Development and Validation of a Novel Nomogram Risk Prediction Model for In-Hospital Death Following Extended Aortic Arch Repair for Acute Type A Aortic Dissection

Abstract

Background:

Extended aortic arch repair (EAR) is increasingly adopted for treating acute type A aortic dissection (ATAAD). However, existing prediction models may not be suitable for assessing the in-hospital death risk in ATAAD patients undergoing EAR. This study aims to develop a comprehensive risk prediction model for in-hospital death following EAR based on patient’s preoperative status and surgical data, which may contribute to identification of high-risk individuals and improve outcomes following EAR.

Methods:

We reviewed clinical records of consecutive adult ATAAD patients undergoing EAR at our institute between January 2015 and December 2022. Utilizing data from 925 ATAAD patients undergoing EAR, we employed multivariable logistic regression and machine learning techniques, respectively, to develop nomograms for in-hospital mortality. Employed machine learning techniques included simple decision tree, random forest (RF), eXtreme Gradient Boosting (XGBoost), and support vector machine (SVM).

Results:

The nomogram based on SVM outperformed others, achieving a mean area under the receiver operating characteristic (ROC) curve (AUC) of 0.842 on training dataset and a mean AUC of 0.782 on testing dataset, accompanied by a Brier score of 0.058. Key risk factors included cerebral malperfusion, mesenteric malperfusion, preoperative critical station, Marfan syndrome, platelet count, D-dimer, coronary artery bypass grafting, and cardiopulmonary bypass time. A web-based application was developed for clinical use.

Conclusions:

We develop a novel nomogram risk prediction model based on SVM algorithm for in-hospital death following extended aortic arch repair for ATAAD with good discrimination and accuracy.

Clinical Trial Registration:

Registration number ChiCTR2200066414, https://www.chictr.org.cn/showproj.html?proj=187074.

Graphical abstract

Keywords

acute type A aortic dissection / extended aortic arch repair / prediction model / machine learning / nomogram

1. Introduction

Acute type A aortic dissection (ATAAD), defined by an intimal tear and the dissection’s propagation between the media and intima layers of the aorta, remains one of the most challenging and highly morbid conditions encountered by cardiovascular surgeons. Despite advancements in surgical techniques and perioperative care, in-hospital mortality rates after ATAAD surgery remain significant, ranging from 5% to 20% in relatively stable patients, and reaching up to 35% in unstable cases [1, 2]. Extended aortic arch repair (EAR), which includes total arch replacement (TAR) combined with a frozen elephant trunk (FET), has gained popularity for its benefits in promoting favorable aortic remodeling, reducing the risk of reintervention, and enabling future completion of descending aortic procedures [3, 4]. Consequently, this approach is increasingly employed for managing ATAAD, particularly in China, where it has emerged as a preferred strategy [5]. However, EAR presents considerable challenges for cardiac surgeons, with its outcome being heavily dependent on the patient’s preoperative condition and the surgical components [6, 7]. Developing a risk prediction model that integrates preoperative characteristics and surgical data to estimate in-hospital mortality after EAR could aid in identifying high-risk patients, optimizing clinical decisions, and potentially improving surgical outcomes.

The widely used European System for Cardiac Operative Risk Evaluation (EuroSCORE) II [8], a standard tool in cardiovascular surgery, was not specifically designed for ATAAD patients and has been shown to have limited effectiveness in predicting ATAAD surgical outcomes [9]. The German Registry of Acute Aortic Dissection Type A (GERAADA) score, developed to estimate mortality risk in ATAAD surgery, may not be suitable for predicting outcomes in EAR procedures, as it is based on data from only 16% of patients who underwent the TAR procedure [6]. Additionally, the GERAADA score primarily considers preoperative risk factors, without including potentially significant intraoperative variables that could influence postoperative outcomes in ATAAD patients. The model developed by Rampoldi et al. [10], based on data from the International Registry of Acute Aortic Dissection in 2007, may also lack relevance due to advancements in surgical techniques over time. For example, only 11.5% of the patients included in their study underwent the TAR procedure [10]. Similarly, an early mortality prediction model for ATAAD repair developed by Zhang et al. [11] was limited by its small sample size and the potential exclusion of critical variables. These limitations suggest that current prediction models may not be well-suited for evaluating in-hospital death risk in ATAAD patients undergoing EAR. Moreover, the four models mentioned above were developed using logistic regression analysis, which is constrained by its assumption of linear relationships between predictors and outcomes. Machine learning offers an alternative approach, as it excels in identifying complex, non-linear patterns [12, 13]. Studies have demonstrated the potential of machine learning shines in analyzing the enormous data [14]. Thus, predictive models based on machine learning algorithms might be superior to those built using traditional logistic regression methods. However, the complexity of machine learning presents challenges in interpreting models and their outputs. Nomograms, known for their simplicity and utility in clinical practice, offer a promising solution. Incorporating variables identified through machine learning into a nomogram-based risk prediction model may be a promising approach to developing a risk prediction model. Nonetheless, data are scarce in this context.

In this study, we developed and validated prediction models using baseline characteristics and surgical data from consecutive ATAAD adult patients who underwent EAR at our institution between January 2015 and December 2022. The primary objective was to create a practical and accurate mortality risk prediction model by comparing the predictive performance and calibration of models constructed through logistic regression and machine learning techniques. We hypothesized that employing machine learning algorithms might provide superior predictive capability compared to traditional logistic regression methods.

2. Materials and Methods

2.1 Study Protocol and Study Population

This study was a single-center, retrospective analysis approved by the Institutional Review Board of our institution (No. B2022-592), with a waiver of individual consent. The study was registered with the Chinese Clinical Trial Registry (Registration number: ChiCTR2200066414, https://www.chictr.org.cn/showproj.html?proj=187074) and adhered to the Declaration of Helsinki. The research was conducted in compliance with the Strengthening the Reporting of Cohort, Cross-sectional and Case-control Studies in Surgery (STROCSS) criteria and aligned with the Transparent Reporting of a Multivariable Prediction Model for Individual prognosis or Diagnosis (TRIPOD) statement.

