Association between systemic sclerosis and left ventricle dysfunction: Findings from observational studies

Abstract

Objectives

Cardiac involvement is common in systemic sclerosis (SSc) patients. In this study, we aimed to systematically evaluate the relationship between SSc and left ventricular dysfunction (LVD), especially the left ventricular diastolic dysfunction, by ultrasound and cardiac magnetic resonance data.

Methods

We searched The Cochrane Library, PubMed and Embase databases collected studies about comparing LVD parameters in SSc patients and controls from establishment to January 2022. Furthermore, we also performed a two-sample MR using summary statistics from genome-wide association studies (GWAS) important LVD parameters, including left ventricular end-diastolic volume (LVEDV), left ventricular mass (LVM) and left ventricular ejection fraction (LVEF).

Results

Our meta-analysis included 31 eligible studies with 1448 SSc patients. According to the results, SSc patients had lower peak of early diastolic flow velocity/peak of late diastolic flow velocity ratio (E/A ratio), E, trans-mitral early filling peak velocity (E′), and left ventricular end-diastolic diameter (LVEDD) compared to controls. The E/E’ ratio, A, left ventricular isovolumetric relaxation time (IVRT), deceleration Time (DT) and left atrial (LA) diameter were higher in SSc patients in comparison with controls. Moreover, we observed that the SSc patients had lower LVEF than controls. And in MR analysis, we also found that SSc was causally correlated with LVEF (OR = 0.9966, 95% CI 0.9935–0.998, P = 0.0398). However, unfortunately, there was no significant correlation between SSC and LVM (OR = 1.0048, 95% CI 0.9919–1.0179, P = 0.4661) and LVEDV (LVEDV OR = 0.9976, 95%CI 0.9888–1.0066, P = 0.6019).

Conclusion

SSc patients had diastolic/systolic dysfunction. However, MR analysis cannot confirm the genetic relationship between SSc and LVDD because of insufficient data. More research is needed to confirm the causal relationship between the two.

• ●SSc patients is more likely to have left ventricular dysfunction.
• ●This result suggests that we should enhance early cardiovascular screening for SSc patients to improve patient prognosis.

1. Introduction

Systemic sclerosis (SSc), also called scleroderma, is an immune connective tissue disease characterized by a unique pathological triad of fibrosis of the skin and internal organs, vasculopathy, as well as immune system dysregulation. It also affects the gastrointestinal tract, lung, heart, and other organ systems with high mortality rate. Weiss et al. reported cardiac involvement in SSc patients in 1943, which has attracted wide attention ever since [1]. Although skin fibrosis is the most characteristic manifestation of SSc, pathological changes in the lungs, gastrointestinal tract, kidneys, and heart are determinants of clinical prognosis. Meanwhile, the estimated clinical prevalence of cardiac involvement in SSc patients has been reported to be 15%–35%, and there is a high mortality rate due to heart disease [2]. Although survival for SSc has improved over the past few decades, the proportion of deaths due to cardiac complications has not changed significantly. Cardiovascular disease still accounts for 20–30% of disease-related mortality [3]. SSc with cardiac involvement, pathologic manifests as myocardial fibrosis, cardiac autonomic dysfunction, small coronary vessels, or small peripheral small blood vessels of the heart, and advanced patients present with pericardial effusion, atrial or ventricular arrhythmias, impaired conduction system, damaged heart valves (less common), myocardial ischemia, myocardial hypertrophy, and heart failure [4]. In recent years, many studies have suggested that once symptoms of cardiac involvement in SSc patients are usually related to late signs of heart failure, accelerated progression is more difficult to control, and mortality is high [5], so timely diagnosis and aggressive control of cardiac involvement in SSc patients is particularly important. LVD, especially LVDD, often appears in the early stage of cardiac lesions [6]. it is also known that LVD can lead to heart failure is consistently linked to high mortality rates. LVD is often diagnosed by imaging method including trans-thoracic echocardiography (TTE). TTE includes M-mode echocardiography, two-dimensional ultrasound, Doppler echocardiography, and tissue doppler imaging (TDI). It is worth mentioning that cardiovascular magnetic resonance (CMR) as a newly developed non-invasive and non-radiative technique also can assess cardiac function and perform tissue characterization. CMR has the capability to detect presentations such as edema, infiltration, ischemia, and fibrosis of the cardiac muscles for the early diagnosis of cardiac involvement in SSc [7,8]. CMR can be used as a useful additional tool to complement the clinical skills of physicians and echocardiography to assess cardiac involvement in asymptomatic SSc patients [9]. In the future, we should also pay more attention to exploring the role of modern imaging technologies, including CMR, in early diagnosis.

Although we know that left ventricular systolic dysfunction is a well-known complication in SSc [10]. LVD remains poorly studied in SSc, especially the LVDD. The prevalence of LVDD in SSc remains controversial. The genetic link between SSc and LVD is even more inconclusive. Therefore, it is necessary to systematically review the relationship between SSc and LVD to help people better understand and knowledge of this severe complication of SSc. All in all, in order to systematic evaluate the relationship between SSc and LVD, we carried out a systematic review and a two-sample Mendelian randomization study (MR) separately by including echocardiographic data and cardiac magnetic resonance data.

