Does regression of skin thickening predict improvement of internal organ involvement and survival in patients with diffuse cutaneous systemic sclerosis? A EUSTAR analysis

Abstract

Objective

Patients with diffuse cutaneous systemic sclerosis (dcSSc) frequently show spontaneous improvement of skin fibrosis. Our aim was to examine whether an improvement in skin fibrosis predicts lower likelihood of visceral organ progression and better survival.

Methods

Patients from the European Scleroderma Trials and Research (EUSTAR) cohort with dcSSc, baseline modified Rodnan skin score (mRSS) ≥7, and valid mRSS at 12±3 months follow up were included. Regression/progression of skin fibrosis was defined as a decrease/increase in mRSS >5 points and ≥25% from baseline to follow up. The outcomes included progression of lung, renal, cardiac and gastrointestinal manifestations using consensus derived definitions and all-cause death. Regressive, stable and progressive patients were compared by univariate, Kaplan-Meier survival curve and Cox regression analysis.

Results

Of 1257 included patients, 883 (70.2%) were stable, 282 (22.4%) regressive, and 92 (7.3%) progressive. Regressive patients, adjusted for baseline mRSS, baseline immunosuppression, baseline FVC, and disease duration, showed a significantly lower probability of FVC decline ≥10% than progressive patients (p=0.00003), lower probability of all-cause mortality during follow up (p=0.035) compared to progressive patients. .Improvement of skin fibrosis was not associated with progression of other organ manifestations.

Conclusion

We found that regression of skin fibrosis is associated with a lower probability of lung progression and better survival at follow up. The link between the disease course of skin and lung fibrosis in SSc can help to better stratify patients in clinical practice and enrich for ILD progressive patients in clinical trials.

Key messages

• Diffuse SSc patients with improvement of skin fibrosis had a lower probability of lung function progression and all-cause mortality than skin progressive patients.

• This allows better risk stratification of SSc patients in clinical practice.

• It could help to improve the design of clinical trials in SSc and better enrichment of ILD progressive patients.

Introduction

Systemic sclerosis (SSc) is a rare autoimmune disease with complex visceral organ involvement leading to substantial morbidity and mortality. As such, interstitial lung disease (ILD), pulmonary arterial hypertension (PAH), scleroderma renal crisis (SRC), and cardiac or gastrointestinal involvement can occur [1]. The identification of patients at high risk for further complications is challenging for physicians.

Skin fibrosis is a hallmark feature of SSc. The modified Rodnan skin score (mRSS) is a semi-quantitative method for the assessment of skin fibrosis. It meets the Outcome Measures in Rheumatoid Arthritis Clinical Trials (OMERACT) criteria as a validated outcome measure and a surrogate for more severe internal organ involvement and mortality in early diffuse SSc. The mRSS has been playing an important role in clinical trials and it has been the most frequently used primary endpoint [2,3,4,5]. It evaluates skin thickness on a scale from 0 (normal) to 3 (severe) at 17 different and standardized body surface areas [6]. Importantly, the mRSS is also the key component of the new revised American College of Rheumatology Composite Response Index in Systemic Sclerosis (revised ACR-CRISS), which is a strong candidate measure as a primary endpoint clinical trials [7, 8]. Consequently, changes of the mRSS strongly influence the CRISS [9].

In our previous prospective, multicentre, real-life cohort study, we have discovered that in patients with diffuse cutaneous SSc (dcSSc) skin worsening within one year is associated with decline in lung function and worse survival during follow-up. Thus, progression of mRSS could be used as a surrogate marker to predict later ILD progression and mortality [10].

However, patients with diffuse SSc are reaching their peak skin score very early in the disease course and later spontaneous regression of mRSS is seen in clinical practice even more often than skin progression. Discovering associations between regression of mRSS and consecutive changes in visceral organ function would therefore be practically relevant. It could further improve clinical management and inform cohort enrichment for clinical trials [11, 12].

Therefore, the aim of our study was to explore associations between regression of mRSS and later visceral organ involvement in dcSSc using the large, longitudinal, real-life EUSTAR registry.

Methods

Patients and study design

This was a post-hoc analysis of prospectively collected data from the European Scleroderma Trials and Research (EUSTAR) registry (a project number CP93). Patients’ visits from 1 January 2009 (because online database with extended data started in 2009) to 7 July 2019 were analyzed. The structure of the EUSTAR database and minimum essential data set have been described [13, 14].

