Introduction

The acute Stanford type A aortic dissection (AAAD) is a devastating condition with poor prognosis1. Technological advances in cardiac surgery have led to personalized treatment options for each situation2. Therefore, preoperative quick screening high-risk patients is of particular importance to tailor and optimize perioperative strategy. In the last few decades, several risk scoring systems for stratification of patients undergoing cardiac surgery have been proposed. Of those, the European System for Cardiac Operative Risk Evaluation II (EuroSCORE II)3, and the German Registry for Acute Aortic Dissection Type A (GERAADA) score4 are the best known. The GERAADA score, which is specifically designed for predicting 30-day mortality in patients undergoing surgery for AAAD, contains a majority of the risk factors implemented in the EuroSCORE II5. The age, creatinine, and ejection fraction (ACEF) score is a novel and simple risk assessment tool which includes only three variables, initially developed in the prediction of mortality in patients undergoing elective cardiac operations6. And further independent validated in several types of cardiovascular surgery, such as valve surgery and ventricular reconstruction7,8,9. However, data on the ACEF score in patients with AAAD are scarce. Consequently, the present study was conducted to assess the predictive value of the ACEF score in patients with AAAD after total arch replacement.

Methods

Patient population

This was a retrospective, single-center observational study of patients with AAAD underwent total arch replacement in our institution between July 2021 and June 2022. The diagnosis was made according to computed tomography angiography (CTA) and echocardiography. All the patients met the indications for total arch replacement which was described in our previous study10. The exclusion criteria were as follows: (1) less than 18 years; (2) sub-acute and chronic AD (the diagnosis was made according to onset time and imaging11); (3) absence of crucial clinical data (required to calculate the ACEF I or II were missing). Figure 1 provides an overview of the study procedures.

Figure 1
figure 1

The flowchart of patient selection and analysis.

Full size image

The present study was following the guidelines of the Declaration of Helsinki and was approved by the Medical Ethics Committee of Fujian Medical University Union Hospital. This retrospective study had the approval from the ethics committee, with the need for informed consent from patients was waived.

Data collection and calculation

The patients were divided into three groups separated by the 33rd and 67th percentiles (tertiles) of ACEF score. Clinical data were collected from electronic medical records by one researcher and checked by another researcher randomly. Preoperative imaging examinations and laboratory tests were performed at emergency admission before any treatment. All in-hospital survival patients were followed up via outpatient clinic visits or telephone interviews after discharge. The follow-up data, including follow-up time and status, was collected.

As suggested by Ranucci et al., the ACEF score was calculated as follows: age (years) / left ventricular ejection fraction (LVEF) (%) + 1 (if serum creatinine > 2 mg/dl)6.

the ACEF II score was calculated as follows: age (years) / LVEF (%) + 2 (if serum creatinine > 2 mg/dl) + 3 (if emergency surgery) + 0.2 × (hematocrit points below 36%)12.

Endpoints

For the purposes of the present study, the primary endpoint was all-cause mortality. The secondary outcome was severe postoperative complications, defined by The International Aortic Arch Surgery Study Group (IAASSG)13, requiring intervention under regional or general anesthesia or requiring new ICU admission or ongoing ICU management for > 7 d or hospitalization for > 30 d, or causing secondary organ failure14.

Statistical analysis

The statistical analysis was conducted in R software version 4.3.2. Continuous variables were reported as mean ± SD or median and quartile range, and compared using analysis of variance or Mann–Whitney test. Categorical variables were expressed as percentages and compared by chi-square analysis. To ensure balance in baseline characteristic, inverse probability of treatment weighting (IPTW) was utilized. Compared to propensity score matching (PSM), IPTW can minimize the influence of confounding variables while maximizing the amount of available information. The sample size formed after IPTW is based on the weight and the pseudo sample size generated by the actual sample15. Discrimination performance of ACEF, ACEF II score in predicting mortality was evaluated by receiver-operating characteristic (ROC) analysis. Univariate Cox proportional hazards model analysis was conducted to determine the independent predictors of 1-year rates of mortality. The variables whose p < 0.10, except the variables of ACEF score, were included in a multivariate model for further analysis. The best cutoff value of ACEF value was identified according to the Youden index. To determine whether the model improved after rounding of continuous ACEF for categorization, Harrell’s C-index, time dependent net reclassification improvement (NRI) and time dependent integrated discrimination improvement (IDI) were calculated. Furthermore, time-dependent (3, 6, 12 month) ROC analysis was conducted. A value of p < 0.05 was considered significant.

Ethics approval

The present study was following the guidelines of the Declaration of Helsinki and was approved by the Medical Ethics Committee of Fujian Medical University Union Hospital.

Patient consent

This retrospective study had the approval from the ethics committee of Fujian Medical University Union Hospital, with the need for informed consent from patients was waived.

