Prognostic factors and tumour markers in early breast cancer; a commentary

Abstract

THIS article is not a review of the literature on prognostic factors and tumour markers in early breast cancer but is a commentary based on some fifteen years’ work on the periphery of the field. The advantage of a commentary over the usual classical review of the literature is that it is obvious from the start that the author has a particular axe to grind and is not impartial. Even formal reviews can be slanted in a number of ways, depending on the interests and bias of the reviewer. For example, the basic biological implications of some of the research in this area is not without interest: other workers may be interested in the taxonomy of the subject; the orderly listing of whether particular factors are tumouror host-specific or whether a marker might be classified as a hormone or not. There is room for all those different viewpoints in cancer research. The main practical aim of work on predictive factors and markers is to see whether they have any immediate use in improving the efficiency of the treatments now available. In realistic terms, it would be desirable to be able to determine one simple figure: the probability that a patient will develop metastatic disease at a given rate (or, less satisfactorily, within a given time, and there is a distinction between these two options). The probability needs to be defined with considerable accuracy, and a finding which scrapes home with the minimal acceptable statistical significance is hardly likely to be useful if the predicted recurrence rate is going to be used for deciding a particular course of treatment. In other words, a probability of 0.05 that, say, serum IgE levels are higher in patients with rapid recurrence than in those with slower rates, based on a series of 20 patients followed for 2 yr, may be acceptable as a basis for a publication but it might be less convincing as a determinant of whether a patient should be given aggressive chemotherapy at the time of mastectomy. In practical terms, it has to be admitted that work in the last decade has failed to provide realistic prognostic indices. It has not been established, beyond reasonable doubt, that action taken as a result of measurements of either single or multiple markers in a medical centre of excellence have added one day to the disease-free interval after mastectomy, or to survival, compared to that found in the generality of patients treated in peripheral hospitals where such facilities are not available. (Pathological staging and histological grading are excluded: hormone receptor site measurements are reviewed elsewhere in this volume.) It might now be useful to look into some of the reasons for failure.