660
The table shows that the ratio of neuroblastoma to Wilms’ low in the 1960s and was still low in Tanzania in 1980-81. In Uganda the ratio (0-3) was higher in 1947-66 than it was later in Kampala (0- 1) and in the West Nile district (zero). The earlier Uganda ratio of 0-3 is substantially below those elsewhere in the world (eg, the US and Manchester, England, 1-35). The ratios in Kenya and Zambia were available only for the late 1960s, and in Zaire (Kinshasa) for 1968-85; all were below 0-1. It would be a pity to pass over this evidence relating to the origins of neuroblastoma because of a contrary clinical impression. tumour was
Clinical Epidemiology Branch, National Cancer Institute, Bethesda, Maryland 20892, USA
ROBERT W. MILLER
DM, Stiller CA, Draper GL, et al, eds. International incidence of childhood (International Agency for Research on Cancer science publication no 87). Lyon: IARC, 1988. 2. Miller RW. Ethnic differences in cancer occurrence: genetic and environmental influences with special reference to neuroblastoma. Progr Cancer Res Therapy 1977; 3: 1-14. 3. Massabi M, Muaka BK, Tamba N. Epidemiology of childhood cancer in Zaire. Lancet 1989; ii: 501. 1. Parkin
cancer
Cerebral
palsy in infants born during trial intrapartum monitoring
of
SiR,—Your readers may have been seriously misled if they read the letter from Dr Hope and Dr Moorcraft (Jan 27, p 238) without reference to our report (Nov 25, p 1233). 16 (78%) of the 22 patients with cerebral palsy in the study had not shown any clinical signs suggestive of intrapartum asphyxia. The fact that the other 6 patients did show signs that could possibly reflect asphyxia does not necessarily mean that the signs were due to intrapartum asphyxia, nor should it be interpreted as meaning that the cerebral palsy was caused by intrapartum asphyxia. Furthermore, even if in some of these 6 patients cerebral palsy was caused by intrapartum asphyxia, it is very questionable whether it could have been prevented by different clinical management, in view of our current knowledge. National Perinatal Epidemiology Unit, Radcliffe Infirmary, Oxford OX2 6HE, UK
Inheritance of
ADRIAN GRANT
glomus
tumours
SIR,-Dr van der Mey and colleagues (Dec 2, p 1291) suggest a new genetic theory to explain some unusual aspects of the inheritance of glomus tumours. In their pedigrees this disease was transmitted almost exclusively by males, both affected and unaffected. Transmission in the maternal line always occurs, possibly with one exception, via unaffected males, and tumours never develop in children of affected mothers. Van der Mey et al suggest that glomus tumour formation is dependent on the mutational activation of a gene that exerts its tumour promoting effects in an autosomal dominant way. This gene is inactivated by genomic imprinting during female oogenesis thereby preventing tumour formation in descendants of affected females, and is reactivated by removal of the imprint in the male germline. Van der Mey et al assume that the imprinting process causes inactivation of the (activated) glomus tumour gene and no return to normal wildtype activity. This assumption implies that the inactivated gene should be functionally recessive to the wildtype gene with normal activity since glomus tumours never develop in individuals with an imprinted inactivated gene and a wildtype gene. These data might equally well be explained by assuming a two-hit model for glomus tumour development, as first proposed by Knudson,’ for the development of retinoblastoma, supplemented with the same assumptions with respect to the imprint status of male and female derived chromosomes. In this model a tumour suppressor gene is inactivated by mutation. This predisposing mutation is inherited in an apparent autosomal dominant way. In affected subjects the second tumour suppressor gene on the
homologous chromosome is already inactivated by genomic imprinting, thereby allowing the tumour to develop. Children of transmitting women, affected or not affected, always receive one active tumour suppressor gene from their father and no tumours will develop in these offspring. If the predisposing mutation in an unaffected male is transmitted to his offspring then a tumour will develop because the mother will always contribute an inactive suppressor gene as a result of the imprinting process in her germline. One assumption of this model is that inactivation of the second gene almost exclusively occurs in prezygotic imprinting and not by postzygotic loss or mutation, although the latter events could explain the single instance in which an affected female was found to transmit the disease to an unaffected child (pedigree N in table i). Up to now there is no direct evidence that tumour suppressor genes can be inactivated by genomic imprinting. However, it has lately been supposed to occur in some familial and sporadic cancers to explain the non-random nature of allele loss in tumours and the absence of linkage between the predisposing mutation and markers from within the presumptive tumour suppressor gene region.2,3 We conclude that this tumour suppressor gene model is, at the moment, equivalent to the dominant oncogene model proposed by van de Mey et al, and that this model should be considered in future research on glomus tumour inheritance. Institute of Human Genetics, Academic Medical Centre, University of Amsterdam, 1105 AZ Amsterdam, Netherlands
TH. J. M. HULSEBOS R. M. SLATER A. WESTERVELD
1. Knudson AG, Jr. Mutation and cancer: statistical study of retinoblastoma. Proc Natl Acad Sci USA 1971; 68: 820-23. 2. Scrable H, Cavanee W, et al. A model for embryonal rhabdomyosarcoma tumorigenesis that involves genome imprinting. Proc Natl Acad Sci USA 1989; 86: 7480-84. 3. Reik W. Genomic imprinting and genetic disorders in man. Trends Genet 1989; 5: 331-36.
