The Stalk Sign in Bladder Cancer: CEUS Versus MRI

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Background: The fibrovascular stalk is characteristic of papillary bladder tumors and may be visualized on MRI as the inchworm sign. Contrast-enhanced ultrasound (CEUS) may depict the same structure dynamically by demonstrating pedicle perfusion before tumor enhancement. We hypothesized that CEUS would be more sensitive than the MRI inchworm sign for detecting a histologically confirmed fibrovascular stalk and that stalk presence would be associated with non-muscle-invasive bladder cancer without completely excluding detrusor muscle invasion. Methods: This retrospective study assessed 80 consecutive patients with one focal bladder lesion; 69 patients with technically adequate imaging and an adequate TURBT reference standard were included in the final paired analysis. Dedicated retrospective rereads of anonymized stored CEUS cine loops and complete mpMRI datasets were independently performed by two experienced readers per modality, who were blinded to the other imaging modality, histopathology, and each other’s assessments. A blinded pathologist assessed fibrovascular stalk presence and detrusor muscle invasion. Paired performance was evaluated using exact McNemar testing. Results: Histopathology identified a stalk in 48 lesions. CEUS detected 46 of 48 stalks, and MRI detected 36, corresponding to sensitivities of 95.8% and 75.0%, specificities of 90.5% and 81.0%, and accuracies of 94.2% and 76.8%, respectively. Interobserver agreement was 94.2% for both signs, with almost-perfect agreement for the CEUS vascular stalk sign (κ = 0.86) and the MRI inchworm sign (κ = 0.88). For MIBC detection, overall CEUS impression and VI-RADS 4–5 showed sensitivities of 66.7% and 87.5% (paired p = 0.063), specificities of 88.9% and 68.9% (paired p = 0.004), and accuracies of 81.2% and 75.4% (paired p = 0.424), respectively. Conclusions: However, the two signs are not operationally equivalent because the MRI inchworm sign additionally requires preservation of the underlying muscular layer, a criterion that may have contributed to its lower observed sensitivity. The stalk strongly favored non-muscle-invasive disease but did not exclude detrusor invasion. CEUS provides complementary perfusion information, whereas MRI-based VI-RADS remains central to local staging.