In this report, we have included: 1) a detailed clinical course, 2) radiological findings, 3) the surgical approach and its results, 4) pathological details, 5) marker expression analysis of tumor cells derived from the CD133-positive cells, and 6) evidence for ex vivo and in vivo behavior including tumor-initiating capacity. Clinically, it is of great interest to have a successful isolation of glioblastoma stem cells from a rare GBM that involves the neurogenic ventricular wall. We have found in this rare case that a tumorigenic CD133-positive progenitor cell phenotype is part of the tumor. The mRNA expression of an array of heterotypic biomarkers may explain the course of this patient's clinical outcome as gene expression indicates the participation of unique cancer-related transcripts specifically related to GBM stem cells, such as caveolin-1 and −2. Their expression in GBM CSC has not been previously reported in the literature.
GBMs usually form in the cerebral white matter, grow quickly, and can become large before producing symptoms. Malignant tumor cells infiltrate from primary tumor sites to nearby tissues, representing the major cause of death in patients. In the clinic, the intrinsic infiltration of single glioma cells into brain parenchyma renders these cancers resistant to the current treatment of surgical removal in combination with radiation-, chemo- and immuno-therapies
. Invariable infiltration into adjacent brain parenchyma, crossing commissures to expand to the opposite cerebral hemisphere, is a hallmark of the malignancy of GBM. Thus, despite recent advances in surgical and medical therapy, the prognosis for patients diagnosed with high-grade GBM remains poor. The realization that a self-replication mechanism may be shared by both normal stem cells and cancer cells has led to the new concept of the cancer stem cell (CSC)
[6, 32]. Similar mechanisms may control normal and cancer stem cell properties. This concept as has been supported by reports that showed the existence of a cancer stem cell population in human brain tumors of both children and adults with different phenotypes
[33–35]. Both normal and tumor stem cell populations are heterogeneous with respect to proliferation and differentiation. The difference between normal neural stem cells and tumor stem cells has not been fully defined
[7, 36], but it has been speculated that brain tumor stem cells may be a cause of the resistance of tumors to conventional treatments, and high recurrence rate
[37–40]. However, targeted elimination of tumor stem cells may be detrimental if it also eliminates normal neural stem cells. In our study, glioblastoma stem cells from a rare GBM that involves the neurogenic ventricular wall may tackle and hijack the source of the normal neural stem cells that reside in neurogenic ventricles.
The hallmark of the malignant glioblastoma is its diverse marker expression. Marker expression in the prognosis of malignant brain tumors has been explored, the main issue being the heterogeneous expression of most of the genes examined
[41–50]. We have presented evidence of the successful isolation and characterization of a small subpopulation of cancer stem cells. The molecular features of these tumor cells may provide potential new therapeutic targets, and therefore strategies that may control them. Certain molecular markers are consistent with those previously reported
. For example, Murat and colleagues (2008) provided the first clinical evidence for the implication of high epidermal growth factor receptor (EGFR) expression associated with resistance to concomitant chemoradiotherapy in a “glioblastoma stem cell" or "self-renewal" phenotype
The clongeneity of these single CD133 positive cells showed biological differences in the growth capacity as shown in Figure
4 and Figure
7. In fact, Dr. Cavenee and Dr. Furnari and colleagues showed that CSCs undergo clonal evolution from a single GBM cancer stem cell to extensive heterogeneity at the cellular and molecular levels
. The single-cell generated heterogeneity confers a biological advantage to the tumor by creating an intratumoral and tumor-microenvironment community that serves to maintain the heterogeneous tumor composition and to promote tumor growth. This tumor community allows interactions between CSCs and/or tumor cells and their environment and between different CSCs and /or tumor cell subclones. Those interactions need to balance out. An inbalance may drive tumor growth, drug resistance, immune suppression, angiogenesis, invasion, migration, or more CSC renewal. We suggested that a delicate balance may be modulated by innovative therapeutics to keep the tumor in surveillance check
. We thought that in the context of stem cell development, there is a parallel with the concept of quiescent or dormant cancer stem cells (CSCs) and their progeny, the differentiated cancer cells; these two populations communicate and co-exist. The mechanism with which determines to extend self-renewal and expansion of CSCs is needed to elucidate.
