Changes in acoustic cardiographic parameters before and after hemodialysis are associated with overall and cardiovascular mortality in hemodialysis patients

Acoustic cardiography can provide simultaneous electrocardiography and acoustic cardiac data to assess the electronic and mechanical heart functions. The aim of this study was to assess whether changes in acoustic cardiographic parameters (ACPs) before and after hemodialysis (HD) are associated with overall and cardiovascular (CV) mortality in HD patients. A total of 162 HD patients was enrolled and ACPs were measured before and after HD, including left ventricular systolic time (LVST), systolic dysfunction index (SDI), third (S3) and fourth (S4) heart sounds, and electromechanical activation time (EMAT). During a follow-up of 2.9 years, 25 deaths occurred with 16 from CV causes. Multivariate analysis showed that high △SDI (per 1; hazard ratio [HR], 2.178; 95% confidence interval [CI], 1.189–3.990), high △EMAT (per 1%; HR, 2.218; 95% CI 1.382–3.559), and low △LVST (per 1 ms; HR, 0.947; 95% CI 0.912–0.984) were independently associated with increased overall mortality. In addition, high △EMAT (per 1%; HR, 2.141; 95% CI 1.117–4.102), and low △LVST (per 1 ms; HR, 0.777; 95% CI 0.637–0.949) were associated with increased CV mortality. In conclusion, the changes in ACPs before and after HD may be a useful clinical marker and stronger prognostic marker of overall and CV mortality than ACPs before HD.

The adjusted Cox regression survival curves for overall survival in the patients according to the 75th percentile of △LVST (9 ms) are shown in Fig. 2. The patients with △LVST < 9 ms had a worse overall survival than those with △LVST ≥ 9 ms (HR, 0.018; 95% CI 0.001-0.455; p = 0.015).

Discussion
To the best of our knowledge, this is the first study to report associations between changes in ACP before and after HD with overall and CV mortality. Our results showed that high △heart rate, low △PR interval, high △SDI, high EMAT, high △EMAT, low △LVST, and low △LVST were significantly associated with increased overall mortality. In addition, low △PR interval, high △EMAT, low LVST and low △LVST were significantly associated with increased CV mortality.
The first important finding of this study is that high △EMAT before and after HD was associated with increased overall and CV mortality. EMAT is defined as the amount of time required for the left ventricle to generate sufficient force to close the mitral valve, and therefore reflects the velocity of force generation during systole 12 . Previous studies have shown a strong association between EMAT and impaired LV contractility, and that EMAT increases steadily with age in patients with heart failure 13,14 . EMAT has also been shown to be a www.nature.com/scientificreports/ useful prognostic marker of CV outcomes. In a study of hospitalized patients with congestive heart failure, Zhang et al. found that elevated EMAT measured at admission was an independent risk factor for major adverse CV events 15 . In patients with acute heart failure syndrome, Chao et al. reported that EMAT could predict CV outcomes independently of LV ejection fraction, LV diastolic function, and serum N-terminal pro-brain natriuretic peptide 16 . In addition, Sung et al. reported that a high %EMAT or S3 strength may indicate an increased risk of de-compensation and mortality in patients with heart failure, and that intensifying diuretic and/or vasodilator therapy may alleviate potential adverse events 17 . But why did the EMAT increase after HD in our study? The potential mechanism may be insufficient coronary perfusion and cause resulting heart ischemia and impaired heart systolic function. Furthermore, there may be a stenosis lesion on coronary artery, so the coronary perfusion is susceptible to volume depletion (decrease in effective circulating volume) by ultrafiltration. In patients with elevated EMAT after HD, this may contribute to higher cardiovascular risk. In the present study, △EMAT was    www.nature.com/scientificreports/ a stronger predictor of adverse outcomes than EMAT before HD. Therefore, dynamic changes in EMAT before and after HD may reflect impaired LV contractility more accurately than resting APCs before HD. The second important finding of this study is that low △LVST before and after HD was associated with increased overall and CV mortality. In 1970, Garrard et al. reported a significant correlation between LVST and angiographically determined LV ejection fraction and left ventricular end-diastolic volume 18 . Patients with LV systolic dysfunction often have a shorter LVST as more time is required to generate sufficient force to open the aortic valve and keep it open in situations with impaired ventricular contraction 19 . Zuber et al. also showed that patients with an LV ejection fraction < 50% had a higher prevalence of S3, lower mean LVST, and higher mean EMAT and EMAT/LVST. The authors concluded that these parameters can improve detection of LV systolic dysfunction 20 . In addition, in patients with intermediate brain natriuretic peptide levels (100 to 500 pg/mL), Shapiro et al. showed that EMAT/LVST could detect LV dysfunction with a high specificity (95%) and moderate sensitivity (55%) and improve the detection of LV dysfunction 21 . In the present study, we found that low △LVST was associated with increased overall and CV mortality, and that this was a stronger predictor of adverse outcomes than LVST before HD.
