Mathematical procedures, some unique to this study, were applied to 114 suggested risk factors, with as many as 10 subsets, from history, physical and x-ray examination on 30,904 breast studies on 7,252 patients referred from the Emory University Clinic since 1963. One-fifth of the cancers were unrelated to symptoms; 82% of the cancers were free of axillary lymph node metastasis. There was no sign or symptom that predicted preclinical cancer. Interaction of numerous indicators subjected to strong statistical procedures could contribute to establishing risk of even early breast cancer. The results of hierarchic discriminant analyses demonstrated the feasibility of using simultaneously large numbers of risk factors in a systematic way to pinpoint patient with mammary cancer. Based on usual clinical and x-ray assessment of the women 12.5% of the noncancer patients required biopsy to demonstrate 70% of the cancers with a cancer to benign rate of 1 to 4. Using the same data with discriminant analyses, 5.6% of the patients would require biopsy, at the rate of one cancer to 1.8 benign lesions; 92% of the cancer patients could be placed in 11.9% of the population. A computerized system has been developed for widespread application to provide the clinician with a highly objective and totally consistent assessment of risk for breast cancer in each of his patients.
Cancer 43:871-877, 1979. HE CONCEPT OF BREAST CANCER as a single T mass, with borderline operability, obvious axillary lymph nodes and questionable distant metastasis is being completely replaced. Through recognition of preclinical breast cancer drastically new and pivotal concepts have evolved that are accompanied by markedly different attitudes and reactions in both physicians and patients. These changes have paralleled the development, adaptation and clinical application of mammography over the past two decades.' ~