Elise Siegert
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Elise Siegert has a total of 606 co-authors. Here are the top co-authors by number of shared publications:
Gabriela Riemekasten
Clinic for Rheumatology and Clinical Immunology, University Hospital Schleswig-Holstein, University of Lübeck, Lübeck, Germany
16 shared publications- Gabriela Riemekasten
Department of Rheumatology and Clinical Immunology, University of Lübeck, Lübeck, Germany
13 shared publications - Oliver Distler
Department of Rheumatology University Hospital Zurich, University of Zurich Zurich Switzerland
12 shared publications Oliver Distler
Department of Rheumatology, University Hospital Zürich, Zürich, Switzerland
10 shared publicationsYannick Allanore
Cochin Hospital, Rheumatology A Department, Paris Descartes University, Paris, France
10 shared publicationsNicolas Hunzelmann
University Hospital Cologne
9 shared publications
Latest Publications
- Open Access
- 11 November 2026
- Published by Wiley in Arthritis & Rheumatology
Initial Treatment Strategies and Outcomes Across Hemodynamic Thresholds and Risk Stratification in Incident Systemic Sclerosis–Associated Pulmonary Arterial Hypertension: A EUSTAR Analysis
Objective: Real‐world initial treatment strategies in systemic sclerosis–associated pulmonary arterial hypertension (SSc‐PAH) remain incompletely described. We aimed to characterize these strategies across hemodynamic thresholds and risk strata and evaluate outcome associations. Methods: We analyzed treatment‐naive patients with SSc‐PAH from the European Scleroderma Trials and Research (EUSTAR) database (2000–2021) who fulfilled the 2022 hemodynamic definition. Initial therapy comprised treatment initiated within four months of right heart catheterization. As‐treated (AT) analyses were primary and intention‐to‐treat (ITT) analyses supportive. Outcomes were all‐cause mortality and 36‐month PAH progression. Results: Among 301 patients, 198 (66%) initiated therapy and 170 maintained therapy for at least three months. Therapy was initiated in 31% with milder hemodynamic impairment (mean pulmonary arterial pressure 21–24 mm Hg and/or pulmonary vascular resistance 2–3 Wood units) versus 79% fulfilling the previous hemodynamic definition; combination therapy increased from 14% to 67% across low‐ to high‐risk strata. Over a median 3.9 years, 125 patients (42%) died and 85 of 227 (37%) progressed. Adjusted ITT analyses showed no association with either outcome, whereas adjusted AT analyses showed lower mortality risk with sustained monotherapy and combination therapy (hazard ratio [HR] 0.50, 95% confidence interval [CI] 0.27–0.92, and HR 0.40, 95% CI 0.18–0.88, respectively) and lower PAH progression risk with sustained initial therapy (HR 0.37, 95% CI 0.18–0.76). In the milder hemodynamic subgroup, the mortality estimate was directionally consistent but imprecise. Conclusion: Initial treatment strategies reflected hemodynamic severity and baseline risk, whereas variation suggested potential implementation gaps. Sustained initial therapy was associated with lower risks of mortality and PAH progression. Contrasting ITT and AT estimates highlight challenges in observational treatment–outcome comparisons. Findings in the milder hemodynamic subgroup support dedicated prospective studies. image
- 11 November 2026
- Published by Wiley in Arthritis & Rheumatology
Open AccessInitial Treatment Strategies and Outcomes Across Hemodynamic Thresholds and Risk Stratification in Incident Systemic Sclerosis–Associated Pulmonary Arterial Hypertension: A EUSTAR Analysis
Objective: Real‐world initial treatment strategies in systemic sclerosis–associated pulmonary arterial hypertension (SSc‐PAH) remain incompletely described. We aimed to characterize these strategies across hemodynamic thresholds and risk strata and evaluate outcome associations. Methods: We analyzed treatment‐naive patients with SSc‐PAH from the European Scleroderma Trials and Research (EUSTAR) database (2000–2021) who fulfilled the 2022 hemodynamic definition. Initial therapy comprised treatment initiated within four months of right heart catheterization. As‐treated (AT) analyses were primary and intention‐to‐treat (ITT) analyses supportive. Outcomes were all‐cause mortality and 36‐month PAH progression. Results: Among 301 patients, 198 (66%) initiated therapy and 170 maintained therapy for at least three months. Therapy was initiated in 31% with milder hemodynamic impairment (mean pulmonary arterial pressure 21–24 mm Hg and/or pulmonary vascular resistance 2–3 Wood units) versus 79% fulfilling the previous hemodynamic definition; combination therapy increased from 14% to 67% across low‐ to high‐risk strata. Over a median 3.9 years, 125 patients (42%) died and 85 of 227 (37%) progressed. Adjusted ITT analyses showed no association with either outcome, whereas adjusted AT analyses showed lower mortality risk with sustained monotherapy and combination therapy (hazard ratio [HR] 0.50, 95% confidence interval [CI] 0.27–0.92, and HR 0.40, 95% CI 0.18–0.88, respectively) and lower PAH progression risk with sustained initial therapy (HR 0.37, 95% CI 0.18–0.76). In the milder hemodynamic subgroup, the mortality estimate was directionally consistent but imprecise. Conclusion: Initial treatment strategies reflected hemodynamic severity and baseline risk, whereas variation suggested potential implementation gaps. Sustained initial therapy was associated with lower risks of mortality and PAH progression. Contrasting ITT and AT estimates highlight challenges in observational treatment–outcome comparisons. Findings in the milder hemodynamic subgroup support dedicated prospective studies. image
- Open Access
- 18 September 2026
- Published by Oxford University Press (OUP) in Rheumatology
- Vol. 65 (10)
Effectiveness of oral anticoagulants in precapillary pulmonary hypertension associated with systemic sclerosis: a EUSTAR cohort study
Objectives: The role of anticoagulation in systemic sclerosis (SSc) with precapillary pulmonary hypertension (PH) remains controversial, with conflicting evidence regarding its impact on survival, including potential harmful effects. Aiming to address this research gap, we evaluated both mortality and PH worsening in a large SSc-precapillary PH cohort, additionally accounting for SSc-specific risk factors in our analyses, to better inform individualized treatment decisions. Methods: This retrospective cohort study included SSc patients from the European Scleroderma Trials and Research database with precapillary PH confirmed by right heart catheterization (RHC). The association of oral anticoagulants (OAC) use with survival and PH worsening was assessed using Kaplan–Meier estimates and multivariable Cox regression models. To minimize confounding, propensity score matching was additionally performed. Results: Among 614 patients included, 143 (23%) received OAC at the time of RHC. At baseline, patients on OAC had worse hemodynamic parameters. No significant association was found between OAC use and survival (HR 0.983 95%CI 0.724–1.334) or PH worsening (HR 1.070 95%CI 0.811–1.412). These findings remained consistent after propensity score matching and in the subgroup of 230 SSc patients with precapillary PH without interstitial lung disease. Conclusion: In this large SSc cohort with precapillary PH, we did not observe any clinical benefit associated with OAC, neither in terms of survival nor of PH worsening. Further prospective studies are needed to determine whether specific patient subgroups may benefit from anticoagulation therapy.
- 18 September 2026
- Published by Oxford University Press (OUP) in Rheumatology
- Vol. 65 (10)
Open AccessEffectiveness of oral anticoagulants in precapillary pulmonary hypertension associated with systemic sclerosis: a EUSTAR cohort study
Objectives: The role of anticoagulation in systemic sclerosis (SSc) with precapillary pulmonary hypertension (PH) remains controversial, with conflicting evidence regarding its impact on survival, including potential harmful effects. Aiming to address this research gap, we evaluated both mortality and PH worsening in a large SSc-precapillary PH cohort, additionally accounting for SSc-specific risk factors in our analyses, to better inform individualized treatment decisions. Methods: This retrospective cohort study included SSc patients from the European Scleroderma Trials and Research database with precapillary PH confirmed by right heart catheterization (RHC). The association of oral anticoagulants (OAC) use with survival and PH worsening was assessed using Kaplan–Meier estimates and multivariable Cox regression models. To minimize confounding, propensity score matching was additionally performed. Results: Among 614 patients included, 143 (23%) received OAC at the time of RHC. At baseline, patients on OAC had worse hemodynamic parameters. No significant association was found between OAC use and survival (HR 0.983 95%CI 0.724–1.334) or PH worsening (HR 1.070 95%CI 0.811–1.412). These findings remained consistent after propensity score matching and in the subgroup of 230 SSc patients with precapillary PH without interstitial lung disease. Conclusion: In this large SSc cohort with precapillary PH, we did not observe any clinical benefit associated with OAC, neither in terms of survival nor of PH worsening. Further prospective studies are needed to determine whether specific patient subgroups may benefit from anticoagulation therapy.
