___________________

¹ This workshop was organized by an independent planning committee whose role was limited to selecting the topics and speakers. This Proceedings of a Workshop was prepared by the rapporteurs as a factual summary of the presentations and discussions that took place at the workshop. Statements, recommendations, and opinions expressed are those of individual presenters and participants and are not endorsed or verified by the National Academies of Sciences, Engineering, and Medicine, and they should not be construed as reflecting any group consensus.

² Genomics is the study of an individual’s complete set of DNA. Scientists hope to understand how diseases, including cancer, form by studying genomics. For more information, please see https://www.genome.gov/About-Genomics/Introduction-to-Genomics (accessed June 20, 2023).

³ Epigenomics is the study of how certain factors chemically alter DNA or associated proteins without changing the nucleotide sequence. Epigenetic changes affect what genes are expressed, and by what degree. For more information, see https://www.genome.gov/about-genomics/fact-sheets/Epigenomics-Fact-Sheet (accessed June 2, 2023).

toxicogenomic effects of environmental hazards,⁴ genetic susceptibility to environmental exposures, and the role of genetic and epigenetic changes in the development of cancer. Public communication around the potential risks of environmental hazards has also been evolving. To explore the state of the science in identifying potential genomic and epigenomic biomarkers of environmental exposures associated with cancers, the National Academies of Sciences, Engineering, and Medicine’s National Cancer Policy Forum convened a virtual public workshop in collaboration with the Roundtable on Genomics and Precision Health on April 13, 2023, and sponsored by the Centers for Disease Control and Prevention (CDC).

This workshop was intended to provide background information to assist the CDC in updating the 2013 guidelines from CDC and the Council of State and Territorial Epidemiologist, Investigating Suspected Cancer Clusters and Responding to Community Concerns.⁵ Stephanie Foster, deputy associate director for science in the CDC’s Division on Environmental Health Science and Practice, provided a brief overview of the role of the Agency for Toxic Substances and Disease Registry (ATSDR) and the National Center for Environmental Health (NCEH)—in investigating cancer clusters. CDC issued its first guidelines for investigating clusters of health events in 1990, and several developments over the past decade led to the CDC/ATSDR effort to updating the guidance for responding to cancer clusters. In particular, Trevor’s Law, passed in 2016 as part of the Frank R. Lautenberg Chemical Safety for the 21st Century Act, provides the U.S. Department of Health and Human Services with the authority to conduct investigations of potential clusters, undertake actions to help address cancer clusters or factors that may contribute to the creation of clusters, and coordinate with other entities and the public (U.S. Congress, 2016). As part of this work, Foster clarified that state, tribal, local, and territorial (STLT) health officials have primary responsibility for evaluating unusual patterns of cancer and bring valuable current and localized data to bear on investigations of known or suspected environmental hazards. STLT partners can draw upon NCEH and ATSDR for technical assistance and partnership on study design, data analysis, and interpretation for epidemiological and environmental investigations of cancer cluster concerns.

While the science is advancing, Foster noted that cancer genomics information currently available did not allow for integration into the most recent

___________________

⁴ Toxicogenomics is the study of how environmental substances affect the genome and epigenome. See https://ntp.niehs.nih.gov/whatwestudy/testpgm/toxicogenomics (accessed June 2, 2023).

⁵ CDC’s 2013 Guidelines on Investigating Unusual Patterns of Cancer and Environmental Concerns are available at https://www.cdc.gov/mmwr/pdf/rr/rr6208.pdf (accessed July 20, 2023).

Guidelines for Examining Unusual Patterns of Cancer and Environmental Concerns (Foster et al., 2022), which was published in 2022 after extensive study and input from cancer registries, STLT partners, and the public.⁶ To determine how scientific advancements in cancer genomics might enhance the CDC’s future efforts in this area, Foster identified three priorities for the workshop: to examine the state of the science of cancer genomics and environmental exposures; to explore how this research might be used to identify and prevent unusual clusters of cancer occurrence, particularly in pediatric populations; and to identify evidence gaps and possible future research directions to further advance knowledge in this area.

Roberta Ness, former dean of the University of Texas School of Public Health (retired), provided an overview of the workshop’s purpose and organization. The workshop was designed with an overarching goal of providing CDC and ATSDR with a better understanding of the connections between genomic and epigenomic markers of environmental exposures and pediatric cancers. In addition, Ness noted that the workshop discussions could enhance CDC’s ability to provide local public health agencies with information, tools, and approaches to assess and address potential cancer clusters.

The workshop presentations and discussions addressed different aspects of cancer genomics information as it relates to cancer cluster investigations. Workshop speakers focused on novel approaches to improve the identification of genomic and epigenomic biomarkers of environmental exposures, statistical and epidemiological approaches to identify linkages of genomic biomarkers to cancer clusters, and an examination of the role of environmental exposures in pediatric cancer. Presentations and discussions also focused on potential strategies for scientists to communicate risk more effectively to communities. Several speakers also provided lessons learned and future considerations to advance the field. This Proceedings of a Workshop summarizes the presentations and discussions from the workshop and highlights observations and suggestions from participants, which are discussed throughout the Proceedings and summarized in Boxes 1 and 2, respectively. Appendix A includes the statement of task for the workshop, and the agenda is provided in Appendix B. Video recordings and speaker presentations are available online.⁷

___________________

⁶ CDC’s 2022 Guidelines on Investigating Unusual Patterns of Cancer and Environmental Concerns are available at https://www.cdc.gov/nceh/cancer-environment/pdfs/Guidelines-for-Examining-Unusual-Patterns-of-Cancer-and-Environmental-Concerns-h.pdf (accessed July 20, 2023).

⁷ See https://www.nationalacademies.org/event/04-13-2023/the-potential-contribution-of-cancer-genomics-information-to-community-investigations-of-unusual-patterns-of-cancer-a-workshop (accessed May 26, 2023).

approach to assess exposures including biological sampling, home sampling, geocoding, and parental interviews (Figure 2).

