© 2003 American Thoracic Society
Sounding the MediastinumCase Western Reserve University Cleveland, Ohio Why is clinical staging of the mediastinum in lung cancer so difficult? Studies show 24–38% of patients are overstaged, and 6–20% are understaged when clinical stage is compared with pathologic stage (1, 2). The importance of stage is clear. Treatment is stage-dependent with surgery the best curative option for patients with Stage IA and IB or IIA and IIB cancers, whereas multimodality therapy is appropriate treatment for Stage IIIA, and palliative therapy is indicated for Stage IIIB and IV. Therefore, accurate staging is imperative to recommend the correct therapy and convey the change in prognosis associated with increasing stage. Yet clinical staging remains an inaccurate process. The poor performance of computed tomography (CT) (3–6) has prompted the search for alternate approaches in preoperative mediastinal staging of non small-cell lung cancer. Against this background, Fritscher-Ravens and colleagues (7) in this issue of the Journal (pp. 1293–1297) have compared the utility of endoscopic ultrasound with fine needle aspirate (EUS-FNA) to positron emission tomography (PET) and CT in evaluation of mediastinal nodes for lung cancer. This study is unique as it is the first prospective direct comparison of PET and EUS-FNA to define utility and cost-effectiveness. Taking a consecutive cohort of 79 patients referred for thoracic surgery with apparent clinically operable, proven or suspected lung cancer, they compared the sensitivity and specificity of each test to identify operable versus inoperable lung cancer (N2–3 nodes pathologically positive). The authors found that the sensitivity of EUS-FNA and PET was superior to CT (63%, 68% and 43%, respectively). EUS-FNA, however, had a superior specificity compared with both PET and CT (100% versus 72% versus 91% respectively). These results have lead the authors to propose that following CT imaging, EUS-FNA is more effective for mediastinal staging than PET. Let's compare the staging course of 100 patients with clinically operable non–small-cell lung cancer and mediastinal disease prevalence similar to the population studied by Fritscher-Ravens and coworkers, applying their values for these two approaches. If all 100 patients undergo an EUS-FNA of the mediastinum after a CT, 35 patients would have a positive cytologic analysis resulting in a pathologic diagnosis of advanced stage disease. Therefore, the workup would be complete with a single test. In the remaining 65 cases EUS-FNA would be negative. These cases would require further testing because of the high false-negative rate (32%): ultimately, an additional 21 patients would be found to have disease. Clearly a negative EUS-FNA requires further evaluation. Examining the EUS-FNA false negatives in this series is revealing. Only one was due to a EUS-FNA sampling error; the rest had a T4 status or distant metastases that EUS-FNA could not identify. Thus, the high specificity of EUS-FNA to evaluate the mediastinum is both an advantage and a limitation requiring a negative EUS-FNA to be paired with another procedure to evaluate potential metastatic sites. If we were to subject the same 100 patients to PET after a CT, 50% of the scans would be positive and 50% would be negative. Because of the high false-positive rate (25%) and false-negative rate (36%) reported in this series, all 100 patients would require further evaluation (3). Cost modeling supports the combination of EUS-FNA with CT as more cost-effective than PET with CT; the model confirms a recent retrospective cost modeling analysis comparing EUS-FNA, PET, CT, mediastinoscopy and bronchoscopy (8). Not included in the analysis is the further savings from studies not performed in patients diagnosed with advanced disease. The clear advantage of EUS-FNA is to avoid further testing in 35% of the population and convert inpatient surgery for diagnosis to a minimally invasive procedure. Importantly, the cases studied by Fritscher-Ravens and coworkers were not selected to undergo EUS-FNA based on favorable CT findings. All patients had the procedure performed regardless of CT findings. What are the limitations of this study? Surprisingly, an analysis of how bronchoscopic transcarinal or transtracheal needle aspiration or CT guided aspiration of the mediastinum may affect EUS-FNA results was not performed. Bronchoscopic needle aspirates of the mediastinum have up to a 50% diagnostic yield in appropriate cases (9). If nodes with staging and diagnostic information can be sampled bronchoscopically, will EUS-FNA be as useful? Though the lymph nodes available to each of the techniques are anatomically separate, metastases are likely distributed throughout the mediastinum not just those accessible by EUS-FNA. Next, the study was performed in a population with a high prevalence of mediastinal node involvement: 55% versus 30% reported in most studies (3, 10, 11). Studies performed in this population will result in a higher true-positive rate, thereby providing a more reliable positive test as in this series (positive predictive value of 100%). Will EUS-FNA perform as well when studied in a population with lower disease prevalence? Related to this issue is the lower negative predictive value for PET than has been reported in other series (3–6), decreasing its utility and improving the effectiveness of EUS-FNA. This also may be related to a high disease prevalence resulting in a low true-negative rate providing a less reliable test. Finally, does this impact patient survival or quality of life? What should we take away from this study? The study shows that EUS-FNA is a procedure with potential to evaluate mediastinal areas that would otherwise only be accessible by an operative procedure. EUS-FNA has the ability to identify nodes greater than 3–4 mm, aspirate samples from nodes less than 1 cm, and be directed to lymph nodes with probable metastatic involvement identified through their echogenic pattern. Finally, EUS-FNA does not require general anesthesia and can be done as a day procedure. The study's limitations are its single institution focus with results that may not be generally translated if unique expertise is required, and a study population with high disease prevalence. The authors' suggestion to alter our staging approach is intriguing and suggests the following. In a setting of high disease prevalence, consider EUS-FNA as the next step after a CT in clinical staging of the mediastinum for non–small-cell lung cancer. In the setting of low disease prevalence and no suggestion of mediastinal involvement on CT, however, EUS-FNA may not be the most effective next staging test. Further analysis, such as PET, is required. Clearly these suggestions need prospective validation in a multicenter study. Who should perform this procedure? Gastrointestinal endoscopists will not be eager to surrender a procedure and technology that they have pioneered. Pulmonologists interested in using the technology will need to learn this technically challenging procedure in a cooperative educational setting. This is a perfect opportunity for subspecialty organizations such as the American Thoracic Society and American Gastroenterological Association to jointly set standards as demand for interdisciplinary application of endoscopic ultrasound technology grows. FOOTNOTES Conflict of Interest Statement: J.A.K. has no declared conflict of interest. REFERENCES
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
