Copy number aberrations in circulating tumor DNA enables prognosis prediction and molecular characterization of breast cancer.

BACKGROUND: Low-pass whole-genome sequencing (LP-WGS)-based circulating tumor DNA (ctDNA) analysis is a versatile tool for somatic copy number aberration (CNA) detection, and this study aims to explore its clinical implication in breast cancer. METHODS: We analyzed LP-WGS ctDNA data from 207 metastatic breast cancer (MBC) patients to explore prognostic value of ctDNA CNA burden and validated it in 465 stage II-III triple-negative breast cancer (TNBC) patients who received neoadjuvant chemotherapy in phase III PEARLY trial (NCT02441933). The clinical implication of locus level LP-WGS ctDNA profiling was further evaluated. RESULTS: We found that a high baseline ctDNA CNA burden predicts poor overall survival and progression-free survival of MBC patients. The post hoc analysis of the PEARLY trial showed that a high baseline ctDNA CNA burden predicted poor disease-free survival independent from pathologic complete response (pCR), validating its robust prognostic significance. The 24-month disease-free survival rate was 96.9% and 55.9% in [pCR(+) and low I-score] and [non-pCR and high I-score] patients, respectively. The locus-level ctDNA CNA profile classified MBC patients into 5 molecular clusters and revealed targetable oncogenic CNAs. LP-WGS ctDNA and in vitro analysis identified the BCL6 amplification as a resistance factor for CDK4/6 inhibitors. We estimated ctDNA-based homologous recombination deficiency status of patients by shallowHRD algorithm, which was highest in the TNBC and correlated with platinum-based chemotherapy response. CONCLUSIONS: These results demonstrate LP-WGS ctDNA CNA analysis as an essential tool for prognosis prediction and molecular profiling. Particularly, ctDNA CNA burden can serve as a useful determinant for escalating or de-escalating (neo)adjuvant strategy in TNBC patients.

Integrative modeling of tumor genomes and epigenomes for enhanced cancer diagnosis by cell-free DNA.

Multi-cancer early detection remains a key challenge in cell-free DNA (cfDNA)-based liquid biopsy. Here, we perform cfDNA whole-genome sequencing to generate two test datasets covering 2125 patient samples of 9 cancer types and 1241 normal control samples, and also a reference dataset for background variant filtering based on 20,529 low-depth healthy samples. An external cfDNA dataset consisting of 208 cancer and 214 normal control samples is used for additional evaluation. Accuracy for cancer detection and tissue-of-origin localization is achieved using our algorithm, which incorporates cancer type-specific profiles of mutation distribution and chromatin organization in tumor tissues as model references. Our integrative model detects early-stage cancers, including those of pancreatic origin, with high sensitivity that is comparable to that of late-stage detection. Model interpretation reveals the contribution of cancer type-specific genomic and epigenomic features. Our methodologies may lay the groundwork for accurate cfDNA-based cancer diagnosis, especially at early stages.

Development and performance evaluation of an artificial intelligence algorithm using cell-free DNA fragment distance for non-invasive prenatal testing (aiD-NIPT).

With advances in next-generation sequencing technology, non-invasive prenatal testing (NIPT) has been widely implemented to detect fetal aneuploidies, including trisomy 21, 18, and 13 (T21, T18, and T13). Most NIPT methods use cell-free DNA (cfDNA) fragment count (FC) in maternal blood. In this study, we developed a novel NIPT method using cfDNA fragment distance (FD) and convolutional neural network-based artificial intelligence algorithm (aiD-NIPT). Four types of aiD-NIPT algorithm (mean, median, interquartile range, and its ensemble) were developed using 2,215 samples. In an analysis of 17,678 clinical samples, all algorithms showed >99.40% accuracy for T21/T18/T13, and the ensemble algorithm showed the best performance (sensitivity: 99.07%, positive predictive value (PPV): 88.43%); the FC-based conventional Z-score and normalized chromosomal value showed 98.15% sensitivity, with 40.77% and 36.81% PPV, respectively. In conclusion, FD-based aiD-NIPT was successfully developed, and it showed better performance than FC-based NIPT methods.

