Cytopathologic diagnoses of fine-needle aspirations from endoscopic ultrasound of the mediastinum: reproducibility of the diagnoses and representativeness of aspirates from lymph nodes.

BACKGROUND: Endoscopic ultrasound-guided fine-needle aspiration biopsy through the esophagus (EUS-FNA) or the bronchial tree (endobronchial ultrasound guided transbronchial needle aspiration [EBUS-TBNA]) may be used to obtain specimens from mediastinal structures. The accuracy of this procedure has been well documented. However, no studies have studied the reproducibility of the pathologic assessment of the aspirated material.
METHODS: A total of 102 slides from EUS-FNA or EBUS-TBNA were assessed 2 times by 4 pathologists who classified each slide to 1 of 5 diagnostic categories and judged if the aspirate came from a lymph node. Between the 2 rounds the criteria to be used in the assessment of the slides were reviewed in a limited education session. The 4 observers had at least 15 years of pathology experience, but their experience in EUS-FNA and/or EBUS-TBNA varied from almost none to more than 10 years. The kappa statistic was applied for the analysis of reproducibility.
RESULTS: The reproducibility of the diagnoses in the first round was good to excellent (kappa, 0.52-0.89). The teaching session led to a significant improvement of the reproducibility between the least and the most experienced observers (kappa ranges of 0.52-0.55 in the first round improved to 0.65-0.71 in the second round).
CONCLUSIONS: The reproducibility of the diagnosis on EBUS-TBNA and EUS-FNA is excellent among pathologists experienced with these types of samples. Pathologists who are generally experienced but have little experience with EBUS-TBNA and EUS-FNA show a steep learning curve. From a pathologic point of view, EBUS-TBNA and EUS-FNA are feasible, but only experienced pathologists should do the assessments.