Microarray analysis verifies two distinct phenotypes of glioblastomas resistant to antiangiogenic therapy

Abstract

Purpose: To identify mechanisms and mediators of resistance to antiangiogenic therapy in human glioblastoma.

Experimental design: We carried out microarray gene expression analysis and immunohistochemistry comparing 21 recurrent glioblastomas progressing during antiangiogenic treatment with VEGF neutralizing antibody bevacizumab to paired pretreatment tumors from the same patients.

Results: Microarray analysis revealed that bevacizumab-resistant glioblastomas (BRG) had two clustering patterns defining subtypes that reflect radiographic growth patterns. Enhancing BRGs (EBRG) exhibited MRI enhancement, a long-established criterion for glioblastoma progression, and expressed mitogen-activated protein kinases, neural cell adhesion molecule-1 (NCAM-1), and aquaporin 4. Compared with their paired pretreatment tumors, EBRGs had unchanged vascularity and hypoxia, with increased proliferation. Nonenhancing BRGs (NBRG) exhibited minimal MRI enhancement but had FLAIR-bright expansion, a newer criterion for glioblastoma recurrence since the advent of antiangiogenic therapy, and expressed integrin α5, laminin, fibronectin1, and PDGFRβ. NBRGs had less vascularity, more hypoxia, and unchanged proliferation than their paired pretreatment tumors. Primary NBRG cells exhibited more stellate morphology with a 3-fold increased shape factor and were nearly 4-fold more invasive in Matrigel chambers than primary cells from EBRGs or bevacizumab-naive glioblastomas (P < 0.05).

Conclusion: Using microarray analysis, we found two resistance patterns during antiangiogenic therapy with distinct molecular profiles and radiographic growth patterns. These studies provide valuable biologic insight into the resistance that has limited antiangiogenic therapy to date.

Figure 1. Two types of bevacizumab-resistant glioblastomas (BRGs)

(A) T1 gadolinium-enhanced volumes as percentages of FLAIR-bright volume in BRGs clustered into 2 groups – enhancing bevacizumab-resistant glioblastomas (EBRGs) had enhancing volumes that were 50–97% of their FLAIR-bright volume, while non-enhancing bevacizumab-resistant glioblastomas (NBRGs) had enhancing volumes that were 0–10% of their FLAIR-bright volumes. Optimal partitioning of tumors based on the percentage of FLAIR bright volume that enhanced (Supplementary Methods) was obtained by division into groups of 11 (high percentage of enhancement) and 10 cases (low percentage of enhancement). (B) T1-gadolinium enhanced axial images from 3 representative EBRGs are shown, with a hematoxylin and eosin staining of the brain-tumor interface to the right at low and high magnification. (C) T1-gadolinium enhanced (upper row) and FLAIR (lower row) axial images from 3 representative non-enhancing bevacizumab-resistant glioblastomas (NBRGs) are shown, with a hematoxylin and eosin staining of the less well defined brain-tumor interface to the right in low (50x, scale bar 200 μm) and high (200x, scale bar 50 μm) power. (D) Example of perivascular tumor invasion (arrow), seen in 2/5 NBRGs and 0/6 EBRGs with tissue from the brain-tumor interface available; 200x; scale bar, 20 μm. (E) mean distance of furthest discontinuous invasion site from the tumor edge was higher in NBRGs than EBRGs (P<0.001).

Figure 2. The microenvironment of bevacizumab-resistant glioblastomas (BRGs) compared to before bevacizumab treatment

(A) representative paired immunostainings of vWF (endothelial marker; first row), CA9 (hypoxia marker; second row), and HIF-1α (hypoxia marker, third row) are shown from tumors that went on to become NBRGs (“pre-NBRG,” first column), NBRGs (second column), tumors that went on to become EBRGs (“pre-EBRG,” third column), and EBRGs (fourth column). (B) fold change in vessel density, percentage of CA9-positive tissue, and HIF-1α staining is shown for NBRGs and EBRGs, with standard deviations. Magnification/scale bars, 100×/100μm (vWF, CA9) or 400×/20 μm (HIF-1α).

