Our center is a major medical institution for the treatment of cerebrovascular diseases in the central region of China, equipped with a comprehensive database containing a large number of cerebrovascular disease cases, clinical data, and sophisticated examination and treatment methods. The samples for this study are all based on data in this database. We collected data on UIAs treated with SAE from 2016 to 2020, incorporating four different stent devices: the LEO, LVIS, Enterprise, and Neuroform Atlas. Exclusion criteria included intraoperative aneurysm rupture, pregnancy, other hypercoagulable states, severe systemic diseases, hematologic disorders, use of flow-diverter and patients with a history of antiplatelet medication use. All patients were divided into three groups based on conditions. The control group consisted of patients who received empirical treatment without platelet aggregation testing. The experimental and prediction groups comprised patients who underwent surgery based on platelet aggregation rate (PAR) as an indicator (generally considered achieving a PAR below 25%). After statistical analysis, the cases were randomly assigned to the experimental and prediction groups.

To reduce the occurrence of TEs we routinely administer 75 mg of clopidogrel and 100 mg of aspirin orally each day for 7 days prior to surgery and continue this DAPT for 1 month postoperatively. For patients who exhibit clopidogrel resistance identified through CYP2C19 genetic testing, we replace clopidogrel with ticagrelor (D = 86) for treatment. These patients are excluded from our study. We perform routine digital subtraction angiography (DSA) before surgery to confirm the status of the aneurysm and blood vessels, ensuring the success of the procedure. A PAR below 25% within 24 h before surgery is considered a criterion for proceeding with the operation.

Interventional surgical approach: All surgeries are performed under general anesthesia with the patient in a supine position and systemic continuous heparinization was administered intraoperatively. The protocol was as follows: heparinization was administered according to the patient’s body weight, with an initial dose of 200/3 U/kg, followed by an additional 1/2 of the previous dose every hour until a final dose of 1,000 U/h was reached, and postoperative heparinization was neutralized by natural neutralization. During the procedure, bilateral femoral arteries were punctured using the Seldinger technique. One femoral artery was used for angiography to localize the aneurysm and assess intraoperative thrombosis, while the other femoral artery was utilized for guidewire-guided SAE. Stent-assisted techniques are generally employed for wide-necked aneurysms and aneurysms in complex locations to achieve satisfactory embolization results. After stent implantation, X-per CT examination was performed to observe whether there was intracranial bleeding. Repeat imaging showed that the aneurysm was well embolized or contrast agent retention, smooth blood flow in the main and branch arteries of the parent artery, and good stent morphology. After surgery, patients undergo routine DAPT and hematologic parameter monitoring. Postoperatively, patients are transferred to the Neuro-Intensive Care Unit (NICU) for 1 day of intensive monitoring and then transferred to a regular ward. Follow-up imaging examinations are promptly conducted postoperatively to evaluate for hemorrhage and infarction.

Cerebrovascular infarction events are defined as neurological symptoms occurring within 7 days after surgery, such as:

Or, based on CT or MRI findings suggestive of thrombotic events. CT scans are used to rule out intracerebral hemorrhage (ICH), or CT angiography shows enhanced intracranial arteries reflecting intracranial arterial infarction (lasting more than 24 h). MRI findings include normal signal intensity on diffusion-weighted imaging (DWI, b = 0) and T2-weighted imaging (T2WI), and high signal intensity on DWI (b = 1,000), indicating fresh infarct lesions.

In our center, we use LTA to measure the PAR. Specifically, platelet-rich plasma (PRP) is placed in a cuvette, and various agonists (such as ADP, adrenaline, collagen, thrombin, arachidonic acid, TXA2, PAF, etc.) are added. The mixture is stirred with a silicon-coated magnetic bead, causing gradual platelet aggregation, resulting in decreased plasma turbidity and increased light transmission. This change is recorded to generate a dynamic curve of platelet aggregation. The aggregation rate and light transmission of PRP are defined as 0%, while the aggregation rate and light transmission of platelet-poor plasma (PPP) are defined as 100%. Platelet aggregation is automatically measured, recorded, and plotted using a platelet aggregometer. The PAR is determined with reference to 100 and 0% aggregation rates. After adding agonists to the sample, the proportion of change within 120 s compared to the reference system is recorded as the 120-s aggregation rate. Generally, the maximum change within 5 min compared to the reference system is considered the maximum aggregation rate for the sample. PAR measurements are conducted within 24 h before surgery for preoperative samples and within 3 days after surgery for postoperative samples.