- Open Access
- 26 June 2026
- Published by Oxford University Press (OUP) in Rheumatology
- Vol. 65 (7)
BCMA-directed therapy with cizutamig induces rapid improvement in long-standing systemic sclerosis-associated interstitial lung disease
- 26 June 2026
- Published by Oxford University Press (OUP) in Rheumatology
- Vol. 65 (7)
Open AccessBCMA-directed therapy with cizutamig induces rapid improvement in long-standing systemic sclerosis-associated interstitial lung disease
- Top Cited PapersOpen Access
- 4 March 2025
- Published by Wolters Kluwer Health in Journal of the American Heart Association
- Vol. 14 (5) , e036730
Evaluation of Systemic Sclerosis Primary Heart Involvement and Chronic Heart Failure in the European Scleroderma Trials and Research Cohort
Background: Systemic sclerosis (SSc) primary heart involvement (SSc‐pHI) is one of the leading causes of mortality in SSc. We aimed to evaluate risk factors for SSc‐pHI and its progression and the outcomes in the EUSTAR (European Scleroderma Trials and Research) cohort. Methods: SSc‐pHI was defined according to the World Scleroderma Foundation/Heart Failure Association definition. Data from 5741 patients with SSc in the EUSTAR cohort were analyzed. Additional cardiovascular data were collected from a subcohort of 838 patients with SSc. Lasso regression was used for risk factor analyses. Kaplan–Meier estimator was used for survival analyses. Progression of SSc‐pHI was evaluated by a study definition developed by rheumatology and cardiology experts. Results: Risk factors for the presence of SSc‐pHI comprised skeletal muscle atrophy (odds ratio [OR], 2.00 [95% CI, 1.00–2.68]), age (OR, 1.91 [95% CI, 1.73–2.03]), male sex (OR, 1.77 [95% CI, 1.42–2.05]), swollen joints (OR, 1.70 [95% CI, 1.47–1.98]), skeletal muscle weakness (OR, 1.38 [95% CI, 1.00–1.85]), and tendon friction rubs (OR, 1.46 [95% CI, 1.00–1.77]) (n=3276). Telangiectasia (OR, 2.10 [95% CI, 1.38–2.72]), intestinal symptoms (OR, 1.70 [95% CI, 1.04–2.42]), age (OR, 1.47 [95% CI, 1.21–1.62]), and antitopoisomerase I antibodies (OR, 1.37 [95% CI, 1.00–1.77]) were associated with an increased risk for new onset of SSc‐pHI (n=1000). Survival rate was significantly lower in patients with SSc‐pHI than in those without ( P value <0.0001, n=3768). Patients with SSc‐pHI had a lower survival rate than patients with interstitial lung disease (n=3365). Swollen joints were associated with an increased risk of progressive SSc‐pHI (OR, 2.49 [95% CI, 1.79–3.52]) (n=595). Tendon friction rubs (OR, 1.21 [95% CI, 0.94–1.90]) increased the risk of heart failure with preserved ejection fraction in patients with SSc‐pHI. Conclusions: We defined progressive SSc‐pHI and identified risk factors for new onset and progression of SSc‐pHI and for SSc‐pHI‐associated heart failure with preserved ejection fraction in the largest cohort with SSc. These findings may guide patient stratification for diagnostic workup and therapy.
- 4 March 2025
- Published by Wolters Kluwer Health in Journal of the American Heart Association
- Vol. 14 (5) , e036730
Top Cited PapersOpen AccessEvaluation of Systemic Sclerosis Primary Heart Involvement and Chronic Heart Failure in the European Scleroderma Trials and Research Cohort
Background: Systemic sclerosis (SSc) primary heart involvement (SSc‐pHI) is one of the leading causes of mortality in SSc. We aimed to evaluate risk factors for SSc‐pHI and its progression and the outcomes in the EUSTAR (European Scleroderma Trials and Research) cohort. Methods: SSc‐pHI was defined according to the World Scleroderma Foundation/Heart Failure Association definition. Data from 5741 patients with SSc in the EUSTAR cohort were analyzed. Additional cardiovascular data were collected from a subcohort of 838 patients with SSc. Lasso regression was used for risk factor analyses. Kaplan–Meier estimator was used for survival analyses. Progression of SSc‐pHI was evaluated by a study definition developed by rheumatology and cardiology experts. Results: Risk factors for the presence of SSc‐pHI comprised skeletal muscle atrophy (odds ratio [OR], 2.00 [95% CI, 1.00–2.68]), age (OR, 1.91 [95% CI, 1.73–2.03]), male sex (OR, 1.77 [95% CI, 1.42–2.05]), swollen joints (OR, 1.70 [95% CI, 1.47–1.98]), skeletal muscle weakness (OR, 1.38 [95% CI, 1.00–1.85]), and tendon friction rubs (OR, 1.46 [95% CI, 1.00–1.77]) (n=3276). Telangiectasia (OR, 2.10 [95% CI, 1.38–2.72]), intestinal symptoms (OR, 1.70 [95% CI, 1.04–2.42]), age (OR, 1.47 [95% CI, 1.21–1.62]), and antitopoisomerase I antibodies (OR, 1.37 [95% CI, 1.00–1.77]) were associated with an increased risk for new onset of SSc‐pHI (n=1000). Survival rate was significantly lower in patients with SSc‐pHI than in those without ( P value <0.0001, n=3768). Patients with SSc‐pHI had a lower survival rate than patients with interstitial lung disease (n=3365). Swollen joints were associated with an increased risk of progressive SSc‐pHI (OR, 2.49 [95% CI, 1.79–3.52]) (n=595). Tendon friction rubs (OR, 1.21 [95% CI, 0.94–1.90]) increased the risk of heart failure with preserved ejection fraction in patients with SSc‐pHI. Conclusions: We defined progressive SSc‐pHI and identified risk factors for new onset and progression of SSc‐pHI and for SSc‐pHI‐associated heart failure with preserved ejection fraction in the largest cohort with SSc. These findings may guide patient stratification for diagnostic workup and therapy.
