Speckle-tracking Echocardiography is More Sensitive in Detecting Subclinical Myocardial Dysfunction in Patients with Rheumatoid Arthritis

Introduction: Patients with rheumatoid arthritis (RA) have shorter life expectancy and their risk of cardiovascular death is more than 50% higher than the rest of the population. Early myocardial dysfunction may be detectable more precisely and sooner using speckle tracking echocardiography. Method: Cross-sectional study enrolled 55 patients with RA (mean age 44.1 years) without known cardiovascular disease and 31 healthy controls, matched for age, sex, blood pressure, BMI and smoking habit. All subjects underwent a standard echocardiographic and Doppler examination (isovolumic contraction and relaxation times (IVCT and IVRT), mitral valve inflow curve (E/A), septal mitral annular motion (e'), and E/e' ratio) as well as the speckle tracking assessment of left ventricle strains and strain rates. Results: In standard echocardiographic examination RA patients exhibited higher indexed left ventricle mass (96.4 ± 20.9 g/m2 vs. 95.8 ± 21.9 g/m2; p=0.013), lower ejection fraction (64 ± 4% vs. 67 ± 4%; p=0.011) and prolonged IVCT (61.5 ± 9.3 ms vs. 53.7 ± 8.95 ms; p=0.001). Diastolic dysfunction was demonstrasted by prolonged IVRT (81.6 ± 9.6 ms vs. 74.6 ± 12.0 ms; p=0.007) as well as by higher E/e’ ratio (8.2 ± 1.8 vs. 7.2 ± 1.5; p=0.009). Speckle tracking method detected decreased global longitudinal epicardial strain (-19.5% vs. -21.5%; p=0.049). Global longitudinal epicardial strain (GLES) correlated with IVCT and IVRT, disease duration and with marker of myocardial damage NTproBNP. RA pts exhibited higher prevalence of markers of myocardial damage (defined as presence NT-proBNP ≥ 125 ng/l or IVRT ≥ 74 ms or IVCT ≥ 57 ms or GLES ≥ -20.0%) 2.2 ± 1.0 vs. 1.3 ± 1.0 (p=0.001), RR 1.97 (95% CI: 1.24–3.15; p=0.004) in comparison with controls. Conclusions: RA patients without known cardiovascular disease exhibited almost two times higher risk for detection of myocardial damage defined as impaired systolic or diastolic function or myocardial contraction deformity parameters or NT-pro-BNP as compared to matched controls. Speckle-tracking echocardiography significantly revealed incipient myocardial dysfunction, which correlates with clinical RA characteristics and other markers of cardiac damage.


Introduction
Diastolic dysfunction is an early marker of myocardial damage in most commonly encountered cardiovascular diseases. Despite the contradictions about its clinical and prognostic value [1], there is a consensus about its increasing importance in non-cardiac diseases such as rheumatoid arthritis (RA) and other collagen diseases due to early occurance [2][3][4]. The early systolic dysfunction is unsatisfactorily studied in RA patients.
Early systolic dysfunction can be detected by assessment of strain (percentage deformation of myocardium normalized to its original shape) and strain rate (how fast the deformation occurs) of global myocardium or predefined myocardial segments in standardized planes. Doppler is an angle-dependent method, therefore not suitable for measuring of strain (deformation) parameters. More informative could be the speckle-tracking echocardiography (STE) [22][23][24][25], which as an angle independent method allows evaluating local myocardial strain and strain rate in predefined segments on the basis of detection and tracking the small natural reflection patterns in ultrasound Bmode picture.
The aim of this study was to detect subclinical pathologies in myocardial function in RA patients using standard and novel echocardiographic methods and to analyze their relationship to clinical characteristics and laboratory risk markers of RA.

