TSCRE: a comprehensive database for tumor-specific cis-regulatory elements.

Cis-regulatory elements (CREs) and super cis-regulatory elements (SCREs) are non-coding DNA regions which influence the transcription of nearby genes and play critical roles in development. Dysregulated CRE and SCRE activities have been reported to alter the expression of oncogenes and tumor suppressors, thereby regulating cancer hallmarks. To address the strong need for a comprehensive catalogue of dysregulated CREs and SCREs in human cancers, we present TSCRE (http://tscre.zsqylab.com/), an open resource providing tumor-specific and cell type-specific CREs and SCREs derived from the re-analysis of publicly available histone modification profiles. Currently, TSCRE contains 1 864 941 dysregulated CREs and 68 253 dysregulated SCREs identified from 1366 human patient samples spanning 17 different cancer types and 9 histone marks. Over 95% of these elements have been validated in public resources. TSCRE offers comprehensive annotations for each element, including associated genes, expression patterns, clinical prognosis, somatic mutations, transcript factor binding sites, cancer-type specificity, and drug response. Additionally, TSCRE integrates pathway and transcript factor enrichment analyses for each study, enabling in-depth functional and mechanistic investigations. Furthermore, TSCRE provides an interactive interface for users to explore any CRE and SCRE of interest. We believe TSCRE will be a highly valuable platform for the community to discover candidate cancer biomarkers.

Do we really need dice? The hidden region-size biases of segmentation losses.

Most segmentation losses are arguably variants of the Cross-Entropy (CE) or Dice losses. On the surface, these two categories of losses (i.e., distribution based vs. geometry based) seem unrelated, and there is no clear consensus as to which category is a better choice, with varying performances for each across different benchmarks and applications. Furthermore, it is widely argued within the medical-imaging community that Dice and CE are complementary, which has motivated the use of compound CE-Dice losses. In this work, we provide a theoretical analysis, which shows that CE and Dice share a much deeper connection than previously thought. First, we show that, from a constrained-optimization perspective, they both decompose into two components, i.e., a similar ground-truth matching term, which pushes the predicted foreground regions towards the ground-truth, and a region-size penalty term imposing different biases on the size (or proportion) of the predicted regions. Then, we provide bound relationships and an information-theoretic analysis, which uncover hidden region-size biases: Dice has an intrinsic bias towards specific extremely imbalanced solutions, whereas CE implicitly encourages the ground-truth region proportions. Our theoretical results explain the wide experimental evidence in the medical-imaging literature, whereby Dice losses bring improvements for imbalanced segmentation. It also explains why CE dominates natural-image problems with diverse class proportions, in which case Dice might have difficulty adapting to different region-size distributions. Based on our theoretical analysis, we propose a principled and simple solution, which enables to control explicitly the region-size bias. The proposed method integrates CE with explicit terms based on L1 or the KL divergence, which encourage segmenting region proportions to match target class proportions, thereby mitigating class imbalance but without losing generality. Comprehensive experiments and ablation studies over different losses and applications validate our theoretical analysis, as well as the effectiveness of explicit and simple region-size terms. The code is available at https://github.com/by-liu/SegLossBias .

GAMMA challenge: Glaucoma grAding from Multi-Modality imAges.

