HIV-related malignancies in children 

Pediatric infection with HIV is a growing problem in the United States. HIV-related malignancies and their clinical course are well described in adults, but little is known about HIV-related cancer in children. Granovsky et al [J Clin Oncol 16:1729-35, 1998] surveyed the Childrenís Cancer Group and NCI cases of childhood cancer that occurred in HIV-positive patients between July 1982 and February 1997. They included all children who were seropositive for HIV either prior to or at diagnosis of a malignancy. A total of 64 HIV infected children with 65 tumors were reported during the time period. Non-Hodgkinís lymphoma (NHL) comprised 65% of all reported malignancies, followed by leiomyosarcomas/leiomyomas (17%), ALL (6%), Kaposiís sarcoma (5%), and Hodgkinís disease (3%). There was one case each of AML, vaginal carcinoma in situ, and tracheal carcinoma. For NHL and the leiomyosarcomas, the clinical characteristics and outcomes were very similar to these same malignancies in HIV-infected adults; the median survival time for NHL and leiomyosarcoma was 6 months and 12 months, respectively.

COMMENT: As in adults, the incidence of HIV-related cancer in children will increase as duration of survival with HIV increases (due to improved use of antiretroviral and supportive therapies). This study could be added to the educational tools used for public health AIDS awareness campaigns to emphasize the devastating consequences of HIV infection. These data should also remind physicians to test children with these malignancies for HIV if risk factors are present. Julie A. Ross

Wilms' in the family?

Familial Wilms' tumor is well recognized (with numerous pedigrees reported) and is inherited as an autosomal dominant trait. In most families the genomic regions 11p13 and 11p15 known to be linked to Wilmsí tumor in other contexts, can be ruled out as the location of the familial predisposition gene. A new report [McDonald et al, Cancer Res, 58:1387-1390, 1998] describes a genome-wide linkage scan in five families with familial Wilmsí tumor and provides evidence femonstrate that each twin shared the same clonotypic TEL-AML1 rearrangement in their leukemic cells. This translocation was not present in normal cells. These data suggest that the TEL-AML1 rearrangement was acquired during fetal hematopoiesis in utero in one twin and transferred to the other twin through placental circulation. These results are similar to what the authors have previously reported regarding MLL-AF4 rearrangemer family studies have indicated linkage at 17q in one large family (Rahman et al, Nature Genetics, 13, 461-462, 1996) but identification of the gene at this locus has not yet been reported. While linkage studies are valuable, identification of the relevant gene can still be a major undertaking, with the time needed depending on the clinical material available, the nature of the gene, and to some extent, good luck. Stella M Davies.

There is more to TEL!

One of the most frequent translocations found in childhood ALL is t(12;21), which fuses the TEL gene with the AML1 gene (see C3 Vol 7 No 2). Approximately 25% of ALL cases diagnosed between the ages of 2 and 10 years have this translocation. Ford et al [Proc Natl Acad Sci 95:4584-4588, 1998] recently studied a pair of monozygotic twins who were diagnosed with ALL at 3 years 6 months and 4 years 10 months, respectively. Using a series of molecular methods, the authors were able to demonstrate that each twin shared the same clonotypic TEL-AML1 rearrangement in their leukemic cells. This translocation was not present in normal cells. These data suggest that the TEL-AML1 rearrangement was acquired during fetal hematopoiesis in utero in one twin and transferred to the other twin through placental circulation. These results are similar to what the authors have previously reported regarding MLL-AF4 rearrangements in infant leukemia (see C3 vol 9 no 1).

COMMENT: These are elegant experiments that take full advantage of an infrequent occurrence, concordant leukemia in older twins. These studies are extremely important in understanding etiology, since they suggest focusing on events that have occurred during pregnancy, even in patients presenting with leukemia beyond infancy. With the findings reported here, as well as those for infants, we may be just beginning to understand the true impact of the fetal origins of leukemia. Julie A.Ross

To everything there is a season

Several studies have examined seasonal variations in the detection or diagnosis of the cancer in an attempt to explore whether an infectious etiology may be important in the development of specific childhood malignancies,. A recent study by Westerbeek et al. [Br J Can 78:119-24,1998] investigated seasonal variations in months of first symptom and months of diagnosis from data collected by The Manchester Childrenís Tumour Registry. Data from a total of 1,669 children younger than 14 diagnosed with leukemia or lymphoma between 1954 and 1996 were analyzed. Edwardís test was used to detect the presence of a sinusoidal curve within a twelve month period. Seasonality was explored for the date of first symptom, which was recorded by the physician and defined as the date the patient was last known or thought to be well and the date of diagnosis. A peak in incidence was found for the date of first symptom for common acute lymphoblastic leukemia (c-ALL) in November with an amplitude of 21.1% (p=0.037). Further, a peak in the date of first symptom was detected in December for Hodgkinís Disease (HD) with an amplitude of 41.0% (p=0.01). No corresponding peaks were found for the date of diagnosis for either c-ALL or HD. Additionally, no statistically significant peaks in date of first symptom or diagnosis were found for any acute non-lymphoblastic leukemia, infant leukemia, or Non-Hodgkinís lymphoma. 