Methods
Cross-sectional study comprised patients aged 25-55 years with at least 5 years history of seropositive RA without known cardiovascular disease. The study group was recruited from outpatient's registers at the 5 th and 1 st Departments of Internal Medicine in Bratislava and the Internal Department in Trnava, between 2012-2013.
Exclusion criteria eliminated patients with a history of congestive heart failure, coronary artery disease, atrial fibrillation, stroke, diabetes mellitus, cancer, chronic kidney disease, or venous thromboembolism. The control group was recruited from cohort of 50 employees of participating departments who were matched for age, sex, blood pressure, BMI and smoking habit roughly in ratio 1 control to 2 patients until the both groups had the same characteristics.
Both groups underwent structuralized interview, basic anthropometric measurements and blood sampling for classic risk factors and safety parameters. N-terminal prohormone brain natriuretic peptide (NTproBNP) was analyzed as standard marker of myocardial damage. The Health Assessment Questionnaire (HAQ) was used to determine the mobility and functional capacity and the composite index Disease Activity Score (DAS28) was used to determine rheumatoid arthritis inflammatory activity.
An echocardiographic examination using an Esaote MyLab 60 ultrasonograph was performed by a single investigator in both groups. Left ventricle systolic function was assessed by calculated ejection fraction (EF -Teicholz), Doppler derived isovolumic contraction time (IVCT) and speckle tracking measured systolic strain and strain rate.
Diastolic function was assessed by isovolumic relaxation time (IVRT), pulsed-wave Doppler mitral valve inflow curve (E/A), and tissue Doppler imaging (TDI) was used to determine the septal mitral annular motion (e'/a'). The e' wave value (cut-off ≤ 8 cm/s) and the E/e' ratio (cut-off ≤ 9 cm/s) are considered as the most sensitive parameters for diastolic dysfunction detection [26]. Indexed left ventricular mass (LVM) and aortic root diameters were measured.
Using speckle-tracking echocardiography we estimated longitudinal epicardial, longitudinal endocardial, radial, transversal, circumferential strain (%) and strain rate (1/s). We measured it separately for six segments in every cross section.
Cross-sectioning was done in longitudinal (LAX) and transversal (SAX) planes at the mitral annulus level. In the LAX plane, longitudinal and transversal strain and strain rate were estimated, whereas in the SAX plane, radial and circumferential strain and strain rate were estimated [27][28][29][30][31][32] and then average in each plane.
A cut-off point for global longitudinal strain was set to -20.0% [33] and cut-off points for cardiac time intervals were given according to generally accepted values for IVRT 74 ms and for IVCT 57 ms [34,35].

Ethics
The study protocol was approved by an Ethics committee of University Hospital of Medical Faculty of Comenius University, Bratislava Slovakia and each participant signed an approved Informed consent.

Statistics
Statistical analyses were performed using SPSS 10.0 software (IBM SPSS Inc., Chicago, Illinois, USA).
Normal distribution of parametric data was tested using the oneway Kolmogorov-Smirnov test, followed by Student's t-test or ANOVA techniques for normally distributed data. Heterogeneously distributed data were either analyzed after log-transformation or they were compared using non-parametrical Mann-Whitney U test for two independent groups using Willcox test. Post hoc analysis by Dunnet or Duncan test was performed, when applicable. The impact of clinical characteristics was evaluated with non-parametric tests of qualitative data (χ 2 test with Yates's correction or Fisher's exact test as appropriate). To quantify the correlation between continuous parameters, Pearson's correlation coefficient was calculated, whereby a value of 1 represents an ideal correlation between two methods [36]. Power of the study was calculated for comparison of global longitudinal epicardial strain by method according to Machin et al. [36,37]. Risk ratio statistics and Bayes theorema (sensitivity, specificity, positive and negative predictive values) were calculated according to Altman [38]. Statistical significance was considered at the level of p ≤ 0.05.

