This retrospective study analyzed 60 patients diagnosed with aortic dissection between November 2019 and November 2023, all with primary entry tears in the descending aorta. In the study design, we used a sample size estimation formula for binary variables based on the incidence rates reported in previous studies. We calculated with an alpha of 0.05 and a power of 0.80, and the results showed that at least 20 patients per group would meet the requirements for model analysis. The study ultimately included 21 patients with RTAD and 39 with TBAD. Demographic and clinical data, including gender, age, height, weight, body mass index (BMI), clinical presentation, medical history (e.g., hyperlipidemia, hypertension, diabetes, coronary artery disease), and prior cardiac/thoracic trauma/surgery, were collected from medical records. Aortic morphological parameters were measured using Carestream Image Suite V4 (New York, USA) and EndoSize (Endovastec, Shanghai, China) software, including:

(1) Maximum/minimum diameters of the aortic root, ascending aorta, proximal/distal aortic arch, and true/false lumen areas. (2) Aortic arch morphology, entry tear location (greater/lesser curvature, anterior/posterior wall), tear size (length $\times$ width), and distance from the tear to the left subclavian artery. (3) Maximum/minimum diameters and true/false lumen areas at the proximal tear site, tear site, distal tear site, celiac trunk level, renal artery level, and aortic bifurcation. (4) Aortic arch angles ($\alpha$ and $\beta$ angles). The study was approved by the Hospital Ethics Committee (Ethics No.: 2024LL0129002), with informed consent obtained from all patients and families.

Continuous variables: Normally distributed data were expressed as mean $\pm$ standard deviation and compared using independent t-tests. Non-normally distributed data were presented as median (interquartile range) and analyzed with nonparametric tests. Categorical variables: Described as frequency (%) and compared using chi-square tests. Variable selection: LASSO regression was applied for dimensionality reduction. Variables with statistical significance in LASSO analysis were incorporated into a multivariate logistic regression model. Odds ratios (ORs) with 95% confidence intervals (CIs) were calculated, and a nomogram was constructed to visualize the prediction model. Model validation: The concordance index (C-index), receiver operating characteristic (ROC) curve, calibration curve, and decision curve analysis (DCA) were performed for internal validation. A two-tailed p 0.05 was considered statistically significant.

The study included 60 patients: 39 in the TBAD group (32 males, 7 females; mean age 54.18 $\pm$ 13.78 years) and 21 in the RTAD group (20 males, 1 female; mean age 49.33 $\pm$ 10.03 years). No significant differences were observed between the two groups in gender, age, height, weight, BMI, hyperlipidemia, hypertension, diabetes, coronary artery disease, or smoking history (all p $>$ 0.05). Details are shown in Table 1.

For maximum aortic root diameter, minimum aortic root diameter, maximum ascending aortic diameter, minimum ascending aortic diameter, maximum proximal aortic arch diameter, minimum proximal aortic arch diameter, maximum distal aortic arch diameter, and minimum distal aortic arch diameter, the p-values of the difference analysis were all less than 0.05, indicating statistically significant differences between the two groups in these variables, with the RTAD group showing smaller diameters compared to the TBAD group. For other parameters, the p-values of the difference analysis were all greater than 0.05, suggesting no statistically significant differences between the two groups in those variables. Details are provided in Table 2.

For the true lumen area of the aortic root, false lumen area of the aortic root, true lumen area of the ascending aorta, false lumen area of the ascending aorta, true lumen area of the proximal aortic arch, false lumen area of the proximal aortic arch, true lumen area of the distal aortic arch, and false lumen area of the distal aortic arch, the p-values of the difference analysis were all less than 0.05, indicating statistically significant differences between the two groups in these variables, with the RTAD group demonstrating higher values compared to the TBAD group. For other parameters (including entry tear area and distance from the tear site to the left subclavian artery), the p-values of the difference analysis were all greater than 0.05, suggesting no statistically significant differences between the two groups in those variables. Details are presented in Table 3.

