HLA and ALL- unconvinced!

A recent study performed by Dorick et al. [Blood 94:694-700, 1999] examined the association of HLA-DR genotype with childhood acute lymphoblastic leukemia (ALL) in 114 patients treated at a single center. The authors reported an association between HLA-DRB1*04, which was apparently stronger in males. They also showed an increase in homozygosity for HLA-DRB4*01, which was entirely due to an increase in homozygosity in males.The authors state that this highly significant result provides the long suspected evidence of the HLA-DR influence in the development of childhood ALL. 

COMMENT: Interest in acute leukemia and HLA type has arisen sporadically since the report by Lilly et al. in 1964 [Lancet 2:1207] describing an increased risk of  spontane-ous and virus-induced leukemia in mice homozygous for the H2K haplotype. Many studies have been performed in this area and results have been equivocal. Serological studies have indicated an association between childhood ALL and HLA-A2 [Tiwari et al, HLA and Disease Associations, New York, NY: Springer-Verlag, 1985]. Another study has shown an association with the HLA class II supertype DR53 (a supertype is a broad functional categorization of HLA type). One of the difficulties affecting all studies of HLA polymorphism and risk of any cancer is the important issue of  multiple comparisons.  If one selects a p-value of 0.05 as significant and makes 20 comparisons one would expect one significant association purely by chance. As the number of comparisons increases, the number of spurious results  will increase. The large number of potential haplotypes and potential ways of grouping them as alleles, as supertypes and as serological specificities in HLA studies, encourages multiple comparisons and subgroup analyses.  In this paper, the authors report that the strength of the association (based on the magnitude of the p-value of the association) is such that it is unlikely that this is a spurious association due to multiple comparisons, and they opted not to apply the usual statistical safeguards.  It is important to note that the magnitude (or lack thereof) of the p-value is not a guide to the strength of the association The signficance of an association can only be properly assessed with knowledge of the number of comparisons made. The variability in findings of studies of HLA and disease over the years in different leukemias leave one concerned that frequently the data being reported might represent statistical background noise. Clear presentation of the number of comparisons made will help to avoid this pitfall. Stella M. Davies

Editors’ Note: New information sources

There are three new published sources of information concerning childhood cancer of interest  to our readers.

1. The much anticipated second volume (first volume was published in 1988) of the International Incidence of Childhood Cancer is now available. (International Agency for Research on Cancer, IARC Scientific Publication Number 144, editors: DM Parkin, E, Kramarova, GJ Draper, E Masuyer, J Michaelis, J Neglia, S Qureshi, CA Stiller; Lyon, France, 1998). This resource provides age-, sex-, (race-, in some instances), and diagnosis-specific incidence rates for over 50 countries (and within countries) using cancer registries primarily for the period 1980-1990. 

2. Julian Little has assembled a very excellent and thorough review (at least 1400 references) of the Epidemiology of Childhood Cancer (IARC Scientific Publications, Number 149, Lyon, France, 1999). Topics covered include: current hypotheses, descriptive epidemiology (time trends, socioeconomic status, etc), incidence patterns and trends of all the major groups of childhood cancer, genetic factors, and finally, exposure-specific topics such as ionizing radiation, electric and magenetic fields, chemicals, maternal exposures, etc. The tables are particularly impressive.

3. The new NCI-sponsored childhood cancer monograph will be published in November 1999. This is an excellent resource, providing information regarding incidence, trends, and risk factors for childhood cancer in the United States. [LAG Ries, MA Smith, JG Gurney, M Linet, T Tamra, JL Young, GR Bunin (eds), Cancer Incidence and Survival Among Children and Adolescents: United States SEER Program 1975-1995, National Cancer Institute, SEER Program. NIH Pub. No. 99-4649, Bethesda, MD, 1999]. The full monograph is web accessible at http://www-seer.ims.nci.nih.gov/Publications/PedMono

As part of the monograph, individuals from the Epidemiology and Cancer Control Strategy Group, of the NCI-supported Children’s Cancer Group, provided summaries of selected risk factors for various childhood cancers (similar to the ‘State of the Epidemiology Report’ we previously provided for childhood leukemia [see C3 vol 5, No 4]. Selected  tables from the monograph will be printed in the next few issues of C3. In this issue, overleaf we highlight osteosarcoma and Ewings’ sarcoma. Julie A. Ross
 

STATE OF THE EPIDEMIOLOGY REPORT: OSTEOSARCOMA, EWINGS’ SARCOMA

From: JG Gurney, AR Swensen, M Bulterys, Malignant Bone Tumors,  in: LAG Ries, MA Smith, JG Gurney, M Linet, T Tamra, JL Young, GR Bunin (eds), Cancer Incidence and Survival Among Children and Adolescents: United States SEER Program 1975-1995, National Cancer Institute, SEER Program. NIH Pub. No. 99-4649, Bethesda, MD, 1999. Below is a compilation of selected risk factors reported to be associated with either osteosarcoma or Ewings’ sarcoma. For ‘known risk factors’,  there is compelling evidence that the factor is strongly associated with the cancer (does not mean that all individuals with the factor develop cancer)  Specific citations are available at: http://www-seer.ims.nci.nih.gov/Publications/PedMono. 
 
 

Selected Risk Factors for Osteosarcoma

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