Hypothesis 1. Autism risk is associated with exposure to intrauterine cytokines and/or hypoxia.

Autism and other neuropsychiatric risks have been linked to intrauterine exposures to cytokines and/or hypoxia. These studies are limited in that they have been forced to rely upon retrospective cohorts, maternal serum samples, maternal medical history, etc., each of which is problematic in terms of bias (true peer "control" sample), ascertainment (retrospective assessment of maternal gestational illness), or true measure of fetal exposure (maternal serum cytokines do not necessarily imply fetal cytokinemia).

The ALSPAC cohort will allow us to avoid most of those pitfalls, in a nested case-control study that will use both standard histopathologists' definitions and image analysis, to determine the presence and severity of

* acute inflammation (most commonly a marker of intraamniotic infection),

* chronic placental inflammation (with a differential diagnosis generally considered to be congenital viral infection or maternal attack against the semi-allogeneic fetoplacental unit), and

* the villous effects of uteroplacental malperfusion (villous fibrosis, excess syncytial basophilia and knotting, infarcts, abruption, generally reflecting reduced fetal oxygenation secondary to the abnormal maternal perfusion of the placenta and/or damage to the placenta).

Hematoxylin and eosin stained slides (and special stains as indicated) will be scored by a team of 3 trained and expert pathologists, and the digitized slides will be evaluated using sets of segmentation and quantification algorithms developed to measure inflammation and hypoxia.These semiquantitative (pathologist') and continuous (image analysis) variables will be correlated with case-control status, with adjustment for gender, gestational age and birth weight at delivery using structual equation models in which the placental histology features are considered to be indicators of underlying latent variables (e.g., maternal acute inflammatory responses, fetal acute inflammatory responses, maternal malperfusion). The model including all

All cases of autism identified in ALSPAC (n=56) will be included in this nested case-control study. We will select four controls for each case to maximize statistical power of the comparisons. The gain in statistical power with the use of additional control subjects is negligible. The large size of the ALSPAC cohort allows us to match on a few key variables that may be differentially distributed among the cases. We propose to match on infant sex and gestational age (=/- 1 week ideally, 2 weeks if necessary). Prematurity is not a major risk factor for autism so we expect matching on gestational age to be feasible.

2. Placental vascular branching is a biomarker for Autism.

Autism is a disorder of abnormal neuronal connectivity and, by extension, abnormal neuronal dendritic branching. The placenta is an organ that is architecturally little more than a thin tissue sheath covering a vascular tree. It is uniquely dependent on coordinated and accurate paternal gene expression, and its growth is regulated by gene families that regulate events that may be key to the connectivity issues described in autism. We hypothesize that the vascular architecture of the placenta is a proxy for aspects of neuronal differentiation, elongation and navigation that are germane to the genesis of autism risk. Thus autism risk will have parallel placental findings of altered villous and vascular branching.

Neurons and the Vasculature: Shared Pathways for Outgrowth, Proliferation and Branching Diversity. Placental blood vessels provide oxygen and nutrients to and remove metabolic products from the embryo/fetus. It has been recognized that the signals and guiding principles to differentiate, elongate and navigate blood vessels towards their destinations are analogous to those in neuronal development. Vascular endothelial growth factor (VEGF) promotes vascularization of the placenta as well as neuronal differentiation. Similarly, a number of factors promote endothelial proliferation and angiogenesis; moreover, their cross-talk-and in particular their defective cross talk-may be at the root of a wide range of both vascular/ neurological and placental diseases .

Antiangiogenesis and Autism: Thalidomide and Valproate. Thalidomide and valproate exposures in early pregnancy are recognized risk factors for autism. Animal models exposed in utero to valproate demonstrate neuroanatomical changes in the brainstem and cerebellum that are comparable to those in brains from children with autism. Both of these drugs are antiangiogenic as well as teratogenic. Mechanisms of such damage involve disruption to vascular growth factors, i.e., VEGF and placental growth factor (PlGF) and their associated signal transduction pathways. Note, VEGF and PlGF are produced and released by the placenta.

Hypothesis: Placental Vascular Branching is an Indicator of Normal Fetal Neuronal Connectivity. Autism is a disorder of abnormal neuronal connectivity and, by extension, abnormal neuronal dendritic branching. The placenta is an organ that is architecturally little more than a thin tissue sheath covering a vascular tree. It is uniquely dependent on coordinated and accurate paternal gene expression, and its growth is regulated by gene families that regulate events that may be key to the connectivity issues described in autism. We hypothesize that the vascular architecture of the placenta is a proxy for aspects of neuronal differentiation, elongation and navigation that are germane to the genesis of autism risk. Thus autism risk will have parallel placental findings of altered villous and vascular branching.

