Published ahead of print on October 25, 2007, doi:10.1164/rccm.200708-1241OC
© 2008 American Thoracic Society doi: 10.1164/rccm.200708-1241OC
Endoscopic Ultrasound Reduces Surgical Mediastinal Staging in Lung CancerA Randomized Trial1 Long Oncologisch Netwerk Gent, Ghent University Hospital, Ghent, Belgium; 2 Department of Respiratory Medicine, Amphia Hospital, Breda, The Netherlands; and 3 Department of Medical Statistics, Ghent University Hospital, Ghent, Belgium Correspondence and requests for reprints should be addressed to Kurt Tournoy, M.D., Ph.D., Ghent University Hospital, Department of Respiratory Medicine, Building 7K12 I.E, De Pintelaan 185, 9000 Ghent, Belgium. E-mail: kurt.tournoy{at}ugent.be
Rationale: Assessment of mediastinal lymph nodes is recommended in patients with non–small cell lung cancer without distant metastases. Linear transesophageal endoscopic ultrasound with real-time guided fine-needle aspiration (EUS-FNA) is a promising, nonsurgical tool for mediastinal staging. Objectives: We conducted a randomized controlled trial comparing surgical staging with EUS-FNA. Methods: Patients with proven or suspected non–small cell lung cancer in whom mediastinal exploration was required were randomly assigned to undergo EUS-FNA or the appropriate surgical staging procedure. When EUS-FNA did not show malignant lymph node invasion, a confirmatory surgical staging procedure was done. A negative surgical staging procedure was followed by thoracotomy with systematic lymph node sampling. The primary endpoint was the rate of surgical staging interventions. The secondary endpoints were test performance of EUS-FNA and surgical staging, morbidity, and length of hospital stay, considering surgical staging was performed as an in-patient procedure. Measurements and Main Results: A total of 40 patients were randomized: 19 to EUS-FNA, and 21 to surgical mediastinal staging. Patient and tumor characteristics were well balanced between both groups. For patients allocated to EUS-FNA, surgical staging was needed in 32% (P < 0.001). The sensitivity to detect malignant lymph node invasion was 93% (95% confidence interval, 66–99%) for EUS-FNA and 73% (95% confidence interval, 39–93%) for surgical staging (P = 0.29). Complication rate was 0% for EUS-FNA and 5% for surgical staging (P = 1.0). The median hospital stay was significantly shorter for EUS-FNA than for surgical staging (0 vs. 2 nights; P < 0.001). Conclusions: EUS-FNA reduces the need for surgical staging procedures in patients with (suspected) lung cancer in whom a mediastinal exploration is needed. Clinical trial registered with www.clinicaltrials.gov (NCT 00119470).
Key Words: lung cancer • staging • mediastinoscopy • endoscopic ultrasound with fine-needle aspiration
In patients with non–small cell lung cancer (NSCLC) in whom there is no evidence for extrathoracic metastasis, assessment of the mediastinal lymph nodes for malignancy is of utmost importance (1–4). Mediastinal staging by surgical techniques (mainly cervical mediastinoscopy) is considered the gold standard (5, 6), although the sensitivity to detect malignant lymph node invasion was shown earlier to be only 81% (7). In addition, surgical staging is invasive, requires general anesthesia, and is subject to potential serious complications (8). Linear transesophageal endoscopic ultrasound with real-time guided fine-needle aspiration (EUS-FNA) is a promising, minimally invasive outpatient technique for the diagnosis and staging of mediastinal lymph nodes in patients with lung cancer. Prospective series show a high sensitivity (0.72–1.00) and specificity (1.00) in computed tomography and/or positron emission tomography (PET) scan–selected patients (9), document the advantage of rapid on site evaluation by a cytopathologist (10, 11), and suggest that EUS-FNA could eliminate the need for surgery in a substantial number of patients (12, 13). We conducted a randomized, controlled trial comparing surgical mediastinal staging with EUS-FNA in patients with (suspected) NSCLC requiring mediastinal lymph node exploration. We compared the number of surgical staging procedures, hospital stay, number of lymph node stations assessed, complication rates, and test characteristics of both approaches.
Patients After approval of the study by the Medical Ethics Committee of the Ghent University Hospital (study no. ECP2004276) and after protocol registration (NCT 00119470), patients were recruited in December 2005 through January 2007 from the institutional thoracic oncology outpatient clinic. Consent to perform all procedures (including thoracotomy) in the Ghent University Hospital was required before screening. Eligible patients had proven or suspected NSCLC, and had a suspected mediastinal lymph node invasion based on computed tomography and/or fluorodeoxyglucose (FDG)-PET scan data. Our guidelines for invasive mediastinal exploration were enlarged (  1-cm short axis) mediastinal lymph nodes and/or FDG uptake in the mediastinal lymph nodes, tumors abutting the mediastinum regardless of FDG uptake in the lymph nodes, and absence of FDG uptake in the primary tumor (14, 15). In addition, the tumor had to be resectable, and the patients had to be functionally operable. A multidisciplinary team approval preceded a possible inclusion. The selection criteria are shown in Table 1. Written informed consent was obtained from all patients before enrolment. The patients were randomly allocated between the following approaches: immediate surgical staging (arm A—gold standard) and EUS-FNA, followed by surgical staging in case of negative EUS-FNA result (arm B).
