This retrospective study was conducted at the Northern Theater General Hospital and was approved by the Ethics Committee under the ethics batch number Y(2022)151. Since this study was retrospective in nature, the Committee waived the requirement for patients’ written informed consent.

This study conducted a continuous screening of patients diagnosed with IMH using computed tomography angiography (CTA) from July 2015 to June 2023. The screening was performed in the Emergency Department and Department of Cardiovascular Medicine of the Northern Theater General Hospital. The inclusion criteria were: (I) age ≥18 years old; (II) type B IMH. The exclusion criteria were: (I) subacute and chronic IMH with onset of more than 15 days; (II) type A IMH; (III) complicated type B IMH; (IV) previous TEVAR or surgical vascular repair; (V) traumatic aortic injury; (VI) IMH combined with Marfan syndrome and other genetic diseases and connective tissue diseases; (VII) incomplete clinical or imaging data; (VIII) lost to follow-up. A total of 197 patients with acute uncomplicated type B IMH who met the criteria were included in the study. They were divided into two groups: stable group (n = 125) and exacerbation group (n = 72). The exacerbation group is defined as patients who, within a 30-day post-discharge observation period, suffer aortic-related adverse events—such as deaths linked to aortic diseases or sudden deaths, or the progression to conditions like AD, PAU, and other complicated AAS events—despite receiving optimized medical therapy (OMT). The stable group is defined as patients who did not suffer any of the previously mentioned aortic-related adverse events during the observation period (Figure 1).

All patients received OMT which included sedation, pain relief, blood pressure reduction, heart rate reduction, and ECG monitoring. The treatment goals were to eliminate pain and control blood pressure and heart rate. To achieve this, drugs were used to maintain blood pressure between 100 and 120 mmHg systolic and 70–80 mmHg diastolic (1 mmHg = 0.133 kPa), and heart rate between 60 and 70 beats/min. Blood pressure management primarily relied on oral anti-hypertensive drugs, with the specific medication(s) chosen based on each patient's individual situation. β-blockers were used to control heart rate, with dosage adjustments or discontinuation if the heart rate dropped below 55 beats/min.

All instances of IMH were diagnosed using thoracoabdominal aorta CTA with 3D reconstruction. Computed tomography was performed using a 256-detector Siemens Sensation. Intravenous administration of 80–150 ml of non-ionic contrast agent was carried out. Scans were conducted from the lower neck to the lower pelvis. All CTA images were obtained from the institutional Picture Archiving and Communication Systems. The location of the aortic lesion, aortic atherosclerosis, ulcer-like projection (ULP), hematoma shape, extent of hematoma involvement, pleural effusion, and pericardial effusion were analyzed. The maximum diameter of the ascending aorta, the maximum descending aorta diameter (MDAD), and the maximum hematoma thickness of the descending aorta (MDHT) were measured. The standard scanning protocol involved confirming the diagnosis with the initial imaging examination within 14 days of onset, followed by CTA scans at 2 weeks and 1 month. When imaging tests indicate that IMH has progressed to AD, PAU, or other high-risk conditions like impending aortic rupture, the patient may consider undergoing preventive invasive interventions to avert aortic rupture or death. The CTA images of all IMH patients were accurately diagnosed and measured by at least two experienced clinicians.

