Endoscopic ultrasound (EUS) is widely accepted as being the most sensitive imaging modality for characterizing lesions of the pancreas. The high spatial resolution of EUS compares favorably with traditional cross-sectional imaging modalities such as magnetic resonance imaging and computed tomography scanning, particularly for lesions measuring < 2 cm. EUS provides detailed information regarding vascular involvement, and allows for simultaneous staging. The higher cost of magnetic resonance imaging and the need for intravenous contrast with computed tomography are additional factors behind the expanding use of EUS in patients with solid pancreatic lesions.
The greatest advantage of EUS is the ability to directly sample pancreatic masses with fine-needle aspiration (FNA). Since the first reports of EUS-FNA in 1991, multiple studies have demonstrated the high accuracy and safety profile of this technique. In most academic and community medical centers, EUS-FNA has replaced percutaneous FNA. Pooled data have demonstrated the accuracy of EUS-FNA as approaching 90% for adenocarcinomas and 70% for neuroendocrine tumors. Metastatic tumors to the pancreas and pancreatic lymphoma can also be characterized with high accuracy, particularly with the advent of EUS-guided core needle biopsy techniques.
Several factors have been associated with improved yields with EUS-FNA of the pancreas. Operator experience remains an important contributor, with an established learning curve for training now supported by society guidelines. Technical factors such as needle size also impact the yield. Smaller-gauge needles (25-gauge) appear to be more effective and safer than larger-bore needles (19-gauge). For specific lesions, 25-gauge needles may improve the yield further by minimizing the blood contamination often encountered in highly vascular lesions such as neuroendocrine tumors. Reducing syringe suction during FNA to < 10 mL also appears to improve the ratio of cellularity to blood contamination. A fanning technique (moving the needle in and out at various angles) improves yield by targeting different aspects of the mass.
In recent years, the role of rapid on-site evaluation (ROSE) has also come to the forefront. ROSE offers the advantage of providing immediate feedback to the endosonographer regarding sample adequacy and cytologic diagnosis. Potential downstream benefits of this technique include reducing the number of needle passes required for tissue acquisition (associated with reduced procedure times and safer outcomes) and the ability to provide more rapid clinical information to both patients and referring providers, thus allowing for the earlier implementation of therapy.
Nonetheless, questions remain before considering this service as the “standard of care.” Do the data definitively support the routine use of this service? Is ROSE cost-effective? How do endosonographers view ROSE in clinical practice? There are growing data supporting ROSE.
To the best of my knowledge, Klapman et al were the first to demonstrate that ROSE can increase the diagnostic yield of EUS-FNA. Outcomes of EUS-FNA from 2 university hospital centers, one with and one without ROSE, and performed by the same endoscopist were reviewed and compared. One endoscopist performed all the procedures at both institutions. Patients treated with ROSE had a diagnosis of positive or negative for malignancy more frequently than those who underwent EUS-FNA without ROSE (P = .001) and were less likely to have an unsatisfactory specimen (P = .035) or undergo a repeat procedure (P = .156).
A recent meta-analysis evaluated 34 distinct studies (for a total of 3644 patients) in which EUS-FNA for a solid pancreatic mass was evaluated. The pooled sensitivity and specificity for EUS-FNA of pancreatic ductal adenocarcinoma was 88.6% and 99.3%, respectively. A meta-regression model demonstrated that rapid on-site cytopathology remained a significant determinant of the accuracy of EUS-FNA (P = .001) after correcting for study population number and reference standard.
A study of the cost-effectiveness of ROSE was presented at the recent Digestive Disease Week meeting. Two competing evaluation strategies of patients with solid pancreatic masses undergoing EUS-FNA were assessed. Under one strategy, slides from each EUS-FNA pass were prepared by an on-site cytotechnologist who performed ROSE to assess for adequacy and to make a preliminary diagnosis. In the other strategy, endosonographers performed several passes until it was believed that an adequate sample had been obtained. No ROSE was performed. Under both strategies, repeat EUS was performed in the case of clinically indicated nondiagnostic results. Factors assessed included diagnostic yield with offsite cytopathology, incremental diagnostic yield with ROSE, incremental adequacy of sampling with ROSE, cytotechnician salary, and incremental diagnostic yield of a second EUS-FNA after an initial nondiagnostic procedure. The results indicated that the overall cost per patient was less with ROSE (US$2061) compared with when ROSE was not used (US$2228).
A randomized controlled trial evaluating the usefulness of ROSE was also presented at the Digestive Disease Week 2013 meeting. In 3 US academic medical centers, patients with solid pancreatic masses were randomized to undergo EUS-FNA with or without ROSE. A significantly higher percentage of patients were diagnosed with suspicious/malignant lesions in the ROSE group compared with the group without ROSE (87.9% vs 70.8%; P = .01). Similarly, specimen adequacy (93.9% vs 81.5%; P = .03) and median number of FNA passes (3 passes vs 7 passes; P < .001) favored those patients randomized to ROSE. Procedural time was slightly longer for patients who underwent ROSE (23.4 minutes vs 19.1 minutes; P = .04). However, the cytopathology review time was reduced in the ROSE group (16.4 minutes vs 27.7 minutes; P < .001)
Another recent study has suggested that optimizing laboratory processing may improve yield without ROSE. It is certainly well known that the most cellular specimen is of little use if improper processing (eg, poor smear quality) precludes evaluation by the cytopathologist. Reducing operator variability in making quality smears with a standardized cytocentrifuge technique coupled with routine cell block preparation reduced the nondiagnostic rate of EUS-FNA of solid pancreatic masses from 22.8% to 4.2% (P < .001).
Where do we stand in 2013? The data convincingly demonstrate that ROSE improves EUS-FNA yield for patients with solid pancreatic masses and that this improvement is associated with the need for fewer FNA passes and repeat procedures. The ability to communicate directly with a cytopathologist intraprocedurally may lead to additional sampling, or different tissue acquisition and processing techniques (such as EUS core needle biopsy) in cases in which special studies may be required for diagnosis (eg, patients with lymphoma or others with overlapping morphological features requiring immunohistochemical stains for a specific diagnosis, such as pancreatic neuroendocrine tumor and acinar cell carcinoma). Preliminary data also have suggested ROSE to be potentially cost-effective when overall patient care costs are considered. From the clinician's perspective, all these attributes add up to improved procedural efficiency and better outcomes for patients. Despite these clear advantages, challenges remain for the widespread implementation of ROSE. These issues center on resource use and availability as well as the reimbursement challenges facing providers. The evolving field of telecytopathology over secured networks may provide an avenue toward overcoming some of these shortcomings.
Despite some limitations, ROSE has a proven track record and should be considered the standard of care for EUS-FNA of solid pancreas masses in 2013. Future developments in telecytopathology will strengthen the collaborative approach to these patients between endonosonography and cytopathology, and will likely evolve into the more widespread implementation of this service.