In view of the outstanding progress of machine learning (ML) and growing cost of health systems, it is a current challenge to incorporate artificial intelligence tools into actual medical practice. Here we explored the feasibility and reliability of using machine learning to perform an important immunological investigation that currently requires experienced biologists : Anti-nuclear cytoplasmic antibodies (ANCAs) are important markers for vasculitis and they may be evidenced by microscopic examination of cells labeled with patients' sera. The use of a reliable ML classifier to discriminate between positive and negative samples would increase the rapidity and decrease the cost of immunofluorescence-based ANCA detection. Here, we tested seven well-documented ML algorithms, ranging from simple models such as decision trees to more complex convolutional neural networks involving millions of adjustable parameter. We studied the feasibility and reliability of classifying 1114 serum samples that had been collected for about 3 years and assayed with conventional procedure. We compared four strategies consisting of assaying either whole microscope fields or individual cell images, and natural images or histograms. The following conclusions were obtained : (i) Several different strategies allowed us to build models stable enough to discriminate between positive and negative samples collected during about 27 months, with a comparison to human classification yielding a kappa index of about 0.7, that may be considered as fairly good and intermediate between the performance of junior and senior biologists. (ii) Task-specific ML models combined with theoretical thinking might provide the most rapid and efficient way of developing a reliable test within the framework of a single institution. (iii) In addition, the interpretability of the simplest model provided some theoretical insight into important classification parameters. (iv) An important point and caveat is that the multiplicity and versatility of currently available tools make it an essential requirement to test repeatedly a given model, that must be chosen as simple as possible, to achieve a reliability compatible with medical use. It is concluded that our study provides a strong incentive to incorporate task-specific ML tools in well defined medical tests, which might reduce the risk of human errors and pave the way to fully automatic procedures.

Despite its wide use to treat various inflammatory diseases, infliximab becomes ineffective in some patients due to inadequate drug levels and production of anti-drug antibodies (ADA). The aim of this study was to compare the prevalence and ADA levels in a large cohort of patients. ADA and infliximab (IFX) through levels measured by enzyme-linked immunosorbent assay were collected from 505 patients within a period of 4 years. The results indicate that (i) 13.5% of patients produce ADA, (ii) male patients were more likely to produce ADA at levels above 10 000 ng/ml than female patients, (iii) ADA levels were lower when associated with immunosuppressant drugs, (iv) there was an inverse relationship between ADA presence and IFX detection, and (v) no correlation was observed between ADA levels and number of injections or brand of IFX administered. This study improves our understanding of the factors promoting IFX immunogenicity and highlights the need to develop personalized treatment strategies.

BACKGROUND AND PURPOSE:Systemic lupus erythematosus is an autoimmune, multisystemic disease affecting all organs in the body. Accrued evidence has elucidated a role for autophagy in the onset and severity of systemic lupus erythematosus. The antimalarial drug hydroxychloroquine constitutes the cornerstone of standard of care for systemic lupus erythematosus, together with glucocorticoids and immunosuppressants or biologics, and all accompanied by various side effects. EXPERIMENTAL APPROACH AND OBJECTIVE:Our study aimed to investigate the potential of GNS561 (ezurpimtrostat) treatment, a small basic lipophilic molecule that induces lysosomal dysregulation, using the pristane-induced lupus mouse model. KEY RESULTS:Compared with hydroxychloroquine, GNS561 treatment presents a more pronounced impact on the development of pathogenic anti-antibodies in pristane-induced lupus mice. Next, focussing on clinical impact, we showed that GNS561 significantly reduced clinical signs of lupus in pristane-induced lupus by preventing the incidence and severity of arthritis, occurrence of nephritis and lung damage. Finally, GNS561 modulated the inflammatory profile in pristane-induced lupus mice through a reduction of the lipogranuloma score. Interestingly, focussing on interferon-α, only pristane-induced lupus mice treated by GNS561 presented a significant decrease of the cytokine, suggesting a higher efficacy for GNS561 in the modulation of lupus-induced inflammation compared with hydroxychloroquine. CONCLUSION:All results show that GNS561, but not hydroxychloroquine, represents as an effective treatment to prevent clinical and inflammatory signs of lupus in this mouse model. IMPLICATIONS:Altogether, this study highlights GNS561 as a promising investigational drug for systemic lupus erythematosus.

