Association of cardiopulmonary bypass with acute kidney injury in patients undergoing coronary artery bypass grafting: a retrospective cohort study.

Abstract

To the Editor: Acute kidney injury (AKI) is a frequent complication of both cardiac and major non-cardiac surgeries. Cardiopulmonary bypass (CPB)-assisted cardiac surgery was associated with a higher incidence of AKI than other surgeries without CPB, suggesting that CPB might be a risk factor for cardiac surgery-associated AKI (CSA-AKI).[1] However, previous studies have demonstrated conflicting results regarding the association of CPB and CSA-AKI, as the diagnosis of AKI has been made according to different criteria.[2-5] The most recent definition of AKI from an AKI Work Group, Kidney Disease: Improving Global Outcomes (KDIGO), allows for the determination of AKI based on a decline in the glomerular filtration rate (GFR) with subsequent increase in serum creatinine or a period of oliguria.[6] We performed a retrospective cohort study to evaluate the association of CPB and postoperative AKI, using the KDIGO definition among patients undergoing on-pump or off-pump coronary artery bypass graft (CABG). From September 29, 2016, to October 07, 2017, patients aged ≥18 years who underwent elective, isolated CABG at Xijing Hospital were included. The exclusion criteria included a history of previous cardiac surgery, hemodynamic instability, estimated GFR of <30 mL·min−1·1.73 m−2 at baseline, or unavailability of a baseline renal function test. Hemodynamic instability here constitutes preoperative cardiogenic shock and use of preoperative intra-aortic balloon pump (IABP), ventricular assist devices, or preoperative cardiopulmonary resuscitation. Our study adheres to the guidelines for Strengthening the Reporting of Observational Studies in Epidemiology. We collected baseline demographic information and data on clinical comorbidities and preoperative medications from all patients from the cardiothoracic surgery database at Xijing Hospital. Data on intraoperative anesthesia management were obtained from anesthesia and medical records of Xijing Hospital. Age, sex, GFR, ejection fraction (EF), European System for Cardiac Operative Risk Evaluation II (EuroSCORE II), and history of diabetes, hypertension, myocardial infarction (MI), stroke, peripheral arterial disease, and chronic obstructive pulmonary disease were determined from the electronic health records. Preoperative use of the following medications was also recorded: angiotensin-converting enzyme inhibitors (ACEIs), angiotensin receptor blockers (ARBs), diuretics, calcium channel blockers, nitrates, and β-blockers, and whether β-blockers were administered in the morning of the day of the surgery. Serum creatinine level and EF at baseline were defined as the latest measurements before surgery. GFR was estimated using the Cockcroft-Gault equation: GFR=[(140 − age) × (weight in kg) × (0.85 if female)]/(72 × serum creatinine). The following intraoperative variables were collected: duration of the operation, number of grafts, adverse events, sufentanyl dosage, and use of IABPs. The primary outcome was the incidence of postoperative AKI. We adopted the definition of AKI by the KDIGO, as a rise in serum creatinine by 0.3 mg/dL in 48 h or a rise in serum creatinine 1.5 times the baseline over 7 postoperative days.[7] As urine output was unavailable after patients were discharged from the intensive care unit (ICU), it was not used to diagnose AKI in this study. Secondary outcomes were in-hospital medical expenses, length of stay, mortality, and composite outcomes including all-cause mortality, MI, stroke, repeat revascularization, or acute renal failure requiring renal replacement therapy. Composite outcomes were determined 1 year after surgery via a follow-up phone call. Definitions of secondary outcomes and telephone follow-ups are described in the Supplementary Materials, https://links.lww.com/CM9/B348. This study was approved by the Institutional Ethics Review Committee of Xijing Hospital (approval No. KY20192156-C-1). It was exempt from the requirement of informed consent due to ethical