Cardiopulmonary bypass time is an independent risk factor for acute kidney injury in emergent thoracic aortic surgery: a retrospective cohort study

A retrospective cohort study was conducted at the Beijing Anzhen hospital from December 2015 to April 2017 in China, Beijing. This study was approved by the ethics committees of this hospital and conducted following the rules of Good Clinical Practice and principles of the Declaration of Helsinki. Individual consent was waived owing to the retrospective study. All patients who underwent aortic total arch replacement (TAR) combined with a frozen elephant trunk (FET) implant for ADTIAD in this timeframe were included. All operations were performed by the identical surgery team.

Trained staff collected detailed data from recruited patients from the electronic medical records at our medical center. The baseline characteristics collected for each patient involved age, gender, height, weight, body mass index (BMI, calculated based on height and weight recorded by the nurse on the day of hospital admission), drinking history, smoking history; Comorbidities: diabetes mellitus, hypertension, previous cerebrovascular disease, left ventricular ejection fraction (LVEF), coronary artery disease, preoperative hemoglobin, hematocrit, preoperative serum creatinine (sCr), preoperative blood urea nitrogen (BUN); eGFR (estimated glomerular filtration rate, calculated based on Epidemiology Collaboration equation), hemopericardium, renal artery dissection or occlusion, Penn class (Class Aa and Non class Aa), kidney malperfusion, acute myocardial infarction (AMI), preoperative shock; Intraoperative data: intraoperative transfusion of packed red blood cells (PRBCs), CPB time, aortic cross clamp time, circulatory arrest time, rectal temperature at circulatory arrest, nasopharyngeal temperature at circulatory arrest, the type of surgery (Bentall+TAR+FET or ascending aorta replacement+TAR+FET), combined with coronary artery bypass grafting (CABG) and combined with ascending aorta to femoral artery bypass surgery (aortic bypass surgery); Postoperative data: reoperation for bleeding, postoperative dialysis, length of intensive care unit (ICU) stay, length of time in hospital, sepsis, in-hospital death. It should be noted that in our study cohort, there was no patient underwent coronary angiography (CAG) within the 24 h before surgery and no patient received aprotinin or underwent statin therapy. Primary indications for postoperative dialysis were volume overload, uremia, anuria, and significant biochemical abnormalities.

The presence or absence of malperfusion was based on the Penn classification which was established and subsequent validated in the last decade [31]. Preoperative shock was defined as a systolic blood pressure < 90 mmHg [11]. Patients with ST elevation on a 12-lead electrocardiogram associated with wall hypokinesis at the corresponding region on echocardiography were considered to have AMI [33]. Renal malperfusion was diagnosed as at least one renal artery dissection with creatinine rise above 50% of the normal upper limit [11].

The primary endpoint event was AKI after thoracic aortic surgery for ADTIAD. Several classification criterions were established to access the postoperative AKI. Recently, KDIGO put forward a new range of guidelines for the classification of postoperative AKI based on the 2 previous classifications, RIFLE and AKIN [15, 24]. For the purpose of this study, postoperative AKI was diagnosed based on the KDIGO criteria: increase in sCr ≥ 0.3 mg/dL (in 48 h) or 1.5 times or greater by baseline (in 7 days).

Age, gender, diabetes mellitus, hypertension, preoperative hemoglobin, hematocrit, preoperative sCr, preoperative BUN, eGFR, renal artery dissection or occlusion, Penn class, kidney malperfusion, AMI, preoperative shock, CPB time, aortic cross clamp time, circulatory arrest time, nasopharyngeal temperature at circulatory arrest, rectal temperature at circulatory arrest, reoperation for bleeding were recorded in all participants. Preoperative sCr, BUN, hemoglobin, hematocrit was recorded based on the results of the initial laboratory test after admission before surgery.

