Project 1 Update Archive


The overall goal of our project is to use biomarkers to improve the risk assessment of benzene at Superfund sites and ambient locations close to known point sources of benzene.  Earlier research indicated that benzene is metabolized by two pathways in humans.  One of these metabolizes benzene only at low environmental exposure levels.  This year we have shown that this pathway is more active in females than males and is inactivated or otherwise inhibited by smoking.  This has important implications for calculating the risk from benzene exposure at low levels of exposure.  We have also shown that the expression of genes in the pathway to myeloid leukemia is changed at low levels of benzene exposure and that there is a group of genes that are consistently altered.  This may provide a gene expression signature for examining effects of benzene at low exposure levels.  These data also suggest that benzene poses a health risk at doses lower than previously thought.


The overall goal of our project is to use biomarkers to improve the risk assessment of benzene at Superfund sites and ambient locations close to known point sources of benzene.

Use toxicokinetics-based statistical models to assess the effects of gender and smoking status on the metabolism of benzene in humans.

We previously showed that benzene metabolism was 9-fold greater at air concentrations below 1 ppm and  that benzene is apparently metabolized by two pathways, one of which accounts for about 70% of the metabolism below 0.1 ppm.  However, following reanalysis of our data (that had been obtained under the Freedom of Information Act), Price et al. (Carcinogenesis, 33, 2094-9, 2012) argued that our conclusions regarding enhanced benzene metabolism below 1 ppm were not justified. In response, we conducted additional analyses and showed that the methods and arguments presented by Price et al. were scientifically unsound and that their results were unreliable (Rappaport, S.M., et al. Carcinogenesis, 34, 2-9, 2013).  Price et al. (Carcinogenesis, 34(7), 1692-6, 2013) then provided additional calculations – based on our response – to suggest that only trivial fractions of benzene were metabolized below 1 ppm and also argued that our original results were biased due to incorrect selection of low-exposed subjects for estimation of background levels of metabolites arising from catabolism.  In our rebuttal (Rappaport, S.M., et al. Carcinogenesis, 34(7), 1689-91, 2013) we showed that Price et al.’s new calculations were also erroneous and that the ‘correct’  calculations provided further support for our conclusions regarding low-dose benzene metabolism.

To interpret our prior results regarding low-dose benzene metabolism, we hypothesized the presence of a second metabolic pathway (in addition to CYP2E1) that metabolizes benzene at air concentrations below 1 ppm (S.M. Rappaport et al.,  Environ Health Perspect 2009, 117, (6), 946-52; S.M. Rappaport,  Chem Biol Interact 2010, 184, (1-2), 189-95).   The hypothesis was tested by fitting Michaelis-Menten-like kinetic models to urinary metabolite data from non-smoking women in the China study.  Early attempts at including male subjects and smoking subjects produced models that failed to converge, in part because most Chinese males and virtually none of the females were smokers. We have now extended our one-pathway and two-pathway models to levels of muconic acid in urine from males and females, smokers and non-smokers from both the China study and an independent investigation of  Italian subjects who were exposed to low benzene concentrations (S. Fustinoni, et al. 2005, Cancer Epidemiol Biomarkers Prevention 14: 2237-2244). (Muconic acid was the only benzene metabolite measured in the Italian subjects).  Using weights of evidence based upon AIC values from the competing models, we find strong evidence that the two-pathway model provides a better fit than the one-pathway model for the production of the muconic acid in both male and female non-smokers from the two populations.  Efforts at modeling data from smokers has been complicated by confounding due to the presence of benzene in mainstream tobacco smoke.

Examine the ability of CYP2F1 and CYP2A13 to metabolize benzene and determine how genetic variation in these enzymes affects benzene metabolism.

Work on this aim continues.

Determine the effect of variation in candidate genes on the hematotoxicity and genotoxicity of benzene in two cohorts of benzene-exposed workers using high-throughput genotyping, fine mapping and resequencing. 

The laboratory analyses using high throughput sequencing for this aim are in progress and have been delayed by QC issues and the addition of more samples to increase power.

Validate changes in the expression of a panel of 16 highly responsive genes in the blood mononuclear cells of benzene-exposed workers using RT-PCR.