The p-values of the difference analysis for both aortic arch type and tear location were greater than 0.05, indicating no statistically significant differences between the two groups in these variables. Details are provided in Table 4.

The p-values of the difference analysis for both the alpha and beta angles of the aortic arch were greater than 0.05, indicating no statistically significant differences between the two groups in these variables. Details are presented in Table 5.

A total of 44 independent variables were included in this study. Dimensionality reduction was performed to screen for the most representative high-risk predictors through LASSO regression analysis. The $\lambda$ value was selected as the optimal one through ten-fold cross-validation using the 1-SE criterion, rather than being manually preset. As the penalty coefficient ($\lambda$) increased, the coefficients of initially included variables were gradually shrunk, with some ultimately compressed to zero, thereby preventing model overfitting. The optimal penalty coefficient ($\lambda$) was identified using 10-fold cross-validation based on the minimum criterion. When $\lambda$ increased to the value corresponding to one standard error ($\lambda$ = 1 SE), it was selected as the optimal value for the model. Results are detailed below (Fig. 4A,B).

Fig. 4.

LASSO regression results for 44 predictor variables. (A) LASSO Dimension Reduction Results. (B) Process of selecting optimal independent variables via ten-fold cross-validation and identifying the optimal $\lambda$ value through LASSO cross-validation. (C) Risk Prediction Model for RTAD Occurrence. (D) ROC/AUC Evaluation of the Prediction Model. (E) Validation of the Prediction Model Robustness. (F) Evaluation of the Prediction Model. Note: Fig. 4A displays the coefficient trajectories of the 44 candidate independent variables. Fig. 4B illustrates the optimal variable selection via LASSO regression with 10-fold cross-validation. Following dimensionality reduction, the retained variables for subsequent analysis include: Minimum ascending aortic diameter (partial regression coefficient: 0.0001503664); Maximum ascending aortic diameter (partial regression coefficient: 0.0008316777); Maximum distal aortic arch diameter (partial regression coefficient: 0.00005818305); Minimum distal aortic arch diameter (partial regression coefficient: 0.0001007072); Proximal aortic arch true lumen area (partial regression coefficient: –0.00000000100702); Constant term: 0.08353068. LASSO, Least Absolute Shrinkage and Selection Operator; ROC, receiver operating characteristic; AUC, area under the curve.

The results of the multivariable logistic regression analysis for the selected predictors are as follows:

(1) Minimum ascending aortic diameter: The regression test p-value was less than 0.05, indicating a statistically significant influence on the dependent variable when controlling for other covariates. The odds ratio (OR) was less than 1 (OR 1), suggesting that an increase in this parameter was associated with a reduced incidence of RTAD outcomes, thus identified as an independent protective factor. (2) Maximum ascending aortic diameter: The regression test p-value was less than 0.05, demonstrating a statistically significant effect. The OR was greater than 1 (OR $>$1), indicating that an increase in this parameter correlated with an elevated incidence of RTAD outcomes, categorizing it as an independent risk factor. (3) Maximum distal aortic arch diameter: The regression test p-value exceeded 0.05, showing no statistically significant impact on the dependent variable after adjusting for confounders. (4) Minimum distal aortic arch diameter: The regression test p-value was less than 0.05, confirming statistical significance. The OR was less than 1 (OR 1), signifying that an increase in this measurement reduced the likelihood of RTAD outcomes, establishing it as an independent protective factor. (5) Proximal aortic arch true lumen area: The regression test p-value was greater than 0.05, indicating no statistically significant association with the outcome variable. Detailed results are summarized in Table 6.

Fig. 4C presents the nomogram constructed using statistically significant variables from the logistic regression equation to predict the risk of RTAD occurrence. In the nomogram: (1) The Points scale (first row) provides a scoring reference for each variable. (2) For any individual patient, the corresponding Points can be assigned based on the value or category of each predictor variable. (3) Total Points are calculated by summing the individual points for all variables. (4) The Total Points are then mapped to the Risk scale (second-to-last row) to obtain the predicted probability of RTAD occurrence.