Results

Baseline characteristics

We totally enrolled 227 patients who had undergone total arch replacement. 2 patients were excluded for missing key data. After exclusions, a total of 225 AAAD patients for further analysis. The median age was 51.24 ± 11.43 years, with 131 (58.22%) males and 94 (41.78%) females. Patients were classified into three groups according to the tertiles of ACEF score: Tertiles 1 (ACEF ≤ 0.73, n = 73), Tertiles 2 (0.73 < ACEF ≤ 0.95, n = 78) and Tertiles 3 (ACEF > 0.95, n = 74). Obviously, significant differences were observed among tertiles in age, serum creatinine and LVEF. In addition, patients with a high ACEF score had a higher proportion of hypertension (p = 0.027) and more likely to receive a simpler root procedure (p = 0.003). In order to make all the data comparable, the IPTW was conducted to match the baseline and surgical variables aside from age, serum creatinine, LVEF. After matching, no else significant differences were detected. Table 1 Summarizes the baseline characteristics comparison between groups before and after IPTW (Fig. 2). The standardized mean differences were assessed graphically by using a Love plot. All covariates that had a standardized mean difference of < 0.25 were determined to be balanced.

Table 1 Baseline characteristics and clinical characteristics according to ACEF score groups before matching and after IPTW.
Full size table
Figure 2
figure 2

A Love plot was used to assessed standardized mean differences before adjusting and after adjusting.

Full size image

Perioperative outcomes

In-hospital death occurred in 22 patients (2 in Tertiles 1, 5 in Tertiles 2, 15 in Tertiles 3, respectively) and the mortality rate was 9.8% (2.74% vs. 6.41% vs. 20.27%, p = 0.001). This difference also remained statistically significant (p = 0.024) after IPTW weighting. Before IPTW, A higher incidence of renal failure in patients with a high ACEF score was observed (16.44% vs. 24.36% vs. 41.89%, p = 0.002), however this effect was no longer found after weighting (p = 0.777). There was a statistically significant difference in hepatic insufficiency among groups (p = 0.035) after IPTW, while it was not apparent before IPTW. No significant differences were observed in the rest complications after IPTW. The comparison of perioperative outcomes between the groups before and after matching is summarized in Table 2. In ROC curve analysis, the discriminative capacity of the ACEF score was better than ACEF II score in predicting in-hospital mortality (Fig. 4a). The AUC of ACEF, ACEF II score was 0.723 (95% CI 0.613–0.833; p = 0.006), 0.642 (95% CI 0.512–0.771; p = 0.029), respectively. But there is no significant difference (p = 0.127).

Table 2 Comparisons of perioperative outcomes according to ACEF score groups after before matching and after IPTW.
Full size table

1-year outcomes

All 203 patients discharged completed the 1-year follow-up. 1-year survival was 90.6% (97.18% vs. 93.15% vs. 79.66%, respectively). The results showed that the baseline ACEF score were significantly affected 1-year survival outcomes (log-rank test, p (Tertiles 1:2) = 0.147, p (Tertiles 1:3) < 0.001, p (Tertiles 2:3) = 0.002). Figure 3 shows the survival curves of Tertiles 1, 2, and 3 by Kaplan–Meier methods. The results of univariable and multivariable Cox proportional hazard analyses are presented in Table 3. To maximize the predictive value, the receiver operating characteristic (ROC) analysis was performed and cut-off values of ACEF in 0.94 (sensitivity, 0.706; specificity, 0.717; AUC, 0.758) were obtained (Fig. 4b). In the univariate analysis, the ACEF [hazard ratio (HR) 2.98; 95% CI 1.06–8.36; p = 0.038] and ACEF > 0.94 [HR 4.95; 95% CI 2.36–10.36; p < 0.001] were a significant predictor of 1-year all-cause mortality. However, ACEF II was not an independent risk factor for death. After adjusting for potential risk factors (BMI, cerebral malperfusion, dissection involving supra-aortic vessels), ACEF (adjusted hazard ratio = 1.68; 95% CI 1.34–4.91; p = 0.036) and ACEF > 0.94 (adjusted hazard ratio = 2.26; 95% CI 1.82–6.20; p < 0.001) remained an independent predictor for 1-year mortality.

Figure 3
figure 3

Kaplan–Meier curves of 1-year survival in patients with different ACEF groups.

Full size image
Table 3 Cox regression analyses for 1-year outcomes according to ACEF score groups.
Full size table
Figure 4
figure 4

ROC curves of ACEF score and ACEF IIin predicting in-hospital (a) ROC curves of ACEF score in predicting in-hospital at different time (b) Time-dependent ROC curve of binary ACEF model (c).

