https://orcid.org/0000-0001-8655-3244
Abstract
Myocardial involvement is common in patients with systemic sclerosis (SSc) and causes myocardial fibrosis and subtle ventricular dysfunction. However, the temporal onset of myocardial involvement during the progression of the disease and its prognostic value are yet unknown. We used cardiovascular magnetic resonance (CMR) to investigate subclinical functional impairment and diffuse myocardial fibrosis in patients with very early diagnosis of SSc (VEDOSS) and established SSc and examined whether this was associated with mortality.
One hundred and ten SSc patients (86 established SSc, 24 VEDOSS) and 15 healthy controls were prospectively recruited. The patients were followed-up for a median duration of 7.0 years (interquartile range 6.0–7.3 years). Study subjects underwent CMR including assessment of myocardial fibrosis [native T1 and extracellular volume (ECV)] and measurement of global longitudinal (GLS) and circumferential (GCS) myocardial strain. Native T1 values and ECV were elevated in VEDOSS and SSc patients compared with controls (P < 0.001). GLS was similar in VEDOSS and controls but significantly impaired in patients with established SSc (P < 0.001). GCS was similar over all groups (P = 0.88). There were 12 deaths during follow-up. Elevated native T1 [hazard ratio (HR) 5.8, 95% confidence interval (CI): 1.7–20.4; P = 0.006] and reduced GLS (HR 6.1, 95% CI: 1.3–29.9; P = 0.038) identified subjects with increased risk of death. Only native T1 was predictive for cardiovascular mortality (P < 0.001).
Subclinical myocardial involvement first manifests as diffuse myocardial fibrosis identified by the expansion of ECV and increased native T1 in VEDOSS patients while subtle functional impairment only occurs in established SSc. Native T1 and GLS have prognostic value for all-cause mortality in SSc patients.
Diffuse myocardial fibrosis as identified by elevated native T1 and ECV is already present in patients with very early disgnosis of SSc to a simlar extend as in patients with established SSc (*P < 0.05). Native T1 also exhibits a strong prognostic value for mortality in SSc patients.
Introduction
Systemic sclerosis (SSc) is a chronic connective tissue disorder characterized by abnormal immune responses, microvascular damage, and multi-organ fibrosis.1 Life expectancy of SSc patients has increased over the last decades.2,3 An expanding therapeutic armamentarium ranging from tailored immunosuppressive therapy to haematopoietic stem cell transplantation notwithstanding,4 there remains a significant disease-specific mortality which can be largely attributed to pulmonary and cardiac involvement.5,6 Recent advances in the treatment of pulmonary arterial hypertension7 and interstitial lung disease8–11 leave cardiac involvement as one of the most important, unaddressed causes of death in SSc patients.3,6
Advanced cardiac involvement of SSc manifests as focal myocardial fibrosis that can be visualized by cardiac magnetic resonance (CMR) on late gadolinium enhancement (LGE) images.12 Even in 26% of SSc patients without clinical suspicion of cardiac involvement, myocardial fibrosis identified by LGE was found and correlated with ventricular arrhythmias.13 Given that the event-free 5-year survival of SSc patients with LGE is <50%,14 the assessment of earlier stages of cardiac involvement is of paramount importance to guide timely decisions on treatment interventions. Novel methods such as native T1 mapping or the quantification of extracellular volume (ECV) have been shown to identify diffuse myocardial fibrosis in SSc patients without manifest LGE.15, 16 In addition, subclinical impairment of ventricular function, as defined by reduced global longitudinal strain (GLS), has been reported in patients with SSc.17 However, the temporal onset of diffuse myocardial fibrosis and subtle functional impairment during the progression of the disease as well as its prognostic value still remains elusive.
Patients with very early diagnosis of SSc (VEDOSS) do not fulfill the 2013 ACR/EULAR SSc classification criteria but exhibit Raynaud’s phenomenon (RP) and other features of SSc, such as puffy fingers, disease-specific autoantibodies, and capillaroscopy changes typical for SSc.18,19 Since up to 79% of the VEDOSS patients progress to definite SSc after up to 9 years of observation,20 they can be considered to be in the initial stage of SSc evolution. While gastrointestinal involvement can already be present in VEDOSS patients,21 the presence and extent of cardiac involvement have not yet been investigated.
Therefore, we conducted this study to investigate the presence of subtle fibrotic and functional myocardial involvement in VEDOSS patients in comparison to patients with established SSc and evaluated its prognostic value.
Methods
Study design
Patients with a diagnosis of systemic sclerosis and age- and sex-matched healthy controls were recruited in the Department of Rheumatology at the University Hospital Zurich from October 2013 through December 2014. Patients were enrolled consecutively if they had systemic sclerosis according to the 2013 ACR/EULAR classification1 or met the VEDOSS criteria19 and had no standard exclusion criteria for CMR.22 VEDOSS patients did not fulfill the 2013 classification criteria for SSc, but showed RP and additional features of SSc (puffy fingers, SSc-specific antibodies, SSc pattern on nailfold capillaroscopy, or any characteristic SSc organ involvement). Patients with primary RP (RP without any of the manifestations above) were excluded. Prerequisite for the administration of gadolinium-based contrast agent (GBCA) was a kidney function with eGFR > 30 mL/min. All selected patients underwent a dedicated CMR examination as part of a standardized annual assessment programme for SSc patients, which followed the EUSTAR recommendations23 and included clinical examination, lung function tests, high-resolution computed tomography (HRCT), and laboratory analysis. Before imaging, written informed consent was obtained from all subjects. The follow-up consisted of a yearly clinical visit and an additional contact with the patient if clinically required. The clinical endpoints were all-cause mortality and cardiovascular mortality. The study protocol was approved as part of the local SSc registry by the ethics committee of the canton of Zurich (EUSTAR BASEC-Nr. PB_2016-01515; VEDOSS BASEC-Nr. PB_2020-00003, BASEC-Nr. 2018-02165).
