The subjects recruited for the present study provided informed consent and consent for publication. Consent for involvement of patients without the ability to suitably make their own decisions was provided by their legal guardian. Surgical specimens from 7 patients (referred to as patients 1–7) with primary high-grade glioma who underwent surgery at the Daping Hospital were collected between December 2016 and December 2017. Tumors were graded according to the World Health Organization (WHO) classification of tumors of the nervous system (20). Each tumor specimen was divided into three sections under sterile conditions. The first section was used to extract primary tumor stem cell spheres, which were subsequently used to establish orthotopic xenograft models. The second section was embedded in paraffin for histopathological analysis and the third section was used for transcriptome sequencing. The study involving patients was approved by the Human Research Ethics Committees of Daping Hospital at the Army Medical University (Chongqing, China; approval no. 2014-9).

The xenografts were grown, and MRI was performed on the tumor-bearing mice during the later stages of tumor growth. Tumor tissues were obtained for histopathological staining and transcriptome sequencing. All non-obese diabetic-severe-combined immunodeficiency (NOD-SCID) nude mice used in the present study were purchased from the Department of Experimental Animals (Daping Hospital, Army Medical University). All protocols involving the use of animals were performed according to the International Principles of Laboratory Animal Care (21) and were approved by the Animal Use Subcommittee of the Daping Hospital at the Army Medical University. Tumor-bearing mice were sacrificed via cervical vertebra dislocation when they exhibited clinical signs that suggested impending death, such as emaciation, weakness or obvious spinal curvature. Confirmation of euthanasia was performed by assessing cardiac arrest and mydriasis. All mice included in the present study exhibited a single tumor, the maximum level of cachexia observed was a body weight loss ≤20% compared with the age-matched controls. The mean diameter of the xenograft was <1.2 cm and the maximum diameter was 1.27 cm.

MRI scans were performed using a 3.0 Tesla MRI scanner (Magnetom Verio; Siemens AG) with a 16-channel head coil. The conventional MRI included axial and sagittal T₁-weighted, and T₂-weighted sequences. The sequence parameters were as follows: T₁-weighted imaging (T₁WI), repetition time (TR)/echo time (TE)=250/2.67 ms, slice thickness=5 mm, field of view (FOV)=230×230 mm; T₂-weighted imaging (T₂WI), TR/TE=4,900/96 ms, SL=5 mm, FOV=230×230 mm; DWI-MRI: TR=6,600 ms, TE=100 ms, b=0, 500, 1,000 sec/mm², slice thickness=5 mm. DCE-MRI was performed as follows: Two sets of T₁-weighted images were scanned with the T₁-vibe sequence [TR/TE=5.08/1.74 ms, FOV=260×260 mm, matrix=138×192, slice thickness=5 mm, flip angle (FA)=2° and 15°] and subsequently 75 consecutive scans were performed using the T₁-twist sequence (TR/TE=4.82/1.88 ms, matrix=138×192, slice thickness=3.6 mm, FOV=260×260 mm, FA=12°) at 5.3 sec intervals. At the 6th phase, 0.1 mmol/kg gadolinium contrast (Ominscan, GE Healthcare) agent was injected via the elbow vein at a rate of 4 ml/sec.

ADC maps were calculated automatically using a Siemens syngo MR Workstation (version VE36A; Siemens AG) based on DWI-MRI scanning. Subsequently, the ADC value of the tumor was calculated using the hot-spot method (22). A total of five regions of interest (ROIs) with lower ADC values were selected in the tumor area and the ADC values of the contralateral healthy brain tissues were measured. The relative apparent diffusion coefficient (rADC) values were calculated from the ratio of the tumor area to the healthy brain tissue area. The rADC value of the tumor was represented by the average rADC value of the five ROIs.

