(1) Seminar on “Transplacental and Transgenerational Carcinogenesis”
A seminar on “Transplacental and Transgenerational Carcinogenesis” was held at Volcano House, Volcanoes National Park, Hawaii, during November 13-14, 1990. Organizers were Dr. Taisei Nomura, Osaka University for Japan, and Dr. Jerry Rice, National Cancer Institute, for the U.S. This meeting was stimulated by the report of M. J. Gardner (Br. Med. J. 300:423-429, 1990) of an excess of childhood leukemia and non-Hodgkin's lymphoma in children born near the Sellafield nuclear power plant in the UK, whose fathers had been employed at the plant prior to the child's conception. The purpose of the seminar was to evaluate and assess current evidence for prenatal and preconception exposures to environmental agents as possible causes of human cancer, drawing on the epidemiologic literature and on experimental studies in animals, for both ionizing radiation and chemical carcinogens. Special attention was focused on the large Japanese data base from the Hiroshima and Nagasaki atomic bombings, as well as on studies undertaken in the U.S. and elsewhere, to look for events similar to the Sellafield cluster in other countries. A more complete meeting report is being prepared for publication.
The first session opened with presentations concerning human transplacental carcinogenesis. In an overview, R.W. Miller (Bethesda) described the first known human transplacental carcinogen, diethylstilbestrol (DES), and how it was discovered through a case-control study of a cluster of young women who developed a cancer of elderly women, clear cell adenocarcinoma of the vagina or cervix. It is now known that DES causes a congenital anomaly, adenosis of the lower female genital tract. This ectopic tissue in place of the normal squamous lining is prone to malignant change. Possibly phenylhydantoin also acts as a teratogen/carcinogen, for four cases of neuroblastoma have been reported in children with the fetal hydantoin malformation syndrome. Other possible transplacental carcinogens are cigarette smoke, including passive exposures, aflatoxin, benzene, artistic painting, and any other known human carcinogen. The effects will probably not occur until adulthood. The Japanese custom of saving the dried stump of the umbilical cord as a memento of the child's birth has been shown to be useful, for example, in detecting mercury exposure at Minamata Bay where an epidemic of neurological disease occurred, including cerebral palsy as an in utero effect.
Y. Yoshimoto (RERF, Hiroshima) reported on an important reanalysis of the in utero data from the Japanese atomic-bomb survivors. The results showed that an excess of cancers was beginning to appear especially after exposure to 1-29 cGy (rad), but was no more than occurred at each succeeding year of age until age 5. In fact, those exposed at 3 or 4 years had somewhat higher rates of adult cancers than did those who were younger. This finding contrasts with Alice Stewart's reports of markedly increased childhood cancers among fetuses exposed to low-dose radiation (pelvimetry mostly) late in gestation.
The subject then turned to transgenerational carcinogenesis from radiation exposure of either parent. K. Mabuchi (RERF, Hiroshima) described cancer in the F1 generation of A-bomb survivors. From 1946-1989, only one person developed leukemia among those whose parents had each been exposed to 1 cGy or more. That person was among 208 in the exposure category, 10-49 cGy. None developed cancer among the 90 whose parents received 50+ cGy.
N. Kamada (Hiroshima University) reported on cytogenetic findings in leukemia among the F1 generation: 20 cases, of which 13 had chronic myelogenous leukemia (CML). All were over age 16 at examination (1972-1988). Both parents had been exposed in seven cases, only the father in five, and only the mother in eight. Cytogenetic abnormalities did not differ from those found in the general population of leukemia patients. All patients with CML had a Philadelphia chromosome. The findings among survivors directly exposed to the A-bomb had many more cytogenetic abnormalities.
S. Abrahamson (University of Wisconsin) examined the implications of Gardner's report on leukemia in children conceived after their fathers had worked at Sellafield. The analysis, based on information consistent with radiation genetics in animal and human studies, indicate that the doses at Sellafield were too low to induce the excess of leukemia observed. If the assumptions on which the analysis was based are correct, the leukemia cluster must have some other explanation.
Since 1948, James V. Neel (University of Michigan) has been seeking genetic changes among Japanese children conceived after their parents were exposed to the atomic bomb. Starting with the imprecise indicators of the time (birth defects, neonatal and infant mortality rates, anthropometric measurements, the sex ratio at birth, and F1 mortality), and moving through cytogenetic studies to biochemical genetics, no evidence of an excess of mutations has been found. Specimens of DNA are now being stored for future studies. In the long history of these studies, from clinical observations to the most sophisticated laboratory means for detecting genetic mutations, no effect has been detectable. It is presumed that an effect did occur, as it does in all species studied experimentally, but is not discernible in the population under observation. The leukemia cluster at Sellafield is thus not in accord with radiation genetics based on extensive studies of diverse effects among A-bomb survivors.
