Familial brain tumors

Taylor et al. have described a new genetic mutation associated with posterior fossa brain tumors in infancy [J Hum Genet 66:1403-6, 2000]. In this family, the proband presented at age 18 months with a cerebellar malignant rhabdoid tumor. Both of the parents are healthy, but the proband’s maternal uncle died at age 2 years from a posterior fossa choroid plexus carcinoma. In addition, the maternal grandfather’s sibling died as an infant from a disease process consistent with malignant brain tumor. Malignant rhabdoid tumors of the CNS and of the kidney have been shown to have loss of heterozygosity at chromosome 22q11. The hSNF5 gene on chromosome 22q11 has recently been identified as a candidate tumor suppressor gene in sporadic CNS and renal malignant rhabdoid tumors. In light of this, the authors of this study sequenced the gene in this multiply affected family. They showed a germline splice site mutation leading to exclusion of exon 7 from mature cDNA and subsequent frameshift in hSNF5. Tumor tissue showed loss of the wild type hSNF5 allele in keeping with a tumor suppressor gene. The authors suggest that germline mutations in hSNF5 cause a novel autosomal dominant syndrome with incomplete penetrance that predisposes to malignant posterior fossa brain tumors in infancy. The proband’s mother carries the mutation but is unaffected, hence the incomplete penetrance. It seems likely that she will remain unaffected as both rhabdoid tumors and choroid plexus carcinomas are uncommon in the adult population. However, it remains possible that she is at continuing risk. 

COMMENT: This interesting family confirms the heritability of hSNF5 mutations. Previous studies have identified truncating mutations of the hSNF5 gene in sporadic renal and CNS malignant rhabdoid tumors [Versteege et al., Nature 294:203-6, 1998 (reviewed in C3 vol. 9, no. 4, 1998); Biegel et al., Cancer Res 59:74-9, 1999]. In addition, in these studies, several infants were noted to have germline mutations as well as somatic mutations in the hSNF5 gene, but in no case could mutation of hSNF5 be found in the parents of affected individuals nor were there signs of affected individuals in other generations. Identification of this gene mutation should lead to the study of further families to confirm the pattern of heritability.  Stella M. Davies

Where’s the bacon? 

A number of epidemiological studies have explored potential associations between consumption of cured meats and the risk of childhood brain cancer, as well as some other childhood malignancies. Blot et al [Nutrition and Cancer 34:111-18] were commis-sioned by an organization representing concerned meat producers to critically review the scientific evidence. Their review included 14 published studies that explored associations between cured meat consumption (mother, father, child) and the development of childhood brain tumors (ten studies) or other malignancies (4 studies). Thirteen of these studies used a case-control approach, while one was an ecological study. Blot et al present results by summation of the evidence across studies including a) total cured meat intake and whether a dose-response was observed, b) observations by individual meats consumed,  and c) the amount of confounding that was adjusted for (e.g., was there an adjustment for fruit and vegetable intake, as this might reduce risk). Overall, significant trends associated with total cured meat consumption (mostly maternal during pregnancy) were observed in 4 of the 10 brain cancer studies. When individual meats were examined (nine studies), there was little consistency across studies. However, several studies reported elevated risks with hot dog consumption, although some of these studies were based on extremely small numbers in the subgroups noted. Socioeconomic status (SES) was both positively and inversely associated with cured meat intake; one study found no effect with SES. In the ecological analysis that examined the per capita consumption of cured meats in the years 1977-78 and 1989-91, a 17% decrease was noted. During this same time period, there had been a 20% increase in the incidence of childhood brain tumors (which is likely due in part to improved detection; See C3 Vol 9 No 5, 1998). Although Blot et al recognize the limitations of interpreting ecological studies, they suggest that the lack of concordance between the two variables is inconsistent with the hypothesis that cured meat intake increases brain tumor occurrence. The authors also review the evidence supporting biological plausibility, since a major finding in favor of a causal association comes from animal studies which demonstrate that in utero exposure to nitrosoureas causes brain tumors in offspring. However, these animal data do not extend to the nitrosoamines, the other class of N-nitroso compounds. Moreover, while it is possible that nitrosoureas can form endogenously in the human stomach through metabolism of nitrosourea precursors, it is not known if they can reach the brain tissue in a human embryo. The authors conclude that a causal association between cured meat consumption and childhood cancers cannot be assessed based on the available evidence---nor do they claim it can be ruled out.  Given the difficulties with assessing diet through a case-control study design, they propose that a cohort study of women of child-bearing age (which would have to be extremely large) may best address this question. The cured meat hypothesis has been extended to include adult brain cancers [Blowers et al, Cancer Causes Control, 8:5-12, 1997], which Blot et al note might be more feasibly addressed in a cohort study.

