Soy: Dr. Jeckyl or Mr. Hyde?

There is laboratory and epidemiological evidence that consumption of soy products may reduce the risk of many adult malignancies including breast and prostate [Messina MJ, et al. Nutr Cancer 21: 113-131, 1994.]. Further, one of the major phytoestrogens in soy, genistein, is under investigation as a possible chemopreventive agent. However, there are an increasing number of studies that suggest that genistein may have deleterious effects, especially if administered prenatally. One recent study reported an increased risk of mammary tumors in the offspring of rats fed genistein during pregnancy [Hilakivi-Clarke et al, Oncol Rep 6:1089-1095, 1999]. Given the widespread use of soy-based formulas and the high levels of genistein that infants are being exposed to, it is important to delineate the possible deleterious effects of soy during the prenatal and neonatal period. In this study by Newbold et al [Cancer Research 61, 4325-4328, 2001], female mice pups were given daily subcutaneous injections of genistein, diethylstilbestrol (DES) (a known carcinogen), or corn oil (as a control) on days 1-5 of neonatal life. At 18 months of age, the reproductive tracts of the female mice were examined for abnormalities. The authors found that the incidence of uterine carcinoma was 35% for genistein and 31% for DES; no cancer was found in the mice treated with corn oil. They conclude that the potential carcinogenicity of genistein may depend on the timing of administration. They also suggest that the use of soy-based formulas in infants, in the absence of medical necessity, should be questioned more thoroughly. 

COMMENT: While preliminary, this study reminds us that not all exposure-disease relationships are straightforward. Thus, what might be beneficial at one point in time, may be harmful at another. The authors indicate that the level of genistein used in these animal experiments was only one order of magnitude higher than the level consumed in 1 liter of soy-based formula. They plan to investigate physiological levels in their animal model as well as determine if an oral administration of genistein would produce a similar effect.  It should be noted that soy products (including infant formula) contain several other ingredients, and it is difficult to directly extrapolate the conclusions observed here to a real-life setting. However, given that genistein pills are currently being marketed as a natural food supplement, it will be important for studies such as these to further define potential deleterious effects. 

Parents perception of childhood cancer studies

There is a perception among researchers and members of ethics committees that questionnaires administered to parents in the conduct of health studies of serious childhood diseases may have an adverse emotional impact, although few studies have addressed this question. Jenkinson et al [Br J Cancer, 84:413-416, 2001] investigated this issue among 751 parents of children with and without cancer who had participated in the UK Childhood Cancer Study.  In the original study, face-to-face interviews were conducted with both parents of children diagnosed with cancer and healthy control children matched by age and sex. The mother’s interview lasted on average 1 ¼ hours, while the father’s interview lasted about 20 minutes. Questions included reproductive history, family medical history, personal habits (smoking, alcohol consumption), and occupational exposures. Approximately 3-4 weeks after the in-person interview, parents were mailed a 20-minute questionnaire to assess their feelings immediately after the interviews and also 2 weeks later. Included in this assessment were feelings of emotional distress or difficulty with particular questions, and whether the parent felt that they had benefited in any way from having taken part in the study. Of the questionnaires mailed, 371 case parents (64%) and 380 control parents (66%) returned completed questionnaires. Over 94% of case parents, and 90% of control parents were moderately or very interested in taking part in the case-control study. When asked about specific interview topics, 19.5% of case parents and 11.1% of control parents found recalling pregnancy history ‘difficult’; 17.9% and 12.7%, respectively, found this area of questioning upsetting. In contrast, less than 3% from each group found questions regarding smoking difficult or upsetting. At the end of the interview, 95% of cases and 96% of controls were satisfied with their contribution. While 18% of case parents reported feeling tense and 14% reported feeling unhappy immediately after the interview, over 90% of case parents felt glad they had taken part a few weeks later. Moreover, 90% of controls were glad or very glad to have taken part. Finally, 38% of case parents and 24% of control parents reported benefiting from the interview.

COMMENT: Ethics committees are increasingly concerned about the perceived emotional impact of parents taking part in interviews for epidemiological studies of childhood cancer. The protection of human subjects is an essential task of both researchers and ethics committees alike. Thus, it is important to investigate this perceived impact by approaching the individuals who have taken part in these types of studies. Previous studies have shown that interviews may have a beneficial effect as they release some of the stress that cases may be experiencing [Funch and Marshall, Am J Pub Health 71:1396-1398, 1981; Savitz et al, Am J Epidemiol 123:362-366, 1986; Taylor et al, J Epidemiol Comm Hlth 45:317-320, 1991]. Little data, however, exist for parents of children with cancer. The results of this study are reassuring as they suggest that the majority of parents of children with and without cancer are glad to take part in studies that address etiology. It must be noted that there is some concern that less than 70% of parents from each group completed the survey, and it would be important for future studies such as this to determine the reasons for non-participation. However, the results of this study emphasize that a fine balance must be struck on ethics committees between the protection of human subjects and the rights of parents to make the final decision concerning their participation in these types of studies. 