Recent developments have shown that transcriptome sequencing (RNA-Seq) has a greater dynamic range than microarrays and the potential to reveal additional transcriptional information. Thus, to increase the innovation potential of our project, we are using RNA-Seq to analyze the transcriptome of mononuclear cells from benzene-exposed and unexposed study subjects. We sequenced RNA from 10 workers occupationally exposed to high levels of benzene (≥5 ppm) in air and 10 matched unexposed control workers, from a large study (n = 125) in which gene expression was previously measured by microarray (Thomas R et al., Env Mol Mutagen 2013). The Pearson correlation between RNA-seq and microarray signals for the 20 subjects was around 0.6. 60% of the transcripts with detected reads from the RNA-seq experiments did not have corresponding probes on the microarrays. Fifty-three percent of the transcripts detected by RNA-seq and 99% of those with probes on the microarray were protein-coding. There was a significant overlap (P < 0.05) in transcripts declared differentially expressed due to benzene exposure using the two technologies. About 20% of the transcripts declared differentially expressed using the RNA-seq data were non-coding transcripts. Six transcripts were determined (false-discovery rate < 0.05) to be alternatively spliced as a result of benzene exposure. Overall, this pilot study shows that RNA-seq can complement the information obtained by microarray in the analysis of changes in transcript expression from chemical exposures and has the potential to reveal a much more comprehensive transcriptome signature. A much larger study is now underway.

Measure benzene-induced changes in global and gene-specific DNA methylation and microRNA expression in the blood cells of benzene-exposed workers.

MicroRNA expression has been analysed using Affymetrix 3.0 GeneChips which contain probes for 1733 mature human miRNAs on RNA isolated from all 125 subjects analyzed previously by gene expression profiling. The statistical analysis is ongoing. The profiling of DNA methylation is highly dependent on the cell types present in a mixed sample (e.g. whole blood) such that artifactual results can be obtained if one does not account for changes in cell make-up. We have therefore decided to analyze purified granulocyte DNA as this is myeloid in origin.

Compare DNA methylation profiles of blood cells from workers with different status of benzene exposure (current, past, or none).

Granulocyte DNA is being analysed.

Overall Project Significance

We have found very strong statistical evidence for two pathways of benzene metabolism with one operating maximally at environmental levels of exposure.  Further evidence of enhanced low-dose benzene metabolism has been derived from published measurements of benzene in inhaled and exhaled air. We have also shown effects of benzene on gene expression at low levels. These data suggest that benzene poses a health risk at doses lower than previously thought.

Future Plans

We will continue to address all specific aims. We are taking advantage of the lower cost and higher throughput of DNA sequencing to better address our original aims and are using the latest information to avoid artifacts in studies of epigenetic changes in the blood of exposed workers.



Currently under review.



Currently under review.



Childhood leukemia is the most common form of childhood cancer affecting 3,000 children annually in the U.S. and many more worldwide. The causes are essentially unknown. The major form of childhood leukemia is ALL (acute lymphoblastic leukemia) and the other is AML (acute myeloid leukemia).

Overall goals
The goal of Project 1 is to determine the role chemical exposures play in the development of childhood leukemia. In one of the largest studies of its kind in the world, bone marrow and/or peripheral blood was obtained on 961 cases of childhood leukemia and peripheral blood and urine obtained from many of the mothers. These have been analyzed with the latest technologies.

Important discoveries so far

  1. Project 1 investigators examined whether the use of paint and petroleum solvents at home before birth and in early childhood influenced the risk of leukemia in children. They found that ALL risk was significantly associated with paint exposure, with a higher risk observed when paint was used post-natally, by an adult other than the mother, or frequently. They found no significant association between solvent use and ALL risk overall, but a significant association with AML risk was observed.
  2. The investigators measured blood levels of selected Superfund chemicals in the blood of mothers of children with leukemia and control mothers, including 33 volatile organic compounds (VOCs), and urine levels for 10 cancer-causing polycyclic aromatic hydrocarbons (PAHs) and cotinine (a biomarker of smoking). House dust was collected simultaneously and a significant positive correlation between levels of nicotine in house dust and those of serum cotinine (a nicotine metabolite) in maternal blood samples was found, showing that nicotine in house dust is a good surrogate marker of adult smoking in the home. House dust was also used as an exposure indicator to examine the risk of childhood leukemia in relation to residential exposure to persistent organochlorine chemicals, such as polychlorinated biphenyls (PCB). Detection of any PCB congener in the dust conferred a 2-fold increased risk of ALL.
  3. Earlier studies had shown that most causal genetic changes found in the children’s leukemia were present before they were born. Recent studies have shown that secondary changes, including those in a cancer gene called RAS, occur post-natally after the child is born.