- Open Access
- 29 January 2025
- Published by Oxford University Press (OUP) in Rheumatology
- Vol. 64 (6) , 3643-3656
A comparative analysis of risk stratification tools in SSc-associated pulmonary arterial hypertension: a EUSTAR analysis
Objectives: The 2022 European Society of Cardiology and European Respiratory Society (ESC/ERS) guidelines for pulmonary arterial hypertension (PAH) recommend risk stratification to optimize management. However, the performance of generic PAH risk stratification tools in patients with SSc-associated PAH remains unclear. Our objective was to identify the most accurate approach for risk stratification at SSc-PAH diagnosis. Methods: In this multicentre, international cohort study from the European Scleroderma Trials and Research (EUSTAR) group database, we screened 11 risk stratification tools upon SSc-PAH diagnosis. We compared the performance of the three top-ranked tools to predict mortality with the ESC/ERS three-strata model, the currently recommended tool for baseline risk assessment. We also assessed the impact of incorporating SSc-specific characteristics into the tools. Kaplan–Meier analyses and Cox regression with area under the ROC curve (AUC) were conducted. Results: The ESC/ERS three-strata model had a lower ability to predict mortality than the ESC/ERS four-strata model, ‘SPAHR updated’ and ‘REVEAL Lite 2’. The ESC/ERS four-strata model divided ‘intermediate-risk’ patients into two groups with significantly different long-term survival rates and is the easiest applicable tool. Incorporating SSc-specific characteristics did not significantly improve the predictive ability of any model, but a low diffusing capacity of the lung for carbon monoxide (DLCO) was an independent predictor of mortality. Conclusion: Considering its ability to predict mortality, risk segregation capabilities and clinical applicability, this study provides a rationale for using the simplified ESC/ERS four-strata model at SSc-PAH diagnosis as an alternative to the comprehensive ESC/ERS three-strata model. We propose considering DLCO as an individual prognostic marker in SSc-PAH.
- 29 January 2025
- Published by Oxford University Press (OUP) in Rheumatology
- Vol. 64 (6) , 3643-3656
Open AccessA comparative analysis of risk stratification tools in SSc-associated pulmonary arterial hypertension: a EUSTAR analysis
Objectives: The 2022 European Society of Cardiology and European Respiratory Society (ESC/ERS) guidelines for pulmonary arterial hypertension (PAH) recommend risk stratification to optimize management. However, the performance of generic PAH risk stratification tools in patients with SSc-associated PAH remains unclear. Our objective was to identify the most accurate approach for risk stratification at SSc-PAH diagnosis. Methods: In this multicentre, international cohort study from the European Scleroderma Trials and Research (EUSTAR) group database, we screened 11 risk stratification tools upon SSc-PAH diagnosis. We compared the performance of the three top-ranked tools to predict mortality with the ESC/ERS three-strata model, the currently recommended tool for baseline risk assessment. We also assessed the impact of incorporating SSc-specific characteristics into the tools. Kaplan–Meier analyses and Cox regression with area under the ROC curve (AUC) were conducted. Results: The ESC/ERS three-strata model had a lower ability to predict mortality than the ESC/ERS four-strata model, ‘SPAHR updated’ and ‘REVEAL Lite 2’. The ESC/ERS four-strata model divided ‘intermediate-risk’ patients into two groups with significantly different long-term survival rates and is the easiest applicable tool. Incorporating SSc-specific characteristics did not significantly improve the predictive ability of any model, but a low diffusing capacity of the lung for carbon monoxide (DLCO) was an independent predictor of mortality. Conclusion: Considering its ability to predict mortality, risk segregation capabilities and clinical applicability, this study provides a rationale for using the simplified ESC/ERS four-strata model at SSc-PAH diagnosis as an alternative to the comprehensive ESC/ERS three-strata model. We propose considering DLCO as an individual prognostic marker in SSc-PAH.