Patients underwent median sternotomy and CPB. Briefly, the procedure is performed with right axillary artery cannulation for CPB and antegrade cerebral perfusion [5–15 mL/(kg•min)] under moderate hypothermic circulatory arrest (HCA). CPB was performed after systemic heparinization [300 U/kg and maintaining an activated clotting time (ACT) longer than 480 s]. Temperature-adjusted flow rates were 2.5 L/(min·m²) at the time of CPB. The mean arterial pressure was commonly maintained between 50 and 70 mmHg. After CPB was established, cooling was initiated. After clamping of the ascending aorta, cardiac arrest was accomplished with cold cardioplegic solution. Whether to perform an aortic valve replacement depended on the condition of the aortic valve. If the classification of aortic regurgitation was moderate or severe, we preferred to perform Bentall procedure (aortic valve replacement combined with ascending aorta replacement). If there was only mild regurgitation, we preferred to perform ascending aorta replacement only. All patients underwent TAR with FET. The method has been described in detail by our research team [26, 37]. An intraoperative stent-graft (MicroPort Medical Company Limited, Shanghai, China) and a four branched prosthetic graft (Maquet Cardiovascular, Wayne, NJ) were employed in this implantation. In brief, cannulation of the right axillary artery was used for CPB and unilateral selective cerebral perfusion (SCP). The distal aorta was transected circumferentially between the origin of the left common carotid artery and the origin of the left subclavian artery. The stent was implanted into the distal aorta. The distal aorta incorporating the stented elephant trunk was firmly attached to the distal end of the four-branched prosthetic graft using the “open” aortic method. Antegrade systemic perfusion was reestablished through the perfusion limb of the four-branched prosthetic graft. The anastomosis to the left common carotid artery was carried out first. After the anastomosis was completed, CPB was gradually returned to normal flow, SCP was discontinued, and rewarming was started. The anastomosis to the left subclavian artery, the innominate artery, and the proximal anastomosis were completed. If the blood pressure of the upper and lower limbs differed significantly and the signs and symptoms of lower limbs ischemia were presented, the ascending aorta to femoral artery bypass surgery was performed.

After excluding the 22 patients, a total of 115 consecutive patients underwent TAR combined with a FET implant for ADTIAD were included in the final analysis.

Continuous variables were provided as median (quartile) or mean ± standard deviation (SD), in the light of the data dispersion. Categorical variables were expressed as percentages (%). The t-tests were used to compare if the continuous variables followed normal distribution, and if the variables were skewed distribution, non-parametric Mann–Whitney U tests were applied. The chi-square test was applied to compare with categorical variates. Logistic regression analysis was applied to identify the predictors of postoperative AKI. Multiple logistic regression analysis was used to evaluate the association between CPB time and postoperative AKI. We constructed four models: (I) adjusted for none; (II) adjusted for demographics: age; gender; (III) adjusted for age; gender; BMI; smoking history; aortic cross clamp time; nasopharyngeal temperature at circulatory arrest; combined with aortic bypass surgery; AMI; intraoperative transfusion of PRBCs; (IV) adjusted for age; gender; BMI; smoking history; aortic cross clamp time; nasopharyngeal temperature at circulatory arrest; combined with aortic bypass surgery; AMI; intraoperative transfusion of PRBCs; renal artery dissection or occlusion; Penn class; kidney malperfusion; preoperative shock; Bentall+TAR+FET. Conforming to the recommendations of the STROBE statement [39], the results were analyzed from unadjusted or minimally adjusted and fully adjusted in parallel. Whether the covariables was adjusted was determined according to the recommendations of the article published by The NEW ENGLAND JOURNAL of MEDICINE [16]: if, when the variable was added to this model, the matched odds ratio was changed by at least 10% then an adjustment was made. In addition, a generalized additive model (GAM) was also applied to identify linear relationships. The propensity score (PS) matching method was used to adjust intergroup differences between the non-AKI and AKI groups. We calculated the PS for each patient by matching variables (age; gender; BMI; diabetes mellitus; hypertension; smoking history; BUN; preoperative sCr; hemoglobin; hematocrit; eGFR). The balance in baseline covariant was assessed through paired t-tests and McNemar tests and standardized the mean differences as appropriate for categorical and continuous variables.

Interaction and stratified analysis were performed based on age (< 60 and ≥ 60 years), gender, BMI (< 24 kg/m², 24-28 kg/m², ≥28 kg/m²), hypertension, drinking history, smoking history, coronary artery disease, eGFR (< 60 mL/min/1.73m², ≥60 mL/min/1.73m²), aortic cross clamp time (< 115 min, ≥115 min), circulatory arrest time (< 27 min, ≥27 min), hemoglobin (< 135 g/L, ≥135 g/L). All of the analysis was performed with the statistical software packages R (http://​www.​R-project.​org, The R Foundation) and Empower Stats (http://​www.​empowerstats.​com, X&Y Solutions, Inc., Boston, MA). A 2-sided significance level of 0.05 was regarded as statistically significant.

A total of 137 consecutive patients with ADTIAD who underwent emergent aortic TAR combined with an FET implant with CPB were included. All aortic TAR combined with a FET implant procedure, with or without aortic valve operation, were eligible. A total of 18 patients requiring RRT before surgery were excluded for the difficulty to evaluate the progression of renal dysfunction and three patients who died intraoperatively or within 24 h after operation were also excluded because no useful data were available to evaluate the AKI. One patient was excluded for incomplete information. As a consequence, a total of 115 consecutive persons were included in the eventual analysis. A flow chart of the screening and registration of study participants was given in Fig. 1.

