Crèche and burn?

As we have mentioned before [see C3 :Vol 12 No 2, Vol 12, No 3; Vol 11, No 4], there is great interest in determining whether childhood leukemia might have an infectious etiology, or reflect an unusual or inappropriate immune response to infection. This hypothesis can be studied directly, by searching for physical signs of past infection, or indirectly, by measuring the extent of exposure to situations that modify the likelihood of infection. Proxy measures of infection such as birth order, paternal occupation, and day care may have an advantage over the direct method of study, since evidence of infection can wane over time and it is often difficult to establish that an infection preceded diagnosis of leukemia. The frequency of infection in children has been shown to increase with the length of time spent and the number of other children in attendance at day care. Moreover, aspects of day care attendance, such as whether a child attended and for how long, are easily determinable through interview.

A California case-control study has taken the most detailed look at childhood ALL and day care to date. Ma X, et al. [Br J Cancer 2002; 86: 1419-24] collected  day care history of 140 newly diagnosed cases of childhood leukemia and the same number of birth registry controls matched on date of birth, gender, race, Hispanic ethnicity, and residence.  The main variables of interest were age at which day care was started, the total number of children in attendance, the mean number of children in attendance, months in attendance, hours per week attended, and  “child-hours”, which was the product of the latter three measures. In crude analysis cases started day care significantly later than did controls, while controls had significantly greater months of attendance, hours per week, and number of children in attendance, and child-hours at day care. However, in multivariate analysis, after adjusting for income, only child-hours was significant.  The odds ratio per thousand child-hours was 0.991 (95% CI: 0.984-0.999).  To illustrate their measure of exposure to day care, the authors gave the example of children with 50 thousand child-hours, who could have acquired them by attending daycare with 15 other children for 25 hours per week, for a total of 30.65 months. Such children would have an odds ratio of 0.64 (95% CI: 0.45-0.95) compared to children who did not attend day care. Similar results were found for the 86 cases of ALL aged 2-5 years, who were likely to have had common ALL, while there was not a significant association of day care with the small number of AML cases.

A French case-control study also recently examined childhood leukemia and day care, among other factors. Perrilat et al. [Br J Cancer 2002; 86: 1064-1069] compared the histories of day care attendance, sibship, breast-feeding, and early childhood infections of 280 cases of childhood acute leukemia cases and 288 hospital controls. The odds ratio comparing children who attended day care to those who didn’t was 0.6 (95% CI: 0.4-1.0) adjusting for socioeconomic and other factors. In a multivariate model, children that had both repeated infections before the age two and who attended day care had an odds ratio of 0.24 compared to those who had neither. However, there was neither a significant association of day care attendance without repeated infections nor vice versa.      

COMMENT:  The California study was well designed. The short time interval between case diagnosis and interview of both cases and controls minimized recall bias. Also, the level of detail in the day care exposure assessment surpassed that of previous studies of the topic. Together these attributes lend credence to the study, with the caveat that day care attendance was censored at the case’s date of diagnosis rather than at some earlier date. As such there is the possibility that cases dropped out of day care because of early symptoms of leukemia, which would mean that the difference in attendance was not etiologically relevant. The French study, though it presented less detailed day care exposure assessment, was interesting for having looked at day care and infections simultaneously. The finding that day care did not modify risk except in the presence of repeated infections supports the idea that the former is a good proxy for the latter. On the other hand, it is not clear why repeated infections should not have modified risk apart from day care. One explanation might be that repeated infections of children do not happen often except in day care. Combined with studies of other infection proxies, evidence that childhood leukemia has an infectious origin is growing. 

NQ01-null mice- a potential model for AML

NADPH-quinone oxidoreductase 1 (NQ01) activity is present in all tissues and is induced in response to antioxidants, xenobiotics, heavy metals, UV light, and ionizing radiation. Through coordinated induction of multiple genes, including NQ01, cells are protected from free radical damage. In humans, a polymorphism exists in the NQ01 gene (P187S) that results in the abolishment of NQ01 protein and activity. Approximately 2-4% of individuals possess this homozygous variant of NQ01. Some studies have suggested an association between the homozygous variant allele and an increased risk of adult leukemia, particularly AML [Rothman N, et al Cancer Res 1997; 57:2839-2842, Smith MT et al Blood 2001; 97:1422-1426, Larson R, et al Blood 1999; 94:803-807]. However, the impact of such a deficiency in NQ01 on hematopoiesis is unknown.  In this report, Long DJ et al [Cancer Research 2002; 62:3030-3036] examined the effects of disruption of the NQ01 gene (null phenotype) in mice. Compared to wild-type NQ01 +/+ mice, NQ01-null mice are normal in appearance, show normal development and behavior, and have similar reproductive capacity. However, the null mice show highly significant differences with respect to certain blood parameters. NQ01 -/- mice demonstrate marked myeloid hyperplasia and significant elevation in blood neutrophils, eosinophils, basophils and megakaryocytes. Further, these mice showed decreases in lymphocytes compared to NQ01 +/+ mice.  No structural chromosomal abnormalities were observed in the mice. The authors conclude that NQ01-null mice are preconditioned for leukemia and may provide a useful model in studies of AML.

