ObjectivesDiagnostic reasoning in systemic lupus erythematosus (SLE) is a complex process reflecting the probability of disease at a given timepoint against competing diagnoses. We applied machine learning in well-characterised patient data sets to develop an algorithm that can aid SLE diagnosis.MethodsFrom a discovery cohort of randomly selected 802 adults with SLE or control rheumatologic diseases, clinically selected panels of deconvoluted classification criteria and non-criteria features were analysed. Feature selection and model construction were done with Random Forests and Least Absolute Shrinkage and Selection Operator-logistic regression (LASSO-LR). The best model in 10-fold cross-validation was tested in a validation cohort (512 SLE, 143 disease controls).ResultsA novel LASSO-LR model had the best performance and included 14 variably weighed features with thrombocytopenia/haemolytic anaemia, malar/maculopapular rash, proteinuria, low C3 and C4, antinuclear antibodies (ANA) and immunologic disorder being the strongest SLE predictors. Our model produced SLE risk probabilities (depending on the combination of features) correlating positively with disease severity and organ damage, and allowing the unbiased classification of a validation cohort into diagnostic certainty levels (unlikely, possible, likely, definitive SLE) based on the likelihood of SLE against other diagnoses. Operating the model as binary (lupus/not-lupus), we noted excellent accuracy (94.8%) for identifying SLE, and high sensitivity for early disease (93.8%), nephritis (97.9%), neuropsychiatric (91.8%) and severe lupus requiring immunosuppressives/biologics (96.4%). This was converted into a scoring system, whereby a score >7 has 94.2% accuracy.ConclusionsWe have developed and validated an accurate, clinician-friendly algorithm based on classical disease features for early SLE diagnosis and treatment to improve patient outcomes.
Mental disorders such as anxiety and depression are prevalent in systemic lupus erythematosus (SLE) patients, yet their association with the underlying disease activity remains uncertain and has been mostly evaluated at a cross-sectional level. To examine longitudinal trends in anxiety, depression, and lupus activity, a prospective observational study was performed on 40 adult SLE outpatients with active disease (SLE Disease Activity Index [SLEDAI]-2K ≥ 3 [excluding serology]) who received standard-of-care. Anxiety and depression were determined at baseline and 6 months by the Hospital Anxiety and Depression Scale. Treatment adherence was assessed with the Morisky Medication Adherence Scale-4. Increased anxiety (median [interquartile range] HADS-A: 11.0 [7.8]) and depression (HADS-D: 8.0 [4.8]) were found at inclusion, which remained stable and non-improving during follow-up (difference: 0.0 [4.8] and −0.5 [4.0], respectively) despite reduced SLEDAI-2K by 2.0 (4.0) (p < 0.001). Among possible baseline predictors, paid employment—but not disease activity—correlated with reduced HADS-A and HADS-D with corresponding standardized beta-coefficients of −0.35 (p = 0.017) and −0.27 (p = 0.093). Higher anxiety and depression correlated with lower treatment adherence (p = 0.041 and p = 0.088, respectively). These results indicate a high-mental disease burden in active SLE that persists despite disease control and emphasize the need to consider socioeconomic factors as part of comprehensive patient assessment.
ObjectiveTo obtain real-world data on outcomes of belimumab treatment and respective prognostic factors in patients with systemic lupus erythematosus (SLE).MethodsObservational study of 188 active SLE patients (median disease duration 6.2 years, two previous immunosuppressive/biological agents) treated with belimumab, who were monitored for SLEDAI-2K, Physician Global Assessment (PGA), LLDAS (lupus low disease activity state), remission (DORIS/Padua definitions), SELENA-SLEDAI Flare Index, SLICC/ACR damage index and treatment discontinuations. Group-based disease activity trajectories were modelled followed by multinomial regression for predictive variables. Drug survival was analysed by Cox-regression.ResultsAt 6, 12 and 24 months, LLDAS was attained by 36.2%, 36.7% and 33.5%, DORIS-remission by 12.3%, 11.6% and 17.8%, and Padua-remission by 21.3%, 17.9% and 29.0%, respectively (attrition-corrected). Trajectory analysis of activity indices classified patients into complete (25.5%), partial (42.0%) and non-responder (32.4%) groups, which were predicted by baseline PGA, inflammatory rash, leukopenia and prior use of mycophenolate. During median follow-up of 15 months, efficacy-related discontinuations occurred in 31.4% of the cohort, especially in patients with higher baseline PGA (hazard ratio [HR] 2.78 per 1-unit; 95% CI 1.32-5.85). Conversely, PGA improvement at 3 months predicted longer drug retention (HR 0.57; 95% CI 0.33-0.97). Use of hydroxychloroquine was associated with lower risk for safety-related drug discontinuation (HR 0.33; 95% CI 0.13-0.85). Although severe flares were reduced, flares were not uncommon (58.0%) and contributed to treatment stops (odds ratio [OR] 1.73 per major flare; 95% CI 1.09-2.75) and damage accrual (OR 1.83 per mild/moderate flare; 95% CI 1.15-2.93).ConclusionsIn a real-life setting with predominant long-standing SLE, belimumab was effective in the majority of patients, facilitating the achievement of therapeutic targets. Monitoring PGA helps to identify patients who will likely benefit and stay on the treatment. Vigilance is required for the prevention and management of flares while on belimumab.