Association between risk for and HLA DR4
pre-eclampsia
SIR,-Dr Kilpatrick and colleagues (Nov 4, p 1063) report results of family study showing an association between HLA DR4 and mild and proteinuric pre-eclampsia in a British (Edinburgh) maternal population. Among 76 parous sisters of women with proteinuric a
pre-eclampsia, they found that sisters with pregnancy-induced hypertension (pre-eclampsia with or without proteinuria) had a higher frequency of HLA DR4 antigen than did normotensive sisters. In addition, they cited unpublished findings in which they found a higher frequency of HLA DR4 antigen in a large sample of pre-eclamptic women and their babies than in appropriate controls. We have completed a study of HLA antigens and pregnancy outcome among a cohort of 715 black (50-9%) and white (49-1%) primigravida who were delivered at a medical centre in southern USA. HLA DR typing was done by the one-colour fluorescence technique with reagents obtained from One Lambda (Los Angeles, USA) and Biotest (Frankfurt, West Germany).1 On the basis of standard criteria for diagnosis of pre-eclampsia and eclampsia,2 6-9% of our cohort had mild non-proteinuric pre-eclampsia, 8-8% had pregnancy-induced hypertension, and 9-5% had combined pre-eclampsia and eclampsia. Whereas black women had higher rates than white women in all three clinical categories (eg, pregnancy-induced hypertension 10-7% vs 6-8%, respectively), differences were not significant and frequencies of HLA DR4 antigen were higher among normotensives in both races (results not shown). We therefore pooled the two racial groups for analyses. With the X2 test (without Yates’s corrections) we found that frequencies of HLA DR4 antigen were significantly higher in normotensive
women
(24-9%)
than in
women
with mild pre-
women with pregnancy-induced eclampsia (10-2%, p
The table shows that the ratio of neuroblastoma to Wilms’ low in the 1960s and was still low in Tanzania in 1980-81. In Uganda the ratio (0-3) was higher in 1947-66 than it was later in Kampala (0- 1) and in the West Nile district (zero). The earlier Uganda ratio of 0-3 is substantially below those elsewhere in the world (eg, the US and Manchester, England, 1-35). The ratios in Kenya and Zambia were available only for the late 1960s, and in Zaire (Kinshasa) for 1968-85; all were below 0-1. It would be a pity to pass over this evidence relating to the origins of neuroblastoma because of a contrary clinical impression. tumour was
Clinical Epidemiology Branch, National Cancer Institute, Bethesda, Maryland 20892, USA
ROBERT W. MILLER
DM, Stiller CA, Draper GL, et al, eds. International incidence of childhood (International Agency for Research on Cancer science publication no 87). Lyon: IARC, 1988. 2. Miller RW. Ethnic differences in cancer occurrence: genetic and environmental influences with special reference to neuroblastoma. Progr Cancer Res Therapy 1977; 3: 1-14. 3. Massabi M, Muaka BK, Tamba N. Epidemiology of childhood cancer in Zaire. Lancet 1989; ii: 501. 1. Parkin
cancer
Cerebral
palsy in infants born during trial intrapartum monitoring
of
SiR,—Your readers may have been seriously misled if they read the letter from Dr Hope and Dr Moorcraft (Jan 27, p 238) without reference to our report (Nov 25, p 1233). 16 (78%) of the 22 patients with cerebral palsy in the study had not shown any clinical signs suggestive of intrapartum asphyxia. The fact that the other 6 patients did show signs that could possibly reflect asphyxia does not necessarily mean that the signs were due to intrapartum asphyxia, nor should it be interpreted as meaning that the cerebral palsy was caused by intrapartum asphyxia. Furthermore, even if in some of these 6 patients cerebral palsy was caused by intrapartum asphyxia, it is very questionable whether it could have been prevented by different clinical management, in view of our current knowledge. National Perinatal Epidemiology Unit, Radcliffe Infirmary, Oxford OX2 6HE, UK
Inheritance of
ADRIAN GRANT
glomus
tumours
SIR,-Dr van der Mey and colleagues (Dec 2, p 1291) suggest a new genetic theory to explain some unusual aspects of the inheritance of glomus tumours. In their pedigrees this disease was transmitted almost exclusively by males, both affected and unaffected. Transmission in the maternal line always occurs, possibly with one exception, via unaffected males, and tumours never develop in children of affected mothers. Van der Mey et al suggest that glomus tumour formation is dependent on the mutational activation of a gene that exerts its tumour promoting effects in an autosomal dominant way. This gene is inactivated by genomic imprinting during female oogenesis thereby preventing tumour formation in descendants of affected females, and is