Down syndrome-leukemia and the CBS gene

As we have discussed previously [see C3 Vol 12, no 2; Vol 11, No 4], there is growing interest in determining the role of polymorphisms in childhood malignancy. The cystathione-B synthase (CBS) gene, at chromosome 21 band q22.3, is of interest since it catalyzes the reaction that forms cystathione, an intermediate step in the synthesis of cysteine. This reaction is part of the folate metabolism pathway. Individuals deficient in CBS enzymatic activity are characterized by mental retardation, elevated homocysteine levels, and skeletal abnormalities, and are at risk of developing blood clots and atherosclerosis. A recent study suggested a relationship between CBS activity, ARA-C treatment, and high event-free survival in children with Down syndrome and acute myeloid leukemia (AML) [Taub J et al Blood 1999; 94:1393-1400]. A polymorphism in the CBS gene results in a 68bp insertion, (844ins68)  and is found in approximately 12% of non-white Hispanics. In this study, Ge Y, et al [Leukemia 2002; 16:2339-2341] investigated the frequency of  CBS polymorphism in 19 children with Down syndrome and AML, 44 children with AML (no Down syndrome), 54 children with Down syndrome (no leukemia) and 20 healthy individuals. Moreover, they attempted to evaluate the functional significance of this polymorphism by treating isolated myeloblasts from the patient groups with ARA-C to evaluate effects. The investigators found a statistically significant higher frequency of the 68bp insertion in children with DS-AML compared to children with AML alone, as well as healthy controls (52.6% versus 11.4% and 10.0%, respectively, p< 0.001). Interestingly, they also found a higher frequency in DS-leukemia cases compared to healthy children with Down syndrome, although the range of the difference was narrower (52.6 % versus 24.1%, p=0.04). Using RT-PCR, the authors did not find any difference in the transcriptional product between the wildtype and 844ins68, suggesting that the variant did not alter CBS transcripts. Finally, in an experiment to explore the functional significance of this polymorphism, the investigators exposed isolated myeloblasts from patients to ARA-C. For patients with the 844ins68, there was nearly a 7-fold higher sensitivity to the drug compared to patients with the wildtype CBS allele. 

COMMENT: While the investigators acknowledge the small sample size in their study, the results do provide evidence that genes involved in folate metabolism may be important in the etiology of Down syndrome-leukemia. Folate metabolism genes have been implicated in Down syndrome [Hobbs CA, et al: Am J Hum Genet 2000; 67:623-30]  as well as leukemia [Wiemels J et al PNAS 2001; 98:4004-9]. There are few clues as to why children with Down syndrome are at such an increased risk of developing leukemia. It is of interest in this study that the frequency of the 844ins68 polymorphism in healthy children with Down syndrome was higher than the frequency in healthy controls as well as other children with AML. We know that the higher risk of leukemia in children with Down syndrome persists into early adulthood, so it is possible that Down syndrome individuals with the CBS polymorphism may be at higher risk. Obviously, this calls for larger studies. 

Millions of people received doses of poliovirus vaccine from early batches that were inadvertently contaminated with simian virus 40 (SV40).This exposure has caused  concern because SV40 encodes for T-antigen, which is an inhibitor of tumor-suppressing proteins [Barbanti-Brodano G et al. Adv. Virus Res. 1998; 50: 69-99]. SV40 DNA or antigen has been detected in a variety of tumors and, SV40 very efficiently induces brain tumors in test animals inoculated at birth [Kirschstein R. and Gerber P. Nature 1962; 195: 299-300]. However, it is very difficult to establish that SV40 infection preceded disease in case-control studies. Therefore, researchers have taken advantage of the natural experiment presented by SV40-contaminated polio-virus vaccine. 

A recent study  [Engels E.A. et al. J Natl Cancer Inst 95(7): 532-539] examined cancer incidence in three cohorts of Danish children: those born in 1946-1952, 1955-1961, and 1964-1970. The first cohort received contaminated vaccine at four years of age or older, the second as infants at about nine months of age, and the third cohort received uncontaminated vaccine.  Poisson regression was used to compare the rates of cancer in each of the cohorts, as reported by the Danish cancer registry.  Comparing both SV40-exposed cohorts to the unexposed cohort, the age-adjusted rate ratio for incidence of all cancers diagnosed up to 1997 was 0.82 (95% CI: 0.78-0.88). They also looked at the incidence of all cancers diagnosed between 0 and 4 years of age in the SV40-exposed cohort and that in the cohorts that were not exposed to SV40 by four years of age (i.e. both the 1946-1952 cohort and the post-1961 cohort) There was no difference in incidence for all cancers (RR: 1.01; 95% CI: 0.86-1.18) but there was a significant excess of ependymomas (RR: 2.59; 95% CI: 1.36-4.92) in exposed children. However, the rate of ependymoma in the exposed cohort was not significantly different from that in the post-1961 unexposed cohort alone (RR: 0.94; 95% CI: 0.59-1.49) Moreover, the researchers fit curves to describe single-year incidence of ependymoma and the best fitting curve suggested that incidence peaked in 1969, long after any child aged 0-4 years would have received SV40-contaminated vaccine.