COMMENT: Studies that have explored seasonal variations in the months of birth or diagnosis of childhood cancer have been inconsistent. Since seasonality is characteristic of certain viruses, it is believed that seasonal observations in the diagnosis of certain childhood cancers could support a viral hypothesis. This study had the advantage of exploring the pattern of the month of first symptom, which may be more etiologically relevant. Interestingly, statistically significant peaks were detected for the date of symptom for c-ALL and HD-- both of which have been hypothesized to be linked with an infectious etiology. However, as the date of first symptom was only loosely defined and no corresponding peak was detected for month of diagnosis for either c-ALL or HD, it is difficult to interpret these findings. Although this study provides modest evidence that an infectious agent may be implicated in the development of c-ALL and HD, a definitive conclusion can only be reached when that agent is identified. Andrine R. Swensen

Rhabdoid tumors: Switching genes on and off

Malignant rhabdoid tumors are extremely aggressive cancers of early childhood. They occur in various locations, mainly the kidney, brain, and soft tissues. Cytogenetic and molecular analyses have shown that the deletion of region 11.2 of the long arm of chromosome 22 (22q11.2) is a recurrent genetic characteristic of malignant rhabdoid tumors, indicating that this locus may encode a tumor suppressor gene. Versteege et al [Nature 394:203206, 1998] have identified a gene deleted in this region to be the hSNF5/IINI-1 gene. This gene encodes a member of the chromatin remodeling SWI/SNF multiprotein complexes. The data they present suggest that loss of function mutations within this gene contributes to oncogenesis in rhabdoid tumors. Studies were performed both in tissue cell culture lines and in primary tumors, eliminating the possibility that this is an artifact of growth in tissue culture.

COMMENT: The gene identified in this study is a member of a family of genes that appear to switch on inducible genes by altering the structure of chromatin. This is a similar activity to the histone deacetylase activity recently described for a number of genes involved in childhood cancer; e.g., the retinoblastoma gene and the partner genes fused with the retinoic acid receptor in APL (t(15;17)). [C3 Vol 9, No. 2]. Understanding the way genes are switched on and off is essential to understanding cancer. The uncontrolled proliferation of a malignancy is a failure of either the ëoffí switch for cell proliferation or the ëoní switch for cell death (apoptosis). Further studies are likely to identify additional genes in this class. Understanding chromatin configuration and its importance in DNA regulation and gene expression is likely to be a major area of interest for the next few years. Stella M. Davies

Undercover: electric blankets or down comforters?

As reported in several past issues of C3 (Vol 8, No 4; Vol 8, No 3), the association between electromagnetic field exposure (EMF) and childhood leukemia is quite controversial. Although EMF exposure through power lines has been studied in great detail, only a few small case-control studies have explored EMF exposure through small household appliances. In this report, Hatch et al evaluated the use of electric appliances by both the mother (during pregnancy) and the child [Epidemiology 9:234-245, 1998]. Data from 640 children under the age of 15 diagnosed with acute lymphoblastic leukemia (ALL) in nine states in the US between 1989 and 1993 and 640 matched control children were available for analyses. There was a statistically significant positive association with maternal use of an electric blanket during pregnancy (OR=1.59, 95% CI=1.11-2.29), while there was an inverse association with sewing machine use (OR=0.76, 95% CI=0.59-0.98). For children, there was a statistically significant positive association with electric blanket use (ever versus never, OR=2.75, 95% CI=1.52-4.98), although there was no obvious trend with increasing usage. There was also a positive association with use of hair dryers, video machines in arcades, and video games connected to television. Finally, increased time spent watching television was associated with an increased risk, although the risk did not vary by usual distance from the television. The authors conclude that due to the inconsistencies in the dose-response relationships, as well as inconsistencies in reporting, the associations reported here do not support a causal relation between magnetic field exposure from appliances and childhood ALL. 

COMMENT: The authors came to an appropriate conclusion; these data do not provide evidence of an association between specific electric appliance use and risk of childhood ALL. Moreover, the authors raise an interesting question regarding recall bias in childhood cancer studies. Most case-control studies rely heavily on recall of past exposures. However, even if the mother of a case child is instructed to recall events prior to her child's diagnosis, it is possible that her responses could reflect changes in a child's behavior after diagnosis. For example, perhaps the child has spent more time watching television after diagnosis (compared with before) due to illness. Current studies are attempting to address these important issues and improve methods for exposure measurement. Julie A. Ross

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