Challenges and opportunities in research on early-life events/exposures and cancer development later in life
- First Online:
- Cite this article as:
- Mahabir, S., Aagaard, K., Anderson, L.M. et al. Cancer Causes Control (2012) 23: 983. doi:10.1007/s10552-012-9962-5
It is becoming increasingly evident that early-life events and exposures have important consequences for cancer development later in life. However, epidemiological studies of early-life factors and cancer development later in life have had significant methodological challenges such as the long latency period, the distinctiveness of each cancer, and large number of subjects that must be studied, all likely to increase costs. These traditional hurdles might be mitigated by leveraging several existing large-scale prospective studies in the United States (US) and globally, as well as birth databases and birth cohorts, in order to launch both association and mechanistic studies of early-life exposures and cancer development later in life. Dedicated research funding will be needed to advance this paradigm shift in cancer research, and it seems justified by its potential to produce transformative understanding of how cancer develops over the life-course. This in turn has the potential to transform cancer prevention strategies through interventions in early-life rather than later in life, as is the current practice, where it is perhaps less effective.
Interest in the role of early-life events/exposures in cancer development has been growing. Two recent commentaries argued that early-life exposures have important consequences in adult cancers [1, 2]. To assess the state of the science and address areas for emphasis and support, the National Cancer Institute (NCI) convened an Expert Panel Workshop on Early-life Events and Cancer on 25 May 2011. The workshop consisted of lectures in the morning and a working group session with outside experts and NCI scientists in the afternoon. The scientists were experts in the epidemiology of adult cancers, puberty and adolescence, animal model systems, epigenetics, IARC priority areas, and cancer registry linkage systems. This commentary is a summary of the workshop presentations and discussions.
The workshop followed up on the Early Reproductive Events and Breast Cancer workshop, held in 2003 (http://www.cancer.gov/cancertopics/causes/ere/workshop-report). With growing evidence for the role of early-life factors related to malignancies other than breast cancer and the need to assess progress from this workshop, the expert panel was convened. The scope of the current workshop was expanded to other cancers in addition to breast cancer with the goal of stimulating and facilitating epidemiologic and molecular research in the area of early-life influences and cancer in adult humans (pediatric cancers were deemed outside the scope of this workshop because they are being covered by other activities).
Definition of early-life
While early-life is often defined as the in utero period, evidence indicates that peri-conceptional and pre-adult exposures, as well as paternal influences, are also important. Thus, the working group (WG) considered the period of early-life broadly as peri-conception through 20 years of age. Early-life can also be categorized by well-defined critical periods or events such as peri-conception, in utero, birth, adrenarche, puberty, and the end of growth/maturation.
State of the current evidence
In human studies, several early-life factors have emerged as important risk factors for cancer development later in life, with growth/stature and age at menarche the most consistent. Early menarche has been an established risk factor for breast cancer for some time, an observation which sparked interest in the possibility that this early time period is critical in the etiology of this disease. Greater growth/stature is a risk factor common to many cancers, including breast, ovary, endometrium, kidney, testicular, colon, and rectal cancer, and malignant melanoma, non-Hodgkins lymphoma, and leukemia [3, 4]. Size at birth (i.e., birth weight and length) also is associated with breast cancer risk. A pooled analysis of 32 studies with 22,058 breast cancer cases reported that birth size and in particular birth length are independent correlates of breast cancer risk in adulthood . Birth size, a possible proxy for the prenatal environment and growth, has not been adequately studied for cancer outcomes. Also, epidemiologic studies have been limited by the lack of known intermediate markers for cancers. In addition to early growth, having been breast-fed, exposures around the time of puberty, and other factors warrant further exploration . Other in utero factors have been linked to later cancer risk in several organs, particularly the breast and testis [7, 8]. For example, pre- and perinatal risk factors, such as adverse maternal health, maternal smoking, and low birth weight increase testicular cancer risk . Early-life infections, likely in combination with immunogenetic factors, have been associated with adult occurrence of several forms of malignancy. Examples include, human T-cell leukemia virus type 1, which is primarily transmitted in breast milk and is endemic in southwestern Japan, the Caribbean, central Africa, parts of South America, and New Guinea, is the causative agent of adult T-cell leukemia and HTLV-1-associated myelopathy/tropical spastic paraparesis . Certain human papilloma virus (HPV) infections, which are usually acquired during adolescence, cause almost all invasive cervical cancer cases worldwide following a decades-long process . Epstein-Barr virus (EBV), which may be acquired very early in life or in adolescence, has been linked with EBV-related malignancies that may occur over a range of ages from childhood (Burkitt lymphoma in African children) to adolescence or young adulthood (nasopharyngeal carcinoma and Hodgkin lymphoma) to young or middle-aged adulthood (AIDS-associated immunoblastic and primary central nervous system lymphomas . A single acute exposure to radiation to those who were in utero or in early childhood at the time of the atomic bombings of Hiroshima and Nagasaki resulted in significantly elevated risk of malignancies decades later . Migrant studies have shown that men of the first generation moving from a low testicular cancer risk area (e.g., Finland) to a high risk area (e.g., Sweden) keep their low risk . Several of these studies point toward the importance of critical time periods for the carcinogenic exposures.
