The first two authors contributed equally to this work. Autism spectrum disorders (ASDs), including autism as well as pervasive developmental disorder not otherwise specified (PDD-NOS) and Asperger's disorder, are characterized by impairments of reciprocal social interactions, problems in communication, and a restricted range of behaviors and interests [1] . Although specific causes of ASDs have yet to be found, many risk factors have been identified that may contribute to the development of

more »

... s. These risk factors include genetics, environmental factors, prenatal and perinatal factors, and neuroanatomical abnormalities [2] . There is ample evidence suggesting that genetic factors play a role in the etiology of ASDs [3] . With the accomplishment of several systematic family studies, it is found that the consistent sibling recurrence rate is substantially higher than the population prevalence [4, 5] . The results of twin studies suggest a dramatically higher concordance rate in monozygotic twins than in dizygotic twins [6] . In addition, by means of single-nucleotide polymorphism (SNP) microarrays and karyotyping, structural variation of chromosomes has been identified in some individuals with ASD, including 9p24.1 as a recurrent and overlapping locus in ASDs [1]. The ABO blood type is controlled by the ABO gene encoding a glycosyltransferase, an enzyme that modifies the carbohydrate content of the red blood cell antigens. The gene consists of seven exons, and is located on chromosome 9q34 [7] . A series of single nucleotide polymorphisms across the seven exons form three different alleles of the ABO gene, referring to the A, B and O alleles [8] . As the ABO gene is located on the ninth chromosome, of which structural variation is also observed in some individuals with ASD, it seemed reasonable to surmise that ABO blood type was functionally related to ASDs. Blood types are inherited from both parents. We hypothesized that if parental ABO blood type was associated with the development of filial ASDs, there would be a higher probability of filial ASDs in parents with a specific ABO blood type. If so, medical workers can utilize the parental ABO blood type as an easy way to roughly predict the morbidity of filial ASDs. To determine whether there is a specific ABO blood type in parents of children with ASD, we surveyed families of children with ASD ascertained through Shanghai Mental Health Center. To ensure an unbiased study population for our investigation, the program was designed to obviate ascertainment biases with the same inclusion criteria including diagnosis and treatment. Any individual information meeting those criteria was collected in an ASD Study Population Database of Shanghai Mental Health Center, with a total number of 440 children (385 male and 55 female) who came to the clinic for medical help from 2009 to 2011. One hundred and two cases of 440 (23.18%) showed histories of treatment for jaundice. Families primarily came from Shanghai. With the information provided by the database, we could collect general materials of family members and conduct follow ups for subsequent studies. All the families agreed to participate in telephone interviews, and 225 mothers and 212 fathers who were reached by telephone knew their ABO blood group status. The control population was accessed from the Shanghai Blood Center and represents all individuals who donated blood between 2009 and 2011. The percentage of A, B, AB and O in the control population were 29.9%, 28.6%, 11.0%, 30.5%, respectively, with a sum of 1695 Table 1 Distribution of ABO blood types (%) in fathers of children with ASD, mothers of children with ASD and the control population. A B AB O Total Father 61(28.8) 55 (25.9) 17 (8.0) 79 (37.3) 212 Mother 60 (26.7) 64 (28.4) 21 (9.3) 80 (35.6) 225 Control population 509 (29.9) 485 (28.6) 186 (11.0) 517 (30.5) 1695 Total 628 604 224 676 individuals. To compare possible differences between distributions of blood types in parents of children with ASD and in the control population, the Chi-square test was carried out. Distribution of ABO blood types in fathers/mothers of children with ASD and the control population showed no statistically significant differences (χ 2 = 7.01, df = 6, P = 0.28) ( Table 1) . Numerous studies have been trying to disentangle the complex pathophysiology of ASDs [9] . In this study, we postulated that specific blood type of parents may contribute to the risk of autism with some unknown genetic mechanism which subsequently disrupts fetal neurodevelopment. However, no significant difference was discovered between the distribution of ABO blood type in parents of children with ASD and that in the control population. This study adds to the evidence that there is no causal association between blood type and childhood ASDs, and has major implications for other hematic and genetic research [10] . Limitations are inevitable in the research. The primary one is the relatively small number of subjects. Moreover, information of blood types in the study was self-reported by parents of children with ASD, which may create recall bias. An additional constraint is the utility of those populations is lessened by the bias caused by an exclusion rate of nearly half of parents who did not know their blood type. All of these underscore the importance of large collaborative studies that have high statistical power to draw a reliable conclusion. Therefore, further study is needed before a conclusion can be reached on this topic.