Automatic Ploidy Prediction and Quality Assessment of Human Blastocyst Using Time-Lapse Imaging.

Assessing fertilized human embryos is crucial for in vitro-fertilization (IVF), a task being revolutionized by artificial intelligence and deep learning. Existing models used for embryo quality assessment and chromosomal abnormality (ploidy) detection could be significantly improved by effectively utilizing time-lapse imaging to identify critical developmental time points for maximizing prediction accuracy. Addressing this, we developed and compared various embryo ploidy status prediction models across distinct embryo development stages. We present BELA (Blastocyst Evaluation Learning Algorithm), a state-of-the-art ploidy prediction model surpassing previous image- and video-based models, without necessitating subjective input from embryologists. BELA uses multitask learning to predict quality scores that are used downstream to predict ploidy status. By achieving an AUC of 0.76 for discriminating between euploidy and aneuploidy embryos on the Weill Cornell dataset, BELA matches the performance of models trained on embryologists' manual scores. While not a replacement for preimplantation genetic testing for aneuploidy (PGT-A), BELA exemplifies how such models can streamline the embryo evaluation process, reducing time and effort required by embryologists.

A non-invasive artificial intelligence approach for the prediction of human blastocyst ploidy: a retrospective model development and validation study.

BACKGROUND: One challenge in the field of in-vitro fertilisation is the selection of the most viable embryos for transfer. Morphological quality assessment and morphokinetic analysis both have the disadvantage of intra-observer and inter-observer variability. A third method, preimplantation genetic testing for aneuploidy (PGT-A), has limitations too, including its invasiveness and cost. We hypothesised that differences in aneuploid and euploid embryos that allow for model-based classification are reflected in morphology, morphokinetics, and associated clinical information. METHODS: In this retrospective study, we used machine-learning and deep-learning approaches to develop STORK-A, a non-invasive and automated method of embryo evaluation that uses artificial intelligence to predict embryo ploidy status. Our method used a dataset of 10 378 embryos that consisted of static images captured at 110 h after intracytoplasmic sperm injection, morphokinetic parameters, blastocyst morphological assessments, maternal age, and ploidy status. Independent and external datasets, Weill Cornell Medicine EmbryoScope+ (WCM-ES+; Weill Cornell Medicine Center of Reproductive Medicine, NY, USA) and IVI Valencia (IVI Valencia, Health Research Institute la Fe, Valencia, Spain) were used to test the generalisability of STORK-A and were compared measuring accuracy and area under the receiver operating characteristic curve (AUC). FINDINGS: Analysis and model development included the use of 10 378 embryos, all with PGT-A results, from 1385 patients (maternal age range 21-48 years; mean age 36·98 years [SD 4·62]). STORK-A predicted aneuploid versus euploid embryos with an accuracy of 69·3% (95% CI 66·9-71·5; AUC 0·761; positive predictive value [PPV] 76·1%; negative predictive value [NPV] 62·1%) when using images, maternal age, morphokinetics, and blastocyst score. A second classification task trained to predict complex aneuploidy versus euploidy and single aneuploidy produced an accuracy of 74·0% (95% CI 71·7-76·1; AUC 0·760; PPV 54·9%; NPV 87·6%) using an image, maternal age, morphokinetic parameters, and blastocyst grade. A third classification task trained to predict complex aneuploidy versus euploidy had an accuracy of 77·6% (95% CI 75·0-80·0; AUC 0·847; PPV 76·7%; NPV 78·0%). STORK-A reported accuracies of 63·4% (AUC 0·702) on the WCM-ES+ dataset and 65·7% (AUC 0·715) on the IVI Valencia dataset, when using an image, maternal age, and morphokinetic parameters, similar to the STORK-A test dataset accuracy of 67·8% (AUC 0·737), showing generalisability. INTERPRETATION: As a proof of concept, STORK-A shows an ability to predict embryo ploidy in a non-invasive manner and shows future potential as a standardised supplementation to traditional methods of embryo selection and prioritisation for implantation or recommendation for PGT-A. FUNDING: US National Institutes of Health.

Weakly-supervised tumor purity prediction from frozen H&E stained slides.

