Acute lymphoblastic leukemia (ALL) is the most common childhood malignancy. There is compelling evidence that childhood ALL is often initiated in utero and ALL risk is affected by neonatal immune status. Children who developed ALL tend to have had more fulminant infections during the first year of life, despite having had less frequent exposure to common infections than controls. One plausible explanation for this apparent immune dysregulation in children who develop ALL is aberrant in utero programming of the immune system. There are several lines of evidence that support this hypothesis. First, we have observed that children who develop ALL later in life had lower levels of IL-10, an immunosuppressive cytokine, at birth. Second, both maternal infections during pregnancy and higher maternal immunoglobulin E (IgE) levels after birth are positively associated with ALL risk, suggesting a role for maternal modulation of fetal immune development in determining ALL risk. Finally, we have identified chemicals associated with increased risk of childhood ALL, some of which are potential modifiers of immune status, including polychlorinated biphenyls, polybrominated diphenyl ethers, polycyclic aromatic hydrocarbons, and tobacco smoke. The objective of this proposed project is to assess, for the first time, the interplay between in utero chemical exposures and the immune status of both the mother and the child in the development of childhood ALL. We hypothesize that specific in utero chemical exposures will impact maternal and neonatal immune status and the altered immune status increases the risk of childhood ALL. To test this hypothesis, we will leverage exceptional resources from two existing NIH-funded studies (through Core A): neonatal dried blood spots collected at birth (neonatal blood spots) from the California Childhood Leukemia Study (CCLS), a population-based case-control study, and paired samples of neonatal blood spots and second trimester maternal serum from the California Mother-Child Birth Cohort. In both populations, we will (1) characterize immune status by measuring immunomodulatory cytokines in biospecimens, (2) examine whether in utero chemical exposures (identified in Project 2) influence immune status, and (3) assess the relation of in utero chemical exposures to the risk of childhood ALL, while accounting for immune status. In the CCLS population, home dust measurements and self-reported chemical exposures will be used to further characterize in utero chemical exposures, and data on maternal infections during pregnancy will be used to help characterize maternal immune status. An engineered mouse model of childhood ALL will be used to elucidate mechanisms by which in utero chemical exposures and maternal/neonatal immune status interact to initiate ALL (through Core C). In addition, data on maternal/neonatal immune status will be incorporated in the analysis of DNA methylation (Project 3). Our proposed project will fill important knowledge gaps in the etiology of childhood ALL, and further the understanding of immunomodulatory effects of in utero chemical exposures.

Though important findings regarding possible risk factors of acute lymphoblastic leukemia (ALL) have been made by CIRCLE and other childhood leukemia (CL) investigators, the timing and mechanisms by which the reported chemicals cause childhood ALL are only partially characterized, and additional risk factors of CL are likely to be discovered using evolving “omics” technology. Because children generally present with childhood leukemia during the first 5 years of life, in utero exposures should be a major focus of investigations regarding the etiology of childhood leukemia. We hypothesize that the ‘fetal exposome’ (representing all fetal exposures) affects causal pathways in childhood leukemia. Using archived neonatal blood spots (ANBS) from California, we propose to measure chemicals representing fetal exposures from 400 ALL cases and 800 matched controls combined from two epidemiologic studies, the California Childhood Leukemia Study (CCLS) and the California Mother-Child Birth Cohort (CA Birth Cohort). Analyses will employ liquid chromatography-high resolution mass spectrometry (LC-HRMS) to perform untargeted omics of small molecules and adducts of reactive electrophiles with human serum albumin (HSA) at cysteine amino acid at position 34 (‘Cys34 adductomics’). This data-driven design will allow us to characterize broad classes of maternal/fetal exposures received during gestation from diverse sources including pollution, diet, drugs and endogenous processes. By comparing omic profiles between childhood leukemia cases and controls, we will find discriminating chemical features in ANBS that will be annotated and used with data in Projects 1 and 3 to investigate causal pathways. In a separate analysis we will compare profiles of small molecules and Cys34 adducts in ANBS with those matched to maternal blood, collected during pregnancy from 200 ALL cases and 400 controls, and will test for associations between maternal exposures and childhood ALL. Finally, while conducting omics analyses of small molecules and adducts, we will target particular biomarkers of interest, including benzene (a known human leukemogen), coffee, smoking, alcohol consumption and reactive oxygen species (ROS). Given the dearth of existing data regarding fetal and maternal exposures during pregnancy, this project offers an exceptional opportunity to explore early-life causes of childhood leukemia and to validate our untargeted methods for pursuing these aims with ANBS. Since other States maintain ANBS for research purposes, a successful outcome from this project would provide important impetus to develop, maintain and exploit repositories of ANBS for investigations of all childhood diseases in the U.S.