The second session concerned paternal versus maternal origin of germline mutations of cancer and other diseases. M. Sasaki (Kyoto University) reported that germinal mutation of retinoblastoma (RB) genes has a strong bias to paternal origins, as does the initial mutation of RB genes in sporadic osteosarcoma. Previous reports claimed that the initial somatic mutation in sporadic RB had an equal contribution from the mother and father, but Sasaki's study of the parental origin of seven such cases indicated that they are mostly derived from the father. The lack of an effect of the father's age on germ-line mutation, suggests that the paternal age effect is post-meiotic rather than pre-meiotic. It is also likely that genomic imprinting is a factor in mutation and pre-mutation, be it spontaneous, radiogenic or chemically induced.
G. Bunin (Children's Hospital of Philadelphia) spoke on epidemiologic evidence for transplacental and transgenerational carcinogenesis. Such studies of paternal occupation showed that fathers of children with Wilms tumor were five times more likely than control fathers to be employed as welders. Employment of either parent in metal-related occupations has been associated with retinoblastoma, brain tumors, acute non-lymphocytic leukemia, and hepatoblastomas. She mentioned many other individual studies that suggest parental exposures as risk-factors for diverse childhood cancers. Without replication, however, the results of these studies must be viewed only as hypotheses. At Sellafield the father's exposure was before, during, and after gestation. She suggested that diseases due to single gene mutations, such as achondroplasia or neurofibromatosis, be studied for exposures that may have caused these new mutations.
J.M. Friedman (University of British Columbia) addressed the question, does paternal exposure to toxic agents cause congenital anomalies? No such effect has been convincingly shown in the human. An increased frequency of genetic disease resulting from exposure of men to potential mutagens may, however, be important at the population level. Assessment of this risk is necessary. Transmission of a toxic chemical or metabolite from the father to the mother after conception could be teratogenic. Genic mutation would have to occur before conception. The frequency of new mutations that produce clinically important disease is unlikely to be greater than one percent. Fewer than half would be apparent at birth. Most studies of paternal exposure to a mutagen and congenital anomalies in the children have used a cohort design. In a perfectly designed and executed study, about 20,000 exposed infants and an equal number of controls would be required to detect a doubling of new mutations. If an increase were found, one would still have to establish that the relationship was causal, a difficult task. Case-control studies are more likely to be successful. A doubling of the odds ratio with exposure in five percent of the general population probably would be detected in a study of 500 cases and equal number of controls.
C.S. Chung (University of Hawaii) re-evaluated data published by Shiono et al (1980) on prenatal X-ray examinations in the Collaborative Perinatal Program conducted by the National Institute of Neurological and Communicative Disorders and Stroke. Nine cases of leukemia or lymphoma were compared with 17 cases of cancer that may be heritable (retinoblastoma, neuroblastoma and Wilms tumor). The results concerning exposure before conception were consistent with Gardner's results, but no effect was found among those exposed in utero, in contrast to the findings by Alice Stewart. Potential selection bias may affect the results.
T. Fujii (Teikyo University, Tokyo) told of her studies concerning transgenerational effects of maternal exposure to immunosuppressive/anticancer drugs in the rat. Low maternal calcium induced by parathyroidectomy during pregnancy produced impaired regulation of serum calcium levels in the offspring or increased their lethality (LD50) from injected calcium chloride -- even in the tenth generation. In another experiment, administration of cyclophosphamide on the tenth day of gestation in rats induced a significant decrease in the pentagastrin-stimulated calcitonin release from C cells in the offspring. On the tenth day, cells of the rat's neural crest are migrating. The same impairment was observed in the next generation.
The third session concerned mechanisms of transplacental and transgenerational carcinogenesis in animals. In a model system, H. Yamasaki (International Agency for Research on Cancer), initiated cancer transplacentally and promoted it postnatally. Skin cancers were induced in mice by transplacental initiation with DMBA, and promoted by postnatal exposure to TPA. Many of the tumors had a specific mutation, the A--> T transversion at the 61st codon of c-Ha-ras. No such mutation occurred when benzo(a)pyrene was the initiating agent. The implication is that fetal c-Ha-ras can be activated transplacentally by a carcinogen through a point mutation. Tumors of the lung, liver and lymphoreticuloendothelial system are common in CD-1 mice. Transplacental DMBA exposure increases their frequency. Why these specific sites? Perhaps activation of the ras gene was involved. The IARC group used Xbal RFLP analysis and differential oligonucleotide hybridization, and found an A-->T transversion at the second position of codon 61 of Ha-ras oncogenes in skin and liver tumors, but not in lung or lymphoreticular tumors. A high proportion of DMBA-induced liver tumors had the mutation, but none of 11 spontaneous hepatomas did. To examine more directly the interaction of carcinogens with fetal oncogenes, the IARC group has developed a method by which Ha-ras mutation can be detected before the appearance of the tumors. Preliminary results indicate that the test is useful for examining the quantitative correlation between oncogene activation and tumor occurrence, as well as chemical specificity of oncogene activation in vivo.