CD133 (prominin-1), a neural stem cell (NSC) marker implicated in brain tumors, notably glioblastoma, was highly expressed in our material. Interestingly, CD133 is also expressed in the glioma cell lines U251 and U87MG
. Remarkably, a recent study showed that the level of membrane particle-associated CD133 is elevated in early stage glioblastoma patients and decreases dramatically in the final stage of the disease
. This change may be used for diagnosing and surveying glioblastoma initiation and progression
[55, 56]. More clinically relevant, CD133 is associated with specific extracellular membrane particles in cerebrospinal fluid, which can be routinely used for diagnosis and prognosis in neurological diseases. Malignant brain tumors have a higher CD133 index than low-grade tumors
. Purified populations of CD133-positive tumor cells injected into the brains of NOD/SCID mice induced tumors that were heterogeneous and had the characteristic of infiltration
[58, 59]. It has also been shown that transplantation of neurospheres derived from glioblastoma tumor cells cultured in EGF and bFGF-containing media drove tumor formation in immune-deficient mouse models
[60, 61]. These CD133-positive tumor cells may be a leading force for reinitiating tumor genesis and progression
. However, there is debate about the lineage relationship between normal NSCs and brain cancer stem cells. It is not yet fully understood if CD133-positive brain CSCs are derived from CD133-positive normal NSCs. Thus, it is still questionable if tumor therapies can be developed for targeted destruction of CSCs without damaging normal NSCs. Dr. Bota and colleagues have recently found that both the proteasome inhibitor bortezomib (BTZ) and the epidermal growth factor receptor tyrosine kinase inhibitor erlotinib (ERL) decreased glioma stem-like cells (GSCs) proliferation but not NSC viability
. Surprisingly, commonly used temozolomide (TMZ) and cisplatin (CIS) were more toxic for NSCs than for GSCs. This in vitro observation may inspire a new journey to search for GSC-specific destruction agents, which are not detrimental to NSCs.
Angiogenesis is a critical component of brain tumor growth. Consistent with our pathological findings, VEGF is highly expressed, confirming that neovasculization is driven by the up-regulation of VEGF around tumors. Recent clinical trials of antivascular endothelial growth factor agents for glioblastoma show promising progression-free and better overall survival rates, even without inhibiting tumor growth
The intermediate filament protein, Nestin, and the RNA-binding protein, Musashi, are expressed by NSCs during CNS development. Their expression in glial tumors correlated with the levels of Cysteine Cathepsins
 that are known as prognostic markers of several tumors
. Nestin is a strong prognostic marker of glioma malignancy; the invasive cells may well be closely related to glioma stem cells
, which our data confirms. Nestin functions in the organization of the cytoskeleton, cell signaling, organogenesis, and cell metabolism. It is down-regulated in mature cells, whereas GFAP, neurofilaments, and PDGFR are expressed in differentiated astrocytes, neurons, and oligodendrocytes, respectively
. Neoplastic transformation up-regulates Nestin expression in astrocytes of the adult CNS, suggesting that its reactivation may relate to tumor genesis
. Nestin has been shown to be a strong prognostic marker for glioma malignancy and its expression correlates with patient survival
. We have found Nestin expressed in both CD133-positive tumor cells and differentiated tumor cells, although the latter with down-regulation, which suggests the existence of residual neural stem cells after induced differentiation.