MRI data acquisition
CMR imaging was performed on a clinical 1.5 T Philips Achieva System (Philips Healthcare, Best, The Netherlands) equipped with a five-channel cardiac receiver array and cardiac synchronization was performed using a vector ECG. Cardiac function was assessed by a contiguous stack of breath-hold balanced steady-state free precession (bSSFP) short-axis cine images covering the entire left ventricle (LV) and bSSFP long-axis cine images in two-chamber, three-chamber, and four-chamber orientation [repetition time/echo time = 3.3/1.6 ms, flip angle = 60°, spatial resolution = 1.5 × 1.5 mm2, slice thickness = 8 mm, min. 25 cardiac phases). In addition, T1- and T2-weighted turbo-spin-echo (TSE) sequences in short-axis orientation were acquired as part of the clinical protocol.
For the determination of myocardial T1 and ECV, a modified Look-Locker inversion (MOLLI) sequence was acquired with a total of eight T1-weighted images split into a set of five and three images acquired after inversion, separated by three heartbeats of recovery [5-(3)-3 scheme].24 Post-contrast T1 mapping was performed 15–20 min after the administration of a bolus injection of 0.2 mmol/kg GBCA (Gadovist, Bayer Schering, Germany). Post-contrast T1 mapping was added to the study protocol in January 2014, so that ECV data for 81 SSc patients are available. The haematocrit was measured in all patients on the day of CMR. LGE images were acquired with an inversion prepared spoiled gradient echo sequence ∼10 min after the bolus injection. The inversion recovery prepulse delay was determined using a Look-Locker sequence.
MRI data analysis
Standard measurements of LV size and function including left ventricular ejection fraction, left ventricular indexed mass, and intraventricular septal wall thickness were obtained using the GTVolume software package (GyroTools LLC, Zurich, Switzerland).
Strain analysis was performed using a certified CMR feature tracking evaluation software (2D CPA MR, Cardiac Performance Analysis MR Version 4, TomTec Imaging Systems, Unterschleissheim, Germany). For the evaluation of LV end-systolic global longitudinal strain (GLS) the three long-axis bSSFP slices and for global circumferential strain (GCS) three short-axis (basal, mid-ventricular, and apical level) bSSFP slices were used. Strain calculations were based on software-driven automatic tracking of the endocardial contours, however, the tracing of each end-diastolic and end-systolic phase was individually checked and adapted, if necessary.
Before calculating the pre- and post-contrast T1 maps, the individual T1 MOLLI images were registered for compensating residual motion by using a non-rigid groupwise image registration method.25 In agreement with current clinical recommendations, native T1 and ECV were evaluated in a septal ROI on the mid-ventricular level.26 ECV was calculated based on the collected haematocrit and native and enhanced T1 values in the myocardium and the blood pool.
Statistical analysis
Continuous variables were expressed as mean ± SD or median and quartiles as appropriate. Categorical variables were expressed as frequency and percentage. Statistical analysis was conducted using MedCalc software (MedCalc 17.9.7, MedCalc Software bvba, Ostend, Belgium). The normal distribution of data was confirmed using the Shapiro–Wilk test. Differences between groups were tested using an unpaired two-tailed Student t-test for normally distributed data and a Mann–Whitney U test for not normally distributed data. If more than two groups were compared, a one-way analysis of variance or Kruskal–Wallis test was applied with Tukey–Kramer post hoc comparison. As appropriate, Pearson or Spearman correlation analyses were used to compare CMR parameters to mRSS and disease duration. The optimal threshold to discriminate patients who died during follow-up was identified by receiver operating characteristic curve (ROC) analysis. Cumulative incidence rates of outcome were estimated using the Kaplan–Meier method and compared with the log-rank test. Patients who were lost to follow-up were censored at the time of the last contact. Cox proportional hazards method was used to identify the predictors of all-cause mortality among patients. The multivariable model was built by stepwise variable selection with entry and exit criteria set at P ≤ 0.2. A P-value of <0.05 was considered statistically significant.
Results
Patient characteristics
Data acquisition and evaluation were successful in 110 of 130 enrolled patients and in all 15 healthy controls. Fifteen patients were excluded due to insufficient image quality for quantitative T1 or strain evaluation, three due to reclassification as undifferentiated connective tissue disease, and two patients withdraw consent. Ninety-four female (85%) and 16 male patients with a mean age of 55 ± 14 years were finally included in the analysis. Of those, 24 were VEDOSS patients and 86 fulfilled the 2013 ACR/EULAR classification criteria for established SSc. Of the patients with established SSc, 61 (71%) also fulfilled the 1980 ACR criteria for SSc indicating patients with more advanced disease.27 Approximately one-fifth of the patients had diffuse cutaneous SSc (21%), while the majority had limited cutaneous SSc (79%). The median disease duration, i.e. time since the onset of the first non-Raynaud’s symptom, was 6.3 years.
There were no significant differences in LV volume, LV-ejection fraction, and LV mass between controls, VEDOSS, and established SSc patients. However, the right ventricular end-diastolic volume was slightly reduced in VEDOSS and SSc patients compared with controls. Two of the VEDOSS patients (8.3%) and nine patients with established SSc (10.5%) showed myocardial LGE. Of those, two had ischaemic, subendocardial LGE while the remaining nine patients had subepicardial or diffuse midmyocardial LGE, mostly located in the basal and mid-ventricular segments (see Supplementary data online, Figure S1). The other baseline characteristics were similar between the three groups (Table 1). An extended description of the systemic sclerosis-specific organ manifestations can be found in Supplementary data online, Table S1.
