These results indicate that SLOS is an uncommon cause of ASD in individuals without clinical signs of SLOS. Because we found no SLOS patients among the 100 patients selected to have a greater representation of autosomal recessive genetic disorders, unrecognized SLOS probably represents less than 1% of autism and, more likely, less than 0.2%. This is consonant with the epidemiology of SLOS and the distribution of DHCR7 alleles. As described by Kelley and Herman 
, at least 80% of DHCR7 mutations are null mutations. Given the previous estimate of a carrier frequency of 1.25% for null alleles, this predicts a prevalence of milder missense mutations of 0.2 to 0.3%, and thus a collective carrier frequency of approximately 1.5% in European-derived populations. The small percentage of SLOS patients whose physical findings are so mild that they could escape recognition until the age when a diagnosis of ASD could be made, most have been compound heterozygotes for two mild missense mutations or, occasionally, a null allele with a unique high residual activity missense allele. Therefore, one can estimate a birth incidence of SLOS mild enough to be “misdiagnosed” as ASD of approximately 1 in 400 – 500,000 births (0.003 × 0.003 × 0.25). If the incidence of non-regressive autistic spectrum disorder is estimated to be 1 in 1,000 births, the incidence of SLOS in autism is about 0.2%, in agreement with the high estimate from our biochemical screening of the AGRE samples. Thus, even using an incidence of non-regressive autism that may underestimate the current incidence of autism, our results indicate that routine screening of ASD for disorders of cholesterol biosynthesis is not indicated. Moreover, our clinical experience with close examination of biochemically minimally affected SLOS patients, whom we find are not truly normal physically, is that the most efficient way to screen for SLOS-caused ASD is a careful physical examination by a geneticist, who at the same time could assess the patient for other genetic conditions that can present as ASD.
The detection of a high incidence of hypocholesterolemia in our patient population is a finding of potential clinical significance regarding the possible role of non-SLOS cholesterol disorders in the etiology of ASD. A limitation of our study is that we lack information regarding parental lipoprotein profiles, which could help identify conditions such as hypobetalipoproteinemia, which are common but not known risk factors for autism. A major limitation of our study sample is the lack of clinical information about diet and other conditions, such as chronic diarrhea, medications, and the use of alternative therapies, that could underlie the hypocholesterolemia found in 19% of our subjects. However, studies indicate that individuals with decreased dietary intake or increased intestinal looses of cholesterol will have increased levels of cholesterol precursors, especially lathosterol (Lund et al., 1989
), which we did not find in comparing the patients with cholesterol levels below 100 mg/dL vs. the full cohort. Rather, individuals with a cholesterol level below 100 mg/dL had a statistically lower level of lathosterol compared to the entire cohort, indicating that the cause of the hypocholesterolemia was decreased cholesterol synthesis rather than increased cholesterol losses from gastrointestinal disturbances of abnormal diets. Instead, the data argue that the ASD patients with low cholesterol levels have intrinsically reduced cholesterol synthesis, similar to patients with Smith-Lemli-Opitz syndrome.
Several studies of ASD patients have reported increased incidences of chronic diarrhea, “fungal overgrowth,” and other intestinal disorders, which could be associated with hypocholesterolemia caused by bile acid and cholesterol malabsorption. However, in view of the rapid behavioral response of SLOS children to cholesterol supplementation, including partial remission of autistic spectrum behaviors, genetic or acquired hypocholesterolemia could be a contributing factor to the development of ASD. Our results, therefore, warrant further study. Many SLOS experts have speculated that the rapid behavioral response of SLOS patients to cholesterol supplementation is mediated by changes in neurosteroid levels, which are known potent modulators of behavior via modulation of GABA receptor synthesis and the responsiveness of receptors to GABA [Kelley and Herman, 2001
]. Indeed, the behavioral profile of SLOS [Tierney et al., 2001
] has several features, such as excessive anxiety, that would be expected in a CNS GABA-deficiency syndrome.
In summary, we have shown that, although ASD is a very common behavioral profile in SLOS, SLOS is a rare cause of ASD, accounting for no more than 1% of ASD. However, the unexpected finding that up to 20% of children from a sample of mostly multiplex ASD sibships have substantial hypocholesterolemia warrants further research, since the study of hypocholesterolemia and its predicted effects on neurosteroid metabolism may offer important insights into the cause and treatment of ASD.