Table 1

 Characteristics of studies in the meta-analysis*

Study (reference)YearPrevalence§NSCLC (%)TBNA results‡‡ (%)Inclusion criteria and comments
TPFNFPTN
TBNA, transbronchial needle aspiration; NSCLC, non-small cell lung cancer; SCLC, small cell lung cancer; LAD, lymphadenopathy; CT, chest computed tomography; CXR: chest radiograph; NA, not applicable; TP, true positive; FN, false negative; FP, false positive; TN, true negative; –, studies in which positive TBNA results were not confirmed surgically but were assumed to be true positives.
*Five studies30,31,33,34,36 reported age (median 60 years) and six studies 30,31,33–36 reported sex characteristics (median proportion male 91%).
†Studies meeting tier 1 criteria surgically confirmed all TBNAs had at least 10 subjects with and without mediastinal lymph node metastasis and used the patient as the unit of analysis.
‡Additional information obtained from original study author(s).
§Prevalence of mediastinal lymph node metastasis. Studies that did not surgically confirm all TBNA results assumed that the false positive rate was zero.
¶All mediastinal and hilar lymph nodes (⩾8 mm) identified by CT scanning in this study were sampled by TBNA. To estimate TBNA diagnostic accuracy for identifying mediastinal metastasis, patients with N0 or N1 disease on pre-TBNA CT scanning were excluded from the analysis (the combined mediastinal/hilar TBNA results for NSCLC were: TP 49, FN 18, FP 0, TN 16). The sensitivity (73% v 73%; p = 0.97) and specificity (100% v 100%; p = 1.0) were similar whether or not N0/N1 disease on CT scanning was excluded.
**Two studies used real-time radiological assistance to guide needle placement during TBNA.
††Three of 42 patients undergoing TBNA had SCLC. Thirty two patients who had hilar lymph node biopsies (not mediastinal) were excluded.
‡‡Number of patients (except for the study by Patelli et al,35 where statistical analysis was by TBNA specimen).
Tier 1 studies
Harrow et al28200034100817148Patients with suspected lung cancer were included. Patients without lung cancer, or TBNA from a lymph node confluent with the tumour mass were excluded by the authors.
Bilaceroglu et al29199860100249022Patients with potentially resectable extrabronchial or endobronchial mass suggestive of lung cancer and without extrathoracic metastases were included. We excluded patients with N0 or N1 disease on pre-TBNA CT.¶
Disdier et al3019985210059112Patients with potentially resectable lung cancer without evidence of extrathoracic metastases and with enlarged mediastinal LAD on CT or CXR were included.
Ratto et al3119883096212033Patients with potentially resectable lung cancer (no preoperative SCLC) without evidence of extrathoracic metastases were included. Only subcarinal lymph nodes were biopsied.
Schenk et al32198629100813250Patients with CXR evidence of lung cancer and who were potentially resectable without evidence of extrathoracic metastases were included. One FP had scanty neoplastic cells without lymphocytes.
Studies not meeting tier 1 criteria
Herth et al33‡ **20028210030608Patients with central lung cancer and enlarged mediastinal LAD by CT without evidence of extrathoracic metastases were included. Needle placement via endobronchial ultrasound.
Wang et al34200210093††10000Inclusion criteria not specified. Unable to calculate specificity (100% prevalence).
Patelli et al352002NA1001275215Patients with NSCLC, N2 disease by CT scanning and a negative bronchoscopy for lung cancer were included. Statistical analysis was by TBNA specimen. Data were collected by retrospective chart review. Non-surgically confirmed negative TBNA specimens (total 49) were assumed to be false negative.
Katis et al361998951002882Patients with CXR evidence of lung cancer and enlarged mediastinal LAD on CT and who were potentially resectable and without evidence of extrathoracic metastases were included.
Rong et al37‡ **19987910026205Patients undergoing thoracotomy for mediastinal adenopathy on CT and suspected lung cancer were included. Bronchoscopist was not experienced with TBNA. Real-time CT assisted needle placement.
Schenk et al381993811003269Patients with resectable lung cancer and paratracheal LAD without extrathoracic metastases were included. Four of the 32 TPs were confirmed surgically.
Schenk et al391989811001434Patients with CXR evidence of lung cancer and who were potentially resectable and without evidence of extrathoracic metastases were included.
Wang et al4019835510013313Patients with suspected lung cancer and who were potentially resectable and without evidence of extrathoracic metastases were included.