LEAP: Using machine learning to support variant classification in a clinical setting.

Advances in genome sequencing have led to a tremendous increase in the discovery of novel missense variants, but evidence for determining clinical significance can be limited or conflicting. Here, we present Learning from Evidence to Assess Pathogenicity (LEAP), a machine learning model that utilizes a variety of feature categories to classify variants, and achieves high performance in multiple genes and different health conditions. Feature categories include functional predictions, splice predictions, population frequencies, conservation scores, protein domain data, and clinical observation data such as personal and family history and covariant information. L2-regularized logistic regression and random forest classification models were trained on missense variants detected and classified during the course of routine clinical testing at Color Genomics (14,226 variants from 24 cancer-related genes and 5,398 variants from 30 cardiovascular-related genes). Using 10-fold cross-validated predictions, the logistic regression model achieved an area under the receiver operating characteristic curve (AUROC) of 97.8% (cancer) and 98.8% (cardiovascular), while the random forest model achieved 98.3% (cancer) and 98.6% (cardiovascular). We demonstrate generalizability to different genes by validating predictions on genes withheld from training (96.8% AUROC). High accuracy and broad applicability make LEAP effective in the clinical setting as a high-throughput quality control layer.

A machine learning model to determine the accuracy of variant calls in capture-based next generation sequencing.

BACKGROUND: Next generation sequencing (NGS) has become a common technology for clinical genetic tests. The quality of NGS calls varies widely and is influenced by features like reference sequence characteristics, read depth, and mapping accuracy. With recent advances in NGS technology and software tools, the majority of variants called using NGS alone are in fact accurate and reliable. However, a small subset of difficult-to-call variants that still do require orthogonal confirmation exist. For this reason, many clinical laboratories confirm NGS results using orthogonal technologies such as Sanger sequencing. Here, we report the development of a deterministic machine-learning-based model to differentiate between these two types of variant calls: those that do not require confirmation using an orthogonal technology (high confidence), and those that require additional quality testing (low confidence). This approach allows reliable NGS-based calling in a clinical setting by identifying the few important variant calls that require orthogonal confirmation. RESULTS: We developed and tested the model using a set of 7179 variants identified by a targeted NGS panel and re-tested by Sanger sequencing. The model incorporated several signals of sequence characteristics and call quality to determine if a variant was identified at high or low confidence. The model was tuned to eliminate false positives, defined as variants that were called by NGS but not confirmed by Sanger sequencing. The model achieved very high accuracy: 99.4% (95% confidence interval: +/- 0.03%). It categorized 92.2% (6622/7179) of the variants as high confidence, and 100% of these were confirmed to be present by Sanger sequencing. Among the variants that were categorized as low confidence, defined as NGS calls of low quality that are likely to be artifacts, 92.1% (513/557) were found to be not present by Sanger sequencing. CONCLUSIONS: This work shows that NGS data contains sufficient characteristics for a machine-learning-based model to differentiate low from high confidence variants. Additionally, it reveals the importance of incorporating site-specific features as well as variant call features in such a model.

Inherited predisposition to breast and ovarian cancer in non-Jewish populations in Israel.

PURPOSE: The contribution of genetic factors to cancer in non-Jewish populations in Israel is understudied. Yet the early, mostly premenopausal age at breast cancer diagnosis is suggestive of an inherited predisposition. METHODS: High-risk cancer cases of non-Jewish origin who were counseled at the Oncogenetics unit, Sheba Medical Center and the oncology institute at the Ziv medical center from January 1, 2000 to December 31 2016 were eligible. DNA extracted from leukocytes was subjected to massive parallel, next-generation sequencing using the Color Genomics platform. Data were analyzed for pathogenic and likely pathogenic mutations using existing pipelines. RESULTS: Overall, 68 cases, each representing a unique high-risk breast/ovarian family, were genotyped: 32 Druze, 26 Muslim Arabs, and 10 Christian Arabs. Fifty-nine had breast cancer (mean age at diagnosis 42.7 ± 7.6 years), and 9 had ovarian cancer (51.6 ± 9.7 years). Overall three pathogenic mutations one each in BRCA1, PALB2, and BRIP1 genes were detected mostly in Druze families. In addition, 29 variants of unknown significance were also detected, and in 36 cases no sequence variants were noted in any of the genotyped genes. CONCLUSION: The contribution of the known cancer susceptibility genes to the burden of inherited breast/ovarian cancer predisposition in non-Jews in Israel is modest. Other genes or molecular mechanisms account for the familial breast/ovarian cancer clustering in this population.