INTRODUCTION:Usual interstitial pneumonia (UIP) is a key pattern of interstitial lung disease (ILD), most commonly linked to idiopathic pulmonary fibrosis (IPF). Identifying underlying autoimmune conditions such as myositis is clinically relevant, yet guidelines provide limited recommendations regarding myositis antibody (MSA) screening. METHODS:We retrospectively analyzed patients with UIP who underwent systematic MSA screening at our center. Clinical and serological characteristics were compared between MSA-positive (MSA+) and MSA-negative (MSA-) groups. Logistic regression was used to identify predictive factors. RESULTS:Among 134 patients, 15 (11 %) were MSA+. Compared with MSA- patients, MSA+ cases were more likely to present with autoimmune disease (p = 0.03), ANA ≥ 1:320 (p < 0.001), and positive rheumatoid factor (p = 0.04). A four-parameter profile combining ANA < 1:320, absence of rheumatoid factor, no hypergammaglobulinemia, and no Raynaud's phenomenon strongly predicted MSA negativity with excellent specificity (100 %), modest sensitivity (35 %), and good overall discriminative ability (AUC = 0.82). DISCUSSION:In patients with UIP, a simple four-parameter profile (ANA < 1:320, absence of rheumatoid factor, absence of hypergammaglobulinemia, and absence of Raynaud's phenomenon) strongly predicts negative MSA status. These findings support a more targeted approach to MSA screening, potentially improving diagnostic accuracy, guiding treatment decisions, and reducing unnecessary testing in UIP.

Anorexia nervosa (AN) is a severe, potentially life-threatening psychiatric disorder characterized by an intense fear of weight gain, a distorted body perception and an extern food restriction leading to an abnormally low body weight. In AN patients, malnutrition is often associated with hepatic cytolysis. A growing body of evidence support an association between AN and auto-immunity. In this exploratory study, we revealed for the first-time the presence of autoantibodies targeting hypoxia-inducible factor 1 alpha (HIF1a) in AN. We evidenced the presence of autoantibodies against HIF1a in 22% of AN patients, which were absent in patients with other metabolic disease as type-I diabetes or in healthy subjects. In addition, we found that HIF1a autoantibodies was associated with hepatic cytolysis in 80% of AN patients and that their levels significantly correlated with those of ALT (alanine aminotransferase). In-vitro experiments demonstrated that HIF1a autoantibodies induced hepatocyte lysis, suggesting a potential link between these autoantibodies and liver dysfunction in AN patients. Altogether, our data reveal an implication of HIF1a autoantibodies in AN with hepatic cytolysis. Future investigation will explore their potential as a diagnostic or a prognostic marker in AN, but also other diseases since dysregulated hypoxia pathway has been implicated in other pathologic conditions.

Objectives To assess the prevalence and clinical significance of autoantibodies against lysobisphophatidic acid (aLBPA) in patients with APS. Methods We conducted a retrospective analysis involving 91 patients with persistent conventional aPLs: 60 patients with at least one clinical event of APS (symptomatic group) and 31 without (asymptomatic group), as well as 33 aPL-negative controls. Detection of aLBPA in serum samples was performed using an enzyme-linked immunosorbent assay (ELISA) specifically designed for this study. Results The prevalence of aLBPA is significantly higher in patients with persistent aPL than that of the control group (P < 0.0001). Among patients with persistent aPL, our findings reveal a significantly higher prevalence of aLBPA in asymptomatic patients compared with their symptomatic counterparts (P = 0.027). Notably, patients positive for IgG aPL alone demonstrated a greater likelihood of presenting clinical events suggestive of APS. Conclusion The combined assay of aLBPA and conventional aPL could be used to stratify patients with persistent aPL. This combined approach could serve as a valuable tool in the management of this complex autoimmune disease, particularly in guiding decisions regarding the initiation of primary thromboprophylaxis in asymptomatic patients with persistent aPL.

