Using Primary Teeth and Archived Dried Spots for Exposomic Studies in Children: Exploring New Paths in the Environmental Epidemiology of Pediatric Cancer

Abstract

It is estimated that 300,000 children 0–14 years of age are diagnosed with cancer worldwide each year. While the absolute risk of cancer in children is low, it is the leading cause of death due to disease in children in high-income countries. In spite of this, the etiologies of pediatric cancer are largely unknown. Environmental exposures have long been thought to play an etiologic role. However, to date, there are few well-established environmental risk factors for pediatric malignancies, likely due to technical barriers in collecting biological samples prospectively in pediatric populations for direct measurements. In this review, we propose the use of novel or underutilized biospecimens (dried blood spots and teeth) and molecular approaches for exposure assessment (epigenetics, metabolomics, and somatic mutational profiles). Future epidemiologic studies of pediatric cancer should incorporate novel exposure assessment methodologies, data on molecular features of tumors, and a more complete assessment of gene-environment interactions.

1. Introduction

It is estimated that 300,000 children 0–14 years of age are diagnosed with cancer worldwide each year.^[13] However, the absolute risk of cancer in children is low compared to adults; incidence is 183 cases per million in the United States.^[63] While relatively rare, these conditions continue to pose a serious public health problem. For instance, in high-income countries, cancer is the leading cause of death due to disease in children after the first year of life.^[13] Furthermore, while survival has continued to improve due to advances in treatment, there is still a high likelihood of relapse, and this young population with long life expectancy often have an increased risk of second cancers, an excess burden of multiple chronic health conditions as a consequence of their therapy, and diminished quality of life.^[31] This is notable, as there are currently 500,000 survivors of pediatric cancer living in the United States.^[39]

Pediatric cancers are heterogeneous and display a markedly different range of tumor types than in adults, including several classes which are largely exclusive to children. Based on data from the Surveillance, Epidemiology, and End Results (SEER) program in the United States,^[63] for children 0 to 14 years, acute lymphoblastic leukemia (ALL) is the most common cancer, accounting for 25.5% of all cancer diagnoses. Acute myeloid leukemia is the next most common type of leukemia in this age group, occurring at a rate one-fifth that for ALL. Central nervous system cancers, primarily occurring in the brain, accounts for 20.8% of cancer diagnoses, and together with acute lymphoblastic leukemia and acute myeloid leukemia make up one-half of cancer diagnoses among children younger than 15 years of age. The most common non-central nervous system solid tumor in the 0- to 14-year age group is neuroblastoma (6.8%), followed by non-Hodgkin lymphoma (6.1%) and Wilms tumor (5.3%). Other diagnoses that individually represented 2% to 4% of cancer diagnoses in this age group include Hodgkin lymphoma, rhabdomyosarcoma, non-rhabdomyosarcoma soft tissue sarcomas, germ-cell tumors, retinoblastoma, and osteosarcoma. Notably, these distributions vary globally, suggesting differences in etiologic factors by geography.^[51]

While key environmental exposures have been identified for adult cancers (e.g., smoking, benzene), much less is known in relation to pediatric cancer. A notable difference between adult cancers and pediatric cancers is the latency period associated with these conditions. For instance, smoking usually starts during adolescence or young adulthood, but associated malignancies do not become apparent until many decades later. However, several pediatric cancers are predominantly diagnosed in infancy (e.g., embryonal tumors such as neuroblastoma) or in early childhood (2 to 5 years of age), such as acute lymphoblastic leukemia. Therefore, the disruptions in molecular processes that may lead to pediatric cancer are likely different from those of adult cancers; at the least, the carcinogenic process in children is necessarily much shorter in time. There is also an emerging consensus that pediatric cancer is a disease that arises from dysregulated development.^[25] This is supported by recent large-scale germline genetic studies of pediatric cancer indicating approximately 8% of cases are due to pathogenic variants in well-established cancer predisposition genes, some of which are known to play a key role in development (e.g., HRAS).^[71] Additionally, genome-wide association studies of pediatric malignancies have pointed to the role of other important developmental genes on cancer risk (e.g., IKZF1).^[51]

In spite of this, there are few well-established environmental determinants of pediatric cancer or modifiers of pediatric cancer risk in the context of underlying genetic susceptibility. While there is extensive evidence that high doses of ionizing radiation are associated with pediatric cancer, the prevalence of this exposure is very low.^[44] Relatively common environmental exposures, including pesticides^[18] and air pollution^[26] have also been explored. While individual studies and meta-analyses have indicated associations between these exposures and some pediatric cancers, effect sizes are relatively modest. One problem with previous studies of environmental exposures includes the limitations of using questionnaire data to estimate exposure or the use of proxies for exposure assessment (e.g., residential information). Additionally, few studies have evaluated how environmental factors might influence genetic susceptibility to pediatric cancer. In this review, we propose the use of novel or underutilized biospecimens and molecular exposure assessment strategies for evaluating the environmental determinants and modifiers of pediatric cancer. Therefore, our objective is to consider novel and underutilized biospecimens (dried blood spots and primary teeth) for molecular-based exposure assessment strategies (e.g., epigenetics, metabolomics) in studies of pediatric cancer.

