2015 Laboratory Investigation  
Vancomycin-resistant Enterococcus faecium (VREf) belonging to clonal cluster 17 (CC17) has been emerging globally since the 1990's and is now among the predominant group of enterococci causing nosocomial infections of the bloodstream, urinary tract, skin and soft-tissues. The aim of this study was to assess the clonality of temporally spaced VREf clinical isolates using whole-genome sequencing analysis. Design: Thirty-four and 113 VREf clinical isolates recovered from patients in a tertiary
more » ... cal center in suburban New York City in 1995 and 2013, respectively, were selected and analyzed. Whole-genome sequencing was performed on the Illumina MiSeq™ or HiSeq™ System by using paired-end methods. Multilocus sequence typing (MLST) and single nucleotide variations (SNVs) data were derived from the genome sequences of each isolate. The genetic relatedness of different enterococcal isolates and sequence types were explored using the goeBURST program. Results: The predominant strain type for VREf isolates from 1995 was ST17 (79.4%), followed by ST18 (8.8%) and ST323 (8.8%). By contrast, a newly described clone, ST736, accounted for 48.7% of VREf isolates in 2013, followed by ST18 (24.7%) and ST412 (21.2%). Population genetic analysis established that ST736 is a novel clone within CC17 with differences in two alleles from its prototype ST17. Conclusions: Whole-genome sequencing analysis of VREf clinical isolates demonstrated a dramatic change in dominant strain types and the emergence of a novel clone ST736 in our patient population over the past 18 years. Background: The development of next generation sequencing (NGS) and associated target sequence enrichment technologies has enabled time and cost effective detection of clinically relevant molecular alterations in hundreds of genes using a single clinical assay, thus paving the path for personalized medicine. Although data generation is no longer the major hurdle, accurate and time-effective bioinformatic and statistical analysis, and biologically meaningful interpretation of the data remain challenging. Design: We created CGA (Clinical Genome Analytics), an automated data analysis pipeline that is implemented using Civet, an in-house engineered XML-based framework for building analytical pipelines for efficient multi-step data processing. Results: CGA includes capabilities for automatic scheduling of jobs, monitoring analysis runs, and automatically saving information about each run (including command line options, versions and standard output/error for each tool). It puts together 17 different tools, from quality control and alignment through local realignment around indels and base quality recalibration to calling SNPs, indels and CNVs, and finally assigning genomic annotations to these variants. CGA takes an average of 6 hr to process samples containing approximately 25M paired-end reads, and ensures high (> 99.4%) sensitivity and specificity for the detection of >= 5% frequency SNPs and indels (<= 50bp) and of copy number changes >= 6. Conclusions: CGA is an automated analysis pipeline that ensures accurate and sensitive detection and clinical annotation of mutations. It is currently being used as part of an NGS-based molecular diagnostic assay that detects actionable mutations in solid tumors in a CLIA-certified laboratory. It provides not only an accurate analysis of samples in a clinically acceptable short turn-around time, but also the ability to document quality metrics and information on all of the tools, their versions and options, and all reference genomes and databases used, which is key to ensure reproducibility and traceability in a clinical set-up. Background: SNP microarrays are being widely used as a tool in a routine oncology diagnostic workflow. The main advantage is its ability to accurately identify the composition of derivative chromosomes & marker chromosomes commonly observed in clonal populations. One of the biggest challenges of interpreting oncology arrays is the lack of bioinformatics tools facilitating the analysis and interpretation of microarray data. Another major challenge is to rule out the common polymorphic genomic variants stored in databases such as DGV, as well as having the ability to tease out potentially relevant copy number variable (CNV) and Loss-Of-Heterozygosity (LOH) regions possibly contributing to the pathogenesis of neoplastic disorder. The aim of this study was to access and configure the utility of an automated bioinformatic pipeline in the analysis and interpretation of SNP Microarrays in neoplasia. Design: For constitution SNP array's our lab has been using various automated bioinformatics pipelines for data analysis and reporting. For oncology SNP array data analysis and interpretation we collaborated with Cartagenia Inc. Aim is to create a different variant assessment and filtration strategies that can be established and automated for the analysis of SNP microarrays in various leukemias and the utility of the Bench platform in the identification of pathogenic changes. We selected array SNP array data from 42 cases of cytogenetically normal cases of CMML, Myelofibrosis and ALL and subjected to the customized bench. Results: Our studies have allowed us to establish specific filters for the determination of potentially significant copy number changes (like mosaic deletions and mosaic duplications). SNP microarrays also have the added advantage of identifying regions of LOH which play a very important role in the 2 hit hypothesis in somatic cancers. Our filtration allowed removing germline mutations, stored in databases such as DGV, as well as alignment of genomic variants with control populations or healthy tissue samples. We have also been able to establish an automated filtration and reporting pipeline that allows the identification and reporting of regions of LOH harboring important cancer genes such as CDKN2A and JAK2. Conclusions: We propose that an automated bioinformatic pipeline for the analysis and interpretation of neoplastic disorders will benefit the labs and oncologists ordering a high-resolution chromosomal microarray in a diagnostic setting. Diagnostic and Educational Uses of Google Glass in Anatomic Pathology Natalie Ciomek, Hongfa Zhu, Carlos Cordon-Cardo. The Icahn School of Medicine at Mount Sinai, New York, NY. Background: Google Glass is a wearable intelligent device permitting the capture and communication of photography and video via its hands-free, voice recognition capabilities. Multiple institutions are