Figure 3. Properties of tumor cells isolated from bevacizumab-resistant glioblastomas (BRGs)

(A) Representative Ki-67 immunostainings are shown for a pre-treatment tumor that became an NBRG (“pre-NBRG,” first column), an NBRG (second column), a pre-treatment tumor that became an EBRG (“pre-EBRG,” third column), and an EBRG (fourth column). 400×, scale bar 50 μm. (B) Fold change in the percentage of Ki-67-positive cells in NBRGs and EBRGs after bevacizumab resistance compared to pre-treatment. (C) Tumor cell density increased in a non-significant (P=0.2–0.4) manner in EBRGs and NBRGs after bevacizumab resistance. (D) Image guided biopsies were taken from enhancing (n=2) or non-enhancing (n=2) regions of bevacizumab-naïve glioblastomas, as well as from EBRGs (n=2) and NBRGs (n=2). Axial T1 gadolinium-enhanced and FLAIR images are shown from one representative case of each of the 4 groups with locations of the image-guided biopsies in green. Primary cultures of glioblastoma cells from these 4 groups stained with DAPI and phalloidin revealed that tumor cells from NBRGs (n=2; right hand column) exhibited round enlarged cytoplasm with abundant actin-positive extensions, consistent with pseudopodia, while tumor cells from enhancing or non-enhancing regions of bevacizumab-naïve glioblastomas (n=2 each; left hand most 2 columns) or EBRGs (n=2; third column from left) exhibited more polarized stellate central morphology with fewer actin-positive extensions. Shown is a representative cell from the image-guided biopsy shown on the MRI images. 400×; Scale bar, 10 μm. (E) Inverse shape factor, a measure of cell dendricity calculated from micrographs representatively illustrated in (D) was elevated in NBRG cells versus EBRG cells or cells from enhancing or non-enhancing regions of bevacizumab-naïve recurrent glioblastomas (P<0.05).

Figure 4. Invasion and migration of tumor cells isolated from bevacizumab-naïve and bevacizumab-resistant glioblastomas (BRGs)

Cells were isolated from image-guided biopsies (green spots on MRIs) taken from the enhancing versus non-enhancing regions of bevacizumab-naïve glioblastomas, and from non-enhancing regions of an NBRG and enhancing regions of an EBRG (n=2 per group; location of biopsies from both patients in each of the 4 groups are shown on T1 gadolinium-enhanced and FLAIR axial MRI images). Shown in the upper graph are the number of cells from enhancing and non-enhancing areas of bevacizumab-naïve recurrent glioblastomas, EBRGs (n=2), and NBRGs invading through matrigel-coated Boyden chambers 15 hours after plating cells from fresh tissue specimens, a measure of invasion, with NBRG cells showing more invasion than cells from EBRGs or enhancing or non-enhancing areas of bevacizumab-naïve recurrent glioblastomas (P<0.05). Shown in the lower graph are the number of cells from these same 4 groups invading through matrigel-coated Boyden chambers 15 hours after plating cells taken from fresh tissue specimens, a measure of migration, with no change between groups (P=0.6).

Figure 5. Microarray analysis of bevacizumab-resistant glioblastomas (BRGs)

(A) Unsupervised clustering of 18 tumor samples (9 before bevacizumab treatment and 9 after developing bevacizumab resistance) revealed no particular pattern of clustering. (B) An unbiased gene selection approach in which 9 differential gene expressions in 9 BRGs compared to their pre-treatment paired specimens were clustered on the high variance probes (top 98th percentile and above, 491 probes encompassing 478 unique genes; plotted data is median-centered) revealed that NBRGs clustered distinctly from EBRGs. There were 43 genes differentially expressed in NBRGs and 146 in EBRGs with adjusted P<0.05, including integrin α5, fibronectin1, laminin, and PDGFRβ in NBRGs and NCAM-1, aquaporin 4, and MAP kinases 4 and 10 in EBRGs. (C) Hierarchical clustering of the patterns of differential gene expression in BRGs compared to their paired pre-treatment specimens. Values at branches are AU p-values (left), BP values (right), and cluster labels (bottom). Clusters with AU≥95 are indicated by red rectangles. Case numbers in blue. (D) Real-time RT-PCR confirmed differential expression of integrin β1 (ITGB1) and CXCL12 in NBRGs (n=5) compared to EBRGs (n=6) (P<0.05).