2. Method

Ethical approval was obtained from the ethics committee of the Third People's Hospital of Chengdu (2019-S-20).

2.1. Meta-analysis

2.1.1. Search strategy

We conducted this study based on requirements from the Meta-Analysis of Observational Studies in Epidemiology group [11]. In our research, A literature review was executed using online databases, including The Cochrane Library, PubMed and Embase databases to collect relevant studies comparing cardiac ultrasound indicators in SSc patients and controls, from inception to January 2022. The search terms included: systemic sclerosis, scleroderma, cardiac function, heart failure, heart function, diastolic function, and diastolic dysfunction. The search details are shown in Fig. 1.

Fig. 1.

Flow chart of the systematic search and of the qualitative synthesis for meta-analysis.

2.1.2. Inclusion selection

(ⅰ) These controlled clinical studies investigating the association between SSc and LVDD; (ii) The case group consisted of patients who met the diagnostic criteria of the American College of Rheumatology (ACR) in 1980 or the 2013 Classification Criteria for Systemic Sclerosis: ACR/European League Against Rheumatism Collaborative Initiative (EULAR). And the control group consisted of people without SSc and other connective tissue diseases; (iii) Containing the following outcome indicators: M-mode echocardiography and two-dimensional echocardiography detection indicators, including left atrial diameter (LA), left ventricular mass index (LVMI), left ventricular posterior wall end-diastolic thickness (posterior wall, PW), ventricular septal end-diastolic thickness (interventricular septal, IVS), and left ventricular end-diastolic diameter, left ventricular ejection fraction (ejection faction, EF); Doppler echocardiography detection indicators, including trans-mitral early filling peak velocity (E) value, trans-mitral late filling peak velocity (A) value, E/A ratio, mitral valve Deceleration Time (DT), and left ventricular isovolumetric relaxation time (IVRT); TDI detection index, including trans-mitral early filling peak velocity (E') value, trans-mitral late filling peak velocity (A′) value, E/E′ ratio, E'/A′ ratio, and left ventricular isovolumetric relaxation time (isovolumetric relaxation time, IVRT').

2.1.3. Exclusion criteria

(i) Articles that are not observational studies. (ii) These subjects were not SSc patients and control groups, and studies in specific populations, such as malignant tumors. (iii) Studies for which echocardiographic measurements were not available.

2.1.3.1. Data extraction

Two researchers (Wei Yan and Qiong Nie) independently reviewed the studies to confirm whether they met the inclusion criteria and then extracted the qualified data. When the opinions are inconsistent, a third reviewer (Qiang Luo) would discuss with us to reach a consensus. Extracts from this systematic review mainly included: first author, publication year, article type, information about participants (number of cases, age, gender, country, course of disease), echocardiographic parameters (E, A, E/A ratio, E′, E/E′ ratio, LVEF, LA, PW, DT, IVS, LVEDD). In order to avoid double counting, when there are more than one publication involving the same research cohort, we selected the most comprehensive set of studies.

2.1.4. Quality evaluation

We evaluated the quality of included studies based on the Newcastle-Ottawa Quality Assessment Scale [12]. Scores were calculated based on three major study components: case selection (A score of 0–4), comparing cases with controls (A score of 0–2), and exposure assessment (A score of 0–3). Higher scores reflect a higher level of methodological quality.

2.1.5. Statistical analysis

We used RevMan 5.4.1 software for meta-analysis. Then, we summarized the effect size of each echocardiographic variable with the mean difference and its 95% confidence interval and represented it with a forest plot. In order to describe the distributions of continuous variables, percentages and mean ± standard (SD)were calculated, standardized mean differences (SMD) can be used to resolve unit inconsistencies. We used the I² index to estimate heterogeneity among included studies. Statistical heterogeneity was considered significant if I² statistics were≥50%. A random-effects model was utilized if significant heterogeneity (p < 0.5, I² > 50%) was observed; If not, we use the fixed effects model. For apparent heterogeneity, we used sensitivity analysis by exclusion of each research, one at a time. In order to assess publication bias, a funnel plot test was utilized. Statistical significance was considered as p < 0.05.

2.2. Mendelian randomization study

In this study, the two-sample MR approach was used to explore the causal association between SSc and important LVD parameters (LVEDV, LVM and LVEF). MR is a powerful statistical method that can be used to make causal inferences in exposure-outcome relationships. Furthermore, because genetic variation is innate, the association of genetic variation with the outcome is consistent with a causal timing, and is not influenced by acquired factors [13]. Hence, MR can effectively overcome the bias caused by confounding and reverse causality.

