Journal Article

POS0265 ESTABLISHING CONSISTENCY IN THE DIGITAL IMAGING ANALYSIS OF HISTOLOGICAL PARAMETERS IN SJOGREN’S DISEASE

Zabotti, A., Kroese, F. G. M., Bombardieri, M., Quartuccio, L., Liefers, S. C., Pitzalis, C., Van der Vegt, B., Pulvirenti, A., Sutcliffe, N., Pontarini, E., Tappuni, A., De Vita, S., Cavallaro, G., Colafrancesco, S., Rizzo, M. T., Verstappen, G. M.

2024

Overview

Background:Digital image analysis (DIA) is increasingly utilized for evaluating salivary gland (SG) histopathology in Sjögren’s disease (SD) [1,2]. DIA, particularly in assessing the area fraction (AF) of lympho-mononuclear cell aggregates, addresses the limitations of the focus score (FS) and minimizes variability between raters and across centers [1]. Utilizing DIA for quantifying intraepithelial B-lymphocytes proves valuable in the assessment of lymphoepithelial lesions [2]. In addition, DIA serves as a precise tool for longitudinally quantification of histopathological parameters before and after treatment, in randomized clinical trials (RCT) [3]. Despite on-going advancements in DIA and its apparent advantages over standard histopathology, its regular implementation is hampered by the absence of standardized protocols for histopathology and DIA parameters.Objectives:The aim is to establish consistency in evaluating salivary gland (SG) histopathology by defining i) standard operating protocols for DIA quantification and ii) identifying histological markers and parameters for monitoring inflammatory changes (Table 1). This harmonization is crucial to implement SG histopathology DIA tools in diagnosing and/or monitoring Sjögren’s disease (SD) histological progression.Methods:Digital slides of Hematoxylin and Eosin (H&E) and immunohistochemistry (IHC) from 21 Sjögren’s disease (SD) patients (11 labial, 10 parotid salivary glands) underwent blind analysis by three independent raters across three centers (UMCG-Groningen, QMUL-London, Rheumatology-Udine). Sequential sections were stained for both H&E and various IHC markers (CD45, CD3, CD20, CD21, CD138, IgG, IgA, and IgM), and the resulting images were shared for analysis. The QuPath open-source software was used to compare assessments for each marker, and the collected data were utilized to calculate the interclass correlation coefficient (ICC) to evaluate the consistency and reproducibility of agreement among the raters (Table 1). Quantitative surrogate markers for T/B cell segregation were evaluated using ROC-curve analysis.Results:A high inter-observer variability was confirmed in the focus score (FS) assessment highlighting a significant improvement when using aggregate area fraction (AF) on H&E staining. The use of CD45 facilitated the quicker identification of focal infiltrates, but did not show a notable enhancement in interclass correlation coefficient (ICC) compared to aggregate AF on H&E. In the quantitative assessment of CD3 (T-cells) and CD20 (B-cells), the percentage of positive cells (of all nucleated cells) exhibited a higher ICC compared to cell-density and positive AF (Table 1). In exploring T/B cell segregation, a key step in the lymphoid organization of inflammatory aggregates in SD, we identified an optimal threshold of 200 B-cells/mm2 to effectively distinguish between tissue sections with segregated and non-segregated aggregates. For identifying aggregates with ectopic lymphoid structure development, the digital count of CD21+ B-cell aggregates demonstrated a higher agreement rate compared to CD21+ AF. Finally, for quantifying plasma cells, the manually determined IgA/IgG shift (≤70% IgA+, ≥30% IgG+) proved to be a more reliable tool than CD138. As a standardization recommendation, we propose quantifying IgG- and IgA-positive plasma cells (as a percentage of positive cells) through digital image analysis (DIA) to objectively calculate the IgA/IgG shift.Conclusion:The proposed workflow for DIA is applicable to both labial and parotid SG, improving the standardization of the quantitative assessment in SG histopathology. This approach identifies histological parameters with high inter-rater agreement across different centers, thereby facilitating the harmonization of SG assessments among diverse cohorts and randomized controlled trials (RCTs).REFERENCES:[1] D. Lucchesi, E. Pontarini, et al. Clin. Exp. Rheumatol. 38 Suppl 1, 180–188 (2020).[2] M. S. van Ginkel, et al. Rheumatology 62, 428–438 (2022).[3] E. Pontarini, et at. Arthritis Rheumatol. (2023), doi:10.1002/art.42772.Acknowledgements:NIL.Disclosure of Interests:Elena Pontarini: None declared, Maria Teresa Rizzo: None declared, Silvia C. Liefers: None declared, Giulia Cavallaro: None declared, Bert van der Vegt: None declared, Serena Colafrancesco: None declared, Alen Zabotti: None declared, Luca Quartuccio: None declared, Alfredo Pulvirenti: None declared, Anwar Tappuni: None declared, Nurhan Sutcliffe: None declared, Salvatore De Vita: None declared, Costantino Pitzalis: None declared, Frans G.M. Kroese: None declared, Michele Bombardieri GSK, Janssen, Janssen, Gwenny M. Verstappen: None declared.

• Publisher: BMJ Publishing Group Ltd and European League Against Rheumatism,Elsevier B.V,Elsevier Limited
• Subject: Aggregates / CD20 antigen / CD3 antigen / CD45 antigen / Cell density / Clinical trials / Digital health / Digital health/Measuring health / Exocrine glands / Histopathology / Image processing / Imaging / Immunoglobulin A / Immunoglobulin G / Immunoglobulin M / Immunohistochemistry / Inflammation / Lymphocytes B / Lymphocytes T / Plasma cells / Rheumatology / Salivary gland / Scientific Abstracts / Sjogren's syndrome / Standardization / Validation