Artificial intelligence model for segmentation and severity scoring of osteophytes in hand osteoarthritis on ultrasound images.

Objective: To develop an artificial intelligence (AI) model able to perform both segmentation of hand joint ultrasound images for osteophytes, bone, and synovium and perform osteophyte severity scoring following the EULAR-OMERACT grading system (EOGS) for hand osteoarthritis (OA). Methods: One hundred sixty patients with pain or reduced function of the hands were included. Ultrasound images of the metacarpophalangeal (MCP), proximal interphalangeal (PIP), distal interphalangeal (DIP), and first carpometacarpal (CMC1) joints were then manually segmented for bone, synovium and osteophytes and scored from 0 to 3 according to the EOGS for OA. Data was divided into a training, validation, and test set. The AI model was trained on the training data to perform bone, synovium, and osteophyte identification on the images. Based on the manually performed image segmentation, an AI was trained to classify the severity of osteophytes according to EOGS from 0 to 3. Percent Exact Agreement (PEA) and Percent Close Agreement (PCA) were assessed on individual joints and overall. PCA allows a difference of one EOGS grade between doctor assessment and AI. Results: A total of 4615 ultrasound images were used for AI development and testing. The developed AI model scored on the test set for the MCP joints a PEA of 76% and PCA of 97%; for PIP, a PEA of 70% and PCA of 97%; for DIP, a PEA of 59% and PCA of 94%, and CMC a PEA of 50% and PCA of 82%. Combining all joints, we found a PEA between AI and doctor assessments of 68% and a PCA of 95%. Conclusion: The developed AI model can perform joint ultrasound image segmentation and severity scoring of osteophytes, according to the EOGS. As proof of concept, this first version of the AI model is successful, as the agreement performance is slightly higher than previously found agreements between experts when assessing osteophytes on hand OA ultrasound images. The segmentation of the image makes the AI explainable to the doctor, who can immediately see why the AI applies a given score. Future validation in hand OA cohorts is necessary though.

Development and validation of a new tool for assessment of trainees' interventional musculoskeletal ultrasound skills.

OBJECTIVES: Interventional musculoskeletal ultrasound (MSUS) procedures are routinely performed in rheumatology practice. However, the efficacy and safety of the procedures rely on the competence of the physician, and assessment of skills is crucial. Thus, this study aimed to develop and establish validity evidence for a tool assessing trainees' interventional MSUS skills. METHODS: An expert panel of rheumatologists modified an existing tool for assessing competences in invasive abdominal and thoracic ultrasound procedures. The new tool (the Assessment of Interventional Musculoskeletal Ultrasound Skills (AIMUS) tool) reflects the essential steps in interventional MSUS. To establish validity evidence, physicians with different levels of interventional MSUS experience were enrolled and performed two procedures on a rubber fantom, simulating real patient cases. All performances were video recorded, anonymized, and assessed in random order by two blinded raters using the AIMUS tool. RESULTS: Sixty-five physicians from 21 different countries were included and categorized into groups based on their experience, resulting in 130 videos for analysis. The internal consistency of the tool was excellent, with a Cronbach's alpha of 0.96. The inter-case reliability was good with a Pearson's correlation coefficient (PCC) of 0.74 and the inter-rater reliability was moderate to good (PCC 0.58). The ability to discriminate between different levels of experience was highly significant (p< 0.001). CONCLUSIONS: We have developed and established validity evidence for a new interventional MSUS assessment tool. The tool can be applied in future competency-based educational programs, provide structured feedback to trainees in daily clinical practice, and ensure end-of-training competence.

Solid validity evidence for two tools assessing competences in musculoskeletal ultrasound: a validity study.

