Thursday, 12 March 2009 Poster Session I. Predictive and prognostic factors S37 Differences in adjuvant chemotherapy benefit (CMF or anthracycline +/− taxane-based regimens) between the 70-gene low risk versus high risk patients were analyzed in a meta-analysis for 5-year distant disease- free survival (DDFS) and for breast cancer-specific survival (BCSS). For 167 additional patients, treated with neoadjuvant chemotherapy (anthracy- cline +/− taxanes +/− trastuzumab), pCR was assessed. Results: In the adjuvant analysis, MammaPrint assigned 772 patients (47%) to low risk and 865 (53%) to high risk, respectively. 349 patients were treated with ET alone, whereas 226 were treated with ET+CT. In the MammaPrint high risk group a substantial benefit for ET+CT was shown: DDFS hazard ratio (HR) 0.28 (0.14−0.56), p < 0.01 and BCSS HR 0.17 (0.06−0.47), p < 0.01. In multivariate analysis including known clinico- pathologic prognostic factors, the results remained robust: DDFS HR 0.40 (0.20−0.82), p = 0.01 and BCSS HR 0.28 (0.10−0.80), p = 0.02. In the low risk group no significant advantage was observed for ET+CT over ET alone: DDFS p = 0.962 and BCSS p = 0.472 in multivariate analysis. In the neoadjuvant patients, a pCR rate of 20% was shown only in MammaPrint high risk patients, whereas no case of pCR occurred in the low risk group. Conclusion: The 70-gene MammaPrint signature is not only a strong and independent prognostic indicator; moreover it is predictive for the benefit of neoadjuvant and adjuvant chemotherapy. While in 70-gene high risk patients a clear benefit for chemotherapy added to hormonal therapy is demonstrated, a MammaPrint low risk result may select patients who do not benefit from additional chemotherapy. 0074 Early prognosis prediction: MammaPrint on core-needle biopsies F. de Snoo 1 , A. Glas 2 , A. Floore 3 , J. Mayordomo 4 , A. Modollel 5 , C. Rolfo 6 , L. van ’t Veer 7 , E. Rutgers 8 , S. Rodenhuis 9 , R. Bender 10 . 1 Medical Affairs, 2 Bioinformatics, 3 Laboratory, Agendia BV, Amsterdam, Netherlands, 4 Oncology, Hospital Clinico Universitario Lozano Blesa, Zaragoza, 5 Oncology, IDOC, Barcelona, 6 Oncology, Clinica Rotger, Palma de Mallorca, Spain, 7 Pathology, 8 Surgical Oncology, 9 Medical Oncology, Netherlands Cancer Institute, Amsterdam, Netherlands, 10 Medical Affairs, Agendia Inc, Huntington Beach, USA Goals: A 70-gene microarray prognosis signature ‘MammaPrint’ has been developed to improve the selection of patients with breast cancer for adju- vant therapy. MammaPrint was originally established on surgical resection specimens. Since most breast cancer patients will undergo core needle biopsies, we investigated whether the MammaPrint prognosis signature could be assessed in core needle biopsies. Methods: We determined MammaPrint outcome in patients from two neo-adjuvant trials; 50 patients from 4 ONCAMI hospitals in Spain and 168 patients from the NKI. Patients were diagnosed with breast cancer stage II−IV who had core needle biopsies (14-gauge) and subsequent neo- adjuvant treatment. Also in the diagnostic setting, Agendia is performing MammaPrint on core needle biopsies. Results: In the two trials, MammaPrint results were obtained in 98% of samples with sufficient tumor cell percentage. In the ONCAMI neo- adjuvant data set four (11%) of the 35 cases were assigned as “low risk” (good prognosis signature) and thirty-one cases (89%) were predicted to be “high risk” (poor prognosis signature). In the NKI study 23 (14%) were classified good prognosis signature, whereas 144 (86%) patients were classified as poor prognosis signature. Of the 75 samples that were sent in following standard diagnostic proce- dures with sufficient RNA and tumor cell percentage 69 were successfully hybridized and received a MammaPrint result. Conclusion: The MammaPrint assay was originally designed for tumor tissue from surgical specimens. We show here that MammaPrint prognosis signatures can be obtained in core needle biopsies. In addition the same biopsy can be used for ER, PR and HER2 gene expression read-out by TargetPrint. The results of this study have broadened the clinical applicability of the MammaPrint prognosis signature. 0075 Use of the genomic test MammaPrint in daily clinical practice to assist in risk stratification of young breast cancer patients G. Kunz 1 , A. Glas 2 , F. de Snoo 2 , I. Simon 2 . 