COMMENT: Developing appropriate animal models for evaluating human exposure-disease relationships is a challenge. Long et al present a convincing argument to explore the use of the NQ01-null mice in studies of human leukemia. In particular, the role of benzene in leukemogenesis using this mouse model would be of interest. As the authors note, bone marrow cells containing the NQ01 homozygous variant appear to be more susceptible to benzene toxicity[Moran JL, et al Proc Natl Acad Sci USA 1999]. 

Another reason to take vitamins?

Few risk factors have been identified for neuroblastoma, although it is one of the most common malignancies in young children. Since maternal vitamin use during pregnancy has been associated with a reduction in neural tube defects and oral clefts, investigators have explored whether use may also reduce the risk of childhood cancers, including neuroblastoma. One population-based case-control study (183 cases, 372 birth certificate controls) in New York State reported that maternal vitamin use during pregnancy was associated with a decreased risk of neuroblastoma (OR=0.5, 95% CI=0.3-0.7) [Michalek AM et al Am J Epidemiol 1996; 143:996-1001]. Vitamin use, however, was ascertained using an open-ended question regarding ‘any other medications’ that were prescribed by a doctor. In the current case-control study, Olshan AF et al [Epidemiology13:575-580, 2002] analyzed data from maternal telephone interviews of 538 children with neuroblastoma and 504 control children obtained through random digit dialing. Self-reported maternal supplemental vitamin use was obtained in the year before and during the pregnancy and included questions regarding brands of vitamins or minerals, reasons for use, whether the supplement was prescribed by a physician, when the supplement was taken (2-12 months prior to pregnancy, month before, and each trimester) and frequency and duration. For the 2-12 month period prior to pregnancy, no association was found with less than daily use of a supplement, while daily or more frequent use decreased risk (OR=0.7, 95% CI=0.4-1.1). Eighty-three per cent of cases and 88% of controls reported any use of a multivitamin from the month before pregnancy through birth suggesting that vitamin supplementation reduced the risk of neuroblastoma (OR=0.6, 95% CI=0.4-1.0). Since many women took vitamin supplements throughout the pregnancy, it was difficult to disentangle the importance of duration and period of use. However, the authors were able to compare women who began daily vitamin use during the second trimester (when she would have known she was pregnant) to non-users and found that use begun this late in pregnancy still reduced risk (OR=0.7; 95% CI=0.4-1.4). The authors found no pronounced modifying (interaction) effects with age at diagnosis, sex, or maternal education. Of the 381 (76%) cases that had MYCN oncogene amplification status available, there was a more pronounced protective effect for the cases with MYCN-unamplified tumors (OR=0.5 for daily vitamin use during second trimester, 95% CI=0.3-0.8) compared to MCYN-amplified tumors (OR=0.8, 95%CI=0.3.-2.4).

COMMENT: In this largest case-control study to date of neuroblastoma, there is evidence that vitamin supplementa-tion during pregnancy may reduce risk in offspring. It was of interest to the authors to determine whether specific vitamins or minerals may be important in reducing risk, although the prevalent use of multivitamins precluded any detailed assessment. In an ad hoc analysis examining vitamin preparations that contained ? 0.4 mg folic acid, they found that daily use during the first trimester had an OR of 0.7 (95% CI=0.5-0.9), which was similar to their overall analysis. Given the recent enrichment of foodstuffs in the US with folic acid, and the notable reduction in neural tube defects (NCHS Vital Statistics Data), it will be of interest to monitor incidence trends in neuroblastoma. 

Briefly noted

Fanconi anemia (FA) is known to be associated with an increased risk of malignancy, both hematopoietic and non-hematopoietic. A new retrospective review of 145 FA patients who had not received a BMT has further quantified the risk [Rosenberg PS, Blood prepublished online, 9/02/02, www.bloodjournal.org]. Among the 145 cases, 9 developed leukemia and 14 developed a total of 18 solid tumors. The observed-to-expected ratios were highly elevated at 785 for leukemia, 4317 for vulvar cancer, 2362 for esophageal cancer and 706 for head and neck cancer. Cause-specific hazards of both death and AML peaked at 1% per year in the teenage years. The hazard of a solid tumor approached 8% per year by age 40. One caution regarding this study- data were collected by questionnaire and less than half of those mailed replied, introducing possible bias. Recent improvements in BMT for this disease may be associated with a further increase in this risk as more patients survive after bone marrow failure. 

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