OBJECTIVES: Musculoskeletal ultrasound (MSUS) is increasingly used by rheumatologists in daily clinical practice. However, MSUS is only valuable in trained hands, and assessment of trainee competences is therefore essential before independent practice. Thus, this study aimed to establish validity evidence for the EULAR and the Objective Structured Assessment of Ultrasound Skills (OSAUS) tools used for assessing MSUS competences. METHODS: Thirty physicians with different levels of MSUS experience (novices, intermediates, and experienced) performed four MSUS examinations of different joint areas on the same rheumatoid arthritis patient. All examinations were video recorded (n = 120), anonymized, and subsequently assessed in random order by two blinded raters using first the OSAUS assessment tool followed by the EULAR tool 1 month after. RESULTS: The inter-rater reliability between the two raters was high for both the OSAUS and EULAR tools, with a Pearson correlation coefficient (PCC) of 0.807 and 0.848, respectively. Both tools demonstrated excellent inter-case reliability, with a Cronbach's alpha of 0.970 for OSAUS and 0.964 for EULAR. Furthermore, there was a strong linear correlation between the OSAUS and the EULAR performance scores and the participants' experience levels (R2 = 0.897 and R2 = 0.868, respectively) and a significant discrimination between different MSUS experience levels (P < 0.001 for both). CONCLUSIONS: MSUS operator competences can be assessed reliably and valid using either the OSAUS or the EULAR assessment tool, thereby allowing a uniform competency-based MSUS education in the future. Although both tools demonstrated high inter-rater reliability, the EULAR tool was superior to OSAUS. TRIAL REGISTRATION: ClinicalTrials.gov, http://clinicaltrials.gov, NCT05256355.

Cytochrome P450 activity in rheumatoid arthritis patients during continuous IL-6 receptor antagonist therapy.

BACKGROUND: Inflammation suppresses cytochrome P450 (CYP) enzyme activity, and single-dose interleukin 6 receptor antagonists (anti-IL-6R) reverse this effect. Here, we assess the impact of continuous anti-IL-6R therapy in patients with rheumatoid arthritis. METHODS: In a clinical pharmacokinetic trial, the Basel cocktail was administered before and after 3 and 12 weeks of anti-IL-6R therapy to assess CYP enzyme activity (registered in the ClinicalTrials.gov database (identifier NCT04842981) on April 13th, 2021). In a retrospective study, the 4ß-hydroxycholesterol/cholesterol ratio was measured as a biomarker for CYP3A4 activity before and after 3 and 6 months of anti-IL-6R therapy. The control group was patients initiating a tumor necrosis factor alfa (TNF-α) inhibitor. RESULTS: In the clinical pharmacokinetic trial (n = 3), midazolam metabolic ratio (CYP3A4) was inconclusive due to the limited sample size. Midazolam AUC and Cmax indicate a weak impact on CYP3A4 activity after 3 weeks of anti-IL-6R therapy compared to baseline (AUC geometric mean ratio (GMR): 0.80, 95% CI: 0.64-0.99 and Cmax GMR: 0.58, 95% CI: 0.37-0.91), which returns to baseline levels after 12 weeks of therapy (AUC GMR 1.02, 95% CI: 0.72-1.46 and Cmax GMR 1.03, 95% CI 0.72-1.47). No effect on the 4ß-hydroxycholesterol/cholesterol ratio was observed in the retrospective study. CONCLUSION: Based on sparse data from three patients, continuous anti-IL-6R therapy seems to cause an acute but transient increase in CYP3A4 activity in rheumatoid arthritis patients, which may be due to a normalization of the inflammation-suppressed CYP activity. Further studies are warranted to understand the mechanism behind this putative transient effect. Trial registration Registered in the ClinicalTrials.gov database (identifier NCT04842981) on April 13th, 2021.

The impact of an ultrasound atlas for scoring salivary glands in primary Sjögren's syndrome: a pilot study.

The objective of this pilot study was to assess the impact of a salivary gland ultrasound (SGUS) atlas for scoring parenchymal changes in Sjögren's syndrome by assessing the reliability of the scoring system (0-3), without and with the use of the SGUS atlas. Ten participants with varying experience in SGUS contributed to the reliability exercise. Thirty SGUS images of the submandibular and parotid gland with abnormalities ranging from 0 to 3 were scored using the written definitions of the OMERACT SGUS scoring system and using the SGUS atlas based on the OMERACT scoring system. For intra-reader reliability, two rounds were performed without and with the atlas-in the 2nd round the 30 images were rearranged in random order by a physician not included in the scoring. Inter-reader reliability was also determined in both rounds. Without using the atlas, the SGUS OMERACT scoring system showed fair inter-reader reliability in round 1 (mean kappa 0.36; range 0.06-0.69) and moderate intra-reader reliability (mean kappa 0.55; range 0.28-0.81). With the atlas, inter-reader reliability improved in round 1 to moderate (mean kappa 0.52; range 0.31-0.77) and intra-reader reliability to good (mean kappa 0.69; range 0.46-0.86). Higher intra-reader reliability was noted in participants with previous SGUS experience. The SGUS atlas increased both intra- and inter-reader reliability for scoring gland pathology in participants with varying SGUS experience suggesting a possible future role in clinical practice and trials. Key Points • Ultrasonography can detect parenchymal changes in salivary glands in patients with Sjögren's disease. • An ultrasound atlas may improve reliability of scoring parenchymal changes in salivary glands.