Baseline characteristics of the study population
| Characteristic | Controls (N = 15) | VEDOSS (N = 24) | Established SSc (N = 86) | P-value |
|---|---|---|---|---|
| Female (n) (%) | 12 (80) | 22(92) | 72 (84) | P = 0.54 |
| Age (years) | 51 ± 16 | 49 ± 13 | 56 ± 13 | P = 0.07 |
| BSA (m2) | 1.70 ± 0.16 | 1.69 ± 0.18 | 1.75 ± 0.22 | P = 0.39 |
| HR (bpm) | 70 ± 12 | 69 ± 13 | 71 ± 13 | P = 0.83 |
| Hct (%) | 40.8 ± 4.0 | 39.9 ± 3.2 | 39.0 ± 4.0 | P = 0.56 |
| Diffuse cutaneous SSc (n) (%) | 0 | 23 (27) | ||
| Modified Rodnan skin score (median) (range) | 0 | 3, (0–29) | ||
| Years since first SSc symptoma (median) (range) | 4.3, (1.0–33.8) | 6.3, (0.1–49.7) | P = 0.41 | |
| NT-proBNPb (median) (min./max. range) | 96 (8–267) | 94, (12–1888) | P = 0.53 | |
| LVEF (%) | 61.2 ± 4.7 | 63.6 ± 5.7 | 64.1 ± 7.2 | P = 0.29 |
| LV EDV (mL) | 127 ± 32 | 112 ± 25 | 121 ± 36 | P = 0.49 |
| LV Massind (g/m2) | 48 ± 12 | 41 ± 8 | 43 ± 16 | P = 0.15 |
| RVEF (%) | 56.7 ± 6.0 | 60.7 ± 6.4 | 61.3 ± 6.9 | P = 0.052 |
| RV EDV (mL) | 145 ± 43 | 109 ± 25c | 115 ± 36c | P = 0.009 |
| LGE Fibrosis (n) (%) | 0 (0) | 2 (8.3) | 9 (10.5) |
| Characteristic | Controls (N = 15) | VEDOSS (N = 24) | Established SSc (N = 86) | P-value |
|---|---|---|---|---|
| Female (n) (%) | 12 (80) | 22(92) | 72 (84) | P = 0.54 |
| Age (years) | 51 ± 16 | 49 ± 13 | 56 ± 13 | P = 0.07 |
| BSA (m2) | 1.70 ± 0.16 | 1.69 ± 0.18 | 1.75 ± 0.22 | P = 0.39 |
| HR (bpm) | 70 ± 12 | 69 ± 13 | 71 ± 13 | P = 0.83 |
| Hct (%) | 40.8 ± 4.0 | 39.9 ± 3.2 | 39.0 ± 4.0 | P = 0.56 |
| Diffuse cutaneous SSc (n) (%) | 0 | 23 (27) | ||
| Modified Rodnan skin score (median) (range) | 0 | 3, (0–29) | ||
| Years since first SSc symptoma (median) (range) | 4.3, (1.0–33.8) | 6.3, (0.1–49.7) | P = 0.41 | |
| NT-proBNPb (median) (min./max. range) | 96 (8–267) | 94, (12–1888) | P = 0.53 | |
| LVEF (%) | 61.2 ± 4.7 | 63.6 ± 5.7 | 64.1 ± 7.2 | P = 0.29 |
| LV EDV (mL) | 127 ± 32 | 112 ± 25 | 121 ± 36 | P = 0.49 |
| LV Massind (g/m2) | 48 ± 12 | 41 ± 8 | 43 ± 16 | P = 0.15 |
| RVEF (%) | 56.7 ± 6.0 | 60.7 ± 6.4 | 61.3 ± 6.9 | P = 0.052 |
| RV EDV (mL) | 145 ± 43 | 109 ± 25c | 115 ± 36c | P = 0.009 |
| LGE Fibrosis (n) (%) | 0 (0) | 2 (8.3) | 9 (10.5) |
Data are presented as mean ± SD or number of patients (%). Differences between groups were compared using a χ2 test, an one-way analysis of variance, or Kruskal–Wallis test as appropriate.
Time since first non-Raynaud’s symptom or time since first Raynaud’s symptom in all patients who had no non-Raynaud`s symptoms.
NT-proBNP values were available for 83 ACR/EULAR patients and all VEDOSS patients. BSA, body surface area; Hct, haematocrit; HR, heart rate; LGE, late gadolinium enhancement; LV EDV, left ventricular end-diastolic volume; LVEF, left ventricular ejection fraction; LV Massind, left ventricular mass indexed to BSA; NT-proBNP, N-terminal prohormone of brain natriuretic peptide; RVEF, right ventricular ejection fraction; RV EDV, right ventricular end-diastolic volume; SSc, systemic sclerosis.
P < 0.05 vs. controls.