considerations, and we obtained the oral consent of patients via telephone follow-up. The patients volunteered to tell their long-term outcomes, to which no patient objected. All tests and confidence intervals (CIs) were two-sided, and statistical significance was set at P < 0.05. All analyses were conducted using R (http://www.R-project.org, c) and Free Statistics software version 1.4 (Beijing FreeClinical Medical Technology Co., LTD, China). Please refer to the supplementary material, https://links.lww.com/CM9/B348 for a detailed statistical analysis. Between September 29, 2016, and October 7, 2017, 242 patients underwent isolated CABG. Eight patients were excluded from the study: two had a history of cardiac surgery, one had an estimated GFR of <30 mL·min−1·1.73 m−2 at baseline, two died prior to the postoperative renal function test, two had duplicated electronic medical records, and one was missing serum creatinine data. In total, 234 patients (195 [83.3%] men and 39 [16.7%] women) were included in the final study analysis; 107 (46%) underwent on-pump CABG and 127 (54%) underwent off-pump CABG [Supplementary Figure 1, https://links.lww.com/CM9/B348]. The mean age of the patients was 61.3 ± 7.6 years. The mean preoperative GFR was 67.0 ± 15.8 mL·min−1·1.73 m−2, the median of EuroSCORE II was 1.2 (interquartile range [IQR]: 0.9–1.7), and the mean preoperative EF was 54.4 ± 7.1. Supplementary Table 1, https://links.lww.com/CM9/B348 shows the main demographic characteristics and incidence of preoperative comorbidities grouped by patients who underwent on-pump or off-pump procedures. Compared with patients who underwent on-pump CABG, more patients in the off-pump group were provided with nitrates and β-blockers preoperatively (66.9% vs. 48.6%, P = 0.007 for nitrates; 92.1% vs. 68.2%, P < 0.001 for β-blockers). Patients in the off-pump group tended to have less vessels grafted, shorter operative duration (142.4 ± 38.0 min vs. 268.1 ± 55.5 min, P < 0.001), less intraoperative sufentanyl administered (250.0 [IQR: 225.0–300.0] μg vs. 450 [IQR: 387.5–500.0] μg, P < 0.001), and higher standardized sufentanyl dose (by operative time and expressed as microgram per minute) (1.8 [IQR: 1.4–2.1] μg/min vs. 1.7 [IQR: 1.4–2.0] μg/min, P = 0.046). During the first 7 postoperative days, 117 AKI events occurred in the entire cohort, including 105 stage I AKIs, 7 stage II AKIs, and 5 stage III AKIs; and one patient required renal replacement therapy. The incidence of postoperative AKI was 69.1% in the on-pump CABG cohort and 33.9% in the off-pump CABG cohort (absolute risk difference, 35.2% [95% CI, 23.3–47.3%]) [Supplementary Table 2, https://links.lww.com/CM9/B348]. The association between CPB exposure and postoperative AKI is shown in Table 1. In the unadjusted model, exposure to CPB was associated with an increased risk of postoperative AKI (risk ratio, 2.04 [95% CI, 1.40–2.97], P<0.001). In the adjusted models, after the potential confounders were adjusted and controlled, the direction of association was unchanged. A potential confounder was statistically significant in the univariate analysis of factors affecting AKI [Supplementary Table 3, https://links.lww.com/CM9/B348]. In the fully adjusted model, in which we adjusted for age, sex, GFR, preoperative use of β-blockers, number of grafts, EuroSCORE II, standardized sufentanyl dose, preoperative use of digoxin, and history of stroke, CPB was an independent risk factor for AKI (relative risk [RR], 1.73, [95% CI, 1.14–2.63], P = 0.01) during the first 7 postoperative days. The association with postoperative AKI was statistically significant in all strata of the population investigated according to age, sex, history of stroke, history of diabetes, history of hypertension, history of MI, preoperative use of ACEI or ARB, and preoperative GFR [Supplementary Figure 2, https://links.lww.com/CM9/B348]. The E-values of the RR for the point estimate and lower confidence bound for the incidence of AKI were 2.85 and 1.54, respectively. Table 1 - Association between CPB and incidence of AKI using poisson regression models. Poisson regresssion models Crude Model 1 Model 2 