BACKGROUND: Glioblastoma multiforme (GBM), a high-grade glioma, is characterized by being diffuse, invasive, and highly angiogenic and has a very poor prognosis. Identification of new biomarkers could help in the further diagnosis of GBM.
OBJECTIVE: To identify ELTD1 (epidermal growth factor, latrophilin, and 7 transmembrane domain-containing protein 1 on chromosome 1) as a putative glioma-associated marker via a bioinformatic method.
METHODS: We used advanced data mining and a novel bioinformatics method to predict ELTD1 as a potential novel biomarker that is associated with gliomas. Validation was done with immunohistochemistry, which was used to detect levels of ELTD1 in human high-grade gliomas and rat F98 glioma tumors. In vivo levels of ELTD1 in rat F98 gliomas were assessed using molecular magnetic resonance imaging.
RESULTS: ELTD1 was found to be significantly higher (P = .03) in high-grade gliomas (50 patients) compared with low-grade gliomas (21 patients) and compared well with traditional immunohistochemistry markers including vascular endothelial growth factor, glucose transporter 1, carbonic anhydrase IX, and hypoxia-inducible factor 1α. ELTD1 gene expression indicates an association with grade, survival across grade, and an increase in the mesenchymal subtype. Significantly high (P < .001) in vivo levels of ELTD1 were additionally found in F98 tumors compared with normal brain tissue.
CONCLUSION: Results of this study strongly suggests that associative analysis was able to accurately identify ELTD1 as a putative glioma-associated biomarker. The detection of ELTD1 was also validated in both rodent and human gliomas and may serve as an additional biomarker for gliomas in preclinical and clinical diagnosis of gliomas.
ABBREVIATIONS: AA, anaplastic astrocytoma
AO, anaplastic oligodendroglioma
CAIX, carbonic anhydrase IX
CLIO, cross-linked iron oxide
EGF, epidermal growth factor
ELTDI, endothelial growth factor, latrophilin, and 7 transmembrane-containing protein 1 on chromosome 1
GAMMA, global microarray meta-analysis
GBM, glioblastoma multiforme
Gd-DTPA, gadolinium-diethylenetriaminepentaacetic acid
iNOS, inducible nitric oxide synthase
LGA, low-grade astrocytoma
TE, echo time
TCGA, The Cancer Genome Atlas
VEGF, vascular endothelial growth factor
VEGFR2, vascular endothelial growth factor receptor 2
*Advanced Magnetic Resonance Center
§Arthritis and Clinical Immunology, Oklahoma Medical Research Foundation, Oklahoma City, Oklahoma
¶Huntsman Cancer Institute, University of Utah Health Sciences Center, Salt Lake City, Utah
‖The Methodist Hospital (BV), The Methodist Neurological Institute, Houston, Texas
Correspondence: Rheal A. Towner, PhD, Advanced Magnetic Resonance Center, Oklahoma Medical Research Foundation, 825 NE 13th Street, Oklahoma City, OK 73104. E-mail: Rheal-Towner@omrf.org
Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Web site (www.neurosurgery-online.com).
Received October 27, 2011
Accepted September 18, 2012
Gliomas represent 40% of all primary central nervous system tumors diagnosed. Among them, glioblastoma multiformes (GBMs) are the most malignant, with a very poor survival time of approximately 15 months for most patients with this tumor.1 High-grade gliomas are the most common primary brain tumors in adults, and their malignant nature ranks them highly regarding cause of cancer death.1 Grading and identification criteria that can be used to provide information regarding tumor behavior include cell proliferation (cellularity and mitotic activity), nuclear atypia, neovascularization, and the presence of necrosis and/or apoptotic regions.2,3 Differences in molecular composition between tissue types or biomarkers can be used diagnostically to classify tumors and assess prognosis. Molecular markers have increasingly been used to assess and manage adult malignant gliomas.2,4-8 The most useful are markers that can predict response to certain therapies and guide clinical decisions. The most recent biomarkers are from genome-wide surveys associating somatic mutations with the risk of glioma development. Molecular biomarkers most commonly used to evaluate adult malignant gliomas from biopsy samples include 1p/19q co-deletion, methylation of the O6-methylguanine–DNA methyltransferase gene promoter, alterations in the epidermal growth factor (EGF) receptor pathway and isocitrate dehydrogenase 1 and isocitrate dehydrogenase 2 gene mutations.2,4-9 Dozens of proteomics-based approaches have sought to find proteins that are unique to gliomas,10 but have been severely limited by issues of sample size, ability to detect low-abundance proteins, and reproducibility. Many of these studies have generated hundreds and even thousands of putative candidates, yet have not been able to follow them up with subsequent validation and characterization.
Via a bioinformatics method developed by our group,11-14 we conducted a global meta-analysis of approximately 18 000 microarray experiments from the National Center for Biotechnology Information database to identify gene sets consistently coexpressed across heterogeneous conditions. After identifying these gene sets, an automated, large-scale analysis of the peer-reviewed literature was conducted12,13 to identify genes that are consistently transcribed with established glioma-related genes, but which have themselves never been associated with gliomas in the literature. This process was used to identify ELTD1 (EGF, latrophilin, and 7 transmembrane domain-containing protein 1 on chromosome 1) as a novel gene that may be an important biomarker for the confirmation and detection of gliomas.
ELTD1 is not well characterized. Based on its sequence, ELTD1 is a member of the secretin family of G protein–coupled peptide hormone receptors and belongs to the EGF-7 transmembrane subfamiliy.15 Structurally, it contains a large extracellular domain with EGF-like repeats, a 7-transmembrane domain, and a short cytoplasmic tail.15 ELTD1 was first identified to be developmentally regulated in rat fetal and postnatal cardiomyocytes.15 ELTD1 has also been identified with its ligand dermatan sulfate in rheumatoid synovial tissue in rheumatoid arthritis patients.16 In more obscure roles, variations in ELTD1 have been thought to be a risk factor for cannabis use disorders,17,18 tick burden in cattle,19 and subcutaneous fat thickness.20 Of more importance to cancer, ELTD1 has been considered an endothelial marker in microvasculature.21 Our goal in this study was to determine whether ELTD1 could be used as a marker for glioma-related processes, and use immunohistochemistry (IHC) and molecular magnetic resonance imaging (MRI) to validate its presence in human and rodent gliomas.
MRI is becoming one of the most commonly used techniques to provide information on brain tumor growth, vasculature, biochemical metabolism, and molecular changes in preclinical models, as MRI is the optimal imaging tool used in the diagnostic process for human gliomas.22 Molecular alterations can be assessed with the use of targeting magnetic resonance (MR) contrast agents, which can specifically indicate levels of cancer biomarkers that may be elevated in malignant tumors.22 The development of targeted imaging ligands attached to MRI contrast agents allows the in vivo evaluation of tumor biology, such as tumor cell apoptosis, angiogenic blood vessels, and the expression of specific tumor antigens or signaling pathways.23 Molecular imaging involves the coupling of a targeting moiety (antibody [Ab] or peptide targeted to a protein of interest) to a reporter molecule (eg, MRI contrast agent). Commonly used MRI contrast agents are gadolinium (Gd)-based compounds and iron oxide–based nanoparticles.
In this study, we identified ELTD1 as a putative glioma-associated marker via a bioinformatic method and experimentally validated its presence in both rodent and human gliomas via IHC and molecular MRI analyses in an F98 rodent glioma model. For IHC, ELTD1 was compared with traditional IHC markers for human gliomas including vascular endothelial growth factor (VEGF), glucose transporter 1 (GLUT-1), carbonic anhydrase IX (CAIX), and hypoxia inducible factor-1α (HIF-1α). ELTD1 expression in human gliomas was also evaluated from gene expression databases (Rembrandt, Erasmus, and The Cancer Genome Atlas [TCGA]) to establish whether this biomarker is differentially expressed in varying glioma grades.
