Abstract P582: Accuracy of Transcranial Doppler in Detection of Intracranial Stenosis in Patients With Sickle Cell Disease

Abstract

Background: Stroke is among the most devastating consequences of sickle cell disease (SCD). Most SCD strokes occur due to intracranial stenosis, that can be detected by increased flow velocities on Transcranial Doppler (TCD). However, increased velocities may also reflect anemia rather than arteriopathy and vessel stenosis. We aimed to assess the accuracy of TCD in detection of intracranial stenosis in SCD. Methods: Pediatric SCD patients with at least one TCD and MRA within one month apart were identified from a retrospectively collected database maintained at our institution from January 2000 to December 2016. Patient demographics, hemoglobin level, transfusion and hydroxyurea status were collected, along with mean flow velocities (mFV) and degree of stenosis from bilateral middle and anterior cerebral, and internal carotid arteries. A mFV of > 200 cm/s and vessel stenosis > 50% were considered abnormal. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were obtained to assess overall accuracy of TCD velocities in relation to vessel stenosis. Multivariate analysis was performed to identify independent factors associated with MRA stenosis. Results: A total of 164 patients were included in the database (median age 12 [IQR 8.9] years, 56% had ischemic strokes) and 64 of them had at least one TCD and MRA one month apart. Of these, 20% had ischemic strokes, 17% had MRA stenosis > 50% and 10% had TCD velocity > 200 cm/s. TCD mFV > 200 cm/s had a high specificity (95%) and NPV (87%) but low sensitivity (29%) and PPV (55%) when compared to MRA stenosis > 50%. As a continuous variable, TCD mFV 137.5 cm/s had the best balance between maximal specificity (77%) and sensitivity (72%). After adjustment for age, hemoglobin level, transfusion status, hydroxyurea, and vessel, for every increase in cm/sec on TCD, there was a 2% increase in the odds of > 50% stenosis on MRA (OR=1.02, 95% CI 1.01-1.03, p<0.001). Conclusion: Our study reports that TCD mFV is a positive predictor of MRA stenosis in SCD, independent of patient characteristics, including hemoglobin. A mFV > 200 cm/s is highly specific but less sensitive to detect stenosis > 50%. Lower mFV cut points may need to be considered for early detection of intracranial stenosis and risk of stroke.