Accepted
Article
This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process, which may
lead to differences between this version and the Version of Record. Please cite this article as
doi: 10.1002/art.40906
DR. VINCENT SOBANSKI (Orcid ID : 0000-0003-3083-2441)
Article type : Full Length
Phenotypes determined by cluster analysis and their survival in the
prospective EUSTAR cohort of patients with systemic sclerosis
Running head: Cluster analysis of EUSTAR cohort
Vincent Sobanski1,2,3,4$, MD, PhD, Jonathan Giovannelli1,2,3$, MD, PhD, Yannick Allanore5, MD, PhD, Gabriela Riemekasten6, MD, PhD, Paolo Airò7, MD, Serena Vettori8, MD, PhD, Franco Cozzi9, MD,
PhD, Oliver Distler10, MD, PhD, Marco Matucci-Cerinic11, MD, PhD, Christopher Denton12, MD, PhD, David Launay1,2,3,4*, MD, PhD, Eric Hachulla1,2,3,4*, MD, PhD, and EUSTAR collaborators
$
and *: these authors have contributed equally
1. Univ. Lille, U995 - LIRIC - Lille Inflammation Research International Center, F-59000 Lille, France 2. INSERM, U995, F-59000 Lille, France
3. CHU Lille, Département de Médecine Interne et Immunologie Clinique, F-59000 Lille, France 4. Referral center for rare systemic autoimmune diseases North and North-west of France 5. Service de Rhumatologie A – – APHP, Université Paris Descartes, Paris, France 6. Rheumatolgy and Clinical Immunology, University Clinic Schleswig-Holstein, University of Lübeck,
Germany
7. UO Rheumatology and Clinical Immunology, ASST Spedali Civili, Brescia, Italy 8. , Napoli, Italy 9. , Padova, Italy
10. Department of Rheumatology - University Hospital Zurich, Switzerland
11. Department Experimental and Clinical Medicine, Division of Rheumatology AOUC, University of Florence, Florence, Italy
Accepted
Article
Corresponding author: Pr David Launay
Service de Médecine Interne ; Hôpital Claude-Huriez, CHRU Lille ; Rue Michel Polonovski F-59037 LILLE Cedex, FRANCE
Tel: + 33320445650 / Fax: + 33320445459 ; E-mail: [email protected]
Abstract
ObjectivesSystemic sclerosis (SSc) is a heterogeneous connective tissue disease. The usual subclassification divides patients into limited (lc) and diffuse cutaneous (dc) subsets relying on the skin extension but may not capture the entire variability of clinical phenotypes. The European Scleroderma Trials and Research (EUSTAR) database is a prospective cohort providing data from 137 European referral centers. We performed a cluster analysis to distinguish and characterize homogeneous phenotypes without any a priori, and examined survival among the obtained clusters.
Methods
A total of 11,318 patients were registered in the EUSTAR database of whom 6,927 patients were included in the study. Twenty-four variables including clinical and serological variables were used for clusterisation.
Results
Clustering analyses provided a first delineation of 2 clusters showing a moderate stability. We further characterized 6 homogeneous groups in an exploratory attempt which differed regarding their clinical features, autoantibody profile, and mortality. Some groups resembled usual dcSSc or
Accepted
Article
lcSSc prototypes, but other exhibited unique features like a majority of lcSSc with a high rate of visceral damage and antitopoisomerase antibodies. Prognosis varied among groups and the presence of organ damage markedly impacted survival beyond the cutaneous involvement.
Conclusion
Our work suggests that restricting the subsets of SSc patients only to the cutaneous involvement may not capture the complete heterogeneity of the disease. Organ damage and antibody profile should be taken into consideration when individuating homogeneous groups of patients with distinct prognosis.
Funding None
Keywords
Systemic sclerosis – cluster analysis – heterogeneity – organ involvement – autoantibody
Introduction
Systemic sclerosis (SSc) is a chronic disease affecting connective tissue and characterized by vascular damage, autoimmunity and fibrosis. The European League Against Rheumatism (EULAR) and the American College of Rheumatology have recently developed new classification criteria for SSc [1]. To date, the sub-classification of SSc patients mainly rely on the cutaneous extension subsets proposed by LeRoy et al. in 1988 [2,3]. It separates patients into two main groups: (i) diffuse cutaneous SSc
Accepted
Article
(dcSSc) associated with early skin changes affecting the trunk and proximal limbs and (ii) limited cutaneous SSc (lcSSc) in which skin fibrosis is limited to the hands, face, feet and forearms. Organ damage can vary between the two subsets, with an early and significant incidence of organ damage (lung fibrosis, gastrointestinal involvement, heart disease and renal crisis) in dcSSc and pulmonary hypertension (PH) in lcSSc [4]. The two subsets also differ in autoantibody profile, with a high prevalence (70-80%) of anticentromere antibodies (ACA) in lcSSc, and a predominant presence of antibodies against topoisomerase I (ATA) in dcSSc (30%) than in lcSSc in the original report [4]. In addition, mortality is higher in patients with dcSSc than in patients with lcSSc [5,6]. Overall, previous studies suggest that lcSSc and dcSSc are two clearly differentiated phenotypes regarding clinical characteristics, serological profiles and prognosis [7].
Yet, past and recent reports of large cohorts have challenged this distinction by highlighting an often neglected heterogeneity among clinical subsets [8-12], as suggested by, for example, lcSSc patients with ATA and severe interstitial lung disease (ILD). One method to deal with heterogeneity is to conduct a cluster analysis in order to organize data from a heterogeneous population into a fairly small number of homogeneous groups. Cluster analysis has been applied to various conditions such as gout [13], chronic heart failure [14], asthma [15], mixed connective tissue diseases [16] or ANCA-associated vasculitis [17]. Cluster analyses have also been carried out in two SSc studies, to our knowledge [18,19]. One of them included patients from the EULAR scleroderma trials and research cohort (EUSTAR) but was centered on capillaroscopy patterns [18]. Another recent study took into account a limited number of cluster variables and a limited number of patients [19].
This work intended to distinguish and characterize homogeneous groups of SSc patients using cluster analysis within the large EUSTAR cohort, and analyze survival between the obtained clusters.
Accepted
Article
Patients and Methods
Patient population
Data were collected from the SSc patients included in the prospective, open and multi-national SSc EUSTAR cohort, started in June 2004 [20-22]. In this work, the EUSTAR database was locked in April 2014. Eligible patients w ≥ 18, fulfilled ACR criteria for SSc [23], and had a calculable SSc duration – i.e. a date of disease onset (defined as the onset of the first non- y ’ ) and at least one date of visit.
All patients agreed to participate in the EUSTAR database by signing informed consent forms approved by the local ethical committees. The study was conducted in accordance with the principles of the Declaration of Helsinki and local laws and guidelines for Good Clinical Practice [21,22].
Definition and selection of variables
The EUSTAR database contains data on demographics, disease features, organ damage, lab parameters, capillaroscopy results, echocardiography, pulmonary function tests (PFTs), and medication. In order to harmonize clinical practices and ensure reliable evaluation of parameters among centers, EUSTAR arranges regular training courses and edits SSc management guidelines [24,25]. Autoantibodies were identified and characterized accordingly to local centers guidelines [21,22]. Clustering variables were selected in order to ensure a global phenotype of SSc patients by: (a) considering clinical relevance and representativeness of disease features, (b) eliminating
redundant variables procuring analogous information and (c) dismissing variables with high rate of missing values. We retained 24 variables, covering: (i) symptoms or organ involvement observed at least once among visits (Raynaud, oesophageal, stomach and intestinal symptoms, digital ulcerations (DU), joint synovitis, joint contractures, tendon friction rubs, muscle weakness, muscle atrophy, arterial hypertension, palpitations, renal crisis), (ii) results of laboratory values (CK elevation,
Accepted
Article
proteinuria, positive antinuclear antibodies (ANA), ACA and ATA), (iii) other tests (restrictive defect on PFTs, plain X-ray lung fibrosis, conduction blocks, abnormal diastolic function, suspected PH by cardiac echography), and (iv) the peak mRSS noticed during the follow-up (Table 1 and eFigure 2). Each variable from (i) to (iii) was treated positive for a defined patient if recorded at least once as a “y ” included visits.
