Linking Population-Based Registries to Study Early-Life Factors and Cancer

June 22, 2016, by DCEG Staff

by Victoria A. Fisher, M.P.H.

Early-life factors from conception to childhood have important consequences for cancer risk later in life. Scientists have found evidence that certain exposures, like birth size and growth/stature, could play a role in future cancer development. 

However, epidemiological studies of these factors face significant methodological challenges, including the long latency period between exposure and disease diagnosis and keeping track of the large number of subjects that must be followed over many decades. As a result, very few prospective studies have been conducted. Participants in retrospective studies may not know or have trouble recalling the circumstances of their birth or events from childhood.

“In our cohort studies, it’s difficult to obtain early-life exposure information because we often enroll people when they’re in middle to older age, to study them at a time when their risk of cancer is high,” said Cari Kitahara, Ph.D., tenure-track investigator in the Radiation Epidemiology Branch.

To tackle these hurdles, DCEG investigators are partnering with international colleagues to apply cutting-edge analytical approaches to existing data resources, like population-based registries. Studies linking birth, school, and hospital registries with cancer registries allow investigators to explore early-life characteristics and cancer; by pooling data from multiple sources, investigators can explore even rare exposures and outcomes.

“Population-based registries are one of the few ways in which we can look at a person’s exposures that occur at or before the time of birth, and then relatively easily learn the diseases they developed later in life,” Dr. Kitahara said. “These registries are a very unique resource.”

In particular, the Nordic countries—Denmark, Finland, Norway, and Sweden—have a long tradition of collecting high-quality data through national population-based registries. Denmark has the oldest cancer registry, established in the 1940s. Birth registries were established later. Reporting is mandatory, thus the quality and completeness of the data are very good.

Collaborative Studies with Nordic Data

Since the mid-2000s, Rebecca Troisi, Sc.D., staff scientist in the Epidemiology and Biostatistics Program, has collaborated with investigators from Denmark, Finland, Norway, and Sweden, to organize pooled linkage studies to explore maternal and early life factors and subsequent cancer risk.

“The birth registries in these countries have information on a woman’s pregnancy and maternal characteristics, as well as the baby’s characteristics; we link this historical information to the cancer incidence data,” Dr. Troisi said. “We can continually do linkages and obtain information across an entire lifespan. These rich datasets allow us to study not only pediatric cancers, but those that arise later in life.”

Dr. Troisi and collaborators have published a number of important findings. They reported an association between high birth weight and increased risks for several childhood tumors (Bjorge, 2013) and breast cancer in adulthood (Troisi, 2013); both low (<2,500 g) and high (≥ 4,000 g) birth weight increased risk for childhood testicular germ-cell cancer (Stephansson, 2011). Pregnancy factors did not appear to play a role in the etiology of bone cancers (Troisi, 2013), and hyperemesis gravidarum (severe nausea and vomiting during pregnancy) did not increase cancer risk in offspring (Vandraas, 2015). Dr. Troisi is currently analyzing how early-life factors relate to the mother’s subsequent cancer risk, after pregnancy.

Other DCEG investigators are utilizing this collaboration to study a variety of exposures and subsequent cancer outcomes. For example, Dr. Kitahara is looking at pregnancy characteristics, complications, and birth outcomes in relation to subsequent thyroid cancer in the mother and offspring.

“I was interested in pregnancy-related factors and thyroid cancer for a while, but was having difficulty finding the right data resource; I needed a very large population,” Dr. Kitahara said. “When I discovered that Rebecca was working with this group to combine registries, it seemed like the perfect opportunity.” This latest study is in the early stages; collaborators in Denmark are performing a new linkage that will allow investigators to analyze different outcomes, including thyroid cancer.

Copenhagen School Health Records Register

Another important resource for epidemiologic studies of early-life factors is the Copenhagen School Health Records Register (CSHRR). In collaboration with Dr. Jennifer Baker, Dr. Thorkild Sørensen, and others in Denmark, DCEG investigators have conducted a number of studies utilizing data from the register.

“There are many advantages to working with these data,” Michael B. Cook, Ph.D., tenure-track investigator in the Metabolic Epidemiology Branch (MEB) said. “They have a large population with excellent quality data collected over several decades. Individual identification numbers allow us to link with data collected later in childhood and from the national cancer registry.”

The CSHRR includes health examination information on nearly 375,000 children (generally between the ages of 7-13) who attended school in Copenhagen, Denmark, from 1936 to 2005. Among many factors, they measured height and weight annually. Originally, the data were collected for record-keeping purposes only, not necessarily for research.

