Sine scleroderma, limited cutaneous, and diffused cutaneous systemic sclerosis survival and predictors of mortality

Background Systemic sclerosis (SSc) is associated with a variability of mortality rates in the literature. Objective To determine the mortality and its predictors in a long-term follow-up of a bi-centric cohort of SSc patients. Methods A retrospective observational study by systematically analyzing the medical records of patients diagnosed with SSc in Toulouse University Hospital and Ducuing Hospital. Standardized Mortality Ratio (SMR), mortality at 1, 3, 5, 10, and 15 years of disease and causes of death were described. Predictors of mortality using Cox regression were assessed. Results Three hundred seventy-five patients were included: 63 with diffuse cutaneous SSc, 279 with limited cutaneous SSc, and 33 with sine scleroderma. The SMR ratio was 1.88 (95% CI 1.46–1.97). The overall survival rates were 97.6% at 1 year, 93.4% at 3 years, 87.1% at 5 years, 77.9% at 10 years, and 61.3% at 15 years. Sixty-nine deaths were recorded. 46.4% were SSc related deaths secondary to interstitial lung disease (ILD) (34.4%), pulmonary hypertension (31.2%), and digestive tract involvement (18.8%). 53.6% were non-related to SSc: cardiovascular disorders (37.8%) and various infections (35.1%) largely distanced those from cancer (13.5%). Four significant independent predictive factors were identified: carbon monoxide diffusing capacity (DLCO) < 70% (HR=3.01; p=0.0053), C-reactive protein (CRP) >5 mg/l (HR=2.13; p=0.0174), cardiac involvement (HR=2.86; p=0.0012), and the fact of being male (HR=3.25; p=0.0004). Conclusion Long-term data confirmed high mortality of SSc. Male sex, DLCO <70%, cardiac involvement, and CRP> 5mg/l were identified as independent predictors of mortality. Supplementary Information The online version contains supplementary material available at 10.1186/s13075-021-02672-y.

Assessment of prognosis is crucial to identify the patient who may benefit from close monitoring and immunosuppressants or autologous hematopoietic stem cells transplantation [20]. Given the wide variability of mortality rate reported in the literature, it appeared essential to obtain a better understanding of SSc prognosis and its associated risk factors in a well-characterized incident SSc cohort. The objective of the present study was to estimate mortality in the Systemic Scleroderma Toulouse Cohort (SSTC) in order to determine risk factors and causes of death.

Data source
Data were obtained from the SSTC which includes incident patients with SSc who fulfilled the 2013 ACR/ EULAR criteria and 2001 LeRoy and Medsger classification [28,29], with retrospective collection of data between January 1, 1978, andMay 30, 2018, and prospective onwards. The SSTC is a bi-centric cohort from the Toulouse University Hospital, a tertiary referral center for SSc, and Joseph Ducuing Hospital, a secondary referral center. Patients were selected through the coding system PMSI (Programme de Médicalisation des Systèmes d'Informations) and through the Doctors' active patient files. Concerning the coding system PMSI, International Classification of Diseases, Tenth Revision (ICD-10) was used (M34, M34.1, M34.8, M34.8+J99.1, M34.8+ G73.7, M34.9), as we described in a previous study [30].
A thorough medical chart review followed by the entry of the standardized data collection form, for all consecutive unselected incident SSc patients were performed. SSc sub-types were classified as "diffuse SSc, " "limited SSc, " and "SSc sine scleroderma" [31]. Localized scleroderma patients (morphea and linear disease) were not included in the SSTC. The standardized data collection covered demographic aspects, disease duration, organ involvement, laboratory data, and drug exposure. Patients with more 25% of missing data were excluded. Disease onset was defined as the date of the first non-Raynaud's phenomenon symptom attributable to SSc. Annual follow-up examinations were carried out. The forms were filled out by SSc specialists. Patients gave informed consent to participate in the SSTC. The data at baseline and at each follow-up evaluations (each annual follow-up and/or acute complication) were collected in the SSTC as part of routine clinical care in accordance to Good Clinical Practice and complied with the requirements of the Commission Nationale Informatique et Liberté (CNIL) (registration No. 914607). In compliance with French regulation relating to clinical non-interventional research, this study does not require ethics committee approval.

Population
For the present study, we included adult (≥18 years) incident SSc subjects who had at least one follow-up visit during the first years in the SSTC and a disease onset between January 1, 2000, and January 1, 2018. Patients were followed until May 31, 2018.

