Arterial spin labeling-ASPECTS and a conventional MRI-based nomogram for predicting prognosis after surgical revascularization in Moyamoya disease

Abstract

Objective: This study aimed to explore the significant factors of prognosis in patients with Moyamoya disease (MMD) after surgical revascularization and to develop a nomogram model for predicting poor prognosis.

Materials and methods: We retrospectively analyzed magnetic resonance imaging (MRI) and clinical data of 128 patients with MMD. The patients were randomly assigned to training and validation cohorts in a ratio of 7:3. Multivariate logistic regression analysis was applied to identify factors significantly associated with prognosis. The predictive efficiencies of the models were evaluated using receiver operating characteristic (ROC) curves and compared using the Delong test. We then developed a nomogram model for prediction and verified it using a validation cohort.

Results: Preoperative arterial spin labeling (ASL)-Alberta Stroke Program Early computed tomography Score (ASL-ASPECTS), admission modified Rankin scale (mRS) score, ivy sign, and Houkin's grade >2 were significantly associated with poor prognosis (mRS > 2). The areas under the curves (AUCs) for predicting poor prognosis were 0.772, 0.855, 0.899, and 0.994 for clinical, conventional MRI, ASL-based, and combination models, respectively. The results of the Delong test demonstrated the superior prediction ability of the combination model compared with the clinical, conventional MRI, and ASL models (all p < 0.001). Calibration curve analysis showed that the predictive probability of the nomogram model was highly consistent in the training cohort. The decision curve showed a net predictive benefit in the validation cohort.

Conclusion: Preoperative ASL-ASPECTS, admission mRS, ivy sign, and Houkin grade >2 were significantly associated with poor prognosis in patients with MMD after surgical revascularization. The nomogram model, including enrolled ASL-ASPECTS and MRI features, may help improve prognosis prediction.

Keywords: MRI; Moyamoya disease; arterial spin-labeling; nomogram; prediction.

Figure 1

Flowchart illustrating the inclusion and exclusion of eligible patients in this study.

Figure 2

ROC analysis of various models for predicting the risk of poor prognosis in MMD patients after bypass treatment.

Figure 3

Nomogram for predicting the poor prognosis of MMD patients after bypass treatment.

Figure 4

ROC analysis of nomogram for predicting the poor prognosis of MMD patients after bypass treatment in raining (A), validation (B) cohorts.

Figure 5

Calibration curves of t analysis of the nomogram model in the training (A), validation (B) cohorts. Decision curves of the nomogram model in the training (C), validation (D) cohorts.

Figure 6

Representative MMD case with poor prognosis (A–D). Female, 60 years old, with an admission mRS score of 2, and an mRS score of 3 after 1 year. T2-FLAIR showing ivy sign (A); Preoperative ASL-ASPECT score was 4 (B). Houkin score was 6 at 1 week postoperatively, and MRA staging was stage 3 (C); DWI showing acute/subacute infarction in the right caudate nucleus (D). The nomogram score was 130, with a probability of poor prognosis of 0.90.

Figure 7

Representative MMD case with good prognosis (A–D). Male, 50 years old, with an admission mRS score of 1, and the 1-year mRS score was 0. FLAIR image showed no ivy sign (A); A Preoperative ASPECTS score was 7 (B); Postoperative Houkin score was 3 at 1 week, and MRA staging was stage 2 (C); Follow-up DWI showed no high signal (D). The nomogram score was 25, with a probability of poor prognosis < 0.1.