H. Watanabe (Research Institute for Nuclear Medicine and Biology, Hiroshima) spoke on the effects in mice of paternal 252Cf-neutron exposure on the F1 offspring. Seven-week-old male C3H male mice were exposed to 252Cf neutrons at doses of 0, 50, 100 or 200 cGy. Two weeks later they were mated with 9-week-old female C57 BL mice. The number of implantations per mouse was not dose-related, but survival was. The frequency of abnormal sperm of the fathers increased linearly with dose. Among the male offspring in the group exposed to 50 cGy, 19 of 44 (43.2 percent) had liver tumors as compared with 1 of 31 (3.2 percent) in the non-irradiated group (p < 0.01). In female offspring, no such increase was observed. The sperm abnormality is thought to have led to dominant lethal mutations. These male mice have a high rate of spontaneous liver tumors, which may be further increased by 252Cf neutron irradiation.
L.M. Anderson (NCI, Bethesda) stated that since 1950, 29 reports and 2 abstracts have presented positive findings of transplacental tumorigenesis among rats or mice treated with six carcinogenic chemicals or x-rays. Five papers described negative results. From these data at least two categories of mechanism may be considered: 1) heritable structural gene change, such as a mutation in an oncogene or at a locus for growth regulation, and 2) alteration in the cellular imprinting of the gametes, an epigenetic change that influences gene expression in the offspring when they reach adulthood, including a role in tumorigenesis. The two mechanisms are differentiated by about a dozen features in the patterns of lineal tumor occurrence. Most positive effects have been noted with potent genotoxicants, but aside from DDT, few chemicals have been studied that might act epigenetically. DES exposure in utero and high dietary fat intake by the offspring increased the frequency of ovarian and mammary tumors in the offspring. In chemically induced tumors of the mouse lung, inheritance was predictable and unaltered through at least three generations, likely the result of a structural gene change. In the rat, the multigenerational effect (tumors of the nervous system, lymphoid, mammary, and pituitary tissue) disappeared by the fifth generation, suggesting that imprinting was the mechanism. X-ray treatment of mice before mating led to increased susceptibility of their offspring to lung tumorigenesis by urethane. On balance the data favor imprinting as the mechanism most involved, or a combination of both mechanisms.
O. Niwa (Research Institute for Nuclear Medicine and Biology, Hiroshima) told of studies concerning radiation-induced germ-line mutations in minisatellite sequences of the mouse. The high variability of these sequences has been exploited for tracking transmission of certain alleles through germ cells. The high copy number of minisatellite sequences and their apparent lack of function are great assets in detecting the genetic effect of radiation on germ cells and their transmission to the next generation. Although somatic mutations were observed after radiation of mice at the MO-2 locus, no germ-line mutations could be found. In contrast, the Pc-1 locus was highly mutable in germ cells even without irradiation (manifested by change in the length of the restriction fragment, indicating that the mutation was not due to a deletion). Unequal crossing over or slippage during DNA replication can best explain these mutations. Among the offspring of irradiated males, similar mutations were detected at the Pc-1 locus.
J.M. Rice (NCI, Bethesda) described molecular approaches to confirm germ cell mutations in experimental transgenerational carcinogenesis. A specific genetic strategy is required to distinguish germ-line from somatic cell mutations. Such a strategy can be defined for specific rodent tumors that regularly contain a transforming protooncogene that is activated by a point mutation. In any tumor due to a germ-line mutation in specific protooncogenes, a transforming allele of the appropriate cellular protooncogene should be detectable not only in the tumor, but also in all normal tissue of the affected individual. Because these genes can be demonstrated in formalin-fixed paraffin-embedded tissues from specimens in storage by means of polymerase chain reaction techniques, one can test for the presence of transforming alleles in each tumor and normal tissue from animals in selected experiments on carcinogenesis. Through the use of paraffin blocks from ten rats with malignant schwannomas that were either naturally occurring or a result of paternal exposure to a carcinogen, the expected T--> A transversion was identified in three tumors, one of which occurred in the offspring of a carcinogen-treated father. The same mutation must now be sought in the normal tissue. Because this was a solitary tumor in an aged animal, it may well be naturally occurring. This experience illustrates how the procedure can be used to seek an extra dividend from the study of archival material.
