Construction method and application of abnormal smoke vascular retraction prediction model after smoke disease cerebral blood supply reconstruction

Abstract

The invention relates to a construction method and application of an abnormal smog vascular retraction prediction model after smog disease cerebral blood supply reconstruction. The prediction model performs prediction by using the age of a patient, an RNF213 gene mutation type, a Suzuki staging and a preoperative PA score. The prediction model is high in accuracy, high in conformity with clinical results, good in calibration degree and capable of providing a powerful basis for preoperative consultation, operation strategy making and postoperative management of patients.

Description

Technical Field

[0001] The present invention relates to the technical field of postoperative effect prediction, and in particular to a method for constructing and applying a prediction model for abnormal moyamoya vascular retraction after cerebral revascularization surgery for moyamoya disease. Background Art

[0002] Moyamoya disease (MMD) is a chronic, progressive cerebrovascular disease characterized by stenosis or occlusion of intracranial arteries, accompanied by the formation of an abnormal vascular network at the skull base. MMD presents insidious onset but carries high rates of disability and mortality, making it a common cause of stroke in children and young adults in my country.

[0003] Abnormal smoke vessels, also known as periventricular anastomosis (PA), are unique abnormal vascular networks in MMD and have received widespread attention in recent years. Studies have shown that the presence of PA is an important potential cause of cerebral hemorrhage in MMD patients. Currently, the main treatment for MMD is cerebral revascularization, but even after surgical treatment, some patients may still experience postoperative rebleeding. Compared with the initial hemorrhage, postoperative rebleeding often has a worse prognosis and a higher mortality rate. Studies have shown that postoperative rebleeding is related to the failure of abnormally dilated PA to effectively shrink after surgery. However, there is currently no reliable model to predict postoperative PA shrinkage, and there is a lack of imaging indicators to accurately assess patients' risk of long-term cerebral hemorrhage after surgery.

[0004] Therefore, developing a model that can predict PA retraction after surgery in MMD patients is particularly important. This model can not only help clinicians evaluate the effectiveness of postoperative treatment, but also more accurately predict the patient's long-term risk of intracerebral hemorrhage, providing a scientific basis for formulating personalized treatment plans and improving patient prognosis. Summary of the Invention

[0005] The present invention aims to provide a method for establishing a prediction model for abnormal moyamoya vascular retraction after cerebral revascularization surgery for moyamoya disease and its application.

[0006] On the one hand, the present invention provides a prediction model for abnormal moyamoya vessel retraction after cerebral revascularization surgery for moyamoya disease, wherein the prediction model uses the patient's age, RNF213 gene mutation type, Suzuki stage and preoperative PA score for prediction.

[0007] In one embodiment, the prediction model uses preoperative variables for prediction, and the preoperative variables include the patient's age, RNF213 gene mutation type, Suzuki stage, and preoperative PA score.

[0008] In one embodiment, the patient's age range is 0-65 years old.

[0009] In one embodiment, the patient's RNF213 gene mutation type is none of the rare and novel variants, p.R4810K variant, rare or novel variant.

[0010] In one embodiment, the patient's Suzuki stage is stage I-II, stage III-IV, stage V-VI.

[0011] In one embodiment, the patient's preoperative PA score refers to the sum of the scores of each PA dilation. PA includes five types: LSA (lenticulostriate artery), TTA (thalamic tubercle artery), TPA (thalamic perforating artery), AchoA (anterior choroidal artery) and PchoA (posterior choroidal artery). The degree of dilation of each PA has three levels: grade 0 (no dilation or extension), grade 1 (dilated or extended perforating artery develops to the level of the lateral ventricle), and grade 2 (dilated or extended perforating artery reaches above the level of the lateral ventricle). Each PA dilation grade 0 and grade 1 are both 0 points, and each PA dilation grade 2 score is 1 point. In one embodiment, the score range of the preoperative PA score is 0-5.

[0012] On the other hand, the present invention provides a method for predicting abnormal moyamoya vascular retraction after cerebral revascularization surgery for moyamoya disease, the method comprising the step of predicting using the prediction model.

[0013] In one embodiment, the method includes the step of predicting using the patient's age, RNF213 gene mutation type, Suzuki stage, and preoperative PA score.

[0014] In one embodiment, the method includes the step of predicting using preoperative variables, wherein the preoperative variables include the patient's age, RNF213 gene mutation type, Suzuki stage, and preoperative PA score.

