| Translating gene expression patterns into IHC diagnostics for breast cancer: novel IHC reagents predict recurrence in a retrospective cohort at the Comprehensive Cancer Institute of Huntsville. Poster at San Antonio Breast Cancer Symposium 2004
Ross DT¹, Ring BZ¹, Seitz RS², Beck R², Shasteen WJ², Estopinal NC³
¹Applied Genomics Inc., Sunnyvale, CA; ²Applied Genomics Inc., Huntsville, AL, ³Comprehensive Cancer Institute of Huntsville, Huntsville, AL
Abstract #3004
Background: Gene expression studies have identified novel, biologically distinct classes of breast cancer which have differing outcomes. In collaboration with scientists at Stanford University, we have used gene expression data to target the production of over 700 novel antisera and screened them using tissue arrays to discover IHC biomarker reagents that can be used to distinguish patients subtypes in retrospective patient cohorts.
Methods: We used tissue arrays containing 400 independent breast cancer samples to screen over 400 novel IHC reagents and identify a Panel of Diversity comprised of approximately 45 reagents that distinguish independent subtypes of breast carcinoma. We assembled a second breast cancer cohort consisting of 550 patients treated at the Comprehensive Cancer Institute of Huntsville between 1989 and 2000 (average follow-up of 65 months) and assessed the pattern of staining of these reagents and their association with tumor recurrence and survival. In a preliminary analysis of our data, we have used regression tree analysis to identify algorithms for the prediction of outcome in both ER+ and ER- patients.
Results: Antibody reagents targeted to so-called basal and luminal subtypes of breast cancer distinguished appropriate subtypes of normal breast epithelial cells confirming the biologic interpretation of breast gene expression studies. A panel of six antibodies (four novel and two commercial) antibodies had a sensitivity of 84% and specificity of 62% (HR=6.7 p<0.0001) for identifying ER-positive patients who recurred. A related panel of seven antibodies (six novel and one commercial) detected 54% of ER-negative patients that remained disease free in the study and contained only 8.9% of recurrers (HR=2.7 p=0.003) The prognostic values of these markers were independent of stage and grade by a Cox multivariate proportional hazard model. Lymph node status remained a significant independent discriminator for the ER-negative patient prognosticator.
Discussion: These studies demonstrate that subtypes of breast cancer originally discovered by gene expression patterns can be distinguished using IHC reagents and that a panel of only several antibodies may be sufficient to predict clinical course. Follow-up studies on independent cohorts are warranted to validate the clinical utility of these prognostic antibody panels. We believe that the expanded Panel of Diversity , which distinguishes broader biologic diversity in breast cancer, should prove useful for retrospective and prospective studies using paraffin embedded diagnostic tumor tissue to identify and validate biomarkers that distinguish responder populations in adjuvant clinical trials.
|