Collected data
Data collected at the inclusion visit were sex, ethnic group, age at disease onset, date of the first Raynaud's phenomenon symptom, date of the first non-Raynaud's phenomenon symptom, body mass index (BMI), smoking habits, SSc sub-types according to Leroy and Medsger [31], presence of arthralgia, myalgia, calcinosis or tendon friction rubs [32], gastrointestinal (GI) complications [33], neurological involvement [34], skin involvement as measured by the Rodnan modified skin score (mRSS) [35], and cardiac [36] and pulmonary evaluation [37], including pulmonary function tests (Forced Vital Capacity (FVC), diffusing capacity of the lung for carbon monoxide (DLCO)) according to the American Thoracic Society and European Respiratory Society (ATS/ERS) consensus standards [38], presence of SSc-related interstitial lung disease (ILD) or pleural effusion on chest X-ray or on high resolution computed tomography (HRCT), results from transthoracic echocardiography (TTE) including left ventricular ejection fraction (LVEF), tricuspid regurgitation velocity (TRV, m/s), and pulmonary arterial systolic pressure (PASP, mmHg) measurement [39]. Pulmonary arterial hypertension (PAH) was confirmed as a mean PAP ≥25 mmHg and was considered a SSc-associated pre-capillary PAH when associated with a pulmonary artery wedge pressure (PAWP) ≤ 15 mmHg [40]. Cardiac involvement was defined by LVEF <50%, and/or TTE abnormalities (pericarditis, cardiac valvulopathy, or diastolic dysfunction), and/or an electrocardiogram (ECG) abnormality (arrhythmia or conduction blocks). Atrial and ventricular arrhythmia were included in arrythmia. Any cardiac symptoms appearing prior to SSc diagnosis were considered as having no connection with the disease under study. The scleroderma renal crisis (SRC) was defined as a new onset of significant systemic hypertension (>150/85 mmHg) and acute renal failure (≥30% reduction in estimated glomerular filtration rate) [2,[41][42][43][44]. Laboratory parameters collected were hemoglobin level, serum creatinine, C-reactive protein, albumin, antinuclear (ANA), anticentromere (ACA), anti-SCL70, anti RNA polymerase III, anti PM/Scl, anti-TIF1-y, and ANCA [45,46]. If the clinical information was missing, this information was counted as missing data.

Mortality data
Survival status was ascertained up until the end of this study based on the records in the database and systematically verified in each medical chart, telephone tracing of patient's general practitioner (GP), telephone tracing of patients in whom no data had been entered for ≥24 months in the database, and systematic interrogation of either the Registre d'état civil [47]. The final status of loss to follow-up was defined as one where no data had been entered for ≥ 24 months with a failure to contact the patient or his GP despite at least two attempts, no death recorded in either the Registre d'état civil [47] and in French database recording deaths "https:// deces. match id. io/ search. "

Calculation of standardized mortality ratio
The SMR and its 95% confidence interval (95% CI) were calculated according to the ratio of observed death in the cohort to the number of deaths of the French age/ sex-matched population [48]. The mortality rates of the general population were obtained from the French National Statistical Agency (Institut national de la statistique et des études économiques -INSEE), and the most recent available data at the time of data analysis were from December 2016 [49]. In relation to subjects lost to follow-up, we performed 2 sensitivity analyses to recalculate SMR, one of which assumed that all such subjects were alive at the end of the study and the other of which assumed that all such subjects were dead at the end of the study.

Causes of death
A standardized death case report form was extracted from the SSTC database. Cause of death was then systematically verified against source documents. The causes of death were categorized as a single primary cause (either SSc or non-SSc-related) and all other SSc organ involvements that contributed to death. Death was attributed to SSc if the cause was identified with the specific organ involved. Death not attributed to SSc in the following cases: • When the specific organ involved was cited in a diagnosis prior to SSc diagnosis, except typical scleroderma involvement • Sudden inexplicable death • Death from events with no direct link to SSc.

Statistical analysis
Data are presented as the mean ± SD for continuous variables, the median and interquartile range for non-normally distributed continuous variables, and the number and percent for categorical variables. Baseline characteristics were compared using ANOVA variance analysis for continuous variables, and a chi-square test or Fisher's exact test for categorical variables depending on the sample size.
Survival analysis was performed using the Kaplan-Meier method with comparisons performed using the log rank test. The primary end point was death from any cause or data censoring. The follow-up period ended in May 30, 2018. Univariable and multivariable Cox proportional hazards models (ascending step-by-step method) were used to determine baseline variables associated with mortality. Variables with p value ≤0.05 in univariate analysis were selected for multivariate analysis. Analysis of the co-linearities variables were used to multivariable Cox model: PAPs >35 mmHg and Raynaud's syndrome appearing after age 45 were not taken into account in the multi-variate analysis given the co-linearities respectively with the DLCO< 70% of the theoretical value and disease onset after the age of 50 years old.
Two-tailed P values less than or equal to 0.05 were considered significant. All statistical analyses were performed using SAS ® software (French version 9.4).