The DCE-MRI data were imported into a GE workstation and OmniKinetics software (version 2.0; GE Healthcare) was used for analysis. An Extend-Tofts model was selected as the hemodynamic model (23). The arterial input function (AIF) was calculated by placing an ROI on the middle cerebral artery and the time and signal intensity curve of the brain functions were obtained from the AIF. Subsequently, the software estimated the K_trans map. Five ROIs with higher K_trans values were selected in the tumor area and the K_trans value of the tumor was represented by the average K_trans value of the five ROIs. The MRI data were analyzed by two experienced researchers.

Fresh glioma tissue from the ~1 mm³ pieces were digested in papain (Worthington Biochemical Corporation) at 37°C for ~15 min, filtered through a 200 µm filter and centrifuged at 300 × g for 3 min at room temperature. The supernatant was discarded and the cells were resuspended in DMEM/F-12 supplemented with N-2, B-27 (all from Gibco; Thermo Fisher Scientific, Inc.), epidermal growth factor (20 ng/ml; Sigma-Aldrich Merck KGaA) and basic fibroblast growth factor (20 ng/ml; PeproTech, Inc.) and cultured in an incubator with 5% CO₂ at 37°C (24).

Xenografts were established in the right basal ganglia of NOD-SCID nude mice. The glioma stem cell spheres were digested with trypsin (Gibco; Thermo Fisher Scientific, Inc.) for 3 min and the supernatant was discarded by centrifugation. The pellet was resuspended in PBS to a density of ~1×10⁴ cells/ml. Following anesthesia with 5% chloral hydrate (300 mg/kg), a total 5 µl cell suspension was aspirated with a micro-injector and the needle was vertically inserted at 1.8 mm posterior and 2.2 mm to the right of the intersection between the midline and the posterior canthus line of the brain in the NOD-SCID nude mice. The needle was initially inserted to a depth of ~3.5 mm and withdrawn to 0.5 mm depth. The cell suspension was slowly injected (at ~1 µl/min) and the needle was withdrawn following ~10 min of cell transfer. Glioma stem cell spheres form each human glioma tissue were implanted into 5 NOD-SCID nude mice. In total, 35 mice were used in the present study.

A Bruker 7.0 T MRI scanner for small animals (BioSpec 70/20 USR; Bruker Corporation) was used and MRI was performed during the later stages of xenograft growth. After anesthesia with isoflurane (3% for induction and 1.5% for maintenance), tumor bearing mice were fixed in a flat position and scanned in the later phases of tumor growth. The MRI scanning included coronal T₂WI, sagittal T₁WI, T₁WI contrast enhanced, DWI-MRI and DCE-MRI. The sequence parameter settings for T₂W1 were as follows: Turbo-RARE sequence, repetition time/echo time=4,000 ms/45 ms, FOV=25×25 mm, matrix sizes=256×256, slice thicknesses=0.5 mm. Similarly for T₁WI, the following parameter settings were used: TR/TE=600 ms/14 ms, FOV=25×25 mm, slice thicknesses=0.5 mm. For DWI, the parameter settings were as follows: TR/TE=3,000 ms/50 ms, FOV=35×35 mm, slice thicknesses=0.5 mm, b=0, 500, 1,000 sec/mm². Finally, for DCE-MRI the FLASH sequence was used with a repetition time/echo time of 25.0 ms/1.8 ms. In addition, the following settings were used: FOV=25×25 mm, matrix sizes=128×128, slice thicknesses=0.5 mm, slices=5, and FA=5°/15°/20°/30°. FAs of 5°, 15°, 20° and 30° were used to perform pre-contrast scans and subsequently 100 consecutive scans were performed with an FA of 15°. At the 4th phase, 0.1 mmol/kg gadolinium contrast agent was administered via tail-vein injection manually within 3 sec.

ADC maps were calculated automatically using Bruker image display and processing software (Paravision version 6.0.1; Bruker Corporation) based on DWI-MRI scanning. A total of five ROIs with lower ADC values were selected in the tumor area and the ADC values of the contralateral healthy brain tissues were measured. The rADC values were calculated from the ratio of the tumor area and the healthy brain tissue. The rADC value of the tumor was represented by the average rADC value of the five ROIs.