Statistical analyses
Cluster analysis
Cluster analysis determines the distances between individuals using the combined values of their measured features to obtain groups of individuals who are more resembling to each other in one group than to those in the other groups. Cluster analysis was carried out from ascendant hierarchical clustering on the 24 selected variables using Ward’ . Results were graphically represented in a dendrogram. We estimated the number of clusters with a visual distance criterion by the horizontal intersection at the highest dissimilarity level on the dendrogram (i.e. where the vertical branches were the longest). In an exploratory approach, we increased the number of clusters considering the sub-optimal visual distance criterion by cutting the dendrogram horizontally at the second level of highest dissimilarity [26].
Evaluation of the cluster-wise stability and reproducibility is a major issue while performing a cluster analysis [27]. To assess it, we conducted 100 iterations of the clustering process (with the number of clusters of the primary analysis) in randomly selected subsets to 50% of the original dataset, and estimated the cluster-wise stability with the Jaccard coefficient (which is a similarity measure between datasets) computed between every cluster of the primary analysis and the most comparable cluster retrieved for each iteration [27]. A Jaccard similarity index ≤ 0.5 indicates a weakly stable and reproducible cluster [28].
Accepted
Article
The main cluster analysis was carried out in patients without missing variable for the 24 selected variables. In order to estimate the impact of late complications on the cluster analysis, we performed a sensitivity analysis by selecting patients with a disease duration more than 10 years (adequate time for the occurrence of organ damage). In order to study the possible impact of rare antibodies on the clustering process, we executed a second sensitivity analysis by adding in
clustering variables anti-RNA polymerase III, anti-PM/Scl and anti-U1RNP antibodies. Finally, a third sensitivity analysis was conducted to evaluate the potential survival bias, restricted to patients with a disease duration at the enrollment visit less than 5 years.
T w “ ” ( w/ / / …) w the clustering process but to describe and interpret the groups of patients in accordance with established practice [13,14].
Survival analysis
Survival was assessed using the disease duration (time from disease onset to the date of the latest news). We found a high percentage (52%) of patients lost to follow-up (i.e. date of last news prior January 2012), responsible for a significant survival overestimation. Because we could not update data with actual vital status, we chose to exclude those patients from the survival analysis. A sensitivity analysis was therefore performed without rejecting those patients. We also performed a y y y ’ definition of disease onset.
Survival rates were examined using several Cox proportional hazards models: (i) unadjusted, (ii) adjusted for age at disease onset, (iii) adjusted for age at disease onset and sex, and (iv) adjusted for age at disease onset, sex and immunosuppressive treatment. The proportional hazards assumption for Cox regression models were assessed by the graphical study of Schonfeld's residues, and the log-linearity assumption for quantitative predictors was assessed using cubic spline functions. Finally, we calculated the C-index for each Cox regression models (i.e. the estimation of the probability of concordance, equivalent to the area under the receiver operating characteristic curve for logistic
Accepted
Article
regression models). Statistical analyses were carried out using R software, version 2.14 [29], working with “ ” “f ” k . The statistical significance threshold was p<0.05.
Results
Population
A total of 11,318 patients were registered within the EUSTAR database as of April 2014
(corresponding to 120 centers), and 34,066 visits were recorded. Of these patients, 2,886 were excluded and 1,505 were not analyzed (due to 1 missing value in the variables used for clustering). Therefore 6,927 patients were incorporated in the cluster analysis (eFigure 1). Patients included were a little older (58.7±13.2 vs. 56.3±13.9 years, p<0.001), and had a longer disease duration (11.4±8.1 vs. 8.7±8.1, p<0.001), rate of dcSSc (42 vs. 38%, p=0.011) and more generally a more severe disease as shown by proportions of organ damages than patients who were not analyzed (Table 1). The median (interquartile range) number of visits per patient was 3 (4).
The proportion of patients included with dcSSc and lcSSc was 42% and 58%, respectively. DcSSc patients were significantly younger than those with lcSSc, and presented a more serious disease. In dcSSc, ACA/ATA were found in 14%/61%, and in lcSSc 54%/23%, respectively (Table 1).
Cluster analysis
Clustering of individuals on the basis of the 24 selected variables yielded an optimal number of 2 clusters: C(A) and C(B) (Figure 1A). Jaccard indexes pointed out a moderate stability: 0.64 for C(A) and 0.66 for C(B). Table 2, eTable 1, Figure 1B and Figure 2 summarized C(A) and C(B)
characteristics. Contingency tables (eTables 2 and 3) show the proportion of ACA and ATA in the different subsets of SSc according to the skin extension (lc/dcSSc).
Accepted
Article
C(A) (n=3149, 45.5%). C(A) contained principally patients with lcSSc (81%). Less than a third of
patients in this cluster had severe organ damage (DU, intestinal symptoms, muscle, joint, cardiac and lung involvement). ACA were present in 54% of patients and ATA in 21%.
C(B) (n=3778, 55.5%). Patients were a little younger than in C(A), with a younger age at the
beginning of the disease. In C(B), 61% of patients had dcSSc. A majority of patients presented with DU, joint contractures, intestinal involvement and ILD. Autoantibody profile was the opposite of C(A) with 54% of ATA-positive patients and 22% of ACA-positive patients.
In an exploratory attempt to decipher the heterogeneity of the disease, we then increased the clusters number. Graphical observation of the dendrogram determined that a suboptimal number of clusters was 6 (Figure 1A). As a consequence, we observed a decrease in Jaccard coefficients
(ranging from 0.32 to 0.68). Table 2, Figure 1B and Figure 3 summarized C1 to C6 characteristics.
C1 (n=1186, 17%). C1 included a majority of lcSSc (89%) female patients, with an older onset, a high
prevalence of GI involvement and a weak proportion of ILD. Most of the patients (79%) were ACA-positive.
C2 (n=720, 18%). C2 was composed mainly by lcSSc patients (71%), with raised frequencies of
suspected PH by echo (39%), ILD (85%) and restrictive defect (61%). ATA were present in 35% of patients and ACA in 24%.
C3 (n=1243, 10%). C3 contained mainly lcSSc (79%) characterized by low prevalence of GI
involvement and ILD. ACA were twice as frequent as ATA (48% vs. 24%, respectively).
C4 (n=1673, 18%). Patients in C4 were mainly lcSSc patients (63%) with a serious disease as
demonstrated by high proportions of cardiac and lung, muscular, joint and GI involvements and DU. ATA were present in 46% of patients and ACA in 29%.
Accepted
Article
C5 (n=1249, 24%). In C5, patients were mainly dcSSc (72%), with a notable proportion of male
patients (19%), GI, joint and cardiac disease and moderate lung involvement. Half of patients in C5 were ATA-positive and 20% ACA-positive.
C6 (n=856, 12%). Patients in C6 patients were characterized by the highest proportion of dcSSc
(92%) and males (22%), the most elevated peak of mRSS (27.2), and a serious disease as shown by high frequencies of GI, joint, muscular, renal, lung and cardiac disease. We found ATA in 77% and ACA in 12% of patients.
Three sensitivity cluster analyses were conducted (i) including only patients with disease duration more than 10 years (eTable 4), (ii) including anti-U1RNP, anti-RNA polymerase III, and anti-PM/Scl antibodies as clustering variables (eTable 5) and (iii) including only patients with a disease duration less than 5 years at the enrollment visit (eTable 6) and rendered similar results.
Survival analyses
Kaplan-Meier curves are shown in Figures 2 and 3 and eFigure 3 and 4. Survival rates are presented in eTable 7 and results of Cox regression analyses in Table 3.
The risk of death was increased for patients with dcSSc compared to patients with lcSSc (hazard ratio (HR) = 2.03 [95%CI 1.61-2.56] in the most adjusted model). An increased risk of death was also present in C(B) compared to C(A) (HR=2.47 [1.86-3.27]). When analyzing 6 clusters, we noticed in the most adjusted model a continuous increasing mortality from C1 to C6. The risk of death had a magnitude superior to those noted in the two previous analyses (i.e. HR=6.14 [3.81-9.89] for C6, compared to C1). C-indexes were similar for the most adjusted models: lc vs. dcSSc, C(A) vs. C(B) and for the 6 clusters (0.78 0.02, 0.78 0.02 and 0.79 0.02, respectively).
The sensitivity analysis taking account of lost to follow-up patients yielded comparable HR when we examined survival in C(A)-C(B) and clusters 1 to 6 (data not shown). We also performed a sensitivity
Accepted
Article
y y ’ (eTable 8), which rendered similar results, albeit the number of patients with available data was lower.