Glaucoma is a chronic neuro-degenerative condition that is one of the world's leading causes of irreversible but preventable blindness. The blindness is generally caused by the lack of timely detection and treatment. Early screening is thus essential for early treatment to preserve vision and maintain life quality. Colour fundus photography and Optical Coherence Tomography (OCT) are the two most cost-effective tools for glaucoma screening. Both imaging modalities have prominent biomarkers to indicate glaucoma suspects, such as the vertical cup-to-disc ratio (vCDR) on fundus images and retinal nerve fiber layer (RNFL) thickness on OCT volume. In clinical practice, it is often recommended to take both of the screenings for a more accurate and reliable diagnosis. However, although numerous algorithms are proposed based on fundus images or OCT volumes for the automated glaucoma detection, there are few methods that leverage both of the modalities to achieve the target. To fulfil the research gap, we set up the Glaucoma grAding from Multi-Modality imAges (GAMMA) Challenge to encourage the development of fundus & OCT-based glaucoma grading. The primary task of the challenge is to grade glaucoma from both the 2D fundus images and 3D OCT scanning volumes. As part of GAMMA, we have publicly released a glaucoma annotated dataset with both 2D fundus colour photography and 3D OCT volumes, which is the first multi-modality dataset for machine learning based glaucoma grading. In addition, an evaluation framework is also established to evaluate the performance of the submitted methods. During the challenge, 1272 results were submitted, and finally, ten best performing teams were selected for the final stage. We analyse their results and summarize their methods in the paper. Since all the teams submitted their source code in the challenge, we conducted a detailed ablation study to verify the effectiveness of the particular modules proposed. Finally, we identify the proposed techniques and strategies that could be of practical value for the clinical diagnosis of glaucoma. As the first in-depth study of fundus & OCT multi-modality glaucoma grading, we believe the GAMMA Challenge will serve as an essential guideline and benchmark for future research.

Targeting proteasomal deubiquitinases USP14 and UCHL5 with b-AP15 reduces 5-fluorouracil resistance in colorectal cancer cells.

5-Fluorouracil (5-FU) is the first-line treatment for colorectal cancer (CRC) patients, but the development of acquired resistance to 5-FU remains a big challenge. Deubiquitinases play a key role in the protein degradation pathway, which is involved in cancer development and chemotherapy resistance. In this study, we investigated the effects of targeted inhibition of the proteasomal deubiquitinases USP14 and UCHL5 on the development of CRC and resistance to 5-FU. By analyzing GEO datasets, we found that the mRNA expression levels of USP14 and UCHL5 in CRC tissues were significantly increased, and negatively correlated with the survival of CRC patients. Knockdown of both USP14 and UCHL5 led to increased 5-FU sensitivity in 5-FU-resistant CRC cell lines (RKO-R and HCT-15R), whereas overexpression of USP14 and UCHL5 in 5-FU-sensitive CRC cells decreased 5-FU sensitivity. B-AP15, a specific inhibitor of USP14 and UCHL5, (1-5 µM) dose-dependently inhibited the viability of RKO, RKO-R, HCT-15, and HCT-15R cells. Furthermore, treatment with b-AP15 reduced the malignant phenotype of CRC cells including cell proliferation and migration, and induced cell death in both 5-FU-sensitive and 5-FU-resistant CRC cells by impairing proteasome function and increasing reactive oxygen species (ROS) production. In addition, b-AP15 inhibited the activation of NF-κB pathway, suppressing cell proliferation. In 5-FU-sensitive and 5-FU-resistant CRC xenografts nude mice, administration of b-AP15 (8 mg·kg-1·d-1, intraperitoneal injection) effectively suppressed the growth of both types of tumors. These results demonstrate that USP14 and UCHL5 play an important role in the development of CRC and resistance to 5-FU. Targeting USP14 and UCHL5 with b-AP15 may represent a promising therapeutic strategy for the treatment of CRC.

Calibrating segmentation networks with margin-based label smoothing.

Despite the undeniable progress in visual recognition tasks fueled by deep neural networks, there exists recent evidence showing that these models are poorly calibrated, resulting in over-confident predictions. The standard practices of minimizing the cross-entropy loss during training promote the predicted softmax probabilities to match the one-hot label assignments. Nevertheless, this yields a pre-softmax activation of the correct class that is significantly larger than the remaining activations, which exacerbates the miscalibration problem. Recent observations from the classification literature suggest that loss functions that embed implicit or explicit maximization of the entropy of predictions yield state-of-the-art calibration performances. Despite these findings, the impact of these losses in the relevant task of calibrating medical image segmentation networks remains unexplored. In this work, we provide a unifying constrained-optimization perspective of current state-of-the-art calibration losses. Specifically, these losses could be viewed as approximations of a linear penalty (or a Lagrangian term) imposing equality constraints on logit distances. This points to an important limitation of such underlying equality constraints, whose ensuing gradients constantly push towards a non-informative solution, which might prevent from reaching the best compromise between the discriminative performance and calibration of the model during gradient-based optimization. Following our observations, we propose a simple and flexible generalization based on inequality constraints, which imposes a controllable margin on logit distances. Comprehensive experiments on a variety of public medical image segmentation benchmarks demonstrate that our method sets novel state-of-the-art results on these tasks in terms of network calibration, whereas the discriminative performance is also improved. The code is available at https://github.com/Bala93/MarginLoss.

