The Siriraj Institutional Review Board approved the research protocol, which conformed to the ethical guidelines of the 1975 Helsinki Declaration (Si 465/2011). The EUS database at a tertiary care center from 2010 to 2015 was retrospectively reviewed. Patients with abdominal lymphadenopathy of unknown etiology who underwent EUS-FNA were identified. The inclusion criteria were as follows: (1) enlarged abdominal lymph nodes detected with MRI or abdominal CT in the absence of an identifiable primary lesion and (2) lymph nodes that can be accessed for EUS-FNA based on abdominal imaging. The exclusion criteria were as follows: (1) superficial lymphadenopathy, (2) known history of malignancy, (3) known history of tuberculosis, (4) definite diagnosis determined using modalities other than EUS-FNA, (5) coagulopathy (international normalized ratio >1.5), and (6) thrombocytopenia (platelet count <50,000 µL).

All EUS procedures were performed with a linear echoendoscope (GF-UC140P-AL5; Olympus Optical Co. Ltd., Tokyo, Japan), with the patient under sedation with propofol, by an experienced endosonographer who had performed >2,000 EUS procedures. Aspiration was performed using a 22-gauge FNA needle (EZ shot; Olympus Medical, Tokyo, Japan and EchoTip; Cook Medical, IN, USA). The enlarged lymph nodes were assessed under three scanning scope positions, as follows: (1) gastric station, to assess the celiac, peripancreatic (body and tail), and periportal nodes; (2) duodenal bulb station, to evaluate the peripancreatic (head) and periportal nodes; and (3) second part of the duodenum, to examine lymph nodes near the uncinate process, mesenteric vessels, and inferior vena cava. Needle puncture was performed through the stomach for position (1) and through the duodenum for positions (2) and (3). The technique involved localizing the target lymph node, Doppler assessment of intervening vessels, needle puncture, tissue aspiration, and specimen collection and processing. Under ultrasound guidance, the lymph node was punctured after the Doppler evaluation confirmed the lack of intervening vessels. The stylet was removed, and negative pressure was applied with 5 mL of suction, followed by moving the needle to and fro within the lymph node using the fanning technique. The aspirated material was expelled onto glass slides for visual inspection, and the specimen from each pass was prepared for cytology and cell block assessment. If inadequate macroscopic material was observed, repeated passes were performed (up to four passes). Rapid on-site evaluation was not performed.

The specimens from each pass were divided for cytology and cell block preparation. The aspirated specimens were expelled on a glass slide, smeared for cytologic assessment, and immediately fixed with 95% ethyl alcohol solution. The alcohol-fixed slides were stained with Papanicolaou stain for cytologic examination. The remaining material was placed in a bottle containing formalin solution and embedded in paraffin for cell block assessment. The slides were stained with hematoxylin and eosin and reviewed for histologic features. Additional staining, tissue cultures, and immunohistochemical (IHC) studies were performed when necessary.

The demographics, clinical presentations, laboratory results, radiologic findings, EUS findings, cytopathologic results, and clinical course of the study participants were collected. Information on lymph node characteristics on EUS examination, such as size, shape, echogenicity, border, and location, was obtained. The number of needle passes was recorded. The definite diagnosis was determined on the basis of the cytologic or histologic results of EUS-FNA specimens, clinical course, laboratory findings, and results of radiologic follow-up for at least 12 months.

The final diagnosis was determined on the basis of the (1) histologic findings of either the surgical specimen or cell blocks, (2) cytologic findings, (3) IHC staining of cell blocks, and (4) clinical course and imaging findings during a minimum follow-up of 12 months. A diagnosis of malignancy was made using the following criteria: (1) surgical histology results suggesting malignancy, (2) cytologic or histologic evidence of malignancy based on EUS-FNA specimens, or (3) cytopathologic results suspicious for malignancy combined with the clinical course and follow-up imaging indicating malignancy. Lymphoma was diagnosed on the basis of histologic findings and IHC staining. The diagnosis of tuberculosis was made on the basis of histologic findings of caseating granuloma or positive acid-fast bacilli (AFB) staining of the cell blocks. The diagnosis of other benign conditions required negative cytopathologic assessment and at least 12 months of clinical and imaging follow-up demonstrating no progression or resolution of the enlarged lymph nodes.

The data were analyzed using the Statistical Product and Service Solution (SPSS) statistics software (version 16; SPSS Inc., Chicago, IL, USA). Descriptive statistics were used for demographic, clinical, and laboratory data. Continuous variables were summarized as mean±standard deviation, and categorical variables were summarized as percentages. Qualitative variables were summarized as counts and percentages. Comparisons between groups were performed using the χ² test for categorical variables and the t-test for continuous variables. The predictive ability of EUS-FNA was analyzed by calculating the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) with 95% confidence intervals (CIs). The area under the receiver operating characteristic curve (AUROC) was estimated to evaluate the overall accuracy of EUS-FNA. Logistic regression analysis was used to determine the factors associated with the presence of malignancy or tuberculosis. The data are presented as odds ratios (ORs) with 95% CIs. Statistical significance was set at p<0.05.

