Recently, it was shown that DMXAA exhibits differential effects on murine and human macrophages, and also that the stimulator of interferon genes, was a receptor for DMXAA. The finding that DMXAA was unable to activate human STING provided a salient explanation for the failure of this agent in the human clinical trials. Our findings using both DMXAA and 2939-cGAMP suggest that STING activation was the common factor leading to M2 macrophage re-polarization, a process that undoubtedly played a role in mediating the vascular disrupting effects of DMXAA we observed on the subcutaneous 344SQ-ELuc tumors. Interestingly, however, we found that the vascular disrupting effects of DMXAA on subcutaneous tumors did not extend to either the 344SQ-ELuc metastases obtained following intracardiac injection of these cells, or to the spontaneously arising tumors in the K-rasLA1/+ GEMM model of NSCLC. Indeed, the majority of successful pre-clinical studies evaluating the utility of DMXAA were carried out in subcutaneous tumor models, with relatively few studies examining the effects of DMXAA on tumors in other anatomical sites. Echoing our results, experiments employing a different vascular disrupting drug, flavone acetic acid, vessel disruption was seen in subcutaneous tumors, but not systemic tumors. Thus, although the inability of DMXAA to activate human STING provided an obvious reason for failure of DMXAA in human cancer trials, our results nevertheless suggest that vascular disruption might not occur in either primary or metastatic human NSCLC if human STING agonists were administered. With regard to correcting this defect, considerable efforts are now underway involving the development of stable cyclic dinucleotide analogs that will allow human STING activation. Differences in the density of TAM infiltration amongst different tumor sites may have been one of several factors accounting for the differential effects of DMXAA on subcutaneous versus 344SQELuc metastases. Supporting the key role of infiltrating TAMs, we found that their depletion in subcutaneous 344SQ-ELuc tumors prevented DMXAA-induced intra-tumoral hemorrhagic necrosis. In the case of 344SQ-ELuc subcutaneous tumors, TAMs were present as a dense rim at the tumor periphery and were thus well positioned to support DMXAA-induced vascular disruption via the production of pro-inflammatory mediators. There were extensive regions of ischemic necrosis invariably present within the subcutaneous 344SQ-ELuc tumors, signifying the availability of macrophage activating environmental factors such as hypoxia and the Toll-like receptor 4-activating protein, high mobility group, HMGB1 protein. Thus, in addition to quantitative differences in macrophage infiltration density between tumor sites, there may have been substantial qualitative differences between the TAMs at different tumor sites.