Solid tumours are an enormous cancer burden and a major therapeutic challenge. The cancer stem cell (CSC) hypothesis provides an attractive cellular mechanism to account for the therapeutic refractoriness and dormant behaviour exhibited by many of these tumours. There is increasing evidence that diverse solid tumours are hierarchically organized and sustained by a distinct subpopulation of CSCs. Direct evidence for the CSC hypothesis has recently emerged from mouse models of epithelial tumorigenesis, although alternative models of heterogeneity also seem to apply. The clinical relevance of CSCs remains a fundamental issue but preliminary findings indicate that specific targeting may be possible.

Cancer stem cells in solid tumours: accumulating evidence and unresolved questions

Stem cells are defined as cells able to both extensively self-renew and differentiate into progenitors. Research data show that more resistant stem cells than common cancer cells exist in cancer patients.To identify unrecognized differences between cancer stem cells and cancer cells might be able to develope effective classification,diagnose and treat ment for cancer.

Stem Cells,Cancer and Cancer Stem Cells

Tissues with defined cellular hierarchies in development and homeostasis give rise to tumors with cellular hierarchies, suggesting that tumors recapitulate specific tissues and mimic their origins. Glioblastoma (GBM) is the most prevalent and malignant primary brain tumor and contains self-renewing, tumorigenic cancer stem cells (CSCs) that contribute to tumor initiation and therapeutic resistance. As normal stem and progenitor cells participate in tissue development and repair, these developmental programs re-emerge in CSCs to support the development and progressive growth of tumors. Elucidation of the molecular mechanisms that govern CSCs has informed the development of novel targeted therapeutics for GBM and other brain cancers. CSCs are not self-autonomous units; rather, they function within an ecological system, both actively remodeling the microenvironment and receiving critical maintenance cues from their niches. To fulfill the future goal of developing novel therapies to collapse CSC dynamics, drawing parallels to other normal and pathological states that are highly interactive with their microenvironments and that use developmental signaling pathways will be beneficial.

/pdf/cancer-stem-cells-in-glioblastoma-56rj4maz9b.pdf

Cancer stem cells in glioblastoma

Mechanisms of Hedgehog pathway activation in cancer and implications for therapy

Overexpression of the polycomb group protein enhancer of zeste homologue 2 (EZH2) occurs in diverse malignancies, including prostate cancer, breast cancer, and glioblastoma multiforme (GBM). Based on its ability to modulate transcription of key genes implicated in cell cycle control, DNA repair, and cell differentiation, EZH2 is believed to play a crucial role in tissue-specific stem cell maintenance and tumor development. Here, we show that targeted pharmacologic disruption of EZH2 by the S-adenosylhomocysteine hydrolase inhibitor 3-deazaneplanocin A (DZNep), or its specific downregulation by short hairpin RNA (shRNA), strongly impairs GBM cancer stem cell (CSC) self-renewal in vitro and tumor-initiating capacity in vivo. Using genome-wide expression analysis of DZNep-treated GBM CSCs, we found the expression of c-myc, recently reported to be essential for GBM CSCs, to be strongly repressed upon EZH2 depletion. Specific shRNA-mediated downregulation of EZH2 in combination with chromatin immunoprecipitation experiments revealed that c-myc is a direct target of EZH2 in GBM CSCs. Taken together, our observations provide evidence that direct transcriptional regulation of c-myc by EZH2 may constitute a novel mechanism underlying GBM CSC maintenance and suggest that EZH2 may be a valuable new therapeutic target for GBM management.