[0015] In one embodiment, the patient was diagnosed with moyamoya disease, underwent surgical revascularization surgery, underwent computed tomography perfusion imaging (CTP) examination before surgery, underwent complete digital subtraction angiography (DSA) examination before surgery and 3-12 months after surgery, and performed exon sequencing of the C-terminal region of the RNF213 gene. The patient had no other systemic diseases or conditions and was between 0 and 65 years old.

[0016] In one embodiment, the surgical revascularization procedure includes indirect bypass surgery and combined bypass surgery.

[0017] In one embodiment, the other systemic diseases or conditions include atherosclerosis, autoimmune diseases (such as systemic lupus erythematosus, antiphospholipid antibody syndrome, nodular peripheral arteritis, Sjögren's syndrome), meningitis, neurofibromatosis, intracranial tumors, Down syndrome, head trauma, radiation injury, hyperthyroidism, Turner syndrome, Alagille syndrome, Williams syndrome, Noonan syndrome, Marfan syndrome, tuberous sclerosis, Hirschsprung's disease, glycogen storage disease type I, Prader-Willi syndrome, Wilms tumor, oxalate storage disease, sickle cell anemia, Fanconi anemia, spherocytosis, eosinophilic granuloma, type II fibrinogen deficiency, leptospirosis, pyruvate kinase deficiency, protein deficiency, fibromuscular dysplasia, osteogenesis imperfecta, polycystic kidney disease, oral contraceptives, and drug poisoning.

[0018] In one embodiment, the other systemic diseases or conditions include atherosclerosis, autoimmune diseases, meningitis, neurofibromatosis, intracranial tumors, Down syndrome, head trauma, radiation injury, hyperthyroidism, Turner syndrome, Alagille syndrome, Williams syndrome, Noonan syndrome, Marfan syndrome, tuberous sclerosis, Hirschsprung disease, glycogen storage disease type I, Prader-Willi syndrome, Wilms tumor, oxalate storage disease, sickle cell anemia, Fanconi anemia, spherocytosis, eosinophilic granuloma, type II fibrinogen deficiency, leptospirosis, pyruvate kinase deficiency, protein deficiency, fibromuscular dysplasia, osteogenesis imperfecta, polycystic kidney disease, oral contraceptives, and drug poisoning.

[0019] In one embodiment, the method includes the step of predicting using a nomogram, wherein the variables of the nomogram include the patient's age, RNF213 gene mutation type, Suzuki stage, and preoperative PA score.

[0020] In one embodiment, the nomogram is as follows Figure 3 As shown or as the website ( https: / / mmdparegression.shinyapps.io / dynnomapp / ) As shown (i.e. Figure 4 shown).

[0021] In one embodiment, the nomogram is as follows Figure 3 As shown, the nomogram is to add up the scores of all variables to obtain a total score, and then project the total score onto the probability scale (Predicted probability of PAR) to obtain the predicted probability of abnormal moyamoya vessel retraction after cerebral revascularization surgery for moyamoya disease.

[0022] In one embodiment, the method comprises the steps of:

[0023] (1) The patient's age is entered into the nomogram to obtain an age score;

[0024] (2) The patient's RNF213 gene mutation type is entered into the nomogram to obtain the RNF213 gene mutation type score;

[0025] (3) The patient's Suzuki stage is entered into the nomogram to obtain the Suzuki stage score;

[0026] (4) The patient's preoperative PA score was entered into the nomogram to obtain the preoperative PA score;

[0027] (5) The total score was calculated by adding the patient's age score, RNF213 gene mutation type score, Suzuki stage score, and preoperative PA score;

[0028] (6) The total score was projected onto the probability scale (Predicted probability of PAR) to obtain the predicted probability of abnormal moyamoya vessel retraction after cerebral revascularization surgery for moyamoya disease.

[0029] In one embodiment, the method comprises the steps of:

[0030] (1) The patient's age is entered into the second row of the nomogram. The score corresponding to the first row of the second row of the nomogram is the age score;

[0031] (2) The patient's RNF213 gene mutation type is entered into the third row of the nomogram. The score corresponding to the third row and the first row of the nomogram is the RNF213 gene mutation type score;

[0032] (3) The patient's Suzuki staging is entered into the fourth row of the nomogram. The score corresponding to the fourth row and the first row of the nomogram is the Suzuki staging score;

[0033] (4) The patient's preoperative PA score is entered into the fifth row of the nomogram. The score corresponding to the first row of the fifth row of the nomogram is the preoperative PA score;

[0034] (5) The total score was calculated by adding the patient's age score, RNF213 gene mutation type score, Suzuki stage score, and preoperative PA score;

[0035] (6) The probability corresponding to the total score obtained and the probability scale in the eighth row of the nomogram is the predicted probability of abnormal moyamoya vessel retraction after cerebral revascularization for moyamoya disease.