Certolizumab pegol, abatacept, tocilizumab or active conventional treatment in early rheumatoid arthritis: 48-week clinical and radiographic results of the investigator-initiated randomised controlled NORD-STAR trial.

BACKGROUND: The optimal first-line treatment in early rheumatoid arthritis (RA) is debated. We compared clinical and radiographic outcomes of active conventional therapy with each of three biological treatments with different modes of action. METHODS: Investigator-initiated, randomised, blinded-assessor study. Patients with treatment-naïve early RA with moderate-severe disease activity were randomised 1:1:1:1 to methotrexate combined with (1) active conventional therapy: oral prednisolone (tapered quickly, discontinued at week 36) or sulfasalazine, hydroxychloroquine and intra-articular glucocorticoid injections in swollen joints; (2) certolizumab pegol; (3) abatacept or (4) tocilizumab. Coprimary endpoints were week 48 Clinical Disease Activity Index (CDAI) remission (CDAI ≤2.8) and change in radiographic van der Heijde-modified Sharp Score, estimated using logistic regression and analysis of covariance, adjusted for sex, anticitrullinated protein antibody status and country. Bonferroni's and Dunnet's procedures adjusted for multiple testing (significance level: 0.025). RESULTS: Eight hundred and twelve patients were randomised. Adjusted CDAI remission rates at week 48 were: 59.3% (abatacept), 52.3% (certolizumab), 51.9% (tocilizumab) and 39.2% (active conventional therapy). Compared with active conventional therapy, CDAI remission rates were significantly higher for abatacept (adjusted difference +20.1%, p<0.001) and certolizumab (+13.1%, p=0.021), but not for tocilizumab (+12.7%, p=0.030). Key secondary clinical outcomes were consistently better in biological groups. Radiographic progression was low, without group differences.The proportions of patients with serious adverse events were abatacept, 8.3%; certolizumab, 12.4%; tocilizumab, 9.2%; and active conventional therapy, 10.7%. CONCLUSIONS: Compared with active conventional therapy, clinical remission rates were superior for abatacept and certolizumab pegol, but not for tocilizumab. Radiographic progression was low and similar between treatments. TRIAL REGISTRATION NUMBER: NCT01491815.

E-learning and practical performance in musculoskeletal ultrasound: a multicentre randomized study.

OBJECTIVES: To examine the effect of pre-course e-learning on residents' practical performance in musculoskeletal ultrasound (MSUS). METHODS: This was a multicentre, randomized controlled study following the Consolidated Standards of Reporting Trials (CONSORT) statement. Residents with no or little MSUS experience were randomized to either an e-learning group or a traditional group. One week before a 2-day face-to-face MSUS course, the e-learning group received access to an interactive platform consisting of online lectures, assignments, and practical instruction videos aligned with the content of the course. The traditional group only received standard pre-course information (program, venue, and time). All participants performed a pre- and post-course practical MSUS examination and were assessed by two individual raters, blinded to the group allocation, using the validated Objective Structured Assessment of Ultrasound Skills (OSAUS) tool. RESULTS: Twenty-eight participants completed the study. There were no statistically significant differences in the pre- or post-course practical MSUS performance between the e-learning group and the traditional group; the mean pre-course OSAUS score (s.d.) in the -learning group was 5.4 (3.7) compared with 5.2 (2.4) in the traditional group (P = 0.8), whereas the post-course OSAUS score in the e-learning group was 11.1 (2.8) compared with 10.9 (2.4) in the traditional group (P = 0.8). There was a significant difference between the mean pre- and post-course scores (5.74 points, P < 0.001). The OSAUS assessment tool demonstrated good inter-rater reliability (intra-class correlation = 0.84). CONCLUSION: We found no significant impact of pre-course e-learning on novices' acquisition of practical MSUS skills. Hands-on training is of the utmost importance and improves MSUS performance significantly. The OSAUS assessment tool is an applicable tool with high interrater reliability. TRIAL REGISTRATION: https://clinicaltrials.gov/ NCT04959162.

Salivary gland ultrasound is associated with the presence of autoantibodies in patients with Sjögren's syndrome: A Danish single-centre study.

