2.1 Patients and clinical specimens

We performed a retrospective study of 80 patients with GBM treated at Sun Yat-sen University Cancer Center (SYSUCC, Guangzhou, Guangdong, China) between January 2000 and December 2012. All included patients were pathologically confirmed as GBM according to the 2016 World Health Organization (WHO) classification for central nervous system tumors. All patients were treatment-naive before surgery.

The formalin-fixed, paraffin-embedded (FFPE) tissue specimens of GBM from the selected patients were retrieved from the Department of Pathology at SYSUCC. Tumor sections were reviewed by two neuropathologists to verify the GBM diagnosis. This study was approved by the Ethics Committee of SYSUCC (case number: GZR2013-052). All patients provided written informed consent.

2.2 Mice

Four-week-old nude mice weighing 16 to 20 g were obtained from the Medical Experimental Animal Center of Guangdong Province (Guangzhou, Guangdong, China). All nude mice were housed under standard conditions and cared for according to the institutional guidelines for animal care. All animal procedures were performed under the ethical standards of the Accreditation of Laboratory Animal Care and approved by the Ethics Committee of SYSUCC.

2.3 Cell culture

The glioma stem-like cells (GSC-1), stably expressing CD133 and sex determining region Y-box 2 (Sox2), were generated from tumor tissue of a patient with GBM(18) and cultured in Dulbecco's Modified Eagle Medium/Nutrient Mixture F-12 (DMEM/F12; Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 20µL/mL B27 (Thermo Fisher Scientific), 20 ng/mL epidermal growth factor (EGF; Peprotech, Rocky Hill, NJ, USA), 20 ng/mL basic fibroblast growth factor (bFGF; Peprotech), and 1% penicillin/streptomycin (Thermo Fisher Scientific, Waltham, MA, USA) at 37°C in 5% CO₂.

2.4 Xenograft specimens

A total of 1 × 10⁶ GSC-1 cells in 200 µL serum-free medium were inoculated subcutaneously into the flanks of the nude mice. Eleven days after the inoculation, a tumor mass was observed. All nude mice were euthanized 18 days after inoculation, and the xenograft tumor tissues were collected, fixed with 10% formalin, and embedded in paraffin for further experiments.

2.5 Immunohistochemistry (IHC)

Standard streptavidin-biotin-peroidase complex method was used for IHC staining. FFPE sections were stained with an antibody against CD34 (1:1000; ab198395, Abcam, Boston, MA, USA) and periodic acid-Schiff (PAS; BD, Franklin Lakes, NJ, USA), as previously described [7]. Briefly, FFPE sections were deparaffinized by xylene and absolute ethanol, rinsed in distilled water and exposed to 3% H₂O₂ for 10 min. Slides were put in ethylenediaminetetraacetic acid (EDTA) antigen-unmasking solution heating in oven at 90°C for 15 min and then cooled to room temperature in unmasking solution. Slides were then washed in phosphate-buffered saline (PBS) and incubated with the primary antibody overnight at 4°C. After being washed in PBS, the slides were incubated with the secondary antibody (1:1000; Golden Bridge, Beijing, China). The nuclear expression of CD34 was visualized with chromogen 3,3ʹ-diaminobenzidine (DAB; Golden Bridge). Slides were then rinsed with distilled water for 5 min and incubated with PAS stain for 15 min, and the cell nuclei were stained with hematoxylin.

2.6 Immunofluorescence (IF) staining

The deparaffinization, rehydration, and antigen retrieval of tumor sections were as described in IHC. Slides were washed with PBS before incubation with the first primary antibody, rabbit monoclonal anti-CD34 (1:1000; ab198395, Abcam), at 4°C overnight. After being washed in PBS, the sections were incubated with mouse monoclonal anti-GFAP (1:1000; ab10062, Abcam) or anti-laminin antibodies (1:1000; ab7463, Abcam). Next, slides were washed with PBS and incubated with Alexa Fluor® 488-conjugated goat anti-rabbit IgG (1:1000; O-11038; Life Technologies, Carlsbad, CA, USA) and Alexa Fluor® 594-conjugated goat anti-mouse IgG (1:1000; AK-A11005; Life Technologies) for 40 min at room temperature. Nuclear staining was performed with 4′,6-diamidino-2-phenylindole (DAPI; D3571; Life Technologies), and the slides were then mounted with Slow Fade Gold antifade reagent (S36938, Thermo Fisher Scientific).

