Growth Hormone: Risks as well as benefits?

Growth hormone, derived from human pituitary gland, was initially used to treat idiopathic primary growth hormone deficiency in children.  However, as treatment for childhood cancer has progressed and survivorship has increased, it is being used more widely in childhood cancer survivors with secondary growth hormone deficiency. Synthetic growth hormone became available in 1985, removing the risk of transmission of Jakob-Creutzfeld disease and broadening its application.  Growth hormone raises serum concentrations of insulin-like growth factor (IGF-1), which is mitogenic and anti-apoptotic, and results from in vitro and animal studies suggest that growth hormone might raise the risk of hyperplasia and malignancy. A. Swerdlow, et al. [Lancet 360: 273-7, 2002] performed a cohort study looking at cancer incidence and mortality in 1,848 patients in the UK who were treated during childhood and early adulthood with growth hormone during the period 1959-1985 (prior to the introduction of synthetic product).  Fifty-three percent of cases were being treated for idiopathic growth hormone deficiency,  21% had had a prior brain tumor; only 22 (1%) cases had prior leukemia or lymphoma. Patients were followed up for cancer incidence until December 1995, and for mortality to December 2000. Risk of cancer in the cohort was compared with that in the general population, controlling for age, sex and calendar period.  The study found that patients treated with human pituitary growth hormone had significantly raised risks of mortality from cancer overall (SMR 2.8, 95% CI=1.3-5.1; 10 cases), colorectal cancer (10.8, 1.3-38.8; 2 cases), and Hodgkin disease (11.4, 1.4-41.3; 2 cases).  Incidence of colorectal cancer was also greatly raised (7.9, 1.0-28.7).  After exclusion of patients whose original diagnosis rendered them at high risk of cancer (chromosome fragility syndromes, prior malignancy), the risks of colorectal cancer incidence and mortality and of Hodgkin disease mortality were increased.  In this study, 3 cancers were reported in persons with prior malignancy, one of which was a bone tumor in a radiation field.  Of note, the 2 deaths from colorectal cancer and the 2 deaths from Hodgkin Disease all occurred in individuals treated for idiopathic growth hormone deficiency.

In a separate study, C. Sklar et al. [J. Clin. Endocrin. 
Metab., 87:3136-3141, 2002] studied 361 growth hormone-treated cancer survivors enrolled in the Childhood Cancer Survivor Study. Importantly, there was no evidence of an increased risk of relapse of the primary malignancy in these cases (RR 0.83, 95% CI 0.37-1.86, p=0.65). The risk of second neoplasm (n=15) was elevated in growth hormone-treated survivors (RR 3.21, 95% CI, 1.88-5.46, p < 0.0001), largely due to an increase in second malignancy in four children with leukemia (2 developed osteogenic sarcoma, 2 brain tumors; 3 of these 4 children are known to have received radiation).

COMMENT: In the past, concern about cancer in patients treated with growth hormone was focused primarily on the risk of relapse or initiation of leukemia, perhaps because available data have largely been for children in whom leukemia is the most common cancer.  No leukemia has occurred as a second malignancy in either cohort.  This finding adds to the evidence from other cohort studies that suggest that leukemia risk is not substantially raised in children without other reasons for cancer susceptibility (eg Fanconi anemia).  In the discussion of both these papers, the authors caution that one should not over-interpret risk based on such small numbers. Note that the number of actual cases of malignancy is small in each area. While numbers are small, the increased frequency of colorectal cancer in the Swerdlow study is of some concern as there is biological plausibility for the role of IGF-1 in the occurrence of colorectal cancer.  While further monitoring will be necessary to determine whether this risk is real, these two papers are reassuring that overall growth hormone therapy is safe for childhood cancer survivors. 

BRCA2 is a Fanconi Anemia Gene

Fanconi anemia (FA) is a genetically complex disorder, with at least 8 different possible genes involved in the phenotype.  Apart from the genes B and D1, these have all been isolated and encode diverse proteins, many of which interact within a cellular complex related to the maintenance of genomic stability.  In a new study, NG Howlett, et al [Science, 297, 606, 2002] show that one of the previously unidentified FA genes, FANCD1, is BRCA2. Heterozygous carriers of BRCA2 mutations inherit a high risk of breast cancer (up to 85%) and other cancers such as ovary and pancreas. Such individuals do not have malformations at other sites.  In the FA study, biallelic BRCA2 mutations were shown in 2 FANCD1 patients.  In an FA-B patient there was a mutation of one BRCA2 allele, and a polymorphic stop variant of the second allele on another patient.  In addition, biallelic mutations were shown in two FA patients in whom the complementation group was unassigned.  These data are quite striking. Biallelic mutation of BRCA2 has not been described in humans before and was thought to be lethal in the embryo, as has been seen in double knockout mice.  It appears, from data in mice and the findings of Howlett et al., that mutations in BRCA2 that involve the 3’ region of the gene and allow expression of some truncated protein are compatible with survival and an FA phenotype.  Mice with these “mild” or hypomorphic mutations appear able to survive for some time and have phenotypes reminiscent of FA including small gonads, skeletal defect, and sensitivity to DNA crosslinking agents.  It is, therefore, possible that many of these so-called hypomorphic BRCA2 mutations will be found in FA patients but that more severe mutations are lethal in embryogenesis and do not occur in humans. Of note, FA has an estimated incidence of less than 1 per 100,000 live births. Less than 5% of FA families are assigned to complementation groups B and D1 and  BRCA2 mutations have a cumulative carrier frequency of approximately 1% of the U.S. population. This BRCA2 carrier frequency predicts a higher incidence of BRCA2 homozygotes than the observed FA incidence, supporting the likelihood that only a subset of BRCA2 biallelic mutant individuals can survive.  It will be fascinating in the future to determine the cancer risk in this small subset of cases, and the relationship with malignancy at sites other than the traditional FA-associated malignancies.