[0036] In one embodiment, the nomogram is as shown on the website ( https: / / mmdparegression.shinyapps.io / dynnomapp / ) As shown (i.e. Figure 4 The method includes the steps of inputting the patient's age, RNF213 gene mutation type, Suzuki stage and preoperative PA score on the website to obtain the predicted probability of abnormal moyamoya vessel retraction after cerebral revascularization for moyamoya disease.

[0037] In one embodiment, the method includes the step of using the predicted probability to determine whether abnormal moyamoya vascular retraction occurs after cerebral revascularization for moyamoya disease.

[0038] On the other hand, the present invention provides an application of the above-mentioned model for predicting abnormal moyamoya vessel retraction after cerebral revascularization surgery for moyamoya disease in predicting abnormal moyamoya vessel retraction after cerebral revascularization surgery for moyamoya disease.

[0039] Beneficial effects

[0040] The prediction model for abnormal moyamoya vessel retraction after cerebral revascularization surgery for moyamoya disease established in the present invention is based on a nomogram established based on four variables: patient age, RNF213 gene mutation, Suzuki stage, and preoperative PA score. The prediction model is easy to understand and operate, can provide satisfactory prediction results, and can better guide clinicians in stratified management and judgment, providing important support for the development of individualized treatment plans for MMD patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 Schematic diagram of the construction method and application process of the PA retraction prediction model for postoperative MMD patients.

[0042] Figure 2 Forest plot showing the contribution of various clinical factors to postoperative PA retraction.

[0043] Figure 3 A nomogram for predicting PA shrinkage after MMD surgery based on multiple indicators including age, RNF213 gene variation, Suzuki stage, and preoperative PA score.

[0044] Figure 4 This is a dynamic nomogram for the web page.

[0045] Figure 5 Survival curve analysis of the PA retraction group and the non-retraction group.

[0046] Figure 6 The calibration curve of the PA retraction probability of the patient predicted by the established model and the actual patient PA retraction probability.

[0047] Figure 7 Schematic diagram of the decision curve for predicting postoperative PA retraction in MMD patients.

[0048] Figure 8 The ROC curve of the model established to predict PA retraction in patients after surgery. DETAILED DESCRIPTION

[0049] The present invention will be further described below with reference to the following embodiments. The following description is merely a preferred embodiment of the present invention and does not limit the present invention in any other form. Any person skilled in the art may utilize the above disclosed technical content to make equivalent embodiments with equivalent variations. Any simple modification or equivalent variation of the following embodiments made in accordance with the technical essence of the present invention without departing from the content of the present invention shall fall within the scope of protection of the present invention.

[0050] Example 1: Construction of a PA retraction prediction model after MMD surgery

[0051] Abnormal moyamoya vessels, also known as periventricular anastomosis (PA), are a unique abnormal vascular network in moyamoya disease (MMD). Studies have shown that the presence of PA is an important potential cause of cerebral hemorrhage in MMD patients. Currently, the main treatment for MMD is cerebral revascularization, but even after surgical treatment, some patients may still experience postoperative rebleeding. Compared with the initial hemorrhage, postoperative rebleeding often has a worse prognosis and a higher mortality rate. Studies have shown that postoperative rebleeding is related to the failure of abnormally dilated PA to effectively shrink after surgery. This study aims to construct a model to predict PA shrinkage after cerebral revascularization in MMD and to assess the risk of long-term cerebral hemorrhage in patients after surgery.

[0052] Schematic diagram of the construction method and application process of the PA retraction prediction model after MMD surgery Figure 1 The specific steps are as follows:

[0053] 1. Data collection and preprocessing

[0054] This study used patient data from patients with MMD admitted to Beijing Tiantan Hospital, affiliated with Capital Medical University. All patients were diagnosed with MMD and received cerebral revascularization therapy. After the completion of systemic treatment, relevant data were recorded through follow-up visits.