Baseline characteristics of the study population
| Characteristic | Controls (N = 15) | VEDOSS (N = 24) | Established SSc (N = 86) | P-value |
|---|---|---|---|---|
| Female (n) (%) | 12 (80) | 22(92) | 72 (84) | P = 0.54 |
| Age (years) | 51 ± 16 | 49 ± 13 | 56 ± 13 | P = 0.07 |
| BSA (m2) | 1.70 ± 0.16 | 1.69 ± 0.18 | 1.75 ± 0.22 | P = 0.39 |
| HR (bpm) | 70 ± 12 | 69 ± 13 | 71 ± 13 | P = 0.83 |
| Hct (%) | 40.8 ± 4.0 | 39.9 ± 3.2 | 39.0 ± 4.0 | P = 0.56 |
| Diffuse cutaneous SSc (n) (%) | 0 | 23 (27) | ||
| Modified Rodnan skin score (median) (range) | 0 | 3, (0–29) | ||
| Years since first SSc symptoma (median) (range) | 4.3, (1.0–33.8) | 6.3, (0.1–49.7) | P = 0.41 | |
| NT-proBNPb (median) (min./max. range) | 96 (8–267) | 94, (12–1888) | P = 0.53 | |
| LVEF (%) | 61.2 ± 4.7 | 63.6 ± 5.7 | 64.1 ± 7.2 | P = 0.29 |
| LV EDV (mL) | 127 ± 32 | 112 ± 25 | 121 ± 36 | P = 0.49 |
| LV Massind (g/m2) | 48 ± 12 | 41 ± 8 | 43 ± 16 | P = 0.15 |
| RVEF (%) | 56.7 ± 6.0 | 60.7 ± 6.4 | 61.3 ± 6.9 | P = 0.052 |
| RV EDV (mL) | 145 ± 43 | 109 ± 25c | 115 ± 36c | P = 0.009 |
| LGE Fibrosis (n) (%) | 0 (0) | 2 (8.3) | 9 (10.5) |
| Characteristic | Controls (N = 15) | VEDOSS (N = 24) | Established SSc (N = 86) | P-value |
|---|---|---|---|---|
| Female (n) (%) | 12 (80) | 22(92) | 72 (84) | P = 0.54 |
| Age (years) | 51 ± 16 | 49 ± 13 | 56 ± 13 | P = 0.07 |
| BSA (m2) | 1.70 ± 0.16 | 1.69 ± 0.18 | 1.75 ± 0.22 | P = 0.39 |
| HR (bpm) | 70 ± 12 | 69 ± 13 | 71 ± 13 | P = 0.83 |
| Hct (%) | 40.8 ± 4.0 | 39.9 ± 3.2 | 39.0 ± 4.0 | P = 0.56 |
| Diffuse cutaneous SSc (n) (%) | 0 | 23 (27) | ||
| Modified Rodnan skin score (median) (range) | 0 | 3, (0–29) | ||
| Years since first SSc symptoma (median) (range) | 4.3, (1.0–33.8) | 6.3, (0.1–49.7) | P = 0.41 | |
| NT-proBNPb (median) (min./max. range) | 96 (8–267) | 94, (12–1888) | P = 0.53 | |
| LVEF (%) | 61.2 ± 4.7 | 63.6 ± 5.7 | 64.1 ± 7.2 | P = 0.29 |
| LV EDV (mL) | 127 ± 32 | 112 ± 25 | 121 ± 36 | P = 0.49 |
| LV Massind (g/m2) | 48 ± 12 | 41 ± 8 | 43 ± 16 | P = 0.15 |
| RVEF (%) | 56.7 ± 6.0 | 60.7 ± 6.4 | 61.3 ± 6.9 | P = 0.052 |
| RV EDV (mL) | 145 ± 43 | 109 ± 25c | 115 ± 36c | P = 0.009 |
| LGE Fibrosis (n) (%) | 0 (0) | 2 (8.3) | 9 (10.5) |
Data are presented as mean ± SD or number of patients (%). Differences between groups were compared using a χ2 test, an one-way analysis of variance, or Kruskal–Wallis test as appropriate.
Time since first non-Raynaud’s symptom or time since first Raynaud’s symptom in all patients who had no non-Raynaud`s symptoms.
NT-proBNP values were available for 83 ACR/EULAR patients and all VEDOSS patients. BSA, body surface area; Hct, haematocrit; HR, heart rate; LGE, late gadolinium enhancement; LV EDV, left ventricular end-diastolic volume; LVEF, left ventricular ejection fraction; LV Massind, left ventricular mass indexed to BSA; NT-proBNP, N-terminal prohormone of brain natriuretic peptide; RVEF, right ventricular ejection fraction; RV EDV, right ventricular end-diastolic volume; SSc, systemic sclerosis.
P < 0.05 vs. controls.
Diffuse myocardial fibrosis can be detected in VEDOSS patients
The myocardial native T1 values differed significantly between the three groups (P = 0.007). VEDOSS patients showed elevated native T1 times when compared with the healthy control group (VEDOSS 1030 ± 33 ms vs. control 994 ± 23 ms, P < 0.05). Likewise, the native T1 values in patients with established SSc were significantly increased compared with controls (ACR/EULAR 1034 ± 52 ms, P < 0.05). Notably, despite their very early disease, there were already similar changes of native T1 in the VEDOSS group as compared with the established SSc patients (Figure 1A).
Comparison of tissue characterization and myocardial strain in controls, patients with VEDOSS and patients with established SSc according to the 2013 ACR/EULAR criteria. (A) Box and whisker plots of the native myocardial T1 times that are significantly elevated in VEDOSS patients and patients with established SSc (ACR/EULAR criteria) when compared with healthy controls. Likewise, (B) exhibits a significant expansion of the ECV in VEDOSS patients and patients with established SSc compared with healthy controls. There are no significant differences in native T1 and ECV between VEDOSS and established SSc patients. ECV values were available for 63 patients with established SSc and 17 VEDOSS patients. (C) Subtle impairment of the systolic left ventricular function, assessed by the end-systolic GLS can be detected in patients with established SSc, while there is no significant difference in GLS between healthy controls and VEDOSS patients. As seen in (D), the end-systolic GCS is similar between all groups (* indicates P < 0.05).
