The goals of my research are: (1) to correlate mutation profiles and exposures in cancers associated with cancer therapy, diagnostics or hormonal therapy; (2) to generate mechanistic models of mutagenesis and pathogenesis; and (3) to identify genetic susceptibility factors based on mechanistic models.
The price of success in cancer treatment is long-term survivors who develop second malignancies (termed iatrogenic, or treatment-related) from the genotoxic effects of cancer therapy. In addition, some hormonal regimens cause liver and endometrial cancers. Prominent examples include lung cancer following radiation therapy for Hodgkin's disease, hepatic angiosarcoma and lung cancer following Thorotrast (thorium dioxide) administration for radiographic contrast, bladder cancer following cyclophosphamide therapy for non-Hodgkin's lymphoma, endometrial cancer following tamoxifen medication for breast cancer, and liver cancer following oral contraceptive use.
We have completed or started mutation spectrum studies of the p53 tumor suppressor gene in all five examples. Our series of lung cancers following radiotherapy for Hodgkin's disease contained an excess of G:C > A:T transitions at non-CpG sites. This result was unexpected because all patients were heavy smokers and most series have found a predominance of G:C > T:A transversions in smoking-associated lung cancers.
Based on these data, we suggest that oxidative damage secondary to radiolysis of water produced most of the p53 mutations in these cancers. If confirmed in a larger series, then we may challenge the dogma that radiation-associated tumors cannot be distinguished from those caused by other factors. Hepatic angiosarcomas and lung cancers linked to thorotrast administration are additional examples of radiationinduced cancers. Thorotrast is a radiologic contrast agent used for cerebral arteriography from the 1930s to the 1950s. It accumulates in liver, spleen and bone marrow, and chronic exposure to alpha particle decay generates hepatic angiosarcomas and lung cancers.
We analyzed the p53 and K-ras genes for evidence of oxidative damage secondary to radiolysis of water. Among seven lung cancers and five angiosarcomas, there were two G:C > A:T transitions at non-CpG sites and two G:C > T:A transversions, a spectrum consistent with oxidative damage secondary to radiolysis of water. Our analysis of hepatocellular carcinomas related to oral contraceptive administration showed a low mutation frequency and a mutation spectrum consistent with mitogenesis and/or oxidative damage secondary to metabolism of estrogen to cathecol estrogens and metabolic redox cycling. Cyclophosphamide-associated bladder cancers contain a notable clustering of G:C > A:T transitions at non-CpG sites in exon 6 of p53, which coincides with hot spots of adduct formation by phosphoramide mustard, a primary metabolite of cyclophosphamide.
These data implicate phosphoramide mustard, rather than acrolein, as the primary mutagen driving these cancers and suggest a testable strategy for chemoprevention. Mutation analysis of the p53 and K-ras genes in tamoxifen-associated endometrial cancers is ongoing. Preliminary data suggest that tamoxifen may cause endometrial cancers by oxidative damage via redox-cycling of a catechol intermediate. If this is the correct mechanism, then it is possible that anti-oxidant therapy may reduce the incidence of these cancers. In addition, specific genetic susceptibility factors might be found to identify in advance the 0.3 percent of women who are likely to develop endometrial cancer while taking tamoxifen.