OBJECTIVES: To investigate whether ultrasound findings of major salivary glands are correlated with serological markers, autoantibodies, patient- or doctor-reported disease activity in a Danish cohort of patients with primary Sjögren's Syndrome (pSS). METHODS: In all, 49 patients at Odense University Hospital with pSS diagnosed according to the 2002 American-European Consensus Group (AECG) classification criteria were included. Patients were characterized using the EULAR Sjögren's Syndrome Disease Activity Index (ESSDAI, score of systemic complications) and EULAR Sjögren's Syndrome Patient Reported Index (ESSPRI), serologic markers, Schirmer's test and salivary test. Salivary gland ultrasound (SGUS) was performed of the submandibular and parotid glands and scored according to the Outcome Measures in Rheumatoid Arthritis Clinical Trials (OMERACT) semi-quantitative scoring system. RESULTS: More patients with abnormal SGUS had antinuclear antibodies (ANA) (p = 0.002), anti-Ro52 (p = 0.001), anti-Ro60 (p<0.001), anti-La (p<0.001) and IgM-RF (p<0.001). Titers for ANA (p = 0.02) and anti-Ro52 (p = 0.03) were higher in patients with abnormal SGUS. Twenty-three of the pSS patients had no pathological findings on SGUS. There was no correlation between SGUS severity and ESSDAI- or ESSPRI-scores. CONCLUSIONS: Abnormal SGUS findings are associated with autoantibodies of high specificity for pSS but not with ESSDAI, ESSPRI or inflammatory markers.

Infliximab biosimilar-to-biosimilar switching in patients with inflammatory rheumatic disease: clinical outcomes in real-world patients from the DANBIO registry.

OBJECTIVE: Successful uptake of biosimilars in rheumatology is limited by lack of real-world evidence regarding effectiveness of biosimilar-to-biosimilar switching. We investigated infliximab biosimilars CT-P13-to-GP1111 switching among patients with rheumatoid arthritis (RA), psoriatic arthritis (PsA) and axial spondyloarthritis (AxSpA). METHODS: Observational cohort study from the DANBIO registry. Patients were classified as originator-naïve or originator-experienced. Retention rates of 1-year GP1111 treatment were explored (Kaplan-Meier). We identified baseline factors (at the time of switch) associated with withdrawal of GP1111 (multivariable Cox-regression analyses with HRs including originator treatment history). Changes in subjective and objective measures of disease activity 4 months before and after the switch were assessed in individual patients. RESULTS: Of 1605 patients (685 RA, 314 PsA and 606 AxSpA, median disease duration was 9 years, 37% in Clinical Disease Activity Index/Ankylosing Spondylitis Disease Activity Score remission), 1171 were originator-naïve. Retention rates at 1-year were 83% (95% CI: 81% to 85%) and 92% (95% CI: 90% to 95%) for the originator-naïve and originator-experienced, respectively. GP1111 retention rates were higher in originator-experienced compared to originator-naïve with RA (HR=0.4 (95% CI: 0.2 to 0.7)) and PsA (HR=0.2 (95% CI: 0.1 to 0.8)), but not significantly for AxSpA: HR=0.6 (95% CI: 0.3 to 1.2). Lower disease activity was associated with higher retention. Changes in disease activity preswitch and postswitch were close to zero. CONCLUSION: This real-world observational study of more than 1600 patients with inflammatory arthritis showed high 1-year retention following a nationwide infliximab biosimilar-to-biosimilar switch. Retention was higher in originator-experienced and in patients with low disease activity, suggesting outcomes to be affected by patient-related rather than drug-related factors.

Ultrasound joint examination by an automated system versus by a rheumatologist: from a patient perspective.

BACKGROUND: The Arthritis Ultrasound Robot (ARTHUR) is an automated system for ultrasound scanning of the joints of both hands and wrists, with subsequent disease activity scoring using artificial intelligence. The objective was to describe the patient's perspective of being examined by ARTHUR, compared to an ultrasound examination by a rheumatologist. Further, to register any safety issues with the use of ARTHUR. METHODS: Twenty-five patients with rheumatoid arthritis (RA) had both hands and wrists examined by ultrasound, first by a rheumatologist and subsequently by ARTHUR. Patient-reported outcomes (PROs) were obtained after the examination by the rheumatologist and by ARTHUR. PROs regarding pain, discomfort and overall experience were collected, including willingness to be examined again by ARTHUR as part of future clinical follow-up. All ARTHUR examinations were observed for safety issues. RESULTS: There was no difference in pain or discomfort between the examination by a rheumatologist and by ARTHUR (p = 0.29 and p = 0.20, respectively). The overall experience of ARTHUR was described as very good or good by 92% (n = 23), with no difference compared to the examination by the rheumatologist (p = 0.50). All (n = 25) patients were willing to be examined by ARTHUR again, and 92% (n = 23) would accept ARTHUR as a regular part of their RA clinical follow up. No safety issues were registered. CONCLUSIONS: Joint ultrasound examination by ARTHUR was safe and well-received, with no difference in PRO components compared to ultrasound examination by a rheumatologist. Fully automated systems for RA disease activity assessment could be important in future strategies for managing RA patients. TRIAL REGISTRATION: The study was evaluated by the regional ethics committee (ID: S-20200145), which ruled it was not a clinical trial necessary for their approval. It was a quality assessment project, as there was no intervention to the patient. The study was hereafter submitted and registered to Odense University Hospital, Region of Southern Denmark as a quality assessment project and approved (ID: 20/55294).