We reviewed the records of 925 adult ATAAD patients (aged >18 years) who underwent EAR, with or without additional procedures, at Zhongshan Hospital of Fudan University from January 2015 to December 2022. Exclusion criteria were as follows: (1) patients undergoing no-arch proximal aortic repair, hemiarch, or partial arch replacement; (2) patients undergoing fully endovascular or hybrid procedures, defined as a combination of surgical and endovascular approaches in the same setting; (3) cases of iatrogenic dissection; and (4) patients with dissection during pregnancy. No-arch proximal aortic repair, hemiarch, partial arch replacement, endovascular procedures, and hybrid procedures were excluded because these different surgical approaches could impact patient outcomes, and the study’s aim was to construct a predictive model specifically for EAR procedures. Including patients who underwent alternative surgical methods would not align with the study’s focus. Iatrogenic dissections were excluded due to their distinct onset mechanisms and management strategies, which might lead to varied clinical outcomes. Additionally, patients with dissection during pregnancy were excluded to account for the unique pathophysiological changes and mortality risks associated with pregnancy, such as amniotic fluid embolism.

2.2 Grouping

Eligible patients were categorized into two groups based on the occurrence of in-hospital death: the death group and the survival group. Baseline characteristics and surgical data were compared between these two groups.

2.3 Variable and Data

This study collected a comprehensive dataset comprising 108 clinical features. These included baseline characteristics (such as demographic, comorbidity, comorbidities, medical history, end-organ malperfusion, preoperative critical conditions, dissection characteristics, and laboratory and transthoracic echocardiographic [TTE] data), surgical details, and in-hospital outcomes. All data were retrieved from the hospital’s electronic database and reviewed using a standardized data collection form. Data collection was conducted by two trained staff members who were unaware of the study’s specific objectives to minimize bias. Discrepancies in data interpretation were resolved through consensus with a third reviewer. An independent database monitoring center was engaged to verify the plausibility of the dataset. Laboratory and TTE data were obtained within the first 24 hours before surgery. Only datasets that were validated through independent monitoring were included in the statistical analysis.

The primary outcome was in-hospital mortality, defined as all-cause deaths occurring within 30 days or any in-hospital deaths beyond 30 days for patients who had not been discharged after the index procedure. Previous cardiac surgery was defined as any prior major cardiac operation involving the opening of the pericardium [8]. A critical preoperative state was defined as the occurrence of one or more of the following occurring preoperatively in the same hospital admission as the operation: cardiac massage, preoperative ventilation prior to arrival in the anesthetic room, hypotension or shock, intra-aortic balloon counterpulsation, ventricular-assist device placement prior to arrival in the anesthetic room, or cardiac tamponade [8]. Malperfusion was defined as inadequate blood supply to specific organs due to aortic dissection, confirmed by clinical signs, symptoms, physical examination findings, and laboratory results [15]. Emergency surgery was defined as a procedure performed before the start of the next working day following the decision to operate [8]. Binary variables were encoded as 0 or 1 (0 = no, 1 = yes). Other categorical variables were preprocessed according to their nature. For example, the Neri classification of coronary involvement [16] was encoded as 0, 1, 2, or 3 (0 = no coronary involvement, 1 = Neri A class, 2 = Neri B class, and 3 = Neri C class). Similarly, the degree of aortic valve stenosis or regurgitation was encoded as 0, 1, 2, or 3 (0 = no stenosis/regurgitation, 1 = mild, 2 = moderate, and 3 = severe).

For handling missing data, the MissForest imputation method (R package “missForest”, https://CRAN.R-project.org/package=missForest) was employed to impute variables with less than 10% missing data. Variables with more than 10% missing data were excluded from the analysis. The distribution of variables before and after imputation was shown in Supplementary Fig. 1.

2.4 Development, Validation, and Comparison of Nomogram Prediction Models

For this study, the dataset was randomly divided into two subsets: a training dataset (70%) for model development and a testing dataset (30%) for model validation. Variables for the nomogram risk prediction models were identified using logistic regression analyses and machine learning techniques, including random forest (RF), decision tree (Dtree), eXtreme Gradient Boosting (XGBoost), and support vector machine (SVM). After variable selection, the optimal parameters for each nomogram risk prediction model were modified within the training dataset and subsequently validated using the testing dataset.

Nomogram risk prediction model 1 (Fit.logistic regression analysis (LR)) was constructed using univariate and multivariate binary logistic regression analyses to select variables. Variables with p 0.05 in the between-group comparisons were included in the univariate logistic regression analysis performed on the training dataset. Subsequently, variables with p 0.05 and odds ratios (OR) not equal to 1, as identified in the univariate analysis, were entered into the multivariate logistic regression analysis within the training dataset. Finally, variables with p 0.05 and OR not equal to 1 from the multivariate analysis were incorporated into the Fit.LR model.

Four additional nomogram risk prediction models were constructed using machine learning algorithms for variable selection. Initially, four machine learning prediction models, RF, Dtree, XGBoost, and SVM, were developed on the training dataset. Variables with p 0.05 in between-group comparisons were included in these models. To prevent overfitting, the machine learning models were constructed using the optimal subset of feature variables obtained via feature selection, selected to maximize model accuracy. The variations in prediction accuracy during the feature selection process are illustrated in Supplementary Fig. 2. The SHapley Additive exPlanations (SHAP) method [17] was applied to assess the significance of each variable in the machine learning models. To avoid overfitting in the nomogram risk prediction models, the principle of 10 events per variable (EPV) [18] was followed, as 84 valid events (in-hospital deaths) were observed in this cohort. Therefore, the top eight most significant variables were included in the final models. These variables were used to construct the nomograms, and the SHAP values for each variable in the machine learning models were shown in Supplementary Fig. 3.

The performance of the nomogram risk prediction models was evaluated by using receiver operating characteristic (ROC) curves, with the area under the ROC curve (AUC) calculated for all models. To ensure robust performance evaluation and to reduce the risk of overfitting, the AUC was computed using the bootstrap method, with 1000 resampling iterations. Nomograms were developed for models with good predictive performance to enhance practical applicability. Calibration curves were generated to assess the agreement between predicted and observed outcomes. The Brier score was used to quantify the difference between predicted and actual outcomes. Decision curve analysis (DCA) evaluated the utility of the predictive models in clinical decision-making by using net benefit as an indicator [19]. Additionally, the net reclassification index (NRI) and integrated discrimination improvement (IDI) were calculated to further evaluate the predictive accuracy and discriminatory ability of models with similar predictive performance. Finally, the nomogram prediction model demonstrating the best overall performance was integrated into a web-based survival calculator for ease of use.