OBJECTIVES:This review aims to identify biological markers associated with the risk of recurrence of thrombotic and/or obstetric events in patients with antiphospholipid syndrome (APS). METHODS:A comprehensive review of literature was conducted to evaluate established and potential novel biological markers associated with thrombosis in APS. To this end, a PubMed literature search was conducted for the last twenty years using the following keywords or their combinations: thrombotic risk, recurrence of thrombosis, risk stratification, severity, predictive value. RESULTS:Previous studies showed that multiple aPL positivity correlates with an increased risk of thrombosis in APS. Moreover, the analysis of N-glycosylation of antiphospholipid antibodies (aPL) revealed that low levels of IgG sialylation, fucosylation or galactosylation increases the pro-inflammatory activity of aPL, predisposing to thrombosis. In addition, quantification of neutrophil extracellular traps (NETs) and antibodies directed against NETs (anti-NETs) in serum demonstrates promising prognostic utility in assessing APS severity. Oxidative stress plays a role in the pathogenicity of APS and paraoxonase 1 (PON1) activity emerges as a promising biomarker of thrombotic risk in APS. Furthermore, identification of novel antigenic targets involved in the pathophysiology of APS, such as lysobisphosphatidic acid (LBPA), had led to the discovery of unconventional aPL, antibodies directed against the LBPA (aLBPA), whose clinical value could make it possible to identify APS patients at high risk of thrombotic recurrence. CONCLUSION:The immunological profile of aPL, N-glycosylation of aPL, quantification of NETs and anti-NETs, analysis of biomarkers of oxidative stress and the discovery of aLBPA offer potential prognostic tools for risk stratification in APS patients.

RésuméMême si ces dernières années la définition du syndrome des antiphospholipides (SAPL) n’a pas subi de changements majeurs, certaines connaissances ont évolué au cours du temps. Dans cet article, nous présenterons les nouveautés en termes de critères de classification, physiopathologie, évaluation du risque prothrombotique, cibles antigéniques et enfin traitements potentiels.

The success of artificial intelligence and machine learning is an incentive to develop new algorithms to increase the rapidity and reliability of medical diagnosis. Here we compared different strategies aimed at processing microscope images used to detect anti-neutrophil cytoplasmic antibodies, an important vasculitis marker: (i) basic classifier methods (logistic regression, k-nearest neighbors and decision tree) were used to process custom-made indices derived from immunofluorescence images yielded by 137 sera. (ii) These methods were combined with dimensional reduction to analyze 1733 individual cell images. (iii) More complex models based on neural networks were used to analyze the same dataset. The efficiency of discriminating between positive and negative samples and different fluorescence patterns was quantified with Rand-type accuracy index, kappa index and ROC curve. It is concluded that basic models trained on a limited dataset allowed for positive/negative discrimination with an efficiency comparable to that obtained by conventional analysis performed by humans (0.84 kappa score). More extensive datasets and more sophisticated models may be required for efficient discrimination between fluorescence patterns generated by different auto-antibody species.

Introduction Apheresis allows the fast removal of autoantibodies in anti-glomerular basement membrane (anti-GBM) disease, and in severe antineutrophil cytoplasmic antibodies (ANCA)-associated vasculitis. The CINEVAS study tested whether immunoadsorption (IA) allowed a faster removal of ANCA and/or anti-GBM antibodies than plasma exchanges (PEx). Methods CINEVAS was a prospective multicenter study comparing IA to PEx in consecutive patients with ANCA and/or anti-GBM vasculitides. The primary objective was the reduction rate in autoantibody titers between the beginning of the first and the end of the seventh apheresis session. Secondary objectives were number of sessions needed to obtain desired reduction rates; reduction rates of total Ig levels; tolerance of sessions; and patients’ outcome. Results The results of 38 patients (16 treated with IA and 22 with PEx), and 43 autoantibodies, were analyzed. There was no difference in the reduction rates in autoantibody titers between IA and PEx over 7 sessions (respectively 98% vs. 96%, P = 0.39). The numbers of sessions needed to obtain undetectable autoantibodies, or 50%, 75%, or 90% reductions, did not differ between techniques. Greater reduction rates of autoantibodies were observed when plasma was separated by filtration compared to centrifugation, with IA and PEx. IA allowed a greater reduction in total IgG levels, and better preservation of total IgA and IgM levels than PEx. PEx sessions required higher volumes of plasma, IA sessions higher volumes of citrate; IA sessions were longer. Conclusions IA and PEx were comparable in ANCA or anti-GBM removal kinetics, despite a faster reduction in total IgG with IA.