The overall increase in childhood cancers points to the role of environmental factors alone, or in combination with genetic susceptibility, and it is clear that “where you live matters,” Metayer emphasized. Residential proximity to pesticide use has repeatedly been shown to increase the risk of childhood leukemia (Bamouni et al., 2022; Malagoli et al., 2016; Park et al., 2020; Patel et al., 2020). Spatial risk analyses and home sampling studies found that Latino communities who work on farms with high pesticide use in the Central Valley in California have a high burden of multiple harmful environmental pollutants (National Institute of Environmental Health Sciences, n.d.). Texas communities with intensive releases of hazardous air pollutants have higher incidences of pediatric liver cancer (Thompson et al., 2008). Living near heavy traffic areas, gas stations, or fracking wells has also been linked to multiple childhood cancers (Boothe et al., 2014; Deziel, 2021; Filippini et al., 2019; Kumar et al., 2018; Mazzei et al., 2022; Ritz et al., 2022).

Emerging tools and strategies are helping scientists to identify environmental factors that increase the risk of childhood cancers, as well as behaviors that can lower this risk (Metayer et al., 2016; Whitehead et al., 2016). For example, molecular epidemiology has linked self-reported maternal smoking during pregnancy with validated epigenetic biomarkers of tobacco exposure in children (de Smith et al., 2017; Gonseth et al., 2016; Kaur et al., 2016; Metayer et al., 2013; Yano et al., 2019). In addition, the exposome model generates new hypotheses

by identifying the totality of exogenous exposures through multiple targeted or untargeted searching methods (Rappaport, 2011). The Developmental Origins of Health and Diseases theory also offers a possible explanation of how in utero and preconception exposures may affect childhood cancer outcomes (Heindel and Vandenberg, 2015).

Despite these advancements, many suspected pediatric cancer clusters are not completely understood. Metayer said that more research is needed to better understand specific cancer subtypes and causal agents. In addition, she emphasized the need for researchers to work together to proactively monitor spatial and

temporal clusters at fine resolutions and to incorporate other advanced molecular tools for assessing germline genetics, tumor genetics, and epigenetics to address unusual patterns of childhood cancers.

___________________

⁸ See https://www.genome.gov/genetics-glossary/Methylation (accessed June 1, 2023).

⁹ See https://www.cancer.gov/publications/dictionaries/cancer-terms/def/methylation (accessed June 1, 2023).

predict future cancer risk, in lieu of traditional exposure assessment methods and metrics. Recent studies have found a correlation between high methylation scores and the incidence of lung and aerodigestive cancers (Mani et al., 2012). Ladd-Acosta said this research also uncovered a group of outliers—people with extremely high smoking-related methylation scores but low or no self-reported smoking. Upon closer examination, these individuals were found to reside in an area with multiple tobacco farms and production facilities, raising important questions about the potential effects of environmental tobacco exposure (such as through clearing land by burning or employment in tobacco production facilities) among nonsmokers.

DNA methylation biomarker tools can be used to detect shared exposures within cancer clusters by collecting methylation patterns from individuals with and without cancer, comparing them to patterns from different exposures, and creating cancer-associated exposure profiles. Ladd-Acosta suggested that combining such profiles with genetic variant patterns and custom exposome arrays could improve cluster detection, adding that DNA methylation biomarker tools can also be used to perform cost-efficient population-wide cancer risk monitoring to identify potential high-risk groups who would benefit from screenings or interventions to reduce risk.

Genomic Methods to Understand the Genetic Origins of Pediatric Cancers

Genomic profiles of children with pediatric cancer are currently being studied to identify prognostic markers of drug response and adverse effects. Jinghui Zhang, chair of computational biology, St. Jude Children’s Research Hospital, discussed how this genetic information could also help to pinpoint the genetic origins and molecular mechanisms of pediatric cancer development by studying tumor and germline genomics.

Whole genome sequencing studies of pediatric cancer tumors have identified 142 significant driver genes, most of which are absent in adult cancers (Ma et al., 2018). In addition, the age of onset is associated with different driver genes and mutations, as well as different clinical outcomes (Brady et al., 2020, 2022). Researchers are studying how patterns of methylation data might be used to identify cluster-specific driver genes, as well as how germline genomes of children with cancer can be used to identify pathogenic mutations (Zhang et al., 2015). As an example, the R337H mutation in the TP53 oncogene is associated with adrenocortical carcinoma tumors and may be useful in predicting disease onset.

Zhang highlighted several resources that can help advance research in this area. One is the Pediatric Cancer Variant Pathogenicity Information Exchange, a platform that classifies the pathogenicity of mutations in children with cancer

for clinical or research purposes (Edmonson et al., 2019). Another is the St. Jude Cloud Platform, a collection of data and analysis tools that includes data from more than 11,000 children with cancer and about 2,500 clinical samples, which researchers can use to examine new approaches to detecting cancer clusters (McLeod et al., 2021).

Using Cancer Mutational Signatures to Explore the Etiology of Pediatric Cancer

Somatic mutations constantly occur in human cells because of normal cell operations as well as malfunctions that may be related to lifestyle and environmental factors. These mutations can strongly affect cancer risk. Ludmil Alexandrov, associate professor, University of California, San Diego, described how his group applies machine learning (ML) approaches to explore cancer etiology by investigating the patterns, or “signatures,” of these mutations.

ML can be used to detect unusual mutational patterns and associate them with mutagenic processes such as cell function, lifestyle factors, or environmental exposure. For example, adult skin cancers associated with ultraviolet (UV) light and adult lung cancers associated with smoking each have specific mutational signatures (Alexandrov, 2015; Alexandrov et al., 2013). Alexandrov and colleagues created a compendium of more than 100 cancer mutational signatures associated with normal and abnormal cellular processes, treatments, environmental exposures, and lifestyle factors (Alexandrov et al., 2020). This resource enabled researchers to uncover the etiology of several germline mutation associated with both specific mutational phenotypes and an increased risk for developing cancer (Alexandrov et al., 2015; Drost et al., 2017; Grolleman et al., 2019). While there are many environment-linked mutational signatures associated with adult cancers, most mutational signatures in pediatric cancers are attributed to endogenous mutagenic processes rather than environmental exposures. However, researchers have identified a signature for UV light exposure that is linked to a specific subtype of acute lymphoblastic leukemia (ALL) in children of non-African origin (Coste et al., 2017; Ma et al., 2018).¹⁰

Mapping the mutational processes associated with different cancer types could help researchers use mutational signatures to identify cancer clusters; however, Alexandrov said that more work is needed to elucidate mutational signatures of environmental exposures, especially for pediatric cancers. So far, analyses of the mutational signatures in 785 cancers representing 27 pediatric cancer types have identified several types of exposures, but they cannot yet be linked

___________________

¹⁰ Presenter provided additional clarification after the workshop, in the course of writing this proceedings.

to specific cancers (apart from those associated with UV light exposure), and a number of signatures still have unknown etiologies (Thatikonda et al., 2023).

