Progress in the field of noninvasive diagnostics for colorectal cancer: a systematic review for the accuracy of blood-based biomarkers for detection of advanced pre-cancerous lesions.

BACKGROUND: Early detection of pre-cancerous adenomas through screening can reduce colorectal cancer (CRC) incidence. Fecal immunochemical tests are commonly used, but have limited sensitivity for pre-cancerous lesions. Blood-based screening may improve test sensitivity. This systematic review and meta-analysis was conducted to evaluate the accuracy of blood-based biomarkers for detection of advanced pre-cancerous lesions. RESEARCH DESIGN AND METHODS: We present the accuracy of blood-based biomarkers for the detection of advanced pre-cancerous lesions. EMBASE, Web of Science and PubMed databases were searched, with study populations limited to adults diagnosed with advanced pre-cancerous lesions at colonoscopy, who had a blood-based biomarker test analyzed with reports of sensitivity and specificity. RESULTS: 69 studies were identified, which assessed 133 unique biomarkers sets. The best performing test was a panel of 6 miRNAs, with a sensitivity of 95% and specificity of 90% for advanced pre-cancerous lesions. Only 6 biomarkers demonstrated sensitivity ≥ 50% and specificity ≥ 90% for the detection of advanced pre-cancerous lesions. CONCLUSION: Many different blood-based biomarkers have been assessed for detection of advanced pre-cancerous lesions, but few have progressed beyond the discovery stage. While some biomarkers have reported high sensitivity and specificity, larger prospective studies in unbiased intended-use screening populations are required for validation.

Accuracy of blood-based biomarkers for screening precancerous colorectal lesions: a protocol for systematic review and meta-analysis.

INTRODUCTION: Colorectal cancer (CRC) is the third most diagnosed cancer and the second most common cause of cancer mortality worldwide. Most CRCs develop through either the adenoma-to-carcinoma or the serrated pathways, and, therefore, detection and removal of these precursor lesions can prevent the development of cancer. Current screening programmes can aid in the detection of CRC and adenomas; however, participation rates are suboptimal. Blood-based biomarkers may help to address these low participation rates in screening programmes. Although blood-based biomarker tests show promise for cancer detection, limited attention has been placed on the sensitivity and specificity for detection of the precursor lesions. The aim of this research is to conduct a systematic review and meta-analysis to evaluate the accuracy of blood-based biomarker tests in detecting advanced precancerous lesions. METHODS AND ANALYSIS: This protocol was informed by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses-Protocols (PRISMA-P) and results will be reported in line with the PRISMA guidelines. Literature searches will be conducted on PubMed, Embase and Web of Science. Two reviewers will conduct the searches, and independently screen them, according to title and abstract and then the full-text versions of those selected articles as well as the risk of bias via the Quality Assessment of Diagnostic Accuracy Studies version 2 (QUADAS-2) tool. The Grading of Recommendations Assessment, Development and Evaluation guidelines will be used to validate the certainty of evidence for recommendations based on the risk of bias findings. Meta-analysis will be conducted where appropriate on groups of studies with low heterogeneity. ETHICS AND DISSEMINATION: No patient data will be included in our review and, therefore, ethics approval is not required. It is anticipated that the review will identify the most promising candidate biomarkers for clinical translation in the screening of advanced precancerous lesions. The results will be published in a peer-reviewed journal. PROSPERO REGISTRATION NUMBER: CRD42021285173.

Neural Shape Parsers for Constructive Solid Geometry.

Constructive solid geometry (CSG) is a geometric modeling technique that defines complex shapes by recursively applying boolean operations on primitives such as spheres and cylinders. We present CSGNet, a deep network architecture that takes as input a 2D or 3D shape and outputs a CSG program that models it. Parsing shapes into CSG programs is desirable as it yields a compact and interpretable generative model. However, the task is challenging since the space of primitives and their combinations can be prohibitively large. CSGNet uses a convolutional encoder and recurrent decoder based on deep networks to map shapes to modeling instructions in a feed-forward manner and is significantly faster than bottom-up approaches. We investigate two architectures for this task-a vanilla encoder (CNN) - decoder (RNN) and another architecture that augments the encoder with an explicit memory module based on the program execution stack. The stack augmentation improves the reconstruction quality of the generated shape and learning efficiency. Our approach is also more effective as a shape primitive detector compared to a state-of-the-art object detector. Finally, we demonstrate CSGNet can be trained on novel datasets without program annotations through policy gradient techniques.