Purpose: As the feasibility of breast preservation in locally advanced breast cancer is currently under evaluation, little information is available correlating mammographic changes to chemotherapy with local outcome. To evaluate the role of mammography in selecting candidates with locally advanced breast cancer for conservative local therapy, we analyzed mammographic changes in the breast to induction chemotherapy and correlated the radiologic appearance with pathologic outcome.
Methods and materials: From 1985 through 1993, 91 patients with Stage III breast cancer were enrolled on a multimodality clinical trial using chemohormonal therapy followed by local treatment and maintenance therapy. Induction therapy consisted of cyclophosphamide, doxorubicin, methotrexate,and 5-fluorouracil with hormonal synchronization using tamoxifen and conjugated estrogens. After nine cycles, surgical biopsies of the breast were performed. Through 1988, clinical examination alone directed the site for postinduction biopsy; for patients treated after 1988, mammography, in addition to physical examination, determined the biopsy location. Local treatment was determined by biopsy result. Patients with a pathologic complete response received radiation only to the breast adn regional nodes, while those with pathologically proven residual disease underwent mastectomy and postoperative radiotherapy. Nine additional cycles of maintenance chemotherapy were administered.
Results: Fifty-five of 91 patients (58%) obtained a clinical complete response (CR) to induction chemotherapy. Twenty-eight of the 53 women with a clinical CR had both pre- and postinduction mammograms. Of these 28 women, 9 obtained a pathologic CR and 19 obtained a pathologic partial response (PR). Fifty-five percent of the pathologic complete responders had resolution of mammographic abnormalities on the postinduction mammograms. Sixty-eight percent (13) of the pathologic partial responders had abnormal mammographic findings. The positive predictive value for residual cancer using physical examination was 92%, while the negative predictive value was only 36%. Among patients with a clinical complete response, the positive and negative predictive values for residual cancer using postinduction mammography were 79% and 56%, respectively. Limitations of mammography included uncertain significance of residual microcalcifications and residual masses on postinduction chemotherapy mammograms.
Conclusions: Although mammography improved the accuracy of noninvasive evaluations in patients with a clinical complete response, pathologic assessment was still required to determine appropriate local therapy. More sensitive imaging modalities or modifications of film-screen mammography may improve noninvasive detection of residual disease following induction chemotherapy.