What we plan to do next
The project has now become part of a new NIEHS/EPA Children’s Environmental Health Center, called the Center for Interdisciplinary Research on Childhood Leukemia and the Environment (CIRCLE) lead by Dr. Buffler.



We investigated the use of nicotine concentrations in house dust as measures of children’s exposure to cigarette smoke in 469 households from the Northern California Childhood Leukemia Study (1999-2007).  House dust was collected using high-volume surface samplers and household vacuum cleaners and analyzed for nicotine via gas chromatography-mass spectrometry.  Nicotine concentrations in house dust were associated with self-reported smoking for periods of months and years before dust collection, and with age of the residence.  Furthermore, the authors found that the relationship between nicotine dust levels and self-reported smoking varied by parental age and socioeconomic status. These findings suggest that house-dust nicotine concentrations reflect long-term exposures to cigarette smoke in the home (published in American Journal of Epidemiology, 2009).  Furthermore, we observed a significant positive correlation between levels of nicotine in house dust and those of serum cotinine (a nicotine metabolite) in 110 maternal blood samples.

We examined determinants of polycyclic aromatic hydrocarbons (PAHs) in home dust samples, and reported correlations with age of the residence, gas heating, and outdoor source of PAHs. Yet, these home characteristics were marginally predictive of observed PAH levels, accounting for only 15% of the variation in PAH levels in house dust.  Therefore, we recommend that PAH concentrations be measured directly in dust samples for use in epidemiologic studies (In Press, Journal of Exposure Science and Environmental Epidemiology).

In biological studies, we analyzed genome-wide DNA methylation in patients from the NCCLS with different subtypes of acute myeloid leukemia (AML) using Illumina’s Infinium HumanMethylation27 BeadChip microarray. This assay interrogates more than 27,000 highly informative CpG sites (~14,000 genes) per sample at single-site resolution, in a 12-sample throughput format. DNA methylation status is more stable than gene expression and can be examined with small amounts of DNA rather than RNA. Overall, methylation profiles differed among the leukemic subtypes, with some apparent similarity between the alterations involving core-binding transcription factors [inv(16), t(8;21)]. Analysis of differential methylation relative to the karyotypically normal subtype, revealed many (~50 at q<0.01) differentially altered CpG sites, in each subtype. This altered methylation status may reflect prior environmental exposures, an area that is being further explored.



The most common mutation in childhood leukemia is a translocation between chromosomes 12 and 21 which fuses the TEL and AML1 genes causing the B cells to grow abnormally as a cancer. Earlier work from this project confirmed that this mutation is a prenatal mutation in most children arising when the child is still a fetus. Other mutations seem to be necessary, however, for the generation of full blown leukemia.  One such frequent mutation in this subtype is partial deletion of the short arm of chromosome 12 called del(12p) which deletes the second copy of the TEL gene. Drs. Patricia Buffler and Martyn Smith sequenced the DNA at nine breakpoints in the chromosome 12 deletions of six childhood leukemia cases to explore the etiology of this genetic event.  The majority of these breakpoints were located in highly repetitive regions of the DNA. This stands in contrast with the TEL-AML1 translocation which infrequently involved breakage in DNA repeats. The two events are therefore quite different and may have different causes. The data are also compatible with a two-stage natural history: TEL-AML1 occurring prenatally, and del(12p) occurring postnatally in more mature cells.  These findings have recently been published in the December 2008 issue of Cancer Research.

The researchers are exploring the causes of the DNA mutations which produce childhood leukemia by asking the parents about their exposures to household chemicals, such as paints and solvents. They recently reported in a paper that is on-line in Environmental Health Perspectives that acute lymphoblastic leukemia (ALL) risk was significantly associated with paint exposure with a higher risk observed when paint was used postnatally, by a person other than the mother, or frequently. Further, the association was restricted to leukemia containing translocations between chromosomes 12 and 21. No significant association was found between solvent use and ALL risk overall but, a second rarer form of childhood leukemia, acute myeloid leukemia (AML) was associated with solvent but not with paint exposure. The association of ALL risk with paint exposure was strong, consistent with a causal relationship, while further studies are needed to confirm the association of AML risk with solvent exposure.