PATIENTS AND METHODS
The human tissue sample portion of the study was conducted in compliance with the University of Utah Health Sciences Center Institutional Review Board. For IHC analysis, GBMs, anaplastic astrocytomas (AAs), and anaplastic oligodendrogliomas (AOs) (high-grade gliomas: 50 patients, 21 female and 29 male; 40 GBMs: 6 AAs, 4 AOs) were compared with tumors classified as low-grade gliomas (21 patients: 10 female, and 11 male; 11 oligodendrogliomas: 10 low-grade astrocytomas (LGAs), including benign oligodendrogliomas. Abs to ELTD1 were available commercially (human specific Abs all used for IHC: all are rabbit polyclonal anti-human Abs; CLS-C40639 [LifeSpan BioSciences, Inc, Seattle, Washington]; NBP1-84775 [Novus Biologicals, Novus USA, Littleton, Colorado]; and PA1-32729 [Thermo Fisher Scientific Inc., Rockford, Illinois]; all human Abs were assessed and found to provide similar results; dilution was 1:500 and rodent Ab specific for both mouse and rat, recommended use for IHC and Western blotting: ETL (N-20): sc-46951; goat polyclonal anti-mouse, peptide mapping near the N-terminus of ETL [Santa Cruz Biotechnology, Inc, Santa Cruz, California; dilution was 1:100]). A lack of cross-reactivity between human and rodent Abs is illustrated in supplementary Ab data (see Appendix 1, Supplemental Digital Content 1, http://links.lww.com/NEU/A501. Lack of cross-reactivity between human and rat ELTD antibodies. IHC slides for human GBMs and rat F98 gliomas, in which human GBM and rat F98 tumor tissues were stained with either anti-human ELTD or anti-rat ELTD antibodies. Note only positive staining of human GBMs when stained with anti-human ELTD Ab or when rat F98 tumors are stained with anti-rat ELTD Ab. First slide is ×10 magnification, second slide is ×20 magnification, and third slide is ×40 magnification.) A toluidine blue (0.1%) counterstain was used (15 seconds). For human tissues, IHC was performed using the Vectastain ABC Kit (Vector Laboratories, Burlingame, California). Negative controls were performed by replacing the primary Ab with nonimmune serum. Slides were examined using an Olympus BX41 microscope (Center Valley, PA, USA). Under ×200 (10 ocular × 20 objective) magnification, slides were scored by 2 investigators blinded to the specimen tumor grade and patient information. A score of 0 to 4 (0, 0-25%; 1, 25%-50%; 2, 50%-75%; 3, 75%-100%; 4, 100%) was assigned based on the number of cells stained in a given field. In previous studies we demonstrated that this method was very reproducible as demonstrated by good interrater reliability (P = .99, 95% confidence interval: 0.99-1.00) and intrarater reliability (P = .96; 95% confidence interval: 0.92-0.99).24 Each investigator reviewed the slide at low power and at random high-power fields independently of the other investigator when determining the IHC score. Scores of 2 to 4 were considered positive expression, whereas scores of 0 and 1 were considered negative expression.
For the rat F98 glioma tissues, IHC fluorescence staining was done for the endothelial cell marker, CD31 (fluorescein isothiocyanate–labeled donkey anti-mouse Ab). For ELTD1, a secondary Ab (Cy3-labeled donkey anti-goat; Jackson ImmunoResearch, Suffolk, England) was used to detect the anti-ELTD1 Ab in the anti-ELTD1 probe. Rat brains were extracted after the 3-hour molecular MRI protocol; the tumor side and contralateral side of the brain were cut and fixed in Z-fixative (zinc formalin: formaldehyde 3.7%, zinc sulfate). The tissue was then washed with phosphate-buffered saline and incubated with 15% sucrose before embedding in an OCT (optimal cutting temperature) compound and freezing in liquid nitrogen. The cryosections were then stained with a secondary Ab (Cy3 [blue]-labeled donkey anti-goat [Jackson ImmunoResearch]) to target the anti-ELTD1 (ETL) Ab, and a fluorescein isothiocyanate (red)–labeled donkey anti-mouse Ab to target the anti-CD31 (CD31; mouse monoclonal anti-rat; Dako Denmark, Glostrup, Denmark) Ab within the brain tissue. The nucleus was stained with 4',6-diamidino-2-phenylindole (blue). Stained tissue slices were examined with a Nikon C1 confocal laser scanning microscope (Nikon Instruments, Melville, New York). Colocalization analysis was done using an Imaris Coloc module (version 6.4), and data were presented as the percentage of colocalization and the Pearson colocalization coefficient.25
Rat Glioma Cell Implantation
The rat portion of the study was conducted in compliance with the Oklahoma Medical Research Foundation Institutional Animal Care and Use Committee. The rat glioma cell implantation model was conducted as previously described by our group.26 Briefly, 3-month-old male Fischer 344 rats (Harlan Inc, Indianapolis, Indiana) were anesthetized and immobilized on a stereotactic unit. A hole was drilled through the skull at 2 mm anterior and 2 mm lateral to the bregma on the right-hand side of the skull. Then 10 000 F98 cells suspended in a 10-µL volume were injected at a depth of 3 mm from the dura into the cerebral cortex at a rate of 2 µL/min followed by a waiting time of 2 minutes. The F98 cells were originally obtained from the ATCC (Manassas, Virginia), passaged once, and then infected with a modified pMMP retrovirus fused with the coding sequences for luciferase and hygromycin (pMMP-LucHygro; obtained from Dr Stephen Lessnick, University of Utah, Huntsman Cancer Institute). They were selected for hygromycin resistance and colonies were screened for luciferase expression. Stocks were frozen 1 passage after infection and colony screening. Cells were also found to have no mycoplasma infection (Hoechst dye 33258 nuclear (DNA) staining detection kit; Roche Diagnostics, Indianapolis, Indiana).
Synthesis of ELTD1 Nanoprobes
The dextran-coated NH2 base iron oxide nanoparticle construct underwent conjugation with an ELTD1-specific Ab using a protocol previously reported by our group26 via the formation of a stable thioether linkage between the activated polyethylene glycol nano particles and activated Ab. The Ab is activated with N-succinimidyl-S-acetylthioacetate to introduce a sulfhydryl group.26 The amine groups were activated with N-succinimidyl 3-(2-pyridyldithio)-propionate.
In vivo MRI experiments on rats with F98 gliomas were carried out with the rats under general anesthesia (1%-2% isoflurane, 0.8-1.0 L/min O2). The MRI equipment used was a Bruker Biospec 7.0-T/30-cm horizontal-bore imaging spectrometer (Bruker BioSpin MRI GmbH, Ettlingen, Germany). Animals were imaged at 7 to 10 days after the cells were injected and then every 2 to 3 days until the desired volume of the tumor was obtained (75-150 mm3). Anesthetized (2% isoflurane) restrained rats were placed in a MR probe and their brains localized by MRI. Images were obtained using a Bruker S116 gradient coil (2.0 mT/m/A), a 72-mm quadrature multirung radiofrequency coil for radiofrequency transmission, and a rat head coil for radiofrequency signal receiving. MRI was performed for the purpose of determining the incidence, number, growth rate, and volume of each tumor for the F98 gliomas. Multiple 1H-MRI slices were taken in the transverse plane using a spin-echo multislice (repetition time, 0.8 seconds; echo time (TE); 23 ms; 128 × 128 matrix; 4 steps per acquisition; 4 × 5-cm2 field of view; 1-mm slice thickness).26-28
For determination of T2* values of the IO nanoprobes in gliomas, a multiple gradient echo method was used with the following parameters: TE (first echo) = 4 ms, echo spacing of 4-ms repetition time = 1500 ms, 10 echoes (TE = 4, 11, 18, 25, 32, 39, 46, 53, 60, 67 ms), 2 coronal (horizontal) slices, number of averages = 2, 256 × 256 matrix, 1-mm slice thickness with a spatial resolution of 0.137 mm/pixel, and an estimated total scan time of 10 minutes. T2* maps were generated from the multiecho data sets. Rat brains were imaged at 0 (pre-nanoprobe [cross-linked iron oxide (CLIO) anti-ELTD1 nanoprobe] or precontrast agent [control CLIO-IgG] administration), 10-minute intervals post-nanoprobe or IgG contrast agent injections for up to 3 hours. Rats were injected with a single intravenous dose via a tail vein catheter with either the anti-ELTD1 nanoprobe (anti-ELTD1 antibodies [goat anti-mouse] tagged with an IO-based contrast agent [CLIO-based] (200 μL/200 g rat; 1 mg Ab/kg; 0.05 mmol Fe3+/kg]), or the normal rat IgG control contrast agent (same dose as anti-ELTD1 nanoprobe).