Peptidases hydrolyze macromolecular components of the extracellular matrix, support the malignant invasive behavior of brain tumor cells, and promote brain tumor progression by advancing tumor angiogenesis
[69–71]. Peptidases consist of matrix metalloproteinases (MMPs), Cathepsins, and Plasminogen activators. Among MMPs, MMP2 and MMP9 strongly correlate with glioma progression
[72–74]. Most importantly, Wong and colleagues found that increased cerebrospinal fluid (CSF) MMP-9 activity could be a biomarker of disease activity in patients with malignant gliomas, before any changes are detectable on MRI
. Lysosomal Cathepsin B is highly expressed in malignant glial cells and endothelial cells of vascularized glioblastoma, an indication of a shorter survival time. Besides invasion, Cathepsin L may play a role in decreased susceptibility of anaplastic glioma cells to apoptosis
[76, 77]. Cathepsin B has been considered a marker for malignancy in the more aggressive type of meningiomas
; developing inhibitors of these peptidases might help control local spread
Originally identified as an oncogenic partner of c-Myc in murine lymphoma genesis, Bmi-1 is a member of the polycomb group transcriptional repressors
[79, 80]. Bmi-1, a proto-oncogene for inhibition of p53 involved in cell cycle and self-renewal, is required for the postnatal maintenance of stem cells in multiple tissues, including the central nervous system (CNS) and peripheral nervous system (PNS). Bmi-1 was highly expressed in the GBM tumor cells we cultured from our case, consistent with a previous report
. Targeting of the Bmi-1 in stem cells by microRNA-128 inhibits glioma proliferation and self-renewal, implying that miRNA-128 may be a therapeutic target agent for the "stem cell-like" characteristics of glioma
Finally, we have found that Caveolin-1 and Caveolin-2 are expressed in our CD133-positive lineage (Figure
9). Interestingly, their expression in GBM CSCs has not been previously reported in the literature. Rather, this has been reported in commercialized glioma non-stem cell lines, such as glioblastoma cell line U87MG
. However, their clinical significance in brain tumor diagnosis and prognosis remains to be determined. Caveolin-1 has been found in detergent-resistant plasma membrane microdomains involved in signaling transduction in many cell types, including neurons and astrocytes
[83–85]. It is a secreted biomarker in some pathological conditions
. In prostate cancer, high preoperative serum Caveolin-1 levels have been established as a biochemical predictor of cancer progression and recurrence
, suggesting a poor prognosis (shorter time to cancer recurrence). Lisanti’s group analyzed breast tissue samples from 154 women diagnosed with breast cancer using immunohistochemical staining of stromal Caveolin-1
. Among each subgroup of patients, as grouped by prognostic factors such as hormone status, disease stage or lymph node status, a loss of stromal Caveolin-1 remained the strongest single predictor of breast cancer patient outcome. Progression-free survival (PFS) was also affected by the loss of stromal caveolin-1. The approximate 5-year survival rate for patients positive for stromal Caveolin-1 was 80% vs. 7% for patients negative for stromal caveolin-1, i.e. a ~11.5-fold reduction in 5-year PFS. Caveolin-1 serves not only as a prognostic marker, but also as a means of therapeutic stratification. Caveolin-1 can be detected at breast cancer diagnosis, which is important because high-risk patients would benefit from more aggressive antiangiogenic therapy. A prognostic biomarker present in the stroma rather than the epithelial cancer cell is a paradigm-shift, since a diagnostic test may not require DNA-based technologies for cost-effective identification for high-risk breast cancer patients at diagnosis.
Despite their clinical importance, little is known about the underlying composition and cellular interactions of tumors that govern their degree of malignancy, and consequently, provide targets to control their growth. The diverse biomarker expression reflects the nature of heterogeneity in the tumor, a mixture of cells at different stages of their development. Indeed, Vescovi’s group discovered that at least two types of CSCs bear quite diverse tumorigenic potential and distinct genetic anomalies, yet derive from common ancestor cells within different regions of the same human GBM
. Thus, therapeutic success relies on an effective strategy to select for a therapy to target some particular stage of tumor cell development at which tumor cells are most susceptible to treatment.
The transition from neural stem cells to cancer cells
 may be activated by expression of some cancer driver, characteristic of dominant clones (single cells), but not in every cell
. Cancer cell phenotypes may be derived from such a few dominant single cells with a continuum from single driver stem cells to cancer cells. We may need to define at what point we call it a cancer cell, for which a treatment is needed. Such a point of time in cancer development, namely the therapeutic window
, may be defined by an integrated genomic
 and epigenomic
[92, 93] analyses through applying next-generation sequencing technology. However, the current whole-genome sequencing mainly on the bulk tumor that also includes stromal and immune cells, does not specifically address the tumor-initiating cells (or CSCs). Developing therapeutic window-specific drugs may be realized by using patient-specific cancer stem cell lines for chemical and genetic screens as described previously
. We need to focus on these tumor-initiating cells at a single-cell level. Glioma stem cell lines derived from patients like the one described in our study may be used for single cell analyses.