EUS-FNA EUS-FNA was performed in an out-patient setting under local anesthesia (with or without moderate sedation) using the curved linear scanning ultrasound endoscope (GF-UCT160-OL5; Olympus, Aartselaar, Belgium) connected to the ultrasound unit (Aloka, Mechelen, Belgium) (10), with monitoring of heart rate and oxygen saturation (16). Punctures were performed with a 22-gauge fine needle (EUS-needle; Olympus). Smears of the aspirates obtained by EUS-FNA were processed on site to evaluate the cellular contents of the air-dried specimens with a quick-staining method (Diff-Quick; Medion Diagnostics, Düdingen, Switzerland). If necessary, several lymph nodes were sampled. Specimens were categorized as positive (tumor cells), negative (lymphoid but no tumor cells), or not representative (necrosis, no lymphoid cells). Punctures were continued until the cytopathologist (L.R.V. or M.P.) was able to make a formal conclusion.
Surgical Staging
Thoracotomy
Data Collection
Outcome Measures and Statistical Analysis
Enrollment and Baseline Characteristics of the Patients We screened 44 patients, and 40 underwent randomization (Figure 1). The demographic and clinical characteristics of the patients in the two study groups were well balanced (Table 2). Imaging of the mediastinum suggested N2 disease in 29 (73%) and N3 in 11 (27%) subjects. PET data were available for 38 patients (95%), and suggested malignant lymph node invasion (FDG uptake) in 33 (83%). A total of 25 patients (63%) had proven NSCLC; in the others, there was a high clinical suspicion based on their presentation and imaging studies.
Procedural Characteristics and Tumor Data The surgical technique for staging the lymph nodes was cervical mediastinoscopy in 20 patients and an anterior mediastinotomy in 1 patient for a solitary paraaortic lymph node (lymph node station 6) (Table 3). Both EUS-FNA and surgical staging procedures took about 35 minute to be completed. With EUS-FNA, it was possible to observe the primary tumor in about one-third of the cases, which was significantly more than with surgical staging. Significantly more lymph nodes were sampled with surgical techniques than with EUS-FNA.
The distribution of the primary lung tumor pathology shows that the majority of the patients had NSCLC (Table 3). However, four patients (10%) had small cell lung carcinoma. In one patient, thoracotomy after a negative mediastinal evaluation showed the presence of a benign schwannoma. The distribution of tumor histology was comparable between the two arms, as was the prevalence of malignant lymph node invasion.
Primary and Secondary Outcomes
No complications were noted in the patients investigated with EUS-FNA. In the patient group randomized for surgical staging, one esophageal perforation was noted. Other complications (e.g., infection or hoarseness) were not noted. Because EUS-FNA is performed under local anesthesia as an outpatient procedure, the length of hospital stay was significantly shorter as compared with surgical staging techniques, which require general anesthesia. The test characteristics were not different between the staging approaches (Table 5). When calculating the sensitivity of the surgical staging technique, also taking into account the six patients of the EUS-FNA study arm who underwent an additional surgical staging, the sensitivity becomes 75% (95% confidence interval, 42–94%), which is also not statistically different from the value obtained with EUS-FNA only (93%; 95% confidence interval, 66–99%).