Animal studies have provided strong support to link early-life exposures with cancer development . Even preconceptional carcinogenesis has been demonstrated for a variety of types of radiation and chemicals, with sensitivity for all stages of germ cell development . Recent data suggest that estrogen exposure in general may predispose to prostate [17, 18], uterine, and vaginal tumors  and may have relevance to endocrine-disrupting chemicals. Animal studies have also demonstrated that exposure to environmental endocrine disruptors cause alterations in mammary gland development and this increases susceptibility to breast cancer depending on the timing of the exposure . Mechanistic studies have been useful for follow-up in human research. For example, the Agouti mouse model was used to convincingly demonstrate that maternal diet affects epigenetic modifications and phenotypic changes in the offspring that are potentially linked to cancer . With strong evidence from a variety of animal model systems, new work to relate the human research to animal studies is needed and emerging [22, 23].
Progress following the 2003 workshop
The Workshop on Early Reproductive Events and Breast Cancer in 2003 highlighted the known relationships between early age at full-term pregnancy and breast cancer risk. Multiple mechanisms have been proposed for the protective effect. These mechanisms include the induction of a stable morphological phenotype, the alteration of mammary stem cell number, an intrinsic altered cell fate defined by a distinct gene expression signature, and an altered systemic hormonal milieu. Whatever the mechanism, the result is a cell population which exhibits a decreased cell proliferation and/or an increased cell apoptosis activity, resulting in a block in promotion of the initiated state. The evidence for these mechanisms is of variable degrees of strength and has evolved since the Workshop. There is little evidence across animal models for an induction of a stable morphological phenotype. There is a great deal of evidence for a state of decreased proliferating/increased apoptotic activity. This evidence is consistent with the changes in gene expression seen in different animal and human studies. There is conflicting evidence for pregnancy leading to a decreased stem cell activity with the majority of the evidence favoring no long-term change in stem cell activity. In contrast, recent experiments have highlighted the role of systemic hormonal changes as a dominant factor in controlling the altered mammary epithelial cell behavior. The role of specific factors produced by the stromal microenvironment has not been extensively studied.
Studies using human samples have expanded in recent years. There have been several recent publications that have defined gene expression changes in the breast epithelial cells, which have produced results consistent with studies in animal models. Such studies have the potential to generate a gene signature that defines the persistent protective state generated by an early first full-term pregnancy. At this time, the individual studies have relatively small population sizes, but the information is highly encouraging for further exploration. More studies and a meta-analysis would put these data and conclusions on a firm footing. Additionally, a proteomic analysis is just in its infancy, and such studies might also be informative. The end result will be to define pathways that are inducible and/or targetable, so viable prevention strategies can be designed and implemented.
There are four major methodological challenges to the study of early-life events/exposures and cancer development in later-life. These are (1) site-specific adult cancers, being rare events with different etiologies; (2) the long latency period, which required epidemiology studies of early-life events/exposures to be large and expensive; (3) the lack of validation of most of the currently used exposure assessment tools for early-life exposures, for example, most studies have used questionnaires developed for exposure assessment in adults; and (4) inadequate development of long-lasting biomarkers correlated with early-life exposures and others correlated with cancer development many years later. The identification of appropriate biomarkers in either of these classes would allow smaller studies, and those without extended follow-up, to make substantial contributions.
Studies that model exposures have been conducted, but early-life exposure assessment procedures have not yet been validated. Researchers require information on how to assess and validate exposures retrospectively in a comprehensive fashion and to develop biomarkers that reflect earlier exposures. Many exogenous and endogenous factors etiologically relevant for the majority of cancers, such as smoking, nutrition, environmental toxicants, chemicals, heavy metals, and hormones, can be measured in biological fluids but reflect recent exposure. The challenge is to assess and integrate these measurements over the life-course to understand how lifetime exposure relates to cancer development.
Birth cohort studies are some of the most valuable resources to address early-life factors and cancer development. However, birth cohort studies carry considerable challenges such as large sample sizes and significant expenses.
