2018 Volume 82 Issue 3 Pages 857-865
Background: The ratio of the early transmitral flow velocity to early diastolic velocity of the mitral annulus (E/e′) is an echocardiographic index of mean left ventricular (LV) filling pressure. We investigated the association between the preoperative E/e′ ratio and postoperative acute kidney injury (AKI) during off-pump coronary artery bypass surgery (OPCAB).
Methods and Results: We reviewed 585 patients who underwent OPCAB and with preserved LV ejection fraction determined by preoperative echocardiography. AKI was determined by the Kidney Disease Improving Global Outcomes (KDIGO) criteria. Multivariable logistic regression analysis was performed. E/e′ was also analyzed as 3 categories (E/e′ <8, 8≤E/e′≤15, and E/e′ >15) and as a continuous variable. A propensity score analysis was used to match the patients with E/e′ >15 and E/e′ ≤15. A preoperative E/e′ >15 was an independent predictor for AKI (odds ratio 3.01, 95% confidence interval 1.40–6.17). E/e′ >15 was also an independent predictor for AKI when E/e′ was analyzed with 3 categories or as a continuous variable. In the matched sample, the incidence of AKI and 1-year mortality was significantly higher in patients with E/e′ >15.
Conclusions: Among patients undergoing OPCAB with preserved LV systolic function, a preoperative E/e′ ratio >15 was an independent predictor of postoperative AKI. Measurement of the preoperative E/e′ ratio may help to assess the risk of postoperative AKI.
Perioperative acute kidney injury (AKI) has been reported to be an important contributor to postoperative morbidity and mortality especially after cardiovascular surgery, including coronary artery bypass graft (CABG).1,2 Off-pump coronary artery bypass surgery (OPCAB) has been associated with less severe and a lower incidence of AKI compared with on-pump CABG.3 Nonetheless, there is still a substantial incidence of AKI after OPCAB and it is caused by transient circulatory arrest, global hypoperfusion, myocardial ischemia-reperfusion injury, and inflammatory response.3
The incidence of diastolic dysfunction in patients who undergo CABG has been reported as high.4,5 Although many studies have reported systolic dysfunction of the heart as a risk factor for postoperative AKI,6,7 the association between preoperative diastolic dysfunction and postoperative AKI has not been investigated. A stiff left ventricle with low compliance would lead to an elevated LV filling pressure (LVFP),8 which is associated with higher morbidity and mortality.4,9 Hemodynamic disturbance during OPCAB has been reported to have a biventricular contribution, and the major cause is believed to be diastolic dysfunction and abnormal filling of both ventricles.10,11 As such, it is possible that an underlying increased LVFP may aggravate the intraoperative hemodynamic disturbance, resulting in a higher rate of postoperative AKI. Although previous studies have reported the effect of diastolic dysfunction on postoperative outcomes, including death or major adverse cardiac events (MACE),4,12 postoperative renal dysfunction was evaluated only as a secondary outcome and the association between increased LVFP in patients with normal LV systolic function and postoperative AKI has not been fully evaluated.
The ratio of the early transmitral blood flow velocity to early diastolic velocity of the mitral annulus (E/e′) has been reported to have good predictive power for mean LV diastolic pressure or mean LVFP and has been used to estimate the diastolic function of the heart.13,14 E/e′ >15 is regarded as a predictor of increased mean LVFP when septal e′ velocities are used. Furthermore, a previous study reported that E/e′ >15 predicts hemodynamic instability during coronary anastomosis in OPCAB, even among patients with preserved LV systolic function.15 Furthermore, an increased E/e′ ratio was found to be associated with deterioration of renal function in patients with essential hypertension16 and in patients undergoing kidney transplantation.17 E/e′ >15 was independently associated with the development of AKI in myocardial infarction patients.18
Therefore, the main hypothesis of the present study was that increased LVFP, measured by a preoperative increased E/e′ (>15), may be associated with postoperative AKI in patients with preserved LV ejection fraction (LVEF). To test our hypothesis, we conducted a retrospective observational study of patients who underwent OPCAB with a preoperative measurement of E/e′.
This study was approved by the Samsung Medical Center Institutional Review Board (2013-12-116), and the requirement for patients’ informed consent was waived, given the retrospective design of the investigation. This observational study was registered at https://clinicaltrials.gov (NCT02081261). The electronic medical records of 877 consecutive adult patients who had undergone CABG at the reporting institution were retrospectively reviewed (Figure 1). Exclusion criteria were: on-pump CABG (n=88), missing preoperative E/e′ (n=67), missing preoperative creatinine value (n=11), preoperative hemodialysis (n=4), decreased LV systolic function with LVEF <50% (n=112) or regional wall motion abnormalities in the LV septal wall (n=10). The remaining 585 patients were analyzed. Details of anesthesia, surgical procedures, and hemodynamic management are described in Supplementary File 1.
