In a new paper in Archives of General Psychiatry that has received considerable media attention, Sullivan et al. (Sullivan et al., 2012) use register data from Sweden and Israel to show higher rates of ASDs among individuals with family histories of schizophrenia and bipolar disorder. The authors interpret these results as indicating that ASD, schizophrenia, and bipolar disorder share etiology. This is a very interesting hypothesis that, if supported, would have important implications for our understanding of the genetic underpinnings of schizophrenia in relation to other conditions. However, two alternative hypotheses not involving shared causation may, at least in part, help to explain their results.

First, a recent set of studies demonstrates that drug treatments for schizophrenia and bipolar disorder increase the incidence of ASDs, or their biologically based phenotypic correlates, in offspring. Croen et al. (Croen et al., 2011) reported that prenatal exposure to antidepressants (SSRIs) was associated with a twofold increase in risk of ASD. It is also notable that hyperserotoninemia has also been found in about one-third of autism cases (Burgess et al., 2006). Fetal exposure to the mood stabilizer valproate has been associated with a sevenfold increase in ASD risk (Bromley et al., 2008), and also serves as a model system for autism in animal studies. Use of clozapine and olanzapine during pregnancy has been associated with increased offspring head circumference (Bodén et al., 2012), which represents another well-validated correlate of autism (Courchesne et al., 2011). Moreover, environmental exposure to three psychoactive drugs (fluoxetine, venlafaxine, and carbamazepine) has been demonstrated to cause gene-expression changes that resemble those seen only in autism (Thomas and Klaper, 2012).

These results may help to explain mother-offspring and sib-sib associations of schizophrenia and bipolar disorder with ASDs. Such effects might be expected to be higher than those seen for fathers, but data were not presented in the report by Sullivan et al. on such parental sex differences. Effects of pharmacological treatment of fathers on ASD risk in offspring apparently have yet to be investigated, although paternal effects on offspring psychopathology and epigenetic profiles have been reported with regard to such factors as age (Hultman et al., 2011), and stress (Essex et al., 2011).

Second, the authors' data may also be attributable in part to false-positive diagnoses of premorbidity to schizophrenia (or bipolar disorder) as ASD in children, and conflation of schizotypal personality disorder (SPD) with high-functioning autism and Asperger's syndrome. Premorbidity to schizophrenia occurs in a notable proportion of cases, and most usually involves "negative symptoms" such as deficits in social interaction and language (discussed in Crespi, 2011). The clearest apparent evidence regarding this hypothesis comes from Sullivan et al. themselves, who noted that in their Study 1, 2,147 individuals had received a diagnosis of both ASD and (at discharge) schizophrenia or bipolar disorder. The authors excluded these cases as involving "diagnostic uncertainty." However, such uncertainties in the retained data may still influence the analyses. Thus, to the extent that individuals with diagnoses of ASD are under the age of onset for schizophrenia or bipolar disorder, they may exhibit false-positive diagnoses of premorbidity to schizophrenia or bipolar disorder as ASDs. Similar considerations apply to sibs differing in age.

Schizophrenia exhibits well-established genetic, symptomatic, and epidemiological overlap with both schizotypal personality disorder (SPD) and bipolar disorder (Carpenter et al., 2009). Additionally, first-order relatives of individuals with schizophrenia or affective psychosis show elevated rates of SPD (Schürhoff et al., 2005). These results indicate that SPD may show conflation in epidemiological data with high-functioning autism or Asperger's, due to the presence in both SPD and high-functioning forms of ASD of general social deficits and abnormalities. The possibility of such conflation is supported by: 1) the authors' finding that their familial association "was principally in cases without clinical indication of mental retardation," and 2) studies showing behavioral overlap of SPD with ASDs (based predominantly on questionnaires) (Barneveld et al., 2011), but a striking lack of data on overlap for developmental, physiological, or neurological phenotypes. Such conflation would falsely connect ASDs (which are actually SPD) with schizophrenia or bipolar disorder. It would appear more useful and realistic to consider the possibility and expected effects of diagnostic uncertainties than to presume that they do not exist.

This second set of considerations also applies to studies that would use GWAS data to evaluate hypotheses of how autism and schizophrenia are related to one another; even a rather small degree of false-positive conflation of premorbidity to schizophrenia with ASD could result in incorrect conclusions regarding the genetic etiologies of these sets of conditions. Such potential problems might be minimized by subsetting ASD cases into autism “sensu stricto,” given that PDD-NOS is the diagnostic category most likely to be conflated with schizophrenia premorbidity.