SMR and survival analysis
During the study period, 69 patients died (18.4%) and 6 patients (only women, one sine scleroderma sub-type and 5 diffuse SSc) were lost to follow-up. The mean ± SD age at the time of death was 69.1 ± 14.8 years for SSc patients, 72 ± 15.5 and 64.2 ± 12.3 years, for women and men, respectively. The age-and sex-adjusted SMR of the cohort was 1.88 (95% CI 1.46-1.97) assuming that all subjects lost to follow-up were alive or 2.04 (95% CI 1.60-2.13) assuming that they were dead. Age-adjusted SMR for men was 3.61 (95% CI 2.35-3.94). Age-adjusted SMR for women was 1.80 (95% CI 1.31-1.92) assuming that all women lost to follow-up were alive or 2.05 (95% CI 1.52-2.18) assuming that they were dead. Age-and sex-adjusted SMR for diffuse subtype was 3.31 (95% CI 1.88-3.76). Age-and sex-adjusted SMR for limited cutaneous and sine scleroderma subtype was 1.74 (95% CI 1.30-1.85) and 1.03 (95% CI 0.20-1.49), respectively, assuming that all patients lost to follow-up were alive or 1.91 (95% CI 1.44-1.98) and 1.37 (95% CI 0.36-1.86) assuming that they were dead.