In terms of inter-disciplinary research, the approach has been to extrapolate results from animal studies to humans. However, the reverse may also be informative to investigate mechanisms. For example, human early-life exposure factors, such as folic acid supplementation during pregnancy, infant formula use for newborns, diagnostic radiation, and common allergy medications in childhood, can be modeled in animal studies to identify molecular mechanisms associated with cancer development, such as metabolic, hormonal, and immunological modulation, gene methylation and imprinting, and DNA repair capacity.
Since very few prospective epidemiology studies of early-life events/exposures have been conducted and because the focus has been on breast cancer, enormous gaps in knowledge are evident, not only in the epidemiology, but in the studies attempting to understand biological underpinnings. General examples of research gaps include questions whether some cancers originate in utero in a minority of the population and what exposures and events in early-life affect the risk of specific cancer development in later-life. In terms of early-life nutrition, there is a wide gap in knowledge regarding how diet in early-life interacts with the gut microbes to affect cancer risk. The gut microbiome in infancy and childhood could be an important link between nutrition and cancer development later in life. Even regarding breast cancer, although later age at first birth is an established risk factor, it is unclear by what mechanisms this effect is mediated, for example, it remains unknown whether pregnancy at an early age changes the phenotype of the breast. Several early-life events are routinely collected in medical records, such as gestational factors, birth weight, and pregnancy complications, but the biological mechanisms that explain the observed epidemiological associations with cancer have not been established. Additional research on stem cells and cell lines may be useful to establish the impact of early-life events and to facilitate further research on mechanisms. In this regard, the variability among different cells in tissues can confound the findings and force researchers to average results, thus masking potential biomarkers of early-life exposure and cancer risk.
Developing and utilizing innovative study frameworks and identifying cohorts (existing or planned) with relevant intermediate markers or risk factors would be useful. Further work is needed in medical record retrieval and questionnaire-based approaches that capture early-life exposures. Additional gaps include identification of available records or databases that can be utilized for research questions in the life-course related to cancer outcomes or cancer risk factors. In addition, identification of new animal models and determination of biological markers sufficiently predictive of early-life events/exposures in humans would aid in moving the research forward. Also, understanding social/behavioral factors in animals could be applied more often as tools for research in humans.
Important research gaps also exist for predictive markers for cancer risk based on what happens biologically at a young age. Markers that predict malignancy or pre-malignant conditions would allow assessment of early-life exposures with relevant outcomes without having to wait 50 years for cancer development. Given recent knowledge of the critical importance of early-life events on development of obesity, diabetes, and cardiovascular disease, more work needs to be done to evaluate the impact of early-life events in relation to later-life risk factors.
Most of cancer research in human populations has focused on a narrow range of exposures in the middle to last 25th percentile of the lifespan. While this narrow age range yields the highest number of cancer cases, it is a phase in which cancer prevention efforts are more difficult and perhaps less effective. The emerging evidence that early-life exposures affect cancer development later in life calls for a refocusing of efforts targeting the early-life spectrum: maternal (as well as paternal), in utero, infancy, childhood, and adolescence exposures. This paradigm shift in cancer research has the potential to translate into substantial gains in cancer prevention and control later in life.
One potential option to quickly expand research efforts on early-life exposures and cancer development is to leverage several existing large-scale prospective cohort studies to collect either retrospective data on early-life and link to cancer outcomes or collect new prospective data on early-life exposures. NCI has funded several prospective cohort studies of adults such as the Nurses’ Health Study (NHS), the Physician Health Study, The Multi-Ethnic Cohort Study, The Southern Community Cohort Study, and Minnesota Breast Cancer Family Study to address cancer etiology. Potential opportunities exist for these and other resources to be exploited for potential ancillary studies of early-life exposures and cancer. The proof of principle for this approach has already been established by some of these existing cohorts. Investigators of the NHS II (pre-menopausal nurses aged 25–43 years) were able to retrospectively collect data on diet during high school. Using these data, the investigators reported that dietary fat and red meat consumption during adolescence elevated risk for breast cancer later in life [24, 25]. Body fatness at young ages has a strong inverse relation to breast cancer throughout life . Investigators have retrospectively collected data on several risk factors and diet during childhood and adolescence. Other studies have reported that weight and adiposity at approximately age 12 were inversely associated with adult mammographic breast density (a strong risk factor for breast cancer)  and prostate cancer . Ongoing studies of special populations may also be informative. For example, opportunities are available to capture events about pregnancy and early childhood form the mothers of participants in longitudinal studies such as in the NHS II . Follow-up of atomic bomb survivors indicated that exposure for those less than age 20 years resulted in higher risk of breast cancer, compared with radiation exposure at older ages . Prospective studies of early development that capture established cancer risk factors such as the onset of puberty, menarche, and maximal attained height, exemplified by the Breast Cancer and the Environment Research Program (BCERP), provide another promising approach to more detailed measurement of critical early-life exposures and pathways to adult cancer . Other opportunities for existing resources include the expansion of enrollment and data collection of ethnic/minority in studies of early-life exposures and cancer.