Flow diagram outlining the inclusion and exclusion criteria and study design. AKI, acute kidney injury, CABG, coronary artery bypass grafting; LV, left ventricular.
Baseline medical or perioperative variables previously reported as associated with postoperative AKI6,7,19–25 were investigated in this study. They included medical history, medication history,26 preoperative cardiovascular status, anesthesia and surgery-related factors, intraoperative hemodynamic variables and laboratory findings, including postoperative nadir platelet count.
The results of preoperative transthoracic echocardiography conducted within 4 weeks prior to OPCAB and performed by experienced sonographers and reported by academic cardiologists were reviewed. Comprehensive echocardiography, including tissue Doppler imaging, was performed using the Acuson SC2000 system (Siemens Medical Solutions USA Inc., Mountain View, CA, USA) with a 4V1c transducer at a frequency of 1.25–4.5 MHz. The standard 2D echocardiography and Doppler data were obtained in the left lateral decubitus position in all patients. LVEF was calculated using a biplane approach and modified Simpson method from apical imaging planes.27 In the recording of tissue Doppler indexes, including e’ and a’ velocities, attention was given to reducing the angle between the ultrasonic beam and annular motion. Left atrial volume index (LAVI) was measured on 2-D echocardiography, calculated by the area-length method and the volume was adjusted to body surface area. LV mass index (LVMI) was calculated using the method of Devereux et al.28 Values recorded in multiple consecutive cycles were averaged to obtain reliable values. In the apical 4-chamber view, early (E) and late (A) transmitral inflow velocities, and early deceleration time were determined using conventional pulsed-wave spectral Doppler echocardiography. Early (e′) and late (a′) diastolic mitral annular peak velocities were measured by pulsed-wave spectral Doppler tissue imaging from the same view, on the septal side of the mitral annulus. These tissue Doppler variables were measured in 92.4% of the eligible patients. Next, the E/A ratio, as well as the E/e′ ratios, were calculated. A cutoff of E/e′ >15 was used, which is the value that is usually considered to indicate increased LVFP and diastolic dysfunction in patients with normal systolic function when the septal e′ velocities are used to calculate the E/e′ ratio.14 The recently updated 2016 American Society of Echocardiography (ASE) recommendations for the evaluation of LV diastolic function by echocardiography suggest different cutoff values for E/e′; 15 for septal e′ and 13 for lateral e′.14 Because the e′ velocity was routinely measured on septal side in all of the study patients, a binary variable of an E/e′ ratio with a cutoff of 15 was used in our logistic regression analysis. In addition, considering the gray zone approach that uses 2 cutoff values of 8 and 15,29 we also analyzed the E/e′ ratio with 3 categories of E/e′: <8, 8≤E/e′≤15, and E/e′ >15. The E/e′ ratio was also analyzed as a continuous variable as a sensitivity analysis. In patients with atrial fibrillation, the echocardiographic measurements, including the E and e′ velocities, were averaged over >3 consecutive cardiac cycles using matched RR intervals for both velocities. All echocardiographic data were measured and obtained in accordance with the guidelines of the ASE.14,27
Outcome Variable and Associated ComorbiditiesThe development of AKI within the first 7 postoperative days was the primary measured outcome. We defined AKI according to the Kidney Disease Improving Global Outcomes (KDIGO) criteria:30 serum creatinine (sCr) increased ≥1.5-fold or increased by ≥0.3 mg/dL from the baseline within 7 postoperative days.30,31
To evaluate whether a preoperative E/e′ ratio >15 is associated with poor clinical outcomes, the following secondary outcome variables were investigated: need for postoperative continuous venovenous hemodialysis, extubation time, the length of hospital and intensive care unit stays, postoperative new-onset atrial fibrillation and in-hospital and 1-year all-cause mortality rates.
Statistical AnalysisMedian [interquartile range] was used for continuous parameters and number (frequency, %) was used for categorical parameters. There was missing data other than preoperative sCr in less than 2% of records. We substituted the most frequent sex-specific value for missing values of categorical variables and sex-specific median values for missing values of continuous variables. Standardized differences and their 95% confidence intervals (CI) were calculated for comparison of patients’ characteristics, perioperative parameters and secondary outcome variables between those with E/e′ ≤15 and E/e′ >15.
