The eyes have it!

A high frequency of second malignancies in patients with bilateral retinoblastoma has been reported for many years. The majority of these malignancies have been sarcomas. A brief paper [Cancer 1998; 83:767-771] by Howrey et al. describes sebaceous gland carcinoma occurring in patients who have received radiation therapy for bilateral retinoblastoma. These lesions commonly present as a small, firm, slowly growing nodule on the upper eyelid and clinically may appear relatively innocuous in the early stages; misdiagnosis is frequent. The authors cite a mean of 2.2 incorrect diagnoses for patients prior to the correct diagnosis of sebaceous gland carcinoma. The mean interval between onset of symptoms and initiation of treatment ranges from 15.4 months to 3 years. These lesions have the potential for metastasis, generally to pre-auricular, parotid, or cervical lymph nodes. Liver, lung, brain and bone metastases have been reported but are rare. When metastatic disease is present, the five year mortality rate is 50-67%, therefore, early and correct diagnosis is important. In the general population, the mean age of diagnosis for sebaceous gland carcinoma is 61 years. In patients with bilateral retinoblastoma and radiation therapy, the mean age of onset is 21 years. Clinical awareness of this entity is important as increasing numbers of patients treated for retinoblastoma are surviving to adulthood. Stella M. Davies 

Radiation and childhood leukemia: of mice and men and women

Controversy exists as to whether paternal exposure to preconceptional x-rays increases the risk of leukemia in offspring [Gardner et al, Br Med J, 300:423-29, 1990; Doll et al, Nature 367:678-80, 1994; Petridou et al, Nature 382:352-52, 1996]. By and large the epidemiologic data are inconclusive. A mouse model (called ëN5í) has been developed whereby external exposure to radiation of male mice increases the risk of leukemia in offspring [NomuraJ Radia Res, 2:64-72, 1991]. Nomuraís work has been expanded by Daher et al [Carcinogenesis 19:1553-58, 1998], who exposed male mice to either an external x-ray dose of 500cGy or intraperitoneal injections of radioactive tritiated water( H³₂0) (internal dose estimate of 150cGy) prior to mating with unexposed females. Main outcomes of interest were fertility and leukemia incidence in offspring. For the irradiated mice, there was a significant reduction in litter size; the greatest impact on fertility was noted at least 17 days out from x-ray exposure. Moreover, paternal x-ray exposure resulted in an approximate doubling of the leukemia risk in offspring. No differences were observed in the litter size of the H³₂0 treated mice. There was, however, an approximate 7-fold increase in the incidence of leukemia in these offspring. For both the irradiated and H³₂0-treated mice, the greatest risk of leukemia in pups occurred in the first few months of life and diminished by the time the mice reached 8 months of age. For both treatment groups, the authors noted an excess number of female pups developing leukemia. 

COMMENT: This study nicely mirrors the epidemiologic data on infant leukemia. As in this animal study, the association between paternal x-ray exposure and childhood leukemia in humans is most apparent in infancy [Shu et al, Cancer Epidemiol Bio Prev 3:645-53, 1996]. Moreover, infant females develop leukemia approximately 50% more often than infant males [Gurney et al, Cancer 75:2186-95, 1995]. In this study, the risk of leukemia was highest in young female pups. Finally, the overall ratio of male:female pup births was 1.01, 0.99, and 1.2 for pregnancies fathered by the control mice, H³₂0-treated mice, and irradiated mice. Thus, in the irradiated mice, there was a significant excess of male births. This suggests that female mice may not have survived their leukemia in utero. Epidemiologic data suggest that fetal loss may be associated with an increased risk of infant leukemia [Yeazel et al, Cancer 75:1718-27, 1995], and that this association may be particularly apparent if the infant afflicted is female [Ross et al, Ann Epidemiol 7:172-79,1997]. It is important to point out, however, that the radiation dose used in this study, 500cGy, is higher than most people would ever receive over a lifetime, much less in one exposure. Therefore, radiation doses more typical of human exposures should be evaluated in this animal model. The N5 mouse model may represent an ideal animal system to further explore these (and perhaps other) epidemiologic observations with respect to infant leukemia. Julie A. Ross 

The paradox of better hygiene: infections and childhood leukemia

The incidence of childhood acute lymphoblastic leukemia (ALL) has exhibited marked geographic and temporal variation during the 20^th century. Incidence rates of childhood ALL are higher in developed countries. Further, within developed countries, higher socioeconomic status has been associated with an increased of risk childhood ALL. It has been suggested that the relationship between improved social conditions and childhood ALL may be due to an unknown environmental agent, or a change in exposure to infectious agents. 

Smith et al. investigated the hypothesis that improved public hygiene conditions (measured by prevalence of hepatitis A infection (HAV)) may be associated with an increased risk of childhood leukemia [Cancer Causes and Control 9:285-98, 1998]. HAV was chosen as an indicator of public hygiene because exposure to this virus, which is transmitted by the fecal-oral route, is common in areas of overcrowding, or areas with inadequate systems for removal and treatment of sewage. This ecologic study compared rates of HAV to rates of leukemia for children diagnosed less than 5 years of age in the United States. U.S. data on seroprevalence of HAV were obtained from the National Health and Nutrition Examination Survey (NHANES II), conducted from 1976-80. Using these data, Smith et al. estimated the HAV force (or exposure to infections) in the United States for this century. The authors conclude that the frequency of HAV decreased substantially prior to the increase in leukemia for white children in the United States. Further analysis of these data led the authors to speculate on some of the characteristics of a ëputative leukemia-inducing agentí. They suggest that this ëagentí is widespread in developed countries and persists in affected individuals. 