BACKGROUND: Estimating tumor purity is especially important in the age of precision medicine. Purity estimates have been shown to be critical for correction of tumor sequencing results, and higher purity samples allow for more accurate interpretations from next-generation sequencing results. Molecular-based purity estimates using computational approaches require sequencing of tumors, which is both time-consuming and expensive. METHODS: Here we propose an approach, weakly-supervised purity (wsPurity), which can accurately quantify tumor purity within a digitally captured hematoxylin and eosin (H&E) stained histological slide, using several types of cancer from The Cancer Genome Atlas (TCGA) as a proof-of-concept. FINDINGS: Our model predicts cancer type with high accuracy on unseen cancer slides from TCGA and shows promising generalizability to unseen data from an external cohort (F1-score of 0.83 for prostate adenocarcinoma). In addition we compare performance of our model on tumor purity prediction with a comparable fully-supervised approach on our TCGA held-out cohort and show our model has improved performance, as well as generalizability to unseen frozen slides (0.1543 MAE on an independent test cohort). In addition to tumor purity prediction, our approach identified high resolution tumor regions within a slide, and can also be used to stratify tumors into high and low tumor purity, using different cancer-dependent thresholds. INTERPRETATION: Overall, we demonstrate our deep learning model's different capabilities to analyze tumor H&E sections. We show our model is generalizable to unseen H&E stained slides from data from TCGA as well as data processed at Weill Cornell Medicine. FUNDING: Starr Cancer Consortium Grant (SCC I15-0027) to Iman Hajirasouliha.

The genomic landscape of metastatic clear cell renal cell carcinoma after systemic therapy.

Primary clear cell renal cell carcinoma (ccRCC) has been previously characterized, but the genomic landscape of metastatic ccRCC is largely unexplored. Here, we performed whole exome sequencing (WES) in 68 samples from 44 patients with ccRCC, including 52 samples from a metastatic site. SETD2, PBRM1, APC and VHL were the most frequently mutated genes in the metastatic ccRCC cohort. RBM10 and FBXW7 were also among the 10 most frequently mutated genes in metastatic tissues. Recurrent somatic copy number variations (CNV) were observed at the previously identified regions 3p25, 9p21 and 14q25, but also at 6p21 (CDKN1A) and 13q14 (RB1). No statistically significant differences were found between samples from therapy-naïve and pretreated patients. Clonal evolution analyses with multiple samples from 13 patients suggested that early appearance of CNVs at 3p25, 9p21 and 14q25 may be associated with rapid clinical progression. Overall, the genomic landscapes of primary and metastatic ccRCC seem to share frequent CNVs at 3p25, 9p21 and 14q25. Future work will clarify the implication of RBM10 and FBXW7 mutations and 6p21 and 13q14 CNVs in metastatic ccRCC.

Containers in Bioinformatics: Applications, Practical Considerations, and Best Practices in Molecular Pathology.

Systematic implementation of bioinformatics resources for next generation sequencing (NGS)-based clinical testing is an arduous undertaking. One of the key challenges involves developing an ecosystem of information technology infrastructure for enabling scalable and reproducible bioinformatics services that is resilient and secure for handling genetic and protected health information, often embedded in an existing non-bioinformatics-oriented infrastructure. Container technology provides an ideal and infrastructure-agnostic solution for molecular laboratories developing and using bioinformatics pipelines, whether on-premise or using the cloud. A container is a technology that provides a consistent computational environment and enables reproducibility, scalability, and security when developing NGS bioinformatics analysis pipelines. Containers can increase the bioinformatics team's productivity by automating and simplifying the maintenance of complex bioinformatics resources, as well as facilitate validation, version control, and documentation necessary for clinical laboratory regulatory compliance. Although there is increasing popularity in adopting containers for developing NGS bioinformatics pipelines, there is wide variability and inconsistency in the usage of containers that may result in suboptimal performance and potentially compromise the security and privacy of protected health information. In this article, the authors highlight the current state and provide best or recommended practices for building, using containers in NGS bioinformatics solutions in a clinical setting with focus on scalability, optimization, maintainability, and data security.

Molecular Evaluation of Low-grade Low-stage Endometrial Cancer With and Without Recurrence.