[0055] Inclusion criteria included: 1) a diagnosis of MMD confirmed by digital subtraction angiography (DSA) that met current diagnostic criteria; 2) surgical revascularization (including indirect bypass and combined bypass); 3) preoperative computed tomography perfusion (CTP); 4) complete DSA examinations before and 3-12 months after surgery; and 5) exon sequencing of the C-terminal region of the RNF213 gene. PA retraction analysis was performed primarily on the hemisphere undergoing initial surgery; the hemisphere that did not undergo surgery or underwent subsequent reoperation served as a control group.

[0056] Exclusion criteria: 1) patients with MMD syndrome due to other systemic diseases or conditions that interfere with study results, including atherosclerosis, autoimmune diseases (such as systemic lupus erythematosus, antiphospholipid antibody syndrome, peripheral arteritis nodosa, Sjögren's syndrome), meningitis, neurofibromatosis, intracranial tumors, Down syndrome, head trauma, radiation injury, hyperthyroidism, Turner syndrome, Alagille syndrome, Williams syndrome, Noonan syndrome, Marfan syndrome, tuberous sclerosis, Hirschsprung's disease, glycogen storage disease type I, Prader-Willi syndrome, Wilms' tumor, oxalate deposition disease, sickle cell anemia, Fanconi anemia, spherocytosis, eosinophilic granuloma, type II fibrinogen deficiency, leptospirosis, pyruvate kinase deficiency, protein deficiency, fibromuscular dysplasia, osteogenesis imperfecta, polycystic kidney disease, oral contraceptives, and drug poisoning; 2) patients lost to follow-up or had a clinical follow-up of less than 12 months.

[0057] A total of 305 MMD patients (305 hemispheres) were enrolled according to the above criteria, with a mean follow-up of 45.9 months. Among them, 103 patients (33.8%) experienced PA retraction after surgery, while the remaining 202 patients (66.2%) did not experience PA retraction.

[0058] The study collected patient data including age, sex, clinical presentation, medical history (hypertension, diabetes, hyperlipidemia), admission mRS (Modified Rankin Scale) score, Suzuki stage, collateral circulation, posterior cerebral artery involvement, pre-infarction stage, preoperative PA score, external cervical compensation, surgery type, surgical side, RNF213 genotype, and postoperative Matsushima grade. The primary follow-up outcome was recurrent stroke (including hemorrhage, infarction, and transient ischemic attack). The total cohort was divided into PA retraction and non-PA retraction groups. Clinical information is summarized in Table 1.

[0059] The preoperative PA score refers to the degree of PA dilation. PA dilation is composed of five clusters of dilated or elongated perforating vessels: the LSA (lenticulostriate artery), TTA (thalamic tuberous artery), TPA (thalamic perforating artery), AchoA (anterior choroidal artery), and PchoA (posterior choroidal artery). PA dilation is categorized into three levels: 0 (no dilation or elongation), 1 (dilated or elongated perforating arteries reaching the level of the lateral ventricles), and 2 (dilated or elongated perforating arteries reaching above the level of the lateral ventricles). Each PA dilation of 0 or 1 level is assigned a score of 0, and each PA dilation of 2 levels is assigned a score of 1. The PA score ranges from 0 to 5 by summing the scores for the LSA, TTA, TPA, AchoA, and PchoA. Preoperative PA scoring criteria can also be found in the article "Association of RNF213 Variants With Periventricular Anastomosis in Moyamoya Disease."

[0060] Table 1: Comparison of clinical baseline characteristics between the PA retraction group and the PA non-retraction group

[0061]

[0062]

[0063] SPSS 26.0, R software (version 4.2.3), and GraphPad Prism (version 9.0) were used for data analysis and graphics. Normally distributed quantitative data were expressed as mean ± standard deviation, and intergroup comparisons were performed using the independent sample t-test. Non-normally distributed quantitative data were expressed as median and quartiles, and intergroup comparisons were performed using the Mann-Whitney U test. Enumeration data were expressed as number of cases and percentages, and intergroup comparisons were performed using the chi-square test or Fisher's exact test. Cumulative stroke incidence curves were constructed using the Kaplan-Meier method, and intergroup differences were compared using the log-rank test.

[0064] The RNF213 gene mutation detection method is as follows:

[0065] (1) Whole blood DNA extraction

[0066] a. Add 5.5 ml of distilled water to the lyophilized QIAGEN protease and mix thoroughly. Dissolve the QIAGEN protease and aliquot, then store at -20°C. Add anhydrous ethanol (96-100%) to Buffer AW1 and AW2.

