Reply to J.W. Locasale et al

We appreciate the thoughtful comments by Locasale et al¹ regarding the tumor biomarker analysis from our recent phase II clinical trial of aflibercept for recurrent glioblastoma (NABTC 06-01).² Our analysis demonstrated that elevated tumor levels of hypoxia-inducible factor 1α and carbonic anhydrase 9 (CA9), both markers of hypoxia, were strongly correlated with patient response to aflibercept, whereas patients with elevated CXCR4 were more likely to be resistant. Because hypoxia is known to induce CXCR4 expression, Locasale et al suggest that there may be a positive feedback loop whereby hypoxia-inducible factor 1α induces CXCR4 expression in resistant tumors, which in turn activates signal transducer and activator of transcription 3 (STAT3).

We strongly agree with the authors' conclusions that both CXCR4 and STAT3 represent potential mediators of resistance to antiangiogenic therapy. There is a complex interaction between the tumor and its microenvironment, where both CXCR4 and STAT3 may be expressed in either or both compartments. Antiangiogenic therapy targets the tumor microenvironment, and not surprisingly, the tumor microenvironment plays a central role in mediating resistance to anti–vascular endothelial growth factor therapy. CXCL12 (formerly known as SDF-1α) elaborated from tumor or the microenvironment can activate CXCR4 on tumor cells in an autocrine or paracrine fashion, leading to downstream activation of the phosphoinositide 3 kinase/AKT, STAT3, and MAPK pathways to promote tumor survival, angiogenesis, metastasis, and invasion.³

STAT3 is a central signaling node in both tumor and infiltrating immune cells.⁴ Preclinical studies have recently demonstrated the central role of STAT3 in regulating glioma proliferation, angiogenesis, and immune evasion and in prosurvival effects on glioma stem cells.^5–7 Although CXCR4 and STAT3 are expressed in glioblastoma cells, they are also expressed on multiple myeloid cell types including monocytes/macropahges, granulocytes, dendritic cells, and myeloid-derived suppressor cells, which may mediate resistance to antianigogenic therapy.^8,9 Hypoxia exacerbated by prolonged or high-dose antiangiogenic therapy may increase CXCL12 expression and the attraction of CXCR4+ myeloid cells,¹⁰ which are responsible for mediating vascular endothelial growth factor–independent angiogenesis. Interestingly, recent studies have demonstrated that hypoxia also attracts and induces regulatory T-cells, which promote immune tolerance and angiogenesis.^11,12 Consistent with this idea, our analysis showed that tumors expressing markers of lymphocyte proliferation such as CXCR4, CD84, ZAP70, and CD27 were resistant to aflibercept. Characterization of a positive feedback loop will ultimately require additional studies, because this level of complexity cannot be gleaned from undissected tumor tissue analysis; rather, it will require analysis of gene expression after microdissection and preclinical studies.

Antiangiogenic therapy has limited efficacy in glioblastoma, because tumors rapidly develop resistance. Careful analysis of tumor tissue and blood biomarker data is critical to develop a full understanding of the potential mechanisms of response and resistance. The data from our trial will inform the next generation of trials using combinations of antiangiogenic agents and targeting resistance pathways, including inhibitors of CXCR4 and STAT3.

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The author(s) indicated no potential conflicts of interest.