Introduction

Because of rapid developments in diagnostic imaging modalities such as ultrasound (US), computed tomographic (CT) scan, and positron emission tomography (PET), increasing number of cases of enlarged abdominal lymph nodes of uncertain origin are being detected. When no primary malignant lesion is evident, the differential diagnosis of abdominal lymphadenopathy can be difficult. Until now, in order to make a definite diagnosis, open laparotomy or laparoscopic tissue sampling had to be performed, but these methods are very invasive and are not cost-effective.

Endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) was developed in the 1990s for the diagnosis of peri-luminal lymphadenopathy adjacent to the gastrointestinal tract [1, 2]. Although the specimens obtained by EUS-FNA can usually differentiate between benign and malignant lymphadenopathy, some cases such as malignant lymphoma can be difficult to diagnose and subtype. EUS-FNA material was originally used for cytology alone, but in the last few years EUS-FNA material has also been used for histological review. In the present study, we evaluated performance characteristics of EUS-FNA for the diagnosis of undiagnosed abdominal lymphadenopathy which could not be diagnosed by other modalities.

Patients and methods

EUS-FNA was performed in 1257 patients from January 1997 to December 2007 at the Aichi Cancer Center Hospital, Nagoya. Patients who underwent EUS-FNA for undiagnosed abdominal lymphadenopathy origin were identified from our prospectively maintained database. Table 1 shows the inclusion and exclusion criteria for undiagnosed lymphadenopathy used in the present study. All patients had given written informed consent to undergo EUS-FNA, and repeat consent for analysis in the present study was not taken. The study was approved by the institutional review board of our hospital.

Table 1 Inclusion and exclusion criteria of EUS-FNA for undiagnosed lymphadenopathy
Full size table

EUS protocol

EUS-FNA was performed as previously described, using a 7.5 MHz convex linear array echoendoscope (GF-UCT240, Olympus Optical Co. Ltd., Tokyo, Japan) and a 22-G needle (NA-200H-8022, Olympus) [3–5]. Three standard scanning positions were used for evaluating intra-abdominal lymphadenopathy: (1) Posterior gastric wall to detect celiac axis and upper peri-aortic nodes, (2) Gastric antrum and duodenal bulb to detect enlarged lymph nodes around the pancreatic head and portal vein, and (3) Descending duodenum to detect lymph nodes around the inferior vena cava, superior mesenteric artery, and the uncinate process. Lymph nodes were sampled by the trans-gastric route for position 1, and by the trans-duodenal route for positions 2 and 3. The aspirated material was divided into 2 parts for cytopathologic and histopathologic assessment, respectively. For all 57 patients, the aspirated material was immediately evaluated by a cytopathologist (TK, YY) for rapid on site diagnosis [4, 5]. The aspirated material was later stained using Papanicolaou’s method. For histopathologic evaluation, the material aspirated with a 22-G needle was directly fixed in 10% formalin in a specimen bottle and later embedded in paraffin. The sections were then stained with hematoxylin-eosin (H&E), as for standard tissue sections. If the cause of lymphadenopathy was considered to be hematopoietic disease, additional samples for flow cytometry were also collected. Immunohistochemical (IHC) staining was selected based on the results of the cytology and the H&E stain. When insufficient material was obtained for a definitive pathological diagnosis, other methods such as CT-guided or US-guided biopsy were attempted.

In each case the final diagnosis of benign or malignant disease was based either on pathology or on a follow up of 12 or more months. Malignancy was diagnosed based on the results of EUS-FNA or percutaneous sampling. Cases with non-malignant aspirates were followed up clinically and with serial imaging. If the lymphadenopathy either regressed or disappeared, and the patient remained clinically well for 12 or more months, a benign etiology was confirmed.

Comparing the results of EUS-FNA with the final diagnosis, the sensitivity, specificity and accuracy of EUS-FNA were calculated.

Patient characteristics

There were 57 patients (31 men, 26 women) with mean age of 62 years (range 32–87 years). The average long axis size of the lymph nodes was 25 mm (range 8–81 mm). We aspirated lymph nodes in 37 patients from position 1 (trans-gastric), in 12 patients from position 2 (duodenal bulb), and in 8 patients from position 3 (descending duodenum).