[0067] b. Before starting, return the blood sample to room temperature for 4 hours. Pipette 200 μl of QIAGEN protease into the bottom of a 15 mL centrifuge tube.

[0068] c. Add 2 mL of whole blood and mix thoroughly. If the sample volume is insufficient, add PBS to make the volume 2 mL and then add the sample to the centrifuge tube.

[0069] d. Add 2.4 ml of Buffer AL, invert the tube 15 times to mix thoroughly, then shake vigorously again for at least 1 minute and incubate at 70°C for 20 minutes.

[0070] e. Add 2 ml of absolute ethanol to the sample, invert the tube 10 times to mix, and then shake it vigorously again.

[0071] f. Transfer the solution to a QIAamp Midi column and place it in a 15 ml centrifuge tube, taking care not to wet the rim. Close the lid and centrifuge at 1850 x g (3000 rpm) for 5 minutes.

[0072] g. Remove the QIAamp Midi column, discard the filtrate, and return the QIAamp Midi column to the 15 ml centrifuge tube. Repeat step h. Close the lid and centrifuge again at 1850 x g (3000 rpm) for 3 minutes.

[0073] h. Remove the QIAamp Midi column, discard the filtrate, and place the QIAamp Midi column back into the 15 ml centrifuge tube.

[0074] i. Wet the rim and add 2 ml of Buffer AW1 to the QIAamp Midi column. Close the lid and centrifuge at 4500 x g (4000 rpm) for 10 minutes. Do not discard the filtrate.

[0075] j. Without wetting the rim, add 2 ml of Buffer AW2 to the QIAamp Midi column. Close the lid and centrifuge at 4500 x g (4000 rpm) for 15 minutes. Place the QIAamp Midi column in a clean 15 ml centrifuge tube and discard the collection tube containing the filtrate.

[0076] k. Pipette 150 μl of distilled water, equilibrate to room temperature (15-25°C), and add to the QIAamp Midi. Incubate at room temperature for 5 minutes, centrifuge at 4500 x g (4000 rpm) for 2 minutes, and re-add the eluate to the column for another elution.

[0077] 1. Check DNA concentration. If the DNA sample concentration is lower than 50 ng / ul, re-extract the blood DNA.

[0078] (2) Polymerase Chain Reaction (PCR)

[0079] a. PCR on DNA samples

[0080] PCR amplification reaction system (50ul):

[0081] 2x Prime Star 25ul PCR Forward Primer (10 μM) 1ul PCR Reverse Primer (10 μM) 1ul DNA (sample) 2ul H2O 21ul

[0082] PCR reaction procedure:

[0083] 98℃ 30sec 99℃ 15sec 60℃ 15sec 72℃ 30sec Cycles 40 4℃ save

[0084] b. Perform gel electrophoresis on the PCR products.

[0085] c. Sanger sequencing was performed on the exons encoding the carboxyl terminus of RNF213 protein (NM_001256071, exons 42-68).

[0086] Primer sequences used

[0087]

[0088] 2. Screening independent influencing factors

[0089] To identify independent predictors of postoperative PA retraction, multivariate regression analysis was performed using a stepwise regression method, including all variables with a P value < 0.05 in the univariate analysis. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated. A nomogram was developed based on the final results of the multivariate analysis.

[0090] First, the screening results of univariate analysis are shown in Table 2 , which showed that P < 0.05 for age, clinical phenotype, Suzuku stage, collateral circulation, posterior cerebral involvement, preoperative PA score, external cervical compensation, surgery type, RNF213 genotype, and postoperative Matsushima grade.

[0091] Table 2: Univariate Logistic regression screening of characteristic variables associated with PA retraction

[0092]

[0093]

[0094] Then, the above factors with P < 0.05 obtained from univariate analysis were included in the multivariate logistic regression model (enter stepwise method), and finally the independent influencing factors of postoperative PA retraction were obtained.

