Something in the air?

 

Administering oxygen to a blue baby is a reflex action for pediatricians and has been seen as a benign and helpful intervention. We should therefore pay particular attention to a new analysis of the classic cohort study, the Collaborative Perinatal Project (CPP), that assesses the effect of perinatal oxygen on risk of childhood cancer [Spector LG et al, J Pediatr 2005; 147:27-31]. The CPP enrolled approximately 48,000 women who had about 60,000 pregnancies at 12 university-affiliated medical centers between 1959 and 1966. Trained observers in delivery rooms recorded the details of 54,795 live births using a standardized methodology. Children received multiple physical examinations in the first year of life and again at 7 years. Parents were interviewed several times in the first 2 years and annually thereafter until their children were aged 7 or 8 years. Fifty-one cancers were observed in the cohort and subsequently confirmed by two pediatricians [Klebanoff MA et al N Engl J Med 1993; 329:905]. Three cancers diagnosed in the first week of life were excluded from the current analysis as the cancer was presumably initiated prenatally. The analysis showed an increased risk of cancer in neonates who received oxygen (Hazard rate (HR) for any oxygen compared to no oxygen 1.77 (95% Confidence Interval (CI)=0.94-3.35), p=0.08). The increased risk was confined to babies receiving oxygen for more than 3 minutes (HR 2.87 (95%CI 1.46-5.66; p=0.002). The frequency and distribution of malignancies reported appears similar to that expected in such a population. Exclusion of children with disorders such as Down syndrome known to increase risk of malignancy did not change the findings.

 

COMMENT: This study presents a carefully performed analysis of an old and valuable dataset. The CPP has already provided analyses on etiology of childhood cancer and the role of vitamin K (none), maternal smoking during pregnancy (none), and simian virus 40 (probably none) [Klebanoff MA et al N Engl J Med 1993;329:905; Klebanoff MA Am J Epi 1996; 144:1028-33; Engels E et al Am J Epi 2004; 164:306-16]. The positive findings of this study are of particular interest because they address an everyday procedure that is assumed to be benign, and there is biological plausibility for oxidative stress in a susceptible child increasing risk of cancer. An illuminating accompanying editorial by Nigel Paneth [J Ped 2005; 147:4-6] points out that neonatologists are already addressing the use of room air instead of oxygen for neonates because of concerns for toxicity of hyperoxia to eyes, lungs and brain. Studies are already available indicating that room air is as effective (or in some studies better) as oxygen for routine resuscitation [Davis PG et al, Lancet 2004; 664:1329]. The study by Spector et al shows that the time might be right for a change in practice, and that old datasets never die.

Stella M. Davies

 

No beans about it: diet and retinoblastoma in Mexico

 

Retinoblastoma, an embryonal tumor of the eye, is the original paradigm for the two-hit model of carcinogenesis [Knudson A. Nature 1986; 323: 643-6] . Mutations of the RB1 gene underlie the disease and may be inherited or occur sporadically, either in gametes or during fetal development. Cases with inherited or sporadic germline mutations typically have bilateral disease; cases that have acquired a somatic mutation in RB1 have unilateral disease. Intake of lutein, folate, and other micronutrients may influence the risk of sporadic retinoblastoma given their importance to retinal development and DNA synthesis. Noting that the Mexican diet tends towards lower fruit and vegetable intake compared to the Unites States, and that Mexican children appear to have an elevated incidence of retinoblastoma, Orjuela and colleagues initiated a case-control study based in Mexico City [Orjuela M et al. Cancer Epi Bio Prev 2005; 14(6): 1433-9] .

 

Children ages 0-6 years with incident diagnoses of retinoblastoma were identified at the Instituto Nacional de Pediatria hospital in Mexico City during 1995-1998. Control children, also ages 0-6 years and with unspecified conditions, were identified at the same institution. Children were excluded from the study if they had a sibling with a malignancy, a recognized genetic syndrome, or a family history of retinoblastoma. Mothers of participating children were interviewed in person regarding demographics, medical history, and diet during pregnancy. The dietary portion of the interview consisted of three open-ended questions in which mothers were asked to list the foods and amounts they typically ate during breakfast, lunch, and dinner while pregnant. Low fruit or vegetable consumption was defined as < 2 servings per day while low bean consumption was < 1 serving per day. Intake of folate, vitamin B6, lutein, alpha-carotene, beta-carotene, lycopene, and beta-cryptozanthine was estimated using nutritional database software adapted for the Mexican diet. Intake of the latter two nutrients was dichotomized into any or no consumption while that of the former nutrients was divided into high or low consumption based on the median value for controls. Logistic regression was used to calculate the odds ratios (OR) and 95% confidence intervals (CI) relating maternal diet during pregnancy to retinoblastoma. Analyses were performed both for all sporadic cases and for unilateral cases alone.