T. Nomura (Osaka University) described his studies, which are of great current interest, concerning radiation-induced transgenerational carcinogenesis in mice. He said that the difference between the Sellafield cluster and the lack of such an effect among Japanese atomic-bomb survivors might be explained by his experimental observations. First there might be a difference in susceptibility to germinal mutations that cause leukemia, as seen in the N5 mouse strain, whose spermatogonia are highly susceptible, as compared with the ICR strain, in which they are not. A similar difference might exist between the people at Sellafield and those in Hiroshima/Nagasaki. Second, the risk at Sellafield was extremely high during the six months before conception, which suggests higher radiosensitivity of the postmeiotic sperm than of spermatogonia. ICR mice showed an excess of leukemia after X-ray exposure of spermatozoa or spermatids, but not after exposure of spermatogonia. High mutational sensitivity of postmeiotic sperm has also been observed in mice by the specific loci method. Too few survivors of the atomic bombs conceived children within 6 months of exposure to determine if leukemia occurred after irradiation of the germ cells that are most sensitive in mice. Third, at Sellafield postnatal exposure to chemicals and/or further radiation (through contamination of homes) might have enhanced the effect of preconception irradiation. This possibility is in accord with studies of mice that show continuing hypersensitivity to tumorigenesis when exposed postnatally to tumor promoting agents.
Sieber (NCI, Bethesda) commented on three epidemiologic studies of radiation carcinogenesis and the need for laboratory research to go beyond the reach of epidemiology. First, the Sellafield cluster of childhood leukemia suffers from lack of accurate dosimetry; the possibility of other explanations, such as occupational chemical exposures; and uncertainty about chronic low-dose effects as compared with those from a single instantaneous exposure, as among Japanese A-bomb survivors, the main source of information on radiation effects. Studies of the interaction of chemicals and radiation would best be evaluated through animal experimentation. Second, a study by the Radiation Epidemiology Branch of NCI found no relationship between type-specific cancer mortality and residence in counties near nuclear facilities as compared with 1) rates before the facilities were activated, and 2) similar counties without such installations over a 35-year interval. As the authors pointed out, mortality is not as good a measure as incidence would have been, had these data been available, and counties are large enough to dilute out more localized increases in mortality. Third, a three-year case-control study has been initiated by the Biostatistics Branch concerning acute lymphocytic leukemia in childhood as related to electromagnetic field (EMF) exposure. Two thousand newly diagnosed cases are being ascertained through the Childhood Cancer Study Group. The possibility that EMF is a promoting influence should not be overlooked, and would best be evaluated through animal experimentation in which chemical carcinogens are administered first.
Overview: The discussion, which will be reported elsewhere by Rice, Anderson and Nomura, featured the inability to explain transgenerational carcinogenesis by conventional genetic theory. When genomic imprinting was suggested as a mechanism, close questioning by the classical geneticists present could not discount this explanation.
(2) Workshop on “Renal Cancer”
A workshop was held at the East-West Center, Honolulu, Hawaii, February 18-19, 1991, to review recent progress in studies of the etiology and pathogenesis of cancer of the kidneys, particularly Wilms tumor and renal cell carcinoma. Attending the meeting were experts in clinical oncology, epidemiology, and laboratory sciences. The scientific diversity of Japanese and U.S. participants fostered opportunities for exchanges across research disciplines.
Kevin Bove (Children's Hospital of Cincinnati) opened the meeting with a presentation on the pathology of renal cancer in children. Most childhood renal tumors are nephroblastoma (Wilms tumor) and its variants that have a relationship to metanephric blastema. Wilms tumor and its variants are presumed to arise by one or more transforming events in renal blastema cell. Kidney specimens containing Wilms tumor also tend to have microscopic nodules containing blastema whose potential to develop into clinically significant tumors is unknown. However, these lesions predict for tumor development in the contralateral kidney. Many of these nodules exhibit regressive features, analogous to observations made of retinoblastomas. Studies are in progress on immunohistological and proliferative characteristics of normal and abnormal blastema occurring as independent nodules in kidneys containing Wilms tumor, Beckwith-Wiedemann syndrome, or sporadic multicystic renal dysplasia. Other childhood tumors in the kidney include mesoblastic nephroma, sarcomas, angiomyolipoma, and renal cell carcinoma.
In answer to questions, Bove stated that the value of the National Wilms Tumor Study (NWTS) has been its vast accumulation of data, and the continual generation of hypotheses. Separation of nephrogenic rests into two classes, intralobar and perilobar, was a major advance, made possible by study of a large number of cases. There are more familial cases of Wilms tumor in the NWTS than there are case-reports in the world literature.
Patients with trisomy 18 have accumulated nodular blastema, but survival is so short that progression to Wilms tumor has not been seen. A registry of cases with trisomy 18 is maintained in San Antonio, Texas, where studies could be made of the lesions of the kidneys and their possible relationship to Wilms tumor.