2. Novel or Underutilized Biospecimens in Environmental Studies of Pediatric Cancer

2.1. Using Dried Blood Spots to Estimate Perinatal Exposures

Newborn screening is a mandatory public health program in most developed countries worldwide that tests newborn babies for selected genetic, endocrine, and metabolic disorders. Under this program, dried blood spots (DBS) are routinely collected within 24–48 hours of birth for >98% of the nearly 4 million infants born in the United States each year.^[56] After screening is complete, residual DBS are stored for confirmation of positive results or reanalysis, if needed. In addition, some states also retain residual DBS for research purpose. Although the duration of retention and storage practice varies widely by state, from 30 days to indefinite, and policies regarding their retention and use for research have rapidly evolved over the past decade,^[48] DBS biobanks offer a valuable biospecimen for direct measurement of early life exposures.^[70]

Etiologic research into the environmental causes of pediatric cancer is hindered by many factors, including the relative rarity of pediatric cancer subtypes, selection bias in case-control studies, and recall bias in questionnaire-based exposure assessments. Additionally, accurate exposure assessment in the absence of biospecimens is notoriously difficult to achieve.^[75] Self-reported exposure data, sometimes collected many years after critical exposure windows, such as pregnancy, gives rise to the potential for exposure misclassification. Alternatively, biospecimens can provide accurate, unbiased estimates of exposures through the use of biomarkers. However, they must be collected prior to the onset of disease for use in etiologic research, because when collected contemporaneously with diagnosis, the disease process may alter exposure, biomarker status, or both (e.g., “reverse causality”). Given the low incidence of pediatric cancers, cohort studies for prospective collection of data and biologic specimens are not economically or practically feasible requiring follow-up of millions of children to accrue sufficient cases to produce statistically meaningful results. Therefore, DBS constitute the only population-based source of pre-diagnostic biospecimens for case-control studies.

The potential uses of DBS in environmental studies of pediatric cancer are extensive. Analytes of environmental exposures detectable in DBS reported to-date include benzene oxide,^[29] lead,^[6] mercury,^[61] arsenic,^[28] cadmium,^[17] cotinine,^[78,87] viral DNA,^[27] organochlorine^[53] and organophosphorus pesticides,^[84] perfluorinated compounds,^[88] polychlorinated biphenyls,^[53] polybrominated diphenyl ethers,^[52] bisphenol A,^[12,30,88] phosphatidylethanol,^[9] and polyfluoroalkyl chemicals.^[43,54] Additionally, there is increasing awareness that epigenetic signatures of environmental exposures may provide effective biomarkers of past exposure.^[46] Recent studies have shown that DNA methylation profiling, including epigenome-wide association studies, is possible using residual DBS as the source material.^{[36,38,58,81,89]}

Residual DBS offer a unique and extremely valuable resource to assess environmental exposures present at birth, with possible extrapolation to earlier in utero exposures. Neonatal biobanks are population-based, can supply samples from representative controls, and can provide basic demographic data through linkage to birth certificates. Further utilization of DBS to elucidate causal environmental agents and biological response associated with pediatric cancer risk could enable primary prevention for modifiable environmental factors.

2.2. Using Primary Teeth to Reconstruct Early-Life Exposures

There is growing evidence that deciduous teeth (i.e., “baby teeth”) serve as a stable repository for exposures occurring during the prenatal period and early childhood. This arises from the developmental biology of teeth—deciduous teeth begin mineralization prenatally towards the end of the first trimester and then continue to develop in an incremental manner forming daily (and even sub-daily) markings.^[66] Thus, naturally shed deciduous teeth are perhaps the only stable fetal tissue that can be attained in childhood non-invasively, which can then be used to reconstruct exposure history at fine-scale temporal resolution.