Model 3 Model 4 Groups RR (95% CI) P value RR (95% CI) P value RR (95% CI) P value RR (95% CI) P value RR (95% CI) P value Off-pump 1.00 NA 1.00 NA 1.00 NA 1.00 NA 1.00 NA On-pump 2.04 (1.40, 2.97) <0.001 1.79 (1.18, 2.71) 0.006 1.77 (1.17, 2.68) 0.007 1.75 (1.15, 2.65) 0.008 1.73 (1.14, 2.63) 0.010 Model 1: adjusted for age, sex, GFR, preoperative use of β-blockers, and number of grafts. Model 2: adjusted for age, sex, GFR, preoperative use of β-blockers, number of grafts, and EuroSCORE II. Model 3: adjusted for age, sex, GFR, preoperative use of β-blockers, number of grafts, EuroSCORE II, and standardized sufentanyl dose. Model 4: adjusted for age, sex, GFR, preoperative use of β-blockers, number of grafts, EuroSCORE II, standardized sufentanyl dose, preoperative use of digoxin, and history of stroke. AKI: Acute kidney injury; CI: Confidence interval; CPB: Cardiopulmonary bypass; EuroSCORE II: European System for Cardiac Operative Risk Evaluation II; GFR: Glomerular filtration rate; NA: Not available; RR: Relative risk. The on-pump CABG group was associated with a significant increase in total costs, postoperative costs, length of ICU stay, and postoperative in-hospital stay compared with the off-pump CABG group [Supplementary Table 2, https://links.lww.com/CM9/B348]. AKI was associated with increased costs and length of hospital stay after adjusting for potential confounders [Supplementary Table 4, https://links.lww.com/CM9/B348]. The CORONARY trial reported a reduction in ICU stay as a secondary outcome in patients undergoing off-pump CABG,[3] which is consistent with our findings. Mediation analysis demonstrated that the increased incidence of AKI mediated a 18.17% increment in postoperative costs, as well as a 17.10% increase in total costs and a 25.69% increase in the length of postoperative in-hospital stay after on-pump CABG [Supplementary Figure 3, https://links.lww.com/CM9/B348]. Follow-up at 1 year was completed in 215 patients (91.5%). All-cause mortality was 5.2% and 4.1% in the off-pump and on-pump groups, respectively (P = 0.73). There was no statistically significant difference in the rate of 1-year composite outcomes between the off-pump and on-pump groups (24.1% vs. 28.6%; P = 0.47) [Supplementary Figures 4, https://links.lww.com/CM9/B348 and 5, https://links.lww.com/CM9/B348]. However, the 1-year all-cause mortality (P = 0.06) and composite outcomes (P = 0.0072) were significantly higher in patients who developed postoperative AKI than in those who did not, regardless of whether the surgery was performed with or without CPB assistance [Supplementary Figures 6, https://links.lww.com/CM9/B348 and 7, https://links.lww.com/CM9/B348]. This single-center observational study was limited by the inherent limitations due to its retrospective nature. We did not collect data on preoperative contrast. However, there was no difference in baseline GFR levels, which may rule out the possibility of contrast-induced kidney damage. In addition, the E-values indicated that the observed RR of 1.73 for incident AKI could only be explained by unmeasured confounders, such as contrast exposure, by a risk ratio of >2.85, and beyond that of the confounders that were measured in this study (lower confidence bound, 1.54). According to the KDIGO definition, the risk of AKI after CABG was significantly higher in the on-pump group than in the off-pump group, with a 73% RR increase for AKI in patients exposed to CPB after adjustment for confounding factors. Meanwhile, our subgroup analysis revealed a consistent and robust association between CPB and the risk of AKI across strata. Additional studies are warranted to replicate and confirm these findings using different surgical procedures and in different populations. If an association or causality is established, this may justify a shift toward less invasive surgical procedures. Furthermore, if CPB is deemed necessary, intensive and comprehensive renal protective strategies should be considered when caring for such patients. Funding This work is supported by a grant from the Natural Science Foundation of China (No. 81970448). Conflicts of interest None.