2.2.1. Data accessed

Summary data of SSc were obtained from a meta-analysis of the largest SSc genome-wide association studies (GWAS) [14], including 9,095cases and 17,584 controls of European ancestry. As gene loci of these metabolites rarely reach genome-wide significance in GWAS, we chosen single nucleotide polymorphisms (SNPs) with suggestive genome-wide significance thresholds (p < 5 × 10⁻⁵) as instrumental variables (IVs) in this research (Table 1 [12], [13], [12], [13],).

Table 1.

Systemic sclerosis and Important LVD Parameters summary data sources.

Trait	Year	Sample size	Case	Control	Population	Reference
Systemic sclerosis	2019	26,679	9095	17,584	Europeans	[12]
Important LV Parameters (LV end-diastolic volume, LV mass and LVEF)	2019	16,920	NA	NA	Europeans	[13]

We selected the important LVD parameters (LVEDV, LVM and LVEF) to explore LVD. Studies of the UK Biobank's cardiovascular magnetic resonance provided these measurements of LVD parameters [15]. During 2006 to 2010, the UK Biobank recruited half a million people aged 40 to 69 based on prospective cohort studies. There were 16,923 European UK Biobank participants (mean age 62.5 years; 45.8% men) without generalized myocardial infarction or heart failure in this study. Moreover, it has collected information on health and lifestyle data, physical measurements, biological samples, genotype, and cMRI-derived cardiac phenotypes. Details of the included date were described in Table 1.

2.3. Instrumental variables selection

Initially, we applied the clump module in PLINK software to calculate the selected IVs. The parameters were set to a genetic distance of 10,000 kb, a threshold of 0.001 for the linkage disequilibrium parameter (r²), and P-value threshold of 5 × 10⁻⁵ for the SNPs. Second, using PhenoScanner and Catalog, we checked whether the included SNPs were related to known confounders. If so, SNP will be excluded. Last, F-statistic will be used to calculate whether the SNPs included in this study were affected by weak IVs [16], and the inclusion of IVs with F-statistics less than 10 can be rejected to avoid bias on the results. The Selection of IVs is shown in Fig. 2, and characteristics of SNPs used in MR analysis showed in Supplementary Table 2.

Fig. 2.

Directed acyclic graphs for the classical Mendelian randomization designs. The arrows denote causal relations between two variables, pointing from the cause to the effect. The causal pathway is blocked if “X” is placed in the arrowed line.

2.3.1. Statistical analysis

Our study investigated the causal relationships between SSc and LV parameters mainly using the IVW fixed effects models [17]. Moreover, as a second sensitivity analysis, we additionally used MR Egger, Maximum likelihood, and Weighted median for the MR analysis.

2.4. Horizontal pleiotropy and heterogeneity tests

First, we used “leave-one-out” sensitivity analyses and funnel plot method to assess the relationship between each IVs and outcome. Then, the MR Egger intercept test was used to explain the horizontal multiplicity and the effect of genetic pleiotropy was considered to be small if the intercept of the MR-egger regression line was close to zero [18]. Finally, Cochran Q statistic was also used to analyze the heterogeneity of the IVs included [19]. The Two Sample MR package in R software was used for statistical analysis. Meanwhile, all results were expressed as OR and its 95% CI, with a P value < 0.05 being a statistically significant difference.

3. Result

3.1. Meta-analysis

3.1.1. Results of literature screening

According to our search criteria, a total of 1249 articles were searched after deduplication. After an initial screening of titles and abstracts, 65 articles were eligible. The full-text screening revealed 31 articles that met our inclusion criteria [[20], [21], [22], [23], [24], [25], [26], [27], [28], [28], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50]]. Finally, 1448 SSc patients and 1006 healthy controls were involved, and the flowchart for retrieval and inclusion of articles is shown in Fig. 1.

3.1.2. Research features

We included 30 case-control studies, and only the study by Maione et al. [30] was a cohort study. At the same time, the included studies are from various countries, including 8 studies from Italy, 6 studies from Turkey, 3 studies from France, 2 studies from Poland and Serbia, and the remaining studies from China, Tunisia, Sweden, Serbia, New Zealand, Korea, Ireland, Greece, Germany and Romania. The included population is basically distributed in Europe and Asia, with only one from Africa and one from Oceania (Table 2 [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50],).

Table 2.

Features of studies included in meta-analysis.