To assess the association between preoperative E/e′ ratio and postoperative AKI with adjustment of confounders, we performed multivariable logistic regression analysis of the entire cohort. In addition, as a secondary analysis, we matched patients with E/e′ ≤15 and E/e′ >15 using a propensity score and compared the incidence of AKI in the matched cohort.
First, logistic regression analyses were used in the entire cohort to evaluate whether a preoperative E/e′ ratio >15 is independently associated with postoperative AKI. Multivariable logistic regression analysis was performed, including all possible risk factors for postoperative AKI, to adjust for as much confounding as possible. We did not perform univariable screening according to a certain P-value criteria before multivariable logistic regression analysis. Continuous variables were not categorized for the logistic regression analysis in order not to lose information. Independent predictors were selected from a list of all potential predictors of univariable analysis by performing backward stepwise variable elimination process with a significant criterion of P<0.20.
Second, a propensity-score analysis was used to match patients with and without AKI, to reduce any potential selection bias from demographic factors and underlying medical conditions. Binary logistic regression analysis, including interaction terms, was used to determine the probability of AKI and non-AKI group assignment, which was used for matching. Patients were matched using a greedy method with 1:1 pairing. A total of 80 patients with E/e′ >15 were matched with those with E/e′ ≤15 using nearest-neighbor matching. The following variables were used as contributors to the propensity score: age, sex, history of diabetes mellitus, hypertension, cerebrovascular accident (CVA), peripheral vascular disease (PVD), dyslipidemia, preoperative hemoglobin, albumin, creatinine clearance, chronic obstructive pulmonary disease and preoperative medication (including angiotensin-converting enzyme inhibitors, angiotensin-receptor blockers, β-blockers, furosemide, aspirin, clopidogrel and statins). We defined the caliper as 0.2 standard deviations of the logit-transformed propensity score. The balance between the 2 groups was tested using standardized difference (>0.20 represents imbalance). The incidence of AKI, as well as of the secondary outcome variables, was compared in the matched cohort.
All statistical analyses were performed using SPSS software version 22.0 (IBM Corp., Armonk, NY, USA). Propensity score-matching was conducted by Propensity Score Matching for SPSS version 3.0.4. G*power (version 3.1.9.2, Universität Düsseldorf, Düsseldorf, Germany) was used for the sample size estimation. A sample size ≥308 patients was calculated under the assumption that the expected odds ratio (OR) of AKI development in patients with E/e′ ratio >15 would be 1.5 and the incidence of AKI in patients with E/e′ ratio ≤15 would be 20% with a type I error of 0.05 and a power of 0.8.32 With our final sample size of 585 and α error of 0.05, the available power was 97% when we assume that OR was 1.5. We had 90% power to detect OR of ≥1.40 with our total sample size of 585 patients.
A total of 585 patients were analyzed after exclusion (Figure 1). Among them, 67 (11.5%) developed AKI as defined by the KDIGO criteria and 17 (2.9%) required hemodialysis within the first 7 postoperative days.
Study patients’ characteristics, perioperative parameters, and echocardiography results are presented in Table 1. Of the 585 patients with preserved systolic function, 81 (13.8%) had an E/e′ ratio >15. These patients were older, more likely to be women, had a higher EuroSCORE, and more frequent PVD than those with an E/e′ ≤15. Patients with an E/e′ >15 had shorter deceleration times, larger E/A ratios, larger LAVI and slower e′ and a′ velocities, lower preoperative albumin and higher C-reactive protein (CRP) levels, and received more packed red blood cell (pRBC) transfusions than patients with an E/e′ ≤15. The propensity score-matched group set comprised 160 patients. There was no unbalanced contributor to the propensity score with a standardized difference >0.20 (Figure 2).