Multiple regions of interest (10 regions of interest within tumor and nontumor tissues) were selected (in various representative tumor and corresponding contralateral nontumor regions by 2 operators with interobserver agreement) from T2-weighted images and T2* maps to calculate relative changes in MR signal intensities and T2 values in F98 glioma-bearing rats administered either the CLIO-anti-ELTD1 nanoprobe (n = 3) or the CLIO-IgG contrast agent (n = 2). MR angiographic images were obtained as previously described.28
A Mathematica-based T2* program was used to crop the data/matrix to an area of interest, and it then fit each pixel of a designated image slice to the T2* relation: I = a + I0 * exp(TE/T2*), where a is a constant determined by the FindFit function of Mathematica, I is the intensity at time t, I0 is the signal intensity at time 0, TE is the set of defined echo times, and T2* is the parameter of interest. The FindFit function of Mathematica optimizes the values not explicitly defined to produce the best possible fit. This method is applied to both pre- and postcontrast images. The percentage of difference is then taken from the fitted data/matrix using the following equation: % difference = [(after-before)/before] × 100. ArrayPlot graphical images are produced for the original cropped image, the fitted pre- and postcontrast image, and the percentage of difference of the fitted images. A contour plot is made using the percentage of difference data to highlight the regions where the greatest intensity change occurred. The contour plot is then overlaid on the original cropped imaged to provide a more visual display of where changes in signal intensity occurred.
Prussian Blue Staining of the Nanoprobes
Detection of the iron oxide–based nanoprobes in tissue cryosections was done using Prussian blue staining, which involves the treatment of sections with acid solutions of ferrocyanides. The ferric ion (+3) present in the iron oxide–based nanoprobes from tissue sections combines with the ferrocyanide and results in the formation of ferric ferrocyanide, visible as a blue pigment in bright-field imaging. Stained tissue slices were viewed and photographed with a Nikon Eclipse 800 microscope (Nikon Instruments).
Gene Expression Analysis
For the glioblastoma expression microarray analysis, raw Affymetric .cel files were downloaded for TCGA (National Cancer Institute) (Cancer Genome Atlas; www.cancergenome.nih.gov/; 529 GBM samples), Rembrandt (National Cancer Institute Repository for Molecular Brain Neoplasia Data; www.rembrandt.nci.nih.gov/; 229 total astrocytomas, of which 151 are GBMs), and Erasmus (National Center for Biotechnology Information Gene Expression Omnibus; GEO Series GSE16011; total of 187 astrocytomas, of which 159 are GBMs), as well as the corresponding clinical annotations for each. The .cel files were then processed using R and Bioconductor, using a custom chip definition files with background correction, log transformation, and quantile normalization performed using the robust multiarray average algorithm implemented in R.
For mesenchymal and proneural gene signature definition, we used a composite of signatures from Phillips et al29 and Verhaak et al.30 For a given tumor, the metagene mesenchymal and proneural signature scores were both calculated. Within a data set, the mesenchymal and proneural metascores were z-score–corrected to allow their comparison. Tumors were then assigned to one of the signatures based on the higher expressing metagene.
Frozen tissue was weighed, 200 mg was thawed in 1 mL red blood cell lysis buffer (Sigma-Aldrich, R7757, St. Louis, MO, USA) with protease inhibitors (Sigma, p8340), Na3V04 (1 mM), dithiothreitol (1 mM) and phenylmethanesulfonyl fluoride; 1 mM, then diced using surgical microscissors. Tissue was centrifuged at 1500 rpm, supernatant removed, and 500 μL lysis buffer containing proteases and phosphatase inhibitors added. Tissues were homogenized with a rotor-stator at 4°C for 1 minute, incubated on ice for 30 minutes with shaking, and then centrifuged (42 000g, 20 minutes, 4°C). The clear supernatant was transferred to a clean 1.5-mL tube. After determining the total protein concentrations, 40 μg of total protein was separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis using a Novex 4% to 12% gel (Invitrogen, Carlsbad, California), and transferred to polyvinylidene difluoride membranes. Western analysis was done using antibodies against ELTD1 (ETL (N-20): sc46951; Santa Cruz Biotechnology Inc). Secondary antibodies were labeled with horseradish peroxidase. The Supersignal West Pico chemiluminescent substrate kit (#34077; Thermo Scientific) was used for detection.
Statistical differences in MR signal intensities and T2* relaxations, which indicated specific binding of the nanoprobes in glioma tissue, were analyzed in the treatment and control groups and in tumor and nontumor regions with an unpaired, 2-tailed Student t test using commercially available software (InStat; GraphPad Software, San Diego, California). A P value of <.05 was considered to indicate a statistically significant difference. For IHC scoring and ELTD1 expression, statistical differences were compared between groups using the Welch 2-sample t test (unpaired, 2 sided), with P values <.05 considered significant.
A global microarray meta-analysis (GAMMA) of all genes differentially expressed across 3651 human 2-color microarray experiments was conducted as previously described11 to identify gene–gene coexpression patterns that were consistent and specific across heterogeneous microarray experiments. The significance and reproducibility of the GAMMA predictions from the 2-color array data have since been corroborated by normalization31 and meta-analysis of 16 000 additional 1-color human microarrays.32 This “guilt-by-association” approach identifies gene sets that are likely to be associated in biologically relevant ways such as phenotype, disease, and genetic network. GAMMA has been used successfully to identify the mitotic role of a formerly uncharacterized gene called C13ORF3 (now Ska3)33 and a role in coagulation for C6ORF105 (now ADTRP)34 and to identify OLFM4 as a novel neutrophil subset marker associated with granule secretion.35
With the GAMMA approach, genes are not analyzed directly, but the top 20 genes most consistently coexpressed with them are analyzed for what they have in common in the peer-reviewed literature using a large-scale computational analysis.12,14 This way, even if a protein has no known function, its function can be inferred. Then, using the Human Proteome Reference Database34 and other experimental sources on protein cellular localizations, we screened this list of predicted glioma-associated proteins for those that were extracellular or membrane bound because these proteins were thought to be ideal targets for molecular imaging probes and targeting therapies because they are more likely to be accessible to injected antibodies. Using our procedure, we identified membrane-bound proteins that have not yet been associated with gliomas, but whose expression consistently correlates with genes reported to be associated with gliomas. This circumvents a problem inherent in the lists of expressed genes derived by microarrays, which identify only those genes that are being actively transcribed at the time of the experiment without detecting proteins that are present but not actively transcribed. That is, GAMMA associates genes frequently cotranscribed regardless of the condition, and then if a statistically significant set of genes has been reported as glioma-associated in the literature, these associations need not be transcriptional to be identified by GAMMA (eg, they could be from proteomics or genome-wide association studies). The enormous sample size of both microarray data and analyzed abstracts enables us to screen out genes that do not pass a threshold of statistical significance. This associative method works for glioma-derived, literature-based associations as well as searches on associated processes (such as angiogenesis, apoptosis, and cell migration), helping corroborate any putative roles in tumorigenesis that we uncover. For each association, we calculate mutual information (a measure of variable dependency) between literature terms to prioritize the strength of association between each protein and a role in gliomas.14 Finally, we obtained increased confidence in the predictions because GAMMA also successfully predicted many established glioma-related genes (eg, EGF receptor, matrix metalloproteinase 2, glial fibrillary acidic protein, fibroblast growth factor 2). These identifications serve as positive controls for predictive capacity. We identified 195 putative candidate markers, all genes predicted or known to be membrane bound and not appearing in any MEDLINE article that mentioned gliomas (or synonymous terms). Of these 195, only 75 had commercial antibodies. ELTD1 was chosen from among this list of 75 because it had the highest score. With this analysis set to stringent thresholds, we empirically observed that the ELTD1 gene is found to be consistently transcribed with known glioma-associated genes (see Appendix 2, Supplemental Digital Content 2, http://links.lww.com/NEU/A502, which list GAMMA predicted associations for the ELTD1 gene; predicted associations that were tested in this study are shown in red).