Advancing the Identification of Pediatric Mutational Signatures

Speakers identified areas of synergy among emerging research strategies, considered data gaps and future research directions, and discussed how this work could inform cluster investigations and cancer prevention efforts.

Speakers discussed similarities and differences among various novel approaches being explored to improve identification of biomarkers of environmental exposures relevant to cancer. Methylation patterns are distinct, spatially and temporally, from mutational signatures, Ladd-Acosta clarified, but both may contribute to cancer etiology, and may even overlap. Zhang noted that methylation could be a very early-onset event in tumor initiation, noting that longitudinal studies of people with Wilms tumors (a rare pediatric kidney cancer) have detected methylation signatures in the germline sample that could serve as a biomarker to evaluate the effect of therapy on the long-term outcome of survivors, or even for early detection (Guerra et al., 2019).

Ladd-Acosta suggested that once researchers have a better understanding of which exposures generate specific methylation patterns, they can further investigate how those relate to specific health contexts and outcomes across broad populations. She added that there are still open questions regarding which other chemical toxicants or psychosocial stressors might have associated methylation patterns, beyond tobacco. In addition, from a methodology perspective she said it would be useful to determine what data could be gleaned from dried-blood spot cards that are used for newborn screening and whether tissue surrogates could be used to identify methylation biomarkers in hard-to-access tumors.

Alexandrov noted that while mutational signatures are associated with very specific chemical compounds, these compounds can be found throughout the environment. As a result, epidemiological studies are needed to pinpoint the exact exposure source. Pediatric cancers have fewer mutations than adult cancers, which may be attributable to adults’ longer exposure to mutagens. Zhang added that discovering more pediatric cancer mutational signatures will require global collaboration to account for the relative rarity of childhood cancers.

Certain signatures are already being used to identify known mutagen exposures for cancer clusters in adult populations, Alexandrov said, but little evidence has been found in pediatric tumors of environmentally driven mutagenesis. He added that most studies in this area to date have comprised analyses of a broad mix of cancer types, rather than cluster investigations, and he said he is not aware of any epidemiological studies based on mutational analysis. He suggested that future investigations of statistically unlikely occurrences of cancers seek evidence of a common causative agent regardless of the population involved. He added that

the cancer occurrences could also be due to underlying socioeconomic structures. Alexandrov also noted that there is still much more to learn about the pediatric leukemia signature associated with UV light exposure, such as how the effects of light are transmitted to the bone marrow. Another challenge is understanding the timing of the exposure and the initiation of cancer. Zhang suggested that the initiation events for childhood cancers could be prenatal. He stated that knowledge about mechanisms contributing to mutation rates and signatures remain incomplete. For example, multiple signatures may be present in patients affected by the same specific germline mutation, and epidemiological studies to identify shared exposures and time lines could be illuminating. Alexandrov said that the timing of mutational signatures can be determined, and many of the mutations believed to drive cancer have been found to occur in the first decade of life, suggesting that childhood exposure increases the risk of developing cancer in adulthood.

Pediatric cancers are rare, but cumulative exposure to chemicals is known to negatively affect children’s health more generally as well. In addition to identifying environmental factors that increase risk, Metayer said it is important to use that knowledge to try to prevent harmful exposures. She noted that her group is translating their work to promote healthy behaviors, reduce hazardous exposures, and create policies to protect children’s health. “Cancer is not the only adverse outcome that is related to those exposures,” Metayer said.

STATISTICAL AND EPIDEMIOLOGICAL APPROACHES TO IDENTIFY GENOMIC BIOMARKERS IN CANCER CLUSTERS

Participants discussed opportunities to use statistical and epidemiological approaches for connecting genomic biomarkers with cancer clusters. Session moderators Wei-Ting Hwang, professor of biostatistics, University of Pennsylvania Perelman School of Medicine, and Sharon Plon, professor and assistant dean of the School of Medicine at Baylor College of Medicine, and chief of the Cancer Genetics Clinic at Texas Children’s Hospital, and several speakers highlighted the role of cancer registries in advancing research through structured data collection and epidemiological methods. They also discussed the role of biological, environmental, and social determinants of health in the occurrence of cancer clusters.

The Role of Central Cancer Registries in the Investigation of Unusual Patterns of Cancer

To help reduce the burden of cancer, cancer registries track local and national cancer rates and trends; provide data for cancer control, prevention, and research; respond to community concerns about potential clusters; and

propose hypotheses of cancer risk and etiology. Recinda Sherman, program manager, North American Association of Central Cancer Registries (NAACCR), described how NAACCR supports these registries by developing data collection standards, evaluating and certifying data, and supporting rigorous and relevant data analysis.

Cancer cluster investigations seek to identify shared causal exposures while simultaneously ruling out others, a challenging task that can often take years. Investigations can reveal clear links between cancer and the environment, but they can also harm communities emotionally or financially if a known health hazard is present but cannot be definitively linked to the cancers (Gundersen, 2000).

Sherman noted that the 2022 CDC Guidelines are a critical resource for registries and health departments investigating clusters. She said that the most recent version recommends proactively addressing and monitoring potential clusters, a resource-intensive activity that cancer registries may not have the budget to support. In addition, she cautioned that the multiple methods suggested in the Guidelines can sometimes lead to conflicting results.

To improve cancer cluster investigations, one NAACCR workgroup is creating a new standard to integrate genomic and biomarker data, which can reveal genetic risks and environmental exposures but are usually stored in isolated, inaccessible documents, into clinical records and cancer registries (Confluence, 2023). One potential solution is through the Virtual Pooled Registry Cancer Linkage System (VPR-CLS), which standardizes and links research cohort data with cancer registry data to support cancer cluster studies. With more funding, she suggested that VPR-CLS could be scaled up to include genomic data.¹¹

Epidemiological Approaches for Assessing Reports of Cancer Clusters and Environmental Exposures

The 2013 CDC Guidelines defined a cancer cluster as a greater-than-expected number of cancer cases that occurs within a group of people in a geographic area over a period of time.¹² Cancer clusters may arise for various reasons, including environmental exposures. Melissa Bondy, chair and professor, associate director of population sciences, and codirector of the Center for Population Health Sciences, Stanford University, discussed epidemiological approaches used to assess potential cancer clusters.