The overall objective of The Biomarkers of Chemical Exposure and Leukemia Risk project is to identify environmental risk factors for various subtypes of childhood leukemia, using novel biomarkers to improve chemical exposure assessment and disease classification. Drs. Buffler and Smith are specifically examining the potential role of the chemicals benzene, polycyclic aromatic hydrocarbons and trichloroethylene. The Northern California Childhood Leukemia Study (NCCLS) completed a full year of participation by its nine collaborating hospitals in 2007. As of November, 2007, 1599 newly diagnosed childhood leukemia cases have been reported to the study. Of these cases, 1239 were eligible, 1071 have consented (i.e., 83% acute lymphoid leukemia [ALL] and16% acute myeloid leukemia [AML]), and 906 have been interviewed. In addition, the NCCLS has interviewed 1067 control subjects. Cytogenetic classification for leukemia cases enrolled in the study is ongoing by the Toxicogenomics Labratory Core.

Using this cytogenetic data, preliminary statistical analyses were conducted to investigate the risk of ALL and home use of paints and petroleum products overall (n=550 ALL cases) and by cytogenetic subtype (available for 70% of ALL to date). Use of paint in the house confers odds-ratios (ORs) of 1.63 (95% confidence interval [CI] 1.24-2.14) for all ALL cases combined, OR= 1.68 (95% CI 0.81-3.49) for ALL cases with “normal” karyotype (n=80), whereas OR=3.91 (95% CI 1.54-9.90) for ALL cases with translocation t(12;21) (n=63). No association was found between petroleum solvent use and ALL overall or for various cytogenetic types. In contrast, an association was observed between AML (n=100) and home use of petroleum solvents (OR=2.54, 95% CI 1.19-5.42), but not paint. Risks were the highest for cases with karyotypes known to confer a good prognosis (i.e., any translocation involving chromosome 21, t(15,17), or inversion of chromosome 16) with OR=4.28, 95% CI 1.04-17.6; these analyses are based on a small number of cases (N=25) and will be updated when cytogenetic characterization is completed for all AML. Associations found between paint exposure and ALL risk and between petroleum solvent exposure and AML risk diverged markedly by cytogenetic subtypes, suggesting that specific pathways and genetic susceptibilities might explain the increased risk (Abstract presented at the 2007 SBRP meeting, Durham, NC and submitted to AACR).

The findings above are based on self-reports of exposures to paints and solvents. In order to improve this exposure assessment, project leaders are collaborating with Dr. Stephen Rappaport to assess the level of protein adducts formed from benzene and other compounds in newborn blood spots. Significant progress has been made in methods development for this aim.

As of October 2007, blood specimens have been collected in 300 mothers and shipped to the CDC laboratories. To date, analysis of 244 whole blood samples for 33 volatile organic compounds (VOCs) showed that, for the most part, the distributions of VOC analytes in the NCCLS samples were similar to the historical NHANES blood VOC data. Parallel to the maternal biospecimen collection, Buffler and Smith have conducted follow-back visits in 611 homes as of November 2007, and collected over 560 air samples (using 3M passive monitors placed in child’s bedroom). To date, 454 monitors have been analyzed to determine benzene concentrations, 340 for 1, 4 – dichlorobenzene and 227 for toluene. Preliminary analyses suggest that season during monitoring (p<0.0001) and having a car parked in an attached garage (p=0.02) were significant predictors of indoor benzene levels, while using mothballs was a predictor of 1, 4 – dichlorobenzene concentrations (p=0.03).



The Biomarkers of Chemical Exposure and Leukemia Risk project is in its early stages of data collection and analysis.  The Northern California Childhood Leukemia Study (NCCLS) completed a full year of participation by its nine collaborating hospitals in 2006.  As of December 1, 2006, 1471 newly diagnosed childhood leukemia cases have been reported to the study.  Of these cases, 963 have consented and 809 have been interviewed.  In addition, the NCCLS has enrolled 991 control subjects, of which 929 have been interviewed.