| Characteristic | E/e′ ≤15 | E/e′ >15 | P value |
|---|---|---|---|
| Patient population, n | 504 (86.2) | 81 (13.8) | |
| Demographic data | |||
| Age (years) | 63 [56–70] | 69 [63–73] | <0.001 |
| Female (n) | 100 (19.8) | 31 (38.3) | <0.001 |
| BMI (kg/m2) | 24.4 [22.6–26.2] | 24.1 [22.8–25.8] | 0.995 |
| EuroSCORE | 3 [2–5] | 6 [3–7] | <0.001 |
| Medical history | |||
| Hypertension (n) | 332 (65.9) | 57 (70.4) | 0.426 |
| Diabetes mellitus (n) | 241 (48.2) | 39 (51.3) | 0.613 |
| Cerebrovascular accident (n) | 62 (12.3) | 13 (16.0) | 0.349 |
| Peripheral vascular disease (n) | 20 (4.0) | 9 (11.1) | 0.012 |
| Dyslipidemia (n) | 112 (22.2) | 19 (23.5) | 0.805 |
| COPD (n) | 5 (1.0) | 2 (2.5) | 0.251 |
| Medications | |||
| ACEI or ARB | 166 (32.9) | 27 (33.3) | 0.944 |
| β-blocker | 136 (27.0) | 27 (33.3) | 0.237 |
| Diuretic | 75 (14.9) | 19 (23.5) | 0.051 |
| Aspirin | 313 (62.1) | 54 (66.7) | 0.430 |
| Clopidogrel | 208 (41.3) | 39 (48.1) | 0.245 |
| Statin | 200 (39.7) | 40 (49.4) | 0.099 |
| Preoperative cardiovascular status | |||
| MI within 4 weeks or UA within 8 weeks (n) | 245 (48.6) | 38 (46.9) | 0.777 |
| CAG within 7 days (n) | 253 (50.2) | 42 (51.9) | 0.782 |
| Previous coronary stent insertion (n) | 91 (18.1) | 14 (17.3) | 0.867 |
| 3-vessel disease (n) | 349 (69.2) | 48 (59.3) | 0.074 |
| Regional wall motion abnormality (n) | 172 (34.1) | 29 (35.8) | 0.768 |
| LVEF (%) | 63 [58–67] | 61 [57–67] | 0.222 |
| Deceleration time (ms) | 249 [186–286] | 172 [164–188] | <0.001 |
| E/e′ ratio | 9.8 [8.1–11.8] | 17.1 [16.3–20.2] | <0.001 |
| E/A ratio | 0.72 [0.64–0.99] | 0.84 [0.70–1.08] | 0.006 |
| e′ velocity, septal (m/s) | 0.06 [0.05–0.07] | 0.05 [0.04–0.05] | <0.001 |
| a′ velocity, septal (m/s) | 0.10 [0.09–0.11] | 0.08 [0.07–0.09] | <0.001 |
| LAVI (mL/m2) | 32 [29–35] | 41 [37–44] | <0.001 |
| LVMI (g/m2) | 104 [80–145] | 150 [115–169] | <0.001 |
| Baseline laboratory findings | |||
| Preoperative platelet count (103/μL) | 195 [162–240] | 187 [144–232] | 0.362 |
| Postoperative platelet nadir (103/μL) | 90 [70–115] | 89 [73–114] | 0.892 |
| Serum creatinine (mg/dL) | 0.89 [0.77–1.06] | 0.92 [0.75–1.18] | 0.240 |
| Albumin (mg/dL) | 4.2 [4.0–4.5] | 4.2 [3.9–4.5] | 0.035 |
| C-reactive protein (mg/L) | 0.12 [0.05–0.40] | 0.21 [0.06–0.67] | 0.025 |
| Operative details | |||
| Emergency case (n) | 65 (12.9) | 15 (18.5) | 0.172 |
| Operation time (min) | 264 [228–316] | 266 [228–322] | 0.417 |
| Anesthesia time (min) | 323 [283–375] | 324 [295–390] | 0.523 |
| Grafts per patient (n) | 4 [3–5] | 4 [3–5] | 0.274 |
| Anesthesia details | |||
| Intraoperative crystalloid use (L) | 2,300 [1,800–2,800] | 2,200 [1,600–2,830] | 0.574 |
| Intraoperative colloid use (L) | 1,000 [500–1,000] | 1,000 [500–1,000] | 0.291 |
| pRBC use during surgery (units) | 2 [0–3] | 2 [0–3] | 0.012 |
| Cell-saver transfusion (mL) | 0 [0–250] | 0 [0–230] | 0.521 |
Data are expressed as mean (SD), median [interquartile range] or number (%). a′ velocity, late diastolic mitral annulus velocity; ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin-receptor blocker; BMI, body mass index; CAG, coronary angiography; COPD, chronic obstructive pulmonary disease; E/A ratio, ratio of early to atrial inflow velocity; E/e′ ratio, ratio of early transmitral flow velocity to early diastolic mitral annulus velocity; EuroSCORE, European System for Cardiac Operative Risk Evaluation Score; LAVI, left atrial volume index; LVEF, left ventricular ejection fraction; LVMI, left ventricular mass index; MI, myocardial infarction; pRBC, packed red blood cells; UA, unstable angina.
Line plots of individual differences in means of contributors to the propensity score before and after propensity score matching (Left) and histograms showing the distribution of standardized differences of the contributors to the propensity score before and after propensity score-matching (Right).
