Gayle Windham, MSPH, PhD1, Meredith Anderson, MPH2, and Judith Grether, PhD1. (1) Division of Environmental and Occupational Disease Control, CA Department of Public Health, 850 Marina Bay Pkwy, Building P, Richmond, CA 94804, (2) California Dept. of Public Health, Impact Assessment, Inc., 850 Marina Bay Pkwy, Bldg P, Richmond, CA 94804
Background: Hormonal etiologies have been proposed to explain some of the observed patterns of autism occurrence.
Objectives: To explore direction for further studies, we linked hormonal data from routine prenatal screening to autism prevalence data in California.
Methods: Our statewide autism surveillance links data from the Department of Developmental Services to birth certificates. This surveillance data was linked to maternal serum triple marker screening data for two birth years, 1996 and 2002. Over 50% of pregnancies were screened, yielding nearly 600,000 singleton births and 2,500 autism cases for analysis. The three markers are 1) maternal serum alpha-fetoprotein (MSAFP), a glycoprotein that may also bind estrogen, 2) unconjugated estriol (uE3), the principal estrogen metabolite found in blood during pregnancy, and 3) human chorionic gonadotropin (hCG), important for the maintenance of pregnancy and steroid regulation. Analyte concentration is measured and also converted to multiples of the median (MoM) for reporting risk of various fetal conditions. We examined the marker levels as continuous and categorical variables, comparing autism cases to remaining births, and will adjust for additional factors in multi-variate models.
Results: For 1996 and 2002 combined, mean and median MSAFP levels were slightly, but statistically significantly, higher in mothers of autistic children. In contrast, uE3 mean levels were significantly lower among case mothers. The pattern persisted when markers were treated as categorical variables, with significant odds ratios around 1.2 for the highest decile of MSAFP and the lowest decile of uE3, compared to the mid quartile range (25th-75th percentiles). Risks of similar magnitude were noted for both the high and low deciles of hCG.
Conclusions: Low and high levels of these markers are associated with adverse fetal outcomes. Should the differences we observe persist after further analyses, it would suggest these hormone markers may also be related to autism, warranting further study.