Background: The positivity of anti-RNP autoantibodies as biological criteria for the diagnosis of mixed connective tissue disease (MCTD) has recently divided the rheumatology community. Autoantigenicity of the U1-snRNP complex tends to generate multiple autoantibodies against RNP-A, -C and -70 KDa or Sm proteins. The aim of this study is to identify the most informative autoantibodies in clinical practice, in particular, to contribute to differential diagnosis between MCTD and systemic lupus erythematosus (SLE). Methods: Sera from 74 patients positive for anti-RNP autoantibodies were selected over a period of one year of laboratory practice. Autoantibodies directed against extractable nuclear antigen, RNP proteins (A, C, 70 KDa) and 40 kDa fragments of RNP-70 KDa were investigated by using quantitative fluoroenzymatic assay and Western blot analysis. Results: Among the 74 patients, 40 patients were diagnosed with SLE, 20 with MCTD, six with another autoimmune disease, three with SARS-CoV-2 infection, three with cancer and two were healthy. No preferential clinical association of IgG or IgM autoantibodies directed against each of the RNP proteins was found between SLE and MCTD. In contrast, the proportion of autoantibodies directed against the RNP component within the U1-snRNP complex showed a significantly higher RNP index in patients with MCTD than in those with SLE (p = 0.011), with good performance (sensitivity: 69.2%, specificity: 88.9%). Conclusions: The analysis of the proportion of the different autoantibodies directed against the U1-snRNP complex is more informative than the analysis of each autoantibody separately. A follow-up of patients could be informative about the interest of the RNP index as a predictor of disease evolution.

COVID-19 has led to more than 5 million deaths to date.[1] COVID-19 presentation can range from a simple asymptomatic viral infection to acute respiratory distress syndrome (ARDS)[2]; 15 to 20% of patients develop severe pneumonia and experience coagulopathy disorders. Severe COVID-19 occurs in the context of hyperinflammation that could potentially generate autoimmune disease.[3] We and others have shown the presence of antiphospholipid (aPL) autoantibodies in patients with COVID-19.[4] However, there remains contention if these aPLs represent transient antibodies associated with infection, or “true” aPL reflective of potential autoimmune-associated coagulopathy, namely, antiphospholipid syndrome (APS).[5] [6] Thus, only if they can be shown to persist for at least 12 weeks, can they be considered essential markers for the diagnosis of APS.[7] [8]

The detection of anti-phosphatidylethanolamine autoantibodies (aPEs) has been proposed to improve the diagnosis and management of patients presenting clinical manifestations of antiphospholipid syndrome (APS), such as thrombosis, and who are persistently negative for conventional markers. After selecting the most specific ELISA for their detection, we evidenced the interest of aPEs in the exploration of thrombosis when APS conventional markers were negative through a 1-year retrospective study including 1131 consecutive patients routinely tested for aPEs. To validate this result, we assessed aPEs in a newly selected population of 77 patients with unexplained deep vein thrombosis (DVT). With a total prevalence of 19.5%, we confirmed the interest of aPE detection in patients with unexplained DVT who were devoid of other aPLs markers. Since endosomal compartment, a source of ROS production, has been recently identified as the cellular target of aPEs in vitro, we then investigated an association between aPE positivity and reactive oxygen species (ROS) production by measuring the production of thiobarbituric acid-reactive substances. We showed, for the first time, a significant association between aPE positivity and systemic ROS production in patients which led us to hypothesize a new mechanism of action of aPEs in thrombosis through a signaling related to oxidative stress.