Author, Year [ref.]	country	Distributed	Subjects (T/C)	Female (T/C)	Average age (T/C)	Age/sex matched (T/C)	disease duration	Quality score	Outcome indicators
Burak ERER2010 [20]	Turkey	Asia	20/20	18/18	44 ± 9.6/41 ± 10	Y/Y	N	4/2/2	⑥⑨⑩⑪
Karadag2020 [21]	Turkey	Asia	47/36	42/33	52.1 ± 12.4/49.4 ± 8.4	Y/Y	8.5 ± 5.9 yrs	2/2/2	②⑤⑥⑦⑧⑨⑩⑫ ⑭
Guerra 2018 [22]	Italy	Europe	52/52	46/46	54.6 ± 16.1/53.9 ± 16.6	Y/Y	6 (3–10) yrs	4/2/2	⑥⑨⑭
Ciurzyński2013 [23]	Poland	Europe	111/21	101/18	54.2 ± 13.7/49.3 ± 10.5	Y/Y	9.4 ± 11.4 yrs	4/2/2	⑦⑧⑨⑩⑪
Cadeddu2015 [24]	Italy	Europe	45/20	36/16	60.4 ± 10.3/60.8 ± 10.8	Y/Y	6.3 ± 5.75 yrs	3/2/2	②⑥⑨
Halil Ataş2015 [25]	Turkey	Asia	41/38	38/35	49.5 ± 11.6/48.5 ± 10.8	N/N	6.3 ± 4.7 yrs	4/2/2	1 ②③④⑤⑥⑩⑭
D'Andrea 2004 [26]	Italy	Europe	23/25	20/20	56.3 ± 8.2/55 ± 9.3	Y/Y	N	3/2/2	④③⑤⑦⑧⑨⑪⑫
Agoston2014 [27]	Italy	Europe	42/42	40/40	50 ± 14/49 ± 13	Y/Y	N	3/1/3	⑥⑨⑭
Zlatanovic2016 [28]	Serbia	Europe	41/30	38/28	56 ± 11/54 ± 9	Y/Y	6 ± 3.8 yrs	4/2/2	④⑤⑥⑨⑩
Zairi2018 [29]	Tunisia	Africa	25/25	24/25	53.64 ± 13.456/60.8 ± 8.72	Y/Y	N	3/2/2	2 ④⑤⑥⑨
Giunta2000 [30]	Italy	Europe	77/36	74/33	53.6 ± 11.6/49.6 ± 10.3	N/N	17.3 ±9yrs	3/2/2	⑥⑦⑧⑩⑪
Meune2008 [31]	France	Europe	100/26	86/21	53.7 ± 13.9/49.6 ± 13.0	Y/Y	7.9 ± 7.9yrs	4/2/2	1 ③④⑤⑥⑨
Maione2005 [32]	Italy	Europe	77/45	71/41	54.4 ± 10.9/51.9 ± 9.7	Y/Y	18.2 ± 9.2yrs	4/2/2	1 ③④⑥⑦⑧⑨⑩
KAZZAM1990 [33]	Sweden	Europe	30/48	15/22	54.5 ± 2.4/54.6 ± 2.1	Y/Y	5.6 (0.5–23) yrs	4/2/2	1 ②③④
Gerede2014 [34]	Turkey	Asia	31/21	29/18	50 ± 9/47 ± 6	N/N	N	2/2/2	1 ④⑤⑥⑦⑧⑨⑩⑪⑫⑭
ARMSTRONG1996 [35]	New Zealand	Oceania	35/35	33/33	45 ± 14/48 ± 15	Y/Y	8 (1–20) yrs	2/2/2	3 ③②⑤⑨①⑩
Lee 2010 [36]	South Korea	Asia	35/35	26/25	49.2 ± 12.8/51.6 ± 8.3	Y/Y	9.8 ± 2.2 mos	3/2/2	1 ⑤⑥④③⑦⑧⑨⑪⑫⑭
Ciurzyńsk2008 [37]	Poland	Europe	51/31	47/28	53.3 ± 15.2/52.68 ± 12.1	Y/N	9 ± 12.4yrs	4/2/2	⑦⑧⑨⑪⑫
D'Alto 2014 [38]	Italy	Europe	74/71	69/66	19-71/18-72	N/N	(12)1–43yrs	3/2/2	2 ④⑤⑥⑦⑧⑨
D'Andrea 2007 [39]	Italy	Europe	33/30	24/23	56.3 ± 8.2/55 ± 9.3	Y/Y	N	3/2/2	3 ③④⑤⑥⑦⑧⑨⑪⑫
Morelli 1996 [40]	Ireland	Europe	76/64	63/56	48.0/44.0	Y/Y	(10)1–33yrs	4/2/2	4 ④⑤
Kepez 2008 [41]	Turkey	Asia	27/26	26/25	46.0 ± 10.0/44.0 ± 10.0	Y/Y	(10)3–18yrs	4/2/3	1 ⑤⑥⑪⑫⑬⑮
Gullulu 2005 [42]	Turkey	Asia	22/22	20/20	50.2 ± 4.8/45.4 ± 13.4	N/Y	103.9 ± 102/3-396mos	2/2/2	2 ⑤⑥⑦⑧⑨⑫ ⑬⑭⑮
Ivanovic 2011 [43]	Serbia	Europe	50/48	45/44	55.0 ± 9.0/54.0 ± 8.0	Y/Y	(9)2–23yrs	4/1/2	③④⑤⑥⑦⑧⑨⑪⑫⑬⑮
Meune 2005 [44]	France	Europe	17/15	14/12	53.4 ± 8.1/53.1 ± 6.2	Y/Y	7.2 ± 3.5 yrs	4/2/2	5 ⑥
Yiu 2011 [45]	China	Asia	104/37	80/26	54.0 ± 12.0/55.0 ± 8.0	Y/Y	5.1 ± 2.3 yrs	4/2/2	6 ⑫⑭
Dimitroulas 2010 [46]	Greece	Europe	52/25	51/17	55.7 ± 10.1/55.0 ± 8.0	Y/Y	11 (0.7–27) yrs	3/2/2	7 ⑦⑧⑨⑪⑫
Schattke 2010 [47]	Germany	Europe	22/22	17/17	57.0 ± 13.4/57.0 ± 13.9	Y/Y	N	3/2/2	⑫
Lindqvis 2005 [48]	Sweden	Europe	26/25	21/21	56 ± 15/56 ± 16	Y/Y	11.8 ± 8.7yrs	3/2/2	3 ④⑤⑥⑦⑧⑨⑫
Poanta 2009 [49]	Romania	Europe	20/15	18/12	52 ± 10.4/48 ± 12.8	Y/Y	46.5 ± 55.7 mos	4/2/2	4 ⑤⑥⑦⑧⑨
Meune 2004 [50]	France	Europe	42/20	35/17	54.3 ± 9.7/53.9 ± 10.0	Y/Y	2.5 ± 1.4 yrs	4/2/2	4 ④⑤⑥