investigating the use of Glass technology in health care, most notably within surgical and internal medicine specialties. Glass utilization within pathology has been limited to very few reports. The first study investigated livestreaming of routine specimen dissection; another study described the acquisition and quality of still images in forensic autopsy findings. To date, no studies have applied the use of Glass-acquired data, specifically videos, for subsequent diagnostic and educational purposes in anatomic pathology. Design: Google Glass (Explorer edition) photography and videography was implemented as an adjunct to routine gross specimen documentation by hand-held digital photography during intraoperative consultations and/or routine permanent processing. Specimens included pulmonary, gynecologic, and gastrointestinal resections. The edited dissection videos were less than 40 seconds in length. Data was obtained, stored, and accessed in the same secure and HIPAA-compliant methodology currently in place for digital gross photography. Specimen photography and videography were evaluated in several clinical and educational capacities, including but not limited to consultations among residents and attendings, gross specimen review at the time of microscopic examination, pathology resident education, and inter-and intradepartmental conferences. Results: Glass technology obtained images and videos with quality comparable to those by digital photography but without disruption or increase in processing time to pre-existing grossing procedures. The use of Glass was described as more convenient and efficient than conventional photography. Dissection videos of complex resection specimens, beginning from their intact and fresh states to their final sectioning, were reported to be of greater diagnostic and educational utility than both photography and gross specimen review of the fixed and sectioned tissues. Conclusions: Glass has demonstrated clinical and educational value within anatomic pathology. The analysis of surgical specimens in pathology almost always requires specimens to become disrupted and altered in fixative following dissection. Maintaining quality photographic and video records of the intact and fresh specimens at the time of initial dissection is important for diagnoses and reporting, medical education, and potential error reduction. FISH Digital Image Capture and Automated Segmentation Improves Laboratory Workflow Efficiency Background: Fluorescence in situ hybridization (FISH) is a laboratory testing modality where fluorescent probes highlight specific chromosomal sequences which are then localized by fluorescence microscopy. Limitations to FISH interpretation include identification of individual cells showing fluorescent signal without cellular overlap. These limitations are exacerbated by manual counting and interpretation, leaving manual FISH analysis subjective and time consuming. As FISH testing is increasingly incorporated in diagnostic pathology, turnaround time with accurate reproducibility of results is of utmost importance to workflow and patient care. We examined if incorporation of automated segmentation protocols on digitally captured images improves cell detection and laboratory workflow efficiency. Design: FISH cases (hematologic and non-hematologic) were analyzed (5 Her2/neu, 3 FKHR, and 8 MYC) using automated FISH analysis systems (GenASIs, Applied Spectral Imaging, Carlsbad, CA). H&E slides were reviewed and marked by a pathologist. Following image capture by GenASIs, manual and automated segmentation protocols were performed on a minimum of 60 cells (hematopoietic cases) and 120 cells (non-hematopoietic cases) from four to five digital image frames. Manual segmentation involved a technologist circling appropriate cells for analysis with the GenASIs mouse. Automated segmentation by the GenASIs Omni software, was performed on identical frames with manual deselection of inappropriate cells. Both methods were compared for efficiency and accuracy. 396A ANNUAL MEETING ABSTRACTS Results: Automated segmentation by GenASIs resulted in an overall signifi cant technician time reduction (range 13-19 min, mean 10.7 min) in comparison to manual segmentation, (range 19-85 min, mean 37.2 min) (p= <0.0001). Automated segmentation maintained signifi cant time reduction over manual segmentation in each test: Her2/neu (p=<0.0001), FKHR (p=0.0004), and MYC (p=0.06). There was no signifi cant difference in the classifi cation time between methods (p=0.46). There was high concordance on both signal scoring pattern (P=<0.0001, r 2 =0.98) as well as 100% concordance on case classifi cation/result interpretation in both methods with gold standard microscopic evaluation. Conclusions: Digital capture and analysis FISH systems perform automated cell detection much more rapidly than manual segmentation yet maintain high levels of accuracy, thus improving laboratory workfl ow. Adoption of digital imaging in the clinical laboratory also provides advantages including long term archival and quality assurance measures. Background: Microscopic examination of trephine bone marrow biopsies is a routine part of the hematopathology examination and accurate estimation of the marrow cellularity (ratio of hematopoietic to fat and stromal cells) is an important diagnostic determinant. Manual (visual) assessment of marrow cellularity lacks standardized criteria, which introduces inter and intraobserver variability and makes comparison of cellularity in sequential samples diffi cult. Reliable cellularity scores are diagnostically essential and clinically important for tracking therapy-related changes and disease progression. The aim of our study is to evaluate feasibility of computational image analysis in an attempt to standardize the cellularity assessment. Design: 104 consecutive bone marrow biopsies were digitally scanned at 20x magnifi cation using an Aperio CS whole slide scanner. 89 cases had numeric manual cellularity scores and qualifi ed for comparison with digital analysis. Images were annotated manually to isolate the region of interest. An in house developed algorithm was used to process images and determine a cellularity percentage. Manual scores were extracted from the diagnostic pathology reports and compared to the automated image analysis scores. Results: Our image analysis technique has a good correlation with manually reported cellularity. Linear regression analysis showed an R-squared value of 0.8337.
doi:10.1038/labinvest.2015.17 fatcat:74eygyy7o5gkrhizoootoyfol4