2.5 Statistical Analysis

The incidence of in-hospital death in the cohort was estimated to be 10%. Based on a margin of error \le0.05 and following established recommendations for developing clinical prediction models [20], the minimum required sample size was calculated to be 139 patients. To improve model stability and ensure the development of a more representative predictive model, the study included a total of 925 patients over an 8-year period.

The Shapiro-Wilks test was used to assess data normality. Continuous variables with a normal distribution were expressed as the mean \pm standard deviation and compared between groups using the independent-sample t-test. Non-normally distributed continuous variables were reported as the median and interquartile range and compared using the Wilcoxon rank-sum test. Categorical variables were presented as frequencies and percentages and were compared between groups using the Chi-square test or Fisher’s exact test when the expected frequency was 5. To identify variables associated with in-hospital death, univariate and multivariate binary logistic regression analyses were conducted to calculate OR and 95% confidence intervals (CI). Statistical significance was set at p 0.05 (two-sided). All statistical analyses were performed using SPSS version 26.0 (SPSS Inc., Chicago, IL, USA) and R version 4.3.2 (R Project for Statistical Computing, https://www.r-project.org).

2.6 Code Availability

R (version 4.3.2) was used for building and validating the predictive models. The complete code for this study was publicly available without restriction at the following repository: https://github.com/qiyi-chen/Nomogram-for-in-hospital-death-following-EAR.

3. Results

The variable distribution before and after imputation was shown in Supplementary Fig. 1, indicating no significant changes in data distribution following imputation. After random allocation, the distribution of variables in the training and testing datasets was presented in Table 1. As shown in Table 1, a statistically significant difference was observed in the time from symptom onset to surgery between the two datasets. However, this variable was not incorporated into the prediction models, and the overall distribution of variables in the training and testing datasets was considered balanced.

3.1 Study Population and Risk Factors for in-Hospital Death

A total of 1064 adult patients underwent surgical repair of ATAAD at our center over an 8-year period. After excluding 139 patients (Fig. 1), 925 eligible patients were included in the analysis and categorized into the death group (n = 84) or the survival group (n = 841). Among the study population, the average age was 51.9 \pm 12.4 years, and the mean body mass index (BMI) was 25.7 \pm 3.7 kg/m², with 731 (79%) patients being male. Significant differences in baseline characteristics and surgical data were observed between the two groups, as detailed in Supplementary Table 1.

The results of univariate and multivariate logistic regression analyses on the training dataset are presented in Table 2. After conducting both univariate and multivariate logistic regression analyses, the following variables were identified as independent predictors of in-hospital death following EAR for ATAAD: cerebral malperfusion, mesenteric malperfusion, critical preoperative status (CPStatus), D-dimer (D2), international normalized ratio (INR), cardiopulmonary bypass (CPB) time, and coronary artery bypass grafting (CABG).

3.2 Variables of Nomogram Risk Prediction Models

The following variables were included in the Fit.LR model after logistic regression analyses: cerebral malperfusion, mesenteric malperfusion, CPStatus, D2, INR, CPB time, and CABG. The Fit.RF model incorporated eight variables: mesenteric malperfusion, cardiac tamponade, D2, INR, platelet count (Plt), albumin levels, CPB time, and intraoperative red blood cell transfusion. For the Fit-Dtree model, the eight variables included were: mesenteric malperfusion, hypotension, Marfan syndrome (MFS), Plt, cardiac troponin T (cTnT), INR, D2, and CPB time. The Fit.XGBoost model utilized the following variables: mesenteric malperfusion, CPStatus, Plt, D2, aspartate aminotransferase (AST), creatinine (Cr), CPB time, and aortic cross-clamp time (ACC). Lastly, the Fit.SVM model analyzed these variables: cerebral malperfusion, mesenteric malperfusion, CPStatus, D2, Plt, CABG, intraoperative blood product transfusion, and CPB time.

3.3 Performances of Nomogram Risk Prediction Models

The predictive performances of all five models were evaluated using both the training and testing datasets. As shown in Table 3 and Fig. 2, the Fit.LR model demonstrated the highest mean AUC value on the training dataset (0.849, 95% CI 0.786 to 0.908), followed by the Fit.SVM model (0.842, 95% CI 0.780 to 0.910), the Fit.XGBoost model (0.835, 95% CI 0.772 to 0.892), the Fit.Dtree model (0.834, 95% CI 0.772 to 0.890), and the Fit.RF model (0.822, 95% CI 0.757 to 0.884). On the testing dataset, the Fit.SVM model achieved the highest mean AUC value (0.782, 95% CI 0.698 to 0.860), followed by the Fit.RF model (0.769, 95% CI 0.688 to 0.857), the Fit.LR model (0.768, 95% CI 0.668 to 0.859), the Fit.XGBoost model (0.766, 95% CI 0.673 to 0.860), and the Fit.Dtree model (0.740, 95% CI 0.636 to 0.860). Among the models, the Fit.Dtree model had the lowest standard error (0.030), followed by the Fit.LR model (0.031), the Fit.SVM model (0.031), the Fit.XGBoost model (0.031), and the Fit.RF model (0.033). Calibration curves showed that the predicted probabilities for all five models were comparable to the actual observations (Fig. 3). All models demonstrated a good fit based on the Hosmer-Lemeshow test. Additionally, the Fit.SVM model had the lowest Brier score (0.058), followed by the Fit.LR model (0.059), the Fit.XGBoost model (0.060), the Fit.Dtree model (0.063), and the Fit.RF model (0.064), indicating effective probability calibration. The DCA curves for each prediction model are presented in Fig. 4. As shown, all five models outperformed the “treat all” and “treat none” strategies across the risk threshold range of 1.8% to 100%, suggesting considerable clinical utility for all models. Among them, the Fit.SVM model exhibited the largest area under the curve, indicating its superior performance across various decision-making scenarios.