Pre-B cell leukemia is the most common cancer in children, and though treatable in most cases, the disease leads to long-term morbidity. Preventing leukemia requires an understanding of its causes. In the previous funding cycle, we found that childhood leukemia tumor cells are profoundly altered from pre-B cell precursors with regards to DNA methylation; i.e., methylation of fully 10% of CpG sites is altered during leukemogenesis. Environmental risk factors, including chemical factors (polycyclic aromatic hydrocarbons) and dietary factors (folic acid), which were found to impact childhood leukemia risk in our epidemiology study (California Childhood Leukemia Study, CCLS), also affect DNA methylation. Genetic risk factors impact DNA methylation locally and genome-wide. While researchers have increasingly focused on the environmental and genetic causes of variation in DNA methylation, few have accounted for the impact of the combination of the two factors. We will address this knowledge gap in the current proposal. First, we will use genome-wide DNA single nucleotide polymorphism (SNP) data to assess the impact of genetic variation on DNA methylation among 200 acute lymphoblastic leukemia (ALL) cases and 400 controls from the California Mother-Child Birth Cohort. A second set of 200 ALL cases and 400 controls of CCLS subjects will be used for replication and meta-analysis. As a second aim, we will use the same two case/control sets to investigate the effects of both genetic and environmental factors on DNA methylation and ALL risk. Chemical risk factors for ALL identified in Project 2 during the next funding cycle, chemical risk factors for ALL that were previously identified by our research group, and immune risk factors for ALL identified in Project 1 will be used in the proposed Project 3 statistical analysis. The biospecimens from 400 cases and 800 controls (two sets together) used in Project 3 will completely overlap with those assessed for cytokines in Project 1 and those assessed for protein adducts and small molecules in Project 2. Finally, we will use a mouse model that recapitulates the features of human childhood leukemia to directly assess the effects of particular in utero chemical exposures on DNA methylation in pre-B cells (our target cell population), as well as on leukemogenesis.

Previous case-control studies of childhood leukemia mostly relied on self-reported exposure which lack specificity and may suffer from recall bias. The proposed study builds upon a large case-control study, the Northern California Childhood Leukemia Study (NCCLS) to improve chemical exposure assessment, using available home dust samples and a variety of biospecimens. Preliminary NCCLS findings suggest that house dust can provide useful quantitative surrogates for in-home exposures to toxic contaminants. In order to assess whether persistent contaminants such as nicotine (a surrogate for ETS), polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs), and polybrominated diphenyl ethers (PBDEs) accumulate in house dust over several years, the proposed study will obtain an additional household-dust sample from each of 150 homes in the NCCLS population for which we have existing dust samples. To further validate the use of house dust as a measure of children's exposures to toxic substances, we will measure nicotine, PCBs, and PBDEs in serum samples obtained from about 250 childhood leukemia cases at diagnosis and then determine correlations between analyte levels in serum and house dust. In addition, the NCCLS obtained archived newborn dried blood spot (DBS) collected at birth. Since blood and DBS contain adducts of potential carcinogens with hemoglobin (Hb) and human serum albumin (HSA), they offer opportunities for quantifying children's exposures and internal doses during one or two months prior to collection. Using methods developed in our laboratory, we will detect and profile cysteinyl adducts of HSA in pre-treatment diagnostic blood from children with leukemia and in newborn DBS from a subset of 200 children (100 cases and 100 controls; these subjects will be among those investigated for DNA methylation patterns in Project 3). By comparing DBS-adduct profiles between childhood leukemia case and control children, we will detect particular adducts that are associated with disease status. Then, after chemically identifying these adducts and their likely precursors, we will pinpoint early life exposures that increase the risk of childhood leukemia, and possible changes in levels of important adducts between birth and diagnosis.