 

Mothers of 101/102 eligible case and 172/175 eligible control children were interviewed. There were 58 cases of unilateral retinoblastoma. Cases were similar to controls with respect to maternal age, morning sickness, vitamin use during pregnancy, alcohol consumption during pregnancy, season of birth, and household income. However, controls were significantly less likely to reside outside of Mexico City (i.e. in rural areas) and their mothers were significantly more likely to have greater than a 9 th grade education. As would be expected, vegetable intake was highly correlated with nutrient intake. For instance, the correlation coefficient (r) was 0.6 relating low vegetable to low folate intake (p < 0.001). There was no significant association of low fruit, bean, lycopene, or beta-cryptozanthine intake with sporadic retinoblastoma, and these variables were not examined further. There were significant associations of retinoblastoma with low intake of vegetables, folate, vitamin B6, and lutein, adjusted for rural residence and maternal education. The respective ORs were 3.4 (95% CI: 2.0-6.0), 4.0 (95% CI: 2.1-7.6), 1.9 (95% CI: 1.1-3.3), and 2.6 (95% CI: 1.5-4.6) for all sporadic cases; for unilateral cases alone the respective ORs were 3.5 (95% CI: 1.8-6.8), 3.2 (95% CI: 1.6-6.7), 1.8 (95% CI: 0.93-3.5), and 2.6 (95% CI: 1.3-5.1). In addition there was a borderline significant association of sporadic retinoblastoma with low alpha-carotene intake (OR = 1.7; 95% CI: 0.97-3.0). Mutual adjustment of vegetable nutrient intake resulted in attenuated, but still significant ORs.

 

COMMENT: This study presents an initial examination of the biologically plausible hypothesis that low intake of nutrients involved in retinal development increases the risk of sporadic retinoblastoma. While the study generally corroborates a previous finding that maternal vitamin supplementation is associated with a reduced risk of retinoblastoma [Bunin G et al. Cancer Res 1989; 49: 5730-5] the study design was limited. In particular, the use of hospital-based cases and controls, though understandable in a developing nation, is less than ideal. One concern is that cases diagnosed at a given hospital derive from a different population, with a different distribution of exposure, than do controls. The reported disparity in maternal education and rural residence suggests this may be the case in this study. On the other hand, the magnitude of the ORs reported for low intake of vegetables and folate were large (for a study of childhood cancer) and thus less likely to be solely the result of selection bias or confounding. These results certainly call for follow-up in a rigorously designed study.

Logan G. Spector

 

Power to push the panic button

 

Considerable controversy exists regarding potential associations between electromagnetic fields [EMFs] and childhood leukemia. A recent meta-analysis of several case-control studies concluded that children living in homes at the highest levels of exposure (> 0.4 µ T) were at a 2-fold increased risk of developing leukemia [Ahlbom A, et al, Br J Cancer 2000; 83:692-698] . However, few individuals would be exposed to these high levels. There also needs to be consideration of timing: would EMF be more important as in utero exposure (thus, possibly acting as an initiator) or after birth (thus possibly acting as a promoter)? Importantly, the biological evidence that EMFs might be related to cancer is inconclusive [Documents of the National Radiological Protection Board 2004; 15 (1)] . Nevertheless, because of public concern and the ubiquity of the exposure, the subject is far from closed. In this most recent study, Draper G et al [Br J Cancer 2005; 330: 1290-1292] mapped the distance between home address at birth for children with cancer compared to control children matched for age, sex, and birth registration data. No interviews were conducted with the families. Children who lived < 200 m from a high voltage power line at birth were 1.69 times (95% CI=1.13-2.53) more likely to develop leukemia compared to children who lived > 600 m. For children who lived between 200 and 600 m the risk was slightly elevated (OR=1.23; 95% CI=1.02-1.49). Interestingly, the authors found no elevated risk or patterns with other childhood cancers including brain tumors. These results persisted even after adjustment for socioecono-mic status, which was measured using a deprivation index created from the census. The authors were circumspect in discussing their findings and noted that the controls could still be insufficiently representative of the underlying population. For example, it is possible that economic or social factors associated with living in close proximity to power lines, which cannot be accounted for by the deprivation adjustment, may help explain these findings. Even if the observation were true, Draper and colleagues note that exposure to high voltage power lines would only account for 5 (1%) of 420 cases of childhood leukemia diagnosed in England and Wales each year. In an accompanying editorial, Watts G [Br J Cancer 2005; 330:1293] notes the lack of biological certainty regarding the dangers of EMF.

 

COMMENT: This mapping study will undoubtedly lead to more confusion, but the authors are to be commended in prudently publishing their findings with appropriate caveats and cautions. We expect to see additional studies in this area. In fact, based on the cell phone/brain tumor controversy, it will not be a surprise if the next round of debate will focus on the new hybrid vehicles, which save gas but can generate high levels of EMF. We already have the title: Baby, you can drive my car!

Julie A. Ross

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