e11117 Background: Prognosis of young women’s breast cancer is influenced by reproductive history with worse prognosis when diagnosis is within five years postpartum. We propose that breast involution accounts for the poor prognosis of postpartum breast cancers (BC), as involution is identified by tissue remodeling programs that share similarities with tumor promoting microenvironments including elevated fibrillar collagen deposition and COX-2 expression. Methods: We developed murine models of postpartum BC. To determine if targeting COX-2 may decrease progression of postpartum BC, mice were treated with NSAID ibuprofen for two weeks during the window of involution. Further, the collagen/COX-2 pathway in human postpartum BC was investigated in breast tissue and DCIS from young women. Results: The normal murine involuting gland showed high levels of radiating collagen fibers organized similar to collagen from invasive tumors. Mammary tumors arising in this collagen-rich microenvironment have increased levels of COX-2. In vitro, tumor cells isolated from PABCs are motile and invasive in a collagen and COX-2 dependent manner. Surprisingly inhibition of COX-2 during involution decreased collagen levels and frequency of radial fibers. Ibuprofen treatment reduced tumor size, collagen architecture, COX-2 expression, and lung cell seeding. In humans, comparison of breast tissue from nulliparous women to involution shows increased levels of collagen and >2x the frequency of radially aligned fibers. In our DCIS study, analysis of COX-2 expression reveals a trend towards increased COX-2 expression in the postpartum DCIS lesions. Conclusions: Our results confirm involution as a driver of postpartum BC and indicate two roles for COX-2 in tumor cell invasion in the postpartum setting: 1) COX-2 activity in the tumor is required for motility and 2) COX-2 activity in the host promotes collagen fibrillogenesis, which promotes metastasis. The human correlative results implicate collagen and COX-2 in the poor prognosis of postpartum BC. Our data preliminarily support the idea of an NSAID intervention study for recently pregnant women at high risk for breast cancer.