EULAR points to consider for minimal reporting requirements in synovial tissue research in rheumatology.

BACKGROUND: Synovial tissue research has become widely developed in several rheumatology centres, however, large discrepancies exist in the way synovial tissue is handled and, more specifically, how data pertaining to biopsy procedure, quality check and experimental results are reported in the literature. This heterogeneity hampers the progress of research in this rapidly expanding field. In that context, under the umbrella of European Alliance of Associations for Rheumatology, we aimed at proposing points to consider (PtC) for minimal reporting requirements in synovial tissue research. METHODS: Twenty-five members from 10 countries across Europe and USA met virtually to define the key areas needing evaluation and formulating the research questions to inform a systematic literature review (SLR). The results were presented during a second virtual meeting where PtC were formulated and agreed. RESULTS: Study design, biopsy procedures, tissue handling, tissue quality control and tissue outcomes (imaging, DNA/RNA analysis and disaggregation) were identified as important aspects for the quality of synovial tissue research. The SLR interrogated four databases, retrieved 7654 abstracts and included 26 manuscripts. Three OPs and nine PtC were formulated covering the following areas: description of biopsy procedure, overarching clinical design, patient characteristics, tissue handling and processing, quality control, histopathology, transcriptomic analyses and single-cell technologies. CONCLUSIONS: These PtC provide guidance on how research involving synovial tissue should be reported to ensure a better evaluation of results by readers, reviewers and the broader scientific community. We anticipate that these PtC will enable the field to progress in a robust and transparent manner over the coming years.

Training and assessment of musculoskeletal ultrasound and injection skills-a systematic review.

OBJECTIVES: To examine how residents are trained and assessed in musculoskeletal US (MSUS), MSUS-guided and landmark-guided joint aspiration and injection. Additionally, to present the available assessment tools and examine their supporting validity evidence. METHODS: A systematic search of PubMed, Cochrane Library and Embase was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and studies published from 1 January 2000 to 31 May 2021 were included. Two independent reviewers performed the search and data extraction. The studies were evaluated using the Medical Education Research Quality Instrument (MERSQI). RESULTS: A total of 9884 articles were screened, and 43 were included; 3 were randomized studies, 21 pre- and post-test studies, 16 descriptive studies and 3 studies developing assessment tools. The studies used various theoretical training modalities, e.g. lectures, anatomical quizzes and e-learning. The practical training models varied from mannequins and cadavers to healthy volunteers and patients. The majority of studies used subjective 'comfort level' as assessment, others used practical examination and/or theoretical examination. All training programs increased trainees' self-confidence, theoretical knowledge, and/or practical performance, however few used validated assessment tools to measure the effect. Only one study met the MERSQI high methodical quality cut-off score of 14. CONCLUSION: The included studies were heterogeneous, and most were of poor methodological quality and not based on contemporary educational theories. This review highlights the need for educational studies using validated theoretical and practical assessment tools to ensure optimal MSUS training and assessment in rheumatology.

Ultrasound joint examination by an automated system versus by a rheumatologist: from a patient perspective