When characterized by ECV, the myocardial fibrosis in the three groups exhibited a similar pattern as native T1 with an even higher level of significance (P < 0.001). Compared with controls, ECV was elevated in the VEDOSS group (VEDOSS 29.1 ± 4.4% vs. control 26.1 ± 2.5%, P < 0.05) and in patients with established SSc (ACR/EULAR 31.4 ± 5.4%, P < 0.05). As in native T1, there was no significant difference in ECV between VEDOSS patients and patients with established SSc (Figure 1B), indicating significant changes already in the very early disease course. The native T1 and ECV values per AHA-segment can be found in the Supplementary data online, Tables S2 and S3. Figure 2 depicts representative mid-ventricular slices of a VEDOSS patient and a patient with established SSc who do not show focal myocardial fibrosis in LGE images but exhibit elevation of native T1 and ECV, indicating diffuse myocardial fibrosis.
Representative mid-ventricular slices of a VEDOSS patient and a patient with established SSc. Both patients do not show local myocardial fibrosis in LGE images but exhibit elevation of native T1 and ECV, indicating diffuse myocardial fibrosis.
Myocardial strain is normal in VEDOSS patients
While LV-ejection fraction was similar in the three groups, there was a difference in end-systolic GLS. Patients with established SSc according to the ACR/EULAR criteria had significantly lower absolute GLS than controls and VEDOSS patients (controls −22.6 ± 1.8% vs. VEDOSS−22.5 ± 2.5% vs. ACR/EULAR −20.4 ± 2.8%, P < 0.001). There was no difference in GLS between VEDOSS patients and controls (Figure 1C).
The end-systolic GCS was similar across all groups (controls −28.5 ± 3.4% vs. VEDOSS −28.6 ± 3.0% vs. ACR/EULAR −27.8 ± 4.5%; P = 0.64, Figure 1D).
Prognostic value of subtle myocardial involvement
In the entire population, 12 of 110 patients died during the follow-up period (median 7.0 years, interquartile range 6.0–7.3 years). In the whole-study cohort the annualized event rate was 1.7% for all-cause mortality. For native T1, a threshold of 1065 ms best-discriminated patients at an elevated risk. In the groups with native T1 > 1065 ms, the annualized rate for death was 4.4% while patients with a native T1 below this threshold had a low annual event rate of 0.95%. By Kaplan–Meier survival curves, there was a significantly greater cumulative incidence of death [hazard ratio (HR) 5.8, 95% confidence interval (CI): 1.7–20.4; P = 0.006] for elevated native T1 (Figure 3A). Patients with an ECV above the optimal threshold of 30% showed a trend to higher mortality during follow-up [HR 3.9, 95% CI: 0.8–20.1; P = 0.076; Figure 3B]. GLS with a threshold of −16.5% identified a small group of patients who were at increased risk of mortality [HR 6.1, 95% CI: 1.3–29.9; P = 0.038; Figure 3C], however, this group also exhibited a significantly lower LV-ejection fraction (see Supplementary data online, Table S4). In a multivariable stepwise Cox regression, the presence of elevated native T1 (HR 4.6, 95% CI: 1.5–14.6; P = 0.01) and GLS (HR 4.2, 95% CI: 1.1–15.7; P = 0.02) were independent predictors of a higher incidence of death of any cause (Table 2). For the analysis of cardiovascular mortality, four patients had to be excluded due to death of unknown cause. In the remaining study population, cardiovascular death occurred in six patients while two patients died of non-cardiovascular cause (see Supplementary data online, Table S5). Native T1 > 1065 ms exhibited a similar discriminatory ability in identifying patients at increased risk of cardiovascular mortality (HR 21.3, 95% CI: 2.4–193.2) as it does for all-cause mortality (Figure 3D). ECV and GLS did not reach statistical significant discriminatory power for cardiovascular mortality (Figures 3E and F).
Kaplan–Meier curves for all-cause mortality and cardiovascular mortality in 110 subjects with VEDOSS or established SSC over a median follow-up duration of 7 years. Shown are estimates of the probability of death from any cause stratified by the native T1 > 1065 ms measured at baseline (A). Elevated ECV > 30% exhibits a similar graph but does not reach statistical significance (B). Impaired GLS identified a small group of patients at high risk of mortality while the majority of deaths (9 of 12) occurred in the group with normal GLS (C). The right panels show the probability of cardiovascular death stratified by native T1 (D), ECV (E), and GLS (F).
Cox proportional hazard regression analyses for the CMR parameters of diffuse fibrosis and impaired strain
| Cox proportional hazard regression | ||||
|---|---|---|---|---|
| Univariate analysis | Multivariate analysis | |||
| Hazard ratio (95% CI) | P-value | Hazard ratio (95% CI) B-exponent | P-value | |
| Native T1 | 5.8 (1.7–20.4) | 0.006 | 4.6 (1.5–14.6) | 0.01 |
| ECV | 3.9 (0.8–20.1) | 0.076 | ||
| GLS | 6.1 (1.3–29.9) | 0.038 | 4.2 (1.1–15.7) | 0.02 |
| GCS | 1.6 (0.5–5.2) | 0.475 | ||
| Cox proportional hazard regression | ||||
|---|---|---|---|---|
| Univariate analysis | Multivariate analysis | |||
| Hazard ratio (95% CI) | P-value | Hazard ratio (95% CI) B-exponent | P-value | |
| Native T1 | 5.8 (1.7–20.4) | 0.006 | 4.6 (1.5–14.6) | 0.01 |
| ECV | 3.9 (0.8–20.1) | 0.076 | ||
| GLS | 6.1 (1.3–29.9) | 0.038 | 4.2 (1.1–15.7) | 0.02 |
| GCS | 1.6 (0.5–5.2) | 0.475 | ||
CI, confidence interval; ECV, extracellular volume; GCS, global circumferential strain; GLS, global longitudinal strain.