Multi-gene panel testing for hereditary cancer predisposition in unsolved high-risk breast and ovarian cancer patients.

PURPOSE: Many women with an elevated risk of hereditary breast and ovarian cancer have previously tested negative for pathogenic mutations in BRCA1 and BRCA2. Among them, a subset has hereditary susceptibility to cancer and requires further testing. We sought to identify specific groups who remain at high risk and evaluate whether they should be offered multi-gene panel testing. METHODS: We tested 300 women on a multi-gene panel who were previously enrolled in a long-term study at UCSF. As part of their long-term care, all previously tested negative for mutations in BRCA1 and BRCA2 either by limited or comprehensive sequencing. Additionally, they met one of the following criteria: (i) personal history of bilateral breast cancer, (ii) personal history of breast cancer and a first or second degree relative with ovarian cancer, and (iii) personal history of ovarian, fallopian tube, or peritoneal carcinoma. RESULTS: Across the three groups, 26 women (9%) had a total of 28 pathogenic mutations associated with hereditary cancer susceptibility, and 23 women (8%) had mutations in genes other than BRCA1 and BRCA2. Ashkenazi Jewish and Hispanic women had elevated pathogenic mutation rates. In addition, two women harbored pathogenic mutations in more than one hereditary predisposition gene. CONCLUSIONS: Among women at high risk of breast and ovarian cancer who have previously tested negative for pathogenic BRCA1 and BRCA2 mutations, we identified three groups of women who should be considered for subsequent multi-gene panel testing. The identification of women with multiple pathogenic mutations has important implications for family testing.

Comparison of molecular subtyping with BluePrint, MammaPrint, and TargetPrint to local clinical subtyping in breast cancer patients.

PURPOSE: To compare breast cancer subtyping with the three centrally assessed microarray-based assays BluePrint, MammaPrint, and TargetPrint with locally assessed clinical subtyping using immunohistochemistry (IHC) and fluorescence in situ hybridization (FISH). METHODS: BluePrint, MammaPrint, and TargetPrint were all performed on fresh tumor samples. Microarray analysis was performed at Agendia Laboratories, blinded for clinical and pathological data. IHC/FISH assessments were performed according to local practice at each institution; estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) assessments were performed on 132 samples, and Ki-67 on 79 samples. RESULTS: The concordance between BluePrint and IHC/FISH subtyping was 94 % for the Luminal-type, 95 % for the HER2-type, and 94 % for the Basal-type subgroups. The concordance of BluePrint with subtyping using mRNA single gene readout (TargetPrint) was 96 % for the Luminal-type, 97 % for the HER2-type, and 98 % for the Basal-type subgroups. The concordance for substratification into Luminal A and B using MammaPrint and Ki-67 was 68 %. The concordance between TargetPrint and IHC/FISH was 97 % for ER, 80 % for PR, and 95 % for HER2. CONCLUSIONS: The implementation of multigene assays such as TargetPrint, BluePrint, and MammaPrint may improve the clinical management of breast cancer patients. High discordance between Luminal A and B substratification based on MammaPrint versus locally assessed Ki-67 or grade indicates that chemotherapy decisions should not be based on the basis of Ki-67 readout or tumor grade alone. TargetPrint serves as a second opinion for those local pathology settings where high-quality standardization is harder to maintain.

Transglutaminase 2 is secreted from smooth muscle cells by transamidation-dependent microparticle formation.

Transglutaminase 2 (TG2) is a pleiotropic enzyme involved in both intra- and extracellular processes. In the extracellular matrix, TG2 stabilizes the matrix by both covalent cross-linking and disulfide isomerase activity. These functions become especially apparent during matrix remodeling as seen in wound healing, tumor development and vascular remodeling. However, TG2 lacks the signal sequence for a classical secretory mechanism, and the cellular mechanism of TG2 secretion is currently unknown. We developed a green fluorescent TG2 fusion protein to study the hypothesis that TG2 is secreted via microparticles. Characterization of TG2/eGFP, using HEK/293T cells with a low endogenous TG2 expression, showed that cross-linking activity and fibronectin binding were unaffected. Transfection of TG2/eGFP into smooth muscle cells resulted in the formation of microparticles (MPs) enriched in TG2, as detected both by immunofluorescent microscopy and flow cytometry. The fraction of TG2-positive MPs was significantly lower for cross-linking-deficient mutants of TG2, implicating a functional role for TG2 in the formation of MPs. In conclusion, the current data suggest that TG2 is secreted from the cell via microparticles through a process regulated by TG2 cross-linking.