- Open Access
Predicting interstitial lung disease progression in patients with systemic sclerosis using attentive neural processes - a EUSTAR study
Background Systemic sclerosis (SSc) is an autoimmune disease with high mortality with lung involvement being the primary cause of death. Progressive interstitial lung disease (ILD) leads to a decline in lung function (forced vital capacity, FVC% predicted) with risk of respiratory failure. These patients could benefit from an early and tailored pharmacological intervention. However, up to date, tools for prediction of individual FVC changes are lacking. In this paper, we aimed at developing a trustworthy machine learning system that is able to guide SSc management by providing not only robust FVC predictions, but also uncertainty quantification (i.e. the degree of certainty of model prediction) as well as similarity-based explainability for any patient P (i.e. a list of past SSc patients with similar FVC trajectories like P). We further aimed to identify the key clinical factors influencing the model’s predictions and to use model-guided data representation to identify SSc patients with similar sequential FVC measurements. Methods We trained and evaluated machine learning (ML) models to predict SSc-ILD trajectory as measured by FVC% predicted values using the international SSc database managed by the European Scleroderma Trials and Research group (EUSTAR), which comprises clinical, laboratory and functional parameters. EUSTAR records patients’ data in annual assessment visits, and, given any visit, we aimed at predicting the FVC value of a patient’s subsequent visit, taking into account all available patient data (i.e. baseline and follow-up visit data up to the time point where we make the prediction). For training of our ML models, we included 2220 SSc patients that had at least 3 recorded visits in the EUSTAR database, were at least 18 years old, had confirmed ILD and sufficient clinical documentation. We developed sequential ML models implementing the attentive neural process formalism with either a recurrent (ANP RNN) or transformer encoder (ANP transformer) architecture. We compared these architectures with baseline sequential models including gated recurrent neural networks (RNNs) and multi-head self-attention transformer-based networks. Baseline non-sequential models included tree-based models such as gradient boosting trees, and regression-based models with varying regularization schemes. Our experiments used stratified 5-fold cross-validation to train and test the models using the average root mean squared error (RMSE), weighted RMSE, and mean absolute error (MAE) as performance metrics. We computed the coverage and Winkler score for uncertainty quantification, SHAP values for grading the input features importance and used the data embeddings of the ANP architectures for both similarity-based explainability and the identification of similar SSc patient journeys. Results Patients’ baseline FVC scores ranged from 22 to 150% predicted with a mean (SD) of 90.53% predicted (21.52). Our deep learning models showed better performance for FVC forecasting, compared to tree- and regression-based models. The top performing ANP RNN architecture was able to closely model future FVC values with average (SD) performance of 8.240 (0.168) weighted RMSE and 6.94 (0.190) MAE that was further used as feature generator for a logistic regression trained to predict a FVC% decline of at least 10% points achieving 0.704 AUC score. In comparison, a naïve baseline using the mean FVC value as a predictor achieved much lower FVC forecasting capabilities, with 18.718 (0.317) weighted RMSE, and 17.619 (0.599) MAE. SHAP value analysis indicated that prior FVC measurements, diffusion of carbon monoxide (DLCO) values, skin involvement, age, anti-centromere positivity, dyspnea and CRP-elevation contributed most to deep-learning-based FVC predictions. Regarding uncertainty quantification, ANP RNN achieved 79% coverage (i.e. the model would provide uncertainty estimates that included the true future FVC value in 79 out of 100 predictions) out of the box, and 90% using an additional conformal prediction module with an corresponding Winkler score of 892 (indicating the width of the uncertainty estimate plus penalty for mistakes), smaller than any other model at the same coverage level. We further demonstrate how the data abstraction provided by the ANP RNN model (embeddings) allows for deriving similar patient trajectories (for similarity-based explanation). Conclusions Our study demonstrates the feasibility of FVC forecasting and thus the ability to predict ILD trajectories in individual SSc patients using deep learning. We show that model predictions can be paired with uncertainty quantification and similarity-based model explainability, which are crucial elements for deploying trustworthy ML algorithms. Our study is thus an important first step towards reliable automated ILD trajectory (i.e. FVC%) prediction system with potential clinical utility.