Health departments and cancer registries assess reported clusters by consulting with the community; gathering and analyzing environmental samples; examining demographic data within the affected community including age, ethnicity,

___________________

¹¹ See https://www.naaccr.org/about-vpr-cls/ (accessed June 27, 2023).

¹² See https://www.cdc.gov/nceh/cancer-environment/about.html (accessed May 31, 2023).

education, occupation, residential history, and access to medical care; and defining the geographic area and the time frame. In doing this work, Bondy said, it is important to take steps to overcome sources of error such as erroneous diagnosis, attribution, and census estimates; to appropriately assess demographic factors; to account for mixtures of different cancers; and, importantly, to account for the possibility that a cluster occurs as a result of chance rather than a shared cause.

Investigations of potential cancer clusters seek evidence of carcinogen exposure at a dose that measurably and unambiguously increases cancer risk, clinical records confirmed through laboratory studies, relevant animal studies, identification of the exposure mechanism (e.g. airborne, waterborne, or viral), and a study of epidemiological patterns to rule out chance occurrence or bias and account for multiple exposures. Studying clusters is complicated by the fact that there are many risk factors that can lead to cancer, as well as confounding demographic factors.

Bondy also emphasized the importance of addressing public perceptions of a cluster in the community. She said that this requires deliberate steps to address community members’ fear, gain their trust, keep them informed, and fully explain the findings. She also suggested that public health agencies should partner with minority-serving organizations to address environmental injustices in areas with high exposures and added that community outreach and monitoring should be initiated promptly after environmental disasters occur.

Genomic Ancestry Inference to Decompose Genetic and Environmental Contributions to Cancer Disparities

Black, Indigenous, and other communities of color face higher rates of new cancer diagnoses and cancer deaths, posing a serious scientific and social challenge, stated King Jordan, professor and director of the Bioinformatics Graduate Program at the Georgia Institute of Technology. He said:

Ancestry affects a person’s health in two ways: geographically defined ancestry accounts for health-related genetic variants, and socially defined ancestry accounts for differences in a person’s social environment and lived experiences, which in turn affect health outcomes. Jordan described how a process called genomic ancestry inference can help to disentangle the complex web of genetic, social, and environmental risk factors that contribute to cancer disparities.

Genomically inferred ancestry is the study of an individual’s genome to determine their cancer risk, beyond their geographic or social ancestry. It offers

a precise and objective measurement and can be expressed in different ways: as a categorical or a continuous variable, at the continental or local level, and genome-wide or for specific chromosomal locations. A recent analysis examined ancestry, survivorship, and molecular signatures (such as mutations, gene expression, and methylation) across different types of cancer and found four cancer types with survival disparities and seven cancer-related genes for which epigenetic changes, more than DNA mutations, contribute to survival disparities, suggesting that population-specific targeted therapies might reduce cancer disparities (Lee et al., 2022).

To improve research on cancer health disparities, Jordan said that registries need more data from diverse and underrepresented populations, researchers need better methods for applying insights across populations and ancestries, and the research teams themselves need more diverse representation.

Advancing the Use of Statistical and Epidemiological Tools for Cancer Cluster Assessment

Speakers discussed current opportunities and limitations in statistical and epidemiological tools for assessing potential cancer clusters. Sherman stated that NAACCR is working on a national childhood cancer registry linking exposure and genomic data to improve cancer cluster investigations, and she posited that a database with even more variables and improved latency could use cancer data more effectively for cluster detection. She cautioned that these investigations should be seen as a piece of a larger picture that enables cancer control or mitigation efforts to occur alongside investigations to improve the health of every community.

Bondy suggested that epidemiological studies of residential data, cross-examined with electronic health records, race, or exposure data and incorporating ML and artificial intelligence (AI) techniques, could be used to identify clusters more quickly. Jordan agreed, noting that explainable AI, an innovation that pinpoints the specific features that contribute to an outcome, could be an especially useful tool to identify cluster causes. In addition to cancer cluster detection, AI and ML techniques will also benefit ancestry research, Jordan noted, especially as more people use home ancestry test kits and the global reference population improves. Matched samples from normal tissues and tumors (which enable comparisons and generate data on germline variants, somatic mutations, gene expression, and epigenetic processes like methylation) remain the “gold standard” of samples for cancer research, however.

Cancer cluster investigations would also benefit from approaches that incorporate social determinants of health, geocoded data, and multiomics molecular data to understand how the legacy of discrimination contributes to cancer disparities, Jordan said. Bondy added that social determinants of health are both

confounders and effect modifiers because they can affect health outcomes differently depending on the exposure.

Sherman suggested that the NAACCR’s VPR-CLS could be adapted to address some of these issues, especially if location data are included. In fact, the NAACCR is currently identifying a minimum set of variables needed to respond as quickly as possible to a potential cluster. She cautioned that while registries do hold the most recent data, reporting and pattern detection can be delayed. In addition, while it is possible to model future trends, she said that an important gap is the lack of methods to proactively predict localized clusters, an approach which would be more equitable than merely responding to concerns, a strategy that favors better-resourced communities.

Several speakers posited that cancer registries are an ideal place to start collecting genomic information, because, as Bondy explained, epidemiologists already refer community members expressing concerns to local registries and public health organizations, who perform the initial evaluations. Foster reiterated that it can take years, or even decades, to confirm a cluster, and she said that the CDC believes that expanding registry data may be useful to enable research into genomic markers of cancer and environmental exposures in clusters.

ENVIRONMENTAL EXPOSURES AND PEDIATRIC CANCER

Toxic exposures, genetics, and infections all play a role in the development of pediatric cancers, said Kim Nichols, director of the Division of Cancer Predisposition at St. Jude Children’s Research Hospital. However, the rarity of these cancers—combined with the challenges of obtaining biological specimens that could demonstrate correlations between cancer and exposures, especially during critical developmental periods that are most likely to influence cancer development—complicates efforts to ascertain the exact relationships among these factors. Speakers focused on evidence and strategies for understanding how environmental exposures influence pediatric cancer.