3.4 Comparison of Nomogram Risk Prediction Models

In summary, the Fit.SVM model outperformed other predictive models based on machine learning algorithms. It achieved the highest AUC values on both the training and testing datasets while demonstrating minimal deviation between predicted results and actual outcomes, as reflected in its lowest Brier score. Consequently, the Fit.SVM model was further compared with the Fit.LR model, which was constructed using logistic regression. Both the Fit.LR and Fit.SVM models showed excellent predictive performance. To further compare these models, the NRI and IDI were calculated. Using the Fit.LR model as the baseline and the Fit.SVM model as the comparator, in-hospital mortality 20% was classified as low risk, while mortality \ge20% was classified as high risk. The Fit.SVM model improved prediction accuracy by 14.3% (NRI 0.143, 95% CI 0.030 to 0.257, p = 0.013) and enhanced overall predictive ability by 6.2% (IDI 0.062, 95% CI 0.020 to 0.104, p = 0.004) compared to the Fit.LR model. These results demonstrated that the Fit.SVM model provided superior predictive performance. A nomogram was developed based on the Fit.SVM model to estimate the probability of a composite endpoint event (Fig. 5). The equation for the Fit.SVM model is as follows: Fit.SVM model = –5.040871 + (0.886100 \times cerebral malperfusion) + (2.608367 \times mesenteric malperfusion) + (1.725816 \times CPStatus) + (0.028642 \times D2) + (–0.006118 \times Plt) + (1.381637 \times CABG) + (0.379383 \times intraoperative blood product transfusion) + (0.010404 \times CPB time). Additionally, a web-based survival calculator based on the Fit.SVM model has been created and can be accessed at: https://heartsugery7zs-hospital.shinyapps.io/DynNomapp/.

3.5 Stratified Analyses

Patients were stratified into low-risk and high-risk groups based on the probability of death predicted by the Fit.SVM model, with a threshold of 20%. As shown in Table 4, 810 low-risk patients were identified in the cohort, with 37 in-hospital deaths, corresponding to a mortality rate of 4.6%. Meanwhile, 115 high-risk patients were identified, of whom 47 died during hospitalization, resulting in a mortality rate of 40.8%. The difference in the incidence of in-hospital death between the low-risk and high-risk groups was statistically significant. The distribution of variables included in the Fit.SVM model also differed significantly between the low-risk and high-risk groups. Notably, variables with a substantial impact on mortality, such as cerebral malperfusion (univariate OR 4.71, 95% CI 2.52 to 8.81, p 0.001), mesenteric malperfusion (univariate OR 19.40, 95% CI 6.77 to 55.58, p 0.001), and CPStatus (univariate OR 7.93, 95% CI 4.29 to 14.68, p 0.001), were predominantly observed in the high-risk group.

4. Discussion

In this cohort of 925 adult patients who underwent EAR for ATAAD over an 8-year period, the in-hospital mortality rate was 9.1%, aligning with findings from previous studies [1, 2]. EAR has become a widely adopted surgical approach in China for ATAAD involving the aortic arch and descending thoracic aorta, with an acceptable in-hospital mortality rate [6, 7]. In recent years, EAR has attracted significant attention and is increasingly utilized in clinical practice. Current guidelines also recommend EAR as a surgical strategy for treating ATAAD [15]. EAR, however, has been considered to be the most difficult and challenging among all kinds of surgical procedures for ATAAD with a high risk for mortality [21]. Identifying high-risk patients and improving outcomes following EAR requires the development of a practical and effective risk prediction model for in-hospital death. In this study, we developed five nomogram models based on different methods for selecting predictive variables to estimate the risk of in-hospital mortality after EAR for ATAAD. Among these, the Fit.SVM model, constructed using the SVM machine learning algorithm, demonstrated excellent predictive performance on both the training and testing datasets. It also showed strong discrimination and calibration capabilities.

In this study, the SHAP method [17] was utilized to evaluate the significance of each variable in the machine learning model, aiding in variable selection through machine learning algorithms. SHAP serves as a robust tool that visualizes the predictions of the final model, making it widely recognized for improving the interpretability of machine learning models. By offering a unified framework, SHAP quantifies the individual contributions of each feature to the prediction, whether positive or negative, thereby enhancing the model’s explainability and transparency [22]. Overfitting, a common issue in machine learning, can undermine the predictive accuracy of models [23, 24]. It typically arises when a model becomes overly complex [25], resulting in erroneous conclusions that may lead to inappropriate clinical decisions. To counteract overfitting, strategies such as reducing noise (irrelevant data), feature selection, early stopping, and k-fold cross-validation are often employed [25, 26]. In this study, variables with p 0.05 between the death and survival groups were selected to minimize irrelevant data. Additionally, the optimal subset of features identified through feature selection was used to reduce the risk of overfitting. Despite these precautions, the possibility of overfitting cannot be entirely excluded, and it may have implications for clinical outcomes.

Through univariable and multivariable logistic regression analyses, cerebral malperfusion, mesenteric malperfusion, CPStatus, D2, INR, CPB time, and the need for CABG were identified as independent risk factors for in-hospital death following EAR. Using the SVM algorithm, eight variables from the 108 variables analyzed were selected and incorporated into the Fit.SVM model: cerebral malperfusion, mesenteric malperfusion, CPStatus, D2, Plt, CABG, intraoperative blood product transfusion, and CPB time. The strong predictive performance of the Fit.SVM model, demonstrated by the ROC curve, calibration curve, and DCA, suggests that the selected combination of variables is well-suited for forecasting outcomes in ATAAD patients. SVM, a machine learning method with exceptional classification and generalization capabilities [27], proved to be an effective tool in this context. Additionally, machine learning algorithms not only identified variables significantly associated with mortality in univariate analysis but also uncovered variables that lacked statistical significance in univariate logistic regression. This highlights the advantages of machine learning over logistic regression, particularly in capturing non-linear relationships between variables and outcomes [12, 13]. Compared to the Fit.LR model, which was based on logistic regression analysis, the Fit.SVM model demonstrated superior prediction accuracy and overall predictive capability. These findings highlight the potential of machine learning algorithms in selecting variables and constructing nomogram-based risk prediction models.

