Page 396 - 2014 Printable Abstract Book
P. 396
(SND03) Potential confounding in studies of radiation exposure from pediatric CT scans and cancer.
Michael Hauptmann, Netherlands Cancer Institute, Amsterdam, Netherlands
Several studies on CT-related radiation exposure among children and cancer risk have been
published (Pearce et al., Lancet 2012; Mathews et al., BMJ 2013; Huang et al., Br J Cancer 2014) or are
currently underway (http://epi-ct.iarc.fr). The Dutch Pediatric CT Study (Meulepas et al., Eur J Epidemiol
2014) has so far recruited over 100,000 children nation-wide who received over 200,000 CT scans since
1983. Analyses of radiation-related cancer risk will be based on linkage with the Dutch Cancer Registry
(complete since 1989) and the Dutch Childhood Oncology Group (complete for childhood leukemia since
1973). To reach the large size required for evaluating the small risks from low-dose radiation exposure,
the Dutch Pediatric CT Study as well as other studies are commonly based on record linkage with limited
subject-specific information on potential confounders. Whereas confounding by lifestyle factors is unlikely
to be important in studies of children and young adults, medical reasons for obtaining a CT scan have been
raised as potentially strong confounders (NCRP report no. 17, 2012; Walsh et al., J Radiol Prot 2014).
Examples are cancer susceptibility syndromes, such as Down syndrome (Hauptmann and Meulepas,
Lancet 2012), and reverse causation by indolent cancers causing symptoms for which a CT is done long
before diagnosis (Shuryak et al., Radiat Res 2014). We designed a theoretical study to evaluate the
likelihood and degree of bias for various scenarios. Since empirical data on the link between potential
confounders and CT scans as well as between potential confounders and cancer risk are often scarce, we
include expert assessment for the identification of realistic scenarios. We provide qualitative and
quantitative analyses of the potential magnitude of such confounding, with focus on cancer susceptibility
syndromes and risks for leukemia as well as brain tumors. The results of this work may provide a starting
point for assessing the likelihood of confounding in current and future record linkage studies on cancer
risk from CT scans. The framework can be updated as new data become available on the likelihood of CT
scans and cancer risk associated with certain scenarios.
SNE. GENETIC ISSUES IN RADIATION EPIDEMIOLOGY
The understanding that radiation induces heritable genetic changes dates back to work by
Hermann Muller in the 1920s. Among the first issues of scientific study in the aftermath of the A-bombings
in Japan in 1945 was whether birth defects were increased in children of mothers exposed to the radiation
from the bombings. In more recent years this has been extended to accessing whether offspring of
exposed parents are at higher risk of disease. Different yet related questions are whether some individuals
are more susceptible to the effects of radiation on cancer or other disease risk because of their genetic
makeup and whether somatic changes in the cancer genome can explain or predict tumor etiology, tumor
aggressiveness, or response to therapy. In recent years the revolutions in genotyping and sequencing
technology have developed extremely powerful tools to address these important questions. Genome-
wide association studies can search the genome for common variants involved in cancer risk and in gene
x environment interactions, and modern high throughput sequencing techniques can comprehensively
search for new changes either in tumor tissue compared to germ line DNA, or in germ line DNA of offspring
compared to parents. This session will tentatively focus on (1) the search, using genome-wide genotyping
arrays in exposed populations, for specific sequence variants that may be involved in gene × radiation
Michael Hauptmann, Netherlands Cancer Institute, Amsterdam, Netherlands
Several studies on CT-related radiation exposure among children and cancer risk have been
published (Pearce et al., Lancet 2012; Mathews et al., BMJ 2013; Huang et al., Br J Cancer 2014) or are
currently underway (http://epi-ct.iarc.fr). The Dutch Pediatric CT Study (Meulepas et al., Eur J Epidemiol
2014) has so far recruited over 100,000 children nation-wide who received over 200,000 CT scans since
1983. Analyses of radiation-related cancer risk will be based on linkage with the Dutch Cancer Registry
(complete since 1989) and the Dutch Childhood Oncology Group (complete for childhood leukemia since
1973). To reach the large size required for evaluating the small risks from low-dose radiation exposure,
the Dutch Pediatric CT Study as well as other studies are commonly based on record linkage with limited
subject-specific information on potential confounders. Whereas confounding by lifestyle factors is unlikely
to be important in studies of children and young adults, medical reasons for obtaining a CT scan have been
raised as potentially strong confounders (NCRP report no. 17, 2012; Walsh et al., J Radiol Prot 2014).
Examples are cancer susceptibility syndromes, such as Down syndrome (Hauptmann and Meulepas,
Lancet 2012), and reverse causation by indolent cancers causing symptoms for which a CT is done long
before diagnosis (Shuryak et al., Radiat Res 2014). We designed a theoretical study to evaluate the
likelihood and degree of bias for various scenarios. Since empirical data on the link between potential
confounders and CT scans as well as between potential confounders and cancer risk are often scarce, we
include expert assessment for the identification of realistic scenarios. We provide qualitative and
quantitative analyses of the potential magnitude of such confounding, with focus on cancer susceptibility
syndromes and risks for leukemia as well as brain tumors. The results of this work may provide a starting
point for assessing the likelihood of confounding in current and future record linkage studies on cancer
risk from CT scans. The framework can be updated as new data become available on the likelihood of CT
scans and cancer risk associated with certain scenarios.
SNE. GENETIC ISSUES IN RADIATION EPIDEMIOLOGY
The understanding that radiation induces heritable genetic changes dates back to work by
Hermann Muller in the 1920s. Among the first issues of scientific study in the aftermath of the A-bombings
in Japan in 1945 was whether birth defects were increased in children of mothers exposed to the radiation
from the bombings. In more recent years this has been extended to accessing whether offspring of
exposed parents are at higher risk of disease. Different yet related questions are whether some individuals
are more susceptible to the effects of radiation on cancer or other disease risk because of their genetic
makeup and whether somatic changes in the cancer genome can explain or predict tumor etiology, tumor
aggressiveness, or response to therapy. In recent years the revolutions in genotyping and sequencing
technology have developed extremely powerful tools to address these important questions. Genome-
wide association studies can search the genome for common variants involved in cancer risk and in gene
x environment interactions, and modern high throughput sequencing techniques can comprehensively
search for new changes either in tumor tissue compared to germ line DNA, or in germ line DNA of offspring
compared to parents. This session will tentatively focus on (1) the search, using genome-wide genotyping
arrays in exposed populations, for specific sequence variants that may be involved in gene × radiation
