[MR-angiography and duplex-ultrasonography: predictive reliability for angiographically determined internal carotid...

A prospective study was undertaken to evaluate whether digital subtraction angiography (DSA) which is still associated with a substantial morbidity can be replaced by less invasive diagnostic modalities such as duplex scanning (DS) and magnetic resonance angiography (MR-A) for the detection of angiographically defined internal carotid artery (ICA) stenosis >/= 70 %.

A total of 47 patients with suspected severe ICA stenosis underwent examination of their carotid arteries using duplex studies, MR-A and DSA. According to the study protocol, the arteriographic diameter reduction (DR) >/= 70 % which had to be predicted by DS and MR-A was determined following the NASCET criteria.

Stroke rate following DSA amounted to 2.1 %. In 94 carotid arteries studied by DSA 34 times a DR >/= 70 % was found. Using ROC curve for determining optimal discriminant value, duplex-derived peak systolic velocity (PSV) >/= 250 cm/s provided a sensitivity of 94.1 %, a specificity of 80 %, a positive predictive value (PPV) of 72.7 % and a negative predictive value (NPV) of 96 % to characterise an ICA stenosis >/= 70 %. Due to an inadequate PPV, PSV failed to suffice as the sole preoperative diagnostic modality even if different PSV velocity cut points were applied. On the other hand, end diastolic velocity (EDV) >/= 150 cm/s provided a PPV of 100 % thereby identifying 16/34 ICA stenoses >/= 70 % in our study. MR-A showed a sensitivity of 91.2 %, a specificity of 88.3 %, a PPV of 81.6 %, and a NPV of 94.6 % to predict an ICA stenosis >/= 70 %.

In our series, both duplex-derived PSV as well as MR-A provided high sensitivity to detect surgically relevant ICA stenosis. However, to select patients for surgery inclusion of EDV proved to be important due to a high PPV and may spare conventional angiography half of patients with stenosis exceeding 70 %.