Factors influencing in-hospital outcomes of ATAAD repair have been extensively documented in the literature [28, 29]. Previous studies have indicated that ATAAD patients presenting with cardiac tamponade, shock, congestive heart failure, cerebrovascular accident, stroke, coma, cerebral malperfusion, coronary malperfusion, or mesenteric malperfusion are classified as unstable and have an in-hospital mortality rate of 35%, compared to stable patients [1, 2]. In this study, cerebral malperfusion, mesenteric malperfusion, CPStatus, D2, INR, CPB time, and the need for CABG were identified as independent risk factors for in-hospital death following EAR through univariable and multivariable logistic regression analyses. Patients in the in-hospital death group exhibited significantly higher rates of cerebral malperfusion, mesenteric malperfusion, CPStatus, and the need for CABG compared to those in the survival group. These findings are consistent with previous research [2, 28, 29, 30, 31, 32, 33]. Malperfusion of critical organs such as the brain and heart, if not promptly relieved from ischemia, often leads to irreversible damage and postoperative complications, contributing to increased mortality rates [31, 32]. The mechanism by which organ malperfusion increases mortality is linked to the harmful cascade of inflammatory responses triggered by ischemia-reperfusion injury, resulting in metabolic acidosis and organ dysfunction [34]. In particular, the diagnosis, management, and decision-making for mesenteric malperfusion remain complex [35]. Patients with ATAAD and mesenteric malperfusion often succumb due to delays in diagnosis. The need for concurrent CABG typically indicates significant coronary malperfusion or hemodynamic instability following cardiac resuscitation [36]. These conditions compromise cardiac function, adversely affecting postoperative survival rates. Numerous studies corroborate our findings, consistently identifying cerebral malperfusion, mesenteric malperfusion, CPStatus, and the need for CABG as predictive factors for in-hospital mortality in patients with ATAAD [10, 37, 38, 39, 40].

In this study, patients in the in-hospital death group exhibited significantly higher levels of D2 and INR, as well as prolonged CPB times, compared to those in the survival group. These findings are consistent with previous research [30, 41, 42, 43, 44]. Elevated D2 levels in acute aortic dissection are strongly associated with activation of the coagulation system within the false lumen [45], reflecting a state of hypercoagulability and secondary hyperfibrinolysis [46]. Prior studies [43, 44] have demonstrated that plasma D2 concentrations correlate with factors such as vessel involvement length, dissection size, and injury characteristics. Patients with elevated D2 levels are more likely to experience organ ischemia and more extensive dissections [47]. Increased D2 levels have also been linked to reduced Plt, higher transfusion requirements during surgery, prolonged operative times, and a greater likelihood of in-hospital mortality. The prognostic significance of preoperative D2 elevation in ATAAD has been widely reported [43, 45, 48]. Preoperative elevation in INR indicates severe coagulopathy, which exacerbates bleeding tendencies. Emergency aortic repair for ATAAD presents a particularly high risk of bleeding due to prolonged CPB time, the induction of moderate to severe hypothermia, and the fragility of the dissected aorta [42]. Patients with elevated preoperative INR face a markedly increased risk of perioperative bleeding, which can lead to in-hospital mortality. Prolonged CPB time has been identified as an independent risk factor for poor outcomes following EAR for ATAAD, consistent with the findings of Macrina et al. [30] and Zhang et al. [41]. The complexity of EAR often necessitates extended CPB durations, which cannot fully replicate the body’s physiological blood supply. Prolonged CPB activates inflammatory responses, disrupts coagulation mechanisms, and causes significant damage to critical organs [48]. It may result in pulmonary dysfunction, systemic inflammatory responses, cytotoxin production, embolism, and reperfusion injury [49], all of which contribute to higher in-hospital mortality rates.

The combination of cerebral malperfusion, mesenteric malperfusion, CPStatus, D2, Plt, CABG, intraoperative blood product transfusion, and CPB time demonstrated significant predictive power for in-hospital mortality. Smith et al. [50] reported a link between massive plasma transfusion and adverse outcomes after cardiac surgery, likely due to complications associated with transfusion. These complications include transfusion-related acute lung injury, transfusion-associated circulatory overload, febrile and allergic reactions, infections, and multi-organ dysfunction, all of which are strongly associated with increased in-hospital mortality risk [51, 52, 53]. The prognostic importance of Plt in predicting mortality from aortic dissection has been extensively studied [54, 55, 56, 57], showing that decreased preoperative platelet levels are correlated with bleeding complications and increased fatality risk. In this study, the SVM algorithm effectively identified platelets as a significant predictor, consistent with previous findings.

This study developed a simple, effective, accurate, and practical model to predict the risk of in-hospital mortality following EAR. Our model offers several advantages over other predictive models [6, 8, 10, 11]. First, we compared multiple modeling approaches and selected the model with the best predictive performance. Second, our model incorporated a wide range of factors, including demographic characteristics, comorbidities, preoperative conditions, laboratory values, TTE data, and surgical details, making it comprehensive in assessing in-hospital mortality risk. Third, it was constructed using clinical data from a large cohort of ATAAD patients undergoing EAR at a high-volume center over an 8-year period. Lastly, we provided nomograms and a web-based calculator, enabling other researchers and clinicians to input their own data to estimate in-hospital mortality risk following EAR. This nomogram-based predictive model empowers surgeons to accurately assess the risk of in-hospital mortality for ATAAD patients undergoing EAR. As an effective visualization tool, it facilitates precise postoperative evaluations and patient risk stratification. By applying this model, surgeons can improve the quality of postoperative care, develop personalized treatment plans, and implement effective surgical strategies, ultimately enhancing survival rates and overall therapeutic outcomes.

Although some of the selected factors in this study are not novel, their combination was employed to develop a nomogram capable of predicting the risks of adverse events. However, external validation using larger sample sizes remains necessary before clinical implementation. This can be achieved through multi-center and/or multinational collaborative efforts. To promote reproducibility and support further validation, the code and model equations have been provided in this paper. By refining parameters such as regression coefficients using multi-center data, potential biases in the model could be minimized, and its predictive performance optimized. Further validation with external datasets will also enhance the model’s interpretability and reliability. To assist clinicians, an online calculator has been developed to support postoperative management.

This study has several limitations. First, it was a retrospective single-center analysis, which may have introduced selection bias. Second, procedures were performed by up to five surgeons, and variations in surgical experience and technique may have contributed to uncertainty in the results. Third, the study cohort included 84 patients aged 70 years or older, accounting for only 9% of the population, with a mortality rate of 11.9%. This suggests that the model may have limited predictive value for elderly patients. Fourth, among the cohort, 71.1% of patients had hypertension, whereas the prevalence of other comorbidities was less than 5%. As a result, the predictive value of this model for patients with multiple comorbidities may be restricted. Fifth, despite the measures taken to prevent it, machine learning algorithms may still be susceptible to overfitting. Finally, external validation with independent cohorts has yet to be conducted.