COMMENT: This was a comprehensive review of the available evidence regarding cured meat consumption and childhood malignancies. The authors reached the appropriate conclusion, although given their detailed assessment of each individual component of causality (dose-response, biological plausibility, consistency across studies, strength of the association, etc), one might conclude that it is unlikely that cured meat consumption is associated with childhood malignancies. Julie A. Ross

Autoimmune disease and childhood cancer

A record linkage study has been conducted in Denmark in order to explore the potential relationship between parental autoimmune diseases and childhood cancer in offspring [Mellenkjaer et al, BJC 2000; 62:1353-1367]. In this analysis, data from the Danish National Registry of Patients from 1977-1993 were combined with data from the Danish Central Population Register (CPR) and the Danish Cancer Registry. A total of 36,654 individuals with auto-immune diseases were identified thorough the Registry of Patients, which includes information on non-psychiatric hospitalizations. Of the 56,052 children of these patients identified through the CPR, 115 developed childhood cancer. The standardized incidence ratio (SIR) of childhood cancer for the children of parents with autoimmune diseases was 1.07 (95% CI 0.89-1.29). The authors observed slightly increased risks of leukemia (SIR 3.1, 95%CI 0.8-7.9) and lymphoma (SIR 1.6 95%CI 1.0-2.4). None of the SIRs were statistically significant. The authors conclude that children of parents with autoimmune diseases are slightly more susceptible to childhood lymphoma and leukemia than children in general. They recommend that further studies need to be conducted to evaluate this association.

COMMENT: The impetus for this study was the hypothesis that both leukemia and autoimmune diseases are the result of an ‘overly florid’ immune response. Contrary to the authors conclusions, parental autoimmune disease does not appear to increase cancer risk in offspring.  In fact, their study provides compelling evidence to the contrary.  None of the SIRs were statistically significant. Additionally,  this study included only children of parents who had been hospitalized for their autoimmune diseases. If an association existed, one would expect to see it in this severely affected  population. Further, 188 children were identified with an autoimmune condition yet none of them had a diagnosis of cancer during childhood. If similar immune responses resulted in both autoimmune diseases and childhood cancer, higher rates of cancer should be observed in children with the condition. Record linkage studies can be a very powerful and inexpensive way to investigate an epidemiologic hypothesis.  The results of this study suggest that parental autoimmune disease is not associated with an increased risk of childhood cancer. 

Briefly noted: Imprinting of IGF2 in neuroblastoma and infant leukemia

Genomic imprinting, or the differential expression of a gene depending on the parent of origin, has been explored in several pediatric malignancies. In particular, loss of imprinting (or biallelic expression) of IGF2 has been reported in Wilms’ tumor, and it has been suggested that this LOI may contribute to the somatic overgrowth sometimes seen in these patients [Ogawa et al, Nature 362:749-51, 1993]. In this report, Hattori et al [Cancer 88:2372-2377, 2000] examined imprinting patterns of IGF2 in patients with infant leukemia (46 cases) and neuroblastoma (59 cases).  Of these patients, 22 of the 46 leukemia specimens, and 16 of the neuroblastoma samples were both informative for the Apa1 polymorphism and had sufficient cDNA for further analysis of IGF2 expression. Two of the 22 infant leukemia samples demonstrated biallelic expression of IGF2, while none of the neuroblastoma specimens showed loss of imprinting. The authors conclude that loss of imprinting at IGF2 does not play a major role in carcinogenesis at these sites. 

COMMENT: The significance of loss of imprinting of IGF2 in human cancer is unclear. Further functional studies are needed to determine whether biallelic expression of IGF2 (and other genes) plays a role in the initiation or progression of neoplastic growth in specific human malignancies. 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