1
|
Sonomoto K, Fujino Y, Tanaka H, Nagayasu A, Nakayamada S, Tanaka Y. A Machine Learning Approach for Prediction of CDAI Remission with TNF Inhibitors: A Concept of Precision Medicine from the FIRST Registry. Rheumatol Ther 2024; 11:709-736. [PMID: 38637465 PMCID: PMC11111643 DOI: 10.1007/s40744-024-00668-z] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 01/24/2024] [Accepted: 03/18/2024] [Indexed: 04/20/2024] Open
Abstract
INTRODUCTION This study aimed to develop low-cost models using machine learning approaches predicting the achievement of Clinical Disease Activity Index (CDAI) remission 6 months after initiation of tumor necrosis factor inhibitors (TNFi) as primary biologic/targeted synthetic disease-modifying antirheumatic drugs (b/tsDMARDs) for rheumatoid arthritis (RA). METHODS Data of patients with RA initiating TNFi as first b/tsDMARD after unsuccessful methotrexate treatment were collected from the FIRST registry (August 2003 to October 2022). Baseline characteristics and 6-month CDAI were collected. The analysis used various machine learning approaches including logistic regression with stepwise variable selection, decision tree, support vector machine, and lasso logistic regression (Lasso), with 48 factors accessible in routine clinical practice for the prediction model. Robustness was ensured by k-fold cross validation. RESULTS Among the approaches tested, Lasso showed the advantages in predicting CDAI remission: with a mean area under the curve 0.704, sensitivity 61.7%, and specificity 69.9%. Predicted TNFi responders achieved CDAI remission at an average rate of 53.2%, while only 26.4% of predicted TNFi non-responders achieved remission. Encouragingly, the models generated relied solely on patient-reported outcomes and quantitative parameters, excluding subjective physician input. CONCLUSIONS While external cohort validation is warranted for broader applicability, this study highlights the potential for a low-cost predictive model to predict CDAI remission following TNFi treatment. The approach of the study using only baseline data and 6-month CDAI measures, suggests the feasibility of establishing regional cohorts to generate low-cost models tailored to specific regions or institutions. This may facilitate the application of regional/in-house precision medicine strategies in RA management.
Collapse
Affiliation(s)
- Koshiro Sonomoto
- Department of Clinical Nursing, School of Health Sciences, University of Occupational and Environmental Health, Japan, 1-1, Iseigaoka, Yahatanishi-ku, Kitakyushu, 807-8555, Japan
- The First Department of Internal Medicine, School of Medicine, University of Occupational and Environmental Health, Japan, 1-1, Iseigaoka, Yahatanishi-ku, Kitakyushu, 807-8555, Japan
| | - Yoshihisa Fujino
- Department of Environmental Epidemiology, Institute of Industrial Ecological Sciences, University of Occupational and Environmental Health, Japan, 1-1, Iseigaoka, Yahatanishi-ku, Kitakyushu, 807-8555, Japan
| | - Hiroaki Tanaka
- The First Department of Internal Medicine, School of Medicine, University of Occupational and Environmental Health, Japan, 1-1, Iseigaoka, Yahatanishi-ku, Kitakyushu, 807-8555, Japan
| | - Atsushi Nagayasu
- The First Department of Internal Medicine, School of Medicine, University of Occupational and Environmental Health, Japan, 1-1, Iseigaoka, Yahatanishi-ku, Kitakyushu, 807-8555, Japan
| | - Shingo Nakayamada
- The First Department of Internal Medicine, School of Medicine, University of Occupational and Environmental Health, Japan, 1-1, Iseigaoka, Yahatanishi-ku, Kitakyushu, 807-8555, Japan
| | - Yoshiya Tanaka
- The First Department of Internal Medicine, School of Medicine, University of Occupational and Environmental Health, Japan, 1-1, Iseigaoka, Yahatanishi-ku, Kitakyushu, 807-8555, Japan.
