Data regarding the surgical outcomes of 345 consecutive patients diagnosed with ATAAD who had undergone TAR combined with FET implantation at Xijing Hospital between September 2019 and September 2023 were analyzed. Computed tomography angiography findings were used to confirm the diagnosis of ATAAD. Seventeen patients with preoperative end-stage renal disease who required dialysis or who were lost to follow-up for the primary endpoints were excluded from the study. The final study cohort comprised 328 patients who were classified into the following cohorts based on AKI severity: non-AKI, mild AKI, and severe AKI.

All patients underwent TAR with FET implantation. This complex procedure involved the use of a four-branch arch Gelweave graft (Vascutek Terumo Inc., Scotland, United Kingdom) for total arch replacement and implantation of a FET device (MicroPort Medical, Shanghai, China) (Supplementary Figure S1) in the descending aorta. Anesthetic induction was commenced using intravenous administration of sufentanil, rocuronium, propofol, and midazolam. Perioperative maintenance of anesthesia was facilitated using a combination of sufentanil, pipecuronium, and midazolam. Cardiopulmonary bypass (CPB) was initiated with cannulation of the brachiocephalic trunk and right atrium, accompanied by left ventricular venting. Systemic perfusion was maintained at a flow rate of 2.2–2.4 L/min/m². Myocardial protection was achieved through antegrade cold del-Nido cardioplegia administration via the coronary orifices following aortotomy. Once the aortic root procedures were completed, moderate hypothermic circulatory arrest ([MHCA], bladder temperature: 25–28°C), and selective antegrade cerebral perfusion (rate, 5–10 mL/kg/min) were routinely applied via the innominate artery and/or the left common carotid artery, and FET and TAR were initiated. Upon completion of the anastomosis, lower body perfusion resumed through the perfusion limb of the four-branch prosthetic graft. Subsequently, the left common carotid artery, proximal aortic stump, left subclavian artery, and iliac artery were sequentially anastomosed to the graft. Rewarming was initiated, and CPB was discontinued. Postoperative analgesia was administered using sufentanil and remifentanil.

The primary outcome was the development of postoperative severe AKI (KDIGO stage 3). Postoperative AKI was defined in accordance with KDIGO guidelines (8). Mild AKI (KDIGO stage 1 or 2) was defined as follows: an increase in the serum creatinine level of ≥26.5 μmol/L within 48 h; or a 1.5-to 2.9-fold elevation from baseline and severe AKI with the level exceeding a 3-fold baseline or ≥ 353.6 μmol/L within 7 days; or the requirement for renal replacement therapy. Myocardial injury was defined as cardiac troponin I (cTnI) of ≥0.1 ng/mL. Hepatic dysfunction was defined as aspartate aminotransferase (AST) or alanine aminotransferase (ALT) levels of >40 IU/L. Renal dysfunction was defined as an endogenous creatinine clearance rate of <85 mL/min. The creatinine clearance rate was estimated using serum creatinine (sCr) values derived from the following Cockcroft–Gault formula:

Paraparesis was defined using the Tarlov score, based on a unilateral or bilateral lower limb muscle strength score of ≤4 combined with iliac artery involvement, as indicated on computed tomography angiography. According to KDIGO guidelines regarding time window restrictions for AKI diagnosis, serum lactate values were collected at nine time points, namely, admission to the operating room, prior to MHCA, at the 5th minute after reinitiation of lower body perfusion, on completion of rewarming, at the end of CPB, at the end of the surgery, and at 4 h, 12 h, and 24 h postoperatively to describe perioperative lactate levels from admission to 24 h postoperatively. Lactate levels at the respective time points were denoted by Lac_adm, Lac_ca, Lac_op, Lacrw, Lac_cpb, Lac_end, Lac_po4h, Lac_po12h, and Lac_po24h. Characteristics denoted with the subscript ‘po12h’ present the levels at the postoperative 12th hour. Clinical data were extracted from hospital databases.