GAMMA scores are based on a combination of (1) how many genes out of the 20 top coexpressed analyzed genes showed associations with gliomas based on published reports and (2) their statistical significance based on random network simulations to estimate the probability that a set of equally frequent terms would be associated with gliomas. Only proteins with P < .01 significance were selected as potential candidates. A flow diagram of the GAMMA approach is illustrated in supplementary data (see Appendix 3, Supplemental Digital Content 3, http://links.lww.com/NEU/A503, which includes steps for the GAMMA approach. Slide 1 describes the steps in the GAMMA approach. Slide 2 provides descriptive diagrams for gene comparisons [Figure S1], gene function associations [Figure S2], and reported commonalities in peer-reviewed literature [Figure S3].).
Immunohistochemistry and Western Blot
From human IHC assessment, ELTD1 was expressed in all gliomas where it was found to have an average IHC score of 2.7 (± 1.16) or more than 67% expressed in high-grade gliomas and an average IHC score of 2.05 (± 1.07) in low-grade gliomas or more than 33% expressed in low-grade gliomas (Figure 1), the difference being statistically significant (P = .03). The percentages of survival for the GBM, AA, and AO patients were 0, 0, and 50%, respectively. The percentages of survival for benign oligodendroglioma and LGA patients were 27.3% and 70.0%, respectively.
ELTD1 compared well with known glioma biomarkers, including VEGF, HIF-1α, GLUT-1, and CAIX. The IHC average score for ELTD1 in high-grade gliomas was slightly lower than the average IHC scores for VEGF, HIF-1α, GLUT-1, and CAIX (Figure 1A). In low-grade gliomas, ELTD1 had an IHC score that was less than those for VEGF, HIF-1α, and CAIX (Figure 1A). Although the recently discovered glioma marker Brevican had 68% expression levels in high-grade glioma patients (Figure 1B), it was found to be the lowest for all biomarkers tested (Figure 1A). Brevican and the biomarker CAIX were not found to be significantly higher (P = .19 for both) when comparing high-grade with low-grade gliomas, whereas all other biomarkers (VEGF, HIF-1α, and GLUT-1) had significance levels of P = .01, P = .02, and P = 0.001, respectively, including ELTD1 (P = .03) and were significantly higher in high-grade compared with low-grade gliomas (Figure 1A).
Figures 2A and B depict representative IHC staining for ELTD1 in human GBMs and control brain tissues, respectively, indicating that GBMs (Figure 2A) had substantially higher levels (including both vascular and glioma cells). Figures 2 C and D show representative IHC staining for ELTD1 in a rat F98 glioma model compared with contralateral brain tissue, respectively, with higher levels detected in glioma tissue (including both vascular and glioma cells) vs contralateral brain tissue (Figure 2D). Figure 2E is a representative Western blot of ELTD1 levels obtained from rat F98 glioma tissues (2 right lanes) compared with normal rat brains (4 left lanes), illustrating high levels of ELTD1 in tumor tissues.
With the use of molecular MRI and iron oxide–based nanoprobes, in vivo ELTD1 levels were detected in rat F98 gliomas (Figure 3). The ELTD1 probe was a dextran-coated iron oxide construct with an anti-ELTD1 Ab coupled to the dextran (Figure 3A). A representative T2* difference image overlaid on top of a T2-weighted morphological MR image is shown in Figure 3B. A corresponding MR angiography image and its difference image (2 hours after administration of the ELTD1 probe minus before injection of the probe) are shown in Figures 3C and 3D, respectively. Note the high signal intensity within the tumor region (see Figure 3B for morphological T2-weighted image). The percentage of change in T2* differences is shown dynamically within representative animals in Figure 3E, where only the glioma region from an F98 glioma–bearing rat administered the ELTD1 probe had a high percentage of T2* differences (12%-14%) compared with the contralateral region (6%-8%) in the same animal or the tumor or contralateral regions of a F98 glioma–bearing rat that was administered the nonspecific IgG probe (<6%). Corresponding quantitative T2* differences in the tumor region of F98 glioma–bearing tumors is shown in Figure 3F (where a T2* difference of 87.44 ± 39.95 in a F98 tumor administered the ELTD1 probe (3 F98 glioma–bearing rats, 10 sample regions per rat, ie, 30 sampling regions in total) was significantly higher (∼3-fold, P < .001) than that measured in the tumor region of the nonspecific IgG control, which was 26.87 ± 35.48 (2 F98 glioma–bearing rats, 10 sample regions per rat, ie, 20 sampling regions in total).
Confirmation of the presence of the iron oxide–based anti-ELTD1 nanoprobes in an F98 glioma–bearing animal (Figure 4A; T2-weighted MR image) is shown in Figures 4Bi and 4Bii. Low levels of the iron oxide particles are also detected in the contralateral brain tissue of an animal administered the ELTD1 probe (Figure 4Ci and ii), as well as less probe within the glioma (Figures 4E and Figure 3F) or contralateral (Figure 4F) brain tissues of an F98 glioma–bearing animal (Figure 4D; T2-weighted MR image) administered the IgG contrast agent.
To establish whether ELTD1 was predominantly an endothelium-associated marker, excised F98 glioma and contralateral brain tissues exposed to the anti-ELTD1 probe were fluorescently labeled with an anti-CD31 Ab (targeted with a fluorescein isothiocyanate–labeled secondary Ab), and a secondary Cy3-labeled Ab against the anti-ELTD1 Ab used in the anti-ELTD1 probe (Figure 5). Colocalization images indicate that ELTD1 colocalized predominantly with endothelial cells (CD31) (yellow = red [CD31] + green [ELTD1]; Figures 5Bi and 5Ci; highlighted rectangular regions). Colocalization analysis indicated that the Pearson colocalization coefficients were 0.8089 (1.0000 would be 100% colocalization) (Figure Bii) and 0.7929 (Figure Ci) for ELTD1 and CD31 in glioma tissue, indicating a high association of ELTD1 with endothelial cells. There is also some evidence that indicates the slight presence of ELTD1 surrounding glioma cells that did not colocalize with CD31 (Figures 5Ci and 5Civ; highlighted circle in Figure 5Ci). Contralateral brain tissue also had low levels of ELTD1 that colocalized with CD31 (Figure 5Ai, highlighted regions; Pearson colocalization coefficient of 0.7149).
