 2010). It is also embraced by patients and families who wish to absolve themselves of responsibility for the consequences of their actions. Schizophrenia may run in families, but James (2016) points out that sexual abuse does too. Promoting biogenetic explanations may also induce pessimism in patients (Kvaale, Haslam, & Gottdiener, 2013) and evoke significantly less empathy in clinicians by exacerbating perceptions of patients as abnormal (Lebowitz & Ahn, 2014). Joseph (2004b) argues that even if a substantial genetic predisposition for schizophrenia was discovered, it would have ‘little meaning’ (p. 345) because researchers and clinicians could better serve patients by understanding how to minimise or eliminate environmental factors (e.g. intrafamilial assault).

Frances (2014) has admitted that schizophrenia is ‘not [a] disease’ but rather ‘a useful construct’ (p. 21). This ‘construct’ includes a variety of experiences and behaviours which are only considered symptoms of schizophrenia when they negatively impact an individual’s personal and professional functioning (Boyle, 2004). As there are no objective tests for schizophrenia and no symptom pathognomonic of the disorder, searching for a specific ‘cause’ for ‘the behaviours and experiences from which schizophrenia has misleadingly been inferred’ (Boyle, p. 455) makes little sense. Rather than accepting the reductionist notion that schizophrenia represents a distinct psychopathology resulting from a biological affliction, researchers should be aware that a variety of stressors have been implicated in experiences related to schizophrenia. For example, Read and colleagues (2003) reported a ‘particularly strong relationship between CA [child abuse] and hallucinations’ (p. 14). Specifically, child physical abuse survivors were 2.5 times, child sexual abuse survivors 3 times, and survivors of both forms of abuse 4 times more likely to experience hallucinations, compared to non-abused patients. Abuse takes many forms and Boyle (2004) reasoned that the behaviours and experiences categorised as ‘schizophrenic’ ‘are amongst the most extreme ways in which people react to or attempt to manage the distress caused by very aversive and threatening circumstances’ (p. 460). Focusing on experiences and behaviours rather than a categorisation inevitably demands an individualised approach to the treatment of people diagnosed with schizophrenia, which few members of the behavioural genetics field or pharmaceutical industry are interested in promoting.

The future?

Evidence supporting a genetic component to schizophrenia was evaluated. Twin studies do not provide valid evidence for the high heritability estimates their proponents claim are the primary ‘cause’ of schizophrenia. Reasons for this include; the false assumptions (eg. EEA) twin studies make, biased researchers, and the disparity in concordance rates between same- and opposite-sex DZ twins (Joseph, 2004b, 1999). Furthermore, adoption studies have been unable to apply adequate research methods to justify acceptance of their dubious findings (see Joseph, 2004b, chapter 7). Critics have been largely ignored by behavioural geneticists, who used high heritability estimates to justify searching for specific genes. After thousands of gene studies, few findings have been reproduced (Frances, 2014; Joseph 2013, 2015a). Moreover, the identified gene associations have, at best, only explained a small fraction of variance (Plomin & Simon, 2013, cited in James, 2014). All of these attempts to understand the genetic etiology of schizophrenia ignore the inconvenient fact that there are still no objective tests to distinguish a ‘normal’ individual from one with schizophrenia (Frances, 2014). Finally, strong evidence suggests childhood adversity plays a key role in the distress that individuals labelled ‘schizophrenic’ experience (Alemany, et al., 2013; Read & Dillon, 2013; Varese, et al., 2012). Researchers would better serve psychologically distressed individuals by understanding the environmental conditions that give rise to their distress, and learning how to minimise or even eliminate them.