This study is our first attempt at using the pooled multisite data to localize genetic locations using autism subphenotypes. It is apparent that pooling data from multiple sources to increase sample size is not a panacea due to the possible presence of genetic heterogeneity (67,68). However, this pooled sample did provide a larger base for us to select phenotypically homogenous subgroups, especially the ASD families with IQ ≥ 70. Future genetic studies may be improved by using the ADI-R item scores or factors derived from the item scores rather than the domain total scores due to the drawbacks discussed above. In addition, a quantitative IQ measurement may be a better heterogeneity-informative subphenotype than the categorical IQ used in this study.