Current Epidemiological Evidence on Environmental Exposures and Pediatric Cancer

Michael Scheurer, professor at Baylor College of Medicine, provided an overview of the current evidence on the role of environmental exposures in pediatric cancers and discussed challenges and opportunities in advancing knowledge on this relationship.

Pediatric cancer incidence has increased by 45 percent since 1975 (SEER, 2023), but only 10 percent of cases are caused by a known genetic variant. This has led to growing concern that environmental exposures are to blame, though the evidence remains limited. Among known human carcinogens, ionizing radia-

tion and exposure to environmental metals are the most well-established risk factors for pediatric cancers. Studies have also linked childhood leukemias to a variety of chemicals including tobacco smoke, polychlorinated biphenyls, benzene, and others. However, most chemical compounds, such as those found in herbicides and pesticides, have not been fully evaluated for risks in children, even those with established links to adult cancers (IARC, 2023; Shakeel et al., 2022).

As other speakers noted, studying the relationship between environmental exposures and pediatric cancer is challenging because of the rarity of these cancers and the difficulty of estimating exposures during the critical developmental periods of preconception, in utero, and early childhood, Scheurer said. Investigations often rely on self-reported questionnaires or residential information that does not accurately measure environmental compounds or the exposure window. To overcome these challenges, Scheurer suggested that researchers should focus on vulnerable populations, who often live in high-exposure areas; initiate large, multi-institutional, and multinational studies to increase sample size; use advanced analytical tools to uncover biomarkers of exposure and timing; evaluate the interactions between exposure and inherited and somatic genomes, especially during key developmental periods; and evaluate the exposures not just for cancer development but also for treatment outcomes and survival rates.

Using Molecular Epidemiology to Investigate Environmental Exposures and Pediatric Cancer

“What we’re trying to do with molecular epidemiology,” said Joseph Wiemels, professor and associate director of shared resources at the University of Southern California, “is open up the black box between exposure and disease.” Noting that environmental exposures are the most likely cause of rising leukemia rates (Barrington-Trimis et al., 2015; Gurney et al., 1996), Wiemels said that the field of molecular epidemiology is moving closer to determining cancer etiology by linking environmental exposures, such as from infections, parental tobacco use, ionizing radiation, pesticides, and more, to molecular processes correlated with cancer, such as gene deletions or mutations and DNA methylation patterns.

Molecular epidemiology studies conducted in utero, when initiating lesions for childhood leukemia are present, and in early childhood, when secondary genetic changes occur that can cause the onset of leukemia, have identified neonatal cytomegalovirus infections as a driver of acute lymphoblastic and hyperdiploidy leukemias (Francis et al., 2017; Gallant et al., 2023). In addition, in utero and early infancy tobacco exposure has been linked with epigenetic biomarkers associated with gene deletions commonly found in ALL tumors, suggesting a correlation between the exposure and ALL (de Smith et al., 2017; Finette et al., 1998; Xu et al., 2021). A meta-analysis of maternal smoking studies also showed strong associations between tobacco use and DNA methylation in newborn

cord blood (Joubert et al., 2016). Researchers are now developing a smoking biomarker assay based on DNA methylation patterns and exploring linkages with genetic polymorphisms, tobacco use, and leukemia risk (Arroyo et al., 2022).

The Exposome and Pediatric Cancer

The exposome is the measure of a person’s total exposures, beginning before birth, and how the exposures may influence health.¹³ Lauren Petrick, associate professor and head of untargeted metabolomics at Mount Sinai and the director of the Center for Metabolomics and Molecular Phenotyping, Sheba Medical Center in Israel, defined exposomics as the direct measurement of the physical, chemical, and biological components of a human’s external environment combined with the internal molecular, physiological, and health-related effects of these exposures (Gao and Snyder, 2021; Petrick and Shomron, 2022; Wishart, 2022). Researchers have used new exposomic tools and techniques, such as biomonitoring with high-resolution mass spectrometry and liquid chromatography, to demonstrate links between various environmental risk factors and adult cancers (Goodrich et al., 2022). These tools have also illuminated the breadth of chemicals and metabolites in the prenatal exposome; however, they have not yet been used to identify many specific factors involved in pediatric cancers (Panagopoulos Abrahamsson et al., 2021; Parada et al., 2019; Sarink et al., 2021; Shearer et al., 2021; Wolff et al., 1993; Yu et al., 2021).

Exposome studies also have challenges and limitations. They are nontargeted analyses, which are advantageous for small sample sizes, but they only uncover information on the current body burden, leaving the timing and magnitude of the exposure unknown. In addition, retrospective biomonitoring is difficult in rare diseases like pediatric cancer, and drawing blood can pose challenges in young children.

Petrick highlighted several potential opportunities to better use these tools for studying pediatric cancers. One option is to use dried blood spot cards, collected at birth for newborn screening, to provide a predisease biological snapshot. These cards have been used to identify metabolites linked with risk factors for development of pediatric leukemia, to support epidemiological evidence of diet as a risk factor of pediatric leukemia, and to discover early life biological interactions that may drive later leukemia development (Metayer et al., 2023; Petrick et al., 2019, 2021; Whitehead et al., 2021).

In addition, capillary blood—which can be collected at home with a simple heel or finger prick—and primary teeth are potential resources for sampling prenatal exposure (Lupo et al., 2021; Petrick et al., 2020; Yu et al., 2021). External,

___________________

¹³ See https://www.cdc.gov/niosh/topics/exposome/default.html (accessed June 1, 2023).

noninvasive devices such as wristbands or mobile apps can also be used to collect exposome data. Finally, exposomics can be integrated with other -omics methods to detect associations with leukemia development, and archived samples can be remined for chemicals of emerging concern in pediatric cancer etiology (Gao et al., 2022; Maitre et al., 2022; Shearer et al., 2021).

