Establishment and external validation of prediction models for risk factors of cerebral hyperperfusion syndrome after combined revascularization in moyamoya disease based on meta-analysis and retrospective cohort studies.

OBJECTIVE

Cerebral hyperperfusion syndrome (CHS) is a complication affecting up to 50% of moyamoya disease (MMD) patients after combined revascularization. The aim of this study was to identify reliable CHS risk factors and develop predictive models.

METHODS

The authors performed a meta-analysis including studies on both combined and direct revascularization to capture potential risk factors, followed by a sensitivity analysis on the combined bypass group. They also performed a retrospective cohort analysis. Then three models were developed using the derivation cohort and externally validated. Significant variables from meta-analysis and logistic regression were used to construct nomograms, the performance of which was evaluated with receiver operating characteristic curves, calibration, and discrimination analyses.

RESULTS

Meta-analysis identified seven significant risk factors: age (effect 2.17, 95% CI 0.28-4.05; p = 0.025), hypertension (risk ratio [RR] 1.42, 95% CI 1.16-1.73; p = 0.001), surgery in the dominant hemisphere (RR 1.80, 95% CI 1.55-2.10; p < 0.001), preoperative hematocrit (effect 2.50, 95% CI 1.58-3.41; p < 0.001), intraoperative high arterial pressure (HAP) (RR 1.78, 95% CI 1.07-2.96; p = 0.026), postoperative white blood cell (WBC) count (effect 1.67, 95% CI 0.46-2.87; p = 0.007), and cerebral blood flow increase rate (effect 0.80, 95% CI 0.40-1.20; p < 0.001). Similar risk factors were confirmed in the retrospective cohort analysis. In a sensitivity analysis focusing on combined revascularization, fewer studies limited inclusion of factors such as temporary occlusion time, intraoperative HAP, postoperative WBC count, and postoperative blood pressure; age and hypertension also lost significance. In the validation cohort, the areas under the curve (AUCs) for model 1 (based on multivariate analysis) and model 2 (based on meta-analysis) were 0.83 (95% CI 0.77-0.90) and 0.84 (95% CI 0.77-0.91), with Youden indices of 0.54 and 0.53, respectively. Model 3 (based on sensitivity analysis) showed lower performance (AUC 0.76, 95% CI 0.66-0.85; Youden index 0.45).

CONCLUSIONS

In the meta-analysis and cohort analysis, the authors identified risk factors for CHS following combined revascularization surgery in patients with MMD, and the nomogram constructed based on factors from meta-analysis demonstrated good predictive performance and clinical utility.