Supplementary Material for: Predictive Value of Endobronchial Ultrasound Strain Elastography in Mediastinal Lymph Node Staging: The E-Predict Multicenter Study Results

posted on 03.06.2020 by Verhoeven R.L.J., Trisolini R., Leoncini F., Candoli P., Bezzi M., Messi A., Krasnik M., deKorte C.L., Annema J.T., vanderHeijden E.H.F.M.
Background: Systematic assessment of lymph node status by endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) is indicated in (suspected) lung cancer. Sampling is herein guided by nodal size and FDG-PET characteristics. Ultrasound strain elastography (SE) might further improve risk stratification. By imaging tissue deformation over time, SE computes relative tissue strain. In several tissues, a lower strain (deformation) has been associated with a higher likelihood of malignancy. Objectives: To assess if EBUS-SE can independently help predict malignancy, and when combined with size and FDG uptake information. Methods: This multicenter (n = 5 centers) prospective trial included patients with suspected or proven lung cancer using a standardized measurement protocol. Cytopathology combined with surgery or follow-up imaging (>6 months) were used as reference standard. Results: Between June 2016 and July 2018, 327 patients and 525 lymph nodes were included (mean size 12.3 mm, malignancy prevalence 0.48). EBUS-SE had an overall AUC of 0.77. A mean strain <115 (range 0–255) showed 90% sensitivity, 43% specificity, 60% positive predictive value, and 82% negative predictive value. Combining EBUS-SE (<115) with size (<8 mm) and FDG-PET information into a risk stratification algorithm increased the accuracy. Combining size and SE showed that the 48% a priori chance of malignancy changed to 11 and 70% in double negative or positive nodes, respectively. In the subset where FDG-PET was available (n = 370), triple negative and positive nodes went from a 42% a priori chance of malignancy to 9 and 73%, respectively. Conclusions: EBUS-SE can help predict lymph node malignancy and may be useful for risk stratification when combined with size and PET information.