For decades teeth have been used to estimate long-term cumulative exposure to metals; however, these methods did not allow for longitudinal exposure assessment.^[60] Recently, high-dimensional analytics combining sophisticated histological and chemical analyses have been developed to reconstruct the early-life exposome (e.g., metals, organic chemicals, nutrients, infection, inflammation) by measuring tooth layers corresponding to specific life stages.^[57] While not yet applied to pediatric cancer, emerging autism and neurodevelopment studies are utilizing primary teeth to reconstruct the early-life exposome to measure the etiologic role of environmental exposures on disease risk.^[57,59] Notably, even if shed after the initiation of cancer therapy, primary teeth can be used to reconstruct exposures prior to cancer diagnosis.^[7]

Compared to using blood or other biological tissues, primary teeth provide a more accurate assessment of 1) exposure dose, 2) the developmental period during the time of exposure, and 3) exogenous exposures during earliest periods of fetal development that often have short half-lives.^[8] Moreover, teeth are easy to transport and store being stable at room temperature. When taken together, these factors point to the importance of utilizing teeth as novel biospecimens in environmental studies of pediatric cancer.

3. Novel or Underutilized Molecular Approaches for Assessing Exposures in Environmental Studies of Pediatric Cancer

3.1. Using Epigenetics as Biomarkers of Environmental Exposures

Epigenetics refer to changes in gene expression that are not due to the DNA sequence. One type of epigenetic process is DNA methylation – the addition of a methyl group at the fifth carbon of a cytosine that is typically followed by a guanine. DNA methylation is the best-studied epigenetic mechanism in humans. The development of DNA methylation arrays has allowed researchers to systematically measure methylation at individual C-phosphate-G sites (CpGs) throughout the genome for a relatively inexpensive cost. Thus, there has been a prolific number of epigenome-wide methylation studies published in the past decade, including those in relation to environmental exposures.^{[32,38,40,46,67,73,77,81,85]}

Because environmental factors can alter DNA methylation at specific CpGs, these specific CpGs can be used as a robust biomarker of exposure in adults^[77,90] and can capture in utero exposures in newborns.^[73] There is evidence that some altered DNA methylation in newborns can persist into childhood and adolescence.^[41,74] There are also studies that have successfully evaluated the impact of air pollution on DNA methylation in children.^[40] Leveraging DNA methylation as a biomarker of exposure may overcome the exposure assessment challenges in previous studies of pediatric cancer, as it can objectively capture exposures in utero and after birth. A limitation of this approach is that large epigenome-wide studies of environmental exposures are needed to identify robust CpGs related to the exposure of interest. Alternatively, epigenome-wide meta-analyses of several smaller studies can overcome this limitation, as meta-analyses of smoking,^[41] alcohol,^[49] and air pollution^[34–35] have identified robust CpGs in relation to the exposure.

DNA methylation may be a novel, cost-effective molecular tool that can be utilized to assess environmental exposures for pediatric cancer research. One DNA sample can provide information on many different environmental factors that may contribute to pediatric cancer. While it is ideal to collect samples prior to the time of diagnosis, ongoing assessments are evaluating if there are metastable epialleles related to pediatric cancer that can evaluated with samples collected at the time of diagnosis.^[23] Because DNA methylation is tissue specific and is influenced by both genetic variation and environmental factors, alterations in DNA methylation may also provide mechanistic insight into treatment response and survivorship, as have been reported,^[14,50] to improve clinical outcomes through epigenetic therapies in children with cancer. As noted, while a few studies have utilized DBS for epigenome-wide assessments, to our knowledge, there have been no assessment exploring the utility of primary teeth to evaluate DNA methylation patterns, providing territory for novel explorations of pediatric cancer risk.

3.2. Using Untargeted Metabolomics as Measure of the Exposome

Metabolomics, the global measurement of small molecules in a biological specimen that correlate with cellular function and dysregulated physiology, holds promise for understanding the environmental determinants of pediatric cancer incidence and survival.^[19,68] Metabolomics experiments most commonly utilize nuclear magnetic resonance (NMR), liquid chromatography-mass spectrometry (LC-MS), or gas chromatography-mass spectrometry (GC-MS) methodologies.^{[11,20,24,86]} The type of technology also determines the class of small molecules that can be measured, while there is some overlap in methodologies. Therefore, the choice of method in a particular study depends on practical considerations such as cost and sample type, as well as the investigator’s objectives. Each may contribute valuable data to studies of pediatric cancer.