5. Conclusions

We developed a novel nomogram-based risk prediction model using the SVM algorithm to predict in-hospital mortality following extended aortic arch repair for ATAAD. The model demonstrated good discrimination and accuracy. The combination of cerebral malperfusion, mesenteric malperfusion, CPStatus, MFS, D2, Plt, CABG, and CPB time was identified as having significant predictive capability.

Availability of Data and Materials

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

References

[1] Evangelista A, Isselbacher EM, Bossone E, Gleason TG, Eusanio MD, Sechtem U, et al. Insights From the International Registry of Acute Aortic Dissection: A 20-Year Experience of Collaborative Clinical Research. Circulation. 2018; 137: 1846–1860. https://doi.org/10.1161/CIRCULATIONAHA.117.031264.

[2] Trimarchi S, Nienaber CA, Rampoldi V, Myrmel T, Suzuki T, Mehta RH, et al. Contemporary results of surgery in acute type A aortic dissection: The International Registry of Acute Aortic Dissection experience. The Journal of Thoracic and Cardiovascular Surgery. 2005; 129: 112–122. https://doi.org/10.1016/j.jtcvs.2004.09.005.

[3] Chen SW, Chen Y, Ma WG, Zhong YL, Qiao ZY, Ge YP, et al. Limited vs. extended repair for acute type I aortic dissection: long-term outcomes over a decade in Beijing Anzhen Hospital. Chinese Medical Journal. 2021; 134: 986–988. https://doi.org/10.1097/CM9.0000000000001416.

[4] Xue Y, Pan J, Cao H, Fan F, Luo X, Ge M, et al. Different aortic arch surgery methods for type A aortic dissection: clinical outcomes and follow-up results. Interactive Cardiovascular and Thoracic Surgery. 2020; 31: 254–262. https://doi.org/10.1093/icvts/ivaa095.

[5] Sun L, Qi R, Zhu J, Liu Y, Zheng J. Total arch replacement combined with stented elephant trunk implantation: a new “standard” therapy for type a dissection involving repair of the aortic arch? Circulation. 2011; 123: 971–978. https://doi.org/10.1161/CIRCULATIONAHA.110.015081.

[6] Czerny M, Siepe M, Beyersdorf F, Feisst M, Gabel M, Pilz M, et al. Prediction of mortality rate in acute type A dissection: the German Registry for Acute Type A Aortic Dissection score. European Journal of Cardio-thoracic Surgery: Official Journal of the European Association for Cardio-thoracic Surgery. 2020; 58: 700–706. https://doi.org/10.1093/ejcts/ezaa156.

[7] Berretta P, Patel HJ, Gleason TG, Sundt TM, Myrmel T, Desai N, et al. IRAD experience on surgical type A acute dissection patients: results and predictors of mortality. Annals of Cardiothoracic Surgery. 2016; 5: 346–351. https://doi.org/10.21037/acs.2016.05.10.

[8] Nashef SAM, Roques F, Sharples LD, Nilsson J, Smith C, Goldstone AR, et al. EuroSCORE II. European Journal of Cardio-thoracic Surgery: Official Journal of the European Association for Cardio-thoracic Surgery. 2012; 41: 734–734–44; discussion 744–5. https://doi.org/10.1093/ejcts/ezs043.

[9] Ge Y, Sun L, Zhu J, Liu Y, Cheng L, Chen L, et al. Can EuroSCORE II predict the mortality and length of intensive care unit stay after total aortic arch replacement with stented elephant trunk implantation for DeBakey type I aortic dissection? The Thoracic and Cardiovascular Surgeon. 2013; 61: 564–568. https://doi.org/10.1055/s-0033-1348197.

[10] Rampoldi V, Trimarchi S, Eagle KA, Nienaber CA, Oh JK, Bossone E, et al. Simple risk models to predict surgical mortality in acute type A aortic dissection: the International Registry of Acute Aortic Dissection score. The Annals of Thoracic Surgery. 2007; 83: 55–61. https://doi.org/10.1016/j.athoracsur.2006.08.007.

[11] Zhang Y, Chen T, Chen Q, Min H, Nan J, Guo Z. Development and evaluation of an early death risk prediction model after acute type A aortic dissection. Annals of Translational Medicine. 2021; 9: 1442. https://doi.org/10.21037/atm-21-4063.

[12] Cutillo CM, Sharma KR, Foschini L, Kundu S, Mackintosh M, Mandl KD, et al. Machine intelligence in healthcare-perspectives on trustworthiness, explainability, usability, and transparency. NPJ Digital Medicine. 2020; 3: 47. https://doi.org/10.1038/s41746-020-0254-2.

[13] Wilkinson J, Arnold KF, Murray EJ, van Smeden M, Carr K, Sippy R, et al. Time to reality check the promises of machine learning-powered precision medicine. The Lancet. Digital Health. 2020; 2: e677–e680. https://doi.org/10.1016/S2589-7500(20)30200-4.

[14] Storm H, Baylis K, Heckelei T. Machine learning in agricultural and applied economics. European Review of Agricultural Economics. 2020; 47: 849–892. https://doi.org/10.1093/erae/jbz033.

[15] Writing Committee Members, Isselbacher EM, Preventza O, Hamilton Black Iii J, Augoustides JG, Beck AW, et al. 2022 ACC/AHA Guideline for the Diagnosis and Management of Aortic Disease: A Report of the American Heart Association/American College of Cardiology Joint Committee on Clinical Practice Guidelines. Journal of the American College of Cardiology. 2022; 80: e223–e393. https://doi.org/10.1016/j.jacc.2022.08.004.

[16] Neri E, Toscano T, Papalia U, Frati G, Massetti M, Capannini G, et al. Proximal aortic dissection with coronary malperfusion: presentation, management, and outcome. The Journal of Thoracic and Cardiovascular Surgery. 2001; 121: 552–560. https://doi.org/10.1067/mtc.2001.112534.

[17] Lundberg SM, Nair B, Vavilala MS, Horibe M, Eisses MJ, Adams T, et al. Explainable machine-learning predictions for the prevention of hypoxaemia during surgery. Nature Biomedical Engineering. 2018; 2: 749–760. https://doi.org/10.1038/s41551-018-0304-0.

