| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Article |
1 Department of General Surgery–Breast Center, The Cleveland Clinic Foundation, The Cleveland Clinic Breast Center, 9500 Euclid Avenue, A10, Cleveland, Ohio 44195
2 Department of Biostatistics and Epidemiology, The Cleveland Clinic Foundation, The Cleveland Clinic Breast Center, 9500 Euclid Avenue, A10, Cleveland, Ohio 44195
3 Department of Surgical Pathology, The Cleveland Clinic Foundation, The Cleveland Clinic Breast Center, 9500 Euclid Avenue, A10, Cleveland, Ohio 44195
4 Department of Hematology and Oncology, The Cleveland Clinic Foundation, The Cleveland Clinic Breast Center, 9500 Euclid Avenue, A10, Cleveland, Ohio 44195
Correspondence: Address correspondence and reprint requests to: Pedro F. Escobar, MD; E-mail: escobap{at}ccf.org.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Methods: Data, collected prospectively between 1997 and 2002, included clinical tumor-node-metastasis stage, age at diagnosis, hormone receptor status, type of preoperative chemotherapy, histological type, surgical procedure, pathologic measurement in centimeters of residual breast tumor, and the number of positive axillary nodes in the surgical specimen. Univariable correlates of residual breast disease were assessed by using the 
2 test. Recursive partitioning analysis was used to determine the prognostic significance of clinical tumor size, residual tumor size, and pathologic node involvement relative to overall survival. Survival was estimated by using the method of Kaplan and Meier and compared by using the log-rank test. A P value of < .05 was considered significant.
Results: Data were available for 85 patients with advanced breast cancer. Although uni-variable analysis identified increasing age, clinically involved axillary nodes, and a higher clinical tumor-node-metastasis stage as predictors of an increased risk of residual disease, recursive partitioning analysis identified more than three involved axillary nodes in the surgical specimen, with or without any measurable residual breast disease, as the most significant predictor of decreased survival (P < .001).
Conclusions: Pathologic axillary node involvement was the most significant predictor of decreased survival for patients who had undergone primary induction chemotherapy for advanced breast cancer.
Key Words: Advanced breast cancer • Primary induction chemotherapy • Clinical outcome • Axillary lymph nodes • Metastatic breast cancer
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
PIC was introduced in the 1970s. Randomized clinical trials5–8 comparing clinical outcomes between patients who underwent PIC for ABC and those who did not concluded that PIC offered similar disease-free and overall survival (OS).9–11 Several studies have identified molecular or pathologic markers that may predict the clinical and pathologic response to PIC or estimate OS, including residual tumor size,12 clinical tumor-node-metastasis tumor stage, HER-2/neu overexpression, hormone receptor status,13,14 and axillary node involvement.15 Another study suggests that axillary lymph node dissection may not be necessary if the clinical response to PIC is deemed complete.16 Generally residual breast disease (with or without axillary node involvement) or complete pathologic remission correlates with disease-free survival and OS for patients with ABC. The purposes of this study were to identify correlates of residual breast disease and to determine the prognostic significance of clinical tumor size, residual breast disease, and axillary lymph node involvement for patients with ABC treated with PIC followed by surgery.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Histopathologic Evaluation After Chemotherapy
The pathologic evaluation of a breast tumor after PIC often differed from those that had not been treated before surgery. For this study, the pathologist determined the location of the lesion by reviewing the imaging studies. If a gross lesion was identified in the resected tissue, measurements were taken in three dimensions, and the area was thoroughly sampled for microscopic examination. If a gross lesion was not identified (a finding not uncommon after PIC), then the region of the prior mass, as determined by imaging and/or clinical examination, was located and thoroughly sampled for microscopic examination. In such cases, if residual tumor was seen microscopically, then the greatest diameter of invasive tumor from the histological section was measured and reported.
Not infrequently after PIC, the microscopic appearance of the remaining gross lesion demonstrated fibrosis with scattered clusters of residual tumor cells (often with tissue edema), which made up no more than 20% to 30% of the grossly identified mass. Before chemotherapy, much of this gross lesion would have been full of tumor cells (as is seen in untreated cases). These cells were eradicated by chemotherapy and so described as infiltrating carcinoma with treatment effect.
All identified lymph nodes were submitted in their entirety for microscopic evaluation. If metastatic tumor was seen, it was measured, and the presence of extracapsular extension, if any, was noted. Occasionally areas of necrosis or fibrosis with edema or inflammation of the node were observed and were believed to be areas of treated metastatic disease. If the node was a sentinel node and tumor was not seen with hematoxylin and eosin stain, then it was further analyzed by using immunohistochemical stains for cytokeratins. Often these were negative, but occasionally isolated tumor cells were seen with immunohistochemical stains only and were reported as such.