Patients from the EUSTAR registry with dcSSc, baseline mRSS ≥ 7, valid mRSS assessment at 12 ± 3 months after baseline and additional ≥ 1 annual follow-up visits were recruited [15,16,17].

Regression of skin fibrosis was defined as decrease in mRSS > 5 and ≥ 25%, and progression of skin fibrosis as increase in mRSS > 5 and ≥ 25% from baseline to 12 ± 3 months follow up. These thresholds are based on the minimally clinical important difference for mRSS as defined previously [12, 18]. Patients were classified as stable skin fibrosis when neither a decrease nor an increase in mRSS > 5 score and ≥ 25% from baseline to 12 ± 3 months was observed. The 12 months follow up for the definition of progression or improvement of skin fibrosis was chosen, because it is the standard follow up in clinical practice and is the preferred duration of randomized controlled clinical trials.

Follow-up and outcome measures

The baseline visit was defined as the first available visit in the database. The first follow-up was defined as the visit with a time interval of 12 ± 3 months after the baseline visit. The duration of follow-up was calculated as the time between the baseline visit and the last annual visit with available data for the respective outcome.

The outcomes defining visceral organ progression were previously developed by consensus of an expert group using the nominal group technique [19]. Organ progression was defined as occurrence of one of the following events during follow-up:

• 1) ILD progression defined as decrease in forced vital capacity (FVC) ≥ 10% from baseline or decrease of FVC ≥ 5–9% combined with diffusing capacity for carbon monoxide (DLCO) ≥ 15%;

• 2) Cardiac progression defined as reduction of left ventricular ejection fraction (LVEF) to < 45% or relative decrease of LVEF > 10% or new onset of any of the following: conduction disturbance or diastolic function abnormalities assessed by echocardiography;

• 3) Intestinal progression defined as new onset of any of the following: malabsorption syndrome, paralytic ileus, weight loss > 10%, oesophageal symptoms (dysphagia, reflux), stomach symptoms (early satiety, vomiting), intestinal symptoms (diarrhoea, bloating, constipation);

• 4) New onset of scleroderma renal crisis (SRC);

• 5) All-cause death.

For the intestinal involvement (3), the analysis was performed with and without stomach symptoms (early satiety, vomiting) and intestinal symptoms (diarrhoea, bloating, constipation), because malabsorption syndrome, paralytic ileus and weight loss > 10% as severe manifestations needed to be weighted stronger, and we investigated whether this separate analysis was leading to different results.

Regarding events with multiple components, an event was defined as present if there was worsening for at least one of the components stated above. Censoring was present if there was no worsening for any of the components. Consequently, an event was missing if at least one of the components was missing at every follow-up visit and worsening was not present for any of the components. Patients with missing information for event could not be used for the survival analyses.

We compared the occurrence of the above-mentioned outcomes during follow up between regressive, stable and progressive dcSSc patients at the first 12 ± 3 months’ observation period.

Statistical analysis

Continuous data are expressed as mean ± SD and categorical data as frequency and percentage (%). Baseline variables were compared between skin regressive, stable and progressive patients by univariate analysis and with Bonferroni correction for multiple comparisons. Chi-squared tests or Fisher’s exact tests were used for categorical variables, and Kruskal-Wallis test for continuous variables. P-values < 0.05 were considered statistically significant.

As measure for median follow-up, the reverse Kaplan-Meier (KM) estimate was calculated [20]. Survival analysis was censored at 7 years, because of low number of patients under observation after 7 years. Associations between skin changes in the first 12 ± 3 months and occurrence of long-term organ outcomes were evaluated by Kaplan–Meier survival analysis and KM estimates with 95% confidence intervals based on a log-log transformation were calculated. Only the first event was considered. Log-rank tests were executed for comparison of stable, progressive and regressive patients. In case of p < 0.05 in the log rank- test, pairwise comparisons with Bonferroni-Holm correction were performed.

For ILD progression, intestinal progression, and all-cause death, a Cox regression model was performed. Log-linearity and proportional hazards assumptions were checked with spline fits and Schoenfeld residuals.

For the analysis of new onset of renal crisis, conduction blocks, and diastolic function abnormalities, patients with these conditions present at baseline were excluded from the KM analysis, as they could not show any event by definition.