Results
The cross-sectional study included 55 patients with confirmed RA without known cardiovascular disease symptomatic for 13 ± 8 years and diagnosed for 11 ± 7 years, control group consisted of 31 healthy persons.  The RA patients were similar to controls in terms of classic risk factors for cardiovascular disease: blood pressure, smoking habit, glycaemia, and total cholesterol and triacylglycerol levels ( Table 1).
No significant difference between groups was found in laboratory parameters, except of explainable higher leucocyte level and lower hemoglobin level in RA patients.
As related to diastolic dysfunction, RA patients shown a higher E/e' ratio (8.2 ± 1.8 vs. 7.2 ± 1.5; p=0.009), but the mean values of e' did not differ between the groups. IVRT was prolonged in RA patients in comparison with controls (81.6 ± 9.6 ms vs. 74.6 ± 12.0 ms; p=0.007).
As related to clinical characteristics and markers, indirect correlations was found between longitudinal epicardial strain rate and the NTproBNP level (r=-0.28; p=0.041).

Discussion
Comparing the standard echocardiographic and Doppler parameters for systolic and diastolic left ventricle function in age, gender, BMI and smoking habits matched RA patients without known cardiovascular disease with controls did not reveal dramatic differences, confirming difficulties in identification of early myocardial abnormality markers in RA patients. Majority of the differences between the groups were within the reference values and out-off reference range parameters were found also in control group. Moreover, none of analyzed parameters were specific for rheumatic arthritis myocardial damage. On other hand with power of our study reaching 0.82 due to homogeneity of both groups (see standard deviations), we were able to discriminate and validate even small differences between RA diseased and controls.
Analysis of four independent myocardial damage parameters (IVRT, IVCT, GLES, NTproBNP) out of generally accepted reference ranges shown more frequent prevalence in RA group (Figure 1). On other side 42% of the controls had two or three out of range parameters; therefore it is possible to attribute the myocardial damage to rheumatic etiology only with some statistical certainty. Considering ≥2 abnormal parameters as cut-off point caused increase of RR for detection of myocardial damage in RA patients two times (RR=1.97). Sensitivity for given association is 76%, with positive predictive value 78% and negative predictive value 56%. Considering ≥3 abnormal parameters do improve the value of RR only negligible (RR=0.01) (data not shown). Bayes analysis confirms that even early stages of the RA are associated with early detectable myocardial damage with acceptable clinical certainty; however the finding is not specific.

Strain and strain rate
The dominant speckle-tracking parameter is considered longitudinal strain and strain rate [29,32,39]. Presented study demonstrated lower trend for the average values of longitudinal strain and strain rate in RA patients than in healthy controls in both global and defined segments of the left ventricle, predominantly affecting apicolateral epicardial segment. Recent study by Midtbo [40] showed that speckle tracking can be sensitive method for finding of subclinical left ventricle dysfunction.
Also study by Fine et al. [23] confirmed decreased global left ventricle longitudinal strain in RA patients and it´s correlations with a reduced HAQ mobility index, and contrary to our patients, lacking correlation with disease duration. Longitudinal strain in Fine's study might have been influenced by more prevalent classic cardiovascular risk factors (hypertension 85%; smoking habit 44%; and dyslipidemia 71%, diabetes mellitus 14%) as well as higher blood pressure in RA patients, which may confuse the results. More reduced HAQ mobility index and lacking correlation with disease duration in Fine et al. study could be explained by higher age (55.7 vs 44.1 years). However, our study with the same risk factors, but strictly balanced, has provided the same results suggesting the causal effect of RA on longitudinal strain.
Strain parameters in our study correlated with NTproBNP and disease duration and similar association was found also by other authors: in a small study of Sitia et al. (22 patients and 20 controls) using analogous inclusion and exclusion criteria as in our study, RA patients exhibited same abnormalities in longitudinal strain correlating with disease duration [22]. Baktir et al. compared 37 RA patients divided in two age-matched groups (45 years) according to disease duration (2.8 vs. 14.6 years) (no healthy control group) and found significantly lower longitudinal strain in patients with advanced disease [24].