Only about 15 percent of children with Wilms tumor do not survive. They have more malignant forms of the neoplasm -- including the sarcomatous type, develop aneuploidy or are casualties of treatment. An attempt should be made to determine if nephrogenic rests are tumorigenic by transplanting them into nude mice.
Rikuo Machinami (Tokyo University) reviewed findings at autopsy of 5512 Japanese cases of renal cancers, 1974-88. Renal cell carcinoma was the diagnosis in 91% of these cases The predominant type was clear cell carcinoma (66% of all cases), and poorly differentiated carcinomas were second in frequency. In many tumors, several histological subtypes were observed, and classification is based on the dominant type. Median age at diagnosis was 64 years. The sex ratio (M/F) is almost 3, but the explanation for the predominance in men is unknown. Analysis of the cases by place of residence showed high renal cancer frequency in two districts, Kanto and Hokuriku, where additional studies might be informative Clinical-pathological analyses show that sarcomatoid renal cell carcinoma exhibited highly malignant behavior. This and other tumors of high nuclear grade gave evidence of high proliferative thrust, as measured by NORs (nucleolar organizing regions) and PCNA (proliferative cell nuclear antigen).
Discussion of the paper brought out the suggestion that case-control studies could be made in areas with high rates of renal cell carcinoma (RCC), such as the Kanto area in Japan and north central areas of the United States. Through the Japanese Autopsy Registry, the histology of RCC under age 20 years might be compared with that of older patients. To do so, one would have to follow back from the published abstracts to the reporting hospitals to obtain pathology specimens for study.
Yoshiaki Satomi (Yokosuka-Kosai Hospital) discussed the epidemiology of renal carcinoma. In 1968, age-adjusted renal cancer mortality rates in Japan were considerably lower than those of several Western nations. However, reported rates have risen steadily in Japan in recent decades. The role of improved diagnosis of renal cancers or increased exposure to renal carcinogens is unclear. Studies in Japan show that the rise in renal cancer rates correlates with increased consumption of meat and cow's milk, though the association may be non-causal. Unusual clinical characteristics of renal carcinomas include frequent late relapses, variable rate of growth, and spontaneous regression after nephrectomy in isolated cases. Recent studies in Japan have reported that some renal cell carcinomas secrete IL-6 in an autocrine manner, and that circulating level of which correlated with tumor growth rate.
As a urologist Dr. Satomi has treated about 500 hundred patients with RCC, none of whom was known to be a familial case. A large cohort study indicated no relation to cigarette smoking, but drinking milk seemed to be a risk factor.
Robert W. Miller (National Cancer Institute) presented the epidemiology of Wilms tumor and renal cell carcinoma. A demographic peculiarity of Wilms tumor in Japan and other Asian countries is that the incidence is about half that in the rest of the world. The age-peak is at 1-3 years, and boys and girls are equally affected. Wilms tumor is not known to be caused by environmental agents. It is occasionally familial. Comparisons of the epidemiologic features of Wilms tumor with those of renal cell carcinoma are shown in Table 1. The first foothold to its etiology was its association with aniridia, described in 1964, and later found to be due to deletion of chromosome 11p13. This finding pointed to the locus of a gene, WT-1, as later proven by molecular biology -- a fine example of the interaction of clinical observations, epidemiology and laboratory research. WT also occurs in the Beckwith-Wiedemaun syndrome (BWS: visceral cytomegaly, large tongue, omphalocele and often congenital hemihypertrophy), hamartomas, and as a double primary with hepatoblastoma or adrenal cortical carcinoma. Thus, four types of growth excess are associated with one another [cells that are too large (cytomegaly), too many throughout the body (hemihypertrophy), or as neoplasms -- benign (hamartoma) or malignant]. The diversity in these forms of overgrowth remains to be explained.
The discussion turned first to possible reasons for the low rates of WT throughout Asia. Could it be due to a less mutable suppressor gene? No ethnic variation in tumor suppressor genes or oncogenes is yet known. Genes for other diseases do vary in their frequency; e.g., the absence of cystic fibrosis of the pancreas in Japan.
Why did Knudson and Strong (1972) in their study of the age-distribution of WT cases find that, unlike the log-linear plot observed for familial cases and for sporadic cases with aniridia, the plot for hemihypertrophy was curvilinear. It was suggested that a) the hypertrophied kidney may provide more target cells than a normal kidney, with disproportionate numbers of progenitor cells to nephrons; and b) that everyone has some asymmetry, which was overdiagnosed after the association with WT was first described.
