P4-09-01: Identification of Poor Prognosis T1T2N0 Luminal ERBB2-ve Breast Carcinomas.

Background: To identify ER+ve ERBB2-ve ductal T1T2N0 carcinomas associated with a poor prognosis remains challenging. We have previously demonstrated that the number of chromosomal breakpoints assessed by CGH could be a marker of worse outcome for breast carcinomas. Our aim was to validate the CGH based signature in a series of luminal ductal and T1T2 N0 carcinoma patients with long-term clinical follow-up. Patients and methods: We analyzed 214 patients treated for an invasive ductal ER+ve ERBB2-ve carcinomas, smaller than 30mm. The training set was composed of 109 patients (10.9 years of median follow-up; 30 cases associated with a metastatic event within less than 4 years/79 control cases with no metastastic event at 5 years) and the validation set of 105 patients (10.5 years of median follow-up; 30 relapses including contra-lateral breast carcinomas, loco-regional relapses and 8 metastatic events). None of the patient received adjuvant chemotherapy. 16 received an adjuvant hormonotherapy (10 in the training and 6 in the validation groups). We genotyped the sample set with the SNP6.0 affymetrix array. After RMA normalisation using Genotyping console, segmentation was performed according to the Zhang and Siegmund maximum method. In the training data set, the number of breakpoints was assessed, linked to outcome and the threshold optimising the sensitivity and specificity was determined (ROC curve). The threshold prognostic value was then tested on the validation series (Kaplan Meier analysis, log rank test, determination of relative risk and its confidence interval with a Cox model). Results: In the training set, median numbers of breakpoints were 7 in cases that experienced a metastatic event after more than 5 years and 40.5 in cases that experienced a metastatic event in less than 4 years. The threshold (Younden index ) was 34 breakpoints with a sensitivity of 0.57 and a specificity of 0.94 (AUC: 0.81[0.71;0.91]). In the validation set, the outcome of patients with more than 34 breakpoints was poorer than that of patients with less than 34 breakpoints ( 34 breakpoints: 11 events out of 22 patients with a median time to progression of 108 months; p 34 versus 34 breakpoints: 4 events out of 22 patients with a median time to progression of 108 months; p=0.009 (logrank test); RR: 5.29 [1.32; 21.26]). Conclusion: We demonstated that patients with T1T2 ( Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P4-09-01.