https://cancerres.aacrjournals.org/content/canres/69/24/9211.full.pdf

EZH2 Is Essential for Glioblastoma Cancer Stem Cell Maintenance

The prognosis of patients suffering from glioblastoma (GBM) is dismal despite multimodal therapy. Although chemotherapy with temozolomide may contain tumor growth for some months, invariable tumor recurrence suggests that cancer stem cells (CSC) maintaining these tumors persist. We have therefore investigated the effect of temozolomide on CD133(+) and CD133(-) GBM CSC lines. Although differentiated tumor cells constituting the bulk of all tumor cells were resistant to the cytotoxic effects of the substance, temozolomide induced a dose- and time-dependent decline of the stem cell subpopulation. Incubation with sublethal concentrations of temozolomide for 2 days completely depleted clonogenic tumor cells in vitro and substantially reduced tumorigenicity in vivo. In O(6)-methylguanine-DNA-methyltransferase (MGMT)-expressing CSC lines, this effect occurred at 10-fold higher doses compared with MGMT-negative CSC lines. Thus, temozolomide concentrations that are reached in patients were only sufficient to completely eliminate CSC in vitro from MGMT-negative but not from MGMT-positive tumors. Accordingly, our data strongly suggest that optimized temozolomide-based chemotherapeutic protocols might substantially improve the elimination of GBM stem cells and consequently prolong the survival of patients.

https://cancerres.aacrjournals.org/content/canres/68/14/5706.full.pdf

Temozolomide preferentially depletes cancer stem cells in glioblastoma.

Underwater images are challenging for correspondence search algorithms, which are traditionally designed based on images captured in air and under uniform illumination. In water however, medium interactions have a much higher impact on the light propagation. Absorption and scattering cause wavelength- and distance-dependent color distortion, blurring and contrast reductions. For deeper or turbid waters, artificial illumination is required that usually moves rigidly with the camera and thus increases the appearance differences of the same seafloor spot in different images. Correspondence search, e.g. using image features, is however a core task in underwater visual navigation employed in seafloor surveys and is also required for 3D reconstruction, image retrieval and object detection. For underwater images, it has to be robust against the challenging imaging conditions to avoid decreased accuracy or even failure of computer vision algorithms. However, explicitly taking underwater nuisances into account during the feature extraction and matching process is challenging. On the other hand, learned feature extraction models achieved high performance in many in-air problems in recent years. Hence we investigate, how such a learned robust feature model, D2Net, can be applied to the underwater environment and particularly look into the issue of cross domain transfer learning as a strategy to deal with the lack of annotated underwater training data. 

Towards Cross Domain Transfer Learning for Underwater Correspondence Search

The adoption of machine learning (ML) technology in real-world settings like medical imaging is currently hampered by a lack of trust in ML models and a lack of labeled data. These two issues are currently addressed in parallel by two subdisciplines of ML: uncertainty quantification (UQ) is concerned with obtaining reliable estimates of a model's confidence in its outputs, and active learning (AL) deals with efficiently training models in low-data regimes. To date, the usefulness of the new methods emerging from the field of UQ for AL remains under-explored. We here take a step to address this by comparing seven different UQ methods on three image classification data sets. Our experiments confirm previous indications in the AL literature that the ranking of sampling strategies can vary greatly across models and data sets. We find that Concrete Dropout, Least Confidence, Smallest Margin, and Entropy sampling consistently outperform Random sampling across data sets, whereas Ensembles, Monte-Carlo Dropout, and Bayes-by-Backprop do not. We also observe that AL training stability is sensitive to data quality.

A Comparison of Uncertainty Quantification Methods for Active Learning in Image Classification

The quality of datasets plays a crucial role in the successful training and deployment of deep learning models. Especially in the medical ﬁeld, where system performance may impact the health of pa-tients, clean datasets are a safety requirement for reliable predictions. Therefore, outlier detection is an essential process when building autonomous clinical decision systems. In this work, we assess the suitability of Self-Organizing Maps for outlier detection speciﬁcally on a medical dataset con-taining quantitative phase images of white blood cells. We detect and evaluate outliers based on quantization errors and distance maps. Our ﬁnd-ings conﬁrm the suitability of Self-Organizing Maps for unsupervised Out-Of-Distribution detection on the dataset at hand. Self-Organizing Maps perform on par with a manually speciﬁed ﬁlter based on expert domain knowledge. Additionally, they show promise as a tool in the exploration and cleaning of medical datasets. As a direction for future research, we suggest a combination of Self-Organizing Maps and feature extraction based on deep learning.

Stefanie Röhrl

Papers

Temozolomide preferentially depletes cancer stem cells in glioblastoma.

Towards Cross Domain Transfer Learning for Underwater Correspondence Search

A Comparison of Uncertainty Quantification Methods for Active Learning in Image Classification

Outlier Detection using Self-Organizing Maps for Automated Blood Cell Analysis

Explainable Feature Learning with Variational Autoencoders for Holographic Image Analysis