“The paper records were going to be destroyed, but Dr. Sørensen saved them and had them electronically recorded,” Dr. Kitahara said. “We are lucky that he recognized their tremendous research value.”

Several published studies have examined early-life anthropometric factors and future cancer risk using the CSHRR. Dr. Cook and collaborators found associations between childhood height with prostate cancer risk (Cook, 2013) and prostate cancer mortality (Aarestrup, 2015). Childhood body mass index (BMI) was weakly associated with prostate cancer risk, but was attenuated when adjusted for height (Aarestrup, 2014). In addition, Dr. Cook found evidence that individuals with higher childhood BMI were at elevated risk of developing esophageal adenocarcinoma, even though these cancers occurred many decades later in life (Cook, 2015).

Also using the CSHRR, Dr. Kitahara and colleagues found that greater BMI and childhood height were positively associated with thyroid cancer risk (Kitahara, 2013). They also found associations between greater birth weight and, in boys, greater height with increased risk of glioma in adulthood (Kitahara, 2014).

Other studies are underway. Katherine A. McGlynn, Ph.D., M.P.H., senior investigator in MEB, is investigating childhood height and future risk for testicular germ cell tumor, and Britton Trabert, Ph.D., M.S., tenure-track investigator in MEB, will be examining birth weight and subsequent ovarian and endometrial cancer risk.

In addition, Drs. Cook and McGlynn are in the process of linking the CSHRR to compulsory military conscription records for men who were drafted between the ages of 18-21. This linkage will enable them to examine not only height and weight in early childhood, but also in early adulthood. The investigators will look at a number of outcomes, including esophageal carcinoma, testicular, and prostate cancer.

“In recent work using other studies, we’ve been able to show that weight change over the adult life course is important for esophageal adenocarcinoma,” Dr. Cook said. “By linking the CSHRR data to military draft records, we hope to delineate between risk associated with childhood and adult BMIs, and to assess the effect on cancer risk of weight change between childhood and adulthood.”

Swedish Registries

A growing body of evidence suggests that testicular germ cell tumors (TGCT) may be influenced by prenatal exposures related to typical genital development. While cryptorchidism is a known risk factor for TGCT, few studies have evaluated the association between other birth defects and risk of TGCT.

Using large, linked population-based registries in Sweden, Drs. Trabert and McGlynn evaluated this important question and demonstrated that hypospadias, inguinal hernia, and other genital malformations are associated with an increased risk of TGCT (Trabert, 2013).

“Previously, it’s been hypothesized that hypospadias could be a part of testicular dysgenesis syndrome,” Dr. McGlynn said. “This analysis provides some of the most definite evidence to date that hypospadias, and perhaps inguinal hernia, should be considered a part of the syndrome.”

Future Directions

While most linked-registry studies have been conducted in the Nordic countries, many U.S. states have used their own data for such research. Dr. Troisi and collaborators at the University of Washington are investigating the possibility of developing a consortium of state birth and cancer registries to provide data on early-life exposures and future cancer in offspring. 

Sharing of data among states would provide the very large numbers of cases necessary to investigate rare tumors, and could address loss to follow-up from out of state migration.

“We think that it would be really worthwhile to do something like this in the U.S.,” Dr. Troisi said. “Our work in the Nordic countries is invaluable, but the population and exposures in those countries are not broadly representative of those in the U.S. Plus, some of the Nordic countries aren’t even as big as our largest states. To do a record-linkage study with even a few states would be a huge accomplishment.”

While investigators continue to conduct big registry-linkage studies looking at early-life epidemiologic factors and cancer risk, there are important next steps to be considered.

“Enormous gaps in knowledge exist, not only in the epidemiology, but in the research efforts to understand biological mechanisms,” Dr. Troisi said. “We hope to expand on that in the future.”

Many early-life factors are determined by a complex interplay of genetic, hormonal, nutritional, and environmental exposures, and the mechanisms that explain the observed associations with cancer have not been established. In addition, it’s not always clear if early-life factors have an independent effect on cancer risk, or if associations are confounded by adult exposures. Additional studies are needed to clarify these questions.

Understanding how cancer develops over the life-course has the potential to transform cancer prevention strategies by targeting interventions to key time periods in early-life when they may be more effective, rather than later in life, as is the current practice. This type of research could also enhance understanding of whether exposure to modifiable factors in early-life affects cancer disparities among racial/ethnic groups.