Another important finding of this study is that high △SDI before and after HD was correlated with increased overall mortality. In patients with heart failure, previous studies have shown that SDI can discriminate patients with severe LV systolic dysfunction (ejection fraction [EF] ≤ 35%) from those with moderate LV systolic dysfunction (35% < EF < 50%) with an SDI > 5 22 . Dillier et al. reported that both SDI (≥ 5) and S3 (4) score were strong independent predictors of mortality and could identify high-risk patients requiring more aggressive monitoring and therapy 23 . In the present study, we found that high △SDI was associated with higher overall mortality, and that it was a stronger predictor of adverse outcomes than SDI before HD.
In this study, the patients with an increased heart rate after HD had higher overall mortality. The reason for an increase in heart rate after HD is not fully understood, but it may be related to progressive volume depletion by ultrafiltration, sympatho-activation and changes in electrolyte levels 24,25 . The autonomic system controls heart rate and rhythm via a balance between parasympathetic and sympathetic systems 26 . A previous study reported that HD patients are characterized by vagal withdrawal and sympathetic overactivity, and that this was linked to a poor CV prognosis 27 . In addition, Barnas et al. investigated hemodynamic patterns during dialysis hypotension, and demonstrated that tachycardia induced by ultrafiltration is part of the spectrum of normal CV autonomic activation. The authors concluded that tachycardia develops by increasing baroreflex-mediated sympathetic activity 24 . Rogovoy et al. investigated sudden cardiac death in ESRD patients using continuous electrocardiography monitoring, and found that sudden increases in heart rate during or after dialysis preceded non-sustained ventricular tachycardia events 26 . In addition, Severi et al. reported a significant increase (11%) in heart rate following changes in calcium, potassium and pH, without significant changes in autonomic activity indices 25 .
Studies on the association between PR interval and mortality have reported inconsistent results. Magnani et al. conducted a prospective cohort study with 2722 older participants (age 70-79 years), and multivariable-adjusted analyses showed that every SD increase (29 ms) in baseline PR interval was associated with 13% increases in the 10-year risk of both incident atrial fibrillation and heart failure. However, they did not find an association between PR interval and all-cause mortality 28 . Skampardoni et al. published a review of patients with chronic kidney disease or dialysis, and showed varying results of the association between PR interval prolongation and CV outcomes and overall mortality 29 . Moreover, Flueckiger et al. reported an association between prolonged PR interval and increased mortality 30 , while Kestenbaum et al. reported that a longer PR interval could not independently predict incident CV events or mortality 31 . In addition, Silva et al. reported that prolonged PR interval could predict bradyarrhythmias 32 . Regarding the difference between PR intervals before and after HD, Badarau et al. reported that an increased PR interval after HD was associated with lower rates of CV events and mortality 33 . During dialysis, changes in electrolyte levels, fluid shift and autonomic response may influence the PR interval 29 . In the present study, the patients who had low △PR interval before and after HD were associated with increased overall and CV mortality. Therefore, further studies are needed to investigate the significance and prognostic value of decreased PR interval after dialysis.
