Risk stratification in symptomatic intracranial atherosclerotic disease with conventional vascular risk factors and cerebral haemodynamics

Abstract

Background and purpose: Symptomatic intracranial atherosclerotic stenosis (sICAS) is associated with a considerable risk of recurrent stroke despite contemporarily optimal medical treatment. Severity of luminal stenosis in sICAS and its haemodynamic significance quantified with computational fluid dynamics (CFD) models were associated with the risk of stroke recurrence. We aimed to develop and compare stroke risk prediction nomograms in sICAS, based on vascular risk factors and these metrics.

Methods: Patients with 50%-99% sICAS confirmed in CT angiography (CTA) were enrolled. Conventional vascular risk factors were collected. Severity of luminal stenosis in sICAS was dichotomised as moderate (50%-69%) and severe (70%-99%). Translesional pressure ratio (PR) and wall shear stress ratio (WSSR) were quantified via CTA-based CFD modelling; the haemodynamic status of sICAS was classified as normal (normal PR&WSSR), intermediate (otherwise) and abnormal (abnormal PR&WSSR). All patients received guideline-recommended medical treatment. We developed and compared performance of nomograms composed of these variables and independent predictors identified in multivariate logistic regression, in predicting the primary outcome, recurrent ischaemic stroke in the same territory (SIT) within 1 year.

Results: Among 245 sICAS patients, 20 (8.2%) had SIT. The D₂H₂A nomogram, incorporating diabetes, dyslipidaemia, haemodynamic status of sICAS, hypertension and age ≥50 years, showed good calibration (P for Hosmer-Lemeshow test=0.560) and discrimination (C-statistic 0.73, 95% CI 0.60 to 0.85). It also had better performance in risk reclassification and provided larger net benefits in decision curve analysis, compared with nomograms composed of conventional vascular risk factors only, and plus the severity of luminal stenosis in sICAS. Sensitivity analysis in patients with anterior-circulation sICAS showed similar results.

Conclusions: The D₂H₂A nomogram, incorporating conventional vascular risk factors and the haemodynamic significance of sICAS as assessed in CFD models, could be a useful tool to stratify sICAS patients for the risk of recurrent stroke under contemporarily optimal medical treatment.

Keywords: atherosclerosis; prospective studies; risk factors; stroke.

Figure 1

The D₂H₂A nomogram for SIT within 1 year in sICAS patients receiving optimal medical treatment. To use the D₂H₂A nomogram, for instance, a 60-year-old (3 points) sICAS patient with a history of hypertension (1 point) and intermediate haemodynamic status of sICAS (7 points) would have 11 points in the D₂H₂A nomogram, and hence a 5.0% probability of having SIT within 1 year under contemporarily optimal medical treatment. sICAS, symptomatic intracranial atherosclerotic stenosis; SIT, recurrent ischaemic stroke in the same territory.

Figure 2

Calibration plot of the D₂H₂A nomogram for recurrent ischaemic stroke in the same territory within 1 year.

Figure 3

DCA for the three nomograms. The DCA shows the net benefits (y axis) of nomograms to stratify subjects for the risk of an outcome across a range of threshold probabilities of the outcome (x axis). (A) The reference lines (grey and black dashed lines) assume no patient or all patients will have an SIT. The grey and black solid lines illustrate the net benefits of nomograms I and D₂H₂A in risk stratification of sICAS patients, across a range of threshold probabilities of 1 year SIT (x axis). With an 8% risk of SIT in the study population (red dashed line), nomograms I and D₂H₂A, respectively, yielded a net benefit of 0.021 and 0.031, suggesting that the D₂H₂A nomogram would identify 10 more patients at risk of SIT within 1 year per 1000 patients, compared with nomogram I. (B) Similarly, nomograms II and D₂H₂A, respectively, yielded a net benefit of 0.020 and 0.031, suggesting that the D₂H₂A nomogram would identify 11 more patients at risk of SIT within 1 year per 1000 patients, compared with nomogram II. DCA, decision curve analysis; sICAS, symptomatic intracranial atherosclerotic stenosis; SIT, recurrent ischaemic stroke in the same territory.