Discussion
This work intended to distinguish homogeneous groups in a substantial population of approximately 7,000 SSc patients with a cluster analysis. The main findings were that: (i) the optimal clustering divided patients into 2 distinct groups by their clinical and serological features, severity and prognosis, which partially overlapped the sub-classification dcSSc/lcSSc, and (ii) in an exploratory attempt, we further described 6 homogeneous subsets of individuals which broadly differed regarding clinical features, autoantibody profiles and survival.
The fact of finding 2 clusters could be considered as a validation of the expected dichotomy between dcSSc and lcSSc. Nevertheless, in C(A) 19% of patients had dcSSc and 21% had ATA. C(B) included up to 39% of patients with lcSSc and 22% with ACA. It appeared that no clear parallelism could be observed between the severity of organ damage and the cutaneous extension of SSc. This is in accordance with recent works. As an example, Nihtyanova et al. demonstrated that the presence of significant organ involvement was a strong predictor of prognosis, in both lcSSc and dcSSc, in a study of nearly 400 consecutive patients followed up for up to 15 years. Notably, survival curves were close for both cutaneous subsets when organ damage was present [30]. Altogether, these results suggest that, while a consensus stands on the relevance and practicality of a SSc sub-classification into lcSSc and dcSSc [31], this binary scheme could seem as a restrictive separation within a continuous spectrum of varying severity primarily driven by organ damage and subsequent prognosis [12].
In an exploratory attempt to study more in-depth the heterogeneity of SSc, we further described 6 additional clusters. Some of the 6 obtained clusters were quite expected as they were consistent
Accepted
Article
with the historical description of lcSSc and dcSSc. Indeed, C1 combined patients with the classical presentation of lcSSc, i.e. older female patients with a small proportion of severe organ damage, the topmost frequency of ACA-positive patients and a general favorable prognosis. C6 resembled the classical description of dcSSc with a high rate of male patients, the highest frequency of ATA-positive patients and severe organ damage, and prognosis. Intriguingly, we observed other subgroups that seemed to stand beyond the sole skin extension. Thereby, C2 was composed principally by lcSSc but with a rather high frequency of ATA-positive patients and a high rate of ILD, and suspected PH. Of note, the prognosis of patients from C2 was significantly unfavorable than those from C1. Similarly, C4 also contained predominantly lcSSc patients often with visceral complication. C5 was represented for the most part by dcSSc but we noted in this group a lower frequency of ILD and suspected PH than in C2, C4 or C6. These findings evoke that sub-classification established solely on skin extension might not be entirely representative of the severity of organ damage and prognosis.
Furthermore, this work highlighted some groups of patients in which the classical relationship lcSSc/ACA and dcSSc/ATA were not obvious. C2 was a representative group of this observation. Indeed, 71% of patients were classified as lcSSc albeit 85% had lung fibrosis. Moreover, we found a relatively small proportion of ACA-positive patients (24%) and a notable rate of ATA positivity (35%), which was rather unanticipated in a group containing a majority of lcSSc. The prognosis of patients from this group was worse than those from C1 in which patients have mainly lcSSc and few organ damage, which once again support Nihtyanova et al. work [30]. Likewise, a Canadian Scleroderma Research Group (CSRG) study examined the clinical features and mortality of ATA-positive lcSSc and ACA-positive dcSSc patients. The autoantibody profile seemed to associate stronger with
demographics and visceral damage than the skin subgroup. Mortality was related with both skin and serologic profile [9]. Kranenburg et al. also indicated that lcSSc patients positive for ATA contrasted with lcSSc patients negative for ATA and dcSSc patients positive for ATA in terms of survival and organ involvement [32]. Altogether these works suggest that sub-classification combining antibody profile and skin extension might predict clinical outcomes more accurately than skin or serology
Accepted
Article
alone [9,32]. Heterogeneity of SSc patients has been discussed over a long period and many publications exist before and after the publication of LeRoy et al. describing the limited and diffuse subset [2,3,33,34]. The significance of serologic profile has also been highlighted by Patterson et al. who characterized 5 groups of patients with homogeneous clinical and organ involvement [11,12]. Significant efforts to classify patients into subsets on the basis of common clinical phenotypes, rather than through a predetermined decision process, have proposed to gather individuals using changes in mRSS over time [34,35], changes in the percent predicted FVC [36,37], or gene expression patterns in the skin [38,39]. Each of these attempts has issued in a small number of subsets that define the range of phenotypes captured by the stratification characteristics [12]. There is a growing attention for a next sub-classification combining patterns of underlying pathogenesis, organ damage and prognosis, in order to personalize disease management and ameliorate outcomes [12,31].
This study presents strengths and limitations. The principal forces are the number of patients recorded in this large prospective and multicentric cohort, and the without any a priori method utilized. The main weakness is that several clinically relevant variables were lacking or were
disregarded considering that the proportion of missing data was too high (e.g. autoantibodies other than ACA/ATA, extension of ILD on high resolution CT-scan, detailed skin involvement, overlap syndromes). In addition, we had to exclude 1505/8432 patients in the cluster analysis because of missing data in any of the selected clustering variables. As those excluded patients were slightly less severe than the included ones, it could affect the extrapolation of our results. Imputation of missing data by a model-based clustering was not performed because we could not assume that these data were missing at random [40,41]. Moreover, several definitions of variables lacked precision (e.g. ILD definitions (X-ray lung fibrosis where high resolution CT-scan is widely used nowadays) and PH (suspicion by cardiac echo but no invasive confirmation)). We also acknowledge that data are missing for a thorough analysis of treatment regimens. Nevertheless, we were able to find for a majority of patients whether or not they had been taking an immunosuppressive drug. Regarding the potential effect of these drugs on survival, immunosuppressive treatment was included as an
Accepted
Article
adjusting variable in the survival analyses. A potentially important bias is the influence of duration of disease on the clustering process, since the frequency of organ damage tends to rise as the disorder progresses. Also, the disease duration at the enrollment visit was relatively high raising the
possibility that study results have been influenced by survival bias. Yet, the sensitivity analyses including only patients with a long disease duration or only patients with a short disease duration yielded similar results. Another limitation is that a significant number of patients were dismissed from the survival analysis because of lost to follow-up. Nevertheless, this exclusion did not alter the survival differences between clusters in a sensitivity analysis. The primary aim of our study was not to assess the prognosis factors of survival in SSc, but to decipher the heterogeneity of SSc by a cluster analysis and describe the survival of the clusters obtained – allowing us to post-hoc validate this approach. In studies assessing the prognosis factors of survival, baseline data are most often used. In our study, we had to include follow-up data in order to identify the occurrence of organ involvements. Therefore we considered that an organ complication was present if the corresponding variable was collected “ ” f
patient. We did not describe the progression of organ involvements in the whole population neither in the different clusters because the limited number of follow-up visits precluded us from
performing a precise temporal description. In the end, the weak reproducibility of the exploratory analysis with 6 clusters restrain the rapid transposition of these results to a new sub-classification (e.g. to allocate an individual to a designated group on the basis of their features). Moreover, earlier studies have shown differences between distinct geographical cohorts [42]. Of note, 95% of patients included in this study were Caucasian. It is likely that inclusion of a higher proportion of Asian or African American patients could have modified the results.
Conclusions
In conclusion, this work corroborate that SSc is a very heterogeneous condition. While a consensus stands on the relevance and practicality of SSc sub-classification into lcSSc and dcSSc, this binary
Accepted
Article
system might omit a wider spectrum of clinical phenotypes characterized not only by skin involvement but also by organ damage, serologic profile and subsequent prognosis. There is an increasing demand for a future SSc classification combining these different patterns, in order to personalize approach to diagnosis and clinical management.
