Stroke outcome and neuroimaging of intracranial atherosclerosis (SONIA): design of a prospective, multicenter trial of diagnostic tests

Background and Relevance: Intracranial atherosclerosis is responsible for 70,000 ischemic strokes each year in the USA. Noninvasive testingsuch as transcranial Doppler ultrasound (TCD) and magnetic resonance angiography (MRA) to identify intracranial atherosclerosis is in widespread use, but has not been rigorously validated against the gold standard, catheter angiography. The recently NIH-funded Warfarin-Aspirin Symptomatic Intracranial Disease (WASID) trial will compare warfarin with aspirin for stroke prevention in patients with intracranial atherosclerosis. WASID requires performance of angiography along with TCD and MRA, providing an opportunity to critically evaluate these noninvasive tests. Main Objective: The purpose of the Stroke Outcomes and Neuroimaging of Intracranial Atherosclerosis (SONIA) study is to develop the noninvasive diagnosis of intracranial atherosclerosis. The primary aim of SONIA is to define velocity values on TCD and anatomic abnormalities on MRA that identify severe (50–99%) intracranial stenosis of large, proximal arteries seen on catheter angiography. SONIA will define the criteria, or ‘cutpoints’, for an abnormal TCD or MRA and show that they perform with a reliable positive predictive value (PPV). Study Design: SONIA will be conducted in collaboration with WASID. Study-wide cutpoints defining positive TCD and MRA have been developed and reviewed by the site investigators of WASID. Hard copy angiography, TCD and MRA generated in WASID will be centrally read in SONIA. TCD and MRA cutpoints seek to achieve a target PPV of 80% for the identification of severe intracranial stenosis on angiography. Conclusions: Central readings will be used to validate the cutpoints and to develop measures of negative predictive value, and inter- and intra-observer variability. Sensitivity and specificity will be determined after adjustment for verification bias and employed in receiver-operator characteristic analyses. SONIA will use these techniques to develop TCD and MRA cutpoints that minimize the clinical consequences of test errors occurring in the noninvasive evaluation of patients with suspected intracranial atherosclerosis.