[0095] 3. Establishing Logistics regression equation and nomogram prediction model

[0096] Multivariate logistic regression analysis was performed, and the results were presented as a forest plot of the contribution of each clinical factor to postoperative PA retraction. Figure 2 As shown. Figure 2 Age is age, Clinical manifestation is clinical manifestation, Ischemia is ischemia, Hemorrhage is hemorrhage, Suzuki stage is Suzuki stage, Collateral circulation is collateral circulation, PCAinvolvement is posterior cerebral involvement, Pre PAscore is preoperative PA score, ECA collateral is extracervical compensation, Surgery type is surgery type, Genotype of RNF213 is RNF213 genotype, Post Matsushimagrading is postoperative Matsushimagrading, Absence of PAR is lack of PA retraction (PA non-retraction group), Presence of PAR is presence of PA retraction (PA retraction group). Figure 2 It can be seen that age (OR=0.974, 95%CI: 0.952-0.997, P=0.025), Suzuki III-IV stage (OR=3.410, 95%CI: 1.030-11.292, P=0.045), preoperative PA score (OR=2.117, 95%CI: 1.548-2895, P<0.001), RNF213 p.R4810K variant (OR = 2.438, 95% CI: 1.177-5.051, P = 0.016), other rare or novel variants (OR = 8.657, 95% CI: 2.060-36.370, P = 0.003), and good postoperative Matsushima grade (OR = 0.063, 95% CI: 0.016-0.255, P < 0.001) were independent predictors of postoperative PA shrinkage, with P values ​​less than 0.05. Independent factors influencing postoperative PA shrinkage included age, preoperative PA score, RNF213 mutation, Suzuku stage III-IV, and postoperative Matsushima grade.

[0097] According to the results of multivariate logistic regression analysis, a nomogram for efficacy prediction was constructed, such as Figure 3 As shown in FIG, the nomogram for predicting PA retraction after MMD surgery established in this embodiment (since the Matsushima classification is postoperative data, preoperative indicators are not included). The nomogram includes a score scale in the first row, where the score ranges from 0 to 100; the second row is the patient's age ( Figure 3 Age in the first row), different ages correspond to a corresponding score in the first row; the third row is the RNF213 gene variation ( Figure 3 RNF213 variant in the first row), no rare and novel variants, p.R4810K variants and rare or novel variants correspond to a corresponding score in the first row; the fourth row is Suzuki stage ( Figure 3 Suzuki stage in the first row), Suzuki I-II, Suzuki III-IV and Suzuki V-VI stages correspond to a corresponding score in the first row; the fifth row corresponds to the preoperative PA score ( Figure 3 Pre PA score in the first row, 0-5 correspond to a corresponding score in the first row; the sixth row is the total score of the patient ( Figure 3 Total Points in the , add the scores of the four indicators in the second to fifth rows in the first row to get the total score of the patient, and project the total score of the patient in the sixth row to the eighth row ( Figure 3 The probability of PA retraction after surgical treatment is obtained based on the Predicted probability of PAR).

[0098] For example, if a patient meeting the criteria for MMD is admitted, and their clinical baseline, imaging features, and genetic data are obtained, such as the patient's age is 16, RNF213 p.R4810K variant, Suzuki stage III, and preoperative PA score is 4. Based on the prediction nomogram, the second row of the age corresponds to the first row, which is 30 points; the third row of the RNF213 p.R4810K variant corresponds to the first row, which is 19 points; the fourth row of the Suzuki stage III corresponds to the first row, which is 25 points; the fifth row of the preoperative PA score corresponds to the first row, which is 80 points. The total score for this patient is 154 points. Projecting the 154 points from the sixth row to the eighth row indicates that the probability of PA retraction in this MMD patient after revascularization is greater than 90%.

[0099] To facilitate clinical application, the present invention constructs a web version of the dynamic nomogram, such as Figure 4 As shown, users can use the website ( https: / / mmdparegression.shinyapps.io / dynnomapp / ) by inputting the corresponding variable values ​​(age, RNF213 mutation status, preoperative PA score, and Suzuki stage) to obtain the probability of postoperative PA retraction.

[0100] Figure 5 Figure 3 is the survival curve analysis diagram of the PA retraction group and the non-retraction group, where Non-PAR is the PA non-retraction group and PAR is the PA retraction group. Figure 5 A indicates that there was no difference in the total events (bleeding and ischemia) after revascularization for moyamoya disease between the PA retraction group and the non-retraction group (P value greater than 0.05); Figure 5 B shows that there was no difference in ischemic events between the PA retraction group and the non-retraction group after revascularization for moyamoya disease (P value greater than 0.05); Figure 5C indicates a difference in bleeding events after revascularization for moyamoya disease between the PA retraction group and the non-PA retraction group (P value less than 0.05). Overall, the long-term bleeding risk after revascularization for moyamoya disease is higher in the non-PA retraction group than in the PA retraction group. PA is a good imaging marker for predicting long-term bleeding after moyamoya disease surgery.