Competing interests
Dr. Sobanski reports consultancies and speaking fees from Grifols (less than $10,000) and research support from Actelion, Grifols, GSK, Octapharma, Pfizer, Shire, outside the submitted work; Dr. Giovannelli has nothing to disclose; Dr. Allanore has nothing to disclose; Dr. Riemekasten has
nothing to disclose; Dr. Airo' has nothing to disclose; Dr. Vettori has nothing to disclose; Dr. Cozzi has nothing to disclose; Dr. Distler reports consultancies and speaking fees from Actelion, Bayer,
BiogenIdec, Böhringer Ingelheim, ChemomAb, espeRare foundation, Genentech, GSK, Inventiva, Lilly, Medac, MedImmune, Mitsubishi Tanabe Pharma, Pharmacyclics, Novartis, Pfizer, Sanofi, Sinoxa, UCB, Roche (less than $10,000 each), and research support from Actelion, Bayer, Böhringer Ingelheim, Mitsubishi Tanabe Pharma and Roche, outside the submitted work; In addition, Dr. Distler has a patent mir-29 for the treatment of systemic sclerosis licensed; Dr. Matucci Cerinic has nothing to disclose; Dr. Denton reports consultancies and speaking fees from Actelion, Boehringer Ingelheim, Bayer, GSK, Inventiva, Roche, Sanofi-Aventis (less than $10,000 each), and research support from Bayer, CSL Behring, GSK, Inventiva, Roche, outside the submitted work; Dr. Launay reports research support from Actelion, GSK, Octapharma, Pfizer, Shire, outside the submitted work; Dr. Hachulla has nothing to disclose.
Accepted
Article
Contributorship
Criterion 1: a) Substantial contributions to study conception and design; and/or b) Substantial contributions to acquisition of data; and/or c) Substantial contributions to analysis and interpretation of data.
Criterion 2: Drafting the article or revising it critically for important intellectual content
Criterion 3: Final approval of the version of the article to be published
Vincent Sobanski (1a, 1b, 1c, 2, 3), Jonathan Giovannelli (1a, 1b, 1c, 2, 3), Yannick Allanore (1a, 1b, 1 , 2, 3), G b k (1b, 2, 3), A ’ (1b, 2, 3), Serena Vettori (1b, 2, 3), Franco Cozzi (1b, 2, 3), Oliver Distler (1b, 1c, 2, 3), Marco Matucci-Cerinic (1b, 1c, 2, 3), Christopher Denton (1b, 1c, 2, 3), David Launay (1a, 1b, 1c, 2, 3), Eric Hachulla (1a, 1b, 1c, 2, 3)
Acknowledgments:
EUSTAR Collaborators (numerical order of centres): Marco Matucci Cerinic, Serena Guiducci, Department of Medicine, Section of Rheumatology, University of Florence, Italy; Ulrich Walker, Department of Rheumatology, University Hospital Basel, Switzerland; Giovanni Lapadula, Florenzo Iannone, Rheumatology Unit-DiMIMP, School of Medicine University of Bari, Italy; Radim Becvar, Institute of Rheumatology, 1st Medical School, Charles University, Prague, Czech Republic; Stanislaw Sierakowsky, Department of Rheumatology and Internal Medicine, Medical University of Bialystok, Bialystok, Poland; Maurizio Cutolo, Alberto Sulli, Research Laboratory and Division of Rheumatology, Department of Internal Medicine, University of Genova, Italy; Gabriele Valentini, Giovanna Cuomo, S V , D f Ex M ‘F-M ’ II , f Rheumatology, Naples, Italy; Gabriela Riemekasten, Department of Rheumatology, University of Lübeck, Germany; Elise Siegert, Department of Rheumatology, Charitè University Hospital, Berlin, German Rheumatism Research Centre Berlin (DRFZ), a Leibniz Institute, Germany; Simona Rednic, I , D f y, y f M y ‘I ’
Accepted
Article
Cluj, Cluj-Napoca, Romania; André Kahan, Department of Rheumatology, University Paris Descartes and Cochin Hospital, Paris, France; P. Vlachoyiannopoulos, Department of Pathopysiology, Medical , y f A , G ; . M , b , ’ O tiva e Cattedra di Reumatologia, IRCCS Policlinico S Matteo, Pavia, Italy; Patricia E. Carreira, Division of Rheumatology, Hospital 12 de Octubre, Madrid, Spain; Srdan Novak, Department of Rheumatology and Clinical Immunology, Internal Medicine, KBC Rijeka, Croatia; László Czirják, Cecilia Varju, Department of Immunology and Rheumatology, Faculty of Medicine, University of Pécs, Hungary; Carlo Chizzolini, Department of Immunology and Allergy, University Hospital, Geneva, Switzerland; Eugene J. Kucharz, Anna Kotulska, Magdalena Kopec-Medrek, Malgorzata Widuchowska,
Department of Internal Medicine and Rheumatology, Medical University of Silesia, Katowice, Poland; Franco Cozzi, Rheumatology Unit, Department of Clinical and Experimental Medicine, University of Padova, Italy; Blaz Rozman, University Medical Center Ljublijana, Division of Internal Medicine, D f y, Lj b , ; M , B , ‘ M ’, Balzan, Malta; Armando Gabrielli, Dipartimento di Scienze Cliniche e Molecolari, Clinica Medica, Università Politecnica delle Marche, Ancona, Italy; Dominique Farge, Chen Wu, Zora Marjanovic, Helene Faivre, Darin Hij, Roza Dhamadi, Department of Internal Medicine, Hospital Saint-Louis, Paris, France; Paolo Airò, Spedali Civili di Brescia, Servizio di Reumatologia Allergologia e Immunologia Clinica, Brescia, Italy; Roger Hesselstrand, Frank Wollheim, Dirk M Wuttge, Kristofer Andréasson, Department of Rheumatology, Lund University, Lund, Sweden; Duska Martinovic, Department of Internal Medicine, Clinical Hospital of Split, Croatia; Alexandra Balbir-Gurman, Yolanda Braun-Moscovici, B. Shine Rheumatology Unit, Rambam Health Care Campus, Rappaport Faculty of Medicine, Technion, Haifa, Israel; F. Trotta, Andrea Lo Monaco, Department of Clinical and Experimental Medicine, Rheumatology Unit, University of Ferrara, Italy; Nicolas Hunzelmann, Department of Dermatology, University Hospital Cologne, Germany; Raffaele Pellerito, Ospedale Mauriziano, Centro di Reumatologia, Torino, Italy; Lisa Maria Bambara, Paola Caramaschi, Unità di Reumatologia, AOUI, Verona Italy; Carol Black, Christopher Denton, Centre for Rheumatology, Royal
Accepted
Article
Free and University College London Medical School, Royal Free Campus, United Kingdom; Nemanja Damjanov, Institute of Rheumatology, Belgrade, Serbia and Montenegro; Jörg Henes, Medizinische Universitätsklinik, Abt. II (Onkologie, Hämatologie, Rheumatologie, Immunologie, Pulmonologie), Tübingen, Germany; Vera Ortiz Santamaria, Rheumatology Granollers General Hospital, Barcelona, Spain; Stefan Heitmann, Department of Rheumatology, Marienhospital Stuttgart, Germany; Dorota Krasowska, Department of Dermatology, Medical University of Lublin, Poland; Matthias Seidel, Medizinische Universitäts-Poliklinik, Department of Rheumatology, Bonn, Germany; Harald
Burkhardt, Andrea Himsel, Klinikum der Johann Wolfgang Goethe Universität, Medizinische Klinik III, Rheumatologische Ambulanz, Frankfurt am Main, Germany; Maria J. Salvador, José Antonio Pereira Da Silva, Rheumatology Department, Hospitais da Universidade, Coimbra, Portugal; Bojana
Stamenkovic, Aleksandra Stankovic, Institute for Prevention, Treatment and Rehabilitation of Rheumatic and Cardiovascular Diseases, Niska Banja, Serbia and Montenegro; Mohammed Tikly, Rheumatology Unit, Department of Medicine Chris Hani Haragwanath, Hospital and University of the Witwatersrand, Johannesburg, South Africa; Lidia P. Ananieva, Lev N. Denisov, Institute of
Rheumatology, Russian Academy of Medical Science, Moscow, Russia; Ulf Müller-Ladner, Marc Frerix, Ingo Tarner, Justus-Liebig University Giessen, Department of Rheumatology and Clinical Immunology, Kerckhoff-Klinik Bad Nauheim, Germany; Raffaella Scorza, U.O. Immunologia Clinica, Centro di Riferimento per le Malattie Autoimmuni Sistemiche, Milano, Italy; Merete Engelhart, Gitte Strauss, Henrik Nielsen, Kirsten Damgaard, Department of Rheumatology, University Hospital of Gentofte, Hellerup, Denmark; Antonio Zea Mendoza, Carlos de la Puente, Sifuentes Giraldo WA, Servicio de Reumatología, Hospital Ramon Y Cajal, Madrid, Spain; Øyvind Midtvedt, Silje Reiseter, Department of Rheumatology, Rikshospitalet University Hospital, Oslo, Norway; Eric Hachulla, David Launay, Department of Internal Medicine, Hôpital Claude Huriez, Lille cedex, France; Guido Valesini, V , D f I M M , ‘ z ’ y f Rome, Italy; Ruxandra Maria Ionescu, Daniela Opris, Laura Groseanu, Department of Rheumatology, St. Mary Hospital, Carol Davila, University of Medicine and Pharmacy, Bucharest, Romania; Roxana
Accepted
Article
Sfrent Cornateanu, Razvan Ionitescu, Ana Maria Gherghe, Alina Soare, Marilena Gorga, Mihai Bojinca, Department of Internal Medicine and Rheumatology, Cantacuzino Hospital, Carol Davila University of Medicine and Pharmacy, Bucharest, Romania; Georg Schett, Jörg HW Distler, Christian Beyer, Department of Internal Medicine 3, University Hospital Erlangen, Germany; Pierluigi Meroni, Francesca Ingegnoli, Division of Rheumatology, Istituto Gaetano Pini, Department of Clinical Sciences and Community Health, University of Milano, Milano, Italy; Luc Mouthon, Department of Internal Medicine, Hôpital Cochin, Paris, France; Filip De Keyser, Vanessa Smith, University of Ghent, Department of Rheumatology, Gent, Belgium; Francesco P. Cantatore, Ada Corrado, U.O.