1 Frauenheilkunde und Geburtshilfe, Johannes Hospital, Dortmund, Germany, 2 Agendia BV, Amsterdam, Netherlands Goals: Most patients with early stage breast cancer will not develop metastases. However, chemotherapy treatment is administered to most patients and many will not benefit but still suffer from often severe side and long-term toxic effects. It is therefore crucial to identify those patients who are at high risk of developing metastases. Using clinical guidelines, many patients are considered to be at intermediate risk. Additional methods should help to better classify this intermediate breast cancer patient group. Methods: A 70-gene expression profile was established as a pow- erful predictor of disease outcome in breast cancer patients. The test (MammaPrint) was validated in independent cohorts and cleared by FDA in 2007. In this study, adaptation of MammaPrint into clinical work-up procedures was investigated. MammaPrint was used to determine risk prognosis in 44 patients with an average age of 44 years treated between 2004 and 2008 in a General Hospital in Germany. The prognosis profile is compared to clinicopathological factors used in current risk classification. Results: MammaPrint indicated a good prognosis profile in 29 patients and a poor prognosis profile in 15 patients. Using the St. Gallen 2007 guidelines for risk assessment, 4 patients were low risk and 6 patients were classified as high risk. The remaining patients (n = 34) were classified as intermediate risk. In the intermediate group, Adjuvant Online classified 19 patients as high risk and 15 as low risk while MammaPrint classified 11 of the intermediate patients as high and 23 as low risk. To assess the accuracy of MammaPrint in this younger patient group, we performed a meta-analysis on a subgroup from previously described retrospective studies (n= 689). Kaplan–Meier analysis showed a significant difference for both the probability of remaining metastasis-free as well as overall survival (p < 0.001) and a better separation than using clinical parameters. 10 year overall survival probability for the good and poor MammaPrint prognosis groups was 90.2% and 65.2%, respectively. Conclusion: Patients with an intermediate clinical risk can be accurately separated into poor and good prognosis risk groups. MammaPrint provides more accurate information on recurrence risk compared to conventional clinicopathological criteria and provides additional guidance in daily clinical practice. 0076 Delayed chemotherapy and/or dose reduction. A model to accurately predict the dose-intensity E. Banu 1 . 1 Medical Oncology, Cancer Institute Ion Chiricuta, Cluj-Napoca, Romania Goals: Dose-intensity (DI) is an important variable when treating breast cancer (BC) patients with chemotherapy. Its impact on the outcome was clearly demonstrated for antracycline-based regimens in adjuvant setting. DI could be important for targeted-therapies. Methods: DI was calculated for a taxane and antracycline regimen administered every 3 weeks. We used FileMaker Pro as database sys- tem. Drug dose reduction (%) and delay between consecutive cycles of chemotherapy were independent variables. Multiple simulations were performed using 76 combinations of both parameters: dose reduction (0, 5, 10, 15, 20, 25, 30%) and delay (−4, −3, −2, −1, 0, 1, 2, 3, 4, 5, 6, 7 days). The dependent variable was the DI (%). Categories of patients were defined function of DI reduction. Linear and multiple regression analyses were performed using a model with two independent and one dependent variables (stepwise selection). SPSS was used to estimate the unstandardized coefficients with 95% confidence interval (CI). The adjusted R-squared expressed the association between independent and dependent variables. Results: The predicted DI value was obtained as follows: DI (%) = 100 − 3.59×delay(days) −0.94×dose reduction(%). The 95% CI: lower bound [99.49 −3.76×delay(days) − 0.99×dose reduction(%)]; upper bound [101.55 − 3.42×delay(days) − 0.88×dose reduction(%)]. The adjusted R-squared were 0.58 and 0.38 for delay and dose reduction, respec- tively. Each day of delay is equivalent with 4% of dose reduction. The expected DI resulted from 10% dose reduction combined with 3 days of