Project investigators have performed molecular cytogenetic classification for 543 leukemia cases using abstracted clinical data and fluorescence in-situ hybridization (FISH) screening.  About 20% had t(12;21) translocation, and 37% had hyperdiploidy.  The investigators are in the process of completing the cytogenetic classification for additional cases enrolled in the study.

Proteomic analysis of biological samples from this project has led to the identification of a potentially important biomarker pro-thymosin-a, a protein that was differentially expressed in acute lymphoblastic leukemia (ALL) versus acute myelogenous leukemia (AML) childhood leukemias.

Project investigators are analyzing 33 different volatile organic compounds (VOCs) in maternal blood samples.  To date the investigators have completed analysis of 163 maternal blood samples.  Parallel to the maternal biospecimen collection, the investigators have conducted follow-back visits in 513 homes as of November 2006, and collected over 400 air samples (using 3M passive monitors placed in childs bedroom), which are being analyzed for benzene and toluene.  Future analyses will also include 1,3-butadiene, p-dichlorobenzene, styrene, naphthalene, perchloroethylene, trichloroethylene, and chloroform.



The case-control study of childhood leukemia in Northern and Central California (NCCLS) has completed a full year of participation by its nine collaborating hospitals. Dr. Buffler and her research team have completed a total of 816 interviews on eligible and consenting childhood leukemia cases since 1995 up to October 2005. Control selection, personal interview, and follow-back (Tier 2) home visits continue to be efficiently managed by the Survey Research Center at UC Berkeley. Collection of biological specimens is progressing well. A total of 92 subjects have completed the Tier 2 visits over the last year (leading to a total of 428 subjects), which include validation of self-reported use of chemicals in the home and collection of dust and air samples. Laboratory analyses of the home samples and planning for statistical analyses related to methodological studies and risk assessment is underway.

In 2005, the NCCLS has published a total of 10 peer-reviewed journal articles, and an additional 4 manuscripts have been submitted for review. In the researchers’ evaluation of the role of infectious exposures in childhood leukemia etiology (294 cases and 376 controls), daycare attendance measured by child-hours was associated with a significantly reduced risk of acute lymphoblastic leukemia (ALL) in non-Hispanic White children (OR=0.42). In addition, self-reported ear infection during infancy was associated with a significantly reduced risk of common ALL (OR=0.32). These associations were not observed in Hispanic children which suggests an important ethnic difference. In the researchers’ evaluation of the effect of vaccinations (323 cases and 409 controls), compared to children who received two or fewer does of Haemophilus influenzae b (Hib) vaccine, those who received three or more doses had a significantly reduced risk of childhood leukemia (OR=0.55).

Project investigators examined the relationship between maternal illness and drug use surrounding the pregnancy and risk of childhood leukemia (365 case-control sets). History of maternal flu/pneumonia before, during, or after pregnancy was associated with a significant increased risk of ALL in the offspring (OR=1.89). Other maternal influences on risk of leukemia in the offspring include a reduced risk associated with iron supplement use (OR=0.67) and an increased risk associated with recreational drug use before, during, or after pregnancy (OR=1.74). In the investigators’ evaluation of reproductive history and birth characteristics (313 ALL, 53 AML cases and 460 controls), an increased risk of acute myeloid leukemia (AML) was associated with a history of miscarriage (OR=2.94). Neither birth weight, birth order, nor parental ages appeared to be an important risk factor of ALL or AML.

The researchers characterized the cytogenetics of 543 childhood leukemia patients enrolled in the NCCLS and compared the cytogenetic profiles between Hispanics and non-Hispanic Whites. The ploidy levels most frequently observed among ALL patients were high hyperdiploidy (51-67 chromosomes) and pseudodiploidy. Among B lineage ALL patients, the percentage of TEL-AML1 translocations was significantly lower in Hispanics (13%) than in non-Hispanic Whites (24%). The researchers are also currently investigating the interplay between genetic polymorphisms and environment exposures on childhood leukemia risk including folate metabolism genes and folate intake, chemical metabolism genes and parental smoking and membrane transporter genes and pesticides.