Reumatologia- F , O ‘ . D'A z ’, F , I y; M . zz , Dipartimento di Medicina, Ospedale San Gerardo, Monza, Italy; Kilian Eyerich, Rüdiger Hein,
Elisabeth Knott, Department of Dermatology and Allergy of the TU Munich, Germany; Piotr Wiland, Magdalena Szmyrka-Kaczmarek, Renata Sokolik, Ewa Morgiel, Marta Madej, Department of
Rheumatology and Internal Diseases, Wroclaw University of Medicine, Wroclaw, Poland; Brigitte Krummel-Lorenz, Petra Saar, Endokrinologikum Frankfurt, Germany; Martin Aringer, Claudia Günther, Division of Rheumatology, Department of Medicine III/Department of Dermatology, University Medical Center Carl Gustav Carus, Technical University of Dresden, Germany; Rene Westhovens, Ellen de Langhe, Jan Lenaerts, Catholic University of Leuven, Department of
Rheumatology, Leuven, Belgium; Branimir Anic, Marko Baresic, Miroslav Mayer, University Hospital Centre Zagreb, Division of Clinical Immunology and Rheumatology, Department of Medicine, Zagreb, Croatia; Sebastião C. Radominski, Carolina de Souza Müller, Valderílio F. Azevedo, Hospital de Clínicas da Universidade Federal do Paraná, Curitiba - Paraná, Brasil; Svetlana Agachi, Liliana Groppa, Lealea Chiaburu, Eugen Russu, Sergei Popa, Municipal Centres of Research in Scleroderma, Hospital ‘ T y’, D f y/D f y, b Hospital, Chisinau, Republic of Moldova; Thierry Zenone, Department of Medicine, Unit of Internal Medicine, Valence cedex 9, France; Simon Stebbings, John Highton, Dunedin School of Medicine, D , w Z ; L , , J O’D , D f M ,
Accepted
Article
University of Otago, Christchurch, New Zealand; Kamal Solanki, Alan Doube, Waikato University Hospital, Rheumatology Unit, Hamilton City, New Zealand; Douglas Veale, Marie O'Rourke,
Department of Rheumatology, Bone and Joint Unit, St. Vincent's University Hospital, Dublin, Ireland; Esthela Loyo, Reumatologia e Inmunologia Clinica, Hospital Regional Universitario Jose Ma Cabral y Baez, Clinica Corominas, Santiago, Dominican Republic; Mengtao Li, Department of Rheumatology, Peking Union Medical College Hospital (West Campus), Chinese Academy of Medical Sciences, Beijing, China; Edoardo Rosato, Antonio Amoroso, Antonietta Gigante, Centro per la Sclerosi Sistemica - Dipartimento di Medicina Clinica, Università La Sapienza, Policlinico Umberto I, Roma, Italy; Cristina-Mihaela Tanaseanu, Monica Popescu, Alina Dumitrascu, Isabela Tiglea, Clinical Emergency Hospital St. Pantelimon, Bucharest, Romania; Rosario Foti, U.O. di Reumatologia, A.O.U. Policlinico Vittorio Emanuele, Catania, Italy; Rodica Chirieac, Codrina Ancuta, Division of
Rheumatology and Rehabilitation GR.T.Popa, Center for Biomedical Research, European Center for Translational Research, "GR.T.Popa" University of Medicine and Pharmacy, Rehabilitation Hospital, Iasi, Romania; Peter Villiger, Sabine Adler, Department of Rheumatology and Clinical
Immunology/Allergology, Inselspital, University of Bern, Switzerland; Paloma García de la Peña Lefebvre, Silvia Rodriguez Rubio, Marta Valero Exposito, Hospital Universitario Madrid Norte Sanchinarro, Madrid, Spain; Jean Sibilia, Emmanuel Chatelus, Jacques Eric Gottenberg, Hélène Chifflot, University Hospital of Strasbourg, Department of Rheumatology, Hôpital de Hautepierre, Service de Rhumatologie, Strasbourg Cedex, France; Ira Litinsky, Department of Rheumatology, Tel Aviv Sourasky Medical Center, Tel Aviv, Israel; Algirdas Venalis, Irena Butrimiene, Paulius Venalis, Rita Rugiene, Diana Karpec, State Research Institute for Innovative Medicine, Vilnius University, Vilnius, Lithuania; Lesley Ann Saketkoo, Joseph A. Lasky, Tulane University Lung Center,
Tulane/University Medical Center Scleroderma and Sarcoidosis Patient Care and Research Center, New Orleans, USA; Eduardo Kerzberg, Fabiana Montoya, Vanesa Cosentino, Osteoarticular Diseases and Osteoporosis Centre, Pharmacology and Clinical Pharmacological Research Centre, School of medicine, University of Buenos Aires, Rheumatology and Collagenopathies Department, Ramos
Accepted
Article
Mejía Hospital, Buenos Aires, Argentina; Massimiliano Limonta, Antonio Luca Brucato, Elide Lupi, USSD Reumatologia, Ospetali Riuniti di Bergamo, Italy; François Spertini, Camillo Ribi, Guillaume Buss, Department of Rheumatology, Clinical Immunology and Allergy, Lausanne, Switzerland; Jean Louis Pasquali, Thierry Martin, Audrey Gorse, Clinical Immunology and Internal Medicine, National Referral Center for Systemic Autoimmune Diseases, Strasbourg, France.
Funding info: no specific funding for this study
Ethical approval information: Local ethics committee.
Data sharing statement: Data may be available from EUSTAR upon separate scientific request.
References
1 van den Hoogen F, Khanna D, Fransen J, et al. 2013 classification criteria for systemic sclerosis: an American college of rheumatology/European league against rheumatism collaborative initiative. 2013. 1747–55. doi:10.1136/annrheumdis-2013-204424
2 Johnson SR, Soowamber ML, Fransen J, et al. There is a need for new systemic sclerosis subset criteria. A content analytic approach. Scandinavian Journal of Rheumatology 2017;00:1–9. doi:10.1080/03009742.2017.1299793
3 Wollheim FA. Classification of systemic sclerosis. Visions and reality. Rheumatology 2005;44:1212–6. doi:10.1093/rheumatology/keh671
4 LeRoy EC, Black C, Fleischmajer R, et al. Scleroderma (systemic sclerosis): classification, subsets and pathogenesis. The Journal of rheumatology 1988;15:202–5.
Accepted
Article
5 Komócsi A, Vorobcsuk A, Faludi R, et al. The impact of cardiopulmonary manifestations on the mortality of SSc: a systematic review and meta-analysis of observational studies. Rheumatology (Oxford, England) 2012;51:1027–36. doi:10.1093/rheumatology/ker357
6 Tyndall AJ, Bannert B, Vonk M, et al. Causes and risk factors for death in systemic sclerosis: a study from the EULAR Scleroderma Trials and Research (EUSTAR) database. Annals of the Rheumatic Diseases 2010;69:1809–15. doi:10.1136/ard.2009.114264
7 Medsger TA. Natural history of systemic sclerosis and the assessment of disease activity, severity, functional status, and psychologic well-being. Rheumatic Diseases Clinics of North America 2003;29:255–73–vi.