Table 2 shows the postoperative outcomes according to preoperative E/e′ in all and the matched cohort. Patients with an E/e′ ratio >15 had longer ICU and hospital stays, and had a higher incidence of 1-year death. During the first 7 postoperative days, AKI occurred in 45 patients (8.9%) with a preoperative E/e′ ≤15 and in 22 patients (27.2%) with a preoperative E/e′ >15 (standardized differences: −0.18, 95% CI −0.28, −0.08). The incidence of AKI was also significantly different between the groups in the matched cohort (n=10, 12.5% for E/e′ ≤15 vs. n=21, 26.3% for E/e′ >15, P=0.028).
| Characteristic | All patients | Matched cohort | ||||
|---|---|---|---|---|---|---|
| E/e′ ≤15 (n=504) |
E/e′ >15 (n=81) |
P value | E/e′ ≤15 (n=80) |
E/e′ >15 (n=80) |
P value | |
| Extubation time (h) | 8 [6–12] | 9 [6–12] | 0.131 | 9 [6–12] | 9 [6–12] | 0.213 |
| ICU stay (h) | 27 [21–46] | 40 [23–71] | 0.020 | 28 [21–44] | 40 [23–40] | 0.001 |
| Hospital stay, postoperative (days) | 6 [5–8] | 7 [6–14] | 0.001 | 6 [5–8] | 7 [5–14] | 0.019 |
| AKI by KDIGO criteria (n) | 45 (8.9) | 22 (27.2) | <0.001 | 10 (12.5) | 21 (26.3) | 0.028 |
| Hemodialysis during postoperative 7 days (n) | 5 (1.0) | 12 (14.8) | <0.001 | 1 (1.3) | 11 (13.8) | 0.005 |
| Atrial fibrillation during postoperative 3 days (n) | 42 (8.3) | 9 (11.1) | 0.398 | 2 (2.5) | 9 (11.3) | 0.056 |
| In-hospital death (n) | 2 (0.4) | 2 (2.5) | 0.091 | 0 | 2 (2.5) | 0.497 |
| 1-year death (n) | 7 (1.4) | 8 (9.9) | <0.001 | 2 (2.5) | 11 (13.8) | 0.017 |
Data are expressed as mean (SD), median [interquartile range] or number (%). The incidences of complications were investigated during postoperative hospital stay except for atrial fibrillation and hemodialysis. AKI, acute kidney injury; ICU, intensive care unit; KDIGO criteria, Kidney Disease Improving Global Outcomes criteria.
Compared with those who did not develop AKI, patients who developed AKI were older, had a higher values for EuroSCORE, preoperative E/e′, sCr, and CRP, and lower a′ velocities and serum albumin concentrations. Patients who developed AKI were also more likely to have a history of hypertension, PVD, diuretic medication, and a greater volume of pRBC transfusion (Table S1).
The results of univariable and multivariable analyses of predictors for AKI within all KDIGO stages are shown in Table 3. Multivariable logistic regression in the entire sample revealed that a preoperative E/e′ >15 was independently associated with postoperative AKI (OR 3.01; 95% CI 1.40–6.17; P=0.006). E/e′ >15 was also an independent predictor for AKI when E/e′ was analyzed with 3 categories. Preoperative E/e′ was also an independent predictor for AKI as a continuous variable (OR 1.09, 95% CI 1.01–1.19; P=0.032).