2.7 Observation and density assessment of microcirculation patterns

The IHC- and IF-stained slides were reviewed by Xin Mei and Yin-Sheng Chen using optical microscope (Olympus Corporation, Shinjuku,Tokyo, Japan) for IHC and confocal laser microscope (Olympus Corporation) for IF following the methods described by Folberg et al. [19]. EDV had a CD34-positive endothelium wall. ECMDV had extracellular matrix surrounding the lumen but without CD34 or GFAP staining. Vessels with only a GBM cell lining were regarded as GDVs, and vessels comprising a mixture of tumor cells and endothelial cells were regarded as MVs (Supplementary Table S1). Vessel density was assessed as described by Paschoal et al. [20]. Briefly, each IHC- or IF-stained section was first observed at low optical power to choose two regions of interest (ROI). The number of vessels expressing the biomarkers was then counted at high magnification (high optical power, 400 ×) within the ROI. Next, the mean microvessel density (MVD) was calculated.

2.8 Three-dimensional (3D) reconstruction of microcirculation patterns

We reconstructed the microcirculation patterns in human GBM tissue sections (1 mm thickness) with IF staining for CD34 and GFAP and analyzed it using Fluoview 3.0 software (Olympus Corporation). Images were stored in TIFF format. One 3D reconstruction picture comprised more than twenty image frames, and the thickness of each layer was 2 µm. The visualization of the 3D shapes of the cell nuclei (blue), endothelial vessels (green), and GBM cells (red) was achieved by representing the traces of these objects using “Reconstruct”. The 3D models were saved as a 360° image series and then exported into Fluoview 3.0 software, where the files were compressed (JPEG).

2.9 Methylation-specific PCR (MSP) of MGMT

Sufficient tissue was available from 61 GBM patients to determine the methylation status of the MGMT promoter region using MSP. Genomic DNA was extracted from paraffin-embedded specimens according to the manufacturer's instructions (TianGen DNA Mini Kit, Beijing, China). The methylation status was determined by performing the bisulfite modification, which converts unmethylated but not methylated cytosines to uracils. The genomic DNA (50 ng) was amplified using primers designed to detect the methylated or unmethylated sequences by MSP. MSP was performed according to the manufacturer's protocol (EZ DNA Methylation Gold kit, Zymo Research, Irvine, CA, USA). The primer sequences were as follows: forward primer, 5′-GTTATGAATGTAGGAGCCCTTATG-3′; reverse primer, 5′-TGACAACGGGAATGAAGTAATG-3′, which were specific for either the methylated or modified unmethylated DNA [21]. DNA obtained from normal peripheral blood lymphocytes (PBLs) served as the negative control, and enzymatically methylated DNA from PBL was used as the positive control. Ten microliters of each 50-µL MSP product were directly loaded onto nondenaturing 6% polyacrylamide gels, stained with ethidium bromide, and examined under ultraviolet illumination.

2.10 Detection of IDH1/2 mutations

Bidirectional Sanger sequencing (Qiagen, Hilden, Germany) was performed to detect the mutations of IDH1/2. The primers for IDH1 (129 bp for codon 132) and IDH2 (290 bp for codon 172) were listed here: IDH1-F-5′-CGGTCTTCAGAGAAGCCATT-3′, IDH1-R-5′-ATTCTTATCTTTTGGTATCTACACC-3′, IDH2-F-5′- GGGGTTCAAATTCTGGTTGA-3′, IDH2-R-5′-ACTGGAGCTCCTCGCCTAG-3′.

Briefly, PCR was performed in a total volume of 60 µL using 15 ng of DNA and Taq Polymerase Platinum in standard PCR buffer (BioRad Laboratories, Pleasanton, CA, USA). An initial denaturation step at 94°C for 2 min was followed by 35 cycles of 94°C for 30 s, 55°C for 30 s, and 68°C for 30 s. PCR products were purified using the QIA quick 96 PCR Purification Kit (28181; Qiagen). The purified PCR products (1.5 µL) were sequenced using Applied Biosystems Big Dye 3.1 dideoxy chain termination sequencing chemistry in conjunction with fluorescent-based capillary sequencers (Model 3730xl, Applied Biosystems Division of Perkin Elmer, Foster City, CA, USA) for electrophoresis, data collection, and base calling.