The statistical analyses were accomplished by the biostatistician Nicole Graf using R programming language (V.3.3.3), packages “prodlim”, “survival”, “survminer”, “ggplot2” and “tableone”.

Results

Baseline characteristics

Out of 17,212 patients from the EUSTAR database and from 103 EUSTAR collaborative partners at time of data extraction, 1257 met the requirement criteria. Of these, 282 (22.4%) were regressive, 883 (70.2%) stable and 92 (7.3%) were progressive for skin fibrosis at 1-year follow-up. Table 1 shows a summary of the clinical and demographic characteristics of the SSc patients.

Table 1 Baseline clinical and demographic characteristics of SSc patients

The mean ± SD age of the 1257 patients at baseline was 51.9 ± 13.8 years, mean disease duration 7.6 years, the median follow-up was 4.2 years (95% CI: 1.96–7.09) and mean ± SD mRSS 16.9 ± 7.8. In most baseline characteristics, the three groups were comparable. After Bonferroni correction, significant differences were found between the groups regarding sex and baseline mRSS. In addition, disease duration at baseline was significantly shorter for skin progressors compared to the others, which confirms previous findings [7, 22, 23].

Associations between skin regression and visceral organ progression

Interstitial lung disease

In total, 320/981 (32.76patients fulfilled the definition of lung progression (= event) during follow up (relative decrease in FVC ≥ 10% or decrease of FVC ≥ 5–9% combined with relative decrease of DLCO ≥ 15%). There were 31/64 (48.4%) events in skin progressive patients, 231/695 (33.2%) in skin stable patients and 58/222 (26.1%) in skin regressive patients.

The probability of FVC decline was differing between the study groups of progressive, stable, and regressive patients (log-rank test p < 0.001; Fig. 1). Pairwise comparisons with Bonferroni-Holm correction were as follows: progressors vs. stable (p = 0.0021), progressors vs. regressors (p = 0.0003), regressors vs. stable (p = 0.129). When controlled for baseline mRSS, baseline immunosuppression, baseline FVC, and disease duration, Cox regression analysis indicated a significantly lower probability of later FVC decline ≥ 10% in the regressive group than in progressive patients (p = 0.00003).

Fig. 1

Probability of lung function progression depending on previous change of skin fibrosis

While the analysis above has been performed for all patients without considering presence of ILD, we repeated the analysis for patients with ILD on HRCT. Four hundred three out of 981 (40.1%) patients with data on lung progression had ILD on HRCT. In skin progressive patients, there were 12/31 (38.7%) events, while in skin stable patients, there were 86/280 (30.7%) events, and in skin regressive patients, there were 34/92 (37.0%) events. Controlling for baseline mRSS, baseline immunosuppression, baseline FVC, and disease duration, skin regressive patients tended to be less likely to face ILD progression compared to progressive patients (p = 0.085).

Cardiac progression

In the overall group 308/838 (36.8%) patients experienced a cardiac progression during the follow-up, with 19/845 (2.2%) having LVEF progression, 156/862 (18.1%) showing new conduction blocks and 189/769 (24.6%) new diastolic function abnormalities. There were 27/61 (44.3%), 227/604 (37.6%) and 54/173 (31.2%) patients with cardiac progression in the group of skin progressive, skin stable und skin regressive patients, respectively. No significant difference in the probability of cardiac progression between skin regressive patients and either stable or progressive patients was found (Supplementary Figure S1).

Intestinal progression

During the follow-up, overall 570/996 (57.2%) patients had intestinal progression, 46/76 (60.5%) of the skin progressors, 415/712 (58.3%) of the skin stable patients and 109/208 (52.4%) of the skin regressors, without significant differences between the groups (p = 0.1) (Fig. 2).

Fig. 2

Probability of intestinal progression depending on previous change of skin fibrosis

Using a more stringent definition of intestinal progression with new onset of malabsorption, weight loss > 10%, and paralytic ileus, 213/785 (27.1%) in the overall group, 20/57 (35.1%) in the progressive group, 148/549 (27.0%) in the stable group and 45/179 (25.1%) in the regressive group had intestinal progression. Although the Kaplan-Meier curve pointed to a slightly more favourable course of disease for regressive patients compared to progressive patients, the log-rank test was not significant (p = 0.3) (Fig. 3). Controlling for baseline mRSS, baseline immunosuppression, and disease duration, did not change this result (p = 0.216 for regressive vs. progressive patients).