| Variable | Univariable analysis | Multivariable analysis | ||
|---|---|---|---|---|
| OR (95% CI) | P value | OR (95% CI) | P value | |
| Demographic data | ||||
| Age, years | 1.06 (1.03–1.09) | <0.001 | 1.05 (1.01–1.09) | 0.023 |
| Female | 1.10 (0.61–2.00) | 0.750 | ||
| Medical history | ||||
| Hypertension | 1.72 (0.95–3.10) | 0.072 | 2.51 (1.03–6.14) | 0.043 |
| Diabetes mellitus | 1.18 (0.70–2.00) | 0.529 | ||
| Cerebrovascular accident | 1.78 (0.92–3.44) | 0.088 | 1.87 (0.79–4.42) | 0.155 |
| Peripheral vascular disease | 3.12 (1.33–7.32) | 0.009 | 2.64 (0.81–8.59) | 0.107 |
| MI within 4 weeks or UA within 8 weeks | 0.73 (0.44–1.23) | 0.240 | ||
| CAG within 7 days | 0.83 (0.50–1.38) | 0.471 | ||
| 3-vessel disease | 0.99 (0.58–1.70) | 0.966 | ||
| EuroScore | 1.14 (1.03–1.26) | 0.009 | ||
| Preoperative laboratory findings | ||||
| Hemoglobin, g/dL | 0.86 (0.76–0.97) | 0.017 | 0.89 (0.76–1.03) | 0.124 |
| Creatinine clearance, mL/h | 0.97 (0.96–0.98) | <0.001 | ||
| Serum albumin, g/dL | 0.36 (0.19–0.71) | 0.003 | ||
| E/e′ ratio >15 | 3.80 (2.14–6.78) | <0.001 | 3.01 (1.40–6.17) | 0.006 |
| E/e′ ratio as a continuous variable | 1.10 (1.04–1.16) | 0.001 | 1.09 (1.01–1.19) | 0.032 |
| E/e′ ratio, categorized as 3 groups | ||||
| >15 | 5.70 (2.30–14.13) | <0.001 | 3.90 (1.48–10.20) | 0.008 |
| 8–15 | 1.65 (0.72–3.80) | 0.240 | 1.32 (0.50–3.49) | 0.572 |
| <8 (baseline) | Baseline | Baseline | ||
| LAVI (mL/m2) | 1.02 (0.97–1.06) | 0.439 | ||
| LVMI (g/m2) | 1.01 (1.00–1.01) | 0.042 | ||
| Postoperative nadir platelet count, ×109/L | 0.99 (0.98–0.99) | 0.032 | 0.99 (0.99–1.00) | 0.109 |
| Preoperative medication | ||||
| ACEI or ARB | 1.46 (0.87–2.46) | 0.156 | 1.80 (0.93–3.48) | 0.083 |
| β-blocker | 0.72 (0.39–1.32) | 0.290 | 0.59 (0.27–1.27) | 0.174 |
| Diuretics (furosemide) | 1.81 (0.98–3.33) | 0.057 | ||
| Cephalosporin, 2nd generation | Constant | – | ||
| New inotrope use between preoperative echocardiography and surgery |
0.86 (0.11–6.87) | 0.884 | ||
| Operative details | ||||
| Emergency surgery | 1.44 (0.74–2.83) | 0.286 | 2.48 (0.95–6.46) | 0.064 |
| Operation time, min | 1.00 (0.98–1.00) | 0.856 | ||
| Anesthesia details | ||||
| pRBC transfusion, units | 1.16 (0.99–1.36) | 0.052 | ||
| Hydroxyethyl starch administration, mL | 1.00 (0.80–1.15) | 0.225 | ||
| Intraoperative dopamine infusion | 0.77 (0.42–1.42) | 0.403 | ||
| Intraoperative epinephrine infusion | 1.52 (0.65–3.55) | 0.338 | ||
| Intraoperative norepinephrine infusion | 1.24 (0.64–2.42) | 0.528 | 1.92 (0.81–4.54) | 0.137 |
| Hemodynamic variables during surgery | ||||
| Mean cardiac index, L/min/m2 | 0.56 (0.21–1.53) | 0.259 | ||
| Mean arterial pressure, mmHg | 1.02 (0.97–1.08) | 0.412 | ||
| Mean SvO2, % | 0.93 (0.87–1.00) | 0.051 | ||
Data are expressed as number (%) or median (interquartile range). CI, confidence interval; OR, odds ratio; SvO2, mixed venous oxygen saturation. Other abbreviations as in Tables 1,2.
In this retrospective observational study, we demonstrated that an increased preoperative echocardiographic index of LVFP, measured as E/e′ >15, was associated with the development of postoperative AKI defined by the KDIGO criteria after OPCAB in patients with preserved LV systolic function. A preoperative E/e′ >15 was a strong independent predictor for AKI in our multivariable analysis. In the matched cohort of patients with E/e′ >15 and ≤15, the incidence of AKI and clinical outcomes were significantly poor among those with an increased E/e′. These results implied that E/e′ >15 may be associated with the development of postoperative AKI in patients with preserved LVEF.