Increasing Evidence-Based Linkages Between Environmental Exposures and Pediatric Cancers

Mark Purdue, senior investigator, Division of Cancer Epidemiology and Genetics at the National Cancer Institute, moderated a discussion on opportunities to advance research on environmental exposures and pediatric cancers. Speakers reiterated that achieving adequate sample sizes will require effective data-sharing collaborations that use the most advanced technologies. Petrick, Scheurer, and Wiemels said that new biosampling methodologies, such as dried blood cards, can be instrumental in generating the data needed to uncover relationships between exposures and cancer, especially during critical developmental windows. Wiemels noted that different states have different regulations regarding the allowable research uses of dried blood cards and emphasized that, to protect privacy, it is important to deidentify samples when using them for research. Scheurer suggested that an opt-out approach (rather than requiring parents to opt in to their child’s data being accessible for research) can help to eliminate sample bias and improve sample quantity.

Petrick noted that a recent advancement in exposomics is the ability to collect measurements for multiple risk factors in a single assay, at low concentrations and small volumes, and to identify factors in combination. Wiemels agreed that the ability to examine multiple exposures simultaneously is a substantial improvement, as is the new landscape of multiomics platforms. He added that new AI techniques will be needed to analyze the large amounts of data generated with this approach.

Genomic assessment tools now provide a more holistic view of childhood cancer risk than studies of individual genes, Wiemels continued. For example, Wiemels said gene-only association assessments on more than 15,000 children with ALL have found 23 different single-nucleotide polymorphisms associated with a very high cancer risk. While the science still has a long way to go, he suggested that similar large-scale studies of gene–environment interactions, in adults or children, could identify individuals with high susceptibility to environmental toxins. That knowledge may not prevent cancer, but it could promote behavior change in certain populations.

Participants also discussed several routes of exposure that could warrant additional study. Scheurer noted that schools are a potential source of exposure for children. He said he was not aware of any studies investigating the issue but

sees it as a ripe opportunity for research and policy change Another potential area of focus is air pollution, which is known to increase the risk of respiratory infections in childhood cancer survivors, although it is unclear if the cause is chronic exposure to pollution or if the pollution is a surrogate for identifying environments in which infectious exposures are likely (Ou et al., 2019). Wiemels suggested that exposure to ethylene oxide, which has been associated with adult cancers and is usually related to occupational exposures, could also be studied in children.

POTENTIAL STRATEGIES FOR COMMUNICATION ABOUT ENVIRONMENTAL EXPOSURES AND CANCER RISK

Communication is critical to effectively engaging communities during the investigation of potential cancer clusters. Speakers discussed communication strategies to inform communities of findings from environmental research and to explain potential exposure risk.

Enhancing Communication and Engagement with Communities About Cancer Clusters

Trevor Schaefer established Trevor’s Trek Foundation with his mother after being diagnosed with brain cancer at age 13. The foundation raises awareness of childhood cancers, facilitates research into its causes, and works to protect communities from environmental contamination. In addition, its Childhood Cancer Cluster Buster Program collects and maps national childhood cancer data, identifies affected communities, and works to remedy environmental issues. The foundation also was instrumental in pushing for Trevor’s Law, a law named in Schaefer’s honor, which mandates federal assistance to communities experiencing environmental contamination and provides a basis for CDC’s Guidelines for investigations of potential clusters.

To improve cancer cluster investigations, Schaefer suggested that relevant nongovernmental organizations (NGOs) should form an online network to track local cancer cases and investigations, which would be publicly available to assure communities that their concerns are being taken seriously. In addition, he suggested shortening the time states have to report cancer cases to 2–3 months to enable quicker identification of potential clusters.

Communicating What is Known in Community Cancer Investigations

Julia Brody, executive director and senior scientist at Silent Spring Institute and research associate in epidemiology at Brown University, urged public health

researchers who are engaged in cancer cluster investigations to focus their communication on what is known about the causes of cancer, while also acknowledging the uncertainty and limitations. Bondy said that these investigations make important contributions to public health, including when the causes of community cancers are not identified. In one example, research that was prompted by high rates of breast cancer on Cape Cod, Massachusetts, led to improved understanding of chemical exposures from consumer products—yielding discoveries that ultimately resulted in product reformulations and new chemical regulations (Brody et al., 2004, 2006; McKelvey et al., 2004; Rudel et al., 2003; Zota et al., 2010). Brody said that researchers have a responsibility to report their findings to the public, especially those who participate in studies, for example by donating tissue samples or allowing researchers to measure chemicals in their homes. People want to know the results, even if they are inconclusive. Effective communication is vital to increasing transparency and trust, advancing equity, and reducing environmental injustice in communities that are underresourced or experiencing structural racism, Brody said, adding that effective communication efforts can also improve study participation rates, further advancing public health.

Silent Spring’s Digital Exposure Report-Back Interface (DERBI) was created to provide researchers with the resources and training needed to effectively report their findings back to the public (Boronow et al., 2017). DERBI generates adaptable and personalized exposure reports, details about the chemicals being studied, exposure reduction strategies, and study-wide results in a format that is accessible by computer, smartphone, or printer (Figure 3).

When reporting back to the public, Brody emphasized that effective communication can advance environmental health literacy, build trust, and facilitate conversations people have with clinicians. In addition to offering individual results to study participants, she said that study-wide results should be publicly shared via news, social media, and community meetings. Information can also be tailored for use in clinical appointments. Communities can benefit from resources to enable connections with other affected areas, similar to the approach used by the PFAS Exchange.¹⁴

Finally, Brody cautioned that researchers should avoid making false reassurances. She stated that cancer is a difficult disease to study because of its long latency and multiple risk factors; it is more accurate and more helpful to focus on what is known, rather than implying that a lack of certainty or proof means that there is no harm. Brody said:

___________________

¹⁴ PFAS are per- and polyfluoroalkyl substances and are known to be toxic. Please see https://pfas-exchange.org for more information (accessed June 20, 2023).

A range of scientific evidence, including results from human epidemiological studies and experimental studies in animals and cells, can help community members interpret their results and evaluate risks.

