P008 (0053) WHOLE-SLIDE-IMAGE ANALYSIS OF THE TUMORMICRONENVIRONMENT IN CLASSICAL HODGKIN LYMPHOMA

doi: 10.1097/01.HS9.0000547860.04632.43
Biology and Microenvironment
Free

Luise Pieper1, Sarah Reinke1, Ron Jachimowicz2,3, Annette Plütschow3, Andreas Engert2,3, Wolfram Klapper1

1Institut für Pathologie, Sektion Hämatopathologie, Universitätsklinikum Schleswig-Holstein, Campus Kiel,2 Innere Medizin I, Universitätsklinikum Köln,3The German Hodgkin Study Group (GHSG)

The composition of the tumormicroenvironment (TME) in classical Hodgkin Lymphoma (cHL) is highly heterogeneous between the histological subtypes but also within each subtype (intertumoral) and even within the involved tissue (intratumoral). Gene expression profiling identified an association of the composition of the TME with outcome. However, this technology cannot easily be applied in clinical practice. Classical immunohistochemistry (IHC) based approaches to quantify TMA components e.g. by manual counting in high-power-fields, are restricted to a small tumor region, suffer from high interobserver-variability and may be biased by intratumoral heterogeneity. To overcome these limitations, we used digital whole-slide-image (WSI) analysis to quantify Hodgkin-Reed-Sternberg-Cells (HRSC) and T-Cells. Diagnostic lymph-node samples, formalin-fixed-paraffin-embedded (FFPE), of 390 patients with advanced-stage cHL, derived from two randomized trials of the GHSG (HD12, HD15), were analyzed using the image analysis program TissueStudio 64 (Definiens AG, Munich, Germany). FFPE samples were cut and -automatically immunohistochemically stained with CD3 (T-cells) and CD30 (HRSC). T-Cells were quantified as respective percentage of all cells in the FFPE-slide, for HRSC the IHC positive area was detected as an indirect cell content marker. Data were correlated with B-cell content assessed by the same method (Jachimowicz et al.).

To test interobserver variability of WSI two independent observers processed image analysis on scanned slides (n = 20) and resulting agreement between both was high (r = 0.93, p < 0.0001).

The tissue of cHL showed a mean T-cell content of 53,53%, mean area covered by HRSC was 5,1%. As expected, we found that T-cell-content was lowest in lymphocyte-depleted cHL and highest in lymphocyte-rich cHL, HRSC area however was highest in lymphocyte-depleted cHL. Patients with a low B-cell content (<21%, Jachimowicz et al.) surprisingly presented also with a low T-cell content (t-test, p < 0.001) and a higher content of HRSC (t-test, p < 0.0001). Analysis of T-cell and HRSC content in respect to clinical data is currently ongoing.

In conclusion, we could show that digital WSI analysis allows objective, accurate and practical quantification of TME in cHL. Insights in the composition of the TME made by using this technique can help to understand more about the biology of TME in cHL and reveal a cell quantification as a possibility to establish prognostic markers.

Copyright © 2018 The Authors. Published by Wolters Kluwer Health Inc., on behalf of the European Hematology Association.