|
Sub-category:
Adjuvant Therapy
Category:
Breast Cancer--Local-Regional and Adjuvant Therapy
Meeting:
2008 ASCO Annual Meeting
Session Type and Session Title:
General Poster Session, Breast Cancer — Local-Regional and Adjuvant Therapy
Abstract No:
541
Citation:
J Clin Oncol 26: 2008 (May 20 suppl; abstr 541)
Author(s):
W. F. Symmans, C. Hatzis, C. Liedtke, C. Desmedt, V. Valero, H. M. Kuerer, G. N. Hortobagyi, M. Piccart-Gebhart, L. Pusztai, C. Sotiriou
Abstract:
Background: The genomic grade index (GGI) is a 97-gene measure of histologic tumor grade. High GGI is associated with decreased relapse-free survival without and with systemic adjuvant endocrine- or chemotherapies. In the current study we examined whether GGI predicts pathologic response to neoadjuvant chemotherapy in 229 patients with HER2-normal breast cancer. We hypothesized that high GGI will predict for greater sensitivity to chemotherapy. Methods: Gene expression data was generated with Affymetrix U133A gene chips from fine-needle aspirations samples prospectively collected before starting therapy. All patients received sequential paclitaxel, 5-fluorouracil, doxorubicin, cyclophosphamide (T/FAC) neoadjuvant chemotherapy. Pathologic response was quantified using the residual cancer burden (RCB) method. The association between the GGI and pathologic response was assessed in univariate and multivariate logistic regression analyses including age, estrogen receptor (ER) and node status, nuclear grade, baseline T size and the GGI. The performance of a multivariate (clinical + GGI) predictor was evaluated in 3-fold cross-validation with 1,000 iterations. Results: GGI correlated with tumor grade, 79% of grade 1 tumors had low GGI whereas 88% of grade 3 tumors had high GGI. Sixty-one percent of grade 2 tumors were reclassified as low GGI. GGI scores were significantly (p<0.04) higher in patients with pathologic complete (pCR) or excellent response (RCB 0/I = pCR or near-pCR) compared to lesser response in both ER + and ER- cancers. In multivariate analysis, ER-status (OR:0.43, 95%CI:0.2-0.91), node status (OR:0.35, 95%CI:0.16-0.8) and GGI (OR:1.86, 95%CI:1.16-3.0) were significant (p<0.03) independent predictors of RCB 0/I response. A multivariate prediction model had mean AUC of 0.73, PPV=0.53, NPV=0.72, sensitivity=38% and specificity=86% in cross validation. Conclusions: High GGI is predictive of increased sensitivity to neoadjuvant T/FAC chemotherapy in both ER- and ER+ patients. GGI can be combined with clinical variables to produce a reasonably accurate multivariate prediction model of pathologic response to chemotherapy.
Abstract Disclosures
Abstracts that were granted an exception in accordance with ASCO's Conflict of Interest Policy
and are designated with a caret symbol (^) here and in the print version.
 Associated Presentation(s):
Other Abstracts in this Sub-Category:
Abstracts by W. F. Symmans :
Presentations by W. F. Symmans :
Educational Book Manuscripts by W. F. Symmans :
|