Gene Expression Analysis
A number of large public gene expression databases including multiple types and grades of gliomas have been established, including Erasmus, Rembrandt, and, more recently, TCGA (GBMs only). To determine whether mRNA levels of ELTD1 were related to glioma grade, survival, or tumor gene expression subtype, we performed analyses of data from these databases (as described in the Methods section). In the 2 databases that included different grades of gliomas (WHO II-IV), there was a very significant association of increased ELTD1 expression with higher grade (Figure 6). In addition, an analysis of survival using the Rembrandt database demonstrated that increased ELTD1 expression was associated with worse survival across glioma grades (ELTD1 Rembrandt Survival supplementary data; see Appendix 4, Supplemental Digital Content 4, http://links.lww.com/NEU/A504. Increased ELTD1 expression is associated with worse prognosis across grades in gliomas [Rembrandt gene expression database; date of query 6/14/2012]. The probability of survival was worse for up-regulated ELTD1 expression compared with improved probability of survival in down-regulated ELTD1 expression. Total number of tumors = 343 [all glioma group], 173 up-regulated, 5 down-regulated, and 165 intermediate. Statistics [from the Rembrandt Web site] are log-rank P value [for significance of difference in survival between groups of samples]: up-regulated vs intermediate = 3.019854E-4; up-regulated vs down-regulated = 0.056211295; down-regulated vs intermediate = 0.2565988951; up-regulated vs all other samples = 1.396275E-4; down-regulated vs all other samples = 0.1333846694; and intermediate vs all other samples = 8.958061E-4.) Analysis of survival within GBM tumors from these databases did not demonstrate significant survival association within grade IV tumors. However, we did find that when we analyzed ELTD1 expression as a function of tumor gene expression subtype30 within grade IV tumors, there was a potential association of higher ELTD1 expression in the mesenchymal vs proneural subtype, which was significant in the Rembrandt data set and showed a trend in the TCGA data set (Figure 7). Taken together, these data indicate that ELTD1 is potentially a strong biomarker of glioma grade and survival and may be preferentially associated with the mesenchymal subtype of GBM.
We have demonstrated that the differential presence of ELTD1 in gliomas compared with nondiseased regions could potentially serve alone or in combination with other glioma-specific biomarkers because it is detected in both human GBM and rodent models for gliomas. Despite current therapies, GBM is a devastating cancer, and the validation of more biomarkers for GBM could be beneficial in the diagnosis and therapeutic intervention of this disease. ELTD1, as shown in our human IHC data (Figures 1 and 2), fares well in comparison with more traditional IHC markers currently used to diagnose GBM.36 IHC staining in human high-grade gliomas (GBM + AAs + AOs) and low-grade gliomas (LGAs + benign oligodendrogliomas) indicated higher levels of ELTD1 for high-grade compared with low-grade gliomas. The level of ELTD1 was similar to currently investigated glioma markers including VEGF, HIF-1α, and GLUT-1. It is well-known that HIF-1α is an important diagnostic marker and can be targeted for therapeutic intervention.24,37-43 ELTD1 also can be detected in an aggressive rodent model for gliomas. Within a rat F98 glioma model, ELTD1 levels from IHC assessment were found to be higher in glioma tissue; however, the contralateral tissue seemed to still have some staining for ELTD1 (Figures 2C and Figure 5A). Western blot data indicate high levels of ELTD1 compared with normal rat brain (Figure 2E). Lower levels of ELTD1 in normal brain tissue may be due to decreased vasculature (compared with a tumor). Human normal brain tissue had very low ELTD1 levels (Figure 2B) compared with extremely high levels in a GBM patient (Figure 2A).
Preclinical glioma models, induced by orthotopic (into native tumor sites) injection of primary tumor cells or tumor cell lines, represent the most frequently used in vivo cancer model systems for glioma research.23,44 F98 glioma cell lines were obtained from chemical induction as a result of administering ethylnitrosourea to pregnant rats, where the progeny developed brain tumors that were isolated, propagated, and cloned in cell culture.45 F98 gliomas are classified as anaplastic malignant tumors, which have an infiltrative pattern of growth, which is an attribute associated with human GBM.45,46 MRI techniques have been used by our group to demonstrate the aggressive nature of F98 gliomas. With the use of diffusion tensor imaging, we were able to demonstrate that the F98 glioma model is much more infiltrative than the rat C6 glioma model.47 With the use of MR angiography, we have shown that the F98 model predominantly uses preexisting blood vessels in tumor angiogenesis, but has longer and thicker new blood vessels compared with other glioma models.48
In this study in the F98 glioma model, we were also able to demonstrate, with the use of molecular MRI and an anti-ELTD1 probe, that substantial levels of ELTD1 are found in the tumor tissue of F98 glioma–bearing animals. ELTD1 was only found to be in high levels within glioma tissue with an approximately ∼4-fold increase compared with contralateral brain tissue (Figure 3F). A decrease in T2 relaxation would indicate the presence of the anti-ELTD1 probe, which would be indicative of the presence of ELTD1. Colocalization images staining for ELTD1 and CD31 and subsequent analysis indicated that most of the ELTD1 detected by fluorescence confocal imaging was associated with endothelial cells (Figure 5). There is also some indication that ELTD1 may be expressed on some glioma cells at much lower levels. Therefore, any increase in ELTD1 will more than likely be associated with increased angiogenesis or neovascularization in gliomas. Decreased levels of ELTD1 in glioma cells compared with high levels in tumor vasculature from the in vivo data may reflect either predominant uptake by the endothelial cells before they reach the glioma cells or decreased uptake of the anti-ELTD1 probe in tumor tissue. IHC staining (Figure 2) indicates ELTD1 staining in tumor cell nuclei as well as around endothelial cells, indicating that the in vivo targeting may be restricted to the distribution of the probe to endothelial cells and only some tumor cells. Prussian blue staining for the anti-ELTD1 probe (Figure 4) seems to indicate intravascular staining based on the well-individualized pattern and distinct shapes, which could indicate an endothelial association and limited distribution of the probe to these cells. However, previous studies using the same probe construct to assess in vivo c-Met28 or VEGF receptor 2 (VEGFR2)26 levels in rat gliomas do not indicate that these probes only reach vascular cells, but do reach glioma cells, which would suggest that the ELTD1 probe has a preferred association with endothelial cells. To confirm ELTD1 expression with vascular endothelial cells, future experiments may also need to evaluate the levels of ELTD1 in association with the inhibition of neovascularization using antiangiogenic therapies (eg, bevacizumab or sunitinib). Additionally, verification of ELTD1 expression in neoplastic cells should be done, such as assessing EGF receptor gene amplification by fluorescence in situ hybridization colocalization with ELTD1-expressed cells.
It is interesting to note that the Human Protein Atlas shows very little positive staining for ELTD1 in malignant gliomas, but strong Ab staining for other cancers, such as thyroid cancer and malignant melanoma (http:///www.proteinatlas.org/ENSG00000162618/cancer; accessed 07/24/12). This database is ideal for the initial determination of general expression levels of a particular protein that could indicate further study. However, more extensive studies, as we have done with the use of IHC staining for ELTD1 in numerous patient samples and in vivo expression levels of ELTD1 in a preclinical model, strongly suggest that malignant gliomas do have high levels of ELTD1. Also worthy of mention, according to SymAtlas (http://biogps.org/#goto=genereport&id=170757; accessed 07/24/12), ELTD1 mRNA expression seems to be highest in hematopoietic stem cells as well as lung and common myeloid progenitor tissues. Whether hematopoietic stem cells are the source of neovascularization that we observe in gliomas would need to be further investigated.
From the gene expression results, we have also demonstrated that there was a strong association of ELTD1 expression with increasing grade (Figure 6). This results in a strong survival association when data across all grades (ELTD1 Survival supplementary data) are compared. However, there was not a survival association with expression level within GBM, suggesting that it is mainly a biomarker of grade. Alternatively, when we looked at GBM tumor subtype, it looked like there was a possible association with the mesenchymal subtype vs the proneural subtype that was significant in Rembrandt, but not in TCGA (Figure 7). It is reasonable to conclude that ELTD1 expression is a strong biomarker of grade (also supported by the IHC data), associated with survival across grades and may be increased in the mesenchymal subtype. ELTD1 expression and associated survival should, in the future, be evaluated by IHC via a glioma tissue microarray to confirm survival differences observed in Rembrandt.