Continuous variables of normal distribution were described by the mean value (±standard deviation), tested by one-way analysis of variance, and compared between cohorts by the least-significant difference method or the Games–Howell method. The continuous variables with non-normal distribution were described by the median (25% quartile; 75% quartile). The Kruskal–Wallis H-test was used for comparisons between multiple cohorts, and the Bonferroni correction was applied. The categorical variables were described by frequency n (%), and the x test and the Bonferroni correction were used to make comparisons among multiple cohorts. To determine the sensitivity and specificity of perioperative lactate levels for predicting severe postoperative AKI, receiver operating characteristic (ROC) curve analysis was utilized to identify the optimal time point to measure lactate levels for severe postoperative AKI. Statistical significance was set at p < 0.05. Variables exhibiting significant differences (p < 0.05) underwent a univariate logistic regression analysis. Subsequently, only variables that maintained statistical significance and passed collinearity diagnostic tests were included in the stepwise multivariate logistic regression analysis. Given the constraints of sample size, multivariate logistic regression was conducted using a stepwise backward elimination method. This approach ensures that only variables with significant associations are included in the model, thereby enhancing the precision and interpretability of the analysis. Subgroup and interaction analyses were performed using univariate and multivariate logistic regression analyses, respectively. Subgroup-specific analysis was stratified according to age, sex, estimated glomerular filtration rate (eGFR), left ventricular ejection fraction (LVEF), and cTnI level to determine the robustness of lactate levels in predicting severe postoperative AKI across diverse demographic and clinical subgroups. All statistical analyses were conducted using SPSS (version 22.0; SPSS Inc., Chicago, IL, United States) software.

The baseline characteristics and surgical details of all the cohorts are presented in Tables 1, 2. Postoperatively, 111 (45.4%) patients developed mild AKI, and 68 (20.7%) developed severe AKI, among whom 48 underwent renal replacement therapy. Moreover, 60% of patients were diagnosed at local hospitals and subsequently transferred to our institution. Influenced by the transfer time, 92 patients (28.0%) underwent surgery <24 h post-onset, 154 (47.0%) underwent surgery between 24 and 48 h post-onset, 36 (11.0%) underwent surgery between 48 and 72 h post-onset, and 46 (14.0%) underwent surgery >72 h post-onset.

Compared with the non-AKI and mild-AKI cohorts, patients in the severe-AKI cohort had a higher preoperative body mass index, white blood cell count, and neutrophil granulocyte (NE), AST, ALT, and cTnI levels, along with higher incidence of hepatic dysfunction (p < 0.05). Furthermore, these patients underwent longer CPB and surgery durations and more complex surgical procedures with coronary artery bypass grafting (p < 0.05). Postoperative outcomes in the severe-AKI cohort were also more complex, with longer ICU stays and higher incidences of cardiovascular, respiratory, and neurological complications and hepatic dysfunction. The corresponding laboratory results at 12 h postoperatively supported this conclusion (p < 0.05), with higher AST, ALT, sCr, cystatin C, and cTnI levels, a lower eGFR, and a longer prothrombin time (PT) in the severe-AKI cohort (p < 0.05). The 30-day mortality rate in the severe-AKI group was significantly higher than that in the other two cohorts (p < 0.05, Table 3).

In all cohorts, lactate levels increased from admission to 24 h postoperatively, with a more significant increase during CPB, peaking at the end of surgery, then decreasing and approaching admission levels at 24 h postoperatively. Lactate levels at each time point were highest in the severe-AKI cohort, followed by the mild-AKI and non-AKI cohorts, with more significant differences at later time points during or after the surgical procedure (p < 0.05). Concerning the decline in postoperative lactate levels, these levels decreased rapidly in the non-AKI and mild-AKI cohorts, whereas the reduction in the severe AKI was less pronounced (Figure 1).

The Mann–Whitney U test results showed significant differences in lactate levels between the severe-AKI and non-AKI cohorts at all time points (p < 0.05). Excluding the time points of admission to the operating room and before MHCA, significant differences in lactate levels were observed between the severe-AKI and non-AKI cohorts (p < 0.05; Figure 1; Supplementary Table E1). The predictive power for severe AKI was well-reflected by lactate levels at all time points, with particularly high predictive efficacy at 12 h postoperatively (Lac_po12h, area under the curve: 0.718; cutoff value: 3.3 mmol/L; sensitivity: 63.2%; specificity: 72.3%; Figure 2).

