| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
ORIGINAL ARTICLE |
Division of Cardiothoracic Surgery, Department of Surgery, Queen Mary Hospital, Hong Kong SAR, China
Alva KY Sit, MRCS, Tel: +852 2855 3973, Fax: +852 2819 4819, Email: alvasit{at}doctor.com, Division of Cardiothoracic Surgery Department of Surgery, Queen Mary Hospital, Hong Kong SAR, China.
ABSTRACT
A potential limitation of integrated positron-emission tomography and computed tomography in non-small-cell lung cancer may be false-positive results due to granulomatous disease. This retrospective study examined the accuracy of this imaging modality for mediastinal nodal staging of non-small-cell lung cancer in Hong Kong where tuberculosis remains endemic. There were 249 lymph node stations evaluated in 107 patients, of whom 38 (36%) had active tuberculosis or evidence of previous tuberculosis. Imaging results were compared with histological findings. The sensitivity, specificity, and accuracy of integrated imaging for mediastinal nodal staging were 52%, 86%, and 80%, respectively; the overall positive-predictive value for mediastinal nodal metastasis was 46%, and the overall negative-predictive value was 89%. The positive-predictive value for mediastinal nodal metastasis was 39% in patients with tuberculosis and 50% in controls; the negative-predictive value was high in both groups (92% and 87%). The likelihood ratio for true positives was 6.47 in patients with tuberculosis vs. 10.97 in controls. This suggests that the reliability of positron-emission/computed tomography may be substantially poorer in patients with tuberculosis. Histological confirmation should be considered mandatory in patients with suspected metastasis on integrated imaging.
Key Words: Lung Neoplasms • Neoplasm Staging • Tomography • Emission-Computed • Tomography • X-Ray Computed • Tuberculosis
INTRODUCTION
Lung cancer remains the leading cause of cancer death worldwide. Surgical resection is the best curative therapy for patients with early stage lung cancer.1 When staging lung cancer patients, the status of the intrathoracic lymph nodes must be considered. Surgery remains the treatment of choice for stage II disease with hilar nodal metastases (N1), whereas combination chemoradiotherapy is preferred for prospectively recognized stage IIIA disease with mediastinal nodal metastases (N2).1,2 This highlights the importance of accurate preoperative nodal staging. Computed tomography (CT) is most commonly used for preoperative noninvasive lung cancer staging, but despite its fine spatial resolution, it cannot reliably differentiate between benign and malignant lesions. Over the past decade, positron-emission tomography (PET) has emerged as a powerful staging tool that can indicate the likelihood of malignancy in a lesion, based on metabolic activity.3 So far, PET has been shown to be useful for the evaluation of indeterminate pulmonary nodules, mediastinal lymph nodes, and extrathoracic metastases.3–6 Although CT provides much more anatomic detail, PET is considered to be more accurate for lung cancer staging. Newer-generation integrated PET/CT imagers combine the advantages of both.7 However, despite the promise shown by PET and PET/CT, most studies on their use in lung cancer have been based on Western populations in which the prevalence of tuberculosis (TB) and similar pulmonary inflammatory diseases is low. Crucially, a key limitation of PET and PET/CT is the relatively disappointing positive-predictive values (PPV) because inflammatory disease may mimic malignant lesions. It has not yet been established whether the excellent results of PET/CT reported in the West can be reproduced in regions where TB remains endemic and relatively prevalent. Hong Kong is still considered a region where TB is endemic.8 The objective of this study was to determine the accuracy and effectiveness of PET/CT scanning for preoperative mediastinal lymph node staging in lung cancer patients being considered for surgery in Hong Kong.
PATIENTS AND METHODS
This was a retrospective cohort study conducted in the cardiothoracic surgery unit of a tertiary-referral university teaching hospital. All patients who underwent PET/CT before lung resection surgery with curative intent for suspected or proven non-small-cell lung cancer (NSCLC) from January 2003 to January 2006 were included. Their demographic and clinical data were retrieved using hospital medical records and the computerized database of patient records of the Hospital Authority of Hong Kong. Patients whose final histology results confirmed a metastatic tumor from an extrapulmonary source, or who were known to have current or previous metastatic lung cancer, and those who received preoperative chemotherapy and/or radiotherapy, were excluded. Patients were also excluded if no lymph nodes were ultimately biopsied during the operation. Patients who were found to have non-resectable disease at surgery, but who nonetheless had lymph nodes biopsied during the operation, were included; some had staging mediastinoscopy prior to thoracotomy, in which frozen sections of nodal biopsies confirmed metastatic disease. Patients suspected of having lung cancer who underwent surgery with mediastinal nodal biopsy, but were subsequently found not to have lung cancer, were not excluded from the analysis, in accordance with intention-to-treat principles.
