부산대학교 도서관

Highlights • Ultrasonographic features of mediastinal lymph nodes can predict malignancy. • Central hilar structure and heterogeneous echogenicity are the strongest predictors. • Ultrasonographic features may reduce unnecessary repeat endobronchial ultrasounds. • Inconsistency in reported predictive ability suggests more research is needed. • Predictive tools can be developed based on these ultrasonographic features. Abstract Objectives The primary objective of this study is to systematically review all pertinent literature related to the use of ultrasonographic features to predict malignancy in mediastinal lymph nodes seen during endobronchial ultrasound (EBUS) procedures. Materials and Methods Two independent reviewers completed the search and review (PubMed, EMBASE, Medline, and Cochrane databases) of the resulting titles and abstracts. Following full-text screening, thirteen articles met the inclusion criteria. Heterogeneity prevented any meta-analysis, instead a narrative review was completed. Results from each included article are categorized by the following ultrasonographic features: shape, echogenicity, margin status, central necrosis, short axis length, and central hilar structure. Diagnostic tools are also described in detail. Results Absence of a central hilar structure and heterogeneous echogenicity were often associated with malignancy; however, consensus was not achieved amongst the included articles. The remaining ultrasonographic features were not consistently associated with malignancy or benign disease status, suggesting a need for prospective analysis. Four diagnostic tools were also assessed. These tools demonstrate that a combination of ultrasonographic features may accurately predict lymph node malignancy rather than a single feature. Conclusion Analysis of ultrasonographic features may prevent the need for repeat EBUS procedures when initial biopsy results are inconclusive. However, prospective external validation of these features is required to determine their true predictive capability. PROSPERO registration number: CRD42017068468. [ABSTRACT FROM AUTHOR]