Open AccessPredicting interstitial lung disease progression in patients with systemic sclerosis using attentive neural processes - a EUSTAR study
Background Systemic sclerosis (SSc) is an autoimmune disease with high mortality with lung involvement being the primary cause of death. Progressive interstitial lung disease (ILD) leads to a decline in lung function (forced vital capacity, FVC% predicted) with risk of respiratory failure. These patients could benefit from an early and tailored pharmacological intervention. However, up to date, tools for prediction of individual FVC changes are lacking. In this paper, we aimed at developing a trustworthy machine learning system that is able to guide SSc management by providing not only robust FVC predictions, but also uncertainty quantification (i.e. the degree of certainty of model prediction) as well as similarity-based explainability for any patient P (i.e. a list of past SSc patients with similar FVC trajectories like P). We further aimed to identify the key clinical factors influencing the model’s predictions and to use model-guided data representation to identify SSc patients with similar sequential FVC measurements. Methods We trained and evaluated machine learning (ML) models to predict SSc-ILD trajectory as measured by FVC% predicted values using the international SSc database managed by the European Scleroderma Trials and Research group (EUSTAR), which comprises clinical, laboratory and functional parameters. EUSTAR records patients’ data in annual assessment visits, and, given any visit, we aimed at predicting the FVC value of a patient’s subsequent visit, taking into account all available patient data (i.e. baseline and follow-up visit data up to the time point where we make the prediction). For training of our ML models, we included 2220 SSc patients that had at least 3 recorded visits in the EUSTAR database, were at least 18 years old, had confirmed ILD and sufficient clinical documentation. We developed sequential ML models implementing the attentive neural process formalism with either a recurrent (ANP RNN) or transformer encoder (ANP transformer) architecture. We compared these architectures with baseline sequential models including gated recurrent neural networks (RNNs) and multi-head self-attention transformer-based networks. Baseline non-sequential models included tree-based models such as gradient boosting trees, and regression-based models with varying regularization schemes. Our experiments used stratified 5-fold cross-validation to train and test the models using the average root mean squared error (RMSE), weighted RMSE, and mean absolute error (MAE) as performance metrics. We computed the coverage and Winkler score for uncertainty quantification, SHAP values for grading the input features importance and used the data embeddings of the ANP architectures for both similarity-based explainability and the identification of similar SSc patient journeys. Results Patients’ baseline FVC scores ranged from 22 to 150% predicted with a mean (SD) of 90.53% predicted (21.52). Our deep learning models showed better performance for FVC forecasting, compared to tree- and regression-based models. The top performing ANP RNN architecture was able to closely model future FVC values with average (SD) performance of 8.240 (0.168) weighted RMSE and 6.94 (0.190) MAE that was further used as feature generator for a logistic regression trained to predict a FVC% decline of at least 10% points achieving 0.704 AUC score. In comparison, a naïve baseline using the mean FVC value as a predictor achieved much lower FVC forecasting capabilities, with 18.718 (0.317) weighted RMSE, and 17.619 (0.599) MAE. SHAP value analysis indicated that prior FVC measurements, diffusion of carbon monoxide (DLCO) values, skin involvement, age, anti-centromere positivity, dyspnea and CRP-elevation contributed most to deep-learning-based FVC predictions. Regarding uncertainty quantification, ANP RNN achieved 79% coverage (i.e. the model would provide uncertainty estimates that included the true future FVC value in 79 out of 100 predictions) out of the box, and 90% using an additional conformal prediction module with an corresponding Winkler score of 892 (indicating the width of the uncertainty estimate plus penalty for mistakes), smaller than any other model at the same coverage level. We further demonstrate how the data abstraction provided by the ANP RNN model (embeddings) allows for deriving similar patient trajectories (for similarity-based explanation). Conclusions Our study demonstrates the feasibility of FVC forecasting and thus the ability to predict ILD trajectories in individual SSc patients using deep learning. We show that model predictions can be paired with uncertainty quantification and similarity-based model explainability, which are crucial elements for deploying trustworthy ML algorithms. Our study is thus an important first step towards reliable automated ILD trajectory (i.e. FVC%) prediction system with potential clinical utility.
- Open Access
- 30 January 2024
- Published by MDPI AG in Diagnostics
- Vol. 14 (3) , 295
The Performance of Pulmonary Function Tests in Predicting Systemic Sclerosis—Interstitial Lung Disease in the European Scleroderma Trial and Research Database
Background and Objectives: In SSc, ILD is a major cause of morbidity and mortality. We aimed to investigate the performance of DLCO (diffusing capacity of lung carbon monoxide) and FVC (forced vital capacity) delta change (Δ) and baseline values in predicting the development of SSc-ILD. Methods: Longitudinal data of DLCO, FVC, and ILD on the HRCT of SSc patients from the EUSTAR database were evaluated at baseline (t0) and after 12 (±4) (t1) and 24 (±4) (t2) months. Results: 474/17805 patients were eligible for the study (403 females); 46 (9.7%) developed ILD at t2. Positivity for anti-topoisomerase antibodies (117 patients) showed an association with ILD development at t2 (p = 0.0031). Neither the mean t0 to t1 change (Δ) of DLCO nor the mean t0 to t1 FVCΔ predicted the appearance of ILD at t2. Investigating the possible role of baseline DLCO and FVC values in predicting ILD appearance after 24 (±4) months, we observed a moderate predictive capability of t0 DLCO < 80%, stronger than that of FVC < 80%. Conclusions: We suggest that an impaired baseline DLCO may be predictive of the appearance of ILD after 2 years of follow-up. This result advances the hypothesis that a reduction in gas exchange may be considered an early sign of lung involvement. However, further rigorous studies are warranted to understand the predictive role of DLCO evaluation in the course of SSc.