This study shows that, in patients with (presumed) NSCLC without extrathoracic metastasis, in whom mediastinal lymph node exploration is required, EUS-FNA reduces the need for surgical staging procedures by two-thirds (68%). We consider this a relevant clinical benefit for the lung cancer patient, because implementation of EUS-FNA avoids the need for general anesthesia and the potential complications associated with surgical staging. At the time the current trial was designed (2004), some retrospective and prospective series already suggested a benefit of EUS-FNA for staging mediastinal lymph nodes in lung cancer (18, 19). Larger series have since been published, mainly focusing on the diagnostic performance of EUS-FNA in the staging of lung cancer (10, 12, 20). These studies invariably suggested a higher accuracy for EUS-FNA as compared with imaging techniques, and reported on the reduction in the number of surgical staging procedures. However, these studies all had similar flaws in their lack of comparison with the current standard of care (being surgical staging) (5, 6). These uncontrolled accuracy studies are important in the process of generating evidence of feasibility and effectiveness, but are, from a conceptual point of view, not suited to provide information on the difference between the gold standard and a new diagnostic technique (21). Randomized studies are therefore of importance to study advancing diagnostic techniques beyond the level of accuracy analyses. We are aware of only one published prospective, randomized trial in this field, comparing EUS-FNA with "conventional" diagnostic workup of patients with NSCLC (22). The authors showed in their preliminary results that the patients investigated with EUS-FNA had less futile thoracotomies, the latter defined as explorative thoracotomy or recurrent disease after resection. However, the design of that study makes the interpretation of results difficult. Because the investigation technique was tailored to the patient despite randomization, EUS-FNA was performed in the majority of the patients. In addition, there are several methodological differences: EUS-FNA was variably performed under general anesthesia, and there was no on-site evaluation of the samples. The latter has been shown to be valuable for EUS-FNA, especially when performed by a dedicated cytopathologist (10, 11). In the present study, patients were all candidates for radical surgery after a diagnostic work-up adhering to the available guidelines. Furthermore, and in contrast with previous series, the pathology assessment of the mediastinal lymph nodes was strictly defined. As such, we provide solid evidence that EUS-FNA should be implemented in the preoperative staging algorithm of patients with (presumed) lung cancer who have no evidence of extrathoracic metastasis. In addition, the current data show that the implementation of EUS-FNA in the staging algorithm of lung cancer results in less complications and a shorter hospital stay as compared with conventional staging. However, the latter statement might not be applicable when surgical staging is performed as an outpatient procedure (23–25). No difference was found in the diagnostic performance between study arms. Although the sensitivity of rapid on-site examination–assisted EUS-FNA to detecting malignant lymph node invasion tended to be higher than that with surgical staging (10, 26), this difference was not statistically significant, even when taking into account the patients who underwent both procedures. Together, the present data are in agreement with the test characteristics that have been published for these techniques (7), and indicate that implementing EUS-FNA does not result in a lower sensitivity. The negative predictive value of EUS-FNA is too low to preclude further lymph node sampling in case of a negative result. This is in agreement with published data (12). The results of the present study do not allow a conclusion about the preferred diagnostic method in the case of negative EUS-FNA. Because it was shown, in a prospective series, that the combination of mediastinoscopy and EUS-FNA improves the sensitivity and negative predictive value (27), we opted to use a mediastinoscopy before referring the patient to thoracotomy. It is impossible to rule out false-positive findings either for EUS-FNA or for surgical staging procedures. This is related to the current study design. However, it seems reasonable to assume that the false-positive rate is low in mediastinal staging (6), although one study showed that sampling a left lower lobe tumor instead of a mediastinal lymph node might result in overstaging (27). Another finding with potential implications is a difference in the completeness of mediastinal staging between the two studied techniques, as the number of sampled lymph nodes was significantly lower in the patients investigated only with EUS-FNA. This has to do with the anatomy of the mediastinum and the interposition of the trachea impeding the investigation of solitary lymph nodes located strictly in the lower right-sided paratracheal area (region 4R). A weakness of this study lies in EUS-FNA missing malignant N3 invasion whenever only N2 is shown. However, this has no implication for treatment allocation, as combination chemoradiotherapy, especially delivered concurrently, is the preferred treatment for locally advanced NSCLC, regardless of the level of mediastinal lymph node involvement (28). Moreover, even if one should consider surgery after induction chemotherapy for IIIA NSCLC, invasive restaging of the mediastinum is currently advocated. Remediastinoscopy has been shown to be inaccurate (29), which is an additional reason for assessing the mediastinum with minimally invasive techniques before any treatment is given. Nevertheless, early data based on selected patients suggest that endobronchial ultrasound with real-time guided transbronchial needle aspiration could overcome the problem of incomplete mediastinal sampling (30–32). Whether the combination of this method and EUS-FNA will become an alternative for overcoming the aforementioned problem is currently under investigation (32, 33). Based on the outcomes of this study, we conclude that EUS-FNA reduces the need for surgical staging procedures in patients with (suspected) lung cancer in whom a mediastinal exploration is needed. Therefore, we propose that, in these patients, the mediastinum should be first investigated with EUS-FNA, followed by surgical staging in the case of negative findings.
The authors wish to thank the patients who participated in this trial. We are grateful for the referring clinicians (J. Boie, K. Carron, S. Fieremans, M. Mallet, S. Marien, D. Ommeslagh, H. Ottervaere, A. van Boxem, J. Van Den Bergh). We also thank our data managers (F. Vandewalle and I. Derudder), nurses (Y. Loyen, V. Vermont, B. De Burchgraeve and G. Van Severen) and lab technician (Y. Sturtewagen), whose contributions were very important.
Originally Published in Press as DOI: 10.1164/rccm.200708-1241OC on October 25, 2007 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form August 22, 2007; accepted in final form October 25, 2007
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