Values are mean (SD) or median (range) unless otherwise stated. T:experiment; C: control; Y: yes; N: no; NR: not reported; yrs: years; mos: months. Outcome indicators: ①l eft atrial diameter (LA), ② left ventricular mass index (LVMI) ③ left ventricular posterior wall end-diastolic thickness (posterior wall, PW)④Left ventricular septal end-diastolic thickness (interventricular septal, IVS) ⑤ left ventricular end-diastolic diameter(LVEDD) ⑥ left ventricular ejection fraction (ejection faction, EF) ⑦ Trans-mitral early filling peak velocity (E) value⑧trans-mitral late filling peak velocity (A) value ⑨ E/A ratio ⑩ mitral valve Deceleration Time (DT) ⑪ left ventricular isovolumetric relaxation time (IVRT) ⑫ trans-mitral early filling peak velocity (E') value ⑬ trans-mitral late filling peak velocity (A′) value ⑭ E/E′ ratio ⑮ E'/A′ ratio ⑯ left ventricular isovolumetric relaxation time (isovolumetric relaxation time, IVRT').

3.1.3. Quality evaluation

The quality scores ranged from 6 to 8 points, with 23 of the 31 studies (74%) scoring above 7 points (Table 2).

3.1.4. Results of meta analysis

All studies used transthoracic echocardiography to examine cardiac function. LV diastolic dysfunction (DD) cannot be diagnosed by any single echocardiographic parameter. Individual and pooled analyses were required for each parameter to confirm the presence or absence of known American Society of Echocardiography (ASE) guideline-defined left ventricular diastolic dysfunction (LVDD) [51]. LV diastolic dysfunction was defined as pulmonary capillary wedge pressure (PCWP) or LV end-diastolic pressure (LVEDP) > 12 mmHg on catheterization [52]. The systematic review showed evidence of LVDD in SSc patients in these studies based on lower E/A ratio, higher E/E′ ratio, and prolonged IVRT or DT and/or increased LA diameter compared to controls. At the same time, we also clarified the presence of LV systolic dysfunction in SSc patients by exploring LVEF (Table 3).

Table 3.

Meta-analysis outcomes LA: left atrial diameter (LA); left ventricular mass index (LVMI), IVRT: left ventricular isovolumetric relaxation time; E/E′ ratio: peak of early diastolic flow velocity/mitral annular early diastolic velocity; A: trans-mitral late filling peak velocity value; DT: mitral valve Deceleration Time; E: trans-mitral early filling peak velocity value; E/A ratio: peak of early diastolic flow velocity/peak of late diastolic flow velocity; LVEDD: left ventricular end-diastolic diameter; E’: mitral annular early diastolic velocity; LVEF: left ventricular ejection fraction; PW: left ventricular posterior wall end-diastolic thickness; IVS: interventricular septal; LVMI: left ventricular mass index; A’: trans-mitral late filling peak velocity value; E’/A′ ratio: trans-mitral early filling peak velocity (E') value/trans-mitral late filling peak velocity (A′) value; 95% CI: 95% confidence interval; I², % ‡: ‡I²index for degree of heterogeneity; Q†: †Cochran's Q statistic for heterogeneity; T²§: §Tau-squared measure of heterogeneity.

