Printed in U SA. RE: "LONG-TERM HORMONE REPLACEMENT THERAPY AND RISK OF BREAST CANCER IN

In their paper on long-term hormone replacement therapy and risk of breast cancer, Newcomb et al. (1) adjusted for multiple risk factors by applying the frequently used logistic regression model implying multiplicative odds. Endogenous gonadal hormones have an important role in causing breast cancer (2). At present, the dominant multi-stage theory of carcinogenesis assumes that cancer arises from one single normal cell turning malignant over the course of a certain number of transitions. An exposure can affect the first transition rates (early-stage carcinogens) or one of the later transition rates (late-stage carcinogens). Hormones are regarded as late-stage carcinogens (promot-ers) for breast cancer (3). Kinetics of late-stage carcinogens may best be described by additive relative risk models (3, 4). Moreover, when at least two causal agents act on th