A total of 803 patients were identified from the EUS database, with 370 patients undergoing EUS-FNA for intra-abdominal solid mass lesions, during the study period. Of them, 328 patients were excluded because of having solid pancreatic masses, pancreatic cystic lesions, subepithelial lesions, periampullary masses, liver masses, and mediastinal masses. Forty-two patients underwent EUS-FNA for isolated abdominal lymphadenopathy of unknown etiology and fulfilled the eligibility criteria. The baseline characteristics of the patients are shown in Table 1. The mean age was 52.7±15.8 years, and 60% were men. A vast majority of the patients had at least one symptom, and only five patients (12%) were asymptomatic. The presenting symptoms included abdominal pain or discomfort (57%), relevant weight loss (26%), fever (17%), anemia (10%), jaundice (7%), and a palpable mass (5%). Most patients (90%) had multiple lymph nodes.

Adequate tissue specimens were obtained from 40 patients (95%), all of whom received a definite diagnosis. The final diagnoses included metastatic cancer (n=16), lymphoma (n=9), tuberculosis (n=8), inflammatory changes (n=6), and amyloidosis (n=1), as shown in Figure 1. Most patients with metastatic cancer had adenocarcinoma of unknown primary site, and the remaining patients had primary pancreatic cancer (n=3), ampullary cancer (n=1), gallbladder cancer (n=1), gastric cancer (n=1), neuroendocrine tumor (n=1), and sarcoma (n=1). Additional IHC staining was performed in 15 cases (35.7%), with a final diagnosis of lymphoma (n=9), metastatic adenocarcinoma (n=2), neuroendocrine tumor (n=1), sarcoma (n=1), amyloidosis (n=1), and reactive lymphadenopathy (n=1). Among the two patients with inadequate specimens, one patient had positive AFB staining of the aspirated material and was treated for tuberculosis, which reduced the size of the enlarged lymph nodes. The other patient was diagnosed with a benign condition based on the stable clinical course and imaging findings after a 1-year follow-up without progression of the lymphadenopathy. Logistic regression analysis was performed to evaluate the association between clinical parameters and diagnoses. In univariate analysis, age ≥55 years was associated with malignancy (OR, 10.0; 95% CI, 2.06 to 48.6; p=0.002) and fever was associated with tuberculosis (OR, 8.3; 95% CI, 1.33 to 52.0; p=0.013).

The number, location, size, shape, border, and echogenicity of lymph nodes were evaluated. Three patients (7%) had six nodes. Thirty-five patients (83%) had to two to five nodes, and four patients (9.5%) had a single node. The most common locations of the targeted lymph nodes included the peripancreatic (57%), celiac (38%), and aortocaval regions. (24%). The mean size of the lymph nodes was 24.6±13.2 mm (range 5-70 mm) in the long axis. Of the visualized lymph nodes, 31 were >20 mm, 15 were 16-20 mm, 14 were 11-15 mm, and 7 were <10 mm in size.

The lymph node shape was described as round, oval, or irregular. With respect to the distribution of the lymph node shape, 24 (42%) were round, 14 (25%) were oval, and 18 (32%) were irregular. The lymph node border was reported as regular or irregular. Twenty-three nodes (53.5%) had regular borders, whereas 20 (46.5%) had irregular borders. The echogenic pattern was described as hypoechoic, hyperechoic, or isoechoic Thirty-six lymph nodes (83.7%) were hypoechogenic, six (13.9%) were hyperechoic, and one (2.3%) was isoechoic.

Logistic regression analysis was performed to assess the association between endosonographic features and diagnoses. Univariate analysis indicated that an oval or irregular shape was strongly associated with tuberculosis or inflammatory changes rather than malignancy (OR, 6.0; 95% CI, 1.08 to 33.3; p=0.029). Lymphadenopathy in the perihepatic area was correlated with tuberculosis (OR, 11.0; 95% CI, 1.75 to 69.1; p=0.005). Moreover, lymph nodes with regular borders were more frequently observed in the lymphoma group than in the non-lymphoma group (OR, 12.6; 95% CI, 1.35 to 117.6; p=0.010), as shown in Figure 2. In addition, all patients diagnosed with tuberculosis had pus material in the aspirates, as shown in Figure 3. Although lymph node echogenicity could not predict the diagnosis, the results showed that all lymph nodes observed in the lymphoma group were hypoechoic. The mean number of needle passes was 2.1±1.5. No complications occurred.

Furthermore, we evaluated the correlation between EUS and CT in evaluating the shape of the lymph nodes. We found a 48% correlation between the two modalities, with irregular shape being the most concordant interpretation.

EUS-FNA showed a respectable overall diagnostic performance for undiagnosed abdominal lymphadenopathy, with an AUROC, sensitivity, specificity, PPV, and NPV of 0.85, 86.7%, 83.3%, 96.3%, and 55.6%, respectively, as shown in Table 1. Moreover, the diagnostic accuracy of EUS-FNA for each diagnosis was assessed. The results showed that EUS-FNA had an excellent accuracy in detecting lymphoma, with a sensitivity and specificity of 100% (95% CI, 67.6 to 100) and 100% (95% CI, 87.9 to 100), respectively, when combined with cytologic assessment, histologic evaluation of cell blocks, and IHC staining. The diagnostic sensitivity decreased to 75%, whereas the specificity and PPV remained 100%, for tuberculosis. However, the two patients whose histology did not reveal granuloma were diagnosed with tuberculosis based on positive AFB staining of the aspirates in one patient and positive tissue culture for Mycobacterium tuberculosis in the other patient.