Metabolomics may be useful for comprehensively characterizing an individual’s “exposome,” as the metabolome is a biochemical read-out of genetic predisposition, diet, environmental exposures, general metabolic health, and other traits.^[62,72] Thus, when applied appropriately, metabolomics may allow for the simultaneous assessment of diverse exposures using a single experimental approach. This is particularly relevant to environmental health research, because humans are not exposed to one chemical at a time or even in small groups, but to complex mixtures. Further, by measuring dysregulated biological pathways, metabolomics enables the measurement of biological response to transient exposures that occurred weeks, months, or years ago and may be long gone from the body.^[21] Thus, metabolomics has potential to accelerate the discovery of exposures associated with pediatric cancer, which has proven a difficult undertaking.

Of particular note to investigators interested in the etiologies of cancer and other childhood diseases, mass spectrometry-based metabolomics have been applied to samples such as archived DBS, cord blood, and deciduous teeth.^[4,42,69] These samples may provide information on exposures during critical developmental periods, without concern for recall bias and, sometimes, with high temporal resolution.

Applications for metabolomics exist along the continuum of pediatric cancer research from etiology to survivorship. In the context of acute lymphoblastic leukemia, metabolomics has been applied to study the tumor microenvironment, treatment response, vincristine-induced neuropathy, and treatment-related fatigue.^{[10,15,22,76,80,82]} In addition, as long-term survival for children with cancer continues to improve, the community has increasingly recognized the critical importance of survivorship research. Two-thirds of survivors report at least one chronic health condition in adulthood, including second malignant neoplasms, cardiovascular disease, and frailty.^[16,37,64] In adult populations, recent studies have demonstrated associations between metabolomic profiles, future cardiovascular disease risk, and all-cause mortality;^[47,65] this suggests the intriguing possibility that metabolomic profiling of pediatric cancer survivors may be useful for predicting outcomes and risk of late effects, particularly cardiovascular events.

Barriers to the broader application of metabolomics include cost; sample type and storage; lack of large, shareable spectral libraries contributing to difficulty in feature identification; heterogeneity in sample and data analysis pipelines; and data dimensionality.^[20,86] However, the field is under active development, and it is anticipated that metabolomics approaches will become more comprehensive and powerful with time as these limitations are addressed. In combination with robust epidemiologic data and additional ‘-omics’ data such as from genomic studies, metabolomics has the potential to be a powerful tool for elucidating the developmental origins of pediatric cancer, understanding treatment response, and managing the health of long-term survivors.

3.3. Using Somatic Mutational Signatures in Pediatric Tumors to Understanding the Carcinogenic Role of Environmental Exposures

Somatic mutations found in cancer genomes may be the consequence of the intrinsic infidelity of the DNA replication machinery, exogenous or endogenous mutagen exposures, enzymatic modification of DNA, or defective DNA repair.^[3] Biological processes that cause mutations in somatic cells leave a mutational signature.^[3] To date, 30 distinct mutational signatures have been identified in 40 different cancer types using whole-genome and whole-exome sequencing, which include associations with age, smoking, and ultraviolet light in addition to other exogenous and endogenous exposures.^[1–2] A previous analysis of smoking-related cancers revealed a mutational signature of tobacco exposure which was present in cancers derived from cells directly exposed to tobacco smoke.^[1] Specifically for pediatric cancers, researchers have identified overlap with previously identified mutational signatures.^[33,55] Additionally, unique signatures have been identified in atypical teratoid rhabdoid tumors (ATRT), ependymoma, B-lineage acute lymphoblastic leukemia, acute myeloid leukemia, neuroblastoma, and Wilms tumor.^[33,55]

Understanding mutational signatures of pediatric cancer can help identify potential endogenous and exogenous exposures that are important in the etiology of pediatric cancers.^[2] Gröbner et al.^[33] noted that approximately 50% of pediatric cancers studied may have a targetable genetic mutation that can be used to improve precision therapeutics in children. Current resources to study mutational signatures in cancer include COSMIC (Catalogue Of Somatic Mutations in Cancer)^[83] and, specifically for pediatric cancers, the PedPanCan^[45] and PeCan Variant Knowledge Base.^[79] Future studies would benefit from focusing on expanding cohorts to include additional types of pediatric cancer to better understand the impacts of these mutational signatures on etiology across cancer types.

4. Conclusion

While there have been tremendous strides in improving outcomes for children with cancer, there is still a great deal of work related to disentangling the etiologic origins of these conditions. Future studies should incorporate novel exposure assessment methodologies using unique biospecimens, such as those proposed here. Additionally, there are several well-established germline genetic variants that have been identified for pediatric cancers, which have yet to be leveraged in studies of environmental exposures in the context of genetic susceptibility.^[71] Through these efforts, it is hoped that our understanding of the causes of pediatric cancer can be better ascertained, leading to novel surveillance or prevention strategies.