Discussion
This study still clearly confirmed an increased mortality risk from SSc, compared with the general healthy population. This is one of the only to describe the mortality rate of SSc sine scleroderma. Male sex, cardiac involvement, systemic inflammation, and altered DLCO were independent mortality risk factors. The majority of deaths (53.6%) was not attributed to SSc directly, especially cardiovascular mortality (37.8%).
Our SMR is lower than those produced in the two most recent meta-analyses focusing on mortality in SSc. Rubio-Rivas et al. [8], analyzing 17 articles from 1964 to 2005, found an SMR of 2.72 (95% CI 1.93-3.83). More recently, Pokeerbux et al. [11] reported an SMR of 5.73 (95% CI 4.68-6.94) from a French multi-centered cohort of patients suffering from SSc. The latter also carried out a meta-analysis of 18 articles with a pooled SMR of 3.45 (95% CI 3.03-3.94). However, the literature reveals a wide variability of mortality rate in SSc, with SMRs ranging from 1.34 to 7.18 [8,10,11,17]. Such variability is probably caused by considerable methodological differences in the various studies carried out, differences in time origin from which survival time is calculated and this disease high heterogeneity. The elevated SMR figure in the and a less extended level of cutaneous involvement (average Rodnan skin score of 6.8 for the patients in the present study vs. 9) [11]. Regarding SRC, we described the same incidence as in the recent German epidemiological study evaluating renal involvement on 2873 patients with SSc [50]. In addition, we kept in our analysis SSc sine scleroderma, which is often excluded from such studies that could explain our relatively low SMR. Indeed, sine   The results are expressed as hazard ratios (HR) with a 95% confidence interval (95% CI), cardiac involvement comprises left ventricle ejection fraction <50%, and/or a TTE anomaly (pericarditis, valvular disease, or diastolic dysfunction), and/or an ECG anomaly (arrhythmia or conduction blocks). PAPs >35 mm Hg and Raynaud's syndrome appearing after age 45 were not taken into account in the multi-variate analysis given the co-linearities respectively with the DLCO< 70% of the theoretical value and disease onset after the age of 50 years old  scleroderma subtype tended to have the best survival in our study with more 80% of survival rate at 15 years; the same result was found by Siméon-Aznar et al. [51]. The SMR of our cohort more approximated "mortality with real life, " taking into account the different subtypes of SSc, in particular sine scleroderma subtype. Our overall survival rates at 1 year, 3 years, 5 years, and 10 years were consistent with those reported previously [8,9,11,23,26]. In addition, we have shown a non-significant higher mortality rate from the diffuse cutaneous sub-type vs. the limited cutaneous subtype and the sine scleroderma subtype of SSc, in line with the published literature [8,11,18,24,25,[52][53][54]. However, the absence of significance may be related to an underpowered sample size for this analysis.
Our study identified 4 independent predictors of mortality: male sex, cardiac involvement, DLCO <70% of predicted, and CRP >5 mg/ml, which were consistent with previous studies [11,23,52,[55][56][57][58][59][60][61]. Indeed, a low DLCO in SSc often associated with ILD and/or PH (essentially pulmonary arterial hypertension and secondary pulmonary hypertension resulting from ILD), the two main causes of SSc-related mortality. Male SSc patients often presented with a worst prognostic than female SSc patients [11,60,62,63], like in our cohort. Cardiac involvement during the course of SSc is also associated with an increased risk of mortality, especially pericardial effusion [11,57,59,64], valvular disease [11], and an LVEF <50% [23]. In our study, cardiac involvement was a composite outcome including LVEF <50%, and/or a TTE anomaly (pericarditis, valvulopathy, or diastolic dysfunction), and/or an ECG anomaly (arrhythmia or conduction blocks), which was strongly associated with a 2.9-fold increase mortality risk (HR=2.86 95% CI 1. 54-5.41). Finally, we confirmed that systemic inflammation was a major pejorative prognostic factor of mortality (HR 2.13 95%CI 1. 11-5.41) like reported by few studies [11,23,53]. Diffuse subtype does not appear as a prognosis factor; this result can be explained by an underpowered sample size of diffuse and anti-SCL70 subtypes for this analysis. In several studies, diffuse subtype is a strong prognosis factor. Thus, in EUSTAR cohort, diffuse subtype does not appear to be a major factor in mortality (HR 0.79 (0.67 to 0.93)) [12]. In the SCleroderma mOrtality p Eustar (SCOpE), a score predicting mortality from EUstar cohort, diffuse subtype was one of the least weighted criteria after DLCO<60%, age, dyspnea, LVEF <50%, FVC<70%, and renal involment [12]. As already discussed, the incidence of this complication has decreased, and in our cohort, we have a lower rate of this involvement that others studies. It was a possible explanation that diffuse subtype is not a risk factor for mortality in our study.
One of the most interesting results of our study was that the majority of identified death in the SSTC cohort was not related to SSc directly. Due to a better understanding of SSc and a better management of this severe systemic disease, SSc-related deaths became less frequent [3,18]. The first cause of non-SSc related in our work was cardiovascular diseases in a higher proportion than in previous large cohort studies (2)(3)(4)(5)(6)(7)(8)(9)(10)(11).9%) [8,9,18,62]. The cardiovascular-related deaths are represented mainly by sudden cardiac arrest. The majority of cardiac arrest cases, occurring at the patient's home, were diagnosed by general practitioner and thus could be subject to a bias. However, 46.4% of the deaths in our cohort were directly attributable to SSc. Our results were consistent with the data from the EUSTAR cohort, finding that nowadays, the two main causes of SScrelated were ILD and PH [3,6,18,23,25,52,53,62]. SSc GI-related mortality was frequent in our cohort (18.2%), and this cause of death represented 7.6% of deaths in metanalysis of Rubi-Rivas et al. [8]. Due to our strict methodology with a thorough medical chart review and systematic telephone tracing of patient's general practitioner to ascertain cause of death, we were perhaps more able to identify this complication and its prognosis.
The major strengths of the present study are that we carried out an exhaustive medical chart review for each included patient, which led to detailed clinical and laboratory characteristics in a large cohort of incident patients. We ascertained vital status and cause of death by systematically interrogated the SSTC database, reviewed each medical chart a second time, called each patient's GP, and interrogated the Registre d'état civil [47]. Other major strengths of this study are the longterm follow-up of patients, a very low attrition rate, as only six patients were lost to follow-up covering a period of 18 years and that we kept for the analysis SSc sine scleroderma which is often excluded from clinical trial and cohort studies. Thus, this study provides the sine scleroderma subtype survival. Probably, our overall SMR is more pertinent, not excluding the subtypes.
Οur study has some limitations. Our study is necessarily limited by its retrospective and bi-centric nature. Because it is a retrospective study, misclassification of systemic sclerosis subtypes is possible and induces a classification bias. Additionally, we did not include information about the treatments used during the follow-up due to heterogeneity in time of treatment initiation and the duration or type of treatment administered and as a consequence the effect of the various treatments in the natural course of the disease could not be eradicated. Concerning causes of deaths, potential interactions between SSc-related and unrelated to SSc

Conclusion
In conclusion, our results show that mortality is still substantial in SSc despite constant therapeutic progress. This is one the first study that ascertained SSc sine scleroderma mortality rate alongside the two other sub type of SSc. This serves as an important reference for future survival analysis and epidemiology surveys. Our study identifies strong predictors of mortality male sex, cardiac involvement, DLCO <70%, and CRP > 5 mg/l. Non-SScrelated death is more frequent than SSc-related deaths, of whom cardiovascular disease is the most common. An early and systematic management of the large proportion of cardiac complications is in order, in hope of extending SSc outcome.