LVFP represents a condition of acute LV loading. Previous studies have demonstrated that this increased LVFP is independently associated with cardiovascular morbidity after cardiovascular surgery or with death in cases of severe aortic stenosis.4,5,9,15 LVFP can be estimated by Doppler echocardiography, which is a widely available non-invasive method.14 Specifically, E/e′ >15 is considered to be an excellent echocardiographic index of increased LVFP,13,14,33 and is also an independent predictor of 30-day and 1-year MACE in patients undergoing CABG.4 Diastolic dysfunction has been associated with higher rates of postoperative death and MACE after cardiovascular surgery, regardless of LVEF.34 However, it should be noted that a septal E/e′ >15 can be used to estimate LV diastolic dysfunction regardless of other echocardiographic measurements, including E/A ratio and LAVI, only in patients with preserved LVEF,14 and E/e′ is meaningful only when the E/A ratio is between 1 and 2 in patients with decreased LVEF. A previous study failed to show a correlation between LVFP and E/e′ in patients with decreased systolic function.35 Therefore, only patients with preserved EF were included in the present study. A recent observational study reported an association between preoperative E/e′ and postoperative AKI;29 however, as they analyzed LVEF as well as E/e′, their analysis of E/e′ may have been confounded by systolic dysfunction.
There are possible explanations as to why increased LVFP as estimated by E/e′ >15 could provoke postoperative AKI in patients undergoing OPCAB. First, a baseline increased LVFP may aggravate intraoperative hemodynamic instability,15 which may contribute to the higher incidence of AKI. Significant displacement of the heart is required to expose the grafting site during OPCAB, and both cardiac output and mixed venous oxygen saturation can decrease.10 Both ventricles contribute to the altered hemodynamics, which has been demonstrated by measuring both of the ventricular end-diastolic pressures.11 These hemodynamic derangements were observed more often in patients with septal E/e′ >15 compared with patients with E/e′ ≤15 in our study as well.15 In our study population, patients with an E/e′ ratio >15 showed decreased intraoperative cardiac indexes and mixed venous oxygen saturation compared with those with an E/e′ ≤15 (Table S2). Intraoperative hypotension and subsequent renal ischemia caused by this exaggerated hemodynamic instability may be associated with postoperative AKI. Intraoperative hypotension itself is associated with postoperative AKI.36
Second, preoperative renal dysfunction is associated with preoperative diastolic dysfunction, as shown in previous studies.37,38 In patients with preoperative diastolic and renal dysfunction, fluid overload may develop during and after cardiac surgery where a significant amount of bleeding, transfusion and fluid shift occurs. Fluid overload may precede postoperative AKI,39 and postoperative fluid overload is reported to be a predictor of clinical outcomes in both adult and pediatric cardiac surgery patients.40 However, these associations were reported only in retrospective studies and a prospective study is required to determine whether fluid overload precedes AKI or is only a secondary effect following AKI in patients with diastolic dysfunction.
In addition to E/e′, other echocardiographic parameters that are associated with increased LVFP should be considered.14,41 The LAVI, although known to reflect chronically elevated filling pressures and being related to E/e′, could be associated with postoperative AKI. High LVFP is also associated with increased LVMI, which may also be related to postoperative AKI. These 2 variables were included in our multivariable analysis. Multivariable logistic regression using backward stepwise variable selection identified that preoperative E/e′ was an independent predictor among these 3 variables, by avoiding multicollinearity among the 3 variables. A recent study comparing the correlation between E/e′ ratio, LAVI, and LVFP found that correlation of the LAVI with LVFP (r=0.23) was weaker than that of E/e′ (r=0.65).41 In addition, the measurement of LAVI may be inaccurate in patients with atrial fibrillation (15% incidence in our study sample). An inaccurately measured LAVI would not correlate with LVFP and therefore the association between LAVI and postoperative AKI may be weakened.
Because it was a major demise of our study that E/e′ >15 was a good indicator of increased LVFP, we need to discuss the accuracy of E/e′ >15 to predict increased LVFP. A recent meta-analysis by Sharifov et al42 studied the diagnostic accuracy of E/e′ for evaluating LVFP and diastolic dysfunction with preserved EF. Evidence to support the accuracy of E/e′ to identify or exclude increased LAFP in preserved EF is not sufficient and a further prospective trial is required for validation. For this validation, a recent multicenter study evaluated the accuracy of estimating LVFP by echocardiography.41 In their results, an average E/e′ ratio was modestly correlated with invasively measured LVFP (r=0.65). The accuracy of diagnosis of elevated filling pressure by echocardiographic index was good, with an overall accuracy of 87%, although the accuracy was less reliable with normal EF (84%) compared with 91% with a decreased EF. However, comprehensive judgment with other indexes, including LAVI and peak tricuspid regurgitation velocity, is important for accurate evaluation of LVFP and diastolic dysfunction in patients with preserved EF.14
Although e′ may reflect global LV relaxation, it is nonetheless a regional parameter. Regional wall motion abnormalities at the Doppler sampling site may influence the measurement of the E/e′ ratio.43 Because the present study data included septal e′ measurements, patients with septal wall motion abnormalities were excluded from the analysis. In addition, the preoperative septal E/e′ ratio can be influenced by various confounders, including drug administration and hemodynamic conditions during the perioperative period. We examined whether hemodynamic instability requiring inotropic support developed during the period between the day of preoperative echocardiography and the day of surgery. There was no significant difference in the incidence of inotropic support between the AKI and non-AKI groups and this incidence was not a significant predictor of AKI development in the univariable logistic regression analysis.