Full size image

Clinical application of ACEF score

A prediction model was constructed from baseline factors including BMI, cerebral malperfusion, dissection involving supra-aortic vessels, the addition of continuous ACEF score or binary ACEF score (transformed to a categorical variable based on the cut-off value generated by ROC analysis) improve Harrell’s C-index for 1-year mortality (0.877 and 0.879). However, the time dependent NRI and IDI tended both to be improved in binary ACEF score model (0.435, p < 0.001, and 0.131, p < 0.001), respectively (Table 4). The time-dependent AUC (Fig. 4c) shows that the AUC of binary ACEF score model continues to increase within 1 year postoperatively (3, 6, 12 month: 0.876, 0.892, 0.904).

Table 4 Predictive accuracy of the ACEF score using Harrell’s C-index for Cox hazard model, time dependent NRI and time dependent IDI.
Full size table

Discussion

The current investigation showed that the predictive value of the ACEF score in early surgical outcome in AAAD patients underwent total arch replacement. Patients in a higher ACEF score possibly indicating a poor baseline characteristic, with more comorbidities and prior significant medical history. This lack of consistency made direct comparison difficult between groups. Thus, we applying the IPTW to balance the baseline and surgical variables without reducing the sample size, making the conclusion more persuasive and scientific. The key advantage of ACEF score is its simplicity and rapidity, no specific software is necessary. The ACEF score could be considered as a useful tool to risk stratification in patients with AAAD before operation in daily clinical work.

Acute type A aortic dissection (AAAD) is a life-threatening pathology. The mortality is reportedly 9.4–22%1,16.Despite the advances in surgical techniques and perioperative care, the short-term outcome is still undesirable with in-hospital mortality at 9.8% and 1-year survival at 90.6%. Previous studies suggested that the AAAD surgical outcome was closely related to preoperative state17. However, a comprehensive evaluation within the limited time is admittedly a challenging task. Therefore, it is necessary to find a rapid and easy method to identify high-risk patients. Then, a comprehensive baseline data evaluation of high-risk patients should be taken to developing an optimize treatment strategies. Currently, ACEF score has been considered as a useful risk stratification tool in many cardiovascular disease patients18,19,20. The advanced age, high level of serum creatinine and the worse cardiac function has also been acknowledged as the important independent predictive factor for AAAD patients21,22,23. In our study, by comparing AAAD patients in different ACEF score, although the baseline has been balanced, a higher ACEF score was strongly associated with a higher in-hospital mortality. In all discharged patients, the 1-year survival still affected by a low baseline ACEF score.

The ACEF score, comprising only three key variables, is undoubtedly not a substitute for well-recognized scores such as the EuroSCORE II and the GERAADA score in the assessment model for Stanford type A acute aortic dissection. It cannot accurately calculate surgical mortality. Its significance lies particularly in the quick and effective identification of high-risk patients. This provides a reference for optimizing clinical decision-making in high-risk patients, including further comprehensive assessment, more aggressive use of support devices, simplified surgical procedures, and close postoperative monitoring. In the meantime, ACEF score also has a certain predictive role in the 1-year outcomes of AAAD patients. For patients in high ACEF score, intensive follow-up, and guidance in early postoperative are necessary.

The ACEF II score, a modified version of the ACEF score that includes the variables emergency surgery and preoperative anemia, is already being widely used24. There is no doubt that emergency surgery and preoperative anemia are both risk factors of thoracic aortic surgery. In the current study, however, ACEF II score did not show a good prediction ability for in-hospital mortality and 1-year survival in AAAD patients underwent total arch replacement. One reason for this might be the vast majority of the AAAD patients (88.00%) underwent emergency surgery and the assignment value of emergency surgery in ACEF II score was much high (3 (if emergency surgery)). It may lead to the overestimation of actual mortality rate. Hence, for better prediction of mortality in AAAD patients, the equations of ACEF II scores should be recalibrated.

It is strange that renal failure is no more a significant variable between groups after weighing the baseline and surgical variables. In the current study, concomitant chronic renal disease is the main cause of the preoperative elevated creatinine level. We can only speculate that postoperative acute renal failure is considerably more often caused by acute ischemia.

Limitation

Several limitations of this study should be considered. On top of all of that, the retrospective nature of this study introduces inherent biases, causality cannot be determined. Next, this study was conducted in a single center, the limited sample size and a short follow-up duration may influence the extrapolation to all institutions. Thus, the multicenter studies with a large sample size and long follow-up period are warranted to better replicate our results. Also important is, despite the comprehensive IPTW weighting, residual confounding might remain.

Conclusions

In this study, the ACEF score, was demonstrated to be associated with in-hospital mortality and 1-year survival of AAAD patients after total arch replacement. As a simple and reliable score, ACEF could be considered as a useful tool to risk stratification in patients with AAAD before operation in daily clinical work.