Parental cancer as a risk factor for nine common childhood malignancies

The nationwide Swedish Family-Cancer Database was used to analyse childhood tumours among 8158 offspring by parental cancers. The results showed 2-fold familial increases for nervous system cancers and lymphomas, a 6.4-fold increase for endocrine tumours, a 60-fold increased risk for retinoblastomas but no excess risk for leukaemia and Wilms tumour. © 2001 Cancer Research Campaign http://www.bjcancer.com

were retrieved from the nationwide Swedish Cancer Registry from years 1958 to 1996. A 4-digit diagnostic code according to the 7th revision of the International Classification of Diseases (ICD-7) was used with 3 or 4 digits; for retinoblastoma and Wilms tumour, pathology codes were also used for classification.
The analysis was carried out on all types of malignancies diagnosed before age 15 between years 1958 and 1996. Explanatory variables included in the statistical analysis were parental cancer status, socio-economic status (SES, 4-category variable: agriculture, professional, worker, other) and area of living (county, five-category variable: Stockholm area, the largest city; Gothenburg-Malmö area, two largest cities in south of Sweden; Götaland, Svealand and Norrland, three geographic regions, from south to north, respectively). The SES and area variables were extracted from the national census 1960 of Statistics Sweden. All analysis also included two variables: age at diagnosis (five-year categories) and year of birth (birth cohort, three categories: 1941-1950, 1951-1960, 1961-1996). Standardized incidence ratios (SIRs) were calculated as the ratio of observed (O) to expected (E) number of cases. The expected numbers were calculated from 5-year-age-and tumour type-specific standard incidence rates (Esteve et al, 1994). Confidence intervals (95% CI) were calculated assuming a Poisson distribution (Esteve et al, 1994).
The Family-Cancer Database covered years 1958 to 1996 from the Swedish Cancer Registry and included 8158 cases of childhood tumours, diagnosed before age 15 (Table 1). The most common types were nervous system tumours (2486), leukaemias (2397) and lymphomas (807).
SIRs for the 9 most common childhood tumours by parental cancer are shown in Table 2. When neither parent had cancer the SIRs ranged from 0.90 to 1.01. When either father or mother had any cancer the SIRs ranged from 0.92 or 0.93 (leukaemia and Wilms tumour, respectively) to 1.22 for lymphoma and 1.44 for retinoblastoma, which were significant. When both parents had any cancer, the SIR for lymphoma was 1.93 and that for endocrine gland tumour 4.84, both statistically significant. In these families, the parental cancer types were heterogeneous and did not reveal a special pattern.

Parental cancer as a risk factor for nine common childhood malignancies
SIRs of childhood cancers by parental nervous system and endocrine gland tumour are shown in Table 3. For nervous system cancer, 28 child-parent pairs were identified, giving a SIR of 2.03. Most tumours (25/28) in offspring were located in the brain. In one pair, even the parental tumour (ependymoma) was diagnosed in childhood. In another pair, both the offspring and the parent had cerebral haemangioblastoma. In 7 pairs, both the offspring and the parent presented with cerebral astrocytoma; in 5 pairs, offspring presented with cerebral astrocytoma, the parents with a cerebral meningioma; in three pairs, offspring presented with cerebral astrocytoma, the parent with peripheral or spinal neurinoma. The remaining pairs were mixed single histologies. Offspring endocrine gland tumours were associated with parental nervous system tumours, giving a SIR of 5.67. Of these, three were adrenal ade1nocarcinomas (offspring) -cerebral astrocytomas (parent), one pair was adrenal ganglioneurinoma -brain neurinoma, and one pair was pancreatic adenoma-brain meningioma.
Parental endocrine gland tumours were associated with thyroid and other endocrine gland tumours in offspring. The thyroid (offspring)-endocrine (parent) pairs were one medullary thyroid cancer-adrenal paraganglioma, one non-medullary thyroid cancer-adrenal adenocarcinoma, and one undifferentiated thyroid cancer-adrenal paraganglioma.
Associations of childhood cancers with parental leukaemia, lymphomas, kidney cancers and retinoblastomas are shown in Table 4. Parental leukaemia and kidney cancer associated with no Among the present childhood malignancies, retinoblastoma and Wilms tumour are well-defined clinical entities. The main cause of retinoblastoma is mutation in the RB1 gene. About 40% of retinoblastomas are hereditary and a quarter of these are new germinal mutations (Lindor et al, 1998). Practically all the hereditary cases are diagnosed in the first years of life, in agreement with our data. The SIR for retinoblastoma, 57.97, was the highest found in the present study. RB1 mutations predispose also to other tumours, such as osteosarcoma. In our material there were too few cases to confirm the relationship. For Wilms tumour, at least three gene loci, WT1, WT2 and WT3 have been reported (Lindor et al, 1998). Less that 1% of Wilms tumour is hereditary and we found no indication of increase in childhood Wilms tumours by parental cancer. Childhood brain cancer is encompassed in many rare cancer syndromes of high risk (Draper et al, 1996), but hereditary effects have been ascribed only to some 2-4% of brain cancers (Bondy et al, 1991;Hemminki et al, 2000c;Narod et al, 1991). In populationbased studies, the familial risk of brain tumours was 2.5, and in offspring of survivors of childhood brain tumours it was 2.0 (Sankila et al, 1998;Hemminki et al, 2000c). However, a study on cancer in parents of childhood cancer probands found no increase in the risk of nervous system cancer between the two generations (Olsen et al, 1995). In a previous study on childhood brain cancer we noted a risk of 10.26 in childhood astrocytoma when a parent had meningioma (Hemminki et al, 2000c). Among adult brain tumours, we have reported a familial risk of 1.70 and another study from Sweden reported an aggregation of adult astrocytomas (Malmer et al, 1999;Hemminki et al, 2000a). Our present findings were in line with the above reports, because 25 of the 28 familial nervous system cancers were brain tumours. Li-Fraumeni syndrome features the occurrence of diverse gliomas (Sedlacek et al, 1998), and it is possible that some of the astrocytoma aggregates were due to this syndrome. The astrocytoma-meningioma combination may be due to neurofibromatosis, particularly of type 2 (Huson, 1998). One child-parent pair presented with brain haemangioblastoma, pathognomonic of von Hippel-Lindau syndrome (Lindor et al, 1998;Hemminki et al, 2000b).
Endocrine gland tumours manifest in several known cancer syndromes. The aggregation of endocrine gland tumours, particularly parathyroid adenomas, among children and parents suggests involvement of multiple endocrine neoplasia 1. In one pair, adrenal paraganglioma in offspring was associated with parental medullary thyroid tumour, signalling the presence of multiple endocrine neoplasia 2 (Hemminki and Dong, 2000). Offspring endocrine tumours were associated with parental nervous system tumours, but not vice versa. The aggregation of nervous system and endocrine tumours is common to neurofibromatosis 1 (Lindor et al, 1998).
Among haematopoietic malignancies, only lymphoma showed familial aggregation. The SIR of 2.19 in children was higher than the risk (1.5) previously found for the adult population from the Database (Hemminki et al, 1998). On the other hand, leukaemia showed no familial effect although such is present in the adult population, showing a relative risk of 1.8 (Hemminki et al, 1998). The common childhood leukaemia, acute lymphocytic leukaemia is uncommon among adults, perhaps explaining the absence of a familial effect.
The present data showed about 2-fold familial effects in nervous system tumours and lymphomas, a 6.4-fold risk in endocrine tumours and close to a 60-fold risk in retinoblastoma. No familial effect was observed in leukaemia and Wilms tumour. Aggregation of nervous system and endocrine gland tumours was noted.