Echo variables	Mean difference (95% CI)	P	I2, %‡	Q†	T2§
LA (n = 10)	1.90 (1.03, 2.76),	<0.0001	77%	38.72	1.13
IVRT (n = 10)	4.3 (2.55, 6.20)	<0.00001	41%	9.00	15.17
E/E′ ratio (n = 8)	0.98 (0.26, 1.96)	0.008	78%	31.33	0.80
A (n = 12)	6.92 (4.16, 9.69)	<0.00001	62%	28.63	13.55
DT (n = 15)	8.12 (2.82, 13.42)	0.03	56%	32.03	54.22
E (n = 14)	−0.27 (−0.40, −0.15)	<0.0001	53%	13.00	27.64
E/A ratio (n = 23)	−0.21 (−0.27, −0.14)	<0.0001	79%	105.53	0.02
LVEDD (n = 20)	−0.17 (−0.34, −0.01)	0.04	58%	45.66	0.08
E’ (n = 7)	−0.44 (−0.61, −0.26)	<0.00001	44%	6.00	10.73
LVEF (n = 24)	−1.29 (-1.95, −0.63)	0.0001	60%	57.51	1.21
PW (n = 16)	0.26 (-0.08, 0.59)	0.13	93%	220.87	0.37
IVS (n = 16)	0.37 (-0.25, 0.99)	0.25	98%	601.20	1.45
LVMI (n = 7)	5.18 (-3.36, 13.72)	0.23	95%	130.46	121.66
A’ (n = 5)	0.13 (-0.22, 0.49)	0.46	60%	9.88	0.10
E’/A′ ratio (n = 4)	−0.22 (-0.49, −0.05)	0.12	77%	13.14	0.06

The results of this meta-analysis are as follows: LA (Fig. 3A) and IVRT (Fig. 3C) were reported in 10 studies. The LA (mean difference 1.90 [95%CI1.03, 2.76], P < 0.0001), and IVRT(mean difference 4.37 [95%CI2.55, 6.20], P＜0.00001) was higher in SSc patients than in controls; At the same time, E/E′ ratio (Fig. 3B) (mean difference 0.98 [95%CI0.26, 1.69] P = 0.008), the A (mean difference 6.92 [95%CI4.16, 9.69]P＜0.00001), and DT (mean difference 8.12 [95%CI2.82, 13.42] P = 0.03), were all higher in SSc patients; while E (mean difference-0.27 [95%CI-0.40, −0.15]P < 0.0001), E/A ratio (Fig. 3D) (mean difference-0.21 [95%CI-0.27, −0.14] P < 0.0001), E’ (mean difference-0.44 [95%CI-0.61, −0.26] P < 0.00001)and LVEDD (mean difference-0.17 [95%CI-0.34, −0.01] P = 0.04) were lower in SSc patients as compared to controls. In addition, we observed that LVEF (Fig. 3E) was available in 24 studies. In SSc patients, LVEF (mean difference-1.29 [95%CI-1.95, −0.63] P < 0.0001) was lower than in controls; There was no significant difference in PW (P = 0.13), IVS (P = 0.25), LVMI (P = 0.23), A' (P = 0.46) and E'/A′ ratio (P = 0.12) between the two groups. The results of the publication bias analysis of the E/A ratios of the two groups showed no publication bias (Fig. 3F).

Fig. 3.

The results of our meta-analysis. A. LA comparison between SSc patients and controls. B. E/E′ comparison between SSc patients and controls. C. IVRT comparison between SSc patients and controls. D. E/A comparison between SSc patients and controls. E. LVEF comparison between SSc patients and controls. F. Funnel-plot for the E/A to assess for publication bias. MD = mean difference.

3.2. Mendelian randomization study

3.2.1. Genetic instrumental variables for SSc

In this study 75 SNPs were used to explore the MR analysis. In addition, we found that the F-statistics were all more than 10, indicating that the included IVs were not prone to influence the results (See Supplementary Table 1).

3.3. MR analysis of SSc and important LVD parameters

In the MR analysis, we observed that SSc were not associated with LV end-diastolic volume (OR = 0.9976, 95%CI 0.9888–1.0066, P = 0.6019) (Table 4; Fig. 4a) and LV mass (OR = 1.0048, 95% CI 0.9919–1.0179, P = 0.4661) (Table 4; Fig. 4b). However, we found a causal association between SSc and LVEF (OR = 0.9966, 95% CI 0.9935–0.998, P = 0.0398) (Table 4; Fig. 4c). The above relationship was also confirmed by Weighted median (OR: 0.9969, 95%Cl: 0.9931–1.0006. P = 0.1030), MR Egger (OR: 0.9978, 95%Cl:0.9938–1.0020, P = 0.3883).

Table 4.

MR analysis of the relationship between SSc and LVD.

Outcome	Exposure	MR Egger		Inverse variance weighted		Weighted median		Maximum likelihood		Heterogeneity P	Pleiotropy P
		Beta ±SE p		Beta ±SE p		Beta ±SE p		Beta ±SE P
LVEDV	SSc	−0.004 ± 0.008	0.6262	−0.002 ± 0.005	0.6019	−0.005 ± 0.005	0.2626	−0.003 ± 0.003	0.4445	0.0944	0.7759
LVM	SSc	−0.018 ± 0.030	0.5737	0.005 ± 0.007	0.4661	0.006 ± 0.008	0.4783	0.005 ± 0.007	0.4599	0.5721	0.4726
LVEF	SSc	−0.002 ± 0.002	0.3883	−0.003 ± 0.002	0.03980	−0.003 ± 0.002	0.1030	−0.003 ± 0.002	0.0671	0.5394	0.4154

Abbreviations: LVEDV: left ventricular end-diastolic volume; LVM: left ventricular mas; LVEF: left ventricular ejection fraction.