Promoting Culturally Sensitive Risk Communication with Communities

Exposure to pollution can contribute to premature death and disease. As a result, said Monica Ramírez-Andreotta, associate professor at the University of Arizona, “your zip code can be more important than your genetic code.” To address the root causes of environmental and health disparities, Ramírez-Andreotta urged scientists working in communities experiencing environmental injustice to use participatory research methods to engage residents, listen to their concerns, and codesign relevant reporting materials. “Environmental racism is occurring,” she said, “and we should be striving for justice.” She emphasized the importance of identifying the social, cultural, and political contexts perpetuating the injustice—rather than attempting to address health at the individual level—to achieve structural change and improve environmental quality and overall health (Davis and Ramírez-Andreotta, 2021; Masuda et al., 2010).

Participatory research can transform investigations through listening, cultural humility, building partnerships, and generating solutions through dialogue.

The research is designed and conducted in cooperation with the affected population (Cornwall and Jewkes, 1995). In this work, Ramírez-Andreotta said it is important to consider community members’ intersectionalities—the multiple dimensions along which an individual may experience structural barriers, such as minority status, educational attainment, languages spoken, and income level—when collecting samples; conducting surveys, interviews, or focus groups; analyzing data; and codesigning reports (Crenshaw, 1989; Davis et al., 2020).

In one project, researchers and residents of a Superfund site collaborated to ascertain the level of contamination in food grown in their gardens. Community members were instructed on how to collect samples and calculate their personal cancer risk scores to determine how much garden-grown food they could safely eat (Ramírez-Andreotta et al., 2013a,b, 2015; Sandhaus et al., 2019). In another study called Project Harvest, researchers and community members codesigned data-sharing materials to understand contaminants in rainwater, how people used rainwater, what regulations were in place, and what strategies could prevent or reduce exposures (Gray, 2018; Ramírez-Andreotta et al., 2023).

There are multiple theoretical frameworks to evaluate project outcomes (Cairo, 2012; Emmett et al., 2009; Finn and O’Fallon, 2019; Odden and Russ, 2019; Ryan and Deci, 2000; Warren et al., 2001). Both the garden project and Project Harvest emphasized public engagement, information codesign, and face-to-face experiences. In addition to greater awareness and self-efficacy, participants gained scientific literacy and data interpretation skills, and researchers gained a better understanding of what motivates people to participate (Sandhaus et al., 2020). Data from these studies become a boundary object that can stimulate dialogue among stakeholders and create structural change (Coombe, 2017; Davis and Ramírez-Andreotta, 2021; Star and Griesemer, 1989).¹⁵

Potential Opportunities to Enhance Community Collaboration in Research

Speakers discussed what motivates communities and researchers to engage with one another and examined opportunities to build more effective relationships in order to uncover environmental contributors to cancer. Brody and Ramírez-Andreotta urged researchers to not overlook the motivations and contributions of study participants and community members. The public is often willing to work hard to participate in investigations of potential environmental exposures or cancer clusters, and many people see these studies as a meaningful

___________________

¹⁵ Star and Griesemer recognized that successful scientific research and advancement requires cooperation between people with different viewpoints. The boundary object facilitates communication between the team members. See https://scalar.usc.edu/works/boundary-objects-guide/boundary-objects for definition (accessed June 12, 2023).

opportunity to contribute to a better future, Bondy said. In return, researchers have a responsibility to communities. Describing community reporting as “the heart of science communication,” Ramírez-Andreotta said that people deserve to know whether they are living near a contaminated site, to understand the risks, and to be involved in the cleanup decisions.

“Communication isn’t just about us doing a better job of communicating facts about science,” said Alejandro Sweet-Cordero, chief of the Division of Pediatric Oncology and the director of Molecular Oncology Initiatives at the University of California, San Francisco. “It’s about giving members of the community agency to understand the environment that they’re living in and providing them with tools to impact that environment.” To facilitate effective communication, Ramírez-Andreotta said that researchers should explain statistical uncertainties or outliers, communicate risks and limitations in the data clearly, encourage questions, recognize the importance of community-collected data, and work to ensure such data sources are standardized and interoperable with federal and state data (Ramírez-Andreotta et al., 2021). Schaefer agreed, noting that communities often need help finding resources, connecting with experts, and sharing data. In addition to partnering with NGOs, Ramírez-Andreotta said that it can be helpful for researchers to hire knowledge liaisons, or promotoras, who are trusted community members who can liaise between researchers and the community, refer participants to relevant resources, and train others in data collection methods.

Noting that the new CDC Guidelines include resources for community participation, Brody said that, while DERBI is not publicly available, teams partnering with CDC can use it through collaborations with Silent Spring Institute, and in the future, DERBI could be made available to communities through CDC or another agency. She added that community-collected environmental exposure data give researchers the opportunity to create publicly accessible health resources and databases that can complement CDC’s National Health and Nutrition Examination Survey, which focuses on dietary exposures, as well as chemical exposures, and the All of Us research program of the National Institutes of Health (NIH), a large database with genetic, health, and other information (NCHS, 2023; NIH, 2023). Another aspect of the PFAS Exchange website allows individuals to enter their own water or blood data and get help with interpretation (Silent Spring Institute, 2021).

Several speakers underscored the importance of continuing to support and expand investigations into the role of environmental exposures in causing cancer clusters. While genetics play a role, Brody emphasized that the increasing rate of pediatric cancers cannot be attributable to genetics alone, pointing to a need for a greater focus on the role of the environment and its interaction with genes. “So far, we’ve devoted a great deal more resources to unfolding the genetics than the environment side of that equation,” she said. She added that families strug-

gling with a pediatric cancer diagnosis should be given straightforward answers without assigning blame or guilt, and Ramírez-Andreotta noted that if families do experience feelings of guilt, that should be acknowledged with empathy.

Stating that “children are essentially canaries in the coal mine,” Schaefer said that environmental factors are likely to have a particularly large influence on cancer development in children because their developing immune systems are more vulnerable to their environments. An important complication in public communication around these issues is that pollution or other environmental exposures are often unknown or even hidden from the public, he added. Families also move, which changes their exposures. In light of that, Ramírez-Andreotta underscored the importance of combining rigorous environmental monitoring with social science methods to capture these details, understand participants’ social and ecological landscape, and recognize their interest and knowledge. “They’re experts in their own right, by living in that space and bearing witness to the scenarios that we need to be aware of,” she said.