Objective The proportion of SLE patients with residual disease activity in routine settings is variable. We assessed disease activity state in patients during their most recent visit, and whether patients with residual activity were offered therapy intensification. Methods Cross-sectional study of consecutive lupus patients in three tertiary centers. Patients were categorized as: i) remission off-therapy, ii) remission on-therapy, iii) low disease activity, and iv) non-optimally controlled disease. Multivariable regression identified factors associated with treatment intensification and ROC analysis calculated the accuracy of SLEDAI-2K to predict this intensification. Results Three hundred and thirty-two patients were included [93.1% female, mean (SD) age 48.5 (14.7) years, median (IQR) disease duration 6.5 (12.4) years]. Mean (SD) total and clinical SLEDAI at last visit were 3.7 (3.0) and 3.0 (2.9), respectively. Although 23.2% of patients were in remission, 40.1% were categorized as non-optimally controlled disease (79.2% due to SLEDAI-2K > 4), but less than 50% of them were offered therapy intensification. Proteinuria (OR 6.78, 95% CI 2.06–22.25), arthritis (OR 5.48, 95% CI 3.20–9.40), and rash (OR 3.23, 95% CI 1.81–5.75) were associated with intensification of therapy, but the accuracy of either total or clinical SLEDAI to predict it was moderate (ROC area under the curve 0.761 and 0.779, respectively). Conclusions About 40% of patients have evidence of residual disease activity and could qualify for novel treatments. Most treatment changes were triggered by active renal, joint, and skin disease, whereas the predictive value of SLEDAI-2K as a metric of disease activity was modest.
BackgroundBelimumab has been introduced in the management of SLE for more than 10 years, however long-term efficacy and safety data are still limited and mostly derive from the extended phase of randomized clinical trials.ObjectivesTo evaluate the long-term survival of belimumab treatment, reasons for treatment cessation and associated predictors in routine care setting.MethodsMulticentre observational study of adult SLE patients who were treated with belimumab according to physician discretion and in line with the EULAR recommendations. Disease activity (Physician Global Assessment [PGA]: scale 0-3; SLE disease activity index-2000 [S2K]), flares (SELENA-SLEDAI Flare Index), organ damage (SLICC damage index [SDI]), co-administered treatments and dosage, adverse events and causes of belimumab discontinuation were monitored prospectively at 3–6-month intervals. Cox-regression analysis was performed to identify factors associated with reduced drug survival.ResultsA total 184 patients treated with belimumab for at least 3 months were included (women 95.6%; mean ± SD age 48.8 ± 13.4 years; disease duration 9.2 ± 11.3 years). Baseline S2K and PGA were 7.5 ± 3.0 and 1.64 ± 0.42, respectively, both demonstrating significant improvement at 6 months (4.5 ± 3.5 and 1.02 ± 0.69, respectively; p<0.001) and 12 months (3.5 ± 3.1 and 0.68 ± 0.55, respectively; p<0.001). Of patients receiving glucocorticoids at onset, 49.0% tapered the dose and 17.6% completely withdrew them. After a median (interquartile range) follow-up of 15.1 (16.9) months, 44.0% of patients discontinued belimumab due to suboptimal efficacy as judged by the treating physician (28.3%), adverse events (including infections) (9.8%) or other causes (e.g., pregnancy, patient decision). Accordingly, efficacy-related drug survival rates at 1 and 2 years were 70% and 61%, respectively, with corresponding safety-related survival rates of 94% and 87%, respectively. Baseline factors associated with belimumab discontinuation due to suboptimal efficacy included PGA >1.50 (hazard ratio [HR] 3.66; 95% confidence interval [95% CI] 1.14–11.73; p=0.029) and severe (RA-like) arthritis (HR 2.56; 95% CI 1.16–5.68; p=0.020) but not disease duration, use of glucocorticoids, active serology or organ damage. Notably, patients with early (3 months) improvement (i.e., any decrease in PGA) showed significantly lower risk for treatment cessation (HR 0.38; 95% CI 0.22–0.67; p=0.001) (Figure 1) and this effect was independent of the