AbstractAbstract #3026Background: Patients with early stage ER+ve breast cancer have excellent prognosis with ca 90% 5 year disease free survival when treated with endocrine therapy. However for patients who relapse during endocrine therapy additional adjuvant therapy options, such as chemotherapy, are indicated. The challenge is to prospectively identify such patients. The Mammostrat® test comprises 5 simple immunohistochemical markers (p53, HTF9C, CEACAM5, NDRG1, SLC7A5) which stratify node negative tamoxifen treated patients into low, moderate and high risk groups. We have now tested the efficacy of this panel in a mixed population of node positive/node negative cases treated in a single centre (Edinburgh Breast Unit) with breast conserving surgery.
 Methods: TMAs from a consecutive series (1981-98) of 1,812 women managed by wide local excision and postoperative radiotherapy (45Gy in 20-25 fractions) were collected following appropriate ethical review. Of 1390 cases stained, 197 received no adjuvant hormonal or chemotherapy, 1044 received tamoxifen only as adjuvant therapy and 149 received a combination of hormonal and chemotherapy. Median age at diagnosis was 57, 71% were post-menopausal, 23.9% node positive, median size was 1.5 cm. Samples were stained, using triplicate 0.6mm2 TMA cores and positivity for p53, HTF9C, CEACAM5, NDRG1, SLC7A5 recorded as previously described. Each case was assigned a Mammostrat score and RFS and OS analysed by marker positivity and Mammostrat score.
 Results: Staining for all 5 antibodies was successful in 1174/1390 (84%) of cases. In the primary analysis of 531 N0/ER+ve Tamoxifen only treated patients Mammostrat was significantly associated with relapse free survival (RFS) in univariate (p=0.025) & multivariate proportional hazards analysis (p=0.01, HR=1.3, 95%C.I. 1.08-1.74). PgR, multifocality and menopausal status were significant co-variates (p<0.05, HR 0.89, 2.0 & 0.6 respectively). The Nottingham prognostic index was non-significant. Of the 5 antibodies, only p53 (p=0.04) was independently predictive of survival.
 In a secondary univariate analysis of 781 patients (including N+ve and chemo/tam treated patients) Mammostrat was predictive of RFS & OS (p<0.01) with NDRG1/CEACAM5/p53 also predictive of RFS(p<0.05). However Mammostrat was not independent of nodal status, pathological size, grade or multifocality in a proportional hazards analysis.
 Discussion: In the Edinburgh BCS population Mammostrat was predictive of RFS (both local and distant relapses) in N-ve/ER+ve patients treated with tamoxifen alone irrespective of menopausal status. There was a strong correlation between Mammostrat scores and grade, however, in a multivariate analysis Mammostrat contributed significantly to prognostication along with PgR, multifocality and menopausal status.Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 3026.