The case-control study of childhood leukemia in Northern and Central California has completed a full year of participation by its nine collaborating hospitals, with 102 newly diagnosed cases (age 0-14 years) ascertained in 2004.  A total of 732 eligible childhood leukemia cases agreed to participate in the study since 1995 up to November 2004.  Control selection, personal interview, and follow-back home visits with environmental household sampling are now being efficiently managed by the Survey Research Center at UC Berkeley.  Collection of biological specimens is progressing well with bone marrow and blood samples obtained in 72% of participating cases, and buccal cells in 99% of cases, controls, and their mothers.  Dr. Buffler and her team have completed 102 additional follow-back home visits, gathering 97 carpet dust samples, 89 air samples, and 10 windowsill dust samples.  Now that the reliability study has been completed, the researchers reduced the time between initial and follow-back interviews from 5-6 to 2-3 months to increase the number of eligible households.

The researchers evaluated the relationship between parental smoking and risk of childhood leukemia (328 cases, 417 controls).  Paternal preconception smoking is significantly associated with an increased risk of acute myeloid leukemia (AML) (odds ratio [OR]=3.8), while an association with acute lymphoblastic leukemia (ALL) is suggestive (OR=1.3, P>0.05).  Maternal smoking before, during, and after pregnancy does not appear to be a risk factor for ALL or AML.  However, paternal preconception smoking combined with maternal postnatal smoking confers a greater risk of ALL than paternal preconception smoking alone.  The researchers completed additional analyses on household pesticide use with a larger sample (382 cases and 482 controls).  Their data strongly support that in utero (OR=1.9) and post-natal (OR=1.6) exposures to indoor insecticides are critical events in the development of childhood leukemia.  The research team assessed the reliability of self-reported pesticide use in a subset (47 cases and 52 controls).  Percent positive agreement for flea control products was 79%; professional pest control, 72%; ants/flies/cockroaches, 66%; and indoor foggers, 48%.  There was no indication of differential misclassification between cases and controls, and Hispanic and non-Hispanic children.

The researchers investigated the interplay between genetic polymorphisms and exposure to traffic-related pollutants, using surrogate measures for traffic density.  Preliminary analyses suggest that children carrying the variant CYP1A1m2 allele and exposed to the highest level of traffic density had a two-fold increased risk of leukemia compared to those with the homozygous wild-type genotype and exposed to the lowest level of traffic density.  More analyses are underway to examine the interaction of metabolic gene polymorphisms (CYPs, GSTs) with parental smoking and household pesticide use, as well as MTHFR genes and dietary folate intake.  Project investigators examined the relationship of RAS gene mutation with cytogenetic, demographic, and parental smoking characteristics.  All findings described above were presented at professional meetings (manuscripts are in preparation or In Press).  Investigators also completed analyses on child (Kwan et al. 2004) and maternal diet (Jensen et al. 2004), breastfeeding (Kwan et al. 2004), vaccination history, family cancer history, daycare exposures, residential history, maternal reproductive history, drug use, and illness.



The epidemiologic study of childhood leukemia in Northern and Central California has completed a full year of participation by its nine collaborating hospitals, with 120 newly diagnosed cases (age 0-14 years) ascertained in 2003. A total of 617 eligible childhood leukemia cases agreed to participate in the study since its inception in 1995 up to August 2003. Control selection, personal interview, and follow-back home visits (including second interview for reliability study and household dust sampling), have been successfully transferred to the Survey Research Center at UC Berkeley. Collection and processing of biological specimens continues to progress smoothly with bone marrow and blood samples obtained in 72% of the eligible and consented cases, and buccal cell specimens available for 96% to 98% of cases, controls, and their mothers. In-depth analyses of the use of birth cohort for control selection indicates that participating birth controls provide a representative sample of children within the birth cohort of cases, therefore reducing potential for selection bias. Analyses using complete phase 1 and 2 data are underway to confirm preliminary findings observed with child’s diet (protective effect of orange/bananas; no association with breastfeeding), and increased risks reported with environmental factors such as household pesticide use and paternal smoking. With investigators from Project 1, Dr. Buffler’s team has submitted a manuscript showing that FHIT gene is methylated in a distinct subset of pediatric acute leukemia and may play an important role in the etiology of the acute lymphocytic leukemia. Several analyses are underway to investigate the interplay between genetic polymorphisms and dietary/environmental risk factors (e.g., MTHFR gene and mother’s folate intake; GST, CYP, MDR genes and environmental chemical exposures).