COMMENT: This study had the advantages of high-quality cancer incidence data and well-defined exposure. The authors confirmed that nearly all poliovirus vaccine used in Denmark between 1955 and 1962 contained SV40, whereas only a portion of vaccine used in other countries was contaminated. Thus, the study credibly confirmed previous findings that SV40 exposure is not associated with cancer in childhood or at older ages. Evidence that SV40 continued to circulate in the population after poliovirus vaccine was cleared of SV40 is limited and contradictory, but even if there was continued SV40 transmission, as the authors pointed out, infection, “would likely have been less frequent, have occurred at older ages, and have arisen from smaller inocula of virus.”  The finding that the incidence of ependymoma was higher among SV40-exposed children was interesting but likely does not indicate causation, since incidence did not decline after exposure to SV40 was considerably lessened, if not eliminated.  Evidence that SV40 causes cancer in humans remains lacking despite suggestive laboratory findings.

NQ01 genotype, MLL+ and MLL- leukemias

A single-nucleotide polymorphism (SNP) of interest in childhood leukemia is that found in the NAD(P)H: quinone oxidoreductase 1 (NQ01) gene. NQ01 protects cells from oxidative stress and toxic quinones such as those found in benzene. A polymorphism at position 609 (C609T) is known to inactivate  the enzyme. As reported previously [Wiemels J et al Cancer Res 1999;59:4095-4099] in a study of 36 cases, children (primarily infants) with MLL translocations in their leukemia cells had a much higher frequency of the inactivating polymorphism  than healthy individuals or non-MLL gene-rearranged B-lineage ALL cases. In a new study by Smith MT et al [Blood 2002; 100:4590-4593], comparisons were made between 39 patients with de novo leukemias (32 infants) characterized by MLL gene translocations, 18 patients with treatment-related MLL leukemias, and 56 patients with de novo B lineage leukemias without MLL gene rearrangements. The investigators found that patients with de novo MLL translocation leukemias were significantly more likely to be heterozygous (lower activity) or homozygous (no activity) at NQ01 C609T compared to patients with non-MLL leukemia (odds ratio (OR)=2.47, 95% CI=1.08-5.68). These results were similar when the MLL-rearranged group was compared to ethnically-matched healthy individuals. As with the Wiemels study, this association was strongest for de novo leukemias characterized by the t(4;11) translocation. Interestingly, there was no difference in genotype frequencies between the MLL-rearranged treatment-related leukemias and the comparison groups, suggesting  possible  important differences in etiological events in de novo MLL infant and childhood leukemia and therapy-related cases. 

COMMENT: This study provides further evidence that the NQ01 gene may be important in the risk of MLL-defined infant leukemia. It will also be important to evaluate gene-environment interactions. For example, does NQ01 genotype interact with reported exposures to solvents? The power of such a study is enhanced by the ability to perform case-case analyses (i.e., MLL + versus MLL-). This approach diminishes the likelihood of recall bias as a mother of child with MLL positive leukemia is unlikely to recall her exposures differently than a mother of a child with MLL negative leukemia. The Children’s Oncology Group has proposed to expand the current case-control study of infant leukemia and increase the number of cases enrolled in order to evaluate these and other potentially important relationships.

Breast feeding and childhood cancer: no association?

A number of recent case-control studies have suggested that breastfeeding may be protect against the development of a number of childhood cancers. Lancashire RJ and Sorahan TJ [Br J Cancer 2003; 88:1036-7] present results from an analysis of the Oxford Survey of Childhood Cancers (OSCC). Briefly, the OSCC began in Oxford in 1955. Interview data were collected from parents (mostly mothers) of all children who died of cancer before their 16th birthday in England, Wales, and Scotland during the period 1953-1984. Several questionnaires were used during this long study and information on breastfeeding during infancy was collected for children who died during the period 1972-1981 and their matched controls. Of the 7590 deaths during this time period, interview data were obtained from 4288 (56%) of cases. Interview data were also obtained from parents of 3827 control children matched on sex and date of birth (461 case children did not have a suitably matched control and therefore were not included in the analysis). Controls were selected from birth registries; after refusals, etc. the investigators reported an overall 54% response rate. After excluding adoptions, children who died in the first year of life, or who demonstrated symptoms in the first nine months of life, 3376 matched pairs remained. Information collected on breastfeeding included ever/never and duration. The authors found no protective effect for reported breastfeeding for childhood cancer overall (OR=1.04, 95% CI=0.93-1.15) nor for ALL or other leukemias. There were also no trends observed with duration of breastfeeding. The authors acknowledge the possibility of bias in this analysis. Participation rates were low, and the deaths of children may have influenced the mother’s recall of breastfeeding.

COMMENT: It is important to put this study in the right context with respect to other recent studies. The low participation rate of case and controls parents is of concern. It is also difficult to evaluate this study in relationship to breastfeeding practices in today’s society. Thus, the results of this study cannot provide a convincing argument that breastfeeding is not protective against childhood cancer.

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