initial PGA level. Baseline use of hydroxychloroquine was associated with prolonged safety-related belimumab survival (HR 0.32; 95% CI 0.12–0.88; p=0.028).Figure 1.Efficacy-related survival of belimumab according to improvement or not of PGA at 3 months since treatment initiation.ConclusionIn real-life setting, about 28% of SLE patients discontinue belimumab due to suboptimal treatment response per physician judgement, especially those with moderate-to-high activity and severe arthritis. Improvement in PGA at 3 months predicts long-term drug maintenance, therefore suggesting its value for patient monitoring. Our data confirm the very good tolerability of belimumab and identify hydroxychloroquine co-administration as a predictor for prolonged safety-related drug survival.AcknowledgementsThe study was partly funded by the Greek Rheumatology Society and the Greek Association of Professional Rheumatologists (ERE-EPERE) and by Pfizer Global Medical GrantsDisclosure of InterestsMyrto Nikoloudaki: None declared, Dionysis Nikolopoulos: None declared, SOFIA KOUTSOVITI: None declared, Irini Flouri: None declared, Noemin Kapsala: None declared, ARGYRO REPA: None declared, PELAGIA KATSIMPRI: None declared, EVANGELOS THEOTIKOS: None declared, Sofia Pitsigavdaki: None declared, Katerina Pateromichelaki: None declared, Anastasios Eskitzis: None declared, ANTONIA ELEZOGLOU: None declared, Prodromos Sidiropoulos: None declared, Antonis Fanouriakis: None declared, Dimitrios Boumpas: None declared, George Bertsias Speakers bureau: GSK, AstraZeneca, Pfizer, SOBI, UCB, Novartis, AENORASIS, Abbvie, Grant/research support from: GSK, Pfizer
BackgroundEarly or pre-clinical forms of lupus encompass a broad range of presentations, spanning from asymptomatic individuals with immunological abnormalities to individuals with autoantibodies and some features suggestive of SLE who do not yet meet the classification criteria. Research on this topic could reveal predictive and diagnostic biomarkers for individuals at-risk for progression to SLE.ObjectivesTo examine the rate of transition from at-risk to classified (ACR 1997 criteria) SLE, and identify demographic and clinical predictors. To prospectively evaluate the sensitivity and accuracy of the newer classification criteria (SLICC 2012, EULAR/ACR 2019) and the SLE Risk Predictive Index (SLERPI)[1] in patients at-risk who progress or not to classified SLE.MethodsThis is a single-centre analysis of individuals at-risk for SLE as part of an ongoing multicentric inception cohort study aiming to identify clinical, environmental and molecular prognostic factors for SLE onset. Enrolled individuals: a) were 18–55 years old; b) had clinical and/or serological features suggestive of SLE; c) had no clinical diagnosis of SLE or other autoimmune rheumatic disease; and d) did not fulfill the ACR 1997 classification criteria. Prospective monitoring at 6-month intervals was performed to determine accrual of classification and non-classification features, and ascertain the disease status (at-risk/undifferentiated connective tissue disease, SLE, other connective tissue disease).ResultsA total 124 subjects were included, all Whites, 94.4% women, with an average (standard deviation) age 36 (11) years. At first assessment, individuals fulfilled 2.25 (0.72) ACR 1997 criteria with ANA being the most prevalent feature (75.8%) followed by low complement (43.5%), arthritis (37.9%), photosensitivity (28.2%), malar rash (23.4%), and non-scarring alopecia (18.5%). After a median follow-up of 16 months, 27 participants (21.8%) fulfilled the ACR 1997 criteria, of whom 8 (6.5%) developed moderate or severe SLE. Multivariable-adjusted logistic regression identified anti-Ro/SSA (odds ratio [OR] 6.93; 95% confidence interval [95% CI] 1.75–27.5, p=0.006), combined low C3 and low C4 (OR 4.82; 95% CI 1.42–16.3, p=0.012) and photosensitivity (OR 3.25; 95% CI 1.17–8.99, p=0.023) as independent predictors for transition to classified SLE. The sensitivity of SLICC 2012, EULAR/ACR 2019 and SLERPI (>7) at baseline for detecting individuals who progressed to SLE (ACR 1997) was 40.7%, 25.9% and 40.7%, respectively, with corresponding specificities of 83.5%, 88.7% and 79.4%.ConclusionAmong individuals at-risk for SLE, about 20% may evolve into classified disease after a medium follow up of 16 months which is predominantly of mild severity. Presence of anti-Ro/SSA, hypocomplementemia, and photosensitivity indicate subjects who at increased risk for transition to SLE. Newer classification systems may capture as many as 40% of progressors with acceptable specificity.References[1]doi: 10.1136/annrheumdis-2020-219069AcknowledgementsThis work was funded by the Foundation for Research in Rheumatology (FOREUM).Disclosure of InterestsNone declared