At the cellular level, pediatric leukemias, like other cancers, are caused by genetic and epigenetic alterations. Distinct subtypes of cytogenetic abnormalities have been recognized in childhood ALL for many years and are cornerstones of leukemia patient management and more recently, epidemiology studies. Recent studies on the epigenetic of leukemia suggest that alterations including DNA methylation are common and potentially important features of childhood ALL. Our preliminary studies and those of others have shown that DNA methylation events in leukemia are complex. Some events are reflective of the normal developmental epigenetic programs and therefore the events simply mirror the normal biology of the progenitor population of cells that were transformed. Other events are acquired during the transformation process. Some may arise from early epigenetic aberrations resulting from environmental exposures and developmental abnormalities. In this project we will attempt to delineate the critical DNA methylation events in pediatric leukemias and furthermore define those influenced by the environment, particularly in the prenatal period. The development of a panel of exposure- and leukemia-related biomarkers is the goal, which could be utilized in future biomonitoring, risk stratification, and epidemiology studies. We will characterize the DNA methylation pattern of normal B-cell progenitors and a series of leukemia bone marrow samples (n=250) using a high dimension technology, to define the DNA methylation events critical in leukemia. Second, we will utilize the same technology to define the DNA methylation pattern at birth on same 250 leukemia cases and their case bloods using neonatal dried blood spot (DBS) DNA. Using novel statistical techniques, we will associate DNA methylation patterns and individual genes to specific and quantitative data on environmental exposures including a chemical analysis of over 120 analytes in house dust, with repeat measurements at 150 houses, and an assessment of albumin-chemical adducts in 200 subjects (from Project 2). Using this information, we will validate a subset of the most informative (i.e., leukemia- and environmental-associated) methylation markers using a distinct set of neonatal DBS that are part of the Northern California Childhood Leukemia Study and a new set of cards from the California Department of Public Health, and seek to understand the role of DNA methylation in gene-environment interaction in leukemia-associated DNA methylation at birth.

The proposed research builds on 13 years of experience with the Northern California Childhood Leukemia Study (NCCLS), a population-based case-control study which includes ~40% of Hispanics, a traditionally under-represented ethnic group. The proposed study will continue to use a multi- and transdisciplinary approach to comprehensively study environmental, molecular, genetic, biologic and epidemiologic factors of childhood leukemia, and to improve chemical exposure assessment. Approximately 1,000 childhood leukemia cases [830 acute lymphocytic leukemia (ALL) and 160 acute myelocytic leukemia (AML)] and 1,230 matched controls have been enrolled with DNA for genotyping available in ~95% of subjects. The investigations proposed will examine how environmental and genetic risk factors for ALL interact, and explore the etiology of rare and understudied subtypes of childhood leukemia such as AML and other groups defined by molecular and epigenetic markers. To achieve adequate statistical power, we propose to enroll an additional ~660 childhood leukemia cases (~530 ALL; 85 AML) and ~660 matched controls using an established comprehensive, rapid reporting network of pediatric oncologists in the 38 California study counties. The resulting total sample size of ~1,660 leukemia cases (1,360 ALL, 245 AML) and 1,890 matched controls will allow us to explore hypotheses related to pre- and post-natal residential exposures to pesticides, persistent organic pollutants, and sources of benzene exposure and increased risks of ALL and AML, and how these risks may vary by leukemia subtype and be modified by variants in metabolizing genes. We will also examine the influence of immune function on the risk of childhood ALL overall, B-cell ALL and common-ALL (c-ALL) subtypes, and how these associations are modified by genes involved in the regulation of immune function. We will conduct large scale genotyping of ~10,000 gene variants, and employ a number of recently developed approaches for analysis of complex multidimensional genetic and epidemiologic data.