[0101] 4. Model evaluation and validation

[0102] The performance of the nomogram was evaluated using calibration curves, decision curve analysis (DCA), concordance index (C-index), and area under receiver operating characteristic (AUC-ROC). The internal validity of the model was assessed by calculating the C-index using 1000 bootstrap samples. For each patient, the total score was derived from the nomogram, and the optimal cutoff value was determined by maximizing the Youden index (sensitivity + specificity - 1) using receiver operating characteristic (ROC) curve analysis.

[0103] Figure 6 The figure shows the calibration curve of the model-predicted PA retraction probability and the actual PA retraction probability of the patient. The long dashed line is a straight line with a slope of 45°, which represents the ideal prediction of the patient's PA retraction probability being equal to the actual PA retraction probability of the patient. The short zigzag solid lines are the calibration curves of the model-predicted PA retraction probability and the actual PA retraction probability of the patient after internal validation and calibration. The long solid line that is closer to the 45° long dashed line is the calibration curve of the total model (red curve), and the rest are calibration curves of each single variable (age, RNF213 gene variation, Suzuki stage, and preoperative PA score). It reflects the degree of consistency between the model-predicted risk and the actual risk, so it can also be called consistency. Good calibration indicates that the prediction model is highly accurate, while poor calibration indicates that the model may overestimate or underestimate the risk of disease occurrence. The total model calibration curve constructed by the inventors is consistent with the ideal calibration curve, indicating that the calibration of our model is good and the prediction model is highly accurate. The present inventors evaluated the calibration of the prediction model by using the calibration curve and the Hosmer Lemeshow goodness of fit test statistic. The larger the corresponding P value, the better the calibration of the prediction model. The present inventors obtained the Hosmer Lemeshow test statistic P = 0.668, indicating that the model has good calibration. Figure 6 In the data, Pre PAscore model is the preoperative PA score, Suzuki stage model is the Suzuki stage, Agemodel is the age, and RNF213 variant model is the RNF213 variant.

[0104] Figure 7Schematic diagram of the decision curve for predicting postoperative PA retraction in MMD patients. It can be seen that the overall model (red) shows a good clinical net benefit compared with other univariate models (age, RNF213 gene variation, Suzuki stage and preoperative PA score), indicating that the model has significant clinical effectiveness. This is a method to evaluate the application value of the prediction model in actual clinical decision-making. It is achieved by comparing the net benefits of different decision-making schemes within a specific threshold range. The DCA curve drawn by the inventor shows that the overall model has a good net benefit compared with the single variable model. At the optimal cutoff value, the nomogram showed an accuracy of 81.0%, a sensitivity of 74.4%, a specificity of 94.1%, a positive predictive value of 96.2%, a negative predictive value of 64.9%, a positive likelihood ratio of 12.33, and a negative likelihood ratio of 0.28 (Table 3). Figure 7 In the data, Pre PA score model is the preoperative PA score, Suzuki stage model is the Suzuki stage, Age model is the age, and RNF213 variant model is the RNF213 variant.

[0105] Figure 8 The ROC curve was used to evaluate the discrimination of the efficacy prediction model and calculate the AUC. Figure 8 The calibration curve shown was used to evaluate the probability of the model predicting postoperative PA retraction and the actual PA retraction. Internal validation of the model was achieved using the bootstrap method. This involved randomly selecting 1000 samples with replacement from the original dataset to form a new sample for internal validation. Finally, the diagnostic performance and optimal cutoff values ​​of each nomogram were determined (see Table 3). The ordinate represents the true positive rate (TPR), which is the number of patients with true PA retraction who were also predicted by the model to have PA retraction divided by the total number of patients with true PA retraction. The abscissa represents the false positive rate (FPR), which is the number of patients without PA retraction who were predicted by the model to have PA retraction divided by the total number of patients without PA retraction. The TPR represents the detection rate of patients with PA retraction, i.e., the number of patients with PA retraction who were actually detected. The FPR represents the false positive rate of patients without PA retraction, i.e., the number of patients with PA retraction who were falsely detected. The area under the ROC curve is 0.89. The closer the area under the curve is to 1, the better the model discrimination. The AUC or C index of this model is 0.89 (95% CI: 0.85-0.92), which proves that the model has good discrimination.