Abstract Background Hypocomplementemic urticarial vasculitis (HUV) is a rare systemic vasculitis. We aimed to describe the kidney involvement of HUV in a multicenter national cohort with an extended follow-up. Methods All patients with HUV (international Schwartz criteria) with a biopsy-proven kidney involvement, identified through a survey of the French Vasculitis Study Group (FVSG), were included. A systematic literature review on kidney involvement of HUV was performed. Results Twelve patients were included, among whom 8 had positive anti-C1q antibodies. All presented with proteinuria, from mild to nephrotic, and 8 displayed acute kidney injury (AKI), requiring temporary haemodialysis in 2. Kidney biopsy showed membrano-proliferative glomerulonephritis (MPGN) in 8 patients, pauci-immune crescentic GN or necrotizing vasculitis in 3 patients (with a mild to severe interstitial inflammation), and an isolated interstitial nephritis in 1 patient. C1q deposits were observed in the glomeruli (n = 6), tubules (n = 4) or renal arterioles (n = 3) of 8 patients. All patients received corticosteroids, and 9 were also treated with immunosuppressants or apheresis. After a mean follow-up of 8.9 years, 6 patients had a preserved renal function, but 2 patients had developed stage 3–4 chronic kidney disease (CKD) and 4 patients had reached end-stage kidney disease (ESKD), among whom 1 had received a kidney transplant. Conclusion Renal involvement of HUV can be responsible for severe AKI, CKD and ESRD. It is not always associated with circulating anti-C1q antibodies. Kidney biopsy shows mostly MPGN or crescentic GN, with frequent C1q deposits in the glomeruli, tubules or arterioles.

La néphropathie lupique est une glomérulonéphrite à dépôts de complexes immuns dans laquelle l’activation du complément joue un rôle clé.

Whereas the detection of antiphospholipid autoantibodies (aPL) in COVID-19 is of increasing interest, their role is still unclear. We analyzed a large aPL panel in 157 patients with COVID-19 according to the disease severity. We also investigated a potential association between aPL and extracellular DNA (exDNA, n = 85) or circulating markers of neutrophil extracellular traps (NET) such as citrullinated histones H3 (CitH3, n = 49). A total of 157 sera of patients infected by SARS-CoV-2 were collected. A large aPL panel including lupus anticoagulant, anti-cardiolipin and anti-beta-2 glycoprotein I (IgG, IgM and IgA), anti-phosphatidylethanolamine IgA, anti-prothrombin (IgG and IgM) was retrospectively analyzed according to the disease severity. We found a total aPL prevalence of 54.8% with almost half of the cases having aCL IgG. Within an extended panel of aPL, only aCL IgG were associated with COVID-19 severity. Additionally, severe patients displayed higher CitH3 levels than mild patients. Interestingly, we highlighted a significant association between the levels of aCL IgG and exDNA only in aCL positive patients with severe disease. In conclusion, we showed a significant link between aPL, namely aCL IgG, and circulating exDNA in patients with severe form of COVID-19, that could exacerbate the thrombo-inflammatory state related to disease severity.

Abstract Due to the role of macrophages in glomerular inflammation, macrophage surface molecules such as CD163 and CD11b represent attractive biomarkers for monitoring Lupus Nephritis (LN). We hypothesize that their urinary levels may reflect kidney disease activity. Here, we first analyzed the levels of urinary soluble CD163 (U-sCD163) and CD11b (U-sCD11b) in a cohort of 40 patients with LN including 23 with active disease. U-sCD163 levels were significantly elevated in active LN and correlated with the renal activity score, in contrast to U-sCD11b. Then, we developed the pristane induced LN mouse model to analyze, in a longitudinal way, the evolution of U-sCD163 levels according to the glomerular inflammation progression and response to treatment. We showed an increase of U-sCD163 levels associated with glomerular immune-complex deposits on mouse kidney biopsies at an early stage of the disease and a correlation with inflammatory markers including interferon-α, C-reactive protein and tumor necrosis factor-α. In addition, we showed that U-sCD163 levels decreased following efficient hydroxychloroquine treatment. Altogether our results led us to conclude that U-sCD163 represent a non-invasive biomarker for LN reflecting glomerular inflammation. Although prospective study in patients with LN, active or not, are necessary, U-sCD163 could be proposed to monitor response to treatment.