During the study period, PET/CT scanning was not available under the public health service provided in Hong Kong. Although patients were advised of the availability of PET/CT in private hospitals in Hong Kong, whether it was performed was ultimately determined by each patient’s willingness to pay for the private service. The PET/CT scans in this study were carried out in several private hospitals in Hong Kong, using intravenous injection of 18 F-fluorodeoxyglucose. To standardize the PET results from different hospitals for the purpose of this study, we took a standardized uptake value 
2.5 on PET/CT plus the radiologist’s report of each scan to indicate the presence of malignant cells in a particular lesion. In all cases, surgery was performed within 30 days of the PET/CT scan. All patients underwent routine preoperative investigations in addition to PET/CT, including CT and fiberoptic bronchoscopy. Mediastinoscopy or mediastinotomy is routinely performed when PET/CT suggests the possibility of mediastinal nodal metastasis.
From 2006 onwards, complete video-assisted thoracic surgery was used for most lung cancer resections at our hospital; these patients were excluded to preserve cohort homogeneity. Thus all lung resection surgery during the study period was performed via an open thoracotomy. The extent of resection ranged from wedge resection to total pneumonectomy, depending on the individual circumstances of each patient. Systematic lymph node sampling (not radical nodal clearance) was carried out at all ipsilateral mediastinal lymph node stations. Lymph node stations were defined using the standard classification of Mountain and Dresler.9 Nodal stations 1 to 9 on the Mountain-Dresler classification were defined as mediastinal lymph node stations in this study; stations 10–14 are the hilar lymph node stations. The histology report on each mediastinal lymph node biopsied during surgery was taken to be the gold standard for determining the presence or absence of metastases in that node. The histology report was compared with the PET result for each mediastinal node. From this, the sensitivity, specificity, PPV, and negative-predictive value (NPV) of PET/CT in detecting nodal metastases were determined for each mediastinal nodal station.
All statistical analyses were performed using SPSS version 11.0 software (SPSS, Chicago, IL, USA). Results were considered significant at p < 0.05.
RESULTS
The demographic and clinical characteristics of all 107 patients who met the selection criteria are shown in Table 1
. Of these, 15 (14%) patients did not undergo thoracotomy after mediastinoscopy or mediastinotomy had confirmed metastases. Of the 92 patients who underwent thoracotomy, 4 (4.3%) were found to have non-resectable disease on-table. Among the 107 study patients, 112 lung lesions were either resected, excised, or biopsied. The histological findings of these lung lesions are shown in Table 2. Benign disease coexisted with malignancy in 5 (4.7%) patients, and 7 (6.5%) were ultimately found to have no malignancy. When the PET/CT findings were compared with the histological results of these 112 lung lesions, PET/CT had a sensitivity of 91%, specificity of 57%, and accuracy of 87% in assessing malignancy in the primary lung lesion. PET/CT in this role had PPV and NPV values of 94% and 47%, respectively. However, on subgroup analysis, if the primary lung lesion was 2 cm in size, PET/CT yielded a slightly higher overall accuracy of 96%.
|
|
and 4. When comparing PET/CT findings with the histological results of all sampled lymph nodes (mediastinal and hilar), PET/CT was noted to have a sensitivity of 35%, specificity of 84%, and accuracy of 74% in detecting nodal metastases. PET/CT falsely identified 31 (12.4%) lymph nodes as being positive for metastasis. If only N2 mediastinal lymph node stations were considered, PET/CT yielded a slightly higher accuracy of 80%. Nonetheless, the PPV remained low at 46%, and 11.5% of mediastinal lymph nodes were falsely suggested to be metastatic by PET/CT. Further subgroup analysis was performed based on whether the mediastinal nodes were located in the superior mediastinal (stations 1–4), aortic (stations 5 and 6), subcarinal (station 7), or inferior mediastinal (stations 8 and 9) areas. However, this failed to demonstrate any significant correlation between PET/CT accuracy and nodal station (Table 4). In the 100 patients with lung cancer in this study, the cTNM staging by preoperative PET/CT scan was compared with the final pTNM staging by histology of the surgical specimens. It was found that 25 (25%) patients were under-staged by PET/CT, and 21 (21%) were over-staged. In particular, PET/CT suggested that 11 (11%) patients had inoperable disease (cTNM stage III/IV), but they were found to have early stage (pTNM stage I/II) disease. On the other hand, 12 (12%) patients were suspected to have early stage (cTNM stage I/II) disease on PET/CT, but they turned out to have advanced (pTNM stage III/IV) disease.
