Endobronchial ultrasound and positron emission tomography positive mediastinal lymph nodes

Lung cancer remains the first cause of cancer-related deaths, with an overall 5-yr survival rate of 15%. The best hope for cure is for patients with potentially resectable disease, but even in these cases, the 5-yr survival rate is only in the order of 40–50% 1. Adequate staging before surgery is, thus, of paramount importance to better stratify the therapeutic approach and to limit the number of futile thoracotomies. In particular, patients with N2 mediastinal lymph node involvement are poor candidates for initial surgical resection, since 5-yr survival rate in this condition ranges, according to various series, from 11–30% only 2.

Computed tomography (CT) has traditionally been used to predict lymph node involvement, and a short axis of lymph node exceeding 1 cm has been chosen to indicate a high probability of neoplastic involvement. Using this criterion, however, the sensitivity and specificity of CT are only 0.57 and 0.82, respectively 3. Recently, the positron emission tomography with 18F-fluoro-2-deoxy-d-glucose (FDG-PET) has been shown to be more accurate than CT in the evaluation of mediastinal lymph node involvement 4. It is now used in many countries, in addition to CT, to further improve the sensitivity and specificity rates of the noninvasive assessment of mediastinal lymph nodes. This is reflected in recent consensus recommending that, in patients with enlarged mediastinal lymph nodes on CT scans who are potential candidates for surgery, a whole-body FDG-PET scan should be performed 5. FDG-PET has a sensitivity of 0.85, a specificity of 0.88, a positive predictive value of 0.78, and a negative predictive value of 0.93 3. It has then been suggested that a negative mediastinal FDG-PET result may obviate the need for invasive lymph node sampling 4, but a positive one needs to be confirmed 4, 5.

In case of positive mediastinal FDG-PET results, further assessment using mediastinoscopy is still considered as the “gold-standard” 3. Mediastinoscopy, however, is a surgical procedure with a mortality rate of 0.2% and a morbidity rate of 0.5–2.5% 6. In addition, it requires general anaesthesia and at least one overnight stay in the hospital 6. Transbronchial needle aspiration (TBNA) of lymph nodes, which was developed >50 yrs ago 7, may circumvent some of the drawbacks of mediastinoscopy. Surprisingly, this technique remains largely underutilised 8 and this is probably explained, at least in part, by concerns about the practice learning curve and by the fact that, until recently, this procedure was not guided by imaging. In a recent study, Herth et al. 9 have shown that TBNA might be assisted by previous localisation of mediastinal lymph nodes using a new tool, endobronchial ultrasound (EBUS). They also showed that this strategy was associated with an increased diagnostic yield of TBNA.

The current authors' reasoned that the diagnostic yield of TBNA might be even higher if, in addition to lymph node sampling after EBUS localisation, evaluation was specifically performed in the lymph node areas showing abnormal results during FDG-PET scan. With this in mind, all patients referred to the endoscopic unit of Institut Bordet, Saint-Pierre Hospital (Brussels, Belgium), for diagnosis and/or staging of mediastinal lymph nodes with abnormal results of FDG-PET scan were assessed with EBUS followed by TBNA, and needle aspiration was concentrated in areas shown to be abnormal by FDG-PET. As a secondary aim, the current authors also tried to assess in which percentage of the cases this strategy might obviate the need for mediastinoscopy or other major surgical diagnostic/staging procedures.

MATERIALS AND METHODS

All the patients with FDG-PET positive lymph nodes referred to the endoscopic unit of Institut Bordet, Saint-Pierre Hospital, for evaluation were considered. Of them, 17 (52%) were referred from eight other hospitals. All patients underwent EBUS examination for localisation of the FDG-PET abnormal lymph node area followed by TBNA. Data were prospectively collected and analysed to assess the diagnostic yield of TBNA in this particular setting. These patients had three different initial clinical presentations. In group I (staging), lung cancer was histologically confirmed, without any evidence for distant metastasis, and FDG-PET was suggestive of mediastinal lymph node involvement. Group II (diagnosis and staging) included patients with lesions that were suspected to correspond to lung cancer, but the initial bronchoscopy was not contributive and FDG-PET was abnormal at the level of the tumour and of the mediastinum. Group III (lymphadenopathy work-up) was more heterogeneous, and included patients with one or more FDG-PET positive lymph nodes without evidence of lung tumour.