The final diagnosis was malignant lymphadenopathy in 37 patients (65%) and benign lymphadenopathy in 20 patients (35%). The mean (±SD) long axis diameters of the lymph nodes in the benign and malignant groups were 20 ± 10 mm and 27 ± 15 mm, respectively (p = 0.06) (Table 2).

Table 2 Patients’ characteristics
Full size table

The etiologic sub-groups among patients with malignant lymphadenopathy included post-operative abdominal nodal recurrence in 16 patients, lymphoma in 12 patients, metastatic cancer with unknown primary in 4 patients, and malignant nodes due to concomitant cancer in 5 patients (gastric cancer, breast cancer, duodenal cancer, testicular cancer, and ovarian cancer). The etiology of benign abdominal lymphadenopathy included reactive nodes in 17 patients, tuberculosis in 2 patients, and sarcoidosis in 1 patient (Table 3).

Table 3 Final diagnosis of undiagnosed lymphadenopathy (n = 57)
Full size table

Results

Diagnostic accuracy

The average number of needle passes was 2.3 (range 1–6), and adequate specimens were obtained using EUS-FNA in 93% (53 of 57) of the patients. For the 53 patients with adequate samples, the sensitivity, specificity and accuracy of EUS-FNA were 94% (32 of 34 patients), 100% (19 of 19 patients) and 96% (51 of 53 patients), respectively (Table 4). Among the 4 patients with inadequate samples, diffuse large B-cell lymphoma (DLBCL) was diagnosed using US-guided and CT-guided percutaneous aspiration in 2 patients, sarcoidosis was diagnosed using CT-guided percutaneous aspiration in 1 patient, and recurrence of cholangiocarcinoma in 1 patient based on his clinical course.

Table 4 Diagnosis accuracy of EUS-FNA for lymphadenopathy
Full size table

Post-operative abdominal lymph node recurrence

Fourteen of 16 patients (88%) with metastatic lymphadenopathy were diagnosed by EUS-FNA. The primary lesions included 4 cholangiocarcinomas, 3 pancreatic cancers, 2 gastric cancers, 2 cancers of the papilla of Vater, 2 colon cancers, 1 lung cancer, 1 esophageal cancer, and 1 ovarian cancer (Table 5). One cholangiocarcinoma and one gastric cancer could not be diagnosed due to insufficient material. The mean interval of recurrence after surgical resection of the primary tumor was 3.25 years (range 6 months–16 years).

Table 5 Final diagnosis of lymphadenopathy suspicious for cancer recurrence
Full size table

The pathological diagnosis was adenocarcinoma in 12 patients, and one each of squamous cell carcinoma and acinar cell carcinoma in patients with resected esophageal cancer and pancreatic cancer, respectively.

Malignant lymphoma

Ten of 12 patients (83%) with lymphoma were diagnosed by EUS-FNA. On cytology, 4 patients were Class V (malignant lymphoma), 1 patient was Class IV (suspicious of malignancy except ductal carcinoma), 4 patients were Class IIIb (suspicious of malignant lymphoma), 1 patient was Class III (atypia, cannot exclude malignancy), and 1 patient was Class II (reactive change) (Table 6).

Table 6 Malignant lymphoma (n = 12)
Full size table

On IHC studies using CD3, CD5, CD10, CD20 and BCL2 monoclonal antibodies, 5 lymphomas were subtyped as follicular lymphomas and 3 lymphomas were subtyped as diffuse large B-cell lymphomas. The overall diagnostic accuracy of EUS-FNA for diagnosing lymphoma was 83%. Two cases of diffuse large B-cell lymphoma (DLBCL) could not be diagnosed by EUS-FNA due to insufficient material. These were eventually diagnosed using another modality (US or CT).

Benign lymphadenopathy

A total of 20 patients had benign lymphadenopathy. Seventeen patients (85%) had reactive lymphadenopathy, 2 had tuberculosis, and 1 had sarcoidosis. The median follow-up period was 6.9 months (range 6.1–71 months). Tuberculosis was diagnosed by polymerase chain reaction, and sarcoidosis was diagnosed on histopathology.

