VILNIUS UNIVERSITY Digital immunohistochemistry platform for the staining variation monitoring based on integration of image and statistical analyses with laboratory information system Arvydas Laurinavicius 12th European Congress on Digital Pathology Paris, France, 18-21 June 2014
Does digital IHC need digital tissue controls? Conventional IHC staining quality is monitored by semiquantitative visual evaluation of tissue (multi)controls Image analysis may require more sensitive quality control Intra- and interlaboratory IHC staining variation may be a major obstacle for DIA adoption We have previously shown intra-laboratory HER2 IHC variation by digital analysis means
HER2 average membrane staining intensity variation 0 1+ 0 1+ 2+ 2+ Batch/Serial section # 3+ 3+ Laurinavicius, Besusparis, et al Diagnostic Pathology 2013.
Integrated IHC staining monitor by and for digital analysis Explore Ki67 IHC staining variation in our laboratory routine, set up LIS integration of the IHC tissue multi-controls, monitored by image analysis, and supported by automated statistical analysis feedback
Consecutive sections (n=69) of one TMA block containing 10 cores of breast cancer tissue used as tissue controls in routine Ki67 IHC testing Ventana slide label BarcodeID was sent to the LIS to register the serial section sequence, then used for statistical analyses Slides stained and scanned (Aperio ScanScope XT), IA performed by the Aperio/Leica Colocalization and Genie Classifier/Nuclear algorithms SQL-based SpectrumPlus/LIS integration ensured automated statistical analysis of the IA data by the SAS Enterprise Guide project Factor analysis and plot visualizations to explore slide-to-slide variation of the Ki67 IHC staining results in the control tissue
Workflow TMA block with 10 control tissue cores Serial sections stored at +4C Used on demand for routine Ki67 TMALAB design applied for the tissue controls Slides stained and scanned Ventana BarcodeID stored in LIS when slide label printed Colocalization Genie/Nuclear performed on the TMA and WSI SAS project rerun from the Spectrum/LIS SQL query Statistical feedback generated
Intraslide Intertissue Intraslide Intertissue
4 8 3 7 2 6 10 1 5 9 Variation of individual 10 spots
Slide-to-slide variation of the TMA multicontrols Total Nuclei Total Stained Area (mm2) Positive Nuclei Brown Intensity The sequence of Ventana slide Barcode on the x axis to represent consecutive serial sections of TMA blocks of the 10 multi-tissue control cores (labelled as SampleID),
Slide-to-slide variation of the TMA multicontrols Percent Positive Nuclei Blue Intensity Positive Density (Pos Nuclei/mm 2 ) Brown/Blue Intensity ratio The sequence of Ventana slide Barcode on the x axis to represent consecutive serial sections of TMA blocks of the 10 multi-tissue control cores (labelled as SampleID),
Factor pattern of Brown and Blue intensity indicators from Colocalization Analysis in 10 individual TMA cores Total Intensity Both Brown and Blue lighter Relatively deeper Blue Brown/Blue balance
4 8 3 2 7 6 10 e.g., Sum (Area) Sum (Total Nuclei) Median (Intensity) etc. 1 5 9 Variation of aggregated data from the 10 TMA spot images super-multi-tissue control
Factor pattern of Colocalization and Genie/Nuclear analysis outputs aggregated from 10 TMA cores IHC positivity High Ki67 LI, Dark Brown More tissue, Dark Blue Amount of Tissue Detected with Darker Blue
Relevant inverse covariation of selected image analysis variables Median Blue Intensity (Colocalization) Area of Analysis (Genie) Aggregated data from 10 TMA cores plotted against the sequence of Ventana Barcode on the x axis
The solution enabled continuous monitor of IHC multi-tissue controls by the means of IA, followed by automated statistical analysis, integrated into the laboratory workflow. Even in consecutive serial tissue sections, tissue-related factors affected the IHC IA results; meanwhile, less intense blue counterstain was associated with less amount of tissue, detected by the IA tools.
Remarks Scanning variation was not addressed in our experiment, same scanner used. The IA issues with hematoxylin counterstain in IHC, have been highlighted, alternative counterstaining and IA techniques have been proposed Brey at al J Histochem Cytochem 2003, Pham et al Diagn Pathol 2007, Stefanovic et albiotechnic & histochemistry 2013, Zehntner et al J Histochem Cytochem 2008, Bernardo et al Microsc Microanal 2009] Another approach - measuring signal-to-noise ratio of the images to evaluate quality before IA (Ali et al Br J Cancer 2013), however, adjustment of the images and/or analyses may require effort. Our data simulate reproducibility of the Ki67 index in the consecutive sections of one 1 mm diameter core: variation of Ki67% (the IA result) was satisfactory (standard deviation in all 10 cores ranged from 3 to 8, relative error within the range of 0.07 to 0.39) the variation of cell numbers detected (the process) was higher. Inter-laboratory IHC staining variation is likely to be more significant: international Ki67 reproducibility study (Polley et al J Natl Cancer Inst 2013) revealed unsatisfactory results of visual estimation which was even worse when the slides were stained locally. Further optimization and standardization of IHC procedures, especially, when applying IA tools with unknown sensitivity to the staining/scanning variation. Ideally, IA tools should be robust and resistant to the IHC staining and scanning variations.
Aida Laurinavičienė and Paulette Herlin Daiva Lesciute- Krilaviciene Yasir Iqbal Benoit Plancoulaine Darius Raudeliunas Raimundas Meškauskas Indra Baltrušaitytė, Justinas Besusparis A COMPREHENSIVE BIOMARKER INTRA- TUMOUR HETEROGENEITY EVALUATION BY DIGITAL IMMUNOHISTOCHEMISTRY IMAGE ANALYSIS This research is funded by European Social Fund under the Global Grant measure.
www.vpc.lt