Air pollution and childhood cancer: Seeing the forest from the trees

 

In this report, Knox EG [Epidemiol Comm Health 2005; 59:101-105] used geographical displays of emissions of different chemicals published in 2001 by the United Kingdom National Atmospheric Emissions Inventory and mapped them to birth addresses of children who had died of leukemia or other cancers between 1953 and 1980 before the age of 16 years. To reduce the potential bias associated with the discrepancy in years between exposure measurement and cancer death, the author subsequently restricted to cases who died in the second half of the study period (1966-1980). Estimates of relative risks were based on being born close to or distant from an emission hazard, and were first calculated among children who had moved more than 1.0 km between birth and death. It was postulated that because many childhood cancers are initiated in utero, children who were born in close proximity to an emission site would be at a higher risk of developing cancer, regardless of subsequent movement out of the area. Further, the ratio of children born near an emission site and subsequently moved was also calculated for each type of emission including carbon dioxide, sulphur dioxide, benzene, carbon monoxide, dioxins, specific metals, and other chemicals. Overall, Knox reported an elevated ratio of outward to inward migration for emission plants producing combustion products and volatile organic compounds. Relative risks were calculated for specific chemicals for children born near high emission zones and suggested elevated RR for 1,3-butadiene, dioxins, benzene, and other chemicals. In contrast, there were a higher number of cases who died in close proximity to emission plants that produced high levels of arsenic, nickel and vanadium. Knox acknowledges limitations to this study design including the time interval between cancer records and emissions data and difficulty with assuming the premise of migration equilibrium. Nevertheless, particularly for 1,3-butadiene, he calls for no further delays in controlling emissions based on ‘the massive evidence' reported here.

 

COMMENT: This type of study is referred to as an ‘ecologic study'. In typical ecologic studies, two groups of data are compared to look for associations. For example, one might compare per capita cell phone usage in the United States to brain cancer incidence over time and find that there is a strong positive correlation (e.g., as cell phone usage increased during the 1990's so did brain cancer incidence). This does not mean, however, that cell phone use causes brain cancer. One could also find that per capita use of VCRs, microwave ovens, etc, also are positively correlated with brain cancer incidence. Ecologic studies can only generate hypotheses. Only with further rigorous study, which includes the collection of individual data, can hypotheses be tested. In the Knox study, one of the major limitations is the use of emissions data collected in 2001 correlated to cancer deaths that occurred at least 21 years ago. Knox argues that since there were only improvements in emissions standards over time, it is likely that most of the factories still existing in 2001 would be at least as active when the children were born. The problem is that we simply do not know how many of these plants were active (or even in existence) that many years ago. Further, since there are no individual data collected (only addresses), it is impossible to quantify exposure or consider potential confounders. Also, this study presumes that the relevant time of exposure for all childhood cancers is the perinatal period. While this may be true for some childhood cancers, it is unlikely to be true for all. Professor Knox is to be commended on the statistical methodology he has developed over the years to investigate cancer clusters. However, because of the major limitations in interpreting these type of data, it can only be said that they are of interest. This study cannot be interpreted to indicate that certain chemicals are likely responsible for several childhood cancers.

Julie A. Ross

 

Fetal hormones and stem cells- it's got potential!

 

We have reported numerous associations between high birth weight and childhood cancer [See C3 Vol 15, No 6; Vol 9, No 2; Vol 10, No 3] . Most recently, we called for studies that attempt to understand the underlying mechanism behind this association. In this report, Baik I et al [Cancer Res 2005; 65:358-363] examined whether stem cell potential as measured by the population size of hematopoietic stem cells [CD34+, CD34+CD38-, and colony-forming unit, CF-GM] in cord blood correlated with levels of selected hormones including insulin-like growth factor-1 (IGF1), IGF binding protein 3 (IGFBP3), estradiol, unconjugated estriol, testosterone, progesterone, pro-lactin, and sex hormone binding globulin (SHBG). Cord blood samples from singleton births of at least 37 weeks were obtained from 40 women (ages 18-39 years). Flow cytometry was used to separate cells and CFU-GM colonies were assayed. Measures of stem cell potential were evaluated as CD34+, CD34+CD38-, and CF-GM counts per 1000 mononuclear cells. The authors report that among the hormones, estradiol, estriol, testosterone, IGF-1, and IGFBP3 were positively associated with stem cell potential. After adjustment for maternal and neonatal characteristics, including mother's age, race of parents, number of live births, gestational length, sex of infant, delivery time, and infant birth weight, only IGF-1 and IGF-BP3 were significantly and positively correlated with all three measures of stem cell potential. For example, a one standard deviation increase in IGF-1 levels was associated with a 40.9% increase (95% Confidence Interval (CI)=10.0, 80.5) in CD34+ cells. In contrast, there was no significant change in CD34 associated with a 1 SD increase in progesterone levels (7.4% increase, 95% CI= -14.2, 34.3). The authors suggest that growth hormones during the perinatal period might increase the number of stem cells and thus increase the total number of replicating immature cells that might be susceptible to malignant transformation.