Locked Nucleic Acid (LNA) Sanger sequencing was performed in a separate reaction by amplifying the mutation hotspot regions using the pyrosequencing PCR primers and LNA (TAGGTCGTCA/3InvdT/) at a final concentration of 70 nmol/L. The PCR product was sequenced using the Big Dye Terminator v1.1 Sequencing Kit (Applied Biosystems; Foster City, CA, USA).

2.11 Statistical analysis

The constituent ratios of the microcirculation patterns were separately calculated as the mean value of each sample stained with IHC and IF. The microcirculaiton pattern was determined by the staining results from either IHC or IF. We then compared the prognosis of patients with different microcirculation patterns. The overall survival was defined from the date of diagnosis to the date of the death or to the last observation date (December 31, 2018). Multivariable analysis was performed including gender, age and MGMT promoter methylation status for the prognosis of GDVs. The hazard ratios and 95% confidence intervals for patient outcome were estimated by univariate and multivariate Cox regression. Kaplan-Meier survival curves were generated for each set. Statistical analyses were performed using SPSS version 19 (SPSS Inc., Chicago, IL, USA), and P values less than 0.05 were considered significant.

3.1 Microcirculation patterns in GBM

Four microcirculation patterns, EDV, ECMDV, GDV, and MV, were assessed in both clinical GBM and xenograft specimens by IHC and IF staining. Among them, EDVs existed in all tissue specimens. However, ECMDVs, GDVs, and MVs were only observed in 27.5%, 43.8%, 52.5% of specimens from GBM patients and 40.0%, 5.0%, 70.0% of the specimens from xenografts, respectively.

Most of the intratumoral vessels (90.2%) in GBM tissues were EDVs. They showed positive staining of CD34 (an endothelial marker) on the internal surface of the lumen and PAS on the vessel wall (Figure 1a and c), indicating the high angiogenic ability of GBM tissues. High microvascular proliferation was also observed by IF staining (Figure 1b and d). Abnormal nuclei were observed in some CD34-positive vessels (Figure 1b, indicated by white arrow).

ECMDVs appeared as PAS-positive vessels with no cell lining but with erythrocyte shadows in the lumen (Figure 2a and c). To verify the distribution of ECMDVs, we treated the tissue sections with an antibody against laminin (an extracellular matrix marker). As distinct from EDVs, the vessel wall of ECMDVs was stained positive for laminin but was negative with CD34 (Figure 2b and d).

GDVs were also observed in tissue specimens containing red blood cells. They appeared to be lined by GBM cells (marked by PAS or GFAP) but not by endothelium (marked by CD34) when detected by IHC staining (Figure 3a and c). Similarly, GDVs were also observed in most of the tissue sections when detected by IF staining. As shown in Figure 3b and d, we observed GFAP-positive but CD34-negative staining of the cells surrounding the lumen in GDVs (expression of GFAP is a characteristic of GBM cells).

MVs are defined as blood vessels containing both GFAP-expressing tumor cells and CD34-positive endothelial cells in the same vessel wall. IHC staining of GBM specimens revealed the presence of MVs. Red cells were observed in some GDVs which were connected with EDVs in both clinical specimens and xenografts (Figure 4a and c). MVs contained endothelial cells (CD34-positive) and GFAP-positive GBM cells contributed to the formation of the vascular tubes in both human GBM specimens and mice xenografts (Figure 4b and d). The cells lined inside of GDVs were CD34 positive by IF staining, which were simillar to the GSC-1 cells expressing both CD34 and GFAP.

3.2 MVD evaluation for four microcirculation patterns in GBM specimens

The MVDs determined by IHC and IF in both clinical and xenograft specimens are listed in Table 1. The MVD of four microcirculation patterns varied slightly and this depended on the detection method used. In the tissue sections stained by IHC, EDVs had the highest MVD and then followed by ECMDVs. However, in the tissue sections stained by IF, EDVs still had the highest MVD, and followed by MVs.