After the exclusion criteria were used, 115 consecutive patients were admitted to this cohort. The average age was 47.8 ± 10.7 years. There were 86 (74.8%) male among these patients. The overall incidence of AKI was 53.0% (61 patients). The average CPB time was 211 ± 56 min. The average preoperative sCr was 86.2 ± 29.1umol/L, BUN was 7.2 ± 2.5 mmol/L. The average eGFR was 88.6 ± 22.6 mL/(min•1.73m²), Hemoglobin was 136.5 ± 17.3 g/L. The incidence of sepsis was 14.8%. A total of 23 patients required RRT. In-hospital mortality was 7.8%. The characteristics of the 115 patients at baseline who underwent thoracic aortic surgery for ADTIAD were shown in Table 1.

The consequences of a univariate analyses were given in Table 2. These results revealed that BMI, eGFR and CPB time were all correlated with AKI. We also found smoking, drinking, preoperative hemoglobin levels, history of diabetes mellitus, hypertension, coronary artery disease, preoperative hematocrit, preoperative sCr, BUN, LVEF, renal artery dissection, Penn class, kidney malperfusion, coronary malperfusion, preoperative shock, aortic cross clamp time, circulatory arrest time and Bentall+TAR+FET were not associated with AKI.

Spline smoothing was applied using GAM to explore the association between CPB time and AKI after adjusting for age; gender; BMI; smoking history; aortic cross clamp time; nasopharyngeal temperature at circulatory arrest; combined with aortic bypass surgery; AMI; intraoperative transfusion of PRBCs; renal artery dissection or occlusion; Penn class; kidney malperfusion; preoperative shock; Bentall+TAR+FET. A linear relationship between CPB time and AKI was shown in Fig. 2. The red points express the fitting spline. The black points express the 95% confidence intervals.

Table 3 revealed the results of multivariate logistic regression analysis models for AKI based on different variable categories (preoperative, intraoperative and variables which are known to be related to AKI) included in each analysis. In adjusted model I, the result showed a significant association between CPB time and AKI [odds ratio (OR) =1.085; 95% confidence interval (CI): 1.007–1.170; P = 0.033]. In adjusted model II (adjusted age; gender), the result remained significant (OR = 1.092, 95% CI:1.012–1.179; P = 0.024). In adjusted model III (adjusted age; gender; BMI; smoking history; aortic cross clamp time; nasopharyngeal temperature at circulatory arrest; combined with aortic bypass surgery; AMI; intraoperative transfusion of PRBCs), the result remained significant (OR = 1.166, 95% CI:1.009–1.349; P = 0.031). In adjusted model IV (adjusted for age; gender; BMI; smoking history; aortic cross clamp time; nasopharyngeal temperature at circulatory arrest; combined with aortic bypass surgery; AMI; intraoperative transfusion of PRBCs; renal artery dissection or occlusion; Penn class; kidney malperfusion; preoperative shock; Bentall+TAR+FET.), the result still remained significant (OR = 1.171; 95% CI:1.002–1.368; P = 0.047).

To reduce the influence of confounding variables, we subsequently utilized the derived PS values to match 61 patients in the AKI group with the patients in the non-AKI group at a ratio of 1:1 using a greedy matching algorithm [4]. Finally, 64 patients were successfully matched, 32 patients with AKI and 32 without AKI. After all PS matches were performed, all variables were shown in Additional file 2: Table S1. The PS matching map was shown in Additional file 1: Figure S1. After PS matching (n = 64 pairs), the association between CPB time and AKI was still statistic significant (OR = 1.128, 95% CI:1.004–1.267; P = 0.043). The results were shown in Additional file 3: Table S2.

Stratified analysis was performed in patients with age (< 60 and ≥ 60 years), gender, BMI (< 24 kg/m², 24-28 kg/m², ≥28 kg/m²), hypertension, smoking history, drinking history, coronary artery disease, eGFR (< 60 mL/min/1.73m², ≥60 mL/min/1.73m²), aortic cross clamp time (< 115 min, ≥115 min), circulatory arrest time (< 27 min, ≥27 min), hemoglobin (< 135 g/L, ≥135 g/L). CPB time was still an independent predictor of post-operation AKI in one of these high-risk subgroups. And, there was no interaction with AKI among these groups. Stratified analysis was given in Fig. 3.

Our research also has several limitations which are worth noting. First, it is a retrospective cohort study, and for this purpose, can constitute an association, but not a causality between CPB time and AKI. Second, aortic TAR combined with a FET implant is a preferred choice to treat ADTIAD at our center, while other centers may select more conventional procedures, this may result in a discrepancy in the outcomes among different studies, but our study cohort also forms a homogenous population who had undergone this procedure, which adds internal validity. Finally, nearly all patients were male in our cohort. Caution is required when extrapolating these findings to female patients.