 

COMMENT: This study is an excellent example of how we might unlock the biological basis of some of the findings from case-control observations in childhood cancer. As the authors note, there is evidence that the size of the stem cell pool, which is largely determine during fetal development, is associated with cancer risk [Albanes D & Winick M. JNCI 1988; 80:772-774; Moolgavkar SH et al JNCI 1980; 65:559-569] . One limitation of this study is the cross-sectional design. It is unknown whether a larger number of stem cells are producing IGF1 or IGF1 is stimulating the population of stem cells. The authors point to experimental evidence that stem cells proliferate in response to IGF1 [Frostad S, et al. Stem Cells 1998; 16:334-342 ]. Neverthe-less, this question cannot be answered in this type of study design. Because this study is limited by sample numbers, it will be of interest to see it replicated. Additionally, animal studies may help provide answers to the question of a temporal relationship.

Julie A. Ross

 

OTX2 in medulloblastoma: brain science at its best

 

Medulloblastoma is the most frequent malignant brain tumor in children and nearly one third of children die within 5 years of diagnosis. Di C et al [Cancer Res 2005; 65:919-924] identified an oncogene that is common in a subset of childhood medulloblastoma that appears to be amenable to therapy with all-trans-retinoic acid. Using digital karyotyping that looks for subchromosomal regions of amplifications or deletions, the investigators initially examined a medulloblastoma cell line for common deletions or amplifications. Along with the previously identified amplification of the C-Myc oncogene on chromosome 8q24.21, loss of chromosome 17p, and gain of chromosome 17q, they discovered amplification of a region on chromosome 14q22.3, which contains the genes OTX2 and C14orf101. Further investigation demonstrated that the OTX2 gene, which is important in normal cerebral development, was the gene amplified in a number of different medulloblastoma cell lines as well as in several clinical medulloblastoma samples. This OTX2 amplification appeared unique to medulloblastoma since it was not observed in other cell lines, including glioblastoma multiforme tumors. Increased OTX2 expression was observed in 21 of 33 medulloblastoma samples tested. Further, anaplastic medulloblastomas, which are characterized by clinical aggressiveness, were more likely to have increased OTX2 expression than classical medulloblastoma. Finally, since exogenous retinoids have been shown to repress OTX2, the authors examined the influence of the retinoid, ATRA, on cell proliferation and apoptosis in cells that expressed the OTX2 transcript versus those that did not. They found the ATRA inhibited the growth of seven cell lines that expressed OTX2, but had little effect on four lines that had absent or minimal OTX2 expression. Di et al suggest that these results provide the conceptual framework for clinical studies using retinoids in the treatment of pediatric medulloblastomas.

 

COMMENT: This is an elegant study. As Di et al suggest, the characterization of unique subgroups of rare tumors by both clinical and molecular presentation may assist in the treatment of these devastating diseases. It is possible that classification of this rare malignancy by OTX2 expression might also assist in understanding the causes of certain pediatric medulloblastomas. The value of molecular characterization has been shown previously in studies of infant leukemia, where exposures have been examined with respect to the presence or absence of an MLL abnormality in the leukemia cells. [Alexander FE, et al. Cancer Res 2001; 61:2542-2546; Spector LG et al, Cancer Epi Biom Prev; in press].

Julie A. Ross

 

When is a Spitz not a Spitzoid melanoma?

 

Melanoma in childhood is extremely rare, but there are reports that it has been increasing in incidence. A distinct type of melanoma in children is Spitzoid melanoma, which is clinically characterized by solitary papules that may or may not exhibit pigmentation. Importantly, Spitz nevi, which are found at all ages, are benign melanocytic neoplasms that are difficult to distinguish from Spitzoid melanoma. Not surprisingly, Spitzoid melanomas have been misdiagnosed as nevi, often leading to fatal consequences. Approximately 90% of melanomas are sporadic, and reports suggest that a high proportion of these melanomas have mutations in the B-RAF gene, suggesting that B-RAF may play a role in the early phase of melanoma. In this report. Gill M et al [Cancer 2004; 101:2636-2640] wished to determine whether B-RAF mutations, as well as mutations in N-RAS and K-RAS might help distinguish a benign spitz nevus from a Spitzoid melanoma. Malignant Spitzoid melanoma specimens were obtained from 9 children under the age of 10 years and compared to 10 typical Spitz nevus specimens (also from children under the age of 10 years). No activating mutations in the B-RAF, N-RAS, or HRAS genes were found in either group of tumor specimens. A number of silent mutations were noted in the H-RAS gene, which were present in both melanoma specimens as well as benign nevi. They conclude that Spitz nevi as well as Spitzoid melanoma are genetically distinct from other melanomas and melanocytic nevi. They further conclude that other unidentified genes may be important in the progression of Spitz nevi to Spitzoid melanoma.

 

COMMENT: While negative, this observation is important in our understanding of the genetics of rare tumors. Some activating mutations found in non-Spitzoid melanoma are also present in benign melanocytic nevi, suggesting an early event in the carcinogenesis process. A next step might be to evaluate the specimens in this report using array techniques to determine whether specific gene expression differences help distinguish a benign lesion from a malignancy.

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