This project builds upon the existing Northern California Childhood Leukemia Study (NCCLS), a large case-control study with over 1,000 cases. Its overall goal is to determine the role exposure to environmental chemicals plays in the etiology of childhood leukemia. We aim to examine the potential role benzene, polycyclic aromatic hydrocarbons (PAH) and other selected Superfund chemicals, such as trichloroethylene (TCE), play in the etiology of childhood leukemia in California. We will use state-of-the-art exposure assessment techniques to accomplish this goal. We further propose to use gene expression profiling and proteomics to develop new biomarkers that aid in the classification of etiologic subtypes of this disease and help find its cause(s). Our hypothesis is that certain subtypes of leukemia will have specific environmental causes. Specifically, we plan to: 1) Extend case ascertainment through 2009, such that biological samples will be available from over 1000 cases; 2) characterize the childhood leukemia cases by proteomics and gene expression profiling; 3) collaborate with the CDC to measure more than 30 volatile hydrocarbons in the blood of mothers of children with leukemia and control mothers (we will compare these measures with results from concomitant home air and water sampling and with surrogate markers of past exposures to VOCs using self-reported exposure to tobacco smoke and estimated traffic density); 4) measure protein adducts of benzene and naphthalene (a representative PAH) in serum from mothers of cases and controls in collaboration with Dr. Rappaport of the North Carolina SBRP; and, 5) measure protein adducts of benzene and naphthalene in the plasma of children with different forms of leukemia and correlate these measurements with the chemical exposure analysis conducted with household samples. The strengths of the study are: a) Large sample size; b) extensive residential exposure assessment; c) measures of current and cumulative exposure to Superfund chemicals in both case and control mothers and children; d) ability to compare exposures to national NHANES data; e) availability of carefully processed biological samples; and, f) application of sophisticated -omic technologies to case characterization. These studies should provide new insights into the role of chemical exposure in the etiology of childhood leukemia and produce new biomarkers of exposure and disease status that can be used in future studies.

Previous attempts to link leukemia risks to children's home exposures relied upon indirect measures of exposures to carcinogens, based on interviews of households and outdoor levels predicted by Geographic Information Systems (GIS). Such indirect exposure measures can lead to systematic biases in estimated relative risks, associated with particular carcinogens. Also, children's exposures to carcinogens vary randomly within households over time, and thus can introduce attenuation biases into exposure-risk analyses. Given the likely magnitudes of these uncertainties in children's exposures to carcinogens, it is surprising that no studies have characterized systematic and random errors in children's home exposures. Here, we propose a validation/reproducibility study in which we will collect repeated air and dust samples in 250 households selected from a large case-control study of childhood leukemia, namely, the Northern California Childhood Leukemia Study (NCCLS). We will estimate the magnitudes of both systematic errors and random errors and calibrate interview/GIS based exposure surrogates against 'true' contaminant levels measured in air/dust samples. Then, we will use the errors estimated from the validation study to correct naïve estimates of childhood leukemia risk (overall and by subtype) associated with particular carcinogens in the main NCCLS (1,100 cases and 1,650 controls). These analyses will also allow us to identify possible time trends and seasonal changes in exposure levels during critical time windows of gestation and early childhood development. We will focus upon 40 known or suspected carcinogens in house dust and indoor air, including pesticides, persistent organic compounds (polychlorinated biphenyls, polycyclic aromatic hydrocarbons, and nicotine, a surrogate for tobacco smoke), and volatile organic compounds. In each household, a total of three samples of vacuum-cleaner dust and indoor air (using passive air monitors) will be tested for the respective persistent and volatile carcinogens. This proposal presents an exciting opportunity to identify and correct for potential biases of estimated childhood leukemia risks, associated with carcinogen exposures in the NCCLS population. The results would not only clarify the potential importance of particular carcinogens to childhood leukemia risk but also would provide unique information regarding home exposures to carcinogens that can be used in environmental epidemiology and risk management.