Prince Masahito (Department of Experimental Pathology, Cancer Institute) and N. Okamato (Tokyo University of Fisheries) described their studies of Wilms tumor in sea eel. These nephroblastomas were observed in nearly 1% of eels reared in a farm for five to nine months and were histologically very similar to human Wilms tumor. Tumors arose from the kidney near the anus and were noted externally by abdominal swelling. Although the cause underlying the development of these tumor is unclear, it is notable that the incidence of these tumors has significantly increased after the spread of indoor eel culture at raised water temperatures. Environmental factors associated with new culture conditions (elevation of NO2-N or NH4-N in cultured fluid, low Ph) have been suggested as important factors for tumorigenesis. The eel genomic clone homologous to human Wilms tumor gene (WT33) was isolated to examine the mutation in tumors. So far, no abnormality was detected by Southern blot analysis.
In the discussion it was noted that the highest frequency of Wilms tumor in eels was one percent, almost entirely in males. The earliest tumors occurred at five months of age. The gene size is about 40 Kb, as in the human. Also, could the extreme crowding in the huge agriculture tanks cause concentration of excreted carcinogens? Studies of such contaminants in the water have not yet been made.
David Housman (Center for Cancer Research, MIT) presented studies on the Wilms tumor gene. By the method of positional cloning, the gene located on chromosome 11q13 that is responsible for Wilms tumor was isolated. The gene had a zinc finger motif and is expressed in kidney. The gene product is located in the nucleus and shows high homology to EGR (early growth response gene) 1 and EGR 2. Constitutional mutation, which caused the deletion of one zinc finger motif, was detected in one Wilms patient. As loss of heterozygosity was not detected in the tumor with a constitutional mutation, the dominant negative model was considered as the mechanism of tumorigenesis. Another gene associated with Wilms tumor is suspected to be in chromosome 11p15, duplication of which causes Beckwith-Wiedemann syndrome. However, as familial Wilms tumor did not show linkage with markers on chromosome 11p13 or p15, a third Wilms gene might also exist somewhere else in our genome.
The discussion brought out that the known tumor suppressor genes have no single pathogenic feature in common, such as zinc fingers. Tumor suppressor genes in the human occupy rate-limiting steps in pathways that lead to tumorigenesis. Different species may have different pathways. Pathologists may learn from molecular studies which precursor lesions are related to Wilms tumor. As a general rule, big deletions inevitably include a house-keeping gene, the loss of which prevents cell survival.
Yasuhiko Kaneko, (Saitama Cancer Center) presented cytogenetic and molecular genetic analyses of Wilms tumor in Japan. 44% of 33 Wilms tumors had hyperdiploidy. The frequent chromosomal aberrations in tumors are + lq/ + i(lq), +6, +8, +12 and +18. del (11) (p13p13) was observed, constitutionally in two cases, and in two tumors. One tumor had inv (11) (p15p21). A reciprocal translocation was seen two cases. By Southern blot analysis one tumor showed homozygous deletion, and hemizygous deletion was observed in 4 cases. By comparing cytogenetic aberrations with histological classification of Wilms tumor, unique chromosomal abnormalities were detected in each histological type. Trisomy of 7, 12, 13, 17 and 18 and 21 were seen in cystic, partially differentiated nephroblastoma. Trisomy and tetrasomy of chromosome 11 were observed in congenital mesoblastic nephroma (CMN). t(2:22) was seen in of the clear cell sarcomas. Among them, deletion of chromosome 11p13 was the most common features except in CMN.
Frederick Li (National Cancer Institute) reviewed data gained from analyses of families at high risk of kidney cancer. In a family with a history of renal carcinoma in seven relatives, three additional cases were detected through screening. All evaluable renal cancer patients in the family had a constitutional 3;8 translocation. Recent molecular and cytogenetic analyses of tumor cells from family members reveal loss of the derivative 8 chromosome, but no changes in the normal chromosomes 3 and 8. A second family with 5 cases of Wilms tumor was observed for a decade when 2 additional cases of Wilms tumor occurred. Analysis of the family results excluded linkage with polymorphic markers for chromosomes 11p13 and 11p15, loci of candidate genes for sporadic (non-familial) Wilms tumor. The data indicate a third gene of unknown location that confers high risk of Wilms tumor. In the dominantly inherited syndrome of breast cancer and diverse childhood tumors (Li-Fraumeni syndrome), recent studies indicate that inherited mutations in the tumor suppressor gene, p53, is the molecular defect in these families. There might be an excess of Wilms tumor in carriers of p53 mutation, but renal carcinomas have rarely occurred in these families.