There are several limitations to this study. First, because relatively few patients died, the statistical power of the study was reduced. Further studies with a larger sample size should be performed to verify our findings. In addition, the quality of acoustic cardiography data is influenced by exogenous and endogenous noises. Finally, the ACPs were measured only before and after HD, and repeated measurements may be necessary.
In conclusion, we demonstrated that high △SDI, high △EMAT and low △LVST were independently associated with increased overall mortality. In addition, high △EMAT and low △LVST were significantly associated with increased CV mortality. Changes in ACP before and after HD (△ACP) may be a useful clinical marker and stronger prognostic marker of overall and CV mortality than ACPs before HD. Acoustic cardiography is a non-invasive cost-effective technique to screen dialysis patients who are at high risk of adverse clinical outcomes.

Materials and methods
Ethics statement. Written informed consent was obtained from each participant in accordance with institutional requirements and the principles of the Declaration of Helsinki. Moreover, the current study was approved by the Institutional Review Board of Kaohsiung Medical University Hospital (KMUHIRB-E(I)-20160071).
Study patients and design. This study was conducted at the dialysis unit of a regional hospital in south- Demographic, medical, and laboratory data collection. Age, sex, smoking status (ever vs. never) and comorbid conditions of the patients were obtained from medical records and interviews. After overnight fasting, blood samples were obtained and biomarkers including triglycerides, hemoglobin, albumin, fasting glucose, uric acid, and total cholesterol, calcium, phosphorus, and potassium were measured. The efficiency of dialysis was determined by Kt/V according to the Daugirdas method 34 . All blood samples were obtained within 1 month of enrollment into the study.
Acoustic cardiography. Acoustic cardiographic examinations were performed in the supine position 30 min before and 30 min after a mid-weekly HD session. Acoustic cardiographic data including heart sounds and related systolic time intervals were analyzed using an AUDICOR system (Inovise Medical). The following ACPs were measured 8,12,22 : 1. EMAT and %EMAT: the time from Q wave onset to the mitral component of the first heart sound (S1). EMAT was defined as the time required for the left ventricle to generate sufficient force to close the mitral valve. 2. LVST: the time from S1 to the second heart sound (S2). 3. The third heart sound (S3): the probability that S3 existed. A single value between 0 and 10 was reported, and the presence of S3 was defined as a value ≥ 5. 4. S4: as with S3, a single value between 0 and 10 was reported, and the presence of S4 was defined as a value ≥ 5. 5. Systolic dysfunction index (SDI): SDI was derived from the nonlinear transformation of [(S3 score ÷ 10) × QRS duration × QR interval × %EMAT] and mapped onto a scale of 0 to 10.
Definition of overall and CV mortality. Cases of CV and overall mortality were verified from medical records by two cardiologists, with disagreements being resolved by a third cardiologist. Patients were followed until death or September 2020. Patients who did not die were censored at the end of follow-up.
Statistical analysis. Data are expressed as percentages, mean ± standard deviation for ACPs, mean ± standard error of mean for △ACP, or median (25th-75th percentile) for triglycerides. The patients were divided into two groups as survivors and non-survivors. Between-group differences in categorical variables were assessed using the chi-square test. Approximately normally distributed continuous variables were analyzed using the independent t-test, and continuous variables with skewed distribution were analyzed using the Mann-Whitney U test. △ACP was defined as the difference between the ACP measurement before and after HD. Multivariable linear regression analysis was used to identify the determinants for △ACP. Multivariable Cox proportional hazard analysis was used to identify associations between an ACP before HD and △ACP and overall and CV mortality. Survival curves for overall and CV mortality were derived using Cox-regression analysis. A p value < 0.05 was considered to be statistically significant. Statistical analyses were performed using SPSS 19.0 for Windows (SPSS Inc. Chicago, USA).