Pathobiology of preoperative chemotherapy: Findings from the National Surgical Adjuvant Breast and Bowel Project (NSABP) protocol B‐18

Abstract

BACKGROUND: Examination was performed on pathologic material from patients enrolled in the National Surgical Adjuvant Breast Project (NSABP) protocol B‐18, in which the clinical effects of preoperative (preop) and postoperative (postop) doxorubicin and cyclophosphamide (AC) were compared. METHODS: Of the total number of 1523 patients, 1234 patients (81%) were in the pathologically evaluable cohort. Six hundred twenty‐six patients had been randomized prospectively to receive AC postop and 608 had been randomized to receive AC preop. Preentry diagnosis was made by fine‐needle aspiration (FNA) and/or Tru‐cut biopsy (TC). AC‐induced and other pathologic changes were identified, and their relation to pathologic response and overall survival (OS) and disease‐free survival (DFS) was determined. Frequencies of the number of lymph node metastases, their size, stromal reaction, and extracapsular extension (ECE) were compared in the two treatment groups, as was their correlation with OS and DFS. Survival estimates were based on 9 years of follow‐up. RESULTS: Approximately 13% of primary breast carcinoma cases exhibited both a clinical complete response (cCR) and a pathologic complete response (absence of invasive tumor [pCR]) to preop AC. An additional 7% of patients exhibited a pCR in the absence of a cCR. A pCR occurred in 38% of those patients determined to have achieved a cCR. Poor nuclear grade of the tumor cells in the pre‐entry FNA and/or TC specimens significantly predicted a pCR. Patients with the latter exhibited a better OS and DFS compared with those with a pathologic partial response (presence of sparse invasive tumor [pPR]) or no pathologic response (pNR). Epithelial alterations considered to be induced in tumors by preop AC were comprised of types 1 and 2 giant cells with meganuclei, apocrine metaplasia, and cytoplasmic vacuolation. They had a high degree of specificity (range, 86–99%) but a low sensitivity (range, 7–38%). All were predictive of a pPR and were found to be related adversely to OS and DFS. A fibrous stromal reaction noted in tumors or their putative sites in the preop group was found to have only modest degrees of specificity (63%) and sensitivity (74%). Moderate/marked sclerosis of basement membranes of the ductal and ductular elements of the terminal ductolobular unit (TDLU) was significantly more frequent in nontumor‐bearing areas of breasts from patients in the preop treatment group compared with those in the postop treatment group (67% vs. 48%; P < 0.0001). The degrees of change in the TDLU in patients in the postop treatment group were found to be unrelated to age. Lymphatic tumor extension in the primary tumor, as well as a positive lymph node status, were less frequent in the preop treatment group compared with the postop treatment group. The OS and DFS were nearly identical in both treatment groups, being 69% and 55% and 70% and 53% in the preop and postop treatment groups, respectively, at 9 years. A fibrous stromal response to lymph node metastases was found to be significant for DFS but not OS. ECE was similar in both groups (55% vs. 48%; P = 0.12). Only 1% of ECE was found to be related to axillary failure in both treatment arms combined. There was no significant difference with regard to the parameters of survival for patients in the postop treatment group whose lymph nodes contained micrometastases (< 2.0 mm) or mini micrometastases (< 1.0 mm) (the latter detected immunohistochemically with anticytokeratin), and a true‐negative lymph node status (not immunohistochemically converted to positive). Conversely, there was no apparent difference with regard to OS in preop treated patients with lymph node micrometastases, mini micrometastases, and macrometastases (P = 0.19). Those with mini micrometastases had a significantly worse OS compared with those with a true‐negative lymph node status (P = 0.0007). DFS remained worse for patients in that treatment group with micrometastases and mini micrometastases compared with those with negative lymph nodes, although it was better than that for patients with macrometastases (P = 0.02). CONCLUSIONS: Poor nuclear grade of tumor cells in the preentry FNA or TC specimens in the preop group was predictive of a pCR. AC‐induced meganuclear giant cells and apocrine changes and nuclear and histologic grades of the primary tumors also were found to be prognostically significant in patients in the preop treatment group, and the latter two variables were found to be significant for those patients in the postop treatment group. No evidence was found to support the need for axillary lymph node radiation for ECE of lymph node metastases. Extended pathologic or immunohistochemical procedures also appear to be unnecessary for the detection of lymph node mini micrometastases, at least when traditional postop chemotherapy is used. The adverse relation between such small metastases and OS and DFS after preop AC appears to be related to the timing of the chemotherapy administration rather than any pathobiologic reasons. Cancer 2002;95:681–95. © 2002 American Cancer Society. DOI 10.1002/cncr.10741