References:

Sullivan PF, Magnusson C, Reichenberg A, Boman M, Dalman C, Davidson M, Fruchter E, Hultman CM, Lundberg M, Långström N, Weiser M, Svensson AC, Lichtenstein P. Family history of schizophrenia and bipolar disorder as risk factors for autism. Arch Gen Psychiatry. 2012 Jul 2:1-5. Abstract

Croen LA, Grether JK, Yoshida CK, Odouli R, Hendrick V. Antidepressant use during pregnancy and childhood autism spectrum disorders. Arch Gen Psychiatry. 2011:68(11):1104-1112. Abstract

Burgess NK, Sweeten TL, McMahon WM, Fujinami RS. Hyperserotoninemia and altered immunity in autism. J Autism Dev Disord. 2006:36(5):697-704. Abstract

Bromley RL, Mawer G, Clayton-Smith J, Baker GA; Liverpool and Manchester Neurodevelopment Group. Autism spectrum disorders following in utero exposure to antiepileptic drugs. Neurology. 2008:71(23):1923-4. Abstract

Bodén R, Lundgren M, Brandt L, Reutfors J, Kieler H. Antipsychotics during pregnancy: relation to fetal and maternal metabolic effects. Arch Gen Psychiatry. 2012:69(7):715-21. Abstract

Courchesne E, Mouton PR, Calhoun ME, Semendeferi K, Ahrens-Barbeau C, Hallet MJ, Barnes CC, Pierce K. Neuron number and size in prefrontal cortex of children with autism. JAMA. 2011:306(18):2001-10. Abstract

Thomas MA, Klaper RD. Psychoactive pharmaceuticals induce fish gene expression profiles associated with human idiopathic autism. PLoS One. 2012;7(6):e32917. Abstract

Hultman CM, Sandin S, Levine SZ, Lichtenstein P, Reichenberg A. Advancing paternal age and risk of autism: new evidence from a population-based study and a meta-analysis of epidemiological studies. Mol Psychiatry. 2011:16(12):1203-12. Abstract

Essex MJ, Thomas Boyce W, Hertzman C, Lam LL, Armstrong JM, Neumann SM, Kobor MS. Epigenetic vestiges of early developmental adversity: childhood stress exposure and DNA methylation in adolescence. Child Dev. 2011 Sep 2. Abstract

Crespi B. One hundred years of insanity: genomic, psychological, and evolutionary models of autism in relation to schizophrenia. In: Ritsner M, ed. Handbook of Schizophrenia Spectrum Disorders, Volume I. New York, NY: Springer; 2011:163-185.

Carpenter WT, Bustillo JR, Thaker GK, van Os J, Krueger RF, Green MJ. The psychoses: cluster 3 of the proposed meta-structure for DSM-V and ICD-11. Psychol Med. 2009: 39(12):2025-42. Abstract

Schürhoff F, Laguerre A, Szöke A, Méary A, Leboyer M. Schizotypal dimensions: continuity between schizophrenia and bipolar disorders. Schizophr Res. 2005:80(2-3):235-42. Abstract

Barneveld PS, Pieterse J, de Sonneville L, van Rijn S, Lahuis B, van Engeland H, Swaab H. Overlap of autistic and schizotypal traits in adolescents with Autism Spectrum Disorders. Schizophr Res. 2011:126(1-3):231-6. Abstract

View all comments by Bernard Crespi

Shared risk for ASDs in bipolar and schizophrenia families is important, and the apparent gradient in risk with schizophrenia being greater than bipolar may be informative. From the view that schizophrenia and bipolar disorder are heterogeneous syndromes, the following is surmised:

• The association with ASDs probably is related to specific aspects of shared features, perhaps impaired social affiliation or neurodevelopmental pathology.
• These features are more common in schizophrenia than in bipolar disorder, so the gradient may be determined by the difference in proportion of relevant probands with the key features.
• If the same data could be reduced to family definition based on probands who have impaired social affiliation and/or neurodevelopmental psychopathology, then the odds ratios might be substantially higher and the ORs for schizophrenia and bipolar disorder might be more similar.

View all comments by William Carpenter

  I recommend the Primary Papers

This impressive study adds to the growing body of evidence demonstrating that heritable factors are shared among autism, schizophrenia, and bipolar disorder. The authors suggest that genetic factors could underlie the findings, but also remind the reader that environmental factors could play a role. They note that twin-based studies of heritability in schizophrenia and autism have demonstrated appreciable contributions for environmental factors that were shared between the affected individuals—usually referred to as common environmental effects. It should be noted that in this context, the word “common” does not equate with “prevalent.” With respect to shared genetic factors, the growing body of evidence regarding structural variation such as copy number variants is impressive. With respect to non-genetic factors, more work is needed—prenatal infection (which could trigger maternal immune activation) and nutrition (e.g., low vitamin D) might be candidate domains. If there are shared environmental risk factors contributing to schizophrenia, bipolar disorder, and autism, and if these were potentially modifiable, then this would be a very attractive proposition from a public health perspective.

The study is also an excellent demonstration of collaborative epidemiology—three datasets from two nations were used to examine the same research questions. This is an efficient way to do science.

View all comments by John McGrath