Intronic Breakpoint Signatures Enhance Detection and Characterization of Clinically Relevant Germline Structural Variants.

The relevance of large copy number variants (CNVs) to hereditary disorders has been long recognized, and population sequencing efforts have chronicled many common structural variants (SVs). However, limited data are available on the clinical contribution of rare germline SVs. Here, a detailed characterization of SVs identified using targeted next-generation sequencing was performed. Across 50 genes associated with hereditary cancer and cardiovascular disorders, a minimum of 828 unique SVs were reported, including 584 fully characterized SVs. Almost 40% of CNVs were <5 kb, with one in three deletions impacting a single exon. Additionally, 36 mid-range deletions/duplications (50 to 250 bp), 21 mobile element insertions, 6 inversions, and 27 complex rearrangements were detected. This data set was used to model SV detection in a bioinformatics pipeline solely relying on read depth, which revealed that genome sequencing (30×) allows detection of 71%, a 500× panel only targeting coding regions 53%, and exome sequencing (100×) <20% of characterized SVs. SVs accounted for 14.1% of all unique pathogenic variants, supporting the importance of SVs in hereditary disorders. Robust SV detection requires an ensemble of variant-calling algorithms that utilize sequencing of intronic regions. These algorithms should use distinct data features representative of each class of mutational mechanism, including recombination between two sequences sharing high similarity, covariants inserted between CNV breakpoints, and complex rearrangements containing inverted sequences.

Small artery remodeling: current concepts and questions.

Blood flow regulation by small arteries and arterioles includes adaptation of both vascular tone and structure. It is becoming clear that tone and remodeling of resistance vessels are highly interrelated. Indeed, concepts pointing to continuous resistance artery adaptation and plasticity are emerging. The purpose of this review is to summarize such concepts and approaches related to vascular organization and remodeling, and to point out the missing links and possible directions for future research. We focus on the individual vessel level. Since several relevant studies are based on isolated vessels, we briefly re-iterate the available isobaric and isometric approaches. We further discuss the major elements of the small artery wall and their relation to the passive and active mechanical properties, as important determinants for vascular remodeling. The cytoskeletal elements and actin re-organization during remodeling are discussed, as well as the re-lengthening of smooth muscle cells during prolonged constriction. We then consider tone as major causal factors in remodeling and discuss the role of vessel wall inflammation. Finally, we illustrate examples of current quantitative, integrative approaches of small artery mechanosensing and adaptation that may lead to a physiomics description of small artery adaptation in health and in diseases such as hypertension.

Controlling technical variation amongst 6693 patient microarrays of the randomized MINDACT trial.

Gene expression data obtained in large studies hold great promises for discovering disease signatures or subtypes through data analysis. It is also prone to technical variation, whose removal is essential to avoid spurious discoveries. Because this variation is not always known and can be confounded with biological signals, its removal is a challenging task. Here we provide a step-wise procedure and comprehensive analysis of the MINDACT microarray dataset. The MINDACT trial enrolled 6693 breast cancer patients and prospectively validated the gene expression signature MammaPrint for outcome prediction. The study also yielded a full-transcriptome microarray for each tumor. We show for the first time in such a large dataset how technical variation can be removed while retaining expected biological signals. Because of its unprecedented size, we hope the resulting adjusted dataset will be an invaluable tool to discover or test gene expression signatures and to advance our understanding of breast cancer.

Identification of a novel GREM1 duplication in a patient with multiple colon polyps.

