POS1051 Lung Clustering Analysis-Based Phenotypes of Rheumatoid Arthritis Using Artificial Intelligence-Based Technology for Chest Computed Tomography

843 Rheumatoid arthritis - comorbidity and clinical aspects POS1050 INCIDENCE RATES OF INTERSTITIAL LUNG DISEASE AMONG PATIENTS WITH RHEUMATOID ARTHRITIS TREATED WITH ABATACEPT: A POST HOC POOLED ANALYSIS FROM A COMPENDIUM OF CLINICAL TRIALS Keywords: Rheumatoid arthritis, Lungs, Disease-modifying drug (DMARDs) P. Dieudé 1, J. Sparks 2, A. Fischer 3, L. Chen 4, K. Lozenski 5, S. Dahan 6, M. Chaballa 7, W. Little 8. 1Paris Cité University, Bichat Hospital, APHP, Department of Rheumatology, Paris, France; 2Brigham and Women’s Hospital, Harvard Medical School, Division of Rheumatology, Inflammation and Immunity, Boston, United States of America; 3Bristol Myers Squibb, Lung Fibrosis, Lawrenceville, United States of America; 4Syneos Health, Biostatistics and Statistical Programming, Richmond, Canada; 5Bristol Myers Squibb, Rheumatology Pipeline and Early Assets, Lawrenceville, United States of America; 6Bristol Myers Squibb, Worldwide Medical, Lawrenceville, United States of America; 7Bristol Myers Squibb, US Medical, Immunology and Fibrosis, Lawrenceville, United States of America; 8Bristol Myers Squibb, Worldwide Rheumatology, Lawrenceville, United States of America Background: Interstitial lung disease (ILD) is a recognized complication of RA. Prior studies have suggested stabilization or improvement of ILD in patients with RA (RA-ILD) treated with abatacept.[1,2] Few studies have evaluated the background incidence rate of RA-ILD. Objectives: To determine the incidence rate of clinically significant ILD in a cohort of patients with RA receiving abatacept + MTX versus placebo + MTX from multiple clinical trials. Methods: This retrospective analysis examined pooled safety data from ten phase 3 clinical trials of patients with RA treated with background MTX in combination with abatacept or placebo. The term ‘interstitial lung’ was used to identify incidences of ILD reported as AEs in the safety data. The exposure period for each patient was censored at first incidence of clinically significant ILD, 56 days after last study drug administration, or 1 day prior to commencement of another study drug, whichever occurred first. Poisson regression models were used to estimate crude incidence rates per 100 person-years for baseline risk factors within treatment groups, and to estimate incidence rate ratios for the placebo + MTX versus abatacept + MTX treatment groups. Disease activity parameters, DAS28 (CRP) and HAQ-DI, were estimated from baseline to the time of ILD diagnosis (as reported by AEs). Results: In total, 3,708 patients (10,521 person-years) treated with abatacept + MTX and 999 patients (938 person-years) treated with placebo + MTX were included. Patients treated with placebo + MTX had a higher incidence rate of ILD per 100 person-years (95% CI) versus those treated with abatacept + MTX (0.43 [0.16–1.14] vs 0.10 [0.05–0.18], respectively; Figure 1). The incidence rate ratio of placebo + MTX versus abatacept + MTX treatment groups for the total population was 4.49 (95% CI, 1.23–13.42). For all subpopulations stratified by baseline risk factors (where ≥ 1 patient in each treatment group had an ILD event), incidence rate ratios were > 2. Patients with RA aged ≥ 55 years, BMI < 30 kg/m 2, no history of smoking, no baseline DMARD use other than MTX, no prior TNF inhibitor use, corticosteroid use, high DAS28 (CRP) status, and RF or anti-citrullinated protein antibody (ACPA) positivity were all significantly less likely to develop ILD if treated with abatacept + MTX versus placebo + MTX (Figure 1). Conclusion: Incidence rates of ILD among patients with RA were significantly lower in those treated with abatacept + MTX than placebo + MTX, both in the total analysis population and subpopulations stratified by baseline risk factors. This suggests a possible protective benefit of abatacept on the incidence rate of RA-ILD. Limitations of this post hoc analysis include the different observation period lengths of the 2 treatment arms due to the original study designs, and the low number of patients with ILD events overall. However, the low incidence rate of ILD observed here is consistent with previously reported studies.