Bronchoscopy was performed using a large working channel bronchoscope (Olympus Excera; Olympus, Tokyo, Japan) while patients were comfortably seated in a 30° recumbent position in a high-backed armchair. Oxygen (2 L·min⁻¹) was administered with nasal prongs, and transcutaneous haemoglobin saturation and cardiac rhythm (Ohmeda Biox 3740; Ohmeda, Louisville, CO, USA) were monitored continuously during the whole procedure. Analgesia of the airways was performed as previously described 10, with conscious sedation using intravenous midazolam; patients were evaluated on a real outpatient basis. A 20-MHz radial mechanical transducer type ultrasonic probe (UM-BS20-26R; Olympus) in a flexible sheet equipped with a balloon at the tip (Maj-643R; Olympus), connected to a EBUS processor (EU-M20; Olympus) was inserted in the channel and used to localise the abnormal FDG-PET lymph nodes and to make precise their relationship with the tracheobronchial tree. Examination was performed with the balloon inflated to optimise contact between the miniaturised probe and the bronchial wall. It was also performed with the small probe applied directly on the wall, without balloon inflation, attempting to better demarcate sampling location (fig. 1⇓). The catheter was then pulled out and a combined cytological and histological needle (BARD-Wang, Billerica, MA, USA) was systematically used for TBNA. In each patient, four to six punctures were performed, beginning with the highest staging node level 11, 12 in the case where several areas were FDG-PET abnormal. In case of failure of EBUS to localise lymph nodes, sampling location was based on CT and PET scan information using the methodology previously described 11. No rapid on-site cytological examination was used.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1—

a) Endobronchial ultrasound image of a lymph node (LN) with a hypoechoic centre at level 4R. The balloon is not inflated and the small ultrasonic probe (P, diameter 2.5 mm) is in direct contact with the tracheal wall. The small size of the probe, without balloon inflation, allows better demarcation of the sampling location. b) Computed tomography scan at level 4R showing two LN, the largest corresponding to the lymph node represented in a).

For the purpose of analysis, TBNA was considered as contributive whenever a clear definite cytological or histological diagnosis (including in the latter case normal lymphoid tissue) was obtained. It was not contributive if no diagnosis (including only rare suspicious cells without lymphocytes on cytological examination) could be reached. In these latter cases, surgical staging/diagnostic procedure was recommended and, if it was not performed, final diagnosis was then based on adequate clinical follow-up. A surgical staging/diagnostic procedure was considered as eluded whenever TBNA provided a positive result in groups I and II, and a definite diagnosis in group III, and on the condition that there was no alternative, less invasive method than surgery.

Unless otherwise specified, results are expressed as mean±sd. No comparison was made between the three different groups of patients, or between patients with successful or unsuccessful previous localisation with EBUS, because the numbers in the study were too low.

