Are there Endoscopic Ultrasound (EUS) criteria that reliably identify malignant adenopathy?

D. O. Faigel, G. G. Ginsberg, S. Fanning, M. L. Kochman

Research output: Contribution to journalArticlepeer-review


EUS criteria to identify malignant nodes (size, round shape, hypoechoic, homogeneous) have been proposed. However, the reliability of these criteria in identifying patients with malignant adenopathy has been challenged, and EUS guided FNA advocated to establish N-stage. Methods: 147 EUS patients (Olympus GFUM20) being evaluated for suspected malignancy in whom nodes were identified were included. EUS data were collected and assessed by a single observer: TNM stage (standard criteria), tumor size, largest node size, number of nodes, proximity of the closest node to the tumor, and on a 5 point subjective scale (1-least malignant 5-most malignant): roundness, echogenicity, homogeneity and the endosonographer's estimate of the likelihood of malignant adenopathy (Likely). A combined criteria index was calculated (Comb= node size+#nodes + l/proximity+roundness+echo+homogeneity) and a logistic regression model was generated based on these characteristics and the pathological N-stage (Model). The patients were followed through surgery or nodal biopsy and associations between EUS findings and pN-stage were assessed (Chi square, Student's T tests) for the entire group and for the subset who had not received preoperative chemo or radiotherapy (C/RT). Results: Pathological N-stage (pN) was available in 69 patients (62 surgical, 7 FNA) with cancers of the esophagus (30), pancreas (19), rectum (10), stomach (7), misc. (3): pN0: 34, pN1: 35. 16 received preoperative C/RT. pN-stage was significantly associated with node size (0.99 vs. 1.33 cm, p=0.05), node size>1.4 cm (p=0.015), Likely (p=0.02) and the Comb index (16.0 vs. 19.0, p=0.03) with a trend towards node number (p=0.09) (n=69). There were no associations with EUS T-stage, tumor size, node proximity, node size>1 cm, roundness, echogenicity, or homogeneity. In the 53 who had not received C/RT, only likelihood (p = 0.05) and Comb (15.2 vs 18.4, p=0.04) were significant. All: SENS SPEC PPV NPV Acc P-value >1.4 cm*37% 88% 76% 42% 62% 0.015 Likely >3 86% 44% 61% 75% 65% 0.01 Likely=5 46% 80% 70% 59% 62% 0.02 Comb>17.5 57% 71% 67% 71% 71% 0.02 Model 69% 65% 69% 68% 68% 0.003 No C/RT: Likely>3 81% 46% 61% 70% 64% 0.03 Likely=5 40% 85% 73% 58% 62% 0.04 Comb>17.5 48% 81% 74% 62% 66% 0.04 Model 70% 63% 67% 65% 66% 0.02*Node size>1.4 cm, Comb>17.5: combined index>17.5, Model: regression model Conclusions: 1. In patients with nodes on EUS there is no single criterion that reliably identifies malignant adenopathy. 2. While endosonographer assessment and combinations of criteria are statistically associated with malignant adenopathy, neither have sufficient accuracy to establish N-stage without biopsy.

Original languageEnglish (US)
Pages (from-to)AB146
JournalGastrointestinal endoscopy
Issue number4
StatePublished - Dec 1 1998

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging
  • Gastroenterology


Dive into the research topics of 'Are there Endoscopic Ultrasound (EUS) criteria that reliably identify malignant adenopathy?'. Together they form a unique fingerprint.

Cite this