8 Nihtyanova SI, Schreiber BE, Ong VH, et al. Prediction of pulmonary complications and long-term survival in systemic sclerosis. Arthritis & Rheumatology 2014;66:1625–35.
doi:10.1002/art.38390
9 Srivastava N, Hudson M, Tatibouet S, et al. Thinking outside the box—The associations with cutaneous involvement and autoantibody status in systemic sclerosis are not always what we expect. YSARH 2015;45:1–6. doi:10.1016/j.semarthrit.2015.04.009
10 Sobanski V, Dauchet L, Lefèvre G, et al. Prevalence of Anti–RNA Polymerase III Antibodies in Systemic Sclerosis: New Data From a French Cohort and a Systematic Review and
M ‐A y . Arthritis & Rheumatology 2014;66:407–17. doi:10.1002/art.38219
11 Patterson KA, Roberts-Thomson PJ, Lester S, et al. Interpretation of an extended auto-antibody profile in a well characterised Australian systemic sclerosis (scleroderma) cohort utilising Principle Components Analysis. Arthritis & Rheumatology 2015;67:3234–44.
doi:10.1002/art.39316
12 Ligon CB, Wigley FM. Editorial: Scleroderma: Bringing a Disease From Black-and-White Into Technicolor. Arthritis & Rheumatology 2015;67:3101–3. doi:10.1002/art.39310
13 Richette P, Clerson P, Périssin L, et al. Revisiting comorbidities in gout: a cluster analysis. Annals of the Rheumatic Diseases 2013;74:142–7. doi:10.1136/annrheumdis-2013-203779
14 Ahmad T, Pencina MJ, Schulte PJ, et al. Clinical implications of chronic heart failure phenotypes defined by cluster analysis. Journal of the American College of Cardiology 2014;64:1765–74. doi:10.1016/j.jacc.2014.07.979
15 Bourdin A, Molinari N, Vachier I, et al. Prognostic value of cluster analysis of severe asthma phenotypes. The Journal of Allergy and Clinical Immunology 2014;134:1043–50.
doi:10.1016/j.jaci.2014.04.038
16 Szodoray P, Hajas A, Kardos L, et al. Distinct phenotypes in mixed connective tissue disease: subgroups and survival. Lupus 2012;21:1412–22. doi:10.1177/0961203312456751
Accepted
Article
17 Mahr A, Katsahian S, Varet H, et al. Revisiting the classification of clinical phenotypes of anti-neutrophil cytoplasmic antibody-associated vasculitis: a cluster analysis. Annals of the Rheumatic Diseases 2013;72:1003–10. doi:10.1136/annrheumdis-2012-201750
18 Ingegnoli F, Ardoino I, Boracchi P, et al. Nailfold capillaroscopy in systemic sclerosis: data from the EULAR scleroderma trials and research (EUSTAR) database. Microvasc Res 2013;89:122–8. doi:10.1016/j.mvr.2013.06.003
19 Leclair V, Hudson M, Proudman SM, et al. Subsets in systemic sclerosis: one size does not fit all. JSRD 2016;1:298–306. doi:10.5301/jsrd.5000212
20 Tyndall A, Mueller-Ladner U, Matucci-Cerinic M. Systemic sclerosis in Europe: first report from the EULAR Scleroderma Trials And Research (EUSTAR) group database. Annals of the
Rheumatic Diseases. 2005;64:1107–7. doi:10.1136/ard.2005.036038
21 Walker UA, Tyndall A, Czirják L, et al. Clinical risk assessment of organ manifestations in systemic sclerosis: a report from the EULAR Scleroderma Trials And Research group database. Annals of the Rheumatic Diseases 2007;66:754–63. doi:10.1136/ard.2006.062901
22 Meier FMP, Frommer KW, Dinser R, et al. Update on the profile of the EUSTAR cohort: an analysis of the EULAR Scleroderma Trials and Research group database. Annals of the Rheumatic Diseases 2012;71:1355–60. doi:10.1136/annrheumdis-2011-200742
23 Preliminary criteria for the classification of systemic sclerosis (scleroderma). Subcommittee for scleroderma criteria of the American Rheumatism Association Diagnostic and Therapeutic Criteria Committee. Arthritis & Rheumatism 1980;23:581–90.
24 Park JW, Ahn GY, Kim J-W, et al. Impact of EUSTAR standardized training on accuracy of modified Rodnan skin score in patients with systemic sclerosis. International Journal of Rheumatic Diseases Published Online First: 5 November 2018. doi:10.1111/1756-185X.13433 25 Kowal-Bielecka O, Fransen J, Avouac J, et al. Update of EULAR recommendations for the
treatment of systemic sclerosis. 2017. 1327–39. doi:10.1136/annrheumdis-2016-209909 26 Han L, Benseler SM, Tyrrell PN. Cluster and Multiple Correspondence Analyses in
Rheumatology: Paths to Uncovering Relationships in a Sea of Data. Rheumatic Diseases Clinics of North America 2018;44:349–360.e29. doi:10.1016/j.rdc.2018.01.013
27 Hennig C. Cluster-wise assessment of cluster stability. Computational Statistics & Data Analysis 2007;52:258–71. doi:10.1016/j.csda.2006.11.025
28 Hennig C. Dissolution point and isolation robustness: Robustness criteria for general cluster analysis methods. Journal of Multivariate Analysis 2008;99:1154–76.
doi:10.1016/j.jmva.2007.07.002
Accepted
Article
30 Nihtyanova SI, Schreiber BE, Ong VH, et al. Prediction of pulmonary complications and long-term survival in systemic sclerosis. Arthritis & Rheumatology 2014;66:1625–35.
doi:10.1002/art.38390
31 Varga J, Hinchcliff M. Connective tissue diseases: systemic sclerosis: beyond limited and diffuse subsets? Nature Publishing Group 2014;10:200–2. doi:10.1038/nrrheum.2014.22
32 Kranenburg P, van den Hombergh WMT, Knaapen-Hans HKA, et al. Survival and organ involvement in patients with limited cutaneous systemic sclerosis and anti-topoisomerase-I antibodies: determined by skin subtype or auto-antibody subtype? A long-term follow-up study. Rheumatology (Oxford, England) Published Online First: 12 August 2016.
doi:10.1093/rheumatology/kew298
33 Giordano M, Valentini G, Migliaresi S, et al. Different antibody patterns and different prognoses in patients with scleroderma with various extent of skin sclerosis. The Journal of rheumatology 1986;13:911–6.
34 Perera A, Fertig N, Lucas M, et al. Clinical subsets, skin thickness progression rate, and serum antibody levels in systemic sclerosis patients with anti–topoisomerase I antibody. Arthritis & Rheumatism 2007;56:2740–6. doi:10.1002/art.22747
35 Shand L, Lunt M, Nihtyanova S, et al. Relationship between change in skin score and disease outcome in diffuse cutaneous systemic sclerosis: application of a latent linear trajectory model. Arthritis & Rheumatism 2007;56:2422–31. doi:10.1002/art.22721
36 Man A, Davidyock T, Ferguson LT, et al. Changes in forced vital capacity over time in systemic sclerosis: application of group-based trajectory modelling. Rheumatology 2015;54:1464–71. doi:10.1093/rheumatology/kev016
37 Schulam P, Wigley F, Saria S. Clustering Longitudinal Clinical Marker Trajectories from Electronic Health Data: Applications to Phenotyping and Endotype Discovery. AAAI 2015. 38 Pendergrass SA, Lemaire R, Francis IP, et al. Intrinsic Gene Expression Subsets of Diffuse
Cutaneous Systemic Sclerosis Are Stable in Serial Skin Biopsies. J Investig Dermatol 2012;132:1363–73. doi:10.1038/jid.2011.472
39 Hinchcliff M, Huang C-C, Wood TA, et al. Molecular Signatures in Skin Associated with Clinical Improvement during Mycophenolate Treatment in Systemic Sclerosis. J Investig Dermatol 2013;133:1979–89. doi:10.1038/jid.2013.130
40 Little RJA, Rubin DB. Statistical Analysis with Missing Data. Hoboken, NJ, USA: : John Wiley & Sons, Inc. 2002. doi:10.1002/9781119013563
41 Storlie CB, Myers SM, Katusic SK, et al. Clustering and variable selection in the presence of mixed variable types and missing data. Statistics in Medicine Published Online First: 17 May 2018. doi:10.1002/sim.7697
Accepted
Article
42 Walker UA, Tyndall A, Czirják L, et al. Geographical variation of disease manifestations in systemic sclerosis: a report from the EULAR Scleroderma Trials and Research (EUSTAR) group database. Annals of the Rheumatic Diseases 2009;68:856–62. doi:10.1136/ard.2008.091348
Accepted
Article
Figure Legends
Figure 1. A. Dendrogram of the 6,927 patients included in the cluster analysis. The length of vertical lines represents the degree of similarity between patients. B. Heat map showing the clinical
characteristics in each cluster.
