Toward a Diagnostic Imaging Algorithm for Undifferentiated Pulsatile Tinnitus

Abstract

Objective Decisions around the diagnostic evaluation for pulsatile tinnitus (PT) remain challenging. We describe the usage patterns and diagnostic accuracy of imaging modalities and propose an evidence-based diagnostic approach for undifferentiated PT. Study Design Retrospective. Setting Single otology/neurotology clinic. Subjects Patients with PT presenting between 2009 and 2020. Main Outcome Measures Sensitivity, specificity, diagnostic yield, and diagnostic accuracy. Results A total of 315 subjects met inclusion criteria (74% female, mean ± SD age = 52 ± 17 years). Subjects were divided into four cohorts based on exam findings: normal (n = 229), venous cohort (n = 34), arterial cohort (n = 16), and outer/middle ear pathology cohort (n = 40). In total, 53% of patients received a nonidiopathic diagnosis for PT. The most common identifiable cause was sigmoid sinus dehiscence (78%) in the venous cohort, carotid stenosis (36%) in the arterial cohort, and glomus tumor (56%) in the outer/middle ear pathology cohort. There was a higher diagnostic rate among patients with positive exam findings compared to those with unrevealing exams (p = 0.04). Imaging studies with the highest diagnostic yield were computed tomography (CT) venography (44%), formal angiography (42%), and magnetic resonance venography (40%); studies with the highest specificity were formal angiography (0.82), CT angiography (0.67), and CT venography (0.67). A diagnostic algorithm is proposed. Conclusions Reaching a diagnosis in patients with PT requires a systematic approach, taking into account both clinical and radiographic information. Physical examination is a key first step for differentiating patients into venous, arterial, and other cohorts to narrow down the likely pathology and determine which radiographic studies have the highest yield and accuracy.