Fig. 3

Probability of severe intestinal progression depending on previous change of skin fibrosis

New onset of scleroderma renal crisis (SRC)

Even though a large cohort was recruited, only a few patients developed new onset SRC during the follow-up. In total 19/1215 (1.6%) had a new onset of SRC and 1/89 (1.1%), 14/856 (1.6%) and 4/270 (1.5%) in the progressive, stable and in the regressive skin group, respectively. Between these groups, there was no significant difference in the probability of a new SRC (Supplementary Figure S2).

All-cause death

During follow-up, 98/1257 (7.8%) of all patients died, with 13/92 (14.1%) progressive, 69/883 (7.8%) stable and 16/282 (5.7%) skin regressive patients. The log-rank test was not significant with p = 0.06 (Fig. 4). When adjusted for baseline mRSS, baseline immunosuppression, and disease duration, the probability of all-cause death during follow up was significantly lower in patients with regression of skin fibrosis compared to skin progressive patients (p = 0.035).

Fig. 4

Probability of all-cause death depending on previous change of skin fibrosis

Discussion

Our study analysed the associations between regression of skin fibrosis and the following occurrence of progressive visceral organ diseases in patients with dcSSc. We were able to show that skin regressive patients have a lower probability of lung progression, better survival. Thus, regression of patients over a 12 months period can be used as surrogate to predict long term outcome. This allows individual stratification of patients into lower risk groups and may support decision making to initiate or continue with therapies. It also helps to identify inclusion and exclusion criteria for clinical trials, in which enrichment for patients with progressive organ involvement during the trial is often required. For example, patients with regression of skin fibrosis in the year before inclusion in a clinical trial targeting SSc-ILD should be avoided based on our data, as a lower number of progressive ILD is expected during the study.

This goes along with a study from the Pittsburgh database with 278 subjects, which concluded that improvement in skin thickening occurs in up to two-thirds of patients who are surviving the first few years of diffuse scleroderma and is associated with improved survival [24].

In addition, our study could confirm the findings of Wu et al., where patients with progressive skin involvement had a significantly higher probability of FVC decline and all-cause death. Furthermore, we observed that patients with progressive skin involvement and ILD at baseline had a higher probability of experiencing FVC decline, compared to non-progressive patients.

The associations of progressors with a worse outcome were seen versus all other patients (non-progressors), whereas the association of regressive patients with better survival in our present study could only be seen versus progressive patients. Together, these data indicate that the disease course of skin fibrosis and ILD is linked together in SSc, and that disease progression or regression in skin fibrosis can be used to for risk assessment of patients [10].

It is noteworthy that there was a time lag between the occurrence of skin progression or regression and ILD progression. In both studies, skin progression or regression was associated with later, but not parallel ILD progression/lack of ILD progression, and the association was continuous over time with stronger effects over longer follow up.

Our study has strengths and limitations. Data for this study were derived from a very large, multicentre, real-life cohort reflecting clinical practice in SSc expert centres. The large number of patients and events allows detecting differences, which are not possible to be analysed in smaller cohorts. Follow up of the patients in the study was long and reached 7 years in a reasonable number of patients. Definitions of organ progression were defined by expert consensus and have been used successfully in other studies. Limitations include missing values and loss to follow-up as typically seen in registry studies. We lack data on co-existing lung diseases, which could possibly influence our lung outcomes. Moreover, since it was an observational study, we could not assess the impact of possible treatments on outcomes. The SSc treatment is often individualised and organ specific, particularly when immunosuppressive therapy is used. As a result, effectively excluding the impact of treatment in an unselected heterogeneous cohort is challenging in observational studies with a significant treatment-by-indication error. However, in our population, there were similar proportions of patients undergoing immunosuppressive treatments in each group. Finally, it was not possible to identify the exact cause of death for all affected patients. As a result, we could only evaluate all-cause mortality, without the association to SSc. Nevertheless, all-cause mortality is seen as a more robust indicator of disease outcome than only SSc-related mortality.

Conclusion

Our study demonstrated that regression of skin fibrosis is associated with a lower probability of lung progression and better survival at follow up. These observations support the link between the disease course of skin and lung fibrosis in SSc and can help to risk stratify patients. It might be helpful to support decision making to initiate or continue with therapies and might be informative for clinical trial design in the enrichment of ILD progressive patients.