| |
Collapse
|
2
|
Fernández-Ávila DG, Patiño-Hernández D, Moreno-Luna S, Brance L, Arbeláez Á, Vilar AC, Lozada C, Ríos C, Toro C, Ramírez C, Pons-Estel G, Ugarte-Gil M, Narváez M, Albanese M, Roa O, Ruiz O, Burgos P, Xavier R, Fuentes Y, Soriano E. Development of a novel clinimetric tool: PAtient Reported Disease Activity Index in Rheumatoid Arthritis (PARDAI-RA) by PANLAR, for the assessment of patients living with rheumatoid arthritis. Clin Rheumatol 2024; 43:1277-1285. [PMID: 38355831 PMCID: PMC10944809 DOI: 10.1007/s10067-024-06868-w] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 09/11/2023] [Revised: 12/13/2023] [Accepted: 01/04/2024] [Indexed: 02/16/2024]
Abstract
BACKGROUND Clinical experience has shown that a single measure is not sufficient to assess disease activity in rheumatoid arthritis (RA). Various clinimetric tools are necessary to address the many clinical situations that can arise. METHODS In order to develop a comprehensive measurement tool, the Pan American League of Associations for Rheumatology searched for the most frequent measures of disease activity applied in RA by means of a semi-systematic review of the available literature. RESULTS We found that the most frequently reported measures of disease activity were the 28-joint Disease Activity Score, C-reactive protein, and the erythrocyte sedimentation rate, followed by patient-reported measures of pain and stiffness and many other composite indices and patient-reported outcome measures. The most frequent physician-reported sign of disease was the swollen joint count, and the most frequently self-reported feature was the increase in disease activity or flares. CONCLUSION In this article, we present a new clinimetric tool developed based on expert consensus and on data retrieved from our search. Disease activity can be better assessed by combining various data sources, such as clinical, laboratory, and self-reported outcomes. These variables were included in our novel clinimetric tool. Key Points • The goal of treatment of RA is to achieve the best possible control of inflammation, or even remission; therefore, disease management should include systematic and regular evaluation of inflammation and health status. • Clinimetric tools evaluate a series of variables (e.g., symptoms, functional capacity, disease severity, quality of life, disease progression) and can reveal substantial prognostic and therapeutic differences between patients. • Our clinimetric tool, which is based on a combination of data (e.g., clinical variables, laboratory results, PROMs), can play a relevant role in patient assessment and care.
Collapse
Affiliation(s)
- Daniel G Fernández-Ávila
- Rheumatology Division, Pontificia Universidad Javeriana - Hospital Universitario San Ignacio, Bogotá, Colombia.
| | | | | | - Lorena Brance
- Rheumatology Division, Universidad Nacional de Rosario, Santa Fe, Argentina
| | | | | | - Carlos Lozada
- Rheumatology Division, University of Miami, Coral Gables, USA
| | - Carlos Ríos
- Universidad de Especialidades Espíritu Santo, Guayaquil, Ecuador
| | - Carlos Toro
- Centro de Referencia en Osteoporosis y Reumatología, Cali, Colombia
| | | | - Guillermo Pons-Estel
- Centro Regional de Enfermedades Autoinmunes y Reumáticas (GO-CREAR), Rosario, Argentina
| | | | | | | | - Orlando Roa
- Rheumatology Division Keralty, Bogotá, Colombia
| | - Oscar Ruiz
- Rheumatology Division Keralty, Bogotá, Colombia
| | - Paula Burgos
- Clinic Immunology and Rheumatology Department, Pontificia Universidad Católica de Chile, Santiago de Chile, Chile
| | - Ricardo Xavier
- Rheumatology Service Hospital de Clinicas de Porto Alegre, Porto Alegre, Brazil
| | | | - Enrique Soriano
- Rheumatology Division, Hospital Italiano de Buenos Aires, Buenos Aires, Argentina
| |
Collapse
|
3
|
Schmukler J, Li T, Gibson KA, Morla RM, Luta G, Pincus T. Patient global assessment is elevated by up to 5 of 10 units in patients with inflammatory arthritis who screen positive for fibromyalgia (by FAST4) and/or depression (by MDS2) on a single MDHAQ. Semin Arthritis Rheum 2023; 58:152151. [PMID: 36586208 DOI: 10.1016/j.semarthrit.2022.152151] [Citation(s) in RCA: 7] [Impact Index Per Article: 7.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 08/11/2022] [Revised: 12/01/2022] [Accepted: 12/06/2022] [Indexed: 12/13/2022]
Abstract
BACKGROUND Patient global assessment (PATGL) is a component of rheumatoid arthritis (RA) and spondyloarthritis (SpA) activity indices, reflecting inflammation in selected clinical trial patients. In routine care, PATGL often may be elevated independently of inflammatory activity by fibromyalgia (FM) and/or depression, leading to complexities in interpretation. A feasible method to screen for FM and/or depression could help to clarify interpretation of high PATGL and index scores, including explanation of apparent limited responses to anti-inflammatory therapies. PATIENTS AND METHODS Patients with RA or SpA in routine care in Barcelona, Chicago, and Sydney complete a 2-page multidimensional health assessment questionnaire (MDHAQ) in 5-10 min. The MDHAQ includes PATGL and three indices, RAPID3 (routine assessment of patient index data) to assess clinical status, FAST4 (0-4 fibromyalgia assessment screening tool) and MDS2 (0-2 MDHAQ depression screen). PATGL was compared for each diagnosis at each site and pooled data in FAST4 positive (+) vs negative (-) and/or MDS2+ vs MDS2- patients using medians and median regressions. RESULTS Median PATGL was 5.0 in 393 RA and 175 SpA patients; 2.0-3.0 in 305 (58.9%) FAST4-,MDS2- patients, 5.5-6.0 in 71 (13.7%) FAST4-,MDS2+ patients, 7.0-7.5 in 50 (9.7%) FAST4+,MDS2- patients, and 7.0-8.0 in 92 (17.8%) FAST4+,MDS2+ patients. Positive FAST4 and/or MDS2 screens were seen in 41% of patients. Results were similar in RA and SpA at 3 settings on 3 continents. CONCLUSION Median 0-10 PATGL varied from 2-3/10 to 5.5-8/10, according to negative vs positive screening for FM and/or depression on a single MDHAQ for busy clinical settings.