|
|
). The PPV for mediastinal nodal metastasis in TB patients was lower than in non-TB patients. The NPV of PET/CT was high in both TB and non-TB patients. The likelihood ratios for true positives with PET/CT nodal staging were 6.47 in TB patients (p =0.02) and 10.97 in non-TB patients (p =0.001). This suggests that the reliability of PET/CT in detecting mediastinal nodal metastases is substantially better in patients without TB than in those with active or previous TB.
|
Although PET and PET/CT have become established as powerful and reliable tools for the routine staging of lung cancer patients, these techniques are not infallible. In addition to failure to detect small metastatic deposits and relative weakness in detecting hilar metastases with central lung tumors, a more consistently noted weaknesses of PET is the relatively low PPV.10,11 This is attributed to high false-positive rates that are due to the high fluorodeoxyglucose uptake by benign lesions, which mimics that of malignant deposits on PET. Such benign lesions include those due to recent radiotherapy, surgery, or inflammatory disease; the latter can be the result of TB and granulomatous infections.
In this study, we showed that PET/CT yields a diagnostic accuracy for mediastinal staging that is comparable to results reported elsewhere in the world (Table 6). The list of primary studies and meta-analyses in this table is hardly exhaustive, but nonetheless the figures are quite typical. The specificity and accuracy of PET/CT in Hong Kong do not fall far short of those seen in other countries. In particular, the NPV is very high. This suggests that it is entirely feasible to follow international guidelines on the use of PET in Hong Kong to exclude mediastinal nodal metastases in lung cancer patients who are potential candidates for surgery.17 Nevertheless, whether a negative PET finding completely obviates the need for mediastinoscopy is still debatable.18,10 Also obvious in Table 6 are the low sensitivity and low PPV compared to reports from the West. The low PPV is consistent with the hypothesis that false positives due to benign pathologies such as TB may have confounded the accuracy of PET/CT in our region. Certainly, the fact that over one third of all patients in this study had current or previous TB suggests that this infection may play some role. The simplest method to verify this would be to check all histology results for false-positive mediastinal lymph node findings. Regrettably, this was not possible in this retrospective study because the pathologists were not alerted to specifically look for nonmalignant diagnoses in the lymph nodes during the study period. As a consequence, the majority of nodes were reported simply as "no malignancy" and it is now impossible to determine whether TB was present in those nodes. We acknowledge that this is a weakness, and we hope to address it in future studies.
|
Where our study has ventured beyond previous studies on PET for lung cancer is in demonstrating that a history of current or previous TB may impact on PET reliability. The likelihood ratio of PET/CT in detecting a true-positive mediastinal metastasis was significantly higher in patients with no history of TB. The clinical significance of this result is that PET/CT can still be employed with confidence in non-TB patients, even in a TB-endemic region. In patients with TB history, PET/CT findings should be interpreted more carefully, and if doubt exists, they may benefit from a more rigorous staging investigation; however, the NPV remains very good, suggesting that PET/CT should not be discarded altogether when investigating such cases.
We acknowledge that some additional limitations exist in this study. Firstly, all patients were potential surgical candidates on the basis of preoperative investigations. Thus there could have been selection bias towards earlier stages of disease, and patients with microscopic nodal metastasis might have been more selectively included. Furthermore, because surgeons are invariably guided by CT and PET/CT findings when operating, there could have been selective targeting of certain nodal stations for sampling during surgery, leading to a degree of verification bias. Also, despite our efforts to maintain cohort homogeneity, the patients received a variety of procedures ranging from mediastinoscopy only to open-and-close thoracotomy and major lung resection with systematic nodal sampling. This may have impacted on the adequacy of mediastinal nodal biopsy in some patients, and the low average lymph node yield per patient in this study suggests this may have been the case. If so, this may be another source of verification bias, putting into question the reliability of the histology results as the gold standard for determining metastasis. On the other hand, we believe that the accuracy achieved in this study compared to previous reports suggests that these issues may not have had an overwhelmingly negative impact.
It was concluded that PET/CT retains good diagnostic accuracy for mediastinal staging of lung cancer in TB-endemic regions. Certain caveats apply: in particular, the low PPV and diminished sensitivity should be borne in mind when interpreting results; and extra caution should be maintained when assessing PET/CT scans in patients with history of TB. However, our results fully support the current lung cancer staging guidelines that advise the use of PET in all surgical candidates, and the need for further invasive staging when suspicion exists.17
Presented at the 18th World Congress of the World Society of Cardiothoracic Surgeons, Kos, Greece, April 30–May 3, 2008.
REFERENCES
Asian Cardiovasc Thorac Ann 2010;
18:33-38
© 2010 by SAGE Publications
DOI: 10.1177/0218492309352119
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