- 30 January 2024
- Published by MDPI AG in Diagnostics
- Vol. 14 (3) , 295
Open AccessThe Performance of Pulmonary Function Tests in Predicting Systemic Sclerosis—Interstitial Lung Disease in the European Scleroderma Trial and Research Database
Background and Objectives: In SSc, ILD is a major cause of morbidity and mortality. We aimed to investigate the performance of DLCO (diffusing capacity of lung carbon monoxide) and FVC (forced vital capacity) delta change (Δ) and baseline values in predicting the development of SSc-ILD. Methods: Longitudinal data of DLCO, FVC, and ILD on the HRCT of SSc patients from the EUSTAR database were evaluated at baseline (t0) and after 12 (±4) (t1) and 24 (±4) (t2) months. Results: 474/17805 patients were eligible for the study (403 females); 46 (9.7%) developed ILD at t2. Positivity for anti-topoisomerase antibodies (117 patients) showed an association with ILD development at t2 (p = 0.0031). Neither the mean t0 to t1 change (Δ) of DLCO nor the mean t0 to t1 FVCΔ predicted the appearance of ILD at t2. Investigating the possible role of baseline DLCO and FVC values in predicting ILD appearance after 24 (±4) months, we observed a moderate predictive capability of t0 DLCO < 80%, stronger than that of FVC < 80%. Conclusions: We suggest that an impaired baseline DLCO may be predictive of the appearance of ILD after 2 years of follow-up. This result advances the hypothesis that a reduction in gas exchange may be considered an early sign of lung involvement. However, further rigorous studies are warranted to understand the predictive role of DLCO evaluation in the course of SSc.
- Open Access
- 27 January 2023
- Published by Oxford University Press (OUP) in Rheumatology
- Vol. 62 (9) , 3067-3074
Anti-acid therapy in SSc-associated interstitial lung disease: long-term outcomes from the German Network for Systemic Sclerosis
Gastroesophageal reflux disease (GERD) occurs frequently in patients with SSc. We investigated whether the presence of GERD and/or the use of anti-acid therapy, specifically proton-pump inhibitors (PPIs), are associated with long-term outcomes, especially in SSc-associated interstitial lung disease (SSc-ILD). We retrospectively analysed patients with SSc and SSc-ILD from the German Network for Systemic Sclerosis (DNSS) database (2003 onwards). Kaplan–Meier analysis compared overall survival (OS) and progression-free survival (PFS) in patients with GERD vs without GERD (SSc and SSc-ILD), and PPI vs no PPI use (SSc-ILD only). Progression was defined as a decrease in either percentage predicted forced vital capacity of ≥10% or single-breath diffusing capacity for carbon monoxide of ≥15%, or death. It was found that 2693/4306 (63%) registered patients with SSc and 1204/1931 (62%) with SSc-ILD had GERD. GERD was not associated with decreased OS or decreased PFS in patients in either cohort. In SSc-ILD, PPI use was associated with improved OS vs no PPI use after 1 year [98.4% (95% CI: 97.6, 99.3); n = 760 vs 90.8% (87.9–93.8); n = 290] and after 5 years [91.4% (89.2–93.8); n = 357 vs 70.9% (65.2–77.1); n = 106; P < 0.0001]. PPI use was also associated with improved PFS vs no PPI use after 1 year [95.9% (94.6–97.3); n = 745 vs 86.4% (82.9–90.1); n = 278] and after 5 years [66.8% (63.0–70.8); n = 286 vs 45.9% (39.6–53.2); n = 69; P < 0.0001]. GERD had no effect on survival in SSc or SSc-ILD. PPIs improved survival in patients with SSc-ILD. Controlled, prospective trials are needed to confirm this finding.
- 27 January 2023
- Published by Oxford University Press (OUP) in Rheumatology
- Vol. 62 (9) , 3067-3074
Open AccessAnti-acid therapy in SSc-associated interstitial lung disease: long-term outcomes from the German Network for Systemic Sclerosis
Gastroesophageal reflux disease (GERD) occurs frequently in patients with SSc. We investigated whether the presence of GERD and/or the use of anti-acid therapy, specifically proton-pump inhibitors (PPIs), are associated with long-term outcomes, especially in SSc-associated interstitial lung disease (SSc-ILD). We retrospectively analysed patients with SSc and SSc-ILD from the German Network for Systemic Sclerosis (DNSS) database (2003 onwards). Kaplan–Meier analysis compared overall survival (OS) and progression-free survival (PFS) in patients with GERD vs without GERD (SSc and SSc-ILD), and PPI vs no PPI use (SSc-ILD only). Progression was defined as a decrease in either percentage predicted forced vital capacity of ≥10% or single-breath diffusing capacity for carbon monoxide of ≥15%, or death. It was found that 2693/4306 (63%) registered patients with SSc and 1204/1931 (62%) with SSc-ILD had GERD. GERD was not associated with decreased OS or decreased PFS in patients in either cohort. In SSc-ILD, PPI use was associated with improved OS vs no PPI use after 1 year [98.4% (95% CI: 97.6, 99.3); n = 760 vs 90.8% (87.9–93.8); n = 290] and after 5 years [91.4% (89.2–93.8); n = 357 vs 70.9% (65.2–77.1); n = 106; P < 0.0001]. PPI use was also associated with improved PFS vs no PPI use after 1 year [95.9% (94.6–97.3); n = 745 vs 86.4% (82.9–90.1); n = 278] and after 5 years [66.8% (63.0–70.8); n = 286 vs 45.9% (39.6–53.2); n = 69; P < 0.0001]. GERD had no effect on survival in SSc or SSc-ILD. PPIs improved survival in patients with SSc-ILD. Controlled, prospective trials are needed to confirm this finding.