There are also opportunities for pooling projects to study the role of early-life exposures and cancer development. Some efforts are already underway in consortia. For example, the International Childhood Cancer Cohort Consortium (I4C) of prospective cohort studies will soon be able to pool data to address various early-life exposures and cancer risk factors. Opportunities also exist to leverage other ongoing studies. For example, The Consortium of Health-Oriented Research in Transitioning Societies (COHORTS) collaboration involves five long-standing prospective cohort studies from Brazil, Guatemala, India, the Philippines, and South Africa, which are being assessed for adult outcomes in the offspring . The Hyperglycemia and Adverse Pregnancy Outcome (HAPO) study is an international 15-center multicultural study of over 25,000 pregnant women, designed to assess the association between maternal hyperglycemia and adverse outcomes .
Childhood outcomes are also being assessed, for example, a subset of HAPO subjects were assessed for the association between maternal glucose at 28 weeks of gestation with obesity in 2-year-old offspring and found only weak associations . Expansion and follow-up of this and other cohorts, which have rich repositories of biomarkers, to assess cancer-related factors such as age at menarche will be important for cancer prevention strategies. There are also several European birth cohorts that are being followed and represent a substantial collective resource that can be harnessed for cancer investigations. The European birth databases and birth cohorts can be linked to various registries to assess the role of early-life exposures in cancer risk, both in children and in adults. In the United States, opportunities exist to assemble many state-based birth and cancer registries. These data can be combined to look for early-life exposures related to neoplasia. Opportunities also exist to create linkages between birth defects and genetic syndromes to study interactions with prenatal factors and cancer risk.
An intermediate alternative between birth cohorts and the recall of adolescent exposures is to enroll participants in a cohort while they are in later childhood and can provide data themselves. This has been done in the Growing Up Today Study (GUTS and GUTS II), which includes approximately 25,000 male and female offspring of the NHS II cohort. GUTS and GUTS II are already providing data, for example, showing that alcohol consumption between 18 and 22 years significantly increase risk for biopsy-confirmed benign breast disease .
In terms of the effects of maternal nutrition and phenotypic outcomes in offspring, including cancer, convincing evidence has emerged from transgenic animal studies pointing to epigenetic switching mechanisms. However, it is not clear whether transgenic mouse models are appropriate for early-life nutrition and cancer in humans. The effort to connect animal studies to human studies is challenging but worth exploring. While better animal models of early-life exposures and cancer development are needed, it seems that simulation models can be incorporated to assess potential variations in early-life exposures.
Probably, the biggest opportunity lies in early-life biomarker assessment and development to predict tumor risk later in life without generally having to wait for many years (30–50 years) for the cancer outcome. A centralized pathology review of specimens collected during the early-life period and archived can be utilized for biomarker development. It is possible that some of the emerging “omics” technologies would be useful for this purpose. In the mean time, at least for breast cancer, there are several established early-life markers of risk: birth weight, breast density, estrogen concentration, puberty, age at menarche.
Biomarkers of significance from animal studies can be explored in human epidemiology studies. However, new discovery-based technologies may yield greater success for early-life cancer studies. Epigenetics is receiving much attention because distinct programs of gene expression underlying the development of different cell lineages and tissues are executed (with minor exceptions) without changes in the genome, so embryonic development can be viewed as an epigenetic affair. In addition, it is also possible that as development proceeds, stem/progenitor cells as well as differentiated cell in a given organism may accumulate distinct epigenetic marks depending on the environment and endogenous clues to which the embryo is exposed. Adverse exposures to environmental factors may deregulate epigenetic buffers resulting in changes that may serve as biomarkers of early-life exposure and represent susceptibility factor later in life. Since there is a different requirement for retaining and removing epigenetic modifications during stem cell and lineage differentiation, different stages of embryonic development may constitute windows of susceptibility to environmental influences. This duality of epigenetic information can sometimes be inherited across multiple generations and may also be a basis for trans-generational inheritance [36, 37]. The intrinsic reversibility of epigenetic alterations may also represent an opportunity for the development of novel strategies for cancer prevention.