Measuring Placental Vascular Abnormality. The placenta is accessible (delivered with each baby born), and can be measured reliably. In a series of papers, we have developed novel methods for the study of the placental vascular arborization. We have introduced a dynamic model of the placental vascular growth based on a well-studied fractal random growth process (DLA). The model has allowed us to associate an abnormal morphology of the placenta with the abnormal vascular branching and connectivity. The standard marker of the branching density of a fractal tree is a version of a fractal dimension. While mathematically elegant, it is difficult to quantify in a practical study. We proposed a new idea, relating the placental vascular branching structure with a metabolic scaling exponent. This quantity is easy to measure as a ratio of logarithms of the fetal birth weight and the placental weight, and is a biologically relevant version of the fractal dimension of the vasculature. Our studies have demonstrated an association of the scaling exponent with the known factors affecting the normal vascular development.

Preliminary Findings on a Female-Specific Risk Factor for Autism: Advanced Paternal Age. Autism and Autism Spectrum Disorders have a strong genetic component, and a predilection for males. Recently, a risk factor for autism, advanced paternal age, was demonstrated to have an effect confined to female offspring, who inherit their second X-chromosome from their fathers [18]. The operative hypothesis is that the X-chromosome, which contains many genes associated with cognition and intelligence, may acquire either mutations or epigenetic alterations over the father's lifespan that then alter those gene's expression in his daughters. We have recently identified a female-specific negative effect of advanced paternal age on birth weight in the National Collaborative Perinatal Project, a well studied national cohort with over 24,000 cases with placental data. The placentas were thinner in this case, indicating a less arborized vascular tree.

Methods of the Proposed Research.

a. Placental Processing: Placentas will be processed according to protocols that have been reviewed and agreed to by the Pathology Department University of Bristol Hospital. Specific features include:

1. Digital photography of the chorionic surface and of slices of the placental disk.

2. Tissue sampling to include:

2 cross sections of the umbilical cord

1 membrane roll

4 samples of grossly normal placenta obtained from each placental quadrant

2 "functional units" (2.5 cm square full thickness regions with a terminal chorionic surface vessel)

Digital photographs will be uploaded to the Placental Analytics FTP site for analysis. Tissue samples will be processed to wax blocks. The blocks will be shipped to Placental Analytics, LLC for slide preparation.

Data extracted from the digital photographs and copies of the files of all digitized histology slides prepared from the placental tissue samples will be returned to ALSPAC as they are generated throughout the course of the study.

Tissue blocks will be retained by Placental Analytics, LLC until completion of the study at which point they will be returned to ALSPAC.

Placental Vascular Analyses: A key to our analysis of the placental vascular branching structure is the connection between the arborization of the vascular tree and the metabolic scaling exponent (a biologically relevant fractal dimension), as well as the morphological factors, such as the shape and the size of the placenta. Using the DLA model of placental angiogenesis, we can connect the variability in the placental shape with a perturbation in the placental vascular architecture. Fourier analysis of the two-dimensional images of the chorionic plate can be used to reliably identify such perturbations which appear early in gestation, and are hypothesized to be connected to Autism risk. An early reduction of the placental angiogenesis is also related to a larger value of the metabolic scaling exponent - a quantity which is particularly easy to measure at the time of delivery, with implications for an early diagnostic of the related Autism risk factors.

We have begun the analysis of 3-dimensional placental shapes, beginning with the study of the thickness and variability of slices perpendicular to the chorionic plate. The thickness of the slices, in particular, is associated with the vascular density. Our preliminary findings show that for female babies, the average thickness is adversely affected by the advanced paternal age, which is a known Autism risk factor. The variability of thickness is also associated with the suppressed angiogenesis in the DLA model [16]. We have also initiated the study of the skeletonized structure of the placental vasculature both for DLA models (with the skeletons of vascular trees traced automatically) and for digital photographs of the chorionic surfaces (with manually traced skeletons). We have developed measures of average density of the large and medium blood vessels in the placental surface, which complement the techniques described above, with similar results. We plan to further this approach by developing automated digital recognition of the large-scale branching structure in the chorionic plate surfaces.