Abstract Background Recurrent glioblastoma (GBM) is today inevitable, even using an aggressive treatment protocol (surgery, radiotherapy, and chemotherapy). It mostly occurs within 2 cm of the resection cavity due to the invasion ability of GBM cells. The aim of this work is to test and qualify a local integrative approach using a nano-based chemokine-cancer cell trap concept to lure the residual GBM cells by releasing chemoattractive molecules and, ultimately, confine and kill them in a biomimetic nanostructured polymeric scaffold. Material and Methods The stromal cell-derived factor 1α (SDF-1α) was used here as a chemoattractant due to its synergy with the CXCR4 receptor, and the correlation of this axis to the migration of infiltrating GBM cells. Freeze-dried sponges (20 to 140μm/pore size) composed of silk fibroin (SF), hyaluronic acid (HA), and heparin (hep) charged with SDF-1α was evaluated as a GBM cell trap. First, the cytocompatibility and biodegradability of the sponges were evaluated both in vitro and vivo. Secondly, molecular and functional responses of GBM cells to SDF-1α treatment were evaluated using U87MG-CXCR4+ cells, that constitutively and stably expressed both the CXCR4 receptor and RFP, as reporters. To assess the cell migration in response to the SDF-1α gradient in vitro, transwell, under-agarose, and agarose drop assays were used. U87MG-CXCR4+ spheroids were cultivated in the sponges to verify the hosting capacity of the scaffolds. Finally, syngeneic and xenogeneic models were developed to evaluate the bioperformance of the sponges in vivo. Results SF-HA-hep sponges presented low cytotoxicity and complete biodegradation after 4 months of implantation. Histological analysis revealed only local reaction, as an immune response in the short-term study, but no signs of acute inflammation or fibrotic process were found at the resection cavity-brain margin in the long-term study, demonstrating fine tolerability in vivo. Relevant molecules involved in chemotaxis (Akt, Erk, and Paxillin) were activated upon treatment with SDF-1α. U87MG-CXCR4+ cells were directionally attracted to the SDF-1α-loaded sponges, and also a larger neurosphere infiltration area was observed in these sponges compared to controls. Preliminary observations have shown cell clusters detached from the main tumor mass towards the cavity with the implanted SDF-1α-loaded sponge. This may be indicative of chemoattraction, as DAPI correlated with Ki67 staining in the migration front. Further analyses including CXCR4, CD44, paxillin, and galectin-1 immunolabeling will be performed to confirm this hypothesis. Conclusion Overall, the results showed that the sponges are biocompatible and the SDF-1α suitable to use as a chemoattractant for GBM cells. This strategy can be used to overcome the treatment resistance in GB, and potentially be transferred to the clinic in the future.