Fig. 4.

(a) Scatter plot to visualize causal effect of systemic sclerosis on left ventricular end-diastolic volume. The slope of the straight line indicates the magnitude of the causal association. IVW indicates inverse-variance weighted; and MR, Mendelian randomization. (b)Scatter plot to visualize causal effect of systemic sclerosis on left ventricular mass. The slope of the straight line indicates the magnitude of the causal association. IVW indicates inverse-variance weighted; and MR, Mendelian randomization. (c) Scatter plot to visualize causal effect of systemic sclerosis on left ventricular ejection fraction. The slope of the straight line indicates the magnitude of the causal association. IVW indicates inverse-variance weighted; and MR, Mendelian randomization.

3.3.1. Horizontal pleiotropy and heterogeneity

In the MR analysis, while incorporating more SNP tools usually results in more accurate results and provides better power for MR, it can also lead to higher levels of pleiotropy and heterogeneity. Therefore, we used a series of methods to assess whether there was significant horizontal pleiotropy and heterogeneity in the study. The MR-Egger intercept and Cochran's Q test reported that there was no susceptibility to horizontal pleiotropy and heterogeneity in this study (LVEDV P = 0.7759, LVM P = 0.4726, LVEF P = 0.4154, All P value > 0.05) (See Table 4). The “leave-one-out” sensitivity analysis showed that the SNPs included in the study did not significant impact the results, demonstrating that the results had significant reliability (see Fig. 4).

4. Discussion

As far as we know, it is a first large-scale comprehensive review and MR analysis examining the association between SSc and LVD. LVD includes LV systolic and diastolic dysfunction. In our work, based on the MR study, the results showed that we found a causal relationship between SSc and LV systolic dysfunction, which was also supported by the meta-analysis. It is worth mentioning that we also found that SSc patients were more likely to have LVDD compared to non-SSc patients. However, we found no evidence of causal effects of SSc on LVDD, suggesting that our MR results may be affected by potential bias.

For LV systolic dysfunction, the results of meta-analysis showed that the SSc patients had lower LVEF than in controls, at the same time, the MR findings also confirmed a causal relationship of SSc with LVEF, which is similar to what was observed in previous studies. And it is well known that LV systolic dysfunction is a common complication of SSc, for example, kazzam et al. conducted a case-control study of 30 patients and 48 controls and found that LV systolic function was frequently impaired in SSc patients [53]. Growing evidence strongly suggests that systolic dysfunction is associated with myocardial fibrosis in SSc patients, and that some changes related to ischemia and inflammation associated with vasospasm may explain the development of myocardial fibrosis leading to systolic dysfunction [2]. For example, in a follow-up study over 7 years, it was confirmed that SSc patients with severe Cardiac Raynaud had a 40.1-fold higher risk of developing LV systolic dysfunction compared to SSc patients without Cardiac Raynaud [54].

For diastolic dysfunction, the meta-analysis showed that SSc patients are more likely to demonstrate features of DD, such as increased E/E′ ratio, reduced E′, lower E/A ratio, larger LA dimension, and so on, indicating that SSc patients are more likely to develop LVDD as compared to controls. In recent years, there have been some reports about LVDD in SSc patients [38,39], mainly manifests as decreased left ventricular dilatation, impaired diastolic function and abnormal diastolic filling of diastolic function, for example, Meune et al. conducted a prospective study with 100 SSc patients and found that the depression of LV systolic and diastolic function is common in SSc patients [31]. At the same time, some studies have shown that LVDD is associated with the high mortality in SSc patients. For example, a large unselected cohort study showed that DD has a high effect on mortality in SSc [55]. Hinze et al. also confirmed that DD is independently associated with an increased risk of death in SSc patients [56]. Furthermore, PAH is one of the most serious complications of SSc and one of the leading causes of death in this population. And right atrial enlargement is a common manifestation of pulmonary hypertension. Recent studies have shown that right atrial enlargement is associated with elevated pulmonary arterial pressure as a result of both diastolic dysfunction and pulmonary hypertension in SSc patients [57]. For example, the Itinér AIR study have demonstrated that 45% of patients with SSc-associated elevated pulmonary artery pressure is due to LVDD [58]. In a study comparing various data from patients with SScPAH and idiopathic PAH, significantly larger LA dimensions were found in patients with SScPAH, reflecting an increased prevalence of LVDD in this group [59]. Therefore, we should also conduct more research to explore the specific mechanism between right-atrial enlargement and diastolic dysfunction in SSc.

However, in the MR study, we failed to find a causal link between SSc and LVEDV as well as LVM. This is mainly because we were not able to find more and better indicators and parameter date representing LVDD leading to positive results, and the echocardiographic indicators that traditionally indicate LVDD mainly include e', E/e' ratio, E/A ratio, etc., so we could not prove the causal relationship between SSc and LVDD.