Maybe she was born with it?

One of the most remarkable recent insights into the etiology of childhood leukemia is the observation that somatic mutations which play a role in leukemogenesis are commonly acquired in utero [Gale et al, Proc. Natl. Acad. Sci. 94, 13950–54, 1997]. A recent study by Yoshioka et al [Cancer Research 61, 3432-3438, 2001] extends previously reported observations on mutation frequencies in cord blood samples.  In this study the authors examined cord blood samples from 20 preterm and 33 full-term deliveries.  Mutation frequencies in the reporter gene HPRT were determined, using well described methods.  In a previous report [Yoshioka et al, Proc. Natl. Acad. Sci. 96, 586-591, 1999] the authors demonstrated an increased mutation frequency in preterm infants compared with full-tem newborns.  In addition, mutation frequencies were higher in female compared with male infants.  In this report the spectrum of mutations seen is described.  The data demonstrate an increased frequency of V(D)J recombinase mediated HPRT deletions in preterm compared with term infants and in female compared with male infants.  The preponderance of the deletions in girls is fascinating in light of the epidemiological observation of a higher frequency of infant leukemia in girls compared with boys (female to male ratio 1.3.  Of note, the preponderance of girls is only seen in cases with an MLL gene rearrangement; in infants without this the typical male preponderance is seen (female to male ratio 0.7) [Ross et al, Leuk. Res. 21, 793-5, 1997]. Yoshioka et al speculate that the increased frequency of mutations seen in pre-term infants may reflect the process of T-cell maturation occurring in the infant. 

COMMENT: This paper describes an increased frequency of deletions in female infants.  This is not the typical abnormality associated with childhood or infant leukemia, in which balanced translocations predominate.  However, if these abnormalities are considered as markers of aberrant V(D)J recombinase activity, which has been associated with MLL rearrangements in other studies, the argument is  plausible [Gu et al, Proc. Natl. Acad. Sci., 89, 10464-68,1992].  Most epidemiologic studies of infant leukemia, however, do not show an increased risk in premature infants; in fact, high birth weight has been associated with an increased risk [reviewed in Ross et al, Cancer Causes & Control 7:553-59, 1996]. Clearly additional factors, environmental, genetic or both, influence the outcome of genetic errors generated in utero. 

Crossing the placenta

In utero events are an increasing focus of epidemiological and biological studies of childhood cancer, yet little is known about direct exposure of the fetus to environmental carcinogens (see above).  Whyatt et al [Cancer Epidemiol. Biomarkers Prev. 10, 581-588, 2001] have examined paired maternal and cord blood samples to identify biomarkers of exposure to tobacco smoke and environmental pollutants.  The authors measured aromatic-DNA adduct levels, polyaromatic hydrocarbon (PAH)-DNA adduct levels, and plasma cotinine levels as a measure of cigarette smoking. This study included 160 Polish women and their infants, selected from areas with high air pollution including PAHs.  In 80 mother-child pairs, blood samples were within one hour of delivery. Notably, higher levels of all three biomarkers were seen in infants compared with paired maternal samples. The difference was statistically significant for aromatic-DNA adducts and plasma cotinines, but not for PAH-DNA adducts. When analyses were restricted to the 80 pairs drawn within one hour of delivery, all 3 biomarkers were significantly increased in the infants.  The authors interpret these results as indicating reduced detoxification ability in the fetus and increased susceptibility to DNA damage. The data also reinforce the adverse consequences of smoking on the fetus.

COMMENT: A significant limitation of many epidemiology studies is the absence of any direct measurement of the exposure under investigation.  Additionally, a lack of understanding of the biologic consequences of exposure in different age groups (including the elderly as well as the very young) hinders interpretation of the data.  More studies such as this, directed at measuring exposures or the consequences of exposure directly will benefit all epidemiology studies. 

C3 Quarterly Newsletter
Children's Cancer Research Fund
Epidemiology Research Unit
Division of Pediatric Epidemiology
Clinical Research
University of Minnesota
420 Delaware St. SE, Box 422
Minneapolis, MN 55455
pedsepi@umn.edu

Editors: 
Stella M. Davies, MD, PhD, and Julie A. Ross, PhD