RESULTS

From January 2003 to June 2004, 33 patients with positive FDG-PET mediastinal (hilar) lymph nodes were referred for TBNA exploration after previous evaluation with EBUS. Their main characteristics are summarised in table 1⇓. The bronchoscopist could always look at the FDG-PET images and the CT scans, but data on quantitative assessment, including standardised uptake value (FDG-PET) or lymph node size (CT scan), were often missing due to nonstandardised initial diagnostic work-up (nine different centres). The procedure, performed under local anaesthetic, was well tolerated and no significant side-effect was encountered (with the exception of some agitation, in one patient, that limited the number of needle punctures to one, and cough that occurred in some patients but without significantly disturbing examination and sampling). FDG-PET scan located abnormal lymph nodes at the level of mediastinum in 30 cases and at the hilar level in three cases. The exact location of the target lesion by EBUS was obtained in 24 (73%) patients. The main reasons for EBUS failure were, in one case only, a technical problem with the probe, and difficulty in obtaining good visualisation of lymph nodes, in particular around the trachea, mainly because of initial lack of EBUS experience and limitations due to local anaesthesia. Indeed, full inflation of the balloon in the large trachea is often followed by balloon detachment and, in this particular case, examination was continued with the probe directly applied on the wall. Under the latter conditions, the small size of the probe (relative to inflated balloon) often allows better demarcation of the sampling location, but adequate contact (and imaging) is only available in about three-quarters of the cases (fig. 1⇑). EBUS localisation of lymph nodes required an additional time of ∼5 min.

A total number of 139 samples (mean number per patient: 4.2±1.5) were obtained in 39 different lymph node stations. The lymph node stations that were most frequently explored with TBNA were 4R (33.3%, 13 patients) and 7 (33.3%, 13 patients), followed by 4L (12.8%, five patients), 11R (10.3%, four patients), 11L (5.1%, two patients) and 10L (2.6%, one patient). In one case (2.6%), it was unclear whether the target lesion was the primitive lesion located in close vicinity to the bronchial tree or contiguous agglomerated lymph nodes. Cytological sampling was obtained in 32 of the 33 patients (97%), and tissue sampling in 23 patients (70%). In two patients with non-neoplastic lymph node disorders, including one case of sarcoidosis, cytological sampling was not contributive, whereas tissue sampling allowed a diagnosis to be reached. In the case with anthracosis, follow-up until 8 months was also uneventful in supporting benign disease. Surprisingly, in one patient with small cell carcinoma, only tissue sampling was diagnostic.

Table 2⇓ shows the final diagnosis, and the overall diagnostic yield of TBNA, as well as the diagnostic yield after previous localisation using EBUS. TBNA allowed diagnosis to be reached in 27 (82%) of the patients, of whom 78% were diagnosed after previous EBUS localisation. Previous localisation with EBUS was associated with a TBNA diagnostic yield of 88% (21 out of 24) and failure of EBUS localisation with a diagnostic yield of 67% (six out of nine). A definite diagnosis was more often possible in case of neoplastic (25 out of 27, 92%) than non-neoplastic involvement (two out of six, 33%). In the six cases (18%) without diagnosis after TBNA, lymph cells were observed in smears by the cytologist in only three cases. Further surgical sampling (mediastinoscopy) was performed in three of these six patients (two from group I and one from group III), and showed lymph node tissue without malignant involvement. In two additional cases, the multidisciplinary medical team considered that the lymph nodes were likely to be invaded (in one case, a synchronous advanced head and neck cancer was also discovered) and in the last case, clinical follow-up was chosen by the referring team and was uneventful at 20 months, supporting non-neoplastic disease.

Surgical sampling (mediastinoscopy or other surgical sampling of mediastinal lymph nodes) was avoided by the positive results of TBNA in 25 cases (76% of the patients). In one case from group II (diagnosis and/or staging), an alternative procedure would have been possible in case of negative TBNA, namely a puncture of suspect liver metastasis.

DISCUSSION

The present study shows that TBNA combined with previous EBUS examination is a very safe and effective method with which to assess patients with FDG-PET positive lymph nodes. This technique could be performed under local anaesthetic, on a real outpatient basis and without any complication, and was associated with a high diagnostic yield. It also allowed avoidance of surgical diagnostic/staging sampling in 76% of the patients. To that extent, this technique should be considered as a primary method of histological sampling in patients with FDG-PET positive mediastinal lymph nodes. In case no lymph node metastases are demonstrated, however, due to low negative predictive value 5, further surgical sampling should be performed.