[0106] Table 3: Diagnostic performance of the nomogram model

[0107] variable Value (95% CI) Area under the ROC curve, C index 0.89(0.85-0.92) Cutoff value 0.232 Accuracy 0.81(0-76-0.85) Sensitivity 0.74(0.68-0.80) Specificity 0.94(0.90-0.99) Positive predictive value 0.96(0.93-0.99) Negative predictive value 0.65(0 57-0.73) Positive likelihood ratio 12.33(8.00-68.00) Negative likelihood ratio 0.28(0.20-0.36)

[0108] The above data show that the model has good discrimination and calibration, and can accurately predict whether abnormal moyamoya vessels will retract after cerebral revascularization surgery for moyamoya disease, thereby accurately predicting the patient's long-term risk of cerebral hemorrhage, providing a scientific basis for formulating individualized treatment plans, and improving the patient's treatment prognosis.

[0109] Although the specific embodiments of the present invention have been described in detail, those skilled in the art will understand that various modifications and changes can be made to the details based on all the teachings published, and these changes are all within the scope of protection of the present invention. The entire invention is given by the appended claims and any equivalents thereof.

Claims

1. A model for predicting abnormal moyamoya vascular retraction after cerebral revascularization for moyamoya disease, characterized in that: The prediction model uses the patient's age, RNF213 gene mutation type, Suzuki stage and preoperative PA score for prediction.

2. A method for predicting abnormal moyamoya vascular retraction after cerebral revascularization surgery for moyamoya disease, characterized in that: The method comprises the step of performing prediction using the prediction model according to claim 1.

3. The method according to claim 2, characterized in that The patient was diagnosed with moyamoya disease and underwent surgical revascularization surgery. He underwent computed tomography perfusion imaging before the surgery, complete digital subtraction angiography before and 3-12 months after the surgery, and exon sequencing of the C-terminal region of the RNF213 gene. The patient had no other systemic diseases or conditions and was between 0 and 65 years old.

4. The method according to claim 2, characterized in that The method comprises the step of using a nomogram for prediction, wherein the variables of the nomogram include the age of the patient according to claim 1, the RNF213 gene mutation type, the Suzuki stage and the preoperative PA score.

5. The method according to claim 4, characterized in that The nomogram is shown in FIG3 or as shown on the website ( https: / / mmdparegression.shinyapps.io / dynnomapp / ) as shown.

6. The method according to claim 5, characterized in that The nomogram is shown in FIG3 . The nomogram is obtained by adding up the scores of all variables to obtain a total score, and then projecting the total score onto a probability scale (Predicted probability of PAR) to obtain the predicted probability of abnormal moyamoya vessel retraction after cerebral revascularization for moyamoya disease.

7. The method according to claim 6, characterized in that The method comprises the steps of: (1) The patient's age is entered into the second row of the nomogram. The score corresponding to the first row of the second row of the nomogram is the age score; (2) The patient's RNF213 gene mutation type is entered into the third row of the nomogram. The score corresponding to the third row and the first row of the nomogram is the RNF213 gene mutation type score; (3) The patient's Suzuki staging is entered into the fourth row of the nomogram. The score corresponding to the fourth row and the first row of the nomogram is the Suzuki staging score; (4) The patient's preoperative PA score is entered into the fifth row of the nomogram. The score corresponding to the first row of the fifth row of the nomogram is the preoperative PA score; (5) The total score was calculated by adding the patient's age score, RNF213 gene mutation type score, Suzuki stage score, and preoperative PA score; (6) The probability corresponding to the total score obtained and the probability scale in the eighth row of the nomogram is the predicted probability of abnormal moyamoya vessel retraction after cerebral revascularization for moyamoya disease.

8. The method according to claim 5, characterized in that The nomogram is shown on the website ( https: / / mmdparegression.shinyapps.io / dynnomapp / ), the method includes the steps of inputting the patient's age, RNF213 gene mutation type, Suzuki stage and preoperative PA score on the website to obtain the predicted probability of abnormal moyamoya vessel retraction after cerebral revascularization for moyamoya disease.

9. The method according to any one of claims 6 to 8, characterized in that: The method comprises the step of using the predicted probability to determine whether abnormal moyamoya vessel retraction occurs after cerebral revascularization surgery for moyamoya disease.

10. Use of the model for predicting abnormal retraction of moyamoya blood vessels after cerebral revascularization for moyamoya disease according to claim 1 in predicting abnormal retraction of moyamoya blood vessels after cerebral revascularization for moyamoya disease.

Images