[18] Bull SB. Sample size and power determination for a binary outcome and an ordinal exposure when logistic regression analysis is planned. American Journal of Epidemiology. 1993; 137: 676–684. https://doi.org/10.1093/oxfordjournals.aje.a116725.

[19] Vickers AJ, Elkin EB. Decision curve analysis: a novel method for evaluating prediction models. Medical Decision Making: an International Journal of the Society for Medical Decision Making. 2006; 26: 565–574. https://doi.org/10.1177/0272989X06295361.

[20] Riley RD, Ensor J, Snell KIE, Harrell FE, Jr, Martin GP, Reitsma JB, et al. Calculating the sample size required for developing a clinical prediction model. BMJ (Clinical Research Ed.). 2020; 368: m441. https://doi.org/10.1136/bmj.m441.

[21] Uchida K, Minami T, Cho T, Yasuda S, Kasama K, Suzuki S, et al. Results of ascending aortic and arch replacement for type A aortic dissection. The Journal of Thoracic and Cardiovascular Surgery. 2021; 162: 1025–1031. https://doi.org/10.1016/j.jtcvs.2020.02.087.

[22] Lundberg SM, Lee S-I. A Unified Approach to Interpreting Model Predictions. 31st Annual Conference on Neural Information Processing Systems (NIPS). Curran Associates Inc. 2017; 30: 4768–4777.

[23] Gianfrancesco MA, Tamang S, Yazdany J, Schmajuk G. Potential Biases in Machine Learning Algorithms Using Electronic Health Record Data. JAMA Internal Medicine. 2018; 178: 1544–1547. https://doi.org/10.1001/jamainternmed.2018.3763.

[24] d’Alessandro B, O’Neil C, LaGatta T. Conscientious Classification: A Data Scientist’s Guide to Discrimination-Aware Classification. Big Data. 2017; 5: 120–134. https://doi.org/10.1089/big.2016.0048.

[25] Kernbach JM, Staartjes VE. Foundations of Machine Learning-Based Clinical Prediction Modeling: Part II-Generalization and Overfitting. Acta Neurochirurgica. Supplement. 2022; 134: 15–21. https://doi.org/10.1007/978-3-030-85292-4_3.

[26] Charilaou P, Battat R. Machine learning models and over-fitting considerations. World Journal of Gastroenterology. 2022; 28: 605–607. https://doi.org/10.3748/wjg.v28.i5.605.

[27] Wu X, Zuo W, Lin L, Jia W, Zhang D. F-SVM: Combination of Feature Transformation and SVM Learning via Convex Relaxation. IEEE Transactions on Neural Networks and Learning Systems. 2018; 29: 5185–5199. https://doi.org/10.1109/TNNLS.2018.2791507.

[28] Harky A, Chan J, MacCarthy-Ofosu B. The future of stenting in patients with type A aortic dissection: a systematic review. The Journal of International Medical Research. 2020; 48: 300060519871372. https://doi.org/10.1177/0300060519871372.

[29] Yamaguchi T, Nakai M, Sumita Y, Miyamoto Y, Matsuda H, Inoue Y, et al. Current status of the management and outcomes of acute aortic dissection in Japan: Analyses of nationwide Japanese Registry of All Cardiac and Vascular Diseases-Diagnostic Procedure Combination data. European Heart Journal. Acute Cardiovascular Care. 2020; 9: S21–S31. https://doi.org/10.1177/2048872619872847.

[30] Macrina F, Puddu PE, Sciangula A, Trigilia F, Totaro M, Miraldi F, et al. Artificial neural networks versus multiple logistic regression to predict 30-day mortality after operations for type a ascending aortic dissection. The Open Cardiovascular Medicine Journal. 2009; 3: 81–95. https://doi.org/10.2174/1874192400903010081.

[31] Tsukube T, Haraguchi T, Okada Y, Matsukawa R, Kozawa S, Ogawa K, et al. Long-term outcomes after immediate aortic repair for acute type A aortic dissection complicated by coma. The Journal of Thoracic and Cardiovascular Surgery. 2014; 148: 1013–1018; discussion 1018–9. https://doi.org/10.1016/j.jtcvs.2014.06.053.

[32] Fukuhara S, Norton EL, Chaudhary N, Burris N, Shiomi S, Kim KM, et al. Type A Aortic Dissection With Cerebral Malperfusion: New Insights. The Annals of Thoracic Surgery. 2021; 112: 501–509. https://doi.org/10.1016/j.athoracsur.2020.08.046.

[33] Di Eusanio M, Trimarchi S, Patel HJ, Hutchison S, Suzuki T, Peterson MD, et al. Clinical presentation, management, and short-term outcome of patients with type A acute dissection complicated by mesenteric malperfusion: observations from the International Registry of Acute Aortic Dissection. The Journal of Thoracic and Cardiovascular Surgery. 2013; 145: 385–390.e1. https://doi.org/10.1016/j.jtcvs.2012.01.042.

[34] Cui X, Lin L, Sun X, Wang L, Shen R. Curcumin Protects against Renal Ischemia/Reperfusion Injury by Regulating Oxidative Stress and Inflammatory Response. Evidence-based Complementary and Alternative Medicine: ECAM. 2021; 2021: 8490772. https://doi.org/10.1155/2021/8490772.

[35] Velayudhan BV, Idhrees AM, Mukesh K, Kannan RN. Mesenteric Malperfusion in Acute Aortic Dissection: Challenges and Frontiers. Seminars in Thoracic and Cardiovascular Surgery. 2019; 31: 668–673. https://doi.org/10.1053/j.semtcvs.2019.03.012.

[36] Zhang K, Dong SB, Pan XD, Lin Y, Zhu K, Zheng J, et al. Concomitant coronary artery bypass grafting during surgical repair of acute type A aortic dissection affects operative mortality rather than midterm mortality. Asian Journal of Surgery. 2021; 44: 945–951. https://doi.org/10.1016/j.asjsur.2021.01.031.

[37] Tolenaar JL, Froehlich W, Jonker FHW, Upchurch GR, Jr, Rampoldi V, Tsai TT, et al. Predicting in-hospital mortality in acute type B aortic dissection: evidence from International Registry of Acute Aortic Dissection. Circulation. 2014; 130: S45–50. https://doi.org/10.1161/CIRCULATIONAHA.113.007117.