Collapse
Affiliation(s)
- Juan Schmukler
- Division of Rheumatology, Department of Medicine, Rush University School of Medicine, Chicago, IL 60612, United States
| | - Tengfei Li
- Department of Biostatistics, Bioinformatics & Biomathematics, Georgetown University, Washington, DC 20057, United States
| | - Kathryn A Gibson
- Department of Rheumatology, Liverpool Hospital, Ingham Research Institute, University of New South Wales, Sydney, NSW 2170, Australia
| | - Rosa M Morla
- Department of Rheumatology, Hospital Clinic Universitari de Barcelona, Institut d´investigacions Biomèdiques August Pi i Sunyer, 08036 Barcelona, Spain
| | - George Luta
- Department of Biostatistics, Bioinformatics & Biomathematics, Georgetown University, Washington, DC 20057, United States
| | - Theodore Pincus
- Division of Rheumatology, Department of Medicine, Rush University School of Medicine, Chicago, IL 60612, United States.
| |
Collapse
|
4
|
De Cock D, Buckinx E, Pazmino S, Bertrand D, Stouten V, Westhovens R, Verschueren P. Belgian rheumatologists' preferences regarding measures of disease activity in patients with rheumatoid arthritis: results from a mixed-methods study. Rheumatol Int 2021; 42:815-823. [PMID: 34687348 DOI: 10.1007/s00296-021-05020-0] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 06/29/2021] [Accepted: 10/02/2021] [Indexed: 12/01/2022]
Abstract
The reliability and clinical usefulness of the different composite disease activity scores and their individual components in Rheumatoid Arthritis (RA) are still debated. This study investigated which measures of disease activity were preferred by rheumatologists. A mixed-method study was performed. First, ten Belgian rheumatologists were invited for individual interviews on their current practice and preferences for measurement of RA disease activity. Results of this qualitative study and evidence from literature served as input for developing a survey. This survey asked rheumatologists to rate preferred standard disease activity score(s), their individual components, ultrasound and related patient-reported outcomes (PROs), by maximum difference scaling. The relative importance score (RIS) for each indicator was calculated using hierarchical Bayes modeling. The qualitative study included 6/10 invited rheumatologists. Composite scores and components were perceived as useful, while PROs were found subjective. Interestingly, ultrasound was used to mediate discrepancies between physician and patient. The survey based on this was sent to 244 Belgian rheumatologists, 83/244 (34%) responded, including 66/83 (80%) complete and 17/83 (20%) incomplete surveys (two missing essential information). Most rheumatologists (75/81, 93%) used a disease activity score and 68/81 (84%) preferred the DAS28-CRP. Swollen joint count obtained the highest mean ± SD RIS (22.54 ± 2.64), followed by DAS28 ESR/CRP (20.61 ± 4.06), ultrasound (16.47 ± 7.97), CRP (13.34 ± 6.11) and physician's global assessment (12.59 ± 7.83). PROs including fatigue, pain, and patient's global assessment, and Health Assessment Questionnaire, obtained the lowest mean RIS (0.34-2.54). Rheumatologists place more faith in self-assessed disease activity components or in laboratory tests. Trust in PROs to evaluate disease activity is low in clinical practice.