We decided to use the KDIGO criteria to diagnose AKI in our study because these criteria are reported to be the most sensitive and more predictive for in-hospital death than the Risk, Injury, Failure, Loss of function and End-stage renal disease (RIFLE) criteria.44 In our study population, the KDIGO criteria were the most sensitive for detecting AKI and was the most predictive of in-hospital death. Although the KDIGO criteria have 3 stages, we used a binary AKI variable because the relatively small numbers of patients with stage 2 and 3 precluded multinomial logistic regression analysis. Many previous studies that reported risk factors for post-cardiac surgery AKI used a binary AKI variable.7,20,22,23 However, stage 1 AKI by KDIGO criteria may be transient and may not be associated with poor clinical outcomes or death,45 although a previous study reported that even small increases in sCr are associated with increased incidence of death.1 Using only the severe stages (e.g., stages 2 or 3) as the outcome variable may select patients with more parenchymal injury,45 but could also reduce sensitivity to detect AKI.
The risk factors for CABG-associated AKI revealed by our study are mostly consistent with those in previous studies, which have reported that decreased LVEF,6,7 poor baseline renal function,7,19 hypoalbuminemia,20,21 and large-volume intraoperative transfusion19 are associated with postoperative AKI.
Preoperative measurements of E/e′ can help physicians plan the perioperative management and apply appropriate preventive measures. First of all, because the hemodynamic instability is frequently encountered in patients with E/e′ >15 and hemodynamic optimization may decrease postoperative renal dysfunction,46,47 efforts to maintain stable hemodynamics during surgery would be required in patients with preoperative E/e′ >15. For this purpose, it may be better to monitor such patients with more advanced hemodynamic variables such as cardiac output or central venous oxygen saturation. Furthermore, patients may be monitored with biomarkers of AKI for early diagnosis of the condition.48 Several protective interventions that are associated with improved clinical outcomes can be performed. Hemodialysis may be initiated early49 or pharmacologic interventions that may protect renal function can be considered.31 The combination of renin-angiotensin system inhibitors, nonsteroidal anti-inflammatory drugs and/or diuretics, which is known to be nephrotoxic, should be avoided if possible.50 As volume overload may develop in patients with renal dysfunction, strict control of fluid administration and avoidance of aggressive fluid intake could potentially mitigate postoperative AKI. E/e′ could be incorporated into the routine preoperative assessment or the inclusion criteria of clinical trials to select high-risk patients.
Study LimitationsFirstly, the relationship between increased preoperative E/e′ and postoperative clinical outcomes including AKI is only an association. Any causal relationship that diastolic dysfunction could lead to postoperative AKI cannot be inferred from our study results. Secondly, as this was a single-center study, the external validity of our results is limited. Thirdly, our study sample included 3 patients with left bundle branch block. The measurement of mitral annulus velocities and thus E/e′ ratio may be inaccurate in patients with left bundle branch block. However, as long as the mitral E and A velocities are not combined, the parameters used to estimate LVFP remain valid.14 Fourthly, a preoperative value of E/e′ may reflect LVFP only at the time of measurement. Consideration of both preoperative and postoperative echocardiographic measurements would provide more reliable information.
In conclusion, we demonstrated a robust and independent association between a preoperative E/e′ >15 and AKI in the first 7 days following OPCAB in patients with preserved LV systolic function. Although there is a study that reported no short-term adverse effects of preoperative diastolic dysfunction in CABG,51 measurement of the preoperative E/e′ ratio may help to assess the risk of postoperative AKI and offer a better opportunity to apply appropriate preventive measures.
The authors thank Dr. Seulggie Choi, Seoul National University for English editing of the manuscript.
This work was not supported by any funding.
Supplementary File 1
Anesthesia, Surgical Procedures, and Hemodynamic Management
Table S1. Patients’ characteristics and perioperative parameters according to AKI
Table S2. Comparison of hemodynamic data between patients with preoperative E/e′ ratio >15 and E/e′ ratio ≤15
Please find supplementary file(s);
http://dx.doi.org/10.1253/circj.CJ-17-0660