reactivated by removal of the imprint in the male germline. Van der Mey et al assume that the imprinting process causes inactivation of the (activated) glomus tumour gene and no return to normal wildtype activity. This assumption implies that the inactivated gene should be functionally recessive to the wildtype gene with normal activity since glomus tumours never develop in individuals with an imprinted inactivated gene and a wildtype gene. These data might equally well be explained by assuming a two-hit model for glomus tumour development, as first proposed by Knudson,’ for the development of retinoblastoma, supplemented with the same assumptions with respect to the imprint status of male and female derived chromosomes. In this model a tumour suppressor gene is inactivated by mutation. This predisposing mutation is inherited in an apparent autosomal dominant way. In affected subjects the second tumour suppressor gene on the
homologous chromosome is already inactivated by genomic imprinting, thereby allowing the tumour to develop. Children of transmitting women, affected or not affected, always receive one active tumour suppressor gene from their father and no tumours will develop in these offspring. If the predisposing mutation in an unaffected male is transmitted to his offspring then a tumour will develop because the mother will always contribute an inactive suppressor gene as a result of the imprinting process in her germline. One assumption of this model is that inactivation of the second gene almost exclusively occurs in prezygotic imprinting and not by postzygotic loss or mutation, although the latter events could explain the single instance in which an affected female was found to transmit the disease to an unaffected child (pedigree N in table i). Up to now there is no direct evidence that tumour suppressor genes can be inactivated by genomic imprinting. However, it has lately been supposed to occur in some familial and sporadic cancers to explain the non-random nature of allele loss in tumours and the absence of linkage between the predisposing mutation and markers from within the presumptive tumour suppressor gene region.2,3 We conclude that this tumour suppressor gene model is, at the moment, equivalent to the dominant oncogene model proposed by van de Mey et al, and that this model should be considered in future research on glomus tumour inheritance. Institute of Human Genetics, Academic Medical Centre, University of Amsterdam, 1105 AZ Amsterdam, Netherlands
TH. J. M. HULSEBOS R. M. SLATER A. WESTERVELD
1. Knudson AG, Jr. Mutation and cancer: statistical study of retinoblastoma. Proc Natl Acad Sci USA 1971; 68: 820-23. 2. Scrable H, Cavanee W, et al. A model for embryonal rhabdomyosarcoma tumorigenesis that involves genome imprinting. Proc Natl Acad Sci USA 1989; 86: 7480-84. 3. Reik W. Genomic imprinting and genetic disorders in man. Trends Genet 1989; 5: 331-36.
Association between risk for and HLA DR4
pre-eclampsia
SIR,-Dr Kilpatrick and colleagues (Nov 4, p 1063) report results of family study showing an association between HLA DR4 and mild and proteinuric pre-eclampsia in a British (Edinburgh) maternal population. Among 76 parous sisters of women with proteinuric a
pre-eclampsia, they found that sisters with pregnancy-induced hypertension (pre-eclampsia with or without proteinuria) had a higher frequency of HLA DR4 antigen than did normotensive sisters. In addition, they cited unpublished findings in which they found a higher frequency of HLA DR4 antigen in a large sample of pre-eclamptic women and their babies than in appropriate controls. We have completed a study of HLA antigens and pregnancy outcome among a cohort of 715 black (50-9%) and white (49-1%) primigravida who were delivered at a medical centre in southern USA. HLA DR typing was done by the one-colour fluorescence technique with reagents obtained from One Lambda (Los Angeles, USA) and Biotest (Frankfurt, West Germany).1 On the basis of standard criteria for diagnosis of pre-eclampsia and eclampsia,2 6-9% of our cohort had mild non-proteinuric pre-eclampsia, 8-8% had pregnancy-induced hypertension, and 9-5% had combined pre-eclampsia and eclampsia. Whereas black women had higher rates than white women in all three clinical categories (eg, pregnancy-induced hypertension 10-7% vs 6-8%, respectively), differences were not significant and frequencies of HLA DR4 antigen were higher among normotensives in both races (results not shown). We therefore pooled the two racial groups for analyses. With the X2 test (without Yates’s corrections) we found that frequencies of HLA DR4 antigen were significantly higher in normotensive
women
(24-9%)
than in
women
with mild pre-
women with pregnancy-induced eclampsia (10-2%, p