Figure 2. A. Main characteristic of the 2 clusters (A) and (B). B. Representative proportions of the main clinical characteristics for each cluster. C. Kaplan-Meier survival curves of the 2 clusters. D. Forrest plot showing mortality hazard ratios and 95% confidence intervals in the 2 clusters.
Figure 3. A. Main characteristics of the 6 clusters C1 to C6. B. Representative proportions of the main clinical characteristics for each cluster. C. Kaplan-Meier survival curves of the 6 clusters. D. Forrest plot showing mortality hazard ratios and 95% confidence intervals in the 6 clusters.
Accepted
Article
This article is protected by copyright. All rights reserved.
Table 1. Characteristics of patients (i) analyzed and not analyzed, and (ii) by cutaneous subset.
EUSTAR population Study population
Patients analyzed (n=6927)
Patients not analyzed (n=1505) dcSSc (42%) lcSSc (58%) N % or mean±SD N % or mean±SD p % or mean±SD % or mean±SD p DEMOGRAPHICS Sex, female 6924 86 1505 83 <0.001 80 91 <0.001 Ethnicity, white/asian/black 3973 95/3/2 1176 87/11/2 <0.001 92/5/3 97/2/1 <0.001 Age, years 6927 58.7±13.2 1505 56.3±13.9 <0.001 55.6±13.0 60.9±13.0 <0.001
Age at first non- y ’ y , years 6927 47.3±13.3 1505 47.6±14.1 0.474 45.6±13.2 48.5±13.3 <0.001
Disease duration, years 6927 11.4±8.1 1505 8.7±8.1 <0.001 10.0±7.4 12.4±8.5 <0.001
Time between, years:
y ’ and first non- y ’ 5868 3.9±8.0 1351 3.4±8.1 <0.001 2.0±5.6 5.2±9.2 <0.001 First non- y ’ and EUSTAR enrolment 4875 9.4±7.8 1271 7.8±7.8 <0.001 8.0±7.3 10.3±8.1 <0.001 EUSTAR enrolment and last visit 4875 2.6±2.5 1271 0.8±1.7 <0.001 2.7±2.6 2.5±2.5 0.031
Body mass index, kg/m2 2483 23.6±4.3 889 24.4±4.8 <0.001 22.9±4.0 24.1±4.4 <0.001
SSc CHARACTERISTICS Autoantibodies
Antinuclear antibodies, presence 6927 96 1412 94 <0.001 97 96 0.400
Anti-centromere antibodies, presence 6927 37 1264 36 0.751 14 54 <0.001
Anti-topoisomerase I antibodies, presence 6927 39 1270 36 0.028 61 23 <0.001
Anti-U1RNP antibodies, presence 4054 5 807 7 0.006 5 5 0.770
Anti-PM/Scl antibodies, presence 3335 3 648 4 0.278 5 2 <0.001
Anti-RNA polymerase III antibodies, presence 3163 4 563 6 0.025 6 3 <0.001
Cutaneous involvement
Skin involvement, dcSSc 6913 42 1437 38 0.011 - - -
mRSS, mean peak value 6927 12.0±9.2 1170 10.9±9.7 <0.001 18.3±9.8 7.5±5.2 <0.001
Gastro-intestinal involvement
Accepted
Article
This article is protected by copyright. All rights reserved.
Stomach symptoms, presence 6927 42 1491 27 <0.001 47 38 <0.001
Intestinal symptoms, presence 6927 43 1497 33 <0.001 44 42 0.027
Joint involvement
Joint contractures, presence 6927 48 1492 35 <0.001 64 36 <0.001
Joint synovitis, presence 6927 26 1496 18 <0.001 32 22 <0.001
Tendon friction rubs, presence 6927 17 1477 8 <0.001 28 9 <0.001
Vascular involvement
y ’ , presence 6927 98 1500 97 <0.001 98 98 0.340
Digital ulceration, history or active 6927 49 1491 35 <0.001 58 42 <0.001
Muscular involvement
Muscle weakness, presence 6927 39 1488 24 <0.001 47 33 <0.001
Muscle atrophy, presence 6927 22 1484 12 <0.001 30 16 <0.001
CK elevation, presence 6927 13 1231 13 0.711 18 9 <0.001
Cardiac involvement
Systemic arterial hypertension, presence 6927 34 1492 27 <0.001 33 35 0.150
Palpitations, presence 6927 39 1483 26 <0.001 41 38 0.014
Conduction blocks, presence 6927 22 1152 14 <0.001 24 20 <0.001
LVEF<50%, presence 4239 5 879 5 0.799 6 4 <0.001
Abnormal diastolic function, presence 6927 33 1116 22 <0.001 34 33 0.588
Pericardial effusion, presence 4442 11 920 8 0.042 13 9 <0.001
Pulmonary hypertension
Pulmonary hypertension by echo, presence 6927 31 1173 22 <0.001 33 29 <0.001
sPAP measured by echo, mmHg 3983 34.5±15.3 727 34.2±15.1 0.041 34.8±16.4 34.2±14.5 0.013
Interstitial lung disease
Plain X-ray lung fibrosis, presence 6927 49 1033 39 <0.001 63 39 <0.001
HRCT lung fibrosis, presence 3424 57 816 53 0.023 68 48 <0.001
Restrictive defect on PFTs, presence 6927 43 1083 33 <0.001 57 32 <0.001
FVC, % predicted 4349 89.3±21.7 903 90.0±21.8 0.437 81.4±21.1 94.9±20.3 <0.001
DLCO, % predicted 6196 61.8±20.1 1026 66.1±21.1 <0.001 57.4±19.9 64.9±19.7 <0.001
6MWD, meters 1179 392±134 338 411±145 0.007 394±137 391±131 0.872
Renal involvement
Accepted
Article
This article is protected by copyright. All rights reserved.
Proteinuria, presence 6927 12 1308 10 0.082 15 9 <0.001
Blood tests
CRP elevation, presence 4736 36 1100 31 <0.001 44 30 <0.001
Hypocomplementemia, presence 4469 11 860 10 0.409 12 11 0.504
Treatment
Steroids, active or past use 4647 43 1081 38 0.006 55 34 <0.001
Prednisone, mg/day 4644 4.4 1080 5.1 0.081 6.0 3.3 <0.001
Immunosuppressive drugs, active or past use 4631 42 1085 44 0.162 60 28 <0.001
Clustering variables are underlined.
b % ± D. f w ’ f b F ’ exact test for categorical variables.
N: number of patients with available data; disease duration: time between the first non- y ’ ; mRSS: mean Rodnan skin score; oesophageal symptoms: dysphagia and/or reflux; stomach symptoms: early satiety and/or vomiting; intestinal symptoms: diarrhoea, bloating and/or constipation; HRCT: high resolution CT-scan; PFTs: pulmonary function tests; FVC: forced vital capacity; DLCO: diffusing capacity of the lung for carbon monoxide; 6MWD: 6-minute walking distance; CRP: C-reactive protein
Accepted
Article
This article is protected by copyright. All rights reserved.
Table 2. Characteristics of patients for the 2 and 6 clusters found in the cluster analysis (n=6927).