Abstract Background: The Arthritis Ultrasound Robot (ARTHUR) is an automated system for ultrasound scanning of the joints of both hands and wrists, with subsequent disease activity scoring using artificial intelligence. The objective was to describe the patient's perspective of being examined by ARTHUR, compared to an ultrasound examination by a rheumatologist. Further, to register any safety issues with the use of ARTHUR. Methods: Twenty-five patients with rheumatoid arthritis (RA) had both hands and wrists examined by ultrasound, first by a rheumatologist and subsequently by ARTHUR. Patient-reported outcomes (PROs) were obtained after the examination by the rheumatologist and by ARTHUR. PROs regarding pain, discomfort and overall experience were collected, including willingness to be examined again by ARTHUR as part of future clinical follow-up. All ARTHUR examinations were observed for safety issues. Results: There was no difference in pain or discomfort between the examination by a rheumatologist and by ARTHUR ( p =0.29 and p =0.20, respectively). The overall experience of ARTHUR was described as very good or good by 92% (n =23), with no difference compared to the examination by the rheumatologist ( p =0.50). All (n =25) patients were willing to be examined by ARTHUR again, and 92% (n =23) would accept ARTHUR as a regular part of their RA clinical follow up. No safety issues were registered. Conclusion: Joint ultrasound examination by ARTHUR was safe and well-received, with no difference in PRO components compared to ultrasound examination by a rheumatologist. Fully automated systems for RA disease activity assessment could be important in future strategies for managing RA patients. Trial registration: The study was evaluated by the regional ethics committee (ID: S-20200145), which ruled it was not a clinical trial necessary for their approval. It was a quality assessment project, as there was no intervention to the patient. The study was hereafter submitted and registered to Odense University Hospital, Region of Southern Denmark as a quality assessment project and approved (ID: 20/55294).

In vitro effect of biological and conventional disease-modifying antirheumatic drugs on fibrocyte differentiation in patients with rheumatoid arthritis and healthy controls.

OBJECTIVE: Fibrocytes are circulating bone-marrow-derived cells that migrate to organs with ongoing repair or inflammation. In the target organ, the cells differentiate, become long and spindle-shaped, and are able to produce extracellular matrix components. In fibrotic diseases, the levels of fibrocytes are increased, both in circulation and the diseased tissue. In rheumatoid arthritis (RA), fibrocytes have been proposed to be involved in the spread of the disease and possibly in RA fibrotic manifestations, as can be seen in RA interstitial lung disease (RA-ILD). Therefore, we aimed to investigate a range of current RA treatment modalities (corticosteroids and conventional and biological disease-modifying antirheumatic drugs (DMARDs)) regarding their effect on in vitro fibrocyte differentiation. METHODS: A total of 10 participants were included (5 patients with RA and 5 healthy controls). Peripheral blood mononuclear cells (PBMCs) were isolated and cultured for 5 days with prednisolone, conventional DMARDs (methotrexate, sulfasalazine, and hydroxychloroquine), and biological DMARDs (etanercept, tocilizumab, adalimumab, abatacept, and rituximab). The numbers of fibrocytes were counted. Dose-response data for abatacept and tocilizumab were collected. RESULTS: Abatacept and prednisolone significantly suppressed differentiation of PBMC into fibrocytes compared with control (p=0.02 and p<0.01, respectively) (n=10). In overall analysis (n=10), abatacept reduced fibrocyte levels with an average of 44% overall and 71% in the RA group compared with the control wells. Tocilizumab reduced the fibrocyte count by 63% overall and 45% in the RA group, although it was not significant (p=0.07 and p=0.06, respectively). Both tocilizumab and abatacept display a dose-response relationship. CONCLUSION: Abatacept and prednisolone suppress the differentiation of mononuclear cells to mature fibrocytes in vitro in patients with RA, and data indicate a similar effect of tocilizumab; this was further supported by the observed dose-response relationship. Clinical trials are needed to compare the effect of these drugs on fibrotic RA manifestations, for example, RA-ILD.

Deep learning detects and visualizes bleeding events in electronic health records.

BACKGROUND: Bleeding is associated with a significantly increased morbidity and mortality. Bleeding events are often described in the unstructured text of electronic health records, which makes them difficult to identify by manual inspection. OBJECTIVES: To develop a deep learning model that detects and visualizes bleeding events in electronic health records. PATIENTS/METHODS: Three hundred electronic health records with International Classification of Diseases, Tenth Revision diagnosis codes for bleeding or leukemia were extracted. Each sentence in the electronic health record was annotated as positive or negative for bleeding. The annotated sentences were used to develop a deep learning model that detects bleeding at sentence and note level. RESULTS: On a balanced test set of 1178   sentences, the best-performing deep learning model achieved a sensitivity of 0.90, specificity of 0.90, and negative predictive value of 0.90. On a test set consisting of 700 notes, of which 49 were positive for bleeding, the model achieved a note-level sensitivity of 1.00, specificity of 0.52, and negative predictive value of 1.00. By using a sentence-level model on a note level, the model can explain its predictions by visualizing the exact sentence in a note that contains information regarding bleeding. Moreover, we found that the model performed consistently well across different types of bleedings. CONCLUSIONS: A deep learning model can be used to detect and visualize bleeding events in the free text of electronic health records. The deep learning model can thus facilitate systematic assessment of bleeding risk, and thereby optimize patient care and safety.

