Correlations among tumor types in mouse cancer bioassays: liver adenomas, liver carcinomas, leukemias and lymphomas

Abstract

In an examination of rodent bioassays Young and Gries [Young S.S., and Gries C.L. Exploration of the negative correlation between proliferative hepatocellular lesions and lymphoma in rats and mice—establishment and implications. Fundam. Appl. Toxicol. 1984: 4: 632-640] and Haseman et al. [Haseman J.K., et al. Body weight-tumor incidence correlations in long-term rodent carcinogenicity studies. Toxicol. Pathol. 1997: 25: 256-263] noticed that there is a negative correlation (anticorrelation) between development of liver tumors and leukemia or lymphomas. If an animal has a lymphoma or leukemia it is less likely to develop liver tumors. These studies noted that this applies to several strains of animals. The anticorrelation appeared in control animals. In this paper we study this anticorrelation in the quarter of a million rodents exposed in the Carcinogenesis Bioassay Database System (CBDS) database of the National Toxicology program in both control and dosed animals. We failed to completely replicate Young and Gries or Haseman et al. However, when benign liver tumors (adenomas or nodules) and malignant liver tumors (carcinomas) are considered separately AND different leukemia and lymphoma types are considered, a strong anticorrelation appears. We identify survival and the time period (in years) during which the bioassay was completed as an important factor in interpreting correlations and anticorrelations. Differences between liver adenomas and carcinomas and lymphoma types contribute to a more general question of grouping of the individual tumor types. In our classification scheme (originally developed about 1986 by Dr. Bailar) an effort is made to distinguish benign and malignant neoplasms, while other investigators group all tumors at a specific site. For many analyses liver adenomas and carcinomas have been lumped together. This is because it is suspected that the diagnoses by pathologists may not distinguish, or ignore the (benign) adenoma when a (malignant) carcinoma is present. Possible biological differences in tumor mechanisms may require separate evaluation of these tumors with all the dangers of “pathologist bias” that this introduces.