COMMENT:The relationship between improved social conditions and increased risk of childhood ALL has stimulated two similar but distinct theories regarding socioeconomic status, exposure to infectious agents, and childhood ALL [Greaves, Leukemia 2:120-5, 1986; Kinlen, Lancet 2:1323-27, 1988]. In this article, Smith et al. increased our understanding of this association by establishing a link between public hygiene and risk of ALL. However, as these links are based on multiple assumptions in an ecologic study, they must be interpreted cautiously. Kinlen, in an editorial published in the same issue of Cancer Causes and Control, further describes the limitation of Smith et al.ís analysis. One of Kinlenís main concerns is that although childhood ALL incidence data are compatible with a leukemic agent transmitted by the fecal-oral route, other possibilities such as droplet-spread infections were not explored (which may be equally important).However, it is encouraging to see new hypotheses developed . Andrine R. Swensen 

Smoke, but is there fire?

Epidemiologic data regarding the contribution of maternal and paternal smoking to the etiology of childhood cancer have been ambiguous. Recent studies have explored biologic markers that have been associated with cancer in newborns of mothers who smoked during pregnancy. A study presented by Steven Hecht et al. to the 1998 American Chemical Society meeting described the detection of metabolites of the tissue specific carcinogen, NNK, in the urine of newborns of smoking mothers. Another study by Finette et al. [Nature Med 4:1142-50, 1998] has examined the occurrence of somatic mutations in the reporter or marker gene, HPRT, in cord blood samples of infants from smoking and non-smoking mothers. While the absolute number of mutations was not increased in infants of smoking mothers, there was an increase in "illegitimate" genomic deletions mediated by VDJ recombinase. 

COMMENT: The data in this study are of interest and contrast with a study of adults who smoke in whom increased frequency of mutation (but no change in the spectrum) has been demonstrated [Vreiling et al., Carcinogenesis 13:1625-31, 1992]. In the current study, it should be noted that 30 HPRT mutants were isolated from 12 newborns with no exposure to smoke and 37 isolates from 12 infants with passive cigarette smoking. Each isolate was treated as an independent event, although a number of isolates came from the same infant. This leaves open the possibility that increased frequencies of particular types of mutation were confined to a limited number of individuals. Individuals can vary in their ability to metabolize carcinogens and repair DNA damage and thus, in their overall susceptibility to mutation and to cancer. Larger studies will be needed to determine differences in individual susceptibility.

While these data provide a potential biological mechanism whereby smoking could be related to cancer in offspring, it should be noted that the epidemiologic data on this issue are unclear. In five different case-control studies, adverse effects were associated with paternal smoking but not with maternal smoking. Other studies have failed to find an association. One of the references cited as supporting a role for exposure to tobacco-born carcinogens [Shu et al, JNCI 88:24-31, 1996] describes a modestly elevated odds ratio (1.6; 95% CI = 1.03-2.36) for paternal smoking one month prior to pregnancy.However, no elevated risk of AML or ALL was seen amongst mothers who smoke themselves. It is difficult to believe that passive smoking from contact with the father would have a more powerful effect than maternal smoking if the effect was due to placental transmission of carcinogens. The relatively modest positive associations reported in these epidemiologic studies suggest that the effect of tobacco is modest, unless there is a small population of highly susceptible individuals in whom the effect is large. Expansion of these biological studies to determine population variability may address this issue.Stella M. Davies

Caveat Emptor!

Increasingly, RT-PCR is being used to identify important genetic mutations in cancer. Whang et al. [JNCI 90:859-61, 1998] have described an important error in the use of this technology. These individuals are studying the potential tumor suppressor gene, PTEN/MMAC1, which has been implicated in a wide range of cancers, including glioblastoma, melanoma, and cancers of the breast, kidney and endometrium. Germ line mutations of PTEN/MMAC1 have been described in the cancer predisposition Cowdenís disease (cancers of heart and thyroid).These workers designed primers to amplify the full length coding region of this RNA. While testing the primers, they were surprised to find that they could identify a product of the appropriate size from all cell lines that they tested, including those that were known to have complete homozygous deletions of the gene. Subsequent work showed that they were actually amplifying a pseudogene.

Pseudogenes are non-expressed, intron-less, mRNA-like sequences of genomic DNA. These are derived from the processed RNA of an expressed gene and frequently contain a number of conserved mutations. At least two other previous studies of PTEN/MMAC1 in malignant tissue have probably mistakenly amplified the pseudogene and reported their findings as the presence of important mis-sense mutations in the malignancy. This mistake can be avoided by running PCR controls that contain no reverse transcriptase. 

Pseudogenes are common throughout the genome and should be considered when ambiguous or suprising results are obtained with PCR. PCR is a powerful technology and results should always be interpreted with caution. Stella M. Davies
 

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