Others have previously used molecular MRI to also assess neovascularization. For example, the expression of cell adhesion molecules, such as integrins, has been found to be up-regulated during tumor growth and angiogenesis, and αVβ3 expression, which has been correlated with tumor aggressiveness, can be measured by MRI with targeted paramagnetic-labeled cyclic arginine-glycine-aspartic acid peptides.23,49 In another study, within U87MG xenograft tumors in nude mice, arginine-glycine-aspartic acid–labeled ultrasmall superparamagnetic iron oxide probes were found to accumulate only within the neovasculature associated with tumors and not within tumor cells.50 Tumor angiogenesis was also monitored via the expression of CD105 in F98 tumor–bearing rats with the use of Gd-diethylenetriaminepentaacetic acid (Gd-DTPA) liposomes targeted to CD105 (CD105-Gd-SLs) and MRI.51
In our laboratories, MRI probes (either Gd or iron oxide based) have also been developed to monitor in vivo levels of molecular markers known to be overexpressed in malignant brain tumors, such as the angiogenic marker VEGFR2)26,52; the tumor cell migration/invasion marker c-Met, a tyrosine kinase receptor for the scatter factor (also known as the hepatocyte growth factor)27,28; and the inflammatory marker inducible nitric oxide synthase (iNOS).53 With the use of a Gd-DTPA–albumin–anti-VEGFR2–biotin probe, regional differences in VEGFR2 levels were detected by MRI in vivo in a C6 glioma model, and probe specificity for glioma tissue, particularly in the peritumor and perinecrotic regions, was confirmed by tagging the biotin moiety of the probe in excised tissues with streptavidin-Cy3.52 The control nonspecific probe had rat IgG conjugated to the albumin instead of the VEGFR2 Ab. A similar result was obtained when an aminated dextran-coated iron-oxide nanoparticles conjugated with an anti-VEGFR2 Ab was used in a C6 glioma model, where distribution of the probe was mainly in the peritumor and perinecrotic regions of the tumor.26 Confirmation of the presence of the nanoprobes was obtained by using Prussian blue stain for the VEGFR2-targeting iron oxide nanoparticles in excised tumor tissues.26 Both Gd- and iron oxide–based probes were also developed to characterize c-Met levels in C6 gliomas. c-Met is a tumor marker that is overexpressed in many malignant cancers, indicative of the invasive nature of a tumor. The distribution of c-Met was found to be more widely dispersed, but mainly concentrated in peritumor regions.27,28 As detected with a Gd-DTPA–albumin–anti-iNOS–biotin (anti-iNOS) probe, iNOS levels were found to vary in different rat glioma models, where the percentage of MRI signal intensity changes were highest in the C6 tumor compared with the RG2- and ethylnitrosourea-induced tumors.53 Dynamic kinetic monitoring of the anti-iNOS probe indicated sustained uptake over 3 hours within tumor tissue regions and no specific uptake of a control Gd-DTPA–albumin-IgG-biotin contrast agent within tumors.53 Fluorescence imaging of the anti-iNOS probe by targeting the biotin moiety with streptavidin-Cy3 verified higher levels of probe uptake in C6 tumors vs RG2 gliomas, despite the increased perfusion and microvascularity detected in the RG2 tumors.53
For this study, an iron oxide–based nanoparticle construct covalently bound to an anti-ELTD1 Ab was used to detect high levels of ELTD1 in the tumor regions of F98 glioma–bearing rats (Figures 4 and 5). Specificity of the ELTD1 probe seems to be associated with neovascularization.
The results presented strongly suggest that the associative analysis method used in this study was able to accurately identify ELTD1 as a glioma-associated biomarker, possibly due to increased angiogenesis. Both ex vivo and in vivo validation studies indicate that ELTD1 is a biomarker that can be used to confirm or detect the presence and grade of gliomas, particularly high-grade gliomas in humans, and that this biomarker may play an important diagnostic role in addition to currently used markers for gliomas, particularly as a histological marker for identifying vascular proliferation.
Supported by Oklahoma Medical Research Foundation, the National Institutes of Health (grant 5P20RR020143-07 to J.D.W.), and Oklahoma Center for the Advancement of Sciences and Technology (OCAST grant AR092-049 to R.A.T.). The authors have no personal financial or institutional interest in any of the drugs, materials, or devices described in this article.
The authors thank Dr Stephen Lessnick, PhD, (University of Utah, Huntsman Cancer Institute) for the kind gift of the modified pMMP retrovirus fused with the coding sequences for luciferase and hygromycin (pMMP-LucHygro). They also thank Charity Njoku, BSc, (OMRF) for the culturing and maintenance of the F98 cells used in the orthotopic rodent model.
2. Louis DN. Molecular pathology of malignant gliomas. Annu Rev Pathol. 2006;1:97–117.
3. Gudinaviciene I, Pranys D, Juozaityte E. Impact of morphology and biology on the prognosis of patients with gliomas. Medicina (Kaunas). 2004;40(2):112–120.
4. Cancer Genome Atlas Research Network. Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature. 2008;455(7216):1061–1068.
5. Riemenschneider MJ, Jeuken JW, Wesseling P, Reifenberger G. Molecular diagnostics of gliomas: state of the art. Acta Neuropathol. 2010;120(5):567–584.
6. Jansen M, Yip S, Louis DN. Molecular pathology in adult gliomas: diagnostic, prognostic, and predictive markers. Lancet Neurol. 2010;9(7):717–726.
7. Colman H, Zhang L, Sulman EP, et al.. A multigene predictor of outcome in glioblastoma. Neuro Oncol. 2010;12(1):49–57.
8. Farias-Eisner G, Bank AM, Hwang BY, et al.. Glioblastoma biomarkers from bench to bedside: advances and challenges. Br J Neurosurg. 2012;26(2):189–194.
9. Silber JR, Bobola MS, Blank A, Chamberlain MC. O(6)-Methylguanine-DNA methyltransferase in glioma therapy: promise and problems. Biochim Biophys Acta. 2012;1826(1):71–82.
10. Niclou SP, Fack F, Rajcevic U. Glioma proteomics: status and perspectives. J Proteomics. 2010;73(10):1823–1838.
11. Wren JD. A global meta-analysis of microarray expression data to predict unknown gene functions and estimate the literature-data divide. Bioinformatics. 2009;25(13):1694–1701.
12. Wren JD, Garner HR. Shared relationship analysis: ranking set cohesion and commonalities within a literature-derived relationship network. Bioinformatics. 2004;20(2):191–198.
13. Giles CB, Wren JD. Large-scale directional relationship extraction and resolution. BMC Bioinformatics. 2008;9(suppl 9):S11.
14. Wren JD. Extending the mutual information measure to rank inferred literature relationships. BMC Bioinformatics. 2004;5(1):145.
15. Nechiporuk T, Urness LD, Keating MT. ETL, a novel seven-transmembrane receptor that is developmentally regulated in the heart. ETL is a member of the secretin family and belongs to the epidermal growth factor-seven-transmembrane subfamily. J Biol Chem. 2001;276(6):4150–4157.
16. Kop EN, Kwakkenbos MJ, Teske GJ, et al.. Identification of the epidermal growth factor-TM7 receptor EMR2 and its ligand dermatan sulfate in rheumatoid synovial tissue. Arthritis Rheum. 2005;52(2):442–450.
17. Agrawal A, Pergadia ML, Saccone SF, et al.. An autosomal linkage scan for cannabis use disorders in the nicotine addiction genetics project. Arch Gen Psychiatry. 2008;65(6):713–721.
18. Agrawal A, Lynskey MT. Candidate genes for cannabis use disorders: findings, challenges and directions. Addiction. 2009;104(4):518–532.