Low-grade, low-stage endometrioid carcinomas (LGLS EC) demonstrate 5-yr survival rates up to 95%. However, a small subset of these tumors recur, and little is known about prognostic markers or established mutation profiles associated with recurrence. The goal of the current study was to identify the molecular profiles of the primary carcinomas and the genomic differences between primary tumors and subsequent recurrences. Four cases of LGLS EC with recurrence and 8 cases without recurrence were evaluated via whole-exome sequencing. Three of the 4 recurrent tumors were evaluated via Oncomine Comprehensive Assay. The resulting molecular profiles of the primary and recurrent tumors were compared. Two of the 3 recurrent cases showed additional mutations in the recurrence. One recurrent tumor included an additional TP53 mutation and the other recurrent tumor showed POLE and DDR2 kinase gene mutation. The POLE mutation occurred outside the exonuclease domain. PIK3CA mutations were detected in 4 of 4 primary LGLS EC with recurrence and in 3 of 8 disease-free cases. LGLS EC with recurrence showed higher MSIsensor scores compared with LGLS without recurrence. The level of copy number gains in LGLS EC with recurrence was larger than LGLS EC without recurrence. This pilot study showed 1 of 3 recurrent cases gained a mutation associated with genetic instability (TP53) and 1 of them also acquired a mutation in the DDR2 kinase, a potential therapeutic target. We also noted a higher level of copy number gains, MSIsensor scores and PIK3CA mutations in the primary tumors that later recurred.

A Deep Learning Approach to Diagnostic Classification of Prostate Cancer Using Pathology-Radiology Fusion.

BACKGROUND: A definitive diagnosis of prostate cancer requires a biopsy to obtain tissue for pathologic analysis, but this is an invasive procedure and is associated with complications. PURPOSE: To develop an artificial intelligence (AI)-based model (named AI-biopsy) for the early diagnosis of prostate cancer using magnetic resonance (MR) images labeled with histopathology information. STUDY TYPE: Retrospective. POPULATION: Magnetic resonance imaging (MRI) data sets from 400 patients with suspected prostate cancer and with histological data (228 acquired in-house and 172 from external publicly available databases). FIELD STRENGTH/SEQUENCE: 1.5 to 3.0 Tesla, T2-weighted image pulse sequences. ASSESSMENT: MR images reviewed and selected by two radiologists (with 6 and 17 years of experience). The patient images were labeled with prostate biopsy including Gleason Score (6 to 10) or Grade Group (1 to 5) and reviewed by one pathologist (with 15 years of experience). Deep learning models were developed to distinguish 1) benign from cancerous tumor and 2) high-risk tumor from low-risk tumor. STATISTICAL TESTS: To evaluate our models, we calculated negative predictive value, positive predictive value, specificity, sensitivity, and accuracy. We also calculated areas under the receiver operating characteristic (ROC) curves (AUCs) and Cohen's kappa. RESULTS: Our computational method (https://github.com/ih-lab/AI-biopsy) achieved AUCs of 0.89 (95% confidence interval [CI]: [0.86-0.92]) and 0.78 (95% CI: [0.74-0.82]) to classify cancer vs. benign and high- vs. low-risk of prostate disease, respectively. DATA CONCLUSION: AI-biopsy provided a data-driven and reproducible way to assess cancer risk from MR images and a personalized strategy to potentially reduce the number of unnecessary biopsies. AI-biopsy highlighted the regions of MR images that contained the predictive features the algorithm used for diagnosis using the class activation map method. It is a fully automatic method with a drag-and-drop web interface (https://ai-biopsy.eipm-research.org) that allows radiologists to review AI-assessed MR images in real time. LEVEL OF EVIDENCE: 1 TECHNICAL EFFICACY STAGE: 2.

Predictive modeling in reproductive medicine: Where will the future of artificial intelligence research take us?

Artificial intelligence (AI) systems have been proposed for reproductive medicine since 1997. Although AI is the main driver of emergent technologies in reproduction, such as robotics, Big Data, and internet of things, it will continue to be the engine for technological innovation for the foreseeable future. What does the future of AI research look like?

Publisher Correction: Building an international consortium for tracking coronavirus health status.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Integrative Molecular Analysis of Patients With Advanced and Metastatic Cancer.