[3] REFERENCES: [1] Mena-Vázquez N, et al. Biomedicines 2022;10:1480. [2] Fernández-Díaz C, et al. Rheumatology 2020;59:3906–3916. [3] Curtis JR, et al. Arthritis Res Ther 2015;17:319. Acknowledgements: This study was sponsored by Bristol Myers Squibb. Professional medical writing and editorial assistance was provided by Ellen Seeley, MSc, of Caudex, and was funded by Bristol Myers Squibb. Disclosure of Interests: Philippe Dieudé Speakers bureau: AbbVie, Boehringer Ingelheim, Bristol Myers Squibb, Chugai, Janssen, Lilly, MEDAC, Novartis, Pfizer, Roche, Consultant of: Boehringer Ingelheim, Bristol Myers Squibb, Chugai, Lilly, Pfizer, Roche, Grant/research support from: Bristol Myers Squibb, Galapagos, Pfizer, Jeffrey Sparks Consultant of: AbbVie, Amgen, Boehringer Ingelheim, Bristol Myers Squibb, Gilead, Inova Diagnostics, Janssen, Optum, Pfizer, Grant/ research support from: Bristol Myers Squibb, Aryeh Fischer Shareholder of: Bristol Myers Squibb, Employee of: Bristol Myers Squibb, Leo Chen Consultant of: Bristol Myers Squibb, Employee of: Syneos Health, Karissa Lozenski Shareholder of: Bristol Myers Squibb, Employee of: Bristol Myers Squibb, Stephanie Dahan Shareholder of: Bristol Myers Squibb, Consultant of: Otsuka Pharmaceutical, Grant/research support from: Bristol Myers Squibb, Janssen Pharmaceuticals, Employee of: Bristol Myers Squibb, Mark Chaballa Shareholder of: Bristol Myers Squibb, Employee of: Bristol Myers Squibb, Wayne Little Shareholder of: Bristol Myers Squibb, Employee of: Bristol Myers Squibb. DOI: 10.1136/annrheumdis-2023-eular.250 POS1051 LUNG CLUSTERING ANALYSIS-BASED PHENOTYPES OF RHEUMATOID ARTHRITIS USING ARTIFICIAL INTELLIGENCE-BASED TECHNOLOGY FOR CHEST COMPUTED TOMOGRAPHY Keywords: Rheumatoid arthritis, Artificial intelligence, Lungs Y. Nakayama1, R. Nakashima1, T. Handa2, K. Tanizawa2, H. Onizawa3, T. Fujii3, K. Murata3, K. Murakami4, A. Onishi3, M. Tanaka3, M. Shirakashi1, R. Hiwa1, H. Tsuji1, K. Kitagori1, S. Akizuki1, H. Yoshifuji1, A. Morinobu1. 1Kyoto University, Department of Rheumatology and Clinical Immunology, Kyoto, Japan; 2Kyoto University, Department of Respiratory Medicine, Kyoto, Japan; 3Kyoto University, Department of Advanced Medicine for Rheumatic Diseases, Kyoto, Japan; 4 Kyoto University, Center for Cancer Immunotherapy and Immunobiology, Kyoto, Japan Background: Lung involvement is a prevalent extraarticular manifestation of rheumatoid arthritis (RA) that remains a significant clinical challenge. Few studies have comprehensively quantified lung abnormalities of RA patients using artificial intelligence-based (AI) technology. Objectives: The aim of this study was to quantify lung lesions in RA patients and classify them based on their lung parameters. Methods: An AI-based quantitative computed tomography (CT) image analysis software (AIQCT) was applied to high-resolution CT scans of RA patients in a cross-sectional manner. AIQCT automatically classified and quantified 10 types of parenchymal image patterns, expressing the volumes as percentages of total lung volume [1]. Hierarchical Ward’s linkage clustering based on these patterns identified five clusters. Visual assessments (ILD, and airway lesions) of HRCT and clinical phenotypes were assessed. Results: A total of 408 RA patients were included in the study. The lung profiles of the five clusters were as follows: Cluster I (68.6%), characterized by nearly normal lungs; Cluster II (23.5%), characterized by slight lung lesions with honeycombs or ground-glass opacities (GGOs); Cluster III (5.6%), characterized by the Ann Rheum Dis: first published as 10.1136/annrheumdis-2023-eular.2178 on 30 May 2023. Downloaded from http://ard.bmj.com/ on August 27, 2024 by guest. Protected by copyright. Scientific Abstracts Scientific Abstracts predominance of GGOs; Cluster IV (1.0%), characterized by a predominant hyperlucent area; and Cluster V (1.2%), characterized by extensive lung abnormalities (Figure 1, Table 