Are the four Wilms tumors among 91 families part of Li-Fraumeni syndrome (LFS)? If so, another locus (p53) for WT in addition to the two on chromosome 11 would be known. If the sub-type of the two cases of RCC among the 91 families were unusual, suspicion would be stronger that they are part of the LFS. One family had seven members with lung cancer, a tumor that was rare in other families. Thus, some families may have an exceptional type-specific cancer link to LFS. Hepatoblastoma has been observed in several Japanese families, but only one case has as yet occurred in a U.S. family.
How to look systematically for LFS? A search through data from the Connecticut Tumor Registry yielded only one family. A search of hospital records, using adrenal cortical carcinoma as a cancer likely to be part of LFS, can be productive, as in a study by Dr. Tsunematsu in Japan. A screening test used by Dr. Stephen Friend can detect the syndrome in families too small to have a substantial aggregations of LFS-related cancers. Random errors in the p53 gene are sought through gel electrophoresis. When found, gene-sequencing is carried out to detect mutation.
Isao Ishikawa (Kanazawa Medical University) presented data on the occurrence of renal cell carcinoma in patients with acquired renal cystic disease. Studies of the natural history of acquired renal cystic disease reveal a more rapid progression of the cysts in males as compared with females. The longer the duration of hemodialysis, the more numerous are the cysts that develop. Successful renal transplant cases show suppression of cysts. A ten-year prospective study has revealed that renal cell carcinoma in dialysis patients arises in 1 per 250 patient-years of observation, with a higher frequency in men than in women. Two patterns of renal cell carcinoma were found in dialysis patients. One type occurred in older patients after a short time on hemodialysis, and involved fewer renal cysts. The second type occurred in relatively young patients on long-term dialysis and with extensive cyst formation. The latter group accounted for a rising cancer frequency after dialysis for more than five years. These findings suggest a causal relation between acquired renal cystic disease and renal cell carcinoma.
In response to questions, it was said that the histology of RCC related to dialysis was 30 percent clear cell type, 10 percent granular cell and 60 percent mixed type. Geographic differences in the occurrence of RCC are not likely to be due to differences in the availability of dialysis, but to differences in interest of physicians in diagnosing the neoplasm.
The expansion of the kidney with progression of cystic disease is due to fluid accumulation, not to the growth of solid tissue. The fluid in autosomal dominant cystic disease of the kidney is high in creatine and very low in!!
!2-micro-globulin, and different from that in cysts in patients with end-stage renal disease on dialysis. In the latter, carcinogenesis is thought to develop from the proximal tubules that give rise to the cysts. In von Hippel-Lindau disease the kidney is much more prone to RCC than are simple renal cysts. Might the primary event in acquired cystic disease of the kidney be epithelial hyperplasia that gives rise to the cysts, and a second event causes carcinogenesis? Doubt was expressed that the primary tubules could be the origin of relatively primitive cancer cells. It may be that there are two kinds of tumors that develop from cysts, different in polycystic disease from those in end-stage renal disease. Susan Naylor (University of Texas, Health Science Center at San Antonio) showed data on hybrid mapping of chromosome 3, and tumor suppression of small cell lung carcinoma by chromosomal transfer. Her group generated a large number of somatic hybrid cell lines which contained very small pieces of human chromosome 3. These cell lines were characterized with a large number of markers. The part(s) of chromosome 3 was transferred into a lung cancer line (h592) by microcell transfer. Suppression of tumor formation was assayed by tumor volume in mice. It was shown that chromosome 3p, around ALAS1, contains the gene suppressing tumor formation of H592. Introduction of a very small piece (approximately -2cM) of chromosome 3 also suppressed A9 tumor formation in nude mice. The segment was derived from p21, the region thought to be involved in small cell lung carcinoma. The segment contained GLB1 and ALAS1, but not 3GTB9 or D3F15S2. One case, in which the cholecystokinin locus was homozygously deleted, was also reported.
Is the clinical problem in small cell carcinoma of the lung due to early metastases rather than growth of the primary tumor? Study was made of the growth of the tumor in nude mice, which does not measure metastases.
Yusuke Nakamura (Department of Biochemistry, Cancer Institute) presented the allelotype study and the deletion mapping of chromosome 3p in 47 renal cell carcinomas (RCC). By allelotype study, the most frequent LOH (loss of heterozygosity) (75%) was observed at one or more loci on chromosome 3p. Nearly 30% LOH was observed with markers on chromosome 5q, 6q, 19q, 17p, and 19p. These results lead us to suspect the existence of tumor suppressor genes associated with the carcinogenesis of RCC on these chromosomes. The correlation of FAL (fractional allelic loss) value to histopathological grade and tumor size was reported. All seven tumors with FAL of > 20% was classified to grade 2 and others were grade 1. Furthermore, 3 of 24 tumors with FAL of < 20% were T3 or T4 although 6 of 8 tumors with FAL of > 20% were T3 or T4. Commonly deleted region in RCC was also reported by using 25 RFLP markers on chromosome 3p. Two separated regions in chromosome p21.3 and p13-14 were suspected to contain two tumor suppressor genes for RCC.