PURPOSE: We developed a precision medicine program for patients with advanced cancer using integrative whole-exome sequencing and transcriptome analysis. PATIENTS AND METHODS: Five hundred fifteen patients with locally advanced/metastatic solid tumors were prospectively enrolled, and paired tumor/normal sequencing was performed. Seven hundred fifty-nine tumors from 515 patients were evaluated. RESULTS: Most frequent tumor types were prostate (19.4%), brain (16.5%), bladder (15.4%), and kidney cancer (9.2%). Most frequently altered genes were TP53 (33%), CDKN2A (11%), APC (10%), KTM2D (8%), PTEN (8%), and BRCA2 (8%). Pathogenic germline alterations were present in 10.7% of patients, most frequently CHEK2 (1.9%), BRCA1 (1.5%), BRCA2 (1.5%), and MSH6 (1.4%). Novel gene fusions were identified, including a RBM47-CDK12 fusion in a metastatic prostate cancer sample. The rate of clinically relevant alterations was 39% by whole-exome sequencing, which was improved by 16% by adding RNA sequencing. In patients with more than one sequenced tumor sample (n = 146), 84.62% of actionable mutations were concordant. CONCLUSION: Integrative analysis may uncover informative alterations for an advanced pan-cancer patient population. These alterations are consistent in spatially and temporally heterogeneous samples.

Deep learning enables robust assessment and selection of human blastocysts after in vitro fertilization.

Visual morphology assessment is routinely used for evaluating of embryo quality and selecting human blastocysts for transfer after in vitro fertilization (IVF). However, the assessment produces different results between embryologists and as a result, the success rate of IVF remains low. To overcome uncertainties in embryo quality, multiple embryos are often implanted resulting in undesired multiple pregnancies and complications. Unlike in other imaging fields, human embryology and IVF have not yet leveraged artificial intelligence (AI) for unbiased, automated embryo assessment. We postulated that an AI approach trained on thousands of embryos can reliably predict embryo quality without human intervention. We implemented an AI approach based on deep neural networks (DNNs) to select highest quality embryos using a large collection of human embryo time-lapse images (about 50,000 images) from a high-volume fertility center in the United States. We developed a framework (STORK) based on Google's Inception model. STORK predicts blastocyst quality with an AUC of >0.98 and generalizes well to images from other clinics outside the US and outperforms individual embryologists. Using clinical data for 2182 embryos, we created a decision tree to integrate embryo quality and patient age to identify scenarios associated with pregnancy likelihood. Our analysis shows that the chance of pregnancy based on individual embryos varies from 13.8% (age ≥41 and poor-quality) to 66.3% (age <37 and good-quality) depending on automated blastocyst quality assessment and patient age. In conclusion, our AI-driven approach provides a reproducible way to assess embryo quality and uncovers new, potentially personalized strategies to select embryos.

Bone biopsy protocol for advanced prostate cancer in the era of precision medicine.

BACKGROUND: Metastatic biopsies are increasingly being performed in patients with advanced prostate cancer to search for actionable targets and/or to identify emerging resistance mechanisms. Due to a predominance of bone metastases and their sclerotic nature, obtaining sufficient tissue for clinical and genomic studies is challenging. METHODS: Patients with prostate cancer bone metastases were enrolled between February 2013 and March 2017 on an institutional review board-approved protocol for prospective image-guided bone biopsy. Bone biopsies and blood clots were collected fresh. Compact bone was subjected to formalin with a decalcifying agent for diagnosis; bone marrow and blood clots were frozen in optimum cutting temperature formulation for next-generation sequencing. Frozen slides were cut from optimum cutting temperature cryomolds and evaluated for tumor histology and purity. Tissue was macrodissected for DNA and RNA extraction, and whole-exome sequencing and RNA sequencing were performed. RESULTS: Seventy bone biopsies from 64 patients were performed. Diagnostic material confirming prostate cancer was successful in 60 of 70 cases (85.7%). The median DNA/RNA yield was 25.5 ng/µL and 16.2 ng/µL, respectively. Whole-exome sequencing was performed successfully in 49 of 60 cases (81.7%), with additional RNA sequencing performed in 20 of 60 cases (33.3%). Recurrent alterations were as expected, including those involving the AR, PTEN, TP53, BRCA2, and SPOP genes. CONCLUSIONS: This prostate cancer bone biopsy protocol ensures a valuable source for high-quality DNA and RNA for tumor sequencing and may be used to detect actionable alterations and resistance mechanisms in patients with bone metastases. Cancer 2018;124:1008-15. © 2017 American Cancer Society.