1). The number of patients in each cluster with ILD and airway lesions, based on visual assessments, were as follows: [ILD] Cluster I, 11/280 (3.9%), Cluster II, 19/96 (19.8%), Cluster III, 5/23 (21.7%), Cluster IV, 0/4 (0%), and Cluster V, 5/5 (100%) (p < 0.001): [airway lesions] Cluster I, 31/280 (11.1%), Cluster II, 29/96 (30.2%), Cluster III, 6/23 (26.1%), Cluster IV, 0/4 (0%), and Cluster V, 1/5 (20%) (p < 0.001). Clinical characteristics of each cluster were described in Table 1. The disease activity of RA in each cluster were as follows: [DAS28ESR, mean (SD)] Cluster I, 2.4 (1.0), Cluster II, 3.0 (1.2), Cluster III, 2.8 (1.0), Cluster IV, 2.2 (0.7), and Cluster V, 3.0 (0.8) (p < 0.001); [HAQ, mean (SD)] Cluster I, 0.4 (0.6), Cluster II, 0.9 (0.9), Cluster III, 1.1 (0.9), Cluster IV, 0.3 (0.3), and Cluster V, 0.8 (0.9) (p < 0.001). Conclusion: This study is the first to classify lung lesions in comprehensive RA patients using quantitative data derived from novel AI technology. The AIQCT-derived clustering of RA patients appears to be associated with their clinical backgrounds and characteristics. REFERENCE: [1] Ann Am Thorac Soc 2022;19:399-406 Table 1. Scores of ten lung parenchymal image patterns and clinical characteristics in each cluster. Cluster I (n=280) II (n=96) AIQCT lung score (%), mean (SD) Normal GGOs Reticulation Consolidation Honeycombs Small nodules Interlobular septum Hyperluency Bronchi Blood vessels Age, mean (SD) Sex; female, n (%) Smoking history, n (%) RF titer, mean (SD) ACPA titer, mean (SD) MTX, n (%) PSL, n (%) bDMARDs, n (%) 92.6 (1.5) 90.7 (2.5) 0.7 (0.3) 1.0 (0.4) 0.1 (0.1) 0.4 (0.4) 0.1 (0.05) 0.3 (0.3) 0 (0.02) 0.08 (0.2) 0.1 (0.06) 0.3 (0.2) 0.1 (0.06) 0.3 (0.2) 0.1 (0.3) 0.7 (1.6) 1.8 (0.4) 2.3 (0.6) 4.4 (1.0) 4 (1.1) 62.9 (12.0) 72.2 (8.0) 240 (86) 81 (84) 94 (34) 25 (26) 94 (165) 151 (216) 132 (189) 181 (306) 201 (72) 64 (67) 45 (16) 32 (33) 136 (49) 45 (47) III (n=23) IV (n=4) V (n=5) p value 87.4 (1.8) 2.9 (1.4) 0.5 (0.5) 0.3 (0.2) 0.01 (0.02) 0.5 (0.6) 0.4 (0.3) 0.1 (0.1) 2.5 (0.7) 5.5 (1.1) 70.4 (9.9) 19 (83) 8 (35) 184 (328) 132 (161) 13 (57) 8 (35) 9 (39) 76.3 (5.1) 0.6 (0.3) 0.2 (0.2) 0.4 (0.6) 0.07 (0.05) 0.1 (0.04) 0.1 (0.07) 17.1 (3.9) 1.9 (0.3) 3.3 (0.6) 68.5 (4.8) 1 (25) 4 (100) 120 (130) 92 (41) 3 (75) 1 (25) 2 (50) 80.2 (1.6) 2.4 (0.5) 3.8 (3.3) 1.6 (1.1) 1.7 (1.8) 0.5 (0.4) 0.5 (0.3) 0.4 (0.4) 4.2 (1.4) 4.7 (0.5) 60.0 (8.0) 5 (100) 0 (0) 288 (425) 677 (1018) 1 (20) 4 (80) 5 (100) <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 0.015 0.019 0.012 <0.001 0.065 <0.001 0.183 Figure 1. Clustering constellation tree diagram for ten lung parenchymal HRCT image patterns of rheumatoid arthritis. Acknowledgements: NIL. Disclosure of Interests: Yoichi Nakayama: None declared, Ran Nakashima: None declared, Tomohiro Handa: None declared, Kiminobu Tanizawa: None declared, Hideo Onizawa: None declared, Takayuki Fujii: None declared, Koichi Murata Speakers bureau: AbbVie GK, Eisai Co., Ltd., Pfizer Inc., Chugai Pharmaceutical Co., Ltd., Mitsubishi Tanabe Pharma Corporation, Pfizer Inc., Bristol-Myers Squibb, Daiichi Sankyo Co. Ltd., and Asahi Kasei Pharma Corp., Grant/research support from: Daiichi Sankyo Co. Ltd, Kosaku Murakami: None declared, Akira Onishi: None declared, Masao Tanaka: None declared, Mirei Shirakashi: None declared, Ryosuke Hiwa: None declared, Hideaki Tsuji: None declared, Koji Kitagori: None declared, Syuji Akizuki: None declared, Hajime Yoshifuji: None declared, Akio Morinobu: None declared. DOI: 10.1136/annrheumdis-2023-eular.2178 POS1052 RHEUMATOID ARTHRITIS ASSOCIATED INTERSTITIAL LUNG DISEASE ACROSS CONTINENTS Keywords: Rheumatoid arthritis, Lungs, Comorbidities S. L. Heckert1, T. Maarseveen1, E. R. Marges2, A. Chopra3, D. Vega-Morales4, R. Du Toit5, L. L. Winchow6, C. E. Toro Gutierrez7, R. Knevel1,8, A. Van der Helm – van Mil1, T. Huizinga1, C. Allaart1, S. A. Bergstra1. 