Accuracy of machine learning for differentiation between optic neuropathies and pseudopapilledema.

BACKGROUND: This study is to evaluate the accuracy of machine learning for differentiation between optic neuropathies, pseudopapilledema (PPE) and normals. METHODS: Two hundred and ninety-five images of optic neuropathies, 295 images of PPE, and 779 control images were used. Pseudopapilledema was defined as follows: cases with elevated optic nerve head and blurred disc margin, with normal visual acuity (> 0.8 Snellen visual acuity), visual field, color vision, and pupillary reflex. The optic neuropathy group included cases of ischemic optic neuropathy (177), optic neuritis (48), diabetic optic neuropathy (17), papilledema (22), and retinal disorders (31). We compared four machine learning classifiers (our model, GoogleNet Inception v3, 19-layer Very Deep Convolution Network from Visual Geometry group (VGG), and 50-layer Deep Residual Learning (ResNet)). Accuracy and area under receiver operating characteristic curve (AUROC) were analyzed. RESULTS: The accuracy of machine learning classifiers ranged from 95.89 to 98.63% (our model: 95.89%, Inception V3: 96.45%, ResNet: 98.63%, and VGG: 96.80%). A high AUROC score was noted in both ResNet and VGG (0.999). CONCLUSIONS: Machine learning techniques can be combined with fundus photography as an effective approach to distinguish between PPE and elevated optic disc associated with optic neuropathies.

Correction: A deep learning model for the detection of both advanced and early glaucoma using fundus photography.

[This corrects the article DOI: 10.1371/journal.pone.0207982.].

A deep learning model for the detection of both advanced and early glaucoma using fundus photography.

PURPOSE: To build a deep learning model to diagnose glaucoma using fundus photography. DESIGN: Cross sectional case study Subjects, Participants and Controls: A total of 1,542 photos (786 normal controls, 467 advanced glaucoma and 289 early glaucoma patients) were obtained by fundus photography. METHOD: The whole dataset of 1,542 images were split into 754 training, 324 validation and 464 test datasets. These datasets were used to construct simple logistic classification and convolutional neural network using Tensorflow. The same datasets were used to fine tune pre-trained GoogleNet Inception v3 model. RESULTS: The simple logistic classification model showed a training accuracy of 82.9%, validation accuracy of 79.9% and test accuracy of 77.2%. Convolutional neural network achieved accuracy and area under the receiver operating characteristic curve (AUROC) of 92.2% and 0.98 on the training data, 88.6% and 0.95 on the validation data, and 87.9% and 0.94 on the test data. Transfer-learned GoogleNet Inception v3 model achieved accuracy and AUROC of 99.7% and 0.99 on training data, 87.7% and 0.95 on validation data, and 84.5% and 0.93 on test data. CONCLUSION: Both advanced and early glaucoma could be correctly detected via machine learning, using only fundus photographs. Our new model that is trained using convolutional neural network is more efficient for the diagnosis of early glaucoma than previously published models.

IVAG: An Integrative Visualization Application for Various Types of Genomic Data Based on R-Shiny and the Docker Platform.

Next-generation sequencing (NGS) technology has become a trend in the genomics research area. There are many software programs and automated pipelines to analyze NGS data, which can ease the pain for traditional scientists who are not familiar with computer programming. However, downstream analyses, such as finding differentially expressed genes or visualizing linkage disequilibrium maps and genome-wide association study (GWAS) data, still remain a challenge. Here, we introduce a dockerized web application written in R using the Shiny platform to visualize pre-analyzed RNA sequencing and GWAS data. In addition, we have integrated a genome browser based on the JBrowse platform and an automated intermediate parsing process required for custom track construction, so that users can easily build and navigate their personal genome tracks with in-house datasets. This application will help scientists perform series of downstream analyses and obtain a more integrative understanding about various types of genomic data by interactively visualizing them with customizable options.