The Invasive Region of Glioblastoma Defined by 5ALA Guided Surgery Has an Altered Cancer Stem Cell Marker Profile Compared to Central Tumour

Abstract

Glioblastoma, a WHO grade IV astrocytoma, is a highly aggressive and heterogeneous tumour that infiltrates deeply into surrounding brain parenchyma, making complete surgical resection impossible. Despite chemo-radiotherapy, the residual cell population within brain parenchyma post-surgery causes inevitable recurrence. Previously, the tumour core has been the focus of research and the basis for targeted therapeutic regimes, which have failed to improve survival in clinical trials. Here, we focus on the invasive margin as defined by the region with 5-aminolevulinic acid (5ALA) (GliolanTM) fluorescence at surgery beyond the T1 enhancing region on magnetic resonance imaging (MRI). This area is hypothesized to constitute unique microenvironmental pressures, and consequently be molecularly distinct to tumour core and enhancing rim regions. We conducted hematoxylin and eosin (H&amp;E), array real time polymerase chain reaction (PCR), and immunohistochemistry staining on various intra-tumour regions of glioblastoma to determine molecular heterogeneity between regions. We analyzed 73 tumour samples from 21 patients and compared cellular density, cell proliferation, and the degree of vascularity. There is a statistically significant difference between the core, invasive margin and other regions for cell density (p < 0.001), cell proliferation (p = 0.029), and vascularity (p = 0.007). Aldehyde dehydrogenase 1 (ALDH1) and Nestin immunohistochemistry were used as a measure of stem-like properties, showing significantly decreased Nestin expression (p < 0.0001) in the invasive margin. Array PCR of the core, rim, and invasive regions showed significantly increased fibroblast growth factor (FGF) and ALDH1 expression in the invasive zone, with elevated hypoxia inducing factor 1-alpha (HIF1α) in the rim region, adjacent to the hypoxic core. The influence of varying microenvironments in the intra-tumour regions is a major key to understanding intra-tumour heterogeneity. This study confirms the distinct molecular composition of the heterogeneous invasive margin and cautions against purported therapy strategies that target candidate glioblastoma stem-like genes that are predominantly expressed in the tumour core. Full characterization of tumour cells in the invasive margin is critical, as these cells may more closely resemble the residual cell population responsible for tumour recurrence. Their unique nature should be considered when developing targeted agents for residual glioblastoma multiforme (GBM).

Keywords: 5ALA; glioma stem cell; hypoxia; intra-tumour heterogeneity; invasion.

Figure 1

Intra-tumour cellular density reveals spatial heterogeneity. Cell density in all samples was assessed on H&E sections with a median of 127 cells per high-powered field (HPF) across all of the tumour regions (range 39–561 cells per HPF). (A,B) Tumour core. (C) Tumour rim. (D,E) Tumour invasive margin. (F) Core samples had a median cell density of 144 cells per HPF (range 66–335), rim samples a median of 152 cells per HPF (56–561) and invasive samples median of 76 cells per HPF (39–287). On pairwise post-hoc t-test, there were significant differences between core and invasive regions (* p = 0.018) and between rim and invasive regions (* p = 0.045) but not between core and rim regions (p = 0.849). (A–E) ×200. Error bars (A–E) 25 µm.

Figure 2

Intra-tumour proliferation index reveals lower growth rate at the GBM invasive margin. Using Ki67 immunohistochemistry we assessed the fraction of proliferating cells in each tumour region across the various samples. (A,B) Median Ki67 positive fraction was 39% for core samples (range 4–61%), (C,D) 26% for rim samples (range 2–45%) and (E) 22% for invasive margin samples (range 4–39%). On post-hoc t-test there was a significant difference between core and invasive samples (* p = 0.028) but not between rim and invasive or rim and core samples. (F) Number of cells positive on Ki67 immunohistochemistry in core, rim and invasive regions of GBM tumours. On post-hoc t-test, there was a significant difference between core and invasive samples, indicated by asterisk. (A–E) ×200. Error bars (A–E) 25 µm.

Figure 3

The GBM invasive margin is characterised by a distinct vasculature. Tumour vascularity was assessed using CD31 (PECAM-1) immunohistochemistry, identifying the number of vascular structures per HPF. (A,B) Core tumour regions showed a median of 9 vessels pre HPF (range 0–28), (C,D) rim regions a median of 7 vessels per HPF (range 0–25) and (E) invasive areas a median of 5 (range 0–12). There was statistically significant variation between areas (p = 0.025 Kruskal Wallis) with significant differences between core and invasive and between core and rim regions on pairwise t-test. (F) Relative levels of CD31 positivity between core, rim and invasive tumour areas. On post-hoc T-test, there was statistically significant variation between areas. (A–E) ×40. Error bars (A–E) 100 µm.

Figure 4

The GBM invasive margin exhibits reduced expression of canonical glioma stem cell markers. (A) Immunohistochemistry against nestin in invasive zone tissue showing low staining (23.1 ± 11%). (B) Immunohistochemistry against core tumour showing high expression of nestin (80.7 ± 31%). (C) Positive staining for ALDH1 in an invasive region of tumour (53.6 ± 29%). (D) Low levels of ALDH1 detected in the tumour core (10.3 ± 4%). (A,B) ×40, (C,D) ×200. Error bars (A,B) 100 µm; (C,D) 25 µm.

Figure 5

The GBM invasive margin is characterized by a unique stem cell transcriptomic profile. Array PCR studies of known stem related markers (including cell cycle regulators, chromosome and chromatin regulators, self-renewal and differentiation markers) demonstrated distinct stem cell expression profiles within different spatial regions of two primary patient GBMs (GBM 7, GBM 8). The specific stem cell markers ALDH1A, ABCG2, and FGF1 were consistently and significantly over-expressed (all p values < 0.05) in the invasive margin of GBM when compared to (A,B) enhancing rim, (C,D) enhancing core and (E,F) non-enhancing core tumour regions. The gap junction protein gene GJB1 was also consistently over-expressed in the invasive region. Other stem cell markers had varied expression between tumour regions on PCR, with expression of an average of 39 markers varying significantly between regions. Neurogenic locus notch homolog protein 1 (NOTCH1), lysine acetyltransferase 2A (KAT2A); sex-determining region Y-box 2 (SOX2); fibroblast growth factor receptor (FGFR1).

Figure 6

Multi-regional surgical sampling of primary glioblastoma multiforme (GBM). T1-weighted MRI scans of a representative patient, with cross-hairs depicting spatially-distinct multi-region surgical sampling. Regions include central enhancing core, necrotic core, medial and posterior rims and invasive margin (Region 5).