Population screening has increased the absolute number of breast cancer cases. In spite of widespread screening, there are still many patients who present with locally advanced breast cancer, often women who are young or who present with a mass between annual screens (“interval cancers”). Molecular profiling provides the opportunity to determine whether mammographically detected cancers are biologically distinct from locally advanced breast cancers. Methods: Tumor samples from the time of diagnosis (2002-2006), from patients with locally advanced breast cancers enrolled on the multi-site I-SPY TRIAL (CALGB 150007/150012 and ACRIN 6657) were compared to stage 1 & 2 tumor samples collected in 2004 from multiple community Netherlands hospitals (RASTER dataset). The NKI 70 gene set was performed on 100 of 221 patients (80 more samples pending) from I-SPY and 228 of 242 patients from the RASTER dataset. The proportion of good vs. poor prognosis profiles were compared. Additionally, we have collected data on response to chemotherapy from the I-SPY patients. Finally, mammograms from the I-SPY and RASTER datasets are being reviewed to enable classification of tumors by presentation (palpable, screen detected, and interval cancer and negative prior mammogram within 1 year). Results: I-SPY data includes 221 patients with tumors ≥3cm in size; median tumor size is 6cm; 55% were <50 years old; 44% were ER negative. Only 20% had an NKI 70 gene good prognosis at time of diagnosis. The percentage of women under <40, 41-50, and >50, with a good prognosis is 10%, 23%, and 22%, respectively. The RASTER dataset included 427 patients from age 30 to 61. NKI 70 gene profiles are available in 219 patients: 66% were Stage 1; 33% were Stage 2. Overall, 51% had a NKI 70 gene good prognosis profile. For women under the age of 40, 31% had a NKI 70 gene poor prognosis. For women 40-50, and >50, the fraction of NKI 70 gene poor prognosis is 55% and 56%, respectively. Data will be presented based on mode of detection: screen vs. palpation (never screened vs. interval). Conclusion: The key observation is that the fraction of good prognosis tumors substantially increases with age. In patients too young for screening, the fraction of poor prognosis tumors is 69%. Patients who present with locally advanced breast cancers who are too young to be screened almost always have a poor prognosis signature. This may extend to those with interval cancers. This suggests that the tumors identified by screening, although they are early stage cancers, are not the precursors of the poor prognosis tumors in young women and those with locally advanced disease. Therefore, we may not be able to rely on current screening approaches to improve outcomes for young women and those with interval cancers. Further analysis will include reviewing interval cancer cases within the I-SPY dataset and assessing prognostic indicators.