Segmentation with mixed supervision: Confidence maximization helps knowledge distillation.

Despite achieving promising results in a breadth of medical image segmentation tasks, deep neural networks (DNNs) require large training datasets with pixel-wise annotations. Obtaining these curated datasets is a cumbersome process which limits the applicability of DNNs in scenarios where annotated images are scarce. Mixed supervision is an appealing alternative for mitigating this obstacle. In this setting, only a small fraction of the data contains complete pixel-wise annotations and other images have a weaker form of supervision, e.g., only a handful of pixels are labeled. In this work, we propose a dual-branch architecture, where the upper branch (teacher) receives strong annotations, while the bottom one (student) is driven by limited supervision and guided by the upper branch. Combined with a standard cross-entropy loss over the labeled pixels, our novel formulation integrates two important terms: (i) a Shannon entropy loss defined over the less-supervised images, which encourages confident student predictions in the bottom branch; and (ii) a Kullback-Leibler (KL) divergence term, which transfers the knowledge (i.e., predictions) of the strongly supervised branch to the less-supervised branch and guides the entropy (student-confidence) term to avoid trivial solutions. We show that the synergy between the entropy and KL divergence yields substantial improvements in performance. We also discuss an interesting link between Shannon-entropy minimization and standard pseudo-mask generation, and argue that the former should be preferred over the latter for leveraging information from unlabeled pixels. We evaluate the effectiveness of the proposed formulation through a series of quantitative and qualitative experiments using two publicly available datasets. Results demonstrate that our method significantly outperforms other strategies for semantic segmentation within a mixed-supervision framework, as well as recent semi-supervised approaches. Moreover, in line with recent observations in classification, we show that the branch trained with reduced supervision and guided by the top branch largely outperforms the latter. Our code is publicly available: https://github.com/by-liu/ConfKD.

Piperlongumine overcomes imatinib resistance by inducing proteasome inhibition in chronic myelogenous leukemia cells.

ETHNOPHARMACOLOGICAL RELEVANCE: Piper longum L., an herbal medicine used in India and other Asian countries, is prescribed routinely for a range of diseases, including tumor. Piperlongumine, a natural product isolated from Piper longum L., has received widespread attention due to its various pharmacological activities, such as anti-inflammatory, antimicrobial, and antitumor effects. AIM OF THE STUDY: Chronic myelogenous leukemia (CML) is a hematopoietic disease caused by Bcr-Abl fusion gene, with an incidence of 15% in adult leukemias. Targeting Bcr-Abl by imatinib provides a successful treatment approach for CML. However, imatinib resistance is an inevitable issue for CML treatment. In particular, T315I mutant is the most stubborn of the Bcr-Abl point mutants associated with imatinib resistance. Therefore, it is urgent to find an alternative approach to conquer imatinib resistance. This study investigated the role of a natural product piperlongumine in overcoming imatinib resistance in CML. MATERIALS AND METHODS: Cell viability and apoptosis were evaluated by MTS assay and Annexin V/propidium iodide counterstaining assay, respectively. Levels of intracellular signaling proteins were assessed by Western blots. Mitochondrial membrane potential was reflected by the fluorescence intensity of rhodamine-123. The function of proteasome was detected using 20S proteasomal activity assay, proteasomal deubiquitinase activity assay, and deubiquitinase active-site-directed labeling. The antitumor effects of piperlongumine were assessed with mice xenografts. RESULTS: We demonstrate that (i) Piperlongumine inhibits proteasome function by targeting 20S proteasomal peptidases and 19S proteasomal deubiquitinases (USP14 and UCHL5) in Bcr-Abl-WT and Bcr-Abl-T315I CML cells; (ii) Piperlongumine inhibits the cell viability of CML cell lines and primary CML cells; (iii) Proteasome inhibition by piperlongumine leads to cell apoptosis and downregulation of Bcr-Abl; (iv) Piperlongumine suppresses the tumor growth of CML xenografts. CONCLUSIONS: These results support that blockade of proteasome activity by piperlongumine provides a new therapeutic strategy for treating imatinib-resistant CML.