Cox proportional hazard regression analyses for the CMR parameters of diffuse fibrosis and impaired strain
| Cox proportional hazard regression | ||||
|---|---|---|---|---|
| Univariate analysis | Multivariate analysis | |||
| Hazard ratio (95% CI) | P-value | Hazard ratio (95% CI) B-exponent | P-value | |
| Native T1 | 5.8 (1.7–20.4) | 0.006 | 4.6 (1.5–14.6) | 0.01 |
| ECV | 3.9 (0.8–20.1) | 0.076 | ||
| GLS | 6.1 (1.3–29.9) | 0.038 | 4.2 (1.1–15.7) | 0.02 |
| GCS | 1.6 (0.5–5.2) | 0.475 | ||
| Cox proportional hazard regression | ||||
|---|---|---|---|---|
| Univariate analysis | Multivariate analysis | |||
| Hazard ratio (95% CI) | P-value | Hazard ratio (95% CI) B-exponent | P-value | |
| Native T1 | 5.8 (1.7–20.4) | 0.006 | 4.6 (1.5–14.6) | 0.01 |
| ECV | 3.9 (0.8–20.1) | 0.076 | ||
| GLS | 6.1 (1.3–29.9) | 0.038 | 4.2 (1.1–15.7) | 0.02 |
| GCS | 1.6 (0.5–5.2) | 0.475 | ||
CI, confidence interval; ECV, extracellular volume; GCS, global circumferential strain; GLS, global longitudinal strain.
Clinical signs of internal organ involvement do not indicate subtle myocardial involvement
Furthermore, we investigated the association between clinical signs for pulmonary or cardiac organ involvement and the presence of early fibrotic myocardial alterations. Patients with pulmonary hypertension as judged by echocardiography or lung fibrosis by HRCT were considered to have clinical evidence of lung involvement. Palpitations, any conduction block or highly elevated NT-proBNP > 650 pg/mL28 defined clinical signs of cardiac involvement. Neither native T1 nor ECV was different between patients with or without clinical signs of internal organ involvement.
Association with disease subtype and classification
The markers of myocardial fibrosis were not different between patients with limited cutaneous SSc and diffuse cutaneous SSc (native T1 P = 0.97, ECV P = 0.89). Myocardial strain, however, exhibited a trend to lower longitudinal function in the diffuse cutaneous SSc group (GLS P = 0.07) and a weaker trend to lower circumferential function (GCS P = 0.13).
When the patients with established SSc according to the newer 2013 ACR/EULAR criteria but not fulfilling the 1980 ACR criteria were compared with patients that were positive for both criteria, no significant difference in myocardial function or fibrosis could be found. The results are summarized in Table 3.
Tissue characterization and myocardial strain in limited cutaneous SSc vs. diffuse cutaneous SSc and 1980 ACR criteria positive patients vs. patients that only fulfill the 2013 ACR/EULAR criteria for SSc
| Limited cutaneous SSc (N = 87) | Diffuse cutaneous SSc (N = 23) | P-value | |
|---|---|---|---|
| Native T1 (ms) | 1033 ± 48 | 1033 ± 53 | P = 0.97 |
| ECV (%) | 31.1 ± 5.5 | 30.1 ± 3.5 | P = 0.89 |
| GLS (%) | −21.1 ± 2.7 | −19.9 ± 3.2 | P = 0.07 |
| GCS (%) | −28.3 ± 4.0 | −26.6 ± 4.7 | P = 0.13 |
| 2013 ACR/EULAR criteria only (N = 25) | 1980 ACR criteria (N = 61) | ||
| Native T1 (ms) | 1029 ± 45 | 1036 ± 54 | P = 0.54 |
| ECV (%) | 31.1 ± 6.5 | 31.6 ± 4.9 | P = 0.37 |
| GLS (%) | −21.0 ± 2.1 | −20.2 ± 3.0 | P = 0.21 |
| GCS (%) | −27.9 ± 3.4 | −27.9 ± 4.9 | P = 0.99 |
| Limited cutaneous SSc (N = 87) | Diffuse cutaneous SSc (N = 23) | P-value | |
|---|---|---|---|
| Native T1 (ms) | 1033 ± 48 | 1033 ± 53 | P = 0.97 |
| ECV (%) | 31.1 ± 5.5 | 30.1 ± 3.5 | P = 0.89 |
| GLS (%) | −21.1 ± 2.7 | −19.9 ± 3.2 | P = 0.07 |
| GCS (%) | −28.3 ± 4.0 | −26.6 ± 4.7 | P = 0.13 |
| 2013 ACR/EULAR criteria only (N = 25) | 1980 ACR criteria (N = 61) | ||
| Native T1 (ms) | 1029 ± 45 | 1036 ± 54 | P = 0.54 |
| ECV (%) | 31.1 ± 6.5 | 31.6 ± 4.9 | P = 0.37 |
| GLS (%) | −21.0 ± 2.1 | −20.2 ± 3.0 | P = 0.21 |
| GCS (%) | −27.9 ± 3.4 | −27.9 ± 4.9 | P = 0.99 |
Data are presented as mean ± SD. Differences between groups were tested using an unpaired two-tailed Student t-test or a Mann–Whitney U as appropriate. ECV, extracellular volume; GCS, global circumferential strain; GLS, global longitudinal strain; SSc, systemic sclerosis.