19. Porto Neto LR, Bunch RJ, Harrison BE, Barendse W. DNA variation in the gene ELTD1 is associated with tick burden in cattle. Anim Genet. 2011;42(1):50–55.
20. Lee KT, Byun MJ, Kang KS, et al.. Neuronal genes for subcutaneous fat thickness in human and pig are identified by local genomic sequencing and combined SNP association study. PLoS One. 2011;6(2):e16356.
21. Wallgard E, Larsson E, He L, et al.. Identification of a core set of 58 gene transcripts with broad and specific expression in the microvasculature. Arterioscler Thromb Vasc Biol. 2008;28(8):1469–1476.
22. Towner RA, He T, Doblas S, Smith N. Assessment of rodent glioma models using magnetic resonance imaging techniques. In: Chen CC, ed. Advances in the Biology, Imaging and Therapies for Glioblastoma. Rijeka, Croatia: InTech; 2011:251–272.
23. Waerzeggers Y, Monfared P, Viel T, Winkeler A, Jacobs AH. Mouse models in neurological disorders: applications of non-invasive imaging. Biochim Biophys Acta. 2010;1802(10):819–839.
24. Flynn JR, Wang L, Gillespie DL, et al.. Hypoxia-regulated protein expression, patient characteristics, and preoperative imaging as predictors of survival in adults with glioblastoma multiforme. Cancer. 2008;113(5):1032–1042.
25. Zinchuk V, Zinchuk O, Okada T. Quantitative colocalization analysis of multicolor confocal immunofluorescence microscopy images: pushing pixels to explore biological phenomena. Acta Histochem Cytochem. 2007;40(4):101–111.
26. Towner RA, Smith N, Asano Y, et al.. Molecular magnetic resonance imaging approaches used to aid in the understanding of angiogenesis in vivo: implications for tissue engineering. Tissue Eng Part A. 2010;16(2):357–364.
27. Towner RA, Smith N, Doblas S, et al.. In vivo detection of c-Met expression in a rat C6 glioma model. J Cell Mol Med. 2008;12(1):174–186.
28. Towner RA, Smith N, Asano Y, et al.. Molecular magnetic resonance imaging approaches used to aid in the understanding of the tissue regeneration marker Met in vivo: implications for tissue engineering. Tissue Eng Part A. 2010;16(2):365–371.
29. Phillips HS, Kharbanda S, Chen R, et al.. Molecular subclasses of high-grade glioma predict prognosis, delineate a pattern of disease progression, and resemble stages in neurogenesis. Cancer Cell. 2006;9(3):157–173.
30. Verhaak RG, Hoadley KA, Purdom E, et al.. Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1. Cancer Cell. 2010;17(1):98–110.
31. Dozmorov MG, Wren JD. High-throughput processing and normalization of one-color microarrays for transcriptional meta-analyses. BMC Bioinformatics. 2011;12(suppl 10):S2.
32. Dozmorov MG, Giles CB, Wren JD. Predicting gene ontology from a global meta-analysis of 1-color microarray experiments. BMC Bioinformatics. 2011;12(suppl 10):S14.
33. Daum JR, Wren JD, Daniel JJ, et al.. Ska3 is required for spindle checkpoint silencing and the maintenance of chromosome cohesion in mitosis. Curr Biol. 2009;19(17):1467–1472.
34. Lupu C, Zhu H, Popescu NI, Wren JD, Lupu F. Novel protein ADTRP regulates TFPI expression and function in human endothelial cells in normal conditions and in response to androgen. Blood. 2011;118(16):4463–4471.
35. Clemmensen SN, Bohr CT, Rørvig S, et al.. Olfactomedin 4 defines a subset of human neutrophils. J Leukoc Biol. 2012;91(3):495–500.
36. Goel R, Muthusamy B, Pandey A, Prasad TS. Human protein reference database and human proteinpedia as discovery resources for molecular biotechnology. Mol Biotechnol. 2011;48(1):87–95.
37. Jensen RL. Brain tumor hypoxia: tumorigenesis, angiogenesis, imaging, pseudoprogression, and as a therapeutic target. J Neurooncol. 2009;92(3):317–335.
38. Gillespie DL, Flynn JR, Ragel BT, et al.. Silencing of HIF-1alpha by RNA interference in human glioma cells in vitro and in vivo. Methods Mol Biol. 2009;487:283–301.
39. Ragel BT, Couldwell WT, Gillespie DL, Jensen RL. Identification of hypoxia-induced genes in a malignant glioma cell line (U-251) by cDNA microarray analysis. Neurosurg Rev. 2007;30(3):181–187.
40. Gillespie DL, Whang K, Ragel BT, Flynn JR, Kelly DA, Jensen RL. Silencing of hypoxia inducible factor-1α by RNA interference attenuates human glioma cell growth vivo. Clin Cancer Res. 2007;13(8):2441–2448.
41. Rong Y, Hu F, Huang R, et al.. Early growth response gene-1 regulates hypoxia-induced expression of tissue factor in glioblastoma multiforme through hypoxia-inducible factor-1-independent mechanisms. Cancer Res. 2006;66(14):7067–7074.
42. Jensen RL. Hypoxia in the tumorigenesis of gliomas and as a potential target for therapeutic measures. Neurosurg Focus. 2006;20(4):E24.
43. Jensen RL, Ragel BT, Whang K, Gillespie D. Inhibition of hypoxia inducible factor-1alpha (HIF-1alpha) decreases vascular endothelial growth factor (VEGF) secretion and tumor growth in malignant gliomas. J Neurooncol. 2006;78(3):233–247.
44. Sibenaller ZA, Etame AB, Ali MM, et al.. Genetic characterization of commonly used glioma cell lines in the rat animal model system. Neurosurg Focus. 2005;19(4):E1.
45. Barth RF, Kaur B. Rat brain tumor models in experimental neuro-oncology: the C6, 9L, T9, RG2, F98, BT4C, RT-2 and CNS-1 gliomas. J Neurooncol. 2009;94(3):299–312.
46. Barth RF. Rat brain tumor models in experimental neuro-oncology: the 9L, C6, T9, F98, RG2 (D74), RT-2 and CNS-1 gliomas. J Neurooncol. 1998;36(1):91–102.
47. Asanuma T, Doblas S, Tesiram YA, et al.. Visualization of the protective ability of a free radical trapping compound against rat C6 and F98 gliomas with diffusion tensor fiber tractography. J Magn Reson Imaging. 2008;28(3):574–587.
48. Doblas S, He T, Saunders D, et al.. Glioma morphology and tumor-induced vascular alterations revealed in seven rodent glioma models by in vivo magnetic resonance imaging and angiography. J Magn Reson Imaging. 2010;32(2):267–275.
49. Sipkins DA, Cheresh DA, Kazemi MR, Nevin LM, Bednarski MD, Li KC. Detection of tumor angiogenesis in vivo by alphaVbeta3-targeted magnetic resonance imaging. Nat Med. 1998;4(5):623–626.
50. Kiessling F, Huppert J, Zhang C, et al.. RGD-labeled USPIO inhibits adhesion and endocytotic activity of alpha v beta3-integrin-expressing glioma cells and only accumulates in the vascular tumor compartment. Radiology. 2009;253(2):462–469.
51. Zhang D, Feng XY, Henning TD, et al.. MR imaging of tumor angiogenesis using sterically stabilized Gd-DTPA liposomes targeted to CD105. Eur J Radiol. 2009;70(1):180–189.
52. He T, Smith N, Saunders D, et al.. Molecular MRI assessment of vascular endothelial growth factor receptor-2 in rat C6 gliomas. J Cell Mol Med. 2011;15(4):837–849.
53. Towner RA, Smith N, Doblas S, et al.. In vivo
detection of inducible nitric oxide synthase in rodent gliomas. Free Radic Biol Med. 2010;48(5):691–703.