Safety and efficacy of faecal microbiota transplantation for active peripheral psoriatic arthritis: an exploratory randomised placebo-controlled trial.

OBJECTIVES: Although causality remains to be established, targeting dysbiosis of the intestinal microbiota by faecal microbiota transplantation (FMT) has been proposed as a novel treatment for inflammatory diseases. In this exploratory, proof-of-concept study, we evaluated the safety and efficacy of FMT in psoriatic arthritis (PsA). METHODS: In this double-blind, parallel-group, placebo-controlled, superiority trial, we randomly allocated (1:1) adults with active peripheral PsA (≥3 swollen joints) despite ongoing treatment with methotrexate to one gastroscopic-guided FMT or sham transplantation into the duodenum. Safety was monitored throughout the trial. The primary efficacy endpoint was the proportion of participants experiencing treatment failure (ie, needing treatment intensification) through 26 weeks. Key secondary endpoints were change in Health Assessment Questionnaire Disability Index (HAQ-DI) and American College of Rheumatology (ACR20) response at week 26. RESULTS: Of 97 screened, 31 (32%) underwent randomisation (15 allocated to FMT) and 30 (97%) completed the 26-week clinical evaluation. No serious adverse events were observed. Treatment failure occurred more frequently in the FMT group than in the sham group (9 (60%) vs 3 (19%); risk ratio, 3.20; 95% CI 1.06 to 9.62; p=0.018). Improvement in HAQ-DI differed between groups (0.07 vs 0.30) by 0.23 points (95% CI 0.02 to 0.44; p=0.031) in favour of sham. There was no difference in the proportion of ACR20 responders between groups (7 of 15 (47%) vs 8 of 16 (50%)). CONCLUSIONS: In this first preliminary, interventional randomised controlled trial of FMT in immune-mediated arthritis, we did not observe any serious adverse events. Overall, FMT appeared to be inferior to sham in treating active peripheral PsA. TRIAL REGISTRATION NUMBER: NCT03058900.

Fibrocytes in early and long-standing rheumatoid arthritis: a 6-month trial with repeated synovial biopsy, imaging and lung function test.

OBJECTIVES: To correlate the level of fibrocytes in peripheral blood, synovial tissue and in vitro culture in rheumatoid arthritis (RA) with changes in disease activity, imaging and pulmonary function. METHODS: Twenty patients with early RA (ERA) and 20 patients with long-standing RA (LRA) were enrolled in a 6-month prospective study. Sixteen patients undergoing wrist arthroscopy were healthy controls. Patients with RA underwent pulmonary function tests, ultrasound and synovial ultrasound-guided needle biopsy of the same wrist at baseline and 6 months. Wrist MRI was performed at baseline (all) and 6 months (ERA). Circulating fibrocytes were measured by flow cytometry, in vitro by the number of monocytes that were differentiated to fibrocytes and in synovial biopsies by counting in histological sections. RESULTS: Fibrocytes were primarily located around vessels and in the subintimal area in the synovium. Fibrocyte levels did not decline during the trial despite effective RA treatment. In the ERA group, increased synovitis assessed by ultrasound was moderate and strongly correlated with an increase in circulating and synovial fibrocyte levels, respectively. Increased synovitis assessed by MRI during the trial in the ERA group was moderately correlated with both increased numbers of circulating and cultured fibrocytes. Absolute diffusion capacity level was overall weakly negatively correlated with the level of circulating and synovial fibrocytes. The decline in diffusion capacity during the trial was moderately correlated with increased levels of synovial fibrocytes. CONCLUSION: Our findings suggest that fibrocytes are involved in RA pathogenesis, both in the synovium and the reduction in lung function seen in a part of patients with RA. TRIAL REGISTRATION NUMBER: NCT02652299.

Active conventional treatment and three different biological treatments in early rheumatoid arthritis: phase IV investigator initiated, randomised, observer blinded clinical trial.