Collapse
Affiliation(s)
- D De Cock
- Skeletal Biology and Engineering Research Centre, KU Leuven, ON IV Herestraat 49, P. O. Box 813, 3000, Leuven, Belgium.
| | - E Buckinx
- Skeletal Biology and Engineering Research Centre, KU Leuven, ON IV Herestraat 49, P. O. Box 813, 3000, Leuven, Belgium
| | - S Pazmino
- Skeletal Biology and Engineering Research Centre, KU Leuven, ON IV Herestraat 49, P. O. Box 813, 3000, Leuven, Belgium
| | - D Bertrand
- Skeletal Biology and Engineering Research Centre, KU Leuven, ON IV Herestraat 49, P. O. Box 813, 3000, Leuven, Belgium
| | - V Stouten
- Skeletal Biology and Engineering Research Centre, KU Leuven, ON IV Herestraat 49, P. O. Box 813, 3000, Leuven, Belgium
| | - R Westhovens
- Skeletal Biology and Engineering Research Centre, KU Leuven, ON IV Herestraat 49, P. O. Box 813, 3000, Leuven, Belgium.,Department of Rheumatology, University Hospitals of Leuven, 3000, Leuven, Belgium
| | - P Verschueren
- Skeletal Biology and Engineering Research Centre, KU Leuven, ON IV Herestraat 49, P. O. Box 813, 3000, Leuven, Belgium.,Department of Rheumatology, University Hospitals of Leuven, 3000, Leuven, Belgium
| |
Collapse
|
5
|
Bugatti S, De Stefano L, Manzo A, Sakellariou G, Xoxi B, Montecucco C. Limiting factors to Boolean remission differ between autoantibody-positive and -negative patients in early rheumatoid arthritis. Ther Adv Musculoskelet Dis 2021; 13:1759720X211011826. [PMID: 34093745 PMCID: PMC8141996 DOI: 10.1177/1759720x211011826] [Citation(s) in RCA: 8] [Impact Index Per Article: 2.7] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 11/16/2020] [Accepted: 03/31/2021] [Indexed: 12/14/2022] Open
Abstract
Background: The patient global assessment of disease activity (PGA) is the major limiting factor to Boolean remission in patients with established rheumatoid arthritis (RA). Here, we investigated the limiting variables to disease remission in patients with early RA treated with conventional synthetic disease modifying anti-rheumatic drugs, also in relation to autoantibody status. Methods: Data were retrieved from 535 early RA patients (<12 months of symptoms) with an observation period of 6–12 months upon initiation of therapy with methotrexate aimed at the achievement of low disease activity based on the 28-joints disease activity score. Near-remission was defined as any of the four core items of Boolean remission >1 with the remaining three all ⩽1. Reasons for missing Boolean remission and predictors of near-remission subcategories were analyzed in relation to baseline disease variables. Results: After 6 and 12 months, near-remission was two-times more frequent than Boolean remission (25.6% and 26.9% at the two time-points). A 28-swollen joint count (SJC28) >1 was responsible for the majority of near-remission (56.2% and 57.6% at 6 and 12 months, respectively), and PGA > 1 accounted for approximatively 35% of the cases. Autoantibody-positivity independently predicted the risk of missing remission because of SJC28 > 1 [adjusted odds ratio (OR) 95% confidence interval (CI) 2.81 (1.59–4.9) at 6 months and 1.73 (1.01–3.01) at 12 months], whilst autoantibody-negativity was an independent predictor of PGA near-remission [adjusted OR (95% CI) 2.45 (1.25–4.80) at 6 months and 5.71 (2.47–13.2) at 12 months]. Conclusion: In early RA, Boolean remission is more frequently missed because of persistent swollen joints. However, barriers to full-remission vary in relation to the autoantibody status. Autoantibody-positive patients more commonly experience residual swollen joints, whilst PGA more frequently impairs remission in autoantibody-negative patients. Efforts to target full-remission in early RA may thus require different strategies according to autoantibody profile.
Collapse
Affiliation(s)
- Serena Bugatti
- Division of Rheumatology, IRCCS Policlinico San Matteo Foundation University Hospital, Viale Golgi 19, Pavia, 27100, Italy
| | - Ludovico De Stefano
- Division of Rheumatology, IRCCS Policlinico San Matteo Foundation, Pavia, Italy
| | - Antonio Manzo
- Division of Rheumatology, IRCCS Policlinico San Matteo Foundation, Pavia, Italy
| | | | - Blerina Xoxi
- Division of Rheumatology, IRCCS Policlinico San Matteo Foundation, Pavia, Italy
| | | |
Collapse
|