Inhibition of proteasomal deubiquitinases USP14 and UCHL5 overcomes tyrosine kinase inhibitor resistance in chronic myeloid leukaemia.

BACKGROUND: Chronic myeloid leukaemia (CML) is a haematological cancer featured by the presence of BCR-ABL fusion protein with abnormal tyrosine kinase activation. Classical tyrosine kinase inhibitor (TKI)-based therapies are available to patients with CML. However, acquired resistance to TKI has been a challenging obstacle, especially stubborn T315I mutation is the most common cause. Therefore, it is especially urgent to find more effective targets to overcome TKI resistance induced by BCR-ABLT315I . Proteasomal deubiquitinases (USP14 and UCHL5) have fundamental roles in the ubiquitin-proteasome system and possess multiple functions during cancer progression. METHODS: The human peripheral blood mononuclear cells were collected to measure the mRNA expression of USP14 and UCHL5, as well as to detect the toxicity effect of b-AP15. We explored the effect of b-AP15 on the activity of proteasomal deubiquitinases. We detected the effects of b-AP15 on BCR-ABLWT and BCR-ABLT315I CML cells in vitro and in the subcutaneous tumour model. We knocked down USP14 and/or UCHL5 by shRNA to explore whether these proteasomal deubiquitinases are required for cell proliferation of CML. RESULTS: In this study, we found that increased expression of the proteasomal deubiquitinase USP14 and UCHL5 in primary cancer cells from CML patients compared to healthy donors. b-AP15, an inhibitor of USP14 and UCHL5, exhibited potent tumour-killing activity in BCR-ABLWT and BCR-ABLT315I CML cell lines, as well as in CML xenografts and primary CML cells. Mechanically, pharmacological or genetic inhibition of USP14 and UCHL5 induced cell apoptosis and decreased the protein level of BCR-ABL in CML cells expressing BCR-ABLWT and BCR-ABLT315I . Moreover, b-AP15 synergistically enhanced the cytotoxic effect caused by TKI imatinib in BCR-ABLWT and BCR-ABLT315I CML cells. CONCLUSION: Collectively, our results demonstrate targeting USP14 and UCHL5 as a potential strategy for combating TKI resistance in CML.

The Mediating Role of Social Support in the Relationship Between Parenting Styles and Adolescent Drug Abuse Identification.

Adolescent drug abuse is a social issue of global concern, causing a serious burden of diseases for individuals, families and society. To design effective prevention and intervention strategies for adolescent drug abusers, the predictive factors associated with drug abuse must be quantified and assessed. This study explores the similarities and differences between the parenting styles of adolescent drug abusers and non-drug abusers and applies a structural equation model to analyze the mechanisms involved between parenting styles, social support and adolescent drug abuse identification. Data were derived from adolescent drug abusers (n = 363) and non-drug abusers (n = 229) between the ages of 18 and 35 in China, and the data were collected and analyzed by SPSS 26 and AMOS 24. The results show that parenting styles significantly predict adolescent drug abuse identification, and different parenting styles have different influencing mechanisms, which further indicates that poor parenting styles are a risk factor for adolescent drug abuse. Additionally, social support plays a mediating role between parenting styles and drug abuse identification (χ2/df = 4.52, CFI = 0.939, TLI = 0.914, RMSEA = 0.077, IFI = 0.939, PCFI = 0.671). The specific pathways involved are as follows: Paternal parenting style → Social support → Drug abuse identification and Maternal parenting style → Social support → Drug abuse identification. However, beyond this, the mediation model of social support shows good adaptability and stability between adolescent drug abusers and non-drug abusers. Since parenting styles and social support are important predictors of adolescent drug abuse, the importance of integrating family-social support antidrug programs into adolescent prevention and intervention strategies should be considered.