Although increasing attention has been paid to LVD in SSc patients, at present, the exact mechanism between SSc and LVD was still unknown. We speculate that it may be related to the following three mechanisms: first of all, it may be related to inflammation. Some studies suggest that inflammatory factors, such as interleukin-6, tumor necrosis factor and C-reactive protein, are all higher among SSc patients in comparison with normal people [41]. Chronic inflammation may accelerate the progression of cardiovascular disease in patients with SSc, and may be a possible underlying cause of LVD in SSc patients. Secondly, histological studies have shown that myocardial necrosis and sporadic fibrosis are not associated with epicardial coronary stenosis [60]. Meanwhile, several trials have assessed myocardial perfusion, providing evidence of reversible ischemia [61]. Therefore, in the early development of SSc disease, cardiac microcirculation dysfunction may be an important cause of heart disease, and it may also contribute to the development of LVD. Finally, myocardial fibrosis is a characteristic change of myocardial involvement in SSc. Myocardial fibrosis generally affects the ventricles bilaterally, resulting in increased ventricular weight, decreased ventricular wall mobility, which affects the diastolic capacity of the ventricular wall during diastole [62].

Our MR analysis has several strengths. Above all, our greatest strength is combining meta-analysis with MR studies. Secondly, we used the CMR parameters to explore the cause relationship. CMR is the most effective method for early diagnosis of cardiac involvement in SSc patients. CMR can reveal mechanisms of diastolic dysfunction, such as subclinical fibrosis, coronary microcirculation disorders, and myocardial inflammation, all of which contribute to the development of diastolic dysfunction in SSc [63]. Furthermore, on the one hand, the large study sample used in the analysis is a major strength of our study, allowing us to conduct comprehensive analysis for LVD and well-powered GWAS to acquire genetic instruments for MR analyses. On the other hand, the consistent causal estimation across four methods (Inverse variance weighted, MR Egger, Maximum likelihood, Weighted median methods) suggests our results are robust.

At last, several limitations of meta-analysis and MR study should also be mentioned. For MR study, First, our study was limited to populations of European ancestry. Although it may reduce the bias caused by population stratification, it is not clear whether the above results can be applied to other populations. Second, we cannot completely exclude the effects of possible gene-environment interactions, other confounding factors such as hypertension, diabetes and other traditional risk factors. Third, although we employ the largest GWAS data to date, further studies still need larger sample sizes to provide a more accurate asses the effect of SSc on LVD. Finally, as with all MR studies, we were unable to consider unobserved pleiotropy and, therefore, may be biased in assessing the relationship between SSc and important LV parameters. For meta-analysis, firstly, we cannot analyze whether cardiac dysfunction and disease activity are related. When collecting data on disease activity in SSc patients, we found that only 5 studies have the evidence of diseases activity [21,25,32,38,49], and a variety of scales were used to assess disease activity in the studies, Therefore, it was not possible to combine evaluation items from different scales for analysis. Additionally, these studies did not provide data on the activity of SSc patients with normal cardiac function or LVDD separately. Secondly, there are two subtypes of SSc: lcSSc and dcSSc, however, this study evaluated all SSc patients and the results may be biased; Thirdly, clinically, LVDD is diagnosed by using a comprehensive evaluation of multiple cardiac Doppler ultrasound indicators, but the original studies included in this study all looked at the abnormality of several important indicators of left ventricular diastolic function to confirm LVDD, which may influence the accuracy of the results. Fourthly, there may have been data of disease duration that were not normally distributed, but raw data could not be obtained after contacting the authors. However, the representation of mean ± standard (SD) has no substantial effect on the results of this study. Finally, the small number of related studies for SSc limits the accuracy of our assessment of LVDD in patients with SSc, which needs a collaborative multi-regional hospital study. The prognosis of patients with SSc needs to be evaluated based on additional more longitudinal studies to estimate changes in the development of DD. At the same time, there is a need for more research to confirm the specific mechanism of LVDD in SSc in the future.

5. Conclusions

Although a robust causal relationship between SSc and important LVDD parameters has not been demonstrated in the MR study. However, SSc patients may be likely to suffer from LVDD according to the results of our meta-analysis. These results suggest that cardiac ultrasound needs to be considered as a routine test in SSc patients. For these patients, left ventricular function should be long-term followed up.

Author contribution statement

Wei Yan: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data; Wrote the paper. Qiang Luo: Analyzed and interpreted the data. Qiong Nie: Performed the experiments. Han Wang: Conceived and designed the experiments; Wrote the paper. Jing Wu: Contributed reagents, materials, analysis tools or data.

Funding statement

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability statement

Data included in article/supp. Material/referenced in article.

Declaration of interest's statement

We declare that we have no financial and personal relationships with other people or organizations that can inappropriately influence our work, there is no professional or other personal interest of any nature or kind in any product, service and/or company that could be construed as influencing the position presented in, or the review of, the manuscript entitled.

Supplementary content related to this article has been published online at [URL].

Appendix A. Supplementary data

The following are the Supplementary data to this article.