Statistical Analysis and Study Design
Recursive partitioning analysis (RPA), also known as classification and regression tree, is a technique used to determine subsets of explanatory variables most important for prediction of some response variable, which may be either binary or time to event, such as survival. RPA separates patients into groups based on outcome; patients in the same group have similar outcomes, and patients in different groups have different outcomes.
In the analysis of residual breast disease, RPA with a binary end point was used to determine whether cutpoints that correlated with residual breast disease could be determined for the two continuous variables: age and clinical tumor size. Five variables were considered as potential correlates of residual breast disease: age, hormone status, clinical tumor size, clinical axillary nodes, and clinical tumor-node-metastasis stage. The 2 test was used to assess the association of these variables with residual breast disease.
RPA with a time-to-event outcome was used to assess the prognostic significance of clinical tumor size, residual tumor size, and residual axillary nodes for OS, which was then estimated by using the method of Kaplan and Meier and compared among groups by using the log-rank test. A P value of < .05 was considered significant.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Clinical histological findings included 63 patients with infiltrating ductal carcinoma, 13 with infiltrating lobular carcinoma, and 9 with either mixed histological characteristics or infiltrating medullary carcinoma. Clinical tumor-node-metastasis staging identified 17 patients with stage II disease, 35 with stage IIIA, 17 with stage IIIB, and 16 with stage IV, including 7 patients who had T2 tumors, 55 who had T3, and 23 who had T4a/b/c. The mean clinical tumor-node-metastasis size was 8.5 cm (SD ± 3.8 cm). Clinical tumor-node-metastasis axillary node status included seven cases classified as N0, 15 classified as NX, and 63 classified as N1.
After completion of PIC, 67 (79%) patients chose mastectomy, and 18 chose partial mastectomy; all patients underwent axillary lymph node dissection. Pathologic evaluation of breast specimens showed complete tumor resolution in 13 specimens, whereas 72 did not show resolution, including 52 in which infiltrating ductal carcinoma was identified, 14 with lobular carcinoma, and 6 with either mixed infiltrating histological characteristics or only ductal carcinoma-in-situ. The mean residual tumor diameter was 3.0 cm, representing a mean reduction of 62%. Pathologic evaluation of axillary lymph nodes showed no metastases for 35 patients, whereas 50 patients did have metastases.
In the assessment of presurgical characteristics versus residual breast disease, RPA of the two continuous variables split patients into two groups based on age (
56 and >56 years) and three groups based on clinical tumor-node-metastasis size (1–6, 7–12, and 13–20 cm). Univariable correlates of residual infiltrating breast cancer were identified to determine whether any presurgical characteristics could help predict which patients would likely have residual breast disease (Table 1
). Age at diagnosis, clinical tumor-node-metastasis axillary node status, and clinical tumor-node-metastasis stage were found to be univariable correlates of residual breast disease, whereas hormone receptor status and clinical tumor-node-metastasis size were not. All patients >56 years of age at diagnosis had residual breast disease, compared with 79% of patients 56 years of age (P = .028). The likelihood of residual breast disease also increased for patients who had a clinical tumor-node-metastasis stage of N1 (P = .038) and as the clinical tumor-node-metastasis stage increased (P = .004).
|
).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The significance of pathologic axillary node involvement for ABC patients has been reported in some studies,21–23 including one that concluded that axillary lymph node status associated with age at diagnosis was the most important predictor of OS.24 Krag and Single25 found that the number of axillary nodes removed was associated with survival, even when those nodes were interpreted pathologically as negative, thus suggesting that axillary lymph node dissection alone may contribute to OS. This finding is in contrast to studies that suggest that a good clinical response to PIC may obviate the need for axillary lymph node dissection16 or that axillary irradiation may be sufficient.26 In this study, the number of involved axillary nodes in the pathologic specimen, with or without residual tumor in the breast, was the most important predictor of OS.
To assess OS more accurately, long-term follow-up is required; longer follow-up captures additional deaths for patients with ABC.24,27,28 Although an increased OS was observed for patients with zero to three involved axillary nodes and residual disease of 0 to .4 cm, in the absence of data from a randomized trial, a definitive conclusion about OS in this subset of patients is unsupported because this finding may reflect selection bias or insufficient follow-up time.
PIC may reduce the size of the primary breast tumor, decrease the number of involved axillary nodes, or both; nevertheless, this theoretical advantage has generally not translated into a significant improvement in OS.26 Neither the significance of histopathologic evaluation nor the standards to measure pathological response in this patient population are universal, but our findings suggest that the degree of pathologic axillary lymph node involvement after PIC is perhaps the most important predictor of OS.
Received for publication July 22, 2005. Accepted for publication November 23, 2005.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
