Children with autism are hard to reach - they typically avoid eye contact, have difficulties with communication and fail to read the emotional signals of others. The genetics of the condition is proving to be similarly elusive. Recent studies have begun to suggest which genes might be involved and what their function might be - but the picture is turning out to be highly complex.
'I never thought it would be so hard', says Professor Tony Monaco, who heads the group at WTCHG working on neurodevelopmental disorders. 'Autism is so heritable - in identical twins where one twin is affected, around 90 per cent of the time the other twin will also have a condition on the autistic spectrum.' Other evidence that the condition clusters in families points to a strong role for a genetic cause. Optimistic by nature, Monaco used to believe that once the technology was available, it would only be a matter of time before it would reveal mutations in one or two major genes.
Autism is a relatively rare condition, so the first task was to collect DNA from families where at least two members were affected. Realising that there was a need to pool these resources internationally, Monaco and his colleague in the Department of Psychiatry, Anthony Bailey, helped to set up the International Molecular Genetic Study of Autism Consortium (IMGSAC), which involves researchers in nine countries in Europe and the USA, with multiple funders including the Wellcome Trust and Medical Research Council. In 1998 they carried out their first genome scans on members of 99 families with autism. At first Monaco's optimism seemed justified, as the scan revealed a strong association between the condition and a region on chromosome 7, confirmed in a later study in 2001.
However, further studies that sequenced genes in promising regions on chromosome 7 and also chromosome 2 could not find any gene that could account for the strong linkage. 'When you found a gene that had mutations or was deleted, you only found it in less than 1 per cent of cases', says Monaco. 'We realised that we had to do much larger studies.' IMGSAC has now joined with other consortia to form the Autism Genome Project, holding genetic data from 3000 families and greatly increasing their 'power' - the ability to find statistically significant links between genes and disease.
Meanwhile the most recent fine scan of the main linked regions (published in April 2009) has turned up a new association, with a gene known as Dock4 on chromosome 7. Although mutations in Dock4 are linked to only a small number of cases, the new discovery adds to a picture of autism as a problem in the wiring of the brain early in development. Dock4 and several other candidate genes play a role in making the initial contacts that establish the brain's network of connections, or synapses. 'That does give us a focus for looking at the biological basis of the condition', says Monaco.'
Monaco and his colleagues have had to accept a major shift in their assumptions about the genetics of autism. They began with the 'common variant, common disease' model, where gene variants that occur in up to 30 per cent of people can cause disease if inherited in the wrong combinations. This model has worked well for diabetes, but for more complex conditions including schizophrenia and autism it does not seem to apply. 'For autism we find one gene, another group finds another - nothing stands up across all the studies', says Monaco. 'We believe that common variants have a role to play, but we won't be able to detect their significance until we get up to much higher numbers. What has come out is a rare variant model.'
Gene variants typically contain one or more 'single nucleotide polymorphisms' or snps - single-letter changes in the spelling of the genetic code. But there are other, less common forms of variation. One such is copy number variation, in which one of the pair of genes at each location in the genome is either deleted or duplicated, so that individuals have one or three copies instead of the usual two. '15-20 per cent of autism cases have copy number variants', says Monaco, 'but each of these copy number variants is contributing to only 1 per cent or so of cases.' The deletions that lead to copy number variation can knock out more than one gene at a time, and thereby increase risk for a range of different disorders, the outcome depending on other genetic and environmental factors.
Monaco and his colleagues are about to move into a new phase of research, hoping not only to take their genome scans into thousands rather than hundreds of samples, but also opening up a new 'translational' front which will provide genetic information to clinicians treating autistic patients. They would like to recruit a network of paediatricians who will take a blood sample from 1000 children consecutively diagnosed with autism according to standard clinical criteria. Researchers will then look for copy number variants using array-based technology. Tests for genetic abnormalities are already used in cases of learning disability to provide genetic counselling to the family. 'No one's done a large study to show its clinical utility in autism', says Monaco. 'If you found a copy number variation, you might be able to test younger siblings to see if they were at increased risk.'
But he does not envisage that that genetics alone will be able to provide early diagnosis in autism, as it does in single gene diseases such as muscular dystrophy and cystic fibrosis. 'It will be very difficult, even at the end of the day, to produce an absolute test for autism at birth', he says. Another commonly-expressed goal of genetic linkage studies - to provide a biological basis for drug treatments - also seems some way off for autism.
Autism researchers liken their field to doing a 500-piece jigsaw puzzle without the picture - and without many of the pieces to hand to start with. 'Slowly you try to build up what the picture looks like', says Monaco. 'We think it's the connections between brain cells that are affected, but we have a lot of work to do to understand how the genes interact with one another.'
E. Maestrini et al, High-density SNP association study and copy number variation analysis of the AUTS1 and AUTS5 loci implicate the IMMP2L-DOCK4 gene region in autism susceptibility. Molecular Psychiatry advance online publication 28 April 2009.
For more information on Dr Monaco's autism research, click here.