The DCE-MRI data were processed following import into OmniKinetics. The K_trans maps were calculated following a ‘reference region’ model proposed by Cárdenas-Rodríguez et al (25), a total of five ROIs with higher K_trans values were selected in the tumor area and the K_trans value of the tumor was represented by the average K_trans values of these five ROIs. The MRI data were analyzed by two experienced radiologist.

Hematoxylin and eosin (H&E) and immunohistochemical staining were performed on surgical specimens and on the corresponding xenograft samples following paraffin embedding, as described previously (26). The antibodies used were raised in rabbits against human CD34 (1:30; cat. no. ab81289; Abcam). The serial sections were prepared at 2 µm thickness and were used for immunohistochemical staining following dewaxing in xylene. Antigen retrieval was performed in a boiling EDTA solution (pH 9.0) for 2.5 min. The sections were washed with PBS following cooling. H₂O₂ (10%) and goat serum (Beyotime Institute of Biotechnology) were used to block endogenous peroxidase activity and nonspecific antigens, respectively. Each slice was incubated overnight with the corresponding primary antibody at 4°C. The specimens were washed with PBS and incubated with horseradish peroxidase-conjugated goat anti-rabbit secondary antibodies at 37°C for 30 min. 3, 3′-diaminobenzidine was used to visualize the antigen signal.

A total of five ROIs were selected using the hot-spot method. The images were visualized by light microscopy with ×200 magnification. The number, diameters and areas of the CD34-positive lumens were measured, and the average values were used as the tumor microvascular density, diameter and area, respectively. All pathological data for the tissues were measured by two highly experienced staff members who were blinded to the experimental groups.

The original tumors (patient tumors 1, 2 and 3) and the corresponding xenografts (xenografts 1, 2 and 3) were collected for transcriptome sequencing (Wuhan Seqhealth). The differences in mRNA expression levels between the original tumors and the xenografts were compared. Total RNA was extracted from the glioma tissues using TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.). Following removal of the ribosomal RNA and double-stranded RNA, the mRNA was reverse-transcribed into double-stranded cDNA. Polymerase chain reaction (PCR) was performed to amplify and establish the RNA library, which was assessed for nucleotide purity. Agarose gel electrophoresis was used to exclude severely degraded nucleic acid samples, and subsequently, the concentration of nucleic acids was measured using Qubit 2.0 (Thermo Fisher Scientific, Inc.). Samples with a concentration >200 ng/µl and a total quantity >0.8 mg were considered as suitable samples. The RNA library was sequenced on an Illumina sequencer (Illumina, Inc.). Gene expression levels were determined by several reads per kilobase per million reads and further processed with sample biological repeat correlation testing. Differentially expressed genes were screened between xenografts and their corresponding original tumors (fold change >2; P<0.05). Gene Ontology (GO) (27,28) analysis was performed to analyze the functions of the differentially expressed genes.

SPSS version 19.0 (IBM, Corp.) was used for statistical analysis. A paired t-test was used to compare differences of rADC, microvessel density, microvessel area and diameter values between original tumors and the corresponding xenografts. A Wilcoxon signed-rank test was used to compare differences of K_trans values between the original tumors and corresponding xenografts. MRI data and pathological result reproducibility were assessed using intraclass correlation coefficient (ICC). P<0.05 was considered to indicate a statistically significant difference.

The 7 glioma patients enrolled in the present study included 6 glioblastoma cases (patients 1, 2, 3, 4, 6 and 7) and 1 anaplastic astrocytoma patient (patient-5) (Table I). Glioma stem cell spheres were extracted from the surgical specimens of the patients (Fig. 1), and the orthotopic glioma xenograft model was successfully established in all NOD-SCID nude mice. The growth pattern of the xenografts were divided into two categories. The first category included 6 cases of tumor with diffuse growth (xenografts 1, 2, 3, 4, 5 and 6), with 5 mice per group. The second category was for the one tumor case with nodular growth (xenograft 7; n=5 mice).