COMMENT:  This paper shows a further elucidation of the complex biochemical pathway involved in FA. A helpful editorial by E. Witt and A. Ashworth [Science 297:534, 2002] shows a model of the interactions between the DNA repair proteins implicated in FA and breast cancer susceptibility.  The FA proteins are thought to be present in the nucleus of the cell as a complex that responds to detection of DNA damage by the ubquitination of a protein FANCD2.  This protein can also be ubquitinated by the ATM gene (mutated in ataxia telangiectasia gene), and interacts itself with the breast cancer susceptibility gene BRCA1.  Taken together, these data suggest a close interaction between DNA damage response genes and breast cancer susceptibility genes, and will stimulate new interest in cancer susceptibility in FA heterozygotes. 

GPA mutations and XRCC1 genotype in cord blood

Several mutational assays are available to measure genomic instability, which is a hallmark of genotoxic exposure. One such measure, the glycophorin A (GPA) somatic mutation assay, uses flow cytometry to identify two classes of variant red blood cells, one of which results from gene activation or gene loss (NO) and the other of which arises through chromosomal recombination or mis-segregation (NN). Approximately 50% of the population is informative for the assay. The GPA assay has been used in high risk groups to monitor the biological effects of atomic bomb radiation as well as high dose chemotherapy[WL Bigbee et al, Mutat Res 240:165-175, 1990]. Recently the GPA assay has been used as a biodosimeter in population studies to detect subtle exposures such as cigarette smoking and occupational and household exposures to pesticides and solvents [SG Grant & WL Bigbee Mutat Res 299:163-172, 1993; Bigbee WL et al, Cancer Epi Biom Prev 5:801-810, 1996]. Approximately 10% of informative individuals in ‘healthy’ populations manifest a significantly higher level of N0 and/or NN variant frequencies (Vfs). While it is unclear which exposure(s) might increase the frequency of GPA mutations, there has been some recent intriguing work evaluating the potential relationship between GPA Vfs and polymorphisms in DNA repair genes. The human DNA repair gene, XRCC1, is involved in the repair of DNA single strand breaks generated in response to either ionizing radiation or alkylating agents. RM Lunn et al. [Cancer Res 59:2557-61, 1999] explored whether hereditary common genetic polymorphisms in XRCC1 were associated with an impact on GPA mutations. A total of 59 individuals informative for the GPA assay, including 49 smokers and 10 non-smokers, were genotyped for three XRCC1 poly-morphisms. Only one of the genotypes, Arg399Gln, was significantly associated with a higher level of GPA NN Vfs. Individuals who smoked and had at least one Gln allele had substantially higher NN Vfs compared to non-smokers. In a new study by CL Relton et al, [Mutation Res 502:61-68, 2002], cord blood from 189 infants informative for the GPA assay was genotyped for two polymorphisms in the XRCC1 gene; a microsatellite repeat region in the 3’ un-translated region of the gene (3’UTR) and the Arg399Gln single nucleotide polymorphism. In this study, 7.4%(n=14) of the cord bloods had Vfs of the NN allele that fell into the extreme category (40-1487 X 10 –6), well-outside of the majority of the population (n=175) (N0 mean (4.8. +- 2.8) X 10 –6; NN mean 2.62 +- 2.0) X 10 –6). Seven different alleles were identified in a polymorphic tandem [AC]n region in the 3’UTR region of the XRCC1 gene. There was no association between the distribution of the [AC]n genotypes and either N0 or NN Vfs among the 175 cord bloods or among the 14 with extreme NN Vfs. However, for the XRCC1 Arg399Gln genotype, there was a much higher frequency of Gln/Gln among the 14 infants who had extreme NN Vfs compared to infants with normal GPA NN Vfs (42.8% versus 14.3%, respectively, p < 0.001). The authors suggest that carriers of the Gln/Gln genotypes are over-represented in the group of individuals with higher NN Vfs. 

COMMENT: This recent work corroborates the earlier work by Lunn et al, and provides further evidence that the Arg399Gln polymorphism in XRCC1 may play an important role in DNA repair, particularly as it relates to recombination and segregation events. Studies of biological markers of effect (GPA mutations) and potential interactions with specific genetic polymorphisms in cord bloods are extremely important in advancing our understanding of possible mechanisms that may be important in childhood malignancy. We anxiously await results of further research in this area. 

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