Inflammatory breast cancer (IBC) is characterized by numerous tumor emboli within lymphatics. In a recent study, we observed tumor embolic budding both in vitro and in vivo within lymphovascular spaces and proposed this to account for the plethora of tumor emboli seen in IBC. These observations did not address, however, how lymphovascular invasion is initiated or the mechanisms involved. In the present study, using the well-characterized patient-derived xenograft (PDX), Mary-X, which exhibited florid lymphovascular invasion (LVI) in athymic mice (LVI) as defined by E-cadherin-positive tumor emboli within lymphatic channels distinguished by podoplanin and LYVE1 membrane and Prox1 nuclear immunoreactivities and spontaneous spheroidgenesis in vitro and human cases of IBC which showed similar LVI, we compared laser-captured microdissected emboli from Mary-X and from the cases of human IBC to non-embolic areas. Mary-X and IBC emboli expressed high levels of E-cadherin and no evidence of epithelial-mesenchymal transition (EMT). Mary-X spheroids expressed high levels of VEGF, especially VEGF-C, and stimulated both vascular and lymphatic endothelial haptotaxis. We then transplanted Mary-X serially into green, cyano, red, and nestin-green fluorescing protein (GFP-, CFP-, RFP-, and nestin-GFP) transgenic reporter mice in various combinations. Multicolor murine imaging studies indicated that reporter-labeled stroma initially encircled clumps of tumor cells and then served as a scaffold that supported nestin-GFP-labeled endothelial haptotaxis resulting in encircling lymphangiogenesis, confirmed by dual LYVE1 immunofluorescence. The present studies demonstrate a possible mechanism of a critical step of the tumor emboli formation of IBC.