2 clusters 6 clusters C(A) C(B) C1 C2 C3 C4 C5 C6 Jaccard index 0.64 0.66 0.39 0.32 0.57 0.38 0.68 0.43 N 3149 3778 1186 720 1243 1673 1249 856 DEMOGRAPHICS Sex, female 90 84 94 88 88 88 81 79 Ethnicity, white/asian/black 94/5/2 96/2/2 97/2/1 88/10/2 94/4/2 96/2/2 94/3/3 96/2/2 Age, years 59.2±13.3 58.2±13.2 61.3±12.9 60±12.8 56.6±13.5 61.2±12.6 55.8±13.2 55.9±13.2
Age at first non- y ’ y , years 47.9±13.3 46.7±13.3 48.9±13.1 48.3±12.8 46.7±13.6 48.1±13.1 46±13.4 45.1±13.4
Disease duration, years 11.3±8.2 11.5±8.1 12.4±8.1 11.8±8.3 9.9±7.9 13.2±8.4 9.8±7.6 10.8±7.5
Time between, years:
y ’ first non- y ’ 4.8±8.7 3.1±7.3 5.4±8.7 4.4±9.1 4.4±8.5 3.9±8.2 2.8±6.6 2.2±6.1 First non- y ’ and EUSTAR enrolment 9.4±7.9 9.3±7.8 10.3±7.9 9.8±8.2 8.2±7.4 10.5±8.1 8.1±7.4 8.6±7.4 EUSTAR enrolment and last visit 2.2±2.3 2.8±2.6 2.5±2.3 2.3±2.5 1.8±2.2 3±2.7 2.4±2.5 2.9±2.5
Body mass index, kg/m2 24.1±4.3 23.2±4.2 24.3±4.4 24.5±4.6 23.6±4 23.6±4.4 23.3±3.9 22.1±4.2
SSc CHARACTERISTICS Autoantibodies
Antinuclear antibodies, presence 96 97 98 94 95 97 95 98
Anti-centromere antibodies, presence 54 22 79 24 48 29 20 12
Anti-topoisomerase I antibodies, presence 21 54 8 35 24 46 50 77
Anti-U1RNP antibodies, presence 5 5 3 8 5 7 3 4
Anti-PM/Scl antibodies, presence 2 4 1 3 1 4 4 6
Anti-RNA polymerase III antibodies, presence 3 5 2 3 4 3 6 6
Accepted
Article
This article is protected by copyright. All rights reserved.
Skin involvement, dcSSc 19 61 11 29 21 37 72 92
mRSS, mean peak value 6.6±4.3 16.5±9.8 6.6±4.2 7.2±4.6 6.3±4.1 9.2±5.3 19±6.7 27.2±8.7
Gastro-intestinal involvement
Oesophageal symptoms, presence 73 88 88 76 58 91 79 95
Stomach symptoms, presence 26 55 52 16 7 60 36 70
Intestinal symptoms, presence 33 50 64 21 11 57 34 63
Joint involvement
Joint contractures, presence 24 67 29 17 23 65 55 91
Joint synovitis, presence 14 37 15 13 15 37 25 53
Tendon friction rubs, presence 4 28 6 3 4 19 19 57
Vascular involvement
y ’ , presence 98 99 99 98 97 99 98 99
Digital ulceration, history or active 32 63 35 24 33 62 50 85
Muscular involvement
Muscle weakness, presence 16 59 27 8 10 69 33 77
Muscle atrophy, presence 6 35 9 3 6 38 17 57
CK elevation, presence 6 18 7 7 5 17 13 26
Cardiac involvement
Systemic arterial hypertension, presence 31 37 38 28 26 44 26 38
Palpitations, presence 25 51 38 32 9 64 28 57
Conduction blocks, presence 12 30 16 14 6 39 16 34
LVEF<50%, presence 3 7 3 3 2 6 5 10
Abnormal diastolic function, presence 24 42 27 33 15 54 24 43
Pericardial effusion, presence 7 14 7 11 4 15 9 18
Pulmonary hypertension by echo, presence 21 39 24 39 8 44 24 50
sPAP measured by echo, mmHg 32.5±13.7 36±16.2 33±14.3 36.7±14.1 29.4±12 37.2±14.6 32.4±12 38.1±22.1 Interstitial lung disease
Plain X-ray lung fibrosis, presence 29 65 8 85 17 72 46 80
Accepted
Article
This article is protected by copyright. All rights reserved.
Restrictive defect on PFTs, presence 24 58 13 61 14 60 42 77
FVC, % predicted 97.8±19.3 82.7±21.1 101.2±17.4 86.7±21.9 99.9±17.7 84.4±20.8 87.5±19.8 72.8±20.3
DLCO, % predicted 68±18.9 56.6±19.7 69.8±17.2 57.7±19.3 72.3±18 55.2±18.8 62.5±20.3 50.6±18.1
6MWD, meters 411±129 381±136 400±135 405±130 427±121 366±133 418±130 362±138
Renal involvement
Renal crisis, history 2 4 2 1 2 4 3 8
Proteinuria, presence 7 16 6 8 7 15 11 26
Blood tests
CRP elevation, presence 24 45 25 29 20 43 36 62
Hypocomplementemia, presence 10 13 13 7 8 14 10 12
Treatment
Steroids, active or past use 27 55 22 45 24 57 44 65
Prednisone, mg/day 2.8±6.4 5.7±7.9 2±4.9 5.5±9.3 2.3±5.6 5.6±7.6 4.6±7.6 7.3±8.8
Immunosuppressive drugs, active or past use 27 54 17 44 27 48 54 66
Mortality
Number of deaths per 1000 patient-years 10.3 22.6 7.5 17.3 9.7 19.1 20.8 31.9
Clustering variables are underlined. Numbers are given as % or mean±SD.
Disease duration: time between the first non- y ’ ; mRSS: mean Rodnan skin score; oesophageal symptoms: dysphagia and/or reflux; stomach symptoms: early satiety and/or vomiting; intestinal symptoms: diarrhoea, bloating and/or constipation; HRCT: high resolution CT-scan; PFTs: pulmonary function tests; FVC: forced vital capacity; DLCO: diffusing capacity of the lung for carbon monoxide; 6MWD: 6-minute walking distance; CRP: C-reactive protein
Accepted
Article
This article is protected by copyright. All rights reserved.
Table 3. Cox regression analysesUnivariable analysis
Multivariable analysis Age at disease onset Age at disease onset
Sex
Age at disease onset Sex
Immunosuppressive treatment N=3352
p N=3352 p N=3352 p N=2887 p
HR (95%CI) HR (95%CI) HR (95%CI) HR (95%CI)
lcSSc Reference Reference Reference Reference
dcSSc 1.90 (1.64-2.19) <0.001 2.39 (2.07-2.77) <0.001 2.14 (1.85-2.48) <0.001 2.03 (1.61-2.56) <0.001
C-Index* 0.60±0.01 0.73±0.01 0.75±0.01 0.78±0.02
Cluster A Reference Reference Reference Reference
Cluster B 2.23 (1.88-2.65) <0.001 2.40 (2.02-2.85) <0.001 2.26 (1.91-2.69) <0.001 2.47 (1.86-3.27) <0.001
C-Index 0.59±0.01 0.72±0.01 0.74±0.01 0.78±0.02
Cluster 1 Reference Reference Reference Reference
Cluster 2 2.32 (1.62-3.31) <0.001 2.10 (1.46-3.00) <0.001 1.97 (1.38-2.82) <0.001 1.64 (0.88-3.03) 0.119 Cluster 3 1.30 (0.89-1.91) 0.172 1.63 (1.11-2.38) 0.012 1.62 (1.11-2.37) 0.013 1.97 (1.10-3.54) 0.023 Cluster 4 2.47 (1.86-3.27) <0.001 2.49 (1.88-3.30) <0.001 2.40 (1.81-3.19) <0.001 2.77 (1.74-4.39) <0.001 Cluster 5 3.03 (2.23-4.11) <0.001 3.77 (2.77-5.12) <0.001 3.37 (2.47-4.58) <0.001 3.22 (1.93-5.36) <0.001 Cluster 6 4.40 (3.30-5.87) <0.001 5.85 (4.38-7.81) <0.001 5.20 (3.89-6.95) <0.001 6.14 (3.81-9.89) <0.001 C-Index 0.63±0.01 0.75±0.01 0.76±0.01 0.79±0.02
(disease onset: first non- y ’ , T b 8 f y y y ’ )
* C-index was calculated for each Cox regression models, and corresponds to the estimation of the probability of concordance, equivalent to the area under the receiver operating characteristic curve for logistic regression models. A value of 1 means a perfect agreement and 0.5 is an agreement that is no better than chance.
