Study group and screening criteria
We retrospectively reviewed the patients with RIAs in Beijing Tiantan Hospital and Beijing and Peking University International Hospital from December 2020 to September 2021. As is presented in Fig. 1, patients were selected according to the following standards: (1) a CTA (computational tomography angiogram) was performed after initial hemorrhage; (2) the patients were sent to our institution within 12 h as soon as the occurrence of symptoms, e.g., when an acute headache or a sudden coma occurred; (3) clinical records were complete, or clinical history could be traced. We further excluded the patients who (1) had other intracranial tumors, angiostenosis or angio-malformation, e.g., meningioma or arteriovenous malformation; (2) had a family history of intracranial aneurysm; (3) had multiple intracranial aneurysms, because it would make it difficult to identify the source of bleeding or rebleeding; (4) had dissection, thrombus or traumatic aneurysms; (5) received special treatment for RIA in other medical institutions before admitted to our institution.
Perioperative management
Once patients were admitted to the medical institutions, they would receive standard medical care, including acute pressure lowering and intensive medical monitoring, followed by the guideline and recommendation [16, 17]. The target of blood pressure was to lower their systolic pressure to 120–140 mmHg.
After initial hemorrhage, patients with Hunt-Hess I–II would receive a surgical intervention within 72 h, who could not benefit from immediate surgical treatment; therefore, we would just give these patients conservative treatment until the Hunt-Hess grade “degraded”; however, once patients’ condition progressively deteriorated, an emergency surgical intervention would be performed. In addition, for patients with Hunt-Hess V, surgical treatment would not be recommended.
Data collection
Clinical information was collected from electronical medical records regarding to age, gender, comorbidities (including hypertension, dyslipidemia, diabetes mellitus, coronary heart disease, and ischemic stroke), aneurysmal subarachnoid hemorrhage history, Hunt-Hess grade at admission, time from admission to rebleeding (time from admission to neurological symptoms) or surgery as well as blood pressure at admission and before rebleeding/surgery. The time interval from admission to rebleeding or surgery was recorded. Modified Fisher scale (mFS) and IA site were collected based on radiological data.
The digital imaging and communications on medicine data of CTA performed after admission were collected and converted into reordered slices (about 0.5 mm per slice). The DICOM data was introduced into Mimics 17.0 (Mimics Research 17.0, Materialize, Belgium) and reconstructed for further studies.
The measurement of morphological parameters was performed by the same neurosurgeons (PJ.J. and J.W.) based on the vascular model. The discrepancies were solved by consulting a senior neuro-interventionist (HW.H.). Aneurysm size (S), diameter of dome (D), perpendicular height (H), diameter of parent artery, vessel angle (VA), aneurysm inclination angle (AA), volume, and surface area were measured here [18]. The parameters mentioned were measured twice by two investigators, the average of whose measurements was taken for further analyses. Aspect ratio (AR), size ratio (SR), undulation index (UI), and nonsphericity index (NSI) were calculated [19]. An irregular shape was identified according to our previous study [15].
The nomogram points were calculated based on hypertension, AR and bifurcation site, which were then transferred into risk probability or rebleeding according to our previous study [15]. The risk probability of rebleeding was used for further analyses.
Identification of rebleeding after admission
In this study, rebleeding was diagnosed based on two aspects: (1) the patients had a sudden disorder of consciousness, a gradually worsening neurological state or convulsion after admission, and (2) the magnitude of subarachnoid, intracerebral or intraventricular blood significantly increased in immediate CTs compared with that at admission, and the magnitude of bleeding did not increase or kept stable at/before admission. Rebleeding was confirmed by experienced neurosurgeons (S.W.) according to the bleeding presentation on medical record and CT after admission.
The patients suffering from rebleeding were categorized as the rebleeding group, otherwise the stable group.
Statistical analysis and model establishment
Categorical variables were presented as numbers (n) and percentage (%). Continuous variables with a normal distribution were presented as means and standard deviation, and medians (m) as well as inter-quartile range (IQR) if possible. We compared the differences between continuous variables of the two groups based on Student’s t tests or Wilcoxon rank sum tests as well as the differences in categorical variables based on chi-square tests or Fisher’s exact tests. The clinical + morphological nomogram was given in our previous study [15]; the risk of rebleeding after admission was calculated, meanwhile PHASES score and ELAPSS score were calculated according to previous protocols [20, 21]. The parameters with significance in univariable analysis were input into Cox regression model to identify the independent risk factors, whose result was presented as hazard ratio (HR) and a confidence interval of 95% (CI). A two-tailed P < 0.05 was considered statistically significant, the SPSS 24.0 (SPSS, Chicago, IL) was adopted for statistical analyses, and a two-sided P < 0.05 was considered as statistical significance.
A calibration curve was derived to assess the calibration of the actual rebleeding percentage through the model. The nomogram was subjected to bootstrapping validation (1000 bootstrap resamples). The predictive accuracy of the model while predicting rebleeding was measured by the area under the curve (AUCs) through receiver operating characteristic curve (ROC) analyses. The models with AUC > 0.7 were considered as useful models for clinical work.