LESSONS LEARNED AND FUTURE CONSIDERATIONS

The final workshop session focused on drawing out lessons learned and future considerations for improving our understanding of the potential contribution of cancer genomics information to community investigations of unusual patterns of cancer. To set the stage for the discussion, Roberta Ness reiterated the urgent need to advance research in this area. Pediatric cancers are rising, she said, and while the exact reasons are not yet clear, advanced technologies and emerging findings in metabolomics, genomics, epigenetics, and biomarkers present exciting opportunities to make progress on elucidating unanswered questions.

Jack Taylor, NIH scientist emeritus, National Institute of Environmental Health Sciences, studied the connection between epigenetics and adult cancers. Environmental exposures can make a person biologically older than their chronological age, and he is investigating whether that change is associated with cancer and other disease risks. He has also investigated mutational specificity in tumors from environmental exposures, epigenetic changes induced by tobacco or alcohol exposure, and associations between epigenetic profiles and cancer risks.

Logan Spector, professor at the University of Minnesota and chair of the Childhood Cancer and Leukemia International Consortium,¹⁶ said that molecular investigations of tumors and the genomes of both affected and unaffected community members represent an important opportunity to find related cases or even a causative exposure. He also reiterated the importance of effective commu-

___________________

¹⁶ See https://www.clic.ngo/ for information on the Childhood Cancer and Leukemia International Consortium.

nication in the critical—and sometimes fraught—relationship between researchers and a public that wants answers during cluster investigations.

Building upon his previous comments at the workshop, Schaefer stated that communities need cancer cluster guidelines that give them a voice and enable public, timely, and honest responses. He said that Trevor’s Trek Foundation bridges the communication gap between communities and public health departments when pediatric cancer cluster concerns arise, and the organization also compiles pediatric cancer data, identifies potential clusters, works with scientists, and focuses on exposure remediation efforts.

Jerald A. Fagliano, clinical professor and department chair at Drexel University, stated that while tremendous progress has been made in understanding how exposures affect the genome, it is difficult to apply that knowledge in cancer cluster investigations, which typically span a very long time. It may take years before a cluster comprising a sufficient number of cases is suspected and recognized, which means scientists must work with incomplete information about potential exposures that may have occurred years or decades ago. Fagliano said that technological advances that improve the ability to reconstruct exposures are encouraging, but he expressed his belief that these approaches are not ready to be integrated into the CDC Guidelines.

REFLECTIONS AND WRAP-UP

To wrap up the workshop, panelists reflected on the nuances of building productive relationships between researchers and communities and highlighted potential paths forward for improving cancer cluster investigations.

Spector said that his research has taught him that the social aspects of investigations are as important as the scientific ones. Although emerging molecular techniques are promising, he agreed with Fagliano that they are not yet ready to be used in cancer cluster investigations, which means that registry data are going to continue to be the initial go-to data source for health departments. Researchers need to anticipate that not everyone in a registry will be willing to engage with a more in-depth investigation. For example, a community that is strongly connected with or economically reliant on industrial activities suspected of being the source of an exposure may be hesitant to engage with an investigation that could end up threatening that industry. To navigate these issues, Schaefer said that it is important to foster strong connections between communities and public health departments. He suggested that health departments should hire community liaisons and compensate them for their time to make those connections, perform environmental exposure histories, and ensure proper data sharing. Most importantly, these liaisons could partner with local NGOs to share the work and address a community’s concerns, improving trust between communities and researchers. Fagliano agreed, stating that researchers

should make every effort to demonstrate their commitment and integrity and gain the community’s trust.

Panelists highlighted some of the factors that can cause communities to mistrust researchers during cluster investigations. Ness said that sometimes media reports can generate public interest in a new technology before it is ready to be employed, which can wind up eroding trust. Taylor added that communities can also lose trust when study participants, who have a right to know the investigation’s outcome, have to wait years for results or are given data that they are not equipped to understand. Ness pointed to scientists’ role in communicating with the public, addressing their concerns, and helping them to initiate cluster investigations, which she said some scientists may be hesitant to do because of the risk of being perceived as advocates. “Many scientists have drawn kind of a hard line between science and advocacy, so we don’t get out and spread the message; we don’t get out and disseminate more well-curated information,” Ness said.

Turning to considerations related to study design and methodology, Taylor noted that the genetic determinants of cancer are, in many ways, much easier to study and understand, but environmental exposures require lengthy investigations and ample resources to identify. “It’s been a really intractable box to split open and figure out,” he said. Ness added that large epidemiological studies need well-planned, intentional designs to minimize the risk of failure. Taylor said that his experience studying childhood cancers has taught him that parental exposures, especially in utero, are a critical component of cancer etiology. Fagliano agreed, noting that two well-known clusters, in Toms River, New Jersey, and Woburn, Massachusetts, had strong associations between maternal prenatal exposure to contaminated drinking water and childhood leukemia. He said that new molecular techniques could help to identify other exposures that use the same genetic pathways or operate via similar mechanisms, and Ness commented that there are several European registries of prenatal biological data that would be a rich data resource for this research.

Fagliano suggested that a national registry linked to banked biological specimens, which could be implemented in the next few decades, would be a powerful resource to assess and respond to potential pediatric cancer clusters and identify past exposures or associations. In addition, he said that health and environmental agencies should be working in parallel to address community concerns and identify hazardous exposures that may be actionable, which would increase trust and potentially prevent future cases. Spector agreed that such a registry would be helpful, but questioned whether it could be a real-time, reactive system unless the time between diagnosing a person with cancer and adding them to a registry were significantly shortened, allowing statistical anomalies to be identified more quickly.

Spector said that the general scarcity of tumor samples for pediatric cancers poses a significant barrier to employing new molecular tools in cancer cluster

investigations. “There are precious few studies that have the trifecta of good exposure data, good germline and pretreatment samples, and then the tumor,” said Spector, adding:

Spector said that funding is another important consideration if health departments and community organizations are going to incorporate molecular techniques into clusters, saying:

Foster said that collaborative efforts among academic researchers, public health practitioners, and affected communities will be vital to advancing data availability and interoperability, building multidisciplinary inputs like cancer registries, and practicing participatory science. Plon suggested that such collaborations could be incentivized by including requirements to partner with communities and make research data publicly available in order to obtain NIH grants and career promotions, similar to the requirements of the federally funded Patient-Centered Outcomes Research Institute.

REFERENCES

This page intentionally left blank.