The discussion examined the utility of allelotyping and karyotyping. If the majority of cells contain the same mutation, it can be detected by allelotype, but if a small proportion of cells contain some mutations, it cannot. Both allelotyping and karyotyping have their advantages. The time may come when a simple method can be devised for allelotyping.
Might a deletion be beneficial to cells, such as deletion of an oncogene? Some immunologic data may indicate that one can kill cells using an antibody against activated ras. Hit-and-run models for tumor progression provide for expressing an activated oncogene temporarily to facilitate one aspect, but handicap a second aspect of metastasis.
Gyula Kovacs (NCI-Frederick Cancer Research and Development Center) showed data on molecular cytogenetics of hereditary and sporadic renal cancers. Combined histologic, cytogenetic and RFLP analysis resulted in the discovery of subtypes of renal cell cancers each characterized by distinct molecular pathology and natural history. The vast majority of renal cancers are nonpapillary, clear cell renal carcinomas which are marked by loss of one of the homologous 3p13-pter segment, trisomy of the 5q22-qter segment and monosomy 14. Renal clear cell carcinomas which develop in patients with von Hippel-Lindau disease share karyotype changes with sporadic cases. The most frequent genetic event leading to the loss of 3p segment is a nonhomologous chromatid exchange between chromosome 3 and others. In contrast, papillary renal cell cancers arise from embryonal rests of the kidney and are always multi-focal. These tumors are marked by a combination of trisomy 7 and 17 and loss of the Y chromosome. During progression, tumors acquire trisomy 16, 12, 20 and other changes. Another tumor, renal oncocytoma is a benign tumor of the kidney. Oncocytoma is characterized by alteration in restriction pattern of the mitochondrial DNA. However, no recurrent chromosome aberration has been found in renal oncocytomas.
Why were interstitial deletions not seen, as others have reported? They were looked for carefully and not observed in four cases with karyotypically normal chromosome 3. In other cases the presence of translocations practically preclude the detection of interstitial deletions. Also, why does 3p13 and 5q translocation occur, but when 3p11 is the breakpoint, the translocation is with 8 or some other chromosome? One possibility is that there may be homology between 3p13 and 5q. With regard to the role of additional chromosomal changes, none was found in tumors under 2 mm in diameter. In tumors under 1 cm in diameter, ten percent had monosomy 14, whereas in larger tumors nearly 50 percent did. In some cases later cytogenetic changes elsewhere suggested clonal evolution.
Mitsuo Oshimura (School of Life Sciences, Tottori University) presented tumor suppression by chromosome transfer. Transfer of chromosome 6, 9 or 11 could suppress the tumorigenicity of a uterine endometrial carcinoma cell line, but chromosome 1 or 19 were unsuccessful for suppression. Tumorigenicity of HT1080 was suppressed by the transfer of chromosome 1 or 11 but not by chromosome 2. Chromosome 11 could suppress better than chromosome 1. This indicates that different members of the family of tumor suppressor genes are present on different chromosomes, that more than one normal chromosome can suppress the tumorigenicity of certain tumors, and that functionally distinct tumor suppressor genes exist. For tumor suppression of renal cell carcinoma cell line, one copy of chromosome 3 is not sufficient and two copies of chromosome 3 were needed to suppress tumor formation in nude mice. Dosage effect for tumor suppression was suspected from this result. Furthermore, the structural difference between tumor and normal cells was also reported. Microvilli was observed in a tumor cell, but not in a normal cell.
In summary, peculiarities in the occurrence of Wilms tumor and renal cell carcinoma may provide clues to the etiology of these neoplasms. Wilms tumor rates differ among nations, and are low in Japan and other parts of Asia. Its occurrence is limited to children, suggesting the etiological role of prenatal events. Renal cell carcinoma rates may also show international differences in rates and case clusters Men are consistently affected more often than women.
Patients have been identified who are at exceptionally high risk of renal cancers. High-risk groups for Wilms tumor include members of cancer families and persons with associated birth defects, such as aniridia. Renal cell carcinoma also occurs in families as well as dialysis patients. Cytogenetic and molecular genetic studies have helped map two Wilms tumor-associated genes to chromosome 11p and identified a possible third locus. For renal cell carcinoma, 3p deletions and translocations appear to inactivate tumor suppressor gene (s) within regions of overlapping deletions among tumors. The integration of clinical, epidemiological and laboratory studies have speeded progress towards understanding of the genesis of these neoplasms.
EPIDEMIOLOGY: WILMS TUMOR COMPARED WITH RENAL CELL CARCINOMA