The authors put their results into perspective – “the population attributable risk ... is estimated to be 3.3%”. This says, in effect, that 96.7% of ASD cannot be attributed to these genetic changes. Moreover, the figure of 3.3% includes ID genes (already excluded) and significance was not adjusted for multiple comparisons.
The paper therefore provides extremely robust evidence that, in all but a few cases, ASD is not a genetic disorder...
By John Stone
The scientific journal Nature has refused publication to a letter from Drs Lorene Amet and Carol Stott in response to the study by the Autism Genome Project Consortium (AGPC, Pinto et al) (HERE ), which was widely publicised as having breakthrough status in the media when it was published on-line in June. The letter echoes and expands upon many criticisms already articulated by Mark Blaxill in Age of Autism (HERE). The Amet/Stott letter shows that so far from having demonstrated the influence of genes in autism the Pinto study illustrates the relatively small contribution of identified abnormalities in all but a very small proportion of cases, leaving the AGPC with the project of conducting ever wider searches for ever more tenuous correlations.The unwillingness of Nature to publish any challenge to the study is morally insupportable.
Autism Spectrum Disorder (ASD), a previously rare neurodevelopmental condition affects today in the UK one child in 64 (1).
There has been extensive debate on whether ASD has a predominantly genetic or environmental etiology. The three largest genome-wide association studies performed on more than 3000 individuals in total, have however so far failed to detect any specific gene association with any consistency across the studies (2-4). Predicated on a genetic etiology, the most recent study by Pinto et al. 2010 (4) looked at the frequency of rare inherited and de novo copy number variants (CNV). The paper concludes that CNVs are implicated in ASD, highlights the potential role of novel ASD genes including SHANK2, SYNGAP1 and DLGP2.
However, does this study point towards, or away from, a genetic etiology? Although seeming to argue that genetic changes underlie ASD development, careful scrutiny reveals that the paper confirms that ASD is not a predominantly genetic disorder.
Before examining the results of the Pinto et al. publication, the underlying assumption that genetic changes underpin ASD warrants evaluation. The paper states “Some 5-15% of individuals with an Autism Spectrum Disorder (ASD) have an identifiable genetic aetiology corresponding to known single-gene disorders (for example, fragile X syndrome) and chromosomal rearrangement (for example, maternal duplication of 15q11-q13)”, but no reference is provided for this statement. One remarks that the oft-cited twin studies that find high concordance between monozygotic twins do not distinguish between genetic and environmental etiologies – because such twins share an identical in utero and postnatal environment.
Taking the first example – fragile X, associated with mutations at the FMR1 locus – is it true that many ASD individuals have FMR1 mutations?
In the early 80s, when autism prevalence was estimated to be around 5 per 10,000 births (5), as many as 10% of ASD individuals were found to have fragile X (6). In 2005 by contrast, when autism prevalence is much increased, to around 1 in 150 (7), fragile X was encountered in less than 1% of the ASD population (8). In the last few years, in a group of 400 ASD children at the Autism Treatment Trust, only three (i.e. less than 1%) have a known genetic condition; one has fragile X, another has a mutation affecting NF-1 and a third one has partial trisomy of chromosome 15.
Today, therefore, it is strictly incorrect to state that any significant proportion of ASD individuals has “known single-gene disorders”. The vast majority do not.
Turning to the publication by Pinto et al., it will be helpful to look in depth at the results reported and the conclusions that can be justified.
Pinto et al. used a SNP microarray to look at signal intensity (i.e. copy number) across the whole genome, and then assemble these into clusters in which adjoining (genetically linked) markers reiteratively have higher (duplication) or lower (deletion) intensities. This was performed on 1274 ASD cases and 1981 controls of European ancestry.
The first analysis – changes in CNV numbers across the whole genome – apparently found no significant changes. To improve statistical power, Pinto et al. then looked at CNVs encompassing known genes (‘genic CNVs’). This second analysis revealed a significant difference between cases and controls (P = 0.012) although the case:control ratio was only 1.19. Here only deletions were of statistical significance (P=0.008).
The third analysis looked at the sizes of the CNVs. Small CNVs (<500 kb) were significant for deletions only (P=0.004), whereas large CNVs (>500Kb) achieved significance for duplications only (P = 0.007) (uncorrected for multiple analyses).
To further expand on these findings, Pinto et al. introduced several gene lists and analysed them separately. The first gene list was ‘ASD implicated’. This comprises 36 genes and 10 loci “strongly implicated” in ASD: in other words, a list of genes that previous studies have highlighted as being associated with some (rare) cases of ASD. This predominantly comprises the new candidate genes identified in the Pinto et al. study. The second list is ‘intellectual disability’ (ID) – “110 genes and 17 loci implicated in intellectual disability but not yet in ASD” and the third list consists of “ASD candidates including 103 genes from previous studies of common and rare variants”.
The first list of 36 genes selected as autism candidates includes a range of rare genetic disorders, such as Joubert Syndrome (9), Cortical dysplasia-focal epilepsy syndrome (10), Rubinstein-Taybi Syndrome (11) and many other similar conditions that have specific clinical, developmental, morphological and behavioural features that are clearly distinct to that of autism. A significant association between known ASD genes (P = 0.0026) was found but here the association was only with duplications (P = 0.00094) and not with deletions (NS, not significant).
Regarding the second list, the observer will be keenly aware that the category (ID) is suspect on statistical grounds. One does not know how many artificial lists were generated and compared. Other possible lists might include ‘any psychiatric disorder’ (including e.g. anxiety, a common co-morbidity of ASD), ‘any central nervous system disorder’ (including e.g. autoimmune neurodegenerative conditions – immune system anomalies are common in ASD), ‘any childhood disorder’; many other such lists are possible. In the absence of any clarification of how many such analyses, therefore, the Bonferroni correction for multiple comparisons argues that ID must be excluded from further analysis.