Considering the temporal sequence of other postoperative systemic complications, their association with AKI, and their influence on clinical outcomes, they were not included in the regression analysis. As postoperative sCr levels serve as a manifestation and diagnostic criterion for AKI, the creatinine level at 12 h postoperatively (sCr_po12h) was not included in the logistic regression analysis despite significant differences (p < 0.05). Additionally, the cystatin C level and the eGFR, which are indicators of renal function, were excluded. Other perioperative indicators that showed statistically significant differences and passed collinearity diagnostics were included in the univariate analysis.

Univariate analysis showed that body mass index, NE, glucose, coronary artery involvement, preoperative hepatic dysfunction, paraparesis, CPB duration, duration of surgery combined with coronary artery bypass grafting, hemoglobin_po12h, AST_po12h, cTnI_po12h, PT_po12h, and Lac_po12h were risk factors for severe AKI in the severe-AKI and non-AKI cohorts. In the severe-AKI and mild-AKI cohorts, except for coronary artery involvement (odds ratio [OR]: 1.604; 95% confidence interval [CI]: 0.857–3.003), preoperative paraparesis (OR: 2.203; 95% CI: 0.919–5.283), and hemoglobin_po12h (OR: 0.989; 95% CI: 0.972–1.005), the aforementioned indicators, preoperative myocardial injury (OR: 1.857, 95% CI: 1.011–3.411), and preoperative eGFR (OR: 0.976, 95% CI: 0.964–0.988) were identified as risk factors for severe AKI (Figure 3).

Owing to the limited sample size, a stepwise backward method was employed to incorporate the independent risk factors identified from the univariate analysis into the multivariate analysis. Lac_po12h, AST_po12h, and operative duration were independent risk factors for severe AKI in all patients after multivariate adjustment. Additionally, a low preoperative eGFR (OR 0.98, 95% CI 0.965–0.995) was identified as an independent risk factor for severe AKI in the severe-AKI + mild-AKI cohort (Figure 4).

In the severe-AKI + non-AKI cohort in relation to predicting severe AKI, Lac_po12h was more effective in patients aged >45 years than in those aged ≤45 years. Lac_po12h was also more effective in patients with an LVEF of ≥55% compared with those with an LVEF of <55%. However, there were no differences in the subgroups according to sex, eGFR, or cTnI levels. In the severe-AKI + mild-AKI cohort in relation to predicting severe AKI, the effect of Lac_po12h was better in subgroups of patients aged >45 years who were male, with an LVEF of ≥55%, and with eGFR of ≥90 mL/min/1.73 m². No significant differences in the cTnI subgroup were observed. In all the subgroups, no interaction was observed between Lac_po12h and severe postoperative AKI (Figure 5).

In the severe-AKI cohort, the incidence of cardiovascular system complications (45.6%), respiratory failure (51.5%), pulmonary infection (48.5%), stroke (13.2%), delirium (32.4%), and hepatic dysfunction (58.8%) was significantly higher (p < 0.05) than those in the other two cohorts. The 30-day postoperative mortality rates in the non-AKI, mild-AKI, and severe-AKI cohorts were 2.7, 2.7, and 48.5%, respectively, indicating significant differences (p < 0.001; Table 3). Lac_po12h levels in all patients who died within 30 days postoperatively were significantly higher than those in patients who survived (p < 0.05). The ROC curve analysis for perioperative lactate levels in relation to postoperative 30-day mortality indicated that lactate levels at four time points from the end of surgery to 24 h postoperatively all had a good predictive effect on postoperative mortality (Lac_end, AUC = 0.639; Lac_po4h, AUC = 0.723; Lac_po12h, AUC = 0.780; Lac_po24h, AUC = 0.811) (Supplementary Figure S2). It could be observed that higher lactate levels at time points further from the end of surgery had a more pronounced predictive effect on postoperative mortality.