The definition of the complicated IMH (8) encompasses the following clinical conditions: expansion of the IMH despite pharmacological management, impending rupture, uncontrolled hypertension, end-organ damage, refractory pain, or evidence of an intimal tear on CTA. Aortic-related adverse events were defined as either aortic disease-related death or sudden death, or AAS composite events such as AD and PAU. AD (9) occurs when an intrusive intimal tear divides the aorta into true and false lumens, with varying degrees of contrast medium filling both lumens. PAU (10) refers to a situation where an aortic atherosclerotic plaque penetrates the internal elastic layer into the media, creating a pocket-like protrusion on the aortic wall that is filled with contrast agent. This condition is characterized by intimal destruction and may lead to the formation of a hematoma around the protrusion. ULP (11) refers to a focal pouch filled with contrast agent that protrudes from the aortic lumen, with a breach diameter greater than 3 mm. It is different from PAU as it does not penetrate the internal elastic membrane and can occur even in the absence of atherosclerosis in patients with the disease. According to the 2014 ESC guidelines, the aorta is divided into four segments (4): the ascending aorta (from the aortic sinotubular junction to the proximal end of the brachiocephalic opening), the aortic arch (from the opening of the brachiocephalic trunk to the opening of the left subclavian artery), the descending thoracic aorta (from the distal end of the opening of the left subclavian artery to the level of the diaphragm), and the abdominal aorta (from the level of the diaphragm to the bifurcation of the iliac artery). If a hematoma involves one segment, it is recorded as 1; if two segments are involved, it is recorded as 2, and so on.

Statistical analysis was conducted using SPSS 26.0 and R 4.2.0. Continuous variables were presented as mean ± SD or median (quartile 1 to quartile 3). Shapiro–Wilk test and generated Quantile-Quantile plots were employed to assess whether the variables follow a normal or near-normal distribution. Independent samples t-test was used to compare normally distributed variables, while the Mann–Whitney U test was used for non-normally distributed variables. Categorical variables were reported as frequencies and percentages and compared using the χ² or Fisher's exact test. We incorporated variables that exhibited significant differences between the two groups (P < 0.1) into univariate logistic regression analyses. Subsequently, the variables with P < 0.05 in the univariate analysis were further analyzed using multivariate analysis to further identify independent predictors. Feature selection was performed using the “glmnet” package, and factors with non-zero coefficients in the LASSO regression model were selected. The optimal penalty coefficient λ was determined using the 10-fold cross-validation method and 1 standard error (1-SE) criterion. The rms package was utilized for constructing the nomogram. To draw the receiver operating characteristic curve (ROC), calculate the area under the curve (AUC), and evaluate the model's differential diagnosis performance, the pROC package was employed. Additionally, a calibration curve was drawn to assess the calibration of the model. The rmda package was utilized for drawing decision analysis curves (DCA) and clinical impact curves (CIC) to evaluate the clinical benefit of the model. Internal validation was performed using the bootstrap method. Statistical significance was determined using a two-sided test with a P-value of less than 0.05.

This study included a total of 197 patients with type B IMH, consisting of 135 males and 62 females, with a mean age of 62.19 ± 9.79. Out of these patients, 72 (36.5%) experienced short-term aortic-related adverse events. Specifically, out of the total number of patients, 61 patients developed PAU, 8 developed AD, and 3 cases resulted in death related to aortic disease or sudden death.

In this study, the prevalence of diabetes in the type B IMH exacerbation group (1.4% vs. 12.0%, P = 0.009) was significantly lower compared to the stable group, while the prevalence of anemia (15.3% vs. 3.2%, P = 0.002) was significantly higher in the exacerbation group. These differences were statistically significant. There were no significant differences between the two groups in terms of age, gender, BMI, admission blood pressure, heart rate, drinking history, smoking history, and clinical symptoms (all P > 0.05). Please refer to Tables 1, 2.

Laboratory examination results revealed that the type B IMH exacerbation group had significantly lower levels of glucose [6.14 (5.22, 7.41) mmol/L vs. 7.08 (6.03, 8.66) mmol/L, P < 0.001] compared to the stable group. On the other hand, the type B IMH exacerbation group had significantly higher levels of creatinine [68.53 (57.35, 87.60) µmol/L vs. 65.60 (49.88, 79.70) µmol/L, P = 0.034], C-reactive protein [11.80 (2.48, 48.90) mg/L vs. 6.15 (2.00, 14.23) mg/L, P = 0.043], and D-dimer level [2.31 (1.01, 4.38) mg/L vs. 1.30 (0.63, 2.87) mg/L, P = 0.008] compared to the stable group. These differences were statistically significant. However, there were no significant statistical differences in white blood cell count, platelet count, serum low-density lipoprotein C, creatine phosphokinase-myocardial band, troponin T, alanine aminotransferase, and aspartate aminotransferase between the two groups (all P > 0.05) (Table 3).