Regarding the third list of genes, there was no significant association between ASD and CNVS associated with the candidates ASD genes. This indicates therefore that the Pinto study failed to replicate the findings of predeceasing genome-wide studies.
Interestingly, the Pinto paper identified in 5.7% of ASD cases one or more de novo CNVs, but somehow there is not even a tentative explanation as to the reason for this. Importantly 69% of the control group consisted of females, with a mean age of 39.2 years. This cannot possibly be equivalent to any typical group of ASD subjects, which must be younger and include a larger proportion of males. It must be possible that the de novo CNV rate is associated with the birth age and gender of the patients, which may account for the somewhat subtle and inconsistent changes detected between the two groups.
Despite the very unclear picture emerging from these data, the authors concluded that “Single-occurrence CNV deletions had increased rates in ASD cases over controls, indicating that some could be pathogenic”. Such a statement is premature, particularly given the level of confusion and the absence of a measurement of potential changes in gene expression. DNA copy number variation is a widespread and common phenomenon among humans. It is estimated that approximately 0.4% of the genomes of unrelated people typically differ with respect to copy number. Recent work in breast and ovarian cancer demonstrates an extremely variable penetrance in monozygotic twins to be functionally independently of CNV profiling (12). A deletion or duplication in a selected gene area does not necessarily imply that there is a phenotypical change associated with this variation. Consistently, in the lists of de novo CNVs identified, many deletions or duplications were not found in the affected ASD sibling. The simplest explanation to these findings is therefore that the CNV changes detected are not related to autism.
What is the importance of these findings? The authors put their results into perspective – “the population attributable risk ... is estimated to be 3.3%”. This says, in effect, that 96.7% of ASD cannot be attributed to these genetic changes. Moreover, the figure of 3.3% includes ID genes (already excluded) and significance was not adjusted for multiple comparisons.
The paper therefore provides extremely robust evidence that, in all but a few cases, ASD is not a genetic disorder.
The complexity of ASD dictates that any assessment of the biomedical relevance of the CNV and potential changes in gene expression must be considered in a more strongly integrated context. The early onset of ASD implicates prenatal and perinatal environmental stress (13), and stress itself is known to foster genome mobilization – Barbara McClintock emphasised "the importance of stress in instigating genome modification by mobilizing available cell mechanisms that can restructure genomes” (14). The phenotypic presentation of today’s autism commonly includes regressive features, together with gastro-intestinal and immune abnormalities (15, 16). These must be taken into account in any search for an etiology. Time will tell if the locus duplications identified by Pinto et al. in a minority of ASD cases are found to be, not a cause of ASD, but instead a marker of adverse environmental influences that can lead to ASD development.
1- Baron-Cohen,S., et al. (2009). Prevalence of autism-spectrum conditions: UK school-based population study. The British Journal of Psychiatry 194, 500-509.
2- Wang,K., et al. (2009). Common genetic variants on 5p14.1 associate with autism spectrum disorders. Nature 459, 528-533.
3- Weiss,L.A., et al.. (2009). A genome-wide linkage and association scan reveals novel loci for autism. Nature 461, 802-808.4- Pinto,D., et al. (2010). Functional impact of global rare copy number variation in autism spectrum disorders. Nature.
5- Croen,L.A., et al. (2002). The changing prevalence of autism in California. J. Autism Dev. Disord. 32, 207-215.
6- Brown,W.T., et al. (1982). Autism is associated with the fragile-X syndrome. J. Autism Dev. Disord. 12, 303-308.
7- Hertz-Picciotto,I., et al. (2006). The CHARGE study: an epidemiologic investigation of genetic and environmental factors contributing to autism. Environ. Health Perspect. 114, 1119-1125.
8- Lathe,R. (2009). Fragile X and autism. Autism 13, 194-197.
9- Brancati,F., Dallapiccola,B., and Valente,E.M. (2010). Joubert Syndrome and related disorders. Orphanet. J. Rare. Dis. 5, 20.
10- Fauser,S., et al. (2006). Clinical characteristics in focal cortical dysplasia: a retrospective evaluation in a series of 120 patients. Brain 129, 1907-1916.
11- Chiang,P.W., Lee,N.C., Chien,N., Hwu,W.L., Spector,E., and Tsai,A.C. (2009). Somatic and germ-line mosaicism in Rubinstein-Taybi syndrome. Am. J. Med. Genet. A 149A, 1463-1467.
12- Lasa,A., Ramon,Y.C., Llort,G., Suela,J., Cigudosa,J.C., Cornet,M., Alonso,C., Barnadas,A., and Baiget,M. (2010). Copy number variations are not modifiers of phenotypic expression in a pair of identical twins carrying a BRCA1 mutation. Breast Cancer Res. Treat.
13- Gillberg,C., and Coleman,M. (1992). The biology of the autistic syndromes (MacKeith Press and Blackwell: Oxford).
14- McClintock, B. (1983). Nobel Lecture: The Significance of Responses of the Genome to Challenge, Nobel Foundation, Stockholm. http://nobelprize.org/nobel_prizes/medicine/laureates/1983/mcclintock-lecture.html.
15- Buie,T., Fuchs,G.J., III, Furuta,G.T., Kooros,K., Levy,J., Lewis,J.D., Wershil,B.K., and Winter,H. (2010). Recommendations for evaluation and treatment of common gastrointestinal problems in children with ASDs. Pediatrics 125 Suppl 1, S19-S29.
16- Amet,L.E., and Line,P. (2010). Autism Syndromes: Clinical issues behind the label. Brit. Med. J. 340, c1127.