Tissue characterization and myocardial strain in limited cutaneous SSc vs. diffuse cutaneous SSc and 1980 ACR criteria positive patients vs. patients that only fulfill the 2013 ACR/EULAR criteria for SSc
| Limited cutaneous SSc (N = 87) | Diffuse cutaneous SSc (N = 23) | P-value | |
|---|---|---|---|
| Native T1 (ms) | 1033 ± 48 | 1033 ± 53 | P = 0.97 |
| ECV (%) | 31.1 ± 5.5 | 30.1 ± 3.5 | P = 0.89 |
| GLS (%) | −21.1 ± 2.7 | −19.9 ± 3.2 | P = 0.07 |
| GCS (%) | −28.3 ± 4.0 | −26.6 ± 4.7 | P = 0.13 |
| 2013 ACR/EULAR criteria only (N = 25) | 1980 ACR criteria (N = 61) | ||
| Native T1 (ms) | 1029 ± 45 | 1036 ± 54 | P = 0.54 |
| ECV (%) | 31.1 ± 6.5 | 31.6 ± 4.9 | P = 0.37 |
| GLS (%) | −21.0 ± 2.1 | −20.2 ± 3.0 | P = 0.21 |
| GCS (%) | −27.9 ± 3.4 | −27.9 ± 4.9 | P = 0.99 |
| Limited cutaneous SSc (N = 87) | Diffuse cutaneous SSc (N = 23) | P-value | |
|---|---|---|---|
| Native T1 (ms) | 1033 ± 48 | 1033 ± 53 | P = 0.97 |
| ECV (%) | 31.1 ± 5.5 | 30.1 ± 3.5 | P = 0.89 |
| GLS (%) | −21.1 ± 2.7 | −19.9 ± 3.2 | P = 0.07 |
| GCS (%) | −28.3 ± 4.0 | −26.6 ± 4.7 | P = 0.13 |
| 2013 ACR/EULAR criteria only (N = 25) | 1980 ACR criteria (N = 61) | ||
| Native T1 (ms) | 1029 ± 45 | 1036 ± 54 | P = 0.54 |
| ECV (%) | 31.1 ± 6.5 | 31.6 ± 4.9 | P = 0.37 |
| GLS (%) | −21.0 ± 2.1 | −20.2 ± 3.0 | P = 0.21 |
| GCS (%) | −27.9 ± 3.4 | −27.9 ± 4.9 | P = 0.99 |
Data are presented as mean ± SD. Differences between groups were tested using an unpaired two-tailed Student t-test or a Mann–Whitney U as appropriate. ECV, extracellular volume; GCS, global circumferential strain; GLS, global longitudinal strain; SSc, systemic sclerosis.
Myocardial involvement is not correlated with SSc disease duration or skin involvement
Disease duration was defined as the time from the onset of first non-RP symptom or time from first RP symptom in patients with RP as only clinical manifestation. No parameter of myocardial fibrosis (native T1 ρ = 0.065, P = 0.51; ECV ρ=0.122, P = 0.28) or myocardial deformation (GLS ρ = 0.116, P = 0.23; GCS ρ = −0.068, P = 0.48) correlated with disease duration. Likewise, no cardiac parameter correlated with skin involvement as quantified by the modified Rodnan skin score (native T1 ρ = 0.163, P = 0.091; ECV ρ = 0.015, P = 0.89; GLS r = 0.113, P = 0.25; GCS ρ = −0.119, P = 0.22).
Discussion
The presented work is the largest study assessing diffuse myocardial fibrosis and subtle systolic impairment in SSc patients, and the first one to differentiate myocardial involvement in very early diagnosis of SSc (VEDOSS) patients vs. patients with established SSc according to the 2013 ACR/EULAR criteria. Our data demonstrate that diffuse myocardial fibrosis is already present in VEDOSS patients to a similar extend as in patients with established SSc. In contrast, subtle impairment of systolic ventricular function was only observed in patients with established SSc diagnosis. LGE, as a sign of manifest SSc cardiomyopathy was present in ~10% of patients within the VEDOSS and the ACR/EULAR SSc groups. Systolic heart failure with ejection fraction < 50% was not found in any SSc group, in accordance with previous studies.29,30 Our finding of impaired GLS in patients with established SSc is in line with previous echocardiography-based investigations which reported impaired longitudinal strain in SSc patients with preserved ejection fraction.17,31 Regarding prognostication, a study by van Wijngaarden et al.32 revealed that left ventricular GLS is the only echocardiographic parameter that independently predicts all-cause mortality and cardiovascular hospitalization. This study aimed for using GLS to identify cardiac involvement in SSc at an early stage, for individualized risk stratification. However, myocardial strain in a very early stage of SSc as in VEDOSS patients has not been investigated, yet. Our finding that native T1 and ECV, but not strain are altered in VEDOSS patients indicates, that markers of myocardial fibrosis may be superior for very early risk stratification in SSc patients.
