Continuous local delivery of interferon-β stabilizes tumor vasculature in an orthotopic glioblastoma xenograft resection model

Abstract

Background: High-grade glioblastomas have immature, leaky tumor blood vessels that impede the efficacy of adjuvant therapy. We assessed the ability of human interferon (hIFN)-β delivered locally via gene transfer to effect vascular stabilization in an orthotopic model of glioblastoma xenograft resection.

Methods: Xenografts were established by injecting 3 grade IV glioblastoma cell lines (GBM6-luc, MT330-luc, and SJG2-luc) into the cerebral cortex of nude rats. Tumors underwent subtotal resection, and then had gel foam containing an adeno-associated virus vector encoding either hIFN-β or green fluorescence protein (control) placed in the resection cavity. The primary endpoint was stabilization of tumor vasculature, as evidenced by CD34, α-SMA, and CA IX staining. Overall survival was a secondary endpoint.

Results: hIFN-β treatment altered the tumor vasculature of GBM6-luc and SJG2-luc xenografts, decreasing the density of endothelial cells, stabilizing vessels with pericytes, and decreasing tumor hypoxia. The mean survival for rats with these neoplasms was not improved, however. In rats with MT330-luc xenografts, hIFN-β resulted in tumor regression with a 6-month survival of 55% (INF-β group) and 9% (control group).

Conclusion: The use of AAV hIFN-β in our orthotopic model of glioblastoma resection stabilized tumor vasculature and improved survival in rats with MT330 xenografts.

Figure 1. Orthotopic glioblastoma xenograft resection model

A, positioning in the stereotactic frame and exposure of the skull; B, creation of a burr hole; C, completed burr hole, ready for implantation of cells; D, pre-resection bioluminescence image; E, neurotip suction used to subtotally remove the tumor; F, post-resection bioluminescence image.

Figure 2. In vitro sensitivity of glioblastomas to rhIFN-β

GBM6-luc, MT330-luc, SJG2-luc, and Hs294T cells were exposed to increasing dose of rhIFN-β for 96 hours, and the number of viable cells was quantified using the AlamarBlue Assay. All three glioblastomas are only slightly sensitive to the direct effects of rhIFN-β in vitro.

Figure 3. AAV serotype tissue specificity

AAV serotypes 2, 5, and 8 containing the firefly luciferase gene were injected into the brain of SCID mice. Serotype 5 showed a high level of local expression, while serotypes 2 and 8 were found to have significant levels of transduction at remote sites.

Figure 4. AAV-2/5 tissue distribution

Genomic DNA was extracted from various tissue sites (tumor, peritumoral ipsilateral hemisphere, uninvolved ipsilateral hemisphere, contralateral hemisphere, testes, and liver) and CMV primers were used to detect the presence of vector by qPCR. High levels of the vector DNA were detected in the tumor and peritumoral tissue, while very low levels were found in the liver and testes. Error bars refer to standard error. n = 3 in each group.

Figure 5. hIFN-β stabilizes tumor vasculature and decreases tumor hypoxia

Xenografts were stained for CD34, αSMA, and CAIX. Images of stained slides, magnified 400 times, were quantified as pixels per HPF and treatment groups were compared. A, data for GBM6 xenografts, *p=0.0006, **p=0.045, ***p=0.0009; B, data for SJG2 xenografts, *p=0.0007, **p=0.0008, ***p=0.0065; C–D, SJG2 xenografts stained for CD34; E–F, SJG2 xenografts stained for αSMA; G–H, GBM6 xenografts stained for CA IX. Error bars refer to standard error. n = 5 in each group.

Figure 6. Effects of AAV hIFN-β on MT330 tumor progression and overall survival

A, bioluminescence signal preoperative, postoperative, 7 days postoperative, and 14 days postoperative for rats with MT330 tumors. B, Kaplan-Meier survival curve for rats with MT330 tumors, with 6-month survival of 9% for GFP (control) (n = 11) and 55% for hIFN-β (n = 11), *p=0.01. Error bars refer to standard error.