3.3 Three microcirculation patterns comprise a novel blood supply system in GBMs

All microcirculation patterns accessed in this study were observed in GBM specimens. Since both ECMDV and GDV are subtypes of VM [22], we analyzed VM vessels in GBM specimens by combining the data of ECMDV and GDV. The percentages of tissue samples that had VM were 38.8% (31/80) in clinical specimens and 20.0% (9/45) in xenograft specimens when detected by IHC. Additionally, when detected by IF, those percentages were 41.3% (33/80) and 20.0% (9/45), respectively.

To explore the 3D structures of the microcirculation patterns in GBM specimens, IF images of dual-stained sections were further analyzed by 3D reconstruction. GFAP-positive GBM cells lined the GDV walls (Figure 5a). Both CD34-negative and CD34-positive GBM cells, along with endothelial cells, lined the MV walls(Figure 5b and c). Moreover, CD34-positive endothelial cells formed EDVs (Figure 5a). GDVs connected with EDVs in some areas of the samples (Figure 5a). Our results therefore confirmed that MVs could be formed by GDVs and EDVs. Three microcirculation patterns (MVs, GDVs and EDVs) connected with each other and comprised a novel blood supply system for GBM.

3.4 Patient characteristics and survival analysis

Sixty-four (80.0%) of the 80 GBM patients with complete follow-up data were selected for survival analysis. The median age of the patients was 47.6 years (range: 9.0-78.0 years), and 19 (29.7%) patients were female. The median Karnofsky Performance Status Score (KPS) was 78 (range: 20-90), and 15 (23.4%) patients had a KPS < 80 (Supplementary Table S2). All patients received standard radiotherapy and chemotherapy after surgery.

The methylation status of the MGMT promoter was detected in the samples from 61 of the 64 patients. We found that the MGMT promoter of 38 (59.4%) patients was methylated. IDH1/2 mutations were examined in eligible samples from 55 of the 64 patients, and 7 (10.9%) samples had IDH1 mutations, while none had IDH2 mutations.

The median survival of the 64 patients was 12.13 months (range: 1.20-149.50 months). Until the last follow-up, 3 (42.9%) of 7 patients with IDH1 mutations survived 149.50, 87.40, and 82.70 months, respectively. GBM patients with MGMT promoter methylation had a median survival of 12.33 months (range: 3.10-149.50 months) compared with 10.00 months (range: 1.20-82.70 months) for patients without MGMT promoter methylation (P = 0.037).

Patients with tumors containing GDVs had a median survival of 9.56 months (range: 1.20-22.83 months) compared with 13.60 months (range: 2.73-149.50 months) for patients without GDVs (P = 0.015; Figure 6a). In the MGMT promoter methylation cohort, patients with GDV-positive tumors had a median survival of 6.76 months (range: 3.53-20.93 months) compared with 14.23 months (range: 3.13-149.50 months) for patients with GDV-negative tumors (P = 0.022; Figure 6b). In the IDH1 wild-type cohort, patients with GDV-positive tumors had a median survival of 10.00 months (range: 1.20-20.93 months) compared with 13.60 months (range: 2.73-83.20 months) for patients with GDV-negative tumors (P = 0.096; Figure 6c). We performed multivariable analysis including gender, age and MGMT promoter methylation status for the prognosis of GDVs. The results showed that the presence of GDVs predicted a poor prognosis for GBM patients (hazard ratio [HR] = 2.21; 95% confidence interval [CI] = 1.13-4.32; P = 0.020), especially for the patients with MGMT promoter methylation (HR = 1.178; 95% CI = 0.65-2.11; P = 0.018). The results of survival analysis for GBM patients with other microcirculation patterns in tumors are summarized in Supplementary Table S3. GDVs might be a negative predictor for the survival of GBM patients, even in those with MGMT promoter methylation.

ETHICS APPROVAL AND CONSENT TO PARTICIPATE

All procedures performed in this study involving animal experiments were in accordance with the ethical standards of the Accreditation of Laboratory Animal Care and approved by the Ethics Committee of Sun Yat-sen University Cancer Center (GZR2013-052).