Next-Generation Rapid Autopsies Enable Tumor Evolution Tracking and Generation of Preclinical Models.

PURPOSE: Patients with cancer who graciously consent for autopsy represent an invaluable resource for the study of cancer biology. To advance the study of tumor evolution, metastases, and resistance to treatment, we developed a next-generation rapid autopsy program integrated within a broader precision medicine clinical trial that interrogates pre- and postmortem tissue samples for patients of all ages and cancer types. MATERIALS AND METHODS: One hundred twenty-three (22%) of 554 patients who consented to the clinical trial also consented for rapid autopsy. This report comprises the first 15 autopsies, including patients with metastatic carcinoma (n = 10), melanoma (n = 1), and glioma (n = 4). Whole-exome sequencing (WES) was performed on frozen autopsy tumor samples from multiple anatomic sites and on non-neoplastic tissue. RNA sequencing (RNA-Seq) was performed on a subset of frozen samples. Tissue was also used for the development of preclinical models, including tumor organoids and patient-derived xenografts. RESULTS: Three hundred forty-six frozen samples were procured in total. WES was performed on 113 samples and RNA-Seq on 72 samples. Successful cell strain, tumor organoid, and/or patient-derived xenograft development was achieved in four samples, including an inoperable pediatric glioma. WES data were used to assess clonal evolution and molecular heterogeneity of tumors in individual patients. Mutational profiles of primary tumors and metastases yielded candidate mediators of metastatic spread and organotropism including CUL9 and PIGM in metastatic ependymoma and ANKRD52 in metastatic melanoma to the lung. RNA-Seq data identified novel gene fusion candidates. CONCLUSION: A next-generation sequencing-based autopsy program in conjunction with a pre-mortem precision medicine pipeline for diverse tumors affords a valuable window into clonal evolution, metastasis, and alterations underlying treatment. Moreover, such an autopsy program yields robust preclinical models of disease.

Whole-Exome Sequencing of Metastatic Cancer and Biomarkers of Treatment Response.

IMPORTANCE: Understanding molecular mechanisms of response and resistance to anticancer therapies requires prospective patient follow-up and clinical and functional validation of both common and low-frequency mutations. We describe a whole-exome sequencing (WES) precision medicine trial focused on patients with advanced cancer. OBJECTIVE: To understand how WES data affect therapeutic decision making in patients with advanced cancer and to identify novel biomarkers of response. DESIGN, SETTING, AND PATIENTS: Patients with metastatic and treatment-resistant cancer were prospectively enrolled at a single academic center for paired metastatic tumor and normal tissue WES during a 19-month period (February 2013 through September 2014). A comprehensive computational pipeline was used to detect point mutations, indels, and copy number alterations. Mutations were categorized as category 1, 2, or 3 on the basis of actionability; clinical reports were generated and discussed in precision tumor board. Patients were observed for 7 to 25 months for correlation of molecular information with clinical response. MAIN OUTCOMES AND MEASURES: Feasibility, use of WES for decision making, and identification of novel biomarkers. RESULTS: A total of 154 tumor-normal pairs from 97 patients with a range of metastatic cancers were sequenced, with a mean coverage of 95X and 16 somatic alterations detected per patient. In total, 16 mutations were category 1 (targeted therapy available), 98 were category 2 (biologically relevant), and 1474 were category 3 (unknown significance). Overall, WES provided informative results in 91 cases (94%), including alterations for which there is an approved drug, there are therapies in clinical or preclinical development, or they are considered drivers and potentially actionable (category 1-2); however, treatment was guided in only 5 patients (5%) on the basis of these recommendations because of access to clinical trials and/or off-label use of drugs. Among unexpected findings, a patient with prostate cancer with exceptional response to treatment was identified who harbored a somatic hemizygous deletion of the DNA repair gene FANCA and putative partial loss of function of the second allele through germline missense variant. Follow-up experiments established that loss of FANCA function was associated with platinum hypersensitivity both in vitro and in patient-derived xenografts, thus providing biologic rationale and functional evidence for his extreme clinical response. CONCLUSIONS AND RELEVANCE: The majority of advanced, treatment-resistant tumors across tumor types harbor biologically informative alterations. The establishment of a clinical trial for WES of metastatic tumors with prospective follow-up of patients can help identify candidate predictive biomarkers of response.