2008 Summer Meeting - Leeds - The Pathological Society of Great ...

P175Systematic Random Sampling with Virtual Slides: A NewSoftware Tool for Tissue ResearchDTreanor 1 ,MDattani 1 ,PQuirke 1 , H Grabsch 11 Pathology and Tumour Biology, Leeds Institute of Molecular Medicine,University of LeedsSystematic random sampling is a powerful tool in tissue based research. Itallows accurate quantification of tissue elements on pathology slides whileminimising the number of samples needed to measure – for example the ratio ofcancer to stroma, or the distribution of blood vessels.Historically optical graticules with conventional light microscopes have beenused for systematic randomised sampling, but this method has significantdrawbacks in terms of flexibility and user acceptability. Commercial software isavailable which adds grids to captured micrographs, but this requires individualsnapshots of tissue to be taken manually. Using virtual slides for this task issimpler, less labour-intensive, and allows recording of results directly intocomputer software.We describe a system using custom made software developed locally whichallows users to select arbitrary areas of any virtual slide and add any number ofsampling points using a systematic grid. The system is web based, allowingresearchers to apply a systematic random grid to any sample from their personalcomputer. Once the sampling points are generated they can be viewed locallyusing virtual slide viewing software and the scores entered rapidly andefficiently by the user for export and further analysis.P177The Digital Pathology Archive: the UMCU initiativePJ Van Diest 11 University Medical Centre Utrecht, Department of Pathology, TheNetherlandsBackground:Our Pathology Department produces yearly about 100.000 slides. After initialdiagnosis, slides often have to be retrieved. Which takes a lot of time, and slidequality deteriorates in time. We have therefore set up infrastructure to create afully digital archive.Methods: Two Aperio scanners processes batches of each 120 slides, recognizethe tissue on low resolution, auto focus on a defined number of spots, and thenscan the tissue in true colourat x20. Logistics to maximize scanning capacityhave been worked out. File size is about 500 megabytes on average. The imagesare stored on a Sun HSM system consisting of a storage server, 6 TB fast harddisks and a tape robot of 120 TB. Access time for the images varies from a fewseconds for the hard disk to several minutes for the tape storage.Conclusions: It is technically and practically feasible to digitize all slidesproduced by an average size pathology lab, and store them on a SAN. Doingclinicopathological conferences digitally will gradually be implemented whichwill save a lot of preparation time and increase flexibility in location of theconferences. Further, consultations will be more efficient. Slide images will beintegrated within the Electronic Patient File, and practical slide sessions forteaching purposes will be done digitally using the Slidepath application.P176Tracking with virtual slides: a tool to study diagnostic error inpathologyDTreanor 1 , D Magee 2 , A Bulpitt 2 ,CLim 3 ,PQuirke 11 Pathology and Tumour Biology, Leeds Institute of Molecular Medicine,University of Leeds, 2 School of Computing, University of Leeds,3 Department of Gastroenterology, Goodhope HospitalBACKGROUND: Diagnostic error in pathology is a significant problem.Studying the reasons for error is difficult because of a lack of data on thediagnostic process - virtual slides are a useful tool to study error in pathology asthey allow unsupervised study of diagnosis.METHODS: Custom software was developed to produce visualisations of thediagnostic track followed by pathologists as they viewed virtual slides. Theseshowed the diagnostic path in 3 dimensions, areas studied for >1000ms, andsubject’s comments about the areas viewed. The system was used to study 2trainee and 2 expert pathologists diagnosing 60 oesophageal biopsies showingBarrett’s oesophagus. Comparisons of the diagnostic tracks between expertsand trainees were studied to classify the reason for errors.RESULTS: 46 cases had an expert consensus diagnosis. The two trainees madeerrors in 21 cases (46%) and 15 cases (33%) respectively, of which 11 and 9were clinically significant errors. Errors were made across the whole spectrumfrom normal biopsies to intramucosal carcinoma, though there was a tendencyfor serious undercalls to be more common than serious overcalls. Detailedexamination of the tracks showed that in all 36 errors there was a error ininterpretation of information; in 3 errors there was an additional component offailure to identify diagnostic features.CONCLUSIONS: Tracking with virtual slides and the resulting visualisationsare a useful tool in studying diagnosis. The tool has the potential for use intraining and assessment in order to improve the quality of diagnosis inpathology.P178Non-Side Population Cell Cycle Suppression by Hoechst 33342Occurs in the Absence of Significant Cell Death in thePancreatic Cancer Line Capan-2N Guppy 1 , WR Otto 2 , MR Alison 11 Institute for Cell and Molecular Science, Barts and The London School ofMedicine and Dentistry, 2 Histopathology Unit, London Research Institute,Cancer Research UKIsolation of a “side population” (SP) via Hoechst (Ho) 33342 efflux assay is apopular tool for adult stem cell identification. High activity of ABCtransportersconfers the enhanced dye efflux capacity of SP cells and cytotoxicdrug resistance. However, Ho33342 is known to be capable of inducingcytotoxic and cytostatic effects, raising the possibility that enhancedclonogenicity and proliferative capacities exhibited by SP cells versus non-SPcells isolated in this way may be an artefact of Ho33342 overload in the non-SPpopulation.SP and non-SP cells were isolated from two human pancreaticadenocarcinoma lines (Panc-1 and Capan-2) via standard Ho33342 effluxassay. Viability, colony forming ability (CFA) and proliferation of SP, non-SPand unsorted populations exposed to Ho33342 ± reserpine (an inhibitor ofABC-transporter activity), reserpine or vehicle alone were examined.Capan-2 SP cells showed elevated CFA versus unsorted Ho-exposedcontrols. However, in unsorted samples, Ho33342 exposure producedsignificant decreases in CFA and population doubling rate, exacerbated by coincubationwith reserpine. Conversely, Panc-1 cells showed no significantdifferences in CFA or proliferation rate between treatments or sub-populations.These differing responses to Ho33342 exposure could not be explained bydifferences in cell cycle position, DNA content or apoptosis levels.Furthermore, the percentage of propidium iodide-negative viable cells did notvary between lines or treatments.Therefore, elevated CFA of Capan-2 SP cells may be an artefact of dyeinducedsuppression of proliferation of the non-SP fraction.76 Summer Meeting (194 th ) 1–4 July 2008 Scientific Programme