Deep immunophenotyping reveals that autoimmune and autoinflammatory disorders are spread along two immunological axes capturing disease inflammation levels and types.

OBJECTIVES: Based on genetic associations, McGonagle and McDermott suggested a classification of autoimmune and autoinflammatory diseases as a continuum ranging from purely autoimmune to purely autoinflammatory diseases and comprising diseases with both components. We used deep immunophenotyping to identify immune cell populations and molecular targets characterising this continuum. METHODS: We collected blood from 443 patients with one of 15 autoimmune or autoinflammatory diseases and 71 healthy volunteers. Deep phenotyping was performed using 13 flow cytometry panels characterising over 600 innate and adaptive cell populations. Unsupervised and supervised analyses were conducted to identify disease clusters with their common and specific cell parameters. RESULTS: Unsupervised clustering categorised these diseases into five clusters. Principal component analysis deconvoluted this clustering into two immunological axes. The first axis was driven by the ratio of LAG3+ to ICOS+ in regulatory T lymphocytes (Tregs), and segregated diseases based on their inflammation levels. The second axis was driven by activated Tregs and type 3 innate lymphoid cells (ILC3s), and segregated diseases based on their types of affected tissues. We identified a signature of 23 cell populations that accurately characterised the five disease clusters. CONCLUSIONS: We have refined the monodimensional continuum of autoimmune and autoinflammatory diseases as a continuum characterised by both disease inflammation levels and targeted tissues. Such classification should be helpful for defining therapies. Our results call for further investigations into the role of the LAG3+/ICOS+ balance in Tregs and the contribution of ILC3s in autoimmune and autoinflammatory diseases. TRIAL REGISTRATION NUMBER: NCT02466217.

Identification of Symptom Phenotypes of Hand Osteoarthritis Using Hierarchical Clustering: Results From the DIGICOD Cohort.

OBJECTIVE: We aimed to delineate phenotypes in hand osteoarthritis (HOA) based on cardinal symptoms (pain, functional limitation, stiffness, and aesthetic discomfort). METHODS: With data from the Digital Cohort Design (DIGICOD), we performed a hierarchical agglomerative clustering analysis based on Australian/Canadian Osteoarthritis Hand Index (AUSCAN) subscores for pain, physical function, stiffness, and visual analog scale for aesthetic discomfort. Kruskal-Wallis and post hoc analyses were used to assess differences between clusters. RESULTS: Among 389 patients, we identified 5 clusters: cluster 1 (n = 88) and cluster 2 (n = 91) featured low and mild symptoms; cluster 3 (n = 80) featured isolated aesthetic discomfort; cluster 4 (n = 42) featured a high level of pain, stiffness, and functional limitation; and cluster 5 (n = 88) had the same features as cluster 4 but with high aesthetic discomfort. For clusters 4 and 5, AUSCAN pain score was >41 of 100, representing only one-third of our patients. Aesthetic discomfort (clusters 3 and 5) was significantly associated with erosive HOA and a higher number of nodes. The highly symptomatic cluster 5 was associated but not significantly with metabolic syndrome, and body mass index and C-reactive protein level did not differ among clusters. Symptom intensity was significantly associated with joint destruction as well as with physical and psychological burden. Patients' main expectations differed among clusters, and function improvement was the most frequent expectation overall. CONCLUSION: The identification of distinct clinical clusters based on HOA cardinal symptoms suggests previously undescribed subtypes of this condition, warranting further study of biological characteristics of such clusters, and opening a path toward phenotype-based personalized medicine in HOA.

Use of machine learning in osteoarthritis research: a systematic literature review.

OBJECTIVE: The aim of this systematic literature review was to provide a comprehensive and exhaustive overview of the use of machine learning (ML) in the clinical care of osteoarthritis (OA). METHODS: A systematic literature review was performed in July 2021 using MEDLINE PubMed with key words and MeSH terms. For each selected article, the number of patients, ML algorithms used, type of data analysed, validation methods and data availability were collected. RESULTS: From 1148 screened articles, 46 were selected and analysed; most were published after 2017. Twelve articles were related to diagnosis, 7 to prediction, 4 to phenotyping, 12 to severity and 11 to progression. The number of patients included ranged from 18 to 5749. Overall, 35% of the articles described the use of deep learning And 74% imaging analyses. A total of 85% of the articles involved knee OA and 15% hip OA. No study investigated hand OA. Most of the studies involved the same cohort, with data from the OA initiative described in 46% of the articles and the MOST and Cohort Hip and Cohort Knee cohorts in 11% and 7%. Data and source codes were described as publicly available respectively in 54% and 22% of the articles. External validation was provided in only 7% of the articles. CONCLUSION: This review proposes an up-to-date overview of ML approaches used in clinical OA research and will help to enhance its application in this field.