OBJECTIVE: To evaluate and compare benefits and harms of three biological treatments with different modes of action versus active conventional treatment in patients with early rheumatoid arthritis. DESIGN: Investigator initiated, randomised, open label, blinded assessor, multiarm, phase IV study. SETTING: Twenty nine rheumatology departments in Sweden, Denmark, Norway, Finland, the Netherlands, and Iceland between 2012 and 2018. PARTICIPANTS: Patients aged 18 years and older with treatment naive rheumatoid arthritis, symptom duration less than 24 months, moderate to severe disease activity, and rheumatoid factor or anti-citrullinated protein antibody positivity, or increased C reactive protein. INTERVENTIONS: Randomised 1:1:1:1, stratified by country, sex, and anti-citrullinated protein antibody status. All participants started methotrexate combined with (a) active conventional treatment (either prednisolone tapered to 5 mg/day, or sulfasalazine combined with hydroxychloroquine and intra-articular corticosteroids), (b) certolizumab pegol, (c) abatacept, or (d) tocilizumab. MAIN OUTCOME MEASURES: The primary outcome was adjusted clinical disease activity index remission (CDAI≤2.8) at 24 weeks with active conventional treatment as the reference. Key secondary outcomes and analyses included CDAI remission at 12 weeks and over time, other remission criteria, a non-inferiority analysis, and harms. RESULTS: 812 patients underwent randomisation. The mean age was 54.3 years (standard deviation 14.7) and 68.8% were women. Baseline disease activity score of 28 joints was 5.0 (standard deviation 1.1). Adjusted 24 week CDAI remission rates were 42.7% (95% confidence interval 36.1% to 49.3%) for active conventional treatment, 46.5% (39.9% to 53.1%) for certolizumab pegol, 52.0% (45.5% to 58.6%) for abatacept, and 42.1% (35.3% to 48.8%) for tocilizumab. Corresponding absolute differences were 3.9% (95% confidence interval -5.5% to 13.2%) for certolizumab pegol, 9.4% (0.1% to 18.7%) for abatacept, and -0.6% (-10.1% to 8.9%) for tocilizumab. Key secondary outcomes showed no major differences among the four treatments. Differences in CDAI remission rates for active conventional treatment versus certolizumab pegol and tocilizumab, but not abatacept, remained within the prespecified non-inferiority margin of 15% (per protocol population). The total number of serious adverse events was 13 (percentage of patients who experienced at least one event 5.6%) for active conventional treatment, 20 (8.4%) for certolizumab pegol, 10 (4.9%) for abatacept, and 10 (4.9%) for tocilizumab. Eleven patients treated with abatacept stopped treatment early compared with 20-23 patients in the other arms. CONCLUSIONS: All four treatments achieved high remission rates. Higher CDAI remission rate was observed for abatacept versus active conventional treatment, but not for certolizumab pegol or tocilizumab versus active conventional treatment. Other remission rates were similar across treatments. Non-inferiority analysis indicated that active conventional treatment was non-inferior to certolizumab pegol and tocilizumab, but not to abatacept. The results highlight the efficacy and safety of active conventional treatment based on methotrexate combined with corticosteroids, with nominally better results for abatacept, in treatment naive early rheumatoid arthritis. TRIAL REGISTRATION: EudraCT2011-004720-35, NCT01491815.

Applying cascaded convolutional neural network design further enhances automatic scoring of arthritis disease activity on ultrasound images from rheumatoid arthritis patients.

OBJECTIVES: We have previously shown that neural network technology can be used for scoring arthritis disease activity in ultrasound images from rheumatoid arthritis (RA) patients, giving scores according to the EULAR-OMERACT grading system. We have now further developed the architecture of this neural network and can here present a new idea applying cascaded convolutional neural network (CNN) design with even better results. We evaluate the generalisability of this method on unseen data, comparing the CNN with an expert rheumatologist. METHODS: The images were graded by an expert rheumatologist according to the EULAR-OMERACT synovitis scoring system. CNNs were systematically trained to find the best configuration. The algorithms were evaluated on a separate test data set and compared with the gradings of an expert rheumatologist on a per-joint basis using a Kappa statistic, and on a per-patient basis using a Wilcoxon signed-rank test. RESULTS: With 1678 images available for training and 322 images for testing the model, it achieved an overall four-class accuracy of 83.9%. On a per-patient level, there was no significant difference between the classifications of the model and of a human expert (p=0.85). Our original CNN had a four-class accuracy of 75.0%. CONCLUSIONS: Using a new network architecture we have further enhanced the algorithm and have shown strong agreement with an expert rheumatologist on a per-joint basis and on a per-patient basis. This emphasises the potential of using CNNs with this architecture as a strong assistive tool for the objective assessment of disease activity of RA patients.