Using a human CYP2C19 gene typing test kit, we identify mutations at the CYP2C19 allele loci CYP2C192 (681G → A, rs4244285), CYP2C193 (636G → A, rs4986893), and CYP2C19*17(−806C → T, rs12248560). Based on the combination of mutation alleles at these loci, patients are categorized into different CYP2C19 enzyme metabolic types: poor metabolizers (*2/*2, *2/*3, or *3/*3) and intermediate metabolizers (*1/*2 or *1/*3).

Patients’ baseline clinical data include age, gender, hypertension, diabetes, smoking status, alcohol consumption, aneurysm neck size, aneurysm dome diameter, location, platelet laboratory parameters, and lipid laboratory parameters. These data are obtained from our center’s database. Hypertension is defined as a history of systolic blood pressure >140 mmHg or diastolic blood pressure >90 mmHg on previous examinations. Hyperlipidemia is defined as a history of diagnosed hyperlipidemia, plasma total cholesterol concentration >5.17 mmol/L, or plasma triglyceride concentration >2.3 mmol/L. Smoking and alcohol consumption are defined as a history of smoking/alcohol consumption for more than 1 year and not quitting. Aneurysm size includes measurements of the aneurysm neck and body, and aneurysm location is categorized as anterior circulation of the brain, internal carotid artery, middle and posterior cerebral artery, or posterior circulation of the brain. Aneurysm measurements are obtained from imaging studies following DSA.

We evaluated the normality of sample distributions using the Shapiro–Wilk test. Normally distributed continuous data were reported as mean (±SD) and analyzed using the t-test. Non-normally distributed continuous data were expressed as median and interquartile range, and analyzed using the Mann–Whitney U test. Count data were presented as percentages and analyzed using the chi-square test and Fisher’s exact tests. Logistic and Cox regression analyses were employed to analyze independent related factors. The statistical significance level was defined as p < 0.05 (*), p < 0.01 (**), p < 0.001 (***), p < 0.0001 (****). Data processing was conducted using SPSS V26.0 and R.

We compared the clinical baseline data tables of the control group (96 cases) and the experimental group (363 cases) (Tables 1, 2). Based on our data comparison, there were no significant differences in patient gender, age, underlying diseases, etc., between the two groups regarding the occurrence of stroke events. However, a statistically significant difference in the indicator of PAR was observed between the two groups regarding the occurrence of stroke events. Additionally, we compared the proportions of perioperative stroke events between the control and experimental groups (9.38% vs. 4.96%), and found improved outcomes in the latter. Our clinical practice and data analysis indicate that guiding perioperative DAPT for UIAs assisted by stent intervention based on PAR testing can improve patient outcomes.

We aim to construct a model to predict the incidence of thrombosis in patients with UIAs undergoing embolization therapy, based on the collected clinical baseline data. Recognizing that various clinical indicators have different impacts on patient prognosis, we employed Lasso regression to select the most clinically relevant indicators. Initially, we used data from the experimental group to screen clinical baseline indicators, followed by Lasso analysis. An appropriate λ value was then selected based on cross-validation curves to determine the final model. Ultimately, we identified the following clinically relevant indicators strongly associated with thrombotic events: smoking, hypertension, hyperlipidemia (HLP), prothrombin time (PT), prothrombin activity (PTA), international normalized ratio (INR) in preoperative coagulation function tests, postoperative platelet count (PLT), D2 dimer (DD), triglycerides (TG), high-density lipoprotein (HDL), preoperative 120 s platelet aggregation rate (Pre 120 s PAR), maximum aggregation rate (Pre Max PAR), postoperative 120 s platelet aggregation rate (Post 120 s PAR), maximum aggregation rate (Post Max PAR), and maximum platelet aggregation rate induced by arachidonic acid (Post AA MAR) (Figure 1A).