Proteomic studies on people with cancer also are available; however, no large-cohort data have been obtained before cancer onset. A transcriptomics profile (gene activation/inactivation or expression profiling), and ultimately a metabolomic profile, may reveal relatively strong risk factors. This is difficult to accomplish because the massive amounts of data resulting from a large number of variables would be linked to an at-risk cohort, and only the easily accessible variables would be identified. Therefore, a well-controlled experimental design in specific cellular subtypes will need to be considered. For example, improving methodologies for better selecting target cells and making future-generation sequencing technology suitable for “single-cell omics” should overcome the current problems associated with the analysis of heterogenous cell populations .
Initiating new birth cohorts in the United States can be difficult because of the size and scope of the National Children’s Study (NCS). The ongoing NCS is a longitudinal birth cohort study that plans to enroll 100,000 children from before they are born until they reach the age of 21 years to investigate the effects of modifiable and genetic factors on health. However, support for birth cohorts should be increased because these studies have significant value. Cohorts from different countries provide a unique opportunity to incorporate ethnicity, assess additional exposures, and apply useful comparisons to early-life exposure. The World Health Organization (WHO) is planning to initiate several birth cohorts in developing countries, which could provide additional opportunities to collect reliable and valid exposures as well as biospecimens to test associations with cancer risk. There are also several European birth cohorts with rich repositories of data and biospecimens. Collectively, birth cohorts around the world have the potential to answer many questions related to early-life factors and cancer development.
NCI has played a modest role in supporting the International Childhood Cancer Cohort Consortia (I4C). If the consortia results look promising, then the aim will be to expand cohort studies beyond childhood cancer, which will take a few decades. The Jerusalem Cohort is an older perinatal study that includes no biological components. Some small European cohorts have brought themselves under the consortia umbrella as well, and there also is interest from low-income countries. Emerging large-scale projects such as the Human, Heredity, and Health (H3) African Initiative can be equipped for research efforts targeting early-life exposures and cancer development in several African countries.
Understanding the biological underpinnings of early-life events/exposures and cancer development later in life requires the collection and storage of biospecimens. New efforts, such as placenta and umbilical cord banking, need to have dedicated funding resources. In some countries, private companies are preserving umbilical cord tissue for at least 10 years as a unique resource. An important issue is whether biorepositories should be encouraged for all studies or only in the context of a formal study design. NCI has been a major proponent of biorepositories for decades; however, the freezers are filled and samples often remain unused—perhaps rampant biospecimen collection is not the best solution. However, many groups have similar or overlapping biospecimen needs—the development of a system that indicates biospecimen availability and how access can be obtained would be useful. Rules for access are scattered and inconsistent. In addition to NIH publishing data-sharing policies, NIH could play an important role in standardizing access for US investigators to biospecimen and data resources from NIH-funded large-scale epidemiology studies. There are additional issues with commercial biobanks, for example, some will not sell samples, while others will sell them to secondary vendors. Also, in prospective biobanking or discovery-based research, it is unclear how to handle patient data. Unfortunately, NIH does not yet have general guidelines in this area. The Genome-Wide Association Studies (GWAS) and the NCI’s Prostate, Lung, Colon and Ovary (PLCO) models may provide some insight into guidelines regarding access and patient data.
Databases and record linkages
Assembling historic databases may prove difficult. However, it is worth undertaking for school-age children studies or adolescent studies in addition to birth cohort studies. It is possible that in the United States, several types of useful data have been linked, but are within different record retrieval systems. Throughout life, people in the United States see physicians, but in the absence of lifetime electronic medical records, it is difficult to conduct longitudinal research for most of the US population. However, some networks (crn.cancer.gov) have been established that could be used for longitudinal research with electronic medical records, although most are relatively recent and researchers may still need to wait for outcomes. Other innovative work may be possible with these records, however. Model systems such as record linkages in Scandanivian countries will be extremely useful for studying early-life exposures and cancer.
Research funding needs
Within NCI, there is no established grant funding mechanism geared toward early-life events/exposures and cancer development. To date, excluding childhood cancers, only a few grants investigating early-life factors and cancer risk in adults have been funded by the NCI. Research on early-life factors has the potential to produce transformative understanding of how cancer develops over the life-course, potentially leading to prevention strategies through interventions in early-life. Clinical partnerships with obstetric and pediatric groups also can be forged, and existing NIH Roadmap Initiatives can be leveraged. This sort of approach likely will have cross-institute appeal. Considering the limited number of funded grants investigating the role of early-life factors and cancer development later in life and the potentially transformative nature of this research paradigm, dedicated research funding should be prioritized by the NCI to encourage this type of research.
Conflict of interest