The results of the imaging examination revealed that the proportion of MDAD (37.97 ± 4.64 mm vs. 35.09 ± 4.76 mm, P < 0.001), MDHT (9.77 ± 3.27 mm vs. 8.83 ± 3.16 mm, P = 0.049), and ULP (41.7% vs. 9.6%, P < 0.001) were significantly higher in the exacerbation group compared to the stable group. These differences were statistically significant (all P < 0.05). However, there were no significant statistical differences between the two groups in terms of pleural effusion, pericardial effusion, maximum ascending aorta diameter, aortic atherosclerosis, number of hematoma-involved segments, IMH passing diaphragm, IMH passing iliac artery, crescent hematoma, and circular hematoma (all P > 0.05) (Table 4).

Logistic regression analysis was conducted to examine the relationship between clinical data, imaging characteristics, and the occurrence of aortic-related adverse events in type B IMH. The results revealed that diabetes (OR = 0.053, 95%CI: 0.005–0.580, P = 0.016), anemia (OR = 6.814, 95%CI: 1.761–26.364, P = 0.005), MDAD (OR = 1.149, 95%CI: 1.066–1.238, P < 0.001), ULP (OR = 8.941, 95%CI: 3.708–21.561, P < 0.001) were independent predictors of aortic-related adverse events. These predictors are summarized in Table 5. To further validate the predictors, LASSO regression analysis was performed, which confirmed the same four predictors as the logistic regression analysis, thereby strengthening our model (Figure 2).

A nomogram risk prediction model for short-term exacerbation in type B IMH patients was established based on the 4 statistically significant influencing factors obtained from Logistic regression and LASSO regression analysis. Each risk factor was evaluated to calculate the predicted probability of short-term exacerbation risk for patients with type B IMH. Figure 3 illustrates that each variable corresponds to a line segment, with the scale indicating the value. The total score is calculated by adding the corresponding scores above each variable, and based on this total score, the predicted probability of the patient's recent deterioration risk is obtained. To address the limitations of nomograms in practical applications, this study developed a web-based nomogram dynamic prediction tool (Figure 4) (https://imh2023.shinyapps.io/DynNomapp/). This tool facilitates easy calculation and assists in clinical decision-making. As an example, if a patient has no history of diabetes or anemia, and exhibits imaging manifestations accompanied by ULP, along with an MDAD of 38 mm, the results indicate a predicted risk of aortic-related adverse events in the patient to be 77.1%. This suggests a high risk of recent deterioration for the patient.

The ROC curve was generated based on the nomogram risk model, and the AUC was 0.813 (95%CI, 0.756–0.872), indicating a strong predictive performance of the model. At a Youden index of 0.334, the model demonstrated a specificity of 0.752 and a sensitivity of 0.736 (Figure 5). To validate the model, the Bootstrap method was employed with 1000 repetitions, resulting in an AUC of 0.794. The calibration curve demonstrated good agreement between the predicted and actual probabilities of the type B IMH short-term exacerbation risk model (Figure 6), with an accuracy rate of 0.724, indicating the model’s reliability. Furthermore, the Hosmer–Lemeshow goodness-of-fit test yielded a non-significant result (χ² = 7.040, P = 0.532), confirming the model's stability.

DCA was utilized to assess the probability of short-term exacerbation in patients with type B IMH, aiming to determine whether the model yielded a significant net benefit. According to this study, DCA revealed that our predictive model generated more clinical net benefit when the risk threshold was set within the range of 0.01–0.7, compared to the options of “no intervention” or “all intervention” (Figure 7). Additionally, CIC plots were employed to illustrate the number of high-risk patients and the number of truly high-risk cases at each high-risk threshold, thereby demonstrating the model's favorable predictability and clinical applicability (Figure 8).