Systolic and diastolic dysfunction affects kidney outcomes in hospitalized patients

The main data source used in this study was obtained from a cohort of patients from Seoul National University Bundang Hospital. A total of 21,574 patients were admitted in this hospital from January 2013 to December 2013. We included patients who underwent an echocardiography examination with tissue Doppler image before hospitalization. Finally, 1327 patients were analyzed in this study.

Demographic and medical data including age, sex and comorbid conditions were obtained from medical records or interviews with patients. The body mass index was calculated as the ratio of weight in kilograms divided by square of height in meters. Data on hemoglobin, serum albumin, fasting glucose, cholesterol level were collected. The baseline value of the estimated glomerular filtration rate (eGFR) was calculated using the Chronic Kidney Disease Epidemiology Collaboration equation [9]. Additionally, information regarding patients’ medications, including aspirin, angiotensin-converting enzyme inhibitors, angiotensin II receptor blockers, alpha/beta-blockers, calcium channel blockers, and diuretics during the study period was obtained from medical records.

AKI was defined as an increase in serum creatinine by ≥0.3 mg/dl or ≥ 1.5 times above baseline during hospital admission, in adherence to the Kidney Disease Improving Global Outcomes guideline [10]. ESRD was defined as commencement of dialysis. All-cause mortality data were obtained from the national database of Statistics Korea. The patients were followed until August 2015, except for the death-censored cases.

Echocardiograms were performed within 1 year before the date of admission using standard techniques to evaluate cardiac function in lateral decubitus position. Left ventricular ejection fraction (LVEF) was calculated using biplane approach and modified Simpson method from apical imaging planes [11]. The left ventricular mass index was calculated using the method of Devereux et al. [12]. In the apical 4-chamber view, early (E) and late (A) transmitral inflow velocities and early (e’) and late (a’) diastolic mitral annular peak velocities were measured by pulsed-wave spectral Doppler tissue images from the same view on the septal side of the mitral annulus.

Statistical analyses ware performed using SPSS version 20.0 (IBM, Armonk, NY, USA). Data are presented as percentages for categorical parameters. Mean (± standard deviations) or median (interquartile ranges) was used for continuous parameters. Odds ratios (ORs) and confidence intervals for AKI according to the echocardiographic parameters were calculated using logistic regression analysis. Survival curves were expressed using Kaplan-Meier method. The Cox proportion hazard model was used to calculate the hazard ratio (HR) of ESRD risk according to the echocardiographic parameters. Comparisons between non-normally distributed continuous variables were performed using the Mann-Whitney U test. Multiple comparisons among the study groups were performed by Kruskal-Wallis test followed by a post hoc test that was adjusted with a less significant difference correction. The discrimination of predicting outcomes by echocardiographic parameters was assessed by calculating the area under the receiver operating characteristic curve. Model performance was additionally assessed using the continuous net reclassification improvement (cNRI) and integrated discrimination improvement (IDI). Both of cNRI and IDI were analyzed by using R software (version 3.4.4; The Comprehensive R Archive Network: http://​cran.​r-project.​org). A difference was considered significant if the value was less than 0.05.

A total of 1327 patients were included in this study and 52.4% were men. The mean age of the patients was 66.2 ± 13.4 years and mean eGFR was 82.5 ± 24.1 ml/min/1.73 m² at baseline. A total of 210 (15.8%) patients developed AKI during admission. The patients’ demographic, medical, and laboratory data, and echocardiographic parameters are shown in Table 1. Multiple variables in baseline characteristics, such as hypertension, diabetes mellitus, and heart failure, were associated with AKI. Among the echocardiographic parameters, LVEF and E/e’ were associated with AKI.

Table 2 shows the ORs of AKI according to the echocardiographic parameters. LVEF and E/e’ were categorized by quartiles of each parameter. Additionally, combined parameter was defined using the median of LVEF and quartile of E/e’. When baseline clinical and laboratory parameters and other echocardiographic parameters were adjusted, the first quartile of LVEF and the fourth quartile group of combined parameters were selected as independent predictors of AKI (all Ps <  0.05). E/e’ showed a tendency for an association with AKI, with a tendency of an increasing OR value. Based on the non-linear relationship curves, a greater E/e’ and smaller LVEF were associated with high risk of AKI (Fig. 1).

Figure 2 shows Kaplan-Meier curves of ESRD based on the echocardiographic parameters. The highest E/e’ group seemed to be at the high risk of ESRD compared with other E/e’ groups, whereas EF was not associated with the risk trend of ESRD. Table 3 shows HRs of ESRD according to the echocardiographic parameters. In univariate analysis, the fourth quartile of E/e’ had a risk of ESRD. When combined parameter was used, the group with low LVEF and high E/e’ showed a tendency of increased risk of ESRD in univariate analysis. However, there were reduced significances between ERSD and echocardiographic parameters in multivariate analysis (Table 3).

Figure 3 shows the risk of ESRD according to the presence of AKI and heart dysfunction. The risk of ESRD was assessed by dividing the groups with and without AKI and by the first to third combined parameter group and fourth combined parameter group. The non-AKI and fourth combined parameter group (HR, 5.95 [0.373–95.107]; P = 0.207), the AKI and first to third combined parameter group (HR, 29.47 [2.499–347.649]; P = 0.007), and the AKI and fourth combined parameter group (HR, 31.31 [2.544–385.238]; P = 0.007) had higher HRs compared with the non-AKI and first to third combined parameter group. These results indicated that heart dysfunction affected the risk of ESRD separately by the presence of AKI.

The areas under the receiver operating characteristic curves (AUCs) predict AKI was 0.548 (0.505–0.591) for LVEF and 0.570 (0.526–0.614) for E/e’. The AUCs of 3-year ESRD were 0.566 (0.426–0.706) and 0.690 (0.544–0.836) for LVEF and E/e’, respectively. We evaluated whether the addition of echocardiographic parameters to risk models increased overall predictability for AKI and 3-year ESRD. The reference risk model was composed of covariates that had been known as a risk factor of cardiovascular diseases in addition to age and sex. Table 4 shows the AUCs for the risk models that predict AKI and ESRD. Compared with the reference model, the AUC of the risk model including LVEF and E/e’ improved predictive capacity of AKI from 0.695 (0.657–0.733) to 0.709 (0.672–0.745) (P = 0.027). When model performance was assessed using the cNRI and IDI indices, the risk models including E/e’ (model 2) or E/e’ and EF together (model 3) had an improved predictability of outcomes compared with the reference model (Table 5). Collectively, the overall predictability of AKI and ESRD was improved when echocardiographic parameters were additionally considered.