[38] Santini F, Montalbano G, Casali G, Messina A, Iafrancesco M, Luciani GB, et al. Clinical presentation is the main predictor of in-hospital death for patients with acute type A aortic dissection admitted for surgical treatment: a 25 years experience. International Journal of Cardiology. 2007; 115: 305–311. https://doi.org/10.1016/j.ijcard.2006.03.013.

[39] Leontyev S, Légaré JF, Borger MA, Buth KJ, Funkat AK, Gerhard J, et al. Creation of a Scorecard to Predict In-Hospital Death in Patients Undergoing Operations for Acute Type A Aortic Dissection. The Annals of Thoracic Surgery. 2016; 101: 1700–1706. https://doi.org/10.1016/j.athoracsur.2015.10.038.

[40] Macrina F, Puddu PE, Sciangula A, Totaro M, Trigilia F, Cassese M, et al. Long-term mortality prediction after operations for type A ascending aortic dissection. Journal of Cardiothoracic Surgery. 2010; 5: 42. https://doi.org/10.1186/1749-8090-5-42.

[41] Zhang K, Pan XD, Dong SB, Zheng J, Xu SD, Liu YM, et al. Cardiopulmonary bypass duration is an independent predictor of adverse outcome in surgical repair for acute type A aortic dissection. The Journal of International Medical Research. 2020; 48: 300060520968450. https://doi.org/10.1177/0300060520968450.

[42] Fujimori T, Kimura N, Mieno M, Hori D, Kusadokoro S, Tanaka M, et al. An increased prothrombin time-international normalized ratio in patients with acute type A aortic dissection: contributing factors and their influence on outcomes. Surgery Today. 2022; 52: 431–440. https://doi.org/10.1007/s00595-021-02399-y.

[43] Itagaki R, Kimura N, Mieno M, Hori D, Itoh S, Akiyoshi K, et al. Characteristics and Treatment Outcomes of Acute Type A Aortic Dissection With Elevated D-Dimer Concentration. Journal of the American Heart Association. 2018; 7: e009144. https://doi.org/10.1161/JAHA.118.009144.

[44] Wen D, Du X, Dong JZ, Zhou XL, Ma CS. Value of D-dimer and C reactive protein in predicting inhospital death in acute aortic dissection. Heart (British Cardiac Society). 2013; 99: 1192–1197. https://doi.org/10.1136/heartjnl-2013-304158.

[45] Huang M, Lian Y, Zeng Z, Li J. D-dimer, C-reactive protein and matrix metalloproteinase 9 for prediction of type A aortic dissection patient survival. ESC Heart Failure. 2024; 11: 147–154. https://doi.org/10.1002/ehf2.14552.

[46] Rylski B, Bavaria JE, Beyersdorf F, Branchetti E, Desai ND, Milewski RK, et al. Type A aortic dissection in Marfan syndrome: extent of initial surgery determines long-term outcome. Circulation. 2014; 129: 1381–1386. https://doi.org/10.1161/CIRCULATIONAHA.113.005865.

[47] Zhao S, Liu Z, Wen M, Zhang H, Wang L, Zhang N, et al. Association between preoperative D-dimer with morphologic features and surgical outcomes of acute type A aortic dissection. Interdisciplinary Cardiovascular and Thoracic Surgery. 2024; 39: ivae193. https://doi.org/10.1093/icvts/ivae193.

[48] Tian L, Fan X, Zhu J, Liang Y, Li J, Yang Y. Plasma D-dimer and in-hospital mortality in patients with Stanford type A acute aortic dissection. Blood Coagulation & Fibrinolysis: an International Journal in Haemostasis and Thrombosis. 2014; 25: 161–166. https://doi.org/10.1097/MBC.0000000000000013.

[49] Gasparovic H, Plestina S, Sutlic Z, Husedzinovic I, Coric V, Ivancan V, et al. Pulmonary lactate release following cardiopulmonary bypass. European Journal of Cardio-thoracic Surgery: Official Journal of the European Association for Cardio-thoracic Surgery. 2007; 32: 882–887. https://doi.org/10.1016/j.ejcts.2007.09.001.

[50] Smith MM, Kor DJ, Frank RD, Weister TJ, Dearani JA, Warner MA. Intraoperative Plasma Transfusion Volumes and Outcomes in Cardiac Surgery. Journal of Cardiothoracic and Vascular Anesthesia. 2020; 34: 1446–1456. https://doi.org/10.1053/j.jvca.2019.12.049.

[51] Clifford L, Jia Q, Subramanian A, Yadav H, Wilson GA, Murphy SP, et al. Characterizing the epidemiology of postoperative transfusion-related acute lung injury. Anesthesiology. 2015; 122: 12–20. https://doi.org/10.1097/ALN.0000000000000514.

[52] Clifford L, Jia Q, Yadav H, Subramanian A, Wilson GA, Murphy SP, et al. Characterizing the epidemiology of perioperative transfusion-associated circulatory overload. Anesthesiology. 2015; 122: 21–28. https://doi.org/10.1097/ALN.0000000000000513.

[53] Kor DJ, Stubbs JR, Gajic O. Perioperative coagulation management–fresh frozen plasma. Best Practice & Research. Clinical Anaesthesiology. 2010; 24: 51–64. https://doi.org/10.1016/j.bpa.2009.09.007.

[54] Wang Y, Qiao T, Zhou J. The short-term prognostic value of serum platelet to hemoglobin in patients with type A acute aortic dissection. Perfusion. 2022; 37: 95–99. https://doi.org/10.1177/0267659120982226.

[55] Xie E, Liu J, Liu Y, Liu Y, Xue L, Fan R, et al. Association between platelet counts and morbidity and mortality after endovascular repair for type B aortic dissection. Platelets. 2022; 33: 73–81. https://doi.org/10.1080/09537104.2020.1847266.

[56] Xie X, Fu X, Zhang Y, Huang W, Huang L, Deng Y, et al. U-shaped relationship between platelet-lymphocyte ratio and postoperative in-hospital mortality in patients with type A acute aortic dissection. BMC Cardiovascular Disorders. 2021; 21: 569. https://doi.org/10.1186/s12872-021-02391-x.

[57] Guvenc O, Engin M. The role of neutrophil-lymphocyte platelet ratio in predicting in-hospital mortality after acute Type A aortic dissection operations. European Review for Medical and Pharmacological Sciences. 2023; 27: 1534–1539. https://doi.org/10.26355/eurrev_202302_31396.