Complications

Procedure-related complication occurred in only one patient (2%). This patient required a total of 10 days hospitalization due to submucosal hemorrhage, but no blood transfusion or surgical intervention was needed.

Discussion

When lymphadenopathy is detected on imaging modalities such as US, CT or PET, a pathological diagnosis is very important for the patient’s treatment. Catalano et al. reported that lymph node size ≥10 mm, well defined margins, round or oval shape, and low echogenicity were morphologic predictors of malignancy [6]. When an enlarged lymph node fulfils all of these criteria, malignancy can be diagnosed with high accuracy. However, a malignant lymph node does not always fulfill all of these criteria, while benign lymph nodes sometimes fulfill all of them. Thus, in an individual patient, morphology alone is seldom helpful.

Since the first description of EUS-FNA in 1992, the list of lesions that can be targeted by EUS-FNA has grown to include pancreatic mass lesions, submucosal tumors, mediastinal mass lesions, intra-abdominal or intra-thoracic lymph nodes and ascites [7]. The diagnosis of postoperative recurrence of carcinoma is often difficult even with CT, MRI or fluorine-18 fluoro-deoxyglucose positron emission tomography (FDG-PET) follow-up studies [8, 9]. Even though PET is considered to be the better predictive method for local recurrence after operation, its sensitivity and negative prediction value for malignant lymphadenopathy are very low [10]. Dewitt et al. reported that EUS-FNA could diagnose recurrence of carcinoma with intra-abdominal or mediastinal lymphadenopathy [11]. In the present study, EUS-FNA was performed in patients with undiagnosed lymphadenopathy, and the overall accuracy rate was 96%. Moreover, a diagnosis of postoperative recurrence could be made from 6 months to 16 years after the surgery. The effectiveness of this method is based on the fact that EUS can be used to identify and sample lymph nodes a few millimeters in size. EUS-FNA is also a safer modality for tissue sampling compared to CT or US guided aspirations because of its higher spatial resolution and shorter needle tract to the target. Also, interposed vessels can be easily avoided by real time imaging with EUS. Hence, if the target lesions are peri-luminal we choose EUS-FNA as the first modality for obtaining a tissue diagnosis.

Malignant lymphoma includes a variety of subtypes, many of which can be cured by chemotherapy and/or radiation therapy. An accurate identification of the subtype of lymphoma should be made because treatment is subtype specific [12]. Immunohistochemistry is necessary for identifying the specific histologic type of lymphoma. Ribeiro et al. [13] reported that malignant lymphoma could be diagnosed by cytology and flow cytometry with an accuracy rate of 81%. In the present study, the diagnostic accuracy rate of cytology for malignant lymphoma was 83%. Usually it is believed that histology is a more sensitive technique than cytology for obtaining a final diagnosis of several diseases. However, cytology has been reported to be an equal or more sensitive technique than histology for the diagnosis of breast or thyroid cancer [14]. We think that the results of cytology are equal to or better than histology for the following reasons: Firstly, we are able to make a smear and evaluate the sample adequacy for diagnosis on-site. If the EUS-FNA specimen is inadequate, the target can be punctured again. Secondly, FNA can sample a wide area of the target by moving the needle. Thirdly, the FNA cytology material can also be processed as a cell block and evaluated with conventional histology.

The optimal needle size (19 G, 22 G, 25 G) that should be used for FNA cytology is controversial. Yasuda et al. reported that a 19-G needle is enough to obtain an adequate tissue sample for both H&E and IHC staining [15]. They could diagnose 44 of 48 malignant lymphomas sampled. Up to now, a 22-G needle has usually been used for EUS-FNA because the specimens obtained by EUS-FNA with this size of needle are sufficient to make a differential diagnosis between malignant and benign lesions. In the present study, sufficient material was obtained and immune stained with a 22 G needle, so that 10 of 12 malignant lymphoma patients could be diagnosed.

Conclusion

EUS-FNA is very useful for making a final diagnosis of undiagnosed abdominal lymphadenopathy. It can also obtain adequate material for IHC staining using a 22-G needle. The material obtained from EUS-FNA is sufficient for classifying the subtype of malignant lymphoma in a large percentage of patients.