Aortic dissection (AD) is a life-threatening cardiovascular disease. Over the years, researchers have focused on identifying its high-risk factors. Studies have reported that age, hypertension, dyslipidemia, and inherited connective tissue disorders are significant risk factors for AD [22]. Additionally, aortic geometric morphology has been shown to play a critical role in the development of both Type A and Type B aortic dissections.

In TAAD, specific geometric features of the proximal aorta, including elongation, angulation, and tortuosity, may significantly influence pathogenesis [23, 24, 25, 26, 27, 28]. Ascending aortic diameter remains the only widely accepted morphological risk factor for TAAD, with prophylactic surgery recommended when the diameter exceeds 55 mm, even in asymptomatic patients [29]. However, Krüger et al. [23] argue that aortic diameter alone may not be the optimal predictor, as most dissections occur at diameters below 55 mm. They propose aortic elongation as a potential risk factor for TAAD [23].

Studies on TBAD highlight increased aortic arch curvature and angulation as independent and specific predictors [30]. Other factors, such as aortic elongation, incremental angulation, and tortuosity indices, are also associated with TBAD [31, 32]. However, spatial geometric measurements on CTA are complex and time-consuming, complicating risk assessment. Sun et al. [33] proposed an easily identifiable morphological parameter—aortic arch type—as a surrogate marker. Type III aortic arch, characterized by elongation, increased angulation, and tortuosity, serves as a comparable identifier for high-risk TBAD patients [33].

Previous studies have broadly focused on geometric features of AD or compared TAAD and TBAD. However, RTAD, a distinct subtype, remains understudied. DiMusto et al. [20] demonstrated that RTAD often occurs when the primary entry tear is near the aortic arch with poor false lumen decompression through distal branches. This may result from increased pressure in the false lumen due to thrombus formation or slow flow, leading to retrograde propagation into the ascending aorta. One possible reason is that the patient population we studied and the models or data used in the studies by DiMusto et al. [20] differ. For example, our study only included primary RTAD, while the studies by DiMusto et al. [20] may have covered different types of aortic dissection, and the sample size and population characteristics may also have been different. In addition, our study emphasized the relationship between changes in aortic geometry and the occurrence of RTAD, while the studies by DiMusto et al. [20] mainly focused on the impact of the tear location. Therefore, different research focuses may have led to these differences. Osswald et al. [21] identified elevated wall shear stress (WSS) at the subclavian artery distal region in RTAD patients using computational fluid dynamics, suggesting WSS as a potential screening marker when combined with clinical risk factors. Dziodzio et al. [19] highlighted the importance of primary entry tear location in porcine models, noting that tears on the aortic arch’s lesser curvature may predispose to RTAD. This study focuses on primary RTAD, distinct from iatrogenic or stent-graft-induced RTAD, emphasizing differences in pathophysiology compared to TBAD and TAAD.

This study identifies reduced minimum ascending aortic diameter, increased maximum ascending aortic diameter, and reduced minimum distal aortic arch diameter as specific predictors for RTAD development in TBAD patients. These geometric changes may increase vascular eccentricity, altering hemodynamics and wall shear stress to influence dissection propagation. Although Dziodzio et al. [19] linked RTAD risk to primary tear location on the lesser curvature, this study found no such association, possibly due to anatomical differences between porcine and human aortas. Additionally, post-TEVAR RTAD risk may correlate with aortic arch angulation and curvature [34], but no significant differences in $\alpha$/$\beta$ angles were observed here, likely due to distinctions between primary and post-TEVAR RTAD.

The RTAD risk prediction nomogram, validated by a C-index of 0.952, ROC-AUC of 0.952, and robust calibration/DCA curves, provides clinical guidance for early intervention. High-risk patients should undergo intensified screening to mitigate retrograde progression risks.

This single-center retrospective study is limited by potential measurement variability despite standardized protocols. The small sample size and selection bias inherent to retrospective designs challenge generalizability. Future multi-center prospective studies with long-term follow-up are needed to validate these findings.