BackgroundMental disorders such as anxiety and depression are highly prevalent in SLE patients,[1] yet their association with the underlying disease activity remains elusive and has been mostly evaluated at cross-sectional level.[2] This is further complicated by the often-increased rates of treatment non-adherence,[3] an important determinant of heightened lupus activity, among patients with depression.[4]ObjectivesTo examine the relationship between longitudinal changes in anxiety (ICD-10-CM F41.9), depression (ICD-10-CM F32.x) and disease activity levels in adult SLE patients. Second, to test the association between the aforementioned mental disorders with treatment adherence and sociodemographic factors.MethodsA prospective 6-month observational study of outpatients aged 18-65 years who fulfilled the EULAR/ACR 2019 classification criteria and had active disease ascertained by a SLEDAI-2K ≥3 and PGA (physician global assessment; scale 0–3) >1. Patients were enrolled by consecutive sampling technique during May-September 2021. Excluding criteria were overlap rheumatic diseases, active neuropsychiatric lupus, ongoing pregnancy or post-partum period, history of dementia or malignancy. Sociodemographic factors (age, disease duration, education level, working status) and comorbidities were collected. Anxiety and depression levels (assessed with the Hospital Anxiety and Depression Scale [HADS-A/D subscales]), disease activity (SLEDAI-2K, PGA), use of medications, and treatment adherence (Morisky Medication Adherence Scale-4 items scale) were monitored during the observation period.ResultsForty SLE patients (39 females) with an average [standard deviation] age 50.5 (10.3) years and disease duration 10.3 (7.0) years, were enrolled. Baseline SLEDAI-2K was 6.0 (2.0) driven predominantly from the musculoskeletal and mucocutaneous domains. The prevalence of anxiety (HADS-A >11) and depression (HADS-D >8) were 42.5% and 45.0%, respectively. During follow-up, disease activity was significantly reduced (average [SD] reduction in SLEDAI-2K: 1.90 [2.80], p<0.001), however, anxiety and depression levels remained unchanged (average [SD] change in HADS-A -0.05 [3.76] and HADS-D 0.53 [3.25], respectively, p>0.300 for both). Accordingly, Spearman’s non-parametric test showed that longitudinal changes in SLEDAI-2K were not significantly correlated with the corresponding changes in the HADS-A (rho = 0.13, p=0.417) or HADS-D (rho = -0.05, p=0.781) scores. Treatment non-adherence was found in 19 patients (47.5%) but did not correlate with anxiety and depression (p>0.500 for both). Notably, mental disorders were not significantly associated with comorbidities (including fibromyalgia) but unemployment status predicted the presence of anxiety (odds ratio 7.73, p-value 0.018).ConclusionAnxiety and depression are frequent comorbidities in active SLE and do not correlate with short-term disease improvement, thus underscoring the need for adjunct treatment. Physician awareness in the detection of treatment adherence is necessary. Larger studies in early disease and with longer follow-up will be required to further explore the possible interaction between of mental disorder and lupus disease course.References[1]Zhang L, et al. BMC Psychiatry. 2017;17(1).[2]Tay SH, et al. Lupus. 2015;24(13):1392–9.[3]Costedoat-Chalumeau N, et al. Clin Pharmacol Ther. 2018;103(6):1074–82.[4]Alsowaida N, et al. Lupus. 2018;27(2):327–32.Disclosure of InterestsNone declared
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
334 Leonard St
Brooklyn, NY 11211
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.