1Leiden University Medical Center (LUMC), Rheumatology, Leiden, Netherlands; 2Leiden University Medical Center (LUMC), Pulmonology, Leiden, Netherlands; 3 Center For Rheumatic Diseases, Rheumatology, Pune, India; 4Hospital General Zona 17, Instituto Mexicano del Seguro Social, Rheumatology, Monterrey, Mexico; 5Stellenbosch University and Tygerberg Academic Hospital, Rheumatology, Department of Medicine, Faculty of Medicine and Health Sciences, Cape Town, South Africa; 6Chris Hani Baragwanath Academic Hospital, Rheumatology, Department of Internal Medicine, Johannesburg, South Africa; 7Centro de Referencia en Osteoporosis & Reumatología, Internal medicine and Rheumatology, Cali, Colombia; 8 Newcastle University, Translational and Clinical Research Institute, Newcastle upon Tyne, United Kingdom Background: Interstitial lung disease (ILD) is a spectrum of inflammatory and fibrotic lung diseases, and can be associated with RA (RA-ILD). The reported prevalence ranges from 1.8 to 58 percent, depending on the definition and diagnostics. Objectives: To investigate the incidence and prevalence of RA-ILD in different countries worldwide. Methods: Patients of 5 countries from 2 observational databases were studied. Patients from India, Mexico, South Africa (two clinics) and Colombia with a physician-based RA diagnosis were selected from the observational METEOR database. From the Leiden EAC in the Netherlands, patients with early RA (1987 ACR/ 2010 ACR/EULAR criteria) were included. A clinical diagnosis of RA-ILD was based on chest X-ray or CT. X-ray was performed at the first visit in India, South Africa and the Netherlands and on clinical indication in all countries. Prevalence of RA-ILD at the end of follow-up was calculated for each country. Incidence rates (IR) were calculated in patients with newly diagnosed RA and an available baseline visit. Patient characteristics were described, comparing patients with and without RA-ILD using appropriate statistical tests. Results: Within the five countries 16,667 patients with RA, both newly diagnosed and with longer RA disease duration, were evaluated. Prevalence and incidence of RA-ILD differed per country. The prevalence of RA-ILD at the end of follow-up was 0.7% (84/11,733) in India (mean follow-up 12±20 months), 2.1% (9/427) in Mexico (mean follow-up 17±27 months), 2.0% (21/1,077) in the Netherlands (mean follow-up 83±71 months), 3.0% (19/629) in South Africa (mean follow-up 20±21 months) and 0.7% (18/2,747) in Colombia (mean follow-up 14±12 months). The IR of RA-ILD in newly diagnosed RA patients in India was 1.6 (95% CI 1.0-2.5) per 1000 person years. In the Netherlands the IR was 3.8 (95% CI 1.6-9.1) per 1000 person years. In South Africa the IR was 6.6 (95% CI 2.5-17.5) per 1000 person years. For Mexico and Colombia, no IR could be calculated. Patient characteristics are described in Table 1. In India and Mexico, patients with RA-ILD were older. In 4 of 5 countries RA-ILD patients were more often male. Higher inflammatory markers and more RF positivity were seen in RA-ILD patients in India and the Netherlands. In South Africa, patients with RA-ILD more often had a history of smoking. ACPA positivity was more frequent in RA-ILD patients from India, but less frequent in RA-ILD patients from South Africa. Conclusion: Prevalence and incidence of RA-ILD varied between the four countries with the highest prevalence and incidence in South Africa. These differences might be partially explained by differences in diagnostic approaches, and potentially in patient populations, used therapies and comorbidity. REFERENCES: NIL. Acknowledgements: NIL. Disclosure of Interests: Sascha L Heckert Grant/research support from: Bristol-Myers Squibb (BMS), Tjardo Maarseveen: None declared, Emiel R Marges: None declared, Arvind Chopra: None declared, David Vega-Morales: None declared, Riette du Toit: None declared, Lai-Ling Winchow: None declared, Carlos Enrique Toro Gutierrez: None declared, Rachel Knevel: None declared, Annette van der Helm – van Mil: None declared, Thomas Huizinga: None declared, Cornelia Allaart: None declared, Sytske Anne Bergstra: None declared. DOI: 10.1136/annrheumdis-2023-eular.2245 Ann Rheum Dis: first published as 10.1136/annrheumdis-2023-eular.2178 on 30 May 2023. Downloaded from http://ard.bmj.com/ on August 27, 2024 by guest. Protected by copyright. 844