A brain tumor research group at the UW Carbone Cancer Center focuses on biological studies of patient-derived glioblastoma multiforme (GBM) cancer stem cells, combined with analysis of patient-matched serum-cultured GBM and an annotated GBM tissue microarray, to identify clinically relevant biomarkers.
John S. Kuo, MD, PhD, FAANS, FACS
Brain Tumor Program
Chair, CNS Tumors Group
Carbone Cancer Center
More than a decade ago, cancer cells with stem-like properties were first identified in human solid tumors (eg, breast cancer, glioblastoma).1,2 Unlike cancer cell lines isolated from surgical specimens and grown in standard media, patient-derived cancer stem cells (CSCs) were defined by their ability to grow as non-adherent sphere cultures in stem cell media, to express progenitor markers and manifest multipotent differentiation, and to generate phenotypically diverse cancer cells in orthotopic xenografts with remarkably high efficiency.3
Many studies support the hypothesis that CSCs are therapeutically resistant cells that can cause tumor recurrence, implying that targeting CSCs could improve outcomes for incurable cancers such as the adult primary brain tumor glioblastoma multiforme (GBM).4 Compared with serum-cultured lines, CSC lines also retain more genetic similarity to patients’ tumors and generate patient-derived xenograft models for clinically relevant studies.5
CSC Studies Yield Clinically Relevant Answers
Our brain tumor research group at the UW Carbone Cancer Center focuses on biological studies of patient-derived GBM CSCs, combined with analysis of patient-matched serum-cultured GBM and an annotated GBM tissue microarray (TMA), to identify clinically relevant biomarkers. Patientderived tumor lines and orthotopic xenografts are also valuable resources for in vitro and in vivo testing of novel agents and therapeutic strategies.Our observation that GBM CSCs can survive in minimal media without exogenous growth factors partly explained the failure of EGFR-targeted therapies for GBM, despite their efficacy for lung and breast cancer. The addition of EGFR inhibitors causes CSCs to rapidly upregulate and increase expression of other EGFR-related receptors such as HER2 and HER3, and multireceptor inhibition was required for therapeutic effect.6 We also characterized the genetic heterogeneity of GBMs by subgrouping CSC lines according to neural lineage markers, and identified cyclic nucleotide phosphodiesterase and cadherin-19 as new GBM biomarkers.7,8
Expression of these biomarkers was associated with less invasive CSC-derived xenografts and better survival of patients with GBM, showing the clinical relevance of CSC-identified tumor biomarkers. Differential microRNA (miR) screening of CSCs compared with normal neural stem cells identified miR-100 as a possible tumor suppressor in GBM, and initial results show promise for miR-100 pathway- based therapeutic strategies.9
GBM is difficult to treat due to cellular invasion and infiltration of surrounding brain tissue. We discovered that organized collagen in the tumor microenvironment is associated with less invasive xenografts and serves as a biomarker for improved patient survival on TMA analysis.10 Collagen is not abundantly expressed in the brain, whereas it is abundantly found in the rest of the body; in contrast, expression of organized collagen is a prometastatic signature for breast cancer and also is associated with worse survival.11 The discovery about tissue-dependent tumorigenic mechanisms could inform future therapeutic development.
Our CSC work also yielded insights for developing cancer therapeutics against ion channels that are aberrantly expressed in many cancers and are involved in tumor proliferation, survival, invasion, and angiogenesis. We analyzed the expression of Ether-à-go-go—related gene (hERG) potassium channels in patient-derived CSC cells and CSC-derived xenografts, and TMA analysis of patient GBM specimens. High hERG expression correlated with increased xenograft proliferation, and patients with higher hERG expression experienced worse survival. Inhibition of hERG also decreased CSC sphere growth and proliferation.
A promising finding from GBM TMA analysis showed that patients who incidentally received off-target hERG inhibitory drugs experienced improved survival, but only for the high hERG-expressing GBM patients.
These data suggest that "non-torsadogenic" hERG inhibition (that does not promote cardiac arrhythmia) is a promising strategy for the subgroup of patients with high hERG-expressing GBM.12
Testing Novel Cancer-Targeting Agents
The study also suggests the utility of hERG as a biomarker for patient selection in clinical trials for hERG-targeted agents, possibly by repurposing commonly prescribed drugs with hERG inhibitory activity such as the antidepressant fluoxetine or the antiepiletic drug phenytoin. This also raises the possibility of GBM adjuvant therapies that might also treat associated comorbidities such as depression or seizure. This paradigm may also apply to other cancer types because hERG misexpression is found in neoplasms such as leukemia, melanoma, pancreatic, colorectal, and gastric cancers.13Our patient-derived CSCs and related xenografts were also used in testing new synthetic cancer-targeting alkylphosphocholine (APC) analogs that were created for diagnostic cancer imaging throughout the body and brain, intraoperative tumor fluorescence visualization, and therapy applications.
APC analogs are selectively retained in 55 tested rodent and human cancers and cancer stem cell models representing over 12 different cancers, whereas normal tissues metabolize and eliminate APC analogs.14 APC analogs show very low background in normal brain tissue and high avidity and specificity for tumors. Multiple APC analogs have demonstrated safety in early phase human clinical trials, and are quite promising for in vivo fluorescent detection of tumor cells.14,15,16
The first iterations of these versatile theranostic APC analogs can be radiolabeled for either PET imaging or molecular radiotherapy, or linked with fluorophores for intraoperative tumor cell visualization.17
These modifications do not change the ability of APCs to target many different cancers throughout the body, including detection of primary brain tumors such as GBM, patient-derived CSCs, and metastatic cancers to the brain from many different systemic cancers.16 APC analogs are a promising multimodal theranostic agent with broad capability to detect many different types of cancer throughout the body and brain, possibly discriminate between tumor recurrence and treatment-related effects (pseudoprogression), provide real-time intraoperative fluorescent tumor visualization, and target therapy to residual or recurrent CSCs and tumor tissue. By definition, each metastasis must be “seeded” by a cancer cell derived from the initial neoplasm that is capable of recapitulating and growing a new cancer lesion in another tissue environment.
This perspective would imply that cancer metastasis results from cells that behave very much like CSCs; therefore, CSC-relevant biology and approaches could also inform future studies and therapeutic advances for metastatic cancer.