The differences in the MRI features between xenografts and original tumors are described in Table II. The original tumor samples did not have clear tumor boundaries or evidence of edema around the tumor mass. However, certain tumor specimens (patients 3 and 7) possessed a boundary between the tumor and the normal brain tissue in certain regions. The intra-tumoral signals were heterogeneous and multiple necrosis and cystic degeneration were evident. DWI indicated significant high signal intensity in the tumor area region (b=1,000 sec/mm²). Following gadolinium contrast enhancement, marked annular contrast enhancement was observed. The tumor area indicated apparent annular high signal intensity as determined by the K_trans map (Fig. 2).

The most significant difference noted in the MRI features between the 6 cases of diffusely grown xenografts and the original tumors was the mild enhancement occurring in the local area of the xenografts. The k_trans value of the xenografts was significantly lower compared with the original tumors. In addition, the internal signal of the tumor was homogeneous in the absence of edema (Fig. 2). This finding was different when compared with the corresponding signal noted in the original tumors (Fig. 2). A clear demarcation between the xenografts and the normal brain tissue was noted in only one case of tumor nodular growth between the two groups (xenografts and human tumor). This was the major difference in the imaging characteristics between these two groups. In addition, the enhancement degree of the xenografts and the K_trans value of the tumor area were significantly lower compared with the original tumors (Fig. 3). The K_trans values of all the xenografts were lower compared with the corresponding values of the original tumors, and the rADC values of all the xenograft samples were higher compared with the corresponding values of the original tumors (P=0.016 and P=0.001, respectively; Figs. 2 and 3; Table III). The measurement results were determined using the ICC test, a measure of consistency of results, and the ICC value was 0.962, suggesting the consistency was good.

Among the 6 cases of orthotopic glioma xenograft models with diffuse growth, xenograft-5 was derived from a case of WHO grade III anaplastic astrocytoma, and the rest were derived from glioblastoma. Xenograft 5 showed diffuse growth with a homogeneous internal signal, mild enhancement present in the local area and almost no high signal on the K_trans map, which were the same as the other xenografts with diffuse growth (Fig. 2). The rADC values of xenograft-5 (0.948±0.033) were not significantly different from those of other xenografts (0.937±0.055) with diffuse growth (P=0.301).

H&E staining of the xenografts and human tumor tissues indicated that the boundary of the xenograft tissues was clear, whereas that of the patient tumor was unclear. An example is shown for patient-7 and the corresponding xenograft, where the patient exhibited nodular growth (Fig. 4). However, the tumor boundary was not clear in xenografts with diffuse growth and the corresponding original tumors (Fig. S1). CD34 staining showed that the microvessel area and diameter of the 6 ×enograft cases that exhibited diffuse growth were significantly lower compared with the patient tumors (P=0.009 and P=0.007, respectively). There were no significant differences in the microvessel density between the xenografts and the patient tumors (Fig. 5). These findings were verified using an ICC test, with an ICC value of 0.955.

The differences in gene expression of the samples from patients 1, 2 and 3, and the corresponding xenografts are presented in Fig. 6. The black dots between the blue lines represent genes with expression differences <2-fold and the red dots represent genes with differences in expression >2-fold. The red dots above the blue lines represent genes with higher expression in the original tumors compared with the xenografts, whereas the red dots below the blue lines represent genes with lower expression in the original tumors compared with the xenografts. Significant differences were noted in the gene expression levels between primary tumors and xenografts.

GO analysis of the differentially expressed genes revealed that tumor cell characteristics and extracellular matrix-associated genes (cell activation, cell adhesion, cell migration, cell motility and extracellular matrix associated genes), angiogenesis-associated genes (angiogenesis and vasculature development) and immune-associated genes (immune response, immune system process and immune effector process) were highly expressed in the original tumors. The expression levels of the genes that were involved in cell cycle and nuclear division were increased in the xenografts (Figs. 7–9).