NTproBNP
NTproBNP is cardiac neurohormone responding to ventricular overload, hypoxia, hypertension and increased ventricular wall tension or hypertrophy and in RA patients it correlates with systolic and diastolic volumes and left ventricular mass as measured by Doppler echocardiography [41][42][43][44]. In our study with participants balanced for blood pressure and other traditional risk factors, we did not detect a correlation between the NTproBNP level and standard echocardiography parameters for systolic (EF, IVCT) and diastolic function (IVRT, E/e') similarly to previous studies in patients with chronic inflammatory disease [45,46]. Both LV mass or volume overload parameters did not explain differences in NTproBNP between RA patients and controls. Identification of causal relationship between echocardiographical parameters with soluble biomarkers and clinical characteristics could be therefore dependent on modulating factors as age, duration of the disease, functional capacity, therapy, concomitant disease and other difficult-to-balance factors.

Cardiac time intervals
Cardiac time intervals (IVRT a IVCT) are powerful and independent predictors of future ischemic heart disease, heart failure or cardiac death in cardiovascular patients [47]. RA patients in our study had a significantly prolonged IVRT, which may indicate some active relaxation disturbance of the left ventricle in early diastole, as already suggested by Lascano et al. [48] and other authors [49,50] and it may tempt to explain its association with increased NTproBNP, however, the correlation is only borderline (r=0.35; p=0.054). Any of mentioned studies, including our one, did not find correlation between IVRT with other clinical characteristics as disease duration or HAQ.
In patients with diastolic dysfunction is frequently diagnosed arterial hypertension with the prevalence of 60-88% [51,52]. Majority of cited RA studies including our [8,10,11,13] investigated groups with comparable blood pressure and therefore, the slightly impaired diastolic parameters and LV mass in RA patients cannot be explained by an effect of hypertension on diastolic function [53].
Systolic IVCT in RA patients was studied only in several studies -Levendoglu et al. found in RA patients in similar age and duration of disease identical IVCT in controls as our study (54 ms) but substantially shorter duration in RA patients (40 ms) which could be caused by high inflammatory activity or lacking biological therapy in their gropus. Alpaslan et al. in RA patients in age 52 years also found shorter IVCT (38 ms vs. 41 ms) in group also without biological therapy [18,49]. Difference in IVCT duration can be also caused by different method of assessment (flow Doppler vs. tissue Doppler). Until now, there are no established reference values and recommended measurement method of IVCT in the guidelines [54][55][56][57].

Study limitations
Major limitation of presented study could be lacking interobserver comparison of speckle tracking measurement which can cause "virtual dyskinesia" due to intraindividual error [58,59]. When the endocardial borders are manually traced even with experienced specialist, subjective evaluation may overestimate the myocardial thickness and may lead to impaired epicardial strain assessment. The measurement of total and segmental strain could be improved by the measurement of isovolumic times, maximum mitral inflow velocities, and mitral annulus motion, which decrease the chance of false positive strain assessment.
On other side, even when reproducibility of the strain measurements is conditioned by inter-individual variability, studies with one investigator [22,25] as well as studies with more operators [23,24] obtained relatively consistent findings.

Conclusion
Standard echocardiographic assessment of myocardial functions in patients with RA without signs of Cardiovascular disease suggested minimal changes in left ventricular systolic function (by IVCT and LVEF by Teicholz) and diastolic function (by E/e' and IVRT). The speckle-tracking method for myocardial strain analysis showed unambiguously systolic impairment of longitudinal strain parameters, which correlates with biomarkers and clinical characteristics of RA.
Clustering of minor impairments of systolic, diastolic and myocardial contraction deformity parameters together with increased NT-pro-BNP is almost two fold higher in rheumatic arthritis patients serving for identification of increased cardiovascular risk in RA patient and giving impetus for further consideration on cardiovascular therapy.

Conflict of Interest Statement
Neither the first author nor any of the co-authors have any conflict of interest with respect to the presented study.

Funding
This work was supported by the Grant of Comenius University (UK/ 488/2011).