The epidemiologic study of childhood leukemia in Northern and Central California ascertained 150 newly diagnosed cases (age 0-14 years) during the last year. Control selection, personal interview, and collection of biological specimens (by Core B) have been progressing smoothly. Follow-back home visits with original respondents are currently being conducted to evaluate the reliability of reported use of household chemicals, to obtain a detailed history of pesticide use during the 12 months before the home visit, and to collect household dust samples. With investigators from Project 1, Dr. Buffler’s team has published two manuscripts detailing the prenatal origin of two translocations, t(8;21) and t(1;19), thus showing that pre-natal and early life exposures are likely to be critical in the development of childhood leukemia. Furthermore, 660 cases and controls have been genotyped for specific polymorphisms of interest. Statistical analyses continue. Preliminary results indicate that certain polymorphisms related to the metabolism of folate play a crucial role in the risk of childhood leukemia.



The epidemiologic study of childhood leukemia in Northern and Central California ascertained 173 newly diagnosed cases during the last year. Control selection, personal interview, and collection of biological specimens (by Core B) have been progressing smoothly. With investigators from Project 1 we have published a manuscript on the prenatal origin of t(8;21) translocation, showing that pre-natal and early life exposures are likely to be critical in the development of childhood leukemia. Interestingly, a statistical analysis of information collected from 162 newly diagnosed cases of leukemia (age 0-14 years) and 162 non-leukemia control children during 1995-1999 indicates that children who were exposed to household pesticides were more likely to be diagnosed with childhood leukemia, especially when the family used professional pest control services. In agreement with our molecular findings, insecticide exposures experienced early in life, i.e. during pregnancy and the first year of life, appear to be more significant than exposures in ages 1-3 years, with the highest risk being observed for exposure during pregnancy.



This project initiated an examination of the relationship between reported chemical and other environmental exposures and the risk of childhood leukemia utilizing molecular biology techniques to characterize leukemia subtypes. Using a case-control design, incident cases and appropriately selected controls in Northern California are interviewed shortly after diagnosis to identify maternal and paternal exposures to selected chemicals during specific time intervals–prior to or at the time of conception, during pregnancy, and while breast feeding and chemical exposures of the child during the first three years of life. In addition to primary exposures, data are gathered on secondary exposures, such as dietary practices, environmental tobacco smoke or genetic susceptibilities of the parents and/or children, which may modify the effects of primary environmental exposures.

The case series and appropriately selected controls identified during the Superfund Research Program are being complemented with a second investigation, childhood Leukemia and Environmental Exposures This second investigation uses the protocol developed in the first study and increases the case series by including incident cases diagnosed in the Sacramento and Central Valleys of California. Also included are an exposure assessment, which evaluates the reliability of self-reported residential pesticide, and other chemical exposures, which occurred during the exposure periods of interest (identified above) and during the year preceding the interview. This exposure assessment includes a follow-up interview to observe chemicals present in the household. The active chemical ingredients in the pesticide products are identified using available EPA databases. The study analysis will include determination of the risk of disease given exposure to selected chemicals as well as the evaluation of the reliability of initial reporting. This intensive evaluation of reported chemical exposures will be made for cases and birth certificate controls who are the most residentially stable (families that have not changed residence in the 12 months preceding initial interview) and for cases who have translocations of interest. It is estimated that approximately 53% of the subjects included in the initial interview (Tier 1) are eligible for the subgroup of families participating in the household reliability study (Tier 2).

This proposal (Tier 3) will address the levels of specified chemicals currently present in the home environment and, by inference, at the time of conception, pregnancy, breast feeding and the child’s first three years of life. Using industrial hygiene techniques, samples of house dust, outdoor soil, and carpet residue will be analyzed to determine the congruence between reported chemical exposures (from Tier 1 and Tier 2) and measured exposures (Tier 3). Approximately 53% of the subjects who participate in Tier 2 will be included in the Tier 3 study.

Methods: Overview: This proposed study for Superfund XII is designed to parallel and extend the methods in place for the Molecular Epidemiology of Childhood Leukemia. Case accrual began at four Bay Area hospitals in 1996 and by 1999 there will be 157 eligible and consenting cases enrolled in the study. We propose to extend the case accrual in the Bay Area study for 3 years (2000-2002) in order to enroll an additional 130-140 cases eligible for Tiers 2 and 3. The protocol for the verification of chemical exposures (Tier 2) is implemented for all study subjects using funds available from and the protocol developed in childhood Leukemia and Environmental Exposures Implementation of Tier 3 measurement of chemical exposures is described below.