In this article, the authors found ELTD1, a transmembrane protein with a large extracellular domain with EGF-like repeats, to be probably a glioma-associated marker by using the bioinformatic method that they had used before. Then they confirmed preliminarily that ELTD1 was expressed in human gliomas and the expression of ELTD1 correlated positively with the grading of gliomas through immunohistochemistry (IHC). The results of in vivo MR experiments on rats with F98 glioma were also promising for ELTD1 to be a biomarker of gliomas, especially of high-grade gliomas and GBM. The results of the study indicate that ELTD1 could be a new useful marker in the diagnosis and grading of gliomas, especially with molecular MRI. With further corroborative studies, ELTD1 could be a clinically useful marker for the diagnosis and treatment of gliomas.
The authors have identified a novel tumor biomarker, ELTD1, for high-grade gliomas via a unique bioinformatics meta-analysis of approximately 18 000 publicly available gene expression profiles. Their protocol identified glioma-associated genes not previously reported in the literature. ELTD1 protein is preferentially expressed by tumor endothelial cells of high-grade gliomas. Antibody-based nanoprobes for ELTD1 were synthesized and shown to localize to an F98 rat glioma orthotopic model. Finally, ELTD1 expression was positively correlated with glioma grade in the Rembrandt and Erasmus databases, associated with decreased survival across all glioma grades, and associated with the mesenchymal GBM subclass.1 This work highlights the utility of comprehensive, unbiased screens for novel glioma biomarkers, for such work opens up new avenues of investigation.
Discovery of novel biomarkers such as ELTD1 may aid in determining more precise subtypes and histopathological classifications of gliomas that potentially have clinical significance in tumor biology and assessing therapeutic response. One crucial future study is to validate ELTD1 protein expression and associated change in patient survival via clinically annotated glioma tissue microarrays to confirm the survival differences observed in Rembrandt. The positive correlation of ELTD1 mRNA expression across increasing grades of glioma; its origins in the epidermal growth factor, 7-transmembrane subfamily; and localized expression to endothelial cells all suggest potential ELTD1 involvement in tumorigenic mechanisms, and further study could yield new therapeutic targets. Such novel biomarkers may also be useful for assessing therapeutic response, especially with newly available targeted therapies. In the era of personalized medicine, research dedicated to systematically characterizing predictive tumor biomarkers is increasingly useful for classifying tumor patients for designing efficient clinical trial assessment of new targeted therapies.
John S. Kuo
1. Phillips HS, Kharbanda S, Chen R, et al.. Molecular subclasses of high-grade glioma predict prognosis, delineate a pattern of disease progression, and resemble stages in neurogenesis. Cancer Cell. 2006;9:157–173. PubMed | CrossRef Cited Here... |
The authors have demonstrated the utility of ELTD1, a novel biomarker likely related to endothelial neovascularization, in the diagnosis and grading of glioblastoma (GBM). Extensive immunohistochemical analysis of the rat and human GBM tissue, as well as molecular MRI analysis of the F98 glioma model in rats, demonstrated a significant association between ELTD1 expression and glioma grade. It is also suggested that increased ELTD1 expression was indicative mesenchymal subtype of GBM. Although an increase in ELTD1 negatively correlated with survival across different grades of glioma, it was not useful for predicting survival of GBM patients.
The authors should be commended for both their advanced approach to identifying a candidate biomarker (via sophisticated bioinformatics to analyze 3651 microarrays, followed by computer-assisted meta-analysis of published literature on the top 20 candidates), as well as their rigorous evaluation in both animal and human tissue. The association between ELTD1 and CD31, an endothelial cell marker, suggests that ELTD1 expression is a marker for endothelial proliferation and neovascularization, important histological and radiological features of aggressive gliomas.1 Another study2 of the transcriptional profile of human GBM tissue suggests similar results. As the authors correctly state, further study of the effects of current antiangiogenic therapy (bevacizumab or sunitinib) on ELTD1 expression could help confirm such an association. When combined with the emerging techniques of molecular MRI, neurooncologists could eventually use ELTD1 probes in the presurgical diagnosis and grading of suspected glioma, as well as quantifying the efficacy of subsequent antiangiogenic therapy.
The goal of biomarker research in glioma is twofold: first, aiding in the diagnosis of the disease and its grades and subtypes, and second, identifying potential therapeutic targets. The authors have taken the first step by providing evidence that ELTD1 is useful in the detection and grading of gliomas in humans and is an important histological marker of their neovascularization.
Michael R. Levitt
Daniel L. Silbergeld
1. Russell SM, Elliott R, Forshaw D, Golfinos JG, Nelson PK, Kelly PJ. Glioma vascularity correlates with reduced patient survival and increased malignancy. Surg Neurol. 2009;72(3):242–246; discussion 246-247. Cited Here...
2. Dieterich LC, Mellberg S, Langenkamp E, et al.. Transcriptional profiling of human glioblastoma vessels indicates a key role of VEGF-A and TGFbeta2 in vascular abnormalization. J Pathol. 2012;228:378–390. View Full Text | PubMed | CrossRef Cited Here... |
Molecular biomarkers have increasingly been used to assess and manage cancer. In this article, the authors used a very high-tech methodology (advanced laboratory and imaging techniques + bioinformatic tools in humans as well as in animal models) to propose ELTD1 as a novel and potential glioma-associated biomarker. ELTD1 expression was found to be higher in high-grade than low-grade gliomas with a clear association with tumor grade and patient survival. Moreover, ELTD1 was validated as a specific marker of neoplastic angiogenesis, showing an extremely lower expression in normal brain tissue than in tumors.
From the technical point of view, the article is highly detailed, giving information for the reproducibility of the experimental design, even if, obviously, many tools described are not familiar to a large neurosurgical audience. Notwithstanding this, the article appears very interesting and informative for the neuroscientific community, and the authors should be commended for their results as well as for the experimental background (7T MR and bioinformatics applied in neuropathology).
It should be emphasized that new biomarkers are proposed daily in the scientific literature; however, just a few of them have been successfully translated into clinical practice. Our opinion is that finding and statistically demonstrating that a biomarker is a meaningful biomarker does not mean that a real clinically useful parameter has been found, but something like an epiphenomenon occurring along the pathology. The proposal of a protein as a biological biomarker or of a morphometric parameter as an image biomarker should undergo systematic validation, after the technical and statistical steps, in a more specific way. The proposed biomarkers should undergo a step-by-step hierarchical validation system before being proposed as clinically meaningful, in the same way as for drugs, from experimentation to clinical application, from bench to bedside.1 Moreover, it is our opinion that molecular biomarkers should be investigated in parallel with objective morphometric parameters; in the case of ELTD1, which is associated with neoangiogenesis, for example, the analyses should run in parallel with the morphometric analyses of the microvessels, in terms of number, density, and geometric complexity of the microvascular networks, which are geometrically different in physiological vs pathological states as well as in different types and subtypes of tumors.2,3 There is no doubt that further investigations could give ELTD1 the “dignity” to become a clinically meaningful biomarker, being added in the near future to the family of other glioma-specific biomarkers.
Antonio Di Ieva
Toronto, Ontario, Canada
1. Hainfellner JA, Heinzl H. Neuropathological biomarker candidates in brain tumors: key issues for translational efficiency. Clin Neuropathol. 2010;29:41–54. PubMed | CrossRef Cited Here... |
2. Di Ieva A. Angioarchitectural morphometrics of brain tumors: are there any potential histopathological biomarkers? Microvasc Res. 2010;80:522–533. Cited Here...
3. Di Ieva A. Fractal analysis of microvascular networks in malignant brain tumors. Clin Neuropathol. 2012;31:342–351. PubMed | CrossRef Cited Here... |
ELTD1; Glioblastoma multiforme; Gliomas; Immunohistochemistry; Molecular magnetic resonance imaging, Rat F98 glioma model
Supplemental Digital Content
Copyright © by the Congress of Neurological Surgeons