The same screening process was subsequently carried out in the control group (Figure 1B). Based on the optimal λ value obtained, the final model revealed gender and postoperative fibrinogen (Post FIB) as clinically relevant indicators strongly associated with TEs in the control group.

Notably, by combining the results of Lasso regression from both the control group (Figure 1A) and the experimental group (Figure 1B), we included the PAR index for factor analysis. A significant correlation between PAR and the occurrence of TEs was discovered, in addition to the basic clinical baseline indicators. Eventually, five PAR indices were selected.

To further investigate these five PAR indices, logistic regression analysis was used to fit the filtered clinical baseline data, including: Clinical for the control group; Clinical, Clinical + PAR, and PAR for the experimental group. These represent four different fitting curves with varying variables. Here, PAR represents the fitting result of the five selected platelet aggregation indices, while Clinical represents the fitting result of other clinical indicators. We then calculated the ROC values for these four fitting curves. Comparison revealed that in the experimental group, Clinical + PAR had the largest area under the curve (AUC = 0.944, Figure 2), indicating the highest predictive performance for thrombotic events during the perioperative period. Additionally, the AUC for PAR in the experimental group was higher than that for Clinical alone (0.901 vs. 0.790, Figure 2), demonstrating significant improvement in predictive effectiveness for TEs when PAR indices were added to the clinical baseline data.

Subsequently, based on data from the experimental group, we calculated the optimal cutoff values for the ROC curves of each index in PAR, ultimately, determining the clinical threshold values for platelet aggregation. Notably, the AUC value for Pre MAR was significantly higher at 0.871 (Figure 3) compared to the other indices: AUC for Pre 120 s PAR was 0.584; Post 120 s PAR was 0.502 (Figure 3); Post MAR was 0.690 (Figure 3); and postoperative maximum aggregation rate induced by arachidonic acid was 0.719 (Figure 3). This indicates that utilizing the Pre MAR index for predicting TEs is more accurate. The optimal cutoff value corresponding to Pre MAR in the ROC was determined to be 19.81. To improve patient prognosis, we recommend DAPT during the perioperative period to ensure that the maximum preoperative PAR is below this threshold value.

To better predict the occurrence of perioperative ischemic events in patients using clinical indicators, we constructed nomograms based on the clinical indicators selected by Lasso regression in the two groups (Figures 4A,B), along with corresponding calibration curves (Figures 4C,D). The Z-test yielded a p-value of 0.486, indicating satisfactory alignment between predicted values and actual values in the experimental group, with a correlation coefficient (Dxy) of 0.906, suggesting satisfactory fit for both curves. Drawing from decision curve analysis (DCA), we noted that the net benefit rate (NB) of the experimental group was significantly higher than that of the control group within the threshold probability range of 0–1 (Figure 4E). Subsequently, we validated the accuracy of the nomogram model in the experimental group using predictive data (Table 3), and the AUC of the ROC results indicated that the model had good prognostic ability (AUC = 0.655, Figure 5). Therefore, we believe that using PAR indicators to guide DAPT during the perioperative period of SAE for UIAs can help reduce the occurrence of perioperative ischemic events, which is clinically significant.

This study has several limitations. First, it is a single-center retrospective study, and the data may be subject to bias. Second, the study mainly focused on ischemic events and did not consider other hemorrhagic complications, apart from intracranial hemorrhage, which may introduce bias when assessing the benefits of antiplatelet therapy. Furthermore, the study did not perform post-thromboembolic event PAR re-testing, so the dynamic changes in platelet aggregation after complications could not be assessed. Additionally, the exclusion of PPI from the study may introduce bias. Future research should continue to collect clinical data from our center and further investigate these issues.