Hereditary mixed polyposis syndrome (HMPS) is a hereditary syndrome that is characterized by multiple colon polyps of mixed pathologic subtypes and an increased risk for colorectal cancer. A 40 kb duplication in the 5' regulatory region of the GREM1 gene was recently found to be the causal mutation in a subset of Ashkenazi Jewish families with HMPS. Given this discovery, the GREM1 5' regulatory region is now analyzed on many different multi-gene cancer panels, however the data on duplications distinct from the 40 kb duplication remains minimal. Herein we report a novel 24 kb tandem duplication of the 5' regulatory region of GREM1 in a patient without Ashkenazi Jewish heritage, who had a family history that was concerning for Lynch syndrome and satisfied Amsterdam II criteria. This is only the third reported GREM1 duplication separate from the 40 kb Ashkenazi Jewish duplication, and is the only reported duplication to selectively involve exon 1 of GREM1. This finding supports comprehensive testing of the GREM1 regulatory region in families of all ethnicities with multiple colon polyps or colon cancer, and when Lynch syndrome is suspected.

Multi-Gene Panel Testing of 23,179 Individuals for Hereditary Cancer Risk Identifies Pathogenic Variant Carriers Missed by Current Genetic Testing Guidelines.

Recent advancements in next-generation sequencing have greatly expanded the use of multi-gene panel testing for hereditary cancer risk. Although genetic testing helps guide clinical diagnosis and management, testing recommendations are based on personal and family history of cancer and ethnicity, and many carriers are being missed. Herein, we report the results from 23,179 individuals who were referred for 30-gene next-generation sequencing panel testing for hereditary cancer risk, independent of current testing guidelines-38.7% of individuals would not have met National Comprehensive Cancer Network criteria for genetic testing. We identified a total of 2811 pathogenic variants in 2698 individuals for an overall pathogenic frequency of 11.6% (9.1%, excluding common low-penetrance alleles). Among individuals of Ashkenazi Jewish descent, three-quarters of pathogenic variants were outside of the three common BRCA1 and BRCA2 founder alleles. Across all ethnic groups, pathogenic variants in BRCA1 and BRCA2 occurred most frequently, but the contribution of pathogenic variants in other genes on the panel varied. Finally, we found that 21.7% of individuals with pathogenic variants in genes with well-established genetic testing recommendations did not meet corresponding National Comprehensive Cancer Network criteria. Taken together, the results indicate that more individuals are at genetic risk for hereditary cancer than are identified by current testing guidelines and/or use of single-gene or single-site testing.

Decomposition cross-correlation for analysis of collagen matrix deformation by single smooth muscle cells.

Microvascular remodeling is known to depend on cellular interactions with matrix tissue. However, it is difficult to study the role of specific cells or matrix elements in an in vivo setting. The aim of this study is to develop an automated technique that can be employed to obtain and analyze local collagen matrix remodeling by single smooth muscle cells. We combined a motorized microscopic setup and time-lapse video microscopy with a new cross-correlation based image analysis algorithm to enable automated recording of cell-induced matrix reorganization. This method rendered 60-90 single cell studies per experiment, for which collagen deformation over time could be automatically derived. Thus, the current setup offers a tool to systematically study different components active in matrix remodeling.

MammaPrint molecular diagnostics on formalin-fixed, paraffin-embedded tissue.

MammaPrint, a prognostic 70-gene profile for early-stage breast cancer, has been available for fresh tissue. Improvements in RNA processing have enabled microarray diagnostics for formalin-fixed, paraffin-embedded (FFPE) tissue. Here, we describe method optimization, validation, and performance of MammaPrint using analyte from FFPE tissue. Laboratory procedures for enabling the assay to be run on FFPE tissue were determined using 157 samples, and the assay was established using 125 matched FFPE and fresh tissues. Validation of MammaPrint-FFPE, compared with MammaPrint-fresh, was performed on an independent series of matched tissue from five hospitals (n = 211). Reproducibility, repeatability, and precision of the FFPE assay (n = 87) was established for duplicate analysis of the same tumor, interlaboratory performance, 20-day repeat experiments, and repeated analyses over 12 months. FFPE sample processing had a success rate of 97%. The MammaPrint assay using FFPE analyte demonstrated an overall equivalence of 91.5% (95% confidence interval, 86.9% to 94.5%) between the 211 independent matched FFPE and fresh tumor samples. Precision was 97.3%, and repeatability was 97.8%, with highly reproducible results between replicate samples of the same tumor and between two laboratories (concordance, 96%). Thus, with 580 tumor samples, MammaPrint was successfully translated to FFPE tissue. The assay has high precision and reproducibility, and FFPE results are substantially equivalent to results derived from fresh tissue.