- Open Access
- 4 January 2023
- Published by Frontiers Media SA in Frontiers in Immunology
- Vol. 13, 1045523
Comprehensive autoantibody profiles in systemic sclerosis: Clinical cluster analysis
Background: Systemic sclerosis (SSc) belongs to the group of connective tissue diseases and is associated with the occurrence of disease-specific autoantibodies. Although it is still controversial whether these antibodies contribute to pathogenesis, there are new insights into the development of these specific antibodies and their possible pathophysiological properties. Interestingly, they are associated with specific clinical manifestations, but for some rarer antibodies this association is not fully clarified. The aim of this study is a comprehensive analysis of the serum autoantibody status in patients with SSc followed by correlation analyses of autoantibodies with the clinical course of the disease.Methods: Serum from SSc patients was analyzed using a line blot (EUROLINE, EUROIMMUN AG) for SSc-related autoantibodies. Autoantibodies to centromere, Topo-1, antimitochondrial antibodies (AMA) M2 subunit, angiotensin II type 1 receptors (AT1R) and endothelin-1 type-A-receptors (ETAR) were also determined by ELISA. We formed immunological clusters and used principal components analysis (PCA) to assign specific clinical characteristics to these clusters.Results: A total of 372 SSc patients were included. 95.3% of the patients were antinuclear antibody positive and in 333 patients at least one SSc specific antibody could be detected. Four immunological clusters could be found by PCA. Centromere, Topo-1 and RP3 all formed own clusters, which are associated with distinct clinical phenotypes. We found that patients with an inverted phenotype, such as limited cutaneous SSc patients within the Topo-1 cluster show an increased risk for interstital lung disease compared to ACA positive patients. Anti-AT1R and anti-ETAR autoantibodies were measured in 176 SSc patients; no association with SSc disease manifestation was found. SSc patients with AMA-M2 antibodies showed an increased risk of cardiovascular events.Conclusion: In our in large cluster analysis, which included an extended autoantibody profile, we were able to show that serologic status of SSc patients provides important clues to disease manifestation, co-morbidities and complications. Line blot was a reliable technique to detect autoantibodies in SSc and detected rarer autoantibodies in 42% of our patients.
- 4 January 2023
- Published by Frontiers Media SA in Frontiers in Immunology
- Vol. 13, 1045523
Open AccessComprehensive autoantibody profiles in systemic sclerosis: Clinical cluster analysis
Background: Systemic sclerosis (SSc) belongs to the group of connective tissue diseases and is associated with the occurrence of disease-specific autoantibodies. Although it is still controversial whether these antibodies contribute to pathogenesis, there are new insights into the development of these specific antibodies and their possible pathophysiological properties. Interestingly, they are associated with specific clinical manifestations, but for some rarer antibodies this association is not fully clarified. The aim of this study is a comprehensive analysis of the serum autoantibody status in patients with SSc followed by correlation analyses of autoantibodies with the clinical course of the disease.Methods: Serum from SSc patients was analyzed using a line blot (EUROLINE, EUROIMMUN AG) for SSc-related autoantibodies. Autoantibodies to centromere, Topo-1, antimitochondrial antibodies (AMA) M2 subunit, angiotensin II type 1 receptors (AT1R) and endothelin-1 type-A-receptors (ETAR) were also determined by ELISA. We formed immunological clusters and used principal components analysis (PCA) to assign specific clinical characteristics to these clusters.Results: A total of 372 SSc patients were included. 95.3% of the patients were antinuclear antibody positive and in 333 patients at least one SSc specific antibody could be detected. Four immunological clusters could be found by PCA. Centromere, Topo-1 and RP3 all formed own clusters, which are associated with distinct clinical phenotypes. We found that patients with an inverted phenotype, such as limited cutaneous SSc patients within the Topo-1 cluster show an increased risk for interstital lung disease compared to ACA positive patients. Anti-AT1R and anti-ETAR autoantibodies were measured in 176 SSc patients; no association with SSc disease manifestation was found. SSc patients with AMA-M2 antibodies showed an increased risk of cardiovascular events.Conclusion: In our in large cluster analysis, which included an extended autoantibody profile, we were able to show that serologic status of SSc patients provides important clues to disease manifestation, co-morbidities and complications. Line blot was a reliable technique to detect autoantibodies in SSc and detected rarer autoantibodies in 42% of our patients.