Tier 1: The four hospitals in the San Francisco Bay Area have been using the study protocol to identify newly diagnosed cases of childhood leukemia since mid-1996. This investigation proposes to extend case ascertainment in this geographic area for three years (2000-2002). These cases and those diagnosed in 1999 will provide subjects for inclusion in the Tier 2 and 3 exposure assessment described below. This is necessary as cases diagnosed in earlier years (1996-1998) will be increasingly likely to have moved or to be unwilling to continue participation in the study. This condition is particularly likely if the case child has died. The expansion of the case series will strengthen the ability of the molecular component of the study to link specific genetic characteristics with specific environmental exposures.

Tier 2: Any case or birth certificate control, irrespective of matching, that meets the criteria for residential stability will be included in Tier 2. Criteria for inclusion in Tier 2 phase of the study will be as follows: the case child was less than six years of age at the time of the Tier 1 interview, children between ages 3 and 5.9 years lived in the current residence at the time the child was 3 years of age, children between ages 1 and 3 have lived in the current residence at least 12 months, and children less than one lived at the current residence since birth. Cases that have translocations of interest, such as t(9;22), t(8;21), and t(11q23), are also included in Tier 2, even if cases are not residentially stable. Approximately 64% of the case subjects diagnosed before age 6 are eligible for the household pesticide survey as they are residentially stable; 35% of the cases diagnosed before age six had never moved in their lifetime).

This exposure assessment includes a survey of pesticides present in the household, interview data to quantify the frequency of use of each pesticide during pregnancy, the first three years of life of the child, and during the past year. Identification of the active ingredient in each product will be made though the EPA pesticide registration number. The exposure assessment interview data will be compared with interview data provided in Tier 1 to evaluate the reliability of the exposure reporting. This exposure assessment activity is included in a separate NIEHS/NCI funded grant ES/CA09137-01A1.

Tier 3: Further exposure assessment is proposed for cases and controls included in Tier 2. Specifically, cases and birth certificate controls, irrespective of matching criteria, that are included in Tier 2, are eligible for Tier 3, if pesticide products listed at the time of interview (Tier 1) are also found in the Tier 2 household survey. In addition, cases with translocations of interest are also eligible for Tier 3, regardless of residential stability in order to support a case/case analysis. Standard techniques will be used in the collection and chemical analysis of houseguest (collected by vacuum and hand dust swipes) and soil samples taken from areas in and around the home of the study subject. Samples of the dust and soil in the home and clothing worn by parents employed in agriculture will be analyzed for the presence of household and agricultural compounds. Household compounds will be selected for analysis from active ingredients of compound found in the homes of Tier 2 subjects. A list of active ingredients prevalent in household pesticide products found in a survey of CA homes will also be included, where different from the compound identified from Tier 2. Agricultural compounds will be selected based on high volume usage in agriculture as well as on toxicity and persistence. Two analyses are possible from Tier 3 data. A case/case analysis involves comparing cases with translocations with cases that do not by exposure status. A case/control analysis of residentially stable participants from Tier 2 will also be included.

The principal advantages of this study design:

The primary design advantage of Tier 3 is to evaluate the real-time exposure pathways by case/control status. Cases with translocations compared to cases without translocations will be stratified by detected and non-detected levels of household and agricultural pesticides, and where possible by active ingredients. Analysis of cases to birth certificate controls by measured levels of pesticides in the home may reveal active ingredients important to the etiology of childhood leukemia. Because the study area includes counties that are highly agricultural, a mixture of urban and agricultural, and highly urban, differences in amounts and types of residues may be determined.



The most important scientific goals which have been achieved during the first year of this study were as follows: 1) the development of a dietary component to assess micronutrients potentially associated with or modifiers of cancer risk in children, 2) the development of a field research protocol to evaluate the etiologic hypotheses regarding early immunologic stimulation and the interaction with chemical exposures, and 3) establishing a research agreement regarding human subjects between two academic centers, four clinical programs and the California State health agency.