Vascular smooth muscle cells remodel collagen matrices by long-distance action and anisotropic interaction.

While matrix remodeling plays a key role in vascular physiology and pathology, the underlying mechanisms have remained incompletely understood. We studied the remodeling of collagen matrices by individual vascular smooth muscle cells (SMCs), clusters and monolayers. In addition, we focused on the contribution of transglutaminase 2 (TG2), which plays an important role in the remodeling of small arteries. Single SMCs displaced fibers in collagen matrices at distances up to at least 300 µm in the course of 8-12 h. This process involved both 'hauling up' of matrix by the cells and local matrix compaction at a distance from the cells, up to 200 µm. This exceeded the distance over which cellular protrusions were active, implicating the involvement of secreted enzymes such as TG2. SMC isolated from TG2 KO mice still showed compaction, with changed dynamics and relaxation. The TG active site inhibitor L682777 blocked local compaction by wild type cells, strongly reducing the displacement of matrix towards the cells. At increasing cell density, cells cooperated to establish compaction. In a ring-shaped collagen matrix, this resulted in preferential displacement in the radial direction, perpendicular to the cellular long axis. This process was unaffected by inhibition of TG2 cross-linking. These results show that SMCs are capable of matrix remodeling by prolonged, gradual compaction along their short axis. This process could add to the 3D organization and remodeling of blood vessels based on the orientation and contraction of SMCs.

The redox state of transglutaminase 2 controls arterial remodeling.

While inward remodeling of small arteries in response to low blood flow, hypertension, and chronic vasoconstriction depends on type 2 transglutaminase (TG2), the mechanisms of action have remained unresolved. We studied the regulation of TG2 activity, its (sub) cellular localization, substrates, and its specific mode of action during small artery inward remodeling. We found that inward remodeling of isolated mouse mesenteric arteries by exogenous TG2 required the presence of a reducing agent. The effect of TG2 depended on its cross-linking activity, as indicated by the lack of effect of mutant TG2. The cell-permeable reducing agent DTT, but not the cell-impermeable reducing agent TCEP, induced translocation of endogenous TG2 and high membrane-bound transglutaminase activity. This coincided with inward remodeling, characterized by a stiffening of the artery. The remodeling could be inhibited by a TG2 inhibitor and by the nitric oxide donor, SNAP. Using a pull-down assay and mass spectrometry, 21 proteins were identified as TG2 cross-linking substrates, including fibronectin, collagen and nidogen. Inward remodeling induced by low blood flow was associated with the upregulation of several anti-oxidant proteins, notably glutathione-S-transferase, and selenoprotein P. In conclusion, these results show that a reduced state induces smooth muscle membrane-bound TG2 activity. Inward remodeling results from the cross-linking of vicinal matrix proteins, causing a stiffening of the arterial wall.

Role of transglutaminases in cuff-induced atherosclerotic lesion formation in femoral arteries of ApoE3 Leiden mice.

UNLABELLED: Transglutaminases play an important role in vascular remodeling, calcification, cell adhesion and endothelial barrier function. In this study we investigate the influence of combined inhibition of both tissue-type transglutaminase (TG2) and the plasma transglutaminase FXIIIA on early lesion development. METHODS: A cuff was placed around the femoral arteries of ApoE3 Leiden mice while fed a Western type diet to induce atherosclerotic lesion development. An osmotic minipump was placed in the intraperitoneal cavity containing an irreversible inhibitor of TG2 and FXIIIA activity ((1,3,4,5-tetramethyl-2-[(2-oxopropyl)thio]imidazolium chloride, Zedira). Atherosclerotic lesion composition was analyzed using immunohistochemistry and RT-PCR. RESULTS: Inhibition of transglutaminases did not influence lesion size or geometric remodeling of the vessels. However, systemic transglutaminase inhibition resulted in 41% less macrophage infiltrate in the media of the vessels. Additional in vitro experiments on HL60 cells confirmed a decreased migratory response during transglutaminase inhibition. CONCLUSION: Inhibition of TG2 and FXIIIA during early development of lesions reduced the macrophage content in the media of atherosclerotic vessels, while not affecting lesion size or geometric remodeling.