While previous studies also found elevated native T1 in SSc patients,33 recent studies reported conflicting data on the prognostic value of diffuse myocardial fibrosis. Ross et al.34 identified no association between native T1 or LGE and ventricular arrhythmias on ambulatory ecg in a small cohort of 32 SSc patients. In contrast, Bordonaro et al.,35 who also investigated a small group of 33 SSc patients, report significant divergence of the event-free survival curves for cardiac death, haemodynamically significant arrhythmia, or heart failure in SSc patients with elevated T1 and ECV. Our data substantiate the predictive value of quantifying native T1 with CMR, which proved to identify subjects with T1 elevation who are at 5.8 times higher risk of death over a median 7-year follow-up period. Of note, of the five fatalities in the group below the optimal T1 threshold, there was one patient with pathologically low native T1 and one patient that died of traumatic brain injury, suggesting that only three patients with normal T1 had a likely disease-related death. While the ECV threshold of >30% did not reach statistical significance to identify patients at risk for the primary outcome, the Kaplan–Meier curves show a similar pattern compared with native T1. There is a remarkable similarity between our optimal prognostic thresholds for native T1 and ECV and the values reported by Bordonaro et al.35 which raises hope for a generalizability of the data, at least for the given T1 mapping sequence at 1.5 T. Interestingly, there were no differences regarding LV function, LV volume, or LGE between the high and low native T1 groups, suggesting that these patients at elevated risk could not be identified by conventional CMR parameters. A small group of patients with severely impaired GLS also proved to be at elevated risk for fatal outcome. However, these patients also had lower LV-ejection fraction as assessed by CMR and the majority of fatalities occurred in the group with normal GLS. Death in patients with SSc is often of multifactorial aetiology which makes it difficult to attribute individual cases to cardiovascular mortality. However, in the subgroup of patient with identifiable cause of death, native T1 remained the only, highly significant predictor of cardiovascular mortality.
Native T1, ECV, and myocardial strain were not significantly different between diffuse cutaneous SSc (dcSSc) and limited cutaneous SSc (lcSSc) patients. This is at variance with a previous study by Ntusi et al. who reported higher native T1 and ECV values in dcSSc vs. lcSSc but included only 10 and 9 patients, respectively. Ultimately, our study did only include few patients with advanced cardiac involvement, consequently differences between dcSSc and lcSSc that might become apparent in very advanced stages of the disease were not within the scope of this study.
The lack of association with SSc subtype, disease duration, and clinical signs of internal organ involvement precludes the assessment of cardiac involvement based solely on clinical data. Moreover, our finding of myocardial fibrosis even before the definite diagnosis of established SSc implies a need for cardiac imaging in patients with very early SSc. Since, 50% of the patients with manifest myocardial fibrosis defined as LGE experience cardiac events over 5-year follow-up,14 the prognostic value of these earlier markers of cardiac involvement may help to define rule-in criteria for patients that are in need of intensive cardiac follow-up. On the other hand, the fact that even after decades of disease duration SSc does not necessarily lead to myocardial involvement raises hope that also rule-out criteria or a non-cardiac phenotype may be defined based on quantitative tissue characterization and myocardial deformation imaging. Normal native T1 and normal myocardial strain could serve as a potential marker for SSc patients with an annual event rate of <1%.
Limitations of this study are the reduced sample size and smaller number of events in patients that had ECV measurements since it may underestimate the prognostic value of ECV. However, because ECV assessment requires the administration of gadolinium-based contrast agent, native T1 may be better suited for screening and routine follow-up CMR examinations in SSc patients. In addition, the lack of quantitative T2 mapping precludes the evaluation of the effect of global myocardial oedema as part of this study. However, with the acquired T2-weighted TSE images, no signs of focal oedema were found in a SSc patient.
Conclusion
In conclusion, diffuse myocardial fibrosis assessed by the expansion of ECV and increased native T1 precedes subtle functional impairment in patients with very early stages of systemic sclerosis (VEDOSS). LV deformation analysis identified impaired strain parameters only for GLS in patients with established SSc. Quantification of native T1 and GLS with CMR showed significant associations with mortality. These CMR parameters might prove useful for identifying suitable patients for intensified cardiac follow-up and those who are at low risk of fatal outcome. No significant differences in cardiac involvement could be found regarding the SSc subtype, disease duration, or skin involvement.
Supplementary data
Supplementary data are available at European Heart Journal – Cardiovascular Imaging online.
Data availability
The data underlying this article are available from the corresponding author on reasonable request.
References
Author notes
Oliver Distler and Robert Manka contributed equally to the manuscript.
Conflict of interest: B.M. has/had grant/research support from Abbvie, Protagen, Novartis Biomedical Research, received speaker fees from Böhringer Ingelheim as well as congress support from Pfizer, Roche, Actelion, Mepha, and MSD. C.M. has received congress support from Actelion and Roche, and personal fees from Boehringer-Ingelheim, Mepha, and MEDtalks Switzerland, outside the submitted work. R.D. reports grants/research support from Pfizer, Actelion, and personal fees (speaker/consultancy) from Actelion and Boehringer-Ingelheim, outside the submitted work. O.D. has/had consultancy relationship with and/or has received research funding from or has served as a speaker for the following companies in the area of potential treatments for systemic sclerosis and its complications in the last 3 years: Abbvie, Acceleron, Alcimed, Amgen, AnaMar, Arxx, AstraZeneca, Baecon, Blade, Bayer, Boehringer-Ingelheim, ChemomAb, Corbus, CSL Behring, Galapagos, Glenmark, GSK, Horizon (Curzion), Inventiva, iQvia, Kymera, Lupin, Medac, Medscape, Miltenyi Biotec, Mitsubishi Tanabe, Novartis, Prometheus, Roche, Roivant, Sanofi, Serodapharm, Topadur and UCB. Patent issued ‘mir-29 for the treatment of systemic sclerosis’ (US8247389, EP2331143). O.D. and B.M. have a patent for ‘mir-29 for the treatment of systemic sclerosis’ (US8247389, EP2331143). All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
